https://orcid.org/0000-0003-1250-5228
Figures
Abstract
Live attenuated vaccines (LAVs) whose virulence would be controlled at the tissue level could be a crucial tool to effectively fight intracellular bacterial pathogens, because they would optimize the induction of protective immune memory while avoiding the long-term persistence of vaccine strains in the host. Rational development of these new LAVs implies developing an exhaustive map of the bacterial virulence genes according to the host organs implicated. We report here the use of transposon sequencing to compare the bacterial genes involved in the multiplication of Brucella melitensis, a major causative agent of brucellosis, in the lungs and spleens of C57BL/6 infected mice. We found 257 and 135 genes predicted to be essential for B. melitensis multiplication in the spleen and lung, respectively, with 87 genes common to both organs. We selected genes whose deletion is predicted to produce moderate or severe attenuation in the spleen, the main known reservoir of Brucella, and compared deletion mutants for these genes for their ability to protect mice against challenge with a virulent strain of B. melitensis. The protective efficacy of a deletion mutant for the plsC gene, implicated in phospholipid biosynthesis, is similar to that of the reference Rev.1 vaccine but with a shorter persistence in the spleen. Our results demonstrate that B. melitensis faces different selective pressures depending on the organ and underscore the effectiveness of functional genome mapping for the design of new safer LAV candidates.
Author summary
Brucellosis is one of the most widespread bacterial zoonoses worldwide. We report here the use of transposon sequencing to compare the bacterial genes involved in the multiplication of Brucella melitensis, a major causative agent of brucellosis, in the lungs and spleens of infected mice. We found 257 and 135 genes predicted to be essential for B. melitensis multiplication in the spleen and lung, respectively, with 87 genes common to both organs, which demonstrates that B. melitensis faces different selective pressures depending on the organ. Then, we selected genes whose deletion is predicted to produce moderate or severe attenuation in the spleen, the main known reservoir of Brucella, and compared deletion mutants for these genes for their ability to protect mice against challenge with a virulent strain of B. melitensis. We observed that the protective efficacy of a deletion mutant for the plsC gene, implicated in phospholipid biosynthesis, is similar to that of the reference Rev.1 vaccine but with a shorter persistence in the spleen. Our results demonstrate the effectiveness of functional genome mapping for the design of new safer live attenuated vaccine candidates.
Citation: Barbieux E, Potemberg G, Stubbe F-X, Fraikin A, Poncin K, Reboul A, et al. (2024) Genome-wide analysis of Brucella melitensis growth in spleen of infected mice allows rational selection of new vaccine candidates. PLoS Pathog 20(8): e1012459. https://doi.org/10.1371/journal.ppat.1012459
Editor: Richard B.S. Roden, Johns Hopkins University School of Medicine, UNITED STATES OF AMERICA
Received: January 3, 2024; Accepted: July 29, 2024; Published: August 26, 2024
Copyright: © 2024 Barbieux et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: We have shared the fastq files containing reads from our Tn-seq analyzes on the Figshare site (available at https://doi.org/10.6084/m9.figshare.26063158.v1). All of the results of our transposon sequencing analyzes are available in S1 Table which is submitted with the manuscript.
Funding: This work was supported by grants from the Fonds National de la Recherche Scientifique (FNRS) (CDR J.0120.18 and CDR J.0157.20 to E.M. and PDR T.0058.20 to X.D.B., Belgium). E.M. is a Senior Research Associate from the FRS-FNRS (Belgium). E.B., G.P. and F-X.S. hold FRIA PhD grants from the FRS-FNRS (Belgium). The funders played no role in study design, data collection, analysis and interpretation of data, or the writing of this manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Brucellae are small Gram-negative facultative intracellular bacteria belonging to the Rhizobiales order within the α2-proteobacteria subgroup [1,2]. They are the causative agent of brucellosis, a common bacterial zoonotic disease responsible for important economic losses and public health issues, particularly in low- and middle-income countries [3–6]. B. melitensis is the species most often involved in ovine and caprine brucellosis and is also the most pathogenic species for humans [7].
Due to the impact of this disease on public health and the damage that it causes to the livestock industry, much effort has been expended to control or eradicate brucellosis in cattle and small ruminants. Three commercially available live attenuated vaccines (LAVs) are used to control Brucella infection in domestic animals: B. abortus S19 and RB51 to prevent brucellosis in cattle, and B. melitensis Rev.1 to protect sheep and goats. While the effectiveness of RB51 is controversial [8], S19 and Rev1 have been used in livestock wherever eradication has been successful. However, both vaccines have serious drawbacks [8–11]. Their attenuation was obtained empirically, and even accidentally for S19, which makes their virulence quite unpredictable. In some cases, they persist for years in the vaccinated host, where they can induce abortions and excretion in milk in animals, posing a high risk to humans for whom they are virulent [12]. Since in low- and middle-income regions mass vaccination is the only way to control the disease, which implies vaccination of pregnant animals, the control of brucellosis is therefore more complicated in these areas [13]. To encourage research in this area, the Bill and Melinda Gates foundation has offered a large prize to reward any new brucellosis vaccine that presents a safer profile.
The development of safer vaccines should involve functional mapping of the Brucella genome to identify the virulence genes that are essential at the different stages of Brucella infection in animals. This would allow for the rational development of a vaccine that persists just long enough in the host to induce frontline and systemic protective immune memory but does not remain long term in the reservoir organs.
Transposon sequencing (Tn-seq) is a powerful approach to rapidly and comprehensively determine an organism’s minimal genetic requirements for growth and survival under a variety of different conditions [14,15]. Previously, we performed various Tn-seq screens of B. melitensis 16M and identified a set of genes predicted as important for growth in 2YT nutrient rich media, in vitro in the RAW 264.7 macrophage cell line and in vivo in lungs from intranasally infected wild-type C57BL/6 mice [16]. Our main findings were that the genes that are essential for the multiplication of B. melitensis in vitro and in vivo are different and vary according to the immune status of the host.
Numerous observations suggest that the host is not a homogeneous environment for Brucella. In our mouse model, the early phase of the protective immune response against B. melitensis varies according to the organs: at 120 hours post-infection, an IL-17RA-dependent (Th17) response controls the infection in the lungs [17] and an IFNγ-dependent (Th1) response controls the infection in the spleen [18]. B. melitensis persists for more than 50 days in the spleen of infected mice but disappears after a few weeks from the lungs [19,20]. In addition, confocal microscopy analysis has demonstrated that Brucella-infected cells in the lung [16,21] and spleen [22,23] display a very different phenotype. Thus, we hypothesize that both the available nutrients as well as the microbicidal mechanisms encountered by B. melitensis in the lungs and spleen are different and that therefore a different set of bacterial genes might be required for the multiplication of Brucella in each organ. To test this hypothesis in this study we used Tn-seq screens to identify genes contributing to the fitness of B. melitensis 16M in spleen from wild-type and IFNγ-R-/- C57BL/6 infected mice. By comparing these genes to those previously found in the lungs [16], we can select genes specifically necessary for the multiplication of B. melitensis in the spleen and test whether they can help us to develop a safer brucellosis vaccine.
Results
Tn-Seq identification of bacterial genes affecting the fitness of B. melitensis in the spleen of wild-type mice
As described previously [16], a B. melitensis 16M library of 3×106 random mutants was constructed using a KanR derivative of the mini-Tn5 transposon. This library was exposed to different selection conditions, such as culture in a 2YT rich medium for 24 hours, and in lungs from intranasally infected C57BL/6 mice for 120 hours. Following selection, surviving bacteria were collected and DNA was extracted and sequenced to identify mini-Tn5 insertion sites. For each selection condition, an average insertion index, called the transposon insertion frequency (TnIF), was associated with each gene of the B. melitensis genome. The TnIF is equal to the log10 (r+1) for a given coding sequence, with r = number of reads aligned in the central 80% of the coding sequence of the considered gene, as described in Materials and Methods.
The 2YT condition is the control condition (CTRL) for all our Tn-seq analyses of B. melitensis 16M, because whatever the condition analyzed, the bacteria are always cultured in 2YT after mice infection. Genes exhibiting a drop in TnIF of more than 0.5 compared to the average TnIF for the entire genome in the 2YT condition are considered to cause a significant attenuation of B. melitensis in 2YT. The threshold of 0.5 was determined based on the standard deviation of the bimodal TnIF distribution of the whole genome. It corresponds to 1.6–2.6 standard deviations, as described in the Materials and Methods. According to this criterion, 2460 of the 3369 genes of B. melitensis 16M are considered not to cause significant growth defect in the 2YT condition. Only these genes are analyzed under the other selection conditions.
ΔTnIF is calculated to quantify the attenuation of a gene in a condition other than 2YT. ΔTnIF = TnIFcdt−TnIFCTRL, where TnIF is computed for the test condition (TnIFcdt) and the control condition (TnIFCTRL).
In the present study, wild-type C57BL/6 mice were intraperitoneally infected with 5x106 CFU of our B. melitensis 16M transposon mutant library and sacrificed at 120 hours post-infection, what corresponds to the peak of infection in the spleen [18]. As described previously [16], this high dose is necessary to administer the entire bank to each mouse and avoid bottleneck effects. Bacteria were extracted from this organ and analyzed as indicated in the Materials and Methods. The intraperitoneal route of infection was used in order to avoid the risk of a bottleneck which would lead to a sampling effect in the library. The TnIF values for each gene in the spleen condition were compared with those obtained previously [16] in the 2YT and lung conditions (S1 Table, page 1). A ΔTnIF for each gene in the lung and spleen conditions was calculated, as indicated previously and summarized in the Material and Methods section.
A dot plot representation was used to visualize the differences between the lung and spleen conditions, where each dot corresponds to a B. melitensis gene defined by ΔTnIF values in the lung and spleen (Fig 1A). Among the 2460 genes analyzed (S1 Table, page 2), 257 genes are predicted, if inactivated, to cause a significant drop in B. melitensis fitness in the spleen condition and the same prediction was made for 135 genes in the lung condition (S1 Table, page 3 and 4, respectively). Only 87 genes are common to these two conditions (S1 Table, page 5), which implies that 170 genes are specific to the spleen condition (S1 Table, page 6) and 48 to the lung condition (S1 Table, page 7).
The figure shows the distribution of the ΔTnIF values of all B. melitensis genes (n = 2460) predicted not to induce an attenuation of fitness in 2YT rich medium (CTRL). Each gene is defined by two ΔTnIF values. (A) The x-axis indicates the ΔTnIF value for the lung of wt mice (TnIFlung—TnIFCTRL) and the y-axis indicates the ΔTnIF value for the spleen of wt mice (TnIF spleen wt—TnIFCTRL) at 120 hours post-infection. These ΔTnIF value comparisons show all the genes associated with a drop in fitness both in the lung and wt spleen (brown area), in the lung specifically (green area) or in the wt spleen specifically (pink area). (B) The x-axis indicates the ΔTnIF value for the spleen of IFNγR-/- mice (TnIF spleen IFNγR-/-—TnIFCTRL) and the y-axis indicates the ΔTnIF value for the spleen of wt mice (TnIF spleen wt—TnIFCTRL) at 120 hours post-infection. These ΔTnIF value comparisons show all the genes restored in the IFNγR-/- spleen (pink area).
A clustering analysis with the STRING database was carried out on the genes presenting a strong drop (-1.0) in ΔTnIF. The 48 genes displaying a ΔTnIF < -1.0 in both the lungs and the spleen included 6 gene clusters (Fig 2). These include a cluster of 9 genes involved in purine (purA,B,C,E,H,Q,S) and histidine (hisA,H) synthesis, a cluster of 2 genes associated with methionine transport (metN,I), a cluster of 2 genes involved in central carbohydrate metabolism (fba, ppdK), a cluster of 2 genes involved in fatty acid oxidation (fadA,J), a cluster of 12 genes necessary for the synthesis of the lipopolysaccharide (LPS) (galU, gmd, per, pgm, manBcore/rfbK, wboA, wboB, wbkD, wbpZ/wbkE, wbkA, wbpL/wbkF, wzt) and a cluster of 12 genes forming the virB operon (vjbR, virB1-11) and encoding the type IV secretion system VirB and its transcriptional regulator VjbR [24,25].
The diagram shows the potential interactions between the 48 genes displaying a ΔTnIF < -1.0 identified in the lungs and spleen of wild-type mice at 120 hours post-infection. The color code represents the different pathways. This clustering analysis was carried out with the STRING database (https://string-db.org).
Among the 12 genes displaying a ΔTnIF < -1.0 in the lungs and > -0.5 in the spleen, a clustering analysis identified a cluster of 6 genes (trpA,B,C,D,E,F) involved in tryptophan synthesis, a cluster of 2 genes (bveA, mprF) associated with polymyxin resistance and a cluster of 2 genes (ctaA,G) associated with respiration that are indispensable in the lungs and not in the spleen (Fig 3A).
The diagram shows the potential interactions between genes displaying a ΔTnIF < -1.0 identified in the lungs or spleen of wild-type mice at 120 hours post-infection (12 genes in the lungs (A) and 34 genes in the spleen (B)). The color code represents the different pathways. This clustering analysis was carried out with the STRING database (https://string-db.org).
Regarding the genes displaying a ΔTnIF < -1.0 in the spleen and > -0.5 in the lungs, we identified 4 small clusters (Fig 3B). A cluster of 3 genes (tac4, gltB, hisE) associated with amino acid biosynthesis, a cluster of 3 genes (znuA,B,C) encoding a zinc transporter, a cluster of 3 genes (czcA, triA, BMEI0381) identified as an efflux transporter and a cluster of 2 genes (BMEI1934, BMEI1935) identified as an ABC transporter. Other 23 genes could not be linked to a gene cluster and 5 of them were unidentified before this study.
To summarize our Tn-seq data, we produced schematic representations of the LPS synthesis pathway (Fig 4) and of the central carbohydrate metabolism (Fig 5) and attempted to give a general view of the metabolism of B. melitensis by connecting these pathways to the tricarboxylic acid cycle, glutamine metabolism as well as the histidine and adenine/adenosine synthesis pathways (Fig 6). Interpretations are proposed in the Discussion section.
Schematic representation of the lipopolysaccharide biosynthesis pathway of B. melitensis with genes attenuated in 2YT rich medium highlighted in red and genes required in the lung and spleen highlighted in blue (adapted from the KEGG PATHWAY database, https://www.genome.jp/kegg/pathway.html).
Schematic representation of the glycolysis, pentose phosphate and Entner-Doudoroff pathways in B. melitensis with genes attenuated in 2YT rich medium highlighted in red and genes attenuated in the lung and spleen, required only in the lung or only in the spleen respectively in blue, pink and green (adapted from the KEGG PATHWAY database, https://www.genome.jp/kegg/pathway.html).
Schematic representation of lipopolysaccharide (LPS) synthesis, the pentose phosphate (PP) pathway, the fatty acid β oxidation, glucose (G) pathway and the tricarboxylic acid (TCA) cycle in B. melitensis with genes attenuated in 2YT rich medium highlighted in red and genes attenuated in the lung and spleen, required only in the lung or only in the spleen respectively in blue, pink and green (adapted from the KEGG PATHWAY database, https://www.genome.jp/kegg/pathway.html).
We attempted to confirm certain predictions from our Tn-seq analyses by producing deletion mutants of a few genes of interest. We selected trpD (BMEI0843), predicted to cause attenuation in the lungs; znuA (BMEII0178) and lysR (BMEI0513), predicted to cause attenuation in the spleen; and gmd (BMEI1426), purH (BMEI0233) as well as the virB operon genes, predicted to cause attenuation in both the lungs and the spleen. The wild-type (wt), Δgmd, ΔlysR, ΔpurH, ΔtrpD, ΔvirB and ΔznuA strains were intranasally administered at a dose of 5x106 CFU to wild-type C57BL/6 mice and the number of bacteria present in the lungs at 120 hours post infection was measured by CFU counting (Fig 7A). The same strains were administered at the same dose by the intraperitoneal route to wild-type C57BL/6 mice and the number of CFUs in the spleen at 120 hours was determined (Fig 7B). The Tn-seq predictions for each of these genes in the lungs and spleen are shown in Fig 7C and 7D, respectively. We observed that the level of persistence of deletion mutants in the lungs and spleen was qualitatively well predicted by our Tn-Seq data. The Δgmd, ΔpurH and ΔvirB strains were significantly attenuated compared to the wild-type strain in the lungs and spleen while the ΔtrpD strain was attenuated only in the lungs and the ΔlysR and ΔznuA strains were only attenuated in the spleen. Overall, our data demonstrate that B. melitensis 16M is subject to different selection pressures in the lungs and spleen of C57BL/6 mice.
Data shown in panels (A) and (B) are bacterial counts (CFU) at 120 hours post-infection in (A) lungs from wild-type mice infected intranasally and (B) the spleen from wild-type mice infected intraperitoneally with wild-type (wt), ΔtrpD, ΔlysR, ΔznuA, ΔvirB, Δgmd and ΔpurH strains of B. melitensis at the dose of 5x106 CFU. Red lines represent the geometric mean. Dotted lines represent the mean of the wild-type condition. Significant differences between wt and the indicated groups are marked with asterisks: *p < 0.1, ***p < 0.001, in a (Wilcoxon-)Mann-Whitney post-test. These results are representative of three independent experiments. Data shown in panels (C) and (D) are the prediction derived from the Tn-Seq analysis expressed in ΔTnIF. ΔTnIF of genes in lungs from wild-type mice infected intranasally (TnIF lung—TnIFCTRL) are represented in (C). ΔTnIF of genes in the spleen from wild-type mice infected intraperitoneally (TnIF spleen—TnIFCTRL) are represented in (D). The color code of the legend indicates whether the genes are specifically attenuated in the lung (green), in the spleen (orange) or in both (blue) at 120 hours post-infection.
Absence of the IFN-γR dependent signaling pathway affects the nature of essential bacterial genes in the spleen
Interferon-γ (IFN-γ) is well known to control Brucella growth in vivo as well as Brucella-induced inflammation [26]. Using the mouse model of intraperitoneal infection by B. melitensis, we have previously shown that production of IFN-γ in the spleen presents a peak at 120 hours post-infection and is mainly due to CD4+ T cells [27]. In principle, comparison of the genes that are essential for the growth of B. melitensis in the spleen of wild-type and IFN-γR-/- mice should identify genes required for resistance to the IFN-γ-dependent immune response.
We infected IFN-γR-/- C57BL/6 mice with 5x106 CFU from our B. melitensis mini-Tn5 library. Mice were sacrificed at 120 hours post-infection, the spleen was collected, and bacteria were isolated and analyzed as described above. We point out that although the IFN-γ pathway is essential for the control of B. melitensis infection in the spleen [18], infected IFN-γR-/- C57BL/6 mice do not yet show any clinical signs at this time of infection. We then compared the ΔTnIF for the wild-type and IFN-γR-/- spleen conditions (S1 Table, page 2) for the 2460 genes which are not predicted to cause growth failure for B. melitensis in the 2YT condition using a dot plot presentation where each dot represents a gene (Fig 1B). Our results showed that among the 257 genes displaying a ΔTnIF < -0.5 in the wild-type spleen condition, only 102 displayed a ΔTnIF < -0.5 in the IFN-γR-/- spleen condition. Thus, 155 B. melitensis genes predicted to cause attenuation in spleen are not predicted to cause attenuation in the spleen of IFNγR-/- mice. Therefore, numerous genes seem to be involved in resistance to the IFN-γ-dependent immune response. S2 Table presents a list of the 19 genes presenting a ΔTnIF < -1.0 (very low fitness) in the wild-type spleen condition and a ΔTnIF > -0.5 (predicted not attenuated) in the IFNγR-/- spleen condition. Unfortunately, a clustering analysis of these genes using the STRING database did not allow us to identify any cluster of genes, making it impossible to propose a hypothesis explaining how these genes are involved in the resistance of B. melitensis to the Th1 immune response.
In order to validate our Tn-seq predictions in the IFNγR-/- spleen condition, we compared the persistence in the spleen of wild-type and IFN-γR-/- mice of a deletion mutant of the lysR gene (BMEI0513, also called lysR21 and vtlR [28]). This gene displayed a ΔTnIF in the wild-type condition of -0.73 and a ΔTnIF of +0.56 in the IFNγR-/- spleen condition. We infected wild-type and IFN-γR-/- mice with 5x106 CFU of B. melitensis 16M or the ΔlysR mutant. Evaluation of the number of bacteria at 120 hours post-infection confirmed the one log-attenuation of the ΔlysR strain in the spleen of wild-type mice while the mutant showed no attenuation in the spleen of IFN-γR-/- mice (Fig 8). These results fully validate our Tn-seq predictions regarding the lysR gene.
Data shown are bacterial count s(CFU) at 120 hours post-infection in the spleen from wild-type and IFN-γR-/- mice infected intraperitoneally with wild-type (wt) or ΔlysR strains of B. melitensis at a dose of 5x106 CFU. Red lines represent the geometric mean. Dotted lines represent the mean of the wild-type condition. Significant differences between wt and the indicated groups are marked with asterisks: ***p < 0.001, in a (Wilcoxon-)Mann-Whitney post-test. These results are representative of two independent experiments.
Selection of new vaccine candidates based on lung and spleen Tn-seq results
Ideally, to be protective, a brucellosis vaccine should multiply long enough in the lungs and spleen to induce mucosal and systemic immunity. Furthermore, an ideal vaccine should also not be able to persist in the spleen, which is the main known reservoir of Brucella in mice. To select genes whose deletion could lead to this type of behavior for B. melitensis, we should have Tn-seq data at different points of infection, for example at 5, 12 and days post infection. Unfortunately, in our model of B. melitensis infection, we only have Tn-seq data in the spleen up to 5 days post-infection. Beyond that period, the drop in the CFU count could lead to a bottleneck effect which causes sampling effects and makes the Tn-seq data uninterpretable. Therefore, we were not able to identify the B. melitensis genes necessary for a long-term chronic infection of the spleen. Accordingly, we decided to select genes whose inactivation is not expected to cause strong attenuation in the lungs, and which are expected to cause moderate or strong attenuation in the spleen at 120 hours post-infection. We hypothesized that genes thought to induce moderate attenuation at 120 hours in the spleen will cause stronger attenuation in the chronic phases of infection.
Bearing the above in mind, from the 3369 genes of the B. melitensis genome we selected the 2460 genes that, when inactivated, were not predicted to cause an attenuation in 2YT medium (TnIF > 2.7 in 2YT condition) (Fig 9). Then, we selected the 2387 genes predicted not to cause strong attenuation in the lungs when inactivated (ΔTnIF > -1.0 in the lung condition). From this category we selected the genes whose deletion was predicted to cause a moderate (ΔTnIF between -1.0 and -2.0 in wild-type spleen condition) or a strong attenuation (ΔTnIF between < -2.0 in wild-type spleen condition) in the spleen (Fig 9). Finally, we verified that these candidates remain predicted as attenuated in IFNγR-/- mice (ΔTnIF < -0.5 in the IFNγR-/- spleen condition) to avoid a possible increase in virulence of the vaccine strain in the event of a deficient Th1 response. From the 3 genes predicted to cause strong attenuation in the spleen (S1 Table, page 7) we selected murI (BMEI0795) and from the 17 genes predicted to cause moderate attenuation (S1 Table, page 8) we selected plsC (BMEI1977). We selected these two genes because although the enzymatic functions of the encoding proteins are known, their role in Brucella pathogenesis has, to our knowledge, not yet been investigated.
We selected genes whose deletion could optimize a strain of B. melitensis to safely induce protective immunity. In practice, we selected the genes able to produce deletion mutants that induce a systemic immune response but do not persist long term in the spleen reservoir. Only genes that are not required to grow on 2YT rich medium were kept (ΔTnIF > 2.7) (2460 genes). In this group, genes that are not strongly attenuated (ΔTnIF > -1.0) in the lungs were conserved (2387 genes). Two different thresholds were then applied for the spleen condition in wild-type (wt) mice: either strongly attenuated (ΔTnIF < -2.0) or moderately attenuated (-2.0<ΔTnIF<-1.0). In addition, the candidate must not revert in immune deficient mice (IFNγR-/-) (ΔTnIF < -0.5). ΔTnIF corresponds to the TnIF tested condition—TnIFCTRL (2YT rich medium).
Tn-seq predictions for these two genes were verified by constructing deletion mutants and measuring their persistence in the spleen (Fig 10). For that purpose, wild-type C57BL/6 mice were infected intranasally or intraperitoneally with 5×106 CFU of the wt, ΔmurI and ΔplsC strains and sacrificed at 120 hours post-infection, and the numbers of bacteria was evaluated in the lungs and spleen by CFU-counting. Our observations confirmed that inactivation of both genes leads to low attenuation in the lungs (-1.0 and -0.5 log CFU, respectively) and that inactivation of murI and plsC induces strong (-2.0 log CFU) and moderate (-1.3 log CFU) attenuation in spleen, respectively.
Data shown in panels (A) and (B) are bacterial counts (CFU) at 120 hours post-infection in (A) lungs from wild-type mice infected intranasally and (B) the spleen from wild-type mice infected intraperitoneally with wild-type (wt), ΔmurI and ΔplsC strains of B. melitensis at a dose of 5x106 CFU. Red lines represent the geometric mean. Dotted lines represent the mean of the wild-type condition. Significant differences between wt and the indicated groups are marked with asterisks: *p < 0.1, **p < 0.01, ***p < 0.001, ****p < 0.0001, in a (Wilcoxon-)Mann-Whitney post-test. These results are representative of three independent experiments. Data shown in panels (C) and (D) are the prediction derived from the Tn-Seq analysis expressed in ΔTnIF of genes in the lungs (C) and spleen (D) from wild-type mice infected intraperitoneally (TnIF spleen—TnIFCTRL).
To ensure that the observed phenotype was indeed consequence of the absence of the murI and plsC genes and not a non-specific effect caused by the deletion, we complemented these mutants with an intact coding sequence on a low copy plasmid, as indicated in the Material and Methods section. We observed that the complemented strains displayed a level of CFU in the lungs and spleen close to that of the wt strain (S1 Fig), demonstrating that the attenuation of the mutants was not due to off-target effects.
Finally, as we planned to test the protective capacity of these vaccine candidates in the intranasal vaccination model with a dose of 105 CFU characterized previously [17,20], we also tested the persistence of the ΔmurI and ΔplsC strains in the lungs and the spleen under these conditions. Wild-type C57BL/6 mice were infected intranasally with 105 CFU of the wt, ΔvirB, ΔmurI and ΔplsC strains as well as with the Rev.1 vaccine. The ΔvirB strain was used as a negative control due to its low persistence in the lungs and spleen. Comparison of our mutant strains with the Rev.1 vaccine will allow us to determine whether our vaccines persist less in animals than the reference vaccine and therefore whether we can expect these vaccine candidates to be safer than Rev.1. As expected, we observed (Fig 11) that the ΔvirB strain was rapidly eliminated from the lungs and did not colonize the spleen. The ΔmurI strain persisted for up to 12 days in the lungs but was unable to colonize the spleen. In contrast, the ΔplsC strain persisted in the lungs at the same level as the ΔmurI strain but succeeded in colonizing the spleen. At 12 days, the level of bacteria in the spleen from the ΔplsC strain and Rev.1 were close to that associated with the wt strain. However, ΔplsC bacterial burden dropped drastically at 28 days post-infection while the wt strain and Rev.1 persisted in the spleen. Under our vaccination conditions, the ΔmurI strain is therefore able to multiply in the lungs but is not able to colonize the spleen, whereas the ΔplsC strain can temporarily colonize the spleen but is unable to establish itself there over the long term.
Wild-type C57BL/6 mice were infected intranasally with 105 CFU of different B. melitensis strains: wild-type, ΔmurI, ΔplsC, Rev.1 or ΔvirB. This dose was used to mimic the infection following vaccination. Data shown are bacterial counts (CFU) at 5 (A), 12 (B) and 28 (C) days post-infection in the lung and at 5 (D), 12 (E) and 28 (F) days post-infection in the spleen. Red lines represent the geometric mean. Dotted red lines represents the mean of the wild-type strain. Significant differences between wt and the indicated groups are marked with asterisks: *p < 0.1, **p < 0.01, ***p < 0.001, ****p < 0.0001, in a (Wilcoxon-)Mann-Whitney post-test. These results are representative of three independent experiments.
Evaluation of the protective capacity of the ΔmurI and ΔplsC vaccine candidates
In order to assess the ability of the ΔmurI and ΔplsC vaccine candidates to induce a long-lasting protective memory, wild-type C57BL/6 mice intranasally received 105 CFU of B. melitensis wt, Rev.1, ΔvirB, ΔmurI and ΔplsC strains and were challenged 6 weeks later via the same route with 105 CFU of the virulent wt strain, as described in the Materials and Methods. This challenge strain expressed a fluorochrome and resistance to kanamycin in order to be differentiated from the vaccination strains. One group of mice (control) was not vaccinated. As expected, we observed that the wt and Rev.1 strains induced a very effective immune response that reduced the challenge strain load in the lungs and spleen by 3 log CFU (Fig 12). In contrast, the ΔvirB strain induced no or very weak detectable protective immunity in the lungs and spleen (-0.5 log CFU), respectively. Finally, ΔmurI induced only weak immunity in the lungs and spleen (-1 log CFU) whereas ΔplsC induced a protective immune response similar to the wild-type and Rev.1 strains.
Wild-type C57BL/6 mice were immunized intranasally with 105 CFU of different B. melitensis strains: wild-type, Rev.1, ΔvirB, ΔmurI or ΔplsC. Wild-type B. melitensis 16M strain and the Rev.1 vaccine were used as positive controls. ΔvirB was used as the negative control, as it is known to be rapidly eliminated from the body and thus does not have time to induce a memory response. ΔmurI and ΔplsC were the selected vaccine candidates. Unvaccinated (control) and vaccinated wild-type mice were challenged intranasally at 6 weeks post-vaccination with 105 CFU of wild-type mCherry–B. melitensis and sacrificed at 12 days post-infection. The data represent the number of CFU/g in the lung (A) and spleen (B). Red lines represent the geometric mean. Dotted lines represent the mean of CFU for unvaccinated mice. Significant differences between the control (unvaccinated) and groups of vaccinated mice are marked with asterisks: ***p < 0.001, ****p < 0.0001, in a (Wilcoxon-)Mann-Whitney post-test. These results are representative of three independent experiments.
Interestingly, these results could be correlated with the induction of specific Brucella IgG2a antibodies since protective strains (wt, Rev.1 and ΔplsC) induced high levels of antibodies while non-protective strains (ΔvirB, ΔmurI) did not (S2 Fig).
Discussion
Brucellosis is one of the most common bacterial zoonoses and has a significant economic and public health impact worldwide. Brucella infections cause abortions and sterility in ruminants and pigs and a severely debilitating febrile illness in humans [3–5], an accidental host that become infected through contact with infected animals or by consuming contaminated animal products. Since cases of human-to-human transmission are extremely rare, efforts to control brucellosis should first focus on farm animals. Control of brucellosis in northern European countries and the USA was possible through a policy combining mass vaccination and slaughter of infected herds, a strategy that is difficult to apply in many southern countries, like African countries, where the state cannot compensate breeders for the slaughter of herds or carry out sufficiently rigorous screening for infected animals. In these countries, the fight against brucellosis relies heavily on vaccines. Unfortunately, the only vaccines that are effective on the ground are live attenuated vaccines and those currently recommended, such as S19, RB51 and Rev.1, have several important shortcomings [9–11].
B. melitensis, which mainly infects goats and sheep, is the most pathogenic Brucella species for humans [7]. The live attenuated B. melitensis Rev.1 strain is recommended by the World Organization for Animal Health (WOAH) for vaccination of small ruminants since conjunctival Rev.1 administration successfully protects these animals [29]. However, this vaccine has several important drawbacks [9,30]. Rev.1 is a natural revertant of a virulent B. melitensis 6056 strain rendered dependent on streptomycin in the mid-1950s, one of the antibiotics recommended for the treatment of human brucellosis. Also, Rev.1 is a smooth strain and as such induces positive serology which interferes with the diagnosis of brucellosis, it causes dose dependent abortions in small ruminants [9,31], it can persist for a long time in the vaccinated animal and be excreted in the milk [32,33] and it is virulent for humans [34,35]. Comparative proteomic analysis of the virulent B. melitensis 16M strain with the attenuated Rev.1 strain in an acid medium in vitro showed that the two strains presented important differences [36]. At least 403 genes involved in complex cellular processes, like metabolism and transport, are expressed at different levels making very complicated to identify the mechanisms of the attenuation of Rev.1.
In this study, we tested the rational development of new attenuated vaccine candidates from the virulent B. melitensis 16M strain based on a functional map of its genome produced by Tn-Seq. We have identified the genes that are essential for the multiplication of B. melitensis 16M in the spleen from wild-type and IFNγR-/- C57BL/6 mice and compared them with the genes previously identified by our team as essential for multiplication in the lungs of C57BL/6 mice [16]. This comparison enabled us to identify the genes essential for the multiplication of B. melitensis in both the lungs and spleen as well as the genes specifically necessary for multiplication in the lungs or spleen. The predictions from our Tn-Seq analyses were validated by constructing deletion mutants for the genes of interest and measuring their persistence in the lungs and spleen of infected mice.
We collected a very large amount of information that will be of interest for future aspects of research on B. melitensis. Here, we attempt to summarize the major trends in B. melitensis metabolism in mice suggested by the analysis of our Tn-Seq data.
A clustering analysis of the 48 genes predicted to be essential in both the lungs and spleen identified 6 clusters involved in purine and histidine biosynthesis, methionine transport, carbohydrate metabolism, fatty acid oxidation, LPS biosynthesis and the type IV secretion system encoded by the virB operon (Fig 2). These data confirm the importance of bacterial genes implicated in LPS synthesis for B. melitensis survival in vivo [37]. While most of the genes involved in the biosynthesis of the LPS from glucose and fructose are not essential for B. melitensis 16M to multiply in rich medium or in the macrophage line RAW 264.7 [16], they become essential in the lungs [16] and spleen (Fig 4). This is consistent with the fact that B. melitensis is extracellular in the early times of intranasal [16] and intraperitoneal infection [38] and therefore exposed to attack by the innate immune system. It is well demonstrated in vitro that rough strains (LPS O-chain deficient) of Brucella are more sensitive to complement [39,40] and bactericidal cationic peptides [41]. Our data also confirm previous works documenting the importance of genes implicated in methionine transport [42], purine metabolism [43] and genes of the virB operon [44], with the exception of virB12 [45] for B. abortus growth in mice.
Regarding central carbohydrate metabolism, bacteria might present these interconnected pathways: classical glycolysis/gluconeogenesis, the pentose phosphate (PP) pathway, the Entner–Doudoroff (ED) pathway and the citric acid cycle (TCA). Numerous in vitro studies (reviewed in [46,47]) have shown that Brucella present a complete TCA cycle but are naturally deficient in the enzyme pfk and in consequence, in glycolysis. Therefore, the attenuation generated by fba (BMEII0423) deletion suggests an important role for gluconeogenesis during Brucella multiplication in lungs and spleen (Fig 5), results in accordance with the observations made by other authors [48,49] In this regard, inactivation of the pyk gene (BMEI0292), which converts phosphoenolpyruvate (PEP) to pyruvate, does not lead to B. melitensis attenuation in any of the conditions tested, while the enzyme PpdK (BMEI1436), which converts pyruvate to PEP, is essential in vivo (Fig 6). In addition to a possible role of PpdK in gluconeogenesis, the produced PEP is necessary to synthesize phenylalanine, tyrosine, tryptophan, glycerolipids, and other PEP-derived molecules [49].
Concerning the ED pathway, the first enzyme of the route, Edd (BMEII0511), is inactive in B. abortus and B. melitensis due to a nonsynonymous mutation [47], and therefore, these two strains exclusively use the PP pathway for hexose catabolism. In keeping with this observation, we found that two key enzymes of the PP pathway become indispensable for B. melitensis 16M multiplication in the lungs and spleen: pgl (BMEII0512) and rpiA (BMEI0974) (Fig 5). We also observed that many enzymes involved in the synthesis of histidine (hisA,H) and adenine (purA,B,C,E,H,S,S) from 5-phosphoribosyl 1-pyrophosphate produced by the PP pathway become indispensable in vivo (Fig 6).
The Tn-seq data suggest that fatty acid catabolism might be important during infection of the lungs and spleen since fadA (BMEII0496) and fadJ (BMEII0497), putatively required for fatty acid utilization, play an essential role during the multiplication of B. melitensis 16M in these organs (Fig 6). In this regard, the glyoxylate shunt is essential for utilizing acetate and fatty acids as carbon sources under physiological conditions requiring gluconeogenesis [50] However, the two enzymes of this cycle, isocitrate lyase aceA (BMEI0409) and malate synthase glcB (BMEI0380), are not predicted to be essential in vivo, data in agreement with previous results in B. abortus [49] and B. suis [48,51]
Also, the Tn-seq data predict that the glutamate synthase (GOGAT) enzyme, formed by gltB (BMEII0040) and gltD (BMEII0039), is essential during B. melitensis 16M multiplication in the spleen. GS-GOGAT is a two reaction-cycle that incorporates ammonium into glutamate to obtain glutamine (glutamine synthetase, GS), followed by the transfer of the amide group into a-ketoglutarate by GOGAT, with an overall yield of one glutamate molecule. In line with these results, mutants in gltB and gltD are attenuated both in cells and mice [52,53] These results might indicate that, although glutamine/glutamate are critical because they serve as the N donor for the biosynthesis of most amino acids, amino sugars, purines, pyrimidines, NAD+ and p-aminobenzoate [54], they are not provided in sufficient amounts in the infected cells of the spleen.
Collectively, these Tn-seq data suggest that B. melitensis could face multiple nutritional stresses in mice, which is consistent with its intracellular localization in activated phagocytic cells. In vivo, B. melitensis seems to have to synthesize histidine and adenine itself and import methionine. It also seems to be dealing with glucose deprivation. In keeping with these hypotheses, activation of the MyD88 pathway in macrophages has been shown to promote M1 metabolic polarization favoring glycolysis and glucose consumption able to directly restrict Brucella proliferation [55]. Thus, in an M1 polarized intracellular environment low in glucose, we hypothesize that B. melitensis would use amino acid such as glutamate and glutamine as well as fatty acids from the host cells to allow LPS biosynthesis and production of essential amino acids like histidine. In support of this, we previously showed by fluorescent microscopic analysis that B. melitensis host cells in the spleen of IL-12-/- BALB/c mice are particularly lipid-rich [23]. It is not known whether B. melitensis has access to large quantities of glutamine within its replication niche in vivo. However, glutamine is an amino acid that is very abundant in mammals and fuels many tissues as well as immune system cells [56]. It is one of the most abundant amino acids in the THP-1 macrophage cell line [57]. Glutamine is a favored source of nitrogen and, for example, is required for adenine biosynthesis via the PurA,B,C,E,H,Q,S enzymes. Glutamine can be also used to produce glutamate via GOGAT [42] or, possibly, via a glutaminase purQ (BMEI1124). The latter was functionally identified as a glutaminase in Bacillus subtilis [58]. Glutamate can be consumed by the TCA cycle to produce pyruvate and ultimately glucose (Fig 6). In addition, in several bacteria, glutamate metabolism has been implicated in the resistance of bacteria to acid stress [59]. The glutamate decarboxylase (GAD) system facilitates intracellular pH homoeostasis by consuming protons in a decarboxylation reaction that produces c-aminobutyrate (GABA) from glutamate. However, B. melitensis strains do not have a functional GAD system [60].
Comparison of our Tn-seq data obtained in the lungs and spleen demonstrates that several sets of genes are specifically required for B. melitensis in the lungs or spleen. Among the 12 genes predicted to be essential in the lungs and not in the spleen, a clustering analysis identified a cluster involved in tryptophan synthesis, a cluster associated with polymyxin resistance, and a cluster associated with respiration that are indispensable in the lungs and not in the spleen (Fig 3A). It is well described that the innate immune response depletes cellular tryptophan in response to infection via the host enzyme indoleamine 2,3-dioxygenase (IDO-1) that converts tryptophan to N-formylkynurenine, which is a potent negative regulator of inflammation. The tryptophan biosynthetic pathway has been shown to be essential for host colonization by Mycobacterium tuberculosis. ΔtrpD M. tuberculosis failed to cause disease in both wild-type and severe combined immune-deficient (SCID) mice [61]. The bveA (BMEII0681) gene of B. melitensis encodes a phospholipase A1 with specificity for phosphatidylethanolamine (PE). By reducing the level of PE in the bacterial membrane, bveA increases the resistance of B. melitensis to polymyxin and is required for persistent infection in mice [62]. To our knowledge, the importance of ctaA (BMEI1172), a heme A synthase, and ctaG (BMEI1463), a cytochrome c oxidase assembly protein, in mice has not been described.
Among the 34 genes predicted to be essential in the spleen and not in the lungs, a clustering analysis identified only four small clusters (Fig 3B): one cluster associated with amino acid biosynthesis, one identified as a zinc transporter, one identified as an efflux transporter and one identified as an ABC transporter. The other 23 genes could not be linked to a gene cluster and 5 of them are unidentified. Zinc (Zn2+) is an essential metal required by bacteria as either a structural or catalytic cofactor but free Zn2+ concentrations in mammalian hosts are very low. The znuABC operon constitutes a high-affinity periplasmic binding protein-dependent ATP-binding cassette (ABC) transport system used by bacteria for the uptake of Zn2+. It has been reported that ΔznuA B. melitensis [63] and ΔznuA B. abortus [64] are attenuated in the spleen of BALB/c mice.
Taken together, these results suggest that B. melitensis faces different nutritional conditions in the lungs and spleen. In particular, B. melitensis is thought to face deprivation of tryptophan in the lungs and of Zn2+ in the spleen. We confirmed using deletion mutants that the trpD (BMEI0844) gene is essential in the lungs and not in the spleen and that the znuA (BMEII0178) gene is essential in the spleen and not in the lungs.
The fact that persistence of B. melitensis in the lungs and spleen requires specific sets of genes suggests that it should be possible to use our Tn-seq data to develop vaccine candidates capable of persisting in organs long enough to induce development of a protective immune memory but unable to colonize the reservoir of the spleen on a long-term basis. Using these criteria (summarized in Fig 9), we selected two candidate genes, murI (BMEI0795) and plsC (BMEI1977). We predicted that inactivation of murI would induce very strong attenuation of B. melitensis in the spleen and that inactivation of plsC would result in moderate attenuation. We constructed deletion mutants of these genes and validated our Tn-seq predictions.
To our knowledge, murI and plsC have not been characterized in Brucella spp. The murI gene is predicted to code for a glutamate racemase, an enzyme involved in peptidoglycan (PG) biosynthesis. By converting L-glutamate to D-glutamate, this enzyme participates in the synthesis of peptide stems of PG, and thus contributes more generally to the integrity and growth of the bacteria [65]. In the literature, the murI gene is known to be the only enzyme that is able to synthesize D-Glu in E. coli [66] and M. tuberculosis [67] and is essential for their growth in vitro. The plsC gene coding for the integral membrane protein PlsC, an Acyl-sn-glycerol-3-phosphate acyltransferase. In mammals, PlsC enzyme is involved in phospholipid biosynthesis and is therefore required for epidermal permeability barrier homeostasis [68]. It could be hypothesized that inactivation of plsC in B. melitensis might change the properties of inner and outer membrane, potentially having pleiotropic effects on mechanisms such as effector secretion, resistance to antimicrobial peptides or alterations in the ability of the bacterium to adapt to environmental changes such as variations in acidity and osmolarity. Identifying the mechanisms explaining attenuation of ΔplsC B. melitensis strain in mice is difficult work that is beyond the scope of our study.
We compared the ability of deletion mutants for the murI and plsC genes to induce protective immunity in wild-type C57BL/6 mice. Surprisingly, we observed that the strongly attenuated ΔmurI strain induces only weak immunity in the lungs and spleen (-1 log CFU) whereas the moderately attenuated ΔplsC strain induces immunity similar to the wild-type and Rev.1 strains. These results suggest that it is essential for the vaccine strain to be able to multiply in the lungs and the spleen to induce the development of protective immunity in these organs. Excessive attenuation prevents activation of the immune system. This suggests that the activation of adaptive immunity by B. melitensis indeed requires the completion of an infection cycle, and that the simple administration of antigens is ineffective in protecting the animal. These results can be correlated with the induction of specific Brucella IgG2a antibodies. Protective wild type, ΔplsC and Rev.1 strains induce high levels of antibodies while non-protective strains such as ΔvirB and ΔmurI strains do not. Very interestingly, we observed that persistence of the ΔplsC strain in mice at 28 days post-vaccination was lower than that of the Rev.1 strain, suggesting that the ΔplsC strain may be safer than the Rev.1 strain.
Our approach has several limitations. First, B. melitensis can infect and persist in many organs in vivo and we only analyzed two of them in C57BL/6 mice. A more extensive functional map, including other organs of interest, such as the lymph nodes and placenta, would be useful to better understand the biology of B. melitensis in the mouse model. Secondly, it is not advisable to make attenuation of a vaccine dependent on the deletion of a single gene. Consequently, it would be useful to carry out new Tn-Seq analyses in the organs of interest using a library constructed from the ΔplsC strain and to compare these results to those already available with the library constructed from the wild-type strain. This approach would identify other genes that can be deleted without the risk of excessively reducing the virulence of the vaccine candidate in the spleen, while guaranteeing the desired level of attenuation. Thirdly, since mice are not the natural host of B. melitensis and in light of the fact that many vaccines developed in mice have failed to induce immunity in cattle, Tn-seq analyses should be carried out in the spleen of goats or sheep in order to validate the results obtained in mice.
Overall, our results demonstrate that B. melitensis faces very different environments in vivo depending on the organs infected. Identifying the nutritional requirements of B. melitensis in vivo may open up new avenues for brucellosis treatment. For example, tryptophan synthase inhibitors from M. tuberculosis have been developed and have been shown to be effective in blocking its growth [69]. Our results also show that the construction of a functional map of the B. melitensis genome using Tn-seq analyses carried out on different organs can be a valuable aid for the development of effective attenuated vaccine candidates in mice, with a safer profile.
Materials and methods
Ethics statement
The procedures used in this study and the handling of the mice complied with current European legislation (Directive 86/609/EEC). The Animal Welfare Committee of the Université de Namur (UNamur, Belgium) reviewed and approved the complete protocol for Brucella melitensis infection (Permit Number: UN-LE-18/309 and UN-LE-23/401).
Mice and bacterial strains
Wild-type C57BL/6 mice were acquired from Harlan (Bicester, UK). IFN-γR-/- C57BL/6 mice [17] were acquired from Dr B. Ryffel (University of Orleans, France). All wild-type and deficient mice used in this study were bred in the animal facility of the Gosselies campus of the Université Libre de Bruxelles (ULB, Belgium).
The wild-type B. melitensis 16M strain used here is a NalR derivative of wild-type B. melitensis 16M [70]. We also used wild-type [22] and ΔvirB1-12 (BMEII0025-35) [71] B. melitensis 16M strains stably expressing the mCherry fluorescent protein under the control of the strong Brucella spp. promoter psecE, also called psog or psojA [72]. B. melitensis Rev.1 is the World Organisation for Animal Health (WOAH) recommended goat and sheep brucellosis vaccine and was obtained from Sciensano (Belgium).
Brucella melitensis was always handled in BSL-3 containment facilities according to Council Directive 98/81/EC of 26 October 1998 and the law of the Walloon government of 4 July 2002.
Transposon mutagenesis
One milliliter of an overnight culture of a nalidixic acid-resistant strain of B. melitensis 16M was mixed with 50 μL of an overnight culture of the conjugative Escherichia coli S17-1 strain carrying the pXMCS-2 mini-Tn5 Kanr plasmid [73]. This plasmid possesses a hyperactive Tn5 transposase allowing for straightforward generation of a high number of Tn mutants, as described previously [73]. The mating mixture was incubated overnight at room temperature (RT) on 2YT agar plates (rich medium, 1% yeast extract, 1.6% peptone, 0.5% NaCl, 2% agar). The resulting B. melitensis Tn mutants were selected on 2YT agar plates supplemented with both kanamycin (10 μg/mL) and nalidixic acid (25 μg/mL). Tn5 mutagenesis generates insertion of the transposon at only one locus per genome, as demonstrated previously for Brucella [74].
Analysis of essential genes for growth on plates
Genomic DNA was extracted from a spleen transposon library using standard techniques and prepared for transposon library sequencing. Briefly, B. melitensis Tn mutants from each plate were collected, mixed and killed by heat (1 hour, 80°C). The lysate was incubated with a mixture composed of Tris (tris-hydroxymethyl-aminomethane 50 mM), EDTA (Ethylenediaminetetraacetic acid, 50 mM), 0.1 M NaCl, Proteinase K (20 mg/mL) and 10% SDS. The mixture was treated with an equal volume of 100% isopropanol to precipitate the DNA, which was washed with 70% ethanol. Genomic DNA was resuspended in deionized water and genomic DNA flanking the Tn5 was sequenced (Fasteris company, Geneva, Switzerland). Libraries were sequenced on an Illumina HiSeq (for spleen from IFNγR-/- mice conditions) or on an Illumina NextSeq (for 2YT, lung and spleen from wild-type mice conditions) with a primer hybridized at the border of the transposon, with its 3’ end pointing toward the flanking genomic DNA. Raw reads for each biological conditions (available at https://doi.org/10.6084/m9.figshare.26063158.v1) were mapped on B. melitensis 16M (accession numbers NC_003317 and NC_003318 for chromosomes I and II, respectively) using BWA [75] and read counts were determined using the samtools suite [76]. To account for truncated but functional products and misannotated start sites, only insertions in the central 80% of each gene were considered.
To determine if an insertional mutant in a defined gene is affected in a condition but untouched in the control 2YT condition, each gene was assigned an insertion index, called the transposon insertion frequency (TnIF), equal to the log10 of its total read count (thus including multiple insertions at the same position) +1, divided by its length (in bp), corresponding to 80% of the internal segment of the coding sequence. For each gene, a ΔTnIF (TnIFcdt−TnIFCTRL) value was calculated, where TnIF was computed for the tested condition (TnIFcdt) and the control condition (TnIFCTRL). The frequency distribution of ΔTnIF values was plotted for both chromosomes and for each condition tested (S3 Fig), to identify the main peak of unaffected ΔTnIF values and its standard deviation. 2% of ΔTnIF values at each extremity were removed to avoid an influence of extreme values. The standard deviation was calculated on this distribution. Depending on the conditions tested, the standard deviation ranged from 0.19 and 0.30 with a mean of 0.24. The ΔTnIF values greater than 0.5 were thus selected as significant, since they correspond to 1.6 to 2.6 standard deviations from the mode, designating genes for which the TnIF value was decreased compared to the control condition.
Construction of deletion mutants to test Tn-seq predictions
Construction of the ΔtrpD (BMEI0843) and Δgmd (BMEI1413) B. melitensis 16M strains has been previously described [16]. The ΔlysR (BMEI0513), ΔmurI (BMEI0795), ΔplsC (BMEI1977), ΔpurH (BMEI0233) and ΔznuA (BMEII0178) deletion strains were constructed in the B. melitensis 16M wt strain by triparental mating to introduce the pNPTS138 KanR plasmid (containing the upstream joined to the downstream region, generated by PCR, for the respective genes of interest for deletion) in the B. melitensis 16M Nal strain using the E. coli MT 607 (pro-82 thi-I hsdR17 (r-m+) supE44 recA56 pRK600) strain (as described in [77]), and allelic replacement was performed as described previously for other gene deletions [78]. These mutant strains do not have inserted antibiotic resistance genes.
Δgmd, ΔlysR, ΔmurI, ΔplsC, ΔpurH, ΔtrpD and ΔznuA deletion strains were conjugated with E. coli S17-1 containing the pSK kanR DsRed plasmid to introduce genes for kanamycin resistance and to express DsRed. Deletion of the genes was checked using the respective primers: the lysR-CheckF and lysR-CheckR primers for lysR, the murI-CheckF and murI-CheckR primers for murI, the plsC-CheckF and plsC-CheckR primers for plsC, purH-CheckF and purH-CheckF primers for purH, the znuA-CheckF and znuA-CheckR primers for znuA. See S3 Table for primer sequences used to amplify upstream and downstream regions of each gene.
The ΔmurI and ΔplsC deletion mutants were complemented with the pMR10 vector containing the gene corresponding amplified with primers, as listed in S3 Table, under the control of the plac promoter (pMR10::trpD, pMR10::murI).
Brucella melitensis infection in vivo
Mice were anesthetized with a cocktail of Xylazine (9 mg/kg) and Ketamine (36 mg/kg) in PBS before being inoculated by intranasal injection with 105 or 5 × 106 CFU of B. melitensis in 30 μL of RPMI, as indicated. Intraperitoneal injection of 5 × 106 CFU of B. melitensis in 500 μL of RPMI was performed without anesthesia as described previously [27]. Control animals were inoculated with the same volume of RPMI. We used a mCherry-expressing wild-type (wt) 16M strain [79], mCherry-expressing ΔvirB 16M strain [22] or DsRed-expressing gene deletion mutants in the wild-type B. melitensis 16M background as indicated for the infections. Cultures were grown overnight with shaking at 37°C in 2YT medium and were washed twice in RPMI 1640 (Gibco Laboratories) (3500 g, 10 min) before inoculation in the mice. The infectious doses were validated by plating serial dilutions of the inoculums. At the selected times after infection, mice were sacrificed by cervical dislocation. Immediately after sacrifice, spleen and/or lung cells were collected for bacterial counting or bacterial DNA extraction. All infections were performed in an Animal Biosafety Level 3 facility.
For bacterial counting, organs were homogenized in PBS/0.1% X-100 Triton (Sigma-Aldrich). We performed successive serial dilutions in RPMI to obtain the most accurate bacterial count and plated them on 2YT medium. The CFU were counted after 4 days of incubation at 37°C.
Protocol for secondary infection with Brucella melitensis
C57BL/6 mice were immunized intranasally (i.n.) with 105 CFU of live wild-type or deletion mutants of B. melitensis 16M, as indicated. The infectious doses were validated by plating serial dilutions of inoculums. Six weeks after immunization, the mice were challenged i.n. with 105 CFU of live wild-type mCherry-B. melitensis and sacrificed at 12 days post-infection. It is important to note that none of the strains used for vaccination has Kan resistance or expresses a fluorochrome. The expression of Kan resistance and the mCherry fluorochrome by the challenge strain makes it possible to easily differentiate the vaccine strains from the challenge strain when counting the bacterial colonies on plates of 2YT-Agar medium with or without kanamycin (50 μg/mL) since bacteria with mCherry form pink colonies on plates.
Enzyme-linked immunosorbent assay (ELISA)
The presence of Brucella melitensis specific murine IgG2a was determined by ELISA. Polystyrene plates (269620; Nunc) were coated with Heat Killed (HK) B. melitensis (107 CFU/mL) and incubated overnight at 4°C. The plates were blocked for 2 hours at RT with 200 μL/well of PBS-1% Bovine Serum Albumin (BSA). Then, plates were incubated with 50 μL/well of plasma in serial dilutions in PBS-0.1% BSA. The plasma of uninfected mice and PBS were used as negative controls. After four washes with PBS, isotype-specific goat anti-mouse HRP-conjugated Ab were added (50 μL/well) at appropriated dilutions (LO-MG2a-9 HRPO from LOIMEX). After 1 hour of incubation at RT, plates were washed four times in PBS and 100 μL/well of TMB substrate solution (BD OptEiA Kit) was added. After 15 minutes of incubation at RT in the dark, the enzyme reaction was stopped by adding 25 μL/well of 2 N H2SO4. The absorbance was measured at 450 nm.
Statistical analysis
We used a (Wilcoxon-)Mann-Whitney test provided by the GraphPad Prism software to statistically analyze our results. Each group of deficient mice was compared with the wild-type mice. We also compared each group with the other groups and displayed the results when required. Values of p < 0.05 were considered to represent a significant difference. *, **, *** denote p < 0.05, p < 0.01, p < 0.001, respectively.
Supporting information
S1 Fig. Complementation of ΔmurI and ΔplsC in the lungs and spleen.
Data shown are bacterial counts (CFU) at the 120 hours post-infection in the lungs (A) and spleen (B) from wild-type mice infected intranasally (A) or intraperitoneally (B) with wild-type (wt), ΔmurI or ΔmurI-complemented, ΔplsC or ΔplsC-complemented strains of B. melitensis at a dose of 5x106 CFU. Red lines represent the geometric mean. Dotted lines represent the mean of the wild-type strain. Significant differences between wt and the indicated groups are marked with asterisks: *p < 0.1, **p < 0.01, ***p < 0.001, ****p < 0.0001, in a (Wilcoxon-)Mann-Whitney post-test. These results are representative of two independent experiments.
https://doi.org/10.1371/journal.ppat.1012459.s001
(TIF)
S2 Fig. Humoral immune response induced by intranasal B. melitensis strains infection.
Wild-type C57BL/6 mice were infected intranasally with a dose of 105 CFU of several strains of B. melitensis (wild-type (WT), ΔplsC, Rev.1, ΔmurI or ΔvirB). Sera were collected at 5 weeks post-infection, and ELISA was performed to determine the isotype distribution of the IgG2a Brucella-specific antibodies. The data represent the means ± SD of the results. O.D, optical density. These results are representative of two independent experiments.
https://doi.org/10.1371/journal.ppat.1012459.s002
(TIF)
S3 Fig. Frequency distribution of ΔTnIF values of genes from chromosomes I and II of B. melitensis for each tested in vivo condition.
For each condition (lungs, spleen, spleen from IFN-γ-/-), the ΔTnIF values are represented by classes of 0.2. The blue histogram shows the distribution for ΔTnIF values for all genes that are untouched in the control 2YT condition. The red color represents the distribution for ΔTnIF values without 2% of the genes at each extremity. SD = standard deviation.
https://doi.org/10.1371/journal.ppat.1012459.s003
(PDF)
S1 Table. List of attenuated B. melitensis genes in vitro and in mice conditions.
Page 1: List of the 3369 genes of B. melitensis. Each gene is associated with the value of TnIF in the 2YT, lungs and spleen conditions as well as the value of ΔTnIF associated with the lungs and spleen conditions. Page 2: List of 2460 genes that are not predicted to induce a growth defect in 2YT (TnIF > 2.7 in 2YT condition). Page 3: List of 135 genes predicted to cause attenuation in the lungs of wild type mice (ΔTnIF < -0.5 in lungs condition). Page 4: List of 257 genes predicted to cause attenuation in the spleen of wild type mice (ΔTnIF < -0.5 in spleen condition). Page 7: List of the 4 genes selected as vaccine candidates showing strong attenuation in the spleen (dTnIF < -2.0 in spleen condition). Page 8: List of the 17 genes selected as vaccine candidates showing moderate attenuation in the spleen (dTnIF < -1.0).
https://doi.org/10.1371/journal.ppat.1012459.s004
(XLSX)
S2 Table. List of B. melitensis genes that are predicted as attenuated in the spleen of wild-type mice but that are predicted as not attenuated in the spleen of IFNγR-/- mice.
Bacterial genes required to grow in the spleen of wild-type (wt) mice infected intraperitoneally (ΔTnIF < -1.0) and that are not attenuated in the spleen of IFNγR-/- mice infected by the same route (ΔTnIF > -0.5). NA = not assigned.
https://doi.org/10.1371/journal.ppat.1012459.s005
(DOCX)
S3 Table. List of primers used in the construction of deletion mutants and for complementation of mutants.
https://doi.org/10.1371/journal.ppat.1012459.s006
(DOCX)
Acknowledgments
We thank K. Willemart, F. Tilquin, E. Carlier and M. Waroquier for their technical support.
References
- 1. Martirosyan A, Moreno E, Gorvel JP. An evolutionary strategy for a stealthy intracellular Brucella pathogen. Immunological Reviews. 2011. pp. 211–234. pmid:21349096
- 2. Atluri VL, Xavier MN, De Jong MF, Den Hartigh AB, Tsolis RM. Interactions of the human pathogenic Brucella species with their hosts. Annu Rev Microbiol. 2011;65: 523–541. pmid:21939378
- 3. Pappas G, Papadimitriou P, Akritidis N, Christou L, Tsianos E V. The new global map of human brucellosis. Lancet Infect Dis. 2006;6: 91–99. pmid:16439329
- 4. McDermott J, Grace D, Zinsstag J. Economics of brucellosis impact and control in low-income countries. OIE Rev Sci Tech. 2013;32: 249–261. pmid:23837382
- 5. Lai S, Zhou H, Xiong W, Gilbert M, Huang Z, Yu J, et al. Changing epidemiology of human brucellosis, China, 1955–2014. Emerg Infect Dis. 2017;23: 184–194. pmid:28098531
- 6. Laine CG, Johnson VE, Scott HM, Arenas-Gamboa AM. Global Estimate of Human Brucellosis Incidence. Emerg Infect Dis. 2023;29: 1789–1797. pmid:37610167
- 7. Moreno E, Moriyón I. The Genus Brucella. The Prokaryotes. 2006.
- 8. Blasco JM, Moreno E, Muñoz PM, C Álvarez R, Moriyón I. A review of three decades of use of the cattle brucellosis rough vaccine Brucella abortus RB51: myths and facts. BMC Vet Res. 2023; 1–21. pmid:37853407
- 9. Blasco JM. A review of the use of B. melitensis Rev 1 vaccine in adult sheep and goats. Prev Vet Med. 1997;31: 275–283. pmid:9234451
- 10. Ficht TA, Kahl-McDonagh MM, Arenas-Gamboa AM, Rice-Ficht AC. Brucellosis: The case for live, attenuated vaccines. Vaccine. 2009;27. pmid:19837284
- 11. Dorneles EMS, Sriranganathan N, Lage AP. Recent advances in Brucella abortus vaccines. Vet Res. 2015;46: 1–10. pmid:26155935
- 12.
Blasco J, Moreno E, Moriyón I. Brucellosis vaccines and vaccine candidates. First Edit. In: Metwally S, Viljoen G, El Idrissi A, editors. Veterinary vaccines Principles and applications. First Edit. Rome: Hobeken; 2021. pp. 295–316.
- 13. Blasco JM, Molina-Flores B. Control and Eradication of Brucella melitensis Infection in Sheep and Goats. Vet Clin North Am—Food Anim Pract. 2011;27: 95–104. pmid:21215893
- 14. Van Opijnen T, Camilli A. Transposon insertion sequencing: a new tool for systems-level analysis of microorganisms. Nat Publ Gr. 2013;11: 435–442. pmid:23712350
- 15. Kwon YM, Ricke SC, Mandal RK. Transposon sequencing: methods and expanding applications. Appl Microbiol Biotechnol. 2015;100: 31–43. pmid:26476650
- 16. Potemberg G, Demars A, Barbieux E, Galia M, Lagneaux M, Comein A, et al. Genome-wide analysis of Brucella melitensis genes required throughout intranasal infection in mice. PLOS Pathog. 2022;16: e1010621. pmid:35771771
- 17. Mambres DH, MacHelart A, Potemberg G, De Trez C, Ryffel B, Letesson J-J, et al. Identification of immune effectors essential to the control of primary and secondary intranasal infection with brucella melitensis in mice. J Immunol. 2016;196:3780–3793. pmid:27036913
- 18. Copin R, De Baetselier P, Carlier Y, Letesson J-J, Muraille E. MyD88-dependent activation of B220-CD11b+LY-6C + dendritic cells during Brucella melitensis infection. J Immunol. 2007;178: 5182–5191. 178/8/5182 [pii].
- 19. Grilló M, Blasco JM, Gorvel JP, Moriyón I, Moreno E. What have we learned from brucellosis in the mouse model? 2012; 1–35.
- 20. Demars A, Lison A, Machelart A, Van Vyve M, Potemberg G, Vanderwinden JJ-M, et al. Route of Infection Strongly Impacts the Host-Pathogen Relationship. Front Immunol. 2019;10: 1589. pmid:31354728
- 21. Archambaud C, Salcedo SP, Lelouard H, Devilard E, De Bovis B, Van Rooijen N, et al. Contrasting roles of macrophages and dendritic cells in controlling initial pulmonary Brucella infection. Eur J Immunol. 2010;40: 3458–3471. pmid:21108467
- 22. Copin R, Vitry M-A, Hanot Mambres D, Machelart A, De Trez C, Vanderwinden J-M, et al. In situ microscopy analysis reveals local innate immune response developed around Brucella infected cells in resistant and susceptible mice. PLoS Pathog. 2012;8: e1002575. pmid:22479178
- 23. Mambres DH, Machelart A, Vanderwinden JM, De Trez C, Ryffel B, Letesson JJ, et al. In situ characterization of splenic brucella melitensis reservoir cells during the chronic phase of infection in susceptible mice. PLoS One. 2015;10: 1–20. pmid:26376185
- 24. Delrue RM, Deschamps C, Léonard S, Nijskens C, Danese I, Schaus JM, et al. A quorum-sensing regulator controls expression of both the type IV secretion system and the flagellar apparatus of Brucella melitensis. Cell Microbiol. 2005;7: 1151–1161. pmid:16008582
- 25. Arocena GM, Sieira R, Comerci DJ, Ugalde RA. Identification of the quorum-sensing target DNA sequence and N-acyl homoserine lactone responsiveness of the Brucella abortus virB promoter. J Bacteriol. 2010;192: 3434–3440. pmid:20400542
- 26. Lacey CA, Chambers CA, Mitchell WJ, Skyberg JA. IFN-γ-dependent nitric oxide suppresses Brucella-induced arthritis by inhibition of inflammasome activation. Journal of Leukocyte Biology. 2019. pp. 27–34. pmid:30748031
- 27. Vitry M-A, De Trez C, Goriely S, Dumoutier L, Akira S, Ryffel B, et al. Crucial role of gamma interferon-producing CD4+ Th1 cells but dispensable function of CD8+ T cell, B cell, Th2, and Th17 responses in the control of Brucella melitensis infection in mice. Infect Immun. 2012;80: 4271–80. pmid:23006848
- 28. Sheehan LM, Budnick JA, Blanchard C, Dunman PM, Caswell CC. A LysR-family transcriptional regulator required for virulence in Brucella abortus is highly conserved among the α-proteobacteria. Molecular Microbiology. 2015. pp. 318–328. pmid:26175079
- 29. ELBERG SS, FAUNCE K. Immunization against Brucella infection. VI. Immunity conferred on goats by a nondependent mutant from a streptomycin-dependent mutant strain of Brucella melitensis. J Bacteriol. 1957;73: 211–217. pmid:13416171
- 30. Schurig GG, Sriranganathan N, Corbel MJ. Brucellosis vaccines: Past, present and future. Vet Microbiol. 2002;90: 479–496. pmid:12414166
- 31. Alton GG. Control of brucella melitensis infection in sheep and goats-a review. Trop Anim Health Prod. 1987;19: 65–74. pmid:3307078
- 32. Banai M. Control of small ruminant brucellosis by use of Brucella melitensis Rev.1 vaccine: Laboratory aspects and field observations. Vet Microbiol. 2002;90: 497–519. pmid:12414167
- 33. Higgins JL, Gonzalez-juarrero M, Bowen RA. Evaluation of shedding, tissue burdens, and humoral immune response in goats after experimental challenge with the virulent Brucella melitensis strain 16M and the reduced virulence vaccine strain Rev. 1. 2017; 1–16.
- 34. Blasco JM, Díaz R. Brucella melitensis Rev-1 vaccine as a cause of human brucellosis. Lancet. 1993;342: 805. pmid:8103891
- 35. Vives-Soto M, Puerta-García A, Rodríguez-Sánchez E, Pereira J, Solera J. What risk do Brucella vaccines pose to humans? A systematic review of the scientific literature on occupational exposure. PLoS Negl Trop Dis. 2024;18: 1–12. pmid:38190394
- 36. Salmon-Divon M, Zahavi T, Kornspan D. Transcriptomic analysis of the brucella melitensisrev.1 vaccine strain in an acidic environment: Insights into virulence attenuation. Front Microbiol. 2019;10: 1–12. pmid:30837973
- 37. González D, Grilló MJ, De Miguel MJ, Ali T, Arce-Gorvel V, Delrue RM, et al. Brucellosis vaccines: Assessment of Brucella melitensis lipopolysaccharide rough mutants defective in core and O-polysaccharide synthesis and export. PLoS One. 2008;3. pmid:18648644
- 38. Vitry M-A, Hanot Mambres D, Deghelt M, Hack K, Machelart A, Lhomme F, et al. Brucella melitensis invades murine erythrocytes during infection. Infect Immun. 2014. pmid:25001604
- 39. Eisenschenk FC, Houle JJ, Hoffmann EM. Mechanism of serum resistance among Brucella abortus isolates. Vet Microbiol. 1999;68: 235–244. pmid:10510042
- 40. Allen CA, Adams LG, Ficht TA. Transposon-derived Brucella abortus rough mutants are attenuated and exhibit reduced intracellular survival. Infect Immun. 1998;66: 1008–1016. pmid:9488389
- 41. De Tejada GM, Pizarro-Cerda J, Moreno E, Moriyon I. The outer membranes of Brucella spp. are resistant to bactericidal cationic peptides. Infect Immun. 1995;63: 3054–3061. pmid:7622230
- 42. Ronneau S, Moussa S, Barbier T, Conde-Álvarez R, Zuniga-Ripa A, Moriyon I, et al. Brucella, nitrogen and virulence. Crit Rev Microbiol. 2016;42: 507–525. pmid:25471320
- 43. Alcantara RB, Read RDA, Valderas MW, Brown TD, Roop RM. Intact purine biosynthesis pathways are required for wild-type virulence of Brucella abortus 2308 in the BALB/c mouse model. Infect Immun. 2004;72: 4911–4917. pmid:15271960
- 44. de Jong MF, Tsolis RM. Brucellosis and type IV secretion. Future Microbiol. 2012;1: 47–58. pmid:22191446
- 45. Sun YH, Rolán HG, Den Hartigh AB, Sondervan D, Tsolis RM. Brucella abortus VirB12 is expressed during infection but is not an essential component of the type IV secretion system. Infect Immun. 2005;73: 6048–6054. pmid:16113325
- 46. Barbier T, Zúñiga-Ripa A, Moussa S, Plovier H, Sternon JF, Lázaro-Antón L, et al. Brucella central carbon metabolism: an update. Crit Rev Microbiol. 2018;44: 182–211. pmid:28604247
- 47. MacHelart A, Willemart K, Zúñiga-Ripa A, Godard T, Plovier H, Wittmann C, et al. Convergent evolution of zoonotic Brucella species toward the selective use of the pentose phosphate pathway. Proc Natl Acad Sci U S A. 2020;117: 26374–26381. pmid:33020286
- 48. Zúñiga-Ripa A, Barbier T, Lázaro-Antón L, de Miguel MJ, Conde-álvarez R, Muñoz PM, et al. The fast-growing Brucella suis Biovar 5 depends on phosphoenolpyruvate carboxykinase and pyruvate phosphate dikinase but not on Fbp and GlpX fructose-1,6-bisphosphatases or isocitrate lyase for full virulence in laboratory models. Front Microbiol. 2018;9: 1–13. pmid:29675004
- 49. Zúñiga-Ripa A, Barbier T, Conde-Álvarez R, Martínez-Gómez E, Palacios-Chaves L, Gil-Ramírez Y, et al. Brucella abortus depends on pyruvate phosphate dikinase and malic enzyme but not on fbp and glpX fructose-1,6-bisphosphatases for full virulence in laboratory models. J Bacteriol. 2014;196: 3045–3057. pmid:24936050
- 50. Maloy SR, Bohlander M, Nunn WD. Elevated levels of glyoxylate shunt enzymes in Escherichia coli strains constitutive for fatty acid degradation. J Bacteriol. 1980;143: 720–725. pmid:7009561
- 51. Köhler S, Foulongne V, Ouahrani-Bettache S, Bourg G, Teyssier J, Ramuz M, et al. The analysis of the intramacrophagic virulome of Brucella suis deciphers the environment encountered by the pathogen inside the macrophage host cell. Proc Natl Acad Sci U S A. 2002;99: 15711–15716. pmid:12438693
- 52. Hong PC, Tsolis RM, Ficht TA. Identification of genes required for chronic persistence of Brucella abortus in mice. Infect Immun. 2000;68: 4102–4107. pmid:10858227
- 53. Wu Q, Pei J, Turse C, Ficht TA. Mariner mutagenesis of Brucella melitensis reveals genes with previously uncharacterized roles in virulence and survival. BMC Microbiol. 2006;6: 1–15. pmid:17176467
- 54.
Gottschalk G. Bacterial Metabolism. 2nd Editio. New York: Springer-Verlag; 1986.
- 55. Lacey CA, Ponzilacqua-Silva B, Chambers CA, Dadelahi AS, Skyberg JA. MyD88-Dependent Glucose Restriction and Itaconate Production Control Brucella Infection. Infect Immun. 2021. pmid:34125603
- 56. Watford M. Glutamine and glutamate: Nonessential or essential amino acids? Anim Nutr. 2015;1: 119–122. pmid:29767158
- 57. Tullius M V., Harth G, Horwitz MA. Glutamine synthetase GlnA1 is essential for growth of Mycobacterium tuberculosis in human THP-1 macrophages and guinea pigs. Infect Immun. 2003;71: 3927–3936. pmid:12819079
- 58. Anand R, Hoskins AA, Bennett EM, Sintchak MD, Stubbe JA, Ealick SE. A model for the Bacillus subtilis formylglycinamide ribonucleotide amidotransferase multiprotein complex. Biochemistry. 2004;43: 10343–10352. pmid:15301532
- 59. Feehily C, Karatzas KAG. Role of glutamate metabolism in bacterial responses towards acid and other stresses. J Appl Microbiol. 2013;114: 11–24. pmid:22924898
- 60. Damiano MA, Bastianelli D, Al Dahouk S, Köhler S, Cloeckaert A, de Biase D, et al. Glutamate decarboxylase-dependent acid resistance in Brucella spp.: Distribution and contribution to fitness under extremely acidic conditions. Appl Environ Microbiol. 2015;81: 578–586. pmid:25381237
- 61. Lott J. Shaun The tryptophan biosynthetic pathway is essential for Mycobacterium tuberculosis to cause disease. Biochem Soc Trans. 2020;48: 2029–2037. pmid:32915193
- 62. Kerrinnes T, Young BM, Leon C, Roux CM, Tran L, Atluri VL, et al. Phospholipase A1 modulates the cell envelope phospholipid content of brucella melitensis, contributing to polymyxin resistance and pathogenicity. Antimicrob Agents Chemother. 2015;59: 6717–6724. pmid:26282427
- 63. Clapp B, Skyberg JA, Yang X, Thornburg T, Walters N, Pascual W pascual. Protective live oral brucellosis vaccines stimulate Th1 and Th17 cell responses. Infect Immun. 2011;79: 4165–4174. pmid:21768283
- 64. Yang X, Becker T, Walters N, Pascual DW. Deletion of znuA virulence factor attenuates Brucella abortus and confers protection against wild-type challenge. Infect Immun. 2006;74: 3874–3879. pmid:16790759
- 65. Fisher SL. Glutamate racemase as a target for drug discovery. Microb Biotechnol. 2008;1: 345–360. pmid:21261855
- 66. Doublet P, Van Heijenoort J, Bohin JP, Mengin-Lecreulx D. The murI gene of Escherichia coli is an essential gene that encodes a glutamate racemase activity. J Bacteriol. 1993;175: 2970–2979. pmid:8098327
- 67. Morayya S, Awasthy D, Yadav R, Ambady A, Sharma U. Revisiting the essentiality of glutamate racemase in Mycobacterium tuberculosis. Gene. 2015;555: 269–276. pmid:25447907
- 68. Lu B, Jiang YJ, Man MQ, Brown B, Elias PM, Feingold KR. Expression and regulation of 1-acyl-sn-glycerol-3-phosphate acyltransferases in the epidermis. J Lipid Res. 2005;46: 2448–2457. pmid:16150824
- 69. Abrahams KA, Cox JAG, Fütterer K, Rullas J, Ortega-Muro F, Loman NJ, et al. Inhibiting mycobacterial tryptophan synthase by targeting the inter-subunit interface. Sci Rep. 2017;7: 1–15. pmid:28842600
- 70. Verger JM, Grayon M, Chaslus-Dancla E, Meurisse M, Lafont JP. Conjugative transfer and in vitro/in vivo stability of the broad-host-range IncP R751 plasmid in brucella spp. Plasmid. 1993;29: 142–146. pmid:8469720
- 71. Casadaban MJ, Cohen SN. Analysis of Gene Control Signals by DNA Fusion and Cloning in Escherichia coli. J Mol Biol. 1980;138: 179–207. pmid:6997493
- 72. Ouahrani-bettache S, Porte F, Teyssier J, Liautard J, Köhler S. pBBR1-GFP: A Broad- Host-Range Vector for Prokaryotic Promoter Studies. Biotechniques. 1999;26: 620–622. pmid:10343896
- 73. Sternon J, Godessart P, Gonçalves de Freitas R, Van der Henst M, Poncin K, Francis N, et al. Transposon Sequencing of Brucella abortus Uncovers Essential Genes for Growth In Vitro and Inside Macrophages. Infect Immun. 2018;86: 1–20. pmid:29844240
- 74. Lestrate P, Delrue R, Danese I, Didembourg C, Taminiau B, Mertens P, et al. Identification and characterization of in vivo attenuated mutants of Brucella melitensis. Mol Microbiol. 2000;38: 543–551. pmid:11069678
- 75. Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics. 2009;25: 1754–1760. pmid:19451168
- 76. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment / Map format and SAMtools. Bioinformatics. 2009;25: 2078–2079. pmid:19505943
- 77. Finan TM, Kunkel B, De Vos GF, Signer ER. Second symbiotic megaplasmid in Rhizobium meliloti carrying exopolysaccharide and thiamine synthesis genes. J Bacteriol. 1986;167: 66–72. pmid:3013840
- 78. Deghelt M, Mullier C, Sternon J-F, Francis N, Laloux G, Dotreppe D, et al. G1-arrested newborn cells are the predominant infectious form of the pathogen Brucella abortus. Nat Commun. 2014;5: 4366. Available: http://www.ncbi.nlm.nih.gov/pubmed/25006695 pmid:25006695
- 79. Shaner NC, Campbell RE, Steinbach PA, Giepmans BNG, Palmer AE, Tsien RY. Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nature biotechnology. 2004. pp. 1567–1572. pmid:15558047