Mycobacterium bovis Mb3523c protein regulates host ferroptosis via chaperone-mediated autophagy

ABSTRACT

The occurrence of necrosis during Mycobacterium bovis (M. bovis) infection is regarded as harmful to the host because it promotes the spread of M. bovis. Ferroptosis is a controlled type of cell death that occurs when there is an excessive buildup of both free iron and harmful lipid peroxides. Here, we demonstrate that the mammalian cell entry (Mce) 4 family protein Mb3523c triggers ferroptosis to promote M. bovis pathogenicity and dissemination. Mechanistically, Mb3523c, through its Y237 and G241 site, interacts with host HSP90 protein to stabilize the LAMP2A on the lysosome to promote the chaperone-mediated autophagy (CMA) pathway. Then, GPX4 is delivered to lysosomes for destruction via the CMA pathway, eventually inducing ferroptosis to promote M. bovis transmission. In summary, our findings offer novel insights into the molecular mechanisms of pathogen-induced ferroptosis, demonstrating that targeting the GPX4-dependent ferroptosis through blocking the M. bovis Mb3523c-host HSP90 interface represents a potential therapeutic strategy for tuberculosis (TB).

Abbreviations: CFU: colony-forming units; CMA: chaperone-mediated autophagy; Co-IP: co-immunoprecipitation; Fer-1: ferrostatin-1; GPX4: glutathione peroxidase 4; HSP90: heat shock protein 90; LDH: lactate dehydrogenase; Mce: mammalian cell entry; MOI: multiplicity of infection; Nec-1: necrostatin-1; PI: propidium iodide; RCD: regulated cell death.

Introduction

Mycobacterium bovis (M. bovis), together with Mycobacterium tuberculosis (Mtb), are composed of the important etiological agents of human tuberculosis, around 10%-15% of human tuberculosis cases worldwide are attributed to infections by M. bovis [1]. It is responsible for approximately $3 billion in global economic losses annually owing to cattle tuberculosis [2]. Furthermore, Same as Mtb as an intracellular pathogen, M. bovis has evolved multiple strategies to modulate host cell functions, facilitating its survival in the hostile intracellular environment of the host. Therefore, a deeper understanding of the molecular mechanisms underlying the interactions between M. bovis and host cells is critical for identifying novel therapeutic targets.

Cell death commonly ensues during pathogenic infections, serving either to aid the host in eliminating the pathogen or to be exploited by the pathogen to enhance its virulence [3,4]. Ferroptosis, a novel form of programmed cell death characterized by the accumulation of iron-dependent lipid peroxides, plays a pivotal role in various infectious diseases [5–7]. Recent evidence indicates that Mtb can induce ferroptosis in macrophages, thereby promoting the progression of tuberculosis (TB) [8]. However, the specific mechanisms and effectors involved are still largely unknown.

The mammalian cell entry (Mce) protein family in mycobacteria is a conserved group of cell wall proteins typically encoded by four mce operons (mce1–4), with the notable absence of the Mce3 operon in M. bovis [9–12]. Previous studies have demonstrated that these Mce family proteins, encoded by the mce operons, exhibit varying expression profiles during different stages of mycobacterial infection, suggesting their diverse regulatory roles in the infection process [13–15]. For instance, peptides targeting mce4F have been shown to inhibit the invasion of alveolar epithelial cells and macrophages by Mtb, indicating a correlation between mce4F and the pathogenicity of mycobacteria [16]. The Mb3523c protein, assumed to be rich in proline and valine, is a member of the Mce4 family and shares 100% homology with the Rv3493c protein of the Mtb strain H37Rv [17]. The specific role of the Mb3523c protein in modulating host cell functions remains largely unexplored. Here, we demonstrate that the Mb3523c protein can regulate chaperone-mediated autophagy (CMA) through interaction with HSP90, promoting GPX4 degradation and thus inducing ferroptosis, leading to the promotion of pathogen pathogenicity and dissemination.

Together, this study provides novel insights into the molecular mechanism underlying Mb3523c -mediated host ferroptosis during M. bovis infection, and suggests a potential Mtb-host interface-based TB treatment.

Results

Virulent M. bovis promotes host cell necrosis in vivo

In order to examine the processes via which M. bovis causes necrosis in the host, we investigated the alterations in mice after being infected with two different strains of M. bovis. M. bovis N# strain was first isolated in 2015 and it caused meningoencephalitis in a few calves [18,19]. M. bovis strain C68004# was first isolated in the 1950s and kept in a laboratory. Preliminary investigations revealed that, compared with M. bovis C68004#-infected mice, those infected with M. bovis N# showed a significant increase in bacterial burden in both the lungs (no significant change at day 14; ~1.07-fold increase at day 28) and spleen (no significant change at day 14; ~1.15-fold increase at day 28) (Figures 1A,B). Simultaneously, a significant increase in the production of proinflammatory cytokines, including IL1B, TNF, and IFNG, was observed in M. bovis N#-infected mice, particularly 28 days post-infection (Figure 1C-E). Gross pathology revealed that M. bovis N#-infected mice exhibited more gray-white nodular lesions in the lungs and splenomegaly, most pronounced at 28 days post-infection (Figure 1F). Ziehl-Neelsen staining subsequently showed that the number of M. bovis N# in the lungs was significantly higher than that of M. bovis C68004#, especially at 28 days post-infection (Figure 1G). Consistent with gross pathology findings, histopathological examination revealed more severe inflammatory lesions, necrosis, and calcification in the lungs of M. bovis N#-infected mice, particularly 28 days post-infection (Figure 1H,I). To further validate our findings, lung cell suspensions were isolated from mice, and necrosis levels caused by different strains were assessed using ANXA5 staining. Results indicated that lung cell necrosis was more extensive in M. bovis N#-infected mice compared to those infected with M. bovis C68004#, particularly at 28 days post-infection (Figure 1J-K).

Figure 1.

M. bovis N# promote M. bovis pathogenicity and dissemination in vivo. C57BL/6 mice were infected with the M. bovis N# or M. bovis C68004# strain for 14 or 28 days. (A) CFU analysis for mycobacterial survival in the lungs of mice. (B) CFU analysis for mycobacterial survival in the spleen of mice. (C) ELISA detected the alteration of IFNG, (D) TNF, and (E) IL1B in serum from mice. (F) the representative images of lung and spleen showed the gross pathological changes of all experimental groups (n = 6). (G) Z&N staining results showing the number of M. bovis in lungs of mice (H) Higher magnification of H&E staining sections of lung showed M. bovis induced lesions. Scale bar: 100 μm. (I) the percentage of lung’s area occupied by inflammatory lesions was quantified by ImageJ software. (J) the lung cell which isolated from the lungs of different group of mice were treated with ANXA5 and detected by flow cytometry. (K) Quantification of the necrosis cell in lung cell from mice. The results shown in these panels are means ± SE of data obtained from three independent experiments. Statistically significant differences were obtained for the indicated groups (paired two-tailed t test, *p < 0.05, **p < .01, ***p < .001).

In conclusion, our findings suggest that M. bovis N# is more likely to induce lung necrosis in mice.

Virulent M. bovis suppresses cell apoptosis and promotes cell necroptosis and ferroptosis

In recent years, RCD (regulated cell death) has been recognized as signaling cascades involving effector molecules and has unique biochemical, morphological, and immunological consequences [20]. RCD can be regulated, in part, by inhibiting the transduction of lethal signals, as well as by improving the ability of cells to make adaptive responses to stress [21]. RCD takes many forms depending on the inducing conditions including necroptosis, pyroptosis, and ferroptosis, and others [22]. Initially, to examine whether M. bovis-induced RCD, we employed a model in which RAW264.7 cell were exposed to M. bovisN# and M. bovis C68004# strain at MOI (multiplicities of infection) of 10 and cell death evaluated by using Live/Dead staining (which selectively detects cells undergoing cell death) and LDH (lactate dehydrogenase) release 24 h post infection. Following infection with M. bovisN#, RAW264.7 cell cultures showed an elevated percentage of propidium iodide (PI)-staining-positive cells (Figure 2A), as well as enhanced LDH release (Figure 2B). In addition, consistent with previous result, M. bovis N# strain induced more proinflammatory cytokines, including IL1B, TNF in RAW264.7 cell (Figure 2C,D). Next, in order to determine which pathway is responsible for the induction of cell death caused by M. bovis, we detected apoptosis firstly and found M. bovis N# strain suppress apoptosis compared with M. bovis C68004# strain (Figure S1A). Then we examined the GPX4 change usually happened for ferroptosis and p-MLKL for necroptosis. We confirmed that M. bovis N# strain significantly suppressed the expression of GPX4 and promoted the phosphorylation of MLKL (Figure 2E). However, we observed that infection of macrophages with both M. bovis C68004# and M. bovis N# induced pyroptosis, but with no significant difference between the strains (Figure S1B). Consistently, this phenomenon was also observed in vivo (Figure 2F). To further validate our conclusions, we treated cells with 10 μM ferrostatin-1 (Fer-1), a ferroptosis inhibitor, 40 μM necrostatin-1 (Nec-1), a necroptosis inhibitor, and 50 μM VX-765, a pyroptosis inhibitor, which significantly reduced both the percentage of cell necrosis and the amount of LDH released.

Figure 2.

M. bovis N # induces cell necrosis. (A) RAW264.7 macrophage cell were uninfected (CT) or infected with M. bovis N# and M. bovis C68004# at a MOI of 10 for 24 h. M. bovis-induced macrophage necrosis in vitro measured by Live/Dead staining; (B) Cells were uninfected or infected with M. bovis N# and M. bovis C68004# at a MOI of 10 for 24 h. LDH release was measured in the supernatants from the macrophage cultures. ELISA detected the alteration of IL1B (C) and (D) TNF in cell supernatant. The data reported in A – D are representative results from at least six independent experiments performed. (E) Western blot detection and the relative ratio of p-mlkl, and GPX4 in cells of different group. GAPDH, TUBA/and MLKL expression served as a cell control and was used for normalization (n = 3/group). (F) Western blot detection and the relative ratio of p-mlkl, and GPX4 in lung of different group of mice. GAPDH, TUBA and MLKL expression served as a lung control and was used for normalization (n = 3/group). (G) RAW264.7 macrophage cell were uninfected (CT) or infected with M. bovis N# and M. bovis C68004# at a MOI of 10 for 24 h. With the treatment of vehicle, Fer-1 (10 μM), VX-765 (50 μM) or Nec-1 (40 μM). M. bovis-induced macrophage necrosis in vitro was measured by Live/Dead staining; (H) RAW264.7 macrophage cell were uninfected (CT) or infected with M. bovis N# and M. bovis C68004# at a MOI of 10 for 24 h. With the treatment of vehicle, Fer-1(10 μM), VX-765 (50 μM) or Nec-1 (40 μM). LDH release was measured in the supernatants from the macrophage cultures. The results shown in these panels are means ± SE of data obtained from three independent experiments. Statistically significant differences were obtained for the indicated groups (paired two-tailed t test, *p < 0.05, **p < .01, ***p < .001).

Although treating cells with pyroptosis inhibitors partially reduced cell death, though this effect was less pronounced than with necroptosis or ferroptosis inhibitors alone (Figure 2G,H). In general, M. bovis N# can cause cell death through several pathways, thereby promoting the spread of the pathogen systematically.

RCD induced by M. bovis Mb3523c protein is dependent on ferroptosis

To elucidate the specific mechanisms by which M. bovis N# induces cell RCD. In our previous study, we utilized tandem mass spectrometry (TMT) for quantitative proteomics technology to identify and analyze the differently expressed proteins of the two strains. The findings revealed the identification of 2,174 common proteins, out of which 479 proteins exhibited differential expression (p < 0.05). Additional study revealed that M. bovis N# exhibited an increase in the expression of 250 proteins compared to M. bovis C68004#. Among these proteins, Mb3523c showed the highest level of upregulation (Figure S2) [23]. To examine Mb3523c-induced necrosis in vitro, we used BCG and BCG: Mb3523c at various MOIs and time points to evaluate necrosis levels. After 24 hours of BCG: Mb3523c infection at a MOI of 10, RAW264.7 cells exhibited the highest percentage of PI-positive cells (Figures 3A,3C) and the strongest LDH release (Figures 3B,3D). Next, to identify the pathway of Mb3523C-induced necrosis, we treated infected cells with Fer-1 and Nec-1. Although both inhibitors significantly reduced cell necrosis levels, necrosis in BCG: Mb3523C-infected macrophages in the Nec-1 group remained significantly higher than in BCG-infected macrophages (Figures 3E,3F), implying that Mb3523C-induced necrosis is ferroptosis-dependent. To validate our findings, we used C11 BODIPY 581/591 probes to quantify lipid peroxide levels in cells post-infection with different strains. Cells infected with BCG: Mb3523c exhibited higher lipid peroxide accumulation. As expected, Mb3523c led to a greater generation of lipid peroxides (green fluorescence). This phenomenon disappeared after treatment with Fer-1 (Figure 3G-I). Next, we examined changes in ferroptosis-related protein GPX4 and Fe²⁺ levels. GPX4 expression was significantly inhibited in macrophages (Figure 3J), while Fe²⁺ accumulation increased (Figure S3A, S3B) after BCG: Mb3523c infection. Interestingly, p-MLKL levels were not significantly affected during this process (Figure 3K). Furthermore, BCG: Mb3523c infection of RAW264.7 cells led to an increased bacterial load and upregulated inflammatory cytokines, including IL1B and TNF (Figure 3L-N). To further validate our findings, we over-expression of Mb3523c in C68004# as C68004:Mb3523c. Consistent with the previous results, the PI-positive cells and LDH release increase in cells infected with M. bovis C68004#: Mb3523c than those infected with M. bovis C68004# (Figure S3C, S3D). GPX4 degradation are higher and more bacterial survival in cells treated with M. bovis C68004: Mb3523c (Figure S3E, S3F). These findings suggest that Mb3523c-induced cell death may depend on ferroptosis.

Figure 3.

Necrosis induced by Mb3523c protein is dependent on ferroptosis. (A-B) RAW264.7 macrophage cell were uninfected (CT) or infected with BCG and BCG: Mb3523c at a MOI of 10 for 6, 12 and 24 h. (A) Necrotic cell death measured by Live/Dead staining. (B) LDH release measured in supernatants from macrophage cultures. (C-D) RAW264.7 macrophage cell were uninfected (CT) or infected with BCG and BCG: Mb3523c at a MOI of 1,5, or 10 for 24 h. (C) Necrotic cell death measured by Live/Dead staining. (D) LDH release measured in supernatants from macrophage cultures. (E-F) RAW264.7 macrophage cell were uninfected (CT) or infected with BCG and BCG: Mb3523c at a MOI of 10 for 24 h. With the treatment of vehicle, Fer-1(10 μM), or Nec-1 (40 μM). (E) Necrotic cell death measured by Live/Dead staining. (F) LDH release measured in supernatants from macrophage cultures. (G) Confocal microscopic analysis for lipid peroxides using BODIPY C11 lipid probe in RAW264.7 cells. Cells infected with BCG and BCG: Mb3523c at a MOI of 10 for 24 h were treated with vehicle or 10 μM Fer-1 for 24 h, and were then treated with 1.5 μM BODIPYC11 lipid probe for additional 20 min. red and green fluorescence represent the non-oxidized and oxidized BODIPY C11, respectively. Scale bars: 10 μm. (H-I) Quantification of the fluorescence intensity of non-oxidized (H) and oxidized BODIPY C11 (I) in cells treated as in (G). (J) Immunoblot analysis of GPX4, and GAPDH in RAW264.7 cells infected with BCG and BCG: Mb3523c at a MOI of 10 for 24 h were treated with vehicle or 10 μM Fer-1 for 24 h. (K) Immunoblot analysis of pp-mlkl, MLKL and GAPDH in RAW264.7 cells infected with BCG and BCG: Mb3523c at a MOI of 10 for 24 h. (L-N) RAW264.7 cells infected with BCG and BCG: Mb3523c.(L) Colony-forming unit (CFU) analysis. (M) ELISA detected the alteration of IL1B. (N) ELISA detected the alteration of TNF. The results shown in these panels are means ± SE of data obtained from three independent experiments. Statistically significant differences were obtained for the indicated groups (paired two-tailed t test, *p < 0.05, **p < .01).

To further investigate the regulatory role of Mb3523c on ferroptosis during M. bovis infection, we deleted the gene encoding Mb3523c in M. bovis N# (ΔMb3523c) and complemented it with the wild-type (WT) M. bovis N# Mb3523c (ΔMb3523c: Mb3523c). Similarly, 24 hours after ΔMb3523c infection at a MOI of 10, RAW264.7 cells exhibited significantly lower cell death levels compared to those infected with the WT and complemented strains (Figure 4A-D). Further experiments revealed that Mb3523c-induced cell necrosis was significantly reduced after treatment with Fer-1instead of Nec-1 compared to other groups (Figures 4E,4F). The generation of lipid peroxides in cells infected with various strains at a MOI of 10 post infection 24 h was detected using the C11 BODIPY 581/591 probe. The ΔMb3523c strain showed a significant reduction in lipid peroxide production (indicated by green fluorescence) compared to M. bovis N# and ΔMb3523c: Mb3523c strains, no difference was observed between various strains after treatment with Fer-1 (Figure 4G-I). GPX4 expression was significantly upregulated when cells were infected with ΔMb3523c strain compared with M. bovis N# and ΔMb3523c: Mb3523c strains (Figure 4J), and Fe²⁺ accumulation was inhibited (Figure 4J, Figure S3G, S3H), with no change observed in p-MLKL levels (Figure 4K). This phenomenon disappeared after treatment with Fer-1. Meanwhile, RAW264.7 cells infected with ΔMb3523c showed a significant reduction in bacterial load (Figure 4L) and expression of inflammatory cytokines, including IL1B and TNF (Figure 4M,N) compared to those infected with the WT and complemented strains.

Figure 4.

Ferroptosis did not arise following the knockdown of Mb3523c. (A-B) RAW264.7 macrophage cell were uninfected (CT) or infected with WT M. bovis N#, ΔMb3523c and ΔMb3523c: Mb3523c strains at an MOI of 10 for 6,12 and 24 h. (A) Necrotic cell death measured by Live/Dead staining. (B) LDH release measured in supernatants from macrophage cultures. (C-D) RAW264.7 macrophage cell were uninfected or infected with WT M. bovis N#, ΔMb3523c and ΔMb3523c: Mb3523c strains at a MOI of 1, 5, or 10 for 24 h. (C) Necrotic cell death measured by Live/Dead staining. (D) LDH release measured in supernatants from macrophage cultures. (E-F) RAW264.7 macrophage cell were uninfected (CT) or infected with WT M. bovis N#, ΔMb3523c and ΔMb3523: Mb3523c strains at a MOI of 10 for 24 h. With the treatment of vehicle, Fer-1(10 μM), or Nec-1 (40 μM). (E) Necrotic cell death measured by Live/Dead staining. (F) LDH release measured in supernatants from macrophage cultures. (G) Confocal microscopic analysis for lipid peroxides using BODIPY C11 lipid probe in RAW264.7 cells. Cells infected with WT M. bovis N#, ΔMb3523c and ΔMb3523c:Mb3523c strains at a MOI of 10 for 24 h were treated with vehicle or 10 μM Fer-1 for 24 h, and were then treated with 1.5 μM BODIPYC11 lipid probe for additional 20 min. red and green fluorescence represent the non-oxidized and oxidized BODIPY C11, respectively. Scale bars: 10 μm. (H-I) Quantification of the fluorescence intensity of oxidized (H) and non- oxidized BODIPY C11 (I) in cells treated as in (G).(J) Immunoblot analysis of GPX4, and GAPDH in RAW264.7 cells infected with WT M. bovis N#, ΔMb3523c and ΔMb3523c:Mb3523c strains at a MOI of 10 for 24 h were treated with vehicle or 10 μM Fer-1 for 24 h. (K) Immunoblot analysis of p-mlkl, MLKL and GAPDH in RAW264.7 cells infected with WT M. bovis N#, ΔMb3523c and ΔMb3523c:Mb3523c strains at a MOI of 10 for 24 h. (L-N) RAW264.7 cells infected with WT M. bovis N#, ΔMb3523c and ΔMb3523c:Mb3523c strains. (L) Colony-forming unit (CFU) analysis for mycobacterial survival. (M) ELISA detected the alteration of IL1B. (N) ELISA detected the alteration of TNF. The results shown in these panels are means ± SE of data obtained from three independent experiments. Statistically significant differences were obtained for the indicated groups (paired two-tailed t test, *p < 0.05**p < .01, ***p < .001).

Similarly, the phenomenon disappeared after treatment with Fer-1. In summary, the cell death induced by Mb3523c was facilitated through the ferroptosis pathway.

M. bovis Mb3523c promotes GPX4 degradation by the lysosome pathway

It has been reported that host ferroptosis induced by mycobacteria is characterized by the suppression of GPX4 [24]. Therefore, we attempted to explore how Mb3523c affects the degradation of GPX4. We conducted RNA sequencing (RNA-seq) analysis on RAW264.7 cells infected with M. bovis N# and ΔMb3523c, identifying 1414 differentially expressed genes. Infection with the ΔMb3523c strain in RAW264.7 cells led to the overexpression of 589 genes and the downregulation of 825 genes compared to the WT M. bovis N# infection (Figure 5A).

Figure 5.

M. bovis Mb3523c promote GPX4 degradation by lysosome pathway. (A) Volcano plot for differentially expressed genes in RAW264.7 cells infected with WT M. bovis N# and ΔMb3523c strain. The red plot represents up-regulated genes (fold change ≥ 0.25 and p value < 0.05), and the green plot represents down-regulated genes (fold change ≤ −0.25 and p value < 0.05). (B) rt-qPCR analysis of GPX4 in RAW264.7–derived macrophages infected with M. bovis N#, ΔMb3523c or ΔMb3523c:Mb3523c. (C) the list of the top 11 enriched gene sets. (D) Immunoblot analysis of LAMP1, and GAPDH in RAW264.7 cells infected with M. bovis N#, ΔMb3523c and ΔMb3523c:Mb3523c strains at a MOI of 10 for 24 h. (E-F) RAW264.7 were treated with 12.5 μM CQ (E) or 200 nM MG-132 (F) for 3 h prior to infection, then were infected with M. bovis N#, ΔMb3523c and ΔMb3523c:Mb3523c strains at a MOI of 10 and treated with or without CQ,MG-132, CQ or MG-132 for 24 h. (E) western blot detection and the relative ratio of GPX4 in cells of different group. GAPDH expression served as a cell control and was used for normalization (n = 3/group). (F) Western blot detection and the relative ratio of GPX4 in cells of different group. GAPDH expression served as a cell control and was used for normalization (n = 3/group).

However, the infection with different strains did not alter GPX4 mRNA levels (Figure 5B). KEGG enrichment analysis of significantly upregulated genes in the WT M. bovis infection group revealed that deletion of the Mb3523c gene significantly downregulated genes involved in the lysosomal degradation pathway (Figure 5C).We then verified the expression of those genes in RAW264.7 cells infected with WT M. bovis N#, ΔMb3523c and ΔMb3523c: Mb3523c strain, and confirmed that Mb3523 significantly promoted the expression of lysosome-regulating genes including LAMP1 (lysosomal-associated membrane protein 1) (Figure 5D).

Next, to verify whether the degradation of GPX4 mediated by Mb3523c is accomplished through the lysosomal pathway alone, we treated cells with the lysosomal pathway inhibitor CQ and the proteasome pathway inhibitor MG-132, respectively. The degradation of GPX4 mediated by Mb3523c was significantly alleviated after inhibiting the lysosomal pathway, but GPX4 remained at a lower level after inhibiting the proteasome pathway (Figures 5E, 5F). Overall, the degradation of GPX4 induced by Mb3523c is associated with the lysosomal pathway.

Mb3523c protein induces chaperone-mediated autophagy through interaction with HSP90

To examine the process by which Mb3523c causes the degradation of GPX4 through the control of the lysosomal pathway, our initial step is to confirm whether Mb3523c can interact with GPX4 and trigger its degradation. No evidence of interaction between GPX4 and Mb3523c was found by the co-immunoprecipitation (co-IP) experiment (Figure 6B). Next, we endeavored to express the Mb3523c protein in HEK293T cells and employed immunoprecipitation-mass spectrometry (IP-MS) techniques to identify potential targets influencing GPX4 degradation. The IP-MS experiments revealed 111 proteins specifically interacting with Mb3523c (Figure 6A and Table S2 and Table S3). Notably, we discovered an interaction between HSP90 and Mb3523c. Recent studies have shown that HSP90 regulates the degradation of GPX4 through chaperone-mediated autophagy (CMA) during ferroptosis. To verify whether the degradation of GPX4 by Mb3523c is associated with HSP90, we examined their interaction via co-IP, finding that Mb3523c interacts with HSP90 (Figure 6B). The colocalization of Mb2523c and HSP90 was observed after plasmid transfection by confocal microscopy (Figure 6C). Despite discovering an interaction between HSP90 and Mb3523c, we have yet to elucidate the mechanism by which HSP90 facilitates GPX4 degradation. To identify this issue, we employed WT M. bovis and ΔMb3523c to infect macrophages and detect the changes of HSP90. The western blot analysis revealed that the absence of Mb3523c protein had a substantial inhibitory effect on the expression of HSP90 (Figure 6D). Furthermore, the suppression of GPX4 produced by M. bovis infection was reversed when cells were treated with the HSP90 inhibitor tanespimycin or transfection with HSP90 siRNA (Figure 6D and Figure S4A). Next, we extracted cellular lysosomes and observed a significant decrease in the GPX4 concentration within the lysosomes that had undergone Mb3523c deletion. Additionally, lysosomes isolated from transfection with HSP90 siRNA cells showed a significant decrease in GPX4 levels. These results suggest that HSP90 is crucial in the CMA pathway mediated by mycobacterium bovis protein Mb3523c (Figure 6E).

Figure 6.

Mb3523c protein induces chaperonin-mediated autophagy through interaction with HSP90. (A) Mass spectrometric analysis of the proteins coimmunoprecipitated by anti-ha antibody. (B) Immunoblot analysis and immunoprecipitation of HA-Mb3523c, GPX4 and HSP90 in HEK293T cells. (C) Confocal microscopy analysis of HSP90 (red) colocalization with HA-Mb3523c (green) in HEK293T cells. (D, F) RAW264.7 were treated with 0.5 μM tanespimycin or 30 μM apoptozole for 3 h prior to infection, then were infected with M. bovis N#, ΔMb3523c and ΔMb3523c:Mb3523c strains at a MOI of 10 and treated with or without tanespimycin, apoptozole, tanespimycin or apoptozole for 24 h. (D) western blot detection and the relative ratio of GPX4, HSP90 in cells of different group. GAPDH expression served as a cell control and was used for normalization (n = 3/group). (E) Expression of GPX4 in the cytoplasm or lysosome was determined by western blot. GAPDH expression served as a lysosome-free cell lysate control and LAMP1 expression served as a lysosome control were used for normalization (n = 3/group). (F) western blot detection and the relative ratio of HSPA8, LAMP2A, and GPX4 in cells of different group. GAPDH expression served as a cell control and was used for normalization (n = 3/group). (G) A549 cells, which were pre-transfected with KFERQ-PA-mCherry, were infected with M. bovis N#, ΔMb3523c and ΔMb3523c:Mb3523c strains at a MOI of 10 for 24 h. Cells were observed under a confocal microscope. The scale bar indicates 5 µm. The white arrows represent each red punctum. (H) Quantification of KFERQ-PA-mCherry expression in A549 cells treated as in (G). (I) A549 cells were transfected or co-transfected with HA-Mb3523c or HA-Mb3523c Δ1–97 and KFERQ-PA-mCherry for 36 h. Cells were immunostained with anti-ha antibody, and then observed under a confocal microscope. The white arrows represent each red punctum. (J) Quantification of KFERQ-PA-mCherry expression in A549 cells treated as in (I). The scale bar indicates 5 µm. Error bars: mean ± SD of 3 independent tests. Student’s t-test; ns: non-significance; *p <.05; **p < 0.01.

Meanwhile, confocal microscopy revealed a significant reduction in the colocalization of GPX4 and LAMP2A following the deletion of the Mb3523c (Figure S4B). Subsequently, we further explored the levels of key CMA proteins in M. bovis-infected cells. As shown in Figure 6F, M. bovis N# infection increased the protein levels of LAMP2A and HSPA8. However, HSPA8 and LAMP2A expression levels significantly decreased following ΔMb3523c infection. Simultaneously, we observed that the reduction in GPX4 levels caused by M. bovis N# infection was significantly mitigated when 30 μM apoptozole, a CMA inhibitor, was administered (Figure 6F). Additionally, HSP90 siRNA cells showed a significant decrease in LAMP2A levels (Figure S4A). KFERQ-PA-mCherry is initially nonfluorescent but exhibits red fluorescence upon irradiation with 405 nm visible light, enabling visualization of lysosomes as red fluorescent puncta. The number of red puncta per cell serves as a reliable indicator of CMA activity. As shown in (Figure 6G-J), the number of red fluorescent puncta increased in A549 cells infected with M. bovis N# or transfected with Mb3523c but not infected with ΔMb3523c.

In summary, Mb3523c modulates GPX4 degradation through the CMA pathway by interacting with HSP90 protein.

Mb3523c protein interacts with HSP90 via amino acid sites Y237 and G241 in the region of 98–242

Next, we wanted to explore the key region or amino acid (aa) site where Mb3523c interacts with HSP90 protein. Initially, the AlphaFold2 online server (https://alphafold.ebi.ac.uk/) was employed for prediction of crystal structure of Mb3523c. It was found that aa1–74 was inside the cell membrane, aa75–97 was anchored to the bacterial cell membrane like a “hammer handle” and aa98–242 was mainly composed of α-helix structure, which formed a folding at the end of the C-terminus and resembled the “hammer head”. During the sway of the “hammer head”, HSP90 protein was recruited to play its function, the details of which are illustrated in Figure S5A and Figure 7A,C. To test this conjecture, we constructed three truncated mutants of Mb3523c, together with the WT, were used to transfect 293T cells, the lysates of which were analyzed by co-IP assays. Only those constructs containing aa98–242 were able to coimmunoprecipitate with the HSP90 protein, while those lacking this domain failed to coimmunoprecipitate, suggesting that Mb3523c interacts with HSP90 through its amino acid regions 98–242 which mainly locate in the outer of membrane (Figure 7B). In addition, as shown in Figure 6I-J, the number of red fluorescent puncta increased in A549 cells transfected with Mb3523c but not transfected with Mb3523c aa1–97.

Figure 7.

The G241 and Y237 of Mb3523c protein are the key sites for interaction with HSP90. (A) Schematic diagram of domain structure of Mb3523c and construction strategy of gfp-tagged WT Mb3523c and its truncation mutants. (B) HEK293T cells were transfected with PCMV-GFP-Mb3523c WT or any one of its mutants for 24 h. The cells were analyzed by co-ip assays using anti-gfp magnetic beads, followed by western blot analyses. Representative results are shown and similar results were obtained in three independent experiments. (C)The predicted model of the Mb3523c-HSP90 interaction complex was obtained using the AlphaFold2 online server. The HSP90 (left) and Mb3523c (right) proteins are marked in blue and pink respectively. The amino acid residues predicted to be possible interaction sites are labeled with one-letter type name in stick representation. (D) HEK293T cells were transfected with HA-Mb3523c, Vector or any one of the six mutants with a single mutation site of N158A, N162A, Y216A, L235A, Y237A or G241A. The cells were analyzed by co-ip assays using anti-ha magnetic beads, followed by western blot analyses. Representative results are shown and similar results were obtained in three independent experiments.

Then, we further analyzed and predicted the binding sites between Mb3523c and HSP90, resulting in six predicted structural models of the Mb3523c-HSP90 complex. Specifically, the binding sites involved Mb3523c residues N158, N162, Y216, L235, Y237, and G241 interacting with HSP90 residues D118, N30, T31, Q23, N158, and Q188 (Figure 7C). To ascertain which of the six predicted amino acid residues in Mb3523c determines its interaction with HSP90, single amino acid substitutions at positions 158 (N→a), 162 (N→a), 216 (Y→a), 235 (L→a), 237 (Y→a), and 241 (G→a) were introduced to generate six Mb3523c mutants. These mutants were then analyzed for their ability to interact with HSP90 protein using co-IP. The results indicated that G241 and Y237 are two key residues in Mb3523c for the interaction with HSP90 (Figure 7D).

Finally, to verify the interaction between Mb3523c and HSP90 under natural infection conditions, we utilized FITC-labeled bacteria and observed that the WT M. bovis N# strain recruited more HSP90 around the bacteria, an effect not observed with the deletion of the Mb3523c protein. This suggests that the surface of M. bovis bacteria has the capacity to recruit HSP90 (Figure S5B).

In summary, the aa 98–242 region of Mb3523c protein is the key region of the interaction between Mb3523C and HSP90 protein. In addition, Y237 and G241 in this region are the key amino acid sites that determine the interaction between Mb3523C and HSP90 protein.

M. bovis Mb3523c induces ferroptosis to promote pathogen pathogenicity and dissemination in vivo

Ferroptosis typically drives tissue necrosis, aiding M. bovis pathogenicity and dissemination in host. Therefore, we evaluated the impact of Mb3523c-induced ferroptosis on pulmonary necrosis and M. bovis dissemination in C57BL/6 mice treated with a ferroptosis inhibitor (Fer-1) or an HSP90 inhibitor (tanespimycin). As expected, mice infected with M. bovis N# or the ΔMb3523c: Mb3523c strain exhibited a higher presence of gray-white nodular lesions in the lungs than those infected with ΔMb3523c (Figure S6A). Histopathological observations, consistent with gross pathological findings, indicated more severe inflammatory lesions, necrosis, and calcification in the lungs of mice infected with either M. bovis N# or the ΔMb3523c: Mb3523c strain (Figures 8A,B), as well as a higher mycobacterial load in the lungs and spleens compared to mice infected with ΔMb3523c (Figures 8C,D). Ziehl-Neelsen staining subsequently showed that the number of M. bovis N# and ΔMb3523c: Mb3523c in the lungs was significantly higher than that of ΔMb3523c (Figure S6B). Consistent with the in vitro data, mice infected with M. bovis N# or the ΔMb3523c: Mb3523c strain exhibited elevated levels of 4-hydroxy-2-noneal (4-HNE, a ferroptosis marker) and decreased GPX4 expression compared to those infected with the ΔMb3523c strain (Figure 8E-G). Simultaneously, treatment with Fer-1 and tanespimycin significantly reduced the pathological differences between M. bovis N# (or ΔMb3523c: Mb3523c)-infected mice and those infected with ΔMb3523c.

Figure 8.

M. bovis Mb3523c induces ferroptosis to promote M. bovis pathogenicity and dissemination in vivo. (A) Higher magnification of H&E staining sections of lung from mice uninfected or infected with M. bovis N#, ΔMb3523c and ΔMb3523c: Mb3523c strain for 4 weeks. Daily intraperitoneal injections of vehicle Fer-1 (3 mg/kg) or tanespimycin (5 mg/kg) were administered to mice from day 15 after M. bovis infection (n = 6). Scale bar: 100 μm (B) the percentage of lung’s area occupied by inflammatory lesions was quantified by ImageJ software treated as in (A). (C) M. bovis survival in the lungs from mice treated as in (A). (D) M. bovis survival in the spleen from mice treated as in (A). (E) Immunohistochemical staining of 4-HNE in lung sections from mice treated as in (A) Scale bars: 50 μm. (F) Quantification of 4-HNE expression in lung sections from mice treated as in(A). (G) Western blot detection and the relative ratio of HSP90, and GPX4 in lungs of different group of mice. GAPDH expression served as a lung control and was used for normalization (n = 3/group).

In summary, our study elucidates the detailed molecular mechanism by which Mb3523c induces host cell ferroptosis. Specifically, Mb3523c interacts with host cell HSP90 through the 241 G and 237Y sites, regulating CMA to suppress GPX4 expression, ultimately leading to ferroptosis-induced pathogenicity and dissemination of M. bovis (Figure 9).

Figure 9.

Schematic model shows that Mb3523c enhances HSP90-mediated CMA to promote ferroptosis-dependent pathogen pathogenicity and dissemination. Briefly, M. bovis surface protein Mb3523c interact with host HSP90, promoting lysosome to process the degradation of GPX4, followed by promoting ferroptosis-dependent pathogen pathogenicity and dissemination.

Discussion

Pathogenic microorganisms manipulate host innate immune responses through various mechanisms, which are key factors enabling their survival and dissemination within host cells.

Current research indicates that attenuated strains of Mtb induce apoptosis in infected cells, while virulent strains are more prone to induce necrosis. Necrosis is a critical factor contributing to the replication and spread of mycobacteria within the host [25,26]. The M. bovis N# strain was first isolated in 2015, exhibiting more virulent transmission and causing meningoencephalitis in a few calves. The M. bovis C68004# strain was first isolated in the 1950s and has been maintained in the laboratory. Our previous studies demonstrated that the M. bovis N# strain infected mice, leading to higher serum levels of IFNG, TNF, and IL22 [18,19]. In this article, we reveal that, compared to the M. bovis C68004# strain, the M. bovis N# strain more readily induces necrosis and inhibits apoptosis in the host. Since necrosis is often accompanied by the expression of proinflammatory cytokines, this also explains why M. bovis N# induces higher pro-inflammatory cytokines production. Overall, this outcome explains the greater pathogenicity of the M. bovis N# strain.

To investigate how the M. bovis N# strain induces necrosis, proteomic analysis was conducted to elucidate the differential protein expression between M. bovis N# and M. bovis C68004# strains. Among the upregulated proteins, Mb3523c showed the highest fold increase. Subsequent gene knockout experiments revealed that Mb3523c specifically induces ferroptosis without affecting necroptosis. This study aims to elucidate the specific mechanisms through which Mb3523c induces necrosis via ferroptosis. The pathway by which M. bovis N# specifically regulates necroptosis remains a subject for future research. Given the complexity of mycobacterial pathogenicity, we hypothesize that multiple pathways may regulate ferroptosis in mycobacteria. For instance, recent studies have found that Mtb-induced macrophage necrosis is associated with decreased glutathione and GPX4 levels, alongside increased free iron, mitochondrial superoxide, and lipid peroxidation-key features of ferroptosis [27]. Further research revealed that PtpA in Mtb targets PRMT6 (protein arginine methyltransferase 6), enhancing the asymmetric dimethylation of histone H3 arginine 2 (H3R2me2a), thereby inhibiting GPX4 expression and ultimately inducing ferroptosis [28]. Additionally, Dai et al. found that Mtb infection enhanced NCOA4-mediated ferritin degradation via MAPK/p38-AKT1 and TRIM21-mediated HERC2 proteasomal degradation, which in turn increased iron bioavailability to intracellular Mtb and promoted bacterial survival in vivo [29]; In addition, Mb3523c may also regulate ferroptosis through other pathways. Recent studies have indicated that the transcription factor BACH1 plays a significant role in the regulation of ferroptosis during Mtb infection by enhancing the induction of several antioxidant genes [30]. Interestingly, this study demonstrated that in the absence of BACH1, the mRNA expression of GPX4 was not significantly affected, but the protein level was markedly increased, a finding consistent with our observations. Although we currently lack evidence to suggest that Mb3523c directly affects BACH1, the possibility that Mb3523c indirectly influences BACH1 to regulate ferroptosis cannot be excluded.

Those lines of research have greatly elucidated the regulatory mechanism of Mtb in inducing host ferroptosis. However, as an adept pathogen, Mtb can evade the immune system through various mechanisms. Our findings enhance the understanding of mycobacterial dissemination mechanisms within the host and provide new insights into the pathways through which mycobacteria induce ferroptosis.

It has been reported that mycobacteria-induced host ferroptosis is characterized by the suppression of GPX4. The use of the ferroptosis inhibitor Fer-1 can mitigate this cell death process [31], and our results also support this conclusion. Consequently, how mycobacteria manipulate GPX4 expression has emerged as a pivotal factor in inducing ferroptosis. Through mass spectrometry analysis, we have unveiled, for the first time, an interaction between HSP90 and Mb3523c. HSP90 is a highly conserved molecular chaperone that is widely present in eukaryotes and some prokaryotes [32,33]. HSP90 plays a crucial role in cells, particularly in helping other proteins fold properly and stabilize [34–36]. Although most studies have reported that CMA delivers selected protein substrates with a pentapeptide CMA-targeting motif to lysosomes by binding to its chaperone HSPA8 and interacting with LAMP2A [37–39], growing evidence suggests that HSP90 also plays an important role in CMA [40,41]. For instance, under oxidative stress conditions, such as treatment with the ferroptosis inducer erastin or glutamate, the interaction between HSP90 and LAMP2A is markedly enhanced. This interaction promotes CMA activation, leading to GPX4 degradation in lysosomes. Consequently, 2-amino-5-chloro-3-dimethylbenzylamine inhibits ferroptosis and CMA by targeting HSP90, thereby preventing GPX4 degradation and mitigating ferroptosis [42]. In acute injury contexts, legumain can interact with HSP90 to transport cytoplasmic GPX4 to the lysosome for degradation [43,44].

Furthermore, HSP90 was identified as a novel target of timosaponin AIII, a steroid saponin that exhibits potent anticancer activity across various cancers. timosaponin AIII binds to HSP90, forming a complex that targets and degrades GPX4, ultimately inducing ferroptosis in non-small cell lung cancer [45]. Similarly, our research clearly demonstrates that GPX4 in the cytoplasm of macrophages infected with M. bovis N# is increasingly transported to the lysosome for degradation. This phenomenon is abrogated upon the knockout of Mb3523c or treatment with an HSP90 inhibitor. Finally, confocal microscopy revealed that CMA pathway activity was significantly reduced following ΔMb3523c infection, and GPX4 degradation caused by M. bovis N# infection was notably mitigated after treatment with apoptozole, a CMA pathway inhibitor. This suggests that Mb3523c suppresses GPX4 expression via the CMA pathway, providing new insights into the mechanism of ferroptosis induced by MTBC. In fact, HSPA8 can recognize and bind proteins with a KFERQ-like motif, a key feature of HSPA8-mediated chaperone-mediated autophagy [46]. In theory, other proteins containing KFERQ-like motifs could also be influenced. However, GPX4 is strongly associated with lipid peroxidation and ferroptosis, making it more pathologically significant. Thus, this study specifically focused on GPX4 degradation via the CMA pathway. However, CMA-mediated degradation of other proteins could potentially serve as an alternative immune evasion mechanism utilized by M. bovis. To further elucidate the specific interaction mechanism between Mb3523c and HSP90, we predicted the transmembrane region of Mb3523c and constructed a truncated version. The co-IP test revealed that aa98–242, the outer membrane region of Mb3523c, is a critical domain for interaction with HSP90. In fact, Mb3523c, a member of the Mce family, has the potential to interact with host proteins. For example, Mce3E in Mtb interacts with and colocalizes with MAPK3 at the endoplasmic reticulum (ER). Mce3E alters the subcellular localization of MAPK3 from the cytoplasm to the ER, reduces MAPK3’s association with MAP2K1, and prevents the nuclear translocation of phospho-MAPK3. This inhibition of the MAPK3 signaling pathway results in suppressed TNF and IL6 expression and enhances mycobacterial survival within macrophages [14].

Similarly, Mce2E inhibits the activation of MAPK/ERK and MAPK/JNK, reducing TNF and IL6 expression in macrophages [13]. Therefore, the interaction between Mb3523c and HSP90 to induce the CMA pathway is plausible, as suggested by our results. We demonstrated for the first time that Mb3523c forms a hammer-like structure extending outside the bacterial membrane, with the “head” of the hammer recruiting HSP90 through direct interaction sites. Mutagenesis studies revealed that the G241 and Y237 residues of Mb3523c are crucial for its interaction with HSP90.

Given the 100% homology between the Rv3493c protein of H37Rv and Mb3523c, future research will focus on designing small molecule inhibitors that target specific sites on Mb3523c. This strategy aims to control the dissemination of MTBC, particularly Mtb and M. bovis, throughout the body.

In summary, our findings provide novel insights into the molecular mechanisms of pathogen-induced ferroptosis, demonstrating that targeting GPX4-dependent ferroptosis by blocking the M. bovis Mb3523c-host HSP90 interface represents a potential therapeutic strategy for tuberculosis.

Materials and methods

Animals and ethics statement

A cohort of 108 female C57BL/6 mice, aged six to eight weeks, was procured from SPF Biotechnology, Beijing, China, and accommodated within the Biosafety Level 3 (BSL-3) laboratory facilities at China Agricultural University. These mice were maintained under standard housing conditions within conventional cages, characterized by a controlled environment maintained at 21 ± 1°C and a relative humidity of 50 ± 10%. A consistent 12:12-hour light-dark cycle was established, with the onset of light at 08:00 hours. Unrestricted access to food and water was ensured for all mice, with equitable distribution across cages.

The execution of animal-based investigations adhered to protocols sanctioned by the Institutional Animal Care and Use Committee (IACUC) of China Agricultural University, Beijing. These protocols were in alignment with guidelines for the humane care and handling of laboratory animals, as delineated by the Ministry of Science and Technology, People’s Republic of China. The ethical review and approval for the animal study methodologies were granted by The Laboratory Animal Ethical Committee of China Agricultural University, Beijing, China, with the approval document numbered 20,110,611–01. All experimental maneuvers involving M. bovis culture and subsequent animal infections were meticulously performed under stringent biosafety protocols within the BSL-3 laboratories at China Agricultural University, Beijing.

Bacterial strains

Mycobacterium bovis BCG (ATCC 35,734), M. bovis C68004# provided by the China Institute of Veterinary Drug Control (CVCC, China), M. bovis N# first isolated in the 1950s and kept in a laboratory, M. bovis N#ΔMb3523c, M. bovis N#ΔMb3523c: Mb3523c (provide by Gene Optimal) were grown in Middlebrook 7H9 medium (BD 271,310) supplemented with 10% oleic acid-albumin-dextrose-catalase/OADC (BD Biosciences 212,351) and 0.05% Tween-80 (G-CLONE, CS9029), or on Middlebrook 7H10 agar (BD 262,710) supplemented with 10% oleic acid-albumin-dextrose-catalase.

Cell culture

The RAW264.7 macrophage cells, HEK293T and A549 cell were obtained from the Cell Culture Center, Peking Union Medical College and cultured in a humidified incubator at 37°C with 5% CO₂ in DMEM (Invitrogen, C11995500BT) supplemented with 10% fetal bovine serum (FBS; Gibco 16,000–044), 100 μL/mL streptomycin, and 100 μL/mL penicillin (Sigma-Aldrich 516,106).

Generation of M. bovis mutants

The process initiated with the construction of homologous exchange sites [47,48], followed by their integration into the genome of a mycobacteriophage to produce a phasmid. Briefly, this allelic exchange substrate designed to replace the Mb3523c gene was introduced into the PacI site of phasmid phAE159 and electroporated into Mycobacterium smegmatis mc2155 to obtain high-titer phage lysates. M. bovis strain was then washed twice with MP buffer containing 50 mm Tris (pH 8.0), 150 mm NaCl, 10 mm MgCl₂, 2 mm CaCl₂, and mixed with phage lysates at 37°C overnight. The cell resuspension was plated on Middlebrook 7H10 agar (BD 262,710) containing 75 µg/mL hygromycin (Sigma, H7772) at 37°C. For Mb3523c mutant complementation and overexpression strains of BCG, the integrative vectors pMV361 and pMV261 (provide by Gene Optimal) were used to create ΔMb3523c: Mb3523c, and BCG: Mb3523c. The primers and plasmids used were listed in Table S1.

Reagents and antibodies

The following chemicals were used: Ziehl-Neelsen (Z&N) stain (Solarbio, G1274), C11 BODIPY 581/591(Cayman Chemical 217,075-36-0), DAPI (Beyotime, C1002), FerroOrange probe (DojinDo, F374) Live/Dead fixable blue dead cell stain kit (Solarbio, CA1630), ferrostatin-1 (Mce, HY-100579), necrostatin-1 (Mce, HY-15760), apoptozole (Mce, HY-15098), tanespimycin (Mce, HY-10211), chloroquine (CQ; Mce, HY-17589A), MG-132 (Mce, HY-13259), VX-765 (Mce, SHY-13205), FITC-label (Solarbio, 3326-32-7), Alexa Fluor 594 affinipure donkey anti-mouse IgG (Invitrogen, A20000). For the antibodies, anti-HA antibody (1:200 for immunofluorescence, 1:3000 for immunoblotting; Abmart, M20003), anti-TUBA antibody (1:2000 for immunoblotting; Sigma-Aldrich, T6199), anti-GPX4 antibody (1:1000 for immunoblotting; ABclonal, A11243), anti-HSP90 antibody (1:1000 for immunoblotting, 1:200 for immunofluorescence; nature bioscience, A92707), anti-HSPA8 antibody (1:1000 for immunoblotting; Abmart, T58015) anti-LAMP1 antibody (1:1000 for immunoblotting; Proteintech 67,300–1-Ig), anti-LAMP2A antibody (1:1000 for immunoblotting; Abmart, TN24835), anti-GAPDH antibody (1:1000 for immunoblotting; Proteintech, 60004–1-Ig), anti-4-HNE antibody (1:300 for immunohistochemistry; Abcam, ab46545), anti-GSDMD antibody (1:1000 for immunoblotting; Abcam, Ab209845), anti-CASP1/caspase 1 (1:1000 for immunoblotting; Proteintech 22,915–1-AP), anti-p-MLKL antibody (1:1000 for immunoblotting; Abcam, Ab196436), anti-MLKL antibody (1:1000 for immunoblotting; Abcam, Ab184718), anti-GFP antibody (1:3000 for immunoblotting; Abmart, M20003)

Bacterial infection and CFU assay

RAW264.7 Cells were seeded in either 12- or 24-well culture plates using Dulbecco’s Modified Eagle Medium (DMEM; Invitrogen, C11995500BT) supplemented with 2% FBS and incubated for 12–18 hours. To establish a cell infection model, RAW264.7 macrophages were infected with M. bovis strains at MOI of 10. After 3 h, the infected cells were washed three times with pre-warmed PBS (solarbio, P1020) and incubated again with the fresh DMEM supplemented with 2% FBS in a CO₂ incubator at 37°C. The initial 3-h incubation was designated as the phagocytosis phase for the macrophages, marking the onset of the post-infection period (0 h). Samples were then collected at 24 h post-infection for further analysis.

For inhibitor pre-treated samples, RAW264.7 were treated with 10 μM Fer-1, 40 μM Nec-1, 50 μM VX-765, 30 μM apoptozole, 0.5 μM tanespimycin, 12.5 μM CQ, 200 nM MG-132 separately, for 3 h prior to infection.

For CFU assay, the macrophages were lysed in PBS containing 0.1% Triton X-100 (Solarbio, T8200), and the lysates were plated on 7H10 agar (BD 262,710). Colonies were then counted after 3–4 weeks.

Mouse model

Mice were anesthetized using Zoletil (50 mg/kg) with M. bovis C68004# strain, M. bovis N#, ΔMb3523c and ΔMb3523c: Mb3523c, via the intranasal route (i.n.) with 2000 CFU. At 2 weeks post-infection, vehicle or Fer-1 (3 mg/kg) and tanespimycin (5 mg/kg) was injected intraperitoneally into mice daily. After 24 h of M. bovis infection, one mouse from each infection group was randomly sacrificed, and lung bacterial burdens were measured by plating lung homogenates on 7H10 agar plates. An average of 2100 CFU was observed one day post-infection. After 2 or 4 weeks of infection, lungs and spleens were harvested and homogenized utilizing small ceramic beads within a tissue homogenization apparatus (Jing Xin Technology, JXFSTPRP-CL). Subsequently, each specimen was subjected to ten-fold serial dilutions in sterile PBS and inoculated in triplicate onto Middlebrook 7H10 agar plates for culture. The plates were incubated for a duration of two to three weeks. CFUs of M. bovis were enumerated following a three-week incubation period at 37°C. Lung specimens from the left lobe were fixed in 10% formalin, embedded in paraffin, sectioned, and stained with H&E for histological examination. A thin section of lung tissues was stained with Z&N staining method to observe the dissemination of M. bovis in the lung tissues of experimental mice, and lung tissues used immunohistochemical staining (Gene technology, GK600710) to observe the level of ferroptosis The H&E Z&N, and immunohistochemical-stained sections were examined by Leica CS2 and analyzed by SlideViewer (v2.5.0.143918), Image scope (v12.3 March 7014) and ImageJ (v1.8.0) software.

ELISA assay

For ELISA, the cell-free culture supernatants were harvested by centrifugation at 11,600×g for 1 min to remove suspension cells and mycobacteria. Levels of TNFIL1B and IFNG protein in the supernatants or serum were quantified by Mlbio and Neobioscience ELISA kits (Mouse IL1B: MM-0040M1; Mouse IFNG ELISA Kit: MM-0182M1; Mouse TNF ELISA Kit:560301), according to the manufacturer’s instructions. Each sample was done in triplicates.

Cell death assay, and LDH assay

For cell death assays, 5 μg/mL PI was added to cells at 37°C for 20 min and cells were measured using the TECAN Spark multifunctional microplate reader. For the LDH assay, the supernatants from RAW264.7 cells infected with various M. bovis strains and LDH regents were incubated in a sterile 96-well plate. Fluorescence was measured at TECAN Spark multifunctional microplate reader at 490 nm absorbance.

Lipid peroxidation and Fe²⁺assay

RAW264.7 cells were seeded at a density of 1 × 10⁵ cells per well in 24-well plates. The cells were incubated with 1.5 μM BODIPY (581/591) C11 or 1 µM FerroOrange probe at 37°C for 30 min, followed by washing in PBS. Confocal images were subsequently captured using a Nikon confocal microscope. The intensity of green fluorescence, indicative of lipid peroxidation levels, and orange fluorescence, indicative of Fe²⁺ levels, were analyzed using ImageJ software.

Flow cytometric analysis of lung cells

Excised lungs were washed with PBS and incubated on ice for 5 minutes. The sectioned lung tissues were then transferred to a 6-well plate containing 5 mL of RPMI-1640 complete medium (ThermoFisher scientific, C22400500BT), 50 μL of collagenase 1A (1 mg/mL; solarbio, C8140), and 50 μL of DNase1 (150 U/mL; Solarbio, D8071). The tissues were digested for 1 h. The digest was centrifuged at 100 ×g for 5 min, and the supernatant was discarded. Then, 3 mL of red blood cell lysis buffer was added and incubated for 5 min. Next, 3% FBS was added to block the reaction. The cell suspension was centrifuged at 100 ×g for 5 min, and the pellet was resuspended in 1 mL of RPMI-1640. The lung cell suspension was then transferred to a 1.5 mL microcentrifuge tube and mixed with 500 μL of 1× binding buffer. The cells were then treated with 5 μL of ANCA5 and 10 μL of PI and incubated for 5 min. The cells were analyzed using a BD FACS ARIA (NJ, USA) and FlowJo software.

Immunoprecipitation and immunoblot analysis

Protein interactions were assessed using immunoprecipitation. Briefly, the cells were also lysed in IP buffer (Beyotime, P1003) containing PMSF (1×10⁶ cells/150 μL). The homogenized samples were centrifuged to collect the supernatants. Add the anti-HA target antibody (1:200) to cell lysate. Gently rotate the cell lysate/antibody mixture overnight at 4°C on a rotator. Capture the immunocomplex by adding 30–40 μL Protein A/G-Agarose slurry (15–20 μL packed beads; Cell Signaling Technology, 37478P) and gently rotating on a rotator for 6 h at 4°C. Collecting the agarose beads by centrifugating for 3 min at 100 ×g and followed by extensive washing with IP buffer. The proteins were then separated on 12% SDS-PAGE and transferred to PVDF membrane (Millipore, IEVH85R). The membrane was incubated with antibody overnight at 4°C. Subsequently, it was incubated with horseradish peroxidase-conjugated IgG secondary antibody (dilution 1: 3000; Cell Signaling Technology, 2729) for 1 h at room temperature. The protein bands were developed and imaged using the ChemiDoc XRS system (Bio-Rad). The relative density of the blots was quantified using ImageJ.

Immunofluorescence and confocal microscopy

Cells were seeded onto coverslips and transfected with plasmids. After 24 h, cells were fixed, permeabilized, and blocked at room temperature. This was followed by overnight incubation with the primary antibody. After washing three times with PBS, fluorescently labeled secondary antibodies were applied and incubated for 1 h at 37°C. Coverslips were then mounted onto slides using DAPI-containing antifade mounting medium (Solarbio, YA0170). Confocal images were captured using a Nikon confocal microscope.

RNA sequencing

Total RNA was extracted from the RAW264.7 cell infected with M. bovis N# and M. bovis N#ΔMb3523c according to the manual of TRIzol® (Vazyme 15,596–026). Sequencing was conducted using the DNBSEQ platform with paired-end 150 bp reads. The adaptors and low-quality bases were assessed using FASTQC (v0.11.3). Trimmed reads were then aligned to the human ensemble 79 (GRCh38.p2) genome using STAR (v2.4.2a). The genes differentially expressed between WT M. bovis and M. bovis ΔMb3523c-infected cells with the fold change > 1.2 and the p-value <0.05 were analyzed by DESeq2 (v1.18.1) in R (v3.4.4). The results of RNA sequencing are available at the Gene Expression Omnibus (PRJNA1111330)

Mass spectrometry

For mass spectrometry, the precipitates immunoprecipitated with anti-HA antibody or control IgG were separated on 12% SDS-PAGE. Subsequently, the identification of protein gel strips is to separate the sample proteins by gel electrophoresis, then obtain the protein gel strips at different positions on the film, extract the peptides after enzymatic digestion, and then uses mass spectrometry to obtain the mass spectrum of the proteins in these gel strips, and finally uses the protein identification software to identify the proteins in the samples.

Lysosome isolation

Lysosomes were harvested by homogenization and sequential centrifugation with a lysosome isolation kit (BestBio, BB3603), according to the manufacturer’s protocol.

Plasmids and small interference RNA transfection

The coding sequences of M. bovis Mb3523c and its truncated mutants or point mutants (aa 1 to 74, aa 75 to 97 and aa 98 to 242) or (N158, N162, Y216, L235, Y237, and G241) were amplified from M. bovis genomic DNA and cloned into PCMV vector to get the PCMV-Mb3523c, PCMV-Mb3523c -aa1–74, PCMV-Mb3523c-aa75–97 and PCMV-Mb3523c aa-98–242 plasmids or PCMV-Mb3523c- N158A, PCMV-Mb3523c-N162A, PCMV-Mb3523c- Y216A, PCMV-Mb3523c- L235A PCMV-Mb3523c- Y237A, PCMV-Mb3523c- G241A. All plasmid constructs were confirmed by sequencing. The sequences of primers used for plasmid construction are listed in Table S1. PCMV-MYC, PMV-261, PMV-361 were preserved in our laboratory.

KFERQ-PA-mcherry-N1 plasmid was donated by Associate Professor Yina Zhang from Henan Agricultural University. KFERQ-PA-mCherry were transfected by Lipo3000 (Invitrogen, L3000075), and then cells were infected by M. bovis N#, ΔMb3523c or ΔMb3523c: Mb3523c. Ultimately, under the illumination of 405 nm visible light, lysosomes are vividly revealed as distinct red fluorescent specks.

RAW264.7 cells were transfected with 10 nmol/L HSP90 siRNA and 10 nmol/L control siRNA for 48 h and infected with M. bovis N#, ΔMb3523c and ΔMb3523c: Mb3523c (MOI = 10 for 24 h).

M. bovis staining

FITC was dissolved in 0.1 M bicarbonate/carbonate buffer (pH 8.5) to a final concentration of 1 mg/ml. Following the growth of the cultured M. bovis to the logarithmic phase, the culture was centrifuged at 1500 ×g for 10 min. At room temperature, the 1 mg/mL FITC solution was used to stain the bacteria on a shaker for 2 h. After staining, the bacteria were washed three times with PBS, subjected to multiple disruptions with a 25 g needle, and then filtered through a 5-μm filter to obtain single cell bacteria. The bacterial OD600 was measured using a spectrophotometer (Bio-Rad, USA). The bacteria were then used to infect RAW264.7 cells and observed under a confocal microscope 24 h post-infection.

Statistical analysis

The statistical analysis was conducted using Graphpad Prism 8. The data from two groups were compared using the t-test of the study. Statistical significance was denoted as *p < 0.05, **p < 0.01, ***p < 0.001. A p-value less than 0.05 was deemed to be statistically significant.

Funding Statement

This work was supported by; “National Key Research and Development Program (Project No. 2021YFD1800405)”; “National Natural Science Foundation of China (Project No.32172800);.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Availability of data and materials

All data generated or analyzed during this study are included in this published article

Ethics approval and consent to participate

Animal experiments were conducted following the protocols, approved by the animal care and use committee (IACUC) of China Agricultural University, Beijing, under the rules for the care of laboratory animals, Ministry of Science and Technology People’s Republic of China. The procedures of the current animal study were evaluated and approved by The Laboratory Animal Ethical Committee of China Agricultural University, Beijing, China with approval number 20,110,611–01.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2025.2468139