Post-translational knockdown and post-secretional modification of EsxA unambiguously determine the role of EsxA membrane permeabilizing activity in mycobacterial virulence

Current genetic studies (e.g. gene knockout) have suggested that EsxA and EsxB function as secreted virulence factors that are essential for Mycobaterium tuberculosis (Mtb) virulence, specifically in mediating phagosome rupture and translocation of Mtb to the cytosol of host cells, which further facilitates Mtb intracellular replicating and cell-to-cell spreading. The EsxA-mediated virulence is presumably achieved by its pH-dependent membrane-permeabilizing activity (MPA). However, the data from recent studies have generated a discrepancy regarding to the role of EsxA MPA in mycobacterial virulence with a major concern that genetic manipulations, such as deletion of esxB-esxA operon, may stimulate genetic compensation to produce artifacts and/or affect other co-dependently secreted factors that could be directly involved cytosolic translocation. To avoid the drawbacks of gene knockout, we first engineered a Mycobacterium marinum (Mm) strain, in which a DAS4+ tag was fused to the C-terminus of EsxB to allow inducible knockdown of EsxB (also EsxA) at the post-translational level. We also engineered a Mm strain by fusing a SpyTag to the C-terminus of EsxA, which allows inhibition of EsxA-ST MPA at the post-secretional level through a covalent linkage to SpyCatcher-GFP. Both post-translational knockdown and post-secretional inhibition of EsxA resulted in attenuation of Mm intracellular survival and virulence in macrophages and lung epithelial cells, which unambiguously confirms the role of EsxA MPA in mycobacterial virulence. Author Summary Genetic studies, such as loss of function by gene deletion and disruption, have suggested that EsxA is a virulence factor essential for mycobacterial virulence. However, its role is questioned because knockout of esxA gene may affect the function or secretion of other related genes. Here, we employed two methods other than gene deletion and disruption to determine EsxA role in mycobacterial virulence. First, we added a degradation signal peptide DAS4+ tag to the C-terminus of EsxB, the chaperon of EsxA so that EsxB-DAS4+ could be degraded by protease ClpXP, whose function can be induced by an inducer, ATC. By this way, we were able to control the amount of EsxB and EsxA at the post-translational level. The results showed that ATC inhibited mycobacterial intracellular survival through down-regulating EsxA and EsxB. Second method is to take advantage of SpyTag(ST) and SpyCatcher(SC) system. Like DAS4+, ST was fused to C-terminus of EsxA without affecting its expression, secretion and MPA. After secretion, EsxA-ST can be specifically recognized by SC-GFP and form a covalent bond between ST and SC, which blocks the MPA, an activity that directly related to mycobacterial virulence. Endogenous expression of SC-GFP in the infected cells inhibited mycobacterial intracellular survival. In summary, our results demonstrate that knockdown of EsxA at the post-translational level or inhibition of EsxA MPA at the post-secretional level, attenuate mycobacterial virulence, and this attenuation is solely attributed to EsxA, not to other factors.


44
Pathogenic mycobacteria, like tuberculosis and leprosy species, have been imposing great threats to public 45 health for decades (1,2). Extensive research on their virulence and pathogenicity is urgently needed to 46 minimize the impacts of pathogenic mycobacteria. Completion of whole genome sequencing of multiple 47 mycobacteria species has facilitated researches on mycobacterial pathogenicity, epidemiology, detection and 48 3 vaccine development (3-9). Various gene editing methods allow researchers to explore genes of interest for 49 their roles in mycobacterial virulence and pathogenicity (10).

50
The esxB-esxA operon is located within the esx-1 locus in Mtb genome that encodes a Type VII secretion 51 system (11). Current genetic studies (e.g. gene deletion, disruption or mutation) have suggested that EsxA and 52 EsxB play an essential role in mycobacterial pathogenicity, intracellular translocation and escape from 53 immune responses (12)(13)(14)(15)(16)(17). EsxA and EsxB are secreted as a heterodimer through the ESX-1 secretion system 54 (18). Our previous studies have demonstrated that EsxA, but not EsxB, exhibits acidic pH-dependent MPA 55 (19). The MPA is uniquely present in the EsxA proteins from pathogenic Mtb and Mm, but is absent in the 56 highly homologous EsxA from non-pathogenic Mycobacterium smegmatis (Ms) (19,20). This suggests that 57 EsxA MPA is the key factor determining the virulence phenotype of mycobacteria. This notion is further  (19,20,24), but they found that Mm was still able to 70 penetrate the phagosome and translocate to the cytosol in the presence of Bafilomycin, a reagent that inhibits 71 intracellular acidification, indicating that phagosome rupture doesn't occur through the acidic pH-dependent 72 4 MPA of EsxA (23). Most recently, Lienard et. al. employed a collection of Mm ESX-1 transposon mutants, 73 including the mutants that disrupt EsxA secretion, to infect macrophages and showed that the transposon 74 mutants without EsxA secretion was still able to permeabilize phagosomes, suggesting that other factors 75 independent of EsxAB play a role in cytosolic translocation (17). Therefore, conclusive evidence is required 76 to resolve the discrepancy. It is possible that knockout of the esxB-esxA operon stimulates mycobacterial 77 genetic compensatory mechanisms (e.g. secondary mutations and altered gene expression) that produce 78 artifacts (25,26), and knockout of esxB-esxA also affects the co-dependently secreted factor(s) (e.g. EspA, 79 EspC or EspB) that could also play roles in cytosolic translocation (18,(27)(28)(29)(30). 80 In order to determine the exact role of EsxA in mycobacterial virulence, in the present study we employed two 81 approaches to avoid the potential artifacts caused by gene knockout. We first constructed a Mm recombinant 82 strain, namely Mm(EsxB-DAS4+), in which a degradation signal peptide DAS4+ was fused to the C-terminus   To avoid the potential artifacts caused by gene knockout, we set off to construct a Mm strain that allows 97 inducible knockdown of EsxA at the post-translational level. Initially, we attached the degradation peptide 98 DAS4+ to the C-terminus of EsxA in the Mm genome. However, EsxA-DAS4+ was not stably expressed in 99 the engineered strain. Since an earlier study has shown that the C-terminal modifications have no significant 100 impact on EsxB's interaction with other proteins (36), we attached DAS4+ to the C-terminus of EsxB (Fig.   101   S1A) and confirmed the construction by PCR (Fig. S1B). Western blot analysis showed that EsxB-DAS4+ 102 was expressed in a similar level as EsxB in Mm(WT) in the total lysates. In the culture filtrate, however, 103 EsxB-DAS4+ was less than EsxB, suggesting that the secretion of EsxB-DAS4+ was partly affected by 104 DAS4+ tag (Fig. 1A). As expected, the intracellular survival of Mm(EsxB-DAS4+) was ~10 folds lower 105 than that of Mm(WT), but it was still ~100 folds higher than that of Mm(ΔEsxA:B) (Fig. 1B).

107
To test the inducible knockdown of EsxB-DAS4+, the Mm(EsxB-DAS4+) liquid culture was treated with 108 ATC (0.5 g/ml) for various times. The lysate was applied to SDS-PAGE, followed by Western blot. The 109 expression of EsxB-DAS4+ was significantly reduced after 6 h of ATC treatment, while ATC had no effect 110 on EsxB even after 48 h treatment on Mm(WT) ( Fig. 2A). Interestingly, the expression of EsxA was 111 similarly diminished after 6 h treatment ( Fig. 2A). Given that EsxA and EsxB reportedly form a 112 heterodimer, the protease may induce degradation of the heterodimer at the post-translational level.

113
Moreover, ATC did not down-regulate the transcription level of esxB-DAS4+ (Fig. 2B) was significantly reduced (Fig 2C, D). Interestingly, EsxA was also significantly reduced (Fig 2E, F), which 118 is consistent to the finding that EsxA and EsxB form a heterodimer.

ATC-induced knockdown of EsxB-DAS4+ attenuated Mm's intracellular survival in mammalian cells
We then tested the effects of ATC induction on Mm's intracellular survival in WI-26 cells. The Mm(EsxA-121 DAS4+)|pGMCKq1 cells were pre-treated with ATC (0.5 μg/mL) before infection and then were applied to 122 infection of WI-26 cells. As expected, the ATC-treated Mm(EsxB-DAS4+)|pGMCKq1 had a significantly 123 lower intracellular survival than that without ATC treatment. As controls, the intracellular survival of 124 Mm(WT) is not affected by ATC (Fig. 3A). Next, we set out to test the effects of ATC treatment during the 125 infection on Mm intracellular survival. First, we titrated the cytotoxic effect of ATC and found that ATC did 126 not have significant cytotoxicity at a concentration up to 5 μg/mL (Fig. 3B). To ensure knockdown of EsxB 127 when Mm(EsxB-DAS4+) is inside the host cell, the highest concentration without cytotoxicity (5 μg/mL) 128 was applied to cell culture medium at the time of infection (0 hpi). At 24 hpi and 48 hpi, the intracellular 129 survival of Mm(EsxB-DAS4+)|pGMCKq1 with ATC addition was significantly lower than the group 130 without ATC (Fig. 3C). ATC had no effect on the intracellular survival of Mm(WT) at 24 hpi, but had a 131 down-regulatory effect at 48 hpi. It could be because that ATC is an antibiotic that inhibits bacterial growth 132 overtime, or it has cytotoxic effects to the host cells. However, we noticed that the ATC unspecific 133 inhibition to Mm(WT) was much less than its specific inhibition for Mm(EsxB-DAS4+) (Fig. 3C). Similar 134 results were acquired from THP-1 cell line (Fig. 3D). Together, knockdown of EsxB-DAS4+ by ATC either 135 before or during the infection attenuates mycobacterial intracellular survival.

138
According to an early report, modification of EsxA N-terminus impaired its secretion and function (37), so 139 we engineered the ST to C-terminus of EsxA with the suicide plasmid (Fig. S2). The insertion of ST was 140 confirmed by PCR using the overlap primers (Fig. S2). To test if EsxA-ST is expressed, the lysate of 141 Mm(EsxA-ST) was incubated with the purified SC-GFP and then applied to SDS-PAGE, followed by 142 Western blot using either anti-EsxA antibody (Fig. 4A) or anti-GFP antibody (Fig. 4B). The results showed 143 that EsxA-ST was successfully expressed, and a portion of EsxA-ST reacted with SC-GFP to form a higher 7 molecular weight complex EsxA-ST-SC-GFP (~ 70 kDa) ( Fig. 4A and B). As a control, SC-GFP did not 145 react with EsxA in the lysate of Mm(WT). Next, we tested if SC-GFP reacts with the bacterial surface-146 associated EsxA-ST. The live mCherry-expressing Mm(EsxA-ST) and Mm(WT) were first incubated with 147 SC-GFP, and then the cells were washed to remove free SC-GFP, which was followed fluorescence 148 microscopy. We found that Mm(EsxA-ST) was labeled by SC-GFP, but Mm(WT) was not, indicating that  are less than 100% and vary, in order to obtain accurate and reliable results, we measured the intracellular 165 survival with two independent approaches, fluorescence microscopy and CFU counting. In the fluorescence 166 microscopy method, we selected the cells containing both mCherry fluorescence (mycobacterial cells) and 167 green fluorescence (either GFP or SC-GFP) and then quantified the Red/Green ratio in each cell (Fig. 6A).
The Red/Green ratio of the cells containing Mm(EsxA-ST) and SC-GFP was significantly lower than that of 169 the cells containing Mm(EsxA-ST) and GFP (Fig. 6B), while the Red/Green ratio of the cells containing 170 Mm(WT) and SC-GFP was similar to the cells containing Mm(WT) and GFP (Fig. 6B), suggesting that SC-171 GFP specifically inhibited Mm(EsxA-ST) intracellular survival. Consistent results were obtained in the CFU 172 counting assay (Fig. 6C), showing that at 48 hpi, the intracellular survival of Mm(EsxA-ST) in the cells 173 expressing SC-GFP was significantly lower than that in the cells expressing GFP.

175
Despite of extensive studies in the past decades, the data regarding to the role of EsxA in Mtb pathogenesis, 176 particularly in phagosome rupture and cytosolic translocation, is still conflicting. The artifacts resulted from 177 gene knockout may be a major reason for the discrepancy and confusion. One, knockout of the esxB-esxA 178 operon may stimulate hosts to produce compensatory mechanisms, such as secondary mutations or 179 alternations of expression of other genes. Two, knockout of EsxAB will affect other factors (e.g. EspA, EspB 180 and EspC) that are co-dependently secreted with EsxAB. Therefore, to avoid the potential artifacts induced 181 by gene deletion, we engineered the Mm(EsxB-DAS4+) strain to allow inducible knockdown of EsxB (also 182 EsxA) at the post-translational level, which presumably minimizes potential compensatory mutations (Figs. 183 1-2). As expected, inducible knockdown of EsxB significantly attenuated Mm intracellular survival (Fig. 3). 184 However, there is still a concern that inducible knockdown of EsxB and EsxA affects the secretion of other 185 co-dependent factors. Therefore, we constructed Mm(EsxA-ST) stain to allow inhibition of EsxA MPA at the 186 post-secretion level. The data showed that attaching ST to the C-terminus of EsxA does not affect the 187 expression, secretion and function of EsxA, and hence it has no effect on Mm virulence (Fig. 4). The MPA of 188 EsxA-ST can only be specifically inhibited by SC-GFP through a covalent bonding (Fig. 5). Endogenous 189 expression of SC-GFP attenuated the intracellular survival of Mm(EsxA-ST) (Fig. 6). Therefore, the data in 190 this study conclusively support that EsxA MPA plays a direct role in mycobacterial intracellular survival and 191 virulence. In fact, the data obtained in this study is highly consistent with our earlier reports. We have found that single-  The DAS4+ system is based on mycobacterial innate protease ClpXP to degrade the intracellular protein (31, 201 32). Since ClpXP's protease expression is dependent on adaptor protein SspB, the inducible SspB expression 202 makes the degradation conditionally (38, 39). Initially, the DAS4+ tag was attached to C-terminus of EsxA, 203 but EsxA-DAS4+ was not stable, so we engineered Mm(EsxB-DAS4+). While EsxB-DAS4+ was stable, it 204 had a lower secretion than Mm (WT) EsxB (Fig. 1A). Thus, Mm(EsxB-DAS4+) had a lower virulence than 205 Mm(WT), but it is still much stronger than Mm(ΔEsxA:B) (Fig. 1B). Interesting, ATC induction also caused 206 degradation of EsxA, which is consistent to the fact that EsxB and EsxA form a heterodimer (40). Moreover,

207
AT C did not affect the mRNA level of EsxB. Together, ATC-induced knockdown of EsxB (also EsxA) occurs 208 at the post-translational stage.

209
The ST/SC system was originally developed to produce synthetic peptides (33). It is based on spontaneous 210 formation of an amide linkage between Lys and Asn, which leads to intramolecular isopeptide bond in the 211 stable pilin protein in Streptococcus pyogenes (41,42). After optimization, ST was determined to be a 13 aa 212 peptide, which binds fast and stably with 138 aa SC under variant conditions (34, 35). Unlike antibody-antigen 213 recognition, ST binds SC through an irreversible covalent linkage, which prompts its usage for protein 214 modification and fluorescent imaging (43,44). In this study, EsxA-ST was properly expressed and secreted to 215 the mycobacterial surface where it was covalently modified with SC-GFP (Fig. 4). Covalent linkage of SC-216 10 GFP to EsxA-ST inhibited its MPA in liposome leakage assay (Fig. 5) and endogenous expression of SC-GFP 217 inhibited Mm(EsxA-ST) intracellular survival in A549 cells (Fig. 6). Based on our experience, the rate of 218 transient transfection and the rate of infection never reach 100% in the target cells. Thus, there were cells that 219 were transfected but not infected and vice versa. In order to accurately measure the effect of SC-GFP 220 expression on mycobacterial infection, we applied two independent approaches, fluorescence quantification 221 and CFU, as described in Fig. 6. The fluorescence quantification is to calculate red/green ratio in the cells 222 with both red (mCherry) and green fluorescence (SC-GFP or GFP). However, the CFU approach is not able 223 to distinguish the heterogeneity of the cells. Because the cells that were infected but not transfected by SC-224 GFP were included into the CFU counting, the SC-GFP specific inhibitory effect was diluted, which may  In summary, using DAS4+ system and ST/SC system we were able to knockdown EsxB and EsxA at the post- SpyCatcher-ELP-GFP protein. 254 We found that the purified SpyCatcher-ELP-GFP protein was toxic to human cells, which is presumably due 255 to the cytotoxic effect of ELP linker between SC and GFP (48). Therefore, we removed ELP link to express  The primers for site-directed mutagenesis, identification and RT-qPCR were all designed according to Mm's 264 12 genomic DNA sequence (GenBank Accession Number: CP000854.1 ) and synthesized by  Louis, MO, USA). The plasmids and primers used in this study are listed in Table 1 To fuse ST or DAS4+ to the C-terminus of EsxA or EsxB, the homologous arms that flank the genes esxA or 269 esxB were amplified from the genomic DNA of Mm, respectively. Then the amplified fragments were inserted 270 into pJSC407-sacB. Similarly as previously described (21)  Immunofluorescence assay was also used to test the inducible knockdown of EsxB-DAS4+. As described 285 above, the Mm(EsxB-DAS4+)|pGMCKq1 pellet was collected at 48 h of post ATC induction. After the 286 bacteria were fixed to the coverslip, the bacteria were incubated with 2% BSA for 30 mins at RT. After that, 287 anti-EsxB polyserum and anti-EsxA polyserum were used as a primary antibody, and the FITC labeled 288 secondary antibody was used to visualize bacteria-associated EsxB-DAS4+ and EsxA under confocal 289 microscopy. To quantify the amount of bacteria-associated EsxB-DAS4+ and EsxA, the layers of FITC (green) 290 and mCherry (red) were extracted from each image to calculate the overlap rate between the green and red 291 signal. Briefly, the areas with green or red signals were first calculated. Then, the area of green that was 292 colocalized with red was calculated. This area's percentage in image's total red area is the overlap rate, which 293 was calculated according to equations in Table 2. Layer extraction and area calculation were achieved with 294 Python 3.7.3 (49). areas that were colocalized with green were calculated. This area percentage in the total green area is the 313 overlap rate we need, which was calculated according to equations in Table 2. Layer extraction and area 314 calculation were achieved with Python 3.7.3 (49).

588
The experiments were duplicated and data is present as mean ± SD . For cytotoxicity data, the statistical 589 analysis was performed with One-way ANOVA, followed by Holm-Sidak multiple comparison. For CFU data,  plasmid pJSC407-sacB was used to insert DAS4+ to the C-terminus of EsxB. Then the pGMCKq1-10M1-ssBopt plasmid that encodes adaptor protein was electroporated into Mm(EsxB-DAS4+).
(B) The genomic DNA of Mm(EsxB-DAS4+) was extracted and applied to PCR to confirm the insertion of DAS4+. The primers that overlap the insert sequence were used for PCR and produced a specific DNA fragment with an expected length ~ 400 bp, which is absent in the genomic DNA extracted from Mm(WT).  The samples were applied to SDS-PAGE, followed by Western blots using the antibody against GFP.