![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Previous Article | Next Article 
Infection and Immunity, August 2007, p. 4050-4061, Vol. 75, No. 8
0019-9567/07/$08.00+0 doi:10.1128/IAI.00486-07
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Departamento de Microbiología y Genética, Edificio Departamental, Universidad de Salamanca, Plaza Doctores de la Reina s/n, 37007 Salamanca, Spain,1 Centro de Investigación y Tecnología Agroalimentaria del Gobierno de Aragón, Unidad de Sanidad Animal, Carretera de Montañana 930, 50059 Zaragoza, Spain,2 INRA, UR1282, Infectiologie Animale et Santé Publique, IASP, Nouzilly F-37380, France,3 Instituto de Agrobiotecnología y Recursos Naturales, CSIC-UPNA, Ctra. Mutilva Baja, 31192 Pamplona, Spain4
Received 4 April 2007/ Returned for modification 10 May 2007/ Accepted 30 May 2007
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
omp22 mutant to nonimmune serum and its difficulty in surviving in the stationary phase might be on the basis of its dramatic attenuation, no defects in the OM able to explain the attenuation of the omp25d mutant were found, especially considering that the fully virulent omp25c mutant displayed more important OM defects. Accordingly, Omp25d, and perhaps Omp22, could be directly involved in the penetration and/or survival of B. ovis inside host cells. This aspect, together with the role of Omp25d and Omp22 in the virulence both of B. ovis in rams and of other Brucella species, should be thoroughly evaluated in future studies. |
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Brucella spp. may have smooth or rough lipopolysaccharide (S-LPS or R-LPS) depending on the presence or absence, respectively, of O-polysaccharide chains. Rough mutants derived from smooth Brucella strains show an important reduction in virulence (1, 26, 30, 40, 45). Accordingly, the O-polysaccharide chains of LPS are thought to be necessary for the pathogenicity of Brucella strains bearing S-LPS. However, B. ovis and B. canis lack O chains in the LPS but are pathogenic for rams and dogs, respectively, and induce long-lasting infections with high levels of splenic colonization in laboratory animals (34). Since O chains mask other components of the Brucella spp. outer membrane (OM), OM proteins (OMPs) are more exposed at the bacterial surface of rough B. ovis and B. canis (4, 7) and their involvement in virulence in rough Brucella strains may be more relevant than in smooth Brucella strains.
The Brucella spp. Omp25/Omp31 family comprises seven homologous OMPs. Omp25 and Omp31 are major OMPs, except Omp31 in B. abortus, which lacks its coding gene (48, 49, 51), and they have been fairly well characterized in several aspects (11). Regarding virulence, mutant B. melitensis, B. abortus, and B. ovis strains with the omp25 gene inactivated have been found to be attenuated in mice, goats (B. melitensis), and cattle (B. abortus) (17-19). Additionally, Omp25 has been shown to inhibit the production of tumor necrosis factor alpha (TNF-
) by human macrophages (35) and to be involved in the permeability of the Brucella membrane, allowing the secretion of periplasmic proteins, in acidic medium (3).
In contrast, little is known about the Omp31b, Omp25b, Omp25c, Omp25d, and Omp22 proteins. Although no studies specifically analyzing the presence of these proteins in each Brucella species have been reported, the five OMPs have been detected experimentally in at least one Brucella species. Thus, Omp31b has been identified in B. abortus and B. suis (13, 42), Omp25b in B. melitensis and B. suis (23, 42, 52), Omp25c in B. melitensis, B. abortus, and B. suis (13, 42, 52), and Omp22 in B. abortus (13, 32). Evidence for the production of Omp25d in the genus Brucella is less conclusive, since the protein has been detected only weakly, by reactivity with an anti-Omp25 monoclonal antibody (MAb), in an omp25-defective B. suis mutant (42). Studies addressing DNA polymorphism of the Omp25/Omp31-encoding gene family in the genus Brucella have provided additional information about the occurrence of these five proteins in the OM of the Brucella species (46): (i) Omp31b is probably absent from the OMs of B. melitensis, B. ovis, and B. canis; (ii) Omp25b is not expected to be present in the OMs of B. abortus, B. ovis, B. canis, B. cetaceae, and B. pinnipediae; and (iii) Omp25c, Omp25d, and Omp22 are conserved proteins that would be located in the OMs of the six classical Brucella species and of B. cetaceae and B. pinnipediae (perhaps with the exception of Omp22 in some B. cetaceae isolates).
In the present work, we evaluated the role of the five members of the Omp25/Omp31 family expected to be located in the OM of B. ovis (46) in terms of the OM properties and virulence of the bacterium. To achieve these objectives, single-mutant strains with the genes omp31, omp25, omp25c, omp25d, or omp22 inactivated were obtained from virulent B. ovis PA and were subsequently complemented with the corresponding wild-type genes.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
pir). Recombinant E. coli cells were grown in Luria-Bertani medium supplemented with 50 µg/ml of the required antibiotic(s) depending on the plasmid. Primers and DNA techniques. Primers (Roche, Germany) (Table 2) were chosen according to the published B. melitensis 16 M and B. suis 1330 nucleotide sequences for chromosomes I and II (GenBank accession numbers AE008917, AE008918, AE014291, and AE014292). PCR was performed with the Expand Long Template PCR system (Roche) following the instructions of the manufacturer with 100 ng of B. ovis strain DNA template extracted with a High Pure PCR template preparation kit (Roche) and 2 µM each primer. DNA was sequenced by primer-directed dideoxy sequencing with an ABI PRISM 377 DNA sequencer (Perkin-Elmer, Foster City, CA).
|
MAbs and immunological techniques. MAbs A59/05F01/C09 (C09) and A19/12B10/F04 (F04) (ascitic fluid), specific for Brucella sp. Omp25, and A59/10F09/G10 (G10) and A01/08H06/G02 (G02) (hybridoma culture medium supernatant), specific for Brucella spp. Omp31 (7, 50), were used. Western blotting with Brucella whole-cell lysates was performed as described previously (50). For indirect enzyme-linked immunosorbent assay (iELISA) (46), parental or mutant B. ovis PA strains cultured for 24 h in the corresponding TSB-YE-HS media and washed with phosphate-buffered saline (PBS; pH 7.2) were used as coating antigens (100 µl/well of a bacterial suspension adjusted to an optical density at 600 nm [OD600] of 1). Then, a saturation step with 5% skim milk in PBS was included (300 µl/well; 30 min at 37°C). After subsequent incubation steps with MAbs and goat anti-mouse immunoglobulin G (Fc-specific) peroxidase conjugate (Sigma, St. Louis, Mo.; 1:4,000), OD405 values were recorded after 30 min of incubation with the substrate solution, and the results were expressed as the mean ± standard deviation (SD) of the OD readings for three wells.
Inactivation of the genes encoding the Omp25/Omp31 family in B. ovis PA by homologous recombination. In this work, we obtained five mutant strains (Table 1) derived from parental B. ovis PA by individually inactivating the genes coding for Omp31, Omp25, Omp25c, Omp25d, and Omp22 (genes BRA0423, BR0701, BR0119, BR0118, and BR1284, respectively, according to the B. suis 1330 genome sequence). For all the genes except omp25, inactivation was performed by replacing most of the gene by a Kan resistance cassette, which was placed in the direction opposite to that of the gene. Inactivation of omp25 was achieved by insertion of the Kan resistance cassette, also in the direction opposite to that of the gene, at the AflII site located just 10 nucleotides downstream from the omp25 ATG start codon.
Briefly, deletion of omp31 from parental B. ovis PA was performed as follows. Primers 31MUT-F and 31MUT-R (Table 2) were used to amplify, according to the B. melitensis 16 M sequence (49), a 1,406-bp fragment comprising omp31 and DNA flanking both sides of the gene. The amplified fragment was cloned into plasmid pGEM-T Easy, excised by EcoRI digestion from the recombinant plasmid obtained, and then cloned into the EcoRI site of pGEM7-Zf. Then, the XcmI-SalI fragment of omp31 was replaced from the resulting pNV31201 recombinant plasmid by the Kanr gene extracted from pUC4K (Amersham Biosciences) to give pNV31211. The DNA insert of pNV31211 was isolated by digestion with SphI and SacI and cloned into pCVD442, resulting in pNV31211-1, which was introduced into B. ovis PA by electroporation as described previously (46). To replace wild-type omp31 by the inactivated omp31 gene of pNV31211-1 from the B. ovis chromosome by double recombination through the homologous regions flanking both sides of omp31, a previously described procedure was followed (21). Mutant B. ovis PNV31A (omp31) was selected by both its resistance to Kan and sucrose and by its susceptibility to Amp.
A similar process was followed to inactivate the other genes, and only the significant differences are mentioned below. The primers used for the inactivation of the other genes (Table 2) were as follows: 25MUTZ-F and 25MUTZ-R (omp25), 25cdMUT-F and 25cMUT-R (omp25c), 25cdMUT-F and 25cdMUT-R (omp25d), and 22MUT-F and 22MUT-R (omp22). Plasmids pNV25mut-1, pNV25c-1, pNV25d-2, and pNV22-1 were the pCVD442-derived plasmids used to electroporate B. ovis PA to obtain the mutant strains described in Table 1. These plasmids respectively contained the omp25, omp25c, omp25d, and omp22 genes inactivated by the Kan resistance cassette, as shown in Table 1, column 3.
To check the mutant strains, a chromosomal region including the entire fragment used to inactivate the omp gene and adjacent DNA on both sides was PCR amplified and sequenced. Southern blot hybridization of HindIII-digested chromosomal DNA with pCVD442 and either pNV31211-1, pNV25mut-1, pNV25c-1, pNV25d-2, or pNV22-1 DIG-labeled probes was also performed. The absence of Omp31 and Omp25 from B. ovis PNV31A and B. ovis PNV25A (Table 1), respectively, was checked by Western blotting with MAbs G02 and C09. CFU in medium with and without Kan were determined for each mutant strain in order to evaluate the stability of the mutations in vitro.
Complementation of mutant strains. Mutant B. ovis PA strains were complemented with the corresponding wild-type genes. Each gene, including the ribosome binding site and the transcription terminator, was PCR amplified with primers 31sd and 31ter (omp31), 25A-F and 25-R (omp25), 25C-F and 25C-R (omp25c), 25D-F and 25D-R (omp25d), or 22-F and 22-R (omp22) (Table 2). PCR products were cloned in pGEM-T Easy, and the DNA inserts of the selected plasmids were subsequently cloned in pBBR1MCS-4 to leave the omp genes under the control of the lacZ promoter. Recombinant pBBR1MCS-4 derivatives pNV31300, pNV25com, pNV25c4, pNV25d4, and pNV22D were introduced in the corresponding mutant B. ovis PA strain by electroporation, and colonies resistant to both Amp and Kan were selected (Table 1). Complementation was verified by extraction of the plasmid DNA and sequencing of the DNA insert. Synthesis of Omp31 in B. ovis PNV31A-com and of Omp25 in B. ovis PNV25A-com (Table 1) was also confirmed by reactivity with the MAbs G02 and C09 in Western blotting assays. Plasmid stability of the complemented mutant strains was tested by plating in culture medium without antibiotic or with Kan and Amp.
Susceptibility assays. For susceptibility assays, B. ovis strains and B. abortus RB51 were cultured for 44 and 24 h, respectively, on TSA-YE-HS supplemented with antibiotic(s) when required. Then, a bacterial suspension was prepared in PBS as detailed for each test.
The susceptibilities of Brucella strains to polymyxin B (Sigma) and sodium deoxycholate (Sigma) were determined following a protocol described previously (38) with some modifications. A bacterial suspension containing approximately 1 x 104 CFU/ml was prepared in PBS, and 100 µl was mixed in wells of a 96-well sterile plate with 100 µl of 2 mg/ml polymyxin B or 0.2 mg/ml sodium deoxycholate (final concentrations in the wells, 1 mg/ml and 0.1 mg/ml, respectively). After a 1-h incubation at 37°C in a 10% CO2 atmosphere, the content of each well was mixed, and 50 µl was spread in triplicate on TSA-YE-HS plates supplemented with antibiotic(s) when required. The CFU obtained after treatment with polymyxin B or sodium deoxycholate were counted, and the percentages of survival were established with respect to the CFU obtained with bacteria incubated in PBS (100% survival). The results were expressed as the mean ± SD of three assays.
Susceptibility to hydrogen peroxide (Sigma) was evaluated by a modification of the disk sensitivity assay described by Elzer et al. in 1994 (22). A bacterial suspension with approximately 109 CFU/ml (OD600 of 0.2) was prepared in PBS, and 100 µl was spread by triplicate on TSA-YE-HS plates supplemented with antibiotic(s) when required. A Whatman 3MM Chr disk (9-mm diameter) was placed at the center of the plate, and 10 µl of 30% H2O2 was deposited on the disk. After 4 days of growth, the diameter of the zone of clearance around the disk was measured in quadruplicate for each plate and the mean diameter was calculated. The results were expressed as the mean ± SD of the diameter (in cm) obtained for the three plates.
The protocol designed to determine the susceptibility of the Brucella strains to nonimmune ram serum was based on that described by Corbeil et al. in 1988 (14). Serum was obtained from a 5-month-old ram born in the brucellosis-free flock of the Centro de Investigación y Tecnología Alimentaria (CITA) del Gobierno de Aragón, Spain. Half of the serum was heated at 56°C for 30 min to remove complement (control serum) and the other half was used fresh. Briefly, 50-µl bacterial suspensions with approximately 2 x 104 CFU/ml in PBS were mixed in wells of a 96-well sterile plate with 150 µl of either fresh serum or heated serum. After a 4-h incubation at 37°C in a 10% CO2 atmosphere, the content of each well was gently mixed by pipetting and 50 µl of each bacterial suspension was spread in triplicate on TSA-YE-HS plates supplemented with antibiotic(s) when required. The CFU obtained after exposure to fresh serum were counted, and the percentages of survival were established with respect to the CFU obtained with the heated serum (100% survival). The results were expressed as the mean ± SD of three assays performed with the same serum.
Autoagglutination test. Bacterial suspensions with an OD600 of 0.8 were prepared in TSB-YE-HS from B. ovis strains previously cultured for 44 h (or from B. abortus RB51 cultured for 24 h) in TSA-YE-HS supplemented with the corresponding antibiotic(s). One ml of each suspension was left in a spectrophotometer microcuvette without agitation, and the OD600 readings were scored for 45 h at selected intervals. One microcuvette with TSB-YE-HS was used as a blank. The results were represented as the mean ± SD of the values obtained for three assays at each time point.
Virulence studies in mice. Female 8-week-old BALB/c mice (Charles River Laboratories, Spain) were accommodated in the animal building of the CITA of Aragón (registration number ES-502970012005) under biosafety conditions. The animals were kept in cages with water and food ad libitum for 1 week before the start of the experiment. The mouse experimental procedures and the facilities used to hold the mice were in accordance with current European legislation (directive 86/609/EEC).
After a preliminary assay to determine the most adequate dose of infection, mice were inoculated intraperitoneally with approximately 5 x 106 CFU of the corresponding B. ovis strain in 0.1 ml of PBS, and splenic bacterial counts were determined for five mice per strain at 1, 2, 3, 5, 7, 9, and 11 weeks postinoculation (p.i.). Experimental procedures (i.e., preparation and administration of inocula, retrospective assessment of the exact inoculating doses, and determination of the number of CFU/spleen) were performed as described previously (12). The numbers of CFU/spleen were transformed logarithmically to normalize the distribution of individual counts. Results were expressed as the mean ± SD (n = 5) of the log CFU/spleen for each strain at each selected p.i. time point.
To check the stability of mutation after in vivo activity, the strains recovered from the spleens of infected mice were rechecked by PCR amplification, Southern blot hybridization, plating in culture medium with or without the required antibiotic, bacteriological typing, and Western blot assays as described above.
Statistical analysis. Statistical comparisons between means were performed using analysis of variance, and the levels of significance of the differences were determined with the post hoc Fisher's protected least significant differences test.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In vitro growth on solid (TSA-YE-HS) and in liquid (TSB-YE-HS) media was not, in general, dramatically affected for the genetically modified B. ovis PA strains. The number of CFU/ml was determined for bacterial suspensions of an OD600 of 0.2 prepared from bacteria cultured for 44 h on TSA-YE-HS plates. B. ovis omp25d and omp22 mutants and their derived complemented strains had counts similar to that obtained with parental B. ovis PA, while omp31 and omp25c mutant strains and their respective complemented strains had CFU/ml values about half as high (the omp31 and omp31-com strains also required one more day than the other strains to develop visible colonies on solid medium) (data not shown). The omp25 mutant strain also afforded CFU/ml values about half as high as those observed with the parental strain, but complementation of this mutant with wild-type omp25 recovered the counts obtained with B. ovis PA (data not shown).
Growth in liquid medium of bacterial cultures starting at an OD600 of 0.05 and incubated with shaking is shown in Fig. 1. The omp31 and omp25 mutant strains, in spite of reaching higher OD values (Fig. 1A), provided CFU/ml numbers in the stationary phase similar to that observed for parental B. ovis PA (data not shown). The omp25c strain seemed to grow slower than the parental strain, reaching OD600 and CFU/ml values in the stationary phase that were slightly lower than those corresponding to B. ovis PA. As determined by OD600 readings and CFU/ml counts, the omp25d mutant grew as well as the parental strain did (Fig. 1A and data not shown). The omp22 mutant grew faster than B. ovis PA, reaching the stationary phase earlier (Fig. 1B). However, although the bacterial counts for the omp22 mutant were in accordance with the OD600 scores and both parameters were higher than those observed with B. ovis PA up to 25 h of incubation, the number of CFU/ml for this mutant underwent a drastic decrease after 46 h of culture; at this time point, it was almost 2 logs lower than that obtained for this mutant at 25 h and that obtained for the parental strain after 46 h of incubation (Fig. 1B).
|
|
omp31 mutant strain by Western blotting (Fig. 2A) or by iELISA (Fig. 2C), and complementation of this strain with wild-type omp31 restored the synthesis of Omp31 (Fig. 2A and C). Upon comparing B. ovis PNV31A-com to parental B. ovis PA, however, slight differences regarding the reactivity of the anti-Omp31 MAbs were found as follows. (i) MAb G02 showed, in both Western blotting and iELISA, a weaker reactivity with the complemented mutant strain PNV31-com (Fig. 2A and C) than with the parental strain (Fig. 2A and C). (ii) Nevertheless, strain PNV31A-com reacted slightly better than B. ovis PA with MAb G10 in iELISA (Fig. 2C). This MAb, raised against B. melitensis Omp31, reacts strongly with Omp31 from B. ovis in iELISA but very weakly in Western blotting (50). The other four mutant strains and the four respective complemented strains behaved in manners similar to that of the parental strain regarding reactivity with the anti-Omp31 MAbs in Western blotting (Fig. 2A) and iELISA (Fig. 2C), with the exception of B. ovis PNV25cA, which reacted more weakly in iELISA with MAb G02 (Fig. 2C).
Likewise, Omp25 was not detected in the B. ovis PNV25A mutant strain by Western blotting or by iELISA with a specific MAb (Fig. 2B and D), and the recovery of the synthesis of the protein was detected in the complemented strain B. ovis PNV25A-com (Fig. 2B and D). Anti-Omp25 MAb C09 developed in Western blotting two protein bands in parental B. ovis PA that correspond to Omp25, as previously shown (10, 29). These Omp25 protein bands were more intensely marked both in mutant B. ovis PNV31A and in its corresponding strain complemented with wild-type omp31 (Fig. 2B) than in the other mutant or complemented strains (Fig. 2B), which displayed the same band profile as the parental strain (Fig. 2B). With regard to iELISA, further differences were found between strains in the reactivity with the two anti-Omp25 MAbs (Fig. 2D). Thus, while the omp31 strain and its complemented strain showed slightly higher reactivity with MAb C09, inactivation of omp25c, omp25d, and omp22 led to a lower reactivity with this MAb, which persisted after complementation with the respective wild-type genes (Fig. 2D). The opposite situation was observed with MAb F04. This MAb did not react with parental B. ovis PA or with the derived PNV25A or PNV25A-com strains. The absence of reactivity with parental and PNV25A-com complemented strains was not surprising, since MAb F04 was previously shown to react in ELISA much more weakly with the B. ovis Omp25 protein than with the B. melitensis protein and not to react at all with B. ovis Omp25 located in the OM of E. coli (10). Surprisingly, however, the mutant omp31 strain, and especially the omp25c, omp25d, and omp22 strains, reacted intensely with MAb F04 (Fig. 2D). This was also the case for the corresponding complemented mutant strains, with the exception of strains PNV31A-com and especially PNV25cA-com, which, although showing reactivities higher than that of the parental strain, afforded reactions less intense than those observed for their corresponding mutant strains (Fig. 2D).
Role of the proteins of the Omp25/Omp31 family in the OM properties of B. ovis PA. Several tests related to the OM properties of Brucella spp. (i.e., autoagglutination and resistance to polymyxin B, sodium deoxycholate, hydrogen peroxide, and nonimmune ram serum) that might be linked to virulence were performed with all the B. ovis strains obtained in this work, and the results were compared with those given by parental B. ovis PA and B. abortus RB51.
When bacterial suspensions were incubated statically at room temperature, the parental rough strain B. ovis PA remained in suspension throughout the experiment, while the rough strain B. abortus RB51 settled quickly, showing a reduction of about 60% in the OD600 after 4 h of incubation and an almost complete clearance after 18 h (Fig. 3). B. ovis PA omp31 and omp25c mutant strains, and to lesser extent mutant PNV25A, also showed a strong capacity to autoagglutinate (Fig. 3). Complementation of mutant strains PNV25A and PNV25cA with wild-type omp25 and omp25c, respectively, restored the phenotype of the B. ovis PA parental strain. However, complementation of strain PNV31A with wild-type omp31, although reducing its autoagglutination properties, did not completely recover the characteristics of the parental strain (Fig. 3). Finally, both the PNV25dA and PNV22A mutant strains and the corresponding complemented strains behaved like the parental strain, remaining in suspension throughout the experiment (data not shown).
|
|
omp31 strain, showed percentages of survival to this detergent that were significantly lower than that observed for parental B. ovis PA (Table 3). The omp25c and omp22 mutant strains were the most susceptible strains (P < 0.0005), with percentages of survival of 39.21% ± 7.92% and 30.68% ± 3.60%, respectively, and none of the complemented mutant strains recovered the resistance levels observed for the parental strain, proving to be as susceptible to deoxycholate as the corresponding mutant strains were (Table 3). Regarding susceptibility to hydrogen peroxide, all the B. ovis PA Omp25/Omp31 family mutant and complemented mutant strains were more susceptible (P < 0.0005) than the parental strain (Table 3). B. abortus RB51 was more resistant (P < 0.005) than B. ovis PA and therefore more resistant than the B. ovis PA mutants (Table 3). Complementation of the mutant strains with the corresponding wild-type genes failed to restore the behavior of the parental strain and, in particular, the complemented mutant strain B. ovis PNV25cA-com was even more susceptible (P < 0.0005) to hydrogen peroxide than its corresponding mutant strain, PNV25cA.
Exposure for 4 h to nonimmune ram serum strongly reduced the viability of B. abortus RB51 (survival of 25.73% ± 4.84%), with no dramatic effect on the survival of parental B. ovis PA (survival of 85.52% ± 5.20%) (P < 0.0005 for the statistical differences between both strains) being observed (Table 3). All the B. ovis PA Omp25/Omp31 family mutant strains were more susceptible than the parental strain to the bactericidal action of serum, with the omp22 mutant strain displaying more defects in this regard (survival of 25.38% ± 3.98%; P < 0.0005). The properties of the complemented strain B. ovis PNV25A-com were similar to those of the parental strain. By contrast, the complementation of the omp31, omp25c, omp25d, and omp22 mutant strains with the corresponding wild-type genes did not restore the survival capacities of the parental strain to nonimmune ram serum, with strain B. ovis PN25cA-com proving to be even significantly more susceptible to the bactericidal action of ram serum than the corresponding omp25c mutant strain (survival of 17.51% ± 14.17% versus 55.68% ± 7.42%; P < 0.0005) (Table 3).
Role of the proteins of the Omp25/Omp31 family in the virulence of B. ovis PA in mice. The virulence of the B. ovis PA Omp25/Omp31 family mutant strains obtained in this work was evaluated in a BALB/c mouse model by intraperitoneal inoculation of 5 x 106 CFU/mouse and subsequent determination of bacterial counts in spleen at selected intervals p.i. Mice inoculated with the B. ovis PA parental strain were used as controls. The exact inoculation doses (CFU/mouse) for each B. ovis strain, determined retrospectively, were as follows: for parental PA, 5.40 x 106; for PNV31A, 0.65 x 106; for PNV25A, 4.40 x 106; for PNV25cA, 5.50 x 106; for PNV25dA, 8.50 x 106; and for PNV22A, 7.00 x 106.
The omp25 and omp25c mutant strains showed kinetics of infection in spleen similar to that displayed by the parental strain, with maximum levels of infection between weeks 2 and 5 p.i. declining slowly thereafter until the end of the experiment at week 11 p.i. (Fig. 4A). Mutant strain PNV31A reached a level of splenic infection equivalent to that seen in the parental strain at week 1 p.i. and also between weeks 7 and 11 p.i. However, maximum spleen colonization was about 1 log lower for the omp31 mutant than for parental B. ovis PA (P < 0.0005 and P < 0.005 at weeks 3 and 5 p.i., respectively) (Fig. 4B).
|
omp25d mutant strain underwent a quick reduction in splenic bacterial counts after week 2 p.i., being 5.5 logs lower than those detected for the parental strain at 3 weeks p.i. (P < 0.0005) and undetectable thereafter in any mice (Fig. 4A). Even more dramatic was the attenuation observed for the omp22 mutant strain, since this strain was isolated from only two out of five mice at week 1 p.i., and it was completely cleared from spleens after week 2 p.i. (Fig. 4A). The complemented mutant strains B. ovis PNV25dA-com and PNV22A-com inoculated into mice at doses of 5 x106 CFU/mice had bacterial counts in spleen at weeks 2 and 9 p.i. that were similar to those observed for the corresponding mutant strains at the same times p.i. (data not shown). |
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The B. ovis PA mutant strain with omp25d inactivated displayed behavior similar to that of the parental strain as regards in vitro growth, in both solid and liquid culture media (Fig. 1A and data not shown). Accordingly, the marked reduction in virulence observed for this B. ovis mutant in mice (Fig. 4A) cannot be justified on this basis. The in vitro growth of the omp22 strain was also similar to that of the parental strain, with the exception of a marked reduction in viability in the stationary phase (Fig. 1B). This fact could denote difficulties for the bacterium to grow in deprived media and/or at acid pH, conditions that it would find inside the host cells, and this could be related to the attenuation observed for the B. ovis PA strain lacking Omp22 (Fig. 4A).
The integrity of the OM is essential for the viability of bacteria. Thus, the removal of one OMP of the Omp25/Omp31 family might lead to a compensatory effect by increasing the synthesis of other members of this family. The possibility of a compensatory effect, analyzed only for the levels of Omp25 and Omp31 (major OMPs for which MAbs are available), was detected only in the omp31 mutant strain. In this strain, the levels of Omp25 seemed to be increased (Fig. 2B and D), as was also the case for the strain complemented with wild-type omp31 (Fig. 2B and D), which also seemed to have levels of Omp31 lower than those for parental B. ovis PA. However, these results should be considered with caution, since the synthesis of the other members of the Omp25/Omp31 family was not evaluated and the limitations to immunological detection of the protein must be taken into account (i.e., the reactivity of a MAb may be influenced by the particular conformation of the protein in each strain, even under the denaturing conditions of sodium dodecyl sulfate-polyacrylamide gel electrophoresis). The existence of a compensatory effect has been described for the genus Bartonella for a group of homologous proteins that are also homologous to the Brucella spp. Omp25/Omp31 family (39). Additionally, there is a work reporting that the inactivation of omp25c, omp25d, and omp22 in B. suis 1330 seems to increase the levels of Omp25b (42), but this effect is not possible in B. ovis, since this Brucella species lacks Omp25b (46). A B. suis strain with the omp31 gene inactivated was not obtained in this previous study (42), and it is therefore impossible to compare the results obtained in our work with B. ovis.
The results for the reactivity in iELISA of MAbs specific to Omp25 and Omp31 with the B. ovis PA Omp25/Omp31 family mutant strains suggest that the conformation of Omp31 and especially of Omp25 would be modified when some of the other OMPs of the family are removed from the OM. The change in the conformation, which also indicates alterations in the OM, was particularly evident when the anti-Omp25 MAbs C09 and F04 were used. The specific epitope of MAb C09 seems to be either less exposed at the surface or disposed in a different conformation in some mutant strains compared to what is seen for parental B. ovis PA (Fig. 2D). The reactivity of MAb F04 in iELISA has previously been shown to be stronger with Omp25 from B. melitensis than with the B. ovis protein, and no reactivity was detected between this MAb and recombinant E. coli bearing the B. ovis Omp25 protein at the surface (10). Surprisingly, although this MAb did not react with parental B. ovis PA or with mutant strain B. ovis PNV25A, it reacted strongly with the other B. ovis PA Omp25/Omp31 family mutants (Fig. 2D). In light of the results obtained from Western blotting and iELISA with the anti-Omp25 MAb C09, this higher reactivity of MAb F04 (except perhaps for mutant PNV31A) does not seem to be related to increased levels of Omp25 but rather seems related to a different topology of Omp25, suggesting modifications of the OM in these B. ovis strains.
Accordingly, the OM alterations detected by reactivity with the MAbs specific for Omp25 and Omp31 do not by themselves justify the marked reduction in virulence observed with mice for the omp25d and omp22 mutant strains, since the omp25c mutant strain exhibited a similar pattern of reactivity and was not attenuated in mice (Fig. 4A).
The autoagglutination properties could not be related with either virulence or the rough phenotype, since parental rough B. ovis PA and the attenuated strains B. ovis PNV25dA and PNV22A did not agglutinate, while rough B. abortus RB51 settled rapidly (Fig. 3 and data not shown). Taking into account that the omp31 mutant of B. ovis PA displayed an autoagglutination pattern similar to that observed for B. abortus RB51 (Fig. 3) and that it has been shown that B. abortus lacks Omp31 (49, 51), the autoagglutination phenomenon of B. abortus RB51 could be related to the rearrangements of the OM caused by the absence of Omp31. However, the possibility of other differences in the OM, related or not to the Omp25/Omp31 family, that might help to explain the different autoagglutination properties of both strains cannot be ruled out. In this respect, other divergences between B. abortus and B. ovis with respect to the members of the Omp25/Omp31 family have been described previously (i.e., the absence of Omp31b and the presence of Omp25 of a smaller size in B. ovis) (10, 46). It has been also observed that the presence of O-polysaccharide chains in LPS prevents the autoagglutination phenomenon, since other smooth strains, including smooth B. abortus 2308, from which rough strain RB51 derives, remain in suspension (data not shown).
The particular properties of the OM are thought to be responsible for the resistance of Brucella spp. to the bactericidal action of cationic peptides compared to what is seen for other gram-negative bacteria, such as enterobacteria (27, 37, 38). This resistance has been associated in part with the presence of O-polysaccharide chains in smooth strains, which would reduce the surface exposure of inner anionic groups (38), although the core-lipid A complex also plays a major role (33, 38). As in previous reports (27, 28, 38, 43), polymyxin B was used in this work to determine the stability of the OM and also as an indicator of the susceptibility of the B. ovis PA mutants to the bactericidal effect mediated by cationic peptides in the host. Surprisingly, rough parental B. ovis PA was much more resistant to a 1-h exposure to 1 mg/ml polymyxin B than rough B. abortus RB51, which did not survive this treatment (Table 3), and it was even slightly more resistant than smooth B. abortus 2308 (data not shown). The discrepancies with the results described in other works reporting that B. ovis Reo 198 is more susceptible to polymyxin B than smooth B. abortus S19 and behaves like rough B. abortus 45/20 (28, 38) might be related to the slight differences between the experimental protocols and/or the different B. ovis and B. abortus strains used. It should be noted that B. ovis Reo 198, employed in other works, is an atypical B. ovis strain, since it does not require a CO2 atmosphere or serum supplementation in the culture medium to grow and it is in addition an avirulent strain (28, 37, 38). In any case, results presented in this work regarding polymyxin B susceptibility reveal differences in the OM properties between B. abortus RB51 and B. ovis PA that do not depend on the presence or absence of O-polysaccharide chains at the surface, since both strains exhibit a rough phenotype.
Regarding the B. ovis PA Omp25/Omp31 family mutant strains, a higher susceptibility to polymyxin B cannot explain the attenuation observed in mice for the omp25d and omp22 B. ovis PA mutant strains, since they behaved like the parental strain (Table 3) and the fully virulent omp25c mutant was significantly more susceptible. However, the possibility that the omp25d and omp22 strains might be more susceptible than parental B. ovis PA to other bactericidal cationic peptides present in mammals cannot be ruled out. On the other hand, the complemented strain PNV25cA-com, which recovered the synthesis of Omp25c but at levels higher than those observed for the parental strain (unpublished results), was significantly more susceptible to polymyxin B than the corresponding mutant strain, suggesting that balanced levels of Omp25c would be required for the stability of the B. ovis PA OM.
Again, no explanation for the attenuation of the omp25d and omp22 strains was found on the basis of susceptibility to sodium deoxycholate, since both strains, although more susceptible than the parental strain (Table 3), did not show any differences with the omp25c strain, which is fully virulent in mice (Fig. 4A). Mutant strain PNV25A was also more susceptible to this anionic detergent, and complementation of the four susceptible mutant strains with the respective wild-type genes did not restore the resistance properties of parental B. ovis PA in any case. Since recoveries of the synthesis of Omp25, Omp25c, and Omp22 were detected in the corresponding complemented strains (Fig. 2 and unpublished results), these results again suggest that balanced levels of the members of the Omp25/Omp31 family would be important to maintain the stability of the OM.
In the present work, susceptibility to hydrogen peroxide was selected as an indicator both of the stability of the OM and of the ability of the B. ovis PA Omp25/Omp31 family mutant strains to resist killing by the oxygen-dependent bactericidal mechanisms of phagocytes. All the B. ovis mutants, and also all the complemented mutants, were significantly more susceptible to H2O2 than the parental strain, again showing the difficulties involved in restoring the phenotype of parental B. ovis PA by complementation with the wild-type gene when the encoded protein is located in the OM. The loss of virulence caused by the absence of Omp25d or Omp22 from the B. ovis PA surface cannot be explained in terms of a higher susceptibility to hydrogen peroxide, although the possibility that the omp25d and omp22 mutants might be more susceptible to intracellular killing mediated by other oxygen-dependent mechanisms cannot be excluded. All the B. ovis strains, including the parental strain, were also more susceptible than B. abortus RB51 (Table 3), again showing up the differences between rough strains from different Brucella species.
Brucella spp. are also at least partially resistant to killing mediated by the action of host nonimmune serum (1, 14, 20, 24, 25), which may contribute to virulence. Since OM properties may influence the behavior of Brucella spp. towards nonimmune serum to a considerable extent, we have evaluated the role of the members of the Omp25/Omp31 family in the survival of B. ovis PA in the presence of nonimmune ram serum. Although the virulence assays were performed in a murine model, ram serum was selected for susceptibility studies because ram is the preferred host for B. ovis and because of the difficulty involved in obtaining the required amount of serum from mice.
Exposure for 4 h to nonimmune ram serum dramatically reduced the viability of rough B. abortus RB51, while the susceptibility of rough B. ovis PA was very low (Table 3). Smooth B. abortus 2308 has previously been reported to be more resistant than the rough strains of the same species to the bactericidal effect of bovine and human nonimmune serum (1, 14, 20, 25), and these differences have been attributed to the presence of O-polysaccharide chains in the LPS of smooth B. abortus 2308 (1, 14, 20, 25). Nevertheless, while rough B. abortus strains are susceptible to exposure to human nonimmune serum, rough B. melitensis mutant strains are not (25). Therefore, in addition to the presence or absence of O-polysaccharide chains in the LPS, other differences, probably in the bacterial surface, must exist between the Brucella species to account for their different levels of susceptibility to nonimmune serum. This idea is also supported by the differences between B. ovis PA and B. abortus RB51 found in this work (Table 3).
Considering that B. abortus lacks a DNA fragment of about 25 kb that includes omp31 (48, 49, 51), the absence of Omp31 or other proteins encoded in this DNA fragment in B. abortus has been suggested as a possible explanation for the differences between B. melitensis and B. abortus as regards their susceptibilities to nonimmune serum (25). In the present work, we observed that the omp31 mutant strain was more susceptible to killing by nonimmune ram serum than parental B. ovis PA, but it was also significantly more resistant than rough B. abortus RB51 (Table 3). Accordingly, by itself the absence of Omp31 in B. abortus RB51 does not explain its higher susceptibility to the bactericidal effect of nonimmune ram serum, and hence other differences, known or unknown, between the B. abortus and B. ovis rough strains used in this work must be involved in their different susceptibilities. In this respect, it should be noted that between the two Brucella species, in addition to the absence of Omp31 in B. abortus, there are other differences regarding the members of the Omp25/Omp31 family, such as the absence of Omp31b in B. ovis (46) and the presence of Omp25 of a smaller size in B. ovis (10).
The absence of Omp22 from the OM of B. ovis PA elicited an important reduction in its resistance to killing by nonimmune ram serum (Table 3), which could account at least in part for the almost complete loss of virulence observed in the murine model for the omp22 mutant strain (Fig. 4A). On the other hand, the attenuation of the omp25d mutant strain (Fig. 4A) cannot be explained exclusively on the basis of its higher susceptibility to nonimmune serum, since it was as susceptible as the fully virulent omp25c strain (Table 3; Fig. 4A). Again, complementation of the mutant strains with the corresponding wild-type genes of the Omp25/Omp31 family did not restore the resistance to nonimmune ram serum observed for parental B. ovis PA (Table 3), even in complemented mutant strains where the synthesis of the removed protein was detected. Additionally, strain PNV25cA-com was significantly more susceptible than the corresponding PNV25cA mutant strain (Table 3), which again shows that a tight balance of the members of the Omp25/Omp31 family is essential for maintaining the properties of the OM.
Regarding the role of the members of the Omp25/Omp31 family in the virulence of B. ovis PA in mice, removal of Omp31 from the OM had no dramatic effect on the persistence of the bacterium in spleen (Fig. 4B). The only remarkable effect was that in comparison with the parental strain, the mutant strain B. ovis PNV31A showed a reduction of about 1 log in the maximum levels of splenic colonization, which were obtained between weeks 2 and 5 p.i. (Fig. 4B). The fact that Omp31 has no relevant effect in the virulence of B. ovis, in spite of being a major OMP, is not surprising, considering that this protein is not synthesized in virulent B. abortus (49, 51) and that it is not essential for the virulence of B. melitensis Rev.1 in either the murine or sheep model (8, 31). It has been shown that Omp31 functions as a hemin-binding protein in B. melitensis, B. suis, and B. ovis, and its synthesis seems to be increased under iron-limiting conditions in the culture medium (15). Despite the relevance that the hemin-binding capacity of Omp31 might have for the virulence of Brucella spp., the results obtained in this work regarding the virulence in mice of the B. ovis PA omp31 mutant (Fig. 4B), together with those reporting that Omp31 is not necessary for the virulence of B. abortus or B. melitensis Rev.1 (8, 31, 49, 51), suggest that Omp31 would not be essential for iron uptake. However, the other members of the Omp25/Omp31 family might also have hemin-binding properties that would compensate the effect of the absence of Omp31 from the OM. In this respect, it should be noted that the hemin-binding protein HbpA from Bartonella quintana shares 32% amino acid sequence identity with Omp31 from B. melitensis (6) and that other four proteins homologous to HbpA and thought to be able to bind hemin exist in B. quintana (39).
The omp25 and omp25c mutant strains displayed kinetics of splenic colonization similar to that observed for the parental strain (Fig. 4A), and therefore Omp25 and Omp25c do not play an important role in the virulence of B. ovis PA, at least under our experimental conditions. The full virulence of the omp25c mutant strain was surprising, considering that it was the B. ovis PA mutant that showed in the tests performed in this work the most evident alterations of the OM (Fig. 2 and 3; Table 3) and that omp25c is highly conserved, with almost identical nucleotide sequences, in all Brucella species (46). On the other hand, the results obtained with the omp25 strain differ from those obtained in a previous work with a omp25 mutant of B. ovis LSU99 (18). The differences between the two studies might be related to the different genetic backgrounds of the respective B. ovis parental strains or to differences in the doses or routes of infection employed in each case. The first of these options is the most likely, since the infection levels reached in spleen with the parental strains were similar in both studies and it has been demonstrated that B. ovis PA splenic growth curves after intravenous or intraperitoneal infection are almost identical (34).
Some indirect evidence has been reported previously about the possible involvement of Omp25 in the virulence of Brucella spp., but this evidence might be dependent on the host and/or Brucella species. Thus, on one hand it has been shown that Omp25 of B. suis inhibits the production of TNF- in human macrophages (35) and that it is involved in the release of periplasmic proteins in acidic conditions (3) and that B. abortus attenuated mutants with the bvrS-bvrR two-component regulatory system impaired do not synthesize Omp25 (32). On the other hand, other works have shown that the protein factor responsible for the inhibition of the TNF- production is not active in mouse macrophages (5) and that attenuated mutants of B. suis bvrS-bvrR showing defects in intracellular multiplication in human macrophages do not present lower Omp25 levels (42).
As discussed above for each test, the marked reduction in virulence observed for the omp25d mutant strain cannot be explained on the basis of the results obtained in any of the analyses regarding the OM properties performed here. Accordingly, Omp25d could be directly involved in the penetration and/or survival of B. ovis inside host cells. The same reasoning could be applied to hypothesize the role of Omp22 in virulence, although as discussed before the higher susceptibility of the omp22 mutant strain to nonimmune serum and its difficulty in surviving during the stationary phase of growth could be related to the striking degree of attenuation observed for this B. ovis PA mutant strain in mice (Fig. 4A). It should also be mentioned that B. abortus attenuated mutants with the bvrS-bvrR two-component regulatory system impaired do not synthesize Omp22 (32).
Complementation of omp25d and omp22 mutant strains of B. ovis PA with the corresponding wild-type genes did not recover the virulence or the OM properties of the parental strain. Although the possibility of a spontaneous mutation being the origin of the defects observed for these mutant strains cannot be completely disregarded, it should be noted that a clear rho-independent transcription terminator is located downstream from all the genes of the Omp25/Omp31 family, and hence the probability of a polar effect on a downstream gene in B. ovis PNV25dA and PNV22A seems to be very low. Moreover, the gene downstream from omp22 is located in the opposite orientation in the Brucella spp. genome. Additionally, by reactivity with specific antisera (unpublished results), Omp25d has not been detected in the complemented strain PNV25dA-com, and very low levels of Omp22 were detected in the complemented strain PNV22A-com in comparison to what was detected for the parental strain. Complementation in trans of Brucella mutant strains with wild-type genes coding for OMPs has frequently proved unsuccessful (3, 18, 44) due to the difficulty involved in recovering levels of the surface protein exactly the same as those found in the parental strain, which is probably essential for maintaining the integrity of the OM, especially in rough strains. This was clearly seen in complemented strain PNV25cA-com, which recovered the synthesis of Omp25c—although at levels higher than those found for parental B. ovis PA (unpublished results)—but had even more OM defects (Table 3) than the corresponding omp25c mutant strain.
The results reported here are pioneering in the study of the virulence factors involved in the pathogenicity of B. ovis, which unlike rough mutant strains obtained from smooth virulent Brucella spp. (1, 26, 30, 40, 45) is a natural rough Brucella species that is virulent in its preferred host and in animal models. Taking into account that Omp25d and Omp22 are essential for the virulence of B. ovis PA in mice and that both proteins are highly conserved with almost identical amino acid sequences in all Brucella species (46), the possibility of a role in the virulence of the other Brucella species should be considered. It would be also of great interest to determine the relevance for the virulence of B. ovis in rams not only of Omp25d and Omp22 but also of the other members of the Omp25/Omp31 family. These aspects, together with the role of these OMPs in intracellular survival, will be addressed in a forthcoming study to be conducted in our laboratory.
|
ACKNOWLEDGMENTS |
|---|
This work was financed by projects G03/204 (Red para Investigación en Brucelosis) and PI050091 from the Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo, Spain and project SA113A06 from the Junta de Castilla y León, Spain. Paola Caro-Hernández and Ana I. Martín-Martín were supported by projects G03/204 and PI050091, respectively.
|
FOOTNOTES |
|---|
Published ahead of print on 11 June 2007.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| 1. | Allen, C. A., L. G. Adams, and T. A. Ficht. 1998. Transposon-derived Brucella abortus rough mutants are attenuated and exhibit reduced intracellular survival. Infect. Immun. 66:1008-1016. |
| 2. | Alton, G. G., L. M. Jones, R. D. Angus, and J. M. Verger. 1988. Techniques for the brucellosis laboratory. Institut National de la Recherche Agronomique, Paris, France. |
| 3. | Boigegrain, R. A., I. Salhi, M. T. Alvarez-Martinez, J. Machold, Y. Fedon, M. Arpagaus, C. Weise, M. Rittig, and B. Rouot. 2004. Release of periplasmic proteins of Brucella suis upon acidic shock involves the outer membrane protein Omp25. Infect. Immun. 72:5693-5703. |
| 4. | Bowden, R. A., A. Cloeckaert, M. S. Zygmunt, S. Bernard, and G. Dubray. 1995. Surface exposure of outer membrane protein and lipopolysaccharide epitopes in Brucella species studied by enzyme-linked immunosorbent assay and flow cytometry. Infect. Immun. 63:3945-3952.[Abstract] |
| 5. | Caron, E., A. Gross, J. P. Liautard, and J. Dornand. 1996. Brucella species release a specific, protease-sensitive, inhibitor of TNF-alpha expression, active on human macrophage-like cells. J. Immunol. 156:2885-2893.[Abstract] |
| 6. | Carroll, J. A., S. A. Coleman, L. S. Smitherman, and M. F. Minnick. 2000. Hemin-binding surface protein from Bartonella quintana. Infect. Immun. 68:6750-6757. |
| 7. | Cloeckaert, A., P. de Wergifosse, G. Dubray, and J. N. Limet. 1990. Identification of seven surface-exposed Brucella outer membrane proteins by use of monoclonal antibodies: immunogold labeling for electron microscopy and enzyme-linked immunosorbent assay. Infect. Immun. 58:3980-3987. |
| 8. | Cloeckaert, A., I. Jacques, M. J. Grilló, C. M. Marín, M. Grayon, J. M. Blasco, and J. M. Verger. 2004. Development and evaluation as vaccines in mice of Brucella melitensis Rev.1 single and double deletion mutants of the bp26 and omp31 genes coding for antigens of diagnostic significance in ovine brucellosis. Vaccine 22:2827-2835.[CrossRef][Medline] |
| 9. | Cloeckaert, A., J. M. Verger, M. Grayon, J. Y. Paquet, B. Garin-Bastuji, G. Foster, and J. Godfroid. 2001. Classification of Brucella spp. isolated from marine mammals by DNA polymorphism at the omp2 locus. Microbes Infect. 3:729-738.[CrossRef][Medline] |
| 10. | Cloeckaert, A., J. M. Verger, M. Grayon, M. S. Zygmunt, and O. Grépinet. 1996. Nucleotide sequence and expression of the gene encoding the major 25-kilodalton outer membrane protein of Brucella ovis: evidence for antigenic shift, compared with other Brucella species, due to a deletion in the gene. Infect. Immun. 64:2047-2055.[Abstract] |
| 11. | Cloeckaert, A., N. Vizcaíno, J. Y. Paquet, R. A. Bowden, and P. H. Elzer. 2002. Major outer membrane proteins of Brucella spp.: past, present and future. Vet. Microbiol. 90:229-247.[CrossRef][Medline] |
| 12. | Conde-Álvarez, R., M. J. Grilló, S. P. Salcedo, M. J. de Miguel, E. Fugier, J. P. Gorvel, I. Moriyón, and M. Iriarte. 2006. Synthesis of phosphatidylcholine, a typical eukaryotic phospholipid, is necessary for full virulence of the intracellular bacterial parasite Brucella abortus. Cell. Microbiol. 8:1322-1335.[CrossRef][Medline] |
| 13. | Connolly, J. P., D. Comerci, T. G. Alefantis, A. Walz, M. Quan, R. Chafin, P. Grewal, C. V. Mujer, R. A. Ugalde, and V. G. DelVecchio. 2006. Proteomic analysis of Brucella abortus cell envelope and identification of immunogenic candidate proteins for vaccine development. Proteomics 6:3767-3780.[CrossRef][Medline]< |
