CD34+CD146+ adipose‐derived stromal cells enhance engraftment of transplanted fat

Abstract Fat grafting is a surgical technique able to reconstruct and regenerate soft tissue. The adipose‐derived stromal cells (ASCs) within the stromal vascular fraction are believed to drive these beneficial effects. ASCs are increasingly recognized to be a heterogeneous group, comprised of multiple stem and progenitor subpopulations with distinct functions. We hypothesized the existence of an ASC subpopulation with enhanced angiogenic potential. Human ASCs that were CD34+CD146+, CD34+CD146−, or CD34+ unfractionated (UF) were isolated by flow cytometry for comparison of expression of proangiogenic factors and endothelial tube‐forming potential. Next, lipoaspirate was enriched with either CD34+CD146+, CD34+CD146−, CD34+ UF ASCs, or was not enriched, and grafted beneath the scalp skin of immunodeficient CD‐1 Nude mice (10 000 cells/200 μL/graft). Fat retention was monitored radiographically more than 8 weeks and fat grafts were harvested for histological assessment of quality and vascularization. The CD34+CD146+ subpopulation comprised ~30% of ASCs, and exhibited increased expression of vascular endothelial growth factor and angiopoietin‐1 compared to CD34+CD146− and CD34+ UF ASCs, and increased expression of fibroblast growth factor‐2 compared to CD34+CD146− ASCs. The CD34+CD146+ subpopulation exhibited enhanced induction of tube‐formation compared to CD34+CD146− ASCs. Upon transplantation, fat enriched CD34+CD146+ ASCs underwent less resorption and had improved histologic quality and vascularization. We have identified a subpopulation of CD34+ ASCs with enhanced angiogenic effects in vitro and in vivo, likely mediated by increased expression of potent proangiogenic factors. These findings suggest that enriching lipoaspirate with CD34+CD146+ ASCs may enhance fat graft vascularization and retention in the clinical setting.


| INTRODUCTION
Autologous fat grafting (AFG) is an increasingly popular reconstructive technique; more than 75 000 AFG procedures were performed in 2018 in the United States alone. 1 Initially popularized for its ability to restore soft tissue volume and soften contour deformities, fat has become increasingly appreciated for its regenerative potential. Today AFG is also used to improve the quality of scarred skin, facilitate wound closure, and reverse the effects of radiation-induced soft tissue damage. 2,3 The adipose-derived stromal cells (ASCs) within the stromal vascular fraction (SVF) of fat are thought to drive these regenerative effects. 4 Evidence in support for this hypothesis has stemmed from several clinical and preclinical studies that demonstrate, compared to fat alone, grafts enriched with ASCs, also known as cellassisted lipotransfer, undergo less resorption and have improved vascularization and histological structure. [5][6][7] Recent work has indicated that ASCs are a heterogeneous population comprised of distinct subpopulations of cells with differing functional capacities; bone morphogenetic protein receptor (BMPR)-1A identifies ASCs with proadipogenic qualities 8 and low expression of CD105 (endoglin) at the cell surface identifies ASCs with enhanced osteogenic capacity. 9 Further defining ASC heterogeneity may improve the efficiency of current fat grafting procedures. Given that grafted fat often undergoes significant resorption, with retention rates ranging from 25% to 80%, 10 and studies suggesting a lack of blood supply as a contributing factor, enriching grafted fat with ASCs exhibiting proangiogenic effects may promote early revascularization and thus better retention. Interestingly, a recent report identified CD146+ mesenchymal cells isolated from human umbilical cords that possessed enhanced ability to promote blood vessel maturation. 11 We therefore hypothesized a similar existence of a subpopulation of ASCs with enhanced angiogenic potential that can increase the viability of fat grafts.

| Human adipose-derived SVF harvest
Fresh human lipoaspirate was obtained from five healthy female donors with no medical comorbidities (age range: 18-62 years) under the Stanford Institutional Review Board approval (IRB: 2188). The fat was washed with phosphate-buffered saline (PBS, Thermo Fisher Scientific, Waltham, Massachusetts, Cat#10010023) and allowed to settle for 30 minutes at 4 C to separate into layers of lipid, fat, and blood. The layer of fat was retrieved with a serological pipette for both isolation of ASCs and for fat grafting. For the fat grafting experiment, a single source of fat was used; 4 mL of fat was set aside for grafting and the remaining fat was digested. The strategy to isolate CD34+CD146+ ASCs was adapted from methodology previously described. 12 (Figures 1 and S1). [15][16][17] Cells were sorted using a 100 μm

Significance statement
This study has identified a subpopulation of adipose-derived stromal cells (ASCs), positive for the surface marker CD146, that have increased expression of proangiogenic genes and enhanced capacity to induce endothelial-tube formation. nozzle, and a flow rate of 1, with the "purity" setting. Purity was tested before sorting by analyzing 2000 freshly sorted cells and was found to be >99% for each of the three ASC subpopulations.

| Angiogenic gene expression
To confirm proangiogenic gene expression, ASC subpopulations were freshly sorted directly into TRIzol (Invitrogen, Carlsbad, California, Cat#15596026). RNA was extracted using the RNeasy Mini Kit (Quiagen; Hilden, Germany, Cat#74104). Reverse transcription was performed using Reverse Transcription Reagents (Invitrogen). Quantitative real-time polymerase chain reaction (qRT-PCR) was performed using an ABI Prism

| Animals
Adult female 60-day-old CD-1 Nude immunocompromised mice (Crl: CD1-Foxn1nu, Charles River) were used for experimentation (total n = 32). Mice were maintained at the Stanford University Research Animal Facility (5 animals/cage) in sterile microinsulators and were given water and rodent chow ad libitum, in accordance with Stanford University guidelines. All experiments were performed in accordance with the Stanford University Animal Care and Use Committee Guidelines under an approved APLAC protocol (APLAC-31212).

| CT scans and reconstructions
Micro-computed tomography (microCT) imaging was performed using a Bruker Skyscan 1276 microCT (Bruker), as previously described. 22 Immediately following fat grafting, mice were imaged to determine baseline fat graft volume. Thereafter, serial imaging was performed every 2 weeks for a total of 8 weeks postgrafting. Three-dimensional reconstructions were performed using cubic-spline interpolation to determine fat graft volume as a percentage of the original transplanted volume. 5 All reconstructions were performed blinded by two investigators (M. R. B. and S. V.) and the mean of each score was calculated. F I G U R E 3 Gene expression, protein expression, and endothelial tube forming assays. A, Expression of three potent proangiogenic genes in CD34+CD146+, CD34+CD146−, and CD34+ UF ASCs was compared by PCR. CD34+CD146+ ASCs expressed significantly more VEGF (***P < .001) and ANGTP1 (***P < .001) than CD34+CD146− and CD34+ UF ASCs, and significantly greater levels of FGF than CD34+CD146− ASCs (**P < .01). B, Representative immunofluorescence images (top row) and binarized images from ImageJ analysis (bottom row) showing results from endothelial tube forming assays. Human microvascular endothelial cells (green due to staining with Calcein AM) were cocultured for 18 hours in transwells with CD34+CD46+ ASCs (left), CD34+CD146− ASCs (middle), and CD34+ UF ASCs (right). Scale bars = 100 μm. C, Graphs showing results from endothelial tube forming assay analysis demonstrating, C(i), greater total pixel density, C(ii), tube length, and, C(iii), total branching length in HMECs cocultured with CD34+CD146+ compared to CD34+CD146− ASCs (all *P ≤ .05). D, Graphs showing the results from enzyme-linked immunosorbent assays comparing protein expression between three ASC subpopulations. CD34+CD146+ ASCs expressed significantly more VEGF (****P < .0001) (D[i]) and ANGTP1 (****P < .0001) (D[ii]) than both CD34+CD146− and CD34+ UF ASCs, as well as significantly greater levels of FGF than CD34+CD146− ASCs (***P < .001) and CD34+ UF ASCs (****P < .0001) (D[iii]). ANGPT1, angiopoietin-1; ASC, adipose-derived stromal cell; FGF, fibroblast growth factor-2; HMEC, human microvascular endothelial cell; PCR, polymerase chain reaction; UF, unfractionated; VEFG, vascular endothelial growth factor 52) to visualize cell nuclei. Fluorescent images were captured with laser scanning confocal microscopy using a Leica TCS SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany). A uniform frame size of 1024 × 1024 was used with the ×25 or ×63 oil objectives. Quantification of fat graft vascularization was performed on fat grafts from five mice per group. To ensure large areas of the fat graft were assessed, tiled (3 × 3) z-stacked (8 μm deep) fluorescent images were taken. To quantify graft vascularization, the percent of pixels positive for CD31 staining was calculated for five representative regions of interest (ROIs) of equal area using ImageJ software (https://imagej.nih.gov/ij/) following methods previously described. [24][25][26][27][28] A threshold was used to select pixels occupied by blood vessels, represented by CD31 staining, and the image was binarized by converting the blood vessels to white (pixel value 255) and background pixels to black (pixel value 0) to form a "mask" of positive CD31 staining. The mean number of pixels positive for CD31 per area was calculated, and the average of the five ROIs per image were calculated to give one value per image.

| Statistical analysis
Continuous data were described using the mean and SD of the mean when parametric, and with the median and the range when nonparametric. Data were reported as frequencies when categorical. Analysis of variance and Bonferroni multiple comparisons were used to compare means between groups. A P value of <.05 was considered significant. All statistical analyses were performed using Prism GraphPad 5.0 (GraphPad Software, Inc., La Jolla, California) statistical software.

| Enriching fat grafts with CD34+CD146+ improves retention
Fat retention was monitored radiographically by Micro-CT every 2 weeks for 8 weeks in total. Results indicated that fat

| Improved histological quality of fat grafts enriched with CD146+ ASCs
H&E-stained sections of explanted grafts from the four groups of mice exhibited different histological characteristics 8 weeks F I G U R E 6 Vascularization of explanted fat grafts 8 weeks postgrafting. A, Immunofluorescence staining for endothelium using CD31 (green) in fat grafts enriched with CD34+CD146+ (far left), CD34+CD146− (middle left), CD34+ UF (middle right) ASCs, or not enriched alone ("fat only," far right) shown at low magnification with DAPI (blue) (top row), with CD31 alone (middle row), and of the selected ROI (white dotted box) at high magnification (bottom row). Scale bars = 100 μm. B, Fat graft vascularization was quantified as ROI area occupied by CD31-positive pixels. Fat enriched with CD34+CD146+ ASCs had greatest vascularization, indicated by increased CD31 immunofluorescence staining compared to fat enriched with CD34+CD146− or CD34+ UF ASCs, and grafts not enriched with ASCs and fat alone; n = 5 per group (*P < .05, **P < .01). ASC, adipose-derived stromal cell; DAPI, 4 0 ,6-diamidino-2-phenylindole; ROI, region of interest; UF, unfractionated postgrafting ( Figure 5A[i], top row). The fat grafts enriched with CD34 +CD146+ ASCs had greater integrity (****P < .0001), fewer cysts/vacuoles (****P < .0001), less inflammation (****P < .0001), and less fibrosis (***P < .001) than fat grafts enriched with CD34+CD146− ASCs, CD34+ UF ASCs, or fat alone ( Figure 5B). Since dead adipocytes cannot be distinguished from living adipocytes using H&E, perilipin staining was used to label viable adipocytes. While grafted fat in all groups of mice contained viable adipocytes, more perilipin staining was observed in mice grafted with CD34+CD146+ and CD34+ UF ASCs ( Figure 5A[ii], top row). Explants were also stained with F4/80+ to label macrophages. We noted grafts enriched with CD34+CD146+ had qualitatively less staining for F4/80, consistent with less inflammation ( Figure 5A[ii], bottom row). Last, Masson's trichrome staining was used to further visualize fibrosis within the grafted fat. Also consistent with ratings of the H&E-stained slides, fat grafts enriched with CD34+CD146+ ASCs had the least staining for collagen (blue) and thus the least fibrosis. The degree of fibrosis was comparable between grafts enriched with CD34+CD146− ASCs and fat grafted without ASC enrichment ( Figure 5A[i], bottom row).  32 The exact biological role of CD146, however, is poorly defined. CD146+ cells play a role in implantation, placentation, and tumor progressionpossibly through enhancing the interaction between endothelial and melanoma cells. 33 Umbilical cord MSCs positive for CD146 have enhanced tube-forming abilities and coexpress the Notch ligand Jag-ged1 (JAG1), a known potent regulator of blood vessel maturation. 11 Pericytes, cells believed to give rise to MSCs, express CD146, and transplantation of purified pericyte populations promotes tissue repair and angiogenesis. 30 Like pericytes, ASCs are predominantly associated with vascular structures within adipose tissue, but unlike pericytes, they are CD34+. 34 Recent work has suggested that CD146 may also mark a proangiogenic subset of ASCs (CD34+) in human adipose tissue. 35,36 The angiogenic qualities of this subpopulation, however, have only been demonstrated using in vitro assays. 35,36 Our work is the first to demonstrate the angiogenic effects of the CD34+CD146+ subpopulation in vivo in the setting of fat grafts.

| Improved vascularization of fat grafts enriched with CD146+ ASCs
The mechanisms by which ASCs promote fat graft vascularization are thought to be largely mediated via paracrine signaling. 4 Here, we show that CD34+CD146+ ASCs highly express VEGF, a potent proangiogenic factor. 37 VEGF induces endothelial mitosis by acting though mitogen-activated protein kinases, 38 and also promotes endothelial cell migration, together leading to angiogenesis of existing vessels. 39 The CD34+CD146+ ASC subpopulation also highly express ANGPT1, another endothelial cell-specific growth factor. ANGPT is reported to stabilize blood vessels and counteract VEGF-induced blood vessel leakage. 40 Thus, together, VEGF and ANGPT1 have an additive effect on angiogenesis. 40 Interestingly, the CD34+CD146+ ASC subpopulation expresses more ANGPT1 than CD34−CD146+ pericytes, and may therefore be of greater therapeutic utility. Finally, the surface marker CD146 can be cleaved by metalloproteases and released by CD146+ cells as a soluble molecule. 41,42 In this form, CD146 can act chemotactically to enhance formation of vascular structures by endothelial cells and promote neovascularization in a hind limb ischemia model in rats. 43 The enhanced vascularization observed when grafted fat is enriched with CD34+CD146+ ASCs likely also explains the observed reduced fibrosis and inflammation when fat grafts are enriched with this ASC subpopulation. Fat necrosis is an inflammatory process which in time leads to tissue fibrosis. Improved delivery of oxygen and nutrients by effective vascularization in the early stages postgrafting may promote adipocyte survival, reduce inflammation, and prevent tissue fibrosis.
Fat grafting procedures are commonly performed surgeries; in the United States, it is estimated that more than 400 000 liposuction procedures are performed each year. 44 Fat grafting is being adopted not only as a technique for restoring contour deformities, but also as a method by which to mitigate fibrosis and the effects of radiation damage on skin. Despite this growing popularity, fat retention rates are variable, especially in irradiated tissue where pathological fibrosis severely diminishes the dermal vascular supply. Enriching fat grafts with proangiogenic ASCs may confer enhanced retention and address the unpredictability of fat grafting. The CD146+ ASC subpopulation, which represents a significant proportion of CD34+ ASCs, may be of particular benefit in this setting. Beyond fat grafting, this proangiogenic ASC subpopulation may help promote vascularization and minimize fibrosis in wounded or irradiated skin. As such, our future aims include investigating the role of CD146+ ASCs in these settings. Finally, while CD146 may mark an angiogenic subset of ASCs, it is likely that other ASC populations play comparable or complementary functions in angiogenesis. We have previously identified a separate population of proangiogenic ASCs, distinguished by the surface marker CD248. 45 Although not investigated in this article, future work may consider the degree to which ASCs coexpress both CD146 and CD248 and how their roles overlap or interact to promote fat graft retention.

| CONCLUSION
We have identified a subpopulation of ASCs that express the CD146 surface marker and exhibit enhanced regenerative qualities. This

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author.