NPC senescence and impaired autophagic activity are strongly correlated with IDD progression
To dissect the role of cellular senescence in IDD progression, we collected intervertebral disc tissue samples from patients. Based on the Pfirrmann MRI classification system, healthy or differently degenerated disc tissues were classified as grade I/II/III/IV. We first analyzed the expression of senescence-associated markers in NP tissues, and these markers were dramatically elevated in degenerated NP tissues (Fig. 1a–c). In addition, linear regression analysis of senescence-associated protein expression levels and IVD degeneration grade showed that senescence-associated markers levels were positively correlated with IVD degeneration grade (Fig. 1d). Furthermore, we developed an in vitro model of NPC degeneration by exposing NPCs to tert-butyl-hydroperoxide (TBHP) based on a previous study . TBHP-treated NPCs consistently exhibited multiple senescence phenotypes, including upregulated expression of senescence-associated markers, a decreased ratio of EdU-positive cells, and increased SA-β-Gal activity (Fig. 1e–g). Autophagy is a regulatory mechanism of cellular homeostasis, and decreased autophagy has been implicated in a variety of degenerative diseases. Indeed, impaired autophagy in IDD tissues was evidenced by reduced LC3-II expression and elevated p62 expression, as indicated by immunohistochemistry (IHC) analysis and immunoblot analysis (Fig. 1c, d, h, i). Subsequently, we determined the autophagic activity of TBHP-treated NPCs by LC3-II autophagic flux. Reduced LC3-II flux was observed in NPCs after TBHP treatment, as shown in Fig. 1j. Moreover, the expression of the autophagic substrate p62 was elevated in NPCs after TBHP treatment (Fig. 1k, Fig. S1b). To further confirm the above findings, NPCs transduced with stubRFP-sensGFP-LC3 were used to monitor autophagosome maturation. The percentage of red-only puncta, representing autolysosomes, was significantly decreased in TBHP-treated NPCs (Fig. 1l). To investigate whether autophagy mediates NPC senescence, we performed senescence analysis on NPCs after knockdown of ATG7 or treatment with two autophagy inhibitors, 3-MA or BafA1. The protein levels of senescence-related markers were elevated strongly in NPCs treated with si-ATG7, 3-MA or BafA1 (Fig. 1m). Consistently, SA-β-Gal staining and EdU incorporation assay indicated that autophagy blockade could accelerate NPC senescence (Fig. 1n, o). This line of evidence suggests that NPC senescence and impaired autophagic activity are strongly correlated with IDD progression and decreased autophagic activity contributes to increased senescence in NPCs in IDD.
Impaired TFEB-regulated lysosomal dysregulation contributes to NPC senescence
Completion of the autophagic process depends on the fusion of newly synthesized autophagosomes with lysosomes, while lysosomal dysfunction could impede the clearance of autophagosomes. We next evaluated the status of lysosomes in degenerated NPCs. We found that TBHP treatment reduced the number of lysosomes, as indicated by LysoTracker staining (Fig. 2a). Further, we used Magic Red to assess the proteolytic activity of lysosomes. The results showed that the red fluorescence intensity of TBHP-treated NPC was comparable to that of BafA1, which is a lysosomal acidification inhibitor and is used here as a positive control (Fig. 2b). To investigate whether TBHP treatment led to lysosomal membrane damage, we used immunofluorescence assays to assess whether the lysosomal marker enzyme cathepsin B leaks into the cytoplasm. The results shown that the intensity of cathepsin B and LAMP2 colocalization was similar between control and TBHP-treated cell. However, LLOMe exposure (positive control) induced significant cathepsin B cytosolic release which indicated by a reduced colocalization between cathepsin B and LAMP2 (Fig. S2a). This suggests that lysosomal stress levels are lower in our model. Degenerated NPCs are accompanied by autophagic failure and lysosomal dysregulation, which are transcriptionally controlled by a master regulator, TFEB. Next, we evaluated TFEB activity in degenerated NPCs. Under TBHP stimulation, TFEB mRNA and protein expression did not change significantly (Fig. S2b, c). However, TFEB nuclear localization was decreased in degenerated NPCs (Fig. 2c, d). Moreover, the phosphorylation level of TFEB, which determines its nuclear translocation, was significantly elevated in degenerated NPCs (Fig. 2e). To assess whether the loss of TFEB activity is associated with any compensatory activation of other TFE/MITF family of transcriptional factors, we evaluated the status of o MITF and TFE3. The results showed that the nuclear translocation of MITF and TFE3 was similar between control and TBHP-treated cell (Fig. S2d, e). Previous studies have shown that mTOR is the major phosphokinase complex regulating TFEB phosphorylation. However, immunofluorescence staining showed that TBHP treatment significantly reduced the nuclear translocation of TFEB in NPC treated with Torin1, a strong mTOR inhibitor, suggesting that TBHP treatment-induced TFEB hyperphosphorylation is due to a regulatory mechanism independent of mTOR (Fig. S2f). The ubiquitin-proteasome pathway is also involved in the regulation of TFEB activity through degrading phosphorylated TFEB. We used MG132, a proteasome inhibitor, to clarify whether the ubiquitin proteasome pathway is involved in the TBHP-induced cytoplasmic retention of TFEB. The results showed that TBHP treatment significantly increased TFEB phosphorylation levels and enhanced its binding with 14-3-3, even when the proteasome pathway was inhibited (Fig. S2g, h). These suggest that TBHP-induced cytoplasmic accumulation of TFEB is not dependent on the proteasome pathway. To further explore the role of TFEB in NPC senescence, we first knocked down TFEB in NPCs with an siRNA (Fig. S2i, j). Immunoblot analysis showed that TFEB knockdown strongly induced impaired autophagy and cellular senescence (Fig. 2f–h). Further, we used the TFEB S211A mutant to sort out the link between TFEB phosphorylation and NPC senescence. We re-expressed wild-type TFEB and S211A mutant in TFEB-depleted NPCs. RT‒qPCR analysis and immunofluorescence analysis showed that the reconstituted expression of S211A mutant in NPCs blocked TBHP-reduced TFEB inactivation (Fig. 2i, j). Importantly, TFEB S211A mutant could rescue the senescence phenotype in TBHP-treated NPCs, as indicated by western blot, SA-β-gal activity analysis and EdU incorporation assay (Fig. 2k–m). Taken together, these results indicated that impaired TFEB-regulated lysosomal dysregulation and autophagic failure contributes to NPC senescence.
Lysine methylation is involved in TFEB inactivation and IDD progression
Activation of TFEB is regulated by a range of protein PTMs, including phosphorylation, ubiquitination, sulfhydration and alkylation. As an integral type of protein modification, lysine methylation is involved in both chromatin-associated and non-chromatin-associated signaling pathways, and these pathways play an important role in various aspects of cellular biology. However, the relationship between lysine methylation and TFEB and its role in IDD remain elusive. First, we assessed the effect of lysine methylation on TFEB activity by using the pan-methylation inhibitor Adox, and the results showed that Adox treatment rescued TBHP-reduced TFEB activity, as evidenced by the elevated nuclear accumulation of TFEB and the upregulated expression of its downstream gene (Fig. 3a–c). Subsequently, Adox treatment effectively alleviated the TBHP-induced increase in senescence-associated marker expression, delayed proliferation, and increased SA-β-Gal activity (Fig. 3d–g, Fig. S3a, b). In addition, Adox partially reversed the autophagic activity of NPCs treated with TBHP, as evidenced by the elevated autophagic flux (Fig. 3h–j and Fig. S3c). To further assess the role of lysine methylation in IDD progression in vivo, a needle puncture-induced IDD rat model was established (Fig. 3k) . X-ray and micro-CT analyses suggested Adox treatment substantially attenuated the loss of disc height (Fig. 3l, q). Results from MRI examination revealed the T2-weighted signals of the IVD in rats treated with PBS were lower than those in rats received Adox injection (Fig. 3l, m). Moreover, H&E and SO&FG staining showed that Adox treatment significantly attenuated IVD histological degeneration, as evidenced by improved extracellular matrix arrangement, elevated NP tissue area and a clearer boundary between NP and AF (Fig. 3n, p and Fig. S3d). Importantly, the expression of senescence-related markers was significantly decreased after the intradiscal injection of Adox, while autophagic activity was restored (Fig. 3o and Fig. S3d). Collectively, these data indicate that reduced TFEB activity in IDD progression may be regulated by lysine methylation.
K141 methylation of PPP1CA disrupts its binding with TFEB
To assess the methylation status of TFEB, we performed LC‒MS/MS in human NPCs (Fig. 4a). However, no TFEB methylation modification sites were found. Surprisingly, protein phosphatase PPP1CA, a TFEB-interacting protein, was found to possess a methylation modification site at lysine 141 (Fig. 4b). This led us to explore the relationship between PPP1CA and TFEB and the role that methylation plays in this relationship. First, an interaction between ectopically expressed TFEB and PPP1CA was observed by reciprocal coimmunoprecipitation (co-IP) in human embryonic kidney-293 T (HEK293T) cells (Fig. 4c, d), as well as between endogenous TFEB and PPP1CA in NPCs (Fig. 4e, f). Furthermore, the level of TFEB phosphorylation was significantly elevated in PPP1CA-knockdown NPCs and the nuclear localization of TFEB, which is mainly regulated by its phosphorylation level, was significantly decreased in NPCs with PPP1CA knockdown (Fig. 4g–i and Fig. S4a and S4b). Next, we evaluated the binding of TFEB to 14-3-3, which is regulated by the TFEB phosphorylation status. The results showed that PPP1CA knockdown enhanced TFEB binding to 14-3-3 (Fig. S4c). However, mTORC1 activity was not significantly altered in PPP1CA knockdown NPCs, as evidenced by the phosphorylation of mTORC1 substrates 4EBP1 and S6K (Fig. S4d). These data indicate that PPP1CA interacts with and dephosphorylates TFEB in an mTORC1 non-dependent manner. Next, we investigated whether lysine 141 of PPP1CA was methylated. Sequence alignment across multiple species demonstrated that lysine 141 of PPP1CA is an evolutionarily conserved residue (Fig. 4j). Using an anti-pan-methylation antibody, we detected methylation in endogenous PPP1CA immunoprecipitates, and PPP1CA methylation levels were elevated in NPCs treated with TBHP (Fig. 4k). Furthermore, we mutated lysine 141 of PPP1CA to arginine or methionine, where lysine to arginine (KR) was used as a methyl-deficient mutation and lysine to methionine (KM) was used as a methyl-mimetic mutation . Strikingly, the methylation levels of both KR and KM mutants were significantly decreased, and TBHP treatment did not increase the methylation levels of the KR/KM mutants compared with those of wild-type PPP1CA (Fig. 4l). These data suggest that PPP1CA is methylated at lysine 141. Subsequently, we evaluated the colocalization of TFEB with PPP1CA and showed that the intensity of TFEB and PPP1CA colocalization was significantly decreased in TBHP-treated NPCs (Fig. 4m). The interaction between endogenous TFEB and PPP1CA was consistently weakened in NPCs treated with TBHP (Fig. 4n). The recruitment of TFEB to lysosomes is closely related to its phosphorylation status, so we next assessed whether TFEB binds to PPP1CA at the lysosome. The results seem to show some binding of PPP1CA/TFEB to lysosomes, however, TBHP treatment did not change their co-localization (Fig. S4e). To investigate whether the attenuated interaction between TFEB and PPP1CA was mediated by PPP1CA methylation, KR mutants were transfected into NPCs. Upon TBHP stimulation, the interaction between Flag-TFEB and wild-type PPP1CA but not the KR mutant was significantly reduced (Fig. 4o). Furthermore, we assessed the interaction between Flag-TFEB and His-PPP1CA in HEK293T cells. The results showed that the KM mutant but not the KR mutant or wild-type protein diminished the binding between Flag-TFEB and His-PPP1CA (Fig. 4p). Collectively, these data indicate that the K141 methylation of PPP1CA disrupts its binding with TFEB.
K141 methylation of PPP1CA modulates TFEB, autophagic activity and cellular senescence in NPCs
To investigate the role of PPP1CA K141 methylation in degenerated NPCs, wild-type PPP1CA or KR mutants were re-expressed in PPP1CA-depleted NPCs (Fig. 5a). RT‒qPCR analysis of TBHP-treated NPCs transduced with KR mutants revealed an attenuated reduction in TFEB downstream gene expression compared to wild-type PPP1CA-transduced NPCs (Fig. 5b). Immunofluorescence (IF) analysis consistently showed that TBHP treatment significantly attenuated TFEB nuclear accumulation in NPCs transduced with wild-type PPP1CA but not in NPCs transduced with KR mutants (Fig. 5c). These data indicate that reduced TFEB activity is regulated by the K141 methylation of PPP1CA. Next, autophagic flux analysis showed that the reconstituted expression of KR mutants in NPCs blocked TBHP-reduced autophagic flux, as evidenced by restored LC3-II accumulation and the percentage of red-only puncta (Fig. 5d, S5a). Furthermore, the TBHP-induced expression of senescence-associated markers was abrogated by the reconstituted expression of KR mutants (Fig. 5e). Moreover, NPCs transduced with KR mutants were resistant to the TBHP-induced expression of senescence-associated secretory phenotype (Fig. 5f). In addition, TBHP-induced elevated SA-β-Gal activity and delayed proliferation were significantly attenuated in NPCs transduced with KR mutants compared with those transduced with wild-type PPP1CA (Fig. 5g, h). Together, these data indicate that the K141 methylation of PPP1CA and its inhibitory effect on TFEB activity is a key mechanism underlying TBHP-induced TFEB inactivation and subsequent impaired autophagy and senescence.
K141 methylation of PPP1CA disrupts PPP1CA/PPP1R9B holoenzyme assembly
The PPP1CA holoenzyme consists of an activated catalytic subunit and a regulatory subunit. The substrate specificity of the PPP1CA catalytic core is dependent on the regulatory subunit [25, 26]. Interestingly, we identified a regulatory subunit, PPP1R9B, in the TFEB-interacting protein (Fig. 6a). To determine whether PPP1R9B targets PPP1CA to TFEB, we first evaluated the binding between PPP1CA, PPP1R9B, and TFEB. Co-IP assays showed ectopically expressed GST-PPP1R9B in the His-PPP1CA and Flag-TFEB immunoprecipitates and vice versa (Fig. 6b, c, e, f). Moreover, endogenous Co-IP results indicated that PPP1CA, PPP1R9B, and TFEB bind to each other (Fig. 6d, g). To further confirm the binding of TFEB, PPP1CA and TFEB, we immunoprecipitated PPP1CA from NPCs and followed by mass spectrometric analysis. We found that both PPP1R9B and TFEB are part of the PPP1CA interactome (Fig. 6a). In addition, knockdown of PPP1R9B in NPCs strongly disrupted the binding of PPP1CA to TFEB (Fig. 6h and Fig. S6a and S6b). We also observed a significant reduction in TFEB activity in NPCs with PPP1R9B knockdown, as indicated by reduced TFEB nuclear localization and elevated phosphorylation levels (Fig. 6i–k). These data suggest that the PPP1CA/PPP1R9B complex can bind to and dephosphorylate TFEB, with PPP1R9B playing a bridging role. Given that the PPP1CA targeting of TFEB is dependent on PPP1R9B and that the K141 methylation of PPP1CA blocks its binding to TFEB, we hypothesize that the K141 methylation of PPP1CA may disrupt the assembly of the PPP1CA/PPP1R9B complex. First, the binding of PPP1CA to PPP1R9B was assessed in TBHP-treated NPCs, and their interaction was significantly weakened (Fig. 6l, m). Then, we further mutated lysine 141 to arginine or methionine. Strikingly, in contrast to wild-type PPP1CA, the KR mutants displayed a negligible response in their binding with PPP1R9B upon TBHP treatment (Fig. 6n). In addition, the KM mutants but not the KR mutants or wild-type PPP1CA exhibited a significant reduction in binding to PPP1R9B (Fig. 6o). Collectively, these findings suggest that the K141 methylation of PPP1CA may disrupt PPP1CA/PPP1R9B holoenzyme assembly and subsequent TFEB activation.
K141 methylation of PPP1CA is meditated by the methyltransferase SUV39H2
To identify the upstream methyltransferase responsible for PPP1CA K141 methylation, we immunoprecipitated PPP1CA from NPCs and subsequently used LC–MS/MS to determine the PPP1CA-binding proteins. Three methyltransferases were identified: G9a, STED1A and SUV39H2 (Fig. 7a and Fig. S7a). Further Co-IP assays indicated that PPP1CA binds to SUV39H2 but not to G9a or SETD1A (Fig. 7b). This finding suggests that SUV39H2 potentially methylates PPP1CA. We detected an interaction between endogenous PPP1CA and SUV39H2 in NPCs (Fig. 7c, d). Subsequently, we also identified a relatively strong interaction between exogenous His-PPP1CA and Mbp-SUV39H2 in HEK293T cells (Fig. 7e, f). We also immunoprecipitated SUV39H2 followed by mass spectrometry to further characterize the interaction of SUV39H2 with PPP1CA. The mass spectrometry results once again demonstrated that the binding of SUV39H2 to PPP1CA (Fig. S7b). These results suggest that SUV39H2 is a PPP1CA-interacting protein. Next, we assessed whether SUV39H2 methylated K141 of PPP1CA. We overexpressed SUV39H2 in HEK293T cells and used the pan-lysine methylation antibody to detect the methylation of ectopically expressed PPP1CA. The results indicated that the methylation level of wild-type PPP1CA but not the KR mutant was increased in the cells with SUV39H2 overexpression (Fig. 7g). In addition, treatment with an inhibitor specific to SUV39H2 (SUV39H2i) strongly reduced the methylation of wild-type PPP1CA but not the KR mutant compared to the control groups (Fig. 7h). Importantly, SUV39H2i treatment or the depletion of SUV39H2 led to a significant reduction in endogenous PPP1CA methylation (Fig. 7i, j and Fig. S7c and S7d). These data suggest that SUV39H2 acts as a methyltransferase and is potentially responsible for methylating K141 of PPP1CA. Furthermore, we assessed the effect of SUV39H2 on the PPP1CA/PPP1R9B/TFEB complex. SUV39H2 knockdown enhanced the interaction of PPP1CA with PPP1R9B or TFEB, while SUV39H2 overexpression produced the opposite result (Fig. 7k, l). Moreover, TBHP treatment-induced TFEB activity reduction was restored by SUV39H2 depletion (Fig. 7m–p). Collectively, these findings suggest that SUV39H2 methylates K141 of PPP1CA and suppresses its phosphatase activity targeting TFEB.
Suppression of SUV39H2 delays NPC senescence and IDD development
Given the regulation of the PPP1CA/PPP1R9B/TFEB complex by SUV39H2, we further investigated whether SUV39H2 expression changes during IDD progression. IHC analysis showed that SUV39H2 was elevated in patients with IDD (Fig. 8a). Furthermore, IF staining indicated that SUV39H2 expression increased as IVD tissue damage worsened in the needle puncture IDD rat model (Fig. 8b and Fig. S8a). Western blot analysis showed that SUV39H2 expression was elevated in degenerated NP tissues (Fig. 8c). Further linear regression analysis showed that SUV39H2 expression levels were positively correlated with IVD degeneration grade and autophagic failure (Fig. 8d and Fig. S8b, c). Additionally, TBHP-treated NPCs consistently displayed a marked increase in SUV39H2 (Fig. 8e). Collectively, these data demonstrate that SUV39H2 is increased in the NP of patients and rats with IDD, implicating a potential role for SUV39H2 in IDD progression. To further explore the role of SUV39H2 in NPC senescence as well as in IDD, we knocked down SUV39H2 in NPCs with an siRNA. NPCs with silenced SUV39H2 expression exhibited a significantly alleviated senescence-related phenotype compared with wild-type NPCs treated with TBHP, as indicated by decreased SA-β-gal activity and an increased ratio of EdU-positive cells (Fig. 8f, g and Fig. S8d and S8e). ELISA analysis and immunoblot analysis demonstrated that the expression of senescence-related markers was also decreased in SUV39H2-knockdown NPCs (Fig. 8h, i). Moreover, SUV39H2 depletion effectively alleviated TBHP-induced autophagic flux blockage (Fig. 8j–l and Fig. S8f–h). Collectively, these observations indicate that SUV39H2 knockdown can reverse impaired autophagy activity and cellular senescence in degenerated NPCs. Given that autophagy is negatively regulated by SUV39H2, we next examined whether the elevated SUV39H2 expression in degenerating NPCs was caused by autophagy decay. The results showed that the expression of SUV39H2 was not significantly altered (Fig. S8i). Next, we used adeno-associated virus (AAV) vectors to target and inhibit SUV39H2 in vivo. We injected AAV vectors carrying a short hairpin RNA targeting SUV39H2 into the NP region of IDD rats every week for four weeks (Fig. 8m). Strikingly, SUV39H2 knockdown significantly retarded the degeneration of IVD, as manifested by radiographic imaging and histological assessments (Fig. 8n, o, q, r and Fig. S8j and S8k). Importantly, the AAV-shSUV39H2-treated group exhibited enhanced autophagic activity and reduced expression of senescence-associated markers (Fig. 8p and Fig. S8j). Together, these findings suggest that the targeted inhibition of SUV39H2 could delay IDD progression.