Type 2 diabetes alters mesenchymal stem cell secretome composition and angiogenic properties

Abstract This study aimed at characterizing the impact of type 2 diabetes mellitus (T2DM) on the bone marrow mesenchymal stem cell (BMMSC) secretome and angiogenic properties. BMMSCs from Zucker diabetic fatty rats (ZDF) (a T2DM model) and Zucker LEAN littermates (control) were cultured. The supernatant conditioned media (CM) from BMMSCs of diabetic and control rats were collected and analysed. Compared to results obtained using CM from LEAN‐BMMSCs, the bioactive content of ZDF‐BMMSC CM (i) differently affects endothelial cell (HUVEC) functions in vitro by inducing increased (3.5‐fold; P < 0.01) formation of tubule‐like structures and migration of these cells (3‐fold; P < 0.001), as well as promotes improved vascular formation in vivo, and (ii) contains different levels of angiogenic factors (e.g. IGF1) and mediators, such as OSTP, CATD, FMOD LTBP1 and LTBP2, which are involved in angiogenesis and/or extracellular matrix composition. Addition of neutralizing antibodies against IGF‐1, LTBP1 or LTBP2 in the CM of BMMSCs from diabetic rats decreased its stimulatory effect on HUVEC migration by approximately 60%, 40% or 40%, respectively. These results demonstrate that BMMSCs from T2DM rats have a unique secretome with distinct angiogenic properties and provide new insights into the role of BMMSCs in aberrant angiogenesis in the diabetic milieu.


Introduction
Type 2 diabetes mellitus (T2DM) is reaching epidemic proportions worldwide. The persistent hyperglycaemic milieu in T2DM is associated with macro-and micro-vascular complications affecting the heart, blood vessels, eyes, kidney and nerves, and the wound-healing process in diabetic patients [1][2][3]. Most of the T2DM pathological complications are associated with impaired vascularization and/or aberrant angiogenesis. Excessive and abnormal angiogenesis plays a pivotal role in diabetic retinopathy and nephropathy [4,5], whereas deficient angiogenesis contributes to impaired wound healing and coronary collateral vessel development [1,3,6]. Despite extensive research, the mechanism(s) behind the abnormal angiogenesis balance in diabetic patients is still poorly understood.
Scientific research has provided evidence that the dysfunction of different types of cells, including endothelial, peripheral blood and bone marrow mesenchymal stem cells (BMMSCs), is related to the reduced blood vessel regenerative potential in T2DM [7][8][9]. Endothelial cells migrate in response to soluble mediators that regulate new blood vessel formation [10], but their function is impaired in the diabetic milieu [7,11,12]. BMMSCs, whose phenotype is strongly determined by their specific environment, have the capacity to home in on sites of injury, proliferate and differentiate into multilineage cell types [13]. The T2DM milieu affects the regenerative potential of BMMSCs as well as their proliferation [8], restricts their multipotency and impairs their capacity to augment post-ischemic neovascularization in diabetic mice [14]. BMMSCs exert a paracrine effect by the release of growth factors and cytokines, such as IGF-1, which stimulate endothelial cell migration [15], inhibit endothelial apoptosis and promote angiogenesis [16]. Moreover, by providing proangiogenic factors, BMMSCs create a favourable microenvironment that promotes neovascularization and tissue regeneration [13,15,17]. Although important in this respect, the impact of T2DM on the BMMSC secretome and its angiogenic properties has not yet been characterized.
We hypothesized, therefore, that the BMMSC secretome in the diabetic milieu has a different composition and affects angiogenesis-related endothelial cell (EC) functions, including proliferation and new blood vessel formation. We tested this hypothesis using Zucker diabetic fatty (ZDF) (Leprfa/fa) rats (a model of T2DM) and Zucker Lean control (LEAN) (Leprfa/+/Lepr+/+) rats (their age-matched littermates) [18]. By focusing on the impact of T2DM on BMMSC secretome functions, this study aimed at both elucidating the contribution of MSCs in diabetic-related vascular complications and establishing relevant aspects for future cell therapies in diabetes.

Animal models
Adult, 8-week-old, male, Swiss nude mice as well as adult, 12-to 13-week-old, male, obese Zucker fa/fa rats (ZDF) and their lean fa/+ littermates (LEAN) were purchased from Charles River (L'Arbresle, France). The animals were housed, operated and killed using procedures in accordance with the European Community Standards on the Care and Use of Laboratory Animals. The experimental procedures were approved by the Ethics Committee of the University Paris-Diderot.
The diabetic condition of the rats used in this study was determined on the day of killing them. For this purpose, venous blood was collected, and the respective glycaemic state was evaluated using a glucometer (Roche Diagnostics, Meylan, France). Plasma concentrations of glucose and fructosamine were determined using commercially available kits (Roche Diagnostics) following the manufacturer's instructions.

Isolation of rat mesenchymal stem cells (BMMSCs)
The femurs and tibiae from each rat were cleaned of connective tissues, and their respective epiphyses were removed to allow insertion of 23gauge needles connected to syringes containing serum-free aMEM (Invitrogen, Cergy Pontoise, France). The cells present in the harvested marrow were then homogenized using complete medium composed of aMEM supplemented with 10% (v/v) foetal calf serum and 1% (v/v) antibiotic/antimycotic (ATB/ATM) solution (PAA Laboratories GmbH, Pasching, Austria). The isolated cells from rats were seeded at 5 9 10 5 cells/cm² and cultured at 37°C in a humidified 5% CO 2 /95% air environment. After 2 days of culture, the supernatant media (containing nonadherent cells) were discarded. Fibroblastic colonies (CFU-F) appeared at day 5 of culture and were all pooled at day 12 (cell passage 1). For amplification of the LEAN-and ZDF-BMMSC populations, the cells were seeded at 10 9 10 3 cells/cm². The supernatant media were changed twice a week. BMMSCs at passage 2-3 were used for the experiments of this study.

Preparation of conditioned media
BMMSCs were seeded at 10 4 /cm² in a culture-treated flask and cultured in calf serum-free aMEM under standard cell culture conditions. After 24 hrs, the supernatant was collected, centrifuged (700 9 g, for 4 min.), aliquoted and frozen at À80°C until further use. The concentration of total proteins in the conditioned medium (CM) of BMMSCs from both ZDF and LEAN rats was determined using the Bradford protein assay and following the manufacturer's instructions. For experiments with neutralizing antibodies, anti-IGF-1, anti-LTBP1 or anti-LTBP2 were added to CM (5 ng/mL) and maintained at 37°C for 1 hr before use in experiments with cells.

Culture of HUVECs
The effects of CM on endothelial cells were studied using human umbilical vein endothelial cells (HUVECs). HUVECs in EGM-2 medium (Lonza, France) supplemented with 5% foetal calf serum and 1% (v/v) ATB/ ATM were cultured in tissue-culture flasks pre-coated with 0.5% gelatin under standard cell culture conditions. When the cells reached 80% confluency, they were passaged by trypsinization (Trypsin/EDTA Solution, Life Technologies, Alfortville, Ile de France, France).

Cell proliferation
Cells were seeded at the density of 3 9 10 3 cells/cm 2 in individual wells of 12-well culture plates in aMEM containing 10% FBS. For each timepoint, BMMSCs were trypsinized, incubated with Trypan blue and counted using a Malassez chamber.

Migration of human umbilical vein endothelial cells in Boyden chambers
HUVEC migration was determined using commercially available Boyden chambers (Corning Costar, Tewksbury MA, USA) whose two compartments were separated by polycarbonate membranes with 8 lm diameter pores. Aliquots of cells (50 9 10 3 cells) in 100 ll of aMEM without foetal calf serum were placed in the upper chamber, and the various media of interest to this study were each placed in the bottom chamber. The cell migration experiments were conducted in a humidified, 37°C, 5% CO 2 /95% air environment for 24 hrs. The cells that had transversed but still adhered on the other side of the membrane separating the Boyden chamber were then stained in situ using May Grunwald-Giemsa stain. All such membranes were excised, mounted on slides, visualized using light microscopy and photographed.

Wound-healing assay
HUVECs were seeded at a density that resulted in~70-80% confluence on the bottom surface of individual wells of 12-well tissue-culture plates within 24 hrs of culture. At that time, each monolayer was scratched across the respective centre using a new 1-ml pipette tip and was rinsed twice with phosphate-buffered saline (PBS) to remove detached cells. The wounded HUVEC samples were then treated with CM from either BMMSCs of diabetic or control rats at 37°C for 6 hrs, rinsed twice with PBS and fixed using 4% paraformaldehyde for 30 min. The scar region on each cell monolayer was visualized using light microscopy and photographed before and after the 6-hr interval. Comparison of the evidence on these two sets of micrographs was used to determine the migration of HUVECs. These data were expressed as the

Formation of tubular-like structures by HUVECs on Matrigel
Each well of 48-well tissue-culture plates was coated with 10 mg/ml Matrigel (growth factor reduced Matrigel, VWR France) and maintained in a humidified, 37°C, 5% CO 2 /95% air environment for 1 hr. Then, 5 9 10 4 cells in various media were added and cultured in a 37°C, 5% CO 2 /95% air environment for 24 hrs. At that time, four random areas were visualized using light microscopy, photographed and analysed for the presence of tubular-like structures. The total length of the formed tubular-like formation was measured using commercially available Ima-geJ software.

3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) assay
HUVECs were seeded at 10 4 cells per each well of 96-well plates in various media (containing 150 lM of H 2 O 2) in a humidified, 37°C, 5% CO 2 /95% air environment for 90 min. At that time, the cells were rinsed with PBS, and the MTT assay (Life Technologies) was performed. The results provided evidence regarding cell viability after challenge by H 2 O 2 .

ROS measurement
To measure reactive oxygen species (ROS) production in HUVECs, DCFH-DA, an oxidation-sensitive indicator, was used. In the presence of ROS, DCFH is oxidized to fluorescent 2 0 ,7 0 -dichlorofluorescein (DCF), which can be measured by fluorometry. For this purpose, HUVECs were seeded at 10 4 cells/well in 96-well cell culture plates. The supernatant medium was removed, and DCFH-DA was added at a concentration of 10 lM per well. These specimens were kept at 37°C for 1 hr. At that time and after rinsing the cells with PBS, CM from BMMSCs of either diabetic or control rats was added. Oxidative stress was induced using tert-butyl hydroperoxide (Luperox, t-BuOOH Sigma-Aldrich (St. Louis, MO, USA) at 100 lM. Fluorescence intensity of the samples was measured using a fluorescence spectrophotometer at excitation and emissions wavelengths of 485 nm and 530 nm, respectively.

RNA extraction, reverse transcription and quantitative polymerase chain reaction (qRT-PCR)
Expression of select angiogenic markers by BMMSCs was determined using total RNA that had been extracted from these cells using TRIzol reagent (Life Technologies) according to the manufacturer's protocols. Real-time PCR was performed using an iCycler thermocycling apparatus (MyiQ TM Single-Color Real-Time PCR, Bio-Rad Laboratories, Marnes-la-Coquette, France). After activation of the DNA polymerase at 95°C for 10 min., cDNA was amplified by performing 40 two-step PCR cycles: a 15 sec. denaturation step at 95°C, followed by a 60 sec. annealing and an extension step at 60°C. The Multiplex used contained 84 angiogenic-related genes (the rat angiogenesis RT² Profiler PCR Array, Qiagen).
Each sample was run in triplicate, and each analysis was repeated on four separate occasions. Data were analysed using MyiQ TM Software (Bio-Rad Laboratories). The data were normalized with respect to results obtained from LEAN-BMMSCs.

ELISA analysis
The level of 7 mediators (VEGF, IL-6, TNFa, MCP-1, FGF-2, IGF1 and PLAU) were determined in the respective supernatant media at each time-point of interest to this study using different ELISA kits and following the manufacturer's instructions. These experiments were conducted in triplicate and repeated on three separate occasions. A fiveparameter regression formula was used to calculate concentrations of the respective mediators using standard curves for each mediator assessed.

Proteomic analysis
Each supernatant CM from LEAN and ZDF-BMMSCs was separately concentrated and desalted using an Amicon Ultra-2 Centrifugal device column (Millipore, France). Briefly, 500 ll of each conditioned medium tested was loaded in a column and centrifuged at 4500 9 g for 75 min. This treatment resulted in a 10-fold increase in protein concentration. The proteins in the concentrated CM were then analysed using a mass spectrometer (LTQ-Orbitrap, Thermo Scientific, Waltham, MA, USA) at the Jacques-Monod Institute (Paris, France). The data were normalized with respect to results obtained from LEAN-BMMSCs.

In vivo experiments
Matrigel plug assay Samples of supernatant CM from either LEAN-BMMSCs or ZDF-BMMSCs, in aMEM either supplemented with 10% foetal calf serum or serum-free, were freeze-dried using a Christ Alpha 1-2 freezer-drier overnight. At that time, 500 lg protein from each dried CM sample was mixed with 500 ll of ice-cold Matrigel (growth factor reduced Matrigel, VWR, France) and was injected subcutaneously into the flanks of 8week-old Swiss nude mice [19]. After 14 days, the mice were killed, and each Matrigel implant was excised, placed in 500 ll of RIPA buffer, and processed using a Retsch MM 300 tissue lyser at 30 pulses/min. for 2 min. The haemoglobin content in each excised implant was determined using a commercially available assay (Hemoglobin Colorimetric Assay kit, Cayman Ann Arbor, MI, USA). Plugs were also embedded in paraffin, sectioned, and stained with Masson's trichrome.

Statistical analysis
Each experiment of this study was performed in triplicate and repeated with three independent experiments using different cell preparations. Numerical data are reported as the mean AE standard error of the mean  (SEM). Student's t-tests for unpaired two-tailed samples were used for statistical analyses of the experimental data and compared to the respective controls. P value of < 0.05 was considered statistically significant.

CM from BMMSCs of diabetic rats affects HUVEC functions
To characterize the impact of diabetes type 2 (T2DM) on the BMMSC secretome, the effect of CM on selected functions (specifically migration, proliferation, viability and reactive oxygen species production) of HUVECs as well as in vitro and in vivo angiogenesis was examined.

CM from BMMSCs of diabetic rats increases HUVEC vascular-like tube formation in vitro
The angiogenic potential of CM collected from either ZDF-BMMSCs (CM-ZDF) or LEAN-BMMSCs (CM-LEAN) was determined using HUVECs on Matrigel in the presence of CM-ZDF, CM-LEAN, aMEM containing 10% FBS (FM) or FBS-free aMEM (NCM) for 24 hrs (Fig. 1A). The CM-LEAN group was the reference in this experiment. The supernatant CM-ZDF induced a significantly (P < 0.01) increased  and B). The protein content in the CM-LEAN and CM-ZDF was similar (Fig. 1C), but lower (P < 0.01) than that in the FM (Fig. 1C). These results suggest that MSCs from diabetic rats secrete soluble, bioactive, angiogenic mediators into their supernatant medium that affect HUVEC tube-like formation.

CM from BMMSCs of diabetic rats promoted increased HUVEC migration in vitro
The chemotactic potential of CM was examined using both the transwell cell migration assay and the in vitro scratch wound-healing assay. HUVEC migration through the porous transwell membrane (8 lm pore diameter) was observed after 24 hrs with CM-LEAN, CM-ZDF or FM, but not when NCM was used as the chemoattractant ( Fig. 2A). Compared with either CM-LEAN or FM, CM-ZDF induced significantly (P < 0.001) increased (3-fold) migration of HUVECs (Fig. 2B). After 6 hrs, HUVEC migration in the scratch wound-healing assay was significantly (P < 0.01) enhanced (1.4-fold) in the presence of CM-ZDF compared to results obtained with either CM-LEAN or FM (Fig. 2C and D). These results provide evidence that BMMSCs from diabetic rats secrete soluble, bioactive, chemotactic mediators into their supernatant media, which promote HUVEC migration.
CM from BMMSCs of diabetic rats does not affect either HUVEC proliferation or viability and has a limited effect on reactive oxygen species production The effect of the supernatant CM collected from either ZDF-BMMSCs or LEAN-BMMSCs after 24 hrs of culture on the proliferation of HUVECs was assessed. For this purpose, HUVECs were cultured in CM-LEAN, CM-ZDF, FM or NCM, and their proliferation was determined over a period of up to 7 consecutive days. Compared to results obtained from the other conditions tested, HUVECs cultured in FM exhibited increased proliferation starting at day 3; this trend increased by 4-fold at day 7 of culture ( Fig. 3A). The HUVECs did not survive in the absence of serum in the supernatant media (Fig. 3A). HUVECs cultured with either CM-LEAN or CM-ZDF survived but did not proliferate over the 7 days of the experiments. Moreover, the number of HUVECs under either CM-ZDF or CM-LEAN was similar for the duration of the experiments (Fig. 3A). The effect of the various supernatant CM on HUVEC viability after challenge with H 2 O 2 was assessed by exposing HUVECs to 150 lM of H 2 O 2 in each medium of interest to this study for 90 min. HUVECs exposed to CM-LEAN, CM-ZDF or NCM had similar numbers of cells, which were significantly (P < 0.001) higher than those observed under FM (Fig. 3B).
Furthermore, upon exposure to either CM-LEAN or CM-ZDF, production of ROS by HUVECs was higher when the cells were exposed to CM-ZDF than to CM-LEAN; this difference in ROS was approximately 15% after 1 hr of measurement (P < 0.001) (Fig. 3C).

CM from BMMSCs of diabetic rats promotes increased neoangiogenesis in vivo
To evaluate the proangiogenic potential of the CM-ZDF in vivo, Matrigel plugs were implanted in a mouse model. These Matrigel plugs, prepared using either CM-ZDF or CM-LEAN (each containing 500 lg/ ml of proteins), as well as either NCM or FM, were injected into the flanks of nude mice and were implanted for a 14-day period. At that time, the explanted plugs containing NCM, CM-LEAN or FM were pale in colour, indicating no or little blood vessel content (Fig. 4A). In contrast, plugs containing CM-ZDF appeared dark red due to the presence of red blood cells, indicating the formation of blood vessels ( Fig. 4A and B). Compared to plugs prepared with either CM-LEAN or NCM, the implanted Matrigel plugs containing CM-ZDF had a significant (P < 0.05) increase (2.5-fold) in haemoglobin content (Fig. 4C). In conjunction with the in vitro results, these in vivo data provided evidence that supernatant CM from ZDF-BMMSCs promotes increased angiogenesis in vivo.
ELISA analysis of the protein content in the conditioned media provided evidence that among the five different mediators tested, only the content of IGF-1 correlated (P < 0.001) with the observed difference in gene expression due to ZDF-BMMSCs (Fig. 5B). The concentrations of IL-6, MCP-1 and PLAU were similar in CM from LEAN-and ZDF-BMMSCs. TNFa and FGF were not detected in all media tested (Fig. 5B), while the VEGF concentration was significantly (P < 0.001) lower in ZDF-BMMSCs. The presence of bioactive IGF-1 in CM from ZDF-BMMSCs was confirmed by the reduced (by 60%; P < 0.001) HUVEC migration when this CM was supplemented with antibodies against IGF-1 ( Fig. 6A and B). Blocking IGF-1 in CM from ZDF-BMMSCs did not affect the formation of tubular structures (Fig. 6C  and D).
Some of these components could be related to angiogenesis and may be involved in the observed enhanced chemotactic response of endothelial cells (Fig. 7D). In fact, among the proteins differentially expressed in ZDF-BMMSC CM, seven proteins, specifically LTBP1 (2fold), LTBP2 (2-fold), OSTP (5-fold), cathepsin D (CATD) (2-fold), SAP (2-fold) and FMOD (5-fold), are related to angiogenesis and were overexpressed (Fig. 7D). Blocking either LTBP1 or LTBP2 in the CM from ZDF-BMMSCs resulted in a decrease (by 40%) of the stimulatory effect of these chemical compounds on HUVEC migration ( Fig. 8A and B) but   did not affect tubular structure formation (Fig. 8C and D). These results indicate that compared to the secretome of control BMMSCs, the secretome of ZDF-BMMSCs contains matrix-related proteins (such as LTBP1 or LTBP2) and promotes angiogenesis (Fig. 8).

Discussion
Despite the fact that BMMSC-derived paracrine factors were reported to be largely involved in neovascularization and tissue repair, the impact of type 2 diabetes mellitus (T2DM) on the BMMSC secretome and its effects on angiogenesis are still unknown [16,17]. We, therefore, performed a comparative analysis of the BMMSC secretome and its effects on endothelial cells, using ZDF rats at 13 weeks to represent early diabetic conditions and their LEAN littermates as controls. The results obtained provided the first evidence that short-term T2DM alters the BMMSC secretome composition and promotes angiogenic capabilities.
The results of this study demonstrated that angiogenesis-related genes are differently expressed in BMMSCs from ZDF rats when compared with results from LEAN animals. In fact, several pro-angiogenic genes were overexpressed, while anti-angiogenic genes were downregulated (Fig. 5). The up-regulated (over 5-fold) genes included IGF-1 and TIE1, which are critical regulators of angiogenesis [20], MCP-1 and IL-6, homing factors for BMMSCs and EC/EPCs, respectively [21,22], and IL-6 and TNFa, critical mediators of the inflammatory process. Nevertheless, the expression of several angiogenic genes did not correlate with the released protein levels in the supernatant CM in this study. Analysis of the protein content in the 24-hr secreted CM from LEANand ZDF-BMMSCs provided evidence that among the different mediators tested, only the content of IGF-1 correlated (P < 0.001) with the observed difference in gene expression by ZDF-BMMSCs (Fig. 5B). However, proteomic analysis did not reveal IGF1 in the CM. This imperfect and variable correlation between ELISA and proteomic analysis can be explained by different levels of sensitivity and dynamic ranges of the methods [23]. The concentrations of IL-6, MCP-1 and PLAU were similar in CM from LEAN-and ZDF-BMMSCs, while that of VEGF was lower in ZDF-BMMSCs. TNFa and FGF were not detected in all media tested. IGF-1 plays a significant role in the physiology of endothelial cells by promoting their migration [20,24]. IGF-1 is also overexpressed by several types of cells implicated in T2DM-proliferative retinopathy [25,26]. Under our experimental conditions, IGF-1 released by ZDF-BMMSCs accounted for approximately 60% of the observed enhanced HUVEC migration ( Fig. 6A and B). The IGF-1 effect is mediated by activation of the IGF-1 receptor, which triggers several signalling pathways and therefore several cellular responses. For instance, the Ras/Raf pathway is critical for proliferative responses, whereas activation of Rac is important for cell migration [27]. This study provides evidence that IGF-1 might be a key factor in the paracrine action of T2DM-BMMSCs on endothelial cell behaviour, thus contributing to the pathogenesis of microvascular complications associated with the diabetic condition, including diabetic retinopathy.
Moreover, proteomic analysis of the T2DM-BMMSC secretome revealed some proteins that are related to angiogenesis through several mechanisms. Specifically, the results showed (i) decreased levels of ab-crystallin, a chaperone for VEGF-A, and (ii) increased levels of LTBP1 and LTBP2, regulators of TGF-b availability [28], as well as of OSTP and FMOD, which are components of the extracellular matrix and might be involved in the paracrine action of T2DM-BMMSCs on endothelial cells [29][30][31][32]. The role of extracellular matrix components in angiogenesis has been highlighted by both in vitro and in vivo studies [33,34]. In this respect, this study revealed a different secretory profile of BMMSCs that may explain how these cells affect the environmental niche and paradoxical angiogenesis in the diabetic milieu. ROS stands at the cross-point of the regulations exerted on the BMMSCs as a signalling molecule in the bone marrow environment [35]. For instance, low levels of ROS are associated with quiescent MSCs, whereas an increased ROS level is typical of proliferating cultures.
In addition, the proteomic analysis of T2DM-BMMSC CM demonstrated a specific secretory phenotype of extracellular matrix remodelling and glucose metabolism. Compared to LEAN-BMMSCs, of the 261 proteins released by ZDF-BMMSCs, 44 were differentially secreted, with 27 at higher and 17 at lower levels (Fig. 7). The 27 overexpressed proteins included extracellular matrix homeostasis and remodelling-related molecules [36]. In contrast, proteins involved in the metabolism of glucose (such as ALDOA, LDHA, KPYM, G6P, PTMA, OAS2, ALD1 and IBP2) were secreted at lower levels (Fig. 7C). Impaired glucose metabolism was already observed in T2DM muscle cells [37], but to our knowledge, this is the first report in the BMMSC secretome. Targeting such molecules may offer insight into the underlying mechanisms of T2DM-BMMSC physiology.
Compared with LEAN-BMMSCs, the ZDF-BMMSC secretome was unique, despite the fact that ZDF-BMMSCs were extensively expanded by ex vivo culture at normal glucose levels. These findings show that the effects of the diabetic milieu were evident on the BMMSCs, even after multiple cell divisions in vitro in a normoglycaemic environment. Moreover, these results corroborate literature reports that aspects of the diabetic condition, particularly hyperglycaemia, induce transmissible alterations in many type of cells, including endothelial, vascular smooth muscle, retinal and cardiac cells [38]. In fact, the process by which hypo-methylation of gene promoters leads to deregulated gene expression in T2DM is known as 'metabolic memory' [38,39] and is implicated in the persistence of vascular complications of diabetes, even after return to normal blood glucose levels [38,39]. Regarding ZDF rats, alteration of hepatic DNA methylation at early diabetic conditions (12 weeks) was previously reported [40]. For this reason, it is possible that the altered secretome observed in ZDF-BMMSCs is due to a 'metabolic memory' phenomenon. Further studies are needed to determine whether such changes are initiated during the pre-diabetic state in the ZDF rat model.
This study is the first to directly compare both in vitro and in vivo the angiogenic potential of BMMSC-CM from T2DM and control animals. In line with previous studies reporting the angiogenic potential of the BMMSC secretome [17,41], BMMSC-CM from both, ZDF rats and age-matched LEAN rats also showed angiogenic properties. These results corroborated previous studies establishing that healthy rat BMMSCs and CM-BMMSC improve the formation of HUVEC tubular structure in both paracrine [41,42] and juxtacrine manners [41]. The CM from BMMSCs, however, affected endothelial cell functions differently. It promoted HUVEC migration and tube formation with no change in proliferation. Such differences are due to changes in the levels of several bioactive compounds, such as IGF-1, LTBP1 and LTBP2, and reflect the interactive and possible synergistic action of those multiple bioactive mediators leading to a promotion of angiogenesis. Another plausible explanation for the differences in effects of CM from BMMSCs on HUVEC functions may be the relative concentrations of paracrine factors. For instance, IGF-1-related effects are mediated by activation of IGF-1 receptors, which trigger, among others, the Rac pathway. This pathway is important for cell migration but not proliferation [27]. Our findings are in line with Gruber et al. [43] who suggested that MSCs paracrine effects might not be responsible for the initiation of the angiogenic process, but rather would play a role in guiding the growing blood vessels. However, in contrast with these data, Kinnaird et al. [17] reported that human BMMSCs MC increases EC proliferation. The reasons from these mixed and discrepant results are not really surprising if one takes into consideration the differences in isolation methods, species (rat versus human) and EC sources use [44]. Moreover, in this study, CM from rat BMMSCs induced vascular formation in matrigel plugs in vivo. These findings are in line with previous studies [45] that show the occurrence of functional blood vessels in rBMMSC-CM loaded Matrigel plugs. Taken together, these data indicate that BMMSCs secretome may contribute to EPCs recruitment and subsequent formation of functional vessels.
Another important result of this study is that the CM from T2DM-BMMSCs increased in vitro endothelial cell migration and tubular structure formation, as well as in vivo vascular formation in matrigel transplants (Fig. 4). The greater angiogenic potential of T2DM-BMMSCs secretome may reflect the interactive actions of multiple bioactive mediators (including IGF-1, LTBP1 and LTBP2) which may alter the functions of endothelial cells, key players in diabetic vascular complications [46]. Angiogenesis results from an equilibrium between pro-and anti-angiogenic factors [46]. The increased angiogenesis observed in this study may arise from a disturbed balance between these pro-and anti-angiogenic factors and highlights the paracrine effect of an altered BMMSCs secretome on endothelial cells. It further illustrates the ambiguous role of angiogenesis in the pathogenesis of diabetic vascular complications. Thus, exacerbated angiogenesis occurs in diabetic retinopathy and nephropathy, whereas reduced angiogenesis contributes to diabetic impaired wound healing. A similar response to that observed in this study had been reported for BMMSCs cultured under in vitro hypoxic and high glucose conditions used to mimic the diabetic environment in vitro [47,48]. This increase in angiogenic response may reflect the 'adaptive' BMMSC response to the severe stress conditions of the diabetic milieu. Because CM delivers paracrine/trophic angiogenic factors, it is important to take these aspects of diabetic pathology into consideration in the development of BMMSC-based therapies aimed at improving tissue repair in diabetes. Last but not least, the findings reported here are somewhat different from the results of Dzhoyashvili et al. [49] who have shown that CM from human diabetic adipose tissue exhibited a reduced angiogenic potential. These apparently contradictory data further illustrate the complexity the pathogenesis of diabetic vascular complications and deserve supplementary investigations to fully decipher the molecular basis of MSC paracrine effects in diabetes.
A limitation of this study is the fact that it is difficult to separate the effects of hyperglycaemia from those of obesity. The ZDF rat was chosen as the animal model because these animals initially develop insulin resistance followed by T2DM at predictable ages and display many of the human T2DM characteristics, including obesity, abnormal blood lipid profiles, and related vascular complications [18], making them a clinically relevant model.
In summary, this study provided the first evidence that compared with control animals, BMMSCs from T2DM rats exhibit an altered secretome pattern and angiogenic properties and underline the critical role of paracrine factors of MSC-mediated angiogenesis in the diabetic milieu. These results provide valuable insight in the vascular complications associated with diabetes and could be used in the development of BMMSC-based therapies to improve diabetic tissue repair.