Angiogenin Released from ABCB5⁺ Stromal Precursors Improves Healing of Diabetic Wounds by Promoting Angiogenesis

Abstract

Severe angiopathy is a major driver for diabetes associated secondary complications. Knowledge on underlying mechanisms essential for advanced therapies to attenuate these pathologies is limited. Injection of ABCB5⁺ stromal precursors (SPs) at the edge of non-healing diabetic wounds in a murine db/db model, closely mirroring human type II diabetes, profoundly accelerates wound closure. Strikingly, enhanced angiogenesis was substantially enforced by the release of the ribonuclease angiogenin from ABCB5⁺ SPs. This compensates for the profoundly reduced angiogenin expression in non-treated murine chronic diabetic wounds. Silencing of angiogenin in ABCB5⁺ SPs prior to injection significantly reduced angiogenesis and delayed wound closure in diabetic db/db mice implying an unprecedented key role for angiogenin in tissue regeneration in diabetes. These data hold significant promise for further refining SPs-based therapies of non-healing diabetic foot ulcers and other pathologies with impaired angiogenesis.

Graphical Abstract

Impaired healing of chronic wounds is mainly caused by a lack of angiogenesis. This is due to insulin resistance enforced - among other reasons - by a significant increase in fat mass resulting in metabolic dysfunction in type II diabetes. In such a hostile diabetic environment the number of stromal precursors and possibly their secretory function severely declines. In particular the pro-angiogenic factor angiogenin is suppressed, leading to disturbed tissue repair. The lack of angiogenin in chronic diabetic wounds can efficiently be supplemented by injection of ABCB5⁺ SPs into diabetic wounds. ABCB5⁺ SPs release high amounts of angiogenin and, in consequence, enhance angiogenesis through its paracrine actions on endothelial cells with subsequent activation of VEGFR signaling in diabetic wounds. In addition, angiogenin also restores wound contraction and matrix deposition, essential phases during wound healing.

Introduction

Diabetes mellitus is an urgent global health problem imposing a significant socioeconomic burden on societies. Around 600 million people are predicted to suffer from diabetes by 2045 (Aldworth et al., 2017) and, in consequence, secondary complications like non-healing wounds will profoundly increase. The most important among them are vascular pathologies including micro- and macroangiopathy. In fact, the vast majority of patients undergoing non-traumatic lower-limb amputations have diabetes (Armstrong et al., 2020). Despite major efforts (Okonkwo et al., 2020, Tahergorabi and Khazaei, 2012), the mechanisms of vasculopathies and reduced angiogenesis are still not sufficiently understood in diabetes. In consequence, there is an outstanding interest to (i) get better insight how angiogenesis is suppressed and (ii) to further advance therapies to improve angiogenesis in chronic wounds in diabetes. Of particular clinical relevance are stem cell-based therapies that will be cost saving in long term in chronic wounds and other conditions (Augustin and Vanscheidt, 2012, Soria-Juan et al., 2019, Thavorn et al., 2020). Stromal precursors in their endogenous niche in the skin are largely unexplored under diabetic conditions. The beneficial effects of stromal precursor therapies - though incompletely understood for diabetes - are primarily due to paracrine signals in wound models (Davey et al., 2014, Jiang et al., 2013, Qi et al., 2014, Vander Beken et al., 2019). We here are interested whether local administration of stromal precursors (SPs) at the wound site enhance angiogenesis and restore delayed wound healing in diabetic wounds. A major obstacle still impeding the successful implementation of SP-based therapies in clinical routine is the lack of a cell surface marker that reliably allows to isolate SPs from the skin to further enrich, expand and reproducibly test them for their paracrine efficiency and potency. Previously, we identified a cell population in the dermis expressing the ATP-binding cassette in addition to well established SP cell markers (Vander Beken et al., 2019). This ABCB5⁺ SPs can suppress effector T-cells, while stimulating regulatory T cells in vitro and in vivo (Schatton et al., 2015). In addition, ABCB5⁺SPs suppress the unrestrained activation of pro-inflammatory M1 macrophages in a murine iron-overload model (Vander Beken et al., 2019). Employing antibodies against ABCB5, it is possible to reliably isolate ABCB5⁺SPs under GMP standards to a homogeneous population which even after in vitro expansion partly maintains their stemness, mesenchymal expression profile and meets stringent quality criteria (Tappenbeck et al., 2019, Vander Beken et al., 2019).

Results

Diabetic conditions suppress angiogenesis and rewire the global transcriptome of endothelial cells

To assess the influence of obesity induced diabetes on angiogenesis, we first performed in vitro tube formation indicative of angiogenesis. For this purpose, human umbilical vein endothelial cells (HUVECs) were co-cultured with fibroblasts from either wild type or db/db mice. The db/db mice closely mimics metabolic changes in human type II diabetes (Chen et al., 1996). Blood vessel (tube) formation was severely suppressed in the presence of db/db mouse fibroblasts as opposed to wild type fibroblasts (Figure 1a and 1b). These results clearly show detrimental consequences of db/db stromal fibroblasts on endothelial cell function.

Figure 1. Diabetic stroma suppresses angiogenesis and leads to profound transcriptomic alterations in endothelial cells.

(a) In vitro angiogenesis from co-cultures of endothelial cells (HUVECs) with wild type mouse fibroblasts (Wt MDFs, middle panel) or db/db mouse fibroblasts (db/db MDFs, right panel). HUVECs cultured alone (left panel) served as a negative control. Scale bars: 50μM.

(b) Quantification of branched vessels in the angiogenesis assay at 14 days of co-culture. Statistical analysis was performed using one-way ANOVA, values are represented as mean ± SEM, n=3.

(c) Scatter plot showing differentially regulated genes and (d) Pathway enrichment analysis demonstrating key significantly enriched signaling pathways, which are upregulated or (e) downregulated in HUVECs following 5 days co-culture with db/db mouse fibroblasts as compared to wild type mouse fibroblasts.

To further identify the underlying mechanisms that impair endothelial cell function, we co-cultured HUVECs with fibroblasts from either wild type or db/db mice for 5 days followed by transcriptomic analysis of FACS-sorted HUVECs. We observed marked alterations in the global transcriptome of HUVECs following co-culture with db/db fibroblasts as opposed to wild type fibroblasts (Figure 1c). Following co-culture with db/db fibroblasts, we observed upregulation of proinflammatory cytokines (IL-6 and IL1β), matrix degrading enzymes (MMP1, MMP7, MMP10 and MMP19), tumor angiogenesis suppressor (TP73) genes as well as genes associated with DNA damage response (RAD51, RAD9B and XRCC2), senescence (CDKN2A or p16) and apoptosis (FASLG) (Figure 1c). By contrast, genes regulating endothelial adhesion and migration including VCAM1, KIF26A and ITGB4 were downregulated in HUVECs co-cultured with db/db fibroblasts as opposed to wild type fibroblasts (Figure 1c).

Furthermore, pathway enrichment analysis revealed overrepresentation of ‘Cytokine signaling’, ‘Apoptosis’, ‘Interleukin-1 signaling’, ‘Collagen degradation’ and ‘DNA repair pathway’ in HUVECS co-cultured with db/db fibroblasts as compared to wild type fibroblasts (Figure 1d). The pathways associated with ‘NOTCH signaling’, ‘AKT signaling’,’Integrin cell surface interactions’ and ‘IL-10 signaling’ were underrepresented in HUVECs co-cultured with db/db fibroblasts as compared to wild type fibroblasts (Figure 1e). Consistent with these findings HUVEC numbers markedly declined after 5 day of co-culture with db/db stromal fibroblasts as opposed to wild type fibroblasts (Supplementary Figure S1a). In aggregate, these findings indicate that adverse signaling of diabetic fibroblasts severely impairs endothelial cell homeostasis and promotes angiogenesis suppressing and angiopathy-associated gene programs.

Vascular network and numbers of stromal precursor diminish in diabetes

Next, we examined angiogenesis in the skin of adult diabetic mice. Severely reduced numbers of CD31⁺ vessels were observed in the skin of diabetic mice as opposed to wild type mice (Figure 2a). We hypothesized that local signals originating from stromal precursor cells are essential for endothelial function and homeostasis. Therefore, we explored the endogenous stromal niche for distinct stromal precursor subsets to understand whether diabetes may affect the endogenous stromal niche. Immunostaining of skin sections was performed with antibodies against ABCB5 that specifically identify a subset of stromal precursors (Vander Beken et al., 2019). Interestingly, a severe reduction in the number of ABCB5⁺ SPs was found in the dermis of db/db mice as opposed to wild type mice (Figure 2b). We also observed a marked decrease in ABCB5⁺ SP numbers in close vicinity to blood vessels in the skin from diabetic mice (Supplementary Figure S1b). The number of ABCB5⁺ SPs was also severely reduced during tissue repair in diabetic mice as compared to higher numbers in wild type mice (Supplementary Figure S1c). We also observed marked reduction in other stromal precursors including PDGFRβ⁺, Nestin⁺ and CxD34⁺ cells (Koliaraki et al., 2020) in db/db mouse skin as compared to wild type (Supplementary Figure S2a). This may imply that metabolic changes in diabetes profoundly affect many types of stromal precursor cells in their endogenous niche.

Figure 2. Diabetes severely affects the vascular and stromal compartment.

(a) Immunostaining and quantification of blood vessels marker CD31 (red) and (b) ABCB5⁺ (green) marking dermal SPs population in wild type and db/db mouse skin. Scale bar: 20μm. Unpaired t-test, values are represented as mean ± SEM, n=3.

(c) Heatmap shows the gene expression profile of arteries and veins isolated from wild type and db/db mice.

(d) Pathway enrichment analysis demonstrating key significantly enriched signaling pathways, which are upregulated or (e) downregulated in db/db mice derived arterial blood vessels compared to wild type.

(f) Western blot analysis of indicated proteins from skin lysates of wild type and db/db mice. This Western blot is representative of three independent experiments. Quantification is shown in Supplementary Figure S3b.

Furthermore, our data unveiled a transcriptomic profile fundamentally different in endothelial cells derived from db/db mice as opposed to wild type (Figure 2c). Genes linked to Notch signaling (Hey2), VEGF signaling (Vegfa and Vegfb, Vegfc, Flt1, Flt4 and Kdr), IGF-1-AKT signaling (Ip6k3), ribosomal biogenesis (Rpl7a and Cars1), and genes encoding for the actin organizing protein Klhl1 and the negative regulator of autophagy Tmem208 were suppressed in either arteries or veins (Figure 2c). By contrast, genes linked to DNA double strand breaks (Aunip), the gene encoding RNA Polymerase II Associated Protein 3 (RPAP3) and the G protein receptor coding gene Gpr165 were upregulated in blood vessels from diabetic mice as opposed to wildtype (Figure 2c). The PCA plot revealed a close clustering of diabetic arteries and veins as compared to a broader distribution of wild type markers, indicating that both type of vessels undergo marked remodeling and transcriptional alteration under diabetic environment (Supplementary Figure S2b). Similar to pathway enrichment analysis of the co-culture experiments with endothelial cells and db/db fibroblasts (Figure 1d and 1e), the signature of endothelial cells from arteries and veins of db/db mice - representing the diabetic microenvironment in vivo - also revealed overrepresentation of ‘Cytokine signaling’, ‘Apoptosis’, ‘Degradation of extracellular matrix’, ‘Interferon signaling’ and ‘DNA damage’ pathways (Figure 2d and Supplementary Figure S2c). Of note, the pathways associated with ‘VEGF signaling’, ‘Insulin IGF-1/AKT signaling’, and ‘rRNA processing’ were underrepresented in blood vessels from diabetic mice as compared to wild type (Figure 2e and Supplementary Figure S2d). These data imply that microenvironmental cues originating in part from the stromal compartment in diabetic mice impact on endothelial cell signaling and function.

In this context, we observed markedly reduced collagen deposition in the dermis of diabetic mice as opposed to wild type mice (Figure 2f, Supplementary Figure S3a and Supplementary Figure S3b). Interestingly, enhanced expression of senescence markers p21 and p16 (Meyer et al., 2017, Ressler et al., 2006) was observed in the skin of 3 month old diabetic db/db mice as opposed to wild type mice (Figure 2f and Supplementary Figure S3b). These data highlight that the dermal stromal compartment, as a part of the endogenous niche, is significantly affected in diabetic mice skin and may disrupt the niche environment of stromal precursors.

ABCB5⁺ SPs enhance angiogenesis and accelerate wound healing in db/db mice

Previously, it was reported that wound healing is impaired in db/db mice (Chen et al., 1996, Xiao et al., 2016). Based on our observations that the stromal compartment including the endogenous niche of stromal precursor cells is profoundly affected in db/db mice, and SPs were reduced in numbers in the db/db skin, we next wished to explore whether introduction of stromal precursor cells at the wound site may be beneficial for the wound healing deficiency in diabetic mice.

To exclude any engraftment or cell fusion effects, we purposely used an established xenotransplant model (Vander Beken et al., 2019) with local injection of human-derived ABCB5⁺ SPs into wounds of db/db mice. Intradermal injection of either phosphate buffered saline (PBS, vehicle) (100μl), 1×10⁶ human dermal fibroblasts (100μl) or human skin derived 1×10⁶ ABCB5⁺ SPs (100μl) was performed at the wound edge in diabetic db/db mice (Figure 3a).

Figure 3. ABCB5⁺ SPs enhance wound healing and stimulate angiogenesis in diabetic mice wounds.

(a) Scheme showing experimental setting.

(b) Time kinetic of wound healing in PBS (vehicle), fibroblasts or ABCB5⁺ SPs-injected wounds of diabetic db/db mice. One way ANOVA, values are represented as mean ± SEM, n=5.

(c) Representative H&E photomicrographs depicting wound architecture in PBS (vehicle), fibroblasts or ABCB5⁺ SPs-injected wounds from diabetic mice. Scale bars, 500 μm.

(d) Immunostaining and quantification of blood vessel formation marked with CD31 (red) in day 10 wounds of db/db mice which had been injected with PBS (vehicle), fibroblasts or ABCB5⁺ SPs after wounding. Scale bars, 50μm. E, epidermis; GT, granulation tissue (n=5). One-way ANOVA, values are represented as mean ± SEM, n=5.

Quantitative analysis of wound areas shows that administration of 1×10⁶ ABCB5⁺ SPs per wound substantially enhanced wound closure at day 3, 5, 7 and 10 (Figure 3b) as opposed to vehicle treated diabetic mice (compare median of filled triangles: db/db mice with vehicle and filled circle: db/db mice with ABCB5⁺ SPs). ABCB5⁺ SPs-injected wounds showed better wound closure at day 5, 7, and 10 as opposed to fibroblasts-injected wounds (compare median of filled rectangles: db/db mice with fibroblasts and filled circles: db/db mice with ABCB5⁺ SPs) (Figure 3b). Injection of fibroblasts improved wound healing at day 7 when compared to PBS injected wounds (compare median of filled rectangles: db/db mice with fibroblasts and filled triangles: db/db mice with vehicle) (Figure 3b). These results suggest a therapeutic benefit of ABCB5⁺ SPs for wounds in diabetic mice.

Consistent with these findings, we observed smaller wound areas and marked matrix deposition (depicted by eosin stained material) following ABCB5⁺ SPs injection (Figure 3c). We next analyzed the newly formed vessels in wounds by immunostaining with antibodies against CD31, indicative of endothelial cells. Interestingly, injection of ABCB5⁺ SPs significantly enhanced angiogenesis in db/db mice wounds compared to the vehicle (PBS) and fibroblasts treated wounds (Figure 3d). These findings suggest that the beneficial role of ABCB5⁺ SPs on diabetic wound therapy is at least in part due to enhanced angiogenesis.

Diabetic microenvironment abrogates angiogenin expression

Next we set out to dissect the underlying mechanisms by which ABCB5⁺ SPs accelerate wound healing and stimulate angiogenesis. We first analyzed the secretome of ABCB5⁺ SPs to address the question whether they release trophic or proangiogenic factors which may enhance angiogenesis and, thus, improve impaired angiogenesis. Employing antibody arrays, we uncovered significant enrichment of the key proangiogenic factor angiogenin (ANG) in supernatants collected from ABCB5⁺SPs (Figure 4a). Angiogenin is a secreted ribonuclease that plays a key role in stimulating angiogenesis (Cho et al., 2005). This 14.1-kDa monomeric protein constitutes the fifth member of the secreted ribonuclease (RNase) family (Riordan, 2001). The strong expression of angiogenin in the antibody array was confirmed by western blot analysis of cell lysates from ABCB5⁺ SPs when compared to low angiogenin release from dermal fibroblasts (Figure 4b). In addition, macrophage derived chemokine (MDC), monocyte chemotactic protein-4 (MCP4), insulin-like growth factor-binding protein 3 (IGFBP3), tissue inhibitor of metalloproteinases 2 (TIMP2) and osteoprotegerin (OPG) were also induced in the supernatants of ABCB5⁺ SPs (Figure 4a).

Figure 4. Increased release of angiogenin from ABCB5⁺SPs substitutes low levels of angiogenin and downstream effectors in diabetic skin.

(a) Secretome analysis of supernatants from ABCB5⁺SPs by an antibody array under unstimulated culture conditions.

(b) Western blot analysis and quantification of angiogenin in the cell lysates of fibroblasts and ABCB5⁺ SPs.

(c) Western blot analysis of angiogenin and indicated proteins in skin lysates of wild type and db/db mice. Quantification is shown in Supplementary Figure S4a.

(d) Immunostaining of phosphorylated ribosomal protein S6 (pS6RP) (red). Its quantification is shown in Supplementary Figure S4b. Scale bar 50μm.

(e) H&E and immunostaining of angiogenin (red) in human acute, diabetic and chronic venous leg ulcer wound biopsies. Scale bars 100μm. Quantification of angiogenin positive cells is shown in Supplementary Figure S5c.

Since we were interested in the question how suppressed angiogenesis could be restored in diabetic wounds, we selected angiogenin as an angiogenesis stimulating factor to be further studied, though other up-regulated proteins in the supernatant of ABCB5⁺SPs may also be relevant to be explored in future studies. Our choice for angiogenin was further inspired by our data of impaired signaling from dysfunctional endothelial cells. AKT signaling as well as angiogenic VEGFR signaling was suppressed both in co-culture of HUVECs with db/db fibroblasts and in vessels harvested from db/db mice. In fact, angiogenin was earlier reported to stimulate VEGF mRNA levels (Kishimoto et al., 2005), and thus - as we hypothesized - may counteract suppressed VEGF signaling in diabetes. Lysates from diabetic mice skin depicted severely reduced angiogenin expression as compared to wild type mice (Figure 4c and Supplementary Figure S4a). This correlates with suppression of growth signaling, with a severe reduction of activated downstream effectors like pAKT, pS6 ribosomal protein, a component of 40S ribosomal subunit involved in translation, and the translation initiation complex protein elf4G (Figure 4c, 4d, Supplementary Figure S4a and S4b), proteins which, independent of angiogenin, had earlier been shown to be involved in ribosomal biogenesis and protein translation (Holz et al., 2005, Shahbazian et al., 2006, Wang et al., 2001). As expected, our data confirm that angiogenin when supplied to HUVEC cultures stimulates VEGFR2 signaling with phosphorylation of the downstream effector c-Src at Tyr416 position. Angiogenin also enforces growth signaling via phosphorylation of the downstream effector pS6 ribosomal protein (Supplementary Figure S5a and S5b) involved in enhancing translation and protein biosynthesis required for proliferation, growth, and migration during angiogenesis. To specifically address the expression of angiogenin in human diabetic wounds, we performed immunostaining from wound sections (Figure 4e). Interestingly, wound biopsies obtained from patients with chronic diabetic foot ulcers and chronic venous leg ulcers depicted a significant decrease in angiogenin expression when compared to acute human wounds (Figure 4e, Supplementary Figure S5c, and Supplementary Table S1). Though correlative, these data imply that reduced angiogenin expression has been conserved across species suffering from chronic wounds.

Angiogenin released by injected ABCB5⁺ SPs promotes wound healing via enhancing angiogenesis in db/db mice

As angiogenin plays a prime role in the coordination of angiogenesis, and angiogenesis is a prerequisite for sufficient oxygen and nutrient supply to meet the high energy demand during tissue repair, we next focused on dissecting the role of angiogenin released from ABCB5⁺ SPs. The expression of angiogenin in ABCB5⁺ is markedly reduced in diabetic skin wounds (Supplementary Figure S6a). The delivery of high number of ABCB5⁺ SPs with a fresh reservoir of angiogenin (Supplementary Figure S6b) in db/db wounds not only triggers new blood vessel formation, but also significantly accelerates wound closure of otherwise impaired wound closure in diabetic mice.

To study whether angiogenin secreted from ABCB5⁺ SPs is causal for enhanced skin healing of diabetic wounds, we set out to effectively silence angiogenin through siRNA technologies (Supplementary Figure S6c). PBS, 1×10⁶ scrambled siRNA control ABCB5⁺ SPs and angiogenin-silenced ABCB5⁺ SPs were injected into the edges of full-thickness wounds of db/db mice. In addition, we included a group of db/db mice which were repetitively injected with 1×10⁶ ABCB5⁺ SPs at days 0, 5 and 10 during wound healing. Analysis of wounds at days 5 and 7 after wounding revealed significantly delayed healing of wounds that had been injected with angiogenin silenced ABCB5⁺ SPs as opposed to wounds injected with scrambled siRNA transfected ABCB5⁺ SPs (Figure 5a). By contrast, scrambled siRNA transfected ABCB5⁺ SPs significantly accelerated wound closure in db/db mice on days 5, 7 and 10 as compared to the PBS treated db/db mice (Figure 5a). These data indicate that angiogenin is at least in part causal for the therapeutic benefit of ABCB5⁺ SPs accelerating impaired wound healing in diabetic mice, though our data cannot exclude the role of other factors released from ABCB5⁺ SPs.

Figure 5. Silencing of angiogenin abrogates therapeutic beneficial effects of ABCB5⁺ SPs on wound closure in diabetic mice.

(a) Comparison of wound closure after a single injection of PBS (vehicle control) or scrambled ABCB5⁺ SPs or angiogenin silenced (ANGSi)-ABCB5⁺ SPs or multiple injections of ABCB5⁺ SPs into wounds of diabetic db/db mice. One way ANOVA, values are represented as mean ± SEM, n=5.

(b) Representative H&E photomicrographs depicting wound architecture in indicated groups. Scale bars, 100μm.

(c) Immunostaining of the blood vessel specific marker CD31 (red) in wounds from indicated groups. Quantification of CD31 is depicted in Supplementary Figure S6d. Nuclei stained with DAPI (blue). Scale bars, 200μm. E, epidermis; granulation tissue (n=5). Wound edges marked with arrows.

Interestingly, repetitive administrations of ABCB5⁺ SPs also showed improved wound closure in db/db mice on days 7 and day 10 (Figure 5a), but less efficient than single injection of ABCB5⁺ SPs. It is most likely that multiple injections may interfere with and disrupt the sequentially occurring classical phases of wound healing. In fact, histology of wounds repetitively injected with ABCB5⁺ SPs depict a stronger fibrotic response when compared to wounds injected with ABCB5⁺ SPs pretreated with scrambled siRNA or the vehicle control (PBS) (Figure 5b). This suggest that timing of the application is important to carefully orchestrate appropriate responses.

Immunostaining for the endothelial marker CD31 demonstrated that angiogenesis was severely reduced in sections from wounds injected with angiogenin silenced ANG Si-ABCB5⁺ SPs as opposed to enhanced vessel formation in wounds injected with single or multiple injections of scrambled siRNA (Scr) transfected ABCB5⁺ SPs (Figure 5c and Supplementary Figure S6d).

We also confirmed a causal role of angiogenin in the in vitro vessel (tube) formation assay where addition of ABCB5⁺ SPs to HUVECs and db/db MDFs co-cultures enhanced tube formation as opposed to co-cultures of HUVECs only with db/db MDFs (Supplementary Figure S6e). By contrast, ABCB5⁺ SPs silenced for angiogenin failed to rescue the db/db MDF-dependent impaired tube formation (Supplementary Figure S6e). In addition, collagen deposition indicative of the matrix restoration phase of wound healing was studied. Interestingly, we found enhanced collagen deposition indicated by blue colored matrix (Massońs Trichrome staining) and enhanced expression of the myofibroblasts marker α-SMA in scrambled (Scr) ABCB5⁺ SPs injected wounds as compared to PBS injected or ANG Si-ABCB5⁺ injected wounds of db/db mice (Supplementary Figure S7a and S7b). This partial restoration of key matrix proteins and enhanced expression of α-SMA facilitates cellular matrix interaction and supports tissue repair and wound contraction of atrophic diabetic dermis. This, in consequence, counteracts wound infection. These results further strengthen our conclusion for a causal role of angiogenin in ABCB5⁺ SPs-mediated repair of chronic wounds in db/db mice.

In aggregate, our study highlights the therapeutic benefits of ABCB5⁺ SPs for the successful treatment of non-healing wounds in a murine model of human type II diabetes.

Discussion

The major finding of our study is that the injection of ABCB5⁺ SPs in a therapeutic intent markedly enhanced angiogenesis and accelerated overall healing of chronic diabetic wounds. We identified the ribonuclease family member angiogenin secreted from ABCB5⁺ SPs at the wound site as a key mediator enforcing new vessel formation in a paracrine manner thereby restoring impaired wound healing in diabetic mice (graphical summary). This finding is of outstanding clinical relevance given that angiogenesis is severely suppressed in patients suffering from diabetic non-healing wounds.

Our data show that the stromal microenvironment of endothelial cells is essential for angiogenesis, and - if compromised like in diabetes mellitus - leads to suppression of tube formation as the most important prerequisite for angiogenesis. In support of this view, our results show that in skin wounds of db/db mice vessel formation is severely reduced in an overall structurally changed skin. The suppressive impact of db/db mice fibroblasts on vessel formation was highlighted in an in vitro angiogenesis assay. Though the angiogenesis suppressing factors from db/db stroma have not been explored in this study, we found that the endothelial signature in response to their noxious impact shares suppression of VEGF and AKT signaling with the endothelial cell signature of vessels from diabetic mice, and this is also true for upregulated pathways including apoptosis, DNA damage and matrix degradation among others. The number of ABCB5⁺SPs as well as other SP populations including PDGFRβ⁺, Nestin⁺ and CD34⁺ are reduced in the skin of diabetic mice adding to profound changes in the stromal niche of the skin of diabetic mice.

Therefore, local transplantation of precursor cells holds substantial promise to rebalance the endogenous stromal niche and resident cells therein and may even be appropriate to refine treatment of non-healing wounds (Goncalves et al., 2016, Silberstein et al., 2016). Our study uncovered angiogenin released from injected stromal ABCB5⁺ SPs at the wound site to be causal for the impressive rescue of both impaired angiogenesis and the non-healing state of wounds in diabetic mice. Alternatively, approaches such as delivery of angiogenin by either recombinant protein or an expression vector needs to be further assessed in preclinical and clinical settings. Earlier, reduced angiogenin concentrations were found in the serum of patients suffering from diabetes mellitus (Siebert et al., 2010). We here show that angiogenin is also markedly less expressed in human diabetic wounds and in chronic venous leg ulcers. Thus, reduced angiogenin levels are conserved across species suffering from chronic wounds though the role of angiogenin in human wounds needs to be further explored. Accordingly, angiogenin was identified to promote cancer, and this knowledge has propelled the development of anticancer therapies (Shapiro et al., 1987). Mutations and polymorphisms of angiogenin have been linked to Parkinson’s disease (Steidinger et al., 2011), Alzheimer’s disease (Kim and Kim, 2012) and diabetic peripheral neuropathy (Greenway et al., 2006, van Es et al., 2009, Wang et al., 2013).

Classically, after entry into endothelial cells via receptor mediated endocytosis (Ferguson and Subramanian, 2018), angiogenin translocate into the nucleus and subsequently into the nucleolus where it stimulates rRNA transcription, most important for ribosomal biogenesis and the control of transcription of other angiogenic factors and growth factors (Ibaragi et al., 2009, Kishimoto et al., 2005). Interestingly, we provide data that angiogenin not only activates VEGF receptor signaling but also stimulates AKT downstream signaling in HUVECs. Angiogenin released from ABCB5⁺ SPs is vital for angiogenesis in an diabetic microenvironment installed by insulin resistant db/db mice stroma (Shao et al., 2000). This restores endothelial cell function may either be due to unlocking VEGF signaling or due to a receptor-independent direct cytoplasmic signaling. A limitation of the employed xenograft model is that it does not allow to study engraftment and differentiation of injected ABCB5⁺ SPs.

Our data on the successful therapy with ABCB5⁺-derived SPs for the treatment of non-healing diabetic wounds have substantial clinical relevance for the intended implementation into clinical routine. The employment of a single marker strategy (ABCB5⁺) allows the enrichment of an easily accessible homogeneous ABCB5⁺-derived SP population from human skin with GMP grade quality, reliable potency ready to use for clinical routine (Tappenbeck et al., 2019, Vander Beken et al., 2019). Thus, the herein described dermal ABCB5⁺ SPs hold substantial promise for the successful clinical therapy of non-healing diabetic wounds. In fact, clinical phase IIA studies have recently been initiated (EudraCT number: 2015-000399-81, 2017-000233-31, 000234-57) with promising results of the first studied patients suffering from diabetic foot ulcer.

Materials and methods

For detailed methods, see the Supplementary Materials and Methods.

Mice experiments

For wound healing studies 8 to 10 weeks old male db/db diabetic (BKS.Cg-Dock7m^+/+ Lepr^db/J) and wild type (C57BLKS/J) mice were used. Mice were directly purchased from Jackson laboratories and housed under pathogen free conditions at the animal care facility of Ulm University in compliance with the German Law for Welfare of Laboratory Animals. All experiments were approved by the RP Tübingen Baden-Württemberg (approval number TVA1289).

Histology and immunostaining

Histology and immunostainings were performed on 5μm thick wound sections or sections from non-wounded skin biopsies by standard procedures as previously described (Singh et al., 2017). We employed primary antibodies against ABCB5 (TICEBA GmbH and Thermo Fisher), CD34 (e-Bioscience), CD31, PS6RP, PDGFRβ (Cell Signaling) and ANG, αSMA (R&D Systems). Respective isotypes were used as negative controls, and AF488- and AF555-conjugated (Life Technologies) or HRP-conjugated secondary antibodies (Vector stain) were employed. Nuclei were counterstained with DAPI. Following staining, tissues were mounted with fluorescence mounting medium (DAKO). Images were captured by Zeiss Axiophot microscope equipped with an AxioCam digital color camera, Zen 3.2 (ZEN lite) and AxioVision software v4.8 (Carl Zeiss).

Statistical calculations

Error bars represent SEM. The significance of differences between two groups was analyzed by Student’s t test or one-way ANOVA, followed by Bonferroni correction for comparing the difference between more than two groups and presented as absolute p values.

Data availability statement

RNA sequencing analyses data has been deposited in the Gene Expression Omnibus (GEO) with accession number ‘GSE181881’. Additional data related to this paper may be requested from the corresponding author.