Abstract
Adipose-derived stromal vascular fraction (SVF) cells have been shown to self-associate to form vascular structures under both in vitro and in vivo conditions. The angiogenic (new vessels from existing vessels) and vasculogenic (new vessels through self-assembly) potential of the SVF cell population may provide a cell source for directly treating (i.e., point of care without further cell isolation) ischemic tissues. However the correct dosage of adipose SVF cells required to achieve a functional vasculature has not been established. Accordingly, in vitro and in vivo dose response assays were performed evaluating the SVF cell vasculogenic potential. Serial dilutions of freshly isolated rat adipose SVF cells were plated on growth factor reduced Matrigel and vasculogenesis, assessed as cellular tube-like network assembly, was quantified after 3 days of culture. This in vitro vasculogenesis assay indicated that rat SVF cells reached maximum network length at a concentration of 2.5 × 105 cells/ml and network maintained at the higher concentrations tested. The same concentrations of rat and human SVF cells were used to evaluate vasculogenesis in vivo. SVF cells were incorporated into collagen gels and subcutaneously implanted into Rag1 immunodeficient mice. The 3D confocal images of harvested constructs were evaluated to quantify dose dependency of SVF cell vasculogenesis potential. Rat- and human-derived SVF cells yielded a maximum vasculogenic potential at 1 × 106 and 4 × 106 cells/ml, respectively. No adverse reactions (e.g., toxicity, necrosis, tumor formation) were observed at any concentration tested. In conclusion, the vasculogenic potential of adipose-derived SVF cell populations is dose dependent.
Introduction
Cell-based therapies are being developed to address tissue ischemia caused by insufficient microvascular perfusion. Microvascular dysfunction is a common occurrence in a wide range of diseases including peripheral limb ischemia, diabetic retinopathy, and myocardial infarction. The common proposed mechanism of action for cell-based therapies to address these clinical conditions is the accelerated formation of a functional microcirculation [reviewed in (2,29)]. Various approaches to stimulate restoration of microvascular function within ischemic tissues have been investigated including systemic and local delivery of angiogenic growth factors as well as genes encoding these factors (15,25) and delivery of potentially therapeutic cells (e.g., BMMSCs, adipose-derived stem and regenerative cells, circulating endothelial progenitor cells) (3,6,8,13,16,19,20,31).
Regenerative and stem cells present in adipose tissue represent an attractive cell therapy to correct tissue ischemia due to the ease of cell isolation, point of care, and the established ability of these cells to stimulate microvascular formation. The primary cell isolate that results from the enzymatic digestion of adipose tissue, the stromal vascular fraction (SVF), is a heterogeneous population of cells that includes endothelial cells, macrophages, stem cells, pericytes, and smooth muscle cells (34,35). This cell population has previously been shown to participate in blood vessel formation through self-assembly of SVF cells into microvascular structures in vivo (11,23). The SVF cell population also expresses known angiogenic factors (32), and thus the formation of new microvasculature following adipose SVF implantation may result through both vasculogenic (26) and angiogenic (4) mechanisms.
As adipose SVF cell-based therapies and indeed all cell-based angiogenic and vasculogenic therapies progress to human clinical trials, it is imperative that the cell dose delivered to the patient be established to achieve clinical efficacy. Clinical trials employing other cell-based technologies to address ischemic tissue have previously used cell concentrations on the order of up to or greater than 109 cells/ml (16,19,31); however, the dose dependency of these therapies has not been established. Lacking in most cell-based therapies is a full understanding of the relationship between cell dose and a measurable efficacy endpoint. Based on previous studies (11,23), we hypothesized that the self-assembly of blood vessels from adipose SVF cell populations is cell concentration dependent. In vitro and in vivo studies were performed and established a relationship between SVF dose and efficacy, measured by an assessment of blood vessel assembly. These studies provide an assessment to plan clinical dosing strategies for future adipose SVF cell-based therapies.
Materials and Methods
Isolation of Rat and Human SVF
Rat adipose SVF cells (rSVF) were derived from the epididymal fat pads of transgenic green fluorescent protein (GFP) adult male Sprague–Dawley rats (Rat Research and Resource Center, University of Missouri, Columbia, MO, USA) as previously described and approved by the University of Louisville IACUC Committee (23). For rSVF quantification, each experiment was repeated three times, and sufficient cells were isolated to complete an experiment; therefore, a total of six rats were used. Human SVF (hSVF) was procured from human female lipoaspirate fat samples (all other identifying information was unavailable) obtained under IRB exemption #09.0037, through nonultrasonic suction-assisted liposuction of abdominal, thigh, buttock, flank, and axillary regions. Briefly, adipose tissue was rinsed with 0.1% bovine serum albumin (BSA) (Sigma-Aldrich, St. Louis, MO, USA) in phosphate-buffered saline (PBS) (Hyclone, Logan, UT, USA), finely minced, and vigorously shaken in 2 mg/ml type I collagenase (Worthington Biochemical Company, Freehold, NJ, USA) for 40 min at 37°C for optimal digestion. The SVF cells were pelleted via centrifugation and buoyant adipocytes discarded. The pellet was then washed once more with 0.1% BSA–PBS in preparation for use (14,23). Viable nucleated cells were counted using NucleoCounter NC-100 (ChemoMetec, Allerod, Denmark).
Matrigel Network Assay
To demonstrate network formation in vitro, we utilized an in vitro Matrigel assay (18) using growth factor reduced (GFR) Matrigel (BD Biosciences San Jose, CA, USA) (1,10). Twenty-four-well tissue culture plates (BD Biosciences) were coated with a thin gel of GFR Matrigel and incubated at 37°C and 5% CO2 for 30 min to allow for polymerization per the manufacturer's instructions. One milliliter of rSVF serial dilutions (10, 5, 2.5, 1.25, and 0.625 × 105 cells/ml) were added on top of the GFR-Matrigel and cultured for 72 h at 37°C and 5% CO2 in complete rat media [Dulbecco's modified Eagle's medium (DMEM high glucose; Invitrogen, Camarillo, CA, USA), 2 mM L-glutamine (Invitrogen), 50 μg/ml ECGS, 5.6 mM HEPES (Invitrogen), and 10% FBS (Hyclone)] (23).
Rat and Human Microvessel Construct Preparation
Freshly isolated rSVF-GFP or hSVF was serial diluted as described in the in vitro assays to yield the same concentrations with an additional concentration of 4 × 106 cells/ ml to test for cell overloading of the construct. Cells were suspended in 250 μl of 3 mg/ml collagen type I (BD Biosciences). Constructs were subcutaneously implanted into the flanks of 8–10-week-old male Rag1–/– immune-compromised mice (Jackson Laboratory, Bar Harbor, ME, USA) (23,28) in accordance with the University of Louisville IACUC approved protocols. Each implant experiment (i.e., rSVF and hSVF quantification) was replicated three times with one mouse for each implant dose, each mouse receiving two implants. The total number of mice used was 36. Implants of rSVF and hSVF were harvested at 2 and 4 weeks, respectively. These different time courses were based on previous experimental data for achieving vascular networks in vivo by these two species (23). Harvested constructs were explanted and fixed with 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA, USA) for quantitative and qualitative analysis. Following fixation, constructs were permeabilized with 0.5% triton (Fisher-Scientific, Pittsburgh, PA, USA) and then blocked for 1 h in fresh 5% goat serum (Sigma-Aldrich) in PBS. rSVF-GFP and hSVF were incubated with GS1-rhodamine or UEA1-fluorescein (Vector Labs, Burlingame, CA, USA) (1:100), respectively, at 4°C overnight for species- specific endothelium labeling. Constructs were then washed multiple times with PBS in preparation for imaging.
Microscopy and Quantitative Vascular Analysis
For analysis of in vitro networks on GFR Matrigel, phase contrast images of vessel network formation were acquired using an Olympus IX71 (Olympus America, San Jose, CA, USA) inverted microscope and images quantified using Metamorph software (Molecular Devices, Sunnyvale, CA, USA). Total network length was measured, and the sum for each respective concentration in micrometers was recorded in Excel (Microsoft, Redmond, WA, USA) and averaged with its triplicate counterparts. A scatterplot of total network length versus cell concentration was then performed using SigmaPlot (Systat Software, San Jose, CA, USA).
For in vivo vascularization analysis, explanted rSVF-GFP and hSVF constructs were imaged using an Olympus MPE FV1000 confocal microscope. Image stacks were analyzed for total vessel length and branch points using Metamorph software. The relative vessel length, determined by combining all image stacks for the entire construct, was then calculated and recorded in Excel to determine concentration averages.
Statistics
One-way analysis of variance (ANOVA) was used to test for mean differences and Tukey's post hoc pairwise comparisons between groups using SigmaPlot software. All quantifications are reported as mean ± SEM. A value of p ≤ 0.05 was considered statistically significant.
Results
rSVF Network Assembly Demonstrates a Concentration Response In Vitro
After 72 h of culture, freshly isolated rSVF undergo self-assembly to form vessel-like network structures (Fig. 1). Concentration dependency was evaluated by measuring total network formation as a function of cell densities of 0.625, 1.25, 2.5, 5 and 10 × 105 cells/ml. At lower concentrations (0.625 and 1.25 × 105 cells/ml) sparse cellular interactions were observed and failed to form definitive interconnected networks. As cellular concentration increased (2.5–5.0 × 105 cells/ml), cells appeared to have sufficient numbers to interact and achieve formation of a more intricate network (Fig. 1). At the highest concentration (1.0 × 106 cells/ml), the cells still assembled into networks, but were thicker and tended toward cellular monolayer. One-way ANOVA indicates that the mean network tubule lengths across all cell concentrations were significantly different (p = 0.008). Tukey's post hoc pairwise analysis indicated that in comparison to the lowest dose (0.625 × 105 cells/ml) the highest concentrations of 2.5 and 5 × 105 cells/ml were significantly different (p = 0.029 and 0.011, respectively), though the highest dose was not significantly different (p = 0.068). Pairwise comparison between the three highest doses showed no significant difference in total network tube length. These data suggest SVF cell network formation is concentration dependent within a range, then plateaus to a constant total level.
In vitro rSVF tube formation is concentration dependent. rSVF tube network formation on Matrigel requires a minimum cell concentration before significant assembly occurs. Quantification of total network tubule length significantly increases at 2.5 (*p = 0.029) and 5 (**p = 0.011) × 105 cells/ml compared to the lowest concentration. Data are mean ± SEM.
rSVF Vasculogenesis In Vivo
As endothelial cells are a major component of the SVF cell population (23), we hypothesized that self-assembly of implanted endothelium into vascular networks would occur and be a primary contributor to SVF-induced vascularization. Since SVF is a heterogeneous mix of cells, and all SVF cells from the GFP rat fluoresce green, we wanted to confirm if the vessel-like structures detected in the implants were being assembled by endothelium. Following explantation, rSVF-GFP collagen constructs were incubated with GS1-rhodamine lectin for labeling rodent endothelial cells (Fig. 2). Our results demonstrate that the SVF cell vasculatures (i.e., GFP+) colocalize with GS1-labeled cells, indicating the SVF-formed microvas-culatures are composed of endothelium. Within each construct, GFP+ vasculatures indicated the persistence of the implanted rSVF (Fig. 2).
In vivo rSVF vessels are endothelial derived. rSVF from GFP transgenic rats self-assemble into vessel structures in vivo. Colocalization of GS1+ rodent endothelium within the implant with rSVF GFP+ vessels confirms endothelial composition (merge). Scale bar: 200 μ.
We next tested if the in vitro concentration response for network formation had a corollary in vivo and if the heterogeneous rSVF cellular mix would have a detrimental effect if the implant was “overloaded” (Fig. 3). Therefore, we generated implant constructs using fresh isolated rSVF-GFP cells at the concentrations utilized in our previous in vitro experiments and added a “cell overload” construct of 4 × 106 SVF cells/ml. After 2 weeks, the implants were harvested, imaged by confocal microscopy, and tube length quantified as a function of cell concentration (Fig. 4). One-way ANOVA indicated that the mean vessel length difference across groups was significantly different with rSVF cell concentration (p = 0.006). Pairwise comparisons indicated there were no significant differences between the mean vessel lengths of the lowest four concentrations (0.625, 1.25, 2.5, and 5 × 105 cells/ ml). The mean vessel length compared across the highest three concentrations (5, 10, and 40 × 105 cells/ml) also were not significantly different from each other. When 10 × 105 cells/ml was compared to the lowest two doses, the mean differences in vessel length indicated a significantly different increase (0.625 vs. 10, p = 0.010; 1.25 vs. 10, p = 0.034). The maximal tube length was quantified at the cell concentration of 10 × 105 cells/ml, and interestingly, at the highest concentration of 40 × 105 cells/ml, total vessel length was less extensive and statistically the same as the 5 × 105 cells/ml. At this maximal concentration, mean vessel length decreased (Fig. 4), and images showed evidence of significant cell clustering, though no tumor formation was indicated (Fig. 3). To get an indication of vascular complexity, we also measured the number of branch points and normalized to vessel length (Fig. 4). One-way ANOVA indicated that the number of branch points were not significantly different across doses when normalized to total vessel length (p = 0.39). These data indicate achieving significant vessel formation in vivo requires a minimal concentration of cells (5–10 × 105 cells/ml) and that there is a comparable response between in vitro network assembly and in vivo vessel formation.
rSVF in vivo vascular self-assembly is concentration dependent. Implantation for 2 weeks of transgenic GFP+ rSVF at different cell concentrations indicates a minimum cellular quantity is required to achieve vascular self-assembly.
rSVF total vessel length but not relative branching is cell concentration dependent. Quantification of the total 2-week implant vessel length formed by rSVF (i.e., GFP+ vessels) indicates a concentration of 10 × 105 cells/ml is required for a statistically significant increase in vessel formation compared to the two lowest cell concentrations (*p < 0.05). However, when the number of vessel branch points was normalized to total vessel length, no difference was detected as a function of cell concentration.
Human SVF Is Concentration Dependent In Vivo
We have recently shown that human SVF is capable of forming vascular networks in vivo that interact with implant parenchymal cells (23). Therefore, we asked if freshly isolated hSVF also exhibited a concentration-dependent vasculogenic response in vivo. We performed collagen implant studies utilizing the same hSVF cell concentrations as in the previous rSVF experiments. A total of three separate, donor fat isolations were used and total vessel length and branching averaged for each concentration. hSVF constructs were harvested, labeled with UEA1-fluorescein lectin (human endothelium), and imaged by confocal microscopy (Fig. 5). hSVF microvessel constructs displayed UEA1+ vasculature, confirming human endothelial origin. One-way ANOVA indicated that the mean vessel lengths were significantly different with cell concentration (p = 0.013) (Fig. 6). Images of the hSVF vasculogenic response showed that at lower concentrations, some vessel formation is evident (Fig. 5), but pairwise comparisons across groups from 0.625 to 10 × 105 cells/ml were not statistically different (Fig. 6). However, at the highest concentration of 40 × 105 cells/ ml, a large increase in vascular network formation was observed that was significantly different from all others except 10 × 105 cells/ml (0.625, p = 0.012; 1.25, p = 0.016; 2.5, p = 0.027; 5, p = 0.041; 10, p = 0.162) (Fig. 6). One-way ANOVA testing of the number of branch points normalized to vessel length indicated cell concentration has a significant effect on vessel architecture (p = 0.025). Pairwise comparison also showed that the highest dose of 40 × 105 cells/ml was significantly different from the lowest number of normalized branch points (p = 0.025). These data support the hypothesis that hSVF vessel formation and complexity is concentration dependent, but it also indicates there may be potential differences between rat and human SVF.
hSVF in vivo vascular self-assembly is concentration dependent. Implantation of hSVF at different concentrations demonstrates a cell concentration dependence to achieve vascular self-assembly as detected by labeling with the lectin UEA1 for recognizing human endothelium.
hSVF total vessel length and vessel branching is cell concentration dependent in vivo. The maximum cell concentration (40 × 105 cells/ml) of hSVF demonstrates a significantly greater capacity for vascular self-assembly than the lowest four concentrations (*p ≤ 0.041). When branch points were normalized to total vessel length for each implanted concentration, statistical difference was only detected between two of the measurements (*p = 0.025), suggesting cell concentration may not affect vascular branching. Data are mean ± SEM.
Discussion
The significant and novel findings from this study are 1) rat and human SVF cell vasculogenic potential is a cell concentration-dependent process, and 2) the concentration required to achieve effective vasculogenesis differs between rat and human SVF cells.
We have previously established that adipose SVF is a rich source of endothelium and other vascular-associated cells (33), and when isolated either as blood vessel fragments or as a fully dissociated single-cell preparation can form neo blood vessels both in vitro and in vivo (7,23). Our objective in this study was to determine the effect of cell concentration on SVF cell vasculogenic potential. As we have previously reported, rat-derived adipose SVF cells undergo vasculogenic assembly into tube-like structures within 18 h of plating on collagen. In the current study, the SVF cell population was seeded onto Matrigel, and vasculogenic tube assembly was observed. However, unlike the regression and dissociation of vessel tube-like structures reported for pure endothelium plated on Matrigel (12), we observed persistence of the tube-like structures, indicating the heterogeneous population of cells in SVF (35) supports vasculogenesis with stability of the structures formed. Quantification of the in vitro vessel-like networks indicated cell concentration dependence.
A dose-dependent vasculogenesis of adipose SVF was also observed in vivo. In these studies, the SVF cell population was first suspended in a collagen type I gel and the entire gel implanted subcutaneously. The total vessel lengths that formed were quantified, and a relationship between SVF concentration and vessel length formed was observed with a plateau at the highest concentrations evaluated. The differences observed in the concentration of SVF required to reach a plateau of vasculogenesis is most likely due to 1) plating cells on the gel surface concentrates the cells while dispersion of cells within the gel results in a more dilute suspension, and 2) in vitro studies utilized a Matrigel matrix, while in vivo studies utilized collagen I. During the in vivo studies, we also tested if the 3D gel construct could be overloaded with cells. This issue is critical for future potential clinical use since 1) it will be necessary to determine how many cells are required to achieve a desired therapeutic response and 2) similar to pharmacological dosing, there may be a cell concentration level above which therapeutic effects diminish or become detrimental. During the rat SVF studies, the maximum concentration of 4 × 106 cells/ml supported vasculogenesis with no adverse effects such as necrotic zones or undifferentiated cell masses. However, the total network length, though not statistically different from lower concentrations, suggested a reduced vasculogenic trend.
While the human adipose SVF cells supported the formation of new vessels following implantation, a delay of approximately 2 weeks was observed to reach maximal vasculogenesis compared to results with rat-derived SVF cells. Even at the highest dose, 4 × 106 cells/ml, the human SVF vasculogenesis was delayed compared to previous studies using rat SVF (23). Mechanisms underlying this species-dependent difference in the rate of vasculogenesis are unknown; however, it is likely that the relative cell composition of SVF differs between rat and human and the growth factors and cytokines necessary to support vasculogenesis are most likely also different between these species. Nevertheless, in both species, SVF cell populations support a dose-dependent vasculogenesis in vivo.
These studies were performed to compare two commonly used sources of fat, the rat epididymal fat pad and human liposuction-derived fat. It is important to note that the harvest of rat adipose from the epididymal fat pad involves blunt dissection of the fat from the underlying tissue, hand mincing, and then enzymatic digestion. In contrast, the human adipose utilized in this study was acquired after tumescent liposuction from subcutaneous fat deposits, which involves the injection of a saline solution that includes lidocaine and epinephrine, then the mechanical suction of the adipose tissue (9). Though in both species adipose is extensively washed, exposure of the human adipose to lidocaine and epinephrine, as well as the comparatively rough extraction procedure represented by liposuction, could affect hSVF cell recovery and function, providing a partial explanation for the delay in vasculogenic potential of the human SVF.
Inherent in the study of cell concentration to establish dose dependency is the methodology used to establish not only the number of cells in the initial cell preparation, but the viability of cells, and the relative concentration of different cell types, especially in the complex cell mixture that is represented by adipose SVF. In the current study, a fluorescence-based automated system was used to quantify both cell number and cell viability in the primary SVF isolates. Previous studies have established fluorescence-based assays, in contrast to methods that use direct particle counting, provide more accurate calculation of SVF cell populations. Our results are in agreement with the studies by Morrison et al. (21) that human adipose SVF isolated using enzymatic digestion results in a cell yield of approximately 1 million cells per gram of fat. There are two limitations to our present study. The dose calculation is based on the entire cell concentration per milliliter and not corrected for cell viability. Cell viability was consistently greater than 80%; thus, a majority of cells were viable. Second, we did not perform FACS analysis of the SVF cell preparations. The primary rationale for not performing FACS is the effect of liposuction and enzymatic digestion on cell surface marker expression used in FACS analysis. Another potential reason for the observed results could be inherent species differences in cellular populations due to evolutionary development (5,24). Multiple studies, including our own, have investigated the cellular components found in adipose tissue (17,23,35), yet it should be kept in mind that cell analysis occurs after enzymatic digestion that could remove surface proteins and thus affect quantitative outcome.
Whatever the difference between rat and human SVF cell functionality, these results strongly support the use of adipose SVF as a potential autologous cell source for therapeutic vascularization. It is well recognized that excess weight and obesity are critical health issues in Western societies (5), and adipose vasculature plays a crucial role in adipose tissue homeostasis (30). Rupnick and colleagues demonstrated that adipose vasculature was sensitive to angiogenesis inhibitors even when it was not actively growing (27). This led the authors to propose that adipose vasculature may reside in a relatively immature state such that it is rapidly responsive to the tissue's metabolic requirements. This relative immaturity may be why adipose SVF cells can self-assemble into a functional vasculature after complete dispersion to single cell components (23).
From this current study and previous studies (22,23), we propose that the heterogeneous cell composition of adipose SVF functions as a collective cellular system, and individual cell compositions in adipose (e.g., endothelium) will not support vasculogenic and angiogenic reconstitution of a complete and functional microcirculation. Furthermore, this study supports the conclusion that reconstitution of a microcirculation through cell transplantation is dose dependent. Human trials of cell-based therapies to accelerate formation of blood vessels must consider the effective dose to support assembly of a new microcirculation.
Footnotes
Acknowledgments
Supported by AHA 115SDG7500025 (N.L.B.), NIH DK078175 (S.K.W.), EB007556 (J.B.H.) grants, and microscopy core supported by NIH/NIGMS P30 GM103507. The authors declare no conflicts of interest.