Nucleotide Sequence Diversity of the bexA Gene in Serotypeable Haemophilus influenzae Strains Recovered from Invasive Disease Patients in Canada

Abstract

The bexA genes of 36 Haemophilus influenzae isolates were sequenced to reveal their nucleotide sequence diversity, which divided them into two groups, similar to clonal divisions I and II. This sequence diversity may lead to false-negative PCR results for H. influenzae infections if bexA is the chosen gene target.

Haemophilus influenzae is a pathogenic, gram-negative coccobacillus that causes a wide variety of infections in children and adults. One of the virulence factors of H. influenzae is the polysaccharide capsule. Six different capsular types have been identified based on specific antisera that recognize antigenic specificities of the different capsular structures. These six capsular types are referred to as serotypes a to f (14). Strains that lack the polysaccharide capsule (nonencapsulated) are termed nontypeable. By using multilocus enzyme electrophoresis (MLEE) to study the population genetics of encapsulated H. influenzae isolates from a wide variety of geographical sources, two phylogenetic groups, clonal divisions I and II, have been identified (12). Most isolates of H. influenzae from clinical infections belong to clonal division I (13). It has been suggested that the patterns of organization of the capsule (cap) loci in clonal division I and II strains are different, partly related to the association of these loci with either the sodC gene or the insertion element IS1016 (16).

The genes involved in the synthesis of the polysaccharide capsule structure are organized in the cap locus, which can be divided into three regions. Region 1 contains genes (bexA to bexD) that encode proteins involved in capsule transport (6). Region 2 contains genes involved in the synthesis of serotype-specific polysaccharide (18), and genes in region 3 appear to be involved in some postpolymerization processing of the capsule, similar to what takes place in Neisseria meningitidis and Mannheimia (Pasteurella) haemolytica as described previously (4, 10, 16). Most invasive H. influenzae infections prior to the introduction of the H. influenzae serotype b (Hib) conjugate vaccine were due to strains of serotype b (20). With the routine use of the Hib conjugate vaccine, infections due to Hib have declined substantially, and now most invasive infections are due to non-Hib strains (2, 17). Nucleic acid amplification methods for the nonculture detection of H. influenzae in clinical specimens are designed to detect the bexA gene (1) as a marker for H. influenzae based on the bexA gene sequence of the Hib Eagan strain (19). However, reports of potential failure to detect some non-Hib serotypes in clinical specimens by using PCR assays that target the bexA gene (1, 15) have prompted us to examine sequence diversity among the bexA genes of non-Hib serotypes. This study compares the bexA gene sequences of 36 strains representing all six serotypes (a, b, c, d, e, and f). To accomplish this, we selected 31 clinical isolates for this study, as well as 5 H. influenzae serotypeable strains from the American Type Culture Collection (ATCC; Manassas, VA) and the National Collection of Type Culture (NCTC; Health Protection Agency, London, United Kingdom): ATCC 9006, serotype a; ATCC 33533, serotype b; ATCC 9007, serotype c; ATCC 9008, serotype d; and NCTC 10479, serotype e. The 31 clinical isolates (8 serotype a, 2 serotype b, 2 serotype c, 1 serotype d, 6 serotype e, and 12 serotype f isolates [including 2 serotype f isolates from a single patient]) were from patients with invasive disease and were provided by provincial public health laboratories across Canada. The identities of the isolates were confirmed by standard biochemical tests (5), and their serotypes were determined by a slide agglutination assay using antisera from commercial sources (Difco, Oakville, Ontario, Canada, and Denka Seiken, Tokyo, Japan).

PCR detection of sodC genes.

Since the sodC gene has been found to be present in phylogenetic division II strains only and not in division I strains, with the exception of those of serotype e (7, 9), we looked for the sodC gene in H. influenzae isolates by PCR. Degenerate universal primers (5′-CAYSAAAAYCCAAGCTG-3′ and 5′-CAYMCGYGSGCCGSCRCCRCC-3′) were used to amplify an approximately 310-bp conserved fragment of the prokaryotic sodC gene under the PCR conditions described by Kroll et al. (8). Of the 36 H. influenzae isolates tested, 21 were found to contain the sodC gene by PCR. These included 2 out of 9 serotype a isolates (ATCC 9006 and one clinical isolate, 2006-0017), all 7 serotype e isolates (including NCTC 10479 and the 6 clinical isolates), and all 12 serotype f clinical isolates. None of the other serotype a (7 isolates), serotype b, serotype c, or serotype d isolates were positive for the sodC gene. All H. influenzae serotype f strains and a minority of H. influenzae serotype a and Hib strains are known to belong to clonal division II. Although classified as clonal division I, strains of H. influenzae serotype e have been described as being distinctively different from other clonal division I strains (12) and are the only clonal division I strains reported to have the sodC gene (7, 9).

Comparison of bexA gene sequences from H. influenzae isolates of serotypes a to f.

A 647-bp fragment of the bexA gene, starting with the fifth nucleotide after the ATG start codon and encompassing the whole 3′ end of the gene to the TAA stop codon, from each of the 36 H. influenzae isolates was sequenced. Primers for the PCR amplification of bexA genes included HI-1 and HI-2, previously described by Falla et al. (3); capfSodC and capfBexA, described by Satola et al. (16); and bexAF3 (5′-GGAGTTGAGCCACAATGAT-3′) and bexAR7 (5′-CGCTATGGGCAAAAGTATCTC-3′). For DNA sequencing, PCR products were purified with the QIAquick PCR purification kit (QIAGEN Inc., Mississauga, Ontario, Canada) and subjected to sequence analysis using DNA analyzer 3730xl (Applied Biosystems, Foster City, CA). Generated DNA sequences of both strands were edited, assembled, and aligned using software from DNAStar, Inc., Madison, WI.

The bexA genes from three serotype b isolates (including ATCC 33533) showed 100% sequence homology to one another as well as to the published sequence of the bexA gene of the Eagan strain, also a serotype b strain. The bexA gene sequences of three serotype c and two serotype d isolates were nearly identical to that of the Eagan strain. The bexA sequences of the serotype c and d isolates differed by only 3 and 2 bp, respectively, from that of the Eagan strain. The two serotype d isolates had identical bexA gene sequences, and there was also no difference in the bexA genes among the three serotype c isolates studied.

The bexA gene sequence data from the nine serotype a isolates showed two distinct groups: sequences from isolates with the sodC gene (presumably clonal division II, including ATCC 9006 and one clinical isolate, 2006-0017) and sequences from isolates without the sodC gene (presumably clonal division I, represented by seven clinical isolates). The bexA gene sequences of the clonal division I serotype a isolates (represented by that of clinical isolate 2006-0018 in Fig. S1 in the supplemental material) showed nearly 100% homology to the bexA gene sequence of the Eagan strain (also clonal division I). Of the 647 bp sequenced, only 1 bp differed between these two serotypes. In contrast, the bexA gene sequences of the clonal division II serotype a isolates (represented by those of ATCC 9006 and the clinical isolate 2006-0017 in Fig. S1 in the supplemental material) showed substantial differences from that of the serotype b Eagan strain, with variations at 97 nucleotide positions. These differences were not concentrated in any specific region but were distributed almost evenly over the entire gene and were also found in the HI-1 and HI-2 primer sites used for the PCR detection of bexA genes in clinical specimens (3, 19). Nucleotide sequence diversity among bexA genes was also found in the DNA probe site used by Sam and Smith (15) for the real-time PCR detection of H. influenzae.

The bexA genes from 12 clinical serotype f isolates showed 100% sequence homology to one another and were very similar to that of the clonal division II serotype a strain ATCC 9006 but differed substantially from the bexA genes of the clonal division I serotype a strains or serotype b Eagan strain. Only 25 bp differed between the bexA genes of serotype f and clonal division II serotype a strains, and these nucleotide differences appeared to be concentrated in the 5′ ends of the genes (Fig. S1 in the supplemental material). Of the 25 bp in the bexA genes of serotype f strains that differed from the nucleotides found in the bexA genes of clonal division II serotype a strains, 23 were identical to the nucleotides found in serotype e strains (Fig. S1 in the supplemental material).

All seven serotype e isolates were found to have identical bexA gene sequences, which differed at 102 bp from the bexA gene sequence of the serotype b Eagan strain. Like those in the clonal division II serotype a isolates, the nucleotide differences found were not concentrated in any particular area of the bexA gene but were evenly distributed throughout the entire gene. Serotype e bexA genes differed at 53 and 31 bp from the bexA genes of clonal division II serotype a and serotype f strains, respectively. The percentages of identity between the bexA genes of all six serotypes of H. influenzae representing both clonal divisions I and II are summarized in Table 1. A phylogenetic tree based on the bexA gene sequence data was generated using BioNumerics software version 3.5 (Applied Maths, Kortrijk, Belgium). Isolates were divided into two clusters similar to the clonal divisions I and II defined by MLEE, with the only exception that serotype e isolates were found to be more closely related to clonal division II isolates than to those belonging to clonal division I (Fig. 1).

TABLE 1.

Comparison of the bexA gene sequences among serotypeable strains of H. influenzae

Serotype corresponding to bexA sequence	% Difference from or identity toabexA sequence from serotype:
	ab	b	c	d	e	f	ac
ab		99.8	99.4	99.5	84.2	84.1	85.0
b	0.2		99.5	99.7	84.2	84.1	85.0
c	0.6	0.5		99.2	83.8	83.6	84.5
d	0.5	0.3	0.8		84.1	83.9	84.9
e	15.8	15.8	16.2	15.9		95.2	91.8
f	15.9	15.9	16.4	16.1	4.8		96.1
ac	15.0	15.0	15.5	15.1	8.2	3.9

^aEach blank space represents the comparison of a sequence to itself; numbers below the blank space in each column are percentages of difference, and those above the blank space are percentages of identity.

^bSerotype a clinical isolate 2006-0018.

^cSerotype a strain ATCC 9006.

FIG. 1.

Phylogenetic tree based on bexA gene sequences of all six serotypes of H. influenzae comprising strains from clonal divisions I and II.

The aim of this study was to examine the sequence diversity of the bexA genes in all six serotypes of H. influenzae, but in particular serotypes e and f. Therefore, the selection of isolates for inclusion in this study had nothing to do with their prevalence in our patient population. Although different from each other, isolates of both serotypes e and f were found to have, within each serotype, 100% sequence homology in their bexA genes.

This study provides new information on the grouping and relationship of typeable H. influenzae strains. The bexA gene sequence data from serotypes a to f appeared to group isolates into two divisions similar to the phylogenetic findings from MLEE, which divide the typeable H. influenzae isolates into clonal divisions I and II (12). If the bexA gene sequence grouping can be confirmed with a larger collection of strains, it will help to track the spread of these clonal groups of typeable H. influenzae since very few laboratories have the capability of carrying out the laborious method of MLEE. Also, the MLEE method is very hard to standardize, and it is equally difficult to compare MLEE data obtained from different laboratories. Therefore, bexA gene sequence analysis may be an alternative to MLEE. Multilocus sequence typing is a more portable method than MLEE and also provides clustering analysis of H. influenzae isolates very similar to that provided by MLEE (11). However, the data presented by Meats et al. (11) suggest possible difficulties in separating H. influenzae isolates into the traditional clonal divisions I and II.

The bexA gene sequence of serotype e was found to be more similar to those of serotype f and two clonal division II serotype a strains (ATCC 9006 and clinical isolate 2006-0017). This finding may suggest that serotype e is potentially more similar to H. influenzae isolates in clonal division II than to those in clonal division I, even though it is classified as clonal division I serotype (see above). The presence of the sodC gene in serotype e isolates may add further support to this hypothesis.

Our data on the bexA gene sequence diversity among the six serotypes of H. influenzae confirmed the nucleotide sequence variations among the serotypes a, e, and f suggested by Corless et al. (1). The real-time PCR method described by these authors failed to detect bexA genes from these serotypes. Sam and Smith (15) also proposed sequence variations in the bexA genes of serotypes e and f as an explanation for the failure to detect these serotypes of H. influenzae by their real-time PCR protocol. However, their study was based on only four strains (two of each serotype). Our data with 18 independently collected isolates (7 serotype e and 11 serotype f strains, since 2 serotype f isolates were from the same patient) provide further proof of these earlier findings.

With more non-Hib strains causing invasive diseases (such as meningitis and septicemia) (2, 17) and the findings of extensive sequence diversity in their bexA genes, our next objective is to look for new PCR primer sites within the bexA gene in order to improve on present PCR detection methods for this pathogen. However, any PCR protocol that targets only the bexA gene will not detect invasive disease cases caused by nontypeable H. influenzae. Our studies in Manitoba as well as data collected from other countries have already identified nontypeable H. influenzae strains as a significant cause of invasive disease (2, 17). Therefore, we are presently also examining a number of H. influenzae-specific gene targets that would allow the detection of both typeable and nontypeable strains.

In conclusion, our data on the bexA genetic diversity should caution laboratories using PCR methods to diagnose H. influenzae infections to be mindful of potential false-negative results if their methods target the amplification of the bexA gene for positive detection.

Nucleotide sequence accession numbers.

The bexA gene sequences of H. influenzae isolates belonging to serotypes a to f have been deposited in GenBank and were given the accession numbers EF490494 to EF490501.