Characterization of Genetic and Phenotypic Diversity of Invasive Nontypeable Haemophilus influenzae
http://www.100md.com
感染与免疫杂志 2005年第9期
Bacterial Pathogenesis Program, Seattle Biomedical Research Institute, Seattle, Washington 98109
Division of Biological Sciences, University of Missouri, Columbia, Missouri 65211
Centro Nacional de Microbiología, Instituto de Salud Carlos III, Carretera de Pozuelo, 28220 Majadahonda, Madrid, Spain
Infectious Diseases and Clinical Microbiology, Nuffield Department of Clinical Laboratory Sciences, John Radcliffe Hospital, Oxford, United Kingdom
Atlanta Veterans Affairs Medical Center and Department of Medicine, Emory University School of Medicine, Decatur, Georgia 30033
Department of Pediatrics, University of North Carolina, Chapel Hill, North Carolina 27599
Department of Pediatrics, University of Arkansas for Medical Sciences, Arkansas Children's Hospital, Little Rock, Arkansas 72202
Institute for Infectious Diseases, University of Bern, Bern, Switzerland
Netherlands Vaccine Institute (NVI), NL-3720 AL Bilthoven, The Netherlands
ABSTRACT
The ability of unencapsulated (nontypeable) Haemophilus influenzae (NTHi) to cause systemic disease in healthy children has been recognized only in the past decade. To determine the extent of similarity among invasive nontypeable isolates, we compared strain R2866 with 16 additional NTHi isolates from blood and spinal fluid, 17 nasopharyngeal or throat isolates from healthy children, and 19 isolates from middle ear aspirates. The strains were evaluated for the presence of several genetic loci that affect bacterial surface structures and for biochemical reactions that are known to differ among H. influenzae strains. Eight strains, including four blood isolates, shared several properties with R2866: they were biotype V (indole and ornithine decarboxylase positive, urease negative), contained sequence from the adhesin gene hia, and lacked a genetic island flanked by the infA and ksgA genes. Multilocus sequence typing showed that most biotype V isolates belonged to the same phylogenetic cluster as strain R2866. When present, the infA-ksgA island contains lipopolysaccharide biosynthetic genes, either lic2B and lic2C or homologs of the losA and losB genes described for Haemophilus ducreyi. The island was found in most nasopharyngeal and otitis isolates but was absent from 40% of invasive isolates. Overall, the 33 hmw-negative isolates were much more likely than hmw-containing isolates to have tryptophanase, ornithine decarboxylase, or lysine decarboxylase activity or to contain the hif genes. We conclude (i) that invasive isolates are genetically and phenotypically diverse and (ii) that certain genetic loci of NTHi are frequently found in association among NTHi strains.
INTRODUCTION
Haemophilus influenzae is a small, fastidious gram-negative coccobacillus that colonizes the human nasopharynx, usually without causing symptoms. When disease occurs, it is usually limited to local infections of the respiratory tract such as otitis media in young children, bronchitis or sinusitis in adults, or pneumonia secondary to chronic pulmonary disease. Invasive disease such as bacteremia or meningitis is currently uncommon in the developed world and is nearly always associated with strains possessing polysaccharide capsules. The six serotypes of H. influenzae have capsules with different carbohydrate structures, with serotype b being the most common. Since the introduction of conjugate vaccines against the type b capsular polysaccharide, the incidence of life-threatening H. influenzae infection has decreased substantially. As expected, the vaccine has not affected the incidence of infections due to unencapsulated (nontypeable) H. influenzae (NTHi). It was once thought that invasive NTHi disease occurred only in children with immunologic or anatomical defects that predispose them to bacterial infections. More recently it has become apparent that NTHi can cause bacteremia and meningitis in otherwise healthy children. One of the first well-documented cases of invasive NTHi was reported in 1996 (33), and since then there have been several publications describing additional cases (5, 6, 8, 34). It is not known whether NTHi isolates from invasive infections possess virulence determinants not shared with isolates associated with otitis media or isolates asymptomatically colonizing the nasopharynx. It is important to determine whether NTHi strains isolated from invasive infections represent distinct lineages or have novel virulence genes not shared with commensals or isolates from localized infections.
The genetic diversity of encapsulated H. influenzae and NTHi has been studied using ribotyping, restriction fragment length polymorphism (RFLP), multilocus enzyme electrophoresis, and multilocus sequence typing (MLST) (4, 26, 32, 43). Phylogenetic trees generated using these methods have shown that whereas encapsulated isolates cluster by serotype, NTHi isolates are much more diverse, with less evidence of clonality (26, 32). The extent to which disease isolates of NTHi are phylogenetically similar has not been studied extensively. van Alphen et al. reported that isolates from acute disease (otitis media or meningitis) are less phylogenetically diverse than isolates from chronic infection (cystic fibrosis or chronic bronchitis) (43). Classically, H. influenzae strains were grouped into "biotypes" by biochemical assays. Kilian (23) described a system for biotyping H. influenzae based on the presence or absence of three enzymatic activities: urease, ornithine decarboxylase, and the production of indole from tryptophan. Biotyping was useful because of weak correlations with pathogenicity. Serotype b strains, particularly those isolated from systemic disease, usually possess all three activities, making them biotype I; conjunctival isolates are usually biotype III and urogenital isolates type IV. Mucosal isolates, whether from healthy children or associated with disease, are more diverse in biotype (23). Indole production, associated with a tryptophanase (tna) gene cluster and with several biotypes, is more frequent in isolates from mucosal and invasive disease than in commensals or in conjunctivitis (25).
Respiratory tract isolates differ in the fermentation of fructose, maltose, and xylose (22, 41). The API 20E system, developed for testing multiple biochemical reactions in clinical isolates of Enterobacteriaceae, has shown several additional activities to be present in some H. influenzae strains but not in others, including arginine dihydrolase and lysine decarboxylase activities and arabinose fermentation (18). For NTHi, none of these biochemical activities has been shown to be strongly associated with a specific disease or with asymptomatic carriage, though it has been noted that lower respiratory tract isolates from patients with cystic fibrosis are more likely to decarboxylate lysine than are other isolates (17).
Surface molecules of both NTHi strains and encapsulated H. influenzae strains have been shown to be heterogeneous in a variety of ways that may affect pathogenesis. The major outer membrane proteins known as P1, P2, and P5 are particularly diverse in sequence (2). Several genes involved in lipopolysaccharide biosynthesis are variably present in H. influenzae strains, and some of these are phase variable. The resulting heterogeneity of lipopolysaccharide structure affects susceptibility to complement-mediated serum killing and the interaction of bacteria with host cells (40). H. influenzae isolates also differ in protein adhesins: most contain genes for one of the adhesins known as Hia and HMW, but not both, and some strains are also able to synthesize fimbriae (38). A recent study reported that otitis isolates are more likely than throat isolates to contain hmw genes, while throat isolates were more likely than isolates from blood or the middle ear to hybridize with a probe for the hifBC genes (required for synthesis of fimbriae) (10).
We undertook these studies to determine whether isolates of NTHi associated with invasive disease differ phenotypically or genetically from isolates cultured from healthy children or associated with acute otitis media. We previously reported that the invasive NTHi strain Int1 (later R2866) differs from most NTHi strains in being unusually resistant to the bactericidal activity of normal adult human serum (49). We have also found that strain R2866 contains several genetic loci that are absent from the sequenced strain Rd KW20 and from several nontypeable strains. These include the gene for an autotransporter, lav, located between holB and tmk (9), the lysogenic bacteriophage HP2 (48), the tna cluster (25), and a 53-kb plasmid similar to other large integrative plasmids of Haemophilus spp. (28). In the present work a collection of 17 NTHi isolates associated with invasive disease were compared with isolates from healthy children or associated with acute otitis media in order to determine whether invasive isolates are genetically or phylogenetically distinct. We assayed the susceptibility of each strain to complement-mediated killing by pooled normal human serum and tested each in a panel of biochemical assays. In addition, we used PCR to screen for the presence of genes for HP2, Lav, fimbriae, and the adhesins Hia and HMW, the 53-kb plasmid, and the lipopolysaccharide biosynthetic locus flanked by infA and ksgA. Invasive strains were genetically heterogeneous and did not differ significantly in the range of serum sensitivity from nasopharyngeal commensals or otitis media isolates. However, several genetic characteristics are not assorted randomly but are strongly associated with one another. In particular, several biochemical and genetic traits were found primarily in strains lacking genes for the hmw adhesin. Within the hmw-negative group we identified a cluster of eight isolates that resembled R2866 in being biotype V and hia positive, lacking an infA-ksgA island, and having the urease locus replaced by a homolog of the gonococcal membrane protein mtrF. These strains were shown by MLST to be members of a phylogenetic cluster.
MATERIALS AND METHODS
Bacterial strains and growth. H. influenzae strains are listed in Table 1. We studied three types of isolates: nasopharyngeal or throat isolates from healthy children, strains cultured from middle ear aspirates from children with otitis media, and blood or cerebrospinal fluid isolates obtained from children with invasive disease. For simplicity, isolates are referred to below as "throat," "otitis," or "invasive." Strains were isolated between 1995 and 2004, at different locations within the United States, Canada, and Europe. For invasive isolates, we limited the study to isolates from children who were considered by the treating physician to have normal anatomy and immune function. Neonates were excluded. Otitis isolates included the recently sequenced strains 86-028NP (31) and R2846 (strain 12). Invasive isolates included R2866 (Int1), which has also been sequenced recently. Bacteria were cultivated at 37°C on chocolate agar supplemented with 1% IsovitaleX (Becton Dickinson and Co., Sparks, MD) or in Difco brain heart infusion broth (Becton Dickinson and Co.) supplemented (sBHI) with hemin (10 μg/ml) and -NAD (10 μg/ml) and with agar for solid sBHI plates. The requirement of all strains for X and V factors was confirmed by disk diffusion. All strains were negative for bexA by PCR, indicating the inability to express capsular polysaccharide. The serotype b strain Eagan was used as a positive control for bexA PCR.
Biochemical assays. Bacteria were suspended at 108 CFU/ml in 0.9% NaCl containing hemin (10 μg/ml) and -NAD (10 μg/ml) and were used to inoculate API 20E biochemical test strips purchased from bioMerieux (St. Louis, MO) (19). Enzyme activities and fermentation reactions were recorded after 24 h of incubation at 37°C in air. Indole, ornithine decarboxylase, and urease activities were used for biotyping (23).
Screen for genetic markers by PCR. Genomic DNA was prepared using the DNeasy tissue kit (QIAGEN, Inc., Valencia, CA). Primers (Table 2) were synthesized by Integrated DNA Technologies, Coralville, IA. For lav, we amplified across the holB-tmk junction (HI0455 to HI0456 in Rd KW20). Positive strains yielded a product of 2 to 2.2 kb. Partial sequencing confirmed that all were related to the published sequence of lav. The remaining isolates did not yield an amplicon or else produced a 150-bp product consistent with no inserted gene. For detection of bacteriophage HP2, we used PCR to amplify regions of rep (common to both phage HP1 and HP2) and orf10 (specific for HP2). For the plasmid, PCR with the primers listed as topoF and topoR gave a product of about 450 bp, and primers AF and AR gave a product of about 500 bp. We used primers designed to amplify conserved regions of the adhesin genes hmwA and hia (36, 44). PCR with the hmw primers gave a product of approximately 1.2 kb. Occasional faint products were sequenced and were scored negative if the sequences were not related to hmwA. hia products were usually 3 to 4 kb but ranged from approximately 0.75 kb to 10 kb. The basis of this size difference was not evaluated. Ten of the hia amplicons were sequenced and found to be similar to the published hia sequence. Of four strains yielding hia products of only 0.75 kb, two were sequenced and found to be homologous to the 3' end of the hia gene. All of these hia-related sequences were scored as positive for hia. The presence and size of the hif locus, required for synthesis of fimbriae, was evaluated by amplification across the purE-pepN junction. Isolates with a complete hif locus produced a product of approximately 7 kb. These PCR fragments were sequenced using the purE primer to detect the presence of hicA and hicB in addition to the hif genes. As seen previously (27), the remaining isolates contained either a shorter purE-pepN insert, reflected by a product of approximately 0.5 to 1 kb (presumed to include hicAB), or no insert, reflected by a PCR product of approximately 250 bp. The genetic island flanked by infA and ksgA was detected by PCR across the infA-ksgA junction. The contents of the 2.2-kb island were determined either by sequencing or by PCR using primers infA and losB-R to detect losAB and primers lic2B-F and lic2BA to detect lic2BC. Routine PCR was carried out using the Biolase DNA polymerase (Bioline USA, Randolph, MA), with the following amplification protocol: reaction tubes were incubated at 94°C for 2 min, then for 36 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min per kb of predicted product. The Expand Long Template PCR system (Roche Diagnostics Company, Indianapolis, IN) was used as directed for amplification of the hif and hia loci. Following electrophoresis, ethidium bromide-stained agarose gels were visualized and photographed using the Fluorochem 8900 digital imaging system (Alpha Innotech Corporation, San Leandro, CA). PCR products to be sequenced were processed using the Qiaquick PCR purification kit (QIAGEN, Inc.) and submitted to the core sequencing facility at Seattle Biomedical Research Institute. Sequences were aligned using the BioEdit Sequence Alignment Editor (Isis Pharmaceuticals, Inc., Carlsbad, CA).
MLST. PCR for the housekeeping genes adk, atpG, adk, atpG, frdB, fucK, mdh, pgi, and recA was carried out using the primers and methods described in reference 26. Sequences were submitted to the MLST website (www.mlst.net) for allele and sequence type (ST) assignment.
Serum bactericidal activity. Log-phase bacteria (2,000 CFU/ml) were incubated for 30 min at 37°C with pooled normal human serum diluted in 10 mM phosphate-buffered saline containing 4 mM KCl and 0.1% gelatin and were then plated to determine bacterial survival. The concentration of serum that killed 50% of bacteria was calculated using XLfit 4.1 (ID Business Solutions, Guildford, United Kingdom) and is referred to as the IC50 of the serum for that strain.
Statistics. Groups were compared using the Fisher exact probability test. Calculations were performed online using VassarStats: Web Site for Statistical Computation (http://faculty.vassar.edu/lowry/VassarStats.html).
Sequence data. Sequences for the genomes of H. influenzae strains Rd KW20, R2846, R2866, and 86-028NP and of Neisseria gonorrhoeae FA1090 and Neisseria meningitidis strains MC58 and Z2491 were accessed through the Microbial Genomes pages at the website of the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/genomes/lproks.cgi). The sequences of the mtrF (strain R2866) and losAB (strain R2846) loci were submitted to NCBI and were assigned accession numbers DQ007025 and DQ007026, respectively.
RESULTS
Biotyping and other biochemical characterization. For H. influenzae type b, invasive isolates nearly always belong to biotype I (positive for all three of the biotyping reactions), with biotype II (negative for ornithine decarboxylase only) seen occasionally (14). In contrast, of the invasive NTHi isolates studied here, only one-third belonged to biotype I and about one-fourth belonged to biotype V (negative for urease only) (Table 3). We noted heterogeneity among strains in several other biochemical reactions. Arginine dihydrolase and arabinose fermentation were seen predominantly for throat isolates, while lysine decarboxylase and growth on citrate were seen less often for middle ear isolates than for the other two groups (Table 3). In most cases these correlations were not statistically significant.
Correlation of the hmwA gene with biochemical reactions and genes for other surface molecules. We evaluated strains for the presence of genes encoding the HMW adhesins by PCR amplification of a conserved region of hmw1A and hmw2A. The results correlated strongly with the biochemical reactions discussed above and with the presence of genes encoding other surface molecules (Table 3). hmw-negative isolates were much more likely than hmw-positive isolates to be positive for indole, ornithine decarboxylase, lysine decarboxylase, and arginine dihydrolase activities, to be negative for urease activity, and to ferment arabinose and grow on citrate. hmw-negative isolates were nearly always positive for hia-related sequences, as reported previously (39). One isolate, R3579, contained sequences for both hmw and hia. This result was confirmed by sequencing the products and by carrying out PCR on five separate colonies. Each of the five colonies yielded both a hia product (approximately 3 kb) and a hmw product, indicating that the culture was not a mixture of a hia-positive and a hmw-positive strain. We used PCR to determine the presence and size of the fimbrial gene cluster flanked by purE and pepN. We found that 11 isolates contained a 7-kb purE-pepN insert consistent with a complete fimbrial biosynthetic locus; 10 of these isolates were hmw negative. Of 15 isolates lacking a purE-pepN insert, 14 were also hmw negative. Of the 20 hmw-positive isolates, 18 had a purE-pepN insert of 0.8 to 1 kb. Such inserts have been reported to contain two genes of unknown function, hicA and hicB. This correlation between hmw and the fimbrial gene cluster has not been reported previously.
Lipopolysaccharide biosynthetic island flanked by infA and ksgA. The lipopolysaccharide biosynthetic genes now known as lic2B and lic2C were initially described for a type b strain, RM7004 (originally strain 760705 from the van Alphen laboratory, isolated from cerebrospinal fluid). In this strain they are flanked by infA and ksgA (homologs of HI0548 and HI0549 in Rd KW20) (16, 20). Comparison of the infA-ksgA region in the available H. influenzae genome sequences showed that in strains Rd KW20 and R2866, the infA and ksgA genes are adjacent. Strain 86-028NP contains a single gene, lic2C, separating infA and ksgA. In strain R2846, infA and ksgA are separated by two genes that are unrelated to lic2B and lic2C. These genes are strongly homologous to losA and losB (also known as lbgB and lbgA), characterized in Haemophilus ducreyi as encoding -1,4-galactosyltransferase and D-glycero-D-manno-heptosyltransferase, respectively (13, 42). The infA-ksgA junction thus appears to be a site of substantial genetic heterogeneity, with at least four complements of genes seen in different isolates (Fig. 1).
We used PCR across the infA-ksgA junction to evaluate the diversity of this region in the same 53 isolates that were used for the biochemical studies described above. Most of the throat and mucosal-disease isolates in our study contained a genetic island of 2.2 kb. Of 12 isolates lacking the island, 6 were invasive, 4 were otitis isolates, and only 2 were throat isolates. Partial sequencing of the amplified ksg-infA regions showed that the distribution of losAB and lic2BC-related sequences differed among the three groups: only one invasive isolate, but about one-quarter of throat and otitis isolates, contained losAB-related sequences. Most of the remaining strains contained lic2B and lic2C, but four contained only lic2C (Table 4).
These data suggest that invasive isolates of NTHi may be less likely than other NTHi isolates to contain an island flanked by infA and ksgA, and also less likely to contain an island consisting of losAB. However, this is based on a fairly small number of isolates. We used PCR to investigate the infA-ksgA regions of 19 additional invasive isolates of NTHi received from laboratories in the United States, Canada, and Switzerland. Nine of these isolates lacked the infA-ksgA island. Thus, of the 72 isolates studied, we identified 21 isolates that lack the island. Of these, 15 were invasive (P = 0.04). The 19 additional invasive isolates also included 7 with sequences related to lic2BC and 3 with sequences related to losAB, consistent with the earlier observation that losAB is relatively rare among invasive isolates.
Identification of group of closely related isolates. The presence or absence of the infA-ksgA island and its content correlate with several of the genetic and biochemical features known to be variably present among H. influenzae strains (Table 4). Isolates lacking the infA-ksgA island were more likely than other isolates to contain lav or the HP2 bacteriophage. Two-thirds of the isolates lacking this island were urease negative but positive for indole and ornithine decarboxylase, making them members of the relatively rare biotype V (23). All biotype V strains were hia positive (P = 0.007). The similarities of the biotype V isolates led us to look carefully for differences among them (Table 5). Although these isolates have several features in common, they are heterogeneous with regard to the presence of lysine decarboxylase activity, production of nitrite from nitrate, arabinose fermentation, and the presence of the lav and HP2 genes. It thus seems unlikely that they represent a recently derived clone. We used MLST to determine the STs for 11 strains that are biotype V, hmw negative, and negative for the infA-ksgA island and for 13 strains of biotypes I, II, and III. Eight of the biotype V strains formed a more distinct cluster than any of the other strains, while one belonged to a neighboring lineage; the remaining two biotype V strains were closely related to each other but not to other strains (Fig. 2).
Replacement of the urease locus by an mtrF homolog. The strong correlation between urease activity and the presence of an infA-ksgA island (P < 0.0005) may be due to genetic linkage, since these two loci are separated by only 4 kb. A search of the available H. influenzae genome sequences revealed that strains Rd KW20, R2846, and 86-028NP each contain a cluster of seven genes (designated HI0535 to HI0541 in Rd KW20) with homology to the well-characterized urease genes of Klebsiella aerogenes (30). In each case this locus is flanked by homologs of aspA and groES. In strain R2866, which is urease negative, the seven-gene cluster is replaced by a single gene with homology to the gonococcal gene mtrF (Fig. 1). Using PCR primers that hybridized to the aspA and groES genes, we were able to amplify the expected 9-kb product from genomic DNA of each of four urease-positive strains. Partial sequencing of the PCR products showed that each was similar to the urease locus of strain Rd KW20. The same PCR primers yielded a 4-kb product from each of nine urease-negative strains, consistent with the sequence of R2866. Partial sequencing of these 4-kb products showed that all were very similar to that of R2866.
RFLP analysis of the hif locus. In previous work (T. Mhlanga-Mutangadura and M. Golomb, unpublished) on the heterogeneity of the hif locus, we had identified three NTHi isolates, R3101, R3151, and R3157, that had hifA sequences nearly identical to that of R2866 and in which the 7-kb hif locus had a PstI RFLP pattern similar to that of R2866. We now extended that work to the strains found in this study to have a 7-kb hif locus. Consistent with previous work, several patterns were identified. R3265, a throat isolate, had the same RFLP pattern as R2866. All of the isolates with this pattern belonged to biotype V and lacked the infA-ksgA island.
Serum resistance. Our prototype invasive nontypeable strain, R2866 (Int1), had previously been shown to survive in 40% normal human serum much longer than the laboratory strain Rd KW20 or the nontypeable strain U11 (49). We hypothesized that the in vitro serum resistance of R2866 reflected a virulence trait that might be shared with other invasive NTHi isolates. In the present study we used a more quantitative assay to evaluate the serum resistance of the 53 clinical isolates described above. We found that the range of IC50s for clinical NTHi isolates was nearly as wide as the difference between encapsulated and unencapsulated strains and that R2866 was one of the most resistant strains in our collection. However, the 16 additional invasive isolates that we tested were not overall more serum resistant than otitis isolates and were only slightly more serum resistant than throat isolates. The IC50 range for the invasive isolates was broad, from 1.9% to 13.3%, and this range was similar to that for the other two groups. It is clear that in vitro serum resistance is neither common to all invasive isolates nor a specific characteristic of this group (Fig. 3).
DISCUSSION
While it is well known that NTHi isolates are more heterogeneous than type b isolates, the extent to which the heterogeneous characteristics are correlated with each other or with the anatomic site of isolation is not well understood. We had previously identified genes encoding the autotransporter Lav and bacteriophage HP2 in the invasive isolate R2866 and hypothesized that they might be involved in the unusual virulence of this strain (9, 48). In this study we found that the lav gene is indeed rare in throat isolates but is present in many otitis isolates and in only about one-fourth of invasive isolates. Only 4 of the 53 isolates studied in this work contained HP2; 3 of these are invasive. R2866 also contains a 53-kb region that appears to be an integrated plasmid similar to those described for other Haemophilus strains (28). We identified sequences characteristic of the plasmid in 11 isolates, only 1 of which was a throat isolate. None of the genetic loci or biochemical reactions we studied were found to be consistently present in invasive isolates. A genetic island flanked by infA and ksgA was identified in most throat and otitis isolates but was absent in 40% of invasive isolates. When present, this island contains lipopolysaccharide biosynthetic genes, as discussed below.
The ease with which H. influenzae strains exchange DNA and the observed genetic heterogeneity among NTHi isolates might have suggested that variable genes would occur randomly. A high rate of recombination in otitis media isolates has been reported (7). However, we found that certain genetic loci occurred together much more often than would occur by chance. Many of the characteristics we examined were significantly associated with the presence of sequences related to hmw or hia genes. Because we detected these genes by PCR amplification of an internal portion of each gene, we cannot be sure that the strains we scored positive for an adhesin actually contain the complete genetic locus required to express HMW or Hia adhesins. Indeed, some strains appeared to contained only a fragment of the hia gene. It is also possible that some of the strains we scored negative contained gene variants that differed slightly at one of the primer binding sites. St. Geme et al. reported that nearly all H. influenzae strains contain genes encoding either the Hia adhesin or the HMW adhesins (39). hia is an allele of the hsf gene found in encapsulated strains. This led to the hypothesis that hia-containing strains may have evolved from an encapsulated ancestor. In support of this idea, the insertion element IS1016, which is found in the cap locus of encapsulated strains, was found to hybridize to genomic DNA of 6 hia-containing NTHi isolates out of 9 isolates tested but not to hybridize to any of the 47 hmw-containing isolates that were tested (39). Our observation that several genetic and biochemical traits are correlated with the presence of a sequence related to hmw or hia is consistent with the idea that these genes are found in two distinct lineages. Indole, ornithine decarboxylase, and lysine decarboxylase were all significantly more prevalent among hia-containing isolates than among hmw-containing isolates. The hmw-containing isolates in our study were uniformly urease positive, while nearly one-third of the hia-containing isolates were urease negative.
Within the hmw-negative, hia-positive isolates, we identified a group of 11 biotype V isolates that shared several characteristics of the invasive strain R2866 (Table 5). All the biotype V strains except one lacked the genetic island between ksgA and infA. Some of the isolates lacking an infA-ksgA island contained a hif locus with a characteristic RFLP pattern. Isolates lacking an infA-ksgA island were also more likely than other isolates to contain lav genes. Bacteriophage HP2 was found only in biotype V strains.
With the exception of the hmw-hia dichotomy (39), associations among variable genetic loci of H. influenzae have not been previously noted. Because natural transformation and other means of genetic exchange readily disrupt linkages, gene associations must be explained by a recent shared lineage or by shared adaptive strategies or requirements for virulence. The urease locus and the infA-ksgA locus are linked within 4 kb and might have persisted in linkage disequilibrium. However, the associations with the hia gene and the hif RFLP pattern cannot be explained in this way. It is possible that the isolates lacking an infA-ksgA island are closely related to each other, although they were derived from different parts of the United States as well as from Canada and Europe. As noted above, the isolates in this group were more likely to have been cultured from blood than from ear aspirates or nasopharyngeal samples. This suggests that they may share virulence determinants. Phylogenetic analysis by MLST showed that 9 of 11 biotype V isolates formed a single cluster, while 2 other biotype V isolates were closely related to each other but not to the cluster of 9. Typing of 13 strains of other biotypes showed some tendency for strains to cluster by biotype and the by presence of an hmw-related sequence. Further evaluation of the correlation between MLST and other characteristics of NTHi is ongoing.
We found that in the urease-negative isolates we studied, the urease locus was invariably replaced by a gene with homology to the mtrF gene found in N. gonorrhoeae and N. meningitidis. In gonococci, mtrF encodes a predicted membrane protein that has been reported to be involved in resistance to detergents and hydrophobic antibiotics, in conjunction with the resistance-nodulation-cell division (RND) pump MtrCDE (45), which does not appear to be present in any of the H. influenzae strains that have been sequenced. A less closely related homolog, abgT, is thought to encode a transporter that allows an Escherichia coli mutant that is defective in p-aminobenzoate synthesis to utilize p-aminobenzoyl-glutamate (21). It thus seems likely that the H. influenzae MtrF homolog is involved in transport, but the substrate cannot be determined without experimental data.
We found that when the infA-ksgA region contains an insert, the locus may consist of lic2BC, as previously seen for type b isolates (15), of lic2C only, or of two genes with homology to losA and losB of H. ducreyi. The prevalence of lic2B and lic2C has been studied previously, though not in invasive isolates. Hood et al. reported that 13 of 27 NTHi strains studied contained lic2C (20). Pettigrew et al. examined 90 throat and 48 otitis isolates and found lic2B in one-half of the otitis isolates but only 14% of the throat isolates (35). In contrast, we found for throat, otitis, and invasive isolates that about half in each group contained lic2BC. However, it is possible that our small sample of throat isolates (17 isolates) had an atypical distribution of genes at this locus. The losAB genes have not been studied previously in H. influenzae. We found that the infA-ksgA island contained losAB in about one-third of our throat and otitis isolates but in very few invasive isolates.
Without further study, it is not possible to determine whether the variation we see in the structure of the infA-ksgA region has a direct role in H. influenzae biology. Certainly the lic2BC and losAB genes are likely to affect the structure of lipopolysaccharide and thereby the interaction of bacteria with the host. It is also possible that these genes are markers for other characteristics that collectively affect the ability of a given strain to colonize or cause disease. We noted that the genome sequences of both R2846 and R2866 contain homologs of losAB at sites distant from the infA-ksgA region. In R2846, the second losA homolog contains a frameshift and appears unlikely to encode an active product.
NTHi strains isolated from the blood of ill children were as likely to be sensitive to serum as throat and otitis isolates. One explanation for this unexpected finding is that our laboratory stocks differ genetically from the bacterial population in the patients during infection. H. influenzae is subject to high-frequency, reversible gain or loss of phenotypes such as hemagglutination (reflecting piliation), colony morphology, or reactivity of lipopolysaccharide with monoclonal antibodies (11, 24, 46). The molecular basis of this phase variation is variation in the length of tandem repeat regions, often within the coding region of genes (47). A growing culture of any H. influenzae strain thus contains a large number of variants that differ from each other in the expression of one or more genes. A change in environmental conditions may result in selection of a different population of variants, and a culture that has been derived from a single colony (as is likely to happen during isolation in a clinical laboratory) and then propagated in the lab may contain a mixture of variants very different from that in the patient's blood. Although R2866, the first invasive NTHi strain studied, has retained serum resistance in the lab, this may be the result of a fortuitous subculture of a resistant colony. The range of serum resistance observed for other clinical isolates may reflect true differences among these isolates but may also indicate that for some isolates we are working with serum-sensitive variants and for other isolates our laboratory stocks happen to consist largely of serum-resistant variants.
Overall, our data indicate that NTHi isolates from invasive disease are not homogeneous: they differ in genotype and in resistance to human serum. It is possible that invasiveness can result from more than one combination of biologic capabilities and that different invasive strains have different sets of genes mediating these functions. It is also possible that all invasive isolates share a gene or group of genes that has not yet been discovered. Several of the biochemical and genetic traits we studied were correlated with each other and particularly with the hia and hmw genes, supporting the idea that NTHi strains may cluster into distinct lineages, although the lines may be blurred by genetic exchange. We have recently determined the complete genome sequence of the invasive strain R2866 and an otitis strain, R2846 (strain 12). Each of these strains contains 30 to 40 genetic loci not present in the previously sequenced strain, Rd KW20. Analysis of these novel loci will add to our understanding of the genetic relationships among NTHi isolates and may identify genetic determinants characteristic of strains associated with invasive disease and with otitis media. This will aid in the understanding of the virulence mechanisms required for nasopharyngeal bacteria to cause mucosal or systemic disease.
ADDENDUM IN PROOF
Since acceptance of this paper, we discovered that the bexA primers listed in Table 2 are able to amplify bexA from serotype b, c and d strains but do not reliably amplify bexA from serotype a or e strains. On testing the strains listed in Table 1 for bexA using the primers described by Falla et al. (J. Clin Microbiol. 32: 2382-2386, 1994), strains R3049, R3172, R3173, R3368, and R3595 were found to be positive. Strain R3368 was biotype I and negative for lysine decarboxylase, arginine dihydrolase, citrate, and arabinose. It was positive for hia and losAB sequences and negative for hmw, lav, HP2, plasmid, hif, and hic sequences. Strains R3049 and R3173 contain the lic2BC genes, and R3172 and R3595 contain losAB. Elimination of R3368-derived data from Table 3 increased each of the reported P values slightly, with each P value remaining <0.02. Eliminating data on the remaining strains from the section entitled "Lipopolysaccharide biosynthetic island flanked by infA and ksgA" did not alter the conclusions.
ACKNOWLEDGMENTS
This work was supported in part by grants AI 44002 and AI 46512 from the National Institute of Allergy and Infectious Disease to A.L.S. and by a grant from the M. J. Murdock Charitable Trust to the Seattle Biomedical Research Institute.
We thank Theodore White and Bruce Torian for critical review of the manuscript. We are grateful for the assistance of David Scheifele (University of British Columbia), Carol Shaw (British Columbia Centre for Disease Control), Stephen Barenkamp (Washington University), Robert Munson and Lauren Bakaletz (Ohio State University), Janet Gilsdorf (University of Michigan), and Bruce Green)Wyeth Vaccines) in obtaining clinical isolates. This publication made use of the Multi Locus Sequence Typing website (http://www.mlst.net/) at Imperial College London, developed by Man-Suen Chan and David Aanensen and funded by the Wellcome Trust. Daniel Godoy, curator of the H. influenzae MLST database, provided assistance with allele and sequence type assignments.
Present address: Department of Veterinary Pathobiology, University of Missouri, Columbia, MO 65211.
REFERENCES
1. Bakaletz, L. O., B. M. Tallan, W. J. Andrzejewski, T. F. DeMaria, and D. J. Lim. 1989. Immunological responsiveness of chinchillas to outer membrane and isolated fimbrial proteins of nontypeable Haemophilus influenzae. Infect. Immun. 57:3226-3229.
2. Barenkamp, S. J. 2004. Rationale and prospects for a nontypable Haemophilus influenzae vaccine. Pediatr. Infect. Dis. J. 23:461-462.
3. Barenkamp, S. J., and E. Leininger. 1992. Cloning, expression, and DNA sequence analysis of genes encoding nontypeable Haemophilus influenzae high-molecular-weight surface-exposed proteins related to filamentous hemagglutinin of Bordetella pertussis. Infect. Immun. 60:1302-1313.
4. Bruce, K. D., and J. Z. Jordens. 1991. Characterization of noncapsulate Haemophilus influenzae by whole-cell polypeptide profiles, restriction endonuclease analysis, and rRNA gene restriction patterns. J. Clin. Microbiol. 29:291-296.
5. Campos, J., M. Hernando, F. Roman, M. Perez-Vazquez, B. Aracil, J. Oteo, E. Lazaro, and F. de Abajo. 2004. Analysis of invasive Haemophilus influenzae infections after extensive vaccination against H. influenzae type b. J. Clin. Microbiol. 42:524-529.
6. Cerquetti, M., M. L. Ciofi degli Atti, G. Renna, A. E. Tozzi, M. L. Garlaschi, P. Mastrantonio, and the Hi Study Group. 2000. Characterization of non-type B Haemophilus influenzae strains isolated from patients with invasive disease. J. Clin. Microbiol. 38:4649-4652.
7. Cody, A. J., D. Field, E. J. Feil, S. Stringer, M. E. Deadman, A. G. Tsolaki, B. Gratz, V. Bouchet, R. Goldstein, D. W. Hood, and E. R. Moxon. 2003. High rates of recombination in otitis media isolates of non-typeable Haemophilus influenzae. Infect. Genet. Evol. 3:57-66.
8. Cuthill, S. L., M. M. Farley, and L. G. Donowitz. 1999. Nontypable Haemophilus influenzae meningitis. Pediatr. Infect. Dis. J. 18:660-662.
9. Davis, J., A. L. Smith, W. R. Hughes, and M. Golomb. 2001. Evolution of an autotransporter: domain shuffling and lateral transfer from pathogenic Haemophilus to Neisseria. J. Bacteriol. 183:4626-4635.
10. Ecevit, I. Z., K. W. McCrea, M. M. Pettigrew, A. Sen, C. F. Marrs, and J. R. Gilsdorf. 2004. Prevalence of the hifBC, hmw1A, hmw2A, hmwC, and hia genes in Haemophilus influenzae isolates. J. Clin. Microbiol. 42:3065-3072.
11. Farley, M. M., D. S. Stephens, S. L. Kaplan, and E. O. Mason, Jr. 1990. Pilus- and non-pilus-mediated interactions of Haemophilus influenzae type b with human erythrocytes and human nasopharyngeal mucosa. J. Infect. Dis. 161:274-280.
12. Geluk, F., P. P. Eijk, S. M. van Ham, H. M. Jansen, and L. van Alphen. 1998. The fimbria gene cluster of nonencapsulated Haemophilus influenzae. Infect. Immun. 66:406-417.
13. Gibson, B. W., A. A. Campagnari, W. Melaugh, N. J. Phillips, M. A. Apicella, S. Grass, J. Wang, K. L. Palmer, and R. S. Munson, Jr. 1997. Characterization of a transposon Tn916-generated mutant of Haemophilus ducreyi 35000 defective in lipooligosaccharide biosynthesis. J. Bacteriol. 179:5062-5071.
14. Harper, J. J., and M. H. Tilse. 1991. Biotypes of Haemophilus influenzae that are associated with noninvasive infections. J. Clin. Microbiol. 29:2539-2542.
15. High, N. J., M. E. Deadman, and E. R. Moxon. 1993. The role of a repetitive DNA motif (5'-CAAT-3') in the variable expression of the Haemophilus influenzae lipopolysaccharide epitope Gal(1-4)Gal. Mol. Microbiol. 9:1275-1282.
16. High, N. J., M. P. Jennings, and E. R. Moxon. 1996. Tandem repeats of the tetramer 5'-CAAT-3' present in lic2A are required for phase variation but not lipopolysaccharide biosynthesis in Haemophilus influenzae. Mol. Microbiol. 20:165-174.
17. Hoiby, N., and M. Kilian. 1976. Haemophilus from the lower respiratory tract of patients with cystic fibrosis. Scand. J. Respir. Dis. 57:103-107.
18. Hollander, R. 1981. Biochemical characterization of Haemophilus strains by using the API 20E and API 50e test system. Zentbl. Bakteriol. Mikrobiol. Hyg. A 250:322-329. (In German.)
19. Holmes, R. L., L. M. DeFranco, and M. Otto. 1982. Novel method of biotyping Haemophilus influenzae that uses API 20E. J. Clin. Microbiol. 15:1150-1152.
20. Hood, D. W., M. E. Deadman, A. D. Cox, K. Makepeace, A. Martin, J. C. Richards, and E. R. Moxon. 2004. Three genes, lgtF, lic2C and lpsA, have a primary role in determining the pattern of oligosaccharide extension from the inner core of Haemophilus influenzae LPS. Microbiology 150:2089-2097.
21. Hussein, M. J., J. M. Green, and B. P. Nichols. 1998. Characterization of mutations that allow p-aminobenzoyl-glutamate utilization by Escherichia coli. J. Bacteriol. 180:6260-6268.
22. Kawakami, Y., Y. Okimura, and M. Kanai. 1982. Occurrence and biochemical properties of Haemophilus species in pharyngeal flora of healthy individuals. Microbiol. Immunol. 26:629-633.
23. Kilian, M. 1976. A taxonomic study of the genus Haemophilus, with the proposal of a new species. J. Gen. Microbiol. 93:9-62.
24. Kimura, A., and E. J. Hansen. 1986. Antigenic and phenotypic variations of Haemophilus influenzae type b lipopolysaccharide and their relationship to virulence. Infect. Immun. 51:69-79.
25. Martin, K., G. Morlin, A. Smith, A. Nordyke, A. Eisenstark, and M. Golomb. 1998. The tryptophanase gene cluster of Haemophilus influenzae type b: evidence for horizontal gene transfer. J. Bacteriol. 180:107-118.
26. Meats, E., E. J. Feil, S. Stringer, A. J. Cody, R. Goldstein, J. S. Kroll, T. Popovic, and B. G. Spratt. 2003. Characterization of encapsulated and noncapsulated Haemophilus influenzae and determination of phylogenetic relationships by multilocus sequence typing. J. Clin. Microbiol. 41:1623-1636.
27. Mhlanga-Mutangadura, T., G. Morlin, A. L. Smith, A. Eisenstark, and M. Golomb. 1998. Evolution of the major pilus gene cluster of Haemophilus influenzae. J. Bacteriol. 180:4693-4703.
28. Mohd-Zain, Z., S. L. Turner, A. M. Cerdeno-Tarraga, A. K. Lilley, T. J. Inzana, A. J. Duncan, R. M. Harding, D. W. Hood, T. E. Peto, and D. W. Crook. 2004. Transferable antibiotic resistance elements in Haemophilus influenzae share a common evolutionary origin with a diverse family of syntenic genomic islands. J. Bacteriol. 186:8114-8122.
29. Mühlemann, K., M. Balz, S. Aebi, and K. Schopfer. 1996. Molecular characteristics of Haemophilus influenzae causing invasive disease during the period of vaccination in Switzerland: analysis of strains isolated between 1986 and 1993. J. Clin. Microbiol. 34:560-563.
30. Mulrooney, S. B., and R. P. Hausinger. 1990. Sequence of the Klebsiella aerogenes urease genes and evidence for accessory proteins facilitating nickel incorporation. J. Bacteriol. 172:5837-5843.
31. Munson, R. S., Jr., A. Harrison, A. Gillaspy, W. C. Ray, M. Carson, D. Armbruster, J. Gipson, M. Gipson, L. Johnson, L. Lewis, D. W. Dyer, and L. O. Bakaletz. 2004. Partial analysis of the genomes of two nontypeable Haemophilus influenzae otitis media isolates. Infect. Immun. 72:3002-3010.
32. Musser, J. M., S. J. Barenkamp, D. M. Granoff, and R. K. Selander. 1986. Genetic relationships of serologically nontypeable and serotype b strains of Haemophilus influenzae. Infect. Immun. 52:183-191.
33. Nizet, V., K. F. Colina, J. R. Almquist, C. E. Rubens, and A. L. Smith. 1996. A virulent nonencapsulated Haemophilus influenzae. J. Infect. Dis. 173:180-186.
34. O'Neill, J. M., J. W. St. Geme III, D. Cutter, E. E. Adderson, J. Anyanwu, R. F. Jacobs, and G. E. Schutze. 2003. Invasive disease due to nontypeable Haemophilus influenzae among children in Arkansas. J. Clin. Microbiol. 41:3064-3069.
35. Pettigrew, M. M., B. Foxman, C. F. Marrs, and J. R. Gilsdorf. 2002. Identification of the lipooligosaccharide biosynthesis gene lic2B as a putative virulence factor in strains of nontypeable Haemophilus influenzae that cause otitis media. Infect. Immun. 70:3551-3556.
36. Rodriguez, C. A., V. Avadhanula, A. Buscher, A. L. Smith, J. W. St. Geme III, and E. E. Adderson. 2003. Prevalence and distribution of adhesins in invasive non-type b encapsulated Haemophilus influenzae. Infect. Immun. 71:1635-1642.
37. Sirakova, T., P. E. Kolattukudy, D. Murwin, J. Billy, E. Leake, D. Lim, T. DeMaria, and L. Bakaletz. 1994. Role of fimbriae expressed by nontypeable Haemophilus influenzae in pathogenesis of and protection against otitis media and relatedness of the fimbrin subunit to outer membrane protein A. Infect. Immun. 62:2002-2020.
38. St. Geme, J. W., III. 2002. Molecular and cellular determinants of non-typeable Haemophilus influenzae adherence and invasion. Cell. Microbiol. 4:191-200.
39. St. Geme, J. W., III, V. V. Kumar, D. Cutter, and S. J. Barenkamp. 1998. Prevalence and distribution of the hmw and hia genes and the HMW and Hia adhesins among genetically diverse strains of nontypeable Haemophilus influenzae. Infect. Immun. 66:364-368.
40. Swords, W. E., P. A. Jones, and M. A. Apicella. 2003. The lipo-oligosaccharides of Haemophilus influenzae: an interesting array of characters. J. Endotoxin Res. 9:131-144.
41. Tiller, F. W. 1982. Biochemical differentiation of Haemophilus influenzae. Additional characterization of biotypes by carbohydrate fermentation patterns. Zentbl. Bakteriol. Mikrobiol. Hyg. A 253:236-246.
42. Tullius, M. V., N. J. Phillips, N. K. Scheffler, N. M. Samuels, J. R. Munson, Jr., E. J. Hansen, M. Stevens-Riley, A. A. Campagnari, and B. W. Gibson. 2002. The lbgAB gene cluster of Haemophilus ducreyi encodes a -1,4-galactosyltransferase and an -1,6-DD-heptosyltransferase involved in lipooligosaccharide biosynthesis. Infect. Immun. 70:2853-2861.
43. van Alphen, L., D. A. Caugant, B. Duim, M. O'Rourke, and L. D. Bowler. 1997. Differences in genetic diversity of nonencapsulated Haemophilus influenzae from various diseases. Microbiology 143:1423-1431.
44. van Schilfgaarde, M., P. van Ulsen, P. Eijk, M. Brand, M. Stam, J. Kouame, L. van Alphen, and J. Dankert. 2000. Characterization of adherence of nontypeable Haemophilus influenzae to human epithelial cells. Infect. Immun. 68:4658-4665.
45. Veal, W. L., and W. M. Shafer. 2003. Identification of a cell envelope protein (MtrF) involved in hydrophobic antimicrobial resistance in Neisseria gonorrhoeae. J. Antimicrob. Chemother. 51:27-37.
46. Weiser, J. N. 1993. Relationship between colony morphology and the life cycle of Haemophilus influenzae: the contribution of lipopolysaccharide phase variation to pathogenesis. J. Infect. Dis. 168:672-680.
47. Weiser, J. N., J. M. Love, and E. R. Moxon. 1989. The molecular mechanism of phase variation of H. influenzae lipopolysaccharide. Cell 59:657-665.
48. Williams, B. J., M. Golomb, T. Phillips, J. Brownlee, M. V. Olson, and A. L. Smith. 2002. Bacteriophage HP2 of Haemophilus influenzae. J. Bacteriol. 184:6893-6905.
49. Williams, B. J., G. Morlin, N. Valentine, and A. L. Smith. 2001. Serum resistance in an invasive, nontypeable Haemophilus influenzae strain. Infect. Immun. 69:695-705.(Alice L. Erwin, Kevin L. )
Division of Biological Sciences, University of Missouri, Columbia, Missouri 65211
Centro Nacional de Microbiología, Instituto de Salud Carlos III, Carretera de Pozuelo, 28220 Majadahonda, Madrid, Spain
Infectious Diseases and Clinical Microbiology, Nuffield Department of Clinical Laboratory Sciences, John Radcliffe Hospital, Oxford, United Kingdom
Atlanta Veterans Affairs Medical Center and Department of Medicine, Emory University School of Medicine, Decatur, Georgia 30033
Department of Pediatrics, University of North Carolina, Chapel Hill, North Carolina 27599
Department of Pediatrics, University of Arkansas for Medical Sciences, Arkansas Children's Hospital, Little Rock, Arkansas 72202
Institute for Infectious Diseases, University of Bern, Bern, Switzerland
Netherlands Vaccine Institute (NVI), NL-3720 AL Bilthoven, The Netherlands
ABSTRACT
The ability of unencapsulated (nontypeable) Haemophilus influenzae (NTHi) to cause systemic disease in healthy children has been recognized only in the past decade. To determine the extent of similarity among invasive nontypeable isolates, we compared strain R2866 with 16 additional NTHi isolates from blood and spinal fluid, 17 nasopharyngeal or throat isolates from healthy children, and 19 isolates from middle ear aspirates. The strains were evaluated for the presence of several genetic loci that affect bacterial surface structures and for biochemical reactions that are known to differ among H. influenzae strains. Eight strains, including four blood isolates, shared several properties with R2866: they were biotype V (indole and ornithine decarboxylase positive, urease negative), contained sequence from the adhesin gene hia, and lacked a genetic island flanked by the infA and ksgA genes. Multilocus sequence typing showed that most biotype V isolates belonged to the same phylogenetic cluster as strain R2866. When present, the infA-ksgA island contains lipopolysaccharide biosynthetic genes, either lic2B and lic2C or homologs of the losA and losB genes described for Haemophilus ducreyi. The island was found in most nasopharyngeal and otitis isolates but was absent from 40% of invasive isolates. Overall, the 33 hmw-negative isolates were much more likely than hmw-containing isolates to have tryptophanase, ornithine decarboxylase, or lysine decarboxylase activity or to contain the hif genes. We conclude (i) that invasive isolates are genetically and phenotypically diverse and (ii) that certain genetic loci of NTHi are frequently found in association among NTHi strains.
INTRODUCTION
Haemophilus influenzae is a small, fastidious gram-negative coccobacillus that colonizes the human nasopharynx, usually without causing symptoms. When disease occurs, it is usually limited to local infections of the respiratory tract such as otitis media in young children, bronchitis or sinusitis in adults, or pneumonia secondary to chronic pulmonary disease. Invasive disease such as bacteremia or meningitis is currently uncommon in the developed world and is nearly always associated with strains possessing polysaccharide capsules. The six serotypes of H. influenzae have capsules with different carbohydrate structures, with serotype b being the most common. Since the introduction of conjugate vaccines against the type b capsular polysaccharide, the incidence of life-threatening H. influenzae infection has decreased substantially. As expected, the vaccine has not affected the incidence of infections due to unencapsulated (nontypeable) H. influenzae (NTHi). It was once thought that invasive NTHi disease occurred only in children with immunologic or anatomical defects that predispose them to bacterial infections. More recently it has become apparent that NTHi can cause bacteremia and meningitis in otherwise healthy children. One of the first well-documented cases of invasive NTHi was reported in 1996 (33), and since then there have been several publications describing additional cases (5, 6, 8, 34). It is not known whether NTHi isolates from invasive infections possess virulence determinants not shared with isolates associated with otitis media or isolates asymptomatically colonizing the nasopharynx. It is important to determine whether NTHi strains isolated from invasive infections represent distinct lineages or have novel virulence genes not shared with commensals or isolates from localized infections.
The genetic diversity of encapsulated H. influenzae and NTHi has been studied using ribotyping, restriction fragment length polymorphism (RFLP), multilocus enzyme electrophoresis, and multilocus sequence typing (MLST) (4, 26, 32, 43). Phylogenetic trees generated using these methods have shown that whereas encapsulated isolates cluster by serotype, NTHi isolates are much more diverse, with less evidence of clonality (26, 32). The extent to which disease isolates of NTHi are phylogenetically similar has not been studied extensively. van Alphen et al. reported that isolates from acute disease (otitis media or meningitis) are less phylogenetically diverse than isolates from chronic infection (cystic fibrosis or chronic bronchitis) (43). Classically, H. influenzae strains were grouped into "biotypes" by biochemical assays. Kilian (23) described a system for biotyping H. influenzae based on the presence or absence of three enzymatic activities: urease, ornithine decarboxylase, and the production of indole from tryptophan. Biotyping was useful because of weak correlations with pathogenicity. Serotype b strains, particularly those isolated from systemic disease, usually possess all three activities, making them biotype I; conjunctival isolates are usually biotype III and urogenital isolates type IV. Mucosal isolates, whether from healthy children or associated with disease, are more diverse in biotype (23). Indole production, associated with a tryptophanase (tna) gene cluster and with several biotypes, is more frequent in isolates from mucosal and invasive disease than in commensals or in conjunctivitis (25).
Respiratory tract isolates differ in the fermentation of fructose, maltose, and xylose (22, 41). The API 20E system, developed for testing multiple biochemical reactions in clinical isolates of Enterobacteriaceae, has shown several additional activities to be present in some H. influenzae strains but not in others, including arginine dihydrolase and lysine decarboxylase activities and arabinose fermentation (18). For NTHi, none of these biochemical activities has been shown to be strongly associated with a specific disease or with asymptomatic carriage, though it has been noted that lower respiratory tract isolates from patients with cystic fibrosis are more likely to decarboxylate lysine than are other isolates (17).
Surface molecules of both NTHi strains and encapsulated H. influenzae strains have been shown to be heterogeneous in a variety of ways that may affect pathogenesis. The major outer membrane proteins known as P1, P2, and P5 are particularly diverse in sequence (2). Several genes involved in lipopolysaccharide biosynthesis are variably present in H. influenzae strains, and some of these are phase variable. The resulting heterogeneity of lipopolysaccharide structure affects susceptibility to complement-mediated serum killing and the interaction of bacteria with host cells (40). H. influenzae isolates also differ in protein adhesins: most contain genes for one of the adhesins known as Hia and HMW, but not both, and some strains are also able to synthesize fimbriae (38). A recent study reported that otitis isolates are more likely than throat isolates to contain hmw genes, while throat isolates were more likely than isolates from blood or the middle ear to hybridize with a probe for the hifBC genes (required for synthesis of fimbriae) (10).
We undertook these studies to determine whether isolates of NTHi associated with invasive disease differ phenotypically or genetically from isolates cultured from healthy children or associated with acute otitis media. We previously reported that the invasive NTHi strain Int1 (later R2866) differs from most NTHi strains in being unusually resistant to the bactericidal activity of normal adult human serum (49). We have also found that strain R2866 contains several genetic loci that are absent from the sequenced strain Rd KW20 and from several nontypeable strains. These include the gene for an autotransporter, lav, located between holB and tmk (9), the lysogenic bacteriophage HP2 (48), the tna cluster (25), and a 53-kb plasmid similar to other large integrative plasmids of Haemophilus spp. (28). In the present work a collection of 17 NTHi isolates associated with invasive disease were compared with isolates from healthy children or associated with acute otitis media in order to determine whether invasive isolates are genetically or phylogenetically distinct. We assayed the susceptibility of each strain to complement-mediated killing by pooled normal human serum and tested each in a panel of biochemical assays. In addition, we used PCR to screen for the presence of genes for HP2, Lav, fimbriae, and the adhesins Hia and HMW, the 53-kb plasmid, and the lipopolysaccharide biosynthetic locus flanked by infA and ksgA. Invasive strains were genetically heterogeneous and did not differ significantly in the range of serum sensitivity from nasopharyngeal commensals or otitis media isolates. However, several genetic characteristics are not assorted randomly but are strongly associated with one another. In particular, several biochemical and genetic traits were found primarily in strains lacking genes for the hmw adhesin. Within the hmw-negative group we identified a cluster of eight isolates that resembled R2866 in being biotype V and hia positive, lacking an infA-ksgA island, and having the urease locus replaced by a homolog of the gonococcal membrane protein mtrF. These strains were shown by MLST to be members of a phylogenetic cluster.
MATERIALS AND METHODS
Bacterial strains and growth. H. influenzae strains are listed in Table 1. We studied three types of isolates: nasopharyngeal or throat isolates from healthy children, strains cultured from middle ear aspirates from children with otitis media, and blood or cerebrospinal fluid isolates obtained from children with invasive disease. For simplicity, isolates are referred to below as "throat," "otitis," or "invasive." Strains were isolated between 1995 and 2004, at different locations within the United States, Canada, and Europe. For invasive isolates, we limited the study to isolates from children who were considered by the treating physician to have normal anatomy and immune function. Neonates were excluded. Otitis isolates included the recently sequenced strains 86-028NP (31) and R2846 (strain 12). Invasive isolates included R2866 (Int1), which has also been sequenced recently. Bacteria were cultivated at 37°C on chocolate agar supplemented with 1% IsovitaleX (Becton Dickinson and Co., Sparks, MD) or in Difco brain heart infusion broth (Becton Dickinson and Co.) supplemented (sBHI) with hemin (10 μg/ml) and -NAD (10 μg/ml) and with agar for solid sBHI plates. The requirement of all strains for X and V factors was confirmed by disk diffusion. All strains were negative for bexA by PCR, indicating the inability to express capsular polysaccharide. The serotype b strain Eagan was used as a positive control for bexA PCR.
Biochemical assays. Bacteria were suspended at 108 CFU/ml in 0.9% NaCl containing hemin (10 μg/ml) and -NAD (10 μg/ml) and were used to inoculate API 20E biochemical test strips purchased from bioMerieux (St. Louis, MO) (19). Enzyme activities and fermentation reactions were recorded after 24 h of incubation at 37°C in air. Indole, ornithine decarboxylase, and urease activities were used for biotyping (23).
Screen for genetic markers by PCR. Genomic DNA was prepared using the DNeasy tissue kit (QIAGEN, Inc., Valencia, CA). Primers (Table 2) were synthesized by Integrated DNA Technologies, Coralville, IA. For lav, we amplified across the holB-tmk junction (HI0455 to HI0456 in Rd KW20). Positive strains yielded a product of 2 to 2.2 kb. Partial sequencing confirmed that all were related to the published sequence of lav. The remaining isolates did not yield an amplicon or else produced a 150-bp product consistent with no inserted gene. For detection of bacteriophage HP2, we used PCR to amplify regions of rep (common to both phage HP1 and HP2) and orf10 (specific for HP2). For the plasmid, PCR with the primers listed as topoF and topoR gave a product of about 450 bp, and primers AF and AR gave a product of about 500 bp. We used primers designed to amplify conserved regions of the adhesin genes hmwA and hia (36, 44). PCR with the hmw primers gave a product of approximately 1.2 kb. Occasional faint products were sequenced and were scored negative if the sequences were not related to hmwA. hia products were usually 3 to 4 kb but ranged from approximately 0.75 kb to 10 kb. The basis of this size difference was not evaluated. Ten of the hia amplicons were sequenced and found to be similar to the published hia sequence. Of four strains yielding hia products of only 0.75 kb, two were sequenced and found to be homologous to the 3' end of the hia gene. All of these hia-related sequences were scored as positive for hia. The presence and size of the hif locus, required for synthesis of fimbriae, was evaluated by amplification across the purE-pepN junction. Isolates with a complete hif locus produced a product of approximately 7 kb. These PCR fragments were sequenced using the purE primer to detect the presence of hicA and hicB in addition to the hif genes. As seen previously (27), the remaining isolates contained either a shorter purE-pepN insert, reflected by a product of approximately 0.5 to 1 kb (presumed to include hicAB), or no insert, reflected by a PCR product of approximately 250 bp. The genetic island flanked by infA and ksgA was detected by PCR across the infA-ksgA junction. The contents of the 2.2-kb island were determined either by sequencing or by PCR using primers infA and losB-R to detect losAB and primers lic2B-F and lic2BA to detect lic2BC. Routine PCR was carried out using the Biolase DNA polymerase (Bioline USA, Randolph, MA), with the following amplification protocol: reaction tubes were incubated at 94°C for 2 min, then for 36 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min per kb of predicted product. The Expand Long Template PCR system (Roche Diagnostics Company, Indianapolis, IN) was used as directed for amplification of the hif and hia loci. Following electrophoresis, ethidium bromide-stained agarose gels were visualized and photographed using the Fluorochem 8900 digital imaging system (Alpha Innotech Corporation, San Leandro, CA). PCR products to be sequenced were processed using the Qiaquick PCR purification kit (QIAGEN, Inc.) and submitted to the core sequencing facility at Seattle Biomedical Research Institute. Sequences were aligned using the BioEdit Sequence Alignment Editor (Isis Pharmaceuticals, Inc., Carlsbad, CA).
MLST. PCR for the housekeeping genes adk, atpG, adk, atpG, frdB, fucK, mdh, pgi, and recA was carried out using the primers and methods described in reference 26. Sequences were submitted to the MLST website (www.mlst.net) for allele and sequence type (ST) assignment.
Serum bactericidal activity. Log-phase bacteria (2,000 CFU/ml) were incubated for 30 min at 37°C with pooled normal human serum diluted in 10 mM phosphate-buffered saline containing 4 mM KCl and 0.1% gelatin and were then plated to determine bacterial survival. The concentration of serum that killed 50% of bacteria was calculated using XLfit 4.1 (ID Business Solutions, Guildford, United Kingdom) and is referred to as the IC50 of the serum for that strain.
Statistics. Groups were compared using the Fisher exact probability test. Calculations were performed online using VassarStats: Web Site for Statistical Computation (http://faculty.vassar.edu/lowry/VassarStats.html).
Sequence data. Sequences for the genomes of H. influenzae strains Rd KW20, R2846, R2866, and 86-028NP and of Neisseria gonorrhoeae FA1090 and Neisseria meningitidis strains MC58 and Z2491 were accessed through the Microbial Genomes pages at the website of the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/genomes/lproks.cgi). The sequences of the mtrF (strain R2866) and losAB (strain R2846) loci were submitted to NCBI and were assigned accession numbers DQ007025 and DQ007026, respectively.
RESULTS
Biotyping and other biochemical characterization. For H. influenzae type b, invasive isolates nearly always belong to biotype I (positive for all three of the biotyping reactions), with biotype II (negative for ornithine decarboxylase only) seen occasionally (14). In contrast, of the invasive NTHi isolates studied here, only one-third belonged to biotype I and about one-fourth belonged to biotype V (negative for urease only) (Table 3). We noted heterogeneity among strains in several other biochemical reactions. Arginine dihydrolase and arabinose fermentation were seen predominantly for throat isolates, while lysine decarboxylase and growth on citrate were seen less often for middle ear isolates than for the other two groups (Table 3). In most cases these correlations were not statistically significant.
Correlation of the hmwA gene with biochemical reactions and genes for other surface molecules. We evaluated strains for the presence of genes encoding the HMW adhesins by PCR amplification of a conserved region of hmw1A and hmw2A. The results correlated strongly with the biochemical reactions discussed above and with the presence of genes encoding other surface molecules (Table 3). hmw-negative isolates were much more likely than hmw-positive isolates to be positive for indole, ornithine decarboxylase, lysine decarboxylase, and arginine dihydrolase activities, to be negative for urease activity, and to ferment arabinose and grow on citrate. hmw-negative isolates were nearly always positive for hia-related sequences, as reported previously (39). One isolate, R3579, contained sequences for both hmw and hia. This result was confirmed by sequencing the products and by carrying out PCR on five separate colonies. Each of the five colonies yielded both a hia product (approximately 3 kb) and a hmw product, indicating that the culture was not a mixture of a hia-positive and a hmw-positive strain. We used PCR to determine the presence and size of the fimbrial gene cluster flanked by purE and pepN. We found that 11 isolates contained a 7-kb purE-pepN insert consistent with a complete fimbrial biosynthetic locus; 10 of these isolates were hmw negative. Of 15 isolates lacking a purE-pepN insert, 14 were also hmw negative. Of the 20 hmw-positive isolates, 18 had a purE-pepN insert of 0.8 to 1 kb. Such inserts have been reported to contain two genes of unknown function, hicA and hicB. This correlation between hmw and the fimbrial gene cluster has not been reported previously.
Lipopolysaccharide biosynthetic island flanked by infA and ksgA. The lipopolysaccharide biosynthetic genes now known as lic2B and lic2C were initially described for a type b strain, RM7004 (originally strain 760705 from the van Alphen laboratory, isolated from cerebrospinal fluid). In this strain they are flanked by infA and ksgA (homologs of HI0548 and HI0549 in Rd KW20) (16, 20). Comparison of the infA-ksgA region in the available H. influenzae genome sequences showed that in strains Rd KW20 and R2866, the infA and ksgA genes are adjacent. Strain 86-028NP contains a single gene, lic2C, separating infA and ksgA. In strain R2846, infA and ksgA are separated by two genes that are unrelated to lic2B and lic2C. These genes are strongly homologous to losA and losB (also known as lbgB and lbgA), characterized in Haemophilus ducreyi as encoding -1,4-galactosyltransferase and D-glycero-D-manno-heptosyltransferase, respectively (13, 42). The infA-ksgA junction thus appears to be a site of substantial genetic heterogeneity, with at least four complements of genes seen in different isolates (Fig. 1).
We used PCR across the infA-ksgA junction to evaluate the diversity of this region in the same 53 isolates that were used for the biochemical studies described above. Most of the throat and mucosal-disease isolates in our study contained a genetic island of 2.2 kb. Of 12 isolates lacking the island, 6 were invasive, 4 were otitis isolates, and only 2 were throat isolates. Partial sequencing of the amplified ksg-infA regions showed that the distribution of losAB and lic2BC-related sequences differed among the three groups: only one invasive isolate, but about one-quarter of throat and otitis isolates, contained losAB-related sequences. Most of the remaining strains contained lic2B and lic2C, but four contained only lic2C (Table 4).
These data suggest that invasive isolates of NTHi may be less likely than other NTHi isolates to contain an island flanked by infA and ksgA, and also less likely to contain an island consisting of losAB. However, this is based on a fairly small number of isolates. We used PCR to investigate the infA-ksgA regions of 19 additional invasive isolates of NTHi received from laboratories in the United States, Canada, and Switzerland. Nine of these isolates lacked the infA-ksgA island. Thus, of the 72 isolates studied, we identified 21 isolates that lack the island. Of these, 15 were invasive (P = 0.04). The 19 additional invasive isolates also included 7 with sequences related to lic2BC and 3 with sequences related to losAB, consistent with the earlier observation that losAB is relatively rare among invasive isolates.
Identification of group of closely related isolates. The presence or absence of the infA-ksgA island and its content correlate with several of the genetic and biochemical features known to be variably present among H. influenzae strains (Table 4). Isolates lacking the infA-ksgA island were more likely than other isolates to contain lav or the HP2 bacteriophage. Two-thirds of the isolates lacking this island were urease negative but positive for indole and ornithine decarboxylase, making them members of the relatively rare biotype V (23). All biotype V strains were hia positive (P = 0.007). The similarities of the biotype V isolates led us to look carefully for differences among them (Table 5). Although these isolates have several features in common, they are heterogeneous with regard to the presence of lysine decarboxylase activity, production of nitrite from nitrate, arabinose fermentation, and the presence of the lav and HP2 genes. It thus seems unlikely that they represent a recently derived clone. We used MLST to determine the STs for 11 strains that are biotype V, hmw negative, and negative for the infA-ksgA island and for 13 strains of biotypes I, II, and III. Eight of the biotype V strains formed a more distinct cluster than any of the other strains, while one belonged to a neighboring lineage; the remaining two biotype V strains were closely related to each other but not to other strains (Fig. 2).
Replacement of the urease locus by an mtrF homolog. The strong correlation between urease activity and the presence of an infA-ksgA island (P < 0.0005) may be due to genetic linkage, since these two loci are separated by only 4 kb. A search of the available H. influenzae genome sequences revealed that strains Rd KW20, R2846, and 86-028NP each contain a cluster of seven genes (designated HI0535 to HI0541 in Rd KW20) with homology to the well-characterized urease genes of Klebsiella aerogenes (30). In each case this locus is flanked by homologs of aspA and groES. In strain R2866, which is urease negative, the seven-gene cluster is replaced by a single gene with homology to the gonococcal gene mtrF (Fig. 1). Using PCR primers that hybridized to the aspA and groES genes, we were able to amplify the expected 9-kb product from genomic DNA of each of four urease-positive strains. Partial sequencing of the PCR products showed that each was similar to the urease locus of strain Rd KW20. The same PCR primers yielded a 4-kb product from each of nine urease-negative strains, consistent with the sequence of R2866. Partial sequencing of these 4-kb products showed that all were very similar to that of R2866.
RFLP analysis of the hif locus. In previous work (T. Mhlanga-Mutangadura and M. Golomb, unpublished) on the heterogeneity of the hif locus, we had identified three NTHi isolates, R3101, R3151, and R3157, that had hifA sequences nearly identical to that of R2866 and in which the 7-kb hif locus had a PstI RFLP pattern similar to that of R2866. We now extended that work to the strains found in this study to have a 7-kb hif locus. Consistent with previous work, several patterns were identified. R3265, a throat isolate, had the same RFLP pattern as R2866. All of the isolates with this pattern belonged to biotype V and lacked the infA-ksgA island.
Serum resistance. Our prototype invasive nontypeable strain, R2866 (Int1), had previously been shown to survive in 40% normal human serum much longer than the laboratory strain Rd KW20 or the nontypeable strain U11 (49). We hypothesized that the in vitro serum resistance of R2866 reflected a virulence trait that might be shared with other invasive NTHi isolates. In the present study we used a more quantitative assay to evaluate the serum resistance of the 53 clinical isolates described above. We found that the range of IC50s for clinical NTHi isolates was nearly as wide as the difference between encapsulated and unencapsulated strains and that R2866 was one of the most resistant strains in our collection. However, the 16 additional invasive isolates that we tested were not overall more serum resistant than otitis isolates and were only slightly more serum resistant than throat isolates. The IC50 range for the invasive isolates was broad, from 1.9% to 13.3%, and this range was similar to that for the other two groups. It is clear that in vitro serum resistance is neither common to all invasive isolates nor a specific characteristic of this group (Fig. 3).
DISCUSSION
While it is well known that NTHi isolates are more heterogeneous than type b isolates, the extent to which the heterogeneous characteristics are correlated with each other or with the anatomic site of isolation is not well understood. We had previously identified genes encoding the autotransporter Lav and bacteriophage HP2 in the invasive isolate R2866 and hypothesized that they might be involved in the unusual virulence of this strain (9, 48). In this study we found that the lav gene is indeed rare in throat isolates but is present in many otitis isolates and in only about one-fourth of invasive isolates. Only 4 of the 53 isolates studied in this work contained HP2; 3 of these are invasive. R2866 also contains a 53-kb region that appears to be an integrated plasmid similar to those described for other Haemophilus strains (28). We identified sequences characteristic of the plasmid in 11 isolates, only 1 of which was a throat isolate. None of the genetic loci or biochemical reactions we studied were found to be consistently present in invasive isolates. A genetic island flanked by infA and ksgA was identified in most throat and otitis isolates but was absent in 40% of invasive isolates. When present, this island contains lipopolysaccharide biosynthetic genes, as discussed below.
The ease with which H. influenzae strains exchange DNA and the observed genetic heterogeneity among NTHi isolates might have suggested that variable genes would occur randomly. A high rate of recombination in otitis media isolates has been reported (7). However, we found that certain genetic loci occurred together much more often than would occur by chance. Many of the characteristics we examined were significantly associated with the presence of sequences related to hmw or hia genes. Because we detected these genes by PCR amplification of an internal portion of each gene, we cannot be sure that the strains we scored positive for an adhesin actually contain the complete genetic locus required to express HMW or Hia adhesins. Indeed, some strains appeared to contained only a fragment of the hia gene. It is also possible that some of the strains we scored negative contained gene variants that differed slightly at one of the primer binding sites. St. Geme et al. reported that nearly all H. influenzae strains contain genes encoding either the Hia adhesin or the HMW adhesins (39). hia is an allele of the hsf gene found in encapsulated strains. This led to the hypothesis that hia-containing strains may have evolved from an encapsulated ancestor. In support of this idea, the insertion element IS1016, which is found in the cap locus of encapsulated strains, was found to hybridize to genomic DNA of 6 hia-containing NTHi isolates out of 9 isolates tested but not to hybridize to any of the 47 hmw-containing isolates that were tested (39). Our observation that several genetic and biochemical traits are correlated with the presence of a sequence related to hmw or hia is consistent with the idea that these genes are found in two distinct lineages. Indole, ornithine decarboxylase, and lysine decarboxylase were all significantly more prevalent among hia-containing isolates than among hmw-containing isolates. The hmw-containing isolates in our study were uniformly urease positive, while nearly one-third of the hia-containing isolates were urease negative.
Within the hmw-negative, hia-positive isolates, we identified a group of 11 biotype V isolates that shared several characteristics of the invasive strain R2866 (Table 5). All the biotype V strains except one lacked the genetic island between ksgA and infA. Some of the isolates lacking an infA-ksgA island contained a hif locus with a characteristic RFLP pattern. Isolates lacking an infA-ksgA island were also more likely than other isolates to contain lav genes. Bacteriophage HP2 was found only in biotype V strains.
With the exception of the hmw-hia dichotomy (39), associations among variable genetic loci of H. influenzae have not been previously noted. Because natural transformation and other means of genetic exchange readily disrupt linkages, gene associations must be explained by a recent shared lineage or by shared adaptive strategies or requirements for virulence. The urease locus and the infA-ksgA locus are linked within 4 kb and might have persisted in linkage disequilibrium. However, the associations with the hia gene and the hif RFLP pattern cannot be explained in this way. It is possible that the isolates lacking an infA-ksgA island are closely related to each other, although they were derived from different parts of the United States as well as from Canada and Europe. As noted above, the isolates in this group were more likely to have been cultured from blood than from ear aspirates or nasopharyngeal samples. This suggests that they may share virulence determinants. Phylogenetic analysis by MLST showed that 9 of 11 biotype V isolates formed a single cluster, while 2 other biotype V isolates were closely related to each other but not to the cluster of 9. Typing of 13 strains of other biotypes showed some tendency for strains to cluster by biotype and the by presence of an hmw-related sequence. Further evaluation of the correlation between MLST and other characteristics of NTHi is ongoing.
We found that in the urease-negative isolates we studied, the urease locus was invariably replaced by a gene with homology to the mtrF gene found in N. gonorrhoeae and N. meningitidis. In gonococci, mtrF encodes a predicted membrane protein that has been reported to be involved in resistance to detergents and hydrophobic antibiotics, in conjunction with the resistance-nodulation-cell division (RND) pump MtrCDE (45), which does not appear to be present in any of the H. influenzae strains that have been sequenced. A less closely related homolog, abgT, is thought to encode a transporter that allows an Escherichia coli mutant that is defective in p-aminobenzoate synthesis to utilize p-aminobenzoyl-glutamate (21). It thus seems likely that the H. influenzae MtrF homolog is involved in transport, but the substrate cannot be determined without experimental data.
We found that when the infA-ksgA region contains an insert, the locus may consist of lic2BC, as previously seen for type b isolates (15), of lic2C only, or of two genes with homology to losA and losB of H. ducreyi. The prevalence of lic2B and lic2C has been studied previously, though not in invasive isolates. Hood et al. reported that 13 of 27 NTHi strains studied contained lic2C (20). Pettigrew et al. examined 90 throat and 48 otitis isolates and found lic2B in one-half of the otitis isolates but only 14% of the throat isolates (35). In contrast, we found for throat, otitis, and invasive isolates that about half in each group contained lic2BC. However, it is possible that our small sample of throat isolates (17 isolates) had an atypical distribution of genes at this locus. The losAB genes have not been studied previously in H. influenzae. We found that the infA-ksgA island contained losAB in about one-third of our throat and otitis isolates but in very few invasive isolates.
Without further study, it is not possible to determine whether the variation we see in the structure of the infA-ksgA region has a direct role in H. influenzae biology. Certainly the lic2BC and losAB genes are likely to affect the structure of lipopolysaccharide and thereby the interaction of bacteria with the host. It is also possible that these genes are markers for other characteristics that collectively affect the ability of a given strain to colonize or cause disease. We noted that the genome sequences of both R2846 and R2866 contain homologs of losAB at sites distant from the infA-ksgA region. In R2846, the second losA homolog contains a frameshift and appears unlikely to encode an active product.
NTHi strains isolated from the blood of ill children were as likely to be sensitive to serum as throat and otitis isolates. One explanation for this unexpected finding is that our laboratory stocks differ genetically from the bacterial population in the patients during infection. H. influenzae is subject to high-frequency, reversible gain or loss of phenotypes such as hemagglutination (reflecting piliation), colony morphology, or reactivity of lipopolysaccharide with monoclonal antibodies (11, 24, 46). The molecular basis of this phase variation is variation in the length of tandem repeat regions, often within the coding region of genes (47). A growing culture of any H. influenzae strain thus contains a large number of variants that differ from each other in the expression of one or more genes. A change in environmental conditions may result in selection of a different population of variants, and a culture that has been derived from a single colony (as is likely to happen during isolation in a clinical laboratory) and then propagated in the lab may contain a mixture of variants very different from that in the patient's blood. Although R2866, the first invasive NTHi strain studied, has retained serum resistance in the lab, this may be the result of a fortuitous subculture of a resistant colony. The range of serum resistance observed for other clinical isolates may reflect true differences among these isolates but may also indicate that for some isolates we are working with serum-sensitive variants and for other isolates our laboratory stocks happen to consist largely of serum-resistant variants.
Overall, our data indicate that NTHi isolates from invasive disease are not homogeneous: they differ in genotype and in resistance to human serum. It is possible that invasiveness can result from more than one combination of biologic capabilities and that different invasive strains have different sets of genes mediating these functions. It is also possible that all invasive isolates share a gene or group of genes that has not yet been discovered. Several of the biochemical and genetic traits we studied were correlated with each other and particularly with the hia and hmw genes, supporting the idea that NTHi strains may cluster into distinct lineages, although the lines may be blurred by genetic exchange. We have recently determined the complete genome sequence of the invasive strain R2866 and an otitis strain, R2846 (strain 12). Each of these strains contains 30 to 40 genetic loci not present in the previously sequenced strain, Rd KW20. Analysis of these novel loci will add to our understanding of the genetic relationships among NTHi isolates and may identify genetic determinants characteristic of strains associated with invasive disease and with otitis media. This will aid in the understanding of the virulence mechanisms required for nasopharyngeal bacteria to cause mucosal or systemic disease.
ADDENDUM IN PROOF
Since acceptance of this paper, we discovered that the bexA primers listed in Table 2 are able to amplify bexA from serotype b, c and d strains but do not reliably amplify bexA from serotype a or e strains. On testing the strains listed in Table 1 for bexA using the primers described by Falla et al. (J. Clin Microbiol. 32: 2382-2386, 1994), strains R3049, R3172, R3173, R3368, and R3595 were found to be positive. Strain R3368 was biotype I and negative for lysine decarboxylase, arginine dihydrolase, citrate, and arabinose. It was positive for hia and losAB sequences and negative for hmw, lav, HP2, plasmid, hif, and hic sequences. Strains R3049 and R3173 contain the lic2BC genes, and R3172 and R3595 contain losAB. Elimination of R3368-derived data from Table 3 increased each of the reported P values slightly, with each P value remaining <0.02. Eliminating data on the remaining strains from the section entitled "Lipopolysaccharide biosynthetic island flanked by infA and ksgA" did not alter the conclusions.
ACKNOWLEDGMENTS
This work was supported in part by grants AI 44002 and AI 46512 from the National Institute of Allergy and Infectious Disease to A.L.S. and by a grant from the M. J. Murdock Charitable Trust to the Seattle Biomedical Research Institute.
We thank Theodore White and Bruce Torian for critical review of the manuscript. We are grateful for the assistance of David Scheifele (University of British Columbia), Carol Shaw (British Columbia Centre for Disease Control), Stephen Barenkamp (Washington University), Robert Munson and Lauren Bakaletz (Ohio State University), Janet Gilsdorf (University of Michigan), and Bruce Green)Wyeth Vaccines) in obtaining clinical isolates. This publication made use of the Multi Locus Sequence Typing website (http://www.mlst.net/) at Imperial College London, developed by Man-Suen Chan and David Aanensen and funded by the Wellcome Trust. Daniel Godoy, curator of the H. influenzae MLST database, provided assistance with allele and sequence type assignments.
Present address: Department of Veterinary Pathobiology, University of Missouri, Columbia, MO 65211.
REFERENCES
1. Bakaletz, L. O., B. M. Tallan, W. J. Andrzejewski, T. F. DeMaria, and D. J. Lim. 1989. Immunological responsiveness of chinchillas to outer membrane and isolated fimbrial proteins of nontypeable Haemophilus influenzae. Infect. Immun. 57:3226-3229.
2. Barenkamp, S. J. 2004. Rationale and prospects for a nontypable Haemophilus influenzae vaccine. Pediatr. Infect. Dis. J. 23:461-462.
3. Barenkamp, S. J., and E. Leininger. 1992. Cloning, expression, and DNA sequence analysis of genes encoding nontypeable Haemophilus influenzae high-molecular-weight surface-exposed proteins related to filamentous hemagglutinin of Bordetella pertussis. Infect. Immun. 60:1302-1313.
4. Bruce, K. D., and J. Z. Jordens. 1991. Characterization of noncapsulate Haemophilus influenzae by whole-cell polypeptide profiles, restriction endonuclease analysis, and rRNA gene restriction patterns. J. Clin. Microbiol. 29:291-296.
5. Campos, J., M. Hernando, F. Roman, M. Perez-Vazquez, B. Aracil, J. Oteo, E. Lazaro, and F. de Abajo. 2004. Analysis of invasive Haemophilus influenzae infections after extensive vaccination against H. influenzae type b. J. Clin. Microbiol. 42:524-529.
6. Cerquetti, M., M. L. Ciofi degli Atti, G. Renna, A. E. Tozzi, M. L. Garlaschi, P. Mastrantonio, and the Hi Study Group. 2000. Characterization of non-type B Haemophilus influenzae strains isolated from patients with invasive disease. J. Clin. Microbiol. 38:4649-4652.
7. Cody, A. J., D. Field, E. J. Feil, S. Stringer, M. E. Deadman, A. G. Tsolaki, B. Gratz, V. Bouchet, R. Goldstein, D. W. Hood, and E. R. Moxon. 2003. High rates of recombination in otitis media isolates of non-typeable Haemophilus influenzae. Infect. Genet. Evol. 3:57-66.
8. Cuthill, S. L., M. M. Farley, and L. G. Donowitz. 1999. Nontypable Haemophilus influenzae meningitis. Pediatr. Infect. Dis. J. 18:660-662.
9. Davis, J., A. L. Smith, W. R. Hughes, and M. Golomb. 2001. Evolution of an autotransporter: domain shuffling and lateral transfer from pathogenic Haemophilus to Neisseria. J. Bacteriol. 183:4626-4635.
10. Ecevit, I. Z., K. W. McCrea, M. M. Pettigrew, A. Sen, C. F. Marrs, and J. R. Gilsdorf. 2004. Prevalence of the hifBC, hmw1A, hmw2A, hmwC, and hia genes in Haemophilus influenzae isolates. J. Clin. Microbiol. 42:3065-3072.
11. Farley, M. M., D. S. Stephens, S. L. Kaplan, and E. O. Mason, Jr. 1990. Pilus- and non-pilus-mediated interactions of Haemophilus influenzae type b with human erythrocytes and human nasopharyngeal mucosa. J. Infect. Dis. 161:274-280.
12. Geluk, F., P. P. Eijk, S. M. van Ham, H. M. Jansen, and L. van Alphen. 1998. The fimbria gene cluster of nonencapsulated Haemophilus influenzae. Infect. Immun. 66:406-417.
13. Gibson, B. W., A. A. Campagnari, W. Melaugh, N. J. Phillips, M. A. Apicella, S. Grass, J. Wang, K. L. Palmer, and R. S. Munson, Jr. 1997. Characterization of a transposon Tn916-generated mutant of Haemophilus ducreyi 35000 defective in lipooligosaccharide biosynthesis. J. Bacteriol. 179:5062-5071.
14. Harper, J. J., and M. H. Tilse. 1991. Biotypes of Haemophilus influenzae that are associated with noninvasive infections. J. Clin. Microbiol. 29:2539-2542.
15. High, N. J., M. E. Deadman, and E. R. Moxon. 1993. The role of a repetitive DNA motif (5'-CAAT-3') in the variable expression of the Haemophilus influenzae lipopolysaccharide epitope Gal(1-4)Gal. Mol. Microbiol. 9:1275-1282.
16. High, N. J., M. P. Jennings, and E. R. Moxon. 1996. Tandem repeats of the tetramer 5'-CAAT-3' present in lic2A are required for phase variation but not lipopolysaccharide biosynthesis in Haemophilus influenzae. Mol. Microbiol. 20:165-174.
17. Hoiby, N., and M. Kilian. 1976. Haemophilus from the lower respiratory tract of patients with cystic fibrosis. Scand. J. Respir. Dis. 57:103-107.
18. Hollander, R. 1981. Biochemical characterization of Haemophilus strains by using the API 20E and API 50e test system. Zentbl. Bakteriol. Mikrobiol. Hyg. A 250:322-329. (In German.)
19. Holmes, R. L., L. M. DeFranco, and M. Otto. 1982. Novel method of biotyping Haemophilus influenzae that uses API 20E. J. Clin. Microbiol. 15:1150-1152.
20. Hood, D. W., M. E. Deadman, A. D. Cox, K. Makepeace, A. Martin, J. C. Richards, and E. R. Moxon. 2004. Three genes, lgtF, lic2C and lpsA, have a primary role in determining the pattern of oligosaccharide extension from the inner core of Haemophilus influenzae LPS. Microbiology 150:2089-2097.
21. Hussein, M. J., J. M. Green, and B. P. Nichols. 1998. Characterization of mutations that allow p-aminobenzoyl-glutamate utilization by Escherichia coli. J. Bacteriol. 180:6260-6268.
22. Kawakami, Y., Y. Okimura, and M. Kanai. 1982. Occurrence and biochemical properties of Haemophilus species in pharyngeal flora of healthy individuals. Microbiol. Immunol. 26:629-633.
23. Kilian, M. 1976. A taxonomic study of the genus Haemophilus, with the proposal of a new species. J. Gen. Microbiol. 93:9-62.
24. Kimura, A., and E. J. Hansen. 1986. Antigenic and phenotypic variations of Haemophilus influenzae type b lipopolysaccharide and their relationship to virulence. Infect. Immun. 51:69-79.
25. Martin, K., G. Morlin, A. Smith, A. Nordyke, A. Eisenstark, and M. Golomb. 1998. The tryptophanase gene cluster of Haemophilus influenzae type b: evidence for horizontal gene transfer. J. Bacteriol. 180:107-118.
26. Meats, E., E. J. Feil, S. Stringer, A. J. Cody, R. Goldstein, J. S. Kroll, T. Popovic, and B. G. Spratt. 2003. Characterization of encapsulated and noncapsulated Haemophilus influenzae and determination of phylogenetic relationships by multilocus sequence typing. J. Clin. Microbiol. 41:1623-1636.
27. Mhlanga-Mutangadura, T., G. Morlin, A. L. Smith, A. Eisenstark, and M. Golomb. 1998. Evolution of the major pilus gene cluster of Haemophilus influenzae. J. Bacteriol. 180:4693-4703.
28. Mohd-Zain, Z., S. L. Turner, A. M. Cerdeno-Tarraga, A. K. Lilley, T. J. Inzana, A. J. Duncan, R. M. Harding, D. W. Hood, T. E. Peto, and D. W. Crook. 2004. Transferable antibiotic resistance elements in Haemophilus influenzae share a common evolutionary origin with a diverse family of syntenic genomic islands. J. Bacteriol. 186:8114-8122.
29. Mühlemann, K., M. Balz, S. Aebi, and K. Schopfer. 1996. Molecular characteristics of Haemophilus influenzae causing invasive disease during the period of vaccination in Switzerland: analysis of strains isolated between 1986 and 1993. J. Clin. Microbiol. 34:560-563.
30. Mulrooney, S. B., and R. P. Hausinger. 1990. Sequence of the Klebsiella aerogenes urease genes and evidence for accessory proteins facilitating nickel incorporation. J. Bacteriol. 172:5837-5843.
31. Munson, R. S., Jr., A. Harrison, A. Gillaspy, W. C. Ray, M. Carson, D. Armbruster, J. Gipson, M. Gipson, L. Johnson, L. Lewis, D. W. Dyer, and L. O. Bakaletz. 2004. Partial analysis of the genomes of two nontypeable Haemophilus influenzae otitis media isolates. Infect. Immun. 72:3002-3010.
32. Musser, J. M., S. J. Barenkamp, D. M. Granoff, and R. K. Selander. 1986. Genetic relationships of serologically nontypeable and serotype b strains of Haemophilus influenzae. Infect. Immun. 52:183-191.
33. Nizet, V., K. F. Colina, J. R. Almquist, C. E. Rubens, and A. L. Smith. 1996. A virulent nonencapsulated Haemophilus influenzae. J. Infect. Dis. 173:180-186.
34. O'Neill, J. M., J. W. St. Geme III, D. Cutter, E. E. Adderson, J. Anyanwu, R. F. Jacobs, and G. E. Schutze. 2003. Invasive disease due to nontypeable Haemophilus influenzae among children in Arkansas. J. Clin. Microbiol. 41:3064-3069.
35. Pettigrew, M. M., B. Foxman, C. F. Marrs, and J. R. Gilsdorf. 2002. Identification of the lipooligosaccharide biosynthesis gene lic2B as a putative virulence factor in strains of nontypeable Haemophilus influenzae that cause otitis media. Infect. Immun. 70:3551-3556.
36. Rodriguez, C. A., V. Avadhanula, A. Buscher, A. L. Smith, J. W. St. Geme III, and E. E. Adderson. 2003. Prevalence and distribution of adhesins in invasive non-type b encapsulated Haemophilus influenzae. Infect. Immun. 71:1635-1642.
37. Sirakova, T., P. E. Kolattukudy, D. Murwin, J. Billy, E. Leake, D. Lim, T. DeMaria, and L. Bakaletz. 1994. Role of fimbriae expressed by nontypeable Haemophilus influenzae in pathogenesis of and protection against otitis media and relatedness of the fimbrin subunit to outer membrane protein A. Infect. Immun. 62:2002-2020.
38. St. Geme, J. W., III. 2002. Molecular and cellular determinants of non-typeable Haemophilus influenzae adherence and invasion. Cell. Microbiol. 4:191-200.
39. St. Geme, J. W., III, V. V. Kumar, D. Cutter, and S. J. Barenkamp. 1998. Prevalence and distribution of the hmw and hia genes and the HMW and Hia adhesins among genetically diverse strains of nontypeable Haemophilus influenzae. Infect. Immun. 66:364-368.
40. Swords, W. E., P. A. Jones, and M. A. Apicella. 2003. The lipo-oligosaccharides of Haemophilus influenzae: an interesting array of characters. J. Endotoxin Res. 9:131-144.
41. Tiller, F. W. 1982. Biochemical differentiation of Haemophilus influenzae. Additional characterization of biotypes by carbohydrate fermentation patterns. Zentbl. Bakteriol. Mikrobiol. Hyg. A 253:236-246.
42. Tullius, M. V., N. J. Phillips, N. K. Scheffler, N. M. Samuels, J. R. Munson, Jr., E. J. Hansen, M. Stevens-Riley, A. A. Campagnari, and B. W. Gibson. 2002. The lbgAB gene cluster of Haemophilus ducreyi encodes a -1,4-galactosyltransferase and an -1,6-DD-heptosyltransferase involved in lipooligosaccharide biosynthesis. Infect. Immun. 70:2853-2861.
43. van Alphen, L., D. A. Caugant, B. Duim, M. O'Rourke, and L. D. Bowler. 1997. Differences in genetic diversity of nonencapsulated Haemophilus influenzae from various diseases. Microbiology 143:1423-1431.
44. van Schilfgaarde, M., P. van Ulsen, P. Eijk, M. Brand, M. Stam, J. Kouame, L. van Alphen, and J. Dankert. 2000. Characterization of adherence of nontypeable Haemophilus influenzae to human epithelial cells. Infect. Immun. 68:4658-4665.
45. Veal, W. L., and W. M. Shafer. 2003. Identification of a cell envelope protein (MtrF) involved in hydrophobic antimicrobial resistance in Neisseria gonorrhoeae. J. Antimicrob. Chemother. 51:27-37.
46. Weiser, J. N. 1993. Relationship between colony morphology and the life cycle of Haemophilus influenzae: the contribution of lipopolysaccharide phase variation to pathogenesis. J. Infect. Dis. 168:672-680.
47. Weiser, J. N., J. M. Love, and E. R. Moxon. 1989. The molecular mechanism of phase variation of H. influenzae lipopolysaccharide. Cell 59:657-665.
48. Williams, B. J., M. Golomb, T. Phillips, J. Brownlee, M. V. Olson, and A. L. Smith. 2002. Bacteriophage HP2 of Haemophilus influenzae. J. Bacteriol. 184:6893-6905.
49. Williams, B. J., G. Morlin, N. Valentine, and A. L. Smith. 2001. Serum resistance in an invasive, nontypeable Haemophilus influenzae strain. Infect. Immun. 69:695-705.(Alice L. Erwin, Kevin L. )