点击显示 收起
Departments of Medicine and Microbiology, New York University School of Medicine
Department of Veterans Affairs Medical Center, New York, New York
Département de Microbiologie et Infectiologie, Hpital St-Luc du Centre Hospitalier de l'Université de Montréal, Montréal, Québec, Canada
Campylobacter fetus causes gastrointestinal and systemic infections in humans. Although relapse is common despite antibiotic treatment, the mechanisms are not well understood. The surface-layer proteins (SLPs) of C. fetus, which are critical in virulence, undergo high-frequency phenotypic switching due to recombination of sap homologues, resulting in antigenic variation. To investigate the mechanisms involved in relapsing C. fetus infections, we compared SLP variation in 4 pairs of C. fetus strains that infect humans; initial and follow-up isolations were performed 20 days to 34 months apart. Of the 4 pairs of strains, 2 had antigenic variation, and another provided evidence for selection for SLP-positive populations. Southern hybridization indicated recombination underlying the SLP variation and up-regulation. The fourth pair had the same SLP antigenic profile and sap homologue hybridization pattern, which is consistent with latency of the original strain in a privileged locus. In total, these findings indicate that relapse may reflect at least 3 differing pathogenetic pathways.
The ability of pathogenic microbes to persist in vertebrate hosts that are protected by adaptive immunity is a central biological problem. Generation of antigenic variation of surface components is one of the mechanisms enabling evasion of host immune responses during long-term colonization [1]. Rearrangement of the genetic loci involved in expression of major antigens, an efficient mechanism for generation of diversity, is employed by a wide range of microbes [26].
Campylobacter fetus, a gram-negative bacterium, is capable of persistently colonizing vertebrate hosts and also can cause infections both in animals important for food production and in humans [714]. Surface-layer proteins (SLPs), which are encoded by the sap homologues, are critical virulence factors during C. fetus infection [1521]. In wild-type C. fetus strain 23D, the 9 (8 complete and 1 partial) sapA homologues are clustered on a 54-kb chromosomal DNA locus (termed the "sap island" [22, 23]) that surrounds the genes required for secretion of the SLPs from the bacterial cell [24].
Genetic recombination resulting in phenotypic variation can occur at high frequencies within the C. fetus sap island [2529]. Recombination among the multiple sapA homologues, mediated by their 5 conserved regions [26, 28], results in phase variation in expression of homologues; the homologue located downstream of the unique sap promoter can be transcribed, whereas all other sapA homologues are transcriptionally silent [3032].
C. fetus can persistently colonize its natural mammalian hosts [29, 33, 34]. The high-frequency DNA rearrangement within the sap island, leading to antigenic variation, implies that persistence may be facilitated by the organism's ability to adapt to dynamic environments by continuous change. In the present study, we investigate whether antigenic variation can be observed in 4 pairs of C. fetus strains isolated from human patients with relapsing infections. Despite seemingly effective antibiotic therapies, relapsing C. fetus infections in humans have been well observed [35]. We now report analyses of genotypic and phenotypic variation of paired C. fetus strains from 4 patients with relapsing infections. The present study provides a better understanding of the mechanisms underlying recurrent infections.
MATERIALS AND METHODS
Strains.
All 4 pairs of C. fetus strains were isolated from patients in the province of Quebec, Canada. For each pair of strains, the clinical characteristics and the interval between isolations was known (table 1). In the present study, C. fetus strains were grown at 37°C under microaerobic conditions, in GasPak jars, by use of a CampyPak Plus gas generator (BBL Microbiology Systems), on brucella agar (Difco Laboratories) containing antibiotics at the following concentrations: 7 U/mL polymyxin B, 10 g/mL vancomycin, 15 g/mL nalidixic acid, and 10 g/mL trimethoprim lactate.
Polymerase chain reaction (PCR) techniques.
The PCR primers used in the present study are listed in table 2. To characterize the strains as either type A or type B, sapA-specific (SAF01 and SAR01) and sapB-specific (SBF01 and SBR01) primers, respectively, were used [36, 37] (table 2). To ascertain the presence of particular sap homologues, homologue-specific primers (A0R7R1) (table 2) were used for PCR amplifications [38]. PCR was performed in reaction volumes of 50 L containing 0.25 U of Taq polymerase (Qiagen), 10 mmol/L MgCl2, and 10 pm of each primer. The PCR protocol (35 cycles) included a denaturing step for 1 min at 94°C, annealing for 1 min at 5°C below the predicted melting temperature of the primers, and extension at 72°C for 1 min per kilobase of amplification product.
Immunoblot assay.
C. fetus whole-cell preparations were prepared as described elsewhere [39]. Protein concentrations were determined by bicinchoninic acid assay (Pierce), and lysates representing 1 g of protein were assayed by electrophoresis on a sodium dodecyl sulfate7% polyacrylamide gel by use of polyclonal rabbit serum against C. fetus 97-kDa SLP, as described elsewhere [39]. The secondary antibody was goat antirabbit IgG alkaline phosphatase conjugate (Chemicon International).
Serum-susceptibility assay.
Determination of the susceptibilities of C. fetus cells to normal human serum (NHS) was performed as described elsewhere [40]. Cells from 48-h cultures were harvested and resuspended in saline with 1 mmol/L calcium chloride; bacterial suspensions then were diluted 10-fold, from 10-1 to 10-7, and the 10-410-7 dilutions were incubated in the presence of 10% NHS or heat-inactivated (56°C) NHS for 60 min at 37°C in an atmosphere with 5% CO2. Triplicate samples then were inoculated onto blood agar plates, and survival rates were determined as described elsewhere [40].
Southern hybridization.
Chromosomal DNA was prepared from C. fetus cells grown on 2 plates for 48 h, by use of the Wizard Genomic DNA Purification Kit (Promega). DNA was digested with HindIII, electrophoresed on a 1% agarose gel, and transferred to a nylon membrane (MSI). The hybridization probe was a PCR product amplified by use of the sapA 5 conserved-region primers [37] or promoter-specific primers [30], with template DNA from strain 23D. Probes were labeled by use of the Renaissance chemiluminescent nonradioactive kit (NEN Research Products).
RESULTS
C. fetus strain typing.
C. fetus SLPs may be type A, type B, or, occasionally, type AB, which is consistent with determination of their serotype and lipopolysaccharide type [36, 37]. To characterize the 4 pairs of strains, we performed PCR using type A (sapA-specific) or type B (sapB-specific) primers. The results showed that all 4 pairs of C. fetus strains from these systemic infections of humans were type A, and not type B or type AB (figure 1A and 1B).
Repertoire of sap homologues for each pair of strains.
Variation of C. fetus SLP expression, mediated by DNA recombination, is due to rearrangement of the unique sap promoter upstream of one of the multiple sap homologues [30]. To determine the repertoire of sapA homologues contained within the sap locus, for each of the 4 pairs of strains, we performed PCR using primers specific for each of the 8 sap homologues that exist in strain 23D [22, 38]. The PCR results showed that the same sapA homologues were present in each pair of strains (figure 1C). Three pairs possessed all 8 sapA homologues that exist in strain 23D [38], but the pair-2 strains did not possess sapA. Thus, the 4 sets of strains were relatively conserved in the sap homologues that they possessed.
To characterize the arrangement of the sap locus in each pair of stains, we performed Southern hybridization using as a probe the 5 conserved region present in all sapA homologues. These results (figure 2A) indicated that each pair shared a characteristic overall hybridization pattern, but that, in pairs 1, 2, and 3, rearrangements had occurred between the strains. In pair 1, strain 2 (CA725) gained 2.7-kb and 2.9-kb bands, compared with strain 1 (CA700). In pair 2, strain 4 (CA740) gained a 4.2-kb band, compared with original strain 3 (29826). In pair 3, strain 6 (51918) lost a 4.2-kb band, compared with strain 5 (51560). For the pair-4 strains (designated "51534" and "CA1124"), there were no differences in sap hybridization patterns. Southern hybridization using the sap promoter as a probe showed that the promoter had been rearranged within populations of the second strains of pairs 1, 2, and 3, compared with populations of their first strains (figure 2B). From these studies, we concluded that, in pairs 1, 2, and 3, the strains were clonal variants of one another, whereas there was no evidence for variation in the populations of the pair-4 strains.
The expressed SLPs.
To examine the phenotype of each pair of strains in terms of SLP expression, we performed SDS-PAGE analysis and then performed immunoblotting using polyclonal antiserum raised to the 97-kDa type A SLP [39] (figure 3). Each pair of strains had the same protein profile, as determined by SDS-PAGE analysis, with the exception of the SLP bands (figure 3A). Immunoblotting indicated that, in pair 1, strain 2 had 2 major bands (97 and 127 kDa), whereas strain 1 had a major band (127 kDa) and a minor band (97 kDa) (figure 3B). In pair 2, strain 3 had a minor band (97 kDa), whereas strain 4 had a major band (97 kDa). In pair 3, both strain 5 (127 kDa) and strain 6 (149 kDa) had a major band. In pair 4, strains 7 and 8 had a major band of the same size (97 kDa).
Since both strains in pair 1 had 2 SLP bands, these could be accounted for by either a mixture of 2 populations of cells expressing different SLP bands or a single population simultaneously expressing 2 different major bands. To address these possibilities, we examined individual colonies of both strains. All 46 colonies of strain 1 (CA700) had the same 97-kDa band; however, 35 colonies had an extra 127-kDa band, and the other 11 colonies had a 131-kDa band (figure 4A). Of 18 colonies of strain 2 (CA725), 13 had a major 97-kDa band, 2 had major 127-kDa bands, and the other 3 did not have any SLP bands (figure 4B). Thus, there was extensive phenotypic (SLP) diversity within the population of cells in pair 1. In pair 2, strain 3 (29826) had a minor 97-kDa band, whereas strain 4 (CA740) had a major 97-kDa band. On the basis of these findings and the findings of Southern hybridization using the promoter probe (figure 2B), we hypothesized that, in the pair-2 populations, the SLP promoter may be subject to high-frequency rearrangement events. Examination for SLP expression showed that none were detected in the 13 colonies of strain 3 (29826) cells or in 4 of the 13 colonies of strain 4 (CA740) cells (figure 4C). These findings suggest that deletions involving the sapA promoter may have been occurring at a high frequency in strains from this pair.
Serum resistance of the pair-2 strains.
All pathogenic C. fetus strains are SLP positive [15]. If the capacity to express SLPs is removed, then the resulting strains are markedly reduced in virulence in vitro [15] and in vivo [16, 18]; SLP-negative strains are sensitive to NHS, whereas SLP-positive strains are resistant to NHS [15]. To further assess the population differences in SLP expression between pair-2 strains 3 (29826) and 4 (CA740), the cells were incubated with NHS to quantitate resistance to NHS. As expected [40], strain 23D was resistant to NHS (>10-1 survival), whereas SLP-negative strain 23B was highly sensitive to NHS (<10-5 survival). After incubation with NHS, the survival rate for strain 3 was 10-2.6, whereas, for strain 4, it was 10-0.8, nearly 2 log10 higher (figure 5). For each of the strains, we selected 10 survivors of incubation with NHS to examine SLP expression. All 10 colonies of NHS-resistant strain 4 expressed 97-kDa SLPs, whereas only 2 of 10 colonies of strain-3 NHS survivors expressed 97-kDa SLPs; the other 8 were SLP negative (data not shown). To determine whether the shift to SLP negativity could be completely explained as its being an in vitro phenomenon, we selected 1 colony each of SLP-positive strain-3 and strain-4 serum survivors. Both continued to be highly resistant to NHS (>10-1 survival), similar to strain 23D (figure 5A). We next examined the susceptibility to NHS of serum-surviving but SLP-negative strain 3. After incubation with NHS, the survival rate was 10-2.4, nearly identical to the result for its parental strain and much higher than that observed for the spontaneously mutated SLP-negative strain 23B (<10-5 survived) (figure 5A). These results show that strain 3 included several populations that differed in SLP expression and in resistance to NHS. A much greater proportion of the cells from subsequent strain 4 were high-level SLP expressers and were 100-fold more resistant to NHS. The results of analysis of this pair of strains suggested that there had been in vivo selection for resistance to NHS.
Genetic characteristics of the pair-2 variants.
To investigate the basis for why pair-2 strains and their derivatives had different susceptibilities to NHS, we compared the sap homologue profiles among single colonies of strains 3, 3S- (resistant to NHS and SLP negative), 3S+ (resistant to NHS and SLP positive), 4, and 4S+ (SLP positive). The results indicated that strain 3S- lost a 4.2-kb band, compared with strain 3S+. As expected, the 4.2-kb band also existed in strain 4 and its selected clone, 4S+, and was not detected in original strain 3 (figure 5B). The sap homologuespecific PCR indicated that sapA4, the SLP-expressing gene in strains 3S+ and 4S+, was deleted in strain 3S- (figure 5C).
To determine whether strain 3S- underwent genotypic and phenotypic variation over time, we passed strain 3S- 10 times in vitro, representing hundreds of generations. We performed Western blotting and PCR for strains 3S- and 3S+, at 0, 5, and 10 passages. By Western blotting, each strain had no change in original SLP phenotype (data not shown). By PCR, strain 3S+ had the expected products in the promoter PCR and in the sapA4 PCR, without change from passage 0, 5, or 10. For strain 3S-, there were no PCR products at passage 0, 5, or 10, also as expected (data not shown). These data confirm the in vitro stability of genotypes and phenotypes related to SLP expression and provide further evidence that the changes we observed with the passed strains reflected in vivo phenomena.
DISCUSSION
We have characterized 4 pairs of C. fetus strains from humans with relapsing infection by comparing their sap type, sap island repertoire and organization, and the SLPs expressed. In each case, both strains in the pair were type A, which appears to be more common in human infection [37], although no systematic study has been done. That 3 pairs of strains possessed all 8 sapA homologues that exist in type A strain 23D and that both strains in the other pair possessed 7 sapA homologues (without sapA) suggest conservation of this particular repertoire of multiple sap homologues among the C. fetus strains infecting humans, a finding that is consistent with our previous observations of 18 C. fetus strains from a variety of sources [37]. Whether the strains we studied possessed any of the 5 other sapA homologues that have been characterized from other C. fetus isolates [38] was not determined.
The sapA 5 conserved-region Southern hybridization (figure 2A) reflected the organization of the sapA homologues, in relation to their genomic positions. All 4 pairs of strains had different sap island profiles, indicating that each had a different epidemiologic source, but, within each pair, there was a highly similar pattern of organization of the sap island, confirming that both strains in each pair had the same ancestor.
Our results indicate that, in 3 of the 4 pairs of strains studied, the strain isolated after relapse of C. fetus infection had undergone change in the predominant SLP expressed. This was illustrated in the pair-1 strains, for which the overall (figure 3) and the detailed (figure 4) analyses showed differences in size and amount of SLP expression. Southern hybridization indicated one basis for these changes: differing rearrangements involving the sapA homologues and the sap promoter. In pair 3, strain 6 produced a 149-kDa SLP instead of the 127-kDa SLP produced by strain 5, a finding consistent with selection for antigenic variation to escape host immune defenses. That the pair-4 strains demonstrated the same SLP expression illustrates a different phenomenon. Strain 7 was isolated from the blood of a patient with septicemia who had been successfully treated with antibiotics. Weeks later, the patient developed chronic arthritis from which synovial fluid was cultured several times; this fluid was found to be positive for C. fetus only once, 34 months later. This long interval is most consistent with latency of low-grade C. fetus infection in a privileged site and then reactivation of the original strain with a change in host status, as has been observed previously [34]. That the pair-4 strains appeared to be identical at both phenotypic and genotypic levels supports the hypothesis of latency, paralleling that documented for other infectious agents, including Mycobacterium tuberculosis, Herpes viruses, Plasmodium species, and Rickettsia prowazekii.
Our analyses of the pair-2 strains provide a new understanding of C. fetus virulence. The first isolate, strain 3, showed little SLP expression, and Southern hybridization did not detect the sap promoter. For this strain, the susceptibility to NHS was intermediate between that of strain 23D and that of the spontaneous SLP-negative strain 23B, in which the sap promoter had been deleted [32]. This suggests the possibility that the strain actually represented a mixture of 2 such populations, with high-frequency switching. The subsequent SDS-PAGE and serum-susceptibility analyses confirmed the existence of 2 such populations that differed in whether they expressed SLP. Within that mixture, a completely resistant SLP-positive population was present. We hypothesize that strain 4, which was isolated 20 days later, emerged from this population (i.e., was selected because of its resistance to NHS). The results of Southern hybridization of the strain-3 derivatives provide support for this hypothesis (figure 5B) and indicate DNA rearrangement as the mechanism underlying SLP expression and resistance to NHS. Thus, in addition to completely NHS-resistant and highly SLP-positive strains causing human infections (as seen in pairs 1, 3, and 4), a strain with an intermediate phenotype but with the capacity to rearrange into a more virulent phenotype also was pathogenic. In total, the results of analyses of these 4 pairs of strains associated with relapse point to at least 3 different mechanisms for relapse of clinical C. fetus infection: (1) rearrangement to up-regulate virulence, (2) rearrangement for antigenic variation, and (3) true latency of a fully virulent strain with later reactivation due to a change in clinical status.
References
1. Finlay BB, Falkow S. Common themes in microbial pathogenicity revisited. Microbiol Mol Biol Rev 1997; 61:13669. First citation in article
2. Borst P. Antigenic variation and allelic exclusion. Cell 2002; 109:58. First citation in article
3. Henderson IR, Owen P, Nataro JP. Molecular switchesthe ON and OFF of bacterial phase variation. Mol Microbiol 1999; 33:91932. First citation in article
4. Seifert HS, So M. Genetic mechanisms of bacterial antigenic variation. Microbiol Rev 1988; 52:32736. First citation in article
5. Bahl H, Scholz H, Bayan N, et al. Molecular biology of S-layers. FEMS Microbiol Rev 1997; 20:4798. First citation in article
6. Beveridge TJ, Pouwels PH, Sara M, et al. Functions of S-layers. FEMS Microbiol Rev 1997; 20:99149. First citation in article
7. Guerrant RL, Lahita RG, Winn WC Jr, Roberts RB. Campylobacteriosis in man: pathogenic mechanisms and review of 91 bloodstream infections. Am J Med 1978; 65:58492. First citation in article
8. Blaser MJ. Campylobacter fetus emerging infection and model system for bacterial pathogenesis at mucosal surfaces. Clin Infect Dis 1998; 27:2568. First citation in article
9. Skirrow MB. Campylobacter and Helicobacter infections of man and animals. In: Parker MT, Collier LH, eds. Principles of bacteriology, virology, and immunity. 8th ed. Vol 2. London: Edward Arnold, 1990:53145. First citation in article
10. Smibert RM. The genus Campylobacter. Annu Rev Microbiol 1978; 32:673709. First citation in article
11. Penner JL. The genus Campylobacter: a decade of progress. Clin Microbiol Rev 1988; 1:15772. First citation in article
12. Young VB, Dangler CA, Fox JG, Schauer DB. Chronic atrophic gastritis in SCID mice experimentally infected with Campylobacter fetus. Infect Immun 2000; 68:21108. First citation in article
13. Harvey S, Greenwood JR. Isolation of Campylobacter fetus from a pet turtle. J Clin Microbiol 1985; 21:2601. First citation in article
14. Tu ZC, Zeitlin G, Gagner JP, Keo T, Hanna BA, Blaser MJ. Campylobacter fetus of reptile origin as a human pathogen. J Clin Microbiol 2004; 42:44057. First citation in article
15. Blaser MJ, Smith PF, Hopkins JA, Heinzer I, Bryner JH, Wang WL. Pathogenesis of Campylobacter fetus infections: serum resistance associated with high-molecular-weight surface proteins. J Infect Dis 1987; 155:696706. First citation in article
16. Blaser MJ, Pei Z. Pathogenesis of Campylobacter fetus infections: critical role of high-molecular-weight S-layer proteins in virulence. J Infect Dis 1993; 167:3727. First citation in article
17. Blaser MJ, Wang E, Tummuru MK, Washburn R, Fujimoto S, Labigne A. High-frequency S-layer protein variation in Campylobacter fetus revealed by sapA mutagenesis. Mol Microbiol 1994; 14:45362. First citation in article
18. Grogono-Thomas R, Dworkin J, Blaser MJ, Newell DG. Roles of the surface layer proteins of Campylobacter fetus subsp. fetus in ovine abortion. Infect Immun 2000; 68:168791. First citation in article
19. Grogono-Thomas R, Blaser MJ, Ahmadi M, Newell DG. Role of S-layer protein antigenic diversity in the immune responses of sheep experimentally challenged with Campylobacter fetus subsp. fetus. Infect Immun 2003; 71:14754. First citation in article
20. Dubreuil JD, Kostrzynska M, Austin JW, Trust TJ. Antigenic differences among Campylobacter fetus S-layer proteins. J Bacteriol 1990; 172:503543. First citation in article
21. Casademont I, Chevrier D, Guesdon JL. Cloning of a sapB homologue (sapB2) encoding a putative 112-kDa Campylobacter fetus S-layer protein and its use for identification and molecular genotyping. FEMS Immunol Med Microbiol 1998; 21:26981. First citation in article
22. Tu ZC, Wassenaar TM, Thompson SA, Blaser MJ. Structure and genotypic plasticity of the Campylobacter fetus sap locus. Mol Microbiol 2003; 48:68598. First citation in article
23. Tu ZC, Ussery DW, Pride DT, Blaser MJ. Genomic characteristics of the Campylobacter fetus sap island. Genome Lett 2003; 2:406. First citation in article
24. Thompson SA, Shedd OL, Ray KC, Beings MH, Jorgensen JP, Blaser MJ. Campylobacter fetus surface layer proteins are transported by a type I secretion system. J Bacteriol 1998; 180:64508. First citation in article
25. Garcia MM, Lutze-Wallace CL, Denes AS, Eaglesome MD, Holst E, Blaser MJ. Protein shift and antigenic variation in the S-layer of Campylobacter fetus subsp. venerealis during bovine infection accompanied by genomic rearrangement of sapA homologs. J Bacteriol 1995; 177:197680. First citation in article
26. Dworkin J, Shedd OL, Blaser MJ. Nested DNA inversion of Campylobacter fetus S-layer genes is recA dependent. J Bacteriol 1997; 179:75239. First citation in article
27. Ray KC, Tu ZC, Grogono-Thomas R, Newell DG, Thompson SA, Blaser MJ. Campylobacter fetus sap inversion occurs in the absence of RecA function. Infect Immun 2000; 68:56637. First citation in article
28. Tu ZC, Ray KC, Thompson SA, Blaser MJ. Campylobacter fetus uses multiple loci for DNA inversion within the 5 conserved regions of sap homologs. J Bacteriol 2001; 183:665461. First citation in article
29. de Vargas AC, Costa MM, Vainstein MH, Kreutz LC, Neves JP. Campylobacter fetus subspecies venerealis surface array protein from bovine isolates in Brazil. Curr Microbiol 2002; 45:1114. First citation in article
30. Dworkin J, Blaser MJ. Generation of Campylobacter fetus S-layer protein diversity utilizes a single promoter on an invertible DNA segment. Mol Microbiol 1996; 19:124153. First citation in article
31. Fujita M, Moriya T, Fujimoto S, Hara N, Amako K. A deletion in the sapA homologue cluster is responsible for the loss of the S-layer in Campylobacter fetus strain TK. Arch Microbiol 1997; 167:196201. First citation in article
32. Tummuru MK, Blaser MJ. Characterization of the Campylobacter fetus sapA promoter: evidence that the sapA promoter is deleted in spontaneous mutant strains. J Bacteriol 1992; 174:591622. First citation in article
33. Wesley IV, Bryner JH. Antigenic and restriction enzyme analysis of isolates of Campylobacter fetus subsp venerealis recovered from persistently infected cattle. Am J Vet Res 1989; 50:80713. First citation in article
34. Neuzil KM, Wang E, Haas DW, Blaser MJ. Persistence of Campylobacter fetus bacteremia associated with absence of opsonizing antibodies. J Clin Microbiol 1994; 32:171820. First citation in article
35. Tremblay C, Gaudreau C, Lorange M. Epidemiology and antimicrobial susceptibilities of 111 Campylobacter fetus subsp. fetus strains isolated in Quebec, Canada, from 1983 to 2000. J Clin Microbiol 2003; 41:4636. First citation in article
36. Dworkin J, Tummuru MK, Blaser MJ. Segmental conservation of sapA sequences in type B Campylobacter fetus cells. J Biol Chem 1995; 270:15093101. First citation in article
37. Tu ZC, Dewhirst FE, Blaser MJ. Evidence that the Campylobacter fetus sap locus is an ancient genomic constituent with origins before mammals and reptiles diverged. Infect Immun 2001; 69:223744. First citation in article
38. Tu ZC, Hui J, Blaser MJ. Conservation and diversity of sap homologues and their organization among Campylobacter fetus isolates. Infect Immun 2004; 72:171524. First citation in article
39. Pei Z, Ellison RT 3rd, Lewis RV, Blaser MJ. Purification and characterization of a family of high molecular weight surface-array proteins from Campylobacter fetus. J Biol Chem 1988; 263:641620. First citation in article
40. Blaser MJ, Smith PF, Kohler PF. Susceptibility of Campylobacter isolates to the bactericidal activity of human serum. J Infect Dis 1985; 151:22735. First citation in article