Two Accessory Proteins Govern MmpL3 Mycolic Acid Transport in Mycobacteria

The cell envelope of Mycobacterium tuberculosis, the bacterium that causes the disease tuberculosis, is a complex structure composed of abundant lipids and glycolipids, including the signature lipid of these bacteria, mycolic acids. In this study, we identified two new components of the transport machinery that constructs this complex cell wall. These two accessory proteins are in a complex with the MmpL3 transporter. One of these proteins, TtfA, is required for mycolic acid transport and cell viability, whereas the other stabilizes the MmpL3 complex. These studies identify two new components of the essential cell envelope biosynthetic machinery in mycobacteria.


Results 121
MmpL3 is stably associated with two proteins of unknown function, MSMEG_0736 122 and MSMEG_5308 123 In order to discover stable binding interactions with MsMmpL3 in situ, we devised 124 a native, stringent, affinity purification. MsMmpL3 was fused to a flexible linker 125 connecting the C-terminus of MmpL3 to monomeric superfolder GFP (msfGFP) at the 126 native chromosomal locus of MmpL3. As mmpL3 is an essential gene, the normal 127 growth rate of this strain suggests that fusion did not disrupt the essential function of the 128 protein. Cell membranes were collected and solubilized with the mild detergent n-129 Dodecyl β-D-maltoside (DDM). Anti-GFP nanobodies covalently linked to a magnetic 130 bead were incubated with detergent-solubilized membranes and then extensively 131 washed with 0.2% DDM containing buffer. Co-purifying proteins were identified via 132 shotgun mass spectrometry (Fig 1). One of the most abundantly co-purifying proteins 133 was a protein of unknown function, MSMEG_0736 (Table 1 and Table S1A,B). In 134 contrast, pulldown of MmpL10, another MmpL transporter, did not copurify 135 MSmeg_0736 or any proteins in common with MmpL3 (Table S1A,B). To validate this 136 interaction, we created a strain in which a msfGFP was fused to MSMEG_0736. When 137 MSMEG_0736-msfGFP was purified from detergent solubilized membranes under the 138 same conditions, the most abundantly copurified protein was MsMmpL3 (Table 1 and  139   Table S1A,B). In a control experiment using MSMEG_0410 (MmpL10) fused to msfGFP 140 as a bait, neither MSMEG_0736, MSMEG_0250 or MSMEG_5308 were co-purified 141 (Table 1 and Table S1A,B). In a biological replicate of the MSmeg_0736 pulldown, we 142 confirmed the identity of the prominent band at approximately 100 kDa as MmpL3 (Fig  143  1B, Table S2). As MSMEG_0736 interacts with MmpL3, and evidence we will present in 144 this paper shows MSMEG_0736 is required for TMM transport, we propose 145 MSMEG_0736 be named "TMM transport factor A", TtfA. 146 Analysis of MsTtfA and MsMmpL3 copurifying proteins identified by anti-GFP 147 nanobody purification showed a third complex member found in both pulldowns, the 148 protein encoded by MSMEG_5308. This seven bladed beta-propeller protein has a 149 homolog in M. tuberculosis, Rv1057, that has been shown to be non-essential, although The M. tuberculosis H37Rv homolog of TtfA is Rv0383c. rv0383c was predicted 158 to be an essential gene in H37Rv based on transposon mutagenesis [19,42], but its 159 essentiality in M. smegmatis and M. tuberculosis is unknown and its molecular function 160 obscure. With no predicted protein domains or homologs of known function, 161 confirmation of its essentiality in both organisms was the first step to analyze its 162 function. To test the essentiality of ttfA in M. smegmatis, we generated a merodiploid 163 strain in which a second copy of ttfA was integrated in the chromosome. We then 164 deleted the endogenous coding sequence, so that the only a single copy of ttfA 165 remained at the attB site. We then attempted to remove the second copy of ttfA 166 from attB by marker exchange with either a vector or a plasmid encoding TtfA and 167 conferring kanamycin resistance, pAJF792 [43]. Only transformation with the plasmid 168 encoding TtfA yielded transformants that were kanamycin resistant and streptomycin 169 sensitive. Similar results were obtained with a plasmid encoding TtfA from M. 170 tuberculosis (Fig 2A). This inability to remove ttfA from attB in our ΔttfA strain suggested 171 that ttfA was required for growth of M. smegmatis (Fig 2A). To further assess the 172 essential role of MsTtfA, we generated CRISPR interference (CRISPRi) strain that 173 allows anhydrotetracycline (ATc) inducible knockdown [44]. Growth inhibition by gene 174 knockdown was visualized by spotting 10-fold serial dilutions on plates with and without 175 ATc, MsTtfA depletion led to an ATc dependent growth defect not seen in the non-176 targeting control (Fig 2A). Gene knockdown of ttfA in M. smegmatis also led to 177 cessation of growth in liquid media between 9 and 12 hours post induction with ATc (Fig  178   2B). To test whether TtfA was essential in M. tuberculosis, we attempted to knockout 179 the gene using a temperature sensitive phage and were unsuccessful, suggesting 180 essentiality. We then generated three ttfA targeting CRISPRi strains with independent 181 guide RNAs. Gene knockdown of ttfA in M. tuberculosis with all three guide RNAs all led 182 to cessation of growth in liquid media after three days after induction with ATc, 183 indicating that TtfA is essential for M. tuberculosis growth in vitro (Fig 2C). 184 To examine the morphologic changes that accompany growth arrest during loss 185 of MsTtfA, we depleted the protein using CRISPRi and tracked morphological changes 186 using a MalF(1,2)-mCitrine expression strain that uniformly labels the cell membrane. 187 Time-lapse microscopy indicated that growth arrest without MsTtfA was characterized 188 by shorter, misshapen cells ( Fig 2D, Movies S1, S2). Quantitation of cell length 189 revealed that MsTtfA depleted cells were significantly shorter (2.88± 0.89 m) as 190 compared to control cells (6.00±2.03 m) (Fig 2E). The short cell phenotype suggested 191 that MsTtfA might be required for cell elongation. These data indicate that TtfA is 192 essential for mycobacterial viability and that the function of this gene is conserved 193 between fast and slow growing mycobacteria.  (Fig 3B), suggesting that the protein is not secreted. 215 Fractionation of the cell lysate showed that MsTtfA-msfGFP localized in the Trition-X100 216 soluble fraction, similar to a membrane protein control FtsY, but not the soluble fraction 217 marked by cytosolic RNAPb, supporting that MsTtfA is membrane anchored, is not 218 secreted, and has a cytoplasmic C-terminus. 219 The essential portion of TtfA is conserved among mycolate producers 220 To further delineate the functional domains of the protein, we examined the 221 conservation of the protein sequence across homologs. BLAST searches identified 222 homologous predicted proteins among mycolate producing organisms ( Fig S1). 223 Alignments of these homologs suggested that amino acids 1 through approximately 205 224 were well conserved, with poor conservation in the C-terminal 73 amino acids (Fig S1). 225 The C-terminal 73 amino acids are also predicted to be disordered [46]. This lack of 226 conservation at the C-terminus was also apparent in the alignment with the MtbTtfA, 227 which we demonstrate above is functional in M. smegmatis (Fig 2A). To assess the 228 functional contribution of these conserved regions, we generated MsTtfA truncations 229 fused at the C-terminus to msfGFP and assessed the ability of these truncations to 230 complement the essential function by marker exchange. Only the plasmid encoding 231 amino acids 1-205 yielded kanamycin resistant, streptomycin sensitive transformants, 232 indicating that amino acids 1-205 were essential ( Fig S2A). 233 After confirming that all of these truncations accumulate as stable proteins at 234 their predicted sizes when expressed in wild type M. smegmatis (Fig S2A), we localized 235 each truncation by fluorescence microscopy. MsTtfA(1-205aa)-msfGFP localized to 236 poles and septa in a pattern similar to the full-length protein (Fig S2B), indicating that 237 the poorly conserved C-terminus is not required for essential function or proper 238 localization. However, truncations shorter than 205AA, which did not complement 239 essential function, also failed to localize to poles and septa, indicating that the first 240 205AA of the protein, including the N terminal transmembrane domain, are required for 241 proper localization and that this localization is tightly linked to its essential function. 242

The N-terminus of TtfA is required for interaction with MmpL3 243
To determine the regions of MsTtfA required for interaction with MmpL3, we 244 immunopurified MsTtfA truncations fused to msfGFP when coexpressed with MmpL3-245 mCherry. MsTtfA-msfGFP was purified from DDM detergent solubilized lysates with 246 GFPTrap resin. Unfused msfGFP did not coprecipitate MmpL3-mCherry, whereas full-247 length MsTtfA-msfGFP copurified with MmpL3-mCherry ( Fig 4A). All truncations were 248 visible at comparable levels in DDM solubilized lysates at their predicted sizes ( Fig 4B). TtfA-msfGFP-MmpL3-mCherry complex ( Fig 4C). These results indicate that active 264 TMM biosynthesis is not required for TtfA-MmpL3 complex formation. 265 MmpL3-GFP has been previously reported to localize to cell poles and septa 266 [49], a finding we confirm with our MmpL3-msfGFP strain, which localizes the MmpL3 CRISPRi led to cessation of growth between 6 and 9 hours, but did not affect 273 localization of TtfA-msfGFP or MmpL3-msfGFP, again indicating that TMM synthesis 274 was not required for localization of either protein to the poles or septa (Fig 5B). Taken cultures as compared to replete cultures (Fig 6A,B). As a control for essential protein 290 depletion, we depleted the essential DnaK chaperone [43] and found no effect on 14 C-291 TMM/ 14 C-TDM, indicating that cell arrest by depletion of any essential protein does not 292 alter TMM and TDM levels ( Fig S4). MSMEG_5308 was also found to co-purify with both MsTtfA and MmpL3 (Fig 1  302 and Table 1). To further investigate this MmpL3 complex member, we generated a C-303 terminal msfGFP fusion to MSMEG_5308 at the chromosomal locus. We then depleted 304 either MmpL3 or TtfA in the MSMEG_5308-msfGFP strain. Either MmpL3 or TtfA 305 depletion, but not non-targeting control, led to accumulation of MSMEG_5308 protein 306 ( Fig 7A). In contrast, CRISPRi depletion of Pks13 led to cessation of cell growth after 6 307 hours of induction, but did not induce MSMEG_5308 accumulation (Fig 7A). 308 We further examined the response of MSMEG_5308 to inhibitors of the 309 TMM/TDM pathway, including early mycolate biosynthesis (isoniazid (INH)), and 310 inhibitors targeting late steps in TMM/TDM transport (SQ109, BM212, and AU1235). SQ109 and AU1235 caused MSMEG_5308-msfGFP accumulation at 1.5 and 3 hours, 315 but INH or BM212 (at 5 and 10 μM) had no effect (Fig 7B and data not shown). The lack 316 of accumulation of MSMEG_5308 with INH treatment or Pks13 depletion suggests that 317 MSMEG_5308 does not accumulate in response to loss of TMM or TDM biosynthesis, 318 but rather inhibition of their transport. 319

MmpL3/TtfA interaction 321
The identification of MSMEG_5308 as an MmpL3/TtfA interacting protein suggested 322 that MSMEG_5308 may co-localize with the MmpL3 complex. Indeed, MSMEG_5308-323 msfGFP localized to cell poles and septa in a pattern similar to both TtfA-msfGFP and 324 MmpL3-msfGFP by live cell fluorescence microscopy ( Fig 8A). To examine the role of 325 MSMEG_5308, we targeted MSMEG_5308 using CRISPRi and verified efficient 326 knockdown using a MSMEG_5308-msfGFP strain ( Fig S5). Depletion of MSMEG_5308 327 had no impact growth or cell morphology, confirming MSMEG_5308 was not essential 328 in M. smegmatis (data not shown). 329 To assess the effect of MSMEG_5308 on MmpL3/TtfA complexes, we isolated 330 TtfA-msfGFP using anti-GFP nanobodies and probed for MmpL3-mCherry in the 331 presence and absence of MSMEG_5308. In DDM solubilized lysates, TtfA copurified 332 with MmpL3-mCherry in MSMEG_5308 depleted lysates similarly to control cells (Fig  333   8B). However, in Triton X-100 solubilized lysates, although the MmpL3-TtfA complex 334 was intact when MSMEG_5308 was present, TtfA-msfGFP did not coprecipitate 335 MmpL3-mCherry in the absence of MSMEG_5308 (Fig 8B). These results indicate that 336 MSMEG_5308 is a nonessential member of the MmpL3 complex that is induced by 337 stress and stabilizes the MmpL3-TtfA protein complex. 338 339

Discussion 340
We have identified two new components of the essential machinery of mycolic acid 341 transport and cell growth in mycobacteria. The MmpL3 transporter was previously 342 known to transport trehalose monomycolate, but its cofactors were unknown. The 343 MmpL3 machinery contains the essential protein TtfA, which we show is required for 344 TMM flipping across the cytoplasmic membrane. A third complex member, 345 MSMEG_5308, while not required for TMM transport, appears to stabilize the MmpL3 346 complex and is upregulated in response to MmpL3 dysfunction. All three of these 347  Similarly, MmpL7 is physically and functionally coupled to PDIM biosynthesis [52]. 363 However, the lack of such coupling in the MmpL3 system may suggest that a coupling 364 protein is required to chaperone the transported glycolipid to the transporter, a function 365 we hypothesize for TtfA. 366 Alternatively, it is possible that that TtfA is a scaffolding protein that nucleates 367 additional essential MmpL3 complex members yet to be elucidated. TtfA has been 368 previously shown to interact with the non-essential vesiclulogenesis regulator VirR in M. 369 tuberculosis, that we also find in our purifications of MsTtfA [53]. 370 The second protein we identify in the MmpL3 complex, MSMEG_5308, is a seven 371 bladed propeller protein. This protein structural motif has been previously described to 372 aid in protein-protein interactions though members are functionally diverse [54-56]. In 373 Mtb, the MSMEG_5308 homolog, Rv1057, is responsive to a variety of membrane 374 stresses as well as MmpL3 depletion. Our data indicates that the function of 375 MSMEG_5308 is to stabilize the MmpL3/TtfA complex. We hypothesize that 376 MSMEG_5308 is upregulated during times of membrane stress in order to stabilize 377

MmpL3 complexes and preserve TMM transport and cell wall biosynthesis in conditions 378
that may dissociate the MmpL3 complex. 379 MmpL3 mediated TMM transport has emerged as an attractive drug target after 380 several high throughput screens identified whole cell active inhibitors that appear to 381 target this transporter. Our identification of two previously unidentified cofactors for 382 MmpL3 will empower future studies to investigate these proteins as drug targets and 383 their potential roles in cellular response and resistance to MmpL3 targeting small 384 molecules. Additionally, future biochemical and structural studies will examine the 385 biochemical and structural organization of this essential mycolic acid transport complex.

Bacterial and DNA manipulations 388
Standard procedures were used to manipulate recombinant DNA and to transform E.  Table S2. 392 Plasmids including relevant features, and primers are listed in Table S3

Microscopy 407
All images were acquired using a Zeiss Axio Observer Z1 microscope equipped with 408 Definite focus, Stage top incubator (Insert P Lab-Tek S1, TempModule S1), Colibri. Samples were analyzed on a Thermo Scientific Orbitrap Fusion mass 476 spectrometry system equipped with an Easy nLC 1200 ultra-high pressure liquid 477 chromatography system interfaced via a nanoelectrospray source. Samples were 478 injected onto a C18 reverse phase capillary column (75 um inner diameter x 25 cm 479 length, packed with 1.9 um C18 particles). Peptides were then separated by an organic 480 gradient from 5% to 30% ACN in 0.1% formic acid over 180 minutes at a flow rate of 481 300 nl/min. The MS continuously collected spectra in a data-dependent fashion over the 482 entire gradient. to just cover the gel pieces. Samples were then placed in a 37ºC room overnight. 510 Peptides were later extracted by removing the ammonium bicarbonate solution, 511 followed by one wash with a solution containing 50% acetonitrile and 1% formic acid. 512 The extracts were then dried in a speed-vac (~1 hr). The samples were then stored at 513 4ºC until analysis. 514 On the day of analysis the samples were reconstituted in 5 -10 µl of HPLC solvent A 515 (2.5% acetonitrile, 0.1% formic acid). A nano-scale reverse-phase HPLC capillary 516 column was created by packing 2.6 µm C18 spherical silica beads into a fused silica 517 capillary (100 µm inner diameter x ~30 cm length) with a flame-drawn tip [60]. After 518 equilibrating the column each sample was loaded via a Famos auto sampler (LC 519 Packings, San Francisco CA) onto the column. A gradient was formed and peptides 520 were eluted with increasing concentrations of solvent B (97.5% acetonitrile, 0.1% formic 521 acid). 522 As peptides eluted they were subjected to electrospray ionization and then entered into 523 an LTQ Orbitrap Velos Pro ion-trap mass spectrometer (Thermo Fisher Scientific, 524 Waltham, MA). Peptides were detected, isolated, and fragmented to produce a tandem 525 mass spectrum of specific fragment ions for each peptide. Peptide sequences (and 526 hence protein identity) were determined by matching protein databases with the contructs were grown to OD600 0.5. For non-targeting and MSMEG_5308 depletion 537 strains were grown with ATc-50ng/ml for 24 hours and Pks13 depletion strains were 538 grown with ATc for 6 hours. Cultures were cooled on ice and cells were harvested by 539 centrifugation (3700g, 10 min, 4°C). Pellets were washed once with 1ml of PBS. Pellets 540 were resuspended in 500ul PBS with 1x HALT protease (Thermo Scientific) and lysed 541 via bead beating (Biospec, Mini-beadbeater-16) 2 times for 1 min with 5 min on ice 542 between. Beads, unbroken cells, and debris were pelleted at 5000g for 10 min at 4°C. 543 Supernatant was collected and an additional 500ul of PBS containing either 1% DDM or 544 1% Triton X-100 was added and incubated at 4°C for 1 hour with rocking. Insoluble 545 material was then pelleted at 21130g for 1 hour at 4°C and the supernatant (~1ml) was 546 collected and added to 20ul pre-washed GFPTrap magnetic agarose beads (Bulldog 547 Bio) and incubated for 2 hours at 4°C with rocking. After incubation beads were 548 collected with a magnet and washed 3 times with 1mL PBS and 0.1% DDM or Triton X-549 100. Elution was done using SDS sample buffer and heating 60°C for 15