Research ArticleWound Healing

Glycosaminoglycan-based hydrogels capture inflammatory chemokines and rescue defective wound healing in mice

See allHide authors and affiliations

Science Translational Medicine  19 Apr 2017:
Vol. 9, Issue 386, eaai9044
DOI: 10.1126/scitranslmed.aai9044

Capturing chemokines in chronic wounds

Chronic, poor healing wounds are characterized by the lack of resolution of initial proinflammatory signaling present during acute injury. Lohmann et al. designed a synthetic hydrogel wound dressing based on heparin, a glycosaminoglycan that can bind and sequester chemokines. The hydrogel mopped up inflammatory chemokines such as MCP-1 and IL-8 from fluid from patients’ chronic venous leg ulcers in vitro and inhibited neutrophil and monocyte migration. Applying the hydrogel to skin wounds in diabetic mice improved wound healing and vascularization and reduced inflammation more effectively than the FDA-approved hydrogel Promogran. Capturing chemokines may be an effective strategy to promote tissue regeneration in chronic wounds.


Excessive production of inflammatory chemokines can cause chronic inflammation and thus impair cutaneous wound healing. Capturing chemokine signals using wound dressing materials may offer powerful new treatment modalities for chronic wounds. Here, a modular hydrogel based on end-functionalized star-shaped polyethylene glycol (starPEG) and derivatives of the glycosaminoglycan (GAG) heparin was customized for maximal chemokine sequestration. The material is shown to effectively scavenge the inflammatory chemokines MCP-1 (monocyte chemoattractant protein–1), IL-8 (interleukin-8), and MIP-1α (macrophage inflammatory protein–1α) and MIP-1β (macrophage inflammatory protein-1β) in wound fluids from patients suffering from chronic venous leg ulcers and to reduce the migratory activity of human monocytes and polymorphonuclear neutrophils. In an in vivo model of delayed wound healing (db/db mice), starPEG-GAG hydrogels outperformed the standard-of-care product Promogran with respect to reduction of inflammation, as well as increased granulation tissue formation, vascularization, and wound closure.


Chemokines are a class of signaling molecules of which some selectively recruit and activate cells during inflammation (1). As a consequence, chemokines can trigger the development of a broad range of inflammatory diseases including chronic wounds (2). During the wound healing process, chemokines are released by tissue-specific cells and resident immune cells at the site of injury to establish a chemoattractant gradient that promotes the invasion of blood-derived immune cells, which are essential for the initial inflammatory phase of acute wound healing (3, 4). In chronic wounds, the healing process is often deadlocked in an unrestrained inflammatory response (5). Under this condition, the uncontrolled activation of polymorphonuclear neutrophils (PMNs) and inflammatory monocytes/macrophages leads to a destruction of the wound tissue and drives the abundant production of inflammatory mediators including chemokines. The persistent chemoattractant gradients—predominantly based on monocyte chemoattractant protein–1 (MCP-1) and interleukin-8 (IL-8)—promote further invasion of these immune cells into the wound tissue, thus perpetuating a vicious circle of chronic inflammation (3, 4, 68).

One important feature of chemokines is their ability to bind to extracellular matrix glycosaminoglycans (GAGs), such as heparan sulfate or heparin (9, 10), a process that is mediated by electrostatic interactions of positively charged amino acid residues of the chemokines and negatively charged sulfate groups of the GAGs. Accordingly, the sulfation degree and concentration of the GAGs modulate multiple binding events and thus the transport of chemokines within extracellular matrices, enabling the formation of long-range chemokine gradients that control immune cell activation and migration (11, 12). Furthermore, GAGs protect chemokines against proteolytic inactivation (10, 13, 14), mediate chemokine oligomerization for receptor activation (9, 15), and accumulate chemokines near the cell surface (16). Therefore, GAG-based engineered materials may allow for modulating chemokine concentrations within tissues to therapeutically attenuate inflammation in chronic wounds (17).

To test this hypothesis, hydrogels composed of star-shaped polyethylene glycol (starPEG) and GAGs were customized to suppress persistent proinflammatory chemokine gradients in chronic wounds. The GAG heparin, carrying the highest anionic charge density of all known biopolymers, was chosen as the chemokine-scavenging component (15) within a sterically easily accessible bulk hydrogel network. Wound dressings formed out of these biohybrid gels were expected to serve as an efficient “molecular sink,” sequestering high amounts of chemokines from the inflamed wound, thereby preventing further recruitment of immune cells to ultimately resolve the inflammatory process. Because the sulfation pattern of heparin governs its cytokine binding profile (18, 19), a set of selectively desulfated heparin derivatives was used for the formation of the biohybrid polymer gels and compared with respect to the resulting scavenging and proregenerative characteristics of the materials (Fig. 1). The scavenging characteristics of the materials were compared in binding assays using recombinant human chemokines MCP-1 and IL-8, conditioned medium containing a mix of cell-derived inflammatory cytokines, and wound exudates of human patients suffering from chronic venous leg ulcers. Transmigration assays and a murine model of full-thickness excisional wounds were performed to investigate the consequences of gel-based chemokine sequestration. A model of delayed cutaneous wound healing (db/db mice) was applied to evaluate the overall proregenerative effect of the starPEG-GAG wound dressings.

Fig. 1. Schematic of the proposed function of starPEG-GAG hydrogels for modulating chemokine gradients in chronic wounds over time.

From left to right: Chemokines are secreted from the injured tissue and mediate immune cell influx. Immune cells from the bloodstream invade the wound bed and produce more inflammatory chemokines, creating a circle of immune cell invasion and chemokine release, which eventually results in the persistent inflammatory environment found in chronic wounds. StarPEG-GAG hydrogel networks [depicted in (A)] can bind and neutralize chemokines through strong electrostatic interactions of heparin derivatives and chemokines [depicted in (B)]. Chemokine sequestration by the hydrogel scaffolds results in a reduced immune cell invasion, which in turn, lowers the concentrations of inflammatory chemokines and ultimately allows the inflammation to resolve.


StarPEG-GAG hydrogels were engineered to scavenge inflammatory chemokines

A previously established modular hydrogel system based on starPEG and heparin (2022) was customized to modulate the inflammatory environment and sequester chemokines from the inflamed wound, thereby preventing further recruitment of immune cells to ultimately resolve the inflammatory process and support cutaneous wound healing (Fig. 1). First, the polymer network properties of the starPEG-GAG hydrogels had to be optimized to meet the requirements of the topical skin application and to maximize chemokine scavenging. Accordingly, the cross-linking degree of all hydrogels, including GAG-free starPEG-starPEG gels and starPEG-GAG gels containing heparin derivatives with different sulfation patterns, was adjusted to result in a storage modulus of ~3.0 ± 1.1 kPa (Young’s modulus of ~9 ± 3.3 kPa, assuming a Poisson ratio of 0.5; see Fig. 2A), which is compliant with the Young’s modulus of human skin (4.5 to 8 kPa) (23), and a mesh size of >10 nm, which exceed the size of wound relevant chemokines to be scavenged [radius of gyration (Rg) of MCP-1 and IL-8, <2 nm; see Fig. 2, A and B] (3, 4, 7). The similar physical properties of the hydrogels ensure that the observed chemokine binding is determined by the net charge and charge distribution within the polymer network rather than by steric effects (Fig. 2C). To alter the chemokine binding affinity, we selectively removed the sulfate groups of heparin according to previously established protocols (Fig. 2D) (24), resulting in hydrogels with variable sulfation patterns and independently adjustable mechanical properties (25), containing GAG units with three sulfate groups (heparin; SH), two sulfate groups (N-desulfated heparin; N-dSH), and one sulfate group (6-ON-desulfated heparin; 6ON-dSH) per disaccharide unit (Fig. 2A).

Fig. 2. Physical and biofunctional characteristics of hydrogel scaffolds formed by cross-linking of starPEG and different heparin derivatives (starPEG-GAG hydrogels).

(A) Storage modulus, mesh size, and sulfate content of the compared hydrogels (PEG/PEG, PEG/SH, PEG/N-dSH, and PEG/6ON-dSH). (B) Schematic representation of hydrogel mesh size (11 nm) and gyration radius (2 nm) of MCP-1 and IL-8. (C) Results of computational docking analyses of both chemokines using ClusPro software. The binding affinity to heparin (Embedded Image) was derived from literature (2633). All other physical properties of the proteins were computed from previously published structural models (80, 81). MW, molecular weight; pI, isoelectric point. (D) Structures of native heparin and selectively desulfated heparin derivatives. (E) Binding kinetics of MCP-1 and IL-8 to the compared hydrogel types. One hundred nanograms of the chemokines was added to 35 μl of the different hydrogels; sequestration was determined after 24 hours as bound amount of chemokine. (F) Chemokine binding of compared hydrogel types after 24 hours at different chemokine concentrations. (G) Estimated chemokine saturation of GAG-based hydrogels assuming one chemokine binding site per heparin molecule. Data show mean values ± SD of three independent experiments. Analysis of variance (ANOVA) with multiple comparisons versus PEG/PEG (E) or PEG/SH (F) using Bonferroni t test: ***P ≤ 0.001, **P ≤ 0.01. ns, not significant.

Next, the binding capabilities of the different hydrogels for the chemokines MCP-1 and IL-8, involved in the attraction of immune cells and abundantly expressed in chronic wounds (3, 4, 7), were investigated systematically. The binding kinetics of the chemokines to the four hydrogel types were determined by incubation with either 100 ng of MCP-1 or IL-8 per 35 μl of scaffold (2.85 ng cytokine/μl gel) (Fig. 2E). After 24 hours, the chemokines administered in the assay were nearly quantitatively scavenged by all of the compared GAG-based hydrogels independent of the hydrogel sulfation patterns. After 3 hours, about 50% of the chemokines were sequestered by the hydrogels formed from the different heparin derivatives. In contrast, only 20% of the chemokines were bound to the PEG/PEG reference gels with no further increase after the first 3 hours of incubation. The observation of a strong interaction between chemokines and starPEG-GAG hydrogels is in line with previous results, reporting dissociation constant (Kd) values of MCP-1 and IL-8 in the lower micromolar range (Kd, 1 to 2 μM) for soluble heparin (Fig. 2C) (2633).

Moreover, the binding capacity of the different materials for chemokines was analyzed by incubation with increasing concentrations of MCP-1 or IL-8 for 24 hours (Fig. 2F). A linear correlation of the amounts of deployed and bound chemokine was found for gels made of all compared heparin derivatives. For all chemokine concentrations, PEG/SH hydrogels scavenged almost 100% of the applied MCP-1 but only 70% of the applied IL-8 (Fig. 2F). The stepwise removal of heparin sulfate groups was found to influence the binding of MCP-1. The removal of the 6-O-sulfate group particularly affected the binding of MCP-1, whereas the removal of the N-sulfate did not alter the overall sequestration (Fig. 2F; MCP-1). In contrast, the sequestration efficacy of the hydrogels for IL-8 was not affected by a specific sulfate position but rather by the overall sulfate content (Fig. 2E, IL-8). Although differences in the overall affinity of the different hydrogels for MCP-1 and IL-8 were detected, no saturation of the scaffolds was observed in the assessed concentration range for either of the chemokines tested. On the basis of the conservative assumption of one chemokine binding site per heparin molecule, it was calculated that even with the high loading amount of 1000 ng chemokine/35 μl scaffold (28.5 ng chemokine/μl gel; see Fig. 2F), far exceeding the chemokine concentrations found in chronic wound environments (MCP-1, 0.9 ng/ml; IL-8, 702 ng/ml; table S1), only less than 5% of the available binding sites of the GAG-containing hydrogels will be occupied. This, in turn, would allow for scavenging of up to 30 μg chemokine/μl hydrogel (100% binding; see Fig. 2G).

StarPEG-GAG hydrogels reduce migration of immune cells

To assess the binding of chemokines to starPEG-GAG hydrogels in a biologically relevant inflammatory context, we incubated the biohybrid scaffolds with conditioned medium from inflammatory activated macrophages or dermal fibroblasts. Both cell types play a central role in the wound healing process (8) and produce a plethora of mediators upon inflammatory activation, including tumor necrosis factor (TNF), IL-1β, IL-6, MCP-1, and IL-8 (fig. S1). This mix of inflammatory cytokines and chemokines closely resembles the signaling environment found in chronic wounds (34, 35). After incubation of the hydrogel scaffolds with conditioned medium for 24 hours, the concentrations of the nonbound cytokines or chemokines were determined by enzyme-linked immunosorbent assay (ELISA). The results demonstrate high binding capacities of the starPEG-GAG hydrogels but almost no binding capacity of the inert PEG/PEG control gels (Fig. 3A and fig. S1). StarPEG-GAG scaffolds were found to preferentially sequester the chemokines MCP-1 and IL-8, whereas other proinflammatory cytokines such as TNF, IL-1β, and IL-6 were unaffected (Fig. 3A). Furthermore, differences in the capacity of the starPEG-GAG hydrogels to bind MCP-1 and IL-8 were observed depending on the sulfation degree of the different heparin derivatives (Fig. 3, A and B). Whereas scaffolds based on fully sulfated heparin (PEG/SH) scavenged about 80% for both chemokines, the scaffolds based on PEG/N-dSH depleted about 50% of MCP-1 and IL-8. Moreover, the hydrogels based on 6ON-dSH (PEG/6ON-dSH) scavenged even less chemokine (MCP-1, 25%; IL-8, 45%). The decrease in the chemokine binding capability correlated with the overall reduced sulfation degree of the particular hydrogels and resembled data obtained with recombinant chemokines (see Figs. 2 and 3B).

Fig. 3. Characterization of binding and functional inactivation (scavenging) of cell-derived chemokines MCP-1 and IL-8 by starPEG-GAG hydrogels.

(A) Chemokine binding assay: Conditioned medium derived from human activated dermal fibroblasts (dFb) or inflammatory macrophages (iMΦ) was incubated with the different hydrogels (PEG/PEG, PEG/SH, PEG/N-dSH, and PEG/6ON-dSH). After 24 hours, the remaining cytokines were quantified by ELISA, and the amounts of hydrogel-bound cytokines were calculated. Bars represent calculated mean values ± SD of experiments using four dermal fibroblast and seven different macrophage donors. (B) Correlation of the sulfation degree of heparin with the chemokine binding capacity of the different hydrogels. (C) Transmigration assay: Conditioned medium from proinflammatory dermal fibroblasts incubated for 24 hours on the different hydrogels (PEG/PEG, PEG/SH, PEG/N-dSH, and PEG/6ON-dSH) or without hydrogel (no hydrogel) was used as chemotactic stimuli in a Transwell migration assay with human monocytes or human PMNs. Migration toward PEG/PEG was defined as 100%. Bars represent calculated means ± SD of experiments using three different monocyte and three different PMN donors. ANOVA with multiple comparisons versus PEG/PEG using Bonferroni t test: ***P ≤ 0.001, **P ≤ 0.01, *P ≤ 0.05.

Next, the impact of the hydrogel scaffolds on the chemoattractant function of MCP-1 and IL-8 was investigated in transmigration assays using primary human monocytes or PMNs, which are predominantly attracted by both chemokines and found abundantly in chronic wounds (36, 37). Conditioned medium from activated dermal fibroblasts was used as chemoattractant stimulus to promote the migration of monocytes and PMNs (fig. S2). In accordance with the effective sequestration of MCP-1 and IL-8, transmigration of both monocytes and PMNs was significantly reduced [monocytes: P < 0.001 (PEG/PEG versus PEG/SH), P = 0.033 (PEG/PEG versus PEG/N-dSH); PMNs: P = 0.005 (PEG/PEG versus PEG/SH), P = 0.009 (PEG/PEG versus PEG/N-dSH] after incubation of conditioned medium with PEG/SH or PEG/N-dSH hydrogels (Fig. 3C). In comparison, no effect of the PEG/PEG hydrogels on the chemoattractant activity of the conditioned medium was detected (Fig. 3C and fig. S2). For PEG/6ON-dSH hydrogels, only a marginal decrease in the migratory activity of the cells was observed, which is in line with its lower capacity to bind MCP-1 and IL-8 (Fig. 3C). Similar effects of chemokine-binding starPEG-GAG hydrogels on the transmigration of monocytes and PMNs were observed in single mediator assays using recombinant human MCP-1 or IL-8, respectively (fig. S3A).

In summary, these data demonstrate that starPEG-GAG hydrogels specifically and efficiently bind MCP-1 and IL-8 in dependence on the particular GAG sulfate content (see fig. S3B), resulting in a depletion of the chemokines from solutions adjacent to the hydrogels, and considerably reduce the functional activity of both chemokines toward immune cells.

Chemokine scavenging decreases immune cell influx and inflammatory signaling in wound sites

To test the chemokine scavenging by starPEG-GAG–based hydrogels in a complex inflammatory environment in vivo, where a multitude of mediators is continuously produced, we used the murine model of full-thickness excisional wounds. Therefore, preformed PEG/SH and PEG/N-dSH hydrogels, the two materials with the highest effect on immune cell migration, were applied onto wounds directly after infliction on the back of 12-week-old C57BL/6 mice. The materials caused no adverse reactions, such as hemorrhage, infection, or increased inflammation, over the entire course of the experiment (Fig. 4A). Five days after wounding, the hydrogels were recovered from the wounds and analyzed for the amount of bound chemokines. High amounts of bound MCP-1 were determined in PEG/SH and PEG/N-dSH hydrogels, whereas significantly less MCP-1 [P = 0.001 (PEG/PEG versus PEG/SH), P < 0.001 (PEG/PEG versus PEG/N-dSH)] was determined in the PEG/PEG control hydrogels (Fig. 4B). IL-8 could not be assessed because no murine homolog of IL-8 is known to date (38). Instead, high amounts of the PMN chemoattractant chemokine epithelial neutrophil-activating peptide 78 [ENA-78 (39)] were detected particularly within the PEG/N-dSH hydrogel (Fig. 4B). To investigate the functional impact of chemokine binding to the starPEG-GAG hydrogels, we assessed the influx of monocytes and PMNs into the wounds. High amounts of PMNs and monocytes were found in wounds treated with PEG/PEG control hydrogels compared to the unwounded skin (Fig. 4C and fig. S4). In wounds treated with PEG/SH or PEG/N-dSH hydrogels, the infiltration of PMNs and monocytes was significantly decreased [PMN: P = 0.05 (PEG/PEG versus PEG/SH), P = 0.05 (PEG/PEG versus PEG/N-dSH); monocytes: P = 0.002 (PEG/PEG versus PEG/SH), P = 0.008 (PEG/PEG versus PEG/N-dSH)], which corresponds to the chemokine sequestration from the wound site (Fig. 4, B and C). Immunofluorescence analysis of wound sections revealed no accumulation of CD11b+ immune cells at the surface or within the network of the starPEG-GAG hydrogels, demonstrating that immune cells do not accumulate within the scaffolds (fig. S5).

Fig. 4. Characterization of scavenging effects of starPEG-GAG hydrogels during wound healing in mice.

Wounds on the backs of C57BL/6 wild-type (WT) mice were inflicted by 6-mm punch biopsy and treated with hydrogel discs (wt: PEG/PEG, PEG/SH, and PEG/N-dSH) for 5 days. (A) Wounds containing hydrogels 5 days after wounding. (B) Analysis of chemokine binding: Hydrogels were recovered from the wounds, and binding of chemokines from the wound site was determined by multiplex immunoassay in supernatants of disrupted hydrogels after mediator displacement. (C) Characterization of immune cell influx: Wounds were harvested and digested to obtain single-cell suspensions to assess influx of monocytes and PMN into the wound area using flow cytometry (fig. S4). Single-cell suspensions from unwounded skin served as control. (D) Chemokine concentrations in wounds: Protein was isolated from whole wound tissue, and concentrations of chemokines were determined by multiplex immunoassay. (E) Characterization of inflammation: RNA was isolated from whole wound tissue, gene expression was analyzed, and expression was calculated compared to unwounded skin. Each symbol represents one wound. Bars represent means ± SD. ANOVA with multiple comparisons versus PEG/PEG using Bonferroni t test or Dunnett’s method: ***P ≤ 0.001, **P ≤ 0.01, *P ≤ 0.05 (B and C). Unpaired t test: **P ≤ 0.01, *P ≤ 0.05 (D and E).

Chemokine concentrations were quantified in the tissue to verify the hydrogel-driven depletion of the chemokines from the wound site. A significant reduction [MCP-1, P = 0.032; ENA-78, P = 0.045; growth-related oncogene–α (GRO-α), P = 0.022; macrophage inflammatory protein–1β (MIP-1β), P = 0.01 (PEG/PEG versus PEG/N-dSH)] of the monocyte- and PMN-attracting chemokines (MCP-1, ENA-78, GRO-α, and MIP-1β) was found after application of starPEG-GAG hydrogels compared to wounds treated with the PEG/PEG control (Fig. 4D). Moreover, the concentrations of the non–heparin-affine proinflammatory mediators TNF and IL-1β were significantly reduced [TNF, P = 0.047; IL-1β, P = 0.016 (PEG/PEG versus PEG/N-dSH)] in the wound tissue (fig. S6). Gene expression analysis of the wound tissue for proinflammatory mediators revealed a significant down-regulation [MCP-1, P < 0.001; GRO-α, P = 0.009; IL-1β, P = 0.001 (PEG/PEG versus PEG/N-dSH); MCP-1, P = 0.002, IL-1β, P < 0.001, TNF, P = 0.004 (PEG/PEG versus PEG/SH)] of the expression of MCP-1, GRO-α, IL-1β, and TNF in wounds treated with PEG/SH or PEG/N-dSH hydrogels compared to wounds treated with the PEG/PEG controls (Fig. 4E). Accordingly, chemokine scavenging and the consequentially decreased immune cell influx result in a markedly reduced inflammatory environment in wounds treated with starPEG-GAG hydrogels.

StarPEG-GAG hydrogels enhance wound healing in conditions of impaired regeneration

Next, we assessed the starPEG-GAG hydrogels under pathological conditions of chronic inflammation and impaired wound healing using 12-week-old diabetic db/db mice. PEG/N-dSH scaffolds with high in vivo chemokine-scavenging activity but low antithrombogenic activity (40) were compared to the standard-of-care product Promogran, a U.S. Food and Drug Administration (FDA)–approved hydrogel-based wound dressing for the management of chronic wounds including diabetic ulcers (4145). Promogran and PEG/N-dSH hydrogels were applied immediately after wounding of the db/db mice. The impact of these wound dressings on the inflammatory response and tissue regeneration was analyzed over the course of wound healing. High amounts of MCP-1 were recovered from the PEG/N-dSH scaffolds from wounds 5 and 10 days after wounding (Fig. 5A). Consequently, significantly decreased concentrations [day 5: MCP-1, P = 0.001; IL-1β, P < 0.05; day 10: IL-1β, P < 0.05] of MCP-1 and IL-1β were found in the tissue of wounds treated with PEG/N-dSH compared to untreated controls, indicating a reduced inflammatory response during wound healing (Fig. 5B). In contrast, in wounds treated with Promogran, significant depletion (MCP-1, P = 0.002; IL-1β, P < 0.05) of MCP-1 and IL-1β was only found 5 days after wounding, whereas at day 10 MCP-1 was increased. This suggests a more effective and sustained specific scavenging of chemokines from the wound site by the PEG/N-dSH hydrogel compared to the unspecific and temporary absorption of some factors (MCP-1 and IL-1β) by Promogran. Accordingly, a significantly greater reduction (day 5, P = 0.041; day 10, P = 0.024) of infiltrating CD11b+ immune cells comprising monocytes and PMN was detected in wounds treated with PEG/N-dSH compared to Promogran 5 and 10 days after wounding (Fig. 5C).

Fig. 5. Chemokine scavenging and promotion of wound healing by starPEG-GAG hydrogels in conditions of impaired skin regeneration.

(A to F) Wounds were inflicted on the back of db/db mice as described in Fig. 4 and treated with PEG/N-dSH or Promogran. (A) Analysis of chemokine binding in hydrogels recovered from the wounds as in Fig. 4. (B) Characterization of inflammation 5 and 10 days after wounding: Determination of protein concentrations of MCP-1 and IL-1β in whole wound tissue as in Fig. 4. dpw, days postwounding (C) Characterization of immune cell invasion. Five dpw: Quantification of CD11b+ cells within the wound tissue by flow cytometry as in Fig. 4; 10 dpw: CD11b immunofluorescence staining of histological sections and quantified cell count of CD11b+ cell within wound area in CD11b immunofluorescence staining. DAPI, 4′,6-diamidino-2-phenylindole. (D) Wounds 10 dpw and analysis of wound closure: Distance between epithelial tips measured in pan-cytokeratin (panCK) immunofluorescence staining. (E) Analysis of granulation tissue. Ten dpw: panCK/αSMA immunofluorescence staining of histological sections. FL, fluorescence. (F) Analysis of angiogenesis 10 dpw: CD31 immunofluorescence staining of histological sections. All immunofluorescence: g, granulation tissue; e, epidermis; f, fat tissue; s, scab. Arrow head marks epithelial tip, and dotted line marks border between epidermis and dermis. Scale bars, 200 μm. Each symbol represents one wound. Bars represent means ± SD. ANOVA with multiple comparisons versus untreated control (ctr) using Bonferroni t test or Dunnett’s method: ***P ≤ 0.001, **P ≤ 0.01, *P ≤ 0.05 (B). Unpaired t test: **P ≤ 0.01, *P ≤ 0.05 (C to F). (G) Exudates obtained from chronic venous leg ulcers of patients (n = 6) were incubated with PEG/PEG and PEG/N-dSH or without hydrogel (control). After 24 hours, concentrations of mediators were determined by multiplex immunoassay. Each symbol represents one exudate sample. Paired t test: **P ≤ 0.01, *P ≤ 0.05.

Examination of the newly formed tissue 10 days after wounding revealed a faster reepithelization in the presence of the PEG/N-dSH hydrogels with wounds already covered by a fully closed neoepithelium (Fig. 5, D and E). The extent of mature α-smooth muscle actin (αSMA)–positive granulation tissue that developed in the presence of the PEG/N-dSH hydrogels was significantly increased (P = 0.047) and characterized by a pronounced cellularity compared to wounds treated with Promogran (Fig. 5E). Quantitative analysis of vascular structures by CD31 staining showed a significantly (P = 0.006) accelerated angiogenic response induced by the PEG/N-dSH hydrogels in comparison to Promogran (Fig. 5F). Bacterial wound infections represent another important reason for impaired wound healing (46). Therefore, we asked the question whether a reduced inflammatory response caused by our starPEG-GAG hydrogels might favor spreading of bacterial infections. PEG/N-dSH hydrogels were applied to wounds of diabetic mice (db/db) that had been infected with Staphylococcus aureus, a bacterial strain frequently found on infected diabetic ulcers (46). Quantification of the bacterial infection 10 days after wounding showed a similar bacterial load as nontreated wounds (fig. S7A). When PEG/N-dSH hydrogels were applied to these infected wounds, significantly lower concentrations (MCP-1, P = 0.048) of MCP-1 and IL-1β were found in the wound areas (fig. S7B); moreover, a substantial amount of MCP-1 could be recovered from the PEG/N-dSH hydrogels that were removed from the infected wounds (fig. S7C). These results confirm that chemokine scavenging by PEG/N-dSH hydrogels suppresses wound inflammation and effectively promotes wound healing even under conditions of impaired skin regeneration and bacterial infection.

StarPEG-GAG hydrogels scavenge chemokines from human wound fluids obtained from chronic venous leg ulcers

As a next step toward clinical application, we tested whether the starPEG-GAG hydrogels are capable of scavenging chemokines in human wound fluids of patients suffering from nonhealing chronic venous leg ulcers superinfected by bacteria such as S. aureus or Pseudomonas aeruginosa (table S2). Wound exudate samples were collected over 8 hours from wounds with an average area of 35 cm2, as previously described [1 ml/8 hours per 35 cm2 wound surface (47)]. Consistent with other studies (7, 48), high concentrations of proinflammatory chemokines and cytokines including MCP-1, IL-8, MIP-1α, MIP-1β, TNF, and IL-1β (table S1) were quantified in these exudates with peak concentrations of IL-8 ranging from 251 to 1231 ng/ml and of MCP-1 from 173 to 2839 pg/ml (table S2).

After incubation with PEG/N-dSH hydrogels, more than 80% of the initial amounts of IL-8 and MCP-1 were depleted from all exudates (Fig. 5G and fig. S8). Furthermore, 65 to 80% of the chemokines MIP-1α and MIP-1β were depleted from the wound fluids (Fig. 5G and fig. S8). In contrast, proregenerative growth factors were bound to a lesser extent (SDF-1α) or not affected at all (VEGF-A, PLGF-1, or HGF) by the PEG/N-dSH hydrogel (Fig. 5G and fig. S8). These findings suggest a rather selective scavenging pattern of the PEG/N-dSH hydrogel preferentially targeting wound healing, impeding proinflammatory chemokines while sparing wound healing promoting proregenerative growth factors within wound sites. Considering the total binding capacity of the scaffolds of hydrogel (30 μg/μl) as described above, even a 50-μm-thick thin hydrogel film could bind substantial amounts of chemokines from the exudate. On the basis of the average chemokine concentrations found in the wound exudates (table S1) and the exudate production rates reported by Eming et al. (47), the total amount of IL-8 and MCP-1 secreted from the wound tissue over a time course of 3 days, a typical duration until a wound dressing is changed in a patient, would occupy less than 1% of the protein binding sites within the hydrogel.


Overexpression of inflammatory chemokines, such as MCP-1 and IL-8, causes a persistent inflammation and markedly impairs cutaneous wound healing (7). Consequently, this class of signal proteins is an attractive target for novel strategies to control inflammation (10). Current therapeutic approaches in wound healing mainly aim at the inactivation of proteases or the promotion of tissue proliferation rather than addressing the misbalance of inflammatory cues in the chronic wound (8, 4954).

As an alternative, we demonstrate that inflammation can be reduced by modulation of chemokine gradients, using hydrogels as effective chemokine scavengers. Acting as a molecular sink, starPEG-GAG hydrogels were shown to deplete chemokines, particularly MCP-1 (3133) and IL-8 (2629), from the wound. The rather selective binding of chemokines to GAGs (9, 55) results from strong electrostatic interactions involving a GAG-binding consensus sequence with positively charged amino acid residues found in nearly all chemokines (18, 31), which was shown to modulate the bioactivity of the chemokines (30). We found a clear correlation of the binding of these chemokines with the overall negative net charge of the GAGs of the starPEG-GAG hydrogels, allowing for an effective fine-tuning of the sequestration by adjustment of the GAG sulfation pattern.

A rational design concept was applied for tailoring the mechanical properties of the starPEG-GAG hydrogels to match the characteristics of human skin (20, 21). With this approach, the GAG volume concentration of the gels was kept invariant when varying the GAG sulfation pattern. A relatively large mesh size of the polymer hydrogel networks—compared to the dimer size of the chemokines—was chosen to exclude any steric effects on the chemokine uptake of the gels. Instead, the chemokine binding to the hydrogels can be solely attributed to the GAG components serving as affinity centers. Because the heparin derivatives constitute a main building block of the hydrogel network, the materials offer a high capacity for binding of MCP-1 and IL-8. The modular starPEG-GAG hydrogel platform thus allows for tuning chemokine affinity and the capacity to modulate chemokine gradients in an inflammatory context.

As proof of concept, we demonstrate that starPEG-GAG hydrogels can neutralize the chemoattractant function of MCP-1 and IL-8 in vitro and in vivo when applied onto excisional wounds. This sequestration effect results in a significantly reduced influx of immune cells into the wound. Heparin has long been recognized for its anti-inflammatory properties (19). A reduced influx of immune cells to inflamed sites after administration of soluble heparin was demonstrated in various in vivo models of acute inflammation (5658). We detected no binding of the less heparin-affine cytokines TNF, IL-1β, and IL-6 by the hydrogels, whereas chemotactic factors like MIP-1α, MIP-1β, and ENA-78 did interact with the starPEG-GAG hydrogels and may therefore be involved in these regulatory mechanisms as well (55). However, as a secondary effect of the reduced influx of immune cells into the wound, the overall expression of inflammatory cytokines and chemokines such as TNF, IL-1β, GRO-α, and MCP-1 was also diminished. The wound environment benefits from the decreased inflammatory signaling and consequently promotes granulation tissue maturation, vascularization, and reepithelialization.

Finally, to underline the translational potential of the explored concept, PEG/N-dSH hydrogels, showing an optimal chemokine-scavenging potential and low antithrombogenic activity (40), were compared to the standard-of-care Promogran in a db/db mouse model of impaired wound healing. Promogran is an FDA-approved hydrogel-based wound dressing for the management of chronic wounds, including diabetic and venous ulcers. The product is composed of collagen and oxidized regenerated cellulose, which has been shown to absorb large amounts of wound exudate and to inactivate proteases, thus improving impaired wound healing (4145). Wound healing in db/db mice is delayed because of prolonged expression of chemokines and subsequently increased influx of immune cells during the late phases of wound healing (5961). Because this phenotype replicates features of nonhealing wounds in patients with diabetic or chronic venous leg ulcers (5), wound healing studies in db/db mice represent a relevant model for chronic wounds in humans. Even in these highly inflamed wounds, the chemokine-scavenging effects of the PEG/N-dSH hydrogel were sufficient to reduce inflammation, enhance wound closure, and promote tissue granulation and vascularization within only 10 days. Furthermore, the hydrogel was found to be superior in rescuing wound healing deficiency compared to Promogran. As a consequence of the particular physicochemical characteristics adjusted in PEG/N-dSH hydrogels, these materials were shown to advantageously combine a very high affinity for chemokines with a rather moderate affinity for several proregenerative growth factors. Moreover, the rather selective scavenging pattern of the PEG/N-dSH hydrogels observed in wound fluids of patients with chronic venous ulcers shifting the balance of the signaling characteristics from a proinflammatory to a proregenerative state further underpins the translational potential of the chemokine-scavenging hydrogels.

Notably, the binding capacity of starPEG-GAG hydrogels for proinflammatory chemokines far exceeds the amounts of inflammatory mediators detected in the exudate of chronic ulcers. Despite their anti-inflammatory effects, PEG/N-dSH hydrogels neither facilitated bacterial growth within wounds nor supported spreading of the infection beyond wound margins (cellulitis). PEG/N-dSH hydrogels retained their chemokine-scavenging properties in wound fluids from patients suffering from ulcers superinfected by bacteria such as S. aureus or P. aeruginosa. We emphasize that application of starPEG-GAG hydrogels do not have an antimicrobial activity. Thus, an application of these hydrogels onto infected wounds will require additional previous therapeutic measures, such as surgical debridement, application of topical antiseptics, or even systemic antibiotics, to remove necrotic tissue and biofilms and to minimize the bacterial load (62). In this context, preloading of the starPEG-GAG hydrogels with antimicrobials as recently reported by Fischer et al. (63) might create valuable additional therapeutic options.

Clinical translation of starPEG-GAG hydrogels as wound dressing material could be facilitated by the fact that both PEG and heparin have been approved and applied clinically for a long time in different settings (15, 6468). The reported approach does not include the administration of any bioactive molecules, such as growth factors. As a next step toward clinical translation, it will be important to evaluate the reported hydrogel system in an animal model of delayed wound healing that more closely resembles the physiology of human skin (69, 70). Moreover, a good manufacturing practice–grade process of the hydrogel-based wound dressings will have to be established as a prerequisite for clinical trials in humans. Finally, because heparin is known to effectively complex various signal molecules (71), the introduced material may similarly aid therapeutic strategies to control cell migration–associated processes in other inflammatory diseases (72, 73).


Study design

This study was designed to test the chemokine sequestering properties of GAG-based hydrogel materials in the context of wound healing. The sulfation pattern and mechanical properties of hydrogel networks were studied using elementary analysis and rheometry. The scavenging characteristics of the materials were compared in binding assays using recombinant human chemokines MCP-1 and IL-8, conditioned medium derived from activated dermal fibrobasts, and inflammatory macrophages, which contain a mix of cell-derived inflammatory cytokines, as well as wound exudates of human patients suffering from chronic venous leg ulcers. Transmigration assays and a murine model of full-thickness excisional wounds were performed to investigate the consequences of hydrogel-based chemokine sequestration in complex conditions of wound healing, in the presence of different, continuously produced mediators. A model of delayed cutaneous wound healing (genetically diabetic db/db mice) was applied to evaluate the overall proregenerative effect of the starPEG-GAG wound dressings.

Preparation of starPEG-GAG hydrogels

StarPEG-GAG hydrogels were prepared as 300-μm-thick planar layers coupled to a glass substrate. Synthesis of the desulfated heparin derivatives and assembly of the starPEG-GAG hydrogels and their modification with the RGD peptide were performed and analyzed as previously described (20, 21, 25, 74). Briefly, heparin (MW, 14,000; Merck), EDC (Sigma-Aldrich), N-hydroxysulfosuccinimide (sulfo-NHS; Sigma-Aldrich), and amine end-functionalized four-arm starPEG (MW, 10,000; Jenchem) were dissolved in deionized, decarbonized water (MilliQ water) on ice. After mixing heparin with EDC and sulfo-NHS (2:1 ratio of EDC/sulfo-NHS), the solution was incubated on ice for 15 min for heparin and mixed with starPEG. The physical characteristics of the different hydrogels were determined by rheometry (ARES LN2, TA Instruments) (20). The following equation was used to estimate the mesh size (ξ) of the gels from the experimentally determined storage modulus (G′ = 3.0 ± 1.1 kPa) based on the rubber elasticity theory (75):Embedded Imagewhere G′ is the storage modulus, NA is the Avogadro constant, R is the molar gas constant, and T is the temperature (21, 75). The theory assumes a purely elastic hydrogel where all chains contribute to the retraction force after small deformation in a similar way (affine deformation), neglects end effects of single chains (all chains have fixed ends toward an elastic background), and excludes any influence of physical entanglements. With this approach, a mesh size of 11.0 ± 1.5 nm was calculated.

For the in vitro experiments, the liquid gel mixture was covalently attached onto aminosilanized glass coverslips and covered with a hydrophobic glass slide. For the in vivo experiments, hydrogels were formed as gel discs using an in situ assembling Michael-type addition chemistry as previously described (22). Briefly, thiol end-functionalized starPEG hydrogels were mixed with maleimide-prefunctionalized heparin, allowing for rapid gelation in the presence of body fluids. The hydrogels were formed as 5-mm discs with a height of about 0.5 mm. Therefore, the hydrogel mixture was polymerized between two hydrophobic glass coverslips. After polymerization, the hydrogels were washed and stored in phosphate-buffered saline (PBS) supplemented with penicillin (100 U/ml), streptomycin (100 μg/ml) (Biochrom), and amphotericin B (250 μg/ml) (Sigma-Aldrich) for 24 hours. Thereafter, hydrogels were washed with PBS without antibiotic. Material characteristics (storage modulus, mesh size, and sulfate content) were determined as previously described (20, 21, 25).

Docking analysis of chemokines with heparin

The docking of heparin to MCP-1 and IL-8 was tested using ClusPro 2.0 (7679) based on previously published protein structures by Baldwin et al. (80) and Lubkowski et al. (81). The docking model with the highest cluster size was rendered using QuteMol (82). Protein characteristics were calculated with PyMOL Molecular Graphics System (Schrödinger) and ExPASy ProtParam (83).

Chemokine binding to starPEG-GAG hydrogels

Surface-bound starPEG-GAG hydrogels (n = 3) were placed in custom-made incubation chambers that allowed only minimal interaction of the protein solution with area not originating from the hydrogel. Human MCP-1 (eBioscience) or IL-8 (72 amino acids; PeproTech) was incubated with the hydrogels at room temperature at a concentration of 100 μg/ml in RPMI supplemented with 2% (v/v) fetal calf serum (FCS) (Life Technologies GmbH). Samples of the supernatant were taken at intervals of 0, 2, 4, 6, 12, and 24 hours, directly frozen in liquid nitrogen, and stored at −80°C until quantification by ELISA. The immobilization capacity was assessed by analyzing the chemokine solution before and after incubation on the hydrogels for their chemokine concentration by ELISA. Additionally, cytokine binding assays were performed with conditioned medium derived from fibroblasts and macrophages or human wound fluids from patients with chronic wounds. Conditioned medium and human wound fluids were transferred onto the hydrogels and incubated at 37°C and 5% (v/v) CO2 in a humidified atmosphere. After 24 hours, supernatants were harvested from the hydrogels, and cytokine concentrations were determined by ELISA or multiplex immunoassay (see quantification of protein expression in the Supplementary Materials).

Preparation of human cells

Human peripheral blood was taken from healthy volunteers after approval of the local ethics committee in compliance with the Declaration of Helsinki (ethic vote no. 064-2009). Monocytes, inflammatory macrophages, and PMNs as well as primary human dermal fibroblasts from healthy breast skin were generated as previously described (8486).

Generation of conditioned medium

Conditioned medium was derived from inflammatory activated macrophages or dermal fibroblasts. Therefore, macrophages were stimulated with lipopolysaccharide (100 ng/ml) (Sigma-Aldrich) in RPMI 1640 (Biochrom) containing penicillin (100 U/ml) and streptomycin (100 μg/ml) (Biochrom) supplemented with 2% FCS (R2 medium). Dermal fibroblasts were seeded at 5 × 104 in R10 medium (10% FCS). After overnight culture, dermal fibroblasts were washed twice and stimulated with TNF (5 ng/ml) and IL-1β (2.5 ng/ml) (Miltenyi Biotec) in R2 medium. After 24 hours, cell-free conditioned medium was collected and stored at −20°C until further use.

Transmigration assay

Transmigration assays were performed with freshly purified PMNs and monocytes. Cells were resuspended in R0 medium and placed (2.5 × 105 per insert) to the upper compartment of Transwell chambers (pore size, 3 μm; ThinCert, Greiner Bio-One). Conditioned medium from dermal fibroblasts or controls as indicated was added to the lower well. PMNs and monocytes were allowed to migrate at 37°C and 5% (v/v) CO2 in a humidified atmosphere for 30 min and 2 hours, respectively. Migrated cells in the lower wells were quantified by counting the cells microscopically.

Human wound fluids

Samples of wound fluids from chronic venous leg ulcer patients (n = 6) were provided by S. Eming (University of Cologne). Wound fluids were generated as previously described (47), covering the wounds with a semipermeable polyurethane film (Hydrofilm) for a maximum of 8 hours, followed by the collection of fluids, centrifugation, and freezing. Bacterial contamination of the wound fluids were determined by bacterial enrichment of the wound fluid samples according to a standard procedure for aerobic and anaerobic bacteria. After bouillon turbidity, inoculation of samples on standard bacterial culture plates was carried out, followed by pure culture and matrix-assisted laser desorption/ionization–time-of-flight mass spectrometry (confidence interval, 99.9%).

Murine wound healing studies

Animal experiments were performed in accordance with institutional and state guidelines and approved by the Committee on Animal Welfare of Saxony (TVV 24/12). C57BL/6 or db/db mice (10 to 12 weeks old) were anesthetized, and full-thickness wounds (including panniculus carnosus) were inflicted with 6-mm dermal biopsy punches on both sites of the shaved back as previously described (87). Different hydrogel discs (5 mm for wt mice or 12 mm for db/db mice) or Promogran (Systagenix), which were prewetted with PBS to allow optimal performance in the typically dry wound bed of mice, were applied to the wounds and covered with Mepitel (Mölnlycke Health Care) and Raucodrape (Lohmann & Rauscher) to keep the materials in place and moisturized. Five (wt, db/db) or 10 (db/db) days after wounding, mice were sacrificed, and wounds were photographed before and after hydrogels were recovered from the wounds. Hydrogels were quick-frozen in liquid nitrogen and stored at −80°C until quantification for bound wound-derived mediators. Wounds were then excised and deep-frozen in tissue freezing medium (Leica) for sectioning and immunofluorescence staining to study tissue expression of CD31, αSMA, panCK, and CD11b. Wounds from mice were further prepared for tissue lysis to isolate RNA for gene expression analysis (see table S3) or to isolate proteins for quantification by ELISA or multiplex immunoassay or to prepare single-cell suspension for immediate quantification of PMN and monocyte infiltration by flow cytometry (fig. S4).

Detection of wound-derived mediators in hydrogels

Hydrogels recovered from wounds 5 or 10 days after wounding were immersed in PBS (Biochrom) and mechanically disrupted for 30 s at 30 Hz using a 7-mm steel bead and a bead mill (TissueLyser LT, Qiagen). For mediator displacement, 1% Triton was added for 5 min before centrifugation. Concentrations of recovered cytokines were determined in the supernatant by ELISA or multiplex immunoassay.

Statistical analysis

All experiments with primary cells were performed with at least three or more different donors, as indicated in the figure legends. In all experiments, at least three specimens from each hydrogel were tested. Data are presented as means ± SD. Statistical analyses were performed using one-way ANOVA with multiple comparisons versus indicated control using Bonferroni t test or, if data failed normality test, using Dunnett’s method or using unpaired t test or, if data failed normality test, by Mann-Whitney rank sum test. Paired t test was used for statistical analysis of results of human wound fluids. Any P values <0.05 were considered to be statistically significant. Individual level data are included in table S4.


Supplemental experimental procedures

Fig. S1. Characterization of conditioned medium from activated primary human inflammatory macrophages and dermal fibroblasts.

Fig. S2. Transmigration assays with monocytes and PMNs using conditioned medium from activated dermal fibroblasts.

Fig. S3. Characterization of scavenging effects of starPEG-GAG hydrogels with recombinant human MCP-1 and IL-8.

Fig. S4. Gating strategy used to identify PMNs and monocytes present in wound tissue.

Fig. S5. Histological characterization of starPEG-GAG hydrogels applied onto wounds.

Fig. S6. Modulation of cytokine protein concentrations in wounds by starPEG-GAG hydrogels.

Fig. S7. Scavenging and pro-wound healing effects of starPEG-GAG hydrogels in wounds infected with S. aureus.

Fig. S8. Characterization of scavenging effects of starPEG-GAG hydrogels in human wound fluids.

Table S1. Characterization of scavenging effects of starPEG-GAG hydrogels in human wound fluids.

Table S2. Chemokine concentration and infection status of patient samples.

Table S3. Primer sequences and annealing temperatures (Tanneal) used for quantitative polymerase chain reaction.

Table S4. Individual level data corresponding to the different figures.


  1. Acknowledgments: We gratefully acknowledge A. Rodloff and S. Wendt (Leipzig University) for intensive support in microbiological analytics. We thank G. Duda (Charité-Universitätsmedizin Berlin) for constructive discussion of the manuscript as well as suggesting the analysis of human wound exudate samples and S. Eming (University of Cologne) for providing us with human wound exudate samples. We thank N. Rein (Leibniz Institute of Polymer Research Dresden) for vital help with the material preparation. We also thank I. Forstreuter (Leipzig University) for excellent assistance in the animal studies. Funding: This work was funded by the German Research Council (DFG SFB-TR67 projects A10, B3, and FR2671/4-1). Author contributions: N.L., L.S., P.A., E.W., R.A.F., and S.F. contributed to the collection of experimental data. N.L., L.S., S.F., and J.C.S analyzed the data. N.L., L.S., J.C.S., C.W., S.F., and U.F. contributed to writing the paper. U.F. and S.F. supervised the research. Competing interests: C.W. and U.F. are inventors on the patent WO2010060485A1 “Bioactive hydrogel” that covers the starPEG-GAG hydrogels used in the study. All other authors declare that they have no competing interests. Data and materials availability: Requests for materials or further information on the material preparation should be addressed to U.F. (freudenberg{at}
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article