Adipose tissue, a subtype of connective tissue, encompasses a diverse array of cell types, including stem/progenitor cells, preadipocytes, mature adipocytes, fibroblasts, vascular endothelial cells, and immune cells (macrophages and leukocytes) (Merrick et al., 2019; Bäckdahl et al., 2021; Wang and Wang, 2023). Mature adipocytes accumulate lipid droplets, which function as the body’s primary energy reservoir. Throughout ontogenesis, adipose tissue exhibits a remarkable degree of plasticity, undergoing significant changes in metabolism, structure, phenotype, and secretome in response to various physiological stimuli (Sakers et al., 2022). In addition to its role as an energy reservoir through triglyceride accumulation, adipose tissue emerges as an important endocrine organ that regulates the overall body metabolism. Mature adipocytes produce a wide range of biologically active molecules referred to as adipokines (adiponectin, leptin, resistin, vaspin, etc.), growth and differentiation factors (FGF21, BMP4, etc.), and microRNAs (Guerre-Millo, 2004; Kershaw and Flier, 2004; Cao, 2014; Kurylowicz, 2021).

In normal metabolism, the physiological expansion of adipose tissue occurs through an increase in the number of adipocytes, a process known as hyperplasia. However, under chronic malnutrition leading to obesity, the increase in adipose tissue volume primarily occurs through hypertrophy, characterized by an enlargement of the preexisting adipocytes (Sakers et al., 2022). In cases of metabolic syndrome, obesity, and insulin resistance, alterations extend beyond mere changes in adipose tissue morphology. Deficiencies emerge in the renewal of adipose tissue cells, proliferation of stem/progenitor cells, differentiation and maturation of adipocytes. This also entails a disturbance in the hormonal and secretory functions of adipose tissue, resulting in changes in the spectrum of secreted adipokines, growth factors, and microRNAs (Kahn et al., 2019).

While adipose tissue may appear uniform at first glance, it is, in fact, remarkably heterogeneous in its cellular composition. Diverse identities of heterogeneous stem/progenitor cell subpopulations, characterized by the expression of specific markers and distinct differentiation trajectories, and committed cells have been discovered (Burl et al., 2018; Schwalie et al., 2018; Merrick et al., 2019; Khazaei et al., 2022). Additionally, multiple genetic lineage tracing and transcriptomics analyses revealed heterogeneity among mature adipocytes within the same depots, which, despite their similar size, exhibit distinct adipokine secretion profiles (Bäckdahl et al., 2021; Corvera, 2021).

Among adipokines produced by adipose tissue, adiponectin notably stands out due to its high concentration in blood, typically ranging from 0.01% to 0.05% and exceeding that of the other hormones and cytokines by 3–6 orders of magnitude (Maeda et al., 2020). Adiponectin plays systemic protective roles against diabetes and atherosclerosis by exerting insulin-sensitizing, anti-inflammatory, and anti-proliferative effects. Moreover, adiponectin enhances fatty acid oxidation in skeletal muscles, regulates blood sugar levels and reduces triglyceride accumulation (Mattu and Randeva, 2013; Khoramipour et al., 2021). Concentration of adiponectin is markedly reduced in pathological conditions, such as type 2 diabetes mellitus (T2DM), obesity, and cardiovascular diseases (Achari and Jain, 2017; Rubina et al., 2021). Adiponectin circulates in three distinct forms in the bloodstream, each possessing distinct biological properties: low molecular weight (LMW), trimer-dimer or middle molecular weight hexamer, and high molecular weight (HMW) form (Khoramipour et al., 2021). Numerous studies indicate that the HMW form of adiponectin exhibits the highest metabolic activity (Hara et al., 2006; Mattu and Randeva, 2013). Adiponectin has three major receptors, AdipoR1, AdipoR2, and T-cadherin (Khoramipour et al., 2021; Rubina et al., 2021). However, T-cadherin was shown to act as an exclusive receptor for hexametric and high-molecular-weight (HMW) adiponectin (Hug et al., 2004; Fukuda et al., 2017). It serves as a major binding partner for native adiponectin in serum, as demonstrated by specific interactions between native adiponectin and T-cadherin (Kita et al., 2019). Moreover, the same authors revealed that T-cadherin knockdown, rather than AdipoRs, significantly reduces the binding of native adiponectin to C2C12 myotubes, HEK293 and CHO cells (Kita et al., 2019). Additionally, single nucleotide polymorphism (SNPs) in CDH13 gene encoding T-cadherin strongly correlates with plasma adiponectin level and cardiovascular diseases in humans (Rubina et al., 2021).

While adiponectin is predominantly synthesized by adipose tissue in adult organisms, the expression and function of its key receptor, T-cadherin, in this tissue remain elusive. Several single-cell RNA sequencing analyses reported T-cadherin expression in mesenchymal stem cells (MSCs) from various sources, indicating that MSCs express T-cadherin at least at the mRNA level. Specifically, T-cadherin mRNA has been detected in MSC cell clusters positive for PDGFRα (Pdgfrα) and Meflin (Islr) alongside well-known mesenchymal markers, such as CD73, CD90, and CD105 (Holley et al., 2015; Nakamura et al., 2020; Ramirez et al., 2020).

One of the most intriguing aspects of T-cadherin biology is its role as a receptor for two distinct ligands: HMW adiponectin and low-density lipoproteins (LDL) (Tkachuk et al., 1998; Rubina et al., 2021). Elevated levels of LDL in blood have been strongly linked to the pathophysiology of metabolic disorders, cardiovascular diseases, type 2 diabetes (T2D), insulin resistance (IR), and systemic inflammation (Kannel, 1987; Bissonnette et al., 2023). When contemplating the functional significance of T-cadherin as a dual receptor, we proposed a concept placing T-cadherin as a master switch. This concept implied that the interaction of T-cadherin with LDL or HMW adiponectin might determine, whether the outcomes are protective or detrimental, depending on the concentrations of these ligands in bloodstream (Balatskaya et al., 2019; Sysoeva et al., 2023).

To explore the potential involvement of T-cadherin in adipogenic differentiation, we performed a series of experiments utilizing MSCs isolated from the adipose tissue of both control and T-cadherin deficient mice. We examined the role of T-cadherin in adipogenic differentiation and lipid accumulation as well as the effects of T-cadherin ligands (LDL and adiponectin) on adipogenic differentiation. This approach aimed to gain deeper insights into whether T-cadherin orchestrates a feedback loop on adipogenesis upon binding with its ligands and to shed light on the potential function of T-cadherin in the adipose tissue.

T-cadherin-deficient mice (Cdh13DExon3 mice) were generated as previously described by crossing Cdh13loxP/loxP mice, possessing 2 loxP sites flanking exon 3 of the Cdh13 gene, and mice constitutively expressing Cre-recombinase (Popov et al., 2023). Adult Cdh13DExon3 mice and wild-type (WT) C57BL/6J mice were maintained in standard polypropylene cages under controlled vivarium conditions (temperature: 20°C–24°C, humidity: 35%–65%, 12-h light/dark cycle) with ad libitum access to food and water. Animal care and handling adhered to the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes (ETS №123). All procedures were complied with Directive 2010/63/EU of the European Parliament and the Council of 22 September 2010 on the protection of animals used for scientific purposes. Animal manipulations received ethical approval from the local ethical committee in accordance with the in-house requirements of the Commission on Bioethics of Lomonosov Moscow State University (license number 3.4).

To test the effects of T-cadherin ligands on adipogenic differentiation, MSCs isolated from subcutaneous adipose tissue of WT and T-cadherin-deficient mice, were seeded into 24-well plates and cultured until a confluent monolayer. To exclude the effect of adiponectin and LDL present in serum, ligands were added into a serum-free media, both for control and adipogenic conditions. 1% BSA was supplemented in the culture media to maintain cell viability. Ligands were added in the following concentrations: 70 μg/mL of low-density lipoproteins (LDL) (Biorbyt, #orb1748137), or 25 μg/mL of high molecular weight adiponectin (HMW adipo) (Biovendor, #RD172023100-C), or 25 μm/mL of low molecular weight adiponectin (LMW adipo) (Abcam, #ab283926). These concentrations were previously validated in our laboratory (Balatskaya et al., 2019). Cells were cultured for 5 days. Next, live cells were stained with Nile red (Sigma Aldrich (19123-10MG) for 40 min in Hank’s buffer (Paneco) stabilized with 50 mM HEPES (Paneco) pH 7.2 at 37°C. All wells were imaged in Whole well mode using a Nikon microscope (Nikon Eclipse Ti equipped with a digital EMCCD camera Andor iXon 897 (Andor Technology, Belfast, United Kingdom), NIS-Elements 5.21.02) with red and green fluorescence registration. Cells at least in two wells per condition were analyzed in each experiment, and the experiment was repeated twice.

To generate the LV-LeGO-hTcad-IRES-eGFP plasmid, hT-cadherin gene was amplified from pCWP1-hTcad plasmid encoding human T-cadherin cDNA (a kind gift from the Laboratory of Cellular Engineering led by T.N. Vlasik, at the Federal State Budgetary Institution National Medical Research Center of Cardiology of the Ministry of Health of the Russian Federation) using the primer pair hTcad-dir (GTC​CTC​CGA​TTG​ACT​GAG​TCG​CCC​GGA​TCC​CCC​GGA​CAA​AAT​GCA​GC) and hTcad-rev (CGG​ATC​CCA​ATT​CGA​TAT​CAA​GCT​GGT​TCA​CAG​ACA​AGC​TAA​GCT​GAA​GAG​GC); IRES-eGFP sequence was amplified from pIRES2-EGFP-p53 plasmid (Addgene plasmid #49242; http://n2t.net/addgene:49242; RRID:Addgene_49242) using the primer pair IRES2-dir (ACC​AGC​TTG​ATA​TCG​AAT​TGG​GAT​CCG) and eGFP-rev (GCT​ATA​CGA​AGT​TAT​TAG​GTC​CCT​CGA​CGT​CTA​GAT​TAC​TTG​TAC​AGC​TCG​TCC​ATG​CG). The successfully amplified fragments were isolated utilizing Cleanup Mini kit (Evrogen, Russia) and assembled using NEBuilder® HiFi DNA Assembly cloning kit (NEB, United States) into BamHI-EcoRI digested LeGO-G2 vector (Addgene plasmid # 25917; https://n2t.net/addgene:25917; RRID:Addgene_25917). To generate control lentivirus (LV-LeGO-eGFP), we used LV-LeGo-G2 (https://www.addgene.org/25917/).

For lentivirus production, HEK 293 TN cells were utilized. Cells were seeded at a density of 5 × 104 cells/cm2 in 25 cm2 tissue culture flasks and cultured in DMEM (Paneco, Russia) supplemented with 10% FBS (Biowest, China), and 1% penicillin/streptomycin (Paneco, Russia) at 37°C and 5% CO2. After 20–22 h, cells were transfected using 40 kDa linear polyethyleneimine (Polysciences, United States) at a mass ratio of 1:3 (DNA:PEI), with the following plasmids: 4 µg of vector genome; 1.3 µg of VSV-G envelope-expressing plasmid pMD2. G (Addgene plasmid # 12259; http://n2t.net/addgene:12259; RRID:Addgene_12259); and 2.6 µg of psPAX2 (Addgene plasmid # 12260; http://n2t.net/addgene:12260; RRID:Addgene_12260). The media was replaced 18–22 h post-transfection. The supernatant was collected 48 h later, clarified using a 0.45 µm filter, and stored at −80°C.

For automated quantification of the ratio of differentiated adipocytes to total cells, we developed an algorithm using artificial intelligence and deep learning techniques from the NIS. ai module within the NIS Elements software package (NIS-Elements Imaging Software, 2024). In each experimental group, the cells were segmented based on the red channel of Nile Red staining using the Segment. ai neural network from the NIS. ai module of NIS Elements 5.42.02.

The Segment. ai model operates by taking an image as an input and generating a binary mask as an output, effectively segmenting the image and highlighting individual areas of interest. It allows segmentation of difficult-to-detect areas by learning on manually segmented dataset. This neural network was fine-tuned on a set of 20 images (1024 × 1024 pixels) with manually labeled masks using the Train Segment. ai function (utilizing 2,000 epochs and Dynamic Range Adaptation option).

The output of this neural network is a binary mask where each cell is segmented and isolated from its neighbors, enabling cell counting of and the identification of the cells with lipid droplets. The total number of cells was determined using the Object Count function from the General Analysis 3 module [refer to (NIS-Elements Imaging Software, 2024), GA3: an analysis pipeline with AI capabilities] NIS Elements 5.42.02 applied to the resulting mask.

To count cells with lipid droplets, droplets were segmented by thresholding the green channel of Nile Red staining, with the threshold value selected manually (consistent across all points within individual experiments). The “Having” function in the General Analysis 3 module was then employed between the masks of all cells and the mask of lipid droplets, resulting in a mask containing only cells with lipid droplets. The Object Count function from the General Analysis 3 module was again used to count these cells. Finally, the ratio was calculated by dividing the number of cells with lipid droplets by the total number of cells for each experimental group.

To further automatically quantify the results of the rescue experiment, NIS Elements AR software was used. Utilizing Oil Red O staining, DAPI staining (blue) and GFP (green) fluorescent images, the maximum intensity values of GFP and Oil Red O signals were computed for each cell. These values quantitatively reflected the proportion of virus-infected cells in the green channel (GFP fluorescence) and the accumulation of lipid droplets visualized through Oil Red O staining under transmitted light. The segmentation of each cell was carried out using the DAPI image (nuclei were detected by the Bright Spots function), followed by the cell segmentation using the Watershed function. GFP images were normalized to the highest signal intensity in the image using the Auto Contrast function, setting the most intense signal to a value of 255. To evaluate the intensity of the red signal in the Oil Red O-stained image, the formula proposed by Eggerschwiler et al. (2019) was applied: for each pixel, the value in the red channel was divided by the sum of the values in the green and blue channels; a result > 1 was considered positive. The peak intensity values of GFP and Oil Red O signals for each cell in the entire cell population were visualized using scatterplots, generated with Seaborn 0.12.2 library in Python 3.11.

Intracellular lipid quantification was conducted following the previously described methods with some modifications (Kraus et al., 2016). Cells were seeded into 24-well plates and induced for adipogenic differentiation as described above. For Oil Red staining, cells were fixed in 4% formaldehyde in PBS for 30–40 min at room temperature. Сells were carefully rinsed 3 times (5–10 min each) with phosphate buffer saline (PBS). Oil Red O Solution (Lonza, Switzerland) was added to cover the wells for 5 min at room temperature.

To quantitatively assess the Oil Red O content bound to neutral lipids within the cells, we conducted dye elution by adding 750 µL of 60% isopropanol to each well, followed by agitation for 20 min at room temperature. An additional 750 µL of 60% isopropanol was added to each well. The absorbance of the resulting eluates was measured using a Bio-Rad SmartSpec Plus spectrophotometer at a wavelength of 500 nm in Greiner Bio-One Semi-micro cuvettes (Greiner Bio-One, Germany). The obtained values were normalized to the protein content in a sample.

Protein concentration in protein extracts was determined using the Bio-Rad Protein Assay (Bio-Rad), based on the Bradford method for protein concentration evaluation. Protein content was measured on a SmartSpec Plus spectrophotometer (Bio-Rad), according to the manufacturer’s protocol, using bovine serum albumin (BSA) dilutions as standards (Bio-Rad). Protein extracts were obtained by adding 200 µL of Laemmli buffer to each well followed by a 15-min incubation.

Initially, RT-qPCR and Western blotting were employed to confirm the absence of full-length T-cadherin expression in T−/− MSCs (Figure 1).

Figure 1. T-cadherin in content in WT MSCs and T−/− MSCs was assed using quantitative RT-PCR (A) and Western blot (B) analysis. MSCs were cultured in standard media.

To address the role of T-cadherin in adipogenic differentiation, we first isolated MSCs (WT MSCs and T−/− MSCs) from subcutaneous adipose tissue of control and T-cadherin-deficient mice and cultured the cells in standard media. We found T−/− MSCs exhibited a tendency for spontaneous adipogenic differentiation at 1–3 passages when cultured in dense monolayers under standard conditions for 4–7 days without any adipogenic inducers. To verify this phenomenon, MSCs isolated from 5 T-cadherin-deficient and 5 WT male mice (aged 8 weeks) were cultured under identical conditions. In 2 out of 5 examined samples of T−/− MSCs we detected foci containing cells with lipid droplets among the cell monolayer. Using fluorescent dye Nile Red staining we further verified the lipid accumulation in these cells. Such foci were not observed in MSCs from WT mice (Figure 2).

Figure 2. Spontaneous adipogenic differentiation in MSCs derived from murine subcutaneous adipose tissue. Upon reaching the confluent monolayer, T−/− MSCs spontaneously formed cell clusters containing lipid droplets (marked by a whit oval) [(C) - Nile Red staining, green fluorescence; (D)–phase-contrast microsopic image]. No clusters of the kind were observed in MSCs derived from control mice (WT MCSs) (A) (Nile Red staining) and (B) (phase-contrast image). Scale bar 100 µm.

Next, to explore the role of T-cadherin in adipogenic differentiation, we compared the potential of MSCs isolated from WT and T-cadherin-deficient mice to accumulate lipids in standard media or upon induction of adipogenic differentiation. For quantitative evaluation of Oil Red O content, MSCs from WT and T-cadherin-deficient mice were seeded into 12-well plates and cultured until they formed a confluent monolayer. Subsequently, cells were either induced towards the adipogenic lineage by adding the adipogenic induction factors as per the Materials and Methods section, or they were maintained in standard media for 10 days. Next, the cells were subjected to dye elution by adding 60% isopropanol to each well. The absorbance of the resulting eluates was measured and normalized to the protein content in each sample (Figure 3).

Figure 3. Performance of quantitative oil red O staining with arbitrary parameters. MSCs derived from WT and T−/− mice were cultured for 10 days in either control media or media for inducing adipogenic differentiation. Subsequent dye elution was performed by adding 60% isopropanol to each well. The absorbance of the eluates was measured using a Bio-Rad SmartSpec Plus spectrophotometer at a wavelength of 500 nm. The obtained values were normalized to the protein content in each sample. Quantitative evaluation was performed using two-way analysis of variance (Two-way ANOVA) followed by a post hoc test. Statistical significance was considered at p < 0.001.

The results obtained revealed a statistically significant difference between the groups, indicating that both, the cell type (cells derived from WT or T-cadherin-deficient mice) and the media type (media for adipogenic induction or control media) independently influenced intracellular lipid accumulation. Specifically, there was a 1.5-fold increase in Oil Red O content in T−/− MSCs vs WT MSCs in adipogenic media (P < 0.001). Additionally, there was a 2.08-fold increase in Oil Red O content in T−/− MSCs in adipogenic media vs. T−/− MSCs in standard media (P < 0.001), a 2.19-fold increase in T−/− MSCs vs. WT MSCs, both in standard media (P < 0.05), and a 3-fold increase in Oil Red O content in WT MSCs in adipogenic media vs WT MSCs in standard media (P < 0.001).

Overall, MSCs derived from T-cadherin-deficient mice (T−/− MSCs) were more prone to spontaneous adipogenic differentiation and were also more susceptible to induction of adipogenic differentiation and lipid accumulation in the presence of adipogenic induction factors compared to WT MSCs.

However, the method used for quantitative evaluation of Oil Red O content does not allow for distinguishing whether there is a difference in the size of lipid droplets or in the number of cells containing them, which contribute to the overall increase in the Oil Red O staining. Specifically, individual cells may harbor large lipid droplets, or alternatively, a large number of cells may accumulate small lipid droplets.

To further explore the role of T-cadherin in shaping lipid droplets and influencing the number of cells accumulating them, MSCs from WT and T-cadherin-deficient mice were seeded in 24-well plates and cultured until a confluent monolayer in standard media. The cells were either induced for adipogenic differentiation or maintained in standard media, both for 10 days. We performed Oil Red O and Nile Red staining combined with fluorescent, light and phase-contrast microscopy to analyze the pattern of lipid droplet accumulation in relation to T-cadherin expression (Figure 4).

Figure 4. Adipogenic differentiation in MSCs derived from murine subcutaneous adipose tissue. (A) - WT MSCs cultured for 10 days in the standard media [(A1) - Nile Red, Scale bar 100 µm; (A2) – phase-contrast, Scale bar 100 µm; (A3) - phase-contrast and Nile Red merge, Scale bar 100 µm; (A4) - Oil Red O inset, Scale bar 50 µm] (lipid droplets marked by an arrow). (B) - T−/− MSCs cultured for 10 days in the standard media [(B1) - Nile Red, Scale bar 100 µm; (B2) – phase-contrast, Scale bar 100 µm; (B3) - phase-contrast and Nile Red merge, Scale bar 100 µm; (B4) - Oil Red O inset, Scale bar 50 µm] (large lipid droplets marked by arrows]. (C) - WT MSCs cultured for 10 days in adipogenic media [(C1) - Nile Red, Scale bar 100 µm; (C2) – phase-contrast, Scale bar 100 µm; (C3) - phase-contrast and Nile Red merge, Scale bar 100 µm; (C4) - Oil Red O inset, Scale bar 50 µm] (D) - T−/− MSCs cultured for 10 days in adipogenic media [(D1) - Nile Red, Scale bar 100 µm; (D2) – phase-contrast, Scale bar 100 µm; (D3) - phase-contrast and Nile Red merge, Scale bar 100 µm; (D4) - Oil Red O inset, Scale bar 50 µm]. (E) - Performance of quantitative Oil red O staining of T−/− MSCs and WT MSCs. The analysis was performed using 10 random fields of view. Statistical differences between the cells were assessed with Kruskal–Wallis test followed by multiple comparisons.

Among T−/− MSCs cultured in adipogenic conditions, we observed cells with large droplets (inset in Figure 4A (4) - Oil Red O staining). WT MSCs cultured in adipogenic conditions were more homogeneous and contained small lipid droplets (inset in Figure 4B (4) - Oil Red O staining).

Single cells containing large lipid droplets were observed among T−/− MSCs cultured in standard media (inset in Figure 4C (4), Oil Red O staining). In WT MSCs cultures maintained in standard media such cells with large lipid droplets were extremely rare and mainly contained small droplets (inset in Figure 4D (4), Oil Red O staining).

To quantify the number of cells containing lipid droplets, we analyzed images obtained after Oil Red O staining of T−/− MSCs and WT MSCs after 10 days of cell culture in adipogenic conditions or standard media. We randomly selected 10 fields of view per well and calculated the percentage of cells with lipid droplets, followed by statistical analyses.

After 10 days of adipogenic differentiation, the percentage of cells with lipid droplets in both T−/− MSCs and WT MSCs significantly increased compared to standard media (a 7.8-fold increase for WT MSCs and a 1.8-fold increase T−/− MSCs). Remarkably, the percentage of these cells in T−/− MSCs was 1.4-fold higher than in WT MSCs in adipogenic conditions (Figure 4E). Statistical significance between the groups was as follows: WT MSCs in standard media vs. all other groups (P < 0.0001); T−/− MSCs in adipogenic condition vs. WT MSCs in adipogenic condition (a 1.4-fold increase) (P < 0.01); T−/− MSCs in standard media vs. T−/− MSCs in adipogenic condition (a 1.8-fold increase) (P < 0.001). The calculations revealed a significantly higher percentage of the cells containing lipid droplets in T−/− MSCs compared to WT MSCs, even under standard media conditions (a 6.3-fold increase) (Figure 4E).

These experiments illustrate that MSCs isolated from the subcutaneous adipose tissue of T-cadherin knockout mice exhibit an increased inclination towards adipogenic differentiation compared to MSCs from control mice. Interestingly, even in the absence of adipogenic induction factors, T−/− MSCs demonstrated a propensity to differentiate into adipocytes characterized by large lipid droplets, and this trend persisted upon the addition of adipogenic inducers.

We observed a steady rise in cellular PPARγ content in WT MSCs under adipogenic differentiation conditions up to day 7, compared to WT MSCs maintained in standard media. A similar trend was noted in T−/− MSCs, although the initial PPARγ levels in these cells were slightly lower (Figure 5A). Meanwhile, we detected a notable difference in C/EBPβ protein levels between WT MSCs and T−/− MSCs: in WT MSCs C/EBPβ protein content was elevated in cells cultured in adipogenic conditions compared to WT MSCs on days 5, 7, and 10. In contrast, the increase in C/EBPβ protein in T−/− MSCs was detected only on day 5 (Figure 5A). Given that C/EBPβ is typically upregulated during the early stages of adipogenesis, we hypothesized that T−/− MSCs progress through adipogenic differentiation at a faster rate than WT MSCs. This hypothesis was further supported by the earlier emergence of adiponectin in T−/− MSCs during adipogenic differentiation. Cellular adiponectin was detected only on the 7th day of adipogenic induction in WT MSCs, while in T−/− MSCs adiponectin was detected starting from the 5th day and was growing steadily from that moment on (Figure 5B). The Western blot analysis of leptin clearly demonstrated a consistent increase in its cellular accumulation T−/− MSCs through day 7, while in WT MSCs the results were less convincing (Figure 5C). The decrease in leptin content on the 10th day may reflect the process of its intensive secretion. Western blot analysis of leptin revealed a consistent increase in its cellular accumulation in T−/− MSCs up to day 7, whereas in WT MSCs, the results were less pronounced (Figure 5C). The observed reduction in leptin levels on day 10 in both cell types may reflect its intensified secretion processes during adipogenic differentiation. Similar dynamics were observed in perilipin protein expression: in WT MSCs, perilipin was detected on day 7 and increased further on day 10. In contrast, T−/− MSCs showed perilipin expression as early as day 5, with a dramatic rise on days 7 and 10, ultimately reaching levels that exceeded the peak expression in WT MSCs (Figure 5D).

Figure 5. Western blot assessment of early and late adipocyte differentiation markers in lysates of WT MSCs and T−/− MSCs cultured in adipogenic (a) or standard (c) media for 10 days Commercially available antibodies were used to evaluate the protein content of PPARγ and C/EBPβ (A), adiponectin (B), leptin (C), and perilipin (D). For loading control, anti-GAPDH antibody was used. (0 – start of the experiment; 5a, 7a, 10a–days following adipogenic differentiation induction; 5c, 7c, 10c–days in the experiment where cells were maintained in media without adipogenic inducers (control).

Given that the functional maturity of adipocytes depends on their capacity to produce and release specific adipose tissue hormones, we further analyzed the secretion of adiponectin and leptin into the culture media by MSCs. Culture media from WT MSCs and T−/− MSCs maintained in either standard or adipogenic media was collected over a 15-day period, with samples taken on days 3, 6, 9, 12, and 15. Comparative analysis of adiponectin concentration into culture media from the cells maintained in adipogenic conditions revealed that adiponectin secretion from T−/− MSCs started as early as day 3, whereas adiponectin secretion from WT MSCs could be detected only from day 6 (Figure 6A). A statistically significant difference in adiponectin secretion was observed throughout the whole experiment until day15 being much lower in WT MSCs compared to T−/− MSCs. A similar pattern was noted for leptin secretion: leptin was detected in T−/− MSCs culture media starting from day 3, whereas in WT MSCs it became detectable only from day 12. Overall, there was a statistically significant difference in leptin secretion throughout the whole experiment. Neither adiponectin nor leptin were detected in the control media (Figure 6B).

Figure 6. Elisa evaluation of adiponectin (A), leptin (B), and perilipin (C) in WT MSCs and T−/− MSCs cultured in adipogenic or standard media for 15 days. For adiponectin and leptin evaluation, culture media from WT MSCs and T−/− MSCs maintained in either standard or adipogenic media was collected over a 15-day period, with samples taken on days 3, 6, 9, 12, and 15. Perilipin 1 content was assessed in cell lysates obtained on the 15th day.

Perilipin is a lipid droplet surface protein required for lipid storage and fatty acid release (Tansey et al., 2004). Therefore, we assessed perilipin intracellular content in lysates of WT MSCs and T−/− MSCs cultured in standard and adipogenic media at the end of the experiment (day 15). Perilipin was detected in lysates from both T−/− MSCs and WT MSCs cultured in control and adipogenic media. While perilipin levels were significantly elevated in adipogenic media compared to standard media in both cell types, no difference was found between T−/− MSCs and WT MSCs (Figure 6C). These results indicate a notably elevated expression of late adipogenic markers at the protein level in T−/− MSCs compared to WT MSCs.

Overall, these findings suggest that T−/− MSCs may be predisposed to adipogenic lineage commitment, as indicated by the elevated levels of adiponectin, leptin, and perilipin in both types of cell lysates and secreted fractions.

At the first stage, fixed with 10% formalin and stained with DAPI, cells were imaged to capture phase-contrast, green, and DAPI fluorescent images. At the second stage, the cells were stained with Oil Red O and re-imaged to visualize red lipid droplets under transmitted light. The NIS. ai module in NIS Elements 5.42.02 facilitated identification of the same cells imaged during the first and the second stages. Therefore, using DAPI and phase-contrast images, the neural network first recognized previously imaged cells from the first stage and identified those expressing GFP, marking viral infection. The network then overlaid these images with those obtained at the second stage, obtained after Oil Red O staining, to visualize the distribution of cells containing both lipid droplets and GFP expression (Figure 7).

Figure 7. (A–C) - T−/− MSCs transduced with a control lentivirus construct (LV-LeGO-eGFP) and cultured for 14 days in adipogenic media. (A) - Oil Red O with phase-contrast; (B) – GFP fluorescence, Scale bar 100 µm. Cells expressing GFP are marked by arrows. (C) – Scatterplot illustrating the dependency of MSC differentiation efficiency on GFP fluorescence intensity. - (D–F) - T−/− MSCs transduced with a lentivirus construct for T-cadherin re-expression (LV-LeGO-hTcad-IRES-eGFP) and cultured for 14 days in adipogenic media. (D) - Oil Red O with phase-contrast; (E) – GFP fluorescence, Scale bar 100 µm. Cells expressing GFP are marked by arrows. (F) – Scatterplot illustrating the dependency of MSC differentiation efficiency on GFP fluorescence intensity. Each dot on the plot represents a single cell. The X-axis represents GFP fluorescence intensity. The Y-axis represents the maximum Oil Red O (ORO) signal, calculated using the formula (red channel value)/(blue channel value + green channel value). The red line on the Y-axis indicates the threshold for positive values; points above this line represents cells with lipid droplets. Green square comprises cells with a GFP signal intensity above 127 expressing the transgenes and stained for Oil Red O.

For the scatterplots, we set a threshold for Oil Red O signal intensity (red line), since the cells were stained with the same reagents, allowing for a direct comparison of Oil Red O intensity. However, GFP signal intensity differed significantly between the cells transduced with LV-LeGO-hTcad-IRES-eGFP and LV-LeGO-eGFP lentiviruses, primarily due to the larger size of LV-LeGO-hTcad-IRES-eGFP construct compared to LV-LeGO-eGFP. Although this precluded the direct comparison based on GFP intensity, it still allowed for identification of transduced cells. GFP signal intensity for each cell type, transduced with either LV-LeGO-hTcad-IRES-eGFP or LV-LeGO-eGFP lentivirus, was represented within a range of 0–255. Cells on each scatterplot with a GFP signal intensity above 127 were deemed to express the transgenes, and double-positive cells (GFP and Oil Red O) were localized within the green square. In T−/− MSCs transduced with control LV-LeGO-eGFP (Figures 7A–C), we detected a significant number of double-positive cells (16 cells) in the green square (Figure 7C), indicating their adipogenic differentiation. In contrast, for T−/− MSCs transduced with LV-LeGO-hTcad-IRES-eGFP (Figures 7D–F), only one cell was detected within the green square (Figure 7F), indicating that re-expression of T-cadherin drastically reduced the cells’ ability to accumulate lipid droplets. This rescue experiment thus confirmed T-cadherin role in lipid droplet accumulation, indicating that T-cadherin re-expression suppressed lipid droplet formation.