Diabetic wound regeneration using peptide-modifiedhydrogels to target re-epithelializationYun Xiaoa,b, Lewis A. Reisb, Nicole Fericb, Erica J. Kneeb, Junhao Gua, Shuwen Caoa, Carol Laschingerb, Camila Londonob,Julia Antolovichc,d, Alison P. McGuigana,b, and Milica Radisica,b,e,1

aDepartment of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, ON M5S 3E5, Canada; bInstitute of Biomaterials andBiomedical Engineering, University of Toronto, Toronto, ON M5S 3G9, Canada; cDepartment of Pharmacology and Toxicology, University of Toronto,Toronto, ON M5S 1A8, Canada; dDepartment of Biochemistry, University of Toronto, Toronto, ON M5S 1A8, Canada; and eToronto General ResearchInstitute, University Health Network, Toronto, ON M5G 2C4, Canada

Edited by Robert Langer, Massachusetts Institute of Technology, Cambridge, MA, and approved August 19, 2016 (received for review July 27, 2016)

There is a clinical need for new, more effective treatments for chronicwounds in diabetic patients. Lack of epithelial cell migration is ahallmark of nonhealing wounds, and diabetes often involves endo-thelial dysfunction. Therefore, targeting re-epithelialization, whichmainly involves keratinocytes, may improve therapeutic outcomes ofcurrent treatments. In this study, we present an integrin-bindingprosurvival peptide derived from angiopoietin-1, QHREDGS (gluta-mine-histidine-arginine-glutamic acid-aspartic acid-glycine-serine), as atherapeutic candidate for diabetic wound treatments by demonstrat-ing its efficacy in promoting the attachment, survival, and collectivemigration of human primary keratinocytes and the activation of pro-tein kinase B Akt and MAPKp42/44. The QHREDGS peptide, both as asoluble supplement and when immobilized in a substrate, protectedkeratinocytes against hydrogen peroxide stress in a dose-dependentmanner. Collective migration of both normal and diabetic human ker-atinocytes was promoted on chitosan–collagen films with the immo-bilized QHREDGS peptide. The clinical relevance was demonstratedfurther by assessing the chitosan–collagen hydrogel with immobilizedQHREDGS in full-thickness excisional wounds in a db/db diabetic mousemodel; QHREDGS showed significantly accelerated and enhancedwound closure compared with a clinically approved collagen wounddressing, peptide-free hydrogel, or blank wound controls. The accelera-tedwound closure resulted primarily from faster re-epithelialization andincreased formation of granulation tissue. There were no observabledifferences in blood vessel density or size within the wound; however,the total number of blood vessels was greater in the peptide-hydrogel–treated wounds. Together, these findings indicate that QHREDGS is apromising candidate for wound-healing interventions that enhance re-epithelialization and the formation of granulation tissue.

diabetic wound healing | peptide | hydrogel | QHREDGS |re-epithelialization

Chronic ulcers are considered a major healthcare challengebecause they affect 6.5 million people in the United States

(1). Nonhealing wounds, including chronic ulcers, can be causedby a number of common diseases and medications, such as vas-cular insufficiency, diabetes mellitus, and local-pressure effects,which disrupt the well-orchestrated cellular and molecular in-teractions during the wound-healing process (2). Specifically,diabetic foot ulcers affect 15% of people with diabetes and are aleading cause of amputation (3). The mechanism underlyingdiabetic chronic wounds remains elusive, and new interventionsfor diabetes-impaired wound healing are needed. After almosttwo decades without new chemical entities approved by the Foodand Drug Administration (FDA) (Regranex was approved in1997), it has been recognized that an optimal wound-healingoutcome requires a multifaceted approach that simultaneouslyaddresses various issues (e.g., persistent inflammation, insufficientangiogenesis, and impaired re-epithelialization) (2).Keratinocytes are the major cell type in the epidermis, the out-

ermost layer of skin. Upon injury, keratinocytes migrate from thewound edge into the wound to re-epithelialize the damaged tissue

and restore the epidermal barrier. The hallmark of nonhealing hu-man wounds is nonmigratory and hyperproliferative keratinocytes,resulting in epidermis thickening at the wound edge and an absenceof wound closure (4). Scarless embryonic wound healing and com-plete healing in animals with a high regenerative potential such asnewts critically depend on rapid re-epithelialization (5, 6).Additionally, nonhealing diabetic wounds are trapped in a state

of prolonged inflammation characterized by supraphysiologicaloxidative stress that can induce keratinocyte injury, dysfunction, andapoptosis (7–9). The supraphysiological oxidative stress results fromthe excess production of reactive oxygen species (ROS) by macro-phages and neutrophils, coupled with an impaired antioxidant de-fense capability in response to hyperglycemia (7). Moreover, alteredextracellular matrix (ECM) composition in nonhealing wounds andan enhanced ECM degradation rate caused by elevated matrixmetalloproteinase levels can impair keratinocyte attachment, lead-ing to aberrant cell signaling and impaired migration (10–12).To address these challenges, we sought to develop a wound-

healing approach that could recapitulate key aspects of scarlessembryonic wound healing by (i) promoting effective keratinocytemigration, (ii) protecting the wound-bed cells against oxidativestress, and (iii) providing a new matrix for cell attachment.Our group has recently described an angiopoietin-1–derived

peptide, QHREDGS (glutamine-histidine-arginine-glutamic acid-aspartic acid-glycine-serine), which interacts with integrins, receptorsthat function in cell adhesion and ECM binding. The QHREDGS

Significance

Current treatments for diabetic chronic wounds fail to achieve ef-fective therapeutic outcomes. The majority of these treatmentsfocus on angiogenesis, but diabetes often involves endothelialdysfunction. A hallmark of regenerative wound healing is rapid,effective re-epithelialization. In this study, we present QHREDGS(glutamine-histidine-arginine-glutamic acid-aspartic acid-glycine-serine), a prosurvival peptide derived from angiopoietin-1, as atherapeutic candidate that targets re-epithelialization. ImmobilizedQHREDGS peptide promoted cell survival against hydrogen perox-ide stress and collective cell migration of both normal and diabetichuman keratinocytes in vitro. The clinical relevance was demon-strated further in type 2 diabeticmice: A single treatmentwith a lowQHREDGS dose immobilized in chitosan–collagen was effective inpromoting wound healing, and a single high-dose peptide treat-ment outperformed a clinically approved porous collagen dressing.

Author contributions: Y.X., A.P.M., and M.R. designed research; Y.X., L.A.R., and C. Londonoperformed research; M.R. supervised research; A.P.M. advised on experiments; Y.X., J.G., S.C.,and C. Laschinger contributed new reagents/analytic tools; Y.X., L.A.R., N.F., E.J.K., J.G., S.C.,C. Laschinger, and J.A. analyzed data; and Y.X., L.A.R., N.F., and M.R. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: m.radisic@utoronto.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1612277113/-/DCSupplemental.

E5792–E5801 | PNAS | Published online September 19, 2016 www.pnas.org/cgi/doi/10.1073/pnas.1612277113

peptide was shown to enhance endothelial cell metabolism, tube-formation kinetics, and survival in response to apoptotic stimuli (13).QHREDGS also was shown to promote neonatal rat cardiomyocyteattachment and survival (14), to inhibit human induced pluripotentstem cell (hiPSC) apoptosis during cell expansion (15), to induceosteoblast matrix deposition and mineralization (16), and to havecardiac-protective effects in a chitosan–collagen hydrogel both invitro and in vivo (17, 18).We therefore hypothesized that the QHREDGS peptide could

promote keratinocyte survival and migration and thereby accel-erate diabetic wound healing. We investigated the effect of theQHREDGS peptide as a soluble supplement on the survival ofnormal human keratinocytes upon oxidative stress. The effect onattachment, survival upon oxidative stress, and collective migra-tion of both normal and diabetic keratinocytes was assessed byimmobilizing the QHREDGS peptide within a chitosan–collagenfilm coating. We further investigated the ability of the QHREDGSpeptide to promote diabetic wound repair in vivo using a full-thickness excision wound model in db/db diabetic mice.

ResultsQHREDGS Peptide Prevents H2O2-Induced Cell Death in Human PrimaryKeratinocytes via Akt and MAPKp42/44 Signaling. To evaluate the ef-fect of the QHREDGS peptide on keratinocytes, we first focusedon normal neonatal human epidermal keratinocytes (HEKs)cultured with the soluble QHREDGS peptide at doses previouslyreported to be effective for endothelial cell survival (low: 100 μM;high: 650 μM) (13). The percentage of proliferating HEKs in thepopulation was not affected by the presence of the QHREDGS

peptide at either concentration as quantified by BrdU incorporation(Fig. 1A). No significant difference in the HEK migration rate wasobserved with the soluble QHREDGS peptide at either concen-tration (SI Appendix, Fig. S1).To investigate the effect of the soluble QHREDGS peptide on

keratinocyte survival under oxidative stress, we preconditionedHEKs by incubating the cells with or without the QHREDGSpeptide and then exposed the HEKs to 500 μM H2O2 for 2 h(Fig. 1B). An endpoint cell integrity assay showed a significantdose-dependent increase in the percentage of viable HEKs in thepresence of supplemented QHREDGS (Fig. 1C).Given that the full-length protein from which the QHREDGS

peptide was derived, angiopoietin-1, is known to protect skin cellsfrom oxidative damage and to increase the activation of the pro-survival Akt and MAPKp42/44 pathway (19), we investigated whetherimproved survival upon H2O2 stress in the presence of theQHREDGS peptide was associated with the up-regulation of Aktand MAPKp42/44 phosphorylation. HEKs treated with 500 μM H2O2showed transient phosphorylation of both Akt (Fig. 1D) andMAPKp42/44 (Fig. 1E) at 15 min by Western blot analysis. Indeed, thepresence of the soluble QHREDGS peptide during preconditioningand H2O2 treatment increased the phosphorylation of Akt andMAPKp42/44, and the increase was dose-dependent (Fig. 1 D and E).

Immobilized QHREDGS Peptide Promotes HEK Attachment, Survival,and Migration in Vitro. Although we observed increased sur-vival without excessive proliferation in the presence of solubleQHREDGS, we did not observe enhanced keratinocyte migra-tion (SI Appendix, Fig. S1). This absence of accelerated migration

Fig. 1. Soluble QHREDGS peptide prevents H2O2-induced cell death in human primary keratinocyteswith up-regulation of Akt and MAPK phosphoryla-tion. (A) Kct-positive HEKs cultured in the presenceor absence of different concentrations of solubleQHREDGS peptide did not incorporate significantlydifferent amounts of BrdU, indicating similar pro-liferation rates in all three conditions (low peptide:100 μM; high peptide: 650 μM). (Scale bars: 50 μm.)n = 3 or 4. Ctrl, control. (B) H2O2 treatment regimenfor HEKs in the presence or absence of QHREDGSpeptide. HEKs were allowed to attach for 2 h and wereserum-starved overnight. For the peptide groups, HEKswere preconditioned with QHREDGS peptide for 2 hand then were treated with 500 μM H2O2 in thepresence or absence of the peptide. (C) HEK survivalafter H2O2 treatment was determined by the EarlyToxCell Integrity assay (Molecular Devices). The QHREDGSpeptide protected HEKs against H2O2-induced celldeath in a dose-dependent manner. (Scale bars:200 μm.) One representative experiment is shown;three independent experiments were performed;each experiment had nine replicates of each condi-tion. (D and E) Immunoblotting with phosphorylatedAkt or MAPKp42/44 and Akt or MAPKp42/44 antibodiesshowed transient activation of Akt and MAPKp42/44

pathways signaling under H2O2 stress. GAPDH wasused as a loading control. The presence of the QHREDGSpeptide in the culture medium up-regulated Akt andMAPKp42/44 phosphorylation. Four independent ex-periments were performed; each experiment had twoor three independent replicates of each condition.Data are presented as mean ± SD; *P < 0.05.

Xiao et al. PNAS | Published online September 19, 2016 | E5793

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

motivated our further optimization of the method by whichwe presented the peptide to the cells. Given that the QHREDGSpeptide is reported to function primarily through integrin interac-tions (13, 15), and an ever-growing body of literature shows thatthe efficacy of integrin ligands is increased when immobilized to amatrix (20, 21), we covalently immobilized the QHREDGS peptideto a chitosan–collagen hydrogel. Chitosan and collagen interactthrough a combination of thermal and ionic mechanisms, stabilizedby polyanion (collagen) and polycation (chitosan) electrostatic in-teractions (22). Conjugation of the QHREDGS peptide to chitosanwas achieved using previously described methods (23), and chitosan–collagen films with or without immobilized QHREDGS peptide werecast in the wells of 24-well or 96-well plates. Quantification usingfluorescently labeled peptide, FITC-QHREDGS, demonstratedeffective immobilization in both low (4.7 ± 0.1 nmol/cm2) and high(13.8 ± 1.4 nmol/cm2) peptide concentrations and the absence of thepeptide in the control condition (Fig. 2A). Normalized to the mass ofchitosan in the films, the amount of immobilized QHREDGS pep-tide was 14.9 ± 0.3 nmol/mg in the low peptide concentration and44.1 ± 4.6 nmol/mg in the high peptide concentration.There was no significant difference in the attachment of HEKs

to the various chitosan–collagen films (Fig. 2B). However, inchitosan-only films, in which adhesion was poor, the QHREDGS

peptide clearly promoted HEK attachment in a dose-dependentmanner (SI Appendix, Fig. S2). This result indicates that, althoughthe QHREDGS peptide can promote HEK attachment, thepresence of collagen adhesion sites in the setting of the chitosan–collagen film masks this effect. Furthermore, Western blot anal-ysis showed the increased activation of Akt and MAPKp42/44during 2-h attachment on chitosan–collagen films in the presenceof immobilized QHREDGS peptide (Fig. 2C).We then investigated the effect of the immobilized QHREDGS

peptide on HEK survival following treatment with 500 μM H2O2.HEKs were allowed to attach to the chitosan–collagen films for 4 hand then were treated with H2O2 for 2 h (Fig. 2D). Subsequent cell-integrity assessment showed an increased percentage of viable HEKsin the presence of the immobilized QHREDGS peptide (Fig. 2E).HEKs treated with 500 μMH2O2 and nontreated controls were alsocompared using Western blot analysis, which showed that thephosphorylation of MAPKp42/44 was increased in the presence ofimmobilized QHREDGS peptide relative to the control (Fig. 2F).Keratinocyte migration is essential for wound healing, because

a wound cannot heal in the absence of re-epithelialization (24).We therefore assessed the effect of the immobilized QHREDGSpeptide on HEK migration in 2D monolayers using an Ibidimigration assay system. Importantly, the Ca2+ concentration in the

Fig. 2. Immobilized QHREDGS peptide in chitosan–collagen films promotes the survival and migrationof human neonatal primary keratinocytes. (A) Quan-tification of the amount of QHREDGS peptide immo-bilized within chitosan–collagen films. n = 3. (B) HEKattachment on the chitosan–collagen films in thepresence or absence of conjugated QHREDGS peptide.Image analysis showed no difference in the number ofattached HEKs (stained with DAPI) among the threegroups. (Scale bars: 200 μm.) Three independent ex-periments were performed; each experiment hadthree independent replicates of each condition. (C)Immunoblotting with anti-phosphorylated Akt orMAPKp42/44 and anti-Akt or MAPKp42/44 showed up-regulation of MAPKp42/44 and Akt activation duringHEK attachment. GAPDH was used as a loadingcontrol. Three independent experiments were per-formed; each experiment had two independentreplicates of each condition. (D) Experimental time-line. Cells were harvested for attachment Westernblotting 2 h after seeding. H2O2 was applied4 h afterseeding, and the cells were harvested 15 min later forWestern blotting. Live/dead staining was performedafter 2 h of H2O2 treatment. (E) HEK survival afterH2O2 treatment was determined by the EarlyTox CellIntegrity assay. QHREDGS peptide in the chitosan–collagen film protected HEKs against H2O2-inducedcell death in a dose-dependent manner. One repre-sentative experiment is shown; three independent ex-periments were performed; each experiment had fourindependent replicates of each condition. (F) Immuno-blotting showed up-regulation of the MAPKp42/44 andAkt phosphorylation in HEKs under H2O2 stress at15 min. GAPDH was used as a loading control. Threeindependent experiments were performed; each ex-periment had two independent replicates of each con-dition. (G) Representative examples of HEK-woundingexperiments on chitosan–collagen films in the presenceor absence of conjugated QHREDGS peptide. (Scalebars: 200 μm.) HEK migration on the QHREDGS-immo-bilized films was accelerated compared with HEKmigration on the control peptide-free films. One rep-resentative experiment is shown; three independentexperiments were performed; each experiment had sixindependent replicates of each condition. Data arepresented as mean ± SD; *P < 0.05.

E5794 | www.pnas.org/cgi/doi/10.1073/pnas.1612277113 Xiao et al.

culture medium was increased from 0.06 mM to 0.12 mM uponinitiation of the migration assay to ensure collective HEK migra-tion (essential for wound healing), as demonstrated by the for-mation of E-cadherin–mediated cell–cell junctions (SI Appendix,Fig. S3A). The presence of the immobilized QHREDGS peptideaccelerated collective HEK migration in a dose-dependent man-ner (Fig. 2G). The accelerated migration was not caused by in-creased proliferation, because there was no difference in celldensity among the three groups as characterized at the migrationendpoint (SI Appendix, Fig. S3B).

Immobilized QHREDGS Peptide Promotes Diabetic Human PrimaryKeratinocytes’ Attachment, Survival, and Migration in Vitro. In di-abetic chronic wounds, keratinocytes experience hyperglycemia andsupraphysiological oxidative stress, which challenge keratinocyte’sproliferation and survival (7, 25). Therefore, we examined the ef-fect of the immobilized QHREDGS peptide on adult diabetichuman epidermal keratinocytes (DHEKs) by seeding DHEKs ontochitosan–collagen films in the presence or absence of immobilizedQHREDGS peptide. Similar to the results with normal HEK cells,we found the presence of the immobilized QHREDGS peptidepromoted DHEK attachment to chitosan-only films (SI Appendix,

Fig. S4) but did not affect DHEK attachment to chitosan–collagenfilms (Fig. 3A).Western blot analysis showed that the activation ofMAPK and Akt was increased in DHEKs in the presence of theQHREDGS peptide during a 2-h attachment (Fig. 3B).Because of prolonged inflammation (25), human chronic wounds

experience three to four times higher oxidative stress (26) and ox-idative damage (27) than acute wounds. To mimic this scenario, weinvestigated the effect of immobilized QHREDGS peptide onDHEK survival following a 2-h treatment with 2 mM H2O2 (anexposure four times higher than used for the HEK survival assay)after DHEKs were allowed to attach to the chitosan–collagenfilms for 4 h (Fig. 3C). Cell-integrity assessment showed thatDHEK survival under H2O2 stress was improved in the presence ofimmobilized QHREDGS peptide (Fig. 3D), despite the higherH2O2 concentration used. DHEKs treated with 2 mM H2O2 andnontreated controls were also compared by Western blot analysis,which showed that the phosphorylation of Akt and MAPKp42/44upon H2O2 treatment was increased in the presence of immobilizedQHREDGS peptide in a dose-dependent manner (Fig. 3E).We then assessed the effect of immobilized QHREDGS peptide

on DHEK migration using the Ibidi migration assay. Importantly,the Ca2+ concentration in keratinocyte growth medium (KGM),

Fig. 3. Immobilized QHREDGS peptide in chitosan–collagen films promotes the survival and migrationof adult DHEKs. (A) DHEK attachment on the chitosan–collagen films in the presence or absence of immobi-lized QHREDGS peptide. Image analysis showed nodifference in the number of attached DHEKs (stainedwith TO-PRO) among the three groups. (Scale bars:200 μm.) Three independent experiments were per-formed; each experiment had three independentreplicates of each condition. (B) Immunoblottingshowed up-regulation of MAPKp42/44 and Akt activa-tion during DHEK attachment. GAPDH was used toensure even loading. Three independent experimentswere performed; each experiment had two in-dependent replicates of each condition. (C) Experi-mental timeline. Cells were harvested for attachmentWestern blotting 2 h after seeding. H2O2 was applied4 h after seeding, and the cells were harvested 15 minlater for Western blotting. Live/dead staining wasperformed after 2 h of H2O2 treatment. (D) DHEKsurvival following H2O2 treatment was determined bythe EarlyTox Cell Integrity assay. The QHREDGS pep-tide in the chitosan–collagen film protected DHEKsagainst H2O2-induced cell death in a dose-dependentmanner. Four independent experiments were per-formed; each experiment had at least six independentreplicates of each condition. (E) Representative im-munoblots of phosphorylated and total MAPKp42/44 orAkt. Quantification of immunoblotting revealed up-regulation of the Akt and MAPKp42/44 activation ofDHEKs under H2O2 stress. GAPDH was used as aloading control. Three independent experiments wereperformed; each experiment had two independentreplicates of each condition. (F) Representative ex-amples of DHEK wounding experiments on chitosan–collagen films with or without immobilized QHREDGSpeptide. (Scale bars: 200 μm.) DHEK migration on thefilms with QHREDGS peptide was accelerated comparedwith the peptide-free control. Three independent ex-periments were performed; each experiment had atleast four independent replicates of each condition.Data are presented as mean ± SD; *P < 0.05.

Xiao et al. PNAS | Published online September 19, 2016 | E5795

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

0.1 mM, was sufficient to ensure collective DHEK migration, asdemonstrated by the formation of E-cadherin–mediated cell–cell junctions, without additionally elevating the Ca2+ concen-tration (SI Appendix, Fig. S5A). The presence of the immobilizedQHREDGS peptide also accelerated DHEK collective migration(Fig. 3F), and this acceleration was not caused by differences incell density as characterized at the migration end point (SI Ap-pendix, Fig. S5B).

QHREDGS-Immobilized Hydrogel Promotes Wound Healing in db/dbDiabetic Mice. We investigated whether the QHREDGS peptideimmobilized to the chitosan–collagen hydrogel could acceleratewound healing in diabetic mice. This hydrogel system was chosenas a delivery vehicle because of its rapid gelation under physio-logical conditions and its persistence for a period of 3 wk in vivo(18). In vivo biocompatibility had been demonstrated in previousmyocardial infarction model studies (18). Therefore, only oneapplication of the hydrogel onto the wounds was needed for the2-wk study. A full-thickness excision wound (Fig. 4A) was createdon 8-wk-old male BKS.Cg-Dock7m +/+ Leprdb/J mice (db/db). Thismodel was selected because the animal is leptin-receptor defi-cient and represents a type II diabetes model characterized byhyperglycemia, obesity, hyperinsulinemia, and impaired wound

healing. Moreover, this strain heals wounds primarily by the for-mation of granulation tissue rather than by contraction (28).Quantification using fluorescently labeled peptide demonstrated

that in reconstituted chitosan solutions the amount of conjugatedQHREDGS peptide was 17.5 ± 2.2 nmol/mg chitosan in low-concentration conditions and 41.5 ± 1.4 nmol/mg chitosan in high-concentration conditions. In the final chitosan–collagen hydrogelthese concentrations were converted to a peptide concentration of43.8 ± 4.4 μM in the low-concentration conditions and 103.8 ±3.5 μM in the high-concentration conditions. With a view to futureclinical translation, we first tested the low-concentration conditionin the in vivo studies to minimize the amount of peptide appliedto the wound. A single application of 50 μL low-concentrationchitosan–collagen hydrogel (2.2 nmol immobilized QHREDGSpeptide; low-peptide treatment) was applied to the wound. Thechitosan–collagen hydrogel alone without the peptide (controltreatment) and a no hydrogel/no peptide (blank condition) wereused as controls. A secondary dressing, Tegaderm film, was appliedon top of the wound, with or without the hydrogel, to maintain amoist environment. As shown in Fig. 4B, the presence of immo-bilized QHREDGS in the hydrogel resulted in significantly smallerwounds on day 14 than in the controls. Image analysis of the woundgross morphology performed by an investigator blinded to the

Fig. 4. A low dose of QHREDGS peptide immobi-lized to chitosan–collagen hydrogel is sufficient topromote wound healing in db/db diabetic mice.(A) Representative image of the 8-mm full-thicknessdorsal wounds on db/db diabetic mice. (B) Repre-sentative gross images of the initial wounds on day0 (D0) and the wounds at 14 d (D14) after treatmentwith no hydrogel (Blank), with peptide-free chitosan–collagen hydrogel (Ctrl), or with a low dose ofQHREDGS peptide conjugated to chitosan–collagenhydrogel (Low peptide). Quantification of the woundsize as a percentage of the original wound arearevealed faster wound closure in the peptide-treated mice at days 8–14. n = 4. (C) Representativeimages of Trichrome-stained tissue sections of woundstreated with no hydrogel, with peptide-free chitosan–collagen hydrogel, or with a low dose of QHREDGSpeptide conjugated to chitosan–collagen hydrogel onday 14. Black arrowheads indicate wound edges; redarrowheads indicate the tips of the healing epithelialtongue. (Scale bar: 3 mm.) (Insets) The tips of the healingepithelial tongue were confirmed by pan-keratin stain-ing. [Low and High magnification images from slidescanned image at 20x using the Aperio ScanScope XT(Aperio Technologies, USA)]. (Scale bar: 50 μm.) (D)Quantification of wound size from histological samplescollected 14 d after treatment. (i) Image analysis showedno significant difference in the wound-edge distanceamong the three groups. (ii) The low-peptide treatmentsignificantly reduced the size of epithelial gap, in-dicating accelerated wound closure compared with theblank and control groups. (iii) The low-peptide treat-ment significantly increased the re-epithelializationpercentage at the end of experiment compared withthe blank and control groups. (iv) The low-peptidetreatment significantly increased the size of the granu-lation tissue compared with the blank and controlgroups. (v) Average thickness of the epidermis within300 μm of the leading edge of the wound. The epi-dermal thickness in the low-peptide group was lowerthan in the blank and control groups. n = 4. Data arepresented as mean ± SD; *P < 0.05.

E5796 | www.pnas.org/cgi/doi/10.1073/pnas.1612277113 Xiao et al.

study groups demonstrated faster wound healing in the low-peptidegroup starting on day 8. Administration of the chitosan–collagenhydrogel without the immobilized QHREDGS peptide (controltreatment) had no significant effect on the wound closure ratecompared with the blank controls.We also examined the wound histology by Masson’s trichrome

staining and confirmed the location of the epithelial tongue usingpan-keratin staining (Fig. 4C). The wound edge was defined as thedistance between the two boundaries of intact skin (thin muscula-ture of the panniculus carnosus).There was no significant differenceamong the three groups in the wound-edge distance, indicating nodifference in wound contraction over 14 d (Fig. 4 D, i). The epi-thelial gap, defined as the distance between the two advancingepithelial tongues (Fig. 4C), was smaller in the low-peptide groupthan in the blank and control groups (Fig. 4 D, ii and SI Appendix,Fig. S6). The re-epithelialization percentage, defined as the ratio ofthe distance that has been re-epithelialized over the wound-edgedistance, was significantly higher in the presence of the QHREDGSpeptide than in the controls (Fig. 4 D, iii and SI Appendix, Fig. S6).The low-peptide group also developed significantly more granula-tion tissue (Fig. 4 D, iv) than did the controls. Moreover, the epi-dermal thickness of the advancing epithelial tongue was significantlysmaller in the peptide group than in the blank and control groups(Fig. 4 D, v). There was no difference in the epithelial thickness ofthe skin remote from the wounds among the three experimentalgroups (SI Appendix, Fig. S7).Immunostaining for CD31 (SI Appendix, Fig. S8 A and C, i–iii)

and smooth muscle actin (SMA) (SI Appendix, Fig. S8 B and C,iv) indicated no significant differences in the blood vessel density(SI Appendix, Figs. S8 C, ii, and S9) and myofibroblast contentamong the groups (SI Appendix, Fig. S8 C, iv), although totalnumber of blood vessels in the wound was significantly higher inthe low-peptide group than in the blank and control groups atday 14 (SI Appendix, SI Results and Fig. S8 C, iii).

High Peptide Hydrogel Outperforms a Clinically Approved WoundDressing in Wound Closure of db/db Diabetic Mice. To demon-strate the full potential of the peptide-modified hydrogels and as abenchmark against an FDA-approved and clinically availabletreatment, a high-peptide hydrogel (high-peptide treatment) wascompared with ColActive collagen dressing (collagen treatment)in vivo. After 21 d, visual inspection indicated essentially closedwounds in animals in the high-peptide group (Fig. 5A). When theanimals were killed, analysis of wound images indicated 100%wound closure in three of the five animals in the high-peptidegroup, in one of five animals in the collagen-treatment group,and in none of the five animals in the blank or the peptide-free hydrogel controls. The advantages of the high-dose peptidehydrogel treatment became apparent at day 8 after treatment (Fig.5B) and persisted throughout the healing period up to day 21, atwhich point the wounds in the high-peptide group were essentiallyclosed. In the linear range of wound closure (Fig. 5B, days 2–16),the fitted line slopes indicated a significantly higher rate of woundclosure in the high-peptide group (8% of wound area/d) than inthe blank (3%/d) and peptide-free hydrogel (4%/d) control groupsand in collagen-treatment (5%/d) groups (SI Appendix, Fig. S10).There was no significant difference in the rate of wound closure inthe peptide-free hydrogel and the ColActive dressing groups; therate of wound closure was significantly faster in the high-peptidehydrogel group than in all other groups (SI Appendix, Fig. S10).Histological analysis (Fig. 5C) indicated that only the animals

in the high-peptide group had a significant difference in thewound-edge distance (Fig. 5 D, i), epithelial gap (Fig. 5 D, ii), re-epithelialization percentage (Fig. 5 D, iii), and epithelial thick-ness (Fig. 5 D, iv) relative to the blank group. In the wounds inthe high-peptide group the epithelial thickness was also com-parable to that of the unwounded epidermis (Fig. 5 D, iv and SIAppendix, Fig. S7B), whereas the other treatment groups had

epithelium that was significantly (more than two times) thickerthan the unwounded epidermis. According to these histologicalmeasures of wound closure and epidermis quality, the high-peptidegroup significantly outperformed both the control group treatedwith peptide-free hydrogel and the collagen treatment group(Fig. 5 D, ii–iv). Notably, in the high-peptide group, one mousewith complete wound closure also exhibited hair regrowth at thewound periphery as seen in gross morphology (SI Appendix, Fig.S11A) and histology (SI Appendix, Fig. S11B) images. Picrosiriusred staining indicated disruption of the native collagen orga-nization in day 21 wounds in blank-, peptide-free hydrogel-, andcollagen-treated mice (SI Appendix, Fig. S12). In contrast, collagenorganization equivalent to the unwounded skin was identified inwounds treated with the high-peptide immobilized hydrogel (SIAppendix, Fig. S12). Staining for the neural marker protein geneproduct 9.5 (PgP9.5) confirmed a low density of positive cells in allgroups, with no appreciable differences among groups (SI Ap-pendix, Fig. S13). This result is consistent with the inability of di-abetic mice to regenerate neurons fully (29–31). There were nosignificant differences in the vascular density among the groups atday 21 (SI Appendix, Fig. S14).To delineate further whether there were differences in blood

vessel densities in the early stages of healing, we compared thetime course of early vessel development in the high-peptide andblank groups. At days 4 and 8 after injury there were no ap-preciable differences in blood vessel density in the two groups (SIAppendix, Fig. S15). We did, however, observe an increase invessel density and in CD31+ area between day 4 and day 8 in thehigh-peptide group but not in the blank control group (SI Ap-pendix, Fig. S15). Blood vessel density and the percentage of areacovered by endothelial cells remained statistically unchangedbetween day 8 (SI Appendix, Fig. S15) and day 21 (SI Appendix,Fig. S14) according to ANOVA analysis. Collectively these re-sults indicate that the treatment with high-dose peptide hydrogelleads to sufficient angiogenesis to drive the rapid formation ofgranulation tissue without the massive vessel overgrowth early onfollowed by a pronounced vessel regression that is often reportedwith treatments that overstimulate early angiogenesis (32–34).

DiscussionAngiogenesis has been a primary target of therapeutic interven-tions in wound healing by local application of growth factors [e.g.,epidermal growth factor (35, 36), fibroblast growth factor (37, 38),vascular endothelial growth factor (39, 40), and angiopoietin (41)]or angiogenic cells (42, 43). However, diabetic patients are oftenreported to have dysfunctional endothelium that cannot respondefficiently to growth-factor stimulation (44, 45). Diabetics alsosuffer from the insufficient recruitment of circulating endothelialprogenitor cells, which are critical for wound repair (46, 47). As aresult, many of these approaches failed to meet the FDA-acceptedprimary efficacy endpoint standard of “complete wound closurewithin 12 weeks for any dose” (48).We chose to take a different approach here, by considering the

hallmarks of true regenerative healing as observed during theclosure of embryonic wounds and scarless healing in model or-ganisms such as flies, zebrafish, and newts. In all these models,rapid, coordinated, and collective migration of epidermal cells iscritical for regenerative healing (5, 6). Hence, we focused ondeveloping a hydrogel treatment that would promote the col-lective migration of keratinocytes and enhanced formation ofgranulation tissue. Different mechanisms have been proposed toexplain the cell–cell interactions in collective cell migration, in-cluding local tractions pulling cooperatively toward unfilledspace (termed “kenotaxis”) (49), mechanical exclusion interac-tions between cells (50), and intercellular adhesion and tension(51). In our in vitro studies, calcium concentration was increasedwhen necessary to ensure that both normal and diabetic kerati-nocytes migrated collectively rather than individually so that

Xiao et al. PNAS | Published online September 19, 2016 | E5797

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

the treatment effect on collective migration could be describedaccurately.Following its discovery (52), there has been a growing recog-

nition that angiopoietin-1 is an important regulator of cellularprocesses including vascular protection (53), cardiac remodeling(54–56), inflammation (57–59), and wound healing (41). Be-cause of the insolubility of angiopoietin-1, its derivatives havebeen developed to promote wound healing by angiogenesis(60–62). Here, we assessed the effect of a water-soluble peptide,QHREDGS, derived from the fibrinogen-like domain of angio-poietin-1 and conserved among species (54), on promoting thesurvival and collective migration of normal and diabetic kerati-nocytes. In this study, supplementation of the culture mediumwith soluble QHREDGS peptide protected human keratinocytesagainst H2O2 stress and induced up-regulation of Akt andMAPKp42/44 phosphorylation. These findings are consistent withprevious reports for angiopoietin-1 in human keratinocytesand melanocytes (19). Here we demonstrate the effect of theQHREDGS peptide on cell migration and cutaneous woundhealing. Our previous studies also reported that a scrambledpeptide (DGQESHR or DQSHGER) does not exert prosurvivaleffects similar to that of the QHREDGS peptide in cardiac cells(14, 63), endothelial cells (13), and induced pluripotent stem

cells (15), motivating the omission of the scrambled peptide inthe studies described here.Importantly, the QHREDGS peptide functions through inter-

actions with β1-containing integrins that are involved in cell–matrix interactions, rather than Tie2 receptors that reside mainlyon endothelial cells (13, 15, 18). Thus the QHREDGS peptide canact on various cell types, including keratinocytes, that express theα3β1-integrin implicated in their enhanced survival and migration(64–66). The integrin interaction also motivated the presentationof the QHREDGS peptide as a matrix-bound ligand. Further-more, the short peptide sequence QHREDGS can be modifiedusing versatile chemical methods and does not require a specificorientation or conformation to function, as does the full-lengthprotein (67), and peptides are less susceptible than full-lengthproteins to degradation caused by proteolysis or hydrolysis duringmodification and after delivery to the native environment (68).Here, we used 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide

(EDC) chemistry to conjugate the QHREDGS peptide to thebackbone of chitosan, which is a zero-length cross-linker that doesnot add any other moieties to the final product; these features areimportant for further clinical translation. Peptide immobiliza-tion also provides the advantages of localized action in the targettissue and a lower total dose than required for application in an

Fig. 5. A high dose of QHREDGS peptide immobi-lized to chitosan–collagen hydrogel outperforms anapproved wound-healing dressing in db/db diabeticmice. (A) Representative gross images of the initialwounds on day 0 (D0) and day 21 (D21) after treat-ment with no hydrogel (Blank), with peptide-freechitosan–collagen hydrogel (Ctrl), with ColActivecollagen dressing (Collagen), or with a high dose ofQHREDGS peptide conjugated to chitosan–collagenhydrogel (High Peptide). (B) Quantification of thewound size as a percentage of the original woundarea revealed a faster wound-closure rate in thepeptide-treated mice. n = 5 per group. (C) Repre-sentative images of Trichrome-stained tissue sectionson day 21. Black arrowheads indicate wound edges;red arrowheads indicate the tips of the healing epi-thelial tongue. [Low magnification images from slidescanned image at 20x using the Aperio ScanScopeXT (Aperio Technologies, USA)]. (Scale bar: 3 mm.)(D) Quantification of wound size from histologicalsamples collected 21 d after treatment. (i) The high-peptide treatment significantly reduced the wound-edge distance comparedwith the blank group. (ii) Thehigh-peptide treatment significantly reduced the sizeof epithelial gap compared with the other threegroups, indicating accelerated wound closure. (iii) Thehigh-peptide treatment significantly increased the re-epithelialization percentage compared with the otherthree groups. (iv) The high-peptide treatment signifi-cantly decreased the epithelial thickness comparedwith the other groups, to a level comparable to un-damaged epithelium. n = 4 or 5 per group. Data arepresented as mean ± SD; *P < 0.05.

E5798 | www.pnas.org/cgi/doi/10.1073/pnas.1612277113 Xiao et al.

encapsulated or soluble form. We confirmed that the prosurvivaleffect of the QHREDGS peptide was preserved after it was con-jugated and immobilized in the chitosan–collagen films and that theimmobilized QHREDGS peptide was able to promote keratinocytecollective migration, whereas the soluble peptide did not. Notably,in the context of the chitosan–collagen film system, the conjugatedQHREDGS peptide is presented in close proximity to collagen,thereby creating an environment reminiscent of the collagen–glycosaminoglycan interaction in the native ECM (69).Using DHEKs from an adult diabetic patient, we determined

that the prosurvival, promigratory effects of immobilizedQHREDGS peptide could be translated to diabetic keratino-cytes. Diabetic chronic wounds often occur in the elderly pop-ulation (the source of the primary DHEKs in this study) becauseof age-associated comorbidities (70), and advanced protein gly-cosylation that can alter the functional properties of importantECM components, such as collagen, has been associated withdiabetes and aging (71). Despite these challenges, the immobi-lized QHREDGS peptide was effective in the in vitro diabeticwound-healing model, suggesting the possibility of in vivo ben-efits for diabetic wound healing.Although our previous in vitro study showed that the QHREDGS

peptide promoted endothelial cell survival, metabolism, and tubeformation mediated by integrin interactions (13), the accelerateddiabetic wound healing shown in the animal studies here was notassociated with the significantly enhanced blood vessel density perarea of granulation tissue. In the peptide-treated group there wasmore granulation tissue and consequently a higher total number ofblood vessels in these wounds, but there was no increase in bloodvessel density relative to the other groups. This finding is consistentwith our previous study in a rat myocardial infarction model, whichshowed an increase in large vessels within the myocardial infarctionborder zone but no difference in the total vascularization (18). An-giogenesis is critically important for wound healing (72). We view thestable and steady growth of blood vessels in our system as an ad-vantage, because previous studies reported leaky vasculature and, insome cases, angioma formation in treatments that overemphasizedangiogenesis (73).Alternatively, the diabetic endothelium is associated with en-

hanced degradation of nitric oxide, an important regulator of in-flammation, angiogenesis, and re-epithelialization, because of thepresence of excessive ROS (74). In a previous study, we showed thatthe QHREDGS peptide could induce enhanced endothelial cellnitric oxide production (13). Hence, it is possible the improvedwound-healing effects induced by the QHREDGS peptide might beattributable in part to nitric oxide production. In future studies, theQHREDGS peptide could be combined with biomaterials (72) and/orpeptides (62) shown to strongly enhance angiogenesis in woundhealing to investigate synergistic effects.Our previous study demonstrated the retention of the chitosan–

collagen hydrogel over 3 wk postinjection in the infarcted ratheart, an extremely dynamic, inflammatory environment (18).In this study, we applied the QHREDGS peptide in the samechitosan–collagen hydrogel as a single treatment and found thateven a low dose was sufficient to accelerate wound healing in aclinically relevant, genetically modified db/db diabetic mousemodel. Moreover, a single application of a high peptide dosesignificantly outperformed a clinically available wound dressing,resulting in complete wound closure at 3 wk. The approvedColActive dressing significantly accelerated wound healing by67% compared with the untreated wounds in the blank group;the high-peptide immobilized hydrogel accelerated healing by167% compared with untreated wounds and by 60% comparedwith the ColActive dressing. The effect of the peptide on the rateof wound closure was dose-dependent, based on a comparison ofdata from days 2–11 from our two independent animal studies.Although we show potential for hair regrowth (SI Appendix, Fig.

S11) and reinnervation (SI Appendix, Fig. S13), it is important to

emphasize that robust and full hair regrowth and reinnervation inthe wounds of diabetic animals is generally not reported (61, 75–85). The inability of these animals to regenerate nerves fully hasalso been recognized (29–31). This failure is in contrast to wound-healing studies in healthy animals, such as burn injury models, thatreported hair regrowth (72) and signs of reinnervation as judgedby the presence of PgP9.5 in the wound (78).The ability to induce an effective wound-healing response with a

single application is an important clinical consideration, because itremoves the need for frequent repeated applications that candisturb the healing process, it reduces the pain and discomfortassociated with frequent dressing changes, and it requires fewerhealthcare resources (70, 86). The use of a single mouse model forthe analysis of therapeutic efficacy could be seen as a potentiallimitation of our study. However, the db/db diabetic mice are thoseused most often as a model of human diabetic chronic wounds(87), and the genetically modified db/db diabetic mouse has beenused previously to support clinical trial development with othermolecules (88).To tease out the extent to which the QHREDGS peptide

could induce wound healing, we selected a larger, 8-mm di-ameter initial wound size rather than the commonly used 6-mmwound (76, 79). A low dose of peptide was tested in vivo basedon the concept that pharmaceutical agents should be applied at aminimal effective dosage to avoid potential side effects [e.g., anincreased rate of mortality secondary to malignancy was reportedfor patients treated with >45 g of Regranex (180 nmol PDGF-BB) (89)]. A high dose was tested in vivo because of the moreprofound effects on keratinocytes observed with the high peptidedose in vitro. Here, a single low dose of 4.4 nmol peptide/cm2 ofwound area was effective in promoting granulation tissue for-mation and wound closure in vivo, whereas daily application of0.4 nmol Regranex/cm2 of wound area was needed to promotegranulation tissue formation but did not shorten the time towound closure in a db/db mouse model (90).The QHREDGS peptide is available by cost-effective synthesis

with a precisely defined composition, offering an additional ad-vantage in potential clinical applications. The cost of the syntheticQHREDGS peptide used in our study was $1.6 Canadian dollars(CAD)/mg, which is ∼20 000-times cheaper than the commerciallyavailable rhPDGF-BB, an active component of the FDA-approvedRegranex, which costs $33,000 CAD/mg (Fisher Scientific). Thechoice of the hydrogel components, collagen and chitosan, alsowas motivated by clinical translation considerations. Chitosan hasbeen approved by the FDA for use in humans in topical appli-cations (e.g., HemCon bandages). Similarly, collagen is a com-ponent of numerous wound dressings currently on the market indifferent formulations such as freeze-dried sheets, pastes, pads, pow-ders, and gels (e.g., INTEGRA Matrix Wound Dressing, BIOSTEPcollagen matrix dressing, BGCMatrix, Stimulen Collagen, ColActive,and others).Here, we have demonstrated that in diabetic wounds keratinocyte

survival and collective migration represent promising alternativetherapeutic targets that support re-epithelialization, a hallmark ofwound regeneration and closure. In a genetically modified diabeticmouse, the immobilized QHREDGS peptide accelerated thewound-healing process by promoting the rate of re-epithelializationand the formation of granulation tissue without significantlyaffecting angiogenesis. Moreover, the immobilized QHREDGSpeptide did not increase the number of α-SMA–positive cells (e.g.,myofibroblasts) that can induce the antithetic wound-regenerationprocesses of wound contraction, collagen overproduction, and scarformation (91).Our reported results give rise to a number of questions. Mainly,

the efficacy of a potential therapeutic intervention providingsynergistic regulation of angiogenesis and re-epithelialization indiabetic chronic wounds is yet unknown. Notably, the chitosan–collagen hydrogel system is capable of simultaneously delivering

Xiao et al. PNAS | Published online September 19, 2016 | E5799

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US

other molecules, because the chitosan backbone is amenable tovarious chemical modifications (92, 93).In conclusion, the QHREDGS peptide promoted keratinocyte

adhesion and collective migration in vitro, as well as survival againstH2O2 stress through Akt and MAPKp42/44 pathways. In vivo, theQHREDGS peptide immobilized to a chitosan–collagen hydrogelaccelerated diabetic wound healing by enhanced re-epithelializationand granulation tissue formation. Together, our data in both normaland diabetic human primary keratinocytes and in a db/db dia-betic mouse model demonstrate the translational relevance of theQHREDGS peptide in treating diabetic wounds. We propose theQHREDGS peptide as a therapeutic candidate for promoting di-abetic wound healing.

Materials and MethodsStudy Design. We hypothesized that the QHREDGS peptide would promotekeratinocyte migration and survival under H2O2 stress and that a singleapplication of QHREDGS-immobilized chitosan–collagen hydrogel wouldaccelerate wound healing in diabetic mice. Endpoints and peptide dosagewere selected on the basis of previous studies and preliminary results (18).For animal studies, mice were randomized, and gross morphology analysiswas performed by an investigator blinded to the study groups.

Details on primary human keratinocytes cell culture, evaluation of solubleQHREDGS in vitro, the proliferation assay, H2O2 treatment of HEKs withsoluble QHREDGS peptide, the conjugation of QHREDGS to chitosan, solventcasting of chitosan–collagen films, coating validation; keratinocyte attach-ment on chitosan-only films, H2O2 treatment of keratinocyte on the chitosan–collagen films, the EarlyTox cell-integrity assay, Western blotting, the migrationassay, calcium increase for HEK migration, immunostaining, histological anal-ysis, and microvessel and positive staining analysis algorithms are given inSI Appendix, SI Materials and Methods.

Animals, Wound Model, and Treatment. The Animal Care Committee of theUniversity of Toronto approved all described animal studies. Eight-week-old,genetically diabetic, maleBKS.Cg-Dock7m +/+ Leprdb/J mice (db/db) (Stock000642) were ordered from Jackson Laboratories. Mice were acclimatized for1 wk, and their blood glucose levels were tested with a glucometer (Accu-Chek;Aviva) to confirm that plasma glucose levels were above 300 mg/dL on theday before surgery.

The chitosan–collagen hydrogel was prepared in a manner similar to thatdescribed previously (23). The final hydrogel consisted of 2.5 mg/mL chitosan

(with or without conjugated QHREDGS peptide) and 2.5 mg/mL type I collagenneutralized by 1 N NaOH and 10× PBS. The final hydrogel solution was mixedthoroughly and kept on ice until use. For in vivo application, the pre-gel so-lution was warmed for about 10min at 37 °C to initiate the gelling process andwas applied to the wound site with a 23.5-gauge needle. To benchmark thepeptide hydrogel versus an FDA-approved therapy, a commercially availableColActive collagen dressing was gifted by Covalon Technologies, Inc.

Micewere anesthetized with inhaled isoflurane (5%), and the dorsal surfaceof the mouse was shaved with an electric shaver, followed by treatment with ahair-removal cream (Veet). Betadine and 70% ethanol were applied in series tothe surgical site. Eight-millimeter biopsy punches (VWR) were used to createmiddorsal full-thickness wounds by excising the epidermis and dermis, in-cluding the panniculus carnosus. Fifty microliters of hydrogel without con-jugated QHREDGS peptide (control group), 50 μL of hydrogel containing atotal of 2.2 nmol peptide (low-peptide group), 50 μL of hydrogel containing atotal of 5.9 nmol peptide (high-peptide group), or a circular (8-mm diameter)ColActive collagen dressing (collagen group) was applied topically to thewound beds, or the wounds were left untreated (blank group). The woundbeds then were covered by Tegaderm film (Figs. 4B and 5A). Buprenorphine(0.03 mg/kg) was given s.c. before and immediately after the surgery as ananalgesic. Thereafter, the mice were housed individually and observed everyother day. Digital photographs of wounds were taken every 2 d at the samedistance by a camera (Canon) with a calibration scale on the side. Mice werekilled using CO2 asphyxiation, followed by cervical dislocation, on day 4(blank and high-peptide groups), day 8 (blank and high-peptide groups), day14 (blank, control, and low-peptide groups), or day 21 (blank, control, col-lagen, and high-peptide groups) as indicated in the Results and figures.

Statistical Analysis. All results are presented as mean ± SD. Statistical analysiswas performed using SigmaPlot 11.0. Differences between experimentalgroups were analyzed using one-way or two-way ANOVA followed by aTukey post hoc test for pairwise comparison. Linear regression was used tocalculate wound-closure rates, and significant differences were determinedby comparing the slopes of the fitted lines. A value of P < 0.05 was con-sidered statistically significant.

ACKNOWLEDGMENTS. We thank Val DiTizio of Covalon Technologies forthe kind gift of ColActive collagen dressing and L. E. Fitzpatrick andM. V. Seftonfor input on animal study and useful discussions. This work was funded by aNational Sciences and Engineering Research Council (NSERC) Steacie Fellowship(to M.R.), Canadian Institutes of Health Research Operating Grants MOP-126027and MOP-137107, NSERC Discovery Grant RGPIN 326982-10, and NationalInstitutes of Health Grant 2R01 HL076485.

1. Singer AJ, Clark RA (1999) Cutaneous wound healing. N Engl J Med 341(10):738–746.2. Eming SA, Martin P, Tomic-Canic M (2014) Wound repair and regeneration: Mecha-

nisms, signaling, and translation. Sci Transl Med 6(265):265sr6.3. Singh N, Armstrong DG, Lipsky BA (2005) Preventing foot ulcers in patients with di-

abetes. JAMA 293(2):217–228.4. Stojadinovic O, et al. (2008) Deregulation of keratinocyte differentiation and acti-

vation: A hallmark of venous ulcers. J Cell Mol Med 12(6B):2675–2690.5. Martin P, Lewis J (1992) Actin cables and epidermal movement in embryonic wound

healing. Nature 360(6400):179–183.6. Fernandez-Gonzalez R, Zallen JA (2013) Wounded cells drive rapid epidermal repair in

the early Drosophila embryo. Mol Biol Cell 24(20):3227–3237.7. Spravchikov N, et al. (2001) Glucose effects on skin keratinocytes: Implications for

diabetes skin complications. Diabetes 50(7):1627–1635.8. Kümin A, Huber C, Rülicke T, Wolf E, Werner S (2006) Peroxiredoxin 6 is a potent

cytoprotective enzyme in the epidermis. Am J Pathol 169(4):1194–1205.9. Lan CCE, Wu CS, Huang SM, Wu IH, Chen GS (2013) High-glucose environment en-

hanced oxidative stress and increased interleukin-8 secretion from keratinocytes: Newinsights into impaired diabetic wound healing. Diabetes 62(7):2530–2538.

10. Tarnuzzer RW, Schultz GS (1996) Biochemical analysis of acute and chronic woundenvironments. Wound Repair Regen 4(3):321–325.

11. Loots MA, et al. (1998) Differences in cellular infiltrate and extracellular matrix of chronicdiabetic and venous ulcers versus acute wounds. J Invest Dermatol 111(5):850–857.

12. Lan CC, Wu CS, Kuo HY, Huang SM, Chen GS (2009) Hyperglycaemic conditionshamper keratinocyte locomotion via sequential inhibition of distinct pathways: Newinsights on poor wound closure in patients with diabetes. Br J Dermatol 160(6):1206–1214.

13. Miklas JW, et al. (2013) QHREDGS enhances tube formation, metabolism and survivalof endothelial cells in collagen-chitosan hydrogels. PLoS One 8(8):e72956–e72956.

14. Rask F, et al. (2010) Photocrosslinkable chitosan modified with angiopoietin-1 pep-tide, QHREDGS, promotes survival of neonatal rat heart cells. J Biomed Mater Res A95(1):105–117.

15. Dang LT, et al. (2014) Inhibition of apoptosis in human induced pluripotent stem cellsduring expansion in a defined culture using angiopoietin-1 derived peptide QHREDGS.Biomaterials 35(27):7786–7799.

16. Feric N, et al. (2014) Angiopoietin-1 peptide QHREDGS promotes osteoblast differ-entiation, bone matrix deposition and mineralization on biomedical materials.Biomater Sci 2(10):1384–1398.

17. Reis LA, et al. (2012) A peptide-modified chitosan-collagen hydrogel for cardiac cellculture and delivery. Acta Biomater 8(3):1022–1036.

18. Reis LA, et al. (2015) Hydrogels with integrin-binding angiopoietin-1-derived pep-tide, QHREDGS, for treatment of acute myocardial infarction. Circ Heart Fail 8(2):333–341.

19. Ismail NS, Pravda EA, Li D, Shih S-C, Dallabrida SM (2010) Angiopoietin-1 reducesH(2)O(2)-induced increases in reactive oxygen species and oxidative damage toskin cells. J Invest Dermatol 130(5):1307–1317.

20. Stupack DG, Cheresh DA (2002) Get a ligand, get a life: Integrins, signaling and cellsurvival. J Cell Sci 115(Pt 19):3729–3738.

21. Hersel U, Dahmen C, Kessler H (2003) RGD modified polymers: Biomaterials forstimulated cell adhesion and beyond. Biomaterials 24(24):4385–4415.

22. Chiu LLY, Radisic M (2011) Controlled release of thymosin β4 using collagen-chitosancomposite hydrogels promotes epicardial cell migration and angiogenesis. J ControlRelease 155(3):376–385.

23. Xiao Y, Reis LA, Zhao Y, Radisic M (2015) Modifications of collagen-based biomate-rials with immobilized growth factors or peptides. Methods 84:44–52.

24. Pastar I, et al. (2014) Epithelialization in wound healing: A comprehensive review.Adv Wound Care (New Rochelle) 3(7):445–464.

25. Schäfer M, Werner S (2008) Oxidative stress in normal and impaired wound repair.Pharmacol Res 58(2):165–171.

26. James TJ, Hughes MA, Cherry GW, Taylor RP (2003) Evidence of oxidative stress inchronic venous ulcers. Wound Repair Regen 11(3):172–176.

27. Yeoh-Ellerton S, Stacey MC (2003) Iron and 8-isoprostane levels in acute and chronicwounds. J Invest Dermatol 121(4):918–925.

28. Scherer SS, et al. (2008) Wound healing kinetics of the genetically diabetic mouse.Wounds 20(1):18–28.

29. Cheng C, et al. (2013) Loss of innervation and axon plasticity accompanies impaireddiabetic wound healing. PLoS One 8(9):e75877.

30. Zochodne DW, Guo GF, Magnowski B, Bangash M (2007) Regenerative failure of di-abetic nerves bridging transection injuries. Diabetes Metab Res Rev 23(6):490–496.

E5800 | www.pnas.org/cgi/doi/10.1073/pnas.1612277113 Xiao et al.

31. Wang F, Gao N, Yin J, Yu FS (2012) Reduced innervation and delayed re-innervationafter epithelial wounding in type 2 diabetic Goto-Kakizaki rats. Am J Pathol 181(6):2058–2066.

32. Martin DC, Semple JL, Sefton MV (2010) Poly(methacrylic acid-co-methyl methacry-late) beads promote vascularization and wound repair in diabetic mice. J BiomedMater Res A 93(2):484–492.

33. Shen YI, et al. (2015) Acellular hydrogels for regenerative burn wound healing:Translation from a porcine model. J Invest Dermatol 135(10):2519–2529.

34. Hanjaya-Putra D, et al. (2013) Integration and regression of implanted engineeredhuman vascular networks during deep wound healing. Stem Cells Transl Med 2(4):297–306.

35. TsangMW, et al. (2003) Human epidermal growth factor enhances healing of diabeticfoot ulcers. Diabetes Care 26(6):1856–1861.

36. Hong JP, Jung HD, Kim YW (2006) Recombinant human epidermal growth factor(EGF) to enhance healing for diabetic foot ulcers. Ann Plast Surg 56(4):394–398, dis-cussion 399–400.

37. Tsuboi R, Rifkin DB (1990) Recombinant basic fibroblast growth factor stimulateswound healing in healing-impaired db/db mice. J Exp Med 172(1):245–251.

38. Obara K, et al. (2003) Photocrosslinkable chitosan hydrogel containing fibroblastgrowth factor-2 stimulates wound healing in healing-impaired db/db mice. Biomaterials24(20):3437–3444.

39. Galiano RD, et al. (2004) Topical vascular endothelial growth factor accelerates di-abetic wound healing through increased angiogenesis and by mobilizing and re-cruiting bone marrow-derived cells. Am J Pathol 164(6):1935–1947.

40. Saaristo A, et al. (2006) Vascular endothelial growth factor-C accelerates diabeticwound healing. Am J Pathol 169(3):1080–1087.

41. Bitto A, et al. (2008) Angiopoietin-1 gene transfer improves impaired wound healingin genetically diabetic mice without increasing VEGF expression. Clin Sci (Lond)114(12):707–718.

42. Jackson WM, Nesti LJ, Tuan RS (2012) Concise review: Clinical translation of woundhealing therapies based on mesenchymal stem cells. Stem Cells Transl Med 1(1):44–50.

43. Mizuno H, Tobita M, Uysal AC (2012) Concise review: Adipose-derived stem cells as anovel tool for future regenerative medicine. Stem Cells 30(5):804–810.

44. Sasso FC, et al. (2005) Increased vascular endothelial growth factor expression butimpaired vascular endothelial growth factor receptor signaling in the myocardium oftype 2 diabetic patients with chronic coronary heart disease. J Am Coll Cardiol 46(5):827–834.

45. Hazarika S, et al. (2007) Impaired angiogenesis after hindlimb ischemia in type 2 di-abetes mellitus: Differential regulation of vascular endothelial growth factor receptor1 and soluble vascular endothelial growth factor receptor 1. Circ Res 101(9):948–956.

46. Loomans CJ, et al. (2004) Endothelial progenitor cell dysfunction: A novel conceptin the pathogenesis of vascular complications of type 1 diabetes. Diabetes 53(1):195–199.

47. Gallagher KA, et al. (2007) Diabetic impairments in NO-mediated endothelial pro-genitor cell mobilization and homing are reversed by hyperoxia and SDF-1 alpha.J Clin Invest 117(5):1249–1259.

48. Eaglstein WH, Kirsner RS, Robson MC (2012) Food and Drug Administration (FDA)drug approval end points for chronic cutaneous ulcer studies. Wound Repair Regen20(6):793–796.

49. Kim JH, et al. (2013) Propulsion and navigation within the advancing monolayersheet. Nat Mater 12(9):856–863.

50. Londono C, et al. (2014) Nonautonomous contact guidance signaling during collectivecell migration. Proc Natl Acad Sci USA 111(5):1807–1812.

51. Vedula SR, et al. (2014) Epithelial bridges maintain tissue integrity during collectivecell migration. Nat Mater 13(1):87–96.

52. Davis S, et al. (1996) Isolation of angiopoietin-1, a ligand for the TIE2 receptor, bysecretion-trap expression cloning. Cell 87(7):1161–1169.

53. Brindle NP, Saharinen P, Alitalo K (2006) Signaling and functions of angiopoietin-1 invascular protection. Circ Res 98(8):1014–1023.

54. Dallabrida SM, Ismail N, Oberle JR, Himes BE, Rupnick MA (2005) Angiopoietin-1promotes cardiac and skeletal myocyte survival through integrins. Circ Res 96(4):e8–e24.

55. Dallabrida SM, et al. (2008) Integrin binding angiopoietin-1 monomers reduce cardiachypertrophy. FASEB J 22(8):3010–3023.

56. Takahashi K, et al. (2003) Adenoviral-delivered angiopoietin-1 reduces the infarctionand attenuates the progression of cardiac dysfunction in the rat model of acutemyocardial infarction. Mol Ther 8(4):584–592.

57. Gamble JR, et al. (2000) Angiopoietin-1 is an antipermeability and anti-inflammatoryagent in vitro and targets cell junctions. Circ Res 87(7):603–607.

58. Kim I, Moon SO, Park SK, Chae SW, Koh GY (2001) Angiopoietin-1 reduces VEGF-stimulated leukocyte adhesion to endothelial cells by reducing ICAM-1, VCAM-1, andE-selectin expression. Circ Res 89(6):477–479.

59. Fuxe J, et al. (2011) Pericyte requirement for anti-leak action of angiopoietin-1 andvascular remodeling in sustained inflammation. Am J Pathol 178(6):2897–2909.

60. Oike Y, et al. (2003) Angiopoietin-related growth factor (AGF) promotes epidermalproliferation, remodeling, and regeneration. Proc Natl Acad Sci USA 100(16):9494–9499.

61. Cho C-H, et al. (2006) COMP-angiopoietin-1 promotes wound healing through en-hanced angiogenesis, lymphangiogenesis, and blood flow in a diabetic mouse model.Proc Natl Acad Sci USA 103(13):4946–4951.

62. Van Slyke P, et al. (2009) Acceleration of diabetic wound healing by an angiopoietinpeptide mimetic. Tissue Eng Part A 15(6):1269–1280.

63. Rask F, et al. (2010) Hydrogels modified with QHREDGS peptide support cardiomyocytesurvival in vitro and after sub-cutaneous implantation. Soft Matter 6(20):5089–5099.

64. Goldfinger LE, et al. (1999) The alpha3 laminin subunit, alpha6beta4 and alpha3beta1integrin coordinately regulate wound healing in cultured epithelial cells and in theskin. J Cell Sci 112(Pt 16):2615–2629.

65. Shang M, Koshikawa N, Schenk S, Quaranta V (2001) The LG3 module of laminin-5harbors a binding site for integrin alpha3beta1 that promotes cell adhesion,spreading, and migration. J Biol Chem 276(35):33045–33053.

66. Manohar A, et al. (2004) Alpha 3 beta 1 integrin promotes keratinocyte cell survivalthrough activation of a MEK/ERK signaling pathway. J Cell Sci 117(Pt 18):4043–4054.

67. Sreejalekshmi KG, Nair PD (2011) Biomimeticity in tissue engineering scaffoldsthrough synthetic peptide modifications-altering chemistry for enhanced biologicalresponse. J Biomed Mater Res A 96(2):477–491.

68. Place ES, Evans ND, Stevens MM (2009) Complexity in biomaterials for tissue engi-neering. Nat Mater 8(6):457–470.

69. Martínez A, Blanco MD, Davidenko N, Cameron RE (2015) Tailoring chitosan/collagenscaffolds for tissue engineering: Effect of composition and different crosslinkingagents on scaffold properties. Carbohydr Polym 132:606–619.

70. Sen CK, et al. (2009) Human skin wounds: A major and snowballing threat to publichealth and the economy. Wound Repair Regen 17(6):763–771.

71. Brownlee M (1995) Advanced protein glycosylation in diabetes and aging. Annu RevMed 46:223–234.

72. Sun G, et al. (2011) Dextran hydrogel scaffolds enhance angiogenic responses andpromote complete skin regeneration during burn wound healing. Proc Natl Acad SciUSA 108(52):20976–20981.

73. Schwarz ER, et al. (2000) Evaluation of the effects of intramyocardial injection of DNAexpressing vascular endothelial growth factor (VEGF) in a myocardial infarctionmodel in the rat–angiogenesis and angioma formation. J Am Coll Cardiol 35(5):1323–1330.

74. Soneja A, Drews M, Malinski T (2005) Role of nitric oxide, nitroxidative and oxidativestress in wound healing. Pharmacol Rep 57(Suppl):108–119.

75. Martino MM, et al. (2014) Growth factors engineered for super-affinity to the ex-tracellular matrix enhance tissue healing. Science 343(6173):885–888.

76. Randeria PS, et al. (2015) siRNA-based spherical nucleic acids reverse impaired woundhealing in diabetic mice by ganglioside GM3 synthase knockdown. Proc Natl Acad SciUSA 112(18):5573–5578.

77. Xie Z, et al. (2013) Dual growth factor releasing multi-functional nanofibers forwound healing. Acta Biomater 9(12):9351–9359.

78. Shen YI, et al. (2016) Engineered human vascularized constructs accelerate diabeticwound healing. Biomaterials 102:107–119.

79. Martino MM, et al. (2011) Engineering the growth factor microenvironment withfibronectin domains to promote wound and bone tissue healing. Sci Transl Med3(100):100ra89.

80. Lim YC, et al. (2015) Proinsulin C-peptide prevents impaired wound healing by acti-vating angiogenesis in diabetes. J Invest Dermatol 135(1):269–278.

81. Balaji S, et al. (2012) Tissue-engineered provisional matrix as a novel approach toenhance diabetic wound healing. Wound Repair and Regeneration 20(1):15–27.

82. Asai J, et al. (2006) Topical sonic hedgehog gene therapy accelerates wound healingin diabetes by enhancing endothelial progenitor cell-mediated microvascular re-modeling. Circulation 113(20):2413–2424.

83. Xu J, et al. (2012) The role of microRNA-146a in the pathogenesis of the diabeticwound-healing impairment: Correction with mesenchymal stem cell treatment.Diabetes 61(11):2906–2912.

84. Albrecht-Schgoer K, et al. (2014) Topical secretoneurin gene therapy accelerates di-abetic wound healing by interaction between heparan-sulfate proteoglycans andbasic FGF. Angiogenesis 17(1):27–36.

85. Roy DC, Mooney NA, Raeman CH, Dalecki D, Hocking DC (2013) Fibronectin matrixmimetics promote full-thickness wound repair in diabetic mice. Tissue Eng Part A19(21-22):2517–2526.

86. Ragnarson Tennvall G, Apelqvist J (2004) Health-economic consequences of diabeticfoot lesions. Clin Infect Dis 39(Suppl 2):S132–S139.

87. Galiano RD, Michaels J, 5th, Dobryansky M, Levine JP, Gurtner GC (2004) Quantitativeand reproducible murine model of excisional wound healing. Wound Repair Regen12(4):485–492.

88. Hanft JR, et al. (2008) Phase I trial on the safety of topical rhVEGF on chronic neu-ropathic diabetic foot ulcers. J Wound Care 17(1):30–32, 34–37.

89. US Food and Drug Administration (2008) FDA announces new labeling changes forRegranex: Product to carry boxed warning (US Food and Drug Administration;Washington, DC).

90. Chan RK, et al. (2006) Effect of recombinant platelet-derived growth factor (Re-granex) on wound closure in genetically diabetic mice. J Burn Care Res 27(2):202–205.

91. Gabbiani G (2003) The myofibroblast in wound healing and fibrocontractive diseases.J Pathol 200(4):500–503.

92. Alves NM, Mano JF (2008) Chitosan derivatives obtained by chemical modificationsfor biomedical and environmental applications. Int J Biol Macromol 43(5):401–414.

93. Dash M, Chiellini F, Ottenbrite RM, Chiellini E (2011) Chitosan-A versatile semi-syntheticpolymer in biomedical applications. Prog Polym Sci 36(8):981–1014.

Xiao et al. PNAS | Published online September 19, 2016 | E5801

APP

LIED

BIOLO

GICAL

SCIENCE

SPN

ASPL

US