Escherichia coli DH5α used for gene cloning was cultured in Luria-Bertani medium with or without agar at 37°C. Mycobacterium tuberculosis H37Ra was used as a model strain in this study and was grown in Middlebrook 7H9 medium (BD Biosciences, Franklin Lakes, NJ, USA) supplemented with 10% (v/v) OADC (oleic acid–albumin–dextrose–catalase; BD Biosciences), 0.2% (v/v) glycerol, and 0.05% (v/v) Tween-80 at 37°C. HEK293T, HeLa, and RAW264.7 obtained from the American Type Culture Collection (ATCC) were grown in DMEM (Gibco, Grand Island, NY, USA) medium supplemented with 10% fetal bovine serum (Gibco), 100 U/ml penicillin, and 100 μg/ml streptomycin (Gibco) at 37°C with 5% CO₂.

The plasmids below were purchased from Clontech (Mountain View, CA, USA): pEGFP-C1, pDsRed2-ER, pEGFP-Actin, and pEYFP-Mem. CdhM and its mutants were cloned into pEGFP-C1 at restriction sites of HindIII and PstI and selected by kanamycin. The pMind vector was used for CdhM knockout in H37Ra as described previously (Liu et al., 2019). pMV261 was used for gene overexpression in H37Ra (Liu et al., 2019). All constructs were examined by sequencing. Plasmids were electroporated into H37Ra and selected on 7H10 medium containing 50 µg/ml kanamycin.

HEK293T or HeLa cells were seeded into 96-well plates a day before transfection, and transfection was performed when confluence reached around 90%. Hieff Trans™ Liposomal Transfection Reagent (YEASEN Biotech, Shanghai, China) was used for transfection. The plasmids without endotoxin were prepared using the EndoFree Maxi Plasmid Kit (TIANGEN Biotech, Beijing, China). The experimental procedures followed the respective manufacturer’s instructions. Around 24 h post transfection, cells were subcultured into the tailor-made 96-well plates with serial dilutions for live-cell imaging using the High Content Imaging System (PerkinElmer, Waltham, MA, USA). The High Content Imaging System could provide suitable conditions, like 37°C with 5% CO₂, for normal cell growth, and trace multiple precise fields of view continuously.

HEK293T cells were seeded into 12-well plates a day before transfection and transfection performed when confluence reached around 90% as above. 48 hours post transfection, cells were washed for a time with PBS and lysed directly with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) followed by flash freezing with liquid nitrogen. Then, samples were sent to Novogene (Novogene, Beijing, China) with dry ice for RNA isolation and sequencing as well as bioinformatics analysis. For RAW264.7, cells were seeded into 12-well plates a day before infection. Then, RAW264.7 cells were infected with CdhM-related strains at a multiplicity of infection (MOI) of 10 and incubated at 37°C with 5% CO₂ for 24 h. Subsequent procedures were the same as HEK293T.

The protocols for transfection or infection were the same as the sample preparation of RNA-seq. Total RNA was isolated from the transfected or infected cells using TRIzol reagen (Invitrogen). The quality of RNA was assessed using NanoDrop 2000 (Thermo Fisher, Waltham, MA, USA). MonScript™ RTIII Super Mix with dsDNase (Monad Biotech, Shanghai, China) was used for reverse transcription according to the manufacturer’s protocol. Then, quantitative real-time PCR (qRT-PCR) was performed for quantification of target gene expression with the SYBR Green PCR mixture (Aidlab, Beijing, China) in Bio-Rad CFX Connect Real-Time System. The relative mRNA expression levels were calculated by normalization to GAPDH. The primers used in this study are listed in Supplementary Table 2 .

Cells were seeded into 6-well plates a day before infection or transfection. The protocols for infection or transfection were the same as above. Then cells were washed three times with cold PBS and harvested from plates to microtubes followed by centrifugation at 1,000 rpm for 5 min. Cell samples were lysed with RIPA lysis buffer (Boster, Pleasanton, CA, USA) supplemented with protease inhibitor cocktail (GlpBio, Montclair, CA, USA) for 1 h on ice followed by centrifugation at 14,000 rpm for 15 min. Then, the supernatant was transferred into new microtubes and the concentration of total proteins was determined by Pierce™ BCA Protein Assay Kit (Thermo Fisher). Proteins were electrophoresed in 12% SDS-PAGE and transferred to a nitrocellulose blotting membrane (GE Healthcare Life Sciences, Chicago, IL, USA). The NC membrane was blocked with 5% BSA in TBST (20 mM Tris–HCl, pH 7.5, 137 mM NaCl, and 0.05% Tween 20) for 2 h at room temperature. Then the blots were incubated with primary rabbit or mouse IgG antibodies, with 1,000 times dilution in 5% BSA, overnight at 4°C, and then incubated with HRP-conjugated goat anti-rabbit or anti-mouse IgG secondary antibodies with 5,000 times dilution in 5% BSA for 2 h. The immunoblots were visualized with Immobilon Western Chemiluminescent HRP Substrate (Millipore, Burlington, MA, USA) according to the manufacturer’s protocol. ImageJ software was used for quantification of immunoblots. β-Actin was used as loading controls. All antibodies are listed as follows: BiP (CST#3177), CHOP (CST#2895), P-eIF2α (CST#3398), and β-actin (CST#8457) were purchased from Cell Signaling Technology (Danvers, MA, USA), and HRP-conjugated goat anti-rabbit (BA1054) and HRP-conjugated goat anti-mouse (BA1050) were purchased from Boster Biological Technology Co., Ltd. (Wuhan, China).

Cells were seeded into 12-well plates a day before infection. The protocols for infection were the same as above. Then cell apoptosis was detected with FITC Annexin V Apoptosis Detection Kit (BD, USA) according to the manufacturer’s protocol, and analysis was performed by flow cytometry (CytoFLEX LX, Beckman, Brea, CA, USA). The principles of this kit are that externalized membrane phospholipid phosphatidylserine (PS), which is a hallmark of apoptotic cells, will be stained with FITC-labeled annexin V protein, and propidium iodide (PI) can only be used to stain the nuclei of dead or membrane damaged cells which occur at the late stage of apoptosis. Thus, early apoptosis cells (FITC Annexin V positive and PI negative) and late apoptosis cells (FITC Annexin V and PI positive) can be distinguished from viable (FITC Annexin V and PI negative) or mechanically damaged cells (FITC Annexin V negative and PI positive).

RAW264.7 cells were seeded into 24-well plates a day before infection. Then it was infected with CdhM-related strains at an MOI of 10 and incubated for 4 h at 37°C with 5% CO₂. After 4 h for phagocytosis, the cells were washed three times with PBS to remove extracellular bacteria, and then a fresh medium with or without gentamycin was added for subsequent cell culture. At 4, 24, and 48 h postinfection (hpi), all bacteria, including intracellular and extracellular, were collected and plated on 7H10 (BD Biosciences) agar plates with serial dilutions. After being incubated at 37°C for about 3 weeks, colonies were counted.

CdhM-related strains were grown in Middlebrook 7H9 medium (BD, USA) supplemented with 10% (v/v) OADC (oleic acid–albumin–dextrose–catalase; BD Biosciences), 0.2% (v/v) glycerol, and 0.05% (v/v) Tween 80, at 37°C and 180 rpm. The OD₆₀₀ of bacteria was measured with a day interval. The mid-logarithmic phase of bacteria was seeded on 7H10 (BD, USA) agar plates for observation of colony morphology.

Statistical analysis was performed using GraphPad Prism 8.0 by one-way or two-way ANOVA followed by Tukey’s multiple comparisons test. Significant differences are marked with ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. All results are graphed as means ± SD for triplicate samples.

A high-throughput screening about the interaction of Mtb secretion proteins and macrophage had been performed in our lab (data not shown). A protein, which is annotated as CDP-diacylglycerol pyrophosphatase (alternative named CDP-diglyceride hydrolase) in KEGG, hereafter designated CdhM, caught our attention. We wondered therefore what will happen if CdhM is transfected into host cells. Surprisingly, a phenotype of intracellular vacuolation surrounding nuclei was observed at about 20 h post transfection ( Figure 1A ). Then, time-lapse live imaging was applied to trace the cells transfected with GFP-CdhM from 20 to 55 h post transfection, and we found that transfection with CdhM caused cell death with condensation morphology eventually, which resembled apoptosis ( Figure 1B ). We asked if those phenotypes are specific and functional. Therefore, the conserved amino acids of CdhM among all homologous sequences were analyzed by ClustalW Multiple Alignment and ten amino acids were identified ( Supplementary Figure 1 ). A single conserved amino acid mutant of CdhM was constructed and transfected into host cells. The mutants, including L44A, G96A, H150A, and W175A, were found unable to cause vacuolation and condensation-like cell death anymore. Among these mutants, L44A, G96A, and W175A still kept the similar subcellular localization with wild-type CdhM, but H150A made an obvious difference (The results of L44A and W175A were similar to G96A, so only the representative results of G96A were presented, similarly in the following.) ( Figures 1B, C and Supplementary Movies 1 – 4 ). It suggests that the phenotypes elicited by CdhM are specific and functional. Then, we compared the subcellular localization of CdhM with several known subcellular markers, such as pEGFP-Actin, pYFP-Membrane, and pDsRed2-ER, which are commercially available, and found that GFP-CdhM has a similar subcellular localization with DsRed2-ER ( Figure 1D ). The subsequent co-localization experiment verified that CdhM and its mutants L44A, G96A, and W175A, except H150A, localized to the ER ( Figure 1E ). We presumed therefore that the phenotype of intracellular vacuolation might be abnormal morphology of the ER, and the result of the condensation-like cell death might be apoptosis mediated by ER stress (Tabas and Ron, 2011).

In order to uncover the potential molecular mechanism of the phenotypes caused by CdhM, a global transcriptome sequencing (RNA-seq) of HEK293T transfected with CdhM was performed. The cells of untransfected (UN), transfected with empty vector (vector), and mutant H150A were set as controls. The quality control and statistics of the RNA-Seq results suggest that the data are reliable ( Figure 2A ). A clustering heatmap of differential gene expression among all four groups indicated that significant differences in gene expression profiling do exist ( Figure 2B ). The differentially expressed genes (DEGs) between CdhM and empty vector were analyzed and shown on a volcano plot ( Figure 2C ). There are 339 genes upregulated and 396 genes downregulated significantly in the cells transfected with CdhM relative to the empty vector. Then Gene Ontology (GO) and KEGG enrichment analyses of these DEGs were implemented, and we found that the top biological processes or signaling pathways are mostly ER-related ( Figures 2D, E ). Hence, we speculated that transfection with CdhM might disrupt the normal function of the ER and that the cells were experiencing ER stress. A clustering heatmap analysis of the 37 DEGs involved in ER stress showed significant differences between CdhM and other control groups ( Figure 2F ). It is consistent with the results in Mtb granulomas (Seimon et al., 2010). Incorporation of the subcellular localization, morphology phenotyping, and transcriptome analysis suggest that CdhM may induce ER stress in host cells.

Based on the above results, we performed more biochemical experiments to validate if CdhM causes ER stress in host cells. As mentioned before, the UPR would be activated when the ER stress happened. BiP, in which the expression was upregulated and dissociated with three sensors of the UPR, is the marker of UPR initiation (Hetz, 2012). On the other hand, activation of the C/EBP homology protein (CHOP) is a marker of the late stage of the UPR, which implicates that cell apoptosis would happen (Tabas and Ron, 2011). Both proteins are therefore well-known indicators for assessing ER stress. Thus, we detected both transcriptional and protein levels of BiP and CHOP via qRT-PCR and Western blot, respectively. As shown in Figures 3A, B, D, E , compared with the empty vector, cells were transfected with CdhM, but its mutants, both BiP and CHOP, were upregulated significantly at both transcriptional and protein levels. It is consistent with the results of RNA-seq. After that, we wondered which signaling pathway of the three of the UPR was activated by CdhM. An unconventional splicing of XBP1 mRNA is the marker for the activation of the IRE1 pathway. A 26-base-pair fragment of XBP1 mRNA would be deleted by the activated IRE1 during the splicing process (Uemura et al., 2009). Thus, the PCR products based on the templates of spliced or un-spliced XBP1 cDNA could be separated by SDS-PAGE, and the percentage of spliced XBP1 in total XBP1 could be used to determine the degree of IRE1 activation. Experimental results show that the percentage of spliced XBP1 in the cells transfected with CdhM is about 50%, which is markedly higher than the empty vector as well as its mutants ( Figure 3C ), indicating that the IRE1-pathway was activated by CdhM. In addition, we also assessed the activation of the PERK pathway via the phosphorylation of eIF2α which is considered as the hallmark of PERK activation (Harding et al., 1999). Indeed, compared with the empty vector and its mutants, transfection with CdhM elevated the phosphorylation levels of eIF2α remarkably ( Figures 3D, E ). It implicated that the PERK pathway was activated by CdhM as well. Taken together, we confirmed that CdhM causes ER stress and activates the UPR through at least two signaling pathways.

Having shown the ability of CdhM to target the ER and induce ER stress alone, we then considered if CdhM would exhibit these abilities during Mtb infection of macrophage. Hence, CdhM-related strains were prepared, such as CdhM-knockout strain (KO), constructed by double exchanges of homologous strands, CdhM-overexpression strain (OE), and mutant of H150A ( Supplementary Figure 2A, B ). Firstly, a transcriptome sequencing of RAW264.7 infected with wild-type (WT) and CdhM-related strains was performed ( Supplementary Table 1 ). A clustering heatmap analysis of 43 DEGs involved in ER stress showed significant differences between KO and WT as well as OE, but the fold changes of the DEGs were not as large as these between the uninfected group (UI) and WT ( Figure 4A ). It suggests that besides CdhM there are other effectors of Mtb contributing to induction of ER stress during infection, which corresponds to the previous research (Choi et al., 2013; Grover et al., 2018). Next, biochemical experiments were performed to investigate the effect of CdhM in ER stress induction during Mtb infection of macrophage. The same as above, we measured the transcriptional and protein levels of ER stress markers BiP and CHOP ( Figures 4B, C, E, F ), and splicing of XBP1 ( Figure 4D ) as well as phosphorylation of eIF2α ( Figures 4E, F ). As shown in the results, compared with wild type and uninfected, overexpression of CdhM, except its mutants, significantly increased the expression levels of BiP and CHOP as well as phosphorylation levels of eIF2α and percentages of spliced XBP1. These results confirmed that CdhM worked in induction of ER stress during Mtb infection of macrophage. However, we were surprised that the CdhM-knockout strain showed no significant differences with the wild type. There are no redundant homologous genes of CdhM in the genome of Mtb. It seems that there were other effectors that compensated for the lack of CdhM in induction of ER stress during Mtb infection of host cells. Collectively, CdhM can also induce ER stress during Mtb infection of the host.

Since the ability of CdhM to induce ER stress during infection had been verified and the upregulation of CHOP in this process was observed too, we accordingly speculated that CdhM might also promote the apoptosis of macrophage during Mtb infection. Hence, the apoptosis of RAW264.7 infected with CdhM-related strains was assessed using FITC Annexin V Apoptosis Detection Kit (BD, USA), and analysis was performed by flow cytometry (CytoFLEX LX, Beckman). As shown in Figure 5A , infection with Mtb notably elevated the apoptosis of RAW264.7 compared to uninfected cells. The percent of apoptosis of CdhM-knockout and H150A mutant was less than that of the wild type but has no statistically significant differences, but they all were less than the CdhM-overexpression strain with statistically significant differences. It corresponds with the results before and proves that CdhM enhances apoptosis of macrophage during Mtb infection.

It is well defined that apoptosis of macrophages would influence the survival of infecting pathogens. Thus, we measured the colony-forming unit (CFU) of CdhM-related strains during infection of RAW264.7. Firstly, at 4 h postinfection (hpi), extracellular bacteria were washed with PBS and fresh mediums with 50 μg/ml gentamycin were added for continuing cell culture. The CFU of bacteria was measured at 24 and 48 hpi, respectively. As shown in Figure 5B , beyond our expectation, there are no obvious differences between all strains. It has already been discussed that apoptosis of macrophages is a double-edged sword during Mtb infection. In general, apoptosis restricts Mtb infection by releasing them into the extracellular environment in the form of apoptotic bodies. These bodies can be phagocytized by other macrophages for the further degradation of Mtb or by the dendritic cells, thereby leading to the activation of T lymphocytes. However, if the following actions are restrained, things might lead to another outcome. Thus, we tried to add mediums without antibiotics for subsequent cell culture after removing the extracellular bacteria at 4 hpi and measured the CFU at 24 and 48 hpi as before. As shown in Figure 5C , compared to wild-type or knockout strains, overexpression of CdhM conferred a proliferative advantage to Mtb during infection of macrophages in vitro. It should be noted that there were no obvious differences in growth curves and colony morphology among CdhM-related strains ( Figures 5D, E ). Thus, the proliferative advantage should be provided by CdhM-induced apoptosis that releases Mtb into the antibiotic-free extracellular environment. Taken together, CdhM may favor Mtb dissemination and proliferation at appropriate situations, like late stage of granulomas, by promoting ER stress-mediated apoptosis of infected macrophages, which liberates them into the extracellular milieu.