Octapeptin C4 Induces Less Resistance and Novel Mutations in an Epidemic Carbapenemase-producing Klebsiella pneumoniae ST258 Clinical Isolate Compared to Polymyxins

Polymyxin B and E (colistin) have been pivotal in the treatment of extensively drug-resistant (XDR) Gram-negative bacterial infections, with increasing use over the past decade. Unfortunately, resistance to these antibiotics is rapidly emerging. The structurally-related octapeptin C4 (OctC4) has shown significant potency against XDR bacteria, including against polymyxin-resistant (Pmx-R) strains, but its mode of action remains undefined. We sought to compare and contrast the acquisition of XDR Klebsiella pneumoniae (ST258) resistance in vitro with all three lipopeptides to help elucidate the mode of action of the drugs and potential mechanisms of resistance evolution. Strikingly, 20 days of exposure to the polymyxins resulted in a dramatic (1000-fold) increase in the minimum inhibitory concentration (MIC) for the polymyxins, reflecting the evolution of resistance seen in clinical isolates, whereas for OctC4 only a 4-fold increase was witnessed. There was no cross-resistance observed between the polymyxin - and octapeptin-induced resistant strains. Sequencing revealed previously known gene alterations for polymyxin resistance, including crrB, mgrB, pmrB, phoPQ and yciM, and novel mutations in qseC. In contrast, mutations in mlaDF and pqiB, 1genes related to phospholipid transport, were found in octapeptin-resistant isolates. Mutation effects were validated via complementation assays. These genetic variations were reflected in phenotypic changes to lipid A. Pmx-R isolates increased 4-amino-4-deoxy-arabinose fortification to phosphate groups of lipid A, whereas OctC4 induced strains harbored a higher abundance of hydroxymyristate and palmitoylate. The results reveal a differing mode of action compared to polymyxins which provides hope for future therapeutics to combat the increasingly threat of XDR bacteria.


INTRODUCTION
corresponding to m/z 973.2 was never observed. The lipid A from the OctC4 induced isolates was 145 substantially different from the Pmx-R isolates and similar to the WT profile, with a major peak 146 of the hydroxymyristate derivative and a significant 5-fold increase in representation of 147 palmitoylation (Fig. 3B, see Fig. S4 in the supplemental material). The Ara4N modification was 148 enhanced compared to WT, but not to the extent seen with Pmx-R isolates. 149 Plasmid loss associated with OctC4 resistance. To ascertain the genetic basis for resistance and 150 subsequent phenotypic traits, four day 20 replicates were selected from each treatment group.

151
Clonal expansion of genomic variations were monitored by selecting four colonies per replicate. 152 Additionally, two colonies from the initial isolate were sequenced. The initial isolate harbored 153 multiple acquired resistance genes targeting aminoglycosides, β-lactams, fosfomycin, quinolones, 154 sulfonamides, tetracycline and trimethoprim, consistent with the parent XDR profile (Table 2). 155 Five plasmid replicons were identified including ColRNAI, IncFIB(K)-Kpn3, IncFII(K), IncN and 156 IncX3. In cross-resistance studies, the only Pmx-R replicate with an alteration in MIC profile to resistance whilst phospholipid transport associated with OctC4 resistance. Genomic 169 alterations identified in polymyxin and OctC4 treated replicates differed significantly. In  replicates, genes predominantly associated with LPS processing and lipid A modifications were 171 altered, including crrB, hepIII, lptC, mgrB, pmrB, phoPQ and yciM (Table 3). An additional TCS  The OctC4 replicates harbored changes in mlaDF, pqiB and traH in all four colonies. Additional 187 genes altered that were apparent in two colonies per replicate included azoR, hinT and rpsA.

188
Strikingly, mlaF (A165P) was impacted in three different OctC4 replicates at the same position 10 induced replicates reduced the MIC by 2-fold, however, consistently only partial growth was 191 observed at 8 µg/ml. This finding validates the partial contribution of these genes to resistance 192 ( Fig. 4I-L). Introduction of WT genes into the initial isolate revealed that the vector and gene did 193 not influence the MIC and confirmed that these alterations are responsible for the resistance 194 observed ( Fig. 4M and N).

195
Polymyxin unfortunately now induce high levels of resistance during therapeutic use, which is 197 further compromised by suboptimal exposure in the clinic due to the risk of nephrotoxicity (29).

198
Resistance in K. pneumoniae appears to be stable and incurs a minimal fitness cost (30, 31). These

205
Following 20 days of increasing sub-lethal antibiotic exposure, no cross-resistance was apparent 206 between polymyxins and OctC4. CST and PMB resulted in similar profiles with the only deviation 207 seen in sample PMB_2 in which susceptibility to meropenem was regained. This is due to the 208 absence of blaKPC-2 and blaOXA-9. Additionally, the homogenous loss of the IncX3 plasmid was 209 identified. Clinically, meropenem is being used in combination with polymyxins, and these results 210 suggest that, in some cases, meropenem may overcome polymyxin resistance (32, 33).

211
Furthermore, previous research has identified the loss of blaKPC plasmids in Pmx-R clinical 212 isolates and suggests that this loss is due to a potential fitness cost (34). Our results show various 213 accounts of plasmid loss in OctC4-exposed replicates, and this corresponded to a reduction in 214 resistance towards cefepime, meropenem and tetracycline. Whether this resembles a fitness cost 215 associated with OctC4 exposure or due to repeated passaging under no selective pressure for the 216 genes harbored on these plasmids warrants further investigation.

217
Interestingly, resistance towards chloramphenicol was diminished in polymyxin and OctC4 218 exposed strains. Resistance towards chloramphenicol can arise from plasmid-encoded 219 chloramphenicol acetyltransferases, alterations in the target 50S ribosomal subunit, or disruptions 220 in porins and efflux pumps (35). The absence of acquired resistance genes and the lack of 221 modifications in these regions may imply either a down-regulation of efflux pumps or an 222 alternative resistance mechanism. The synergistic mechanism of polymyxins and chloramphenicol 223 have been extensively studied; however, this finding potentially indicates a novel loss of 224 chloramphenicol resistance upon gaining resistance towards these lipopeptides (36,37), and is also 225 seen with the octapeptins.

226
The mutations observed in polymyxin resistance induced ST258 strains can be compared to those 227 we have previously identified in closely related polymyxin-resistant clinical ST258 isolates, 228 2_GR_12, 4_GR_12, 10_GR_13, 13_GR_14 and 14_GR_14 (24). As in this study, the vast 229 majority of resistance was attributed to mgrB (60%), albeit not via an IS element disruption 230 commonly observed in the clinic. Additional mutations were also identified in phoPQ 231 accompanying the mgrB disruption, which was also apparent in this study (CST_3, CST_4, 232 PMB_2). Other mutations in crrB, mgrB, pmrB, phoPQ and yciM in acquisition of polymyxin 233 resistance have previously been described in resistant strains (16,17,38). Taken together, this isolates, mutations were present in multiple genes within a replicate. CST_3 harbored a deleterious 241 mutation in mgrB (M1I) and additional alterations in phoQ (P420A, G434C) increased tolerance 242 to CST. This was also the circumstance for CST_4 (mgrB: N42I, phoQ: D417N). PMB_2 243 possessed a resistance conferring mutation in mgrB (W47L) and phoP (R81C). The notion that 244 one alteration in TCS drives resistance, the circumstance for the majority of clinical isolates is well 245 accepted (39). However, our findings contradict this concept. 246 We also identified alterations in another TCS, QseBC, which is known to facilitate cross-talk with 247 PmrAB in Escherichia coli (40). In E. coli, PmrB acts as a noncognate partner to the QseBC TCS 248 and has the capability to not only phosphorylate PmrA, but also QseB. The absence of QseC was 249 shown to impact virulence due to the accumulation of phosphorylated QseB and in particular, 250 alterations in the histidine kinase domain attenuates its ability to de-phosphorylate QseB (40, 41).

251
Furthermore, the deletion of qseC and pmrA, promoting phosphorylation of QseB by PmrB, 252 stimulated tolerance to PMB (42). This signalling pathway remains severely under characterized 253 in K. pneumoniae. We observed partial tolerance to PMB when a frameshift mutation was apparent 254 at nucleotide 118; however, full resistance in PMB_4 was promoted by alterations in mgrB (D29Y) 255 and yciM (V43G), which has recently been identified to cause resistance (38). Conversely, PMB_3 256 also harbored a frameshift mutation early in the coding sequence of qseC 257 (GCCTGAGCCTGC17Δ fs ), although an additional I283L change in the histidine kinase region 258 resulted in an MIC of 4 µg/ml. This did not explain the full resistance profile exhibited by PMB_3 259 and due to the presence of both alleles during complementation, the true extent of resistance cannot 260 be deduced. Considering PMB_3 still resulted in the addition of Ara4N to lipid A, we speculate 261 that due to the perturbation in the QseC kinase, this is increasing the accumulation of  Table S1 in the supplemental material). Clinical breakpoints were determined in 300 concordance to CLSI guidelines (51) and tigecycline as per The European Committee on Lipid A modifications. Lipid A was extracted using the ammonium hydroxide-isobutyric acid 317 protocol as previously described (52). Day 20 cultures were grown overnight in LB supplemented 318 with antibiotic (see Table S1 in the supplemental material). Overnight inoculums were subcultured

718
b Nucleotide variations present in ≥90 % of reads and ≥50 X coverage compared to initial strain,

719
20_GR_12. Δ symbolises a deletion, -in front of the nucleotide position indicates an alteration 720 upstream and fs represents a frameshift mutation.

721
c The introduction of a truncation in the protein downstream of the alteration is noted as tr .