CRISPR-Cas blocks antibiotic resistance plasmid transfer between Enterococcus faecalis strains in the gastrointestinal tract

The emergence of antibiotic resistant bacteria is a major public health issue. Antibiotic-resistant bacteria can emerge via the horizontal acquisition of antibiotic resistance genes. This process is especially relevant in the emergence of antibiotic-resistant enterococci, which are among the leading causes of nosocomial infections in the United States. Enterococcus faecalis is host to a class of plasmids, the pheromone-responsive plasmids (PRPs), that mediate inter- and intraspecies transfer of antibiotic resistance genes and other virulence traits. Interestingly, hospital-associated E. faecalis generally lack complete CRISPR-Cas genome defense systems, which reduce the acquisition frequency of PRPs in commensal isolates. The absence of CRISPR-Cas likely contributes to the emergence of multidrug-resistant E. faecalis. Here, we assessed the impacts of in vitro growth conditions (planktonic versus biofilm), production of a PRP-encoded bacteriocin, and in vivo mouse gastrointestinal colonization on CRISPR-Cas efficacy. We found that bacteriocin production significantly impacts PRP transfer efficiency and CRISPR-Cas efficacy under in vitro but not in vivo conditions. Strikingly, we observed that CRISPR-Cas completely prevented PRP acquisition in 85% of mice, with or without a PRP-encoded bacteriocin, a much more pronounced effect than that observed in vitro. Our data suggest that the in vitro and in vivo activities of CRISPR-Cas in E. faecalis are differentially regulated. These results demonstrate that native CRISPR-Cas in E. faecalis confers robust in vivo defense against antibiotic resistance plasmids. Our results are significant because they demonstrate that CRISPR-Cas has a profound effect on the dissemination of antibiotic resistance genes in the gastrointestinal tract.

lack complete CRISPR-Cas genome defense systems, which reduce the acquisition 23 frequency of PRPs in commensal isolates. The absence of CRISPR-Cas likely 24 contributes to the emergence of multidrug-resistant E. faecalis. Here, we assessed the 25 impacts of in vitro growth conditions (planktonic versus biofilm), production of a PRP-26 encoded bacteriocin, and in vivo mouse gastrointestinal colonization on CRISPR-Cas 27 efficacy. We found that bacteriocin production significantly impacts PRP transfer 28 efficiency and CRISPR-Cas efficacy under in vitro but not in vivo conditions. Strikingly, 29 we observed that CRISPR-Cas completely prevented PRP acquisition in 85% of mice, 30 with or without a PRP-encoded bacteriocin, a much more pronounced effect than that 31 observed in vitro. Our data suggest that the in vitro and in vivo activities of CRISPR-Cas 32 in E. faecalis are differentially regulated. These results demonstrate that native 33 CRISPR-Cas in E. faecalis confers robust in vivo defense against antibiotic resistance 34 plasmids. Our results are significant because they demonstrate that CRISPR-Cas has a 35 profound effect on the dissemination of antibiotic resistance genes in the 36 gastrointestinal tract. 37

Introduction 50
Enterococcus faecalis is a gram-positive bacterium and native inhabitant of the 51 gastrointestinal tracts (GI) of humans and other animals (1). E. faecalis is also an 52 opportunistic pathogen that is among the leading causes of hospital-acquired infections 53 in the United States (2-4). Hospital-associated E. faecalis strains are often multidrug-54 resistant (MDR) and can encode resistance to vancomycin, leaving few treatment 55 options (3,5). 56 57 E. faecalis strains acquire novel genetic traits, including antibiotic resistance, via 58 horizontal gene transfer (HGT) (6-8). A common method of HGT in E. faecalis is 59 mediated by the narrow host range pheromone-responsive plasmids (PRPs). The PRPs 60 are a unique class of plasmids that are highly co-evolved with E. faecalis (9, 10). PRPs 61 are large (~60 kb) and conjugate efficiently, resulting in transfer frequencies reaching 1 62 7 sequence identity with the repB gene of pAM714, and acquisition of pAM714 is 119 increased by ~80-fold in E. faecalis T11RFΔcas9 after 18 hours of biofilm mating on an 120 agar surface (35). To gain insight into the kinetics of plasmid transfer, and whether or 121 not culture condition impacts CRISPR-Cas defense, we sampled planktonic and biofilm 122 (agar spread plate) mating reactions over an 18-hour period (Figure 1). All mating 123 reactions were inoculated at a donor to recipient ratio of 1:9. The donor strain used in 124 these experiments was E. faecalis OG1SSp(pAM714). The recipient strains were 125 T11RF and a T11RF derivative with an in-frame deletion of cas9, T11RFΔcas9 (see 126 Table 1 for a list of strains). 127

128
The number of pAM714 transconjugants (i.e., the number of T11RF or T11RFΔcas9 129 cells that acquired pAM714) was used to compare the kinetics of conjugative plasmid 130 transfer in the presence and absence of CRISPR-Cas defense. After only 30 minutes of 131 mating, we observed ~10 3 -10 4 transconjugants for both recipient strains (T11RF and 132 Δcas9) under both planktonic and biofilm conditions (Figure 1A and 1B). Under 133 planktonic mating conditions, the number of T11Δcas9(pAM714) transconjugants 134 remained stable over all time points while the number of T11RF(pAM714) 135 transconjugants decreased ( Figure 1A). This suggests that CRISPR-Cas in T11RF is 136 actively targeting pAM714 in planktonic conjugation by preventing plasmid acquisition 137 and/or by curing T11RF transconjugants of pAM714 (38). The number of 138 transconjugants produced in biofilm matings were comparatively higher (Figure 1B). 139 Both recipient strains achieved transconjugant yields similar to those previously 140 reported for an 18-hour conjugation reaction in a biofilm (35). CRISPR-Cas had a 8 significant impact on transconjugant yield in only a subset of time points, with the most 142 pronounced effect observed at 18 hours (Figure 1B). 143 144 Impact of cytolysin on CRISPR-Cas efficacy. PRPs can encode bacteriocins that 145 promote their maintenance in polymicrobial communities (17,39). pAM714 encodes a 146 bacteriocin, cytolysin, that is bactericidal against gram-positive bacteria including E. 147 faecalis (20). The operon encoding cytolysin also encodes a membrane-associated 148 immunity factor that protects the cytolysin-producing cell from the activity of cytolysin 149 (40). Therefore, pAM714-containing cells (and by extension, newly generated 150 transconjugants) are immune to the bacteriocin. To determine if cytolysin production 151 impacts CRISPR-Cas anti-plasmid defense, we utilized another derivative of pAD1, 152 pAM771, which is cytolysin-negative (41, 42). Under planktonic conditions using 153 OG1SSp(pAM771) as a plasmid donor, we obtained a similar number of 154 T11RF(pAM771) and T11RFΔcas9(pAM771) transconjugants across all time points 155 ( Figure 1C). Moreover, we observed a statistically significant impact of CRISPR-Cas on 156 pAM771 acquisition at only one time point, and the magnitude of the effect was small. 157 These results are in contrast to those observed for pAM714 planktonic matings (Figure 158 1A). These results suggest that cytolysin production influences CRISPR-Cas efficacy at 159 a population level under planktonic mating conditions. In pAM771 biofilm matings, we 160 observed an overall increase in transconjugant numbers by at least one log for all time 161 points and for both wild-type and Δcas9 recipient strains ( Figure 1D), as compared to 162 pAM714 ( Figure 1B). However, the impact of CRISPR-Cas was similar and did not 163 appear to be influenced by cytolysin production.

165
Recipient strain densities are substantially impacted by cytolysin in planktonic 166 mating reactions. Under planktonic mating conditions, we observed a decrease in 167 pAM714 transconjugants over time ( Figure 1A) and a difference in the overall number of 168 T11RF(pAM714) transconjugants compared to T11RF(pAM771) transconjugants after 5 169 and 18 hours of conjugation (Figure 1A and 1C). We hypothesized that pAM714-170 encoded cytolysin impacts plasmid transfer kinetics under planktonic mating conditions. 171 172 Cells possessing pAM714 (i.e., donor strains) are protected from the action of cytolysin 173 through production of an immunity factor (40). T11RF and T11RFΔcas9 recipients do 174 not encode the immunity factor and are vulnerable to the activity of cytolysin. We 175 analyzed total T11RF or T11RFΔcas9 densities in our conjugation reactions (the same 176 reactions as shown in Figure 1). In planktonic mating reactions, T11RF density 177 remained high (~10 8 CFU/mL) and stable when the donor strain harbored pAM771 178 ( Figure 2A). However, in the presence of pAM714 donors, there was a linear decrease 179 in T11RF density that initiated after only 30 minutes of mating (Figure 2A). A similar 180 trend was also observed for the T11RFΔcas9 recipient strain ( Figure 2B). We conclude 181 that under planktonic mating conditions, cytolysin production from pAM714 continuously 182 depletes the recipient cell population, and this is independent of CRISPR-Cas. 183

184
We performed the same analysis for biofilm matings and found that the degree of cell 185 death attributed to cytolysin production from pAM714 was less robust ( Figure 2C and 186 D). Significant changes in T11RF and T11RFΔcas9 did not occur until the later time 187 points of conjugation. From this, we conclude that cytolysin activity has a greater impact 188 on the population dynamics of conjugation reactions under planktonic mating conditions 189 as compared to biofilm mating conditions. 190

191
To obtain better resolution of the recipient cell sensitivity to cytolysin activity, we utilized 192 a cytolysin immunity assay (40). We sought to determine the time point in our pAM714 193 matings at which the recipient populations became immune to cytolysin activity. To 194 isolate only recipient cells from mating reactions, we sampled from our conjugation 195 reactions at set time points, and then performed an outgrowth for 18 hours in antibiotic 196 medium that would eliminate OG1SSp(pAM714) donors (see Materials and Methods). 197 These cultures were then assessed for cytolysin susceptibility in the cytolysin immunity 198 assay. We note that, during this outgrowth, it is possible that additional pAM714 transfer 199 may occur between recipient and transconjugant cells present in these populations. At 24 hours post co-colonization, we observed pAM714 transfer in only one of ten mice 233 when T11RF was the recipient ( Figure 4A). Transconjugants were not recovered in this 234 mouse at subsequent time points. Conversely, pAM714 transconjugants at densities up 235 to 10 6 CFU/g of feces were observed for eight of ten mice colonized with T11RFΔcas9 236 recipients over the course of the experiment ( Figure 4A). We hypothesized that the absence of CRISPR-Cas in the progenitors of hospital-301 adapted E. faecalis allowed for their rapid evolution of multidrug resistance via 302 horizontal gene transfer (32). The results of our current study demonstrate that 303 CRISPR-Cas is, in fact, a robust barrier to plasmid acquisition in vivo. In this study, we 304 did not provide erythromycin during the in vivo mating period, therefore we did not 305 determine the impact of antibiotic selection for the plasmid on experimental outcomes. 306 We predict that there would be very strong selection for CRISPR-Cas mutants or fully 307 CRISPR-Cas-deficient strains that lack this barrier to plasmid acquisition. Further, the 308 impact of CRISPR-Cas should be assessed in multiple in vivo models. Here, we 309 induced gastrointestinal dysbiosis with antibiotics, allowed the mice to recover for one The impact of CRISPR-Cas and cytolysin production on in vivo PRP transfer should be 318 assessed in these models, as well. 319 320 Our study identified key differences concerning CRISPR-Cas defense efficacy, 321 depending on the experimental conditions used and the accessory factors (in this case, 322 bacteriocin production) encoded by the plasmid studied. From the five conditions tested 323 in our study, we observed that CRISPR-Cas activity can range from having no impact 324 on plasmid acquisition to completely blocking plasmid acquisition between the same 325 donor and recipient pair of E. faecalis strains (see Fig S1 for  Cas is attenuated (45). Specifically, E. faecalis can transiently tolerate both an active 342 CRISPR-Cas system and one of its targets, albeit at a fitness cost (34,38,45). 343 Overexpression of cas9 overcomes this phenotype, greatly increasing the efficacy of 344 defense and preventing CRISPR/target co-maintenance (45). Little is known about the 345 regulation of CRISPR-Cas systems in E. faecalis. Our results suggest that they are 346 differentially regulated in vitro and in vivo. There are several possible mechanisms for 347 our observations, including differential transcriptional regulation of cas genes in vitro 348 and in vivo, or differential expression or activity of anti-CRISPR regulatory factors (46) in 349 E. faecalis, although to our knowledge none have yet been identified in this species. 350 Future studies will seek to elucidate the mechanisms underlying the robust in vivo anti-351 plasmid activity of E. faecalis CRISPR-Cas. 352 353

Materials and Methods 354
Bacteria and reagents used. Strains used in this study are shown in Table 1. E. 355 faecalis strains were cultured in brain heart infusion (BHI) broth or on BHI agar at 37°C. 356 Antibiotic concentrations used were as follows: rifampicin, 50 µg/mL; fusidic acid, 25 357 µg/mL; spectinomycin, 500 µg/mL; streptomycin, 500 µg/mL; erythromycin, 50 µg/mL. following day, these donor-deficient cultures were used in a soft agar overlay on normal 387 BHI agar. After allowing the overlay to dry, 5 µL of OG1SSp(pAM714) was spotted on 388 the overlay as an indicator for cytolysin susceptibility. Plates were incubated overnight 389 at 37°C and were inspected for zones of inhibition the following day. 390 391 Mouse model of E. faecalis colonization. Seven days prior to bacterial colonization, 392 6-8 week old C57BL/6 mice (Jackson laboratories) were gavaged with 100 µL of an 393 antibiotic cocktail (streptomycin 1 mg/mL, gentamicin 1 mg/mL, erythromycin 200 394 µg/mL), and given a water bottle ad libitum with the same antibiotic cocktail for 6 days 395 following gavage. 24 h prior to bacterial inoculation, antibiotic water was removed and 396 replaced with standard sterile antibiotic-free water. Bacteria were grown overnight in 397 BHI, and mice were gavaged with 1e 9 CFU/mL in PBS of each bacterial strain as 398   The recipient CFU/mL was determined for both planktonic (A and B) and biofilm (C and 590 D) mating reactions with pAM714 (cyl+; open, red symbols) and pAM771 (cyl-; closed 591 green symbols) donors. Data shown are the average and standard deviation from a 592 minimum of three independent trials for each time point for both mating conditions. 593 Statistical significance was assessed using a two-tailed Student's t-Test; P-values, 594 *<0.05, **<0.01 and ***<0.001. cyl = cytolysin. with T11RF (closed squares) or T11RFΔcas9 (closed triangles) recipients, and a no-618