Role of antibodies in immunity to Bordetella infections

The persistence of Bordetella pertussis and B. parapertussis within vaccinated populations and the reemergence of associated disease highlight the need to better understand protective immunity. The present study examined host immunity to bordetellae and addressed potential concerns about the mouse model by using a comparative approach including the closely related mouse pathogen B. bronchiseptica. As previously observed with B. pertussis, all three organisms persisted throughout the respiratory tracts of B-cell-deficient mice, indicating that B cells are required for bacterial clearance. However, adoptively transferred antibodies rapidly cleared B. bronchiseptica but not human pathogens. These results obtained with the mouse model are consistent with human clinical observations, including the lack of correlation between antibody titers and protection, as well as the limited efficacy of intravenous immunoglobulin treatments against human disease. Together, this evidence suggests that the mouse model accurately reflects substantial differences between immunities to these organisms. Although both B. pertussis and B. parapertussis are more closely related to B. bronchiseptica than they are to each other, they share the ability to resist rapid clearance from the lower respiratory tract by adoptively transferred antibodies, an adaptation that correlates with their emergence as human pathogens that circulate within vaccinated populations.     

Multiple roles for Bordetella lipopolysaccharide molecules during respiratory tract infection

Bordetella pertussis, Bordetella parapertussis, and Bordetella bronchiseptica are closely related subspecies that cause respiratory tract infections in humans and other mammals and express many similar virulence factors. Their lipopolysaccharide (LPS) molecules differ, containing either a complex trisaccharide (B. pertussis), a trisaccharide plus an O-antigen-like repeat (B. bronchiseptica), or an altered trisaccharide plus an O-antigen-like repeat (B. parapertussis). Deletion of the wlb locus results in the loss of membrane-distal polysaccharide domains in the three subspecies of bordetellae, leaving LPS molecules consisting of lipid A and core oligosaccharide. We have used wlb deletion (Deltawlb) mutants to investigate the roles of distal LPS structures in respiratory tract infection by bordetellae. Each mutant was defective compared to its parent strain in colonization of the respiratory tracts of BALB/c mice, but the location in the respiratory tract and the time point at which defects were observed differed significantly. Although the Deltawlb mutants were much more sensitive to complement-mediated killing in vitro, they displayed similar defects in respiratory tract colonization in C5(-/-) mice compared with wild-type (wt) mice, indicating that increased sensitivity to complement-mediated lysis is not sufficient to explain the in vivo defects. B. pertussis and B. parapertussis Deltawlb mutants were also defective compared to wt strains in colonization of SCID-beige mice, indicating that the defects were not limited to interactions with adaptive immunity. Interestingly, the B. bronchiseptica Deltawlb strain was defective, compared to the wt strain, in colonization of the respiratory tracts of BALB/c mice beginning 1 week postinoculation but did not differ from the wt strain in its ability to colonize the respiratory tracts of B-cell- and T-cell-deficient mice, suggesting that wlb-dependent LPS modifications in B. bronchiseptica modulate interactions with adaptive immunity. These data show that biosynthesis of a full-length LPS molecule by these three bordetellae is essential for the expression of full virulence for mice. In addition, the data indicate that the different distal structures modifying the LPS molecules on these three closely related subspecies serve different purposes in respiratory tract infection, highlighting the diversity of functions attributable to LPS of gram-negative bacteria.     

Role of Bordetella O antigen in respiratory tract infection

Lipopolysaccharide (LPS), as the major surface molecule of gram-negative bacteria, interacts with the host in complex ways, both inducing and protecting against aspects of inflammatory and adaptive immunity. The membrane-distal repeated carbohydrate structure of LPS, the O antigen, can prevent antibody functions and may vary as a mechanism of immune evasion. Genes of the wbm locus are required for the assembly of O antigen on the animal pathogen Bordetella bronchiseptica and the human pathogen B. parapertussis. However, the important human pathogen B. pertussis lacks these genes and a number of in vitro and in vivo characteristics associated with O antigen in other organisms. To determine the specific functions of O antigen in these closely related Bordetella subspecies, we compared wbm deletion (Deltawbm) mutants of B. bronchiseptica and B. parapertussis in a variety of assays relevant to natural respiratory tract infection. Complement was not activated or depleted by wild-type bordetellae expressing O antigen, but both Deltawbm mutants activated complement and were highly sensitive to complement-mediated killing in vitro. Although the O-antigen structures appear to be substantially similar, the two mutants differed strikingly in their defects within the respiratory tract. The B. parapertussis Deltawbm mutant was severely defective in colonization of the tracheas and lungs of mice, while the B. bronchiseptica Deltawbm mutant showed almost no defect. While in vitro characteristics such as serum resistance may be attributable to O antigen directly, the role of O antigen during infection appears to be more complex, possibly involving factors differing among the closely related bordetellae or different interactions between each one and its host.     

Genetic Basis for Lipopolysaccharide O-Antigen Biosynthesis in Bordetellae

Bordetella bronchiseptica and Bordetella parapertussis express a surface polysaccharide, attached to a lipopolysaccharide, which has been called O antigen. This structure is absent from Bordetella pertussis. We report the identification of a large genetic locus in B. bronchiseptica and B. parapertussis that is required for O-antigen biosynthesis. The locus is replaced by an insertion sequence in B. pertussis, explaining the lack of O-antigen biosynthesis in this species. The DNA sequence of the B. bronchiseptica locus has been determined and the presence of 21 open reading frames has been revealed. We have ascribed putative functions to many of these open reading frames based on database searches. Mutations in the locus in B. bronchiseptica and B. parapertussis prevent O-antigen biosynthesis and provide tools for the study of the role of O antigen in infections caused by these bacteria.     

Comparative toll-like receptor 4-mediated innate host defense to Bordetella infection

Bordetella pertussis, B. parapertussis, and B. bronchiseptica are closely related species associated with respiratory disease in humans and other mammals. While B. bronchiseptica has a wide host range, B. pertussis and B. parapertussis evolved separately from a B. bronchiseptica-like progenitor to naturally infect only humans. Despite very different doubling times in vitro, all three establish similar levels of infection in the mouse lung within 72 h. Recent work has revealed separate roles for Toll-like receptor 4 (TLR4) in immunity to B. pertussis and B. bronchiseptica, while no role for TLR4 during B. parapertussis infection has been described. Here we compared the requirement for TLR4 in innate host defense to these organisms using the same mouse infection model. While B. bronchiseptica causes lethal disease in TLR4-deficient mice, B. pertussis and B. parapertussis do not. Correspondingly, TLR4 is critical in limiting B. bronchiseptica but not B. pertussis or B. parapertussis bacterial numbers during the first 72 h. Interestingly, B. bronchiseptica induces a TLR4-dependent cytokine response that is considerably larger than that induced by B. pertussis or B. parapertussis. Analysis of their endotoxins using RAW cells suggests that B. bronchiseptica lipopolysaccharide (LPS) is 10- and 100-fold more stimulatory than B. pertussis or B. parapertussis LPS, respectively. The difference in LPS stimulus is more pronounced when using HEK293 cells expressing human TLR4. Thus, it appears that in adapting to infect humans, B. pertussis and B. parapertussis independently modified their LPS to reduce TLR4-mediated responses, which may compensate for slower growth rates and facilitate host colonization.     

Role of systemic and mucosal immune responses in reciprocal protection against Bordetella pertussis and Bordetella parapertussis in a murine model of respiratory infection

The roles of systemic humoral immunity, cell-mediated immunity, and mucosal immunity in reciprocal protective immunity against Bordetella pertussis and Bordetella parapertussis were examined by using a murine model of respiratory infection. Passive immunization with serum from mice infected with B. pertussis established protective immunity against B. pertussis but not against B. parapertussis. Protection against B. parapertussis was induced in mice that had been injected with serum from mice infected with B. parapertussis but not from mice infected with B. pertussis. Adoptive transfer of spleen cells from mice infected with B. pertussis or B. parapertussis also failed to confer reciprocal protection. To examine the role of mucosal immunity in reciprocal protection, mice were infected with preparations of either B. pertussis or B. parapertussis, each of which had been incubated with the bronchoalveolar wash of mice that were convalescing after infection with B. pertussis or B. parapertussis. Such incubation conferred reciprocal protection against B. pertussis and B. parapertussis on infected mice. The data suggest that mucosal immunity including secreted immunoglobulin A in the lungs might play an important role in reciprocal protective immunity in this murine model of respiratory infection.     

O antigen protects Bordetella parapertussis from complement

Bordetella pertussis, a causative agent of whooping cough, expresses BrkA, which confers serum resistance, but the closely related human pathogen that also causes whooping cough, Bordetella parapertussis, does not. Interestingly, B. parapertussis, but not B. pertussis, produces an O antigen, a factor shown in other models to confer serum resistance. Using a murine model of infection, we determined that O antigen contributes to the ability of B. parapertussis to colonize the respiratory tract during the first week of infection, but not thereafter. Interestingly, an O antigen-deficient strain of B. parapertussis was not defective in colonizing mice lacking the complement cascade. O antigen prevented both complement component C3 deposition on the surface and complement-mediated killing of B. parapertussis. In addition, O antigen was required for B. parapertussis to systemically spread in complement-sufficient mice, but not complement-deficient mice. These data indicate that O antigen enables B. parapertussis to efficiently colonize the lower respiratory tract by protecting against complement-mediated control and clearance.     

The Bordetella bronchiseptica type III secretion system inhibits gamma interferon production that is required for efficient antibody-mediated bacterial clearance

Several species of pathogenic microorganisms have developed strategies to survive and persist in vital organs which are normally maintained as sterile by the generation of strong immune responses. Here, we report an immunomodulation involving the Bordetella bronchiseptica type III secretion system (TTSS) which contributes to bacterial survival in the lower respiratory tract of the host. The prolonged persistence of B. bronchiseptica that was observed in gamma interferon (IFN-gamma)-/- mice indicates that the efficient clearance of bacteria from the lower respiratory tract requires not only B cells and antibodies but also IFN-gamma production. Our data also suggest that interleukin-10 (IL-10)-producing splenocytes are generated early during infection and that IL-10 inhibits IFN-gamma-producing cells and delays the clearance of B. bronchiseptica from the lungs. The TTSS of B. bronchiseptica inhibits the generation of IFN-gamma-producing splenocytes and is required for long-term bacterial persistence in the lower respiratory tract in wild-type mice. This suggests that a mechanism involving the modulation of IFN-gamma production by the TTSS facilitates B. bronchiseptica survival in the lower respiratory tract.     

Suppression of serum antibody responses by pertussis toxin after respiratory tract colonization by Bordetella pertussis and identification of an immunodominant lipoprotein

Pertussis toxin (PT), a virulence factor secreted by Bordetella pertussis, contributes to respiratory tract infection and disease caused by this pathogen. By comparing a wild-type (WT) B. pertussis strain to a mutant strain with an in-frame deletion of the ptx genes encoding PT (DeltaPT), we recently found that the lack of PT confers a significant defect in respiratory tract colonization in mice after intranasal inoculation. In this study, we analyzed serum antibody responses in mice infected with the WT or DeltaPT strain and found that infection with the DeltaPT strain elicited greater responses to several B. pertussis antigens than did infection with the WT, despite the lower colonization level achieved by the DeltaPT strain. The same enhanced antibody response was observed after infection with a strain expressing an enzymatically inactive PT; but this response was not observed after infection with B. pertussis mutant strains lacking filamentous hemagglutinin or adenylate cyclase toxin, nor when purified PT was administered with the DeltaPT inoculum, indicating a specific role for PT activity in this immunosuppressive effect. In particular, there were consistent strong serum antibody responses to one or more low-molecular-weight antigens after infection with the DeltaPT strain. These antigens were Bvg independent, membrane localized, and also expressed by the closely related pathogens Bordetella parapertussis and Bordetella bronchiseptica. Two-dimensional gel electrophoresis and mass spectrometry were used to identify one of the immunodominant low-molecular-weight antigens as a protein with significant sequence homology to peptidoglycan-associated lipoprotein in several other gram-negative bacterial species. However, a serum antibody response to this protein alone did not protect mice against respiratory tract infection by B. pertussis.     

Detection of Bordetella bronchiseptica by the polymerase chain reaction.

Polymerase chain reaction (PCR) assays were developed that enabled not only discriminative detection of three Bordetella species, B. pertussis, B. parapertussis, and B. bronchiseptica (Bspp PCR), but also specific detection of B. bronchiseptica (Bb PCR). An upstream sequence of the flagellin gene was used as a target DNA region. This sequence contained differences in B. pertussis, B. parapertussis, and B. bronchiseptica DNA. These species could then be differentiated using two different sets of primers, Bspp and Bb. When oligonucleotide Bspp primers were used, PCR products were obtained from the three species of Bordetella. A fragment of the expected size (164 bp) was amplified using B. bronchiseptica and B. parapertussis DNA, but a fragment with a distinct molecular weight was amplified with B. pertussis DNA (195 bp). This Bspp PCR was specific and sensitive, but it could not differentiate between B. parapertussis and B. bronchiseptica. When Bb primers were used, a 237-bp PCR product was detected only from B. bronchiseptica DNA. No PCR products were identified after Bb PCR amplification of DNAs either from B. parapertussis isolates or B. pertussis isolates, nor from other respiratory pathogen DNAs tested. This second PCR assay had a sensitivity limit of less than 10 organisms of B. bronchiseptica after detection with a specific probe. The specificity and the sensitivity of the fla PCR assay were evaluated with purified DNA, as was its capacity for detecting the bacteria in human clinical samples and in lungs of infected mice.     

Interleukin-1 receptor signaling is required to overcome the effects of pertussis toxin and for efficient infection- or vaccination-induced immunity against Bordetella pertussis

Interleukin-1 receptor-deficient (IL-1R(-/-)) mice are healthy despite being colonized by commensal microbes but are defective in defenses against specific pathogens, suggesting that IL-1R-mediated effects contribute to immune responses against specific pathogenic mechanisms. To better define the role of IL-1R in immunity to respiratory infections, we challenged IL-1R(-/-) mice with Bordetella pertussis and Bordetella parapertussis, the causative agents of whooping cough. Following inoculation with B. pertussis, but not B. parapertussis, IL-1R(-/-) mice showed elevated bacterial numbers and more extensive inflammatory pathology than wild-type mice. Acellular B. pertussis vaccines were not efficiently protective against B. pertussis in IL-1R(-/-) mice. B. pertussis-stimulated dendritic cells from IL-1R(-/-) mice produced higher levels of tumor necrosis factor alpha (TNF-α) and IL-6 than wild-type cells. Moreover, elevated levels of gamma interferon (IFN-γ) and TNF-α but lower levels of IL-10 were detected during B. pertussis infection in IL-1R(-/-) mice. Since B. parapertussis did not cause severe disease in IL-1R(-/-) mice, we hypothesized that the extreme requirement for IL-1R involves pertussis toxin (Ptx), which is expressed only by B. pertussis. An isogenic Ptx-deficient B. pertussis strain had only a modest phenotype in wild-type mice but was completely defective in causing lethal disease in IL-1R(-/-) mice, indicating that the particular virulence of B. pertussis in these mice requires Ptx. Ptx contributes to IL-1β induction by B. pertussis, which is involved in IL-10 induction through IL-1R signaling. IL-10 treatment reduced B. pertussis numbers in IL-1R(-/-) mice, suggesting that the lower IL-10 responses partially account for the uncontrolled inflammation and bacterial growth in these mice.     

Interactions of pulmonary collectins with Bordetella bronchiseptica and Bordetella pertussis lipopolysaccharide elucidate the structural basis of their antimicrobial activities

Surfactant proteins A (SP-A) and D (SP-D) play an important role in the innate immune defenses of the respiratory tract. SP-A binds to the lipid A region of lipopolysaccharide (LPS), and SP-D binds to the core oligosaccharide region. Both proteins induce aggregation, act as opsonins for neutrophils and macrophages, and have direct antimicrobial activity. Bordetella pertussis LPS has a branched core structure and a nonrepeating terminal trisaccharide. Bordetella bronchiseptica LPS has the same structure, but lipid A is palmitoylated and there is a repeating O-antigen polysaccharide. The ability of SP-A and SP-D to agglutinate and permeabilize wild-type and LPS mutants of B. pertussis and B. bronchiseptica was examined. Previously, wild-type B. pertussis was shown to resist the effects of SP-A; however, LPS mutants lacking the terminal trisaccharide were susceptible to SP-A. In this study, SP-A was found to aggregate and permeabilize a B. bronchiseptica mutant lacking the terminal trisaccharide, while wild-type B. bronchiseptica and mutants lacking only the palmitoyl transferase or O antigen were resistant to SP-A. Wild-type B. pertussis and B. bronchiseptica were both resistant to SP-D; however, LPS mutants of either strain lacking the terminal trisaccharide were aggregated and permeabilized by SP-D. We conclude that the terminal trisaccharide protects Bordetella species from the bactericidal functions of SP-A and SP-D. The O antigen and palmitoylated lipid A of B. bronchiseptica play no role in this resistance.     

The O antigen enables Bordetella parapertussis to avoid Bordetella pertussis-induced immunity

Bordetella pertussis and Bordetella parapertussis are closely related endemic human pathogens which cause whooping cough, a disease that is reemerging in human populations. Despite how closely related these pathogens are, their coexistence and the limited efficacy of B. pertussis vaccines against B. parapertussis suggest a lack of cross-protective immunity between the two. We sought to address the ability of infection-induced immunity against one of these pathogens to protect against subsequent infection by the other using a mouse model of infection. Immunity induced by B. parapertussis infection protected against subsequent infections by either species. However, immunity induced by B. pertussis infection prevented subsequent B. pertussis infections but did not protect against B. parapertussis infections. The O antigen of B. parapertussis inhibited binding of antibodies to the bacterial surface and was required for B. parapertussis to colonize mice convalescent from B. pertussis infection. Thus, the O antigen of B. parapertussis confers asymmetrical cross-immunity between the causative agents of whooping cough. We propose that these findings warrant investigation of the relative role of B. parapertussis in the resurgence of whooping cough.     

Clearance of Bordetella parapertussis from the lower respiratory tract requires humoral and cellular immunity

Bordetella parapertussis and Bordetella pertussis are closely related species that cause whooping cough, an acute, immunizing disease. Their coexistence in the same host populations at the same time and vaccine studies showing that B. pertussis vaccines have little effect on B. parapertussis infection or disease suggest that the protective immunity induced by each does not efficiently cross protect against the other. Although the mechanisms of protective immunity to B. pertussis have been well studied, those of B. parapertussis have not. The present study explores the mechanism by which B. parapertussis is cleared from the lower respiratory tract by anamnestic immunity. Serum antibodies are necessary and sufficient for elimination of this bacterium, and CD4(+) T cells, complement, and neutrophils are required for serum antibody-mediated clearance. Mice lacking immunoglobulin A had no defect in their ability to control or clear infection. Interestingly, serum antibody-mediated clearance of B. parapertussis did not require Fc receptors that are required for antibody-mediated clearance of B. pertussis. Together these data support a model for the mechanism of protective immunity to B. parapertussis that is similar but distinct from that of B. pertussis.     

Reciprocal protective immunity against Bordetella pertussis and Bordetella parapertussis in a murine model of respiratory infection

The protective immunity induced by infection with Bordetella pertussis and with Bordetella parapertussis was examined in a murine model of respiratory infection. Convalescent mice that had been infected by aerosol with B. pertussis or with B. parapertussis exhibited a protective immune response against B. pertussis and also against B. parapertussis. Anti-filamentous hemagglutinin (anti-FHA) serum immunoglobulin G (IgG) and anti-FHA lung IgA antibodies were detected in both mice infected with B. pertussis and those infected with B. parapertussis. Antibodies against pertussis toxin (anti-PT) and against killed B. pertussis cells were detected in mice infected with B. pertussis. Pertactin-specific antibodies and antibodies against killed B. parapertussis cells were detected in mice infected with B. parapertussis. Spleen cells from mice infected with B. pertussis secreted interferon-gamma (IFN-gamma) in response to stimulation by FHA or PT. Spleen cells from mice infected with B. parapertussis also secreted IFN-gamma in response to FHA. Interleukin-4 was not produced in response to any of the antigens tested. The profiles of cytokine secretion in vitro revealed induction of a Th1-biased immune response during convalescence from infection by B. pertussis and by B. parapertussis. It is possible that Th1 and Th2 responses against FHA might be related to the reciprocal protection achieved in our murine model.     

Antibody-mediated bacterial clearance from the lower respiratory tract of mice requires complement component C3

To assess the contribution of complement to respiratory immunity in the context of a natural bacterial infection, we used mice genetically deficient in complement components and the murine pathogen Bordetella bronchiseptica. Complement component C3 was not required for the control of bacterial infection or for the generation of infection-induced protective immunity. However, C3-deficient (C3(-/-)) mice were severely defective, compared to wild type, in vaccine-induced protective immunity. Adoptively transferred immune serum from convalescent wild-type or C3(-/-) animals rapidly cleared B. bronchiseptica from the lungs of wild-type mice but did not affect its growth in C3(-/-) mice, indicating that the defect is not in the generation of protective immunity, but in its function. Immune serum was effective in C5-deficient mice but had little effect in the lungs of mice lacking either Fcgamma receptors (FcgammaR) or CR3, suggesting bacterial clearance is not via direct complement-mediated lysis. Together, these data indicate that complement is required for antibody-mediated clearance of Bordetella and suggest the mechanism involves C3 opsonization of bacteria for phagocytosis that is both CR3- and FcgammaR-dependent.     

Nucleotide sequence homology to pertussis toxin gene in Bordetella bronchiseptica and Bordetella parapertussis.

Multiple strains of Bordetella parapertussis and B. bronchiseptica were examined for the presence of nucleotide sequences which hybridized with a cloned 4.5-kilobase (kb) fragment of B. pertussis DNA containing the genes responsible for pertussis toxin expression. All six B. parapertussis strains tested had nucleic acid sequences that hybridized with the cloned 4.5-kb fragment in Southern blot analyses. When the B. parapertussis DNA was digested with restriction endonuclease PstI, the pattern of hybridization was identical to that obtained with B. pertussis. Only five of the seven B. bronchiseptica strains tested had sequences that hybridized with the 4.5-kb fragment. Three of these B. bronchiseptica strains had a hybridization pattern identical to B. pertussis upon PstI digestion and Southern blot analyses. Two B. bronchiseptica strains were shown to lack a PstI cleavage site downstream from the region analogous to that coding for the pertussis toxin structural genes. Monoclonal antibody analyses were unable to detect pertussis toxin subunits S1 and S2 in Western blots with cellular material or culture supernatant from several B. bronchiseptica and B. parapertussis strains that possessed the DNA homologies. In addition, preliminary Northern hybridizations with RNA isolated from B. bronchiseptica and B. parapertussis strains suggested that the homologous regions were not transcribed. The data show that the gene coding for the toxic component of B. pertussis is common in other Bordetella species, though the gene probably is not expressed.     

Comparative phenotypic analysis of the Bordetella parapertussis isolate chosen for genomic sequencing

The genomes of three closely related bordetellae are currently being sequenced, thus providing an opportunity for comparative genomic approaches driven by an understanding of the comparative biology of these three bacteria. Although the other strains being sequenced are well studied, the strain of Bordetella parapertussis chosen for sequencing is a recent human clinical isolate (strain 12822) that has yet to be characterized in detail. This investigation reports the first phenotypic characterization of this strain, which will likely become the prototype for this species in comparison with the prototype strains of B. pertussis (Tohama I), B. bronchiseptica (RB50), and other isolates of B. parapertussis. Multiple in vitro and in vivo assays distinguished each species. B. parapertussis was more similar to B. bronchiseptica than to B. pertussis in many assays, including in BvgS signaling characteristics, presence of urease activity, regulation of urease expression by BvgAS, virulence in the respiratory tracts of immunocompromised mice, induction of anti-Bordetella antibodies, and serum antimicrobial resistance. In other assays, B. parapertussis was distinct from all other species (in pigment production) or more similar to B. pertussis (by lack of motility and cytotoxicity to a macrophage-like cell line). These results begin to provide phenotypes that can be related to genetic differences identified in the genomic sequences of bordetellae.     

Species specificity of Bordetella adherence to human and animal ciliated respiratory epithelial cells.

Bacteria of the genus Bordetella adhere preferentially to ciliated respiratory epithelial cells. We investigated the specificity of this unique tropism by assessing the concentration-dependent adherence of the three Bordetella species to ciliated cells from different hosts. Bordetella pertussis and Bordetella parapertussis adhere better to human ciliated cells than to those from rabbits, mice, or hamsters. In contrast, Bordetella bronchiseptica demonstrates preferential adherence to nonhuman mammalian ciliated cells of rabbits, mice, and hamsters. There was no attachment of any Bordetella organisms to chicken ciliated cells. These observations suggest that specificities of attachment may explain the marked predominance of B. pertussis as the cause of whooping cough in humans and of B. bronchiseptica as a respiratory pathogen of many nonhuman mammals.