Introduction

The United States is facing a rapid increase in its elderly population. According to U.S. Census Bureau projections, the number of Americans age sixty-five and older is expected to more than double between 2012 and 20601. Concomitant with an aging population is an increased risk for vasculopathies, including cardiovascular disease, peripheral vascular disease and impaired wound healing2,3. One of the main underlying causes for this is an impairment of new blood vessel formation, or neovascularization4,5. Progenitor cell populations, such as mesenchymal stem cells (MSCs), have been shown to promote neovascularization following therapeutic application through the secretion of pro-vascular cytokines6,7.

Unfortunately, aging is known to have a negative impact on the regenerative capacities of most tissues8 and recent evidence suggests that stem cells are also susceptible to biologic aging9,10. Aged MSCs in particular have been shown to possess impairments in proliferation and differentiation capacity9,10,11, as well as alterations in therapeutic gene expression and cytokine production12,13. However, the mechanisms underlying control of the aging process in human cells remain poorly understood14, limiting the development of cell based therapies.

Only in the last few years have high-resolution measurement tools evolved to interrogate heterogeneous cell populations like mesenchymal progenitors with sufficient precision to detect potentially subtle population shifts15. Our laboratory has recently developed a novel method that combines high throughput single cell gene expression analysis with complex mathematical modeling to identify critical perturbations in cell subpopulations16. Applying this approach to study diabetes, a condition also associated with a global breakdown of reparative processes, we identified critical disruptions in “progenitor cell ecology” which were responsible for impairments in the formation of new blood vessels in response to ischemia (neovascularization)17,18. Given that aging is likewise associated with significant neovascular impairments19,4 and that vascular disease is the leading cause of death in the aged population20, we sought to evaluate the cellular dynamics of aging in this context.

Wound healing is a common and easily measured indicator of impaired neovascularization and is clinically impaired in aging21. Progenitor cell-based therapy using allogenic or autologous cells applied to the wound site represent an attractive strategy to treat impaired wound healing in advanced age. Here we evaluate the direct effect of aging on MSC functionality, specifically determining the ability of these cells to support vascular network formation in vitro and in vivo. Building on previous work demonstrating that MSCs accelerate wound vascularization and closure6,7, we also sought to determine whether the delivery of aged MSCs would similarly improve wound healing, or whether these cells were impaired and thus limited in their potential clinical effectiveness.

Results

Aging does not influence MSC frequency, viability, or proliferative capacity

We first assessed whether aging affected the MSC phenotype. Consistent with previous studies22,23, the frequency of MSCs within adipose tissue (as determined by the percentage of CD45-/CD31-/CD34+ cells within the SVF) was unaffected by age (Figure 1A–B). Furthermore, aging had no effect on adipose derived mesenchymal stem cell (ASC) viability and proliferation following hydrogel seeding in vitro (Figure 1C–D). Because these population-level phenotypic similarities did not explain the signaling and functional deficiencies associated with aged progenitor cells13, we next analyzed ASC subpopulation dynamics via single cell interrogation of young and aged cells.

Figure 1
figure 1

Assessment of age on ASC phenotype.

(A) Flow cytometric analysis determining the % of CD45- cells (top row) and ASCs (CD45-/31-/34+ cells; bottom row) within the SVF of young and aged mice. (B) Quantification of CD45-/31-/34+ ASCs in young and aged SVF reveals no significant difference in ASC frequency across samples. (C,D) Live/dead and MTT proliferation assays revealed no significant differences in the viability or proliferation of young and aged ASCs following in vitro hydrogel seeding.

Full size image

Aging selectively depletes a putatively vasculogenic cell subpopulation

Utilizing a previously described microfluidic-based single-cell gene expression platform16, the transcriptional profiles of 75 individual cells per group were simultaneously evaluated for approximately 70 gene targets related to stemness, vasculogenesis and tissue regeneration (Supplemental Table 1). In this analysis, ASCs isolated from both young and aged mice displayed significant heterogeneity at the single-cell level (Figure 2A–B). Differences in the transcriptional profiles of genes related to cell stemness, vasculogenesis and tissue remodeling, such as the metalloproteinase Adam10, the chemokines Angpt1 and 2 and the transcription factors Hif1a and Mef2c24,25,26,27,28, were also observed (Figure 2C). Additionally, Kolmogorov-Smirnov analysis of these single-cell data confirmed differential expression of the anti-oxidative enzyme Sod2 in aged versus young ASCs (p < 0.01).

Figure 2
figure 2

Single cell transcriptional analysis of young and aged ASCs.

(A) Hierarchical clustering of cells from young (left) and aged (right) mice. Gene expression is presented as fold change from median on a color scale from yellow (high expression, 32-fold above median) to blue (low expression, 32-fold below median). (B) Whisker plots presenting raw qPCR cycle threshold values for each gene across all young and aged ASCs. Individual dots represent single gene/cell qPCR reactions, with increased cycle threshold values corresponding to decreased mRNA content. Cycle threshold values of 40 were assigned to all reactions that failed to achieve detectable levels of amplification within 40 qPCR cycles. (C) Median-centered Gaussian fit curves of selected genes relating to cell stemness and vasculogenesis displaying grossly differential expression profiles between young and aged cells. Non-parametric two-sample Kolmogorov-Smirnov analysis confirmed the differential expression of the anti-oxidative enzyme Sod-2 in aged versus young ASCs (p < 0.01). The left bar for each panel represents the fraction of qPCR reactions that failed to amplify in each group. (D,E) Partitional clustering of young and aged cells based on the expression patterns of all 71 genes (k = 2). (F) Pie charts representing the fraction of ASCs comprising each cluster (Young [red], Aged [green]), with selected cluster 1-defining genes listed below.

Full size image

To further examine this niche, the super-set of transcriptional profiles of aged and young cells was subjected to a partitional clustering algorithm16. This analysis identified two distinct transcriptionally defined ASC clusters in each group, with the first cluster possessing considerably fewer aged cells (Figure 2D–F). Critically, this subpopulation was characterized in part by the increased expression of genes associated with stemness, tissue remodeling and vasculogenesis, such as Mef2c, Hif1a, Adam10, Ccl2 and Cxcl1224,25,26,29,30 (Figure 2F, Supplemental Table 2), suggesting it may represent a more pluripotent subset of this heterogeneous cell collection.

Canonical pathways whose expression was significantly upregulated in this cluster were determined using Ingenuity Pathway Analysis (IPA). These included multiple cardiovascular processes, including development, differentiation and hypoxic signaling, as well as the degradation of superoxide radicals, lending additional support to a pro-vasculogenic role for these cells (Figure 3A). Furthermore, the top molecular network associated with these genes based on the Ingenuity Knowledge Base appears to link key mediators of neovascularization (HIF1a, VEGF and CXCL12)31,32 and wound healing (Ccl2, P38, MAPK and IL1)33,34,35,36 (Figure 3B). Collectively, these data support the depletion of a putatively vasculogenic subpopulation of ASCs as a potential mechanism underlying age-related impairments in ASC regenerative potential, particularly in the context of ischemic wound healing.

Figure 3
figure 3

Pathway analysis of a pro-vascular ASC subpopulation.

(A) Canonical pathways enriched for genes whose expression was significantly increased in cells comprising Cluster 1 in Figure 2C–D. (B) Top scoring Ingenuity Pathway Analysis (IPA)-constructed transcriptome network based on genes that were significantly increased in Cluster 1 cells. These significant “seed” genes are colored in red to distinguish them from the remaining “inferred” entities in the network.

Full size image

Aging disrupts MSC signaling and angiogenic potential

To investigate whether the subpopulation differences discovered at the single cell level were artifactual, we assessed the effect of age on the ASC in situ environment. Consistent with an age-related signaling dysfunction in this setting, the expression of multiple growth factors (Ffg2, Pdgfa, p < 0.05), as well as their receptors (Fgfr2, Pdgfra, p < 0.01), was diminished in aged adipose tissue (Figure 4A). Similar negative effects on paracrine signaling could be detected in isolated aged ASCs seeded within hydrogel bioscaffolds in vitro. Specifically, aged ASCs displayed decreased transcriptional expression and production of several vasculogenesis-related and anti-oxidative molecules, such as Angpt1, Vegfa and Sod3 (p < 0.05) (Figure 4B–C).

Figure 4
figure 4

Analysis of ASC neovascular potential.

(A) Transcriptional profiles of murine fat pads exhibit dysfunctional in situ signaling in aged samples. (B–C) RT-PCR and ELISA quantifying the relative expression of selected vasculogenic and anti-oxidant genes in young versus aged ASCs in vitro. (D) Matrigel co-culture of ASCs and HUVECs under hypoxic conditions in vitro. (E) In vivo aged and young ASC-seeded matrigel plugs harvested on day twelve stained for the endothelial cell marker CD31. Scale bar = 25 μm. ** indicates p ≤ 0.01; * indicates p ≤ 0.05.

Full size image

Given the significant signaling disruption observed in aged samples, we next sought to directly examine the potential of aged ASCs to support vasculogenesis via cytokine signaling in vitro and in vivo. To analyze the ability of ASCs to promote endothelial cell sprouting (an in vitro surrogate for vascular formation), aged and young ASCs were co-cultured with HUVEC cells on matrigel under hypoxic conditions. Indicative of a reduced cytokine stimulatory capacity with aging, young ASCs supported significantly greater HUVEC tubule formation than their aged counterparts (11.4 vs. 3.1 tubules/HPF, p < 0.01) (Figure 4D).

To confirm that the in vitro vasculogenic impairments in aged ASCs were also present in vivo, a matrigel plug vascularization assay was performed37. Interestingly, while ASC survival was not significantly different across groups (Supplemental Figure 1), consistent with our in vitro findings plugs containing aged ASCs were significantly less vascularized (0.02 vs 0.12% CD31 staining/HPF, p < 0.05) (Figure 4E). Together, these data demonstrate that aging significantly impairs the potential of ASCs to promote neovascularization both in vitro and in vivo.

Aged MSCs are unable to improve cutaneous wound healing in aged mice

To evaluate the translational potential of autologous aged ASCs for the promotion of wound healing, cell-seeded hydrogels were applied to aged murine excisional cutaneous wounds utilizing a previously established model38. While wounds treated with young ASC-seeded hydrogels showed significantly improved healing rates as early as day six (Figure 5A–B) and resulted in significantly faster wound closure times as compared to controls (14.0 vs 15.8 days, p < 0.05), aged ASCs failed to enhance wound healing rates (Figure 5B) or closure times (15.4 vs. 15.8 days, p = 0.89) (Figure 5C).

Figure 5
figure 5

Application of young but not aged ASCs improves excisional wound healing.

(A,B) Gross appearance and area measurements of humanized excisional murine wounds treated with aged and young ASCs. Application of young ASCs resulted in accelerated wound healing from day six on compared to wounds treated with aged ASCs or no cells. (C) Application of young ASCs resulted in an accelerated time to wound closure compared to wounds treated with aged ASCs or no cells. * indicates p ≤ 0.05.

Full size image

Aged MSCs fail to enhance wound vascularity or modulate wound anti-oxidative or vasculogenic profiles

To better characterize the therapeutic dysfunction of aged ASCs in vivo, immunohistochemical staining of day four wounds was performed for the anti-oxidative and pro-vasculogenic molecules SOD-3 and VEGF. Diminished levels of both SOD-3 (Figure 6A) and VEGF (Figure 6B) were found in wounds treated with aged versus young ASCs, with the aged cells displaying a therapeutic efficacy similar to that of the no cell control. Consistent with this signaling dysfunction, healed wounds in the aged ASC treatment group displayed significantly less neovascularization (0.15 vs. 0.52% CD31 staining/HPF, p < 0.01) (Figure 6C), with the aged ASC group again showing no significant increase over acellular controls. These data further underscore the significance of the impaired regenerative potential of aged ASCs in vivo.

Figure 6
figure 6

Immunohistochemical assessment of wound protein expression and vascularity following ASC treatment.

(A,B) Representative images and quantification of day four wounds demonstrating significantly decreased levels of the anti-oxidative enzyme SOD-3 and the vasculogenic cytokine VEGF in wounds treated with aged versus young ASCs. (C) Immunohistochemical staining of closed wounds for the endothelial cell marker CD31 revealed a significant increase in vascular density in wounds treated with young ASCs as compared to aged ASC and no cell controls. Scale bar = 25 μm. ** indicates p ≤ 0.01; * indicates p ≤ 0.05.

Full size image

Discussion

In this study, we identified age-related alterations in MSC subpopulation dynamics at a single cell level that may explain the reduced vasculogenic potential of aged MSCs. These findings are congruent with the concept that complex cell populations such as progenitor cells are characterized by a level of functional heterogeneity that may be critical to their biological role and that the selective loss of different functional cell populations and thereby complexity may be one of the fundamental mechanisms of aging. This idea is supported by some theoretical work in the aging literature39,40 as well as evidence that certain stem/progenitor cell populations disappear during the aging process41,42,43,44,45,46,47,48,49,50.

Studying the effects of aging on progenitor cell populations is difficult, however, as traditional population-level approaches rely on pooled RNA or protein from hundreds of thousands of cells and are thus unable to detect differential expression among rare cells and/or subgroups. Only recently have high-throughput techniques evolved to interrogate samples with single cell resolution. These systems make use of microfluidic technology to achieve massively parallel single-cell gene expression analysis, the resulting data from which can provide novel insights into the relationships among cells in complex tissues51,52,53. Our laboratory has developed a novel method that combines single cell transcriptional analysis with advanced mathematical modeling to characterize heterogeneity in putatively homogeneous populations, as well as identify critical perturbations in cell subpopulations (cellular ecology) under pathologic conditions16,54,55,56,57. Utilizing this methodology, we have previously demonstrated that impaired neovascularization is linked to selective depletion of progenitor cell populations in diabetes, a disease associated with many of the same vasculopathies found in aging17,18. By evaluating whether a similar disruption in progenitor cell populations is associated with impaired neovascularization in advanced age, we set out to determine the clinical potential of this cell source, as well as provide fundamental insights into the effect of aging on progenitor cells.

Highlighting the value of this approach, age did not have an effect on ASC frequency or population-level phenotype that would explain their reduced vasculogenic potential12,13. This finding is consistent with previous reports of ASC cell-surface markers remaining intact with aging58. Conversely, high-resolution transcriptional analysis identified age-related differences in key genes involved in ischemic neovascularization, including the hypoxic transcription factor Hif1a and chemokine Cxcl12 (Sdf1), as well as the anti-oxidative enzyme Sod2. When subjected to partitional clustering, aged animals also had significantly fewer cells in the subcluster defined by vasculogenesis-related genes, such as Cxcl12, Hif1a, Sod2 and Ccl2. These results suggest that deficits in the neovascular potential of aged progenitor cells may be the result of selective subpopulation depletion.

Consistent with these single-cell data, we found that while young ASCs improved wound healing in aged mice, this effect was absent when using ASCs isolated from aged mice. Mechanistically, this decrease in ASC in vivo therapeutic efficacy was likely due to the depletion of a functional subset of ASCs expressing anti-oxidative and pro-regenerative cytokines, as multiple pro-regenerative pathways within the wound that were upregulated with application of young ASCs were not affected following application of aged cells. These data support a decrease in progenitor cell heterogeneity and removal of critical cell subsets as potential key factors leading to the loss of cellular complexity and tissue homeostasis in aging, which would have considerable implications for new diagnostic and therapeutic approaches in this population. Specifically, adding back the depleted cellular subpopulations (as was done with the utilization of young cells in this work) may be an effective strategy to combat not only impaired wound healing, but also other age related sequelae.

Taken together, these findings demonstrate that MSCs are susceptible to biological aging that results in impairments in their regenerative capacity. Single-cell transcriptional interrogation of these cells suggests that depletion of a pro-vasculogenic subpopulation may underlie many of the well-known consequences of aging such as neovascular dysfunction. Given the significant healthcare burden associated with age-related morbidity and the limitations in current therapeutic approaches grounded in existing theories of aging59,60, this more quantitative understanding of how aging causes progenitor cell dysfunction suggests the need for novel therapeutic and diagnostic tools addressing age-associated impairments in cellular heterogeneity.

Methods

Animals

Young (3 months) and aged (21 months) wild-type (WT, C57BL/6) mice were obtained from the National Institute on Aging (NIA, Bethesda, MD). All protocols were approved by the Stanford Administrative Panel on Laboratory Animal Care and all experiments were performed in accordance with relevant guidelines and regulations.

ASC in situ analysis

Young and aged murine inguinal fat pads were harvested and manually disrupted for real-time quantitative PCR (RT-PCR) as described below.

ASC harvest and culture

ASCs were isolated from young and aged murine inguinal fat pads, minced and digested for one hour at 37°C using collagenase I (Roche Applied Science, Indianapolis, IN). After centrifugation, the pelleted stromal vascular fraction (SVF) was cultured in supplemented DMEM. Cells were used at or before passage two and all analyses were run in triplicate unless otherwise stated.

In vitro HUVEC matrigel tubulization assay

2 × 104 PKH26-labeled young or aged ASCs were mixed with 2 × 104 calcein-labeled human umbilical vein endothelial cells (HUVECs) (Life Technologies, Grand Island, NY) and cultured for 12 hours under hypoxic conditions on a 24-well plate coated with growth factor reduced Matrigel (BD Biosciences, Franklin Lakes, NJ). HUVEC tubule counts were determined in 5 random high-power fields per well using an inverted Leica DMIL microscope.

In vivo matrigel plug assay

8 × 105 PKH26-labeled young or aged ASCs were suspended in 250 µl of growth factor reduced Matrigel (BD Biosciences) and injected in a subcutaneous fashion on the dorsum of 8-12 week old WT mice. Plugs were harvested at day 12, embedded in OCT (Sakura Finetek USA, Inc., Torrance CA), sectioned and immunohistochemically stained for CD31 as described below. ASC cell density was determined based on cell counts per high power field.

In vitro hydrogel bioscaffold seeding

1 × 105 ASCs were suspended in 15 µl of growth media and seeded within a 5% collagen-pullulan hydrogel bioscaffold as previously described6,7. Seeded scaffolds were placed in growth media and incubated at 37°C in 5% CO2 prior to proliferation and survival analyses, or RNA/protein harvest.

In vitro viability and proliferation

A live-dead assay was performed to assess ASC viability at days 1, 3 and 8 following hydrogel seeding, according to manufacturer's instructions (Live/Dead Cell Viability Assay, Life Technologies, Grand Island, NY). ASC proliferation was assessed at days 1, 3 and 8 following hydrogel seeding using an MTT assay (Vybrant MTT Cell Proliferation Assay Kit, Invitrogen, Grand Island, NY).

Real-time quantitative PCR (RT-PCR)

Total RNA was isolated from ground fat pads or plated/hydrogel-seeded ASCs using the RNeasy Mini Kit (Qiagen, Germantown, MD) and transcribed to cDNA (Superscript First-Strand Synthesis Kit, Invitrogen, Grand Island, NY). Real-time qPCR reactions were performed using Taqman gene expression assays (Applied Biosystems, Foster City, CA) for murine Angpt1 (Angiopoietin 1, Mm00456503_m1), Vegfa (Vascular endothelial growth factor-A, Mm01281447_m1), Sod3 (Superoxide dismutase 3, Mm01213380_s1), Sod2 (Superoxide dismutase 2, Mm00449726_m1), Angpt2 (Angiopoitin 1, Mm00545822), Fgf2 (Fibroblast growth factor 2, Mm00433287_m1), Fgfr2 (Fibroblast growth factor receptor 2, Mm01269930_m1), Pdgfa (Platelet derived growth factor-A, Mm01205760_m1) and Pdfgra (Platelet derived growth factor receptor-A, Mm01205760_m1) 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). See supplemental table 3 for overview.

In vitro cytokine quantification

Total protein was collected from ASC-seeded hydrogels using RIPA buffer (Sigma-Aldrich) in combination with a protease inhibitor. Protein levels of VEGF and SOD3 were quantified using murine ELISA kits (R&D Systems, Minneapolis, MN and USCN, Wuhan, China).

In vivo excisional wound model

Twenty-one month old male C57Bl/6 mice were randomized into three treatment groups: young or aged murine ASC-seeded hydrogel and unseeded hydrogel control (n = 5 per group). As previously described38, two 6 mm full-thickness cutaneous wounds were excised on either side of the midline, with each wound stented with silicone rings sutured in place to prevent wound contraction. For mice in the unseeded hydrogel group, a 6 mm piece of hydrogel saturated with PBS was placed in each wound bed. For mice in the ASC-seeded hydrogel groups, a 6 mm piece of hydrogel seeded with 2 × 105 young or aged ASCs was placed in the wound bed following creation. All wounds were covered with an occlusive dressing (Tegaderm, 3M, St. Paul, MN). Digital photographs were taken on day 0, 4, 6, 8, 11, 13, 15 and 18. Wound area was measured using Image J software (NIH, Bethesda, MD).

Immunohistochemistry (IHC)

Wounds from the excisional model were harvested upon closure and immediately embedded in OCT (Sakura Finetek USA, Inc.). To assess vasculature in healed wounds and matrigel plug samples, seven micron thick frozen sections were immunohistochemically stained for CD31 (1° - 1:100 Rb α CD31, Ab28364, Abcam, 2° - 1:400 AF547 or AF488 Gt α Rb, Life Technologies). To assess in vivo signaling during wound healing, day four wounds from a separate cohort of mice were similarly harvested and subjected to immunohistochemical staining for SOD3 (1° - 1:100 Rb α SOD3, Ab21974, Abcam; 2° - 1:200 AF547 Gt α Rb, Life Technologies) and VEGF (1° - 1:100 Rb α VEGF, Ab1316, Abcam, Cambridge, MA; 2° - 1:200 AF547 Gt α Rb, Life Technologies). For all IHC analyses, nuclei were stained with DAPI and Image J (NIH) was used to binarize images taken with the same settings. Intensity thresholds were used to quantify CD31 staining based upon pixel-positive area per high power field.

Microfluidic single-cell gene expression analysis

Young and aged primary ASCs with the surface marker profile CD45-/CD31-/CD34+ (to exclude contaminating hematopoietic and endothelial cells found within the SVF) were sorted as single cells using a Becton Dickinson FACSAria flow cytometer into 6 µl of lysis buffer. Reverse transcription and low cycle pre-amplification were performed using Cells Direct (Invitrogen) with Taqman assay primer sets (Applied Biosystems) as per the manufacturers specifications. cDNA was loaded onto 96.96 Dynamic Arrays (Fluidigm, South San Francisco, CA) for qPCR amplification using Universal PCR Master Mix (Applied Biosystems) with a uniquely compiled Taqman assay primer set as previously described55.

Statistical analysis

Results are presented as mean ± standard error of the mean (SEM). Data analysis was performed using a Student's t-test. Results were considered significant for p ≤ 0.05. For the single cell gene expression analysis, a two sample Kolmogorov-Smirnov (K-S) test was used to compare empirical distributions, using a strict cutoff of p<0.01 following Bonferroni correction for multiple samples. Transcriptionally defined subpopulations were determined using an adaptive fuzzy c-means clustering algorithm employing a standard Euclidean distance metric, as previously described55. Each cell was assigned partial membership to each cluster based on similarities in expression profiles, with optimally partitioned clusters flattened and sub-grouped using hierarchical clustering in order to facilitate visualization of data patterning within and across these clusters. Canonical pathway calculations and network analyses were performed using Ingenuity Pathway Analysis (IPA, Ingenuity Systems, Redwood City, CA) based on genes differentially expressed among cell subpopulations.