Pro-angiogenic Activity Discriminates Human Adipose-Derived Stromal Cells From Retinal Pericytes: Considerations for Cell-Based Therapy of Diabetic Retinopathy

Abstract

Diabetic retinopathy (DR) is a frequent diabetes-associated complication. Pericyte dropout can cause increased vascular permeability and contribute to vascular occlusion. Adipose-derived stromal cells (ASC) have been suggested to replace pericytes and restore microvascular support as potential therapy of DR. In models of DR, ASC not only generated a cytoprotective and reparative environment by the secretion of trophic factors but also engrafted and integrated into the retina in a pericyte-like fashion. The aim of this study was to compare the pro-angiogenic features of human ASC and human retinal microvascular pericytes (HRMVPC) in vitro. The proliferation and the expression of ASC and HRMVPC markers were compared. Adhesion to high glucose-conditioned endothelial extracellular matrix, mimicking the diabetic microenvironment, was measured. The angiogenesis-promoting features of both cell types and their conditioned media on human retinal endothelial cells (EC) were assessed. To identify a molecular basis for the observed differences, gene expression profiling was performed using whole-genome microarrays, and data were validated using PCR arrays and flow cytometry. Based on multiplex cytokine results, functional studies on selected growth factors were performed to assess their role in angiogenic support. Despite a distinct heterogeneity in ASC and HRMVPC cultures with an overlap of expressed markers, ASC differed functionally from HRMVPC. Most importantly, the pro-angiogenic activity was solely featured by ASC, whereas HRMVPC actively suppressed vascular network formation. HRMVPC, in contrast to ASC, showed impaired adhesion and proliferation on the high glucose-conditioned endothelial extracellular matrix. These data were supported by gene expression profiles with differentially expressed genes. The vessel-stabilizing factors were more highly expressed in HRMVPC, and the angiogenesis-promoting factors were more highly expressed in ASC. The vascular endothelial growth factor receptor-2 inhibition efficiently abolished the ASC angiogenic supportive capacities, whereas the addition of angiopoietin-1 and angiopoietin-2 did not alter these effects. Our results clearly show that ASC are pro-angiogenic, whereas HRMVPC are marked by anti-angiogenic/EC-stabilizing features. These data support ASC as pericyte replacement in DR but also suggest a careful risk-to-benefit analysis to take full advantage of the ASC therapeutic features.

Keywords: angiogenesis; angiopoietin; diabetic retinopathy; human adipose-derived stromal cells; human retinal pericytes; vascular–endothelial growth factor.

FIGURE 1

Morphology of adipose-derived stromal cells (ASC) and human retinal microvascular pericytes (HRMVPC) and the growth curves differ, whereas the mesenchymal stromal cells (MSC) marker expression is indistinguishable. (A) Representative images of ASC, Bmi-HRMVPC and HRMVPC. (B) Growth curves of ASC (three different donors), HRMVPC (one donor), and Bmi-HRMVPC cell line determined by kinetic live imaging (****p ≤ 0.0001, two-way ANOVA with Sidak’s post hoc test). (C) Surface expression of MSC positive and negative markers for ASC (n = 7 different donors), HRMVPC (HRMVPC and Bmi-HRMVPC groups, n = 6, biological and technical replicates from three different donors), and endothelial cells (EC; n = 3, human umbilical vein endothelial cell and human retinal microvascular endothelial cell groups) (*p ≤ 0.05 and **p ≤ 0.01, two-way ANOVA with Tukey’s post hoc test; significant differences to EC not depicted).

FIGURE 2

Human retinal microvascular pericytes (HRMVPC) show impaired interaction with high glucose (HG)-conditioned endothelial cells (EC) extracellular matrix (ECM) compared to adipose-derived stromal cells (ASC). The EC, either human umbilical vein endothelial cells (HUVEC) or human retinal microvascular endothelial cells (HRMVEC), were grown until confluence in normal glucose (NG), HG, or mannitol-supplemented medium. Then, the EC ECM was prepared as described in section “Materials and Methods”. ASC or HRMVPC were seeded on top, and the increase in confluence was monitored using live cell imaging over 48 h. (A) Confluence of ASC seeded on HUVEC ECM conditioned by NG, HG, or mannitol (n = 4 biological replicates; non-significant, two-way RM-ANOVA with Dunnett’s post hoc test). (B) Confluence of HRMVPC seeded on HG-conditioned HRMVEC ECM compared to NG-conditioned ECM (n = 5 independent experiments with HRMVPC from one donor; non-significant, two-way RM-ANOVA with Sidak’s post hoc test). (C) To account for differences in proliferative capacity, confluence data were normalized against the NG-conditioned HUVEC ECM (line at value 1). ASC vs. Bmi-HRMVPC (n = 3 independent experiments with ASC from different donors; non-significant, two-way RM-ANOVA with Sidak’s post hoc test). (D) ASC vs. HRMVPC interaction with NG and HG-conditioned HUVEC ECM (n = 5 independent experiments with ASC from different donors; *p ≤ 0.05 for ASC vs. HRMVPC for the first 14 h, two-way RM-ANOVA with Sidak’s post hoc test). (E) ASC vs. HRMPVC interaction with NG and HG-conditioned HRMVEC ECM (n = 3 independent experiments with ASC from different donors; *p ≤ 0.05 for HRMVPC interaction with NG and HG-conditioned HRMVEC ECM for all time points, two-way RM-ANOVA with Sidak’s post hoc test).

FIGURE 3

Pro-angiogenic activity discriminates human adipose-derived stromal cells (ASC) and retinal pericytes. (A) Vascular network formation in CC angiogenesis assay using network branch point metrics (1/mm²) of ASC/human retinal microvascular endothelial cells (HRMVEC), human retinal microvascular pericytes (HRMVPC)/HRMVEC, and HRMVPC/human umbilical vein endothelial cells (HUVEC) monitored using live imaging (four different ASC donors, two and three independent experiments with HRMVPC/HRMVEC and HRMVPC/HUVEC, respectively). Medium change is indicated by the box. ASC/HUVEC cocultures are not shown. (B) Vascular network formation in a BM angiogenesis assay (n = 11 paired experiments with ASC from 11 different donors and HRMVPC from three different donors in independent experiments; *p ≤ 0.05, one-way ANOVA with Tukey’s post hoc test). (C,D) Vascular network formation in CC angiogenesis assay, adding ¼ ECGM-2 as control or CC angiogenesis assay-derived CM ASC or CM HRMVPC to ASC/HRMVEC (C) or HRMVPC/HRMVEC (D) cocultures (n = 3, with three different ASC donors, independent experiments for two HRMVPC donors; *p ≤ 0.05 to ****p ≤ 0.0001, two-way RM-ANOVA with Tukey’s post hoc test). (E) Representative images showing the expression of α-smooth muscle actin (α-SMA) in ASC/EC cocultures (dTomato EC, red; α-SMA green; DAPI, blue nuclear counter stain and merge). Scale bar, 50 μm.

FIGURE 4

Differential marker expression of adipose-derived stromal cells (ASC) and human retinal microvascular pericytes (HRMVPC) assessed by microarray, PCR array, flow cytometry, and immunofluorescence. (A) Volcano plot visualizing the microarray data showing the magnitude of change (log₂-fold change, x-axis) vs. statistical significance [-log₁₀(p-value), y-axis] of gene expression of ASC vs. HRMVPC (each n = 3 biological replicates). Most differentially expressed genes and putative pericyte markers are labeled. (B) Validation of microarray results using PCR array. The correlation is high (Spearman correlation R = 0.95, p < 2.2e-16). Most differentially expressed genes and putative pericyte markers are labeled. (C) Marker expression in ASC, HRMVPC, and EC measured by flow cytometry. Percent positivity calculated against the unstained control [n = 7 different ASC donors, n = 6 HRMVPC with three donors in independent experiments, n = 3 human umbilical vein endothelial cells and human retinal microvascular endothelial cells (HRMVEC); ***p ≤ 0.001 and ****p ≤ 0.0001, two-way ANOVA with Tukey’s post hoc test, p-values for EC not shown]. Correlation to microarray and PCR array is poor (Spearman correlation R = 0.4, p = 0.2) using mean fluorescence intensity values. (D) Representative pictures of marker expression (green) in ASC, HRMVPC, and the respective cocultures measured by two-photon microscopy. dTomato HRMVEC are shown in red. Scale bar, 100 μm.

FIGURE 5

Conditioned medium composition differs significantly between adipose-derived stromal cells (ASC) and human retinal microvascular pericytes (HRMVPC). Angiogenic growth factor levels (A–I) of conditioned cell culture supernatants (CM) of mono- or cocultures measured by flow cytometry-based plex assay (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001, one-way ANOVA with Tukey’s post hoc test) (ASC from six different donors, HRMVPC from two different donors in independent experiments, ¼ ECGM-2 n = 2 independent experiments).

FIGURE 6

VEGFR-2 inhibition efficiently abolishes adipose-derived stromal cells (ASC) angiogenic supportive capacities, whereas Ang-1 and Ang-2 addition do not alter effects. (A) Vascular network formation in CC angiogenesis assay. VEGF and the VEGF inhibitor ZM 3238811 were added to ASC-human retinal microvascular endothelial cell (HRMVEC) or human retinal microvascular pericyte (HRMVPC)/HRMVEC cocultures (three different ASC donors and one HRMVPC donor; **p ≤ 0.01, two-way RM-ANOVA with Tukey’s post hoc test). (B) Vascular network formation in CC angiogenesis assay. CM ASC or CM HRMVPC are added to ASC/HRMVEC cocultures with the VEGF inhibitor ZM 323881 (three different ASC donors; *p ≤ 0.05, two-way RM-ANOVA with Tukey’s post hoc test). (C) Vascular network formation in BM angiogenesis assay assessing CM ASC, CM HRMVPC, and VEGF₁₆₅ in the absence and the presence of the VEGF inhibitor ZM 323881 (different ASC donors indicated by different symbols; *p ≤ 0.05, two-way ANOVA with Sidak’s post hoc test). (D) Vascular network formation in BM angiogenesis assay assessing CM ASC, CM HRMVPC, and control condition in the absence and the presence of recombinant Ang-1 or Ang-2 (different ASC donors or independent HRMVPC experiments indicated by different symbols; non-significant, two-way ANOVA with Tukey’s post hoc test).