University of Southern California Dissertations and Theses

Spatially controlled tissue differentiation using the synthetic receptor SynNotch

(USC Thesis Other)

Spatially controlled tissue differentiation using the synthetic receptor SynNotch

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2024 Mher Garibyan
SPATIALLY CONTROLLED TISSUE DIFFERENTIATION USING THE SYNTHETIC
RECEPTOR SYNNOTCH
by
MHER GARIBYAN
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 2024

ii
Dedication
This dissertation is dedicated to my parents who built the foundation that made this possible.

iii
Acknowledgements
I have been dealt a very favorable hand in the game of life, and I am thankful to God and
the universe for the many blessings. I am eternally grateful for the opportunities and more
importantly, the love and support from my friends and family who have made this all possible.
First, my parents. Your sacrifice for providing a proper upbringing for me and my brother
have been the base that I’ve built on. You inspire me and I aim to make you both proud in
everything I do. You are both so strong and have triumphed over obstacles that would have made
many people give up. To my brother, I am proud of the man you are becoming. You have more
potential than you realize. Keep growing. I love you all.
To my grandma, aunts, uncles, and many cousins. You have each played an important role
in my life. You’ve inspired me, given me friendship and care, and helped make me into the man
that I am. Thank you for all that you’ve done for me. This would not be possible without you all.
You can count on me as I have counted on you.
To my closest friends, the brotherhood we have is truly special. To call you my friends is
an understatement because I consider you brothers. We’ve known each other for over 15 years,
and I’m looking forward to the years to come. I am proud of what each of you are accomplishing
in your life and the vision you are building. Thank you for always being there when I need you.
To my lab mates in both the McCain and Morsut labs. It’s been an honor to work alongside
you and I have learned so much from you over the past 5 years. I want to thank Tyler Hoffman,
my co-first author on our paper, specifically, for being such a pleasure to work with. The hardest
experiments of our paper were made possible by your hard work and dependable planning.
To Dr. Megan McCain and Dr. Leonardo Morsut. It was an absolute pleasure to be a part
of your labs and to help bring our collective goals to life. The project was not easy, and many

iv
sacrifices had to be made to ensure that it was completed, but I am so proud of what we
accomplished together. You inspire me greatly and have pushed me to be a better student, scientist,
and most importantly, a better person.

v
Table of Contents
Dedication………………………………………………………………………………..………..ii
Acknowledgements………………………………………………………………………………iii
List of Tables…………………………………………………………………………………….vii
List of Figures……………………………………………………………………………..…….viii
Abbreviations……………………………………………………………………………….…..xvii
Abstract………………………………………………………………..………………………...xix
Chapter 1: Introduction
Tissue Engineering background…………………………………………………………...1
Animal Models…………………………………………………………………………….5
Organoids………………………………………………………………………………….7
Organ-on-a-chip Systems………………………………………………………………….9
Next-Generation Tissue Patterning Systems…………………………………………….12
Micropatterning…………………………………………………………………………..14
Microcontact Printing……………………………………………………………15
Synthetic biology………………………………………………………………………...17
Synthetic Receptors……………………………………………………………...20
SynNotch Background…………………………………………………………...23
Chapter 2: Engineering Programmable Material-To-Cell Pathways Via Synthetic Notch
Receptors To Spatially Control Cellular Phenotypes In Multi-Cellular Constructs
Abstract…………………………………………………………………………………..25
Results and Figures………………………………………………………………………26
Discussion………………………………………………………………………………..71
Materials and Methods…………………………………………………………………...79
Funding Sources and Acknowledgements……………………………………………….94
Chapter 3: Creating Engineered Muscle-Tendon-Bone Tissue for Regenerative Medicine
Introduction………………………………………………………………………………95
Results……………………………………………………………………………………97
Discussion and Future Directions………………………………………………………103
Materials and Methods………………………………………………………………….105
Chapter 4: Future work, outlook on the field, and concluding remarks

vi
Limitations and Future Work…………………………………………………………...107
Concluding Remarks……………………………………………………………………111
References……………………………………………………………………………………...112

vii
List of Tables
Table 1. DAVID pathway analysis and Assigned Identity of clusters. Pathways rank from 1-10,
1 being the pathway with most genes activated in its given pathway. Pathways in bold are
associated with the cluster's assigned identity. For cluster 0 and 2, identity related pathways
outside the top 10 were included with their rank in parenthesis. Each cluster was compared to the
gene expression of all clusters combined together. List of genes were selected through filtering
with P < 0.01 and Log2FC > 0.5.
Table 2. DAVID pathway analysis and comparison of UP and DOWN regulated pathways
between all muscle-like clusters. Pathways rank from 1-10, 1 being the pathway with most genes
UP or Down regulated between muscle-like clusters. Identity related pathways outside the top 10
were included with their rank in parenthesis. List of genes were selected through filtering with P
< 0.01 and Log2FC > 0.5.

viii
List of Figures
Figure 1. Treating PDMS coverslips with APTES and glutaraldehyde before FN coating enhances
cell adhesion. Brightfield images on Day 3 of UVO + FN treated PDMS coverslips vs APTES/Glut
+ FN treated coverslips. Both conditions were coated with FN at 50 µg/mL……………………..32
Figure 2. Wash step with media (DMEM + 10% FBS) before cell seeding reduces ligand mobility
and increases ligand to cell-activation colocalization. Image showing GFP channels prior to cell
seeding and downstream day 1 mCherry signal…………………………………………………..33
Figure 3. Microcontact printed GFP patterns spatially activate mCherry reporter via synNotch
pathway activation. (A) Visual schematic showing the process of stamp preparation and
concurrent coverslip preparation for the microcontact printing of GFP patterns. (B) Binary mask
of the stamp features that contains 100µm square features with 100µm gaps and resulting GFP
fluorescence image following microcontact printing (C) Schematic of anti-GFP/tTA synNotch
receiver fibroblasts seeded onto GFP-patterned substrate to demonstrate local activation based on
the presence of GFP. Portions of the cells were made transparent to visualize the underlying pattern
of GFP. (D) Binary masks and fluorescence microscopy images of 500µm and 250µm side square
GFP patterns, separated by 250, 350, 500, and 1000µm gaps, followed by mCherry fluorescence
images of anti-GFP synNotch receiver cells two days following seeding. Dotted white lines
separate regions with different interspaces and the solid white lines present the location of GFP
patterns enlarged in the following panel. (E) Plot profile of normalized mCherry intensity on 2, 5,
and 10 days following seeding onto 500x1000µm (top), 250x250µm (bottom) or GFP pattern
(square side x interspace). Green indicates the regions containing GFP. (F) Quantifications of day
2 Pearson’s Correlation Coefficient (PCC) comparing Binary Mask with mCherry channel across
all patterns. Scrambled binary mask images were compared to day 2 mCherry channels as a
negative control. Data represents mean ± s.d, n=7-8, not significant p>0.05 (ns), p<0.01(**),
p<0.0001(****). (G) Binary masks, brightfield microscopy images, and GFP fluorescence images
of concentric circles and letter patterns of GFP, followed by mCherry fluorescence microscopy
images taken two days following seeding onto the GFP pattern. Dotted white rectangles in the
brightfield indicate the region of interest shown in higher magnification on the bottom of the same
image. Scale bars, 1mm.………………………………………………………………………….36

ix
Figure S3. (A) Day 2 fluorescence images showing mCherry activation in anti-GFP reporter
fibroblasts on no GFP, 10μg/mL microcontact printed GFP, and 100μg/mL microcontact printed
GFP. Scale bars, 200μm. (B) Violin plot of mCherry intensity, quantified with flow cytometry,
dose response to 0, 10, 50, 100, and 200μg/mL concentrations of microcontact printed GFP after
48 hours. Dotted line indicates the threshold value to designate mCherry-positive cell. Percent of
mCherry expressing cells quantified by flow cytometry 48 hours after seeding onto GFP
microcontact printed substrate with varying GFP concentrations. Data represents mean ± s.d, n=6-
8, p<0.05(*) (C) Microcontact printed 100μm GFP squares with 100μm interspace length and
resulting mCherry activation signals taken two days after seeding. Scale bars, 1mm. (D)
Fluorescence images of mCherry activation from days 1-12 on 250 and 500μm width GFP squares
with varying interspace lengths. (E) Heatmap demonstrating Pearson correlation coefficient
between mCherry and GFP signals of each square width and interspace length at Days 2, 5, and
10. Color map represents the mean coefficient, n=7-8. (F) Brightfield images of anti-GFP reporter
fibroblasts 2 days following uniform seeding onto 500 and 250μm GFP squares. Scale bars,
500μm….………………………………..……………………………….………………………39
Figure 4. Patterned GFP and mCherry spatially activate respective reporter genes via synNotch
activation in dual receiver fibroblasts. (A) Schematic showcasing: (left) dual-ligand microcontact
printing, and (right) seeding of dual-receiver L929 cell where anti-GFP synNotch drives miRFP
reporter gene and anti-mCherry synNotch orthogonally activates BFP. (B) Fluorescence
microscopy images of microcontact printed GFP and mCherry perpendicular rows of 500µm width
and subsequent dual reporter expression taken 24 hours after uniform seeding. Scale bars, 500μm.
(Right) Normalized plot profiles of miRFP intensity across each row axis of engineered dualreceivers 24 hours following seeding onto perpendicular GFP and mCherry patterns. Green bars
indicate regions containing GFP. Line profiles represent mean ± s.d, n=7. (Below) Normalized plot
profiles of BFP intensity across each row axis of engineered dual-receivers 24 hours following
seeding onto perpendicular GFP and mCherry patterns. Red bars indicate regions containing
mCherry. Line profiles represent mean ± s.d, n=7. Regions of interest on mCherry (red border), on
GFP (green border), or on intersecting patterns (yellow border) enlarged in the following panels.
Higher magnification fluorescence microscopy images of GFP and mCherry perpendicular rows
and subsequent dual reporter expression/brightfield image. Regions of interest on GFP and

x
mCherry intersection (blue border) or non-patterned region (black border) enlarged in the
following panels. Scale bars, 500μm. (C) Higher magnification fluorescence microscopy images
to demonstrate BFP and miRFP expression by dual-receiver fibroblasts on mCherry, GFP,
intersection, and between patterns. Scale bars, 100μm. (D) Percent reporter activation
quantification on the regions containing no ligand, GFP only, mCherry only, and GFP and
mCherry. Results show the percent of cells that are not expressing BFP and miRFP, expressing
both BFP and miRFP, expressing BFP only, or expressing miRFP only, or. n = 4………………..43
Figure 5. Microcontact printed GFP patterns spatially activate myoD and initiate myotube
differentiation in embryonic fibroblasts via synNotch. (A) Schematic of embryonic fibroblasts
expressing anti-GFP synNotch activating myoD and mCherry transgenes seeded onto GFP
patterned PDMS substrate. (B) Sarcomeric α-actinin staining on isotropic GFP patterns on a PDMS
substrate, stained three days following seeding on GFP patterns. (-)GFP indicates image taken on
GFP-negative regions, (+)GFP indicates images taken within GFP-positive regions. Scale bars,
1mm. (C) Heatmap of hierarchical clustering of fibroblast parental cells without GFP, fibroblasts
engineered with anti-GFP synNotch activating myoD and mCherry with and without GFP, and
C2C12 myoblasts (skeletal muscle positive control). Z-Score is calculated by (Gene expression
value in sample of interest) - (Mean expression across all samples) / Standard Deviation. n=2-4.
(D) Volcano plot of gene expression data showing differentially expressed genes of anti-GFP
synNotch activating myoD and mCherry (anti-GFP/myoD) on GFP vs off GFP. n=2-4. (E) GO
term analysis showcasing enriched muscle pathways in myoD expressing synNotch receiver cells
on GFP vs off GFP. n= 2-4. (F) Fluorescence microscopy images of anti-GFP/myoD synNotch
receiver fibroblasts stained for α-actinin 4 days following seeding onto micromolded gelatin
substrates in the presence or absence of GFP. Scale bars, 500μm. (G) Myogenic index, quantified
with image analysis, in the presence or absence of GFP on isotropic or micromolded gelatin. Data
represents mean ± s.d, n=5, p<0.0001(****). Myotube alignment of α-actinin stained myotubes on
GFP-conjugated micromolded gelatin compared to GFP-conjugated isotropic gelatin substrate,
quantified by image analysis. Line plot represents average angles of orientation distribution of 5
individual images from an individual sample. Data represents mean ± s.d. (H) Binary mask used
to generate stamps for microcontact printing followed by fluorescence microscopy images of GFP
ligand and subsequent α-actinin staining for each pattern: 500µm curves, 200µm curves, 500µm

xi
rows, and 200µm rows. Samples were stained three days following uniform seeding onto GFP
patterns. (I) Day 3 Myogenic Index, quantified with image analysis, on and off GFP for each
pattern (isotropic, 200, 500 curves, 200, 500 straight rows). Data represents mean ± s.d, n=3-6,
p<0.01(**), p<0.001(***), p<0.0001(****). (J) Orientation Order Parameter measured across all
patterns quantified with image analysis. Significance values compared to isotropic control. Data
represents mean ± s.d, n=3-8, p<0.05(*), p<0.01(**)…………………………………………….49
Figure S4. (A) Fluorescence images of anti-GFP and anti-mCherry dualreporter fibroblasts in the
presence of no ligand (control) or plate-dried GFP, mCherry, or both GFP and mCherry. Images
taken 24 hours following seeding. Scale bars, 200μm. (B) Percent Reporter Activation of reporter
activation in dual-receiver fibroblasts measured via flow cytometry on no ligand, GFP only,
mCherry only, or both GFP and mCherry. Surfaces were uniformly adsorbed with the ligands.
Data represents mean ± s.d, n=4. (C) Schematic of dual-reporter L929 cell where anti-GFP
synNotch drives miRFP reporter gene and anti-mCherry synNotch orthogonally activates BFP (D)
Fluorescence images of platedried GFP and mCherry droplets and subsequent dual reporter
expression and brightfield taken 48 hours after uniform seeding of engineered fibroblasts. Scale
bars, 2mm. (E) Normalized plot profiles of miRFP and BFP intensity across the x-axis 48 hours
following seeding onto GFP and mCherry droplet pattern. Line profiles represent mean ± s.d, n=4.
(F) Fluorescence and brightfield images of dual reporter expression 48 hours following uniform
seeding onto perpendicular GFP and mCherry patterns. Scale bars, 500μm. (G) Dual-positive
miRFP and BFP masks, created using ImageJ image calculator AND function for BFP and miRFP
signal above a threshold. Mask was superimposed onto fluorescence images and highlighted with
a yellow border. Scale bars, 500μm.……………………………………………………………...50
Figure S5. (A) Fluorescence images of anti-GFP synNotch MyoD fibroblasts stained for α-actinin
7 days following seeding onto flat micromolded gelatin substrates in the presence of
transglutaminase-conjugated GFP (100μg/mL). Scale bars, 500μm. Myotube alignment of αactinin stained myotubes on GFP-conjugated isotropic (no topography) gelatin substrate,
quantified by image analysis. Line plot represents angles of orientation distribution of 5 individual
images from an individual sample. (B) Nuclei Alignment quantifications of anti-GFP synNotch
MyoD fibroblasts in the presence or absence of GFP on isotropic or micromolded gelatin. Data

xii
represents mean ± s.d, n=5, p<0.0001(****). (C) Binary Mask and fluorescence images of antiGFP synNotch MyoD fibroblasts seeded onto adhesion-restricted GFP patterned surfaces (500 and
200 μm rows) stained for α-actinin after 3 days. Scale bars, 500μm. Plot profile of normalized αactinin staining intensity and GFP signal on adhesion-restricted 500μm and 200μm GFP row
patterns. (D) Fluorescence images of nuclei staining for each pattern: 500μm curves, 200μm
curves, 500μm rows, and 200μm rows. Samples were stained three days following uniform seeding
onto GFP patterns. (E) Day 3 brightfield image of spatial myotube formation on 500μm width
rows. Zoomed in region, indicated with a dotted white line, showing phenotype difference onpattern (myotubes) vs off-pattern (fibroblasts). (F) Higher magnification fluorescence images of
α-actinin stained cells on isotropic or 500μm row GFP patterns. Scale bars, 200μm. (G) Plot
profiles of normalized α-actinin expression across 500 and 200μm rows on non-restricted GFP
patterns. Green lines indicate the region containing GFP. (H) Staining validation of mouse αactinin antibody on myotubes differentiated on GFP patterns, Day 3. (I) Volcano plot of gene
expression data showing differentially expressed genes of anti-GFP synNotch mCherry reporter
fibroblasts on GFP vs off GFP. n=2…..…………………………………………………………..52
Figure 6. mCherry activates ETV2 and reporter BFP to induce endothelial differentiation in
embryonic fibroblasts via synNotch. (A) Schematic of embryonic fibroblasts cell line (C3H)
expressing anti-mCherry/Gal4 synNotch activating ETV2 and BFP transgenes seeded onto
mCherry patterned substrate. (B) Fluorescence microscopy images of BFP reporter (top),
endothelial markers VEGFR2 (middle) and VE-Cadherin (bottom) stained 3 days following
seeding onto control wells (-mCherry) or plate-dried mCherry (+mCherry). Scale bars, 200μm.
(C) Percent of cells expressing BFP (left) and VEGFR2 (right) in the presence (+mCherry) or
absence (-mCherry) of mCherry, quantified with flow cytometry. Data represents mean ± s.d, BFP
n=12-13, VEGFR2 n=4-5, p<0.001(***), p<0.0001(****). (D) Heatmap of hierarchical clustering
of fibroblast parental cells without mCherry, fibroblasts engineered with anti-mCherry synNotch
activating ETV2 and BFP with and without mCherry, and BEnd.3 endothelial cells (positive
control) (E) Volcano plot of gene expression showing differentially expressed genes of antimCherry synNotch cells activating ETV2 and BFP on mCherry vs off mCherry, n=2. (F) Binary
mask used to generate stamps for microcontact printing of 500µm rows followed by fluorescence
microscopy images of day 3 nuclei staining, BFP expression, and VEGFR2 immunostaining. Scale

xiii
bars are 1mm. Plot profile of normalized BFP and VEGFR2 intensity on Day 3 following seeding
onto 500µm mCherry rows. Red bars indicate the regions containing mCherry. Line profiles
represent mean ± s.d, n=2. (G) Binary mask used to generate stamps of vasculature-like pattern
followed by fluorescence microscopy images of day 3 nuclei staining, BFP expression, and
VEGFR2 immunostaining……..…………………………………………………………………56
Figure S6. (A) Fluorescence images of anti-mCherry synNotch fibroblasts with inducible ETV2
– BFP expression in the presence of varying plate-dried mCherry concentrations (0-100μg/cm2).
BFP signal is shown in grayscale for visualization, images taken 1, 2, and 3 days following seeding.
Scale bar, 200μm. (B) Normalized BFP intensity of anti-mCherry synNotch fibroblasts seeded on
varying concentrations of plate-dried mCherry up to 3 days. Data represents mean ± s.d, n=3. (C)
Staining validation of anti-mCherry synNotch fibroblasts seeded on 15μg/cm2 mCherry with
endothelial marker VEGFR2. Panel shows sample stained with or without an anti-VEGFR2
primary antibody. Inducible BFP reporter and nuclei stained with HSC NuclearMask Deep Red.
Scale bars, 200μm. (D) Day 3 brightfield image of spatial endothelial-progenitor cell activation on
vascular-like pattern. Dotted white line indicates the region of interest enhanced in the following
panel. Scale bars, 1mm and 500μm, respectively. (E) Flow cytometry gating strategy to exclude
debris (Gate 1), isolate singlets (Gate 2), and determine BFP-expressing cells (top) or VEGFR2
expressing cells (bottom). (F) Measurement of CDH5 (VE-Cadherin) via immunostaining and
flow cytometry with and without presence of mCherry. Data represents mean ± s.d, n=3-4,
p<0.01(**). (G) Principal Component Analysis (PCA) comparing the transcriptome of unmodified
C3H fibroblasts, cell type-specific positive control cells (C2C12 and Bend.3), and receiver cells
that expressing mCherry, MyoD and mCherry, or ETV2 and BFP in the presence or absence of
their corresponding ligand. N = 2-4………………………………………………………………57
Figure 7. Spatially controlled co-transdifferentiation into myogenic and endothelial lineages of
dual-lineage synNotch-engineered cells culturing on micropatterned ligands. (A) Schematic
showing dual protein patterning technique using capillary-driven microfluidic patterning, based
on shallow and deep channels, to generate parallel lines of GFP and mCherry. Feature size, 500
µm. Schematic of dual-lineage mouse embryonic fibroblasts (C3H line) expressing anti-GFP/tTA
synNotch that activates MyoD and miRFP as well as an anti-mCherry/Gal4 synNotch that

xiv
orthogonally activates ETV2 and BFP transgenes, seeded onto GFP and mCherry patterned
substrate. (B) Fluorescence images of GFP and mCherry ligand (left), reporter genes expression
(center) and brightfield (right) of dual-lineage cells 3 days following uniform seeding onto
micropatterned ligands. Dotted white rectangles represent regions of interest for quantification.
Scale bars, 1mm. Plot profiles on the right show normalized fluorescence intensity of ligands and
reporters taken in the region of interest. Green and red bars indicate regions containing GFP or
mCherry, respectively. (C) Center: merged fluorescence image of α-actinin and VEGFR2
immunostaining on dual-lineage cells uniformly seeded onto GFP and mCherry pattern. Dotted
green lines represent region where the GFP signal has been subtracted to visualize VEGFR2
staining. Scale bar, 1mm. Around the central image, higher magnification fluorescence images
taken within distinct regions of the pattern are shown: on mCherry (red border), interface between
mCherry and GFP regions (yellow border), on GFP (green border), and off pattern (beige border).
Scale bars, 200µm. Higher magnification of interface (blue border), visualizing α-actinin and
VEGFR2 staining is shown on the far right with a scale bar of 50µm. D) Schematic showing the
different ligand patterns used in single-nuclei sequencing experiments. T-Distributed Stochastic
Neighbor Embedding plot results of dual-lineage fibroblasts cultured on the four different
patterning conditions. Fibroblast-like cluster contains seven individual clusters, shown here as one
beige color. n=2. E) Percent of Fibroblast-like, muscle-like, and endothelial-like cells across the
four patterning conditions (left). Percent ratio of which muscle-like clusters make up the total
muscle-like cells in each patterning condition (right). n=2. F) Plot showing average expression and
percent expression of selected fibroblast, muscle, and endothelial markers in all clusters across
different patterning conditions. n=2….……………………………….………………………….62
Figure S7. (from Tyler Hoffman) (A) Fluorescence images of anti-GFP and anti-mCherry duallineage synNotch fibroblasts with orthogonally inducible MyoD-miRFP and ETV2 – BFP,
respectively, expression in the presence of varying plate-dried mCherry concentrations (0-
100μg/cm2). BFP signal is shown in grayscale for visualization, images taken 1, 2, and 3 days
following seeding. Scale bars, 200μm. (B) Normalized BFP intensity of dual-lineage synNotch
fibroblasts seeded on varying concentrations of plate-dried mCherry up to 3 days. Data represents
mean ± s.d, n=3. (C) Fluorescence images of anti-GFP and anti-mCherry dual-lineage synNotch
fibroblasts with orthogonally inducible MyoD-miRFP and ETV2 – BFP expression in the presence

xv
of varying plate-dried GFP concentrations (0-100μg/ cm2). miRFP signal is shown in grayscale
for visualization, images taken 1, 2, and 3 days following seeding. Samples from each GFP
concentration were fixed and stained for α-Actinin following three days of culture. Scale bars,
200μm……..……………………………………………………………………………………..64
Figure S8. (A) Images of dual-lineage receiver cells cultured on GFP only, mCherry only, or both
GFP and mCherry. Surfaces were uniformly adsorbed with the ligands. Flow cytometry panel (XAxis: VEGFR2-APC, Y-Axis: BFP) of dual-lineage cells seeded on 15μg/cm2 plate-dried
mCherry unstained (left) or stained for VEGFR2 (center) and dual-lineage cells seeded on
15μg/cm2 plate-dried of both GFP and mCherry stained for VEGFR2 (right). (C) Fluorescence
images of GFP and mCherry droplet patterns (left), subsequent spatial reporter activation (center)
taken 48 hours after uniform seeding of engineered dual-fate fibroblasts. Scale bars, 2mm. Green
(within GFP pattern) and red (within mCherry pattern) borders indicate regions of interest with
corresponding higher magnification brightfield and fluorescence images (right). Scale bars,
200μm. (D) Brightfield image of dual-lineage cells seeded on droplet pattern taken 48 hours after
uniform seeding of engineered fibroblasts. Scale bars, 2mm. (E) Merged fluorescence images to
demonstrate the effects of subtracting the GFP pattern (high GFP intensity excluded) in visualizing
the α-actinin and VEGFR2 staining. Scale bars, 1mm. (E) Fluorescent microscopy images of duallineage cells on dual-ligand patterns showing α-actinin staining. VEGFR2 staining, and Nuclei
expression. Plot profiles of immunostaining of α-actinin and VEGFR2 showing fluorescence
intensity across distance………….…………………………………..…………………………..65
Figure S9. (A) T-Distributed Stochastic Neighbor Embedding (t-SNE) plot analysis showing
twelve cell clusters based on gene expression profiles from all four conditions. (B) Selected marker
gene analysis for Fibroblast-like, Endothelial-like, and the four Muscle-like clusters……………67
Figure 8. Lentiviral transduction and FACS gating strategy for MSC GFP senders and anti-GFP
mCherry reporters. (A) MSCs expression of GFP after 48 hours of viral transduction (right) vs
without virus (left). (B) Different concentrations of receptor and downstream transgenes used to
transduce MSCs to engineer anti-GFP mCherry reporters, resulting in varying levels of basal
mCherry leaky expression. (C) Histogram showing fluorescence in FITC channel of parental MSC

xvi
(left) and MSC transduced with membrane presenting GFP ligand plasmid (GFP sender high
virus). Cells were sorted in the FITC + side of the gate, collecting high GFP expressing MSCs. (D)
Texas Red channel and Forward Scatter Area plot of Parental MSC and anti-GFP mCherry
transduced MSC (high receptor, low transgene). Cells were sorted using low, medium, and high
mCherry fluorescence gates……………..……………………………………………………….99
Figure 9. MSCs engineered with anti-GFP mCherry reporter are activated by MSC GFP senders
and surface presented GFP. (A) Day 2 images of low and high gate sorted anti-GFP mCherry
reporters cultured with GFP senders. (B) Day 2 images of high gate sorted anti-GFP mCherry
reporters cultured on varying concentrations of 5 µL GFP droplets demonstrating spatial
expression of mCherry………..……………………………….………………………………..102

xvii
Abbreviations
SynNotch – Synthetic Notch
GFP – Green Fluorescent Protein
MyoD – Myoblast Determination Protein 1
Etv2 – Ets Variant Transcription Factor 2
NSF – National Science Foundation
EGF – Epidermal Growth Factor
IVF – In vitro Fertilization
iPSC – induced Pluripotent Stem Cells
hESC – Human Embryonic Stem Cells
PDMS – Polydimethylsiloxane
PEGDMA - Polyethylene Glycol and Dimethacrylate
CAR – Chimeric Antigen Receptor
DREADD – Designer Receptors Exclusively Activated by Designer Drugs
CNO – Clozapine N-oxide
AgRP – Agouti-Related Protein
GPCRs – G protein-coupled Receptors
MESA – Modular Extracellular Sensor Architecture
ECM – Extracellular Matrix
CRISPR – Clustered Regularly Interspaced Short Palindromic Repeats
FN – Fibronectin
UV-O – Ultraviolet-Ozone
DMEM – Dulbecco's Modified Eagle Medium
FBS – Fetal Bovine Serum
APTES – (3-Aminopropyl)triethoxysilane
DLP – Digital Light Processing
tTA – Tetracycline-controlled Transactivator
PCC – Pearson’s Correlation Coefficient
BFP – Blue Fluorescent Protein
VEGFR2 – Vascular Endothelial Growth Factor Receptor 2
miRFP – near-infrared Fluorescent Protein

xviii
DAPI – 4′,6-diamidino-2-phenylindole
MTG – Microbial Transglutaminase
DAVID – Database for Annotation, Visualization, and Integrated Discovery
PCA – Principal Component Analysis
OOP – Orientation Order Parameter
MSCs – Mesenchymal Stem Cells
Scx – Scleraxis
Runx2 – Runt-related Transcription Factor 2

xix
Abstract
The field of tissue engineering has been heavily focused on recapturing the structure and
function of native tissue in vitro with the goal of creating reproducible platforms for drug
screening, disease modeling, and regenerative medicine. However, a system that gives researchers
the precision and control to replicate the complex cellular architecture of native tissues does not
exist and is a major challenge in the field of tissue engineering. Current in vitro tissue engineering
methods, such as organoids or organ-on-a-chip systems, lack reproducibility and/or microscale
precision. To overcome these challenges, we propose a system that offers microscale control over
spatial gene induction and cell transdifferentiation. This system combines Synthetic Notch
(synNotch) engineered cells with microcontact printed synNotch activating ligands. SynNotch
receptors are exogenous cell membrane receptors that give researchers the ability to control
modular gene expression upon cell-contact with a ligand of choice. Microcontact printing offers
microscale precision over protein patterning on a material surface. Combining the two, we
developed a tissue engineering system where spatially controlled proteins via microcontact
printing activate transdifferentiation pathways in cells via synNotch, resulting in spatially
controlled gene expression and cell transdifferentiation. We created cells that co-transdifferentiate
fibroblasts into either muscle or endothelial precursor cells, depending on the ligand that they are
in contact with. Our project aims to build a co-transdifferentiated tissue consisting of myotubes,
the functional unit of skeletal muscle, and endothelial cells, the main cell type found in vasculature.
Doing so expands the synthetic biology toolkit by offering an approach to transdifferentiate multicellular tissues with microscale precision. This technology advances the field closer to replicating
the complex cellular architecture found in vivo and benefits the areas of drug screening, disease
modeling, and regenerative medicine.

1
Chapter 1 - Introduction
Tissue Engineering Background:
The field of tissue engineering has made incredible advancements to date and has come a
long way since the term was first coined by Dr. Yuan-Cheng Fung in 1987 during a National
Science Foundation (NSF) panel meeting in Washington, DC1
. Tissue engineering is a very broad
field, so to hone in on an area that is more applicable to my work, this section will focus on the
history and advancements of in vitro tissue models.
In vitro tissue models are an indispensable tool in biological research. Modern day models
aim to create more human relevant tissue platforms and provide a customizable alternative to basic
cell culture, animal models, and human testing with the goal of improving our approach to drug
screening, disease modeling, and regenerative medicine. The problem in the current day is that
scientific research, predominantly in areas such as drug development and disease modeling, still
heavily rely on animal models. This is because the current in vitro alternatives are still too
expensive to manufacture and too immature to rely on for conclusive results especially for
translational studies. This being said, the field has advanced remarkably over the last century and
I will begin by highlighting some of those advancements.
The first ever recorded instance of in vitro cell growth was performed by Ross Harrison in
1907, where he extracted frog neuroblasts and cultured them in lymph medium with a technique
called a hanging drop culture 2
. He placed the cells on a glass slide and inverted it over a dish,
creating a sealed and moist environment for tissue growth. Using a microscope, he recorded the
growth of the nerve fibers within the droplet, indicating the ability of cells to develop and
differentiate outside of the body, mimicking their behavior in vivo. The ability to culture cells
outside an animal was revolutionary and is the cornerstone that all other tissue engineering

2
advancements were, and are currently, built on. Although Dr. Harrison did not continue in this
direction, others such as Dr. Alexis Carrel and Dr. Montrose Burrows refined his cell culture
technique to maintain tissue cultures for longer periods and applied this technique to many other
cell types such as cartilage, skin, and kidneys from varying species3
. They also were the first to
work with cancerous cells to showcase subculturing, which is the start of new cultures from old
ones, demonstrating that primary tissues are not continuously needed to expand cell cultures. In
1912, using cells from a chicken embryo heart, Dr. Carrel showed the first example of a “cell-line”
which possessed indefinite growth as long as the cells were kept in a healthy state 4
. Fast forward
to 1951, Dr. George Otto Gey from Johns Hopkins University established the first immortalized
human cell line by extracting cervical cancer cells from Henrietta Lacks to create HeLa cells.
Today, this is viewed as an unethical event in scientific history as her consent was never given.
The ethical issues surrounding their origin have led to improvements in the way that patient
samples are collected and used in research, with more attention to informed consent and privacy
for the patient. This being said, it cannot be underestimated how impactful HeLa cells have been
over the years such as their role in the development of common lab techniques, early cancer
research, and gene mapping 5
.
In 1977, continuing the advancement of in vitro human tissues, Dr. James G. Rheinwald
and Dr. Howard Green published a paper that highlighted the use of epidermal growth factor (EGF)
to increase the lifetime of epidermal keratinocytes in culture. These were the first human noncancerous cells that were successfully grown in vitro for 50-150 passages and were the building
blocks for the development of bioengineered skin 6
.
Following this success, another milestone in the field was the advancement from 2D to 3D
cultures. The basic 2D cultures failed to capture the complex interactions of human tissues and are

3
limited in the transferable information they provide for drug screening and disease modeling. A
notable milestone in the development of 3D cell cultures was spearheaded by Joseph Vacanti and
Robert Langer in 1988 7
. Up until this point, cell transplantation involved injecting cell suspensions
into tissues. Their publication showcased the use of an implantable cell scaffold created with
biodegradable polymers. They seeded the scaffold with hepatocytes, pancreatic islet cells, or small
intestinal cells from rats or mice. The implantation of this scaffold proved that the cells were still
viable inside the host animal with the scaffold, demonstrating the first successful implantation of
pancreatic islets. This was the first case of using materials in conjunction with cells to enhance the
survivability of implanted cells, marking another step of complexity beyond the basic cell culture
methods. Another area of the field which was growing at this time was the evolution of 3D
spheroids. This particular technique mimicked several aspects of in vivo tissues such as cell-ECM
interactions and nutrient/oxygen gradients that were not observable in 2D cell cultures8
.
A natural advancement of the spheroid technique became possible with the introduction of
human embryonic stem cells (hESC) cells by James Thompson in 1998 9
. These pluripotent stem
cells were capable of differentiating into virtually any cell type. Researchers had their hands full
with the tasks of finding which cocktail of growth factors and culturing conditions resulted in
specific cell types. As these protocols were being established, a new 3D model for study began to
emerge called organoids 10,11. Unlike spheroids which were cultured with pre-differentiated cells,
organoids differed by consisting of stem cells and differentiated with specific protocols to create
organoids of a specific lineage. These allowed scientists to grow their target tissue in a 3D structure
for study. However, hESC did not come without some hurdles. First, there were ethical concerns
surrounding their use because they have to be extracted from human embryos, sacrificing a human
embryo in the process. Second, because of this, they were still not very accessible as they would

4
need donors or in vitro fertilization (IVF)12. Third, their potential for regenerative medicine is
limited as hosts will react with an allogenic immune response and reject the cells. If hESC are used
for transplantation, the patient may need to take immunosuppressive drugs with harmful side
effects13.
Alleviating these drawbacks was the development of induced pluripotent stem cells
(iPSCs) by Shinya Yamanaka and Kazutoshi Takahashi in 2007 14. This was a groundbreaking
discovery. They discovered a cocktail of four factors, Oct3/4, Sox2, Klf4, and c-Myc, called the
Yamanaka factors, which reprogrammed adult dermal fibroblasts back into a stem-like call state
that possessed pluripotency. Firstly, this solved most ethical concerns revolving around hESC as
iPSC could be taken from adult human tissue usually by drawing blood or extracting fibroblasts
from a minimally invasive skin biopsy 15. Second, this method also ensures easier accessibility as
obtaining blood or skin cells is much easier than destroying a human embryo for hESC. Third,
iPSC, being able to be reprogrammed from the same patient that may need a treatment, are much
less likely to be rejected by the patient’s immune system, making them more attractive for
regenerative medicine 16. Shinya Yamanaka alongside John Gurdon, who discovered that cells
were not tied to their singular differentiation fate, received a joint Nobel prize for their
achievements in 2012. This revolutionary work was a major milestone that greatly improved the
viability and potential of in vitro tissue models, among other fields of biology, as scientists now
had more accessible stem cell lines for creating their chosen model system.

5
Animal Models:
The most used technique for the end stages of drug development and disease modeling are
animal models, which have been integral to scientific research for centuries. As early as the 2nd
century, Galen, an ancient Greek physician, performed anatomical studies on pigs and primates as
an alternative to studying human anatomy 17. Frederick Banting and Charles Best used dogs to
discover insulin in 1921 18. To test whether insulin would treat diabetes, they induced diabetes in
a set of dogs by removing their pancreas. They then saved and treated these dogs by extracting the
insulin from a healthy dog and injecting it into the dog with diabetes. Later the following year,
they treated a boy that was suffering from diabetes with insulin extract from a cow’s pancreas.
This discovery has been incredibly revolutionary for the treatment of diabetes and has saved
countless lives.
In the 1950’s, mice and monkeys were used as the animal models to develop the polio
vaccine19. Albert Sabin and Jonas Salk developed oral and injectable vaccines, respectively,
against the polio virus using mice and monkeys for safety and efficacy testing20,21. Starting from
the 19th century, animal models saw an expansion and rodents were the most common model for
study. Although far different than humans, the rodents still shared enough biological and
behavioral similarities to provide valuable insight into physiological and pathological processes.
For example, the simple animal model has the complexity of interacting tissues and organs, which
for the most part, are more similar than they are different to the human’s with the basic functions
being the same. The genetic manipulability, that came much later, was also an advantage of animal
models as many of the same genes, and their effects, are found in the human system. This allows
researchers the ability to investigate the role of certain genes in a disease phenotype or
development.

6
Of course, there are many disadvantages to using animal models. Although they are more
similar than not to the human system, this dissimilarity is actually very crucial as there are still
significant genetic, anatomical, and physiological differences between animals and humans. They
simply do not reflect the human response 100%. Further, maintaining animal colonies is expensive
and very time-consuming, especially compared to the basic cell culture approach. The larger the
animal and closer to the human system, the more costly it is both in terms of economic and time
investment. A primate is much more similar to a human in terms of anatomical and genetic
makeup, but is exponentially more expensive than a rodent model for study. Also, animal models
are not easily modifiable genetically and usually require multiple lineages of breeding in some
cases to create the genetic modification needed for study. Finally, there are a great deal of ethical
concerns revolving around animal research. There have been many advances in this area to
improve the lives and health of the animal models leading up to their experimental demise.
While animal models have played and continue to play a crucial role in advancing
biomedical knowledge, the ethical and scientific limitations necessitate alternative models.
Innovations like organ-on-a-chip systems and organoids present promising alternatives, paving the
way for more human-relevant, ethical, and efficient research methodologies. This being said, many
of these systems still provide very niche insights and are too immature to completely alleviate the
need of animal models. As these technologies mature, a holistic integration of both traditional and
novel methods will be pivotal in driving scientific discoveries.

7
Organoids
Organoids are 3D tissue cultures derived from stem cells that aim to mimic some structural
and/or functional characteristics of their in vivo counterparts. One of the first examples of
organoids came in 2009 where researchers used stem cells expressing Lgr5 undergo single crypt
to multi-crypt fission events and generate villus-like structures22. As a breakthrough development,
this methodology was then used by multiple other groups to target a variety of tissues such as the
stomach, liver, and brain 23–25.
In 2013, human pluripotent stem cells were used to generate cerebral organoids26. The
study shows that the cerebral organoids exhibit characteristics of human cortical development such
as the progenitor zone organization with outer radial glial cells. The group also demonstrated the
use of iPS cells with RNAi technology to model microcephaly, which causes a newborn's head to
grow much smaller and has been difficult to study in mice. This showcased a disease-modeling
use for organoids, highlighting the application and use of the tissue-engineering approach.
By this time, organoids were already a very popular area of research and the advancement
of them was growing quickly. Highlighting personalized medicine, organoids derived from cystic
fibrosis patients were used to predict patient-specific responses to drugs targeting the disease27.
Following this, the natural next step for organoids was to combine multiple different cell types in
them with the goal of studying their function and response to drugs in a more holistic manner. In
2018, Calvin Kuo’s lab aimed to co-culture the primary tumor epithelia with tumor-infiltrating
lymphocytes to reconstitute a more complete tumor microenvironment28. A limitation of organoids
is that the size is restricted due to limited perfusion of nutrients to the core of the organoid, resulting
in cell death in the core. An approach to solve this problem is to mimic what nature has done;

8
create blood vessels. Following this idea, the first vascularized organoids were demonstrated in
2019 by Wimmer et al.29. hPSCs were used to generate self-organizing blood vessel organoids that
recapitulate exhibited morphological and functional features of human microvasculature. This
breakthrough opens the door for creating larger organoids that more closely resemble their in vivo
counterparts.
These key developments in organoid technology have unlocked new possibilities in the
field of tissue engineering, enabling scientists to study human organ development and disease in
unprecedented detail. The many advancements in organoid technology over the years have led to
this system offering a multitude of advantages over basic tissue culture and animal models.
Organoids provide a platform with a significantly increased physiological relevance compared to
basic 2D tissue culture. The 3D structure of organoids intrinsically increases the physiological
relevance due to the fact that in-vivo tissues are grown in 3D, not 2D. The 3D setting better
replicates the cell-cell and cell-matrix interactions. Alongside this, the fact that organoids are
versatile and can be made to mimic specific organ systems, such as the heart or kidneys, allows
researchers to investigate drug responses or other biological phenomena in the isolated target
tissue. Additionally, they are a higher throughput option compared to animal models, which are
more expensive and slower to work with. They are also able to be developed from patient derived
iPSC, creating a more personalized model for study vs animal models.
This being said, organoids as a tissue engineering approach have some notable limitations
as well. Most organoids are limited in size due to a lack of vascularization. This inherently creates
a functional and architectural difference between organoids and their real-life counterparts. The
complexity and maturation, as a result, are also limited. The fact that organoids mainly mimic one

9
specific tissue at a time also has a downside as well in that the effect of surrounding organs and
tissues is inherently lost, decreasing the predictive confidence of drug screening results and other
observations in translation to the human body. Organoids also suffer from batch-to-batch
reproducibility issues. Organoid differentiation is also limited in user-control, as the user mainly
supplies growth factors and the tissue forms in response to those factors without precise patterning
cues from the researcher.
Altogether, organoids will continue to have a great, positive impact on biological research,
but the downsides necessitate more advancements or alternative approaches in the tissue
engineering toolkit.
Organ-On-A-Chip Systems:
Around this time, there was a novel tissue engineering approach that was gaining traction
called organ-on-chip systems. This approach began to take shape in the late 2000’s, but its roots
can be traced back to the advances in microfabrication techniques for the microelectronics industry
in the second half of 1900’s 30. These same microfabrication techniques were modified to create
microfluidic chips that aimed to provide a customizable growth environment for cells that better
mirrored their in vivo environment. Before microfluidics, most cultures were grown in a static
environment which is vastly different from what is felt by the cells in vivo. In the late 1990’s and
early 2000’s these microfluidics chips became a popular technique for culturing cells in vitro.
Some advantages they provided were the ability to control fluid flow rate and the ability to generate
concentration gradients of signaling molecules, allowing researchers to study phenomena that
wasn’t previously possible 31. They also provided a level of spatial control over the structure of
the tissue.

10
In a 2010 landmark study led by Donald Ingber at the Wyss Institute at Harvard, a lungon-a-chip microfluidic device was created that mimicked the microarchitecture of the lunch. The
device separated epithelial cells and endothelial cells with a contractible layer, mimicking the
alveolar capillary interface of a lung 32. One side of the device was exposed to the air (inner lung
layer) and one side was exposed to media flow (blood facing layer). Applying cyclic suction to the
side channels, the team simulated the mechanical signals during breathing. From this study, the
team of researchers demonstrated an inflammatory response to nanoparticles, drug testing efficacy,
and even an example of a pulmonary infection. This milestone served as an example of just how
complex organ-on-a-chip systems can get to model specific organ microenvironments. Going a
step further, "Human Body-on-a-Chip" or "Multi-Organ-on-a-Chip" platforms are among the most
ambitious directions in the field of organ-on-a-chip technology. These systems aim to link together
multiple individual organ-on-a-chip models to create an integrated, system-level model of human
physiology. The hope is that by modeling the interconnectedness of the human body, these systems
can provide more accurate predictions of the effects of drugs, toxins, or diseases than could be
achieved with single organ-on-a-chip models33.
Organ-on-a-chip systems have rapidly developed over the years. They offer more user
control over the structure of the environment that the cells are cultured in compared to organoids.
This allows researchers to develop very specific chips that target niches within a target tissue. For
example, users can control the stiffness of the substrate, the shear stress of the media flow, and
incorporate biochemical cues and gradients. Alongside this, these systems offer less variability
between samples, leading to more reproducible results. Just like organoids, an advantage these
chips offer over animal models is the possibility of using patient derived cells, which allows
researchers to model patient-specific outcomes. Additionally, in contrast to organoids, organ-on-

11
a-chip systems are often fabricated using Polydimethylsiloxane (PDMS), a transparent
biocompatible polymer that allows imaging using common fluorescent microscopes during an
experiment. With all of these advantages, organ-on-a-chip systems are a powerful tool for tissue
engineering in areas such as disease modeling and drug screening.
This being said, microfluidic chips do have some drawbacks such as limited accessibility
as not every lab has the specialized equipment or expertise to manufacture the complex chips.
Simple things such as an air bubble could clog the fluidic channels and render an experiment
useless, which means that a level of troubleshooting and practice is needed to successfully
incorporate microfluidic chips into a lab. Creating more complex systems such as the lung-on-achip mentioned above requires even more experience in the field. Given the time needed and the
complexity of fabrication steps, these systems are relatively costly, especially with the initial setup
and training, creating a barrier for labs or institutions with less funding.
Another limitation that most organ-on-a-chip systems have is the restriction on the spatial
control of cell seeding. These systems offer layer-by-layer or chamber-by-chamber spatial control.
However, spatial control in the X and Y directions within these layers is not easy to achieve,
especially in the micrometer scale that is seen in-vivo. While this is a step above spatial control
that is offered in 2D cultures, organoids, and animal models, there is still room for improvement.
This results in, yet again, a limited tissue maturity and complexity when compared to in vivo
tissues. In addition to this, the cells seeded in these systems are usually pre-differentiated, which
is different from how tissues form in the body and in development, creating functional differences
to the in vivo target tissue and lowered reliability with clinical translation of results 34.

12
Next-Generation Tissue Patterning Systems:
Of course, these drawbacks have been a recent target to solve for the tissue engineering
field. One approach that aimed to improve the spatial resolution of cell-types came from the field
of optogenetics. In a study led by Dr. Bursac and Dr. Gersbach, light inducible gene cassettes were
engineered into cells and activated via patterned light to induce patterned cell differentiation35.
The study used light inducible Cre-recombinase system, that when under activation by blue light,
leads to downstream expression of the user-controlled gene cassette. In the paper, they showcased
millimeter scale precision of myoD expression and myotube formation. This was a great first proof
of concept and a step in the right direction to create a novel tissue engineering approach with higher
spatial control than existing methods. This being said, the scale at which they activated their lightinducible cells was still not sufficient enough for replicating the architecture of most tissues. In
addition to this, the myotube differentiation lacked robustness due to the low percentage of total
cells that formed myotubes in the patterned regions.
Another example of a different approach to solving this problem was the use of a bioprinter
with 3 different extrusion nozzles. Cells with different dox-inducible gene cassettes were fed
through the nozzles onto a 2-D culture surface, spatially controlling their placement. Once the cells
are patterned, doxycycline-containing media is used to activate the gene modules in the cells, one
cell type remains as pluripotent hiPSC while the other two differentiate toward neuronal lineage
or endothelial lineage36. This approach increased the spatial resolution of the differentiation down
to around 200-300 µm. This being said, the nature of the seeding technique is intrinsically limiting
to the possible patterns of differentiation, as the cells can only be seeded parallel to each other and
in rows.

13
Another recent advancement in controlling cell adhesion has been demonstrated by the use
of DNA “velcro” technology for selective patterning of ssDNA-labeled cells with long-term
culture on mechanically defined polyacrylamide hydrogels37. This novel technique allows the user
to spatially control cell adhesion of two different cell types. The paper demonstrated this approach
by patterning the interior mesenchyme and the interior epithelium by patterning fibroblasts that
are surrounded by epithelial cells or by patterning epithelial cells that are surrounded by
fibroblasts, respectively. They found notable differences in the protein expression at the boundary
between these two cell types that depended on which cell type was patterned in the exterior and
which was patterned in the interior, highlighting the importance of spatial organization in in vitro
tissue formation.
As shown, there are many creative approaches to building in vitro tissues in the field of
tissue engineering, each with their pros and cons. Still, there is room for improvement. The
following sections will highlight technologies and areas of research that we will utilize for our
approach in creating a novel tissue engineering approach that alleviates some of the cons with
previous advancements.

14
Micropatterning
Cell and material micropatterning consist of many techniques that allow researchers to
control the microenvironment of cells, influencing their shape, size, and interaction with other
cells. There are several micropatterning techniques, each with their own specific use-cases and
limitations.
Photolithography is a technique involving the use of light to transfer a geometric pattern
from a photomask to a light-sensitive chemical (photoresist) on a substrate. It's commonly used in
microfabrication to create micro- and nano-scale devices. It has been adapted for biological
purposes, using soft lithography to create patterns of adhesive proteins on a substrate to guide cell
adhesion (more on this later).
Electropatterning is another micropatterning technique and aims to guide cell adhesion
and alignment. Cells respond to electric fields by aligning along the field lines, allowing
researchers to influence their alignment direction and pattern. For example, in a paper from 2006,
dielectrophoretic forces were used to position cells within a prepolymer solution prior to
crosslinking the polymer 38. The researchers were able to form cell patterns with micron scale
resolution. This same paper also combined this method with photopatterning, another
micropatterning technique, which was used to selectively crosslink the hydrogel with 100µm
resolution. These two methods together were used to gain hierarchical control over cell positioning
over a vast range of lengths (microns to centimeters).
Another relatively recent advancement, Inkjet and 3D bioprinting are techniques that allow
for precise spatial control of cells and extracellular matrix materials. Cells and materials are
deposited layer by layer, creating 2D or 3D structures that more closely mimic in vivo tissue. A
study from 2012 showcased the use of bioprinting where the authors used inkjet technology to

15
create 3D cartilage from polyethylene glycol and dimethacrylate (PEGDMA)39. The researchers
seeded the 3D constructs with human chondrocytes and implanted them into osteochondral plugs
to successfully repair defects.
One specific micropatterning technique, called microcontact printing, was especially
interesting for us. The following section will highlight the advantages of the technology.
Microcontact Printing:
Microcontact printing is a technique used to pattern proteins or other molecules with
micrometer-scale precision on a variety of substrates. The technique was first introduced by
George Whitesides and Amit Kumar at Harvard University in 1993 40. This technique was
introduced to pattern self-assembled monolayers of gold. This opened a new approach for
controlling tissue structure as researchers used this method to pattern biomolecules, such as
extracellular matrix proteins, with high precision to guide cell adhesion and migration41,42.
Another example of this is a study from 2011 that showcased the ability to introduce and control
asymmetry in patterned cells. Asymmetry of tissues in vivo is a well-conserved biological property
that can also be a disease marker in some cases43. Using microcontact printing to study different
cell types in disease and non-diseased phenotypes is just one significant impact of the technology.
However, despite its undeniable contributions to the field, microcontact printing is not
without its limitations. Primarily, the use of traditional photolithography methods has put a softcap on the aspect ratios that are achievable with microcontact printing. Creating a stamp with thin
features, such as 10µm, that are separated by larger gaps, for example 500µm, is difficult because
of “roof collapse” due to the gap being much larger than the stamp region and too close to the

16
substrate because of the difficulty in creating tall and thin features using the photosensitive
polymer 44,45. This creates a constraint on the aspect ratios achievable for stamps. The technique is
also limited by its inherent planar nature, restricting its applicability to the production of threedimensional structures, which are crucial for replicating the complexity of native tissues.
Moreover, the reliance on manual procedures in traditional µCP can lead to variability and
inconsistency in the resulting patterns. This has been partially addressed by the advent of more
advanced, automated microcontact systems, but these come with their own set of challenges,
including the need for sophisticated equipment and higher operating costs46.
This being said, microcontact printing has been mainly used to restrict cell adhesion onto
the pattern, which is a different approach compared to our goals of using microcontact printing to
pattern protein for cell signaling, not adhesion.

17
Synthetic Biology:
Synthetic biology, an interdisciplinary branch of biology and engineering, involves the design and
construction of new biological parts, devices, and systems, and the redesign of existing natural
biological systems for useful purposes. The field has its roots in the earlier disciplines of genetic
engineering and molecular biology, but it has expanded far beyond these initial areas. The
beginnings of synthetic biology can be traced back to advancements in recombinant DNA
technology in the 1970s. This period saw the advent of techniques to manipulate DNA sequences
in vitro, giving birth to genetic engineering. The most noteworthy milestone from this era was the
development of the first genetically modified organism by Herbert Boyer and Stanley Cohen in
1973, where they successfully introduced a recombinant DNA molecule into Escherichia coli,
marking the dawn of modern biotechnology47.
Michael Elowitz and Stanislas Leibler created the first synthetic biological oscillator,
termed the 'repressilator', in E. coli. The repressilator is a synthetic gene network that periodically
induces the expression of a transgene, in this case the production of green fluorescent protein,
causing the bacteria to flash on and off. This was the first example of an oscillating and artificial
biological clock. This study demonstrated the feasibility of engineering complex behaviors into
biological systems48.
In the same year, Timothy Gardner, Charles Cantor and James Collins engineered a genetic
'toggle switch' in E. coli49. The toggle switch is a synthetic gene network that can exist in one of
two stable states and can be flipped from one state to the other by external signals. It consisted of
two repressible promoters arranged in a mutually inhibitory network, which resulted in a synthetic,
bistable gene-regulatory network. This study, along with the repressilator, marked the advent of
synthetic biology as we know it today. These early studies were crucial in demonstrating that

18
biological systems could be engineered at the genetic level to perform new functions. These initial
experiments in bacteria paved the way for more complex synthetic gene networks in mammalian
cells and other organisms. The principles and techniques developed in these early studies continue
to underpin much of the work in synthetic biology today.
This being said, the beginnings of synthetic biology were exclusively practiced and studied
in bacteria. It was not until a few years later that the transition of these synthetic gene networks
transitioned from prokaryotic to eukaryotic systems. The transition from bacterial to mammalian
systems in synthetic biology was not a sudden one, but a gradual process that continues to this day.
While bacteria are still the most common and convenient system for many types of synthetic
biology research, the use of mammalian cells, yeast, and other eukaryotic organisms has been
steadily increasing, allowing synthetic biology to tackle more complex biological questions and
applications. A prominent example of this was demonstrated by Martin Fussenegger which
described a synthetic epigenetic system able to switch between two stable transgene expression
states that respond to alternate drug administrations50. The researchers implemented streptogramin
and macrolide inducible promoters on two separate transgenes. The two transgenes express
repressors for each other. The addition of either erythromycin or pristinamycin inhibits the
repression of one of the two transgenes and consequently induces expression of the previously
repressed transgene.
The advent of precise gene-editing tools like CRISPR-Cas9 has also greatly facilitated the
engineering of mammalian cells and brought synthetic biology into a new era. Marking the birth
for the CRISPR-Cas9 technology was the discovery of Jennifer Doudna and Emmanuelle
Charpentier when they published their groundbreaking genome editing study in 2012 51. This
technology was showcased in bacteria, and then shortly later, demonstrated to work in eukaryotic

19
cells as well52. This technology was groundbreaking because of the precision it provided for
targeted genome editing to add, delete, or modify a gene. The efficiency and ease of use was
unprecedented compared to existing technologies such as Zinc Finger Nucleases53. On top of this,
CRISPR can be applied to virtually any organism, making it a very versatile technology. Another
exciting potential from CRISPR is the potential application of it for correcting gene defects at their
source. That being said, CRISPR is not without its dangers and ethical concerns. A big concern is
the use of CRISPR to modify the human germline with heritable changes, creating “designer
babies”, or reviving extinct species. Not all of these are inherently bad, but the irresponsible
application of CRISPR can result in unintended consequences. The scientific community holds
this responsibility very highly and most of us hope security measures are enforced.
On the flip side, it has been a great benefit to synthetic biologists. It has become a
foundational technology that allows synthetic biologists more precision and control over the
construction of synthetic gene circuits. For example, CRISPRi and CRISPRa systems were
developed to either repress or activate endogenous genes via endonuclease-deficient Cas954. This
advancement allows scientists to control endogenous genes without modifying or introducing
exogenous genes. Another advancement in the synthetic biology field was the use of dCas9 to
build logic gates into E. coli55. This approach allowed them to target endogenous pathway switches
by building a CRISPR/Cas genetic circuit that interfaces with the endogenous regulatory networks.
This was done by targeting the dCas9 protein to E. coli transcription factor (malT). CRISPR
continues to advance and is an indispensable tool in the field of synthetic biology, among others.

20
Synthetic Receptors:
Natural cells possess many types of endogenous receptors that allow the cells to sense,
process, and respond to external stimuli in their environment. This is crucial for living organisms
to function as it enables all the cell-cell and ECM-cell communications that ensures that different
parts of an organism are in sync. A relatively recent development in the field of synthetic biology
is the engineering of various synthetic receptors. Synthetic receptors are engineered to respond to
additional/replacement exogenous signals or to respond differently to endogenous signals. This
provides a method for researchers to control cellular behavior and function in new and precise
ways.
The predecessor to the modern synthetic receptors, the Chimeric Antigen Receptor (CAR)
was introduced in 1989 by Dr. Zelig Eshhar and his colleagues56 . In this breakthrough paper, they
demonstrated the fusing of an antibody-derived recognition domain with a T-cell derived signaling
domain to create a receptor that could redirect T cells to target and kill cells that are expressing
that specific, user-controlled antigen. Although they were not called synthetic receptors, they do
technically fit the description of them today. Since then, the CAR system has been heavily updated
and improved with various methods to combat the many hurdles it had to achieve its FDA-approval
in 2017.
Before then, the second-generation CARs incorporated co-stimulatory signaling domains
in addition to the primary activation domain to improve T cell killing precision. This significantly
improved the efficacy of CAR T cells in preclinical models. A study by Brentjens in 2007 added
CD28 as a costimulatory domain alongside CD19 antigen that is found on B cells. It was used to
enhance antitumor efficacy of T cells in mouse models 57. The third generation of CARs combines

21
multiple costimulatory domains, such as CD28-41BB or CD28-OX40. This being said, it was a
2nd generation CAR, called Kymriah, which was given FDA approval in 2017 58
Advancing the toolkit came the first optogenetic receptor, channelrhodopsin-2, that was
developed first in mammalian cells in 2005 59. A landmark study by Dr. Nagel investigated the use
of ChR2 in mammalian neurons to demonstrate millisecond-timescale control of neuronal spiking.
They also showed that the system offered control of excitatory and inhibitory synaptic
transmission. Their system allowed them to study the dynamics of exciting the neurons at different
frequencies. For example, they found that higher frequencies of excitation (30 Hz+), lead to
decreased successful spikes. This study established optogenetics as a powerful tool for controlling
neuronal activity and studying the function of neural circuits, paving the way for a revolution in
neuroscience.
The first, modern generation of synthetic receptors were introduced in the early 2000’s and
were typically based on existing cellular receptors that were modified to respond to drugs or other
exogenous ligands. The first generation was based on G-protein coupled receptors and was referred
to as DREADDs (Designer Receptors Exclusively Activated by Designer Drugs). A landmark
study based on this technology was conducted using the DREADD technology to chemogenetically control both neuronal and non-neuronal signaling in mice60. Clozapine N-oxide (CNO)
was used as the chemical signaling ligand because of its ability to penetrate the central nervous
system. Cre dependent hM3Dq receptor was used for expression and activation of the agoutirelated protein (AgRP) neurons which was thought to promote feeding and weight gain. The acute
activation of agRP neurons rapidly induced feeding, reduced energy expenditure, and increased
fat stores in mice. This solidified the knowledge on how AgRP neurons affect feeding habits in
mice and demonstrated an elucidating use-case for DREADD technology.

22
The toolkit of synthetic receptors has been rapidly expanding ever since DREADD was
showcased. Deisseroth's group took optogenetics one step further in 2011 by creating OptoXRs -
light-activated G protein-coupled receptors (GPCRs) 61. They used rhodopsin, a light-sensitive
protein from the retina, to engineer GPCRs that can be activated by light. OptoXRs expanded the
scope of optogenetics beyond neurons, allowing researchers to control a wide range of cellular
processes with light.
Another recent advancement has been Modular Extracellular Sensor Architecture (MESA)
receptors in 2014 62. Prior to MESA, no existing technology enabled cells to sense exclusively
extracellular ligands without relying on native receptors or signal pathways that may crosstalk with
native cellular components. MESA solved this problem by being orthogonal to native cellular
processes through responding to exogenous, orthogonal ligands and activating orthogonal
transgene activation with non-native transcription factors and promoters. Aside from being
maximally orthogonal to native cellular processes, another advantage and innovation of MESA is
that it is very modular in nature. This enables the user to control what ligand activates the receptor
(the input) and what transgene is expressed upon activation (the output). Also important to note is
that the activating ligand can be soluble and does not need to be tethered onto a substrate.
One year later, a photoactivatable CRISPR-Cas9 system was built for optogenetic genome
editing63. The photoactivatable Cas9 system consists of split Cas9 fragments and photo-inducible
dimerization domains called Magnets that dimerize in response to blue light. Although this was
not the first optogenetic switch developed, it was the first time the CRISPR system was modified
to respond to blue light activation, further increasing the synthetic biology toolkit.
It is important to note that multiple receptors and systems from the synthetic biology toolkit
can be used synergistically within one engineered cell line.

23
SynNotch Background:
SynNotch, short for Synthetic Notch, is a synthetic receptor that is derived from the
endogenous Notch receptor, developed in the Wendell Lim Lab in 2016 64 . Endogenous Notch
binds to its cognate ligand, Delta, and is involved in various crucial cell activities such as
development, differentiation, apoptosis. The SynNotch receptor replaces both the extracellular
binding domain of endogenous Notch as well as the intracellular signaling domain. This gives
SynNotch possessing cells the ability to recognize and respond to a specific and modular external
ligand, usually a protein. The extracellular domain can be switched out to bind to GFP or mCherry,
for example. The intracellular domain can be switched to a synthetic transcription factor that
allows the user to control transgene activation orthogonally with endogenous Notch, which
remains active and functional in these cells as well.
The activation mechanism of SynNotch is fairly unique among other synthetic receptors65.
Once the SynNotch receptor makes contact and binds to the presented ligand from a neighboring
cell, there is a pulling force that occurs which causes the receptor to stretch, exposing the cleavage
site for enzymes to cleave the intracellular domain. Only after this cleavage can the intracellular
domain, or transcription factor, be able to travel to the nucleus of the cell to initiate transcription
of the modular transgene. One important aspect of SynNotch is that the presented ligand needs to
be tethered onto a surface in order to generate pulling force on the SynNotch receptor so that the
intracellular domain is cleaved, activating the pathway. In short, soluble ligands cannot activate
any SynNotch pathway because the free floating protein does not provide any resistance to the
receptor, thus the cleavage site of the receptor is not accessible to the enzymes for cleavage. To

24
summarize, these cells can be customized to bind to different proteins and respond by activating
user prescribed transgene(s).
An example of how SynNotch has been used to signal cell differentiation is from a paper
published in 2016 by Dr. Leonardo Morsut and Dr. Kole Roybal from the Wendell Lim lab 66.
They used a variety of SynNotch pathways with different sender ligands to receptor and transgene
combinations to control SynNotch activation. One way they did this was to first pattern a small
droplet of cells that presented CD19 as their membrane ligand followed by seeding receiver cells
with anti-CD19 SynNotch activating myoD, the master myogenic regulator. Within two days, they
were able to verify the transdifferentiation of C3H embryonic fibroblasts into multi-nucleated
myoblast-like cells.
In the context of tissue engineering, using cells to signal other cells to differentiate is not
easy to spatially control. We had the idea of patterning our ligand, for example GFP, onto a
material surface instead of having it be presented by a neighboring cell. As long as the ligand is
presented in the correct orientation and provides enough pulling force for the receptor, it should
activate.
Some of the advantages of patterning our protein onto a material surface over presenting it
from sender cells are:
1) Greater spatial control over the pattern by utilizing precise protein patterning techniques
such as microcontact printing.
2) No need to allocate space in the tissue to sender cells and can have a continuous 2D layer
of receiver cells that we are interested in.
The following chapter will expand on this.

25
Chapter 2 - Engineering Programmable Material-To-Cell Pathways Via Synthetic Notch
Receptors To Spatially Control Cellular Phenotypes In Multi-Cellular Constructs
Abstract
Synthetic Notch (synNotch) receptors are modular synthetic components that are
genetically engineered into mammalian cells to detect signals presented by neighboring cells and
respond by activating prescribed transcriptional programs. synNotch has been used to program
therapeutic cells and pattern morphogenesis in multicellular systems. More recently, evidence that
it can be activated from material-presented ligands has started to emerge. However, whether this
can provide a flexible platform across different biomaterials, and with a focus on spatial patterning
of the ligands at different resolution for applications that require spatial precision and extracellular
matrix (ECM) manipulations, such as tissue engineering, has not been demonstrated so far. To
address this and achieve microscale control over synNotch activation in cell monolayers, we
microcontact-print synNotch ligands onto a surface, and demonstrate orthogonal control of
intracellular gene activations with two types of synNotch ligands generating with up to four
distinct reporter gene profiles. Additionally, we showcase this technology by cotransdifferentiating fibroblasts into skeletal muscle or endothelial cell precursors in user-defined
spatial patterns towards the engineering of muscle tissue with prescribed vascular networks. This
approach extends the synNotch toolkit and provides novel avenues for spatially controlling cellular
phenotypes in mammalian multicellular systems, with many broad applications in developmental
biology, synthetic morphogenesis, human tissue modeling, and regenerative medicine.

26
Results and Figures
A fundamental goal for the emerging area of synthetic morphogenesis and tissue
engineering is the ability to design and spatially control gene expression patterns within a
multicellular construct. Intricate patterns of gene expression control the proper organization and
physiology of cells, tissues, and organs and are a hallmark of complex multicellular systems across
the tree of life. Individual cells express genetic networks that drive or support cell fate commitment
and functional behaviors, like motility and proliferation. During embryonic development, initially
uniform cell ensembles activate genetic networks in designated spatial regions to generate tissues
with distinct geometrical patterns. The spatial organization of cells within a tissue endows them to
coordinate and accomplish complex functions, such as absorption or contractility. In vivo, spatial
domains of gene expression are driven by genetically-encoded communication networks involving
intra-cellular67,68, inter-cellular69–73, and cell-to-extracellular matrix (ECM) and ECM-to-cell74–76
components. Several of these networks are active in organoids in vitro, which self-organize and
replicate select microscale architectural features similar to native tissues. However, genetic
networks in organoids are spatially activated in an autonomous way and some genetic networks
fail to activate at all, leading to heterogeneity and stunted tissue structures. Because selforganization is convoluted with differentiation and other complex cell behaviors, in vitro methods
for arbitrarily engineering and interrogating spatial gene expression patterns and their impact
would augment our understanding of biological systems11,77–79. Advanced technologies for
spatially controlling gene expression would also enable tissues to be engineered with user-defined
cellular compositions and geometries, which would be impactful for the fields of regenerative
medicine, Organs on Chips, and lab-grown protein-rich food sources80–83.

27
Classically, tissue engineers have focused on influencing cell differentiation and behavior
by engaging endogenous cell surface receptors. For example, natural ligands, such as extracellular
matrix proteins, can be presented to cells in user-defined spatial arrangements via microfabricated
biomaterials to control adhesion, alignment, or differentiation84–86. Because these approaches rely
on engagement of endogenous receptors, such as integrins, stereotyped and often complex
behaviors are activated in responding cells. However, with these approaches, users are confined to
the limited library of endogenous ligands and receptors and their pre-existing downstream
pathways, many of which are multifaceted with ambiguous outcomes. Recently, synthetic
receptors have been developed that endow cells with orthogonal, customizable signaling
capabilities22. Thus, we reasoned that these receptors could be leveraged to spatially control gene
expression patterns in engineered tissues with more precision than endogenous receptors.
Specifically, we turned to a class of synthetic receptors based on native Notch signaling, named
synthetic Notch or synNotch66. SynNotch are a class of synthetic receptors composed of chimeric
protein domains: an antibody-based binding extracellular domain (e.g. anti-GFP nanobody), the
Notch juxtamembrane and transmembrane domains, and orthogonal transcription factors (e.g.
Gal4) as the intracellular domain. SynNotch receptors have many desirable features that could be
exploited to spatially control gene expression: (i) the receptor is not activated by soluble factors;
(ii) the ligand is customizable and can be an orthogonal inert molecule, such as a fluorescent
protein (e.g. GFP); (iii) receptor activation can drive customizable cellular responses, such as
differentiation, when combined with complementary genetically engineered cassettes.

28
SynNotch has previously been used to generate spatial patterns of gene expression in 2D
(concentric rings66 and in 3D (polarized and layered spheroids87) by using neighboring cells (i.e.,
sender cells) to present synthetic ligands to cells expressing synNotch (i.e., receiver cells). Cellular
ligand presentation, however, has the disadvantage that controlling the geometry of synthetic
ligands necessitates controlling the location of sender cells, making the problem circular. Evidence
suggests that a pulling force between sender and receiver cells is necessary to initiate signal
transduction in the receiving cell, similar to endogenous Notch receptors. Due to this feature,
synNotch has also been activated by synthetic ligands passively adsorbed onto cell culture
surfaces66, tethered by DNA linkers to microbeads88, and attached to atomic force microscopy
probes26. More recently, an approach to specifically activate synNotch from culture surfaces was
developed under the acronym MATRIX89. In this approach, surfaces are functionalized with
antibodies (e.g. GFP-TRAP) that capture soluble synNotch ligands (e.g. GFP), which can then
activate synNotch receptors (e.g. anti-GFP synNotch) in receiver cells to regulate CRISPR-based
transcriptome modifiers, modulate inflammatory niches, and mediate stem cell differentiation.
Wedge-shaped culture inserts were also used to functionalize surfaces with coarse spatial control.
However, whether synNotch ligands can be directly conjugated to a wider range of natural or
synthetic biomaterials to activate synNotch, and whether this approach could be extended to
pattern gene expression and/or differentiation and co-differentiation of multiple cell fates within
the same culture with micron scale precision, has not yet been shown.
Here, our objective was to develop generalizable, user-defined, material-to-cell pathways
for spatially controlling genetic networks and differentiation in multicellular constructs via
synNotch and microcontact printed culture surfaces. We show that this approach is generalizable
to multiple synNotch receptors and can activate distinct synthetic pathways in cells with two

29
synNotch receptors (i.e., dual-receiver cells). We then show that this approach can be extended to
spatially control patterns of gene expression and cell fate by transdifferentiating embryonic
fibroblasts into either skeletal muscle precursors or endothelial cell precursors in tissue-relevant
geometries. Finally, we demonstrate a method for spatially controlling the co-transdifferentiation
of fibroblasts to one of two cell fates (endothelial cell precursors or skeletal muscle precursors) in
a continuous tissue construct. This was achieved by generating dual-lineage fibroblasts expressing
two independent synNotch receptors (one for endothelial transdifferentiation, and one for muscle
transdifferentiation) and culturing these cells on a surface with the two synthetic cognate ligands
patterned via a microfluidic device. These novel methods for generating spatial patterns of gene
expression and cell fate add a powerful and flexible functionality to the synthetic biology toolbox
for controlling and investigating multicellular organization.
Developing a microcontact printed, non-restrictive adhesion substrate:
To spatially control the patterning of the synNotch cognate ligands, we will be relying on
microcontact printing. To use microcontact printing for spatially controlling synNotch activation,
we first needed to get over a hurdle. To date, microcontact printing was primarily used to pattern
cell adhesion. Cells would adhere to the patterned ECM proteins and did not adhere anywhere else
on the surface. For our case, we need cells to adhere everywhere on the cultured surface while only
being activated on the pattern of proteins. It’s also important to note that we microcontact print
our ligand (GFP) onto a PDMS coated glass coverslip. This is because PDMS has a changeable
elastic modulus which allows the user to fine-tune its stiffness to match the tissue of interest. The
downside of using PDMS is that it is inherently non-cell adhesive and needs to be chemically
altered to ensure cell-adhesion. To solve these issues, multiple approaches were tested.

30
1) Patterning GFP first, then incubating fibronectin on top. The problem with this method is
that by coating fibronectin on top of the GFP, we automatically cover up the GFP signal.
To allow the GFP signal to still reach the cells, we have to use 10x less fibronectin than is
needed for optimal cell adhesion which creates a surface that is not cell-adhesive enough
and results in our tissues delaminating shortly after seeding. Also it’s important to note that
to get protein to transfer over to the PDMS, we UV-O treat our PDMS coverslips for 8
minutes to create a temporary hydrophilic surface for efficient protein transfer from stamp
to substrate.
2) Coating with fibronectin first, then patterning GFP on top. This method allowed us to coat
with fibronectin up to the 50µg/mL concentration that is needed for optimal cell adhesion,
but we still observed partial or whole tissue delamination. The UV-O treatment of our
PDMS does not result in a very stable covalent bond of the fibronectin to the PDMS. When
the entire surface is seeded with cells, as opposed to small regions with restrictive adhesion,
it causes more stress on the substrate and has a larger area of failure chance, increasing the
chances of tissue delamination. The protein signaling did work however, so we were
mainly focused on getting a stronger FN to PDMS bond after this point.
3) Dopamine coating before/after GFP patterning. A previous paper 90 showcased the use of
dopamine to enhance the adhesion properties of PDMS. However, the paper did not
demonstrate the adhesion properties of their method over time. Trying this approach
enhanced our adhesion on day 1, but this quickly diminished as our tissues were also
delaminating before our final time-point of days 2-3. Myotubes were delaminating and
clumping up, rendering this approach useless for our target goal.

31
Finally, the approach that did work was a chemical treatment of our PDMS that resulted in
a covalent bond between the fibronectin and our PDMS surface. To do this, we took inspiration
from another paper that reported increased adhesion of FN to PDMS and increased downstream
cell-adhesion by employing a treatment of APTES followed by glutaraldehyde to create aldehyde
groups on the PDMS surface that covalently bind to FN after overnight incubation with the protein
91. This approach worked extremely well for cell-adhesion, however, it created a new issue with
protein smearing, or ligand mobility92 Ligand mobility happens when a patterned protein is
smeared across a surface when liquid is pipetted across the patterned surface. We solved this issue
by including a wash step with media (DMEM + 10% FBS) to overcome ligand mobility. Although
we don’t know exactly how this works, we believe it is similar to a blocking step where the FBS
in the media binds to the free GFP before it can smear or bind to the FN layer. Once these issues
were resolved, then we started exploring the dynamics of our system and working towards
differentiating in vitro tissues with spatial control.

32
Figure 1. Treating PDMS coverslips with APTES and glutaraldehyde before FN coating
enhances cell adhesion. Brightfield images on Day 3 of UVO + FN treated PDMS coverslips vs
APTES/Glut + FN treated coverslips. Both conditions were coated with FN at 50 µg/mL.

33
Figure 2. Wash step with media (DMEM + 10%FBS) before cell seeding reduces ligand
mobility and increases ligand to cell-activation colocalization. Image showing GFP channels
prior to cell seeding and downstream day 1 mCherry signal.

34
Spatial activation of synNotch via microcontact printing
Our goal here is to dictate synNotch activation patterns within multicellular tissue
constructs at a spatial resolution similar to the cellular length scale. To achieve this, we adapted
microcontact printing techniques designed to transfer microscale patterns of proteins (classically
ECM proteins) onto culture surfaces93,94. Our goal was to microcontact print GFP onto uniformly
cell-adhesive surfaces (Fig. 3A). To achieve this, we treated PDMS-coated coverslips with APTES
and glutaraldehyde to induce covalent bonding of proteins91 and then coated the surface with
fibronectin for uniform cell adhesion. To optimize the transfer of GFP onto the fibronectin layer,
we created simple, featureless PDMS stamps by cutting cylinders from PDMS using a biopsy
punch. We coated and incubated these stamps with 0-200 μg/mL GFP solutions and then inverted
them onto fibronectin-coated coverslips. Finally, we seeded coverslips with receiver cells
expressing anti-GFP/tTA synNotch receptors that activate an mCherry reporter. These cells
formed a confluent monolayer and demonstrated a GFP dose-dependent increase in mCherry
fluorescence that saturated at roughly 100 μg/mL GFP (Fig. S3A-B), indicating that surfaces dualfunctionalized with fibronectin and GFP maintained cell adhesion and activated synNotch.
To induce activation of synNotch in small groups of cells within a multicellular tissue, we
next developed an approach to microcontact print arrays of GFP squares with features ranging
from 100 µm to 1 mm. PDMS stamps for microcontact printing are classically cast on silicon wafer
templates fabricated by cleanroom-based photolithography95. However, this approach is not
suitable for our feature sizes because they are large (100 μm to 1 mm) relative to the height of
photoresist conventionally used for photolithography (1-10 μm). PDMS stamps with high feature
to height ratios are susceptible to buckling and transfer of GFP outside the intended regions95. To

35
overcome this, we used a digital light processing (DLP) 3-D printer to rapidly print templates with
taller features in a photocrosslinkable resin. We first 3-D printed a template comprising an array
of 100 μm sided-squares with 100 μm interspaces, which is roughly the resolution limit of the 3-
D printer. The height of the features was set as 100 μm to minimize buckling. As shown in Figure
3B, PDMS stamps fabricated in this way could successfully transfer GFP onto covalently coated
FN coverslips in the intended 100 μm x 100 μm pattern, demonstrating successful microcontact
printing using PDMS stamps cast on 3D printed templates.

36
Figure 3. Microcontact printed GFP patterns spatially activate mCherry reporter via
synNotch pathway activation. (A) Visual schematic showing the process of stamp preparation
and concurrent coverslip preparation for the microcontact printing of GFP patterns. (B) Binary
mask of the stamp features that contains 100µm square features with 100µm gaps and resulting
GFP fluorescence image following microcontact printing (C) Schematic of anti-GFP/tTA
synNotch receiver fibroblasts seeded onto GFP-patterned substrate to demonstrate local activation
based on the presence of GFP. Portions of the cells were made transparent to visualize the
underlying pattern of GFP. (D) Binary masks and fluorescence microscopy images of 500µm and
250µm side square GFP patterns, separated by 250, 350, 500, and 1000µm gaps, followed by

37
mCherry fluorescence images of anti-GFP synNotch receiver cells two days following seeding.
Dotted white lines separate regions with different interspaces and the solid white lines present the
location of GFP patterns enlarged in the following panel. (E) Plot profile of normalized mCherry
intensity on 2, 5, and 10 days following seeding onto 500x1000µm (top), 250x250µm (bottom) or
GFP pattern (square side x interspace). Green indicates the regions containing GFP. (F)
Quantifications of day 2 Pearson’s Correlation Coefficient (PCC) comparing Binary Mask with
mCherry channel across all patterns. Scrambled binary mask images were compared to day 2
mCherry channels as a negative control. Data represents mean ± s.d, n=7-8, not significant p>0.05
(ns), p<0.01(**), p<0.0001(****). (G) Binary masks, brightfield microscopy images, and GFP
fluorescence images of concentric circles and letter patterns of GFP, followed by mCherry
fluorescence microscopy images taken two days following seeding onto the GFP pattern. Dotted
white rectangles in the brightfield indicate the region of interest shown in higher magnification on
the bottom of the same image. Scale bars, 1mm.

38
We next used these techniques to fabricate stamps and microcontact print arrays of GFP
squares with sides ranging from 250 μm to 1000 μm and interspaces of 250 μm or 500 μm onto
PDMS-coated coverslips pre-coated with fibronectin. The feature height for these stamps ranged
from 100 μm to 500 μm, depending on square sizes and interspaces. Microcontact printed surfaces
were then seeded with receiver cells with anti-GFP/tTA synNotch receptors that activate an
mCherry reporter (Fig. 3C). After two days, mCherry expression was detected within the
multicellular tissue in patterns that overlapped with the original design to different extents,
depending on the pattern (Fig. 3D-E). To quantify the spatial fidelity of synNotch activation, we
calculated the Pearson’s correlation coefficient between the binary pattern design and the mCherry
images (Fig. 3F and S3E). As expected, the correlation coefficient was highest for tissues with the
largest squares (500 μm sides) and largest interspaces (1000 μm). The correlation coefficient
decreased as features and/or gaps decreased. However, for all tissues with square sizes and
interspaces greater than 100 μm (Fig. S3C), the correlation coefficient between the mCherry image
and the binary pattern was significantly higher compared to the correlation coefficient between the
mCherry image and a scrambled binary pattern with the same number of white pixels. The
correlation also decreased with time due to weakening of reporter activation (Fig. 3E, S3D).
Together, these data indicate that the minimum feature size for this approach is approximately 250
μm. Based on this conclusion, we designed other arbitrary patterns with minimal feature sizes of
250 μm, including concentric circles and letters. Qualitatively, we observed similar agreement
between the binary pattern, GFP fluorescence, and mCherry fluorescence (Fig. 3G), demonstrating
versatility of pattern designs.

39

40
Figure S3. (A) Day 2 fluorescence images showing mCherry activation in anti-GFP reporter
fibroblasts on no GFP, 10μg/mL microcontact printed GFP, and 100μg/mL microcontact printed
GFP. Scale bars, 200μm. (B) Violin plot of mCherry intensity, quantified with flow cytometry,
dose response to 0, 10, 50, 100, and 200μg/mL concentrations of microcontact printed GFP after
48 hours. Dotted line indicates the threshold value to designate mCherry-positive cell. Percent of
mCherry expressing cells quantified by flow cytometry 48 hours after seeding onto GFP
microcontact printed substrate with varying GFP concentrations. Data represents mean ± s.d, n=6-
8, p<0.05(*) (C) Microcontact printed 100μm GFP squares with 100μm interspace length and
resulting mCherry activation signals taken two days after seeding. Scale bars, 1mm. (D)
Fluorescence images of mCherry activation from days 1-12 on 250 and 500μm width GFP squares
with varying interspace lengths. (E) Heatmap demonstrating Pearson correlation coefficient
between mCherry and GFP signals of each square width and interspace length at Days 2, 5, and
10. Color map represents the mean coefficient, n=7-8. (F) Brightfield images of anti-GFP reporter
fibroblasts 2 days following uniform seeding onto 500 and 250μm GFP squares. Scale bars,
500μm.

41
Our next goal was to scale-up this approach to spatially activate multiple distinct genetic
programs in the same multicellular tissue. Previous studies have demonstrated that two synNotch
receptors can be integrated into a single dual-receiver cell23. Thus, we asked if culturing dualreceiver cells on a surface patterned with two synthetic ligands in distinct arrangements would
generate a tissue with corresponding patterns of distinct genetic programs (Fig. 4A). We first
generated a dual-receiver fibroblast cell line (L929) that harbors an anti-GFP/tTA synNotch
receptor that activates an miRFP reporter and an anti-mCherry/Gal4 synNotch that activates a
BFP reporter (Fig. S4C). To validate the responses to synthetic ligands of these cells, we seeded
them on a culture surface microcontact printed with a uniform layer of GFP, mCherry, or both.
As shown in Fig. S4A-B, miRFP was expressed only on GFP surfaces and BFP was expressed
only on mCherry surfaces, demonstrating orthogonal activation of the two pathways. On surfaces
with both GFP and mCherry, both miRFP and BFP were expressed, indicating activation of both
pathways. Next, to prototype the generation of spatial patterns of gene expression starting from a
uniform population of dual-receiver cells, we adsorbed GFP from a droplet in one corner of a
culture surface and a droplet of mCherry in the opposing corner. Dual-receiver cells cultured
uniformly on the surface activated miRFP and BFP in a spatial pattern corresponding to the GFP
and mCherry droplets, respectively, demonstrating macroscale spatial control over the activation
of two synNotch pathways in one cell population (Fig. S4D-E). Finally, to provide more precise
spatial control over the patterns, we microcontact printed an array of 500 µm-wide rows of GFP
with 500 μm interspacing. We then stamped perpendicular mCherry rows by manually
positioning the orientation of the stamp (Fig. 4B). When seeded with dual-receiver cells, we
observed rows of miRFP-expressing cells perpendicular to rows of BFP-expressing cells (Fig.

42
4B and Fig. S4F), as expected. At the GFP and mCherry intersections, cells expressed both
miRFP and BFP (Fig. 4C and Fig. S4G), indicating activation of both synNotch pathways,
generating 4 reporter “states” for the initially uniform population of engineered cells (BFP-
/miRFP-, BFP-/miRFP+, BFP+/miRFP-, BFP+/miRFP+) within the 1.5 mm2 tissue.
Additionally, we quantified the percent of BFP and miRFP expression in cells on different
regions of the pattern with image analysis (Fig. 4D). Roughly 60-70% of dual-receiver cells on a
region with a single ligand (GFP or mCherry) expressed the matching reporter (miRFP or BFP,
respectively). On the GFP and mCherry intersections, roughly 50% of dual-receiver cells
expressed both BFP and miRFP. These values were similar to the percent reporter activation
measured by flow cytometry in dual-receiver fibroblasts cultured on surfaces uniformly adsorbed
with one or both ligands (Fig. S4B). Thus, two independent synNotch genetic programs can be
spatially controlled by culturing dual-receiver cells on user-defined patterns of the two synthetic
ligands, to generate a multicellular tissue with up to four spatially controlled reporter gene
expression states.

43
Figure 4. Patterned GFP and mCherry spatially activate respective reporter genes via
synNotch activation in dual receiver fibroblasts. (A) Schematic showcasing: (left) dual-ligand
microcontact printing, and (right) seeding of dual-receiver L929 cell where anti-GFP synNotch
drives miRFP reporter gene and anti-mCherry synNotch orthogonally activates BFP. (B)
Fluorescence microscopy images of microcontact printed GFP and mCherry perpendicular rows
of 500µm width and subsequent dual reporter expression taken 24 hours after uniform seeding.
Scale bars, 500μm. (Right) Normalized plot profiles of miRFP intensity across each row axis of
engineered dual-receivers 24 hours following seeding onto perpendicular GFP and mCherry
patterns. Green bars indicate regions containing GFP. Line profiles represent mean ± s.d, n=7.
(Below) Normalized plot profiles of BFP intensity across each row axis of engineered dualreceivers 24 hours following seeding onto perpendicular GFP and mCherry patterns. Red bars
indicate regions containing mCherry. Line profiles represent mean ± s.d, n=7. Regions of interest
on mCherry (red border), on GFP (green border), or on intersecting patterns (yellow border)
enlarged in the following panels. Higher magnification fluorescence microscopy images of GFP
and mCherry perpendicular rows and subsequent dual reporter expression/brightfield image.
Regions of interest on GFP and mCherry intersection (blue border) or non-patterned region (black

44
border) enlarged in the following panels. Scale bars, 500μm. (C) Higher magnification
fluorescence microscopy images to demonstrate BFP and miRFP expression by dual-receiver
fibroblasts on mCherry, GFP, intersection, and between patterns. Scale bars, 100μm. (D) Percent
reporter activation quantification on the regions containing no ligand, GFP only, mCherry only,
and GFP and mCherry. Results show the percent of cells that are not expressing BFP and miRFP,
expressing both BFP and miRFP, expressing BFP only, or expressing miRFP only, or. n = 4.

45
Spatial control of concurrent differentiation to skeletal muscle and endothelial cell precursors
Beyond expression of fluorescent reporter proteins, synNotch receptors have also been used to
activate transgenes that control cell phenotypes or behaviors via overexpression of transcription
factors, such as Snail for epithelial to mesenchymal transitions or myoD for transdifferentiation of
fibroblasts to skeletal muscle precursors66. Thus, we next tested if synthetic ligands presented by
materials could drive overexpression of functional transcription factors that induce
transdifferentiation. We first generated a receiver fibroblast cell line (C3H) expressing an antiGFP/tTA synNotch receptor that activates myoD (Fig. 5A). When these receiver cells were
cultured on surfaces uniformly printed with GFP, they transdifferentiated to multinucleated, αactinin positive myotubes (Fig. 5B). To further characterize changes in phenotype, we performed
bulk RNA sequencing on unmodified C3H fibroblasts, receiver cells cultured on surfaces with or
without GFP, and C2C12 myotubes. We observed that culturing receiver cells on GFP surfaces
led to 3064 differentially expressed genes. According to hierarchical clustering, receiver cells on
GFP were most similar to C2C12 myotubes (Fig. 5C). Receiver cells on GFP also over-expressed
several muscle-specific genes, such as Myh2, Myh4, and Ttn, and down-regulated expression of
fibroblast genes, such as Col1a1 and Pdgfrb (Fig. 5D). GO-term analysis indicated that several
pathways related to muscle development and differentiation were enriched in receiver cells on
surfaces with GFP compared to without GFP (Fig. 5E). In contrast, receiver cells expressing an
anti-GFP/tTA synNotch receptor that activates mCherry did not over-express muscle-specific
genes or pathways, and only led to 33 differentially expressed genes, when cultured on surfaces
with or without GFP (Fig. S5I). Together, these data indicate that surfaces with GFP specifically
induced the transdifferentiation of receiver cells expressing an anti-GFP synNotch receptor that
activates MyoD to myogenic precursors.

46
Our next goal was to combine the synNotch receptor technology with surface micropatterning to
engineer aligned muscle tissue. Previous studies have shown that micromolded gelatin hydrogels
are favorable for myotube adhesion and alignment96,97. Thus, we asked if this type of surface could
be used to both transdifferentiate and align synNotch-induced myotubes. We constructed gelatin
hydrogels that are either isotropic or micromolded with 10 μm ridges separated by 10 μm spacing
and then enzymatically conjugated GFP to the surface. Receiver cells cultured on GFP
transdifferentiated to α-actinin positive myotubes, independent of surface topography, and receiver
cells consistently aligned to micromolded ridges (Fig. 5A-B, Fig. S5A-B), independent of
activation state. However, only receiver cells cultured on micromolded GFP hydrogels fused into
aligned myotubes (Fig. 5A-B and S5A-B), demonstrating that transdifferentiation and cell
alignment were controlled independently. We did observe a slight but non-significant increase in
nuclei alignment for cells cultured on micromolded gelatin hydrogels with GFP compared to
without GFP (Fig. S5B), possibly because cell fusion induced by MyoD caused a modest
improvement in cell alignment.
Another approach for engineering aligned muscle tissues is to culture muscle cells on
microcontact printed lines of matrix proteins45. We tested if this approach was compatible with
synNotch by microcontact printing lines of a mixture of fibronectin and GFP. When the same
receiver cells were cultured on these surfaces, they transdifferentiated into aligned myotubes (Fig.
S5B), indicating that microcontact printing matrix proteins and synthetic ligands can also be used
to both control tissue architecture and transdifferentiation.
In the approaches described above, a population of fibroblasts was uniformly
transdifferentiated to myoblasts. Our next goal was to selectively transdifferentiate fibroblasts to
myoblasts in a spatially controlled manner as a first step towards generating tissues with multiple

47
distinct cell types arranged in prescribed patterns. To achieve this, we used the approach described
above (Fig. 3) to microcontact print rows of GFP on fibronectin-coated surfaces. To test if we
could achieve spatially controlled differentiation, we printed thin or thick, curved or straight, rows
and then seeded the printed surfaces with fibroblasts harboring an anti-GFP synNotch receptor that
activates myoD (Fig. 5E). After three days, we fixed and stained tissues for α-actinin and
quantified the myogenic index on and off the pattern by using the binary pattern as a mask (Fig.
5F). Myogenic index was significantly higher on-pattern compared to off-pattern for all
geometries, demonstrating local geometric control of transdifferentiation. We also quantified the
coherency of the tissues as a proxy for alignment and observed higher coherency for tissues on the
straight rows compared to the curved rows, where 200 μm rows improved myotube orientation
(Fig. 5G). Thus, we can selectively transdifferentiate fibroblasts to myoblasts in a geometrically
prescribed way while also controlling the global alignment of the tissue, demonstrating that we
can separately and concurrently control local differentiation and tissue architecture.

48

49
Figure 5. Microcontact printed GFP patterns spatially activate myoD and initiate myotube
differentiation in embryonic fibroblasts via synNotch. (A) Schematic of embryonic fibroblasts
expressing anti-GFP synNotch activating myoD and mCherry transgenes seeded onto GFP
patterned PDMS substrate. (B) Sarcomeric α-actinin staining on isotropic GFP patterns on a PDMS
substrate, stained three days following seeding on GFP patterns. (-)GFP indicates image taken on
GFP-negative regions, (+)GFP indicates images taken within GFP-positive regions. Scale bars,
1mm. (C) Heatmap of hierarchical clustering of fibroblast parental cells without GFP, fibroblasts
engineered with anti-GFP synNotch activating myoD and mCherry with and without GFP, and
C2C12 myoblasts (skeletal muscle positive control). Z-Score is calculated by (Gene expression
value in sample of interest) - (Mean expression across all samples) / Standard Deviation. n=2-4.
(D) Volcano plot of gene expression data showing differentially expressed genes of anti-GFP
synNotch activating myoD and mCherry (anti-GFP/myoD) on GFP vs off GFP. n=2-4. (E) GO
term analysis showcasing enriched muscle pathways in myoD expressing synNotch receiver cells
on GFP vs off GFP. n= 2-4. (F) Fluorescence microscopy images of anti-GFP/myoD synNotch
receiver fibroblasts stained for α-actinin 4 days following seeding onto micromolded gelatin
substrates in the presence or absence of GFP. Scale bars, 500μm. (G) Myogenic index, quantified
with image analysis, in the presence or absence of GFP on isotropic or micromolded gelatin. Data
represents mean ± s.d, n=5, p<0.0001(****). Myotube alignment of α-actinin stained myotubes on
GFP-conjugated micromolded gelatin compared to GFP-conjugated isotropic gelatin substrate,
quantified by image analysis. Line plot represents average angles of orientation distribution of 5
individual images from an individual sample. Data represents mean ± s.d. (H) Binary mask used
to generate stamps for microcontact printing followed by fluorescence microscopy images of GFP
ligand and subsequent α-actinin staining for each pattern: 500µm curves, 200µm curves, 500µm
rows, and 200µm rows. Samples were stained three days following uniform seeding onto GFP
patterns. (I) Day 3 Myogenic Index, quantified with image analysis, on and off GFP for each
pattern (isotropic, 200, 500 curves, 200, 500 straight rows). Data represents mean ± s.d, n=3-6,
p<0.01(**), p<0.001(***), p<0.0001(****). (J) Orientation Order Parameter measured across all
patterns quantified with image analysis. Significance values compared to isotropic control. Data
represents mean ± s.d, n=3-8, p<0.05(*), p<0.01(**).

50

51
Figure S4. (A) Fluorescence images of anti-GFP and anti-mCherry dualreporter fibroblasts in the
presence of no ligand (control) or plate-dried GFP, mCherry, or both GFP and mCherry. Images
taken 24 hours following seeding. Scale bars, 200μm. (B) Percent Reporter Activation of reporter
activation in dual-receiver fibroblasts measured via flow cytometry on no ligand, GFP only,
mCherry only, or both GFP and mCherry. Surfaces were uniformly adsorbed with the ligands.
Data represents mean ± s.d, n=4. (C) Schematic of dual-reporter L929 cell where anti-GFP
synNotch drives miRFP reporter gene and anti-mCherry synNotch orthogonally activates BFP (D)
Fluorescence images of platedried GFP and mCherry droplets and subsequent dual reporter
expression and brightfield taken 48 hours after uniform seeding of engineered fibroblasts. Scale
bars, 2mm. (E) Normalized plot profiles of miRFP and BFP intensity across the x-axis 48 hours
following seeding onto GFP and mCherry droplet pattern. Line profiles represent mean ± s.d, n=4.
(F) Fluorescence and brightfield images of dual reporter expression 48 hours following uniform
seeding onto perpendicular GFP and mCherry patterns. Scale bars, 500μm. (G) Dual-positive
miRFP and BFP masks, created using ImageJ image calculator AND function for BFP and miRFP
signal above a threshold. Mask was superimposed onto fluorescence images and highlighted with
a yellow border. Scale bars, 500μm.

52

53
Figure S5. (A) Fluorescence images of anti-GFP synNotch MyoD fibroblasts stained for α-actinin
7 days following seeding onto flat micromolded gelatin substrates in the presence of
transglutaminase-conjugated GFP (100μg/mL). Scale bars, 500μm. Myotube alignment of αactinin stained myotubes on GFP-conjugated isotropic (no topography) gelatin substrate,
quantified by image analysis. Line plot represents angles of orientation distribution of 5 individual
images from an individual sample. (B) Nuclei Alignment quantifications of anti-GFP synNotch
MyoD fibroblasts in the presence or absence of GFP on isotropic or micromolded gelatin. Data
represents mean ± s.d, n=5, p<0.0001(****). (C) Binary Mask and fluorescence images of antiGFP synNotch MyoD fibroblasts seeded onto adhesion-restricted GFP patterned surfaces (500 and
200 μm rows) stained for α-actinin after 3 days. Scale bars, 500μm. Plot profile of normalized αactinin staining intensity and GFP signal on adhesion-restricted 500μm and 200μm GFP row
patterns. (D) Fluorescence images of nuclei staining for each pattern: 500μm curves, 200μm
curves, 500μm rows, and 200μm rows. Samples were stained three days following uniform seeding
onto GFP patterns. (E) Day 3 brightfield image of spatial myotube formation on 500μm width
rows. Zoomed in region, indicated with a dotted white line, showing phenotype difference onpattern (myotubes) vs off-pattern (fibroblasts). (F) Higher magnification fluorescence images of
α-actinin stained cells on isotropic or 500μm row GFP patterns. Scale bars, 200μm. (G) Plot
profiles of normalized α-actinin expression across 500 and 200μm rows on non-restricted GFP
patterns. Green lines indicate the region containing GFP. (H) Staining validation of mouse αactinin antibody on myotubes differentiated on GFP patterns, Day 3. (I) Volcano plot of gene
expression data showing differentially expressed genes of anti-GFP synNotch mCherry reporter
fibroblasts on GFP vs off GFP. n=2.

54
To exploit the modularity of this technology, we next tested if transdifferentiation to
another cell fate could be activated by a similar approach. Due to the universal need for
vascularization in engineered tissue constructs, including muscle, we focused on
transdifferentiating fibroblasts into endothelial cells precursors, which was previously shown via
doxycycline-inducible overexpression of the master transcription factors ETV298,99. Thus, we
generated fibroblast receiver cells engineered with an anti-mCherry/Gal4 synNotch receptor that
activates an ETV2-BFP cassette (Fig. 6A). We then passively adsorbed mCherry onto culture
surfaces, cultured receiver cells on them for three days, and fixed and stained the cells for
endothelial cell precursor markers. As shown in Fig. 6B-C and Fig. S6A-E, the fibroblasts
transdifferentiated to VEGFR2-positive endothelial precursors that also expressed VE-cadherin on
their membrane. We also evaluated the differentiation trajectory by performing bulk RNA
sequencing of receiver cells cultured on surfaces with or without mCherry, unmodified C3H
fibroblasts, and Bend.3 endothelial cells as a positive control. We detected that culturing receiver
cells on mCherry surfaces led to 3022 differentially expressed genes. Receiver cells cultured on
mCherry preferentially clustered with Bend.3 cells (Fig. 6D) and overexpressed endothelialrelated genes, such as KDR and CDH5, compared to cells cultured on surfaces without mCherry
(Fig. 6E). CDH5 (VE-Cadherin), a later-stage endothelial marker, was also detected at the protein
level with flow cytometry (Fig. S6F). These data demonstrate that receiver cells expressing an
anti-mCherry synNotch receptor that activates ETV transdifferentiated to endothelial cell
precursors via mCherry adsorbed on a culture surface.

55

56
Figure 6. mCherry activates ETV2 and reporter BFP to induce endothelial differentiation in
embryonic fibroblasts via synNotch. (A) Schematic of embryonic fibroblasts cell line (C3H)
expressing anti-mCherry/Gal4 synNotch activating ETV2 and BFP transgenes seeded onto
mCherry patterned substrate. (B) Fluorescence microscopy images of BFP reporter (top),
endothelial markers VEGFR2 (middle) and VE-Cadherin (bottom) stained 3 days following
seeding onto control wells (-mCherry) or plate-dried mCherry (+mCherry). Scale bars, 200μm.
(C) Percent of cells expressing BFP (left) and VEGFR2 (right) in the presence (+mCherry) or
absence (-mCherry) of mCherry, quantified with flow cytometry. Data represents mean ± s.d, BFP
n=12-13, VEGFR2 n=4-5, p<0.001(***), p<0.0001(****). (D) Heatmap of hierarchical clustering
of fibroblast parental cells without mCherry, fibroblasts engineered with anti-mCherry synNotch
activating ETV2 and BFP with and without mCherry, and BEnd.3 endothelial cells (positive
control) (E) Volcano plot of gene expression showing differentially expressed genes of antimCherry synNotch cells activating ETV2 and BFP on mCherry vs off mCherry, n=2. (F) Binary
mask used to generate stamps for microcontact printing of 500µm rows followed by fluorescence
microscopy images of day 3 nuclei staining, BFP expression, and VEGFR2 immunostaining. Scale
bars are 1mm. Plot profile of normalized BFP and VEGFR2 intensity on Day 3 following seeding
onto 500µm mCherry rows. Red bars indicate the regions containing mCherry. Line profiles
represent mean ± s.d, n=2. (G) Binary mask used to generate stamps of vasculature-like pattern
followed by fluorescence microscopy images of day 3 nuclei staining, BFP expression, and
VEGFR2 immunostaining.

57

58
Figure S6. (A) Fluorescence images of anti-mCherry synNotch fibroblasts with inducible ETV2 –
BFP expression in the presence of varying plate-dried mCherry concentrations (0-100μg/cm2).
BFP signal is shown in grayscale for visualization, images taken 1, 2, and 3 days following seeding.
Scale bar, 200μm. (B) Normalized BFP intensity of anti-mCherry synNotch fibroblasts seeded on
varying concentrations of plate-dried mCherry up to 3 days. Data represents mean ± s.d, n=3. (C)
Staining validation of anti-mCherry synNotch fibroblasts seeded on 15μg/cm2 mCherry with
endothelial marker VEGFR2. Panel shows sample stained with or without an anti-VEGFR2
primary antibody. Inducible BFP reporter and nuclei stained with HSC NuclearMask Deep Red.
Scale bars, 200μm. (D) Day 3 brightfield image of spatial endothelial-progenitor cell activation on
vascular-like pattern. Dotted white line indicates the region of interest enhanced in the following
panel. Scale bars, 1mm and 500μm, respectively. (E) Flow cytometry gating strategy to exclude
debris (Gate 1), isolate singlets (Gate 2), and determine BFP-expressing cells (top) or VEGFR2
expressing cells (bottom). (F) Measurement of CDH5 (VE-Cadherin) via immunostaining and
flow cytometry with and without presence of mCherry. Data represents mean ± s.d, n=3-4,
p<0.01(**). (G) Principal Component Analysis (PCA) comparing the transcriptome of unmodified
C3H fibroblasts, cell type-specific positive control cells (C2C12 and Bend.3), and receiver cells
that expressing mCherry, MyoD and mCherry, or ETV2 and BFP in the presence or absence of
their corresponding ligand. N = 2-4.

59
We also used Principal Component Analysis (PCA) to compare the transcriptome of
unmodified C3H fibroblasts, cell type-specific positive control cells (C2C12 and Bend.3), and
receiver cells that express fluorescent proteins, MyoD, or ETV2 in the presence or absence of their
corresponding ligand. As shown in Fig. S6G, the presence of the respective receptor-ligand pair
pushed receiver cells away from the unmodified C3H fibroblasts and towards the expected muscle
or endothelial cell line. Receiver cells expressing fluorescent proteins also clustered with
unmodified C3H cells in both the presence and absence of their respective ligand, as expected.
Receiver cells expressing anti-GFP synNotch that activates MyoD also had a significant shift from
the negative control cells towards C2C12 cells in the absence of GFP, suggesting non-specific
activation of the receptor. This was not observed for receiver cells expressing the anti-mCherry
synNotch that activates ETV2.
To test if we can also control the geometry of transdifferentiation for the endothelial
lineage, we generated uniformly adhesive surfaces and then microcontact printed mCherry in
varying designs. We designed a pattern to replicate a branching network structure typical of
vascular beds48 and showed the formation of a tissue consisting of activated cells in the
corresponding pattern surrounded by a uniform layer of fibroblasts (Fig. 6D and S6D). Fibroblast
receivers are activated by mCherry and express VEGFR2 based on the original ligand patterning,
where we evaluated a vascular branching pattern and 500 μm rows (Fig. 6F,G and S6D). Thus,
similar to the myogenic synNotch cells, microcontact printed ligands can activate SynNotchinduced transdifferentiation to endothelial precursors with spatial control.
Finally, we asked if we could engineer a tissue construct in which multiple distinct cell
fates are arranged in user-specified geometries. To do so, we first engineered a “dual-lineage” cell
line with two synNotch pathways: an anti-GFP/tTA synNotch receptor that activates myoD-

60
miRFP and an anti-mCherry/Gal4 synNotch that activates ETV2-BFP (Fig. 7A center). To test the
functionality and orthogonality of these pathways, we cultured these cells on surfaces with a
uniform coating of GFP or mCherry for three days and then stained for markers of differentiation.
As shown in Fig. S8A-C, cells transdifferentiated to α-actinin-positive muscle precursor cells or
VEGFR2-positive endothelial precursor cells, respectively. As a curiosity, we evaluated the effects
of culturing cells on both ligands, which would induce overexpression of both myoD and ETV2
in the same cells. In this case, it seemed that transdifferentiation to both pathways was impaired,
as these cells did not differentiate towards skeletal muscle nor express endothelial cell markers
(Fig. S8A). To prototype simple spatial activation, we used a micropipette to deposit droplets of
GFP and mCherry in opposing corners of a culture surface (Fig. S8B-C). Dual lineage cells
cultured on this surface activated the fluorescent protein reporters with expected spatial control,
and displayed multinucleation in the GFP-coated region, indicating feasibility for spatial activation
of differentiation.
Our next goal was to pattern multiple synNotch ligands onto a surface simultaneously and
with spatial control. To do so, we adapted approaches for controlling the distribution of multiple
streams of liquids with an open capillary microfluidic device100. Briefly, the intended fluid paths
are created as shallow channels that are laterally open and adjacent to deep channels. Fluids
preferentially travel along the shallow channels instead of the deep channels because of greater
surface tension in shallow channels. We used this concept to design a microfluidic device for
delivering solutions of GFP and mCherry as interdigiting 500 µm wide rows (Fig. 7A - left) and
fabricated it by casting PDMS on 3-D printed inverse templates. Air vents and GFP and mCherry
reservoirs were punched into the PDMS and the device was attached to a culture surface and loaded
with GFP and mCherry solutions. After overnight incubation, the PDMS device was removed and

61
remaining solutions were briefly air dried, leaving behind interdigitating rows of GFP and
mCherry adsorbed on the surface (Fig. 7B). When dual-lineage cells were cultured on these
surfaces, cells adhered uniformly to the entire surface and proceeded to transdifferentiate to
myoblasts or endothelial cells in a pattern corresponding to the intended pattern of ligands (Fig.
7B-C). As highlighted in Fig. 7C, α-actinin-positive muscle precursor cells were confined to the
GFP rows, VEGFR2-positive endothelial precursor cells were confined to the mCherry rows, and
intermixing of these two cell types was observed at the interface between GFP and mCherry. Cells
on the unpatterned regions remained fibroblasts.

62

63
Figure 7. Spatially controlled co-transdifferentiation into myogenic and endothelial lineages
of dual-lineage synNotch-engineered cells culturing on micropatterned ligands. (A)
Schematic showing dual protein patterning technique using capillary-driven microfluidic
patterning, based on shallow and deep channels, to generate parallel lines of GFP and mCherry.
Feature size, 500 µm. Schematic of dual-lineage mouse embryonic fibroblasts (C3H line)
expressing anti-GFP/tTA synNotch that activates MyoD and miRFP as well as an antimCherry/Gal4 synNotch that orthogonally activates ETV2 and BFP transgenes, seeded onto GFP
and mCherry patterned substrate. (B) Fluorescence images of GFP and mCherry ligand (left),
reporter genes expression (center) and brightfield (right) of dual-lineage cells 3 days following
uniform seeding onto micropatterned ligands. Dotted white rectangles represent regions of interest
for quantification. Scale bars, 1mm. Plot profiles on the right show normalized fluorescence
intensity of ligands and reporters taken in the region of interest. Green and red bars indicate regions
containing GFP or mCherry, respectively. (C) Center: merged fluorescence image of α-actinin and
VEGFR2 immunostaining on dual-lineage cells uniformly seeded onto GFP and mCherry pattern.
Dotted green lines represent region where the GFP signal has been subtracted to visualize
VEGFR2 staining. Scale bar, 1mm. Around the central image, higher magnification fluorescence
images taken within distinct regions of the pattern are shown: on mCherry (red border), interface
between mCherry and GFP regions (yellow border), on GFP (green border), and off pattern (beige
border). Scale bars, 200µm. Higher magnification of interface (blue border), visualizing α-actinin
and VEGFR2 staining is shown on the far right with a scale bar of 50µm. D) Schematic showing
the different ligand patterns used in single-nuclei sequencing experiments. T-Distributed
Stochastic Neighbor Embedding plot results of dual-lineage fibroblasts cultured on the four
different patterning conditions. Fibroblast-like cluster contains seven individual clusters, shown
here as one beige color. n=2. E) Percent of Fibroblast-like, muscle-like, and endothelial-like cells
across the four patterning conditions (left). Percent ratio of which muscle-like clusters make up
the total muscle-like cells in each patterning condition (right). n=2. F) Plot showing average
expression and percent expression of selected fibroblast, muscle, and endothelial markers in all
clusters across different patterning conditions. n=2.

64
Figure S7. (from Tyler Hoffman) (A) Fluorescence images of anti-GFP and anti-mCherry duallineage synNotch fibroblasts with orthogonally inducible MyoD-miRFP and ETV2 – BFP,
respectively, expression in the presence of varying plate-dried mCherry concentrations (0-
100μg/cm2). BFP signal is shown in grayscale for visualization, images taken 1, 2, and 3 days
following seeding. Scale bars, 200μm. (B) Normalized BFP intensity of dual-lineage synNotch
fibroblasts seeded on varying concentrations of plate-dried mCherry up to 3 days. Data represents
mean ± s.d, n=3. (C) Fluorescence images of anti-GFP and anti-mCherry dual-lineage synNotch
fibroblasts with orthogonally inducible MyoD-miRFP and ETV2 – BFP expression in the presence
of varying plate-dried GFP concentrations (0-100μg/ cm2). miRFP signal is shown in grayscale
for visualization, images taken 1, 2, and 3 days following seeding. Samples from each GFP
concentration were fixed and stained for α-Actinin following three days of culture. Scale bars,
200μm.

65
Figure S8. (A) Images of dual-lineage receiver cells cultured on GFP only, mCherry only, or both
GFP and mCherry. Surfaces were uniformly adsorbed with the ligands. Flow cytometry panel (XAxis: VEGFR2-APC, Y-Axis: BFP) of dual-lineage cells seeded on 15μg/cm2 plate-dried

66
mCherry unstained (left) or stained for VEGFR2 (center) and dual-lineage cells seeded on
15μg/cm2 plate-dried of both GFP and mCherry stained for VEGFR2 (right). (C) Fluorescence
images of GFP and mCherry droplet patterns (left), subsequent spatial reporter activation (center)
taken 48 hours after uniform seeding of engineered dual-fate fibroblasts. Scale bars, 2mm. Green
(within GFP pattern) and red (within mCherry pattern) borders indicate regions of interest with
corresponding higher magnification brightfield and fluorescence images (right). Scale bars,
200μm. (D) Brightfield image of dual-lineage cells seeded on droplet pattern taken 48 hours after
uniform seeding of engineered fibroblasts. Scale bars, 2mm. (E) Merged fluorescence images to
demonstrate the effects of subtracting the GFP pattern (high GFP intensity excluded) in visualizing
the α-actinin and VEGFR2 staining. Scale bars, 1mm. (E) Fluorescent microscopy images of duallineage cells on dual-ligand patterns showing α-actinin staining. VEGFR2 staining, and Nuclei
expression. Plot profiles of immunostaining of α-actinin and VEGFR2 showing fluorescence
intensity across distance.

67
Figure S9. (A) T-Distributed Stochastic Neighbor Embedding (t-SNE) plot analysis showing
twelve cell clusters based on gene expression profiles from all four conditions. (B) Selected marker
gene analysis for Fibroblast-like, Endothelial-like, and the four Muscle-like clusters.

68
Table 1. DAVID pathway analysis and Assigned Identity of clusters. Pathways rank from 1-10, 1
being the pathway with most genes activated in its given pathway. Pathways in bold are associated
with the cluster's assigned identity. For cluster 0 and 2, identity related pathways outside the top
10 were included with their rank in parenthesis. Each cluster was compared to the gene expression
of all clusters combined together. List of genes were selected through filtering with P < 0.01 and
Log2FC > 0.5.

69

Table 2. DAVID pathway analysis and comparison of UP and DOWN regulated pathways between
all muscle-like clusters. Pathways rank from 1-10, 1 being the pathway with most genes UP or
Down regulated between muscle-like clusters. Identity related pathways outside the top 10 were
included with their rank in parenthesis. List of genes were selected through filtering with P < 0.01
and Log2FC > 0.5.

70
To further evaluate the extent of dual-lineage transdifferentiation, we performed singlenuclei RNA sequencing on the dual-lineage cell line after three days of culture on substrates with
no ligand, GFP-only rows, mCherry-only rows, and interdigitating GFP-mCherry rows, patterned
using the technique shown in Fig. 7D. Due to the amount of culture area outside of the pattern,
ligands activated approximately half of the total sequenced cells. T-Distributed Stochastic
Neighbor Embedding (t-SNE) plot analysis identified twelve cell clusters based on gene
expression profiles from all four conditions (Fig. S9A). We analyzed the clusters for signature
genes and performed pathway analysis with DAVID101,102 (Table 1) to assign each cluster to a
putative cell type identity, resulting in seven fibroblast clusters, four muscle-like clusters, and one
endothelial-like cluster. As shown in Fig. 7E, more cells were induced towards the myogenic
lineage on GFP-only patterns and more cells were induced towards the endothelial lineage on
mCherry-only patterns, as expected. On the dual GFP-mCherry pattern, both myogenic and
endothelial clusters were detected (Fig. 7D-E). Selected marker gene analysis (Fig. 7F) showed
that fibroblast marker genes were overall down-regulated on patterns with GFP and/or mCherry,
as expected. Correspondingly, muscle-specific genes and endothelial-specific genes were overexpressed on patterns with GFP and/or mCherry, respectively (Fig. 7F). With this analysis, we
also detected the expected expression of the transgenes (transgenic myoD and BFP). Interestingly,
on the dual pattern, we observed more cells in the muscle-like 4 cluster compared to the other three
patterns, indicating that this cell identity may be unique to co-differentiation. The four muscle
clusters all express similar muscle marker genes, but at different relative levels. Pathway analysis
of differentially expressed genes revealed the four muscle clusters differ in pathways related to
cell cycle, ribosome, and differentiation (Table 2), suggesting that these four clusters may
represent similar cells at slightly different phases of the cell cycle or stages of differentiation.

71
Alternatively, co-differentiation may have unique impacts on cell phenotype, but additional
replicates and/or longer culture times are needed to reach a more clear conclusion. Thus, in
summary, by activating synNotch receptors with microfabricated biomaterials, we induced a single
population of fibroblasts to differentiate into a tissue with three distinct cell populations (skeletal
muscle precursors, endothelial precursors, and fibroblasts) patterned in user-defined microscale
geometries. Of note, these tissues were maintained in standard cell culture media, without the need
for soluble differentiation factors or biophysical stimulation to drive cell fates.
DISCUSSION
SynNotch was originally developed in cell lines and primary T-cells for applications in cell
therapy103–107. More recently, it has been implemented in different cell types, including a
transgenic mouse where synNotch is used for contact-dependent labeling of cells including
endothelial cells, hepatocytes, fibroblasts, pericytes, in vivo tissues108; and embryonic stem cells109
where it was used for controlling differentiation. In this study, we engineered several material-tocell signaling pathways to spatially activate user-defined genetic programs in multicellular
systems. We achieved this by engineering cells with synNotch receptors to define cellular inputs
and outputs while concurrently engineering materials to present synthetic ligands with different
ranges of spatial control. The variety of materials for synthetic ligand presentation yields powerful
and highly flexible tools for activating material-to cell pathways. Due to the functional modularity
of synNotch receptors, material-activated pathways can theoretically be used to drive any number
of transcriptional programs or differentiation pathways. These generalizable technologies are a
novel approach for dictating spatial patterning of gene expression in multicellular constructs,
without the need for soluble differentiation factors.

72
Because of the highly powerful level of transcriptional control over cell behaviors in
natural systems, many efforts in synthetic biology have focused on engineering sophisticated
transcriptional circuits110,111. In the area of stem cell and cell differentiation, genetic
overexpression of master transcription factors has demonstrated robust control over cell
differentiation112,113. Initially, the effect of only a handful of master transcription factors on cell
differentiation was known. However, more recently, approaches that collect entire organism
transcription factor libraries have become available, making it feasible to induce multiple
differentiation pathways with technologies such as synNotch114,115. The capacity to induce master
transcription factors with user-defined spatial control has inspired several recent advances, such
as engineering cells with light-activatable signaling pathways to gain spatiotemporal control over
cell behaviors with light116,117. For example, myoD overexpression has been induced by a
genetically integrated optogenetic switch that can be activated spatially35. Although optogenetic
approaches have the potential for powerful spatiotemporal control over cell behaviors, and have
recently been shown to be multimerized to up to three orthogonal pairs118, optogenetic
technologies require sophisticated light manipulation devices, which can be difficult to scale and
have limited penetration into 3-D tissues, and have not yet demonstrated robust multi-cell fate
control.
With our previous development of synNotch, we generated a novel way to activate userdefined genetic programs via user-defined ligands presented by neighboring cells. Here, we
advanced this technology to a new level by activating synNotch via micropatterned substrates for
tissue engineering. Importantly, we showed that this novel approach can be used to define spatial
patterns of not only gene expression, but also differentiation of two orthogonal lineages. Bulk RNA

73
sequencing and principal component analysis of single-lineage receiver cell lines on materials with
single ligands validated that cells were transdifferentiating towards the intended myogenic or
endothelial lineages. However, we also observed cells with synNotch activating MyoD clustered
further from the unmodified parental fibroblast cell line compared to cells with synNotch
activating ETV2, without any ligand. This is likely due to leakiness of the MyoD transgene and/or
the strength of the transcription factor itself relative to ETV2. Different transcription factors likely
have different activation amplitudes and dynamics and it would be crucial to identify appropriate
signal-to-noise ratios for each specific application by, for example, generating synNotch receiver
cells with different amounts of receptor and target gene constructs and assessing experimentally
which combination works more efficiently for the transgene of interest. Future advancements in
computational modeling and design can likely also help screen these combinations more
efficiently.
In terms of heterogeneity, we observed bimodal and therefore incomplete activation of
synNotch by ligands presented by materials, similar to other studies that have presented synNotch
ligands from cells or other materials. Across all materials we tested, we found that synNotch
activation reached a plateau in response to increasing ligand concentration, beyond which
synNotch activation did not increase. Thus, we likely reached the saturation point of ligand
presentation by the material and synNotch signaling itself seems to be the main factor limiting
activation. We correspondingly observed an imperfect differentiation efficiency, which is likely a
compounded effect of the heterogeneity of synNotch activation and the known heterogeneity of
transcription-factor-mediated differentiation, especially at the early time points that we
investigated in this study. These are major limitations of synNotch but will continue to improve as
the technology evolves.

74
In terms of dynamics, we found that different ligand-presenting materials yielded different
temporal patterns of synNotch activation. For example, synNotch activation peaked at three days
and then subsided when ligands were microcontact printed on PDMS, whereas activation was more
sustained when synNotch was activated by ligands conjugated to 3-D hydrogels. This could be
caused by differences in the conjugation of the ligands to the materials, such as the strength of the
material-ligand bond or ligand orientation, and/or differences in ligand-receptor engagement and
the activation of the synNotch receptor itself. The mechanism of transduction by synNotch
receptors is thought to proceed similarly to endogenous Notch receptors: there, in the core
regulatory region of endogenous Notch receptors, a pulling force is generated upon ligand binding,
which exposes a protease cleavage site for a protease that is constitutively active in the membrane;
this cleavage then liberates the intracellular domain which is a transcription co-activator119,120. The
mechanisms of activation of Notch and synNotch receptors via cell-presented ligands have been
compared, individuating possible mechanisms of activation that distinguish different synthetic and
natural receptor constructs121. Thus, the mechanism of activation of synNotch by materialpresented ligands may also differ from cell-presented ligands and may differ for different materials
with various chemical and mechanical properties. Increased mechanistic understanding of
synNotch activation by materials could yield increased capacity for spatial, and perhaps temporal,
control of gene expression.
Another interesting result from our study is the impact of both ligands on dual receiver
cells. For cells with two synNotch pathways that activate fluorescent reporters, both reporters were
expressed in cells cultured on both ligands. However, for cells with two synNotch pathways that
active myoD or ETV2 cultured on both ligands, both myogenic and endothelial differentiation

75
programs were stunted (S8A). To achieve dual differentiation in the presence of two ligands,
lineage bifurcation moduli (such as lateral-inhibition) or cross-inhibition to prevent the opposing
lineage could lead to a salt-and-pepper or checkerboard pattern of differentiation in regions with
both ligands. These types of approaches could also be combined with the MATRIX system to
provide ligands at specific time points and achieve more advanced spatial and temporal control
over differentiation, resulting in more complex tissue patterns.
To achieve greater spatial control, we microfabricated PDMS stamps and microfluidic
devices to pattern synthetic ligands onto 2-D surfaces. By fabricating these components on 3-D
printed templates instead of classical photolithography-based wafers, we achieved a wider range
of pattern designs and more rapid prototyping capabilities, with the tradeoff that spatial resolution
was confined to 100 μm or above. To pattern synthetic ligands at sub-cellular spatial resolution,
photolithography would still be required. The two PDMS-based patterning technologies that we
used also have tradeoffs. Microcontact printing can generate essentially any geometrical pattern
(including isolated islands) but cannot precisely register multiple ligands since each stamp must
be positioned manually. Conversely, registering the placement of multiple ligands is possible with
a microfluidic device, but pattern geometries are limited to continuous channels connected to a
reservoir. Thus, these constraints must be considered when choosing a patterning modality.
Together with other approaches, such as MATRIX89, these new approaches expand the library of
engineered biomaterials that activate synNotch.
Our most sophisticated tissue construct comprised interdigitating rows of skeletal muscle
and endothelial cells, with some intermingling of the cells at the interface. Importantly, the skeletal
muscle cells and endothelial cells were co-transdifferentiated from a single population of
fibroblasts. This approach is in contrast to conventional tissue engineering techniques, which

76
usually differentiate individual cell types in isolation and then combine them. Our approach may
better mimic natural tissue morphogenesis, where multiple cell fates emerge simultaneously from
a uniform cell population. Studies have also shown that supporting cells, such as endothelial cells,
improve the maturation of human induced pluripotent stem cell-derived cardiomyocytes122,123, and
that co-differentiation of different lineages concurrently more closely recapitulate the conditions
occurring during embryonic development124 An interesting hypothesis to explore with our
technology is whether co-differentiation of supporting cells (e.g., endothelial cells) adjacent to
parenchymal cells (e.g., muscle cells) has additional benefits for phenotypic maturity. For
example, our single-nuclei sequencing revealed one muscle-like cluster that was overrepresented
on the dual-ligand pattern compared to the GFP-only pattern. There are many potential
explanations for this, such as: (i) these muscle-like cells were uniquely influenced by the presence
of the co-differentiating endothelial cells; (ii) these muscle-like cells were located at the GFPmCherry boundary and thus were activated predominantly by GFP but also by mCherry to a lower
extent; and/or (iii) these muscle-like cells were coincidentally captured at a unique stage of cell
cycle or differentiation but are otherwise similar to the other muscle-like clusters. To tease apart
these different possibilities, additional replicates and longer time points will be needed.
Furthermore, the differentiation protocols we used here were very simplified; i.e., only three days
of differentiation in a basal medium with no additional soluble differentiation factors. Initial cell
state changes are expected in this timeframe, based on previous studies with induced transcription
factors, but more complete lineage conversion to mature cell types will require more time for
differentiation and potentially supplementation with soluble differentiation factors for some cell
types.

77
Patterning multi-lineage tissues with spatial control has also been achieved with multimaterial extrusion bioprinting. In a recent example, human stem cells were engineered with
doxycycline-inducible transcription factors for endothelial or neural cell fates36. Wildtype or
engineered cells were embedded in individual bioinks, merged into a tri-layer filament, and
extruded through a nozzle in user-defined patterns. Doxycycline and differentiation media were
then added to induce the co-differentiation of neural stem cells, endothelial cells, and neurons.
Although this is a powerful approach for multi-lineage tissue engineering, it does have its own
limitations, such as the reliance on diffusion-limited soluble differentiation factors, restrictions on
spatial resolution imposed by the nozzle, and a somewhat limited library of printable materials125.
However, we envision many synergistic opportunities for bioprinting and synNotch technologies
to be used together by, for example, bioprinting hydrogels that are functionalized with synNotch
ligands, such as the gelMA and fibrinogen that we synthesized in this paper. Optogenetic
technologies118, as mentioned above, can also be integrated to add more temporal control of cell
phenotype. Overall, the ongoing integration of synthetic biology, biomaterials, and
microfabrication technologies will further advance the capabilities for tissue engineering126.
Beyond bioprinting, we anticipate that our approach for activating synthetic pathways for
transdifferentiation by a material can be combined with other complementary technologies for
deriving complex in vitro tissues, such as organoids127. To generate organoids, stem cells are
exposed to natural ligands that orchestrate their self-organization into complex cellular
arrangements. However, although cellular complexity at the microscale in organoids is remarkably
similar to endogenous organs, users lack geometric control over the arrangement of cells at higher
levels, leading to tissue constructs that are largely heterogeneous and poorly reproducible with unnatural architectural features. Combining synthetic biology and organoids is a recognized frontier

78
of the field77,128–131 and synNotch-mediated spatial patterning technologies, such as those presented
here, could represent a step in the direction of ultimate user-control of cell behaviors across
multiple spatial scales for engineering in vitro multicellular systems.

79
Materials and Methods:
Genetic Constructs Design:
GFP and mCherry-responsive synNotch construction: pHR_SFFV_mycLaG17_synNotch_TetRVP64 (Addgene plasmid# 79128) and pHR_EF1a_flagLaM4_synNotch_Gal4-VP64, built from pHR_EF1a_flag-LaM4_synNotch_TetRVP64
(Addgene plasmid#162237) and HR_pGK_LaG17_synNotch_Gal4VP64 (Addgene plasmid#
79127). The response-element plasmids pHR_TRE_MyoD-P2A-mCherry, pHR_TRE_MyoDP2A-miRFP703_PGK_PuromycinR, and pHR_UAS_ETV2-P2A-tBFP_PGK_HygromycinR
(with and without transcription factor) were generated from pHR_TRE, pHR_5x Gal4 UAS
(Addgene plasmid# 79119), mouse MyoD (NP_034996.2), and mouse ETV2 (NP_031985.2).
All constructs were cloned via In-Fusion HD Cloning (Takara Bio).
Lentivirus Production:
Lentivirus was produced by cotransfecting pHR cloned plasmids with vectors encoding packaging
proteins (psPAX2, pVSVG) using Lipofectamine LTX (ThermoFisher) into 70-80% confluent
HEK-293T cells within 6-well plates. Viral supernatants were collected 2-3 days after transfection,
sterile filtered with 0.45μm PES (Genesee Scientific), and used directly or 10x concentrated using
LentiX Concentrator (Takara Bio) following manufacturers instructions prior to adding to cell
lines.
Cell Culture:

80
L929 mouse fibroblast cells (ATCC# CCL-1), HEK293 cells (Takara 632180), C3H/10T1/2 Clone
8 (ATCC# CCL-226), and NIH/3T3 (ATCC# CRL-1658) were cultured in DMEM
(ThermoFisher) supplemented with 10% Fetal Bovine Serum (ThermoFisher) and 100U/mL
penicillin/streptomycin (ThermoFisher). Cultures were maintained in a 37°C incubator with 5%
CO2 and relative humidity (VWR).
Cell Line Engineering:
For viral transduction, 20-50μL concentrated (or equivalent non-concentrated) viral supernatant(s)
were added to 5-10x104 suspended cells supplemented with 10μg/mL polybrene (Sigma), then
transferred into a 12-well plate for 2-3 days before changing to fresh media. Following
transduction, all applicable cell lines were selected using Puromycin (L929 – 10μg/mL, C3H – 1
μg/mL) and Hygromycin B (L929, C3H – 400μg/mL, MedChem Express) for the expression of
transgenes. Cells were sorted for the coexpression of each component via fluorescence-activated
cell sorting on a FACS ARIA II (Beckton-Dickinson) by staining with appropriate fluorescently
tagged anti-Myc and anti-Flag antibody for 30 minutes at 4°C (Cell Signaling Technologies) or
expression of the transgenes. A bulk-sorted polyclonal population of engineered cells were used
for experiments, unless otherwise noted. For single-cell clonal populations, single cells were sorted
individually into 96-well plates from selected and stained populations using a FACS ARIA II.
GFP and mCherry Production:
GFP, mCherry, and GFP-LACE (pET28-His6-GFP-C-LACE, gift from Jeffrey Bode Addgene
plasmid # 133913) were purified as an N-terminal hexahistidine fusion protein. To express GFP,
BL21(T1R) E. coli cells were grown to an optical density of 0.5 from an overnight-grown glycerol

81
stock, chilled to 25°C, induced with 1 mM IPTG and allowed to express for 5 hours. To express
mCherry, BL21-AI E.coli (Thermo Fisher) were transformed with mCherry-pBAD (gift from
Michael Davidson & Nathan Shaner & Roger Tsien, Addgene plasmid # 54630), grown to an
optical density of 0.6 from an overnight-grown glycerol stock, induced with 0.04% w/v LArabinose (Sigma), and allowed to express for 5 hours, based on previous studies
(https://dergipark.org.tr/en/pub/iarej/issue/44303/429547). The proteins were purified by
NEBExpress Ni Spin Columns (New England Biolabs) following manufacturer's instructions,
dialyzed against 1x PBS overnight at 4°C, sterile filtered, and frozen at -80°C until use.
Gelatin Hydrogel Surface Conjugation:
Gelatin hydrogels were fabricated as previously described96,132. Briefly, a 30W Epilog Mini 24
laser engraver (100% speed, 25% power, 2500 Hz) was used to cut a 150-mm polystyrene dish
into 260-mm2 hexagons. Each hexagon was masked with tape, and an inner circle was cut (18%
speed, 6% power, 2500 Hz) and removed, exposing a polystyrene surface which was then treated
with plasma (Harrick Plasma) for 10 minutes to improve gelatin adherence to polystyrene. Equal
volumes of a 20% porcine gelatin solution (Sigma) and 8% MTG (Ajinomoto) solution were mixed
and 200 μl were added to each coverslip. Flat or 10x10 μm micromolded PDMS stamps were
immediately applied to shape surface topography. After an overnight incubation to solidify, the
hydrogels were rehydrated in water, and the stamp was removed. Coverslips were stored in PBS
at 4°C until cell seeding. (This protocol was performed by Stephanie Do)
PDMS stamps with 10x10 μm grooves of 2 μm height were fabricated with standard
photolithograpy and soft lithography techniques43. Flat PDMS was used as a control substrate with

82
no topography. A 1:1 ratio of GFP (500µg/ml, 200 µg/ml, and 20 µg/ml) and MTG (8% w/v)
solution were added to a parafilm surface, and the gelatin coverslip was inverted onto the GFPMTG droplet for 10 minutes132. The coverslips were then incubated for 1 hour at 37°C. Following
incubation, the coverslips were washed 3 times with warm PBS.
L929 anti-GFP synNotch receiver cells were seeded at a density of 350,000 cells per coverslip and
cultured for 72 hours. To detach the cells from the gelatin for flow cytometry analysis, the gelatin
hydrogels were minced with a sterile X-acto knife and incubated in a 4 mg/ml collagenase IV
solution for 45 minutes at 37°C. Digested gelatin was then filtered with a 40 µm cell strainer.
C3H anti-GFP synNotch MyoD expressing cells were seeded at a density of 500,000 cells per
coverslip, and cultured for 4 days. Coverslips were washed three times with warm PBS, fixed with
ice-cold methanol and immunostained with mouse α-actinin primary antibody (Sigma, A7811) at
1:200 dilution for two hours. Coverslips were then stained with the secondary antibody goat antimouse conjugated to Alexa Fluor 546 and 4′,6-diamidino-2-phenylindole (DAPI) at 1:200
dilutions for 2 hours. ProLong gold antifade mountant (Thermofisher) was used to mount cells on
glass coverslips.
Myotube count was performed in ImageJ through size and intensity thresholding of the α-actinin
signal. The total number of cells per image was calculated by size and intensity thresholding of
the DAPI signal. Myogenic index was determined by dividing the number of co-localized nuclei
within the α-actinin signal by the total number of nuclei in the field of view. The Orientation Order
Parameter of both the myotubes and nuclei were determined by first analyzing images of the α-

83
actinin and DAPI signal, respectively, using the OrientationJ Distribution plugin in ImageJ83. This
plugin was used to generate a histogram with the number of pixels locally oriented along every
angle at 0.5 degree increments. This histogram was then analyzed using MATLAB code to
calculate the Orientation Order Parameter133, which ranges from 0 for completely randomized
systems to 1 for perfectly aligned systems.
Microcontact Printing:
For all surfaces except Fig. S5B (see below), 18 mm glass coverslips were spin-coated with PDMS,
treated with UV ozone for 3 minutes, and incubated in 10% APTES in ethanol for 2 hours at 50°C.
Coverslips were then rinsed with water and incubated in 2% glutaraldehyde solution in ethanol at
room temperature for 1 hour41. Coverslips were then rinsed again and inverted onto 150µL droplets
of 50 µg/mL fibronectin in distilled water in a Petri dish, which was then sealed with Parafilm and
incubated overnight at 4°C. Cylindrical isotropic stamps were cut from a slab of PDMS using an
8mm diameter biopsy punch. GFP at 0, 10, 50, 100, or 200 μg/mL was coated onto the isotropic
stamps and left for 1.5 hours at room temperature until the solution was dry. The stamps were then
briefly dipped into sterile water, air-dried using compressed air, and inverted onto fibronectincoated coverslips. C3H anti-GFP synNotch mCherry expressing cells were seeded onto patterned
coverslips at 650,000 cells per coverslip in a 12-well plate.
To generate micropatterns of GFP, Solidworks was used to design desired patterns (square arrays
ranging from 100 μm to 1 mm, concentric circles, aligned rows, and letters), which were then
printed into templates using a digital light processing (DLP) 3-D printer (CADworks3D). After
3D printing, the templates were placed in 200 proof isopropyl alcohol overnight to ensure all

84
uncured resin was removed. The templates were then UV cured for 1 hour (back side for 20
minutes, feature side for 40 minutes) to finalize the curing process. PDMS (Sylgard 187) was
poured into the templates, desiccated for 30 minutes, and cured overnight in a 65°C oven. PDMS
stamps were then removed from the templates. Microcontact printing with these PDMS stamps
was performed the same as with isotropic stamps at 100 μg/mL GFP. For perpendicular row
patterns, one stamp with aligned rows was coated with GFP (200 μg/mL) and another similar stamp
was coated with mCherry (200 μg/mL). These stamps were manually positioned sequentially in a
perpendicular orientation. Coverslips were stored dry at 4°C until use and incubated in
DMEM+10% FBS for a minimum of 1 hour prior to cell seeding. Coverslips were then seeded
with C3H anti-GFP synNotch mCherry expressing cells, C3H anti-GFP synNotch MyoD
expressing cells, or monoclonal dual receiver cells at a concentration of 650,000 cells per coverslip
in a 12-well plate.
For the surfaces in Fig. S5B, PDMS-coated coverslips were treated with UV ozone for 8 minutes,
then microcontact printed with a stamp coated with a mixture of 100 µg/mL GFP and 50 µg/mL
FN. After patterning, coverslips were incubated in 2% Pluronic in distilled water for 15 minutes
at room temperature and rinsed with PBS45.
Dual Ligand Patterning with Capillary Microfluidic Device:
Solidworks was used to design a 4-row capillary fluidic device with two disconnected inlets.
Shallow channels (100µm distance from the substrate) were designed to guide protein solutions.
These shallow channels were surrounded by 1 mm deep channels, intended as voids. The
inverse design was 3-D printed using a DLP printer (CADWorks), which was then replica

85
molded in PDMS. Inlets were created using 1.5mm biopsy punches and air ventilation punches
were created on two opposite sides ends of the device to allow optimal pressure for capillary
fluid transfer. Prior to protein patterning, the feature side surface of the device was UV plasma
treated for 7 seconds, creating a hydrophilic surface for capillary action. The device was then
placed facedown onto tissue culture treated Ibidi 2-wells. GFP (500µg/mL) and mCherry
(1000µg/mL) were then pipetted into separate inlets and filled their respective shallow channels.
The device was then placed into a petri dish and parafilm sealed before overnight incubation at
4°C. The next day, the device was incubated without parafilm at room temperature for 15
minutes. The fluidic device was then carefully removed from the Ibidi wells to minimize liquid
disruption. The Ibidi well was left at room temperature for 15 minutes for the protein solutions to
dry. The Ibidi well was then UV treated under the biosafety cabinet for 1 hour to sterilize.
DMEM with 10% FBS was pipetted into the wells and incubated for 1 hour before cell
seeding. Dual-lineage cells were seeded at 1.9x105
/cm2 and cultured for three days prior to
fixation and staining for α-actinin and VEGFR2.
Data Quantification:
We studied the spatial control dynamics over time by measuring the Pearson’s Coefficient on days
2, 5, and 10. This was done using the JACoP plugin in ImageJ by comparing the binary mask to
the mCherry channel. Line plots were quantified using the “Plot Profile” feature in ImageJ and
normalized to individual images. The number of nuclei in the field of view of each image were
counted, and the number of myotubes was counted using ImageJ. Myogenic index was calculated
through dividing the number of nuclei within each myotube by the total number of nuclei. The
GFP channel was used to determine which nuclei were on and off-pattern. For calculating on-

86
pattern myogenic index, all nuclei outside the GFP pattern were excluded. The sarcomeric αactinin mask was overlaid on the on-pattern nuclei and used to calculate the myogenic index.
Similarly, to quantify off-pattern myogenic index, we excluded all nuclei located within the GFP
patterns and used the same sarcomeric α-actinin mask to measure the myogenic index. For
Coherency quantification, 200 and 500µm rows and curves were thresholded using the same
methods used to create the myotube mask, except instead of using the thresholded image to create
a selection/mask, we quantified the thresholded myotube image itself. The OrientationJ plugin on
ImageJ was used to quantify the coherency of all patterns. Since curved rows are not straight, we
need to straighten them to get a fair quantification of how the myotubes align with the curves. A
fragmented line was drawn manually following the GFP pattern of the curve and used to straighten
the myotube threshold image before quantifying the coherency. All data was processed in
GraphPad Prism 9 and validated with statistics.
Orientation Order Parameter: Matlab Code
To calculate the Orientation Order Parameter (OOP), OrientationJ Distribution plugin and an
Orientation Order Parameter code in MATLAB were used. First, the OrientationJ Distribution
plugin was used to obtain a list of the number of pixels (Y) oriented along every 0.5 degrees (X).
This matrix of X and Y values were then inputted into the MATLAB OOP code. The OOP code
converts the degrees to radians and uses sine and cosine to determine the direction of each myotube
using a vector. Each myotube is represented by a vector in the 2D space.
ri = [rix ,riy ]

87
This was then converted into a 2x2 matrix by computing the outer product of the vectors and
normalizing.
Next, the largest eigenvalue in the tensor corresponds to the OOP. A score of 0 is a completely
randomized orientation in the field of view whereas a score of 1 equates to perfect alignment of
the myotubes.
Code: OOPcalc
function OOP = OOPcalc(orientation_Histogram)
clc;
input = orientation_Histogram;
list = [];
for i=1:180
%if input(i,1) ~= 0
temp = input(i,1)*ones(1,input(i,2));
list = [list, temp];
%end
end
angles = deg2rad(list);
r(1,:) = cos(angles);
r(2,:) = sin(angles);
OOT_All = zeros(2,2,length(r));
for i=1:2
for j=1:2
OOT_All(i,j,:)=r(i,:).*r(j,:);
end
end
OOT_Mean = mean(OOT_All,3);
OOT = 2.*OOT_Mean - eye(2);
[directions,orient_parameters] = eig(OOT);
[Orientation_order_parameter,I] = max(max(orient_parameters));
display(['Orientational Order Parameter = ',...
num2str(Orientation_order_parameter)])
OOP = Orientation_order_parameter; % Assign the result to the output variable OOP
end
Code: Orientation_Histogram
% column 1: angles from 1 to 180

88
% column 2: random frequencies for each angle
%orientation_Histogram = [(1:180)', randi([0, 50], 180, 1)];
%Input data from the table of imageJ into brackets below
orientation_Histogram = []
% Call OOPcalc function with the created histogram
OOP = OOPcalc(orientation_Histogram);
These two MATLAB scripts were used to calculate the Orientation Order Parameter (OOP)
Plate-drying of Ligand:
For single-ligand activation, ligands (mCherry and GFP) were prepared at 100μg/mL in sterile
DI water and added at 15μg/cm2
. For dual-ligand patterning, 8uL droplets of ligand at 200μg/mL
in sterile DI water were deposited in distinct regions within each well. Plates were left to dry in
the biosafety cabinet overnight, protected from light and then washed once with PBS prior to cell
seeding. Anti-mCherry synNotch ETV2-BFP or dual-lineage fibroblasts were seeded at 5-
20x104 cells/cm2 and cultured for three days prior to flow cytometry analysis or fixation and
staining for VEGFR2.
Staining:
Flow Cytometry: cells were detached using TrypLE (ThermoFisher) and washed once prior to
incubation with fluorescently-tagged antibodies in PBS+5%FBS for 30 minutes-1 hour at 4°C.
Following, cells were washed twice with PBS+5%FBS and filtered through 35μm cell strainer
prior to analysis with ARIA II.

89
Following culture, cells were washed once with PBS, fixed with 4% paraformaldehyde or 10%
ice-cold methanol for 10 minutes and then washed 3x with PBS for 5 minutes each. Samples were
stained immediately or further permeabilized with 0.1% Triton X-100 in PBS for 5-10 minutes
and then washed 3x with PBS for 5 minutes. Cells were blocked for 1 hour with 2% BSA at room
temperature, then incubated with primary antibodies for 2 hours at room temperature or overnight
at 4°C. Following three washes with PBS, samples were incubated with secondary antibodies for
one hour at room temperature, then washed again prior to imaging directly or staining nuclei with
NucBlue (15 minutes, ThermoFisher) or Nuclear Mask Deep Red (30 min, ThermoFisher).
Samples on coverslip were mounted with gold-antifade mounting solution (ThermoFisher).
Imaging/Microscopy:
Unless otherwise stated, a digital Microscope (Keyence BZ-X) was used to image experiments.
Tiling was done with the built-in Keyence software. All images within individual experiments
were taken with the same settings (Light strength, exposure, No LUT). BFP, GFP, mCherry, and
miRFP signals were captured using the respective Filter cubes: BFP, GFP, TexasRed, Cy5-NX.
RNA Sequencing
For bulk RNA sequencing analysis, GFP or mCherry solid circle patterns were created via
microcontact printing with 8mm diameter stamps. 5x104 of the following cells were droplet seeded
with and without the presence of their respective ligands: C3H parental (no-ligand only), antiGFP/tTA synNotch that activates mCherry, anti-GFP/tTA synNotch that activates myoD and
mCherry, anti-mCherry/Gal4 synNotch that activates ETV2 and BFP, C2C12 cell line (no-ligand

90
only), and BEnd.3 cell line (no-ligand only). To ensure all cells were cultured on the activating
ligand, cells were seeded as 50μL droplets within the borders of the patterned ligands and allowed
to adhere for 30 minutes before pipetting in the rest of the culture media. The same cell seeding
strategy was performed for conditions without ligands. Cells were cultured for before total
RNA extraction using miRNeasy kit per manufacturer’s protocol(Qiagen). An RNA cleanup kit
was used to further purify/clean the samples (Zymo RNA Clean and Concentrator). RNA samples
were then sequenced with an Illumina NovaSeq 6000 (Novogene Corporation Inc).
Fastqc files were trimmed with trimmomatic v0.39 using default settings. Trimmed fastQ files
were aligned to GRCm38 reference genome supplemented with custom transgenic sequences using
STAR v2.7.10b (134) with default parameters. Transcriptome alignments were quantified using
featureCount (135) using the custom gene annotation file combining GENCODE annotation file
and transgenes. Gene counts were imported into R and differentially expressed genes were
identified with DESEq2 v1.38.3 (136) with padj = 0.05 as the threshold. GO analysis was performed
on the differentially expressed genes using the clusterProfiler v4.6.2 (137) package.
10x Single-Nuclear RNA Sequencing
For single-nuclear RNA sequencing analysis, patterns were prepared using capillary fluidic device
to contain either GFP in both inlets, mCherry in both inlets, or GFP/mCherry in one inlet each to
generate a dual-ligand pattern. 3.3x104 dual-lineage fibroblasts were seeded in a 30uL droplet on
top of the patterns, allowed to attach for 30 minutes, prior to adding additional media. Cells were
cultured for an additional 3 days before collection. To collect, cells were trypsinized and cells were
lysed in IGEPAL CA630-containing lysis buffer for 7 minutes to isolate individual nuclei.

91
Library construction was performed according to the manufacturer’s protocol (10x Genomics
single cell 3’ v3.1 protocol). Briefly, after resuspension and counting, 16,000 GCs per experiment
were resuspended in master mix and loaded (together with partitioning oil and gel beads), onto
each lane of an 8 lane chip G to generate the gel bead-in-emulsion (GEMs). Reverse transcription
was primed with an oligonucleotide carrying an Illumina TruSeq R1 read-sequencing primer, a 16
nucleotide 10x cell barcode, a 12 nucleotide UMI, and a 30 nucleotide anchored poly dT sequence.
Full length cDNA was amplified from heteroduplex RNA:cDNA using 12 cycles of PCR. The
full-length cDNA was cleaned up on SPRIselect beads, and QCed on Qubit and BioAnalyzer. One
fourth of the resulting ds cDNA was fragmented and prepared for sample index PCR, with 11
cycles of amplification. After QC, the libraries were pooled and submitted for sequencing on 2
lanes of a 10B 100 flowcell on the Illumina NovaSeqX sequencer, targeting a minimum read depth
per cell of 25,000. Sequencing was performed at the UCSF CAT, supported by UCSF PBBR,
RRP IMIA, and NIH 1S10OD028511-01 grants.
FASTQ files were processed with 10x Genomics’ Cell Ranger analysis pipelines. The read count
matrix generated by CellRanger was then analyzed using Seurat v5.0.2. 32288 genes were detected
across no ligand (7877 and 7745 cells tested for replicates), GFP pattern (10539 and 11923 cells
tested for replicates), mCherry pattern (10671 and 10068 cells tested for replicates), and Dual
Pattern (8710 and 16266 cells tested for replicates). Cells that had unique feature counts with at
least 700 genes but no more than 7000 genes and cells that had <55% mitochondrial counts were
filtered and normalized based on the feature expression and total expression of each cell. The
normalized expression data were then used for subsequent analysis.

92
Principal component analysis was performed after merging replicates and integrating all the
conditions. Highly variable genes in each sample after linear transformation and the first 30 PC
scores were used for tSNE analysis to cluster the cells into 12 groups (FindNeighbors and
FindClusters functions implemented in the Seurat package, dims = 30, resolution = 0.4). The
marker genes of each cluster were identified using FindAllMarkers or FindMarkers function with
default parameters. Clusters were annotated using signature genes and DAVID pathway analysis
to identify fibroblast-, muscle-, or endothelial- like cell types across the different conditions. tSNE
clusters that were enriched in proliferation, extracellular matrix, or EGF pathways were identified
as fibroblasts. Clusters that were enriched in lineage-specific markers, muscle or angiogenesis
pathways, were used to identify muscle- and endothelial- like clusters, respectively.
A pseudobulk method was applied to investigate gene expression among different conditions at
the population level. Specifically, the raw gene counts of each sample were extracted after filtering.
The counts were then aggregated to the sample level and the expression of genes of interest
including transgenes were examined across conditions.
Statistics:
Individual data points in graphs represent distinct samples. Statistics were calculated in Prism,
using Unpaired T-test two-tailed or one-way Anova between groups. * p<.05, ** p=<.01, ***
p=<.001, **** p=<.0001
Company Catalog Target Fluorophore Dilution
Cell Signaling
Technologies
3739S Anti-Myc
(EQKLISEEDL)
PE FC: 1:50
Cell Signaling
Technologies
15008S Anti-FLAG
(DYKDDDDK)
AlexaFluor
488
FC: 5uL/106
cells
Abcam ab45688 Anti-Fibronectin IF 1:500

93
BioLegend 136401 Anti-VEGFR2 IF: 1:100
BioLegend 136405 Anti-VEGFR2 APC FC: 1:100
Sigma A7811 Anti-α-Actinin
(Sarcomeric)
IF: 1:200
ThermoFisher A-11030 Anti-Mouse IgG AlexaFluor
546
IF: 1:200
Abcam ab150153 Anti-Rat IgG AlexaFluor
488
IF: 1:200
ThermoFisher A-2144 Anti-Rabbit IgG AlexaFluor
647
IF: 1:1000
ThermoFisher A-21235 Anti-Mouse IgG AlexaFluor
647
IF 1:200
ThermoFisher H10294 HSC NuclearMask
Deep Red
IF: 1:250
Abcam ab228551 Hoechst 33342 IF: 1:20,000
ThermoFisher R37605 Hoechst 33342 IF:
2drops/mL
Contributions
M.G., T.H., T.M., S.D., A.R.M., N.C., S.L., M.L.M., L.M. designed the experiments. M.G.,
T.H., T.M., S.D., N.P., R.E.L., J.S., B.J., performed the experiments. M.G., T.H., T.M., S.D.,
R.E.L., B.J., analyzed the data. M.G., T.H., T.M., S.D., A.K., S.L., M.L.M., L.M. contributed to
data interpretation and discussion. M.G., T.H., T.M., S.D., S.L., M.L.M., L.M. wrote the
manuscript.
Conflict of interest
The technology transfer office of USC with the authors have filed patent disclosures with the
technology described here; LM is an inventor on a previous synNotch patent for applications in
cancer cell therapy licensed to Gilead; MLM is an inventor on a patent on gelatin hydrogels
licensed to Emulate.

94
Funding Sources and Acknowledgments
The authors acknowledge Marion Johnson and all the members of the Morsut, McCain, Li,
Khademhosseini Lab for insightful discussions and suggestions on the project. The authors
acknowledge their family and friends that support them always and in particular for the times of
this work that took place during the COVID-19 pandemic. Research reported in this publication
was supported by NIGMS of the National Institutes of Health under award number R35GM138256
(LM); NSF RECODE from CBET-2034495 (MLM, LM); USC Department of Stem Cell Biology
and Regenerative Medicine Startup Fund (LM), Viterbi Center for CIEBOrg (LM MLM), TH
acknowledges support from the Ruth L. Kirschstein National Research Service Award
T32HL069766 and the UCLA Eli and Edythe Broad Center of Regenerative Medicine and Stem
Cell Research, Research Award. Fellowships for students: CIRM fellowship for SD, Fellowships
from BME Department for first year for MG, NC and SD. Grace True, Finacy Jin for technical
support for the project. SL acknowledges the support of the Innovation Award from the UCLA Eli
and Edythe Broad Center of Regenerative Medicine and Stem Cell Research, and grants
(GM143485 and NS126918) from the National Institute of Health.
Patent Submission: Synthetic Notch Receptors for Use in Customized Spatial Control of
Multiple Gene Expressions and Uses Thereof
The work me and my colleagues have completed has resulted in a patent submission.

95
Chapter 3 - Creating Engineered Muscle-Tendon-Bone Tissue for Regenerative Medicine
Introduction
One of the fields that we predict our tissue engineering approach will be valuable in is the
field of regenerative medicine. Currently, there are no in-vitro models that reproducibly
recapitulate a seamless transition of muscle-tendon-bone tissue, which is a key part of the
musculoskeletal system. Musculoskeletal diseases, such as arthritis, are common and extremely
detrimental in affected patients 138,139. These diseases can significantly impair the quality of life,
causing pain, reduced mobility, and disability. Thus, creating effective models to study and
eventually treat these conditions is of paramount importance.
To create an in-vitro model to better study these diseases, we will build a muscle-tendonbone tissue junction using the SynNotch system. Mesenchymal stem cells (MSCs) are a great
candidate for this as they are promising for tissue regeneration given their immunologic tolerance
and multipotency, especially towards the muscle and skeletal lineages140. MSCs have been shown
to differentiate into various cell types, including osteoblasts (bone cells), myocytes/myotubes
(muscle cells), and tenocytes (tendon cells), under appropriate conditions. This multipotent
capability, combined with their ability to modulate immune responses, makes MSCs a versatile
and valuable cell source for creating complex tissue constructs.
MyoD overexpression has been shown to transdifferentiate hMSCs into multinucleated
myotubes141. Exploring this system in MSCs will be the first step of this project, as we already
have experience with the myogenic lineage and myoD expression via SynNotch. The potential of
MSCs to differentiate into tenocytes has also been explored in the literature. Ectopic expression of

96
Scleraxis (Scx) has been found to drive human bone-marrow derived MSCs towards tendonprogenitors 142. Taking this a step further, a study that utilized both gene expression and scaffold
engineering demonstrated the tendon tissue formation using overexpression of Scx in parallel with
a knitted silk-collagen scaffold143. The osteogenic differentiation potential of MSCs is also welldocumented. Specifically, Runx2 has been heavily studied and validated as an osteogenic master
regulator144. One particular study found overexpressing Runx2 in human MSCs resulted in
osteogenic programming due to the expression of osteogenic markers and mineralized matrix145.
This project aims to harness the SynNotch system's precise spatial control over gene
expression and MSCs' regenerative potential to engineer a functional muscle-tendon-bone
interface. By integrating MSCs with engineered SynNotch receptors and using material
engineering and protein micropatterning to spatially control ligand presentation, we aim to direct
the differentiation of MSCs into specific cell types in a spatially controlled manner. We aim to test
and use myoD, Scx, and Runx2 to transdifferentiate MSCs into muscle, tendon, and bone,
respectively. This approach will enable us to create a more accurate in-vitro model of the
musculoskeletal junction, facilitating better studies of disease mechanisms, drug testing, and
ultimately, the development of regenerative therapies.

97
Results
The first step in establishing an in-vitro model for the seamless transition of muscle-tendonbone tissue is to create the reporter line for fine-tuning the dynamics of the SynNotch system in
MSCs. This will ensure a baseline understanding of the MSC system without activating specific
bias with transcription factors.
To generate MSC GFP senders and anti-GFP mCherry reporters, we utilized lentiviral
transduction followed by fluorescence-activated cell sorting (FACS). Figure 8A demonstrates the
successful expression of GFP in MSCs post-transduction. GFP channel images show a clear
distinction between non-transduced MSCs and those transduced with a high virus titer, where GFP
expression is evident.
Furthermore, we engineered MSCs to express anti-GFP SynNotch receptors and
downstream mCherry reporters. As shown in Figure 8B, varying virus concentrations resulted in
different levels of mCherry expression, highlighting basal leaky expression. Cells transduced with
high virus concentrations exhibited more pronounced mCherry fluorescence compared to those
with low virus concentrations. From these four concentrations, the high receptor and the low
transgene population looked to be the most promising for FACS due to its relatively low basal
mCherry expression and healthier-appearing morphology from the images.
To quantify GFP expression and sort cells, we performed FACS. Figure 8C presents
histograms of fluorescence intensity in the FITC channel for both parental MSCs and GFP senders.
The data reveal that approximately 64.7% of GFP sender MSCs fell into the high FITC+ gate,
indicating successful GFP expression. These high expressing cells were bulk sorted and used in
following experiments. In Figure 8D, the Texas Red channel and Forward Scatter Area plots
further illustrate the sorting of anti-GFP mCherry reporters based on mCherry fluorescence

98
intensity. Cells were sorted into low, medium, and high mCherry expression gates, with higher
gate cells showing stronger mCherry signals.

99

100
Figure 8. Lentiviral transduction and FACS gating strategy for MSC GFP senders and antiGFP mCherry reporters. (A) MSCs expression of GFP after 48 hours of viral transduction (right)
vs without virus (left). (B) Different concentrations of receptor and downstream transgenes used
to transduce MSCs to engineer anti-GFP mCherry reporters, resulting in varying levels of basal
mCherry leaky expression. (C) Histogram showing fluorescence in FITC channel of parental MSC
(left) and MSC transduced with membrane presenting GFP ligand plasmid (GFP sender high
virus). Cells were sorted in the FITC + side of the gate, collecting high GFP expressing MSCs. (D)
Texas Red channel and Forward Scatter Area plot of Parental MSC and anti-GFP mCherry
transduced MSC (high receptor, low transgene). Cells were sorted using low, medium, and high
mCherry fluorescence gates.

101
To assess the functional activation of anti-GFP mCherry reporters by GFP senders, we cocultured low and high gate-sorted reporters in 1:1 ratio with GFP senders. Figure 9A shows Day 2
images where both low and high gate-sorted anti-GFP mCherry reporters displayed activation
when cultured with GFP senders, evidenced by mCherry expression. The intensity of mCherry
expression was higher in the high gate-sorted cells, corroborating the sorting data from Figure 8.
The high sorted cell line for anti-GFP mCherry reporters worked best here, with low basal
activation and high positive mCherry response. We expanded and used these cells for the following
experiments.
We then tested the spatial activation of anti-GFP mCherry reporters using surfacepresented GFP via droplet seeded and dried GFP. Figure 9B illustrates the mCherry expression in
high gate-sorted anti-GFP mCherry reporters cultured on varying concentrations of GFP droplets
(20 µg/mL, 200 µg/mL, 500 µg/mL). The images demonstrate a clear spatial expression of
mCherry correlating with the concentration of surface-presented GFP, confirming that our system
can drive patterned tissue differentiation through localized SynNotch activation. We found that the
cells activated best on 200 µg/mL GFP and that 500 µg/mL was not needed.

102
Figure 9. MSCs engineered with anti-GFP mCherry reporter are activated by MSC GFP
senders and surface presented GFP. (A) Day 2 images of low and high gate sorted anti-GFP
mCherry reporters cultured with GFP senders. (B) Day 2 images of high gate sorted anti-GFP
mCherry reporters cultured on varying concentrations of 5 µL GFP droplets demonstrating spatial
expression of mCherry.

103
Discussion and Future Directions
Our results indicate that the SynNotch system can be effectively employed to engineer
MSCs with spatial control over gene expression through cell presented ligands and surface
presented ligands. An immediate next-step forward would be to test these cells on microcontact
printed patterns and compare them to our L929 and C3H fibroblasts from Chapter 2. Although
preliminary, one initial difference I have noticed between C3H and the new MSC line is that the
MSCs seem to have a much brighter mCherry response when co-cultured with GFP senders vs
from surface dried GFP while the C3H have a more similar mCherry response between the two
activation mechanisms. The underlying factor would have to be further explored. There are many
possible explanations such as differences in force required for SynNotch activation in MSCs or
differences in optimal substrate material for MSC activation. Nonetheless, the preliminary results
are promising for the use of MSCs in our goal to achieve a muscle-tendon-bone tissue construct.
Next steps towards this goal would be to create myogenic, osteogenic, and tenocyte
SynNotch lines. The easiest and most familiar first step would be to re-create the myogenic
SynNotch system in these MSCs. I have started on this process, but the results are preliminary and
in progress.
After this, the MSC system will be used to create anti-mCherry synNotch that controls
activation of Runx2 for bone differentiation as well as anti-streptavidin synNotch that activates
Scx for tenocyte differentiation. It is crucial to test these pathways separately before combining
them into one cell line that is capable of three orthogonal differentiation paths. The spatial
transdifferentiation and maturity of each of these lineages must be validated. Additionally, the
patterning approach would have to be further developed to spatially control patterning of three
ligands. Additionally, to achieve a more accurate tissue construct, the substrates would have to be

104
fine-tuned per cell lineage. For example, a scaffold with different soft-to-hard stiffnesses will be
needed to recapitulate the proper mechanical environment for muscle, tendon, and bone. These
multiple areas of advancements, both in cells and material engineering, will be developed in
parallel with the target goal in mind.

105
Material and Methods
Genetic Constructs Design:
GFP sender plasmid: pHR_EGFPligand (Addgene #79129)
GFP responsive synNotch receptor: pHR_SFFV_myc-LaG17_synNotch_TetRVP64 (Addgene
plasmid# 79128) and
Anti-GFP mCherry transgene: pHR_TRE_mCherry
All constructs were cloned via In-Fusion HD Cloning (Takara Bio).
Lentivirus Production:
Lentivirus was produced by cotransfecting pHR cloned plasmids with vectors encoding packaging
proteins (psPAX2, pVSVG) using Lipofectamine LTX (ThermoFisher) into 70-80% confluent
HEK-293T cells within 6-well plates. Viral supernatants were collected 2 days after transfection,
sterile filtered with 0.45μm PES (Genesee Scientific), and used directly with MSCs.
Cell Culture:
C57BL/6 Mouse Bone Marrow Mesenchymal Stem Cells (Cyagen) were cultured on tissue culture
dishes with MEM-Alpha (ThermoFisher, 12561072) supplemented with MSC qualified fetal
bovine serum (ThermoFisher, 12662029 ) and 100U/mL penicillin/streptomycin (ThermoFisher).
Cell Line Engineering:
For viral transduction, 200-400μL viral supernatant(s) were added to 5-10x104 suspended cells
supplemented with 10μg/mL polybrene (Sigma), then transferred into a 12-well plate for 2-3 days

106
before changing to fresh media. Following transduction, cells were sorted for the via fluorescenceactivated cell sorting on a FACS ARIA II (Beckton-Dickinson) by basal expression of the
transgenes. A bulk-sorted polyclonal population of engineered cells were used for experiments
shown above.
Plate-drying of GFP:
5uL droplets of GFP at 200μg/mL in sterile DI water were deposited in 96-well plates. Plates
were left to dry in the biosafety cabinet overnight, protected from light and then incubated with
MEM-alpha with 15% FBS for 30 minutes prior to cell seeding. Anti-GFP mCherry reporter
MSCs were seeded at 1x105 cells per 96-well, cultured, and imaged.
Co-culture of GFP sender and anti-GFP mCherry reporter:
1:1 ratio of MSC GFP senders and anti-GFP mCherry reporters were seeded into 96-wells for a
total of 5 x 104 of each cell type. Cells were cultured and imaged.

107
Chapter 4 - Limitations and future work, and concluding remarks
Limitations and Future Work
With any new advancement in science, it is crucial to acknowledge the challenges and
limitations of the technology. Here, describe the limitations of our system followed by potential
solutions to the limitation.
First limitation of our tissue engineering approach is that the spatial resolution can be
affected by cell motility. Cells are not restricted in their movement to where the pattern is applied,
meaning they can be seeded onto the protein pattern, activate synNotch, and travel off–pattern.
This is something we mitigated by simply seeding our cells at 100% confluence, ensuring that
there is less room for the cells to travel off-pattern. While this approach was effective for our
purposes, it may not be suitable for cases where cells need to be seeded at lower densities.
However, this phenomenon was seen to varying degrees depending on the cell type and gene we
induced. For example, we noticed that on mCherry patterns that activated ETV2 in C3H cells, the
cells tended to stay on pattern, even at lower densities. This was not the case with GFP activating
myoD in C3H cells, highlighting potential considerations of the activating ligand and downstream
transgene of choice for future experiments. To address this limitation, several strategies can be
considered. One solution would be enhancing cell adhesion to the patterned areas using additional
surface coatings or adhesion molecules, such as additional fibronectin, could help reduce cell
motility.
Another limitation of our approach was using lentivirus and FACS for engineering our
cells. This approach worked for our C3H cells, but did not result in 100% expression in all cells.
Also, in each given cell, the gene was expressed at a different level of expression. On top of this,

108
we observed transgene loss over time. We also created a clonal population of synNotch cells using
Lentivirus, but we did not see many benefits in solving the aforementioned issues. One way to
possibly overcome these limitations would be to use CRISPR to deliver the genes into a safe harbor
locus. This being said, we have not tried this because for our purposes and proof of concept, it was
not necessary to have a higher expressing population. One thing to also take into consideration is
that too much gene expression is harmful for the system and will result in more leaky expression
off-pattern.
While on the topic of lentiviral infection; we tried this approach with hiPSC cells but failed.
The cells did not survive lentiviral transduction and experiments are underway to test non-viral
methods such as PiggyBac.
An area of conversation that has come up frequently with our approach is its use in vivo.
We have not conducted any experiments in vivo, but it is definitely an area worth exploring in the
future. A few things to keep in mind are transgene loss and safety precautions. If these cells are to
be transplanted into an animal/human, the modified cells must both be able to express the
transgenes long enough to solve/perform the task they are designed to while also not being
carcinogenic to the host. Alongside this, immune rejection is another concern for in vivo studies
and all of these need to be addressed before experiments can be conducted. To advance towards
clinical applications, preclinical studies in animal models should be conducted to evaluate the
long-term safety, efficacy, and immune response to the transplanted cells. Developing strategies
to enhance immune compatibility, such as using autologous cells or gene editing to create universal
donor cells that evade immune detection, will be essential. Incorporating safety features such as
inducible suicide genes or kill switches that can be activated if the cells begin to proliferate
uncontrollably can further ensure the safety of this technology.

109
While our SynNotch-based tissue engineering system shows great promise, several
challenges must be addressed to optimize its functionality and safety. By pursuing these potential
solutions and continuing to refine our approach, we can advance towards the goal of developing
clinically relevant tissue engineering solutions that better recapitulate the architecture and function
of native tissues to ultimately impact regenerative medicine and disease modeling.
The field of tissue engineering has been one with many creative advancements. Medical
problems that previously had no hope now have cures and direction thanks to the combined effort
of the global scientific community. This being said, there will alway be new challenges to
overcome. The future of synthetic biology and tissue engineering will be driven by
interdisciplinary collaboration. Combining expertise from bioengineering, materials science, and
computer science will be essential to address the complex challenges and advance the potential of
these technologies. Communication has become increasingly easier, making collaborations
between labs with complementary skill sets able to combine their expertise in a synergistic manner.
It was in this way that our project was successful, combining the device fabrication and
micropatterning expertise from the Laboratory for Living Systems (Dr. McCain’s lab) with
synthetic biology experience from the Tissue Development Engineering lab (Dr. Morsut’s lab).
It is obvious that the rapid advancements of software and A.I. technology is going to
exponentially boost scientific discoveries in the years to come. This human and machine
collaboration is one that will continue to change the trajectory of humanity. Of course, it's not
without its caveats. The safety of any technology must be at the forefront of its research. As the
field progresses, it is imperative to develop robust ethical and regulatory frameworks. Transparent
communication and inclusive discussions with both the policymakers and the public are needed in
order to ensure safety. That being said, I am extremely excited to see how all the recent

110
technological advances will aid in the developments of tissue engineering in the context of organ
(re)generation and in-vivo therapeutic applications.
It is truly an exhilarating time to be a scientist.

111
Concluding Remarks
One thing that drew me to the labs of Dr. Megan McCain and Dr. Leonardo Morsut was
the “sci-fi” feel of their labs. Device fabrication that facilitates the enhanced culturing of in-vitro
tissues and synthetic biology focused projects that push the boundaries of life were areas that
immediately captured my interest. Science has always been my favorite subject and one of my
favorite cartoons was “Dexter’s Laboratory”, a show where a kid-scientist created impossible
inventions. Discovery and building through experimentation were things that inherently fascinated
me. Although I have not created nearly as many inventions/advancements as Dexter has, I am
happy with how my PhD project turned out and the results of my dedication. I may be biased, but
our project that resulted in spatially controlled patterns of tissue transdifferentiation is still mindblowing to me. It’s a project I’m always happy to explain to someone who is interested in what I
do in the lab and it never fails to amaze people who are unfamiliar with the recent biological
technologies.
Seeing first-hand how our project has grown over the past 5 years has been an experience
that has, and will continue to have, a profound impact on my life. The time management, planning,
execution, presentation, and team-work skills I’ve learned during this PhD are invaluable to me
and I am honored and privileged to have had this opportunity. Experiencing this journey alongside
so many intelligent and hardworking people has further motivated me to achieve increasingly
higher goals in life.
Thank you,
Mher Garibyan

112
References:
1. Tissue Engineering | Embryo Project Encyclopedia. https://embryo.asu.edu/pages/tissueengineering.
2. Harrison, R. G. Observations on the living developing nerve fiber. Proc. Soc. Exp. Biol.
Med. 4, 140–143 (1906).
3. Carrel, A. & Burrows, M. T. CULTIVATION OF TISSUES IN VITRO AND ITS
TECHNIQUE. J. Exp. Med. 13, 387–396 (1911).
4. Carrel, A. ON THE PERMANENT LIFE OF TISSUES OUTSIDE OF THE ORGANISM.
J. Exp. Med. 15, 516–528 (1912).
5. Masters, J. R. HeLa cells 50 years on: the good, the bad and the ugly. Nat. Rev. Cancer 2,
315–319 (2002).
6. Rheinwald, J. G. & Green, H. Epidermal growth factor and the multiplication of cultured
human epidermal keratinocytes. Nature 265, 421–424 (1977).
7. Vacanti, J. P. et al. Selective cell transplantation using bioabsorbable artificial polymers as
matrices. J. Pediatr. Surg. 23, 3–9 (1988).
8. Achilli, T.-M., Meyer, J. & Morgan, J. R. Advances in the formation, use and
understanding of multi-cellular spheroids. Expert Opin. Biol. Ther. 12, 1347–1360 (2012).
9. Embryonic Stem Cell Lines Derived from Human Blastocysts | Science.
https://www.science.org/doi/abs/10.1126/science.282.5391.1145.
10. A brief history of organoids | American Journal of Physiology-Cell Physiology.
https://journals.physiology.org/doi/full/10.1152/ajpcell.00120.2020.
11. Hofer, M. & Lutolf, M. P. Engineering organoids. Nat. Rev. Mater. 6, 402–420 (2021).
12. Templeton, A., Morris, J. K. & Parslow, W. Factors that affect outcome of in-vitro
fertilisation treatment. The Lancet 348, 1402–1406 (1996).
13. Immunosuppressive therapy mitigates immunological rejection of human embryonic stem
cell xenografts | PNAS. https://www.pnas.org/doi/abs/10.1073/pnas.0805802105.
14. Takahashi, K. et al. Induction of Pluripotent Stem Cells from Adult Human Fibroblasts by
Defined Factors. Cell 131, 861–872 (2007).
15. Malik, N. & Rao, M. S. A Review of the Methods for Human iPSC Derivation. in
Pluripotent Stem Cells: Methods and Protocols (eds. Lakshmipathy, U. & Vemuri, M. C.)
23–33 (Humana Press, Totowa, NJ, 2013). doi:10.1007/978-1-62703-348-0_3.

113
16. Accept or Reject: The Role of Immune Tolerance in the Development of Stem Cell
Therapies and Possible Future Approaches - Richard Haworth, Michaela Sharpe, 2021.
https://journals.sagepub.com/doi/full/10.1177/0192623320918241.
17. Gross, C. G. Galen and the Squealing Pig. The Neuroscientist 4, 216–221 (1998).
18. Tan, S. Y. & Merchant, J. Frederick Banting (1891–1941): Discoverer of insulin. Singapore
Med. J. 58, 2–3 (2017).
19. Bandyopadhyay, A. S., Garon, J., Seib, K. & Orenstein, W. A. Polio Vaccination: Past,
Present and Future. Future Microbiol. 10, 791–808 (2015).
20. Sabin, A. B. & Boulger, L. R. History of Sabin attenuated poliovirus oral live vaccine
strains. J. Biol. Stand. 1, 115–118 (1973).
21. Salk, J. E., Bennett, B. L., Lewis, L. J., Ward, E. N. & Youngner, J. S. STUDIES IN
HUMAN SUBJECTS ON ACTIVE IMMUNIZATION AGAINST POLIOMYELITIS: 1.
A PRELIMINARY REPORT OF EXPERIMENTS IN PROGRESS. J. Am. Med. Assoc.
151, 1081–1098 (1953).
22. Sato, T. et al. Single Lgr5 stem cells build crypt-villus structures in vitro without a
mesenchymal niche. Nature 459, 262–265 (2009).
23. Organoids as Model for Infectious Diseases: Culture of Human and Murine Stomach
Organoids and Microinjection of Helicobacter Pylori.
https://app.jove.com/t/53359/organoids-as-model-for-infectious-diseases-culture-humanmurine.
24. Prior, N., Inacio, P. & Huch, M. Liver organoids: from basic research to therapeutic
applications. Gut 68, 2228–2237 (2019).
25. Koo, B., Choi, B., Park, H. & Yoon, K.-J. Past, Present, and Future of Brain Organoid
Technology. Mol. Cells 42, 617–627 (2019).
26. Lancaster, M. A. et al. Cerebral organoids model human brain development and
microcephaly. Nature 501, 373–379 (2013).
27. Dekkers, J. F. et al. A functional CFTR assay using primary cystic fibrosis intestinal
organoids. Nat. Med. 19, 939–945 (2013).
28. Neal, J. T. et al. Organoid Modeling of the Tumor Immune Microenvironment. Cell 175,
1972-1988.e16 (2018).
29. Human blood vessel organoids as a model of diabetic vasculopathy | Nature.
https://www.nature.com/articles/s41586-018-0858-8.
30. Hajam, M. I. & Khan, M. M. Microfluidics: a concise review of the history, principles,
design, applications, and future outlook. Biomater. Sci. 12, 218–251 (2024).

114
31. Toh, A. G. G., Wang, Z. P., Yang, C. & Nguyen, N.-T. Engineering microfluidic
concentration gradient generators for biological applications. Microfluid. Nanofluidics 16,
1–18 (2014).
32. Huh, D. et al. Reconstituting Organ-Level Lung Functions on a Chip. Science 328, 1662–
1668 (2010).
33. Sung, J. H. et al. Recent Advances in Body-on-a-Chip Systems. Anal. Chem. 91, 330–351
(2019).
34. Mittal, R. et al. Organ-on-chip models: Implications in drug discovery and clinical
applications. J. Cell. Physiol. 234, 8352–8380 (2019).
35. Polstein, L. R., Juhas, M., Hanna, G., Bursac, N. & Gersbach, C. A. An Engineered
Optogenetic Switch for Spatiotemporal Control of Gene Expression, Cell Differentiation,
and Tissue Morphogenesis. ACS Synth. Biol. 6, 2003–2013 (2017).
36. Skylar-Scott, M. A. et al. Orthogonally induced differentiation of stem cells for the
programmatic patterning of vascularized organoids and bioprinted tissues. Nat. Biomed.
Eng. 6, 449–462 (2022).
37. Prahl, L. S., Porter, C. M., Liu, J., Viola, J. M. & Hughes, A. J. Independent control over
cell patterning and adhesion on hydrogel substrates for tissue interface mechanobiology.
iScience 26, (2023).
38. Lin, R.-Z., Ho, C.-T., Liu, C.-H. & Chang, H.-Y. Dielectrophoresis based-cell patterning
for tissue engineering. Biotechnol. J. 1, 949–957 (2006).
39. Direct Human Cartilage Repair Using Three-Dimensional Bioprinting Technology | Tissue
Engineering Part A. https://www.liebertpub.com/doi/full/10.1089/ten.tea.2011.0543.
40. Kumar, A. & Whitesides, G. M. Features of gold having micrometer to centimeter
dimensions can be formed through a combination of stamping with an elastomeric stamp
and an alkanethiol ‘“ink”’ followed by chemical etching. Appl. Phys. Lett. 63, 2002–2004
(1993).
41. Bernard, A. et al. Printing Patterns of Proteins. Langmuir 14, 2225–2229 (1998).
42. Théry, M. et al. The extracellular matrix guides the orientation of the cell division axis. Nat.
Cell Biol. 7, 947–953 (2005).
43. Micropatterned mammalian cells exhibit phenotype-specific left-right asymmetry | PNAS.
https://www.pnas.org/doi/abs/10.1073/pnas.1103834108.
44. Jagota, A., Sharp, K. G. & Kristunas, D. F. Stamp Deformation During Microcontact
Printing. MRS Online Proc. Libr. OPL 728, S8.41 (2002).

115
45. Perl, A., Reinhoudt, D. N. & Huskens, J. Microcontact Printing: Limitations and
Achievements. Adv. Mater. 21, 2257–2268 (2009).
46. Bou Chakra, E., Hannes, B. & Cabrera, M. A new instrument for automated microcontact
printing with applications to mems and biochips fabrication. in CFM 2007 - 18ème Congrès
Français de Mécanique (ed. Mécanique, A. F. de) (AFM, Maison de la Mécanique, 39/41
rue Louis Blanc - 92400 Courbevoie, Grenoble, France, 2007).
47. Construction of Biologically Functional Bacterial Plasmids In Vitro | PNAS.
https://www.pnas.org/doi/abs/10.1073/pnas.70.11.3240.
48. Elowitz, M. B. & Leibler, S. A synthetic oscillatory network of transcriptional regulators.
Nature 403, 335–338 (2000).
49. Gardner, T. S., Cantor, C. R. & Collins, J. J. Construction of a genetic toggle switch in
Escherichia coli. Nature 403, 339–342 (2000).
50. Kramer, B. P. et al. An engineered epigenetic transgene switch in mammalian cells. Nat.
Biotechnol. 22, 867–870 (2004).
51. Jinek, M. et al. A Programmable Dual-RNA–Guided DNA Endonuclease in Adaptive
Bacterial Immunity. Science 337, 816–821 (2012).
52. Komor, A. C., Badran, A. H. & Liu, D. R. CRISPR-Based Technologies for the
Manipulation of Eukaryotic Genomes. Cell 168, 20–36 (2017).
53. Genome Engineering With Zinc-Finger Nucleases | Genetics | Oxford Academic.
https://academic.oup.com/genetics/article/188/4/773/6063261.
54. Kampmann, M. CRISPRi and CRISPRa Screens in Mammalian Cells for Precision Biology
and Medicine. ACS Chem. Biol. 13, 406–416 (2018).
55. Nielsen, A. A. & Voigt, C. A. Multi‐input CRISPR/Cas genetic circuits that interface host
regulatory networks. Mol. Syst. Biol. 10, 763 (2014).
56. Gross, G., Waks, T. & Eshhar, Z. Expression of immunoglobulin-T-cell receptor chimeric
molecules as functional receptors with antibody-type specificity. Proc. Natl. Acad. Sci. 86,
10024–10028 (1989).
57. Kowolik, C. M. et al. CD28 Costimulation Provided through a CD19-Specific Chimeric
Antigen Receptor Enhances In vivo Persistence and Antitumor Efficacy of Adoptively
Transferred T Cells. Cancer Res. 66, 10995–11004 (2006).
58. Bach, P. B., Giralt, S. A. & Saltz, L. B. FDA Approval of Tisagenlecleucel: Promise and
Complexities of a $475 000 Cancer Drug. JAMA 318, 1861–1862 (2017).

116
59. Nagel, G. et al. Light Activation of Channelrhodopsin-2 in Excitable Cells of
Caenorhabditis elegans Triggers Rapid Behavioral Responses. Curr. Biol. 15, 2279–2284
(2005).
60. Krashes, M. J. et al. Rapid, reversible activation of AgRP neurons drives feeding behavior
in mice. J. Clin. Invest. 121, 1424–1428 (2011).
61. Airan, R. D., Thompson, K. R., Fenno, L. E., Bernstein, H. & Deisseroth, K. Temporally
precise in vivo control of intracellular signalling. Nature 458, 1025–1029 (2009).
62. Modular Extracellular Sensor Architecture for Engineering Mammalian Cell-based Devices
| ACS Synthetic Biology. https://pubs.acs.org/doi/full/10.1021/sb400128g.
63. Nihongaki, Y., Kawano, F., Nakajima, T. & Sato, M. Photoactivatable CRISPR-Cas9 for
optogenetic genome editing. Nat. Biotechnol. 33, 755–760 (2015).
64. Roybal, K. T. et al. Engineering T Cells with Customized Therapeutic Response Programs
Using Synthetic Notch Receptors. Cell 167, 419-432.e16 (2016).
65. Manhas, J., Edelstein, H. I., Leonard, J. N. & Morsut, L. The evolution of synthetic receptor
systems. Nat. Chem. Biol. 18, 244–255 (2022).
66. Morsut, L. et al. Engineering Customized Cell Sensing and Response Behaviors Using
Synthetic Notch Receptors. Cell 164, 780–791 (2016).
67. Lim, H. Y. G. & Plachta, N. Cytoskeletal control of early mammalian development. Nat.
Rev. Mol. Cell Biol. 22, 548–562 (2021).
68. Takebayashi-Suzuki, K. & Suzuki, A. Intracellular Communication among Morphogen
Signaling Pathways during Vertebrate Body Plan Formation. Genes 11, 341 (2020).
69. Bocci, F., Onuchic, J. N. & Jolly, M. K. Understanding the Principles of Pattern Formation
Driven by Notch Signaling by Integrating Experiments and Theoretical Models. Front.
Physiol. 11, 929 (2020).
70. Boareto, M. Patterning via local cell-cell interactions in developing systems. Dev. Biol. 460,
77–85 (2020).
71. Lau, S., Slane, D., Herud, O., Kong, J. & Jürgens, G. Early Embryogenesis in Flowering
Plants: Setting Up the Basic Body Pattern. Annu. Rev. Plant Biol. 63, 483–506 (2012).
72. McGlinn, E. & Tabin, C. J. Mechanistic insight into how Shh patterns the vertebrate limb.
Curr. Opin. Genet. Dev. 16, 426–432 (2006).
73. Levin, M. Bioelectric signaling: Reprogrammable circuits underlying embryogenesis,
regeneration, and cancer. Cell 184, 1971–1989 (2021).

117
74. Linask, K. K., Manisastry, S. & Han, M. Cross Talk between Cell–Cell and Cell–Matrix
Adhesion Signaling Pathways during Heart Organogenesis: Implications for Cardiac Birth
Defects. Microsc. Microanal. 11, 200–208 (2005).
75. Perris, R. & Perissinotto, D. Role of the extracellular matrix during neural crest cell
migration. Mech. Dev. 95, 3–21 (2000).
76. Janmey, P. A., Wells, R. G., Assoian, R. K. & McCulloch, C. A. From tissue mechanics to
transcription factors. Differentiation 86, 112–120 (2013).
77. Organoids by design | Science. https://www.science.org/doi/full/10.1126/science.aaw7567.
78. Brassard, J. A. & Lutolf, M. P. Engineering Stem Cell Self-organization to Build Better
Organoids. Cell Stem Cell 24, 860–876 (2019).
79. Shao, Y. & Fu, J. Engineering multiscale structural orders for high-fidelity embryoids and
organoids. Cell Stem Cell 29, 722–743 (2022).
80. Wolf, K. J., Weiss, J. D., Uzel, S. G. M., Skylar-Scott, M. A. & Lewis, J. A.
Biomanufacturing human tissues via organ building blocks. Cell Stem Cell 29, 667–677
(2022).
81. Ingber, D. E. Human organs-on-chips for disease modelling, drug development and
personalized medicine. Nat. Rev. Genet. 23, 467–491 (2022).
82. Underhill, G. H., Galie, P., Chen, C. S. & Bhatia, S. N. Bioengineering methods for
analysis of cells in vitro. Annu. Rev. Cell Dev. Biol. 28, 385–410 (2012).
83. Rao, K. M., Choi, S. M. & Han, S. S. A review on directional muscle cell growth in
scaffolding biomaterials with aligned porous structures for cultivated meat production.
Food Res. Int. 168, 112755 (2023).
84. Rizwan, M. et al. Viscoelastic Notch Signaling Hydrogel Induces Liver Bile Duct Organoid
Growth and Morphogenesis. Adv. Healthc. Mater. 11, 2200880 (2022).
85. Rexius-Hall, M. L., Ariyasinghe, N. R. & McCain, M. L. Engineering Shape-Controlled
Microtissues on Compliant Hydrogels with Tunable Rigidity and Extracellular Matrix
Ligands. in Programmed Morphogenesis: Methods and Protocols (eds. Ebrahimkhani, M.
R. & Hislop, J.) 57–72 (Springer US, New York, NY, 2021). doi:10.1007/978-1-0716-
1174-6_5.
86. Photopatterned biomolecule immobilization to guide three-dimensional cell fate in natural
protein-based hydrogels | PNAS. https://www.pnas.org/doi/abs/10.1073/pnas.2014194118.
87. Toda, S., Blauch, L. R., Tang, S. K. Y., Morsut, L. & Lim, W. A. Programming selforganizing multicellular structures with synthetic cell-cell signaling. Science 361, 156–162
(2018).

118
88. Huang, X. et al. DNA scaffolds enable efficient and tunable functionalization of
biomaterials for immune cell modulation. Nat. Nanotechnol. 16, 214–223 (2021).
89. Lee, J. C. et al. Instructional materials that control cellular activity through synthetic Notch
receptors. Biomaterials 297, 122099 (2023).
90. Xi, Z.-Y., Xu, Y.-Y., Zhu, L.-P., Wang, Y. & Zhu, B.-K. A facile method of surface
modification for hydrophobic polymer membranes based on the adhesive behavior of
poly(DOPA) and poly(dopamine). J. Membr. Sci. 327, 244–253 (2009).
91. Kuddannaya, S. et al. Surface Chemical Modification of Poly(dimethylsiloxane) for the
Enhanced Adhesion and Proliferation of Mesenchymal Stem Cells. ACS Appl. Mater.
Interfaces 5, 9777–9784 (2013).
92. Wu, D. et al. Ligand Mobility-Mediated Cell Adhesion and Spreading. ACS Appl. Mater.
Interfaces 14, 12976–12983 (2022).
93. Chen, C. S., Mrksich, M., Huang, S., Whitesides, G. M. & Ingber, D. E. Geometric Control
of Cell Life and Death. Science 276, 1425–1428 (1997).
94. Geisse, N. A., Sheehy, S. P. & Parker, K. K. Control of myocyte remodeling in vitro with
engineered substrates. Vitro Cell. Dev. Biol. - Anim. 45, 343–350 (2009).
95. Qin, D., Xia, Y. & Whitesides, G. M. Soft lithography for micro- and nanoscale patterning.
Nat. Protoc. 5, 491–502 (2010).
96. Santoso, J. W. et al. Engineering skeletal muscle tissues with advanced maturity improves
synapse formation with human induced pluripotent stem cell-derived motor neurons. APL
Bioeng. 5, 036101 (2021).
97. Bettadapur, A. et al. Prolonged Culture of Aligned Skeletal Myotubes on Micromolded
Gelatin Hydrogels. Sci. Rep. 6, 28855 (2016).
98. Direct Reprogramming of Human Dermal Fibroblasts Into Endothelial Cells Using
ER71/ETV2 | Circulation Research.
https://www.ahajournals.org/doi/full/10.1161/CIRCRESAHA.116.309833.
99. Morita, R. et al. ETS transcription factor ETV2 directly converts human fibroblasts into
functional endothelial cells. Proc. Natl. Acad. Sci. 112, 160–165 (2015).
100. Lee, S. H. et al. Capillary Based Patterning of Cellular Communities in Laterally Open
Channels. Anal. Chem. 82, 2900–2906 (2010).
101. Sherman, B. T. et al. DAVID: a web server for functional enrichment analysis and
functional annotation of gene lists (2021 update). Nucleic Acids Res. 50, W216–W221
(2022).

119
102. Huang, D. W., Sherman, B. T. & Lempicki, R. A. Systematic and integrative analysis of
large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44–57 (2009).
103. Choe, J. H. et al. SynNotch-CAR T cells overcome challenges of specificity, heterogeneity,
and persistence in treating glioblastoma. Sci. Transl. Med. 13, eabe7378 (2021).
104. Hyrenius-Wittsten, A. et al. SynNotch CAR circuits enhance solid tumor recognition and
promote persistent antitumor activity in mouse models. Sci. Transl. Med. 13, eabd8836
(2021).
105. Moghimi, B. et al. Preclinical assessment of the efficacy and specificity of GD2-B7H3
SynNotch CAR-T in metastatic neuroblastoma. Nat. Commun. 12, 511 (2021).
106. Allen, G. M. et al. Synthetic cytokine circuits that drive T cells into immune-excluded
tumors. Science 378, eaba1624 (2022).
107. Ruffo, E. et al. Post-translational covalent assembly of CAR and synNotch receptors for
programmable antigen targeting. Nat. Commun. 14, 2463 (2023).
108. Zhang, S. et al. Monitoring of cell-cell communication and contact history in mammals.
Science 378, eabo5503 (2022).
109. Malaguti, M., Lebek, T., Blin, G. & Lowell, S. Enabling neighbour labelling: using
synthetic biology to explore how cells influence their neighbours. Dev. Camb. Engl. 151,
dev201955 (2024).
110. Purnick, P. E. M. & Weiss, R. The second wave of synthetic biology: from modules to
systems. Nat. Rev. Mol. Cell Biol. 10, 410–422 (2009).
111. Meng, F. & Ellis, T. The second decade of synthetic biology: 2010–2020. Nat. Commun.
11, 5174 (2020).
112. Firas, J., Liu, X., Lim, S. M. & Polo, J. M. Transcription factor-mediated reprogramming:
epigenetics and therapeutic potential. Immunol. Cell Biol. 93, 284–289 (2015).
113. Takahashi, K. & Yamanaka, S. A decade of transcription factor-mediated reprogramming
to pluripotency. Nat. Rev. Mol. Cell Biol. 17, 183–193 (2016).
114. Ng, A. H. M. et al. A comprehensive library of human transcription factors for cell fate
engineering. Nat. Biotechnol. 39, 510 (2021).
115. Joung, J. et al. A transcription factor atlas of directed differentiation. Cell 186, 209-229.e26
(2023).
116. Johnson, H. E. & Toettcher, J. E. Illuminating developmental biology with cellular
optogenetics. Curr. Opin. Biotechnol. 52, 42–48 (2018).

120
117. Johnson, H. E., Djabrayan, N. J. V., Shvartsman, S. Y. & Toettcher, J. E. Optogenetic
Rescue of a Patterning Mutant. Curr. Biol. 30, 3414-3424.e3 (2020).
118. Müller, K., Engesser, R., Timmer, J., Zurbriggen, M. D. & Weber, W. Orthogonal
Optogenetic Triple-Gene Control in Mammalian Cells. ACS Synth. Biol. 3, 796–801 (2014).
119. Gordon, W. R. et al. Mechanical Allostery: Evidence for a Force Requirement in the
Proteolytic Activation of Notch. Dev. Cell 33, 729–736 (2015).
120. Gordon, W. R. et al. Structural basis for autoinhibition of Notch. Nat. Struct. Mol. Biol. 14,
295–300 (2007).
121. Khamaisi, B., Luca, V. C., Blacklow, S. C. & Sprinzak, D. Functional Comparison between
Endogenous and Synthetic Notch Systems. ACS Synth. Biol. 11, 3343–3353 (2022).
122. Tissue-embedded stretchable nanoelectronics reveal endothelial cell–mediated electrical
maturation of human 3D cardiac microtissues | Science Advances.
https://www.science.org/doi/full/10.1126/sciadv.ade8513.
123. Giacomelli, E. et al. Human-iPSC-Derived Cardiac Stromal Cells Enhance Maturation in
3D Cardiac Microtissues and Reveal Non-cardiomyocyte Contributions to Heart Disease.
Cell Stem Cell 26, 862-879.e11 (2020).
124. Recapitulating human cardio-pulmonary co-development using simultaneous multilineage
differentiation of pluripotent stem cells | eLife. https://elifesciences.org/articles/67872.
125. Daly, A. C., Prendergast, M. E., Hughes, A. J. & Burdick, J. A. Bioprinting for the
Biologist. Cell 184, 18–32 (2021).
126. Hoffman, T. et al. Tissue Engineering: Synthetic Biology and Tissue Engineering: Toward
Fabrication of Complex and Smart Cellular Constructs (Adv. Funct. Mater. 26/2020). Adv.
Funct. Mater. 30, 2070169 (2020).
127. Sasai, Y. Next-Generation Regenerative Medicine: Organogenesis from Stem Cells in 3D
Culture. Cell Stem Cell 12, 520–530 (2013).
128. The hope and the hype of organoid research | Development | The Company of Biologists.
https://journals.biologists.com/dev/article/144/6/938/48390/The-hope-and-the-hype-oforganoid-research.
129. Takebe, T., Zhang, B. & Radisic, M. Synergistic Engineering: Organoids Meet Organs-ona-Chip. Cell Stem Cell 21, 297–300 (2017).
130. Aydin, O. et al. Principles for the design of multicellular engineered living systems. APL
Bioeng. 6, 010903 (2022).
131. Kamm, R. D. et al. Perspective: The promise of multi-cellular engineered living systems.
APL Bioeng. 2, 040901 (2018).

121
132. Besser, R. R. et al. Enzymatically crosslinked gelatin–laminin hydrogels for applications in
neuromuscular tissue engineering. Biomater. Sci. 8, 591–606 (2020).
133. Feinberg, A. W. et al. Controlling the contractile strength of engineered cardiac muscle by
hierarchal tissue architecture. Biomaterials 33, 5732–5741 (2012).
134. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinforma. Oxf. Engl. 29, 15–
21 (2013).
135. Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for
assigning sequence reads to genomic features. Bioinforma. Oxf. Engl. 30, 923–930 (2014).
136. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion
for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
137. Wu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data.
Innov. Camb. Mass 2, 100141 (2021).
138. Fernandes, G., Gonçalves, D. A. G. & Conti, P. Musculoskeletal Disorders. Dent. Clin. 62,
553–564 (2018).
139. Firestein, G. S. Evolving concepts of rheumatoid arthritis. Nature 423, 356–361 (2003).
140. Pittenger, M. F. et al. Multilineage Potential of Adult Human Mesenchymal Stem Cells.
Science 284, 143–147 (1999).
141. Gonçalves, M. A. et al. Genetic Complementation of Human Muscle Cells via Directed
Stem Cell Fusion. Mol. Ther. 16, 741–748 (2008).
142. Alberton, P. et al. Conversion of human bone marrow-derived mesenchymal stem cells into
tendon progenitor cells by ectopic expression of scleraxis. Stem Cells Dev. 21, 846–858
(2012).
143. Chen, X. et al. Scleraxis-overexpressed human embryonic stem cell-derived mesenchymal
stem cells for tendon tissue engineering with knitted silk-collagen scaffold. Tissue Eng.
Part A 20, 1583–1592 (2014).
144. Komori, T. Regulation of Osteoblast Differentiation by Runx2. in Osteoimmunology (ed.
Choi, Y.) 43–49 (Springer US, Boston, MA, 2010). doi:10.1007/978-1-4419-1050-9_5.
145. Thiagarajan, L., Abu‐Awwad, H. A. M. & Dixon, J. E. Osteogenic Programming of Human
Mesenchymal Stem Cells with Highly Efficient Intracellular Delivery of RUNX2. Stem
Cells Transl. Med. 6, 2146–2159 (2017).

Abstract (if available)

Abstract The field of tissue engineering has been heavily focused on recapturing the structure and function of native tissue in vitro with the goal of creating reproducible platforms for drug screening, disease modeling, and regenerative medicine. However, a system that gives researchers the precision and control to replicate the complex cellular architecture of native tissues does not exist and is a major challenge in the field of tissue engineering. Current in vitro tissue engineering methods, such as organoids or organ-on-a-chip systems, lack reproducibility and/or microscale precision. To overcome these challenges, we propose a system that offers microscale control over spatial gene induction and cell transdifferentiation. This system combines Synthetic Notch (synNotch) engineered cells with microcontact printed synNotch activating ligands. SynNotch receptors are exogenous cell membrane receptors that give researchers the ability to control modular gene expression upon cell-contact with a ligand of choice. Microcontact printing offers microscale precision over protein patterning on a material surface. Combining the two, we developed a tissue engineering system where spatially controlled proteins via microcontact printing activate transdifferentiation pathways in cells via synNotch, resulting in spatially controlled gene expression and cell transdifferentiation. We created cells that co-transdifferentiate fibroblasts into either muscle or endothelial precursor cells, depending on the ligand that they are in contact with. Our project aims to build a co-transdifferentiated tissue consisting of myotubes, the functional unit of skeletal muscle, and endothelial cells, the main cell type found in vasculature. Doing so expands the synthetic biology toolkit by offering an approach to transdifferentiate multicellular tissues with microscale precision. This technology advances the field closer to replicating the complex cellular architecture found in vivo and benefits the areas of drug screening, disease modeling, and regenerative medicine.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Engineering multicellular oscillators based on SynNotch-mediated intercellular dialogue

PDF

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

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

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

PDF

Single-cell analysis with high frequency ultrasound

PDF

Establishing the feasibility of spatial genomics using formalin-fixed human post-mortem brain tissue

PDF

Micelle nanoparticles for the detection and treatment of atherosclerosis

PDF

Mechanistic investigation of pro-angiogenic signaling and endothelial sprouting mediated by FGF and VEGF

PDF

Extracellular matrix regulation of mitochondrial function in engineered cardiac myocytes

PDF

Exploring the role of vegfr transcripts during vascular development in the zebrafish embryo

PDF

Proliferation and maturation events in second heart field cells during cardiovascular development activated by the Delta like ligand-4 mediated notch signaling

PDF

Multi-modality intravascular imaging by combined use of ultrasonic and opticial techniques

PDF

Microfluidic cell sorting with a high frequency ultrasound beam

PDF

Adaptive patterning during skin formation: assembly of vasculature and adipose tissue

PDF

Slit/Robo signaling underlies the spatial patterning of spiral ganglion neurons to shape the peripheral auditory circuitry assembly

PDF

Mechanisms of airway multiciliated cell differentiation

PDF

Nr5a2 balances lineage decisions to promote diverse connective tissue fates in the jaw

PDF

Dissecting the role of a non-canonical interleukin-6 cytokine signaling in systemic chronic inflammation and multimorbidity induced by a high-fat diet

Asset Metadata

Creator Garibyan, Mher (author)

Core Title Spatially controlled tissue differentiation using the synthetic receptor SynNotch

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Degree Conferral Date 2024-08

Publication Date 07/12/2024

Defense Date 06/10/2024

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

Tag differentiation,endothelial,microcontact printing,micropatterning,muscle,OAI-PMH Harvest,patterning,spatial control,synbio,SynNotch,synthetic biology,Synthetic Notch,tissue differentiation,tissue engineering

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McCain, Megan (committee chair), Almada, Albert (committee member), Morsut, Leonardo (committee member), Shen, Keyue (committee member)

Creator Email mgaribya@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113997OH4

Unique identifier UC113997OH4

Identifier etd-GaribyanMh-13229.pdf (filename)

Legacy Identifier etd-GaribyanMh-13229

Document Type Dissertation

Format theses (aat)

Rights Garibyan, Mher

Internet Media Type application/pdf

Type texts

Source 20240715-usctheses-batch-1182 (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.

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