Recombinant human SDF-1 (300 ng) from Peprotech (UK), was incubated with 5 µl of neutrophil degranulates (NDs, containing mixtures of neutrophil proteases) or with purified active human MMP-9 (recombinant human MMP-9 produced, purified and activated as described in (29)) at a molar ratio of 1/100 (protease/SDF-1), unless indicated otherwise, in a buffer containing 150 mM NaCl, 10 mM CaCl₂, 30 mM Tris and 0.01% Tween-20 (pH 7.4), for 4 h at 37°C. For proteolysis experiments by CD26 (R&D Systems, UK), SDF-1 was incubated with a molar ratio of 1/200 (CD26/SDF-1) in 25 mM Tris buffer (pH 8), for 2 h at 37°C. In experiments with COAM, SDF-1 was pre-incubated with COAM (mass excess; as indicated) for 1 h before proteolysis by NDs, MMP-9 or CD26. Neutrophil degranulates were collected from human blood granulocytes stimulated with N-formylmethionyl-leucyl-phenylalanine (fMLP) from Sigma (USA), as previously described (30).

COAM synthesis involved two-step oxidation of amylose, as previously described (31). Briefly, the first step oxidation of amylose by sodium periodate is followed by a second oxidation step by sodium chlorite. These reactions yield COAM, a compound with multiple carboxyl groups. COAM batches prepared were free of endotoxin and of protein contaminants, determined by Limulus amebocyte lysate test and SDS-PAGE electrophoresis followed by Coomassie Brilliant Blue or silver staining, respectively (32).

After proteolysis, SDF-1 samples were separated in Novex 16% tricine gels (Thermo Fisher Scientific, USA) and stained with the SilverQuest Silver Staining Kit. Digital images of the gel were then captured. Cleaved SDF-1 forms were visualized as stained protein bands, migrating at lower molecular weights than the original intact SDF-1. For Edman sequencing, proteins blotted onto PVDF membranes were stained with Coomassie Brilliant Blue, excised and analyzed with a PPSQ-51A protein sequencer (Shimadzu, Japan).

For Western blot analysis of SDF-1 proteolysis, SDF-1 samples were separated in Novex 16% tricine gels. For Western blot analysis of SDF-1 signaling responses in THP-1 cell lysates, samples were separated in Novex 4-20% Tris-Glycine gels (Thermo Fisher Scientific). Lysate proteins were then transferred onto Immun-Blot FL PVDF membranes (Bio-rad, USA) and blocked for 1 h with Intercept Blocking Buffer (LI-COR Biosciences, USA). For proteolysis analysis, membranes were incubated overnight at 4°C with a mouse monoclonal antibody against SDF-1 (clone K15C, 1:2000, Merck Millipore, USA) and with a rabbit polyclonal antibody against SDF-1 (1:1000, Cell Signaling Technology, USA). For signaling analysis, membranes were incubated with a rabbit antibody against p-ERK1/2 Thr202/Tyr204 (1:1000, Cell Signaling Technology) and with a mouse antibody against ERK1/2 (1:2000, Cell Signaling Technology). After washing, the blot membranes were incubated for 1h at room temperature with the secondary antibodies: anti-mouse IgG 800CW and anti-rabbit IgG 680RD for proteolysis analysis or anti-rabbit IgG 800CW and anti-mouse IgG 680RD for signaling analysis (1:20000, LI-COR Biosciences). After sufficient washes, the membranes were scanned on an Odyssey Imaging System (LI-COR Biosciences). Relative quantification of intact (uncut) SDF-1 (in green) normalized to total SDF-1 (in red) and of p-ERK1/2 normalized to total ERK1/2 were performed with the use of Image Studio Lite Software (LI-COR Biosciences).

Human monocytic THP-1 cells were cultured in RPMI1640 medium supplemented with glutamine (GlutaMAX, Gibco), 10% FBS and 3% sodium bicarbonate. For signaling experiments, THP-1 cells were serum-starved in RPMI medium without phenol red for 24 h. Cells were stimulated with 30 ng/mL SDF-1 forms (with or without prior proteolysis) for 2 minutes and cell lysates were obtained with the use of RIPA buffer and protease inhibitors.

THP-1 cells (10×10⁶/mL) were stained with 2.5 µM of the fluorescent dye Fura-2AM (Invitrogen, USA) in HEPES buffered saline (HBS) containing 0.1% bovine serum albumin (BSA) and 0.01% Ploronic-F127 (Sigma), for 30 minutes at 37°C. After staining, cells were washed twice with HBS and resuspended at 1x10⁶ cells/mL. For each individual test condition, 1.8×10⁶ cells were preheated for 10 min at 37°C prior stimulation with 30 ng/mL SDF-1 and SDF-1 cleaved by MMP-9 or CD26 in the presence or absence of COAM. At 120 seconds of recording, buffer control (Vehicle) or SDF-1 forms were added. Fura-2 fluorescence was measured on a LS50 B luminescence spectrometer (Perkin Elmer, USA) at 510nm after excitation at 340 and 380nm. Intracellular Ca²⁺ concentrations ([Ca²⁺]i) were calculated using the Grynkiewicz equation (33). R_max and R_min values were determined via the treatment of cells with 50 µM digitonin and 10 mM EGTA (Sigma) in 20 mM Tris buffer (pH 8.5; Merck, Germany), respectively. Results were analyzed with WinLab32 software (Perkin Elmer, USA). Fluorescence intensities of unloaded cells (not treated with Fura-2AM) were used to correct the results for auto-fluorescence. To quantify the peak of intracellular calcium concentration, the baseline measurement was subtracted from the maximum value of the curve (i.e. peak level).

For receptor internalization experiments, THP-1 cells were stimulated with 100 ng/mL SDF-1 and SDF-1 cleaved by MMP-9 in the presence or absence of COAM for 15 minutes. Next, the cells were washed with ice-cold PBS and incubated for 15 min with the Fc-receptor-blocking antibodies anti-CD16/anti-CD32 (BD Biosciences Pharmingen) and with a Zombie Aqua™ viability dye (BioLegend, USA) for 15 min at 4°C. After washing with cold PBS + 2% FBS, surface receptors were stained with phycoerythrin (PE)-conjugated anti-CXCR4 clone QA18A64 (Biolegend) for 30 min at 4°C. Cells were washed and analyzed on a Flow Cytometer BD LSR Fortessa X20 with DIVA software. The results were further analyzed with the FlowJo software (BD). CXCR4 mean fluorescence intensity (MFI) was calculated and the results were expressed as percentage relative to the unstimulated condition (100% CXCR4 expression).

Chemotactic responses of SDF-1 were measured in vitro using a multiscreen disposable device with a 96-well filter (5 µm pore size) and a 96-well receiver plate. THP-1 cell suspensions (100 µl in the filter plate at cell concentration of 3.5x10⁶ cells/mL) and test samples (150 µl in the receiver plate at concentration of 30 ng/mL) were diluted in RPMI medium (without phenol red) supplemented with 0.1% BSA. After 3 h of migration at 37°C, the upper 96-well filter plate was removed and the cells in the lower receiver plate were quantified with the luminescence ATP detection assay system (ATPlite) (Perkin Elmer). ATP concentrations were proportional to the cell numbers. The chemotaxis index was calculated by dividing the luminescence value of the test sample by the luminescence value of the control buffer. Results of test samples were normalized relatively to SDF-1 control samples, considered as 100% THP-1 cell migration. Chemotatic responses into SDF-1-containing fibrin hydrogels were also measured in vitro in the presence and absence of COAM with the use of multiscreen plates. Hydrogels (50 μl fibrin hydrogel with SDF-1 at 1 μg/mL) were placed at the bottom of the well (receiver plate). After 4 h of migration, hydrogels were digested with Nattokinase (50 FU/mL) and migrated cells were counted using a Fortessa Flow Cytometer.

MMP-9 activity tests were performed as previously described (34). Briefly, activated MMP-9 was incubated with COAM. Next, dye-quenched gelatin (DQ™-gelatin, Invitrogen) was added at a final concentration of 2.5 µg/mL and the increase in fluorescence over time was measured with the CLARIOstar microplate reader (BMG Labtech).

To measure the CD26 activity, the chromogenic substrate Gly-Pro p-nitroanilide p-toluenesulfonate (GP-p-NA) was used. CD26 was incubated with COAM and mixed with 500 µM Gly-Pro-pNA in 25 mM Tris Buffer (pH 8.0). CD26 activity was determined kinetically (every 5 minutes for 2h at 37°C) by measuring the enzymatic conversion of GP-p-NA to p-NA (absorbance at 405 nm).

Fibrin hydrogels were prepared as described previously (35, 36). Briefly, plasminogen-depleted human fibrinogen (Enzyme Research Laboratories, USA) was prepared in fibrinogen buffer (20 mM HEPES and 150 mM NaCl). Thrombin (Sigma) and factor XIII (Fibrogammin, CSL Behring, Germany) were prepared in thrombin buffer (20 mM HEPES, 150 mM NaCl, 40 mM CaCl₂ and 0.1% BSA) and kept in a water bath at 37°C for 30 min to activate factor XIII to factor XIIIa. Fibrinogen and thrombin components were mixed to prepare hydrogels with 3.5 mg/ml fibrinogen, 0.1 U/ml thrombin and 0.1 U/ml factor XIII. SDF-1, COAM or SDF-1/COAM were incorporated into fibrinogen component preparation before mixing with thrombin components.

Full-thickness excisional wounds were induced on the back of C57BL/6J mice. All procedures were conducted in accordance with the regulations of the European Union (directive 2010/63/EU) and were approved by the local Animal Ethics Committee of KU Leuven (License numbers P270/2015, P128/2019). Briefly, mice were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg) and the hair was removed using a shaver and hair removal cream. The skin was lifted at the mid-dorsal line, folded and punched through with a 4mm disposable sterile biopsy punch, creating one wound on each side of the dorsal midline. The procedure was repeated one more time, generating 4 wounds per animal. Fibrin hydrogels containing SDF-1, COAM or SDF-1/COAM were applied locally in a volume of 25 µl per wound. The control mice were treated only with fibrin hydrogels (Vehicle). SDF-1 was used at 1 µg and COAM at 100 µg (mass excess of 100x) in the hydrogels applied to the wounds. Digital images of the wounds were captured and the wound sizes were quantified with the software ImageJ. Wound healing analysis was expressed as percent healed in relation to wound diameter on day 0 and expressed as area under the curve (AUC).

Data analysis was done using GraphPad Prism 9.1.2. Data were shown as mean ± SEM. A normality test was applied and all data followed a normal distribution. Therefore, the Student´s two-tailed unpaired t-test was used when analyzing two groups. Otherwise, the one-way ANOVA test was used. A two-way ANOVA test was used for the analysis of wound repair over time. Statistical significance was determined at p < 0.05.

SDF-1 is a substrate for many proteases, including MMPs and serine proteases. Studied MMPs cleave SDF-1 between the 4^th (serine/Ser/S) and 5^th amino acid (leucine/Leu/L) in the N-terminal region and inactivate SDF-1 (15) ( Figure 1A ). Using different ratios of protease versus substrate we confirmed the cleavage of SDF-1 by MMP-9 and MMP-2, illustrated by the lower molecular protein band corresponding to the cleaved SDF-1 form ( Figure 1B ). Interestingly, when SDF-1 was incubated in the presence of COAM (100x mass excess versus SDF-1) prior to proteolysis by MMPs, the reduction in molecular weight as a result of SDF-1 cleavage was not observed, suggesting that COAM protected SDF-1 from proteolysis by MMP-9 and MMP-2 ( Figure 1B ). Similar results were obtained when proteolysis was induced by neutrophil degranulates (NDs) ( Figure 1B ), containing proteases with different catalytic mechanisms (37). The latter experiments suggested that COAM protected SDF-1 against proteolysis in an unselective way. N-terminal sequencing analysis of SDF-1 incubated with MMP-9 resulted in the sequence LSYR, confirming the expected cleaved site by MMP-9 between the serine and leucine amino acids ( Figure 1C ; sample 2). In samples, in which SDF-1 was incubated with COAM before the cleavage by MMP-9, both KPVSL and LSY sequences were detected, which corresponded to uncut and cleaved forms, respectively ( Figure 1C ; sample 3). In the latter specific sample, the most predominant form detected was the uncut form, since higher percentages of the amino acids detected in each cycle corresponded to the uncut SDF-1 form (71% Edman cycle 1 corresponding to Lys, 87% cycle 2 corresponding to Pro and 95% cycle 3 corresponding to Val).

For complementation, we performed additional experiments in which we used a monoclonal antibody able to detect only the intact form of SDF-1 (visualized with the use of green fluorescence), together with a polyclonal antibody able to recognize both cleaved and uncut forms (indicated by red fluorescence). Marginal detection of the uncut form was observed when SDF-1 was incubated with MMP-9, confirming the high efficiency of MMP-9 in cleaving SDF-1. In contrast, pre-incubation with COAM at 100x excess resulted in significant protection against SDF-1 cleavage (around 80% of uncut form remaining) ( Figure 1D ). At the intermediate COAM concentration tested (10x) a protection of around 12% against degradation was observed. Together, these data confirmed that COAM indeed protected SDF-1 against degradation by MMP-9 and other proteases.

SDF-1 binds to the cell surface receptor CXCR4 and triggers activation of diverse signaling pathways, to promote CXCR4-mediated chemotaxis. Therefore, it was relevant to study whether SDF-1 (after proteolysis by MMP-9 in the absence or presence of COAM) was biologically active and able to activate CXCR4 or was devoid of activity. To test this, we performed functional assays using THP-1 cells, a monocytic cell line that expresses the CXCR4 receptor (38) ( Figure 2A ). First, we evaluated intracellular calcium mobilization ([Ca²⁺]i) following THP-1 cell stimulation with SDF-1 forms. The clear intracellular calcium signal seen upon stimulation with intact SDF-1 was completely abolished by MMP-9 proteolysis of SDF-1 ( Figure 2B ). Pre-incubation with COAM 100x maintained the ability of SDF-1 to induce calcium signals at similar levels as observed by the uncut form. Moreover, COAM 10x pre-incubation was sufficient to induce a significant SDF-1 mediated calcium signal ( Figure 2B ). The phosphorylation of the kinase ERK1/2 was also investigated in response to SDF-1 stimulation. Whereas stimulation with uncut SDF-1 resulted in increased ERK1/2 phosphorylation, SDF-1 cleaved by MMP-9 failed to trigger ERK1/2 activation. Interestingly, both COAM 10x and 100x treatments were able to induce ERK1/2 phosphorylation at comparable levels to the uncut SDF-1 ( Figure 2C ).

Another response that is induced by SDF-1 binding to CXCR4 is the recruitment of β-arrestin, which can lead to receptor internalization and enhancement of MAPK pathway activation. Therefore, we also tested the ability of SDF-1 forms to induce CXCR4 internalization. SDF-1 forms incubated with COAM 100x in the presence of proteases promoted CXCR4 internalization at similar levels to the uncut form (25% of receptor internalization), whereas cleaved SDF-1 had no effect on surface receptor expression ( Figure 2D ). Similar to the effect observed in calcium signaling experiments, COAM 10x induced less but still significant receptor internalization.

Lastly, SDF-1-induced chemotaxis was evaluated and, as expected, the ability of SDF-1 to induce cell migration was completely inhibited with the cleaved SDF-1 form and entirely restored with the COAM 100x pre-incubated form ( Figure 2E ). Once more, the intermediate concentration of COAM (10x) induced intermediate but significant cell migration.

Together, these data indicate that COAM protects SDF-1 against processing and inactivation by MMP-9, maintaining SDF-1 biological activity.

Dipeptidyl peptidase IV (DPPIV) or CD26 is a serine protease known to cleave SDF-1 between the 2^nd (proline/Pro/P) and 3^rd amino acid (valine/Val/V) in the amino terminus ( Figure 3A ). Considering the fact that CD26 is an important inflammatory regulator of SDF-1 activity (39), we investigated whether the proteolytic protection of SDF-1 by COAM is also applicable to CD26 proteolysis. Surprisingly, we observed that COAM inhibited SDF-1 degradation by CD26 more efficiently than the protection observed for MMP-9. A complete protection against proteolysis was already seen upon pre-incubation with an equal mass of COAM ( Figure 3B ). Similar levels of protection were observed with COAM 10x and 100x mass excess pre-incubation. We confirmed this finding through the functional evaluation in the intracellular calcium signal assay in response to SDF-1 stimulations and reinforced once more that SDF-1 bound to COAM is still active and able to induce cellular responses at similar levels to the intact form ( Figure 3C ).

To investigate if COAM could have a direct effect on the catalytic activities of the used proteases or if other mechanisms may be involved, we performed MMP-9 and CD26 enzymatic activity assays. No major changes in MMP-9 or CD26 enzymatic activities were detected in the presence of COAM ( Figure 4A ), thus confirming that COAM did not interfere directly with enzyme activities. These findings supported the idea that direct binding of COAM to SDF-1 and protection against proteolysis are involved in the effects observed above. Due to the negatively charged nature of COAM and the positive charges in SDF-1, we hypothesized that the interaction between these two components is direct and dependent on electrostatic charges. To confirm this, we compared proteolysis experiments in the presence and absence of an excess of NaCl to neutralize the electrostatic interactions. While at physiological salt concentrations (150mM NaCl) the SDF-1 proteolytic protection by COAM was clearly observed, an excess of NaCl (500mM NaCl) inhibited the ability of COAM to protect SDF-1 from MMP-9 and CD26 degradation ( Figure 4B ). Importantly, such high NaCl concentrations did not interfere with the capacity of MMP-9 and CD26 to cleave SDF-1. To illustrate possible electrostatic interactions of the negatively charged COAM with SDF-1, the amino-acid sequence of SDF-1, highlighting the positively (blue) and negatively (red) charged amino acids, and the surface positions of these charges onto the folded protein (3D structure) are shown in Figure 4C . Together, these data suggest that the beneficial effect of COAM is associated with its capacity to bind to SDF-1 and protect it against proteolysis in a charge-dependent way.

In various in vivo studies SDF-1 had beneficial effects on wound healing (12–14). Considering the proteolytic inactivation of SDF-1 by proteases present under inflammatory conditions (e.g. neutrophil proteases, MMP-9, CD26) and our finding of protection of SDF-1 against proteolysis with the use of COAM, we hypothesized that the combination of COAM with SDF-1 might be a good application to deliver SDF-1 into wounds and to avoid the fast degradation by several proteases present in the wound environment. To investigate this, we delivered SDF-1 and COAM incorporated into a fibrin hydrogel and analyzed the wound repair over time. We compared the effect of SDF-1/COAM combination treatment with control (Fibrin Gel), COAM (Fibrin Gel + COAM) and SDF-1 (Fibrin Gel + SDF-1). We observed that wounds treated with the combination SDF-1/COAM had a significant reduction in wound size on days 1 and 3 when compared to wounds of the control group (Fibrin Gel) ( Figures 5A, B ). When compared to wounds treated with COAM (Fibrin Gel + COAM), the SDF-1/COAM combination showed wound size reduction only at day 1 post-injury. The difference between wounds treated with SDF-1/COAM combination and wounds treated only with SDF-1 (Fibrin Gel + SDF-1) extended until 7 days post-wounding. Analysis of the areas under the curve (AUC) better illustrated these differences and highlighted the positive effect of SDF-1/COAM combination when compared to Fibrin Gel and Fibrin Gel + SDF-1 groups ( Figure 5C ).