Regulation of Virulence Genes in Bordetella pertussis

National Institute Of Diabetes And Digestive And Kidney Diseases

Investigators

Linked publications, trials & patents

Abstract

Nearly all virulence factors in Bordetella pertussis are activated by a master two-component system, BvgAS, composed of the sensor kinase BvgS and the response regulator BvgA. When BvgS is active, BvgA is phosphorylated (BvgAP), and virulence activated genes are expressed (the Bvg(+) mode). When BvgS is inactive and BvgA is not phosphorylated, virulence repressed genes are induced (the Bvg(-) mode). The Bvg(i) mode represents an intermediate state, with an intermediate concentration of BvgAP where kinase-on and kinase-off BvgS proteins may co-exist in equilibrium. Virulence genes include those encoding adhesins, such as fhaB (filamentous hemagglutinin), which are needed to adhere to the ciliated epithelial cells within the upper respiratory tract, and toxins, which cause the major symptoms of whooping cough disease. Several of these BvgA-activated gene products are components of the acellular pertussis vaccine used in the U.S. and Western Europe. We previously used transcriptome sequencing (RNA-seq) and reverse transcription-quantitative PCR (RT-qPCR) to define the BvgAS-dependent regulon of B. pertussis Tohama I. Our analyses revealed more than 550 BvgA-regulated genes, of which 353 were newly identified. BvgA-activated genes include those encoding two-component systems, multiple other transcriptional regulators, and the extracytoplasmic function (ECF) sigma factor brpL, which is needed for type 3 secretion system (T3SS) expression, further establishing the importance of BvgAP as an apex regulator of transcriptional networks promoting virulence. Most importantly, we showed for the first time that genes for multiple and varied metabolic pathways are significantly upregulated in the B. pertussis Bvg(-) mode. These include genes for fatty acid and lipid metabolism, sugar and amino acid transporters, pyruvate dehydrogenase, phenylacetic acid degradation, and the glycolate/glyoxylate utilization pathway. Our results suggested that metabolic changes in the Bvg(-) mode may be participating in bacterial survival, transmission, and/or persistence and identified >200 new Bvg(-) mode genes that could be tested for function. To expand this work we used the RNA-seq data set to conduct a genome-wide transcriptomic search for non-coding small RNAs (sRNAs) in B. pertussis. sRNAs play a crucial role in post-transcriptional regulation of gene expression in all organisms. A major class of sRNAs in bacteria regulates translation and mRNA stability by base pairing with their target mRNAs via an interaction facilitated by the RNA chaperone Hfq. In pathogens, Hfq and Hfq-dependent sRNAs regulate a wide spectrum of virulence gene expression and are involved in key steps of the infection process. To identify sRNAs in B. pertussis, WT and bvgAS- strains were grown both without MgSO4 (nonmodulating conditions, resulting in the BvgA(+) mode) and with MgSO4 (modulating conditions, resulting in the BvgA(-) mode). To process the data, we performed a computational analysis using the prokaryotic sRNA search program, ANNOgesic, which was recently developed to surpass the limitations of current bacterial sRNA search programs. We picked 20 candidates to analyze by Northern blots and Hfq-binding studies. Our study demonstrates that combining RNA-seq, ANNOgesic, and molecular techniques is a successful approach to identify various BvgAS-dependent and Hfq binding sRNAs, which may unveil the roles of sRNAs in pertussis pathogenesis. BvgAP-mediated regulation is accomplished by dimers of BvgAP, whose C-terminal domains (CTDs) sit upstream of the promoter at imperfect 14 bp palindromes with centers separated by 22 bp (2 helical turns). In the case of the B. pertussis promoter for fhaB, the locations of 3 BvgAP dimers have been determined upstream of the transcription start site. However, precise three-dimensional structural information has not been available for how these dimers and RNA polymerase are located and interact at this, or any other, BvgAP-activated promoter. Previous work has indicated that the sequence of BvgA shares conservation with NarL, an E. coli response regulator, and that like NarL, BvgA is composed of 3 domains: an N-terminal domain (NTD), a linker, and a CTD that interacts with the DNA. Consequently, we used the full length structure of NarL (NarLFL) to model a full length structure for BvgA (BvgAFL) and the structure of a NarLCTD dimer on the NarL DNA site to model a BvgACTD dimer on a BvgA DNA site. We found that the AlphaFold2 predicted structure and our independently modeled structure of BvgAFL were quite similar, suggesting that our modeled BvgAFL structure is highly probable. In BvgAFL, the NTD and CTD are close and are separated by a flexible linker that can accommodate movement of the NTD relative to the CTD. To test how close the NTD is relative to the CTD in the free protein, we generated the BvgA variant S96C/T194C since residues 96 and 194 are predicted to be adjacent within the FL structure. Using the crosslinker BMB, we found that these residues are between 8 and 12 angstroms apart when using the free protein. This distance, which is somewhat greater than that predicted by the models, is consistent with linker flexibility. Interestingly, the BMB crosslink is observed both in the presence or absence of BvgA phosphorylation, indicating that phosphorylation of BvgA is not required for this conformation. The modeled BvgFL structure also provided the information needed to select residues within BvgA for conjugation with the chemical cleaving reagent FeBABE. From the FeBABE footprinting analysis, we were able to determine the positions of BvgANTD, linker, and BvgACTD relative to the DNA. The BvgAFL model predicts that the position of the NTD precludes DNA binding by the CTD, even if the distance between S96 and T194 is the maximum of 12 angstroms. Thus, a BvgACTD dimer cannot interact with the DNA without movement of the NTDs to either specific or random locations. Our FeBABE footprinting analyses conjugated NTD residues indicated that within the transcription complex the BvgANTDs move to specific locations and revealed the side of the NTD that faces the DNA. From this work, we proposed a speculative model for how binding of BvgAP dimers at all three PfhaB binding sites might work. In this scenario, a dimer of BvgACTDs occupies each of the three binding sites, requiring the movement of the NTDs to new locations in the transcription complex as compared to their locations in BvgAFL. Furthermore, because of the predicted bending of the DNA from BvgA binding, we posited that the upstream sites could possibly bend back toward the RNAP/core promoter region. Although the accuracy of this transcription model awaits experimental visualization, our work demonstrates how FeBABE footprinting can provide useful information for future attempts to visualize the architecture of protein/DNA complexes.