Elevated Thrombospondin 2 Contributes to Delayed Wound Healing in Diabetes Free

A serious complication of diabetes is impaired wound healing, which can lead to chronic wounds, hospitalization, and limb amputation in patients with diabetes (1). The extracellular matrix (ECM) coordinates healing by providing structure and facilitating cell-cell and cell-matrix interactions (2,3). Diabetes-related damage to ECM disrupts these functions and contributes to delayed healing (3); for example, damaged ECM impairs fibroblast migration and proliferation (4). Though ECM is critical to wound healing, its role in impaired healing in diabetes has not been extensively studied.

One ECM component involved in wound healing is thrombospondin 2 (TSP2), a matricellular protein that influences cell-matrix interactions (5). TSP2 knockout (KO) mice exhibit accelerated healing, concomitant with altered ECM and increased blood vessel density (6). Conversely, elevated TSP2 expression, as observed in Akt1 KO mice, is associated with delayed wound healing (7). Recent work has provided evidence that TSP2 expression is elevated in retina of patients with diabetes and diabetic rats (8) and endothelial progenitor cells (EPCs) in diabetic mice (9). Despite its role in wound healing, TSP2 expression in diabetes and function in diabetic wounds remain unknown. Here, we describe TSP2 expression in diabetes, investigate the contribution of TSP2 to wound healing, and characterize its regulation by hyperglycemia.

TSP2 antibody was custom-made by GenScript. Actinomycin D, cycloheximide, 2-deoxyglucose, l-glucose, azaserine (Sigma-Aldrich), glucosamine (MP Biomedicals), and BAY11-7082 (Santa Cruz Biotechnology) were prepared according to the manufacturer’s instructions.

Skin samples from four patients with type 2 diabetes (T2D) undergoing limb amputation and foreskin samples from one neonatal and three adult patients with no history of diabetes were obtained through Yale Pathology Tissue Services. Tissue underwent bead mill homogenization and centrifugation for immunoblot analysis.

TSP2 KO mice (C57BL/6) have previously been described (10). db/db TSP2 KO (DKO) animals were generated by breeding TSP2 KO mice with db/+ heterozygotes obtained from The Jackson Laboratory [no. 00697; B6.BKS(D)-Lepr^db/J] and subsequent breeding of db/+ TSP2 KO animals. Matched db/db mice were generated using the same breeding scheme, and C57BL/6 mice served as controls to match the background of db/db and TSP2 KO mice. C57BL/6 mice were given a single injection of streptozotocin according to standard protocols (11). Fasting blood glucose was measured with a OneTouch Ultra monitor. Body composition was analyzed with EchoMRI (Echo Medical Systems). Ten- to twelve-week-old db/db, DKO, STZ, and wild-type (WT) mice were used for wound experiments and dermal fibroblast (DF) isolation. All procedures were approved by the Institutional Animal Care and Use Committee at Yale University.

Full-thickness, nonsplinted excisional wounds were generated as previously described (6,12–14). Briefly, mice were anesthetized and depilated, and two 6-mm wounds were created with a biopsy punch (Acuderm) and left uncovered. Animals were euthanized 7, 10, or 14 days post-wounding and wounds excised with 3 mm surrounding tissue. Wounds were fixed overnight in zinc-buffered formalin and embedded in paraffin. Sections were stained with Masson trichrome according to standard protocols and analyzed by immunohistochemistry for TSP2 (GenScript), vimentin (EMD Millipore), CD31 (dianova), and α-smooth muscle actin (αSMA) (Dako).

DFs were isolated from dorsal skin as previously described (7). NIH/3T3 cells (ATCC) and DFs were maintained in DMEM, high glucose (Gibco) supplemented with 10% FBS and 1% penicillin/streptomycin. For experiments requiring 5 or 30 mmol/L glucose, glucose-free DMEM was supplemented with d-glucose (Hardy Diagnostics).

RNA isolation, reverse transcription, quantitative RT-PCR, and Western blotting with primary antibodies against TSP2, β-actin (Sigma-Aldrich), HSP90 (Santa Cruz), O-GlcNAc (Cell Signaling Technology), O-linked N-acetylglucosamine transferase (OGT) (Cell Signaling Technology), and p65 (Cell Signaling Technology) were performed as previously described (15).

Scratch migration assays were performed as previously described (7). Briefly, DFs were plated in 5 mmol/L glucose DMEM. When cells reached confluence, a scratch was created; then, cells were washed with PBS, treated with 5 or 30 mmol/L glucose DMEM, and imaged after 24 h.

Luciferase assays were performed as previously described (16).

OGT siRNA was synthesized by Thermo Fisher Scientific, and RelA cFlag pcDNA3 was a gift from Stephen Smale, University of California, Los Angeles, School of Medicine (no. 20012; Addgene). NIH/3T3s were transfected with OGT siRNA, nonsilencing siRNA (Dharmacon), RelA cFLAG pcDNA3, or empty vector (Addgene) overnight. Then cells were treated with 5 or 30 mmol/L glucose DMEM for 48–72 h.

Comparisons were made using Student t test, one-way ANOVA followed by Tukey post hoc test, or Fisher exact test as specified in figure legends, with P < 0.05 considered statistically significant. All statistical analyses were performed using GraphPad Prism. Data are presented as mean ± SEM.

Complete data sets generated and analyzed during the current study are available from T.R.K. upon request.

TSP2 level in skin from T2D patients undergoing lower-limb amputation and skin from patients with no history of T2D was assessed by immunoblotting. Results showed a 2.9-fold increase in TSP2 level in T2D skin compared with normal skin (Fig. 1A and B). Consistent with findings in human tissue, TSP2 level was also elevated in diabetic mice. DFs were isolated from WT and db/db mice, and immunoblotting revealed a 2.3-fold increase in TSP2 level in the latter (Fig. 1C and D). TSP2 was also elevated in db/db wounds at day 10 (Fig. 1E–G), the point at which we have previously shown that TSP2 reaches its peak in WT wounds (6). TSP2 wound deposition was also increased in a model of type 1 diabetes (Supplementary Fig. 1).

For investigation of the impact of increased TSP2, a diabetic DKO mouse model was generated. Genotype was confirmed by PCR, Western blot, and immunohistochemistry (Supplementary Fig. 2). DKO mice developed obesity and hyperglycemia comparable with db/db controls (Fig. 2A and Supplementary Fig. 3).

Healing in DKO mice was evaluated through a full-thickness excisional wound model (12,13). DKO mice exhibited improved healing, with more rapid reduction of wound size (Fig. 2B and C). DKO wounds were significantly smaller by 10 days post-wounding (Fig. 2B and C) and exhibited improved closure by reepithelialization compared with db/db (Fig. 2D). Moreover, histological evaluation of Masson trichrome-stained tissue sections showed significantly increased ECM deposition in DKO mice (Fig. 2E–I).

Immunostaining for vimentin, a mesenchymal marker expressed by fibroblasts, was increased in DKO wounds at 7 days (Fig. 3A–C). Consistent with these results, DKO DFs exhibited improved migration in a scratch migration assay (Supplementary Fig. 4), rescuing the defect in db/db DFs. At 10 days post-wounding, there was no difference in CD31 immunostaining between genotypes (Fig. 3D–F), but the number of αSMA-positive vessels was increased in DKO mice, indicating an increased number of mature blood vessels (Fig. 3G–I). It should be noted that our observations are based on the examination of few selected time points and it is possible that additional differences between DKO and db/db wounds exist, especially given the large changes in wound closure rate observed between days 10 and 14.

Due to the limited growth capacity of DFs, we used NIH/3T3 fibroblasts to explore TSP2 regulation. NIH/3T3s cultured in high glucose (HG) media exhibited elevated TSP2 (Supplementary Fig. 5), consistent with findings in db/db mice and DFs.

HG did not alter TSP2 mRNA or protein stability, and in luciferase assays HG doubled TSP2 promoter activity (Supplementary Fig. 6), indicating that increased TSP2 is a result of increased transcription. Transcriptional changes can result from activation of intracellular pathways that rely on glycolytic intermediates (17). For determination of whether glucose metabolism is required to increase TSP2 expression, NIH3T3s were treated with l-glucose, which cannot enter cells, or 2-deoxyglucose, which cannot be metabolized beyond the first step of glycolysis. Neither l-glucose nor 2-deoxyglucose treatment increased TSP2 expression (Supplementary Fig. 6), demonstrating the necessity of glucose catabolism and glucose-6-phosphate, the first intermediate of glycolysis, or its derivatives.

To identify the glucose-related pathway responsible for TSP2 expression, we inhibited each individually including the hexosamine pathway, which converts glucose-6-phosphate to uridine diphosphate N-acetyl-glucosamine (UDP-GlcNAc). Attachment of UDP-GlcNAc at Ser/Thr sites can alter protein activity and induce changes in gene expression (18). The pathway was blocked with azaserine (200 μmol/L), which inhibits its rate-limiting enzyme, glutamine:fructose-6-phospate aminotransferase (GFAT), and normalized TSP2 levels in HG (Fig. 4A–C). Similarly, siRNA-mediated knockdown of OGT, the enzyme that attaches UDP-GlcNAc to proteins, abolished elevated TSP2 expression in HG (Fig. 4D and E). For confirmation of the role of the hexosamine pathway in TSP2 expression, NIH/3T3s were treated with 1 mmol/L glucosamine, an intermediate downstream of glucose-6-phosphate. Glucosamine treatment increased TSP2 expression after 24 h (Fig. 4F and G). Together, these results establish a role for the hexosamine pathway in regulation of TSP2 expression.

We next aimed to identify the transcription factor responsible for TSP2 expression by focusing on factors that are activated by O-GlcNAc modification and have putative binding sites on the TSP2 promoter, such as nuclear factor-κB (NF-κB) (19,20). The p65 subunit of NF-κB can be directly modified by O-GlcNAc in HG, which increases its transcriptional activity and expression of its target genes (21). For evaluation of the role of activated NF-κB, NIH/3T3s were treated with the NF-κB inhibitor BAY11-7082 for 48 h. BAY11-7082 treatment resulted in lower TSP2 expression in HG than in control cells (Fig. 4H and I). Further, overexpression of the p65 (RelA) subunit in NIH/3T3s was sufficient to increase TSP2 expression in normal glucose after 48 h (Fig. 4J and K). These data suggest that TSP2 expression in HG is mediated by hexosamine pathway–dependent NF-κB signaling.

In this work, we describe TSP2 expression in diabetes and provide insights into the mechanism of its upregulation via the hexosamine pathway. The most striking result is that diabetic mice lacking TSP2 exhibit improved healing associated with increased cell migration and blood vessel maturation. Our findings support the hypothesis that ECM contributes to poor healing in diabetes, which has been suggested by studies of ECM-targeting proteases (22).

Our work shows that overexpression of TSP2 in diabetes contributes to compromised healing through its impact on cell migration. Analysis of wound angiogenesis at day 10, when TSP2 KO wounds differ most significantly from WT wounds (6), revealed more αSMA-positive vessels in DKO wounds, indicative of enhanced blood vessel maturation and consistent with studies of db/db wounds treated with TSP2 KO decellularized skin slabs (13). Accelerated vessel maturation in DKO wounds likely results from the influence of TSP2 on migration of fibroblasts, which facilitate repair and migration of other cell types, into the wound at day 7.

The mechanism underlying TSP2 overexpression in diabetes is poorly understood and limited to observations in EPCs (9). By focusing on fibroblasts, the primary TSP2-producing cells during healing (7), we determined that TSP2 overexpression results from increased flux through the hexosamine pathway. This stands in contrast to the reactive oxygen species–mediated mechanism described previously (9); reduction of oxidative stress in db/db EPCs abolished TSP2 overexpression and elevated TSP2 expression in EPCs in HG was correlated with increased oxidative stress (9). This discrepancy may be explained by cell type–specific expression of glucose-responsive transcription factors.

The hexosamine pathway regulates gene expression through posttranslational modification of transcription factors (18). Because of the NF-κB response element in the TSP2 promoter and evidence that NF-κB is activated by the hexosamine pathway (19,20), we explored its role in fibroblasts. Treatment with an NF-κB inhibitor normalized TSP2 levels in hyperglycemia, and overexpression of p65 increased TSP2 expression in normoglycemia, suggesting NF-κB participation in TSP2 overexpression. This contrasts with a previous study in which TSP2 increased NF-κB activity, but NF-κB did not regulate TSP2 expression in endothelial cells lacking cytochrome P-4501B1; this difference may result from cross talk between NF-κB and reactive oxygen species (23). Hexosamine pathway–induced activation of NF-κB has previously been associated with glycosylation and acetylation of specific NF-κB residues in human embryonic kidney cells (HEK293T) and smooth muscle cells (21,24); this or a similar mechanism might be involved in DFs.

In summary, we have identified TSP2 as a contributor to impaired healing in diabetes and a new mechanism of TSP2 regulation by glucose. Additional study of the function of TSP2 in diabetic fibroblasts will provide further insight into its role in diabetes complications. Significantly, our study highlights the importance of ECM in facilitating optimal cell-cell and cell-matrix interactions during healing.

Acknowledgments. The authors thank Ashley Bauer, Department of Pharmacology, Yale School of Medicine, and Dr. Nathan L. Price, Interdepartmental Program in Vascular Biology and Therapeutics, Yale School of Medicine, for invaluable technical assistance.

Funding. This work was supported by National Institutes of Health grants HL107205 and GM072194 and the Gruber Foundation Science Fellowship (to B.K.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. B.K. and T.R.K. designed the study, performed experiments, analyzed data, and wrote the manuscript. T.B., H.X., A.H.M., and A.K.L. designed and performed experiments and discussed data. J.W. performed experiments. C.F.-H. designed the study, contributed to discussion, and reviewed and edited manuscript. T.R.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at Experimental Biology 2018, San Diego, CA, 21–25 April 2018.