Identification of B. holmesii Cases in Massachusetts

Culture-confirmed B. holmesii cases identified during 2005–2009 by the State Laboratory Institute at the MDPH were included in our analysis. According to MDPH guidelines, a nasopharyngeal swab was cultured if the patient was <11 years of age or had a cough for <14 days. For all other patients (>11 years of age and >14 days of cough), a serum test was performed. Details on culturing methods and Bordetella spp. identification tests performed have been described (21). A total of 41 B. holmesii infections were reported; the case records, including symptomology for 26 of these, are maintained in the Massachusetts Virtual Epidemiologic Network.

Bacterial Strains and Growth

B. pertussis strain 536 (24) and B. parapertussis strain CN2591 (25) have been described. B. holmesii strain P3421 was isolated in Massachusetts and used for animal experiments. Bacteria were maintained on Bordet-Gengou agar (Difco, Sparks, MD, USA) supplemented with 10% sheep’s blood (Hema Resources, Aurora, OR, USA) without antimicrobial drugs (B. holmesii) or with 20 µg/mL streptomycin (Sigma-Aldrich, St. Louis, MO, USA) (B. pertussis or B. parapertussis). Liquid cultures were grown overnight in Stainer-Scholte broth at 37°C to mid-log phase (26,27).

Phylogenetic Analysis

Phylogenetic analyses were performed on the basis of atpD, rpoB, tuf, and rnpB gene sequences as described (16). Gene amplifications were completed on 30 B. holmesii isolates obtained from MDPH or the CDC, B. pertussis strain 536, B. parapertussis strain 2591, B. bronchiseptica strain RB50, B. avium strain 197N, and B. hinzii strain BC304. Concatenated sequences were aligned and used to construct unweighted pair group method using average linkages trees in MEGA4 software (www.megasoftware.net/mega4/mega.html).

Animal Experiments

All protocols were approved by Institutional Animal Care and Use Committee (IACUC). All animals were C57BL/6 mice and were handled in accordance with institutional guidelines (IACUC approval no. 31297). Animal experiments were performed as described, with 4 mice per group, and were performed in replicate (28–33). For vaccinations, sedated 4–6-week-old mice were vaccinated by intraperitoneal injection of 10⁸ CFU of heat-inactivated (65°C for 30 min) bacteria in 200 µL of phosphate-buffered saline (PBS; Omnipur, Gibbstown, NJ, USA) for whole-cell B. holmesii vaccine (wH) or wP B. pertussis vaccine; one fifth human dose of Adacel (Sanofi-Pasteur, Swiftwater, PA, USA) (0.5 µg PT, 1 µg FHA, 0.6 µg pertactin, 5 µg fimbriae 2 and 3 per mouse) with Imject Alum (Thermo Scientific) (aP); or only Imject Alum in 200 µL PBS on days 14 and 28 before challenge (28,33).

For challenge, 50 µL PBS containing 5 × 10⁵ CFU of B. pertussis or B. parapertussis or 10⁷ CFU of B. holmesii was added by pipetting onto the external nares of sedated mice (29). A larger inoculum of B. holmesii was used to achieve reproducibility and detect B. holmesii from the respiratory tract at later time points, because it is cleared more rapidly from the lower respiratory tract than are B. pertussis or B. parapertussis. For adoptive transfer of serum antibodies, mice were vaccinated with the indicated bacteria on days 0 and 14, and serum samples were collected on day 28 from vaccinated or naive animals. Serum samples of 200 µL were intraperitoneally injected immediately before mice were inoculated with 5 × 10⁶ CFU of B. holmesii (30,32). Bacterial numbers were quantified as described (30).

Splenocyte Restimulations

Splenocytes were isolated from vaccinated mice as described (31,33) and stimulated with either media alone or media containing 10⁷ CFU (multiplicities of infection of 5) of the indicated heat-killed bacteria (28,31). After 3 days, the supernatants were collected and analyzed for interferon (IFN) γ and interleukin-10 (IL-10) production by using ELISA (R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s instructions.

Titer ELISAs

Antibody titers were determined as described (34–36). In brief, wP- or wH-induced/naive serum samples (1:200 dilution) or aP/adjuvant-induced serum samples (1:50 dilution) from each individual mouse were added to and serially diluted 1:2 across plates coated with heat-inactivated exponential-phase bacteria. After incubation, samples were probed with 1:4,000 dilution of goat anti-mouse Ig horseradish peroxidase–conjugated antibodies (Southern Biotech, Birmingham, AL, USA). Titers were determined by using the endpoint method (33).

Western Blot Analysis

Lysates containing 10⁷ CFU of indicated heat-killed bacteria were subjected to 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis under denaturing conditions. Polyvinylidene fluoride membranes (Millipore, Billerica, MA, USA) were probed overnight with either naive serum (1:100 dilution) or wH- (1:500 dilution), wP- (1:500 dilution), aP- (1:100 dilution) induced serum. A 1:10,000 dilution of goat anti-mouse Ig horseradish peroxidase–conjugated antibodies was used as the detector antibody (35,37). The membrane was visualized with ECL Western Blotting Detection Reagent (Pierce Biotechnology, Rockford, IL, USA).

Statistical Analysis

Mean ± SE values were determined for all appropriate data. Two-tailed, unpaired Student t tests, analysis of variance and Tukey’s simultaneous test in Minitab (www.minitab.com) with similar significance were used to determine statistical significance between groups.

B. holmesii Endemicity in Massachusetts

Figure 1

Figure 1. . . . Bordetella holmesii cases in Massachusetts, USA. A) Nasopharyngeal specimens culture-positive for B. holmesii infection as confirmed by the Massachusetts Department of Public Health, by year, 2005–2009. B) Age...

In 1999, Yih et al. reported an increase in culture-positive B. holmesii cases from 1995 to 1998 (0.2% to 0.6%) (23). Here, collaborating with the same MDPH research team, we report the numbers of B. holmesii culture-positive nasopharyngeal specimens submitted to the MDPH during 2005–2009. Over these 5 years, B. holmesii was isolated from the nasopharyngeal swabs of 41 patients who had similar respiratory symptoms, which is 8 more total cases than observed by Yih et al. during 1994–1998 (33 total cases) (23). At least 2 isolates were recovered each year, and 17 cases were identified in 2006, the highest number observed (Figure 1, panel A). The rate of B. holmesii–positive nasopharyngeal swabs ranged from 0.1% to 0.4%, in line with previous results (Table). Similar to observations made in prior years, 71% of cases occurred in persons 10–19 years of age (Figure 1, panel B), compared with >80% of cases during 1994–1998.

Symptom documentation was obtained for 26 of the 41 cases. All 26 of these patients had a cough; 17 (65%) had a paroxysmal cough, 6 (23%) had post-tussive vomiting, and 4 (15%) had an inspiratory whoop. Nineteen patients (73%) exhibited >1 of these classic symptoms of whooping cough and met the World Health Organization clinical case definition for pertussis (www.who.int/immunization_monitoring/diseases/pertussis_surveillance/en/index.html). No data were collected regarding any previous underlying diseases or potential co-infections among these patients. Although we have no evidence that B. holmesii is the causative agent of these pertussis-like illness, these data suggest that B. holmesii is consistently present in the nasopharynx of a small number of patients who have respiratory infections in Massachusetts.

Phylogenetic Relationships among B. holmesii Isolates

Figure 2

Figure 2. . . . Phylogenetic tree showing 30 Bordatella holmesii isolates, B. pertussis 536, B. parapertussis 2591, B. bronchiseptica RB50, B. avium 197N, and B. hinzii BC304. Tree was constructed on the...

To evaluate the phylogenetic relationships among B. holmesii isolates, an unweighted pair group method using average linkages tree was constructed on the basis of concatenated nucleotide sequences amplified from regions of atpD, rpoB, tuf, and rnpB genes (16). Twelve isolates were obtained from CDC (designated with the letter ‘G’), while the remaining isolates were obtained from the MDPH. Consistent with previous findings (16), all the B. holmesii isolates tested were more closely related to B. hinzii and B. avium than to the classical bordetellae (Figure 2). Although single-nucleotide polymorphisms exist among B. holmesii isolates, pairwise comparisons among the 10 nasopharyngeal isolates from Massachusetts showed >99% sequence identity among 2,958 aligned bases, indicating their close relatedness. By comparison, the CDC isolates were more diverse, with the lowest percent sequence identity 97.5% between CDC strains G7851 and G4363. Moreover, B. holmesii isolates from blood and nasopharyngeal specimens do not cluster separately, which suggests that evolutionary relationship among B. holmesii isolates is not associated with the anatomic site of isolation.

B. holmesii Susceptibility to B. pertussis Vaccine-induced Immunity

Most records for the identified B. holmesii culture-positive cases did not include information regarding vaccination history; however, 16 patients that were culture positive for B. holmesii received >3 doses of pertussis vaccine. The 2010 coverage among children in Massachusetts for DTaP (diphtheria toxoid–tetanus toxoid–acellular pertussis vaccine) 4-dose vaccine was estimated to be >91%, ranking third among the 50 United States (www.cdc.gov/vaccines/stats-surv/imz-coverage.htm). However, adolescents and adults have lower B. pertussis vaccine coverage and waning immunity against B. pertussis. More adolescents and adults compared with children were culture-positive for B. holmesii in Massachusetts, which suggests that B. pertussis vaccine may confer some level of protection against B. holmesii in recently vaccinated persons, similar to the cross-protection against B. bronchiseptica in a murine model of infection (38). Although DTaP was not designed to prevent B. holmesii infections, it is critical to evaluate whether B. pertussis vaccines confer cross-protection against B. holmesii.

Figure 3

Figure 3. . . . Results of testing of Bordatella pertussis and B. holmesii vaccines and protection against B. holmesii in mice. A) Mice vaccinated with whole-cell pertussis vaccine (wP) versus naive mice;...

To test whether B. pertussis vaccines provide cross-protection against B. holmesii, we vaccinated C57BL/6 mice with wP or aP. These mice or vaccine-naive mice were then challenged with B. holmesii or B. pertussis and euthanized 3 days later. Because B. holmesii colonization efficiency in the murine respiratory tract is low compared with the classical bordetellae, possibly because of the decreased attachment to mouse respiratory epithelium (A.T. Karanikas and E. T. Harvill, unpub. data), a higher challenge dose was used than for B. pertussis. wP vaccination reduced B. pertussis numbers in the lungs by >99.99% compared with naive mice; however, wP failed to reduce B. holmesii numbers (Figure 3, panel A). In fact, wP vaccination appeared to increase B. holmesii numbers in the lungs compared with naive mice. Although aP-vaccinated mice also reduced B. pertussis numbers in their lungs by ≈98% (Figure 3, panel B), they were not capable of reducing B. holmesii numbers. Together, these data indicate that wP- or aP-induced immunity does not protect against B. holmesii infections.

To determine whether B. holmesii immunization induces protection against itself or cross-protection against B. pertussis and/or B. parapertussis, we vaccinated C57BL/6 mice with heat-killed B. holmesii (wH) and challenged these or naive mice with B. pertussis, B. parapertussis, or B. holmesii; the mice were then euthanized on day 3 postinoculation for bacterial quantification. Similar numbers of B. pertussis were recovered from the lungs of naive and wH-vaccinated mice (Figure 3, panel C), an indication that B. holmesii vaccination failed to reduce B. pertussis numbers within 3 days. However, B. holmesii vaccination reduced the B. parapertussis load by ≈70% in the lungs, which indicates a modest cross-protection provided by B. holmesii against B. parapertussis. Compared with naive mice, mice who received B. holmesii vaccination had a ≈97% reduction of B. holmesii in the lungs (Figure 3, panel A). This experiment indicates that B. holmesii antigens induce efficient protective immunity against B. holmesii but have little effect against B. parapertussis and are even less protective against B. pertussis.

T-cell Responses to wH and wP/aP for B. holmesii

Because wP and aP fail to reduce B. holmesii numbers, we hypothesized that the pertussis vaccines may induce a different T-cell response than the wH vaccine does. To determine whether T-cell responses after vaccination are cross-reactive, splenocytes from naive or wH-, wP-, or aP-vaccinated mice were stimulated with media or heat-killed B. pertussis, B. holmesii, or B. parapertussis for 72 hours. Cell culture supernatant cytokine concentrations of IFN-γ and IL-10, representing Th1 and Treg cytokines, respectively, were determined. wH vaccination did not induce IFN-γ and IL-10 production by splenocytes at greater levels than those produced by naive splenocytes (Figure 4). However, splenocytes from wP-vaccinated mice produced high levels of IFN-γ and IL-10 on stimulation with heat-killed B. pertussis, B. parapertussis, or B. holmesii, which indicates that wP induces a strong cross-reactive T-cell response to B. holmesii. Although splenocytes from aP-vaccinated mice produced little IFN-γ, they produced ≈2,500 pg/mL IL-10 on stimulation with heat-killed B. pertussis, B. holmesii, or B. parapertussis, which indicates cross-reactive T-cell responses following aP vaccination. These data indicate that B. pertussis vaccines can induce cross-reactive T-cell responses to B. holmesii and thus do not explain the lack of cross-protection between these species.

wP/aP-induced Antibodies for B. holmesii

Figure 5

Figure 5. . . . Antibody responses to whole-cell pertussis vaccine (wP), aceullar pertussis vaccine (aP), and whole-cell Bordetella holmesii vaccine (wH). A) Specific Ig titers of serum antibodies for B. pertussis (B.p.),...

Njamkepo et al. observed that proteins in B. holmesii cell lysates are not recognized by antibodies specific for B. pertussis vaccine antigens (14). To determine whether B. holmesii and B. pertussis vaccine–induced antibodies are cross-reactive, Ig titers of wH-, wP-, or aP-induced serum samples were examined by using ELISA with heat-inactivated bacteria as antigens. The B. holmesii–specific Ig titer of wH-induced serum was >450,000, which is ≈10-fold and 25-fold higher than B. parapertussis– and B. pertussis–specific serum titers, respectively (Figure 5, panel A); this result indicates partial cross-reactivity. B. pertussis–specific Ig titers of wP- and aP-induced serum antibodies were 290,000 and 2,500, respectively. wP- and aP-induced antibodies bind less well to B. parapertussis (Figure 5, panel A) and confer little protection against B. parapertussis in vivo (33). wP- or aP-induced antibodies bound even less well to B. holmesii. A similar trend was observed when live bacteria were used to coat the ELISA plates (data not shown), which rules out the possibility that heat inactivation selectively destroys cross-reactive antigens.

To compare the antigens recognized by serum samples from different groups, Western blot analyses were performed on B. pertussis, B. parapertussis, or B. holmesii lysates probed with naive serum or wH-, wP-, or aP-induced serum. wH-induced serum antibodies recognized B. holmesii antigens of various molecular masses but lacked cross-recognition of some higher molecular mass B. pertussis and B. parapertussis antigens (Figure 5, panel B). The wP- and aP-induced serum antibodies bound less efficiently to B. parapertussis antigens than to B. pertussis antigens, consistent with published data (33). Although wP-induced antibodies recognized some B. holmesii antigens, they lacked recognition of higher molecular mass (>60 kDa) B. holmesii antigens. aP-induced antibodies only poorly recognized a single B. holmesii antigen. Together, these data suggest that antibodies generated following B. pertussis vaccination do not efficiently recognize B. holmesii antigens.

B. holmesii–specific Antibodies and wP-induced Immunity against B. holmesii.

On the basis of our data, we discerned that wP induces sufficient T-cell responses but that antibody responses are not sufficient to confer cross-protection against B. holmesii. If the lack of antibody cross-recognition is the only reason wP and aP vaccines are ineffective against B. holmesii infection, then adding B. holmesii–specific antibodies to wP should render the vaccination effective against B. holmesii.

Figure 6

Figure 6. . . . Supplementation of whole-cell pertussis vaccine (wP) with B. holmesii- but not B. pertussis- specific antibodies. Groups of four wP-vaccinated C57BL/6 mice were left untreated (none) or treated with...

To test this hypothesis, C57BL/6 mice were wP vaccinated or left untreated and later received either naive serum or wP- or wH-vaccinated mouse serum before B. holmesii challenge. B. holmesii lung colonization was determined 3 days postinoculation. Neither naive serum nor wP-induced serum reduced B. holmesii numbers, but vaccinated mice that received B. holmesii–immune serum had substantially lowered B. holmesii numbers in the lungs (Figure 6). The finding indicates that the addition of B. holmesii–specific antibodies to wP increases its efficacy against B. holmesii.