University of Southern California Dissertations and Theses

Amelogenin-ameloblastin protein interaction and function in dental enamel formation

(USC Thesis Other)

Amelogenin-ameloblastin protein interaction and function in dental enamel formation

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Amelogenin-Ameloblastin Protein Interaction and
Function in Dental Enamel Formation
By
Rucha Arun Bapat

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
(CRANIOFACIAL BIOLOGY)

May 2020

Copyright 2020 Rucha Arun Bapat
ii

Dedication
I dedicate this thesis to my parents Mr. Arun Rajaram Bapat, and Mrs. Hema Arun Bapat.
Words cannot express my love and gratitude towards them.

iii

Acknowledgements
First, I want to thank my PhD mentor Dr. Janet Moradian-Oldak for providing expert scientific
guidance and academic advice during this project. Along with her scholarly inputs, I want to thank
her for being a strong female role model in my life. I want to thank Dr. Michael Paine for always
being approachable and supportive throughout my journey from a masters to a PhD student. I
would like to thank my PhD thesis committee members Dr. Baruch Frenkel, Dr. Tobias Ulmer,
and Dr. Jian Xu for their advice during this project. I want to acknowledge the enamel research
community, scientists from around the world whom I met at conferences, whose research I read
and studied which made my project possible, and who have inspired me to continue studying this
unique bioceramic enamel.
I am incredibly grateful to my lab colleagues and friends, Dr. Jingtan Su for sharing his knowledge
of biochemistry and molecular biology, and tirelessly answering all my questions; Gayathri
Visakan and Dr. Natalie Kegulian for their keen eye in editing many of my posters and
presentations; and Dr. Kaushik Mukherjee for his guidance with electron microscopy. I would also
like to thank lab alumni Dr. Parichita Mazumder, Dr. Karthik Chandrababu and Dr. Saumya
Prajapati for guiding me during the first year of my PhD.
I want to acknowledge the USC/Norris Cancer Center Cell and Tissue Imaging Core and USC
Core Center of Excellence in Nano Imaging for providing training and access to transmission
electron microscopy, and the Agilent Center for Excellence in Biomolecular Characterization for
performing mass spectrometry. I want to acknowledge Thach-Vu Ho, Janice Bea, Magdalena
Morales, Elsa Miranda, and all the staff, faculty members, and colleagues at the Center for
Craniofacial Molecular Biology of USC for creating a very supportive and scientifically stimulating
work environment.
iv

I want to thank all of my friends and roommates at USC who made Los Angeles feel like a second
home.
Finally, I want to thank my parents from the bottom of my heart for encouraging me to follow my
dreams of pursuing an unconventional research career path after completing dental school. They
have instilled their work ethic in me and have always inspired independent thinking and decision
making. I feel immensely lucky for having them in my life.

v

Table of Contents
Dedication ................................................................................................................................ ii
Acknowledgements .................................................................................................................iii
List of Tables ............................................................................................................................ x
List of Figures .........................................................................................................................xii
List of Schematics ..................................................................................................................xx
Abbreviations ......................................................................................................................... xxi
Abstract ................................................................................................................................. xxii
1. Introduction ....................................................................................................................... 1
1.1 Amelogenesis ............................................................................................................ 1
1.1.1 Pre-secretory and secretory stages ...................................................................... 2
1.1.2 Transition and maturation stages .......................................................................... 5
1.2 Amelogenins .............................................................................................................. 6
1.2.1 Biochemical properties .......................................................................................... 7
1.2.2 Functions .............................................................................................................. 9
1.3 Ameloblastin .............................................................................................................10
1.3.1 Biochemical properties .........................................................................................11
1.3.2 Functions .............................................................................................................12
vi

1.4 Amelogenin-ameloblastin interactions ...................................................................15
1.5 Hypothesis, aims, and scope of the work ...............................................................17
2. Direct evidence of amelogenin-ameloblastin interaction ..............................................19
2.1 Materials and methods .............................................................................................21
2.1.1 Recombinant amelogenin and ameloblastin expression and purification ..............21
2.1.2 Antibodies ............................................................................................................22
2.1.3 Co-immunoprecipitation .......................................................................................23
2.1.4 SDS-PAGE and Western blots .............................................................................24
2.1.5 Porcine enamel matrix protein extraction .............................................................25
2.1.6 Mass Spectrometry ..............................................................................................26
2.2 Results .......................................................................................................................27
2.2.1 Characterization of recombinant amelogenin and ameloblastin ...........................27
2.2.2 Characterization of porcine enamel matrix protein extract ....................................28
2.2.3 Co-immunoprecipitation using anti-Amel antibody column ...................................29
2.2.4 Co-immunoprecipitation using anti-Ambn antibody column ..................................33
2.2.5 Antibody cross-reactivity controls .........................................................................36
2.2.6 Analysis of porcine EMP elution by mass spectrometry .......................................36
2.3 Discussion.................................................................................................................38
3. Investigation of the binding region of Ambn in Amel-Ambn co-assembly ..................42
Part I .....................................................................................................................................42
3.1 Background ...............................................................................................................42
vii

3.2 Materials and methods .............................................................................................44
3.2.1 Expression and purification of recombinant proteins ............................................44
3.2.2 Ambn synthetic peptides ......................................................................................44
3.2.3 Co-immunoprecipitation .......................................................................................45
3.2.4 SDS-PAGE and Western blots .............................................................................46
3.3 Results .......................................................................................................................46
3.3.1 Characterization of AmbnΔ5 and AmbnΔ6 ...........................................................46
3.3.2 Anti-Amel and anti-Ambn M300 antibody cross-reactivity.....................................48
3.3.3 AmbnΔ5 mutant fails to bind to rAmel in co-IP .....................................................49
3.3.4 N-terminal fragment of Ambn peptide encoded by exon 5 binds directly to Amel .50
Part II ....................................................................................................................................52
3.4 Background ...............................................................................................................52
3.5 Materials and methods .............................................................................................54
3.5.1 Recombinant proteins and peptides .....................................................................54
3.5.2 Obtaining rAmel nanospheres ..............................................................................54
3.5.3 Co-assembly of AB2-rAmel and rAmbn-rAmel .....................................................55
3.5.4 Dynamic light scattering .......................................................................................55
3.5.5 Transmission electron microscopy .......................................................................56
3.6 Results .......................................................................................................................57
3.6.1 Buffer control observed by DLS ...........................................................................57
3.6.2 Characterization of rAmel nanospheres ...............................................................58
viii

3.6.3 Ambn peptide encoded by exon 5 (AB2) dis-assembles rAmel nanospheres .......59
3.6.4 Full-length rAmbn co-assembles with rAmel in solution .......................................63
3.7 Discussion.................................................................................................................67
4. Colocalization of Amel-Ambn and Ambn-ameloblast cell membrane ..........................73
4.1 Background ...............................................................................................................73
4.2 Material and methods ...............................................................................................76
4.2.1 Tissue preparation ...............................................................................................76
4.2.2 Immunohistochemical labeling for Amel and Ambn ..............................................77
4.2.3 Quantitative colocalization (Manders’ colocalization coefficient) ..........................78
4.2.4 Alizarin red S staining ..........................................................................................79
4.2.5 AB2 labeling with FITC ........................................................................................79
4.2.6 AB2-FITC binding to ALC.....................................................................................80
4.2.7 In situ cell membrane staining with DiD, co-labeled with Ambn ............................80
4.3 Results .......................................................................................................................81
4.3.1 Amel and full-length Ambn colocalize within ameloblasts and at the secretory face
of ameloblasts ....................................................................................................................81
4.3.2 N-terminal fragments of Ambn colocalize with Amel in the bulk of developing
enamel .. ............................................................................................................................86
4.3.3 Colocalization of N-terminal fragments of Ambn with Amel in maturation stage
enamel .. ............................................................................................................................90
4.3.4 Ambn colocalizes with ameloblast cell membrane in situ .....................................92
4.3.5 Ambn binds to ALC cell membranes via its exon 5 encoded region .....................96
4.4 Discussion.................................................................................................................97
ix

5. Conclusions and future directions ............................................................................... 102
Bibliography .......................................................................................................................... 107

x

List of Tables
Table 1.1. Summary of techniques used to study protein-protein interactions in developing
enamel matrix ...........................................................................................................................15
Table 2.1. Amelogenin and ameloblastin fragments identified in the porcine enamel extracellular
matrix ........................................................................................................................................21
Table 2.2. List of Ambn and Amel antibodies used in co-immunoprecipitation experiments. .....22
Table 2.3. Fragments identified by mass spectrometry from porcine co-IP elution from Amel
column. † marks fragment novel to elution, not present in whole porcine extract. .....................37
Table 2.4. Summary of Western blot bands representing porcine Amel and Ambn fragments
detected from co-IP elution fractions by Western Blot ...............................................................38
Table 3.1. Masses of Ambn synthetic peptides .........................................................................45
Table 3.2. The table lists protein concentrations for each TEM experiment, and the volume of
protein added to the TEM grid ...................................................................................................56
Table 3.3. Summary of DLS and TEM measurements of nanoparticle sizes from proteins or
protein mixtures. .......................................................................................................................70
Table 4.1. Primary antibodies used for immunohistochemical labeling of Amel and Ambn .......77
Table 4.2. Summary of average Manders’ colocalization coefficients (MCC) for all ROIs within a
given area from Figure 4.6, labeled with anti-Amel and anti-Ambn-M300 antibodies. Note- MCC
Values bellow 0.5 indicate lack of colocalization (in grey). ........................................................82
xi

Table 4.3. Summary of average Manders’ colocalization coefficients for all ROIs within a given
area from Figure 4.8, labeled with anti-Amel and anti-Ambn-N18 antibodies. There is no
discernible ‘secretory front’ in transition and maturation stages (N/A in the Table) but Amel-Ambn
colocalization is observed where ameloblasts end and enamel matrix begins. This area is
considered a part of the enamel matrix for the purpose of calculations. ....................................86
xii

List of Figures
Figure 1.1 SEM images showing the rod-interrod structure of mouse (A) and human enamel (B).
Inset of mouse enamel shows individual HAP crystals assembling to form one enamel rod
(Moradian-Oldak, 2012; Nanci, 2012) ........................................................................................ 2
Figure 1.2. TEM images showing stages of amelogenesis: Secretory (S), transition (T), and
maturation (M) stages. Ameloblasts (Am), enamel (E), Tomes’ processes (red arrow). Modified
from Nanci (2012). ..................................................................................................................... 3
Figure 1.3. TEM image showing secretory face of Tomes’ processes (TP), distal end (DE),
proximal side (PS), rod enamel (R), interrod enamel (IR). Scale bar- 1μm (Nanci, 2012) ........... 4
Figure 1.4. A. Enamel in 12 year old patient with AMELX deletion causing amelogenesis
imperfecta (Hu et al., 2012); B. SEM image of normal mouse enamel showing decussating pattern
of enamel rods; C. SEM image of AmelX KO knockout mouse enamel lacking rod-interrod
architecture (Bidlack et al., 2017). .............................................................................................. 6
Figure 1.5. Sequence alignment between murine and porcine amelogenin. Signal peptide is gray,
conserved N-terminal region is purple, hydrophilic C-terminal region is pink. ............................. 7
Figure 1.6. Mouse ameloblastin sequence. Signal peptide in gray; exon 5 encoded region in
purple, with self-assembly domain underlined; exon 6 encoded region in pink. .........................11
Figure 1.7. A. TEM image showing ameloblastin localizes in enamel rod sheath (Uchida et al.,
1995); B. Confocal image showing immunohistochemically labelled Ambn fragments in rod
sheaths, modified from Mazumder et al. (2016) supplement. ....................................................12
xiii

Figure 2.1. Photographs of A. a 6 month old porcine jaw, unerupted second molar crown within
its bony crypt shown by white arrow. B. Scraping newly formed enamel with a razor blade on a
clean glass plate. ......................................................................................................................25
Figure 2.2. Mass spectra for rM179; Inset: 12% SDS-PAGE gel showing rP172 (lane 1), ladder
(lane 2), and rM179 (lane 3). .....................................................................................................27
Figure 2.3. Mass spectra for rAmbn; Inset: 12% SDS-PAGE showing rAmbn (lanes 1&2). ......27
Figure 2.4. Porcine enamel matrix extract characterization. A. 12% SDS-PAGE stained with
coomassie blue showing all the detectable proteins and protein fragments from porcine EMP
extract, B. Western blot against Ambn showing porcine Ambn as 2 bands (arrows), C. Western
blot against Amel showing characteristic porcine Amel 18 k and 20 k bands (arrows). ..............28
Figure 2.5. Western blot labeled with anti-Amel antibody showing rAmel pulled down with rAmbn
(seen in Figure 2.6) in the elution fractions from an anti-Amel antibody column. Controls 1 to 4 do
not contain any rAmel. ..............................................................................................................29
Figure 2.6. A. Western blot with anti-Ambn antibody (M300) showing rAmbn pulled down in the
elution fractions from an anti-Amel antibody column. B. Graph quantifying band volume (amount
of protein) in elution and control fractions from A. .....................................................................30
Figure 2.7. A. Western blot labeled with anti-Ambn antibody showing ameloblastin washes; B.
Western blot labeled with anti-Amel antibody showing amelogenin washes. Wash 4 has negligible
amounts of protein in both. ........................................................................................................31
Figure 2.8. Western blots labeled with anti-Ambn (A) and anti-Amel (B) antibodies showing Amel-
Ambn co-eluted from porcine enamel matrix protein extract from an anti-Amel ab column. Putative
Amel-Ambn complex marked by an asterisk (*). ........................................................................32
xiv

Figure 2.9. Western blot labeled with anti-Amel antibody showing rAmel pulled down with rAmbn
(seen in Figure 2.10), from an anti-Ambn ab column. There is significantly less rAmel detected in
control fractions (Cntrl E1 and Cntrl E2). ...................................................................................33
Figure 2.10. Western blot labeled with anti-Ambn antibody (R&D) showing rAmbn pulled down
with rAmel (seen in Figure 2.9) in elution fractions from an anti-Ambn ab column. Amount of
protein in control fractions is insignificant (Cntrl E1 and Cntrl E2)..............................................34
Figure 2.11. Western blots labeled with anti-Ambn (A) and anti-Amel (B) antibodies showing
native Ambn and Amel co-eluted from porcine enamel matrix protein extract from an anti-Ambn
(R&D) ab column. .....................................................................................................................35
Figure 2.12. A. Western blot labeled with Anti-Ambn antibody (R&D) showing no cross-reactivity
with rAmel (lane2); B. Western blot labeled with anti-Amel antibody, showing no cross-reactivity
towards rAmbn (lane 1); C. 12% SDS-PAGE to show loading controls in each blot. .................36
Figure 2.13. Summary of results for recombinant and native Amel and Ambn co-
immunoprecipitation. Western blots showing Amel and Ambn labeled in A. elution fractions of
anti-Amel antibody co-IP column, B. elution fractions of anti-Ambn antibody co-IP column. ......38
Figure 3.1. Mouse ameloblastin sequence showing the region deleted in AmbnΔ5 (exon 5
encoded region) in purple and the region deleted in AmbnΔ6 (exon 6 encoded region) in pink.
Signal peptide is in grey. ...........................................................................................................44
Figure 3.2. Cartoon showing sequences of Ambn synthetic peptides; AB2 (encoded by exon 5) -
underlined, AB2N- blue, AB2C- purple, and AB4 (encoded by exon 6) - pink. ...........................44
Figure 3.3. 12% SDS gels stained by coomassie blue showing, A. AmbnΔ5 and AmbnΔ6.
Modified from supplemental data, Su et al. (2019b); and B. full-length rAmbn. C. Western blot
xv

labeled by anti-Ambn M300 antibody showing AmbnΔ5 and AmbnΔ6. Calculated molecular
weights of each are mentioned on the Western blot. .................................................................47
Figure 3.4. A. Western blot labeled with anti-Amel antibody shows no cross-reactivity towards
AmbnΔ6 but a faint band is observed in the lane with AmbnΔ5. B. Western blot labeled with anti-
Ambn M300 antibody showing that it does not cross-react with rAmel. .....................................48
Figure 3.5. Western blot of co-IP elution fractions labeled in tandem with anti-Ambn M300 and
anti-Amel antibodies. Lane 1- full-length rAmbn co-eluting with rAmel (positive control); Lane 2-
AmbnΔ6 co-eluting with rAmel; Lane 3- AmbnΔ5 does not bind to rAmel, and only rAmel was
identified in the elution; Lane 4- standard protein ladder; Lane 5- full-length rAmbn control, Lane
6- rAmel control .........................................................................................................................49
Figure 3.6. 16% SDS-PAGE gel stained by silver staining showing elution fractions of co-IP
experiments. Ambn peptides were used as prey and rAmel as bait. Lane 1- exon 6 AB4 does not
bind to rAmel. Lane 2- exon 5 AB2 binds to rAmel and co-elutes. Lane 3- N-terminus of exon 5
AB2N binds to rAmel and co-elutes. Lane 4- C-terminus of exon 5 AB2C does not bind to rAmel.
Lane 5- rAmel control. Lane 6- standard protein ladder. Lane 7, 8, and 9- AB2, AB2N and AB2C
controls respectively. Lane 10- Low molecular weight ladder. ...................................................50
Figure 3.7. DLS peaks showing the % mass distribution of hydrodynamic radii of rAmel
assemblies in 25 mM Tris-HCl, pH 7.4-7.6, at 22° C. ................................................................57
Figure 3.8. Histogram of DLS peaks from Figure 3.7, showing a homogenous population of
nanospheres with RH 7.1-14.9 or ~20 nm average diameter .....................................................58
Figure 3.9. Negatively stained TEM images A. Tris-HCl control showing no structures resembling
nanospheres, 200,000 X magnification. B. 0.3 mg/mL rAmel, 40,000 X magnification showing
xvi

typical rAmel nanospheres ~20 nm in diameter. Inset shows rAmel nanospheres at 200,000 X.
.................................................................................................................................................59
Figure 3.10. Histogram of the change in size distribution of rAmel+AB2 over time, from DLS
analysis. Pink bars show rAmel+AB2 measured immediately after mixing (0 h) and purple bars
show the change in their size after 3 h incubation at RT. ..........................................................60
Figure 3.11. TEM image of AB2 particles ~25 nm in diameter. Scale bar 100 nm. ...................61
Figure 3.12. TEM images of negatively stained rAmel (A) and rAmel+AB2 (B) at 40k X
magnification showing reduced particle size of rAmel upon addition of AB2. .............................62
Figure 3.13. Dot plot comparing the diameters of rAmel nanospheres and rAmel+AB2 co-
assemblies. Average diameter of rAmel nanospheres was 24.4 nm (red dots) which decreased to
an average of 18.4 nm when AB2 was added to rAmel (yellow dots), n= 100, counted in 4 images,
2 rAmel only and 2 rAmel + AB2; p << 0.0001, n = 100 .............................................................62
Figure 3.14. DLS peaks showing size distribution of rAmel and rAmbn particles A. rAmel particles
are majority nanospheres (~22 nm diameter). B. rAmbn distribution is more heterogenous.
Average particle size rAmbn ~14 nm in diamter. .......................................................................64
Figure 3.15. A. DLS peaks showing particle size distribution of rAmel+rAmbn at 1:3 ratio by
weight. B. Histogram showing heterogeneous size distribution of rAmel+rAmbn particles. Average
particle size was ~ 18 nm. .........................................................................................................65
Figure 3.16. TEM image of negatively stained rAmbn (0.04 mg/mL) showing smaller spherical
assemblies (white arrow heads) ~20 nm in diameter and some larger ~100 nm long aggregates
(center). ....................................................................................................................................66
xvii

Figure 3.17. TEM image of negatively stained rAmel-rAmbn mixture showing spherical co-
assemblies of ~12 nm average diameter. ..................................................................................66
Figure 3.18. Dot plot comparing the diameters of rAmel nanospheres and rAmel+rAmbn co-
assemblies. Average diameter of rAmel nanospheres was 22 nm (blue dots) which decreased to
an average of 12 nm when rAmbn was added to rAmel (orange dots). n = 100, p <<< 0.0001..67
Figure 4.1. A,B,D&E. Confocal images of P8 enamel cross section showing Amel-Ambn N-
terminal fragments colocalization in rod-sheaths in characteristic “fish-net” pattern. C.
Pseudocolored fluorescent resonance energy transfer (FRET) image showing FRET efficiencies
from highest (red, 0.89, positive FRET signal, Amel-Ambn fragments closest) to lowest (purple,
0, negative FRET) (Mazumder et al., 2016). .............................................................................74
Figure 4.2. Cartoon of Ambn exon 5 encoded fragment showing the amphipathic helix motif in
pink and part of the Ambn self-assembly motif (putative Amel-binding region) in the purple box.
.................................................................................................................................................75
Figure 4.3. Cartoon showing two sectioning orientations used for P8 mandibles. A. in the sagittal
plane, B. in the transverse plane, anteroposteriorly. ..................................................................77
Figure 4.4. Mouse Ambn sequence depicting the antigen epitope for M300 antibody in purple 78
Figure 4.5. Confocal tile-scan image of P8 incisor showing Amel-Ambn colocalization in sagittal
section. Amel is labeled with Alexa 488 (green), and Ambn-M300 by Alexa 594 (red). Nuclei are
labeled with DAPI (blue). White squares show areas selected for high resolution images
representing secretory, transition and maturation stages. Arrowheads show areas selected for
colocalization analysis- within ameloblasts, at the secretory front / Tomes’ processes, and within
the enamel thickness. ...............................................................................................................84
xviii

Figure 4.6. High resolution confocal images of P8 incisors from secretory (A), transition (C), and
maturation (E) stages of amelogenesis. B, D, and F. Manders’ colocalization coefficients for A, C,
and E respectively. ....................................................................................................................85
Figure 4.7. Confocal tile-scan image showing P8 mouse incisor labeled with anti-Amel (green)
and anti-Ambn (red) N-18 antibody. Nuclei are labeled with DAPI (blue). Am: Ameloblasts, E:
Enamel .....................................................................................................................................88
Figure 4.8. High resolution confocal images of P8 incisor secretory (A), transition (C), and
maturation (E) stages. B, D, and F. Manders' coefficients of colocalization for A, C, and E
respectively. E: Enamel, TP: Tomes’ processes, Am: ameloblasts. ...........................................89
Figure 4.9. A. Confocal images of aneteroposterior section of P8 incisor showing Amel-Ambn
fragments colocalizing in the characteristic decussating pattern of mouse enamel rods. OE- outer
enamel, BE- bulk enamel. B&C. Amel and Ambn channels respectively. ..................................90
Figure 4.10. A. Confocal image of anteroposterior section of enamel showing Amel-Ambn
colocalization within the surface enamel (SE) or final enamel, outer enamel (OE) and bulk enamel
(BE); B and C. Amel and Ambn channels respectively; D (inset). SEM image from Lacruz et al.
(2012) showing surface enamel, outer enamel and bulk enamel. ..............................................91
Figure 4.11. Optical microscope images of Alizarin red S stained images of P8 incisor at 40 X
magnification. A. secretory stage, B. transition stage, C. maturation stage. Am- ameloblasts, TP-
Tomes’ processes, E- enamel, D- dentin. ..................................................................................92
Figure 4.12. Confocal tile-scan showing colocalization of Ambn (green) with ameloblast cell
membrane (red) in secretory and transition stage (arrowheads). Nuclei in blue. .......................93
xix

Figure 4.13. Confocal images showing A. colocalization of Ambn with Tomes’ processes of
ameloblast cells (white arrowhead); B. Ambn labeled with M300, and secondary antibody
conjugated with Alexa 488. White arrowhead shows Ambn at the apical end of ameloblasts; C.
ameloblast cell membranes labeled with DiD. White arrowhead shows Tomes’ processes. ......94
Figure 4.14. Confocal images showing A. colocalization of Ambn (green) with ameloblast cell
membrane (red) in transition stage. Nuclei in blue. Arrowhead shows Ambn-cell membrane
colocalization at the apical end; B. Ambn labeled with M300 and Alexa 488. Arrowheads show
localization of Ambn within cells and at the apical end of ameloblasts; C. ameloblast cell
membrane labeled with DiD. .....................................................................................................95
Figure 4.15. Confocal image showing FITC labeled AB2 colocalizing with ALC membrane. A.
Merged image of AB2-FITC colocalizing with cell membrane (DiD), B. AB2-FITC channel, C. DiD
channel. ....................................................................................................................................96

xx

List of Schematics
Scheme 2.1. Co-immunoprecipitation protocol .........................................................................23
Scheme 3.1. Protocol for co-IP between rAmel (bait) and mutant AmbnΔ5 or Ambn Δ6 or Ambn
peptides (prey). .........................................................................................................................45
Scheme 4.1. Schematic showing some of the ameloblast derived cell lines. Blue hexagon
mentions primary cell lines and white hexagons are immortalized cell lines. .............................75
Scheme 5.1. Model summarizing the aims and conclusions of this project. Amel-Ambn interact
through their N-terminal regions, and co-assemble to form spherical assemblies (Aims I and II).
Their co-assemblies are detected in vivo at the secretory face of ameloblasts. ....................... 103

xxi

Abbreviations
ALC- Ameloblast lineage cells
Amel/rAmel- Amelogenin/ recombinant Amelogenin
Ambn/rAmbn- Ameloblastin/ recombinant Ameloblastin
Co-IP- Co-immunoprecipitation
DLS- Dynamic light scattering
ECM- Extracellular matrix
EMP- Enamel matrix proteins
HAP- Hydroxyapatite
HPLC- High performance liquid chromatography
MS/Mass Spec- Mass spectrometry
PTM- Post-translational modification
SDS-PAGE- Sodium dodecyl sulphate-polyacrylamide gel electrophoresis
TEM- Transmission electron microscopy

xxii

Abstract
Introduction: Dental enamel formation is guided by a transient extracellular organic matrix which
gives rise to an almost completely inorganic enamel containing organized hydroxyapatite (HAP)
crystals. The most abundant enamel matrix proteins (EMPs) amelogenin (Amel) and ameloblastin
(Ambn) have been studied for over 2 decades to uncover their role in enamel formation. Literature
on Amel and Ambn has revealed that, a) Amel and Ambn are co-secreted through the same
secretory vesicles, b) Amel and Ambn co-localize in developing molars, c) Amel-Ambn double
null mice show a more severe enamel phenotype than the single mutant mice, and d) Amel
synthesis is downregulated in Ambn mutant mice with deletion of exons 5 and 6. Based on these
data, I hypothesized that Amel and Ambn interact and co-assemble during enamel formation. One
of the proposed functions of this interaction is to adhere the enamel extracellular matrix to
ameloblast cells.
Materials and methods: Direct binding between Amel-Ambn in vitro was identified using co-
immunoprecipitation (co-IP). Co-IP experiments were performed using both recombinant Amel
and Ambn as bait. Experiments were repeated with native porcine EMP extract to isolate native
Amel-Ambn complexes using co-IP. The region of Ambn directly binding to Amel was identified
using 2 Ambn mutants with deletions of exons 5 (AmbnΔ5) and exon 6 (AmbnΔ6), as well as
Ambn synthetic peptides encoding exons 5 and 6. A second set of in vitro techniques of dynamic
light scattering (DLS) and transmission electron microscopy (TEM) was used to observe the co-
assembly of Amel with Ambn at room temperature, near physiological pH. In vivo Amel-Ambn co-
localization was demonstrated using post-natal day 8 wild-type mouse incisors sectioned in
sagittal and transverse planes. Amel was co-labeled with Ambn fragments containing either the
N-terminus or the C-terminus using two different anti-Ambn antibodies and one anti-Amel
antibody. Colocalization between Ambn and ameloblast cell membrane was analyzed by co-
xxiii

staining the cell membranes of ameloblasts using the lipophilic dye DiD with immunolabeled
Ambn. Ameloblast lineage cells (ALC) and fluorescently tagged Ambn synthetic peptide encoding
exon 5 were used to confirm that Ambn binds to the ameloblast cell membrane via its exon 5
encoded fragment. Imaging of immunohistochemically labeled mouse tissues was performed on
Leica SP8 confocal microscope and colocalization was quantified using Manders’ colocalization
coefficient.
Results: Recombinant Amel and Ambn directly bound to each other and the binding was
independently confirmed by using Amel and Ambn as bait proteins. Isolation of native Amel-Ambn
complexes from porcine EMP extracts indicated that post-translational modifications neither
prevented not aided this binding. AmbnΔ5 failed to interact with Amel and interaction of Amel with
Ambn synthetic peptides further confirmed that Ambn bound to Amel via the N-terminus of its
exon 5 encoded fragment. Amel-Ambn co-assemblies detected by DLS and TEM resembled Amel
nanospheres in shape but were smaller in size, about 12-15 nm in diameter at near physiological
pH. In enamel biomineralization, Amel-Ambn binding and co-assembly may be essential for
maintaining HAP crystal organization and rod-interrod arrangement. We further suggest that Amel
nanosphere like structures observed in vivo could be Amel-Ambn co-assemblies.
In mouse incisor, the C-terminal containing Ambn (mostly full-length Ambn) co-localized with Amel
within ameloblasts and at the secretory face of ameloblasts from secretory to transition stages of
enamel formation. At maturation stage, the proteins continued to colocalize within ameloblasts,
but there was no colocalization observed within the enamel matrix. On the other hand, the N-
terminal containing fragments of Ambn colocalized with Amel throughout the thickness of the
developing enamel from secretory to maturation stages. Co-localization of Amel-Ambn was also
observed within the enamel rod architecture in transverse sections of maturation stage incisor
tips. These in vivo images supported our idea that Amel-Ambn co-assembly is essential for
maintenance of rod-interrod architecture.
xxiv

Previous work from our lab has shown that Ambn can interact with cell membrane mimicking lipid
vesicles. Confirming these in vitro data, our in vivo analysis showed that Ambn colocalized with
the ameloblast cell membrane at the apical end from secretory to transition stages. Fluorescently
labeled Ambn exon 5 peptide localized on the cell membrane of ameloblast lineage cells
confirming the role of exon 5 in Ambn cell interaction. From these data we proposed a second
function for Amel-Ambn interaction that Ambn can act as a cell adhesion molecule between
ameloblasts and enamel extracellular matrix by binding to ameloblast cell membrane and Amel
in the extracellular matrix.
Conclusion and future directions: Our approach of combining in vitro analysis of recombinant
and native Amel and Ambn, as well as in situ techniques using wild-type mice allowed us to
observe Amel-Ambn interaction in sound enamel. I showed direct Amel-Ambn binding and
identified the Amel-binding domain at the N-terminus of exon 5 encoded fragment of Ambn. I
proposed a two-fold function of this binding, in biomineralization to allow Amel to bind to HAP
crystals more efficiently after the cleavage of its C-terminus and in ameloblast-extracellular matrix
adhesion by interaction of Ambn with ameloblast cell membrane. Exon 5 encoded region of Ambn
was found to be particularly essential for both Amel and cell membrane interactions. This region
contains the Ambn self-assembly domain which may be involved in Amel-Ambn co-assembly.
Further investigation is needed to confirm this hypothesis and to detect the presence of Amel-
Ambn co-assemblies in vivo. Towards the C-terminal of Ambn self-assembly domain lies the
amphipathic helix motif of Ambn which can bind to ameloblast cell membranes as shown in our
recent publication. How the intrinsically disordered protein Ambn modulates itself to allow multi-
targeting of its exon 5 encoded fragment remains to be investigated. By understanding these
functions of Amel-Ambn interaction, we are opening more avenues for cell-free enamel repair.
Currently, Amel derived peptides are being used to restore enamel however, the newly developed
xxv

enamel-like layer lacks the rod-interrod architecture. Greater understanding of EMP interactions
can lead to improved techniques of enamel repair in the future.
1

1. Introduction
Enamel formation or amelogenesis is an intricate and unique process, the likeness of which is not
observed in any other mineralized tissue. Enamel forming cells ameloblasts secrete enamel
specific proteins amelogenin, ameloblastin, enamelin, amelotin, and ODAM (odontogenic
ameloblast associated protein) along with proteinases metalloproteinase-20 (MMP-20) and
kallekrein-4 (KLK-4) (Bartlett and Simmer, 1999; Iwasaki et al., 2005; Moffatt et al., 2008;
Moradian-Oldak et al., 1996; Simmer and Fincham, 1995; Snead et al., 1985). Ameloblasts and
enamel matrix proteins (EMPs) are transient in nature (Shibata et al., 1995) making them
especially challenging to study. Mature enamel is cell free, almost entirely composed of inorganic
hydroxyapatite (HAP) and contains less than 5% organic matter in the form of water and
fragments of EMPs (Glimcher et al., 1977; Robinson et al., 1971). Any mutations in the EMPs or
proteinases lead to a hereditary disease called Amelogenesis Imperfecta (AI) (Smith et al., 2017).
Therefore, in order to understand enamel formation in health and disease, it is essential to study
the interactions and functions of enamel matrix proteins. The aim of this project is to characterize
the binding between two of the most abundant EMPs- amelogenin and ameloblastin in vivo and
in vitro in order to provide insight into their cooperative function during enamel formation.
1.1 Amelogenesis
Enamel is the hardest mineralized tissue in the human body. The mechanical properties of enamel
are the result of its specialized hydroxyapatite (HAP) structure. Each enamel rod is surrounded
by remnants of the organic matrix (Baldassarri et al., 2008). The rod-interrod architecture of
enamel (Figure 1.1) is described as keyhole pattern in human (Meckel et al., 1965) and
decussated pattern in mouse enamel (Møinichen et al., 1996). Each rod is made up of thousands
2

of needle like HAP crystals synthesized by a single ameloblast (Skobe, 2006). Ameloblasts and
therefore enamel are ectodermal in origin unlike other mineralized tissues in the body like bone,
dentin, and cementum which are mesenchymal in origin. However, ecto-mesenchymal
interactions are essential for ameloblasts to start secreting enamel and fully differentiate
(Zeichner-David et al., 1995). Ameloblasts go through three primary stages of development
namely secretory, transition, and maturation (Figure 1.2). These stages described below are
characterized by the ameloblast morphology, function, and contents of the enamel extracellular
matrix.
1.1.1 Pre-secretory and secretory stages
Pre-ameloblasts are derived from the inner enamel epithelium of the enamel organ, following the
formation of odontoblasts from the dental papilla. The important event of reversal of cell polarity
of the inner enamel epithelial cells takes place at this point, wherein the basement membrane is
Figure 1.1 SEM images showing the rod-interrod structure of mouse (A) and human enamel (B). Inset
of mouse enamel shows individual HAP crystals assembling to form one enamel rod (Moradian-Oldak,
2012; Nanci, 2012).
3

eliminated, the nuclei move towards the stratum intermedium, and a robust protein synthesis
complex (rough endoplasmic reticulum and Golgi) develops, giving rise to pre-ameloblasts.
Pre-ameloblasts are in direct contact with the initial dentin matrix secreted by odontoblasts. As
pre-ameloblasts mature into ameloblasts they secrete enamel matrix proteins directly on the
surface of the dentin matrix. The secretory stage ameloblasts (Figure 1.2, S) are characterized
by tall columnar morphology (40-50 μm), with a polarized nucleus at the basal end and a
specialized secretory structure called Tomes’ process at the apical (secretory) end. The
ameloblasts are supported by the stratum intermedium and stellate reticulum at the basal end.
Their apical secretory processes or Tomes’ processes are aligned at an angle to the secretory
front of ameloblasts (Skobe, 2006). This angle is contiguous with the angle at which enamel rods
are deposited. The enamel matrix proteins are secreted from two sides of the Tomes’ processes.
A distal side and a proximal side (Figure 1.3), which give rise to the rod and interrod enamel
respectively. As ameloblasts deposit enamel matrix proteins, they move away from the dentino-
enamel junction (DEJ). This movement of ameloblasts is essential for the enamel prism
architecture and it has been argued that it plays as much, if not a bigger role in the formation of
the rod-interrod arrangement as the amelogenin structural matrix (Simmer et al., 2012).
Figure 1.2. TEM images showing stages of amelogenesis: Secretory (S), transition (T), and maturation (M)
stages. Ameloblasts (Am), enamel (E), Tomes’ processes (red arrow). Modified from Nanci (2012).
4

Immunolabeling experiments have shown that amelogenin and ameloblastin are co-secreted
within the same vesicles from secretory stage ameloblasts (Zalzal et al., 2008). Efforts to localize
different enamel matrix proteins during secretory stage have detected predominantly amelogenin
and its fragments in the enamel thickness, whereas ameloblastin and enamelin the secretory face
of ameloblasts (Gallon et al., 2013; Mazumder et al., 2014; Nanci et al., 1998). Ameloblastin N-
terminal containing fragments have also been identified within the enamel thickness and co-
localize with amelogenin in the rod sheath space (Mazumder et al., 2016).
The extracellular matrix in secretory stage is primarily organic and composed of amelogenin,
along with ameloblastin, enamelin, and proteinase MMP20. The enamel matrix proteins are
cleaved soon upon secretion by MMP20, primarily at their C-termini (Lu et al., 2008). Primitive
mineral ribbons of calcium phosphate are also visible in early secretory stage.
Figure 1.3. TEM image showing secretory face of Tomes’ processes (TP), distal end (DE), proximal side
(PS), rod enamel (R), interrod enamel (IR). Scale bar- 1μm (Nanci, 2012)
5

1.1.2 Transition and maturation stages
Ameloblasts in transition and maturation stages are characterized by the lack of Tomes’
processes (Figure 1.2, T&M). By the end of the secretory stage, almost entire thickness of the
enamel matrix is deposited. In maturation stage, the enamel crystals increase in bulk rather than
number. Although, secretion of full-length proteins is minimum in this phase, protein synthesis still
continues. This is particularly true in case of ameloblastin which has been detected within
maturation stage ameloblasts, even after cessation of amelogenin synthesis (Nanci et al., 1998).
As more calcium phosphate is deposited in the enamel matrix, proteins are further cleaved by
KLK-4 and removed by the ameloblasts. The maturation stage ameloblasts can be differentiated
into ruffle ended and smooth ended. The ruffle ended ameloblasts are primarily responsible for
secretion of calcium ions into the maturing matrix whereas the smooth ended ameloblasts remove
organic material. During the change from rough to smooth morphology the pH of the enamel
matrix is also modulated. Neutral to slightly alkaline pH is needed for HAP crystal growth, which
is observed in the presence of smooth ended ameloblasts and more acidic (6.5) pH is seen in the
presence of ruffle ended ameloblasts. This suggests that, although ruffle-ended ameloblasts
pump calcium ions into the enamel matrix, the actual incorporation of the ions into the crystals
and crystal growth occurs in the smooth ended phase (Lacruz et al., 2010). Enamel matrix
proteins ODAM and amelotin are also secreted in this stage and localize at the junction of enamel
matrix and maturation stage ameloblasts (Iwasaki et al., 2005; Park et al., 2007). The stellate
reticulum and stratum intermedium fuse to form the papillary layer, which is interspersed by blood
vessels. This layer is proposed to be the source of calcium ions being directed to the ameloblasts
(Smith, 1998).
Towards the end of maturation stage, ameloblasts appear almost cuboidal and epithelial like, and
eventually fuse with the papillary layer forming reduced enamel epithelium (REE). As the tooth
6

erupts, REE merges with the oral epithelium and ameloblasts cease to exist. The absence of
ameloblasts in erupted teeth is one of the reasons why enamel cannot repair or regenerate, and
what makes studying amelogenesis more challenging.
1.2 Amelogenins
Amelogenins are a group of alternatively spliced proteins secreted by ameloblasts, which make
up the bulk of the extracellular matrix during enamel development. The largest amelogenin (Amel)
is about 22 kDa, and is an intrinsically disordered protein (IDP) (Delak et al., 2009).
Amelogenins are indispensable for the development of enamel rod-interrod architecture (Gibson
et al., 2001; Moradian-Oldak, 2012; Sasaki et al., 2007). The amelogenin gene is located on X
and Y chromosomes in humans and cattle, but only the X chromosome copy is present in mice
and rats (Snead et al., 1985). In human, majority of secreted Amel is from the X chromosome and
only 10% protein is from the Y chromosome gene when present. To study the function of
amelogenin, AmelX knockout (KO) mouse was engineered (Gibson et al., 2001). In the mutant
Figure 1.4. A. Enamel in 12 year old patient with AMELX deletion causing amelogenesis imperfecta (Hu
et al., 2012); B. SEM image of normal mouse enamel showing decussating pattern of enamel rods; C.
SEM image of AmelX KO knockout mouse enamel lacking rod-interrod architecture (Bidlack et al., 2017).
7

mouse, a thin white and opaque layer of enamel was formed. The mineral phase of AmelX KO
enamel was primarily octacalcium phosphate (OCP) instead of HAP (Hu et al., 2016). At the
structural level, it lacked the rod-interrod architecture characteristic of enamel (Figure 1.4 B) and
hence lacked the physical and mechanical properties of normal enamel. In humans, there are at
least 20 known mutations identified in the amelogenin gene (AMELX) leading to amelogenesis
imperfecta (AI, Figure 1.4 A), which mostly presents as pitted/ridged enamel which abrades
rapidly exposing the underlying dentin (Smith et al., 2017).
1.2.1 Biochemical properties
Amelogenin protein has 180 amino acids in mouse (M180) and 173 (P173) in pig (Figure 1.5). It
has a hydrophilic C-terminus and a hydrophobic core. The C-terminus is cleaved soon upon
secretion by MMP-20 giving rise to the predominant amelogenin found in enamel matrix (Ryu et
al., 1999). Amel forms supramolecular assemblies called as nanospheres above or close to
physiological pH (Fincham et al., 1995). The nanospheres are about 20 nm in diameter and are
formed by a hierarchical assembly mechanism, going from monomers to oligomers to
nanospheres (Bromley et al., 2011; Fang et al., 2011).
Figure 1.5. Sequence alignment between murine and porcine amelogenin. Signal peptide is gray,
conserved N-terminal region is purple, hydrophilic C-terminal region is pink.
8

A single nanosphere of full length amelogenin consists of about 40 Amel monomers (Chen et al.,
2011; Moradian-Oldak et al., 1998), with the C-terminus facing out (exposed to the buffer side)
and hydrophobic core towards the center of the nanosphere (Fang et al., 2011). In the absence
of the C-terminus, nanospheres can still form but they take longer to assemble and are larger
than the full length Amel nanospheres (Moradian-Oldak et al., 2002). The N- and C-termini of
Amel are highly conserved throughout evolution (Toyosawa et al., 1998). Two self-assembly
domains have been identified by the two-hybrid system at the N- and C-termini of Amel termed
as domain A and domain B. Domain A consists of 42 residues at the N-terminus of Amel and
domain B is at the C-terminus, containing 17 residues (Paine and Snead, 1997). Deletion of A-
domain leads to a heterogeneous distribution of Amel nanoparticles ranging from 2 to 38 nm
whereas deletion of the B-domain causes formation of large Amel particles ~50 nm in size. Even
a point mutation in the A-domain leads to disturbances in nanospheres formation, suggesting its
importance in amelogenin self-assembly (Moradian-Oldak et al., 2000). A-domain is also involved
in potential Amel-Ambn interactions (Su et al., 2016) which is described in detail in Chapter 2.
Amelogenin has a single post-translational modification (PTM), a phosphorylation on Serine-16
(Fincham et al., 1994). The phosphorylated amelogenin stabilizes amorphous calcium phosphate
(ACP), whereas the unphosphorylated form promotes formation of hydroxyapatite (HAP) from
ACP in vitro (Kwak et al., 2009). Both forms of amelogenin have been identified in vivo, the
phosphorylated form being more abundant in newly forming enamel matrix and dephosphorylated
(or unphosphorylated) form being more abundant in older enamel matrix (Green et al., 2019). The
phosphorylated form maybe essential to prevent early precipitation of HAP, before the entire
thickness of enamel is laid down, whereas the unphosphorylated (or dephosphorylated) form will
be necessary to mature the HAP crystals during later stages of enamel development.
9

1.2.2 Functions
Amelogenin and its alternatively spliced products make up almost 80% of the enamel extracellular
matrix. The full-length amelogenin alone cannot completely rescue the Amel KO mouse
phenotype (Li et al., 2008) however, a knock-in mouse line expressing only full-length Amel
displays normal enamel (Snead et al., 2011). This demonstrates the complexity of amelogenin’s
alternatively spliced isoforms, the functions which are not fully known. In vitro studies using
constant composition crystallization system have allowed the capture of intermediate stages
starting from calcium-phosphate pre-nucleation clusters to HAP nanorods. These studies
identified that amelogenins can stabilize Ca-P pre-nucleation clusters which can partly explain
how Amel mediates controlled crystal growth during enamel mineral formation in vivo (Wang et
al., 2007; 2008; Yang et al., 2010). Amel nanospheres chains have been observed attached to
the lateral sides, along the length of the forming HAP crystals in vitro, in mouse and bovine
developing enamel (Fincham et al., 1995). Amel prevents untimely fusion of early mineral crystals,
allowing the growth along the c-axis until long thin ribbons of HAP cover the entire thickness of
enamel (Moradian-Oldak and Goldberg, 2005; Moradian-Oldak et al., 2003). Amelogenin also
controls phase transformation of ACP to apatite. Amelogenin with a phosphorylation at serine 16
stabilizes amorphous calcium phosphate (ACP), whereas non-phosphorylated Amel promotes
formation of hydroxyapatite (HAP) from ACP in vitro (Kwak et al., 2009). Recent analysis of Amel
KO (AmelX
-/-
) and heterozygous (AmelX
+/-
) mice at nanoscale has shown that AmelX KO mouse
enamel is capable of forming mineral ribbons (Hu et al., 2016). These ribbons eventually fuse and
grow into plate like octacalcium phosphate (OCP) crystals. This advocates a strong contribution
of Amel in controlling the phase transformation of enamel crystals, by converting ACP to HAP and
preventing precipitation of OCP which would indirectly maintain the crystal morphology. Simmer
et al suggested that it is possible that the mineral ribbons fail to lengthen in AmelX
-/-
mice because
the expanding Amel matrix in which the mineral ribbons extend is absent. Amelogenin was also
10

reported to affect OCP and apatite crystal morphology in in vitro experimental systems resulting
in their elongation (Iijima and Moradian-Oldak, 2005; Iijima and Moradian–Oldak, 2004; Moradian-
Oldak et al., 2003).
Subsequent cleavage of Amel into smaller fragments by MMP-20 is responsible for the increase
in the bulk of enamel, allowing the crystals to grow in width. Amel is trapped within HAP crystals
in mice lacking MMP-20, which makes the crystals short, flat, and plate-like (Prajapati et al., 2016).
It has been shown that the full-length amelogenin has the highest capacity to aggregate apatite
crystals (Moradian-Oldak et al., 1998), however majority of Amel present in the extracellular
matrix lacks the calcium binding C-terminus (Moradian-Oldak, 2001). This suggests a potential
role of non-amelogenins like enamelin and ameloblastin which have a strong affinity towards
calcium (Yamakoshi et al., 2001), in mediating the interaction between Amel and HAP.
1.3 Ameloblastin
Although amelogenin is the most abundant enamel matrix protein, a thin rough layer of enamel
still forms in patients with AMELX mutations and in AmelX KO mouse (Gibson et al., 2001). This
suggests that amelogenin matrix alone cannot support enamel formation and must work together
with non-amelogenin enamel matrix proteins. The most abundant non-amelogenin is
ameloblastin. It is also intrinsically disordered and thought to play a multifaceted role involving
cell signaling as well as biomineralization (see ‘ameloblastin function’). Ameloblastin is co-
secreted with amelogenin through the same secretory vesicles (Zalzal et al., 2008). There are 3
ameloblastin mutations so far identified in patients with Amelogenesis Imperfecta (Lu et al., 2018;
Poulter et al., 2014; Prasad et al., 2016). To study the effects of lack of function of Ambn, a mutant
mouse lacking exons 5 and 6 of Ambn was designed. This mouse model lacked true enamel layer
on the surface of dentin (Fukumoto et al., 2004), highlighting the importance of Ambn in enamel
11

formation. A true ameloblastin knockout mouse was engineered recently (Liang et al., 2019) which
showed very similar phenotype to the exon 5-6 deletion mutant. An Amel-Ambn double mutant
mouse was later produced by crossing the two mutants and the enamel showed significantly more
phenotypes than the single mutants (Hatakeyama et al., 2009). This further supports the idea that
Amel-Ambn cooperate with each other during enamel formation.
1.3.1 Biochemical properties
Similar to Amel, Ambn is an intrinsically disordered protein. It predominantly has a polyproline II
type confirmation (Wald et al., 2011). Ambn is processed by enamel matrix proteins at its C-
terminal immediately following its secretion (Iwata et al., 2007). Ameloblastin 13, 15, 27, and 29
kDa C-terminal proteolytic cleavage products have been isolated from pig enamel matrix
(Yamakoshi et al., 2001). Full-length Ambn localizes in the region between secretory face of
ameloblasts and enamel matrix, whereas Ambn fragments containing the N-terminus region are
observed throughout the thickness of enamel in rod sheath space (Krebsbach et al., 1996;
Mazumder et al., 2016). Three different cell binding domains have been identified on Ambn in
several species, which are heparin binding, fibronectin binding, and integrin binding domains
(Beyeler et al., 2010; Černý et al., 1996; Sonoda et al., 2009; Su et al., 2019b). However, these
are not conserved throughout evolution. One alternately spliced isoform of Ambn is found in
humans and pigs which lacks a 15 amino acid fragment at the N-terminus of exon 6 (Hu et al.,
1997; MacDougall et al., 2000). Possible function of this isoform is unknown.
Figure 1.6. Mouse ameloblastin sequence. Signal peptide in gray; exon 5 encoded region in purple,
with self-assembly domain underlined; exon 6 encoded region in pink.
12

Wald et al. (2017) identified a Tyrosine/Phenylalanine -x-x- Tyrosine/ Leucine/ Phenylalanine -x-
Tyrosine/ Phenylalanine (Y/F-x-x-Y/L/F-x-Y/F) motif essential for the self-assembly of amelogenin
and ameloblastin. On Ambn this motif lies at the N-terminus of exon 5 (Figure 1.6). When this
motif is disrupted, Ambn loses its ability to self-assemble.
Ambn is phosphorylated and highly glycosylated. There is at least a 20 kDa difference between
the calculated mass of Ambn (~44 kDa) and native Ambn detected on SDS gel (~65 kDa), which
can be partially attributed to the presence of glycosylations. As determined by Edman degradation
the post-translational modifications at the N-terminus of Ambn are, hydroxylation and
phosphorylation at proline 11 and serine 17 respectively, and an O-linked glycosylation at serine
86 (Kobayashi et al., 2007). At the C-terminal end the threonine 361 is O-glycosylated (Yamakoshi
et al., 2001).
1.3.2 Functions
Ameloblastin is involved in mineral formation as well as maintaining ameloblast cell morphology
and function (Fukumoto et al., 2005). It plays an important role in crystal growth and maintenance
of rod-interrod structure (Paine et al., 2003). N-terminal fragments of Ambn accumulate in the rod
Figure 1.7. A. TEM image showing ameloblastin localizes in enamel rod sheath (Uchida et al., 1995);
B. Confocal image showing immunohistochemically labelled Ambn fragments in rod sheaths, modified
from Mazumder et al. (2016) supplement.
13

sheath space in mature enamel in characteristic honeycomb pattern (Uchida et al., 1991), and
the C-terminal region is shown to have calcium binding properties (Yamakoshi et al., 2001).
As mentioned earlier, in mutant mice with deletion of Ambn exons 5 and 6 (Ambn
-5,6/-5,6
) true
enamel is absent and instead small discrete mineralized nodules are observed on the surface of
dentin (Fukumoto et al., 2004; 2005). Ameloblastin C-terminal 13, 15, 27, and 29 kDa fragments
isolated from pig enamel matrix bind to calcium with the highest affinity among all EMPs
(Yamakoshi et al., 2001). The data suggest that Ambn (along with other Ca binding EMPs like
enamelin) maybe regulating the calcium ion concentrations in forming enamel by binding to
excess calcium and slowly releasing it as the hydroxyapatite (HAP) crystals form. This would
allow for controlled deposition and organized crystal growth.
When the self-assembly domain of Ambn (located on N-terminal of exon 5) is mutated in mice,
enamel has reduced mineral density, but its thickness remains unchanged (Wald et al., 2017).
This shows that supramolecular assemblies of Ambn participate in the rod-interrod organization.
Further support for this is provided by transgenic mice (Tg) overexpressing Ambn. The Tg mouse
enamel looks normal at the macroscopic level. However, when observed by scanning electron
microscopic (SEM) it is more porous (Paine et al., 2003). The interrod enamel of Tg mice expands
at the cost of rod enamel. Transgenic enamel crystallites are short and wide and do not span the
entire length of enamel. Outermost aprismatic enamel layer is also absent in the Tg enamel.
These findings directly demonstrate the role of Ambn in maintaining enamel rod-interrod
morphology (Paine et al., 2003).
Three AMBN gene mutation have been so far identified in AI patients. Deletion of exon 6 (Poulter
et al., 2014), a proline to serine point mutation at the C-terminus of Ambn (Lu et al., 2018), and a
splice site mutation, which leads to retention of intron 6 or deletion of exon 7 (Prasad et al., 2016).
The clinical phenotype of these mutations is similar with respect to lack of true enamel and small
14

teeth with exposed dentin. However, subtle differences in the microscopic structure of enamel
offer clues to Ambn function in biomineralization. In the exon 6 deletion patients, the rod-interrod
structure is disrupted but the proline to serine point mutation patients have dentin
hypomineralization as well as tooth root deformities. The role of Ambn in dentin and root formation
needs to be further investigated.
Ameloblastin can modulate ameloblast cell function which may have an indirect effect on enamel
mineralization. The Ambn
-5,6/-5,6
mouse (Fukumoto et al., 2004; 2005) ameloblasts are small and
round instead of tall columnar and show loss of nuclear polarization. The cells separate from the
underlying dentin, re-enter cell cycle, and unlike WT ameloblasts they actively divide. The cells
continue to express other EMPs leading to the formation of calcified nodules within the cell layer.
Amelogenin synthesis is also reduced in this mutant at both mRNA and protein levels. The lack
of a true enamel layer in Ambn
-5,6/-5,6
mice could be a combined effect of the mutant Ambn failing
to carry out its biomineralization functions and the disrupted ameloblast layer. Primary ameloblast
cell cultures from Ambn
-5,6/-5,6
mice proliferate about four times as rapidly as their WT counterparts.
When treated with recombinant Ambn, the proliferation rate of Ambn
-5,6/-5,6
ameloblasts reduces
significantly. Interestingly, Ambn does not affect the rate of proliferation of other cell types like
human kidney cells. These results indicate that Ambn may have a function in controlling
ameloblast cell proliferation and maintaining the morphology of ameloblast cell layer during
enamel formation (Fukumoto et al., 2004; 2005).
Above evidence supports a double function of ameloblastin; in ameloblast cell maintenance as
well as controlling enamel crystal growth and morphology. The intrinsically disordered nature of
ameloblastin described earlier, is a key factor in allowing the protein to enter into different
conformations in order to interact with multiple targets.

15

1.4 Amelogenin-ameloblastin interactions
Techniques to Study Protein-Protein
Interactions
References
Yeast-two-Hybrid system Paine and Snead (1997)
Chemical Crosslinking Brookes et al. (2000)
Enzyme-Linked Immunosorbent Assay (ELISA) Ravindranath et al. (2004)
Immunogold Labeling and TEM for localization Zalzal et al. (2008)
Co-immunoprecipitation Fan et al. (2009)
Quantitative Co-localization Mazumder et al. (2016); Mazumder et al. (2014)
Biophysical Techniques like CD, DLS Su et al. (2016)
Table 1.1. Summary of techniques used to study protein-protein interactions in developing enamel matrix
Different techniques have been utilized in the field of enamel research to identify the relationships
between various secreted enamel matrix proteins (Table 1.1). Some of the significant
contributions to our understanding of protein-protein interactions in enamel formation are
described below.
The self-assembly domains of Amel (A-domain: 42 residues at the N-terminus of amelogenin and
B-domain: 17 residues at its C-terminus) were identified using yeast-2-hybrid system (Paine and
Snead, 1997). The advantage of this technique is that it can be scaled up to scan large number
of proteins, like entire cDNA libraries to detect possible binding partners. Using this, Wang et al.
(2005) identified Amel and Ambn interactions with a number of secreted proteins and integral
membrane proteins. Some of the notable binding partners identified for Amel were- Cd63, protein
phosphatase, and procollagen. It was later confirmed that Amel reuptake during enamel
maturation can occur via Cd63 positive vesicles (Shapiro et al.; Zou et al., 2007). In case of Ambn,
Wang et al. (2005) identified three integral membrane proteins (Gpsn2, Itm2a, and Agpat4) as
binding partners to Ambn, supporting its role in ameloblast cell interaction. Ambn also bound to
Cd63, which could mean it has similar reuptake mechanism as Amel. Interactions of other enamel
16

matrix proteins amelotin and ODAM were similarly identified using yeast-2-hybrid system (Holcroft
and Ganss, 2011). Neither amelogenin-ameloblastin binding, nor Ambn self-assembly could be
detected using yeast-2-hybrid system (Paine et al., 1998). Later it was suggested that Amel-Ambn
interaction occurs via the lectin-like carbohydrate binding domain of Amel binding to the
glycosylations present on Ambn (Ravindranath et al., 2004; Ravindranath et al., 1999).
Brookes et al. (2000) confirmed localization of full-length, nascent Amel versus Amel fragments
by using chemical crosslinking. Smaller Amel fragments are present in the deeper enamel,
whereas full length Amel is only present close to the secretory face of ameloblasts. N-terminal
regions of the proteolytic cleavage products of Amel are in close proximity until the tyrosine rich
amelogenin polypeptide (TRAP) region is cleaved (Brookes et al., 2000). The N-terminal regions
are probably involved in self-assembly of the cleavage products and hence are temporospatially
close together.
Co-secretion of Amel-Ambn was detected by immunogold labeling rat incisor sections. Most
secretory granules (~70%) showed presence of for both Amel and Ambn. Ameloblastin was
labeled at the secretory face of ameloblasts more prominently than within the enamel thickness
whereas, amelogenin was present throughout (Zalzal et al., 2008). There were about 13%
granules containing ameloblastin only, whereas only 1% had amelogenin only (Zalzal et al.,
2008).
Further, using fluorescence resonance energy transfer or FRET, amelogenin and ameloblastin
N-terminal fragments were shown to be within 5-7 nm of each other in the sheath space of enamel
rods (Mazumder et al., 2016), suggesting that they are close enough to directly interact.
Biophysical techniques like circular dichroism (CD) spectroscopy, dynamic light scattering (DLS),
and fluorescence spectroscopy have been used to understand how protein-protein interactions
17

change the secondary structure and biophysical characteristics of proteins. Upon interaction with
32 kDa enamelin, the size distribution of recombinant amelogenin particles change in a pH
dependent manner (Fan et al., 2009; Fan et al., 2011), suggesting their possible interaction.
Similar conformational changes in Amel have been detected in the presence of Ambn peptide
encoded by exon 5. Specifically the TRAP region of Amel, which constitutes the tyrosine rich N-
terminal fragment accumulating in the maturation stage ameloblasts, shows a more alpha-helical
conformation in the presence of Ambn synthetic peptide encoding exon 5 (Su et al., 2016). This
supports the idea that Amel-Ambn can bind directly in the absence of post-translational
modifications.
1.5 Hypothesis, aims, and scope of the work
From the short review of enamel matrix proteins and amelogenesis, it is clear that although we
have come a long way in understanding this unique developmental process, there are still some
gaps in our knowledge particularly concerning the cooperative behavior and functions of enamel
matrix proteins. From Amel knockout and Ambn mutant mice it is clear that, enamel formation can
take place in the absence of the most abundant extracellular matrix protein, but true enamel fails
to form in the absence of functional ameloblastin. Their co-localization and co-secretion further
suggest their possible interaction. Therefore, we hypothesized that amelogenin and ameloblastin
interact and co-assemble in forming enamel. One of the proposed functions of this interaction is
to adhere the amelogenin rich enamel extracellular matrix to ameloblast cells.
I investigated this hypothesis using the following three aims
Aim I. To demonstrate Amel-Ambn interaction in vitro using recombinant and native proteins
Aim II. To identify the binding region of Ambn and to study Amel-Ambn co-assembly
18

Aim III. To analyze the co-localization of Amel-Ambn, and Ambn-cell membrane in vivo
I used various biochemical, biophysical, and in situ techniques to study each aim. In Chapter 2, I
demonstrate the co-immunoprecipitation of recombinant as well as native porcine amelogenin
and ameloblastin, confirming direct Amel-Ambn binding in vitro. In Chapter 3, I identify the binding
domain of Ambn involved in its interaction with Amel, using 2 recombinant Ambn mutants and
Ambn synthetic peptides (Su et al., 2019b). This chapter also illustrates Amel-Ambn co-assembly
using classical biophysical techniques of dynamic light scattering (DLS) and transmission electron
microscopy (TEM) to analyze the effect of Ambn on Amel nanospheres. To understand the
interaction of these proteins in vivo, I utilize the continuously growing mouse incisor model in
Chapter 4 and analyze Amel-Ambn colocalization from secretory to maturation stages of enamel
formation. Ameloblastin and ameloblast cell membrane interaction is also demonstrated in this
Chapter by identifying membrane localization of Ambn within Ameloblast Lineage Cells (Nakata
et al., 2003) and developing mouse incisor.

19

2. Direct evidence of amelogenin-ameloblastin
interaction
Background
The release of enamel matrix proteins by ameloblasts during enamel formation is spatiotemporally
tightly controlled. Previous research has shown that amelogenin (Amel) and ameloblastin (Ambn)
are co-secreted through the same secretory vesicles (Nanci et al., 1998; Zalzal et al., 2008). The
proteins and their N-terminal fragments colocalize at the secretory front of enamel and within
enamel rod sheath space in developing mouse molars (Mazumder et al., 2016; 2014). In
ameloblastin exon 5-6 deletion mutant mouse, the synthesis of Amel is reduced at both mRNA
as well as at protein levels (Fukumoto et al., 2004; 2005). Further, in Amel X knockout mouse,
although lacking a prismatic structure, a thin layer of enamel still forms (Gibson et al., 2001). The
observations that enamel crystals still form may suggest that Amel does not function alone during
enamel formation but rather cooperates with the non-amelogenin proteins such as enamelin or
ameloblastin. Therefore, for the purpose of this project it is reasonable to hypothesize that
amelogenin and ameloblastin directly bind to each other to fulfill certain functions during enamel
formation. This hypothesis is further supported by the recent double knock-out animal model
where both Amel and Ambn expressions are blocked resulting in the formation of a severe enamel
malformation phenotype (Hatakeyama et al., 2009).
In this Chapter, I will demonstrate the binding of full-length Amel and Ambn through co-
immunoprecipitation (co-IP). Co-immunoprecipitation is a classical technique used to identify and
confirm direct binding between pairs of proteins. It is commonly used to identify receptors or
downstream targets of a known ligand (Markham et al., 2007). Typically, co-IP uses an antibody
20

immobilized on an inert resin column (usually agarose); a bait protein is reversibly bound to the
antibody, and a prey protein or a protein mixture containing possible targets is incubated in the
column to capture the binding partner(s). Bound proteins can be eluted and identified using
Western blots or mass spectrometry.
Co-IP has been previously used in the enamel field to confirm the interaction between the 32 kDa
enamelin and recombinant amelogenin. Amelogenin antibody was stabilized on a protein A
column and incubated with a mixture of recombinant amelogenin and native porcine 32 kDa
enamelin (Fan et al., 2009). Amelogenin-enamelin complex was detected by Western blots with
anti-amelogenin and anti-enamelin antibodies (Fan et al., 2009). Co-IP was also used to confirm
the interactions between amelotin and ODAM, and ameloblastin and ODAM, first detected by
yeast-2-hybrid system (Holcroft and Ganss, 2011).
I used co-IP to analyze the binding between recombinant Amel and recombinant Ambn
synthesized in E. coli. First, anti-Amel antibody column was used with Amel as the bait protein
and Ambn as the prey protein. I then confirmed the data by reversing the setup where anti-Ambn
antibody column was used and Ambn and Amel were used as bait and prey proteins respectively.
Co-IP elutions were analyzed using SDS-PAGE and Western blots. Co-IP was repeated with
native porcine enamel matrix protein extract to determine the effect of post-translational
modifications on Amel-Ambn interaction. Six month old developing porcine second molar enamel
provides a relatively large and easily accessible concentration of enamel matrix proteins. The
protein extract has been extensively characterized in the literature in order to identify the
fragments of enamel matrix proteins in vitro. The most abundant Amel and Ambn fragments are
listed in Table 2.1.
Protein
Fragments: mass by mass
spec, “mass based on SDS gel”
Characteristics and references
21

Amel
P161 or “20k”
C-terminal (12 AA) cleaved Amel
(Yamakoshi et al., 1994)
P148: 16919.5 Da or “18k”
Most abundant Amel lacks ~25 AA at C-
terminus (Fincham et al., 1994; Yamakoshi
et al., 1994)
TRAP: 5403.5 Da
Tyrosine rich Amel polypeptide- N-terminal
45 AA proteolytic cleavage product
(Fincham and Moradian-Oldak, 1993)
LRAP: 5344.1 Da or “6.5k”
Leucine rich Amel polypeptide- smallest
alternatively spliced product (Fincham and
Moradian-Oldak, 1993)
Ambn
“13 k”, “15 k”, “27 k”, “29 k”
calcium binding non-amelogenins
C-terminal cleavage products of
ameloblastin with a strong affinity to
calcium (Yamakoshi et al., 2001)
Table 2.1. Amelogenin and ameloblastin fragments identified in the porcine enamel extracellular matrix
Upon repeating co-IP with porcine enamel matrix protein (EMP) extract, the elution fractions from
were analyzed by Western blots. This was followed by mass spectrometry analysis to identify
proteolytic cleavage products in the elution fractions not detectable by Western blot antibodies.
2.1 Materials and methods
2.1.1 Recombinant amelogenin and ameloblastin expression and purification
Recombinant porcine amelogenin (rP172) or recombinant mouse amelogenin (rM179) were
expressed and purified from BL21 E. coli as published earlier (Ryu et al., 1999). For co-IP
experiments with anti-Amel antibody column rP172 was used whereas, for anti-Ambn antibody
column experiments rM179 was used. Due to high degree of sequence similarity between the
porcine and murine proteins the protein-protein interactions results are unaffected. Briefly, rAmel
expressed in E. coli was precipitated by using saturated ammonium sulfate to achieve a final
concentration of 20%. The mixture was centrifuged and the pallet was dissolved in 0.1 % trifluoro
acetic acid (TFA). Final purification of rAmel was done by reversed phase HPLC. Recombinant
Amel expressed in E.coli lacks the first methionine and a phosphorylated Ser at position 16.
22

Recombinant Ambn (mouse sequence) was also expressed in BL21 E. coli. Ambn is ~42 kDa
(calculated mass) and intrinsically disordered making it unstable within bacterial cells. To stabilize
the protein and prevent early lysis in the E. coli system, the protein was expressed with an S-tag
(a 15 amino acid oligopeptide derived from pancreatic ribonuclease A), a Thioredoxin (Trx) tag
(for stability) and a histidine-tag containing six histidine residues (for Nickel column purification).
Protein from lysed E. coli was first concentrated by Ni-NTA column (QIAgen), then dialyzed
through a 10,000 Da dialysis membrane against ice cold phosphate buffer. The tags were cleaved
by enzyme enterokinase at 37 °C and protein was purified by HPLC. Purified rAmel and rAmbn
from HPLC peaks were lyophilized and stored at -20° C until further use.
2.1.2 Antibodies
Antibodies used in the co-IP experiments in the columns and for Western blots are listed in Table
2.2.
Protein Antibody Host Epitope
Amelogenin
Custom antibody, gift
from Dr. Malcolm Snead
Chicken
Polyclonal against the full-length
protein.
Ameloblastin
M300, sc-50534, Santa
Cruz Biotech
(discontinued)
Rabbit
Polyclonal against 300 amino acids
excluding the N-terminus. Epitope-
proline108 to proline407
Anti-ameloblastin ab,
AF3026, R&D systems
Goat
Polyclonal, against the entire protein
except the signal peptide. Epitope-
valine27 to proline407
Table 2.2. List of Ambn and Amel antibodies used in co-immunoprecipitation experiments.
The anti-Ambn M300 antibody was discontinued by the company halfway through this project
hence I had to transition to the anti-Ambn R&D antibody. Western blots against recombinant
mouse Ambn and immunohistochemistry data in mouse tissue have shown no differences in the
epitope recognition of these two antibodies. Differences were noted in recognition of porcine
Ambn and they are mentioned in the results section.
23

2.1.3 Co-immunoprecipitation
For Amel-Ambn co-immunoprecipitation, Pierce co-IP kit (Thermofisher Scientific) was used.
Manufacturer’s protocol was followed to bind 10 μg of anti-amelogenin antibody (gift from Dr. M.
Snead) to the antibody coupling resin. Excess antibody was washed by Dulbecco’s PBS provided
in the kit. Lyophilized rAmel was weighed and dissolved in distilled water at 1 mg/mL
concentration. Ten microliter rAmel solution (10 µg protein) was added to the column as bait
protein and bound overnight at 4° C with gentle shaking. Columns were washed 5-6 times with
Dulbecco’s PBS to remove any unbound bait protein. Washes were analyzed by nanodrop during
the experiment and were retained to be analyzed later by Western blots to ensure that majority of
unbound proteins were removed.
Lyophilized rAmbn was dissolved in distilled water at 1 mg/mL concentration. Ten microliters
rAmbn solution (10 µg protein) was applied to the column for 4 h at 4° C as prey protein. The IP
column was washed 5-6 times to remove unbound prey protein. Washes were analyzed as
described above. Bound proteins were eluted with the elution buffer provided in the kit. Each
elution volume was 60 μL and elution washes were repeated 4-5 times. Elution fractions were
Scheme 2.1. Co-immunoprecipitation protocol
24

lyophilized and dissolved in 20 μL distilled water to concentrate the proteins so as to be detected
by Western blots.
To confirm the findings of co-IP, experimental setup was reversed. Anti-Ambn antibody (R&D)
was bound to the antibody coupling resin. Co-IP was repeated by using Ambn as bait and rAmel
as a prey protein. Concentrations of protein and antibody remained the same as above. A
flowchart describes the co-IP workflow in Scheme 2.1.
Anti-Amel and both anti-Ambn antibodies were tested for cross-reactivity against Ambn and Amel
respectively using Western blots. Either one of the following two types of control experiments
were performed with the co-IP. One control experiment was performed with an IP column without
any antibody coupled to the resin. The second control experiment was performed using the control
resin provided in the co-IP kit which cannot bind to antibodies. Protein concentrations and number
of washes in the control experiments were the same as test co-IP experiments.
2.1.4 SDS-PAGE and Western blots
Twelve percent sodium dodecyl sulfate (SDS) gels, tris-glycine running buffer, and loading buffer
without DTT (no cysteines present in Amel or Ambn) were prepared as per the standard protocol
(Simpson, 2006). Proteins were electrophoresed at 120 V for 90 minutes at room temperature
and transferred to a PVDF membrane using wet blotting method (120 V for 60 minutes on ice).
Membranes were blocked in 5% non-fat milk in phosphate buffered saline, with 0.1% tween-20
(PBST), and incubated in primary antibodies diluted in blocking solution, overnight at 4 °C.
Concentrations for primary antibody (ab) were- 1:1000 for anti-Amel, 1:500 for anti-Ambn M300,
and 1:1000 for anti-Ambn R&D. Membranes were washed with PBST for 15 minutes, 3 times
before being incubated in corresponding secondary antibodies diluted in PBST (1:100), at room
temperature for 3 hours. Secondary antibodies were conjugated with horse radish peroxidase
25

(HRP) and blots were visualized using Amersham ECL Western blotting detection reagents (GE
Healthcare) on an Azure cSeries imaging system. When sample amounts were insufficient, same
membranes were labeled with both anti-Amel and anti-Ambn antibodies in tandem.
2.1.5 Porcine enamel matrix protein extraction
Previously published protocol (Uchida et al., 1995) was followed for native porcine enamel matrix
protein (EMP) extraction. Briefly, 6 month-old pig jaws were ordered from Sierra for Medical
Science packing company (Whittier, CA) and unerupted second molars were extracted from their
bony sockets (Figure 2.1 A). They were cleaned with ice cold phosphate buffered saline (PBS,
pH 7.4) and newly formed enamel was scraped with a sharp razor blade on a clean glass plate
(Figure 2.1 B). Enamel scrapings were stirred overnight at 4° C in 0.5 M acetic acid. The slurry
was desalted using Amicon Ultra 15 centrifugal filters to remove calcium phosphate. The desalted
supernatant was lyophilized and proteins were characterized using Western blots to identify Amel
and Ambn fragments. EMP extract was dissolved in distilled water at 5 mg/mL to be used for co-
IP. One hundred microliters of porcine EMP solution (500 µg total protein) was incubated in the
Figure 2.1. Photographs of A. a 6 month old porcine jaw, unerupted second molar crown
within its bony crypt shown by white arrow. B. Scraping newly formed enamel with a razor
blade on a clean glass plate.
26

antibody column overnight at 4° C. Unbound protein were removed with washes and elution
fractions analyzed similar to recombinant co-IP.
2.1.6 Mass Spectrometry
Eluted fractions of porcine EMP extract from anti-Amel antibody co-IP column were analyzed by
mass spectrometry. Mass spec experiments were performed by Dr. Dmitry Eremin (Dr. Valery
Fokin’s lab) at the Agilent Center of Excellence in Biomolecular Characterization of University of
Southern California. Mass spectra were measured using Agilent 6545XT qToF instrument
coupled with 1290 LC system. Spectra were recorded with m/z 100–3000 range. For external
calibration and tuning, a low-concentration tuning mix solution by Agilent Technologies was
utilized. For LC separation of protein samples (4 µL injection) PLRP-S 1000Å, 2.1 x 50 mm, 5 µm
column was used with gradient H2O/MeCN elution. All the MS spectra were recorded at 2 Hz.
Spectra were processed using Agilent BioConfirm 10.0 software package.

27

2.2 Results
2.2.1 Characterization of recombinant amelogenin and ameloblastin
Amelogenin and ameloblastin proteins synthesized from E. coli were characterized using SDS-
PAGE and Western blots. The recombinant proteins were also analyzed by mass spectrometry
to determine their purity and to get baseline readings on the mass spec system. Figure 2.2 shows
recombinant porcine and murine amelogenins on an SDS-PAGE gel (inset) and mass spectra for
Figure 2.2. Mass spectra for rM179; Inset: 12% SDS-PAGE gel showing rP172
(lane 1), ladder (lane 2), and rM179 (lane 3).
Figure 2.3. Mass spectra for rAmbn; Inset: 12% SDS-PAGE showing
rAmbn (lanes 1&2).
28

rM179. Figure 2.3 shows characterization of recombinant mouse ameloblastin on SDS-PAGE and
the mass spectra for the same.
Recombinant Ambn contains four additional peptides smaller than the full-length rAmbn, in low
concentration. These are most likely the remnants from the enterokinase digestion of rAmbn,
which were co-eluted with rAmbn from the HPLC reverse phase column.
2.2.2 Characterization of porcine enamel matrix protein extract
As mentioned earlier, rAmel and rAmbn are synthesized without any post-translational
modifications (PTMs). Therefore, porcine enamel matrix extract was used to confirm that protein-
protein interaction between Amel-Ambn occurs in native proteins in the presence of PTMs and
proteolytic cleavage fragments. Porcine enamel matrix protein (EMP) extract was first
characterized by SDS-PAGE and Western blots to identify the Amel and Ambn fragments present
(Figure 2.4). Twelve percent SDS-PAGE (Figure 2.4 A) shows the most abundant fragments of
the porcine extract, with more resolution of 50-20 kDa fragments. Western blot against Ambn
Figure 2.4. Porcine enamel matrix extract characterization. A. 12% SDS-PAGE stained with coomassie
blue showing all the detectable proteins and protein fragments from porcine EMP extract, B. Western blot
against Ambn showing porcine Ambn as 2 bands (arrows), C. Western blot against Amel showing
characteristic porcine Amel 18 k and 20 k bands (arrows).
29

using R&D antibody detected porcine Ambn at ~150 kDa (arrows, Figure 2.4 B). This appearance
of a very high molecular weight band may be due to dimerization of Ambn. Figure 2.4 C shows
porcine Amel fragments detected by Western blot. Characteristic “18k” and “20k” fragments of
porcine Amel can be identified (Figure 2.4 C arrows), along with a smaller fragment ~13 kDa in
size corresponding to the central region of Amel (amino acids 46 to 148).
2.2.3 Co-immunoprecipitation using anti-Amel antibody column
2.3.3a. rAmbn co-eluted with rAmel
Direct binding between rAmel and rAmbn was confirmed by using anti-Amel ab column.
Recombinant Amel was used as bait and rAmbn was used as prey. Recombinant Ambn bound to
rAmel and was identified by Western blots (M300 antibody) in elution fractions (Figure 2.5 A,
Elution fractions 1 to 4). Recombinant Amel in the same elution fractions is shown in Figure 2.6
Western blot (22 kDa bands). A dimer of rAmel can also be observed in elution 1 (band ~45 kDa).
Figure 2.5. Western blot labeled with anti-Amel antibody showing rAmel pulled down with rAmbn (seen
in Figure 2.6) in the elution fractions from an anti-Amel antibody column. Controls 1 to 4 do not contain
any rAmel.
30

A negative control experiment lacking rAmel bait protein was performed to eliminate non-specific
binding. Figure 2.6, Controls 1 to 4 show absence of bait rAmel in controls. However, rAmbn was
observed in the control elution fractions (Figure 2.5 A control elutions 1 to 4). Hence, the rAmbn
bands in Figure 2.5 A were quantified by AzureSoft analysis software to compare the amount of
bound protein between test and control elution fractions. As shown in Figure 2.5 B, the amount of
rAmbn in control elution 3 is almost half that of the corresponding test elution. Test elution 4
continues to show high levels of rAmbn whereas, in control elution 4 protein levels reduce
significantly. This confirmed that although some non-specific binding between rAmbn and the co-
IP resin was observed, the interaction between rAmel and rAmbn was more robust. Quantification
of Western blot bands was performed only in this experiment because of the appearance of rAmbn
in control elution fractions.
After each bait and prey protein incubation, the column was washed 5-6 times. Wash fractions
were analyzed with Western blots to confirm the absence of unbound proteins in the column
Figure 2.6. A. Western blot with anti-Ambn antibody (M300) showing rAmbn pulled down in the
elution fractions from an anti-Amel antibody column. B. Graph quantifying band volume (amount of
protein) in elution and control fractions from A.
31

before elutions. Figure 2.7 A&B show prey (rAmbn) and bait (rAmel) washes analyzed by Western
blots to confirm that wash 4 has minimal of protein.

2.3.3b. Native Amel-Ambn complexes co-elute from whole porcine enamel
protein extract
Using anti-Amel antibody in the co-IP column, Amel and Ambn were pulled down from the whole
porcine enamel matrix protein extract. The porcine EMP extract contains Amel, Ambn, their
proteolytic fragments, alternatively spliced isoforms, as well as other enamel matrix proteins and
proteinases. Fifity five proteins and proteolytic fragments of different masses were identified from
the porcine EMP extract by mass spectrometry (see section 2.3.6). Using anti-Amel antibody in
the co-IP column, Amel-Ambn complexes co-eluted from this mixture. Figure 2.8 A shows porcine
Ambn detected by anti-Ambn (M300) ab and Figure 2.8 B shows porcine Amel detected by anti-
Amel ab from co-IP elution 1 (E1).
The antibody against Ambn (M300) detected three bands (Figure 2.8 A) - one ~85 kDa (asterisk),
and two just under 50 kDa (arrows).
Figure 2.7. A. Western blot labeled with anti-Ambn antibody showing ameloblastin washes; B. Western
blot labeled with anti-Amel antibody showing amelogenin washes. Wash 4 has negligible amounts of
protein in both.
32

It is likely that these bands represent Amel-Ambn complexes. Full-length Ambn complexing with
20 kDa Amel would be ~85 kDa, whereas, 29 kDa fragment of Ambn binding to 18 kDa Amel,
would lead to a 47 kDa band. We plan to investigate these bands in the future with mass
spectrometry to identify the constituents.
Two most abundant Amel bands of “18 k” and “20 k” were observed in the Western blot with anti
Amel antibody (Figure 2.8 B arrows). Below the 18 kDa band, ~13 kDa band is seen which may
represent the 46 to 148 amino acid fragment (central region) of Amel. A higher molecular weight
band of Amel ~45 kDa is also seen in Figure 2.8 B. This is the dimer of Amel often observed in
high concentration preparations of recombinant as well as native SDS-PAGE/ Western blots. A
high molecular weight band at ~85 kDa was observed in the Amel antibody Western blot (Figure
2.8 B) which matches the Ambn band in Figure 2.8 A (both marked by asterisks). These
observations suggest that all of the Amel-Ambn complexes detected were not dissociated during
SDS-PAGE and are being detected by Amel as well as Ambn Western blots.

Figure 2.8. Western blots labeled with anti-Ambn (A) and anti-Amel (B) antibodies showing Amel-Ambn
co-eluted from porcine enamel matrix protein extract from an anti-Amel ab column. Putative Amel-Ambn
complex marked by an asterisk (*).
33

2.2.4 Co-immunoprecipitation using anti-Ambn antibody column
2.3.4a. rAmel co-elutes with rAmbn
Experimental setup was reversed to confirm the findings of the co-IP. Anti-ameloblastin ab (R&D)
was bound to the co-IP column and rAmbn was used as a bait protein to determine if rAmel could
be co-eluted as a prey. Results are shown in Figures 2.9 and 2.10. Recombinant Amel was pulled
down with rAmbn from anti-Ambn ab column. For these experiments, control resin provided in the
co-IP kit was used. This resin lacks active binding sites for antibody coupling. Control experiments
were performed with the equal quantities of antibodies and proteins loaded in the control column
as the test column. Figure 2.9 shows rAmel (22 kDa) in control and test elution fractions detected
by Western blots. The amount of rAmel in the control elution fractions (Cntrl E1, Cntrl E2) is
significantly lower than the amount of rAmel in test elution fractions (E1 & E2), suggesting strong
specific binding of Amel to Ambn.
Figure 2.9. Western blot labeled with anti-Amel antibody showing rAmel pulled down with rAmbn (seen
in Figure 2.10), from an anti-Ambn ab column. There is significantly less rAmel detected in control
fractions (Cntrl E1 and Cntrl E2).
34

Presence of rAmbn in anti-Ambn ab column co-IP elutions (E1 & E2) is shown in Figure 2.10 anti-
Ambn Western blot (R&D). Recombinant Ambn was virtually absent in control elution fractions
(Cntrl E1, Cntrl E2). No cross-reactivity was observed between anti-Ambn antibody (R&D) and
rAmel (Figure 2.12 A) eliminating the possibility of false positive results.
2.3.4b. Porcine Amel-Ambn complexes co-elute from porcine enamel
protein extract
Similar to the recombinant co-IP experiments, porcine extract co-IP was repeated with anti-Ambn
ab column so that Ambn would act as a bait for Amel prey. Amel and Ambn proteins co-eluted
from porcine enamel matrix protein extract from the anti-Ambn ab column are shown in Figure
2.11. Figure 2.11 A shows porcine Ambn detected in elution E1 at ~150 kDa. This is where the
M300 and R&D antibodies differ in detecting porcine Ambn as mentioned earlier. In this Western
blot labeled with anti-Ambn R&D ab, porcine Ambn appears as a single band at ~150 kDa. This
very high mass may be due to dimerization of Ambn but must be confirmed by mass spectrometry.
Proteolytic cleavage products of Ambn were not detected by this antibody. Figure 2.11 B shows
the most abundant fragments of porcine Amel, the “18k” and “20k” detected from the same elution.
Figure 2.10. Western blot labeled with anti-Ambn antibody (R&D) showing rAmbn pulled down with rAmel
(seen in Figure 2.9) in elution fractions from an anti-Ambn ab column. Amount of protein in control
fractions is insignificant (Cntrl E1 and Cntrl E2).
35

It is interesting to note that the 13 kDa central region of Amel was not identified in this elution but
only the 2 largest proteolytic cleavage products of Amel were detected. This could be because
Ambn binds at the N-terminus of Amel within the tyrosine rich Amel polypeptide domain, which is
present in the “18k” and “20k” fragments but absent in the 13 kDa fragment.
These results, taken together with the anti-Amel antibody column results presented earlier, clearly
demonstrate that recombinant and native porcine amelogenin and ameloblastin bind to each other
in vitro.
Figure 2.11. Western blots labeled with anti-Ambn (A) and anti-Amel (B) antibodies showing native Ambn
and Amel co-eluted from porcine enamel matrix protein extract from an anti-Ambn (R&D) ab column.
36

2.2.5 Antibody cross-reactivity controls
An important caveat of co-immunoprecipitation is that, non-specific binding of prey proteins to
antibodies in the column can lead to false positive results. Figure 2.12 shows cross-reactivity test
for anti-Amel and anti-Ambn antibodies. Western blot in Figure 2.12 A is labeled with anti-Amel
antibody (R&D) which shows rAmbn band in lane 1. No cross-reactivity between anti-Ambn
antibody and rAmel in lane 2. Similarly, anti-Amel antibody did not cross-react with rAmbn (Figure
2.12 B, lane 1). Recombinant Amel was detected by anti-Amel antibody as the full-length (~22
kDa) and dimer at ~45 kDa (Figure 2.12 B, lane 2). These controls eliminate the chances of false
positive results from co-immunoprecipitation.
2.2.6 Analysis of porcine EMP elution by mass spectrometry
Absence of proteolytic cleavage products in Western blot of porcine EMP could either be because
of the lack of antibody bindning to those small proteolytic fragments or due to low abundance of
these fragments. To overcome this limitation, elutions from anti-Amel ab column co-IP were
analyzed using mass spectrometry. Complete porcine enamel matrix extract was analyzed first
Figure 2.12. A. Western blot labeled with Anti-Ambn antibody (R&D) showing no cross-reactivity with
rAmel (lane2); B. Western blot labeled with anti-Amel antibody, showing no cross-reactivity towards rAmbn
(lane 1); C. 12% SDS-PAGE to show loading controls in each blot.
37

as a baseline and 55 different EMP fragments of masses ranging from 4 kDa to 19 kDa were
detected (data not shown). Ten fragments out of 55 (listed in Table 2.3) were bound to the anti-
Amel co-IP column. Four fragments were detected only in the co-IP elution, and not in the whole
porcine EMP extract (marked by † in Table 2.3). It is likely that these four fragments were
concentrated from the porcine EMP extract by binding to Amel, while they were too low in
concentration to be detected in the whole porcine EMP extract. From the list of masses, Amel
fragments were identified using previously published literature. All the detected fragments from
co-IP elution and identifed Amel fragments are listed in Table 2.3. Direct identification of Ambn
fragments was not possible from these data because the exact in vivo masses of porcine Ambn
proteolytic products have not been clarified in literature.
# Protein fragments Mass (Da)
1
Unidentified
† 4525.1
2 5160.4
3 5175.4
4 † 5176.4
5 Dephosphorylated TRAP 5323.5
6 Lucine rich amelogenin polypeptide (LRAP) 5343.7
7 Tyrosine rich amelogenin polypeptide (TRAP) 5403.5
8
Unidentified
† 7334.5
9 † 7405.5
10 16830.4
11
Dephosphorylated P148 16838.4
12
Phosphorylated P148 16918.4
13 Maybe P147 16933.4
14 Unidentified 18510.3
Table 2.3. Fragments identified by mass spectrometry from porcine co-IP elution from Amel
column. † marks fragment novel to elution, not present in whole porcine extract.
Note: The protocol for rAmbn expression and purification was published as a book chapter: Su,
J., Bapat, R.A., and Moradian-Oldak, J. (2019a). The expression and purification of recombinant
mouse ameloblastin in E. coli. In Odontogenesis (Springer), pp. 229-236.
38

2.3 Discussion
In this chapter, direct binding between recombinant and native amelogenin and ameloblastin
proteins was demonstrated (Figure 2.13 and Table 2.4). Results from this Chapter are
summarized below.
Anti-Amel antibody co-IP column
Amel bait and Ambn prey
Anti-Ambn antibody co-IP column
Ambn bait and Amel prey
Amel fragments
Figure 2.8 B
Ambn fragments
(M300), Figure 2.8 A
Amel fragments
Figure 2.11 B
Ambn fragments
(R&D), Figure 2.11 A
~85 kDa ~85 kDa 20 kDa ~150 kDa
~45 kDa ~47 kDa 18 kDa
20 kDa ~44 kDa
18 kDa
13 kDa
Table 2.4. Summary of Western blot bands representing porcine Amel and Ambn fragments detected from
co-IP elution fractions by Western Blot

Co-IP experiments were repeated using both anti-Amel and anti-Ambn antibodies in the co-IP
columns, which made the data more robust and allowed identification of Amel-Amel or Ambn-
Ambn complexes. In Figure 2.6 (elution fraction E1) and Figure 2.9 (elution fractions E1 and E2),
Figure 2.13. Summary of results for recombinant and native Amel and Ambn co-immunoprecipitation.
Western blots showing Amel and Ambn labeled in A. elution fractions of anti-Amel antibody co-IP
column, B. elution fractions of anti-Ambn antibody co-IP column.

39

a dimer of rAmel was detected. Similar phenomenon was also seen when porcine Amel-Ambn
complex was eluted using anti-Amel column (Figure 2.8 B). It is not possible to differentiate
whether Amel self-assemblies are being detected from the porcine extract or free Amel in solution
is complexing with antibody bound Amel in the column. Though in both scenarios, it can be
concluded that Amel-Amel binding or Amel self-assembly is strong and cannot be easily
dissociated despite heating and using SDS in the loading buffer to denature the proteins.
Results from porcine enamel matrix protein extract co-IP are summarized in Table 2.4. Native
porcine ameloblastin has been characterized in literature to be around 65 kDa in SDS-PAGE.
However, in these experiments Ambn was consistently observed to be running higher than
anticipated. In Figure 2.8 A, three bands labeled with M300 Ambn antibody were detected in the
western Blot of porcine Ambn fraction eluted from anti-Amel co-IP column. One darker band
around 85 kDa and 2 lighter bands, one just under 50 kDa and another ~44 kDa. Bands were
consistent across all 3 experiments repeated. In the case of amelogenin, multiple fragments were
identified in the elution fraction of anti-Amel antibody column which are listen in Table 2.4 (Figure
2.8 B). These fragments may have directly bound to the anti-Amel antibody in the column or
assembled with Amel bound to the column. A band, possibly representing Amel-Ambn complex,
was detected at ~85 kDa by both anti-Amel and anti-Ambn Western blots in porcine elution
fraction (Figure 2.8, asterisks). Confirmation of these results by mass spectrometry will provide
definitive identification of the constituents of this band.
When anti-Ambn antibody (R&D) was used in the co-IP column, only 2 Amel fragments were
detcted (Figure 2.11 B). Both these fragments contain the N-terminal tyrosine rich amelogenin
polypeptide and the Amel self-assembly motif (Paine et al., 1998; Wald et al., 2017). Thirteen kilo
dalton Amel fragment (amino acids 46 to 148) lacking this domain was not pulled down with Ambn.
This supports the idea that this domain may be essential for Amel’s interaction with Ambn (Su et
al., 2016). In Figure 2.11 A (anti-Ambn antibody column co-IP) Ambn band was detected as high
40

as 150 kDa by the R&D antibody. There was a concern that this might be the co-IP antibody
eluting from the column along with bound proteins. However, this band matched the Ambn
fragments detected by R&D antibody from whole porcine extract in Figure 2.4, dismissing the
concern. The ~150 kDa Ambn band does not match those stated in literature for porcine Ambn.
We could be detecting Ambn self-assemblies (dimer would be ~130 kDa), or Amel-Ambn
complexes (~105 kDa if one Ambn binds to an Amel dimer). It is important to note that although,
sites of glycosylation on Ambn have been identified (Kobayashi et al., 2007) how many
glycosylations are present on each modified amino acid, and how much mass do they exactly
contribute to the final protein is still unknown. Due to the hydrophobic, intrinsically disordered and
aggregative nature of enamel matrix proteins, they always run higher on SDS-PAGE than their
actual mass even under denaturing conditions. The best way to confirm these findings would be
by mass spectrometry.
The mass spectrometry system that is available to us within the University does not have the tools
to identify unknown protein fragmnets from an open database. The system can generate a report
by comparing the detected masses to a given protein sequence. Mass spectrometry detected 55
individual fragments in the porcine EMP extract. Upon co-immunoprecipitation, 14 fragments were
identifeid in the elution, out of which 4 were unique to the co-IP elution only and not present in the
whole porcine EMP extract (marked by † in Table 2.3). It is likely that the co-IP concentrated these
very low abundance fragments from the whole porcine extract. Amel proteolytic fragments and
alternatively spliced products are highly characterized in literature (Fincham et al., 1994; Fincham
and Moradian-Oldak, 1993; Yamakoshi et al., 1994). Based on that, masses of P148, TRAP and
LRAP (both phosphorylated and dephosphorylated) were manually matched to mass spectra
generated for the porcine enamel matrix co-IP elution (Table 2.3). Matching Ambn fragments from
the same elution is more challenging. The sequences of Ambn proteolytic cleavage products are
known, but their in vivo masses have not been yet clarified, and the mass each glycosylation adds
41

to Ambn is yet somewhat ambiguous. Careful analysis of the mass spectra is required to identify
Ambn fragments based on N-terminal sequencing data in literature (Yamakoshi et al., 2001) which
is something we plan to do in the future.
Co-immunoprecipitation of recombinant as well as native amelogenin and ameloblastin has
shown that this interaction does not require nor is hampered by post-translational modifications.
Ravindranath et al., (2004; 1999) suggested that Amel interacts with Ambn through its trityrosil
motif, which acts as a lectin-like domain to bind to glycosylations on Ambn. Results with
recombinant Ambn bindng to rAmel suggest that the lectin like domain of Amel may not be the
sole mechanism of Amel-Ambn interaction. The trytyrosil motif is present within the A-domain of
Amel which is essential for Amel self-assembly (Paine et al., 1998). The same trytyrosil region
has also been identified as the self-assembly motif ‘Y/F-x-x-Y/L/F-x-Y/F’ in intrinsically disordered
proteins by Wald et al (2017). This region is a part of tyrosin rich amelogenin polypeptide or TRAP.
Previous work from our lab shows that TRAP interacts with a synthetic peptide encoding exon 5
of ameloblastin in the absence of post-translational modifications in vitro, resulting in structural
changes in TRAP (Su et al., 2016). Thus this trytyrosil region maybe functioning as a co-assembly
domain for Amel-Ambn binding, similar to its sel-assembly function. The lectin-like carbohydrate
binding properties of this region may work as an adjunct mechanism for Amel-Ambn interaction.

42

3. Investigation of the binding region of Ambn in
Amel-Ambn co-assembly
In Chapter 2, I concluded that recombinant and native amelogenin and ameloblastin interact with
each other directly in vitro. In the first part of Chapter 3, I will first investigate the region of
ameloblastin (Ambn) essential for binding to amelogenin (Amel). I will utilize two Ambn mutants
and various Ambn synthetic peptides along with recombinant Amel (rAmel) to determine which
region of Ambn is interacting directly with Amel. In the second part of this Chapter, I will study the
co-assembly of rAmel and recombinant Ambn (rAmbn) using dynamic light scattering (DLS) and
transmission electron microscopy (TEM) to understand how Ambn affects the assembly of Amel
nanospheres.
Part I
3.1 Background
Ameloblastin is an intrinsically disordered enamel matrix protein with functions involving
maintenance of ameloblasts (Fukumoto et al., 2004; 2005) and enamel biomineralization (Paine
et al., 2003). The importance of the exon 5 encoded region of Ambn has been particularly shown
in the recent years. In the Ambn exon 5-6 deletion mutant mouse (Ambn
-5,6/-5,6
) ameloblasts lacked
the characteristic tall columnar morphology and separated from underlying dentin (Fukumoto et
al., 2004; 2005). Amel synthesis was reduced in this mouse model at mRNA and protein levels
but the synthesis of other enamel matrix proteins remained unaffected. Initially, this model was
identified as Ambn knockout but it was later discovered that a truncated form of Ambn lacking
exons 5 and 6 encoded regions continued to be secreted in this mutant mouse (Wazen et al.,
43

2009). Ameloblastin was non-functional in its truncated form which highlighted the importance of
exons 5 and 6 in Ambn function. The effect of Ambn exon 5-6 deletion on Amel synthesis
suggested that Ambn may be interacting with Amel through the exon 5-6 encoded region.
The significance of Ambn exon 5 encoded peptide was underlined when Wald et al. (2013)
identified and characterized a self-assembly motif present at the N-terminus of this region. The
motif Y/F-x-x-Y/L/F-x-Y/F is essential for Ambn self-assembly and was later identified within the
self-assembly domain of Amel (Paine et al., 2000; Wald et al., 2017). Possible interaction between
rAmel and exon 5 encoded region of Ambn was suggested through biophysical experiments
conducted by Su et al. (2016). Using circular dichroism and fluorescence spectroscopy Su et al.
(2016) showed that a synthetic peptide encoding exon 5 of Ambn (named AB2, see materials and
methods) promotes a more alpha helical structure in rAmel. This structural change was also
observed in the N-terminal Amel proteolytic product TRAP (tyrosine rich amelogenin polypeptide)
upon mixing with AB2. The TRAP region contains two Y/F-x-x-Y/L/F-x-Y/F self-assembly domains
(Wald et al., 2017). Taken together with data from Wald et al. (2017) this suggests that Amel and
Ambn may be interacting via their respective self-assembly motifs.
Therefore, in Part I of this Chapter, I hypothesize that Ambn interacts with Amel through the N-
terminus of its exon 5 encoded region. I investigated the hypothesis by performing co-IP
experiments between Ambn mutants or Ambn synthetic peptides and rAmel. Two recombinant
Ambn mutants each with deletion of either exon 5 or exon 6 were used as prey proteins in anti-
Amel antibody column with full-length recombinant Amel as bait. The results were confirmed using
co-IP between Ambn synthetic peptides encoding exon 5 or exon 6 regions as prey proteins and
rAmel as bait protein. Finally, peptides encoding the N-terminal and C-terminal halves of exon 5
were used as prey in the co-IP to specifically identify the Amel-binding region of Ambn.
44

3.2 Materials and methods
3.2.1 Expression and purification of recombinant proteins

Two Ambn mutants, one with deletion of exon 5 (Figure 3.1, in purple) and other with deletion of
exon 6 (Figure 3.1, in pink), designated as AmbnΔ5 and AmbnΔ6 respectively, were designed by
Dr. Jingtan Su (Su et al., 2019b). These proteins were expressed and purified in BL21 E. coli
similar to full-length recombinant Ambn described in Chapter 2 (Su et al., 2019a). Briefly, AmbnΔ5
and AmbnΔ6 were expressed with a Thioredoxin tag, S-tag, and His-tags and initially purified by
Ni-NTA agarose columns (QIAgen). Proteins were dialyzed against ice cold phosphate buffer (pH
7.4) and the tags were cleaved using enzyme enterokinase (NEB Inc.). The proteins were purified
from their cleavage products using reverse phase HPLC.
Recombinant porcine amelogenin (rP172) was purified as described in Chapter 2 (Ryu et al.,
1999).
3.2.2 Ambn synthetic peptides

Figure 3.1. Mouse ameloblastin sequence showing the region deleted in AmbnΔ5 (exon 5 encoded
region) in purple and the region deleted in AmbnΔ6 (exon 6 encoded region) in pink. Signal peptide
is in grey.
Figure 3.2. Cartoon showing sequences of Ambn synthetic peptides; AB2 (encoded by exon 5) -
underlined, AB2N- blue, AB2C- purple, and AB4 (encoded by exon 6) - pink.
45

Encoding region Peptide Mass (Da)
Exon 5 AB2 4284.85
Exon 5 N-terminus AB2N 2242.65
Exon 5 C-terminus AB2C 2060.22
Exon 6 AB4 6840.73
Table 3.1. Masses of Ambn synthetic peptides
Ameloblastin synthetic peptides were ordered from Chempeptide Limited (Shanghai, China) (Su
et al., 2016). The peptides are listed in Table 3.1 and their sequences are shown in Figure 3.2.
The peptides were designed to represent exon 5, N-terminus of exon 5, C-terminus of exon 5,
and exon 6 encoded regions of Ambn in order to confirm the findings of co-IP experiments
performed using Ambn mutants. The peptides were analyzed by reverse phase HPLC using an
analytical column (C8, Vydac) to confirm their purity (data not shown).
3.2.3 Co-immunoprecipitation
Co-IP was performed using the protocol described in Chapter 2. Briefly, 10 µg anti-Amel antibody
was bound to the co-IP resin column. Recombinant Amel was used as bait protein. To analyze
whether AmbnΔ5 and Ambn Δ6 can interact with rAmel, the mutant proteins were used as prey.
To further confirm the findings, Ambn synthetic peptides (Table 3.1) were used as prey and the
co-IP was repeated. The elution fractions were lyophilized and dissolved in 20 µL distilled water.
Scheme 3.1. Protocol for co-IP between rAmel (bait) and mutant AmbnΔ5 or Ambn
Δ6 or Ambn peptides (prey).
46

Fractions were analyzed by SD-PAGE and Western blots. Scheme 3.1 describes the co-IP
workflow used in this Chapter.
3.2.4 SDS-PAGE and Western blots
Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blots were
performed as described in Chapter 2. Proteins were blotted to PVDF membranes using wet
blotting method. AmbnΔ5 and AmbnΔ6 were visualized using Western blots labeled with anti-
Ambn M300 antibody (1:500, Santa Cruz Biotech, now discontinued). Anti-Amel antibody from
Dr. M. Snead’s lab (1:1000) was used for Western blots against rAmel. Elution fractions from co-
IP experiments were lyophilized and dissolved in 20 µL of distilled water. To achieve detectable
levels of protein in the Western blots, all 20 µL of the elution fraction were applied to one SDS-
PAGE gel. The PVDF membrane obtained from this gel was labeled in tandem with anti-Ambn
and anti-Amel antibodies to visualize the proteins present in the elution.
Due to lack of antibodies against Ambn synthetic peptides, silver stained 16% polyacrylamide
gels were used to visualize the peptides from co-IP elution fractions. The high percentage of
polyacrylamide in the gels was necessary to resolve small molecular weight Ambn synthetic
peptides (Table 3.1). Gels were stained with silver staining using standard protocol (Switzer III et
al., 1979).
3.3 Results
3.3.1 Characterization of AmbnΔ5 and AmbnΔ6
Mutants AmbnΔ5 and AmbnΔ6 were characterized by SDS-PAGE and Western blots. Figure 3.3
A&B are 12% Tris-glycine gels stained with Coomassie blue stain adapted from supplemental
data from Su et al. (2019b). Figure 3.3 A shows AmbnΔ5 and AmbnΔ6 bands at slightly lower
47

than 50 kDa and ~40 kDa respectively. Figure 3.3 B shows full-length rAmbn band at ~51 kDa for
comparison. Figure 3.3 C is a Western blot of AmbnΔ5 and AmbnΔ6 labeled with anti-Ambn M300
antibody. The masses denoted in the Western blot are the molecular weights of AmbnΔ5
(36547.32 Da) and AmbnΔ6 (33991.45 Da) calculated from ExPASy ProtParam web-based tool
(Gasteiger et al., 2005). The proteins tend to run higher on SDS-PAGE gels than their calculated
weights due to their intrinsically disordered nature.

Figure 3.3. 12% SDS gels stained by coomassie blue showing, A. AmbnΔ5 and AmbnΔ6. Modified from
supplemental data, Su et al. (2019b); and B. full-length rAmbn. C. Western blot labeled by anti-Ambn M300
antibody showing AmbnΔ5 and AmbnΔ6. Calculated molecular weights of each are mentioned on the
Western blot.
48

3.3.2 Anti-Amel and anti-Ambn M300 antibody cross-reactivity
To analyze the cross-reactivity between anti-Amel antibody and Ambn mutants, Western blot
containing rAmel, AmbnΔ5, and AmbnΔ6 was labeled with anti-Amel antibody (Figure 3.4 A).
Lane 2 containing AmbnΔ5 showed a faint band in this Western blot. When co-IP was performed
using AmbnΔ5 as prey, it did not bind to anti-Amel antibody column (see below). Therefore, if this
band represents non-specific binding of anti-Amel antibody to AmbnΔ5, the reactivity is very low
and only observed when very high concentration of AmbnΔ5 are present. Anti-Amel antibody did
not show any cross-reactivity towards AmbnΔ6 (Figure 3.4 A, lane 3). Figure 3.4 B shows Western
blot labeled with anti-Ambn M300 antibody. It does not show any cross-reactivity towards rAmel
present in lane 1. Full-length rAmbn is detected in lane 2.
Figure 3.4. A. Western blot labeled with anti-Amel antibody shows no cross-reactivity towards AmbnΔ6 but
a faint band is observed in the lane with AmbnΔ5. B. Western blot labeled with anti-Ambn M300 antibody
showing that it does not cross-react with rAmel.

49

3.3.3 AmbnΔ5 mutant fails to bind to rAmel in co-IP
Figure 3.5 shows results for co-IP performed using full-length rAmbn, AmbnΔ5, and AmbnΔ6 as
prey proteins in anti-Amel antibody column. Recombinant Amel was used as bait. The Western
blot membrane was first labeled with anti-Ambn M300 antibody, images were taken with ECL
Chemiluminescence reagents to identify Ambn bands and the same membrane was relabeled
with anti-Amel antibody. Figure 3.5 lane 1 shows full-length rAmbn and rAmel co-eluting in the
elution fraction. This is consistent with the findings from Chapter 2 and was treated as a positive
control for the experimental set-up. Figure 3.5 lane 2 shows AmbnΔ6 co-eluting with rAmel from
the anti-Amel co-IP column. This showed that Ambn lacking the exon 6 encoded fragment
retained its ability to bind to rAmel. Figure 3.5 lane 3 shows only rAmel in the elution fraction,
AmbnΔ5 was not detected in the Western blot, meaning that Ambn lacking exon 5 encoded region
cannot interact with Amel. This identified the exon 5 encoded region of Ambn as being essential
for its binding to Amel.
Figure 3.5. Western blot of co-IP elution fractions labeled in tandem with anti-Ambn M300 and anti-Amel
antibodies. Lane 1- full-length rAmbn co-eluting with rAmel (positive control); Lane 2- AmbnΔ6 co-eluting
with rAmel; Lane 3- AmbnΔ5 does not bind to rAmel, and only rAmel was identified in the elution; Lane 4-
standard protein ladder; Lane 5- full-length rAmbn control, Lane 6- rAmel control
50

3.3.4 N-terminal fragment of Ambn peptide encoded by exon 5 binds directly to
Amel
To confirm that Ambn exon 5 encoded region can directly bind to rAmel, co-IP was repeated by
using Ambn synthetic peptides as prey. Experimental set-up was the same as above. The column
contained anti-Amel antibody and rAmel was used as bait protein. Elution fractions of these
experiments were analyzed by silver stained 16% SDS-PAGE gel. When Ambn peptide encoding
exon 5 (AB2) was used as prey, AB2 co-eluted with rAmel from the co-IP column (Figure 3.6, lane
2). This confirmed the findings of AmbnΔ5, Δ6 co-IP shown in Figure 3.5. The results presented
in Figure 3.6 also confirmed that the lack of binding between AmbnΔ5 and rAmel is not due to
misfolding of Ambn lacking exon 5 but indeed due to the absence of the binding region on Ambn.
Co-IP was repeated by using the Ambn peptide AB4 encoding exon 6. AB4 did not bind to rAmel
Figure 3.6. 16% SDS-PAGE gel stained by silver staining showing elution fractions of co-IP
experiments. Ambn peptides were used as prey and rAmel as bait. Lane 1- exon 6 AB4 does not bind
to rAmel. Lane 2- exon 5 AB2 binds to rAmel and co-elutes. Lane 3- N-terminus of exon 5 AB2N binds
to rAmel and co-elutes. Lane 4- C-terminus of exon 5 AB2C does not bind to rAmel. Lane 5- rAmel
control. Lane 6- standard protein ladder. Lane 7, 8, and 9- AB2, AB2N and AB2C controls respectively.
Lane 10- Low molecular weight ladder.
51

and only rAmel was observed in the elution fraction (Figure 3.6, lane 1) concluding that the exon
6 encoded region of Ambn cannot bind to rAmel. The N- and C- terminal fragments of exon 5
encoded region (peptides AB2N and AB2C respectively) were used as prey in order to further
narrow down the binding region within AB2. It was found that the peptide AB2N or the N-terminus
of AB2 binds to rAmel (Figure 3.6, lane 3) but peptide AB2C (C-terminus of AB2) cannot bind to
rAmel (Figure 3.6, lane 4). This narrowed the Amel-binding domain of Ambn within 20 amino acids
at the N-terminus of exon 5 encoded peptide. This region contains the Y/F-x-x-Y/L/F-x-Y/F self-
assembly domain of Ambn (Wald et al., 2017). Taking into consideration that AB2 affects the
secondary structure of TRAP in vitro (Su et al., 2016) which contains Amel self-assembly
domains, we conclude that, Amel-Ambn self-assembly domains are also co-assembly domains
and are responsible for their interaction with each other.
In order to address the co-assembly of Amel-Ambn, in Part II of Chapter 3, I analyzed the effect
of Ambn synthetic peptide AB2 and full-length rAmbn on rAmel nanosphere assembly.

52

Part II
3.4 Background
Amelogenin undergoes stepwise self-assembly starting from intrinsically disordered monomers
(~3.5-4.5 nm diameter), to metastable oligomers (~7.5 nm diameter), and finally nanospheres of
~20 nm diameter at or above pH 7.4 (physiological pH) (Bromley et al., 2011; Du et al., 2005). In
vivo, material resembling globular structures, periodic helical material or stippled material
(Smales, 1975; Travis and Glimcher, 1964; Watson, 1960) has been observed in developing
enamel since the 1960s. After the discovery of Amel nanospheres, it was clear that this material
contained Amel.
Full-length Amel is relatively short-lived within developing enamel. Its C-terminus is cleaved soon
upon secretion by MMP-20 (Bartlett and Simmer, 1999). Therefore, different Amel proteolytic
cleavage products were studied to understand how nanospheres formed from these fragments
may be functioning in vivo. Nanospheres of native full-length Amel (P173) containing hydrophilic
C-terminus function to separate initial mineral ribbons and prevent their early fusion (Moradian-
Oldak et al., 2003). Upon cleavage of the 12 C-terminal, hydrophilic amino acids (resulting in
P161), Amel assembled into larger diameter nanospheres (~20-28 nm) due to fusion of the
particles from hydrophobic interactions. Further cleavage of C-terminal region involving large
hydrophobic residues gives rise to P148, which is the most abundant Amel fragment in vivo. P148
nanospheres were smaller, about 10-14 nm in diameter (Moradian-Oldak et al., 2003). The apatite
binding potential of P161 and P148 was significantly lower than the full length protein due to lack
of C-terminus (Moradian-Oldak et al., 2002). In order to bind to the lateral surfaces of developing
hydroxyapatite (HAP) crystals, P161 and P148 native Amel nanospheres may be cooperating
with non-amelogenin proteins. Non-amelogenins like enamelin and ameloblastin as well as their
53

proteolytic fragments have a strong affinity towards calcium (Yamakoshi et al., 2001). This
suggests that the in vivo globular structures of organic material may be co-assemblies of Amel
and non-amelogenins enamelin or Ambn.
For the purpose of this project, I will be focusing on Amel-Ambn co-assembly. I have
demonstrated direct Amel-Ambn binding in vitro in Chapter 2 and identified the region of Ambn
binding to Amel in Part I of this Chapter. Amel-Ambn co-secretion (Zalzal et al., 2008) and
colocalization (Mazumder et al., 2016; 2014) in developing enamel further support the idea that
Amel-Ambn co-assemble in vivo. To investigate this, I used the Ambn peptide AB2 containing the
Amel-binding domain and full-length rAmbn mixed with rAmel and analyzed the changes in
particle size distribution by dynamic light scattering (DLS) and transmission electron microscopy
(TEM).
These techniques have been used and well established over the years to study the self-assembly
of Amel nanospheres (Bromley et al., 2011; Fincham et al., 1995; Moradian‐Oldak et al., 1994).
Dynamic light scattering (DLS) uses the principles of Tyndall effect and Brownian motion wherein
constantly moving particles in a solution scatter incident light. The scattered light can be collected
by a detector to calculate the sizes of the particles in the solution (Hassan et al., 2015). The output
is given as mass percent or intensity percent of hydrodynamic radii (R H) of different particles in
the solution. The measurements in DLS are based on the Rayleigh approximation which makes
the assumption that the particles are spherical in shape in order to calculate the correlation
between intensity of scattered light and particle size (Stetefeld et al., 2016). Therefore, to visualize
the true nature of rAmel-AB2 and rAmel-rAmbn particles, transmission electron microscope (TEM)
was used. Samples from DLS were directly transferred to TEM grids and negatively stained in
order to confirm the DLS data.
54

3.5 Materials and methods
3.5.1 Recombinant proteins and peptides
Recombinant mouse Amel (rM179) was used for these experiments. Amel was expressed in BL21
E. coli, precipitated with 20% ammonium sulfate and purified by reversed phase HPLC (Ryu et
al., 1999). The protocol is described in detail and rAmel has been characterized in Chapter 2.
Ambn peptide AB2 mentioned in Part I of this Chapter was used for initial DLS and TEM
experiments. AB2 is the exon 5 encoded region of Ambn which harbors the Amel-binding domain
at its N-terminus. DLS experiments were conducted at room temperature and AB2 is more stable
than full-length Ambn at room temperature. Hence, AB2 was used for optimization of DLS and
TEM protocols.
Recombinant mouse Ambn (rAmbn) was expressed and purified from BL21 E. coli and the
detailed protocol is described in Chapter 2 (Su et al., 2019a).
3.5.2 Obtaining rAmel nanospheres
To obtain Amel nanospheres, protocol published by Bromley et al. (2011) was followed with some
modifications. First, lyophilized rAmel was dissolved in distilled deionized (dd) water at 1 mg/ml.
The solution was mixed with gentle shaking for 24 h at 4° C to assure complete dissolution of
rAmel in water. Twenty five millimolar Tris-HCl buffer was prepared in dd water and its pH was
adjusted to 7.6 at room temperature (22° C). The buffer was autoclaved and small quantities (~1
ml) of buffer were filtered with 0.22 µm PVDF syringe filters (Millex-GV, Millipore Sigma) before
each use. Recombinant Amel solution was diluted in filtered 25 mM Tris-HCl to achieve final
concentration of 0.3 mg/mL. The solution was mixed with gentle shaking for 2 h at 4° C to assure
complete dissolution of rAmel in the buffer. Final pH of rAmel-Tris-HCl solution was 7.4-7.6
55

measured at room temperature. Solution was centrifuged at 10,000 RPM for 10 minutes at room
temperature to remove large aggregates before being analyzed by DLS and TEM.
3.5.3 Co-assembly of AB2-rAmel and rAmbn-rAmel
Ambn synthetic peptide AB2 or rAmbn were dissolved in dd water at 1 mg/mL and stirred for 24
h at 4° C to dissolve completely. For determining the co-assembly between AB2 and rAmel, the
protocol for obtaining rAmel nanospheres was modified by addition of 10 µL of 1mg/mL AB2. Final
concentrations were 0.1 mg/mL AB2 and 0.3 mg/mL rAmel in 25 mM Tris-HCl to at pH 7.6, to
achieve AB2: rAmel ratio of 1:3 by weight. Due to aggregative nature of rAmbn, the concentrations
were halved for the analysis of rAmbn-rAmel co-assembly. Five microliters of rAmbn and 15 µL
rAmel (each 1 mg/mL in dd water) were added to 80 µL, 25 mM Tris-HCl to achieve final
concentrations 0.05 mg/mL rAmbn and 0.15 mg/mL rAmel. The ratio between rAmbn:rAmel
remained 1:3 by weight. The final pH of the solutions was between 7.4-7.6 measured at room
temperature. AB2-rAmel or rAmbn-rAmel solutions were centrifuged at 15,000 RPM for 15
minutes at 4° C to remove aggregates prior to analysis by DLS or sample preparation for TEM.
3.5.4 Dynamic light scattering
Dynamic light scattering (DLS) experiments were conducted using a Wyatt DynaPro Nanostar
DLS instrument (Wyatt Technology, Santa Barbara, CA). Dynamics 7.0 software was used to
analyze and plot the data. Twenty microliters of rAmel or AB2/rAmbn-rAmel solution in 25 mM
Tris-HCl were applied to a DLS flow cell using specialized extended length pipette tips. The cell
was cleaned thoroughly with 1 % Tergazyme, deionized water and ethanol and air dried before
and after each experiment. Twenty microliters of 25mM Tris-HCl buffer without proteins was used
to detect the baseline hydrodynamic radii (RH) to identify background values. All DLS experiments
were conducted at room temperature of 22° C. The software uses Stokes-Einstein equation to
56

convert the light scattered by the particles collected by the detector into R H. Principles of Rayleigh
sphere model were utilized during these calculations with the underlying assumption that the
particles in solution are spherical (Stetefeld et al., 2016). The software obtained a set of 10
readings 10 times thereby, 100 readings were acquired for each experiment. Data were plotted
using mass percent of particles on the Y axis against R H on the X axis. The software also
calculated polydispersity values which can be described as the standard deviation or the range
of size distribution of particles in the solution (Moradian‐Oldak et al., 1994). Any sample with
polydispersity higher than 20% was considered heterogeneous in nature.
3.5.5 Transmission electron microscopy
Protein or
protein mixture
Protein concentration
Volume added to
TEM grids
rAmel 0.3 mg/mL 10 µL
AB2-rAmel
0.4 mg/mL (0.1 mg/mL AB2 + 0.3 mg/mL
rAmel)
7.5 µL
rAmbn 0.04 mg/mL 10 µL
rAmbn-rAmel
0.2 mg/mL (0.05 mg/mL rAmbn + 0.15 mg/mL
rAmel)
10 µL
Table 3.2. The table lists protein concentrations for each TEM experiment, and the volume of protein added
to the TEM grid
Table 3.2 lists the proteins and protein mixtures used in TEM experiments and the concentrations
used for each. Droplets of volumes specified in Table 3.2 were placed on the TEM grids (400
mesh carbon-coated grids, Ted Pella Inc.) for 1 minute and excess sample was absorbed on a
filter paper. Grids were stained with 2% uranyl acetate for 2 minutes, rinsed with water for 30 s
and air dried for 5 minutes. Negative control sample was prepared with 10 µL, 25 mM Tris-HCl
buffer only, stained as above. Grids were imaged on JEOL JEM-2100F transmission electron
microscope operating at an accelerating voltage of 120 kV (USC Core Center of Excellence in
57

Nano Imaging) or 80 kV (USC/Norris Cancer Center Cell and Tissue Imaging Core). Images were
analyzed using Gatan Digital Micrograph software version 3.31.2360.0 and ImageJ version 1.52a.
3.6 Results
3.6.1 Buffer control observed by DLS
In order to establish the baseline values obtained by DLS and TEM from the buffer, 25 mM Tris-
HCl was analyzed first without any protein. The buffer was treated exactly the same as the
samples and was centrifuged before placing in DLS flow cell. DLS for 25 mM Tris-HCl at pH 7.4,
22° C was repeated 5 times. Four out of 5 experiments showed inconclusive readings. Either the
hydrodynamic radii were calculated as zero or the machine presented with an error in reading the
measurements. The remaining 1 experiment showed 2 sets of particles 1-10 nm and 100 nm in
RH, however the intensity correlation function in this experiment showed high signal-to-noise ratio
suggesting presence of dust or other dirt particles. It was therefore concluded that the baseline
DLS readings for 25 mM Tris-HCl buffer were zero and it does not generate background signals.
0
10
20
30
40
50
60
70
0.60 1.62 4.34 11.66 31.31 84.09 225.86
Mass %
rAmel RH (nm)
Figure 3.7. DLS peaks showing the % mass distribution of hydrodynamic radii of rAmel
assemblies in 25 mM Tris-HCl, pH 7.4-7.6, at 22° C.
58

3.6.2 Characterization of rAmel nanospheres
Figure 3.7 shows the size distribution of rAmel nanospheres in 25 mM Tris-HCl, pH 7.4-7.6, at
22°C. Each data series in the graph represents a set of 10 readings from a DLS experiment. It
was observed that, up to 70% of rAmel particles had ~12 nm hydrodynamic radius (R H) or ~24
nm diameter. The particles between 1.6-3 nm RH represent rAmel monomers. DLS data for rAmel
were consistent across 5 repeated experiments. DLS data are an excellent representation of
particle size distribution in the sample. Figure 3.8 shows a histogram of average particle size
distribution of rAmel from Figure 3.7. The hydrodynamic radii between 7.1-14.9 nm represent the
average size of rAmel nanosphere population. This population was fairly homogenous in nature
and DLS detected low polydispersity in rAmel samples.
An average of 69.51% of particles are present as nanospheres (between 7.1 and 14.9 nm R H,
Figure 3.8). The larger particles (>19 nm R H) may be rAmel aggregates but can also be rAmel
nanosphere chains which the DLS software treats as spherical assemblies. The rAmel particles
were visualized with TEM to confirm that they were indeed nanospheres.
Figure 3.8. Histogram of DLS peaks from Figure 3.7, showing a homogenous population of
nanospheres with RH 7.1-14.9 or ~20 nm average diameter
0
10
20
30
40
50
60
70
80
0.6-5.5 7.1-14.9 19.1-84.09 107-370
Mass %
R
H
(nm)
Nanospheres
59

Figure 3.9 A shows a TEM image of 25 mM Tris-HCl buffer at pH 7.4, stained with 2% uranyl
acetate at 200,000 X magnification. The Tris-HCl control sample appears clean and does not
show any particles resembling nanospheres or other spherical structures. Figure 3.9 B shows a
negatively stained TEM image of rAmel nanospheres at 40,000 X magnification with the
nanospheres at 200,000 X magnification (from a different grid) in the top right corner inset. The
particle sizes of 100 nanospheres within this image were measured using ImageJ and the average
diameter of the particles was found to be 22.5 nm. This size is consistent with the rAmel
nanosphere size described in literature. The TEM image supports the findings of DLS and
confirms the presence of rAmel nanospheres in the solution.
3.6.3 Ambn peptide encoded by exon 5 (AB2) dis-assembles rAmel
nanospheres
3.6.3a. DLS analysis of rAmel-AB2
Figure 3.9. Negatively stained TEM images A. Tris-HCl control showing no structures resembling
nanospheres, 200,000 X magnification. B. 0.3 mg/mL rAmel, 40,000 X magnification showing typical
rAmel nanospheres ~20 nm in diameter. Inset shows rAmel nanospheres at 200,000 X.
60

To analyze the effect of AB2 on the formation of nanospheres, protocol for obtaining rAmel
nanospheres was followed with the addition of AB2. The pink bars in Figure 3.10 represent the
size distribution immediately upon addition of AB2 to rAmel (0.1 mg/ml AB2 and 0.3 mg/ml rAmel,
in 25 mM Tris-HCl, pH 7.4-7.6, at 22° C). Particles larger than 100 nm radius appeared in the
solution as soon as AB2 was added. An average of 40.58% of the rAmel-AB2 particles were
between RH 107-370 nm (Figure 3.10, pink bar) at this stage. The average mass percentage of
rAmel nanospheres was dramatically decreased from 69.51% (Figure 3.8, bar labeled
nanospheres) to 2.1% (Figure 3.10 pink bar, 7.1-14.9 nm RH). This suggested that AB2 interacted
with rAmel immediately upon addition and interrupted the formation of nanospheres. The
hydrophobic interactions between N-terminus of AB2 harboring the Amel-interacting domain
(recognized by co-IP in Part I of this Chapter) and N-terminus of rAmel may have caused the
formation of large particles. Upon incubating rAmel-AB2 solution for 3 h at room temperature, the
particle size distribution changed significantly (Figure 3.10, purple bars). Note that, the sample
was not centrifuged between 0 h and 3 h measurements, but was left undisturbed in the DLS flow
cell inside the DLS machine at a controlled temperature of 22° C.
Figure 3.10. Histogram of the change in size distribution of rAmel+AB2 over time, from DLS analysis.
Pink bars show rAmel+AB2 measured immediately after mixing (0 h) and purple bars show the change
in their size after 3 h incubation at RT.
0
5
10
15
20
25
30
35
40
45
0.6-5.5 7.1-14.9 19.1-84.09 107-370
Mass %
Average diameter (nm)
rAmel+AB2 at 0 h
rAmel+AB2 at 3h
61

After incubating rAmel with AB2 for 3 h, a heterogeneous particle size distribution was observed.
Largest population by mass was 42.2 % of particles between 0.6-5.5 nm RH (1.2-11 nm diameter).
This group can represent a mixed population of monomeric rAmel (~3.5-4.5 nm diameter), and
rAmel oligomers (~7.5 nm diameter). It is possible that AB2 stabilized rAmel monomers and
oligomers before formation of nanospheres due to its positive charge. The small population of
nanospheres (14.6%, 7.1-14-9 nm radius) could represent the rAmel which did not come into
direct contact with AB2. There were some large aggregates present even at the end of 3 h
(27.39%, 107-370 nm radius). DLS was did not detect AB2 only at concentrations as high as 0.3
mg/mL (data not shown) probably due to its very small size (~4.2 kDa). Therefore there is a strong
evidence that 0.6-5.5 nm RH particles are not AB2 but rather rAmel monomers or oligomers
associated with AB2.
3.6.3b. TEM analysis of rAmel-AB2 mixture
TEM images of AB2 only (0.3 mg/mL) were obtained first at 10,000 X magnification and 80 kV
accelerating voltage. Although, AB2 was stained with 2% uranyl acetate (UA), it gained a positive
stain and formed ring like structures ~ 25-30 nm in diameter (Figure 3.11).
Figure 3.11. TEM image of AB2 particles ~25 nm in diameter. Scale bar 100 nm.
62

TEM image of negatively stained rAmel and rAmel+AB2 are shown in Figure 3.12 A and B
respectively. Both images were taken at 40,000 X magnification, 120 kV accelerating voltage. The
diameters of 100 nanospheres were measured in 2 separate TEM images of rAmel only (0.3
mg/mL) and rAmel-AB2 (0.3 mg/mL rAmel, 0.1 mg/mL AB2) using ImageJ. Recombinant Amel
nanospheres had an average diameter of 24.0-24.3 nm. Upon addition of AB2, the nanosphere
size decreased to average diameter 18.2-18.5 nm. Dot plot representing the change in particle
Figure 3.12. TEM images of negatively stained rAmel (A) and rAmel+AB2 (B) at 40k X magnification
showing reduced particle size of rAmel upon addition of AB2.
Figure 3.13. Dot plot comparing the diameters of rAmel nanospheres and rAmel+AB2 co-assemblies.
Average diameter of rAmel nanospheres was 24.4 nm (red dots) which decreased to an average of 18.4
nm when AB2 was added to rAmel (yellow dots), n= 100, counted in 4 images, 2 rAmel only and 2 rAmel
+ AB2; p << 0.0001, n = 100
0
5
10
15
20
25
30
35
0 20 40 60 80 100
Diameter (nm)
rAmel rAmel+AB2
63

size is shown in Figure 3.13. The difference in average size of the particles was statistically highly
significant (p <<< 0.0001) as determined by t-test (unequal variances).
Although these measurements do not directly represent the measurements from DLS, it supports
the general trend observed in DLS that rAmel nanosphere formation was interrupted by AB2, with
the end result of overall smaller particle size distribution as the result of AB2 addition.
DLS and TEM data of rAmel-AB2 interaction confirmed the co-IP result from Part I of this Chapter
that AB2 can bind directly bind to Amel. These experiments also helped establish protocols in
order to study the co-assembly of rAmel with full-length rAmbn.
3.6.4 Full-length rAmbn co-assembles with rAmel in solution
Although use of AB2 provided insight into its co-assembly with rAmel, AB2 or exon 5 encoded
region of Ambn is not a native proteolytic product of Ambn. Therefore, to confirm the data obtained
by AB2-rAmel interaction and to predict how the co-assembly may be taking place in vivo, DLS
and TEM experiments were repeated using full-length rAmbn and rAmel. The ratio between
rAmbn: rAmel was kept the same as AB2: rAmel (1:3 by weight).
3.6.4a. DLS analysis of rAmel-rAmbn
First, DLS analysis of rAmbn was performed at a concentration of 0.1 mg/mL to determine the
particle distribution in solution. It was challenging to obtain clean DLS data with low polydispersity
and low signal-to-noise ratio from rAmbn due to aggregation and difficulty in dissolving rAmbn in
near neutral pH Tris- HCl. It was found that most reliable DLS data were acquired from a fresh
batch of recombinant Ambn without storing the protein for extended time at -20° C. Despite these
efforts, rAmbn polydispersity was higher than 20% in majority of the experiments. Figure 3.14
A&B compare the particle size distribution of rAmel with rAmbn.
64

Figure 3.14 A shows rAmel, 0.3 mg/mL in 25 mM Tris-HCl buffer at pH 7.4-7.6 and 22° C. Sample
had low polydispersity and approximately 65% of the particles by mass were of similar size,
between 7-12 nm RH or 14-24 nm diameter corresponding to nanospheres (similar to
nanospheres characterized in section 3.6.2). Figure 3.14 B shows rAmbn particle size distribution
at 0.1 mg/mL in 25 mM Tris-HCl buffer at pH 7.4-7.6 and 22° C. The average particle size of
rAmbn was approximately 5.5 nm RH. Only 25-30% of particles by mass were of similar size and
the sample had polydispersity values between 22-25%.
Unlike the sudden change in size distribution observed upon addition of AB2 to rAmel, there was
no significant difference in the size distribution of rAmel immediately upon addition of rAmbn (data
not shown). This may be because of the lower concentrations of both rAmel and rAmbn (0.15 and
0.05 mg/mL respectively) or because initial rAmbn particles are similar in size to rAmel
nanospheres or it could also be that these intrinsically disordered full-length proteins need time
to orient accurately for co-assembly in order to facilitate protein-protein interactions.
After 2 h incubation at room temperature, rAmel-rAmbn solution depicted an overall shift toward
smaller radius (Figure 3.15 A), similar to the trend observed in rAmel-AB2. The mixed sample
was more heterogeneous than either rAmel alone or rAmbn alone. To understand the size
distribution of this heterogeneous population, a histogram was plotted (Figure 3.15 B) for average
Figure 3.14. DLS peaks showing size distribution of rAmel and rAmbn particles A. rAmel particles are
majority nanospheres (~22 nm diameter). B. rAmbn distribution is more heterogenous. Average particle
size rAmbn ~14 nm in diamter.
65

mass percent against hydrodynamic radius. The histogram revealed most abundant particle size
of 9.11 nm RH or ~18 nm in diameter. There was a wide distribution of sizes from 5.56 to 11.66
nm RH. It is likely that strong hydrophobic forces cause rAmel-rAmbn co-assembled particles to
be smaller than rAmel particles. To reveal the true nature of these assemblies and to obtain exact
measurements, TEM analysis was performed.
3.6.4b. TEM analysis of rAmel-rAmbn mixture
First, baseline images of rAmbn only were obtained with 0.04 mg/mL rAmbn. 0.1 mg/mL rAmbn
was diluted to 0.04 mg/mL in dd water to reduce the amount of aggregates on TEM grids. The
rAmbn was analyzed at a different facility from the rest of the TEM images, with 80 kV accelerating
voltage, 10,000 X magnification. TEM did not reveal self-assembly of rAmbn as it has been
previously published by Wald et al. (2013). Figure 3.16 shows negatively stained full-length rAmbn
at 10,000 X magnification. Globular/spherical particles about 20 nm in diameter (Figure 3.16,
white arrow heads), very similar to rAmel in shape and size were observed in rAmbn. However,
these spherical assemblies were few and interspersed by larger aggregates more than 100 nm
long and ~20 nm thick (Figure 3.16, center of the image).
Figure 3.15. A. DLS peaks showing particle size distribution of rAmel+rAmbn at 1:3 ratio by weight. B.
Histogram showing heterogeneous size distribution of rAmel+rAmbn particles. Average particle size was
~ 18 nm.
66

TEM results of rAmel mixed with rAmbn are shown in Figure 3.17. Recombinant Amel and rAmbn
were incubated for 2 h at room temperature to mimic the conditions under which DLS
measurements were taken. Grids were negatively stained with 2% uranyl acetate as described
earlier. TEM revealed that rAmel and rAmbn co-assembled to form spherical structures similar to
rAmel nanospheres in shape but smaller in size. The diameters of 100 particles were measured
Figure 3.16. TEM image of negatively stained rAmbn (0.04 mg/mL) showing smaller spherical assemblies
(white arrow heads) ~20 nm in diameter and some larger ~100 nm long aggregates (center).
Figure 3.17. TEM image of negatively stained rAmel-rAmbn mixture showing spherical
co-assemblies of ~12 nm average diameter.
67

from Figure 3.17 using ImageJ. The average diameter of rAmel-rAmbn co-assemblies was found
to be ~12 nm, as compared to 22 nm average diameter of rAmel nanospheres (Figure 3.18). No
large aggregates like those in rAmbn were observed in rAmel-rAmbn mixture. Further, only 8
particles out of the 100 counted were larger than 15 nm in diameter. Therefore it can be said that
more than 90% of the particles in rAmel-rAmbn TEM consisted of co-assemblies. The rAmel was
in turn stabilized rAmbn and prevented its aggregation. Further investigation is required to
determine the subunits of Amel-Ambn co-assemblies.
3.7 Discussion
Amelogenin and ameloblastin interactions have been suggested for over two decades
(Hatakeyama et al., 2009; Mazumder et al., 2016; Mazumder et al., 2014; Nanci et al., 1998;
Zalzal et al., 2008) but direct evidence of their binding using co-immunoprecipitation (co-IP) was
shown for the first time in Chapter 2. In this Chapter, I took the established protocol for co-IP
further to identify the Amel-binding domain of Ambn. Previous research had identified that the a
0
5
10
15
20
25
30
35
0 20 40 60 80 100
Diameter (nm)
rAmel nanospheres diameter
rAmel+rAmbn co-assemblies diameter
Figure 3.18. Dot plot comparing the diameters of rAmel nanospheres and rAmel+rAmbn co-assemblies.
Average diameter of rAmel nanospheres was 22 nm (blue dots) which decreased to an average of 12 nm
when rAmbn was added to rAmel (orange dots). n = 100, p <<< 0.0001

68

synthetic peptide derived from exon 5 encoded region of Ambn affects the secondary structure of
Amel (Su et al., 2016). In this Chapter using co-IP I further confirmed that the exon 5 encoded
fragment interacts directly with Amel via its N-terminal 18 amino acid region.
For this purpose, I used two recombinant Ambn mutants AmbnΔ5 and AmbnΔ6, which lack
sequences encoded by exon 5 and exon 6 respectively (Su et al., 2019b). Co-IP columns were
prepared by binding anti-Amel antibody to the resin and recombinant Amel (rAmel) was used as
bait protein. AmbnΔ5 and AmbnΔ6 proteins were used as prey. AmbnΔ6 bound to rAmel and co-
eluted with rAmel but AmbnΔ5 failed to bind to rAmel. Figure 3.5 Western blot labeled in tandem
with anti-Ambn (M300) and anti-Amel antibodies shows the elution fractions.
The amino acid regions encoded by exon 5 of Ambn contains important self-assembly domain
(Wald et al., 2013; 2017) and lack of exons 5&6 in mutant mice is known to give rise to a truncated
non-functional form of Ambn (Fukumoto et al., 2004; Wazen et al., 2009). Therefore it was
necessary to confirm that the inability of AmbnΔ5 to bind to rAmel was not due to its misfolding
or loss of function. To this end, Ambn synthetic peptides representing fragments encoding exons
5 and 6 were used as prey proteins in the co-IP. In Figure 3.6 it can be observed that Ambn
peptides AB2 and N-terminal fragment AB2N co-elute with rAmel (lanes 2 and 3) whereas
peptides AB4 and AB2C fail to bind to rAmel (lanes 1 and 4). Note the weak band of AB2N as
compared to AB2 (lanes 2 and 3). Although equal amounts of each of the prey proteins were
added to the co-IP columns in all experiments, significantly less amount of AB2N seems to have
bound to rAmel as compared to AB2. We propose that, even though AB2C region cannot bind
independently to Amel, it may have a role in maintaining the structural orientation of AB2N for its
binding to Amel. Therefore in the absence of AB2C, peptide AB2N cannot bind to Amel as
effectively as the complete sequence of AB2.
69

Another notable observation from Figure 3.6 gel is AB2N runs faster than AB2C even though,
AB2N is larger in size than AB2C (controls towards right, Figure 3.6). The difference might be
because of the hydrophilic nature of AB2C (hydropathicity GRAVY score 1.688) as compared to
AB2N which is hydrophobic (hydropathicity GRAVY score 0.285). The binding of hydrophobic
proteins to SDS molecules is more efficient so they tend to run faster whereas hydrophilic proteins
do not bind as well to SDS and run slower (Shirai et al., 2008).
These experiments confirm that the amelogenin-binding region of ameloblastin lies within the N-
terminal 20 amino acids encoded by exon 5 of Ambn. This evolutionarily highly conserved region
contains the self-assembly domain Y/F-x-x-Y/L/F-x-Y/F (Wald et al., 2017). This region is not
spliced out of any alternatively spliced isoforms of Ambn (MacDougall et al., 2000). The MMP20
cleavage sites for Ambn are before and after the exon 5 encoded region, keeping this domain of
essential residues intact during secretory stage of enamel formation (Chun et al., 2010). Amel
also contains two Y/F-x-x-Y/L/F-x-Y/F domains at its N-terminus, within the 42 amino acid region
called A-domain essential for Amel self-assembly (Paine et al., 2000). N-terminal fragments of
Amel, including TRAP which contains the A-domain, accumulate in maturation stage enamel with
N-terminal fragments of Ambn and are within 5-7 nm of each other (Mazumder et al., 2016).
Therefore, we suggest that the self-assembly domains of Amel-Ambn act as co-assembly
domains and are involved in Amel-Ambn binding.
To improve our understanding of Amel-Ambn interaction, it is imperative to learn about co-
assembly of these proteins. The co-assembly may be essential in vivo to facilitate the interaction
between Amel and hydroxyapatite after the cleavage of C-terminal of Amel (Moradian-Oldak et
al., 2002); or adhesion between extracellular matrix and ameloblast cells (explored in Chapter 4)
Biophysical techniques of DLS and TEM were used to examine the co-assembly of rAmel with
rAmbn. Dynamic light scattering is an important tool to determine the hydrodynamic radii of
70

particles in solution. It can also detect the presence of aggregates in the samples, and its intensity
correlation function can validate how “clean” the sample is by displaying the signal-to-noise ratio.
This is critical in case of proteins like rAmel and rAmbn with a tendency to aggregate. DLS also
determines the “dispersity” of particle sizes. If the particles in a solution are distributed over a wide
range of hydrodynamic radii, the sample is defined to be polydispersed (polydispersity of > 20%).
If the size distribution is narrow, the sample is defined as monodispersed. DLS makes an ideal
precursor to TEM because samples can be retrieved from the flow-cell and applied directly to
TEM grids. Table 3.3 summarizes the DLS and TEM measurements for each experiment.
Protein
DLS (diameters
calculated from RH)
TEM (diameters
measured by ImageJ)
rAmel 20 nm 24 nm
AB2 undetectable
positively stained rings
25-30 nm
rAmbn ~14-18 nm, heterogeneous
20 nm spheres + >100
nm aggregates
rAmel+AB2 heterogeneous 0.6 to >100 18 nm
rAmel+rAmbn 18 nm 12 nm
Table 3.3. Summary of DLS and TEM measurements of nanoparticle sizes from proteins or protein
mixtures.
For initial experiments, Ambn peptide AB2 was added to rAmel instead of full-length rAmbn to
establish the experimental protocols. AB2 directly binds to rAmel, it is less aggregative and less
labile at room temperature as compared to full-length rAmbn.
DLS of rAmel shows monodispersed (polydispersity < 20%) nanospheres at pH 7.4, 0.3 mg/mL
concentration (Figure 3.7). Immediately upon addition of AB2 to rAmel, particles larger than 100
nm R H formed. These could be a result of hydrophobic forces between AB2 and rAmel leading to
immediate aggregation (Figure 3.10). As time passed, these intrinsically disordered proteins likely
oriented themselves in order to achieve specific interactions between N-terminal domains of AB2
and rAmel. This interaction was evident after 3 h incubation of rAmel-AB2 at 22° C. It appeared
71

that there was an overall reduction in particle sizes. This finding was supported by TEM analysis
where the average size of rAmel-AB2 nanospheres was 18 nm as compared to 24 nm of rAmel
nanospheres (Table 3.3).
AB2 is not a native proteolytic cleavage product of Ambn. Therefore it was not possible to
extrapolate the results of AB2-Amel interaction to in vivo events. Hence, DLS and TEM analyses
were repeated with full-length rAmbn and rAmel in order to predict their co-assembly, to help
elucidate the events in vivo.
The DLS results displayed continuation of the same trend as rAmel-AB2. The sizes of rAmel-
rAmbn particle assemblies were smaller than rAmel nanospheres however, it was difficult to
obtain monodispersed samples of rAmel-rAmbn mixture (Figure 3.15). TEM images of rAmbn did
not show very obvious self-assemblies like those displayed by Wald et al. (2013). Rather, large
aggregates interspersed with 20 nm diameter rAmbn spheres were observed in rAmbn samples
(Figure 3.16). Upon mixing with rAmel, the aggregates disappeared and an even field of ~12 nm
in diameter globular particles was obtained (Figure 3.17). Less than 8% of these particles were
15 nm in diameter or larger, which eliminates the possibility that the globular particles consisted
purely rAmbn or rAmel self-assemblies.
Amel-Ambn co-assemblies have to be systematically dissected to understand how Amel-Ambn
together form these smaller nanosphere like structures. However, the driving force behind the co-
assembly must be hydrophobic interactions between Ambn exon 5 encoded region containing the
Amel-binding domain and the hydrophobic N-terminal of Amel.
Biological Significance of Amel-Ambn co-assembly: We have previously shown the co-
localization of Amel and Ambn N-terminal fragments in the rod-sheath space of maturing enamel
(Mazumder et al., 2016). In this study, Amel N-terminal TRAP fragment and Ambn N-terminal 17
72

kDa fragments were isolated from maturation stage molars. The distance between these
fragments within the rod-sheath space was calculated to be between 5-7 nm using fluorescent
resonance energy transfer (FRET). In the light of data from the current Chapter, it can be said
that accumulation of these proteins around the prisms is not random but a controlled event in
which Amel-Ambn N-terminal fragments co-assemble to maintain rod-interrod organization.
Ambn mutant mice designed by Wald et al. (2017) further supported that Amel-Ambn co-assembly
may be essential for enamel rod-interrod organization. Key tyrosine, leucine, and phenylalanine
residues of the Y/F-x-x-Y/L/F-x-Y/F self-assembly motif of Ambn were mutated to glycine in the
mutant mice. These mice exhibited loss of organization in amelogenin containing enamel
extracellular matrix, and there was expansion of interrod material at the cost of rod. It is known
that in full-length Amel nanospheres, the C-terminal hydrophilic tail faces the external environment
(Fang et al., 2011), allowing Amel to bind to newly formed mineral ribbons in nascent enamel.
This prevents untimely fusion of newly formed crystals (Moradian-Oldak et al., 2003). Soon, the
C-terminus of Amel is cleaved by MMP-20 and the calcium binding ability of Amel significantly
decreases (Moradian-Oldak et al., 2002). After this stage, Amel can effectively bind to the mineral
phase only with the help of a non-amelogenin mediator like Ambn. It is likely that Amel co-
assembles with Ambn on the crystal surfaces in order to maintain rod boundaries. If Amel-Ambn
interaction is taking place through the Y/F-x-x-Y/L/F-x-Y/F self-assembly domain as predicted,
mutation of its key residues to glycine would disrupt Amel-Ambn co-assembly. This will lead to
disordered Amel matrix organization as Amel cannot bind to crystal surfaces efficiently and can
explain the eventual loss of rod interrod arrangement.

73

4. Colocalization of Amel-Ambn and Ambn-
ameloblast cell membrane
4.1 Background
In the previous chapters I have provided in vitro data to conclude that amelogenin (Amel) and
ameloblastin (Ambn) directly bind to each other, through the N-terminus of Ambn sequence
encoded by exon 5. In this Chapter, I investigated Amel-Ambn colocalization in vivo, in developing
mouse incisor from secretory to maturation stage. In order to provide insight into the function of
Amel-Ambn interactions in enamel formation, I explored ameloblast-cell membrane colocalization
in vivo.
Amelogenin-ameloblastin interaction in vivo: Amel-Ambn are co-secreted through the same
secretory vesicles (Zalzal et al., 2008). Immunogold labeled images of developing rat incisors
identified Amel and Ambn gold particles together in almost 70% of the secretory vesicles.
Amelogenin synthesis is reduced at mRNA and protein levels in mutant mice expressing exon 5-
6 deled Ambn (Fukumoto et al., 2004; Fukumoto et al., 2005). The cooperative role of Amel and
Ambn was also made evident by the phenotype of Amel-Ambn double mutant mouse
(Hatakeyama et al., 2009). This mouse showed additional defects as compared to Amel or Ambn
single mutants, calcified nodules seen in Ambn mutant were absent and the double mutant
enamel thickness was significantly reduced. Studies from our laboratory using mouse molars
have shown that full length Amel and Ambn colocalize at the secretory face of ameloblasts
(Mazumder et al., 2014). It was reported that their N-terminal fragments colocalize and are within
5-7 nm of each other, in the rod-sheath space around maturing enamel rods (Figure 4.1)
(Mazumder et al., 2016).
74

It is challenging to analyze all stages of amelogenesis in a developing mouse molar because, in
molars at any given point in time the enamel organ is at one stage of development. Therefore to
analyze Amel-Ambn colocalization in different stages of amelogenesis, molars need to be
harvested from different animals at different time points. However, mouse incisors are
continuously growing and retain secretory, transition and maturation stage ameloblasts
throughout the lifetime of the mouse. Therefore, I utilized the continuously growing mouse incisor
as a model in this project, along with two different antibodies against Ambn to analyze Amel-Ambn
colocalization from secretory to maturation stage. This provided a complete picture of how the
colocalization of full-length (largely un-cleaved by proteinases) Ambn with Amel and the N-
terminal fragments of Ambn with Amel changes during amelogenesis.
Ambn-ameloblast cell interaction: Interaction of ameloblastin and ameloblast cells has been
proposed by several independent groups. In Ambn exon 5-6 deletion mutant mouse, ameloblast
separate from underlying dentine and revert into cell cycle (Fukumoto et al., 2004). A number of
cell binding domains have been identified on Ambn including an integrin binding domain, a
thrombospondin-like cell adhesion domain or fibronectin binding domain, and a heparin binding
Figure 4.1. A,B,D&E. Confocal images of P8 enamel cross section showing Amel-Ambn N-terminal
fragments colocalization in rod-sheaths in characteristic “fish-net” pattern. C. Pseudocolored fluorescent
resonance energy transfer (FRET) image showing FRET efficiencies from highest (red, 0.89, positive
FRET signal, Amel-Ambn fragments closest) to lowest (purple, 0, negative FRET) (Mazumder et al.,
2016).
C
75

domain (Beyeler et al., 2010; Černý et al., 1996; Sonoda et al., 2009). However, none of these
domains are conserved throughout evolution. Recently, direct binding of Ambn with cell
membrane mimicking lipid vesicles in vitro was shown (Su et al., 2019b). A membrane leakage
assay and cryo-transmission electron microscopy were used to detect the disruption of lipid
vesicles, and circular dichroism and intrinsic tryptophan fluorescence were used to detect the
changes in the secondary structure of Ambn. It was concluded that Ambn can directly bind to
phospholipid bilayers (like cell membranes) through an amphipathic helix motif within its exon 5
encoded region (Figure 4.2).
In this Chapter, I examined the colocalization of Ambn with ameloblast cell membrane in situ,
within developing mouse incisor by co-staining the ameloblast cell membranes with Ambn
immunolabel. I also analyzed the interaction of Ambn exon 5 peptide AB2 with ameloblast lineage
cells (ALC) (Nakata et al., 2003). Some of the commonly used ameloblast like cell lines and the
Scheme 4.1. Schematic showing some of the ameloblast derived cell lines. Blue hexagon mentions
primary cell lines and white hexagons are immortalized cell lines.
Figure 4.2. Cartoon of Ambn exon 5 encoded fragment showing the amphipathic helix motif in
pink and part of the Ambn self-assembly motif (putative Amel-binding region) in the purple box.
76

animals they are derived from are outlined in Scheme 4.1. The cells used in this project,
Ameloblast lineage cells (ALC) were obtained from Dr. Toshihiro Sugiyama (Akita University,
Japan). ALC line was created by Nakata et al. (2003) using spontaneous immortalization of dental
epithelial cells from new born mouse molars germs. In order to have a cell line resemble native
ameloblasts, enamel epithelial cells from molar germs were isolated by selecting clonal colonies
having characteristic epithelial morphology in culture conditions unfavorable for growth of
fibroblasts. ALC line has been used for studying differentiation in ameloblasts (Takahashi et al.,
2007), effect of ameloblastin on Hertwig’s epithelial root sheath cells (Hirose et al., 2013), and to
study ameloblastomas (Sweeney et al., 2014). In this project, ALC were used to study the biding
of FITC labeled peptide AB2 to the cell membrane of ameloblasts.
4.2 Material and methods
4.2.1 Tissue preparation
Post-natal day 8 (P8, day of birth assigned as day zero) wild-type C57BL/6 mice were euthanized
following protocols from the Institutional Animal Care and Use Committee (IACUC) of the
University of Southern California. The mandibles were dissected, debrided of soft tissue and fixed
in 4 % paraformaldehyde (PFA) in phosphate buffered saline (PBS) at 4° C for 24 h. Fixed
mandibles were washed with PBS three times and decalcified in 10% ethylene diamine tetra-
acetic acid (EDTA) in PBS containing 0.2% PFA and 0.1% glutaraldehyde at 4° C for 8-10 days.
The EDTA solution was replaced 4 times during decalcification. The samples were then
dehydrated using sequentially increasing concentrations of ethanol and embedded in paraffin
according to standard histological protocol (Fischer et al., 2006). Paraffin blocks were sectioned
into 7 µm thick sections, maintaining the structure of the developing incisor. The mandibles were
sectioned in two different longitudinal orientations- along the sagittal plane and anteroposteriorly
along the transverse plane (Figure 4.3).
77

4.2.2 Immunohistochemical labeling for Amel and Ambn

Primary Antibody Source Host Dilution
Anti-amelogenin
Gift form Dr. Malcolm
Snead
Chicken 1:1000
Anti-ameloblastin
M300
Santa Cruz biotech, sc-
50534 (discontinued)
Rabbit 1:500
Anti-ameloblastin
N-18
Santa Cruz biotech, sc-
33100 (discontinued)
Goat 1:100
Table 4.1. Primary antibodies used for immunohistochemical labeling of Amel and Ambn
Single step deparaffinization and epitope retrieval were performed by submerging the slides in
10mM sodium citrate buffer (pH 6.0) with 0.05% Tween20 at 60° C overnight. The slides were
then cooled to room temperature and washed in 10 mM Tris buffered saline (TBS, pH 7.4).
Endogenous peroxidase activity was blocked by using 0.3% hydrogen peroxide, and non-specific
binding sites on the tissue were blocked by 1% bovine serum albumin (BSA) in TBS for 30 minutes
each.
Primary antibodies, their dilutions, and sources are listed in Table 4.1. The epitope of M300
antibody is shown in Figure 4.4 in purple. The epitope of N-18 antibody is 18 amino acids at the
N-terminus of human Ambn, but its exact sequence is not known. Slides were incubated overnight
with primary antibodies diluted in antibody dilution solution (0.1% BSA, 0.3% Triton X-100 in TBS).
Figure 4.3. Cartoon showing two sectioning orientations used for P8 mandibles. A. in the sagittal plane,
B. in the transverse plane, anteroposteriorly.
78

Proteins were labeled with appropriate secondary antibodies conjugated with FITC or Alexa 488
for Amel and TRITC or Alexa 594 for Ambn, diluted 1:100 in TBS for 4 hours in dark. Coverslips
were placed using Vectashield Hard Set mounting medium with DAPI (Vector labs). This protocol
was published as a book chapter ‘immunohistochemical colocalization of amelogenin and
ameloblastin in developing enamel matrix’, Bapat and Moradian-Oldak (2019). The slides were
examined using a Leica SP-8 confocal microscope. Colocalization patterns within the developing
incisor were analyzed with Leica Application Suite LAS-X version 1.8.1.13759.
4.2.3 Quantitative colocalization (Manders’ colocalization coefficient)
For colocalization analysis, approximately 6 x 10 µm regions of interest (ROIs) were selected
along two different areas of the incisor, 1) inside ameloblast cells, 2) at the secretory front/ Tomes’
processes of ameloblasts. For colocalization analysis within the thickness of enamel, large ROIs
~25 x 60 µm size were selected. For Ambn M300 labeled sections, threshold was 30% and
background was 25% whereas for Ambn N-18 labeled sections, high threshold of 60% was used
with background 25%. Manders’ colocalization coefficient (MCC) (Manders et al., 1993) was
calculated using the following equations.
𝑴 𝒓𝒆𝒅
=
∑ 𝑹 𝒊 ,𝒄𝒐𝒍𝒐𝒄 𝒊 ∑ 𝑹 𝒊 𝒊 , 𝑴 𝒈𝒓𝒆𝒆𝒏
=
∑ 𝑮 𝒊 ,𝒄𝒐𝒍 𝒐𝒄 𝒊 ∑ 𝑮 𝒊 𝒊
Figure 4.4. Mouse Ambn sequence depicting the antigen epitope for M300 antibody in purple
79

where Ri,coloc = Ri if Gi > 0 and Ri,coloc = 0 if Gi = 0; and Gi,coloc = Gi if Ri > 0 and G i,coloc = 0 if Ri = 0.
Ri,coloc, Ri, Gi,coloc, and Gi were calculated by the Leica LAS-X software and M red and M green were
calculated in Microsoft Excel 2013. MCC provides one value for each channel being examined.
This ascertains that, each coefficient is directly proportional to the fluorescent intensities of that
channel and cannot be skewed by the differences in the intensity of the second fluorophore. If the
value for either of the colocalization coefficients is 1, it indicates complete colocalization of that
fluorophore with the other.
4.2.4 Alizarin red S staining
Alizarin red S stain is commonly used for staining calcium. Alizarin red S solution was prepared
fresh at a concentration of 0.02 g/ml in distilled, deionized water. The pH of the solution was
adjusted to 4.1-4.3 with 10% ammonium hydroxide. Mandibles used in Alizarin red S staining
were processed the same way as described in section 4.2.1. Slides were deparaffinized with
xylene and rehydrated using decreasing grades of ethanol. Samples were immersed in Alizarin
red S solution for 3 minutes or until red-orange color was visible. Excess stain was blotted and
slides were dehydrated in acetone, followed by acetone-xylene 1:1 solution for 30 seconds each.
Slides were cleared in xylene and coverslips were mounted with a xylene based mounting
medium. Slides were imaged using Keyence BZX710 microscope.
4.2.5 AB2 labeling with FITC
Ambn peptide AB2 (see details in Chapter 3) was tagged with Fluorescein-5-isothiocyanate
(FITC) for visualization after being added to cell cultures. Fresh FITC solution was prepared by
dissolving FITC (Thermofisher) in dimethyl sulfoxide (DMSO) at 5 mg/mL in dark. Peptide AB2
was mixed with FITC solution in 1:100 molar ratio in pH 10.0, 100 mM sodium bicarbonate
(NaHCO3) buffer for 4 h in dark with constant mixing. Excess FITC was removed by buffer
80

exchange using a Microcon-3kDa centrifugal filter unit (Millipore) until AB2-FITC solution was
clear.
4.2.6 AB2-FITC binding to ALC
ALC were cultured in Dulbecco’s modified eagle medium (DMEM) with 100 U/ml penicillin, 100
mg/ml streptomycin and 10% heat inactivated fetal bovine serum. Peptide AB2 labeled with FITC
was exogenously added to ALC to study its cell membrane binding properties. First, ALC
membranes were stained with lipophilic dye 1,1'-Dioctadecyl-3,3,3',3'-
Tetramethylindodicarbocyanine Perchlorate (DiD, ThermoFisher Scientific) in DMEM at 1:100
dilution for at least 5 h up to overnight. Cells were gently washed with fresh DMEM and incubated
with 40 uM AB2-FITC in PBS for 15 min. Cells were washed 3 times with PBS to remove excess
AB2-FITC and fixed with 4 % PFA in PBS for 20 min at room temperature. Nuclei were stained
with DAPI diluted 1:1000 in DMSO for 10 min. Coverslips were placed with Prolong Gold antifade
mounting media (Invitrogen).
4.2.7 In situ cell membrane staining with DiD, co-labeled with Ambn
To stain mouse incisor tissue with DiD, IHC protocol was modified and extended across 4 days
as follows. After epitope retrieval, sections were incubated for 45 minutes in 0.3% Triton X-100 in
TBS. Cell membrane staining was performed with 5 mM DiD diluted 1:100 in DMSO overnight at
40° C in a hot air oven. Sections were labeled with anti-Ambn primary antibody M300, diluted
1:250 in 0.1% BSA in TBS at room temperature overnight. Secondary antibody conjugated with
Alexa 488 (ThermoFisher Scientific) was used to label ameloblastin. DAPI (1:1000) was added to
the secondary antibody dilution solution. To prevent the DiD from leaching out, no mounting media
were used and coverslips were placed using TBS and sealed with clear nail varnish. Slides were
imaged immediately with Leica SP-8 confocal microscope.
81

4.3 Results
4.3.1 Amel and full-length Ambn colocalize within ameloblasts and at the
secretory face of ameloblasts
Figure 4.5 shows a confocal tile scan image of P8 mouse incisor sagittal section labeled with anti-
Amel and anti-Ambn M300 primary antibodies. Amel and Ambn secondary antibodies were
conjugated with Alexa 488 (green) and Alexa 594 (red) respectively. The image is a digitally
stitched picture of 696 individual images taken to capture the entire length of the mouse incisor
secretory, transition, and maturation stages (white squares). The predominant staining for Amel
is observed within the enamel thickness while ameloblasts predominantly show more labeling for
Ambn. Amel and Ambn colocalization can be clearly observed at the junction of the ameloblasts
and enamel in orange (white arrowheads). Colocalization is also present within ameloblasts and
is observed better in higher resolution (Figure 4.6). This image presents an excellent overview of
distribution of Amel, its fragments and Ambn. Anti-Amel antibody labels the full-length Amel as
well as its cleavage products therefore, the entire thickness of the enamel matrix shows Amel
signal. Anti-Ambn M300 antibody mostly labels full-length Ambn therefore, Ambn signal is only
observed within ameloblasts where newly synthesized protein is present and at the junction of
ameloblasts and the enamel extracellular matrix (Tomes’ processes) where Ambn has not yet
undergone proteolytic cleavage.
Figure 4.6 shows high resolution (63 X magnification) confocal images of regions marked by white
squares in Figure 4.5. Figure 4.6 A depicts secretory stage ameloblasts showing their typical tall
columnar morphology and Tomes’ processes. Colocalization at the Tomes’ processes (TP) can
be clearly observed. Within the ameloblasts, the signal intensity for Ambn is higher hence the
appearance is more red than orange. Manders’ colocalization coefficient eliminates the visual
bias. MCC for Amel-Ambn (M300) colocalization are summarized in Table 4.2.
82

Area of
interest
Secretory stage Transition stage Maturation stage
M-green
(Amel)
M-red
(Ambn)
M-green
(Amel)
M-red
(Ambn)
M-green
(Amel)
M-red
(Ambn)
Enamel matrix 0.146 0.244 0.015 0.038 0.007 0.004
Secretory front/
Tomes’
processes
0.840 0.904 0.014 0.035 0.003 0.018
Within
ameloblasts
0.938 0.950 0.974 0.969 0.919 0.914
Table 4.2. Summary of average Manders’ colocalization coefficients (MCC) for all ROIs within a given area
from Figure 4.6, labeled with anti-Amel and anti-Ambn-M300 antibodies. Note- MCC Values bellow 0.5
indicate lack of colocalization (in grey).
Secretory stage: Figure 4.6 B shows MCC values for the areas marked by arrows in Figure 4.6
A. At the secretory front of ameloblasts, the average colocalization coefficient for Amel is 0.84,
which means about 84% of labeled Amel colocalizes with Ambn. Whereas the average
colocalization coefficient for Ambn in the same region is 0.904 meaning more labeled Ambn
(~90%) colocalizes with Amel. The “free” Amel not bound to Ambn may be a part of Amel self-
assembly or could be bound to other enamel matrix proteins like enamelin.
Transition stage: Figure 4.6 C shows transition stage ameloblasts. In this stage Tomes’ process
have retreated and the junction between ameloblasts and enamel matrix appears smooth at this
resolution. There was no colocalization observed at the enamel-cell interface in this stage. Within
transition stage ameloblasts, Amel-Ambn colocalization can be observed as protein synthesis
continues. Here the MCC values for both channels are ~0.97 (Figure 4.6 D) meaning almost all
of labeled Amel-Ambn colocalize within ameloblasts. This is in agreement with previous findings
that Amel-Ambn are synthesized and secreted through the same secretory pathway (Zalzal et al.,
2008).
Maturation stage: In Figure 4.6 E, maturation stage ameloblasts and enamel matrix are
observed. The enamel in this stage contains significantly high levels of hydroxyapatite, which is
removed during EDTA demineralization. Therefore large empty spaces are observed in the
83

residual enamel matrix where Amel fragments are concentrated. The ameloblasts continue Amel-
Ambn synthesis as they approach maturation stage and MCC within cells continue to be high
(~0.91, Figure 4.6 F).
The data in these images depict how full-length (largely uncleaved) Ambn colocalizes with Amel
from secretory to maturation stages of enamel formation. Note the considerably high signal for
Ambn within ameloblasts as compared to Amel. This could simply be due to the difference in the
intensity of fluorophores but it could also be because Ambn is interacting with ameloblast cells
performing functions not yet fully understood.

84

Figure 4.5. Confocal tile-scan image of P8 incisor showing Amel-Ambn colocalization in sagittal section. Amel is labeled with Alexa 488 (green),
and Ambn-M300 by Alexa 594 (red). Nuclei are labeled with DAPI (blue). White squares show areas selected for high resolution images
representing secretory, transition and maturation stages. Arrowheads show areas selected for colocalization analysis- within ameloblasts, at the
secretory front / Tomes’ processes, and within the enamel thickness.
85

0.5
0.6
0.7
0.8
0.9
1
1 3 5 7 9 11 13 15 17 19 21 23 25 27
ROIs
Amel within ameloblasts
Ambn within ameloblasts
D
0.5
0.6
0.7
0.8
0.9
1
1 2 3 4 5 6 7 8 9 101112131415
ROIs
Amel at secretory front
Ambn at secretory front
Amel within ameloblasts
Ambn within ameloblasts
B
0.5
0.6
0.7
0.8
0.9
1
1 3 5 7 9 11 13 15 17 19 21
ROIs
Amel within ameloblasts
Ambn within ameloblasts
F
Figure 4.6. High resolution confocal images of P8 incisors from secretory (A), transition (C), and
maturation (E) stages of amelogenesis. B, D, and F. Manders’ colocalization coefficients for A, C,
and E respectively.
86

4.3.2 N-terminal fragments of Ambn colocalize with Amel in the bulk of
developing enamel
Figure 4.7 shows digitally stitched tile-scan image of 1024 individual frames of P8 incisor labeled
with anti-Amel and anti-Ambn N-18 antibody. Secondary antibodies for Amel and Ambn were
conjugated with FITC and TRITC respectively. The anti-Ambn N-18 antibody labels N-terminal
containing fragments of Ambn hence the signal for Ambn can be observed throughout the entire
thickness of the enamel matrix from secretory to maturation. This image presents an overview of
the colocalization between Amel and N-terminal fragments of Ambn. The proteins colocalize
within ameloblasts, at the secretory face (Tomes’ processes) of ameloblasts, and within the
thickness of enamel (Figure 4.7, white arrowheads). The intensity for both Amel and Ambn signals
increases from secretory stage towards the tip of the incisor (Figure 4.7, white squares). As more
full-length proteins are cleaved, more epitopes are exposed for the antibodies to bind and hence
the signal intensity increases. The colocalization within ameloblasts is clearly evident in the high-
resolution images (Figure 4.8). TRITC presents more background than Alexa 594, which explains
the red artifacts in the pulp and surrounding enamel organ. High threshold value of 60% was used
to quantify colocalization for these images to eliminate high background signals from TRITC.
Area of interest
Secretory stage Transition stage Maturation stage
M-green
(Amel)
M-red
(Ambn)
M-green
(Amel)
M-red
(Ambn)
M-green
(Amel)
M-red
(Ambn)
Enamel matrix 0.903 0.858 0.944 0.916 0.999 0.998
Secretory front/ Tomes’
processes
0.890 0.871 N/A N/A
Within ameloblasts 0.879 0.856 0.987 0.980 0.876 0.845
Table 4.3. Summary of average Manders’ colocalization coefficients for all ROIs within a given area from
Figure 4.8, labeled with anti-Amel and anti-Ambn-N18 antibodies. There is no discernible ‘secretory front’
in transition and maturation stages (N/A in the Table) but Amel-Ambn colocalization is observed where
ameloblasts end and enamel matrix begins. This area is considered a part of the enamel matrix for the
purpose of calculations.
87

Figure 4.8 shows high resolution confocal images of secretory (Figure 4.8, A), transition (C) and
maturation (E) stages of amelogenesis from Figure 4.7. The data are summarized in Table 4.3.

Secretory stage: Figure 4.8, A- MCC values were calculated within all 3 areas of interest, within
ameloblasts, within Tomes’ processes, and within the thickness of enamel matrix (white arrows).
They are depicted in Figure 4.8 B. Overall, about 87% of each Amel and Ambn colocalized in the
secretory stage in all three areas. Within ameloblasts, although the MCC values are high, the
signal intensity for both Amel and Ambn is lower than that observed within enamel. As MCC
negates the effect of signal intensities, this difference is not reflected in the colocalization
quantification. The difference is signal intensities could be caused by less N-terminus being
exposed within the cells as compared to enamel.
Transition stage: Figure 4.8 C- There are no Tomes’ processes in this stage. Although
colocalization persists at the junction of ameloblasts and enamel matrix the area is not considered
separate from enamel matrix itself due to lack of obvious junctional structures i.e. Tomes’
processes. MCC increase for both Amel and Ambn in this stage. Within the enamel matrix
colocalization coefficient for Amel is 0.94 and for Ambn it is 0.91 (Figure 4.8, D). As mentioned
earlier, this effect is probably due to more exposed epitopes from proteolytic cleavage of full-
length Amel and Ambn.
Maturation stage: Figures 4.8 E- As enamel matures, N-terminal of Amel-Ambn fragments co-
assemble within the space around forming enamel rods, leading to almost 99% of both Amel and
Ambn colocalization (Figures 4.8 F). Similar to M300 labeled section above, Amel- Ambn continue
to colocalize within maturation stage ameloblasts even though the signal intensities are very low.
88

Figure 4.7. Confocal tile-scan image showing P8 mouse incisor labeled with anti-Amel (green) and anti-Ambn (red) N-18 antibody.
Nuclei are labeled with DAPI (blue). Am: Ameloblasts, E: Enamel
89

Figure 4.8. High resolution confocal images of P8 incisor secretory (A), transition (C), and
maturation (E) stages. B, D, and F. Manders' coefficients of colocalization for A, C, and E
respectively. E: Enamel, TP: Tomes’ processes, Am: ameloblasts.
0.5
0.6
0.7
0.8
0.9
1
1 2 3 4 5 6 7 8 9 10 11 12
ROIs
Amel within ameloblasts
Ambn within ameloblasts
Amel at secretory face
Ambn at secretory face
Amel within enamel
Ambn within enamel
B
0.5
0.6
0.7
0.8
0.9
1
1 2 3 4 5 6 7 8 9 10 11 12
ROIs
Amel within ameloblasts
Ambn within ameloblasts
Amel within enamel
Ambn within enamel
D
0.5
0.6
0.7
0.8
0.9
1
1 3 5 7 9 11 13 15
ROIs
Amel within ameloblasts
Ambn within ameloblasts
Amel within enamel
Ambn within enamel
F
90

4.3.3 Colocalization of N-terminal fragments of Ambn with Amel in maturation
stage enamel
Images from Figures 4.6 and 4.8 do not show the finer details of maturation stage enamel matrix.
To achieve maximum retention of the enamel matrix during tissue processing, anteroposterior
sections of mandibles were made in the transverse plane (Figure 4.3, B). This section allows the
visualization of enamel prism arrangement. Figure 4.9 shows the tip of P8 incisor (maturation
stage) immunolabeled for Amel and Ambn. Amel and N-terminal fragments of Ambn colocalize
within the maturation stage mouse incisor tip in characteristic decussating pattern observed in
mouse enamel rods. Figure 4.9 B shows Amel, and Figure 4.9 C shows Ambn fragments labeled
using N-18 antibody. Amel-Ambn colocalization in the pattern of enamel rods supports the
findings from Chapter 3 that, Amel-Ambn co-assemble around enamel rods to allow Amel to
effectively bind to HAP.
Figure 4.9. A. Confocal images of aneteroposterior section of P8 incisor showing Amel-Ambn
fragments colocalizing in the characteristic decussating pattern of mouse enamel rods. OE- outer
enamel, BE- bulk enamel. B&C. Amel and Ambn channels respectively.
91

Further evidence of the role Amel-Ambn co-assembly in enamel rod formation s provided in Figure
4.10. This figure shows Amel-Ambn colocalization in the labial aspect of incisor within superficial
enamel (SE) or final enamel, outer enamel (OE), and bulk enamel (BE, Figure 4.10, A). The Amel
and Ambn channels are shown in Figure 4.10 B&C. The pattern of protein co-assembly in
maturing enamel closely resembles prism structure observed in non-demineralized mouse
enamel SEM images (Lacruz et al., 2012) (Figure 4.10 D). These sections are not fully
demineralized and some calcium phosphate is retained which maintains the structure of the
enamel (confirmed by Alizarin Red S staining for calcium, Figure 4.11). Colocalization of Amel
and Ambn on the retained prism structure is a clear indication of their cooperative function in
forming enamel prism architecture.

Figure 4.10. A. Confocal image of anteroposterior section of enamel showing Amel-Ambn
colocalization within the surface enamel (SE) or final enamel, outer enamel (OE) and bulk enamel (BE);
B and C. Amel and Ambn channels respectively; D (inset). SEM image from Lacruz et al. (2012)
showing surface enamel, outer enamel and bulk enamel.
92

Figure 4.11 shows Alizarin red S stained P8 incisor. Alizarin Red S labels calcium, and darker
red-orange color represents higher amounts of calcium present in the section. Secretory stage
enamel matrix retains more calcium (Figure 4.11 A) as compared to transition stage (Figure 4.11
B) which has more calcium than the maturation stage (Figure 4.11 C). It is likely that EDTA
chelation is not effective when calcium is bound to organic matrix in secretory stage. As the
proteins are removed in the maturation stage, the calcium is easily chelated and enamel is almost
completely demineralized (Figure 4.11 C). The decussating pattern of enamel prisms is visible in
transition stage (Figure 4.11 B).
4.3.4 Ambn colocalizes with ameloblast cell membrane in situ
Figure 4.12 shows a digitally stitched tile-scan image consisting 119 individual frames. Cell
membranes in the tissue are stained with DiD dye. Ameloblastin is labeled with M300 antibody
and secondary antibody conjugated with Alexa 488. The white arrowheads indicate secretory and
transition stages. Ambn colocalizes with the apical end of ameloblast cell membrane in the
regions indicated by arrowheads. Figures 4.13 and 4.14 highlight Ambn-cell membrane
colocalization in secretory and transition stages.
Figure 4.11. Optical microscope images of Alizarin red S stained images of P8 incisor at 40 X
magnification. A. secretory stage, B. transition stage, C. maturation stage. Am- ameloblasts, TP- Tomes’
processes, E- enamel, D- dentin.
93

Figure 4.12. Confocal tile-scan showing colocalization of Ambn (green) with ameloblast cell membrane (red) in secretory and
transition stage (arrowheads). Nuclei in blue.
94

Figure 4.13 shows ameloblast cells in secretory stage. Figure 4.13 C is the red channel image
showing DiD stained cell membrane of ameloblasts including Tomes’ processes which can be
clearly observed towards the apical ends (white arrowhead in 4.13 C). Figure 4.13 B (green
channel image) displays Ambn localizing within ameloblasts as well as in the Tomes’ processes
(white arrowhead in 4.13 B). Figure 4.13 A is the merged image of both channels and the
colocalization of Ambn on ameloblast cell membrane at the apical end of ameloblasts can be
observed (white arrowhead in 4.13 A). It can be argued that this is simply Ambn being secreted
through the Tomes’ processes however, the transition stage image 4.14 reveals otherwise.
Figure 4.13. Confocal images showing A. colocalization of Ambn with Tomes’ processes of
ameloblast cells (white arrowhead); B. Ambn labeled with M300, and secondary antibody conjugated
with Alexa 488. White arrowhead shows Ambn at the apical end of ameloblasts; C. ameloblast cell
membranes labeled with DiD. White arrowhead shows Tomes’ processes.
95

It has been established in the literature that active secretion of full-length Ambn is significantly
reduced in transition stage as compared to secretory stage (Brookes et al., 2011). Yet, Ambn
synthesis continues in this stage and it is labeled in Figure 4.14 B. Ambn signal can be observed
apical to the nucleus, which is the location of protein synthesis apparatus (Golgi and endoplasmic
reticulum). Ambn further localizes at the apical end of ameloblasts adjacent to the enamel matrix
(Figure 4.14 B arrowheads, enamel matrix removed during processing). Figure 4.14 C shows
ameloblast cell membrane labeled with DiD. Figure 4.14 A (arrowhead) shows colocalization of
Ambn and ameloblast cell membrane. Ambn localizes at the apical end of ameloblasts, at the
junction of cells and enamel matrix even when active secretion of proteins is depleted. Taken
together with Ambn-Amel interaction data, and Amel-Ambn colocalization at the secretory front,
Figure 4.14. Confocal images showing A. colocalization of Ambn (green) with ameloblast cell
membrane (red) in transition stage. Nuclei in blue. Arrowhead shows Ambn-cell membrane
colocalization at the apical end; B. Ambn labeled with M300 and Alexa 488. Arrowheads show
localization of Ambn within cells and at the apical end of ameloblasts; C. ameloblast cell membrane
labeled with DiD.
96

this is strong evidence to conclude that Ambn acts as a cell adhesion molecule between
ameloblasts and enamel rich extracellular matrix.
4.3.5 Ambn binds to ALC cell membranes via its exon 5 encoded region
There is in vitro evidence from Su et al. (2016) that Ambn can directly interact with liposomes
used as models for cell membranes via its region encoded by exon 5. To show interactions
between Ambn and cell membrane in cell culture models, FITC labeled AB2 was exogenously
added to ameloblast lineage cells (ALC) and its localization was determined using confocal
microscopy. Figure 4.15 shows AB2-FITC localizing on the ALC membranes. Figure 4.15 B is the
green channel showing AB2-FITC on ALC and Figure 4.15 C shows DiD labeled cell membranes
of ALC. Figure 4.15 A is the merged image showing AB2 colocalizing with the DiD labeled cell
membrane. Ambn peptide AB2 can bind to ameloblast lineage cell membrane directly, confirming
Figure 4.15. Confocal image showing FITC labeled AB2 colocalizing with ALC membrane. A. Merged
image of AB2-FITC colocalizing with cell membrane (DiD), B. AB2-FITC channel, C. DiD channel.
97

the in vitro finding that Ambn can bind to ameloblast cell membrane through its exon 5 encoded
region.
Note: Ambn-cell membrane colocalization data are included in our new manuscript titled ‘An
evolutionarily conserved helix mediates ameloblastin-cell interaction’, Jingtan Su, Rucha Arun
Bapat, Gayathri Visakan, Janet Moradian-Oldak, Journal of Dental Research, accepted, March
2020.
The protocol for Amel-Ambn co-labeling and colocalization in mouse incisors was published as a
book chapter: Bapat, R.A., and Moradian-Oldak, J. (2019). Immunohistochemical Co-Localization
of Amelogenin and Ameloblastin in Developing Enamel Matrix. In Odontogenesis (Springer), pp.
219-228.
4.4 Discussion
In this chapter, I analyzed the co-localization of Amel-Ambn, and Ambn-ameloblast cell
membrane in developing mouse incisors. The continuously growing mouse incisor presents an
opportunity to visualize all stages of amelogenesis in a single sample. The changes in
colocalization of Amel and Ambn can be compared between secretory, transition and maturation
stages in the mouse incisor as they serve as direct internal controls for each other.
Amel and Ambn colocalization- In Chapters 2 and 3, I determined that Amel and Ambn can bind
to each other directly in vitro. Their co-localization in vivo confirms that they perform cooperative
functions during enamel development. Colocalization of Ambn with ameloblast cell membrane
suggests a possible function of Ambn as an adhesion molecule between ameloblasts and
amelogenin containing enamel extracellular matrix.
98

In this Chapter, Manders’ colocalization coefficient (MCC) (Manders et al., 1993) was used to
quantify colocalization between Amel and Ambn. MCC has been previously used to quantify
Amel-enamelin and Amel-Ambn colocalization in mouse molars (Gallon et al., 2013; Mazumder
et al., 2014). MCC is different from Pearson’s correlation coefficient or colocalization ‘rate’ (Dunn
et al., 2011) because it eliminates the bias created by differences in the signal intensities, by
calculating separate colocalization coefficients for each channel being analyzed. This makes
MCC one of the more accurate measures of colocalization.
Full-length Ambn colocalizes with Amel within ameloblasts and at Tomes’ processes in the
secretory stage, and only within ameloblasts in transition and maturation stages (Figures 4.5 and
4.6). The M300 antibody epitope (Figure 4.4) excludes the N-terminus of Ambn, allowing the
visualization Ambn which has its C-terminus intact. Therefore the signal from this antibody is
mainly observed within ameloblasts and at the Tomes’ processes where full-length Ambn is
abundant. Within ameloblasts, the Ambn signal intensity is much stronger than that of Amel. This
could be because when Ambn is packaged inside vesicles the C-terminal portion of Ambn is more
exposed and N-terminus is hidden (possibly buried in surrounding hydrophobic regions of the
protein maybe due to self-assembly). It could be a mechanism to prevent untimely Amel-Ambn
binding within the secretory vesicles in which they are packaged together (Zalzal et al., 2008).
Preventing Amel-Ambn interactions might be important for allowing free Ambn molecules to act
as signaling molecules. A cell signaling function of Ambn has been suggested by (Fukumoto et
al., 2004; Fukumoto et al., 2005) and the reason Ambn signal is stronger inside ameloblast could
also be because it is actively participating in intracellular functions as well as being secreted, while
all of synthesized Amel is constantly being secreted.
The N-18 antibody labels the fragments containing N-terminus of Ambn, including full-length
Ambn. The exact sequence of the epitope for this antibody is unknown but it approximately lies
at the N-terminus of the M300 epitope. Using this antibody Ambn was labeled throughout the
99

thickness of developing enamel and colocalization was show within ameloblasts, at the secretory
face of ameloblasts (Tomes’ processes), and within the enamel matrix in secretory stage (Figures
4.7 and 4.8). Colocalization between Amel and Ambn fragments continued in transition and
maturation stages, with no significant changes in Manders’ colocalization coefficients. The signal
intensity for both Amel and Ambn increased in maturation stage enamel as compared to secretory
stage, due to accumulation of proteolytic cleavage products of proteins. Within ameloblasts, the
signal intensity for anti-Ambn N-18 antibody was significantly lower as compared to anti-Ambn
M300 antibody labeling. This supports the notion proposed earlier that N-terminal fragments of
Ambn are not exposed within ameloblasts.
Figures 4.9 and 4.10 show colocalization of Amel-Ambn on the prism architecture of mature
enamel. Figure 4.10 also shows Amel and Ambn colocalizing in the characteristic ladder like
arrangement of mouse enamel matrix, which is disrupted in mice with mutant Ambn (Wald et al.,
2017) lacking a functional self-assembly domain. This suggests that Amel-Ambn interaction and
co-assembly is required for the organization of Amel in the extracellular matrix.
Technical Challenges of Membrane Staining- To identify the colocalization between Ambn and
ameloblast cell membrane, I stained ameloblast cell membranes with DiD and Ambn with anti-
Ambn M300 antibody. The common method of labeling cell membranes in formalin fixed paraffin
embedded (FFPE) tissue is the use of wheat germ agglutinin (lectin) conjugates which bind to
glycolipids or glycoproteins on the cell surface. However, Ambn and enamel matrix protein
enamelin are heavily glycosylated (Kobayashi et al., 2007) and can bind to lectins. Therefore to
avoid false positive results DiD was adapted to be used in FFPE tissue. DiD is a lipophilic organic
dye, which interacts with many chemicals used in tissue processing for FFPE tissues, including
Triton X-100 and paraffin itself. Commonly used coverslip mounting media contain glycerol or
xylene which also interact with DiD. I overcame these challenges by using Triton X-100 before
DiD, to prevent it from leaching out and eliminating mounting media for coverslips. The downside
100

of this was that the slides had to be imaged immediately, bleached easily from the confocal lasers,
and could not be stored for more than 2-3 days.
The resulting images showed the cell membranes of all the cells in the samples stained
specifically with DiD, while Ambn was co-labeled with M300 antibody. Ambn colocalized with
apical end of ameloblast cell membrane from secretory to maturation stage (Figure 4.12).
Ambn-Cell Membrane Interactions- Fukumoto et al. (2004) have shown that non-functional
ameloblastin causes ameloblasts to separate from the underlying enamel matrix. The true Ambn
KO mouse (Liang et al., 2019) also displayed a similar phenotype. By showing that Ambn binds
to Amel in vitro and colocalizes with Amel and ameloblast cell membrane in vivo at the apical end,
we provide evidence that Ambn acts as a cell adhesion molecule between ameloblasts and Amel-
rich enamel extracellular matrix. The role of enamel matrix proteins, particularly Ambn, in
maturation stage ameloblast needs to be further investigated
Figure 4.15 showed exon 5 encoded peptide AB2 binds to the cell membranes of ameloblast
lineage cells. This provided in vivo evidence for biophysical studies conducted by Su et al. (2019b)
concluding that exon 5 encoded fragment of Ambn interacts directly with cell membranes via an
amphipathic helix motif. N-terminal four amino acids of the amphipathic helix motif overlap with
the C-terminus of the Ambn self-assembly domain (Figure 4.2). How the same motif performs
both functions of Ambn-Amel binding and Ambn-cell binding is still unknown. The intrinsically
disordered nature of Ambn must be playing an important part in making this multi-targeting
possible. Zalzal et al. (2008) found that not all the Ambn was packaged with Amel during secretion
but there were some secretory vesicles containing Ambn only. The independently packaged
Ambn could have a different function (cell binding function), than Ambn packaged with Amel which
would perform the biomineralization function.
101

We conclude that one of the functions of Amel-Ambn interaction is to maintain the adhesion
between enamel extracellular matrix and ameloblasts.

102

5. Conclusions and future directions
Dental enamel is a unique ectodermal hard tissue which forms from a transient protein matrix
secreted by ameloblast cells. Mature enamel is cell-free and unlike bone or dentin cannot
physiologically repair or regenerate. Mutations in enamel matrix proteins cause a disease called
Amelogenesis Imperfecta (AI), which typically requires patients to undergo intensive full-mouth
rehabilitation. To understand enamel formation in health and disease, studying enamel matrix
proteins and their interactions is of utmost importance.
Amelogenin (Amel) and ameloblastin (Ambn) are two of the most abundant proteins in the
developing enamel matrix. Amel and Ambn are co-secreted through the same secretory vesicles
and colocalize in the developing molars (Mazumder et al., 2014; Zalzal et al., 2008). The aim of
this project was to establish amelogenin and ameloblastin interaction in vitro (Aims I and II) and
in vivo (Aim III) in order to better understand their independent and cooperative functions during
enamel formation.
I identified direct binding between recombinant and native Amel and Ambn using co-
immunoprecipitation. I characterized the Amel-binding domain on Ambn to be at the N-terminus
of its exon 5 encoded region. I then determined that Amel and Ambn co-assemble to form
spherical nanostructures in vitro. Lastly, I analyzed the colocalization between Amel and Ambn,
as well as Ambn and ameloblast cell membranes in vivo, in developing mouse incisor. I concluded
that Amel and Ambn interact during enamel formation and their cooperative functions may include
maintaining enamel rod architecture and adhering ameloblast cells to the enamel matrix. The
scope of my three specific aims are summarized in Scheme 5.1.

103

Scheme 5.1. Model summarizing the aims and conclusions of this project. Amel-Ambn interact
through their N-terminal regions, and co-assemble to form spherical assemblies (Aims I and II).
Their co-assemblies are detected in vivo at the secretory face of ameloblasts.
104

Previous studies from our laboratory have shown the Amel-Ambn colocalization in developing
mouse molars (Mazumder et al., 2014) but data on direct protein-protein interactions have been
limited. Fluorescent resonance energy transfer (FRET) data from Mazumder et al. (2016) showed
that Amel and Ambn N-terminal fragments are 5-7 nm apart in enamel rod sheath space which
strongly suggested their interaction. To confirm this finding, my first aim for this project was to
detect in vitro evidence of direct binding between Amel and Ambn. Co-immunoprecipitation (co-
IP) is an ideal technique because it requires minimal quantities of proteins and can be adapted to
be used with recombinant as well as native protein extracts. In order to eliminate the probability
of false positive results I applied both Amel and Ambn as the bait proteins for co-IP experiments
and used recombinant as well as native proteins. It was observed that recombinant (not post-
translationally modified) and native (with post-translational modifications) Amel and Ambn bind to
each other and co-elute from the co-IP columns. Porcine enamel matrix protein extract was used
for the native protein co-IP, which contains a mixture of alternatively spliced isoforms and
cleavage products of proteins (Hu et al., 1997; Yamakoshi et al., 2001). The Western blots used
for detecting the constituents of co-IP elution fractions were not suitable for differentiating the
isoforms and cleavage products from the most abundant protein fragments. Therefore, we
collaborated with the Agilent Center for Excellence in Biomolecular Characterization at USC, to
analyze the elution fractions by mass spectrometry. The fragments and alternatively spliced
isoforms of Amel were identified from the mass spectra based on previous literature (Fincham et
al., 1991; 1994; Moradian-Oldak, 1993). However, the identification of Ambn fragments in the
elution fraction was more challenging. Ambn fragments from porcine enamel matrix have been
sequenced (Chun et al., 2010; Iwata et al., 2007) but their calculated molecular weights are
considerably lower than their native masses due to post-translational modifications and the exact
differences are not fully known. One of the steps in the near future would be to identify and
characterize Ambn fragments from native Amel-Ambn co-IP elution fractions.
105

The mass spec data would also confirm the second in vitro finding of this thesis which is, Ambn
interacts with Amel through the N-terminus of its exon 5 encoded region. Ambn exon 5 and 6
deletion mutants, and synthetic peptides encoding exons 5 and 6 were employed to determine
which regions of Ambn are essential for interaction with Amel. The N-terminal 20 amino acid
region from the exon 5 encoded peptide was found to bind directly to Amel. In the absence of this
region, Ambn could not bind to Amel. Notably, this region harbors the Ambn self-assembly motif
Y/F-XX-L/F/Y-x-Y/F (Wald et al., 2017). Based on my in vitro data, it is reasonable to suggest that
this self-assembly motif functions as a co-assembly motif between Amel and Ambn. It would be
interesting to test the binding of Ambn synthetic peptides with single amino acid mutations in the
self-assembly motif, to confirm whether it acts as a co-assembly motif. Specially, the Ambn
mutation used by Wald et al. (2017) replacing the key tyrosine residues of the self-assembly
domain with glycine would be an important candidate to analyze with co-IP, to determine the
effects of the mutation on Amel binding.
Co-assembly of Amel and Ambn was analyzed by dynamic light scattering (DLS) and
transmission electron microscopy (TEM). Spherical co-assemblies approximately 12 nm in
diameter were detected from Amel-Ambn mixture at pH 7.5 and 22° C. Spherical beaded
structures lining newly forming enamel prisms have been previously detected in enamel matrix in
vivo (Wen et al., 2001), however their constituent proteins have not been analyzed. We have
already analyzed Amel-Ambn colocalization in vivo at the micron scale (Aim III). Re-examining
the protein nanosphere structures in vivo using immuno-electron microscopy techniques, to
identify the constituent proteins would provide nano-scale evidence of Amel-Ambn binding in
enamel formation and support our hypothesis that Amel-Ambn co-assemblies are required for
association of Amel with HAP crystals to maintain enamel rod architecture.
To analyze the in vivo colocalization of Amel-Ambn at the micron-scale, I used the continuously
growing mouse incisor model. I used two different antibodies against Ambn and co-labeled it with
106

Amel to conclude Amel and N-terminal fragments of Ambn colocalize from secretory to maturation
stage of enamel development within ameloblasts and in the entire thickness of the enamel matrix.
Ambn also colocalizes with ameloblast cell membrane at Tomes’ processes in secretory stage
and continues to colocalize with the apical end of ameloblasts in transition stage when very little
protein is actively being secreted. Based on these data and earlier studies showing ameloblasts
detach from underlying matrix in Ambn mutant mice (Fukumoto et al., 2004; Liang et al., 2019),
the function of Ambn as an adhesion molecule between ameloblasts and Amel rich enamel matrix
was proposed.
In conclusion, although the majority of enamel extracellular matrix is formed by Amel, it cannot
function alone. In Amel knockout mice, although thin and disordered, enamel still forms and
ameloblasts are normal (Gibson et al., 2001). In Ambn knockout or mutant mice, true enamel is
absent and instead calcified nodules are observed within a distorted layer of ameloblasts
(Fukumoto et al., 2004; Liang et al., 2019). Amel is a relatively small hydrophobic, intrinsically
disordered protein which forms supramolecular structures to architect enamel rod-interrods.
However, the hydrophobicity and early C-terminal (hydrophilic calcium binding region) cleavage
of Amel raises the question of how it interacts with hydroxyapatite and maintains enamel prism
structure. Interaction of Amel with Ambn (and possibly other calcium binding acidic enamel matrix
protein enamelin) bridges this gap and expands our understanding of the structural role of Amel
and its fragments in enamel formation. Based on the findings of my project, we suggest that a
dynamic interaction between amelogenins and non-amelogenins (particularly Ambn) is required
for enamel formation, in a manner similar to how collagen requires non-collagenous proteins for
bone and dentin formation (Roach, 1994).

107

Bibliography
Baldassarri, M., Margolis, H., and Beniash, E. (2008). Compositional determinants of mechanical
properties of enamel. Journal of dental research 87, 645-649.
Bapat, R.A., and Moradian-Oldak, J. (2019). Immunohistochemical Co-Localization of
Amelogenin and Ameloblastin in Developing Enamel Matrix. In Odontogenesis (Springer), pp.
219-228.
Bartlett, J.D., and Simmer, J.P. (1999). Proteinases in developing dental enamel. Crit Rev Oral
Biol Med 10, 425-441.
Beyeler, M., Schild, C., Lutz, R., Chiquet, M., and Trueb, B. (2010). Identification of a fibronectin
interaction site in the extracellular matrix protein ameloblastin. Experimental cell research 316,
1202-1212.
Bidlack, F.B., Xia, Y., and Pugach, M.K. (2017). Dose-dependent rescue of KO amelogenin
enamel by transgenes in vivo. Frontiers in physiology 8, 932.
Bromley, K.M., Kiss, A.S., Lokappa, S.B., Lakshminarayanan, R., Fan, D., Ndao, M., Evans, J.S.,
and Moradian-Oldak, J. (2011). Dissecting amelogenin protein nanospheres: characterization of
metastable oligomers. J Biol Chem 286, 34643-34653.
Brookes, S.J., Kingswell, N.J., Barron, M.J., Dixon, M.J., and Kirkham, J. (2011). Is the 32-kDa
fragment the functional enamelin unit in all species? Eur J Oral Sci 119 Suppl 1, 345-350.
Brookes, S.J., Kirkham, J.L., S.P., Shore, R.C., Wood, S.R., and Robinson, C. (2000). Spatially
related amelogenin interactions in developing rat enamel as revealed by molecular cross-linking
studies.pdf>. Archives of Oral Biology 45, 937-943.
Černý, R., Slaby, I., Hammarström, L., and Wurtz, T. (1996). A novel gene expressed in rat
ameloblasts codes for proteins with cell binding domains. Journal of bone and mineral research
11, 883-891.
Chen, C.-L., Bromley, K.M., Moradian-Oldak, J., and DeYoreo, J.J. (2011). In situ AFM study of
amelogenin assembly and disassembly dynamics on charged surfaces provides insights on
matrix protein self-assembly. Journal of the American Chemical Society 133, 17406-17413.
Chun, Y.H., Yamakoshi, Y., Yamakoshi, F., Fukae, M., Hu, J.C., Bartlett, J.D., and Simmer, J.P.
(2010). Cleavage site specificity of MMP-20 for secretory-stage ameloblastin. J Dent Res 89, 785-
790.
Delak, K., Harcup, C., Lakshminarayanan, R., Sun, Z., Fan, Y., Moradian-Oldak, J., and Evans,
J.S. (2009). The tooth enamel protein, porcine amelogenin, is an intrinsically disordered protein
with an extended molecular configuration in the monomeric form. Biochemistry 48, 2272-2281.
Du, C., Falini, G., Fermani, S., Abbott, C., and Moradian-Oldak, J. (2005). Supramolecular
Assembly of Amelogenin Nanospheres into Birefringent Microribbons. Science 307, 1450-1454.
108

Dunn, K.W., Kamocka, M.M., and McDonald, J.H. (2011). A practical guide to evaluating
colocalization in biological microscopy. Am J Physiol Cell Physiol 300, C723-742.
Fan, D., Du, C., Sun, Z., Lakshminarayanan, R., and Moradian-Oldak, J. (2009). In vitro study on
the interaction between the 32kDa enamelin and amelogenin. Journal of Structural Biology 166,
88-94.
Fan, D., Iijima, M., Bromley, K.M., Yang, X., Mathew, S., and Moradian-Oldak, J. (2011). The
cooperation of enamelin and amelogenin in controlling octacalcium phosphate crystal
morphology. Cells Tissues Organs 194, 194-198.
Fang, P.-A., Conway, J.F., Margolis, H.C., Simmer, J.P., and Beniash, E. (2011). Hierarchical
self-assembly of amelogenin and the regulation of biomineralization at the nanoscale.
Proceedings of the National Academy of Sciences 108, 14097-14102.
Fincham, A., Hu, Y., Lau, E., Slavkin, H., and Snead, M. (1991). Amelogenin post-secretory
processing during biomineralization in the postnatal mouse molar tooth. Archives of oral biology
36, 305-317.
Fincham, A., Moradian-Oldak, J., Diekwisch, T., Lyaruu, D., Wright, J., Bringas Jr, P., and Slavkin,
H. (1995). Evidence for amelogenin" nanospheres" as functional components of secretory-stage
enamel matrix. Journal of structural biology 115, 50-59.
Fincham, A., Moradian-Oldak, J., and Sarte, P. (1994). Mass-spectrographic analysis of a porcine
amelogenin identifies a single phosphorylated locus. Calcified tissue international 55, 398-400.
Fincham, A.G., and Moradian-Oldak, J. (1993). Amelogenin post-translational modifications:
carboxy-terminal processing and the phosphorylation of bovine and porcine "TRAP" and "LRAP"
amelogenins. Biochemical and biophysical research communications 197, 248-255.
Fischer, A.H., Jacobson, K.A., Rose, J., and Zeller, R. (2006). Preparation of cells and tissues for
fluorescence microscopy. Basic Methods in Microscopy: Protocols and Concepts from Cells: A
Laboratory Manual; Spectro, DL, Goldman, RD, Eds.
Fukumoto, S., Kiba, T., Hall, B., Iehara, N., Nakamura, T., Longenecker, G., Krebsbach, P.H.,
Nanci, A., Kulkarni, A.B., and Yamada, Y. (2004). Ameloblastin is a cell adhesion molecule
required for maintaining the differentiation state of ameloblasts. J Cell Biol 167, 973-983.
Fukumoto, S., Yamada, A., Nonaka, K., and Yamada, Y. (2005). Essential roles of ameloblastin
in maintaining ameloblast differentiation and enamel formation. Cells Tissues Organs 181, 189-
195.
Gallon, V., Chen, L., Yang, X., and Moradian-Oldak, J. (2013). Localization and quantitative co-
localization of enamelin with amelogenin. J Struct Biol 183, 239-249.
Gasteiger, E., Hoogland, C., Gattiker, A., Wilkins, M.R., Appel, R.D., and Bairoch, A. (2005).
Protein identification and analysis tools on the ExPASy server. In The proteomics protocols
handbook (Springer), pp. 571-607.
109

Gibson, C.W., Yuan, Z.A., Hall, B., Longenecker, G., Chen, E., Thyagarajan, T., Sreenath, T.,
Wright, J.T., Decker, S., Piddington, R., et al. (2001). Amelogenin-deficient mice display an
amelogenesis imperfecta phenotype. J Biol Chem 276, 31871-31875.
Glimcher, M., Brickley-Parsons, D., and Levine, P. (1977). Studies of enamel proteins during
maturation. Calcified tissue research 24, 259-270.
Green, D.R., Schulte, F., Lee, K.H., Pugach-Gordon, M., Hardt, M., and Bidlack, F.B. (2019).
Mapping the tooth enamel proteome and amelogenin phosphorylation onto mineralizing porcine
tooth crowns. Frontiers in physiology 10, 925.
Hassan, P.A., Rana, S., and Verma, G. (2015). Making sense of Brownian motion: colloid
characterization by dynamic light scattering. Langmuir 31, 3-12.
Hatakeyama, J., Fukumoto, S., Nakamura, T., Haruyama, N., Suzuki, S., Hatakeyama, Y., Shum,
L., Gibson, C.W., Yamada, Y., and Kulkarni, A.B. (2009). Synergistic roles of amelogenin and
ameloblastin. J Dent Res 88, 318-322.
Hirose, N., Shimazu, A., Watanabe, M., Tanimoto, K., Koyota, S., Sugiyama, T., Uchida, T., and
Tanne, K. (2013). Ameloblastin in Hertwig’s epithelial root sheath regulates tooth root formation
and development. PloS one 8, e54449.
Holcroft, J., and Ganss, B. (2011). Identification of Amelotin‐and ODAM‐interacting enamel matrix
proteins using the yeast two‐hybrid system. European journal of oral sciences 119, 301-306.
Hu, C.C., Fukae, M., Uchida, T., Qian, Q., Zhang, C.H., Ryu, O.H., Tanabe, T., Yamakoshi, Y.,
Murakami, C., Dohi, N., et al. (1997). Sheathlin: cloning, cDNA/polypeptide sequences, and
immunolocalization of porcine enamel sheath proteins. J Dent Res 76, 648-657.
Hu, J.C.-C., Chan, H.-C., Simmer, S.G., Seymen, F., Richardson, A.S., Hu, Y., Milkovich, R.N.,
Estrella, N.M., Yildirim, M., and Bayram, M. (2012). Amelogenesis imperfecta in two families with
defined AMELX deletions in ARHGAP6. PLoS One 7.
Hu, Y., Smith, C.E., Cai, Z., Donnelly, L.A., Yang, J., Hu, J.C., and Simmer, J.P. (2016). Enamel
ribbons, surface nodules, and octacalcium phosphate in C57BL/6 Amelx(-/-) mice and Amelx(+/-
) lyonization. Mol Genet Genomic Med 4, 641-661.
Iijima, M., and Moradian-Oldak, J. (2005). Control of apatite crystal growth in a fluoride containing
amelogenin-rich matrix. Biomaterials 26, 1595-1603.
Iijima, M., and Moradian–Oldak, J. (2004). Interactions of amelogenins with octacalcium
phosphate crystal faces are dose dependent. Calcified tissue international 74, 522-531.
Iwasaki, K., Bajenova, E., Somogyi-Ganss, E., Miller, M., Nguyen, V., Nourkeyhani, H., Gao, Y.,
Wendel, M., and Ganss, B. (2005). Amelotin—a novel secreted, ameloblast-specific protein.
Journal of dental research 84, 1127-1132.
Iwata, T., Yamakoshi, Y., Hu, J.C., Ishikawa, I., Bartlett, J.D., Krebsbach, P.H., and Simmer, J.P.
(2007). Processing of ameloblastin by MMP-20. J Dent Res 86, 153-157.
110

Kobayashi, K., Yamakoshi, Y., Hu, J.-C., Gomi, K., Arai, T., Fukae, M., Krebsbach, P., and
Simmer, J. (2007). Splicing determines the glycosylation state of ameloblastin. Journal of dental
research 86, 962-967.
Krebsbach, P.H., Lee, S.K., Matsuki, Y., Kozak, C.A., Yamada, K.M., and Yamada, Y. (1996).
Full-length sequence, localization, and chromosomal mapping of ameloblastin. A novel tooth-
specific gene. J Biol Chem 271, 4431-4435.
Kwak, S.-Y., Wiedemann-Bidlack, F.B., Beniash, E., Yamakoshi, Y., Simmer, J.P., Litman, A.,
and Margolis, H.C. (2009). Role of 20-kDa amelogenin (P148) phosphorylation in calcium
phosphate formation in vitro. Journal of Biological Chemistry 284, 18972-18979.
Lacruz, R.S., Nakayama, Y., Holcroft, J., Nguyen, V., Somogyi-Ganss, E., Snead, M.L., White,
S.N., Paine, M.L., and Ganss, B. (2012). Targeted overexpression of amelotin disrupts the
microstructure of dental enamel. PLoS One 7, e35200.
Lacruz, R.S., Nanci, A., Kurtz, I., Wright, J.T., and Paine, M.L. (2010). Regulation of pH During
Amelogenesis. Calcif Tissue Int 86, 91-103.
Li, Y., Suggs, C., Wright, J.T., Yuan, Z.-a., Aragon, M., Fong, H., Simmons, D., Daly, B., Golub,
E.E., and Harrison, G. (2008). Partial rescue of the amelogenin null dental enamel phenotype.
Journal of Biological Chemistry 283, 15056-15062.
Liang, T., Hu, Y., Smith, C.E., Richardson, A.S., Zhang, H., Yang, J., Lin, B., Wang, S.K., Kim,
J.W., and Chun, Y.H. (2019). AMBN mutations causing hypoplastic amelogenesis imperfecta and
Ambn knockout‐NLS‐lacZ knockin mice exhibiting failed amelogenesis and Ambn tissue‐
specificity. Molecular genetics & genomic medicine 7, e929.
Lu, T., Li, M., Xu, X., Xiong, J., Huang, C., Zhang, X., Hu, A., Peng, L., Cai, D., and Zhang, L.
(2018). Whole exome sequencing identifies an AMBN missense mutation causing severe
autosomal-dominant amelogenesis imperfecta and dentin disorders. International journal of oral
science 10, 1-9.
Lu, Y., Papagerakis, P., Yamakoshi, Y., Hu, J.C., Bartlett, J.D., and Simmer, J.P. (2008).
Functions of KLK4 and MMP-20 in dental enamel formation. Biol Chem 389, 695-700.
MacDougall, M., Simmons, D., Gu, T.T., Forsman‐Semb, K., Kärrman Mårdh, C., Mesbah, M.,
Forest, N., Krebsbach, P.H., Yamada, Y., and Berdal, A. (2000). Cloning, characterization and
immunolocalization of human ameloblastin. European journal of oral sciences 108, 303-310.
Manders, E., Verbeek, F., and Aten, J. (1993). Measurement of co‐localization of objects in dual‐
colour confocal images. Journal of microscopy 169, 375-382.
Markham, K., Bai, Y., and Schmitt-Ulms, G. (2007). Co-immunoprecipitations revisited: an update
on experimental concepts and their implementation for sensitive interactome investigations of
endogenous proteins. Analytical and bioanalytical chemistry 389, 461-473.
Mazumder, P., Prajapati, S., Bapat, R., and Moradian-Oldak, J. (2016). Amelogenin-Ameloblastin
Spatial Interaction around Maturing Enamel Rods. J Dent Res 95, 1042-1048.
111

Mazumder, P., Prajapati, S., Lokappa, S.B., Gallon, V., and Moradian-Oldak, J. (2014). Analysis
of co-assembly and co-localization of ameloblastin and amelogenin. Front Physiol 5, 274.
Meckel, A., Griebstein, W., and Neal, R. (1965). Structure of mature human dental enamel as
observed by electron microscopy. Archives of oral biology 10, 775-783.
Moffatt, P., Smith, C.E., St‐Arnaud, R., and Nanci, A. (2008). Characterization of Apin, a secreted
protein highly expressed in tooth‐associated epithelia. Journal of cellular biochemistry 103, 941-
956.
Møinichen, C., Lyngstadaas, S., and Risnes, S. (1996). Morphological characteristics of mouse
incisor enamel. Journal of anatomy 189 (Pt 2), 325-333.
Moradian-Oldak, A.G.F.a.J. (1993). Amelogenin Post-translational Modifications. Biochemical
and Biophysical Research Communications 197, 248-255.
Moradian-Oldak, J. (2001). Amelogenins: assembly, processing and control of crystal
morphology. Matrix Biol 20, 293-305.
Moradian-Oldak, J. (2012). Protein-mediated enamel mineralization. Frontiers in bioscience: a
journal and virtual library 17, 1996.
Moradian-Oldak, J., Bouropoulos, N., Wang, L., and Gharakhanian, N. (2002). Analysis of self-
assembly and apatite binding properties of amelogenin proteins lacking the hydrophilic C-
terminal. Matrix Biology 21, 197-205.
Moradian-Oldak, J., and Goldberg, M. (2005). Amelogenin supra-molecular assembly in vitro
compared with the architecture of the forming enamel matrix. Cells Tissues Organs 181, 202-218.
Moradian-Oldak, J., Iijima, M., Bouropoulos, N., and Wen., H.B. (2003). Assembly of Amelogenin
Proteolytic Products and Control of Octacalcium Phosphate Crystal Morphology. Connective
Tissue Research 44, 58-64.
Moradian-Oldak, J., Leung, W., Simmer, J.P., Zeichner-David, M., and Fincham, A.G. (1996).
Identification of a novel proteinase (ameloprotease-I) responsible for the complete degradation of
amelogenin during enamel maturation. Biochemical Journal 318, 1015-1021.
Moradian-Oldak, J., Paine, M.L., Lei, Y.P., Fincham, A.G., and Snead, M.L. (2000). Self-assembly
properties of recombinant engineered amelogenin proteins analyzed by dynamic light scattering
and atomic force microscopy. J Struct Biol 131, 27-37.
Moradian-Oldak, J., Tan, J., and Fincham, A.G. (1998). Interaction of Amelogenin with
Hydroxyappetite Crystals. Biopolymers 46, 225-238.
Moradian‐Oldak, J., Simmer, J., Lau, E., Sarte, P., Slavkin, H., and Fincham, A. (1994). Detection
of monodisperse aggregates of a recombinant amelogenin by dynamic light scattering.
Biopolymers: Original Research on Biomolecules 34, 1339-1347.
112

Nakata, A., Kameda, T., Nagai, H., Ikegami, K., Duan, Y., Terada, K., and Sugiyama, T. (2003).
Establishment and characterization of a spontaneously immortalized mouse ameloblast-lineage
cell line. Biochemical and biophysical research communications 308, 834-839.
Nanci, A. (2012). Ten Cate's Oral Histology: Development, Structure, and Function, 8th edn
(Elsevier Health Sciences).
Nanci, A., Zalzal, S., Lavoie, P., Kunikata, M., Chen, W., Krebsbach, P.H., Yamada, Y.,
Hammarstrom, L., Simmer, J.P., Fincham, A.G., et al. (1998). Comparative immunochemical
analyses of the developmental expression and distribution of ameloblastin and amelogenin in rat
incisors. J Histochem Cytochem 46, 911-934.
Paine, M.L., Krebsbach, P.H., Chen, L.S., Paine, C.T., Yamada, Y., Deutsch, D., and Snead, M.L.
(1998). Protein-Protein Interactions- Criteria Defining the Assembly of the Enamel Organic Matrix.
Journal of Dental Research 77, 496-502.
Paine, M.L., and Snead, M.L. (1997). Protein interactions during assembly of the enamel organic
extracellular matrix. Journal of Bone and Mineral Research 12, 221-227.
Paine, M.L., Wang, H.-J., Luo, W., Krebsbach, P.H., and Snead, M.L. (2003). A transgenic animal
model resembling amelogenesis imperfecta related to ameloblastin overexpression. Journal of
Biological Chemistry 278, 19447-19452.
Paine, M.L., Zhu, D.-H., Luo, W., Bringas Jr, P., Goldberg, M., White, S.N., Lei, Y.-P., Sarikaya,
M., Fong, H.K., and Snead, M.L. (2000). Enamel biomineralization defects result from alterations
to amelogenin self-assembly. Journal of structural biology 132, 191-200.
Park, J.C., Park, J.T., Son, H.H., Kim, H.J., Jeong, M.J., Lee, C.S., Dey, R., and Cho, M.l. (2007).
The amyloid protein APin is highly expressed during enamel mineralization and maturation in rat
incisors. European journal of oral sciences 115, 153-160.
Poulter, J.A., Murillo, G., Brookes, S.J., Smith, C.E.L., Parry, D.A., Silva, S., Kirkham, J.,
Inglehearn, C.F., and Mighell, A.J. (2014). Deletion of ameloblastin exon 6 is associated with
amelogenesis imperfecta. Human Molecular Genetics 23, 5317-5324.
Prajapati, S., Tao, J., Ruan, Q., De Yoreo, J.J., and Moradian-Oldak, J. (2016). Matrix
metalloproteinase-20 mediates dental enamel biomineralization by preventing protein occlusion
inside apatite crystals. Biomaterials 75, 260-270.
Prasad, M.K., Geoffroy, V., Vicaire, S., Jost, B., Dumas, M., Le Gras, S., Switala, M., Gasse, B.,
Laugel-Haushalter, V., and Paschaki, M. (2016). A targeted next-generation sequencing assay
for the molecular diagnosis of genetic disorders with orodental involvement. Journal of Medical
Genetics 53, 98-110.
Ravindranath, H.H., Chen, L.-S., Zeichner-David, M., Ishima, R., and Ravindranath, R.M. (2004).
Interaction between the enamel matrix proteins amelogenin and ameloblastin. Biochemical and
biophysical research communications 323, 1075-1083.
113

Ravindranath, R.M., Moradian-Oldak, J., and Fincham, A.G. (1999). Tyrosyl Motif in Amelogenins
BindsN-Acetyl-D-glucosamine. Journal of Biological Chemistry 274, 2464-2471.
Roach, H. (1994). Why does bone matrix contain non‐collagenous proteins? The possible roles
of osteocalcin, osteonectin, osteopontin and bone sialoprotein in bone mineralisation and
resorption. Cell biology international 18, 617-628.
Robinson, C., Weatherell, J., and Hallsworth, A. (1971). Variation in composition of dental enamel
within thin ground tooth sections. Caries research 5, 44-57.
Ryu, O.H., Fincham, A.G., Hu, C.-C., Zhang, C., Qian, Q., Bartlett, J.D., and Simmer, J.P. (1999).
Characterization of Recombinant Pig Enamelysin Activity and Cleavage of Recombinant Pig and
Mouse Amelogenins. Journal of Dental Research 78, 743-750.
Sasaki, T., Takagi, M., and Yanagisawa, T. (2007). Structure and function of secretory
ameloblasts in enamel formation. Paper presented at: Ciba Foundation Symposium 205‐Dental
Enamel: Dental Enamel: Ciba Foundation Symposium 205 (Wiley Online Library).
Shapiro, J., Wen, X., Okamoto, C., Wang, H., Lyngstadaas, S., Goldberg, M., Snead, M., and
Paine, M. (2007). Cellular uptake of amelogenin, and its localization to CD63, and Lamp1-positive
vesicles. Cellular and molecular life sciences 64, 244-256.
Shibata, S., Suzuki, S., Tengan, T., and Yamashita, Y. (1995). A histochemical study of apoptosis
in the reduced ameloblasts of erupting mouse molars. Archives of oral biology 40, 677-680.
Shirai, A., Matsuyama, A., Yashiroda, Y., Hashimoto, A., Kawamura, Y., Arai, R., Komatsu, Y.,
Horinouchi, S., and Yoshida, M. (2008). Global analysis of gel mobility of proteins and its use in
target identification. Journal of Biological Chemistry 283, 10745-10752.
Simmer, J.P., and Fincham, A.G. (1995). Molecular Mechanisms of Dental Enamel Formation.
Critical Reviews in Oral Biology & Medicine 6, 84-108.
Simmer, J.P., Richardson, A.S., Hu, Y.-Y., Smith, C.E., and Hu, J.C.-C. (2012). A post-classical
theory of enamel biomineralization… and why we need one. International journal of oral science
4, 129-134.
Simpson, R.J. (2006). SDS-PAGE of proteins. Cold Spring Harbor Protocols 2006, pdb. prot4313-
pdb. prot4313.
Skobe, Z. (2006). SEM evidence that one ameloblast secretes one keyhole‐shaped enamel rod
in monkey teeth. European journal of oral sciences 114, 338-342.
Smales, F. (1975). Structural subunit in prisms of immature rat enamel. Nature 258, 772-774.
Smith, C. (1998). Cellular and chemical events during enamel maturation. Critical Reviews in Oral
Biology & Medicine 9, 128-161.
114

Smith, C.E., Poulter, J.A., Antanaviciute, A., Kirkham, J., Brookes, S.J., Inglehearn, C.F., and
Mighell, A.J. (2017). Amelogenesis imperfecta; genes, proteins, and pathways. Frontiers in
physiology 8, 435.
Snead, M.L., Lau, E.C., Zeichner-David, M., Fincham, A.G., Woo, S.L., and Slavkin, H.C. (1985).
DNA sequence for cloned cDNA for murine amelogenin reveal the amino acid sequence for
enamel-specific protein. Biochemical and biophysical research communications 129, 812-818.
Snead, M.L., Zhu, D.-H., Lei, Y., Luo, W., Bringas Jr, P.O., Sucov, H.M., Rauth, R.J., Paine, M.L.,
and White, S.N. (2011). A simplified genetic design for mammalian enamel. Biomaterials 32,
3151-3157.
Sonoda, A., Iwamoto, T., Nakamura, T., Fukumoto, E., Yoshizaki, K., Yamada, A., Arakaki, M.,
Harada, H., Nonaka, K., and Nakamura, S. (2009). Critical role of heparin binding domains of
ameloblastin for dental epithelium cell adhesion and ameloblastoma proliferation. Journal of
Biological Chemistry 284, 27176-27184.
Stetefeld, J., McKenna, S.A., and Patel, T.R. (2016). Dynamic light scattering: a practical guide
and applications in biomedical sciences. Biophysical reviews 8, 409-427.
Su, J., Bapat, R.A., and Moradian-Oldak, J. (2019a). The expression and purification of
recombinant mouse ameloblastin in E. coli. In Odontogenesis (Springer), pp. 229-236.
Su, J., Chandrababu, K.B., and Moradian-Oldak, J. (2016). Ameloblastin peptide encoded by
exon 5 interacts with amelogenin N-terminus. Biochemistry and Biophysics Reports 7, 26-32.
Su, J., Kegulian, N.C., Bapat, R.A., and Moradian-Oldak, J. (2019b). Ameloblastin Binds to
Phospholipid Bilayers via a Helix-Forming Motif within the Sequence Encoded by Exon 5. ACS
omega 4, 4405-4416.
Sweeney, R.T., McClary, A.C., Myers, B.R., Biscocho, J., Neahring, L., Kwei, K.A., Qu, K., Gong,
X., Ng, T., and Jones, C.D. (2014). Identification of recurrent SMO and BRAF mutations in
ameloblastomas. Nature genetics 46, 722.
Switzer III, R.C., Merril, C.R., and Shifrin, S. (1979). A highly sensitive silver stain for detecting
proteins and peptides in polyacrylamide gels. Analytical biochemistry 98, 231-237.
Takahashi, S., Kawashima, N., Sakamoto, K., Nakata, A., Kameda, T., Sugiyama, T., Katsube,
K., and Suda, H. (2007). Differentiation of an ameloblast-lineage cell line (ALC) is induced by
Sonic hedgehog signaling. Biochem Biophys Res Commun 353, 405-411.
Toyosawa, S., O’hUigin, C., Figueroa, F., Tichy, H., and Klein, J. (1998). Identification and
characterization of amelogenin genes in monotremes, reptiles, and amphibians. Proceedings of
the National Academy of Sciences 95, 13056-13061.
Travis, D.F., and Glimcher, M.J. (1964). The structure and organization of, and the relationship
between the organic matrix and the inorganic crystals of embryonic bovine enamel. The Journal
of cell biology 23, 447-497.
115

Uchida, T., Fukae, M., Tanabe, T., Yamakoshi, Y., Satoda, T., Murakami, C., Takahashi, O., and
Shimizu, M. (1995). Immunochemical and immunocytochemical study of a 15 kDa non-
amelogenin and related proteins in the porcine immature enamel: proposal of a new group of
enamel proteins ‘sheath proteins’. Biomedical Research 16, 131-140.
Uchida, T., Tanabe, T., Fukae, M., Shimizu, M., Yamada, M., Miake, K., and Kobayashi, S. (1991).
Immunochemical and immunohistochemical studies, using antisera against porcine 25 kDa
amelogenin, 89 kDa enamelin and the 13–17 kDa nonamelogenins, on immature enamel of the
pig and rat. Histochemistry 96, 129-138.
Wald, T., Bednarova, L., Osicka, R., Pachl, P., Sulc, M., Lyngstadaas, S.P., Slaby, I., and
Vondrasek, J. (2011). Biophysical characterization of recombinant human ameloblastin. Eur J
Oral Sci 119 Suppl 1, 261-269.
Wald, T., Osickova, A., Sulc, M., Benada, O., Semeradtova, A., Rezabkova, L., Veverka, V.,
Bednarova, L., Maly, J., and Macek, P. (2013). Intrinsically disordered enamel matrix protein
ameloblastin forms ribbon-like supramolecular structures via an N-terminal segment encoded by
exon 5. Journal of Biological Chemistry 288, 22333-22345.
Wald, T., Spoutil, F., Osickova, A., Prochazkova, M., Benada, O., Kasparek, P., Bumba, L., Klein,
O.D., Sedlacek, R., and Sebo, P. (2017). Intrinsically disordered proteins drive enamel formation
via an evolutionarily conserved self-assembly motif. Proceedings of the National Academy of
Sciences, 201615334.
Wang, H., Tannukit, S., Zhu, D., Snead, M.L., and Paine, M.L. (2005). Enamel matrix protein
interactions. J Bone Miner Res 20, 1032-1040.
Wang, L., Guan, X., Du, C., Moradian-Oldak, J., and Nancollas, G.H. (2007). Amelogenin
promotes the formation of elongated apatite microstructures in a controlled crystallization system.
The Journal of Physical Chemistry C 111, 6398-6404.
Wang, L., Guan, X., Yin, H., Moradian-Oldak, J., and Nancollas, G.H. (2008). Mimicking the self-
organized microstructure of tooth enamel. The Journal of Physical Chemistry C 112, 5892-5899.
Watson, M. (1960). The extracellular nature of enamel in the rat incisor. J Electron Microsc 37,
189-198.
Wazen, R.M., Moffatt, P., Zalzal, S.F., Yamada, Y., and Nanci, A. (2009). A mouse model
expressing a truncated form of ameloblastin exhibits dental and junctional epithelium defects.
Matrix Biology 28, 292-303.
Wen, H.B., Fincham, A.G., and Moradian-Oldak, J. (2001). Progressive accretion of amelogenin
molecules during nanospheres assembly revealed by atomic force microscopy. Matrix Biology 20,
387-395.
Yamakoshi, Y., Tanabe, T., Fukae, M., and Shimizu, M. (1994). Porcine amelogenins. Calcified
tissue international 54, 69-75.
116

Yamakoshi, Y., Tanabe, T., Oida, S., Hu, C.-C., Simmer, J., and Fukae, M. (2001). Calcium
binding of enamel proteins and their derivatives with emphasis on the calcium-binding domain of
porcine sheathlin. Archives of oral biology 46, 1005-1014.
Yang, X., Wang, L., Qin, Y., Sun, Z., Henneman, Z.J., Moradian-Oldak, J., and Nancollas, G.H.
(2010). How amelogenin orchestrates the organization of hierarchical elongated microstructures
of apatite. The Journal of Physical Chemistry B 114, 2293-2300.
Zalzal, S.F., Smith, C.E., and Nanci, A. (2008). Ameloblastin and amelogenin share a common
secretory pathway and are co-secreted during enamel formation. Matrix Biol 27, 352-359.
Zeichner-David, M., Diekwisch, T., Fincham, A., Lau, E., MacDougall, M., Moradian-Oldak, J.,
Simmer, J., Snead, M., and Slavkin, H. (1995). Control of ameloblast differentiation. International
Journal of Developmental Biology 39, 69-92.
Zou, Y., Wang, H., Shapiro, J.L., Okamoto, C.T., Brookes, S.J., Lyngstadaas, S.P., Snead, M.L.,
and Paine, M.L. (2007). Determination of protein regions responsible for interactions of
amelogenin with CD63 and LAMP1. Biochemical Journal 408, 347-354.

Abstract (if available)

Abstract Introduction: Dental enamel formation is guided by a transient extracellular organic matrix which gives rise to an almost completely inorganic enamel containing organized hydroxyapatite (HAP) crystals. The most abundant enamel matrix proteins (EMPs) amelogenin (Amel) and ameloblastin (Ambn) have been studied for over 2 decades to uncover their role in enamel formation. Literature on Amel and Ambn has revealed that, a) Amel and Ambn are co-secreted through the same secretory vesicles, b) Amel and Ambn co-localize in developing molars, c) Amel-Ambn double null mice show a more severe enamel phenotype than the single mutant mice, and d) Amel synthesis is downregulated in Ambn mutant mice with deletion of exons 5 and 6. Based on these data, I hypothesized that Amel and Ambn interact and co-assemble during enamel formation. One of the proposed functions of this interaction is to adhere the enamel extracellular matrix to ameloblast cells. ❧ Materials and methods: Direct binding between Amel-Ambn in vitro was identified using co-immunoprecipitation (co-IP). Co-IP experiments were performed using both recombinant Amel and Ambn as bait. Experiments were repeated with native porcine EMP extract to isolate native Amel-Ambn complexes using co-IP. The region of Ambn directly binding to Amel was identified using 2 Ambn mutants with deletions of exons 5 (AmbnΔ5) and exon 6 (AmbnΔ6), as well as Ambn synthetic peptides encoding exons 5 and 6. A second set of in vitro techniques of dynamic light scattering (DLS) and transmission electron microscopy (TEM) was used to observe the co-assembly of Amel with Ambn at room temperature, near physiological pH. In vivo Amel-Ambn co-localization was demonstrated using post-natal day 8 wild-type mouse incisors sectioned in sagittal and transverse planes. Amel was co-labeled with Ambn fragments containing either the N-terminus or the C-terminus using two different anti-Ambn antibodies and one anti-Amel antibody. Colocalization between Ambn and ameloblast cell membrane was analyzed by co-staining the cell membranes of ameloblasts using the lipophilic dye DiD with immunolabeled Ambn. Ameloblast lineage cells (ALC) and fluorescently tagged Ambn synthetic peptide encoding exon 5 were used to confirm that Ambn binds to the ameloblast cell membrane via its exon 5 encoded fragment. Imaging of immunohistochemically labeled mouse tissues was performed on Leica SP8 confocal microscope and colocalization was quantified using Manders’ colocalization coefficient. ❧ Results: Recombinant Amel and Ambn directly bound to each other and the binding was independently confirmed by using Amel and Ambn as bait proteins. Isolation of native Amel-Ambn complexes from porcine EMP extracts indicated that post-translational modifications neither prevented not aided this binding. AmbnΔ5 failed to interact with Amel and interaction of Amel with Ambn synthetic peptides further confirmed that Ambn bound to Amel via the N-terminus of its exon 5 encoded fragment. Amel-Ambn co-assemblies detected by DLS and TEM resembled Amel nanospheres in shape but were smaller in size, about 12-15 nm in diameter at near physiological pH. In enamel biomineralization, Amel-Ambn binding and co-assembly may be essential for maintaining HAP crystal organization and rod-interrod arrangement. We further suggest that Amel nanosphere like structures observed in vivo could be Amel-Ambn co-assemblies. In the mouse incisor, the C-terminal containing Ambn (mostly full-length Ambn) co-localized with Amel within ameloblasts and at the secretory face of ameloblasts from secretory to transition stages of enamel formation. At maturation stage, the proteins continued to colocalize within ameloblasts, but there was no colocalization observed within the enamel matrix. On the other hand, the N-terminal containing fragments of Ambn colocalized with Amel throughout the thickness of the developing enamel from secretory to maturation stages. Co-localization of Amel-Ambn was also observed within the enamel rod architecture in transverse sections of maturation stage incisor tips. These in vivo images supported our idea that Amel-Ambn co-assembly is essential for maintenance of rod-interrod architecture. Previous work from our lab has shown that Ambn can interact with cell membrane mimicking lipid vesicles. Confirming these in vitro data, our in vivo analysis showed that Ambn colocalized with the ameloblast cell membrane at the apical end from secretory to transition stages. Fluorescently labeled Ambn exon 5 peptide localized on the cell membrane of ameloblast lineage cells confirming the role of exon 5 in Ambn cell interaction. From these data we proposed a second function for Amel-Ambn interaction that Ambn can act as a cell adhesion molecule between ameloblasts and enamel extracellular matrix by binding to ameloblast cell membrane and Amel in the extracellular matrix. ❧ Conclusion and future directions: Our approach of combining in vitro analysis of recombinant and native Amel and Ambn, as well as in situ techniques using wild-type mice allowed us to observe Amel-Ambn interaction in sound enamel. I showed direct Amel-Ambn binding and identified the Amel-binding domain at the N-terminus of exon 5 encoded fragment of Ambn. I proposed a two-fold function of this binding, in biomineralization to allow Amel to bind to HAP crystals more efficiently after the cleavage of its C-terminus and in ameloblast-extracellular matrix adhesion by interaction of Ambn with ameloblast cell membrane. Exon 5 encoded region of Ambn was found to be particularly essential for both Amel and cell membrane interactions. This region contains the Ambn self-assembly domain which may be involved in Amel-Ambn co-assembly. Further investigation is needed to confirm this hypothesis and to detect the presence of Amel-Ambn co-assemblies in vivo. Towards the C-terminal of Ambn self-assembly domain lies the amphipathic helix motif of Ambn which can bind to ameloblast cell membranes as shown in our recent publication. How the intrinsically disordered protein Ambn modulates itself to allow multi-targeting of its exon 5 encoded fragment remains to be investigated. By understanding these functions of Amel-Ambn interaction, we are opening more avenues for cell-free enamel repair. Currently, Amel derived peptides are being used to restore enamel however, the newly developed enamel-like layer lacks the rod-interrod architecture. Greater understanding of EMP interactions can lead to improved techniques of enamel repair in the future.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Ameloblastin-protein interactions pattern enamel matrix

PDF

Chitosan‐amelogenin hydrogel‐saliva interactions: towards optimization of a protocol for enamel repair

PDF

Amelogenin peptides for biomimetic remineralization of enamel and dentin

PDF

Role of MMP-20 in preventing protein occlusion in enamel apatite crystals: relevance in enamel biomineralization and biomimetics

PDF

Enamelysin (MMP20) and Kallikrein 4 (KLK4) functions during enamel formation

PDF

Leucine-rich amelogenin peptide induces osteogenesis in mouse embryonic stem cells

PDF

Transcellular calcium transport in amelogenesis

PDF

Global analysis of the molecular activities defining maturation-stage amelogenesis

PDF

Dynamics of amelogenin self-assembly during in vitro proteolysis by Mmp-20

PDF

Modulation of C/EBP alpha in the regulation of mouse amelogenin transcription during tooth formation

PDF

Integrative analysis of gene expression and phenotype data

PDF

Transcriptional co-activation functions of Msx homeodomain proteins by activating Hsf proteins

PDF

Extracellular heat shock protein-90alpha (eHsp90α): mechanisms of secretion, quantitation and function

PDF

Antibody mediated osseous regeneration

PDF

Protein arginine methyltransferases in murine skull development

PDF

Integrin expression and signaling during palatal fusion

PDF

Identifying and characterizing a unique enhancer for Y537S mutant estrogen receptor (ER) in breast cancer cells

PDF

The structure and function of membrane curving proteins on different membrane shapes and their regulation by post-translational modifications

PDF

IL-7R and c-Kit signaling in thymopoiesis

PDF

The role of neuregulin receptors in cell differentiation and the response to inflammatory cytokines in the intestinal epithelium

Asset Metadata

Creator Bapat, Rucha Arun (author)

Core Title Amelogenin-ameloblastin protein interaction and function in dental enamel formation

School School of Dentistry

Degree Doctor of Philosophy

Degree Program Craniofacial Biology

Publication Date 05/04/2020

Defense Date 03/13/2020

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

Tag ameloblastin,amelogenin,dental enamel,enamel matrix proteins,OAI-PMH Harvest,protein-protein interactions

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Moradian-Oldak, Janet (committee chair), Frenkel, Baruch (committee member), Paine, Michael (committee member), Ulmer, Tobias (committee member), Xu, Jian (committee member)

Creator Email ruchabap@usc.edu,ruchbapat@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-296129

Unique identifier UC11664045

Identifier etd-BapatRucha-8420.pdf (filename),usctheses-c89-296129 (legacy record id)

Legacy Identifier etd-BapatRucha-8420.pdf

Dmrecord 296129

Document Type Dissertation

Rights Bapat, Rucha Arun

Type texts

Source 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 a...

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