IFN-γ synergizes with TNF-α to induce distinct morphological changes of MSCs
To investigate the effects of IFN-γ and/or TNF-α on MSCs, we firstly assessed the morphological features of MSCs when exposed to vehicle, IFN-γ and TNF-α alone or in combination under different concentrations, as previous studies showed that morphology can predict MSCs mineralization[27] and immunosuppressive capacity.[11] After 24 hours of treatment, we observed that MSCs synergistically treated with IFN-γ and TNF-α (MSCs-IT) showed a gradual change from a fibroblastic-like spindle shape to an enlarged widespread morphology, and the cells grown in a disordered way, which was different from the other groups (Figure 1a). In addition, the cells looked like a flat spider web and the blurred cell membrane hinted that these cells may have “ruptured” per high-power field (Figure 1b, 8d). Except for disordered growth direction, enlarged cell size and stretched cell shape, the immunofluorescence staining showed altered polarization and disruption of actin cytoskeletal network in these cells. Yet notably, the nuclei were intact (Figure 1c). All of these alterations in morphology indicated that IFN-γ and TNF-α have a synergistic effect on MSCs.
Biological functions of MSCs are impaired when exposed to IFN-γ and TNF-α in combination
After cytokine stimulation for 72-96 hours at the end of morphology observation, the MSCs were harvested and re-seeded onto 6-well culture plates. Unexpectedly, the adherent cells of MSCs-IT decreased and many floating cells could be seen (Figure 1d), indicating that those cells were functionally impaired.
Firstly, the reduced proliferation potential and impaired migration capacity of MSCs-IT were evidenced through CCK-8 (Figure 1e) and wound-healing assays (Figure 1f, g). Inspired by a previous study, in which IFN-γ and TNF-α synergistically induced apoptosis of mouse MSCs,[21] we next tested the apoptotic rate of MSCs. Nevertheless, MSCs-IT appeared not to be an apoptotic phenotype in this scenario. (Figure 2a, b). Furthermore, the flat and enlarged morphology of MSCs-IT prompt us to detect its senescence-associated β-galactosidase activity. The results showed that MSCs-IT had no higher proportion of cells with senescent property (Figure 2c, d).
The results above demonstrated that MSCs-IT were functionally impaired in regard to their capacities of proliferation and migration, which, however, was not associated with increased cell apoptosis or senescence.
IFN-γ and TNF-α synergistically decreased the tri-lineage differentiation capacities of MSCs
Given that the growth and proliferation of MSCs-IT were impaired, we hypothesized that the differentiation capacity may also be affected. In vitro differentiation assays of MSCs into adipocytes, osteocytes and chondrocytes were performed after priming, and the final results indicated that the tri-lineage differentiation capacities of MSCs-IT were entirely impaired.
Firstly, we assessed their ability to differentiate into adipogenic cells. As shown by oil red O staining in Figure 3a, b, both IFN-γ and TNF-α solely inhibited the adipogenic differentiation ability of MSCs in contrast with the vehicle, and further the combination of the cytokines inhibited that process more profoundly.
Secondly, we assessed their osteogenic differentiation ability and a similar result was obtained that the MSCs-IT group exhibited the weakest capability to differentiate into osteocytes with IFN-γ or TNF-α alone inhibited that process too yet slightly, which was demonstrated by alizarin red S staining (Figure 3c, d).
Finally, in the chondrogenic induction assay, the results showed that IFN-γ and TNF-α synergistically led to the greatest inhibition on chondrocyte differentiation, as alcian blue staining suggested (Figure 3e, f). Noteworthy, despite the dampening effect of IFN-γ on this process, TNF-α itself promoted, instead of inhibited, the chondrogenic differentiation of MSCs (Figure 3e, f). From this point of view, the action of IFN-γ and TNF-α on MSCs indeed was synergized rather than a cumulative effect.
Immunosuppressive capacities of MSCs are weakened by synergistic effect of IFN-γ and TNF-α
As IFN-γ and TNF-α had synergistically showed some impairing effects on MSCs in previous experiments, we assessed the immunosuppression effects of IFN-γ and/or TNF-α treated MSCs on CD3+ T cells’ activity by flow cytometry-based measurements of proliferation (CFSE dilution), CD25 expression, and production of IFN-γ and TNF-α.
Firstly, the proliferation index of activated T cells uniformly stained by CFSE was assessed in direct interaction with irradiated MSCs in culture. As initially shown in Figure S1a, the proliferated T cell masses could be directly observed under microscope. Compared with vehicles, in which T cells were suppressed profoundly, MSCs primed with sole TNF-α (MSCs-T) seemed to have intact suppression ability whereas both MSCs primed with sole IFN-γ (MSCs-I) and MSCs-IT showed an impaired phenotype in regard to visible T cell masses, with MSCs-IT had the lowest observed suppression on T-cell proliferation masses. It was further evidenced by flow cytometry that the proliferation index of both CD4+ and CD8+ cells co-cultured with MSCs-IT were significantly increased (p < 0×001 for both) (Figure S1a and Figure 4a, b). It should be noted that the suppression ability of MSCs-I was attenuated too (p < 0×001 for both), although not significant as MSCs-IT, which was inconsistent with previous studies[11, 12] showed that IFN-γ stimulation could enhance the immunosuppressive properties of MSCs, and a longer prime time in the present study may account for it. In brief, these results suggested that IFN-γ and TNF-α could synergistically impair the immunosuppressive activity of MSCs on T cell proliferation.
Secondly, we measured the frequency of T cells with up-regulated growth factor receptor CD25 to assess the activation extent of T cells after co-culture. Under these conditions, naïve MSCs significantly decreased both the proportions of CD4+CD25+ and CD8+CD25+ cells (p < 0×001 for both) (Figure S1b and Figure 4c, d). In contrast, the highest frequency of CD4+CD25+ cells in MSCs-IT group was observed, which was significantly higher than that of vehicles (p < 0×01) (Figure S1b and Figure 4c). The suppression ability of MSCs-I on CD4+CD25+ cells was also slightly impaired (p < 0×05) whereas MSCs-T maintained almost intact (Figure S1b and Figure 4c). With regard to CD8+CD25+ cells, MSCs-IT again showed an attenuated suppression capacity in comparison with vehicles (p < 0×01) (Figure S1b and Figure 4d). However, MSCs-I appeared a similar impairment as MSCs-IT in this aspect (p < 0×001) (Figure S1b and Figure 4d). Taken together, these data suggested that IFN-γ and TNF-α had also synergistically damped MSCs in regard to the regulation of CD25 expression on T cells.
Finally, the percentages of IFN-γ- and TNF-α-producing cells were evaluated. Untreated MSCs had dramatically decreased the percentages of both IFN-γ- and TNF-α-producing CD4+ T cells and TNF-α-producing CD8+ T cells when compared with monocultures of purified CD3+ T cells (p < 0×01) (Figure 3f-h). As expected, MSCs-IT had the lowest power of decreasing IFN-γ-producing CD4+ and CD8+ T cells among other groups (p < 0×05) (Figure 3e, g); yet no significant influence on the production of TNF-α either in CD4+ or CD8+ populations (Figure 3f, h). Notably, MSCs-T decreased all cytokine-releasing cells more profoundly in contrast with vehicles (p < 0×05), indicating an enhanced immunosuppressive capacity (Figure 3e-h). Collectively, IFN-γ and TNF-α synergistically attenuated the inhibitory capacity of MSCs on T cells in terms of its IFN-γ production.
In this part, although the extent of injured immunosuppression of MSCs-IT varied, IFN-γ and TNF-α still synergistically impaired the immunosuppressive capacities of MSCs.
Impaired MSCs-IT have normal function of producing and releasing IDO
It is well known that human MSCs-mediated T cell immunosuppression involves cell contact-dependent mechanisms and soluble factors such as indoleamine 2,3-dioxygenase (IDO).[28, 29] To figure out the potential mechanism of MSCs-IT impairment in our cell contact co-culture system, we measured both secretory and intracellular IDO of MSCs. Flow cytometric analysis of IDO-expressing intensity in MSCs through mean fluorescence intensity (MFI) was conducted after intracellular immunofluorescent staining of IDO protein. As shown in Figure 5a, b, MSCs-IT exerts the highest production level of IDO. Further, the tryptophan metabolic activity of secretory IDO was determined and the results showed that the IDO secretory capacity of MSCs-IT was still intact and the activity was even greater than MSCs-I (Figure 5c) during the 3 days of incubation.
Conclusively, the immunosuppressive deficiency of MSCs-IT was not mediated through IDO insufficiency or exhaustion.
The signaling of IFN-γ might be enhanced by TNF-α through up-regulation of IFN-γR1
From the above results, we confirmed that IFN-γ and TNF-α definitely have a synergistic effect on human MSCs, but the underlying mechanism is yet to be elucidated.
Initially, we wanted to know how this synergistic role was played. As both IFN-γ and TNF-α are soluble cytokines, we tested the possibility that the cytokines affect the expression of their own or partner’s receptors. The expression level (depicted as MFI) of IFNG-R1, IFNG-R2 and TNF-R1 were analyzed after 20 h of cytokine priming. Notably, the abundance of IFNG-R1 (Figure 5d) was increased by approximately 35.6% after TNF-α priming, compared with IFN-γ itself (p < 0×01). Combined IFN-γ and TNF-α priming increased about 16×4% of IFNG-R1, yet it was not statistically significant (p = 0×096). For IFNG-R2 (Figure 5e), both TNF-α alone or in combination with IFN-γ did not elevate the receptor abundance significantly in comparison with IFN-γ alone (p = 0×229, p = 0×076, respectively). Interestingly, TNF-α decreased the abundance of its own receptor TNF-R1 dramatically when compared with the vehicle (p < 0×001) and IFN-γ group (p < 0×001), indicating that the expression of TNF-R1 is controlled by a negative feedback mechanism (Figure 5f).
Taken together, the signaling of IFN-γ might be enhanced by TNF-α through up-regulation of IFNG-R1, which is possible to be one interplay mechanism of the two cytokines.
IFN-γ and TNF-α synergistically induce MSCs impairment via necroptosis
To clarify the pattern and underlying molecular mechanism of MSCs impairment in this scenario, we further conducted a bulk RNA-sequencing of cytokine primed MSCs.
After cleaning and quality control, RNA-Sequencing data showed high correlation of expression profiles in each paired group when analyzed with Pearson’s correlation and principal component analysis (Figure 6a, b). A differentially expressed genes (DEGs) analysis was then performed to identify gene expression changes between groups. Strikingly, MSCs-IT showed significantly different gene expression mode in comparison with MSCs-vehicle and MSCs-T; and a slight similarity between MSCs-IT and MSCs-I was seen, as shown by heat map plotted with 3680 DEGs (log2 [fold-change] > 1 and p < 0×05) (Figure 6c, d). This result indicated that the alterations in gene expression of MSCs-IT might be resulted from an enhancement of IFN-γ signaling after the addition of TNF-α.
This inference was supported by analyzing the 3245 DEGs between MSCs-IT and MSCs-T (the two groups showed the greatest difference in both morphology and DEGs) (Figure 7a, b). Functional annotation was analyzed through GO-Cellular Component (CC) alone, as the primed MSCs-IT showed a significant alteration in cellular morphology. The significantly enriched top 20 CC terms in the enrichment analysis of DEGs were shown in Figure 7c, and the most enriched were terms associated with cell morphology changes, such as plasma, cytosol, membrane, cytoplasm, cytoskeleton, and spindle etc., which was in consistent with the previous results of cell morphology. Next, a KEGG pathway enrichment analysis was conducted to explore the most significantly enriched pathways (Figure 7d). The most enriched pathways included cytokine-cytokine receptor interaction, necroptosis, apoptosis, and TNF signaling pathways etc., among which the “necroptosis” was paid special attention to.
Necroptosis is a form of programmed cell death characterized by a similar morphology of necrosis, including cell swelling, membrane rupture, and releasing of cellular contents, such as inflammatory cytokines, damage-associated molecular patterns (DAMPs), and chemokines, thereby leading to a variety of inflammatory responses subsequently. It can be triggered by various cytokines and pattern recognition receptors through phosphorylation of receptor interacting protein kinase 1 (RIPK1), which then phosphorylates and activates RIPK3 to form a RIPK1-RIPK3-MLKL (mixed lineage kinase domain-like protein) complex responsible for disrupting plasma membrane integrity via membrane pore executed by MLKL oligomer formation.[30-32]
Further analysis revealed that all genes associated with necroptosis (RIPK1, RIPK3, MLKL, TRADD, TICAM1, ZBP1, TLR3, and TLR4)[33] were significantly up-regulated in the group synergistically primed with IFN-γ and TNF-α (Figure 8a-c), indicating that the morphologically and functionally altered MSCs-IT had undergone necroptosis. Most importantly, this manner of cell death evidenced the results seen in our previous experiments.
Necroptosis of MSCs induced by IFN-γ and TNF-α is reduced by RIPK3 and/or MLKL inhibitors
As necroptosis is a form of regulated cell death, we investigated whether inhibitors of RIPK1, RIPK3 and MLKL can reverse or rescue that process, hoping to validate the results of RNA-Sequencing at the same time.
Strikingly, observed by phase contrast microscope, RIPK3 inhibitor GSK-872[26] and MLKL inhibitor NSA[25] remarkably attenuated the necroptotic morphology of MSCs-IT, while the rescue effect of RIPK1 inhibitor necrostatin-1 (Nec-1) [24] was not so evident (Figure 8d).
These results indicated that IFN-γ and TNF-α indeed induced necroptosis of MSCs collaboratively, and it was a RIPK1-independent process initiated through not yet known alternative pathways to activate RIPK3 and MLKL.