The nucleotide sequence of a 5.1 kb region in the Haemophilus influenzae type b capsulation locus has been determined and found to contain four open reading frames: bexD, bexC, bexB, and bexA. Comparison of the deduced products of bexC, bexB, and bexA to known proteins, and TnphoA mutagenesis, suggests that they form components of an ATP-driven polysaccharide export apparatus. Furthermore, close sequence similarity between BexA and BexB and products of the kpsT and kpsM genes at the Escherichia coli K5 capsulation locus (Smith et al., 1990--accompanying paper) suggests that capsulation genes in these organisms may have a common ancestry.

The population of capsulate Haemophilus influenzae is divided into two phylogenetic divisions. Here we show that in division I strains the capsulation (cap) gene cluster lies between direct repeats of a novel insertion sequence (IS)-like element, IS1016. cap has apparently been mobilized in the chromosome as a compound transposon by IS1016, and the repeats have provided a molecular substrate for reversible cap gene amplification, with augmentation of capsule production, through unequal homologous recombination. Such amplification has occurred in serotype b strains, but in these a large direct repeat of cap genes has become fixed in the population. We have found a 1.2 kb deletion at one end of this duplicated capb locus, removing most of one copy of the polysaccharide export gene bexA. We have shown that this makes capsulation dependent on preservation of the direct repeat structure in order to avoid recombination-mediated loss of the other copy of bexA. Type b strains with this cap configuration are disseminated worldwide and currently cause nearly all invasive Haemophilus infections, leading us to speculate that the 1.2 kb deletion occurred in an ancestral type b strain and conferred significant biological advantage.

Copper-zinc superoxide dismutase ([Cu,Zn]-SOD) is widely found in eukaryotes but has only rarely been identified in bacteria. Here we describe sodC, encoding [Cu,Zn]-SOD in Haemophilus influenzae and H. parainfluenzae, frequent colonists and pathogens of the human respiratory tract. In capsulate H. influenzae, sodC was found in only one division of the bacterial population, and although the protein it encoded was clearly [Cu,Zn]-SOD from its deduced sequence, it lacked enzymatic activity. In H. parainfluenzae, in contrast, active enzyme was synthesized which appeared to be secreted beyond the cytoplasm when the gene was expressed in Escherichia coli minicells. The origin of gene transcription differed between the Haemophilus species, but protein synthesis from cloned genes in vitro was comparable. A C-T transition was found in the H. influenzae sequence compared with the H. parainfluenzae sequence, leading to a histidine, known to be crucial in eukaryotic [Cu,Zn]-SOD for copper ion coordination and so for enzymatic activity, to be changed to tyrosine. This is speculated to be the cause of inactivity of the H. influenzae enzyme. Secreted SODs have only been described in a few bacterial species, and this is the first identification of [Cu,Zn]-SOD in a common human upper respiratory tract colonist. The role of secreted bacterial SODs is unknown, and we speculate that in Haemophilus species the enzyme may confer survival advantage by accelerating dismutation of superoxide of environmental origin to hydrogen peroxide, disruptive to the normal mucociliary clearance process in the host.

Cloned Haemophilus influenzae type b capsulation genes were used as hybridization probes to isolate DNA from the capsulation loci (cap) of other serotypes of H. influenzae. Mapping of the resulting clones and Southern hybridization analysis of chromosomal DNAs from type a, b, c, and d strains showed that in each strain cap was organized in the same way: a central DNA segment specific to each serotype flanked by DNA segments of common structure. We infer that enzymes necessary for the synthesis of specific capsular polysaccharide are encoded in the central segment of cap, while proteins involved in a more general way in the process of capsulation are encoded in the flanking segments. Studies of the function of the DNA in one of these non-serotype-specific flanking segments (J. S. Kroll, I. Hopkins, and E. R. Moxon, Cell 53:347-356, 1988) have previously identified a gene encoding a protein necessary for polysaccharide export, an event now deduced to proceed by a mechanism independent of the nature of the disaccharide subunit in the polysaccharide. The near-total duplication of cap that has been found in most type b strains was not found at the analogous locus in the other serotypes. This reinforces our previous hypothesis, based on were to be found between strains in different divisions, while similarities existed between strains studied from one division, regardless of serotype.In this work, we showed that the cap locus in strains segregating to phylogenetic division I has a physical organization independent of serotype and that the capsular phenotype expressed appears to depend on the nature of a cassette of DNA inserted into common flanking sequences in cap. MATERIALS AND METHODSMedia and culture conditions. H. influenzae strains were grown in brain heart infusion broth supplemented with 2 jig of NAD and 10 ,ug of hemin per ml. Translucent brain heart infusion plates were prepared with 1% agar and supplemented with 10% Levinthal base (1). Colony phenotypes were assessed by viewing with obliquely transmitted light as previously described (5)