Proteomic Analysis of Potential Keratan Sulfate, Chondroitin Sulfate A, and Hyaluronic Acid Molecular Interactions

Abstract

Purpose.: Corneal stroma extracellular matrix (ECM) glycosaminoglycans (GAGs) include keratan sulfate (KS), chondroitin sulfate A (CSA), and hyaluronic acid (HA). Embryonic corneal keratocytes and sensory nerve fibers grow and differentiate according to chemical cues they receive from the ECM. This study asked which of the proteins that may regulate keratocytes or corneal nerve growth cone immigration interact with corneal GAGs.

Methods.: Biotinylated KS (bKS), CSA (bCSA), and HA (bHA) were prepared and used in microarray protocols to assess their interactions with 8268 proteins and a custom microarray of 85 extracellular epitopes of nerve growth-related proteins. Surface plasmon resonance (SPR) was performed with bKS and SLIT2, and their k _a, k _d, and K _D were determined.

Results.: Highly sulfated KS interacted with 217 microarray proteins, including 75 kinases, several membrane or secreted proteins, many cytoskeletal proteins, and many nerve function proteins. CSA interacted with 24 proteins, including 10 kinases and 2 cell surface proteins. HA interacted with 6 proteins, including several ECM-related structural proteins. Of 85 ECM nerve-related epitopes, KS bound 40 proteins, including SLIT, 2 ROBOs, 9 EPHs, 8 Ephrins (EFNs), 8 semaphorins (SEMAs), and 2 nerve growth factor receptors. CSA bound nine proteins, including ROBO2, 2 EPHs, 1 EFN, two SEMAs, and netrin 4. HA bound no ECM nerve-related epitopes. SPR confirmed that KS binds SLIT2 strongly. The KS core protein mimecan/osteoglycin bound 15 proteins.

Conclusions.: Corneal stromal GAGs bind, and thus could alter the availability or conformation of, many proteins that may influence keratocyte and nerve growth cone behavior in the cornea.

Glycosaminoglycans (GAGs) are long, unbranched polysaccharide chains composed of repeating disaccharide units, often either d-glucuronic acid (GlcA), l-iduronic acid (IdoA), or d-galactose (Gal) linked to either d-N-acetylglucosamine (GlcNAc) or d-N-acetylgalactosamine (GalNAc). One or both sugars can be sulfated. The carboxylate (–COO⁻) and sulfate (–SO₄ ) groups make GAGs highly negatively charged polymers. When most GAG chains are synthesized, they are covalently linked at their reducing end to core proteins, thus forming proteoglycans (PGs). The adult human corneal stroma contains (by weight) 65% keratan sulfate (KS) and 30% chondroitin/dermatan sulfate (CS/DS), ¹ first identified as major GAGs of the cornea by Meyer et al. ^2,3 Corneal KS chains are attached to the core proteins lumican, keratocan, and mimecan/osteoglycin, ^4–6 and corneal CS/DS chains are attached to decorin and biglycan. ^7,8 In addition, corneal stromal cells in culture, ⁹ early embryonic corneal endothelial cells, ¹⁰ and anterior stromal cells in injured adult corneas ¹¹ also make hyaluronic acid (HA), unique among GAGs because it is neither covalently attached to a core protein nor sulfated. 

GAGs influence many biological functions, including cell division, differentiation, signal transduction, adhesion, migration, peripheral nerve extension or regeneration, and responses to growth factors, ^12–16 sometimes affecting cell adhesion and/or proliferation differently in different ECM compositions. ¹⁷ In exerting these influences, GAGs interact with proteins in many different ways. Clusters of basic amino acids in BBXB and BBBXXB sequences (where B is a basic amino acid and X is a hydrophilic amino acid) have been proposed as consensus GAG binding sites in some proteins. ¹⁸ In other cases, interactions between GAGs and proteins appear to be more directly charge-dependent, either because the degree of sulfation of the GAG determines the strength of the interaction, ¹⁹ or because an increase in the number of GAG chains on a core protein increases the interactions of that PG with other proteins. ²⁰  

Little is known about the protein interactions of the corneal GAGs KS, CSA, and HA. KS, composed of repeating -[3Galβ1–4GlcNAcβ1]- disaccharide units, has an average molecular mass of approximately 15 kDa. In addition to corneal core KS proteins, KS can be attached to fibromodulin, aggrecan, and PRELP in cartilage and to osteoadherin in bone. ²¹ Although often an ECM component, KS can also be on cell surfaces ^22,23 or in the cytosol. ²⁴ KS chains are sulfated to different degrees in different tissues, ²⁵ at different stages of development within a given tissue, ²⁶ and in different disease states. ²⁷ KS chains bind the cardiotoxins CTX A3 and Tγ ²⁸ and insulin-like growth factor binding protein-2 (IGFBP2). ²⁹ KS core proteins keratocan and lumican bind CXC chemokine KC during corneal inflammatory response. ³⁰ No binding partners for mimecan/osteoglycin have been reported. 

CS is composed of repeating -[4GlcAβ1–3GalNAcβ1]- disaccharide units, with sulfate groups covalently attached to C-4 of GalNAc (CSA), C-6 of GalNAc (CSC), C-2 of GlcA and C-6 of GalNAc (CSD), or C-4 and C-6 of GalNAc (CSE). ³ CS GAGs average approximately 21.5 kDa in molecular mass, can be sulfated to different degrees in different tissues, and are linked to core proteins. Sulfated CS binds the secreted growth factor Nodal. ³¹ CSA binds opticin (OPT) ³² and Indian hedgehog, ³³ but not pleiotrophin, ³⁴ FGF-3, -5, -6, -8, -22, ³⁵ or BMP4. ³⁶  

Nonsulfated anionic HA, first discovered in ocular vitreous humor, ³⁷ is composed of repeating -[4GlcAβ1–3GlcNAcβ1]- disaccharide units, is synthesized on the plasma membrane with no attachment to any core protein, and is very large, with molecular masses from 100 to 10,000 kDa. Nonsulfated HA has the same chemical composition in any tissue in which it occurs. HA noncovalently binds several PGs (aggrecan, ³⁸ versican, ³⁹ neurocan, ⁴⁰ and brevican ⁴¹ ), link proteins, ⁴² TNF-stimulated gene 6 (TSG-6), ⁴³ and the receptor CD44 ⁴⁴ through a linking region known as the PG tandem repeat (PTR) module. ⁴⁵ Other HA-binding proteins include fibronectin, ⁴⁶ receptor for HA-mediated mobility (RHAMM), ^47,48 and intracellular adhesion molecule-1 (ICAM-1). ⁴⁹ There is increasing evidence that HA occurs intracellularly, ^50,51 as well as extracellularly. 

In developing chick embryos, the neurorepellant ECM component semaphorin 3A (SEMA3A), secreted by the lens, strongly influences when neural crest–derived corneal peripheral nerve axons invade the corneal anterior stroma and extend and branch through the anterior stroma and corneal epithelium. ⁵² Corneal nerve growth cone penetration occurs, even though anterior stromal cells and corneal epithelial cells express SLIT2, a strong negative regulator of nerve growth, throughout chick embryonic development. ^53,54 It has been suggested that, in the posterior E9 cornea, ECM-accumulated, highly sulfated KS binds and thereby blocks the diffusion of lens SEMA3A, thus allowing anterior stromal penetration of corneal nerve growth cones to begin on E9, and that corneal nerves extend into the anterior stroma and eventually into the corneal epithelium, to avoid increasing ECM accumulation of highly sulfated KS from posterior to anterior in the developing corneal stroma. ⁵³  

To test this hypothesis and identify other proteins that may interact with corneal GAGs, bovine corneal KS, sturgeon notochord CSA, and chicken cockscomb HA were biotinylated and used to screen an Invitrogen Proteomics Array of more than 8000 intracellular and extracellular (Version 4 [v4]; Invitrogen, Carlsbad, CA) proteins and a separate custom array of 85 extracellular epitopes of neuroregulator receptors, ligands, and growth factors. The biotinylated corneal KS core protein mimecan (OGN) was also used to screen the v4 array. The least prevalent corneal KS core protein by % weight, it may play a major role in some functions, depending on its localization, and nothing is currently known about the proteins that it may bind. Corneal KS, sturgeon notochord CSA, and HA bound many of these proteins selectively, suggesting that KS, CSA, and HA could alter the availability or conformation of many proteins that may influence corneal keratocyte and nerve growth cone behavior. 

Methods

Materials

Bovine corneal KS and sturgeon notochord CSA were purchased from Northstar BioProducts of Associates of Cape Cod, Inc. (ACC; formerly Seikagaku America; East Falmouth, MA). Biotinylated chicken cockscomb HA (bHA), N-ethyl-N′-(3-dimethylaminopropyl)-carbodiimide (EDC), and biotin-X-hydrazide were purchased from Sigma-Aldrich (St. Louis, MO). MES-buffered saline packs (BupH) and the biotin quantitation assay contained in a biotinylation kit (EZ-Link Sulfo-NHS-LC-Biotinylation Kit; product no. 21435) were purchased from Thermo Scientific Pierce Protein Research Products (Rockford, IL). A centrifugal filter (regenerated cellulose 3000 MWCO; Ambion Ultra) was purchased from Millipore Biomanufacturing and Life Research Products (Billerica, MA). Protoarray v4 proteins, Alexa-Fluor 647-labeled goat anti-rat IgG, V5-his-tagged calmodulin kinase 1, and Alexa-Fluor 647-labeled anti-V5 antibody were supplied by Invitrogen as part of their protoarray service. All extracellular epitopes of nerve-related proteins listed in Table 1, mimecan protein, and rat anti-mouse mimecan antibody were purchased from R&D Systems (Minneapolis, MN). Surface plasmon resonance (SPR) analysis was conducted on a commercial system (model 3000; GE Health Care; Biacore Life Sciences Division [GE/Biacore], Piscataway, NJ), and the data were analyzed (BIAevaluation software 4.1; GE/Biacore). Streptavidin (SA)-coated chips, biotin capture kits, and HEPES buffer HBS-P (with 0.005% [vol/vol] surfactant P20, 10 mM HEPES ([pH 7.4], 150 mM NaCl) running buffer were also purchased from GE/Biacore. 

Table 1.

 

Protein Epitopes Used in Nerve-Related Protoarray

Protein	Symbol	With IgG1	Without IgG1
Human IgG1 Pro100-Lys330		X
Human Integrin alpha V beta 3	αVβ3		#
Human Integrin alpha 3 beta 1	α3β1		#
Human FGF acidic	FGFa		#
Human FGF basic (157 aa),	FGFb		#
Human EGF	EGF		#
Human PDGF	PDGF		#
Human TGF-beta 1	TGFβ1		#
Human NT-3	NT3		#
Human TrkA	Ntrk1	X	X
Human TrkB	Ntrk2		#
Human TrkC	Ntrk3	X	X
Human NGFR/TNFRSF16	NGFR		X
Human Neuregulin-1, SMDF isoform	NRG1		#
Human ErbB2	ERBB2	X	X
Human ErbB4	ERBB4	X	X
Mouse SLIT3	SLIT3		#
Rat ROBO1	ROBO1	X	X
Human ROBO2	ROBO2	X	X
Mouse Ephrin-A1	EFNA1	X	X
Mouse Ephrin-A2	EFNA2	X	X
Human Ephrin-A3	EFNA3	X	X
Human Ephrin-A4	EFNA4	X	X
Human Ephrin-A5	EFNA5	X	X
Mouse Ephrin-B1	EFNB1	X	X
Mouse Ephrin-B2	EFNB2	X	X
Human Ephrin-B3	EFNB3	X	X
Human EphA1	EPHA1	X	X
Mouse EphA2	EPHA2	X	X
Mouse EphA3	EPHA3	X	X
Mouse EphA4	EPHA4	X	X
Rat EphA5	EPHA5	X	X
Mouse EphA6	EPHA6	X	X
Mouse EphA7	EPHA7	X	X
Mouse EphA8	EPHA8	X	X
Rat EphB1	EPHB1	X	0
Mouse EphB2	EPHB2	X	X
Mouse EphB3	EPHB3	X	0
Mouse EphB4	EPHB4	X	X
Mouse EphB6	EPHB6	X	X
Human Semaphorin 3A	SEMA3A	X	0
Human Semaphorin 3E	SEMA3E		#
Mouse Semaphorin 3F (Trunc)	SEMA3F	X	X
Human Semaphorin 4A	SEMA4A	X	0
Human Semaphorin 6A	SEMA6A	X	X
Human Semaphorin 6B	SEMA6B	X	X
Human Neuropilin-1	NPN1		#
Human Neuropilin-2	NPN2	X	X
Chicken Netrin-1	NTN1		#
Chicken Netrin-2	NTN2		#
Human Netrin-4	NTN4		#
Mouse DCC	DCC	X
Rat UNC5H1	NC5H1	X
Rat UNC5H2	NC5H2	X
Human UNC5H3	UNC5H3	X
Human UNC5H4	NC5H4	X

 Total number of proteins = 85. X, incorporated in protoarray; #, incorporated in protoarray, but only available without IgG₁ fusion; 0, no IgG₁ cleaved form available.

Preparation of Biotinylated KS

Seikagaku America purified KS by chromatography and chondroitinase ABC treatment, leaving a residual core peptide at the reducing terminus of KS chains. The average molecular mass of the purified KS was approximately 15 kDa. Attachment of biotin to free carboxy groups in the residual core peptide was performed by using the EDC protocol. ^55,56 KS (1 mg) was dissolved in 0.5 mL 0.1 M MES ((2-N-morpholino)ethanesulfonic acid; pH 4.5), containing 2 mg/mL biotin-LC-hydrazide and 20 mg/mL EDC. The mixture was rotated end-over-end at room temperature for 24 hours. With a centrifugal filter (regenerated cellulose 3000 MWCO, Amicon Ultra; Millipore), biotinylated KS (bKS) molecules were extensively washed with deionized H₂O to remove salt and free biotin and then were lyophilized. The yield of bKS was 0.8 mg. The 2-(4′-hydroxyazobenzene) benzoic acid (HABA) assay was used for measuring the level of biotin incorporation according to the manufacturer's protocol (EZ-Link Sulfo-NHS-LC-Biotinylation Kit; Thermo Scientific). Briefly, a calibration curve was obtained by adding biotin to 500 μL avidin-HABA reagent and measuring the decrease in absorbance at 500 nm. The biotin content of bKS was estimated by mixing 50 μL of the sample and 450 μL of the avidin-HABA reagent and comparing the absorbance at 500 nm with the standard. A biotinylation ratio of 1.43 moles biotin per mole KS was obtained, suggesting that one to two molecules of biotin were attached to each KS chain. Since there are no free carboxyl groups in the KS disaccharide, no biotin molecules were attached along the sides of the KS chain. 

Preparation of Biotinylated CSA

Sturgeon notochord CSA was highly purified by chromatography by Seikagaku America. The average molecular mass of CSA was ∼16 kDa. Purified CSA was labeled (EZ-link Biotin-LC-Hydrazide; Thermo Scientific), and biotin was attached to the carboxyl groups of uronic acid residues along the CSA GAG chain, by using the EDC protocol, ^55,56 beginning with 2 mg CSA dissolved in 1 mL 0.1 M MES (pH 4.5) containing 2 mg/mL biotin-LC-hydrazide and 20 mg/mL EDC. The yield of the biotinylated CSA (bCSA) was 1.7 mg. The HABA assay was used to determine a biotinylation ratio of 2.78 moles biotin per mole CSA, which is equivalent to 7.6 residues of biotin for every 100 disaccharide residues. 

Biotinylated HA

Biotinylated chicken cockscomb HA (bHA) was purchased directly from Sigma-Aldrich. According to the certificate of analysis, the molecular mass of the HA was 850 kDa. Biotin was attached by the manufacturer to carboxyl groups of glucuronic acid along the HA GAG chain, resulting in a degree of substitution of 6 mol%, which is equivalent to 6 residues of biotin for every 100 disaccharide residues. 

GAG Protoarrays

Protoarray Version 4 (Invitrogen) comprises 8268 human proteins, most of which are produced in baculovirus expression systems as N-terminal glutathione S-transferase fusion proteins and purified using glutathione S-Sepharose under nondenaturing conditions. These proteins, therefore, do not carry any of the posttranslational modifications, such as glycosylation side chains, that they may carry when synthesized in their normal human cell environment. All proteins were dissolved in printing buffer (PB) (50 mM HEPES, 200 mM NaCl, 0.08% Triton X-100, 25% glycerol, and 20 mM reduced glutathione, pH 7.5) at ∼50 ng/μL and spotted on nitrocellulose-coated glass slides. Protein purification and array printing were done at ∼6°C. A searchable list of these proteins may be viewed at: http://hdl.handle.net/2097/1716, including the amino acid sequence of the protein version spotted at each site, an estimate of the amount of protein in each spot for all GST-tagged proteins, based on an assay with an anti-GST antibody, and an Invitrogen account number or uIOH number for each protein that can be used at Invitrogen's website to find additional information about that protein. Reproducibility of protein binding for a test ligand within a single v4 assay run was assessed as follows: (1) All proteins were spotted in duplicate at each location; (2) some v4 proteins were spotted at several different locations, either as slight variants containing small amino acid variations of other v4 proteins or as larger variants containing only partial sequences of the full-length protein whose name they bear in the searchable list; (3) some kinases, recognized by a uIOH number starting with P or PV, have His tags rather than GST tags, and generally adhered to the nitrocellulose-coated glass slides at higher concentrations than GST-tagged proteins. In conjunction with the KS v4 studies, a custom plate containing SEMA3A-Fc, spotted in duplicate in a four-step concentration gradient (5, 10, 20, and 40 ng/μL), was also prepared. 

Printed plates were washed with PB and then blocked with SMI buffer (50 mM Tris [pH 7.5], 5 mM Mg₂SO₄, 0.1% Tween-20, and 1% BSA, with the 1% BSA acting as the blocking agent). They were then exposed to 1.5, 15, or 150 ng/μL of biotinylated test GAG suspended in SMI buffer for 90 minutes at 4°C. The arrays were washed three times, 5 minutes per wash, with 5 mL SMI buffer at 4°C. Parallel negative (buffer alone) and positive (100 nM Alexa Fluor 647-conjugated staurosporine, known to bind one of the proteins in the protoarray) control arrays were run simultaneously. Arrays were exposed to 1 μg/mL Alexa Fluor 647-conjugated streptavidin in SMI for 30 minutes at 4°C. They were washed three times, 5 minutes per wash, in 5 mL SMI buffer at 4°C and then quickly rinsed in water before drying and scanning (Axon 4000B fluorescence scanner; Molecular Devices, Sunnyvale, CA). Data were acquired (GenePix Pro 6.0 software; Molecular Devises) and analyzed (ProtoArray Prospector software; Invitrogen). 

Four criteria were used to define a significant interaction between a v4 protein and bGAGs: (1) A Z-score, or normalized fluorescence signal, greater than 3.0 SD above the mean human protein signal for nonreactive proteins in the test array, and a Z-score less than 1.0 in the negative control assay; (2) a Z-factor greater than 0.5, meaning that the signal was at least two times greater than the negative control; (3) a Cl P < 0.05, where the Cl P-value assigns a probability that the signal is derived from the distribution of signals arising from a set of defined negative controls; (4) a replicate spot CV less than 50%, where CV is the coefficient of variation for the assay from duplicate spots, determined by dividing the SD of the spot signals for all the spots by their mean spot intensity. A Z-score > 3 required a signal of ∼15,600 RFU (relative fluorescent units) or greater to be classified as a positive binding protein for KS, and a signal of ∼13,000 RFU or greater to be classified as a positive binding protein for CSA. For HA, there were only two proteins that met these stringent threshold criteria, and thus a Z-score > 1 was used to identify additional potential hits. The custom SEMA3A-Fc plate, probed with bKS at the same time as the KS-v4 protoarray, was scored based on the presence of signal that was significantly higher than the signal seen when the custom array was probed with detection reagent alone (result included as the final listing in Table 2). 

Table 2.

 

Proteins that Interact with Highly Sulfated Keratan Sulfate GAG Chains

Protein	Signal	Database ID	Invitrogen ID	Protein Description
ABL1	43,459	NP_005148.2	P3049	v-abl Abelson murine leukemia viral oncogene homologue 1
DNAJC8	41,722	NP_055095.2	NM_014280.1	DnaJ (Hsp40) homologue, subfamily C, member 8
PQBP1	39,994	NP_005701.1	NM_005710.1	Polyglutamine binding protein 1
ABL1	38,005	NP_005148.2	PV3864	v-abl Abelson murine leukemia viral oncogene homologue 1, E255K
RPS6KB2	37,356	NP_003943	PV3831	Ribosomal protein S6 kinase, 70kDa, polypeptide 2
TBK1	37,330	NP_037386.1	PV3504	TANK-binding kinase 1
PLK1	36,949	NP_005021.2	PV3501	Polo-like kinase 1 (Drosophila)
ABL1	36,821	NP_005148.2	PV3866	v-abl Abelson murine leukemia viral oncogene homologue 1, T351I
CAPRIN1	35,834	NP_005889.3	NM_005898.4	Cell cycle associated protein 1
COL23A1	35,723	NP_775736.2	BC042428.1	Collagen, type XXIII, alpha 1
ABL1	35,400	NP_005148.2	PV3865	v-abl Abelson murine leukemia viral oncogene homologue 1, G250E
AXL	35,361	NP_068713	PV3971	AXL receptor tyrosine kinase
NUAK1	35,266	NP_055655.1	PV4127	NUAK family, SNF1-like kinase 1
ABL1	38,005	NP_005148.2	PV3863	v-abl Abelson murine leukemia viral oncogene homologue 1, Y253F
PDAP1	34,768	NP_055706	NM_014891.1	PDGFA associated protein 1
AFF4	34,570	NP_055238.1	BC025700.1	AF4/FMR2 family, member 4
CCDC43	34,393	NP_653210	BC047776.2	Coiled-coil domain containing 43
EPB49	34,349	NP_001107608	BC052805.1	Dematin: erythrocyte membrane protein band 4.9
CHUK	34,299	NP_001269	PV4310	Conserved helix-loop-helix ubiquitous kinase
F11R	34,234	NP_058642.1	NM_016946.3	F11 receptor
CDC2	34,115	NP_001777.1	PV3292	Cell division cycle 2, G1 to S and G2 to M
MYLK2	32,835	NP_149109.1	PV3757	Myosin light chain kinase 2, skeletal muscle
RAD51AP1	32,494	NP_006470.1	NM_006479.2	RAD51 associated protein 1
CCNT1	32,383	NP_001231	PV4131	Cyclin T1
KCNAB2	32,108	NP_003627.1	NM_003636.1	Potassium voltage-gated channel, shaker-related subfamily, beta 2
KIT V654A	31,817	NP_000213	PV4132	v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homologue
DIDO1	31,785	NP_149072.1	BC000770.1	Death inducer-obliterator 1
RPS6KA1	31,745	NP_002944.2	PV3680	Ribosomal protein S6 kinase, 90kDa, polypeptide 1
CCDC55	31,679	NP_115517.1	NM_032141.1	Coiled-coil domain containing 55
ADD2	31,550	NP_059522.1	BC065525.1	Adducin 2 (beta)
PRR15	31,499	NP_787083	NM_175887.2	Proline rich 15
BTK	31,423	NP_000052	PV3363	Bruton agammaglobulinemia tyrosine kinase
ITGA6	31,406	NP_000201.2	NM_000210.1	Integrin alpha 6
PHF15	31,163	NP_056103	BC00492.1	PHD finger protein 15
UBE2S	31,160	NP_055316	BC004236.2	Ubiquitin-conjugating enzyme E2S
RPS6KA2	30,817	NP_066958.2	PV3846	Ribosomal protein S6 kinase, 90kDa, polypeptide 2
STK25	30,743	NP_006365.2	PV3657	Serine/threonine kinase 25 (STE20 homologue, yeast)
HSPC148	30,545	NP_057487.2	BC040946.1	CWC 15 homologue (S. cerevisiae)
MAP2	30,395	NP_002365.3	NM_031845.1	Microtubule-associated protein 2
NTRK3	30,375	NP_002521	PV3617	Neurotrophic tyrosine kinase, type 3
BMX	30,146	NP_001712	PV3371	BMX non-receptor tyrosine kinase
NEK2	29,418	NP_002488	PV3360	NIMA (never in mitosis gene a)-related kinase 2
NEK3	29,363	NP_689933	PV3821	NIMA (never in mitosis gene a)-related kinase 3
ARHGAP17	29,340	NP_001006635.1	NM_001006634.1	Rho GTPase activating protein 17
DAPK2	29,080	NP_055141.2	PV3614	Death-associated protein kinase 2
MAP2K2	29,037	NP_109587.1	PV3615	Mitogen-associated protein kinase kinase 2
UBXD3	28,418	NP_689589	NM_152376.2	UBX domain containing 3
PAK4	28,292	NP_005875.1	NM_005884.2	p21(CDKN1A)-activated kinase 4
ANKRD50	27,779	NP_065070	BC024725.1	Ankyrin repeat domain 50
ABL2	27,620	NP_009298.1	PV3266	v-abl Abelson murine leukemia viral oncogene homologue 2
CACNB1	27,417	NP_000714.3	NM_000723.3	Calcium channel, voltage-dependent, beta 1 subunit
TSSK2	27,281	NP_443732.3	PV3622	Testis-specific serine kinase 2
SCEL	27,277	NP_659001.1	BC020726.1	Sciellin
SERBP1	27,009	NP_001018077.1	BC003049.1	SERPINE1 mRNA binding protein
RPS6KA4	26,944	NP_003933	PV3782	Ribosomal protein S6 kinase, 90kDa, polypeptide 4
TBC1D10C	26,829	NP_940919.1	NM_198517.2	TBC1 domain family, member 10C
KIAA1706	26,773	NP_085139.2	NM_030636.1	KIAA1706 protein
WIBG	26,756	NP_115721.1	NM_032345.1	Within bgcn homologue (Drosophila)
E2F8	26,667	NP_078956.2	BC028244.1	E2F transcription factor 8
FKBP3	26,667	NP_002004	NM_002013.2	FK506 binding protein 3
TCEAL2	26,595	NP_525129	NM_080390.3	Transcription elongation factor A (SII)-like 2
VRK3	26,225	NP_057524.3	NM_016440.1	Vaccinia related kinase 3
AURKB	25,925	NP_004208.2	PV3970	Aurora kinase B
STK24	25,564	NP_003567.2	PV3560	Serine/threonine kinase 24 (STE20 homologue, yeast)
CSNK1G2	25,425	NP_001310.3	PV3499	Casein kinase 1, gamma 2
DAPK3	25,279	NP_001339	PV3686	Death-associated protein kinase 3
DCX	25,123	NP_835365.1	NM_178152.1	Doublecortin: Doublecortex; lissencephaly, X-linked
SMITH	25,068	NP_066972.1	Smith	SMITH antigen (Sm)
SCYE1	24,980	NP_001135887.1	BC01451.1	Small inducible cytokine subfamily E, member 1, endothelial monocyte activating
PRKG2	24,956	NP_006250.1	PV3973	Protein kinase, cGMP-dependent, type II
PRR16	24,762	NP_057728.1	BC038838.1	Proline rich 16
DDX42	24,754	NP_031398.2	BC015505.1	DEAD (Asp-Glu-Ala-Asp) box polypeptide 2
NEK4	24,659	NP_003148.2	PV4315	NIMA (never in mitosis gene a)-related kinase 4
TTK	24,534	NP_003309.2	PV3792	TTK protein kinase
WWP2	24,484	NP_008945.2	BC000108.1	WW domain containing E3 ubiquitin protein ligase 2
ARPP-19	24,474	NP_006619.1	NM_006628.4	Cyclic AMP phosphoprotein, 19kD
CSNK1E	24,328	NP_001885.1	PV3500	Casein kinase 1, epsilon
ABLIM1	24,223	NP_002304.3	BC002448.2	Actin binding LIM protein 1
MAP2K6	24,152	NP_002749.2	PV3318	Mitogen-associated protein kinase kinase 6
PAK1	23,993	NP_002567	PV3820	p21/Cdc42/Rac1-activated kinase 1 (STE20 homologue, yeast)
ANKS4B	23,921	NP_665872.2	NM_145865.1	Ankyrin repeat and sterile alpha motif domain containing 4B
DCAMKL2	23,892	NP_689832	PV4297	Doublecortin and CaM kinase-like 2
ZNF706	23,792	NP_057180.1	NM_016096.1	Zinc finger protein 706
IRS1	23,732	NP_005535.1	BC053895.1	Insulin receptor substrate 1
FAM128A	23,684	NP_001078834.1	BC018206.1	Family with sequence similarity 128, member A
ACTR1B	23,582	NP_005726.1	NM_005735.2	Centractin beta (yeast): actin-related protein homologue B
KIAA1143	23,465	NP_065747.1	BC016790.1	KIAA1143
CTNNA1	23,428	NP_001894.2	BC031262.1	Catenin (cadherin-associated protein), alpha 1, 102kDa
EPHA2	23,315	NP_004422.2	PV3688	Ephrin receptor A2
HOMER2	23,290	NP_004830.2	BC012109.1	Homer homologue 2 (Drosophila)
NTRK2	23,143	NP_006171	PV3616	Neurotrophic tyrosine kinase, receptor, type 2
TCP10L	23,126	NP_653260.1	NM_144659.1	t-complex 10 (mouse)-like
LIMCH1	22,817	NP_055803.2	BC023546.2	LIM and calponin homology domains 1
YES1	22,580	NP_005424.1	P3078	v-yes-1 Yamaguchi sarcoma viral oncogene homologue 1
EIF5	22,457	NP_001960.2	BC032866.2	Eukaryotic translation initiation factor 5
KIF2C	22,401	NP_006836.2	BC014924.1	Kinesin family member 2C
EIF3S4	22,358	NP_003746.2	BC014924.1	Eukaryotic translation initiation factor 3, subunit G
SRPK1	22,335	NP_003128.3	PV4215	SFRS protein kinase 1
TRPT1	22,333	NP_113660.1	NM_031472.1	tRNA phosphotransferase 1
SMTNL2	22,279	NP_940903.2	NM_198501.1	Smoothelin-like 2
LARP4	22,254	NP_954658.1	NM_199188.1	La ribonucleoprotein domain family, member 4
NMT1	22,231	NP_066565.1	NM_021079.2	N-myristoyltransferase 1
DAP	22,223	NP_004385.1	NM_004394.1	Death-associated protein
MAPRE1	22,209	NP_036457.1	NM_012325.1	Microtubule-associated protein, RP/EB family, member 1
CSNK1G3	22,208	NP_004375.2	PV3838	Casein kinase 1, gamma 3
NTRK1	22,192	NP_002520.2	PV3134	Neurotrophic tyrosine kinase, receptor, type 1
KIF3B	22,121	NP_004789.1	NM_004798.2	Kinesin family member 3B
EBAG9	21,974	NP_004206.1	NM_004215.2	Estrogen receptor binding associated, antigen 9
FGFR1OP	21,817	NP_008976.1	NM_007045.2	FGFR1 oncogene partner
NPM1	21,797	NP_002511.1	BC021983.1	Numatrin: nucleophosmin (nucleolar phoshoprotein B23)
LARP6	21,777	NP_060827.2	NM_018357.2	La ribonucleoprotein domain family, member 6
LYN	21,548	NP_002341.1	P2907	v-yes-1 Yamaguchi sarcoma viral oncogene homologue
RTF1	21,542	NP_055953.3	NM_105138.2	Rtfl, Paf1/RNA polymerase II complex component, homolog
FES	21,524	NP_001996.1	PV3354	Feline sarcoma oncogene
HMGB1	21,497	NP_002119.1	NM_002128.2	High-mobility group box 1
SLC4A1AP	21,473	NP_060628.2	NM_018158.1	Solute carrier family 4 (anion exchanger) member 1, adaptor protein
NMT2	21,461	NP_004799.1	BC006376.1	N-myristoyltransferase 2
G3BP1	21,423	NP_938405.1	NM_198395.1	GTPase activating protein (SH3 domain) binding protein 1
IRAK4	21,417	NP_057207	PV3362	Interleukin-1 receptor-associated kinase 4
PSRC1	21,391	NP_116025.1	NM_032636.2	Proline/serine-rich coiled-coil 1
RBM8A	21,196	NP_005096.1	NM_005105.2	RNA binding motif protein 8A
EPHA8	21,107	NP_065387	PV3844	Ephrin receptor A8
LMNA	21,093	NP_733821.1	BC033088.1	Lamin A/C
SYTL2	20,914	NP_116561.1	NM_032943.2	Synaptotagmin-like 2
TSSK1B	20,661	NP_114417	PV3505	Testis-specific serine kinase 1B
RGS14	20,566	NP_006471.2	NM_006480.4	Regulator of G-protein signaling 14
AURKA	20,210	NP_940839	PV3612	Aurora kinase A
NKIRAS1	19,916	NP_065078.1	NM_020345.3	NFKB inhibitor interacting RAS-like 1
PPID	19,915	NP_005029.1	NM_005038.1	Cyclophilin D: peptidylprolyl isomerase D
MARK4	19,852	NP_113605	PV3851	MAP/microtubule affinity-regulating kinase 4
NEK1	19,713	NP_006613	PV4202	NIMA (never in mitosis gene a)-related kinase 1
LENG1	19,379	NP_077292.1	NM_024316.1	Leukocyte receptor cluster member 1
SETBP1	19,196	NP_056374.2	BC062338.1	SET binding protein 1
PPP1R8	19,041	NP_054829.2	NM_014110.3	Protein phosphatase 1, regulatory (inhibitor) subunit 8
CFDP1	18,952	NP_006315.1	NM_006324.1	Craniofacial development protein 1
DDX19B	18,769	NP_009173.1	NM_007242.3	DEAD (Asp-Glu-Ala-Asp) box polypeptide 19B
ZAK	18,704	NP_598407	PV3882	Sterile alpha motif and leucine zipper containing kinase AZK
ASXL1	18,629	NP_056153.2	BC064984.1	Additional sex-combs like 1 (Drosophila)
FGFR1	18,588	NP_000595	PV3146	Fibroblast growth factor receptor 1 (fms-related tyrosine kinase 2)
ERBB2	18,556	NP_004439	PV3366	v-erb-b2; neuro/glioblastoma derived oncogene homologue (avian)
PTK2B	18,455	NP_775266.1	BC036651.2	Protein tyrosine kinase 2 beta
DDX54	18,433	NP_001104792.1	BC001131.1	DEAD (Asp-Glu-Ala-Asp) box polypeptide 54
MEOX1	18,335	NP_004518.1	NM_004527.2	Mesenchyme homeobox 1
OCEL1	18,291	NP_078854.1	NM_024578.1	Occludin/ELL domain containing 1
ING5	18,217	NP_115705.2	NM_032329.4	Inhibitor of growth family, member 5
ASXL2	18,195	NP_060733.4	BC042999.2	Additional sex-combs like 2 (Drosophila)
TFPT	18,176	NP_037474.1	BC001728.1	TCF3 (E2A) fusion partner (in childhood leukemia)
IFRD2	18,021	NP_006755.4	BC001327.1	Interferon-related developmental regulator 2
MARK2	17,836	NP_059672.2	PV3878	MAP/microtubule affinity-regulating kinase 2
SIRPG	17,784	NP_061026.2	BC064532.1	Signal-regulatory protein gamma
KDR	17,770	NP_002244	PV3660	Kinase insert domain receptor (a type III receptor tyrosine kinase)
MRPL1	17,755	NP_064621.3	BC015109.1	Mitochondrial ribosomal protein L1
RPAP3	17,682	NP_078880.1	BC056415.1	RNA polymerase II associated protein 3
FAM50A	17,629	NP_004690.1	NM_004699.1	Family with sequence similarity 50, member A
UBE2E2	17,443	NP_689866.1	NM_152653.1	Ubiquitin-conjugating enzyme E2E 2
CDKN1B	17,376	NP_004055.1	NM_004064.2	Cyclin-dependent kinase inhibitor 1B (p27, Kip1)
PKN2	17,294	NP_006247	PV3879	Protein kinase N2
LIG3	17,224	NP_039269.2	NM_013975.1	Ligase III, DNA, ATP-dependent
MESDC2	17,222	NP_055969.1	BC012746.1	Mesoderm development candidate 2
TRUB1	17,203	NP_631908.1	NM_139169.2	TruB pseudouridine (psi) synthetase homologue 1
MAPKAPK3	17,148	NP_004626.1	BC001662.1	Mitogen-activated protein kinase-activated protein kinase 3
SRC	17,080	NP_005408	P3044	v-src sarcoma (Schmidt-Ruppin A-2) viral oncogene homologue
BAIAP2	16,925	NP_059344.1	BC014020.1	BAI1-associated protein 2
STAC	16,897	NP_003140.1	BC020221.1	SH3 and cystein rich domain
TPPP	16,891	NP_008961.1	NBM_007030.1	Tubulin polymerization promoting protein
FMNL1	16,811	NP_005883.2	BC021906.1	Formin-like 1
HMGN1	16,811	NP_004956.5	NM_004965.3	High-mobility group nucleosome binding domain 1
GLB1L	16,727	NP_078782.3	NM_024506.3	Galactosidase, beta 1-like
FLT3	16,478	NP_004110	PV3967	Fms-related tyrosine kinase 3
PIP4K2C	16,448	NP_079055.3	NM_024779.2	Phosphotidylinositol-5-phosphaste 4-kinase, type II, gamma
GSK3B	16,258	NP_002084	PV3365	Glycogen synthetase kinase 3 beta
ODF2	16,221	NP_002531.3	BC010629.1	Outer dense fiber of sperm tails 2
ZHX2	16,169	NP_055758.1	NM_014943.3	Zinc fingers and homeoboxes 2
DNM2	16,068	NP_001005360.1	BC054401.1	Dynamin 2
LAD1	16,058	NP_005549.2	NM_005558.2	Ladinin 1
COASY	16,042	NP_001035994.1	BC020985.1	Coenzyme A synthetase
BRD3	15,978	NP_031397.1	BC032124.1	Bromodomain containing 3
HCK	15,939	NP_002101	P2908	Hemopoietic cell kinase
EFCAB4A	15,902	NP_775855.3	BC033196.1	EF-hand calcium binding domain 4A
MAP3K10	15,869	NP_002437	PV3877	Mitogen-activated protein kinase kinase kinase 10
BCAR3	15,837	NP_003558.1	BC039895.1	Breast cancer anti-estrogen resistance 3
COIL	15,815	NP_004636.1	NM_004645.1	Coilin
PCDHGC3	15,791	NP_115779.1	NM_032403.1	Protocadherin gamma subfamily C, 3
EPHA3	15,698	NP_005224	PV3359	Ephrin receptor A3
AMMECR1L	15,601	NP_113633.2	NM_031445.1	AMME chromosomal region gene 1 -like
ZMYM3	15,595	NP_005087.1	BC013009.1	Zinc finger, MYM-type 3
SEMA3A-Fc	5,913	NP_006071.1		Semaphorin 3A, extracellular domain

  n = 217 of 8,268 screened.

To further investigate binding of corneal stromal GAGs to ECM nerve-related molecules, we purchased selected extracellular epitopes of growth factors, nerve growth cone guidance ligands, and their receptors from R&D Systems, and asked Invitrogen to create a custom nerve-related protein array and assay binding of bKS, bCSA, and bHA to these proteins using their standard protocol. The protein epitopes used in this custom array are shown in Table 1. To maximize their tertiary structure and biological activity, R&D Systems expresses many extracellular epitopes as fusion proteins with an added human IgG1 Pro100-lys330 domain (Accession # P01857). To assess whether GAGs react with this IgG1 epitope, or whether the fused IgG1 epitope might block the reaction of a GAG with a target protein, Invitrogen conducted a small test protoarray using bKS as the representative GAG, the IgG1 domain alone, NTR3, NTR3-Ig, EFNA1-Ig, and Semaphorin 3A-Ig. bKS reacted with NTR3-Ig, EFNA1-Ig, and SEMA3A-Ig, but not with IgG1 alone or with NTR3 from which the IgG1 domain had been cleaved. These results suggest that GAGs do not bind the IgG1 epitope, and that the presence of the IgG1 fusion domain does not block the reaction of a GAG with a target protein. For thoroughness, both the IgG1-tagged version and the IgG1-cleaved version for every protein for which both forms were available were included in the custom array. Each protein was spotted in duplicate from solutions of 10, 25, or 50 ng/μL in PB and bGAGs were used at 1.5, 15, or 150 ng/μL in SMI buffer. Proteins were scored as GAG binders if the difference between the RFU value measured on the negative control assay (buffer only) and the assay probed with the bGAG was greater than 2000 RFU for any concentration of the probe GAG. 

Mimecan Protoarray

The KS core protein mimecan was used by Invitrogen to probe its v4 protoarray, according to the same format as described for the GAG protoarrays, except that the test ligand was purified mouse mimecan, tested at 5 and 50 ng/μL. After incubation of the protoarray slides with mimecan, the slides were washed, incubated with rat anti-mouse mimecan, and then with Alexa-Fluor 647-labeled goat anti-rat IgG. For the mimecan protoarray, the negative control was buffer alone, and the positive control was V5-his tagged yeast calmodulin kinase 1, developed with Alexa-Fluor 647-labeled anti-V5 antibody. The reaction of calmodulin kinase 1 with calmodulin protein in the v4 protoarray served as an indicator that the assay functioned properly. Data were acquired (GenePix Pro 6.0 software; Molecular Devices) and analyzed (ProtoArray Prospector software; Invitrogen) as for the GAG protoarrays. The criteria for a positive reaction between mimecan and a v4 protein were as described for the v4 bGAG assays. 

Surface Plasmon Resonance

SA sensor chips were conditioned with three consecutive 1-minute injections of 1 M NaCl in 50 mM NaOH before ligand was immobilized. bKS was dissolved in 0.01 M HEPES (pH 7.4), 0.15 N NaCl, and 0.005% (vol/vol) HSB-P buffer at 0.01 mg/mL and immobilized on flow cell (Fc) lanes 2 to 4 of an SA chip, using an injection time of 1 minute at a flow rate of 10 μL/min. A control of biotin alone was then immobilized on Fc lane 1 of each SA chip, with a saturated solution of biotin in HSB-P buffer under the same protocol. The four lanes of the chip were then washed with HSB-P buffer according to the manufacturer's protocol (GE/Biacore). Successful immobilization of biotinylated-KS was confirmed by a 290 to 300 RU increase (where one RU is equivalent to 1 pg protein/mm² on the sensor's surface) in the appropriate sensor chip lane. RU increase for the biotin lane was less than 50 RU. All data were blanked against the biotin control reference Fc (Fc1) and double-referenced against an HBS-P buffer injection (SLIT2 at 0 nM). Protein samples were diluted in HSB-P. Protein solutions were passed over the biotin-KS-coated sensor surface (KINJECT mode, model 3000; GE/Biocore) at a flow rate of 20 μL/min with a total injection volume of 100 μL, which flowed over the chip lanes sequentially from lane 1 to 4. After each run of a given protein concentration, dissociation was conducted in HSB-P buffer for 600 seconds, and regeneration of the sensor surface was conducted in 1 M NaCl in 50 mM NaOH. The response was monitored as a function of time (sensorgram) at 25°C, using automated in-line reference subtraction to subtract biotin control ligand binding in Fc1 from biotin-KS ligand binding in Fc2, 3, and 4. The kinetic parameters k _a and k _d (association and dissociation rate constants, respectively) were analyzed simultaneously using global fit. System software (BIAevaluation software ver 3.1; GE/BIAcore) simultaneously fitted the sensorgrams obtained at different concentrations of ligand, fixing each kinetic parameter to a single value for each set of experimental data. An apparent equilibrium dissociation constant (K _D) was calculated as the ratio k _d /k _a with the maximum capacity (R _max) of the surface floated during the fitting procedure. 

Results

KS, CSA, and HA Binding of Human Cellular Proteins

Bovine corneal KS binding to a very wide variety of proteins was assessed by Invitrogen with their v4 Protoarray, with more than 8000 human proteins or protein variations. A complete and searchable list of all v4 proteins, including an Invitrogen identification number, their location on v4 protoarray plates, an estimation of their concentration if they are GST-tagged proteins, and their exact amino acid sequence may be viewed at their website (http://hdl.handle.net/2097/1716). When screened at 150 ng/μL (10 μM), KS gave positive binding responses with 217 v4 proteins. With the exception of ABL1 homologue 1 variant a, each protein that KS bound is listed only once in Table 2, at the level of its highest binding, even when alternative forms or amounts of the protein, present in the different locations on the protoarray slides, also interacted with KS. An expanded version of Table 2, showing all proteins at all locations on the v4 protoarray slides that reacted with KS can be viewed at http://hdl.handle.net/2097/1716. Extracellular SEMA3A linked to IgG₁, obtained from R&D Systems, also was included in this analysis, and the response of 150 ng/μL bKS to 40 ng/μL SEMA3A-Fc is listed at the end of Table 2. The reproducibility of the KS-v4 protein binding results is supported in several ways: (1) The coefficient of variation of bKS binding to duplicate spots of a given protein at any one location in the v4 protoarray plates was never more than 12% for any positive KS-binding protein (see expanded Table 2 at http://hdl.handle.net/2097/1716); (2) binding of bKS to ABL1 homologue 1 variant a, occurred at all five duplicate locations of ABL1 homologue 1 variant a in the v4 protoarray in which the variation between the ABL1 homologue 1 variant a locations was only 1 amino acid, with very similar binding strengths of 43,459, 38,005, 36,821, 35,400, and 35,199 (Table 2). In contrast, bKS did not bind severely truncated versions of ABL1 homologue 1 variant a spotted at two other locations (see locations 2178 and 2878 in the total v4 proteins list at http://hdl.handle.net/2097/1716); and (3) 150 ng/μL bKS bound SEMA3A-Fc plated at 40 ng/μL in the first small custom protoarray shown at the end of Table 2 (5913 RFU) at a level very similar to its binding to SEMA3A-Fc plated at 50 ng/μL in the second larger custom protoarray shown in Table 5 (8203 RFU). 

Nine of the 217 respondents were still positive when the screening was conducted at 15 ng/μL KS (ABL1, TBK1, PLK1, AXL1, NUAK1, CHUK, CCNT1, KCNAB2, and RPS6KA1), whereas only KCNAB2 was positive when the screening was conducted at 1.5 ng/μL KS. Of proteins that bound 150 ng/μL KS, 75 were protein kinases, comprising several families: (1) 10 cytoplasmic tyrosine kinases: ABL2, ABL1, BMX, BTK, PTK2B, SRC, FES, HCK, YES, and LYN; (2) 15 cytoplasmic domains of receptor tyrosine kinases: EPHA2, EPHA3, EPHA8, Ntrk1, Ntrk2, Ntrk3, wt KIT, two KIT point mutations, CSF1R, FGFR1, KDR, one FLT3 point mutation, ERBB2, and AXL; (3) four MAP kinase family members: MAP2K2, MAP2K6, one MAP2K6 point mutation, and MAP3K10; (4) four NIMA-related kinases: NEK1, NEK2, NEK3, and NEK4; (5) casein kinases epsilon 1, gamma 2, and gamma 3; (6) aurora kinases AURKA and AURKB; (7) death-associated protein kinases DAPK2, DAPK3; (8) ribosomal protein S6 kinases RPS6KB2, RPS6KA1, RPS6KA2, and RPS6KA4; (9) testis-specific serine kinases TSSK1B and TSSK2; (10) MAP/microtubule affinity regulating kinases MARK4 and MARK2; and (11) STE20-related kinases PAK1, STK24, and STK25. In addition, there were 13 membrane or secreted proteins that bound to 150 ng/μL KS: COL23A1, KCNAB2, ITGA6, CACNB1, EBAG9, SIRPG, PCDHGC3, SCYE1, F11R, CAPRIN1, ADD2, SYTL2, and BAIAP2. KS also bound 18 proteins involved in actin or microtubule binding or regulation: ABLIM1, CTNNA1, EPB39, LIMCH1, PAK4, ADD2, MAP2, AURKA, AURKB, DMN2, TPPP, MARK2, MARK4, MAPRE1, KIF2C, KIF3B, CTNNA1, and ACTR1B; and 20 proteins known or thought to be involved in nervous tissue function: Ntrk1, Ntrk2, Ntrk3, EPHA2, EPHA3, EPHA8, CAPRIN, BAIAP2, MAP2K10, SYTL2, DCX, HOMER2, TPPP, GSK3B, KCNAB2, MAP2, MARK4, PQBP1, ACTR1B, and extracellular SEMA3A. 

In comparison, sturgeon notochord CSA bound only 23 v4 proteins when the screening was conducted at 150 ng/mL, listed in Table 3. Of these, seven (CSNK1A1, CHUK, PAK1, RPS6KA5, FLT1, MAP4K4, and GADD45GIP1) displayed their highest binding at a bCSA screening concentration of 15 ng/μL, whereas only three (PLK1, HADH, and NUDT16L1) were still positive when the screening was performed at 1.5 ng/μL. Of the CSA-binding proteins, 10 were kinases: 2 were independent preparations of PLK1, CHUK, PAK1, RPS6KA5, FLT1, MAP4K4, CSNK1D, and PAK4; 2 were nuclear proteins identified with the cell cycle or DNA or RNA binding: NUDT21 and GADD45GIP1; and 2 were cell surface or scaffold proteins: ITGA6 and HOMER2. Again, reproducibility of bCSA binding in this proteomics protocol is supported in several ways. (1) The coefficient of variation between adjacent duplicate protein spots never exceeded 16% (data not shown). (2) PLK1 was spotted in four different duplicate locations in the v4 protoarray. One duplicate PLK1 version at one location was the highest binder of bCSA (64679RFU; Table 3) and had a His-tag instead of the usual GST tag on most v4 proteins, which caused it to be printed at a higher concentration than GST-tagged proteins. A second PLK1 duplicate spot location was the lowest reported binder of bCSA (19806RFU; Table 3), and was plated at a lower concentration than was retained at the highest binding location. PLK1s at a third and fourth location were plated at concentrations one fourth that of the concentration at the second location, the version at the fourth location was severely truncated, and neither location gave a positive signal for bCSA binding. These results demonstrate a concentration dependency for CSA-protein binding. 

Table 3.

 

CSA Interaction with Invitrogen Proteins

Protein	Signal	Accession No.	Invitrogen ID	Protein Description
PLK1	64,679	NP_005021	PV3501	Polo-like kinase 1 (Drosophila)
ITGA6	62,823	NP_000201.2	NM_000210.1	Integrin, alpha 6, transcript variant 2
HADH	61,418	NP_005318.2	BC000306.1	Hydroxyacyl-Coenzyme A dehydrogenase
KCNAB1	60,955	NP_751892.1	NM_172160.1	Potassium voltage-gated channel, shaker-related subfamily, beta 1
NUDT16L1	60,163	NP_115725.1	NM_032349.1	Nudix (nucleoside diphosphate linked moiety X)-type motif 16–1
CSNK1A1	59,748	NP_001883	PV3850	Casein kinase 1, alpha 1
CHUK	54,364	NP_001269	PV4310	Conserved helix-loop-helix ubiquitous kinase
COASY	47,848	BC020985.1	BC020965.1	Coenzyme A synthase
KRTAP13-1	47,443	NP_853630.2	NM_181599.1	Keratin associated protein 13–1
PAK1	39,059	NP_002567	PV3820	p21/Cdc42/Rac1-activated kinase 1 (STE20 homologue, yeast)
RPS6KA5	38,929	NP_004746.2	PV3681	Ribosomal protein S6 kinase, 90kDa, polypeptide 5
FLT1	37,750	NP_002010	PV3666	Fms-related tyrosine kinase 1 receptor
MAP4K4	30,089	NP_004825	PV3687	Mitogen-activated protein kinase kinase kinase kinase 4
KIAAA0515	29,800	BC012289.1	BC012289.1	KIAAA0515
HOMER2	28,772	NP_004830.2	BC01209.1	Homer homologue 2 (Drosophila)
GADD45GIP1	27,379	NP_443082.2	BC013039.1	Growth arrest and DNA-damage-inducible gamma interacting prot.
CSNK1D	26,494	NP_620693	PV3665	Casein kinase 1, delta
SLAIN2	26,324	NP_065897.1	BC031691.2	SLAIN motif family, member 2
C11orf66	25,433	NP_659454.1	BC053995.1	Chromosome 11, open reading frame 66
ACOX1	24,045	NP_004026.2	BC010425.1	Acyl-coenzyme A oxidase 1, palmitoyl
PAK4	23,005	NP_005875.1	NM_005884.2	p21(CDKN1A)-activated kinase 4
NUDT21	22,423	NP_008937.1	NM_007006.1	Nudix (nucleoside diphosphate linked moietyX)-type motif 21
PLK1	19,806	NP_005021	NM_005030.2	Polo-like kinase 1 (Drosophila)

  n = 23 of 8,268 proteins screened.

In contrast, chicken cockscomb HA bound only six of the >8000 v4 proteins (Table 4). Of these six proteins, three (KRTAP13–1, IRS1, and HADH) achieved their highest binding at an HA screening concentration of 150 ng/μL, one (LGALS8) showed its highest binding at an HA screening concentration of 15 ng/μL, and two (GFAP and CABP4) displayed their highest binding at an HA screening concentration of 1.5 ng/μL. Overall, the signal intensity for all three bHA screenings was low compared with the signal intensities for bKS and bCSA screenings, with only KRTAP13-1 generating a signal intensity above the lowest positive v4 intensities obtained with bKS and bCSA. Again, the coefficient of variation between adjacent duplicate spots was never higher than 16% (data not shown), suggesting good binding reproducibility for bHA. In addition, although only one of the six v4 proteins bound by bHA was present more than once in the v4 protoarray, a complete version of galectin 8 (LGALS8) and a variant version lacking an internal segment of the protein and plated at approximately one tenth the concentration of the complete version, bHA bound the more concentrated complete version at a level significantly above background noise, but did not bind the much less concentrated variant lacking the internal segment. Whereas these results could reflect either the avidity of HA for LGALS8 or the ability of HA to discriminate between full length and internally truncated LGALS8, they suggest that the protoarray protocol used in our experiments allowed bHA to bind protein spots discriminately. 

Table 4.

 

HA Interaction with Invitrogen Proteins

Protein	Signal	Accession No.	Invitrogen ID	Protein Description
KRTAP13-1	38,193	NP_853630.2	NM_181599.1	Keratin-associated protein 13-1
IRS1	14,035	NP_005535.1	BC053895.1	Insulin receptor substrate 1
GFAP	4,593	NP_002046.1	NM_002055.1	Glial fibrillary acidic protein
HADH	4,296	NP_005318.2	BC000306.1	Hydroxyacyl-coenzyme A dehydrogenase
LGALS8	2,944	NP_006490.3	BC015818.1	Lectin, galactoside-binding, soluble, 8 (galectin 8)
CABP4	2,443	NP_660201.1	BC033167.1	Calcium binding protein 4

  n = 6 of 8268 proteins screened.

KS, CSA, and HA Binding of Extracellular Epitopes of Nerve-Related Proteins

The anterior cornea is one of the most highly innervated tissues on the surface of the human body. ⁵⁷ Because KS, CSA, and HA are present in very high concentrations in the corneal ECM, we asked whether these GAGs bind some of the neuroregulatory molecules or epitopes of membrane-bound neuroregulatory molecules whose mRNAs are known to be expressed in the developing cornea. ⁵³ Because of our previous localization of SLIT2 mRNA in the anterior cornea and the high amino acid sequence homology between SLIT orthologues, ⁵⁸ we included SLIT3 in our custom array, because it was the only SLIT orthologue R&D Systems had available at the time. 

Bovine corneal bKS bound 44 of the 85 nerve-related proteins or protein epitopes for the nerve-related custom protoarray analysis, listed in Table 5. In addition to the high similarity of bKS binding to SEMA3A-FC shown in both the first and the second custom array experiments, reproducibility of bKS nerve-related protein binding is supported by its concentration dependency. For the nerve-related protein custom arrays, all proteins were plated in duplicate spots at three or four separate locations at different concentrations and screened with three different GAG concentrations. For many KS interactant proteins, there was a concentration-dependent increase in RFU signal, with increasing signals being observed both with higher KS concentrations and at spots corresponding to higher concentrations of protein on the arrays (data not shown). 

Table 5.

 

KS Interaction with Nerve-Related Extracellular Molecules and Epitopes

Protein	Signal	Accession No.	Protein Description
SLIT3	41,488	Q9WVB4	Mus Ser27-His901; secreted axon guidance factor; Robo ligand.
DCC-Fc	39,177	NP_031857	Deleted in colorectal carcinoma, Mus Met1-Asn1097 + LINK + HumIgG Pro100-Lys 330; attractive response toward netrin-1.
ERBB4	33,716	Q15303	Hum Glu26-Arg649. Type 1 glycoprotein, Tyr protein kinase, and epidermal growth factor receptor.
SEMA3F-Fc	26,174	QO88632	Mus Ala19-Pro775 + link + HumIgG Pro100-Lys 330. Secreted, sema domain, Ig domain, short basic domain, neuroguidance.
ERBB2-Fc	25,936	NP_004439	Hum Thr23-Thr652 + Link + HumIgG Pro100-Lys 330. Type 1 glycopro, Tyr protein kinase and epidermal growth factor receptor.
EFNA3-Fc	19,524	NP_004943.1	Ephrin A3 extracellular domain Hum(aa 1–209) + link + HumIgG Pro100-Lys 330; binds Eph receptors in developing neural tissue.
ROBO1-Fc	19,284	055005	Hum Met1-Ala16 + rat Lys 19-Lle 560+link+HumIgG Pro100-Lys 330. Neural tissue negative Slit receptor.
EFNA1-Fc	16,592	P52793	Ephrin A1: Mus Asp19-Ser182 + link + HumIgG Prb100-Lys 330. Binds Eph receptors in developing neural tissue.
EFNB3-Fc	16,459	NP_001397.1	Ephrin B3: Hum Met1-Ser224 + Link + HumIgG Pro100-Lys 330. Binds Eph receptors in developing neural tissue.
EFNB1-Fc	14,809	Q544L9	EphrinB1: Mus Lys30-Ser229 + Link + HumIgG Pro100-Lys 330. Binds Eph receptors in developing neural tissue.
EFNB2-Fc	13,578	AAA82934	Ephrin B2: Mus Arg27-Ala227 + Link + HumIgG Pro100-Lys 330. Binds Eph receptors in developing neural and angiogenic tissue.
EPHB4-Fc	13,620	P54760	Ephrin B4: Mus Leu16-Ala539. Tyr kinase receptor, binds EFNB2. Inhibits cell-cell adhesion and angiogenesis.
SEMA6A-Fc	12,914	NP_065847.1	Semaphorin 6A: Hum Met1-Thr649 + Link + HumIgG Pro100-Lys 330. Anchored transmembrane signaling molecule. Negative axon guidance.
EPHA1-Fc	12,040	NP_005223.4	Ephrin A1: Hum Met1-Glu547 + Link + HumIgG Pro100-Lys 330. Tyr receptor kinase. Binds EFNA1 in neuronal and angiogenic tissue.
EPHB3-Fc	10,730	NP_034273.1	EphB3: Mus Met1-Thr530 + Link + HumIgG Pro100-Lys 330. Transmembrane Tyr kinase receptor; binds EFNB1, -B2, -B3.
ROB02-Fc	10,188	NP_002933.1	Hum Met1-Ala312 + Link + Hum ROB02 Pro313-Pro859 + HumIgG Pro100-Lys 330. Neural tissue negative Slit receptor; permissive of neurite outgrowth via ROB01-ROB02 interaction.
SEMA3E	10,117	0150041	Semaphorin 3E: Mus Thr25-Ser775(Arg557Alaand Arg560A1a) secreted, axon and vascular tip cell chemorepellant; binds plexin.
ERBB4-Fc	9,140	Q15303	HumGlu26-Arg649 + Link + HumIgG Pro100-Lys 330. Type I
			glycoprotein, tyr kinase and epidermal growth factor receptor.
SEMA3A-Fc	8,203	Q14563	Semaphorin3A: Hum Lys26-Val771 + link + HumIgG Pro100-Lys 330. Secreted, binds nerve cell Neuropilin1+PlexinA receptor complex; neurorepellant.
EFNA2-Fc	8,113	NP_031935.3	Mus Met1-Asn184 + Link + HumIgG Pro100-Lys 330. Binds EPH A2, -A3, -A4, -A5, -A6, -A7, -A8 receptors. Fens in neuroguidance and angiogenesis.
Ntrk3-Fc	7,250	NP_002521	Neurotrophic tyrosine kinase, type 3; binds secreted NT-3.
EPHA8-Fc	7210	NP_031965.2	Ephrin A8 Mus Met1-Arg540 + Link + HumIgG Pro100-Lys 330. Transmembrane Tyr kinase receptor; binds EFNA5, -A3,-A3.
SEMA6A	7,110	NP_065847.1	Semaphorin 6A: Hum Met1-Thr649. Anchored transmembrane signalling molecule. Negative axon guidance.
NTN1	6,956	AAC52971	Netrin1: Mus Va122-Ala603. Secreted neuroguidance molecule; interacts with DCC, UNC5 family, and neogenin receptors.
NRG1	6,607	NP_039253.1	Human neuregulin1, SMDF isoform, lacks Ig-like domain, Transmembrane domain, and cytoplasmic tail. Secreted neuroattractant.
SEMA6B-Fc	6,607	Q9H3T3	Semaphorin 6B: Hum CD33Met1-Ala16 + Hum Sem6B Leu26-Ser603 + Link + HumIgG Pro100-Lys 330. Type I glycoprotein with semaphorin domain. Binds Plexin4 w/o neuropilin1; axon repulsion.
EFNA5-Fc	6,397	NP_001953.1	Ephrin A5: Hum Met1-Asn203 + Link + HumIgG Pro100-Lys 330. Binds EPHsSA2-A8. Neurorepellant in neural tissue.
UNC5H2-Fc	5,989	NP_071543.1	Rat Met1-Asp373 + Link + HumIgG Pro100-Lys 330. Transmembrane NTN receptor that mediates neurorepulsion with NTN1.
EPHA6-Fc	5,886	NP_031964.2	Ephrin A6: Mus Met1-Gln546 + Link + HumIgG Pro100-Lys 330. Transmembrane EFN receptor.
NGFR-Fc	5,833	NP_001057.1	Low affinity nerve growth factor receptor: Hum p75 Met1-Asn250 + Link + HumIgG Pro100-Lys 330. Binds NGF, NT3, NT4.
PDGFAB	5,617	NP_036933.1	Dimerized rat PDGFA Ser87-Aug196 and rat PDGFB
		NP_113712.1	Ser74-Thr182. Binds to receptor kinases PDGF-α and PDGF-β.
EFNA4-Fc	5,590	NP_031936.2	Ephrin A4: Mus Met1-Gly176 + Link + HumIgG Pro100-Lys 330. Binds EPHA2-A7, -B1. Important in neurogenesis and angiogenesis.
EPHA7-Fc	5,435	NP_001116361.1	Ephrin A7: Mus Met1-Pro549 + Link + HumIgG Pro100-Lys 330. Tyrosine kinase receptor for EFNA1-A5. Neurogenesis and angiogenesis
NTN2	4,894	Q90923	Netrin 2: chicken CD33 Met1-Met17 + chicken NTN2 Ala16-Pro 581. Secreted neuroguidance molecule; interacts with DCC, UNC5 family, and neogenin receptors.
NTN4	4,785	AAG53651	Netrin 4: Hum Met1-Lys628. Secreted neuroguidance molecule; interacts with DCC, UNC5 family, and neogenin receptors.
EPHB1-Fc	4,184	XP_001081949.1	EPHB1: Hum CD33 Met1-Ala16 + Rat EPHB1 Met18-Gln538 + Link+HumIgG Pro100-Lys 330.Binds EFNA1, -A3, -A4, -B1, -B2, -B3.
EPHA3-Fc	3,279	NP_034270.1	Ephrin A3: Mus Met1-His541 + Link + HumIgG Pro100-Lys 330. Binds EFNA1–A5, -B1.
EPHA4-Fc	3,014	NP_031962.2	Ephrin A4: Mus Met1-Thr547 + Link + HumIgG Pro100-Lys 330. Binds EFNA1–A5, -B2, -B3.
SEMA4A-Fc	2,996	NP_071762.2	Semaphorin 4A: Hum Gly32-His 683 + Link + HumIgG Pro100-Lys 330. Transmembrane; fcns in immune and nervous tissue.
SEMA3F	2,377	O88632	Semaphorins F: Mus A1a19-Pro775(arg583A1a and Arg586A1a). Secreted, interacts with plexinA3 and neuropilin2, chemorepellant.

  n = 40 of 85 proteins screened. EFN A, GPI-membrane-anchored; EFN B transmembrane. Only membrane-bound or Fc-clustered ligands (Ephrins) can activate EPHs.

In agreement with the results of preliminary tests (see the Methods section), in 20 of the 29 sets of IgG₁-tagged/cleaved target protein epitopes, bKS bound the IgG₁-tagged version of the protein, but not the IgG₁-cleaved version. In two additional cases (SEMA3F and SEMA6A), KS bound the IgG₁-tagged version better than the IgG₁-cleaved version; in one case (ERBB4), KS bound the IgG₁-cleaved version better than the IgG₁-tagged version; and in six cases, KS bound neither version (Ntrk1, EPHA2, EPHA5, EPHB2, EPHB6, and NRP2). 

Of the 85 nerve-related epitopes in the protoarray, bKS bound most strongly the neuroregulatory ligand SLIT3; KS also bound the extracellular epitopes of SLIT receptors ROBO1 and -2. The second and third most strongly bound epitopes were deleted in colorectal carcinoma (DCC), a netrin receptor, and ERBB4, a neuregulin receptor. KS bound all the neuroguidance SEMA epitopes in the protoarray, but did not bind either of the SEMA receptor epitopes, neuropilin-1 or -2 (NRP1 or -2). There were no NRP coreceptor plexin epitopes available for the nerve-related protoarray. In the EFN/EPH neuroguidance family, KS bound the extracellular epitopes of all the EFN ligand epitopes in the protoarray, but only 9 of 13 EPH receptor epitopes in the protoarray. In the NTN/DCC/UNC5H family, KS bound all the NTN-soluble ligand epitopes in the array, the one DCC receptor epitope in the array, and one of the four uncoordinated 5H (UNC5H)-negative NTN receptor epitopes in the array. In the neuroattractant NT/Ntrk family, KS did not bind the soluble NT-3 neuroattractant ligand, but did bind the NT3-specific Ntrk3 receptor epitope and the less NT isoform-specific NGFR receptor. KS did not bind the other two NT receptors, Ntrk1 and -2. In the Schwann cell–regulating NRG/ERBB family, KS bound both the soluble NRG1 SMDF Schwann cell attractant/stimulatory ligand and the two SMDF ERBB receptor epitopes that were in the protoarray. Of the other soluble growth factors in the protoarray, KS bound PDGF, but did not bind EGF, acidic or basic FGF, or TGFβ1. Finally, KS did not bind either of the assembled integrin epitopes in the protoarray, integrin alpha V beta 3 (αVβ3), or integrin alpha 3 beta 1 (α3β1). 

In comparison, sturgeon notochord CSA bound only 9 of the 85 nerve-related proteins or protein epitopes in the nerve-related protoarray (Table 6). Secreted neuroguidance ligand NTN4 and one of its UNC5H negative receptor epitopes were the strongest CSA binders, whereas the DCC-positive receptor epitope bound less strongly, and the other NTN and UNC5H isoforms in the protoarray were not bound by CSA. Two neuroregulatory SEMA isoform epitope ligands were slightly bound, along with two EFN ligands, one EFN receptor EPH epitope isoform, and one SLIT neuroguidance receptor ROBO epitope isoform. Of the 29 epitopes that were present in the protoarray in both the IgG₁-tagged and IgG₁-cleaved forms, CSA bound 5, and in each case it bound the tagged version, but not the cleaved version. Binding strengths for nerve-related CSA binders were significantly lower than were binding strengths for KS binders (compare Table 6 to Table 5). 

Table 6.

 

CSA Interaction with Extracellular Nerve-Related Molecules and Epitopes

Protein	Signal	Database ID	Protein Description
NTN4	7,468	AAG53651	Netrin 4: Hum Met1-Lys628. Secreted neuroguidance molecule.
UNC5H1-Fc	4,515	NP_071542.1	Uncoordinated: Hum CD33 Met1-Ala16 + Rat UNC5H1 Gln26-Asp358). Receptor for secreted NTN axon guidance ligands.
SEMA6B-Fc	4,264	Q9H3T3	Semaphorin 6B: Hum CD33Met1-Ala16 + Hum Sem6B Leu26-Ser603 + Link + HumIgG Pro100-Lys 330. Transmembrane glycoprotein that fens in neuroguidance.
EFNA3-Fc	4,144	NP_004943.1	Ephrin A3 extracellular domain Hum(aa 1–209) + link + HumIgG Pro100-Lys 330; binds Eph receptors in developing neural tissue.
EPHA1-Fc	3,062	P52793	Ephrin A1: Mus Asp19-Ser182 + link + HumIgG Pro100-Lys 330. Binds Eph receptors in developing neural tissue.
DCC-Fc	2,933	NP_031857	Deleted in colorectal carcinoma, Mus Met1-Asn1097 + LINK + HumIgG Pro100-Lys 330; attractive response toward netrin-1.
SEMA3E	2,591	0150041	Semaphorin 3E: Mus Thr25-Ser775 (Arg557Alaand Arg560Ala) secreted, axon and vascular tip cell chemorepellant; binds plexin.
ROB02-Fc	2,546	NP_002933.1	Hum Met1-Ala312 + Link + Hum ROB02 Pro313-Pro859 + HumIgG Pro100-Lys 330. Neural tissue negative Slit receptor.
EPHB2-Fc	2,442	P29323	Ephrin B2: Hum Met1-Leu543 + Link + HumIgG Pro100-Lys 330. Transmembrane receptor; interacts with EFNA5 and EFNB ligands.

  n = 9 of 85 proteins screened.

bHA did not bind any proteins in the nerve-related extracellular epitope protoarray. 

KS Core Protein Mimecan Interactions with v4 Proteins

The KS core protein mimecan (OGN) bound 15 of the 8000+ proteins in the v4 array, listed in Table 7. Of these 15 proteins, mimecan bound SLAIN2, a ubiquitously expressed β-tubulin-like protein of unknown function that is widely conserved in vertebrates, ⁵⁹ with a higher signal than any signal generated by bKS, bCSA, or bHA binding to any protein. The next highest binder, UBXD3, generated one fifth the signal intensity, and the other 13 binders gave signal intensities below one fifth the mimecan-SLAIN2 intensity. Mimecan binders included two members of the casein kinase 1 family: delta and epsilon; two different potassium channel proteins: KCNAB1 and KCNAB2; and three kinases that have roles in microtubule formation or stabilization: MARK2, AURKA, and PLK1. 

Table 7.

 

Mimecan Interaction with Invitrogen v4 Proteins

Protein	Signal	Accession No.	Invitrogen ID	Protein Description
SLAIN2	50,042	NP_065897.1	BC031691.2	SLAIN motif family, member 2
UBXD3	12,008	NP_689589.1	NM_152376.2	UBX domain protein 3
DYRK3	9,363	NP_003573	PV3837	Dual specificity serine/threonine and tyrosine kinase 3
APEX2	8,757	NP_055296.2	NM_014481.2	Apurinic/apyrimidinic endonuclease 2, nuclear gene for mito protein
MARK2	8,389	NP_059672.2	PV3878	MAP/microtubule affinity-regulating kinase 2
CSNK1D	7,668	NP_620693	PV3665	Casein kinase 1, delta (serine/threonine kinase)
AURKA	7,657	NP_940839	PV3612	Aurora kinase A, serine/threonine, assoc. w centrosomes
KCNAB1	7,601	NP_751891.1	NM_172159.2	Potassium voltage-gated channel, shaker-related, beta 1
STK40	6,187	NP_114406.1	NM_03217.1	Serine/threonine kinase 40
C11orf63	5,318	NP_954575.1	NM_199124.1	Chromosome 11 ORF 63, variant 2
PLK1	4,143	NP_005021	PV3501	Polo-like kinase 1 (Drosophila) associated with cell mitosis
TCP10L	3,902	NP_653260.1	NM_144659.1	t-Complex 10 (mouse)-like
KCNAB2	3,808	NP_003627.1	NM_003636.1	Potassium voltage-gated channel, shaker-related subfamily, beta 2
CENK1E	3,308	NP_001885	PV3500	Casein kinase 1, epsilon, association with DNA damage repair and cytokine
PAK6	2,995	NP_064553	PV3502	p21(CDKN1A)-activated kinase 6, in testis and prostate

  n = 15 of 8,268 screened.

Kinetic Studies of KS Interaction with SLIT2 as Determined by SPR

Our protoarray study revealed SLIT3 to be one of the most avid binders to KS. After the nerve-related protoarray had been completed, R&D Systems made SLIT2 available. Because SLIT2 mRNA is expressed in the anterior cornea ⁵⁴ and has been shown to influence nerve growth cone migration differently, depending on whether it is bound by ECM molecules, ^60,61 we characterized the interaction of KS with SLIT2 in more detail, performing SPR, with bKS immobilized on streptavidin-coated sensor chips. The kinetic analyses were performed by injecting SLIT2 at various concentrations over immobilized KS. As revealed by a typical sensorgram shown in Figure 1, binding of SLIT2 to KS was concentration dependent. For the analyses shown in Figure 1, the SLIT2 injection volume was 50 μL. Kinetic analyses for each SLIT2 concentration were performed in triplicate. As shown in Figure 2, SLIT2 binds to KS in a saturable manner, as evidenced by the logarithmic trend line. Kinetic data were fitted to a 1:1 Langmuir binding model (Table 8), allowing the calculation of the k _a (kinetic association rate constant), the k _d (kinetic dissociation rate constant), and the K_D (apparent equilibrium dissociation constant). The K _D for SLIT2 binding to KS was 28 nM, suggesting that KS binds SLIT2 with relatively high affinity. 

Figure 1.

 

SPR sensorgrams of SLIT2-KS interactions. Top to bottom: concentrations of SLIT2 were 150, 130, 70, 50, 20, and 10 nM, respectively.

Figure 2.

 

SLIT2 binds to KS in a concentration-dependent and saturable manner with a logarithmic trend line.

Table 8.

 

Langmuir 1:1 Binding Model Data for Interaction between SLIT2 and Biotinylated KS

	Keratan Sulfate
k a × 103, M−1	1.17 ± 0.24
k d×10−7, s−1	3.45 ± 2.37
K D, nM	27.77 ± 15.35

  k _a, kinetic association rate constant; k _d, kinetic dissociation rate constant; K _D, apparent equilibrium dissociation constant.

By protoarray analysis, KS bound SEMA3A much less strongly than SLIT3, but binding was still significantly above background (Tables 2, 5). Repeated attempts to detect SEMA3A binding to immobilized KS using SPR were unsuccessful, using SEMA3A concentrations as high as 2 μM or 250 ng/μL. 

Discussion

In both the v4 protoarray and a custom extracellular nerve-related protoarray, corneal KS bound the greatest number of proteins. In contrast, sturgeon notochord CSA bound significantly fewer v4 proteins and extracellular nerve-related proteins than did KS. Nonsulfated cockscomb HA bound the fewest v4 proteins in comparison to corneal KS and sturgeon notochord CSA, and no extracellular nerve-related proteins at all. Mimecan, included in this study because it is the KS core protein about which the least protein-binding information is known, bound slightly more v4 proteins than did sturgeon notochord CSA and was not tested with our custom extracellular nerve-related protein array. However, binding strengths of 9 of the 23 sturgeon notochord CSA-binding candidates were higher than the binding strength of the strongest corneal KS-binding candidate. In contrast, binding strengths of five of the six nonsulfated HA-binders were lower than the strengths of any positive highly sulfated corneal KS- or highly sulfated CSA-binding proteins. Both corneal KS and sturgeon notochord CSA bound preferentially with the biologically active (IgG₁-tagged) form of almost all extracellular nerve-related epitope pairs, suggesting that the GAG-protein interactions observed in this study have potential in vivo relevance. Although generally regarded as extracellular, KS ^24,62–66 and HA ^50,51 have also been detected intracellularly. These GAG–protein binding profiles demonstrate that GAGs found in the corneal stroma may interact with many proteins, both intracellularly and extracellularly, and thus could influence the availability and/or the configuration of these proteins for subsequent interaction with their target cellular ligands/receptors. 

Our study is the first to examine binding of corneal KS, sturgeon notochord CSA, or nonsulfated HA to a wide variety of cellular and ECM proteins using high-throughput protoarray techniques with the proteins immobilized on a fixed surface and the GAGs free in solution. Other protoarray studies examining GAG–protein interactions have immobilized naturally occurring and synthetic GAGs on a fixed surface and probed them with solutions containing a limited number of known proteins. ^67,68 That the protoarray format can reveal new protein–GAG interactions is substantiated by the fact that in a study by Wang et al. ⁶⁸ dermatan sulfate GAG was found to bind to an antibody not previously thought to bind it. In our study the protoarray format first revealed KS binding to a SLIT family member, which was then further defined for KS binding to SLIT2 by SPR studies. It is not clear from our and other comparative protein-GAG interaction studies whether it is better to attach the GAG or the protein to the fixed matrix, and what is optimal may differ for different GAGs and different proteins. Attaching some GAGs ⁶⁹ and proteins ⁷⁰ to a matrix has been found to impose structural constraints that have diminished their binding to some proteins. Also, some GAGs ^67,71 and some proteins may attach better to a given matrix than others. In our study, HA in solution did not bind the known HA-binders aggrecan, link protein, and fibronectin affixed to a matrix, but it has been shown to bind link module when HA is the fixed moiety. ^45,72 In our study, KS in solution bound SEMA3A in the protoarray format where SEM3A was the fixed moiety, but not in the SPR format where KS was the fixed moiety. 

Although there are several indicators of reproducibility for the protein binders for KS, CSA, and HA detected in our protoarray protocol, our protocol failed to detect binding of some of these GAGs to proteins they have been reported to bind in other studies, such that these known GAG binders did not serve as positive controls in our protoarray study. The most probable reason that the v4 versions of aggrecan, link protein, or fibronectin did not serve as positive controls for HA binding in our proteomics study is that the v4 versions of these proteins are produced in insect cells and are therefore not glycosylated in the ways that these molecules are when produced under their usual human cell biosynthetic conditions. Deglycosylated aggrecan cannot bind HA, ⁷³ and it is likely that the same is true for nonglycosylated link protein. In addition, fibronectin binding of HA requires that the fibronectin be both glycosylated and aggregated, and nonaggregated fibronectin bound to an affinity column does not bind HA. ⁷⁴ In the same vein, although KS had been shown to bind IGFBP2 before this proteomics study, IGFBP2 could not serve as a positive control for KS binding in our proteomics screening format because IGFBP2 binds GAGs only if IGF-I or -II are present. ⁷⁵ These considerations suggest that the proteins identified as GAG binders in our protoarray format bind those GAGs without being posttranslationally glycosylated or the participation of additional cofactors. 

In many GAG-protein interaction studies, biotinylation is a common method of labeling either the proteins or the GAGs. In some protoarray studies, soluble protein ligands are labeled with biotin for detection after GAG binding. ^67,68 We used immobilized proteins with solubilized bGAGs for detection. Others have also used bGAGs to study GAG interactions with immobilized proteins. ^{71–72,75–78} However, biotinylation procedures can biotinylate different positions within a GAG, ⁶⁹ as occurred in our study with biotinylation of KS at only one end of its chain and biotinylation of CSA and HA all along their GAG backbones. In addition, biotinylation can have different efficiencies with different GAGs, again as it did in our study. These differences in biotinylation may have different effects on the functional properties of GAGs. ^69,71,79 Similar position, efficiency, and steric hindrance effects may perturb the binding properties of proteins if they are the biotinylated moieties in GAG–protein binding studies. Also, many GAG–protein interaction studies using column chromatography or SPR protocols use the biotin of the bGAG to attach the GAG to the fixed substrate, ^32,69,80 as we did in our SPR experiments. This method could magnify structural aberrations in a GAG to a greater or lesser extent, depending on where the GAG is biotinylated. Thus, protoarray techniques can suggest false GAG–protein interactions, and proteins identified by protoarray studies as GAG–binding proteins should be considered candidate GAG–binders until additional techniques confirm the GAG-protein binding. 

GAGs are defined by their primary disaccharide structure, and therefore KS, CS, and HA have the same primary structure from one species to another. However, the extent and positions of sulfation of GAG disaccharides can vary considerably for any one kind of GAG, both from species to species, and also from tissue to tissue within a given species. Bovine and human corneal KS chains have similar charge densities and chain lengths, and the three most prevalent capping structures at the nonreducing termini are identical in relative proportions. ⁸¹ Therefore, it is likely that human corneal KS chains would bind v4 human proteins in a manner very similar to that displayed by bovine corneal KS. In contrast, sturgeon notochord CSA is 100% sulfated ⁸² (every disaccharide is monosulfated at C-4 of GalNAc), which is a higher percentage of disaccharide sulfation than is found in most human ⁸³ or bovine ⁸² corneal CSA. The extent and positions of sulfate groups within CS chains have been shown to significantly affect CS binding to protein targets, ^33,34,84 leading to the proposal of a sulfation code ⁸⁵ for GAGs that governs their ability to interact with proteins. It is not known whether human corneal CSA chains may be asymmetrically sulfated, such that they have regions of high (100%) disaccharide sulfation contiguous with regions of low disaccharide sulfation. If they do, 100% sulfated regions (sulfation hot spots) of such human corneal CSA would be expected to show protein interactions similar to those seen with sturgeon notochord CSA, whereas the low sulfated regions may bind very different proteins. HA is never sulfated, so HA from any species or tissue should be equivalent in its binding properties with human proteins, as long as their average chain lengths are similar. 

In our GAG protoarray studies, Invitrogen optimized conditions for KS binding in their protoarray protocol and then used the same general conditions for CSA and HA binding. It is possible that under other binding conditions, highly (or asymmetrically) sulfated CSA and/or HA would react with more candidate proteins. Alternatively, spotting complete posttranslationally modified forms of proteins such as aggrecan, link protein, or fibronectin on microarray plates may also increase the number and kinds of proteins that bind GAGs in a protoarray format. No mimecan binding proteins were known before our study, so it is difficult to assess whether the binding conditions used in this study were optimal for mimecan. 

Because corneal KS, sturgeon notochord CSA, and HA are large linear polyanionic molecules, it is probable that electrostatic interactions are part of the mechanism involved in the protein-KS, -CSA, and -HA binding observed in our protoarray studies, as has been demonstrated in many heparin- and heparan sulfate-protein interactions. ⁸⁶ Recent studies of reactions of yeast ⁸⁷ and human ^88,89 protein arrays with a range of biotinylated linear polyanionic macromolecules (actin, tubulin, heparin, heparan sulfate, and DNA) have defined a wide array of polyanionic binding proteins (PABPs), with sulfated GAGs binding the greatest number of different PABPs in each study. In addition, lack of reduction in the number of PABPs bound by increasing binding solution salt concentrations suggests that there are also noncoulumbic (e.g., hydrophobic and hydrogen binding) molecular interactions involved in protein-GAG binding. ^88,89 Moreover, reacting proteins did not need to be positively charged overall, nor to have a specific GAG-binding signature sequence motif, ¹⁸ but only needed to have some positive subdomain along their amino acid sequence to be capable of polyanionic binding. ^86–89 The authors argue that many proteins contain subdomains of amino acids that are probably unfolded in solution ⁹⁰ and that these disordered regions may allow them to interact with many different binding partners with various degrees of strength or specificity by mechanisms such as electrostatic interaction that do not require tertiary structure complementarity. Similar considerations also apply to interactions of proteins like kinases with smaller anionic molecules such as single nucleotides. ⁸⁹ In our study, 75 of the 217 KS-candidate v4 protein binders and 9 of the 23 CSA-candidate v4 protein binders were kinases. The catalytic domains of all kinases contain two invariant lysines and an arginine, with the serine/threonine kinases containing an additional invariant lysine, and the tyrosine kinases containing additional histidine, arginine, and lysine residues, ⁹¹ suggesting that kinase catalytic domains may have an overall basic charge that would allow them to interact electrostatically with polyanionic GAGs. Viewed more broadly, 28 of the top 50 KS-candidate binders and 11 of the 23 CSA-candidate binders could be classified as PABPs on the basis of their known binding to actin, microtubules, DNA, RNA, nucleotides, or nucleosides. HA's top protein interactants could also be viewed as potential PABPs. Keratin-associated protein 13-1 (KRTAP13-1), contains one partial (-RPR-) basic amino-acid–rich GAG-binding motif ¹⁸ in its short sequence. ⁹² Insulin receptor substrate 1 (IRS1), possesses numerous basic amino-acid–rich segments in its primary amino acid sequence and two ATP-binding sites. ^93,94  

The nature of the interactions between mimecan and its 15 candidate v4 protein binding partners remains unresolved. The revelation that mimecan binds SLAIN2, a β-tubulin-like protein, and three kinases that regulate microtubule behavior suggests a possible role for mimecan in microtubule function in vivo. It is probable that the protein-binding behavior of mimecan core protein is altered when it is glycosylated with KS chains. Although all three corneal KS core proteins are small leucine-rich repeat proteins, mimecan is smaller than the other two proteins. ^4–6 It would be interesting to know whether the protein-binding properties of KS are affected differently when it is attached to different core proteins. 

SLIT-1, -2, and -3 orthologues usually function as chemorepellants of migrating neuronal growth cones. ⁹⁵ However, an N-terminal fragment of SLIT2, purified from rat brain, causes DRG sensory axons to elongate and branch. ⁹⁶ Full-length neurorepellant SLIT2 can be transformed into neuroattractant and bifurcation N-terminal SLIT2 by interaction with ECM components, ⁶¹ including rat brain heparan sulfate proteoglycan glypican-1(GPC1). ⁹⁷ Heparan sulfate O-sulfate groups are critical for GPC1-SLIT2 binding. ⁹⁸ The mechanism of the proteolytic SLIT cleavage that results from GPC1-SLIT binding is not yet known. Our protoarray and SPR demonstrations that SLIT2 binds significantly to highly sulfated corneal KS suggests that polyanionic, highly O-sulfated ECM KSPG may function as a SLIT2-cleavage facilitator in the cornea, thus converting SLIT2 into its neuroattractant/neurobifurcating form in the area of the developing cornea where corneal nerves are extending and branching extensively. ^53,54 This possible chaperone role for polyanionic KS in SLIT2 transformation indeed would mimic the role of polyanionic regions in other better-known chaperone–protein interactions. ^99–101 In addition, under protoarray conditions, ECM neurorepellant SEMA3A was identified as a candidate KS binder. Perhaps association and dissociation constants for SEMA3A with KS could not be determined by SPR because of increased steric hindrance in KS function when it is the moiety attached to the fixed substrate. Our observations suggest that corneal KS is very influential in regulation of corneal innervation during development. 

Footnotes

 Supported by a National Institutes of Health (NIH) Microarray Supplement Grant 3R01EY000952–35S1 to NIH 5R01EY000952, and NIH 2 R01EY000952 (GWC). The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the National Eye Institute or the National Institutes of Health.

Footnotes

 Disclosure: A.H. Conrad, None; Y. Zhang, None; E.S. Tasheva, None; G.W. Conrad, None