An Invasive Haemophilus haemolyticus Isolate

LETTER

Haemophilus haemolyticus is generally considered a human commensal and is found in the nasopharynx of a minority of individuals as well as in subgingival plaque (7). Most authors consider H. haemolyticus to be either nonpathogenic (2, 4, 7, 18) or a rare pathogen (1). The latter assessment appears to be based on two reports of endocarditis from the early part of the 20th century (3, 13); however, in neither case were requirements for the X and V factors reported, and thus, no species designation can be reliably made.

The only method of differentiation between Haemophilus influenzae and H. haemolyticus in clinical microbiology laboratories is based on the ability of H. haemolyticus to produce a zone of clear hemolysis on ovine, bovine, or equine blood agar plates (1, 6, 7, 9). However, the hemolytic activity may be lost on subculture (7), and it was recently shown that many isolates of H. haemolyticus are nonhemolytic even on primary isolation (11, 14).

Invasive H. influenzae disease is a reportable condition in Oklahoma, and all H. influenzae isolates from normally sterile sites are required to be submitted to the Oklahoma State Department of Health (OSDH) Public Health Laboratory (PHL) for confirmation and serotyping. For use in our ongoing studies of H. influenzae iron/heme acquisition, we obtained from the OSDH PHL all invasive H. influenzae isolates from 2003 and 2004. Seven of these isolates were subjected to multilocus sequence typing (MLST) as described by Meats et al. (12). We were consistently unable to amplify the gene fragment for the fucK gene from one strain, designated HI2028 and isolated from the blood of a 1-year-old male. Failure to amplify the fucK gene fragment from presumptive H. influenzae isolates has been considered an indicator of a misidentified strain (15). However, failure to detect the fucK gene cannot be considered conclusive since some strains of H. influenzae have recently been shown to lack the fucose operon (17, 19).

Based on the failure to amplify the fucK gene product, we considered that strain HI2028 may have been misidentified as H. influenzae and thus we proceeded to amplify 1,491 bp of the 16S rRNA gene from HI2028 using the primers UFPL and URPL (10). PCR amplicons were directly sequenced and yielded 1,448 bp of the 16S rRNA gene. Homology searches against the nucleotide collection using the MEGABLAST algorithm (http://blast.ncbi.nlm.nih.gov) with the determined 1,448 bp of the 16S rRNA gene from HI2028 indicated that the strain was H. haemolyticus but cannot be considered a definitive identification of the isolate as such.

A multilocus sequence analysis system based on 5 gene loci (adk, pgi, recA, infB, and 16S rRNA) has been described to separate H. haemolyticus from H. influenzae (11, 18). The sequences of these genes from HI2028 were determined and concatenated as described by McCrea et al. (11). The partial gene sequences from the 200 Haemophilus strains in the original study (11) were retrieved and concatenated. A bootstrap consensus tree using 1,000 replicates was constructed, rooted to Escherichia coli, using the neighbor-joining method in MEGA 5.05 (20). The strain HI2028 clustered with strains designated H. haemolyticus (data not shown). Additionally, the 5 gene sequences were retrieved for 14 available H. influenzae genomic sequences, for 5 recently published H. haemolyticus genomic sequences (5), and for the H. haemolyticus type strain ATCC 33390. Figure 1 shows strain HI2028 clustering with the H. haemolyticus strains in a neighbor-joining tree generated as described above. Additional characteristics of HI2028 are shown in Table 1.

Fig 1.

Neighbor-joining dendrogram of Haemophilus strains. The tree, rooted to E. coli strain MG1655, is based on concatenated adk, pgi, recA, infB, and 16S rRNA gene sequences, with bootstrap values of greater than 50% of 1,000 bootstraps indicated. Strains are genome-sequenced H. influenzae strains unless indicated otherwise. Hh, H. haemolyticus strains, including the type strain ATCC 33390 and five recently sequenced isolates (5); Hp, H. parainfluenzae strains. HI2028 is the strain identified in this report.

Table 1.

Characteristics of strain HI2028 in comparison to H. haemolyticus, H. influenzae, and the Biotype IV cryptic genospecies of H. influenzae

Characteristic	HI2028	H. haemolyticus	H. influenzae	Biotype IV cryptic genospeciesa
X factor required	+	+	+	+
V factor required	+	+	+	+
H2s productionb	+	+	V	V
Indole productionb	+	Vc	V	−
Urease activityb	+	+	V	+
Orinthineb decarboxylase activity	−	−	V	+
Hemolysisd	−	V	−	−
Protein P6 amino acids at residues 33, 42, 59, 61, and 152e	G, S, N, E, A	G, S, N, E, A	A, A, D, T, A	G, S, N, E, S

^aThe cryptic Biotype IV genospecies causes urogenital and neonatal infections (16) and is closely related to H. haemolyticus (14).

^bPhenotype was determined as described by Kilian (6).

^cV, results are variable among strains (7).

^dLack of hemolysis with strain HI2028 was determined on 5% defibrinated horse blood using BBL haemophilus identification quad plates (BD).

^eThe amino acids at specific residues within the P6 protein have been used to differentiate H. influenzae and H. haemolyticus (14).

In summary, we have identified a strain of nonhemolytic H. haemolyticus, originally designated H. influenzae, from a patient with bacteremia. In reporting the H. haemolyticus genomic sequence, Jordan et al. stated that several presumptive H. influenzae isolates from invasive disease had been confirmed as H. haemolyticus, although no data were presented (5). Our finding supports the assertion of Jordan et al. and suggests that H. haemolyticus may occasionally be found among presumed H. influenzae clinical isolates and should be considered a more frequent cause of invasive disease than currently thought. Such misidentification of H. haemolyticus strains as H. influenzae may potentially lead to errors in national burden of invasive H. influenzae disease estimates as is the case with other clinical microbiology laboratory errors (8).

Nucleotide sequence accession numbers.

The partial gene sequences for adk, pgi, recA, infB, and the 16S rRNA from HI2028 have been deposited in GeneBank with the respective accession numbers JQ254883 to JQ254887. The GeneBank accession number for the gene encoding the P6 protein of HI2028 is JQ247079.