Mesenchymal Stem/ Stromal Cells metabolomic and bioactive factors profiles: A comparative analysis on the umbilical cord and dental pulp derived Stem/ Stromal Cells secretome

Mesenchymal Stem/ Stromal Cells assume a supporting role to the intrinsic mechanisms of tissue regeneration, a feature mostly assigned to the contents of their secretome. A comparative study on the metabolomic and bioactive molecules/factors content of the secretome of Mesenchymal Stem/ Stromal Cells derived from two expanding sources: the umbilical cord stroma and the dental pulp is presented and discussed. The metabolic profile (Nuclear Magnetic Resonance Spectroscopy) evidenced some differences in the metabolite dynamics through the conditioning period, particularly on the glucose metabolism. Despite, overall similar profiles are suggested. More prominent differences are highlighted for the bioactive factors (Multiplexing Laser Bear Analysis), in which Follistatin, Growth Regulates Protein, Hepatocyte Growth Factor, Interleukin-8 and Monocyte Chemotactic Protein-1 dominate in Umbilical Cord Mesenchymal Stem/ Stromal Cells secretion, while in Dental Pulp Stem/ Stromal Cells the Vascular Endothelial Growth Factor-A and Follistatin are more evident. The distinct secretory cocktail did not result in significantly different effects on endothelial cell populations dynamics including proliferation, migration, tube formation capacity and in vivo angiogenesis, or in chemotaxis for both Mesenchymal Stem/ Stromal Cells populations.


Introduction
Mesenchymal Stem/ Stromal Cells (MSCs) are at the forefront of research for the development of cell-based therapies, due to their capacity to self-renew and differentiate into several cell types, to secrete soluble factors with paracrine actions, as well as due to their immunosuppressive and immunomodulatory properties [1][2][3][4][5][6].
Currently, the umbilical cord stroma (Whärton jelly) and the dental pulp may come to gain ground as sources for MSCs-based therapies, due to the non/ minimally invasive and ethically accepted collection procedures (umbilical cords and extracted healthy teeth were previously considered medical waste), as well as for the increasingly available private and public banking options worldwide [12].
The first evidence of the MSCs contribution to the healing processes was assigned to their specific differentiation skills, replacing the damaged native cells in their functions [13,14]. However, current trends demonstrate that in some instances MSCs remain undifferentiated at the lesion site or in its vicinity, for limited periods of time, or even that only minimal percentages of the MSCs would effectively differentiate and integrate host tissues [15]. Regardless of their differentiation into tissue specific phenotypes, MSCs are often correlated to improved regenerative outcomes [16].
These observations were then attributed to the secretion products of those MSCs [17][18][19] and, in recent years, research has focused on deepening the knowledge on the effective composition of the MSCs secretion, in the form of soluble molecules or extracellular vesicles [6,[20][21][22][23].
In most tissues, the key for regenerative efficiency is the re-vascularization of the lesion site and MSCs have been associated with improved angiogenesis in a number of models of disease [24,25]. As such, MSCs assume a supporting role to the intrinsic mechanisms of tissue regeneration, promoting the re-vascularization processes, providing adequate perfusion to active healing sites, as well as urging resident regenerative populations to home towards these locations [26].
Further, some groups investigated the extent to which the presence of the cells themselves was absolutely essential to the observation of beneficial effects, since regenerative benefit can be observed by the application of MSCs secretion products alone, conventionally designated as the secretome [8,[17][18][19]27].
The secretome comprises a range of bioactive molecules/factors secreted to the extracellular space. Its composition is particular to individual cells and tissues, and is modulated in response to physiological and/or pathological stimuli [24]. The application of these cell-based products may bring several advantages to the advanced therapies field, namely the decreased cell number requirements and allocated cell storage necessities, ease of tailoring, quality control and dosing, reduced risk of rejection and malignancy, as well as the ready availability for administration in acute scenarios [28].
Therefore, studies on the composition of the MSCs secretome through metabolic analysis are a valuable tool to the comprehension of the underlying mechanisms to MSCs dynamics and therapeutic effects [29][30][31][32].
Metabolomic profiling techniques [33][34][35][36][37] yield information on targeted metabolites' structure and quantitative distribution [33,34], and despite the significant progress made in the field of structural biology and bio-chemistry, the development and application of these techniques towards the MSCs secretome are still sparse [8].
We recently demonstrated the application of proton NMR spectroscopy and implementation of appropriate one (1D) and two (2D) dimensional NMR techniques to the analysis of the metabolic composition of Umbilical Cord Stem/ Stromal Cells (UC-MSCs) conditioned media and changes in the metabolic profile of the culture media in the process of conditioning [21].
Alongside the metabolite content, a wide range of growth factors, cytokines, chemokines and extracellular matrix components have already been identified in the CM obtained from differently sourced MSCs [38,39], and many of them are known to impact on most tissues structure, function and regeneration [25,40,41]. Beyond modulating their surrounding environment MSCs are sensitive themselves to signaling factors, altering their secretory profile in response to, as an example, the presence of inflammatory cytokines. This evidence may contribute to the understanding of the eventual difference in the tissue response to cells or to their secretome alone [42], and to the development of strategies to manipulate secretion profiles towards specific needs.
In a recent study, we determined that UC-MSCs CM becomes rich in a range of proliferative and anti-apoptotic factors, particularly in transforming growth factor beta 1 (TGF-β1), epidermal growth factor (EGF), granulocyte and granulocyte-macrophage colony-stimulating factor (G-CSF and GM-CSF), platelet derived growth factor (PDGF-AA) and vascular endothelial growth factor (VEGF). Also, several other chemokines such as monocyte chemotactic protein-1 e 3 (MCP-1 and MCP-3), chemokine (C-C motif) ligand 5 (CCL5 or RANTES), GRO, and interleukin 8 (IL-8) were observed in increasing levels [21]. This previous analysis enabled the primary definition of UC-MSCs CM composition [21], which we herein expand and compare to Dental Pulp Stem/ Stromal Cells (DPSCs) CM profile.
The first part of the present study focuses on a comparative analysis of the metabolomic and bioactive factors secretion profile of UC-MSCs and DPSCs secretome, through the conditioning process. This aimed to characterize the metabolite profile of the collected CM and to enlighten on the metabolic pathways presiding to the process. Further, the bioactive factors content of the CM was analyzed, focusing on a series of growth factors, cytokines and chemokines, aiming at the comparison of the regenerative potential of the two MSCs population. Finally, the effects of the characterized CMs were addressed on in vitro and in vivo models of angiogenesis.

Reverse transcriptase polymerase chain reaction.
Reverse transcriptase Polymerase chain reaction (RT-PCR) and qPCR targeting specific genes expressed by pluripotent stem cells was performed. Gene DNA sequences were downloaded from GenBank (www.ncbi.nlm. nih.gov/genbank) and aligned using the Clustal Omega bioinformatic tool from EMBL-EBI (http://www.ebi.ac.uk/Tools/msa/clustalo). The primers sequences are listed in Table 1.
Cultures were harvested and pelleted for total RNA extraction (High Pure RNA Isolation kit (Roche TM )). DNA traces were eliminated with DNase I, and RNA quantity and quality assessed by Nanodrop ND-1000 Spectrophotometer, reading from 220 nm to 350 nm, and stored at -80˚C. After, cDNA was synthesized from the purified RNA (kit Ready-To-Go You-Prime First-Strand Beads (GE Healthcare 1 )).The synthesized cDNA, corresponding to the mRNA present in the sample, was run for the expression of six genes: two housekeeping genes (β-actin and GAPDH) and four genes used as pluripotent stem cells markers (c-kit, Oct-4, Nanog and ALP). Quantitative PCR (qPCR) was performed in a CFX96 TM (BioRad 1 ) apparatus using the iQ TM SYBR 1 Green Supermix (BioRad 1 ). Each pair of primers targeting a gene was used to analyze its expression in the UC-MSCs and DPSCs cDNA, in duplicate, along with a negative control, through defined temperature cycles [95˚C for 4 minutes, 35 cycles comprising 95˚C for 20 seconds, 55˚C for 20 seconds and 72˚C for 30 seconds ending with Real-Time acquisition, and final extension of 72˚C for 7 minutes], and the number of cycle threshold for each well were recorded. The plate containing the amplified genes or qPCR products was kept in ice and observed in a 2% agarose gel (NuSieve 1 3:1 Agarose (Lonza)) to check and reinforce the identity of the amplicons, in horizontal electrophoresis apparatus. Under 120 V potential difference for 40 minutes to separate the amplicons. Gel was then observed under UV light and pictures recorded using the GelDoc 1 2000 (BioRad 1 ) and Quantity One 1 software (BioRad 1 ).

Multilineage differentiation.
Multilineage differentiation of the UC-MSCs and DPSCs was induced towards Osteogenic, Adipogenic and Chondrogenic phenotypes using specific differentiation media. Differentiation efficiency was assessed by Alizarin Red S, Oil Red O and Alcian Blue/ Sulfated Glycosaminoglycans (GAGs) quantification, as detailed in [43].

Conditioned medium collection and analysis.
For the production of CM, both cellular populations were plated at 6000 cells/ cm 2 , in triplicates, and cultured in standard culture medium (αMEM 10% FBS) until approximately 80% confluence was reached. Then, the cultures were washed to remove any trace of FBS and other supplements and placed in plain DMEM/F12 GlutaMAX™ (10565018, Gibco 1 ) for 24 and 48 hours. Conditioned media samples were collected at both time-points, centrifuged 1800 xg for 10 min to remove any cellular debris and preserved at -80˚C until analysis.

Metabolomic analysis.
Proton NMR spectroscopy was used to identify and quantify metabolites in plain/unconditioned and conditioned media collected at two time-points (24 and 48 hours) of UC-MSCs and DPSCs in proliferation. A total 600 μL aliquot of each sample was placed into 5 mm NMR tubes and 50 μL deuterium oxide (D 2 O) containing 0.05 mM sodium trimethylsilyl-[2,2,3,3-d4]-propionate (TSP) was added as a chemical shift and quantitation standard. All NMR spectra were recorded at 300K on a Bruker Avance III 600 HD spectrometer, equipped with CryoProbe Prodigy. 1 H NMR spectra with water suppression using a 1D NOESY (noesygppr1d) and a Carr-Purcell-Meiboom-Gill (CPMG, cpmgpr1d) pulse sequences [44,45] were acquired with spectral width of 10000 Hz, 32 K data points and 32 scans. The 1D NOESY spectra were collected using 5 s relaxation delay and mixing time of 100 ms. The CPMG experiments (cpmgpr1d) were acquired with relaxation delay 4 s, spin-echo delay between 0.4 and 0.6 ms, and a loop for T2 filter of 20 was used. All free induction decays (FIDs) were processed by 0.3 Hz line broadening and zero filling to 64 K, manually phased, and baseline corrected. Two dimensional 1H/1H COSY and TOCSY spectra were recorded in phase sensitive mode and with water suppression; a relaxation delay of 2 s, 16 or 32 scans, a total 2K data points in F2 and 256 or 512 data points in F1 over a spectral width of 10000 Hz. Two dimensional 1H/13C heteronuclear single quantum coherence (HSQC) experiments were carried out with a spectral width of 10000 Hz for 1H and 27000 Hz for 13C, relaxation delay 1,5 s, Fourier transform (FT) size 2K × 1K. The quantitative distribution of NMR-detectible metabolites in the samples was determined from the integral intensity of characteristic signals in 1 H NMR spectra of the samples referenced to the integral intensity of TSP signal, considering the number of the contributing nuclei for that particular resonance signal [46].

Bioactive factors detection and quantification.
Multiplexing LASER Bead Analysis was performed by Eve Technologies (Calgary, Alberta, Canada) using the Bio-Plex™ 200 system (Bio-Rad Laboratories, Inc., Hercules, CA, USA), similarly to previously described in [47]. The multiplexing technology is based on color-coded polystyrene beads with unique color / fluorophore signatures, and a dual-laser system and a flow-cytometry system. One laser activates the fluorescent dye within the beads which identifies the specific analyte, the second laser excites the fluorescent conjugate (streptavidin-phycoerythrin), and the amount of the conjugate detected by the analyzer is in direct proportion to the amount of the target analyte. The results are quantified according to a standard curve.

Effects of MSCs CM on angiogenesis
2.3.1. Conditioned medium concentration protocol. Conditioned Medium obtained from the UC-MSCs and DPSCs was concentrated 5 times (5x), using Pierce™ Protein Concentrator, 3k MWCO, 5-20 mL (88525, Thermo Scientific). Tubes were sterilized as per manufacturer's instructions, through immersion in 70% (v/v) ethanol for 30 minutes, followed by 2 cycles of 70% ethanol and DPBS (14190144, Gibco 1 ) centrifugations at 3000 xg, for 10 minutes each). Tubes were air dried for 10 minutes and CM placed ate the top compartment. Samples were centrifuged until 5x concentration of sample volume in the upper compartment was achieved.

Human umbilical vein endothelial cells expansion and maintenance.
Human Umbilical Vein Endothelial Cells (UVECs) were obtained from Sigma 1 (200P-05N; Lot No. 3257, Cell Applications, Inc) and expanded using specific expansion medium (211-500, Cell Applications, Inc). Cells were maintained at 37˚C and 95% humidified atmosphere with 5% CO 2 environment. Passage 4-5 cultures were utilized in the presented assays.

Cell viability assay.
Cells were seeded in 24 well plate at 6000 cells/cm 2 . The cells were allowed to adhere in UVECs expansion medium, overnight. After this period, culture media was replaced by CM supplemented culture media [composed of 80% UVECs expansion medium and 20% 5x concentrated UC-MSCs' CM or DPSCs' CM; control medium was composed of 80% UVECs expansion media and 20% DPBS].
At every time point (0, 24, 48, 72 and 96 hours), the culture media was removed and fresh adequate culture media was added to each well, with 10% (v/v) of 10x PrestoBlue 1 cell viability reagent (A13262, Thermo Fisher Scientific, Molecular Probes, Invitrogen), and UVECs were incubated for 1 hour at 37˚C, 5% CO 2 . The absorbance of the cell culture supernatant was read at 570 nm and 595 nm in a Thermo Scientific Multiskan FC plate reader, to assess for changes in cell viability. Absorbance readings were normalized, and data corrected to unseeded control wells' readings.

Senescence and apoptosis assays.
To assess for signs of senescence in the cellular populations, β-Galactosidase activity was assayed. Cells were plated in 96-well plates, similarly as described for the cell viability assessment. After 48 hours in CM-supplemented and control media, culture media was removed, and cells were washed once with DPBS and incubated with β-Galactosidase Assay Reagent (75705, Thermo Fisher Scientific), for 30 minutes at 37˚C. Absorbance was read at 405 nm.
Cultured UVECs were further assessed for apoptosis events, through the detection of Annexin V and Propidium Iodine (PI) staining (BMS500FI, eBioscience). For such, cells cultured for 2 days in each condition were harvested using Trypsin-EDTA, counted and resuspended in provided Binding Buffer. Cells were incubated with Annexin V-FITC and PI, as per manufacturer's protocol, and FACS analysis was performed using a Coulter Epics XL Flow Cytometer (Beckman Coulter Inc., Miami, FL, USA). Flow cytometry data was processed using FlowJo Engine X10.4 (v3.05478, LLC).
to prevent Matrigel 1 gelation. The dorsum was virtually divided into four quadrants (each assigned to an experimental group) and a final volume of 500 μL was injected subcutaneously. Upon injection, the formation of a protruding plug was confirmed. Animals were recovered from anaesthesia. After 7 days, animals were placed again under general anaesthesia and sacrificed by lethal intraperitoneal injection of 5% sodium pentobarbital (Eutasil 1 200 mg/mL, Ceva).
Injection sites were confirmed and dorsal subcutaneous tissue enclosing the vascularised plug were collected and preserved in paraformaldehyde (3.7-4% buffered to pH 7, Panreac AppliChem 1 ).
Collected tissues were processed for routine histopathologic analysis. Sequential sections (3 μm) were prepared and stained with haematoxylin-eosin (H&E) and assessed for the presence and pattern of endotheliocyte/ capillary penetration into the matrix plug. For the immunohistochemical study, sections were deparaffinized in xylene and rehydrated in sequential graded alcohols. Antigen retrieval was performed in EDTA, pH 8,0 for 30 minutes in water bath 100º C. The NovolinkTM Max-Polymer detection system (Novocastra, Newcastle, UK) was used for visualization, according to the manufacturer's instructions. Slides were then incubated with anti-VEGFR2 antibody (clone 55B11; Cell Signaling Technology, Boston, USA), diluted 1:300, overnight at 4ºC. Colour was developed with 3.3-diamino-benzidine (DAB; Sigma, St. Louis, MO, USA) and sections were then counterstained with haematoxylin, dehydrated and mounted. Sections of normal rat skin were used as positive control and negative controls were performed by replacing the primary antibody with another of the same immunoglobulin isotype.
The tissue areas containing Matrigel 1 plugs and the surrounding tissue were microscopically evaluated in order to identify the regions of highest vascular density. Vessels which showed unequivocal brown VEGFR2 immunostaining were counted manually at higher magnification (×40) in at least 10 different regions of each sample.

Statistical analysis
Data was plotted and treated using GraphPad Prism version 6.00 (GraphPad Software, La Jolla California USA). Data was further analyzed for significant differences. Multiple comparison tests were performed by one-way ANOVA supplemented with Tukey's HSD post-hoc test. Differences were considered statistically significant at P < 0.05. Significance of the results is indicated according to P values with one, two, three or four of the symbols ( � ) corresponding to 0.01�P<0.05; 0.001�P< 0.01; 0.0001�P<0.001 and P<0.0001, respectively.
RNA was purified from both UC-MSCs and DPSCs and converted to cDNA using adequate procedures. Primers targeting two housekeeping genes (β-actin and GAPDH) and four typical pluripotency markers (c-kit, Oct-4, Nanog and ALP) were used to support its identity. Table 2 details on the average of Threshold cycle (Ct) values. Total RNA was successfully purified from UC-MSCs and DPSCs using the High Pure RNA Isolation kit (Roche), obtaining a concentration of 883.2 ng/μl from UC-MSCs and 158.7 ng/μl from DPSCs. Volumes were adjusted to use only 1.5 μg of total RNA to synthetize corresponding cDNA. This cDNA was used as template in the qPCR reaction. The Ct values show a reasonable amplification of the target genes, in both cell types, resulting in an active expression of these genes in both UC-MSCs and DPSCs. The agarose gel confirms the identity of the genes from the observed molecular weight.
The strongest expression is observed in the housekeeping genes, while the lowest expression was observed for the Oct-4 sequence (as evidenced by the lowers and higher ΔCt values determined for the amplification cycles, respectively) ( Table 2).
The differentiation capacity of MSCs towards three mesodermal lineages was confirmed (Fig 2). Macroscopic observation revealed calcified matrix formation, evidenced by ARS staining in both UC-MSCs and DPSCs under osteodifferentiation conditions. Semi-quantitative   difference in terms of sGAGs production within the differentiated and the undifferentiated wells.

Conditioned medium collection and analysis
Populations from both tissue sources were successfully seeded and reached the desired density within 3 to 4 days. For conditioning time, cells remained well attached and preserved their characteristic spindled shape, although a decrease in the proliferation rate could be empirically observed.

Metabolomic analysis
Proton NMR spectroscopy was used to define the metabolic profiles of UC-MSCs and DPSCs through the analysis of produced CMs. To identify metabolite changes induced by DPSCs and UC-MSCs conditioning, various 1D and 2D NMR experiments were performed, and 1 H NMR spectra acquired using appropriate NMR techniques for suppression of the solvent and other interfering signals. The same experimental conditions (temperature, sample/solvent volume, standard (TSP) concentration, NMR acquisition and processing parameters) were used in order to avoid external interferences and to identify particular metabolite variations. The assignment of the proton resonances in NMR spectra and the assessment of the metabolite composition of the samples were achieved by high resolution 1 H NMR spectra analysis, considering specific NMR parameters which reflect structural characteristics of the respective species. The results were verified by the implementation of appropriate 2D NMR ( 1 H/ 1 H COSY, 1 H/ 1 H TOCSY, 1 H/ 13 C HSQC) spectroscopic techniques and compared to available data in the literature [33,34]. The assignment of the resonance signals of metabolites in 1 H NMR spectra of the samples studied is presented in Table 3. Characteristic resonance signals of metabolites identified in 1 H NMR spectra of Plain (unconditioned), UC-MSCs and DPSCs CM are labelled in Fig 3. The visual inspection of the 1 H NMR spectra (Fig 3) suggests similar profile shapes but different concentration distribution of the metabolites. To assess characteristic metabolic changes occurring during hMSCs conditioning, comparative quantitative analysis using 1 H NMR spectra of the samples from the three groups was performed. The metabolites concentration was determined from the integral intensity of characteristic signals in 1 H NMR spectra of the samples referenced to the integral intensity of TSP signal used as an internal standard and considering the number of the contributing nuclei for that particular resonance signal. The concentration of the metabolites identified in the 1 H NMR Where: • Wx represent the mass of the metabolite; • Mx and Mstd are the molar masses of the metabolite and the standard (TSP was used), respectively; • Ix and Istd are the integrated signal area of the metabolite and the standard, respectively; • Nx and Nstd are the number of protons in the integrated signal area of the metabolite and the standard; • mStd represent the mass of the standard.
The concentration of the metabolites (μM) identified in each group, as well as the chemical shift and assignment of the signal used for quantification are presented in Table 4. Statistical significance of encountered differences in Metabolite concentration (μM) is presented in S1 Table. All 1 H NMR spectra are dominated by the proton resonance signals of glucose but several low molecular weight compounds such as amino acids (e.g. alanine, tyrosine), organic acids (e.g. lactate, acetate, citrate, formate) and choline are identified (Fig 3).
Additionally, in the spectra of the Plain Medium, resonance signals characteristic for L-alanyl-L-glutamine (GlutaMAX TM I (L-Ala) at 1.42 and GlutaMAX TM II (L-Glu) at 2.37 ppm) dipeptide from GlutaMAX TM are clearly observed. The intensity of these resonances significantly decreases through the cell conditioning process, which is especially evident after 48 hours of culture conditioning. Glutamine and alanine, the resulting compound from Gluta-MAX TM and glutamate's metabolism, expectedly rise through culturing times. Remarkably, UC-MSCs' CM increase in these metabolites' content occurs at 24 hours, before significant GlutaMAX TM consumption can be observed. For the DPSCs population, this process appears to occur at a slower rate (Fig 4A and Fig 4B).
Glucose is the most prominent metabolite identified in the spectra and therefore in greatest concentration. In the media collected after 24 hours of conditioning of UC-MSCs, a significant increase of α and β-Glucose content is observed, when compared to the Plain Medium (which was supplied to the culture at the start of the conditioning period), as well as when compared to the DPSCs at the same conditioning time. This initial spike in glucose content decreases in the following hours, reaching back to values not significantly different from the Plain Medium. Human DPSCs present no significant variation in glucose content throughout the study period.
Pyruvate is an energetic compound that is catabolised through aerobic or anaerobic pathways. The analysis of the Plan Media confirms it is one of the supplemented metabolites. Both cell types appear to present a tendency for pyruvate consumption in the 48 hours of conditioning (Fig 4C and Fig 4D), but significance in such reduction is only found for UC-MSCs. Catabolites resulting from its metabolism are also detected in the spectra, such as acetate, lactate and ethanol. Acetate accumulated in UC-MSCs cultures, particularly in the first 24h ( Fig  4C). In DPSCs CM, Acetate concentrations remained stable though the evaluation (Fig 4D). Lactate, on the other hand, accumulated strongly within the first 24 hours in both cases, with little increase afterwards, reaching maximum concentration at 48 hours of incubation. Ethanol presented a steady rate of accumulation in the culture supernatant for both UC-MSCs and DPSCs.
Other compounds such as choline, formate, tyrosine and nicotinamide are also detected at low levels in both Plain and Conditioned media.

Bioactive factors detection and quantification.
Multiplexing LASER Bead Analysis was performed for the detection of a total of 57 Bioactive factors in the CM obtained from UC-MSCs and DPSCs, after 24 and 48 hours of conditioning. Obtained measurements were normalized to the Unconditioned/ Plain medium (naturally devoid of artificial bioactive factors), run as blank sample. Fluorescent readings were plotted against a standard curve and calculated concentrations are presented in Table 5. Statistical significance of encountered differences in the detected Bioactive factors (pg/mL) is presented in S2 Table. After 24 hours of conditioning most factors presented no or little expression. Some of the factors gained expression after 48 hours, becoming detectable and, occasionally, in significant concentrations. Overall, none to minimal signal intensity was detected in 36 of the 57 assayed factors, combining the two timepoints.  A similar pattern of superior secretion by UC-MSCs is observed throughout, highlighting HGF, IL-8, and MCP-1 expressive quantities. Exception is appointed to follistatin and, most significantly, to VEGF-A, where DPSCs depict increased levels when compared to UC-MSCs.

Effects of MSCs CM on angiogenesis
3.3.1. Cell viability, senescence and apoptosis assays. The CM obtained from MSCs cultures was employed as culture medium supplement for in vitro culture of UVECs, to assess for its ability to sustain the activity and proliferation of the endothelial populations. PrestoBlue 1 viability was used as a metabolic indicator. Obtained corrected absorbance values are presented in Fig 5 and Table 6 (statistical differences on observed measurements detailed in S3 Table). At 24 hours, UVECs cultured in medium supplemented with DPSCs CM depict slightly increased activity, with no other noteworthy differences. As time progresses (48 and 72 hours), both DPCSs and UC-MSCs CM supplemented groups sustain superior cellular activity than Complete and Control media. At about 72 hours of culture, the UVECs cultured in CM supplemented media reached confluence and entered the plateau stage. The activity of Control medium group also decreased from that point onwards, although cellular confluence was unmatched to the CM supplemented groups. At 96 hours, only the Complete medium continued to present an increase in cellular metabolism but remained slightly inferior to that provided by DPCSs CM.
As for apoptosis events in the cultured population, none of the CM supplemented groups revealed noteworthy impact in cellular viability (Fig 5B, Table 7 and S4 Table). The most striking observation is that the Control group presented increased events of early and late stage apoptosis, significantly decreasing the percentage of viable cells. As for the groups supplemented with DPSCs or UC-MSCs CM, the first presented increased cellular viability, as opposed to the early apoptotic and cell death events identified in the UC-MSCs CM supplemented group. The CM supplemented groups matched the events of each stage recorded for UVECs cultured in their standard Complete media, except for the early apoptosis, where DPSCs CM provided an inferior percentage.

Migration assays
After wound induction on the cellular monolayers with the pipette tip, the area of the produced defect did not differ between groups/wells, deeming them adequate for comparison through the healing time.
In one instance, we investigated the chemotactic effect of the MSCs CMs on endothelial populations, represented by HUVECs. No significant differences in recovered area of the wound were identified until 14 hours into the assay, when DPSCs CM supplemented groups depicted increased percentage of covered wound area, when compared to the Complete and Control media (0.0001�P<0.001) (Fig 6 and Table 8). At the final assessment time (16 hours), both CM supplemented groups presented over 90% area coverage, while Complete and Control media groups remained under 85% coverage of the originally induced 'scratch' (0.0001�P<0.001 to P<0.0001).
Further, we assessed the effect of the CMs on the chemotaxis of the two MSCs populations under study. After 6 hours of 'scratching' and media replacement of the MSCs monolayers, no differences were observed in cellular migration in Control and Complete Medium groups between, and within the two MSCs sources (Fig 7 and Table 9). The CM supplemented groups depicted superior area, with exception for CMs supplemented DPSCs, which did not differ significantly from the same population cultured in αMEM 10% FBS. No differences were patent in UC-MSCS and DPSCs supplemented with either CM (0.0001�P<0.001 to P<0.0001).
At the final assessment of 24 hours, UC-MSCs in CM supplemented media presented increased migration into the area than the un-supplemented Control and αMEM 10% FBS groups. They also presented increased motility than the DPSCs populations (0.0001�P<0.001 to P<0.0001). No difference in response to the two CMs was observed to within each MSCs type, but UC-MSCs responded with superior migration when paired to the DPSCs (0.0001�P<0.001 to P<0.0001).

In vitro endothelial tube formation assay.
The in vitro angiogenic potential of the CMs was evaluated through the tube formation capacity of UVECs, cultured in Complete, Control, UC-MSCs and DPSCs CM supplemented media. Control group presented the least effective tube formation capacity, while CM supplemented groups provided enhanced in vitro angiogenic potential (Fig 8).
It was noticeable that UVECs exposed to UC-MSCs and DPSCs CM supplemented media formed consistently well-defined tubular networks, with evenly distributed and regularly shaped loops in the Matrigel 1 matrix. Complete medium supplementation resulted in defined tubes, although loop distribution appears less homogeneous than in the CMs groups. Occasional events of incomplete tube connection from the branching points were observed. In the Control group, the dominant observation is the incomplete bridging attempts and short unconnected tubes, resulting in large and hill shaped loops.
Regarding the number of branching points, no differences were encountered between groups. In terms of total loops (formed from the connection of the branched tubes) and total tube length, Complete medium and CM supplemented groups showed significative superior performance, when compared to Control medium. Although both CMs provided increased results over the Complete medium group, statistical significance was only noted for the DPSCs CM supplemented group. No significant difference was found between the two CMs. The surface area covered by UVECs reflected the tube formation efficiency observed (Fig 8 and  Table 10).

3.3.4.
In vivo vascularization assay. After 7 days of subcutaneous implantation, the Matrigel 1 plugs were evidenced and collected. Discrete vascular penetration was observed macroscopically in all groups.
Microscopically, the Complete Medium and the UC-MSCs CM groups presented apparent increased capillary density than the Control group (Fig 9). In the CM groups, the capillary penetration reached deeper into the matrix. These observations were further confirmed through VEGFR2 staining of the microvessels within the subcutaneous tissue at the penetrating interface with the Matrigel 1 plug (Fig 9 and Table 11).

Discussion and conclusions
The disclosure of the composition of MSCs populations' secretome is key to the comprehension of the underlying mechanisms of their therapeutic action. Therefore, it is key to understand the basal profile of secretion of these MSCs populations, to grant basis for the selection of the MSCs systems most suitable for each intended therapeutic application. For such, we conducted an analytical study focusing on the investigation of the metabolomic and bioactive factors composition of the secretome of hMSCs originated from two of the most promising sources for medical applications: the UC-MSCs and the DPSCs. These sources may come to gain ground for MSCs-based therapies due to the non-/ minimally invasive and ethically accepted collection procedures, as well as for the increasingly available private and public banking options worldwide. The international scientific community has long debated the criteria for MSCs character assignment and issued a conciliatory guiding line of features for the definition of populations classified as MSCs [2], considering their phenotypic and differentiation capacities, which was confirmed on the hMSCs herein explored. We observed an increased capacity for DPSCs to differentiate towards mineralising populations, as reported by [49], but did not confirm their decreased capacity for adipogenesis since both UC-MSCs and DPSCs displayed comparable Oil Red O uptake. Further, a panel of specific genes has been reported to correlate with the proliferative capacity of undifferentiated MSCs and pluripotency/ multilineage differentiation capacity. Both UC-MSCs and DPSCs presented comparable expression of ALP, c-kit and Nanog. ALP has been described to relate to the proliferative stage of undifferentiated stem cells of embryonic and mesenchymal origin, while Nanog expression in MSCs is associated with the transition from in vivo quiescence to adaptation to in vitro growth condition. Both genes are described to downregulate as lineage commitment occurs, except for the osteoblastic phenotype conversion, where it is characteristically upregulated [50, 51]. C-kit is the gene coding for the receptor for the stem cell factor, also observed highly proliferative populations and related to differential lineage differentiation capacity [52]. Oct-4 was expressed to a lower extent in both cellular populations. Oct-4, along with Nanog, is associated to MSCs populations plasticity [53]. Other authors report that Oct-4 was not identified in cultured human adult MSCs [51].
These populations were therefore expanded in culture and elicited to secrete bioactive molecules to the surrounding media, through the conditioning process. The conditioning process was handled under serum-free conditions, so that the only factors detected in the CM were those produced by the UC-MSCs and DPSCs and not those bared to the system by exogenous supplementation.
Addressing the obtained CM from a metabolomic perspective (through 1 H-NMR spectroscopy), we identified glucose as the dominant metabolite in the spectra. Nuschke et al. recently investigated the metabolism of glucose in MSCs populations (bone marrow) and determined this to be a major limiting factor for MSCs survival. They describe a steady consumption rate for glucose in semi-and confluent cultures [54], a tendency that we could not observe in the cultured UC-MSCs and DPSCs. Indeed, UC-MSCs present an increase in both α-and β-glucose content, suggesting a possible engagement to gluconeogenesis and the utilization of other Table 9. Migration assay of hMSCs. At 0, 6 and 24 hours of expansion in control or CMs supplemented media, wounded area in 10 5 pixels. Values are presented as Mean ± SD. bioenergetic substrates for energy production. The observed profiles suggest that, although significant amounts of glucose are provided by plain media, conditioning cells do not solely rely on glycolytic pathways for ATP and NADP production. Pyruvate is therefore suggested as one of the crucial energetic metabolites that enters a variety of metabolic pathways, for the aerobic production of ATP [through its conversion into acetate/acetylCoA and entrance into the citric acid cycle]. Pyruvate is detected at relatively low concentrations, with a small tendency for consumption as conditioning time progresses. Pyruvate is initially provided by the media, but the maintenance of its concentration in the media suggest intrinsic production by the MSCs, through non-glucose dependent alternative synthesis pathways. Acetate levels, an intermediate product for the aerobic pathway of pyruvate catabolism, also remain fairly constant, as it is not expected to accumulate, but to continue its path through the citric acid cycle towards energy production. The build-up of choline and formate (related to the production of acetate) converge to this assumption.

Area (10 5 pixels)
Evidence of anaerobic metabolism is also observed throughout the conditioning period, providing accumulation of ethanol and lactate, both of which derived mainly from fermentation processes of the same energetic precursor pyruvate.
The observed accumulation of metabolites related to both bioenergetic processes supports the combination of both aerobic and anaerobic pathways for energy production by conditioning MSCs populations.
Other substrates dynamics are worth highlighting. GlutaMAX TM is a formulation whose hydrolysis results in the release of alanine and, ultimately, of L-glutamine, which is an essential nutrient in cell cultures for energy production as well as protein and nucleic acid synthesis [55]. This profile of consumption of GlutaMAX TM and build-up of its by-products is observable in both cellular populations. Little consumption is detected at 24 hours, but the process becomes very significant after 48 hours in culture. Additionally, UC-MSCs seem to entail on this metabolic pathway faster that DPSCs. The metabolic dynamics observed supports the survival and biosynthetic pathway endured by MSCs that, amongst other compounds, results in the production and/or release of bioactive molecules into their surroundings.
Through the evaluation of the bioactive factors content of the UC-MSCs and DPSCs secretome, the paracrine function of MSCs in therapeutic setups is substantiated. This topic has been granted greater attention in research when compared to the metabolomic composition of the MSCs secretome, and various cell sources have been researched besides the bone marrow standard [ We identified a series of pro-proliferative/ anti-apoptotic (TGFβs, FGF-2, G-CSF, HGF and VEGFs) and chemotactic factors (RANTES, GRO, MCP and MDC), as well as relevant cytokines (IL-6 and IL-8). Some other factors are described in the literature, which we were not able to detect in significant amounts, such as GM-CSF [65]. PDGF is another angiogenic factor Table 10. Tube formation assay. After 12 hours of UVECs exposure to Complete, Control, UC-MSCs or DPSCs supplemented media: total branching points, total loops, covered area and total tube length observed in each group. Values are presented as Mean ± SD. previously reported to be produced by DPSCs [70] that we were unable to detect in the conditioned media. Exploring into the specific functions of the identified factors, the VEGF family members are primary factors in the MSCs pro-angiogenic character [76,77]. Interestingly, most authors do not take into account the existence of the different isoforms of VEGF, and rather quantify its presence in bulk. We herein demonstrate that the differential detection of VEGF isoforms is of great importance, since the assayed cellular sources demonstrated clear differences in the isoforms profile content. Human DPSCs were here demonstrated to predominantly secrete VEGF-A (that connects to endogenous re-vascularization response [78]), while UC-MSCs were more prone to VEGF-C secretion (prone to neurogenesis induction, without exerting angiogenic effects [79]), which may impact on their potential for tissue regeneration therapies. Janebodin and colleagues [62] also report source dependent differential isoform secretion, and demonstrated that DPSCs secreted superior levels of VEGF-A than BM-MSCs. They also report the secretion of VEGF-D, and to significantly greater extents than BM-MSCs, while we were unable to detect significant VEGF-D signal, especially after 48 hours of secretion. VEGFs (and Ang) directly stimulate endothelial cells populations to proliferate and migrate, and to organise in tube like vascular networks of increased stability and maturation [56]. Other mitogens are described for endothelial populations [80], such as FGF-2 and PLGF, which we failed to observe in relevant amounts [81], in line with the recent reports for UC-MSCs [64] and for DPSCs [63,69]. HGF, the dominant bioactive factor identified in the UC-MSCs secretome, is recognised as anti-apoptotic and pro-mitogenic in various tissues [82], and was previously identified in UC-MSCs [77] and adipose derived-MSCs [65]. The strong expression of TGF-βs (particularly TGF-β1) by UC-MSCs could raise flags to their potential effect on the stimulation of fibroblastic population at a lesion site, but the combined secretion with antagonistic molecules, such as HGF and Follistatin may be the key to the antifibrotic character of MSCs populations [83]. IL-8 similarly displays pro-angiogenic features [65], promoting endothelial proliferation, migration and in vitro tube network formation [73].

Tube Formation Assay
These observations lead to the conclusion that both cell sources successfully supplied endothelial populations with stimulatory factors for their in vitro proliferation, migration and tube formation capacity, and forestalled senescence and apoptosis factors deprived endothelial cells. Indeed, the CM efficiently outweighed the complete endothelial cell expansion media utilised as positive control.
IL-8 and G-CSF are potent chemoattractant molecules [84], with recognised role in the homing of endogenous and delivered MSCs to lesion sites [85]. We observed great amounts of MCP-1 secretion by UC-MSCs, but not from DPSCs, colliding with Bronckaers' observations [69]. Potapova [58] reports BM-MSCs to also secrete MCP-1, in inferior proportion than IL-8 and IL-6. We observed an inverse pattern, since UC-MSCs secreted much expressive concentrations of MCP-1 instead of the ILs. Other entities such as eotaxin, fractalkine, GRO, and MDC and RANTES have been demonstrated to be chemoattract to MSCs [26,86,87]. The UC-MSCs maintain increased secretion of the molecules and, similarly to what was observed in the endothelial populations, the action of the CMs in the migration of the MSCs populations did not display equivalent differences. Conditioned media from both UC-MSCs and DPSCs promoted overall increased migration of the MSCs, seldom comparable to the stimuli provided by standard expansion media. The UC-derived populations appear to be more responsive to the CM vehiculated stimuli. Additionally, the UC-MSCs also appear to respond to the bioactive factors produced by themselves during the migration challenge, as the un-supplemented controls display some degree of in-defect proliferation, exceeding that of DPSCs. The distinct response profile may relate to the expression of chemokine receptors by MSCs, that varies though species, culturing time and, suggestively, though tissue source for isolation [26]. Our report on the composition of the CMs is therefore consistent with the properties attributed to MSCs, suggesting that UC-MSCs provide a wider variety and greater concentration of relevant growth factors and cytokines. The proposed difference in the secretory potential of UC-MSCs and DPSCs was explored in vitro and in vivo focusing on their effects on endothelial populations, as well as on the MSCs themselves. The observed effects on the endothelial populations discarded the initial assumption that UC-MSCs presented increased potential than DPSCs, since both CM presented comparable stimulatory and protective effects on this population. Significant differences between the two media performances were seldom observed and, whenever present, indicated superior performance to DPSCs-CM supplemented groups. The variety of growth factors and observed effects emphasises on the complementarity of functions between the known and unknown factors that compose MSCs secretion cocktail [64]. This complementarity is also well patent in some works, where the abolishment of one factor only partially attenuate the effects displayed [59].
Amid the direct effects of the MSC's secretome on the angiogenic populations and, hence, on the revascularization of lesion sites, many of the bioactive factors herein identified exert additional effects on other cellular populations, which contribute to their its significantly broad therapeutic potential in multiple applications. The MSCs secretome has been described to positively interact with gastric mucosa epithelial cells [88], dermal fibroblasts [89,90] and keratinocytes [90], pulmonary alveolar and small airway epithelial cells [91], retinal cells [92], hepatocytes [93], central and peripheral neuronal cells [94,95], and osteoblasts [96], as well as on interacting inflammatory populations [96].
To the authors' knowledge, this is the first comparative approach to the umbilical cord and dental pulp derived stem/ stromal cells populations' metabolomic and bioactive secretion profiles. In summary, the present work provided insight on the metabolic profile of UC-MSCs and DPSCs in culture during the serum-free conditioning process through which MSCs conditioned media was obtained. Some differences were evidenced in the metabolite dynamics on the CM from MSCs derived from the umbilical cord stroma and from the dental pulp through the conditioning period (particularly on glucose metabolism), but similar global metabolic profiles are suggested. Limited literature is available to compare on the metabolic profiles of MSCs.
More prominent differences are highlighted for the bioactive factors content of these CMs, where FST, HGF, G-CSF, IL-8 and MCP-1 dominate in the UC-MSCs secretion, while VEGF-A and FST are the most prominently produced by DPSCs.
Finally, the distinct secretory cocktail did not result in significantly different effects on endothelial cell populations (proliferation, migration and tube formation capacity). The apparent decreased chemotactic factors content of DPSCs was additionally refuted by the comparable capacity to induce MSCs migration in vitro.
Supporting information S1