The stringent response reflects the adaptation of a bacterium to nutrient stress via a complex differential gene expression pattern, affecting a large number of structural and regulatory target genes and resulting in a pleiotropic phenotype. In an earlier study, some of us showed that rsh deletion mutants were characterized by altered morphology, lack of expression of virB, and reduced survival in cellular and murine models of infection . In this study, we investigated the global expression profile of the stringent response using a DNA microarray approach with the aim to characterize the (p)ppGpp-dependent regulatory network, and we also focused on methionine biosynthesis, the sole amino acid whose biosynthesis pathway was controlled by stringent response in B. suis.
A comparison of the wild-type and rsh mutant transcriptomes showed that approximately 12% of the B. suis genome were under the control of Rsh, of which 52% were up-regulated and 48% were down-regulated. The differentially transcribed genes were classified into 19 functional categories.
A key element in the establishment of Brucella infection is the ability of the bacterium to resist to acid pH and nutrient deprivation within the macrophage host cells, at least in the early phase of infection. Both signals are essential for a strong induction of the T4SS in B. suis, although RNA blot experiments with a virB5-probe reveal some virB-expression in minimal medium at pH 7 , corresponding to our experimental conditions in this transcriptome study. Our transcriptome study showed that virB-expression was Rsh-dependent, which is in accordance with previous work of some of us . On a fine-tune level, the gene encoding the alpha-subunit of the Integration Host Factor (IHF) and hutC, which both control transcription of the virB operon via specific promoter binding sites, were also up-regulated under nutrient stress conditions. In E. coli, it has been described that expression of IHF is induced under stringent response conditions and in early-stationary phase [46, 47]. The Rsh-dependent up-regulation of IHF-expression in Brucella confirmed the regulation pathway previously suggested: Brucella senses nutrient starvation via Rsh, resulting in (p)ppGpp production, which will then increase transcription of IHF, affecting the activity of the Brucella virB promoter . However, the fact that overexpression of the virB operon in a B. suis Δrsh background did not restore the parental phenotype by itself (data not shown) is an additional indication that the stringent response has a pleiotropic effect on virulence factor expression. Among transcriptional regulators participating in Brucella virulence, expression of MucR was shown to be Rsh-dependent. Studies on a mucR-mutant of B. melitensis suggest that MucR regulates genes involved in nitrogen metabolism and stress response , as well as in lipid A-core and cyclic-β-glucan synthesis of this pathogen . In this context it is important to mention that recent work on the transcriptome analysis of stringent response in the α-proteobacterium and plant symbiont R. etli reported the observation that Rsh positively controls expression of key regulators for survival during heat and oxidative stress . Remarkably, about 6% of the (p)ppGpp-regulated genes in B. suis are transcriptional regulators of, most-often, uncharacterized function.
Several consecutive genes, potentially involved in resistance to acid pH, were also induced by Rsh: genes encoding the glutamate decarboxylase A (BRA0338), the glutaminase (BRA0340), and hdeA (BRA0341). Glutamate decarboxylase, which is involved in resistance of Brucella microti to pH 2.5  is not functional in B. suis. This overexpression could be the remnant of a function that has been lost during the evolution of Brucella. However, this gene potentially forms an operon with hdeA (unpublished results), which plays a role in the resistance to acid stress in B. abortus. These results suggest that Rsh may prepare the bacteria to respond to a possible acid stress subsequent to nutrient deficiency.
Experimental evidence indicates that production of reactive oxygen intermediates (ROIs) represents one of the primary antimicrobial mechanisms, and ROI formation has been described during macrophage infection by Brucella. Among the genes positively regulated by Rsh that contribute to the virulence of Brucella, we identified sodC encoding the Cu-Zn superoxide dismutase. A sodC mutant of B. abortus was more sensitive to killing by O2
− than the wild-type, and exhibited an increased susceptibility to killing in cultured macrophages and in a murine model of infection . We suggest that the lack of up-regulation of this protective gene might participate in the attenuation previously described for the Δrsh mutants of Brucella[22, 23]. In addition, it has been reported recently for biofilm-forming P. aeruginosa that an active stringent response increased not only antibiotic tolerance, but also antioxydant defenses by increasing production of active catalase and SOD against endogenous oxydant production . This indicated at least a partial conservation of stringent response-dependent adaptation strategies of different bacterial pathogens to various stress conditions encountered during chronic infection or biofilm formation.
One of the hallmarks of stringent response is the down-regulation of protein synthesis, together with induction of amino acid biosynthesis pathways. Based on the transcriptome results obtained for the well-studied stringent response of E. coli[7, 53] and on earlier reports , a major event of the stringent response consists in the repression of the translation apparatus, including ribosomal proteins. As expected, the transcriptome data of our study clearly demonstrated that the regulation of the translation apparatus is (p)ppGpp-dependent, as we identified 29 genes coding for ribosomal proteins that were down-regulated. The microarrays used in this study lack the genes of the rRNA operons of Brucella, explaining why these genes were not also identified as being down-regulated in the wild-type strain during stringent response. Regarding induction of amino acid biosynthesis pathways, it has been described that these pathways differ, depending on the bacterial species studied. In E. coli, the stringent response positively controls the biosynthesis of the branched-chain amino acids, glutamine/glutamate, histidine, lysine, methionine and threonine . In B. subtilis, proteome and transcriptome analysis has shown induction of enzymes involved in branched-chain amino acid biosynthesis during stringent response . The assessment of the amino acid biosynthesis pathways positively regulated by Rsh during the stringent response in B. suis revealed that only methionine biosynthesis was controlled by (p)ppGpp. As we previously put forward the hypothesis of a nutrient-poor environment during macrophage infection, based on the observation of the attenuation of several mutants affected in amino acid and nucleotide biosynthesis pathways [22, 56], we verified the possibility that Rsh-dependent methionine synthesis might be crucial during infection. We therefore constructed a ΔmetH allelic exchange mutant of B. suis, which was auxotrophic. Intracellular growth of the ΔmetH mutant, however, was not impaired, indicating that the lack of methionine biosynthesis in the Δrsh mutant did not contribute to its attenuation and that methionine was available in the Brucella-containing vacuole. Our results are in contrast to those described by Lestrate et al., who observed attenuation of a B. melitensis metH transposon mutant in vitro and in vivo, but, surprisingly, in the absence of any auxotrophic phenotype in minimal medium. In addition, genome sequencing of B. melitensis has since revealed that a gene encoding an alternative enzyme functionally replacing MetH is not present . Interestingly, in the α-proteobacterium and plant symbiont R. etli, (p)ppGpp-dependent upregulation of any amino acid biosynthesis pathways during stationary phase could not be evidenced. Rather, amino acid biosynthesis was down-regulated in a (p)ppGpp-independent manner under these growth conditions . It therefore appears that induction of amino acid biosynthesis during stringent response is not a key feature in α-proteobacteria.
The nar operon, which encodes the respiratory nitrate reductase in Brucella, was also induced under stringent conditions, as evidenced by the microarray analysis and by dosage of the nitrites produced by the nitrate reductase. Nar-induction was shown by the Rsh-dependent, positive control of the genes narG, narJ, and narK, involved in the first step of denitrification consisting in the reduction of nitrate to nitrite, and in nitrite extrusion towards the periplasm [59, 60]. This denitrification pathway may allow Brucella to survive under low oxygen tension, using nitrogen oxides as terminal electron acceptors . Denitrification can also be used by brucellae to detoxify NO produced by activated murine macrophages during infection, and part of this denitrification island is important for virulence of B. suis in vivo. Recent proteomic and gene fusion studies with B. suis evidenced the induction of the nar operon under microaerobic conditions . In E. coli, expression of the narGHJI operon is induced by low oxygen tension and by the presence of nitrate , which is imported by NarK, also responsible for nitrite export . One might speculate that our observations made in GMM broth were due to low-oxygen exposure of the strains, but this appears unlikely as pre-cultures were diluted in minimal medium under vigorous shaking for 4 hours. In M. tuberculosis, oxygen concentration-independent expression of nar has been described , and nitrate reductase component genes narH and narI have been identified as being under the positive control of RelMtb (Rsh) under comparable nutrient starvation culture conditions . It is therefore conceivable that the Brucella nitrate reductase is expressed in a stringent response-dependent manner under aerobic conditions, and further induced under hypoxia. In analogy to the observations made for the nar-operon, genes encoding cbb3-type cytochrome c oxidase were also observed being under the positive control of Rsh in B. suis, despite normal oxygenation. We hypothesize that stringent response, mediating adaptation of the pathogen to nutrient stress, therefore also represents a first step of potential adaptation to successive reduction of oxygen concentrations encountered by the pathogen in the host cells, the target organs and granulomes or abcesses.
The gene sodC, encoding Cu, Zn superoxide dismutase, was also positively regulated by the product of Rsh. The mutant was significantly more sensitive to O2
- radicals then the wild-type in vitro, already at short times of treatment. It has been described previously that a sodC mutant of B. abortus exhibited much greater susceptibility to killing by O2
- than the parental strain . The same mutant was also much more sensitive to killing in cultured macrophages, as well as in the murine model of in vivo infection, due to its inability to detoxify the O2
- generated by the respiratory burst of the phagocytes . In addition to the previous observation of some of us that virB expression is Rsh-dependent , we therefore have now validated by a biological survival assay a second Brucella gene product whose activity was necessary for intramacrophagic and intramurine replication and whose expression was controlled by Rsh during starvation. In B. suis, stringent response therefore also participated in protection from oxidative stress. A similar observation has been reported lately for Pseudomonas aeruginosa, where stringent response mediates increase of antioxidant defenses .
Interestingly, both urease operons of B. suis were regulated during stringent response in GMM: The ure-1 operon, responsible for the urease activity observed in most species [65, 66], was less expressed in the (p)ppGpp-producing wild-type strain than in the mutant, whereas ure-2 was more expressed under these conditions. In rich broth, i.e. under non-stringent conditions, urease activity, correlating with expression, has been described to be at its maximum in the absence of ammonium chloride, and enzymatic activity decreases with increasing ammonium concentrations . Interestingly, a similar negative regulation of urease expression during stringent response has been described in Corynebacterium glutamicum. The functions of the ure-2 cluster have been unraveled only recently: it is composed of genes encoding a urea and a nickel transport system within a single operon . Our results are in agreement with those published by Rossetti et al., comparing the transcriptional profiles of B. abortus under logarithmic and stationary growth phase conditions: the ure-2 operon is induced during stationary phase, known to be triggered by stringent response .
The observation that flaF and fliG, belonging to flagellar loci I and loci III, respectively, were down-regulated, suggested that the flagellar apparatus of Brucella was repressed under stringent conditions. This result is in agreement with the E. coli transcription profile of the stringent response, where the expression of the flhDC flagella master regulator is rapidly down-regulated . Hence, bacteria appear to shut down transcription of the flagellar cascade under starvation. This strategy makes physiological sense, as it would avoid the expending scarce energy resources for one of the bacteria’s largest macromolecular complexes. In brucellae, the biological function of the flagellar components has not been determined yet, especially in the context of host cell infection. Very recently, down-regulation of flagellar genes expression by the transcriptional regulator MucR has been described in B. melitensis. Our observation that MucR is positively regulated during stringent response in B. suis, whereas genes encoding the flagellar apparatus are down-regulated under the same conditions, indicates that such a link also exists in B. suis and, more generally, that (p)ppGpp is located high up in the hierarchy of gene regulation in Brucella spp.
Altogether, the transcription analysis of the stringent response in the facultatively intracellular pathogen B. suis confirmed the pleiotropic character of the (p)ppGpp-mediated adaptation to poor nutrient conditions. Several of the genes identified as being under (p)ppGpp control have also been described previously as being essential for the virulence of the pathogen , establishing a link between stringent response and virulence. Among these, five genes encoded proteins involved in cell envelope biosynthesis, of which four are essential for intramacrophagic replication of Brucella. In addition, four outer membrane proteins (Omps) encoded by the genes BR0119, BR0971 (both putative Omps), BRA0423 (“Omp31-2”), and BR1622 (“Omp31-1”), all under the positive control of (p)ppGpp during stringent response, have also been identified previously as being up-regulated in intramacrophagic B. suis at 48 h post infection . At this rather late stage of intracellular infection, B. suis shows a high level of replication, and these four genes/proteins were described as being up-regulated in both studies, indicating the potential importance of these Omps throughout the various stages of intramacrophagic infection.
Last not least, the function of 33% of the genes encoding Rsh-dependent transcripts remains unknown. This illustrates the complexity of the processes involved in adaptation to nutrient starvation while bearing in mind that knowledge of a substantial part of the Brucella genome is still limited. In this context, it is also worthwhile to mention that in E. coli, stringent response induces the alternative, stationary-phase sigma factor RpoS [7, 53]. In brucellae and other α-proteobacterial species, however, a functional analogue of RpoS was unknown. Only very recently, an intact general stress response system (GSR), including genes encoding the regulator phyR and the alternative sigma factor rpoE1, has been described for B. abortus. The transcriptome analysis of the stringent response in B. suis described here did not give any indication that stringent response might directly affect expression of these two GSR-related genes. Either stringent response controls another, yet uncharacterized alternative sigma factor in Brucella, or binding of (p)ppGpp to the RNA polymerase core affects competition between sigma factors in favor of an alternative sigma factor, which could then be RpoE1, increasing specific expression of stress-related genes.