Mesenchymal Stromal Cells: Inhibiting PDGF Receptors or Depleting Fibronectin Induces Mesodermal Progenitors with Endothelial Potential

Realizing the full therapeutic potential of mesenchymal stromal/stem cells (MSCs) awaits improved understanding of mechanisms controlling their fate. Using MSCs cultured as spheroids to recapitulate a three-dimensional cellular environment, we show that perturbing the mesenchymal regulators, platelet-derived growth factor (PDGF) receptors or fibronectin, reverts MSCs toward mesodermal progenitors with endothelial potential that can potently induce neovascularization in vivo. MSCs within untreated spheroids retain their mesenchymal spindle shape with abundant smooth muscle α-actin filaments and fibronectin-rich matrix. Inhibiting PDGF receptors or depleting fibronectin induces rounding and depletes smooth muscle α-actin expression; these cells have characteristics of mesenchymoangioblasts, with enhanced expression of mesendoderm and endoderm transcription factors, prominent upregulation of E-cadherin, and Janus kinase signaling-dependent expression of Oct4A and Nanog. PDGF receptor-inhibited spheroids also upregulate endothelial markers platelet endothelial cell adhesion molecule 1 and vascular endothelial-cadherin and secrete many angiogenic factors, and in vivo they potently stimulate neovascularization, and their MSCs integrate within functional blood vessels that are perfused by the circulation. Thus, MSC potency and vascular induction are regulated by perturbing mesenchymal fate.


INTRODUCTION
Fulfilling the potential of mesenchymal stromal (stem) cells (MSCs) in regenerative medicine requires an improved understanding of the mechanisms that control their fate. It is well known that, in adherent cultures, MSCs adopt a myofibroblast-like contractile phenotype and can differentiate along mesenchymal lineages in response to defined supplements [1,2]. Less is known about MSCs in vivo and even the lineage specification of their embryonic precursors is ill-defined [3,4]. MSCs occupy perivascular niches throughout the body [5] and in the bone marrow can undergo osteogenic differentiation and support hematopoiesis [6,7]. Although MSCs can express endothelial markers in vitro in response to growth factors [8], or to cell density-dependent Notch signals [9], their ability to form functional vascular endothelium and contribute to new blood vessel formation in vivo remains uncertain. We report that MSC fate is changed by perturbing mesenchymal regulators, which in turn stimulates neovascularization and their integration into functional blood vessels.
MSCs are derived predominantly from the mesodermal lineage, but also from endoderm by epithelial-mesenchymal transition and from ectodermal neural crest [10][11][12]. During development, the mesoderm forms distinct mesenchymal and hemato-endothelial lineages. Using embryonic stem cells directed toward mesendoderm, one group identified a common mesoderm-derived precursor for MSCs and endothelial cells, which they termed a mesenchymoangioblast [3,4]. Others described a bone marrow mesodermal progenitor cell population with dual mesenchymal and endothelial differentiation potential [13]. These data point to a mesodermal cell stage with potential to form mesenchyme or endothelium.
Platelet-derived growth factor (PDGF) receptors (PDGFR) are markers and critical regulators of mesenchyme [14][15][16]. Knockout mice showed that loss of PDGFRa or PDGF-A disrupts mesenchymal tissue formation, whereas loss of PDGFRb disrupts pericytes and smooth muscle [17,18]. Knockout of PDGFRa caused death of 50% of embryos before E10 and the rest shortly after birth [19], while in chick, signaling through PDGFRa was required for mesodermal cell migration [20]. We have shown that PDGFR signaling in MSCs regulates migration, proliferation, and cytoskeletal organization, through RhoA/Rho kinase (ROK) signaling [21] and by crosstalk with fibronectin (FN)-activated integrin a5b1 [22] and neuropilin-1 [23]. We showed that FN/a5b1 activates PDGFRb in the absence of PDGF growth factors, and is also required to potentiate PDGF-BB-mediated PDGFRb activation [22]. FN, a chordate innovation, is an extracellular adhesive glycoprotein [24], which controls the deposition of fibrillar matrices by mesenchymal cells [25], and thus tissue formation. FN-null mice are early embryonic lethal due to multiple cardiovascular defects [26]. PDGFRb signaling enhances FN expression [27], and together they are potent drivers of mesenchyme.
We have tested the hypothesis that disrupting mesenchymal regulators can alter the fate of human bone marrowderived MSCs. Cell cytoskeleton was modified by inhibiting PDGFRs or by depleting FN, within three-dimensional (3D) spheroids. Resulting MSCs were rounded rather than spindleshaped, with depleted smooth muscle a-actin (SMA) filaments and greatly reduced migratory capacity. They were mesenchymoangioblast-like with enhanced transcription factors such as EOMES, Foxh1, and Mixl1. These cells also exhibited marked upregulation of E-cadherin, Oct4A, and Nanog as well as endothelial markers platelet endothelial cell adhesion molecule 1 (PECAM-1) and vascular endothelial (VE)-cadherin and angiogenic growth factors. They had endothelial-like organization, and markedly enhanced neovascularization and integration into new functional blood vessels that were perfused by the circulation in vivo. Thus, perturbation of mesenchymal regulators modulates MSC fate and angiogenic potential in vivo. This discovery offers opportunities for therapeutic revascularization.

Quantitative RT-PCR Analysis
For RNA isolation; 12 identically cultured spheroids were pooled together for analysis. RNA was isolated using Trizol reagent (Life Technologies, Grand Island, NY, http://www.lifetechnologies.com) followed by digestion with RNase-free DNase (Promega, Madison, WI, http://www.promega.com). First strand cDNA synthesis was performed using AMV reverse transcriptase (Roche, Basel, Switzerland, http://www.roche.-com), and real-time quantitative PCR using GoTaq qPCR kit (Qiagen, Valencia, CA, http://www.qiagen.com). Gene expression was determined relative to GAPDH using the DC t method. All primer sequences are provided in Supporting  Information Table S1.

siRNA Knockdown
MSCs were transfected with 10 nM small interfering RNAs (siR-NAs) using Lipofectamine RNAi MAX reagent (Life Technologies, Grand Island, NY, http://www.lifetechnologies.com), then cultured for 24 hours in the presence of the reagent and siRNA. Transfected MSCs were then trypsinized and used to form spheroids in the presence of fresh Lipofectamine reagent and siRNA which remained in the culture at 37 C for 5 days. Validated siR-NAs were used to knockdown FN (1027417) (Qiagen, Valencia, CA, http://www.qiagen.com); scrambled siRNA (Qiagen, Valencia, CA, http://www.qiagen.com) was used as a control.

Immunoblotting
For protein isolation; 12 identically cultured spheroids were pooled together for analysis. Protein lysate isolation, immunoblotting, and quantification were performed as previously described [21]. Details of the antibodies used are given in Supporting Information Table S2.

Matrigel In Vitro Network and Implant Analysis
For network formation, spheroids were dissociated into single cells by incubation with 0.001% (wt/vol) collagenase type IV (Sigma Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) in phosphate buffered saline for 1 hour at 37 C, followed by Try-pLE Select (Life Technologies, Grand Island, NY, http://www.lifetechnologies.com) digestion for 10 minutes. MSCs were seeded onto growth factor reduced Matrigel (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com) and incubated at 37 C. For implants, spheroids were covered with Matrigel and incubated at 37 C.

Matrigel Plug In Vivo Angiogenesis Assay
Spheroids suspended in 200 ml growth medium and 500 ml of growth factor reduced Matrigel were injected subcutaneously into the flanks of 6-8-week-old C57BL/6 mice. Five mice were each injected with 10 control spheroids and five with 10 PDGFR-IV spheroids. Two weeks after injection, mice were sacrificed, plugs removed, and immediately fixed in 4% (wt/ vol) paraformaldehyde for 24 hours at 4 C, then incubated in 30% (wt/vol) sucrose solution for 24 hours at 4 C. Plugs were then embedded in OCT and 10 mm sections obtained for immunofluorescence analysis which was performed as previously described [28].
To study whether new blood vessels were functional and integrated with the circulation, 200 ml of 50 mg/ml FITCdextran 2,000 kDa (Sigma Aldrich, St. Louis, MO, http:// www.sigmaaldrich.com) was injected into the tail vein of mice, and allowed to perfuse for 10 minutes before the mice were sacrificed, and the plugs excised and processed, as above.

Whole Mount Immunofluorescence Analysis and Microscopy
Information Table S2. Images were collected on a Nikon C1 confocal using a TE2000 PSF inverted microscope, using 360 /NA 1.40 Plan Apo or 320/NA 0.50 Plan Fluor objectives and 33 confocal zoom. Different sample images detecting the same antibodies were acquired under constant acquisition settings. Images were processed using Nikon EZ-C1 FreeViewer v3.3 software. Brightfield images were collected on an Olympus BX51 widefield microscope, using a 310/NA 0.3 UPlan F1 objective. Images were captured with a CoolSNAP camera system and processed using MetaMorph imaging v5.0 software.

Statistical Analysis
Results are expressed as the mean 6 SD. Statistical differences between sets of data were determined using a paired Student's t test, with p < .05 considered significant.

RESULTS
PDGFRs regulate the formation of mesenchymal cells from mesoderm [14,16]. These cells in turn make tissues by depositing and embedding themselves within a FN-rich extracellular matrix [24,25]. FN/a5b1 is functionally linked with PDGFRb, both by activating PDGFRb in the absence of PDGF growth factors and by potentiating PDGF-BB-induced PDGFRb signaling [22]. Here, we used 3D spheroid cultures to test the hypothesis that cytoskeletal changes induced by inhibiting these functionally integrated mesenchymal regulators, PDGFR signaling and FN, revert MSCs toward a mesodermal progenitor with endothelial potential.

PDGFR Signaling Modulates MSC Spheroid Assembly and Organization
PDGFR signaling regulates the expression of contractile SMA filaments [14][15][16], which are not only a characteristic functional feature of vascular smooth muscle cells [30] but also abundantly expressed by MSCs [21], and are widely used as a distinctive mesenchyme marker. Using live cell imaging, we monitored the assembly of MSCs into control and PDGFR inhibitor-IV spheroids during the first 20 hours of formation (Supporting Information Video S1). While control MSC suspensions rapidly aggregated to form a spheroid, their assembly in the presence of PDGFR inhibitor-IV was markedly slower. After 6 hours of seeding, control MSCs formed a spheroid-like structure which contains a distinct outer ring of cells ( Fig.  1A(i)); this feature was not detected during the assembly of PDGFR inhibitor-IV spheroids. After 20 hours of seeding, control MSCs formed a spherical structure (Fig. 1A(ii)), whereas PDGFR inhibitor-IV spheroids had a larger diameter but more flattened structure. These distinctive organizational features were maintained throughout a 5-day culture period (Fig.  1A(iii)); MSCs within control 3D spheroids retained their characteristic spindle-shaped morphology and expressed widespread abundant extended SMA filaments and FN ( Fig. 1B(i)). However, MSCs within PDGFR inhibitor-IV spheroids were more rounded, and SMA and FN abundance were markedly reduced ( Fig. 1B(i)). Within the core of the spheroids, MSCs appeared viable and expressed abundant SMA filaments ( Fig.  1B(ii)), which were diminished by PDGFR inhibitor-IV. Compared to control spheroids at day 5, PDGFR inhibitor-IV spheroids expressed 2.9 6 0.3-fold lower SMA protein (Fig. 1C).
PDGFRb is known to regulate FN expression [27]. Having shown that PDGFR inhibitor-IV spheroids contained markedly reduced SMA and FN, we used siRNA knockdown to investigate how FN influences the fate of MSCs within 3D spheroids. Compared to spheroids formed from scrambled siRNA knockdown MSCs, spheroids assembled by FN knockdown MSCs displayed a 79% 6 9% reduction in FN protein ( Fig.  1D) expression, respectively, by day 5 of culture. Like the PDGFR inhibitor-IV treated spheroids (Fig. 1A), these FN knockdown spheroids had larger diameters than control spheroids (Fig. 1E). The FN knockdown MSCs also appeared more rounded and in closer contact than control cells and, compared to scrambled siRNA knockdown spheroids, they expressed markedly reduced SMA filaments ( Fig. 1B(iii)) and had a 70% 6 8% decrease in SMA protein expression (Fig.  1F). FN knockdown spheroids also displayed significantly diminished PDGFRb phosphorylation (Y751 80% 6 7% and Y1021 85% 6 8% reduction, respectively) ( Fig. 1G, 1H). These data show that PDGFRs, and FN which activates PDGFRs, both regulate SMA contractile filaments and coordinate MSC spheroid organization and contractility.
Thus, in comparison to adherent MSC cultures, spindleshaped MSCs within 3D control spheroids increased the expression of Oct4A and Nanog transcripts, although only low levels of protein were detected. However, inhibition of PDGFR signaling or FN knockdown within spheroids, which induced more rounded MSCs, prominently upregulated Oct4A and Nanog protein levels.

MSC Spheroids Display a Mesendoderm Expression Profile
Having established that MSCs cultured as 3D spheroids markedly upregulate the expression of the pluripotency transcrip-tion factors Nanog and Oct4A, we investigated expression levels of mesendoderm and endoderm transcription factors expressed by spheroids at day 5 (Fig. 3). Bipotent mesendoderm cells are precursors for both endoderm and mesoderm during ESC differentiation [31]. Mesendoderm has been characterized as goosecoid (Gsc) 1 E-cadherin (CDH1) 1 PDGFRa 1 which can give rise to Gsc 1 CDH1 1 PDGFRa 2 endoderm and Gsc 1 CDH1 2 PDGFRa 1 mesoderm progenitors [31]. Compared to untreated adherent MSCs, control spheroids expressed increased transcripts of mesendoderm transcription factors; Nodal, Mixl, EOMES, Foxh1, and Gsc (Fig. 3A), but PDGFR inhibitor-IV spheroids expressed higher levels of Mixl, EOMES, and Foxh1 (Fig. 3A). Control spheroids also expressed increased levels of the endoderm transcription factors; Foxa1, Whole mount immunofluorescence analysis of (i, ii) control spheroids (con) and PDGFR inhibitor-IV spheroids (IV), (iii) scrambled control siRNA spheroids (Scr #) and FN knockdown spheroids (FN #) cultured for 5 days, showing FN (red) and SMA (green) expression, with DAPI-stained nuclei (blue). Scale bars 5 50 mm. (C): Immunoblot analysis of SMA expression within control (Con) and PDGFR inhibitor-IV (IV) spheroids after 5 days culture, with b-actin as a loading control. Histogram shows SMA expression relative to b-actin and normalized to control spheroid level. *, p < .001 compared with day 5 control spheroids, using paired t test n 5 3 separate experiments, error bars represent SD. (D): Immunoblot analysis of FN expression within control scrambled siRNA spheroids (Scr #) and FN knockdown spheroids (FN #) cultured for 5 days, with b-actin as a loading control. RNA expression is relative to GAPDH and normalized to the level of scrambled control spheroids at day 5. Histogram shows protein expression relative to b-actin and normalized to control siRNA spheroid level. *, p < .001 compared with control siRNA spheroids, using paired t test n 5 3 separate experiments, error bars represent SD. (E): Bright-field images of spheroids assembled using MSCs treated with scrambled (Scr) or FN small interfering RNAs (siRNA) and cultured for 5 days. Scale bars 5 200 mm. (F-H): Immunoblot analysis of SMA, PDGFRb Y751, and Y1021 phosphorylation levels, within scrambled control siRNA spheroids (Scr #) and FN knockdown spheroids (FN #) after 5 days culture, with bactin or PDGFRb as a loading control. Histograms show SMA expression relative to b-actin and PDGFRb Y751 and Y1021 relative to PDGFRb, normalized to scrambled control siRNA spheroid level. *, p < .001 compared with control spheroids, using paired t test n 5 3 separate experiments, error bars represent SD. Abbreviations: FN, fibronectin; PDGFR, platelet-derived growth factor receptor.
To investigate further the increased expression of pluripotency and endoderm proteins within PDGFR inhibitor-IV spheroids, a stem cell proteome array was used to detect relative expression levels simultaneously (Supporting Information Fig.  S3A). Compared to control spheroids, PDGFR inhibitor-IV spheroids showed a distinct increase in the pluripotent markers; Oct4, Nanog, and Sox2, and endoderm markers; Sox17, AFP, Quantitative RT-PCR analysis of Oct4A and Nanog expression within high-density adherent mesenchymal stromal/stem cells (MSCs) (Adh Con), high-density adherent MSCs exposed to PDGFR inhibitor-IV (Adh IV), control spheroids (Sph Con), and PDGFR inhibitor-IV spheroids (Sph IV), cultured for 5 days. Data are relative to GAPDH and normalized to adherent control levels. *, p < .001 compared with adherent controls, # , p < .001 compared with control spheroids, using paired t test n 5 3 separate experiments, error bars represent SD. (B): Quantitative RT-PCR analysis of Oct4A and Nanog expression within control spheroids (Sph Con) and PDGFR inhibitor-IV spheroids (Sph IV), cultured for 5 days, and human embryonic stem cell line Hues1. Data are relative to GAPDH and normalized to control spheroid levels. *, p < .001 compared with control spheroids, using paired t test n 5 2 separate experiments, error bars represent SD. (C): Quantitative RT-PCR analysis of Oct4A and Nanog expression within control spheroids (Con) and PDGFR inhibitor-IV spheroids (IV), cultured for cultured for 1, 2, 3, and 5 days. Data are relative to GAPDH and normalized to the level of control spheroids at day 1. *, p < .001 compared with day 1 control spheroids, using paired t test n 5 3 separate experiments, error bars represent SD. (D, E): Whole mount immunofluorescence analysis of (D) control spheroids (Con) and PDGFR inhibitor-IV spheroids (IV), (E) scrambled control siRNA spheroids (Scr #), and FN knockdown spheroids (FN #) cultured for 5 days, showing Nanog (red) and Oct4A (green) expression, with DAPI-stained nuclei (blue). Scale bars 5 50 mm. (F-I): Immunoblot analysis of Oct4A and Nanog expression within (F, G) control (Con) and PDGFR inhibitor-IV (IV) spheroids and (H, I) scrambled control siRNA spheroids (Scr #) and FN knockdown spheroids (FN #) after 5 days culture, with b-actin as a loading control. Histograms show Oct4A and Nanog expression relative to b-actin and normalized to control spheroid or scrambled control siRNA control spheroid levels. *, p < .001 compared with control spheroids, using paired t test n 5 3 separate experiments, error bars represent SD.
Foxa2, and CDH1. These data verify that PDGFR inhibitor-IV spheroids express mesendoderm and endoderm markers including CDH1, indicating features of a mesendoderm progenitor.
Proteome array analysis demonstrated that, compared to scrambled siRNA knockdown spheroids, FN knockdown spheroids also upregulated Oct4, Nanog, and Sox2 as well as endoderm markers; Sox17, AFP, Foxa2, and CDH1 (Supporting Information Fig. S3B). Thus, FN knockdown also resulted in features of a mesendoderm progenitor, although the relative levels of these markers were lower than those induced by PDGFR inhibition (Supporting Information Fig. S3A). These data indicate that inhibition of PDGFR signaling, or FN depletion, within spheroids reverts MSCs toward a more multipotent state.

MSC Spheroids Exhibit Endothelial Characteristics
Using mesoderm progenitor cells derived from ESCs, activation of PDGFRb signaling can induce differentiation to vascular smooth muscle cells, which express abundant SMA, whereas stimulation with VEGF-A can induce differentiation toward endothelial cells [35]. In addition, adherent MSCs cultured in two-dimensional (2D) at high density acquire a more rounded cobblestone-like morphology, and can undergo differentiation toward an endothelial lineage [9]. We therefore explored the possibility that MSCs within PDGFR inhibitor-IV spheroids, which lack PDGFR signaling, have depleted SMA and are at high density, may exhibit an endothelial rather than mesenchymal disposition. Accordingly, the expression of two characteristic endothelial markers, PECAM-1 and VE-cadherin, was determined in PDGFR inhibitor-IV spheroids.
FN knockdown spheroids, which caused similar effects on cell shape and phenotype (Fig. 1), increased PECAM-1 protein (1.9 6 0.17-fold) (Fig. 4F). The effects of EGFR, FGFR, VEGFR, ROK, Rac, MEK, or PI3K signaling inhibitors on PECAM-1 and VE-cadherin transcripts were also examined, but only ROK or Rac inhibition markedly increased PECAM-1 expression (Fig.  4G). Rac or ROK-inhibited spheroids were shown to increase , within high-density adherent mesenchymal stromal/stem cells (MSCs) (Adh con), high-density adherent MSCs exposed to PDGFR inhibitor-IV (Adh IV), control spheroids (Sph con), and PDGFR inhibitor-IV spheroids (Sph IV), cultured for 5 days. Data are relative to GAPDH and normalized to adherent control levels. *, p < .001 compared with adherent controls, # , p < .005 compared with control spheroids, using paired t test n 5 3 separate experiments, error bars represent SD. (D): Quantitative RT-PCR analysis of CDH1 within control spheroids (Sph Con) and PDGFR inhibitor-IV spheroids (Sph IV), cultured for 5 days, and human ESC line Hues1. Data are relative to GAPDH and normalized to control spheroid levels. *, p < .001 compared with spheroid controls, using paired t test  Fig. S4C); however, this level was lower than that expressed by PDGFR inhibitor-IV or FN-depleted spheroids.

MSC Spheroids Secrete Angiogenic Factors
To identify proteins secreted by MSC spheroids which may regulate the endothelial features identified, we used a human angiogenesis proteome array to analyze supernatants from adherent 2D MSCs at high density, or 3D MSC spheroids, cultured in the presence or absence of PDGFR inhibitor-IV for 3 days (Fig. 5).
Taken together, these data demonstrate that the 3D culture of MSCs as spheroids markedly increased the secretion of potent angiogenic-related proteins, and dramatically upregulated the expression of PECAM-1 and VE-cadherin transcripts, compared with 2D culture of MSCs at high density. Furthermore, PDGFR inhibitor-IV and FN-depleted spheroids displayed an increase in PECAM-1 protein when compared with control spheroids, indicating that inhibition of mesenchymal drivers is a primary determinant in the acquisition of these endothelial features.

MSC Spheroids Facilitate Blood Vessel Formation Within Matrigel
Endothelial cells form capillary-like network structures when cultured on Matrigel [36]. We therefore tested the ability of MSCs which had been cultured as spheroids, to form similar Matrigel-induced networks. Control and PDGFR inhibitor-IV spheroids were cultured for 5 days, then MSCs dissociated into single cells and cultured on Matrigel under identical conditions (i.e., no PDGFR inhibitor-IV added) for a further 2 days. PECAM-1 and VE-cadherin expressions were then determined by immunofluorescence analysis (Fig. 6A, 6B). MSCs derived from control spheroids readily established network structures with elongated branch points. These networks were composed of spindle-shaped cells which expressed both PECAM-1 and VE-cadherin (Fig. 6A). However, some cells expressed neither PECAM-1 nor VE-cadherin ( Fig. 6A (ii, iv)). Similarly MSCs derived from PDGFR inhibitor-IV spheroids also readily formed widespread network structures, but in contrast, these had short branch points and the networks were composed of rounded cells, surrounded by distinctive PECAM-1 and VE-cadherin expression (Fig. 6B). Notably in this case, from >25 different images captured, every cell was positive for both PECAM-1 and VE-cadherin (Fig. 6B (ii-iv)).
As a prelude to examining the effects of MSC spheroids on angiogenesis in vivo, we cultured MSC spheroids within a 3D Matrigel plug in vitro. Control and PDGFR inhibitor-IV spheroids were cultured for 5 days. The intact spheroids were then implanted into Matrigel and cultured under identical conditions (i.e., no PDGFR inhibitor IV added) for a further 5 days. Control spheroids rapidly developed outgrowths of spindle-shaped cells by day 1, which became extensive by day 5 (Fig. 6C). These cellular outgrowths were negative for PECAM-1 and VE-cadherin ( Fig. 6D (i-ii)), but positive for SMA (Fig. 6D (iii)). In contrast, PDGFR inhibitor-IV spheroids developed a ring of nonmigratory rounded cells around the spheroid periphery (Fig. 6E), which exhibited abundant PECAM-1 and VE-cadherin expression (Fig.   Figure 5. Angiogenic factors secreted by mesenchymal stromal/stem cell (MSC) spheroids. A human angiogenesis array kit (ARY007) (R&D Systems, Minneapolis, MN, http://www.rndsystems.com) was used to determine simultaneously the relative expression levels of 55 different angiogenic proteins, secreted by high-density adherent MSCs (Adh Con), high-density adherent MSCs exposed to PDGFR inhibitor-IV (Adh IV), control spheroids (Sph Con), and PDGFR inhibitor-IV spheroids (Sph IV), cultured for 3 days. Prior to analysis, adherent MSCs and spheroids were cultured for 2 days, then fresh growth medium added and cultured for a further 3 days for analysis. Supernatants from 24 identical adherent or spheroid cultures were pooled for analysis. Data are normalized to the DNA content of each MSC culture, and relative to high-density adherent MSCs (Adh con) *, p < .001 compared with adherent controls, # , p < .005 compared with control spheroids, using paired t test n 5 1 experiment, error bars represent SD between two repeats.

PDGFR-Inhibited MSCs Integrate with Functional In Vivo Blood Vessels
To determine the effects of MSC spheroids on in vivo angiogenesis, 10 control or PDGFR inhibitor-IV spheroids were suspended in Matrigel without any additional growth factors, and implanted into mice for 14 days, then human and murine PECAM-1 expression determined by immunofluorescence (Fig. 7). In addition, new functional blood vessels connected to the circulation were identified by FITC-dextran perfusion.
Examination of excised Matrigel plugs revealed that the control spheroids contained few human PECAM-1 positive cells present (Fig. 7A), but these spheroids were infiltrated by murine PECAM-1 positive blood vessels; the absence of FITC-dextran staining suggested that these vessels were not attached to the circulation (Fig. 7B). Similarly, the Matrigel surrounding the control spheroids was permeated with murine PECAM-1 positive blood vessels; however, few human PECAM-1-positive cells were observed (Fig. 7B), and these vessels were also not associated with FITC-dextran perfusion (Fig. 7D). Analysis of the peripheral tissue revealed numerous blood vessels (Fig. 7C), but only minimal human PECAM-1-positive staining was detected (Fig. 7D), which again was not detected near FITC-dextran staining (Fig. 7F).
In marked contrast, PDGFR inhibitor-IV spheroids contained numerous human PECAM-1-positive cells and these spheroids were also infiltrated by murine blood vessels (Fig.  7G), which were connected to the host circulation as indicated by abundant FITC-dextran perfusion (Fig. 7H). The Matrigel surrounding PDGFR inhibitor-IV spheroids was also permeated with both murine and human PECAM-1-positive cells, which associated together in vascular-like assemblies (Fig. 7I). Some of these human PECAM-1-positive vascular-like assemblies within Matrigel were perfused with FITC-dextran showing PECAM-1 (red) and VE-cadherin (green) expression and (D (iii)) SMA (green) expression, with DAPI-stained nuclei (blue). Scale bars 5 200 mm (i) (D (iii)) and 50 mm (ii) (F (iii)). (Fig. 7J). Human PECAM-1-positive cells derived from PDGFR inhibitor-IV spheroids were also detected in the peripheral tissue, where they clearly integrated into blood vessels with murine PECAM-1 positive cells (Fig. 7K). These human PECAM-1-positive vessels in tissue were connected to the host vasculature and perfused with FITC-dextran (Fig. 7L). Thus, PDGFR inhibitor-IV spheroids are a potent source and stimulant for neovascularization.

DISCUSSION
Mesenchymal cells make tissues by depositing and embedding themselves in an extracellular matrix and growth factor-rich microenvironment, which they remodel and maintain throughout life. Although this extrinsic niche dictates the behavior of MSCs, their therapeutic potential remains severely constrained by lack of mechanistic insight into how the niche controls their fate. Here, we have shown that disrupting the extracellular matrix molecule FN or the functionally linked PDGFR, which together regulate mesenchyme [22], converts MSCs rapidly from SMA-rich spindle-shaped contractile cells to rounded E-cadherin-rich cells. These cells exhibit enhanced expression of markers for pluripotency, mesendoderm, endoderm, and angiogenic markers, and display potent angiogenic behavior in vivo. Thus, blocking natural mesenchymal signals offers an effective strategy for reprogramming mesenchymal cells and for therapeutic revascularization.
We have developed a novel approach to modulate MSC fate that does not require the use of viral vectors or exogenous DNA. Using a spheroid MSC culture model that recapitulates physiological features of a 3D cellular environment and cell-cell interactions, MSCs were induced toward a more multipotent state. Untreated MSCs within spheroids retained their spindle-shape and mesenchymal character with a SMArich cytoskeleton and profuse FN matrix. Although culturing MSCs as 3D spheroids was sufficient to induce some upregulation of mesendodermal and endodermal markers, inhibition of PDGFRs or FN knockdown induced rapid cell rounding along with significant further induction of pluripotency markers Oct4A and Nanog, demonstrating that MSCs have the potential to revert to a premesenchymal state, which is accentuated when mesenchymal signals are inhibited. We have previously shown using 2D cultured MSCs that PDGFR signaling inhibition changes their shape and cell fate [28]. In this study, depletion of the extracellular matrix component FN was also shown to modulate MSC shape and direct their fate, emphasizing the crucial role played by cell shape in MSC fate decisions. Furthermore, we were able to demonstrate that these cells had strong angiogenic potential in vitro and in vivo. Thus, disrupting mesenchymal signals in these 3D cultures induced a mesenchymoangioblast-like state [3,4]. It has been unclear whether MSCs possess the ability to transdifferentiate to functional endothelial cells; most reports have relied on exogenous VEGF-A supplementation approaches [8]. We previously showed that high-density cultures exhibited Notch-dependent endothelial potential in vitro and in CAM assays [9]. Here, by demonstrating that MSCs derived from PDGFR-inhibited spheroids are able to induce, and integrate with functional blood vessels perfused by the circulation in vivo, we have directly shown their angiogenic and vascular potential. By inhibiting mesenchymal signals, we were able to induce endothelial fate, including the reinstigation of robust cell-cell interactions, as judged by E-cadherin expression. Thus cell-cell contacts within PDGFR-inhibited spheroids may be just as important in driving the angiogenic features, as increased expression of embryonic transcription factors.

CONCLUSION
In summary, we have shown that the mesenchymal fate of MSCs can be modulated by re-engineering the relationship between cells and their local matrix, without the need for viral delivery of exogenous transcription factors. By blocking mesenchymal drivers, these cells can be reverted to a mesenchymoangioblast-like state and thence to functional endothelial-like cells in vivo. As these strategies target the natural mechanisms that manipulate mesenchymal fate, they have great potential for future revascularization therapies.