A RECOMBINANT SYSTEM TO MODEL PROTEOGLYCAN AGGREGATE
INTERACTIONS AND AGGRECAN DEGRADATION
by
HAZUKI ELEANOR MIWA
Submitted in partial fulfillment of the requirements
For the degree of Doctor of Philosophy
Dissertation Advisors:
Dr. Thomas A. Gerken and Dr. Thomas M. Hering
Department of Biochemistry
CASE WESTERN RESERVE UNIVERSITY
January 2006
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the dissertation of
______
candidate for the Ph.D. degree *.
(signed)______(chair of the committee)
______
______
______
______
______
(date) ______
*We also certify that written approval has been obtained for any proprietary material contained therein.
Copyright © 2005 by Hazuki Eleanor Miwa
All rights reserved
Dedications
I would like to dedicate this thesis to my parents, Johji and Satsuki Bernice Miwa
and my daughter, Miko Nina Miwa. Table of Contents
Title page
Committee Sign-off Sheet
Copyright page
CWRU Waiver
Dedication page
Table of Contents vi
List of Tables xv
List of Figures xvi
Acknowledgements xxi
List of Abbreviations xxiii
Abstract xxix
Chapter 1. Introduction and Background
1.1 General introduction
1.1.1 Cartilage and osteoarthritis 1
1.2 Aggregate interactions
1.2.1 Structure and function of proteoglycan aggregates in the cartilage extracellular
matrix 4
1.2.2 Functions of the link protein family members 6
1.2.3 Divalent cation binding properties of link protein 8
1.2.4 Structural organization of aggrecan 9
- vi - 1.2.5 Structure and biosynthesis of KS and CS on aggrecan 12
1.2.6 Structural characteristics of link protein and the homologous G1 domain of
aggrecan and other lecticans 14
1.2.7 HA binding site of PTR domain 19
1.3 Aggrecan degradation
1.3.1 Aggrecan catabolism by members of ADAMTS family 21
1.3.2 Neo-epitope antibodies 24
1.3.3 Regulation of ADAMTS4 activity 28
1.3.4 Aggrecan structure and substrate specificity of ADAMTS4 31
1.4 Focus of thesis work 36
Part I. Characterization of Recombinant Link Protein and Recombinant Aggrecan
Chapter 2. Expression, Purification, and Refolding of a Pair of Recombinant
Proteoglycan Tandem Repeat Domains of Link Protein from Escherichia coli
Summary 38
2.1 Introduction 39
2.2 Results and Discussion
2.2.1 Cloning of link protein fragments from bovine and human link protein 41
2.2.2 Expression and purification of MBP/full-length and truncated recombinant
bovine and human link protein fusion proteins in E. coli 45
2.2.3 Factor Xa digestion of MBP-bovine PTR1+2 fusion protein 49
2.2.4 Factor Xa digestion of MBP-human link proteins 54
2.2.5 Enterokinase digestion of MBP/E-bPTR1+2 56
- vii - 2.2.6 Refolding of monomeric PTR1+2 domains 58
2.2.7 Zinc (II) binding of various link protein constructs 62
2.2.8 Alternative approaches for expressing recombinant PTR1+2 domains 65
2.3. Conclusions 75
2.4. Experimental Procedures
2.4.1 Materials 76
2.4.2 Construction of E. coli link protein expression vectors 77
2.4.3 Expression and purification of MBP/full-length and truncated recombinant
bovine and human link protein fusion proteins in E. coli 80
2.4.4. Factor Xa digestion of MBP fusion link proteins 82
2.4.5 SDS-PAGE and Western blot analysis 82
2.4.6 Enterokinase digestion of MBP fusion link protein 83
2.4.7 Sephacryl S-300 chromatography and Refolding of MBP/bPTR1+2 84
2.4.8 Zinc binding analysis of MBP/rhLP fusion proteins expressed in E. coli 85
2.4.9 Enzyme linked immunosorbent assay (ELISA) 86
2.4.10 Construction of bovine PTR1+2 into the pBAD/Thio-TOPO vector 86
2.4.11 Pilot expression of thioredoxin fusion PTR1+2 domains in E. coli 87
2.4.12 Construction of bovine PTR1+2 into the pPICZα vector 88
2.4.13 Pilot expression of PTR1+2 domains in Pichia pastoris 89
2.4.14 Large-scale expression and purification of the PTR1+2 domains in Pichia
pastoris 90
2.4.15 N-terminal amino acid sequencing 91
- viii - Acknowledgements 92
Chapter 3. Biochemical and Functional Characterization of Full-length Recombinant
Aggrecan and Cartilage Link Protein Expressed in Mammalian Cells
Summary 93
3.1 Introduction 94
3.2 Results
3.2.1 Construction of recombinant aggrecan expression vectors 95
3.2.2 Mammalian expression of recombinant aggrecan 97
3.2.3 Hydrodynamic sizes of recombinant aggrecan monomer and attached
3.2.4 Expression of recombinant link protein 108
3.2.5 Zinc binding of human recombinant link protein 115
3.2.6 Recombinant proteoglycan ternary aggregate formation 116
3.3 Discussion 121
3.4 Experimental Procedures
3.4.1 Materials 128
3.4.2 Purification of aggrecan and link protein from bovine cartilage and chondrocyte
cultures 129
3.4.3 Construction of full-size aggrecan expression vectors 132
3.4.4 Construction of link protein expression vectors 134
3.4.5 Cell culture 135
3.4.6 Expression and purification of recombinant aggrecan in mammalian cells 136
- ix - 3.4.7 Composite agarose polyacrylamide gel analysis and chemiluminescent Western
blot analysis of aggrecan 137
3.4.8 3.0% and 3.5 % SDS-PAGE gel analysis of full-sized recombinant aggrecan 138
3.4.9 Aggrecan monomer size determination by Sepharose CL-2B chromatography139
3.4.10 GAG size determination by Sepharose CL-6B chromatography 139
3.4.11 Expression and purification of recombinant bovine link protein 140
3.4.12 PNGase F digestion of recombinant link protein 141
3.4.13 Biotin labeled HA binding of link protein 142
3.4.14 Trypsin digestion of proteoglycan aggregates 142
3.4.15 Zinc (II) binding analysis of human recombinant link protein 143
3.4.16 Analysis of proteoglycan aggregate formation 144
Acknowledgements 145
Part II. Aggrecan Degradation
Chapter 4. Characterization of the Substrate Specificity of ADAMTS4 against
Aggrecan Core Protein
Summary 146
4.1 Introduction 149
4.2 Results
4.2.1 Aggrecan structure and aggrecan catabolites 151
4.2.2 Characterization of ADAMTS4-p68 154
4.2.3 Substrate specificity of p68 and p40 156
- x - 4.2.4 KS and CS affect substrate specificity of ADAMTS4 159
4.2.5 N-linked oligosaccharides inhibit the cleavage within the IGD by ADAMTS4
165
4.2.6 Characterization of FLAG-tagged full-length aggrecan 169
4.2.7 Representative digestion of wild-type FLAG-rbAgg expressed in COS-7 cells
with ADAMTS4 173
4.2.8 Anti-NITEGE reactive fragments have intact FLAG epitope and are differentially
glycosylated 175
4.2.9 KS is not required for ADAMTS4 cleavage within the IGD of FLAG-rbAgg 177
4.2.10 Substrate specificity of ADAMTS4 on chondroitin sulfate-free FLAG-aggrecan.
179
4.2.11 Construction of mutagenized full-length bovine aggrecan expression vectors.
185
4.2.12 Expression of full-length FLAG-rbAgg mutant aggrecans and their
susceptibility to ADAMTS4-p40 191
4.2.13 Substrate specificity of ADAMTS4-p68 on mutant aggrecans lacking
potentially glycosylated residues 197
4.2.14 The role of the extended structure N-terminal to the ADAMTS4 cleavage site
within the IGD. 200
4.2.15 Representative digestion of wild-type FLAG-rbAgg expressed in COS-7 cells
with MMP13 205
4.2.16 MMP13 digestion of mutant aggrecans 206
- xi - 4.2.17 Keratan sulfate synthesis in COS-7, CHO-K1, and RCS cells 210
4.2.18 Co-expression of sulfotransferase and FLAG-rbAgg construct 214
4.2.19 Susceptibility of KS-FLAG-rbAgg to ADAMTS4 216
4.3 Discussion 220
4.3.1 Effects of chondroitin sulfate and keratan sulfate on substrate specificity of
ADAMTS4 220
4.3.2 Substrate specificity of ADAMTS4-p68 222
4.3.3 Substrate specificity of ADAMTS4-p40 224
4.3.4 Effects of Keratan sulfate substitution on cleavage within the IGD 225
4.3.5 Effects of non-GAG oligosaccharides on aggrecan cleavage by ADAMTS4 227
4.3.6 S377 is important for substrate recognition by ADAMTS4 228
4.3.7 The potential role of the T352IQTVT357 sequence on ADAMTS4 cleavage of
aggrecan 229
4.3.8 Sulfation of CS by KS sulfotransferases? 232
4.3.9 The susceptibility of KS-FLAG-rbAgg to ADAMTS4 234
4.3.10 MMP13 cleavage sites within the aggrecan core protein 236
4.3.11 Conclusions 237
4.4 Experimental Procedures
4.4.1 Materials 239
4.4.2 Site-directed mutagenesis 240
4.4.3 ADAMTS4 digestion of de-glycosylated cartilage-derived steer aggrecan 241
4.4.4 Removal of N-linked oligosaccharides from cartilage-derived aggrecan 241
- xii - 4.4.5 Secondary structure prediction 242
4.4.6 ADAMTS4 digestion of recombinant aggrecan (FLAG-rbAgg) 242
4.4.7 MMP13 digestion of recombinant aggrecan (FLAG-rbAgg) 243
4.4.8 3.0 % SDS-PAGE of full-size FLAG-rbAgg 244
4.4.9 Western blot analysis 244
4.4.10 Semi-quantification of enzymatically cleaved products 245
4.4.11 Immunocytochemistry 245
4.4.12 Expression of keratan sulfate in COS-7, CHO, and RCS cells 246
4.4.13 Co-expression of FLAG-rbAgg with sulfo- and glycosyl- transferases 247
Acknowledgements 248
Chapter 5. Summary and Future Studies
5.1 General summary 249
5.2 Future studies
-Proteoglycan aggregate interactions-
5.2.1 Refolding PTR1+2 domains from E. coli 252
5.2.2 Functional characterization of the cartilage link protein-HA interaction 253
5.2.3 Effect of glycosylation on HA binding of the G1 domain of aggrecan and link
protein 257
-Aggrecan degradation-
5.2.4 The presence of chondroitin sulfate and keratan sulfate on aggrecan core protein
affects the substrate specificity of ADAMTS4 isoforms 259
5.2.5 Characterization of glycosylation in the IGD of recombinant aggrecan expressed
- xiii - in COS-7 cells 260
5.2.6 The role of Ser 377 in ADAMTS4 recognition 261
5.2.7. The requirement of clusters of hydrophobic residues N-terminal to the
ADAMTS4 cleavage site 262
5.2.8 Keratan sulfate biosynthesis 264
5.2.9 Substrate specificity of sulfotransferases 267
5.2.10 Analysis of glycosaminoglycan-microstructure by FACE 268
5.2.11 Susceptibility of deglycosylated and recombinant mutant aggrecans to
ADAMTS5 269
Bibliography 272
- xiv - Tables
1-I Hyaluronan binding proteins containing a single or double PTR domain(s) 18
1-II Sites within the aggrecan core protein specifically cleaved by members of MMPs and
ADAMTS families 25
2-I Primers for construction of various human link protein constructs 42
2-II Primer sets used for amplification of full-length and truncated human link protein
sequences 43
2-III Apparent molecular mass of MBP fusion proteins on SDS-PAGE gels 48
2-IV N-terminal amino acid sequence of Ni2+ purified 8-A-4 anti-LP weakly reactive
protein from Pichia pastrois 74
4-I Antibodies used in this study for identifying aggrecan fragments generated by
digestion by ADAMTS4 or MMP13 153
4-II Various cell lines tested in Chapter 3 for a transient expression of FLAG-rbAgg 172
4-III Primer sequences and template plasmids used for site-direct mutagenesis 188
4-IV Probability of having an extended secondary structure in the sequence
(E349DITIQTVTQPD360) 192
- xv - Figures
1-1 Toluidine blue-stained sections knee joint cartilage. 3
1-2 Schematic drawing of a proteoglycan aggregate comprised of hyaluronan, aggrecan monomers, and link proteins. 5
1-3 Schematic diagram of cartilage-link protein. 7
1-4 Sequence alignment of bovine cartilage link protein and the G1 domain of bovine aggrecan. 11
1-5 Keratan sulfate and chondroitin sulfate polysaccharide structures and glycosylation linkages. 13
1-6 Sequence alignment of human PTR domains from the members of the human link protein and lectican families. 16
1-7 HA-binding site on a single PTR domain of human TSG-6. 20
1-8 Structure of aggrecan monomer and the major proteolytic cleavage sites within the bovine aggrecan core protein by MMP13 and ADAMTS4 and 5. 26
1-9 Sequence alignment of IGD and nodal regions in the CS-2 domains neighboring the aggrecanase cleavage sites in bovine aggrecan. 27
1-10 ADAMTS4 isoforms and GAG binding motifs. 30
1-11 ADAMTS4 activation pathway and substrate specificity of each isoform. 32
1-12 Distribution of oligosaccharide chains in the hyaluronan-binding region (HABR) of calf and steer aggrecan. 35
2-1 Schematic representation of recombinant MBP/link protein constructs expressed in E. coli 44
- xvi - 2-2 Expression and purification of MBP/bPTR1+2. 47
2-3 Factor Xa digestion of MBP/bPTR1+2. 51
2-4 Factor Xa digested PTR1+2 analyzed with N-terminal amino acid sequencing. 52
2-5 Factor Xa-digested MBP/bPTR1+2 analyzed on SDS-PAGE/Western blot. 53
2-6 Factor Xa-digested MBP/hLP constructs analyzed on SDS-PAGE/Western blot. 55
2-7 Enterokinase-digested MBP/E-bPTR1+2 analyzed on SDS-PAGE/Western blot. 57
2-8 Sephacryl S-300 size exclusion chromatography of factor Xa-undigested and digested
MBP/PTR1+2 in 6 M GnHCl. 59
2-9 Western blot analysis of dialyzed fractions from the Sephacryl S-300 gel filtration chromatography of factor Xa-digested MBP/bPTR1+2. 61
2-10 Zinc binding of MBP/bPTR1+2 by zinc affinity chromatography. 63
2-11 Zinc binding of full-length and truncated human MBP fusion link protein 64
2-12 Expression of thioredoxin PTR1+2 fusion protein in E. coli. 66
2-13 Expression of PTR1+2 domains in KM71 (Pichia pastoris). 70
2-14 Expression of PTR1+2 domains in GS115 (Pichia pastoris). 71
2-15 Purification of bPTR1+2 from Pichia pastoris (cell lysates) on Ni (II) chelate chromatography. 72
2-16 Ni2+ purified protein analyzed with N-terminal amino acid sequencing. 73
3-1 Alignment of bovine aggrecan cDNA clones and PCR products used to construct full-sized cDNA expression vector insert. 96
3-2 Composite agarose-acrylamide gel/ Western blot analysis. 98
3-3 Secretion of full-sized FLAG-rbAgg expressed in COS-7 cells. 100
- xvii - 3-4 Chondroitin sulfate substitution of FLAG-rbAgg expressed in COS-7 cells. 102
3-5 Recombinant aggrecan expression in wild-type CHO-K1 and in CS-deficient
CHO-745 cells. 103
3-6 Chondroitinase ABC susceptibility of FLAG-rbAgg expressed in various cell lines.
105
3-7 Aggrecan monomer size determination by Sepharose CL-2B size exclusion
chromatography. 107
3-8 Analysis of chondroitin sulfate substitution and chain length. 109
3-9 Expression of bovine and human link protein in COS-7 cells. 111
3-10 Biochemical characterization of recombinant link protein. 113
3-11 Zinc binding of human link protein expressed in COS-7 cells. 117
3-12 Proteoglycan ternary aggregation analyses by associative Sepharose CL-2B
chromatography. 118
3-13 Cartilage-derived aggrecan and link protein interact with hyaluronan to form
proteoglycan aggregates. 120
4-1 Sites recognized by specific neo-epitope antibodies. 152
4-2 ADAMTS4-p68 used in this work. 155
4-3 Schematic representation of aggrecan fragments that can be generated by ADAMTS4
digestion of aggrecan core protein. 157
4-4 Substrate specificity of ADAMTS4-p68 and p40. 160
4-5 Analysis of de-glycosylated steer aggrecan. 162
4-6 Aggrecanase digestion of native (undigested) and de-glycosylated cartilage-derived
- xviii - steer aggrecan. 164
4-7 Semi-quantitation of the effects of KS and CS removal on cleavage within the IGD.
166
4-8 ADAMTS-4 digestion of native and de-N-linked aggrecan. 168
4-9 Schematic representation of FLAG-rbAgg. 171
4-10 ADAMTS4 digestion of FLAG-aggrecan. 174
4-11 A lack of ADAMTS4 cleavage site N-terminal to the E373-A374 cleavage site. 176
4-12 Keratanase susceptibility of anti-NITEGE reactive fragments. 178
4-13 CS and KS stubs remain on aggrecan core protein after chondroitinase ABC and keratanases digestions. 180
4-14 Schematic representation of aggrecan varying in CS structure within the CS domains. 181
4-15 Substrate specificity of ADAMTS4-p68 and p40 on CS modified FLAG-rbAggs.
184
4-16 Summary of FLAG tagged recombinant aggrecan mutants. 187
4-17 Alignment of the aggrecan IGD sequence from different species. 189
4-18 Chemical structures of threonine, serine, asparagine, glutamine, and valine. 190
4-19 Full-sized wild type and mutant FLAG-rbAgg expressed in COS-7 cells. 194
4-20 ADAMTS4-p40 digestion of FLAG-rbAgg mutants. 196
4-21 ADAMTS4-p68 digestion of group (A) mutants. 199
4-22 Representative ADAMTS4-p68 digestion of group (B) mutants. 202
4-23 Representative ADAMTS4-p68 digestion of group (C) mutants. 203
- xix - 4-24 Representative ADAMTS4-p68 digestion of triple valine mutant
(T352V-T35V-T357V) 204
4-25 MMP13 digestion of FLAG-rbAgg. 207
4-26 MMP13 digestion of wild-type and mutant FLAG-rbAggs. 209
4-27 Immunocytochemical analysis of KS production in cell lines transiently
overexpressing sulfotransferases. 212
4-28 N- and O-linked KS production in COS-7 and CHO cells overexpressing
6-O-sulfotrtansferases and Core 2 GlcNAc transferase. 213
4-29 3.0 % SDS-PAGE analysis of FLAG-rbAgg expressed in COS-7 cells along with
sulfotransferases. 215
4-30 Substrate specificity of ADAMTS4-p68 and p40 on KS-FLAG-rbAgg. 218
4-31 Schematic model showing the glycosaminoglycan-dependent substrate specificity of
ADAMTS4-p68 and ADAMTS4-p40. 226
4-32 Summary of site-directed mutagenesis studies. 231
5-1 Potential residues in bovine cartilage link protein that may be involved in
HA-binding. 256
5-2 Alignment of active forms of ADAMTS4 and ADAMTS5. 271
- xx - Acknowledgements
Firstly, I would like to thank Dr. Thomas M. Hering and Dr. Thomas A. Gerken for
advising me with my Ph.D. thesis project and for their continuous support for the past
seven years. Thanks to their wise mentoring, I have gained great knowledge, and I hope
some wisdom, in the field of biochemistry, especially in cartilage biochemistry and
I would also like to thank Dr. William Merrick and Dr. Edward Stavnezer for serving
on my committee from the beginning and for their helpful advise. I would also like to
thank Dr. Brian Johnstone for joining my thesis committee and continuing to serve on my
committee even after moving to Oregon Health and Science University in Oregon. I am also grateful to him for his continuous support and helpful advise on my thesis project. I also thank my former committee member Dr. Ki-Joon Shon for his support in serving on my committee. I also like to thank Dr. Jung Yoo for his generous support.
I would like to thank all the people in the Departments of Orthopaedics, Pediatrics,
and Biochemistry for their support over these years, especially those in Dr. Hering’s and
Dr. Gerken’s laboratories.
I would like to thank my co-worker and my friend Mrs. Lori Duesler for helping me
with all aspects of my life for the last six years. I was able to attend many scientific
meetings because of your help in taking care of my daughter during my absence. I would
also like to thank you for lending me your shoulder when I needed one. I would have
never reached this far without your help. Thank you so much.
I would like to thank my past co-worker Mr. John Kollar, Mr. Tru D. Huynh and Dr.
- xxi - Leila Jabbour for giving me technical advise, especially when I was just starting my work
in the laboratory.
I would like to thank my friends Ms. Akiko Yasukawa, Ms. Momo Kameyama, Mr.
Hirotaka Shimizu, and Mr. Yohei Kawashima for helping me live through the early days
in graduate school
I would like to thank my mother Mrs. Satsuki Bernice Miwa for providing me unconditional support throughout my life. I also thank my mother for her critical reading of my entire thesis to “polish” my grammar. I would like to thank my father Dr. Johji
Miwa, my life time mentor, who also introduced me into this wonderful field of science so that I could fulfill my never- ending curiosity to solve the mysteries of life
I would like to thank my grandparents Dr and Mrs. Harold and Eleanor Halcrow and
Mr. and Mrs. Johichi and Suzue Miwa for giving me guidance and love and I grew up.
I would like to thank my aunt Beth and uncle Jeff Johnson for always caring for my
daughter, and me and for offering me with happy home that I can visit for Holidays.
I like to thank my brother Tetsuji Miwa and my sister Tamaki Miwa for being my
bother and sister. I am so lucky to have you two as my siblings.
Lastly, I like to thank my beautiful darling daughter Miko Nina Miwa. I appreciate
you so much for being as cooperative as a three-year old could be, when mommy had to
write this thesis, and for cheering me up with your smile everyday. Even your little
disturbances actually helped ease my stress (maybe) on many a day. Although I may not
be able to spend as much time with you as your friends’ mothers do with them, I am
always with you, and I always love you. Miko, you are my angel.
- xxii - Abbreviations
A1A1D1 Bovine cartilage aggrecan purified by sequential associative, associative, dissociative cesium chloride gradient centrifugation
ADAMTS A disintegrin and metalloproteinase with thrombospondin motifs
ADAMTS4 Aggrecanase-1
ADAMTS4-p40 p40 form of ADAMTS4
ADAMTS4-p68 p68 isoform of ADAMTS4
ADAMTS5 Aggrecanase-2
Agg Aggrecan
Agg1 Higher molecular weight full-length aggrecan
Agg2 Lower molecular weight full-length aggrecan
Amp Ampicillin
AP Alkaline phosphatase
BAC Bovine articular chondrocyte
BCIP 5-bromo, 4-chloro, 3-indolyphosphate
bHA biotinylated hyaluronan
BSA Bovine serum albumin
C2GnT Core 2 N-acetylglucosaminyltransferase
CAPS 3-(cyclohexylamino)-1-propanesulfonic acid
CAPAGE Composite agarose polyacrylamide gel electrophoresis
cDNA complimentary DNA
CGn6ST Corneal N-acetylglucosamine 6-O-sulfotransferase
- xxiii - CHAPS 3-[(3-Cholamidopropyl)dimethylammonio]propanesulfonic acid
Ch'aseABC Chondroitinase ABC
CHO-745 Chinese hamster ovary pgsA 745 xylosyltransferase mutant cell line
COS-7 SV-40 transformed kidney African green monkey
CS Chondroitin sulfate
CS-1 domain Chondroitin sulfate-rich domain 1
CS-2 domain Chondroitin sulfate-rich domain 2
DAPI 6'-diamidino-2-phenylindole
DEAE Diethylaminoethyldiam, diameter
DMEM Dulbecco modified Eagle medium
DPBS Dulbecco phosphate buffer saline
DTT Dithiothreitol
EDTA Ethylenediaminetetraacetic acid
ECL Enhanced chemiluminescence
EGF epidermal growth factor
ELISA Enzyme-linked immunosorbent assay
EM Electron microscopy
EST Expressed sequence tag
FBS Fetal bovine serum
FITC Fluorescein isothiocyanate
- xxiv - FLAG DYKKDDDDK peptide sequence
FLAG-rbAgg Flag tagged recombinant bovine aggrecan
Fuc Fucose
G1 domain Far N-terminal globular domain of aggrecan
G2 domain A globular domain located between the IGD and KS domain of aggrecan
G3 domain Far C-terminal globular domain of aggrecan
GAG Glycosaminoglycan
Gal Galactose
GalNAc N-acetyl galactosamine
GalNAc-S N-acetyl galactosamine 4 or 6 sulfate
GlcNAc N-acetyl glucosamine
GlcA Glucuronic acid
GnHCl Guanidine hydrochloride
HA Hyaluronan
HABR Hyaluronan binding region
HP His-patch
HPLC High performance liquid chromatography
HRP Horse radish peroxidase
IGD Interglobular domain
Ig-fold Immunoglobulin-like domain
IL-1 Interleukin-1
- xxv - IPTG Isopropyl-b-D-thiogalactopyranoside kD Kilodalton(s)
KS Keratan sulfate
KS domain KS-rich domain
KS-FLAG-rbAgg Sulfotransferases co-transfected FLAG rbAgg
KSG6ST Keratan sulfate galactose-6-O-sulfotransferase
LP Link protein
LP1 Link protein with two N-glycans
LP2 Link protein with one N-glycans
LP3 Link protein isolated from trypsin digested HABR complex
Man Mannose
MBP Maltose binding protein
MBP/bPTR1+2 MBP fusion recombinant bovine PTR1+2 domains with factor Xa site
MBP/E-bPTR1+2
MBP fusion recombinant bovine PTR1+2 domains with enterokinase site
MBP/hLP MBP fusion recombinant human full-length link protein
MBP/hIg-fold MBP human recombinant Ig-fold domain
MBP/hPTR1+2 MBP human recombinant PTR1 and 2 domains
MBP/hPTR1 MBP human recombinant PTR1 domain
MBP/hPTR2 MBP human recombinant PTR2 domain
MEM Eagle's minimum essential medium
MES 2-(N-morpholino)ethane sulfonic acid
- xxvi - MMP Matrix metalloproteinase
MT-MMP Membrane-type matrix metalloproteinase
NBT Nitroblue tetrazolium
NMR Nuclear magnetic resonance
OA Osteoarthritis
O/N Over night
PAGE Polyacrylamide gel electrophoresis
PBS Phosphate-buffered saline
PCR Polymerase chain reaction
PDI Protein disulfide isomerase
PMSF Phenyl methyl sulfonyl fluoride ppGalNAcT UDP-GalNAc polypeptide:N-acetylgalactosaminyltransferases
PTR Proteoglycan tandem repeat domain
PTR1 First proteoglycan tandem repeat
PTR2 Second proteoglycan tandem repeat
PTR1+2 PTR1 and PTR2 domains
PVDF Polyvinylidene fluoride rbAgg Recombinant bovine aggrecan
RIPA Radio immuno precipitation assay
RCS Rat chondrosarcoma
SDS Sodium dodecyl sulfate sGAG Sulfated glycosaminoglycan
- xxvii - Sia Sialic acid
T352,5,7Q T352Q, T355Q, T357Q triple mutant
T352,5,7V T352V, T355V, T357V triple mutant
T352,355,357Q T352Q, T355Q, T357Q triple mutant
T352,355,357V T352V, T355V, T357V triple mutant
TBS Tris-buffered saline
TGF-β Transforming growth factor-β
T/C-28a2 Immortalized human juvenile costal chondrocyte
TKMS buffer Tris, potassium, magnesium, sucrose buffer
Xyl Xylose
- xxviii - A Recombinant System to Model Proteoglycan Aggregate
Interactions and Aggrecan Degradation
Abstract
By
Hazuki Eleanor Miwa
The proteoglycan aggregate is a major component of the articular cartilage extracellular matrix comprising hyaluronan (HA), aggrecan, and link protein. Aggrecan, heavily substituted with chondroitin sulfate (CS) and keratan sulfate (KS) glycosaminoglycans, contributes to cartilage hydration by binding water, thus lending articular cartilage its resistance to compressive deformation. A loss of aggrecan is observed in the early stages of osteoarthritis, which may relate to age-dependent changes in biochemical properties of proteoglycan aggregates. The research reported in this thesis was focused on producing recombinant molecules for modeling proteoglycan aggregate assembly and aggrecan proteolytic degradation. First, the proteoglycan tandem repeat (PTR) domains of link protein, which are responsible for HA binding, were expressed in E. coli to develop a tool for studying the PTR-HA interaction. Soluble monomeric PTR domains were obtained by a novel refolding procedure. Second, full-length recombinant link protein and aggrecan were expressed in a mammalian expression system as functional molecules capable of forming a ternary complex with HA. Furthermore, the expression of recombinant aggrecan in various mammalian cell lines allows the production of differently
- xxix - glycosylated aggrecans for comparison of glycosylation-specific functional differences.
Last, both cartilage-derived and recombinant wild-type and mutagenized aggrecans were
used to study proteinase-dependent aggrecan degradation observed in the cartilage extracellular matrix. The present work demonstrated that sulfated glycosaminoglycans
covalently attached to both cartilage-derived and recombinant aggrecans regulate the
susceptibility of aggrecan to ADAMTS4. A study using mutagenized recombinant
aggrecan suggested that potentially glycosylated threonine and serine residues N-terminal
and C-terminal to the ADAMTS4 cleavage site within the interglobular domain of
aggrecan influence the rate of cleavage by ADAMTS4. The degree of hydrophobicity
N-terminal to the ADAMTS4 cleavage site also affects aggrecan’s susceptibility to
ADAMTS4. Recombinant aggrecan was substituted with KS by co-transfecting aggrecan with KS-sulfotransferases. KS-substituted aggrecan, which also had altered CS components, was found to be differentially susceptible to different isoforms of
ADAMTS4. This work demonstrates the usefulness of the recombinant system for studies of molecular interactions with HA, aggrecan, and link protein, as well as for in vitro
analyses of matrix proteinase-mediated degradation of aggrecan.
- xxx - Chapter 1 Introduction and Background 1.1 General introduction
1.1.1 Cartilage and osteoarthritis
The extracellular matrix (ECM) of articular hyaline cartilage, the major load bearing
tissue found in joints, is mainly composed of proteoglycan aggregates and type II
collagen fibrils. The proteoglycan aggregate ternary complex comprises hyaluronan (HA;
a polysaccharide glycosaminoglycan (GAG)), aggrecan (Agg; a proteoglycan), and link
protein (LP; a glycoprotein). In articular cartilage, the meshwork of type II collagen
fibrils provides tensile strength, while the embedded proteoglycan aggregates, which are
rich in negatively charged glycosaminoglycans (GAGs), contribute to cartilage hydration by their water binding properties. Together these molecules lend cartilage its resistance to compressive deformation. It has been suggested that the loss of proteoglycan aggregates and aggrecan monomers from the cartilage ECM is associated with osteoarthritis (OA).
It has been estimated that by the age of 40, about 70 % of the U.S. population, and
by the age of 65, over 80 % of the U.S. population have developed OA in one or more joints (Swagerty and Hellinger, 2001). Health related expenses and work loss due to this disorder are enormous, yet no cure for OA is available to date. Understanding of the biochemical and physiological abnormalities found in OA cartilage compared with normal cartilage is an important starting point for the development of therapeutics against arthritic diseases.
The anatomical hallmark of OA is destroyed articular cartilage, which covers the
surface of bones in joints. Histological analysis of normal articular cartilage shows a
- 1 - smooth surface with abundant glycosaminoglycans stained with Toluidine blue (Fig. 1-1
a). In contrast, damaged OA cartilage shows a prominent loss of proteoglycan and HA
staining from the cartilage surface, which is associated with cartilage fibrillation (Fig. 1-1
b). This analysis clearly shows that a loss of proteoglycan is one of the major changes that can be observed in OA cartilage. Although it was long believed that the main cause of OA is mechanical wear and tear of articular cartilage, several reports now suggest that the components of proteoglycan aggregates undergo many biochemical modifications in an age-dependent manner, which might make older cartilage more susceptible to developing OA. For example, both link protein and aggrecan syntheses are down regulated (Bolton et al., 1996; Bolton et al., 1999) and the rate of proteoglycan aggregate assembly is decreased (Bayliss et al., 2000) with age. In addition, an increased susceptibility of aggrecan isolated from older individuals to matrix degrading enzymes has also been observed (Roughley et al., 2003). Age-dependent structural changes observed in aggrecan’s glycosylation are thought to be one of the major factors affecting the susceptibility of aggrecan to proteolytic degradation (Pratta et al., 2000). In this thesis, we describe the development of recombinant molecules that can serve as in vitro models for studying the mechanisms of proteoglycan aggregate assembly and proteolytic degradation of aggrecan.
- 2 -
Fig. 1-1 Toluidine blue-stained sections knee joint cartilage. (a) Normal and (b) damaged human OA knee joint articular cartilage. Normal cartilage has a smooth surface and dark Toluidine blue staining indicating the presence of GAGs, whereas damaged OA cartilage has a rough fibrillated surface and is weakly stained with toluidine blue (Bar; 400 μm). (Histological preparation by Ms. Teresa Pizzuto).
- 3 - 1.2 Aggregate interactions
1.2.1 Structure and function of proteoglycan aggregates in the cartilage extracellular matrix
Aggrecan in the cartilage ECM interacts in a noncovalent ternary complex with HA and cartilage link protein (Fig. 1-2 a). Electron microscopic studies have shown that more than one hundred aggrecan monomers can bind to one HA polysaccharide chain
(Buckwalter and Rosenberg, 1982), which serves as the backbone structure of proteoglycan aggregates (Fig. 1-2 b). The main function of link protein is to stabilize the interaction between HA and aggrecan by binding to both molecules (Choi et al., 1985)
(Fig. 1-2 a and b). Additionally, in the presence of link protein, proteoglycan aggregates are found to have three times more aggrecan monomers attached to HA suggesting the importance of link protein in the assembly of proteoglycan aggregates (Buckwalter et al.,
1984). In the following section of this chapter, the structures and functions of both link protein and aggrecan are discussed.
Although the assembly of proteoglycan aggregates appears to take place in the ECM, newly secreted aggrecan does not initially have maximum HA binding capacity (Oegema,
1980). Therefore, after being secreted into the ECM, aggrecan apparently must acquire the structural conformation(s) optimal for its interaction with HA in the ECM. Later studies showed that disulfide bonds in the G1 domain require time to mature, and this process may be affected by the presence of soluble ligands such as HA, metal ions, and other environmental factors (e.g., redox state, local pH (Sah et al., 1990), and ionic strength).
- 4 -
Fig. 1-2 Schematic drawing of a proteoglycan aggregate comprised of hyaluronan, aggrecan monomers, and link proteins. (a) Multiple aggrecan monomers bind to the hyaluronan polysaccharide chain. This interaction is stabilized by link protein. (b) Aggrecan binds to hyaluronan (HA) and link protein via the G1 domain forming a ternary complex. Highly charged KS and CS polysaccharides are located between the G2 and G3 domains of aggrecan.
- 5 - As part of this study, the use of recombinant link protein and recombinant aggrecan is
discussed as models for understanding their interactions with each other and with HA.
This model system may help clarify the mechanisms of proteoglycan aggregate assembly
that takes place in the ECM.
1.2.2 Functions of the link protein family members
Four homologous members of the link protein family have been identified showing
abundant expression in cartilage (cartilage link protein (HAPLN-1)), brain (brain link
protein-1 (HAPLN-2) and brain link protein-2 (HAPLN-4)), and the cardiovascular system (HAPLN-3). Link protein consists of three globular domains including the
N-terminal immunoglobulin-like (Ig-fold) domain and the paired proteoglycan tandem repeat (PTR) domains. In the case of cartilage link protein, the Ig-fold domain binds to aggrecan and the PTR domains bind to HA (Neame and Barry, 1993) (Fig. 1-3). The PTR domains are also called link modules because of their HA binding ability. Cartilage link protein was the first member of this family to be identified; therefore, when the term
“link protein” is used, it commonly refers to cartilage link protein.
Recently, it has been demonstrated that cartilage link protein not only binds to
aggrecan but also binds to other members of the lectican (large HA-binding aggregating proteoglycan) family including versican (Shi et al., 2004) and neurocan (Rauch et al.,
2004) in vitro. Members of the link protein and the lectican families co-localize in many tissues indicating their physiological interactions in vivo.
- 6 -
Fig. 1-3 Schematic diagram of cartilage-link protein. The Ig-fold domain is mainly involved in aggrecan binding and the PTR 1 and 2 domains are involved in HA binding.
- 7 - For example, brain link proteins 1 and 2 co-localize with versican and brevican,
respectively (Bekku et al., 2003; Oohashi et al., 2002). In addition, a new member of
the link protein family (HAPLN-3) was shown to co-localize with versican in arterial
smooth muscle cells (Ogawa et al., 2004). Interestingly, in the mammalian genome, each
link protein gene is physically located next to a gene of a lectican family member (Spicer
et al., 2003). Although no binding studies have been performed, the high sequence
homology of cartilage link protein to the other three link proteins suggests that these link
proteins may also bind to both HA and other lecticans.
In addition to its role in holding the components of the proteoglycan aggregate
together, cartilage link protein has also been shown to play a role in regulating the
expression of cartilage ECM components. Liu and co-authors have shown that the
N-terminal peptide of link protein, generated by proteolytic cleavage with matrix
metalloproteinases (MMPs) between His16 and Ile17, functions as a growth factor to
up-regulate the expression of aggrecan and type II collagen (Liu et al., 1997; Liu et al.,
2000; McKenna et al., 1998).
1.2.3 Divalent cation binding properties of link protein
Cartilage link protein has been shown to bind to zinc and other divalent cations
(Rosenberg et al., 1991). Link protein exists in a monomer-hexamer equilibrium in
physiologic solvents, and dissociates to form dimers when binding HA polysaccharides
(Rosenberg et al., 1991). In the presence of zinc, the hexameric link protein complex is found to be insoluble. Link protein can bind to zinc and HA simultaneously suggesting
- 8 - that link protein interacts with these molecules at distinct, but as yet unidentified binding sites (Rosenberg et al., 1991). The physiological significance of divalent cation binding is still unknown. Work from Dr. Hering’s laboratory suggests that zinc is required for the proper folding of recombinant link protein expressed in E. coli (Varelas et al., 1997). On the other hand, Rosenberg and co-workers showed that zinc significantly reduces the solubility of link protein in the absence of HA (Rosenberg et al., 1991). Dimeric link protein bound to HA polysaccharides, however, is completely soluble in the presence of zinc at concentrations that would precipitate unbound hexameric LP complexes
(Rosenberg et al., 1991).
1.2.4 Structural organization of aggrecan
Aggrecan, one of the major proteoglycans residing in the cartilage ECM, contributes significantly to cartilage hydration. Aggrecan is a member of lectican family that binds to
HA via its N-terminal domain and also to other oligosaccharides via its C-terminal domain (Yamaguchi, 2000). Aggrecan consists of a core protein of about 2400 amino acids (varying in length among species), which is heavily glycosylated with sulfated glycosaminoglycans (sGAGs). As shown in Fig. 1-2 b, aggrecan is composed of three globular domains that are linked together by intervening extended domains. The G1 domain at the N-terminus of the aggrecan core is responsible for aggrecan’s non-covalent binding to HA, which is further stabilized by the binding of cartilage link protein
(Caterson and Baker, 1978; Heinegard and Hascall, 1974).
- 9 - The G1 regions of aggrecan and link protein show structural and sequence similarities
(Fig. 1-4) and comprise three functional domains termed the Ig-fold, PTR1, and PTR2
domains, which are also referred to A, B, and B’, domains respectively (Doege et al.,
1991). The G1 and G2 domains are separated by the interglobular domain (IGD), which is sensitive to proteolytic cleavage by members of the ADAMTS (a disintegrin and metalloproteinases with thrombospondin motifs) and MMPs families (Fosang et al.,
1996; Lark et al., 1995; Tortorella et al., 1999) (see Figs. 1-2 b and 1-7). Although the
G2 domain shows high sequence homology to the HA-binding region (PTR1 and 2 domains) of the G1 domain, the G2 domain apparently lacks HA-binding activity
(Fosang and Hardingham, 1989; Watanabe et al., 1997). The G3 domain, which has epidermal growth factor (EGF) repeats, C-type lectin, and complement regulatory protein-like motifs, is located at the far C-terminal of the aggrecan core protein.
Chondroitin sulfate (CS) and keratan sulfate (KS) chains are attached to the core protein mainly in the extended regions, termed the KS, CS-1, and CS-2 domains, located between the G2 and G3 domains (Fig. 1-2 b). KS is also found in the G1 domain and in the IGD (Barry et al., 1995). Aggrecan isolated from cartilage ECM shows considerable heterogeneity owing to the presence of varying amounts of KS and CS in the individual aggrecan molecules. Furthermore, proteolytically degraded aggrecan fragments, which are typically cleaved at the C-terminus and lacking the G3 domain (Dudhia et al., 1996), also contribute to its heterogeneity in cartilage.
- 10 -
Fig. 1-4 Sequence alignment of bovine cartilage link protein and the G1 domain of bovine aggrecan. Sequence in red indicates the signal peptide, in green the Ig-fold, and in blue the PTR1+2 domains. Letters with asterisks are conserved between link protein and aggrecan. Initiation Met is designated as residue number 1 in this figure. Note, however, that for bovine aggrecan residue numbers used in the text, the first amino acid (Val) of secreted mature aggrecan is designated as residue number 1, unless stated otherwise.
- 11 - 1.2.5 Structure and biosynthesis of KS and CS on aggrecan
Sulfated glycosaminoglycans (i.e., CS and KS) covalently attached to the aggrecan
core protein not only play a role in cartilage hydration, but also are suggested to take part
in regulating its susceptibility to proteolytic degradation by members of the ADAMTS
family, which will be discussed later. The GAG structures in cartilage aggrecan seem to
change in an age-dependent manner on aggrecan found in cartilage (Barry et al., 1995).
Many enzymes (described below) are involved in GAG biosynthesis, and the abundance and type of glycosyltransferases and sulfotransferases expressed may change in an age-related manner that could contribute to the age-specific glycosylation.
Keratan sulfate is an oligomer composed of sulfated N-acetlyllactosamine
disaccharide repeats (KS repeats). The degree of KS sulfation, fucosylation, and
sialylation and the number of disaccharide repeats vary between tissues, proteoglycans,
and even the sites of substitution within a protein. KS chains can be attached to Asn
(N-link) or Ser/Thr (O-link) structures on the proteoglycan core protein (Fig. 1-5 a and b)
(Funderburgh, 2000). Both N-linked and O-linked KS are found on aggrecan expressed in cartilage. N-linked KS is attached to Asn in the core protein via a complex-type N-linked branched oligosaccharide (Fig. 1-5 a), whereas O-linked KS is attached to Thr/Ser in the core protein via core 2 branched structures (Funderburgh, 2000). Therefore, it has been suggested that the core 2 β 1,6-N-acetylglucosaminyltransferase, which catalyzes the formation of core 2 branching at Gal C6, is critical for the biosynthesis of O-linked
KS in aggrecan (Funderburgh, 2000) (Fig. 1-5 b).
- 12 -
Fig. 1-5 Keratan sulfate and chondroitin sulfate polysaccharide structures and glycosylation linkages. (a) N-linked KS, (b) O-linked KS, and (c) O-linked CS. Monosaccharide abbreviations, GlcA, glucuronic acid; GalNAc-S, N-acetylgalactosamine 4 (or 6) sulfate; Gal, galactose; Xyl, xylose; GlcNAc, N-acetylglucosamine; Man, mannose; Sia, sialic acid; Fuc, fucose; GalNac, N-acetylgalactosamine.
- 13 - Chondroitin sulfate is attached via Ser residues in the Ser-Gly sequence (Fig. 1-5 c).
CS biosynthesis is initiated by the addition of xylose at Ser, which is catalyzed by
xylosyltransferase (Silbert and Sugumaran, 2002). Subsequently, the CS linkage region is
formed by the action of GalI, GalII, and GluA transferases (Silbert and Sugumaran, 2002).
Finally, CS chains are elongated by CS synthases that catalyze the alternative addition of
GalNAc and GluA to the elongating CS chain (Silbert and Sugumaran, 2002). As the
chain elongates, GalNAc can be sulfated either at the 4-O- or 6-O- position or at both, by
the action of chondroitin sulfate sulfotransferases (Kitagawa et al., 2001; Sato et al.,
2003; Silbert and Sugumaran, 2002; Yada et al., 2003; Yada et al., 2003).
1.2.6 Structural characteristics of link protein and the homologous G1 domain of
aggrecan and other lecticans
The N-terminal G1 domains of members of the lectican family (i.e., aggrecan,
versican, neurocan, and brevican) show high homology to link protein especially in their
PTR 1 and 2 domains (Fig. 1-6). While all members of the link protein and lectican
families carry a pair of PTR domains, the HA-binding proteins TSG-6 (tumor necrosis
factor stimulated gene-6) and CD44 have only one PTR domain (Table 1-I). This observation has lead to the idea that one PTR domain may be sufficient for achieving efficient HA binding. Watanabe and co-authors (Watanabe et al., 1997) demonstrated, however, that aggrecan needs both PTR domains for efficient binding to HA. This result suggests that link protein may also require a pair of PTR domains for optimal HA binding, although Dr. Hering’s laboratory and others had reported that one PTR domain is
- 14 - adequate in vitro (Grover and Roughley, 1994; Varelas et al., 1995). Nevertheless, further studies are required to resolve these conflicting results.
As described earlier, it has been suggested that aggrecan undergoes a conformational maturation in the ECM before it acquires high binding affinity for HA (Oegema, 1980).
Link protein, however, may help aggrecan bind to HA in a low affinity conformation
(Bayliss et al., 2000). It is of interest to know whether link protein also undergoes such conformational maturation in the ECM. In addition, glycosylation of the HA-binding region of aggrecan (G1 domain) and of CD44 has also been suggested to play a role in regulating their binding to HA (Bartolazzi et al., 1996; Watanabe et al., 1997).
Interestingly, T42, the O-glycosylation site in the G1 domain that may affect HA binding of aggrecan (Watanabe et al., 1997), is variably substituted in an age-dependent manner; it is substituted with KS only in steer aggrecan, but not in calf aggrecan (Barry et al., 1995). This may contribute to the age-related differences observed in the assembly of proteoglycan ternary aggregates. Link protein also has two N-glycosylation sites; therefore, it is important to understand the effect of the glycosylation of link protein on
HA binding as well.
- 15 -
Fig. 1-6 Sequence alignment of human PTR domains from the members of the human link protein and lectican families. From top to bottom are human TSG-6 and (a) PTR1 or (b) PTR2 from human cartilage link protein, HAPLN-3 (cardiovascular), brain link protein-1, brain link protein-2, aggrecan, versican, brevican, and neurocan. Letters in red are non-charged polar residues. Letters in blue are charged polar residues. Letters in green are non-polar residues. Shaded letters in yellow are conserved between residues of TSG-6 involved in HA binding (see Fig. 1-7). Shaded letters in blue are conserved between residues of TSG-6 involved in HA binding, but not involved in HA binding of neurocan. Shaded letters in gray are residues involved in HA binding of TSG-6, but not conserved in the link protein and lectican families.
- 16 -
Fig. 1-6 (Continued)
- 17 -
Table 1-I
HA binding proteins containing a single or double PTR domain(s)
Number of PTR domain(s) HA binding protein Major tissues expressed Ubiquitous (Day and Prestwich, Single CD44 2002) TSG-6 (tumor necrosis factor-stimulated gene-6) Cartilage (Maier et al., 1996), etc. Lymph vessel endothelium (Banerji LYVE-1 et al., 1999) Stabilin-1 Myeloid cell (Nagase et al., 1996) KIA0527 Brain (Nagase et al., 1998) Cartilage (Treadwell et al., 1980), Double Cartilage link protein (HALPN1) etc. Brain link protein-1 (HAPLN2) Brain (Hirakawa et al., 2000) Brain link protein-2 (HAPLN4) Brain (Bekku et al., 2003) Cardiovascular system (Ogawa et al., HAPLN3 2004) Cartilage (Heinegard and Gardell, Aggrecan 1967), etc. Versican Ubiquitous Brevican Brain (Yamada et al., 1994) Neurocan Brain (Rauch et al., 1992)
- 18 - 1.2.7 HA binding site of PTR domain
In 1996, the three dimensional NMR solution structure of a single PTR domain from
TSG-6 was solved (Kohda et al., 1996). Based on the tertiary structure of TSG-6, site-directed mutagenesis studies have been conducted on TSG-6 and CD44 to identify the residues involved in HA binding (Bajorath et al., 1998; Day et al., 1996; Kahmann et al., 2000; Kohda et al., 1996; Mahoney et al., 2001). Five residues (Lys11, Tyr12, Tyr59,
Phe70, and Tyr78) of TSG-6 were found to be most important for HA binding (Mahoney et al., 2001) (Fig. 1-7). All three of the tyrosine residues (shaded in yellow in Fig 1-6) of the first PTR domain (PTR1) are conserved in both the link protein and the lectican families, while they are not fully conserved in the second PTR domain (PTR2) (shaded in gray in Fig 1-6). The positively charged K11 in TSG-6 is conservatively replaced with arginine in the link protein and lectican families. The hydrophobic phe70, however, is not conserved among the link protein and lectican families (shaded in gray in Fig. 1-6).
Recently, it has been suggested that the link protein and lecticans having two PTR domains exhibit different HA binding characteristics from those of TSG-6 and CD44 with a single PTR domain by using recombinant neurocan as experimental model (Rauch et al.,
2004). This was demonstrated by showing that the positively charged residues conserved among all the PTR domains involved in HA binding of TSG-6 (K11) and CD44 (R41)
(shaded in blue in Fig 1-6) are not involved in HA binding of neurocan (Rauch et al.,
2004). This difference may relate to some structural differences between the second PTR domain of link protein and the lectican families and the single PTR domain of CD44 and
TSG-6.
- 19 -
Fig. 1-7 HA-binding site on a single PTR domain of human TSG-6. The amino acids shown in red are involved in HA binding, whereas those in green are not. Mutation of the purple residues resulted in changes of the native ternary structure; therefore, their involvement in HA binding cannot be determined. Figure adapted from (Mahoney et al., 2001).
- 20 - Therefore, homology modeling of link protein based on the solved solution structure of
the single PTR from TSG-6 may not be appropriate for estimating the HA binding sites in
link protein and lecticans having paired PTR domains. Elucidation of the
three-dimensional structure of paired PTR domains is a priority for investigating the
mechanisms of their interactions with HA under normal physiological conditions.
Structural determinations may also reveal the nature of the previously described
conformational “maturation” of the G1 domain of aggrecan in the ECM (Oegema, 1980).
In this study, an optimal expression system will be discussed for producing full-length
link protein or a pair of PTR domains (PTR1+2), which may be useful for structural and
functional studies enabling a better understanding of their HA-binding characteristics.
1.3 Aggrecan degradation
1.3.1 Aggrecan catabolism by members of the ADAMTS family
Aggrecan degradation can be observed both in normal aggrecan turnover and in
pathological conditions. It is believed, however, that when the rate of aggrecan
catabolism exceeds aggrecan anabolism, cartilage starts to degenerate. The loss of
aggrecan significantly contributes to cartilage degeneration found in osteoarthritis.
Furthermore, it is now widely accepted that aggrecan degradation found in osteoarthritis is largely protease-dependent (Lohmander et al., 1993; Patwari et al., 2005). Two families of enzymes are thought to be involved in degrading aggrecan in articular cartilage.
Historically, members of the MMP family were considered to be likely suspects, since these proteinases were found to degrade aggrecan in vitro at S341-F342 (bovine) or
- 21 - N341-F342 (human) in the IGD and at other sites as shown in Table 1-II. Later, fragments having the N-terminal sequence of A374RGS were identified in synovial fluid from patients with osteoarthritis, joint injury, and inflammatory joint disease (Lohmander et al., 1993; Sandy et al., 1992). These fragments can be generated by proteolytic cleavage of aggrecan at E373-A374 (bovine and human) in the IGD with the release of the C-terminal fragments into the synovial fluid. This suggested that the unidentified proteinase termed “aggrecanase,” which can cleave aggrecan at E373-A374, might be the primary proteinase responsible for aggrecan degradation in the cartilage ECM under pathological conditions. In addition, it was reported that aggrecanase-mediated fragments were identified in the culture medium of cartilage explants and chondrocyte cultures undergoing matrix degradation (Lark et al., 1995) to further suggest the involvement of aggrecanase in the pathological conditions. Conversely, it was suggested that
MMP-mediated cleavage of aggrecan is not responsible for the release of the majority of aggrecan fragments from cartilage (Little et al., 1999; Sandy and Verscharen, 2001;
Tortorella et al., 1999).
The quest to search for the identity of this aggrecanase lasted almost nine years until
Tortorella and co-authors identified the first aggrecanase as ADAMTS4 (aggrecanase-1)
(Tortorella et al., 1999). It is now known that many other members of the ADAMTS family (Tang, 2001) also possess “aggrecanase” activity, including ADAMTS-5 as shown in Table 1-II. ADAMTS4 and ADAMTS5 are suggested to be the most potent
“aggrecanases” identified and are likely to play major roles in detrimental degradation of aggrecan in the cartilage ECM. Both ADAMTS4 (aggrecanase-1) and ADAMTS5
- 22 - (aggrecanase-2) (Abbaszade et al., 1999; Tortorella et al., 1999) have been isolated,
cloned, and extensively characterized for their substrate specificity. The five major
ADAMTS4 and 5 cleavage sites identified in aggrecan are shown in Fig. 1-8. As seen in
Fig. 1-9, ADAMTS4 and 5 cleave the C-terminal of a glutamate (P1) followed by glycine, alanine, or leucine at the P1’ site. Sequence N-terminal to the P1 site are rich in threonine, serine, and hydrophobic residues (Fig. 1-9). Notably, all the cleavage sites within the
CS-2 domains lack Ser-Gly sequences within the region directly N-terminal to the P1 site, hence creating gap regions having no CS. The absence of CS in close proximity to the site of scission may be important for efficient substrate recognition by ADAMTS4 and 5.
Additionally, it is now shown that ADAMTS4 can cleave at the MMP cleavage site
(S341-F342, bovine sequence) after prolonged incubation (Westling et al., 2002). It is not presently known whether this cleavage by ADAMTS4 also occurs in vivo. Tortorella and co-workers (Tortorella et al., 2000) have proposed that recombinant ADAMTS4 cleaves the E1666-G1667 site preferentially first and that other sites are subsequently cleaved as described in Fig. 1-8. Later, however, it was found that ADAMTS4 has a number of isoforms with different substrate specificities (Gao et al., 2004; Kashiwagi et al., 2004).
Therefore, this observation may only be applicable for a specific isoform of ADAMTS4
(to be discussed later in this chapter). Although Tortorella and co-workers reported earlier
that the cleavage site in the IGD is thought to be less favored than sites in the CS-2
domain (Tortorella et al., 2000), cleavage in the IGD is most detrimental to cartilage
integrity, since it separates the HA binding G1 domain that anchors aggrecan in the ECM
from sGAG-rich regions (KS, CS-1, and CS-2 domains) that are important for cartilage
- 23 - hydration (see Figs. 1-2 and 1-8). Therefore, many laboratories have aimed to characterize the mechanism of cleavage in the IGD (Horber et al., 2000; Hughes et al.,
1997; Mercuri et al., 2000). From the standpoint of designing pharmacological interventions, however, it is also important to understand the mechanism of the initial cleavages in the CS domains and subsequent cleavage within the IGD.
1.3.2 Neo-epitope antibodies
In studies related to aggrecan catabolism, it is important to identify aggrecan catabolites in normal and osteoarthritic cartilage. Such identification became possible with the availability of a number of antibodies specific for the N-terminal or C-terminal peptide “neo-epitopes” generated after aggrecanase- and MMP-mediated catabolism of the aggrecan core protein (Hughes et al., 1995; Lark et al., 1995; Lee et al., 1998; Sandy and Verscharen, 2001; Sztrolovics et al., 1997; Tortorella et al., 2000). For example, the cleavage within the IGD is identified by anti-NITEGE373 and anti-A374RGSV neo-epitope antibodies that bind to fragments generated by ADAMTS4 and 5 cleavage at E373-A374.
Similarly, the cleavage within the CS-2 domain can be identified by the presence of fragments reactive to anti-TAGELE1480, anti-TFKEEE1666 and other antibodies. In this thesis, we will use these antibodies to study the substrate specificity of ADAMTS4. A detailed explanation of neo-epitope antibodies is given in Chapter 4.
- 24 -
Table 1-II
Sites within the aggrecan core protein specifically cleaved by members of MMPs
and ADAMTS families
Protease Major cleavage sites family Protease (shown in mature bovine aggrecan sequence) MMP MMP13 S341-F342, P384-D385 (Fosang et al., 1996) MMP8 S341-F342, E373-A374 (Fosang et al., 1994) MMP2 S341-F342 (Fosang et al., 1992) MMP9 S341-F342 (Fosang et al., 1992) MMP7 S341-F342, D444-L445 (Fosang et al., 1992) MMP14 S341-F342, D444-L445, Q354-T355 (Fosang et al., 1998) MMP1 S341-F342 (Fosang et al., 1993) MMP10 S341-F342 (Mercuri et al., 1999) MMP3 S341-F342 (Fosang et al., 1991) E373-A374, E1480-G1481, E1666-G1667, E1771-A1772, E1871-L1872 (Tortorella et al., 2000), S341-F342 (Westling ADAMTS ADAMTS4 et al., 2002) E373-A374, E1480-G1481, E1666-G1667, E1771-A1772, ADAMTS5 E1871-L1872 (Tortorella et al., 2002) E1871-L1872 (Kuno et al., 2000; Rodriguez-Manzaneque et ADAMTS1 al., 2002) ADAMTS9 E1771-A1772 (Somerville et al., 2003) ADAMTS8 E373-A374 (Collins-Racie et al., 2004)
- 25 -
Fig. 1-8 Structure of aggrecan monomer and the major proteolytic cleavage sites within the bovine aggrecan core protein by MMP13 and ADAMTS4 and 5. Aggrecan has three globular domains, G1, G2, and G3. The CS-1 and CS-2 domains are highly substituted with chondroitin sulfate (CS) chains. Keratan sulfate (KS) is located predominately in the KS domain and also in the IGD. A cleavage in the IGD domain results in complete loss of the sulfated glycosaminoglycan-rich C-terminal region of aggrecan. Numbers indicate the order of cleavage preference (Tortorella et al., 2002; Tortorella et al., 2000).
- 26 -
Fig. 1-9 Sequence alignment of IGD and nodal regions in the CS-2 domains neighboring the aggrecanase cleavage sites in bovine aggrecan. The sequences preceding the aggrecanase cleavage site in the IGD and CS-2 domains are rich in threonine and serine residues (black bold) as well as hydrophobic residues, such as valine, leucine, and proline (green bold). Residues highlighted in yellow are conserved. Note that these sequences are commonly found in highly O-glycosylated regions.
- 27 - 1.3.3 Regulation of ADAMTS4 activity
It has been suggested that the expression of aggrecanase activity (later designated as
ADAMTS4 and 5) is upregulated by the action of a pro-inflammatory cytokine, interleukin-1 (IL-1), TGF-β, and all-trans retinoic acid (Bonassar et al., 1997; Yamanishi et al., 2002). More recently, it has been reported that IL-1 stimulates ADAMTS4 activity
to cleave within the IGD by upregulating the conversion of ADAMTS4 from a less active
form into a more active form by promoting C-terminal truncation through the action of
MT4-MMP (Patwari et al., 2005).
The unprocessed ADAMTS4 (p100) (837 amino acid) contains the prodomain
(residue numbers, 1-212), the catalytic domain (213-436), the disintegrin-like motif
(437-519), the thrombospondin-1 like motif (520-576), the cysteine-rich domain
(577-685), and the spacer domain (686-837) (Fig. 1-10) (Flannery et al., 2002). Wang and co-authors showed that full-length recombinant ADAMTS4 (p100) expressed in HEK293
cells is secreted into medium as a processed form (p68, 213-837) lacking the prodomain,
which is cleaved off in the trans-Golgi network by proprotein convertase (Wang et al.,
2004). Gao and co-authors suggested that the p68 form of ADAMTS4 cleaves aggrecan in the CS-2 domain effectively, but does not cleave within the IGD by using rat aggrecan as the experimental substrate (Gao et al., 2002). Later, Kashiwagi and co-authors also demonstrated that the p68 form is most effective in cleaving within the CS-2 domain, but not in the IGD (Kashiwagi et al., 2004). ADAMTS4 acquires classical “aggrecanase activity” as defined by the ability to cleave within the IGD (at E373-A374) by a subsequent C-terminal truncation, which is mediated either by autoproteolytic digestion
- 28 - (Flannery et al., 2002), or by the action of MT4-MMP (Gao et al., 2004). At least two
forms of ADAMTS4 (p53 and p40) can result from C-terminal truncation (Fig. 1-10).
The p53 form lacks the spacer domain, whereas the p40 form lacks both the cysteine-rich
and the spacer domains. Flannery and co-authors have shown that both the cysteine-rich
and the spacer domains contain multiple GAG-binding motifs; and both p53 and p40
effectively cleave aggrecan within the IGD (Flannery et al., 2002; Gao et al., 2002;
Kashiwagi et al., 2004). Although, the thrombospondin motif also contains a
GAG-binding motif, both p53 and p40 have significantly lower affinity to GAG chains
compared with that of p68 (Flannery et al., 2002). It was proposed that p53 and p40 can
bind to GAGs via the GAG-binding motif (a thrombospondin motif) only with moderate
affinity, and therefore are able to dissociate from GAGs more easily than can p68, which
has a stronger affinity for GAGs. Hence unlike p68, the p53 and p40 forms are able to
dissociate and bind to the next cleavage site more rapidly than can the p68 form of
ADAMTS4 and therefore have greater “aggrecanase” activity (Flannery et al., 2002).
Although, the in vitro incubation of p68 resulted in the generation of C-terminal truncated ADAMTS4 (p53 and p40) by autoproteolytic cleavage (Flannery et al., 2002),
Gao and co-authors have shown that p53 is generated by the activity of MT4-MMP in both chondrosarcoma cell cultures and IL-1 stimulated bovine cartilage explants (Gao et
al., 2004; Patwari et al., 2005). Patwari and co-authors further suggested that the IL-1
stimulated cartilage explants have aggrecanase activity that is upregulated via increased
MT4-MMP activity (Patwari et al., 2005). This result indicates that MT4-MMP plays a central role in regulating aggrecanase activity (Patwari et al., 2005).
- 29 -
1 213 437 520 577 686 837
Pro Catalytic Disintegrin TS Cys-rich Spacer
Y590NHR p100
p68 ADAMTS4 isoforms p53 p40
GAG-binding motif
Fig. 1-10 ADAMTS4 isoforms and GAG binding motifs. ADAMTS4 consists of 6 functional domains and motifs. The prodomain is removed before ADAMTS4 is secreted (Wang et al., 2004). Three major isoforms, p68, p53, and p40, have proteolytic activity that cleaves aggrecan (Flannery et al., 2002). The p68 form tightly binds to GAGs, whereas p53 and p40 have reduced binding affinity for having fewer GAG binding sites (Flannery et al., 2002). Note that the anti-ADAMTS4 antibody used in Chapter 4 recognizes the Y590NHR sequence within the cysteine-rich domain.
- 30 - Based on these most current observations (Flannery et al., 2002; Gao et al., 2004; Patwari et al., 2005), the model shown in Fig. 1-11 can be proposed, which describes the
ADAMTS4 activation pathway.
1.3.4 Aggrecan structure and substrate specificity of ADAMTS4
It has been suggested that there is an age-dependent change in the susceptibility of aggrecan (Pratta et al., 2000; Roughley et al., 2003) to ADAMTSs and such changes may be due to age-dependent changes in aggrecan glycosylation (Barry et al., 1995). As described earlier, aggrecan undergoes extensive post-translational modifications and is highly substituted with O-linked and N-linked oligosaccharides, which comprise about
90% of its molecular mass.
The majority of these oligosaccharides are CS polysaccharide chains attached in the
CS-1 and CS-2 domains, and O-linked KS polysaccharides attached in the KS domain
(Fig. 1-8). O-linked KS chains are highly substituted in the KS domain and some O-link and N-link KS chains are also found in the G1 domain and the IGD (Barry et al., 1995).
On the other hand, CS chains are located mostly in the CS-1 and CS-2 domains (at
Ser-Gly). Furthermore, there are up to nine potential N-linked glycosylation sites
(Asn-X-Ser/Thr motif) in the G1, IGD, G2, and G3 domains. Barry and co-workers showed that there is an age-dependent variance in glycosylation sites in the HA-binding region of aggrecan (HABR), which includes the G1 domain and a part of the IGD domain
(Barry et al., 1995). For example, T42 was substituted with KS only in the steer but not in the calf aggrecan as described earlier (Fig. 1-12).
- 31 -
Substrate 1 213 437 520 577 686 837 specificity p100 Pro Catalytic Disintegrin TS Cys-rich Spacer Not active Proprotein convertase (trans Golgi)
p68 Catalytic Disintegrin TS Cys-rich Spacer CS-2 only MT4-MMP (ECM)
p53 Catalytic Disintegrin TS Cys-rich IGD & CS-2
??? (ECM)
p40 Catalytic Disintegrin TS IGD & CS-2
Fig. 1-11 ADAMTS4 activation pathway and substrate specificity of each isoform. The N-terminal prodomain of full-length ADAMTS4 (p100) is removed by proprotein convertase in the trans Golgi network (Wang et al., 2004). Secreted p68 binds to MT4-MMP on the cell surface where C-terminal truncation takes place to obtain p53 by removal of the spacer domain (Gao et al., 2004). The p53 remains on the cell surface by binding to GAGs on membrane-anchored syndecan-1 (Gao et al., 2004). By an unknown mechanism, a cysteine-rich domain can be removed and p40 is released into the medium (Gao et al., 2004).
- 32 - They also found that KS is linked at either N368 or T370 in the aggrecan of the steer, but not in that of the calf. Furthermore, KS chains at T352, T355, and/or T357 in the steer
were found to be longer than those in the calf. Pratta and co-workers focused on
age-dependent variations in the glycosylation of these sites because of their close
proximity to the “aggrecanase” cleavage site at E373-A374 (NITEGE373↓-A374RGSV)
(Pratta et al., 2000). They found that aggrecan isolated from older animals is more
susceptible to aggrecanase-mediated cleavage. Similar observations were made with the
in vitro digestion of human neonatal and adult aggrecan and fetal and adult cow aggrecan
with ADAMTS4, showing that in both species, adult aggrecan was more susceptible to
these enzymes (Roughley et al., 2003). On the contrary, ADAMTS5 showed little
variation in its substrate specificity against the aggrecans isolated from different age
groups (Roughley et al., 2003). Although the mechanism is not known, the age-dependent
difference in aggrecan’s susceptibility is more profound in cow than in human (Roughley
et al., 2003). These authors suggested that the increased susceptibility of older aggrecan to aggrecanase might be correlated with the age-dependent change in glycosylation in the
IGD (Fig. 1-12). They further suggested that the low susceptibility of aggrecan from fetal cow might be due to the presence of non-KS oligosaccharide near the cleavage site within the IGD (Roughley et al., 2003). Although Barry and co-authors did not observe N-linked
KS oligosaccharide in calf aggrecan at N368 (Barry et al., 1995), Roughley and co-authors suggested the presence of non-KS N-linked oligosaccharide at this site of aggrecan isolated from younger individuals (Roughley et al., 2003).
To investigate whether the KS and other oligosaccharide substitutions on aggrecan
- 33 - affect its susceptibility to aggrecanase, Pratta and co-workers treated cartilage-derived
aggrecan with keratanases and chondroitinase to remove KS and CS, respectively, and
characterized changes in its susceptibility to aggrecanase (Pratta et al., 2000). They found
that when aggrecan was treated with keratanase, the aggrecanase-dependent cleavage
within the IGD was absent, whereas the treatment of aggrecan with chondroitinase had little effect on cleavage within the IGD.
As discussed above, ADAMTS4 has been shown to have multiple sGAG-binding
motifs, which are able to bind to GAGs (Flannery et al., 2002). Furthermore, Tortorella
and co-workers showed that synthetic peptides having the sequence of an sGAG binding
motif (in the thrombospondin motifs) were important for substrate recognition (Tortorella
et al., 2000). Therefore, they suggested that KS might be required for ADAMTS4 to
recognize aggrecan as a substrate. However, several contradictory results have also been
reported, which refute the necessity of KS for aggrecanase cleavage. For example, it has
been shown that aggrecan produced in rat chondrosarcoma cells and recombinant aggrecan expressed in insect cells, both of which lack KS, are also cleaved by
aggrecanase (Lark et al., 1995; Mercuri et al., 1999) supporting the idea that KS
substitution might not be a necessary factor for aggrecanase to cleave substrates. Rather,
it is possible that KS might influence the rate of cleavage by ADAMTS4.
- 34 -
Fig. 1-12 Distribution of oligosaccharide chains in the HA-binding region (HABR) of calf and steer aggrecan. Figure was adapted and modified from (Barry et al., 1995). Calf (immature) and steer (mature) bovine aggrecans have different oligosaccharide distribution in the HA-binding region (HABR). Note the absence of KS chains in calf aggrecan at T42 and N368 and/or T370.
- 35 - 1.4 Focus of thesis work
In the work described in Chapters 2 and 3, bovine and human cartilage link protein
and bovine aggrecan have been expressed in several heterologous expression systems,
and biochemically and functionally characterized. In Chapter 2, full-length and truncated
link protein constructs were expressed in E. coli to obtain recombinant proteins for use in
structural and functional studies. In Chapter 3, full-length recombinant link protein and
aggrecan were expressed in mammalian cells and the biochemical properties of each were
compared with those of cartilage-derived molecules. The aim of this aspect of the work
was to obtain functional recombinant proteins that could be used as experimental models
for investigating the mechanisms of proteoglycan ternary aggregate assembly and
catabolism.
To begin to study the catabolic events of osteoarthritis, both full-length recombinant
bovine aggrecan (characterized in Chapter 3) and cartilage-derived aggrecan were used as
experimental substrates for ADAMTS4 to further investigate the hypothesis that
glycosaminoglycans can alter aggrecan’s susceptibility to ADAMTS4 by the following
experiments. First, we biochemically characterized recombinant aggrecan expressed in
several mammalian cell lines. The characterization of recombinant aggrecan expressed in
COS-7 and other cell lines with different glycosylation potentials is described in Chapter
3. Second, we describe in Chapter 4 whether substitution of aggrecan with KS, CS, or other oligosaccharides affects the susceptibility of aggrecan to cleavage within the aggrecan core protein by ADAMTS4. This was addressed by enzymatically removing KS or CS (or both) from cartilage-derived aggrecan and by utilizing recombinant aggrecan
- 36 - expressed in cell lines with altered KS or CS glycosylation as an experimental substrate
for ADAMTS4. Lastly in Chapter 4, we performed site-directed mutagenesis studies to
characterize the effects of potentially glycosylated specific amino acid residues on
ADAMTS4 recognition and digestion. These studies were conducted by mutagenizing
potential glycosylated threonine, serine, and asparagine residues within the IGD of
aggrecan and comparing the resulting mutant aggrecans’ susceptibility to ADAMTS4
with that of wild-type aggrecan.
A general summary of this thesis work and a discussion of potential future studies to
extend our findings are presented in Chapter 5. Such studies include the functional analysis of recombinant link protein and recombinant aggrecan and further characterization of the effects of glycosylation on aggrecan catabolism by ADAMTS4.
- 37 - Chapter 2
UExpression,U Purification, and Refolding of a Pair of Recombinant Proteoglycan
Tandem Repeat Domains of Link Protein from Escherichia coli
Summary
Several approaches were explored for expressing proteoglycan tandem repeat (PTR)
domains from bovine link protein for structural and hyaluronan (HA) interaction studies.
A pair of recombinant PTR domains (PTR1+2) from bovine cartilage link protein was
expressed in E. coli as a fusion protein with maltose binding protein (MBP). It was found
that under reducing conditions, the purified MBP/PTR1+2 fusion protein ran on
SDS-PAGE with an apparent molecular mass (67 kDa) close to the calculated molecular
mass (64.9 kDa), while under non-reducing conditions, it ran as a series of heterogeneous
high molecular mass multimers. These results suggest that MBP/PTR1+2 was
multimerized by improper folding and intermolecular disulfide bond formation between
the eight cysteine residues present in the PTR domains. Furthermore, when the PTR
domains were cleaved from MBP by factor Xa protease digestion, the PTR domains
appeared to form insoluble aggregates. However, the cleaved PTR domains could be
refolded after separation on Sephacryl S-300 chromatography under reducing and
denaturing conditions, followed by slow removal of the reducing and denaturing reagents
by sequential dialysis. After this procedure, most of the PTR domains ran as monomers
under non-reducing conditions on SDS-PAGE, giving non-aggregating species.
Full-length and variably truncated (Ig-fold, PTR1+2, PTR1, and PTR2 domains) recombinant human link protein constructs were also expressed in E. coli and purified.
- 38 - These MBP-fusion proteins and bovine MBP/PTR1+2, which were not subjected to the
refolding protocol, bound zinc as previously shown by others for native link protein.
These proteins will, however, require additional refolding prior to use in structural and
functional studies since they also form high molecular mass aggregates on non-reducing
SDS-PAGE. Once refolded, these constructs may be useful for studying the structure and
function of the individual functional domains (Ig-fold, PTR1, and PTR2) of cartilage link
protein. Because the yield of refolded protein from the E. coli/MBP-system is relatively
low and requires additional purification steps, the PTR1+2 domains were also expressed
in an E. coli/thioredoxin-system and in a yeast (Pichia pastoris) expression system.
Unfortunately, the thioredoxin/PTR1+2 expressed in E. coli was only recovered from
inclusion bodies, and the PTR1+2 domains expressed in yeast with a signal sequence for
secretion were not detectible in the culture medium. Therefore, the E. coli/MBP-system
was found to be the best non-mammalian system for expressing PTR1+2 domains among
the systems tested.
2.1. Introduction
Cartilage link protein (HAPLN1) is a well-studied hyaluronan binding protein,
which stabilizes the interaction between aggrecan and hyaluronan (HA) in the formation
of proteoglycan aggregates (see Fig 1-2). Although it has been suggested that link protein
and aggrecan form a ternary complex with HA in a 1:1 (link protein: aggrecan) mole ratio
(Neame and Barry, 1993), a more recent study showed that the proteoglycan aggregates
isolated from mature cartilage contain a higher content of aggrecan (G1 domain) to link
- 39 - protein (a 2-3:1 ratio) than did the aggregates isolated from newborn and younger
cartilage, which showed roughly a 1:1 ratio (Wells et al., 2003). Link protein consists of
three domains, the Ig-fold domain, which is responsible for aggrecan binding, and two
PTR domains (PTR1 and PTR 2), which are responsible for HA binding. The PTR1+2
domains are also found in all members of the link protein and lectican families (see Fig.
1-6). Amino acid sequences of PTR domains within these families are highly conserved
among each other with homology ranging from 42 to 62%. Earlier studies using a series
of truncated cartilage link protein constructs (Grover and Roughley, 1994) and work from
Dr. Hering’s laboratory (Varelas et al., 1995) suggested that a single PTR domain is
sufficient for HA binding. However, more recent work using a truncated G1 domain of
aggrecan has suggested that both PTR 1 and PTR 2 are required for functional HA
binding (Watanabe et al., 1997). Elucidation of the three-dimensional structure of the
PTR1+2 domains should help to resolve the conflicting data regarding the mechanism of
their HA binding.
In order to conduct structural and functional studies of link protein, the full-length
and truncated recombinant cartilage bovine and human link protein constructs were
expressed in an E. coli expression system. The goal of this study was to obtain in high yield properly folded PTR1+2 domains to study their solution structure by NMR with
15 13
P P PP PN and PP PC labels. Since link protein has 10 cysteine residues that can form disulfide
bonds, a major challenge of expressing link protein in E. coli was devising a refolding
protocol for obtaining sufficient quantities of correctly folded protein for structural and
functional studies. As expected from earlier studies in Dr. Hering’s laboratory (Varelas et
- 40 - al., 1995; Varelas et al., 1997), both full-length and truncated proteins were found to form high molecular mass aggregates. In this work, we developed a method for successfully refolding aggregated PTR1+2 domains into monomers. This approach is more promising than alternative approaches that were attempted in this study using different expression systems to obtain functional recombinant PTR1+2 domains. The further optimization of the refolding protocol to obtain a sufficient quantity of monomers will permit structural and functional studies in the future.
2.2. Results and Discussion
U2.2.1U Cloning of link protein fragments from bovine and human link protein
Full-length and truncated link protein constructs were amplified by PCR from
bovine and human link protein cDNAs by using the primer sets described in the
“Experimental Procedures” and in Table 2-I, respectively. The combinations of upper and
lower primers used to amplify the specific regions of human link protein are indicated in
Table 2-II. Obtained PCR products were ligated into the pMALc2X vector to generate the
constructs described in Fig. 2-1.
- 41 -
Table2-I
Primers for construction of various human link protein constructs Sequences underlined represent the restriction sites used to ligate the PCR products into the XmnI-XbaI digested pMALc2X vector. All of the forward primers (F1, F2, and F3) contain blunt end restriction sites. The italic C (in F1) indicates a silent mutation introduced into the primer to generate a HincII restriction site. Letters in bold of the reverse primers (R4, R5, and R6) are stop codons.
Primer Sequence RE site
F1 5'-GCGCUUGTCGACUCATCTTTCAGACAACTAT-3' HincII
F2 5'-GCGCUUCACGTGUGTATTCCCTTACTTTCCA-3' PmlI
F3 5'-GCGCUUGCCGGCUCGTTTTTACTATCTGATC-3' NaeI
R4 5'-TTAAUTCTAGAU UTCAATTGAAATTGGATGTAAA-3' XbaI
R5 5'-TTAAUTCTAGAU UTCAGTTGTATGCTCTGAAGCA-3' XbaI
R6 5'-TTAAUTCTAGAU UTCAACCTTGTAAGTCCAGTGC-3' XbaI
- 42 -
Table2-II
Primer sets used for amplification of full-length and truncated human link protein
sequences
Domains (residues) Forward primer Reverse primer Full-length (16-354) F1 R4 Ig-fold (16-158) F1 R6 PTR1+2 (159-354) F2 R4 PTR1 (159-258) F2 R5 PTR2 (259-354) F3 R4
- 43 -
Fig. 2-1 Schematic representation of recombinant MBP/link protein constructs expressed in E. coli. Full-length link protein (residues; 16-354) comprises three functional domains containing a total of 10 cysteine residues. Locations of the primers (described in Table 2-I) are indicated by red arrowheads. Plasmid clone nomenclature is indicated on the left and the calculated molecular mass of each clone is indicated on the right. Numbers in parentheses are calculated molecular masses without the MBP. (Secreted form of link protein starts at residue D16). Disulfide bonds are paired as follows. Ig-Fold, C61-C139; PTR1, C181-C252, C205-C226; PTR2, C279-C349, C304-C325 (Neame et al., 1986).
- 44 - U2.2.2 Expression and purification of MBP/full-length and truncated recombinant bovine and human link protein fusion proteins in E. coli
Full-length and truncated cartilage link protein constructs were expressed in E. coli as fusion proteins with maltose binding protein (MBP). In these constructs, each recombinant protein’s N-terminus is linked to the MBP’s C-terminus separated by a factor Xa protease-cleavage peptide sequence (LEGR) (Fig. 2-1). The expression of the recombinant bovine PTR1+2 MBP fusion protein (MBP/bPTR1+2) was induced with
IPTG. Soluble cell lysates were applied to amylose affinity resin, which binds the maltose binding protein (MBP) fusion protein for protein purification, and the bound protein was eluted with maltose as described in “Experimental Procedures.” A band of the expected size (67 kDa) was observed in the pooled fraction (Fig. 2-2 a, lane 3) eluted from the amylose resin by 10 mM maltose (Fig. 2-2 b). The same size band was observed in cell lysates from the 3 h induction (67 kDa) (Fig. 2-2 a, lane 2), but was absent in the 0 h induction (Fig. 2-2 a, lane 1). Full-length and truncated human link protein constructs
(MBP/hLP, MBP/hIg-fold, MBP/hPTR1+2, MBP/hPTR1, and MBP/hPTR2) were expressed and purified in the same way as for MBP/bPTR1+2, except that these proteins were expressed in BL-21 codon-PLUS RIL cells instead of K12 PR745 cells. BL-21 codon-PLUS RIL cells contain additional tRNAs for mammalian codons not abundant in
E. coli tRNAs. Apparent molecular masses of all the expressed recombinant proteins are shown in Table 2-III. The apparent molecular mass of the human recombinant link protein was slightly larger than the calculated molecular mass (Table 2-III). This could be due to the gradient gel system used for the experiment described in Fig. 2-5, which was to
- 45 - calculate the molecular mass of intact MBP fusion proteins. As it will be discussed later, in this particular experiment, the factor Xa cleavage of recombinant protein was monitored, and a gradient gel was used to separate bands in a broad size range. The intact
recombinant protein migrated in the upper portion of the gel, where resolution is poor and
size estimation is inaccurate. On the other hand, bovine MBP/bPTR1+2 was determined
to have the correct molecular mass on a 10% non-gradient SDS-PAGE gel (Table 2-III).
Typical yields for all of these purified recombinant proteins from 1l of E. coli rich broth ampicillin culture were approximately 8-12 mg of culture with little variation between the different recombinant proteins.
- 46 -
Fig. 2-2 Expression and purification of MBP/bPTR1+2. (a) Cell lysates of 0-h (lane 1) and 3-h induction (lane 2) with IPTG. (b) Soluble cell lysates were loaded onto the amylose resin and bound proteins were eluted with 10 mM maltose. Protein elution was monitored by UV absorbance at 280 nm. Pooled fractions indicated under the bar were concentrated, and 3 μg of purified protein was separated on a 10% SDS-PAGE gel (lane 3) shown in (a). The arrow indicates that purified MBP/bPTR1+2 migrated at 67 kDa. The gel was stained with Coomassie brilliant blue R-250 (CBB).
- 47 -
Table 2-III
Apparent molecular mass of MBP fusion proteins on SDS-PAGE gels
Origin Domains Apparent molecular mass Calculated molecular mass Bovine MBP/PTR1+2 67 kDa 64.9 kDa MBP/LP 97 kDa 80.9 kDa MBP/Ig-fold 74 kDa 58.5 kDa Human MBP/PTR1+2 72 kDa 64.9 kDa MBP/PTR1 60 kDa 54.0 kDa MBP/PTR2 67 kDa 53.4 kDa
- 48 - U2.2.3 Factor Xa digestion of MBP-bovine PTR1+2 fusion protein
To separate the bovine PTR1+2 domains from their fusion partner MBP, the purified
fusion protein (MBP/bPTR1+2) was digested with factor Xa for 24 h or 48 h at either 37
ºC or 4 ºC and analyzed by SDS-PAGE. Coomassie brilliant blue staining of the gel
indicated that the optimal digestion was obtained after a 48-h incubation at 37 ºC (Fig.
2-3, lane 4). Two sizes of factor Xa-mediated products were generated, which may be due to cleavage within the PTR1+2 domain. Previous work from Dr. Hering’s laboratory has suggested the presence of a secondary cleavage site within the PTR1 domain. One of the
169
P potential secondary sites within the PTR1 domain has a sequence (URLGRUPP P-YNLNF)
similar to the sequence recognized by factor Xa (ULEGRU-XXXX). This site is located
10-amino acid residues C-terminal to the first amino acid of the PTR1 domain and is present in both human and bovine PTR1 domains. In order to determine the secondary factor Xa cleavage site within the PTR1+2 domain, MBP/bPTR1+2 was digested with factor Xa, separated on an 18% SDS-PAGE gel, and transferred to two PVDF membranes.
One of the membranes was immunoblotted with 8-A-4 anti-LP antibody (Fig. 2-4, lane 1),
which can interact with epitopes within the PTR1 or PTR2 domains, and the other
membrane was stained with Coomassie brilliant blue (CBB) (Fig. 2-4, lane 2). A CBB
stained band, which was also reactive to 8-A-4, ran at 23 kDa (PTR1+2’) and then was
excised and subjected to N-terminal amino acid sequencing analysis (Fig. 2-4, arrow).
Although the sequencing results contained multiple sequences, the data clearly indicated
170
P the presence of the YPP PNLNFHEA sequence that is adjacent to the putative factor Xa
169
P cleavage site RLGRP P in the PTR 1 domain. This result suggests that the majority of the
- 49 - digested PTR1+2 domain is cleaved by factor Xa at R169-Y170 within the PTR1 domain.
The products generated by factor Xa digestion of the MBP/bPTR1+2 fusion protein
were further analyzed by separation on a SDS-PAGE gel and either stained with
Coomassie brilliant blue (Fig. 2-5 a) or electrophoretically transferred to a PVDF
membrane for Western blot analysis with 8-A-4 anti-LP antibody (Fig. 2-5 b). Both undigested MBP/bPTR1+2 and factor Xa-digested PTR1+2 domains ran as monomers under reducing conditions (Fig. 2-5, lanes 1 and 2), whereas they formed non-specific high molecular mass aggregates under non-reducing conditions indicative for the formation of intermolecular disulfide bonds between the eight cysteines present in
PTR1+2 domains (Fig. 2-5, lanes 3 and 4). While both monomeric and multimeric
MBP/bPTR1+2 were able to enter the stacking gel (Fig. 2-5, lanes 3), once they were digested with factor Xa, neither PTR1+2 nor undigested MBP/bPTR1+2 were able to enter the stacking gel, indicating the formation of insoluble aggregates (Fig. 2-5, lanes 4).
This result suggests that the “cleaved” PTR1+2 domains may aggregate with the undigested MBP/bPTR1+2, which were otherwise soluble prior to digestion with factor
Xa.
- 50 -
Fig. 2-3 Factor Xa digestion of MBP/bPTR1+2. MBP/bPTR1+2 (lane 1) was digested with factor Xa at either 37 ºC (lanes 2-4) or 4 ºC (lanes 5-7) for 0, 24, and 48 h. Protein fragments were separated on a 15% SDS-PAGE gel and stained with CBB. The asterisks indicate the locations where the PTR1+2 and PTR1+2’ run on the gel for the 37 ºC digestion.
- 51 -
8A4 CBB
81 Intact MBP/bPTR1+2 51 MBP 33
28 Sequenced 20 kDa
1 2
Fig. 2-4 Factor Xa digested PTR1+2 analyzed with N-terminal amino acid sequencing. MBP/bPTR1+2 (25 μg) was digested with factor Xa (0.5 μg) for 48 h at 37 ºC, separated on an 18% SDS-PAGE gel, transferred to two PVDF membranes in 10 mM CAPS (pH 10.5) (glycine-free), and the membranes were either immunoblotted with 8-A-4 anti-LP antibody (lane 1) or stained with CBB (lane 2). A band run at 23 kDa (arrow) was subjected to the N-terminal amino acid sequencing.
- 52 -
Fig. 2-5 Factor Xa-digested MBP/bPTR1+2 analyzed on SDS-PAGE/Western blot. Undigested (lanes 1 and 3) and factor Xa-digested MBP/bPTR1+2 (lanes 2 and 4) were electrophoresed on an 18% SDS-PAGE gel under reducing (lanes 1 and 2) and non-reducing (lanes 3 and 4) conditions and either stained with (a) CBB or (b) transferred to a PVDF membrane, immunoblotted with 8-A-4 anti-LP (PTR1/2) antibody, and visualized with the NBT/BCIP system.
- 53 - U2.2.4 Factor Xa digestion of MBP-human link proteins
The MBP human link protein fusion proteins were digested with factor Xa. The effective digestion time varied among different truncation mutants of human link protein
(Fig. 2-6). The full-length and Ig-fold proteins appear to be cleaved faster by factor Xa than were PTR1+2 and PTR1 (Fig. 2-6 c and d), but their cleavage products were further degraded (Fig. 2-6 a, arrowhead at 15 kDa and b, asterisk). PTR2 appears to be relatively stable, since only a single band corresponding to PTR2 is observed (Fig. 2-6 g). Although
PTR1+2 and PTR1 appeared to be more stable than LP and Ig-fold, a secondary cleavage site was observed (Fig. 2-6 c and f, bands PTR1+2’ and PTR1’). This is most likely due to cleavage at R169-Y170, which is cleaved in the bovine PTR1+2 domain by factor Xa as described before. Western blot analysis of the factor Xa-digested PTR1 and PTR2 domains suggests that the secondary cleavage site is only present in the PTR1 domain, since a doublet is only observed in the PTR1 domain (Fig. 2-6 f), but not in the PTR2 domain (Fig. 2-6 g) after factor Xa digestion. Within 15 min digestion, two sizes of 8-A-4 anti-LP reactive PTR1 domain appeared (Fig. 2-6 f, PTR1 and PTR1’). Within 60 to 120 min digestion, only the lower molecular mass species is observed (Fig. 2-6 f, PTR1’).
Therefore, this suggests that the PTR1 domain was completely digested at the secondary cleavage site (Fig. 2-6 f, PTR1’). Furthermore, it is also suggested that cleavage occurred at the authentic factor Xa site primarily and subsequently cleaved at R169-Y170 within the PTR1 domain.
- 54 -
Fig. 2-6 Factor Xa-digested MBP/hLP constructs analyzed on SDS-PAGE/Western blot. Full-length and truncated mutants were digested with factor Xa at 37 ºC for up to 48 h. Protein fragments were separated on 10-20% SDS-PAGE gels and either stained with CBB (a-e), or transferred to PVDF membranes, immunoblotted with 8-A-4 anti-LP (PTR1/PTR2) antibody (f and g), and visualized with the ECL plus system. The asterisk indicates the area containing Ig-fold degradative products. Open arrows refer to contaminating protein present in (a, c-e). PTR1’ and PTR1+2’ refer to products cleaved at the potential secondary cleavage site at R169-Y170.
- 55 - U2.2.5 Enterokinase digestion of MBP/E-bPTR1+2U
Since a factor Xa susceptible site is present in the PTR1 domain, we generated a new MBP fusion construct containing an enterokinase cleavage site (DDDDK↓), instead of a factor Xa cleavage site (LEGR↓) to avoid any secondary cleavage within the link
protein. We chose enterokinase, since no sequence resembling the enterokinase cleavage
site is found in the PTR1+2 domain sequence. Purified fusion protein (MBP/E-bPTR1+2)
was digested with enterokinase for 20 h at room temperature and analyzed by
SDS-PAGE/Western blot (Fig. 2-7). The result shows that the recombinant
MBP/E-bPTR1+2 was highly resistant to digestion by enterokinases (Fig. 2-7 lanes 2-8),
although the enzyme (Enterokinase Max) purchased from Invitrogen showed somewhat
stronger activity (Fig. 2-7, lanes 2-5) than the enzyme from New England Biolabs
(Enterokinase, light chain) (Fig. 2-7, lanes 6-8). Furthermore, one of the cleaved products
that reacted with the 8-A-4 antibody was much larger (45 kDa) than the calculated
molecular mass of PTR1+2 (22 kDa) for unknown reasons, unless it formed a dimer
under reducing conditions. These results suggest that compared with enterokinase, factor
Xa is more effective in cleaving MBP from the PTR1+2 domains, even though it also
cleaves within the PTR1 domain. Since the factor Xa susceptible site within the PTR1
domain is located only 10 amino acids C-terminal from the original N-terminus (V159)
of PTR1 and those 10 amino acids do not contain cysteine residues, which are important
for protein folding, the factor Xa construct (MBP/bPTR1+2) was used for further studies.
- 56 -
Fig. 2-7 Enterokinase-digested MBP/E-bPTR1+2 analyzed on SDS-PAGE/Western blot. MBP/E-bPTR1+2 was digested with different concentrations of Enterokinase purchased either from Invitrogen (Enterkinase Max) (lanes 2-5) or New England Biolabs (Enterokinase, light chain) (lanes 6-8) at room temperature for 20 h. Protein fragments were separated on a 15% SDS-PAGE gel, transferred to a PVDF membrane, immunoblotted with 8-A-4 anti-LP antibody, and visualized with the NBT/BCIP system.
- 57 - U2.2.6 Refolding of monomeric PTR1+2 domainsU
Even after MBP/bPTR1+2 was digested with factor Xa for 48 h at 37 ºC, a
considerable amount of MBP/bPTR1+2 was not cleaved (Figs. 2-3, 2-4, and 2-5). In
addition, it was apparent that the PTR1+2 domains form high molecular mass aggregates.
Therefore, in order to isolate PTR1+2 domains after the factor Xa digestion and to obtain
a soluble monomeric form of PTR1+2, factor Xa-digested MBP/bPTR1+2 was separated on Sephacryl S-300 gel filtration chromatography in the presence of 6 M GnHCl with/without the reducing reagent DTT (Fig. 2-8). The peaks are identified in the right panel of Fig. 2-8. The results show that the high molecular mass aggregates were not disassociated into monomers by denaturation in 6 M GnHCl alone (Fig. 2-7 a and c, i.e.,
peaks 1 and 4, elute at VB0B BB). Peaks 1 and 4, however, disappear and apparently shift to peaks 3 and 8 in the presence of both DTT and 6 M GnHCl, indicative of the disassociation into monomers (Fig. 2-8 b and d, peaks 3 and 8). This suggests that high molecular mass aggregates were formed largely due to the formation of intermolecular disulfide bonds. Interestingly, when MBP/bPTR1+2 was not digested with factor Xa, a detectible amount of monomeric fusion protein is observed even in the absence of DTT
(Fig. 2-8 a, peak 2). However, the entire factor Xa-digested MBP/bPTR1+2 was eluted with the void volume, which presumably included both intact MBP/bPTR1+2 and
“cleaved” PTR1+2 (Fig. 2-8 c, peak 4). Again, this result suggests that the “cleaved”
PTR1+2 may HTH insolubilizeT THT other proteins, which would otherwise be soluble in the
absence of “cleaved” PTR1+2. A similar observation was made with an SDS-PAGE
analysis run under non-reducing conditions, which is described in Fig. 2-5.
- 58 - Peaks 1·············MBP/PTR1+2 aggregates 2, 3, 6·····MBP/PTR1+2 monomer 4·············MBP/PTR1+2 and PTR1+2 aggregates 5, 7·········MBP monomer 8·············PTR1+2 monomer
Fig. 2-8 Sephacryl S-300 size exclusion chromatography of factor Xa-undigested and digested MBP/PTR1+2 in 6 M GnHCl. Elution profiles of undigested MBP/PTR1+2 without (a) or with (b) DTT, and digested MBP/PTR1+2 without (c) or with (d) DTT. Numbers with an asterisk indicate the characteristic peaks obtained in each run. Major proteins that may be present in each numbered peak are shown in the legend on the right. The doubleheaded arrow in (d) indicates the fractions analyzed by Western blot in Fig.
2-9 after dialysis. VBBtB B is 83 ml (not shown in the graph). See “Experimental Procedures” for details of calibration. Elutions of protein standards are shown at the top of the chromatographs.
- 59 - To identify the proteins in each fraction of the factor Xa-digested MBP/bPTR1+2 (Fig.
2-8 d), each fraction (Nos. 29-39) was dialyzed against Tris-buffered saline (TBS) with
decreasing concentrations of GnHCl to slowly remove GnHCl as described in the
“Experimental Procedures.” Analysis of the dialyzed samples on SDS-PAGE under
non-reducing conditions revealed the presence of monomeric PTR1+2, which was
reactive to 8-A-4 anti-LP antibody (Fig. 2-9 b). This procedure was also effective in
separating “cleaved” MBP (Fig. 2-8, peak 7) and “intact” MBP/bPTR1+2 from “cleaved”
PTR1+2 domains, since these materials have larger molecular masses and were eluted in
the earlier fractions (Fig. 2-9, Nos. 29-35). Two anti-LP positive bands with apparent
molecular masses of 28 and 23 kDa were observed in the later fractions of peak 8 (Fig.
2-9, Nos. 34-38), consistent with the presence of a secondary factor Xa cleavage site
(R169-Y170) in the PTR1 domain, as discussed above. Although this procedure will be
useful for the isolation and refolding of the monomeric PTR1+2 domains, it is possible
that incorrectly matched intramolecular disulfide bonds may be formed. Future studies
will go one step further, to isolate correctly folded monomeric PTR1+2 by HPLC as
described by Day and co-authors, who isolated correctly refolded E. coli-expressed
TSG-6, which has 2 disulfide bonds (Day et al., 1996). This will be further discussed in the “future studies” in Chapter 5.
- 60 -
Fig. 2-9 Western blot analysis of dialyzed fractions from the Sephacryl S-300 gel filtration chromatography of factor Xa-digested MBP/bPTR1+2. Dialyzed fractions under the doubleheaded arrow of Fig. 2-8 (d) were separated on 18% SDS-PAGE gels either under (a) reducing or (b, c) non-reducing conditions, electrophoretically transferred to PVDF membranes, and immunoblotted with (a, b) 8-A-4 anti-LP antibody and visualized with the NBT/BCIP system or stained with (c) CBB.
- 61 - U2.2.7 Zinc (II) binding of various link protein constructs
In the present work, we also determined the zinc binding properties of recombinant
link proteins. It was first reported by Rosenberg and co-workers that link protein is a
metalloprotein capable of binding to divalent cations (Rosenberg et al., 1991). Previously,
we have shown that the full-length link protein and the PTR1 domain of link protein
expressed in fusion with MBP were able to bind zinc, but not MBP alone (Varelas et al.,
1995; Varelas et al., 1997), suggesting that a zinc-binding motif is present in the PTR1
domain. In this work, all of the soluble recombinant proteins generated in E. coli
including the MBP/hIg-fold were capable of binding to zinc-chelate affinity
chromatography (Figs. 2-10 and 2-11), being eluted at a low pH of 3.5. When factor
Xa-digested “non-refolded” MBP/bPTR1+2 was applied to the zinc column, free MBP
was eluted in the flow-through fractions (Fig. 2-10 d, lane 3), whereas intact
MBP/bPTR1+2 and bPTR1+2 were eluted with the low pH buffer (Fig. 2-10 c, lane 4)
suggestive of the specific interaction of PTR1+2 domains with zinc. Since the other
expressed link protein domains (i.e., Ig-fold, PTR1, and PTR2) were able to bind to zinc, it is likely that they might contain additional functional zinc binding motifs. In cartilage-derived link protein, there are no free cysteine residues since all of the cysteines are found paired in disulfide bonds (Neame et al., 1986). It is possible that cysteine residues, unpaired due to the misfolding of the protein, may be mediating the zinc binding of recombinant link protein constructs expressed in E. coli. Therefore, this experiment should be repeated using properly folded protein to correctly assess the zinc binding of each domain.
- 62 -
Fig. 2-10 Zinc binding of MBP/bPTR1+2 by zinc affinity chromatography. Elution profile of (a) MBP/bPTR1+2 or (b) factor Xa-digested MBP/bPTR1+2 from zinc-chelate chromatography. Protein was subjected to zinc chelate chromatography in a high pH buffer (0.15 M sodium acetate (pH 7.9), 0.2 M NaCl) and eluted with buffer at low pH 3.5. Samples not subjected to the zinc column (lanes undigested and digested) and peaks eluted from the zinc column (lanes 1-4 correspond to peaks 1-4) were separated on 15% SDS-PAGE gels under reducing conditions, transferred to PVDF membranes, and immunoblotted with (c) 8-A-4 anti-LP antibody and (d) anti-MBP antibody. Bands were visualized with the ECL plus system.
- 63 -
Fig. 2-11 Zinc binding of full-length and truncated human MBP fusion link proteins. Elution profiles of (a) MBP/hLP, (b) MBP/hIg-fold, (c) MBP/hPTR1+2, (d) MBP/hPTR1, and (e) MBP/hPTR2 from zinc-chelate chromatography are shown. The elution of each MBP recombinant protein was determined by ELISA with anti-MBP antibody (405 nm). Note that the fusion proteins were not digested with factor Xa.
- 64 - U2.2.8 Alternative approaches for expressing recombinant PTR1+2 domains
As described above, the MBP/PTR1+2 fusion protein expressed in E. coli formed non-specific aggregates and required additional denaturing/reducing and refolding procedures to obtain the monomeric form of PTR1+2. Therefore, we tried two different expression systems for expressing recombinant bovine PTR1+2, which might not require refolding to achieve a native conformation. First, we attempted to express PTR1+2 domains as a fusion protein with thioredoxin (Thio/PTR1+2). It has been suggested that thioredoxin catalyzes the rearrangement of mismatched disulfide bonds until the correct matches are made, and therefore promotes proper protein folding (Pigiet and Schuster,
1986). In Fig. 2-12, Western blot analysis with anti-LP antibody shows that recombinant
Thio/PTR1+2 was successfully expressed upon protein induction with arabinose for 4 h
(Fig. 2-12, lanes 2), and these anti-LP reactive bands were not present at the 0-h time point (Fig. 2-12, lanes 1). Recombinant protein, however, was highly degraded and only a negligible amount of recombinant protein was recovered from the soluble fraction of cell lysates (Fig. 2-12, lanes 3). This result suggests that most of the thioredoxin fusion
PTR1+2 domains were in insoluble inclusion bodies.
- 65 -
Fig. 2-12 Expression of thioredoxin PTR1+2 fusion protein in E. coli. Cell lysates of 0-h (lanes 1) and 4-h induction (lanes 2) with 0.4% arabinose and soluble fraction of cell lysates (lanes 3) were separated on a 4-20% gradient SDS-PAGE gel. Protein on a gel was either stained with (a) CBB, or was (b) electrophoretically transferred to a PVDF membrane and immunoblotted with 8-A-4 anti-LP antibody.
- 66 - Since the thioredoxin PTR1+2 fusion protein was not soluble, we decided to try
expressing the PTR1+2 domains in a yeast expression system (i.e., Pichia pastoris).
Unlike E. coli, yeast cells possess an endoplasmic reticulum (ER) in which protein folding and disulfide bond formation take place. In eukaryotes, protein disulfide isomerase (PDI), which resides in the ER, catalyzes the cleavage and reformation of disulfide bonds until a protein acquires its properly folded structure. The other advantage of using yeast is that yeast typically gives better yields of recombinant proteins than do other eukaryotic expression systems. The vector used for this study (pPICZα) is designed to have a signal-peptide at the N-terminus that allows the secretion of the recombinant
+
P protein. We expressed PTR1+2 domains in two Pichia pastoris strains, GS115 (MutPP P),
which has the wild-type ability to metabolize methanol, and KM71, which has a reduced
ability to metabolize methanol, to obtain optimal protein expression. The protein
expression of 4 clones (pBLP206-1, pBLP206-5, pBLP206-6, and pBLP206-11)
containing PTR1+2 and one clone with no insert (negative) (strain; KM71) was induced
with methanol for 144 h and both cell lysates and culture supernatants were analyzed.
The results show that proteins reactive with the anti-LP antibody are present in the
fractions of cell lysates at 144 h (Fig. 2-13 a, lanes 2, 4, 6, and 8), but absent at the 0-h
time point (Fig. 2-13 a, lanes 1, 3, 5, and 7). The sizes of these bands, however, are
significantly larger (between 60 –and 100 kDa) than the calculated molecular mass of
PTR1+2 (22.5 kDa). In contrast, no bands were observed in the supernatant (Fig. 2-13 b).
+
P Protein expression in the MutPP P phenotypic GS115: pBLP205-21 was induced with
methanol for 96 h and both supernatants (Fig. 2-14, lanes 5 and 6) and cell lysates (Fig.
- 67 - 2-14, lanes 7 and 8) were analyzed by Western blot. As controls, the culture supernatant
isolated from methanol-induced GS115 containing lacZ (Fig. 2-14, lanes 1 and 2) and an
empty vector (Fig. 2-14, lanes 3 and 4) were also analyzed. The results showed that
anti-LP-reactive bands are only present in the cell lysates isolated from the
methanol-induced PTR1+2 clone; 205-21 (Fig. 2-14, lane 8).
Since the specific expression of 8-A-4 anti-LP reactive protein in cell lysates of
205-21 was observed, we attempted to purify the protein by a nickel-charged chelate
affinity chromatography (Fig. 2-15) through an engineered polyhistidine tag, which is
located at the C-terminus of PTR1+2 domains. To purify the anti-LP reactive proteins
from the cell-lysate of the clone 205-21 after 120 h protein induction with methanol (Fig.
2+
P 2-15 b and c, lanes 2), the total cell lysate was applied to NiPP P chelate affinity
chromatography (Fig. 2-15 a), washed (Fig. 2-15 b and c, lanes 4-6), and eluted at pH 3.0
(Fig. 2-15 b and c, lanes 7). Although the fraction eluted at pH 3.0 contained a protein
that ran as a single band at 35 kDa under reducing conditions, the size of a band similar
to that of PTR1+2 domains did not strongly react with the 8-A-4 anti-LP antibody (Fig.
2-15 b and c, lanes 7, arrow). Therefore, to identify the single band obtained after
2+
P purification by NiP P affinity chromatography, we analyzed the N-terminal sequence of the unknown protein (Fig. 2-15 b, arrow). The purified sample (5 μg) was separated on 10%
SDS-PAGE and transferred to a PVDF membrane followed by staining with CBB. The
same amount of sample separated on 10% SDS-PAGE stained with CBB is shown in Fig.
2-16. Under reducing conditions, a single band was obtained. Under non-reducing
conditions, two bands were obtained (Fig. 2-16, lane 2), which appear to be derived from
- 68 - the single band observed under reducing conditions (Fig. 2-16, lane 1). Since, if this protein is the PTR1+2 domains containing disulfide bonds, it will run faster under non-reducing conditions, a lower molecular mass band that ran faster at 30 kDa (Fig.
2-16, lane 2) was excised and was subjected to N-terminal sequencing. The sequence obtained shows that the isolated protein was not from the PTR1+2 domains, but rather is closely related to alcohol dehydrogenase from Pichia stipitis based on a protein-protein
BLAST search of the obtained N-terminal sequence (Table 2-IV). The identical
N-terminal sequence was obtained by sequencing the concentrated peak C without further purification by SDS-PAGE, suggesting that the major protein in this preparation is an alcohol dehydrogenase-like protein. The expression of alcohol dehydrogenase is highly upregulated when the recombinant protein expression is induced with methanol. It is not
2+
P clear, however, why this protein was purified from the NiPP P affinity chromatography.
- 69 -
Fig. 2-13 Expression of PTR1+2 domains in KM71 (Pichia pastoris). Protein expression was induced with methanol for 144 h at 30 ºC in PTR1+2 clones, 206-1 (lanes 1 and 2), 206-5 (lanes 3 and 4), 206-6 (lanes 5 and 6), 206-11 (lanes 7 and 8), and no insert (lanes 9 and 10). Both (a) cell lysates and (b) supernatants at 0 h (lanes 1, 3, 5, and 7) and 144 h (lanes 2, 4, 6, and 8) were electrophoresed on a 16.5% Tris-Tricine SDS-PAGE gel under reducing conditions, transferred to PVDF membranes, and immunostained with 8-A-4 anti-LP antibody. An asterisk indicates three major bands reactive to 8-A-4.
- 70 -
lacZ Negative 205-21 205-21 (Sup.) (Sup.) (Sup.) (lys.) MW 0 96 0 96 0 96 0 96
106 77 50 35 28
kDa
1 2 3 4 5 6 7 8
Fig. 2-14 Expression of PTR1+2 domains in GS115 (Pichia pastoris). Protein expression was induced with methanol for 96 h at 30 ºC in PTR1+2 clones, lacZ (lanes 1 and 2), no insert (negative) (lanes 3 and 4), and 205-21 (lanes 5 - 8). Both supernatants at 0 h (lanes 1, 3, and 5) and 96 h (lanes 2, 4, and 6), and cell lysates at 0 h (lane 7) and 96 h (lane 8) were electrophoresed on a 16.5% Tris-Tricine SDS-PAGE gel under reducing conditions, transferred to a PVDF membrane, and immunostained with 8-A-4 anti-LP antibody. Arrows indicate 8-A-4 reactive bands, which only appeared after the 96-h induction with methanol.
- 71 -
Fig. 2-15 Purification of bPTR1+2 from Pichia pastoris (cell lysates) on Ni (II) chelate chromatography. (a) Cell-lysates from the 120-h induction with methanol were TM applied to a nickel chelate ProBondPP P column (Invitrogen) and washed with the native binding buffer (pH 7.8) followed by washing with the washing buffer (pH 6.0 and pH 5.5). Finally the bound protein was eluted with pH 3.0 elution buffer. (b and c) Cell lysates of 0-h (lanes 1) and 120-h (lanes 2) induction with methanol, flow-through fraction (lanes 3), wash fraction (lanes 4), peaks A (lanes 5), B (lanes 6), and C (lanes 7) were electrophoresed on a 15% Tris-SDS-PAGE gel, transferred to a PVDF membrane, and immunoblotted with (b) 8-A-4 anti-LP antibody. (c) The same membrane was subsequently stained with CBB.
- 72 -
2+
P Fig. 2-16 NiPP P purified protein analyzed by N-terminal amino acid sequencing. Proteins eluted in peak C (Fig. 2-15) were concentrated and about 5 μg of protein was loaded onto a 10% SDS-PAGE gel under reducing (lane 1) and non-reducing (lane 2) conditions in duplicate. One gel was stained with CBB, which is shown in the figure. Proteins on the other gel were transferred to a PVDF membrane in 10 mM CAPS (pH 11.0) (glycine-free), and the membrane was stained with CBB. A band that ran faster under non-reducing conditions (arrow) was subjected to N-terminal amino acid sequencing.
- 73 -
Table 2-IV
2+
P N-terminal amino acid sequence of Ni PP P purified 8-A-4 anti-LP weakly reactive
protein from Pichia pastoris The band separated on a 10% SDS-PAGE gel was excised and subjected to N-terminal amino acid sequencing (Fig. 2-15, lane 2, arrow). The unknown protein is most closely related to alcohol dehydrogenase from Pichia stipitis based on results of a BLAST search.
Protein N-terminal sequence Strain
Unknown (Peak C) SPTIPTTQ(K/L)AV I F E T TG Pichia pastoris
Alcohol dehydrogenase SP_IPTTQ K AVIFETNG Pichia stipitis
- 74 - 2.3. Conclusions
In this study, both full-length and truncated link proteins were expressed in E. coli as fusion proteins with MBP. Although the recombinant fusion proteins were purified from soluble fractions, they are highly aggregated as demonstrated on non-reducing
SDS-PAGE and gel filtration analysis. Since they do not aggregate under reducing
SDS-PAGE, they were apparently aggregated by the formation of mismatched intra- and intermolecular disulfide bonds. By using the MBP bovine PTR1+2 fusion protein, we developed a refolding protocol and were successful in obtaining monomeric PTR1+2 domains. This procedure, however, may require additional purification step(s) and analysis before protein with the correct disulfide bond pairing is obtained, since mismatched intramolecular disulfide bonds could still be present. Furthermore, the yield was relatively low and the method was somewhat time consuming.
To resolve this problem, we expressed the bovine PTR1+2 domains in E. coli as a fusion protein with thioredoxin, which is thought to assist the folding of proteins containing multiple disulfide bonds (Pigiet and Schuster, 1986). The expressed recombinant protein, however, formed exclusively insoluble inclusion bodies and was less soluble than the MBP fusion protein.
In a further attempt to obtain properly folded recombinant link protein in sufficient yield to conduct NMR structural studies, we used a eukaryotic expression system (Pichia pastoris) to express secreted recombinant PTR1+2 domains. Our results, however, show that no PTR1+2 domains were detected in the culture supernatant. Furthermore, although a small amount of anti-LP reactive protein was detected in the cell lysates of Pichia
- 75 - pastoris stably transfected with the PTR1+2 construct, the purified protein was not the recombinant PTR1+2.
We conclude from these studies that expression of link protein domains in E. coli
may ultimately be the best approach to generate material for NMR structural analysis. We
are encouraged by the success of our novel procedure for isolation and refolding of the
highly disulfide-bonded PTR domains.
In the next phase of this project, we shifted our focus to the production of full-length link protein in mammalian cells. This work is discussed in the following chapter.
2.4. Experimental procedures
U2.4.1 MaterialsU
pMALc2X, pMALc2E, pMALp2E, amylose resin, factor Xa, MBP polyclonal
antibody, enterokinase light chain, and restriction enzymes were purchased from New
TM England Biolabs (Beverly, MA). The EasySelectPP P Pichia expression kit,
TM pBAD/Thio-TOPO vector, TOP 10 cells, EnterokinaseMax, ProBondPP P column, and all
primers were purchased from Invitrogen (Carlsbad, CA). A cDNA for human link protein
(pSP8.1DBS) was a generous gift from Dr. Jayesh Dudhia (Royal Veterinary College,
London, UK). BL-21 codon-PLUS-RIL cells were purchased from Stratagene (La Jolla,
CA). Monoclonal 9/30/8-A-4 anti-LP antibody (supernatant) was purchased from the
Developmental Studies Hybridoma Bank, University of Iowa (Ames, IA). CHAPS, taq
polymerase, and T4 ligase were purchased from Roche Applied Science (Indianapolis,
IN). Protein electrophoresis kits and SDS-PAGE pre-cast gels were purchased from
- 76 - Bio-Rad (Hercules, CA). Sephacryl S-300, high-flow rate chelating resin, ECL plus,
Hyperfilms, anti-rabbit-IgG-horseradish peroxidase (HRP), and anti-mouse-IgG-HRP
were purchased from Amersham Biosciences (Piscataway, NJ). Immobilon-P,
Immobilon-PQ PVDF membranes, and Centriplus ultrafiltration devices were purchased
from Millipore (Bedford, MA). Anti-mouse-IgG-alkaline phosphatase (AP) conjugated
antibody and NBT/BCIP were purchased from Promega (Madison, WI). ELISA 96-well
high binding plates were purchased from Costar (Cambridge, MA). DTT and pNPP
tablets were purchased from Sigma-Aldrich (St. Louis, MO). DNA gel extraction kits and
plasmid prep kits were purchased from Qiagen (Valencia, CA). Acid-washed beads (0.5
mm) were purchased from Biospec Products, Inc. (Bartlessville, OK). All other
chemicals were purchased either from Fisher Scientific (Pittsburgh, PA) or
Sigma-Aldrich.
U2.4.2 Construction of E. coli link protein expression vectors
An expression vector for bovine link protein PTR1+2 (pBLP68-88) was constructed
in the pMALc2X vector as follows. A pair of PTR domains (PTR1+2) of bovine cartilage
link protein was amplified from a bovine cDNA clone (pBLP23-12A) (Varelas et al.,
1997) with primers 5’-ACGTUTACGTA UGTATTCCCTTATTTTCCA-3’ with a SnaBI site
and 5’-GTACUGGATCCUGATGATGTAGCCTAAACAGTT- 3’ with a BamHI site using
pBLP23-12A as a template. The PCR products were ligated into the pMAL-c2X vector,
which was digested with XmnI and BamHI. A clone containing the bovine DNA
sequence was designated pBLP68-88. Note that this construct contained mutations of C
- 77 - to T at 782 and A to G at 880 of the bovine link protein cDNA (GenBank Accession No.
BTU02292), which resulted in changes of two amino acid residues, P225L and N258D,
respectively.
An expression vector for bovine link protein PTR1+2 (pBLP203-36) was
constructed as follows in the pMALc2E vector. A cDNA (pBLP153-5) of full-length
bovine cartilage link protein used as a template was generated as follows. The mRNA
was freshly isolated from bovine articular chondrocytes, and a cDNA for bovine link
protein was generated by reverse transcription-PCR with primers
5’-AGCAGGACTTGAGAGCATCTG-3’ (upper) and
5’-GATGATGTAGCCTAAACAGTT -3’ (lower), and the PCR product was TA-cloned
into the pCRII vector (pBLP153-5). A pair of PTR domains of bovine cartilage link protein was amplified from a bovine cDNA clone (pBLP153-5) with primers
5’-UGAATTCUGTAGTATTCCCTTATTTTCCA-3’ (upper) with an EcoRI site and
5’-UTCTAGAUTTA(ATG)B6B BGTTGTATGCTCTGAAGCAGTA-3’B (lower) with an XbaI site
to incorporate a poly-histidine tag at the C-terminus with a stop codon. The EcoRI-XbaI
digested PCR product was ligated into the EcoRI-XbaI digested pMAL-p2E vector. A clone containing the bovine DNA sequence was designated pBLP204-2. The pBLP204-2 was then digested with EcoRI-XbaI to cut out PTR1+2 and then subcloned into the
EcoRI-XbaI-digested pMALc2E vector. A clone containing the bovine DNA sequence was designated pBLP203-36.
Full-length and truncated human link protein expression vectors were constructed as
follows. The DNA fragments of human cartilage link protein were amplified by PCR with
- 78 - the cDNA pAP8.1DBS as a template using the primer sets described in Table 2-I. Upper
primers carry various blunt cut restriction sites to insert the digested products into the
XmnI-digested pMALc2X vector so that fragments of the link protein sequence
immediately follow the sequence of the factor Xa site described in Fig. 2-1. All of the
lower primers carry the XbaI site.
Polymerase chain reaction products containing the sequences of the Ig-fold, PTR1+2,
and PTR2 domains were digested with XbaI followed by HincII, PmlI, and NaeI,
respectively, and then were ligated into the XmnI-XbaI-digested pMALc2X vector. Since
the 5’ end of the PCR product was not digested, the isolated clones (p207-41 (Ig-fold), p207-36 (PTR1+2), and p207-82 (PTR2)) containing the inserts were further digested with HincII (p207-41), PmlI-SalI (p207-36), and NaeI-SalI (p207-82), respectively. An insert cut from p207-41 was ligated into the XmnI-digested pMALc2X vector, and inserts cut from p207-36 and p207-82 were ligated into the XmnI-SalI-digested pMALc2X vector. TOP10 cells were transformed with plasmids, and positive clones containing inserts in the sense orientation were screened by DNA sequencing. These clones were designated pBLP251-45 (Ig-fold), pBLP251-21 (PTR1+2), and pBLP251-85 (PTR2).
Polymerase chain reaction products containing the full-length and PTR1 domain
sequences were digested with XbaI followed by HincII (full-length) and PmlI (PTR1),
and ligated into the XmnI-XbaI-digested pMALc2X vector. TOP 10 cells were
transformed with plasmids, and positive clones containing the inserts in the sense
orientation were confirmed by DNA sequencing. These clones were designated
pBLP251-6 (full-length) and pBLP251-62 (PTR1). All of the vectors except pBLP68-88
- 79 - were used to transform BL-21 codon plus-RIL competent cells that are designed to obtain
optimal expression of proteins containing "mammalian" codons rarely used in E. coli.
2.4.3 Expression and purification of MBP/full-length and truncated recombinant bovine
and human link protein fusion proteins in E. coli
Maltose binding protein fusion proteins were expressed and purified as described
below. Briefly, a glycerol stock of transformed E. coli (K12 PR745 and BL-21
codon-PLUS RIL cells) was streaked and cultured on a rich broth ampicillin plate (1%
tryptone, 0.5% yeast extract, 0.5% NaCl, 0.2% glucose, 2% agarose, and 50 mmol
ampicillin) overnight at 37 ºC. A single colony was picked and inoculated to 10 ml of rich
broth containing ampicillin (1% tryptone, 0.5% yeast extract, 0.5% NaCl, 0.2% glucose,
and 50 mmol ampicillin). The culture was grown with shaking overnight at 37 ºC. One
liter of rich broth amp was inoculated with 10 ml of the overnight culture and incubated
for 2 to 3 h until the OD600 reached 0.400-0.500. Protein expression was induced by
adding 3 ml of 0.1M IPTG (0.3 mM final concentration) to the culture, which was then
incubated with shaking for 3 h at 37 ºC. The E. coli culture was collected at 0-h
(pre-induction) (400 μl) and 3-h (post-induction) (200 μl) time points, microcentrifuged
and resuspended in 1 x SDS sample buffer (0.06125 M Tris-HCl (pH 6.8), 10% glycerol,
-6
P 2% SDS, 1% 2-mercaptoethanol, and 5 x 10P P% bromophenol blue) and analyzed on a
10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS- PAGE) gel as
described by Laemmli (Laemmli, 1970). Gels were stained with 0.1% Coomassie brilliant blue R-250 (w/v) (50% methanol (v/v), 10% acetic acid (v/v)) and destained in destaining
- 80 - solution (50% methanol (v/v), 10% acetic acid (v/v).
To purify MBP fusion protein, the bacterial culture was centrifuged at 5,000 rpm for
20 min with an SLA3000 rotor. The supernatant was discarded; the cell pellets were re-suspended in 50 ml of column buffer (20 mM Tris (pH 7.4), 200 mM NaCl, 1 mM
EDTA), and distributed into two 50 ml conical tubes (25 ml for each). Cell suspensions were frozen at - 20 ºC overnight and thawed on ice the next day. Cell walls were digested by adding lysozyme at 1 mg/ml followed by incubation on ice for 30 min. Cell lysates were sonicated on ice with a 50-60% pulse for a total of 5-10 min (1 min for each episode) and then were centrifuged at 11,000 rpm for 20 min with an SS-34 rotor.
Supernatants were diluted with 700 ml of column buffer and filtered through a 0.45 μm membrane with a Millipore filtering system. The crude extract was applied to an amylose-resin column (20 ml), which was pre-equilibrated with 200 ml of column buffer.
Flow through was reapplied to the column. The column was washed with 500 ml or more of column buffer until OD280 reached 0.01 or lower. Maltose binding protein fusion protein was then eluted with 100 ml of 10 mM maltose in column buffer. One hundred-eight drops (3 ml) of each fraction were collected, and OD280 was measured to monitor the elution of fusion protein. Fractions having an OD280 of 0.09 or above were pooled and concentrated with Centriprep 30 (Millipore) at 3,000 rpm with an SS-34 rotor until the total volume reached 12 ml. The sample was distributed into 1 ml amounts and stored at - 20 ºC until it was needed for further analysis. Purified MBP fusion protein (3
μg) was analyzed on a 10% SDS-PAGE gel as described above.
- 81 - 2.4.4. Factor Xa digestion of MBP fusion link proteins
MBP-fusion proteins (25 μg/25 μl) were digested with 0.5 μg of factor Xa for 48 h
at 37 ºC in a total of 25.5 μl to 27 μl of column buffer. Fusion proteins (5 μg) were
separated on a 15% or 18% SDS-PAGE gel and either stained with CBB or
electrophoretically transferred to Immobilon-P polyvinylidene difluoride (PVDF)
membranes in 25 mM Tris, 192 mM glycine, 20% methanol at 70 V for 1 h to be
analyzed by Western blot with 8-A-4 anti-LP antibody (1/100) as described below.
CBB-stained gels were destained in 50% methanol, 10% acetic acid, and 40% water.
2.4.5 SDS-PAGE and Western blot analysis
Western blot analysis was performed as follows. Samples were separated on an
SDS-PAGE gel and electrophoretically transferred to a PVDF membrane by using the
Biorad mini gel system. For colorimetric Western blot analysis, the PVDF membrane was
blocked in 5% non-fat dry milk dissolved in Western B1 buffer (10 mM Tris-HCl (pH
8.0), 100 mM NaCl, 0.05% Tween-20), rinsed in B1 buffer, and then incubated with
primary antibody, 8-A-4 anti-LP antibody (1/100) in 5% non-fat dry milk, for 1 h. After
washing 3 times for 5 min each, the membrane was incubated with anti-mouse-IgG-AP
conjugated secondary antibody (1/7500) for 1 h. The membrane was washed for 10 min,
3 times, and then incubated with BCIP (33 μl), NBT (66 μl) in 10 ml of Western B2
buffer (100 mM Tris (pH 9.5), 100 mM NaCl, and 5 mM MgClBB2B)B to visualize the bands.
The membrane was incubated until bands of the desired intensity were obtained.
For chemiluminescent Western blot analysis, the PVDF membrane was blocked in
- 82 - 5% non-fat milk then incubated with primary antibody, 8-A-4 anti-LP (1/100) or
anti-MBP (1/10000), for 1 h. After washing 3 times for 5 min each, the membrane was
incubated with anti-mouse-IgG-HRP or anti-rabbit-IgG-HRP conjugated secondary
antibody (1/10000) for 1 h. The membrane was washed for 15 min, 4 times, immersed in
the ECL Plus solution for 5 min, and then exposed on a Hyperfilm. The exposed film was
developed with a Kodak X-OMAT film developer.
2.4.6 Enterokinase digestion of MBP fusion link protein
Pooled fractions from the amylose resin elutions were packed into a dialysis bag and
concentrated by permitting solvent evaporation in air at room temperature for overnight
in the presence of 1 mM phenyl methyl sulfonyl fluoride (PMSF). The dialysis bag was
then placed into 100 volumes of E-reaction buffer (50 mM Tris-HCl (pH 8.0), 1 mM
CaClB2B B,B and 0.1% Tween-20 (v/v)) for 2 h dialysis and then for overnight in the same
volume of fresh buffer. The concentrations of the fusion proteins were determined by the
Bradford BAC protein assay. The final concentration was approximately 0.4 mg/ml.
MBP-fusion proteins (40 μg/100 μl) were digested with 1, 5, 10, and 20 U of
enterokinase for 20 h at room temperature in a total of 100 μl to105 μl of E-reaction
buffer. Fusion proteins (3 μg) were separated on a 15% SDS-PAGE gel, electrophoretically transferred to a PVDF membrane, and analyzed by Western blot using the 8-A-4 anti-LP antibody (1/100) as described above.
- 83 - 2.4.7 Sephacryl S-300 chromatography and Refolding of MBP/bPTR1+2
The MBP/bPTR1+2 expressed in E. coli was either undigested or digested with factor Xa in column buffer and applied to a Sephacryl S-300 column (1.0 cm x 108 cm), under reducing-denaturing conditions. At first, the column was calibrated with blue
dextran (MW>2 MDa; VBB0B),B phenol red (MW=354.4; VBtB B),B ovalbumin (43 kDa), chymotrypsinogen A (25 kDa), and ribonuclease A (13.7 kDa) in 20 mM Tris-HCl (pH
8.3), 6 M GnHCl. Then each sample was analyzed as follows. Briefly, 1 ml of
MBP/bPTR1+2 (0.87 mg/ml) was either digested or undigested with 20 μl of factor Xa
(1.0 mg/ml) at 37 ºC for 48 h. Guanidine hydrochloride (GnHCl) (0.57 g) and DTT (3 mg) were then added to the samples, which were bubbled with nitrogen gas for 1 min, sealed and incubated at 4 ºC for 24 h. The sample marked with 0.1% phenol red was loaded onto Sephacryl S-300 chromatography, equilibrated, and eluted at 4 ºC with a buffer bubbled with nitrogen gas (20 mM Tris-HCl (pH 8.3), 6 M GnHCl, 2 mM DTT).
Each fraction (55 drops: 1.5 ml) was collected and analyzed for its protein and link protein content by UV absorbance at 280 nm and immunoblot assay (8-A-4), respectively.
To refold the digested PTR1+2 in each fraction, individual fractions of Nos. 29 to 39
(corresponding to elution volumes of 41.8 ml to 58.9 ml) were dialyzed for the indicated times against 500 ml of column buffer containing the progressively decreasing concentrations of GnHCl as follows: 3 M GnHCl overnight; 1.5 GnHCl M GnHCl (1:4)
1.25 h; 0.75 M GnHCl (1:8) 1.25 h; 0.375 M GnHCl (1:16) 1.25 h; no GnHCl 1.25 h; and no GnHCl overnight). Finally the dialyzed fractions were separated on 18% SDS-PAGE
- 84 - gels, either stained with CBB or transferred to PVDF membranes and immunostained with 8-A-4 anti-LP as described above.
2.4.8 Zinc binding analysis of MBP/rhLP fusion proteins expressed in E. coli
High-flow rate chelating resin was washed twice with 2 volumes of water and then
incubated with 1 volume of 1 M ZnClB2B B B overnight at 4 ºC to prepare zinc charged chelating Sepharose. A zinc-binding assay for undigested MBP/bPTR1+2 was performed as follows. Zinc-charged chelating Sepharose was packed into a column and washed with
5 column volumes of water followed by 2 column volumes of zinc-binding buffer (ZBB);
0.15 M sodium acetate (pH 7.9), 0.2 M NaCl. Then, MBP/bPTR1+2 (0.87 mg) diluted in
8 ml total of ZBB was applied to the zinc column and washed with 60 ml of ZBB (pH
7.9) followed by 50 ml of ZBB (pH 3.5). The samples were chromatographed at 4 ºC and collected as 1.5 ml fractions. The protein content in the elution was determined by UV absorbance at 280 nm. The peak fractions (20 μl) were indicated with asterisks in Fig.
2-10 and were analyzed on SDS-PAGE/Western blot with anti-LP (8-A-4) and anti-MBP antibodies as described above.
Zinc binding assays for the undigested full-length and truncated recombinant human link protein fused with MBP were performed as follows. Zinc-charged chelate Sepharose
(1 ml of a 50% slurry) was packed into a 1 ml syringe and washed with 2ml of water followed by 5 ml of ZBB (pH 7.9). Each MBP fusion link protein sample (60 μg/500 μl) was dialyzed against ZBB (pH 7.9) overnight at 4 ºC and applied to the zinc-charged chelate Sepharose. The column was washed with 10 ml of ZBB (pH 7.9), and then the
- 85 - bound protein was eluted with 10 ml of ZBB (pH 3.5) followed by 5 ml of 0.05 M
EDTA/ZBB (pH 7.9).
2.4.9 Enzyme Linked Immunosorbent Assay (ELISA)
Enzyme Linked Immunosorbent Assay was performed with anti-MBP antibody to
monitor the MBP-fusion protein elution from zinc-charged chelate Sepharose. From each
0.5 ml fraction collected, 200 μl was plated into each well of a 96-well ELISA
high-binding plate and incubated overnight at 4 ºC. Samples were discarded, and each
well was blocked with 1% non-fat dry milk/TBS (50 mM Tris-HCl (pH 8.0), 200 mM
NaCl) for 2 h at room temperature or overnight at 4 ºC. Blocking buffer was discarded,
and each well was incubated with 100 μl of the anti-MBP polyclonal antibody (the
primary antibody) diluted (1/10,000) in TBS for 1 h at room temperature. The primary
antibody was discarded and each well was washed with 3 x 200 μl of TBS. The
anti-rabbit-IgG-AP antibody (the secondary antibody) (1/7500) (100 μl) was added to
each well and incubated for 1 h at room temperature. The secondary antibody was
discarded and each well was washed with 3 x 200 μl of TBS. Substrate (pNNP;
SigmaFast tablet) mix (75 μl) was added to each well and incubated for 1 h at 37 ºC. The
reaction was quenched by adding 25 μl of 3 M NaOH solution and the absorbance was
read at 405 nm.
2.4.10 Construction of the bovine PTR1+2 into pBAD/Thio-TOPO vector
A cDNA clone encoding the PTR1+2 domains of bovine link protein was amplified
- 86 - from cDNA generated from bovine articular chondrocyte d 0 culture by PCR with the
following primer set 5’-GTAGTATTCCCTTATTTTCCA-3’ (upper),
5’-GATGATGTAGCCTAAACAGTT-3’ (lower) and TA cloned into the
pBAD/Thio-TOPO bacterial expression vector. Transformants were screened and a
plasmid with an insert in the sense direction was picked and named pBLP201-7. A
purified plasmid pBLP201-7 was also used to transform BL-21 codon-PLUS-RIL
competent cells, which was called pBLP201-67 (later renamed pBLP201-7B).
2.4.11 Pilot expression of thioredoxin fusion PTR1+2 domains in E. coli
His-patch thioredoxin fusion PTR1+2 was expressed and purified as described in the
manufacturer’s protocol. Briefly, a glycerol stock of pBLP201-7B was streaked and
cultured on an LB plate (1% tryptone, 0.5% yeast extract, 1% NaCl, 2% agarose, and 50
mM ampicillin) overnight at 37 ºC. A single colony was picked and inoculated to 2 ml of
LB amp (1% tryptone, 0.5% yeast extract, 1% NaCl, and 50 mM ampicillin). The culture
was grown with shaking overnight at 37 ºC, and then 10 ml of LP amp was inoculated
with 100 μl of the overnight culture and incubated for 2 to 3 h until OD600 reached 0.400.
Protein expression was induced by adding arabinose (0.4% final concentration) to the
culture, which was then incubated with shaking for 4 h at 37 ºC. The bacterial culture (1 ml) was microcentrifuged at 5,000 rpm for 5 min. The supernatant was discarded and the cell pellets were resuspended and boiled for 5 min in 100 μl of 1 X SDS-PAGE sample buffer to be analyzed by SDS-PAGE/Western blot. To prepare cell lysis under native conditions, the cell pellet from a 5-ml culture was resuspended in 250 μl of lysis buffer
- 87 - (50 mM potassium phosphate (pH 7.8), 400 mM NaCl, 100 mM KCl, 10% glycerol,
0.5% Triton-X-100, and 0.5% imidazole). Samples were placed on ice and sonicated for
10 sec (Fisher Model 60 Sonic Dismembrator) followed by microcentrifugation at the
maximum speed. The soluble fraction (supernatant) was separated and analyzed by
SDS-PAGE/Western blotting (8-A-4) as described above.
2.4.12 Construction of the bovine PTR1+2 into pPICZα vector
A yeast Pichia pastoris expression vector encoding the PTR1+2 domains of bovine
link protein was prepared as follows. A DNA fragment that encodes a polyhistidine tag at
the C-terminus was excised from pBLP204-2 (see #2.4.2) by EcoRI-XbaI digestion. The
digested fragment was purified with a DNA gel extraction kit and ligated into the
EcoRI-XbaI-digested pPICZαA vector. Three clones containing the PTR1+2 domains
were confirmed by DNA sequencing and named pBLP202-48, pBLP202-51, and
pBLP202-52. The expression construct pBLP202-51 was used for the generation of stable
lines.
The plasmids pBLP202-51 and pPCIZαA without an insert (negative control) were
linearized with SacI. Two Pichia pastoris strains (GS115, KM71) were transformed with the linearized construct by electroporation with the Biorad GenePulser as described in the
manufacturer’s protocol. Colonies were grown in YPDS (1% yeast extract, 2% peptone,
2% dextrose, 1 M sorbitol, 2% agar) plates with 100 μg/ml Zeocin for 4 d at 30 ºC. Since
colonies grew too close together, a few colonies were re-streaked onto new YPDS plates
with 100, 500, 1000, or 2000 μg/ml of Zeocin and grown for 3 d at 30 ºC. Three colonies
- 88 - were picked from each plate and grown in 2 ml of YPD (1% yeast extract, 2% peptone,
2% dextrose) (+100 μg/ml Zeocin) for overnight at 30 ºC. At this point, glycerol stocks
+
P were produced. In addition, clones in GS115 were confirmed for the MutPP P phenotype as described in the manufacturer’s protocol by their growth on MMH (1.34% YNB, 4 x
-5 -3 10PP P% biotin, 0.5% methanol, and 4.0 x 10PP P% histidine, and 1.5% agar) and MDH
-5 -3
P (1.34% YNB, 4 x 10P P% biotin, 2% dextrose, and 4.0 x 10PP P% histidine, and 1.5% agar) plates for 2 d at 30 ºC. The positive clones were screened by direct PCR. Positive clones in GS115 named pBLP205-21, pBLP205-22, and pBLP205-24 were used for further analysis. Clones in strain KM71 were named pBLP206-1, pBLP206-5, pBLP206-6, and pBLP206-11 and were used for further analysis.
2.4.13 Pilot expression of PTR1+2 domains in Pichia pastoris
A single colony from each streaked plate was inoculated in 25 ml of BMGY (1% yeast extract, 2% peptone, 100 mM potassium phosphate (pH 6.0), 1.34% YNB, and 4 x
-5 10PP P% biotin, and 1% glycerol) and cultured in a 250 ml baffled flask at 30 ºC with shaking overnight until the OD600 reached 2.0. Cells were harvested by centrifuging at
3000 x g for 5 min at room temperature. The supernatant was discarded and the cell pellet was resuspended to an OD600 of 1.0 in BMMY (1% yeast extract, 2% peptone, 100 mM
-5 potassium phosphate (pH 6.0), 1.34% YNB, and 4 x 10PP P% biotin, and 0.5% methanol) to induce the expression of recombinant PTR1+2 domains. The culture was placed in a 1 l baffled flask, and the flask was covered with 2 layers of sterile cheesecloth and incubated at 30 ºC for 96 h (GS115) or 144 h (KM71). Every 24 h, 100% methanol was added to a
- 89 - final concentration of 0.5%. Culture (1ml) was collected and microcentrifuged for 2 min to separate supernatant and cell pellets for analysis of protein expression. The cell pellet from 1 ml of culture was resuspended in 100 μl of breaking buffer and vortexed for 30 sec, followed by incubation on ice for 30 sec. A cycle of vortexing and incubation on ice was repeated for additional 7 times and was followed by microcentrifugation at the maximum speed for 10 min at 4 ºC. The supernatant (7 μl (KM71) or 17 μl (GS115)) was analyzed on a 16.5% Tris-Tricine CBB-stained SDS-PAGE gel. For secreted expression, culture supernatant (7 μl (KM71) or 17μl (GS115)) was analyzed on a 16.5% Tris-Tricine gel.
2.4.14 Large-scale expression and purification of PTR1+2 domains in Pichia pastoris
A single colony from a plate streaked with pBLP205-24 was inoculated in 25 ml of
BMGY and cultured in a 250 ml baffled flask at 30 ºC with shaking overnight until the
OD600 reached 2.0. The entire overnight culture was inoculated into 1 l of BMGY and
grown in a 3-l baffled flask at 30 ºC with vigorous shaking until the culture OD600
reached 2.3. Cells were harvested by centrifuging at 3000 x g for 5 min at room
temperature. The supernatant was discarded, and the cell pellet was resuspended to an
OD600 of 2.0 in BMMY to induce the expression of recombinant PTR1+2 domains.
One liter of culture was placed in a 3-l baffled flask, and the flask was covered with 2
layers of sterile cheesecloth and incubated at 30 ºC for 120 h. Every 24 h, 100% methanol
was added to a final concentration of 0.5%. After 120 h of incubation, cells were
harvested by centrifugation at 3,000 x g for 5 min at room temperature and resuspended
- 90 - and washed in 50 ml of breaking buffer containing 1 mM PMSF. Cells were then centrifuged at 3,000 x g for 5 min. Cell pellets were resuspended in 50 ml of breaking buffer with 1 mM PMSF and 40 ml of acid-washed beads were added. Cell pellets were lysed as described above with 8 cycles of vortexing and incubation on ice. Cell lysates were centrifuged at 3,000 x g for 20 min. Supernatants were collected and further centrifuged at 11,000 x g for 20 min at 4 ºC with an SS-34 rotor. Supernatants were
2+ TM 2+
P diluted with 350 ml of native NiPP P-binding buffer, applied to a ProBondPP P (NiPP P charged chelating Sepharose) column, and sequentially washed with 500 ml of native binding buffer (20 mM sodium phosphate (pH 7.8), 50 mM NaCl), 200 ml of washing buffer (20 mM sodium phosphate (pH 6.0), 50 mM NaCl), and 300 ml of washing buffer (20 mM sodium phosphate (pH 5.5), 50 mM NaCl). Finally the protein was eluted with 150 ml of elution buffer (20 mM sodium phosphate (pH 3.0), 50 mM NaCl).
2.4.15 N-terminal amino acid sequencing
N-terminal amino acid sequencing analyses were performed on an Applied
Biosystems Procise 494 protein sequencer according to the manufacturer’s protocol using pulsed-liquid cycles. Protein was isolated by separation on an SDS-PAGE gel either under reducing (Fig. 2-4) or non-reducing (Fig. 2-17) conditions. Proteins were then electrophoretically transferred to a PVDF membrane (Immobilon-PQ) either in 10 mM
CAPS (pH 10.5), 3 mM DTT, and 15% methanol (Fig. 2-4) or in 10 mM CAPS (pH 11)
(Fig. 2-17). The membrane was stained with CBB and bands (indicated in the figure legends) were excised and subjected to N-terminal sequencing. For the N-terminal
- 91 - sequence described in Table 2-IV, the obtained sequence was then subjected to a
protein-protein BLAST search using the NCBI web site
(HTH http://www.ncbi.nlm.nih.gov/BLAST/T THT ).
Acknowledgements
I would like to thank Dr. Judith Varelas for the construction of pBLP68-88, Mr. John
Kollar for the construction of pBLP153-5, preparation of bovine articular chondrocyte cDNA and technical advice, Mr. Tru D. Huynh for technical advice, Mr. Patrick Klepcyk for DNA sequencing, and Ms. Cheryl L. Owens and Mr. Jason Rarick for amino acid sequencing.
- 92 - Chapter 3
Biochemical and functional characterization of full-length recombinant aggrecan
and cartilage link protein expressed in mammalian cells
Summary
Full-length recombinant aggrecan and link protein were expressed in mammalian
cells and have been biochemically and functionally characterized for use in functional studies. Previously, these studies have been done using tissue-isolated proteoglycans or
truncated recombinant molecules. One reason for this is that these molecules contain
many cysteine residues and are difficult to express as correctly folded proteins. As
described in Chapter 2, we have expressed link protein, which shares sequence homology
to the G1 domain of aggrecan, in E. coli. It is prone to aggregate, however, due to the formation of inappropriate inter- and intra-molecular disulfide bonds. Aggrecan is extensively modified with glycosaminoglycans; therefore, expression systems used to express recombinant aggrecan should be capable of adding glycosaminoglycans. For this reason, mammalian expression systems will be most suitable for obtaining properly glycosylated molecules.
Bovine recombinant aggrecan was expressed in various mammalian cell lines,
which include COS-7, CHO-K1, HeLa, human immortalized chondrocytes (T/C/-28a2),
and CHO (pgsA-745) mutant xylosyltransferase-deficient cells, as a full-length
proteoglycan having both N-terminal G1 and C-terminal G3 domains. Importantly,
aggrecan was expressed and secreted from a CHO mutant cell line deficient in producing
chondroitin sulfate (CS), suggesting that CS modification is not required for aggrecan
- 93 - secretion. A comparison of the hydrodynamic sizes of COS-7 cell-expressed aggrecan with aggrecan isolated from either cartilage or bovine chondrocytes revealed that monomeric aggrecans derived from cartilage and chondrocyte cultures were larger than that of recombinant aggrecan. In contrast, the average molecular mass of sulfated glycosaminoglycans (sGAGs), comprising predominantly CS, on recombinant aggrecan was larger than that of either cartilage- or chondrocyte-derived aggrecan. These findings suggest that the CS occupies fewer sites on the core protein of COS-7 cell-expressed aggrecan compared to those on cartilage- and chondrocyte-derived aggrecan. Bovine cartilage link protein expressed in COS-7 cells was shown to contain variably substituted
N-linked oligosaccharides at two potential N-linked sites similar to what has been observed previously. Recombinant link protein was able to bind hyaluronan (HA) and zinc, as previously observed for the cartilage-derived link protein. Recombinant aggrecan forms large proteoglycan aggregates with HA that could be further stabilized by recombinant link protein, suggesting that both proteins possess the appropriate functional binding capacities. These results suggest that recombinant aggrecan can be used for studies of molecular interactions with HA and recombinant link protein.
3.1. Introduction
In the previous chapter, the biochemical characterization of full-length and truncated recombinant link protein expressed in E. coli was discussed. As expected, recombinant link protein with multiple cysteine residues is prone to aggregate, since E. coli lacks an endoplasmic reticulum (ER) that contains protein disulfide isomerase (PDI). Protein
- 94 - disulfide isomerase helps to correctly fold cysteine-containing proteins by shuffling
disulfide bonds until those that are thermodynamically favored are formed. We therefore
attempted to express a construct containing both of the proteoglycan tandem repeat
(PTR1+2) domains of link protein in a yeast expression system (Pichia pastoris), which
was expected to yield properly folded protein. In this expression system, however, the
recombinant PTR1+2 domains with a poly-histidine tag (for nickel binding) failed to be
secreted. Scant protein in the cell-lysate was reactive to the anti-link protein antibody, but
could not be purified by nickel chelate affinity chromatography.
We therefore expressed link protein in mammalian COS-7 cells, in an expression
system that was also found to be suitable for the expression of aggrecan. Recombinant
aggrecan constructs were also expressed in HeLa, human immortalized juvenile
chondrocyte (T/C-28a2), CHO-K1, and xylosyltransferase-deficient CHO pgsA-745 cell
lines to characterize the cell-specific glycosylation of aggrecan. The development of this model system is the first step toward the eventual goal of manipulating intermolecular interactions within the proteoglycan aggregate and exploring mechanisms of glycosylation-dependent aggrecan catabolism by proteases active in normal and osteoarthritic cartilage, which will be described in Chapter 4.
3.2. Results
3.2.1 Construction of recombinant aggrecan expression vectors
Full-length bovine aggrecan mammalian expression vectors without a tag
(pBAGG64-5) and with a FLAG tag (pBAGG71-28) at the N-terminus were constructed
- 95 -
Fig. 3-1 Alignment of bovine aggrecan cDNA clones and PCR products used to construct full-sized cDNA expression vector inserts. Strategies for construction of (a) pBAGG64-5 (non-tagged aggrecan), and (b) pBAGG71-28 (FLAG-tagged aggrecan). Gray bars in (b) are “adaptor” sequences ligated into the pFLAG-CMV-1 vector. Designations of plasmids containing cDNA inserts are shown at left. Restriction endonuclease cleavage sites used for ligation of overlapping fragments are indicated at the top and by dashed vertical lines. Narrow bars indicate the full length of each cDNA clone. Thick bars indicate the portion of cDNA incorporated into the final full-length aggrecan insert. Numbers indicate the order of fragment assembly.
- 96 - by successive ligations of cDNA clones and PCR products (Fig. 3-1). The strategies for
constructing pBAGG64-5 are described in Fig. 3-1 (a) and pBAGG71-28 in Fig. 3-1 (b).
Recombinant aggrecan expressed with pBAGG64-5 will be referred to as rbAgg and that
with pBAGG71-28 will be referred to as FLAG-rbAgg in this thesis.
3.2.2 Mammalian expression of recombinant aggrecan
The pBAGG64-5 expression vector was transiently transfected into COS-7 cells.
Expressed recombinant aggrecan (rbAgg) was partially purified from conditioned
medium by Sephadex G-50 size exclusion chromatography, and separated by composite
agarose polyacrylamide gel electrophoresis (CAPAGE), which separates the fully
glycosylated proteoglycans electrophoretically on the basis of their charge to mass ratio
(McDevitt and Muir, 1971; Varelas et al., 1991). Immunoblot reactivity with both the
N-terminal G1- and C-terminal G3-specific antibodies confirmed the expression and
secretion of the full-length aggrecan core protein (Fig. 3-2, lanes 1 and 3). However, the
electrophoretic mobility of the full-length rbAgg produced in COS-7 cells (Fig. 3-2, lanes
1 and 3) was slower than that of steer bovine cartilage aggrecan purified from the
A1A1D1 fraction of a cesium chloride gradient (so called “steer aggrecan,” “A1A1D1,”
or “cartilage-derived aggrecan” in this work) (Fig. 3-2, lanes 2 and 4). Judging from its
barely detectable reactivity to the monoclonal antibody 5-D-4, specific for highly sulfated
KS (Fig. 3-2, lane 5), rbAgg contains much less KS than does steer aggrecan (Fig. 3-2,
lane 6), which may be one of the reasons for its slower mobility on CAPAGE. Note that
the anti-KS (5-D-4) antibody is specific for highly sulfated keratan sulfate, and the
- 97 -
Fig. 3-2 Composite agarose-polyacrylamide gel/Western blot analysis. Recombinant aggrecan (rbAgg; lanes 1, 3, and 5) and cartilage-derived steer aggrecan (A1A1D1; lanes 2, 4, and 6) were detected with antibodies directed toward the G1 domain (αG1-2; lanes 1 and 2) and the G3 domain (Lec7; lanes 3 and 4), and the highly sulfated KS chains (5-D-4; lanes 5 and 6). Note that each lane was probed separately on a different membrane (i.e., blots were not stripped and reprobed). (Data obtained from Tru D. Huynh).
- 98 - minimum size requirement of KS for this antibody is linear pentasulfated hexasaccharides (Mehmet et al., 1986). On CAPAGE, the negative charge on KS would enhance its mobility relative to that of non-KS containing aggrecan.
The pBAGG71-28 expression vector was transiently expressed into COS-7 cells, and the cell lysates and conditioned medium were analyzed on 3.5% SDS-PAGE/Western blot under reducing conditions. Unlike CAPAGE, the SDS-PAGE gel separates molecules more closely according to their molecular mass. The major band in the cell lysates
(approx. 750 kDa called Agg2) is barely detectible in the medium; instead, a larger molecular mass band (approx. 850 kDa called Agg1) is observed. Both of these bands
(Agg1 and Agg2) are reactive to antibodies against both FLAG and the G1 domain at the
N-terminus and the G3 domain at the C-terminus, indicating that these bands are full-length. The Agg2 species is likely to be an incompletely processed biosynthetic intermediate form of aggrecan, and is likely to differ from Agg1 in its posttranslational modifications (e.g., glycosylation, processing of N-linked oligosaccharides, etc.). Several fragments migrate faster than the major band at approximately 750 kDa in cell lysates that are only reactive to anti-FLAG and anti-G1 domain antibodies (Fig. 3-3, lanes 1 and
3, dotted line arrow). These fragments are absent in the medium further suggesting that only the completely translated protein, which contains the G3 domain, is secreted into medium (Fig. 3-3, lanes 2, 4, and 6).
To confirm the presence of CS on recombinant aggrecan, COS-7 cell-expressed
FLAG-rbAgg was digested with chondroitinase ABC to determine its susceptibility (Fig.
3-4). The result shows that the diffuse band was sharpened somewhat after chondroitinase
- 99 -
Fig. 3-3 Secretion of full-sized FLAG-rbAgg expressed in COS-7 cells. Cell lysates and medium of FLAG-rbAgg-expressing COS-7 cells were acetone precipitated and reconstituted in 10 M urea (40 μl) and 5 X SDS-PAGE sample buffer (10 μl), boiled, separated on a 3.5% SDS-PAGE gel, transferred to a PVDF membrane, and immunoblotted with antibodies against FLAG epitope (M2), the G1 (αG1-2), and the G3 domains (Lec7). Bands were visualized with ECL.
- 100 - digestion. This result is consistent with the presence of CS on FLAG-rbAgg since the
removal of CS causes a reduction in microheterogeneity and decreases the molecular
mass of FLAG-rbAgg (Fig. 3-4).
We also expressed the pBAGG71-28 expression vector in four other mammalian cell lines (T/C-28a2, HeLa, CHO-K1, and CHO pgsA-745) to characterize the cell-specific glycosylation of aggrecan. CHO pgsA-745 cells (CHO-745) (Esko et al., 1985) are deficient in xylosyltransferase activity, which is required for CS and heparin sulfate (HS) biosynthesis; therefore, the obtained aggrecan should be free of CS (Fig. 3-5). The rbAgg constructs expressed in both CHO-K1 and CHO-745 cells were secreted into the culture medium at full-length as evidenced by their immunoreactivity to both anti-FLAG and anti-G3 domain antibodies (Fig. 3-5, lanes 2 and 4). The rbAgg expressed in CHO-745 migrated as two sharp bands at approximately 850 kDa and 750 kDa (Agg1 and Agg2)
(similar to what was observed in COS-7 cells, see Fig. 3-3), whereas rbAgg expressed in
CHO-K1 ran as a high molecular mass smear (Agg1) and one sharp band (Agg2), suggesting that CS is absent in FLAG-rbAgg expressed in CHO-745 cells, but present in that of wild-type CHO-K1 cells. Note that in both cell lines (and the cell lines discussed below), the Agg1 band is the dominant Agg species secreted into the media. Importantly,
FLAG-rbAgg expressed in CHO-745 was not abnormally deposited in the cell lysates and was fully secreted into the medium (Fig. 3-5, lanes 1 and 3), suggesting that the CS modification is not required for aggrecan secretion.
- 101 -
Fig. 3-4 Chondroitin sulfate substitution of FLAG-rbAgg expressed in COS-7 cells. FLAG-rbAgg (3.2 pmol) expressed in COS-7 cells was purified from conditioned medium, undigested or digested with chondroitinase ABC (0.08 U) at 37 ºC for 1 h, separated on a 3.5% SDS-PAGE gel, transferred to a PVDF membrane, and immunoblotted with anti-FLAG (M2) and anti-G3 domain (Lec7) antibodies. Bands were visualized with ECL.
- 102 -
Fig. 3-5 Recombinant aggrecan expression in wild-type CHO-K1 and in CS-deficient CHO-745 cells. FLAG-rbAgg was transiently expressed in CHO-K1 (wild-type) (lanes 1 and 2) and CHO-745 (xylosytransferase-deficient) cells (lanes 3 and 4). Cell lysates (15 μl out of 1 ml cell lysates) (lanes 1 and 3) and culture medium (150 μl out of 10 ml medium) (lanes 2 and 4) were acetone precipitated. Pellets were reconstituted in 10 M urea (40 μl) and 5 X SDS-sample buffer (10 μl), incubated at 100 ºC for 10 min, separated on a 3.5% SDS-PAGE gel, electrophoretically transferred to a PVDF membrane, and subsequently immunoblotted with antibodies against (a) FLAG and (b) the G3 domain. Arrows indicate FLAG-rbAgg. Bands with asterisks appear to be a result of non-specific binding, because these bands are also present in the medium from non-transfected cells (data not shown). Bands were visualized with ECL.
- 103 - To further characterize the FLAG-rbAgg construct, FLAG-rbAggs expressed in
T/C-28a2, HeLa, CHO-K1, and CHO-745 cells were isolated from culture supernatants by Sephadex G-50 size exclusion and DEAE Sephacel ion exchange chromatography.
Recombinant aggrecans were digested with chondroitinase ABC and separated on a
3.0% SDS-PAGE gel for Western blot analysis with anti-FLAG and anti-G3 domain antibodies (Fig. 3-6). The intact FLAG-rbAgg expressed in HeLa was not detected on this gel (Fig. 3-6, lanes 1, 3, and 5) and those expressed in T/C-28a2 and CHO-K1 cells showed some degree of microheterogeneity; however, after chondroitinase ABC digestion, it was clearly resolved (Fig. 3-6, lanes 2, 4, and 6). This result suggests that rbAggs are substantially substituted with CS. The lack of FLAG and anti-G3 antibodies staining of some of FLAG-rbAggs prior to chondroitinase ABC digestion may result fromxtensive band broadening due to CS microheterogeneity, failing to enter the gel due to its extremely high molecular mass, or having a low affinity for the
PVDF membrane. In fact, the intact steer aggrecan could not be visualized on these gels unless it was digested with chondroitinase ABC and keratanases (Fig. 3-6, lanes
9-11). On the other hand, both intact and chondroitinase ABC-treated FLAG-rbAgg expressed in the xylosyltransferase-deficient CHO-745 cell line migrated at the same positions confirming the absence of CS (Fig. 3-6, lanes 7 and 8). Note that aggrecan isoforms Agg1 and Agg2 were detected in all of the cell lines tested in this study (data not shown for T/C-28a2).
- 104 -
A1A1D1 T/C-28a2 HeLa CHO-K1 CHO-745 Keratanases - - + Ch’aseABC - + - + - + - + - + + (a) FLAG- (c) G1- （N-terminus) （N-terminus)
Agg1 Agg1 Agg2
250 250
(d) (b) -G3 -G3 （C-terminus) （C-terminus) Agg1 Agg1 Agg2
250 250 kDa kDa
1 2 3 4 5 6 7 8 9 10 11 Fig. 3-6 Chondroitinase ABC susceptibility of FLAG-rbAgg expressed in various cell lines. Purified FLAG-rbAgg (0.8 pmol) isolated from media of transiently transfected T/C-28a2 (lanes 1 and 2), HeLa (lanes 3 and 4), CHO-K1 (lanes 5 and 6), and CHO-745 (lanes 7 and 8) cells were either undigested or digested with chondroitinase ABC, separated on a 3.0% SDS-PAGE gel, transferred to a PVDF membrane, and subsequently immunoblotted with antibodies against (a) anti-FLAG (M2) and (b) the G3 domain (Lec7). Steer aggrecan isolated from cartilage by a cesium chloride gradient (A1A1D1) (0.8 pmol) was either undigested (lane 9) or digested with chondroitinase ABC (lane 10) followed by keratanase, keratanase II, and endo-β-galactosidase (lane 11), separated on a 3.0% SDS-PAGE gel, transferred to a PVDF membrane, and subsequently immunoblotted with antibodies against (c) anti-G1 domain (αG1-2) and (b) the G3 domain (Lec7). Bands were visualized with ECL.
- 105 - 3.2.3 Hydrodynamic sizes of recombinant aggrecan monomer and attached
glycosaminoglycans
We showed that 3.0 or 3.5% SDS-PAGE analysis was useful for separating high
molecular mass molecules; however, the molecular mass of highly charged molecules might not be truly represented. Furthermore, appropriate molecular standards are not readily available to correctly estimate the molecular mass of full-length aggrecan.
Therefore, we compared the hydrodynamic sizes of rbAgg (non-FLAG-tagged) expressed in COS-7 cells, bovine articular chondrocyte (BAC)-derived aggrecan, and cartilage-derived (A1A1D1) steer aggrecan monomers by size exclusion chromatography on Sepharose CL-2B under dissociative conditions in the presence of 4 M GnHCl (Fig.
3-7). The exclusion limit of Sepharose CL-2B is approximately 20 to 40 MDa. The approximate Kav values for rbAgg, BAC, and A1A1D1 were 0.61, 0.30, and 0.50,
respectively (see “Experimental Procedures” for the definition of Kav). These values
indicate that BAC has the largest molecular mass followed by A1A1D1 and rbAgg.
To estimate the sizes of individual sulfated GAG (sGAG) chains on rbAgg
expressed in COS-7 cells, aggrecan samples were exhaustively treated with the
non-specific protease papain, which fully digests aggrecan and releases individual sGAG chains bound to short peptide fragments. Molecular masses of the released sGAGs were determined by gel filtration chromatography on Sepharose CL-6B. The approximate Kav values for rbAgg, BAC, and steer A1A1D1 aggrecans were 0.51, 0.57, and 0.67, respectively (Fig. 3-8, solid lines).
- 106 -
Fig. 3-7 Aggrecan monomer size determination by Sepharose CL-2B size exclusion chromatography. Sepharose CL-2B elution profiles of (a) rbAgg expressed in COS-7 cells, (b) bovine articular chondrocytes (BAC), and (c) A1A1D1 preparations under dissociative conditions. BAC and rbAgg elutions were determined by 35S-labeled sGAG chain content, and A1A1D1 elution was monitored for sGAG content by the Safranin-O assay (OD536). Arrows indicate void (V0) and exclusion (Vt) volumes. (Data obtained from Tru D. Huynh).
- 107 - This analysis indicates that the GAG chains of rbAgg are longer that those of BAC and cartilage-derived A1A1D1 aggrecan. These values correspond to reference fractions of
CS (Wasteson, 1971) having molecular masses of approximately 20,000 (rbAgg), 15,000
(BAC), and 9,000 Da (A1A1D1). To determine the type of GAG chains on each of these aggrecan preparations, we performed sequential chondroitinase ABC/papain digestions.
Following sequential digestion (Fig. 3-8, dotted lines) of both BAC and rbAgg, the peak of 35S-labeled material was shifted towards a lower molecular mass. Uronic acid analysis of A1A1D1 steer aggrecan following chondroitinase ABC/papain digestion showed a similar shift relative to full-sized sGAG chains as determined by Safranin-O analysis (Fig.
3-8 c). Although a small amount of chondroitinase ABC-resistant material was observed in the rbAgg profile, the majority of the sGAGs on BAC aggrecan and rbAgg were susceptible to chondroitinase ABC digestion (Fig. 3-8 a and b). This result suggests that most of the sGAGs on BAC and rbAgg are CS.
3.2.4 Expression of recombinant link protein
Retention of aggrecan in the cartilage ECM requires the formation of large proteoglycan aggregates, in which link protein stabilizes the non-covalent binding of aggrecan monomers to hyaluronan (HA). In the previous chapter, we attempted to express the HA binding PTR1+2 domains from bovine link protein to conduct structural and functional studies. As we described, however, it was difficult to produce functional full-length link protein and truncated mutants in non-mammalian expression systems.
- 108 -
Fig. 3-8 Analysis of chondroitin sulfate substitution and chain length. Sepharose CL-6B elution profiles of papain (pap) (―) digested and sequential chondroitinase ABC and papain (---) digested (a) rbAgg, (b) BAC aggrecan, and (c) A1A1D1 steer BAC aggrecans; and rbAgg elutions of papain with/without chondroitinase ABC digests were monitored for their 35S-labeled sGAG chain content. A1A1D1 steer aggrecan elution of papain digests was monitored for sGAG content by the Safranin-O assay (OD536), and the elution of chondroitinase ABC/papain digests was monitored for uronic acid content by the carbazole assay (OD520). (Data obtained from Tru D. Huynh).
- 109 - Therefore, we expressed full-length recombinant bovine and human link protein (rbLP
and rhLP, respectively) in COS-7 cells (Fig. 3-9, lanes 1-4) to conduct functional studies.
Human link protein was also expressed with a poly-histidine tag at the C-terminus (Fig.
3-9, lanes 5 and 6). SDS-PAGE analysis showed that under reducing conditions, all of the
secreted link proteins appear at the monomer molecular mass as doublets, consistent with differential N-linked glycosylation (Fig. 3-9 a). Under non-reducing conditions, the link protein also ran at the size of the monomer, although some oligomeric species are observed (Fig. 3-9 b). This is consistent with correct folding and disulfide bond formation, in sharp contrast to results obtained using prokaryotic expression. COS-7 cells transfected with pcDNA3 (no insert) produced no exogenous or endogenous link protein (Fig. 3-9 lanes 7 and 8).
Since cartilage link protein contains two N-linked glycosylation sites and doublets
were observed in both rbLP and rhLP (Fig. 3-9), it is likely the doublets (LP1 and LP2)
represent different N-glycan occupancies. Conditioned media containing rbLP and rhLP
were therefore digested under reducing conditions with PNGase F, which removes
N-glycans, and yielded a single band (Fig. 3-10 a, lanes 2 and 4, arrow 3) lower in
molecular mass than the doublets (LP1 and LP2) (Fig. 3-10 a, lanes 1 and 3, arrows 1 and
2). This result indicates that the two forms of both bovine and human recombinant link
protein reflect variable substitution with N-linked oligosaccharides (Fig. 3-10 a), as was
observed previously with link protein isolated from bovine chondrocytes (Hering and
Sandell, 1990).
- 110 -
Fig. 3-9 Expression of bovine and human link protein in COS-7 cells. Media supernatant and cell lysates of COS-7 cells transfected with pBLP255-1 (lanes 1 and 2), pHLP252-6 (lanes 3 and 4), pHLP252-70 (lanes 5 and 6), or pcDNA3 (lanes 7 and 8) were separated either on a 10% SDS-PAGE gel (lanes 1 and 2) or on 10 - 20% gradient gels (lanes 3-8) under (a) reducing or (b) non-reducing conditions. Proteins were transferred to PVDF membranes and immunoblotted with 8-A-4 anti-LP antibody. Bands were visualized with ECL plus.
- 111 - Two potential N-glycosylation sites are predicted in both bovine and human link
proteins (see Fig. 3-10 c). In bovine link protein, it has been reported that the first
N-linked site (Asn21) is variably substituted (Le Gledic et al., 1983; Mort et al., 1985).
When rbLP was digested with trypsin in the presence of rbAgg and HA, both LP1 and
LP2 bands were cleaved to generate a single band (Fig. 3-10 d), which is similar in size
to trypsin-digested cartilage-derived LP3, a form that lacks the N-terminal sequence
including Asn21. Therefore, recombinant link protein also is likely to be always
substituted with oligosaccharide at Asn56, but variably substituted at N21.
For both human and bovine LPs, the apparent molecular mass of LP1 is
approximately 5 kDa larger than that of LP2, and the molecular mass of human LP1 and
LP2 are both slightly larger than those of bovine LP1 and LP2 (by 1 to 2 kDa),
respectively. Significant differences in the ratios of LP1 to LP2 of recombinant human
and bovine link protein were observed under both reducing and non-reducing conditions.
Human LP1 accounts for about 90% of total LP, while bovine LP1 comprised about 65 to
75% of total LP. Human and bovine link proteins are over 96% homologous, with only 13
non-identical amino acid residues among a total of 354 residues (Dudhia and Hardingham,
1990; Hering et al., 1995). Only one non-identical residue is found near the second
N-linked site (Asn56), which is presumably fully glycosylated in both LP1 and LP2. In contrast, 4 non-conserved residues occur near the first N-linked site (Asn21) (Fig. 3-10 c),
including one-non-conserved His18 (bovine) and Ile18 (human). We suggest that the
sequence variations in this region may account for the differences in glycosylation
patterns of the two species, and that Asn21 may be more efficiently glycosylated in
- 112 -
Fig. 3-10 Biochemical characterization of recombinant link protein. (a) Conditioned media containing recombinant bovine (rbLP; lanes 1 and 2) and human (rhLP; lanes 3 and 4) link proteins were undigested (lanes 1 and 3) or digested with PNGase F (lanes 2 and 4), and were separated on a 12% SDS-PAGE gel under reducing conditions, transferred to a PVDF membrane, and immunoblotted with 8-A-4 anti-LP antibody. Arrows 1, 2, and 3 indicate positions of LP1, LP2, and de-N-glycosylated LP, respectively. (b) Conditioned media containing recombinant bovine (lanes 1 and 3) and human (lanes 2 and 4) link protein were separated on a 12% SDS-PAGE gel under non-reducing conditions, transferred to a PVDF membrane, and blotted with either 8-A-4 (lanes 1 and 2) or biotinylated HA (lanes 3 and 4). (c) Amino acid sequence alignment of recombinant bovine and human link protein between –10 to +10 (initiation Met is designated as a residue number 1) of the two potential N-glycosylation sites (Asn21 and Asn56). Asterisks indicate residues that are non-conserved between bovine and human link proteins. The letters shaded in gray indicate nonconservative residues. Gray letters indicate residues that are part of a signal peptide. (d) Trypsin undigested (lane 1) or digested (lane 2) HA-rhLP-rbAgg complex were separated on a 12% SDS-PAGE gel under reducing conditions, transferred to a PVDF membrane, and immunoblotted with 8-A-4 anti-LP antibody. Bands in (a and b) were visualized with ECL, and those in (d) were visualized with ECL plus.
- 113 -
Fig. 3-10 (Continued)
- 114 - human link protein.
To investigate whether the different glycosylation patterns on LP1 and LP2 affect their ability to bind HA, unpurified LP1 and LP2 from media supernatants were separated by SDS-PAGE under non-reducing conditions and HA binding was assessed by probing with biotinylated-HA. As shown in Fig. 3-10 b, bovine link protein appeared to bind HA more efficiently than does human link protein under these conditions; however, for each species, non-reduced LP1 and LP2 showed similar binding affinity to biotinylated-HA.
Note that LPs migrate faster under non-reducing conditions than under reducing conditions because of their globular structure lent by the formation of disulfide bonds
(see Fig. 3-10 a and b).
Since bovine LP1 and LP2 both exhibited similar levels of binding to HA, rbLP was purified by HA-immobilized Sepharose-affinity chromatography without concern that one or the other glycoform would be selectively purified. Interestingly, however, it appeared that the LP2 form was enriched after HA-Sepharose affinity purification of recombinant link protein comprising over 50% of total link protein (compare Fig. 3-10 a lane 1, and d lane 1). This result suggests that LP2 may have stronger affinity to HA.
More rigorous HA-binding analyses are required to confirm this observation. For aggregate formation studies, LP was further purified by removing potentially contaminating HA by dissociative CL-2B gel filtration.
3.2.5 Zinc binding of human recombinant link protein
It has been demonstrated that the human link protein isolated from cartilage is
- 115 - capable of binding zinc and other divalent cations (Rosenberg et al., 1991). In the present
work, zinc binding to recombinant human link protein (rhLP) expressed in COS-7 cells
was also demonstrated (Fig. 3-11) by its binding and subsequent elution from Zn2+- charged chelating Sepharose. About 75% of the rhLP bound to the zinc column at pH 7.8
(Fig. 3-11 a, lanes 2 – 5, b) and remained bound at pH 6.0 (lane 6) and 5.5 (lane 7). At pH
3.5, 40% of the bound protein eluted (Fig. 3-11 a and b, lane 8). The remaining link protein was only eluted with 0.05 M EDTA (Fig. 3-11 a and b, lane 9), which completely strips zinc from the chelating Sepharose.
3.2.6 Recombinant proteoglycan ternary aggregate formation
To determine whether rbAgg and rbLP are capable of forming proteoglycan ternary
aggregates in the presence of HA, we monitored the gel filtration elution profiles of the
recombinant preparations during Sepharose CL-2B chromatography under associative
conditions (0.5 M GnHCl). The elution profiles of individual rbAgg and rbLP are given
in Fig. 3-12. Whereas all (100%) of the rbLP was eluted immediately prior to the bed
volume (Vt) (Fig. 3-12 b), one third (33%) of rbAgg was eluted at the void column
volume (V0), and the remaining two-thirds eluted between V0 and Vt (Fig. 3-12 a). The
high molecular mass V0 material is presumably due to the self-aggregation of aggrecan,
which is commonly observed in the absence of detergent (Sandy and Plaas, 1989). When
both rbLP and rbAgg were added to HA, 98% of the rbLP and 91% of the rbAgg were shifted to a common higher molecular mass species co-eluting in the V0 (Fig. 3-12 d and
e), indicating the specific interaction of the G1 domain of rbAgg and recombinant link
- 116 -
Fig. 3-11 Zinc binding of human link protein expressed in COS-7 cells. (a) Unpurified medium supernatant from COS-7 cells overexpressing human link protein was incubated with zinc-charged chelating Sepharose. A percentage (fraction % in the figure) of each fraction was run on a 10 - 20% SDS-PAGE gradient gel, transferred to a PVDF membrane, and blotted with 8-A-4: flow through (lane 1); four washes with high pH buffer (pH 7.8, lanes 2 - 5); one wash with buffer (pH 6.0, lane 6); and one additional wash with buffer (pH 5.5, lane 7). Bound protein was eluted with buffer (pH 3.5, lane 8) and with 0.05 M EDTA (lane 9). (a) Shows the portion of the total volume that was loaded onto the gel. (b) Shows the portion of link protein eluted in each fraction. Relative density of each band was measured as described in the “Experimental Procedures.” Bands were visualized with ECL plus.
- 117 -
Fig. 3-12 Proteoglycan ternary aggregation analyses by associative Sepharose CL-2B chromatography. (a) rbAgg (0.2 nmol) and (b) rbLP (1.0 nmol) were chromatographed individually on Sepharose CL-2B, or as a mixture of rbAgg (0.2 nmol) and HA (20 μg) without (c) rbLP or with (d, e) rbLP (1.0 nmol) under associative conditions. Column fractions were immunoassayed for rbAgg (a, c, and d) or rbLP (b and e) as described in “Experimental Procedures” by anti-G1 (2194) and anti-LP (8-A-4) antibodies, respectively. The numbers inset in the graph indicate the proportions (%) of rbAgg and rbLP eluted in the void column (V0) as aggregates (a, c-e) or eluted in the elution volume between V0 and bed volume (Vt) (b).
- 118 - protein with HA to form proteoglycan ternary aggregates. On the other hand, in the absence of recombinant link protein, only 65% of total rbAgg aggregated (Fig. 3-12 c), indicating that rbLP indeed significantly enhanced rbAgg complex formation.
A similar result was obtained by using cartilage-derived link protein, which formed aggregates with A1A1D1 cartilage-derived steer aggrecan in the presence of HA (Fig.
3-13). Note that in the aggregation study (Figs. 3-12 d and e and 3-13 c), LP (1.0 nmol) was added at 5 times excess to aggrecan (0.2 nmol) to ensure the aggregation of aggrecan by HA and link protein. Since all LP is eluted in the void in the presence of HA, link protein can bind to HA alone in keeping with the fact that LP was purified based on its ability to bind to HA by affinity chromatography.
- 119 -
Fig. 3-13 Cartilage-derived aggrecan and link protein interact with hyaluronan to form proteoglycan aggregates. Link protein, aggrecan, and HA interactions were determined by Sepharose CL-2B chromatography under associative conditions. (a) Cartilage-derived link protein (LP3) (1.0 nmol), (b) cartilage-derived steer aggrecan (A1A1D1) (0.2 nmol), and (c) a mixture of LP3 and A1A1D1 with HA (20 μg). The contents of aggrecan and link protein in each fraction were determined as described in “Experimental Procedures” by the sGAG assay and 8-A-4 anti-LP immunoblotting, respectively
- 120 - 3.3 Discussion
We have successfully used a mammalian expression system to express and purify full-length recombinant bovine aggrecan (rbAgg) and recombinant link protein that are capable of forming a ternary aggregate with hyaluronan (HA). Recombinant aggrecan expressed in various cell lines was shown to be full-length by its reactivity to antibodies against the G1 and/or FLAG (N-terminus) and G3 (C-terminus) domains (Figs. 3-2, 3-3 and 3-4). Furthermore, no C-terminal truncated recombinant aggrecan was secreted into the medium, supporting a previous report that the G3 domain might be required for aggrecan secretion (Zheng et al., 1998). One of the major challenges of expressing aggrecan in a heterologous system is to assess glycosaminoglycan substitution on the core protein. We showed that FLAG-rbAgg expressed in COS-7, T/C-28a2, HeLa, and
CHO-K1 cells was modified with CS, but apparently to different extents. In contrast,
FLAG-rbAgg expressed in CHO-745 xylosyltransferase deficient cells apparently has no
CS, as expected.
In this work, we used a 3.0 or 3.5% SDS-PAGE gel system to resolve full-length recombinant aggrecan. Although the high molecular mass species with abundant CS chains may not be resolved by this system, the different electrophoretic patterns of the full-length aggrecans observed in each cell line indicate that FLAG-rbAgg is differentially glycosylated in each cell line. Indeed, intact cartilage-derived steer aggrecan (1.5 to 2.5 MDa) and intact recombinant aggrecan expressed in HeLa cells (Fig.
3-6) either did not migrate into this gel, was not efficiently transferred, or did not efficiently bind the membrane, suggesting that this gel system is only useful for
- 121 - separating aggrecans with a molecular mass less than 1 to 1.5 MDa. The other drawback
is a lack of appropriate molecular weight standards for estimating the molecular mass of
large molecules. Furthermore, highly charged and glycosylated proteins may be less
efficiently transferred to a PVDF membrane, thus leading to under representation of these
species on Western blots. This problem could possibly be corrected by using PVDF or
nitrocellulose membranes pre-treated with cationic reagents (Karlsson et al., 2000).
Nevertheless, this system could be useful for screening cell lines that are effective in
adding glycosaminoglycans to large molecular mass recombinant molecules such as
full-length aggrecan.
We have also observed two sizes of full-length FLAG-rbAgg isoforms, which we
called Agg1 (higher molecular mass) and Agg2 (lower molecular mass) in this study.
These two species could not be resolved on 4 - 15% gradient SDS-PAGE gels (see Fig.
4-10, band #1). Since both isoforms react to the antibodies against the N-terminal FLAG
and/or G 1 domain and C-terminal G3 domain, the size difference of these forms may
arise from differences in glycosylation. These two full-length aggrecan isoforms, however, are not due to a difference in the amount of CS, since both Agg1 and Agg2 were also observed in recombinant aggrecan from CHO-745 cells. This was further suggested by the results showing that the chondroitinase ABC digestion of Agg1 does not produce
Agg2. Further analysis is required to identify the factors responsible for the size difference observed in these two forms of recombinant aggrecan.
It has been suggested that the modification of aggrecan by CS is important for
efficient secretion (Kiani et al., 2002; Kiani et al., 2001). We have shown, however, that
- 122 - “CS-free” recombinant aggrecan expressed in xylosyltransferase-deficient CHO-745 cells
is secreted into the medium in a similar fashion to that of CHO-K1 and COS-7 cells,
which produce CS. Furthermore, aggrecan was apparently not abnormally accumulated
intracellularly, as core protein was nearly undetectable in the CHO-745 cell lysates.
These results suggest that CS substitution is not essential for aggrecan secretion in this
heterologous expression system. Finally, the ability to express CS-free recombinant
aggrecan will permit an analysis of interactions between ADAMTS4 and the aggrecan core protein. Conventional enzymatic approaches using chondroitinase ABC for removing CS leave short oligosaccharides (so called “CS stubs”). Therefore, differences in ADAMTS4-aggrecan interactions resulting from removing CS by chondroitinase ABC may be difficult to interpret due to incomplete removal of CS. Furthermore, CS-stubs may not have a uniform structure, since the enzyme susceptibility of each CS chain varies with differences in microstructure.
Since it was difficult to estimate the molecular mass of large, highly charged
molecules substituted with GAGs by SDS-PAGE gel analysis, we further characterized
the recombinant aggrecan by size exclusion chromatography to compare its
hydrodynamic size with that of cartilage-derived steer and chondrocyte-derived
aggrecans. The hydrodynamic size of monomeric recombinant aggrecan expressed in
COS-7 cells is significantly smaller than that of aggrecan isolated from primary
chondrocyte culture and of cartilage-derived aggrecan. Interestingly, the apparent lengths of the sGAG chains on recombinant aggrecan, released after exhaustive papain treatment, appear longer than those from either cartilage- or chondrocyte-derived aggrecan. Taking
- 123 - these results together, it can be concluded that rbAgg may be substituted with longer
GAG chains at fewer sites than either cartilage- or chondrocyte-derived aggrecan. This
may be the result of a cell-specific difference in the type and abundance of the glycosyltransferases involved in the initiation (xylosyltransferase) and elongation (GalI,
GalII, GlcA transferases, and CS synthases, etc.) of CS chains (Silbert and Sugumaran,
2002). To account for our results, COS-7 cells may therefore have less xylosyltransferase
activity than do chondrocytes or have a xylosyltransferase with different
peptide-substrate specificity from that expressed in bovine chondrocytes. It could also be
envisioned that, while COS-7 cells may initiate fewer CS chains than do chondrocytes,
each CS chain could extend more rapidly to produce longer CS chains. The different
SDS-PAGE mobilities observed for FLAG-rbAgg expressed in the different cell lines may also result from differences in these enzyme activities.
Although recombinant aggrecan from COS-7 cells did not react with the 5-D-4
anti-KS antibody, which specifically recognizes highly sulfated KS, it contains a small
but detectable amount of a chondroitinase ABC-resistant sulfated species. This may be
due to the presence of small amounts of CS resistant to chondroitinase ABC digestion,
KS that are not detectable by the anti-KS antibody, and/or aggrecan substituted with other sulfated GAGs, which are typically not present in aggrecan isolated from cartilage such
as heparan sulfate (HS). On the other hand, we observed that that KS is essentially absent on chondrocyte-derived aggrecan, which is evident by the lack of chondroitinase
ABC-resistant [35S]sulfated species (Fig. 3-8 b). Previous work by others, however, has
shown the presence of KS on culture-derived aggrecan isolated from both the media and
- 124 - cell layers of articular chondrocyte cultures (Wong-Palms and Plaas, 1995). It has been
shown that culture conditions can significantly alter the KS production in primary
keratocytes, which normally produce KS (Beales et al., 1999) and that fetal bovine serum
(FBS) and TGF-β down-regulate KS biosynthesis (Funderburgh et al., 2001; Nakazawa et
al., 1998). Under our culture conditions (with added FBS), it is possible that the
chondrocytes may have lost the capability to produce KS, since chondrocytes were
cultured in the presence of FBS.
We also expressed recombinant link protein from transiently transfected COS-7 cells
with the intention of generating fully recombinant proteoglycan aggregates to model the
interaction between link protein-HA, aggrecan-HA, and link protein-aggrecan. The
majority of recombinant link protein expressed in COS-7 cells migrates on SDS-PAGE
gels as an apparent monomer under non-reducing conditions, consistent with proper
folding (e.g., correct disulfide bonds). In addition, recombinant link protein is substituted with N-linked oligosaccharides generating two forms (LP1 and LP2), which apparently represent LP isoforms with different glycosylation patterns (one or two N-linked glycans).
It has been suggested that one N-linked site (Asn56) is always substituted with
oligosaccharide (Le Gledic et al., 1983; Mort et al., 1985). The appearance of a single LP
fragment after trypsin digestion of recombinant aggregates, which removes the variably
glycosylated N-terminal region, is evidence that Asn56 is always substituted with an
oligosaccharide chain.
The HA affinity binding assay with biotinylated-HA indicated that both LP1 and
LP2 exhibited similar binding to HA. We observed, however, that HA-purified
- 125 - recombinant bovine link protein contains more LP2 than LP1 (Fig. 3-10 d, lane 1). This
suggests that differential substitution of the alternatively N-linked glycosylated site
(Asn21) may exert an effect on HA binding. A more rigorous HA binding study with
isolated LP1 and LP2 should be conducted to characterize the difference in their HA
binding affinity, which will be discussed in Chapter 5. We also observed differences in
the LP1 to LP2 ratio of bovine and human link protein constructs expressed in COS-7
cells. Although human and bovine link proteins show 96.5% sequence homology, their
sequences are less conserved around the region of the first N-linked oligosaccharide
substitution site (Asn21), thus offering an explanation for the differences in
oligosaccharide occupancies at this site. Even though we have not investigated additional
functional aspects of the glycosylation in this region, it is worth noting that a synthetic
N-terminal human link protein peptide (DHLSDNYTLDHDRAIH), containing the
N-glycosylation site (Asn21), has been shown to function as a growth factor up-regulating both proteoglycan and collagen biosynthesis (Liu et al., 2000; McKenna et al., 1998). The influence of the N-glycosylation of this peptide, however, has not been investigated. It is possible that glycosylation and sequence variations in the N-terminus may potentially affect the growth factor activity of this link protein-derived peptide.
It has been shown that link protein isolated from cartilage strongly binds to divalent cations including zinc (Rosenberg et al., 1991). We have also observed that recombinant human link protein expressed in COS-7 cells is capable of binding to zinc-charged chelating Sepharose. Recombinant link protein expressed in COS-7 cells is predominantly monomeric under non-reducing conditions with functional HA and
- 126 - aggrecan binding activity and is therefore likely to be folded properly. For this reason, the
demonstration of its zinc binding is likely to be physiologically relevant and not an artifact of misfolding, as we believe may be occurring with the E. coli-expressed proteins.
Although the role of zinc binding to link protein is not presently known, Dr. Hering’s laboratory showed that zinc may be required for proper folding of link protein (Varelas et al., 1997), and it may also serve as a reservoir of Zn2+ in cartilage. Since zinc-binding
sites within the link protein have not been identified, our construct should be useful for
conducting site-directed mutagenesis studies to identify a zinc-binding motif.
Lastly, we showed that recombinant link protein and aggrecan are capable of
forming proteoglycan aggregates as demonstrated by CL-2B gel filtration
chromatography (Fig. 3-12). Although recombinant aggrecan appears to partially
self-associate (as observed by others in the absence of detergent (Sandy and Plaas, 1989)),
we show that aggrecan was largely eluted in the void volume of a CL-2B column in the
presence of HA, and the aggregation was further enhanced by recombinant link protein.
Thus, both recombinant link protein and recombinant aggrecan appear to be properly
folded as evidenced by retention of their functional binding properties. Finally, and most
importantly, recombinant aggrecan also requires link protein for enhanced aggregation
with HA.
In summary, we have shown that functional recombinant aggrecan and link protein
can be expressed in a number of mammalian cell lines, producing aggrecan with cell-type
specific post-translational glycosylation, presumably varying in the amount of CS. A
detailed analysis of GAG substitution on each recombinant aggrecan in conjunction with
- 127 - functional analysis will provide additional information on the functions of GAGs on the
properties of aggrecan and its ability to form aggregates. In addition, recombinant link
protein expressed in COS-7 cells should be useful for ascertaining the functional
significance of link protein in stabilizing the proteoglycan ternary complex.
3.4. Experimental Procedures
3.4.1 Materials
The pED expression vector was a generous gift from Wyeth Research (Cambridge,
MA). All the PCR primers, pCRII, pcDNA3, and pBAD/Thio-TOPO vectors, cell
culture medium, and lipofectamine plus reagents were purchased from Invitrogen
(Carlsbad, CA). TransIT-LT1 was purchased from Mirus (Madison, WI).
pBluescriptIISK(+) vector was purchased from Stratagene (La Jolla, CA). Polyclonal
antibody against the G1 domain of aggrecan (αG1-2) was a generous gift from Dr. John
D. Sandy (Sandy and Verscharen, 2001) (Shriners Hospital for Children, Tampa, FL).
Polyclonal antibody against the G3 domain of aggrecan (Lec7) (Sandy et al., 2000) was a
generous gift from Dr. Kurt Doege (Louisiana State University Health Sciences Center,
Shreveport, LA). Polyclonal anti-G1 antibody (2194) (Sztrolovics et al., 2002) was a generous gift from Dr. John Mort (Shriners Hospital for Children, Montreal, Canada).
Purified monoclonal 5-D-4 antibody against highly sulfated keratan sulfate, chondroitinase ABC (protease-free), and chondroitinase ABC, endo-β-galactosidase,
keratanase, and keratanase II were purchased from Seikagaku America, Inc. (Falmouth,
MA). COS-7, HeLa, CHO-K1, and CHO pgsA-745 (CHO-745) (Esko et al., 1985) cells
- 128 - were from ATCC (Manassas, VA). The immortalized juvenile chondrocyte T/C-28a2 cell
line was a generous gift from Dr. Mary Goldring (Harvard Institutes of Medicine, Boston,
MA). PNGase F was purchased from New England Biolabs (Beverly, MA).
COMPLETETM protease inhibitor mix (+EDTA) was purchased from Roche Applied
Science (Indianapolis, IN). Trypsin and Cellgro culture media were purchased from
Mediatech, Inc. (Herndon, VA). ITS culture supplements were purchased from BD
Biosciences (San Jose, CA). DEAE Sephacel, Sephadex G-50, Sepharose CL-2B and
CL-6B, EAH-Sepharose, and ECL kit were purchased from Amersham Biosciences
(Piscataway, NJ). Centriplus, Microcon, and Centricon Plus-20 were purchased from
Millipore (Bedford, MA). Rooster comb hyaluronan, Sephadex G50, anti-FLAG M2 antibody, pFLAG-CMV-1 vector, and HPR conjugated Streptavidin were purchased from
Sigma-Aldrich (St. Louis, MO). Biotin-LC-hydrazide was purchased from Pierce
(Rockford, IL). Alcian blue sGAG assay kit was purchased from Kamiya Biomedical
Company (Seattle, WA). All other chemicals were purchased from Fisher Scientific
(Pittsburgh, PA) and Sigma-Aldrich (St. Louis, MO). Origins of all the other materials used in this chapter can be found in the “Materials” section of Chapter 2.
3.4.2 Purification of aggrecan and link protein from bovine cartilage and chondrocyte
cultures
Trypsin-derived LP3 forms of link protein were prepared from steer
metacarpophalangeal joint cartilage by using methods as described (Brewton and Mayne,
1992; Heinegard and Axelsson, 1977; Varelas et al., 1997). Bovine cartilage aggrecan
- 129 - (A1A1D1) was prepared from bovine articular cartilage essentially as described by
Rosenberg and co-authors with modifications (Rosenberg et al., 1991). Briefly, articular cartilage from 1 to 2 yr old steers was dissected from metacarpalphalangeal joint surfaces, and was added to ice-cold 4 M GnHCl, 0.15 M sodium acetate (pH 6.3), 50 mM EDTA containing protease inhibitors. The cartilage was extracted at 5 ºC for 48 h, filtered and dialyzed at 5 ºC for 16 h against 20 volumes of 0.15 M sodium acetate (pH 6.3), 5 mM
EDTA containing protease inhibitors. Equilibrium density gradient centrifugation under associative conditions was carried out in 2.5 M CsCl at 5 ºC for 48 h at 40,000 rpm. The gradient was divided into three equal fractions called A1 to A3, according to the convention of Heinegard (Heinegard, 1972). To fraction A1 (bottom 1/3 of the associative gradient), two volumes of 5.5 M GnHCl, 0.15 M sodium acetate (pH 6.3), 5 mM EDTA containing protease inhibitors was added. The solution was stirred for 4 h and dialyzed overnight against 20 volumes of 0.15 M sodium acetate (pH 6.3), 5 mM EDTA containing protease inhibitors. Equilibrium gradient centrifugation under associative conditions was performed in 3.5 M CsCl at 5 ºC for 48 h at 40,000 rpm. Fraction A1A1
(bottom 1/3 of second associative gradient) was subjected to equilibrium gradient centrifugation under dissociative conditions. The gradient was divided into three equal fractions, called D1 to D3, and fraction D1 was subsequently dialyzed against 0.1 M sodium acetate (pH 6.3), 5 mM EDTA, 0.05 M sodium acetate (pH 6.3), 2.5mM EDTA, and three changes of water. The sample was frozen and lyophilized.
Chondrocytes were isolated from metacarpophalangeal joint cartilage of 1 to 2 yr old steers and plated at high density as previously described (Kuettner et al., 1982).
- 130 - Briefly, chondrocyte cultures were maintained in Ham’s F-12 medium supplemented with
10% FBS. For metabolic labeling, chondrocytes were grown in medium in the presence
of [35S]sulfate (10 μCi/ml) from d 2 to 3 of culture. Medium was collected and GnHCl
was added to 4 M final concentration. Bovine articular chondrocyte (BAC) aggrecan secreted into conditioned medium was purified by Sephadex G-50 gel filtration and
DEAE Sephacel ion exchange chromatography (Yanagishita et al., 1987). Briefly, 8 ml of conditioned medium containing 4M GnHCl was applied to a G-50 size-exclusion chromatography column (18 cm x 1.3 cm) equilibrated and eluted with 8 M urea, 0.05 M sodium acetate (pH 7.0), and 0.15 M NaCl. Fractions eluted at the column void were incubated with 0.5 volume of DEAE Sephacel equilibrated with 8 M urea, 0.05 M sodium acetate (pH 7.0), and 0.05M NaCl at room temperature for 2 h with agitation.
Resins were washed progressively with 3 x 1 volume of 8 M urea, 0.05 M sodium acetate
(pH 7.0) containing 0.05 M NaCl and with 3 x 1 volume of 8 M urea, 0.05 M sodium acetate (pH 7.0) containing 0.25 M NaCl. Bound proteins were then eluted with 3 x 0.75 volume of 8 M urea, 0.05 M sodium acetate (pH 7.0), and 1.0 M NaCl. Elutants were then concentrated, and the buffer exchanged for the desired buffer for further analysis with Centricon Plus-20 followed by Centricon (YM-30) or Microcon (YM-30).
Culture-derived aggrecan was quantified relative to a standard curve of dry-weight of cartilage-derived (A1A1D1) steer aggrecan by immunoassay with an antibody directed toward the G1 domain (αG1-2). The molecular mass of A1A1D1 was approximated as
2.5 MDa for the purpose of molar calculations.
- 131 - 3.4.3 Construction of full-size aggrecan expression vectors
An expression vector encoding full-length bovine aggrecan with no-tag
(pBAGG64-5) was generated as is shown in Fig. 3-1 (a). Vectors containing the
following inserts were used for sequential ligations and transformations to produce the
full-length aggrecan (7431 bp) expression plasmid pBAGG64-5 in the pED vector.
Plasmid pBAGG23-14 was a 2116 bp fragment comprising residues 1 to 2116 of the full
aggrecan sequence (GenBank Accession No. BTU76615) (Hering et al., 1997). Plasmid
pBAGG23-13 was a 2511 bp fragment comprising residues 428 to 2938; pBAGG57-10
was a 1017 bp PCR product from 2401 to 3417; pBAGG55-2 contained the sequence from 3163 to 6557; and pBAGG67-1 was a PCR product from 5815 to 7392. All inserts were in pBluescriptIISK(+) except for pBAGG67-1, which was in pCRII. Inserts of pBAGG23-13, pBAGG23-14, and pBAGG55-2 were isolated from a bovine chondrocyte
λgt10 cDNA library (Hering et al., 1995). The vector pBAGG57-10 insert was amplified by RT-PCR from bovine chondrocyte RNA with primers
5'-TGCAGAATTCTTCTGCTTCCGAGGTGTTTCA-3' and
5'-ACTGACTAGTTCATCCAGAAGGAAGTCCACTGATA-3', and the insert of pBAGG67-1 was similarly amplified with primers
5'-TTGGACCAAAGTGGGCTTCAG-3' and
5'-ACTGTCTAGAGGCTAGGCTTCTGGGCTCAGC-3'. The DraIII (of insert)-BamHI
(of insert) fragment of pBAGG23-13 and DraIII (of insert)-BamHI (of vector)
pBAGG23-14 were excised. The DraIII-BamHI fragment from pBAGG23-13 was ligated into the linearized pBAGG23-14 to produce vector pBAGG50-10. The ApaI (of
- 132 - insert)-SpeI (of vector) fragment of pBAGG57-10 and pBAGG55-2 (both sites in insert)
were excised. The ApaI-SpeI fragment from pBAGG57-10 was ligated into the linearized pBAGG55-2 to produce plasmid pBAGG59-10. The HindIII (of vector)-ApaLI (of insert)
fragment was excised from both pBAGG50-10 and pBAGG59-10. The HindIII-ApaL1
fragment from pBAGG50-2 was ligated into the linearized pBAGG59-10 to produce
vector pBAGG62-11. The SpeI (of insert)-XbaI (of vector) fragments of both
pBAGG62-11 and pBAGG67-1 (both sites within insert) were excised. The SpeI-XbaI
fragment from pBAGG67-1 was ligated into the linearized pBAGG62-11 to produce
vector pBAGG63-12. The entire SalI-XbaI insert of pBAGG63-12 was ligated into pED
linearized with SalI and XbaI to produce plasmid pBAGG64-5.
An expression vector encoding the full-length bovine aggrecan with a FLAG-tag at
the N-terminus (pBAGG71-28) was generated as is shown in Fig. 3-1 (b). An "adapter"
fragment was amplified from bovine articular chondrocyte cDNA using upper and lower
primers designed to permit the insertion of a full-length aggrecan cDNA into the
pFLAG-CMV-1 vector. The upper primer sequence
(5'-GATCAAGCTTGTAGAAGTTTCAGAACCTGAC-3') matched aggrecan residues
418-438 (Accession BTU76615), and contained a HindIII site, which places Val-1 of the
mature aggrecan polypeptide in frame, and was separated by one leucine residue from the
vector FLAG sequence (DYKDDDDK). The lower primer sequence
(5'-AAGTCGGGCTTTGCCGTGAGG-3') was complementary to aggrecan residues
1554-1574. To ensure the generation of flanking sticky ends, the resulting 1147 bp
product was ligated into the pCRII vector (pBAGG66-6). The vector pBAGG66-6 was
- 133 - digested with HindIII at the HindIII site of the upper "adapter" primer, and at a HindIII site within the aggrecan sequence (at 1437 to1442). This HindIII-HindIII fragment was ligated into the pFLAG-CMV-1 vector linearized with HindIII to produce the plasmid pBAGG68-5. The plasmid pBAGG68-5 was cut with SacI and XbaI, and was ligated with a SacI-XbaI (of vector) fragment from plasmid pBAGG58-3 (from 492 to 3417 in
BTU76615) to generate the plasmid pBAGG70-10. The plasmid 70-10 was cut at the SrfI site (from 1835 to 1842 of BTU76615) and the XbaI site of the vector, which was blunt-ended with T4 polymerase. This linearized product was ligated with the SrfI (of aggrecan)-NotI (of vector) fragment (containing 1835 to 7392 of BTU76615) of pBAGG63-12 to generate plasmid pBAGG71-28 (11,739 bp). This final product was confirmed by restriction mapping with SpeI and SrfI.
3.4.4 Construction of link protein expression vectors
An expression vector encoding full-length link protein was generated as follows.
The coding sequence for link protein was amplified by PCR and ligated into the pcDNA3 vector as follows with bovine (pBLP153-5) and human (pSP8.1DBS) cDNAs as templates. Full-length link protein sequences of both bovine and human starting from a native Kozak sequence were amplified with an upper primer
5’-TATGGATCCAAGATGAAGAGTCTACTTCTT-3’ with a BamHI site and a lower primer 5’-TTAATCTAGATCAGTTGTATGCTCTGAAGCA- 3’. Bovine and human link proteins share identical sequences in these primer regions. To generate the human link protein with a poly-histidine tag at the C-terminus, a lower primer
- 134 - 5’-TCTAGATTA(ATG)6GTTGTATGCTCTGAAGCAGTA-3’ was used. PCR products of
bovine link protein and human link protein without/with a poly-histidine tag were
TA-cloned into the pBAD/Thio-TOPO vector to produce plasmid pBLP253-23,
pHLP209-6, and pHLP209-7, respectively, which were digested with BamHI to excise
the link protein sequences and were subsequently ligated into BamHI-digested pcDNA3.
Clones containing the bovine and human link protein (without/with poly-histidine tag)
sequences in the sense orientation were confirmed by DNA sequencing and were
designated pBLP255-1, pHLP252-6, and pHLP252-70, respectively. Note that
pBLP153-5 contained a silent mutation of T to C at 678 of the bovine link protein cDNA
(GenBank Accession No. BTU02292).
3.4.5 Cell culture
COS-7 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM)
supplemented with 10% FBS, 1 mM sodium pyruvate, and antimycotic-antibiotics (100
units/ml penicillin, 100 μg/ml streptomycin, and 250 ng/ml amphotericin B). CHO-K1
and CHO-745 cells were maintained in F12K supplemented with 10% FBS, L-glutamine,
and pen/strep (100 units/ml penicillin and 100 μg/ml streptomycin). HeLa cells were
maintained in MEM with Earl’s salt supplemented with 10% FBS, 1mM sodium pyruvate,
and non-essential amino acids. T/C-28a2 cells were maintained in DMEM-Ham’s F-12
medium supplemented with 10% FBS. One day prior to transfection, each cell line was
plated at 65% confluency and grown overnight at 37 ºC in 5% CO2 in the absence of
antibiotic supplements.
- 135 - 3.4.6 Expression and purification of recombinant aggrecan in mammalian cells
COS-7, HeLa, T/C-28a2, CHO-K1, and CHO-745 cells plated on 100 mm plates were transiently transfected with 10 μg (per one 100 mm plate) of pBAGG64-5 or pBAGG71-28 by using lipofectamine plus reagent, except for T/C-28a2 cells, which were transfected with TransIT-LT1 as described in the manufacturer’s protocol. Seventy-two hours after transfection, both medium (10 ml per one plate) and cells were collected and total cell lysates were isolated by RIPA buffer (1x PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% SDS) with 10 μg PMSF/ml RIPA and 30 μl aprotinin (Sigma cat.
# A6279)/ml RIPA using a standard protocol as follows. Briefly, cell layers (100 mm dish) were rinsed with 6 ml of PBS, and the cells were scraped into 0.6 ml of RIPA buffer on ice. Subsequently, the plate was washed once with 0.3 ml of RIPA buffer and combined with the first cell lysate. Ten microliters of PMSF (10 mg/ ml) was added to the lysates, DNA was sheared with a 21 gauge syringe, and the mixture was incubated on ice for 1 h. Cells were microcentrifuged at 10,000 x g for 10 min at 4 ºC. The supernatant was collected as total cell lysates (approximately 1 ml). For purification of recombinant aggrecan, culture media were harvested after 72 h transfection and GnHCl was added to a final concentration of 4 M and purified by Sephadex G-50 and DEAE Sephacel icon exchange chromatography as described above. The yield of rbAgg (non-tagged) and
FLAG-rbAgg was approximately 6 pmol and 20 pmol per one 100 mm plate, respectively.
Metabolically radiolabeled recombinant rbAgg was prepared as follows. COS-7 cells were plated on six-well tissue culture plates and transfected with pBAGG64-5 as
- 136 - above. Twenty-four hour post-transfection, medium was replaced with fresh medium containing 30 μCi/ml [35S]sulfate. After 72 h, the medium was harvested and fresh medium was added. After an additional 72 h, the medium was harvested and samples from the two time points were pooled for analysis.
3.4.7 Composite agarose polyacrylamide gel analysis and chemiluminescent Western blot analysis of aggrecan
Recombinant aggrecan purified by a G-50 size exclusion column from COS-7 cell-conditioned medium was applied to composite agarose polyacrylamide gel
(CAPAGE) as described (Varelas et al., 1991). Proteins separated on a gel were electrophoretically transferred to a PVDF membrane in transfer buffer (25 mM Tris, 192 mM glycine, 15% methanol) overnight at 22 V at 4 ºC. The membrane was blocked with
5% non-fat-milk (in TBST; 20 mM Tris-HCl (pH 7.6), 137 mM NaCl, 0.1% Tween-20) for 1 h at room temperature, and incubated with appropriate primary antibodies overnight at 4 °C as described below. The membrane was then incubated with either anti-rabbit IgG or anti-mouse IgG HRP-conjugated secondary antibody (1/5000) diluted in 5% milk/TBS-T, and bands were detected with ECL. To detect the G1 domain in aggrecan preparations, the membrane was incubated with anti-G1 domain (αG1-2) antibody
(1/500) overnight at 4 ºC. To detect the G3 domain, the membrane was incubated with anti-G3 domain (Lec7) antibody (1/500). To detect highly sulfated keratan sulfate, the membrane was incubated with 5-D-4 anti-KS antibody (1/500).
- 137 - 3.4.8 3.0% and 3.5 % SDS-PAGE gel analysis of full-sized recombinant aggrecan
Conditioned medium (100 or 150 μl) and cell lysates (10 or 15 μl) isolated from each cell line expressing rbAgg were acetone-precipitated with 8 volumes of ice-cold acetone and incubated at –20 ºC overnight. The precipitants were microcentrifuged at
3000 x g for 20 min at 4 ºC. Pellets were reconstituted in 40 μl of 10 M urea and 10 μl of
5 X SDS-Sample buffer (60mM Tris-HCl (pH 6.8), 25% glycerol, 2% SDS, 0.1 % bromophenol red, and 14.4 mM 2-mercaptoethanol) at 100 ºC for 10 min.
Sephadex G-50 and DEAE Sephacel-purified recombinant aggrecan in 20 mM
Tris-HCl (pH 7.2), 150 mM NaCl, and 5 mM CaCl2 was digested in 26 mM Tris (pH 8.0),
12 mM sodium acetate, 105 mM NaCl, and 3.5 mM CaCl2 with chondroitinase ABC
(0.01 U/0.4 pmol of aggrecan) for 1 h at 37 ºC. For digestion of A1A1D1 steer aggrecan,
the chondroitinase ABC-treated sample was further treated with keratanase (0.01 U/0.4
pmol of aggrecan), keratanase II (0.0002 U/0.4 pmol of aggrecan), and
endo-β-galactosidase (0.0002 U/0.4 pmol of aggrecan) for 2 h at 37 ºC. Deglycosylated
samples were then mixed with 1/5 volume of 5 X SDS-sample buffer and incubated at
100 ºC for 10 min. Reduced-denatured proteins were separated by electrophoresis on
either 3.0% or 3.5% SDS-PAGE gels, transferred to PVDF membranes, and subsequently
immunoblotted with anti-FLAG (M2) antibody (1/500), anti-G1 (αG1-2) antibody
(1/500), and anti-G3 (Lec7) antibody (1/3000) by incubation overnight at 4 ºC and
detection with ECL as described previously. Subsequent imunoblotting of a membrane was performed by stripping the membrane in a stripping buffer (62.5 mM Tris-HCl
(pH6.7), 2% SDS, and 100 mM 2-mercaptoethanol) at 50 ºC for 15-30 min. The
- 138 - membrane was then blocked and re-probed with a different antibody. Here, we called this
procedure “cycles of stripping and reprobing.” This procedure is routinely used in the
work described in this thesis except for the experiment described in Fig. 3-2.
3.4.9 Aggrecan monomer size determination by Sepharose CL-2B chromatography
To determine monomer size, aggrecan samples were dissolved in 4 M GnHCl, 0.05
M sodium acetate (pH 6.0), 0.5% CHAPS, applied to Sepharose CL-2B columns (120 cm
x 0.65 cm), and eluted with the same buffer at room temperature. Collected fractions were analyzed by Safranin O assay (Carrino et al., 1991) for cartilage-derived aggrecan or aliquots were counted for [35S] radioactivity for chondrocyte-derived aggrecan or
recombinant aggrecan. The Kav value for each sample was determined by the following
equation: Kav = (Ve-V0)/(Vt-V0).
3.4.10 GAG size determination by Sepharose CL-6B chromatography
To determine the length of GAG chains, [35S]sulfate-radiolabeled bovine articular
chondrocyte-derived aggrecan and recombinant bovine aggrecan were dissolved in a
buffer containing 100 mM Tris-HCl (pH 6.5) and 50 mM sodium acetate, and then digested with chondroitinase ABC (0.01 units/0.4 pmol of aggrecan) at 37 ºC for 1 h.
Following chondroitinase ABC digestion, the samples were further digested with papain
(25 μg/ 400 pmol aggrecan) in 0.1 M sodium acetate, 10 mM EDTA, 10 mM cysteine
HCl (pH 6.1), for 18 h at 60 ºC. Another sample was digested with papain alone. Samples were then chromatographed on Sepharose CL-6B and two/thirds (600 μl) of each fraction
- 139 - was monitored for radioactivity as described above for rbAgg and BAC aggrecan. For papain digestion of A1A1D1 steer aggrecan, GAG elution was monitored by Safranin-O assay. For papain/chondroitinase ABC digestion, GAG (i.e., CS) elution was monitored for uronic acid elution by the carbazole assay (Bitter and Muir, 1962). GAG chain lengths were calculated relative to standard Kav values (Wasteson, 1971).
3.4.11 Expression and purification of recombinant bovine link protein
COS-7 cells plated on 100 mm culture plates were transiently transfected with 15 µg
(per one 100 mm plate) of pBLP255-1, pHLP252-6, or pHLP252-70 by using lipofectamine plus reagent as described above. Twenty-four hours after the transfection, medium was replaced with 10 ml of serum-free DMEM supplemented with 1 mM sodium pyruvate and ITS culture supplements and incubated for additional 72 h. Both medium and cells lysates, which were isolated with RIPA buffer as described above, were separated on SDS-PAGE gels, transferred to PVDF membranes, immunoblotted with
9/30/8-A-4 anti-LP antibody (1/100) for 1 h, followed by incubation with anti-mouse-IgG-HRP (1/5000) and detection with ECL plus. Medium was harvested and
GnHCl was added to a final concentration of 4 M to proceed to link protein purification.
Recombinant bovine link protein (rbLP) was purified by affinity chromatography with HA-immobilized to EAH-Sepharose as described (Tengblad, 1981). Harvested conditioned medium (100 ml) in 4 M GnHCl was incubated with 8 ml of HA-Sepharose in a dialysis bag (Mr<12,000-14,000) and dialyzed against 900 ml of 0.5 M sodium acetate (pH 5.7) at 4 °C for 24 h. Link protein-bound HA-Sepharose was packed into a
- 140 - column and washed with 210 ml of 0.5 M sodium acetate (pH 5.7), followed by 0.5 M sodium acetate buffer (pH 5.7) with increasing concentrations of sodium chloride, 120 ml of 1 M NaCl, 120 ml of 1 M-3 M NaCl (gradient), and 120 ml of 3 M NaCl. Finally, link protein was eluted with 50 ml of 4 M GnHCl and 0.5 M sodium acetate (pH 5.7). Link protein elution was monitored by dot blot analysis with 8-A-4 anti-LP antibody. Elutants were pooled and then concentrated using Centriplus (YM-10) to a final volume of 1 ml by centrifugation. To remove potential HA contaminants, HA-purified-link protein was then subjected to CL-2B chromatography, which was equilibrated and eluted with 4 M
GnHCl, 0.5 M sodium acetate (pH 5.7). The content of link protein in the elution was determined by dot blot imunoblotting analysis with 8-A-4 anti-LP antibody. Fractions containing link protein were pooled and concentrated to obtain the final product, which was used for further analysis. Purified recombinant link protein was quantified relative to a standard curve of dry-weight cartilage-derived trypsin-treated LP3 by immunoassay with 8-A-4 anti-LP antibody. The molecular mass of LP3 was defined as 40 kDa for the purpose of molar calculation. The yield of link protein purified from HA-affinity chromatography followed by CL-2B gel filtration was about 175 pmol per one 100 mm confluent culture dish.
3.4.12 PNGase F digestion of recombinant link protein
Conditioned media (100 μl) harvested from COS-7 cells transfected with recombinant bovine and human link proteins were heat-denatured in the presence of 0.5%
SDS and 1% 2-mercaptoethanol prior to PNGase F digestion. Pre-treated link protein was
- 141 - then digested with PNGase F (500 U PNGase F/100 μl of conditioned medium) for 3 h at
37 °C. Digested samples (27.5 μl) were loaded onto a 12% Tris/HCl SDS-PAGE gel
under reducing conditions, transferred to a PVDF membrane, and immunoblotted with
8-A-4 anti-LP antibody (1/100) as described above.
3.4.13 Biotin labeled HA binding of link protein
Conditioned media (25 μl) harvested from COS-7 cells transfected with recombinant
bovine and human link proteins were separated on a 12% SDS-PAGE gel under
non-reducing conditions, transferred to a PVDF membrane, and either immunoblotted with 8-A-4 anti-LP antibody (1/100) as described above or probed with biotinylated-HA
(bHA) (Melrose et al., 1996), which was prepared by using EZ-link biotin-LC-hydrazide as described in the manufacturer’s protocol. Following incubation with bHA (10 μg/ml), the membrane was incubated with HPR-conjugated streptavidin (2 μg/ml) and blots were detected with ECL.
3.4.14 Trypsin digestion of proteoglycan aggregates
Recombinant link protein (25 pmol), recombinant aggrecan (20 pmol) and 0.5 μg of
HA (from rooster comb) in 20 μl of 100 mM Tris-HCl (pH 7.5) and 500 mM NaCl was dialyzed against 50 ml of 4 M GnHCl, 50 mM sodium acetate for 1 h followed by dialysis against 50 ml of 100 mM Tris-HCl (pH 7.5) and 500 mM NaCl for 1 h at 4 ºC.
The total sample volume was then adjusted to 47 μl by 100 mM Tris-HCl (pH 7.5) and
500 mM NaCl. HA-LP-Agg complex (23.5 μl) was digested with 2.5 μl of trypsin (1 mg/
- 142 - ml in 0.1 M HCl) or mock digested with 2.5 μl of 0.1 M HCl for 18 h at 37 ºC. The
reaction was terminated by adding 6.25 μl of 5 X SDS sample buffer and incubating at
100 ºC for 15 min. Twenty-six microliters of each sample was separated on a 12%
SDS-PAGE gel under reducing conditions, transferred to a PVDF membrane, and
analyzed by chemiluminescent Western blotting with anti-LP antibody as described
below.
3.4.15 Zinc (II) binding analysis of human recombinant link protein
Conditioned medium (1 ml) harvested from COS-7 cells transfected with
recombinant human link protein was incubated with Zn2+-charged chelating Sepharose
(20 μl) (prepared as described in Chapter 2) at 4 ºC overnight. Unbound supernatant was
collected for analysis, and the resin was subsequently washed with 4 x 1 ml of native
binding buffer (20 mM sodium phosphate (pH 7.8), 500 mM NaCl), 1 x 500 μl of washing buffer (20 mM sodium phosphate (pH 6.0), 500 mM NaCl,), and 1 x 500 μl of washing buffer (20 mM sodium phosphate (pH 5.5), 500 mM NaCl). Finally, the bound proteins were eluted with 200 μl of elution buffer (20 mM sodium phosphate (pH 3.5),
500 mM NaCl) followed by chelating with 100 μl of 0.05 M EDTA. Flow-through, washes, and elutions were separated on a 12% SDS-PAGE gel under reducing conditions, transferred to a PVDF membrane, and analyzed by chemiluminescent Western blotting with anti-LP antibody as described above. The film was scanned and the relative intensities of blots were measured with Scion Image software using the “Gelplot2” macros (http://www.scioncorp.com/).
- 143 - 3.4.16 Analysis of proteoglycan aggregate formation
Both recombinant bovine aggrecan (rbAgg) (0.2 nmol) and recombinant link protein
(rbLP) (1.0 nmol) were chromatographed on CL-2B columns either individually or as a mixture with HA (20 μg) by using a method essentially as described by Thornton and co-authors (Thornton et al., 1987). Each sample totaling 390 μl was dissociated in 4 M
GnHCl, 0.1% (w/v) bovine serum albumin (BSA) and allowed to reassociate by dialyzing against 0.5 M GnHCl, 0.5 M sodium acetate (pH 7.4), and 0.1% (w/v) BSA in the presence of COMPLETETM protease inhibitor mix at 4 ºC for 24 h. The same solvent was also used to equilibrate and elute the column. The samples were chromatographed at 4 ºC on a column (112.5 cm x 0.65 cm) at a flow rate of approximately 6.0 ml/h and collected as 1.0 ml fractions. The contents of recombinant link protein and recombinant aggrecan in the elution were determined by dot blot immunoblotting analysis with 8-A-4 anti-LP and (2194) anti-G1 domain antibodies (a gift from Dr. John Mort), respectively. Blots were detected with ECL as described. The film was scanned and the relative intensities of blots were measured with NIH Image software using the “Gel Plotting Macros”
(http://rsb.info.nih.gov/nih-image/Default.html). The maximum value was adjusted to
100, and values relative to the maximum were calculated. The content of A1A1D1 cartilage-derived steer aggrecan in the elution was determined by sGAG assay kit as described in the manufacturer’s protocol. V0 and Vt were determined by the elution of cartilage-derived proteoglycan aggregates and phenol red, respectively.
- 144 - Acknowledgements
I would like to thank Dr. Judith Varelas for preparing A1A1D1 steer aggrecan and
LP3 link protein, Mr. Tru D. Huynh for constructing pBAGG64-5 and pBAGG71-28 and
conducting the experiments shown in Figs. 3-2, 3-7, and 3-8, and Mr. Patrick Klepcyk for performing DNA sequencing.
- 145 - Chapter 4
Characterization of Substrate Specificity of ADAMTS4 against
Aggrecan Core Protein
Summary
In the previous chapter, we described the biochemical characteristics of recombinant
aggrecan expressed in COS-7 cells and other mammalian cell lines. Apparently, each cell
line expressed aggrecan with different structural characteristics presumably due to
cell-type-specific glycosylation. In the present chapter, we describe the use of both
cartilage-derived steer and recombinant aggrecans for the study of ADAMTS4 substrate
specificity in vitro. The main focus of this part of the work is to begin to elucidate the
effects of aggrecan’s glycosylation on its susceptibility to cleavage by ADAMTS4. Since
age-related changes have been observed in aggrecan glycosylation, this study was designed to explore the mechanism of differential susceptibility of aggrecan having
altered glycosylation to degradation by ADAMTS4. Our hypothesis is that changes in
glycosylation, which may occur during development, aging, or disease, may produce
aggrecan with altered ADAMTS4 susceptibility. By using cartilage-derived steer
aggrecan, we show that both chondroitin sulfate (CS) and keratan sulfate (KS) play a role
in regulating the substrate specificity of the p68 form of ADAMTS4. Enzymatic removal
of either CS or KS results in inhibition of cleavage within the CS-2 domain or the IGD,
respectively. Interestingly when both CS and KS are removed, the rate of cleavage within
the IGD is increased to a level close to that of native aggrecan. We also showed by using
recombinant aggrecan expressed from COS-7 and CS-deficient mutant CHO cells that CS
- 146 - is essential for efficient cleavage within the CS-2 domain by the p68 form of ADAMTS4.
On the other hand, recombinant aggrecan expressed in COS-7 cells with negligible KS was cleaved within the IGD. Therefore, the presence of KS is not absolutely required to cleave aggrecan within the IGD.
We also investigated the substrate specificity of the p40 form of ADAMTS4 that lacks the cysteine-rich and spacer domains. It was shown that neither steer nor recombinant aggrecan was cleaved within the CS-2 domain by p40, suggesting that either the cysteine-rich and/or spacer domain is required for cleavage within the CS-2 domain, presumably via binding of the cysteine-rich and/or spacer domains to CS for substrate recognition. On the other hand, p40 effectively cleaved recombinant aggrecan within the
IGD, whereas steer aggrecan was not efficiently cleaved within the IGD by using this enzyme isoform. Differential glycosylation between steer and recombinant aggrecan may explain this difference.
We have constructed a series of mutant aggrecans lacking potentially glycosylated threonine (T352, T355, T357, T370, and T381), serine (S377), and asparagine (N368) residues within the IGD near the ADAMTS4 cleavage site at E373-A374 to investigate the roles of specific potentially glycosylated residues on aggrecan degradation by the p68 form of ADAMTS4. The T357Q, T370Q, and T381Q mutants expressed in KS-deficient
COS-7 cells showed slightly higher rates of cleavage by ADAMTS4-p68 at E373-A374 within the IGD compared with wild-type aggrecan. This result suggests that short
O-linked oligosaccharides likely to be substituted on these residues may inhibit cleavage within the IGD. In contrast, the mutation of S377 to glutamine results in a significant
- 147 - reduction in aggrecan’s susceptibility to p68. The T352Q and T355Q mutants also appear to cleave somewhat slower than the wild type. Interestingly, even though the
T352Q-T355Q-T357Q was cleaved at a rate similar to that of wild-type aggrecan, the
T352V-T355V-T357V mutant (hydrophobic) was cleaved faster than the wild type. This suggests that ADAMTS4-68 may directly interact with this region (TIQTVT) via a hydrophobic interaction, which was further enhanced by replacing three hydrophilic threonines to hydrophobic valines (T352IQTVTÆV352IQVVV). No significant changes are observed in the rate of cleavage at E373-A374 of N368Q, T352V, T355V, T357V, and
T352Q-T355Q-T357Q mutants compared with that of wild-type aggrecan within the given time course. Based on our mutagenesis studies, we suggest that non-KS O-linked oligosaccharides may interfere with cleavage at E373-A374 and that ADAMTS4 may directly interact with the T352IQTVT357 sequence N-terminal to the E373-A374 cleavage site. Furthermore, S377 is essential for efficient cleavage at E373-A374, suggesting that the residues C-terminal to the E373-A374 cleavage site are also important for substrate recognition by ADAMTS4.
- 148 - 4.1. Introduction
Aggrecan is one of the major proteoglycans in cartilage extra cellular matrix (ECM), which significantly contributes to cartilage hydration through its sGAGs (CS and KS, see
Fig. 1-5), which are covalently attached to the aggrecan core protein. Obviously the loss of aggrecan from the cartilage ECM of joints is highly detrimental. In fact, extensive degradation of aggrecan is one of the events found in the early stages of osteoarthritis
(Sandy et al., 1992). In the course of investigating the mechanisms of osteoarthritis, it is now widely accepted that that such degradation can be attributed to certain members of a proteolytic enzyme group in the ADAMTS family (Tortorella et al., 1999). Several studies have also shown that the substrate specificity of aggrecanases (e.g., ADAMTS4) is greatly influenced by aggrecan glycosylation, which can significantly vary with age and among species (Barry et al., 1995; Pratta et al., 2000; Roughley et al., 2003). As described in Chapter 1, it has been suggested that the substrate specificity of ADAMTS4 is regulated by the presence of KS on the aggrecan core protein, which shows an age-dependent increase, as well as by changes in the specific pattern of core protein substitution (Barry et al., 1995). Adult aggrecan, which has a higher content of KS chains compared to aggrecan isolated from younger animals (Barry et al., 1995), is more susceptible to being cleaved at E373-A374 by aggrecanases from interleukin-1(IL-1)-stimulated cartilage explants and by purified ADAMTS4 (Pratta et al.,
2000; Roughley et al., 2003), compared with aggrecan isolated from younger animals and humans that presumably have non-KS oligosaccharides (Roughley et al., 2003). However, many recombinant constructs and rat aggrecan lacking KS also are cleaved between
- 149 - E373-A374, leaving this area controversial (Horber et al., 2000; Lark et al., 1995;
Mercuri et al., 1999). CS also appears to play a role in regulating aggrecanase activity
(Kashiwagi et al., 2004; Sugimoto et al., 1999; Tortorella et al., 2000), as Flannery and co-authors have found that ADAMTS4 has multiple GAG binding sites, which may interact with CS (Flannery et al., 2002). In addition, the several ADAMTS4 isoforms have been shown to have significantly different substrate specificities (Kashiwagi et al.,
2004). This may be due to the different numbers of GAG-binding motifs present in each isoform (see Fig. 1-10). To elucidate the mechanism for the effects of KS, CS, and other oligosaccharides on aggrecanase activity, we have used recombinant aggrecan as well as cartilage-derived steer aggrecan (Pratta et al., 2000; Tortorella et al., 2000; Tortorella et al., 2000) as experimental substrates for in vitro digestion studies. Importantly, by using
recombinant aggrecan, we can manipulate its glycosylation through expression in
different cell lines, mutagenesis of potential glycosylation sites, or co-expression of
glycosyl- and sulfo-transferases involved in GAG biosynthesis. Furthermore,
mutagenesis studies allow for identification of amino-acid residues that may be important
for ADAMTS4 recognition independent of glycosylation. On the basis of this work, we
have modeled the enzyme-substrate interactions involved in the cleavage of aggrecan by
ADAMTS4.
- 150 - 4.2. Results
4.2.1 Aggrecan structure and aggrecan catabolites
Aggrecan catabolism is mediated by a number of proteases in the ADAMTS and
MMP families. Most recent studies have suggested that ADAMTS4 and 5 are the key players in degrading aggrecan, especially in osteoarthritic cartilage (Glasson et al., 2005;
Lark et al., 1995; Little et al., 1999; Stanton et al., 2005; Tortorella et al., 1999). Fig. 4-1
shows the structure of aggrecan and the sites cleaved by ADAMTS4 and 5. In order to
identify the fragments generated by ADAMTS4 and other aggrecanases, a number of
neoepitope antibodies have been developed by others (Table 4-I). Such antibodies
recognize a specific peptide sequence only when it is located at the terminal of a peptide
fragment. The antibody does not recognize intact peptides containing non-terminal
sequences. All neoepitope antibodies and other antibodies used in the current work are
described in Fig. 4-1. The detailed substrate specificity of each antibody is summarized in
Table 4-I.
- 151 -
MMP13 VDIPES341-342FFGV
G1 KS (5-D-4) G3
GELE1480-1481GRGT KEEE1666-1667GLGS ADAMT S 4 & 5 TAQE1771-1772AGEG VSQE1871-1872LGQR
keratanases Chondroitinase ABC
CS stubs (3-B-3)
Fig. 4-1 Sites recognized by specific neoepitope antibodies. Neoepitope antibodies recognize the sequences shown in blue only after cleavage has occurred. Other antibodies recognize the domains and GAGs shown in green. Digestion of KS (orange) and CS (purple) by keratanases and chondroitinase ABC, respectively, leaves short oligosaccharide stubs on the aggrecan core protein. 3-B-3 antibodies recognize these CS stubs having 6-O-sulfate on remaining GalNAc attached to Ser on the core protein after chondroitinase ABC digestion. Each stub contains a hexasaccharide chain (GlcA-GalNAc(S)- GlcA-Gal-Gal-Xyl-). Note that the susceptibility of CS to chondroitinase ABC may vary depending on the degree and site of CS sulfation and buffer conditions (Prabhakar et al., 2005).
- 152 -
Table 4-I
Antibodies used in this study for identifying aggrecan fragments generated by aggrecan
digestion with ADAMTS4 or MMP13
Type Antibody Epitope Poly or mono The terminal 6-sulfated disaccharide attached to CS-linkage region (hexasaccharide CS stubs) generated after chondroitinase ABC digestion (Christner et al., sGAG 3-B-3 1980). Monoclonal Pentasulfated hexasaccharides or longer KS chains 5-D-4 (Mehmet et al., 1986) Monoclonal FLAG M2 FLAG epitope (DYKKDDDDK) Monoclonal G1 αG1-2 G1 domain (Sandy and Verscharen, 2001) Polyclonal 2194 G1 domain (Sztrolovics et al., 2002) Polyclonal G3 Lec-7 G3 domain (Sandy et al., 2000) Polyclonal anti-NITEGE Neoepitope (ADAMTS) NITEGE373 neoepitope (Sztrolovics et al., 1997) Polyclonal BC-3 (ADAMTS) 374ARGSV neoepitope Monoclonal anti-VDIPEN VDIPEN341 (human) or VDIPES341 (bovine) (MMP) neoepitope (Sztrolovics et al., 1997) Polyclonal BC-14 (MMP) 342FFGV neoepitope Monoclonal anti-KEEE/GL GS KEEE1666 and 1667GLGS neoepitopes (ADAMTS) (Sztrolovics et al., 2002) Polyclonal anti-GELE/GR GT GELE1480 and 1481GRGT neoepitopes (ADAMTS) (Sztrolovics et al., 2002) Polyclonal
- 153 - 4.2.2 Characterization of ADAMTS4-p68
As described in Chapter 1, active ADAMTS4 has at least four isoforms (p100, p68, p53, and p40), which vary in their molecular mass and substrate specificity (see Fig.
1-10). The p100 form is only present in the cell, since its prodomain is cleaved in the trans-Golgi by proprotein convertase to generate p68 before secretion (Wang et al., 2004).
The p68 form tightly binds to GAGs (Flannery et al., 2002) and has only been shown to cleave within the CS-2 domain (Gao et al., 2002; Kashiwagi et al., 2004). In these studies, it was shown that for ADAMTS4 to cleave within the IGD, p68 must be further processed to remove the C-terminal “spacer” domain to generate p53 (Gao et al., 2002). The p53 form may be subsequently processed to remove its cysteine-rich domain to generate p40
(Flannery et al., 2002). Both p53 and p40 are able to cleave within the IGD between
E373-A374 (Flannery et al., 2002; Gao et al., 2002; Kashiwagi et al., 2004). This processing can occur either via autoproteolytic degradation (Flannery et al., 2002), or via
MT4-MMP (Gao et al., 2004)-mediated C-terminal processing of ADAMTS4 (see Fig.
1-10). In the current work, we have used purified recombinant human ADAMTS4, which was secreted as its p68 form (Westling et al., 2002). As shown in Fig. 4-2, fresh samples of this enzyme only contain the p68 form. Note, however, that the anti-Y590NHR antibody used does not recognize the p40 form of ADAMTS4. Therefore, the possibility is not excluded that p40 is present in this preparation. Nevertheless, since p40 is typically converted from p53, the complete absence of p53 is likely to indicate the absence of p40 as well. A commercially available p40 form (residues 213-579) of recombinant
ADAMTS4 was also used to compare its substrate specificity with p68.
- 154 -
183 114 81 64 p68
50
37
26 kDa
Fig. 4-2 ADAMTS4-p68 used in this work. ADAMTS4-p68 was incubated in SDS sample buffer at 100 ºC for 10 min immediately after being thawed and separated on a 10% SDS-PAGE gel (150 V, 1 h), transferred to a PVDF membrane (80 V, 1 h), and immunoblotted with anti-Y590NHR antibody (1/500) overnight followed by incubation with anti-rabbit-IgG-HRP (1/5000) for 3 h. A band was visualized with ECL. Note that anti-YNHR recognizes a sequence within the cysteine-rich domain (see Fig. 1-10).
- 155 - 4.2.3 Substrate specificity of p68 and p40
Gao and co-authors have suggested that the p68 isoform only has negligible activity
to cleave at E373-A374, which was demonstrated by digesting rat aggrecan. On the other hand, it was demonstrated with both rat and bovine aggrecan that the p53 and p40
isoforms have strong cleavage activity within the IGD (Flannery et al., 2002; Gao et al.,
2002). Furthermore, Kashiwagi and co-authors have suggested that the p40 isoform can
also cleave within the CS-2 and IGD, although significantly less potently than can the
p53 form (Kashiwagi et al., 2004). In this chapter, we have used both the p68 and p40
isoforms of ADAMTS4, which were readily available, to characterize the susceptibility
of aggrecan to ADAMTS4.
Since ADAMTS4 can rapidly undergo autoproteolysis to generate different isoforms
(Flannery et al., 2002), it is reasonable to assume that even when the enzymes are kept on
ice until used after thawing to minimize autoproteolysis and denaturation, the populations
of ADAMTS4 isoforms and their enzymatic activity in each experiment may vary.
Therefore, each of our time course digestions should be treated separately, and only
experiments performed side by side should be directly compared. The identifiable
fragments obtained by digesting aggrecan with ADAMTS4, which are discussed in the
experiments below, are summarized in Fig. 4-3. For simplicity, a unique number was
assigned for use throughout this work to identify the fragments obtained in each
experiment, and these bear no relationship to fragments designated by numbers in
previous studies by others (Sandy and Verscharen, 2001).
- 156 - MW 1 500 kDa
2 374ARGS 450 kDa
3 KEEE1666 350 kDa
4 GELE1480 350 kDa
5 374ARGS KEEE1666 250 kDa
6 374ARGS GELE1480 250 kDa
7 1481GRG 200 kDa
8 1667GLGS 150 kDa
9 1772AGEG 140 kDa
10 1872LGQE 100 kDa
11 1481GRG VSQE1871 ???? 100 kDa
12 NITEGE373 75 kDa (66.5ｋDa) 12’ NITEGE373 70 kDa (61.5 kDa) 13 65 kDa
14 1481GRG KEEE1666 60 kDa
15 33 kDa
Fig. 4-3 Schematic representation of aggrecan fragments that can be generated by ADAMTS4 digestion of aggrecan core protein. Numbers on the left are used throughout this chapter to identify the fragments. The apparent molecular mass of each fragment determined by its mobility on 4-15% SDS-PAGE gels is shown on the right. Molecular mass is based on sizes of steer aggrecan fragments, except those in parentheses are of recombinant aggrecan. The N-terminal valine of the mature secreted form of bovine aggrecan is designated as residue number 1.
- 157 - In order to understand the substrate specificities of the enzymes used in this work, the A1A1D1 fraction of a cesium chloride gradient preparation of cartilage-derived steer aggrecan was digested with both p68 and p40 (20 ng each) (Fig. 4-4). Note that 20 ng of
ADAMTS4-p68 and ADAMTS4-p40 cleave 40 nM (0.8 pmol/20 μl) of recombinant aggrecan at a similar rate within the IGD (see Fig. 4-15 A and B, lanes 1-4, band #12, p.184). The Western blot analyses with anti-G1- and anti-G3-reactive antibodies suggest that the full-length aggrecan and other large fragments were rapidly degraded to smaller fragments by ADAMTS4-p68 (Fig. 4-4 A, lanes 1-4). The high molecular mass bands
(250 kDa and larger) reactive to the anti-G1 domain that rapidly disappear, and fragments
(bands #12 and 12’) at 65-75 kDa that appear, suggest cleavage within the IGD (Fig. 4-4
A, a).
The disappearance of full-length aggrecan (approx. 500 kDa) reactive to the anti-G3 domain (Fig. 4-4 A b, band #1) antibody, and the appearance of smaller fragments reactive to the anti-G3 domain antibody (Fig. 4-4 A b, bands #9, 10, 13, and 15), indicates cleavage within the CS-2 domain. Therefore, with this experiment, we can show that
ADAMTS4-p68 cleaves steer aggrecan efficiently within both the CS-2 domain and the
IGD (Fig. 4-4 A, lanes 1-4). On the other hand, when aggrecan is cleaved with
ADAMTS4-p40, the high molecular mass bands reactive to both the anti-G1 and anti-G3 antibodies do not disappear (Fig. 4-4 A, lanes 5-8). This result is consistent with the observation made by others showing that p40 has less potent aggrecan cleaving activity than do p68 and p53 (Kashiwagi et al., 2004). Flannery and co-authors show that p40 cleaves aggrecan within the IGD (Flannery et al., 2002); however, our result clearly
- 158 - showed that a large portion of the aggrecan core protein remains intact even after 2 h of
digestion with ADAMTS4-p40, suggesting that p40 poorly cleaves within the IGD and
CS-2 domain of steer aggrecan (Fig. 4-4 A, lanes 5-8). The ADAMTS4 isoforms present at each time course were also analyzed by Western blot, which shows that the p68
form is largely intact over the incubation period of 2 h with a small amount of p53
observed (Fig. 4-4 B, lanes 1-4). This result indicates that the ADAMTS4-p68 used in this study is likely a mixture comprising mainly the p68 with some smaller isoforms.
4.2.4 KS and CS affect substrate specificity of ADAMTS4
It has been reported that the enzymatic removal of sGAGs attached to the aggrecan core protein alters aggrecan’s susceptibility to aggrecanases, suggesting that sGAGs play
a role in regulating the aggrecanase (e.g., ADAMTS4) activity (Kashiwagi et al., 2004;
Pratta et al., 2000; Tortorella et al., 2000). To further characterize such a role of sGAGs,
cartilage-derived steer aggrecan was pre-treated with keratanases (keratanase and
keratanase II), and/or chondroitinase to remove KS and/or CS and then digested with
purified ADAMTS4-p68 to compare their susceptibility to ADAMTS4.
To first confirm the enzymatic removal of CS and KS from steer aggrecan, we used an Alcian blue sGAG assay to measure the total sGAGs content and specific imunoblotting assay for the KS content. Fig. 4-5 A shows that the removal of both CS and
KS was completed (de-CS, de-KS aggrecan), since no sGAGs were detected on the aggrecan core protein by Alcian blue (right-most bar). With the Alcian blue assay, however, the keratanase treated sample, de-KS, did not show a reduction in the level of
- 159 - p68 p40 A Time (h) 0 0.5 1 2 0 0.5 1 2 a 1 1 250 160 105 (a) G1 75 *12 *12 50 12’ 12’ 35 30 25 b 1 250 160 7 105 9 75 *10 * (b) G3 50 13 35 30 15 25 B 1 2 3 4 5 6 7 8 75 p68 ADAMTS4 50 p53 (YNHR) kDa
Fig. 4-4 Substrate specificity of ADAMTS4-p68 and p40. Steer aggrecan (0.8 pmol) was digested either with ADAMTS4-p68 (20 ng) or ADAMTS4-p40 (20 ng) in 20 μl of buffer (20 mM Tris (pH 7.2), 150 mM NaCl, and 5 mM CaCl2) for 0, 30, 60, and 120 min at 37 ºC. Reactions were terminated with 22 mM EDTA. Samples were then deglycosylated with chondroitinase ABC for 1 h followed by keratanase II and endo-β-galactosidase for 2 h at 37 ºC, separated on a 4-15% SDS-PAGE gradient gel (150 V, 1 h), transferred to a PVDF membrane (22 V, O/N), and immunoblotted with anti-G1 (αG1-2) and anti-G3 (Lec-7) and anti-YNHR antibodies by cycles of stripping and reprobing. Anti-G1 and G3 reactive bands were visualized with ECL and anti-YNHR reactive bands were visualized with ECL plus. Bands identified by asterisks are from chondroitinase ABC (equally present in all samples), which cross-reacts with the polyclonal antibodies, therefore providing an internal loading control.
- 160 - sGAGs. Similarly the sGAG content of de-CS appears to be higher than what it would be
expected, since CS comprise about 85% of the total sGAGs on aggrecan core protein. It
has been suggested that the presence of KS can interfere with chondroitinase ABC
digestion, according to the manufacturer’s protocol. Therefore, it is possible that the
removal of CS was not completed when only chondroitinase ABC was used for removing
CS (Fig. 4-5, de-CS). The presence of CS may also interfere with the susceptibility of KS
to keratanases. However, from Fig. 4-5 B, it is clear that all the KS was indeed removed
after the keratanase digest, since the keratanase-digested steer aggrecan samples (de-KS
and de-CS, de-KS aggrecans) are not reactive to anti-KS 5-D-4 antibody, whereas
non-keratanase-treated aggrecan samples (native and de-CS aggrecans) are highly
reactive to this antibody. Therefore, the observed Alcian blue response may also reflect its
different sensitivity to KS and CS.
We then digested these variably deglycosylated aggrecans (de-KS, de-CS, and de-CS & de-KS) with ADAMTS4-p68 and compared their susceptibility to native
aggrecan. At first, the rate of cleavage within the IGD was measured by the appearance of
anti-NITEGE (Fig. 4-6 e-h, bands #12 and 12’), anti-G1 (Fig. 4-6 a-d, bands #12 and 12’),
and anti-ARGSV (Fig. 4-6 q-t, band #6) antibody-reactive fragments. Note that
anti-NITEGE reactive fragments are typically found in doublets (#12 and 12’). The
results show that the keratanase-digested aggrecan (de-KS aggrecan) exhibits a
significant reduction in its susceptibility to ADAMTS4 cleavage between the E373-A374
(Fig. 4-6 a-h, q-t, bands #12, 12’ and 6). Interestingly, when aggrecan was digested with
both chondroitinase and keratanases (de-CS & de-KS aggrecan), cleavage between
- 161 - (a)
(b) 2 μg Doubling dilutions
Native A1A1D1 de-CS A1A1D1
de-KS A1A1D1
de-CS & de-KS A1A1D1
Fig. 4-5 Analysis of de-glycosylated steer aggrecan. The de-glycosylation of steer aggrecan with various enzymes was confirmed by the Alcian blue sGAG assay for total sGAGs and immunoblot assay with anti-KS antibody. Native is non-deglycosylated, de-CS is digested with chondroitinase ABC, de-KS is digested with keratanase and keratanase II, and de-CS & de-KS was digested with chondroitinase ABC and keratanases. (A) Alcian blue sGAG assay was used to assess the amount of sGAG on each aggrecan. Percentage of Alcian blue response (OD620) after the digestion of steer aggrecan with chondroitinase and/or keratanases (sGAG assay kit) relative to total response for native aggrecan is shown. (B) To show the complete removal of KS from keratanase-digested aggrecans (de-KS, and de-CS & de-KS), samples were immobilized on a PVDF membrane and immunoblotted with anti-keratan sulfate (5-D-4) antibody. Serial dilutions of native and de-glycosylated steer aggrecan were immobilized on a PVDF membrane and reacted with the 5-D-4 anti-KS antibody. Steer aggrecan treated with keratanases showed no reactivity to 5-D-4.
- 162 - E373-A374 occurred almost as rapidly as for native aggrecan (Fig, 4-6 a, e & d, h #12
and 12’). On the other hand, it is apparent that the overall cleavage within the CS-2
domain was significantly inhibited in both de-CS and de-CS & de-KS aggrecan compared
with native and de-KS aggrecans based on the patterns of fragments reactive to the
anti-CS-stub antibody (3-B-3) (Fig. 4-6 i-l) and the anti-G3 domain (Fig. 4-6 u-x) antibodies. Note, particularly, bands #9 and #10 representing cleavage at E1771-A1772 and E1871-L1872 within the CS-2 domain are prominent in native and de-KS aggrecans
(Fig. 4-6 (i, j) and (u, v) vs. (k, l) and (w, x)), but only slightly observed in the de-CS and de-CS & de-KS aggrecans. It is also apparent that cleavage at E1480-G1481 within the
CS-2 domain is significantly delayed in both de-CS and de-CS & de-KS aggrecan as assessed by comparing the generation of the anti-ARGSV and anti-TAGELE reactive fragments (Fig. 4-6 m,n vs. o, p, bands #4 and 6). Especially, fragment (#4) was never observed in the de-CS aggrecan (Fig. 4-6 o). These results further suggest that the removal of the CS significantly inhibits cleavage within the CS-2 domain. Nevertheless, it was apparent that even though the cleavage within the CS-2 domain was significantly inhibited compared with that of the native aggrecan, the de-CS & de-KS aggrecan was still cleaved within the CS-2 domain more rapidly than within the IGD, which was apparent from the generation of the G1-TAGELE1480 fragment (#4) (Fig. 4-6 p, band #4).
It is noteworthy that a prominent band (#7) identified as the G1481RGT-G3 fragment
was generated in the chondroitinase-treated samples (Fig. 4-6 k, l, o, p, w, and x). This fragment was generated by cleavage at E1480-G1481, which is barely detectible in the
early time course of native aggrecan (Fig. 4-6, u). Since the C-terminal fragment
- 163 -
Fig. 4-6 ADAMTS4-p68 digestion of native (undigested) and de-glycosylated cartilage-derived steer aggrecan. Undigested steer aggrecan A1A1D1 (native aggrecan) and steer aggrecan digested with keratanases (de-KS aggrecan), chondroitinase ABC (de-CS), and chondroitinase and keratanases (de-CS & de-KS aggrecan) (3.2 pmol each) were digested with ADAMTS4-p68 (8 ng) for 0, 7.5, 15, 30, 60, 120, and 240 min or mock digested without the ADAMTS4-p68 for 240 min at 37 ºC in 10 μl of buffer. Generated products were de-glycosylated with chondroitinase ABC, keratanase II, and keratanase (only if they were not deglycosylated prior to ADAMTS4 digestion), loaded onto 4-15% SDS-PAGE gels, transferred to PVDF membranes, and immunoblotted with antibodies against the chondroitinase-digested G1 domain (a-d), anti-NITEGE neoepitope (e-h), CS stubs (i-l), anti-GELE and GRGT di-neoepitope (m-p), ARGSV neoepitope (r-t), and G3 domain (u-x). Bands identified by asterisks are from chondroitinase ABC (equally present in all samples), which cross-reacts with the polyclonal antibodies, thereby providing an internal loading control. Note that anti-ARGSV antibody is a relatively weak antibody compared to the others (q-t). (Red arrows point to each numbered band).
- 164 - (G1481---E1666) (Fig. 4-6 m and n, band #14) generated by cleavage at E1480-G1481
and at E1666-A1667 appears quickly in native and de-KS aggrecan, it suggests that
cleavage at either E1480-G1481 and/or at E1666-A1667 is the site most preferred within
the CS-2 domain by ADAMTS4-p68, and that down-stream cleavage rapidly occurs after
that. In native and de-KS aggrecans, a fragment #7 is not detected, since it has already
been degraded to smaller fragments in the first 7.5 min of the time course.
To semi-quantitatively compare the rates of cleavage within the IGD at E373-A374
by ADAMTS4-p68 of aggrecan pre-treated with different glycosidases, the intensities of both anti-NITEGE (Fig. 4-6 e-g, band #12, 12’) and anti-G1 antibody-reactive (Fig. 4-6 a-d, band #12, 12’) fragments were measured as described in the “Experimental
Procedures” and were plotted as shown in Fig. 4-7 a and b. Both plots show that the removal of KS alone significantly inhibits cleavage within the IGD (Fig. 4-7, pink); whereas, when CS is also removed with KS (Fig. 4-7, green), the level of cleavage at the
E373-A374 bond returns to a level similar to that of native aggrecan or of de-CS aggrecan (Fig. 4-7, black and purple). Again, these results suggest that KS is only required for cleavage at E373-A374, in the presence of CS (i.e., native aggrecan).
4.2.5 N-linked oligosaccharides inhibits the cleavage within the IGD by ADAMTS4
Poon and co-authors have recently reported studies on the aggrecanase digestion of truncated recombinant aggrecan (a G1-G2 construct) expressed in primary keratocytes, which contains an N-linked KS chain at N368 (Poon et al., 2005). Interestingly, they report that the removal of the N-linked KS by PNGase F under denaturing conditions
- 165 - (a) NITEGE
(b) G1
Fig. 4-7 Semi-quantitation of the effects of KS and CS removal on cleavage within the IGD. The intensities of the (a) anti-NITEGE and (b) anti-G1 domain (NITEGE overlapping) reactive fragments generated by ADAMTS4-p68 digestion shown in Fig. 4-6 were semi-quantified as described in “Experimental Procedures.” The relative density of each band was measured to compare the rates of cleavage at E373-A374 among the different aggrecan preparations. The black line is native aggrecan; pink line, de-KS aggrecan; purple line, de-CS aggrecan; and green line, de-CS & de-KS aggrecan.
- 166 - completely inhibits cleavage within the IGD at E373-A374 (Poon et al., 2005). Prior to
this publication, we had investigated the role of aggrecan N-linked oligosaccharides on
ADAMTS4 activity by digesting A1A1D1 steer aggrecan with PNGase F under
non-denaturing conditions (de-N-linked aggrecan) and compared the changes in its
susceptibility to ADAMTS4-p68 with that of undigested steer aggrecan (native
aggrecan).
To achieve complete removal of N-linked oligosaccharides by PNGase F, proteins
are typically reduced-denatured by incubating samples at 100 ºC for 10 min in the
presence of SDS and 2-mercaptoethanol. However, we felt it was important to avoid
denaturing aggrecan prior to ADAMTS4 digestion, as this would be a more physiological
state. Therefore, in our studies the PNGase F treatment was performed under native
conditions. Fig. 4-8 shows that the rate of cleavage of de-N-linked aggrecan within the
IGD was greater than that of native aggrecan as demonstrated by the appearance of the
NITEGE-fragments (Fig. 4-8 c, d, lanes 2-8). Note that the masses of the anti-NITEGE
reactive fragments (bands #12 and 12’) that are digested with PNGase F are 5 –to 10 kDa smaller than those of native aggrecan suggesting successful removal of N-glycans present in these fragments. Furthermore, in native aggrecan even after a 480-min digestion, the high molecular mass species reactive to the anti-G1 domain was present, even though the accumulation of the anti-NITEGE reactive fragments reached a plateau at the 60-min time point. This result suggests that a certain population of steer aggrecan is highly resistant to cleavage within the IGD, which may be due to differential glycosylation observed in aggrecan isolated from the same source. Control incubations of PNGase F-
- 167 -
Fig. 4-8 ADAMTS-4 digestion of native and de-N-linked aggrecan. A1A1D1 steer aggrecan (3.2 pmol) was (a, b) untreated or (c, d) treated with PNGase F, then digested with ADAMTS4-p68 (16 ng) for 0, 7.5, 15, 30, 60, 120, 240, and 480 min or mock digested without the ADAMTS4 for 480 min at 37 ºC. Generated products were de-glycosylated with chondroitinase ABC, keratanase, and keratanase II, separated on 4-15% SDS-PAGE gels, transferred to PVDF membranes, and immunoblotted with the (a, c) anti-G1 domain and (b, d) anti-NITEGE neoepitope antibodies.
- 168 - treated aggrecan without ADAMTS4-p68 for 480 min demonstrate no aggrecanase-like activity from PNGase F (i.e., no anti-NITEGE reactive bands are observed (Fig. 4-8 d,
lane 9)); however, significant amounts of a G1-reactive fragment that is smaller than the
size of the anti-NITEGE-reactive fragment is observed in control digests of aggrecan (Fig.
4-8 c, lane 9). This fragment (Fig. 4-8, band a) was slightly stained with anti-VDPIPEN
antibody suggesting that the PNGase F-treated A1A1D1 preparation might be
contaminated with MMP-like protease. In fact, the overall pattern of the anti-G1 domain
reactive species is significantly altered in de-N-linked aggrecan (Fig. 4-8 c), suggesting
that non-specific proteolytic degradation of aggrecan takes place in the presence of
PNGase F. We can conclude, however, that the removal of accessible N-linked
oligosaccharide by PNGase F did not result in inhibition of the cleavage at the
E373-A374 peptide bond as recently reported by Poon and co-authors (Poon et al., 2005).
The difference in our results may be due to Poon and co-authors utilizing denaturing
conditions to remove the N-linked oligosaccharides. This may result in greater removal of
N-linked oligosaccharides or a change in the native conformation of aggrecan, both of
which could affect aggrecanase binding and/or peptide bond hydrolysis. Furthermore,
they do not provide data confirming the absence of protease contamination in the
glycosidases they used for their experiments, which may be another confounding factor.
4.2.6 Characterization of FLAG-tagged full-length aggrecan
We used the FLAG tagged full-length bovine aggrecan as discussed in Chapter 3
(FLAG-rbAgg) (Fig. 4-9) to study the substrate specificity of ADAMTS4. This construct
- 169 - has a signal peptide followed by a FLAG-epitope at the N-terminus allowing the secretion of recombinant aggrecan with an N-terminal FLAG peptide. The advantage of using a FLAG-tagged aggrecan is that the FLAG epitope allows identification of the
N-terminal fragments generated by ADAMTS4 digestion. In addition, FLAG-rbAgg can be expressed in cell lines such as T/C-28a2 and primary bovine chondrocytes, which make endogenous aggrecan. We also found that the FLAG-tagged aggrecan plasmid produces a relatively higher yield of protein than non-tagged aggrecan, and the efficiency of site-directed mutagenesis by using the FLAG construct was higher than that for the non-tagged construct.
As described in Chapter 3, we successfully expressed full-length FLAG-rbAgg in the various cell lines summarized in Table 4-II, including xylosyltransferase-deficient
CHO-745 cells, which are deficient in the synthesis of CS and heparan sulfate (HS). Our
(Dr. Hering’s) laboratory has expressed recombinant aggrecan in COS-7 cells, and the majority of the work presented in this thesis was performed with the same cell line. After exploring the use of several other cell lines, we concluded that COS-7 cells provided several distinct advantages. COS-7-derived aggrecan is suitable for general characterization of ADAMTS4 substrate specificity, since the product is less degraded after several purification steps, is produced in higher yield, and shows less heterogeneity, as demonstrated by 3.0% SDS-PAGE gel analysis, than aggrecan expressed in other cell lines. We believe the latter feature will simplify the characterization of fragments generated by ADAMTS4 digestion.
- 170 -
FLAG-bovine aggrecan
FL G1 IGD G KS CS1 CS2 G3
1 2349 N-link KS chains O-link
CS chains
Fig. 4-9 Schematic representation of FLAG-rbAgg. The full-length bovine aggrecan construct was designed to contain a signal peptide for protein secretion followed by an N-terminal single “FLAG” epitope (DYKDDDDK) and full-length bovine aggrecan starting from Val-1. A detailed characterization of this construct is described in Chapter 3. .
- 171 -
Table 4-II
Various cell lines tested in Chapter 3 for transient expression of FLAG-rbAgg Table shows the name of the cell line, animal species, source of tissue, antibody reactivity to recombinant aggrecan, and chondroitinase ABC susceptibility, and notes for special characteristics.
Antibody Ch’ase ABC
Cell line Species Source of tissues FLAG G3 KS CS Notes
African green Kidney epithelial COS-7 monkey SV40-transformed + + - + (Gluzman, 1981) Cervical epithelial HeLa Human adenocarcinoma + + - + Costal chondrocyte Endogenous aggrecan T/C-28a2 Human SV40-transformed + + + + (Kokenyesi et al., 2000) Core 2 GlcNAc transferase deficient
CHO-K1 Chinese hamster Ovary + + - + (Bierhuizen and Fukuda, 1992) Xylosyltransferase deficient CHO-745 Chinese hamster Ovary + + + - (Esko et al., 1985)
- 172 - 4.2.7 Representative digestion of wild-type FLAG-rbAgg expressed in COS-7 cells with
ADAMTS4
FLAG-rbAgg expressed in COS-7 cells was digested with ADAMTS4-p68 and
ADAMTS4-p40 to characterize their substrate specificities (Fig. 4-10). The membrane was stripped and reprobed with different antibodies (Fig. 4-1 and Table 4-I) in order to identify the fragments (Fig. 4-10). The right hand panel of Fig. 4-10 shows the predicted structures of the aggrecan fragments, deduced from their reactivity to specific neoepitope antibodies and from their molecular mass.
The time-course digestion of rbAgg with ADAMTS4-p68 (Fig. 4-10) resulted in the
expected cleavages within the CS-2 domain at E1666-G1667 (Fig. 4-10 d) and
E1480-G1481 (Fig. 4-10 e) demonstrated by their reactivity to anti-KEEE1666/G1667LGS and anti-GELE1480/G1481RGT di-neo-epitope antibodies. Cleavage within the IGD at
E373-A374 (Fig. 4-10 c and f) is also evident by the generation of fragments (#12 and
#6) reactive to the anti-NITEGE and anti-ARGSV neoepitope antibodies. These results show that ADAMTS4-p68 cleaves the full-length FLAG-rbAgg rapidly within the CS-2 domain (bands # 3 and 4) and then subsequently within the IGD (bands # 5 and 6). On the other hand, when FLAG-rbAgg is digested with ADAMTS4-p40 (Fig. 4-10, lanes 5 and 6), cleavage within the IGD is significantly preferred over the cleavage within the
CS-2 domain, which was essentially negligible (i.e., loss of bands #3-5) (Fig. 4-10 A d, e, and g, lane 5). This results in the generation of an ARGSV-G3 (#2) fragment, which is not found in an ADAMTS4-p68 digestion (Fig. 4-10 g, lane 5), since ADAMTS4-p68 normally cleaves the CS-2 domain sooner than the cleavage within the IGD.
- 173 -
Fig. 4-10 ADAMTS4 digestion of FLAG-aggrecan. Wild-type FLAG-rbAgg (0.8 pmol) was digested either with 20 ng of ADAMTS4-p68 (lanes 2-4) or ADAMTS4-p40 (lanes 5) at 37 ºC for the indicated times. Aggrecan fragments were deglycosylated with chondroitinase ABC (protease-free) at 37 ºC for 1 h, then separated on a 4-15% SDS-PAGE gradient gel, transferred to a PVDF membrane, and subsequently immunoblotted with antibodies against various epitopes in FLAG-rbAgg in A: (a) anti-FLAG (M2), (b) anti-G1 domain (αG1-2), (c) anti-NITEGE, (d) anti-KEEE and GLGS, (e) anti-GELE and GRGT, (f) anti-ARGS (BC-3), and (g) anti-G3 domain (Lec7). Lane 6 is a shorter exposure of lane 5 in (g). Predicted structures of each fragment are shown to the right. Bands identified by asterisks are from chondroitinase ABC (equally present in all samples), which cross-reacts with the polyclonal antibodies, therefore providing an internal loading control. B, the same membrane probed with anti-ADAMTS4 YNHR antibody, which recognizes a sequence within a cysteine-rich domain of ADAMTS4. All the blots were visualized with ECL, except that the BC-3 and anti-YNHR blots were visualized with ECL plus. Blue figures and letters indicate the location of the epitope for the antibody used in each blot.
- 174 - Note that band #2 could not be detected by the anti-ARGSV antibody due to its low
sensitivity. To follow the ADAMTS4-p68 conversion to the p53 form, the membrane was
also immunoblotted with anti-ADAMTS4 (p68 and p53) antibody. The results show a
little autoproteolytic conversion of p68 to p53 during the 3-h time point (Fig. 4-10 B),
suggesting that the pattern of digestions shown here may be mediated by both
ADAMTS4-p68 and C-terminal processed isoforms of ADAMTS4.
4.2.8 Anti-NITEGE reactive fragments have intact FLAG epitope and are differentially
glycosylated
In Fig. 4-10 c, two bands (#12 and 12’) reactive to the anti-NITEGE antibody are
observed. To carefully characterize the structure of the two bands, wild-type rbAgg was
digested with ADAMTS4-p68, separated on a 4-15% SDS-PAGE gel and immunoblotted
with anti-FLAG (N-terminus), anti-G1 domain, and anti-NITEGE (C-terminus)
antibodies. Samples were electrophoresed at 100 V (typically run at 150 V) in this
experiment to clearly resolve two bands. Two bands (Fig. 4-11, band #12 and 12’), which
are reactive to anti-NITEGE antibody, are also reactive to anti-FLAG and anti-G1
domain antibodies (Fig. 4-11). This result suggests that the two bands are not generated
by cleavage N-terminal to the E373-A374 cleavage site. Therefore, the two bands are most likely generated because of differential glycosylation in this region. Interestingly, band #12 is about 5 kDa larger than band #12’, which is a typical size for one N-linked oligosaccharide chain. Since there are four potential N-linked glycosylation sites (Asn107,
Asn220, Asn314, and Asn368) and a number of O-linked glycosylation sites within this
- 175 -
FLAG G1 NITEGE
75 kDa 12 12’ 50 kDa 1 2 3
Fig. 4-11 Lack of an ADAMTS4 cleavage site N-terminal to the E373-A374 cleavage site. FLAG-rbAgg (0.8 pmol) was digested with 20 ng of ADAMTS4-p68 at 37 ºC for 3 h. Aggrecan fragments were deglycosylated with chondroitinase ABC (protease-free) at 37 ºC for 1 h, separated on a 4-15% SDS-PAGE gradient gel, transferred to a PVDF membrane, and subsequently immunoblotted with anti-FLAG (M2) (lane 1), anti-G1 domain (aG1-2) (lane 2), and anti-NITEGE (lane 3) antibodies in this order by cycles of stripping and reprobing. Bands were visualized with ECL.
- 176 - fragment, however, the actual differentially glycosylated sites cannot be determined only from these observations. Mutagenesis of potential glycosylation sites should allow
identification of the site that may be differentially glycosylated.
4.2.9 KS is not required for ADAMTS4 cleavage within the IGD of FLAG-rbAgg
Since the FLAG-rbAgg expressed in COS-7 cells was cleavable between
E373-A374 as shown in Fig. 4-10, we wanted to determine whether the anti-NITEGE fragment (FLAG-V1---NITEGE373), which contains the G1 domain and part of the IGD,
was substituted with KS, as it has been suggested that KS potentiates cleavage at
E373-A374 (Poon et al., 2005). The ADAMTS4-digested FLAG-rbAgg was digested
with chondroitinase ABC followed by keratanase II and endo-β-galactosidase to remove
any keratan sulfate susceptible to these enzymes (Fig. 4-12, lanes 1 and 2). For
comparison, the ADAMTS4-digested steer aggrecan (A1A1D1) was also digested with
chondroitinase ABC followed by keratanase II and endo-β-galactosidase (Fig. 4-12, lanes
3 and 4). The anti-NITEGE fragment derived from the steer aggrecan showed a high
degree of heterogeneity and was susceptible to keratanase digestion as demonstrated by a
shift in electrophoretic mobility and increase in the band intensity (Fig. 4-12 lanes 3 and
4). On the other hand, the anti-NITEGE fragments (doublets) derived from ADAMTS4
digestion of FLAG-rbAgg showed neither a shift in their mobility nor a change in band
intensity, suggesting the absence of KS on these fragments (Fig. 4-12 lanes 1 and 2).
Since the NITEGE fragments were generated by ADAMTS4 from FLAG-rbAgg lacking
KS, ADAMTS4 does not require KS within this region to cleave aggrecan at E373-A374.
- 177 -
FLAG-rbAgg A1A1D1 Keratanases - + - + 250 160 105 75 50
kDa 1 2 3 4
Fig. 4-12 Keratanase susceptibility of anti-NITEGE reactive fragments. FLAG-rbAgg expressed in COS-7 cells and cartilage-derived steer aggrecan (A1A1D1) were digested with ADAMTS4-p68 for 24 h at 37 ºC. Aggrecan fragments were then treated with chondroitinase ABC (1 h, 37 ºC) followed by digestion without (lanes 1 and 3) or with (lanes 2 and 4) keratanase II and endo-β-galactosidase (2 h, 37 ºC). Samples were separated on a 10% SDS-PAGE gel, transferred to a PVDF membrane, and immunoblotted with anti-NITEGE neoepitope antibody.
- 178 - 4.2.10 Substrate specificity of ADAMTS4 on chondroitin sulfate-free FLAG-aggrecan
Previously (section #4.2.4) we showed that the removal of CS significantly reduces the efficiency of cleavage of steer aggrecan within the CS-2 domain by ADAMTS4-p68.
At least three possible mechanisms may account for such inhibition. First, since
ADAMTS4 binds to sGAGs, efficient cleavage within the CS-2 domain may require that
ADAMTS4 bind to the CS chains attached to aggrecan in the CS-2 domain. Second, since the enzymatic removal of CS leaves hexasaccharide stubs (Fig. 4-13 a) on the core protein, CS-stubs may inhibit the cleavage of the CS-2 region by ADAMTS4. Last, the highly extended structure observed by atomic force and electron microscopic analysis of the CS domains of aggrecan produced by 100 to 150 negatively charged CS chains covalently attached to CS domains (Fig. 4-14 a) may be disrupted upon CS removal, which may inhibit ADAMTS4 cleavage (Fig. 4-14 b and c) (Morgelin et al., 1989; Ng et al., 2003). Indeed, electron microscopic analysis showed that the chondroitinase
ABC-digested aggrecan has a shorter CS domain compared with that of native aggrecan
(Morgelin et al., 1989).
To address these possibilities, we used xylosyltransferase-deficient CHO-745 cells
(Esko et al., 1985) to produce recombinant aggrecan free of CS and hexasaccharide stubs
(Fig. 4-13 c) to determine its susceptibility to ADAMTS4-p68 and p40. Compared to aggrecan substituted with CS (Fig. 4-14 a) or with CS stubs (Fig. 4-14 b), FLAG-rbAgg expressed in CHO-745 would have no sugar moiety on the sites normally substituted with
CS (Fig. 4-14 c) and would not have the highly extended conformation of native aggrecan.
- 179 - (a) CS stub β4 β3 (GalNAc-GlcA)n-GalNAc-GlcA-Gal-Gal-Xyl-Ser
S
(b) N-linked KS stub
(Gal-GlcNAc)n-Gal-GlcNAc-Man β4 β3 β4 S S Man-GlcNAc-GlcNAc-Asn β4 β4 Sia-Gal-GlcNAc-Man Fuc
(c) O-linked KS stub
(Gal-GlcNAc)n-Gal-GlcNA β4 β3β4 β6 S S GalNAc-Thr/Ser Sia-Galβ3 α3
Fig. 4-13 CS and KS stubs remain on aggrecan core protein after chondroitinase ABC and keratanase digestions. (a) Chemical structure of the CS stub generated by complete digestion of CS by chondroitinase ABC and (b, c) KS stubs generated from (b) N- and (c) O-linked KS digestion by keratanase II, keratanase and/or endo-β-galactosidase. Monosaccharide abbreviations, GlcA, glucuronic acid; GalNAc-S, N-acetylgalactosamine 4 (or 6) sulfate; Gal, galactose; Xyl, xylose; GlcNAc, N-acetylglucosamine; Man, mannose; Sia, sialic acid; Fuc, fucose; GalNAc, N-acetylgalactosamine.
- 180 -
Fig. 4-14 Schematic representation of aggrecan varying in CS structure within the CS domains. Proposed structural models of aggrecan having (a) intact CS (approximately 10 kDa per CS chain), (b) CS-stubs that have hexasaccharides, which are generated by chondroitinase ABC digestion, (c) no CS, which is expressed in CHO-745 xylosyltransferase-deficient cells completely lacking carbohydrates on sites normally substituted with CS.
- 181 - To determine the cleavage pattern of aggrecan with or without CS-stubs,
FLAG-rbAgg was expressed in COS-7 and CHO-745 cells. FLAG-rbAgg expressed in
COS-7 cells was digested with chondroitinase ABC to produce CS stubs and cleavage by
ADAMTS4-p68 and p40 was compared with intact FLAG-rbAgg. Cleavage within the
IGD was determined by the appearance of an anti-NITEGE reactive fragment (see Fig.
4-3, bands #12 and 12’) and the disappearance of large molecular mass fragments
reactive to anti-G1 domain antibody (see Fig. 4-3, bands #1, 3, and 4). The cleavage
within the CS-2 domain was determined by the disappearance of full-length aggrecan
(see Fig. 4-3, band #1) and the appearance of smaller fragments (see Fig. 4-3, bands
#7-10), which are reactive to anti-G3 antibody.
As shown in Fig. 4-15, COS-7 cell-expressed FLAG-rbAgg is rapidly cleaved by
ADAMTS4-p68 within both the CS-2 domain (Fig. 4-15 A, a and b, bands #4, 7, and 9)
and the IGD (Fig. 4-15 A, b and c, lanes 1-4, band #12). When COS-7 cell-expressed
FLAG-rbAgg was digested with chondroitinase ABC (de-CS FLAG rbAgg), it was
cleaved slower within the CS-2 domain by ADAMTS4-p68 compared with
CS-substituted aggrecan from COS-7 cells (Fig. 4-15 A, a, lanes 1-4 vs. lanes 5-8, bands
#1 and 7), but cleavage within the IGD was not affected (Fig. 4-15 A, b and c, lanes 1-4
vs. lanes 5-8, band #12). Similarly, CHO-745-derived recombinant aggrecan lacking CS was extremely resistant to cleavage within the CS-2 domain by ADAMTS4-p68 (Fig.
4-15 A, a, lanes 9-12, bands 1 and 7), whereas cleavage within the IGD was comparable to that of COS-7 cell-expressed aggrecan (Fig. 4-15 A, b and c, lanes 9-12, bands #12 and
12’). These results suggest that CS substitution is required for efficient cleavage by
- 182 - ADAMTS4-p68 within the CS-2 domain, whereas CS has little effect on cleavage within the IGD. Furthermore, the inhibition of cleavage in the CS-2 domain of de-CS aggrecan is not caused by the CS-stubs, since CS-free aggrecan is also poorly cleaved within the
CS-2 domain (Fig. 4-15 A, a, lanes 9-12, see band #1).
When FLAG-rbAgg was digested with ADAMTS4-p40, all of the aggrecans tested were rapidly cleaved within the IGD (Fig. 4-15 B, b and c, bands #12 and 12’). On the other hand, cleavage within the CS-2 domain by p40 was negligible for all of the aggrecans (Fig. 4-15 B, a, bands #1 and weak band #7). The p40 form of ADAMTS4, which has only weak binding affinity to sGAG (Flannery et al., 2002), exhibits very weak cleavage activity within the CS-2 domain (Fig. 4-15 B, a). These results suggest that the specific cleavage within the CS-2 domain by p68 may be mediated by its binding to CS via multiple GAG-binding motifs present in the spacer and cysteine-rich domains of
ADAMTS4-p68 (Flannery et al., 2002), which are absent in the p40 form of ADAMTS4.
On the other hand, cleavage within the IGD by either p68 or p40 is not affected by the presence or absence of CS.
- 183 - COS-7 COS-7 (de-CS) CHO-745 (A) p68 Time (m) 0 45 90 180 0 45 90 180 0 45 90 180 (a) G3 1 1 1 250 7 160 7 7 105 9 1 1 1 4 4 (b) G1 250 4 160 105 75 12 12 12 50 12’ (c) NITEGE 75 12 12 50 12 12’ 1 2 3 4 5 6 7 8 9 10 11 12
(B) p40 Time (m) 0 45 90 180 0 45 90 180 0 45 90 180 1 1 1 (a) G3 250 160 7 7 105 1 1 1 (b) G1 250 160 105 75 12 50 12 12 12’ (c) NITEGE 75 12 12 12 50 12’ kDa 1 2 3 4 5 6 7 8 9 10 11 12
Fig. 4-15 Substrate specificity of ADAMTS4-p68 and p40 on CS-modified FLAG-rbAggs. COS-7-expressed FLAG-rbAgg (0.8 pmol) undigested (lanes 1-4) or predigested with chondroitinase ABC (lanes 5-8) and CHO-745-expressed FLAG-rbAgg (lanes 9-12) were digested either with (A) ADAMTS4-p68 (20 ng) or (B) ADAMTS4-p40 (20 ng) in 20 μl of buffer (20 mM Tris (pH 7.2), 150 mM NaCl, and 5 mM CaCl2) for 0, 45, 90, and 180 min at 37 ºC. The reaction was terminated with 22 mM EDTA. Samples were then deglycosylated with chondroitinase ABC, separated on 4-15% SDS-PAGE gradient gels (100 V), transferred to PVDF membranes (22 V, O/N), and immunoblotted with (a) anti-G3 (Lec7), (b) anti-G1 (αG1-2), and (c) anti-NITEGE antibodies by cycles of stripping and reprobing. Bands were visualized with ECL.
- 184 - 4.2.11 Construction of mutagenized full-length bovine aggrecan expression vectors.
Above we have shown that the removal of KS alone from steer aggrecan significantly inhibits cleavage within the IGD by ADAMTS4-p68 (see Fig. 4-6, de-KS aggrecan). When both CS and KS were removed, however, inhibition within the IGD was not as significant (see Fig. 4-6, de-CS & de-KS aggrecan). As shown in Fig. 4-12,
FLAG-rbAgg, which was not substituted with KS but substituted with CS potentially at fewer sites compared to steer aggrecan (see discussion of Chapter 3), was cleaved by
ADAMTS4-p68 at E373-A374, suggesting that KS is not absolutely required for cleavage within the IGD. It is possible, however, that both KS and non-KS oligosaccharides covalently attached to aggrecan may influence the rate of degradation and the sites susceptible to cleavage.
In order to investigate the effects of KS and other oligosaccharides near the
E373-A374 ADAMTS4 cleavage site, we constructed a series of mutant FLAG-rbAggs
with substitutions for potentially glycosylated residues. Reports by others suggest that
T352, T357, and N368 and/or T370 are substituted with KS in bovine and porcine
aggrecan (Barry et al., 1992; Barry et al., 1995). Horber and co-authors have shown by
digesting a series of truncation mutants of human IGD by ADAMTS4 that at least
residues E349 to F386 (38 mer) are required for efficient cleavage within the IGD at
E373-A374 (Horber et al., 2000). Therefore, we performed site-directed mutagenesis on
all the potential O-linked (Thr and Ser) and N-linked (Asn) glycosylation sites within the
E349 to F386 peptide sequence of aggrecan as described in Fig. 4-16 by using the
Quick-change-XL-site-directed mutagenesis kit and determined whether these sites
- 185 - influenced cleavage at E373-A374 by ADAMTS4. Primer sets used for site-directed
mutagenesis are shown in Table 4-III. These residues are also highly conserved among
different species (Fig. 4-17) except for S377, which is replaced with Asn in rat and Thr in
pig aggrecan. Mutants were categorized into three groups. The first group (A) consists of mutations near the E373-A374 ADAMTS4 cleavage site, where all the potential O-linked threonine and serine residues and the potential N-linked asparagine residue were mutated to glutamine. The second group (B) has mutations within the T352IQTVT357 sequence, which has a cluster of potentially glycosylated threonine residues and is predicted to have an extended secondary structure (see Fig. 4-16, 2˚ structure). Each threonine, serine, and asparagine was mutated to glutamine to conserve polarity and relative hydrophobicity to
maintain structural similarity (Fig. 4-18 a-c, and d). The last group (C) consists of the group (B) mutations except that the threonine residues were mutated to valine instead of glutamine. The rational for mutating threonine to valine was based on their beta-branched structural similarity. It is also recognized, however, that this change will introduce a
non-polar residue in place of a polar residue (OHÆCH3) (Fig. 4-18 a and e). These
mutations were made to determine whether the polarity and/or the hydroxyl group of
threonine play a significant role in ADAMTS4 recognition independent from the effects
of glycosylation. These more non-polar substitutions could also alter peptide
conformation in this region. As described above, since T352IQTVT357 is predicted to have an extended structure, the potential effects of a mutation on its secondary structure
(residues 349 to 360) were also analyzed by using available secondary structure
prediction software.
- 186 -
Fig. 4-16 Summary of FLAG-tagged recombinant aggrecan mutants. All threonine, serine, and asparagine residues that can be substituted with N- or O-linked oligosaccharides close-in proximal to the ADAMTS4 cleavage site in the IGD were mutated to either glutamine or valine. Also the predicted secondary structure in this region is shown. C stands for random coil and E stands for the extended structure. See the “Experimental Procedures” for the secondary structure prediction. The T352Q-T355Q-T357Q and T352V-T355V-T357V triple mutants are also called T352, 5, 7Q and T352, 5, 7V, respectively.
- 187 - Table 4-III
Primer sequences and template plasmids used for site-directed mutagenesis Mutations shown in bold indicate mutants tested for ADAMTS4 digestion. Within the primer sequences, codons in bold indicate the mutation, and underlined letters are where the bases are modified.
Primer Template Mutation (s) Upper primer sequence Lower primer sequence location plasmid Clone(s) RE
5'-GAGGAGGACATCCAGATCC 5'-CAGGTCACCGTCTGGATCT T352Q AGACGGTGACCTG-3' GGATGTCCTCCTC-3' 1459-1490 71-28 457-24 BstYI 5'-CATCACCATCCAGCAGGTG 5'-CAGGCCAGGTCACCTGCTG T355Q ACCTGGCCTG-3' GATGGTGATG-3' 1467-1495 71-28 457-43 BtuAI 5'-CATCCAGACGGTGCAGTGG 5'-CACGTCAGGCCACTGCACC T357Q CCTGACGTG-3' GTCTGGATG-3' 1473-1500 71-28 457-70 BstEII 5'-CATCCAGACGGTGCAGTGG 5'-CACGTCAGGCCACTGCACC T352Q- T357Q CCTGACGTG-3' GTCTGGATG-3' 1473-1500 457-24 460-6 T352Q-T355Q- 5'-CATCCAGATCCAGCAGGTG 5'-CAGGCCACTGCACCTGCTG T357Q CAGTGGCCTG-3' GATCTGGATG-3' 1467-1495 460-6 461-4 5'-CCTGCCCCGACAGATCACT 5'-CCTCAGTGATCTGTCGGGG 459-1 N368Q GAGG-3' CAGG-3' 1509-1531 71-28 462-4 5'-CCCCGAAATATCCAGGAGG 5'-GGCTTCACCCTCCTGGATAT T370Q GTGAAGCC-3' TTCGGGG-3' 1513-1539 71-28 459-5 5'-GAAGCCCGAGGCCAGGTGA 5'-GCCGTGAGGATCACCTGGC S377Q TCCTCACGGC-3' CTCGGGCTTC-3' 1534-1563 71-28 460-3 5'-GGCAGCGTGATCCTCCAGG 5'-CGGGCTTTGCCTGGAGGAT T381Q CAAAGCCCG-3' CACGCTGCC-3' 1543-1563 71-28 459-16
5'-GAGGAGGACATCGTCATCC 5'-CACCGTCTGGATGACGATG T352V AGACGGTG-3' TCCTCCTC-3' 1459-1490 71-28 458-3 PflFI 5'-CATCACCATCCAGGTGGTG 5'-GGCCAGGTCACCACCTGGA T355V ACCTGGCC-3' TGGTGATG-3' 1467-1495 71-28 458-18 BstXI 5'-CATCCAGACGGTGGTCTGG 5'-CACGTCAGGCCAGATCACC T357V CCTGACGTG-3' GTCTGGATG-3' 1473-1500 71-28 458-29 BstEII 5'-CATCCAGACGGTGGTCTGG 5'-CACGTCAGGCCAGATCACC T352V-T357V CCTGACGTG-3' GTCTGGATG-3' 1473-1500 458-3 459-24 T352V- T355V- 5'-CATCACCATCCAGGTGGTG 5'-GGCCAGGTCACCACCTGGA T357V ACCTGGCC-3' TGGTGATG-3' 1467-1495 459-24 460-11
Residue numbers are annotated to full-length mature bovine aggrecan (Val-1).
The nucleotide numbers used in the primers are based on their Accession number
BTU76615.
- 188 -
Fig. 4-17 Alignment of the aggrecan IGD sequence from different species. Letters in bold are highly conserved threonine, asparagine, and serine residues, which may be substituted with oligosaccharides. (Numbering at the top is based on the bovine sequence).
- 189 -
(a) Threonine (b) Serine (c) Asparagine
OH OH O NH2 C H C CH3 CH2
CH2 + - + - H3N C COO H3N C COO + - H3N C COO H H
H
(d) Glutamine (e) Valine O NH 2 CH3 C H C CH3 CH2 + - H3N C COO CH2 H + - H3N C COO
H
Fig. 4-18 Chemical structures of threonine, serine, asparagine, glutamine, and valine.
- 190 - The result shows that all of the mutants are likely to maintain the same extended
secondary structure (Table 4-IV).
4.2.12 Expression of full-length FLAG-rbAgg mutant aggrecans and their susceptibility
to ADAMTS4-p40
Initially when starting the mutagenesis studies, we expressed mutants in COS-7 cells.
Although, we found that the COS-7 cells were not substituting aggrecan with detectable levels of KS, we continued to use them for the following reasons. Fetal aggrecan, which is highly resistant to cleavage within the IGD, is thought to be substituted with short non-KS oligosaccharides near the ADAMTS4 cleavage site (E373-A374) (Roughley et al., 2003). Therefore, the expression of aggrecan in COS-7 cells, which can substitute recombinant molecules with both N- and O-linked oligosaccharides (Nehrke and Tabak,
1997), may be a good model for fetal aggrecan. When KS is digested with keratanase, KS stubs are generated that are structurally similar to short oligosaccharides (Fig. 4-13 b and c). Furthermore, at that time, there was no useful alternative mammalian expression system that was known to substitute KS on recombinant molecules. Therefore, we decided to use the COS-7 cell-expressed recombinant aggrecan to focus on elucidating the effects of non-KS, N- and O-linked oligosaccharides, near the E373-A374 cleavage site. We expected this to help us understand the effects of glycosylation on fetal or calf aggrecan, which is poorly substituted with KS and highly resistant to ADAMTS4 cleavage at E373-A374 (Barry et al., 1995; Pratta et al., 2000; Roughley et al., 2003).
- 191 -
Table 4-IV
Probability of having an extended secondary structure in the sequence
(E349DITIQTVTQPD360) Secondary structure predictions for the 32-mer sequence (F342 to E373) of wild type and mutants were obtained by submitting the sequences to “JUFO: Secondary structure prediction for proteins” available at (http://www.jens-meiler.de/jufo.html) (Meiler et al., 2001). Numbers in red indicate more than 10% decreased probability for an extended structure compared to wild type. Numbers in blue indicate more than 10% increased probability for an extended structure compared to the wild-type residue. Larger numbers indicate higher probabilities for having an extended structure.
Residues Mutant E349 D350 I351 T352 I353 Q354 T355 V356 T357 W358 P359 D360 WT 0.264 0.565 0.693 0.671 0.558 0.488 0.484 0.347 0.476 0.187 0.201 0.17 T352Q 0.23 0.558 0.734 0.696 0.660 0.504 0.494 0.35 0.481 0.197 0.202 0.181 T355Q 0.242 0.525 0.713 0.632 0.517 0.500 0.529 0.431 0.537 0.217 0.202 0.159 T357Q 0.252 0.544 0.686 0.623 0.552 0.506 0.486 0.398 0.550 0.218 0.204 0.186 T352,5,7Q1 0.215 0.497 0.707 0.581 0.498 0.426 0.494 0.422 0.469 0.224 0.203 0.167 T352V 0.276 0.601 0.742 0.682 0.635 0.553 0.549 0.412 0.551 0.204 0.202 0.168 T355V 0.271 0.580 0.708 0.680 0.663 0.559 0.532 0.441 0.552 0.215 0.203 0.17 T357V 0.269 0.579 0.7 0.67 0.584 0.518 0.516 0.419 0.553 0.236 0.204 0.17 T352,5,7V2 0.272 0.623 0.764 0.718 0.728 0.592 0.575 0.52 0.596 0.276 0.205 0.18 1 T352Q-T355Q-T357Q triple mutant
2 T352V-T355V-T357V triple mutant
- 192 - In order to confirm the expression of mutant aggrecans in COS-7 cells, intact wild-type and mutant FLAG-rbAggs were purified by Sephadex G-50 size exclusion and DEAE
Sephacel ion exchange chromatography, separated on a 3.0% SDS-PAGE gel, and immunoblotted with anti-FLAG antibody. Both wild-type and mutant aggrecans exhibited similar electrophoretic patterns (Fig. 4-19) having both Agg1 and Agg2 forms of
FLAG-rbAgg as described in Chapter 3. This result suggests that the mutations introduced in the IGD of aggrecan did not cause truncation or abnormal expression of the
FLAG-rbAgg.
We next tested if these mutant aggrecans could be cleaved within the IGD by
ADAMTS4. Both wild-type and mutant aggrecan were digested with ADAMTS4-p40, which effectively cleaved recombinant aggrecan at E373-A374 (Fig. 4-20) according to the representative digestion of FLAG-rbAgg (Fig. 4-10). The sizes of anti-FLAG reactive bands, which overlap with the anti-NITEGE reactive fragment, were compared to determine if the mutations caused any change in the electrophoretic mobility of these fragments due to elimination of active glycosylation sites. As shown by Fig. 4-20, both the N368Q and T370Q mutants gave a band smaller in size (by 5 kDa) compared with the wild-type and the other mutant aggrecans suggesting that these mutations result in the elimination of the N-glycosylation of N368 (Fig. 4-20 A). It is also worth noting that both
N368Q and T370Q strongly reacted with the anti-N368IT370EGE antibody with similar affinity to that of the wild type (compared with the reactivity to anti-FLAG and anti-NITEGE antibodies). Thus, mutations at the N or T of the NITEGE sequence do not seem to significantly affect its sensitivity to the anti-NITEGE neoepitope antibody,
- 193 -
Fig. 4-19 Full-sized wild-type and mutant FLAG-rbAggs expressed in COS-7 cells. Full-length wild-type and mutant FLAG-rbAggs were expressed and purified from COS-7 cell-conditioned media. Each FLAG-rbAgg (3.2 pmol) was separated on 3.0% SDS-PAGE gels (100V), transferred to PVDF membranes (22V, O/N), and immunoblotted with anti-FLAG (M2) antibody. Bands were visualized with ECL.
- 194 - although we have not performed detailed analysis on the antibody’s affinity to these
mutants. Therefore, we used the anti-NITEGE antibody to compare the rate of cleavage
within the IGD at E373-A374 between the wild-type and mutant FLAG-rbAggs
described later. In addition, all of the other mutants could be cleaved within the IGD
and gave doublet bands (#12 and 12’), both of which were reactive to anti-NITEGE
antibody except for the N368Q and T352V-T355V-T357V mutants, which only gave a
single band. As discussed earlier, the two bands are likely the result of differential
glycosylation in these fragments. Since the N368Q mutant resulted in only a single band, it is possible that N368 is differentially glycosylated in the wild type. However, the
T370Q mutant, which also lacks the N-glycosylation site at N368, has two bands reactive to anti-NITEGE antibody. This result suggests that a site other than N368Q may be differentially glycosylated due to a mutation at T370. On the other hand, since the size of a single band of T352V-T355V-T357V is the same as the higher molecular mass band of the wild type, it is possible that these mutations result in increased occupancy of oligosaccharides at N368 or other differentially glycosylated sites. Overall, these results
suggest that the glycosylations at a particular site can be altered by mutation in
neighboring residues. Therefore, when the mutagenesis studies are conducted, these
possibilities have to be taken into account when interpreting the results.
We also observed a slight change in the intensity of each band (Fig. 4-20), which
may reflect their susceptibility to ADAMTS4-p40. Especially, the S377Q mutant
appeared less susceptible to ADAMTS4-p40, compared with the wild type and other
mutants as determined by their reactivity to the anti-NITEGE antibody (Fig. 4-20).
- 195 -
Fig. 4-20 ADAMTS4-p40 digestion of FLAG-rbAgg mutants. Wild-type and mutant aggrecans (0.8 pmol) were digested with 20 ng of ADAMTS4-p40 for 3 h at 37 ºC. Samples were treated with chondroitinase ABC, separated on 4-15% SDS-PAGE gels at 150 V, transferred to PVDF membranes (22V, O/N) and immunoblotted with anti-FLAG (M2) and anti-NITEGE antibodies. The sizes of higher molecular mass bands reactive to the anti-NITEGE antibody are 66.5 kDa, except for the N368Q and T370Q mutants, which are 61.5 kDa. The lower molecular mass bands reactive to anti-NITEGE antibody are 61.5 kDa.
- 196 - Time course digestions, however, should be conducted to determine the difference in the
rate of cleavage at E373-A374. Note that the anti-FLAG antibody apparently bound
weakly to the lower molecular mass anti-NITEGE band (Fig. 4-20, band #12’). This
appears to be an artifact, since, when the membrane was initially probed with anti-FLAG
antibody, the lower molecular mass band was similarly reactive to the anti-FLAG
antibody (see, Fig. 4-11). Therefore, when the membrane was stripped one or more times,
the anti-FLAG antibody apparently becomes less sensitive especially to the lower
molecular mass band (#12’).
4.2.13 Substrate specificity of ADAMTS4-p68 on mutant aggrecans lacking potentially
glycosylated residue3
Wild-type and group (A) mutant FLAG-rbAggs (see Fig. 4-16) expressed in COS-7
cells were purified and digested with ADAMTS4-p68 for 0, 45, 90, and 180 min at 37 ºC,
treated with chondroitinase ABC, separated on 4-15% SDS-PAGE gels, and transferred to
PVDF membranes (Fig. 4-21). Susceptibility to cleavage by ADAMTS4-p68 within the
IGD was compared by imunoblotting with antibody against the NITEGE neoepitope at
E373-A374, (Fig. 4-21).
As described earlier, Poon and co-authors reported that the N-linked KS at N368 potentiates the cleavage at E373-A374 by showing that the removal of KS at N368 by
PNGase F completely abolished this cleavage (Poon et al., 2005). We have mutated each
of two residues within the N-glycosylation motif of N368-I-T370, one of which is
potentially O-glycosylated (T370). Our result showed, however, that the T370Q and
N368Q mutant lacking N-linked oligosaccharides at N368 was cleaved by
- 197 - ADAMTS4-p68 as well as by p40 (see Figs. 4-21 and 4-20). Furthermore, the T370Q
mutant appeared to be cleaved faster than wild-type aggrecan. The N368Q mutant, on the
other hand, was cleaved at a rate similar to wild-type aggrecan. Since T370Q and N368Q
both lack N-linked oligosaccharides, the difference seen between T370Q and N368Q
suggests that the N-linked oligosaccharides at N368 may have little effect on cleavage by
ADAMTS4 and the observed differences may be related to the presence/absence of an
O-linked oligosaccharide on T370 or be due to the glutamine, itself. However, since it appears that the N368Q mutant reached a plateau a little sooner than the wild type, it may require a shorter time course for detailed analysis of this mutant. In this cell line (COS-7), we do not expect the N-linked oligosaccharides at this site to contain KS. It is interesting to speculate that KS substitution on this N-linked oligosaccharide may modulate cleavage.
Fig. 4-21 also shows that the T381Q mutant is cleaved faster than wild-type aggrecan suggesting that T381 may be substituted with O-linked oligosaccharides that are inhibitory to cleavage by ADAMTS4. On the other hand, cleavage at E373-A374 was significantly inhibited in the S377Q mutant. This result suggests that either glycosylation at S377 is required for efficient ADAMTS4 cleavage or that the SÆQ mutation is
inhibitory. We have focused on the glycosylation sites within the IGD N-terminal to the
E373-374 cleavage site, because the glycosylation of residues C-terminal to the
E373-A374 cleavage site have not yet been investigated. Our results, however, show that
both threonine and serine residues that are C-terminal to the ADAMTS4 cleavage site at
E373-A374 also play a role in ADAMTS4 substrate recognition.
- 198 -
(a) (b) Time (m) 0 45 90 180
WT
N368Q
T370Q
S377Q
T381Q
Fig. 4-21 ADAMTS4-p68 digestion of group (A) mutants. (A) Wild-type and mutant FLAG-rbAggs (0.8 pmol) were digested with ADAMTS4-p68 (20 ng) in 20 μl of reaction buffer (20 mM Tris-HCl (pH 7.2), 150 mM NaCl, and 5 mM CaCl2) for 0, 45, 90, and 180 min at 37 ºC, and products generated were digested with chondroitinase ABC (0.02 U/ reaction), separated on 4-15% SDS-PAGE gels (150 V), electrophoretically transferred to PVDF membranes, and immunoblotted with anti-NITEGE antibody. (B) Plots of relative density of each band reactive to anti-NITEGE antibody vs. digestion time. Wild type is shown in black; N368Q, sky blue; T370Q, pink; S377Q, green; and T381Q, gray. The molecular masses of the major bands in each digestion are approximately 66.5 kDa, whereas those of N368Q and T370Q are 61.5 kDa.
- 199 - 4.2.14 The role of the extended structure N-terminal to the ADAMTS4 cleavage site within the IGD.
The wild-type and group (B) and (C) mutant FLAG-rbAgg constructs were digested with ADAMTS4-p68, and each digestion was analyzed as described above for the group
(A) mutants. The mutants having threonine residues mutated to glutamine residues within the cluster of threonine residues (T352IQTVT357) are shown in Fig. 4-22. In addition, the
T370Q mutant was also digested along with the wild type and group B mutants for comparison. Based on the appearance of the anti-NITEGE reactive fragment, it is apparent that the T370Q aggrecan, as we have already shown (see Fig. 4-21), and T357Q mutants show enhanced cleavage compared to the wild-type aggrecan, whereas the
T352Q and T355Q mutants show inhibited cleavage, but not as significantly as that observed in the S377Q mutant. On the other hand, the T352Q-T355Q-T357Q mutant is cleaved similarly to the wild-type aggrecan, perhaps due to a balance between positive and negative effects.
We then digested the group (C) mutants with ADAMTS4-p68 to compare their susceptibility to that of wild-type aggrecan. The results show that all of the mutants
T352V, T355V, and T357V (Fig. 4-23) are cleaved at a rate similar to wild-type aggrecan.
The time course of T352V-T355V-T357V cleavage relative to wild type was analyzed in another experiment conducted under similar conditions with more time points (Fig. 4-24).
It was apparent that T352V-T355V-T357V shows enhanced cleavage compared to wild type. We presently do not understand this behavior; perhaps, ADAMTS4 binds to this region via hydrophobic interactions due to a cluster of valines or due to change in the
- 200 - secondary structure optimal for interacting with ADAMTS4. The other difference
observed between the T352V-T355V-T357V triple mutant and wild type is that cleavage
of T352V-T355V-T357V at E373-A374 generated mostly the higher molecular mass band that corresponds to fragment #12, as observed (see Fig. 4-20 C, lane 5) following cleavage by ADAMTS4-p40. The typical doublet bands reactive to anti-NITEGE
antibody (#12 and 12’) are likely due to variations in glycosylation, since both bands
react with anti-FLAG antibody as well as with anti-NITEGE antibody, which react with
the N-terminus and C-terminus of the fragment (Fig. 4-11). Therefore,
T352V-T355V-T357V triple mutations may have affected glycosylation. Since the results
of the mutations of threonine to valine do not correlate with the results obtained for the
group (B) mutants, it is possible that the altered rate of cleavage observed in the group
(B) and (C) mutants may not entirely be due to the loss of specific glycosylation sites
within the T352IQT355VT357 sequence, but rather changes in structure and hydrophobicity.
As mentioned earlier, the TIQTVT sequence is predicted to have an extended structure
(Fig. 4-16), and this structure may be important for substrate recognition by ADAMTS4.
Since our result suggested that the T352V-T355V-T357V triple mutant, which makes this
part of the sequence extremely hydrophobic (from T352IQTVT to V352IQVVV), might be
recognized by ADAMTS4 more effectively, it is interesting to speculate that ADAMTS4
may normally interact with this region via reversible hydrophobic interactions and that
glycosylation at this site may serve to regulate its binding.
- 201 -
(a) (b) Time (m) 0 10 60 360 WT
T352Q
T355Q
T357Q
T352,5,7Q
T370Q
Fig. 4-22 Representative ADAMTS4-p68 digestion of group (B) mutants. (A) Wild-type and mutant FLAG-rbAggs (0.8 pmol) were digested with ADAMTS4-p68 (20 ng) in 20 μl of reaction buffer (20 mM Tris-HCl (pH 7.2), 150 mM NaCl, and 5 mM CaCl2) for 0, 10, 60, and 360 min at 37 ºC, and products generated were digested with chondroitinase ABC (0.02 U/ reaction) in 100 mM sodium acetate (pH 6.5), 14 mM Tris-HCl, 100 mM NaCl for 1 h at 37 ºC, separated on 4-15% SDS-PAGE gels, electrophoretically transferred to PVDF membranes, and immunoblotted with the anti-NITEGE antibody. (B) Plots of relative density of each band reactive to the anti-NITEGE antibody vs. digestion time. Wild-type is shown in black; T352Q, pink; T355Q, green; T357Q, purple; T352Q-T355Q-T357Q, brown; and T370Q, pink (dots). The molecular mass of these bands is approximately 66.5 kDa for the larger bands and 61.5 kDa for the smaller bands and the larger band of T370Q.
- 202 -
(a) (b) Time (m) 0 45 90 180 WT
T352V
T355V
T357V
Fig. 4-23 Representative ADAMTS4-p68 digestion of group (C) mutants. (a) Wild-type and mutant FLAG-rbAggs (0.8 pmol) were digested with ADAMTS4-p68 (20 ng) in 20 μl of reaction buffer (20 mM Tris-HCl (pH 7.2), 150 mM NaCl, and 5 mM CaCl2) for 0, 45, 90, and 180 min at 37 ºC, and products generated were digested with chondroitinase ABC (0.02 U/reaction), separated on 4-15% SDS-PAGE gels, electrophoretically transferred to PVDF membranes, and immunoblotted with the anti-NITEGE antibody. (b) Plots of relative density of each band reactive to anti-NITEGE antibody vs. digestion time. Wild-type is shown in black; T352V, blue; T355V, pink; T357V, green, and T352V-T355V-T357V, brown. The molecular mass of these bands is approximately 66.5 kDa for the larger bands and 61.5 kDa for the smaller bands.
- 203 - (a)
Time (m) 0 11.25 22.5 45 90 180 360
WT
T352,5,7V
(b)
Fig. 4-24 Representative ADAMTS4-p68 digestion of a triple valine mutant (T352V-T355V-T357V). (a) Wild-type and mutant FLAG-rbAgg (40 nM) were digested with ADAMTS4-p68 (20 ng) in 20 μl of reaction buffer (20 mM Tris-HCl (pH 7.2), 150 mM NaCl, and 5 mM CaCl2) for 0, 11.25, 22.5, 45, 90, 180, and 360 min at 37 ºC, and products generated were digested with chondroitinase ABC (0.02 U/ reaction), separated on 4-15% SDS-PAGE gels, electrophoretically transferred to PVDF membranes, and immunoblotted with anti-NITEGE antibody. (b) Plots of relative density of each band reactive to anti-NITEGE antibody vs. digestion time. Wild-type is shown in black and T352V-T355V-T357V is in brown. The molecular mass of upper (major) bands is approximately 66.5 kDa and the lower band of wild type is 61.5 kDa.
- 204 - 4.2.15 Representative digestion of wild-type FLAG-rbAgg expressed in COS-7 cells with
MMP13
MMP13 is one of the major MMPs shown to cleave aggrecan within the IGD at
S341-F342, but recent reports suggest that MMPs are not involved in detrimental aggrecan degradation in osteoarthritis (Arner, 2002; Little et al., 1999). It may, however, play a bigger role during the endochondral bone development (Stickens et al., 2004).
To characterize the substrate specificity of MMPs against recombinant aggrecan,
FLAG-rbAgg was digested with MMP13. The pattern of time course digestion was analyzed by Western blot with, anti-FLAG, anti-FFGVG, anti-G3 domain, and anti-VDIPEN antibodies (Fig. 11 a-d). For the MMP13-mediated catabolites, we used a unique numbering system. Based on this time course, we have generated a model for
MMP13-mediated degradation of recombinant aggrecan (Fig. 11 e). The digestion of recombinant aggrecan with MMP13 generates the expected anti-FFGVG (Fig. 11 b) and anti-VDIPEN (Fig. 11 d) antibody-reactive fragments from cleavage within the IGD at
S341-F342 (Fig. 11 e, cleavage #2).
Our analysis has revealed a major site within the CS domain (Fig. 11 e, cleavage #1) giving a high molecular mass band with an intact FLAG-tagged N-terminus (Fig. 11 a, lane 10, band II (~355 kDa)) reactive to the anti-FLAG antibody, and other bands, which are non-FLAG reactive, but anti-FFGV positive (Fig. 11 b, bands III (~290), IV (270), V
(190), and VII (140 kDa)) that appeared at the later time point by cleavage (Fig. 11 e, cleavages #4, #6, and #8) N-terminal to the initial cleavage site (Fig. 11 e, cleavage #1).
This suggests that cleavage within the CS domain to generate a band II occurs more
- 205 - rapidly than cleavage at S341-F342. The C-terminal fragment (band VI (160 kDa))
generated by cleavage in the CS domain was also further degraded (Fig. 11 e, cleavages
#3, #5 and #7) to smaller fragments (Fig. 11 c, bands VIII (110 kDa), IX (100 kDa), and
X (65 kDa)). These results suggest that MMP13 is capable of cleaving the aggrecan core
protein at multiple sites, many of which have not previously been identified.
4.2.16 MMP13 digestion of mutant aggrecans
Pratta and co-workers have shown that both chondroitinase and keratanase digestion
of cartilage-derived aggrecan has no effect on aggrecan cleavage by MMP13 (Pratta et al.,
2000), suggesting that glycosylation does not affect cleavage by MMP13, unlike
ADAMTS4. Nevertheless, since the MMP cleavage site is also proximal to the sites
mutated in this work, we characterized the effects of mutations in the IGD on the rates of
cleavage at S341-F342 by MMP13. Selected aggrecan mutants (T352Q, T355Q, T357Q,
T352Q-T355Q-T357Q, N368Q, and T370Q) were digested with MMP13, and their
susceptibility to cleavage at S341-F342 was monitored by the appearance of
anti-VDIPEN reactive fragments (Fig. 4-26). The plots show that MMP13 cleaves these
mutants at rates and levels very similar to those of wild-type recombinant aggrecan.
These results suggest that the differences in aggrecan susceptibility observed in the
mutants with ADAMTS4-p68 are specific to ADAMTS4. Therefore, MMP13 seems less
likely to be affected by glycosylation within the IGD than ADAMTS4.
- 206 -
Fig. 4-25 MMP13 digestion of FLAG-rbAgg. Wild-type FLAG-rbAgg (0.8 pmol) was digested with recombinant human MMP13 (20 ng) in 20 μl of reaction buffer (20 mM Tris-HCl (pH 7.2), 150 mM NaCl, and 5 mM CaCl2) for 0, 5, 30, and 180 min at 37 ºC. Reactions were terminated by 21 mM EDTA and deglycosylated with chondroitinase ABC (0.02 U/reaction) for 1 h at 37 ˚C. Deglycosylated fragments were separated on a 4-15% gradient gel, transferred to a PVDF membrane, and subsequently immunoblotted with (a) anti-FLAG (M2) (lanes 1-4), (b) anti-FFGVG (BC-14) (lane 5-8), (c) anti-G3 domain (lanes 9-12), and (d) anti-VDIPEN(S) (lanes 13-16) antibodies. (a, c, and d) were visualized with ECL, (b) was visualized with ECL plus. Bands identified by asterisks are from chondroitinase ABC (equally present in all samples), which cross-reacts with the polyclonal antibodies, therefore providing an internal loading control. A band identified with XI* is FLAG-reactive band, which has not been completely stripped off. Unique Roman numerals are given to visible fragments. (e) Proposed MMP13-mediated recombinant aggrecan digestion pathway. Fragments identified with asterisks are hypothetical. Each Arabic number indicates a different cleavage site. Number orders do not necessarily correspond to the order of cleavage. The predicted sequence of cleavage runs from top to bottom at each site.
- 207 -
Fig. 4-25 (continued)
- 208 -
(a) (b)
Time (m) 0 5 30 180 WT
N368Q
T370Q
T352Q
T355Q
T357Q T352,5,7Q
Fig. 4-26 MMP13 digestion of wild-type and mutant FLAG-rbAggs. (a) Wild-type and mutant FLAG-rbAgg (0.8 pmol) were digested with MMP13 (20 ng) for 0, 5, 30, and 180 min at 37 ºC, and products generated were digested with chondroitinase ABC (0.02 U/reaction) in 100 mM sodium acetate (pH 6.5), 14 mM Tris-HCl, 100 mM NaCl for 1 h at 37 ºC, separated on 4-15% SDS-PAGE gels, electrophoretically transferred to PVDF membranes, and immunoblotted with anti-VDIPEN antibody. (b) Plots of relative density of each band reactive with anti-VDIPEN antibody vs. digestion time. Wild type is shown in black; T352Q, purple; T355Q, blue; T357Q, red; T352Q-T355Q-T357Q, brown; N368Q, sky blue; and T370Q, pink. Note that the band at 180 min of the T357Q mutant is smeared and may over represent the actual intensity.
- 209 - 4.2.17 Keratan sulfate synthesis in COS-7, CHO-K1, and RCS cells
Previous work described in Chapter 3 (see Fig. 3-2) suggested that the COS-7 cell line synthesizes very little or no KS that is detectible by the 5-D-4 anti-KS antibody. Akama and co-authors found that HeLa cells were able to produce 5-D-4 positive KS when they were co-transferred with corneal GlcNAc 6-O-sulfotransferase (CGn6ST) and keratan sulfate Gal-6-O-sulfotransferase (KSG6ST) (Akama et al., 2001). To achieve KS substitution of aggrecan, we transfected the COS-7, CHO-K1, and rat chondrosarcoma
(RCS) cell lines with vectors encoding two sulfotransferases, CGn6ST and KSG6ST, that are involved in KS biosynthesis (Akama et al., 2001) to test whether the addition of sulfotransferases enabled these cells to produce KS (Fig. 4-27). Since cartilage aggrecan contains both N-linked and O-linked KS, we also co-transfected both COS-7 and
CHO-K1 cells with the O-linked core 2 GlcNAc transferase (C2GnT), an enzyme involved in O-linked KS biosynthesis (Fig. 4-28). We chose to use CHO-K1 cells to test the hypothesis that C2GnT is required for O-linked KS synthesis, since CHO-K1 is
C2GnT deficient (Bierhuizen and Fukuda, 1992) and should not make O-linked KS unless the cells were transfected with the C2GnT construct.
First, we co-expressed CGn6ST and KSG6ST in COS-7, CHO, and RCS cells and analyzed the KS synthesis by immunocytochemistry with the anti-KS 5-D-4 antibody. As shown in Fig. 4-27, RCS, COS-7, and CHO cells were able to produce 5-D-4 positive KS only when they were co-transfected with expression vectors encoding CGn6ST and
KSG6ST, as demonstrated by immunocytochemistry. We then digested the cell lysates of transfected COS-7 and CHO cells with PNGase F to remove N-linked KS for Western
- 210 - blot analysis (Fig. 4-28). 5-D-4 positive bands remaining on the gel would only be
KS-substituted proteoglycans having O-linked KS chains. CHO-K1 should only produce
O-linked KS if transfected to overexpress C2GnT. In both cell lines, PNGase F treatment substantially eliminated immunostaining by the 5-D-4 anti-KS antibody. Therefore, most of the KS chains produced by both cell lines appear to be N-linked. A low amount of
PNGase F-resistant KS (i.e., presumably O-linked KS) is observed in the COS-7 cells, which have an endogenous C2GnT activity (Bierhuizen and Fukuda, 1992) in the presence or absence of overexpressed C2GnT (Fig. 4-28 b, lanes 2 and 4). In contrast,
CHO-K1 cells, which have no endogenous C2GnT activity (Bierhuizen and Fukuda,
1992), showed very little of the PNGase F resistant KS in the absence of C2GnT (Fig
4-28, lane 2). In the cell lysates of CHO-K1 cells overexpressing C2GnT, however,
PNGase F-resistant KS proteoglycans (Fig. 4-28, lane 4, arrow) were observed. This result indicates that COS-7 cells may substitute aggrecan with O-linked KS following co-transfection with CGn6ST and KSG6ST, but that CHO-K1 may require expression of
C2GnT as well. This could be used to express aggrecan having both N- and O- linked KS
(COS-7), or only N-linked KS (CHO-K1). Overall, however, both cell lines appear to mainly produce N-linked KS even in the presence of C2GnT.
- 211 -
RCS COS-7 CHO-K1
pcDNA3 (no insert)
CGn6ST KSG6ST
Fig. 4-27 Immunocytochemical analysis of KS production in cell lines transiently overexpressing sulfotransferases. Rat chondrosarcoma (RCS), COS-7, and CHO-K1 cells were either transfected with pcDNA3 with no inserts or with pcDNA3-CGn6ST and pcDNA3-KSG6ST. Cells were immunostained with anti-KS (5-D-4) antibody. Green is 5-D-4 and blue is DAPI (nuclei). Bar = 80 μm
- 212 -
Fig. 4-28 N- and O-linked KS production in COS-7 and CHO cells overexpressing 6-O-sulfotransferases and core 2 GlcNAc transferase. Both COS-7 and CHO-K1 cells were transiently transfected either with pcDNA3 (lanes 1 and 5), pcDNA3-CGn6ST and pcDNA3-KSG6ST (lanes 2 and 6), pcDNAI-C2GnT (lanes 3 and 7), pcDNA3-CGn6ST, pcDNA3-KSG6ST, and pcDNAI-C2GnT (lanes 4 and 8). Membrane fractions of cell lysates were either (a) undigested or (b) digested with PNGase F, separated on 4-15% gradient SDS-PAGE gels, electrophoretically transferred to PVDF membranes, and immunoblotted with anti-KS (5-D-4) antibody. Bands were visualized by a 20 sec exposure with ECL.
- 213 - 4.2.18 Co-expression of sulfotransferase and FLAG-rbAgg construct
Since COS-7 cells were able to produce KS by co-expressing CGn6ST and KSG6ST
constructs, we then attempted to co-express the FLAG-rbAgg construct with CGn6ST and KSG6ST in COS-7 cells to produce KS-substituted FLAG-rbAgg. The C2GnT construct was co-transfected in order to maximize the probability of obtaining O-linked
KS. To determine the presence of KS on FLAG-rbAgg co-expressed with CGn6ST,
KSG6ST, and C2GnT, we analyzed purified recombinant aggrecan on a 3.0% SDS-PAGE gel (Fig. 4-29). Both FLAG-rbAgg expressed with/without sulfo- and glycosyl-transferases were reactive to the anti-FLAG, anti-G1, and anti-G3 antibodies suggesting that they are expressed and secreted in full-length form (Fig. 4-29 a-c). The electrophoretic pattern of FLAG-rbAgg expressed in COS-7 cells without CGn6ST,
KSG6ST, and C2GnT was clearly different (Fig. 4-29, lanes 1 and 3) from that co-expressed with CGn6ST, KSG6ST, and C2GnT. FLAG-rbAgg co-expressed with sulfo- and glycosyl-transferases showed a higher degree of microheterogeneity on the gel
(Fig. 4-29, lane 3) compared to FLAG-rbAgg overexpressed alone in COS-7 cells (Fig.
4-29, lane 1). To determine the presence of KS on FLAG-rbAgg, it was first digested with chondroitinase ABC and then immunoblotted with 5-D-4 anti-KS antibody (Fig.
4-29 d). The chondroitinase ABC-digested KS-FLAG-rbAgg co-expressed with transferases reacted with the 5-D-4 antibody suggesting the presence of KS.
FLAG-rbAgg expressed without sulfo- and glycosyl-transferases only weakly reacted with the anti-KS antibody. Unexpectedly, the diffuse band for KS-FLAG-rbAgg collapses into a faster migrating band after chondroitinase ABC digestion indicating that the
- 214 - FLAG-rbAgg KS-FLAG-rbAgg Ch’ase ABC - + - + (a) FLAG
Agg1 Agg1 Agg2
(b) G1
Agg1 Agg1 Agg2
(c) G3
Agg1 Agg1 Agg2 Agg2
(d) KS
Agg1 Agg1 Agg2
1 2 3 4 Fig. 4-29 3.0% SDS-PAGE analysis of FLAG-rbAgg expressed in COS-7 cells along with sulfotransferases. Purified FLAG-rbAgg (0.8 pmol) isolated from transiently transfected COS-7 cells non-co-transfected (lanes 1 and 2) or co-transfected with CGn6ST, KSG6ST, and C2GnT (lanes 3 and 4) were either undigested (lanes 1 and 3) or digested (lanes 2 and 4) with chondroitinase ABC, separated on a 3.0% SDS-PAGE gel, transferred to a PVDF membrane, and subsequently immunoblotted with antibodies against (a) FLAG (M2), (b) the G1 domain (αG1-2), (c) the G3 domain (Lec7), and (d) KS (5-D-4).
- 215 - microheterogeneity is primarily due to CS (Fig. 4-29, lanes 4). It has been reported that
chondroitin 6-O-sulfotransferase (C6ST-1), for example, can transfer sulfate to both KS
and CS (Habuchi et al., 1993). Increased sulfation of CS by KS-sulfotransferases has not been previously demonstrated, but is a possible explanation for the observed increase in microheterogeneity of CS chains. Therefore, in addition to enhancing KS biosynthesis, overexpression of sulfotransferases may also have resulted in altered CS biosynthesis.
Further analysis is required, however, to confirm and extend this observation.
4.2.19 Susceptibility of KS-FLAG-rbAgg to ADAMTS4
To determine the cleavage pattern of KS-FLAG-rbAgg expressed in COS-7 cells
and to compare the rate of cleavage to those expressed in COS-7 cells without
co-expression of sulfo- and glycosyl-transferases (FLAG-rbAgg), we digested both
recombinant aggrecans with ADAMTS4-p68 and ADAMTS4-p40 (Fig. 4-30). We have
determined that the number of units of p68 and p40 used in each digest would cleave
FLAG-rbAgg within the IGD at a similar rate (see Fig. 4-15 c, lane 1-4). The membrane
was stripped and reprobed with different antibodies (Fig. 4-1 and Table 4-I) in order to
identify the fragments (Fig. 4-30). The time-course digestion of KS-FLAG-rbAgg with
ADAMTS4-p68 (Fig. 4-30 A, lanes 5-8) shows that KS-FLAG-rbAgg was poorly
cleaved within the CS-2 domain (Fig. 4-30 A, a, bands #1 and 7) as well as within the
IGD judging from the lack of fragments derived from cleavage at E373-A374 compared
with that of FLAG-rbAgg (Fig. 4-30 A, lanes 1-4 vs. 5-8, bands #12 and 12’). On the
other hand, KS-FLAG-rbAgg was cleaved similarly to FLAG-rbAgg by ADAMTS4-p40
- 216 - (Fig. 4-30 B). Cleavage within the CS-2 domain was inhibited as described earlier (Fig.
4-30 B, a, bands #1 and 2), but cleavage within the IGD (Fig. 4-30 B, bands #12 and 12’)
was comparable with that by ADAMTS4-p68 (Fig. 4-30 A, bands #12 and 12’).
Interestingly, the anti-NITEGE antibody only reacted with the lower molecular mass
band of KS-FLAG-rbAgg that was cleaved by ADAMTS4-p40 at E373-A374 (Fig.
4-30 C, lane 4, band #12’). When the same membrane was probed with anti-G1 domain
antibody, the higher molecular mass band appeared (Fig. 4-30 C, lane 3). On the other
hand, both higher and lower molecular mass bands (#12 and 12’) generated by the
ADAMTS4-p40 cleavage of FLAG-rbAgg were reactive to both anti-G1 and
anti-NITEGE antibodies (Fig. 4-30 C, lanes 1 and 2). This result suggests that the
neoepitope (NITEGE) may be masked by some form of post-translational modification, most likely by sulfated oligosaccharides (e.g., KS, etc.) that are only present in
KS-FLAG-rbAgg, but not in FLAG-rbAgg. We noted, however, that the upper band was
still not reactive to anti-NITEGE antibody after keratanase II and endo-β-galactosidase
digestion (data not shown). Since keratanase digestion does not completely remove the
oligosaccharides on the peptide, it is likely that either the KS stubs or oligosaccharides resistant to keratanase cleavage may be masking the NITEGE epitope.
- 217 -
Fig. 4-30 Substrate specificity of ADAMTS4-p68 and p40 on KS-FLAG-rbAgg. COS-7-expressed FLAG-rbAgg (rbAgg, lanes 1-4) (0.8 pmol) and KS-FLAG-rbAgg (KS-rbAgg, lanes 5-8) (0.8 pmol) were digested either with (A) ADAMTS4-p68 (20 ng) or (B) ADAMTS4-p40 (20 ng) in 20 μl of buffer (20 mM Tris (pH 7.2), 150 mM NaCl, and 5 mM CaCl2) for 0, 45, 90, and 180 min at 37 ºC. The reaction was terminated with 22 mM EDTA, deglycosylated with chondroitinase ABC, separated on 4-15% SDS-PAGE gradient gels (100 V), transferred to PVDF membranes (22 V, O/N), and immunoblotted with (a) anti-G3 (Lec7), (b) anti-G1 (αG1-2), and (c) anti-NITEGE antibodies by cycles of stripping and reprobing. Bands were visualized with ECL. Bands identified by asterisks are the chondroitinase ABC enzyme (equally present in all samples), which cross-reacts with the polyclonal antibodies, therefore providing an internal loading control. (C) Enlarged image of 180 min time points (lanes 4 and 8) of fragments #12 and 12’ generated by ADAMTS4-p40 cleavage in (B, b and c).
- 218 -
Fig. 4-30 (Continued)
- 219 - 4.3 Discussion
It has been suggested that the substrate specificity of ADAMTS4 is regulated by sGAGs covalently attached to the aggrecan core protein (Kashiwagi et al., 2004; Poon et al., 2005; Pratta et al., 2000; Tortorella et al., 2000), and the presence of GAGs is essential for efficient cleavage within the aggrecan core protein by ADAMTS4 (Sugimoto et al., 1999; Tortorella et al., 2000). Since aggrecan undergoes age-related changes in its glycosylation, understanding the glycosylation-dependent changes in its susceptibility to
ADAMTS4 is particularly important to explain why aggrecan isolated from older individuals is more susceptible to degradation by ADAMTS4 (Pratta et al., 2000;
Roughley et al., 2003). In the present work, we used both cartilage-derived steer aggrecan and recombinant aggrecan as in vitro experimental substrates for ADAMTS4 to clarify some of the contradicting observations made by others regarding the substrate specificity influenced by the presence of sGAGs.
4.3.1 Effects of chondroitin sulfate and keratan sulfate on substrate specificity of
ADAMTS4
In this work, it was clearly demonstrated that CS and KS influence cleavage at distinct sites by using steer aggrecan as an experimental substrate for ADAMTS4-p68
(see Fig. 4-31). We showed that while CS is required for cleavage within the CS-2 domain (Fig. 4-31 A, a and b), it is not required for cleavage within the IGD (Fig. 4-31 A, c). Conversely, KS is required for efficient cleavage within the IGD, but has no effect on cleavage within the CS-2 domain (Fig. 4-31 A, b). Our novel finding is that the inhibition
- 220 - of cleavage at E373-A374 by enzymatic removal of KS is reversed by the removal of CS
(Fig. 4-31 A, d). It is also interesting to note that in the absence of both CS and KS, the
cleavage within the CS-2 domains is still preferred over cleavage within the IGD,
suggesting that ADAMTS4-68 would only cleave within the IGD after cleaving within
the CS-2 domain, except when CS is removed (Fig. 4-31 A, c). Previous studies by others
have only evaluated cleavage at E373-A374 by the appearance of anti-ARGSV reactive
fragments and have concluded that the removal of both CS and KS resulted in inhibition
of cleavage within the IGD, since anti-ARGSV reactive fragments were either absent or
weakly detected (Kashiwagi et al., 2004; Pratta et al., 2000; Tortorella et al., 2000). Our
study using cartilage-derived steer aggrecan also showed that when both CS and KS were
removed, the anti-ARGSV reactive fragment (which typically has the TAGELE1480
C-terminus) only starts to appear at a later time point compared to that for cleavage of native aggrecan. We believe, however, that this result has been misinterpreted by others as representing significant inhibition of cleavage within the IGD. When we compared this result with the appearance of the G1-NITEGE fragment, it is clear that cleavage within the IGD is not significantly inhibited when both KS and CS were removed.
We also show that recombinant aggrecan lacking KS can be cleaved by ADAMTS4 at E373-A374. These results suggest that KS is not absolutely required to cleave aggrecan within the IGD, but may only be required in the presence of abundant CS. Therefore, recombinant aggrecan having a smaller amount of CS (see Figs. 3-7 and 3-8 in Chapter 3) may not require KS for efficient cleavage at E373-A374 (Fig. 4-31 A, e). On the other hand, recombinant aggrecan expressed in COS-7 cells digested with chondroitinase and
- 221 - recombinant aggrecan expressed in CHO-745 cells lacking CS (CS-free aggrecan) were resistant to ADAMTS4 digestion within the CS-2 domain, while cleavage within the IGD was not affected (Fig. 4-31 A, f and g). This result is consistent with the observation that
CS is required on steer aggrecan for efficient cleavage within the CS-2 domain regardless of the presence or absence of KS. Furthermore, the inhibition within the CS-2 domain by enzymatic removal of CS is not due to an inhibitory effect of CS-stubs, since
CHO-745-expressed aggrecan, which has no CS-stubs, is not cleaved within the CS-2 domain. However, it still remains to be determined whether efficient cleavage within the
CS-2 domain is due to a CS-ADAMTS4 interaction, or to the more extended secondary structure of CS-substituted aggrecan making the cleavage site more accessible to
ADAMTS4. Previous EM studies have shown a distinct shortening of the CS domain upon chondroitinase ABC digestion (Morgelin et al., 1989). Therefore, it is possible that
CS removal can mask the cleavage sites due to a change in its secondary structure within the CS domains. Recombinant aggrecan expressed in CHO-745 cells appeared to be more resistant to cleavage within the CS-2 domain than chondroitinase ABC-digested recombinant aggrecan expressed in COS-7 cells. This may indicate that CS stubs can support cleavage within the IGD to some extent.
4.3.2 Substrate specificity of ADAMTS4-p68
In spite of reports by others suggesting the ADAMTS4-p68 only poorly cleaves within the IGD (Gao et al., 2004; Gao et al., 2002; Kashiwagi et al., 2004), our result showed that both cartilage-derived and our recombinant aggrecans were cleaved within the IGD
- 222 - by purified ADAMTS4-p68 as summarized in Fig. 4-31 A. It is possible that this cleavage
was mediated by small amounts of p53 (and p40) that were detectible by imunoblotting
analysis. It is interesting to note that ADAMTS4-p68 apparently cleaves within the CS-2
domain before IGD cleavage is observed, unless CS was enzymatically removed or was
absent on the aggrecan core protein. This result suggests that p68 may bind CS tightly
and is therefore unavailable for IGD cleavage. In the absence of CS, or following cleavage within the CS-2 domain, which may be accompanied by autoproteolytic cleavage of p68 to a form having less CS affinity (i.e., p53 and p40), ADAMTS4-p68 (if no CS is present) or smaller processed ADAMTS4 isoforms are available to the IGD to cleave at E373-A374 as was suggested by Flannery et al., 2000. This may not be the case, however, since cleavage inhibition within the CS-2 domain does not result in an increased rate of cleavage within the IGD. If the same enzyme mediates cleavage, it is assumed that the inhibition of one of the sites should facilitate cleavage at the other site.
Therefore, it is possible that the majority of cleavages within the CS-2 domain and the
IGD occur independently and are mediated by different isoforms; namely cleavage within the CS-2 domain would be mediated by p68 and within the IGD by smaller isoforms of
ADAMTS4.
To test this hypothesis, it is essential to obtain a mutant ADAMTS4-p68 that is
deficient in undergoing autoproteolysis but retains its activity to cleave aggrecan.
Alternatively, the rate of cleavage of aggrecan incubated with the p68 form of
ADAMTS4 may be compared with that of aggrecan digested with p68 that has been
pre-incubated in the absence of substrates for a given time to allow autoproteolysis.
- 223 - 4.3.3 Substrate specificity of ADAMTS4-p40
As summarized in Fig. 4-31, we made a number of significant observations using
the p68 and p40 isoforms of ADAMTS4 to digest both steer and recombinant aggrecans.
First, while both steer and recombinant aggrecans were efficiently cleaved by
ADAMTS4-p68 within both the IGD and CS-2 domains (Fig. 4-31 A, a), steer aggrecan
was poorly cleaved by p40 within both the IGD and CS-2 domains (Fig. 4-31 B, a) by
using the same amount of substrate and enzyme. In contrast, all recombinant aggrecans
tested were efficiently cleaved within the IGD by ADAMTS4-p40 (Fig. 4-31 B, b), and
cleavage was not affected by the presence or absence of CS. However, since the major
biochemical difference between steer and recombinant aggrecans is their glycosylation,
the glycosylation of aggrecan may have a significant effect on the substrate specificity of
ADAMTS4-p40 as it does for ADAMTS4-p68. Therefore, it will be interesting to
investigate the susceptibility of de-glycosylated steer aggrecan to ADAMTS4-p40 to determine if such a difference is due to the presence of certain glycosaminoglycans.
Interestingly, ADAMTS4-p68 digestion of CHO-745-expressed “CS-free”
recombinant aggrecan resembled the pattern of cleavage obtained by digesting all types
of recombinant aggrecan with ADAMTS4-p40, which only cleaves within the IGD (Fig.
4-31 A, g and B, b). These results suggest that the cysteine-rich and/or the spacer domain,
which are lacking in p40 (see Fig. 1-10), are required to cleave the CS-2 domain; and this
may be mediated by their binding to CS via multiple GAG binding motifs (Flannery et al.,
2002). It still remains to be solved, however, why p40 recognizes and cleaves
recombinant aggrecan within the IGD in a non-CS-dependent manner. It is possible that
- 224 - the ADAMTS4-p40 has a stronger affinity to the exosite amino acid sequence within the
IGD, but low affinity to the exosite sequence within the CS-2 domain. The exosite and/or
substrate-binding site of ADAMTS4, which may only be exposed when the p68 form is
converted to the p40 form of ADAMTS4, may mediate this interaction. The resolution of
this question awaits the determination of the three-dimensional structure of ADAMTS4,
which is not available at this writing.
4.3.4 Effects of Keratan sulfate substitution on cleavage within the IGD
Aggrecan isolated from younger individuals is speculated to have shorter KS chains or non-KS oligosaccharides substituted at sites within the IGD that normally are
substituted with long KS chains in the adult (Barry et al., 1995; Roughley et al., 2003).
Direct visualization by rotary shadowing electron microscopic analysis revealed that
IGDs substituted with KS appeared to be more extended than IGDs lacking KS (Mercuri
et al., 1999). Therefore, short non-KS oligosaccharides may have an inhibitory effect on
ADAMTS4-mediated cleavage within the IGD either through shortening the length of the
IGD, thereby affecting enzyme accessibility, and/or through steric hindrance (Roughley
et al., 2003). Our results, however, show that when both CS and KS are removed,
cleavage within the IGD was increased to a level similar to that of native aggrecan. This
result suggests that KS may be important to retain ADAMTS4-p68 bound to the IGD in
equilibrium with ADAMTS4-p68 bound to CS in the CS domains. Alternatively, we can
speculate that upon removal of CS in addition to KS, this equilibrium may shift in favor
- 225 -
Fig. 4-31 Schematic model showing the glycosaminoglycan-dependent substrate specificity of ADAMTS4-p68 and ADAMTS4-p40. Preferred cleavage sites of (a) Native, (b) de-KS, (c) de-CS, (d) de-CS & KS, (e) native recombinant aggrecan, (f) de-CS COS-7 aggrecan, and (g) CHO-745-expressed aggrecan by (A) ADAMTS4-p68 or (a) native steer aggrecan and (b) recombinant aggrecans by (B) ADAMTS4-p40. Numbers in arrows indicate the order of cleaving within the aggrecan core protein (IGD (left) vs. CS-2 (right)). Arrows with a red “X” indicate significantly reduced cleavage at sites within the IGD or CS-2 domain.
- 226 - of the IGD binding site, due to collapse of previously extended CS domains and the
resulting change in secondary structure.
4.3.5 Effects of non-GAG oligosaccharides on aggrecan cleavage by ADAMTS4
As described above, fetal aggrecan, which may have short oligosaccharides in the
IGD, is resistant to ADAMTS4 cleavage (Roughley et al., 2003). In the present work, we determined whether the potential glycosylation sites within the IGD can affect aggrecan’s susceptibility to ADAMTS4 by generating a series of mutant aggrecans lacking potentially glycosylated resides. To express the mutants, we used COS-7 cells, which synthesize minimal KS and likely substitute aggrecan with short non-KS oligosaccharides.
As summarized in Fig. 4-32, three mutants, T357Q, T370Q, and T381Q, which lack potential O-linked and N-linked oligosaccharides (N368) were cleaved faster within the
IGD than the wild type by ADAMTS4-p68, suggesting that the elimination of these potential glycosylation sites enhances cleavage within the IGD.
Poon and co-authors recently reported that N-linked KS at N368 was required for efficient cleavage within the IGD and that the enzymatic removal of N-linked KS completely abolished cleavage at E373-A374 (Poon et al., 2005). In this study, a recombinant truncated aggrecan construct containing the sequence from the G1 to G2 domains (lacking the KS, CS, and G3 domains) was expressed in primary keratocytes
(Poon et al., 2005). Our results, however, suggest that elimination of the N-linked oligosaccharide at N368 instead enhances cleavage at E373-A374. First, mutation of
- 227 - N368 or T370 to glutamine, which results in elimination of the N-linked glycosylation of
N368, did not result in inhibition of cleavage at E373-A374, but the T370Q mutant rather
showed enhanced cleavage compared with wild type. Second, PNGase F treatment of
cartilage-derived aggrecan to remove N-linked oligosaccharides resulted in an increase in susceptibility at E373-A374. Lastly, our results suggest that KS is only required in the
presence of abundant CS. We must point out that Poon and co-authors heat-denatured
their aggrecan construct in the presence of SDS to remove N-linked KS by PNGase F
prior to digesting with aggrecanase (Poon et al., 2005). The heat denaturation of aggrecan
may have irreversibly changed its conformation and therefore cannot be used to assess
the in vivo susceptibility of aggrecan to digestion by ADAMTS4.
4.3.6 S377 is important for substrate recognition by ADAMTS4
The mutation of S377 to glutamine results in a significant reduction in aggrecan’s
susceptibility to ADAMTS-p68. A serine or threonine hydroxyl-amino acid residue at the
P4’ substrate position is highly conserved among aggrecans from different species, with
the exception of rat, in which it is replaced by asparagine. Serine and threonine residues
are also found in the P4’ position of the E1480-G1481 and E1666-A1667 cleavage sites
within the CS-2 domain (see Fig. 1-9), suggesting that serine and threonine or
oligosaccharides attached to these amino acid residues may be involved in substrate
recognition by ADAMTS4. Rat aggrecan is susceptible to aggrecanase cleavage at the
E373-A374 site, but the rate of rat aggrecan cleavage has not been compared to that of
other species. It remains possible that the bulky glutamine side chain may inhibit
- 228 - cleavage at E373-A374 preventing binding at the P4’ site. It will be interesting to
investigate the susceptibility of mutant aggrecans having S377 mutated to asparagine, alanine, or threonine to determine the significance of the P4’ site in enzyme-substrate recognition.
4.3.7 The potential role of the T352IQTVT357 sequence on ADAMTS4 cleavage of aggrecan
The mutagenesis studies showed that the T352Q and T355Q mutation negatively
affects cleavage at E373-A374 by ADAMTS4, whereas the T357Q mutation positively affects cleavage at E373-A374. Since O-linked glycosylation at a particular residue is influenced by neighboring residues (Gerken et al., 2004) and since three potential
O-glycosylation sites, T352, T355, and T357, are in a cluster, it is likely that any mutation within this cluster may affect glycosylation at the other residues. Therefore, we also generated a triple mutant to eliminate all the threonines in the cluster
(T352Q-T355Q-T357Q). This mutant, however, showed no difference in its susceptibility
to ADAMTS4 compared with wild-type aggrecan, which may be due to a balance
between the positive and negative effects of each threonine mutated to glutamine.
We also mutated clusters of threonines in the sequence (T352IQTVT357) to valines to
determine whether the polarity of threonine plays any role in enzyme susceptibility.
Mutating of T352, T355, and/or T357 to valine results in no significant difference in the
substrates’ susceptibility to ADAMTS4 at E373-A374 compared to that of wild-type
aggrecan. Although the introduction of valine significantly increased the hydrophobicity in this region, it is not predicted to alter the potential extended structure of
- 229 - (I351TIQTVT357). Interestingly, although the T352Q-T355Q-T357Q mutant showed no significant difference in its susceptibility to ADAMTS4 at E373-A374 compared to that of wild-type aggrecan, the T352V-T355V-T357V mutant showed enhanced cleavage at
E373-A374. It is possible that hydrophobic residues in this region of aggrecan may play a role in ADAMTS4 binding and the glycosylation of this region may regulate this binding.
In fact, sequences N-terminal to the ADAMTS4 cleavage sites within the IGD and CS-2 domain are rich in hydrophobic residues (see Fig. 1-9), suggesting that hydrophobic residues may be important for exosite interactions. Furthermore, unpublished results from
Dr. Hering’s laboratory showed that cleavage of the V356A-V361A-E362D triple mutant at E373-A374 was significantly inhibited, suggesting that the hydrophobic residues
(especially valine residues) are important for efficient cleavage. In addition, Westling and co-authors showed with mutated recombinant versican that replacing the Val residue with
Lys at P18 resulted in a reduction in ADAMTS4 cleavage suggesting the involvement of hydrophobic valine residues in ADAMTS4-substrate interactions (Westling et al., 2004).
Alternatively, the observed enhanced cleavage of the triple T352V-T355V-T357V mutant at E373-A374 may reflect the presence of a species glycosylated near the E373-A374 site, which may enhance ADAMTS4 cleavage. Interestingly, only a single band reactive to the anti-NITEGE antibody was observed in the ADAMTS4-digested T352V-T355V-T357V mutant, whereas ADAMTS4-digestion of wild-type aggrecan results in doublet bands reactive to the anti-NITEGE antibody (bands #12 and #12’). The N368Q mutant also gives a single anti-NITEGE reactive band, but it is smaller (61.5 kDa) than that of the
T352V-T355V-T357V mutant (66.5 kDa). This suggests that the T352V-T355V-T357V
- 230 -
Fig. 4-32 Summary of site-directed mutagenesis studies. Susceptibility of each mutant was categorized into 4 groups based on their initial rate of cleavage by ADAMTS4-p68. Mutants showing enhanced cleavage are shown in blue. Mutants cleaved at a rate similar to wild type are shown in black. Mutants showing reduced cleavage are shown in green. The mutant showing the most significant inhibition at E373-A374 is shown in red.
- 231 - mutant has nearly 100% occupancy of oligosaccharides at a site that is normally not
100% glycosylated in the wild type. This is a novel finding suggesting that the surrounding residues not immediately proximal to the glycosylation sites can have profound effects on the level of glycosylation at each site. This change in the glycosylation profiles could also affect susceptibility to ADAMTS4 at E373-A374.
4.3.8 Sulfation of CS by KS sulfotransferases?
In this work, we also attempted to characterize the effects of KS on ADAMTS4 cleavage by using recombinant aggrecans. However, we could not find established cell lines that could stably add KS onto recombinant molecules. In the course of searching for such appropriate cell lines, we found a report by Akama and co-authors describing the KS production in HeLa cells, which normally do not produce KS, by overexpressing keratan sulfate Gal 6-O-sulfotransferase (KSG6ST) and corneal GlcNAc 6-O-sulfotransferase
(CGn6ST), two enzymes involved in KS biosynthesis (Akama et al., 2001). Since COS-7 cells do not produce detectible amounts of KS, we attempted to generate aggrecan substituted with KS (KS-FLAG-rbAgg) in COS-7 cells by co-expressing aggrecan with these enzymes. We found that aggrecan co-expressed with KS sulfotransferases was distinctly more heterogeneous than recombinant aggrecan expressed in non-co-transfected COS-7 cells. Unexpectedly, however, chondroitinase ABC digestion of KS-FLAG-rbAgg resulted in a smaller apparent molecular mass and a more distinct band, suggesting that the observed microheterogeneity is due to CS. FLAG-rbAgg expressed without sulfotransferases showed a smaller decrease in molecular mass and
- 232 - heterogeneity. This result suggests that that COS-7 cells may be producing a CS that is
perhaps more uniform in size or charge, and possibly more resistant to chondroitinase
ABC digestion. Therefore it is apparent that KS sulfotransferases significantly altered the
CS composition from that normally produced in native COS-7 cells. The specificity of different sulfotransferases is highly variable. Chondroitin sulfate 6-O-sulfotransferase I, for example, has broad substrate specificity and can transfer sulfate to both chondroitin sulfate and keratan sulfate (Habuchi et al., 1993). Keratan sulfate Gal
6-O-sulfotransferase has been extensively characterized for its substrate specificity and it
is shown to have neither chondroitin 4-O-sulfotransferase nor chondroitin
6-O-sulfotransferase activity (Fukuta et al., 1997), which is required for sulfation of chondroitin sulfate. On the other hand, CGn6ST has not been characterized to show whether or not it possesses chondroitin sulfotransferase activity. Since our study implies that either one or both of the sulfotransferases may have chondroitin sulfotransferase activity, CGn6ST is a good candidate for such activity. Alternatively, KS substitution of the proteoglycan may affect CS biosynthesis indirectly, resulting in greater microheterogeneity of the CS component.
Fetal aggrecan, which is less susceptible to degradation by ADAMTS4 (Roughley et
al., 2003), has much longer and more densely packed CS chains compared to aggrecan
isolated from adult cartilage (Ng et al., 2003). Furthermore, a recent study in Dr. Hering’s
laboratory suggested that CS attached to calf aggrecan is more susceptible to chondroitinase ABC digestion compared with steer aggrecan (unpublished work).
Chondroitinase ABC shows the highest hydrolyase activity to hydrolyze 4-O-sulfated
- 233 - chondroitin sulfate followed next by 6-O-sulfated chondroitin sulfate, dermatan sulfate, and non-sulfated chondroitin (Prabhakar et al., 2005). Furthermore, it was shown that the optimal buffer condition for hydrolyzing CS differs between different types of CS disaccharide units (Prabhakar et al., 2005). Aggrecan isolated from younger individuals has a higher 4-O–sulfated GalNAc to 6-O-sulfated GalNAc ratio compared with aggrecan isolated from older individuals (Murata and Bjelle, 1979; Roughley and White, 1980).
Therefore, the susceptibility to chondroitinase ABC may be affected by the composition of differentially sulfated CS disaccharide units in aggrecan isolated from humans and animals of different ages. It will be important to identify the fine structures of CS on aggrecan to investigate whether the degrees of sulfation and chain lengths have any similar effects on interactions with ADAMTS4. The identification of the fine structure of
CS on recombinant aggrecan that was co-expressed with sulfotransferases
(KS-FLAG-rbAgg) may also allow for elucidation of the observed resistance to
ADAMTS4-p68 digestion. The fine structure of CS may be analyzed by fluorophore-assisted carbohydrate electrophoresis (FACE), a technique that allows the identification of chain length, oligosaccharide type, and the position of sulfate and other modifications (Calabro et al., 2000).
4.3.9 The susceptibility of KS-FLAG-rbAgg to ADAMTS4
When KS-FLAG-rbAgg was digested with ADAMTS4-p68 and p40,
KS-FLAG-rbAgg appeared to be cleaved more effectively by the p40 form than the p68 form of ADAMTS4. Especially, when the rate of cleavage was compared with that of
- 234 - FLAG-rbAgg with or without sulfotransferase-co-transfection, it was apparent that
FLAG-rbAgg was cleaved significantly slower than non-co-transfected FLAG-rbAgg by
ADAMTS4-p68, but no significant difference was observed in their susceptibility to
ADAMTS4-p40. Because of the aforementioned changes in CS in aggrecan co-expressed with KS sulfotransferases, ADAMTS4-p68 may bind to the CS region with a greater affinity, changing the equilibrium of unbound enzyme available to interact with the cleavage site in the IGD. Furthermore, since cleavage within the CS-2 domain was also inhibited in KS-FLAG-rbAgg by ADAMTS4-p68, it is possible that p68 might irreversibly bind on KS-FLAG-rbAgg so that the overall protease activity of ADAMTS4 is inhibited. On the other hand, ADAMTS4-p40 has less affinity for CS chains, since it lacks the C-terminal GAG binding motifs, and may therefore be available to bind and cleave aggrecan within the IGD, even with the altered CS substitution observed in
KS-FLAG-rbAgg. We must also point out, however, that the susceptibility of
KS-FLAG-rbAgg to p68 showed some degree of variation between different aggrecan samples isolated from different transient transfections. This is likely due to variations in co-transfection efficiency of the four constructs (rbAgg, CGn6ST, KSG6ST, and C2GnT).
We have observed that when FLAG aggrecan was co-transfected with increasing concentrations of the CGn6ST and KSG6ST constructs, the susceptibility of those aggrecans to ADAMTS4-p68 decreased in proportion to increasing concentrations of
CGn6ST and KSG6ST (data not shown). Therefore, it is likely that the variation in co-transfection efficiency would affect KS-FLAG-rbAgg’s susceptibility to
ADAMTS4-p68 as well. We should be able to obtain less variable results upon
- 235 - generation of a cell line stably expressing sulfotransferases.
Typical digestions of FLAG-rbAgg by ADAMTS4-p68 and ADAMTS4-p40
invariably produce two bands that are reactive to anti-NITEGE antibody. Interestingly,
when KS-FLAG-rbAgg is digested by p40, only the lower molecular mass band reacts
with the anti-NITEGE antibody, even though the presence of the slightly higher
molecular mass band is evident by reaction with the anti-G1 domain and anti-FLAG
antibodies (data not shown). This result suggests that the higher molecular mass band
may be substituted with oligosaccharides that mask the epitope of the anti-NITEGE
antibody. Since this masking effect is not observed in the FLAG-rbAgg not co-transfected
with sulfotransferases, it is possible that the epitope is masked with KS or other sulfated
oligosaccharides.
4.3.10 MMP13 cleavage sites within the aggrecan core protein
MMP13 is a potent collagenease that cleaves type II collagen in cartilage. It is also
well known for cleaving aggrecan (Fosang et al., 1996; Little et al., 2002), and it is
suggested to play a role in normal bone development in which cartilage ECM is degraded
and replaced with bone at the growth plate (Stickens et al., 2004). On the other hand, a
number of reports have suggested that MMPs are not involved in detrimental cartilage degeneration, since mostly the ADAMTS4 (and other aggrecanase)-mediated aggrecan fragments are found in patients with osteoarthritis and joint injuries (Arner, 2002;
Lohmander et al., 1993). The result suggests that MMP13 can extensively degrade
recombinant aggrecan by cleavage at multiple sites. It was shown that MMP13 not only
- 236 - cleaves at S341-F342 within the IGD, but also cleaves within the CS; and it appears that the cleavage within the CS domain is preferred over cleavage within the IGD. Aggrecan found in cartilage is known to be extensively processed at the C-terminus and often lacks the G3 domain. It is possible that such C-terminal processing is mediated by the action of
MMP13, or related MMPs. It will be of interest, therefore, to identify the cleavage sites within the CS domain.
4.3.11 Conclusions
Based on our study using both steer aggrecan and recombinant aggrecan expressed in various cell lines as experimental substrates for ADAMTS4 digestion, we have begun to characterize the substrate specificity of both p68 and p40 forms of ADAMTS4. We concluded that CS is essential for efficient cleavage within the CS-2 domain, and that this interaction may involve the cysteine-rich or spacer domain of ADAMTS4 for substrate recognition. We have found that KS may potentiate ADAMTS4 cleavage within the IGD, but only in the presence of abundant CS.
That the elimination of threonine residues within the IGD can enhance cleavage at
E373-A374 provides indirect evidence that the presence of non-KS O-linked oligosaccharides may be inhibitory to cleavage within the IGD, whereas the N-linked oligosaccharide at N368 apparently has a minimal effect on cleavage by ADAMTS4.
The next step will be to identify the exact glycosylation sites within the IGD to confirm this inhibitory effect. Horber and co-authors have suggested that this region may be important for efficient cleavage at E373-A374 (Horber et al., 2000). Our study
- 237 - provides further support for this hypothesis, as several residues have been identified
within the TIQTVT sequence N-terminal to the E373-A374 cleavage site that are
important for substrate recognition. In addition, it was demonstrated that at least one
residue (S377) C-terminal to the ADAMTS4 cleavage site at E373-A373 may also
interact with ADAMTS4 suggesting that the sequence C-terminal to the ADAMTS4
cleavage site also influences the efficiency of substrate recognition.
Finally, we have demonstrated that recombinant aggrecan substituted with KS can
be produced in COS7 cells when co-transfected with sulfotransferases of the KS
biosynthesis pathway. This KS-substituted recombinant aggrecan has altered
susceptibility to cleavage by ADAMTS4. The mechanism for the different susceptibility
may involve KS substitution directly, or may be the result of an unexpected increase in
the microheterogeneity of CS.
In summary, we have used a novel system for the expression of full-length
recombinant aggrecan with post-translational glycosylation and amino acid residues
modified by mutagenesis to explore cellular mechanisms for modulating
enzyme-substrate specificity of ADAMTS4. We have confirmed and extended previous
studies indicating a role for glycosaminoglycans and other oligosaccharides in protection
of the substrate, or enhancement of cleavage. Regulation of these post-translational glycosylations by the chondrocytes may not only influence the structure and function of the extracellular matrix, but may also control the rate of turnover of aggrecan during development, maintenance, and disease.
- 238 - 4.4. Experimental Procedures
4.4.1 Materials
Expression vectors for human corneal GlcNAc 6-O-sulfotransferase
(pcDNA3-hCGn6ST) and human keratan sulfate Gal 6-O-sulfotransferase
(pcDNA3-hKSG6ST) were generous gifts from Dr. Michiko Fukuda (The Burnham
Institute, La Jolla, CA) (Akama et al., 2001). An expression vector for human core 2
UDP-GlcNAc:Gal beta 1-3-GalNAc-R (GlcNAc to GalNAc) beta 1-6GlcNAc transferase
(pcDNAI-C2GTI) was a generous gift from Dr. Minoru Fukuda (The Burnham Institute,
La Jolla, CA) (Bierhuizen and Fukuda, 1992). The anti-YNHR antibody against
ADAMTS4 was a generous gift from Dr. John D. Sandy (Shriners Hospital for Children,
Tampa, FL). The anti-TFKEEE/GLGSV, anti-TAGELE/GRGTI, anti-NITEGE, and anti-VDIPEN (also reactive with VDIPES (Sztrolovics et al., 2002)) neoepitope antisera
(Sztrolovics et al., 1997) were generous gifts from Dr. John Mort (Shriners Hospital for
Children, Montreal, Canada). Anti-ARGSV (BC-3) and anti-FFGV (BC-4) antibodies were purchased from Abcam (Cambridge, MA). Swarm rat chondrosarcoma cell line was a generous gift from Dr. Jim Kimura (Henry Ford Hospital, Detroit, MI). Recombinant human ADAMTS4-p68 was a generous gift from Dr. Elisabeth A Morris (Wyeth
Research, Cambridge, MA). Recombinant poly-histidine-tagged ADAMTS4-p40 was purchased from Calbiochem (San Diego, CA). Recombinant human MMP13 was purchased from MP Biomedicals (Irvine, CA). The Quick change XL site-directed mutagenesis kit was purchased from Stratagene (La Jolla, CA). A1A1D1 (also referred to as steer aggrecan or cartilage-derived aggrecan) and recombinant aggrecan
- 239 - (FLAG-rbAgg) were isolated as described in Chapter 3. Origins of all the other materials
used in this chapter can be found in the “materials” sections of Chapters 2 and 3.
4.4.2 Site-directed mutagenesis
Various point mutations were made in the full-length aggrecan construct
pBAGG71-28 (see Chapter 3) by site-directed mutagenesis with the Quick change XL
site-directed mutagenesis kit as described in the manufacturer’s protocol. A series of
primer sets to mutagenize potential glycosylation sites into non-glycosylation sites is
described in Table 4-III. All the PCR reactions were conducted as follows by using high fidelity PfuTurbo DNA polymerase: 1 min of 95 ºC followed by 18 cycles of (50 sec of
95 ºC, 1 min of 60 ºC, 32.5 min of 68 ºC), then 7 min of 68 °C. The entire PCR reaction
was digested with 40 U of Dpn I to cleave the methylated template. DpnI-treated PCR
products (4 μl) were transformed into XL10-Gold (E. coli) and grown at 37 ºC overnight
on low salt LB ampicillin agar plates. Some mutants were screened by direct-PCR
followed by restriction enzyme digestion prior to performing DNA sequencing.
Restriction enzymes used for such screening of each mutant are described in Table 4-III.
All of the direct PCR reactions were conducted as follows by using taq DNA polymerase:
1 min of 94 ˚C followed by 30 cycles of (1 min of 94 ºC, 1 min of 59 ºC, 1 min of 72 ºC),
then 10 min of 72 ºC. All the site-directed point mutations were confirmed by DNA
sequencing.
- 240 - 4.4 3. ADAMTS4 digestion of de-glycosylated cartilage-derived steer aggrecan.
A1A1D1 steer aggrecan was digested with various combinations of deglycosylation
enzymes (chondroitinase ABC (0.01 U/0.4 pmol of aggrecan), keratanase I (0.01U/0.4
pmol of aggrecan), and keratanase II (0.0002 U/0.4 pmol of aggrecan)) in 100 mM
Tris-HCl and 100 mM sodium acetate (pH 6.5) for 4 h at 37 ºC. Control (native) aggrecan
was also mock digested at 37 ºC for 4 h without glycosidases. The buffer for these
samples was then exchanged for A1A1D1-reaction buffer (50 mM Tris-HCl (pH 7.5), 100
mM NaCl, 20 mM CaCl2). Removal of sulfated GAGs was confirmed by the Alcian Blue
sGAG assay, and removal of KS was confirmed by their reactivity to 5-D-4 (anti-KS
antibody) by immunoblot analysis as described in Chapter 3. Each aggrecan (3.2 pmol) was then digested with 8 ng of ADAMTS4-p68 in 10 µl of reaction buffer for the indicated time at 37 ºC, and the reaction was terminated by 20 mM EDTA. Each aggrecan digest was then incubated at 37 ºC for 4 h with chondroitinase ABC, keratanase, and keratanase II in 100 mM Tris-HCl and 100 mM sodium acetate (pH 6.5) with
COMPLETE TM protease inhibitor mix as described above, if they were not predigested to remove GAGs before the ADAMTS4 digestion. Therefore, no samples were digested twice with the same glycosidase. Samples were then analyzed by 4-15%
SDS-PAGE/Western blot as described below.
4.4.4Removal of N-linked oligosaccharides from cartilage-derived aggrecan.
Steer aggrecan was digested with PNGase F (1000 U/40 pmol) in G7 buffer (New
England Biolabs) for 4 h at 37 ºC to remove N-linked oligosaccharides under native
- 241 - conditions. Control native aggrecan was mock digested at 37 ˚C for 4 h in the absence of
PNGase F. The buffer for both samples was then exchanged for A1A1D1-reaction buffer
(50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 20 mMCaCl2). Each aggrecan (3.2 pmol) was
digested with 16 ng of ADAMTS4-p68 for the indicated time, and the reaction was
terminated by 20 mM EDTA. Aggrecan fragments were then de-glycosylated as
described above and analyzed by 4-15% SDS-PAGE/Western blot.
4.4.5 Secondary structure prediction
Secondary structures of wild type and mutants of the 32-mer sequence
(F342FGVGGEEDITIQTVTWPDVELPLPRNITEGEN 373) were predicted by submitting
the sequence to “JUFO: Secondary structure prediction for proteins” available at
(http://www.jens-meiler.de/jufo.html) (Meiler et al., 2001). The probability of having
unknown, helical, and extended structures was given. The result shows the probability of
assuming an extended structure. The maximum probability was set to 1.
.
4.4.6 ADAMTS4 digestion of recombinant aggrecan (FLAG-rbAgg)
Typically, recombinant aggrecan (FLAG-rbAgg) (0.8 pmol) was digested with 20 ng of ADAMTS4-p68 or ADAMTS4-p40 in rbAgg-reaction buffer (20 mM Tris-HCl (pH
7.2), 150 mM NaCl, and 5 mM CaCl2) at 37 ºC. Twenty nanograms of ADAMTS4-p68
and ADAMTS4-p40 cleaves 0.8 pmol of recombinant aggrecan in 20 μl of buffer at
similar rates within the IGD (see Fig. 4-15 A and B, lane 1-4, band #12). Digestions were
quenched by adding EDTA to a final concentration of 21 mM. The aggrecan fragments
- 242 - were digested with chondroitinase ABC (0.01 U/0.4 pmol of aggrecan) at 37 ºC for 1 h in
buffer containing 8 mM sodium acetate, 10 mM Tris/HCl (pH 8.0) with COMPLETE TM protease inhibitor mix. For some experiments, aggrecan products were further digested with keratanase II (0.0002 U/0.4 pmol of aggrecan), keratanase (0.01 U/0.4 pmol of aggrecan), and/or endo-β-galactosidase (0.0002U/0.4 pmol of aggrecan) for 2 h at 37 ºC in a buffer adjusted to contain 18 mM sodium acetate and 20 mM Tris-HCl (pH 6.5). If different conditions were used, they would be noted in the figure legend. Samples were then analyzed by 4-15% SDS-PAGE/Western blot as described below.
4.4.7 MMP13 digestion of recombinant aggrecan (FLAG-rbAgg)
Pro-MMP13 was incubated in 1 mM APMA for 2 h at 37 ºC to remove the
prodomain for enzyme activation immediately prior to use for aggrecan digestion.
Recombinant aggrecan (0.8 pmol) was digested with 20 ng of MMP13 in a buffer
containing 20 mM Tris/HCl (pH 7.2), 150 mM NaCl, and 5 mM CaCl2 at 37 ºC for the
indicated times. The reactions were quenched by adding EDTA to a final concentration of
21 mM. The aggrecan fragments were digested with chondroitinase ABC (0.01 U/0.4 pmol of aggrecan) at 37 ºC for 1 h in buffer containing 8 mM sodium acetate, 10 mM
Tris/HCl (pH 8.0) with COMPLETE TM protease inhibitor mix. For some experiments,
the reactions were quenched by adding EDTA (20 mM), 10 mM sodium acetate (pH 6.5) with COMPLETE TM protease inhibitor mix and then digested with chondroitinase ABC
(0.01 U/0.4 pmol of aggrecan). Samples were then analyzed by 4-15 %
SDS-PAGE/Western blot as described below.
- 243 - 4.4.8 3.0 % SDS-PAGE of full-size FLAG-rbAgg
FLAG-rbAgg samples were separated by 3.0 % SDS-PAGE gel (100 V, 4ºC),
transferred to PVDF membranes (22V, 4 ˚C, overnight) and immunoblotted as described in Chapter 3.
4.4.9 Western blot analysis
Samples separated on a SDS-PAGE gel were electrophoretically transferred onto a
PVDF membrane for chemiluminescent Western blot analysis as described in Chapter 3
by using the following sets of primary antibodies. In most experiments, the same
membrane was stripped and reprobed to be immunoblotted with multiple antibodies. To
detect the FLAG epitope at the N-terminus of FLAG-rbAgg, the membrane was probed with anti-FLAG (M2) antibody. To detect the G1 domain, the membrane was probed with anti-G1 (aG1-2) antibody. To detect the G3 domain, the membrane was probed with
anti-G3 (Lec7) antibody. To detect the cleavage at the E1666-A1667 bond within the
CS-2 domain, the membrane was probed with anti-TFKEEE1666/G1667LGSV polyclonal
antibody. To detect the cleavage at the E1480-G1481 bond within the CS-2 domain, the
membrane was probed with anti-TAGELE1480/G1481RGTI polyclonal antibody. To detect
the cleavage at the E373-A374 bond within the IGD, the membrane was probed with
anti-NITEGE373 antibody for detection of the N-terminal fragment and with
anti-A374RGSV antibody for detection of the C-terminal fragment. To detect
ADAMTS4-p68 and p53, the membrane was probed with anti-YNHR antibody. To detect the cleavage at the S341-F342 bond within the IGD, the MMP cleavage site, the
- 244 - membrane was probed with anti-VDIPEN(S)341 neoepitope antiserum for detection of the
N-terminal fragment and with anti-F342FGVG antibody for the detection of the
C-terminal fragment. Most of the antibodies were diluted at 1:500 except BC-3 and
BC-14, which were diluted at 1:100 and incubated with the membrane overnight at 4 ºC.
Membranes were then either incubated with anti-mouse-IgG-HRP (1/5000) or with anti-rabbit-IgG-HRP (1/5000) for 1 to 3 h. All the blots were then visualized with ECL, except for blots using BC-3, BC-14, and anti-YNHR antibodies, which were detected with ECL plus. Exceptions to this procedure are noted in the figure legends.
4.4.10 Semi-quantification of enzymatically cleaved products
The amounts of ADAMTS4- and MMP13-mediated aggrecan fragments were quantified by Scion image software (http://www.scioncorp.com/). Developed films were scanned with an Epson Perfection 4870 PHOTO scanner without any “Adjustments.” The relative density of each band was measured by using Gelplot2 macros in the
“uncalibrated OD” mode, and the baseline was arbitrarily defined to subtract the noise from the background. The numbers obtained were referred to as “Relative density” in this work. Since no calibration was made, only the Western blots performed at the same time were used for these comparisons.
4.4.11 Immunocytochemistry
COS-7, CHO-K1, and rat chondrosarcoma (RCS) cells were plated on chamber slides and grown overnight at 37 ºC with 5% CO2. COS-7, CHO-K1 cells were
- 245 - transfected using lipofectAMINE PLUS, and RCS cells were transfected by TransIT-LT1
either with pcDNA3 (no insert) or pcDNA3-CGn6ST and pcDNA3-KSG6ST as
described in the manufacturer’s protocol. Twenty-four hours after transfection, cells were rinsed with DMEM, and then serum-free fresh DMEM medium supplemented with 1%
ITS and 1 mM sodium pyruvate was added. Forty-eight hours post-transfection, cells were washed with PBS and fixed with methanol/acetone (1/1) solution for 10 min at room temperature. Cells were then blocked with 5% BSA/PBS for 30 minutes followed by incubation with anti-KS (5-D-4) antibody (1/100-1/200) diluted in 1% BSA/PBS for 1 to 2 h. Finally, cells were incubated with anti-mouse IgG-FITC (1/500) for 45 min. After cells were extensively washed with PBS, cover-slips were mounted on slides with
Vectashield/DAPI, and fluorescent signals were detected by fluorescence microscopy
(Nikon).
4.4.12 Expression of keratan sulfate in COS-7, CHO, and RCS cells
COS-7, CHO-K1 cells were plated on 100 mm plates and grown overnight at 37 ºC with 5% CO2. Cells were transiently transfected by lipofectAMINE PLUS with the
combinations of pcDNA3-CGn6ST (3 μg/100 mm plate), pcDNA3-KSG6ST (3 μg/100
mm plate), and pcDNAI-C2GnT (3 μg/100 mm plate) as described in the manufacturer’s
protocol. Twenty-four hours after transfection, 10 ml of fresh DMEM medium
supplemented with 1% ITS and 1 mM sodium pyruvate were added to the plate.
Seventy-two hours after transfection, cells were collected and membrane fractions of the
cell lysates were isolated as described (Akama et al., 2001). Briefly, cells were rinsed
- 246 - twice with 6 ml of PBS. Cells were then scraped in 6 ml of PBS and centrifuged at 1,000 x g for 5 min. The cell pellet was resuspended in 960 µl of TKMS buffer (20 mM
Tris-HCl (pH 7.6), 25 mM KCl, 2.5 mM MgCl2, 0.25 M Sucrose, 100 mg/ml of PMSF) and treated with “freeze and thaw cycles” 5 times. Cells were centrifuged at 9,000 x g for
10 min. The pellet was then resuspended in 240 µl of TKMS buffer with 1% triton-X and incubated on ice for 10 min. Cell lysates were microcentrifuged at 9,000 x g for 10 min.
The supernatant fraction is the membrane fraction of proteins. Isolated proteins were then precipitated with 8 volumes of ice-cold acetone overnight at –20 ºC. Precipitants were reconstituted in 0.5% SDS, and the protein concentration was calculated by UV absorbance at A280/A260. To remove N-glycans, 50 μg (COS-7) or 37.5 μg (CHO-K1) of membrane proteins were heat-denatured in the presence of 1% 2-mercaptoethanol and
0.5% SDS. Proteins were then digested with PNGase F (10 U/μg of protein) for 3 h at 37
ºC in G7 buffer (New England Biolabs) and 1% NP-40. Samples were then separated on a 4-15% gradient gel, electrophoretically transferred to a PVDF membrane, and immunoblotted with 5-D-4 anti KS antibody. Bands were visualized with ECL plus.
4.4.13 Co-expression of FLAG-rbAgg with sulfo- and glycosyl- transferases
To generate FLAG-rbAgg substituted with KS, COS-7 cells were transfected with
FLAG-rbAgg (10 μg/100 mm plate), CGn6ST (1 μg/100 mm plate), KSG6ST (1 μg/100 mm plate), and C2GnT (1 μg/100 mm plate) by lipofectAMINE PLUS reagents and
FLAG-rbAgg secreted into the culture medium was purified as described in Chapter 3.
Note that DNA plasmids were premixed before diluting in DMEM.
- 247 - Acknowledgements
I would like to thank Mr. Patrick Klepcyk and Ms. Diane Kocka for performing
DNA sequencing, Mrs. Lori Duesler and Mr. Tru Huynh for technical assistance for Figs.
4-2 and 4-25, respectively, and Mr. Aaron Hunyady for maintaining RCS cells.
- 248 - Chapter 5
Summary and Future Studies
5.1. General summary
In this thesis, the construction, expression, purification, and initial characterization
of recombinant link protein and recombinant aggrecan were described. Using these
recombinant constructs as model systems, the assembly of the proteoglycan ternary
complex and the degradation of aggrecan by ADAMTS4 and MMP13 were studied. Our
objective early in this project was to produce link protein domains in a recombinant
system that would be suitable for structural studies. Several expression systems in E. coli
and yeast were tested for expression of recombinant proteoglycan tandem repeat (PTR)
domains from link protein for use in structural studies to model their interaction with hyaluronan (HA). The PTR domains expressed in the E. coli/MBP expression system
gave the best results among all the systems tested, based on the fact that MBP/PTR fusion
proteins were soluble and obtained with relatively high yield (10 mg/l of culture).
Although they were misfolded due to the formation of intermolecular disulfide bonds, we
were able to obtain monomeric forms of the PTR domains by a novel refolding procedure.
Further optimization of this procedure should allow us to obtain PTR domains that could
be used for structural and functional studies.
Since further optimization was required to obtain PTR domains that would be useful
for both structural and functional studies, recombinant full-length cartilage link protein
was expressed in COS-7 cells to mainly focus on conducting functional studies. The
majority of the secreted link protein was in soluble and monomeric form. Recombinant
- 249 - link protein was functional in that it bound HA, a property consistent with the protein being properly folded. Similarly, recombinant full-length aggrecan expressed in mammalian cell lines was functional and interacted with HA to form proteoglycan aggregates. The presence of recombinant link protein further enhanced the interaction between aggrecan and HA. These results suggest that these recombinant molecules could be used to model the interactions between aggrecan and HA, link protein and HA, and aggrecan and link protein. Furthermore, differentially glycosylated recombinant aggrecan was obtained by expressing the recombinant construct in various mammalian cell lines. It was suggested that each cell line added oligosaccharides to recombinant aggrecan to different extents. This approach should be useful for elucidating the function of
oligosaccharides covalently attached to aggrecan core protein by further characterization of specific oligosaccharides substituted in each cell line.
Recombinant aggrecan can be cleaved by ADAMTS4 and MMP13, which are two
major proteases known to effectively cleave the aggrecan core protein. By using both
cartilage-derived and recombinant aggrecan, it was demonstrated that the
glycosaminoglycan side chains play important roles in regulating the substrate specificity
of ADAMTS4. It was also demonstrated that CS substitution on aggrecan is essential for
efficient cleavage within the CS-2 domain by ADAMTS4-p68. It is likely that p68
recognizes CS by its spacer or cysteine-rich domain, since upon their removal (forming
ADAMTS4-p40) this activity is lost. On the other hand, although it was shown that KS
substitution potentiates the cleavage within the IGD in cartilage-derived aggrecan, KS is
not essential in the absence of CS. Recombinant aggrecans with no KS and apparently
- 250 - fewer CS chains were efficiently cleaved within the IGD.
The mutagenesis study showed that the elimination of potential O-linked
oligosaccharide substitution sites (T357, T370, and T381) enhanced cleavage at
E373-A374 by ADAMTS4-p68, suggesting that in the presence of these oligosaccharides
cleavage may be inhibited by steric hindrance. On the other hand, it was indicated that
S377 is required for efficient cleavage at E373-A374. It was also suggested that the extended secondary structure and/or hydrophobicity of amino acids at the N-terminal to the ADAMTS4 cleavage site (i.e., T352IQTVT357) may be important for ADAMTS4
substrate recognition. The triple mutant having the (Q352IQQVQ357) sequence instead of
the wild-type (T352IQTVT357) did not have significantly different susceptibility to
ADAMTS4 at E373-A374 compared with the wild type. In contrast, the other triple
mutant having the (V352IQVVV357) sequence may be recognized by ADAMTS4 more
efficiently, since this mutant showed enhanced cleavage at E373-A374.
We attempted to produce recombinant aggrecan substituted with KS by
co-expressing aggrecan with sulfotransferases (corneal GlcNAc 6-O-sulfotransferase and
keratan sulfate 6-O-sulfotransferase) and core 2 GlcNAc transferase involved in KS
biosynthesis. It was shown that these co-transfected COS-7 cells mainly produce
N-linked KS with small amounts of O-linked KS. This may be due to a lack of the
endogenous enzymes required for O-linked KS biosynthesis other than sulfotransferases
and core 2 GlcNAc transferase. Aggrecan co-expressed with sulfo- and core 2 GlcNAc
transferases was found to be substituted with KS. This is of major interest, since KS
substitution of proteoglycans expressed in well-established cell lines has not yet been
- 251 - reported. Unexpectedly, it was also found that the expressed aggrecan became distinctly more heterogeneous, appearing as a broad high molecular mass smear on a 3.0%
SDS-PAGE gel. This microheterogeneity could be eliminated by chondroitinase ABC digestion, resulting in a more well-defined band. This microheterogeneity is clearly related to CS but not KS, and may be due to variable changes in length and/or sulfation.
This latter may result from the sulfation of CS by the same sulfotransferases involved in
KS biosynthesis, thus producing CS with altered sulfate composition.
In summary, by using the recombinant model system, we obtained functional recombinant link protein and recombinant aggrecan that can be used to model proteoglycan aggregate interactions, aggrecan glycosylation, and aggrecan degradation mediated by ADAMTS4 and MMP13. Below, we proposed a few future studies to expand the findings presented in this work.
5.2. Future studies
-Proteoglycan aggregate interactions-
5.2.1 Refolding PTR1+2 domains from E. coli
In Chapter 2, we described a novel refolding procedure that was employed to refold
PTR1+2 domains, which were expressed in E. coli and formed high molecular mass aggregates due to inappropriate intermolecular disulfide bonds, into apparent monomers.
Since the PTR1+2 domains contain a total of eight cysteine residues, however, it is possible that incorrectly matched intramolecular disulfide bonds are formed. Future studies will be aimed at isolating correctly folded monomeric PTR1+2 by HPLC as
- 252 - described by Day and co-authors who isolated correctly refolded E coli-expressed TSG-6,
which has 2 disulfide bonds (Day et al., 1996). Day and co-authors isolated several peaks
eluted from HPLC and performed trypsin digestion. Digested fragments were then subjected to HPLC, and the elutants were analyzed by N-terminal sequencing to
determine the peptide fragments co-eluted in each fraction. The disulfide bond linkage
can be mapped, because the co-eluted fragments are covalently bound to each other by the disulfide bond(s) (Day et al., 1996). We will use this method to isolate monomeric
PTR 1+2 domains having correct disulfide bonds (PTR1: C181-C252 and C205-C226,
PTR2: C279-C349 and C304 and C325) (Neame et al., 1986). Once we have established
a purification protocol to obtain properly folded PTR1+2 domains, we will start to label
the protein with 13C and 15N to study the solution structure of PTR1+2 domains by NMR.
5.2.2 Functional characterization of the cartilage link protein-HA interaction
In the present work, it was shown that both link protein and aggrecan expressed in
COS-7 cells were able to form proteoglycan aggregates with HA, thus demonstrating
their functional HA-binding properties. Unlike TSG-6 and CD44, each of which has only
one PTR domain, both link protein and aggrecan have paired PTR domains. It has
recently been suggested that proteins having a single PTR or paired PTR domains have
different mechanisms for interacting with HA (Rauch et al., 2004). Rauch and co-authors
have mutated two arginine residues (Arg169 and Arg269) in the two PTR domains of
neurocan. These residues are highly conserved between the single and paired PTR
domains and are also involved in the HA interaction with TSG-6 and CD44. In most PTR
- 253 - domains, the amino acid residues at these locations are either positively charged arginines
or lysines. Mutations of the lysine in TSG-6 (Lys11) (Mahoney et al., 2001) and arginine
in CD44 (Arg41) (Bajorath et al., 1998) resulted in a significant reduction of HA-binding
compared with that of the wild type, suggesting that they are involved in HA binding.
However, the mutation of arginine residues in these positions of the PTR1 (Arg169) and
PTR2 (Arg269) domains from neurocan showed no reduction in HA binding (Rauch et al.,
2004). This suggests that neurocan and other members of the lectican and link protein
families with two PTR domains may have different HA binding characteristics from those
of TSG-6 and CD44 (Bajorath et al., 1998; Mahoney et al., 2001).
By using the truncated G1 domain of aggrecan mutants as an experimental model,
Watanabe and co-authors have suggested that the Ig-fold domain is also involved in
HA-binding by showing that the truncated G1 domain lacking the Ig-fold domain has weaker affinity to HA compared with the full-length G1 domain (IG-fold, PTR1+2 domains) (Watanabe et al., 1997). Since link protein and the G1 domain of aggrecan are highly homologous, the Ig-fold of link protein may also be involved in HA binding.
Based on these observations, we hypothesize that the different amino acid residues, which are conserved among link protein and lectican families but not conserved in TSG-6 and CD44, may also be involved in HA-binding. We will determine which residues are involved in HA binding of link protein. We can systematically mutate the conserved polar and charged residues that may be involved in HA binding, beginning with the link protein expressed in COS-7 cells that exhibits functional HA binding as described in Chapter 3.
The candidate residues that may be mutated are indicated with asterisks in Fig. 5-1. Since
- 254 - Watanabe and co-authors suggested the involvement of Trp75 in HA binding, we will
especially focus on the involvement of the Ig-fold domain in HA binding by mutating
Trp75 and its surrounding positively charged residues and investigating its HA binding affinity with full-length link protein (Watanabe et al., 1997). In addition, we will mutate positively charge residues conserved among link protein and lectican families. We will also investigate the residues that may be involved in the HA binding of link protein based on mutagenesis studies conducted on TSG-6 and CD44, both of which have a single PTR domain (Bajorath et al., 1998; Mahoney et al., 2001). These workers have generated a homology model of the first and second PTR domains of cartilage link protein based on their solution structure of human TSG-6 and have predicted potential HA binding sites and residues that may be involved in HA binding. These residues may also be mutated to investigate whether a single PTR domain or paired PTR domains differ significantly in their HA binding characteristics.
- 255 -
Fig. 5-1 Potential residues in bovine cartilage link protein that may be involved in HA-binding. Mahoney and co-authors have proposed that the highlighted (in gray) residues may be involved in HA binding (Mahoney et al., 2001), including the boxed Arg169 and Lys269 (Mahoney et al., 2001), by using TSG-6 as a model. However, the same conserved positively charged residues Arg169 and Arg269 are not involved in HA binding of neurocan (Rauch et al., 2004). The underlined residues are additional ones that we propose may be involved in HA binding owing to their polarity and charge. Trp75 in the Ig-fold domain of aggrecan is involved in HA binding (Watanabe et al., 1997). This residue is also conserved in link protein and may be involved in HA binding. These findings suggest that neighboring positively charged residues may also be involved in HA binding. (Note that the initiation Met is designated as residue number 1.)
- 256 - 5.2.3 Effect of glycosylation on HA binding of the G1 domain of aggrecan and link
protein
Watanabe and co-authors described that the deglycosylated G1 domain of aggrecan
had a lower affinity for HA than did the intact glycosylated G1 domain (Watanabe et al.,
1997). According to this report, when T42 was mutated to alanine, the mutant had significantly lower affinity for HA. T42 is found to be O-glycosylated with KS in steer aggrecan but is not glycosylated in calf aggrecan (Barry et al., 1995). Therefore, age-dependent differences in the affinity of the G1 domain for HA may be related to differential glycosylation within the G1 domain. As described in Chapter 1, aggrecan undergoes conformational maturation to acquire optimal HA binding (Oegema, 1980). In another system, it has been shown that covalently attached oligosaccharides can affect the protein conformation (Chen et al., 2002; Krishnan et al., 1999). N-oligosaccharides are also suggested to play a role in protein folding (Helenius and Aebi, 2004). Thus, oligosaccharides attached to the aggrecan core protein may also affect the secondary structure of aggrecan. We hypothesize that the conformational maturation of aggrecan
(Oegema, 1980) is affected by glycosylation within the G1 domain that would alter its
HA binding property. We therefore will compare the HA binding of recombinant wild-type and mutant (T42Q) aggrecans expressed in COS-7 cells. Watanabe and co-authors produced a double mutant (T42A; T43A) that lacks both potential O-linked glycosylation sites. Although this mutant has significantly lower affinity for HA, it does not specify the amino acid that affects HA binding. Furthermore, since polar threonine residues were mutated to hydrophobic alanine residues, the possibility remains that
- 257 - changes in polarity are responsible for the altered HA binding rather than the
oligosaccharides attached onto these threonine residues. We have generated a T42Q
mutant that specifically lacks the O-linked KS site but still retains its polarity. We will
compare the HA binding affinity of the wild-type and T42Q mutant aggrecans. Since
Watabnabe and co-authors also suggested that the total de-glycosylation of the G1
domain resulted in a reduction of HA binding, we will also investigate the roles of three
N-linked glycosylation sites (Asn108, Asn220, and Asn314) found in the G1 domain of
bovine aggrecan on HA binding. Each of these residues would be mutated to glutamine
and the change in its HA binding investigated.
We have also observed slight changes in the binding to HA of link protein that is
differentially N-linked glycosylated (LP1 vs. LP2). Although biotinylated HA bound to
LP1 and LP2 with apparently equal affinity, LP2 appears to bind more efficiently to
HA-Sepharose since the LP2 form of link protein is slightly enriched after total link
protein is purified from HA-Sepharose affinity chromatography (see Fig. 3-10). This
suggests that the additional N-linked oligosaccharide (at Asn21) present only in LP1 may
interfere with HA binding. Thus, we will investigate the affinity to HA of LP having
different numbers of N-linked oligosaccharides. LP1 and LP2 can be separated by lectin
(wheat germ agglutinin) affinity chromatography (Choi et al., 1985). Upon separation of these two forms of LP, we will conduct HA binding assays by using BIACORE or other types of binding assays to investigate the differences in their affinity for HA.
- 258 - -Aggrecan degradation-
5.2.4 The presence of chondroitin sulfate and keratan sulfate on aggrecan core protein
affects the substrate specificity of ADAMTS4 isoforms
It is suggested from the work described in Chapter 4 that CS is required for efficient
cleavage by ADAMTS4-p68 within the CS-2 domain. KS, on the other hand, potentiates
cleavage within the IGD, but it is not absolutely required. In addition, the ADAMTS4
isoform p40 also appeared to be significantly affected by the presence of GAGs, since
cartilage-derived steer aggrecan has very low susceptibility to ADAMTS4-p40, whereas
recombinant aggrecan was cleaved effectively within the IGD at E373-A374 but not in
the CS-2 domain. To test this possibility, we would enzymatically remove CS and/or KS
from cartilage-derived aggrecan and study the substrate specificity of ADAMTS4-p40.
Our results suggest that ADAMTS4 recognizes CS within the CS-2 domain via
GAG-binding domains within the cysteine-rich or spacer domains that are lacking in
ADAMTS4-p40. We would speculate that the removal of CS would not significantly
affect the substrate specificity of ADAMTS4-p40 within the CS-2 domain; however, the removal of KS and/or CS may affect cleavage within the IGD, since recombinant aggrecan that is poorly substituted with CS and KS can be cleaved within the IGD.
The degree and site of CS sulfation may also play a role in regulating the
susceptibility to ADAMTS4 since calf and steer are both substituted with CS but appear
to have different susceptibility within the CS-2 domain (Roughley et al., 2003). The
pattern of CS sulfation changes in an age-dependent manner where older aggrecan has a
higher degree of GalNAc sulfated at the 6 position compared to that sulfated at the 4
- 259 - position. It is interesting to speculate that the sulfation site might also strongly influence
the susceptibility of aggrecan to ADAMTS4. We will investigate this possibility by
analyzing the fine structure of CS isolated from steer and calf aggrecan as well as the
recombinant aggrecan isolated from different cell lines and by studying the pattern of CS
sulfation on aggrecan and its correlation to ADAMTS4 susceptibility. Initially, we will generate recombinant aggrecan having CS with no sulfation by growing cells in chlorate
(Humphries et al., 1989), which is known to inhibit sulfation on CS. It will be interesting to characterize the substrate specificity of ADAMTS4 to sulfation-free aggrecan to determine the role of sulfation on degradation by ADAMTS4.
5.2.5 Characterization of glycosylation in the IGD of recombinant aggrecan expressed in
COS-7 cells
The effects of glycosylation sites on recombinant aggrecan cleavage by ADAMTS4
were discussed in Chapter 4. Mutagenized aggrecan lacking potential glycosylation sites
in the IGD exhibited positively and negatively altered susceptibility to ADAMTS4 (see
Chapter 4 summarized in Fig. 4-32). Efforts to determine the glycosylation sites of
COS-7-expressed aggrecan are ongoing in Dr. Hering’s laboratory. To identify the sites substituted with oligosaccharides in the sequences (F342 ----- E373) located between the
MMP13 and the ADAMTS4 cleavage sites, the recombinant aggrecan was digested with
MMP13, and peptides carrying F342 at the N-terminus were isolated by anti-F342FGV
(BC-14) (Abcam, Cambridge, MA) antibody-immobilized Sepharose affinity chromatography. The amount of peptide isolated by this procedure, however, was not
- 260 - sufficient to perform amino acid sequencing by Edman degradation for identification of glycosylated sites (signals of the glycosylated amino acids will diminish or shift, allowing identification of the glycosylated position). To resolve this problem we will modify two steps in this protocol. At first, we will increase the amount of starting material (full-length recombinant aggrecan) to be digested with MMP13. Second, after the digested materials have been bound to BC-14-immobilized Sepharose, more stringent elution will be used to elute protein fragments remaining bound to the antibody. In the original protocol, 0.1 M glycine (pH 3.5) or 0.1 M acetic acid was used to elute the fragments bound to BC-14. Since the BC-14 antibody is covalently immobilized to the
Sepharose, however, the co-elution of antibody from the Sepharose is not a concern.
Alternatively, we will use a synthetic peptide that would have higher affinity to the
BC-14 antibody or SDS-PAGE buffer to elute the BC-14-bound fragments. The elutants will be exchanged to a buffer compatible with amino acid sequencing by dialysis. This will permit the unambiguous identification of residues within the IGD that are glycosylated in recombinant aggrecan.
5.2.6 The role of Ser 377 in ADAMTS4 recognition
The S377Q mutant was significantly less susceptible to cleavage by both
ADAMTS4-p68 and p40 compared with the wild-type and other mutant aggrecans. This result suggests that S377 may interact with ADAMTS4 via either the catalytic domain, disintegrin, or the thrombospondin motifs that are present in the p40 form. As discussed in Chapter 4, S377 is replaced with asparagine in rat aggrecan that can be cleaved at
- 261 - E373-A374 (Gao et al., 2002). Since glutamine and asparagine differ only in the lengths
of their side chains (see Fig. 4-18), it is possible that the more bulky side chain of
glutamine may interfere with the substrate-enzyme interaction. We will mutate this
residue to asparagine in recombinant bovine aggrecan and characterize its susceptibility
to ADAMTS4. It is also possible that Ser377 may be glycosylated and that the
oligosaccharide is required for efficient cleavage at E373-A374. We will mutate Ser to
Thr, which is a better substrate for O-linked glycosylation, and characterize the
ADAMTS4 susceptibility. Mutation of Ser to Thr would also conserve the hydroxyl
group that may be involved in the interaction with ADAMTS4. The presence of
glycosylation would be analyzed by amino acid sequencing as described in the previous
section (#5.2.5), except that the recombinant aggrecan will be digested with ADAMTS4
and the N-terminal of A374RGSV will be sequenced. Peptides will be isolated by
anti-A374RGSV (BC-3) (Abcam, Cambridge, MA) antibody-immobilized Sepharose
affinity chromatography. Furthermore, we will also mutate the neighboring residues to
characterize their susceptibility to ADAMTS4.
5.2.7 The requirement for clusters of hydrophobic residues N-terminal to the ADAMTS4
cleavage site
The mutagenesis studies described in this thesis and those conducted in Dr. Hering’s laboratory suggest that the hydrophobic residues located N-terminal to the ADAMTS4 cleavage site may be required for efficient cleavage by ADAMTS4, since the removal of valines (V356A-V361A-E362D triple mutant) yields significant inhibition (unpublished
- 262 - work), whereas the introduction of valines (T352V-T355V-T357V triple mutant) results in enhanced cleavage at the E373-A374 site. In addition, Westling and co-authors also suggested using versican as a substrate to study the involvement of Val at P18 on
ADAMTS4 substrate recognition (Westling et al., 2004). At least two mechanisms may be considered to explain such observations.
First, valine residues appear to play a significant role in maintaining the stretched
structure N-terminal to the ADAMTS4 cleavage site, based on the results obtained by the
secondary structure prediction. Therefore, it is possible that ADAMTS4 requires an
extended structure in this region for efficient substrate recognition. To investigate this
possibility, we will introduce amino acids such as proline into the middle of the
T352IQTVT357 sequence that can disrupt the secondary structure in this region and
observe its susceptibility to ADAMTS4.
Second, it is also possible that hydrophobic residues are important for efficient
recognition by ADAMTS4 via hydrophobic interaction. The intensity of the hydrophobic
interaction may be regulated by the threonine residues that can also be glycosylated. As
described in Fig. 1-9, each sequence located N-terminal to the four different cleavage sites within the CS-2 domain is also highly conserved to contain threonines, serines, and hydrophobic residues such as valines, leucines, and isoleucines. The well-balanced distribution of these residues may be important for substrate recognition by ADAMTS4 via exosite interaction. We will mutate these residues to either remove or introduce hydrophobic residues to determine the change in their susceptibility to ADAMTS4 compared with that of the wild type.
- 263 - 5.2.8 Keratan sulfate biosynthesis
In the present work, we have attempted to introduce KS into COS-7 cells that do not normally produce detectible amounts of KS (reactive with anti-KS 5-D-4 antibody). In this system, however, most of the KS produced in these cells appears to be N-linked.
Since over 90% of the KS attached to the native aggrecan core protein is O-linked, it will be desirable to further investigate the biosynthesis of O-linked KS relative to recombinant aggrecan and the cells in which it is expressed. The initiation of O-linked KS is catalyzed by polypeptide N-acetylgalactosaminyl transferases (ppGalNAcTs), which attach
GalNAc to Thr or Ser. To date, 17 ppGalNAcTs have been identified in humans (protein database search) and it has been suggested that each ppGalNAcT may possess a different range of substrate (polypeptide) specificities. Unfortunately, neither the consensus sequence for the KS attachment, nor the ppGalNAcTs responsible for KS biosynthesis in cartilage are presently known.
We have searched the existing human EST libraries (dbEST Library ID.8940
(normal) and ID.8936, ID.10848, ID.10412 (osteoarthritic)) and found relatively high
expression levels of ppGalNAcT-15 (ppGalNAcT15) in both normal (8940) and
osteoarthritic (8936) adult human cartilage. In addition to ppGalNAcT15, relatively low
expression levels of ppGalNAcT-like4 (in ID.8936), ppGalNAcT2 (in ID.8940 and ID.
10848), ppGalNAcT11 (in ID.8940 and ID. 10848), and ppGalNAcT10 (in ID.10412 and
ID. 10848) are evident in cartilage tissues. On the other hand, the EST library constructed
from fetal bovine cartilage (ID.5532) only contains ppGalNAcT1, which is not found in
any of the EST libraries constructed from human adult cartilage. Since many O-linked
- 264 - oligosaccharides produced in cartilage are KS chains attached to aggrecan core protein, we suggest that one or more of these ppGalNAc transferases expressed in adult cartilage may be involved in KS biosynthesis. Furthermore, the expression levels of these enzymes may regulate the age-dependent change in KS biosynthesis observed in cartilage, which may be related to the development of osteoarthritis.
We propose to determine the substrate specificity of ppGalNAc transferases
expressed in cartilage. We hypothesize that each ppGalNAcT expressed in cartilage has a
unique substrate specificity and that some of them are involved in KS biosynthesis. To test this hypothesis, we will use two types of synthetic peptides to conduct in vitro
enzymatic glycosylation by purified soluble ppGalNAcTs. The first type is a set of
synthetic random peptides that have a single Thr or Ser surrounded by random sequences
of amino acids that are commonly found in the IGD of aggrecan (see Fig. 1-9). These peptides will be used to identify the consensus sequence necessary for the addition of
GalNAc by these enzymes. Currently, the substrate specificities of ppGalNAcT1, ppGalNAcT2, and ppGalNAcT10 are being investigated in Dr. Gerken’s laboratory by this approach. We will also obtain other enzymes (ppGalNAcT11, ppGalNAcT2, and ppGalNAcT-like4) that are expressed in cartilage from our collaborators. We will then attempt to deduce the putative aggrecan O-linked sites, based on the specificity of these enzymes. As a second approach, synthetic peptides of the sequences found in the IGD with putative KS sites will be used as substrates for experiments similar to those described above. The significance of this study is that we will be able to experimentally deduce the putative KS sites in aggrecan, which may affect the rate of aggrecan cleavage
- 265 - in vivo. As described in Chapter 1, it has been shown that the bovine and porcine IGD
(FFGVGGEEDITIQTVTWPDVELPLPRNITEGE) contains at least 2 to 3 O-linked KS sites (shown in bold) and bovine IGD shows age-dependent variation in its glycosylation
(Barry et al., 1995). Therefore, it is a priority to identify the enzymes capable of initiating
O-linked KS synthesis on this peptide. Furthermore, the involvement of these enzymes in
KS synthesis on aggrecan will also be tested in vivo by the co-expression of ppGalNAcTs with aggrecan in commonly used mammalian cell lines, such as COS-7 cells. We described in Chapter 4 that the recombinant aggrecan expressed in COS-7 cells is poorly substituted with O-linked KS oligosaccharides. We will test whether the co-expression of aggrecan with ppGalNAcTs would significantly increase the O-linked oligosaccharides on aggrecan.
We will also determine the expression level of ppGalNAcT and the isoforms expressed in cartilage isolated from different human or bovine developmental stages. We hypothesize that the specific ppGalNAcTs involved in KS biosynthesis are upregulated in the older chondrocytes and expressed at low levels in fetal chondrocytes. We also hypothesize that these ppGalNAcTs are co-expressed with aggrecan. To test this hypothesis we will conduct real-time PCR to determine the type and level of ppGalNAcTs expression. Co-localization of these enzymes with aggrecan will also be investigated by in situ hybridization.
- 266 - 5.2.9 Substrate specificity of sulfotransferases
As described above, we attempted to produce KS-substituted aggrecan in COS-7
cells by co-expressing aggrecan with two KS sulfotransferases and core 2 GlcNAc
transferase. Our preliminary data suggested, however, that the sulfotransferases might
also catalyze the sulfate transfer to CS. It has been reported that some sulfotransferases
can add sulfate to both KS and CS (Habuchi et al., 1993), although there are no reports on
chondroitin sulfotransferase activity of corneal GlcNAc 6-O-sulfotransferase (CGn6ST).
Keratan sulfate Gal 6-O-sulfotransferase (KSG6ST), on the other hand, was shown to
have no chondroitin sulfotransferase activity (Fukuta et al., 1997). Therefore, it is possible that CGn6ST can catalyze the transfer of sulfate on CS. Chondroitin disaccharide units can be sulfated either at the 4-O or 6-O position of GalNAc. From the stereochemical point of view, CGn6ST may be able to transfer sulfate on the 6-O position of GalNAc in CS. Since the 6-O-sulfated GalNAc content within the CS-chains increases with age, which also coincides with increasing KS in adult aggrecan, it is interesting to speculate that such changes may relate to an increased activity of CGn6ST. Thus it would be interesting to investigate the expression level of GlcNAc 6-O-sulfotransferase in cartilage and also its chondroitin 6-O-sulfotransferase activity. This can be done by the in vitro sulfation assay described by Fukuta, Habuchi, and co-authors, who characterized the substrate specificity of chondroitin 6-O-sulfotransferase I and keratan sulfate Gal 6-O sulfotransferase (Fukuta et al., 1997; Habuchi et al., 1993).
- 267 - 5.2.10 Analysis of glycosaminoglycan-microstructure by FACE
The present work and that by others suggest that the presence or absence of
particular glycosaminoglycans covalently attached to the aggrecan core protein
significantly influence their susceptibility to ADAMTS4 (Kashiwagi et al., 2004; Pratta et
al., 2000; Roughley et al., 2003; Tortorella et al., 2000). It is also speculated that the
difference in microstructure of each glycosaminoglycan chain may differentially
influence the susceptibility of aggrecan to ADAMTS4. For example, although both
human and bovine aggrecans are substituted with CS, human aggrecan is relatively resistant to cleavage within the CS-2 domain compared with bovine aggrecan (Roughley et al., 2003). This may be due to differences in the fine structure of CS. The age and
species-specific structural differences observed in KS microstructure may also contribute
to differences in their susceptibility to ADAMTS4 (Brown et al., 1998). Fosang and
co-authors (Fosang et al., 2004) reported that KS present in the IGD shows significant
differences in monosaccharide composition (e.g., fucosylation, sialylation, etc.), length,
and degree of sulfation from that within the KS domain.
In the present work, we observed that recombinant aggrecan expressed in COS-7
cells overexpressing KS-sulfotransferases was resistance to cleavage by ADAMTS4-p68.
Although it was apparent that both CS and KS biosynthesis were affected by
KS-sulfotransferases, the mechanism remains unclear, especially for the observed
increase in CS microheterogeneity. The apparent increase in KS may be due to
coordinated elongation and sulfation of GlcNAc in KS by CGn6ST, in combination with
the increase in sulfation of polylactosamine by both hCGn6ST and hKSG6ST and
- 268 - resulting in reactivity with the KS-specific 5-D-4 antibody. Since changes occurred in
both GAG types, and we have shown that CS as well as KS may modulate ADAMTS4
activity, it was difficult to identify the specific factor that was responsible for resistance
to cleavage by ADAMTS4-p68 (see Fig. 4-30). To better understand GAG-related
inhibition of cleavage, a more thorough analysis of the GAG chains is required. Therefore,
we propose to conduct fluorophore-assisted carbohydrate electrophoresis (FACE)
analysis, which allows identification of oligosaccharide composition, chain length, and
degree of sulfation on each oligosaccharide chain. FACE analysis has been conducted
with both CS and KS (Calabro et al., 2000; Calabro et al., 2001; Plaas et al., 2001), and,
therefore, is the method of choice for this investigation. The elucidation of the fine structure of GAGs covalently attached to recombinant aggrecan expressed in sulfotransferase-expressing cells as well as cartilage-derived aggrecan should help to understand the molecular characteristics that contribute to glycosylation-dependent differences in aggrecan’s susceptibility to ADAMTS4.
5.2.11 Susceptibility of deglycosylated and recombinant mutant aggrecans to ADAMTS5
ADAMTS5, also known as aggrecanase-2 (Abbaszade et al., 1999) was identified as
having aggrecanase activity after ADAMTS4 (aggrecanase-1) was first discovered as an enzyme with aggrecanase activity (Tortorella et al., 1999). The active forms (excluding the prodomain) of ADAMTS4 and ADAMTS5 show significant sequence homology, especially in the catalytic, disintegrin, and thrombospondin motifs (Fig. 5-2). Both
ADAMTS4 and ADAMTS5 have common cleavage sites in aggrecan (i.e., E373-A374,
- 269 - E1480-G1481, E1666-A1667, E1771-A1772, E1871-L1872). ADAMTS5 also cleaves at a unique site that is not cleaved by ADAMTS4, located between G1481 and E1666 of aggrecan (Tortorella et al., 2002).
Initially, it was believed that ADAMTS4 was the dominant aggrecanase involved in detrimental cartilage degeneration in osteoarthritis, since its expression is upregulated by pro-inflammatory cytokines, whereas ADAMTS5 is expressed constitutively (Arner,
2002; Yamanishi et al., 2002). Recently, however, two groups reported that ADAMTS5 is the major enzyme responsible for causing osteoarthritis in a murine model (Glasson et al.,
2005; Stanton et al., 2005). Although it is not known at this time whether ADAMTS5 plays a similar role in humans, it will be important to understand the substrate specificity of ADAMTS5. Interestingly, although aggrecan isolated from older animals or humans is more susceptible to ADAMTS4 than that from younger individuals, ADAMTS5 shows less variation with age in its substrate specificity (Roughley et al., 2003). In the work described in Chapter 4, it was apparent that the presence of GAGs significantly affected the preferred cleavage sites within the aggrecan core protein by ADAMTS4; namely, CS promotes cleavage within the CS-2 domain, whereas KS promotes cleavage within the
IGD. Therefore, we would investigate whether ADAMTS5 activity is also affected by the presence of GAGs on the aggrecan core protein. Furthermore, our mutant aggrecans had altered susceptibility to ADAMTS4. Therefore, it is worth investigating whether these mutants show susceptibility to ADAMTS5 similar to those observed using ADAMTS4.
- 270 -
Fig. 5-2 Alignment of active forms of ADAMTS4 and ADAMTS5. Conserved residues are highlighted in blue and similar residues are highlighted in gray. Sequence domains are identified by letter color: catalytic domain, orange; disintegrin domain, blue; thrombospondin domain, pink; cysteine-rich domain, green; spacer domain, purple; and TSP-motif, moss-green. Underlined sequences are putative GAG binding sites (Flannery et al., 2002; Tortorella et al., 2000).
- 271 - Bibliography
Abbaszade, I., R.Q. Liu, F. Yang, S.A. Rosenfeld, O.H. Ross, J.R. Link, D.M. Ellis, M.D.
Tortorella, M.A. Pratta, J.M. Hollis, R. Wynn, J.L. Duke, H.J. George, M.C.
Hillman, Jr., K. Murphy, B.H. Wiswall, R.A. Copeland, C.P. Decicco, R. Bruckner,
H. Nagase, Y. Itoh, R.C. Newton, R.L. Magolda, J.M. Trzaskos, and T.C. Burn.
1999. Cloning and characterization of ADAMTS11, an aggrecanase from the
ADAMTS family. J Biol Chem. 274:23443-50.
Akama, T.O., J. Nakayama, K. Nishida, N. Hiraoka, M. Suzuki, J. McAuliffe, O.
Hindsgaul, M. Fukuda, and M.N. Fukuda. 2001. Human corneal GlcNac
6-O-sulfotransferase and mouse intestinal GlcNac 6-O-sulfotransferase both
produce keratan sulfate. J Biol Chem. 276:16271-8.
Arner, E.C. 2002. Aggrecanase-mediated cartilage degradation. Curr Opin Pharmacol.
2:322-9.
Bajorath, J., B. Greenfield, S.B. Munro, A.J. Day, and A. Aruffo. 1998. Identification of
CD44 residues important for hyaluronan binding and delineation of the binding
site. J Biol Chem. 273:338-43.
Banerji, S., J. Ni, S.X. Wang, S. Clasper, J. Su, R. Tammi, M. Jones, and D.G. Jackson.
1999. LYVE-1, a new homologue of the CD44 glycoprotein, is a lymph-specific
receptor for hyaluronan. J Cell Biol. 144:789-801.
Barry, F.P., J.U. Gaw, C.N. Young, and P.J. Neame. 1992. Hyaluronan-binding region of
aggrecan from pig laryngeal cartilage. Amino acid sequence, analysis of N-linked
oligosaccharides and location of the keratan sulphate. Biochem J. 286 (Pt
- 272 - 3):761-9.
Barry, F.P., L.C. Rosenberg, J.U. Gaw, J.U. Gaw, T.J. Koob, and P.J. Neame. 1995. N-
and O-linked keratan sulfate on the hyaluronan binding region of aggrecan from
mature and immature bovine cartilage. J Biol Chem. 270:20516-24.
Bartolazzi, A., A. Nocks, A. Aruffo, F. Spring, and I. Stamenkovic. 1996. Glycosylation
of CD44 is implicated in CD44-mediated cell adhesion to hyaluronan. J Cell Biol.
132:1199-208.
Bayliss, M.T., S. Howat, C. Davidson, and J. Dudhia. 2000. The organization of aggrecan
in human articular cartilage. Evidence for age-related changes in the rate of
aggregation of newly synthesized molecules. J Biol Chem. 275:6321-7.
Beales, M.P., J.L. Funderburgh, J.V. Jester, and J.R. Hassell. 1999. Proteoglycan
synthesis by bovine keratocytes and corneal fibroblasts: maintenance of the
keratocyte phenotype in culture. Invest Ophthalmol Vis Sci. 40:1658-63.
Bekku, Y., W.D. Su, S. Hirakawa, R. Fassler, A. Ohtsuka, J.S. Kang, J. Sanders, T.
Murakami, Y. Ninomiya, and T. Oohashi. 2003. Molecular cloning of Bral2, a
novel brain-specific link protein, and immunohistochemical colocalization with
brevican in perineuronal nets. Mol Cell Neurosci. 24:148-59.
Bierhuizen, M.F., and M. Fukuda. 1992. Expression cloning of a cDNA encoding
UDP-GlcNAc:Gal beta 1-3-GalNAc-R (GlcNAc to GalNAc) beta 1-6GlcNAc
transferase by gene transfer into CHO cells expressing polyoma large tumor
antigen. Proc Natl Acad Sci U S A. 89:9326-330.
Bitter, T., and H.M. Muir. 1962. A modified uronic acid carbazole reaction. Anal Biochem.
- 273 - 4:330-4.
Bonassar, L.J., J.D. Sandy, M.W. Lark, A.H. Plaas, E.H. Frank, and A.J. Grodzinsky.
1997. Inhibition of cartilage degradation and changes in physical properties
induced by IL-1beta and retinoic acid using matrix metalloproteinase inhibitors.
Arch Biochem Biophys. 344:404-12.
Brewton, R.G., and R. Mayne. 1992. Mammalian vitreous humor contains networks of
hyaluronan molecules: electron microscopic analysis using the
hyaluronan-binding region (G1) of aggrecan and link protein. Exp Cell Res.
198:237-49.
Brown, G.M., T.N. Huckerby, M.T. Bayliss, and I.A. Nieduszynski. 1998. Human
aggrecan keratan sulfate undergoes structural changes during adolescent
development. J Biol Chem. 273:26408-14.
Buckwalter, J.A., and L.C. Rosenberg. 1982. Electron microscopic studies of cartilage
proteoglycans. Direct evidence for the variable length of the chondroitin
sulfate-rich region of proteoglycan subunit core protein. J Biol Chem. 257:9830-9.
Buckwalter, J.A., L.C. Rosenberg, and L.H. Tang. 1984. The effect of link protein on
proteoglycan aggregate structure. An electron microscopic study of the molecular
architecture and dimensions of proteoglycan aggregates reassembled from the
proteoglycan monomers and link proteins of bovine fetal epiphyseal cartilage. J
Biol Chem. 259:5361-3.
Calabro, A., V.C. Hascall, and R.J. Midura. 2000. Adaptation of FACE methodology for
microanalysis of total hyaluronan and chondroitin sulfate composition from
- 274 - cartilage. Glycobiology. 10:283-93.
Calabro, A., R. Midura, A. Wang, L. West, A. Plaas, and V.C. Hascall. 2001.
Fluorophore-assisted carbohydrate electrophoresis (FACE) of
glycosaminoglycans. Osteoarthritis Cartilage. 9 Suppl A:S16-22.
Carrino, D.A., J.L. Arias, and A.I. Caplan. 1991. A spectrophotometric modification of a
sensitive densitometric Safranin O assay for glycosaminoglycans. Biochem Int.
24:485-95.
Chen, P.Y., C.C. Lin, Y.T. Chang, S.C. Lin, and S.I. Chan. 2002. One O-linked sugar can
affect the coil-to-beta structural transition of the prion peptide. Proc Natl Acad Sci
U S A. 99:12633-8.
Choi, H.U., L.H. Tang, T.L. Johnson, and L. Rosenberg. 1985. Proteoglycans from bovine
nasal and articular cartilages. Fractionation of the link proteins by wheat germ
agglutinin affinity chromatography. J Biol Chem. 260:13370-6.
Christner, J.E., B. Caterson, and J.R. Baker. 1980. Immunological determinants of
proteoglycans. Antibodies against the unsaturated oligosaccharide products of
chondroitinase ABC-digested cartilage proteoglycans. J Biol Chem. 255:7102-5.
Collins-Racie, L.A., C.R. Flannery, W. Zeng, C. Corcoran, B. Annis-Freeman, M.J.
Agostino, M. Arai, E. DiBlasio-Smith, A.J. Dorner, K.E. Georgiadis, M. Jin, X.Y.
Tan, E.A. Morris, and E.R. LaVallie. 2004. ADAMTS-8 exhibits aggrecanase
activity and is expressed in human articular cartilage. Matrix Biol. 23:219-30.
Day, A.J., R.T. Aplin, and A.C. Willis. 1996. Overexpression, purification, and refolding
of link module from human TSG-6 in Escherichia coli: effect of temperature,
- 275 - media, and mutagenesis on lysine misincorporation at arginine AGA codons.
Protein Expr Purif. 8:1-16.
Day, A.J., and G.D. Prestwich. 2002. Hyaluronan-binding proteins: tying up the giant. J
Biol Chem. 277:4585-8.
Doege, K.J., M. Sasaki, T. Kimura, and Y. Yamada. 1991. Complete coding sequence and
deduced primary structure of the human cartilage large aggregating proteoglycan,
aggrecan. Human-specific repeats, and additional alternatively spliced forms. J
Biol Chem. 266:894-902.
Dudhia, J., and T.E. Hardingham. 1990. The primary structure of human cartilage link
protein. Nucleic Acids Res. 18:1292.
Esko, J.D., T.E. Stewart, and W.H. Taylor. 1985. Animal cell mutants defective in
glycosaminoglycan biosynthesis. Proc Natl Acad Sci U S A. 82:3197-201.
Flannery, C.R., W. Zeng, C. Corcoran, L.A. Collins-Racie, P.S. Chockalingam, T. Hebert,
S.A. Mackie, T. McDonagh, T.K. Crawford, K.N. Tomkinson, E.R. LaVallie, and
E.A. Morris. 2002. Autocatalytic cleavage of ADAMTS-4 (Aggrecanase-1)
reveals multiple glycosaminoglycan-binding sites. J Biol Chem. 277:42775-80.
Fosang, A.J., K. Last, Y. Fujii, M. Seiki, and Y. Okada. 1998. Membrane-type 1 MMP
(MMP-14) cleaves at three sites in the aggrecan interglobular domain. FEBS Lett.
430:186-90.
Fosang, A.J., K. Last, V. Knauper, G. Murphy, and P.J. Neame. 1996. Degradation of
cartilage aggrecan by collagenase-3 (MMP-13). FEBS Lett. 380:17-20.
Fosang, A.J., K. Last, V. Knauper, P.J. Neame, G. Murphy, T.E. Hardingham, H.
- 276 - Tschesche, and J.A. Hamilton. 1993. Fibroblast and neutrophil collagenases
cleave at two sites in the cartilage aggrecan interglobular domain. Biochem J. 295
(Pt 1):273-6.
Fosang, A.J., K. Last, P.J. Neame, G. Murphy, V. Knauper, H. Tschesche, C.E. Hughes, B.
Caterson, and T.E. Hardingham. 1994. Neutrophil collagenase (MMP-8) cleaves
at the aggrecanase site E373-A374 in the interglobular domain of cartilage
aggrecan. Biochem J. 304 (Pt 2):347-51.
Fosang, A. J,. K. Last, A.H., Plaas, C.J. Poon. 2004. Keratan sulphate in the aggrecan
interglobular domain has microstructure that is distinct from keratan sulphate
elsewhere on aggrecan. Ortho. Res. Soc. Trans. 29: 232
Fosang, A.J., P.J. Neame, T.E. Hardingham, G. Murphy, and J.A. Hamilton. 1991.
Cleavage of cartilage proteoglycan between G1 and G2 domains by stromelysins.
J Biol Chem. 266:15579-82.
Fosang, A.J., P.J. Neame, K. Last, T.E. Hardingham, G. Murphy, and J.A. Hamilton. 1992.
The interglobular domain of cartilage aggrecan is cleaved by PUMP, gelatinases,
and cathepsin B. J Biol Chem. 267:19470-4.
Fukuta, M., J. Inazawa, T. Torii, K. Tsuzuki, E. Shimada, and O. Habuchi. 1997.
Molecular cloning and characterization of human keratan sulfate
Gal-6-sulfotransferase. J Biol Chem. 272:32321-8.
Funderburgh, J.L. 2000. Keratan sulfate: structure, biosynthesis, and function.
Glycobiology. 10:951-8.
Funderburgh, J.L., M.L. Funderburgh, M.M. Mann, L. Corpuz, and M.R. Roth. 2001.
- 277 - Proteoglycan expression during transforming growth factor beta -induced
keratocyte-myofibroblast transdifferentiation. J Biol Chem. 276:44173-8.
Gao, G., A. Plaas, V.P. Thompson, S. Jin, F. Zuo, and J.D. Sandy. 2004. ADAMTS4
(aggrecanase-1) activation on the cell surface involves C-terminal cleavage by
glycosylphosphatidyl inositol-anchored membrane type 4-matrix
metalloproteinase and binding of the activated proteinase to chondroitin sulfate
and heparan sulfate on syndecan-1. J Biol Chem. 279:10042-51.
Gao, G., J. Westling, V.P. Thompson, T.D. Howell, P.E. Gottschall, and J.D. Sandy. 2002.
Activation of the proteolytic activity of ADAMTS4 (aggrecanase-1) by
C-terminal truncation. J Biol Chem. 277:11034-41.
Gerken, T.A., and Rarick, J. 2004 Approaches for characterizing ppGalNAc transferase
peptide substrate specificity. Glycobiology 14: abstract 486
Glasson, S.S., R. Askew, B. Sheppard, B. Carito, T. Blanchet, H.L. Ma, C.R. Flannery, D.
Peluso, K. Kanki, Z. Yang, M.K. Majumdar, and E.A. Morris. 2005. Deletion of
active ADAMTS5 prevents cartilage degradation in a murine model of
osteoarthritis. Nature. 434:644-8.
Gluzman, Y. 1981. SV40-transformed simian cells support the replication of early SV40
mutants. Cell. 23:175-82.
Grover, J., and P.J. Roughley. 1994. The expression of functional link protein in a
baculovirus system: analysis of mutants lacking the A, B and B' domains.
Biochem J. 300 (Pt 2):317-24.
Habuchi, O., Y. Matsui, Y. Kotoya, Y. Aoyama, Y. Yasuda, and M. Noda. 1993.
- 278 - Purification of chondroitin 6-sulfotransferase secreted from cultured chick
embryo chondrocytes. J Biol Chem. 268:21968-74.
Heinegard, D. 1972. Extraction, fractionation and characterization of proteoglycans from
bovine tracheal cartilage. Biochim Biophys Acta. 285:181-92.
Heinegard, D., and I. Axelsson. 1977. Distribution of keratan sulfate in cartilage
proteoglycans. J Biol Chem. 252:1971-9.
Heinegard, D., and S. Gardell. 1967. Studies on protein-polysaccharide complex
(proteoglycan) from human nucleus pulposus. I. Isolation and preliminary
characterisation. Biochim Biophys Acta. 148:164-71.
Helenius, A., and M. Aebi. 2004. Roles of N-linked glycans in the endoplasmic reticulum.
Annu Rev Biochem. 73:1019-49.
Hering, T.M., J. Kollar, and T.D. Huynh. 1997. Complete coding sequence of bovine
aggrecan: comparative structural analysis. Arch Biochem Biophys. 345:259-70.
Hering, T.M., J. Kollar, T.D. Huynh, and L.J. Sandell. 1995. Bovine chondrocyte link
protein cDNA sequence: interspecies conservation of primary structure and
mRNA untranslated regions. Comp Biochem Physiol B Biochem Mol Biol.
112:197-203.
Hering, T.M., and L.J. Sandell. 1990. Biosynthesis and processing of bovine cartilage
link proteins. J Biol Chem. 265:2375-82.
Hirakawa, S., T. Oohashi, W.D. Su, H. Yoshioka, T. Murakami, J. Arata, and Y. Ninomiya.
2000. The brain link protein-1 (BRAL1): cDNA cloning, genomic structure, and
characterization as a novel link protein expressed in adult brain. Biochem Biophys
- 279 - Res Commun. 276:982-9.
Horber, C., F.H. Buttner, C. Kern, G. Schmiedeknecht, and E. Bartnik. 2000. Truncation
of the amino-terminus of the recombinant aggrecan rAgg1mut leads to reduced
cleavage at the aggrecanase site. Efficient aggrecanase catabolism may depend on
multiple substrate interactions. Matrix Biol. 19:533-43.
Hughes, C.E., F.H. Buttner, B. Eidenmuller, B. Caterson, and E. Bartnik. 1997.
Utilization of a recombinant substrate rAgg1 to study the biochemical properties
of aggrecanase in cell culture systems. J Biol Chem. 272:20269-74.
Hughes, C.E., B. Caterson, A.J. Fosang, P.J. Roughley, and J.S. Mort. 1995. Monoclonal
antibodies that specifically recognize neoepitope sequences generated by
'aggrecanase' and matrix metalloproteinase cleavage of aggrecan: application to
catabolism in situ and in vitro. Biochem J. 305 (Pt 3):799-804.
Humphries, D.E., G. Sugumaran, and J.E. Silbert. 1989. Decreasing sulfation of
proteoglycans produced by cultured cells. Methods Enzymol. 179:428-34.
Kahmann, J.D., R. O'Brien, J.M. Werner, D. Heinegard, J.E. Ladbury, I.D. Campbell, and
A.J. Day. 2000. Localization and characterization of the hyaluronan-binding site
on the link module from human TSG-6. Structure Fold Des. 8:763-74.
Karlsson, M., I. Edfors-Lilja, and S. Bjornsson. 2000. Binding and detection of
glycosaminoglycans immobilized on membranes treated with cationic detergents.
Anal Biochem. 286:51-8.
Kashiwagi, M., J.J. Enghild, C. Gendron, C. Hughes, B. Caterson, Y. Itoh, and H. Nagase.
2004. Altered proteolytic activities of ADAMTS-4 expressed by C-terminal
- 280 - processing. J Biol Chem. 279:10109-19.
Kiani, C., L. Chen, Y.J. Wu, A.J. Yee, and B.B. Yang. 2002. Structure and function of
aggrecan. Cell Res. 12:19-32.
Kiani, C., V. Lee, L. Cao, L. Chen, Y. Wu, Y. Zhang, M.E. Adams, and B.B. Yang. 2001.
Roles of aggrecan domains in biosynthesis, modification by glycosaminoglycans
and product secretion. Biochem J. 354:199-207.
Kitagawa, H., T. Uyama, and K. Sugahara. 2001. Molecular cloning and expression of a
human chondroitin synthase. J Biol Chem. 276:38721-6.
Kohda, D., C.J. Morton, A.A. Parkar, H. Hatanaka, F.M. Inagaki, I.D. Campbell, and A.J.
Day. 1996. Solution structure of the link module: a hyaluronan-binding domain
involved in extracellular matrix stability and cell migration. Cell. 86:767-75.
Kokenyesi, R., L. Tan, J.R. Robbins, and M.B. Goldring. 2000. Proteoglycan production
by immortalized human chondrocyte cell lines cultured under conditions that
promote expression of the differentiated phenotype. Arch Biochem Biophys.
383:79-90.
Krishnan, P., A.M. Hocking, J.M. Scholtz, C.N. Pace, K.K. Holik, and D.J. McQuillan.
1999. Distinct secondary structures of the leucine-rich repeat proteoglycans
decorin and biglycan. Glycosylation-dependent conformational stability. J Biol
Chem. 274:10945-50.
Kuettner, K.E., B.U. Pauli, G. Gall, V.A. Memoli, and R.K. Schenk. 1982. Synthesis of
cartilage matrix by mammalian chondrocytes in vitro. I. Isolation, culture
characteristics, and morphology. J Cell Biol. 93:743-50.
- 281 - Kuno, K., Y. Okada, H. Kawashima, H. Nakamura, M. Miyasaka, H. Ohno, and K.
Matsushima. 2000. ADAMTS-1 cleaves a cartilage proteoglycan, aggrecan. FEBS
Lett. 478:241-5.
Laemmli, U.K. 1970. Cleavage of structural proteins during the assembly of the head of
bacteriophage T4. Nature. 227:680-5.
Lark, M.W., J.T. Gordy, J.R. Weidner, J. Ayala, J.H. Kimura, H.R. Williams, R.A.
Mumford, C.R. Flannery, S.S. Carlson, and M. Iwata. 1995. Cell-mediated
catabolism of aggrecan. Evidence that cleavage at the "aggrecanase" site
(Glu373-Ala374) is a primary event in proteolysis of the interglobular domain. J
Biol Chem. 270:2550-6.
Le Gledic, S., J.P. Perin, F. Bonnet, and P. Jolles. 1983. Identity of the protein cores of the
two link proteins from bovine nasal cartilage proteoglycan complex. Localization
of their sugar moieties. J Biol Chem. 258:14759-61.
Lee, E.R., L. Lamplugh, C.P. Leblond, S. Mordier, M.C. Magny, and J.S. Mort. 1998.
Immunolocalization of the cleavage of the aggrecan core protein at the
Asn341-Phe342 bond, as an indicator of the location of the metalloproteinases
active in the lysis of the rat growth plate. Anat Rec. 252:117-32.
Little, C.B., C.R. Flannery, C.E. Hughes, J.S. Mort, P.J. Roughley, C. Dent, and B.
Caterson. 1999. Aggrecanase versus matrix metalloproteinases in the catabolism
of the interglobular domain of aggrecan in vitro. Biochem J. 344 Pt 1:61-8.
Little, C.B., C.E. Hughes, C.L. Curtis, M.J. Janusz, R. Bohne, S. Wang-Weigand, Y.O.
Taiwo, P.G. Mitchell, I.G. Otterness, C.R. Flannery, and B. Caterson. 2002. Matrix
- 282 - metalloproteinases are involved in C-terminal and interglobular domain
processing of cartilage aggrecan in late stage cartilage degradation. Matrix Biol.
21:271-88.
Liu, H., L.A. McKenna, and M. Dean. 1997. An N-terminal peptide from link protein
stimulates synthesis of cartilage proteoglycans. Biochem Soc Trans. 25:427S.
Liu, H., L.A. McKenna, and M.F. Dean. 2000. An N-terminal peptide from link protein
can stimulate biosynthesis of collagen by human articular cartilage. Arch Biochem
Biophys. 378:116-22.
Lohmander, L.S., P.J. Neame, and J.D. Sandy. 1993. The structure of aggrecan fragments
in human synovial fluid. Evidence that aggrecanase mediates cartilage
degradation in inflammatory joint disease, joint injury, and osteoarthritis. Arthritis
Rheum. 36:1214-22.
Mahoney, D.J., C.D. Blundell, and A.J. Day. 2001. Mapping the hyaluronan-binding site
on the link module from human tumor necrosis factor-stimulated gene-6 by
site-directed mutagenesis. J Biol Chem. 276:22764-71.
Maier, R., H.G. Wisniewski, J. Vilcek, and M. Lotz. 1996. TSG-6 expression in human
articular chondrocytes. Possible implications in joint inflammation and cartilage
degradation. Arthritis Rheum. 39:552-9.
McDevitt, C.A., and H. Muir. 1971. Gel electrophoresis of proteoglycans and
glycosaminoglycans on large-pore composite polyacrylamide-agarose gels. Anal
Biochem. 44:612-22.
McKenna, L.A., H. Liu, P.A. Sansom, and M.F. Dean. 1998. An N-terminal peptide from
- 283 - link protein stimulates proteoglycan biosynthesis in human articular cartilage in
vitro. Arthritis Rheum. 41:157-62.
Meiler, J., M. Muller, A. Zeidler, and F. Schmaschke. 2001 Generation and evaluation of
dimension-reduced amino acid parameter representations by artificial neural
networks. J Mol Model 7: 360-9
Mehmet, H., P. Scudder, P.W. Tang, E.F. Hounsell, B. Caterson, and T. Feizi. 1986. The
antigenic determinants recognized by three monoclonal antibodies to keratan
sulphate involve sulphated hepta- or larger oligosaccharides of the
poly(N-acetyllactosamine) series. Eur J Biochem. 157:385-91.
Melrose, J., Y. Numata, and P. Ghosh. 1996. Biotinylated hyaluronan: a versatile and
highly sensitive probe capable of detecting nanogram levels of hyaluronan
binding proteins (hyaladherins) on electroblots by a novel affinity detection
procedure. Electrophoresis. 17:205-12.
Mercuri, F.A., K.J. Doege, E.C. Arner, M.A. Pratta, K. Last, and A.J. Fosang. 1999.
Recombinant human aggrecan G1-G2 exhibits native binding properties and
substrate specificity for matrix metalloproteinases and aggrecanase. J Biol Chem.
274:32387-95.
Mercuri, F.A., R.A. Maciewicz, J. Tart, K. Last, and A.J. Fosang. 2000. Mutations in the
interglobular domain of aggrecan alter matrix metalloproteinase and aggrecanase
cleavage patterns. Evidence that matrix metalloproteinase cleavage interferes with
aggrecanase activity. J Biol Chem. 275:33038-45.
Morgelin, M., M. Paulsson, A. Malmstrom, and D. Heinegard. 1989. Shared and distinct
- 284 - structural features of interstitial proteoglycans from different bovine tissues
revealed by electron microscopy. J Biol Chem. 264:12080-90.
Mort, J.S., B. Caterson, A.R. Poole, and P.J. Roughley. 1985. The origin of human
cartilage proteoglycan link-protein heterogeneity and fragmentation during aging.
Biochem J. 232:805-12.
Murata, K., and A.O. Bjelle. 1979. Age-dependent constitution of chondroitin sulfate
isomers in cartilage proteoglycans under associative conditions. J Biochem
(Tokyo). 86:371-6.
Nagase, T., K. Ishikawa, N. Miyajima, A. Tanaka, H. Kotani, N. Nomura, and O. Ohara.
1998. Prediction of the coding sequences of unidentified human genes. IX. The
complete sequences of 100 new cDNA clones from brain which can code for large
proteins in vitro. DNA Res. 5:31-9.
Nagase, T., N. Seki, K. Ishikawa, M. Ohira, Y. Kawarabayasi, O. Ohara, A. Tanaka, H.
Kotani, N. Miyajima, and N. Nomura. 1996. Prediction of the coding sequences
of unidentified human genes. VI. The coding sequences of 80 new genes
(KIAA0201-KIAA0280) deduced by analysis of cDNA clones from cell line
KG-1 and brain. DNA Res. 3:321-9, 341-54.
Nakazawa, K., I. Takahashi, and Y. Yamamoto. 1998. Glycosyltransferase and
sulfotransferase activities in chick corneal stromal cells before and after in vitro
culture. Arch Biochem Biophys. 359:269-82.
Neame, P.J., and F.P. Barry. 1993. The link proteins. Experientia. 49:393-402.
Neame, P.J., J.E. Christner, and J.R. Baker. 1986. The primary structure of link protein
- 285 - from rat chondrosarcoma proteoglycan aggregate. J Biol Chem. 261:3519-35.
Nehrke, K., and L.A. Tabak. 1997. Biosynthesis of a low-molecular-mass rat
submandibular gland mucin glycoprotein in COS7 cells. Biochem J. 323 (Pt
2):497-502.
Ng, L., A.J. Grodzinsky, P. Patwari, J. Sandy, A. Plaas, and C. Ortiz. 2003. Individual
cartilage aggrecan macromolecules and their constituent glycosaminoglycans
visualized via atomic force microscopy. J Struct Biol. 143:242-57.
Oegema, T.R., Jr. 1980. Delayed formation of proteoglycan aggregate structures in
human articular cartilage disease states. Nature. 288:583-5.
Ogawa, H., T. Oohashi, M. Sata, Y. Bekku, S. Hirohata, K. Nakamura, T. Yonezawa, S.
Kusachi, Y. Shiratori, and Y. Ninomiya. 2004. Lp3/Hapln3, a novel link protein
that co-localizes with versican and is coordinately up-regulated by
platelet-derived growth factor in arterial smooth muscle cells. Matrix Biol.
23:287-298.
Oohashi, T., S. Hirakawa, Y. Bekku, U. Rauch, D.R. Zimmermann, W.D. Su, A. Ohtsuka,
T. Murakami, and Y. Ninomiya. 2002. Bral1, a brain-specific link protein,
colocalizing with the versican V2 isoform at the nodes of Ranvier in developing
and adult mouse central nervous systems. Mol Cell Neurosci. 19:43-57.
Patwari, P., G. Gao, J.H. Lee, A.J. Grodzinsky, and J.D. Sandy. 2005. Analysis of
ADAMTS4 and MT4-MMP indicates that both are involved in aggrecanolysis in
interleukin-1-treated bovine cartilage. Osteoarthritis Cartilage. 13:269-77.
Pigiet, V.P., and B.J. Schuster. 1986. Thioredoxin-catalyzed refolding of
- 286 - disulfide-containing proteins. Proc Natl Acad Sci U S A. 83:7643-7.
Plaas, A.H., L.A. West, and R.J. Midura. 2001. Keratan sulfate disaccharide composition
determined by FACE analysis of keratanase II and endo-beta-galactosidase
digestion products. Glycobiology. 11:779-90.
Poon, C.J., A.H. Plaas, D.R. Keene, D.J. McQuillan, K. Last, and A.J. Fosang. 2005.
N-linked keratan sulphate in the aggrecan interglobular domain potentiates
aggrecanase activity. J Biol Chem.
Prabhakar, V., I. Capila, C.J. Bosques, K. Pojasek, and R. Sasisekharan. 2005.
Chondroitinase ABC I from Proteus vulgaris: cloning, recombinant expression
and active site identification. Biochem J. 386:103-12.
Pratta, M.A., M.D. Tortorella, and E.C. Arner. 2000. Age-related changes in aggrecan
glycosylation affect cleavage by aggrecanase. J Biol Chem. 275:39096-102.
Rauch, U., S. Hirakawa, T. Oohashi, J. Kappler, and G. Roos. 2004. Cartilage link protein
interacts with neurocan, which shows hyaluronan binding characteristics different
from CD44 and TSG-6. Matrix Biol. 22:629-39.
Rauch, U., L. Karthikeyan, P. Maurel, R.U. Margolis, and R.K. Margolis. 1992. Cloning
and primary structure of neurocan, a developmentally regulated, aggregating
chondroitin sulfate proteoglycan of brain. J Biol Chem. 267:19536-47.
Rodriguez-Manzaneque, J.C., J. Westling, S.N. Thai, A. Luque, V. Knauper, G. Murphy,
J.D. Sandy, and M.L. Iruela-Arispe. 2002. ADAMTS1 cleaves aggrecan at
multiple sites and is differentially inhibited by metalloproteinase inhibitors.
Biochem Biophys Res Commun. 293:501-8.
- 287 - Rosenberg, L., H.U. Choi, L.H. Tang, S. Pal, T. Johnson, D.A. Lyons, and T.M. Laue.
1991. Proteoglycans of bovine articular cartilage. The effects of divalent cations
on the biochemical properties of link protein. J Biol Chem. 266:7016-24.
Roughley, P.J., J. Barnett, F. Zuo, and J.S. Mort. 2003. Variations in aggrecan structure
modulate its susceptibility to aggrecanases. Biochem J. 375:183-9.
Roughley, P.J., and R.J. White. 1980. Age-related changes in the structure of the
proteoglycan subunits from human articular cartilage. J Biol Chem. 255:217-24.
Sah, R.L., A.J. Grodzinsky, A.H. Plaas, and J.D. Sandy. 1990. Effects of tissue
compression on the hyaluronate-binding properties of newly synthesized
proteoglycans in cartilage explants. Biochem J. 267:803-8.
Sandy, J.D., C.R. Flannery, P.J. Neame, and L.S. Lohmander. 1992. The structure of
aggrecan fragments in human synovial fluid. Evidence for the involvement in
osteoarthritis of a novel proteinase which cleaves the Glu 373-Ala 374 bond of
the interglobular domain. J Clin Invest. 89:1512-6.
Sandy, J.D., and A.H. Plaas. 1989. Studies on the hyaluronate binding properties of newly
synthesized proteoglycans purified from articular chondrocyte cultures. Arch
Biochem Biophys. 271:300-14.
Sandy, J.D., V. Thompson, K. Doege, and C. Verscharen. 2000. The intermediates of
aggrecanase-dependent cleavage of aggrecan in rat chondrosarcoma cells treated
with interleukin-1. Biochem J. 351:161-6.
Sandy, J.D., and C. Verscharen. 2001. Analysis of aggrecan in human knee cartilage and
synovial fluid indicates that aggrecanase (ADAMTS) activity is responsible for
- 288 - the catabolic turnover and loss of whole aggrecan whereas other protease activity
is required for C-terminal processing in vivo. Biochem J. 358:615-26.
Sato, T., M. Gotoh, K. Kiyohara, A. Kameyama, T. Kubota, N. Kikuchi, Y. Ishizuka, H.
Iwasaki, A. Togayachi, T. Kudo, T. Ohkura, H. Nakanishi, and H. Narimatsu.
2003. Molecular cloning and characterization of a novel human beta
1,4-N-acetylgalactosaminyltransferase, beta 4GalNAc-T3, responsible for the
synthesis of N,N'-diacetyllactosediamine, galNAc beta 1-4GlcNAc. J Biol Chem.
278:47534-44.
Shi, S., S. Grothe, Y. Zhang, M.D. O'Connor-McCourt, A.R. Poole, P.J. Roughley, and J.S.
Mort. 2004. Link protein has greater affinity for versican than aggrecan. J Biol
Chem. 279:12060-6.
Silbert, J.E., and G. Sugumaran. 2002. Biosynthesis of chondroitin/dermatan sulfate.
IUBMB Life. 54:177-86.
Somerville, R.P., J.M. Longpre, K.A. Jungers, J.M. Engle, M. Ross, S. Evanko, T.N.
Wight, R. Leduc, and S.S. Apte. 2003. Characterization of ADAMTS-9 and
ADAMTS-20 as a distinct ADAMTS subfamily related to Caenorhabditis elegans
GON-1. J Biol Chem. 278:9503-13.
Spicer, A.P., A. Joo, and R.A. Bowling, Jr. 2003. A hyaluronan binding link protein gene
family whose members are physically linked adjacent to chondroitin sulfate
proteoglycan core protein genes: the missing links. J Biol Chem. 278:21083-91.
Stanton, H., F.M. Rogerson, C.J. East, S.B. Golub, K.E. Lawlor, C.T. Meeker, C.B. Little,
K. Last, P.J. Farmer, I.K. Campbell, A.M. Fourie, and A.J. Fosang. 2005.
- 289 - ADAMTS5 is the major aggrecanase in mouse cartilage in vivo and in vitro.
Nature. 434:648-52.
Stickens, D., D.J. Behonick, N. Ortega, B. Heyer, B. Hartenstein, Y. Yu, A.J. Fosang, M.
Schorpp-Kistner, P. Angel, and Z. Werb. 2004. Altered endochondral bone
development in matrix metalloproteinase 13-deficient mice. Development.
131:5883-95.
Sugimoto, K., M. Takahashi, Y. Yamamoto, K. Shimada, and K. Tanzawa. 1999.
Identification of aggrecanase activity in medium of cartilage culture. J Biochem
(Tokyo). 126:449-55.
Swagerty, D.L., Jr., and D. Hellinger. 2001. Radiographic assessment of osteoarthritis.
Am Fam Physician. 64:279-86.
Sztrolovics, R., M. Alini, P.J. Roughley, and J.S. Mort. 1997. Aggrecan degradation in
human intervertebral disc and articular cartilage. Biochem J. 326 (Pt 1):235-41.
Sztrolovics, R., R.J. White, P.J. Roughley, and J.S. Mort. 2002. The mechanism of
aggrecan release from cartilage differs with tissue origin and the agent used to
stimulate catabolism. Biochem J. 362:465-72.
Tang, B.L. 2001. ADAMTS: a novel family of extracellular matrix proteases. Int J
Biochem Cell Biol. 33:33-44.
Tengblad, A. 1981. A comparative study of the binding of cartilage link protein and the
hyaluronate-binding region of the cartilage proteoglycan to
hyaluronate-substituted Sepharose gel. Biochem J. 199:297-305.
Thornton, D.J., J.K. Sheehan, and I.A. Nieduszynski. 1987. A study of the interaction
- 290 - between cartilage proteoglycan and link protein. Biochem J. 248:943-51.
Tortorella, M., M. Pratta, R.Q. Liu, I. Abbaszade, H. Ross, T. Burn, and E. Arner. 2000.
The thrombospondin motif of aggrecanase-1 (ADAMTS-4) is critical for
aggrecan substrate recognition and cleavage. J Biol Chem. 275:25791-7.
Tortorella, M.D., T.C. Burn, M.A. Pratta, I. Abbaszade, J.M. Hollis, R. Liu, S.A.
Rosenfeld, R.A. Copeland, C.P. Decicco, R. Wynn, A. Rockwell, F. Yang, J.L.
Duke, K. Solomon, H. George, R. Bruckner, H. Nagase, Y. Itoh, D.M. Ellis, H.
Ross, B.H. Wiswall, K. Murphy, M.C. Hillman, Jr., G.F. Hollis, and E.C. Arner.
1999. Purification and cloning of aggrecanase-1: a member of the ADAMTS
family of proteins. Science. 284:1664-6.
Tortorella, M.D., R.Q. Liu, T. Burn, R.C. Newton, and E. Arner. 2002. Characterization
of human aggrecanase 2 (ADAM-TS5): substrate specificity studies and
comparison with aggrecanase 1 (ADAM-TS4). Matrix Biol. 21:499-511.
Tortorella, M.D., M. Pratta, R.Q. Liu, J. Austin, O.H. Ross, I. Abbaszade, T. Burn, and E.
Arner. 2000. Sites of aggrecan cleavage by recombinant human aggrecanase-1
(ADAMTS-4). J Biol Chem. 275:18566-73.
Treadwell, B.V., L. Shader, C.A. Towle, D.P. Mankin, and H.J. Mankin. 1980.
Purification of the 'link proteins' from bovine articular cartilage and comparison
with 'link proteins' from nasal septum. Biochem Biophys Res Commun. 94:159-66.
Varelas, J.B., J. Kollar, T.D. Huynh, and T.M. Hering. 1995. Expression and
characterization of a single recombinant proteoglycan tandem repeat domain of
link protein that binds zinc and hyaluronate. Arch Biochem Biophys. 321:21-30.
- 291 - Varelas, J.B., C. Roy, and T.M. Hering. 1997. A structural requirement of zinc for the
folding of recombinant link protein. Arch Biochem Biophys. 347:1-8.
Varelas, J.B., N.R. Zenarosa, and C.J. Froelich. 1991. Agarose/polyacrylamide minislab
gel electrophoresis of intact cartilage proteoglycans and their proteolytic
degradation products. Anal Biochem. 197:396-400.
Wang, P., M. Tortorella, K. England, A.M. Malfait, G. Thomas, E.C. Arner, and D. Pei.
2004. Proprotein convertase furin interacts with and cleaves pro-ADAMTS4
(Aggrecanase-1) in the trans-Golgi network. J Biol Chem. 279:15434-40.
Wasteson, A. 1971. A method for the determination of the molecular weight and
molecular-weight distribution of chondroitin sulphate. J Chromatogr. 59:87-97.
Watanabe, H., S.C. Cheung, N. Itano, K. Kimata, and Y. Yamada. 1997. Identification of
hyaluronan-binding domains of aggrecan. J Biol Chem. 272:28057-65.
Wells, T., C. Davidson, M. Morgelin, J.L. Bird, M.T. Bayliss, and J. Dudhia. 2003.
Age-related changes in the composition, the molecular stoichiometry and the
stability of proteoglycan aggregates extracted from human articular cartilage.
Biochem J. 370:69-79.
Westling, J., A.J. Fosang, K. Last, V.P. Thompson, K.N. Tomkinson, T. Hebert, T.
McDonagh, L.A. Collins-Racie, E.R. LaVallie, E.A. Morris, and J.D. Sandy. 2002.
ADAMTS4 cleaves at the aggrecanase site (Glu373-Ala374) and secondarily at
the matrix metalloproteinase site (Asn341-Phe342) in the aggrecan interglobular
domain. J Biol Chem. 277:16059-66.
Westling, J., P.E. Gottschall, V.P. Thompson, A. Cockburn, G. Perides, D.R. Zimmermann,
- 292 - and J.D. Sandy. 2004. ADAMTS4 (aggrecanase-1) cleaves human brain versican
V2 at Glu405-Gln406 to generate glial hyaluronate binding protein. Biochem J.
377:787-95.
Wong-Palms, S., and A.H. Plaas. 1995. Glycosaminoglycan addition to proteoglycans by
articular chondrocytes--evidence for core protein-specific pathways. Arch
Biochem Biophys. 319:383-92.
Yada, T., M. Gotoh, T. Sato, M. Shionyu, M. Go, H. Kaseyama, H. Iwasaki, N. Kikuchi,
Y.D. Kwon, A. Togayachi, T. Kudo, H. Watanabe, H. Narimatsu, and K. Kimata.
2003. Chondroitin sulfate synthase-2. Molecular cloning and characterization of a
novel human glycosyltransferase homologous to chondroitin sulfate
glucuronyltransferase, which has dual enzymatic activities. J Biol Chem.
278:30235-47.
Yada, T., T. Sato, H. Kaseyama, M. Gotoh, H. Iwasaki, N. Kikuchi, Y.D. Kwon, A.
Togayachi, T. Kudo, H. Watanabe, H. Narimatsu, and K. Kimata. 2003.
Chondroitin sulfate synthase-3. Molecular cloning and characterization. J Biol
Chem. 278:39711-25.
Yamada, H., K. Watanabe, M. Shimonaka, and Y. Yamaguchi. 1994. Molecular cloning of
brevican, a novel brain proteoglycan of the aggrecan/versican family. J Biol Chem.
269:10119-26.
Yamaguchi, Y. 2000. Lecticans: organizers of the brain extracellular matrix. Cell Mol Life
Sci. 57:276-89.
Yamanishi, Y., D.L. Boyle, M. Clark, R.A. Maki, M.D. Tortorella, E.C. Arner, and G.S.
- 293 - Firestein. 2002. Expression and regulation of aggrecanase in arthritis: the role of
TGF-beta. J Immunol. 168:1405-12.
Yanagishita, M., R.J. Midura, and V.C. Hascall. 1987. Proteoglycans: isolation and
purification from tissue cultures. Methods Enzymol. 138:279-89.
Zheng, J., W. Luo, and M.L. Tanzer. 1998. Aggrecan synthesis and secretion. A paradigm
for molecular and cellular coordination of multiglobular protein folding and
intracellular trafficking. J Biol Chem. 273:12999-3006.
- 294 -