An increase in the incidence of gonorrhea was observed in Israel in 1998, accompanied by the appearance of quinolone resistant N. gonorrhoeae
(QRNG) isolates [5
]. The incidence of gonorrhea peaked in 2002 and declined sharply thereafter. The rate of isolation of QRNG strains also declined after 2001. During the epidemic period, QRNG strains were detected in several parts of Israel. Pulse field gel electrophoresis analysis of isolates from the Negev region in southern Israel, Jerusalem, Haifa, and Tel Aviv showed that all the QRNG strains were closely related [7
]. In the Tel Aviv area, there were 200 cases of gonorrhea in 2000 and 325 cases in 2001. The isolates for the present study were obtained from January 2000 through October 2001 in Tel Aviv, Israel [5
]. Of 80 isolates collected during this time period that were previously genotyped by molecular probes for the variable regions of the porB
], we selected 24 fluoroquinolone resistant and 24 sensitive strains for more detailed genotyping by sequencing of fragments of multiple housekeeping genes.
Phylogenetic relationships among the ciprofloxacin sensitive and resistant strains based on 9 housekeeping genes were estimated using the statistical parsimony procedure and graphically depicted as a network of gene genealogies (Figure ). The lines on the network indicate mutational connections among the unique genotypes with the number of substitutions separating these sequences in parentheses adjacent to the line. Genotypes were designated by strain number (CX), ciprofloxacin susceptibility (R = resistance, S = sensitive), contact information (P = paid commercial sex worker, F = non-paid sex worker, U = unknown), and month and year of isolation. There was no clustering of strains by contact information or date of isolation. Eighteen of 24 resistant strains had the identical sequence or genotype, represented in the genealogy by C7.R.U.0701. C2.R.U.0701 and C74.R.U.0800 differed by one mutational step from the most common genotype, and C10.R.U.0701 differed by one nucleotide substitution from C2.R.U.0701. Thirteen mutational steps separated these resistant strains from the most recent common ancestor for all the other strains in the data set. C21.R.U.0801 and C24.R.U.1001 had identical genotypes and were separated by 10 mutational steps from the next most closely related strains, which were sensitive isolates. The remaining resistant isolate, C3.R.U.0701, was distantly related to both the other resistant strains and was separated by 4 mutational steps from the most closely related sensitive strains. As a group, the sensitive strains were more genetically diverse than the resistant strains with 13 unique genotypes among 24 strains. However, 9 strains had the same genotype, which is represented in the network by C5.S.U.0801. For the resistant strains, we calculated time of divergence from the most recent common ancestor, and mean years and 95% highest posterior density (HPD) limits are shown in parentheses adjacent to the line connecting genotypes. C2R.U.0701 and C74.R.U.0800 diverged from the majority resistant genotype on average in early 1998 or 1997, respectively, and possibly as recently as mid-1999 or mid-1998, respectively. C10.R.U.0701 is also likely to have diverged from the other resistant strains sometime in 1998 or 1999.
Figure 1 Statistical parsimony network of the 9 housekeeping genes included in the MultiLocus Sequence Typing (MLST) scheme. Number of steps is indicated between parentheses. Mean divergence time estimate in years and 95% highest posterior density (HPD) limits (more ...)
The allelic profiles of the 48 strains produced 13 different STs. The eBURST program assigned these STs to four clonal complexes (Figure ). The majority resistance genotype (represented by C7.R.) was grouped with C2.R, C74.R and C10.R, supporting the recent evolutionary relationships between these strains revealed by the statistical parsimony procedure. C11.S and C27.S were single locus variants of each other, consistent with the close evolutionary relationship (one mutational step) found by the statistical parsimony procedure. Two pairs of double locus variants were identified; C21.R (representative of 2 resistant strains) and C59.S, and C37.S and C1.S (representative of 3 sensitive strains). In the statistical parsimony analysis these strains were separated by 10 and 9 mutational steps, respectively. Although C3.R and C73.S were separated by 4 mutational steps they did not form a clonal complex by eBURST analysis. All the remaining strains were separated by more than 10 mutational steps in the parsimony analysis.
Figure 2 eBURST diagram displaying the relatedness of 48 Israeli isolates. All the STs are displayed in a single diagram using settings for a population snapshot. ST names correspond to the strain number (CXX), followed by the letter R (resistant) or S (sensitive) (more ...)
The network of genotypes of the fluoroquinolone resistance genes (Figure ) showed a much lower level of genetic diversity over all than that of the housekeeping genes. Twenty-one of 24 resistant strains had the identical genotype, with only those three strains that were also distantly related at the housekeeping gene loci exhibiting different quinolone resistance genotypes. Among the quinolone sensitive strains, there were six genotypes, with 50% of strains having the identical genotype, which is represented in the network by C.5.S.U.0301.
Figure 3 Statistical parsimony network of two quinolone resistance genes, gyrA and parC Number of steps is indicated between parentheses. N = number of sequences. C1.S.U.0701 = C9.S.U.0701, C11.S.U.0701, C23.S.U.0801, C27.S.U.0901, and C37.S.P.0600. C5.S.U.0301 (more ...)
The phylogeny for porB sequences was estimated by the Bayesian method and evolutionary relationships among genotypes are displayed as a 50% majority-rule consensus tree (Figure ). There was no clustering of strains by contact information or date of isolation. Thirteen resistant strains had the identical porB sequence. Six strains differed from the majority clade by a 3–9 base pair deletion. C2.R.U.0701 differed by a 3 bp deletion and two nucleotide substitutions. The majority clade was estimated to have diverged from the most recent common ancestor approximately 23 years ago. C2.R.U.0701 formed a branch sister to the majority clade with a mean divergence time of 16 years. C3.R.U.0701 was distantly related to the other porB sequences in the data set. Three resistant strains, C28.R.P.1001, C35.R.P.0600 and C45.R.F.0900, all from men who reported contact with sex workers, had the same porB sequence as that of five sensitive strains. The five sensitive strains shared a common housekeeping and fluoroquinolone resistance genotype, which was also the most prevalent genotype among the sensitive strains (represented by C5.S.U.0801 on the housekeeping and quinolone resistance gene genealogies). The three resistant strains were identical at all 9 housekeeping and 2 fluoroquinolone resistance gene loci to the most common resistant genotype and shared no alleles at these loci in common with the five sensitive strains.
Bayesian 50% majority-rule consensus tree of porB sequences. Mean divergence time in years and 95% highest posterior density (HPD) limits are indicated in brackets. Clade support posterior probabilities (if ≥ 50%) are shown over the branches.
To determine whether positive selection has been a force in the evolution of N. gonorrhoeae
, positively selected sites were identified by estimating the per site nonsynonymous/synonymous rate ratio and by evaluating changes in amino acid properties (Table ). With the exception of glnA
, there were very few positively selected sites (1–3) in the housekeeping genes and generally the same amino acid was found with high frequency among the resistant and sensitive strains. At five sites in glnA
the amino acid found most often in the resistant strains was a low frequency amino acid in the sensitive strains. At amino acid position 284 in pilA
all the sensitive strains had a threonine and 23 of 24 resistant strains had an alanine. Two positively selected sites were identified in gyrA
(amino acid position 91 and 95) and one site in parC
(position 86). The resistant strains had mutations, Ser91Phe and Asp95Asn in gyrA
and Asp86Asn in parC
, that are consistent with quinolone resistance determining mutations identified in other studies. Seventeen positively selected sites were identified in porB
and many sites were polymorphic. A previous study showed that a single amino acid mutation to Lys at residue 120 of the Por IB protein confers full intermediate level resistance to penicillin and tetracycline and a single Asp mutation at either position 120 or 121 (22 and 23 in our shorter sequences) confers partial resistance [19
]. By our analysis, both amino acid residues are under strong positive selection. A single Lys was found at position 120 in all 24 resistant strains and 15 sensitive strains, and 2 sensitive strains had an Asp at either position 120 or 121 (Table ).
Table 1 Positively selected amino acid sites (AA site). Sites based on dN/dS > 1 under PAML and change in amino acid property by TREESAAP (TS) and absolute amino acid frequencies at those sites in susceptible (AA-S), resistant (AA-R), C3 strain (AA-C3R), (more ...)
Amino acid mutations at residues 120 and/or 121 that confer resistance to penicillin and tetracycline in N. gonorrhoeae.
The historical demography of N. gonorrhoeae
in Israel over the past 24 years was estimated from the housekeeping gene and porB
sequences using a Bayesian MCMC method that allows the inference of past population dynamics from contemporary sequences (Figure ). After decreasing slowly from 1976 to approximately 1993, the effective strain population size declined sharply from 1993 to 1999. However, since 1999, it has been flat to slightly increasing. The census population of gonococcal infections in Israel, represented by the clinical case-incidence rate per 100,000 persons, declined from 40 in 1970 to 4 in 1987 and to 0.9 in the mid 1990's [20
]. In 1999, the rate rose to 3.9 and further increased to 8.3 in 2000. The genetic data support this demographic information on the declining infection rate since the 1970s and the recent rise of infection rates since the late1990's
Figure 5 N. gonorrhoeae population dynamics in Israel over the past quarter century. Relative genetic population size (Neτ, where Ne is effective population size and τ is number of generations) was estimated from 9 housekeeping genes (red lines) (more ...)