Description of new mutations in the rpoB gene in rifampicin-resistant Neisseria meningitidis selected in vitro in a stepwise manner

Many reports have described single point mutations leading to rifampicin resistance in a variety of micro-organisms, for instance, in Escherichia coli (Jin & Gross, 1988), Mycobacterium tuberculosis (Telenti et al., 1993) and Staphylococcus aureus (Aubry-Damon et al., 1998). The mutations described provided the first evidence that mutations within a defined subgenic segment of the ß-subunit of DNA-dependent RNA polymerase (encoded by rpoB) may be the mechanism that causes rifampicin resistance. A couple of different point mutations in a subgenic rpoB fragment have also been described for Neisseria meningitidis (Carter et al., 1994; Nolte, 1997; Stefanelli et al., 2001). The hitherto-described point mutations were considered to be responsible for the single-step development of resistance. Additional mechanisms such as mutations in membrane efflux pumps, which were originally described for Neisseria gonorrhoeae resistant to hydrophobic agents (Hagman et al., 1995; Pan & Spratt, 1994), have been discussed as a reason for high-level resistance (Abadi et al., 1996). However, nothing is currently known about the possible stepwise acquisition of rifampicin resistance, neither whether this can indeed happen nor which molecular event might be the mechanism. The current work was therefore intended to select primary susceptible meningococci towards rifampicin resistance in vitro in a stepwise manner. The basic idea behind our approach was that, if we selected single colonies that appeared to be less susceptible to rifampicin than the majority of the respective plate culture on an E-test, we should be able to force the meningococci to stepwise higher MICs. Subsequent sequencing of the subgenic rpoB fragment should lead to the identification of point mutations that confer resistance by means of a cumulative mechanism. Fourteen N. meningitidis strains (Table 1), which were originally isolated from routine diagnostic material sent to the German Reference Centre for Meningococci (at the time that this study was initiated, the Centre was located at Hygiene-Institute, University of Heidelberg; now located at the Hygiene-Institute, University of Würzburg), were used. In order to select stepwise for rifampicin resistance, E-tests were performed using rifampicin E-tests strips (AB Biodisk; Solna) according to the recommendations of the manufacturer. Following 18 h incubation, those colonies growing slightly into the inhibition ellipse were subcultured and used for a subsequent E-test. This procedure was repeated until a strain was found to be resistant. The criterion for resistance was an MIC 4 µg ml^-1, as recommended in the NCCLS guidelines for organisms other than Haemophilus ssp., N. gonorrhoeae and streptococci.

Table 1. Meningococcal strain pairs used in this work Amino acid positions refer to the position in the complete RpoB protein. The phenotype gives the serological formula of the original strains.

The DNA of each strain was subjected to PCR in order to amplify a 790 bp subgenic rpoB fragment (fragment B in Fig. 1) as described previously (Nolte, 1997). Amplification products were purified and analysed by cycle sequencing. A SequiTherm Excel II Long-Read DNA sequencing kit ALF (Epicentre Technologies) and the sequencing primers NmB9F and NmB24R (sequences given in Table 2; positions on rpoB are shown in Fig. 1) were used for automated sequencing on an ALFexpress sequencer (Amersham Pharmacia Biotech).

Table 2). The flanking genes rpoC and rplL are not given to scale. Enlarged is the 1352 bp amplicon (fragment A), derived by using primers U3 and U5, shown as an open bar for the original rifampicin-susceptible recipient strain and as a hatched bar for the resistant donor strain. Restriction sites for HpyCH4III are shown as open circles; a fifth site, characterizing the donor strain, is shown as a filled circle. The position of the meningococcal uptake sequence is indicated by a box. The shaded bar shows the 790 bp fragment (B) amplified using RPORR1 and RPORR5, spanning the cluster of rifampicin resistance-conferring mutations (rifR cluster; C). Previously known mutations are indicated by black triangles (data combined from this paper and from Carter et al., 1994; Nolte, 1997; Stefanelli et al., 2001); the mutation leading to a substitution at Ala558 is not shown as this mutation does not confer resistance. Position His552 is a hot spot; three different mutations have been described to date.

Table 2. Primers used in this study All primers except NmB9F and NmB24R, which were sequencing primers only, and U5, which was used for amplification only, were used for amplification and sequencing. Primer sequences were obtained from the sequence of the entire rpoB gene of N. meningitidis (EMBL/GenBank/DDBJ accession no. Z54353; O. Nolte, unpublished). Tm values were supplied by the manufacturer of the oligonucleotides (TIBMolBiol); for amplification, the annealing temperature was set to 2 °C below the given Tm. NA, Not applicable.

For the transformation experiments, an amplicon of 1352 bp (fragment A in Fig. 1) of rpoB, covering the Ser⁶⁰⁰ mutation, was amplified (primers U3/U5; Table 1, Fig. 1) from strain 93/96 (MIC 24 µg ml^-1). The amplified sequence harboured the meningococcal uptake sequence (5'-GCCGTCTGAA; Goodman & Scocca, 1988, 1991) close to its 5' end.

Strain LB2927, isolated from a case of invasive meningococcal disease, was used as a recipient strain (MIC for rifampicin: 0.047 µg ml^-1). Transformation of LB2927 with PCR amplicons of type A was done as described elsewhere (van der Ley & Poolman, 1992) with modifications. Briefly, starting from a pure culture, meningococci were grown in 3 ml standard cell culture medium RPMI 1640 (Gibco-BRL) supplemented with 10 % fetal calf serum to mid-exponential phase. The meningococci were spun down and resuspended in 100 µl RPMI 1640. DNA was added to the bacterial suspension to a final concentration of 3 µg ml^-1 with and without MgCl₂ at a final concentration of 1 mM, and the mixture was then subjected to a 42 °C heat-shock in a water bath for exactly 45 s. The cultures were incubated overnight in order to enable recombination and to ensure phenotypic expression of the integrated DNA. Transformed meningococcal cultures were streaked onto freshly prepared GC agar (chocolate agar supplemented with IsoVitalex) containing either 32 or 256 µg rifampicin ml^-1 (GC agar produced in-house, rifampicin from Sigma) as well as on GC agar without antibiotic (growth control).

All colonies that were obtained after transformation on rifampicin-agar plates were used in a simple assay to discriminate between transformed meningococci and bacteria that had become resistant through spontaneous point mutations. Briefly, the PCR amplicon used for transformation displayed five sites specific for the restriction enzyme HpyCH4III and the corresponding sequence of the acceptor strain harboured only four sites (Fig. 1). Subgenic fragment A (1352 bp) covering these sites was amplified by adding a toothpick of bacterial growth to PCR mixtures. Following amplification, the DNA was digested using 2 U HpyCH4III (New England Biolabs). The banding patterns were checked on 2 % agarose gels. A pattern of five bands (of 577, 398, 206, 137 and 34 bp) indicated that a grown colony was resistant due to spontaneous mutation; bands of 547, 398, 206, 137, 34 and 30 bp indicated that a colony represented a colony grown from transformed meningococci. Transformants and 12 randomly selected spontaneously mutated isolates were analysed further by sequencing.

Fourteen meningococcal strains were selected in a stepwise manner towards rifampicin resistance. The majority of the resulting resistant strains displayed final MICs of >256 µg ml^-1 (mean number of 5.43 ± 0.51 subcultures in order to achieve resistance). Sequence analysis of the 790 bp subgenic rpoB fragment (fragment B) derived from each of the strain pairs (i.e. the original susceptible strain and the corresponding fully resistant one) yielded only a single nucleotide difference within the region sequenced. The only exception was the strain pair B4241 and 71/96, which differed by two nucleotides. Analysis of the deduced amino acid sequences showed that each of these point mutations lead to amino acid substitutions, which are given in Table 1. The most frequent amino acid substitution observed in our strain collection was the change at His⁵⁵², which has been reported previously (Carter et al., 1994; Nolte, 1997; Stefanelli et al., 2001). In addition, four hitherto-undescribed amino acid substitutions were found in six strains. Three strains (MIC 8, 16 and >256 µg ml^-1) displayed a substitution of Ser⁶⁰⁰ by Leu, one strain displayed a change from Ser⁵³⁴ to Pro (MIC 8 µg ml^-1) and another strain a change at position 526, where Ser was changed to Ala (MIC 24 µg ml^-1). Finally, we observed a mutation leading to a change at position Ala⁵⁵⁸ to Val, which occurred as a second substitution besides the well-known His⁵⁵² in strain 71/96. All mutations, including those that have been published previously, are shown in Fig. 1.

Transformation
In order to demonstrate that mutations observed after stepwise selection are responsible for acquired resistance, PCR-amplified subgenic 1352 bp fragments (fragment A in Fig. 1) harbouring the mutation conferring the amino acid substitution Ser⁶⁰⁰ were used for transformation experiments. The DNA sequence of the amplicon of the donor strain 93/96 (MIC 24 µg ml^-1) was identical with the sequence of another strain (299/95) found to display an MIC of about 8 µg ml^-1 (Table 1). The deduced amino acid sequence of the DNA fragment used for transformation was identical in strain 299/95 (MIC 8 µg ml^-1) and strain 348/95 (MIC 256 µg ml^-1).

The 1352 bp fragment of the donor strain harboured five sites for the restriction enzyme HpyCH4III, whereas the corresponding fragment of the susceptible recipient harboured only four sites (Fig. 1), resulting in different RFLP patterns following digestion of the amplified fragment A. In addition, the subgenic fragment A of the recipient strain LB2927 differed from the respective fragment of the resistant donor strain by a number of silent mutations (genetic marker), which are described in detail by Nolte (1997). In brief, the donor strain belonged to sequence type 2 in that paper, characterized by three consecutive triplets each mutated in the third position (C→A at position 1707, C→A at position 1710 and C→T at position 1713 of the gene) not present in the recipient strain, which belonged to sequence type 1.

Following four independent transformations, 198 resistant colonies were observed (80 on agar plates containing 32 µg rifampicin ml^-1 and 118 on plates containing 256 µg ml^-1). Colonies resistant through spontaneous mutations were observed exclusively after transformation without the addition of MgCl₂. Only three resistant colonies could be identified by their HpyCH4III restriction fragment pattern at rpoB as transformants, all of which grew after transformation in the presence of 1 mM MgCl₂. Two of the three transformants were undoubtedly confirmed following sequence analysis of the 1352 bp fragment: (i) the Ser⁶⁰⁰-conferring mutation was present and (ii) the genetic marker, absent in the original recipient strain, was present in both recipient strains after transformation. The third transformant, however, harboured the genetic marker sequence of the donor strain but did not display the Ser⁶⁰⁰ mutation. Rather, the mutation in codon 552 (nucleotide position 1654) was found, leading to a substitution of His by Tyr in the deduced amino acid sequence.

Of the 195 colonies found to be resistant due to spontaneous mutations, 12 were selected randomly for sequence determination of the 1352 bp subgenic fragment A. All of the sequences were identical and did not contain the genetic marker. All 12 strains, however, were found to have acquired resistance due the mutation C→T at position 1654, leading to the resistance-conferring mutation at amino acid position His⁵⁵² of the protein.

The overall mutation frequency leading to spontaneous mutants was calculated to be 2.59x10^-7 and the transformation efficiency was determined to be 1.93x10^-9.

The transformed meningococci, as well as the spontaneously mutated strains, grew with comparable efficacy on plates containing both 32 and 256 µg rifampicin ml^-1.

The results described here indicate that the molecular mechanism of acquisition of rifampicin resistance by meningococci is more complicated than was known hitherto. The description of four new mutations, all observed in strains selected in vitro rather than in naturally occurring strains, that have not to our knowledge been reported previously for N. meningitidis raises the total number of known mutations to ten. In addition, the strains examined here displayed various levels of resistance ranging from 8 to over 256 µg ml^-1. However, although the strains were selected stepwise, all but one of the resistant strains were found to have only one altered amino acid site within the rpoB fragment analysed. This result agrees with the observation that, although a stepwise increase in MIC was observed, resistance was acquired within one step once the strains under examination exceeded an MIC of 4 µg ml^-1.

It was discussed before that particularly high-level rifampicin resistance may be acquired by a point mutation in rpoB and by another, still-uncharacterized mechanism(s) (Abadi et al., 1996). However, we did not find alterations in the mtrR promoter region when checking for the presence of an 158 bp insertion (results not shown). It could be questioned that only a fragment of rpoB was sequenced, leaving the possible occurrence of other mutations that we did not identify. Indeed, Jin & Gross (1988) described 19 different mutations that conferred resistance among 45 rifampicin-resistant strains of E. coli. Amongst the mutations described, two are outside the region sequenced so far in meningococci. This means that, in theory, more than the particular point mutations that we have detected could have conferred resistance by means of a cumulative mechanism. To rule out the contribution of other undetected mutations to rifampicin resistance in the strains studied, a PCR-amplified subgenic rpoB fragment harbouring the Ser⁶⁰⁰ mutation, which was found in strains displaying different MICs, was transformed into a wild-type recipient N. meningitidis strain. The resulting transformants were found to be resistant to levels of 256 µg rifampicin ml^-1. These results suggest that high-level resistance can be transferred by a single point mutation into a susceptible acceptor strain. However, transformation rates in our experiments were extremely low. This may be explained because we used PCR amplicons rather than genomic DNA for transformation.

A number of the spontaneous mutant strains were also sequenced at the rpoB locus in order to describe their resistance mechanisms. Interestingly, all 12 isolates studied displayed the mutation at position 1656, leading to a substitution at amino acid position 552, known to be the most frequent described substitution in rifampicin-resistant meningococci. This mutation seems therefore to be a hot spot in the subgenic region of rpoB, both in resistant strains from the field as well as in experimental strains. Spontaneous mutants were observed on plates containing both 32 and 256 µg rifampicin ml^-1.

In summary, we have described new mutations that lead to rifampicin resistance in meningococci, but we did not find evidence for more than a one-step mechanism of acquisition of resistance. Our data cannot explain the presence of different levels of resistance. For instance, the newly described mutation Ser⁶⁰⁰→Leu was found in three strains that displayed MICs of about 8, 16 and >256 µg ml^-1. Transformed meningococci, having adopted the Ser⁶⁰⁰ mutation, however, were found to be of the high-level resistance phenotype. It should be considered that a phenotypic adaptation confers high-level resistance to rifampicin rather than additional mutations or additional genetic mechanisms acting synergistically with the rpoB mutations described already.

The results described in this paper were presented in part at the 51st Conference at the Deutsche Gesellschaft für Hygiene & Mikrobiologie, 30 September4 October 2001, Aachen, Germany.