Evolution, persistence, and host adaption of a gonococcal AMR plasmid that emerged in the pre-antibiotic era

Abstract

Plasmids are diverse extrachromosomal elements significantly that contribute to interspecies dissemination of antimicrobial resistance (AMR) genes. However, within clinically important bacteria, plasmids can exhibit unexpected narrow host ranges, a phenomenon that has scarcely been examined. Here we show that pConj is largely restricted to the human-specific pathogen, Neisseria gonorrhoeae. pConj can confer tetracycline resistance and is central to the dissemination of other AMR plasmids. We tracked pConj evolution from the pre-antibiotic era 80 years ago to the modern day and demonstrate that, aside from limited gene acquisition and loss events, pConj is remarkably conserved. Notably, pConj has remained prevalent in gonococcal populations despite cessation of tetracycline use, thereby demonstrating pConj adaptation to its host. Equally, pConj imposes no measurable fitness costs and is stably inherited by the gonococcus. Its maintenance depends on the co-operative activity of plasmid-encoded Toxin:Antitoxin (TA) and partitioning systems rather than host factors. An orphan VapD toxin encoded on pConj forms a split TA with antitoxins expressed from an ancestral co-resident plasmid or a horizontally-acquired chromosomal island, potentially explaining pConj's limited distribution. Finally, ciprofloxacin can induce loss of this highly stable plasmid, reflecting epidemiological evidence of transient reduction in pConj prevalence when fluoroquinolones were introduced to treat gonorrhoea.

Fig 1. pConj is conserved and largely restricted to the gonococcus.

(A) Map of pConj showing four regions involved in replication/partitioning (including parAB), mating bridge formation (trb), conjugation (tra) and the genetic load region which includes three toxin-antitoxin (TA) related loci. (B) Appearance of plasmids in N. gonorrhoeae based on WGS at PubMLST and previous reports; pConj first occurred in strains with pCryp. Years during which penicillin and tetracycline monotherapy were recommended in the USA are as shown. (C) pConj is present in four species of Neisseria; only species with ≥40 WGS are included. pConj was not detected in N. bergeri, N. cinerea and N. subflava. Only N. gonorrhoeae carried tetM⁺ pConj. (D) There are two phylogenetic groups of pConj, depending on the presence of ε:ζ2 or ε:ζ3 TA systems. Sequences were aligned independently of tetM and the surrounding transposon. Each dot represents an isolate, colour-coded according to year of isolation. (E) Despite pConj’s conservation, the plasmid is found in multiple lineages across the gonococcal population. Each dot represents an isolate, colour-coded according to pConj carriage.

Fig 2. pConj has persisted in the absence of tetracycline.

Prevalence of pConj in WGS of isolates from (A) USA and (B) UK. Doxycycline was recommended as part of a dual therapy for treatment before 2014 in USA; pConj carriage (including markerless plasmids) remained constant during/after doxycycline treatment (n = 2,517 isolates), as determined by unpaired t-test. pConj has remained prevalent in the UK after tetracycline treatment (2007–2016, n = 3,240 isolates). ns, p > 0.05.

Fig 3. TIS does not identify chromosomal genes involved in pConj maintenance.

Transposon insertion sites of significant hits at the 16th generation (q < 0.01, Bio-Tradis; log fold change < 0) visualised in Artemis; gene orientation is shown. Graphs show the distribution, numbers and orientation of transposon insertion mutant library obtained under both control and tetracycline conditions (two independent biological repeats). Each vertical line indicates a UIS, with the height reflecting the number of mutants at each site. Red and blue lines indicate transposon insertions in the forward and reverse direction, respectively. Blue lines are plotted in front of red, masking some red insertions. The maximum number of insertions displayed in each panel is marked individually.

Fig 4. TA systems cooperate to maintain pConj.

(A) Schematic of the genetic load region. gfp:kan was used to monitor the presence of pConj. Genes and related functions are annotated based on PubMLST. (B) Replacement of the genetic load region with gfp:kan (pCΔGL) resulted in detectable plasmid loss. (C) pConj lacking individual TA systems (pCtΔvapD, pCt:ζ1^K115A, pCtΔζ2) were stably maintained, whereas loss of combinations of TA systems (pCtΔvapD:ζ1 and pCtΔTA) resulted in plasmid loss within 40 generations (LOD = 0.3%). All assays consist of three independent repeats and results were analysed with one-way ANOVA with Sidak’s multiple comparisons, shown as mean ± SD; * p ≤ 0.05).

Fig 5. Interaction between mobile genetic elements is associated with pConj carriage.

(A) VapD^pConj is homologous to other VapDs. Shading intensity is proportionate to percentage conservation. AlphaFold predicted structure of VapD^pConj (red) has similar architecture to VapD from H. pylori (PDB 3UI3, dark blue) and H. influenzae (PDB 6ZN8, light blue). (B) VapD^pConj is toxic in E. coli, and is neutralised by VapX^pCryp. Viability of E. coli (CFU/ml) after induction with 0.2% of L-arabinose is shown. (C) pConj carriage in N. gonorrhoeae with vapX (n = 9,316) compared to isolates without vapX (n = 302) (D) VapD^pConj toxicity in E. coli is neutralised by VapX^Nm. (E) pConj carriage in N. meningitidis isolates with vapX (n = 194) compared to isolates without vapX (n = 16,718). Toxin-antitoxin assays were carried out in three independent repeats, analysed with two-way ANOVA with Tukey’s multiple comparisons and shown as mean ± SD. In B and D the number of CFU at 360 min post-induction are shown. pConj carriage was analysed with Fisher’s exact test. ** p ≤ 0.01, **** p ≤ 0.0001.

Fig 6. pConj is stably maintained by multiple mechanisms.

(A) pConj is rapidly lost from strains lacking TA and/or partitioning systems (i.e. within 6–9 generations). Assays were carried out three times and analysed with Mann-Whitney test, LOD = 0.3%, shown as mean ± SD. ** p ≤ 0.01. Treatment of (B) FA1090 or (C) 60755 with ciprofloxacin promotes pConj loss. Bacteria were grown overnight on GCB before incubation in 0.5 MIC of ciprofloxacin or spectinomycin. Colonies (n ≥ 200; LOD = 0.5%) were tested for pConj carriage. Assays were carried out three times and analysed with student’s unpaired t-test, shown as mean ± SD. * p ≤ 0.05, *** p ≤ 0.001. (D) As pConj enters N. gonorrhoeae, interactions between pCryp and pConj via VapXD contribute to the stabilisation of pConj in the gonococcus as well as its maintenance. pConj encodes TA and partitioning systems–therefore, pConj has developed multiple fail-safe mechanisms to ensure it is maintained in N. gonorrhoeae. While pConj is stably maintained in N. gonorrhoeae, antibiotic use facilitated spread of both itself and pbla, resulting in the spread of plasmid-mediated antimicrobial resistance. Currently, despite the lack of tetracycline use, pConj persists in N. gonorrhoeae; as such, treatment regimens no longer suggest the use of penicillin or tetracycline.