De novo hem- and lymphangiogenesis by endothelial progenitor and mesenchymal stem cells in immunocompetent mice

Abstract

Cellular pro-angiogenic therapies may be applicable for the treatment of peripheral vascular diseases. Interactions between mesenchymal stem cells (MSCs) and endothelial progenitor cells (EPCs) may provide such a treatment option. With the exception of some studies in man, experiments have only been performed in immunodeficient mice and rats. We studied an immunocompetent syngeneic mouse model. We isolated MSCs from bone marrow and EPCs from the lung of adult C57/Bl.6 mice and co-injected them in Matrigel subcutaneously in adult C57/Bl.6 mice. We demonstrate development of both blood vessels and lymphatics. Grafted EPCs integrated into the lining of the two vessel types, whereas MSCs usually did not incorporate into the vessel wall. Injections of each separate cell type did not, or hardly, reveal de novo angiogenesis. The release of VEGF-A by MSCs has been shown before, but its inhibitors, e.g., soluble VEGF receptors, have not been studied. We performed qualitative and quantitative studies of the proteins released by EPCs, MSCs, and cocultures of the cells. Despite the secretion of VEGF inhibitors (sVEGFR-1, sVEGFR-2) by EPCs, VEGF-A was secreted by MSCs at bioavailable amounts (350 pg/ml). We confirm the secretion of PlGF, FGF-1, MCP-1, and PDGFs by EPCs/MSCs and suggest functions for VEGF-B, amphiregulin, fractalkine, CXCL10, and CXCL16 during MSC-induced hem- and lymphangiogenesis. We assume that lymphangiogenesis is induced indirectly by growth factors from immigrating leukocytes, which we found in close association with the lymphatic networks. Inflammatory responses to the cellular markers GFP and cell-tracker red (CMPTX) used for tracing of EPCs or MSCs were not observed. Our studies demonstrate the feasibility of pro-angiogenic/lymphangiogenic therapies in immunocompetent animals and indicate new MSC/EPC-derived angiogenic factors.

Electronic supplementary material

The online version of this article (doi:10.1007/s00018-013-1460-8) contains supplementary material, which is available to authorized users.

Introduction

Supply with nutrients and removal of metabolites are of utmost importance for tissue homeostasis. These functions are secured by the vascular system: both the blood vessels and the lymphatics. Construction of efficient vascular networks is a major issue of tissue engineering. Thereby, the systemic application of pro-angiogenic factors, such as vascular endothelial growth factor-A (VEGF-A), to induce hemangiogenesis (development of blood vessels), or VEGF-C, to induce lymphangiogenesis, seems to be risky because of possible adverse effects [1]. In order to induce effects only locally, the application of mesenchymal stem cells (MSCs) has evolved as a possible method. MSCs were initially studied because they can give rise to a spectrum of differentiated cell types. It was then observed that MSCs regulate the microenvironment through trophic and immunomodulatory factors [2]. Thereby, MSCs provide an angiogenic environment by producing the growth and differentiation factors VEGF-A, placenta growth factor (PlGF-2), hepatocyte growth factor (HGF), the matrix metalloproteases MMP-9 and MMP-3, and induce blood vessel formation when co-injected with endothelial progenitor cells (EPCs) in mice [3]. The development of lymphatics has not yet been studied in such experiments. Lymphatics are of great importance for the circulation of immune cells [4], however, co-injections of MSCs and EPCs have mainly been performed in immunodeficient mice and rats [3, 5–8]. Alternatively, autologous tissue-engineered grafts seeded with EPCs and MSCs were studied in sheep [9], and studies are already being performed in humans [10, 11]. Given the fact that MSCs have immunomodulatory functions, preclinical studies should be performed in immunocompetent models.

Recently, we isolated EPCs from the lungs of adult Balb/c and C57/Bl.6 mice. The cells express numerous markers characteristic of vascular endothelial cells, and differentiate in vitro and in vivo into blood endothelial cells (BECs) and lymphatic endothelial cells (LECs) [12]. This was verified in clonal EPCs obtained from limiting dilution assays. Mass spectroscopic analysis of the proteins secreted from these cells identified more than 35 typical markers of endothelial and hematopoietic precursor cells [13]. The hem- and lymphangiogenic potential of our lung-derived EPCs was studied in vivo using Matrigel supplemented with fibroblast growth factor-2 (FGF-2) and VEGF-A [12].

The differentiation potential of human and murine endothelial cells injected into nude mice is well documented [6, 14, 15], although the extent of homing into the vasculature may vary considerably [16]. The co-injection of EPCs and MSCs may provide a superior angiogenic microenvironment [6], but, as pointed out above, should be performed in immunocompetent mice, since immune cells are potent modulators of both hem- and lymphangiogenesis [17, 18]. Here, we have used lung-derived EPCs and bone marrow-derived MSCs in a syngeneic mouse model (B57/Bl.6) to study both hem- and lymphangiogenesis. We localized the precursor cells in the vascular walls (EPCs) and the microenvironment (MSCs), and characterized pro- and anti-angiogenic factors released by the two cell types. We show that the co-injected cells produce new blood vessels and lymphatics, and that VEGF-A, expressed by MSCs, is present at bioavailable amounts, despite the secretion of naturally occurring VEGF inhibitors (sVEGFR-1, sVEGFR-2) by EPCs. Furthermore, we identified additional proangiogenic proteins, whose functions have to be characterized in this experimental setting in future studies.

Materials and methods

Animals

Isolation of lung-derived EPCs and bone marrow-derived MSCs as well as the transplantation experiments were performed with C57/Bl.6 mice. For the transplantation experiments, we used female mice (age 8–12 weeks). All experiments were approved by the local institutional animal care committee and by the Lower Saxony state council on animal care (LAVES), and corresponded to the requirements of the American Physiological Society.

Isolation of endothelial progenitor cells and mesenchymal stem cells

All cells were isolated from C57/Bl.6 mice. The isolation of endothelial progenitor cells (EPCs) from the lungs of adult mice has been described before [12]. Mesenchymal stem cells (MSCs) were isolated from bone marrow as described by Soleimani and Nadri [19]. Cells were used between passages 6 and 10 (MSCs) and passages 8 and 24 (EPCs).

Cell growth and co-culture of EPCs and MSCs

As in our previous studies [12, 20], MSCs and EPCs were cultured in DMEM enriched with FCS, either alone or in direct co-culture in 24- and six-well plates (Nunc). For the production of conditioned supernatants, fetal calf serum (FCS) was omitted. EPCs were cultured in gelatine-coated wells or cell culture flasks. For the generation of conditioned supernatants, cells were seeded (in triplicates) at a density of 1.5 × 10⁴ cells/cm². Supernatants were harvested on day 3 (72 h), centrifuged and stored at −20 °C for ELISA.

Green fluorescent protein transfection of primary cells

The production of primary cells transfected with lentiviral constructs encoding green fluorescent protein (GFP) was described in detail before [12].

Fluorescence labeling of MSCs in vitro

For some experiments, cultured MSCs were incubated with 10 mM of the fluorescence dye “cell-tracker red” CMPTX for 30 min at 37 °C according to the instructions of the manufacturer (Invitrogen) and as reported before [21]. Cells were then trypsinized and mixed with EPCs for some in vivo studies. Red fluorescence of the cells was still detectable after 7–9 days.

Quantification of VEGF-A, VEGF-C, PlGF, soluble VEGFR-1/R-2 in cell supernatants

Quantification by sandwich ELISA of total VEGF-A as well as free, bioactive VEGF-A was performed as described in our previous study [22]. Mouse VEGF-A164 was used as a standard. Briefly, total mouse VEGF-A was measured with a standard sandwich ELISA and bioactive VEGF-A was determined by using immobilized, soluble sFlt-1 (sVEGFR-1) as a capture molecule in a sandwich ELISA. Measurements of mouse PlGF, sFlt-1 and sFLK-1 (sVEGFR-2) were made with duo set ELISAs purchased from R&D Systems (Rüsselsheim, Germany; also see Bergmann et al. [23]). Mouse VEGF-C was measured with a sandwich ELISA reported before [24]. Cell supernatants were at least diluted 1:3 with the recommended diluents. All assays were performed in duplicates.

Mouse angiogenesis array kit

A mouse angiogenesis array kit (R&D Systems) was used to detect 53 secreted proteins in cell supernatants related to angiogenesis or endothelial cell growth and differentiation. The supernatants were depleted from floating cells by centrifugation and stored at −20 °C until analysis. A volume of 1,000 μl conditioned medium was used for each single membrane. The membranes were exposed to X-ray films for 1–3 min. The X-ray film was transmission scanned (Epson Perfection V750) and dots were quantified based on their pixel density (ImageJ 1.40 g; National Institutes of Health, Bethesda, Maryland, USA).

Assessment of vessel formation in vivo

For the in vivo experiments, MSCs and EPCs were trypsinized and counted. For the experiments with each single cell type, 1 × 10⁶ cells were used; or 1.5–2 × 10⁶ cells when the two cell types were combined. Cells were centrifuged and the pellets were dissolved in 300 μl of cold Matrigel (without the addition of growth factors). Matrigel with or without cells was injected subcutaneously into the left and right dorsal lumbar region (two injections per animal). The mice were shaved and mildly sedated (ketamine 100 mg/kg bw, ip) before the injection. After 7–9 days, the animals were killed, the Matrigel plugs photographed in situ, excised and fixed with 4 % paraformaldehyde (PFA), immersed with saccharose and embedded in tissue freeze medium (Neg-50, Richard-Allan Scientific, Kalamazoo, MI, USA). The numbers of experiments were: Matrigel without cells (11), Matrigel with EPCs (5), Matrigel with MSCs (6), Matrigel with EPCs and MSCs (26). In the latter group, we used GFP-transfected EPCs in four experiments, GFP-transfected MSCs in four experiments, and in eight experiments the MSCs were labeled with cell-tracker red.

In situ microscopy

Vascularization of Matrigel plugs was documented in situ with a stereomicroscope (Stemi SV11, Zeiss, Oberkochen, Germany) equipped with a Sony DSC-S75 digital camera. Magnification two to fivefold.

Histology and immunohistology

Frozen specimens were sectioned at 16–20 μm and mounted on slides. Hematoxylin and eosin (H&E) staining was performed according to Meyer. For the immunofluorescence studies, non-specific binding of Matrigel proteins was blocked by the incubation with 2 % bovine serum albumin (BSA) for 1 h before the incubation with the primary antibodies. Primary antibodies were: anti-mouse CD31 (rat clone MEC13.3; 1:50, BD), anti-GFP (rabbit polyclonal, 1:1000, Abcam, UK), anti-podoplanin (Syrian hamster clone 8.1.1, 1:1000, Hybridoma Bank, Iowa), anti-Prox1 (rabbit polyclonal, 1:500, Reliatech, Germany), anti-αSMA (rabbit polyclonal, 1:200, Abcam, UK), anti-mouse CD45 (rat monoclonal, 1:50, BD), anti-F4/80 (rat monoclonal, 1:100, Sanbio, The Netherlands) and anti-mouse PDGFR β (rat monoclonal, 1:50, Angio-Proteomie, USA). After incubation with the primary antibody for 1 h, sections were rinsed, and the secondary antibodies were applied: goat anti-rat Alexa594, donkey anti-rabbit Alexa488, goat-anti hamster Alexa594, goat anti-rat Alexa488 (all from Invitrogen). In some experiments, MSCs were detected by their red fluorescence based on their labeling with cell tracker red. DAPI was used to stain all nuclei blue in the sections. The sections were mounted under coverslips with Fluoromount-G (Southern Biotechnology, USA) and studied with Axio Imager Z1 (Zeiss).

Quantitative analyses and statistics

Images for quantification blood and lymphatic vessels were analyzed in microscopic fields at 400× magnification (three different animals for each group [EPC + MSC, only Matrigel, only MSC]; 123 and 42 fields for blood vessels and lymphatics, respectively). The number of blood vessels and lymphatics per microscopic field (area) were evaluated with the Kruskal–Wallis test followed by Dunn’s post hoc test. To study immunological effects of GFP or cell-tracker-red we performed quantitative analyses of immune cells. We compared each two groups of Matrigel plugs. One group contained cells without any labeling (MSC or EPC/MSC [three animals each, 19 fields]), the other group comprised GFP-transfected MSC or MSC with cell-tracker-red in combination with EPC [three animals each, 20 fields]. CD45+ leukocytes were counted in microscopic fields at 200× magnification. In the same way, we compared the number of Prox1⁺ lymphatic endothelial cells in the two groups (MSC with cell-tracker red/EPC vs. MSC/EPC without any dye) with an unpaired t test (Mann–Whitney test). Statistical analyses were carried out with GraphPad Prism 5 software (Version 5.03; GraphPad Software, Inc., La Jolla, CA, USA). Data are presented as mean ± SEM. Statistical significance is indicated in the figures by asterisks: * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

Results

In vivo transplantation

By subcutaneous injection into C57/Bl.6 mice we tested the pro-angiogenic capacity of EPCs and MSCs in vivo. Each single cell type or a mixture of the two was mingled with Matrigel and injected into the dorsal lumbar region of 8 to 12-week-old mice. In eight experiments, the MSCs were pre-labeled with a red-fluorescent cell tracker, and in four experiments we used GFP-transfected EPCs. After 7–9 days, the specimens were studied. In all of the 26 experiments employing the combination of EPCs and MSCs, a dense vascular network was seen in the Matrigel plug, whereas in the Matrigel controls and in the experiments with each single cell type blood vessels could not, or hardly, be detected with the stereomicroscope (Fig. 1a–g). The formation of new blood vessels was efficiently induced only by the combination of the two progenitor cell types. This was confirmed by immunohistological studies with anti-CD31/PECAM-1 antibodies, which recognize both blood vascular and lymphatic endothelial cells (Fig. 1h–l).

Fig. 1.

Differential in vivo formation of vascular networks by EPCs and MSCs. Typical results obtained 7–9 days after subcutaneous injection into C57/Bl.6 mice. a Control Matrigel plug without cells. b Mixture of MSCs and EPCs (2 × 10⁶ cells, ratio 1:1) in Matrigel. c Matrigel plug without cells. d 1 × 10⁶ EPCs. e 0.5 × 10⁶ MSCs. f, g Mixture of 2x10⁶ EPCs and MSCs. h–l CD31 staining of representative plugs. h Matrigel plug without cells. i 1 × 10⁶ EPCs. k 0.5 × 10⁶ MSCs. l Mixture of 2 × 10⁶ EPCs and MSCs. Numerous CD31⁺ vessels are present in the specimens of EPC/MSC mixtures. Very rarely, MSCs alone induced vessel formation. No vessels very found in EPC specimens and Matrigel controls. Bar 30 μm in h–l)

In order to study if the vessels in the grafts were graft- or host-derived, we used lentiviral GFP-transfection of EPCs in combination with anti-GFP antibodies. We regularly found a mosaic of GFP⁺ and GFP⁻ endothelial cells in CD31⁺ vessels in the Matrigel plugs, showing functional interactions of host-derived vessels with the grafted EPCs (Fig. 2, Suppl. video 1). The majority of the endothelial cells in the vessels shown in Fig. 2 were obviously derived from the host.

Fig. 2.

Functional integration of EPCs into the vasculature. Matrigel plugs containing EPCs and MSCs (ratio 1:1) were evaluated. EPCs were pre-labeled by lentiviral GFP expression. Immunofluorescence studies with anti-CD31 (red) and anti-GFP (green) antibodies. a, d Merged pictures. b, e Anti-CD31. c, f Anti-GFP. Note integration of EPCs (arrows) into CD31⁺ vessels. Bars 30 μm in a–c and 50 μm in d–f)

To the best of our knowledge, development of lymphatics has not yet been studied in this type of experiment. We therefore used antibodies against the two LEC markers podoplanin and Prox1. We found GFP⁺ EPCs in podoplanin⁺ lymphatics (Fig. 3, Suppl. video 2). Co-staining with antibodies against CD31 and Prox1 confirmed the presence of numerous lymphatics in the Matrigel plugs (Fig. 4a–d), although this staining does not differentiate between graft- and host-derived cells. Quantification of all CD31⁺ vessels, as well as the podoplanin⁺ lymphatics, revealed a significantly higher number of blood vessels and lymphatics in the EPC/MSC co-injections (Fig. 3g, h). Previous studies provided evidence for inflammation-induced lymphangiogenesis. We therefore studied the infiltration of the Matrigel plugs by host leukocytes, and observed an intimate contact of the newly formed lymphatics with CD45+ leukocytes (Fig. 4e–g). A subgroup of these leukocytes was made up of F4-80- and CD11b-positive macrophages (Fig. 4h, i). Our studies provide evidence for de novo lymphangiogenesis induced by the co-injection of lung-derived EPCs and bone marrow-derived MSCs, and their interaction with host-derived leukocytes. To test the functionality of the lymphatics, we injected Berliner blue into two of the plugs, hoping to identify lymphatic collectors that drain the dermal territory. We were not successful, however; these two negative experiments do not rule out the possibility that there may be functional connections to the host lymphatics.

Fig. 3.

Integration of EPCs into lymphatics. Matrigel plugs containing EPCs and MSCs (ratio 1:1) were evaluated. EPCs were pre-labeled by lentiviral GFP expression (green). Immunofluorescence studies with anti-podoplanin (red) and anti-GFP (green) antibodies. a, d Merged pictures. b, e Anti-podoplanin (lymphatic marker). c, f Anti-GFP. Note integration of EPCs (arrows) into podoplanin⁺ lymphatics. Bars in a–f 30 μm. g, h Quantification of CD31⁺ vessels (g) and podoplanin⁺ lymphatics (h) in the various experimental groups. Note the significantly higher numbers of both all vessels (CD31) and lymphatics in the EPC/MSC mixture per area. Area = microscopic field at ×400 magnification. In the EPC group, we observed scattered CD31+ cells, but they were not arranged in vessel-like structures. Since the number of vessels was zero, we abstained from quantification of the vessels in this group. Statistical significance is indicated. **p ≤ 0.01, ***p ≤ 0.001

Fig. 4.

Verification of lymphatics in Matrigel plugs and association with host leukocytes. Matrigel plugs containing EPCs and MSCs (ratio 1:1) were evaluated. a–c Pan-endothelial marker CD31 (green) and lymphatic marker Prox1 (red, nuclear staining) were applied. Note the presence of CD31⁺/Prox1⁻ blood vessels, and CD31^weak/Prox1⁺ lymphatics (arrows). d Control without primary antibodies. e–g Staining of lymphatics with anti-podoplanin (red) and leukocytes with anti-CD45 (green). Leukocytes are in close contact with the lymphatics. h, i Staining of lymphatics with anti-podoplanin (red) and macrophages with F4-80 and anti-11b (green). Macrophages are in close contact with the lymphatics. Bars 20 μm

We were furthermore interested to study the localization and differentiation potential of the grafted MSCs in the co-injection experiments. In order to identify them in the plugs we pre-labeled them with cell-tracker red. Staining with anti-CD31 antibodies showed that MSCs did not differentiate into endothelial cells (Fig. 5a). We hardly found any CD31⁺ MSC. Usually, the MSCs were located at some distance from the endothelium and obviously did not contribute to the vascular wall. This was confirmed by co-staining with antibodies against α-smooth muscle actin (αSMA). We hardly found any αSMA⁺ MSCs (that could have represented pericytes or smooth muscle cells) adjacent to a vascular structure (Fig. 5b). Most likely, the αSMA⁺ cells are myofibroblasts of the host that invaded the Matrigel. Additionally, the MSC marker PDGFRβ was found in MSCs (Fig. 5c–e), but again, these cells were hardly found in intimate contact with vascular walls.

Fig. 5.

Localization and differentiation of MSCs. Matrigel plugs containing EPCs and MSCs (ratio 1:1) were evaluated. MSCs were pre-labeled with cell tracker (red). Immunofluorescence studies with a anti-CD31 (green), b anti-αSMA (green), and c–e anti-PDGFRβ (green) antibodies. MSCs (red) were regularly scattered throughout the Matrigel plug and almost never appeared to be integrated into the vascular walls. Bars 50 μm in a 30 μm in b and 15 μm in c–e

In a subgroup of our experiments, we labeled cells with GFP or cell-tracker red. There are controversial reports on possible immunogenic effects of GFP in immunocompetent animals. To investigate immunogenic effects of GFP, we counted the number of CD45⁺ leukocytes in two groups of Matrigel plugs: MSCs without any labeling vs. GFP-transfected MSCs (Fig. 6a, b, g). In the GFP⁺ group, approximately 28 % (±5.74) of the counted cells were CD45⁺ leukocytes, whereas in the GFP-negative group there were approximately 14 % (±0.92) leukocytes. This difference, however, was not statistically significant (p = 0.11) and the immunohistology did not reveal a clustering of leukocytes around GFP-positive cells (Fig. 6a, b).

Fig. 6.

Studies on the effects of GFP and cell-tracker red. a, b Immunostaining with antibodies against the common leukocyte marker CD45 (red) reveals almost equal numbers of leukocytes in the Matrigel plugs of transplanted MSC that were not GFP-transfected (a) and in plugs with GFP-transfected MSCs (green) (b). c, d Matrigel plugs containing EPCs and MSCs without any pre-labeling (c) in comparison to EPCs and MSCs, the latter being pre-labeled with cell-tracker red (ct-red) (d). Anti-CD45 staining (green) shows a lower number of leukocytes in the ct-red labeled group. e, f The same specimens as in c and d. Nuclear lymphatic endothelial marker Prox1 (green) shows the development of lymphatics in the Matrigel plugs. Bar 50 μm in a–f. g–j Quantification of the results. g There is no statistically significant difference in the number of CD45⁺ leukocytes between the MSC vs. MSC-GFP groups (p = 0.11). h Leukocyte numbers are significantly lower in the EPC + MSC-ct-red group than in the unlabeled group. j Prox1⁺ lymphatic endothelial cell numbers are significantly lower in the EPC + MSC-ct-red group than in the unlabeled group. Statistical significance is indicated. **p ≤ 0.01

To study potential immunogenic effects of cell-tracker red, we counted the CD45-positive leukocytes in the groups of EPC/MSC vs. EPC/MSC cell-tracker-red (Fig. 6c, d, h). Here, the number of CD45-positive cells was significantly higher (p ≤ 0.01) in the group of EPC/MSC without any dye (22 % ± 2.9 vs. 8 % ± 1.2). The immunohistology did not reveal any clustering of leukocytes around cell-tracker-positive cells (Fig. 6c, d). In this experiment, we also counted the number of Prox1-positive lymphatic endothelial (Fig. 6e, f, j). There was a significantly (p ≤ 0.01) higher number of Prox1-positive cells in the group of the cell-tracker-negative EPC/MSCs (6 % ± 1.3 vs. 2 % ± 0.4, respectively). Together, these data show a higher number of leukocytes and correspondingly a higher number of lymphatics in the group of the unlabeled EPC/MSCs.

In vitro protein expression

We have previously shown that subcutaneously injected lung-derived mouse EPCs form blood vessels and lymphatics in Matrigel supplemented with VEGF-A and FGF-2 [12]. The secretion of VEGF-A by MSCs has already been documented [3, 25]. VEGF effects are naturally inhibited by soluble VEGF receptors, which are produced either by differential splicing or by shedding from the cell surface [26–28]. Bioavailability of VEGF-A is therefore dependent on both its own expression and the expression of its inhibitors. We analyzed the secretion of soluble VEGFR-1 (sVEGFR-1) and soluble VEGFR-2 (sVEGFR-2) and found that EPCs at different passage numbers, but not MSCs, secreted the soluble receptors. After 3 days, conditioned media of EPCs contained up to 10 ng/ml sVEGFR-1 and 800 pg/ml sVEGFR-2 (Fig. 7a, b). MSCs were below detection limit. In contrast, MSCs released high amounts of VEGF-A (up to 4 ng/ml) into the media, whereas EPCs were below detection limit (Fig. 7c). In co-cultures of MSCs and EPCs, the VEGF-A secreted by MSCs was obviously bound by soluble VEGF receptors (Fig. 7c). A substantial amount of VEGF-A (approx. 350 pg/ml) in the MSC supernatant was bioavailable, and even in co-cultures of MSCs and EPCs, bioavailable VEGF-A could still be measured (Fig. 7d).

Fig. 7.

Release of soluble (s) VEGFR-1 and sVEGFR-2 by EPCs, and VEGF-A by MSCs. a, b Supernatants from MSCs and EPCs of passage 14 (p14) and p35 were used for sandwich ELISA. Detection limits were: sVEGFR-1: 300 pg/ml; sVEGFR-2: 80 pg/ml. Cells were plated at 1.5 × 10⁴ cells/cm² and supernatants were harvested after 3 days. c, d Sandwich ELISA of conditioned media from MSCs, EPCs and co-cultures. c Total VEGF-A secretion. d Secretion of bioavailable VEGF-A, measured with sVEGFR-1 as capture molecule. Detection limits were: Total and bioavailable VEGF-A 40 pg/ml. Experiments were performed in triplicate. Bars represent mean values ± SD

In the in vivo assays, hem- and lymphangiogenesis are most sufficiently induced when EPCs and MSCs are co-injected, suggesting that the two cell types release growth factors that act synergistically. We therefore analyzed proteins released from MSCs and EPCs with a mouse angiogenesis array kit, which detects 53 different proteins related to angiogenesis and endothelial cell growth. Several pro-angiogenic proteins could be detected in the supernatants from both EPCs and MSCs, such as FGF-1, IGF binding proteins, and MMP-3—and its inhibitor TIMP-1 (Fig. 8). It has been reported that placenta growth factor (PlGF), also known as progenitor-mobilizing activity, is highly expressed and released from both EPCs and MSCs [29]. Indeed, our analysis by sandwich ELISA confirmed that both cell types released high PlGF levels in the ng/ml range. Thereby, MSCs secreted up to fivefold more PlGF than EPCs (5.69 vs. 1.13 ng/ml) (Fig. 8, and data not shown).

Fig. 8.

Secretion of angiogenesis regulators from MSCs and EPCs. Mouse angiogenesis antibody-array analysis of conditioned media from MSCs and EPCs cultured for 72 h. One milliliter of the conditioned media was applied to each array. Graphs represent the mean pixel density of each of the two spots. Densitometric analysis was performed with ImageJ 1.40 g. Note, in case of PlGF-2, the densitometric analysis reached its limits and was not able to report the extremely high values produced by MSCs

On the other hand, CXCL1/GRO a pro-angiogenic chemokine, CXCL16, ADAMTS1, Platelet-derived growth factor (PDGF-BB) and PF4/CXCL4 were predominantly secreted by EPCs. In contrast, MSCs expressed higher levels of PAI-1, PDGF-AA, fraktalkine/CX3CL1, and MCP-1/CL2. Consistent with our ELISA data, we found VEGF-A in MSCs, but also observed moderate expression of VEGF-B and amphiregulin in MSCs (Fig. 7). Amphiregulin seems to be an important inducer of VEGF-A expression and may be directly responsible for the VEGF-A release from MSCs [30]. The reliability of our data is supported by the observation that endoglin/CD105, a member of the TGFβ receptor complex, is only expressed in EPCs. We have also analyzed the supernatants from EPC/MSC co-cultures, to see if direct cell interactions may alter the secretion pattern. However, this was not the case. The expression profile did not significantly differ from the single cultures, with one exception, VEGF-A. VEGF-A was not found in the EPC/MSC co-cultures, indicating its complete binding to EPCs (Suppl. Fig. 1).

Discussion

We have recently reported that endothelial progenitor cells (EPCs) isolated from the lungs of adult mice contribute to blood vessels and lymphatics, when injected subcutaneously into mice. In these experiments we used Matrigel that was supplemented with VEGF-A and FGF-2 [12]. The lung of adult rodents obviously possesses a high vasculogenic potential, since large numbers of EPCs have also been found in rats [15]. The specific reason for this may reside in the fact that adult rodents potently regenerate their lungs after resection, which is associated with angiogenesis [31]. Regenerative potential of the lung has been observed in various mammals [32], whereas there is only minor evidence for compensatory growth of the lungs in the human [33]. Upon infection, e.g., with Mycoplasma pulmonis, robust lymphangiogenesis has been observed in the lungs of mice [34], and it is likely that similar mechanisms take place in the human, since acute infection is a potent inducer of lymphangiogenesis in man, too [18]. Nevertheless, one should act with caution before translating the finding of lung-derived EPCs into man. Even in animals, their contribution to hem- and lymphangiogenesis within the lungs has not yet been determined.

EPCs and MSCs interact to produce blood vessels and lymphatics

Our studies confirm and expand previous data that have shown cooperative interactions between EPCs and MSCs during de novo blood vessel formation. Thereby, in vivo studies were mostly performed on immunodeficient mice and rats [3, 6–8]. Similar studies were performed with EPCs and adipose progenitor cells (APCs) and have also shown a synergistic vasculogenic effect [35]. The combined application of EPCs and smooth muscle progenitor cells (SMPCs) increased the foot perfusion in experimental hindlimb ischemia in nude mice significantly more than each single cell type [36]. Our studies were performed in a syngeneic model using C57/Bl.6 mice. We also observed robust vessel formation in Matrigel plugs containing a 1:1 mixture of EPCs and MSCs. The endothelial lining of the vessels was made up of both graft- and host-derived endothelial cells showing functional interactions between graft and host. Additionally, we observed the development of lymphatics in the Matrigel plugs, identified by two characteristic markers, the glycoprotein podoplanin and the transcription factor Prox1 [37, 38]. Like the blood vessels, the lining of lymphatics was formed by a mosaic of graft- and host-derived endothelial cells. However, we were not able to demonstrate functionality of this lymphatic network in two injection experiments with Berliner blue, but it may be too early to draw conclusion from just two experiments.

In our studies, we could not find evidence for the direct incorporation of MSCs into the walls of blood vessels or lymphatics. We identified MSCs by pre-labeling with a red fluorescent cell tracker. The cells were regularly distributed in the Matrigel plugs and were found to be negative for αSMA, a marker of pericytes and smooth muscle cells. This is in contrast to previous studies describing that MSCs and adipose stromal cells (ASCs) occupy perivascular positions in graft- and host-derived vessels, and express αSMA [3, 5, 6, 35]. The αSMA-positive cells observed in our studies are obviously host myofibroblasts that invaded the Matrigel. If these differences may be due to the immune status of the animals remains to be studied. Interestingly, in a model of distal femoral artery ligation in Balb/C mice, MSCs improved limb function without being incorporated into functional collaterals [25].

Immunogenic effects of cell tracers

In a subgroup of our experiments, we labeled cells with lentiviral expression-vectors for GFP or with cell-tracker red (CMPTX). Since we used immunocompetent animals, an immunogenic effect of the tracers must be considered. In the literature, the immunogenic effects of GFP and EGFP are discussed controversially. Transfection studies using GFP expression-vectors in different cell lines and studies on GFP-transgenic mice have revealed inconsistent results concerning the immunogenicity and cytotoxicity of GFP. Whereas the study by Liu and colleagues [39] described the induction of apoptosis in cells expressing GFP, Grabski et al. [40] did not observe an impact of GFP-encoding lentiviral vectors on the phenotype and function of dendritic cells. Furthermore, no difference in mature T cell proliferation and no immunological differences were found between GFP⁺ and parental GFP- C57BL/6 mice [41].

Here we observed no immunogenic effects of GFP and cell-tracker red. The number of CD45⁺ leukocytes was slightly but not significantly higher in the GFP⁺ group. There was no clustering of leukocytes around the GFP-positive MSCs. There was a weak tendency for higher numbers of leukocytes in the GFP⁺ group, which may explain why there are controversial opinions about the GFP effects.

To the best of our knowledge, data on the immunogenicity of CMPTX do not exist. In our cell-tracker-labeled group the number of leukocytes is even lower, and this goes along with a lower number of lymphatics in this group. The data clearly show that the cell-tracker is not immunogenic, and provides further evidence that lymphangiogenesis in the specimens is induced indirectly by leukocytes, which are attracted by the interacting EPCs and MSCs. We would also like to point out that in a series of experiments where we labeled EPCs with lentiviral luciferase expression-vectors, the cells vanished from the hosts shortly after the injection. Here, an immunogenic response against luciferase appears to be likely.

Molecular interactions between EPCs and MSCs

VEGF-A

Our studies are consistent with previous ones that have demonstrated the secretion of VEGF-A from MSCs [3, 25]. VEGF-A secretion was also found in ASCs [42]. However, these studies did not measure the bioavailability of VEGF-A, which can be regulated by its natural inhibitors, the soluble VEGF receptors. These bind VEGFs and PlGF, and thus regulate the availability of the ligands to the transmembrane receptors [26–28, 43]. We could previously show that amniotic fluid contains high amounts of soluble VEGF receptors that act as a powerful sink for VEGF-A and PlGF [28, 44]. Regulation of VEGF bioavailability seems to be an important process for the maturation of blood vessels [43]. Here we show that MSCs release high amounts of VEGF-A (up to 4 ng/ml per day), whereas EPCs produce soluble VEGF receptors. Under co-culture conditions, the detectable amount of total VEGF-A is reduced by about 90 %, indicating that most of the ligand is bound by EPC-secreted VEGF receptors. Nevertheless, a substantial amount of VEGF-A in the MSC supernatant is bioavailable, and even in co-cultures of MSCs and EPCs, bioavailable VEGF-A is still present. This is in line with the findings by Melero-Martin and coworkers [6], who quantified the de novo-formed blood vessels in a graded series of EPC/MSC mixing ratios, and observed the strongest effect with 20 % EPCs and 80 % MSCs. Also, when only EPCs are mixed with Matrigel, the addition of exogenous growth factors such as VEGF-A and bFGF is needed to activate their growth and differentiation into functional blood vessels [12, 45].

MSCs may be capable of inducing VEGF-A by autocrine stimulation via amphiregulin (Areg), a 78-aa glycoprotein of the EGF family of proteins. Like previous studies [3], we found Areg and VEGF-B in MSCs. Areg stimulates VEGF-A expression in granulosa cell lines [46] and in fibroblast-like synovial cells [30]. VEGF-B can form heterodimers with VEGF-A [47]. This can lead to heterodimerization of VEGFR-1 and VEGFR-2 on endothelial cells, which is important for homeostasis of the endothelium [48].

PlGF

Another important factor that regulates bioavailability of VEGF-A is PlGF. We found high and very high secretion of PlGF by EPCs (1.13 ng/ml) and MSCs (5.69 ng/ml), respectively.

The moderate expression of PlGF by endothelial cells is well established and discussed as an autocrine stimulation and maintenance factor for the endothelium [28, 49, 50]. The release of PlGF by MSC has been reported before [3, 25]. Besides being involved in the binding of soluble and transmembrane VEGFR-1, thereby increasing the bioavailability of VEGF-A to VEGFR-2, the high concentrations measured in our conditioned media from both EPCs and MSCs may also have functions for tissue repair. In myocardial ischemia, PlGF obviously induces cardiac repair by the up-regulation of antiapoptotic and angiogenic factors. Furthermore, it may be involved in progenitor cell proliferation and their recruitment into sites of tissue repair [29]. The high secretion of PlGF with pleiotropic effects by MSCs and EPCs may therefore have a much broader implication for tissue repair and regeneration than originally expected.

PDGF

In the conditioned media, we detected the secretion of PDGF-BB by EPCs, and PDGF-AA by both EPCs and MSCs. The release of PDGF from EPCs has been discussed as a mechanism for the recruitment of MSCs to newly formed vessel, and their differentiation into pericytes or smooth muscle cells [5]. The paracrine factor PDGF-BB is a ligand of the PDGF-beta receptor, which has been used a marker for MSCs [3]. However, in our in vivo studies, PDGF-AA and PDGF-BB secretion was obviously not sufficient to induce recruitment of MSCs to the vascular wall. If this may be a difference between immunodeficient and immunocompetent mice remains to be studied. We have shown previously that PDGF-B induces the differentiation of fibroblasts into myofibroblasts [51], and we assume that the scattered host-derived αSMA-positive cells in our experiments are derived from a comparable mechanism. The induction of vessel formation in the EPC/MSC mixtures does not seem to be dependent on the recruitment of MSCs to perivascular locations, but rather on the paracrine release of pro-angiogenic factors.

Other growth factors

Consistent with our previous studies [12], we show that EPCs are bipotent and differentiate into both blood vascular and lymphatic endothelial cells. The growth factors that induce lymphangiogenesis most potently are VEGF-C and VEGF-D [4, 52, 53]. However, we could not detect VEGF-C in the supernatants of MSCs or EPCs. However, there may be an alternative source for VEGF-C and VEGF-D: activated mononuclear cells (M2 macrophages) expressing the tyrosine kinase Syk [54]. In our study, the secretion of MCP-1, PlGF, fractalkine [55], CXCL16 [56], and CXCL10 [57] may attract monocytes/macrophages that release VEGF-C or VEGF-D. With the pan-leukocyte marker CD45 we have shown that the newly forming lymphatics are in close contact with leukocytes, and the number of leukocytes correlates positively with the number of lymphatics. Among the lymphatics there are F4-80 and CD11b-positive macrophages. Lymphangiogenesis may then be induced by these VEGF-C- and VEGF-D-expressing macrophages.

Electronic supplementary material

Below is the link to the electronic supplementary material.