The C. difficile clnRAB operon initiates adaptations to the host environment in response to LL-37

To cause disease, Clostridioides (Clostridium) difficile must resist killing by innate immune effectors in the intestine, including the host antimicrobial peptide, cathelicidin (LL-37). The mechanisms that enable C. difficile to adapt to the intestine in the presence of antimicrobial peptides are unknown. Expression analyses revealed an operon, CD630_16170-CD630_16190 (clnRAB), which is highly induced by LL-37 and is not expressed in response to other cell-surface active antimicrobials. This operon encodes a predicted transcriptional regulator (ClnR) and an ABC transporter system (ClnAB), all of which are required for function. Analyses of a clnR mutant indicate that ClnR is a pleiotropic regulator that directly binds to LL-37 and controls expression of numerous genes, including many involved in metabolism, cellular transport, signaling, gene regulation, and pathogenesis. The data suggest that ClnRAB is a novel regulatory mechanism that senses LL-37 as a host signal and regulates gene expression to adapt to the host intestinal environment during infection.

8 175 RNA-seq results from the clnR mutant and parent strain were validated by qRT-PCR for 176 several apparent ClnR-dependent genes, as well as analysis of expression in the clnAB mutant and 177 complemented strains ( Table S6). Comparisons of clnR and clnAB mutant expression revealed that 178 the regulator and transporter disruptions resulted in disparate effects on the transcription of some 179 genes (vanZ1, cdd4 and iorA) ( Table S6). These disparate effects are most prominent in the 180 presence of LL-37, highlighting that ClnR activity is dependent on LL-37. In some cases, the clnR and 181 clnAB mutants had similar levels of expression, suggesting that the ClnAB transporter function is 182 important for the activation of ClnR, whereas in other cases (vanZ1, cdd4 and iorA) expression 183 diverged in the clnR and clnA mutants, suggesting a role for the ClnAB transporter in the regulation of 184 some LL-37-dependent genes, independent of ClnR. Complementation of clnR and clnAB was 185 performed by restoring the entire clnRAB operon in trans, as restoration of expression was not 186 possible using only the disrupted clnR or clnAB, respectively (Fig. S2B). As a result, the clnR 187 complemented strain expresses the acquired copy of clnRAB and the native clnAB. Similarly, the 188 clnAB complement expresses the native clnR and the acquired clnRAB. The altered ratios of 189 components in these complement strains (Fig. S2B) may account for some discrepancies noted in 190 these strains, further highlighting the complicated relationship between ClnR and ClnAB in regulation. 191 These complex gene regulatory patterns suggest that multiple factors are involved in the transcription 192 of some ClnR-regulated genes, and that ClnR has both direct and indirect effects on the expression 193 of some loci.

195 ClnR conditionally represses and induces clnRAB in response to LL-37 196
To understand the molecular mechanism of ClnR function, we further explored regulation of 197 the clnRAB locus. ClnR is annotated as a GntR-family transcriptional regulator, and protein sequence 198 comparisons suggest that it is a member of the YtrA sub-family of GntR regulators (Fig. S4). GntR-199 family regulators are most common among bacteria that inhabit complex environmental niches (29).
200 The YtrA sub-family regulators are often found in conjunction with ABC-transporters and are typically 10 227 The clnR mutant grew less well than other strains in MM, MM with glucose, and MM with NAG, 228 suggesting that ClnR is important for the utilization of peptides, glucose and NAG (Fig. 5). When a 229 low concentration of LL-37 (0.5 µg/ml; 1/30 MIC) was added to the growth medium, the clnR and 230 clnAB mutants grew better than the parent strain in base MM, MM with mannitol, MM with NAG, and 231 MM with EA supplementation (Fig. 5), suggesting that ClnRAB is important for growth in a variety of 232 nutrients in the presence of LL-37. These results indicate that the changes in growth observed with 233 low levels of LL-37 are due to changes in ClnR-dependent bacterial metabolism, rather than the 234 antimicrobial activity of LL-37. Moreover, the data strongly suggest that the growth and metabolism 235 delays observed in LL-37 are mediated by ClnR through repression and activation of metabolic gene 236 expression (Tables S4, S5).

238 ClnRAB modulates growth and virulence in vivo 239
As LL-37 is a host-produced peptide and C. difficile inhabits the gastrointestinal tract, the 240 natural consequences of ClnR-LL-37-dependent gene regulation would appear during the growth of 241 the pathogen in the host intestine. To examine the effects of cln mutants in vivo, we used the hamster 242 and mouse models of CDI. Like humans, hamsters and mice are sensitive to colonization by C. 243 difficile, especially after receiving antibiotics, and both produce cathelicidins similar to LL-37 ( Fig. S3) 244 (34-36). Syrian golden hamsters are acutely susceptible to infection by C. difficile, with as few as 1 to 245 10 CFU needed to produce fulminant disease (37). Mice are naturally not as susceptible to CDI as 246 hamsters and require 10 4 -10 7 CFU to achieve colonization, usually with low morbidity (38). For these 247 reasons, the hamster model is most useful for examining early colonization and virulence, while mice 248 allow for assessment of long-term C. difficile colonization (39). 249 Hamsters were infected with spores of 630∆erm, clnR, or clnAB strains and monitored for 250 symptoms of infection as described in the Methods. Hamsters infected with either the clnR or clnAB 251 mutant strains succumbed to infection more rapidly than animals inoculated with the parent strain, 252 indicating that the clnR and clnAB mutants are more virulent (mean time to morbidity: 46.0 ± 12.2 h 11 253 for 630∆erm, 32.5 ± 5.8 h for clnR, 35.2 ± 6.1 h for clnAB; Fig. 6A). C. difficile disease is mediated by 254 the two primary toxins, TcdA and TcdB. To determine whether the increased virulence of the cln 255 strains was related to increased toxin levels, we extracted RNA from cecal samples collected from 256 animals at the time of morbidity and performed digital droplet PCR for absolute quantification of tcdA, 257 tcdB, and clnR expression (Fig. S5). No significant differences in toxin expression were apparent at 258 the time of morbidity. Because only one timepoint could be assessed, the results do not resolve 259 whether the clnR and clnA mutants had altered toxin expression during the course of infection. But, 260 the time from infection to morbidity for clnR and clnAB infections indicate that these mutants produce 261 toxin earlier in the course of infection, resulting in earlier symptoms of disease and morbidity. 262 To assess C. difficile colonization by the different strains, hamster fecal samples were taken at 263 12 h post-infection and plated onto selective medium. C. difficile was recovered from fecal samples in 264 significantly more animals infected with the clnR strain than in the 630∆erm-infected group, 265 suggesting that the clnR mutant colonizes the hamster intestine more rapidly (Fig. 6B). In addition, 266 the clnR mutant reached a higher bacterial burden at the time of morbidity (1.2 x 10 7 CFU/ml for 267 630∆erm, 2.7 x 10 7 CFU/ml for clnR; Fig. 6C). These results illustrate that the clnRAB operon plays a 268 significant role in the colonization and virulence dynamics of C. difficile hamster infections. 269 The colonization results in hamsters suggested a role for ClnRAB in colonization dynamics, 270 which was further examined in the mouse model. Mice were infected with spores of either 630∆erm, 271 clnR, clnAB, clnR Tn::clnRAB, or clnAB Tn::clnRAB strains and monitored for colonization and 272 disease as described in the Methods. Mice infected with clnR lost less weight and recovered more 273 quickly than mice infected with 630∆erm (Fig. 7A). Additionally, mice infected with clnR cleared the 274 bacteria more quickly, with fewer animals having detectable CFU in their feces ( Fig. 7B; Fig. S6).
275 While the impact on short-term colonization and virulence in hamsters and longer-term colonization 276 and persistence in mice are contrasting, the results from both animal models strongly suggest that 277 ClnR contributes to the ability of C. difficile to initiate colonization, cause disease, and persist in the 278 intestinal environment.

279
Considering that differences in either sporulation or germination rates can also influence 280 virulence and bacterial burden in vivo, we assessed sporulation and germination for the clnR and 281 clnAB mutants for defects in either process. No significant difference in sporulation or germination 282 rates was observed for either mutant (Fig. S7). 283 284 ClnRAB and LL-37 promote toxin production 285 Since toxin production is the primary virulence factor leading to C. difficile symptoms, we 286 further investigated the effects of LL-37 and ClnRAB on toxin production under more controlled 287 conditions in vitro. qRT-PCR analysis of tcdA and tcdB transcription was assessed for the cln mutants 288 and parent strain during logarithmic phase growth in BHIS medium, with or without added LL-37. As 289 shown in Fig. S8A, B, LL-37 exposure resulted in increased expression of tcdA (4.6-fold) and tcdB 290 (2.2-fold) in wild-type cells. In contrast, the clnR and clnAB mutants demonstrated lower expression of 291 toxins in LL-37, suggesting that ClnRAB is partially responsible for LL-37-dependent regulation of 292 toxin expression in vitro. Toxin expression is known to be controlled by several regulatory factors, 293 many of which respond to low nutrient availability and/or the transition to stationary phase growth (40). 294 To determine which of the toxin regulators may be influenced by LL-37, we examined expression of 295 regulators and regulator-dependent factors, including tcdR, sigD, ilcV (as an indicator of CodY 296 activity), and CD0341 (as an indicator of CcpA activity) ( Table S8). Of these, only ilvC expression is 297 statistically altered in LL-37; however, the increase in ilvC expression is far more modest (2.2-fold 298 increase) than would be expected with robust CodY activation (18, 19).

299
Because toxin production is typically low at mid-logarithmic phase in BHIS medium, we also 300 examined toxin protein levels after 24 h growth in TY medium. Western blot analysis indicated that 301 TcdA levels were significantly lower in the clnR mutant in TY medium, relative to the parent strain.
302 When cells were grown in medium supplemented with LL-37, final TcdA levels decreased about 3-303 fold in the parent strain (Fig. S8C). In contrast, TcdA levels did not change for the clnR and clnAB 304 mutants in LL-37. While these findings contradict the induction of tcdA expression observed at log-13 305 phase in BHIS medium, the data support the observation that LL-37 and ClnRAB influence toxin 306 expression, and that the outcome of this regulation on toxin production is dependent on growth 307 conditions. These observations provide further evidence that the ClnRAB system is involved in toxin 308 production and that this system is necessary for the influence of LL-37 on toxin production. 309 310 ClnR acts as a DNA-binding regulator that binds multiple promoters 311 As a predicted GntR-family transcriptional regulator, we hypothesized that ClnR binds DNA.
312 Because expression results suggested that ClnR is autoregulatory (Fig. 4), we initially tested whether 313 ClnR directly regulates its own promoter. We produced recombinant His-tagged ClnR and performed 314 gel shifts with fluorescein-labeled DNA of the 84 bp upstream of the predicted clnR transcriptional 315 start site. This DNA fragment was selected because it encompasses a predicted σ A -dependent 316 promoter with -10 (at -52 to -47 bp) and -35 (at -73 to -68 bp) consensus sequences and a tandem 317 repeat sequence (at -46 to -16 bp) that includes a possible ClnR-binding site (Fig. S9). Incubation of 318 His-ClnR with this DNA fragment resulted in a shift visible after electrophoresis, both with and without 319 LL-37 (Fig. 8A). The apparent K d value for this interaction was calculated to be 118 nM (± 40 nM) 320 without LL-37 and 85 nM (± 7 nM) with LL-37, indicating that the affinity of ClnR for this DNA 321 sequence does not change significantly in the presence of LL-37 in these conditions. 322 Additional ClnR-regulated promoters were examined for direct binding, including predicted 323 upstream promoter elements for the metabolic operons grd (CD630_23540), mtl (CD630_23340), and 324 ior (CD630_23810); other transcriptional regulators, including sigU (csfU, CD630_18870) and 325 CD630_16060, as well as the uncharacterized vanZ ortholog (CD630_12400). ClnR bound to all of 326 these promoter sequences but exhibited specificity for PvanZ, PCD1606, and PsigU, with or without 327 LL-37 (Fig. 8B-D). Binding was less specific for Pgrd, Pmtl and Pior under the conditions tested (Fig.  328 S10). The calculated apparent K d for PvanZ was 141 nM (± 59 nM) without LL-37 and 139 nM (± 33 329 nM) with LL-37, the apparent K d for PCD630_16060 was 1.9 µM (± 0.2 µM) without LL-37 and 2.6 µM 330 (± 0.5 µM) with LL-37, and the apparent K d for PsigU was 2.5 µM (± 0.2 µM) without LL-37 and 4.2 14 331 µM (± 3.4 µM) with LL-37 (Fig. 8). Because the apparent K d values for these targets are similar both 332 with and without LL-37, it does not appear that LL-37 influences ClnR binding of these targets in 333 these in vitro conditions. 334 335 ClnR directly binds LL-37 336 Although the EMSA did not uncover differences in ClnR-DNA binding in the absence or 337 presence of LL-37 in vitro for the DNA targets examined, our expression and growth data suggest 338 that ClnR regulates transcription in response to LL-37. Based on clnR mutant phenotypes for cells 339 grown with and without LL-37, we hypothesized that ClnR directly binds LL-37 to regulate ClnR 340 activity. To test this hypothesis, we performed surface plasmon resonance (SPR) to examine the 341 binding kinetics of ClnR with LL-37. These experiments showed that His-ClnR interacts with LL-37 342 with a K d of 83 ± 14 nM (Fig. 9A). This K d value is similar to the apparent K d values that were 343 calculated for the affinity of His-ClnR for Pcln DNA. This result suggests that the concentrations of 344 ClnR needed to interact with both LL-37 and Pcln DNA are similar, and that interactions of these 345 three components could occur simultaneously. Furthermore, the interaction between His-ClnR and 346 LL-37 is specific, as SPR using scrambled LL-37 found no apparent interaction between these 347 molecules (Fig. 9B) Fig. 3). The regulation of metabolic pathways in response to LL-37 was 360 previously observed in other pathogens, including S. pyogenes, E. coli, P. aeruginosa, and S. 361 pneumoniae, but their role in the bacterial response to LL-37 has not been clear (43, 46-48). Our data 362 indicate that the regulation of genes by LL-37/ClnRAB in vivo has notable effects on C. difficile 363 colonization and virulence (Fig. 6, Fig. 7). Results from the mouse infection model suggest that 364 disruption of ClnRAB results in dysregulation of metabolism that significantly hinders the ability of C. 365 difficile to colonize; whereas in the exquisitely toxin-sensitive hamster model, the metabolism defects 366 in cln mutants quickly progress to nutrient deprivation and toxin production. These effects are not 367 unexpected, given that nutrient deprivation is demonstrably the primary factor driving C. difficile toxin 368 expression (20, 32, 40, 49-54). C. difficile possesses an unusual metabolic repertoire for energy 369 generation, including solventogenic fermentation (55, 56), Stickland (amino acid) fermentation (31), 370 and autotrophic growth via the Wood-Ljungdahl pathway (57). However, the importance of most of 371 these individual metabolic pathways for growth and virulence in vivo has not been determined. 372 Because ClnR is a global regulator that negatively and positively influences the expression of multiple 373 metabolic pathways, many of which are constitutively expressed in a clnR mutant, we cannot infer 374 which of these pathways are most influential for host pathogenesis. Determining how and which 375 ClnR-controlled pathways and mechanisms influence disease could expose potential vulnerabilities of 376 C. difficile that may be exploited to prevent infections. 377 Overall our results indicate that ClnRAB responds directly and specifically to LL-37 without 378 conferring LL-37 resistance and suggest that ClnR responds to LL-37 as an indicator of the host 379 environment, conferring a colonization advantage. The clnRAB locus is highly conserved in C. difficile, 380 with representation at  99% amino acid sequence identity in over 500 strains at the time of this 381 publication (NCBI, BLASTp). The data strongly suggest that ClnR acts as a pleiotropic regulator in C.