Molecular control of gene expression by Brucella BaaR, an IclR-type transcriptional repressor

The general stress response sigma factor σE1 directly and indirectly regulates the transcription of dozens of genes that influence stress survival and host infection in the zoonotic pathogen Brucella abortus. Characterizing the functions of σE1-regulated genes therefore would contribute to our understanding of B. abortus physiology and infection biology. σE1 indirectly activates transcription of the IclR family regulator Bab2_0215, but the function of this regulator remains undefined. Here, we present a structural and functional characterization of Bab2_0215, which we have named Brucella adipic acid-activated regulator (BaaR). We found that BaaR adopts a classic IclR-family fold and directly represses the transcription of two operons with predicted roles in carboxylic acid oxidation. BaaR binds two sites on chromosome II between baaR and a divergently transcribed hydratase/dehydrogenase (acaD2), and it represses transcription of both genes. We identified three carboxylic acids (adipic acid, tetradecanedioic acid, and ϵ-aminocaproic acid) and a lactone (ϵ-caprolactone) that enhance transcription from the baaR and acaD2 promoters. However, neither the activating acids nor caprolactone enhanced transcription by binding directly to BaaR. Induction of baaR transcription by adipic acid required the gene bab2_0213, which encodes a major facilitator superfamily transporter, suggesting that Bab2_0213 transports adipic acid across the inner membrane. We conclude that a suite of structurally related organic molecules activate transcription of genes repressed by BaaR. Our study provides molecular-level understanding of a gene expression program in B. abortus that is downstream of σE1.

cellular environments (1). A fundamental mechanism required for physiological adaptation is the regulation of gene expression. The Gram-negative bacterium, Brucella abortus, is a facultative intracellular pathogen that can cause abortion in mammals and is among the most common zoonotic pathogens globally (2,3). B. abortus encodes an extracytoplasmic function-type sigma factor, E1 , which directly and indirectly regulates transcription of ϳ100 genes in response to environmental perturbation. E1 -dependent transcription confers resistance to multiple environmental stressors and is required for maintenance of chronic B. abortus infection in a mouse model (4 -7). Assigning physiological and/or biochemical functions to E1regulated genes is therefore important for understanding B. abortus stress physiology and infection biology.
Specifically, we have solved an X-ray crystal structure of BaaR, which revealed a classic IclR fold (11) with a C-terminal ligand-binding domain (LBD) 3 and an N-terminal helix-turn-helix (HTH) DNA-binding domain. We demonstrate that deleting baaR (⌬baaR) results in strong transcriptional up-regulation of four genes located adjacent to baaR on chromosome II of B. abortus 2308. This regulated gene set shares sequence similarity with characterized dicarboxylic acid (dca) ␤-oxidation operons from Acinetobacter and Pseudomonas (36 -39). We found that BaaR functions as a transcriptional repressor by binding to two conserved palindromic motifs between baaR and bab2_0216 (previously annotated as acaD2 (acyl-CoA dehydrogenase 2) (40)). A set of structurally related organic molecules enhanced transcription from a BaaR-dependent transcriptional reporter in vivo. Activation of transcription by one of these molecules, adipic acid, required the transporter gene bab2_0213, providing evidence that bab2_0213 encodes an adipic acid transporter. However, none of the small molecules that activate transcription in vivo were found to bind directly to purified BaaR in vitro. We thus conclude that the molecular signal(s) that directly control BaaR-dependent transcription in the cell are distinct from the activating molecules we identified via selective addition to the growth medium.

Regulation of baaR by E1 is likely indirect, and baaR does not contribute to B. abortus hydrogen peroxide (H 2 O 2 ) resistance
Previously, wild-type (WT) B. abortus strain 2308 and B. abortus strain 2308 harboring an in-frame deletion of the general stress regulator rpoE1 (encoding E1 ) were subjected to H 2 O 2 stress, and differences in gene expression between the two strains were assessed by RNA-Sequencing (RNA-seq) (4). Transcription of baaR was decreased 2-fold in the ⌬rpoE1 strain relative to the WT strain (Fig. 1A), suggesting that baaR is directly or indirectly activated by E1 . We could not identify a E1 -binding site in the baaR promoter region, which suggests that this regulatory effect is indirect. To evaluate the contribution of baaR to E1 -dependent protection against H 2 O 2 stress, we measured the survival of the WT, ⌬rpoE1, and ⌬baaR strains after treatment with 5 mM H 2 O 2 . Survival of ⌬rpoE1, assessed by enumerating colony-forming units (CFU) on solid medium after hydrogen peroxide treatment, was reduced by ϳ1 order of magnitude, whereas ⌬baaR survival did not differ from WT survival (Fig. 1B). We thus concluded that decreased expression of baaR in ⌬rpoE1 does not contribute to the viability defect of ⌬rpoE1 under the assayed condition.
Remarkably, compared with typical IclR protein sequences, BaaR has an unusually long N-terminal region. This extension is ϳ38 amino acids (from residue Met-1 to Gln-38) and is similar in size to the N termini of Acinetobacter DcaR and DcaS and Pseudomonas P1630 ( Fig. 2C and Fig. S1). For crystallization, a trimmed version (residues Lys-21-Pro-284) of BaaR was used; the N terminus appeared to be mostly unstructured with the exception of residues Gly-31-Asp-36, which adopted an ␣-helical fold (␣1Ј) in three of four monomers (Fig. 2, A and C). The functional relevance of this region is unknown; we cannot exclude the possibility that an alternative start codon is used for protein translation in vivo.
The BaaR LBD consists of five ␣-helices (␣5, ␣6, ␣7, ␣8, and ␣9) and six anti-parallel ␤-strands (␤3, ␤4, ␤5, ␤6, ␤7, and ␤8) arranged in a semi-circular ␤-scaffold. The ␤-scaffold is partially occluded by ␣6 and folds to form a cavity in the LBD (Fig.  2, A and C). In the LBD cavity of each IclR protein, we observed extra electron density consistent with a bound acetate molecule (Fig. 2, A and B). This molecule is likely derived from the crystallization solution, which contained 200 mM calcium acetate. Notably, the acetate occupied the same position as pyruvate or glyoxylate molecules that have previously been reported to bind to the LBD of Escherichia coli IclR (Fig. 2B) (41,42). Additional structural alignments and a schematic of hydrogen bonding to bound acetate (and its corresponding electron density in the BaaR cavity) are presented in Fig. S2.
Protein sequence alignment of BaaR, Acinetobacter DcaR (ACIAD1688) and DcaS (ACIAD1684), and Pseudomonas PA1630 revealed high sequence identity and common secondary structural features (Fig. 2C). All have HTH domains with long N termini; these proteins also have LBD cavities comprising very similar residues, suggesting that they recognize and respond to identical or similar small molecules and interact with target DNA in a comparable manner ( Fig. 2C and Fig. S1). The two BaaR dimers present in the asymmetric unit are highly superimposable and displayed few structural differences (r.m.s.d ϭ 0.76). IclR-family proteins have been found to adopt dimeric or tetrameric conformations in X-ray structures, in solution, or in their interactions with DNA (11, 41, 43-45).

Transcription regulation by Brucella BaaR
Therefore, even if BaaR crystallizes as a dimer in vitro, it potentially exists in different oligomeric states in vivo.

BaaR represses the transcription of a set of genes with similarities to the dca operon
Our crystal structure clearly identified BaaR as an IclR-family transcriptional regulator capable of binding a carboxylic acid (acetate) via its LBD. To identify genes specifically regulated by BaaR, we measured the global transcriptomic profile of the ⌬baaR strain using RNA-seq and compared it with that of the WT strain (Table S2). Among the differentially expressed genes, four genes were highly up-regulated in ⌬baaR relative to WT (Fig. 3A). These four genes, bab2_0213, bab2_0214, bab2_0216, and bab2_0217, are contiguous and adjacent to baaR (bab2_0215). Genes bab2_0213, bab2_0214, and bab2_0215 are located on the opposite strand from bab2_0216 and bab2_0217, and the two gene sets are likely divergently transcribed (Fig. 3B). bab2_0213 is annotated as a MucK transporter, a member of the major facilitator superfamily (MFS). A transporter related to bab2_0213 in Acinetobacter sp. ADP1 is involved in cis,cis-muconic acid uptake (39). bab2_0214 is annotated as an acyl-CoA dehydrogenase; this family of flavo-Figure 2. X-ray crystal structure of the B. abortus IclR protein, BaaR (Bab2_0215; residues 21-284). A, BaaR adopts the classic fold characteristic of dimeric IclR proteins. The dimer (monomer A ϩ monomer B) contains an N-terminal DNA-binding domain with a winged HTH fold (Phe-39 -Asp-101), ␣-helical linker (␣4, residues Ile-102-Leu-115), and a C-terminal LBD (Met-116 -Pro-284). The Lys-21-Gln-38 portion of the N-terminal extension containing ␣-helix ␣1Ј is also represented. An acetate molecule, likely derived from the crystallization buffer, was found in the LBD of each monomer. In monomer A, ␤-strands are represented as yellow arrows and ␣-helices as blue cylinders. Monomer B is in gray. The C and N termini of monomer A are in red. B, close-up view of the LBD cavity in monomer A. The acetate molecule superimposes with glyoxylate and pyruvate, two ligands found in the E. coli IclR LBD cavity. PDB structures 5WHM, 2O9A, and 2O99 were used to structurally align the LBD structures of B. abortus BaaR and E. coli IclR bound to glyoxylate and pyruvate, respectively. The corresponding r.m.s.d. were 1.27 and 1.29 Å, respectively, for the BaaR-LBD structure when compared with E. coli IclR-LBD bound to glyoxylate or pyruvate. C, amino acid sequence alignment of B. abortus BaaR (Bab2_0215), Acinetobacter DcaR (ACIAD1688) and DcaS (ACIAD1684), and Pseudomonas DcaR (PA1630). Residues in the black boxes are identical; residues in the gray boxes are homologous, and residues in white do not share any homology. B. abortus BaaR secondary structures are shown above the alignment; ␤-strands are represented as yellow arrows and ␣-helices as blue cylinders. A light green line and light red line delimit the HTH domain and LBD, respectively. Residues found in the B. abortus BaaR LBD cavity are indicated by red triangles. For each amino acid sequence, the residues are numbered, starting with methionine.

Transcription regulation by Brucella BaaR
proteins catalyzes ␣,␤-dehydrogenation of fatty acid acyl-CoA conjugates (46,47). bab2_0216 and bab2_0217 are annotated as pseudogenes. However, this annotation appears to be incorrect: a search of Pfam (48) using the primary structures of these loci suggested that bab2_0216 (acaD2), is in fact two fused, in-frame genes that correspond to an enoyl-CoA hydratase/ isomerase and a 3-hydroxyacyl-CoA dehydrogenase. Both of these enzymes are predicted to be involved in ␤-oxidation of fatty acids (49,50). Our transcriptomic data corroborate a recent proteomic study in B. abortus in which a small peptide corresponding to the Bab2_0216 N terminus was identified by MS (51), suggesting that the full-length "fusion" protein is expressed. This hypothesis is supported by known examples of functional proteins (FadB and FadN in E. coli) that simultaneously contain an enoyl-CoA hydratase/isomerase and a 3-hydroxyacyl-CoA dehydrogenase in a single polypeptide (52).
bab2_0217 encodes a protein with homology to the CoAtransferase family III, a class of enzymes that catalyzes reversible transfer reactions of CoA groups from CoA-thioesters to free acids (53). An early stop codon present in the sequence of this gene truncates the corresponding protein by 45 residues relative to related CoA-transferase family III proteins. The adjacent gene bab2_0218 includes these 135 nucleotides missing from bab2_0217; the functional significance of bab2_0217 truncation is not known.
Two additional gene sets were differentially regulated in ⌬baaR relative to WT. The first corresponds to 12 adjacent genes (bab2_1036 -bab2_1047) (Fig. 3A) with predicted enzyme functions similar to those associated with the bab2_0213-bab2_0217 operon. Two of the most strongly upregulated genes, bab2_1046 and bab2_1045, are annotated as enoyl-CoA hydratase/isomerase and 3-hydroxyacyl-CoA dehydrogenase, respectively, and may have functionally redundant roles with bab2_0216. Gene bab2_1036 is annotated as a CoAtransferase family III protein and may be redundant with bab2_0217. Expression of the second gene set, corresponding to bab2_0277-bab2_0282 (Fig. 3A), was significantly lower in ⌬baaR relative to WT. The majority of these genes encode components of an ATP-binding cassette (ABC) transport system predicted to be involved in branched-chain amino acid transport. Comparison of the amino acid sequence of the cognate periplasmic binding protein (PBP) with the sequences of PBPs that co-crystallize with bona fide ligands revealed that Bab2_0282 shares 54% identity and 72% similarity with a Burkholderia mallei PBP (Protein Data Bank (PDB) code 3I09) that co-crystallizes in a closed conformation with acetoacetate. Residues involved in the interaction with acetoacetate in this PBP are conserved in Bab2_0282 (Fig. S3), suggesting that the B. abortus bab2_0278 -bab2_0282 ABC transporter operon is involved in uptake of acetoacetate or a related molecule. Acetoacetate is produced in the liver during ketogenesis. Under certain conditions, the acetyl-CoA formed in the liver from ␤-oxidation of fatty acids can be converted into ketone bodies (acetoacetate, D-␤-hydroxybutyrate, and acetone) for export to other tissues (54,55). This Brucella ABC transporter operon is proximal to a gene (bab2_0277) annotated as a glucose/methanol/choline oxidoreductase, which is also expressed at significantly lower levels in ⌬baaR. In B. mallei, a similar gene (BMA2933) also co-occurs with the related ABC transporter, suggesting that this system may have a function in ketone metabolism.
Finally, genes bab1_0303 (UreG1 urease accessory protein), bab1_0578 (BetI TetR transcriptional regulator), bab1_0914 (DUF1127), and bab2_0548 (ABC transmembrane transporter) also showed significant differential regulation in our data set. The functional significance of these transcriptional changes is not known (see Table S2 for the full data set). A, gene transcription levels were compared between a B. abortus strain harboring a deletion in baaR and the WT strain. The x axis represents the log 2 fold change in gene transcription between the B. abortus WT and ⌬baaR strains. Negative numbers represent genes down-regulated, and positive numbers represent genes up-regulated in the ⌬baaR background. The y axis FDR-adjusted p values were transformed on a Ϫlog 10 scale; a difference in gene transcription levels represented by Ϫlog 10 FDR-adjusted p values of Ͼ1 was considered significant. Genes with high fold changes and high FDR-adjusted p values that were located in the same genomic regions were grouped. Genes present in the bab2_0213-bab2_0217 locus are shown in red (bab2_0213, MFS transporter; bab2_0214, acyl-CoA dehydrogenase; bab2_0216 (acaD2), pseudogene/enoyl-CoA hydratase-isomerase/3-hydroxyacyl-CoA dehydrogenase; bab2_0217, pseudogene/CoA-transferase family III); genes present in the bab2_1036 -bab2_1047 locus are shown in blue (bab2_1036, CoA-transferase family III; bab2_1037-bab2_1040, ABC transporter operon; bab2_1041, IclR; bab2_1042, pyruvate dehydrogenase E2 component; bab2_1043, periplasmic binding protein; bab2_1044, pseudogene/dehydrogenase E1 component/ transketolase; bab2_1045, 3-hydroxyacyl-CoA dehydrogenase; bab2_1046, enoyl-CoA hydratase; bab2_1047, acyl-CoA dehydrogenase); and genes present in the bab2_0277-bab2_0282 locus are shown in green (bab2_0277, glucose/methanol/choline oxidoreductase; bab2_0278 and bab2_0279, branched-chain amino acid transport system permease proteins/ABC transporter; bab2_0280 and bab2_0281, branched-chain amino acid transport system ATP-binding proteins/ABC transporter; bab2_0282, branched-chain amino acid transport system substrate-binding protein/ABC transporter). Other genes, highlighted in gray, were also strongly regulated, but no clear connection between those genes and the BaaR protein could be made (bab1_0303, UreG1 urease accessory protein; bab1_0578, BetI TetR transcriptional regulator; bab1_0914, DUF1127; bab2_0548, ABC transmembrane transporter). B, genomic organization of the bab2_0213-bab2_0217 operon. Genes bab2_0213/bab2_0214/baaR and genes bab2_0216/bab2_0217 are on opposite strands and divergently transcribed. Gene baaR (black arrow) represses its own expression and that of the bab2_0213-bab2_0217 genes.

Transcription regulation by Brucella BaaR BaaR binds palindromic motifs in its own promoter region
To evaluate the mechanism of BaaR-dependent transcriptional repression, we measured binding between purified BaaR and a region of DNA corresponding to its promoter. Because the IclR protein family recognizes and specifically interacts with palindromic motifs (11), the DNA sequence corresponding to the promoter region between baaR and bab2_0216 was analyzed for palindromes. We identified two similar palindromic regions that were 44 and 36 nucleotides away from the bab2_0216 and baaR start codons, respectively. These two regions were named BaaR-binding site 1 (BBS1) and BaaR-binding site 2 (BBS2), and each consisted of two overlapping palindromic sequences (Figs. 4 and 5). A 375-nucleotide fluorescent DNA probe corresponding to the promoter region between baaR and bab2_0216 was used for gel-shift experiments. As expected, 5 ng of this "long" fluorescent DNA probe corresponding to the WT sequence (L WF , long, and WT, fluorescent) was shifted in size on the gel in the presence of increasing concentrations (2-500 nM) of purified BaaR, confirming that BaaR interacts with the DNA region between bab2_0215 and bab2_0216 (Fig. 4A, To determine whether this interaction was specific, we conducted a series of control experiments using unlabeled probes as well as mutated probes in which the palindromic motifs were randomized to maintain base content and length. As observed previously, 5 ng of the L WF probe were shifted in size when mixed with increasing concentrations (62-500 nM) of BaaR (Fig. 4A, panel 2). However, this shift was weakened when 5 ng of the L WF probe were mixed with 500 nM BaaR, a concentration at which all of the DNA probe was shifted, and increasing concentrations (5-200 ng) of an equivalent unlabeled (nonfluorescent) DNA probe (L WU , long, and WT, unlabeled), indicating that the fluorescent and unlabeled probes competitively interacted with BaaR (Fig. 4A, panel 3). This result suggests that the interaction between BaaR and its promoter is specific. In a related control, 5 ng of the L WF DNA probe and 50 ng of a BBS-mutated unlabeled DNA probe (L MU , long, mutated, and . Electrophoretic mobility shift assay of the interaction between purified BaaR protein (residues 9 -283) and the bab2_0215-bab2_0216 intergenic region. A, panel 1, a long WT fluorescent DNA probe (L WF , 375 nucleotides long) corresponding to the intergenic region between baaR and bab2_0216, including BBS1 and BBS2, was mixed with increasing concentrations (2,4,8,16,31,62,125,250, and 500 nM) of purified BaaR and run on a 5% native acrylamide gel. 1st lane represents 5 ng of L WF alone (i.e. without BaaR) and was used as a size control. Panel 2, the L WF (5 ng) was mixed with increasing concentrations (62,125,250, and 500 nM) of purified BaaR and run on a 5% native acrylamide gel. The 1st lane (control) represents 5 ng of L WF alone (i.e. without BaaR). Panel 3, the L WF (5 ng) was mixed with 500 nM purified BaaR and increasing concentrations (5, 50, 100, and 200 ng) of a long WT unlabeled DNA probe (L WU ), corresponding to the intergenic region between baaR and bab2_0216, including BBS1 and BBS2. The 1st lane (control) represents 5 ng of L WF alone (i.e. without BaaR). Panel 4, the conditions used in this assay were similar to those used in panel 2, except that 50 ng of the L MU DNA probe (long mutated and unlabeled), corresponding to the intergenic region between baaR and bab2_0216 and carrying mutated BBS sequences, was added. The 1st lane (control) corresponds to 5 ng of L WF alone (i.e. without BaaR). Panel 5, the conditions used in this assay were similar to those used in panel 2, except that 5 ng of L MF probe (long mutated and fluorescent) was used instead. The 1st lane (control) corresponds to 5 ng of L MF alone (i.e. without BaaR) and was used as a size reference. B, gel-shift assays in panels 1-5 were performed as described in A; however, instead of using a DNA probe targeting the full-length baaR-bab2_0216 intergenic region, a shorter (196 nucleotides long) WT fluorescent DNA probe (S WF ) targeting only half of this region, carrying BBS2 but not BBS1, was used. A schematic representation of the different DNA probes used in this assay is presented at the top of each panel.

Transcription regulation by Brucella BaaR
unlabeled) were mixed together. This mutated probe did not disrupt the BaaR/L WF interaction, suggesting BaaR binding requires the BBS1 and BBS2 regions (Fig. 4A, panel 4). Finally, when we performed a gel-shift assay with a fluorescent DNA probe carrying mutated BBS1 and BBS2 (L MF , long, mutated, and fluorescent), no shift was observed, even at the highest BaaR concentrations (Fig. 4A, panel 5). We conclude that BaaR constitutively and specifically interacts with the bab2_0215-bab2_0216 promoter region, and this interaction requires BBS1 and BBS2.
Equivalent gel-shift assays were conducted using a shorter DNA probe (196 nucleotides) harboring only one BBS (BBS2) (Fig. 4B). This probe behaved exactly the same as the long probe in the presence of the purified BaaR (Fig. 4B). Control experiments performed with the different unlabeled or mutated DNA probes also confirmed the specificity of this interaction (Fig. 4B).
As discussed above, BaaR contains ϳ38 additional amino acids at its N terminus. To assess the role of this N-terminal extension in the interaction between BaaR and its target DNA, we performed gel-shift assays using a BaaR mutant protein missing these additional N-terminal amino acids. We observed no significant differences in probe shift between the short (residues Val-40 -Pro-284) and WT BaaR proteins; both resulted in shifts of the long (289 nucleotides, contains BBS1 and BBS2) and short (196 nucleotides, contains BBS2) fluorescent DNA probes (Fig. S4).

Palindromic regions between baaR and bab2_0216 are required for recognition by BaaR
To identify the motifs in the BBS region between baaR and bab2_0216 required for interaction with BaaR, each palindrome of BBS2 was independently mutated by randomizing the corresponding sequences. We then measured binding of the corresponding 196-nucleotide fluorescent DNA probe to BaaR by gel-shift assay. Compared with the WT BBS2-containing probe, no gel shift was observed when the first half of the first palindromic motif (region 1) was mutated, suggesting no interaction between the DNA probe and BaaR (Fig. 5, A and B). Mutation of BBS2 region 2, corresponding to overlapping regions between the first and second palindromic motifs, also ablated binding of the probe to BaaR (Fig. 5, A and B). Mutation of the second half of the second palindrome (region 3) resulted in an intermediate effect; a partial gel shift of the corresponding DNA probe was still observed with the highest BaaR concentrations (250 and 500 nM) (Fig. 5, A and B). As expected, mutating regions 1-3 completely ablated a gel shift (Fig. 5B). We conclude that regions 1 and 2, present in the TTTCGC/GC-GAAA palindrome, are required for interaction with BaaR and that region 3 plays a minor role in the BaaR/DNA interaction. We note that DNA regions upstream of bab2_0277-bab2_0282 and bab2_1036 -bab2_1047 do not contain similar palindromic sequences recognized by BaaR. This suggests that the transcriptional regulation of these loci observed in our RNA-seq data are indirect.

Transcription regulation by Brucella BaaR BaaR interacts with BBS1 and BBS2 in vivo and regulates transcription of the bab2_0213-bab2_0217 operon
We next evaluated whether BaaR interacts with the baaR-bab2_0216 promoter region in vivo and whether both BBSs are required for transcriptional regulation in B. abortus 2308. We transformed WT and ⌬baaR strains with a plasmid carrying a transcriptional fusion of the intergenic region between bab2_0215 and bab2_0216 to lacZ. This lacZ transcriptional reporter construct (P baaR -lacZ) contains the baaR promoter region, including both BBS1 and BBS2.
In the WT strain, ␤-gal activity was low, suggesting constitutive repression of transcription by BaaR. In the ⌬baaR strain, ␤-gal activity was increased by a factor of 10 relative to WT, suggesting that deletion of baaR derepresses transcription from These reporter genes are a transcriptional fusion between the lacZ and the baaR-bab2_0216 intergenic region encompassing WT or mutant (mut) BBS1 and BBS2. Panel 2, the WT and ⌬baaR B. abortus strains were transformed with different versions of the 1/2P baaR -lacZ reporter gene (boxed in green). These reporter genes corresponded to a transcriptional fusion between lacZ and half of the baaR-bab2_0216 intergenic region carrying the WT or mutated BBS2. BBS2 regions 1 (R1), 2 (R2), and 3 (R3) were also mutated individually. Panel 3, negative controls (boxed in pink), the WT and ⌬baaR B. abortus strains were transformed with the pMR15 empty vector (EV), and a ⌬baaR strain was complemented with the WT baaR allele and transformed with P baaR -lacZ and 1/2P baaR -lacZ reporter genes. B, as in A, the WT and ⌬baaR B. abortus strains were transformed with different versions of the P bab2_0216 -lacZ (boxed in violet) or 1/2P bab2_0216 -lacZ (boxed in green) reporter genes. These reporter genes correspond to a transcriptional fusion between lacZ and the reverse complement of either the full-length (encompassing BBS1 and BBS2) or half (BBS1 only) of the baaR-bab2_0216 intergenic region.

Transcription regulation by Brucella BaaR
its own promoter (Fig. 6A, panel 1). Surprisingly, randomizing the sequences of the BBS1 and BBS2 site (BBS1 mut /BBS2 mut ) resulted in low ␤-gal activity in both strains. It is likely that these tandem BBS1 and BBS2 mutations disrupt the ability of RNA polymerase to induce transcription from this promoter (Fig. 6A, panel 1).
Mutation of BBS1 alone (BBS1 mut /BBS2 WT ) had little effect on BaaR-dependent transcriptional repression; ␤-gal activity was low in the WT and high in the ⌬baaR strain (Fig. 6A, panel  1). However, in the WT strain, the measured ␤-gal activity was higher from the BBS1 mut /BBS2 WT than from the BBS1 WT / BBS2 WT reporter. This indicates that the integrity of both BBSs is likely required for efficient repression by BaaR, perhaps by permitting multiple BaaR dimers to interact simultaneously. Mutation of BBS2 alone (BBS1 WT /BBS2 mut ) resulted in low ␤-gal activity in both the WT and ⌬baaR strains, further supporting that RNA polymerase cannot efficiently induce transcription from this sequence (Fig. 6A, panel 1). We conclude that the integrity of BBS2 is essential for the proper interaction of transcription factors at this promoter.
When half of the intergenic region present between baaR and bab2_0216 was fused to lacZ, the corresponding reporter gene (1/2P baaR -lacZ) behaved similarly to the BBS1 mut / BBS2 WT reporter; ␤-gal activity was low in the WT and high in the ⌬baaR strain (Fig. 6A, panel 2). Again, transcription from the corresponding reporter with only one BBS (BBS2 WT ) was not as repressed as that from the reporter containing both BBS sequences (BBS1 WT /BBS2 WT ) (Fig. 6A, panel 2). Mutation of BBS2 (BBS2 mut ) resulted in low ␤-gal activity in both strains and could be attributed to a possible disruption in transcription (Fig. 6A, panel 2). To overcome this problem, we mutated smaller regions of BBS2 by randomizing the corresponding sequences. When BBS2 region 1 (BBS2 R1 ) or 2 (BBS2 R2 ) was mutated, the corresponding ␤-gal activity was high in both the WT and mutant strains (Fig. 6A, panel 2). From these data, we conclude that the integrity of the transcription initiation site is preserved in these constructs. This result also confirmed that BaaR in the WT strain fails to interact with BBS2 R1 or BBS2 R2 . When region 3 (BBS2 R3 ) was mutated, intermediate ␤-gal activity was measured in both strains (Fig. 6A, panel 2), which confirmed our previous in vitro observations (Fig. 5B). Taken together, these results suggest that BBS2 is required for BaaR-dependent transcriptional regulation of bab2_ 0215, bab2_0214, and bab2_0213. However, the presence of BBS1 ensures even greater transcriptional repression. WT or ⌬baaR strains carrying the empty vector (pMR15) or the baaR complemented strain were used as controls (Fig. 6A,  panel 3).
We next evaluated whether transcription from bab2_0216 -bab2_0217 exhibited a similar transcriptional profile to baaR. We constructed a new lacZ transcriptional reporter in which the reversed and complemented intergenic sequence between baaR and bab2_0216 was fused to lacZ (P bab2_0216 -lacZ). This construct was transformed into the WT or ⌬baaR strains. The ␤-gal activity in ⌬baaR was 5-fold higher than that in WT, providing evidence that BaaR represses transcription from this reporter as well (Fig. 6B, panel 1). The ␤-gal activity observed under this reporter was generally lower than that under P baaR -lacZ, suggesting weaker transcription from this promoter. When BBS1 and BBS2 were mutated simultaneously (BBS1 mut / BBS2 mut ) by randomizing the corresponding sequences, both the WT and ⌬baaR strains exhibited increased ␤-gal activity (Fig. 6B, panel 1). However, in the WT strain, these activities were higher and comparable with the ␤-gal activity levels measured in the ⌬baaR strain carrying the BBS1 WT /BBS2 WT reporter gene. In both strains, mutation of BBS2 only (BBS1 WT / BBS2 mut ) had no effect on ␤-gal activity, whereas mutation of BBS1 only (BBS1 mut /BBS2 WT ) induced greater activity (Fig. 6B,  panel 1).
Finally, a reporter containing half of the bab2_0216 promoter region was evaluated (1/2P bab2_0216 -lacZ) (Fig. 6B, panel  2). Transcriptional activity under this reporter in the WT strain was 5-fold lower than that in the ⌬baaR strain and was comparable with that in the BBS1 WT /BBS2 mut reporter strains (Fig.  6B, panel 2). When BBS1 was mutated, activity increased in both strains and was comparable with that in the BBS1 mut / BBS2 mut reporter strains (Fig. 6B, panel 2).
Together, these results provide evidence that BBS1 is required for BaaR-dependent transcriptional regulation of bab2_0216 and bab2_0217. BBS1 and BBS2 mutations did not disrupt the ability of RNA polymerase to induce transcription from this promoter, suggesting that the bab2_0216 -bab2_0217 initiation site does not overlap with BBS1 or BBS2.

⑀-Aminocaproic acid derepresses transcription from a BaaR-regulated promoter
BaaR, like other IclR proteins, has a C-terminal LBD with a cavity accommodating small molecules. Interaction with specific molecules can positively or negatively modify the affinity of the protein for its target DNA (11,14,17,41,42,44,56). We screened for small molecules that affect transcription from a BaaR-regulated reporter plasmid. Specifically, we transformed WT Brucella ovis, a closely related Biosafety Level 2 (BSL2) surrogate for B. abortus, with the P baaR -lacZ reporter and inoculated this strain into 96-well plates containing 480 distinct, individual small molecules. A single molecule, ⑀-aminocaproic acid, activated transcription from the P baaR -lacZ reporter under this cultivation condition (Fig. 7A). We confirmed this hit in a B. abortus strain carrying the same reporter plasmid (Fig. 7B). ⑀-Aminocaproic acid is a six-carbon molecule with a carboxyl and amine group. It is a lysine derivative and analog used in clinical settings to promote blood clotting (57,58). In this same transcription induction screen, acetate and acetoacetate did not affect transcription (Table S3). We thus conclude that neither acetoacetate nor acetate are ligands that induce transcription, even though we observe acetate bound to the LBD in the BaaR crystal structure.

Adipic acid, tetradecanedioic acid, and ⑀-caprolactone also derepress transcription
Given the results of our initial screen, we evaluated other related small molecules for their ability to activate transcription from a BaaR-regulated reporter. Selection of these molecules was based on their metabolic and physiological properties and Transcription regulation by Brucella BaaR chemical and structural similarities shared with ⑀-aminocaproic acid (Fig. S5). To narrow the number of molecules to be tested, the sequence of BaaR was compared with those of IclR proteins previously described to regulate similar metabolic pathways. BaaR shares 59 and 54% identity with DcaS (ACIAD1684) and DcaR (ACIAD1688), two Acinetobacter IclR proteins potentially involved in the transcriptional regulation of a dca ␤-oxidation operon (36,37,59). BaaR also has high identity (66%) to the Pseudomonas DcaR protein (PA1630) involved in regulation of ⑀-caprolactam catabolism and ␤-oxidation (36,38). In Acinetobacter ADP1 and Pseudomonas aeruginosa, dca operons have been previously described as essential for growth on adipic acid or ⑀-caprolactam as the sole carbon sources (36,38). Interestingly, a cis,cis-muconic transporter (ACIAD1681) is proximal to the Acinetobacter dca operon (39); this transporter is 67% identical to Bab2_0213 in B. abortus. The ability of adipic acid, ⑀-caprolactam, and cis,cismuconic acid to derepress transcription of P baaR -lacZ was therefore investigated in B. abortus. Addition of cis,cis-mu-conic acid did not enhance transcription, but 4 mM adipic acid strongly activated transcription ( Fig. 7B and Fig. S5). To evaluate the specificity of this activation, shorter or longer dicarboxylic acids (C3 to C14) were assessed as well. Only tetradecanedioic acid, a C14 dicarboxylic acid, significantly derepressed transcription from our reporter (Fig. 7B and  Fig. S5).
The trans,trans-muconic acid, a dicarboxylic acid closely related to cis,cis-muconic acid, was also evaluated, but it had no effect on transcription from the reporter (Figs. 7B and Fig. S5). Interestingly, the six-carbon fatty acid caproic acid had no effect on BaaR-dependent repression of P baaR -lacZ reporter gene transcription, although its cyclic form, ⑀-caprolactone (60), significantly derepressed transcription (Fig. 7B and Fig.  S5). Conversely, ⑀-caprolactam, a cyclic ⑀-aminocaproic acid (61), had no effect on transcription (Fig. 7B and Fig. S5). As a control, we cultivated the reporter strain in the presence of DMSO, which was used to solubilize most of the small molecules evaluated. Transcription from this reporter was Figure 7. Screening of small molecules to identify a BaaR effector. A, WT B. ovis strain carrying the P baaR -lacZ reporter gene was grown overnight (37°C, 5% CO 2 ) on phenotype microarrays (Biolog), and ␤-gal activity was measured under each condition. For each plate (PM1, PM2A, PM3B, PM4A, and PM5), molecules 1-96 are delimited on the x axes. Molecule G9 on plate PM3B, corresponding to ⑀-aminocaproic acid, is indicated with a black arrow, and its molecule structure is shown. A list of the different growth conditions evaluated and the corresponding ␤-gal activities measured can be found in Table S3. B, WT B. abortus strain carrying the P baaR -lacZ reporter was grown overnight in the presence of 4 mM of the indicated molecules or DMSO. Di-unsaturated and saturated dicarboxylic acids are highlighted by a pink or violet rectangle, respectively. As negative controls (gray rectangle), B. abortus carrying the pMR15 empty vector control (EV) was grown in the presence of 4 mM ⑀-aminocaproic acid, adipic acid, tetradecanedioic acid, ⑀-caprolactone, or DMSO. Error bars represent standard deviations. C, as in B, a WT B. abortus strain carrying the P bab2_0216 -lacZ reporter gene was grown overnight in the presence of 4 mM ⑀-aminocaproic acid, ⑀-caprolactone, adipic acid, tetradecanedioic acid, or DMSO. Error bars represent standard deviations.

Transcription regulation by Brucella BaaR
unchanged by addition of the DMSO (Fig. 7B). We also confirmed that the different molecules had no effect on an empty vector control strain (Fig. 7B). All molecules were also assessed in a B. abortus strain carrying the P bab2_0216 -lacZ reporter. The ␤-gal activity was significantly enhanced in the presence of 4 mM ⑀-aminocaproic acid, adipic acid, tetradecanedioic acid, and ⑀-caprolactone (Fig. 7C).

Evidence that adipic acid is transported by Bab2_0213, an MFS transporter
We next evaluated whether the molecules found to induce transcription from a BaaR-dependent reporter (P baaR -lacZ) are specifically transported by the MucK-like transporter Bab2_0213. We also tested whether Bab2_0214 (acyl-CoA dehydrogenase), Bab2_0216 (enoyl-CoA hydratase/isomerase and 3-hydroxyacyl-CoA dehydrogenase), and Bab2_0217 (CoA-transferase family III) affect BaaR-regulated transcription. We generated in-frame deletions of each of these genes in B. abortus and transformed the corresponding null mutant strains with the P baaR -lacZ reporter plasmid. Each deletion strain was grown in the presence of increasing concentrations of the different activating molecules, and ␤-gal activity was measured under each condition.
Transcription reporter activity in all four null mutant strains was the same as that in the WT in the presence of ⑀-aminocaproic acid, tetradecanedioic acid, or ⑀-caprolactone (Fig. 8, A-C). However, the strain harboring the bab2_0213 deletion exhibited significantly lower transcriptional activity than the WT strain and other deletion strains in the presence of adipic acid (Fig. 8D). Complementation of ⌬bab2_0213, by re-in-troducing a WT copy of bab2_0213, restored the WT transcriptional phenotype (Fig. 8E). We conclude that, at the concentrations tested, Bab2_0213 is involved in transport of adipic acid. It is not known how ⑀-aminocaproic acid, tetradecanedioic acid, or ⑀-caprolactone is transported. We further conclude that deletion of bab2_0214, bab2_0216, or bab2_0217 did not affect the response of B. abortus to these inducing molecules. This provides evidence that none of these enzymes are required for BaaR-dependent transcriptional regulation by adipic acid, ⑀-aminocaproic acid, tetradecanedioic acid, or ⑀-caprolactone.

Adipic acid does not support growth of B. abortus in GMM
A physiological role for adipic acid in Brucella cell physiology has not been defined. To test whether adipic acid can support growth of B. abortus, we cultivated the WT 2308 strain in Gerhardt's minimal medium (GMM) lacking one of its potential carbon sources (glycerol, lactate, or glutamate) and supplemented with 1 mM adipic acid (Fig. S6A). We observed substantial growth after 72 h in GMM containing all three molecules (glycerol, lactate, and glutamate) or missing glycerol. No growth was observed in GMM missing lactate or glutamate. Addition of adipic acid did not improve or rescue growth of B. abortus in any condition tested, suggesting it cannot be used as a major carbon source under these cultivation conditions. We postulated that constitutive expression of the bab2_ 0213-bab2_0217 locus might provide a growth advantage to B. abortus in a minimum medium containing adipic acid, so we repeated these same growth experiments with the baaR deletion strain. Again, we observed no growth in GMM lack-

Transcription regulation by Brucella BaaR
ing lactate or glutamate, with or without 1 mM adipate (Fig.  S6B).

Adipic acid, ⑀-aminocaproic acid, tetradecanedioic acid, and ⑀-caprolactone do not interact directly with BaaR
Given the ability of particular organic acids to derepress transcription from a BaaR-dependent reporter in vivo, we next evaluated whether adipic acid, ⑀-aminocaproic acid, tetradecanedioic acid, or ⑀-caprolactone modify BaaR binding to its target DNA. We mixed a fluorescent DNA probe corresponding to the BBS2 region with purified BaaR and (separately) 1 mM of each small molecule. We conducted this experiment using a range of BaaR concentrations.
None of the molecules tested had any effect on BaaR binding to the probe (Fig. 9, A-E). However, the possibility that these molecules interact with the BaaR LBD without affecting the interaction of BaaR with DNA could not be ruled out in this assay. Therefore, we also performed isothermal titration calorimetry (ITC) measurements, which allow characterization of the affinities between IclR proteins and small molecules at micromolar levels (41,56,62). Specifically, we performed ITC measurements between purified BaaR LBD and adipic acid, which is likely transported by Bab2_0213 and thus a bona fide direct activating signal. We observed no interaction between adipic acid and the LBD of BaaR, suggesting an indirect effect of adipic acid in vivo (Fig. 9F).

Deleting baaR does not affect B. abortus intracellular entry or replication
As outlined earlier, the general stress response sigma factor E1 indirectly activates transcription of BaaR. Given the known role of E1 in B. abortus infection biology (4, 7), we evaluated whether BaaR contributes to B. abortus infection in vitro. Cultured THP-1 macrophage-like cells were infected (multiplicity of infection ϭ 100) with WT and ⌬baaR B. abortus strains. No significant differences in WT and ⌬baaR CFU recovered from infected THP-1 cells were observed at any time point evaluated (1, 24, and 48 h post-infection), suggesting that BaaR and bab2_0213-bab2_0217 repression are not required for entry or replication in a THP-1 in vitro infection model (Fig. S7). This result is consistent with a previous study in which Brucella melitensis 16M with deletion of a gene orthologous to baaR, BMEII1022, was not attenuated in mouse spleen colonization at 1 week post-infection (63).

Discussion
As a facultative intracellular pathogen, B. abortus is exposed to many different microenvironments during its life cycle. The general stress response sigma factor, E1 , controls the transcription of dozens of genes and is required for B. abortus survival under stress conditions in vitro and chronic infection conditions in a mouse disease model (4,5,7,9,10). In this study, we present structural and functional studies of a transcriptional regulator of the IclR family, BaaR, which is indirectly activated by E1 . Our experiments provide molecular-level understanding of baaR regulatory function in B. abortus but demonstrate that this gene is not a major contributor to oxidative stress survival or mammalian cell colonization.
BaaR strongly represses transcription from two divergently transcribed operons (Figs. 3 and 6) homologous to the dca ␤-oxidation operons of Acinetobacter spp. and Pseudomonas spp., which have been implicated in growth on adipic acid or ⑀-caprolactam as a sole carbon source (Fig. 10) (36 -38). In Acinetobacter and Pseudomonas, it has been postulated that IclR transcriptional regulators control dca operon expression, although this hypothesis has not been tested experimentally. When compared with other IclR proteins, BaaR is closely related to Acinetobacter DcaR (ACIAD1688; 54% identity) and DcaS (ACIAD1684; 59% identity) and Pseudomonas DcaR (PA1630; 66% identity) (Fig. 10). All four of these proteins possess an N-terminal extension compared with other IclR proteins, and very similar LBD cavities (Figs. 2C and Fig. S1), suggesting they recognize structurally related small molecules. The function of the extended N termini in these related proteins is

Transcription regulation by Brucella BaaR
unclear, and deletion of the N terminus of BaaR did not affect DNA binding in vitro (Fig. S4). It is possible that in a cellular context, this structural region affects protein/DNA interactions or protein stability in the presence of specific molecular signals.
In vivo, BaaR represses transcription of the bab2_0213-bab2_0217 locus (including its own gene, bab2_0215). Adipic acid and other related organics relieve this repression, resulting in initiation of bab2_0213-bab2_0217 transcription. It remains unclear how transcription of this operon is maintained despite increasing concentrations of BaaR in the cell. It is conceivable that like the TtgV protein of Pseudomonas putida (64,65) and other proteins belonging to the IclR family (14,15,66,67), reduced affinity of BaaR for DNA upon ligand binding is sufficient to overcome the effects of increased concentrations in the cell.
When compared with the Acinetobacter dca locus, the B. abortus bab2_0213-bab2_0217 dca-like operon (regulated by BaaR) appears incomplete. Indeed, genes essential for the ␤-oxidation of dicarboxylic acids in Acinetobacter are degenerate or absent in B. abortus (Fig. 10). It is therefore difficult to conclude that bab2_0213-bab2_0217 is truly involved in ␤-oxidation metabolism. The RNA-seq studies presented here identified additional genes, present in the bab2_1036 -bab2_1047 locus, that are strongly up-regulated after deletion of baaR. Based on sequence homology, these genes are paralogs and likely have functions that overlap or complement bab2_0214 -bab2_0217 (Figs. 3 and 10). It is conceivable that like Pseudomonas PAO1, multiple clusters of dca-like genes have redundant functions in dissimilation of short-or medium-chain dicarboxylic acids (Fig. 10) (36). No transporter related to bab2_0213 was found proximal to this set of paralogous genes or in B. abortus genome, suggesting that bab2_0213-bab2_0217 and paralogs could process a shared substrate that is transported by Bab2_0213. Bab2_0213 has high sequence identity with MucK-like MFS transporters located proximal to the Acinetobacter (ACIAD1681 (mucK) and ACIAD1694 (dcaK); 67 and 52%) and Pseudomonas (PA1019; 59%) dca operons. In Acinetobacter, MucK is essential for growth in the presence of cis,cis-muconic acid but not adipic acid. A potential role for DcaK in adipic acid transport has been discussed, although it has never been experimentally investigated (Fig. 10) (36,38,39). Our results provide evidence that Bab2_0213 is involved in adipic acid transport, although we cannot rule out the possibility that this protein could also transport other types of molecules. Homology between Bab2_0213 and MFS transporters associated with the dca genes in Acinetobacter or in Pseudomonas clearly suggests that the molecules transported by these systems are structurally related to adipic acid or cis,cis-muconate.
The RNA-seq analysis of WT B. abortus and the ⌬baaR strain also revealed a predicted ABC transport system (bab2_0278 -bab2_0282) that was indirectly activated by BaaR (Fig. 3). The primary structure of the PBP in this system (Bab2_0282) is 54% identical and 72% homologous to a Burkholderia mallei PBP that co-crystallized with bound acetoacetate (PDB code 3I09). Residues involved in this interaction with acetoacetate are also present in Bab2_0282 (Fig. S3), suggesting that Bab2_0282 may transport acetoacetate or a closely related molecule. In mammals, acetyl-CoA formed in the liver during fatty acid ␤-oxidation can be converted into acetoacetate and released into the bloodstream as an energy source during periods of starvation or intense physical activity (54,55). We have no evidence that exogenous acetoacetate is actively transported by B. abortus, although a metabolic connection between this ABC transporter and the bab2_0213-bab2_0217 locus might

Transcription regulation by Brucella BaaR
exist. As discussed earlier, the bab2_0213-bab2_0217 locus is missing a thiolase enzyme required to perform the last step of a ␤-oxidation reaction that leads to the production of an acetyl-CoA and an acyl-CoA molecule (Fig. S8) (52). None of the five putative thiolases (bab1_0486, bab1_1783, bab2_0443, bab2_0606, and bab2_0790) present in the B. abortus genome have altered expression in the ⌬baaR background (Table S2), suggesting that theses genes are not involved in performing an ultimate thiolysis step. Instead, as a last step, we propose that the CoA-transferase family III enzyme (Bab2_0217) present in the bab2_0213-bab2_0217 locus may catalyze a reversible transfer of CoA from CoA-thioesters to free acids (Fig. S8) (53,68), leading to the formation of ketones. Such a reaction has been previously described in an engineered E. coli strain producing isopropanol (69) and in acetone production in Clostridium acetobutylicum (70). In these strains, a CoA-transferase was involved in the conversion of acetoacetyl-CoA into acetoacetate, using butyrate or acetate as CoA acceptors (69,70). Such a reaction in B. abortus may explain why the bab2_0278 -bab2_0282 ABC transport system is down-regulated when the bab2_0213-bab2_0217 locus is overexpressed. Cytoplasmic accumulation of ketones, such as acetoacetate, may repress expression of the bab2_0278 -bab2_0282, which would be no longer needed for ketone transport.
It remains unclear what role adipic acid would play in this hypothetical metabolic process. As shown in Fig. 9, adipic acid does not directly interact with BaaR, suggesting that a degradation product is instead the real activating signal. However, deletion of the bab2_0213-bab2_0217 enzymes did not affect the ability of adipic acid to induce the transcription of our reporter gene (Fig. 8). Assuming that these enzymes are indeed functional, our data provide evidence that Bab2_0214, Bab2_0216, and Bab2_0217 are not involved in its synthesis. In a defined medium, adipic acid does not support growth of WT or ⌬baaR strains (Fig. S6). In a previous study (36), it was shown that Acinetobacter can use adipic acid as a sole carbon source and that this required the presence of the acyl-CoA dehydrogenase DcaA, the enoyl-CoA hydratase DcaE, and the 3-hydroxyacyl-CoA dehydrogenase DcaH (which respectively share 78, 55, and 52% identity with Bab2_0214 and Bab2_0216 (Fig. 10)). The presence of a thiolase (DcaF) in this dca operon might explain the differences in growth between B. abortus and Acinetobacter in presence of adipic acid (Fig. 10) (36).
In conclusion, we have demonstrated that adipic acid and the related molecules ⑀-aminocaproic acid, tetradecanedioic acid, and ⑀-caprolactone activate transcription of the bab2_0213-bab2_0217 locus in B. abortus, although these molecules do not activate transcription through a direct interaction with BaaR. This suggests that the actual activating molecule in vivo is either a metabolic product of these compounds or that these inducing compounds activate some undefined metabolic pathway involved in the synthesis of the BaaR-activating signal. The possible relevance of such molecules in the life cycle of B. abortus and the functional significance of baaR transcriptional activation by E1 remain undefined. Adipic acid is produced by oxidation of fatty acids but is not naturally abundant. We note that very little is known about the B. abortus life cycle outside the host. In the wild, B. abortus can persist for weeks in aborted fetuses, a major source of contagion (71). However, B. abortus can also persist for weeks in soil or on vegetation. How B. abortus survives in these harsh and competitive environments is unknown, and it is possible that the genes investigated in this study enable B. abortus to metabolize unusual substrates found outside the mammalian host.
To conclude, we propose a model (Fig. 11) whereby dimeric BaaR constitutively interacts with the DNA region between baaR and bab2_0216, repressing divergent transcription on both strands. When present in the environment, adipic acid is likely transported by the MFS MucK transporter Bab2_0213, resulting in derepression of transcription from the BaaR-inhibited promoters. The uptake of structurally related molecules, including ⑀-aminocaproic acid, ⑀-caprolactone, and tetradecanedioic acid, also induces transcription, although our data provide evidence that these molecules are transported by a genetically distinct system. Once in the cytoplasm, these molecules are predicted to interact with an unknown metabolic/ regulatory process that leads to production of an intracellular ligand that binds to and regulates BaaR. Interaction with a ligand likely induces structural changes in BaaR that result in dissociation from DNA. The subsequent derepression of the bab2_0213-bab2_0217 locus may increase adipic acid uptake, creating a positive feedback loop. Derepressed expression of the bab2_0213-bab2_0217 locus also indirectly enhances expres- Figure 11. Model of BaaR-dependent gene regulation. The IclR-family protein BaaR (Bab2_0215; in red) constitutively interacts with BBS1 and BBS2, repressing transcription of the bab2_0213-bab2_0216 operon (yellow, green, brown, blue/red, and orange arrows). When present, adipic acid (yellow circles) is specifically transported by the Bab2_0213 transporter (green pipe). The uptake of other molecules, such as ⑀-aminocaproic acid, ⑀-caprolactone, or tetradecanedioic acid, may involve an unknown transporter (green ? box). Once in the cytoplasm, these molecules may activate an unknown metabolic process (orange ? box), leading to the synthesis of an intracellular molecular signal (blue circles) specifically recognized by BaaR. This is proposed to induce structural changes in the BaaR protein and dissociation from DNA. When transcription of the bab2_0213-bab2_0217 operon is derepressed, adipic acid uptake is increased, and derepression of transcription from the operon is enhanced. Derepression apparently modulates expression of bab2_1036 -bab2_1047, bab2_0277-bab2_282, and bab2_0213-bab2_0217 indirectly (gray dashed lines).

Transcription regulation by Brucella BaaR
sion of a paralogous gene set, bab2_1036 -bab2_1047, while attenuating expression of a potential ketone ABC transport system (bab2_0278 -bab2_282).

Materials and methods
All experiments using live B. abortus 2308 were performed in Biosafety Level 3 facilities according to United States Centers for Disease Control (CDC) select agent regulations at the University of Chicago Howard Taylor Ricketts Laboratory.

Chromosomal deletions in B. abortus
The different B. abortus deletion strains (⌬bab2_0213, ⌬bab2_0214, ⌬baaR (i.e. bab2_0215), ⌬bab2_0216, and ⌬bab2_ 0217) were constructed using a double-recombination strategy. Briefly, after PCR amplification using KOD Xtreme Hot start DNA polymerase (EMD Millipore) with B. abortus chromosomal DNA as a template, the corresponding PCR products were purified using the GeneJET PCR purification kit (Thermo Fisher Scientific). The in-frame deletion alleles (carrying 5Јand 3Ј-flanking sequences of B. abortus locus tags) were digested with restriction enzymes (New England Biolabs) before ligation using T4 DNA ligase (New England Biolabs) or were directly Gibson assembled (New England Biolabs) into the suicide plasmid pNPTS138, 4 which carries the nptI gene for initial selection and the sacB gene for counter-selection on sucrose. Plasmids were then transformed in E. coli Top10 strain, and single colonies carrying the plasmid with the correct insert were screened by PCR and Sanger sequenced. Sequenced plasmids were then purified from E. coli using the GeneJET plasmid miniprep kit (Thermo Fisher Scientific). The WT B. abortus 2308 strain was transformed with pNPTS138 carrying the different deletion alleles by electroporation, and single crossover integrants were selected on Schaedler Blood Agar (SBA) plates supplemented with kanamycin (50 g/ml). Counter-selection for the second crossover event was carried out by growing kanamycin-resistant colonies overnight under nonselective conditions and then plating on SBA plates supplemented with 5% sucrose (150 mM). PCR screening to identify colonies containing the deletion alleles was performed, and the products were Sanger sequenced for verification. If necessary, genetic complementation of the deletion strain was carried out by transforming the B. abortus deletion strains with the pNPTS138 plasmid carrying the WT allele locus. The primers, restriction enzymes, plasmids, and strains used are listed in Tables S4 and S5.

RNA extraction and sequencing
Three independent cultures of the B. abortus WT and ⌬baaR strains (see Table S5 for strain information) were grown overnight in 5 ml of Brucella broth (BB) (BD Biosciences) at 37°C with shaking at 220 rpm. The next morning, the OD 600 of the cultures was adjusted to 0.1 in 3 ml if BB and grown for another 5 h at 37°C with shaking at 220 rpm. Cultures (final OD 600 ϳ0.3-0.4) were then pelleted, and the cells were immediately frozen at Ϫ80°C until RNA extraction. A hot-phenol RNA extraction protocol was performed for each pellet. Cell pellets were resuspended in 400 l of commercial phosphate-buffered saline (PBS) buffer and transferred to tubes containing 200 l of SDS lysis buffer (20% SDS w/v, 0.5 M EDTA) preheated at 95°C. The tubes were gently inverted and incubated at 95°C for 5 min. After incubation, the contents of each tube were transferred to new tubes containing 600 l of acid phenol/chloroform preheated at 65°C, vortexed for 5 s, and incubated for 10 min at 65°C. Samples were then centrifuged for 10 min at 2500 ϫ g. After centrifugation, the top aqueous phase was transferred to 600 l of acid phenol/chloroform at room temperature, vortexed, and centrifuged for 10 min at 2500 ϫ g. The top aqueous phase was transferred to 500 l of chloroform/ isoamyl alcohol (24:1) at room temperature, mixed, and centrifuged for 10 min at 2500 ϫ g. The top aqueous phase was then transferred to a new tube containing 500 l of 100% isopropyl alcohol. Precipitation of the RNA was performed overnight at Ϫ80°C. RNA was then pelleted at maximum speed for 30 min at 4°C, and the pellets were washed twice with 70% ethanol, air-dried, and resuspended in 80 l of RNase-free water. To remove any co-purified genomic DNA, 10 l of 10ϫ buffer and 10 l of Turbo DNase (Thermo Fisher Scientific) were added to each sample for 2 h at 37°C. The samples were then purified using purification columns (from the RNeasy MinElute Cleanup kit; Qiagen) and eluted with 30 l of DNase (RNase)free water. The eluate was further digested on the column by adding 7 l of Turbo DNase, 7 l of 10ϫ buffer, and 56 l of RNase-free water. The reaction was incubated at room temperature for 30 min and purified using a Qiagen RNA purification kit according to the manufacturer's protocol.
For RNA-seq analysis, rRNA was depleted from WT and ⌬baaR samples using Ribo-Zero rRNA Removal (Gram-negative bacteria) kit (Epicenter). Libraries were prepared with Illumina TruSeq RNA kit according to manufacturer's instructions and were then quantified using a 2100 Bioanalyzer (Agilent) and sequenced on a HiSeq2500 (Illumina). The obtained RNAseq reads were aligned to the genome sequence of B. abortus 2308 (RefSeq AM040265) using the readmapper tool in CLC Genomics Workbench (Qiagen) (mismatch cost ϭ 2; insertion cost ϭ 3; deletion cost ϭ 3; length fraction ϭ 0.8; and similarity fraction ϭ 0.8). Differential expression analysis of normalized data and false-discovery rate (FDR) p values were calculated in CLC Genomic Workbench. RNA-seq data sets are available at the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) at accession number GSE107825.

lacZ transcriptional reporter construction
The different lacZ reporter genes used in this study were built as follows: the WT DNA fragments corresponding to the different portions of the bab2_0215 (baaR)-bab2_0216 intergenic region were PCR-amplified using B. abortus chromosomal DNA as a template. A gBlocks-synthesized DNA fragment (Integrated DNA Technology, IDT) corresponding to the baaR-bab2_0216 intergenic region carrying mutated BBSs was used as a PCR template to introduce mutations in BBS1 and BBS2. To mutate BBS1 and BBS2, the corresponding palindromic sequences were randomized to maintain the length and base content equivalent to the native sequences. DNA frag-4 M. R. K. Alley, unpublished data.

Transcription regulation by Brucella BaaR
ments corresponding to baaR and bab2_0216 promoter regions carrying a WT and a mutated BBS were generated by overlapping PCRs. Mutation of BBS2 region 1, region 2, or region 3 were introduced using overlapping primers carrying specific mutations. Primer sequences and information are available in Table S4. As described previously, PCRs were performed using with KOD Xtreme Hot start DNA polymerase (EMD Millipore), gel-purified with GeneJET PCR purification kit (Thermo Fisher Scientific), and digested with restriction enzymes (New England Biolabs). Ligation with the linearized pMR15 plasmid was performed using T4 DNA ligase (New England Biolabs). After ligation and transformation into an E. coli Top10 strain, single colonies carrying plasmids with the different inserts were PCR-screened and sent for sequencing. The plasmids were then purified with the GeneJET plasmid miniprep kit (Thermo Fisher Scientific) and transformed in WT B. ovis or in the different B. abortus genetic backgrounds (WT, ⌬bab2_0213, ⌬bab2_0214, ⌬baaR, ⌬bab2_0216, and ⌬bab2_0217) used in this study. Transformants were then selected on SBA plates supplemented with kanamycin (50 g/ml). Primer, restriction enzyme, plasmid, and strain information are available in Tables S4 and S5.

Screening for ligands that derepress BaaR in vivo
A WT B. ovis strain carrying the pMR15-P baaR -lacZ reporter plasmid (see Table S5 for strain information) was harvested from a fresh SBA plate supplemented with kanamycin (50 g/ml) and resuspended at OD 600 ϭ 0.1 in 50 ml of BB containing 50 g/ml of kanamycin. 100 l of this B. ovis suspension was then used to inoculate each well of the phenotype MicroArray plates (Biolog). Biolog plates PM1, PM2A, PM3B, PM4A, and PM5 were used for this assay. After growth overnight at 37°C in the presence of 5% CO 2 , the plates were removed from the incubator, and the OD 600 and ␤-gal activities were measured under each growth condition using the Tecan Spark 20 M plate reader.

Additional small molecule screening in Brucella
Depending on their solubility characteristics, all ligands used in this study were freshly prepared in DMSO or sterile water. When needed, the pH was adjusted to 7.4. Caproic acid, ⑀-caprolactam, ⑀-caprolactone, ⑀-aminocaproic acid, glutaric acid, adipic acid, pimelic acid, suberic acid, sebacic acid, and tetradecanedioic acid were purchased from Sigma. The cis,cis-and trans,trans-muconic acids were purchased from Acros-Organics. Succinic acid was purchased from Thermo Fisher Scientific. Malonic acid was purchased from MP-Biomedicals.
For ␤-gal transcriptional activity measurements, B. abortus liquid cultures were prepared as follows: B. abortus strains carrying the different lacZ transcriptional fusions were harvested from fresh SBA plates supplemented with kanamycin (50 g/ml) (strain information is available in Table S5). Cells were resuspended in 1 ml of BB, and the corresponding OD 600 was measured by spectrophotometry (Thermo Fisher Scientific Genesys 20). These B. abortus suspensions were used to inoculate (at OD 600 ϭ 0.1) culture tubes containing 2 ml of BB supplemented with kanamycin (50 g/ml). These culture tubes also contained different concentrations of small molecules (from 50 M to 4 mM). After overnight growth at 37°C and 220 rpm shaking, ␤-gal activity was measured. Each condition was independently tested at least three times using different clones for each time.

Measurement of ␤-gal activity
In this study, all ␤-gal activity measurements were performed in 96-well plates, and the absorbance was measured using the Tecan Spark 20 M or Infinite 200 PRO plate reader.
To assess regulation of reporter gene transcription by BaaR, the different B. abortus strains were grown on SBA plates supplemented with 50 g/ml kanamycin. After incubation for 2-3 days at 37°C and 5% CO 2 , the cells were harvested and resuspended in 1 ml of BB; 200 l of each culture tube were transferred to a clear Corning flat-bottom 96-well plate, and the OD 600 was measured using the Tecan plate reader. The ␤-gal activities of four different clones for each strain were independently measured at least twice. A representative data set is presented in Fig. 6.
For ligand screening on 96-well plates, Phenotype Micro-Array (Biolog) plates were prepared as described earlier, and the OD 600 was measured using the Tecan plate reader. This initial screen was conducted once.
For ligand screening in culture tubes, 200 l each culture tube were transferred to a clear flat-bottom 96-well plate, and the OD 600 was measured using the Tecan plate reader. In Figs. 7 and 8, each condition was independently tested at least three times using different clones for each time.
For ␤-gal activity measurements, between 2.5 and 10 l of cell suspension were mixed and lysed with 25 l of chloroform in a 96-well chloroform-resistant plate, and 125 l of Z-buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, adjusted to pH 7) were then added to each well. After addition of 42 l of a 4 mg/ml O-nitrophenyl-␤-D-galactopyranoside solution to each well, the reaction was developed at room temperature and stopped by adding 83 l of a 1 M sodium carbonate solution. For each reaction, the incubation time was recorded; 200 l of each reaction were then transferred to a clear flat-bottom 96-well plate, and the OD 420 was measured using the Tecan plate reader. Calculation of ␤-gal activity was performed using Equation 1, where A 420 is the absorbance measured at 420 nm of the ␤-gal reaction; blankA is a blank reaction containing no cells; 0.25 is the total volume of the reaction (in milliliters); A 600 is the absorbance measured at 600 nm of the cell suspension; blankB is a blank containing Brucella broth alone; t is the incubation time (min); and v is the volume of cells used in the reaction (in milliliters).

Growth assays in Gerhardt's minimal medium (GMM) in presence or absence of adipic acid
B. abortus WT and ⌬baaR strains were grown in GMM, pH 6.8 (NaCl 7.5 g/liter, K 2 HPO 4 10 g/liter, sodium thiosulfate 0.1 g/liter, glycerol 30 g/liter, lactate 5 g/liter, L-glutamic acid 1.5  (72). Four different GMM solutions were tested: with all three carbon sources present (glycerol, lactate, and glutamate), without glycerol, without lactate, and without glutamate. The same conditions were also tested but in presence of 1 mM adipic acid. In plastic culture tubes, 2 ml of each GMM solution were inoculated with the WT and ⌬baaR strains at a starting OD 600 of 0.05. Cultures were grown for 3 days at 37°C/220 rpm, and optical densities of the cultures were assessed every 24 h by transferring 200 l of each culture in 96-well plate and using a Tecan plate reader for measurements. Growth curves were performed in triplicate with two different strains.

Transcription regulation by Brucella BaaR
Oxidative stress assay B. abortus WT, ⌬baaR, and ⌬rpoE1 strains (strain information is available in Table S5) grown on SBA plates for 48 h were harvested and resuspended in GMM, pH 6.8 (72). Each sample was then adjusted to a final cell density of 1 ϫ 10 8 CFU/ml in 2 ml of GMM, pH 6.8. The test group was subjected to oxidative stress by the addition of 5 mM H 2 O 2 (final concentration); the control group was mock-treated with sterile water. After 1 h of incubation in a shaking incubator at 37°C, the cultures were serially diluted in PBS and plated on Tryptic Soy Agar (TSA) plates for viable CFU counting. Three independent cultures per condition were prepared for each strain.

Cell culture and macrophage infection assay
Human monocytic THP-1 cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% heat-inactivated fetal bovine serum and 2 mM L-glutamine. Differentiation of the cells into inactivated macrophages was induced by the addition of 40 ng/ml phorbol 12myristate 13-acetate (Sigma) for 48 h at 37°C in a 5% CO 2 atmosphere. Prior to infection, bacteria were harvested from freshly plated SBA plates and resuspended in sterile RPMI 1640 medium, and cell densities were adjusted. For infection assays, 5 ϫ 10 4 THP-1 cells were infected with 5 ϫ 10 6 cells of the WT or ⌬baaR B. abortus strains to achieve a multiplicity of infection of 1:100 in 96-well plates. To synchronize the infections after the addition of the Brucella cells, the plates were centrifuged at 200 ϫ g for 5 min. After 1 h of incubation at 37°C in a 5% CO 2 atmosphere, the medium was removed and replaced with RPMI 1640 medium supplemented with gentamicin (50 g/ml) and incubated for 20 min at 37°C in a 5% CO 2 atmosphere to kill extracellular bacteria. To determine the numbers of intracellular bacteria at 1, 24, and 48 h post-infection, the cells were washed once with PBS and lysed with 0.1 ml of PBS complemented with 0.1% Triton X-100. The lysate was serially diluted and plated on tryptic soy agar plates for CFU counting. This infection experiment was performed in triplicate using two different clones of each strain.

Protein expression plasmids
DNA fragments corresponding to the full-length BaaR (residues 1-284) protein, the BaaR protein deleted for the first 8 (9 -283), 20 , or 39 (40 -284) N-terminal amino acids, and the BaaR LBD(115-284) were amplified by PCR using KOD Xtreme Hot start DNA polymerase (EMD Millipore). The primers used for cloning are listed in Table S4. After purification using the GeneJET PCR purification kit (Thermo Fisher Scientific), the PCR products were directly Gibson assembled (New England Biolabs) or digested with restriction enzymes (New England Biolabs) and ligated using T4 DNA ligase (New England Biolabs) into a linearized pET28a plasmid. Fragments corresponding to BaaR (residues 9 -283) and BaaR (residues 21-284) were cloned, respectively, into pMCSG81 and pMCSG73 plasmids using a ligation-independent procedure (73,74). E. coli Top10 strains were then transformed with the different plasmids by electroporation, and transformants were selected on Luria broth (LB, Thermo Fisher Scientific) agar plates supplemented with 50 g/ml kanamycin. After PCR screening and sequencing, the corresponding plasmids were purified using the GeneJET plasmid miniprep kit (Thermo Fisher Scientific) and transformed into E. coli BL21-Gold(DE3) or Rosetta(DE3)(pLysS) strains (Stratagene) for protein expression (see Table S5 for strain information).

Protein expression and purification
An overnight LB (Thermo Fisher Scientific) pre-culture (100 ml) was used to inoculate 1 liter of LB supplemented with the appropriate antibiotics. Overexpression of the different Histagged BaaR proteins was induced at an OD 600 of ϳ0.8 (37°C, 220 rpm) by adding 1 mM isopropyl ␤-D-1-thiogalactopyranoside (GoldBio). After 5 h of induction, cells were harvested by centrifugation at 11,000 ϫ g for 20 min at 4°C. The cell pellets were resuspended in 20 ml of a buffer containing 10 mM Tris-HCl, pH 8.8, 150 mM NaCl, and 10 mM imidazole and supplemented with 50 l of a DNase I solution at 5 mg/ml and onehalf tablet of complete protease inhibitor mixture (Roche Applied Science). The cells were then disrupted by one passage in a microfluidizer (Microfluidics LV1). The resulting cell lysate was clarified by centrifugation at 39,000 ϫ g for 20 min at 4°C. Purification of the His-tagged proteins was performed by nickel affinity chromatography (nitrilotriacetic acid resin; GE Healthcare). After binding of the clarified lysate to the column, three washing steps were performed using 10, 30, and 75 mM imidazole Tris-NaCl buffers, followed by elution with 200 and 500 mM imidazole Tris-NaCl buffers. All purification steps were carried out at 4°C. Protein purity was assessed by running the eluate on a 14% SDS-polyacrylamide gel and staining the gel with Coomassie Blue. Proteins were then dialyzed overnight at 4°C against 2 liters of Tris-NaCl buffer (10 mM Tris, pH 8.8, 150 mM NaCl, 1.5 mM EDTA). The protein concentrations were estimated using a colorimetric Bradford protein assay method kit (Thermo Fisher Scientific). Protein samples were concentrated using a centrifugal filter (-kDa molecular mass cutoff, Amicon-Millipore). If necessary, protein aliquots were flash-frozen in liquid nitrogen after addition of glycerol to a final concentration of 50%.

Electrophoretic mobility shift assay (EMSA)
For gel-shift assays, fluorescent DNA probes corresponding to different portions of the WT bab2_0215-bab2_0216 inter-

Transcription regulation by Brucella BaaR
genic region were PCR-amplified using B. abortus chromosomal DNA as a template. Upstream primers positioned before BBS1 or BBS2 and a downstream fluorescent primer positioned at the beginning of the baaR gene were used to generate the "long" (375 or 289 nucleotides) and "short" (196 nucleotides) fluorescent DNA probes. The fluorescent primer (Integrated DNA Technology, IDT) was labeled with the Alexa Fluor 488 dye with an excitation wavelength of 492 nm and an emission wavelength of 517 nm. Using the same set of primers, a gBlocks DNA fragment (Integrated DNA Technology, IDT) corresponding to the bab2_0215-bab2_0216 intergenic region carrying mutated BBSs was also used as a PCR template to introduce mutations in BBS1 and BBS2. To maintain the length and base content of the mutated regions, the nucleotide sequences of regions 1-3 of BBS1 and the nucleotide sequences of regions 1-3 of BBS2 were randomized. The resulting PCR products corresponded to the long (375 nucleotides) or short (196 nucleotides) mutant fluorescent DNA probes. Mutations in BBS2 regions 1-3 were introduced using three different sets of overlapping primers containing BBS2 regions 1-3 mutations, respectively. The different fluorescent DNA probes were also used as templates to generate nonfluorescent DNA probes (short/long/WT/mutated) using a nonfluorescent downstream primer during PCR amplification. Primer sequences and information regarding the gBlocks gene fragment are available in Table S4.
The PCR products were run on a 1% agarose gel and gelpurified using the GeneJET PCR purification kit (Thermo Fisher Scientific). DNA concentrations were measured using the Nanodrop One (Thermo Fisher Scientific).
All EMSAs were performed using a BaaR protein corresponding to residues 9 -283. Two additional constructs (a full-length (residues 1-284) and a shorter (residues 40 -284) BaaR protein) were also used to evaluate the importance of the BaaR N-terminal region. All gel-shift assays were performed using previously published protocols (75,76). Briefly, 10-l reactions were prepared as follows: each reaction consisted of a specific concentration of BaaR protein mixed with 5 ng of a fluorescent DNA probe and 1 l of a 10ϫ binding buffer (100 mM Tris, pH 8.8, 10 mM EDTA, 10 mg/ml BSA, 20 mM CaCl 2 , 500 mM KCl, 1 mM DTT, 50% glycerol, 20 mM MgCl 2 ). To control for interaction specificity, fluorescent or nonfluorescent DNA probes (short/long/ WT/mutated) were also added to the mix at defined concentrations. The effect of specific small molecules was evaluated by adding defined concentrations of each molecule to the mix. Protein dialysis buffer was used to bring the total volume to 10 l. Samples were incubated in the dark at room temperature for at least 30 min and then loaded on a fresh 5% native acrylamide gel pre-run for 20 min in 1ϫ Tris acetate EDTA buffer (40 mM Tris, 20 mM acetic acid, and 1 mM EDTA). After running the gel for 1 h at 4°C/110 V in the dark, the gels were imaged using the Bio-Rad Chemidoc MP imaging system with a 3-min exposure and the manufacturer's parameters for Alexa Fluor 488 detection. Each specific condition was tested at least twice.

ITC ligand-binding assay
All samples (proteins and ligands) were degassed for 10 min prior to ITC measurements, and final dilutions were made using dialysis buffer (10 mM Tris, pH 8.8, 150 mM NaCl, and 1.5 mM EDTA). The ligands were injected into a 200-l sample cell containing 50 M purified BaaR LBD (residues 115-284) (protein expression strain information is available in Table S5). A 1 M adipic acid solution was prepared using the dialysis buffer and adjusted to pH 8.8. This same solution was then diluted to 10 mM using the same dialysis buffer. The ligand solution (1 l) was injected into the cell every 2 min, with 20 injections total performed. Measurements were performed twice at 25°C using an iTC200 micro-calorimeter (MicroCal, GE Healthcare).

Protein expression and purification for crystallization
For crystallization, the E. coli strain carrying the pMCSG73 vector was used to overexpress BaaR (residues 21-284) (see strain information listed in Table S5). The pMCSG73 is a bacterial expression vector harboring a tobacco vein mottling virus-cleavable N-terminal NusA tag and a TEV-cleavable N-terminal His 6 and StrepII tag (74). A 2-liter culture of enriched M9 medium was grown at 37°C with shaking at 190 rpm. At OD 600 ϳ1, the culture was cooled to 4°C and supplemented with 90 mg of L-seleno-L-methionine (Se-Met, Sigma) and 25 mg of each methionine biosynthetic inhibitory amino acid (L-valine, L-isoleucine, L-leucine, L-lysine, L-threonine, and L-phenylalanine). Protein expression was induced overnight at 18°C using 0.5 mM isopropyl ␤-D-1-thiogalactopyranoside. After centrifugation, cell pellets were resuspended in 35 ml of lysis buffer (500 mM NaCl, 5% (v/v) glycerol, 50 mM HEPES, pH 8.0, 20 mM imidazole, and 10 mM ␤-mercaptoethanol) per liter of culture and treated with lysozyme (1 mg/ml) and 3 ml of E. coli cells expressing the tobacco vein mottling virus protease. The cell suspension was sonicated, and debris was removed by centrifugation. The Se-Met protein was purified via Ni 2ϩ -affinity chromatography using the AKTAxpress system (GE Healthcare). The column was washed with 20 mM imidazole (lysis buffer) and eluted in the same buffer containing 250 mM imidazole. Immediately after purification, the His tag was cleaved at 4°C for 24 -48 h using a recombinant His-tagged TEV protease, resulting in an untagged protein with an N-terminal Ser-Asn-Ala peptide. A second Ni 2ϩ -affinity chromatography purification was performed to remove the protease, noncleaved protein, and affinity tag. The purified protein was then dialyzed against 20 mM HEPES, pH 8.0, 250 mM NaCl, and 2 mM DTT buffer. Protein concentrations were determined by UV absorption spectroscopy (280 nm) using the NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific). The purified Se-Met BaaR protein was concentrated to 44 mg/ml using a centrifugal filter (10-kDa MWCO, Amicon-Millipore).

Crystallization
Initial crystallization screening was carried out using the sitting-drop, vapor-diffusion technique in 96-well CrystalQuick plates (Greiner Bio-one). Trays were prepared using a Mosquito robot (TTP LabTech) and commercial crystallization kits (MCSG-1-4, Anatrace). The drops were prepared by mixing equal volumes (0.4 l) of the purified protein (44 mg/ml), and Transcription regulation by Brucella BaaR the crystallization solution was equilibrated against 135 l of the same crystallization solution. After 1 week at 16°C, we obtained monoclinic crystals in condition no. 95 of the MSCG-4 screen corresponding to 200 mM calcium acetate, 100 mM HEPES, pH 7.5, and 10% (w/v) PEG 8000. Prior to flashfreezing in liquid nitrogen, crystals were washed for a few seconds in the crystallization solution containing up to 12% ethylene glycol for cryoprotection.

Crystallographic data collection and data processing
Se-Met crystal diffraction was measured at a temperature of 100 K using a 2-s exposure/degree of oscillation. Crystals diffracted to a resolution of 1.95 Å and the corresponding diffraction images were collected on the ADSC Q315r detector with an X-ray wavelength near the selenium edge of 12.66 keV (0.97927 Å) for SAD phasing at the 19-ID beamline (SBC, Advanced Photon Source, Argonne, IL). Diffraction data were processed using the HKL 3000 suite (77). The scaled amplitudes revealed that the crystal belonged to the P2 1 space group with the following cell dimensions: a ϭ 74.15 Å, b ϭ 112.46 Å, c ϭ 83.65 Å, and ␤ ϭ 115.86°(see Table S1).
The structure was determined by SAD phasing using SHELX C/D/E, mlphare, and dm, and initial automatic protein model building with Buccaneer software, all implemented in the HKL3000 software package (77). The initial model was manually adjusted using COOT (78) and iteratively refined using COOT, PHENIX (79), and/or REFMAC (80); 5% of the total reflections was kept out of the refinement in both REFMAC and PHENIX. The final structure converged to an R work of 17.7% and R free of 21.2% and includes four protein chains (A, residues 21-284; B, 21-283; C, 21-284; and D, 20 -284) forming two dimers, one ethylene glycol molecule, seven acetate molecules, one calcium ion, and 315 ordered water molecules. The BaaR protein contained three N-terminal residues (Ser-Asn-Ala) that remain from the cleaved tag and were not visible in the structure. The stereochemistry of the structure was checked using PROCHECK (81), and the Ramachandran plot and was validated using the PDB validation server. Coordinates of BaaR have been deposited in the PDB (PDB code 5WHM). Crystallographic data and refined model statistics are presented in Table S1. Diffraction images have been uploaded to the SBGrid data server (Data DOI: 10.15785/SBGRID/491).
Amino acid sequences were aligned using the M-COFFEE Multiple Sequence Alignment Server (82) and shaded using BoxShade. Figures of the structures, structural alignments, and r.m.s.d. calculations were performed using PyMOL (PyMOL Molecular Graphics System, version 1.7.4; Schrödinger, LLC). The XtalPred server (83) and Dali server (84) were used to identify proteins with the highest structural and sequence homo-logies. A structural model of Bab2_0282 based on the B. mallei PBP structure (PDB code 3I09) was generated using the ExPASy SWISS-Model server (85).