The Gonococcal Genetic Island defines distinct sub-populations of Neisseria gonorrhoeae

Abstract

The incidence of gonorrhoea is increasing at an alarming pace, and therapeutic options continue to narrow as a result of worsening drug resistance. Neisseria gonorrhoeae is naturally competent, allowing the organism to adapt rapidly to selection pressures including antibiotics. A sub-population of N. gonorrhoeae carries the Gonococcal Genetic Island (GGI), which encodes a type IV secretion system (T4SS) that secretes chromosomal DNA. Previous research has shown that the GGI increases transformation efficiency in vitro, but the extent to which it contributes to horizontal gene transfer (HGT) during infection is unknown. Here we analysed genomic data from clinical isolates of N. gonorrhoeae to better characterize GGI+ and GGI- sub-populations and to delineate patterns of variation at the locus itself. We found the element segregating at an intermediate frequency (61%), and it appears to act as a mobile genetic element with examples of gain, loss, exchange and intra-locus recombination within our sample. We further found evidence suggesting that GGI+ and GGI- sub-populations preferentially inhabit distinct niches with different opportunities for HGT. Previously, GGI+ isolates were reported to be associated with more severe clinical infections, and our results suggest this could be related to metal-ion trafficking and biofilm formation. The co-segregation of GGI+ and GGI- isolates despite mobility of the element suggests that both niches inhabited by N. gonorrhoeae remain important to its overall persistence as has been demonstrated previously for cervical- and urethral-adapted sub-populations. These data emphasize the complex population structure of N. gonorrhoeae and its capacity to adapt to diverse niches.

Keywords: Gonococcal Genetic Island (GGI); Neisseria gonorrhoeae; pangenome; population genetics.

Fig. 1.

The core genomes of N. gonorrhoeae GGI+ and GGI− isolates are genetically differentiated. Core genome phylogeny inferred using RAxML; tips are coloured by presence or absence of the GGI. In total, there were 201 isolates (including reference strain NCCP11945) with 123 being GGI+ and 78 being GGI−. Tree is midpoint rooted and scale bar represents nucleotide substitutions per site.

Fig. 2.

The pangenome of GGI+ isolates is larger than GGI− isolates while accessory gene content is shared between groups. (a) Rarefaction and accumulation curves of core and total gene content in GGI+ and GGI− isolates show a larger pangenome among GGI+ isolates. All groups contain an equal number of isolates: GGI+ isolates were randomly sampled without replacement 100 times to the size of the GGI− dataset (n=78). (b) Heatmap comparing accessory gene frequencies in GGI+ and GGI− groups. Accessory genes are maintained at similar frequencies between the two populations (except for the GGI, shown in purple in the bottom right of the plot), as indicated by the density of genes along the diagonal (equal frequencies in both populations). This suggests that there are no barriers to accessory gene flow between groups.

Fig. 3.

Core genome networks indicating differences in patterns of recombination between GGI+ and GGI− isolates. Core genome networks from GGI+ (a) and GGI− (b) isolates made using SplitsTree. The PHI test for recombination was significant (P<0.05) for both groups, indicating evidence of recombination. Network scale bars represent substitutions per site.

Fig. 4.

Isolates have similar frequencies of recombination within each group, but GGI+ isolates appear more differentiated from one another. Recombination tracts predicted by Gubbins in the separate core genomes of GGI+ (a) and GGI− (b) isolates, plotted with a recombination-adjusted tree using Phandango. In total, 59 % of the GGI− core alignment and 68 % of the GGI+ core alignment was affected by recombination. Recombinant fragments shared between isolates within the dataset are shown in red, while those which are determined to have originated outside of the dataset are in blue. Branch lengths in the core genome phylogeny of GGI+ isolates are significantly longer than GGI− isolates (Mann–Whitney U test, P=7.9e-04) (c). These data indicate different evolutionary histories between the two groups, and greater core genome differentiation among GGI+ isolates.

Fig. 5.

Multiple variants in mntH associated with GGI presence. Weir and Cockerham's F _ST (wcFst) values of core genome variants are plotted against the core genome position. The significance threshold was estimated by taking the maximum wcFst value from 100 random permutations of phenotypes. Variants with non-significant wcFst values are shown in grey, and those with significant values in black. Additionally, variants that are homoplasy outliers (see Methods) are shown in red. Twelve variants all within mntH have markedly high wcFst values, indicating a strong association with the presence of the GGI.

Fig. 6.

Mobile genetic elements (MGEs) accumulate in GGI− isolates while prophages are exchanged freely. (a) Multi-dimensional scaling (MDS) of pairwise mash distances of prophage sequences as annotated by ProphET. Each point represents a single annotated prophage from a given isolate, coloured by GGI presence in that isolate. Prophage clusters are not separated by GGI presence/absence, indicating that prophages are exchanged between groups. (b) MGEs (specifically insertion sequences; ISs) were annotated using MobileElementFinder and the total number of ISs per strain was determined and plotted by GGI presence/absence. A Mann–Whitney U test shows significantly higher numbers of MGEs per strain in GGI− isolates (P=6.6e-04).

Fig. 7.

The GGI is largely conserved with minimal variation in gene content and sequence. (a) Locus map showing the pangenome of the GGI is skewed towards core genes, with only a few accessory GGI genes. Dark blue genes are core GGI genes (≥99 % frequency), light blue are soft core (95–99 % frequency), grey are shell (15–95 % frequency) and white are cloud (<15 % frequency) accessory GGI genes. Dotted line indicates the continuation of the operon on the next line. (b) Core GGI network revealing evidence of intra-locus recombination. Network of core GGI genes (53 in total) from GGI+ isolates made in SplitsTree. The PHI test for recombination was significant (P<0.05) indicating evidence of recombination. (c) Pairwise $\pi$ N/ $\pi$ S values for GGI genes indicating varying selective pressures between genes. Nucleotide diversity values at non-synonymous and synonymous sites calculated with Egglib for each pairwise comparison of the same GGI gene between two GGI+ isolates. Values are plotted as boxplots with outliers shown as points.

Fig. 8.

GGI is gained and lost, and switched between four different subtypes. (a) PCA of core genome alignment of 53 genes present in ≥99 % of GGI+ isolates. Analysis was performed using the adegenet package in R showing that the GGI clusters into four distinct groups (named I–IV). (b) Maximum-likelihood phylogeny inferred using RAxML of core GGI gene alignment, coloured by GGI group. (c) Ancestral state of the GGI locus was inferred using the R package Ape, plotted as a phylogenetic tree inferred using RAxML. Ancestral state probabilities are labelled at each node using a pie chart, coloured according to GGI group. The state of the GGI changes a total of 14 times on this phylogeny which are marked by red letters: G (gained) five times, L (lost) three times and S (switched) GGI group six times.