Genomic Sequence of an Otitis Media Isolate of Nontypeable Haemophilus influenzae: Comparative Study with H. influenzae Serotype d, Strain KW20

Abstract

In 1995, the Institute for Genomic Research completed the genome sequence of a rough derivative of Haemophilus influenzae serotype d, strain KW20. Although extremely useful in understanding the basic biology of H. influenzae, these data have not provided significant insight into disease caused by nontypeable H. influenzae, as serotype d strains are not pathogens. In contrast, strains of nontypeable H. influenzae are the primary pathogens of chronic and recurrent otitis media in children. In addition, these organisms have an important role in acute otitis media in children as well as other respiratory diseases. Such strains must therefore contain a gene repertoire that differs from that of strain Rd. Elucidation of the differences between these genomes will thus provide insight into the pathogenic mechanisms of nontypeable H. influenzae. The genome of a representative nontypeable H. influenzae strain, 86-028NP, isolated from a patient with chronic otitis media was therefore sequenced and annotated. Despite large regions of synteny with the strain Rd genome, there are large rearrangements in strain 86-028NP's genome architecture relative to the strain Rd genome. A genomic island similar to an island originally identified in H. influenzae type b is present in the strain 86-028NP genome, while the mu-like phage present in the strain Rd genome is absent from the strain 86-028NP genome. Two hundred eighty open reading frames were identified in the strain 86-028NP genome that were absent from the strain Rd genome. These data provide new insight that complements and extends the ongoing analysis of nontypeable H. influenzae virulence determinants.

In 1995 Haemophilus influenzae strain Rd, a rough derivative of H. influenzae serotype d strain KW20 (strain Rd hereafter), became the first free-living organism to have its genome sequenced to completion (34). Importantly, this also helped establish the large-scale shotgun approach, mated with the utilization of a scaffolding library and computer-assisted assembly, as a rational and expeditious approach for the sequencing of small bacterial genomes. Strain Rd was chosen as the prototypic bacterium for complete genome sequencing as it has a genome size representative of other bacteria and a G+C content close to that of the human genome. Additionally, at the time of sequencing, a physical map of the strain Rd genome did not exist, so this genome was a good test for the approach of shotgun sequencing, scaffolding, and assembly (34).

Although strain Rd is the exemplar organism for the current small-genome sequencing rationale and an important model organism for studying H. influenzae biology, strain Rd is a poor model for the study of pathogenicity caused by members of the genus Haemophilus. Serotype b strains of H. influenzae cause invasive diseases, for example, meningitis, and nontypeable H. influenzae (NTHi) strains principally have a role in localized respiratory disease, particularly in otitis media, acute sinusitis, and community-acquired pneumonia and have important consequences in patients with chronic obstructive pulmonary disease or cystic fibrosis (59, 74, 95, 100, 108). Strain Rd, however, is a derivative of a serotype d strain. Serotype d strains are rarely associated with disease (23, 44, 94, 102).

Because one of the most useful sets of data in the study of an organism's biology is its genomic sequence, a number of investigations have identified and characterized genes found in H. influenzae type b strains, H. influenzae biogroup Aegyptius strains or in nontypeable strains that are not present in strain Rd (13, 17, 30, 60, 64, 82, 106). Previously we carried out partial analyses and comparisons of the genomes of two NTHi strains, 1885MEE and 86-028NP, that were isolated from the middle ear and nasopharynx, respectively, of children with chronic otitis media (72). A genomic DNA-based microarray approach was employed to compare the genomes of strains 1885MEE and Rd as well as the genomes of strains 1885MEE and 86-028NP. In concert, a bioinformatics approach was used to compare the published genome of strain Rd with that of strain 86-028NP, which had been sequenced to threefold coverage. These analyses suggested that the genomes of strains 1885MEE and 86-028NP are more similar to each other than to the genome of strain Rd. Moreover, both analytical approaches identified a number of genes present in the NTHi strains that were absent from strain Rd. These included genes encoding proteins involved in protection against reactive oxygen species (tsaA), with possible roles in adhesion, biofilm formation, and pH regulation (tnaA), and a member of a family of virulence-associated autotransporters (lav).

Although this analysis was of great utility, due to of the fragmentary nature of the data available, it was desirable to expand this analysis. We thus completed sequencing the genome of strain 86-028NP and compared the completed genome to the genome of strain Rd. This allowed us to identify genes common to both strains and, importantly, strain 86-028NP-specific genes that may be involved in virulence. As strain 86-028NP has been extensively characterized, both in chinchilla models of otitis media and at the genetic level, having a full understanding of the genetic make-up of strain 86-028NP will complement existing data as well as allow a better understanding of the processes involved in the pathogenesis of otitis media and, by extension, other diseases in which H. influenzae is involved.

MATERIALS AND METHODS

Choice of strain sequenced.

Previous studies have indicated that NTHi strains are heterogeneous with respect to outer membrane protein profiles (8, 75), enzyme allotypes (76), and lipooligosaccharide genes (18). Furthermore, attempts to subtype NTHi did not identify a suitable candidate for genomic sequencing. We therefore chose the low-passage isolate strain 86-028NP, which was recovered from the nasopharynx of a child with chronic otitis media. This strain has subsequently been well characterized both in vitro (6, 46), in chinchilla models of otitis media (5, 58, 63, 111), and partially at the genomic level (72).

Library construction.

Chromosomal DNA was prepared from strain 86-028NP using Puregene reagents (Gentra Systems, Minneapolis, MN). For the initial shotgun sequencing of the genome, libraries of 1 to 2 kb and 2 to 4 kb of genomic DNA were constructed in pUC18 as previously described (72). For the scaffolding library, genomic DNA was manually sheared into a mean fragment size of 40 kb using a Hamilton syringe. After end repair, fragments were fractionated using a 0.7% low-melting-temperature agarose gel. Fragments larger than 30 kb were excised, and an in-gel ligation to pEpiFOS-5 was performed. The ligation mixture recovered from the gel was packaged into lambda phage in vitro and used to transfect EPI100 cells (Epicentre, Madison, WI).

Sequencing.

For the shotgun portion of the sequencing, cycle-sequencing reactions were run using PE Big-Dye terminators and universal primers (M13 forward and reverse) as previously described (72). To end-sequence the scaffolding library, the plasmid was first purified using an R.E.A.L. Prep 96 plasmid kit (QIAGEN Inc., Valencia, CA), then amplified using a TempliPhi DNA amplification kit (Amersham Biosciences Corp., Piscataway, NJ) before running reactions using PE Big-Dye terminators and pEpiFOS-5 forward and reverse sequencing primers (Epicentre, Madison, WI). The reactions for the clean-up portions of the project were run using PE Big-Dye terminators and custom primers (Integrated DNA Technologies, Coralville, IA). Excess dye terminators were removed with Sephadex G50 columns in a 96-well format, and sequence was determined on either an ABI 3700 or an ABI 3100 capillary electrophoresis DNA sequencer (Applied Biosystems, Foster City, CA).

Genome closure.

Paired end sequences from the scaffolding library and PCR were used to order the contigs and to add sequence in areas of low sequence coverage. Paired custom primers (Integrated DNA Technologies, Coralville, IA) were designed to bind at the ends of each contig as well as regions flanking areas of low sequence coverage. The intervening regions were amplified with a standard PCR protocol (98) using Taq polymerase (Roche Diagnostics, Indianapolis, IN) and sequenced on both strands. rRNA operons and the high-molecular-weight gene clusters were completely sequenced using clones from the scaffolding library as templates.

Assembly.

Phred/Phrap was used for data assembly, employing the default assembly parameters (32, 33, 36) as described (72). Assemblies were checked using the paired-end sequence data from 507 clones using the Seqman II program from the DNASTAR suite.

Data analyses.

Coding regions were identified using Glimmer2 (v2.13) trained on the 1,178 longest open reading frames (ORFs) identified by the Glimmer2 long-orfs program (25). Automated annotation by similarity was done by searching the Glimmer ORF set against the strain Rd proteome, the SwissProt database, the NCBI COGs database, and the KEGG database. The strain Rd database was compared bidirectionally with the strain 86-028NP ORF set using Tricross to determine high-confidence regions of similarity and to produce the dotplot comparison of genome organization (87).

The automatically predicted annotation information was further manually curated using Artemis (97) for visualization and demarcation of genomic regions of interest, and a custom FileMaker Pro database was generated which was then used to apply manual revisions and archive data related to the functional assignment. FASTA analyses were used for the primary automated comparisons. The strong synteny between the strain 80-028NP and strain Rd genomes allowed assignment of a function to the majority of the genes automatically, with similarity held to 90% or better at the amino acid level for matching. The nearly one-to-one mapping from the strain 86-028NP genome to the strain Rd genome was confirmed by assembly of the strain Rd ORFs onto the strain 86-028NP genome sequence and the reverse assembly of the strain 86-028NP ORFs onto the Rd genome, using the SeqMan program with the assembly criterion of 80% identity at the nucleotide level.

Manual BLAST analyses were used to explore the potential function of ORFs that did not show strong similarity to known genes. Manual curation of the automatic assignments was carried out to conform annotations to the current literature and repair the few places where the automated algorithm was easily led astray (notably the high-molecular-weight gene clusters, the hemoglobin-binding proteins, and the hsd gene clusters, whose high family similarity confounds automated assignment).

The tRNA genes were identified by tRNAscan-SE v1.11 (61). The rRNA operons were identified based on 16, 23, and 5S rRNA similarity with strain Rd and the ClustalW alignment of the neighborhoods containing these genes to determine the boundaries of the semiconserved regions.

Nucleotide sequence accession number.

The genome sequence is available from GenBank, where it has been assigned accession number CP000057. A list of the ORFs that are unique to each strain is given in Tables S1 and S2 at our website, www.microbial-pathogenesis.org/H.influenzae.86028/.

RESULTS AND DISCUSSION

Comparison between the genomes of NTHi strain 86-028NP and H. influenzae strain Rd.

The genomic sequence of strain 86-028NP contains 1,913,428 bp. This is approximately 4% larger than the strain Rd genome (1,830,137 bp) (34). There are also a larger number of genes in strain 86-028NP, 1,821 compared to 1,743 in strain Rd. A graphical representation of the strain 86-028NP genome is shown in Fig. 1. The gene complement was compared to that of strain Rd using the Seqman program in the DNASTAR suite. With 80% identity at the nucleotide level as a cutoff value, 280 ORFs were identified in the 86-028NP genome that were absent from the strain Rd genome, and 169 ORFs were identified in the strain Rd genome that are absent from the strain 86-028NP genome.

FIG. 1.

Schematic diagram of the organization of the H. influenzae strain 86-028NP genome. NTHI0001 is at the top of the figure, and increasing coordinates as annotated proceed clockwise around the circle. From the periphery in, the diagram indicates the following information. Labeled arrows indicate the identity, position, and direction of key gene loci. The maroon and orange double band indicates ORF location and coding direction. Clockwise-facing (positive strand as annotated) ORFs are indicated in the outer maroon band, while counterclockwise-facing ORFs are indicated in the inner, orange band. tRNA and rRNA operon positions and directions are indicated by the ring of purple and cyan arrows, purple for tRNAs and cyan for rRNA operons. The triple band of dark red, green, and dark blue demarcations indicates the major predicted COG classification for the corresponding ORF. Dark red indicates information storage and processing, green indicates cellular processes and signaling, and blue indicates metabolism. The prevailing gene order, in comparison to the strain Rd genome in a sliding window 10 ORFs wide, is indicated by the colored and offset background. Yellow inset regions are highly similar to strain Rd and follow the same gene order. Pink outset regions are highly similar to strain Rd and are in the reverse gene order from the strain 86-028NP genome. Regions with neither yellow nor pink background have no significant similarity to strain Rd.

Strain 86-028NP, like strain Rd, has six ribosomal operons. Using tRNAscan-SE v1.11, 58 tRNA genes were identified in the strain 86-028NP genome, representing the 20 common amino acids. The tRNA-Glu, tRNA-Ala, and tRNA-Ile genes were located in spacer regions between the 16S and 23S rRNA genes. A tRNA gene containing the UCA anticodon was also identified. This anticodon corresponds to an opal stop codon and is typically associated with an opal-suppressing tRNA that incorporates selenocysteine. The tRNA is adjacent to two genes encoding selB (NTHI0836), a Sec tRNA-specific elongation factor, and selA (NTHI0835), the enzyme that converts serine to dehydroalanine preparatory to forming selenocysteine by incorporation of selenium (35). The selD gene (NTHI0297), encoding selenophosphate synthetase, was also identified. The importance of this selenocysteine system is evidenced by the coding sequence for the alpha subunit of formate dehydrogenase (NTHI0007) containing an in-frame TGA stop codon that is presumably read as a selenocysteine codon. The in-frame TGA stop codon was previously noted in the current annotation of the strain Rd formate dehydrogenase gene (GenPept accession P46448).

A gross comparison between the genomes involving analysis of the gene order of strain 86-028NP and that of strain Rd reveals a single major rearrangement in the form of a large inversion (Fig. 1 and 2). This 471-kb inversion represents almost 25% of the strain 86-028NP genome and is bounded by NTHI1391 and NTHI1394 (homologues of HI1218 and HI1645, respectively) and by NTHI1949 and NTHI1950 (homologues of HI1219 and HI1647, respectively). HI1219 and HI1646 are partially duplicated genes in strain Rd annotated as cmkA and cmkB (cytidylate kinases). One cmk gene (NTHI1949) is present in strain 86-028NP with a small cmk-like fragment between NTHI1391 and NTHI1394. Several clones from the scaffolding library overlap each end of the inversion in the 86-028NP genome, validating our assembly. Within this large inversion are several insertions, the largest of which are approximately 13 kb, 27 kb, and 51 kb in size. These regions contain predominantly hypothetical and conserved hypothetical genes as well as a number of homologues of phage genes. For example, the 27-kb insertion contains remnants of HP1- and HP2-like phage genes. The largest insert is bounded by homologues of integrase genes. In strain Rd, a mu-like phage is localized to this region (67). This phage is not present in the strain 86-028NP genome. Also within the large inverted region is a 21-kb inversion that restores synteny with the Rd genome.

FIG. 2.

Tricross dotplot diagram of the gene organization of strain 86-028NP compared to strain Rd. Each Tricross two-way hit between strain 86-028NP and strain Rd ORFs attaining a translation-to-translation FASTA E-score of 10⁻¹⁶ or better, is plotted in the diagram.

In addition to the large inversion, strain 86-028NP has other regions of divergence from colinearity with the strain Rd genome. These include nine regions greater than 5 kb, which contain sequences with no apparent homology to DNA that is present in strain Rd. Two of these regions contain the high-molecular-weight adhesins that are discussed below. Hypothetical genes predominate in six of the unique regions. The ninth region is approximately 56 kb in size. It lies between NTHI0100 and NTHI0165. BLASTn analysis indicates that genes in this region, designated ICEHin86-028NP, have high homology to genes in the H. influenzae type b plasmid ICEHin1056 (66). ICEHin1056 is a member of an extended family of genomic islands that are defined by a series of common core genes (66). ICEHin86-028NP possesses homologues of 45 ICEHin1056 ORFs. These include ORFs near the 5′ end of ICEHin86-028NP, including the defined core genes, that primarily encode proteins with putative roles in plasmid replication and conjugation and ORFs near the 3′ end that primarily encode conserved hypothetical proteins with motifs that suggest that they may be either membrane associated or exported. Notably, ICEHin86-028NP lacks the genes encoding proteins involved in tetracycline, chloramphenicol, and β-lactam resistance found in ICEHin1056. Scattered within ICEHin86-028NP are a transposase, resolvases, and a putative integrase regulator, suggesting that ICEHin86-028NP is a composite element derived from several mobile genetic elements.

ICEHin1506 has a sequence designated as an attP site 5′ of the first gene. In strain 86-028NP, a perfect copy of this attP site is present 5′ to NTHI0101 and a copy of this attP site, with a single nucleotide change, is present 3′ of NTHI0164. attP sites are implicated in the incorporation of mobile genetic elements into bacterial chromosomes to form genomic islands, possibly suggesting a mechanism by which this large section of genetic material became integrated into the strain 86-028NP genome (27). ICEHin86-028NP has a G+C content of 39%, lower than any of the other related genomic islands and close to strain 86-028NP's overall genome G+C content of 38%. This implies a long-term genomic association for this element. The presence of this element with its complement of genes homologous to those in ICEHin1506 (27), which are thought to encode membrane-associated and secreted proteins, may have important implications for the virulence of strain 86-028NP.

Several members of the family Pasteurellaceae, including Haemophilus ducreyi, Pasteurella multocida, and Actinobacillus actinomycetemcomitans, produce well-characterized protein toxins. In contrast, H. influenzae does not appear to produce protein toxins, and genes encoding putative protein toxins were not identified in the strain 86-028NP genome. In H. influenzae, the genes encoding glycosyltransferases, responsible for endotoxin biosynthesis, and genes encoding proteins that give the bacteria enhanced “fitness” during the process of infection have generally been considered virulence determinants. These genes include those that encode adhesins, the heme and haemoglobin binding proteins, as well as the genes that encode proteins that protect against oxidative stress. In the following sections, we discuss these genes, among others, in strain 86-028NP and note whether there are homologues in strain Rd.

Contingency genes.

H. influenzae has a limited number of two-component regulatory systems and other global regulators. Moxon and coworkers have argued that loci termed “simple contingency loci” provide an alternative mechanism for regulating gene expression, thus increasing the fitness of an organism by contributing to that organism's ability to rapidly respond to changing environmental conditions. These loci contain short tandem sequence repeats either within or 5′ to a coding region. During DNA replication, addition or loss of a repeat within a reading frame results in an alteration in the reading frame. When localized 5′ to a coding region, addition or loss of a repeat results in a change in promoter activity (11). Loci containing simple sequence repeats have been studied extensively in H. influenzae, for example (50). Several of the loci described in the following sections as phase variable contain simple sequence repeats.

Adhesins.

Strain 86-028NP possesses a number of genes whose products' primary function is in adherence to host cells (Table 1). One of these, the outer membrane protein P5, has previously been identified and its function was carefully dissected (54, 58, 77-79, 101). Recently, we demonstrated that strain 86-028NP possesses a gene cluster containing four genes that are homologues of pilABCD from strain Rd, Actinobacillus pleuropneumoniae, and Pasteurella multocida (3, 28, 96, 107). These genes, together with the comE gene and genes yet to be identified, encode a type IV pilus that has a role in adherence of strain 86-028NP to nasopharyngeal tissues (57).

TABLE 1.

Strain 86-028NP genes encoding proteins with a role in adherence

NTHi no.a	HI no.b	Gene name	Function	Contingency repeats
354		hap	Adhesion and penetration protein Hap
406	296	pilD	Putative type 4 prepilin-like protein specific leader peptidase (EC 3.4.23.43)
407	297	pilC	Putative type IV pilin secretion protein
408	298	pilB	Putative type IV pilin secretion protein
409	299	pilA	Type IV pilin subunit protein
1332	164	ompP5	Outer membrane protein P5 (OMP P5-homologous adhesin)
1448		hmw2C	HMW2C, putative glycosyltransferase involved in glycosylation of HMW1A and HMW2A
1449		hmw2B	HMW2B, OMP-85-like protein required for HMW1A and HMW2A secretion
1450		hmw2A	HMW2A, high-molecular-weight adhesin 2	ATCTTTC repeated 23 times; 5′ of gene
1983		hmw1A	HMW1A, high-molecular-weight adhesin 1	ATCTTTC repeated 17 times, 5′ of gene
1984		hmw1B	HMW1B, OMP-85-like protein required for secretion of HMW1A and HMW2A
1985		hmw1C	HMW1C, putative glycosyltransferase involved in glycosylation of HMW1A and HMW2A

^aAssigned locus number for each strain 86-028NP gene listed.

^bLocus number assigned by TIGR for the corresponding Rd homolog.

Strain 86-028NP possesses two high-molecular-weight (HMW) adhesin gene clusters that are absent in strain Rd. The high-molecular-weight adhesins were first characterized in NTHi strain 12, which has two HMW gene clusters, each encoding three proteins (HMWA, HMWB, and HMWC). HMWA is the structural component of the adhesin, HMWB has a role in transmembrane translocation, and HMWC is required for glycosylation of HMWA (7, 9, 37, 110). Similarly, strain 86-028NP's two HMW gene clusters contain homologues of the hmwA, hmwB, and hmwC genes in the same gene context as in strain 12 (16). The HMW1A and HMW2A proteins from strain 86-028NP are 72% identical, with the major area of divergence, including a 41-amino-acid insertion in HMW2A, toward the C termini. The paired HMWB and HMWC proteins from strain 86-028NP are 99% identical. The sequence ATCTTTC is repeated 17 times upstream of hmw1A and 23 times upstream of hmw2A. In strain 12, 16 repeats of this sequence are found 5′ of each hmw gene cluster (7).

Hap is an autotransported protein with a domain homologous to the catalytic domain of immunoglobulin A1 proteases. The NTHI0354 gene encodes a protein with 83% identity to Hap from NTHi strain N187 (109). Strain 86-028NP, along with other NTHi strains that possess HMW1 and HMW2, lacks the gene encoding Hia, another Haemophilus adhesin (10). Strain 86-028NP also lacks the hif gene cluster, encoding the hemagglutinating pilus, as we previously reported (72).

Lipooligosaccharide synthesis.

The structure, biosynthesis, and role in virulence of H. influenzae lipooligosaccharide has been studied extensively. Table 2 contains a list of genes involved in lipooligosaccharide biosynthesis. Strain 86-028NP has the full complement of genes required to synthesize the heptose-2-keto-3-deoxyoxtulosonic acid-lipid A portion of lipooligosaccharide. The lgtF and lpsA genes encode glycosyltransferases that add glucose and glucose or galactose to heptose residues 1 and 3, respectively. Both of these genes are present in the strain 86-028NP genome, and therefore it is likely that carbohydrate chains can be extended from the heptose 1 and heptose 3 residues of the strain 86-028NP lipooligosaccharide (49).

TABLE 2.

Strain 86-028NP genes encoding proteins with a role in lipooligosaccharide biosynthesis

NTHi no.a	HI no.b	Gene name	Function	Contingency repeats
68	58	kdsB	3-Deoxy-d-manno-octulosonic acid cytidylyltransferase
69	59	lpxK	Tetraacyldisaccharide 4′-kinase
72	60	msbA	Lipid A export ATP-binding protein msbA
296	199	msbB	Lipid A biosynthesis (KDO)2-(lauroyl)-lipid IVA acyltransferase
365	258	lgtC	UDP-galactose-lipooligosaccharide galactosyltransferase	GACA repeated 10 times, in frame
366	260	orfM	Xanthosine triphosphate pyrophosphatase
367	260.1	kdkA	3-Deoxy-d-manno-octulosonic acid kinase
368	261	opsX	ADP-heptose-lipooligosaccharide heptosyltransferase I
383	275	lpt6	PE-tn-6-lipooligosaccharide phosphorylethanolamine transferase
471	351	galE	UDP-glucose 4-epimerase
472	352	lic3A	CMP-Neu5Ac-lipooligosaccharide alpha-2-3-sialyltransferase	CAAT repeated 18 times, in frame
512	391		Predicted acyltransferase	AGCA repeated 8 times, in frame
649	523	waaQ	ADP-heptose-lipooligosaccharide heptosyltransferase III
677	550	lic2A	UDP-galactose-lipooligosaccharide galactosyltransferase	CAAT repeated 14 times, in frame
772	652	kdtA	3-Deoxy-d-manno-octulosonic acid transferase
773	653	lgtF	UDP-glucose-lipooliqosaccharide glucosyltransferase
892	735	lpxH	UDP-2,3-diacylglucosamine hydrolase
899	740	pgmB	Phosphoglucomutase
913		lex2B	UDP-glucose-lipooliqosaccharide glucosyltransferase
926	765	lpsA	Lipooligosaccharide glycosyltransferase
976	812	qalU	UTP-glucose-1-phosphate uridylyltransferase
1034		lic3A2	CMP-Neu5Ac-lipooligosaccharide alpha-2-3-sialyltransferase	CAAT repeated 18 times, in frame
1037	873	rmlB	dTDP-glucose 4,6-dehydratase
1082	915	lpxD	UDP-3-O-[3-hydroxymyristoyl] glucosamine N-acyltransferase
1180	1005		Predicted PE-lipooligosaccharide phosphorylethanolamine transferase
1220	1060	lpxB	Lipid-A-disaccharide synthase
1222	1061	lpxA	Acyl-[acyl-carrier-protein]-UDP-N-acetylglucosamine O-acyltransferase
1224	1064		Predicted PE-lipooligosaccharide phosphorylethanolamine transferase
1272	1105	rfaF	ADP-heptose-lipooligosaccharide heptosyltransferase II
1278	1114	rfaD	ADP-l-glycero-d-manno-heptose-6-epimerase
1312	1144	lpxC	UDP-3-O-[3-hydroxymyristoyl] N-acetylglucosamine deacetylase
1350	1181	lpcA	Phosphoheptose isomerase
1474	1578	lgtD	Putative UDP-GlcNAc-lipooligosaccharide N-acetylglucosamine glycosyltransferase
1576	1557	kdsA	Phospho-2-dehydro-3-deoxyoctonate aldolase and 3-deoxy-d-manno-octulosonic add 8-phosphate synthetase
1594	1540	licD	Phosphorylcholine transferase
1595	1539	licC	Protein LicC, CTP-phosphocholine cytidylyltransferase
1596	1538	licB	Protein LicB, putative choline uptake protein
1597	1537	licA	Protein LicA, choline kinase	CAAT repeated 15 times, in frame
1606	1527	htrB	Lipid A biosynthesis lauroyl acyltransferase
1607	1526	rfaE	ADP-heptose synthase
1664	1337	mrsA	Predicted phosphomannomutase
1750			Putative glycosyltransferase, glycosyltransferase family 8 protein	GACA repeated 14 times, in frame
1769			Putative glycosyltransferase	CCAA repeated 17 times, out of frame
1891	1279	slaB	CMP-Neu5Ac synthetase
1921	1244		Possible polysaccharide biosynthesis protein
2002	1695	lsgF	Putative UDP-galactose-lipooligosaccharide galactosyltransferase
2003	1696	lsgE	Putative UDP-galactose-lipooligosaccharide galactosyltransferase
2004	1697	lsgD	Putative UDP-GlcNAc-lipooligosaccharide N-acetylglucosaminyl glycosyltransferase
2005	1698	lsgC	Putative UDP-galactose-lipooligosaccharide galactosyltransferase
2006	1699	lsgB	CMP-N-acetylneuramlnate-beta-galactosamide-alpha-2,3-sialyltransferase
2007	1700	lsgA	Putative lipooligosaccharide flippase
2025	1716	wecA	Undecaprenyl-phosphate alpha-N-acetylglucosaminyl 1-phosphate transferase

^aAssigned locus number for each strain 86-028NP gene listed.

^bLocus number assigned by TIGR for the corresponding Rd homolog.

In the serotype b strain RM153, the lic2C gene encodes a glucosyltransferase that adds glucose to heptose 2 (49). In the strain 86-028NP genome, this gene contains a frame shift. The phase-variable lic2A and licA genes, encoding a galactosyltranferase and choline kinase, respectively, are present in the strain 86-028NP genome (45, 48, 117). The lex2B gene, which encodes a glucosyltransferase in the serotype b strain DL42 as well as a number of other serotypeable strains, is present in the strain 86-028NP genome (41, 53). Five-prime to the lex2B gene in strain DL42 is the short phase-variable lex2A gene. In strain 86-028NP, this gene is out of frame compared to the DL42 sequence (GenBank accession U05670), due to the loss of one tetranucleotide repeat and a 5-bp deletion.

Recently, Hood and coworkers described a locus in strain Rd, designated hmg, that contains HI0866 through HI0874 (51). With the exception of a homologue of rmlB, these genes are absent from the strain 86-028NP genome. This includes the siaA gene, which encodes a sialyltransferase recently shown to be important in biofilm formation in NTHi strain 2019 (40, 56). Two copies of a homologue of the lic3A gene, encoding an alternative sialyltransferase, were identified in the strain 86-028NP genome (47, 56), as well as a copy of the lsgB gene, which encodes another sialyltransferase (56).

Iron acquisition.

H. influenzae strains have an absolute requirement for either heme or iron, together with protophorphyrin IX, the immediate precursor of heme (31, 118). Table 3 contains a list of genes involved in iron acquisition. Three hemoglobin and hemoglobin-haptoglobin binding proteins HgpA, HgpB, and HgpC were identified in H. influenzae type b strain HI689 (55, 69, 91). In strain HI689, these genes have CCAA tetranucleotide repeats and are known to be regulated by slip-strand mispairing. Two of these genes are present in strain 86-028NP. They both contain CCAA repeats; the hgpB gene is in frame, while the hgpC gene is out of frame. The derived amino acid sequence of a third gene that contains CCAA repeats is 45% identical to hgpA. We have designated this gene hgpD. This gene is out of frame.

TABLE 3.

Strain 86-028NP genes encoding proteins with a role in iron acquisition

NTHi no.a	HI no.b	Gene name	Function	Contingency repeats
177	97	hitA	hFbpA, iron utilization periplasmic protein
179	98	hitB	hFbpB, iron(III)-transport system permease protein
180	99	hitC	hFbpC, iron utilization ATP-binding protein
202	113	hemR	Hemin receptor
284	190	fur	Ferric uptake regulation protein
369	262	hxuC	Heme/hemopexin-binding protein C (heme:hemopexin utilization protein C)
370	263	hxuB	Heme/hemopexin-binding protein B (heme:hemopexin utilization protein B)
371	264	hxuA	Heme/hemopexin-binding protein A (heme:hemopexin utilization protein A)
477	359	hfeD	Putative ABC-type chelated iron transport system, permease component
478	360	hfeC	Putative ABC-type chelated iron transport system, permease component
479	361	hfeB	Putative ABC-type chelated iron transport system, ATPase component
481	362	hfeA	Putative periplasmic chelated iron binding protein
736		hgpD	Hemoglobin-haptoglobin binding protein D (hemoglobin-haptoglobin utilization protein D)	CCAA repeated 17 times, out of frame
782	661	hgpB	Hemoglobin-haptoglobin binding protein B (hemoglobin-haptoglobin utilization protein B)	CCAA repeated 12 times, in frame
840	712	hgpC	Hemoglobin-haptoglobin binding protein C (hemoglobin-haptoglobin utilization protein C)	CCAA repeated 20 times, out of frame
1021	853	hbpA	Heme-binding protein A (hemin-binding lipoprotein)
1168	994	tbpA	Transferrin-binding protein 1
1169	995	tbpB	Transferrin-binding protein 2
1329	1160	hemH	Ferrochelatase
1390	1217	hup	Heme utilization protein
2035	1728		Mn2+ and Fe2+ transporter of the NRAMP family

^aAssigned locus number for each strain 86-028NP gene listed.

^bLocus number assigned by TIGR for the corresponding Rd homolog.

Homologues of the hxuABC genes of H. influenzae type b that encode heme and heme-hemopexin complexes (19-21) as well as a homologue of the hemR receptor were identified. Strain 86-028NP also has the gene encoding the heme-binding lipoprotein HbpA (43). Downstream of hbpA is NTHI1022, a hypothetical gene whose product is a member of COG0748, a cluster that includes putative heme utilization proteins. A homologue of the hup gene, recently identified in H. influenzae type b, that encodes a general heme utilization protein was also identified (68).

In addition to the heme transport systems, iron transport systems were also identified. The hitABC genes encode the FbpABC proteins, respectively, members of a highly specific ferric iron ABC transport system that was elegantly characterized by complementing a siderophore-deficient Escherichia coli strain with the hitABC genes cloned from an H. influenzae type b strain (1). Transferrin-binding proteins 1 and 2 encoded by tbpAB (38, 39) as well as genes designated hfeABCD that are homologues of an ABC transport system involved in iron uptake, originally characterized in Yersinia pestis (12), were identified. This gene cluster is also present in strain Rd. NTHI2035 encodes a putative homologue of the NRAMP family of Mn²⁺ and Fe²⁺ transporters (92).

As noted above, H. influenzae can use iron, together with protoporphyrin IX, as a source of heme for growth in vitro. The hemH gene encoding ferrochelatase, which catalyzes the incorporation of iron into protoporphyrin IX (99), was identified. The gene encoding the global regulator, Fur, was also identified (2, 105).

Oxidative stress.

Although necessary for growth, the active acquisition of iron can have deleterious effects on bacterial cells. Through the Fenton reaction, iron can react with hydrogen peroxide and generate highly reactive hydroxyl radicals. These products have profound effects, including lipid peroxidation and damage to both iron-containing enzymes and DNA (52). The best-known defense system against hydroxyl radicals consists of superoxide dismutases A and B, which convert highly reactive superoxide to hydrogen peroxide, which is then converted, by catalase, into water and oxygen (26).

Strains 86-028NP and Rd contain the sodA gene (NTHI1251) but lack the sodB gene. Both strains also possess a catalase gene, hktE (NTHI1099) (14), the oxyR gene (NTHI0704), encoding a primary regulator of genes involved in protection against oxidative stress (62, 83), and the gene encoding a chimeric peroxidase termed Prx/Grx that has a glutathione-dependent role in protection against small alkyl hydroperoxides (81, 115, 116). We previously identified NTHI0212, a gene encoding a homologue of the P. multocida peroxiredoxin, TsaA, that is absent in strain Rd (72). Strain 86-028NP, however, lacks AhpF, a dedicated alkyl hydroperoxide reductase known to be involved in the reduction of TsaA in Salmonella spp. (84).

Further protection against oxidative stress may be afforded by the ferritin-like proteins encoded by the ftnA and ftnB (NTHI1773 and NTHI1772, respectively) genes. Overexpression of these proteins was shown to protect an iron-overloaded E. coli fur mutant against oxidative damage (113). A conserved hypothetical gene, NTHI1817, encodes a protein with homology to a DNA-binding ferritin-like protein. This is a member of the Dps family of nonspecific DNA binding proteins, which in Salmonella enterica have roles in protection against oxidative stress, both in the presence of iron and during phagocytosis, and are important for virulence in a murine model of Salmonella infection (42). In E. coli, Dps was shown to preferentially bind iron that had been oxidized by hydrogen peroxide, thus having an important role in abrogating the production of hydroxyl radicals generated via the Fenton reaction (120).

Secretion.

In addition to the Sec system, strain 86-028NP has genes that encode the TatA, TatB, and TatC proteins, cytoplasmic membrane-associated proteins that are involved in Sec-independent transport of proteins with twin arginines in their signal peptides (NTHI0279, NTHI0280, and NTHI0282) (15, 119). As previously reported, strain 86-028NP possesses NTHI0585, the gene encoding the autotransported protein Lav (72). This protein is absent in strain Rd, present in Neisseria spp., and appears, within Haemophilus, to be restricted to pathogenic strains (24). Strain 86-028NP also has the gene encoding an immunoglobulin A protease (NTHI1164) (86), and as noted above, the gene encoding the Hap adhesin. Both are proteins of the autotransporter class. As described above, the HMW adhesins are members of the two-partner secretion pathway group of proteins.

Outer membrane proteins.

A number of outer membrane protein (OMP)-encoding genes have been identified by homology to those in other Haemophilus isolates. These include the major OMPs that were all originally identified in H. influenzae type b; the surface-expressed P1 (NTHI0522), the porin P2 (NTHI0225), the phosphomonoesterase and heme transporter P4 (NTHI0816), the adhesin P5 (NTHI1332) and the lipoprotein P6 (NTHI0501) (70, 71, 73, 88-90). Strain 86-028NP also shares a number of minor OMPs with other Haemophilus strains. These include D15 and the transferrin binding proteins from H. influenzae type b as well as a homologue of OMP26, which was identified in NTHi strain 289 (29, 38, 39, 112). All have subsequently been characterized in NTHi strains and analyzed as potential vaccine candidates (4, 22, 65, 74, 85).

Restriction enzymes systems.

Strain 86-028NP lacks the HindII and HindIII type II restriction systems (34, 80, 104). In contrast, genes encoding the HaeII system that was originally identified in H. aegyptius (103) are present in the strain 86-028NP genome but absent in strain Rd. Both strain 86-028NP and strain Rd have Hsd type I restriction systems encoding a methyltransferase (HsdM), a sequence recognition protein (HsdS), and a restriction enzyme (HsdR) (93). These genes are adjacent in the strain Rd genome (HI1285-HI1287). The 86-028NP genome contains three hsd-like loci that each contain four genes. One hsd system is encoded by NTHI1838 to NTHI1843. In this gene cluster, NTHI1841 encodes a hypothetical protein. A second hsd-like locus is encoded by NTHI0314 to NTHI0318. In this gene cluster, NTHI0316 encodes a putative anticodon nuclease. This hsd-like system may be similar to the prr system in E. coli (114). A third hsd locus is encoded by NTHI0188 to NTHI0193. In this gene cluster, NTHI0190 encodes a predicted transcriptional regulator with a helix-turn-helix domain.

Summary.

We have sequenced and annotated the genome of strain 86-028NP, a clinical isolate of a nontypeable H. influenzae strain isolated from a child with chronic otitis media. The genome is approximately 1.91 kb in size, slightly larger than the strain Rd genome. We identified a number of regions of gross genome rearrangement relative to the strain Rd genome as well as a number of genes specific to strain 86-028NP. These data will be instrumental to the long-term goal of increasing our understanding of the pathogenesis of diseases caused by NTHi.