Obesity induced by high-fat diet is associated with critical changes in biological and molecular functions of mesenchymal stromal cells present in visceral adipose tissue

Gene ontology and pathway analyses show differences in the core activities of MSCs obtained from ND and HFD mice

We evaluated the overrepresented ontological terms belonging to the GO classes of biological process (BP) and molecular function (MF). This procedure identified hundreds of ontologies in cell lysates and secretomes obtained from both ND- and HFD-MSCs (Supplementary File 2). We then performed a Venn analysis to identify both the unique and the common ontologies among the cellular and secreted proteomes of MSC samples. The cellular protein analysis revealed 69 BP and 13 MF ontologies specific to ND-MSCs, as well as 30 BP and 4 MF ontologies specific to HFD-MSCs (Supplementary File 2). The evaluation of secreted proteins allowed us to identify 50 BP and 6 MF ontologies specific to ND-MSCs, as well as 40 BP and 11 MF ontologies specific to HFD-MSCs (Supplementary File 2).

Then, we organized the identified specific ontologies into classes of the cell’s core activities to determine the functional effects of obesity on MSCs. We identified eight core activities: cell cycle/cell division, DNA damage, RNA ana-/catabolism, vesicle trafficking, immune system regulation, platelet functions, response to stress, and signaling (Tables 1–4).

Reactome analysis identified 258 pathways in the proteomes of ND-MSCs and 256 in the proteomes of HFD-MSCs. The pathways in the corresponding secretomes were 230 and 264, respectively. We then performed Venn analysis to determine the cell type-specific pathways; we identified 19 pathways exclusively present in proteomes of ND-MSCs and 17 pathways exclusively present in proteomes of HFD-MSCs. The corresponding secretomes contained 8 and 32 exclusive pathways, respectively (Supplementary File 3). These specific pathways were also classified according to the core activities indicated above (Tables 1–4).

We then analyzed the canonical pathways with IPA software, which identified 292 pathways in the proteomes of ND-MSCs and 281 in the proteomes of HFD-MSCs. The numbers of canonical pathways in the corresponding secretomes were 230 and 264, respectively. Venn analysis determined the cell type-specific pathways: 23 pathways were exclusively present in the proteomes of ND-MSCs, and 12 were exclusively present in those of HFD-MSCs. The corresponding secretomes showed 20 and 9 exclusive pathways, respectively (Supplementary File 3). The pathways were classified according to the identified core activities (Tables 1–4).

MSC cellular protein content is profoundly modified by obesity

Both ND and HFD proteomes are enriched in the specific signaling that is related to the G₂/M phase and mitosis, such as mitotic assembly checkpoint and the Aurora kinase and PLK1 pathways for ND samples (Tables 1, 2). The DNA damage task in ND samples appears to be more focused on single-strand DNA repair, given the presence of specific pathways associated with transcription coupled-nucleotide excision repair (TC-NER). The TC-NER is the most versatile and important DNA repair mechanism that allows cells to cope with daily DNA damage events that may occur during a cell’s lifetime [17]. Ubiquitin and the ubiquitin-like proteins (UBL) SUMO and NEDD8 play a role in regulating the cellular response to DNA damage [18]. The ND cellular proteome is endowed with proteins involved in neddylation (NEDD8) pathways. In contrast, the HFD samples are endowed with proteins involved in DNA double-strand break repair—such as SUMO ligases, which add SUMO peptides to BRCA1, HERC1, RNF168, MDC1, and TP53BP1 [19]. This result may indicate that HFD-MSCs have to cope with stronger genotoxic injuries compared to ND-MSCs. Indeed, obesity—with its related increases in reactive oxygen species (ROS) and in pro-inflammatory factors—may promote DNA damage events.

Both ND and HFD proteomes are enriched in proteins involved in RNA metabolism; ND samples are biased toward anabolic processes, such as positive regulation of RNA biosynthetic processes and of transcription, while HFD samples contain more factors involved in RNA catabolism (RNA phosphodiester bond hydrolysis, endonucleolytic regulation of RNA stability). Notably, HFD proteomes are endowed with proteins involved in the processing of capped intronless pre-mRNAs (Table 2). The intronless genes (IGs) are a small proportion of all genes, and the majority of IGs encode for G-coupled receptors and for other proteins with signal transduction activity. Another big IG group encodes for histones and chromatin-remodeling proteins. In addition, some IGs encode for genes involved in inflammation (15 interferon encoding genes) [20]. Most of the proteins encoded by IGs may be part of the pathways and ontological groups identified in the HFD samples, such as those regulating inflammation and signaling (see below).

MSCs have great vesicle trafficking, since they release factors that are synthesized on the endoplasmic reticulum and go through the Golgi apparatus to the exocytotic route; moreover, they play a role in MHC I antigen processing [21]. In this context, it is not surprising that both ND and HFD proteomes are enriched in proteins that regulate Golgi organization, Golgi-to-plasma membrane transport, cargo recognition, and clathrin-mediated endocytosis (Tables 1, 2).

MSCs play a role in immune system regulation and hemostasis [5, 22]. Indeed, ND proteomes specifically contain proteins involved in platelet ligand-dependent activation and the ensuing degranulation with the release of platelet agonists, glycosidases, acid proteases, adhesive proteins, prothrombotic factors, and pro-inflammatory factors [6, 7]. ND proteomes are also endowed with proteins involved in T cell receptor signaling and in PI3K signaling in B lymphocytes. In contrast, HD proteomes do not contain proteins involved in platelet functions, and the proteins involved in immune regulation are primarily associated with pro-inflammatory pathways (e.g., Interleukin-1 family signaling) (Tables 1, 2) [23].

Cellular response to stress is different in ND and HFD samples. In the latter proteome, we identified ontologies and pathways generally associated with cellular response to organonitrogen compounds. This class of chemical compounds contains numerous diverse molecules that may have a role in either cell physiological functions or cell stress. ND proteomes, in contrast, are enriched in factors that can actively cope with endogenous and external reactive oxygen stress, such as pathways related to NADH dehydrogenase and glutathione redox activity [24].

In terms of the key signaling circuitries that govern the above-indicated activities, a comparison of Reactome and canonical pathway analyses allowed us to identify transduction signal pathways in cells under investigation. The ND-MSCs rely on signaling coming from TGF-β, VEGFR2, HMGB1, and Leptin pathways, while the HFD-MSCs depend on MET, WNT, and FGFR2 signal transduction (Figure 4A, 4B). The four paths of ND-MSCs rely on RAS/MAPK downstream signaling. The TGF-β path also requires Gαq/RHO/RHOCK signaling, while the others may work through PI3K/AKT signals. Finally, HMGB1 and leptins may also have a connection with the NF-κB path (Figure 4A).

The MET and FGFR2 pathways identified in HFD-MSCs act downstream via PI3K/AKT, while WNT appears to work through canonical β-catenin signaling (Figure 4B).

Indications of regulators that control expression of proteins present in HFD and ND samples

LC/MS-MS analysis of cell proteomes may not allow the identification of all proteins present in a cell lysate, since cellular proteins range widely and dynamically in their concentrations. This enables the signal from the most abundant proteins to mask the signal of less-abundant ones, such as transcription factors and other regulatory molecules. We used the IPA Causal Network Analysis (CNA) to overcome this methodological limit. The CNA allows us to infer upstream molecules that control the expression of the genes/proteins in a given dataset.

We focused our interest on transcription factors that can be involved in the regulation of the genes that encode the proteins present in our proteome samples. The Venn analysis identified 70 upstream regulators exclusively present in ND-MSCs and 33 exclusively present in HFD-MSCs (Supplementary File 3). Among these regulators, we identified 6 transcription factors (GLI1, NR1I2, NR1I3, SMAD4, KLF6, and ARNT2) that can control the expression of proteins present in ND-MSCs, and we identified 2 (RARA and NKX2-3) that appear to be involved in gene expression regulation in HFD-MSCs.

Obesity-affected secretome composition of MSCs

In healthy conditions, MSCs secrete factors that are related to the specific cellular functions (core activities) we identified in our analysis of the cellular proteome content. ND-MSCs released proteins involved in immune system regulation and platelet activities, such as leukocyte migration and chemotaxis, natural killer cell signaling, chemokine signaling, class I MHC mediated antigen processing/presentation, and hemostasis. ND-MSCs also secreted proteins involved in cell response to stress, such as enzymes with disulfide oxidoreductase activity (detoxification of ROS and response to endoplasmic reticulum stress) (Table 3 and Supplementary Files 2, 3). The secreted proteins of ND-MSCs are enriched in signaling pathways that express MSC physiological functions, such as PAK signaling, CXCR4 signaling, LXR/RXR activation, and aryl hydrocarbon receptor signaling. The P21-activated kinases (PAKs) are serine/threonine protein kinases involved in pathways that play a role in migration, proliferation, apoptosis, mitosis, and vesicle-mediated transport processes [25]. The CXCR4 surface chemokine receptor pathway is involved in the mobilization of MSCs [26]. The activation of the LXR/RXR pathway improves glucose tolerance through coordinated regulation of glucose metabolism in liver and adipose tissues [27]. The aryl hydrocarbon receptor signaling originally functions primarily as a sensor of toxic substances and as a regulator of enzymes involved in stress relief and detoxification [28].

Several changes were seen in the secretome composition of HFD-MSCs. We found factors involved in immune system regulation, such as positive regulation of lymphocytes activation, granzyme A signaling, and methionine degradation. The stress response factors released by HFD-MSCs were those that are involved in regulating heath shock response, chaperone-mediated autophagy, and superoxide radical degradation. The secretomes of HFD-MSCs contain factors involved in PDGF signaling that, in addition to their mitogenic activity, may regulate local inflammation and promote atherosclerosis [29]. Enrichment in MET signaling found in the proteomes of HFD-MSCs was also detected in their secretomes, and this underscore the key importance of such a network for their functions. Two signaling paths identified in the secretomes of HFD-MSCs were related to senescence and aging phenomena: the sirtuin signaling pathway, and the defective B4GALT7 pathway associated with progeroid syndromes [30, 31].

Quantitative changes in the protein expression of MSCs following HFD treatment further confirm that obesity promotes senescence and impairment in physiological functions

The analyses described thus far focused on identifying biological processes, molecular functions, and related pathways that were specific to either health or pathological conditions of the MSCs. These analyses were conducted considering only the presence or absence of each protein in a given dataset (as compared to reference mouse protein databases). Further insights were obtained by evaluating changes in the expression level of each protein, by means of quantitative proteome analysis. We compared the expression levels of cellular proteins of HFD-MSCs with those of ND-MSCs and considered only the differences in values that were highly significant (p < 0.001). We identified 61 upregulated and 74 downregulated proteins in the proteomes of HFD-MSCs and 35 upregulated and 31 downregulated proteins in their secretomes (Supplementary File 4 and Figure 5). We then evaluated how many of these upregulated proteins were related to senescence, using SeneQuest (https://senequest.net) software and literature data to discover proteins associated with senescence. We found that 37 of the 61 upregulated cellular proteins in HFD-MSCs were related to senescence and that 20 of the 35 upregulated secreted factors belong to pathways associated with senescence phenomena (Figure 6) [32, 33].

These results corroborate the finding that obesity promotes senescence of MSCs through production of SASP factors.

Animal work and diet-induced obesity formation

All experiments in this study were carried out in compliance with the approval of Erciyes University’s Animal Experiments Ethical Committee (14^th December 2016, 16/164). The animals were obtained from Erciyes University GENKÖK, the transgenic animal department. We fed five C57BL6 mice (3 weeks old) with a Normal Diet (ND) and five other C57BL6 mice with a High-Fat Diet (HFD) for 20 weeks. In the HFD, the pellets contained 60% fat (Custom Diet, Safe, Inserm, France), while the ND pellets contained 5.7% fat. After 20 weeks, when the average weight difference was at least 10 grams, the mice were euthanized.

Insulin (ITT) and glucose (GTT) tolerance tests

Insulin and glucose tolerance tests were performed before the mice were euthanized. For the GTT, the mice were food-deprived for 12 hours before the test; then, 0.2 g/mL glucose were injected intraperitoneally, and values were measured with Accu-Chek Performa nano (Roche, Switzerland) at 6 time-points (0, 15, 30, 60, 90, and 120 min post-injection). The same food deprivation strategy was used to perform the ITT, whereafter a 0.8 mU/μL insulin solution was injected intraperitoneally and the blood glucose level was measured at five time points (0, 15, 30, 60, and 90 min post-injection).

Isolation and characterization of visceral adipose MSCs

Animals were euthanized by cervical dislocation. The visceral adipose tissue was isolated and cut into small pieces with surgical blades. Then, tissue samples were incubated for 15 minutes at 37° C in a solution containing 2.5 mg/ml type II collagenase (Sigma-Aldrich, MO, USA). Samples were filtered on cell strainers (70 μm mesh), centrifuged, and washed three times with phosphate buffer saline (PBS). Cells were then plated onto 75-mm flasks with alpha-MEM containing 15% fetal bovine serum (FBS) and were incubated for 7 days to reach confluence (P0). The cells were then trypsinized and characterized by flow cytometry analysis, which was performed with BD FACSAriaIII and BD FACSDiva 8.0.1. software (BD Bioscience, CA, USA). We used CD90 as a positive marker and CD29 and CD45 as negative markers. After P0, cells were cultivated for another 15 days (Passage 3 - 4) for all the biological and proteomic assays.

Cell cycle analysis

For each analysis, 5x10⁴ cells were collected by trypsin treatment and then, after PBS washing, were fixed in 70% ethanol overnight at -20° C. The samples were then washed with PBS and subsequently dissolved in a hypotonic buffer containing propidium iodide (Millipore, MA, USA). Analyses were performed using Muse Cell Analyzer (Millipore, MA, USA) following the manufacturer’s instructions.

Senescence-associated β-galactosidase assay

Cells grown in 6 well plates were fixed using a 0.2% glutaraldehyde solution for 5 min at room temperature (RT). Then, the cells were washed with PBS and stained with 40mg/mL X-gal staining solution, as described in [59]. Blue-stained cells were counted from 3 different regions of each well, and the percentage of senescent cells was determined. In identifying senescent cells, we also considered other properties, such as cell size, multi-nuclei presence, and granularity.

Annexin V and dead cell assay

Apoptosis was detected using the Annexin V and Dead Cell kits (Millipore, MA, USA) on a Muse Cell Analyzer (Millipore, MA, USA) following the manufacturer’s instructions.

Cell proliferation assay

Cell proliferation was determined by MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. First, 3,000 cells were seeded in each well of 96 well plates. At 12, 36, 60, and 84 hours, we added 1 mg/mL of MTT reagent to the cell media. After 4 hours of incubation at 37° C, the media were removed, and 100 μL DMSO was added to solubilize the salt crystals. The absorbance values were then measured at 560/750 nm using Glomax MultiWP Reader (Promega, WI, USA).

Whole cell sample preparation for mass spectroscopy

Samples were prepared with the InStage Tip digestion method described by Kulak and colleagues [60]. We collected 1x10⁶ cells each from the NF and HFD groups. After PBS washing, cells were stored at -80° C until lysis step. Then, 100 μL Lysis Buffer (6M Guanidinium chloride, 40mM CAA, 10mM TCEP, 25mM Tris-HCl pH:8,5) was added onto cell pellets and vortexed; subsequently, lysates were boiled for 5 minutes and then incubated on ice for another 5 minutes. After the cold incubation step, samples were sonicated in an ice-filled ultrasonic water bath for 5 minutes. Samples were then centrifuged at 20,000 g for 15 minutes, and proteins containing supernatants were collected.

We mixed 20 μL of each supernatant with 280 ng Lys-C (Promega, WI, USA) containing a 40 μL dilution buffer (25 mM Tris-HCl pH 8.5, % 10 ACN), which we put into InStage tips previously prepared by using 3 SDB-RPS extraction disks (3M Emporem, MN, USA). Mixtures were incubated overnight at 37° C. Subsequently, 1,000 ng Trypsin-Gold (Promega, WI, USA) was added to the Stage Tips, mixed well, and incubated for 4 hours. Following the incubation step, a 140 μL loading buffer (1% TFA) was added to each tip and centrifuged at 2,000 g; then, the peptide-loaded disks were washed four times with a 100 μL washing buffer. Peptides were eluted from disks in three fractions, according to their hydrophobic properties, by using 60 μL of each of the three elution buffers: SDB-RPS1 (100 mM Ammonium formate, 35% ACN, 0.5% Formic Acid), SDB-RPS2 (100 mM Ammonium formate, 55% ACN, 0.5% Formic Acid), and Buffer X (80% ACN, 0.125% Ammonia). Samples were lyophilized with SpeedVac and stored at -20° C until the LC-MS/MS analysis.

Secretome sample preparation for mass spectroscopy

MSC cultures were incubated in serum-free media for 24 hours; then, 5 mL of culture medium (secretome) was collected from each culture dish without disturbing the attached cells. Culture debris was removed by centrifugation at 10,000 g for 10 minutes, and supernatants were used for the StartaClean beads protein pooling. Collected secretomes were incubated overnight with the beads; then, the beads were washed twice with TE Buffer (50 mM Tris 10 mM EDTA pH 7) and dried with a vacuum concentrator.

The dried beads were resuspended at 2% (w/v) in a RapiGest (Agilent, CA, USA) solution containing TEAB (Sigma, MO, USA). Then, TCEP (Sigma, MO, USA) was added to the solution at a final concentration of 20 mM. Samples were incubated at 60° C for 30 minutes and cooled on ice. IAA (Bio Rad, CA, USA) was added to sample solutions, and samples were incubated at RT for 15 min. Then, 200 ng Lys-C (Promega, WI, USA) was added to each sample and incubated for 4 hours at 37° C. After Lys-C incubation, 800 ng Trypsin-Gold (Promega, WI, USA) was added to each sample and incubated overnight. Samples were centrifuged at 10,000 g for 1 minute; peptides containing supernatants were collected and acidified with 1% TFA before being loaded into Stage Tips. These tips were prepared with C18 material: they were washed with buffer B (% 0.1 Acetic Acid, 80% ACN) and equilibrated with buffer A (% 0.1 Acetic Acid). Acidified samples were loaded onto Stage Tips, and peptide-bounded tips were washed twice with buffer A. Following the washing, buffer B was added to the tips, and samples were eluted into collecting tubes with a syringe. Samples were dried with a vacuum concentrator and stored at -20° C until LC/MS analysis.

LC-MS/MS analysis

LC-MS analysis was performed with AB Sciex Triple ToF 5600+ (AB SCIEX, CA, USA) integrated with LC-MS/MS Eksigent ekspert™ nanoLC 400 System (AB SCIEX, CA, USA). Peptides were separated using nanoACQUITY UPLC 1,8 μM HSS T3 C18 column (Thermo Fisher, MS, USA) in the trap-elute mode. In order to separate the peptides, 4–40% ACN gradient was used for 240 minutes. Data dependent acquisition (DDA) MS/MS analysis of separated peptides was performed after electrospray ionization. Raw data analysis—generated by instrument reporting—and multiple analytical data measurements in each sample were performed with Analyst® TF v.1.6 (AB SCIEX, CA, USA). The peptides and the ion-product of the MS and MS/MS data were evaluated with PeakView (AB SCIEX, CA, USA). Generated peak-lists were evaluated in consideration of the UniProtKB-based reference library of the Mus musculus species on our server with ProteinPilot 4.5 Beta (AB SCIEX, CA, USA).

Label-free quantification

Quantitative analysis of data was completed by using MaxQuant (1.16.14.0) (https://www.maxquant.org). MS mass tolerance determined as 0.5 Da, MS/MS mass tolerance was 20 ppm. Carbamidomethyl was determined as fixed modification, oxidation and acetylation (protein N-term) were determined as variable modifications. Quantitative changes between groups were analyzed by using Limma, which is an R/Bioconductor software package (https://bioconductor.org/packages/release/bioc/html/limma.html). Visualization of volcano plots was performed with R (R 4.0.2) a software of Limma.

Bioinformatic evaluation

The cellular proteins and secreted factors identified by LC/MS were analyzed with PANTHER (http://www.pantherdb.org), Reactome Knowledgebase (https://reactome.org), and Ingenuity Pathway Analysis (IPA) (http://www.ingenuity.com/products/ipa). The significance ratio was p < 0.01 for all bioinformatic evaluations.

With PANTHER, we performed a Gene Ontology (GO) analysis according to two ontological terms: biological processes and molecular functions. In this analysis, we used statistics overrepresentation to compare classifications of multiple clusters of lists to a reference list in order to statistically identify the over- or under-representation of PANTHER ontologies. We followed the developers’ instructions for running a PANTHER analysis [61].

For the Reactome analysis, the proteins identified by LC/MS were mapped to specific pathways. Over-representation and pathway-topology analyses were conducted. Over-representation analysis is based on statistical hypergeometric distribution; it evaluates whether certain specific Reactome pathways are enriched in the submitted data. This analysis produced a probability score, wherein false discovery rate (FDR) was corrected for using the Benjamani-Hochberg method. We followed the developers’ instructions for running a Reactome analysis [62, 63].

The identified proteins in the different experimental conditions were also imported into IPA to attribute them to canonical pathways. Fischer's exact test was used to calculate a p-value that would determine the probability that the association between genes in the dataset and the canonical pathway could be explained by chance alone.

Biostatistics

Biostatistical analyses were conducted using Graphpad Prism software. Two-way ANOVA was used for analyses of cell cycles, and unpaired t-tests were used for the data from other biological assays.