Animals

Wild-type (WT) (C57BL/6) and type 2 diabetic (DM2) mice (BKS.Cg-m +/+Lepr ^db ) were purchased from Jackson Laboratories (Bar Harbor, ME, USA) to study the effect of diabetes on ASC physiology. Selected assays were also performed on ASCs isolated from WT mice following induction of type 1 diabetes (DM1) via streptozotocin injection (STZ) (Sigma-Aldrich, St. Louis, MO, USA) as previously described[20]. Only mice with blood glucose levels of greater than 350 mg/dL were considered diabetic and used for further analysis. All protocols were approved by the Stanford Administrative Panel on Laboratory Animal Care.

ASC niche analysis

WT and DM2 murine inguinal fat pads were harvested and manually disrupted for real-time quantitative polymerase chain reaction as described below.

ASC harvest and culture

ASCs were isolated from WT, DM1, and DM2 murine inguinal fat pads, minced, and digested for 1 hour at 37°C using collagenase I (Roche Applied Science, Indianapolis, IN, USA). The reaction was stopped with the addition of supplemented media, pelleted via centrifugation, and subjected to erythrocyte lysis in accordance with the instructions of the manufacturer (Sigma-Aldrich). The remaining cells were pelleted again, forming the stromal vascular fraction (SVF). The SVF was either cultured under standard conditions (37°C in 5% CO₂) in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (Life Technologies, Grand Island, NY, USA) containing 1 g/L of glucose to purify for ASCs or used immediately for flow cytometric/single-cell analyses. Cultured cells were used at or before passage 2, and all analyses were run in triplicate unless otherwise stated.

In vitro Matrigel tubulization assays

PKH26-labeled WT and DM2 ASCs alone or mixed with calcein-labeled human umbilical vein endothelial cells (HUVECs) (Life Technologies) were cultured for 12 hours under hypoxic conditions on a 24-well plate (4 × 10⁴ cells per well) coated with growth factor-reduced Matrigel (BD Biosciences, Franklin Lakes, NJ, USA). ASC and HUVEC tubule counts were determined in five random high-power fields per well, respectively, by using an inverted Leica DMIL microscope (Leica Microsystems, Wetzlar, Germany).

In vivo Matrigel plug assay and CD31 immunohistochemistry

WT or DM2 ASCs (8 × 10⁵) (cultured not more than two passages) were suspended in 250 μL of growth factor-reduced Matrigel (BD Biosciences) and injected in a subcutaneous fashion on the dorsum of WT mice (n = 4). Plugs were harvested at day 10, and 7-μm-thick frozen sections were immunohistochemically stained for the commonly used vascular marker platelet/endothelial cell adhesion molecule 1 (PECAM1/CD31, a transmembrane glycoprotein expressed on the surface of platelets, endothelial cells, and subsets of hematopoietic cells but particularly concentrated at the intercellular junctions of endothelial cells), followed by ImageJ (National Institutes of Health, Bethesda, MD, USA) quantification[7, 21].

In vitro adipogenic differentiation

WT and DM2 ASCs were seeded in standard six-well tissue culture plates (1.5 × 10⁵ cells per well), and adipogenic differentiation medium—consisting of DMEM (1 g/L glucose), 10% fetal bovine serum, 1% penicillin/streptomycin, 10 μg/mL insulin, 1 μM dexamethasone, 0.5 mM methylxanthine, and 200 μM indomethacin—was added after cell attachment. Oil red O staining was performed after 7 days of incubation.

In vitro osteogenic differentiation

WT and DM2 ASCs were seeded in standard six-well tissue culture plates (1.0 × 10⁵ cells per well) and grown to at least 80% confluence before being cultured in osteogenic differentiation medium, which consisted of DMEM (1 g/L glucose) supplemented with 10% FBS, 1% penicillin/streptomycin, 100 μg/mL ascorbic acid, and 10 mM β-glycerophosphate. Photometric quantification of Alizarin red stain was performed after 14 days to assay extracellular mineralization as previously described[22].

In vitro hydrogel bioscaffold seeding

WT and DM2 ASCs (1 × 10⁵) were suspended in 15 μL of growth media and seeded within a previously described 5% collagen-pullulan hydrogel bioscaffold[7, 8]. Seeded scaffolds were placed in growth media and incubated at 37°C in 5% CO₂ prior to proliferation and survival analyses and RNA/protein harvest.

In vitro proliferation and survival

After hydrogel bioscaffold seeding, a live-dead assay (Live/Dead Cell Viability Assay) was performed at multiple time points to assess WT and DM2 ASC viability in accordance with the instructions of the manufacturer (Life Technologies). ASC proliferation was compared between hydrogel-seeded WT and DM2 cells at multiple time points by using an MTT assay (Vybrant MTT Cell Proliferation Assay Kit; Invitrogen, Grand Island, NY, USA).

Real-time quantitative polymerase chain reaction

Total RNA was isolated from ground WT and DM2 fat pads or hydrogel-seeded ASCs by using an RNeasy Mini Kit (Qiagen, Germantown, MD, USA) and transcribed to cDNA (Superscript First-Strand Synthesis Kit; Invitrogen). Real-time quantitative polymerase chain reactions (qPCRs) were performed by using TaqMan gene expression assays (Applied Biosystems, Foster City, CA, USA) for murine Mmp-9 (matrix metalloproteinase 9, Mm00442991_m1), Cxcl-12 (stromal cell-derived factor-1/Sdf-1, Mm00445552_m1), Vegf-a (vascular endothelial growth factor-A, Mm01281447_m1), Eng (endoglin, Mm00468256_m1), Hgf (hepatocyte growth factor, Mm01135193_m1), Mmp-3 (matrix metalloproteinase 3, Mm00440295_m1), Cxcr-4 (chemokine receptor 4, Mm01292123_m1), Fgf-2 (fibroblast growth factor 2, Mm00433287_m1), Fgfr-2 (fibroblast growth factor receptor 2, Mm01269930_m1), Pdgf-a (platelet-derived growth factor-A, Mm01205760_m1), Pdfgr-a (platelet-derived growth factor receptor-A, Mm01205760_m1), and Angpt-1 (angiopoietin 1, Mm00456503_m1) by using a Prism 7900HT Sequence Detection System (Applied Biosystems). Expression levels of the target genes were normalized to Actb (beta actin, Mm01205647_g1) or B2m (beta-2-microglobulin, Mm00437764_m1).

Angiogenesis array

Angiogenic cytokine protein production from hydrogel-seeded WT and DM2 ASCs was quantified by using a Mouse Angiogenesis Array Kit (R&D Systems, Minneapolis, MN, USA). Pixel density of each spot in the array was quantified and normalized to controls by using ImageJ (National Institutes of Health).

In vivo murine ischemia model

A model of graded soft tissue ischemia was created on the dorsum of WT mice, as described previously[23]. Briefly, a full-thickness three-sided peninsular flap was elevated, and a thin silicone sheet was inserted to separate the skin from the underlying fascia. A size-matched hydrogel bioscaffold was inlaid between the silicon sheet and skin flap either alone or following the seeding of 2.5 × 10⁶ WT or diabetic (type 2 or 1) ASCs (cultured not more than two passages) (n = 5). Unseeded scaffolds were used as controls. The skin flap was then sutured into place, and the demarcating necrotic area was quantified 10 days post-op. The proximal surviving flap was processed for neovascular growth via CD31 immunohistochemistry.

Microfluidic single-cell gene expression analysis

ASCs isolated from freshly harvested WT, DM2, and DM1 SVF (obtained as described above) by using the surface marker profile CD45^-/CD31^-/CD34⁺ (to exclude contaminating CD45⁺ hematopoietic and CD31⁺ endothelial cells found within the SVF and to select for a putatively stem-like subset of CD34⁺ cells[22, 24]) were analyzed and sorted as single cells by using a Becton Dickinson FACSAria flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA) into 6 μL of lysis buffer. Propodium iodide exclusion was used to ensure that only live cells were sorted. Reverse transcription and low cycle pre-amplification were performed by using Cells Direct (Invitrogen) with TaqMan assay primer sets (Applied Biosystems) in accordance with the specifications of the manufacturers. cDNA was loaded onto 96.96 Dynamic Arrays (Fluidigm, South San Francisco, CA, USA) for qPCR amplification by using Universal PCR Master Mix (Applied Biosystems) with a uniquely compiled TaqMan assay primer set as previously described[18].

Statistical analysis

Results are presented as mean ± standard error of the mean. Data analysis was performed by using a Student t test. Results were considered significant for P values of not more than 0.05. For single-cell transcriptional data, a Kolmogorov-Smirnov (K-S) test was used to compare empirical distributions, followed by an adaptive fuzzy c-means clustering algorithm as previously described[18].

Diabetes negatively impacts the ASC niche and impairs ASC promotion of neovascularization

To determine the impact of diabetes on the ASC niche in situ, the transcriptional profiles of DM2 and WT murine fat pads were analyzed. Indicative of a dysfunctional signaling environment in situ, the expression of multiple growth factors and cytokines (Vegf-a, Fgf-2, Pdfg-a, and Sdf-1; P <0.02) as well their associated receptors (Cxcr-4, Fgfr-2, and Pdgfr-a; P <0.01) was significantly decreased in the setting of diabetes (Figure 1A).

To next assess the potential of isolated diabetic ASCs to support the generation of vascular structures, we conducted a Matrigel tubule formation assay. When seeded alone, DM2 ASCs displayed significantly fewer tubular structures compared with WT cells (5.2 versus 30.6 tubules per HPF, respectively; P <0.001) (Figure 1B), suggesting a deficient response to extracellular matrix stimuli. ASCs are known to mediate angiogenesis mainly through paracrine mechanisms, however, as opposed to direct endothelial differentiation[25]. To analyze the ability of ASCs to promote endothelial sprouting, WT and DM2 ASCs were co-seeded with HUVEC cells by using the same model. Interestingly, although both ASC groups similarly localized around the HUVECs, consistent with their perivascular nature, the DM2 ASCs supported significantly less HUVEC tubule formation compared with WT ASCs (16.9 versus 27.1 tubules per HPF; P = 0.001) (Figure 1C), indicating that diabetic ASCs possess a reduced stimulatory capacity.

To determine whether the observed in vitro findings translated into impaired in vivo neovascularization, we performed an assay in which Matrigel plugs seeded with DM2 or WT ASCs were implanted into subcutaneous pockets on the dorsum of WT mice, and new vessel formation was histologically assessed after 10 days (a time point at which neovascular processes are active following injury[26]). Consistent with our in vitro findings, plugs containing DM2 ASCs displayed significantly reduced levels of vascularization (as determined by CD31 immunohistochemistry) compared with plugs seeded with WT ASCs (0.38 versus 1.16% CD31 staining per HPF; P = 0.02) (Figure 1D). Further exploring the effect of diabetes on ASC stemness, DM2 ASCs also displayed impairments in adipogenic and osteogenic differentiation in vitro (Figure 2), consistent with a loss of cell stemness and a phenotypic imprinting despite correction of the glucotic environment. Together, these data demonstrate that diabetes significantly impairs the stemness and potential of ASCs to promote neovascularization both in vitro and in vivo.

Optimization of ex vivo environment does not restore diabetic ASC function

The delivery of cells to a wound site is particularly challenging because of a hostile environment characterized by inflammation and poor nutritional support. We have recently developed a biomimetic hydrogel scaffold, which has been shown to support MSC survival, promote ‘stem-like’ features, and improve pro-angiogenic functionality[7, 8]. To further characterize the effect of diabetes on the angiogenic potential of ASCs and to determine whether this function can be restored, we examined the cellular response to seeding in this biomimetic hydrogel. Although WT and DM2 ASCs displayed a similar seeding efficiency and proliferation capacity within the hydrogel (Figure 3), diabetic ASCs possessed a markedly abnormal morphology (Figure 4A). Specifically, DM2 ASCs appeared to be rounded and clumped and failed to form the cytoplasmic extensions seen with WT cells, indicative of a lack of cell interaction with the scaffold microenvironment. Importantly, this phenotypic dysfunction in DM2 ASCs was accompanied by a decrease in transcriptional expression and production of several vasculogenesis-related and tissue remodeling mediators, including VEGF, MMP-9, and HGF (P ≤0.03) (Figure 4B,C).

Diabetic ASCs fail to improve ischemic wound healing in vivo

Diminished perfusion and neovascularization are important factors contributing to impaired diabetic healing; so to evaluate the regenerative potential of diabetic ASCs under representative conditions, cell-seeded biomimetic hydrogels were used in an adapted murine model of skin flap ischemia (Figure 5A,B)[27]. Although the application of WT ASCs showed a significant improvement in tissue survival at day 10 as compared with controls (48.0% versus 22.3% flap survival; P = 0.02), flaps treated with DM2 ASCs developed ischemic tissue necrosis at a level similar to that of the cell-free controls (21.0% versus 22.3% flap survival; P = 0.87) (Figure 5C). This difference in tissue survival correlated with a deficiency of DM2 ASCs in promoting new blood vessel formation within the surviving proximal flap as compared to WT ASCs (1.12% versus 2.34% CD31 staining per HPF; P = 0.04) (Figure 5D,E). This time point is of particular relevance as it is within the window of physiologic neovascularization that occurs following injury but prior to vascular regression that occurs as part of the normal tissue remodeling process[26]. Moreover, these results were recapitulated with ASCs isolated from streptozotocin-induced (DM1) mice, indicating that the observed findings were not model-specific, but rather reflect the impact of prolonged hyperglycemia on ASC physiology (Figure 5C-E). In aggregate, these data confirm the impaired vasculogenic potential of diabetic ASCs in vivo, despite delivery within a biomimetic scaffold.

Diabetes selectively depletes a subpopulation of ASCs with a highly vasculogenic transcriptional profile

Because cell culture and/or surface marker based enrichment of ASCs from the heterogeneous SVF is often employed prior to application, understanding the effect of diabetes on ASC population dynamics in the SVF, as well as following purification, is critical. To first determine the influence of diabetes on global ASC levels within the SVF, a flow cytometric analysis was performed on WT, DM2, and DM1 samples for the presence of putative ASCs (CD45 ^- /CD31 ^- /CD34⁺ cells). Indicative of a detrimental effect of diabetes on the global ASC population, both the DM2 and DM1 SVF displayed a reduced frequency of putative ASCs when compared with WT samples (P <0.01) (Figure 6).

Although this global ASC depletion provides an explanation for the dysfunction of diabetic SVF, a finer degree of granularity is needed to analyze purified ASCs for phenotypic differences, especially as there is evidence that even sorted ASCs consist of several subpopulations with varying levels of functionality[19]. To determine the effect of diabetes on ASC subpopulation dynamics, a single-cell interrogation of WT, DM2, and DM1 ASCs was conducted. For this, we employed microfluidic-based single-cell gene expression analysis as previously described[18]. Transcriptional profiles were simultaneously evaluated for approximately 70 gene targets across 75 individual cells per group, specifically looking at genes relating to stemness, vasculogenesis, and tissue regeneration (Additional file1: Table S1). These cells were primary ASCs isolated based on a primitive surface marker definition (CD45^-/CD31^-/CD34⁺) intended to exclude contaminating hematopoietic and endothelial cells found within the SVF.

When this approach was used, cells isolated from both WT and diabetic mice were found to display significant heterogeneity at the single-cell level (Figure 7A and Additional file2: Figure S1). Moreover, Kolmogorov-Smirnov analysis of these single-cell data identified multiple genes that were differentially expressed in WT versus type 2 or 1 diabetic ASCs, including genes related to tissue remodeling, vasculogenesis, and cell differentiation, such as the metalloproteinases Mmp-3 and Adam-10, the chemokine Ccl-2, and the transcription factors Hif-1a and Mef-2c (P <0.01) (Figure 7B)[28–32]. To determine whether these differences were the result of changes in specific subpopulations of diabetic or WT ASCs, partitional clustering was applied to the super-set of transcriptional profiles encompassing WT and diabetic cells[18]. This analysis identified three distinct transcriptionally defined ASC subpopulations, which exhibited strikingly different profiles across conditions (Figure 7C-E). Interestingly, a distribution analysis of the cells comprising the third cluster revealed that there were considerably fewer diabetic ASCs compared with WT cells in this subpopulation (Figure 7 F), with this cluster characterized in part by the elevated expression of genes associated with vasculogenesis, such as Angpt-1, Sdf-1, Mmp-3, and Fzd-4 (Figure 7G)[29, 33–35]. These data suggest that this subpopulation may be primed to support vasculogenesis, and its depletion in the diabetic state provides a potential mechanism for the impairment of diabetic ASC vasculogenic potential.