The unconventional cytoplasmic sensing mechanism for ethanol chemotaxis in Bacillus subtilis

Motile bacteria sense chemical gradients using chemoreceptors, which consist of distinct sensing and signaling domains. The general model is that the sensing domain binds the chemical and the signaling domain induces the tactic response. Here, we investigated the unconventional sensing mechanism for ethanol taxis in Bacillus subtilis. Ethanol and other short-chain alcohols are attractants for B. subtilis. Two chemoreceptors, McpB and HemAT, sense these alcohols. In the case of McpB, the signaling domain directly binds ethanol. We were further able to identify a single amino-acid residue Ala431 on the cytoplasmic signaling domain of McpB, that when mutated to a serine, reduces taxis to ethanol. Molecular dynamics simulations suggest ethanol binds McpB near residue Ala431 and mutation of this residue to serine increases coiled-coil packing within the signaling domain, thereby reducing the ability of ethanol to bind between the helices of the signaling domain. In the case of HemAT, the myoglobin-like sensing domain binds ethanol, likely between the helices encapsulating the heme group. Aside from being sensed by an unconventional mechanism, ethanol also differs from many other chemoattractants because it is not metabolized by B. subtilis and is toxic. We propose that B. subtilis uses ethanol and other short-chain alcohols to locate prey, namely alcohol-producing microorganisms. Importance Ethanol is a chemoattractant for Bacillus subtilis even though it is not metabolized and inhibits growth. B. subtilis likely uses ethanol to find ethanol-fermenting microorganisms for prey. Two chemoreceptors sense ethanol: HemAT and McpB. HemAT’s myoglobin-like sensing domain directly binds ethanol, but the heme group is not involved. McpB is a transmembrane receptor consisting of an extracellular sensing domain and a cytoplasmic signaling domain. While most attractants bind the extracellular sensing domain, we found that ethanol directly binds between inter-monomer helices of the cytoplasmic signaling domain of McpB, using a mechanism akin to those identified in many mammalian ethanol-binding proteins. Our results indicate that the sensory repertoire of chemoreceptors extends beyond the sensing domain and can directly involve the signaling domain.

and Fig. S3A). We next aligned the amino-acid sequences 231 spanning residues 392 to 434 on the N-helix and neighboring residues 578 to 620 on 232 the C-helix of McpB, McpA, TlpA, and TlpB. (Fig. S3B). Among the 20 putative binding 233 residues, Thr 424 , Asp 427 , and Ala 431 on the N-helix and Glu 581 and Lys 585 on the C-helix 234 were not conserved between the four chemoreceptors and, thus, were targeted for The resulting STD spectra showed reduced peaks near 1.05 ppm and 3.51 ppm as 273 compared to wild-type McpB C (Fig. 4E). The 295 Our analyses identified another interesting feature of ethanol binding, namely that it 296 is able to penetrate the surface of the McpB cytoplasmic domain to bind within the core 297 of the coiled coil. In particular, we observed that ethanol entered between the individual 298 helices of the four-helix bundle at two locations in the methylation-helix region: one 299 involving N-helix residues 393-400 and C-helix residues 613-617 and another involving 300 N-helix residues 382-387 and C-helix residues 628-631 (Fig. S5A). While ethanol 301 binding to these regions was observed in both the wild-type and A431S mutant 302 simulations, the wild-type binding events resulted in longer dwell times, giving rise to the 303 difference in ethanol coordination observed in these regions (Fig. S5A). Preliminary 304 analysis of the two sites, however, suggests they do not themselves play a significant 305 role in signaling. The latter is located outside the region involved in ethanol sensing (see 306 Fig. 3C) and the former, except for residue Glu 399 , is highly conserved among the four 307 chemoreceptors (see Fig. S3B). Indeed, we did not observe a significant reduction in 308 response to ethanol compared to the wild-type control when we tested a mutant 309 expressing mcpB-E399K as its sole chemoreceptor in the capillary assay (569 ± 29.1 310 cells versus 586.1 ± 9.0 cells, respectively). Nevertheless, these observations hint at a 311 signaling mechanism in which ethanol may penetrate to the core of the cytoplasmic 312 domain where it can affect the packing and overall stability of the bundle.

313
To further investigate the above packing hypothesis, we analyzed the strength of   The resulting 1 H and STD spectra with the HemAT N showed clear peaks near 1.05 ppm 332 and 3.51 ppm, which correspond to the -CH 3 and the -CH 2 moieties of ethanol.

333
Additionally, the ratio of areas in the STD spectra compared to 1 H spectra was 0.27 for 334 the -CH 3 moiety and 0.85 for the -CH 2 moiety, suggesting that -CH 2 moiety of ethanol is 335 closer to the protein than its -CH 3 moiety. The STD spectra with the HemAT signaling 336 domain, however, showed negligible peaks near the expected chemical shift values 337 (1.05 ppm and 3.51 ppm) (Fig. 5A). These results collectively indicate that ethanol 338 binds the sensing domain of the HemAT.

339
The sensing domain of the HemAT dimer is composed of a four-helical bundle as its 340 core and a heme group in each subunit (Fig. 5B), which is known to bind molecular oxygen (33). UV-spectral analyses have shown that the oxygen molecule binds the 342 heme group by forming hydrogen bonds with 6-coordinate ferrous heme (34, 35). To 343 determine whether the heme group also interacts with ethanol, we conducted UV

360
We found that B. subtilis performs chemotaxis to multiple short-chain alcohols.  Among the alcohols tested, ethanol is the most likely physiological attractant, 373 because it is produced by many microorganisms and is prevalent in nature (10). As a 374 consequence, we focused on this chemical. Curiously, ethanol is not consumed by B. 375 subtilis, suggesting that it is used for purposes other than nutrition. One possibility is 376 that B. subtilis uses ethanol to locate prey, which could potentially explain why B.

377
subtilis is attracted to a chemical that nominally inhibits its growth. The most likely prey 378 are Crabtree-positive yeast such as Saccharomyces cerevisiae, which produce ethanol 379 at high concentrations even during aerobic growth (37). Indeed, B. subtilis inhibits the 380 growth of S. cerevisiae through the production of antifungal compounds (38) (Fig. S6).
In addition to fermenting microorganisms, plants can also produce ethanol at   However, this bacterium consumes alcohols. In addition, it does not directly sense these 413 alcohols but rather the byproducts of their degradation, namely carboxylic acid.

414
Perhaps the most interesting aspect of ethanol taxis involves the sensing 415 mechanism. Typically, small-molecule attractants bind the extracellular sensing domain.

416
The main exceptions are PTS sugars, which are sensed indirectly through the PTS  Analysis of these binding sites suggests that ethanol preferentially binds helices with 451 amphipathic surfaces (45, 55, 56). The sensing mechanisms for these proteins typically 452 involve replacement of water molecules with ethanol within small hydrophobic cavities 453 between two or more helices. Indeed, an analogous mechanism appears to be 454 employed by the B. subtilis chemoreceptors.

565
The collected HemAT N protein samples were concentrated using an Amicon 566 ultrafiltration cell (Millipore) and dialyzed into dialysis buffer (50 mM Tris, 300 mM NaCl, 567 pH 8) at 4 °C and aliquots were stored at -80°C.  cultures were grown to stationary phase prior to sonication (7 x 10 s pulses), and 635 soluble cell extracts were obtained by centrifugation (7000 x g at 4°C for 10 min).

636
Alcohol dehydrogenase enzyme assays were performed as described previously (63).

637
Briefly, the assay reactions were prepared with 22 mM sodium pyrophosphate (pH 8.8),  Raw data for all experiments are provided as Data set S1.