Introduction

Takayasu’s arteritis (TAK) is a chronic large-vessel vasculitis (LVV) characterised by inflammation of the aorta and its major branches.1 TAK occurs predominantly in young women. Severe ischaemic complications, such as ischaemic stroke and acute coronary syndrome, pose it to be a disabling and fatal disease. With the advancement in therapies, the systemic inflammation can be rapidly controlled. However, subclinical vascular inflammation and the subsequent arterial remodelling remain unsolved in TAK.2 The detailed underlying mechanism is still unclear.

It has been proposed and widely accepted that inflammatory cells infiltration in vascular wall initiates from the adventitial vasa vasorum and mediates the inflammatory injury and tissue remodelling in LVV.3 4 In another word, vascular cells are more regarded as the target cells of inflammatory damage rather than pathogenic components of vascular inflammation. Nevertheless, due to the fact that the lesional inflammation is limited to large blood vessel in TAK, it is reasonable to speculate that vascular cells can act as local initiator or amplifier of inflammatory and autoimmune process in vascular lesion. Vascular smooth muscle cells (VSMCs) constitute the major cellular components of arterial system and have high functional plasticity. To our knowledge, the role of VSMCs in TAK development, especially in vascular inflammation, is still unelucidated.

Cellular senescence is a stable and irreversible state of proliferative arrest associated with multiple functional variations known as senescence-associated secretory phenotype (SASP), involving enhanced secretion of proinflammatory cytokines, chemokines and matrix metalloproteases, contributing to the amplification of local inflammation and tissue remodelling.5 On the other hand, recent studies suggested that prolonged exposure to inflammatory cytokine induces ageing-associated change in vascular cells.6 Persistent vascular inflammation in TAK provides rationales for vascular cell senescence, although TAK has a clear predilection for young women. In addition, the manifestations of vascular senescence, such as atherosclerosis7 and vascular calcification,8 are constant in patients with TAK without traditional cardiovascular risk factors and result in poor outcomes.9 These findings prompted us to propose the hypothesis that inflammatory milieu in TAK induces VSMCs senescence, which in turn contributes to vascular inflammation.

In the present study, through combinational approaches, we investigated whether premature VSMCs senescence is present in patients with TAK, its association with vascular inflammation, as well as the key prosenescent factor and the downstream signalling pathway in TAK.

ResultsVSMCs of patients with TAK have the feature of premature senescence and contribute substantially to vascular inflammation

We first searched for the histological evidence of cellular senescence. According to a multimarker algorithm for assessment of cellular senescence,10 GL13, a biotinylated Sudan Black B-based chemical reagent that detects the lipofuscin, was used for identification of senescent cells in formalin-fixed and paraffin embedded vascular samples using immunohistochemistry and immunofluorescence. The expression of p16, another prevalently used cellular senescence marker, was also adopted for assessing senescent state of cells at tissue level. We found that both the proportions of GL13-positive cells and those of p16-positive cells were significantly higher in vascular samples of TAK than control samples, and these cells were largely distributed in the media of the involved artery (figure 1A,B). The results of Schmorl method for lipofuscin detection further confirmed our finding (online supplemental figure S1B).

Figure 1

The feature of premature senescence in VSMCs of patients with TAK. (A) Left: Representative images for single GL13 staining with immunohistochemistry (upper, ×30) and immunofluorescence (lower, ×20) in vascular sections of patients with TAK and control individuals, showing higher proportions of GL13-positive cells and their distribution in intima (I), media (M) and adventitia (A) of TAK arteries; Right: Comparison of GL13-positive cells ratios between vascular samples of patients with TAK (n=22) and control individuals (n=10). (B) Left: Representative fluorescence images showing p16 expression and the distribution of p16-positive cells in intima (I), media (M) and adventitia (A) of vascular sections from patients with TAK and control individuals (×20); Right: Comparison of p16-positive cells ratios between vascular samples of patients with TAK (n=22) and control individuals (n=10). (C) Costaining of GL13 (purple) and vascular cell marker (α-SMA, CD31 or Vimentin; green) in the vascular sections of patients with TAK and control individuals by immunofluorescence (×40). (D) Costaining of p16 (green) and vascular cell marker (α-SMA, CD31 or Vimentin; red) in the vascular sections of patients with TAK and control individuals by immunofluorescence (×40). (E) Quantification analysis showing the proportions of GL13-positive (left) and p16-positive (right) vascular cells (VSMCs, α-SMA(+); endothelial cells, CD31(+); fibroblasts, Vimentin(+)) in the vascular sections of patients with TAK and control individuals by immunofluorescence. (F) Comparison of age in TAK patients with different GL13-positive VSMCs ratios (left) and p16-positive VSMCs ratios (right). (G) Costaining of α-SMA (Green) and either Lamin B1 (red, ×60) or Ki67 (white, ×30) in the vascular sections of patients with TAK and control individuals by immunofluorescence. (H) Quantification analysis showing the proportions of Lamin B1-positive VSMCs (left) and Ki67-positive VSMCs (right) in the vascular sections of patients with TAK and control individuals. (I) UMAP plot showing cellular subsets in vascular lesion of patients with TAK (n=3) and control individuals (n=3). (J) UCell analysis to compare the activation status of cellular senescence pathway in VSMCs of patients with TAK (n=3) and control individuals (n=3). (K) Pie plot showing the distribution of total UCell score of cellular senescence pathway among different cells subpopulations in vascular samples of TAK. ns, non-significant, **p<0.01, ****p<0.0001. TAK, Takayasu’s arteritis; VSMCs, vascular smooth muscle cells.

In vascular samples of TAK, the colocalisation between α-SMA (VSMCs marker) and either GL13 staining or p16 was detected while no colocalisation between either CD31 (endothelial cell marker) or Vimentin (fibroblast marker), and senescent markers (GL13 and p16) was observed (figure 1C,D). Furthermore, significantly higher ratios of senescent VSMCs (GL13+ α-SMA+cells and p16+ α-SMA+cells) were found in vascular samples of TAK, compared with those of control samples (figure 1E). Notably, there was no difference in age among patients with various proportions of GL13-positive or p16-positive VSMCs (figure 1F). Primarily cultured VSMCs from vascular lesion of patients also exhibited more intense SA-β-gal staining and higher p16 expression levels, compared with those from control samples, and this trend was also age-independent (online supplemental figure S1C,D). As for other features of cellular senescence, the proportions of Lamin B1-positive VSMCs and Ki-67-positive VSMCs were significantly decreased in patients with TAK (figure 1G,H).

This led us to seek out the transcriptomic evidence of VSMCs senescence in vascular lesion of TAK. We revisited a recently published bulk RNA-Seq dataset on aortic tissues from a total of 14 individuals including 8 patients undergoing renal transplantation and 6 patients with TAK (Cohort 1, online supplemental figure S1E). KEGG analysis, together with GSEA, suggested that cellular senescence pathway was upregulated in vascular tissue of patients with TAK, compared with control samples (online supplemental figure S1F). In addition, to validate our findings, bulk RNA-Seq was performed on fresh frozen aortic tissue of another four young patients with TAK undergoing bypass surgeries and three control samples of elderly individuals with atherosclerosis (cohort 2, online supplemental figure S1E). The results of KEGG analysis and GSEA showed similar trends (online supplemental figure S1G). Single-cell RNA sequencing of vascular lesion and UCell analysis were then performed to investigate the activation status of cellular senescence pathway in vascular cells of TAK. Cluster identification revealed distinct cell subgroups including vascular cells (VSMCs, endothelial cells and fibroblasts) and immune cells (macrophages, B cells, T cells and mast cells) (figure 1I, online supplemental figure S1H). Notably, substantially higher UCell scores for cellular senescence pathway in both VSMCs and macrophages, but not in other types of cells including fibroblasts, mast cells, endothelial cells, B cells, and T cells, were observed in patients with TAK, compared with their counterparts of control samples that were obtained from three elderly individuals (figure 1J, online supplemental figure S1I). The colocalisation between CD68 (macrophage marker) and GL13, as well as that between CD68 and p16 were also detected in artery tissue of TAK, consistent with the transcriptome results (as shown in online supplemental figure S1J). Interestingly, among all cell types in vascular samples of TAK, we found that the activation status of cellular senescence pathway in VSMCs accounted largely for overall activation status by comparing the total UCell scores of different cell subpopulations (figure 1K).

We then investigated whether VSMCs senescence contributes to systemic inflammation or vascular inflammation in TAK via SASP. The expression profiles of several well-defined ageing-associated cytokines, including interleukin-6 (IL-6), IL-7, IL-8, IL-13, IL-15, IL-1α, IL-1β and CXCL1,11 in vascular tissue lysates and paired serum collected before operation from 22 patients with TAK and 10 control subjects were revealed using Luminex Multi-Analyte Profiling technology (online supplemental figure S2A). Increased expression levels of IL-6, IL-8, IL-1α, IL-1β and CXCL1 in vascular lesions of patients with TAK but not in paired serum were detected (figure 2A). Notably, our single-cell transcriptomic data suggested that VSMCs of patients with TAK expressed higher mRNA expression levels of these cytokines than VSMCs of control individuals (figure 2B), and that VSMCs are the most prominent source of Il6 (encodes IL-6) and the second most predominant source of Cxcl8 (encodes IL-8), Il1a (encodes IL-1α), Il1b (encodes IL-1β), Cxcl1 (encodes CXCL1) in the vascular lesion of TAK (figure 2C). In addition, Il6-expressing VSMCs, Cxcl8-expressing VSMCs, Il1a-expressing VSMCs, Il1b-expressing VSMCs and Cxcl1-expressing VSMCs had higher UCell scores for cellular senescence pathway than their counterpart that do not (figure 2D), which was validated by the colocalisation or adjacency between either GL13 or p16 and these cytokines in the VSMCs of patients with TAK that was revealed in immunofluorescence (figure 2E,F). Moreover, higher expression levels of these cytokines in vascular lysates but not in paired serum were detected in patients with higher proportions of GL13-positive or p16-positive VSMCs (figure 2G,H, online supplemental figure S2B,C). As to the association with markers of systemic inflammation, no significant difference in erythrocyte sedimentation rate and C reactive protein levels was observed among patients with various proportion of GL13-positive or p16-positive VSMCs (figure 2I,J).

Figure 2

The SASP of VSMCs contributes to the vascular inflammation of TAK. (A) Heatmap of senescence-associated inflammatory cytokines levels in serum and paired vascular lysates from patients with TAK (n=22) and control individuals (n=10). (B) Comparison of senescence-associated inflammatory cytokines expression in VSMCs of patients with TAK and control individuals based on single-cell transcriptomic analysis. (C) Pie plot showing the expression distribution of senescence-associated inflammatory cytokines among different cell types in vascular lesion of TAK. (D) Histogram showing UCell scores of cellular senescence pathway in VSMCs expressing or not expressing senescence-associated inflammatory cytokines. (E) Costaining of GL13 (purple), α-SMA (green) and senescence-associated inflammatory cytokines (IL-6, IL-8, CXCL1, IL1β and IL1α; red) in the vascular sections of patients with TAK by immunofluorescence (×20). (F) Costaining of p16 (green), α-SMA (red) and senescence-associated inflammatory cytokines (IL-6, IL-8, CXCL1, IL1β and IL1α; purple) in the vascular sections of patients with TAK by immunofluorescence (×20). (G) Comparison of senescence-associated inflammatory cytokines expression levels in vascular lysates of TAK patients with different GL13-positive VSMCs ratios. (H) Comparison of senescence-associated inflammatory cytokines expression levels in vascular lysates of TAK patients with different p16-positive VSMCs ratios. (I) Comparison of ESR and CRP in TAK patients with different GL13-positive VSMCs ratios. (J) Comparison of ESR and CRP in TAK patients with different p16-positive VSMCs ratios. (K) Upper: Inflamed areas of vascular samples of TAK were determined by H&E staining and marked by the dotted lines (upper left: ×5; upper right: ×40); costaining of GL13 (green), IL-6 (purple) and α-SMA (red) showing the spatial association between double positive GL13/IL-6 VSMCs and inflamed areas in vascular samples of TAK (middle left: ×5; middle right: ×40); costaining of p16 (Purple), IL-6 (Red) and α-SMA (Green) showing the spatial association between double positive p16/IL-6 VSMCs and inflamed areas in vascular samples of TAK (lower left: ×5; lower right: ×40). (L) Comparison of GL13-positive VSMCs (upper) and p16-positive (bottom) VSMCs ratios at the site of inflammation and at the adjacent non-inflamed regions in TAK artery with transmural inflammation. (M) Comparison of GL13-positive VSMCs (upper) and p16-positive (bottom) VSMCs ratios between vascular samples of TAK with limited disease (limited to adventitia) and extended disease (transmural inflammation). ns, non-significant, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CRP, C reactive protein; ESR, erythrocyte sedimentation rate; SASP, senescence-associated secretory phenotype; TAK, Takayasu’s arteritis; VSMCs, vascular smooth muscle cells.

To better illustrate the spatial relationship between senescent VSMCs and inflammatory cells in TAK, inflammatory areas were determined by H&E staining (figure 2K). Further, multicolour immunofluorescence staining showed the adjacency between senescent VSMCs (either GL13-positive or p16-positive VSMCs) and inflammatory cells, including those expressing IL-6 (figure 2K). Consistently, senescent VSMCs were more prominent in inflamed areas than non-inflamed areas in vascular samples of TAK (figure 2L). Also, the proportion of senescent VSMCs was significantly higher in patients with extended disease (transmural inflammation) than that in patients with inflammation limited to adventitia (online supplemental figure S2D, figure 2M), demonstrating a link between senescent VSMCs and the extension of histological inflammation in TAK arteries. In addition, persistent vascular inflammation underlies chronic damage in involved arteries. In this study, senescent VSMCs could also be detected in areas around calcified plaque (online supplemental figure S2E), suggesting that VSMCs senescence might be associated with both active ongoing disease and chronic damage in TAK.

Altogether, these findings provided comprehensive evidence for the presence of premature senescence and SASP in VSMCs and its important contribution to vascular wall inflammation by increased secretion of proinflammatory factors in TAK.

Proinflammatory microenvironment of TAK exerts the prosenescent effects on VSMCs via IL-6 signalling

TAK is characterised by marked inflammatory cell infiltration in vascular lesion, especially in media and adventitia. To test if the inflammatory milieu in TAK is an important trigger of VSMCs senescence, coculture experiments of T/G HA-VSMCs and peripheral blood mononuclear cells (PBMCs) of patients with active/inactive TAK were conducted (figure 3A). T/G HA-VSMCs cocultured with PBMCs of patients with active disease (ATA PBMC) exhibited the highest p16 expression and the most intense SA-β-gal staining, followed by inactive TAK PBMCs (ITA PBMC)-treated VSMCs and untreated VSMCs (Control) (figure 3B,C). In addition, T/G HA-VSMCs treated with TAK-PBMC conditioned medium (TAK-PBMC CM) had comparable p16 expression levels and staining intensity of SA-β-gal with corresponding TAK PBMC-cocultured VSMCs (figure 3A–C), suggesting the prosenescent effects of immune cells from patients with TAK on VSMCs in a cell contact-independent way. That is, this effect might be mediated by soluble inflammatory cytokines.

Figure 3

Prosenescent effects of TAK inflammatory milieu on VSMCs via IL-6 signalling. (A) Schematic diagram of VSMCs culture experiment. (B, C) p16 expression (B) and SA-β-gal staining (C) of TAK PBMC-cocultured T/G HA-VSMCs and TAK-PBMC CM-treated T/G HA-VSMCs. (D) Venn diagram depicting the number of upregulated DEGs that were listed in cytokine-cytokine receptor interaction pathway based on the bulk RNA-Seq analysis of the two datasets (cohorts 1 and 2). (E) Volcano plot showing the common upregulated DEGs that were listed in cytokine-cytokine receptor interaction pathway based on the bulk RNA-Seq analysis of the two datasets (cohorts 1 and 2). (F) Comparison of Il6r expression in VSMCs of patients with TAK and control individuals based on single-cell transcriptomic analysis. (G) Pie plot showing the expression distribution of Il6r among different cell types in vascular lesion of TAK based on single-cell transcriptomic analysis. (H) Histogram showing UCell scores of cellular senescence pathway in VSMCs expressing or not expressing Il6r. (I) Ucell analysis to compare the activation status of IL-6-JAK-STAT3 pathway in VSMCs of patients with TAK and control individuals. (J) Correlational analysis between UCell scores of IL-6-JAK-STAT3 pathway and cellular senescence pathway in VSMCs of patients with TAK. (K, L) SA-β-gal staining (K) and p16 expression (L) showing the effects of blocking IL-6 signalling via IL6R knockdown or anti-IL-6 neutralising antibody on TAK-PBMC CM-induced VSMCs senescence. (M, N) SA-β-gal staining (M) and p16 expression (N) showing the prosenescent effects of IL-6 stimulation on primary VSMCs from patients with TAK. ns, non-significant, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. PBMC CM, peripheral blood mononuclear cells conditioned medium; TAK, Takayasu’s arteritis; VSMCs, vascular smooth muscle cells.

During the disease progression of TAK, VSMCs were exposed to a complex composition of inflammatory cytokines. We next investigated the key cytokines and the downstream signalling pathway that mediates the prosenescent effects on VSMCs in TAK. The upregulated differentially expressed genes in the above two bulk RNA-seq datasets (cohorts 1 and 2) were both found to be enriched in the cytokine–cytokine receptor interaction pathway (online supplemental figure S1F,G). Further, there were 19 common cytokine-encoding genes or cytokine-receptor-encoding genes that were upregulated in both datasets (figure 3D). Interestingly, Il6, together with Il6r (encodes IL-6R), was the only co-upregulated genes that encode a cytokine and its receptor respectively, providing a mechanism for IL-6 signal amplification (figure 3E). Although VSMCs expressed relatively low levels of membrane bound IL-6R (mIL6R) in physiological conditions,12 13 single‐cell sequencing data analysis showed upregulated expression of Il6r in VSMCs of patients with TAK (figure 3F) that was validated by immunofluorescence performed on vascular tissues (online supplemental figure S3A) and suggested that among different types of vascular cells in TAK, VSMCs were the ones with the highest total expression level of Il6r (figure 3G). In addition, higher UCell scores for cellular senescence pathway were detected in Il6r-expressing VSMCs than their counterpart that do not (figure 3H). Furthermore, VSMCs of patients with TAK had significantly higher UCell scores for IL-6-JAK-STAT3 pathway than those in control samples (figure 3I). More importantly, there was a significant correlation between the UCell scores for IL-6-JAK-STAT3 pathway and those for cellular senescence pathway in VSMCs of TAK (figure 3J). Notably, immunofluorescence analysis revealed significantly upregulated mIL6R expression in VSMCs of patients with TAK while only a very small amount of IL6R expression was observed in extracellular space surrounding VSMCs (online supplemental figure S3A), suggesting that IL-6 might exert prosenescent effect largely through classical signalling, rather than trans-signalling. Collectively, these results provided the rationale for a link between IL-6 signalling activation and VSMCs senescence in TAK, mainly from a transcriptomic perspective.

To investigate whether IL-6 signalling mediates the prosenescent effects of TAK-PBMC CM on VSMCs, IL-6 and IL6R expression in PBMCs that were obtained from TAK patients with inactive or active disease and were used for VSMCs senescence induction, were tested by ELISA performed on supernatants from PBMCs cultures and flow cytometry. Significantly higher IL-6 expression in PBMCs was detected in TAK patients with active disease than those with inactive disease (online supplemental figure S3B) while no significant difference in mIL6R or soluble IL6R (sIL6R) expression in PBMCs was observed between patients with active or inactive TAK (online supplemental figure S3C). Transcriptome differences between T/G HA-VSMCs treated with TAK-PBMC CM and those treated with RPMI 1640 medium as control were further revealed by bulk RNA-Seq. KEGG pathway enrichment analysis and GSEA suggested the co-upregulation of cellular senescence pathway and IL-6-JAK-STAT3 pathway in TAK-PBMC CM-treated T/G HA-VSMCs (online supplemental figure S3D,E). Blocking IL-6 signalling via receptor knockdown, anti-IL-6 neutralising antibody, or anti-IL6R neutralising antibody could suppress TAK-PBMC CM-induced VSMCs senescence (online supplemental figure S3F, figure 3K,L, online supplemental figure S3G,H). Also, treating with recombinant human IL-6 led to cellular senescence in primary cultured VSMCs from patients with TAK, in a dose-dependent manner (figure 3M,N). Taken all together, these results indicated that IL-6 is both an important inducer and a downstream effector of VSMCs senescence in the context of TAK, forming a positive feedback loop.

IL-6 signalling drives senescence-associated mitochondrial dysfunction in VSMCs of TAK

We next investigated the intracellular prosenescent factors in VSMCs of TAK and TAK-PBMC CM-treated VSMCs at the organelle level. It is generally accepted that cellular senescence is closely interconnected with mitochondrial dysfunction. In addition, mitochondria, typically comprise 3%–5% of VSMCs volume (typically 3%–8% in skeletal muscle), maintain vascular tone by energy provision. Functionally, VSMCs possess distinctively enhanced nonphosphorylating respiration than cardiac muscle cell and skeletal muscle cell, suggesting stronger potential in ROS production.14 In the present study, significantly lower UCell scores for oxidative phosphorylation (OXPHOS) pathway, as well as higher UCell scores for reactive oxygen species pathway, glycolysis pathway and mitochondrial fusion pathway were observed in VSMCs from vascular lesions of TAK than those from control samples (figure 4A–D), providing transcriptomic evidence for senescence-associated mitochondrial dysfunctions in VSMCs of TAK.15 GSEA results also suggested that TAK-PBMC CM-treated VSMCs exhibited a downregulation of OXPHOS pathway (online supplemental figure S4A). Collectively, we asked whether inflammatory environment in the context of TAK fuels senescence-associated mitochondrial dysfunction in VSMCs.

Figure 4

IL-6 signalling induces senescence-associated mitochondrial dysfunction in VSMCs of TAK. (A–D) Ucell analysis to compare the activation status of oxidative phosphorylation pathway (A), reactive oxygen species pathway (B), glycolysis pathway (C) and mitochondrial fusion pathway (D) in VSMCs of patients with TAK (n=3) and control individuals (n=3). (E, F) Electron microscopy images (E) and fluorescence images of MitoTracker Green staining (F) showing TAK-PBMC CM-induced mitochondrial morphologic changes in T/G HA-VSMCs and the effects of blocking IL-6 signalling via IL6R knockdown. (G, H) JC-1 staining (G) and mitoSOX staining (H) showing TAK-PBMC CM-induced changes in the mitochondrial membrane potential, mitochondrial ROS production in T/G HA-VSMCs, and the effects of blocking IL-6 signalling via IL6R knockdown. *p<0.05, **p<0.01, ****p<0.0001. PBMC CM, peripheral blood mononuclear cells conditioned medium; ROS, reactive oxygen species; TAK, Takayasu’s arteritis; VSMCs, vascular smooth muscle cells.

We did find elongated and hyperfused mitochondrial network, a characteristic mitochondrial phenotype in senescent cell, in TAK-PBMC CM-treated VSMCs by TEM and MitoTracker Green staining (figure 4E,F, online supplemental figure S4B,C). In addition, TAK-PBMC CM-treatment led to other senescence-associated mitochondrial dysfunction, including disrupted MMP and increased mtROS production, revealed by JC-1 staining (increased monomeric JC-1 and reduced aggregate JC-1) and mitoSOX staining, respectively (figure 4G,H, online supplemental figure S4D,E). More importantly, blocking IL-6 signalling via receptor knockdown or anti-IL6R neutralising antibody suppressed TAK PBMC CM-induced senescence-associated mitochondrial dysfunction in VSMCs (figure 4E–H, online supplemental figure S4B–E). Stimulation with recombinant human IL-6 is also sufficient to induce mitochondrial elongation and enhanced mtROS generation in primary VSMCs from patients with TAK (online supplemental figure S4F,G). Taken together, in the context of TAK, IL-6 signalling can play a central role in the induction of senescence-associated mitochondrial dysfunction in VSMCs.

IL-6-induced prosenescent noncanonical mitochondrial-localised phosphorylated STAT3 (Tyr705) in VSMCs of TAK

We further investigated the mechanism by which IL-6 signalling contributes to VSMCs senescence. The activation of STAT3 via phosphorylation is the most classical molecular events downstream of IL-6 signalling. Indeed, increased phosphorylation of STAT3 at Tyr705 (but not at Ser727) were detected in IL-6-treated primary VSMCs from patients with TAK (figure 5A, online supplemental figure S5A).

Figure 5

IL-6-induced prosenescent mitochondrial-localised Tyr705 phosphorylated-STAT3 and mitochondrial elongation in VSMCs of TAK. (A) The influence of IL-6 treatment on STAT3 phosphorylation in Tyr705 (pT705) and Ser727 (pS727), the expression of GRIM19 and mitochondrial fusion-related protein (MFN1, MFN2, OPA1) in primary VSMCs from patients with TAK, and the effect of ERK1/2 inhibitor, JAK1/2 inhibitor, JAK1/3 inhibitor or pan-PI3K inhibitor treatment. (B) Fluorescence images showing the expression of pT705/pS727 STAT3 and HSP60 in untreated and IL-6-treated primary VSMCs from patients with TAK. (C) The effect of human IL-6 stimulation (100 ng/mL) on (Tyr705)/(Ser727) phosphorylated-STAT3 levels in mitochondrial, nuclear and WCL of primary VSMCs from patients with TAK. (D) p16 expression in primary VSMCs treated with IL-6 at various concentrations and the effects of additional STAT3 inhibition (STAT3-IN-3), specific inhibition of STAT3 phosphorylation in Tyr705 (cryptotanshinone), MFN2 inhibition (MFI8) or mitochondrial STAT3 inhibition (mitoCur-1). (E) SA-β-gal staining of IL-6-treated primary VSMCs from patients with TAK and the effects of additional mitoCur-1, MFI8 or cryptotanshinone treatment. (F) Fluorescence images showing the expression of MFN2 and HSP60 in untreated and IL-6-treated primary VSMCs from patients with TAK, and the effects of mitoCur-1 or cryptotanshinone treatment. ns, non-significant, *p<0.05, **p<0.01, ****p<0.0001. MFN1, mitofusin 1; MFN2, mitofusin 2; OPA1, optic atrophy protein 1; TAK, Takayasu’s arteritis; VSMCs, vascular smooth muscle cells; WCL, whole cell lysates.

It is well known that phosphorylation status controls the subcellular location of STAT3. Conventionally, phosphorylation at serine 727 is associated with its mitochondrial location while phosphorylation at tyrosine 705 is related to its nuclear location. Interestingly, unlike previous studies, IL-6 stimulation led to the colocalisation between phosphorylated STAT3 (Tyr705) and mitochondrial matrix marker HSP60 in primary VSMCs from patients with TAK at different time points (figure 5B, online supplemental figure S5B), suggesting its mitochondrial localisation. Using biochemical fractionation of IL-6-treated primary VSMCs, we further confirmed mitochondrial-specific enrichment of phosphorylated STAT3 (Tyr705) after IL-6 stimulation (figure 5C). Also, the expression level of gene associated with retinoic and interferon-induced mortality 19 protein (GRIM19), an electron transport chain component that mediates STAT3 import into mitochondria, was significantly elevated on IL-6 treatment (figure 5A). In addition, mitochondrial-targeted STAT3 inhibitor mitoCur-1, a TPP-based curcumin analogue, was used to investigate the association between mitochondrial STAT3 and IL-6-induced VSMCs senescence in vitro. Treatment with mitoCur-1 or cryptotanshinone (specific inhibitor of STAT3 phosphorylation at Tyr705) both significantly downregulated p16 expression and SA-β-gal staining intensity in IL-6-stimulated primary VSMCs from patients with TAK (figure 5D,E). Collectively, these findings suggested that IL-6-induced mitochondrial-localised phosphorylated STAT3 (Tyr705) exerted prosenescent effects on VSMCs of TAK.

Classic downstream signals of the IL-6/IL-6R signalling pathway include: JAK/STAT pathway, MAPK/ERK pathway and PI3K/Akt/mTORC1 pathway.16 We sought to explore whether single use of JAK inhibitor, ERK inhibitor or PI3K inhibitor can block the STAT3 phosphorylation at Tyr705 in IL-6-treated primary VSMCs. Notably, JAK1/2 inhibitor ruxolitinib, JAK1/3 inhibitor tofacitinib, ERK1/2 inhibitor SCH772984 or PI3K inhibitor TG100713 had no or only limited effects in vitro (figure 5A, online supplemental figure S5A), suggesting that these treatments could not act as surrogate measures of specific STAT3 inhibition at Tyr705 or mitochondrial STAT3 inhibition.

IL-6-induced mitochondrial phosphorylated STAT3 (Tyr705) prevented MFN2 from proteasomal degradation and promoted senescence-associated mitochondrial dysfunctions in VSMCs of TAK

Along with its antisenescent effects, mitoCur-1 or cryptotanshinone treatment also led to more fragmented mitochondrial network (often seen in young cells) in IL-6-stimulated primary VSMCs from patients with TAK (figure 5F, online supplemental figure S5C). This led us to address whether phosphorylated STAT3 (Tyr705) regulates the expression of mitochondrial fusion proteins including mitofusin 1/2 (MFN1/2) and OPA1 and promotes mitochondrial hyperfusion. We first investigated the possibility that mitochondrial fusion proteins were involved in IL-6-induced primary VSMCs senescence. Upregulated protein expression of MFN2, but not MFN1 and OPA1, was found in IL-6-treated primary VSMCs (figure 5A, online supplemental figure S5A). Notably, like mitoCur-1, treatment with MFN2 inhibitor MFI8 suppressed IL-6-induced mitochondrial elongation, mitochondrial ROS production and VSMCs senescence (online supplemental figure S5C,D, figure 5D,E). These results together indicated the critical role of MFN2 in IL-6-induced primary VSMCs senescence.

We then proceed to explore whether mitochondrial phosphorylated STAT3 (Tyr705) interacts with MFN2 to drive IL-6-induced primary VSMCs senescence. Compared with vascular samples from control individuals, upregulated expression and non-nuclear localisation of phosphorylated STAT3 (Tyr705), as well as increased MFN2 expression coemerged in vascular lesion of TAK (online supplemental figure S6A). Moreover, mitoCur-1 or cryptotanshinone significantly reduced protein expression levels of MFN2 in IL-6 treated primary VSMCs (figure 5F) while there was no detectable difference in the Mfn2 mRNA levels in primary VSMCs across treatment groups (online supplemental figure S6B), suggesting a post-transcriptional or post-translational effect of mitochondrial STAT3 on MFN2 expression levels, for example, by direct binding. To probe this possibility further, the binding mode between phosphorylated STAT3 (Tyr705) and MFN2 were predicted using protein–protein docking (online supplemental figure S6C,D). The physical interaction between phosp