Five M. haemophilum reference strains (all clinical isolates) were available for 16S and internal transcribed spacer (ITS) sequencing. Three strains were provided by the National Institute for Public Health and the Environment and 2 were provided by the Institute for Tropical Medicine (Antwerp, Belgium). The 25 mycobacterial strains used for specificity testing included M. tuberculosis complex, M. kansasii, M. xenopi, M. avium, M. intracellulare, M. gordonae, M. chelonae, M. fortuitum, M. marinum, M. scrofulaceum, and M. malmoense. A complete list of all strains (species and subspecies) has been published in Bruijnesteijn et al. (8). The strains were cultured in liquid Dubos medium at 35°C. The M. haemophilum strains were cultured at 30°C on solid Löwenstein-Jensen (LJ) medium with added iron citrate or in liquid Mycobacteria Growth Index Tube (MGIT) medium with X-factor-strip added (Becton-Dickinson, Alphen a/d Rijn, the Netherlands).

Clinical materials were decontaminated with a Nalc-NaOH decontamination protocol (15). Auramine staining was performed on the decontaminated materials for detection of acid-fast rods. Standard mycobacterial culturing was performed at 35°C in liquid MGIT medium and on solid LJ medium. M. haemophilum–specific culturing was performed at 30°C on LJ medium with added iron citrate and in MGIT medium with X-factor-strip added. Mycobacterial species were identified by using the Inno-Lipa and more recently using the Inno-Lipa V2 assay (InnoGenetics, Gent, Belgium) (16). When no growth was detected after 12 weeks of incubation, the culture results were listed as negative. Samples were also investigated for the presence of other bacterial pathogens by conventional bacterial cultures and by PCR for B. henselae (17).

All clinical materials were processed as described in Bruijnesteijn et al. (8). DNA was extracted from bacterial strains and clinical materials according to the method of Boom et al. (18) with an overnight incubation with proteinase K.

Genus-specific primers for sequencing the total ITS region of mycobacteria were described by Frothingham et al. (19). Primers described previously for a genus specific real-time PCR (8) were also used for sequencing a part of the ITS region directly from clinical materials. Using these primer sets combined, we applied a seminested PCR approach to increase the amount of amplicon. The part of the ITS region used in this real-time PCR shows considerable variation between mycobacterial species (20) (Figure). The primers used in the real-time PCR are genus-specific, and for the design of the M. haemophilum–specific minor groove binding (MGB) probe, the intraspecies and interspecies variation in the amplified ITS region was investigated. Alignments were made of the sequences of the M. haemophilum strains and of different mycobacterial species. The Mycobacterium genus–specific probe is described in Bruijnesteijn et al. (8).

The M. haemophilum–specific probe sequence was checked by using the primer 3 program (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi/) (21) and oligo-analyzer 3.0 (http://biotools.idtdna.com/analyzer/) (IDT Biotools, Coralville, IA), to ensure minimal self-complementary and to prevent the presence of secondary structures. By using the unique features of the MGB group (22), a short and highly specific probe could be designed. The probe was designed on the antisense strand to ensure an A/T rich MGB-site. An NCBI BLAST search was performed to check the specificity of the probe. The primers were prepared by Biolegio (Biolegio, Malden, the Netherlands), and the MGB probe was generated by ABI (Applied Biosystems Inc, Nieuwekerk a/d IJssel, The Netherlands). The broad range primers P1 and P4 were used for 16S sequencing. Primers and probes are listed in Table 1, and their positioning in the genome is illustrated in the Figure.

A plasmid with the ITS equence of M. haemophilum was prepared by cloning the PCR product in a vector and was subsequently quantified (IQ corporation, Groningen, the Netherlands). Dilution series of this plasmid were tested in duplicate in the genus-specific and the M. haemophilum–specific real-time PCR.

After amplification, PCR products were subjected to a cycle sequencing reaction with the Big Dye Terminator Cycle Sequencing ready reaction kit (Applied Biosystems). Samples underwent electrophoresis and sequences were detected and analyzed on ABI model 310 DNA sequencer (Applied Biosystems).

Real-time PCR was performed in 50 μL of reaction mixture consisting of 25 μL of 2x IQ supermix (Bio-Rad, Veenendaal, the Netherlands), 20 pmol of each primer, 12.5 pmol of the genus-specific probe or 10 pmol of the M. haemophilum–specific probe, and 10 μL template. The PCR thermal profile consisted of an initial incubation of 3 min at 95°C for activation of the enzyme, followed by 50 cycles of 30 s at 95°C, 40 s at 55°C, and 30 s at 72°C. Amplification, detection, and data analysis were performed with an iCycler IQ real-time detection system (Bio-Rad). The reaction mix and PCR profile were similar for both the genus-specific probe and the M. haemophilum probe.

Each DNA extract was tested by real-time PCR for the detection of the genus Mycobacterium and species M. haemophilum. As positive control for the genus-specific real-time PCR and extraction protocol, a dilution of M. bovis was used.

In 16 (18%) of 89 patients from the CHIMED study, a mycobacterial infection was suspected, but initially no species identification could be established. After a positive signal was generated by the genus-specific real-time PCR and negative results from the cultures, the amplicons generated in real-time PCR were sequenced to determine the species. Sequencing of the ITS fragment formed in the real-time PCR was difficult, owing to the small amount of amplicon, but eventually the sequences of 4 patient samples were successfully derived. On 4 more samples, a seminested PCR was performed to increase the amount of specific amplicon. This enhancement of PCR resulted in the successful amplification and sequencing of all fragments. No variation was encountered between the ITS sequences of all 8 strains analyzed here. Because no M. haemophilum ITS sequences were available in the public databases, 3 complete ITS sequences from M. haemophilum strains isolated from different CHIMED patients were determined and submitted to the NCBI database (accession numbers AY579398, AY579399, and AY579400). After specific culturing, the identity of the strains was confirmed by comparing partial 16S-gene sequences to the sequences in the NCBI (http://www.ncbi.nlm.nih.gov/) and the RIDOM database (http://www.ridom.com/). A variable part of the 16S gene of 330 base pairs was analyzed, and a 100% agreement was obtained with 16S sequences of 7 available M. haemophilum strains, including ATCC 29548. The M. haemophilum strains had at least 4 mismatches in the analyzed 16S PCR fragment in comparison to other mycobacterial species; therefore, all these strains were M. haemophilum. The identity was also confirmed because of a minimum of 4 mismatches in the 16S fragment between the M. haemophilum sequence and other mycobacterial species.

The real-time PCR was designed to the ITS region. The same conserved primers were used as described previously. The obtained ITS sequences were used to select an M. haemophilum–specific probe.

The detection limit of the M. haemophilum–specific real-time PCR was assessed at 1 copy per reaction by using a dilution series of the plasmid standard. The mycobacterial genus–specific PCR was tested simultaneously with the M. haemophilum–specific PCR and resulted in the same analytical sensitivity. As determined previously, the sensitivity of the primer set in clinical materials was estimated to be 1,100 CFU in pus (8). Specificity testing of the M. haemophilum–specific real-time PCR with 25 other species and subspecies showed no aspecific reactions. All 50 Bartonella-positive samples from the control group remained negative in the real-time PCR assay as well.

Of 16 patients with evidence for M. haemophilum infection, 9 (56%) were positive on auramine staining, and 9 (56%) were positive in M. haemophilum–specific cultures. In addition, in 1 patient (6%), the pathogen was able to grow on and in normal mycobacterial cultures. Thirteen patients (81%) had positive specimens in mycobacterial genus–specific real-time PCR, 11 of which were also positive in the M. haemophilum–specific real-time PCR (Table 2). In contrast, 2 genus-specific M. haemophilum–negative specimens were positive in the M. haemophilum–specific real-time PCR. Thus, the 2 PCRs combined yielded 15 positive (94%) patients. These 4 samples with inconsistent results all had high threshold cycle values, indicating that the amount of bacterial DNA present was close to the detection limit of the assays. This finding was confirmed by retesting the samples 5 times. As expected, this resulted in 2 or 3 positive reactions for any sample that had undergone both assays. No correlation was found between the threshold cycle values in the real-time PCR assay and the culture or auramine-staining results. All 9 patients with positive specimens by auramine staining also had positive results in the real-time PCR assay. Three patients’ conditions were diagnosed by real-time PCR only. Only 1 patient had a positive culture while results of the real-time PCR or auramine staining remained negative. Real-time PCR on the isolate cultured from this patient resulted in a positive signal.

The M. haemophilum–specific culturing method was less sensitive than the real-time PCR assay. The materials from the first 6 patients were cultured specifically for M. haemophilum after negative results were obtained from conventional culturing methods. The stored decontaminated materials were thawed and incubated at 30°C on enriched media. From these 6 patients, 2 samples (33%) yielded positive cultures. The materials from the 10 other patients were cultured directly and yielded positive results from 7 (70%) patients. M. haemophilum–specific real-time PCR was performed additionally on all positive cultures to confirm the specific growth of M. haemophilum.

From the 16 patients positive for M. haemophilum, 22 samples were collected: 9 tissue biopsy specimens and 13 fine needle aspirates. Of these samples, 19 (86%) were positive in the real-time PCR assay, while 16 (73%) samples yielded positive results in auramine staining with 11 positive (50%) or 9 positive (36%) results, respectively. No discrepancies were encountered in the real-time PCR assay when all samples instead of patients were considered. Application of the real-time PCR assay increased the diagnostic yield by 23%.