Background
Critical limb ischemia (CLI) is the most severe form of atherosclerotic peripheral arterial disease (PAD). Due to the progression of vascular risk factors, PAD incidence is predicted to double by 2050 [
1]. The management of patients is limited to surgical revascularization. However, around up to 30% of patients are not eligible for such procedures due to poor distal vascular bed. Therefore, these “no-option” patients (NO-CLI) experience a high risk of major amputation [
2] and are exposed to a high level of cardiovascular death [
3].
In such context, cell therapy (CT) has been proposed for NO-CLI patients to promote angiogenesis and improve tissue perfusion [
4]. To date over 120 phase I/II or III clinical trials have investigated a variety of cell therapies [
5]. Recent meta-analysis [
6] are in favor of a clinical improvement of treated patients but results remain divided [
7,
8]. In most cases, non-selected autologous bone marrow derived cells (BMCs) were used. They were obtained from elderly ill patients and exhibit low proangiogenic potential [
9]. In most studies BMCs were not well characterized.
We recently published a multicenter clinical trial evaluating the effect of autologous BMCs versus placebo in CLI-patients [Bone Marrow Autograft in Limb Ischemia (BALI)]. This study was in favor of efficacy but the rate of amputation remained elevated in treated patients [
10]. Inflammation was associated with poor outcome [
11]. In order to improve our understanding, BMCs were extensively characterized and we observed a great variability in BMCs composition between patients [
11].
BMCs contained mature and immature cells [hematopoietic stem cells (HSCs), endothelial progenitor cells]. Interestingly, they also contained a rare subset of mesenchymal stem cells (MSCs). These non-hematopoietic mononuclear cells reside in the bone marrow (BM) stroma [
12]. MSCs can be obtained in adults from many tissues such as adipose tissue (ADSCs), peripheral blood, synovial membrane and dental pulp. Fetal/neonatal tissues [e.g., umbilical cord blood, umbilical cord Wharton’s jelly (WJ-MSCs), amniotic fluid, placenta] are also a potential source of MSCs [
13]. Still, human BM remains the main source of MSCs. They show an extensive capacity of differentiation into osteoblasts, chondrocytes, adipocytes, astrocytes and skeletal muscle cells [
14]. MSCs can migrate, proliferate in areas of ischemia and can promote regeneration of damaged tissues and reducing inflammation [
15]. They may be good candidates for CT as they combine proangiogenic, anti-inflammatory and immunomodulatory properties [
16].
CT protocols in CLI-generally require hundreds of millions of MSCs per treatment [
17‐
19]. Therefore, in vitro cell expansion is needed. In autologous situation, the patient’s age and clinical characteristics influence the culture efficiency [
20]. Major efforts have been made to improve culture conditions and favor endothelial induction by adding supplements containing pro-angiogenic factors [vascular endothelial growth factor (VEGF), epidermal growth factor (EGF), fibroblast growth factor-2 (FGF2) and insulin-like growth factor-1 (IGF-1)] [
21‐
27]. The so called “endothelial cell-specific growth medium” (EGM-2) improves the proliferation rate and may induce the acquisition of endothelial markers. These stimulated-MSCs (S-MSCs) can form functional blood vessels in collagen-plug implanted in mice [
21] but have never been tested in a hind limb ischemia model (HLIM).
The purpose of this study was: (1) to evaluate the presence of MSCs in CLI-patients BMCs, (2) to analyze the phenotype of S-MSCs, and (3) to compare the effect of MSCs and S-MSCs in a murine HLIM in comparison with BMCs.
Discussion
Among innovative therapies, CT is a good candidate to treat patients with severe PAD. BMCs are considered as “gold standard” in CLI-CT because it was used by the founder trial [
4]. In spite of encouraging clinical studies, the efficacy of CT remains controversial [
7,
8]. This may be partially explained by the heterogeneity of autologous BMCs [
11]. In such BMCs, HSCs are extremely rare and may not support proangiogenic properties of BMCs [
5]. Still, these autologous BMCs may be the source of other types of proangiogenic stem cells. Indeed undifferentiated MSCs were obtained from BMCs CLI-patients as previously described [
11,
29,
36].
MSCs are well-recognized for their high proliferation and differentiation potential. MSCs support a significant paracrine effect through the secretion of proangiogenic cytokines which provide anti-apoptotic effects and stimulate revascularization [
15].
In the first step of our study, we confirmed the presence of MSCs in CLI-BMCs. We next evaluated if the culture of MSCs in a medium enriched in endothelial growth factors (EGM-2) could improve the proangiogenic properties of MSCs that we named S-MSCs. S-MSCs that we obtained from CLI-patients showed a morphology which was indistinguishable from MSCs obtained in CFU-F medium. However, S-MSCs’ doubling time was shorter for when compared with MSCs’, in agreement with published data [
21,
25,
37].
We next characterized S-MSCs in FC and observed that these cells did not express endothelial markers (CD31, Ve-Cadherin, VEGF R1 and VEGF R2). Using other sources of MSCs derived from umbilical cord blood [
25,
26], adipose tissue [
22,
23,
27] or BM, but from healthy donors [
21], an endothelial differentiation of MSCs was reported when cultured in EGM-2 medium [
37]. Oswald et al. was the first to report that a VEGF induction gave rise to the endothelial differentiation of MSCs expression classical endothelial markers but not CD31 [
38]. Tancharoen et al. also, showed that S-MSCs expressed endothelial markers including CD31 but with a much lower level than HUVECs [
39]. In agreement with our data, other published studies on WJ-MSCs [
32], BM-MSCs [
40] or diabetic patients ADSCs [
41], stated that endothelial cell growth supplement alone could not per se induce the expression of molecular markers of ECs. This illustrates the difficulty encountered to define such cells when cultured in different conditions. The term of “endothelial-like cells” has been employed by some authors [
21,
32,
40,
42] in spite of a high heterogeneity in both culture conditions and MSCs origin. Interestingly, it was suggested that shear stress could be a major parameter for endothelial differentiation [
40].
Our data show that MSCs and S-MSCs express comparable proteomic and transcriptomic profiles except for VCAM1 which is significantly higher in S-MSCs. VCAM1 mediated the interaction of MSCs with ECs, which is essential for MSCs homing [
43]. This expression of VCAM1 was previously reported in S-MSCs obtained in the presence of VEGF-A [
38]. A role of EGF, which is present in EGM-2 medium, has been shown to increase VCAM1 and MSCs adhesion to ECs [
44].
To further characterize S-MSCs, we analyzed their secretion capacity. For this, we quantified a set of ten growth factors in S-MSCs culture supernates in comparison with MSCs. Our data indicate major differences between these two types of cells: in comparison with MSCs, S-MSCs produce high levels of PDGF-AA, Angio-1 and LIF whereas the secretion of pro-inflammatory IL-6 is significantly lower. CXCL12 is not secreted by S-MSCs whereas it is found at high concentration in the MSCs supernatant. The loss of capacity of S-MSCs to secrete CXCL12 could be the consequence of EGM-2 induction. Using a supplementation with platelet-rich plasma, which contains PDGF, TGF
β1, FGF2, IGF-1, VEGF and EGF (some of them being also present in EGM-2 medium), Goedecke et al, have reported a loss of CXCL12 secretion associated with a defect in HSCs migration [
45]. The analysis of S-MSCs supernatants indicated the diminution in the concentration of FGF2, EGF and IGF-1 present in the EGM-2 medium. This may be the consequence of a consumption or could be explained by the endocytosis of a ligand-receptor complex.
In order to further characterize the secretome, we evaluated the capacity of supernates to induce tube-like structures from HMEC-1. We observed that S-MSCs culture supernates had the strongest ability to induce tube-like structures. This is in agreement with the results of growth factor quantification showing that S-MSCs supernates contain higher concentrations of proangiogenic factors. Taken together, these results suggest that the secretome analysis allows to differentiate S-MSCs from MSCs. A further issue would be to elucidate the potency of each growth factor present in conditioning media. Furthermore, the existence of possible autocrine mechanisms has to be considered.
Our study shows that S-MSCs can be functionally distinguished from MSCs by their stronger capacity to form pseudo-tubes in vitro. This confirms most reports having used EGM-2 induction [
21‐
23,
25,
40,
41,
46]. However, Choi et al, raised the point that MSCs are able to form tube-like structures in Matrigel but the observation of these pseudo-tubes by electron microscopy revealed the absence of lumen [
32]. Although commonly used, Matrigel cannot be considered as a specific angiogenic assay [
47].
It was therefore mandatory to evaluate in vivo, the proangiogenic potential of MSCs and S-MSCs. In a well-established mouse model of HLIM [
30], our results clearly indicate that MSCs from CLI-patients completely restore blood flow (primary endpoint) in comparison with BMCs that we considered as the “gold standard”. Such results are in agreement with those obtained by Iwase et al, who concluded that BM-MSCs were superior to BMCs in promoting neovascularization [
48]. The limit of this study was that BM-MSCs were obtained from young rats. Indeed, it has been shown that age could impair stem cell properties [
9]. In the context of CLI, this alteration could be further increased in case of associated risk factors [
20]. In contrast, MSCs obtained from CLI-patients conserve proangiogenic properties through a paracrine mechanism [
15,
29,
49]. This may be at least partly explained by the fact that, BM-MSCs from CLI-patient have a similar secretome profile compared to that of healthy MSCs [
29,
50]. In agreement with this hypothesis, Smadja et al, 2012 and Gremmels et al, 2014 showed, in a HLIM model, that MSCs from CLI-patients had a comparable proangiogenic potential to MSCs isolated from age matched patients free of any cardiovascular disease [
29,
36].
To our knowledge, this is the first report to establish the property of S-MSCs obtained from CLI-patients infused in HLIM model. Interestingly, S-MSCs restored completely blood flow and earlier than MSCs. The infusion of BMCs led to a highly heterogeneous response in terms of extend and kinetics of blood flow recovery. This fits with the heterogeneity of BMCs in terms of cellular content. In contrast, the infusion of MSCs and S-MSCs restored blood flow in a homogeneous manner. The selection and amplification in culture may allow to obtain a consistent favorable response to infusion, and this whatever the patient’s clinical status.
We further demonstrated that MSCs and S-MSCs could improve clinical recovery as well as limb salvage. In agreement with Iwase et al, MSCs were more effective than BMCs to save limb [
48]. Our results indicate that S-MSCs were also more effective than MSCs.
Interestingly, the analysis of the
gastrocnemius muscle clearly shows that S-MSCs not only enhance neovascularization but also favor the development of vascular muscle cells suggesting the formation of mature and stable capillaries [
51]. Indeed, arteriogenesis is more capable of restoring tissue blood supply than angiogenesis [
52]. Collateral vessels have the capacity to carry a larger volume of blood than sprouting capillary networks [
53].
In contrast, the analysis of the
semimembranosus muscle indicates an enlargement of the vessels and limited ischemia. This is explained by the anatomic situation of this muscle which is located proximally to the ligation site. It was reported that, following a femoral artery ligation, an arteriogenic process is initiated by the increase of blood pressure in the pre-existing collaterals that circumvent the obstruction and can supply blood flow to the distal tissue [
54]. In contrast,
gastrocnemius muscle being downstream from the ligation site is more sensitive to ischemia and therefore prone to undergo a process of angiogenesis. In this respect, our results clearly show the favorable effect of cell infusion. This effect may be more the consequence of a paracrine effect considering the high potential of secretion of such cells. The proangiogenic effect is unlikely related to endothelial differentiation of infused cells since no human DNA was detected at the end of the experiment. Previous studies have illustrated the rapid disappearance of infused cells [
55,
56].
A major advantage of S-MSCs may be their capacity to improve skeletal muscle repair after ischemia. This effect could explain the efficacy of S-MSCs to reduce the amputation rate. To evaluate muscle repair, we quantified myofibers with central nucleus location. Indeed, muscle regeneration is characterized by the activation of myogenic cells which leads to the formation of new myofibers. These fibers are recognized by a centrally located myonucleus [
34,
35,
57,
58].
MSCs have been evaluated in 13 clinical trials in the last decade [
17‐
19,
59‐
66] and have included a total population of 216 MSCs-treated patients. These studies have established the safety of such cells [
18,
67], which may be at least partly explained by their immunomodulatory properties [
16,
68,
69]. Clinical trial results were in favor of efficacy, one randomized clinical trial concluding that MSCs were more effective than BMCs [
17].
Our study pointed out the importance of the secretome. This should be extended to the analysis of secretome content including extracellular vesicles, exosomes, microRNAs, mRNAs, long non-coding RNAs, circular RNAs which may have an interest in CLI-therapeutics [
15,
70].
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.