Structural and functional insights into the interaction of sulfated glycosaminoglycans with tissue inhibitor of metalloproteinase-3 – A possible regulatory role on extracellular matrix homeostasis

Accepted Manuscript Structural and functional insights into the interaction of sulfated glycosaminoglycans with tissue inhibitor of metalloproteinase-3 – a possible regulatory role on extracellular matrix homeostasis Sandra Rother, Sergey A. Samsonov, Tommy Hofmann, Joanna Blaszkiewicz, Sebastian Köhling, Stephanie Moeller, Matthias Schnabelrauch, Jörg Rademann, Stefan Kalkhof, Martin von Bergen, M. Teresa Pisabarro, Dieter Scharnweber, Vera Hintze PII: DOI: Reference:

S1742-7061(16)30426-3 http://dx.doi.org/10.1016/j.actbio.2016.08.030 ACTBIO 4393

To appear in:

Acta Biomaterialia

Received Date: Revised Date: Accepted Date:

14 June 2016 22 July 2016 17 August 2016

Please cite this article as: Rother, S., Samsonov, S.A., Hofmann, T., Blaszkiewicz, J., Köhling, S., Moeller, S., Schnabelrauch, M., Rademann, J., Kalkhof, S., von Bergen, M., Teresa Pisabarro, M., Scharnweber, D., Hintze, V., Structural and functional insights into the interaction of sulfated glycosaminoglycans with tissue inhibitor of metalloproteinase-3 – a possible regulatory role on extracellular matrix homeostasis, Acta Biomaterialia (2016), doi: http://dx.doi.org/10.1016/j.actbio.2016.08.030

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Structural and functional insights into the interaction of sulfated glycosaminoglycans with tissue inhibitor of metalloproteinase-3 – a possible regulatory role on extracellular matrix homeostasis Sandra Rother1, Sergey A. Samsonov2, Tommy Hofmann3, Joanna Blaszkiewicz4,5, Sebastian Köhling4,5, Stephanie Moeller6, Matthias Schnabelrauch6, Jörg Rademann4,5, Stefan Kalkhof3,7, Martin von Bergen3,8,9, M. Teresa Pisabarro2, Dieter Scharnweber1 and Vera Hintze1* 1

Institute of Materials Science, Max Bergmann Center of Biomaterials, TU Dresden, Budapester

Str. 27, 01069 Dresden, Germany 2

Structural Bioinformatics, BIOTEC TU Dresden, Tatzberg 47-51, 01307 Dresden, Germany

3

Department of Molecular Systems Biology, Helmholtz-Center for Environmental Research-

UFZ, Permoserstraße 15, 04318 Leipzig, Germany 4

Institute of Pharmacy & Institute of Chemistry and Biochemistry, Freie Universität Berlin,

Königin-Luise-Str. 2 + 4, 14195 Berlin, Germany 5

Institute of Medical Physics and Biophysics, Universität Leipzig, Härtelstr. 16/18, 04107

Leipzig, Germany 6 7

Biomaterials Department, INNOVENT e.V., Prüssingstraße 27 B, 07745 Jena, Germany

Current address: Bioanalytics, University of Applied Sciences Coburg, Friedrich-Streib-Straße

2, 96450 Coburg, Germany 8

Institute of Biochemistry, Faculty of Biosciences, Pharmacy and Psychology, Permoserstraße

15, University of Leipzig, Germany 9

Department of Chemistry and Biosciences, Aalborg University, Fredrik Bajers Vej 7H, 9220

Aalborg, Denmark Corresponding Author: *E-mail: [email protected], Phone: +49351 46339389, Fax: +49351 46339401

1

Abstract An imbalance between tissue-degrading matrix metalloproteinases (MMPs) and their counterparts’ tissue inhibitors of metalloproteinases (TIMPs) causes pathologic extracellular matrix (ECM) degradation in chronic wounds and requires new adaptive biomaterials that interact with these regulators to re-establish their balance. Sulfated glycosaminoglycans (GAGs) and TIMP-3 are key modulators of tissue formation and remodeling. However, little is known about their molecular interplay. GAG/TIMP-3 interactions were characterized combining surface plasmon resonance, ELISA, molecular modeling and hydrogen/deuterium exchange mass spectrometry. We demonstrate the potential of solute and surface-bound sulfated hyaluronan (sHA) and chondroitin sulfate (sCS) derivatives to manipulate GAG/TIMP-3 interactions by varying GAG concentration, sulfation degree and chain length. Three GAG binding sites in the N- and C-terminal domains of TIMP-3 were identified. We reveal no overlap with the matrix metalloproteinases (MMP)-binding site, elucidating why GAGs did not change MMP-1/-2 inhibition by TIMP-3 in enzyme kinetics. Since we prove that GAGs alone have a low impact on MMP activity, sHA and sCS offer a promising strategy to possibly control ECM remodeling via stabilizing and accumulating TIMP-3 by maintaining its MMP inhibitory activity under GAGbound conditions. Whether GAG-based functional biomaterials can be applied to foster chronic wound healing by shifting the MMP/TIMP balance to a healing promoting state needs to be evaluated in vivo.

Keywords: Tissue inhibitor of metalloproteinase-3, glycosaminoglycans, hyaluronan/ sulfated hyaluronan, matrix metalloproteinase, molecular modelling, chronic wound healing

2

1. Introduction The demographic changes in developed countries are accompanied by an increase in comorbidities like diabetes, obesity and peripheral vascular diseases. As a consequence, the incidence of chronic wounds rises especially in the geriatric population resulting in a substantial decrease of the patients ‘quality of life and in increased health care costs highlighting the need for new strategies and functional biomaterials for chronic wound treatment.[1,2] Tissue homeostasis and remodeling is a well-orchestrated process critical for maintaining normal tissue functions. In this context the extracellular matrix (ECM) as a complex, highly dynamic meshwork synthesized by the local cells of a tissue plays a crucial role. Timely proteolytic ECM degradation via matrix metalloproteinases (MMPs) is important for tissue repair and remodeling during wound healing since it affects a variety of biological processes (e.g. cell migration, morphogenesis, bioavailability of growth factors).[3][4][5] MMPs are secreted by inflammatory cells (neutrophils, macrophages) and wound cells (epithelial cells, fibroblasts, vascular endothelial cells) especially after injury and infection while some (e.g. MMP-2) are also expressed in resting tissue. [6][7] Their activity is regulated by their biosynthesis, the proMMP/zymogen activation, the compartmentalization and the inhibition especially via tissue inhibitors of metalloproteinases (TIMPs).[8] All four members of the TIMP family are potent inhibitors of the proteolytic MMP activity and, in some cases, of the disintegrin metalloproteinases (ADAMs) and the ADAMs with thrombospondin motifs (ADAMTS) as well.[9] The N-terminal domain of TIMPs binds to the active site of their target mature metalloproteases via a 1:1 non-covalent interaction thereby blocking the access of substrates to the catalytic site while the C-terminal domain can form complexes with proMMPs.[10] TIMP-3 is distinct from the other soluble TIMPs as it is bound to sulfated glycosaminoglycans (sGAGs)

3

in the ECM.[11] The role of TIMP-3 as a major MMP regulator in situ is highlighted as its absence resulted in lung emphysema-like alveolar damage and fast epithelial cell apoptosis in mice.[12,13] Furthermore, TIMP-3 is a competitor of vascular endothelial growth factor, a regulator of hematopoiesis and bone formation via affecting stem cell proliferation, differentiation and migration.[14,15] There are increasing indications that sGAGs contain information in form of a “sulfation code” which alter the location and activity of growth factors or enzymes depending on the GAG polysaccharide backbone, the number and positioning of the sulfate moieties.[16] Several MMPs are known to bind to heparin (HEP) with a wide variation in binding affinity.[17,18] There are evidences for the binding of MMP-1 and -2 to heparan sulfate (HS) and the influence of native sGAGs on MMP activity.[19,20] Sulfated hyaluronan (sHA) derivatives are promising bioactive components for the development of functional ECM-mimetic biomaterials enabling a detailed analysis of the structure-function relationship. Their advantages compared to native sGAGs are their defined structural and chemical properties (distinct carbohydrate backbone, defined molecular weight, chain length, sulfation degree and pattern) and good availability.[21] Hence, sHA-based materials are also more applicable to clinical translation. Increasing evidence suggests an influence of GAG derivatives in biomaterials on matrix remodeling. Previous studies revealed a reduced MMP-1 synthesis of human dermal fibroblasts after cultivation on collagenbased artificial ECMs (aECMs) containing high-sulfated HA (sHA3) or over-sulfated chondroitin sulfate (sCS) derivatives.[22] Furthermore, a decreased MMP-2 activity of human mesenchymal stromal cells after culturing on aECMs with sHA3 compared to non-sulfated HA was reported.[23] A recent study indicates that TIMP-3 and HEP are as well important modulators of ECM turnover being crucial for a variety of physiological and pathological

4

conditions.[24] Pathologic conditions such as chronic wounds and diseases like cancer, osteoarthritis and atherosclerosis often correlate with an increased MMP and decreased TIMP expression leading to an inappropriate ECM degradation and thereby tissue damage.[25,26] Therefore, the development of new adaptive functional biomaterials shifting the TIMP/MMP balance to a healing phenotype is warranted. For that purpose, an in-depth understanding of the underlying mechanisms and especially the possible role of GAGs as potential modulators of the MMP and TIMP activity could foster the development of novel GAG-containing biomaterials as treatment options possibly able to modify the ECM turnover towards patient-specific needs. In this study, we evaluated the interaction of native and chemically sulfated GAG derivatives on the activity of TIMP-3 and MMPs by combining surface plasmon resonance (SPR), ELISA and enzyme activity analyses with molecular modelling techniques as well as amide hydrogen/deuterium exchange (HDX) mass spectrometry. We demonstrate which structural properties of GAGs are essential for tuning the TIMP-3 binding. Therefore, possible GAGbinding sites of TIMP-3 were elucidated offering access to a deeper understanding of the molecular requirements for these interactions. In addition, the influence of GAG and TIMP-3/GAG interactions on MMP activity in vitro was elucidated in detail. 2. Materials and methods 2.1 Materials HA (from Streptococcus, MW = 1.1 106 g/mol) was obtained from Aqua Biochem (Dessau, Germany), sulfur trioxide/dimethylformamide complex (SO3-DMF, purum, ≥ 97%, active SO3 ≥ 48%) and sulfur trioxide/pyridine complex (SO3-pyridine, pract.; ≥ 45% SO3) from Fluka Chemie (Buchs, Switzerland). HEP from porcine intestinal mucosa and biochemical reagents were from Sigma-Aldrich (Schnelldorf, Germany) and CS (porcine trachea, mixture of 70%

5

chondroitin-4-sulfate and 30% chondroitin-6-sulfate) was from Kraeber (Ellerbek, Germany). Oligomeric HA (general formula: GlcA – (GlcNAc – GlcA)2 – GlcNAc), CS (degree of sulfation per repeating disaccharide unit (D.U.): 1; general formula: GlcA – GalNAc,4S or 6S – GlcA – GalNAc,4S or 6S – GlcA – GalNAc, 4S or 6S) and HEP (degree of sulfation: 2; general formula: GlcA,2S or IdoA,2S – GlcNS,6S – (IdoA,2S – GlcNS,6S)2, a mixture of approx. 75% IdoA,2S – GlcNS,6S and saccharides with varying degree and pattern of sulfation) with a degree of polymerization of six (dp 6) were obtained from Iduron (Manchester, UK). The catalytic domains of human recombinant MMP-1 (BML-SE180) and MMP-2 (BML-SE237) were purchased from Enzo Life Science (Lörrach, Germany). Recombinant human TIMP-3 (973-TM010), recombinant human MMP-1 (901-MP-010), recombinant human MMP-2 (902-MP-010) as well as monoclonal, mouse antihuman TIMP-3 (MAB973) and monoclonal, biotinylated, mouse antihuman TIMP-3 (BAM9731) antibodies were obtained from R&D Systems (WiesbadenNordenstadt, Germany). Series S Senor Chips CM5TM were purchased from GE Healthcare Europe GmbH (Freiburg, Germany). 2.2 Preparation of polymeric glycosaminoglycan derivatives Low molecular weight HA, low- and high-sulfated HA derivatives (sHA1, sHA3) and oversulfated CS (sCS3) were produced and characterized as described previously.[27,28] The average number of sulfate groups per D.U. (degree of sulfation) is 1.0 and 2.8 for sHA1 or sHA3 and 3.1 for sCS3. The weight-average molecular weight analyzed by gel permeation chromatography (GPC) using laser light scattering detection is 48255 Da for HA (polydispersity index (PD) detected by GPC = 2.3), 31056 Da for sHA1 (PD = 2.2), 28745 Da for sHA3 (PD = 1.7) and 19915 Da for sCS3 (PD = 1.5). 2.3 Preparation of sulfated hyaluronan oligosaccharides

6

HA tetrasaccharide (dp 4) and HA hexasaccharide (dp 6) as precursors for synthesis were obtained by enzymatic digestion of HA in acidic sodium acetate buffer using bovine testis hyaluronidase. To fix the anomeric configuration of the reducing end and to reduce the reactivity of this position, the HA oligomers were reacted with 2-chloro-1,3-dimethylimidazolinium chloride and sodium azide furnishing the anomeric azides, respectively. The remaining hydroxylgroups were converted to sulfate esters using sulfur trioxide pyridine complex followed by purification of the final products as sodium salts (psHA, dp 4, dp 6).[29] All products were characterized by NMR spectroscopy and mass spectrometry. In addition, selective sulfation of the primary alcohol moieties of the HA (dp 4) were conducted also with sulfur trioxide pyridine complex. Purification using anion exchange chromatography furnished the di-sulfated tetrasaccharide as a sodium salt (sHA1, dp 4). The degree of sulfation per repeating disaccharide unit (D.U.) is 1.0 for sHA1 (dp 4) and 4.0 for psHA (dp 4, dp 6), respectively. Figure 3 a-c displays the structures of the synthesized HA tetrasaccharide in comparison to the non-sulfated HA tetrasaccharide. 2.4 Enzyme kinetic analysis 20 µl 0.450 U/µl catalytic MMP-1 or -2 domain and 20 µl GAGs (0.025-25 mM D.U.) were incubated in assay buffer (50 mM HEPES (N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid), 10 mM CaCl2 and 0.05% L-23, pH 7.5) at 37°C for 1 hour to complete 90 µl in a white half-area 96-well plate (Corning, Tewksbury, USA). Similar molar concentrations of D.U. for each GAG were used to compare the same number of possible binding sites of interactions since GAG polysaccharides are polydisperse. 10 µl 40 µM substrate (Mca-Pro-Leu-Dpa-Ala-Arg-NH2 [Mca = (7-methoxycoumarin-4-yl)-acetyl, and Dpa = N-3-(2,4-dinitrophyenyl)-L-α-β-diaminopropionyl]; Enzo Life Science) were added to start analysis. The fluorescence signal increase (λex

7

= 328 nm, λ em = 420 nm) was measured for 10 min in 30 s intervals using a Tecan infinite M200 pro plate reader and Tecan-i-Control software (Tecan, Crailsheim, Germany). The slope of linear region of the regression curve was used to determine the remaining activities. 20 µl of a 6.5 µM NNGH (N-isobutyl-N-(4-methoxyphenylsulfonyl)glycyl hydroxamic acid; Enzo Life Science) and buffer without GAGs served as positive control and negative control (100% activity). Full-length MMP-1 and -2 were used after allosteric zymogen activation with 1 mM paminophenylmercuric acetate (APMA) at 37°C for 2 or 1 hour according to the manufacturer (R&D Systems). Then, the enzymes were diluted in TIMP buffer (50 mM Tris, 10 mM CaCl2, 150 mM NaCl, 0.05 % Brij-L23, pH 7.5) to a concentration of 3.8 nM in the assay. TIMP buffer was used for all dilutions and kinetic analysis of these enzymes. The inhibition of full-length MMPs by 5 nM or 2.5 nM TIMP-3 in the absence or presence of GAGs (0.25 -2.5 mM D.U.) was determined after 30 min pre-incubation of GAG/TIMP-3 at 37°C. 2.5 Analysis of TIMP-3 binding to GAG surfaces GAGs were immobilized via their reducing ends to 96-well microtiter plates (Nunc MaxiSorpTM) as described previously [30,31]. As control of nonspecific binding, uncoated wells were incubated with 2% bovine serum albumin (BSA) dissolved in Tris-buffered saline (TBS, pH 7.4). Afterwards, 50 µl of 0-1.8 nM TIMP-3 in 1% BSA dissolved in phosphate-buffered saline (PBS, pH 7.4) were added to the wells and incubated o/N at 4°C. Then, the supernatants were collected and stored at -80°C for following Sandwich-ELISA. The amount of GAG-bound TIMP-3 after o/N incubation at 4°C was quantified indirectly with specific antibodies (Sandwich-ELISA) in supernatants containing non-bound TIMP-3 using a TIMP-3 calibration curve ranging from 04000 pg/ml in 1%BSA/PBS according to the manufactures protocol (R&D Systems, WiesbadenNordenstadt, Germany).

8

2.6 SPR analysis of TIMP-3 interaction with GAGs Binding analysis was performed using a BiacoreTM T100 (GE Healthcare). TIMP-3 was immobilized onto Series S Sensor Chips CM5TM via amine coupling at 25°C as described by the manufacturer (GE Healthcare). On average 830 RU of TIMP-3 were immobilized using a concentration of 10 µg/ml. For interaction studies with oligomeric GAGs TIMP-3 was immobilized in values of 4700 RU at 50 µg/ml. An activated and afterwards deactivated flow cell was used as reference surface without immobilized protein. TIMP buffer was used as running buffer and the interactions were studied at 37°C. This buffer was also used for GAG dilutions. 100 µM - 600 µM Polymeric GAGs and 10 µM - 500 µM D.U. oligomeric GAGs were injected for 300 s at 30 µl/min and binding levels were recorded 10 s before end of injection (values represent the reference subtracted sensorgrams relative to a baseline report point). The same molar concentrations of D.U. were chosen for GAG comparison as they represent the possible binding sites of interactions. GAG binding levels were corrected for the molecular weight of the polymer because the mass increase results in higher binding signals. The injection was followed by a dissociation phase in running buffer of 10 min at a flow rate of 30 µl/min. The chip surface was regenerated after each sample injection with 5 M NaCl. The baseline was allowed to stabilize for at least 5 min with running buffer prior to injection of the next sample. The determined values were normalized to the binding response of 100 µM sHA3 or psHA (dp 6), respectively. The Biacore T100 evaluation software was used to evaluate the binding parameters. Specific Biacore sensorgrams were obtained by double referencing. 2.7 Molecular Modeling 2.7.1 Comparative modeling

9

The structure of the N-terminal domain of TIMP-3 (residues 24 to 143) was obtained from the Xray structure of TIMP-3 complexed with ADAM-17 protease (PDB ID: 3CKI, 2.30 Å). The Cterminal domain (residues 144 to 199) was modeled based on its homology to TIMP-1 (34% identity and 62% similarity). For this, the X-ray structure of TIMP-1 complexed with MMP-3 was used as structural template (PDB ID: 1UEA, 2.80 Å). The obtained three-dimensional (3D) model of TIMP-3 was refined by applying molecular dynamics (MD) (vide infra for protocols used). 2.7.2 Molecular docking Autodock 3 (AD3)[32] was used for docking HA (dp 4), HA (dp 6), N3-hyaluronan-6-sulfate (dp 4), N3-psHA (dp 4), N3-psHA (dp 6), chondroitin-4-sulfate (dp 6), chondroitin-6-sulfate (dp 6), HEP (dp 6) to TIMP-3. The GAG molecules were considered completely flexible in the docking calculations. A grid box with dimensions of 63 Å × 63 Å × 63 Å and a grid spacing of 0.5 Å containing the full surface of TIMP-3 was used. 100 independent runs of the Lamarckian genetic algorithm with an initial population size of 300 and a termination condition of 105 generations or 9995 × 105 energy evaluations were carried out. The 50 top docking results were clustered using the DBSCAN algorithm [33]. Three representative poses from each of the obtained clusters were selected for MD calculations. 2.7.3 Molecular dynamics simulations The structures of the GAG/TIMP-3 complexes obtained from the docking calculations were refined by applying MD simulations carried out with AMBER 14.[34] Parameters from the ff14SB and GLYCAM-06j [35] force fields were used for protein and GAGs, respectively. The complexes were solvated in a TIP3P octahedral periodic box with a minimal distance to the periodic box border of 8 Å, and counter ions were used to neutralize the system. Two energy-

10

minimization steps were carried out: first 0.5 × 103 steepest descent cycles and 103 conjugate gradient cycles with harmonic force restraints on solute atoms, and then 3 × 103 steepest descent cycles and 3 × 103 conjugate gradient cycles without constraints. Afterwards, the system was heated up to 300 K for 10 ps, equilibrated for 50 ps at 300 K and 106 Pa in isothermal isobaric ensemble (NPT). Finally, a 20 ns of productive MD run was carried out at constant pressure as an NTP ensemble. The SHAKE algorithm, 2 fs time integrations 8 Å cutoff for non-bonded interactions, and the Particle Mesh Ewald method were used. For each GAG from the complex, pyranose rings were harmonically restrained in 4C1 (for IdoA2S in 1C4) conformations. Free energy calculations and per residue decomposition were performed using Molecular MechanicsGeneralized Born Surface Area (MM-GBSA) with gb=2 implemented in AMBER for 100 frames evenly distributed in the last 10 ns of the productive MD run. Analysis of the trajectories was done using the cpptraj module of AMBER and VMD for visualization.[36] 2.7.4 Electrostatic potential calculations Electrostatic potential isosurfaces were calculated using the PBSA program from AmberTools with a grid spacing of 1 Å. 2.8 Hydrogen/deuterium exchange mass spectrometry 1 µl of 500 ng/µl (20 pM) TIMP-3 dissolved in 200 mM TRIS-HCl buffer (pH 7.4) in the presence and absence of 10 µg/ml (9 nM) psHA tetrasaccharide (dp 4) was mixed with deuterated buffer to a final concentration of 90 % D2O. The protein was incubated for either 1, 60, or 1440 minutes to allow backbone amide hydrogen/deuterium exchange. All experiments were performed in three technical replicates. H/D-exchange was quenched by adding 35 µl of ice cooled 0.1% formic acid to adjust the pH to 2.5. The sample was immediately injected and the sequence specific H/D-exchange was determined by LC-MS using an ESI-TOF-MS (Xevo, Waters Corporation) online coupled to a nano-uHPLC system equipped with an automatic H/D11

exchange workbench (nanoACQUITY UPLC and HDX manager, Waters Corporation). The HDX manager allows for an online enzymatic digestion (at 20 °C) and desalting and chromatographic peptide separation (at 0 °C). The online digestion was performed using an immobilized pepsin column (2.1 × 30 mm Waters) in 0.1 % formic acid at a flow rate of 10 µL/min. Proteolytic peptides were trapped and desalted online (Acquity, UPLC BEH C18 1.7 µm trapping column, Waters Corporation). After 2.5 min the peptides were eluted and separated by reversed phase chromatography (Acquity UPLC BEH C18 1.7 µm, 1 mm × 100 mm column, Waters Corporation) using a 7 min linear acetonitrile gradient (8– 80 %) containing 0.1% formic acid at a flow rate of 40 µL/min. Peptides were detected using ESI-TOF-MS (Xevo G2S TOFMS, Waters) utilizing a lock-mass correction. Mass spectra were acquired in a MSE mode over the m/z range of 50–2000. The ESI cone voltage was set to 40 V with a desolvatisation gas temperature of 200 °C. Back exchange was measured by a 1440 min sample quenched with deuterated FA in the same way as previous samples. Data analysis including peptide identification and the determination of the deuterium uptake was performed using Waters Dynamix software. The calculation of the sequence-specific H/D-exchange ratios were performed as been previously descripted.[37] HDX data where analyzed by structural means using the previously generated model. 2.9 Statistical analysis Experiments were performed at least in triplicate. Results are presented as mean ± standard deviation. Data were analyzed by one-way ANOVA (analysis of variance) or two-way ANOVA. Bonferroni post-hoc test was applied to evaluate differences between the groups. P values < 0.05 were considered statistically significant. 3. Results 3.1 Influence of GAGs on MMP activity

12

The catalytic domain as well as the full-length enzyme after activation were used to study the functional requirements for a possible GAG/MMP-interaction. Distinct effects of GAGs on MMP activity required concentrations in the mM range related to the molecular weight of disaccharide units (D.U.) and were more pronounced for MMP-2 than for MMP-1 (Figure 1). Native GAGs (HA, CS, HEP) and sCS3 led to a concentration-dependent accelerated substrate hydrolysis by both catalytic domains. However, sHA3 did not significantly alter their activity, while high concentrations of sHA1 slightly decreased the substrate cleavage. Lower GAG concentrations (2.5 mM D.U.) led to no observable activity changes in the presence of HA, CS and sHA3, while HEP and sCS3 increased their activity. To better reflect the native situation, full-length MMP-1 and -2 were analyzed as well. Results were opposed to those obtained for the catalytic domains showing a decreased activity in the presence of 25 mM D.U. for almost all GAGs. This effect was most pronounced for sHA3 with a remaining MMP activity of about 50% (Figure 1 c/d). In case of MMP-1 (Figure 1 c) the inhibitory effects increased depending on sulfation degree in the order HA ≤ sHA1 < sHA3 and CS < sCS3. In contrast, the inhibition of MMP-2 showed no clear effect of sulfation (Figure 1 D). 2.5 mM D.U. GAG did not alter the MMP-1 activity while CS reduced the MMP-2 activity by about 25%. However, the presence of other GAGs at this concentration led to no or only a marginal reduction of the MMP-2 activity. 3.2 Tuning the interaction of TIMP-3 with GAG derivatives The interaction of polymeric GAGs with TIMP-3 was investigated via SPR (Figure 2 a-c). The binding strength of all GAGs increased with higher concentration and sulfation degree. The highest binding levels were found for sHA3 and sCS3. CS displayed comparable binding strength to sHA1. HA displayed no or only marginal binding responses. To validate the GAG/TIMP-3 interaction characteristics the binding of TIMP-3 to GAG surfaces was

13

investigated by ELISA (Figure 2 d). The amount of GAG-bound TIMP-3 again revealed a concentration- and sulfation-dependent binding with high-sulfated GAGs showing the strongest interaction compared to GAGs with lower sulfation degree. All GAG-coated surfaces were significantly different from the BSA control, indicating a specific interaction of TIMP-3 with GAG surfaces. The TIMP-3 binding capacities of sHA1 and CS or sHA3 and sCS3 were comparable. Together these results revealed a direct interaction of TIMP-3 with solute and surface-bound polymeric sGAGs. Further requirements regarding the GAG size, sulfation degree and pattern and carbohydrate backbone were examined using SPR measurements with tetra- and hexasaccharides (indicated by dp 4 or dp 6) and TIMP-3 (Figure 3). The structures of the synthesized HA oligosaccharides in comparison to HA (dp 4) are displayed in Figure 3 a-c. Non-sulfated HA oligosaccharides showed only a slight binding to TIMP-3 (Figure 3 a, f). Binding of sHA derivatives was observed in the presence of tetra- and hexasaccharides (Figure 3 b, c, f). However, CS (dp 6) and HEP (dp 6) displayed binding responses that were even lower than for HA oligosaccharides with a fast GAG dissociation after injection indicating that longer oligosaccharide chains are required for the interaction with these GAGs (Figure 3 f). The binding of the investigated tetrasaccharides was sulfation dependent. The binding levels at 500 µM D. U. increased in the following order: CS (dp 6) = HEP (dp 6) < HA (dp 4) = HA (dp 6) < psHA (dp 6) < sHA1 (dp 4) < psHA (dp 4) (Figure 3 f). Furthermore, for all GAG oligosaccharides the binding to TIMP-3 was found to be concentration-dependent. However, the psHA (dp 4) displayed higher binding levels than the corresponding psHA (dp 6) at all concentrations suggesting differences in their binding profiles. 3.3 Molecular modeling and structural analysis of TIMP-3/GAG interactions Our calculations for TIMP-3 clearly suggest a region with a positive electrostatic potential, which could be involved in GAG binding (Figure 4 a). Blind docking calculations for different

14

GAGs and further MD-based Molecular Mechanics-Poisson Bolzmann Surface Area (MMPBSA) free energy decomposition per residue analysis also revealed solutions in this region, corresponding to 24 residues (with contribution ∆G < -1.0 kcal/mol) which can be clustered into three binding sites (A, B and C) of TIMP-3 being involved in GAG-binding as shown in Table 1 and Figure 4 c. Table 1. Binding sites of TIMP-3 involved in GAG recognition.

Binding Number of site

amino acids

A

8

Comprising amino acids Arg-48, Gly-49, Phe-50, Arg-84, Lys-123, Leu-125, Arg-163, Lys-165 Arg-20, Lys-22, Lys-42, Lys-45, Lys-76, Tyr-77, Arg-100, Trp-101,

B

13 Arg-109, Lys-110, Asn-113, Lys-165, Arg-173 Arg-20, Lys-22, Lys-42, Met-44, Lys-45, His-55, Trp-101, Arg-109,

C

10 Lys-110, Asn-113

Among these residues Lys-76, Lys-165 and Arg-163 were included in the motifs previously suggested by mutagenesis to be important for GAG-binding (Figure 4 b).[38] psHA (dp 4) was selected to determine the residues being involved in GAG-binding by analyzing HDX of the backbone amides of TIMP-3 in complex with and without psHA. Information about 87% of the TIMP-3 sequence could be obtained with an average resolution of 4.5 amino acids. The amino acid sequences Arg-20 – Ala-21, Lys-22 – Tyr-39, Thr-105 – Leu-106, Ser-107 – Asn-113, Gln-155 – His-158, and Ile-162 – Trp-171 were found to be most affected (Figure 4). Compared to molecular docking these sequences and any other significant shielding along the sequence overlapped with 70% of binding sites B and C, while binding site A of TIMP-3 was fully covered by psHA (dp 4) during HDX experiments with the exception of the undetectable peptides 117-133, which should contain two proposed amino acids involved in GAG-binding (Figure 4/5). 15

For any amino acid predicted via molecular modeling a shielding of TIMP-3 in the presence of psHA (dp 4) could be observed (Figure 4 c/d, Figure 5). The shielding of TIMP-3/psHA (dp 4) was less pronounced than expected due to the rather small sterical dimension of the tetrasaccharides, even though psHA (dp 4) was theoretically able to build many hydrogen bonds with corresponding backbone elements of TIMP-3. Nevertheless, HDX exchange findings clearly indicated that the binding of the small ligand psHA (dp 4) to GAG binding sites of TIMP3 is not only of ionic nature since HDX is not capable to detect ionic bonds and measured shielding effects usually consist of a range of hydrogen-bonding to sterical shielding of the backbone amides.

3.4 Influence of GAGs on the MMP inhibition via TIMP-3 The impact of GAG-binding to TIMP-3 on the inhibitory potential towards MMP-1 and -2 was studied via enzyme kinetic analysis in the presence of pre-formed TIMP-3/GAG complexes in comparison to TIMP-3 alone (Figure 6). TIMP-3 concentrations which led to a remaining MMP activity of 40%, were chosen to compare the impact of GAG derivatives on TIMP-3/MMP inhibition. GAG concentrations of up to 2.5 mM D.U. did not significantly alter the inhibition of MMPs by TIMP-3. Furthermore, an additional MMP inhibition due to GAGs could not be detected. Molecular modeling was used to compare the location of the positive electrostatic potential on the TIMP-3 surface and the binding site of the ADAM-17 protease (PDB ID: 3CKI), which binds TIMP-3 like MMPs. This approach was chosen since only the X-ray structures of TIMP-3 with this protease are available. In line with our experimental data (Figure 6 a/b/c), the results suggest that binding of GAGs by TIMP-3 not affect TIMP-3/proteases interactions (Figure 6 d).

16

4. Discussion There is an increasing need for functional biomaterials promoting wound healing and regeneration due to the growing incidence of pathological skin conditions especially in the elderly population. One example are chronic wounds, which are often associated to an imbalance between the activity of matrix-degrading MMPs and their natural inhibitors, TIMPs.[26] Hence, identifying and analyzing the modulatory potential of GAG derivatives to influence the interaction with TIMP-3 in detail is of particular interest since these GAGs can be used to functionalize biomaterials. However, to date there is only a limited number of studies often restricted to the effects of HEP, which is a highly variable GAG structure.[20,24] We therefore the molecular mechanisms by which native and structurally as well as chemically defined sGAG derivatives affect the MMP and TIMP-3 activities as well as their interplay. Enzyme kinetics revealed that sulfated GAGs alter the MMP-1 and -2 activity at concentrations in the mM D.U. range (Figure 1). Compared to their native inhibitor TIMP-3, which exhibits IC50 values in the low nM range (Figure 6), GAGs are weak MMP inhibitors. This is in accordance to findings for HEP showing a weak interaction with MMP-1 and proMMP-2.[19,39] In the present study native GAGs (e.g. HEP) led to enhanced activities of the catalytic domains of MMP-1 and -2. In accordance to our findings zymographic detection of pro- and active fulllength of MMP-1, -7 and -13 was reported to be enhanced in the presence of HEP.[20] The authors suggested HEP-induced conformational changes that increase the MMP activity, facilitate refolding or reduce autolysis. An effect of GAGs on the activation of proMMPs by disrupting the zinc-thiol interaction due to allosteric interaction of the prodomain with GAGs

17

leading to conformational perturbation of the cysteine switch is reported[7], which further explains the GAG-induced zymogen activation. To better reflect the in vivo situation, the impact of GAGs on full-length MMPs was examined as well and exposed a reduced enzyme activity of MMP-1 and -2 at high concentrations of solute GAGs. This was more pronounced for MMP-2 than for MMP-1 and in contrast to the findings with the catalytic domains. The reduced activity cannot be simply attributed to the overall charge density of the GAGs since a higher sulfation degree did not automatically result in a stronger inhibition. However, the presence of sHA3 led to the strongest inhibition of both full-length enzymes. The findings are supported by results of Kliemt et al. demonstrating a reduced MMP-2 activity in the presence of sHA3-containing aECMs.[23] The difference to the findings of Yu and Woessner (2001) showing a GAG-induced MMP activation of active full-length enzymes can be explained by the different forms in which the enzyme is present within the assays (bound within the gel during zymography vs. solute during enzyme kinetics). For an enhanced zymographic detection, the MMP needs to be in direct contact with the GAG within the gel.[20] Thereby, the MMP had a reduced flexibility in the gel probably allowing a close interaction of HEP with the catalytic domain, which led to an enhanced substrate hydrolysis. In contrast, MMPs in solution have more freedom to interact with solute GAGs and moreover other noncatalytic protein regions may sterically hinder the interaction of GAGs with the catalytic domain. The opposed GAG-induced effects on the activity of the catalytic domains and the full-length enzymes indicate a possible participation of non-catalytic MMP domains (e.g. hinge-region, hemopexin-like domain) on GAG-mediated changes of MMP activity. Accordingly, binding of HEP to the hemopexin-like domain of MMP-2 was reported.[40] Furthermore, sulfate groups of the GAGs possibly sequester divalent cations such as Ca2+, even though a direct interactions of HEP and calcium ions have not yet been revealed.[41] As the catalytic domain of MMPs 18

contains 2-3 calcium ions, besides the structural zinc ion[42], any changes at these positions may induce altered enzyme activities. Regarding collagen-based aECMs containing only 2-7% of the initially applied 2.5 mM D.U. GAGs[43] we expect the remaining GAG amounts to be no obstacle for the aECM degradation via MMPs. In addition, van der Boom et al. found no direct relation between synovial fluid levels of GAGs and MMP activity.[44] Amongst the known MMP inhibitors, TIMP-3 is of particular interest because is known to be GAG-bound within the ECM. Accordingly, a dose- and sulfation-dependent binding of native and chemically sulfated polymeric GAGs to TIMP-3 was found via SPR and ELISA (Figure 2). In line with previous studies on the GAG interaction with TGF-β1 and BMP-2[28,45], the binding strengths of HA or CS to TIMP-3 were marginal, while those of the high-sulfated GAGs where much higher. In contrast to these studies, our experiments revealed no significant differences between polymeric CS and sHA1, as well as sCS3 and sHA3, in their binding to TIMP-3, indicating that the carbohydrate backbone has no detectable influence on the binding strength of the TIMP-3/GAG interaction at a comparable sulfation degree. This suggests that the interactions of GAGs with TIMP-3 are mainly of electrostatic nature formed primarily between the negatively charged sulfate of the GAG and positively charged amino acids of the protein.[41] Robinson et al. also revealed a sulfation-dependent binding of TIMP-3 to GAGs for CS and HEP. In contrast, they detected no clear correlation between the GAG-bound TIMP-3 amounts and the sulfation degree when comparing HS and HEP samples and suggest that different GAGbinding sites of TIMP-3 might be responsible for this effect.[46] Competition SPR experiments revealed that the interaction of TIMP-3 with HEP was mainly dependent on the HEP chainlength as well as the N-sulfo and 6-O-sulfo groups of HEP. Other GAGs like HS, CS type A and C exhibited weak binding affinities towards TIMP-3.[47]

19

To further clarify these contradictory findings, the molecular mechanism of GAG-binding to TIMP-3 was characterized using defined sulfated oligomeric HA derivatives in comparison to commercially available oligosaccharides (Figure 3). The interaction of these GAGs with TIMP-3 indeed requires distinct structural GAG properties. Concerning the sulfation requirements sHA1 (dp 4) which is exclusively sulfated at the C-6 position of the glucosamine units was still able to bind to TIMP-3. The sulfation at this position was sufficient for a significantly stronger interaction with TIMP-3 in comparison to non-sulfated HA. In addition, as sHA1 (dp 4) exhibited a stronger binding to TIMP-3 than the CS or HEP hexasaccharides, we conclude that the carbohydrate backbone as well as the sulfation pattern of the GAG significantly affect the binding of oligosaccharides to TIMP-3 since these results cannot be simply attributed to the negative net charge of the sugars. Interestingly, the corresponding polymeric GAGs exhibited a clear sulfation dependence. As the exact sulfation pattern of neither the CS hexasaccharide nor polymeric HEP were available a coherent conclusion of this aspect is challenging. However, we suggest that differences in their sulfation pattern might be responsible for this as well. Surprisingly, our findings show that in case of high-sulfated oligomeric GAGs (e.g. psHA (dp 4)) the binding cannot be enhanced by increasing the chain length. This may indicate the binding of the very short tetrasaccharide to additional regions of TIMP-3. Accordingly, Zhang et al. reported that even a small disaccharide of HEP was sufficient for TIMP-3 binding during SPR.[47] To corroborate these findings, additional molecular modelling and HDX experiments were performed to determine the GAG-binding sites of TIMP-3 (Figure 4, Figure 5). The Nterminal domain of TIMP-3 has been suggested as main HEP-binding region since it contains two regions rich in basic amino acids.[11] Lee et al. demonstrated by mutagenesis of TIMP-3 that the N- as well as the C-terminal domain participate in GAG-binding. They identified Lys-26, Lys-27, Lys-30, Lys-76 of the N-terminal domain and Arg-163 and Lys-165 of the C-terminal 20

domain as amino acids involved in the GAG interaction.[38] The contribution of Lys-76, Arg163 and Lys-165 are in line with the present blind docking and MD based free energy calculations. In addition, 21 amino acids were identified as further important amino acids for GAG-recognition (Table 1). Our MD-based conclusions were supported by the results obtained from HDX mass spectrometry performed with TIMP-3 and psHA (dp 4), the oligomeric GAG exhibiting the strongest interaction with TIMP-3. Taken together with SPR results, both findings strongly suggest for the first time that TIMP-3 has three potential GAG-binding sites (Table 1, Figure 5). Consequently, these regions cannot be occupied by one single oligomeric GAG molecule. Thus, multiple GAGbinding modes can occur on TIMP-3, which was also observed in the HDX experiments. This high GAG-binding capacity of TIMP-3 may be an explanation for the SPR detected higher binding of psHA (dp 4) compared to the longer corresponding hexasaccharide (psHA, dp 6), which proves the potential of short sulfated HA derivatives to interact with TIMP-3 in a defined manner. Furthermore, our findings clearly prove that the N- and C-terminal domains of TIMP-3 contribute to GAG binding. The biological consequences of GAG-binding to TIMP-3 in vitro, studied via enzyme kinetics, demonstrated that this interaction however does not affect the inhibitory activity of TIMP-3 against MMP-1/-2 (Figure 6 a, b, c). This is in line with the respective GAG-binding sites on TIMP-3 that do not overlap with the MMP-binding region as shown by molecular modeling (Figure 6 d). In contrast, Butler et al. reported a slightly increased inhibition of MMP-2 by TIMP-3 in the presence of HEP. However in accordance to our results, HS, de-N-sulfated HEP, dermatan sulfate or HA had no or only marginal effects on the inhibition of TIMP-3 towards MMP-2.[48]

21

To further validate the present findings HA oligomers with higher dp and different sulfation patterns as well as GAG derivatives containing iduronic acid residues should be investigated to analyze in greater detail the impact of chain length, substituent distribution and sugar backbone on the TIMP-3/GAG interaction. These alterations could result in a further improved control of the sequestration of TIMP-3 by GAGs without changing the inhibitory potential towards MMP-1 and -2, two major proteases involved in wound healing. Finally, the in vitro approaches performed within this study are limited regarding their possible biocomplexity and cannot completely recapitulate the in vivo conditions present in chronic wounds or osteoarthritis. Whether the presented evidence of the regulatory functions of GAGs on TIMP-3 converts into a defined restored MMP/TIMP balance in vivo requires extensive testing. Even after a variety of clinical trials with different MMP inhibitors effective targets for the treatment of acute and chronic wounds are still missing.[49] Thus, using chemically sulfated GAGs to functionalize biomaterials for stabilizing and accumulating TIMP-3, while maintaining its MMP inhibitory activity under GAG-bound conditions in vitro, seems a promising approach for altering the MMP/TIMP ratio. 5. Conclusion Sulfated GAGs such as sHA derivatives are promising candidates for the treatment of chronic wounds since they sequester TIMP-3 in a defined manner compared to native GAGs without altering its inhibitory potential towards the MMPs-1 and -2, two major proteases involved in wound healing. Here, we describe a novel option for a strategy to regulate TIMP-3 levels with sulfated HA derivatives that might be generally applied in biomaterials to protect tissue from excessive proteolytic degradation associated with several pathologic conditions (e.g. cancer, osteoarthritis). By potentially prolonging the beneficial presence of TIMP-3 in chronic wounds

22

sGAGs-containing biomaterials could slow down matrix degradation processes towards patient specific needs. Therefore due to possibly shifting the MMP/TIMP balance to a healing phenotype sGAGs might improve e.g. chronic wound healing. Acknowledgements: This work was supported by the German Research Council DFG (SFBTRR 67, A2, A3, A7, A8, Z3, Z4).

23

References [1]

C.K. Sen, G.M. Gordillo, S. Roy, R. Kirsner, L. Lambert, T.K. Hunt, F. Gottrup, G.C. Gurtner, M.T. Longaker, Human skin wounds: A major and snowballing threat to public health and the economy, Wound Repair Regen. 17 (2009) 763–771.

[2]

S. Gist, I. Tio-Matos, S. Falzgraf, S. Cameron, M. Beebe, Wound care in the geriatric client., Clin. Interv. Aging. 4 (2009) 269–287.

[3]

H. Nagase, R. Visse, G. Murphy, Structure and function of matrix metalloproteinases and TIMPs, Cardiovasc. Res. 69 (2006) 562–573.

[4]

Q. Yu, I. Stamenkovic, Cell surface-localized matrix mealloproteinase-9 proteolytically activates TGF-beta and promotes tumor invasion and angiogenesis, Genes Dev. 14 (2000) 163–176.

[5]

E. Hadler-Olsen, B. Fadnes, I. Sylte, L. Uhlin-hansen, J.O. Winberg, Regulation of matrix metalloproteinase activity in health and disease, FEBS J. 278 (2011) 28–45.

[6]

D. Gibson, B. Cullen, R. Legerstee, K. Harding, G. Schultz, MMPs Made Easy, Wounds Int. 1 (2009) 1–6.

[7]

A. Tocchi, W.C. Parks, Functional interactions between matrix metalloproteinases and glycosaminoglycans, FEBS J. 280 (2013) 2332–2341.

[8]

A. Page-McCaw, A.J. Ewald, Z. Werb, Matrix metalloproteinases and the regulation of tissue remodelling., Nat. Rev. Mol. Cell Biol. 8 (2007) 221–233.

[9]

G. Murphy, Tissue inhibitors of metalloproteinases, Genome Biol. 12 (2011) 233.

[10] K. Maskos, Crystal structures of MMPs in complex with physiological and pharmacological inhibitors, Biochimie. 87 (2005) 249–263. [11] W.H. Yu, S.S.C. Yu, Q. Meng, K. Brew, J.F. Woessner, TIMP-3 binds to sulfated glycosaminoglycans of the extracellular matrix, J. Biol. Chem. 275 (2000) 31226–31232. [12] J.E. Fata, K.J. Leco, E.B. Voura, H.E. Yu, P. Waterhouse, G. Murphy, R.A. Moorehead, R. Khokha, Accelerated apoptosis in the Timp-3 – deficient mammary gland, J. Clin. Invest. 108 (2001) 831–841. [13] K.J. Leco, P. Waterhouse, O.H. Sanchez, K.L.M. Gowing, A.R. Poole, A. Wakeham, T.W. Mak, R. Khokha, Spontaneous air space enlargement in the lungs of mice lacking tissue inhibitor of metalloproteinases-3 (TIMP-3), J. Clin. Invest. 108 (2001) 817–829. [14] J.H. Qi, Q. Ebrahem, N. Moore, G. Murphy, L. Claesson-Welsh, M. Bond, A. Baker, B. Anand-Apte, A novel function for tissue inhibitor of metalloproteinases-3 (TIMP3): inhibition of angiogenesis by blockage of VEGF binding to VEGF receptor-2., Nat. Med. 9 (2003) 407–415. [15] Y. Shen, I.G. Winkler, V. Barbier, N. a. Sims, J. Hendy, J. Le, J.P. Lévesque, Tissue Inhibitor of Metalloproteinase-3 ( TIMP-3 ) Regulates Hematopoiesis and Bone Formation In Vivo, PLoS One. 5 (2010) 1–13. [16] C.I. Gama, L.C. Hsieh-wilson, Chemical approaches to glycosaminoglycan code, Curr. Opin. Chem. Biol. 9 (2005) 609–619.

deciphering

the

24

[17] I.E. Collier, S.M. Wilhelm, a Z. Eisen, B.L. Marmer, G. a Grant, J.L. Seltzer, C.S. Kronberger, E. a Bauer, C.S. He, G.I. Goldberg, H-ras oncogene transformed human bronchial epithelial cells (TBE-1) secrete a single metalloproteinase capable of degrading basement membrane collagen, J. Biol. Chem. 263 (1988) 6579–6587. [18] K. Imai, Y. Okada, Purification of matrix metalloproteinases by column chromatography., Nat. Protoc. 3 (2008) 1111–1124. [19] W. Yu, J.F. Woessner, Heparan Sulfate Proteoglycans as Extracellular Docking Molecules for Matrilysin (Matrix Metalloproteinase 7), J. Biol. Chem. 275 (2000) 4183–4191. [20] W.H. Yu, J.F. Woessner, Heparin-enhanced zymographic detection of matrilysin and collagenases., Anal. Biochem. 293 (2001) 38–42. [21] J. Schiller, J. Becher, S. Möller, K. Nimptsch, T. Riemer, M. Schnabelrauch, Synthesis and Characterization of Chemically Modified Hyaluronan and Chondroitin Sulfate, Mini. Rev. Org. Chem. 7 (2010) 290–299. [22] A. van der Smissen, V. Hintze, D. Scharnweber, S. Moeller, M. Schnabelrauch, A. Majok, J.C. Simon, U. Anderegg, Growth promoting substrates for human dermal fibroblasts provided by artificial extracellular matrices composed of collagen I and sulfated glycosaminoglycans, Biomaterials. 32 (2011) 8938–8946. [23] S. Kliemt, C. Lange, W. Otto, V. Hintze, S. Moller, M. Von Bergen, U. Hempel, S. Kalkhof, O. Differentiation, S. Kliemt, C. Lange, W. Otto, V. Hintze, S. Mo, M. Von Bergen, U. Hempel, S. Kalkhof, Sulfated Hyaluronan Containing Collagen Matrices Enhance Cell- Matrix-Interaction, Endocytosis, and Osteogenic Differentiation of Human Mesenchymal Stromal Cells, J. Proteome Res. 12 (2013) 378–389. [24] L. Troeberg, C. Lazenbatt, F. Anower-e-khuda, C. Freeman, O. Federov, H. Habuchi, O. Habuchi, K. Kimata, H. Nagase, M.F. Anower-E-Khuda, C. Freeman, O. Federov, H. Habuchi, O. Habuchi, K. Kimata, H. Nagase, Sulfated Glycosaminoglycans Control the Extracellular Trafficking and the Activity of the Metalloprotease Inhibitor TIMP-3, Chem. Biol. 21 (2014) 1300–1309. [25] M. Egeblad, Z. Werb, New functions for the matrix metalloproteinases in cancer progression., Nat. Rev. Cancer. 2 (2002) 161–174. [26] R. Lobmann, a. Ambrosch, G. Schultz, K. Waldmann, S. Schiweck, H. Lehnert, Expression of matrix-metalloproteinases and their inhibitors in the wounds of diabetic and non-diabetic patients, Diabetologia. 45 (2002) 1011–1016. [27] V. Hintze, S. Moeller, M. Schnabelrauch, S. Bierbaum, M. Viola, H. Worch, D. Scharnweber, Modifications of hyaluronan influence the interaction with human bone morphogenetic protein-4 (hBMP-4), Biomacromolecules. 10 (2009) 3290–3297. [28] V. Hintze, A. Miron, S. Moeller, M. Schnabelrauch, H.-P.P. Wiesmann, H. Worch, D. Scharnweber, Sulfated hyaluronan and chondroitin sulfate derivatives interact differently with human transforming growth factor-β1 (TGF-β1), Acta Biomater. 8 (2012) 2144– 2152. [29] J. Köhling, S., Künze, G., Lemmnitzer, K., Bermudez, M., Wolber, G., Schiller, J., Huster, D., Rademann, Chemoenzymatic synthesis of spin-labeled, sulfated tetrahyaluronan for elucidation of its complex with interleukin-10, Chemistry (Easton). 22 (2016) 5563–74.

25

[30] A. Satoh, K. Kojima, T. Koyama, H. Ogawa, I. Matsumoto, Immobilization of saccharides and peptides on 96-well microtiter plates coated with methyl vinyl ether-maleic anhydride copolymer., Anal. Biochem. 260 (1998) 96–102. [31] J. Salbach-Hirsch, J. Kraemer, M. Rauner, S.A. Samsonov, M.T. Pisabarro, S. Moeller, M. Schnabelrauch, L.C. Hofbauer, V. Hintze, B. Diseases, S. Bioinformatics, D. Scharnweber, L.C. Hofbauer, V. Hintze, The promotion of osteoclastogenesis by sulfated hyaluronan through interference with osteoprotegerin and receptor activator of NF-κB ligand/osteoprotegerin complex formation, Biomaterials. 34 (2013) 7653–7661. [32] G.M. Morris, D.S. Goodsell, R.S. Halliday, R. Huey, W.E. Hart, R.K. Belew, A.J. Olson, M.E.T. Al, Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function, J. Comput. Chem. 19 (1998) 1639–1662. [33] M. Ester, H.P. Kriegel, J. Sander, X. Xu, A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise, Second Int. Conf. Knowl. Discov. Data Mining, Menlo Park. Calif. AAAI Press. (1996) 226–231. [34] D.A. Case, T. Darden, T.E.C. Iii, C. Simmerling, S. Brook, A. Roitberg, J. Wang, U.T. Southwestern, R.E. Duke, U. Hill, R. Luo, U.C. Irvine, D.R. Roe, R.C. Walker, S. Legrand, J. Swails, D. Cerutti, J. Kaus, R. Betz, R.M. Wolf, K.M. Merz, M. State, G. Seabra, P. Janowski, F. Paesani, J. Liu, X. Wu, T. Steinbrecher, H. Gohlke, N. Homeyer, Q. Cai, W. Smith, D. Mathews, R. Salomon-ferrer, C. Sagui, N.C. State, V. Babin, T. Luchko, S. Gusarov, A. Kovalenko, J. Berryman, P.A. Kollman, Amber 14, Univ. California, San Fr. (2015). [35] K.N. Kirschner, W.R. Yongye AB, Tschampel SM, González-Outeiriño J, Daniels CR, Foley BL, GLYCAM06: a generalizable biomolecular force field. Carbohydrates., J. Comput. Chem. 29 (2008) 622–655. [36] W. Humphrey, A. Dalke, K. Schulten, VMD : Visual Molecular Dynamics, J. Mol. Graph. 14 (1996) 33–38. [37] T. Hofmann, S.A. Samsonov, A. Pichert, K. Lemmnitzer, J. Schiller, D. Huster, M.T. Pisabarro, M. von Bergen, S. Kalkhof, M. Von Bergen, S. Kalkhof, Structural analysis of the interleukin-8/glycosaminoglycan interactions by amide hydrogen/deuterium exchange mass spectrometry, Methods. 89 (2015) 45-53. [38] M.H. Lee, S. Atkinson, G. Murphy, Identification of the Extracellular Matrix (ECM) binding motifs of tissue inhibitor of metalloproteinases (TIMP)-3 and effective transfer to TIMP-1, J. Biol. Chem. 282 (2007) 6887–6898. [39] T. Crabbe, C. Ioannou, A.J.P. Docherty, Human progelatinase, Computing. 438 (1993) 431 –438. [40] U.M. Wallon, C.M. Overall, The Hemopexin-Like Domain (C-Domain) of Human Gelatinase-A (Matrix Metalloproteinase-2) Requires Ca2+ for Fibronectin and HeparinBinding - Binding-Properties of Recombinant Gelatinase a C-Domain to ExtracellularMatrix and Basement-Membrane Components, J. Biol. Chem. Vol 272 (1997) 7473–7481. [41] D. Xu, J.D. Esko, Demystifying Heparan Sulfate-Protein Interactions., Annu. Rev. Biochem. (2014) 1–29. [42] H. Nagase, J.F. Woessner, H. Nagase, J.F. Woessner, Matrix Metalloproteinases, J. Biol.

26

Chem. 274 (1999) 21491–21494. [43] D. Scharnweber, L. Hübner, S. Rother, U. Hempel, U. Anderegg, S. a. Samsonov, M.T. Pisabarro, L. Hofbauer, M. Schnabelrauch, S. Franz, J. Simon, V. Hintze, Glycosaminoglycan derivatives: promising candidates for the design of functional biomaterials, J. Mater. Sci. Mater. Med. 26 (2015) 232. [44] H. Brommer, P.A.J. Brama, A. Barneveld, R. van den Boom, M.R. van der Harst, H. Brommer, P.A.J. Brama, A. Barneveld, P.R. van Weeren, J. DeGroot, Relationship between synovial fluid levels of glycosaminoglycans, hydroxyproline and general MMP activity and the presence and severity of articular cartilage change on the proximal articular surface of P1., Equine Vet. J. 37 (2005) 19–25. [45] V. Hintze, S.A. Samsonov, M. Anselmi, S. Moeller, J. Becher, M. Schnabelrauch, D. Scharnweber, M.T. Pisabarro, Sulfated glycosaminoglycans exploit the conformational plasticity of bone morphogenetic protein-2 (BMP-2) and alter the interaction profile with its receptor, Biomacromolecules. 15 (2014) 3083–3092. [46] D.E. Robinson, D.J. Buttle, R.D. Short, S.L. McArthur, D. a. Steele, J.D. Whittle, Glycosaminoglycan (GAG) binding surfaces for characterizing GAG-protein interactions, Biomaterials. 33 (2012) 1007–1016. [47] F. Zhang, K. Lee, R. Linhardt, SPR Biosensor Probing the Interactions between TIMP-3 and Heparin/GAGs, Biosensors. 5 (2015) 500–512. [48] G.S. Butler, S.S. Apte, F. Willenbrock, G. Murphy, R.B.Y. Polyanions, G.S. Butler, S.S. Apte, F. Willenbrock, G. Murphy, Human Tissue Inhibitor of Metalloproteinases 3 Interacts with Both the N- and C-terminal Domains of Gelatinases A and B, J Biol Chem. 274 (1999) 10846–10851. [49] G. Dormán, S. Cseh, I. Hajdú, L. Barna, D. Kónya, K. Kupai, L. Kovács, P. Ferdinandy, Matrix metalloproteinase inhibitors: a critical appraisal of design principles and proposed therapeutic utility, Drugs. 70 (2010) 949–964.

27

Figure captions Figure 1. Effects of polymeric GAGs on the MMP-1 and -2 activity. Enzyme activities of catalytic domains of MMP-1 (a) or MMP-2 (b), (c) full-length MMP-1, (d) full-length MMP-2 after 1 hr of incubation with GAGs at 37°C. Two-way ANOVA: #, p < 0.05 vs. Ctrl; ##, p < 0.01 vs. Ctrl; ###, p < 0.001 vs. Ctrl; *, p < 0.05 vs. respective treatment; **, p < 0.01 vs. respective treatment; a, p < 0.001 vs. HA; b, p < 0.001 vs. CS; c, p < 0.001 vs. sHA1; d, p < 0.001 vs. HEP and e, p < 0.001 vs. sCS3. Figure 2. Interaction of polymeric GAGs with TIMP-3. SPR analysis of TIMP-3 with GAGs (100 µM - 600 µM D.U.) (a-c). The sensorgrams in (a) and (b) show the binding of sHA1 or sHA3 with increasing concentrations. Normalized binding levels for HA, CS, sHA1, HEP, sCS3 and sHA3 are depicted in (c). Significant differences calculated by two-way ANOVA are shown by *p < 0.05, **p < 0.01, ***p < 0.001. Binding of TIMP-3 to GAG surfaces determined by ELISA is displayed in (d). Two-way ANOVA: a, p < 0.001 vs. BSA; b, p < 0.01 vs. BSA; c, p < 0.05 vs. BSA; d, p < 0.001 vs. sHA3 or sCS3. Figure 3. Interaction of GAG oligosaccharides with TIMP-3. Structures of HA tetrasaccharides: (a) HA (dp 4), (b) low-sulfated HA tetrasaccharide with exclusive sulfation of the primary OH group (sHA1, dp 4), (c) per-sulfated HA tetrasaccharide (psHA, dp 4). Sulfate residues (SO3Na) were displayed as gray circles. (d)-(f) show the SPR analysis for the interaction of these GAGs (10 µM - 500 µM D.U.) with TIMP-3. Representative sensorgrams are displayed for sHA1 (dp 4) (d) and psHA (dp 4) (e). In (f) normalized binding levels are shown. One-way ANOVA: *** (p < 0.001). Figure 4. Binding of GAGs to TIMP-3. (a) The positive electrostatic potential isosurface (blue, isovalue 1 kcal mol-1 e-1) suggesting a GAGs binding region on the surface of TIMP-3. (b)

28

Residues contributing to GAG binding according to a previously study[38] are highlighted in red. (c) Residues found to highly contribute to GAG binding for different clusters of docking solutions. (d) Residues revealed by HDX analysis of the binding of psHA (dp 4) to TIMP-3 are shown in yellow and those obtained by molecular docking in red. Figure 5. Heatmap of the GAG/TIMP-3 interaction. The shielding of TIMP-3/psHA (dp 4) along the protein sequence is shown for the HDX MS incubation times 1 min, 1 hr and 24 hr. Residues are colored according to a scale ranging from green (no shielding) to red (strong difference compared to deuteration of unbound TIMP-3). The predicted GAG-binding sites A, B and C determined by MD are highlighted in gray. Figure 6. Influence of GAGs on the inhibition of MMP-1 and -2 by TIMP-3 and computational modeling of the TIMP-3/ADAM protease structure complex. (a, b) Remaining activity of MMP1 in the presence of 5 nM TIMP-3 w/o 2.5 mM D.U. polymeric or oligomeric GAGs. In (c) the activity of MMP-2 in the presence of 2.5 nM TIMP-3 w/o 2.5 mM D.U. GAG is shown. Oneway ANOVA: ###p < 0.001 vs. Ctrl. or vs. HA tetrasaccharides w/o TIMP-3 (d) The TIMP3/ADAM complex (PDB ID: 3CKI) is shown in cartoon representation (upper panel), as a molecular surface (middle panel) and with a positive electrostatic potential (EP) isosurface (blue, isovalue 1 kcal mol-1 e-1) of TIMP-3 suggesting possible GAG-binding regions (lower panel).

29

30

31

32

33

34

35

36