University of Southern California Dissertations and Theses

Engineering 2D & 3D microphysiological systems for interrogating skeletal muscle tissues

(USC Thesis Other)

Engineering 2D & 3D microphysiological systems for interrogating skeletal muscle tissues

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

ENGINEERING 2D & 3D MICROPHYSIOLOGICAL SYSTEMS FOR
INTERROGATING SKELETAL MUSCLE TISSUES

by

JEFFREY WINSTON SANTOSO

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
BIOMEDICAL ENGINEERING

AUGUST 2022

Copyright 2022 Jeffrey Winston Santoso
ii

Acknowledgements
I offer my deepest gratitude to my advisor, Dr. Megan McCain, whose guidance
has let me achieve so many of my personal, academic, and career goals and learn how
to set up new ones. I admire your dedication and sincerity in helping each of us in the lab
find our own success. I am so proud to have contributed a small part in the growth of your
lab and am looking forward to seeing how it will continue to push tissue engineering
research towards new horizons.

Thanks to my remaining defense committee members – Drs. Dion Dickman, Justin
Ichida, Leonardo Morsut, and Keyue Shen – for their service, mentorship, and feedback
on these inherently interdisciplinary projects. This appreciation also extends to Drs. Carrie
Miceli and Florian Barthelemy (and the rest of the Center for Duchenne Muscular
Dystrophy at UCLA). I hope our existing collaborations will continue to flourish and that
new ones will soon begin.

To my colleagues in the Laboratory for Living Systems Engineering – Davi,
Nethika, Andrew, Nathan, and Joycelyn: as my predecessors, you set the foundation of
the lab – not just the physical infrastructure (which was a pain in itself), but also a lab
culture that values inclusion, curiosity, growth, pride, and service. Thank you for your
letting me learn from your different perspectives and leadership styles. I hope I passed
that same spirit on to Natalie, Nina, Mher, James, Stephanie, and our many
undergraduate researchers. Additional appreciation to our postdocs Megan R. and
iii

Patrick for their friendship and advisement, and special thanks to Gio, Divya, and Finacy
for putting up with me as a graduate mentor. Keep up the good work, everyone!

There are numerous other colleagues and friends that have positively impacted
me during the PhD journey to thank (there are many others), but I am particularly grateful
for:
- My DRB neighbors in Shen lab – Yuta, Hao, and Hydari, in particular – you were
the first people I met at USC and have been nearly daily pillars of support for the
past six years.
- My colleagues in Chung lab – Deborah, Noah, and Jonathan – I will miss the
mentally therapeutic activity of spending too much money on food and drinks all
around LA.
- Mischal, Chris, and William – whatever you get paid is at least a magnitude less
than what you deserve for the work you put in daily as not only student advisors,
but also event organizers, recruiters, rulemasters, and all-around disaster
prevention czars.
- Gino, Jason, Duy, Joel, Sam, Will, Daniel, Allen, Simon, Eric, Darryl, and Albert –
the original crew. Thanks for being a constant presence I can come home to for
10+ years.

I would also not be here without Dr. Brittany Davis – you gave me my first lab
opportunity working alongside you in Dr. George Truskey’s lab at Duke. I literally cannot
think of a bigger inflection point in my entire life; thank you for opening the door to a
iv

research career for me. The way you valued my learning experience continues to
motivate me to mentor others with the same level of urgency as I would have for the
pursuit of my own growth.

To my parents, grandparents, brother, and extended family – I know several
sacrifices were made so that I could be where I am today. I will always appreciate the
opportunity I was given to lead the life that I want. Your support means so much to me.

And to anyone who is reading this, or any part of this behemoth of a document,
thank you for your interest in my journey and my work. It was not always easy, but I
thoroughly enjoyed my time as a BME PhD student at USC.

-JWS, 04/15/2022

Table of Contents
Acknowledgements ..........................................................................................................ii
List of Tables ................................................................................................................. viii
List of Figures ..................................................................................................................ix
Abstract ..........................................................................................................................xv
Chapter 1 Introduction ..................................................................................................... 1
1.1. Neuromuscular Tissue Development and Physiology ................................ 2
1.1.1. Skeletal Muscle Tissue and Mechanisms of Contraction ..................... 3
1.1.2. Motor Neurons and the Neuromuscular Junction ................................. 5
1.2. Pathological Degeneration of Motor Unit Components ............................... 8
1.2.1. Genetically Inherited Neuromyopathies – Spinal Muscular
Atrophy, Duchenne Muscular Dystrophy, and Charcot-Marie-Tooth
Disease ................................................................................................ 9
1.2.2. Sporadically Acquired Neuromyopathies – Myasthenia Gravis .......... 11
1.2.3. Neuromuscular Diseases with Mixed Etiology - Amyotrophic
Lateral Sclerosis................................................................................. 12
1.3. Case Study: Existing Approaches to Modeling Amyotrophic
Lateral Sclerosis ....................................................................................... 14
1.3.1. In Vivo Animal Models ........................................................................ 15
1.3.2. Traditional In Vitro Models and Microphysiological Systems .............. 16
1.3.3. Outlook – Current Challenges in Modeling Neuromyopathies ............ 22
1.4. Objectives and Aims ................................................................................. 23
Chapter 2 Implementing 2D and 3D Skeletal Muscle Microphysiological Systems
with Advanced Structural and Functional Markers ....................................... 25
2.1. Introduction .............................................................................................. 25
2.2. Materials and Methods ............................................................................. 26
2.2.1. 2D Gelatin Hydrogel Substrate Fabrication ........................................ 26
2.2.2. Skeletal Muscle Cell Culture .............................................................. 28
2.2.3. 2D Myotube Structural Characterization ............................................ 30
2.2.4. Quantifying Contractile Stresses on Muscular Thin Films .................. 32
2.2.5. RNAseq and Bulk Transcriptomic Analysis ........................................ 33
2.2.6. 3D Myobundle Mold Fabrication ........................................................ 35
2.2.7. Seeding and Culturing 3D Myobundles .............................................. 36
2.2.8. 3D Myotube Structural Characterization ............................................ 37
2.2.9. Quantifying Contractile Force Generation of 3D Myobundles ............ 37
2.2.10. Statistical Analysis ........................................................................... 38

2.3. Results ..................................................................................................... 40
2.3.1. 2D Skeletal Myotube Morphology ...................................................... 40
2.3.2. Myotube Contractile Stress Generation on Muscular Thin Films ....... 45
2.3.3. Transcriptomic analysis of engineered chick muscle tissues on
gelatin hydrogels ................................................................................ 47
2.3.4. 3D Skeletal Myotube Morphology ...................................................... 52
2.3.5. Myobundle Contractile Stress Generation .......................................... 54
2.4. Discussion ................................................................................................ 55
Chapter 3 Coculture of Skeletal Myotubes and Motor Neurons to Engineer
Functional Neuromuscular Junctions ........................................................... 63
3.1. Introduction .............................................................................................. 63
3.2. Materials and Methods ............................................................................. 64
3.2.1. 2D Coculture of Skeletal Muscle and Healthy hiPSC-derived
Motor Neurons ................................................................................... 64
3.2.2. 2D NMJ Structural Characterization ................................................... 65
3.2.3. 2D NMJ Electrophysiology ................................................................. 66
3.2.4. Seeding and Culturing Innervated 3D Myobundles ............................ 66
3.2.5. 3D NMJ Structural Characterization ................................................... 67
3.2.6. Innervated Myobundle Contractile Force Generation ......................... 67
3.2.7. Statistical Analysis ............................................................................. 67
3.3. Results ..................................................................................................... 68
3.3.1. Synaptic Structure of Cocultures on Gelatin Hydrogels ..................... 68
3.3.2. Synaptic Activity of Cocultures on Gelatin Hydrogels......................... 71
3.3.3. Morphology of Innervated Myobundles .............................................. 72
3.3.4. Contractile Activity of Innervated Myobundles ................................... 75
3.4. Discussion ................................................................................................ 76
Chapter 4 Modeling Muscular Dystrophy Using Human-Derived Cells on
Micromolded Gelatin Hydrogels ................................................................... 81
4.1. Introduction .............................................................................................. 81
4.2. Materials and Methods ............................................................................. 83
4.2.1. Inducible Directly Reprogrammable Myotube (iDRM) Derivation
and Dermal Fibroblast Characteristics ............................................... 83
4.2.2. iDRM Culture ..................................................................................... 84
4.2.3. Antisense Oligonucleotide (AO) Transfection .................................... 85
4.2.4. RNA Isolation and PCR ...................................................................... 86
4.2.5. Immunofluorescence Staining and Microscopy .................................. 87
4.2.6. Statistical Analysis ............................................................................. 88

4.3. Results ..................................................................................................... 89
4.3.1. Engineering DMD and LGMD2A/R1 iDRM myotubes on
micromolded gelatin hydrogels .......................................................... 89
4.3.2. Morphology of muscle fibers in healthy and DMD patients ................ 93
4.3.3. Evaluation of dystrophin rescue in DMD iDRM myotubes treated
with exon skipping AOs ...................................................................... 96
4.4. Discussion .............................................................................................. 100
Chapter 5 Concluding Remarks and Future Work ....................................................... 106
5.1. Modeling amyotrophic lateral sclerosis with a
neuromuscular microphysiological system ............................................. 106
5.2. Skeletal muscle tissue engineering for consumable products ................ 107
5.3. Visualization and analysis of 3D myobundle architecture in situ ............ 108
5.4. Limitations to our approaches and other future improvements ............... 108
5.5. Funding Sources .................................................................................... 110
References .................................................................................................................. 112

viii

List of Tables
Table 2-1: Components of growth and differentiation media for skeletal muscle
tissues ......................................................................................................... 29

ix

List of Figures
Figure 1-1: Internal structure of skeletal muscle tissue. Skeletal muscle is made up for
muscle fibers surrounded by connective tissue and blood vessels supplying oxygen and
nutrients. Each muscle fiber is made up of bundles of myofibers, or myotubes, and each
myotube contains the contractile protein units called sarcomeres. Sarcomeres consist of
overlapping thick myosin filaments and thin actin filaments. Shortening of the sarcomere
in bulk tissue during contraction allows for bodily movement. Adapted from [6]. ............. 2
Figure 1-2: Calcium ions mediate skeletal muscle contraction upon release into the
cytoplasm. When a muscle action potential is propagated through T-tubules, DHP-
ryanodine complex receptors release calcium ions from the sarcoplasmic reticulum into
the myotube cytoplasm. These ions bind to troponin, a protein complex that must be
bound to remove actin filament-blocking protein tropomyosin. This allows for myosin
crossbridges to bind to actin, generating contractile force during the power stroke. ATP is
used to unbind and reenergize the myosin crossbridge. Adapted from [6]. ..................... 4
Figure 1-3: The structure and functional mechanism of the neuromuscular junction
(NMJ). A) Scanning electron micrograph showing motor axon terminals embedded in the
sarcolemma of the myofiber. B) Internal structure of the NMJ. C) Summary of signal
transmission at the NMJ. An action potential arrives at the axon terminal, inducing
extracellular calcium ions to enter via voltage-gated channels. This induces acetylcholine
to be released into the synaptic cleft, where they bind to acetylcholine receptors on the
myotube sarcolemma, resulting in sodium ion entry. The change in ion concentration
causes a depolarization across the membrane that is propagated down the muscle fiber
as a muscle action potential. The depolarization ends as ion channels close due to
acetylcholine breakdown by acetylcholinesterase, and choline is reabsorbed into the
presynaptic membrane. The membrane resets to resting potential and ends the muscle
contraction stimulus. Adapted from [6]. ........................................................................... 7
Figure 1-4: Distribution of the most common genetic mutations responsible for sporadic
and familial forms of ALS. C9ORF72 repeat expansions are detected in the majority of
familial ALS and a substantial portion of sporadic ALS. Adapted from [22]. .................. 14
Figure 1-5: In vivo NMJs are characterized by distinct invaginations of acetylcholine
receptors localized with motor neuron axon terminals that are poorly recapitulated in
x

various approaches taken by current in vitro models. A) A mouse NMJ exhibiting distinct,
pretzel-like co-localization of presynapse and acetylcholine receptor marker. Adapted
from [67]. B) Poor NMJ morphology in ESC-derived motor neuron and C2C12 myotube
coculture. Adapted from [61]. C) Engineered 3D myobundles result in aligned tissues but
show poor acetylcholine receptor aggregation even when innervated by stem-cell derived
motor neurons, limiting NMJ maturation. Adapted from [54]. D) Patterned human
myotubes innervated by hiPSC-derived motor neuron axons traversing through
microchannels are more anatomically relevant, but still lack proper NMJ morphology.
Adapted from [64] under CC-BY. .................................................................................. 22
Figure 2-1: Fabrication of micromolded gelatin hydrogel muscular thin films (MTFs).
Laser-cut polystyrene coverslips are masked with tape, which is then engraved. Tape A
is removed, and plasma treatment applied, increasing adhesiveness in the unmasked
region. Tape B and C are then removed and 10% gelatin, 4% transglutaminase solution
is pipetted onto the coverslip. A PDMS stamp is placed on the solution, which is cured
overnight at room temperature. The hydrogel is then rehydrated and the PDMS stamp
and tape D are removed. MTF cantilevers are then laser engraved into the hydrogel. [81]
...................................................................................................................................... 28
Figure 2-2: Summary of 3D myobundle fabrication procedure. (a) Assembly of PDMS
device parts through plama bonding. (b) Overview of seeding and culturing myoblasts
and motor neurons in a fibrin hydrogel. ......................................................................... 35
Figure 2-3: Standard curve for displacement of PDMS rods using known weights. ..... 38
Figure 2-4: Structure of engineered muscle tissues over three weeks of differentiation.
(a) Representative images of C2C12, chick, and human muscle tissues after one and
three weeks in culture on micromolded gelatin hydrogels. α-actinin (red), DAPI (blue).
Scale bar: 100 μm. (b) Number of nuclei normalized to field of view area and (c) myogenic
index for each myoblast cell source at weekly timepoints. ............................................ 41
Figure 2-5: Myotube and sarcomeric organization over weeks of differentiation. A)
Representative images of individual myotubes and sarcomeres from C2C12, chSKM and
hSKM tissues. α-actinin (red), DAPI (blue). Scale bar = 50 μm for tissues, 10 μm for
myotubes. B) Average myotube widths, C) sarcomeric indexes, and D) sarcomeric
lengths over differentiation time. ................................................................................... 44
xi

Figure 2-6: Muscular thin films (MTF) were analyzed with image processing MATLAB
software based on deflection of cantilevers. A) Representative images of a single MTF at
basal, twitch, and tetanus contraction. B) Based on Stoney’s equation, a trace of
contractile stress over time of a single cantilever can be plotted. Twitch and tetanus stress
were stimulated at 2 and 20 Hz, respectively. C) Schematic of MTFs laser-engraved into
micromolded gelatin hydrogels. D) 3D perspective of cantilevers at different stages of
contraction. .................................................................................................................... 45
Figure 2-7: Stress generation by myotube MTFs. A) Basal stresses generated by
unstimulated myotubes on MTFs. B) Twitch stresses generated by stimulating MTFs at 2
Hz. C) Tetanus stresses generated by stimulating MTFs at 20 Hz. D) Average tetanus-
to-twitch ratios of MTFs. ................................................................................................ 46
Figure 2-8: Global transcriptome of chick muscle tissues cultured for one week and three
weeks. (a) Hierarchical heat map clustered by features for week 1 (blue) and week 3 (red)
samples. (b) PCA plot of week 1 (blue) and week 3 (red) chick tissues. Individual tissues
are labeled as “week in culture-sample number” in each panel [80, 81]........................ 47
Figure 2-9: Differentially expressed genes in chick muscle tissues cultured for one week
(W1) and three weeks (W3). (a) Volcano plot of all genes up- and downregulated or not
significantly changed in tissues cultured for three weeks compared to one week.
Differentially expressed genes (DEG) were defined as those with p < 0.05 and a
magnitude fold change of greater than 1.5. Normalized gene expression of (b) regulators
of myogenic proliferation and metabolism, (c) contractile proteins, (d) neurotrophic
factors, and (e) signaling molecules in acetylcholine receptor clustering. *p < 0.05;
**p < 0.01; ***p <0.001; ****p < 0.0001 [80, 81]. ............................................................. 51
Figure 2-10: IPA-generated predictive mechanistic networks. In chick muscle tissues
cultured for three weeks compared to one week, pathways for (a) organization of
sarcomeres and (b) innervation of muscle are predicted to be activated. (c) The pathway
for formation of neuromuscular junctions is expected to be inhibited [80, 81]. .............. 52
Figure 2-11: Hematoxylin and eosin stains of 3D myobundles after one and two weeks
of differentiation. Scale bar = 200 μm. ........................................................................... 53
Figure 2-12: 3D myobundle morphology after 2 weeks of differentiation. (a)
Representative myobundle cryosections. α-actinin (red), synapsin (green), DAPI (blue).
xii

Scale bar = 200 μm (left), 50 μm for zoomed in images (right). (b) Number of nuclei
normalized to field of view area and (c) myogenic index at weekly timepoints. ............. 54
Figure 2-13: Contractile force generation by 3D fibrin chick myobundles. (a) Electrically
stimulated twitch and tetanus forces, (b) tetanus-to-twitch ratio, and (c) rise and fall τ after
one and two weeks of differentiation. *p < 0.05; **p < 0.01 ............................................ 55
Figure 3-1: Structure of NMJs formed between engineered muscle tissues and hiPSC-
derived motor neurons. hiPSC-derived motor neurons cocultured with (a) C2C12, (b)
human, and (c) chick muscle tissues for one week and (d) chick muscle tissue for three
weeks. α-actinin (red), DAPI (blue), synapsin-1 (green), and bungarotoxin (white). Scale
bar, 50 μm. Percentage of (e) synapsin and (f) bungarotoxin area per field of view. (g)
Size of individual bungarotoxin clusters. (h) Percentage of bungarotoxin clusters co-
localized with synapsin. *p < 0.05; **p < 0.01; ***p <0.001; ****p < 0.0001. .................... 70
Figure 3-2: Synaptic activity of NMJs formed between engineered muscle tissues and
hiPSC-derived motor neurons. (a) Recordings from representative myotubes after one
and three weeks of coculture with hiPSC-derived motor neurons. (b) mEPSP frequency,
(c) amplitude, (d) rise time constant, and (e) decay time constant as a function of coculture
time. (f) Membrane potential of myotubes used for recordings. * denotes p < 0.05. ...... 71
Figure 3-3: Hematoxylin and eosin stains of innervated 3D myobundles after one and
two weeks of differentiation. Scale bar = 200 μm. *p < 0.05 .......................................... 72
Figure 3-4: 3D myobundle morphology after 2 weeks of coculture with motor neurons.
(a) Representative myobundle cryosections. α-actinin (red), synapsin (green), DAPI
(blue). Scale bar = 200 μm (left), 50 μm for zoomed in images (right). (b) Number of nuclei
normalized to field of view area, (c) myogenic index at weekly timepoints, and (d) synapsin
area coverage per field of view. *p < 0.05; **p < 0.01 .................................................... 74
Figure 3-5: Myobundle contractile force characterization over 2 weeks of monoculture or
coculture. Electrically stimulated (a) twitch and (b) tetanus forces, (c) tetanus-to-twitch
ratio, and (d) rise and (e) fall τ. *p < 0.05; **p < 0.01, ***p < 0.01 ................................... 75
Figure 4-1: Overview of the experimental design and culture timeline. (a) A skin punch
of 3mm diameter is performed on the forearm and the tissue dissociated to obtain
fibroblasts. Following infection with lentiviruses encoding hTERT and MyoD, cells
(renamed induced directly reprogramed myotubes – iDRM) can form myotubes upon
xiii

induction of MyoD and fusion inducing media. Polystyrene coverslips are plasma-treated
to enhance gelatin hydrogel adhesion prior to molding of the surface using soft-
lithography. Hydrogels are rehydrated and stored in PBS prior to cell seeding. (b) A typical
timeframe for the generation of myotubes from iDRM with or without AO treatment.
Analysis for experiments was performed after one (Day 9) or two (Day 16) weeks of
differentiation. ................................................................................................................ 86
Figure 4-2: iDRM derived from LGMD2A/R1 or DMD dermal fibroblasts show distinct
maturation on micromolded gelatin hydrogel coverslips. α-actinin staining was performed
to analyze the maturation of myotubes from a healthy donor (1001) or patients presenting
with LGMD2A/R1 (1077 - c.550delA, c.1342C>G and 1081 delta 17-24) or DMD (1015-
del45, 1023 –del3-23 and 1003 – del46-51). For each cell line, cells were differentiated
for one or two weeks. Images are shown at 20x magnification on the left, where the scale
bar represents 100 μm, and at 60x magnification in the upper right panel where the scale
bar represents 10 μm. Representative myotubes shown in the lower right panel enable
visualization of sarcomeres where the scale bar represents 10μm. .............................. 91
Figure 4-3: Morphological quantification of DMD or LGMD2A/R1 or iDRM reveal distinct
changes in mutation-specific muscle morphology. Different parameters were measured
for each cell line using α-actinin staining: (a) α-actinin area coverage per field of view; (b)
myogenic index representing the proportion of nuclei in myotubes (containing at least 3
nuclei); (c) minor axis length as a proxy for myotube width; (d) myotube alignment. Bars
represent SEM. *p < 0.05; p values reflect a student’s t test. Each experiment (n = 3) is
represented by different color dots. ............................................................................... 93
Figure 4-4: DMD patients’ biopsies show abnormal α-actinin striations. Skeletal muscle
biopsies from a healthy donor and patients presenting Duchenne Muscular Dystrophy
(1023 del3_23; 1015-del45). (a) Co-staining for α-actinin and laminin are shown. (b) The
total area of the fibers (µm
2
) was measured on the biopsies (number of fibers analyzed
between 250-658) and sorted by 500 μm range. (c) The mean average of the fibers is
also represented. Bars represent SEM. ****p < 0.0005; p values reflect a student’s t-test.
Immunostains were taken at 20x, where the scale bar represents 100 μm. .................. 94
Figure 4-5: DMD patients’ biopsies show dystrophic features. Muscle biopsy sections
from patients presenting Duchenne Muscular Dystrophy (1023 del3_23; 1015-del45)
xiv

assessed for Dystrophin protein using dys2 antibody targeting the C-terminus and laminin
using L9393 antibody. Hematoxylin/eosin staining for each biopsy is showed to visualize
their morphological features and proceed to morphological assessment. Scale bar
represents 100 μm. ....................................................................................................... 95
Figure 4-6: Dystrophin rescue by DMD exon skipping in DMD 1015 and DMD 1003
iDRMs. DMD 1015 (delta 45) and DMD 1003 (delta 46–51) iDRMs were cultured for 1 or
2 weeks before the addition of antisense oligonucleotides and, 2 days later,
immunostaining or RNA extraction (on pooled triplicate) were performed. (a) Dystrophin
expression in healthy and DMD iDRM tissues was visualized using Dys2 antibody (c-ter).
Images are shown at 20x magnification in the top panel, where the scale bar represents
100 μm, and at 60x magnification in the lower panel to enable visualization of sarcomeres,
where the scale bar represents 30 μm. After RT-PCR, samples were run on chips to
analyze exon skipping. The percentage of skipped mRNA over the total (skipped plus
unskipped mRNA) is indicated for (b) DMD 1015 and (c) DMD 1003. .......................... 97
Figure 4-7: Effects of dystrophin rescue by exon skipping on α-actinin expression and
organization. DMD 1015 (delta 45) or DMD 1003 (delta 46–51). (a) DMD 1015 or (b) DMD
1003 cells were cultured for 1 or 2 weeks before the addition of antisense oligonucleotides
and α-actinin staining was performed to analyze the maturation of myotubes. Images are
shown at 20x magnification in the left panel, where the scale bar represents 100 μm, and
at 60x magnification in the upper right panel, where the scale bar represents 50 μm.
Representative myotubes shown in the lower right panel enable visualization of
sarcomeres, where the scale bar represents 10 μm. .................................................... 99
Figure 4-8: Morphological quantification of AO-treated DMD 1015 and DMD 1003 iDRMs.
Different parameters were measured for each cell line using α-actinin staining: (a) α-
actinin area coverage per field of view; (b) myogenic index representing the proportion of
nuclei in myotubes (containing at least three nuclei); (c) minor axis length as a proxy for
myotube width; and (d) myotube alignment. Bars represent SEM. Each experiment (n =
3) is represented by different color dots. ..................................................................... 100

xv

Abstract
Neuromuscular diseases involve degeneration of the motor unit, which includes
the motor neuron and the skeletal muscle fibers they innervate at the neuromuscular
junction (NMJ). These diseases vary widely in pathogenesis, but generally start with
ambulatory deficits that increase in severity. Compounded with ensuing cardiac and
respiratory complications, these diseases result in low quality of life and, often, ultimately
death. Physiologically relevant in vitro models of skeletal muscle tissue are critically
needed to probe the mechanisms of disease progression on a cellular and molecular
level. However, traditional attempts to culture skeletal muscle in vitro have resulted in
unaligned, immature myotubes with limited functionality, preventing deeper investigation
of key processes such as sarcomere development or motor neuron integration and the
formation of neuromuscular junctions. Failure to reach a basic level of tissue maturity
limits translatable conclusions when using these models to study neuromyopathies or
muscular dystrophies.
Recently, advances in stem cell biology and micro- or nanofabrication techniques
have been applied to better recapitulate the structure and function of a basic unit of a
tissue or organ system. These platforms, referred to as microphysiological systems
(MPS), act as an intermediate platform between in vivo and in vitro systems, allowing for
medium throughput, relevant tissue structure, and quantifiable functional outputs. By
balancing biological complexity of these engineered systems to recapitulate features of
interest with simplicity and speed of fabrication, it is possible to maximize the efficiency
of collecting relevant results for specific research questions. Our objective was to apply
tissue engineering techniques to culture 2D and 3D skeletal muscle MPS towards
xvi

implementing more translatable disease models, yielding data with increased clinical
relevance.
First, we engineered and characterized functional skeletal muscle tissues in 2D on
micromolded gelatin hydrogels and in 3D suspended in fibrin hydrogels. Specifically, we
assessed muscle tissue density, fusion, and contractile force generation from several
common commercial sources. Second, we developed and implemented robust coculture
protocols for skeletal myotubes and human-induced pluripotent stem cell (hiPSC)-derived
motor neurons. These engineered tissues exhibited distinct, organized co-localization of
presynapse and acetylcholine receptors expected in developing NMJs and manifest
spontaneous synaptic electrophysiological activities. Lastly, as proof-of-concept of
translatability to clinical human relevance, we demonstrated the utility of our 2D gelatin
hydrogel system to recapitulate human muscular dystrophy mutation-specific phenotypes
and their responses to potential therapeutics.
Together, these developed technologies can be extended in the future to
supplement 1) currently used disease models to better understand skeletal muscle or
NMJ degeneration and 2) drug screening platforms for identifying potential therapeutics
of neuromuscular diseases. Ideally, these tools will expediate the clogged therapeutic
development pipeline by producing data with higher physiological relevance, potential
subject specificity, and throughput. Moving forward, integration of neuromuscular MPS
with other relevant cell types, gene editing technologies, or optogenetics may further
expand throughput and common use of tissue-engineered constructs in the research
space. Ultimately, these outcomes should prevent or relieve the negative impacts of
neuromuscular disease, especially for an increasingly prone aging population.
1

Chapter 1 Introduction
Neuromuscular diseases are characterized by pathological impairment of the
motor system and progressive muscular atrophy, and collectively affect 160 per 100,000
people worldwide [1]. These disorders can be inherited or acquired and are variable in
anatomical origin. For example, Duchenne muscular dystrophy is caused by a defect in
the cytoskeleton of skeletal muscle, whereas Charcot-Marie-Tooth disease results from
defects in the myelin sheath of peripheral nerve cells, and amyotrophic lateral sclerosis
(ALS) is characterized by the degeneration and dysfunction of neuromuscular junctions
(NMJs) [1]. NMJs are the chemical synapses between motor neurons and skeletal muscle
fibers. When a motor neuron is activated by an action potential, it releases acetylcholine
(Ach) at the NMJ, which diffuses across the synaptic cleft, binds to acetylcholine receptors
(AchR) on the muscle fiber, and initiates muscle shortening. Diseases that affect the
skeletal muscle fiber, motor neuron, or NMJ are typically diagnosed in patients using
electromyography and nerve conduction studies. Although these approaches can
determine the affected nerves or muscles, they provide limited insight into neuromuscular
physiology and pathology, which is needed to develop more effective treatments [2]. In
vitro models of skeletal muscle are valuable research tools for interrogating the
physiology and pathophysiology of the neuromuscular system with modularity, ease of
access, and medium- to high-throughput efficiency. However, current models suffer from
several drawbacks, hindering the acquisition of clinically relevant mechanistic and
pharmacological data. In this chapter, we will review the structure and function of skeletal
muscle fibers and their integration with motor neurons. We also review current models for
(neuro)myopathies and discuss both their contributions to understanding disease and
2

their limitations. Lastly, we identify the specific challenges that must be addressed to
engineer functionally and physiologically relevant neuromuscular tissues in vitro for
disease modeling and drug screening applications.
1.1. Neuromuscular Tissue Development and Physiology
Neuromuscular junctions (NMJs) refer to the chemical synapses that relay action
potentials from the motor neurons situated in the spinal column to skeletal muscle fibers,
resulting in the muscle contractions that allow for voluntary bodily movement. In humans,
the neurotransmitter acetylcholine (Ach) diffuses across the NMJ and binds to
acetylcholine receptors (AchR), initiating a calcium-mediated shortening of the muscle
that is controlled by the interactions of sarcomeric proteins [3]. Here, we describe the
physiology of skeletal muscle tissue, the motor neuron, and the mechanism of voluntary
contraction via electro-chemical signaling propagating through the motor unit.
Figure 1-1: Internal structure of skeletal muscle
tissue. Skeletal muscle is made up for muscle
fibers surrounded by connective tissue and
blood vessels supplying oxygen and nutrients.
Each muscle fiber is made up of bundles of
myofibers, or myotubes, and each myotube
contains the contractile protein units called
sarcomeres. Sarcomeres consist of overlapping
thick myosin filaments and thin actin filaments.
Shortening of the sarcomere in bulk tissue
during contraction allows for bodily movement.
Adapted from [6].

3

Understanding the structure-function relationship of the motor unit is key to recapitulating
and interrogating muscle models with physiological relevance, which will in turn yield
more translatable insights into disease progression and therapeutic response
mechanisms.

1.1.1. Skeletal Muscle Tissue and Mechanisms of Contraction
Skeletal muscle tissue is responsible for the force generation the human body
relies on for controlled voluntary movement. The tissue consists of bundles of aligned,
multinucleated muscle fibers surrounded by connective tissues. Muscle fibers, or
myotubes, are formed during embryonic development by the fusion of mononucleated
myoblast precursors. Postnatally, skeletal muscle differentiation is complete and
myotubes do not proliferate, but quiescent satellite cells expressing transcription factor
Pax7 remain adjacent to the basal lamina, and can be induced to proliferate and
differentiate into new myoblasts after injury [4]. In addition to hypertrophy of uninjured
myotubes, differentiation of these myoblasts can renew slight to intermediate damage of
skeletal muscle tissue.
Skeletal muscle exhibits striations made up of myofibrils, arrangements of myosin
thick filaments and actin thin filaments in an overlapping structure (Figure 1-1). The
overlapping regions of thick and thin filaments are connected by cross-bridge structures
that extend from the myosin protein. During contraction, these cross-bridges pull on actin
thin filaments to create shortening of the sarcomere, allowing for muscle movement. This
process is regulated by the presence of calcium in the myotube cytoplasm. At rest, the
concentration of ionized calcium is around 10
-7
M. However, calcium is released in large
4

concentrations into the cytoplasm by the sarcoplasmic reticulum upon an action potential
stimulus that is propagated transversely through a muscle bundle by voltage-sensitive
calcium channels in transverse tubules (Figure 1-2). These cytosolic calcium ions then
bind to troponin complexes on actin thin filaments, removing sterically hindering
tropomyosin molecules and exposing actin sites to myosin cross-bridges. The release of
ADP and a phosphate group from the energized myosin heads results in shortening of
the sarcomeric structure, a process known as the power stroke. ATP rebinds to myosin
and is hydrolyzed to reenergize the myosin crossbridge, and the process either repeats
for further sarcomeric shortening or the filament structure resets to its starting length.

Figure 1-2: Calcium ions mediate skeletal
muscle contraction upon release into the
cytoplasm. When a muscle action
potential is propagated through T-tubules,
DHP-ryanodine complex receptors
release calcium ions from the
sarcoplasmic reticulum into the myotube
cytoplasm. These ions bind to troponin, a
protein complex that must be bound to
remove actin filament-blocking protein
tropomyosin. This allows for myosin
crossbridges to bind to actin, generating
contractile force during the power stroke.
ATP is used to unbind and reenergize the
myosin crossbridge. Adapted from [6].

5

1.1.2. Motor Neurons and the Neuromuscular Junction
During human embryonic development, myotubes express diffuse clusters of
AchR at the myotube membrane, which bind the excitatory neurotransmitter Ach. Efferent
motor neurons develop from progenitor cells in the neural tube and project their axons
towards the maturing myotubes, with axon migration positively reinforced towards areas
of higher AchR concentration [5]. At the same time, Schwann cells developing in the
neural crest lay the myelin sheath that insulates the length of the axons. Concurrently,
vesicles of Ach released by the synaptic terminals of motor neurons induce not only
nonspecific muscle contraction, but also disrupt unenervated AchR clusters in the
myotube sarcolemma, causing their dispersal. However, innervated clusters interact with
the proteoglycan agrin, secreted in the synaptic basal lamina, and promotes further
aggregation of AchR and motor neuron innervation of myotubes through activation of
Lrp4-MuSK signaling [5]. In humans, modification and development of the motor neuron-
myotube motor end plate interface continues past the twentieth weeks of fetal life [6]. The
concentration of receptors increases to roughly 10,000 receptors per square micron, and
distinct pretzel-like invaginations in the myotube sarcolemma are formed at axon terminal
innervation point. This increase in surface area, and thus number of receptors, improves
reliability and consistency of signal transmission. Mature myotubes are innervated by a
single motor neuron, although one motor neuron may branch and innervate multiple
myotubes, which contract together as a single motor unit. Each of these synapses is
referred to as an NMJ.
The action potential that stimulates skeletal muscle contraction described in the
previous subsection is first propagated through the motor neurons of the somatic nervous
6

system, the voluntary subset of the peripheral nervous system. Action potential signals
from the central nervous system are relayed at high velocities down myelinated axons to
the NMJ, where neurotransmitter vesicles are released into the synaptic cleft from the
axon terminal. In humans and other vertebrates, Ach is the main neurotransmitter
released in the somatic nervous system, and all muscle responses are excitatory. That
is, all NMJ activity leads to contraction of the muscle fiber, and relaxation of myotubes is
only possible with inhibition of motor neuron activity. The mechanism of signal
transmission through the NMJ starts with the depolarization of the plasma membrane at
the axon terminal when an action potential arrives, activating voltage-gated calcium
channels that result in vesicles of Ach being released into synaptic cleft (Figure 1-3). The
binding of Ach, a diffusion-limited process, allows sodium ions to enter and potassium
ions to exit the myotube via channels. The myotube’s total depolarization at this location
can be quantified as an end-plate potential, which can be further quantized as multiple
miniature end-plate potentials resulting from the release of multiple Ach vesicles. After
the membrane is depolarized at the endplate, the muscle fiber continues to propagate the
action potential down the fiber, at the same time activating the calcium-mediated
contraction discussed previously. Lastly, free Ach released into the NMJ is broken down
by the enzyme acetylcholinesterase into choline. Bound Ach reaches equilibrium with free
Ach, eventually causing ion channels in the endplate to close. Thus, the membrane resets
to resting potential and ends the muscle contraction stimulus. Depending on the
frequency of actional potentials transmitted by the nervous system through the NMJ, both
singular (twitch) and sustained (tetanus) contractions can be achieved. At the tissue and
7

organ scale, these processes allow for voluntary movement by force exertion of several
skeletal myotubes on tendons and bones.

One difficulty of studying neuromuscular systems, their pathologies, and potential
routes for therapy is the complexity of the functional system. For example, while the NMJ
is made up of the motor neuron and myotube, there are a variety of supporting cells not
described in detail in this work, such as astrocytes and Schwann cells, that play a role in
NMJ development and maintenance in vivo. Choosing the appropriate subsets of the
neuromuscular system to study may be essential for specific experiments, especially for
disease modeling. Additionally, the size of the NMJ also poses difficulties in studying
Figure 1-3: The structure and functional
mechanism of the neuromuscular junction
(NMJ). A) Scanning electron micrograph
showing motor axon terminals embedded in
the sarcolemma of the myofiber. B) Internal
structure of the NMJ. C) Summary of signal
transmission at the NMJ. An action potential
arrives at the axon terminal, inducing
extracellular calcium ions to enter via voltage-
gated channels. This induces acetylcholine to
be released into the synaptic cleft, where they
bind to acetylcholine receptors on the myotube
sarcolemma, resulting in sodium ion entry. The
change in ion concentration causes a
depolarization across the membrane that is
propagated down the muscle fiber as a muscle
action potential. The depolarization ends as ion
channels close due to acetylcholine
breakdown by acetylcholinesterase, and
choline is reabsorbed into the presynaptic
membrane. The membrane resets to resting
potential and ends the muscle contraction
stimulus. Adapted from [6].

8

structure and functionality. Vertebrate NMJs vary somewhat in size and extent of folding,
and invertebrate NMJs may vary even more widely, with human AchR clusters exhibiting
areas in the few hundreds of square microns and a synaptic distance of tens of
nanometers [7]. Thus, species-specific differences in NMJ structure/function must also
be considered. As will be discussed in the next section, the variety of etiologies in
neuromuscular pathologies also means the ideal platform to study them should also be
modular and scalable. Overall, the principles of physiological NMJ structure and function
outlined briefly in this subsection act as the foundation for investigations into
understanding neuromuscular pathology.
1.2. Pathological Degeneration of Motor Unit Components
Many neuromuscular pathologies, external drugs or toxins [3], and even the natural
effects of aging or low physical activity [8] can disrupt neuromuscular signaling. While the
root mechanisms for neuromuscular degeneration are widespread, in general, the
breakdown of the motor unit results in muscle atrophy, ultimately leading to fatal
ambulatory, respiratory, and metabolic deficits. To date, widespread treatments of
neuromuscular diseases are mostly palliative in nature, leaving an unmet need of true
therapeutics. This section will review both genetic and environmental etiologies of
neuromuscular diseases, along with some bottlenecks to diagnosing and developing
therapies for them. Finally, we will address one specific neuromuscular disease,
amyotrophic lateral sclerosis (ALS), and challenges to its study and therapy in greater
detail.
9

1.2.1. Genetically Inherited Neuromyopathies – Spinal Muscular Atrophy, Duchenne
Muscular Dystrophy, and Charcot-Marie-Tooth Disease
Several neuromyopathies can be mostly attributed to hereditary mutations in a
single or a group of genes that affect the viability or utility of specific protein. For example,
spinal muscular atrophy (SMA) is an autosomal recessive neuromuscular disease caused
by the individual’s failure to produce functional survival of motor neuron (SMN) protein
due to mutations in the SMN1 gene on chromosome 5q13 [9]. A shortage of functional
SMN introduces deficits in translation regulation and neuronal trafficking, leading to motor
neuron death and NMJ degeneration. In the past, diagnosis was dependent on
electromyography and creatine kinase tests to discriminate from other pathologies after
disease onset, but with the advent of prenatal genetic screens, it has become possible to
detect SMA and other genetically transmitted neuromuscular disorders prior to birth.
Although other factors play into disease phenotype, it is even possible to predict the
severity of SMA based on the number of SMN2 gene copies that can also generate SMN
at lower concentrations [10]. However, even considering advances in diagnosis based on
genetic sequencing, there exist limited options to treat, rather than manage, the disease.
Despite a 2019 landmark FDA approval and Novartis commercialization of Zolgensma as
a one-time intravenous gene therapy treatment for children under two years of age with
SMA type 1, questions remain regarding its long-term efficacy [11]. Furthermore, while
infantile SMA is the most common with a prevalence of 60% of all cases, the drug has
not been approved for SMA types characterized by later onset and lesser severity. To
date, it also happens to be the world’s most expensive drug [11]. While the development
of Zolgensma gives promise as a roadmap for developing gene therapies of other
10

inherited neuromyopathies, there are clearly still many concerns, particularly regarding
the inefficiency of the therapeutic development pipeline.
Other genetic disorders that affect components of the motor unit may originate
from skeletal muscle fiber deficits. Duchenne muscular dystrophy is caused by various
loss-of-function mutations in the DMD gene on the X-chromosome coding for the
dystrophin protein [12]. Dystrophin plays an important role in linking the intracellular
muscle fiber cytoskeleton to the extracellular matrix, and deficits in its expression are
correlated with muscle degeneration and fibrosis [13] and irregularities in calcium
homeostasis [14]. These dysfunctions lead to a progressive loss of ambulation in young
males onsetting around 3 years of age and is ultimately lethal. Interestingly, milder and
delayed forms of this muscular dystrophy can arise from mutations that maintain the DMD
gene reading frame, and thus mutation- and patient-specific phenotypes make modeling
Duchenne particular difficult with animal models, such as the mdx mouse, which exhibit
minimal clinical symptoms not representative of more severe Duchenne cases [15]. Such
hurdles in researching the myopathy has prevented development of a curative treatment;
steroids that delay disease progression, physiotherapy, and other palliative care remains
the only option for patients [12]. While gene therapies such as genome editing and exon
skipping strategies that restore the DMD reading frame are promising routes of therapy
[16], they require further study using patient-specific models to fully characterize their
potential utility.
Other inherited neuromuscular diseases, such as Charcot-Marie-Tooth (CMT)
diseases, still lack clinical data supporting any effective medications, and treatment is
almost wholly dependent on physical therapy and pain management [17]. CMT diseases
11

exhibit hallmarks of demyelinating and/or axon degeneration due to mitochondrial
clustering, leading to sensory and motor neuron dysfunction. One difficulty of studying
CMT diseases, the most commonly inherited neurological disorders, compared to SMA is
the genetic variety of the disease, with 870 mutations on over 80 genes identified so far.
This partially explains the wide range of age onset and disease symptoms, which often
involve involuntary contraction of the distal limbs and loss of sensation. Much of the lack
of therapeutic development results from this patient-specific genetic variation, despite
similarities in disease phenotype, making both recruitment for clinical trials and transgenic
animal models costly and low-throughput [17]. Thus, in addition to specific genetic
mutation disease models, there may also be a need for even more nuanced patient-
specific models that can foster the development of several treatment strategies, allowing
for more time and cost-efficient studies of hereditary neuromuscular diseases.
1.2.2. Sporadically Acquired Neuromyopathies – Myasthenia Gravis
Neuromyopathies can also exhibit less inheritability and occur sporadically (i.e.,
without family history), presenting unique sets of challenges in diagnosis and therapeutic
development. In the case of myasthenia gravis, an individual develops an autoimmune
disorder causing the loss of neuronal signal transmission efficiency as IgG1 and IgG3
autoantibodies selectively destroy acetylcholine receptors, culminating in muscle atrophy
due to decreased stimulation [18]. Because the disease occurs spontaneously, screening
is rarely an option, and the disease usually remains undetected until ocular muscles are
significantly affected. Thus, understanding how the disease develops mechanistically
may be key to not only disease therapy, but also prevention. One challenge in studying
the disease, though, has been the difficulty of studying the mechanism of autoimmunity
12

development. Because of the sporadic nature of the disease, it has been problematic to
narrow down associations between widely different predisposed patient groups, which
include younger women, older men, and those exhibiting thymomas. Genome-wide
association studies have begun to identify associations with immune system regulators
and environmental factors, but studies performed with the mostly commonly used
experimental autoimmune myasthenia gravis (EAMG) mouse models fail to recapitulate
the autoimmune severity in human subjects [18]. With these spontaneously developed
neuromuscular diseases, in vitro models may be useful in more closely investigating and
recapitulating the disease milieu expected in different myasthenia gravis subtypes, with
greater control and cost efficiency than the currently used mouse model. However, due
to immaturity of current neuromuscular models, the electrophysiological characterization
of the NMJ necessary to investigate acetylcholine receptor degradation in a functional
manner has not been possible – this approach has not been successfully documented
[19]. These technological and biological limitations hinder the development of
therapeutics for sporadic neuromuscular diseases.
1.2.3. Neuromuscular Diseases with Mixed Etiology - Amyotrophic Lateral Sclerosis
The aforementioned neuromyopathies and the current approaches to their
diagnosis and treatment showcase the unmet needs that must be fulfilled to more
efficiently develop therapies. Amyotrophic lateral sclerosis (ALS) is the most common
motor neuron disease in adults that is characterized by progressive muscle weakness
leading to respiratory failure, and is particularly difficult to study due to its particularly
varied epidemiology and etiology [20]. Thus, as a case study, we will discuss its
challenges to model in the laboratory setting in further detail. ALS exhibits both familial
13

(10%) and sporadic (90%) forms and shares characteristics and major challenges to
treatment of both hereditary and spontaneously acquired diseases [21]. Firstly, ALS
pathogenesis is complicated, with over 50 genes identified that directly cause either
familial or sporadic disease pathogenesis or alter related function, including vesicle
trafficking, axonal structure, and cytoskeletal stability [21-23]. Additionally, environmental
factors (e.g., pollutants such as pesticides or flame retardants, or occupational hazards
like military-related trauma) have been correlated with ALS pathogenesis in multivariable
statistical models, but few rigorous investigations have been performed due to small
numbers of clinical cases and a lack of appropriate neuromuscular models for controlled
studies [24]. On top of external environmental factors, the natural process of age-related
muscle loss and dysfunction is unsurprisingly tied to NMJ degeneration and ALS etiology,
with a mean disease onset age of 65 [8, 25]. The influence of age-related degeneration
in ALS is unclear but may be related to misfolded protein aggregation and oxidative
stress. Thus, with many factors responsible for ALS pathology, diagnosis is mainly
restricted to limited genetic screens and post-symptomatic tests. Like other
neuromuscular diseases, developing treatments and studying disease pathogenesis has
been difficult and costly due to the myriad of genetic and environmental differences
between patients. To date, only the antiglutamatergic compound riluzole and the
antioxidant edaravone have been approved for pharmalogical treatment of ALS, both of
which just mediate already damaged neuromuscular systems, and assuage symptoms
and extend survival for only a few months [26]. A modular in vitro model that accounts for
patient-specific genetic differences while being capable of recapitulating specific disease
microenvironment (e.g., oxidative stress, chronic inflammation, presence of toxic
14

compounds, altered matrix composition, etc…) may supplement current animal models
and clinical trials for greater therapeutic development efficiency for ALS.
While several genes have been identified or proposed to involve ALS
pathogenesis, the most common involve C9ORF72, SOD1, TARDBP, and FUS, usually
in some combination, accounting for 60-70% of familial ALS and about 10% of sporadic
ALS; the most common mutation, a repeat expansion mutation of C9ORF72,
encompasses greater than 30% of all familial ALS cases [26, 27] (Figure 1-4). While
much work has identified both key genes and some environmental factors that affect ALS
pathogenesis, there is still much work to be done to identify related gene mutations,
understand the mechanisms of disease progression, and possible ways to address them.

1.3. Case Study: Existing Approaches to Modeling Amyotrophic Lateral Sclerosis
Several in vivo and in vitro systems have been utilized for ALS research, shedding
light on mechanisms behind specific forms of ALS and potential routes to true therapy.
However, there are several limitations to the available models for studying ALS that limit
the relevance and efficiency of experimental data. This section will review both in vivo
and in vitro models developed to study ALS currently reported in the literature. We will
Figure 1-4: Distribution of the most common
genetic mutations responsible for sporadic and
familial forms of ALS. C9ORF72 repeat
expansions are detected in the majority of familial
ALS and a substantial portion of sporadic ALS.
Adapted from [22].

15

also highlight challenges of using these models and identify criteria that novel modeling
platforms must meet in order to help overcome the pitfalls of currently used models to
study ALS.
1.3.1. In Vivo Animal Models
For mechanistic and pharmacological studies of ALS, transgenic animal models
are routinely implemented. In particular, the SOD1 mutant mouse is extensively used
because it most closely recapitulates the phenotype of human ALS, and several
mechanisms of motor neuron pathology have been revealed through its study [28, 29].
However, the various strains of SOD1 mouse do not capture all human ALS phenotypes,
such as changes in RNA metabolism, and despite having large prevalence relative to
most other mutations, SOD1 mutations only account for 2% of all human ALS cases [29].
Other concerns include differences in endogenous protein concentrations, as the
synthesis rate of human SOD1 in the mouse models are varying magnitudes higher than
is expected in control mice, leading to murine-specific pathology such as vacuolization
artifacts that is likely irrelevant to human ALS progression [30, 31]. Thus, the translation
between experimental results to clinical relevance can be enigmatic. This system is also
not easily modular, and different mutations require different strains that can be time-
consuming and costly to develop. Non-mammalian transgenic models, such as drosophila
[32, 33] and zebrafish [34], have also been developed for ALS mutations and have
similarly provided insight into disease mechanisms, potentially identifying routes for
therapy. These models exhibit higher utility due to their lower cost, relatively easier
maintenance, more accessible imaging, and more flexible scalability. At the same time,
many of these models still suffer from limited modularity and patient-specific relevance.
16

Thus, although in vivo models have led to clinically significant data, patient-specific
pathogenesis is still difficult to represent completely in non-human, in vivo systems.
1.3.2. Traditional In Vitro Models and Microphysiological Systems
In vitro models of NMJs have the potential to alleviate many of the problems
presented by ALS animal models. In vitro models are higher throughput, more time and
cost-effective, and more easily modular and scalable. Furthermore, while traditional in
vitro models would not yield a whole organism or multi-organ response, careful
experimental design should allow for recapitulation of the most basic units of the tissue
or organ system of interest. For neuromuscular diseases, the motor unit is the
fundamental element that must be studied, although supporting cell types such as
Schwann cells and lymphocytes may play relevant if not essential roles for more nuanced
studies, such as tissue regeneration after trauma [35, 36]. Here, we describe various
approaches to skeletal muscle and motor neuron monocultures, coculture systems, and
the needs still unmet in the field of engineering physiologically relevant NMJs.
Skeletal muscle culture in vitro has traditionally involved the seeding of primary or
cell line-derived myoblasts onto matrix protein-coated substrates such as collagen,
promoting their adherence, and inducing differentiation into fused, multinucleated
myotubes through tissue confluence and serum concentration reduction [37]. Using this
simple technique, over the last several decades, many myoblast sources have been
widely utilized to study skeletal myotube maturation, metabolism, and pathophysiology,
including immortalized myoblast cell lines like murine C2C12 [38-41], primary animal and
human myoblasts [39, 41, 42], and, more recently, human induced pluripotent stem cell
(hiPSC)-derived myoblasts [43]. Immortalized cell lines have been useful in recapitulating
17

basic myogenic properties in a cost-effective manner, but often lose true myogenic
properties following immortalization and multiple passaging rounds, while primary cells
capture true skeletal muscle physiology, but human-derived samples may be limited in
supply and can vary in myogenic purity [39]. hiPSC-derived myoblasts may have the
greatest human relevance with the potential for being a reliable and renewable source
but may require further optimization of maturity to reach functional relevance [44].
Therefore, cell source must be carefully considered in experimental design.
While easy to prepare and fruitful for many decades, classical methods of skeletal
muscle culture on protein-coated surfaces suffer from several drawbacks that limit the
utility of most in vitro neuromuscular models. Firstly, without cues for spatial organization,
skeletal muscle myotubes will not develop into the aligned tissues observed
physiologically, but rather develop into branched structures with mostly random
orientation. Myotube disorganization prevents accurately studying both the internal
sarcomeric structure key to force generation and the detection of generated forces on a
more macro-tissue level. Secondly, delamination of muscle tissue from coated culture
substrates, often caused by increased stress generation of maturing myotubes or a loss
of myogenic properties over time in cell lines, prevents many of these systems from
carrying out experiments past two weeks [45, 46]. The limited timescale available is
especially concerning for traditional 2D culture use in neuromuscular cultures, as the
maturation of the motor unit may require longer culture time [6]. Even for studies into
myopathies in which the role of motor neurons in fiber degeneration is secondary,
delamination in these systems results in significant decreases in tissue and nucleic
density and myogenic index (percent of nuclei fused into myotubes), hindering cell-cell
18

and cell-matrix interactions that regulate myotube geometry, alignment, and fusion
capability [47]. Because alignment and fusion occur concurrently with early myofibril
protein aggregation, degeneration of these factors further limits sarcomeric development
[48]. Deterioration of the in vitro muscle tissue induces sarcomeric degeneration and
dispersal of sarcomeric proteins, leading to unstable and non-contractile myotubes. Thus,
even the myotubes that remain on culture substrate exhibit stunted maturation and limited
to no functionality.
Additionally, there is a wide range of contractile stresses that are generated by
skeletal myotubes depending on the experimental platform, cell source, specific muscle
interrogated, and genetic and environmental factors influencing tissue function [49].
Because contraction is a key functional output and indicator of skeletal muscle health and
maturity, studies utilizing conventional in vitro cultures that render contractile assessment
impossible due to misalignment or delamination possess significant drawbacks. Even in
the more complex models described, the majority of tissue-engineered muscle tissues
exhibit specific forces in the ones to tens of kPa, a magnitude below the hundreds of kPa
that is measured from freshly isolated native human tissues [49]. Engineering skeletal
muscle constructs that more closely mimic physiological skeletal muscle function is
essential for producing in vitro NMJs that can be studied with clinical relevance.
To address poor skeletal muscle tissue alignment, maintenance, and function in
vitro, tissue engineers have adapted various strategies. One approach has been to isolate
whole muscle from animal models directly for immediate use in functional experiments
such as measuring contractility and fatigue, allowing for intact study of native muscle
including its native matrix and surrounding supporting tissue; however, while ex vivo
19

tissue explants from humans and animals can be used for functional research studies,
they are often insufficient throughput-wise due to limitations in availability and cost, and
direct control over the microenvironment cannot be easily achieved [50]. Several groups
have incorporated micro- or nanofabrication techniques to skeletal muscle culture to
better recapitulate structure and function of the native tissue using physiologically relevant
constructs of myotubes known as microphysiological systems (MPS) [51]. For example,
spatial micropatterning of matrix protein [52] or micromolding of hydrogels [46, 53, 54] via
soft lithography methods has been used to align striated muscle cell types, allowing for
the formation of aligned tissues whose contractile stress generation can be quantified
though integration with cantilever-based systems, such as muscular thin films [55-57], or
other deflection-based systems [58]. These systems also allow for at least medium-
throughput functional testing that addresses issues of cost and testing efficiency.
However, delamination often remains an issue after a few weeks, preventing the
maturation time necessary for use in neuromuscular systems. Other systems, such as
engineered 3D myobundles, may have the potential for eliminating issues with
delamination by encapsulation of muscle fibers within a gelled matrix that aligns myotubes
based on inherent tensile cues as the introduced matrix compacts. These 3D structures
provide a more physiological microenvironment of matrix-myotube interaction and often
display higher tissue densities, but data remain mostly unavailable after 2 weeks of
culture [59-62]. Furthermore, these 3D systems are complex, often requiring technical
expertise and a high number of cells per construct, limiting throughput [60, 61]. Thus,
while many complex microphysiolgical systems have been proposed over the last several
years, there are still competing benefits and downsides to each model.
20

Regarding motor neuron culture, issues of cell source, culture lifetime, and
physiological relevancy also exist. However, recent advances in differentiation of hiPSC
cells into motor neuron progenitors has allowed for increased human relevancy in
experiments with greater experimental control and accessibility [63]. hiPSC-derived motor
neurons from control and ALS patients have been used to probe mechanisms of ALS and
identify molecules that lead to functional recovery [22, 23, 64]. Moving forward, using
hiPSC-derived cells will be significant in yielding human and patient-specific data with
clinical relevance, especially as primary human motor neurons are not easily attainable.
However, monocultures of hiPSC-derived motor neurons preclude investigation into
dysfunction at the neuromuscular interface, which is an important aspect of the pathology.
To address this, multiple approaches to coculture motor neurons and skeletal muscle
have been attempted, including seeding motor neurons directly on top of myotubes [65-
67], isolating motor neurons and myotubes into microfabricated compartments connected
by axon-permissive microchannels [68, 69], and culturing motor neuron spheroids
adjacent to or mixed into 3D engineered myotube bundles in microfluidic devices [59, 70].
While direct seeding is the simplest method, separating motor neuron bodies from
myotubes is somewhat more reminiscent of native physiology, as only axons stem from
the spine to innervate skeletal muscle. Many microchannel-based platforms, though,
show decreased viability of motor neurons in cultures of greater than a week compared
to unenclosed motor neurons, likely due to biocompatibility and diffusion-limit issues [71].
Pillar-based separation, where undissociated motor neuron spheroids are physically
barred and innervation of skeletal muscle is dependent on axon outgrowth [59], allows for
looser motor neuron separation and may help ease issues with viability, albeit with less
21

spatial control of innervation that regularly spaced microchannels would allow. 3D
bundles recapitulate skeletal muscle fibers more physiologically in tissue density and
matrix distribution, but also render spatial distribution of motor neurons difficult to control
and may make potential NMJs difficult to access for imaging or functional testing.
Perhaps most pressing, despite this wide array of approaches, NMJs in vitro fail to
recapitulate the mature, pretzel-like folded structures of NMJs in vivo [72], instead
presenting patchy, blot-like structures of presynapse and AchR localization in vitro
(Figure 1-5) [59, 66, 69]. Thus, despite multiple studies showing evidence that skeletal
muscle can be induced to contract via electrical stimulation or chemical stimulation of
motor neurons [59, 60, 69, 70], there is difficulty in understanding the available data’s
translation into clinical significance due to a mismatch in the expected structure-function
relationship of the engineered NMJs. Hypotheses for why the expected structures are not
observed include limited time of coculture and degeneration of the tissue due to a failure
to maintain skeletal muscle tissue or motor neuron viability. These failures also impact
the ability to perform long-term disease modeling, as well as chronic exposure to chemical
toxins or potential therapeutics. Numerous barriers must still be overcome for in vitro
models of neuromuscular systems to play a convincingly legitimate role in studying and
addressing the pathology of neuromuscular diseases.
22

1.3.3. Outlook – Current Challenges in Modeling Neuromyopathies
From the obstacles described in detail in the last subsection, the challenges of
interrogating skeletal muscle tissue and the NMJ in vitro must be addressed to improve
the study of neuromuscular diseases and their potential therapies [73]. Without models
Figure 1-5: In vivo NMJs are characterized by distinct invaginations of acetylcholine receptors
localized with motor neuron axon terminals that are poorly recapitulated in various approaches taken
by current in vitro models. A) A mouse NMJ exhibiting distinct, pretzel-like co-localization of
presynapse and acetylcholine receptor marker. Adapted from [72]. B) Poor NMJ morphology in ESC-
derived motor neuron and C2C12 myotube coculture. Adapted from [66]. C) Engineered 3D
myobundles result in aligned tissues but show poor acetylcholine receptor aggregation even when
innervated by stem-cell derived motor neurons, limiting NMJ maturation. Adapted from [59]. D)
Patterned human myotubes innervated by hiPSC-derived motor neuron axons traversing through
microchannels are more anatomically relevant, but still lack proper NMJ morphology. Adapted from
[69] under CC-BY.

23

that more properly replicate physiological muscle fiber and NMJ structure and function,
mechanistic and pharmacokinetic studies cannot be performed in vitro reliably.
Microphysiological models combining engineering techniques with developments in
biology and stem cell development have the potential to improve upon the current in vitro
models for studying skeletal muscle by extending culture lifetime, permitting functional
interrogation, and advancing neuromuscular tissue maturation. As the general population
age expectancy increases and becomes more prone to deleterious effects of
neuromyopathies [74], the need and demand for modular and higher-throughput systems
to study the motor unit reliably and accurately will likely increase dramatically. In the next
subsection, we will outline current aims that should be taken towards developing ideal
systems of in vitro motor units for modeling disease.
1.4. Objectives and Aims
Rationale: As described above, neuromuscular diseases lead to fatal muscular
atrophy. Outside of palliative care, this disease currently lacks effective therapeutics.
Currently, the study of motor unit components and strategies to prevent or reverse its
degeneration can be difficult to carry out with current models due to species-specific
differences, short cell culture lifetimes, immaturity of engineered tissues, and low-
throughput platforms, among other issues, that lead to a lack of clinically relevant data.
Thus, our goal is to apply engineering techniques to biological methods to improve upon
current approaches to culturing in vitro neuromuscular tissues. The developed
neuromuscular tissue MPSs will recapitulate key structural characteristics of maturing
skeletal muscle tissue. Ideally, it should be capable of modeling diseased neuromuscular
24

tissues, capturing disease pathogenesis and clinically relevant responses to potential
therapeutic compounds.
Objective: Engineer and validate MPS for healthy and diseased neuromuscular
tissues with improved maturity and scalability compared to current in vitro approaches.
Aim 1: Implement modular skeletal muscle MPSs to evaluate and characterize
accessible cell sources that yield optimal sarcomere structure and contractile stress
generation over 2 or more weeks in culture.
Aim 2: Coculture engineered skeletal muscle and hiPSC-derived motor neurons to
evaluate NMJ morphology and synaptic electrophysiology over time.
Aim 3: Translate engineered skeletal muscle MPSs to characterize human
mutation-specific disease phenotypes and responses to potential therapeutics.
Impact: Engineering physiologically relevant neuromuscular tissues in
microphysiological systems capable of functional outputs provide researchers with the
opportunity to develop more clinically relevant insights into the basic mechanisms behind
neuromuscular biology and the pathology of neuromuscular diseases, potentially
increasing the efficiency and output of the currently congested therapeutic development
pipeline.

25

Chapter 2 Implementing 2D and 3D Skeletal Muscle Microphysiological Systems
with Advanced Structural and Functional Markers
Aim 1: Implement modular skeletal muscle MPSs to evaluate and characterize
accessible cell sources that yield optimal sarcomere structure and contractile stress
generation over 2 or more weeks in culture.

2.1. Introduction
Skeletal muscle systems have been challenging to model in vitro for several
reasons. First, while many skeletal muscle cell sources are commercially available, they
hold respective disadvantages. Cell lines do not completely mimic mature structure and
function of native cells, while primary cells have variability and limited passage lifetime,
leading to complications in availability and reproducibility [75]. Mutation-specific, human-
relevant primary cells are also difficult to come by for many general research labs.
Second, conventional substrates fail to re-create critical features of native tissue,
including compliance or cellular alignment, which has been shown to dictate muscle-
specific gene expression; this restricts myotube sarcomeric maturation and functional
capacity [76]. One reason for NMJ immaturity is that most coculture systems are not
reported beyond two weeks, and a longer duration may be required for NMJ maturation
[22, 59]. The immaturity of the muscle tissue, whose proliferation and sarcomeric
development vary widely depending myoblast source and culture conditions, is a major
factor preventing further NMJ development [77]. Last, limited functional assays with at
least medium throughput exist, hampering investigations into the effects of drugs or
mutations. Often, functional outputs, such as contractility, are often insufficiently
26

quantified because of poor myogenic development, technological restrictions, or
misalignment of muscle tissue [46, 60, 78, 79]. Compounded, these shortcomings
preclude formation of physiologically relevant neuromuscular tissues with accessible
functional outputs for disease modeling.
We aim to characterize and identify ideal sources for long-term skeletal muscle
culture and incorporation into functional neuromuscular tissues. This section describes a
protocol using micromolded gelatin hydrogels, minimally modified from procedures
previously described [46, 54, 55], to culture aligned myotubes derived from commonly
used, well-characterized myogenic sources (C2C12, primary chick, and primary human
myoblasts) [80, 81]. We then integrated muscular thin films (MTFs), a cantilever-based
assay [55-57, 81], to assess contractile stresses generated by the tissues. Because 2D
sheets of muscle fail to recapitulate the complex cell-cell and cell-matrix interactions in
3D space, we also describe a more architecturally accurate myobundle model based on
3D printing approaches [82] and fibrin-based hydrogels [60-62], which could also be
assessed for contractile force generation. These data demonstrate our platforms’
reliability, modularity, and eventual utility for coculture with motor neurons for engineering
functional in vitro models of NMJs. Integration of human- and mutation-specific cells onto
these long-maintenance platforms will also yield more relevant mechanistic conclusions.
2.2. Materials and Methods
2.2.1. 2D Gelatin Hydrogel Substrate Fabrication
The substrate preparation procedure is summarized in Figure 2- 1 [81, 83]. 50 mm
x 15 mm polystyrene Petri dishes were covered with tape (Patco, 3900R) and laser-cut
into 260 mm
2
hexagons using a 30W Epilog Mini 24 Laser Engraver (100% speed, 25%
27

power, 2500 Hz). The hexagon shape was chosen to maximize the quantity of substrates
per dish. Within each hexagon, circles were laser-cut into the tape (18% speed, 6%
power, 2500 Hz). For MTF substrates, two additional rectangular areas (4.1 mm x 9.9
mm) were laser-cut into the tape. Circles of tape were then removed such that only the
edges of the hexagons of all substrates and the rectangular areas of MTF substrates
remained taped.
Standard photolithography and soft lithography techniques were used to fabricate
PDMS stamps with 10 micron-wide lines separated by 10 micron-wide spacing and
approximately 2 microns in depth [46, 54, 84]. Stamps were sonicated in 95% ethanol
prior to use. Equal volumes of 20% w/v 175g Bloom Type A porcine gelatin (Sigma,
G2625) in ultrapure water at 65 ℃ and 8% Activa TI transglutaminase (TG) (Ajinomoto,
1002) in ultrapure water at 37 ℃ were combined to form a 10% gelatin – 4% TG solution.
The hydrogel concentration was chosen based on the results of previous experiments
[46]. The mixture was homogenized for 30 seconds and degassed for 20 seconds in a
centrifugal mixer (Thinky USA, AR-100).
Polystyrene hexagon substrates were treated with plasma (Harrick Plasma, PDC-
001-HP) in ambient air for ten minutes to selectively increase hydrogel adhesiveness at
the exposed areas. After plasma treatment, the tape rectangles were removed from MTF
substrates. 10% gelatin – 4% TG solution was pipetted onto substrates within 20 minutes
of plasma treatment. PDMS stamps were slowly pressed on top of the hydrogel solution,
which was left to cure overnight at room temperature (b). The taped edges ensured
consistency in the height of the hydrogels, as described previously [46, 55]. Following the
curing step, substrates were rehydrated with ultrapure water. Stamps and remaining tape
28

were carefully removed in the direction of the pattern to prevent pattern disruption. To
fabricate MTFs, hydrogels were dried for 30 minutes at room temperature and then two
rows of four cantilevers (3.4 x 1.4 mm, separated by 0.8 mm) were laser-cut twice (15%
speed, 4% power, 2500 Hz and then 13% speed, 3% power, 2500 Hz) in the hydrogel
above the un-activated rectangular regions of the polystyrene. Substrates were
transferred to 12-well plates, rinsed in PBS, and kept at 4 ℃ for up to a week prior to cell
seeding. Before seeding, substrates were treated with a UVO Cleaner Model 342 (Jelight
Company) for one minute for sterilization.

2.2.2. Skeletal Muscle Cell Culture
C2C12 myoblasts and primary human skeletal muscle myoblasts (hSKM) were
purchased from ATCC and Lonza, respectively. Cells were initially thawed and cultured
in their respective growth media (Table 2-1) in T175 flasks, similar to previously reported
literature [46, 60]. When reaching 80% confluence, cells were passaged using 1X trypsin-
EDTA solution (Sigma, T3924). Select C2C12 and hSKM myoblasts were cryopreserved
Figure 2-1: Fabrication of
micromolded gelatin hydrogel
muscular thin films (MTFs). Laser-
cut polystyrene coverslips are
masked with tape, which is then
engraved. Tape A is removed, and
plasma treatment applied,
increasing adhesiveness in the
unmasked region. Tape B and C
are then removed and 10%
gelatin, 4% transglutaminase
solution is pipetted onto the
coverslip. A PDMS stamp is
placed on the solution, which is
cured overnight at room
temperature. The hydrogel is then
rehydrated and the PDMS stamp
and tape D are removed. MTF
cantilevers are then laser
engraved into the hydrogel. [81]
29

in their respective GM with 10% DMSO at passage two and three, respectively. For all
experiments, only myoblasts from passages 3 to 6 were used, with media changes every
other day. To transfer myoblasts to hydrogel substrates in 12-well plates, cells were
trypsinized from T175 flasks and seeded at 500,000 cells per substrate. Cells were
maintained in GM until confluence (3-4 days), at which point GM was shifted to
differentiation media (DM) to initiate myoblast fusion and differentiation into myotubes.
DM was refreshed every other day until experimental endpoints.
Day 11 chick embryos were purchased from AA Lab Eggs and thigh muscle tissue
was isolated and minced using forceps and scalpel. Four collagenase (Worthington
LS004177, Lot 43K144303B) (1 mg/mL in Hank’s Balanced Salt solution) digestions were
performed for three minutes each at 37 °C, with mechanical dissociations in between
each digestion by pipetting up and down with a 10 mL serological pipette. Two
consecutive 30-minute preplating steps at 37°C in T75 and T175 flasks were used to
purify myoblasts from the dissected tissue, similar to previously described protocols for
isolating neonatal rat cardiomyocytes [85-87]. Chick skeletal muscle (chSKM) myoblasts
were then seeded onto substrates with 500,000 cells or T175 flasks for expansion in
appropriate growth media [78]. Myoblasts from passages 0 to 3 were used for
experiments with media refreshed every other day, following the same differentiation
timeline as mentioned previously.
Table 2-1: Components of growth and differentiation media for skeletal muscle tissues
Cell Type Component Concentration Product Information
C2C12
(ATCC, CRL-
1772)
Growth Media
high glucose DMEM 4.5 g/L Invitrogen, 11995-040
fetal bovine serum 10% Hyclone, SH3007103
penicillin-streptomycin 1% Lonza, 17-602E
Differentiation Media
30

high glucose DMEM 4.5 g/L Invitrogen, 11995-040
horse serum 2% Hyclone, SH3007103
penicillin-streptomycin 1% Lonza, 17-602E
cytarabine 10 μM Sigma, C1768
chSKM (AA
Lab Eggs)
Growth Media
low glucose DMEM 1.0 g/L Invitrogen, 11885-084
horse serum 10% Hyclone, SH3007103
chicken serum 5% Gibco, 16110082
vitamin B12 4 μg/mL Sigma, V-2876
penicillin 500 units/mL Sigma, P-4687
calcium chloride 3 mM Sigma, 449709
Differentiation Media
DMEM/F12 50% Gibco, 11320-033
Neurobasal 50% Gibco, 21103-049
N-2 Supplement 0.5 X Gibco, 17502-048
B-27 Supplement 0.5 X Gibco, 17504-044
vitamin C 0.1 mM Sigma, A92902
Glutamax 1 X Gibco, 35050079
vitamin B12 4 μg/mL Sigma, V-2876
penicillin 500 units/mL Sigma, P-4687
cytarabine 10 μM Sigma, C1768
hSKM
(Lonza, CC-
2580)
Growth Media
low glucose DMEM 1.0 g/L Invitrogen, 11885-084
fetal bovine serum 8% Hyclone, SH3007103
SkGM SingleQuots,
insulin removed
1X Lonza, CC-1439
penicillin-streptomycin 1% Lonza, 17-602E
Differentiation Media
low glucose DMEM 1.0 g/L Invitrogen, 11885-084
horse serum 2% Hyclone, SH3007103
insulin 10 μg/mL Lonza, CC-1439
penicillin-streptomycin 1% Lonza, 17-602E
cytarabine 10 μM Sigma, C1768

2.2.3. 2D Myotube Structural Characterization
After one, two, and three weeks in DM, tissues on non-MTF substrates were rinsed
three times with PBS and fixed using ice cold methanol for ten minutes. Immunostaining
for sarcomeric ɑ-actinin (Sigma, A7811), a sarcomeric protein marker, was then
performed at a 1:200 dilution. During secondary staining, goat anti-mouse conjugated
Alexa Fluor 546, 647 fluorophore conjugated α-bungarotoxin (BTX), an acetylcholine
31

receptor marker, and 4′,6-diamidino-2-phenylindole (DAPI) were added to the staining
solution at dilutions of 1:200. Coverslips were mounted onto glass slides with ProLong
gold antifade mountant (ThermoFisher Scientific, P36930) and stored at -20 °C prior to
imaging. Widefield or confocal microscopy was performed using a Nikon Eclipse Ti
microscope at 20X air and 60X oil objectives with Confocal Module Nikon C2 and Andor
Zyla sCMOS camera.
Following image acquisition, structural characteristics of the myotube cultures
were quantified to determine maturity. Nucleic density was quantified by analyzing
stitched images of 24 fields of view per tissue (total area = 11.1 mm
2
). Some images were
further cropped to reduce noise from edge effects or visible aberrations in the culture
substrate. Custom CellProfiler (Broad Institute) code was used to determine the number
of nuclei per image based on intensity and size thresholding of the DAPI signal. Nucleic
density was calculated as the number of nuclei divided by the area of the stitched image.
To quantify myogenic index, custom CellProfiler code was used to mask ɑ-actinin
coverage of stitched images, representing myotubes. Then, co-localization of nuclei
within each mask was determined. The proportion of nuclei co-localized in ɑ-actinin-
positive masked areas to the total number of nuclei was reported as the myogenic index,
a global measure of myoblast maturation into myotubes.
To quantify sarcomere formation, we determined a sarcomeric index like those
previously reported [88, 89]. 2D fast Fourier transforms (FFT function) were performed
using ImageJ on five randomly selected 50 micron-wide myotube sections on each tissue.
The data from the Fourier transforms were collapsed radially to generate 1D power
spectrum profiles, which were then normalized to have an integrated area of one.
32

MATLAB curve fitting software was then used to split the radial profiles into aperiodic
(decaying exponential) and periodic (sum of multiple Gaussian functions) components.
The fitted aperiodic component was subtracted from the total and the area under the
periodic component was taken as the sarcomeric index. Sarcomere length was calculated
using automated z-disc detection code from regions of interest [90]. Myotube width was
measured using ImageJ by averaging the widths of each myotube in 5 square fields of
view (0.1 mm
2
), chosen randomly from the taken stitched image.
2.2.4. Quantifying Contractile Stresses on Muscular Thin Films
After one, two, and three weeks in DM, tissues on MTF substrates were transferred
to a 35 mm Petri dish and placed on the stage of a Nikon SMZ745T stereomicroscope to
quantify contractile stress generation, as previously described [55]. Briefly, substrates
were rinsed three times in 37 ℃ Tyrode’s solution (5.0 mM HEPES, 1.0 mM magnesium
chloride, 5.4 mM potassium chloride, 135.0 mM sodium chloride, 0.33 mM sodium
phosphate, 1.8 mM calcium chloride, 5.0 mM glucose). Fine tweezers were used to lift
the gelatin hydrogel cantilevers from the surface of the polystyrene substrate. After
peeling all films, platinum field stimulation electrodes were introduced into the dish and
used to apply 20 V at 2 Hz or 20 Hz to induce twitch or tetanus contractions, respectively.
Videos were recorded at 100 frames per second with a Basler acA640-120um USB 3.0
camera. Custom ImageJ (NIH) and MATLAB (Mathworks) software was used to trace the
radius of curvature during contractions of each MTF cantilever in the presence of
33

electrical pacing, as previously described [91]. The stress generated by each cantilever
in each frame was determined by a modified Stoney’s equation:

which accounts for each cantilever’s radius of curvature and hydrogel thickness and
compressive elastic modulus [92]. Twitch and tetanus forces can be isolated by looking
at frames of maximum contractions. The tetanus-to-twitch ratio was determined by
dividing the tetanus by the twitch stress for each cantilever, and further serves as a metric
of myotube physiological development. The gelatin thickness was measured to be 90
microns by embedding fluorescent beads and quantifying the total height of detectable
fluorescent signal with a Nikon Eclipse Ti microscope, and the elastic modulus was
determined to be 108.3 ± 10.8 kPa by compression testing with an Instron 5942 single
column tabletop tester and Bluehill 3 testing software, as previously reported [83].

2.2.5. RNAseq and Bulk Transcriptomic Analysis
Engineered chick tissues were lysed using 1 ml of TRIzol reagent (Thermo Fisher
Scientific) and RNA isolation was performed per the reagent manufacturer's protocol.
Briefly, at 4 °C, lysate underwent two chloroform phase separation steps to isolate RNA
from other cellular components. RNA was then precipitated with isopropyl alcohol and
two 75% ethanol washes were performed to remove trace amounts of lysing agent and
phenols to improve RNA purity. Total RNA was resuspended in 10 mM Trizma HCl
Equation 2-1: Stoney’s equation: σ cell =
contractile stress generated on a single MTF by
myotubes, E – elastic modulus of gelatin
hydrogel, t s – hydrogel thickness, – Poisson’s
ratio, R – radius of curvature, t c – cell thickness
34

(pH = 8.0) and stored at −80 °C prior to sequencing with an Illumina NovaSeq 6000
system (performed by Novogene Corporation, Inc.). Sequences (GEO Series accession
No. GSE172606 [80]) were filtered and normalized with fragments per kilobase of
transcript per million mapped reads—upper quartile (FPKM-UQ) normalization using
Partek Flow Genomic Analysis software. Data were visualized through principal
component and gene specific analysis with p = 0.05 and a fold change of 1.5 to define
differentially expressed genes. Fold-change data from Partek Flow were uploaded to
Qiagen Ingenuity Pathway Analysis (IPA) software with the same fold-change cutoffs to
perform pathway analysis and generate predictive mechanistic network schematics.
Specific networks were generated by overlaying imported fold-change data onto
interaction networks from the library.
35

2.2.6. 3D Myobundle Mold Fabrication
The fabrication of the 3D myobundle culture device is summarized in Figure 2-2a.
Master Mold Resin (CADWorks) casts for polydimethylsiloxane (PDMS 184, Dow
SYLGARD)-based myobundle molds were designed using TinkerCAD (Autodesk Inc.)
and printed using a CADWorks3D M50 Series digital light processing printer
(CADWorks3D). Resin casts were designed with two 13 x 2 x 1 (length x width x height)
mm half-chambers based on the optimal dimensions of previous studies [82, 93]. Towards
the ends of the myobundle chamber, the chamber width increases to 3 mm to allow for
better visualization of downstream analysis. Next, we cured PDMS within the resin casts
and removed the two PDMS device halves. A laser engraver (70% speed, 45% power,
5000 Hz frequency) was used to engrave 500 μm-wide rods into a 0.25 mm-tall silicone
sheet (Rogers HT-6240-0.25), which was then sandwiched between the two PDMS
device halves after plasma treatment in ambient air for 10 minutes. The fully assembled
Figure 2-2: Summary of 3D myobundle fabrication procedure. (a) Assembly of PDMS device parts through
plama bonding. (b) Overview of seeding and culturing myoblasts and motor neurons in a fibrin hydrogel.

36

device (shown from the top in Figure 2-2a) consists of a 13 x 2 x 2.25 (l x w x h) mm
chamber with the inner rod edges 10 mm apart. Devices were bonded on to 6-well plates
using 8 minutes of treatment with UVO Cleaner Model 342 (Jelight Company). Plates with
assembled devices were wrapped with parafilm and stored at ambient conditions prior to
one day before seeding, during which they were sterilized under UV light in a biosafety
cabinet overnight.
2.2.7. Seeding and Culturing 3D Myobundles
The seeding protocol is summarized in Figure 2-2b. Sterilized PDMS myobundle
devices were treated for one hour with 1% Pluronic (w/v) solution which is then aspirated
away completely. Chick myotubes were isolated and maintained as previously described.
Prior to seeding, 1.5 million chick myoblasts were resuspended in 32.5 μL of GM and 4
μL of thrombin. A separate vial of 20 μL GM 20 μL fibrinogen (20 μg/mL), and 20 μL
Matrigel was mixed and kept on ice. The myoblast-thrombin mixture is then mixed rapidly
without the formation of bubbles with the fibrinogen solution on ice. 70 μL of solution (~1
million cells) is pipetted into each device, which starts to gel within 30 seconds. After
allowing the hydrogel to solidify at 37 °C for 30 minutes, chick GM with the addition of 1.5
mg/mL aminocaproic acid (ACA, Sigma Aldrich), an anti-fibrinolytic agent, was added to
submerge myobundles. After 4 days of culture in growth media with ACA, myobundles
were removed from the bottom of the 6-well plate and cultured on a rocker in chick DM
with ACA supplement, allowing for increased diffusion of nutrients over 2 weeks of culture
time. Media change occurred every other day.
37

2.2.8. 3D Myotube Structural Characterization
Myobundles after one or two weeks of differentiation were washed with PBS and
fixed with ice cold methanol for 30 minutes at 4 °C. After three additional PBS rinses,
myobundles were incubated in 30% sucrose overnight at 4 °C to displace water and
prevent ice crystal formation during freezing, as previously described [82]. Samples were
then immersed in optimal cutting temperature compound in 25 x 20 x 5 mm cryomolds
and incubated at room temperature for 30 minutes. Then, samples were frozen by
cryomold immersion in liquid nitrogen-chilled isopentane for 1 minute. Frozen
myobundles were stored at -80 °C until time of sectioning. A Leica CM3050 S Research
Cryostat was used to generate 12 μm thick longitudinal sections, which were mounted
onto VWR Superfrost Plus Micro Slides. Sections were then stained for H&E following
standard protocols or incubated in 10% goat serum for 1 hour before following the same
staining and mounting protocols as described in section 2.2.4. Sections were also imaged
with the same microscopy methods detailed in section 2.2.4. Analyses for nuclei density
and myogenic index were also performed as detailed in section 2.2.4. Additional
measurements of myobundle width were taken using ImageJ by taking the maximum
width in longitudinal cryosections of the myobundle.
2.2.9. Quantifying Contractile Force Generation of 3D Myobundles
In culture, myobundles spontaneously contracted a few days after seeding, but
generally lost most spontaneous activity by one week of differentiation. After one and two
weeks in DM, myobundles were transferred to a 35 mm Petri dish and placed on the stage
of a Nikon SMZ745T stereomicroscope to quantify contractile stress generation in
Tyrode’s solution, similar to MTF substrates. Bundles were oriented perpendicularly to
38

the stimulating electrodes, and twitch and tetanus contractions were recorded with the
same protocol as gelatin hydrogel MTFs. Deflection of both PDMS anchor rods was
analyzed using ImageJ angle tool to determine the maximum angle of displacement
during twitch and tetanus contractions. The angle of displacement was used to calculate
displacement of the middle of the PDMS rod, which was then correlated to the contractile
force generation of the myobundle by referencing a linear standard curve (Figure 2-3).
The standard curve was created by characterizing the angle of deflection created by
applying known weights to PDMS rods. Time dynamics of contraction were further
analyzed by selecting a rectangular region of interest on the PDMS rod and tracking
changes in the total pixel intensity over time. These data were imported into MATLAB and
the rise and fall time constants (tau) were calculated using the risetime and falltime
functions.

2.2.10. Statistical Analysis
The data were tested for normality and analyzed using 1) student’s t-test/Mann-
Whitney or 2) two-way ANOVA and compared to each other using Tukey’s multiple
Figure 2-3: Standard curve for
displacement of PDMS rods using
known weights.
39

comparisons test (GraphPad Prism 7.04), as appropriate. Comparisons with p-values less
than 0.05 were considered statistically significant. In our figures, significance bars are
only drawn to signify statistically different groups between cell sources at the same time
point or at different time points for a single cell source. In figures, * denotes p < 0.05, ** p
< 0.01, *** p <0.001, **** p < 0.0001, unless otherwise noted.

40

2.3. Results
2.3.1. 2D Skeletal Myotube Morphology
We successfully engineered skeletal muscle tissues from C2C12, hSKM and
chSKM myoblasts on micromolded gelatin hydrogels. These myoblast sources were
chosen based on ease of access, including relatively low costs and documented isolation
and usage. Seeding myoblasts from these sources on the hydrogels resulted in aligned
skeletal muscle cultures exhibiting sarcomeric formations and contractile ability better
than myoblasts seeded on traditional substrates such as glass or polystyrene, as
previously reported [46, 54]. Recapitulating this native tissue architecture was key in
developing characteristic myotube structure and contractile ability. One primary aim of
this study was to determine which cell source yields the most consistent, mature skeletal
muscle tissues on the hydrogels over longer culture periods, that would allow for better
development of neuromuscular junctions when cocultured with motor neurons. We first
evaluated maturity by quantifying morphological characteristics of myotubes over three
weeks of differentiation.
To compare culture stability and myotube maturation over time, we calculated
nucleic density and myogenic index over 3 weeks of differentiation by evaluating at α-
actinin and DAPI immunostaining. Mature muscle tissues should maintain relatively stable
nucleic densities, as myotubes may continue to fuse but not proliferate post-
differentiation. Sustained proliferation after inducing differentiation may be indicative of
fibroblastic cells, which may play a role in delamination by displacing myotubes. By three
weeks of differentiation, nucleic density did not vary significantly between cultures from
the three sources investigated, indicating similar cellularity distributions among all tissues
41

at the end of culture period (Figure 2-4). In C2C12 cultures, nucleic density remained
relatively constant throughout the experiments and despite noticeable decreases for
hSKM cultures due to delamination of myotubes after 3 weeks, there was no statistically
significant differences between the two cell sources at any time point. While chick
myotubes had statistically significant differences between 1 and 3 weeks of culture, the
large deviation in nucleic density at week 1 likely due to initial variations in isolated
myoblast purity is largely absent by the third week of differentiation.
Continued fusion during differentiation should increase myogenic index as more
nuclei are incorporated into α-actinin-positive myotubes. Only chSKM tissues showed
Figure 2-4: Structure of engineered muscle tissues over three weeks of differentiation. (a) Representative
images of C2C12, chick, and human muscle tissues after one and three weeks in culture on micromolded
gelatin hydrogels. α-actinin (red), DAPI (blue). Scale bar: 100 μm. (b) Number of nuclei normalized to field
of view area and (c) myogenic index for each myoblast cell source at weekly timepoints. *p<0.05, **p<0.01,
***p<0.001, ****p<0.00001 [81].
42

statistically significant increases in myogenic index over three weeks. The myogenic
index of chSKM at week 3 was significantly higher than that of C2C12 and hSKM, whose
myogenic index deteriorate rapidly as myotubes detach from the hydrogel surface.
Together, these data show that chSKM were not only much more stable, but also
exhibited a more mature phenotype, highlighting the cell source’s utility for a long-term
model of muscular physiology and function.
Probing individual myotube morphology, we quantified myotube width, as well as
consistency of sarcomeric formation frequency with sarcomeric index and length (Figure
2-5). We expect healthy muscle tissues to reach a stable width and maintain their size
throughout culture period, and this was observed in chSKM (Figure 2-5b). However,
C2C12 cultures increased in width to a point where week 3 samples were statistically
significantly different from the chSKM samples. hSKM widths also increased with time
compared to chSKM, but to a lesser extent. The increase in myotube width was
associated with increased vacuolization of myotubes that had not delaminated, and may
be indicative of myotubes heading towards an apoptotic state [94]. We hypothesize that
hypertrophy also occurred as a compensation mechanism for displaced myotubes.
chSKM myotubes also displayed more consistently developed sarcomeres, indicated by
a stable sarcomeric index which was significantly higher at week 3 compared to C2C12
and hSKM (Figure 2-5c). Myotube sections with high consistency in sarcomeric formation
and spacing have higher sarcomeric index due to sharper and higher magnitude peaks
at the fundamental frequency of the periodic signal, which result from a more
homogenous distribution of ɑ-actinin signal in the spatial domain. chSKM myotube
stability was further evidenced by low variations in average sarcomeric length at the third
43

week of differentiation (Figure 2-5d). In contrast, C2C12 myotubes exhibited similar
localization of α-actinin striations with higher variance and hSKM sarcomere length was
not detectable, signifying both poor myotube maturity and degradation of the contractile
protein over time.

44

Figure 2-5: Myotube and sarcomeric organization over weeks of differentiation. A) Representative images
of individual myotubes and sarcomeres from C2C12, chSKM and hSKM tissues. α-actinin (red), DAPI
(blue). Scale bar = 50 μm for tissues, 10 μm for myotubes. B) Average myotube widths, C) sarcomeric
indexes, and D) sarcomeric lengths over differentiation time. *p<0.05, **p<0.01, ***p<0.001, ****p<0.00001
[81].

45

2.3.2. Myotube Contractile Stress Generation on Muscular Thin Films
Having evaluated tissue structure through immunofluorescence analysis, we used
MTFs to evaluate 2D tissue contractile ability on the tissue-level scale. MTFs were
engraved to functionalize the substrate, allowing for the detection of deflections of the
hydrogel during contraction. Myotubes from all cell sources cultured on the hydrogel
MTFs were capable of being reliably paced for at least one week post-differentiation. This
included both commonly measured twitch contractions at a lower frequency of 2 Hz and
sustained tetanus contractions at 20 Hz, which stimulated maximal contraction by
preventing relaxation in between voltage pulses (Figure 2-6). Prior to electrical
stimulation, basal stress was recorded as the baseline contractile stress of myotubes
when films were lifted off the polystyrene surface. There were no significant differences
between any groups except between chSKM and hSKM MTFs at week 3, which arose
Figure 2-6: Muscular thin
films (MTF) were analyzed
with image processing
MATLAB software based on
deflection of cantilevers. A)
Representative images of a
single MTF at basal, twitch,
and tetanus contraction. B)
Based on Stoney’s equation,
a trace of contractile stress
over time of a single
cantilever can be plotted.
Twitch and tetanus stress
were stimulated at 2 and 20
Hz, respectively. C)
Schematic of MTFs laser-
engraved into micromolded
gelatin hydrogels. D) 3D
perspective of cantilevers at
different stages of
contraction. Adapted from
[81].

46

due to severe delamination of hSKM, resulting in the presence of few myotubes (Figure
2-7a).
chSKM MTFs generated increasing twitch stresses with increasing differentiation
time, and by three weeks of differentiation, exhibited significantly higher twitch stresses
compared to all other timepoints from any cell source (all p < 0.0001) (Figure 2-7b).
Myotubes from other cell sources decreased in generated twitch stresses over time. The
same trend was observed with the generated tetanus stresses (Figure 2-5c), with chSKM
after three weeks differentiation producing an average tetanus much higher than those of
C2C12 and hSKM. The three-fold magnitude difference in stress was a clear indicator of
sustained performance of chSKM over time. Interestingly, the twitch-to-tetanus ratio of
Figure 2-7: Stress generation by myotube MTFs. A) Basal stresses generated by unstimulated myotubes
on MTFs. B) Twitch stresses generated by stimulating MTFs at 2 Hz. C) Tetanus stresses generated by
stimulating MTFs at 20 Hz. D) Average tetanus-to-twitch ratios of MTFs. *p<0.05, **p<0.01, ***p<0.001,
****p<0.00001 [81].
47

chSKM was slightly supraphysiological (physiological ratios range between 4-10,
depending on muscle location) at week one, compared to below physiological ratios of
C2C12 and hSKM, which were comparable to those previously reported (Figure 2-7d)
[39]. However, by week three the ratio for chSKM MTFs fell to expected values and
remained significantly higher than C2C12 and hSKM MTFs. This decrease in ratio can be
explained by more consistent twitch contractions in response to stimulation over time,
rather than a decrease in generated tetanus forces, serving as further evidence of
myotube development.
2.3.3. Transcriptomic analysis of engineered chick muscle tissues on gelatin hydrogels
To identify differentially expressed genes (DEG) that correlate with the observed
enhanced myogenic quality and contractility in chick tissues, we performed bulk RNAseq
on engineered chick muscle tissues after one and three weeks of differentiation. This
allowed for a better understanding on the molecular level of what pathways were being
activated to promote and maintain myogenic health. Figure 2-8a shows a global heat
Figure 2-8: Global transcriptome of chick muscle tissues cultured for one week and three weeks. (a)
Hierarchical heat map clustered by features for week 1 (blue) and week 3 (red) samples. (b) PCA plot of
week 1 (blue) and week 3 (red) chick tissues. Individual tissues are labeled as “week in culture-sample
number” in each panel [80, 81].

48

map of the transcriptome of multiple chick tissues after one and three weeks. Principal
component analysis (PCA) identified that roughly 69% of the variation in the data could
be explained by the first three principal component vectors containing the weighted
contributions of each gene in the data set (Figure2-8b). Notably, several genes involved
in sarcomere structure (MYH1G, TNNI1, ACTN1, HYAL1, DES) [95, 96], fast/slow fiber
phenotype specification (MYH1G, SOX5) [95, 97], calcium signaling
(CAPN6, CAPN10, PVALB) [95], and mitochondrial respiration (MB, MFN1, MFN2) [95]
have high component loadings, indicating that these structures and processes are
actively remodeling in chick muscle tissues with increasing culture time. Intriguingly,
synapse markers or other key players in NMJ formation (AGRN, CHRNA1, PLEKHG5)
[95] also had high loading, but were downregulated over time, as discussed in more detail
below.
Next, we performed gene specific analysis between chick muscle tissues cultured
for one week and three weeks. As shown by the volcano plot in Figure 2-9a, 2244 of the
15 394 detected genes were differentially expressed. We mined the dataset and identified
several differentially expressed genes with known role(s) in muscle development and
physiology. Myostatin (MSTN or GDF8) and myogenic factor 5 (MYF5), which are
markers of proliferating myoblasts [98, 99], are downregulated, consistent with increased
fusion of myoblasts into myotubes. Myoglobin (MB), an oxygen transporter and buffering
protein in striated muscle [95], exhibited one of the highest fold-changes in normalized
gene count ( ∼12×), possibly to meet the higher mitochondrial respiratory demands of
contractile myotubes compared to myoblasts (Figure 2-9b). MB upregulation is also
expected during the maturation of slow-twitch fibers that consume oxygen and expend
49

energy over sustained time periods [100]. The development of contractile units was also
demonstrated by the upregulation of several sarcomere proteins (Figure 2-9c), including
the chicken-specific myosin heavy chain 1G (MYH1G, similar to human MYH1), myosin
heavy chain 7 (MYH7), and alpha-actinin (ACTN1). Thus, analysis of muscle-related
genes in the bulk transcriptome suggests that changes in aerobic respiration, myoblast
proliferation and fusion, and sarcomere development are in part responsible for the
structural and functional improvements observed in chick muscle tissues after three
weeks in culture. Due to their role in promoting motor neuron survival [101], we also
compared the expression of brain-derived neurotrophic factor (BDNF), ciliary
neurotrophic factor (CNTF), and glial cell-derived neurotrophic factor (GDNF) (Figure 2-
9d). Of these three factors, only BDNF was significantly upregulated. Finally, we
compared the expression of genes related to the AGRN-LRP4-MUSK complex (Figure
2-9e). In native NMJs, agrin (AGRN), a proteoglycan secreted by motor neurons that
stabilizes the developing synapse, binds to activated complexes of low-density lipoprotein
receptor-related protein 4 (LRP4), a sarcolemma receptor for agrin, and muscle
associated receptor tyrosine kinase (MUSK), a sarcolemma receptor that congregates
LRP4. Motor neuron-derived AGRN binding to the LRP4-MUSK complex clusters
acetylcholine receptors [102, 103], initiating synaptic differentiation. Our data show
that MUSK is significantly downregulated after three weeks in chick muscle tissues
despite an increase in LRP4. This downregulation of MUSK is likely a response of the
muscle tissue to a lack of agrin, which is secreted by motor neurons that are absent in
these tissues. Although the chick muscle tissues did express AGRN, the amount of AGRN
produced by the muscle tissue was likely insufficient to induce acetylcholine receptor
50

clustering. Additionally, compared to neuron-secreted isoforms of AGRN, muscle-
secreted AGRN isoforms generally lack an insertion of amino acids at a specific sequence
location, referred to as the B site (Z site in mammals). The absence of these amino acid
insertions reduces the potency of AGRN to cluster acetylcholine receptors by several
magnitudes [104].
Next, we used Ingenuity Pathway Analysis (IPA) to generate predictive
mechanistic networks based on the transcriptome data. We found that the network for
organization of sarcomeres was predicted to be activated due to upregulation
of FN1, ROCK2, EDN1, GATA4, MYLK, and TGFB1 (Figure 2-10a). The network for
innervation of muscle was also predicted to be activated due to an upregulation of
neurotrophic factors (IGF1, BDNF, CNTF, and GDNF) and a downregulation
of MUSK (Figure 2-10b). Formation of NMJs was predicted to be inhibited based on
changes in BDNF, AGRN, and MUSK (Figure 2-10c), which can likely be attributed to
the lack of motor neurons in these tissues. Overall, the transcriptomic data corroborate
the structural and functional data and provide additional insights into the expression of
genes that promote sarcomere development and innervation in engineered muscle
tissues.

51

Figure 2-9: Differentially expressed genes in chick muscle tissues cultured for one week (W1) and three
weeks (W3). (a) Volcano plot of all genes up- and downregulated or not significantly changed in tissues
cultured for three weeks compared to one week. Differentially expressed genes (DEG) were defined as
those with p < 0.05 and a magnitude fold change of greater than 1.5. Normalized gene expression of (b)
regulators of myogenic proliferation and metabolism, (c) contractile proteins, (d) neurotrophic factors, and
(e) signaling molecules in acetylcholine receptor clustering. *p < 0.05; **p < 0.01; ***p <0.001; ****p < 0.0001
[80, 81].
52

2.3.4. 3D Skeletal Myotube Morphology
To create a model with more physiologically relevant cell-matrix and cell-cell
interactions, we engineered 3D skeletal fibrin myobundles derived from chSKM myoblasts
in assembled chambers derived from 3D printed resin molds and a laser-engraved
commercial PDMS sheet. The ends of bundle molds were designed to be wider than the
main length of the bundle to better visualize the PDMS rods for functional assay, as
discussed in the next section. Myobundles were formed from the compaction of fibrin-
Matrigel hydrogels on laser-cut PDMS rods, which created a tension force that aligned
myotubes during differentiation. Compaction of the myobundles led to aligned myotube
structures after one week of differentiation on a rocker, and tissue structure was
Figure 2-10: IPA-generated predictive mechanistic networks. In chick muscle tissues cultured for three
weeks compared to one week, pathways for (a) organization of sarcomeres and (b) innervation of muscle
are predicted to be activated. (c) The pathway for formation of neuromuscular junctions is expected to be
inhibited [80, 81].
53

maintained following an addition week of differentiation. H&E staining showed dense
muscle tissue formation at both time points, with the myobundle width staying relatively
constant (Figure 2-11). Immunostaining showed myotubes display well-formed
sarcomes, and confirmed higher nuclei density and comparable myogenic index
compared to 2D cultures on gelatin hydrogels at the same time points (Figure 2-12).
However, few changes are seen after one week of myobundle differentiation in either
metric. The absence of synapsin stain with the excpetion of minor background noise
confirmed that myobundles were free of neuronal cells.

Figure 2-11: Hematoxylin and eosin stains of 3D myobundles after one and two weeks of differentiation.
Scale bar = 200 μm.
54

2.3.5. Myobundle Contractile Stress Generation
Like our 2D tissues on gelatin hydrogels, engineered fibrin myobundles were
paceable at 2 and 20 Hz, yielding twitch and tetanus contractions, respectively.
Myobundles appeared to generate higher contractile twitch and tetanus forces as time in
culture increases, although not to a statistically significant degree (Figure 2-13a). These
values fall under similar magnitudes to those generated in other engineered 3D systems,
which generally present at a physiologically embryonic level [49, 61]. The tetanus-to-
twitch ratio does not appear to change with culture time, remaining around 1.5 after two
weeks of differentiation (Figure 2-13b). This also falls below physiological levels of 4 to
Figure 2-12: 3D myobundle morphology after 2 weeks of differentiation. (a) Representative myobundle
cryosections. α-actinin (red), synapsin (green), DAPI (blue). Scale bar = 200 μm (left), 50 μm for zoomed
in images (right). (b) Number of nuclei normalized to field of view area and (c) myogenic index at weekly
timepoints.
55

10, similar to several past studies [39]. Interestingly, the time dynamics of the detected
contractions showed a significant decrease in both the rise and fall time constant (τ,
defined as the time to reach 63.2% of the final value) when comparing bundles from one
to two weeks of differentiation (Figure 2-13c). This may indicate changes in fiber make-
up over time towards a more glycolytic-fast phenotype, which imply further metabolic and
structural muscle specialization [105].
2.4. Discussion
A primary goal of this aim was to engineer reliable, modular 2D and 3D systems
capable of identifying optimal skeletal muscle sources that can maintain aligned skeletal
muscle tissue for multiple weeks. This would help accomplish the overall aim to engineer
a stable skeletal muscle microphysiological system that improves upon those currently
described in the literature. Long-term skeletal muscle culture beyond the two-week limit
most previous studies have reported is key since cultures on the magnitude of weeks or
months may be necessary for the development of physiologically relevant cultures
suitable for chronic disease modeling. Delamination of muscle tissue from culture
substrate prevents many of these systems from carrying out experiments past a few
weeks. Significant decreases in nucleic density and myogenic index are expected with
Figure 2-13: Contractile force generation by 3D fibrin chick myobundles. (a) Electrically stimulated twitch
and tetanus forces, (b) tetanus-to-twitch ratio, and (c) rise and fall τ after one and two weeks of
differentiation. *p < 0.05; **p < 0.01
56

high rates of delamination, as this hinders cell-cell and cell-matrix interactions, which have
been shown to affect myotube geometry, alignment, and fusion capability [47]. In our 2D
gelatin hydrogel system, these trends were clearly captured in the decline of morphology
characterization metrics of C2C12 and hSKM myotubes over time. Because alignment
and fusion occur concurrently with early myofibril protein aggregation, degeneration of
these factors further limits sarcomeric development [48], which explains our lower
sarcomeric indexes of C2C12 and hSKM myotubes compared to chSKM. As C2C12 and
hSKM myotubes continued to deteriorate, sarcomeric degeneration resulted in diffuse α-
actinin localization, leading to unstable and increased sarcomeric length. Importantly,
chSKM cultures on both our gelatin hydrogels and fibrin myobundles retained stable
myotube morphology and improved myogenic index and sarcomeric formation after three
weeks of culture, essential for engineering consistent myogenic tissues for
neuromuscular models. It should be noted that increases in chSKM nucleic density on
gelatin hydrogels was statistically substantial from one to three weeks in culture, but the
fact that myogenic index increased significantly suggests any tissue loss during the
culture period is not myogenic in origin.
Important functional outputs of skeletal muscle, such as contractility, are also often
insufficiently quantified in traditional in vitro platforms of unaligned skeletal muscle tissue
culture in dishes [46, 60, 78, 79]. Quantifying contraction of aligned muscle tissue is
feasible for 3D myobundles, but fabrication and data acquisition are more cumbersome.
Other attempted strategies to evaluate contractile force involve recording large samples
of single-myotube measurements, such as traction force microscopy [106] or micro-
cantilever deflection interrogated with a laser-photodetector system [107, 108]. However,
57

these systems do not seem to increase culture lifetimes and render myotube-to-myotube
interactions difficult to assess [58]. To allow for a more complete investigation of
physiological skeletal muscle tissue functionality over significant time periods, we chose
to integrate MTF technology into our hydrogel, which can be run at moderate throughput
while still replicating skeletal muscle tissue alignment, despite lacking three-dimensional
architecture.
After peeling cantilevers from the polystyrene substrate, a basal stress exerted by
passive myotubes deforms each MTF in the absence of stimulation. Interestingly, there
were no large discrepancies in the basal contraction among any groups except for week
three hSKM MTFs, which suffered heavy delamination of myotubes. Thus, while stresses
were not normalized on a per cantilever basis, when paired with relatively stable nucleic
densities over time, the basal stress data suggest that differences in contractile stress
output are not solely a result of differences in cell number per MTF. Pacing MTFs at 2 Hz
allowed for observation of singular cantilever contractions, from which twitch stress was
quantified. Distinctly, chick myotubes generated increasing twitch forces, on average
increasing an order of magnitude from one to three weeks. In contrast, C2C12 and hSKM
MTFs could not maintain contractile stresses when paced, quickly deteriorating in
performance after two weeks due to delamination of myotubes and loss of sarcomeric
structure. The same trends are observed in tetanus stress generation of the MTFs. Chick
myotube MTFs could generate maximal contractile stresses in the 100s of kPa range, in
the same magnitude as documented specific forces generated by native human tibialis
anterior and soleus muscles; additionally, most previous studies utilizing tissue-
engineered fibers derived from human, murine or avian myoblasts have achieved specific
58

forces in the low tens of kPa at most [49, 61]. The tetanus-to-twitch ratio of our chick
MTFs also matched more closely to that of in vivo rat and human muscle tissue, while our
C2C12 and hSKM MTFs fall between values documented with other engineered versions
of muscle tissues [49]. Possible factors that have enhanced the contractile functionality
of our system with chick myotubes compared to past studies include the use of a naturally
derived gelatin substrate more adherent to cells, a more compliant substrate appropriate
for myogenic tissue, and the use of freshly isolated primary cells.
To elucidate the molecular changes that may be responsible for the improved
contractile function of engineered chick muscle tissues on gelatin hydrogels, we
performed RNAseq analysis after one and three weeks of culture to identify genes that
are differentially expressed at distinct stages of maturation. We observed downregulation
of endogenous myostatin (MSTN) and myogenic factor 5 (MYF5), which increases
expression of the master myogenic regulatory gene MYOD1 [98] and decreases myoblast
proliferation [109, 110], respectively. Thus, downregulation of these genes would be
expected to push myoblasts to exit the cell cycle, fuse, and differentiate into myotubes.
Expression of sarcomere proteins, such as α-actinin (ACT1), troponin I (TNNI1), and
myosin heavy chain (MYH1G), was significantly higher after three weeks of culture.
While ACT1 and TNNI1 expressions are expressed early in embryonic and neonatal
myotube development, the increased expression of MYH1G, an adult chicken myosin
isoform prevalent in fast-twitch type IIx fibers, is indicative of progression toward more
mature muscle phenotypes [79]. This is especially interesting because engineered
tissues often exhibit low proportion of fast-twitch fibers [111, 112]. The development of
adult slow-twitch fibers was also evidenced by upregulation of slow myosin heavy chain
59

isoform MYH7 [113]. However, other mature sarcomere markers, such as myomesins
(MYOM1, MYOM2, MYOM3) that are expected to be upregulated [79, 114] and non-
muscle myosins (MYH9, MYH10) that are expected to be downregulated [79], remain
largely unchanged, indicating that not all sarcomere proteins follow the expected
developmental trajectory in vitro. However, our Ingenuity Pathway Analysis still showed
that, overall, pathways for sarcomere organization are activated in chick muscle tissues
cultured for three weeks compared to one week.
We also investigated the expression of genes important for motor neuron survival
and integration. The expression of neurotrophic factors BDNF, CNTF, and GDNF was of
particular interest because they promote motor neuron differentiation and survival and
thus are often added to the differentiation media for hiPSC-derived motor neurons [63]. Of
these factors, BDNF, which has been shown to promote innervation of rat diaphragm
muscle post-injury [115], was significantly upregulated in chick muscle tissues cultured
for three weeks. BDNF is produced by skeletal muscle in response to contraction
[116] and thus the upregulation of BDNF could be caused by the increased contractility
of chick muscle tissues at three weeks. Higher levels of BDNF may also enhance lipid
oxidation, an adaptation of skeletal muscle to facilitate increased energy expenditure
[117]. However, the timing of BDNF expression is important to consider for motor neuron
integration, as BDNF also inhibits NMJ maturation after initial innervation [118]. As shown
by our Ingenuity Pathway Analysis, neurotrophins generated by muscle tissue, like BDNF,
inhibit AGRN expression [119], which is needed for activation of the LRP4-MUSK
complex and clustering of acetylcholine receptors to form mature NMJs [120]. In other
words, as previously observed [121], BDNF is needed for the growth phase of
60

neuromuscular tissues but not the synaptogenesis phase. For this reason, BDNF and
other neurotrophic factors are usually removed from culture media after one week of
coculture with muscle tissue. We also observed that FGF2, which encodes for fibroblast
growth factor 2 (FGF2), was upregulated in chick tissues at three weeks. FGF2 interacts
with neuronal receptors, such as fibroblast growth factor receptor 1, to slow axonal growth
and promote synapse formation by counterbalancing the growth promoting effects of
neurotrophins [122]. Some genes for key proteins in NMJ formation and maintenance
(AGRN, CHRNA1, and PLEKHG5) were downregulated in chick muscle tissues after
three weeks. However, this is not too surprising because agrin is primarily secreted by
motor neurons to induce acetylcholine receptor clustering [123] and the tissues used for
RNAseq were muscle tissues without motor neurons. Thus, the activation of genes
important for NMJ formation and stabilization in muscle is likely dependent on the
presence of motor neurons.
While our 2D platform generates well-formed, functional muscle tissue likely
conducive to motor neuron integration, it does a poor job recapitulating the native cell-cell
and cell-matrix environment of muscle fibers. Bulk skeletal muscle at the fascicle level is
made up of bundles of muscle fibers, which in turn are bunched together with connective
and fatty tissue [124]. The bundled structure provides an important structure for
interactions between cells, extracellular matrix, and their responses via signal
propagation to physical and chemical stimuli [125]. Thus, despite a generally much more
complicated fabrication and culture procedure [73], engineered 3D systems may better
recapitulate responses in disease modeling and drug screening studies.
61

The tissue architecture of our fibrin myobundle system yielded similar conclusions
about the utility of primary chick muscle as a reliable cell source for skeletal muscle tissue
engineering. H&E staining after one and two weeks of differentiation indicated hydrogels
compacted to a final density by one week. Tissues were roughly 5 times as dense as
those on gelatin hydrogels, which more closely resembles that of native tissue [126].
Myobundles also exhibited with a slightly lower but comparable myogenic index to 2D
cultures. Sarcomeres could be easily identified in immunostaining; however, the variation
in where myotubes are sectioned in 3D space made visualization of sarcomeres
inconsistent, and thus further analysis on pattern homogeneity was not performed.
Additionally, the bundled architecture preserves cell-cell and cell-matrix interactions that
are lost in 2D culture. In fibrin constructs, these 3D matrix interactions have been shown
to more dynamically regulate adhesion complexes, actin architecture, nuclear shape, and
tissue stiffness compared to traditional 2D culture [127]. The recapitulation of this
architecture may be more important for applications studying the behavior of bulk muscle
tissue at larger scales, such as chronic muscle wasting in disease or building muscle
tissues in the context of tissue regeneration or cultivated meats [128].
On a functional level, 3D myobundles generated higher twitch and tetanus forces
at two weeks of differentiation compared to one week, albeit not at a statistically significant
level. While these values are embryonic in presentation, similar to comparable past
studies [49, 61], evidence of increased stress generation is promising towards generating
more reliable functional models, perhaps with longer culture times. Similar to our 2D
cultures, myobundles did not receive any outside stimulation, which has been shown to
maintain and improve contractile function by shifting glycolytic and fatty acid metabolic
62

flux towards more physiological levels [129]. Thus, introducing stimulation through
electrical, mechanical, or biological (i.e., coculture with motor neurons) means may also
further improve contractile stress generation even further. It should also be noted that to
more directly compare contractile ability, detected forces would ideally be normalized to
DNA or protein content [60] or cross-sectional area [49], accounting for construct size.
The time dynamics of myobundle contraction also show statistically significant decreases
in both rise and fall time. This seems to support our hypothesis that a lack of stimulatory
activity leads to more glycolytic activity, and that fast-to-slow shifts with more energy
efficient ATP generation may occur given electrochemical signals or fatty acid
supplementation in media [130].
Ultimately, it is clear from tracking increases in twitch and tetanus stresses over
time in 2D and 3D culture that long-term maintenance is key for engineering skeletal
muscle tissues with more physiological contractile ability and overall clinical relevance.
While these experiments show the utility of chick myotube skeletal myotube cultures, the
platform can be modified or optimized for other cell sources, such as hiPSC-derived
myoblasts, further adding human relevance. Integration of other mechanically or
biologically engineered technologies, such as microfluidic bioreactors [131] or
optogenetic channels [132], or other supporting cell types, such as Schwann cells [133],
macrophages [134], and motor neurons (described in the next chapter) [81], will likely
pave the way for more effective and personalized therapies for debilitating
neuromyopathies.

63

Chapter 3 Coculture of Skeletal Myotubes and Motor Neurons to Engineer
Functional Neuromuscular Junctions
Aim 2: Coculture engineered skeletal muscle and hiPSC-derived motor neurons
to evaluate NMJ morphology and synaptic electrophysiology over time.

3.1. Introduction
Mechanistic studies into the degeneration of the NMJ have mainly relied on
transgenic animal models, such as the SOD1 mutant mouse for ALS [29]. These animal
models do not capture the entirety of the human ALS phenotype and pathogenesis.
However, skeletal muscle culture in vitro has possessed shortcomings preventing the
study of physiologically relevant NMJs outside of animal models. On the other side of the
basic motor unit, studies utilizing mutation-specific hiPSC-derived motor neurons have
allowed for studies with genetic specificity, clearly advantageous for studying
neuromyopathies possibly stemming from multiple mechanisms, like ALS [22].
Monocultures of motor neurons, however, have shown limited genetic and functional
maturation when compared with motor neurons cultured with myotubes [135, 136], and
are incapable of modeling neuromuscular signal transmission alone. Thus, we aimed to
develop a consistent coculture protocol of motor neurons and our 2D and 3D skeletal
muscle microphysiological systems, platforms that would allow for more efficient and
modular studies of various neuromyopathies.
With the more stable skeletal muscle tissue microphysiological systems described
in the last section, we now aim to develop protocols to engineer highly functional NMJs
in vitro by long-term coculture of hiPSC-derived motor neurons with chick myotubes. To
64

evaluate the maturity of the engineered control healthy NMJs, we quantify both structural
and functional properties. Specifically, we looked at myotube architecture, the distribution
of AchR, and presynapse protein synapsin was tracked over time, as well as the percent
of clusters that are innervated. Our data in 2D or 3D show that over several weeks of
coculture, premature NMJs form with some organized structure and functional ability to
propagate signals and stimulate muscle contraction as measured with sharp-electrode
electrophysiology. Successfully engineering NMJs in neuromuscular microphysiological
systems with structural and functional maturity will significantly increase the utility of in
vitro systems to supplement current neuromyopathy animal models and preclinical testing
of drugs.
3.2. Materials and Methods
3.2.1. 2D Coculture of Skeletal Muscle and Healthy hiPSC-derived Motor Neurons
We performed coculture (up to three weeks) of C2C12, hSKM, and chSKM with
hiPSC-derived motor neurons to engineer functionally NMJs on 2D gelatin hydrogels. The
lifetime limits of conventional skeletal muscle culture prevent long-term culture of in vitro
neuromuscular models, which may explain immaturity of NMJ structure or
electrophysiology in previously described models [59, 65, 66]. Briefly, to generate hiPSC-
derived motor neurons, human lymphocytes attained from the NINDS Biorepository at the
Coriell Institute for Medical Research were reprogrammed into iPSCs as described
previously, with the use of episomal vectors containing Oct4, Sox2, Klf4, L-Myc, Lin28,
and a p53 shRNA [22, 137]. Then, OLIG2-positive motor neuron progenitors in the form
of embryoid bodies were generated and expanded from the iPSCs in a chemically defined
neural medium with a small molecule cocktail (see chick differentiation medium from
65

Table 2-1). Subsequent enrichment of the progenitors into functional motor neurons
(>90%) occurred over another 16 days with the use of a Notch inhibitor (compound E)
and activation of Hedgehog signaling (purmorphamine) [22, 63]. Three to four days after
inducing differentiation in myotube monocultures, motor neuron embryoid bodies
differentiated for 31 days were slowly dissociated mechanically with a P1000 manual
pipettor, and dissociated neurons were seeded dropwise at a density of 750,000 cells per
substrate. Cocultures were maintained in chSKM differentiation media supplemented with
10 μM Rho-associated protein kinase inhibitor (Selleck, S1049) (removed after one day)
and 10 ng/mL BDNF, CNTF and GDNF for one week (removed after one week) [63, 78].
Media was refreshed every other day.
3.2.2. 2D NMJ Structural Characterization
Tissues were fixed using ice-cold methanol for ten minutes at room temperature
and incubated with antibodies for synapsin-1 (Cell Signaling Technology, D12G5, 1:200),
followed by goat anti-chicken antibody conjugated to Alexa Fluor 488, goat anti-mouse
antibody conjugated to Alexa Fluor 546 or α-bungarotoxin conjugated to Alexa Fluor 555,
goat anti-mouse antibody conjugated to Alexa Fluor 633, and DAPI (all 1:200). Samples
were imaged using the same confocal microscope and settings as described for
monocultures in section 2.2.3.
The area and co-localization of synapsin-1 and bungarotoxin were analyzed with
custom NIS Elements AR Analysis 5.02.00 macros [33]. Motor neuron axons and clusters
of acetylcholine receptors were defined by intensity thresholding of the synapsin-1 and
bungarotoxin stains, respectively. For the synapsin-1 stain, area masks were created to
66

fill holes. Data are reported as absolute area or proportion of the area in a field of view.
Each data point represents two fields of view averaged per coverslip.
3.2.3. 2D NMJ Electrophysiology
Muscle tissue cultures were rinsed and resuspended in Tyrode's solution with
20 μm blebbistatin to prevent spontaneous or neuron-induced muscle contraction that
would disrupt the measuring electrode with motion artifacts. Sharp electrode (electrode
resistance between 10 and 20 MΩ) intracellular current-clamp recordings were performed
in individual myotubes at room temperature with an Olympus BX61 WI microscope using
a 40×/0.80 NA water-dipping objective and acquired using an Axoclamp 900 A amplifier,
Digidata 1440 A acquisition system, and pClamp 10.5 software (Molecular Devices)
[138]. Sweeps were digitized at 10 kHz and filtered at 1 kHz. Miniature excitatory
postsynaptic potentials (mEPSPs) were recorded in the absence of stimulation for 5–
10 min. Individual mEPSPs were selected manually by detecting signals over a set noise
threshold that fit an mEPSP waveform. If present, muscle action potentials generated
from spontaneous currents were ignored based on their characteristic waveform, which
is quite distinct from mEPSPs [139, 140]. From at least six mEPSPs per myotube,
amplitude, frequency, and rise and decay time constants (τ) were quantified. The resting
membrane potential of each myotube was also quantified. Data were analyzed using
Clampfit (Molecular devices), MiniAnalysis (Synaptosoft), Excel (Microsoft), and
SigmaPlot (Systat) software. Each data point represents a recording from one myotube.
3.2.4. Seeding and Culturing Innervated 3D Myobundles
The seeding protocol is summarized in Figure 2-2b. Five motor neuron organoids
were mechanically dissociated 20 times each using a P1000 and P200 pipette tip, in
67

succession. Then, the motor neurons were added to the chick myoblast solution
described in section 2.2.7, and seeding protocols remained as described. The fibrin
hydrogel was allowed to solidify at 37 °C for 30 minutes, and chick GM with the addition
of 1.5 mg/mL aminocaproic acid (ACA, Sigma Aldrich), and 10 ng/mL of BDNF, CNTF
and GDNF were added to submerge myobundles. Media was exchanged after 2 days,
and after 4 days in the GM with ACA and neurotrophic factors, media was switched to
DM with ACA and neurotrophic factors and cultured on a rocker for increased diffusion of
nutrients. Media was then exchanged every other day, with neurotrophic factors removed
following one week of differentiation, as previously described.
3.2.5. 3D NMJ Structural Characterization
Myobundles were washed with PBS 3 times and then fixed using ice-cold methanol
for twenty minutes at 4 °C. Following 3 PBS rinses at 5 minutes each, myobundles were
resuspended in 30% sucrose in PBS and allowed to sit for a minimum of 12 hours to
prevent ice crystal formation during freezing. Cryosectioning and immunostaining was
performed as described in section 2.2.8 and 3.2.2, respectively.
3.2.6. Innervated Myobundle Contractile Force Generation
Contractile force recordings and subsequent analysis were performed using the
same methods described in section 2.2.9. In the absence of electrical stimulation, the
impact of neurotransmitter on bulk innervated muscle contraction was also performed by
adding 50 μM of glutamate to stimulate motor neurons to release neurotransmitter.
3.2.7. Statistical Analysis
The data were tested for normality and analyzed using student’s t-test/Mann-
Whitney (GraphPad Prism 7.04), as appropriate. Comparisons with p-values less than
68

0.05 were considered statistically significant. In our figures, significance bars are only
drawn to signify statistically different groups between cell sources at the same time point
or at different time points for a single cell source. In figures, * denotes p < 0.05, ** p <
0.01, *** p <0.001, **** p < 0.0001, unless otherwise noted.
3.3. Results
3.3.1. Synaptic Structure of Cocultures on Gelatin Hydrogels
Chick tissues spontaneously contracted for several days following motor neuron
seeding before contracting sporadically. No spontaneous contractions were observed in
C2C12 and human cocultures at any timepoint. After one week of coculture, motor
neurons extended axons onto myotubes from all myoblast sources (Figure 3-1a-c). To
evaluate NMJ formation, we quantified the area and co-localization of synapsin and
bungarotoxin as pre- and postsynaptic markers, respectively. Synapsin area was similar
for all tissues after one week (Figure 3-1e), indicating similar levels of motor neuron
adhesion and spreading. In contrast, the area (Figure 3-1f) and cluster size (Figure 3-
1g) of bungarotoxin were significantly higher for chick tissues compared to C2C12 and
human tissues. Bungarotoxin clusters were barely present in muscle tissue monocultures
for all myoblast cell sources [81], suggesting that formation of these structures is
promoted and maintained by motor neurons. However, even in cocultured tissues,
bungarotoxin clusters did not strongly co-localize with synapsin for any muscle tissue after
one week (Figure 3-1h), indicating poor NMJ formation. To promote NMJ formation, we
maintained cocultured tissues from all myoblast sources for three weeks. However,
delamination of C2C12 and human myotubes prevented the survival of cocultured
tissues, and thus analysis of NMJ formation, beyond one week. In contrast, chick muscle
69

tissues cocultured with hiPSC-derived motor neurons were stable for three weeks and
displayed evidence of continued NMJ maturation (Figure 3-1d), including increased
synapsin area (Figure 3-1e) and increased co-localization of synapsin and bungarotoxin
clusters (Figure 3-1h). The area and cluster size of bungarotoxin slightly decreased from
one to three weeks (Figure 3-1f, g), which is expected as non-innervated clusters
dissociate.

70

Figure 3-1: Structure of NMJs formed between engineered muscle tissues and hiPSC-derived motor
neurons. hiPSC-derived motor neurons cocultured with (a) C2C12, (b) human, and (c) chick muscle tissues
for one week and (d) chick muscle tissue for three weeks. α-actinin (red), DAPI (blue), synapsin-1 (green),
and bungarotoxin (white). Scale bar, 50 μm. Percentage of (e) synapsin and (f) bungarotoxin area per field
of view. (g) Size of individual bungarotoxin clusters. (h) Percentage of bungarotoxin clusters co-localized
with synapsin. *p < 0.05; **p < 0.01; ***p <0.001; ****p < 0.0001 [81].
71

3.3.2. Synaptic Activity of Cocultures on Gelatin Hydrogels
To evaluate synaptic activity, we performed intracellular sharp electrode
electrophysiology experiments [33, 141] to detect miniature excitatory postsynaptic
potentials (mEPSPs) in chick muscle tissues cocultured with hiPSC-derived motor
neurons (Figure 3-2a). Blebbistatin, a myosin inhibitor, was applied to the tissue to cease
spontaneous contractions and motion artifacts that would obscure motor neuron
signaling. mEPSPs were more frequent in tissues cocultured for three weeks compared
to one week (Figure 3-2b), without a significant change in amplitude (Figure 3-2c). The
average rise time of mEPSPs also did not change from one week to three weeks (Figure
3-2d) but decay time significantly increased (Figure 3-2e). Resting membrane potential
Figure 3-2: Synaptic activity of NMJs formed between engineered muscle tissues and hiPSC-derived motor
neurons. (a) Recordings from representative myotubes after one and three weeks of coculture with hiPSC-
derived motor neurons. (b) mEPSP frequency, (c) amplitude, (d) rise time constant, and (e) decay time
constant as a function of coculture time. (f) Membrane potential of myotubes used for recordings. * denotes
p < 0.05 [81].
72

decreased slightly from one week to three weeks (Figure 3-2f) approaching physiological
levels of approximately −70 mV [142]. Together with the structural analysis, these data
indicate that improving both the maturity and culture lifetime of the engineered muscle
tissue enhanced the development of NMJs with hiPSC-derived motor neurons.
3.3.3. Morphology of Innervated Myobundles
To investigate formation of NMJs in 3D space, we seeded myobundles with motor
neuron organoids mixed into the fibrin hydrogel solution. After a week of coculture,
innervated myobundles did not compact as much as myobundles cultured in the absence
of motor neurons. Indeed, the width of myobundles seeded with motor neurons was over
100 μm thicker on average (Figure 3-3). From the H&E staining, bundles were visually
less dense after a week of coculture. It is possible that clusters of motor neurons
prevented immediate and efficient packing of skeletal muscle fibers, despite the tensile
forces applied as the hydrogel compacts around the PDMS anchor rods. However, by two
weeks, monoculture and coculture myobundles did not show significantly different widths,
although the clusters do appear to disrupt muscle alignment somewhat, as highlighted in
the region marked with the dotted yellow line (Figure 3-3).
Figure 3-3: Hematoxylin and eosin stains of innervated 3D myobundles after one and two weeks of
differentiation. Scale bar = 200 μm. *p < 0.05
73

Immunostaining confirmed the presence of well-striated chick myotubes in
innervated myobundles after 1 and 2 weeks of coculture (Figure 3-4). Nuclear staining
also confirmed slightly lower nuclei density in cocultures compared to monocultures after
a week, although not to a statistically significant degree. By the two-week timepoint, there
is little difference in nuclei density between myobundles with or without motor neurons.
Interestingly, the myogenic index is higher at each weekly timepoint in cocultures relative
to their monoculture counterparts, indicating better fusion. Furthermore, the extension of
axons as visualized through the presynapse marker synapsin suggested that axons can
traverse through the myobundle matrix and that a larger proportion of myotubes may be
innervated.
74

Figure 3-4: 3D myobundle morphology after 2 weeks of coculture with motor neurons. (a) Representative
myobundle cryosections. α-actinin (red), synapsin (green), DAPI (blue). Scale bar = 200 μm (left), 50 μm for
zoomed in images (right). (b) Number of nuclei normalized to field of view area, (c) myogenic index at weekly
timepoints, and (d) synapsin area coverage per field of view. *p < 0.05; **p < 0.01

75

3.3.4. Contractile Activity of Innervated Myobundles
Innervated myobundles spontaneously contracted after a few days following
seeding, which continued throughout the culture period. This was in sharp contrast with
monoculture myobundles, which stopped contracting spontaneously after several days.
To characterize the contractile force generation of innervated myobundles, bundles were
prepared and stimulated at 2 and 20 Hz to generate twitch and tetanus stresses, as
previously described. Interestingly, no significant changes were observed in the twitch or
tetanus forces with longer time in culture, nor was there a noticeable change in tetatnus-
to-twitch ratio (Figure 3-5a-c). However, compared to monocultures, twitch and tetanus
forces were significantly higher at corresponding timepoints, and the tetanus-to-twitch
Figure 3-5: Myobundle contractile force characterization over 2 weeks of monoculture or coculture.
Electrically stimulated (a) twitch and (b) tetanus forces, (c) tetanus-to-twitch ratio, and (d) rise and (e)
fall τ. *p < 0.05; **p < 0.01, ***p < 0.01
76

ratio was also slightly elevated. Thus, the presence of motor neurons appears to increase
the contractile ability of myobundles.
The time dynamics of contractions also showed interesting differences between
myobundles with or without motor neurons. In monoculture, the rise and fall time
constants noticeably drop from one to two weeks of culture (Figure 3-5d, e). However,
this drop is not clearly observed in innervated myobundles. These differences in
contraction speed and relaxation suggest differences in muscle fiber phenotype may be
present in the myobundles. Innervated myobundles appear to exhibit a slower response
characteristic of slow-oxidative fibers, while monoculture myobundles tended to adopt a
fast-glycolytic phenotype with quicker dynamics [143].
In the presence of 50 mM glutamate, which has been used to stimulate motor
neurons to fire in coculture, we also observed uncoordinated contractions at various
regions on the myobundle. These contractions were not observed in monoculture
myobundles. These results further suggest the formation of functional NMJs in our fibrin
myobundle platform. However, uncoordinated contractile performance may suggest
uneven distribution of innervation, or uneven diffusion of glutamate into the myobundle.
Further analysis of the NMJ structure is required to confirm these either hypothesis.
3.4. Discussion
Patient-specific modeling of the NMJ in vitro has been limited by the stunted
maturation of NMJs formed by hiPSC-derived motor neurons and engineered muscle
tissues [59, 65, 66, 69]. Here, we attempted to improve the structure and function of NMJs
formed by hiPSC-derived motor neurons by optimizing the maturity of the engineered
muscle tissue in 2D or 3D cultures. In prior attempts to coculture muscle tissues and
77

primary [65, 144], embryonic stem cell-derived [103], or iPSC-derived [69, 145-
147] motor neurons, NMJs were relatively immature, exhibiting low co-localization of pre-
and postsynaptic markers [59, 65, 144, 148] and low frequency of spontaneous or
induced synaptic activity [59, 65, 66] compared to native muscle [149, 150].
In our cocultured tissues on gelatin hydrogels, hiPSC-derived motor neurons
projected axons onto myotubes and several myotubes exhibited clusters of acetylcholine
receptors after one week, a pattern observed during embryonic development [103].
However, similar to previous in vitro approaches, most clusters of acetylcholine receptors
did not co-localize with axons in any cocultured tissues after one week, indicating
relatively immature NMJs. Cocultured C2C12 and human tissues detached prior to the
three-week timepoint, preventing NMJ maturation. In contrast, hiPSC-derived motor
neurons cocultured with chick tissues continued to extend axons that increasingly co-
localized with acetylcholine receptors over three weeks. In parallel, clusters of
acetylcholine receptors lacking innervation dissipated, likely because of spontaneous
contraction of the chick myotubes. This trend is consistent with reduced acetylcholine
receptor clusters in contracting muscle fibers from three-week-old spinal cord explant-
muscle cocultures [151]. A similar phenomenon has also been observed in NMJ formation
during embryonic development [103]. Spontaneous contractions during early muscle
tissue differentiation have also been shown to precede striation of Drosophila muscle
fibers [152] and are a key predictor of advanced Z-line development and maintenance of
rat myotubes in vitro [153].
To quantify synaptic activity, we performed electrophysiology experiments to
detect mEPSPs in chick myotubes cocultured with hiPSC-derived motor neurons on
78

micromolded gelatin hydrogels. mEPSPs are generated by spontaneous neural activity
and release of neurotransmitter into the NMJ, as shown in Drosophila [33, 141] and spinal
cord explant-muscle cocultures on gelatin [151]. At the one-week timepoint, our
measurements were similar to un-aligned, nine-day cocultures of C2C12 myotubes and
mouse embryonic stem cell-derived motor neurons [66]. However, at the three-week
timepoint, the amplitude and frequency of mEPSPs increased, approaching levels
observed in Drosophila [33, 141], mice [150] and spinal cord explant-muscle cocultures
[139]. The mEPSP decay time also increased from one to three weeks, consistent with
an increase in acetylcholine receptors on the myotube membrane that increase the time
required for acetylcholine to reach equilibrium, be degraded by acetylcholinesterase, and
be reabsorbed into the pre-synaptic terminal. Finally, the membrane potential of
innervated myotubes in our system ranged from −40 to −70 mV, roughly the same as
myotubes cocultured with spinal cord explants[139] and approaching the adult human
value of −70 mV [142]. Thus, our in vitro system recapitulated some key steps in the
native development of NMJs. Our platforms may serve as a useful tool for investigating
NMJ function or dysfunction, compared to traditional in vitro models.
The 2D gelatin hydrogel platform maintained cocultures very well, but as described
in Chapter 2, there 3D myobundles better recapitulate the cell-cell and cell-matrix
interactions expected in native tissue. Like monoculture myobundles, innervated
myobundles show tissue densities more closely resembling native tissue [126]. In our
study, we saw a significant increase in 3D myobundle myogenic index when cultured with
motor neurons. Other groups have not seen a large increase in myogenic index in the
presence of motor neurons [144], but the final myogenic index reported is similar. The
79

appearance of strongly striated myotubes showed great promise that chick fibrin
myobundles generate well-developed muscle tissue in the presence of motor neurons.
While attempts were made to quantify this difference in comparison to monoculture
myobundles, the density of the muscle tissue made it difficult to differentiate myotubes
from each other, preventing accurate measures of myotube width, as was performed in
2D. Furthermore, sarcomere patterning analysis was made difficult due to uneven
expression of α-actinin in sections, due to the more varied spatial distribution of the 3D
tissue. These challenges are common in 3D tissue engineering approaches and are a
disadvantage that must be considered when choosing an in vitro platform.
Other groups have shown evidence of increased contractile force generation when
muscle is regularly stimulated through electrical [154] or biological (neuronal) [155]
means, which we also demonstrated in our study. The inclusion of such stimulation may
be necessary to generate skeletal muscle mature enough for clinical relevance. The
development of adult myosin fiber types in particular is highly dependent on the
modulation of various stimulatory parameters [154-156], including the frequency of
electrical stimulation or the number of motor neurons/extent of innervation. Mechanical
stretch may also play important roles in recapitulating proper tissue architecture, including
enhanced myotube fusion [157] and axonal projection extension [158]. Innervation of
muscle tissue has also been shown in vitro to shift muscle fiber phenotype towards a
slow-oxidative profile [130]. Indeed, the contraction rise and fall time constants observed
show evidence of this trend since they are slower than monoculture contraction times.
Comparing the RNAseq data from our 2D gelatin hydrogels chick myotube cultures to
native chicken tissue (unpublished data), we saw a lack of innervation resulted in more
80

glycolytic tissues compared to physiological baselines. This indicates the potential for our
engineered systems to model processes with fiber type bias, such as aging [159], various
disease [143], and exercise [160] with greater physiological relevance than traditional in
vitro cultures.
It should be noted that our systems do fall short in capturing several other important
physiological aspects, such as myelinated axon fibers that are in spinal cord explants due
to the presence of glial cells [151, 161]. To address this, iPSC-derived glial cells [162] and
astrocytes [163] could potentially be integrated into our 2D or 3D systems to advance the
maturity and relevance of the NMJs. Additionally, electrophysiology experiments in the
presence of drugs known to effect NMJ activity, such as the acetylcholine receptor
antagonist tubocurarine or excitatory neurotransmitter N-methyl-D-aspartic acid (NMDA)
[164, 165], should be performed to verify appropriate physiological responses.
Homogeneity of motor neuron distribution in the 3D fibrin hydrogel model may also be
improved by incorporating elements conducive to muscle, motor neuron, and/or NMJ
development. These include extracellular matrix proteins and proteoglycans like laminin,
vitronectin, or agrin [65, 166-168]. Overall, our more reliable coculture models open the
door for more complex and physiologically accurate studies into neuromuscular tissue
structure and function.

81

Chapter 4 Modeling Muscular Dystrophy Using Human-Derived Cells on
Micromolded Gelatin Hydrogels
Aim 3: Translate engineered skeletal muscle MPSs to characterize mutation-
specific disease phenotypes and responses to potential therapeutics

4.1. Introduction
Although chick muscle tissues were optimal for skeletal muscle maintenance, NMJ
formation, and contractile and synaptic function in the tissue engineering work presented
so far, a major drawback is their non-human origin. The commercial human myoblasts
tested in previous chapters exhibited relatively low myotube formation and stability,
precluding robust sarcomere patterning in monoculture and NMJ formation in coculture.
These cells were cryopreserved and reported to have relatively low myoblast purity by
the vendor. Freshly isolated myoblasts from patient biopsies [169] or myoblasts subjected
to rigorous purification [170] may generate myotubes with higher levels of structural and
functional maturity while maintaining mutation-specific responses. Regardless of the
source or culture procedure, primary human myoblasts are relatively inaccessible to many
researchers, are generally collected in low quantities, have limited passage lifetimes, and
are susceptible to patient-dependent variability [60], limiting scalability. More recently,
protocols for differentiating myoblasts from hiPSCs have been established [44], although
they tend to suffer from low purity or yield [171] and limited maturity [172], which would
likely render limited clinical significance in studying myopathies.
Perhaps a simpler approach to achieve a human-relevant muscle source than
hiPSC differentiation is to reprogram dermal fibroblasts directly to myoblasts by induced
82

expression of MyoD, a master regulator of muscle cell development. Other studies have
expressed an inducible MyoD construct in dermal fibroblasts partially immortalized
through expression of human telomerase reverse transcriptase (hTERT), termed iDRM
(induced directly reprogrammed myotubes), which can be expanded as fibroblasts and
then reprogrammed to myoblasts and multi-nucleated myotube-like cells upon MyoD
induction [173-177]. The utility of iDRM has been documented, having generated iDRM
from patients with Duchenne muscular dystrophy (DMD) [177], a myopathy where the
crucial structural protein dystrophin loses functionality due to frameshift mutations. These
studies evaluated exon skipping strategies aimed at dystrophin rescue by reframing the
RNA through exclusion of an additional exon induced by treatment with specific antisense
oligonucleotides, hereby referred to as AO [177]. However, while low levels of rescued
dystrophin protein were detected for certain DMD patient lines, myotubes were randomly
organized and could not survive longer than a week, which may not be sufficient to detect
rescue [177]. Our engineered platforms with better maintenance ability could be applied
to study mutation-specific disease presentation and potential rescue with more accuracy.
Translating our platforms capable of long-term muscle tissue maintenance to
culture more human-relevant, mutation-specific tissues would be a key step in
demonstrating our systems’ utility in generating more clinically relevant data in disease
modeling and therapeutic screening studies. Therefore, in this chapter, we will show the
ability of our gelatin hydrogels to model mutation-specific Duchenne muscular dystrophy
and Limb Girdle 2A/R1 muscular dystrophy phenotypes [178]. Demonstrating the ability
of system to show nuances between patient-derived cells serve as evidence that our
approach has the potential to yield clinically significant information.
83

4.2. Materials and Methods
4.2.1. Inducible Directly Reprogrammable Myotube (iDRM) Derivation and Dermal
Fibroblast Characteristics
MyoD reprogrammed skin punch fibroblasts (iDRM) were derived from three DMD
subjects [179, 180]. As reported, iDRM derived from DMD subject 1003, who lost
ambulation at 13.5 years, has an out of frame deletion of DMD exons 46-51 and is
predicted not to express the dystrophin protein [180]. DMD subject 1023 has an in-frame
deletion of DMD exons 3 to 23 with sub-typical, albeit significant, sarcolemmal dystrophin
protein expression [179] and remains ambulatory at age 20. Subject 1015 lost ambulation
at age 15 and harbors a deletion of DMD exon 45, low levels of self-corrected exon 44
“skipped” mRNA in derived iDRM [180], and an increased frequency of clusters of
dystrophin-positive revertant fibers. Muscle biopsies were only available from subjects
1015, 1023, and a healthy 20-year-old female volunteer who served as a positive control
for normal dystrophin expression and localization.
MyoD reprogrammed skin fibroblasts (iDRM) from two LGMD2A/R1 were
established for this study. Subject 1077 is a 52-year-old male who lost ambulation at 44
years old with compound heterozygous mutations: c.550delA (p.T184Rfs*36) and
c.1342C>G (p448R>G) in CAPN3. Both mutations are predicted to be pathogenic
(https://databases.lovd.nl/shared/genes/CAPN3). Subject 1081 is a 37-year-old female
presenting a homozygous deletion of CAPN3 exons 17 to 24 (c.(1914+1_1915-
1)_(*544_?)del). Deletion of these exons leads to a short truncated non-functional CAPN3
protein [181]. All these variants leading to LGMD2A/R1 have been previously described
in other patients [181-183].
84

Skin punches and muscle biopsies were obtained with informed consent from
patients of the Center for Duchenne Muscular Dystrophy (CDMD) at UCLA under
University of California Los Angeles IRB ‐approved protocols #11 ‐001087 and #18-
001547. Open or needle skeletal muscle biopsies were performed and processed as
previously described [184, 185]. Muscle samples were sectioned at 10-micron thickness
and stained with hematoxylin and eosin for global histology assessment or
dystrophin/laminin using standard immunohistochemistry [184], where dystrophin NCL-2
(Leica, Buffalo Grove, IL, USA) at a dilution of 1:50 and laminin L9393 (Sigma, St. Louis,
MO, USA) at a dilution of 1:25 were used. Digital histological images were acquired using
standard light and fluorescence using an Axioplan 2 microscope (Carl Zeiss Inc, USA).
Pictures were then processed with Axiovision software (Zeiss) and/or ImageJ software.
4.2.2. iDRM Culture
iDRM were generated from dermal fibroblasts cultured from skin punches of 3mm
diameter obtained from each participant. After the establishment of a fibroblast culture,
cells were immortalized using a lentivirus encoding hTERT and subsequently infected
with a lentivirus encoding a tamoxifen-inducible MyoD to allow commitment to skeletal
muscle lineage as, we have previously reported for derivation of DMD 1023 and DMD
1015 iDRM [177, 179] (Figure 4-1a).
iDRM were cultured as previously described, modified by plating on coverslips with
micromolded gelatin hydrogels (described in section 2.2.1) at the time of fibroblast plating,
prior to MyoD induction (Figure 4-1b) [186]. Briefly, cells were kept in fibroblast growth
media (DMEM [+ phenol red, high glucose] [Thermo Fisher Scientific, Grand Island, NY,
USA] + 15% fetal bovine serum [Omega Scientific, Tarzana, CA, USA] + 1% nonessential
85

amino acids [Thermo Fisher Scientific, Grand Island, NY, USA] + 1%
penicillin/streptomycin [Thermo Fisher Scientific, Grand Island, NY, USA]). To induce
differentiation to myotubes, cells were first incubated in fibroblast growth media containing
5uM of 4-OH tamoxifen 4OH-tamoxifen (Sigma, St. Louis, MO, USA; dissolved in ethanol)
for 48 h. On day 3, cells were washed in PBS (Thermo Fisher Scientific), and fusion media
containing 1 mM 4OH-tamoxifen was added (1:1 Ham’s F-10:DMEM [phenol red free,
high glucose], 2% horse serum [Thermo Fisher Scientific, Grand Island, NY, USA], 2%
insulin-transferrin-selenium) (Thermo Fisher Scientific, Grand Island, NY, USA). During
the first week of differentiation, SB431542 (a TGF beta inhibitor) was added at a final
concentration of 5uM. All lines were kept for up to 2 weeks in fusion conditions prior to
analysis and media was changed every other day (Figure 4-1b).
4.2.3. Antisense Oligonucleotide (AO) Transfection
On day 7 (week 1) or 14 (week2) of differentiation, iDRM cells (1015 or 1003) were
transfected with 25nm or 250nM of 2-O-methyl AO targeting exon 44 (using H44A-
TGTTCAGCTTCTGTTAGCCACTGA and 45 (using H45A -
CCAATGCCATCCTGGAGTTCCTGTAA)[187], respectively, using oligofectamine
(Thermo Fisher Scientific) transfection reagent according to our previously published
protocol [177].

86

4.2.4. RNA Isolation and PCR
On the day of analysis, duplicate or triplicate coverslips from each condition were
removed from the well and the remnant cells growing on the side of the coverslips were
scraped directly and combined in Trizol (Thermo Fisher Scientific). Total RNA was
Figure 4-1: Overview of the experimental design and culture timeline. (a) A skin punch of 3mm diameter
is performed on the forearm and the tissue dissociated to obtain fibroblasts. Following infection with
lentiviruses encoding hTERT and MyoD, cells (renamed induced directly reprogramed myotubes – iDRM)
can form myotubes upon induction of MyoD and fusion inducing media. Polystyrene coverslips are
plasma-treated to enhance gelatin hydrogel adhesion prior to molding of the surface using soft-
lithography. Hydrogels are rehydrated and stored in PBS prior to cell seeding. (b) A typical timeframe for
the generation of myotubes from iDRM with or without AO treatment. Analysis for experiments was
performed after one (Day 9) or two (Day 16) weeks of differentiation. Adapted from [178].
87

isolated using the Purelink RNA mini kit (Thermo Fisher Scientific). Exon skipping
analysis was performed accordingly to our previously published protocol and skipping
efficiency is indicated by the ratio of skipped mRNA transcript over the total mRNA
transcripts (skipped plus unskipped) [177]. For both cell lines, a nested PCR was
performed to amplify the targeted DMD region: between exons 43 and 52 using previously
described primers for cell line CDMD1003 (Ex42-o, 5’-GTCCGTGAAGAAACGATGATG-
3’ + Ex53-0, 5’-CTCCGGTTCTGAAGGTGTTC-3’ and Ex43-i, 5’-
TCTCTCCCAGCTTGATTTCC-3’ and Ex52-i, 5’- TCTAGCCTCTTGATTGCTGG-3’), or
between exons 42 and 46 (Ex42-o, 5’-CAATGCTCCTGACCTCTGTGC-3’ + Ex46-o,
5‘GCTCTTTTCCAGGTTCAAGTGG-3’ and Ex43-i, 5’-
GTCTACAACAAAGCTCAGGTCG-3’ + Ex46-i, 5’-
GCAATGTTATCTGCTTCCTCCAACC-3’) for cell line CDMD1015.
4.2.5. Immunofluorescence Staining and Microscopy
iDRM were seeded at 350,000 cells onto micromolded gelatin hydrogel coverslips
in 12 well plates and MyoD expression was induced with tamoxifen as described above.
During all procedures, cells were kept at room temperature unless specified otherwise.
After myotube formation, cells were fixed with acetone for dystrophin staining or ice-cold
methanol for α-actinin staining. Primary antibodies were incubated overnight at 4 °C:
dystrophin NCL-2 (Leica, Buffalo Grove, IL, USA) at a dilution of 1:100 and α-actinin
(Sigma, St. Louis, MO, USA) at a dilution of 1:250. Secondary antibodies were used at a
dilution of 1:500 goat anti-mouse IgG (SA5-10173) and goat anti-rabbit (35553) from
ThermoFisher Scientific for 1 h. Coverslips were mounted in ProLong gold antifade with
DAPI (Thermo Fisher Scientific). Confocal fluorescence microscopy was performed using
88

a Confocal Module Nikon C2 with 20X air or 60X oil objectives. Z-stacks were acquired
(step size: 1 μm) and average intensity projections were used for data analysis on ImageJ
(NIH).
To quantify myogenic index, CellProfiler was first used to mask α-actinin signal.
Then, the proportion of total nuclei in masked areas that contained at least 3 nuclei was
taken as the myogenic index; we chose to count areas with at least 3 nuclei as stringent
filter of multi-nucleated myotubes. The myotube coverage was defined as the area
percentage of positive α-actinin signal after auto-thresholding using ImageJ. As a proxy
for myotube width, we used an automated CellProfiler calculation to define a minor axis
length based on the ellipse encompassing a mask of α-actinin that defined a single
myotube. . To quantify the degree of myotube alignment based on α-actinin staining, the
ImageJ plugin OrientationJ was used with a Gaussian filter window of σ = 10 pixels to
calculate the coherency and orientation angles at each pixel. Then, a global orientation
order parameter (OOP) was calculated from the orientation angles of at least three
images per sample, as previously described [188, 189].
4.2.6. Statistical Analysis
Student’s t-test were performed between week 1 and week 2 groups for each
patient iDRM lines or between them, or between AO and non-treated patient iDRM lines.
Comparisons with p-values less than 0.05 were considered statistically significant.
89

4.3. Results
4.3.1. Engineering DMD and LGMD2A/R1 iDRM myotubes on micromolded gelatin
hydrogels
To engineer mutation-specific DMD muscle tissues in vitro, we isolated dermal
fibroblasts from DMD 1015, DMD 1023, DMD 1003 and one control subject (1001). All
fibroblasts were engineered to iDRM by expressing a tamoxifen-inducible MyoD and
hTERT (Figure 4-1a), as described previously [180, 186]. iDRM were seeded on gelatin
hydrogels micromolded with 10 µm-wide alternating grooves affixed to coverslips (Figure
4-1a). iDRM were then triggered to become myoblasts with induction of MyoD 1-2 days
after seeding and maintained in differentiation media for one to two weeks to induce fusion
into myotubes (Figure 4-1b).
To characterize the morphological events characteristic of muscle development,
we co-visualized nuclei and sarcomeric α-actinin in engineered iDRM tissues after one
and two weeks in culture. All cell lines formed multi-nucleated, aligned myotubes at both
timepoints (Figure 4-2), although to different degrees. To compare this, we quantified
myotube coverage (α-actinin positive proportion), myogenic index (proportion of nuclei in
α-actinin positive myotubes with more than 3 nuclei), myotube width (minor axis length),
and myotube alignment (Figure 4-3). Healthy 1001 iDRM formed myotubes with stable
sarcomeric α-actinin coverage and myogenic index throughout the two-week culture
period. Organized striations were detectable at week 1 and remained constant through
week 2. Myotube width peaked in the first week and declined slightly by week two, likely
due to detachment of some of the most mature myotubes. Myotube alignment for healthy
90

iDRM was also high and indicated successful tissue patterning by the micromolded
gelatin hydrogels.
DMD 1015 iDRM performed relatively similar to the healthy 1001 iDRM, in terms
of sarcomeric α-actinin coverage (Figure 4-3a), myogenic index (Figure 4-3b), and
myotube width (Figure 4-3c), with a potential modest delay in the time to peak myotube
width relative to healthy control. Similar to the healthy control, striations in DMD 1015
myotubes were also detectable (Figure 4-2). We attempted to quantify α-actinin
striations, but the heterogeneity of the cells precluded a systematic and meaningful
analysis. DMD 1023 and DMD 1003 iDRM also differentiated into multi-nucleated
myotubes with near-typical myogenic index and myotube width. These results are similar
to previous studies that have shown relatively unimpaired fusion in hiPSC-derived
myoblasts from DMD [190] or LGMDR9 [191] patients. However, striations in DMD 1023
and DMD 1003 myotubes were qualitatively impaired relative to healthy 1001 myotubes.
Likewise, myotube coverage was lower in DMD 1023 and DMD 1003 compared to healthy
1001.
91

Myotube alignment was also high for the healthy and DMD myotubes (Figure 4-
3d), indicating that all cell lines were instructed by the micromolded features. Myotube
alignment trended lower for DMD 1003 compared to DMD 1015 and DMD 1023, possible
Figure 4-2: iDRM derived from LGMD2A/R1 or DMD dermal fibroblasts show distinct maturation on
micromolded gelatin hydrogel coverslips. α-actinin staining was performed to analyze the maturation of
myotubes from a healthy donor (1001) or patients presenting with LGMD2A/R1 (1077 - c.550delA,
c.1342C>G and 1081 delta 17-24) or DMD (1015-del45, 1023 –del3-23 and 1003 – del46-51). For each
cell line, cells were differentiated for one or two weeks. Images are shown at 20x magnification on the
left, where the scale bar represents 100 μm, and at 60x magnification in the upper right panel where the
scale bar represents 10 μm. Representative myotubes shown in the lower right panel enable visualization
of sarcomeres where the scale bar represents 10μm. Adapted from [178].
92

due to the greater severity of the mutation in this line. The healthy 1001 myotubes also
had relatively low alignment, likely because myotubes tended to bridge across the
micromolded features in the control line compared to the DMD lines.
To determine if our approach was suitable for other genetic muscle diseases, we
similarly isolated fibroblasts from two LGMD2A/R1 subjects, engineered them to iDRM,
and differentiated them to myotubes on micromolded gelatin hydrogels. LGMD2A/R1 line
1077 exhibits compound heterozygous mutations c.550delA (p.T184Rfs*36) and
c.1342C>G (p448R>G) in CAPN3. LGMD2A/R1 line 1081 exhibits homozygous deletion
of CAPN3 exons 17 to 24 (c.(1914+1_1915-1)_(*544_?)del). Whereas LGMD2A/R1 1077
showed lower α-actinin coverage (Figure 4-3a) and normal or lower myogenic index
(Figure 4-3b) compared to the healthy control, LGMD2A/R1 1081 had similar sarcomere
α-actinin coverage and a much higher myogenic index. For certain timepoints,
LGMD2A/R1 1077 had lower myotube width while LGMD2A/R1 1081 had higher myotube
width, compared to control (Figure 4-3c). However, either LGMD2A/R1 1077 nor
LGMD2A/R1 1081 developed well-organized, discrete striations (Figure 2). Both
LGMD2A/R1 1077 and LGMD2A/R1 1081 exhibited high degrees of myotube alignment
(Figure 4-3d). Thus, our engineered tissues revealed distinct differences in disease- and
mutation-specific muscle morphology, which is an advantage of using patient-derived
cells. Importantly, the variability within each cell line is relatively modest.
93

4.3.2. Morphology of muscle fibers in healthy and DMD patients
As a baseline comparison for our culture models, we evaluated morphological
differences in muscle biopsy tissue sections from a healthy subject and the two DMD
subjects for which we had access to muscle tissue sections, DMD 1015 and DMD 1023.
DMD 1015 exhibits an out of frame deletion of DMD exon 45 predicted to lead to total
loss of dystrophin expression. DMD 1023 exhibits an in-frame deletion of DMD exons 3
to 23 predicted to express a partially functional dystrophin protein lacking the actin binding
site. As expected, muscle fibers in the healthy subject robustly expressed sarcomeric α-
Figure 4-3: Morphological quantification of DMD or LGMD2A/R1 or iDRM reveal distinct changes in
mutation-specific muscle morphology. Different parameters were measured for each cell line using α-
actinin staining: (a) α-actinin area coverage per field of view; (b) myogenic index representing the
proportion of nuclei in myotubes (containing at least 3 nuclei); (c) minor axis length as a proxy for myotube
width; (d) myotube alignment. Bars represent SEM. *p < 0.05; p values reflect a student’s t test. Each
experiment (n = 3) is represented by different color dots. Adapted from [178].

94

actinin throughout their cross-sectional area, were encircled by dystrophin, and were
relatively consistent in size (Figure 4-4). Both DMD biopsies demonstrated
weaker sarcomeric α-actinin staining, variations in muscle fiber size, and fibro-fatty
infiltrates, concordant with a DMD phenotype (Figure 4-5). Sarcomeric α-actinin
expression (Figure 4-5), and fiber size were lowest in DMD 1023 tissues despite
significant expression of internally truncated sarcolemmal dystrophin protein in every
myofiber (Figure 4-4, 4-5). Whereas most 1015 fibers do not express dystrophin, we
observe some patches of dystrophin expressing revertant fibers and higher α-actinin
Figure 4-4: DMD patients’ biopsies show abnormal α-actinin striations. Skeletal muscle biopsies from a
healthy donor and patients presenting Duchenne Muscular Dystrophy (1023 del3_23; 1015-del45). (a)
Co-staining for α-actinin and laminin are shown. (b) The total area of the fibers (µm
2
) was measured on
the biopsies (number of fibers analyzed between 250-658) and sorted by 500 μm range. (c) The mean
average of the fibers is also represented. Bars represent SEM. ****p < 0.0005; p values reflect a student’s
t-test. Immunostains were taken at 20x, where the scale bar represents 100 μm. Adapted from [178].

95

expression relative to 1023, in keeping with findings regarding α-actinin coverage in the
iDRM culture models (Figure 4-2, 4-3). However, it is important to consider that the
degeneration process in vivo is vastly more complex and on a much longer timescale
than what can be observed in vitro. Thus, although it is informative to compare patient
muscle biopsies to in vitro engineered muscle tissues, this should be done with much
caution and conservativeness.

Figure 4-5: DMD patients’ biopsies show dystrophic features. Muscle biopsy sections from patients
presenting Duchenne Muscular Dystrophy (1023 del3_23; 1015-del45) assessed for Dystrophin protein
using dys2 antibody targeting the C-terminus and laminin using L9393 antibody. Hematoxylin/eosin
staining for each biopsy is showed to visualize their morphological features and proceed to morphological
assessment. Scale bar represents 100 μm. Adapted from [178].

96

4.3.3. Evaluation of dystrophin rescue in DMD iDRM myotubes treated with exon skipping
AOs
Development of precision medicines requires human mutation specific platforms
for assessing mechanism of action and efficacy. Antisense oligonucleotide (AO) DMD
exon skipping drugs function to reframe mutant DMD mRNA through removal of an “extra”
exon, enabling rescue of an internally deleted but partially functional dystrophin protein.
“Exon skipping” drugs are mutation-specific, in so far as only some DMD mutations
adjacent to the targeted region can be reframed. 1015 mutation can be rendered in frame
by exclusion of exon 44, whereas 1003 mutation is amenable to reframing by targeting
exon 45.
97

To determine if AO treatment rescued dystrophin expression in iDRM, DMD 1015
and DMD 1003 iDRM were exposed to exon skipping drug targeting exon 44 or exon 45,
respectively, after one or two weeks of differentiation on micromolded hydrogels. We
performed immunostaining on healthy and DMD iDRM engineered tissues treated with or
without AO to determine if we could observe rescue of dystrophin (Figure 4-6a).
Figure 4-6: Dystrophin rescue by DMD exon skipping in DMD 1015 and DMD 1003 iDRMs. DMD 1015
(delta 45) and DMD 1003 (delta 46–51) iDRMs were cultured for
1 or 2 weeks before the addition of antisense oligonucleotides and, 2 days later, immunostaining or RNA
extraction (on pooled triplicate) were performed. (a) Dystrophin expression in healthy and DMD iDRM
tissues was visualized using Dys2 antibody (c-ter). Images are shown at 20x magnification in the top
panel, where the scale bar represents 100 μm, and at 60x magnification in the lower panel to enable
visualization of sarcomeres, where the scale bar represents 30 μm. After RT-PCR, samples were run on
chips to analyze exon skipping. The percentage of skipped mRNA over the total (skipped plus unskipped
mRNA) is indicated for (b) DMD 1015 and (c) DMD 1003. Adapted from [178].

98

Myotubes generated from the healthy iDRM demonstrated dystrophin staining at both
timepoints (Figure 4-6a). In DMD 1015 iDRM myotubes, low levels of dystrophin protein
were detected without AO treatment after two week of culture on the micromolded
platform, consistent with detection of low levels of skipped DMD message in the absence
of AO. Exposure to AO increased the rescued exon 44 deleted DMD mRNA and
dystrophin proteins expression and after both one and two weeks of culture, consistant
with levels of exon 45 skipped mRNA induced (Figure 4-6b). In DMD 1003 iDRM
myotubes, dystrophin was not detectable in 1003 in the absence of skipping drug.
Exposure to exon 45 skipping AO rescued low levels of dystrophin protein when AO is
added after one week and to a greater extent when AO was added after 2 weeks of
differentiation in culture (Figure 4-6). Thus, both DMD 1015 and DMD 1003 demonstrated
brighter dystrophin staining with the addition of skipping AO.
We next measured the effects of dystrophin rescue induced by exon skipping on
myotube morphology for DMD 1015 and DMD1003 at week 1 or 2 of differentiation
(Figure 4-7). We performed staining for α-actinin and nuclei (Figure 4-7) and calculated
myotube coverage (Figure 4-8a), myogenic index, (Figure 4-8b), myotube width (Figure
4-8c), and myotube alignment (Figure 4-8d). DMD 1015 iDRM myotubes developed
striations at week 1 and 2, regardless of exon skipping and dystrophin rescue. There were
also no noticeable changes in myogenic index, myotube coverage, myotube width, or
myotube alignment in 1015 due to AO treatment, except for slight but non-significant
increases in the first three metrics after two weeks of AO treatment. The effects on
dystrophin rescue in DMD 1003 were more noticeable, with exon skipping AO inducing
greater myotube coverage, myogenic index, and myotube alignment at both timepoints.
99

However, these increases did not reach statistical significance. Nonetheless, taken
together, these findings highlight a potential role for rescued dystrophin protein in
maturation or stability of myotubes and sarcomeres.

Figure 4-7: Effects of dystrophin rescue by exon skipping on α-actinin expression and organization. DMD
1015 (delta 45) or DMD 1003 (delta 46–51). (a) DMD 1015 or (b) DMD 1003 cells were cultured for 1 or
2 weeks before the addition of antisense oligonucleotides and α-actinin
staining was performed to analyze the maturation of myotubes. Images are shown at 20x magnification
in the left panel, where the scale bar represents 100 μm, and at 60x magnification in the upper right
panel, where the scale bar represents 50 μm. Representative myotubes shown in the lower right panel
enable visualization of sarcomeres, where the scale bar represents 10 μm. Adapted from [178].

100

4.4. Discussion
Development of genetic medicines aimed at repairing or replacing DMD and
LGMDDR1/LGMD2A mutations is an active area of research and several drugs have
been already approved or are in human pre-clinical or clinical trials [192]. However, there
is a paucity of DMD and LGMD mutation specific human culture models for assessing
mechanism of action or improving efficacy of therapeutic strategies. Here, we report that
culturing heathy subject, three DMD, and two LGMD-derived iDRM on engineered gelatin
micromolded coverslips induces tissue alignment, prolongs culture lifetime, and promotes
Figure 4-8: Morphological quantification of AO-treated DMD 1015 and DMD 1003 iDRMs. Different
parameters were measured for each cell line using α-actinin staining: (a) α-actinin area coverage per
field of view; (b) myogenic index representing the proportion of nuclei in myotubes (containing at least
three nuclei); (c) minor axis length as a proxy for myotube width; and (d) myotube alignment. Bars
represent SEM. Each experiment (n = 3) is represented by different color dots. Adapted from [178].

101

myotube development in culture over two weeks. This relatively long culture period we
previously showed compatible with primary chick myotubes [80, 83] enabled us to
characterize myotube development and dystrophin rescue in human DMD myotubes in
response to AO skipping drugs. We also show that our approach is compatible with iDRM
from LGMD2A/R1 patients, demonstrating its modularity for modeling other human-
relevant forms of muscular dystrophy.
Although all healthy and DMD iDRM demonstrated myogenic potential and high
degree of alignment, DMD 1003 showed the most severe defects in α-actinin expression,
followed by DMD 1023. We were largely unable to distinguish DMD 1015 from healthy
control using these measures. The in vitro results for DMD 1015 and DMD 1023 mirror
the trends of the patient muscle biopsies, for which defects were more substantial for
DMD 1023 compared to DMD 1015. Thus, our combination of patient-derived iDRM with
micromolded gelatin hydrogels replicated select histological phenotypes of native muscle
and further demonstrate that muscle development in vitro is regulated by distinct DMD
mutations.
iDRM DMD 1003 harbors a deletion of exons 46-51 of DMD, which encodes an
out of frame mRNA, produces no dystrophin protein, and demonstrates defects in α-
actinin coverage and the development of striations. These data support a role for
dystrophin in the development or stabilization of myotubes and sarcomeres. Such a
suggestion is consistent with reports that the dystrophin-glycoprotein complex slows
depolymerization of actin filaments in vitro [193, 194]. Alternatively, low α-actinin
coverage and myotube width may be secondary to impaired cell survival or earlier
developmental defects dependent on dystrophin. DMD reframing and dystrophin rescue
102

induced within DMD 1003 by exposure to exon 45 skipping AO modestly improved
measures of myotube coverage, fusion, and alignment, further supporting a role for
dystrophin in these processes and highlight the value of this model as a screening tool
for drug development and optimization of precision medicines.
iDRM DMD 1023 has an in-frame mutation DMD 3-23 mutation, which creates
substantial amounts of dystrophin properly localized at the sarcolemma but lacking the
primary DMD actin binding site [179]. We identified defects in myotube maturation and
organization, potentially highlighting requirements for dystrophin actin binding in
myogenesis and myotube sarcomere maturation, consistent with several reports [195].
Additional experiments will be needed to support this hypothesis.
iDRM DMD 1015 has an exon 45 deletion of DMD amenable to reframing by exon
44 AO skipping. We have previously demonstrated that iDRM DMD 1015 constitutively
expresses low levels of exon 45 skipped and reframed DMD mRNA, even in the absence
of AO, as we have reported for several DMD iDRM with exon 45 deletions [180]. Likewise,
immunostaining frozen muscle biopsy sections from the DMD 1015 fibroblast donor
showed some clusters expressing low levels of dystrophin rescue in vivo. The
predisposition of exon 45 deletion DMD mutants to self-correct by skipping exon 44 to
produce low levels of dystrophin has been suggested as the molecular basis of mild
disease progression relative to typical DMD frameshifting mutations and highlights that
even low levels of dystrophin can have functional consequences. We also detected low
levels of exon 45 skipped DMD mRNA in the absence of any treatment and increased
exon 44 exclusion and robust induction of dystrophin protein expression in cells exposed
to skipping AO. Unlike the other two DMD patient-derived iDRM, DMD 1015 behaved
103

much closer to wildtype, with no defect in α-actinin coverage or striation. It is possible that
the low levels of rescued dystrophin constitutively expressed in DMD 1015 are sufficient
to partially overcome the developmental defects in myotube development or sarcomere
stability observed in dystrophin null DMD 1003. Exon 44 skipping AO enhances increased
dystrophin expression and induced modest improvements in myogenic index, myotube
coverage, and myotube width, further supporting a role for rescued dystrophin in
facilitation of myotube maturation in DMD 1015.
While differentiation on micromolded gelatin hydrogels enabled visualization of
rescued dystrophin expression, we did not observe the expected patterning of dystrophin.
In native muscle, dystrophin is enriched in costamere protein assemblies, which
circumferentially align with the α-actinin enriched Z disk and couple force-generating
sarcomeres with the sarcolemma. Similarly, although we did detect some punctate
sarcomere-like structures in myotubes derived from select iDRM lines, the overall maturity
of the myofibrils and sarcomeres was limited, especially compared to myotubes derived
from primary myoblasts. Similar issues related to myofibril immaturity have routinely been
observed in myotubes derived from a variety of reprogrammed [196] and iPSC-derived
myoblasts [44, 172] and likely reduce the baseline and drug-induced differences in the
phenotypes of healthy and diseased cells. Extending culture time [80] , integrating
supporting cell types [80, 133, 134], providing electrical [154, 197] or mechanical
stimulation [198, 199], or engineering 3-D tissues such as those described in chapter 1
[61, 62, 82, 93, 200, 201] or earlier exposure to AO could help induce muscle maturation
and proper localization of dystrophin and α-actinin. However, most of these interventions
104

reduces throughput and increases cost, which are significant drawbacks for rapidly
screening panels of candidate drugs in a multitude of patient-specific lines.
LGMD2A/R1 1077 express mutations (c.550delA (p.T184Rfs*36) and c.1342C>G
(p448R>G) within the catalytic PC1 and the calcium-binding and phospholipid-binding C2
domains, respectively and these iDRM are show defects in formation of myotube
maturation and structure. Patient cells derived from patient LGMD2A/R1 1081, present
with a homozygous deletion of the entire c-terminal region of the protein, where able to
form myotubes and showed no quantitative defects in α-actinin coverage or axis length,
but did not develop proper maturation of the clearly defined α-actinin marked sarcomeres
upon visually inspection. It is unclear why LGMD 1077 iDRM demonstrates significant
defects in myotube maturation, whereas 1081 has an exceptionally high MI and is
otherwise near normal.
One advantage of differentiating myotubes in culture, is that it allows assessment
of sequential myoblast activation, fusion, and the development of mature myofibers with
organized sarcomeres, responsible for striated muscle patterning and required for force
generation. Thus, iDRM differentiated on micromolded platforms may aid in dissecting
requirements for each of these developmental events [202]. However, factors other DMD
or LGMD mutation might influence iDRM performance including presence/absence of
DMD or LGMD2A/R1 disease modifier genes or artifacts of culture selection. Therefore,
comparisons between individually derived iDRM to determine relative myogenic activity
of particular mutations should be made with caution. Rather, creation of isogenic iDRM
expressing defined DMD or LGMD mutations and full-length dystrophin or a panel of
distinct patient iDRM with similar mutations may be necessary to confirm preliminary
105

observations made regarding effects of DMD or LGMD mutation based on comparison
between two lines. Alternatively, experiments where iDRM DMD 1003 and 1015 activity
is measured before and after exposure to a potential therapeutic or disease modifying
entity, such as exon skipping AO, do not suffer from this criticism and thus are likely to
prove valuable for precision drug optimization and screening.
Our results establish that patient-derived iDRM differentiate into aligned myotubes
on micromolded gelatin hydrogels and enable evaluation of myotube formation as a
function of patient-specific mutations. Our novel experiments serve as proof-of-concept
that applying our engineered systems with patient-derived cells is a promising approach
for screening and testing personalized therapies for muscular dystrophies and other
genetic neuromyopathies in vitro. Together, our findings highlight the benefit of human
cell models combined with engineered scaffolds that more closely mimic the in vivo
microenvironment and encourage the development of new strategies to promote further
maturation for proper characterization of morphological differences and evaluation of the
efficacy of new therapies. Moving forward, using the micromolded gelatin hydrogel and
myobundle system with patient-derived myogenic and neuronal tissues will further
increase the utility of in vitro cultures, as described in the next chapter.

106

Chapter 5 Concluding Remarks and Future Work
5.1. Modeling amyotrophic lateral sclerosis with a neuromuscular
microphysiological system
The developed neuromuscular microphysiological platforms should be capable of
robustly developing more functionally and anatomically relevant NMJs in vitro. In addition
to the contractile stress generation baselines established here, we can further
characterize control values for healthy skeletal muscle and NMJ function based on and
electrophysiological activity, values that should better recapitulate physiology due to the
relative development of our cultures.
We can use the engineered NMJs in 2D or 3D space to study specific diseases,
such as C9ORF72 ALS and restoration of function. Our tools are particularly useful for
diseases like ALS, which have a proposed genetic and environmental origin, due to the
ability to use different patient-derived cells (e.g., iPSCs or iDRMs) while also easily
introducing environmental factors, as appropriate. Because the platform is more easily
modular, scalable, and accessible compared to traditional neuromuscular cultures, we
expect the platform will supplement the therapeutic development pipeline and potentially
reduce time and cost for drug discovery. The platform will be capable of helping evaluate
the safety and efficacy of therapies with higher accuracy and efficiency than traditional
neuromuscular in vivo models or in vitro cultures alone, making it potentially
advantageous for both basic science and clinical research of neuromuscular function.
5.2. Skeletal muscle tissue engineering for consumable products
As population growth, economic development, and urbanization in the mid-21
st

century surges, demands for consumable meats have skyrocketed [203]. Large-scale
107

expansion of animal agriculture is mired with potential issues of zoonotic illness
proliferation, greenhouse gas emissions, and animal welfare concerns [203]. A potential
approach to alleviating these pressures is the use of cell-cultivated meats, or animal-
derived cells grown and differentiated in some sort of scaffolding, which may alleviate
environmental and ethical concerns while (ideally) preserving nutritional content and
taste/texture of the cultivated meat.
Several challenges exist in the development of palatable cultured meats:
integration of different cell types, lowering media costs in culture, and consumer
willingness to eat lab-grown meats are just a few of many obstacles towards mainstream
commercialization. The experiments described here yield some direction to addressing
these concerns. For example, our RNASeq analysis on chick muscle tissues cultured on
our 2D gelatin hydrogel platform revealed that over 3 weeks of culture, tissues express a
higher amount of both fast- and slow-fiber adult myosin isoforms (Figure 2-9) [81]. In
comparison with native tissue (unpublished data), these engineered tissues adopt a more
glycolytic-fast phenotype. However, our myobundle experiments show potential shifts,
towards a more oxidate phenotype can occur based on different culture conditions, such
as extent of innervation. Because fiber type is known to affect taste (slow fibers adopt a
more umami flavor) [204], further investigation into controlling fiber-type in vitro may yield
useful information towards controlling cultured meat taste. Thus, while tangential in
ultimate goals, studying in vitro skeletal muscle development in more physiologically
relevant engineered systems may yield several insights into this burgeoning field.
108

5.3. Visualization and analysis of 3D myobundle architecture in situ
The analysis of NMJ formation in animal or in vitro studies is complicated by the
low accessibility for most imaging modalities. While cryosectioning, as performed in our
studies, can yield well-preserved samples for microscopy, the practice can be time-
consuming and often takes a trained and experienced handler for optimal staining
[205]. Cryosectioning also limits visualization of 3D structures to a single plane, which
takes away from innervated skeletal muscle’s well-defined, cell-matrix architecture.
Optical tissue clearing techniques and subsequent confocal, super-resolution stimulated
emission depletion (STED), or other microscopy imaging may allow for the in situ
observation of NMJs while avoiding potential fiber ripping due to cryosectioning [82]. To
the best of our knowledge, clearing techniques for skeletal muscle tissue have not been
explored thoroughly, despite some successful approaches [82, 206] that clear very thin
tissue sections optimally. Thus, further investigation is necessary into clearing parameters
(e.g., detergent selection, environmental conditions, physical disruption, etc.) to limit any
damage to muscle-specific structures that occur during lipid removal, which would prevent
accurate analysis of NMJ formation and degeneration.
5.4. Limitations to our approaches and other future improvements
With the work presented in this dissertation, we contributed to the understanding
of NMJ formation and function in vitro in 2D and 3D space. We provide evidence that
longer cultures on engineered platforms allow for a more nuanced investigation into
relatively well-developed neuromuscular tissue. The integration of human and mutation-
specific cells allowed us to also probe these platforms’ potential in studying and
addressing muscular dystrophies. Further, as we have described, balancing 2D systems
109

with higher throughput and accessibility and 3D systems that display higher physiological
relevance allows researchers to study the aspects of skeletal muscle most relevant to
their research question. However, there are also several valid criticisms of these models,
which could be better addressed in future study.
In both our 2D and 3D systems, the architecture of innervated tissue is not fully
representative of native human tissue. Motor neuron bodies remain secluded in the spinal
cord while axons traverse up to 1 meter to innervate skeletal muscle fibers [207]. This
elongation of motor neuron axons has been shown to affect cytoskeletal properties, local
protein translation, and mitochondrial homeostasis of human iPSC-derived motor
neurons. However, because the platforms described in this dissertation are mixed in with
skeletal muscle tissue, this elongation is hard to observe and quantify, and motor neuron
bodies interact with myotubes in a non-physiological manner. Other systems that take
advantage of microgrooves that only permit axon traversal [208] or microchannels
between muscle and neuron populations [69] have shown promise in creating up to
centimeter-scale axon lengths, but further study is needed as these studies showed little
evidence of mature NMJ formation.
A compartmentalized system may also provide the opportunity for easier
separation of neuron bodies and skeletal muscle tissue for a variety of relevant assays.
One challenge we faced in analyzing 3D myobundles was the inability to clearly delineate
innervated muscle cells in the bulk tissue. Thus, while membrane potential readings and
potential mEPSPs were observed (not published), it remains a challenge to quantify the
electrophysiological activity of innervated myobundles consistently and unbiasedly.
Proper compartmentalization would yield a more physiological structure and a better idea
110

of where/which myotubes are innervated. Physical separation of culture into proximal,
medial, and distal regions of the motor neuron axon in addition to the skeletal muscle
tissue may further reveal important differences in the transcriptome at these different
sections of the neuromuscular tissue [208]. Overall, several ongoing advances in stem
cell biology and microfabrication/microfluidics will help address current limitations in our
engineered neuromuscular microphysiological systems, and allow for broader
applications in basic science, therapeutic development, and beyond.
5.5. Funding Sources
We thank the NINDS Biorepository at the Coriell Institute and the Center for
Duchenne Muscular Dystrophy at the University of California-Los Angeles for providing
the human cell lines used for this study. We also acknowledge the W.M. Keck Foundation
Photonics Center Cleanroom for photolithography equipment and facilities. We also thank
Dr. Eddih Loh for his input and acknowledge the USC Libraries Bioinformatics Service for
their assistance in interpreting RNASeq data. This work was further funded by NSF GRFP
Grant No. DGE1418060, NSF CAREER Award No. 1944734 (Dr. Megan McCain), the
USC Broad Innovation Award, the ALS Association Starter grant 18-IIA-401, the Rose
Hills Foundation Innovator grant, the USC Provost’s Fellowship, the USC Viterbi Internal
Center Incubator and Center for Integrated Electronic and Biological Organisms
(CIEBOrg), the Donald E. and Delia B. Baxter Foundation, the Tau Consortium, the Frick
Foundation for ALS Research, the Muscular Dystrophy Association, the New York Stem
Cell Foundation, the Alzheimer’s Drug Discovery Foundation, the Association for
Frontotemporal Degeneration, the Pape Adams Foundation, the John Douglas French
Alzheimer’s Foundation, the Harrington Discovery Institute, the Merkin Family
111

Foundation, and the California Center for Rare Diseases within the Institute of Precision
Health at UCLA.

112

References
[1] J.C.W. Deenen, C.G.C. Horlings, J.G.M.V. Jan, A.L.M. Verbeek, B.G.M.v. Engelen,
The Epidemiology of Neuromuscular Disorders: A Comprehensive Overview of the
Literature, Journal of neuromuscular diseases 2(1) (2015) 73-85.

[2] B.M. Morrison, Neuromuscular Diseases, Seminars in neurology 36(5) (2016) 409-
418.

[3] A.J. Vander, J.H. Sherman, D.S. Luciano, Human physiology : the mechanisms of
body function, 8th ed., McGraw-Hill, Boston, 2001.

[4] H. Yin, F. Price, M.A. Rudnicki, Satellite cells and the muscle stem cell niche,
Physiological reviews 93(1) (2013) 23-67.

[5] H. Anne, H. Heli, N. Oula, P. Taina, Collagen XIII and other ECM components in the
assembly and disease of the neuromuscular junction, The Anatomical Record (2019).

[6] A. Fidzianska, Human ontogenesis. II. Development of the human neuromuscular
junction, J Neuropathol Exp Neurol 39(5) (1980) 606-15.

[7] C.R. Slater, The Structure of Human Neuromuscular Junctions: Some Unanswered
Molecular Questions, Int J Mol Sci 18(10) (2017).

[8] Y.C. Jang, H. Van Remmen, Age-associated alterations of neuromuscular
junctionAge-associated alterations of neuromuscular junction, Exp Gerontol. 46(2-3)
(2011) 193-198.

[9] W. Li, How do SMA-linked mutations of SMN1 lead to structural/functional deficiency
of the SMA protein?, PLOS ONE 12(6) (2017) e0178519.

[10] M. Jedrzejowska, M. Milewski, J. Zimowski, J. Borkowska, A. Kostera-Pruszczyk, D.
Sielska, M. Jurek, I. Hausmanowa-Petrusewicz, Phenotype modifiers of spinal muscular
atrophy: the number of SMN2 gene copies, deletion in the NAIP gene and probably
gender influence the course of the disease, Acta Biochim Pol 56(1) (2009) 103-8.

[11] T. Zuroske, Upcoming market catalysts in Q2 2019 Nature Reviews Drug Discovery,
2019.

[12] W. Florian Barthélémy and Nicolas, Personalized gene and cell therapy for Duchenne
Muscular Dystrophy, Neuromuscular Disorders 28(10) (2018) 803-824.

[13] D.J. Blake, A. Weir, S.E. Newey, K.E. Davies, Function and Genetics of Dystrophin
and Dystrophin-Related Proteins in Muscle, Physiological Reviews 82(2) (2002) 291-329.
113

[14] P. Gailly, New aspects of calcium signaling in skeletal muscle cells: implications in
Duchenne muscular dystrophy, Biochimica et Biophysica Acta (BBA) - Proteins and
Proteomics 1600(1) (2002) 38-44.

[15] J.W. McGreevy, C.H. Hakim, M.A. McIntosh, D. Duan, Animal models of Duchenne
muscular dystrophy: from basic mechanisms to gene therapy, Disease models &
mechanisms 8(3) (2015) 195-213.

[16] F. Barthélémy, R.T. Wang, C. Hsu, E.D. Douine, E.E. Marcantonio, S.F. Nelson, M.C.
Miceli, Targeting RyR Activity Boosts Antisense Exon 44 and 45 Skipping in Human DMD
Skeletal or Cardiac Muscle Culture Models, Mol Ther Nucleic Acids 18 (2019) 580-589.

[17] D. McCorquodale, E.M. Pucillo, N.E. Johnson, Management of Charcot-Marie-Tooth
disease: improving long-term care with a multidisciplinary approach, J Multidiscip Healthc
9 (2016) 7-19.

[18] W.D. Phillips, A. Vincent, Pathogenesis of myasthenia gravis: update on disease
types, models, and mechanisms, F1000Research 5(1513) (2016).

[19] R. Mantegazza, C. Cordiglieri, A. Consonni, F. Baggi, Animal models of myasthenia
gravis: utility and limitations, Int J Gen Med 9 (2016) 53-64.

[20] E. Talbott, A. Malek, D. Lacomis, Chapter 13 - The epidemiology of amyotrophic
lateral sclerosis, Handbook of Clinical Neurology 138 (2016) 225 - 238.

[21] K. Boylan, Familial Amyotrophic Lateral Sclerosis, Neurol Clin 33(4) (2015) 807-30.

[22] Y. Shi, S. Lin, K.A. Staats, Y. Li, W.-H. Chang, S.-T. Hung, E. Hendricks, G.R.
Linares, Y. Wang, E.Y. Son, X. Wen, K. Kisler, B. Wilkinson, L. Menendez, T. Sugawara,
P. Woolwine, M. Huang, M.J. Cowan, B. Ge, N. Koutsodendris, K.P. Sandor, J. Komberg,
V.R. Vangoor, K. Senthilkumar, V. Hennes, C. Seah, A.R. Nelson, T.-Y. Cheng, S.-J.J.
Lee, P.R. August, J.A. Chen, N. Wisniewski, V. Hanson-Smith, T.G. Belgard, A. Zhang,
M. Coba, C. Grunseich, M.E. Ward, L.H. van den Berg, R.J. Pasterkamp, D. Trotti, B.V.
Zlokovic, J.K. Ichida, Haploinsufficiency leads to neurodegeneration in C9ORF72
ALS/FTD human induced motor neurons, Nature Medicine 24 (2018) 313.

[23] E.R. Seminary, S.L. Sison, A.D. Ebert, Modeling Protein Aggregation and the Heat
Shock Response in ALS iPSC-Derived Motor Neurons, Frontiers in neuroscience 12
(2018) 86-86.

[24] F.C. Su, S.A. Goutman, S. Chernyak, B. Mukherjee, B.C. Callaghan, S. Batterman,
E.L. Feldman, Association of Environmental Toxins With Amyotrophic Lateral Sclerosis,
JAMA Neurol 73(7) (2016) 803-11.

[25] M.R. Turner, J. Barnwell, A. Al-Chalabi, A. Eisen, Young-onset amyotrophic lateral
sclerosis: historical and other observations, Brain 135(9) (2012) 2883-2891.
114

[26] H.P. Nguyen, C. Van Broeckhoven, J. van der Zee, ALS Genes in the Genomic Era
and their Implications for FTD, Trends in Genetics 34(6) (2018) 404 - 423.

[27] S. Lattante, S. Ciura, G.A. Rouleau, E. Kabashi, Defining the genetic connection
linking amyotrophic lateral sclerosis (ALS) with frontotemporal dementia (FTD), Trends in
Genetics 31(5) (2015) 263-273.

[28] M. Li, V.O. Ona, C. Guégan, M. Chen, V. Jackson-Lewis, L.J. Andrews, A.J.
Olszewski, P.E. Stieg, J.-P. Lee, S. Przedborski, R.M. Friedlander, Functional Role of
Caspase-1 and Caspase-3 in an ALS Transgenic Mouse Model, Science 288(5464)
(2000) 335-339.

[29] T. Philips, J.D. Rothstein, Rodent Models of Amyotrophic Lateral Sclerosis, Current
protocols in pharmacology 69 (2015) 5.67.1-5.67.21.

[30] P.A. Jonsson, K.S. Graffmo, P.M. Andersen, T. Brännström, M. Lindberg, M.
Oliveberg, S.L. Marklund, Disulphide-reduced superoxide dismutase-1 in CNS of
transgenic amyotrophic lateral sclerosis models, Brain 129(2) (2006) 451-464.

[31] J.-P. Julien, J. Krinz, Transgenic mouse models of amyotrophic lateral sclerosis,
Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease 1762(11) (2006) 1013
- 1024.

[32] M.R. Watson, R.D. Lagow, K. Xu, B. Zhang, N.M. Bonini, A drosophila model for
amyotrophic lateral sclerosis reveals motor neuron damage by human SOD1, The Journal
of biological chemistry 283(36) (2008) 24972-24981.

[33] S. Perry, Y. Han, A. Das, D. Dickman, Homeostatic plasticity can be induced and
expressed to restore synaptic strength at neuromuscular junctions undergoing ALS-
related degeneration, Human Molecular Genetics 26(21) (2017) 4153-4167.

[34] P.J. Babin, C. Goizet, D. Raldúa, Zebrafish models of human motor neuron diseases:
Advantages and limitations, Progress in Neurobiology 118 (2014) 36-58.

[35] A. Barik, L. Li, A. Sathyamurthy, W.-C. Xiong, L. Mei, Schwann Cells in
Neuromuscular Junction Formation and Maintenance, The Journal of Neuroscience
36(38) (2016) 9770-9781.

[36] J. Mark, A. Nadia, T.W. Jason, Y. Jean, S. Zohaib, S. Chaichontat, Q. Ying, B. Nenad,
Incorporation of macrophages into engineered skeletal muscle enables enhanced muscle
regeneration, Nature Biomedical Engineering (2018) 1-13.

[37] C. Neville, N. Rosenthal, M. McGrew, N. Bogdanova, S. Hauschka, Chapter 5
Skeletal Muscle Cultures, in: C.P. Emerson, H.L. Sweeney (Eds.), Methods in Cell
Biology, Academic Press1997, pp. 85-116.
115

[38] F. Stetzkowski-Marden, K. Gaus, M. Recouvreur, A. Cartaud, J. Cartaud, Agrin elicits
membrane lipid condensation at sites of acetylcholine receptor clusters in C2C12
myotubes, J Lipid Res 47(10) (2006) 2121-33.

[39] C.S. Cheng, Y. El-Abd, K. Bui, Y.E. Hyun, R.H. Hughes, W.E. Kraus, G.A. Truskey,
Conditions that promote primary human skeletal myoblast culture and muscle
differentiation in vitro, Am J Physiol Cell Physiol 306(4) (2014) C385-95.

[40] Q. Zhao, S.T. Yang, J.J. Wang, J. Zhou, S.S. Xing, C.C. Shen, X.X. Wang, Y.X. Yue,
J. Song, M. Chen, Y.Y. Wei, Q.P. Zhou, T. Dai, Y.H. Song, TNF alpha inhibits myogenic
differentiation of C2C12 cells through NF-κB activation and impairment of IGF-1 signaling
pathway, Biochemical and Biophysical Research Communications 458(4) (2015) 790-
795.

[41] M. Menconi, P. Gonnella, V. Petkova, S. Lecker, P.O. Hasselgren, Dexamethasone
and corticosterone induce similar, but not identical, muscle wasting responses in cultured
L6 and C2C12 myotubes, Journal of Cellular Biochemistry 105(2) (2008) 353-364.

[42] F.M. Vallette, M. Vigny, J. Massoulie, Muscular differentiation of chicken myotubes
in a simple defined synthetic culture medium and in serum supplemented media:
Expression of the molecular forms of acetylcholinesterase, Neurochem Int 8(1) (1986)
121-33.

[43] Y. Miyagoe-Suzuki, S.i. Takeda, Skeletal muscle generated from induced pluripotent
stem cells - induction and application, World journal of stem cells 9(6) (2017) 89-97.

[44] L. Rao, Y. Qian, A. Khodabukus, T. Ribar, N. Bursac, Engineering human pluripotent
stem cells into a functional skeletal muscle tissue, Nature Communications 9(1) (2018)
126.

[45] L. Ramon C. J, S. Annemie M. W. J, K. Marco C. J. M, W. Emiel F. M, J.-H. Yvonne
M. W, Inflammatory cytokines inhibit myogenic differentiation through activation of nuclear
factor-κB, The FASEB Journal 15(7) (2001) 1169-1180.

[46] A. Bettadapur, G.C. Suh, N.A. Geisse, E.R. Wang, C. Hua, H.A. Huber, A.A. Viscio,
J.Y. Kim, J.B. Strickland, M.L. McCain, Prolonged Culture of Aligned Skeletal Myotubes
on Micromolded Gelatin Hydrogels, Scientific reports 6 (2016) 28855.

[47] S.M. Hindi, M.M. Tajrishi, A. Kumar, Signaling mechanisms in mammalian myoblast
fusion, Science signaling 6(272) (2013) re2-re2.

[48] J.L. Myhre, D.B. Pilgrim, At the Start of the Sarcomere: A Previously Unrecognized
Role for Myosin Chaperones and Associated Proteins during Early Myofibrillogenesis,
Biochemistry Research International 2012 (2012) 16.
116

[49] C.S. Cheng, B.N. Davis, L. Madden, N. Bursac, G.A. Truskey, Physiology and
metabolism of tissue-engineered skeletal muscle, Exp Biol Med (Maywood) 239(9) (2014)
1203-14.

[50] K.H. Park, L. Brotto, O. Lehoang, M. Brotto, J. Ma, X. Zhao, Ex vivo assessment of
contractility, fatigability and alternans in isolated skeletal muscles, Journal of visualized
experiments : JoVE (69) (2012) e4198-e4198.

[51] J.P. Wikswo, The relevance and potential roles of microphysiological systems in
biology and medicine, Experimental biology and medicine (Maywood, N.J.) 239(9) (2014)
1061-1072.

[52] R.M. Duffy, Y. Sun, A.W. Feinberg, Understanding the Role of ECM Protein
Composition and Geometric Micropatterning for Engineering Human Skeletal Muscle,
Ann Biomed Eng (2016).

[53] A. Agarwal, Y. Farouz, A.P. Nesmith, L.F. Deravi, M.L. McCain, K.K. Parker,
Micropatterning Alginate Substrates for In Vitro Cardiovascular Muscle on a Chip,
Advanced Functional Materials 23(30) (2013) 3738-3746.

[54] G.C. Suh, A. Bettadapur, J.W. Santoso, M.L. McCain, Fabrication of Micromolded
Gelatin Hydrogels for Long-Term Culture of Aligned Skeletal Myotubes, Methods Mol Biol
1668 (2017) 147-163.

[55] M.L. McCain, A. Agarwal, H.W. Nesmith, A.P. Nesmith, K.K. Parker, Micromolded
gelatin hydrogels for extended culture of engineered cardiac tissues, Biomaterials 35(21)
(2014) 5462--5471.

[56] G. Wang, M.L. McCain, L. Yang, A. He, F.S. Pasqualini, A. Agarwal, H. Yuan, D.
Jiang, D. Zhang, L. Zangi, J. Geva, A.E. Roberts, Q. Ma, J. Ding, J. Chen, D.Z. Wang, K.
Li, J. Wang, R.J. Wanders, W. Kulik, F.M. Vaz, M.A. Laflamme, C.E. Murry, K.R. Chien,
R.I. Kelley, G.M. Church, K.K. Parker, W.T. Pu, Modeling the mitochondrial
cardiomyopathy of Barth syndrome with induced pluripotent stem cell and heart-on-chip
technologies, Nat Med 20(6) (2014) 616-23.

[57] A. Grosberg, A.P. Nesmith, J.A. Goss, M.D. Brigham, M.L. McCain, K.K. Parker,
Muscle on a chip: In vitro contractility assays for smooth and striated muscle, Journal of
Pharmacological and Toxicological Methods 65(3) (2012) 126--135.

[58] G.A. Truskey, Development and application of human skeletal muscle
microphysiological systems, Lab on a Chip 18(20) (2018) 3061-3073.

[59] T. Osaki, S.G.M. Uzel, R.D. Kamm, Microphysiological 3D model of amyotrophic
lateral sclerosis (ALS) from human iPS-derived muscle cells and optogenetic motor
neurons, Science Advances 4(10) (2018).
117

[60] B.N. Davis, J.W. Santoso, M.J. Walker, C.S. Cheng, T.R. Koves, W.E. Kraus, G.A.
Truskey, Human, Tissue-Engineered, Skeletal Muscle Myobundles to Measure Oxygen
Uptake and Assess Mitochondrial Toxicity, Tissue engineering. Part C, Methods 23(4)
(2017) 189-199.

[61] L. Madden, M. Juhas, W.E. Kraus, G.A. Truskey, N. Bursac, Bioengineered human
myobundles mimic clinical responses of skeletal muscle to drugs, eLife 4 (2015).

[62] B.N.J. Davis, J.W. Santoso, M.J. Walker, C.E. Oliver, M.M. Cunningham, C.A.
Boehm, D. Dawes, S.L. Lasater, K. Huffman, W.E. Kraus, G.A. Truskey, Modeling the
Effect of TNF-α upon Drug-Induced Toxicity in Human, Tissue-Engineered Myobundles,
Annals of Biomedical Engineering (2019).

[63] Z.-W.W. Du, H. Chen, H. Liu, J. Lu, K. Qian, C.-L.L. Huang, X. Zhong, F. Fan, S.-
C.C. Zhang, Generation and expansion of highly pure motor neuron progenitors from
human pluripotent stem cells, Nature communications 6 (2015) 6626.

[64] S. Sances, L.I. Bruijn, S. Chandran, K. Eggan, R. Ho, J.R. Klim, M.R. Livesey, E.
Lowry, J.D. Macklis, D. Rushton, C. Sadegh, D. Sareen, H. Wichterle, S.-C. Zhang, C.N.
Svendsen, Modeling ALS with motor neurons derived from human induced pluripotent
stem cells, Nature neuroscience 19(4) (2016) 542-553.

[65] C.L. Happe, K.P. Tenerelli, A.K. Gromova, F. Kolb, A.J. Engler, Mechanically
patterned neuromuscular junctions-in-a-dish have improved functional maturation,
Molecular biology of the cell 28(14) (2017) 1950-1958.

[66] J.A. Umbach, K.L. Adams, C.B. Gundersen, B.G. Novitch, Functional Neuromuscular
Junctions Formed by Embryonic Stem Cell-Derived Motor Neurons, PLoS ONE 7(5)
(2012).

[67] G. Xiufang, G. Mercedes, S. Maria, H.V. Herman, J.H. James, Neuromuscular
junction formation between human stem cell-derived motoneurons and human skeletal
muscle in a defined system, Biomaterials 32(36) (2011) 9602-9611.

[68] E.E. Zahavi, A. Ionescu, S. Gluska, T. Gradus, K. Ben-Yaakov, E. Perlson, A
compartmentalized microfluidic neuromuscular coculture system reveals spatial aspects
of GDNF functions, Journal of cell science 128(6) (2015) 1241-1252.

[69] N. Santhanam, L. Kumanchik, X. Guo, F. Sommerhage, Y. Cai, M. Jackson, C.
Martin, G. Saad, C.W. McAleer, Y. Wang, A. Lavado, C.J. Long, J.J. Hickman, Stem cell
derived phenotypic human neuromuscular junction model for dose response evaluation
of therapeutics, Biomaterials 166 (2018) 64-78.

[70] M. Afshar Bakooshli, E.S. Lippmann, B. Mulcahy, N. Iyer, C.T. Nguyen, K. Tung, B.A.
Stewart, H. van den Dorpel, T. Fuehrmann, M. Shoichet, A. Bigot, E. Pegoraro, H. Ahn,
118

H. Ginsberg, M. Zhen, R.S. Ashton, P.M. Gilbert, A 3D culture model of innervated human
skeletal muscle enables studies of the adult neuromuscular junction, Elife 8 (2019).

[71] L.J. Millet, M.E. Stewart, J.V. Sweedler, R.G. Nuzzo, M.U. Gillette, Microfluidic
devices for culturing primary mammalian neurons at low densities, Lab Chip 7(8) (2007)
987-94.

[72] B.M. Carlos, M.S. Jessica, C.S. Dylan, G.E. Leonid, M.G. Sarah, Z. Chao, M.S.
Kevan, C.S. Gary, TrkB kinase activity maintains synaptic function and structural integrity
at adult neuromuscular junctions, Journal of Applied Physiology 117(8) (2014) 910-920.

[73] J.W. Santoso, M.L. McCain, Neuromuscular disease modeling on a chip, Disease
Models & Mechanisms 13(7) (2020) dmm044867.

[74] B. Martinsen, P. Dreyer, Ageing with Muscular Disease, The open nursing journal 10
(2016) 131-138.

[75] K. Alastair, P. Neel, W. Jason, B. Nenad, In Vitro Tissue ‐Engineered Skeletal Muscle
Models for Studying Muscle Physiology and Disease, Advanced Healthcare Materials
7(15) (2018) 1701498.

[76] A.J. Engler, S. Sen, H.L. Sweeney, D.E. Discher, Matrix Elasticity Directs Stem Cell
Lineage Specification, Cell 126(4) (2006) 677--689.

[77] R. Siller, S. Greenhough, I.-H. Park, G.J. Sullivan, Modelling human disease with
pluripotent stem cells, Current gene therapy 13(2) (2013) 99-110.

[78] E.Y. Son, J.K. Ichida, B.J. Wainger, J.S. Toma, V.F. Rafuse, C.J. Woolf, K. Eggan,
Conversion of mouse and human fibroblasts into functional spinal motor neurons, Cell
stem cell 9(3) (2011) 205-218.

[79] J. White, M.V. Barro, H.P. Makarenkova, J.W. Sanger, J.M. Sanger, Localization of
sarcomeric proteins during myofibril assembly in cultured mouse primary skeletal
myotubes, Anatomical record (Hoboken, N.J. : 2007) 297(9) (2014) 1571-1584.

[80] J.W. Santoso, M.L. McCain, Engineering skeletal muscle tissues with advanced
maturity improves synapse formation with human induced pluripotent stem cell-derived
motor neurons, NCBI Gene Expression Omnibus (GEO), 2021.

[81] J.W. Santoso, X. Li, D. Gupta, G.C. Suh, E. Hendricks, S. Lin, S. Perry, J.K. Ichida,
D. Dickman, M.L. McCain, Engineering skeletal muscle tissues with advanced maturity
improves synapse formation with human induced pluripotent stem cell-derived motor
neurons, APL Bioengineering 5(3) (2021) 036101.

119

[82] N.R. Ariyasinghe, J.W. Santoso, D. Gupta, M.J. Pincus, P.R. August, M.L. McCain,
Optical Clearing of Skeletal Muscle Bundles Engineered in 3-D Printed Templates, Annals
of Biomedical Engineering (2020).

[83] D. Gupta, J.W. Santoso, M.L. McCain, Characterization of Gelatin Hydrogels Cross-
Linked with Microbial Transglutaminase as Engineered Skeletal Muscle Substrates,
Bioengineering (Basel) 8(1) (2021).

[84] D. Qin, Y. Xia, G.M. Whitesides, Soft lithography for micro- and nanoscale patterning,
Nature protocols 5(3) (2010) 491-502.

[85] P.L. Kuo, H. Lee, M.A. Bray, N.A. Geisse, Y.T. Huang, W.J. Adams, S.P. Sheehy,
K.K. Parker, Myocyte shape regulates lateral registry of sarcomeres and contractility, Am
J Pathol 181(6) (2012) 2030-7.

[86] M.L. McCain, T. Desplantez, N.A. Geisse, B. Rothen-Rutishauser, H. Oberer, K.K.
Parker, A.G. Kleber, Cell-to-cell coupling in engineered pairs of rat ventricular
cardiomyocytes: relation between Cx43 immunofluorescence and intercellular electrical
conductance, Am J Physiol Heart Circ Physiol 302(2) (2012) H443-50.

[87] A.W. Feinberg, P.W. Alford, H. Jin, C.M. Ripplinger, A.A. Werdich, S.P. Sheehy, A.
Grosberg, K.K. Parker, Controlling the contractile strength of engineered cardiac muscle
by hierarchal tissue architecture, Biomaterials 33(23) (2012) 5732-41.4

[88] G. Wang, M.L. McCain, L. Yang, A. He, F.S. Pasqualini, A. Agarwal, H. Yuan, D.
Jiang, D. Zhang, L. Zangi, J. Geva, A.E. Roberts, Q. Ma, J. Ding, J. Chen, D.-Z. Wang,
K. Li, J. Wang, R.J.A. Wanders, W. Kulik, F.M. Vaz, M.A. Laflamme, C.E. Murry, K.R.
Chien, R.I. Kelley, G.M. Church, K.K. Parker, W.T. Pu, Modeling the mitochondrial
cardiomyopathy of Barth syndrome with induced pluripotent stem cell and heart-on-chip
technologies., Nature medicine 20(6) (2014) 616--23.

[89] K.K.W. Wieland, A.L. Wolfgang, H.P.W. Manfred, Sarcomere Length–tension
Relationship of Rat Cardiac Myocytes at Lengths Greater than Optimum, Journal of
Molecular and Cellular Cardiology 32(2) (2000) 247-259.

[90] T.A. Morris, J. Naik, K.S. Fibben, X. Kong, T. Kiyono, K. Yokomori, A. Grosberg,
Striated myocyte structural integrity: Automated analysis of sarcomeric z-discs, PLOS
Computational Biology 16(3) (2020) e1007676.

[91] A. Grosberg, P.W. Alford, M.L. McCain, K.K. Parker, Ensembles of engineered
cardiac tissues for physiological and pharmacological study: Heart on a chip, Lab on a
Chip 11(24) (2011) 4165-4173.

[92] A.W. Feinberg, A. Feigel, S.S. Shevkoplyas, S. Sheehy, G.M. Whitesides, K.K.
Parker, Muscular Thin Films for Building Actuators and Powering Devices, Science
317(5843) (2007) 1366-1370.
120

[93] M. Costantini, S. Testa, E. Fornetti, A. Barbetta, M. Trombetta, S.M. Cannata, C.
Gargioli, A. Rainer, Engineering Muscle Networks in 3D Gelatin Methacryloyl Hydrogels:
Influence of Mechanical Stiffness and Geometrical Confinement, Frontiers in
bioengineering and biotechnology 5 (2017) 22-22.

[94] M. Battistelli, S. Salucci, S. Burattini, E. Falcieri, Further considerations on in vitro
skeletal muscle cell death, Muscles Ligaments Tendons J 3(4) (2013) 267-74.

[95] G. Stelzer, N. Rosen, I. Plaschkes, S. Zimmerman, M. Twik, S. Fishilevich, T.I. Stein,
R. Nudel, I. Lieder, Y. Mazor, S. Kaplan, D. Dahary, D. Warshawsky, Y. Guan-Golan, A.
Kohn, N. Rappaport, M. Safran, D. Lancet, The GeneCards Suite: From Gene Data
Mining to Disease Genome Sequence Analyses, Current Protocols in Bioinformatics
54(1) (2016) 1.30.1-1.30.33.

[96] S. Calve, J. Isaac, J.P. Gumucio, C.L. Mendias, Hyaluronic acid, HAS1, and HAS2
are significantly upregulated during muscle hypertrophy, Am J Physiol Cell Physiol 303(5)
(2012) C577-88.

[97] B. Della Gaspera, A. Chesneau, L. Weill, F. Charbonnier, C. Chanoine, Xenopus
SOX5 enhances myogenic transcription indirectly through transrepression,
Developmental Biology 442(2) (2018) 262-275.

[98] D. Joulia, H. Bernardi, V. Garandel, F. Rabenoelina, B. Vernus, G. Cabello,
Mechanisms involved in the inhibition of myoblast proliferation and differentiation by
myostatin, Exp Cell Res 286(2) (2003) 263-75.

[99] D. Montarras, J. Chelly, E. Bober, H. Arnold, M.O. Ott, F. Gros, C. Pinset,
Developmental patterns in the expression of Myf5, MyoD, myogenin, and MRF4 during
myogenesis, New Biol 3(6) (1991) 592-600.

[100] G.A. Ordway, D.J. Garry, Myoglobin: an essential hemoprotein in striated muscle,
Journal of Experimental Biology 207(20) (2004) 3441-3446.

[101] N.J. Lamas, B. Johnson-Kerner, L. Roybon, Y.A. Kim, A. Garcia-Diaz, H. Wichterle,
C.E. Henderson, Neurotrophic Requirements of Human Motor Neurons Defined Using
Amplified and Purified Stem Cell-Derived Cultures, PLOS ONE 9(10) (2014) e110324.
[102] S.J. Burden, N. Yumoto, W. Zhang, The role of MuSK in synapse formation and
neuromuscular disease, Cold Spring Harbor perspectives in biology 5(5) (2013) a009167-
a009167.

[103] A. Heikkinen, H. Härönen, O. Norman, T. Pihlajaniemi, Collagen XIII and Other ECM
Components in the Assembly and Disease of the Neuromuscular Junction, Anat Rec
(Hoboken) 303(6) (2020) 1653-1663.

121

[104] R.W. Burgess, Q.T. Nguyen, Y.-J. Son, J.W. Lichtman, J.R. Sanes, Alternatively
Spliced Isoforms of Nerve- and Muscle-Derived Agrin: Their Roles at the Neuromuscular
Junction, Neuron 23(1) (1999) 33-44.

[105] I. Bourdeau Julien, C.F. Sephton, P.A. Dutchak, Metabolic Networks Influencing
Skeletal Muscle Fiber Composition, Frontiers in cell and developmental biology 6 (2018)
125-125.

[106] B. Li, M. Lin, Y. Tang, B. Wang, J.H. Wang, A novel functional assessment of the
differentiation of micropatterned muscle cells, J Biomech 41(16) (2008) 3349-53.

[107] A.S.T. Smith, C.J. Long, K. Pirozzi, S. Najjar, C. McAleer, H.H. Vandenburgh, J.J.
Hickman, A multiplexed chip-based assay system for investigating the functional
development of human skeletal myotubes in vitro, Journal of Biotechnology 185 (2014)
15-18.

[108] K.L. Pirozzi, C.J. Long, C.W. McAleer, A.S.T. Smith, J.J. Hickman, Correlation of
embryonic skeletal muscle myotube physical characteristics with contractile force
generation on an atomic force microscope-based bio-microelectromechanical systems
device, Applied Physics Letters 103(8) (2013) 083108.

[109] C. Lindon, D. Montarras, C. Pinset, Cell Cycle–regulated Expression of the Muscle
Determination Factor Myf5 in Proliferating Myoblasts, Journal of Cell Biology 140(1)
(1998) 111-118.

[110] S. Biressi, C.R.R. Bjornson, P.M.M. Carlig, K. Nishijo, C. Keller, T.A. Rando, Myf5
expression during fetal myogenesis defines the developmental progenitors of adult
satellite cells, Developmental biology 379(2) (2013) 195-207.

[111] D. Bloemberg, J. Quadrilatero, Rapid determination of myosin heavy chain
expression in rat, mouse, and human skeletal muscle using multicolor
immunofluorescence analysis, PLoS One 7(4) (2012) e35273.

[112] N.R. Martin, S.L. Passey, D.J. Player, A. Khodabukus, R.A. Ferguson, A.P.
Sharples, V. Mudera, K. Baar, M.P. Lewis, Factors affecting the structure and maturation
of human tissue engineered skeletal muscle, Biomaterials 34(23) (2013) 5759-65.

[113] D. Quiat, K.A. Voelker, J. Pei, N.V. Grishin, R.W. Grange, R. Bassel-Duby, E.N.
Olson, Concerted regulation of myofiber-specific gene expression and muscle
performance by the transcriptional repressor Sox6, Proceedings of the National Academy
of Sciences of the United States of America 108(25) (2011) 10196-10201.

[114] R. Schoenauer, S. Lange, A. Hirschy, E. Ehler, J.-C. Perriard, I. Agarkova,
Myomesin 3, a Novel Structural Component of the M-band in Striated Muscle, Journal of
Molecular Biology 376(2) (2008) 338-351.
122

[115] B. Ghosh, Z. Wang, J. Nong, M.W. Urban, Z. Zhang, V.A. Trovillion, M.C. Wright,
Y. Zhong, A.C. Lepore, Local BDNF Delivery to the Injured Cervical Spinal Cord using an
Engineered Hydrogel Enhances Diaphragmatic Respiratory Function, J Neurosci 38(26)
(2018) 5982-5995.

[116] V.B. Matthews, M.B. Åström, M.H.S. Chan, C.R. Bruce, K.S. Krabbe, O. Prelovsek,
T. Åkerström, C. Yfanti, C. Broholm, O.H. Mortensen, M. Penkowa, P. Hojman, A.
Zankari, M.J. Watt, H. Bruunsgaard, B.K. Pedersen, M.A. Febbraio, Brain-derived
neurotrophic factor is produced by skeletal muscle cells in response to contraction and
enhances fat oxidation via activation of AMP-activated protein kinase, Diabetologia 52(7)
(2009) 1409-1418.

[117] D.E. Kelley, Skeletal muscle fat oxidation: timing and flexibility are everything, The
Journal of clinical investigation 115(7) (2005) 1699-1702.

[118] W. Song, X.A. Jin, Brain-derived neurotrophic factor inhibits neuromuscular junction
maturation in a cAMP-PKA-dependent way, Neuroscience Letters 591 (2015) 8-12.
[119] P.P. Li, R. Madhavan, H.B. Peng, Differential regulation of axonal growth and
neuromuscular junction assembly by HGF/c-Met signaling, Dev Dyn 241(10) (2012) 1562-
74.

[120] Y. Zong, B. Zhang, S. Gu, K. Lee, J. Zhou, G. Yao, D. Figueiredo, K. Perry, L. Mei,
R. Jin, Structural basis of agrin-LRP4-MuSK signaling, Genes & development 26(3)
(2012) 247-258.

[121] P.P. Li, C. Chen, C.W. Lee, R. Madhavan, H.B. Peng, Axonal filopodial asymmetry
induced by synaptic target, Mol Biol Cell 22(14) (2011) 2480-90.

[122] P.P. Li, J.J. Zhou, M. Meng, R. Madhavan, H.B. Peng, Reciprocal Regulation of
Axonal Filopodia and Outgrowth during Neuromuscular Junction Development, PLOS
ONE 7(9) (2012) e44759.

[123] G. Bezakova, J.P. Helm, M. Francolini, T. Lømo, Effects of purified recombinant
neural and muscle agrin on skeletal muscle fibers in vivo, The Journal of cell biology
153(7) (2001) 1441-1452.

[124] F. Carlo Gotti and Alberto Sensini and Andrea Zucchelli and Raffaella Carloni and
Maria Letizia, Hierarchical fibrous structures for muscle ‐inspired soft ‐actuators: A review,
Applied Materials Today 20 (2020) 100772.

[125] A.R. Gillies, R.L. Lieber, Structure and function of the skeletal muscle extracellular
matrix, Muscle & Nerve 44(3) (2011) 318-331.

[126] V. Mudera, A.S. Smith, M.A. Brady, M.P. Lewis, The effect of cell density on the
maturation and contractile ability of muscle derived cells in a 3D tissue-engineered
123

skeletal muscle model and determination of the cellular and mechanical stimuli required
for the synthesis of a postural phenotype, J Cell Physiol 225(3) (2010) 646-53.

[127] S. Chiron, C. Tomczak, A. Duperray, J. Lainé, G. Bonne, A. Eder, A. Hansen, T.
Eschenhagen, C. Verdier, C. Coirault, Complex interactions between human myoblasts
and the surrounding 3D fibrin-based matrix, PloS one 7(4) (2012) e36173-e36173.

[128] M.J. Post, S. Levenberg, D.L. Kaplan, N. Genovese, J. Fu, C.J. Bryant, N.
Negowetti, K. Verzijden, P. Moutsatsou, Scientific, sustainability and regulatory
challenges of cultured meat, Nature Food 1(7) (2020) 403-415.

[129] A. Khodabukus, L. Madden, N.K. Prabhu, T.R. Koves, C.P. Jackman, D.M. Muoio,
N. Bursac, Electrical stimulation increases hypertrophy and metabolic flux in tissue-
engineered human skeletal muscle, Biomaterials 198 (2019) 259-269.

[130] A. Khodabukus, Tissue-Engineered Skeletal Muscle Models to Study Muscle
Function, Plasticity, and Disease, Frontiers in Physiology 12 (2021).

[131] K. Shimizu, H. Araki, K. Sakata, W. Tonomura, M. Hashida, S. Konishi, Microfluidic
devices for construction of contractile skeletal muscle microtissues, J Biosci Bioeng
119(2) (2015) 212-6.

[132] R. Raman, C. Cvetkovic, S.G.M. Uzel, R.J. Platt, P. Sengupta, R.D. Kamm, R.
Bashir, Optogenetic skeletal muscle-powered adaptive biological machines, Proceedings
of the National Academy of Sciences 113(13) (2016) 3497-3502.

[133] K.B. Santosa, A.M. Keane, A. Jablonka-Shariff, B. Vannucci, A.K. Snyder-Warwick,
Clinical relevance of terminal Schwann cells: An overlooked component of the
neuromuscular junction, Journal of Neuroscience Research 96(7) (2018) 1125-1135.

[134] M. Juhas, N. Abutaleb, J.T. Wang, J. Ye, Z. Shaikh, C. Sriworarat, Y. Qian, N.
Bursac, Incorporation of macrophages into engineered skeletal muscle enables
enhanced muscle regeneration, Nature Biomedical Engineering (2018) 1-13.

[135] L. Shi, A.K. Fu, N.Y. Ip, Molecular mechanisms underlying maturation and
maintenance of the vertebrate neuromuscular junction, Trends Neurosci 35(7) (2012)
441-53.

[136] M. Bucchia, S.J. Merwin, D.B. Re, S. Kariya, Limitations and Challenges in Modeling
Diseases Involving Spinal Motor Neuron Degeneration in Vitro, Frontiers in Cellular
Neuroscience 12(61) (2018).

[137] K. Okita, Y. Matsumura, Y. Sato, A. Okada, A. Morizane, S. Okamoto, H. Hong, M.
Nakagawa, K. Tanabe, K.-i. Tezuka, T. Shibata, T. Kunisada, M. Takahashi, J.
Takahashi, H. Saji, S. Yamanaka, A more efficient method to generate integration-free
human iPS cells, Nature Methods 8(5) (2011) 409.
124

[138] X. Chen, W. Ma, S. Zhang, J. Paluch, W. Guo, D.K. Dickman, The BLOC-1 Subunit
Pallidin Facilitates Activity-Dependent Synaptic Vesicle Recycling, eNeuro 4(1) (2017)
ENEURO.0335-16.2017.

[139] K. Saito, T. Kobayashi, V. Askanas, W.K. Engel, K. Ishikawa, ELECTRICAL
PROPERTIES OF HUMAN MUSCLE CULTURED IN MONOLAYER ANEURALLY AND
COCULTURED WITH FETAL RAT SPINAL CORD , Biomedical Research
11(1) (1990) 19-28.

[140] S. Gao, M. Zhen, Action potentials drive body wall muscle contractions in
Caenorhabditis elegans, Proceedings of the National Academy of Sciences
108(6) (2011) 2557-2562.

[141] X. Li, P. Goel, J. Wondolowski, J. Paluch, D. Dickman, A Glutamate Homeostat
Controls the Presynaptic Inhibition of Neurotransmitter Release, Cell reports. 23(6) (2018)
1716-1727.

[142] J.E. Hall, Guyton and Hall textbook of medical physiology, Guyton and Hall textbook
of medical physiology (2016).

[143] J. Talbot, L. Maves, Skeletal muscle fiber type: using insights from muscle
developmental biology to dissect targets for susceptibility and resistance to muscle
disease, Wiley interdisciplinary reviews. Developmental biology 5(4) (2016) 518-534.

[144] A.S.T. Smith, S.L. Passey, N.R.W. Martin, D.J. Player, V. Mudera, L. Greensmith,
M.P. Lewis, Creating Interactions between Tissue-Engineered Skeletal Muscle and the
Peripheral Nervous System, Cells Tissues Organs 202(3-4) (2016) 143-158.

[145] X. Guo, V. Smith, M. Jackson, M. Tran, M. Thomas, A. Patel, E. Lorusso, S.
Nimbalkar, Y. Cai, C.W. McAleer, Y. Wang, C.J. Long, J.J. Hickman, A Human-Based
Functional NMJ System for Personalized ALS Modeling and Drug Testing, Advanced
Therapeutics n/a(n/a) (2020) 2000133.

[146] T. Osaki, S.G.M. Uzel, R.D. Kamm, On-chip 3D neuromuscular model for drug
screening and precision medicine in neuromuscular disease, Nature Protocols 15(2)
(2020) 421-449.

[147] R.R. Besser, A.C. Bowles, A. Alassaf, D. Carbonero, I. Claure, E. Jones, J. Reda,
L. Wubker, W. Batchelor, N. Ziebarth, R. Silvera, A. Khan, R. Maciel, M. Saporta, A.
Agarwal, Enzymatically crosslinked gelatin–laminin hydrogels for applications in
neuromuscular tissue engineering, Biomaterials Science 8(2) (2020) 591-606.

[148] Julius A. Steinbeck, Manoj K. Jaiswal, Elizabeth L. Calder, S. Kishinevsky, A.
Weishaupt, Klaus V. Toyka, Peter A. Goldstein, L. Studer, Functional Connectivity under
Optogenetic Control Allows Modeling of Human Neuromuscular Disease, Cell Stem Cell
18(1) (2016) 134-143.
125

[149] M.J. Dennis, L. Ziskind-Conhaim, A.J. Harris, Development of neuromuscular
junctions in rat embryos, Developmental Biology 81(2) (1981) 266-279.

[150] C.B. Mantilla, J.M. Stowe, D.C. Sieck, L.G. Ermilov, S.M. Greising, C. Zhang, K.M.
Shokat, G.C. Sieck, TrkB kinase activity maintains synaptic function and structural
integrity at adult neuromuscular junctions, Journal of Applied Physiology 117(8) (2014)
910-920.

[151] T. Kobayashi, V. Askanas, W. Engel, Human muscle cultured in monolayer and
cocultured with fetal rat spinal cord: importance of dorsal root ganglia for achieving
successful functional innervation, The Journal of Neuroscience 7(10) (1987) 3131-3141.
[152] M. Weitkunat, M. Brasse, A.R. Bausch, F. Schnorrer, Mechanical tension and
spontaneous muscle twitching precede the formation of cross-striated muscle in vivo,
Development (Cambridge, England) 144(7) (2017) 1261-1272.

[153] P.G. De Deyne, Formation of sarcomeres in developing myotubes: role of
mechanical stretch and contractile activation, American Journal of Physiology-Cell
Physiology 279(6) (2000) C1801-C1811.

[154] Z. Chen, B. Li, R.Z. Zhan, L. Rao, N. Bursac, Exercise mimetics and JAK inhibition
attenuate IFN-γ-induced wasting in engineered human skeletal muscle, Sci Adv 7(4)
(2021).

[155] R.P. Rimington, J.W. Fleming, A.J. Capel, P.C. Wheeler, M.P. Lewis, Bioengineered
model of the human motor unit with physiologically functional neuromuscular junctions,
Scientific Reports 11(1) (2021) 11695.

[156] S. Schiaffino, C. Reggiani, Fiber types in mammalian skeletal muscles, Physiol Rev
91(4) (2011) 1447-531.

[157] R. Tatsumi, S.M. Sheehan, H. Iwasaki, A. Hattori, R.E. Allen, Mechanical Stretch
Induces Activation of Skeletal Muscle Satellite Cells in Vitro, Experimental Cell Research
267(1) (2001) 107-114.

[158] K.S. Katiyar, L.A. Struzyna, S. Das, D.K. Cullen, Stretch growth of motor axons in
custom mechanobioreactors to generate long-projecting axonal constructs, J Tissue Eng
Regen Med 13(11) (2019) 2040-2054.

[159] R. Nilwik, T. Snijders, M. Leenders, B.B. Groen, J. van Kranenburg, L.B. Verdijk,
L.J. van Loon, The decline in skeletal muscle mass with aging is mainly attributed to a
reduction in type II muscle fiber size, Exp Gerontol 48(5) (2013) 492-8.

[160] J.M. Wilson, J.P. Loenneke, E. Jo, G.J. Wilson, M.C. Zourdos, J.S. Kim, The effects
of endurance, strength, and power training on muscle fiber type shifting, J Strength Cond
Res 26(6) (2012) 1724-9.
126

[161] T. Mars, K.J. Yu, X.M. Tang, A.F. Miranda, Z. Grubic, F. Cambi, M.P. King,
Differentiation of glial cells and motor neurons during the formation of neuromuscular
junctions in cocultures of rat spinal cord explant and human muscle, J Comp Neurol
438(2) (2001) 239-51.

[162] W. Zheng, Q. Li, C. Zhao, Y. Da, H.-L. Zhang, Z. Chen, Differentiation of Glial Cells
From hiPSCs: Potential Applications in Neurological Diseases and Cell Replacement
Therapy, Frontiers in Cellular Neuroscience 12(239) (2018).

[163] V. Soubannier, G. Maussion, M. Chaineau, V. Sigutova, G. Rouleau, T.M. Durcan,
S. Stifani, Characterization of human iPSC-derived astrocytes with potential for disease
modeling and drug discovery, Neuroscience Letters 731 (2020) 135028.

[164] X. Guo, M. Gonzalez, M. Stancescu, H.H. Vandenburgh, J.J. Hickman,
Neuromuscular junction formation between human stem cell-derived motoneurons and
human skeletal muscle in a defined system, Biomaterials 32(36) (2011) 9602-11.

[165] K.E. Personius, B.S. Slusher, S.B. Udin, Neuromuscular NMDA Receptors
Modulate Developmental Synapse Elimination, J Neurosci 36(34) (2016) 8783-9.

[166] X. Guo, K. Greene, N. Akanda, A. Smith, M. Stancescu, S. Lambert, H.
Vandenburgh, J. Hickman, In vitro Differentiation of Functional Human Skeletal Myotubes
in a Defined System, Biomaterials science 2(1) (2014) 131-138.

[167] P. Molnar, W. Wang, A. Natarajan, J.W. Rumsey, J.J. Hickman, Photolithographic
patterning of C2C12 myotubes using vitronectin as growth substrate in serum-free
medium, Biotechnology progress 23(1) (2007) 265-268.

[168] R. Besser, R. Maciel, M. Dvornik, D. Carbonero, I. Claure, E. Jones, J. Reda, M.
Saporta, A. Agarwal, Micromolded gelatin hydrogels crosslinked with laminin for disease
on a chip applications, Transactions of the Annual Meeting of the Society for Biomaterials
and the Annual International Biomaterials Symposium, 2019, p. 118.

[169] C.C. Lee, A. Hoang, D. Segovia, A. Herbst, F. Barthelemy, E. Gibbs, R. Crosbie,
S.F. Nelson, C. Miceli, J. Wanagat, Enhanced Methods for Needle Biopsy and
Cryopreservation of Skeletal Muscle in Older Adults, Journal of cytology & histology 11(2)
(2020) 10.37421/jch.2020.11.553.

[170] A. Shahini, K. Vydiam, D. Choudhury, N. Rajabian, T. Nguyen, P. Lei, S.T.
Andreadis, Efficient and high yield isolation of myoblasts from skeletal muscle, Stem Cell
Research 30 (2018) 122-129.

[171] N. Jiwlawat, E. Lynch, J. Jeffrey, J.M. Van Dyke, M. Suzuki, Current Progress and
Challenges for Skeletal Muscle Differentiation from Human Pluripotent Stem Cells Using
Transgene-Free Approaches, Stem cells international 2018 (2018) 6241681-6241681.
127

[172] J. Lainé, G. Skoglund, E. Fournier, N. Tabti, Development of the excitation-
contraction coupling machinery and its relation to myofibrillogenesis in human iPSC-
derived skeletal myocytes, Skeletal Muscle 8(1) (2018) 1.

[173] S. Chaouch, V. Mouly, A. Goyenvalle, A. Vulin, K. Mamchaoui, E. Negroni, J. Di
Santo, G. Butler-Browne, Y. Torrente, L. Garcia, D. Furling, Immortalized skin fibroblasts
expressing conditional MyoD as a renewable and reliable source of converted human
muscle cells to assess therapeutic strategies for muscular dystrophies: validation of an
exon-skipping approach to restore dystrophin in Duchenne muscular dystrophy cells,
Hum. Gene Ther. 20(7) (2009) 784-90.

[174] E.Y. Kim, P. Page, L.M. Dellefave-Castillo, E.M. McNally, E.J. Wyatt, Direct
reprogramming of urine-derived cells with inducible MyoD for modeling human muscle
disease, Skelet Muscle 6 (2016) 32.

[175] N. Wein, A. Vulin, A.R. Findlay, F. Gumienny, N. Huang, S.D. Wilton, K.M. Flanigan,
Efficient Skipping of Single Exon Duplications in DMD Patient-Derived Cell Lines Using
an Antisense Oligonucleotide Approach, J Neuromuscul Dis 4(3) (2017) 199-207.

[176] G.C. Kendall, E.I. Mokhonova, M. Moran, N.E. Sejbuk, D.W. Wang, O. Silva, R.T.
Wang, L. Martinez, Q.L. Lu, R. Damoiseaux, M.J. Spencer, S.F. Nelson, M.C. Miceli,
Dantrolene enhances antisense-mediated exon skipping in human and mouse models of
Duchenne muscular dystrophy, Sci. Transl. Med. 4(164) (2012) 164ra160.

[177] F. Barthelemy, R.T. Wang, C. Hsu, E.D. Douine, E.E. Marcantonio, S.F. Nelson,
M.C. Miceli, Targeting RyR Activity Boosts Antisense Exon 44 and 45 Skipping in Human
DMD Skeletal or Cardiac Muscle Culture Models, Mol Ther Nucleic Acids 18 (2019) 580-
589.

[178] F. Barthélémy, J.W. Santoso, L. Rabichow, R. Jin, I. Little, S.F. Nelson, M.L.
McCain, M.C. Miceli, Modeling Patient-Specific Muscular Dystrophy Phenotypes and
Therapeutic Responses in Reprogrammed Myotubes Engineered on Micromolded
Gelatin Hydrogels, Frontiers in Cell and Developmental Biology 10 (2022).

[179] E.M. Gibbs, F. Barthelemy, E.D. Douine, N.C. Hardiman, P.B. Shieh, N. Khanlou,
R.H. Crosbie, S.F. Nelson, M.C. Miceli, Large in-frame 5' deletions in DMD associated
with mild Duchenne muscular dystrophy: Two case reports and a review of the literature,
Neuromuscul. Disord. 29(11) (2019) 863-873.

[180] R.T. Wang, F. Barthelemy, A.S. Martin, E.D. Douine, A. Eskin, A. Lucas, J. Lavigne,
H. Peay, N. Khanlou, L. Sweeney, R.M. Cantor, M.C. Miceli, S.F. Nelson, DMD genotype
correlations from the Duchenne Registry: Endogenous exon skipping is a factor in
prolonged ambulation for individuals with a defined mutation subtype, Hum. Mutat.
(2018).

128

[181] K.S. ARPAN D BHATT1, APURVA PUVAR, CHAITANYA G JOSHI, MADHVI
JOSHI, A Case of Limb Girdle Muscular Dystrophy Type 2A from India: Copy Number
Variation Analysis using Targeted Amplicon Sequencing, Journal of Clinical and
Diagnostic Research. 13(4) (2019).

[182] M. Krahn, R. Bernard, C. Pecheux, H. Hammouda el, B. Eymard, A. Lopez de
Munain, A.M. Cobo, N. Romero, A. Urtizberea, F. Leturcq, N. Levy, L.N. Calpain Study
Group of the French, Screening of the CAPN3 gene in patients with possible LGMD2A,
Clin. Genet. 69(5) (2006) 444-9.

[183] B.R.R. Nallamilli, S. Chakravorty, A. Kesari, A. Tanner, A. Ankala, T. Schneider, C.
da Silva, R. Beadling, J.J. Alexander, S.H. Askree, Z. Whitt, L. Bean, C. Collins, S.
Khadilkar, P. Gaitonde, R. Dastur, M. Wicklund, T. Mozaffar, M. Harms, L. Rufibach, P.
Mittal, M. Hegde, Genetic landscape and novel disease mechanisms from a large LGMD
cohort of 4656 patients, Ann Clin Transl Neurol 5(12) (2018) 1574-1587.

[184] F. Barthelemy, J.D. Woods, S. Nieves-Rodriguez, E.D. Douine, R. Wang, J.
Wanagat, M.C. Miceli, S.F. Nelson, A well-tolerated core needle muscle biopsy process
suitable for children and adults, Muscle Nerve 62(6) (2020) 688-698.

[185] C.C. Lee, A. Hoang, D. Segovia, A. Herbst, F. Barthelemy, E. Gibbs, R. Crosbie,
S.F. Nelson, C. Miceli, J. Wanagat, Enhanced Methods for Needle Biopsy and
Cryopreservation of Skeletal Muscle in Older Adults, J Cytol Histol 11(2) (2020).

[186] F. Barthelemy, D. Wang, S.F. Nelson, M.C. Miceli, Validation and Detection of Exon
Skipping Boosters in DMD Patient Cell Models and mdx Mouse, Methods Mol. Biol. 1828
(2018) 309-326.

[187] S.D. Wilton, A.M. Fall, P.L. Harding, G. McClorey, C. Coleman, S. Fletcher,
Antisense oligonucleotide-induced exon skipping across the human dystrophin gene
transcript, Mol. Ther. 15(7) (2007) 1288-96.

[188] M.L. McCain, S.P. Sheehy, A. Grosberg, J.A. Goss, K.K. Parker, Recapitulating
maladaptive, multiscale remodeling of failing myocardium on a chip., Proceedings of the
National Academy of Sciences of the United States of America 110(24) (2013) 9770--5.

[189] A.P. Petersen, N. Cho, D.M. Lyra-Leite, J.W. Santoso, D. Gupta, N.R. Ariyasinghe,
M.L. McCain, Regulation of calcium dynamics and propagation velocity by tissue
microstructure in engineered strands of cardiac tissue, Integrative Biology 12(2) (2020)
34-46.

[190] L. Caputo, A. Granados, J. Lenzi, A. Rosa, S. Ait-Si-Ali, P.L. Puri, S. Albini, Acute
conversion of patient-derived Duchenne muscular dystrophy iPSC into myotubes reveals
constitutive and inducible over-activation of TGFβ-dependent pro-fibrotic signaling,
Skeletal Muscle 10(1) (2020) 13.
129

[191] C. Ortiz-Cordero, C. Bincoletto, N.R. Dhoke, S. Selvaraj, A. Magli, H. Zhou, D.-H.
Kim, A.G. Bang, R.C.R. Perlingeiro, Defective autophagy and increased apoptosis
contribute toward the pathogenesis of FKRP-associated muscular dystrophies, Stem Cell
Reports 16(11) (2021) 2752-2767.

[192] O. Sheikh, T. Yokota, Developing DMD therapeutics: a review of the effectiveness
of small molecules, stop-codon readthrough, dystrophin gene replacement, and exon-
skipping therapies, Expert Opin Investig Drugs 30(2) (2021) 167-176.
[193] I.N. Rybakova, K.J. Amann, J.M. Ervasti, A new model for the interaction of
dystrophin with F-actin, J. Cell Biol. 135(3) (1996) 661-72.

[194] I.N. Rybakova, J.M. Ervasti, Dystrophin-glycoprotein complex is monomeric and
stabilizes actin filaments in vitro through a lateral association, J. Biol. Chem. 272(45)
(1997) 28771-8.

[195] G.B. Banks, P. Gregorevic, J.M. Allen, E.E. Finn, J.S. Chamberlain, Functional
capacity of dystrophins carrying deletions in the N-terminal actin-binding domain, Hum.
Mol. Genet. 16(17) (2007) 2105-13.

[196] S.M. Boularaoui, K.M.A. Abdel-Raouf, N.S.A. Alwahab, M.E. Kondash, G.A.
Truskey, J.C.M. Teo, N. Christoforou, Efficient transdifferentiation of human dermal
fibroblasts into skeletal muscle, Journal of Tissue Engineering and Regenerative
Medicine 12(2) (2018) e918-e936.

[197] T. Nedachi, H. Fujita, M. Kanzaki, Contractile C2C12 myotube model for studying
exercise-inducible responses in skeletal muscle, Am J Physiol Endocrinol Metab 295(5)
(2008) E1191-204.

[198] Y.J. Chang, Y.J. Chen, C.W. Huang, S.C. Fan, B.M. Huang, W.T. Chang, Y.S. Tsai,
F.C. Su, C.C. Wu, Cyclic Stretch Facilitates Myogenesis in C2C12 Myoblasts and
Rescues Thiazolidinedione-Inhibited Myotube Formation, Front Bioeng Biotechnol 4
(2016) 27.

[199] P. Heher, B. Maleiner, J. Prüller, A.H. Teuschl, J. Kollmitzer, X. Monforte, S.
Wolbank, H. Redl, D. Rünzler, C. Fuchs, A novel bioreactor for the generation of highly
aligned 3D skeletal muscle-like constructs through orientation of fibrin via application of
static strain, Acta Biomaterialia 24 (2015) 251--265.

[200] S.G.M. Uzel, R.J. Platt, V. Subramanian, T.M. Pearl, C.J. Rowlands, V. Chan, L.A.
Boyer, P.T.C. So, R.D. Kamm, Microfluidic device for the formation of optically excitable,
three-dimensional, compartmentalized motor units, Science Advances 2(8) (2016)
e1501429.

[201] T.N.a.E.J.a.L.A.M.a.T.N.a.B. Majid Ebrahimi and Heta Lad and Aurora Fusto and
Yekaterina Tiper and Asiman Datye and Christine, De novo revertant fiber formation and
130

therapy testing in a 3D culture model of Duchenne muscular dystrophy skeletal muscle,
Acta Biomaterialia 132 (2021) 227-244.

[202] J. White, M.V. Barro, H.P. Makarenkova, J.W. Sanger, J.M. Sanger, Localization of
sarcomeric proteins during myofibril assembly in cultured mouse primary skeletal
myotubes, Anat Rec (Hoboken) 297(9) (2014) 1571-84.

[203] N.R. Rubio, N. Xiang, D.L. Kaplan, Plant-based and cell-based approaches to meat
production, Nature communications 11(1) (2020) 6276-6276.

[204] Y. Komiya, W. Mizunoya, K. Kajiwara, I. Yokoyama, H. Ogasawara, K. Arihara,
Correlation between skeletal muscle fiber type and responses of a taste sensing system
in various beef samples, Animal Science Journal 91(1) (2020) e13425.

[205] A. Kumar, A. Accorsi, Y. Rhee, M. Girgenrath, Do's and don'ts in the preparation of
muscle cryosections for histological analysis, Journal of visualized experiments : JoVE
(99) (2015) e52793-e52793.

[206] W.L. Zhang, S.H. Liu, W.C. Zhang, W. Hu, M. Jiang, A. Tamadon, Y. Feng, Skeletal
Muscle CLARITY: A Preliminary Study of Imaging The Three-Dimensional Architecture
of Blood Vessels and Neurons, Cell J 20(2) (2018) 132-137.

[207] J. Sabry, T.P. O'Connor, M.W. Kirschner, Axonal transport of tubulin in Ti1 pioneer
neurons in situ, Neuron 14(6) (1995) 1247-56.

[208] C. Hagemann, C. Moreno Gonzalez, L. Guetta, G. Tyzack, C. Chiappini, A. Legati,
R. Patani, A. Serio, Axonal Length Determines Distinct Homeostatic Phenotypes in
Human iPSC Derived Motor Neurons on a Bioengineered Platform, Advanced Healthcare
Materials n/a(n/a) 2101817.

Abstract (if available)

Abstract Neuromuscular diseases involve degeneration of the motor unit, which includes the motor neuron and the skeletal muscle fibers they innervate at the neuromuscular junction (NMJ). These diseases vary widely in pathogenesis, but generally start with ambulatory deficits that increase in severity. Compounded with ensuing cardiac and respiratory complications, these diseases result in low quality of life and, often, ultimately death. Physiologically relevant in vitro models of skeletal muscle tissue are critically needed to probe the mechanisms of disease progression on a cellular and molecular level. However, traditional attempts to culture skeletal muscle in vitro have resulted in unaligned, immature myotubes with limited functionality, preventing deeper investigation of key processes such as sarcomere development or motor neuron integration and the formation of neuromuscular junctions. Failure to reach a basic level of tissue maturity limits translatable conclusions when using these models to study neuromyopathies or muscular dystrophies.
Recently, advances in stem cell biology and micro- or nanofabrication techniques have been applied to better recapitulate the structure and function of a basic unit of a tissue or organ system. These platforms, referred to as microphysiological systems (MPS), act as an intermediate platform between in vivo and in vitro systems, allowing for medium throughput, relevant tissue structure, and quantifiable functional outputs. By balancing biological complexity of these engineered systems to recapitulate features of interest with simplicity and speed of fabrication, it is possible to maximize the efficiency of collecting relevant results for specific research questions. Our objective was to apply tissue engineering techniques to culture 2D and 3D skeletal muscle MPS towards implementing more translatable disease models, yielding data with increased clinical relevance.
First, we engineered and characterized functional skeletal muscle tissues in 2D on micromolded gelatin hydrogels and in 3D suspended in fibrin hydrogels. Specifically, we assessed muscle tissue density, fusion, and contractile force generation from several common commercial sources. Second, we developed and implemented robust coculture protocols for skeletal myotubes and human-induced pluripotent stem cell (hiPSC)-derived motor neurons. These engineered tissues exhibited distinct, organized co-localization of presynapse and acetylcholine receptors expected in developing NMJs and manifest spontaneous synaptic electrophysiological activities. Lastly, as proof-of-concept of translatability to clinical human relevance, we demonstrated the utility of our 2D gelatin hydrogel system to recapitulate human muscular dystrophy mutation-specific phenotypes and their responses to potential therapeutics.
Together, these developed technologies can be extended in the future to supplement 1) currently used disease models to better understand skeletal muscle or NMJ degeneration and 2) drug screening platforms for identifying potential therapeutics of neuromuscular diseases. Ideally, these tools will expediate the clogged therapeutic development pipeline by producing data with higher physiological relevance, potential subject specificity, and throughput. Moving forward, integration of neuromuscular MPS with other relevant cell types, gene editing technologies, or optogenetics may further expand throughput and common use of tissue-engineered constructs in the research space. Ultimately, these outcomes should prevent or relieve the negative impacts of neuromuscular disease, especially for an increasingly prone aging population.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Spatially controlled tissue differentiation using the synthetic receptor SynNotch

PDF

Engineering scalable two- and three-dimensional striated muscle microtissues for human disease modeling

PDF

New approaches for precisely engineering heterotypic muscle tissues by naturally and synthetically controlling cell fate

PDF

Application of small molecules as self-reprogramming agents for the reconstruction of musculoskeletal tissues within 3-D tissue-specific scaffolds

PDF

Engineering a post-infarct human myocardium on a chip to reveal oxygen-dependent crosstalk in cardiac fibroblasts and myocytes

PDF

Extracellular matrix regulation of mitochondrial function in engineered cardiac myocytes

PDF

2D ultrasonic transducer array’s design and fabrication with 3D printed interposer and applications

PDF

Membrane-bound regulation of hematopoietic stem cells

PDF

Engineering genetic tools to illustrate new insights into the homeostatic control of synaptic strength

PDF

Development and implementation of a modular muscle-computer interface for personalized motor rehabilitation after stroke

PDF

Physiologically-based modeling to predict monoclonal antibody pharmacokinetics in humans from physiochemical properties

PDF

Single-cell analysis with high frequency ultrasound

PDF

Metabolic profiling of single hematopoietic stem cells for developing novel ex vivo culture strategies

PDF

Transcriptomic maturation of developing human cone precursors in fetal and 3D hESC-derived tissues

PDF

Micelle nanoparticles for the detection and treatment of atherosclerosis

PDF

High resolution elastography in ophthalmology

PDF

Improving antitumor efficacy of chimeric antigen receptor-engineered immune cell therapy with synthetic biology and combination therapy approaches

PDF

Engineering solutions for biomaterials: self-assembly and surface-modification of polymers for clinical applications

PDF

Effects of air polishing for the treatment of peri-implant diseases: a systematic review and meta-analysis

PDF

Machine learning methods for 2D/3D shape retrieval and classification

Asset Metadata

Creator Santoso, Jeffrey Winston (author)

Core Title Engineering 2D & 3D microphysiological systems for interrogating skeletal muscle tissues

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Degree Conferral Date 2022-08

Publication Date 05/07/2022

Defense Date 04/21/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag coculture,disease modeling,hydrogel,neuromuscular junction,OAI-PMH Harvest,organoid,tissue engineering

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McCain, Megan Laura (committee chair), Dickman, Dion (committee member), Ichida, Justin (committee member), Morsut, Leonardo (committee member), Shen, Keyue (committee member)

Creator Email jeffreyw.santoso@gmail.com,jwsantos@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111296823

Unique identifier UC111296823

Document Type Dissertation

Format application/pdf (imt)

Rights Santoso, Jeffrey Winston

Type texts

Source 20220509-usctheses-batch-940 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu