The ESX-1 secretion system senses bacterial contact and prepares mycobacteria for environmental adaptation

The ESX-1 system (6-kDa early secretory antigenic target (ESAT-6) secretion system-1) is essential for Mycobacterium tuberculosis pathogenesis and conjugal transfer in Mycobacterium smegmatis, yet little is known about how its function is regulated. Live-cell fluorescence microscopy showed natively expressed ESX-1 was organized into distinct foci predominantly observed at cell-cell contacts. These foci formed when two cells touched and required a fully assembled ESX-1 system in both bacteria, suggesting the generation of an ESX-1 megacomplex across multiple membranes. The emergence of ESX-1 foci and ESX-1 secretion was environmentally dependent: foci formed in low nitrogen environments in which secretion was suppressed, yet with increasing concentrations of nitrogen, ESX-1 systems diffused along the plasma membrane and secretion was activated. Genome-wide transcriptional profiling revealed ESX-1 dependent induction of genes required for the SOS response and error prone DNA replication in high nitrogen. Based on these findings, we propose a new model of ESX-1 function where ESX-1 localization and secretion are responsive to nitrogen levels and form an integral node in the mycobacterial response to neighboring cells and environmental adaptation.

there is growing evidence that ESX-1 secretion systems are regulated by environmental factors 74 during M. tuberculosis infection (Berthet et al., 1998;Fortune et al., 2005). In addition, in M. 75 smegmatis, ESX-1 secretion was found to be active when cells were grown on Sauton's 76 medium, and largely inactive in 7H9 medium (Converse and Cox, 2005). The ramifications of 77 ESX-1 regulation are yet to be explored. 78

79
In this work we constructed functional GFP fusions expressed at endogenous levels to study the 80 localization of ESX-1 components. We found that ESX-1 formed discrete foci at either side of a 81 cell-cell contact in cells grown in 7H9 medium, a condition in which ESX-1-mediated EsxB 82 secretion was inhibited. Conversely, ESX-1 was localized diffusely around the membrane in 83 Sauton's medium, when ESX-1 secretion of EsxB was active. We show that the increase in 84 nitrogen levels in Sauton's medium was sufficient to induce both re-localization and activation of 85 ESX-1 secretion of EsxB. We used RNAseq to probe the physiological function of ESX-1 and 86 discovered that ESX-1 was necessary for activating the mycobacterial SOS response to 87 nitrogen addition. Taken together, these findings, suggested an unexpected function of the 88 ESX-1 secretion system in regulating stress responses in high nitrogen environments that may 89 inform on its role in mycobacterial pathogenesis. 90 91

ESX-1 forms stable foci at cell-cell contacts 93
Prior studies reported the localization of heterologous ESX-1 components upon overexpression 94 of plasmid-based EGFP fusions (Soler-Arnedo et al., 2020; Wirth et al., 2012), which in some 95 cases, leads to non-physiological localization. To investigate the localization of native ESX-1 96 expressed at endogenous levels, we introduced an EGFP tag into multiple ESX-1 components 97 in the chromosome. We determined that EccCb1-EGFP was a functional EGFP fusion, as 98 shown by a secretion assay probing for EsxB in the culture medium although other fusions were 99 not functional ( Figure 1A). We grew these cells to exponential phase in 7H9 liquid medium and 100 then mounted them into microfluidic chambers for time lapse spinning disc confocal microscopy. 101

102
In the minority of cells that were not physically contacting another one, EccCb1-EGFP exhibited 103 a dim localization around the whole plasma membrane ( Figure 1B, top panel). This localization 104 was accentuated by time averaging the images (t avg.). Kymography analysis showed that this 105 intensity was maintained over time ( Figure 1C, top). However, most cells clumped together in 106 large aggregates in 7H9. In cells that were contacting others, we found that ESX-1 components 107 formed discrete foci at regions of cell-cell contact ( Figure 1D middle and bottom panels). These 108 foci were observed at the contact site between cells that occurred either along the cell body 109 sites and are not limited to cell poles. We also confirmed that EccCb1-EGFP plasmid-based 114 overexpression caused foci formation at the poles, regardless of cell-cell contact (Supplemental 115 Figure 1) suggesting that previously reported polar localization may be due to overexpression of 116 and subsequent EGFP self-interaction (Landgraf et al., 2012). In addition, we show that 117 endogenous expression of the monomeric construct of EGFP, mEGFPmut3, yields similar 118 observations of foci at cell-cell junctions and does not differ from our EGFP observations 119 (Supplemental Figure 2). 120

121
We used time-lapse microscopy to investigate the dynamics of focus formation ( Figure 1D). 122 Growing cells were observed while contacting each other and forming an ESX-1 focus on the 123 cell-cell contact point. A representative example is shown in Figure 1C. At t = 0 min when the 124 cells were near but not touching each other, there were no detectable foci. At t = 30 min multiple 125 dim foci appeared along the contact region. By t = 35 min onward there was a single, persistent 126 ESX-1 focus on the contact site that persisted. A kymograph of the entire time-lapse acquisition 127 at the cell-cell contact site illustrates this behavior over time ( Figure 1E). EGFP intensity plots 128 show that the foci form within 5 minutes ( Figure 1F

ESX-1 foci form large oligomers at cell-cell contact sites 134
The discrete ESX-1 foci suggested that ESX-1 formed large oligomeric assemblies at the 135 membrane at cell-cell contacts. To quantify the number of ESX-1 complexes at these sites we 136 compared the fluorescent intensity of EccCb1-EGFP foci to the MotB-EGFP complex which has 137 been reported to contain 22 +/-4 EGFP molecules in each focus (Coffman and

ESX-1 forms a megacomplex across two contacting cells 274
We determined whether ESX-1 complexes at cell-cell contacts were a one or two-sided 275 interaction. To address this question, we used a co-culturing experiment in which strains 276 expressing EccCb1-EGFP were mixed with ecc knockout strains marked with a cytoplasmic 277 mCherry and assayed whether EccCb1-EGFP foci were detected at cell-cell contact sites. As 278 shown by representative time averaged images, focus formation was induced between wild-type 279 and ∆esxB strains, whereas all ecc knockout strains were largely unable to induce foci formation 280 in the other cell ( Figure 2C). Focus formation in PmCherry wild-type and ∆esxB cells retained ~ 281 99% focus formation at cell-cell contacts with EccCb1-EGFP cells ( Figure 2D). Focus formation 282 at cell-cell contacts between PmCherry -ecc knockouts and EccCb1-EGFP labeled cells was as 283 follows: in ∆eccB1-PmCherry 5%, ∆eccCa1-PmCherry 3%, ∆eccD1-PmCherry 13%, and ∆eccE1-PmCherry 284 10% ( Figure 2D). In all instances, foci still formed between EccCb1-EGFP labeled cells, 285 indicating that focus formation was unaffected by the co-culture milieu ( Figure 2C We next investigated if focus formation was dependent on environmental conditions. 328 Observations of ESX-1 focus formation were made in 7H9 medium, however, previous studies 329 (Converse and Cox, 2005) showed that cultures grown in 7H9 medium suppressed ESX-1 330 secretion of EsxB, while those grown in Sauton's media were proficient in EsxB secretion 331 (reproduced in Figure 3A). We tested whether these differences in medium affected ESX-1 332 focus formation. Cells were grown to exponential phase in either 7H9 or Sauton's medium and We systematically altered the level of nitrogen from 1.02 mM nitrogen (M63) to 6.26 mM (M63 349 N+) characteristic respectively of 7H9 and Sauton's. Secretion of the ESX-1 substrate, EsxB, into 350 the spent media only occurred in the M63 N+ medium ( Figure 3B). We confirmed that this 351 secretion of EsxB was dependent on the ESX-1 system ( Figure 3C). We next tested whether cells 352 lacking ESX-1 show clumping in low nitrogen (M63) and dispersal in high nitrogen (M63 N+). 353 Cultures were analyzed by macroscopic analysis of growths in a culture tube, as reported 354 prior (DePas et al., 2019). The growths showed that deletion mutants of ESX-1 did not have 355 differential growth from wild-type cells in either medium (Supplemental Figure 4).  To investigate if nitrogen was sufficient to trigger the dissolution of ESX-1 foci, we imaged 427 EccCb1-EGFP grown in M63 supplemented with varying concentrations of nitrogen. EccCb1-428 EGFP cultures grown in M63 medium exhibited foci formed at 100% of cell-cell contacts, similar 429 to those observed in 7H9 medium ( Figure 4C and 4D). Stepwise addition of NH4Cl to the growth 430 medium led to dissipation of foci as a function of NH4Cl concentration; at 5 mM NH4Cl 15% of 431 contacts retained focus formation, at 10 mM NH4Cl 3% of contacts retained focus formation and 432 at 20 mM NH4Cl 0% of contacts exhibited focus formation ( Figure 4C and 4D).  , amt1 and amtB), the GlnR 506 operon and nitrate assimilation genes, in M63 N+ ( Figure 5A). The differential transcription of 507 these genes remained unchanged in our ESX-1 deletion strains, which was confirmed by RT-508 qPCR ( Figure 5B). Thus, ESX-1 was not required for the general transcriptional response to 509 nitrogen ( Figure 5A). replication machinery genes, imuA, imuB, dnaE2, and recA ( Figure 5D). Interestingly, baseline 527 levels of these elements were generally higher in ESX-1 knockout cultures grown in M63 528 medium, which is especially evident in recA. This suggests ESX-1 mediates the adaptational 529 response to environmental triggers, such as nitrogen. In the absence of ESX-1, a growing 530 culture might not sense environmental conditions fully and compensates by harboring a higher 531 baseline level of SOS induced error prone DNA replication machinery. 532 Environmental bacteria live in nutritionally dynamic conditions and require mechanisms to sense 591 and respond to local overgrowth. Our studies show that ESX-1 alters its function in response to 592 nitrogen levels. At low nitrogen levels, cells do not secrete; instead ESX-1 assembles into a 593 megacomplex across spanning both plasma membranes within minutes of cell-cell contact (Fig-594 ure 1). At high nitrogen levels, cells secrete ESX substrates, and the megacomplexes disap-595 pears. ESX-1 does not appear to be required for forming cell-cell contacts themselves, as ESX-596 1 mutants still clump (Supplemental Figure 4). Rather, ESX-1 may contribute to sensing cell 597 contacts to regulate downstream signaling pathways. beyond the scope of this work. The formation of ESX-1 foci at cell-cell contacts is certainly 619 consistent with a role in conjugation at junctions but we speculate that ESX-1 may contribute to 620 cell-cell communication in a broader sense ESX-1 is not required for DNA transfer per se, and it 621 has a negative inhibitory role in selecting a specific acceptor strain. Its role in SOS response 622 may also be seen as part of conjugation preparations for horizontal gene transfer, as the donor 623 cell is preparing its genome for rapid transfer of DNA to an acceptor strain (Guerin et al., 2009). 624 In this light, DNA replication induced by the error prone polymerase, DnaE2, would trigger 625 production of DNA to prepare the donor strain for DNA transfer. In the absence of ESX-1, 626 conjugation is less regulated in the donor strain; any given donor strain is prepared for rapid 627 DNA transfer to an acceptor strain (Derbyshire and Gray, 2014;Gray et al., 2016). This may be 628 caused in part by the constitutive increase in basal transcription of SOS response elements 629 ( Figure 5), which as such could represent preparation for conjugation (Flint et al., 2004). Targeting plasmid pKM468 and gene targeting ultramers (Supplemental Table 1 which had been coated with PDMS prior to imaging. PDMS was prepared by mixing a 1:10 ratio 720 of curing agent to PDMS and applied to the channels. Excess PDMS was removed from the 721 channels by blowing air through individual channels, then the PDMS was cured by incubation in 722 an oven set to 80°C for 20 minutes. Cells were allowed to settle to the bottom of the channel for 723 10 minutes and then washed with appropriate medium. The chamber was kept at 37°C in a 724 temperature-controlled enclosure (OkoLab) throughout imaging. Cultures were inoculated into the appropriate medium from a colony growing on a 7H9 plate. 738 The cultures were then transferred and grown to mid-log phase (OD600 nm ~0.6-0.8). A second 739 transfer of the cultures was done into medium lacking tween and allowed to reach an OD600 nm of 740 0.8. Cells were harvested by centrifugation at 3000 RPM in an Eppendorf centrifuge. The 741 supernatant was filtered through a 0.22 µm filter and concentrated 500-fold using a 3,000 742 MWCO Amicon ultra-15 filter (25 mL to 50 µL). Whole cell pellets were re-suspended in PBS. 743 Whole cell resuspensions and culture concentrates were incubated with SDS loading dye and 744 analyzed by SDS-PAGE using Invitrogen Bis Tris SDS gels. For western blotting of M. 745 smegmatis EsxB a polyclonal mouse antibody was raised using an EsxB antigenic peptide 746 sequence. The anti-GroEL antibody produced in rabbit (Sigma G6532-.5ML) was used to 747 represent sample integrity. 748 749 RNAseq sample preparation 750 RNA was extracted from cultures grown to OD600 nm ~ 0.7-0.8. Three biological replicates (single 751 colonies) were prepared for every strain assayed. The NEB RNA extraction kit was used as 752 directed except for the lysis step, which was completed by bead beating the cells with 0.1 mm 753 zirconia beads. To deplete rRNA, we used the NEB rRNA kit as directed. Samples were 754 prepared using the NEBNext Ultra RNA library prep kit for Illumina, and barcoding oligos were 755 used to pool the libraries. Two technical replicates were prepared for analysis. The resulting 756 pools were sequenced with the Illumina NovaSeq in 75 nt single end reads, resulting in 15M 757 reads per sample. All constructed strains are reported in Supplementary Table 1