Failure to detect capsule gene bexA in Haemophilus influenzae types e and f by real-time PCR due to sequence variation within probe binding sites

Haemophilus influenzae can be typed into six capsular types (a to f) and non-typable, non-capsulate (NC) strains. The bexA gene, which is involved in exportation of capsular material, is conserved in all six capsular types (Kroll et al., 1988, 1989, 1990). Gel-based PCR detection of bexA, distinguishing capsulate from NC strains, and detection of capsule type-specific targets is the gold standard by which surveillance of trends in invasive H. influenzae types is achieved (Falla et al., 1994; Gonin et al., 2000; Slack et al., 1998).

Real-time PCR is increasingly used as an alternative to conventional PCR as it offers several advantages. It is more rapid and enables specific product confirmation, using a closed system that reduces the number of manipulations and risk of contamination.

While evaluating a real-time PCR assay for detection of bexA in H. influenzae strains, we noticed that, unlike strains of H. influenzae types a to d, H. influenzae types e (HIE) and f (HIF) failed to produce detectable signals with bexA probes. The amplicons of 12 capsulate H. influenzae strains were sequenced to investigate these findings.

Bacterial isolates.
The 20 cultures used to test the assay were clinical isolates, except for three strains from the National Collection of Type Cultures (Colindale, UK): H. influenzae types a (n = 1), b (n = 3, including NCTC 8467^T), c (n = 2), d (n = 2), e (n = 2) and f (n = 2); capsule-deficient type b (n = 2) and NC strains (n = 2); Haemophilus aphrophilus (n = 2), Haemophilus haemolyticus (n = 1, NCTC 10659^T) and Neisseria meningitidis (n = 1, NCTC 10026^T). DNA was extracted from a heavy sweep using the Easy-DNA kit (Invitrogen) according to the manufacturer's instructions.

Detection of bexA by real-time PCR.
Previously published primers bexAF (5'-CGTTTGTATGATGTTGATCCAGA) and bexAR (5'-TGTCCATGTCTTCAAAATGATG-3') target a 343 bp region within the bexA gene (van Ketel et al., 1990). These were used with newly designed adjacent hybridization probes bexA^abcdD (donor, 5'-GAGAAACGCAAAGACCGTTCT-fluorescein-3') and bexA^abcdA (acceptor, 5'-LC Red 640-TCATTTTAGTTTCACATAGCCCG-phosphate-3'). A LightCycler (Roche Diagnostics) was used, with reaction mixtures containing 2 µl LightCycler DNA Master Hybridization Probes (Roche), 3 mM MgCl₂, 10 pM each primer, 3 pM donor probe, 2 pM acceptor probe, 2 µl DNA extract and PCR-grade water to a final volume of 20 µl. All temperature transition rates were 20 °C s¹. Reaction conditions were 95 °C for 0 s, followed by 50 cycles of heating to 95 °C for 0 s, 52 °C for 2 s and 72 °C for 10 s. Melting curve analysis was carried out at 95 °C for 0 s, 43 °C for 5 s and 95 °C for 0 s at a transition rate of 0.2 °C s¹. Accumulation of specific PCR product was detected by measuring fluorescence in real time and analysed with LightCycler software. The amplicons were also subjected to agarose gel electrophoresis.

Sequencing of bexA amplicons.
Amplicons were purified using the QIAquick PCR Purification kit (Qiagen), according to manufacturer's instructions. The reaction mixture, comprising 1 µl Thermo Sequenase DYEnamic Direct Cycle Sequencing kit (Amersham Pharmacia Biotech), 1 pmol Cy5-labelled forward primer bexAF, 0.5 pmol Cy5.5-labelled reverse primer bexAR and 3 µl purified DNA, was processed in a Gene Amp PCR System 9700 thermal cycler (Applied Biosystems), using the following conditions: 94 °C for 5 min, followed by 25 cycles of 94 °C for 30 s, 52 °C for 30 s and 70 °C for 60 s. Sequencing was carried out on a Long Read Tower (Visible Genetics) under standard conditions and analysed with OpenGene System software (Visible Genetics).

On melting-curve analysis, the bexA-specific PCR amplicon produced a melting temperature of 62.5 °C. All 12 capsulate H. influenzae isolates produced PCR products of 343 bp as shown by agarose gel electrophoresis, but the characteristic melting curve was seen only with isolates of H. influenzae types a to d (Fig. 1). To investigate this, the bexA amplicons of all 12 capsulate H. influenzae strains were sequenced.

(15K): Fig. 1. Melting curves for bexA PCR amplicons, using bexAabcd probes. A, H. influenzae types a to d; B, H. influenzae types e and f, NC H. influenzae and negative controls.

Comparison of bexA sequences with the control H. influenzae type b (HIB) strain (NCTC 8467^T) showed a similarity of 99.0100 % for the eight tested strains of types a to d, but this was lower at 86.186.7 % for the four strains of HIE and HIF (sequence alignment data available as supplementary data in JMM Online). The two HIE strains had identical sequences, as did the two HIF strains. No capsulate H. influenzae strain showed nucleotide differences within the primer binding sites. However, each strain of HIE and HIF showed nine base pair changes within the probe binding sites (Fig. 2). This explains why all capsulate strains produced an amplicon of the correct 343 bp size with agarose gel electrophoresis; however, the probes failed to produce detectable melting curves with bexA products of HIE and HIF due to sequence variation.

(7K): Fig. 2. Alignment of part of the bexA gene sequence of HIB, HIE and HIF, showing differences within the hybridization probe binding sites (underlined).

With this in mind, a new set of probes was designed for HIE and HIF: bexA^efD (5'-GAAAAGCGCAAAGACCGTTCC-fluorescein-3') and bexA^efA (5'-LC Red 640-TTATTTT GGTTTCACACAGTCCA-phosphate-3') and tested on all isolates. Only HIE and HIF produced melting curves, with melting temperatures of 54 °C and 49 °C, respectively (Fig. 3).

(12K): Fig. 3. Melting curves for bexA PCR amplicons, using bexAef probes. HIE, H. influenzae type e; HIF, H. influenzae type f; A, H. influenzae types a to d, NC H. influenzae and negative controls.

Earlier conventional PCR studies successfully detected bexA in capsulate H. influenzae of all six types, although less-specific gel-based systems were used to confirm the product (Falla et al., 1994; Gonin et al., 2000; van Ketel et al., 1990). Only one other study has used real-time PCR to detect H. influenzae and attributed the failure to detect bexA from types a, d, e and f to sequence variation, but did not confirm this with sequencing (Corless et al., 2001). Our findings indicate that the HIE and HIF strains used in this study had sufficient bexA sequence variation compared with types a to d to preclude detection of all six serotypes using a single set of hybridization probes.

There are few data comparing bexA sequences among different capsule types. One study applied a 50 bp bexA probe to EcoRI-digested DNA from capsulate H. influenzae strains and found that types a, c and d produced identical fragments, while bound fragments for types e and f were of a different size (Kroll et al., 1988). Genetic analysis of HIF has shown 8994 % similarity with HIB in region I of the cap locus, where bexA is located (Satola et al., 2003). Another study found as much as 16.5 % nucleotide difference between bexA sequences of two genetically divergent HIB strains (Kroll et al., 1990).

In this study, sequence variation within the bexA target region for HIE and HIF was suspected only because gel electrophoresis showed a product of the correct size, despite failure to produce a melting curve. This has implications for the surveillance of H. influenzae disease in this era of HIB vaccination, particularly if a highly specific method such as real-time PCR is used to type strains. Any capsulate H. influenzae strain with significant bexA sequence variation may be undetected and misclassified as NC. As the proportion of disease caused by NC types e and f strains has increased (Campos et al., 2004; Slack et al., 1998; Urwin et al., 1996), it is crucial that all capsule types are accurately identified.

We thank Dr Mary Slack and Suzanne Stringer of the Health Protection Agency Haemophilus Reference Unit, Oxford, UK, for providing the bacterial isolates and technical advice.