A dual role of the transcriptional regulator TstR provides insights into cyanide detoxification in Lactobacillus brevis

Summary

In this study we uncover two genes in Lactobacillus brevis ATCC367, tstT and tstR, encoding for a rhodanese and a transcriptional regulator involved in cyanide detoxification. TstT (LVIS_0852) belongs to a new class of thiosulfate:cyanide sulfurtransferases. We found that TstR (LVIS_0853) modulates both the expression and the activity of the downstream-encoded tstT. The TstR binding site was identified at −1 to +33, from tstR transcriptional start site. EMSA revealed that sulfite, a product of the reaction catalyzed by TstT, improved the interaction between TstR:P_tstR, while Fe(III) disrupted this interaction. Site-directed mutagenesis in TstR identified M64 as a key residue in sulfite recognition, while residues H136-H139-C167-M171 formed a pocket for ferric iron coordination. In addition to its role as a transcriptional repressor, TstR is also involved in regulating the thiosulfate:cyanide sulfurtransferase activity of TstT. A 3-fold increase in TstT activity was observed in the presence of TstR, which was enhanced by the addition of Fe(III). Overexpression of the tstRT operon was found to increase the cyanide tolerance of L. brevis and Escherichia coli. The protein-protein interaction between TstR and TstT described herein represents a novel mechanism for regulation of enzymatic activity by a transcriptional regulator.

Introduction

Lactobacilli are widely used in the food industry for fruit and vegetable fermentation, and the production of yogurt and cheese (Lorca et al., 2007a). These microorganisms are members of the human gastrointestinal tract and their use as probiotics has increased over the last several years (Lebeer et al., 2008). As part of the molecular characterization of probiotic traits, it is of great interest to elucidate the regulatory mechanisms by which beneficial microbes, such as lactobacilli, tolerate a wide variety of stress conditions (for a review see Lorca and Font de Valdez, 2009). More specifically, we have used Lactobacillus brevis as a model to understand how commensal bacteria regulate the transcription of certain genes in response to environmental stressors, such as antibiotics (Pagliai et al., 2010) or flavonoids (Pande et al., 2011). L. brevis has been isolated from various niches, including beer and the intestinal tracts of mammals (Mikelsaar et al., 1998); however, it is most commonly found in decaying plant matter and fermented fruit (Kandler and Weiss, 1986). Such plants and their derivative niches may contain cyanogenic glycosides, which upon hydrolysis generate the toxic cyanide (Lei et al., 1999; Saidu, 2004) that L. brevis may detoxify in order to persist. Cyanide toxicity has been linked to the chelation of di- and trivalent metals in several key metalloenzymes, resulting in the unavailability of essential metal cofactors and the inhibition of cell growth (Luque-Almagro et al., 2007, Cipollone et al., 2008). In some cyanogenic Pseudomonas, exposure to cyanide and its metabolic products has been linked to the induction of an oxidative stress response (Fernandez and Kunz, 2005; Luque-Almagro et al., 2007).

The transformation of cyanide to a less toxic compound by thiosulfate:cyanide sulfurtransferases or rhodaneses (E.C. 2.8.1.1) is one of the main pathways for cyanide detoxification in living organisms (Vennesland et al., 1982; Saidu, 2004; Cipollone et al., 2007a). However, limited information is available concerning the mechanism of gene regulation for this group of enzymes in bacteria. In Pseudomonas aeruginosa PAO1, the rhdA gene (encoding a rhodanese) is constitutively expressed throughout its entire growth cycle, thus increasing these organisms’ biological fitness (Cipollone et al., 2007b). In contrast, the glpE rhodanese in E. coli is encoded within an operon regulated by multiple transcription factors such as the DeoR-homolog, GlpR, as well as catabolic repressor CRP and the respiratory regulators FNR and ArcA/ArcB (Iuchi and Lin, 1993; Zeng et al., 1996). Similarly, PspE is induced by the presence of glycerol in the growth media (Cheng et al., 2008). In L. brevis, we found a rhodanese-related sulfurtransferase (LVIS_0852) that seems to form a transcriptional unit with a MarR-homolog transcriptional regulator.

Members of the MarR family of transcriptional regulators are small proteins capable of modulating a variety of processes, such as the expression of virulence determinants and the catabolism of aromatic compounds (for a review see Perera and Grove, 2010). Moreover, a large proportion of MarR homologs has been extensively characterized and was found to mediate multidrug resistance and redox-associated stress responses. While the majority of MarR homologs sense small molecules, some members such as OhrR, YodB, MgrA, and MexR are examples of MarR homologs that modulate the bacterial response to oxidative stress (Hong et al., 2005; Chen et al., 2009; Chen et al., 2010; Chi et al., 2010; Antelmann and Helmann, 2011). As a general mechanism for the latter group of proteins, the thiol group of cysteine residue(s) suffers reversible or irreversible modifications, producing an active or inactive transcriptional regulator (Antelmann and Helmann, 2011).

In this report, we determined that LVIS_0852 (TstT) and LVIS_0853 (TstR) are involved in cyanide detoxification in L. brevis. We elucidated the mechanism of gene regulation for LVIS_0853 by small molecules and further characterized its downstream rhodanese-related sulfurtransferase, whose activity is increased through a novel interaction with the TstR transcriptional regulator.

Results

LVIS_0852 encodes for a member of a new subfamily of single-domain thiosulfate:cyanide sulfurtransferases

A systematic annotation of the L. brevis ATCC 367 genome classified LVIS_0852 as a rhodanese-related sulfurtransferase. In silico analysis confirmed that the encoded polypeptide belongs to the Rhodanese superfamily (RHOD). Within the RHOD superfamily, several modules encoding active site motifs have been identified. Those modules were previously used to cluster the proteins into different subfamilies and to establish a direct correlation between these structural features and their biological role (Bordo and Bork, 2002; Cipollone et al. 2008). To determine in which subfamily LVIS_0852 clusters, a multiple protein alignment was conducted using 235 rhodanese sequences including the subfamilies previously described by Bordo and Bork (2002). The phylogenetic tree generated revealed that LVIS_0852 and its homologs were not included in any of the subfamilies previously reported and may constitute a new subfamily of rhodaneses (Fig. 1). A close examination of the canonical motifs described for each of these subfamilies unveiled that LVIS_0852, and its homologous proteins, displayed a highly conserved motif [DW(T/N)GGT]. This motif is different from the signature present in most active rhodaneses [CXGGXR], in which the cysteine acts as a nucleophile residue (Bordo and Bork, 2002). In LVIS_0852, the cysteine position is instead occupied by a highly conserved aspartic acid (Fig. S1A), as described in non-active rhodaneses.

Fig. 1.

Neighbor-joining tree for representative members of the Rhodanese Superfamily. The tree was generated from a multiple alignment of 235 rhodanese modules using MUSCLE (Edgar, 2004). The following proteins were used as seeds for each rhodanese subfamily: LVIS_0852 (gi|116333498, TstT), RdhA (gi|226942893), Rhobov (gi|29135275), UBPY (gi|5058999), cdc25B (gi|11641413), Acr2 (gi|18414234), ThiI (gi|446385299), ArsR (gi|15608812), YbbB (gi|16128487), MoeB (gi|226942433), GlpE (gi|16131299), PspE (gi|16129269), YnjE (gi|3916038), YibN (gi|84028011), YceA (gi|2851415), and SseA (gi|401186).

To determine if this predicted new subfamily has enzymatic activity, LVIS_0852 was cloned and purified. The overexpression of LVIS_0852 produced a soluble polypeptide with a yield of 4 mg/liter. LVIS_0852 has an apparent molecular mass of 45 kDa in solution, as determined by size exclusion chromatography (Fig. S2A), suggesting the protein behaves as a trimer in its native conformation. The LVIS_0852-homolog LP_1913 is the only member with a solved 3D structure (PDB# 3FNJ). The asymmetric unit in LP_1913 is integrated by three monomers, suggesting its biological assembly may be a trimer. Purified LVIS_0852 was tested for thiosulfate:cyanide sulfurtransferase (TST; E.C. 2.8.1.1) and 3-mercaptopyruvate sulfurtransferase (MST; E.C. 2.8.1.2) activities. LVIS_0852 exhibited TST activity (49.9 ± 5.5 U mg⁻¹, Fig. 2A), while it did not have MST activity. Rhodaneses are capable of catalyzing the mobilization of the sulfane sulfur from thiosulfate to cyanide, generating as final products sulfite and the less toxic thiocyanate (Singleton and Smith, 1988; Cipollone et al., 2007a). Therefore, steady-state kinetics varying the concentration of one substrate at a time were performed. The kinetic parameters are summarized in the Table 1. It was found the enzyme displayed sigmoidal saturation kinetics for both substrates (Fig. 3A). The Hill coefficient (n), estimated by a non-linear regression analysis, was 2.9 ± 0.9 for CN⁻ and 4.4 ± 0.5 for S₂O₃²⁻, indicating positive cooperativity. The estimated Hill coefficient indicates the presence of four or more binding sites per catalytic unit, assuming infinite cooperativity.

Fig. 2.

Thiosulfate:cyanide sulfurtransferase [TST] activity. (A) The TST activity was determined in presence or absence of SO₃²⁻, EDTA, Fe(II), or Fe(III). (B) TST activity was performed using a mixture of equimolar concentrations of TstT and TstR, in presence or absence of SO₃²⁻, EDTA, or Fe(III). Either TstR or LVIS_0553 were assayed alone as controls. (C) TST activity was tested with a combination of equimolar concentrations of TstT with TstR mutants in the ligand binding site #2 and #3 (the residues mutated are indicated in brackets). Likewise, the presence or absence of Fe(III) is denoted.

Table 1.

Kinetic parameters of TstT and TstT:TstR complex.

Substrate		Vmax a	k0.5 a	n a	kcat-App
S2O32−		U · mg−1	mM		s−1
	TstT	68.6 ± 2.6	19.9 ± 0.5	4.4 ± 0.5	17.6
	TstT:TstR	61.5 ± 4.7	5.9 ± 0.9	1.6 ± 0.3	15.8
CN−		U · mg−1	mM
	TstT	30.9 ± 1.9	7.1 ± 0.7	2.9 ± 0.9	7.9
	TstT:TstR	60.4 ± 3.0	4.2 ± 0.5	1.6 ± 0.2	15.5

^aEstimated by nonlinear regression analysis using MicroCal Origin 8.

Fig. 3.

Thiosulfate:cyanide sulfurtransferase steady-state saturation kinetics of (A) TstT, or (B) TstT:TstR complex. Mean activities are representative of three independent experiments. The saturation kinetics experiments were conducted in presence of an excess of CN⁻, and in absence of iron.

The catalytic ability of LVIS_0852 decreased by 40 % in presence of 100 μM sulfite (Fig. 2A), while complete inhibition was observed at concentrations above 200 μM. The inhibitory effect of sulfite is a common feature of characterized rhodaneses (Cheng et al., 2008). We also determined the effect of adding EDTA chelating agent on TST activity. It was observed that the addition of 100 μM EDTA resulted in a 35 % decrease in TST activity. These results were interesting, as metals have not been previously described to contribute to TST activity. We tested the effect of adding several cations on the activity of LVIS_0852. The addition of 10 μM Fe(III) or Fe(II) increased LVIS_0852 activity by 50 % (Fig. 2A), while the addition of Ca(II), Co(II), Cu(II), Mg(II), or Ni(II) did not affect TST activity (data not shown). We speculate the effect observed in presence of Fe(II) was due to its rapid oxidation to Fe(III), since Fe(II) is highly unstable in solution at neutral pH (Morgan and Lahav, 2007) and the enzymatic activity is increased upon addition of Fe(III). We tested the metal content in the purified LVIS_0852 protein using Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES). However, no metal was detected. It is possible that iron was lost during the consecutive dialysis steps performed prior to the measurement by ICP-AES. Based on the results obtained with EDTA, where the TST activity was not completely inhibited, Fe(III) may not play a classical role as enzyme cofactor. Taken together, these findings suggest LVIS_0852 belongs to a new family of active rhodaneses with an unknown catalytic mechanism.

LVIS_0852:LVIS_0853 complex shows increased TST activity

LVIS_0852 is encoded immediately downstream of LVIS_0853, a member of the MarR family of transcriptional regulators. The analysis of the LVIS_0853 structural model revealed the presence of a cysteine residue in a predicted C-terminal helix (Fig. S1B and 6C). Based on both the exposed nature of this residue and its location, we tested the possibility that it may contribute to the activity of LVIS_0852. To explore this, LVIS_0853 was cloned and purified as a soluble polypeptide with a yield of 30 mg/liter. LVIS_0853 behaves as a dimer in solution with an apparent molecular mass of 45 kDa (Fig. S2B). TST activity was measured after mixing equimolar concentrations of LVIS_0852 and LVIS_0853 (Fig. 2B). Compared to LVIS_0852 alone, the TST activity increased 2-fold [100.9 ± 8.6 U mg⁻¹] when both proteins were present. As observed above for LVIS_0852, 100 μM EDTA decreased the activity of the complex [71.8 ± 5.7 U mg⁻¹]. Remarkably, a 2.8-fold increase [139.9 ± 7.4 U mg⁻¹] was obtained with the addition of 10 μM Fe(III). To determine if the increased activity is due to LVIS_0853 specific contribution, LVIS_0553 was used instead. LVIS_0553 is a MarR homolog that shares a similar fold to LVIS_0853 (Pagliai et al., 2010). The addition of LVIS_0553 at an equimolar ratio to LVIS_0852 had no effect on the TST activity [52.3 ± 4.1 U mg⁻¹]. Neither LVIS_0853 nor LVIS_0553 showed TST activity when tested alone. Taken together, these data suggest the TST activity is entirely performed by LVIS_0852 and enhanced by LVIS_0853 and Fe(III).

Fig. 6.

TstR possesses three ligand binding sites. (A) Model of L. brevis TstR using the structure of HucR (PDB# 2FBK; Bordelon et al., 2006) as a template. Chain A is depicted in light gray, while chain B is colored in dark grey. The putative ligand binding sites are indicated with dashed boxes. (B) Close view of binding site #1. In black are denoted the side chains of M64, F65, L74, L75, V78, and K82. (C) Binding site #2 is formed by the H136, H139, C167, and M171 residues. (D) Binding site #3 may coordinate iron with the side chains of E27, H147, E148, and E151 residues.

Overall, the saturation kinetics performed with the enzyme complex (LVIS_0852:LVIS_0853) follow the similar cooperative pattern described before (Fig. 3B). In the presence of an excess of S₂O₃²⁻ in the reaction mix, the complex displayed similar k_0.5 for CN⁻ to LVIS_0852 alone. However, the Hill coefficient for LVIS_0852:LVIS_0853 was significantly lower, suggesting less cooperativity. Interestingly, the enzyme affinity for S₂O₃²⁻ was 4 times higher when CN⁻ was in excess (Table 1). These data suggest LVIS_0852:LVIS_0853 complex is capable of quickly detoxifying cyanide at low thiosulfate concentrations. These results become more relevant considering the sulfane donors (i.e thiosulfate) are the limiting factor in the rhodanese reaction (Schulz, 1984). Since the mechanisms of catalysis are yet to be elucidated, it is difficult to propose a model to explain the observed results. One possibility is that the cysteine of LVIS_0853 enhances the recruitment of Fe(III), resulting in an improved catalysis. Simultaneously, in the LVIS_0852:LVIS_0853 complex, better hydrophobic compartments can be formed to enhance the interaction with the substrates, hence improving the enzyme affinity. Based on these results, LVIS_0852 was given the name TstT, and LVIS_0853 was named TstR for Thiosulfate:cyanide sulfurtransferase Regulatory protein.

TstR and TstT interact in vivo, and this interaction is modulated by Fe(III)

To confirm that the increased in vitro TST activity obtained with a mix of TstT and TstR was caused by a physical interaction between these proteins, in vivo experiments were conducted using a modified E. coli two-hybrid system (Charity et al., 2007). The fusion of the tstR gene to the ω subunit of the RNAP was generated using the plasmid pBRGP-ω, while the plasmid pACTR-AP-Zif served to merge the tstT gene to the zinc finger DNA-binding domain of the murine Zif268 protein (see “Experimental Procedures”). The interaction between TstT and TstR resulted in the transcriptional activation of the β-galactosidase reporter gene (1527 ± 117 AU, Table 2). As a control, LVIS_0553 was fused to the ω subunit of the RNAP. The interaction between LVIS_0553 and TstT was not significant (620 ± 52 AU). Since Fe(III) was found to modulate TST activity, we hypothesized that it may also be involved in the interaction between TstT and TstR. The amount of Fe(III) present in the media was increased from 10 to 100 μM, which led to a 2.5-fold increase in the β-galactosidase activity (3864 ± 162 AU, Table 2). These results suggest the increased TST activity is likely the result of a close interaction between TstT and TstR, which is modulated by the availability of Fe(III).

Table 2.

In vivo assessment of the interaction between TstT and TstR using a bacterial two-hybrid system.

strain	Cloned gene		β-galactosidase activity [AU]		fold induction
	pBRGP-ω	pACTR-AP-Zif	10 μM Fe(III)	100 μM Fe(III)
FP02	tstR	tstT	1527 ± 117	3864 ± 162	2.5
FP03	tstR[H136A-H139A-C167A]	tstT	1889 ± 33	2892 ± 299	1.5
FP04	tstR[H136A-H139A-H147A-E148A-C167A]	tstT	1694 ± 35	2521 ± 221	1.5
FP05	tstR	--	644 ± 25	ND	ND
FP06	--	tstT	376 ± 69	ND	ND
FP07	LVIS_0553	tstT	620 ± 52	ND	ND

ND: not determined

TstR binds in the leader region of the tstRT promoter

The tstT gene is encoded on the minus strand and immediately downstream of tstR, a member of the MarR family of transcriptional regulators. Comparative genomics, using the Ortholog Neighborhood Viewer at the JGI Integrated Microbial Genomes website (http://img.jgi.doe.gov/) and manual genome inspection, revealed that this arrangement is conserved in Lactobacillus species L. buchneri NRRL B-30929 (NC_015428), L. plantarum WCFS1 (NC_004567), L. kisonensis F0435 (NZ_AGRJ01000104), L. rhamnosus GG (AP011548), and L. zeae KCTC 3804 (NZ_BACQ01000050) (Fig. S1A). Moreover, this putative operon is conserved in the following other genera: Listeria ivanovii FSL F6-596 (NZ_ADXI01000891), Pediococcus claussenii ATCC BAA-344 (NC_016605), Staphylococcus aureus A5948 (NZ_ACKD01000020), S. lugdunensis HKU09-01 (NC_013893), S. epidermidis W23144 (NZ_ACJC01000133), Enterococcus malodoratus ATCC 43197 (NZ_KE136486), Streptococcus criceti HS-6 (NZ_AEUV02000002), and St. downei F0415 (AEKN01000001). The conservation of these operons correlated with the presence of the DW(T/N)GGT motif in the TstT orthologs with the exception of L. kisonensis, where glycine replaces the conserved aspartic acid.

Analysis of the genomic environment of the tstRT operon using bacterial promoter algorithms (http://www.softberry.com) identified a putative promoter in the 5′ region of tstR (P_tstR), and a Rho-independent termination in the 3′ region of tstT. The transcriptional start site for tstR was experimentally determined by 5′RACE-PCR and it was found 40 bp upstream of the translational start codon (Fig. 4B). A 224 bp fragment (base −153 to +71) containing the predicted promoter region was amplified by PCR, and electrophoretic mobility shift assays (EMSAs) confirmed that TstR binds with high affinity within this region (Fig. S3A).

Fig. 4.

Identification of TstR binding site in tstR promoter [P_tstR]. (A) DNase I footprint assay identified a protected region located upstream of tstR (base −1 to +33). The electropherogram shows a fragment of the digested probe in absence (black) or presence (white) of TstR, highlighting the protected region. The nucleotide sequence protected by TstR is shown as a circled box at the bottom of panels A and B. (B) Analysis of P_tstR. Predicted Shine-Dalgarno sequence (SD), and −10 and −35 of the P_tstR are underlined. The transcription start site (+1) is depicted in a triangle. IR, inverted repeat. Black arrows below the sequence denote the position of the primers used to generate the DNA probe for EMSA assays. (C) EMSA competition assays. The biotin labeled P_tstR promoter with 10 nM of TstR were mixed with increasing concentrations of four different unlabeled double-stranded DNA fragments (C-1, C-2, C-3, and C-4), encompassing the region identified by DNase I footprint in panel A.

To determine the binding site of TstR, a DNase I footprint assay was conducted. Two small protected zones were identified within a region of 34 nucleotides located from base −1 to base +33 on the minus strand (Fig. 4A). This protected region includes one inverted repeat (5′-TAgTTAGCCGGCTAAtTA-3′) flanked on both sides by AT-rich sequences (Fig. 4B).

To confirm whether this protected region is the binding site for TstR, a competition assay was performed. A variety of oligonucleotide combinations (Table S1) was used to generate four different unlabeled double-stranded DNA fragments (C-1, C-2, C-3, and C-4) that encompass the region identified by the DNase I footprint assay (Fig. 4C). While the C-1 fragment comprised the entire protected region, the probes C-2, C-3, and C-4 were partially overlapped, covering different sections of the TstR DNA binding site (C-2, +2 to +16; C-3, +9 to +24; and C-4, +18 to +33). The TstR:P_tstR complex was incubated with increasing concentrations of the unlabeled DNA probes. At a 1:10 ratio of P_tstR:C-1 DNA, more than 50 % of the TstR:P_tstR complex was disrupted, while complete disruption of the complex was observed at a 1:100 ratio. The overlapping fragments C-2, C-3, and C-4 were not able to compete with the P_tstR DNA. These results indicate that the entire region encompassed by the C-1 sequence is required for the binding of TstR to P_tstR.

In vitro TstR binding is modulated by sulfite and ferric iron

The effect of sulfite and iron was tested on the TstR:P_tstR interaction, since these compounds were found to modulate TstT enzymatic activity (Fig. 2). In EMSA experiments, the reducing compounds dithiothreitol (DTT), sulfite, and to a lesser extent cysteine, improved the binding of TstR to its cognate DNA site (Fig. 5A) while thioglycolate did not affect the binding of TstR. The interaction between TstR and P_tstR was not modified by other sulfur-related compounds including sulfate or thiosulfate. The observation that TstR activity is not affected by all reducing agents has previously been reported (for a review see Barron, 2006). Some factors that contribute to this variability are the amount and location of freely reacting –SH groups, as well as the redox potential of each compound.

Fig. 5.

Small molecules modulate TstR binding to P_tstR. (A) EMSA experiments were conducted using 10 nM TstR in presence of 100 μM sulfur-related compounds, as indicated on top of each lane. (B) EMSA experiments were carried out using 20 nM TstR with the addition of 20 μM metal chlorides, as indicated on top of each lane. No protein was added to the first lane.

Similarly, EMSA experiments were conducted with Fe(III) or Fe(II) and compared to other metal chlorides. The addition of 20 μM Fe(III) disrupted the interaction between TstR and P_tstR, while Fe(II) showed variable results, decreasing the TstR:P_tstR interaction to a lesser extent. Other metal chlorides such as Zn(II), Ca(II), Co(II), Cu(II), Mg(II), Mn(II), and Ni(II) had no effect (Fig. 5B). The specific effect of Fe(III) on the TstR:P_tstR interaction was studied by adding different iron chelators to the reaction mix during EMSA experiments. When deferiprone [a Fe(III) specific chelating agent (Ibrahim et al., 2006)] was added to the reaction, 20 μM Fe(III) was unable to disrupt the TstR:P_tstR interaction (Fig. S4A). However, when Fe(III) was tested in the presence of 2,2′-bipyridyl (a Fe(II) specific chelating agent), no protection was observed (Fig. S4B). Similar experiments were performed to determine if the effects observed with Fe(II) in Fig. 5B were due its oxidation to Fe(III) at neutral pH. It was found that Fe(II) did not modify the TstR:P_tstR interaction in presence or absence of the specific chelators (Fig. S4C). These results indicate that the unpredictable oxidation rates of Fe(II) to Fe(III) on EMSA experiments resulted in the effects observed in Fig. 5B.

Identification of the TstR binding site for sulfite and Fe(III)

TstR was modeled in silico at the Swiss-model workspace on automated mode (Schwede et al. 2003; Arnold et al., 2006), and refined using the I-TASSER server (Zhang, 2007; Roy et al., 2010) to determine putative binding sites for the ligands (Fig. 6A). The structure of HucR from Deinococcus radiodurans (PDB# 2FBK, Bordelon et al., 2006) was retrieved as the best hit (6.7e–25), although they share low sequence identity (23 %). Another structural homolog used to build the model was Spo143 from Silicibacter pomeroyi (PDB# 3CDH). The latter structure contains five sulfate molecules, four of which are located on the DNA binding domain. The fifth molecule was found in the interface of helices 3 and 4 of 3CDH. The model for TstR was then aligned to 3CDH, and a pocket formed by residues M64, F65, L74, L75, V78, and K82 was identified. This ligand binding site (#1) is located at the junction of the dimerization domain between helices Aα3–Aα4 (A, chain A) and Bα1 (B, chain B) in close proximity to the sulfate molecule in 3CDH (Fig. 6B). We hypothesized these residues serve as the ligand binding site for sulfite in TstR. Since the distinctive residues that coordinate iron are not present in this location (i.e. cysteine, histidine, tyrosine, aspartic or glutamic acid) it is unlikely that pocket #1 contains the iron binding site in TstR.

A visual inspection of the TstR model identified at least two putative iron binding sites. Binding site #2 is located in close proximity to ligand binding site #1, and it may coordinate iron with residues C167 and M171 from the Aα7 helix, along with residues H136 and H139 from the Bα6 helix (Fig. 6C). Binding site #3 is predicted to coordinate iron with the side chains of residues H147, E148, and E151 from the Bα7 helix, in addition to residue E27 from the Aα2 helix (Fig. 6D). Site-directed mutagenesis to alanine was conducted for residues in all three putative ligand binding sites. The mutant proteins were purified and tested for DNA binding and their contribution to the TST activity.

EMSA studies revealed that the binding of mutant TstR[M64A] (from binding site #1) to P_tstR was not affected by increasing concentrations of sulfite (up to 250 μM) (Fig. 7A–B and S3B). The other mutations in binding site #1 of TstR (F65A, L74A, L75A, V78A, and K82A) had no effect on the sensing of sulfite. The ability of TstR[M64A] to sense Fe(III) was then tested. As observed with the wild type protein, Fe(III) disrupted the TstR[M64A]:P_tstR interaction at similar concentrations (data not shown). These results indicate that M64 mediates specific sulfite binding to TstR, but it is not involved in iron recognition. Interestingly, M64 is conserved among all close homologs of TstR, highlighting its role as a regulatory residue.

Fig. 7.

M64 mediates sulfite sensing in TstR. EMSA experiments were conducted using (A) 10 nM WT TstR with increasing concentrations of sulfite; (B) 10 nM TstR[M64A] with increasing concentrations of sulfite; or (C) 10 nM TstR[M64A] with increasing concentrations of DTT. The concentration of the ligands in the reaction mix is indicated on top of each panel. No protein was added to the first lane.

Similarly, TstR mutants in the ligand binding site #2 were assayed for their ability to sense Fe(III) in EMSA experiments. While most of the TstR single mutants presented similar DNA binding affinities as compared to the wild type, the TstR[C167A] or TstR[H136A-H139A-C167A] showed a 10-fold increase in binding to P_tstR (Fig. S3C). The Fe(III) binding capabilities of the latter two mutants were then tested. It was found that 8 times more Fe(III) was required to disrupt the TstR[H136A-H139A-C167A]:P_tstR interaction when compared to the wild type TstR (Fig. 8A–B), while TstR[C167A] was partially impaired in its capability to bind Fe(III) (Fig. 9B).

Fig. 8.

Mutants in TstR binding pockets #2 and #3 are affected in DNA binding and ferric iron response. EMSA assays were carried out using (A) 10 nM of WT TstR; (B) 2.5 nM of TstR[H136A-H139A-C167A]; or (C) 10 nM of TstR[H147A-E148A-E151A]. The presence or absence of each key component in the EMSA reaction mix is depicted in top of each panel. Since the mutant proteins bind DNA with different affinities, the TstR : Fe(III) ratio is indicated for each case. No protein was added to the first lane.

Fig. 9.

In vitro DTT specifically reduces TstR C167 increasing DNA binding. EMSA experiments were conducted using (A) 10 nM TstR, incubated with increasing concentrations of Fe(III), in the presence or absence of 100 μM DTT; or (B) 2 nM TstR[C167A], incubated with increasing concentrations of Fe(III), in the presence or absence of 20 μM DTT. The presence or absence of both DTT and Fe(III) in the reaction mix are indicated in the top of each panel. In addition, the DTT : Fe(III) ratio is indicated. No protein was added to the first lane.

The role of the ligand binding site #3 was assessed using the TstR[H147A-E148A-E151A] triple mutant. This mutant required 2 times more Fe(III) to disrupt the TstR[H147A-E148A-E151A]:P_tstR complex when compared to the wild type TstR (Fig. 8C). These results suggest the presence of at least two Fe(III) binding sites in TstR, a high affinity binding site (#2), and a low affinity binding site (#3). It is also possible that site #2 indeed binds metal, and that the mutation in the nearby site #3 affects the conformation of site #2. A comparison of binding site #2 and #3 to TstR homologs indicated that residues in binding site #2 are not conserved, and no homologs were identified with a cysteine in the C-terminal domain. In contrast, the ligand binding site #3 is conserved among homologs of TstR. However, in the modeled TstR proteins from L. rhamnosus and L. zeae, the binding sites #2 and #3 appear to be merged into one site and mediated by a cysteine residue (data not shown).

TstR mutants that showed impaired ligand sensing were tested for their ability to increase the enzymatic activity of TstT. The complex TstT:TstR[H136A-H139A-C167A] (binding site #2) promoted a low increase in the TST activity (73.6 ± 7.7 U mg⁻¹) when compared to the wild type TstT:TstR (100.9 ± 8.6 U mg⁻¹) (Fig. 2C). Further addition of Fe(III) did not significantly modify the enzymatic activity of the TstT:TstR[H136A-H139A-C167A] complex (85.4 ± 5.7 U mg⁻¹). In contrast, the TstT:TstR[H147A-E148A-E151A] complex presented similar values of TST activity (104.0 ± 6.8 U mg⁻¹) when compared to the wild type complex. Further addition of Fe(III) increased the enzymatic activity by 50 % (149.4 ± 12.3 U mg⁻¹). These results indicate that binding site #2 is the significant site for Fe(III) coordination and stimulation of the TST activity.

To determine if Fe(III) is involved in mediating interactions between TstR and TstT, two-hybrid experiments were performed using TstR mutants in ligand binding sites #2 and #3. At low Fe(III) concentrations (10 μM), the TstT:TstR[H136A-H139A-C167A] interaction (binding site #2) yielded a β-galactosidase activity similar to the wild type TstT:TstR complex. Comparable results were obtained using a mutant of TstR with combined substitutions in ligand binding sites #2 and #3 (TstR[H136A-H139A-H147A-E148A-C167A]). The activity of the reporter gene was weakly stimulated (1.5-fold) with the addition of 100 μM Fe(III) (Table 2), contrasting previous observations with the wild type proteins (2.5-fold increase). These results indicate that the binding of Fe(III) in TstR site #2 plays a crucial role in the enhancement of the enzymatic activity of TstT, the modulation of DNA binding, and protein-protein interactions. Conversely, binding site #3 may only provide a low affinity binding site that affects the TstR:P_tstR interaction.

TstR responds to redox conditions by sensing antagonistic ligands

The improved interaction between TstR and P_tstR observed in presence of sulfite and DTT may be in part attributable to the reducing environment induced by these compounds. Since we observed antagonistic effects of Fe(III) and DTT on TstR:P_tstR interactions (Fig. 5), EMSA experiments combining these two chemicals were performed. To this end, TstR was incubated with increasing concentrations of Fe(III) in the presence of 100 μM DTT or 10 mM Tris buffer pH 8.0 as a control, and further assayed in EMSA experiments (Fig. 9A). It was found that the reduced form of TstR required 4 times more Fe(III) to disrupt the TstR:P_tstR interaction, when compared to the untreated form.

The addition of specific reducing compounds may act either by reducing Fe(III) to Fe(II), rendering it unable to disrupt the TstR:P_tstR interaction, or by reducing specific amino acids in TstR, and consequently becoming unavailable to interact with Fe(III). The later possibility would be indicative of a redox mechanism in which Fe(III) may provide a mild oxidative environment. We tested if the addition of an oxidant agent such as hydrogen peroxide would provoke similar changes in TstR behavior. We found that concentrations of H₂O₂ up to 1 mM did not affect the TstR:P_tstR interaction (data not shown).

The next step was to determine the specific role of DTT on TstR binding. Since DTT and sulfite increase binding of TstR to P_tstR, we tested the effect of DTT on the TstR[M64A] that no longer senses sulfite. It was found that TstR[M64A] binding is increased in presence of DTT (Fig. 7C), as observed for the wild type TstR (Fig. S5A). These results confirmed the specific role of sulfite as a co-repressor of TstR, while the observed effect of DTT on TstR might be through the reduction of a specific residue.

The TstR[C167A] mutant showed increased binding to DNA when compared to the wild type protein (Fig. S3C). It was observed that the addition of increasing concentrations of DTT did not affect the binding of TstR[C167A] to DNA (Fig. S5B). In presence of DTT, higher concentrations of Fe(III) were required to disrupt the TstR[C167A]:P_tstR complex (Fig. 9B), similar to the results obtained in the wild type TstR. Taken together, these results indicate that in vitro DTT may reduce Fe(III) to Fe(II), decreasing the concentration of Fe(III) available to interact with TstR. DTT also reduces the C167 located in the Fe(III) binding pocket, but it did not impair Fe(III) binding to TstR. In summary, TstR may be responding to redox conditions by binding antagonistic ligands [Fe(III) and sulfite] on two different pockets.

TstR and TstT are involved in cyanide detoxification

Based on the biochemical characterization of TstT and the ligands recognized by TstR, we hypothesized that TstR and TstT are involved in the detoxification of cyanide. Tolerance to cyanide was first evaluated in L. brevis. Under static conditions, L. brevis grew with up to 10 mM KCN, while its growth under aeration at 200 rpm was severely affected at much lower concentrations (500 μM). Due to the inability to obtain a knockout mutant in L. brevis, we reasoned that if the tstRT operon is involved in detoxifying cyanide, its overexpression would result in increased tolerance to KCN. To this end, tstR and tstT genes (including the complete P_tstR) were cloned in the low copy vector pRV610 (Crutz-Le Coq and Zagorec, 2008). L. brevis cells, carrying either the complete tstRT (strain FP08) or the empty vector as a control (FP09), were grown in LAPT broth under aeration at 200 rpm, in absence or presence of KCN. A significant 23 % decrease (p-value < 0.001) in the optical density (OD₆₀₀) was observed at late exponential phase for strain FP09 in the presence of 500 μM KCN, when compared to a culture without cyanide (Fig. 10A). In contrast, growth of strain FP08 was not affected by 500 μM KCN throughout the entire growth curve (Fig. 10A). The role of Fe(III) in modulating the tolerance to cyanide was then tested by adding the Fe(III)-specific chelating agent deferiprone, due to the presence of iron rich components in the complex growth media used for these experiments. The addition of 1 mM deferiprone resulted in a 30 % decrease in growth upon cyanide exposure (p-value < 0.001) (Fig. 10A). These results suggest that the TstT:TstR complex plays a role in the detoxification of cyanide in presence of Fe(III) in L. brevis.

Fig. 10.

Overexpression of the tstRT operon increased cyanide tolerance in L. brevis and E. coli. (A) The control vector pRV610 or pFP08 (pRV610::P_tstR-tstR-tstT) were used to transform L. brevis ATCC 367, generating strains FP09 (left, open symbols) and FP08 (right, closed symbols), respectively. Cultures were grown in LAPT medium, supplemented with 200 μM FeCl₃ (squares), 200 μM FeCl₃ plus 500 μM KCN (diamonds), or 200 μM FeCl₃ plus 500 μM KCN and 1 mM deferiprone (triangles). The values are means ± S.D. (B) Strains FP01 (pACTR-Zif, pBRGP-ω; squares), FP02 (pACTR-tstT-Zif, pBRGP-tstR-ω; circles), FP05 (pACTR-Zif, pBRGP-tstR-ω; diamonds), and FP06 (pACTR-tstT-Zif, pBRGP-ω; triangles) were grown to late-exponential phase in LB medium in the presence of different concentrations of KCN (50–400 μM). Cultures grown in the absence of KCN correspond to control conditions and represent 100 % of growth. The values are means ± S.D.

To establish a link between the increased tolerance to KCN and the tstRT operon, the expression levels of the tstR and tstT genes were determined. L. brevis cells were grown under aeration in LAPT broth supplemented with 500 μM KCN and collected at early exponential phase (OD₆₀₀ = 0.3). The mRNA levels were quantified by qRT-PCR. It was observed that the presence of KCN induced the expression of both tstR and tstT genes by 1.8- and 2.4-fold, respectively. However, the addition of 50 μM Fe(III) had no further effect on the expression of either gene, possibly due to the presence of iron in the media components. These results indicated that cyanide (or an unknown derivative) may act as an inducer of tstRT expression in vivo; however, KCN or ferricyanide did not modify the binding of TstR to its cognate DNA in vitro (data not shown).

To determine if the resistance to cyanide is mediated by TstR, TstT, or the TstT:TstR complex, the genetic fusions generated for the two-hybrid system in E. coli were used (Fig. 10B). It was found that E. coli FP02, carrying both tstR and tstT, had 67 % more resistance to 200 μM KCN than control strain (FP01) carrying the empty vectors (p-value < 0.005). An intermediate effect (36 % increased resistance) was observed for strain FP06 (carrying only tstT) when compared to the strain FP01 (p-value < 0.005). No significant difference was observed with strain FP05 (carrying only tstR) when compared to the control (FP01) (Fig 10B). These results suggest that TstT by itself can improve the tolerance to cyanide, although the TstT:TstR complex is more efficient in detoxifying cyanide in vivo.

Discussion

In the present study, we showed that the tstRT operon from L. brevis is involved in the tolerance to cyanide. We demonstrated that TstT and TstR undergo a physical interaction to catalyze the enzymatic transformation of cyanide to a less toxic compound. Detailed phylogenetic analyses indicated that TstT forms a separate cluster, and is not related to other characterized active or inactive rhodaneses. A biochemical characterization unveiled that Fe(III) stimulates TstT activity, a new feature among rhodaneses. Interestingly, TstT activity was further increased by the addition of the TstR regulator. The enzymatic activity of the TstT:TstR complex was similar to previously characterized rhodaneses with a single active domain (Vazquez et al., 1987a; Ray et al., 2000; Giuliani et al., 2007; Cheng et al., 2008). In contrast to the Michaelis-Menten behavior described for most of the biochemically characterized rhodaneses (Alexander and Volini, 1987; Pagani et al., 1993; Trevino et al., 1998; Ray et al., 2000; Giuliani et al., 2007; Cipollone et al., 2008; Cheng et al., 2008), saturation kinetic experiments with TstT and the TstT:TstR complex showed enzyme cooperativity, similar to Rhodopseudomonas palustrius rhodanese (Vazquez et al., 1987b). The k_m of rhodaneses following Michaelis-Menten behavior has been reported in the range of 3 to 80 mM for thiosulfate and cyanide (Alexander and Volini, 1987; Pagani et al., 1993; Trevino et al., 1998; Ray et al., 2000; Giuliani et al., 2007; Cipollone et al., 2008; Cheng et al., 2008), with a turnover number (k_cat) of 10 to 250 s⁻¹ (Alexander and Volini, 1987; Ray et al., 2000; Bauer and Papenbrock, 2002; Cheng et al., 2008). TstT and the TstT:TstR complex displayed a k_0.5 in a similar range of concentrations described above and with an equivalent apparent k_cat (Table 1).

Protein-protein interaction phenomena involving sulfurtransferase activity have been recently described in the process of sulfur compound homeostasis (i.e. IscS and FdhD, Thome et al., 2012; NifS or IscS and RhdA, Cartini et al., 2011). However, scarce information is available regarding protein-protein interaction events involving transcriptional regulators and enzymes (Garcia et al., 2011; Wilke et al., 2008). However, in the reported cases (i.e. Garcia et al., 2011; Wilke et al., 2008), a protein-protein interaction event has an impact on the transcriptional regulation of genes. Here we provide novel evidence supporting the modulation of thiosulfate:cyanide sulfurtransferase activity in L. brevis facilitated by the interaction between TstT and TstR transcriptional regulator.

We proposed a new mechanism of sensing redox conditions mediated by the availability of specific ligands, Fe(III) and sulfite (Fig. 11). When the availability of Fe(III) is low (reduced conditions), TstR binds to its cognate promoter, repressing the expression of the tstR and tstT genes. At high concentrations (indicative of an oxidative environment), Fe(III) binds with high affinity to TstR, disrupting the TstR:P_tstR interaction, thus allowing the expression of the tstRT operon. Subsequently, TstR and TstT undergo a physical interaction which stimulates the sulfurtransferase activity, resulting in increased tolerance to cyanide. The reducing environment offered by sulfite significantly improved the interaction between TstR and P_tstR, thus preventing transcription of the operon. Consistent with molecules having physiological relevance, the concentrations of sulfite and ferric iron required to modulate TstR activity were in the low micromolar range. It is also possible that a Fe(III):cyanide derivative may be involved in signaling; however, under our experimental conditions we were not able to identify it.

Fig. 11.

Proposed model for TstR regulation of the tstRT operon. At low concentrations of Fe(III) (i.e. reducing conditions), TstR binds to its cognate promoter, repressing the expression of the tstR and tstT genes. Under oxidative conditions Fe(III) would bind with high affinity to TstR, disrupting the TstR:P_tstR interaction. As a consequence of the expression of the tstRT operon, TstR interacts with TstT stimulating the sulfurtransferase activity, increasing the tolerance to cyanide. The later step is regulated by the reducing environment offered by sulfite (the product of sulfurtransferase activity), which would create a negative feedback loop by improving the interaction between TstR and P_tstR, hence preventing transcription of the operon.

Limited information is available about MarR homologs modulated by iron or other metals. In P. aeruginosa, expression of the pqrABC operon is regulated by the redox status of the iron-containing prosthetic group of PqrR (Rungrassamee et al., 2009). In contrast, AdcR from Streptococcus pneumonia binds zinc in order to repress the high-affinity zinc-specific uptake system adcCBA (Reyes-Caballero et al., 2010). In addition, zinc binding occurs at two different sites per monomer (Guerra et al., 2011). A structural comparison of AdcR revealed that one of the binding sites for zinc is located in the same place as the predicted binding site for sulfite in TstR. As such, the co-repressor activity of sulfite observed in TstR is in agreement with the event described in AdcR. These findings are significant since the binding of small molecules to other MarR members often results in the induction of gene expression for genes that are being regulated. However, in TstR we observed a second molecule, Fe(III), binds at a different site, resulting in decreased DNA binding affinity. These results indicate that TstR is able to induce or repress gene expression by responding to two different molecules. This feature has been previously reported in the unrelated IclR family of transcriptional regulators, where pyruvate and glyoxylate antagonize the expression of the aceBAK operon (Lorca et al., 2007b). The mechanism described herein for the binding of iron to TstR is similar to previously reported mechanisms for MarR members involved in redox sensing. In MgrA, SarZ, and some OhrR homologs, the oxidation of a unique cysteine, and its further reaction with an external small molecule thiol, leads to an altered protein conformation (mixed disulfide form). As a consequence, the interaction with DNA is reduced, and the regulon expression is promoted (Perera and Grove, 2010). TstR is different, since the redox sensing is attained through the sensing of two different molecules that bind in different pockets of the protein.

The modulatory effect of ferric iron observed in vitro correlated with the in vivo induction of the tstRT operon, in L. brevis. We determined that these effects were the result of the specific binding of ferric iron to TstR. Iron is considered one of the most important micronutrients for all living organisms, and is required for a wide variety of essential functions (i.e., as an enzyme cofactor; Escolar et al., 1999). It is also widely used to trigger the expression of virulence genes in pathogenic bacteria (Guedon and Helmann, 2003). Although it has been reported that iron does not play an essential role in lactic acid bacteria, where manganese and cobalt mimic the role of iron (Krewulak and Vogel, 2008), our results clearly indicate that iron, but no other metal, is capable of modulating the interaction between TstR and TstT, resulting in increased thiosulfate:cyanide sulfurtransferase activity.

We determined that this interaction mediates the increased tolerance to cyanide in L. brevis, and E. coli. Similar findings regarding the importance of iron in lactic acid bacteria were confirmed in L. sakei, where a complete iron uptake system was recently described (Duhutrel et al., 2010).

The biological link between Fe(III) and cyanide has been established in the cyanide degrading bacterium Pseudomonas pseudoalcaligenes (Huertas et al., 2006). It was observed that, in addition to specific enzymes that degrade the toxic compound (i.e., to ammonia), the microorganism should have a resistance mechanism (i.e. a cyanide insensitive oxidase) or detoxification system (i.e., rhodaneses), and an efficient uptake system for Fe(III) from the environment (i.e., siderophores). Although we have yet to establish the link between Fe(III) and cyanide in L. brevis, we hypothesize that the expression of the tstRT may be related to the metabolism of cyanogenic glycosides. These compounds are abundant in many natural environments (i.e., cassava) where this species is found (Lei et al., 1999).

In conclusion, we provide evidence of a dual activity for a MarR transcriptional regulator. Besides its role as a transcriptional repressor, TstR increased the thiosulfate:cyanide sulfurtransferase activity performed by the protein encoded by its downstream gene, tstT. This enhancement effect is the product of a protein-protein interaction modulated by the presence of ferric iron. The physical interaction observed between TstR and TstT represents a novel mechanism of regulation for the enzymatic activity performed by a transcriptional regulator.

Experimental Procedures

Bacterial strains

Lactobacillus brevis ATCC 367 was obtained from the American Type Culture Collection (ATCC, Manassas, VA). L. brevis was grown at 30 °C in MRS broth (Difco Laboratories, Detroit, MI). When indicated, L. brevis cells were grown under aeration at 200 rpm in LAPT media (Raibaud et al., 1961). Escherichia coli DH5α and JM109 cells, used to carry and propagate all vectors, were grown in Luria-Bertani medium (LB, Difco) at 37 °C, under aerobic conditions.

When required, different CN⁻ : Fe(III) proportions were added to the growth media based on the tolerance to cyanide of the bacterial strain, and the minimum amount of iron that will improve the tolerance to cyanide.

When appropriate, media was supplemented with erythromycin, 5 μg ml⁻¹, (for L. brevis), or ampicillin, 100 μg ml⁻¹, tetracycline 10 μg ml⁻¹, kanamycin 50 μg ml⁻¹ (for E. coli). All antibiotics and chemicals were purchased from Sigma-Aldrich (St. Louis, MO). All strains used for this study are shown in Table 3.

Table 3.

Strains and plasmids used in this study.

Name	Genotypea	Origin/reference
Bacterial Strains
E. coli DH5α	φ80 dlacZΔM15Δ(lacZYA-argF)U169 recA1 endA1 hsdR17 (rK− mK+) supE44 thi-1 gyrA relA1	Laboratory stock
E. coli BL21 (DE3)	F− ompT gal dcm lon hsdSB(rB− mB−) λ(DE3 [lacI lacUV5-T7 gene 1 ind1 sam7 nin5])	Stratagene
E. coli JM109	endA1, recA1, gyrA96, thi-1, hsdR17 (rK− mK+), relA1, supE44, Δ(lac-proAB), [F’ traD36, proAB, laqIqZΔM15]	New England Biolabs
E. coli KDZif1ΔZ	araD (gpt-lac)5, rpsL (Strr), ΔspoS3::cat (Camr) [F’ lacIq (Z321(-61) lacZYA*) Kanr]	Vallet-Gely et al., 2005
L. brevis ATCC 367	Standard strain (wild type)	ATCC, Manassas, VA
FP01	KDZif1ΔZ pACTR-Zif, pBRGP-ω. Tetr, Apr.	This work
FP02	KDZif1ΔZ pACTR-tstT-Zif, pBRGP-tstR-ω. Tetr, Apr.	This work
FP03	KDZif1ΔZ pACTR-tstT-Zif, pBRGP-tstR[H136A-H139A-C167A]-ω. Tetr, Apr.	This work
FP04	KDZif1ΔZ pACTR-tstT-Zif, pBRGP-tstR[H136A-H139A-H147A-E148A-C167A]-ω. Tetr, Apr.	This work
FP05	KDZif1ΔZ pACTR-Zif, pBRGP-tstR-ω. Tetr, Apr.	This work
FP06	KDZif1ΔZ pACTR-tstT-Zif, pBRGP-ω. Tetr, Apr.	This work
FP07	KDZif1ΔZ pACTR-tstT-Zif, pBRGP-LVIS_0553-ω. Tetr, Apr.	This work
FP08	ATCC 367 pFP08. Emr.	This work
FP09	ATCC 367 pRV610. Emr.	This work
Plasmids
p15TV-L	Expression vector for purification of proteins by nickel affinity cromatography. Apr.	Pagliai et al., 2010
pRV610	Minimum replicon of pRV500 fused to pBluescript replicon with a MCS-αlacZ. Apr, Emr.	Crutz-Le Coq and Zagorec, 2008
pBRGP-ω	Vector used for translational fusion to the N-terminus of the ω subunit of E. coli RNAP. Apr.	Vallet-Gely et al., 2005
pACTR-AP-Zif	Vector used for translational fusion to the N-terminus of the zinc finger DNA-binding domain of the murine Zif268 protein. Tetr.	Vallet-Gely et al., 2005
pFP02a	pBRGP-tstR-ω. Apr.	This work
pFP02b	pACTR-tstT-Zif. Tetr.	This work
pFP03	pBRGP-tstR[H136A-H139A-C167A]-ω. Apr.	This work
pFP04	pBRGP-tstR[H136A-H139A-H147A-E148A-C167A]-ω. Apr.	This work
pFP07	pBRGP-LVIS_0553-ω. Apr.	This work
pFP08	PtstR-tstR-tstT transcriptional fusion in pRV610 carrying the L. brevis sequence from −152 to +928. Apr, Emr.	This work

^aThe positions indicated are relative to the tstR transcriptional start site.

DNA manipulations and gene cloning

Standard methods were used for chromosomal DNA isolation, restriction enzyme digestion, agarose gel electrophoresis, ligation, and transformation (Sambrook et al., 1989). Plasmids were isolated using QIAprep^® Spin Miniprep Kit (Qiagen, Valencia, CA) and PCR products were purified using QIAquick^® purification kits (Qiagen).

For protein expression and purification, tstT and tstR genes were amplified from L. brevis ATCC367 chromosomal DNA using PCR and cloned into the p15TV-L plasmid as described previously (Pagliai et al., 2010). The primers utilized are described in Table S1. Site-directed mutagenesis was performed using the QuikChange Site-directed Mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer’s protocol using the plasmid p15TV-tstR as the template. All selected amino acids were changed to alanine. Mutations were verified by DNA sequencing using T7 primers.

For the bacterial two-hybrid system, tstR gene from L. brevis was amplified by PCR, and fused to the ω subunit of the RNAP by cloning into the NdeI and NotI sites of the pBRGP-ω plasmid. The tstT gene was fused to the zinc finger DNA binding protein of the murine Zif268 protein via insertion into the NdeI and NotI sites of the plasmid pACTR-AP-Zif. The recombinant clones were selected by transformation in E. coli JM109, confirmed by sequencing, and transformed into the reporter strain KDZif1ΔZ by standard methods (Sambrook et al., 1989). The reporter strain KDZif1ΔZ harbors an F’ episome containing the lac promoter derivative placZif1-61, driving the expression of a linked lacZ reporter gene (Charity et al., 2007).

Protein purification

Protein purification was carried out as described previously (Lorca et al. 2007b). Briefly, the His-tagged fusion proteins were overexpressed in E. coli BL21-Star(DE3) cells (Stratagene). The cells were grown in LB broth at 37 °C. Protein expression was induced with 0.5 mM IPTG when OD₆₀₀ = 0.5. After addition of IPTG, the cells were incubated with shaking at 17 °C overnight. The cells were harvested and the pellet was resuspended in binding buffer (500 mM NaCl, 5 % Glycerol, 50 mM Tris pH 8.0, and 5 mM Imidazole), and stored at −80 °C. A French Press was used to lyse the thawed cells. The lysate was clarified by centrifugation (15,000 rpm for 30 min at 4 °C), and applied to a metal chelate affinity-column charged with Ni²⁺ (Ni-NTA Superflow, Qiagen), previously equilibrated in binding buffer. The column was then washed (binding buffer with 15 mM imidazole), and the protein was eluted in elution buffer (binding buffer with 250 mM imidazole). The purified proteins were dialyzed against 10 mM Tris pH 8.0, 500 mM NaCl, and 2.5 % glycerol and stored in aliquots at −80 °C. Protein concentration was estimated using the Bio-Rad protein assay kit (Bio-Rad). The concentration of Fe(III) was determined using inductively coupled plasma atomic emission spectrometry (ICP/AES), following EPA method 200.7 (Martin et al., 1994) at the IFAS Analytical Research Laboratory, University of Florida, Gainesville.

Size-exclusion chromatography

100 μl of protein samples were prepared using 10 mM Tris pH 8.0, 500 mM NaCl, and 10 μM TstR or TstT. After 20 min of incubation at 30 °C, samples were injected onto a prepacked Superose 12 10/300 GL gel filtration column (GE Healthcare, Pittsburgh, PA), connected to a LCC-501 plus (GE Healthcare), and equilibrated with 10 mM Tris pH 8.0 and 500 mM NaCl. Filtration was carried out at 4 °C, using a flow rate of 0.5 ml/min. The eluted proteins were monitored continuously for absorbance at 280 nm using a UV-M II monitor (GE Healthcare). Blue dextran 2000 was used to determine the void volume of the column. A mixture of protein molecular weight standards, containing IgG (150 kDa), BSA (66 kDa), Albumin (45 kDa), Trypsinogen (24 kDa), Cytochrome C (12.4 kDa), and Vitamin B12 (1.36 kDa) was also applied to the column under similar conditions. The elution volume and molecular mass of each protein standard was then used to generate a standard curve from which the molecular weight of eluted proteins was determined. The theoretical molecular weight of TstR and TstT were predicted from the amino acid sequence using the Compute pI/Mw tool at the ExPASy Proteomics Server (http://ca.expasy.org/tools/pi_tool.html).

DNase I Footprint

Protection assays were performed on the minus and plus strands by using 5′-VIC or 5′-FAM labeled probes generated by PCR, using primers LVIS853fprint_FAM Fw and LVIS853fprint_VIC Rv (Table S1). The reaction mixture was the same as used for EMSA assays, except that 2.5 ng μl⁻¹ labeled probe, 0.25 μg of TstR, 0.5 mM CaCl₂, 2.5 mM MgCl₂, and 0.006U of DNase I (New England Biolabs, Ipswich, MA) were added in a 200 μl reaction. The reaction was incubated for 20 min at 37 ºC. A digestion reaction without TstR was included as a control. The reaction was terminated by incubating 20 min at 70 ºC with the addition of 10 mM EDTA, pH 8.0 (Pagliai et al., 2010). The digested DNA and sequencing reaction products were analyzed at the Plant and Microbe Genomics facility, Ohio State University, Columbus, with a 3730 DNA analyzer, and the protected regions were identified with GeneMarker (Soft Genetics), as described earlier (Zianni et al., 2006).

Electrophoretic Mobility Shift Assays (EMSA)

EMSA analysis of TstR was performed using proteins purified according to the procedures described above. The fragment containing the putative binding site for TstR (Fig. 4) was generated by PCR using biotin-prelabeled (5′-end) primers (Table S1), and subsequently purified using QIAquick spin columns (Qiagen). The optimized reaction mix for EMSA (20 μl) contained 0.36 nM of a 5′-labelled DNA fragment, 50 mM Tris-HCl, pH 7.5, 150 mM KCl, 10 mM MgCl₂, 0.01 % Triton X100, 25 ng μl⁻¹ Poly(dI-dC) nonspecific competitor DNA, purified TstR protein (0–20 nM), and ligand (0–1 mM), unless otherwise specified. After incubation for 20 min at 37 °C, samples were separated on 6 % acrylamide/bis-acrylamide non denaturing gels, in 0.5X Tris borate-EDTA buffer, pH 8.3 (TBE). Electrophoresis was performed at 4 °C in 0.5X TBE as a running buffer. The DNA was then transferred from the polyacrylamide gel to a Hybond-N⁺ membrane (GE Healthcare) by electroblotting at 380 mA for 45 min in 0.5X TBE. Transferred DNA was UV cross-linked, and biotin labeled DNA was detected using Phototope-Star Detection Kit (New England Biolabs). Membranes were exposed to Kodak X-ray film.

For EMSA competition assays, different oligonucleotides encompassing the protected region identified in the DNase I footprint assay were synthesized (Table S1). Annealing was carried out by mixing equimolar concentrations (100 μM) of each complementary oligonucleotide in 0.25 mM Tris-HCl, pH 8.0. The mixture was incubated at 95 °C for 5 min and then chilled on ice to further allow them to anneal at RT.

5′RACE-PCR

The transcription start site of the tstR gene was determined by a modified 5′RACE-PCR protocol. L. brevis cells were grown in MRS broth until exponential phase and the total RNA was extracted using the RiboPure-Bacteria kit (Ambion, Austin, TX) following the manufacturer’s protocol. 2.5 μg of RNA was first treated with 20 U of the Calf intestine alkaline phosphatase (New England Biolabs) for 1h to remove the 5′-PO₄ from degraded RNAs. The reaction was ended, and the RNA was extracted with phenol:chloroform:isoamylalcohol precipitation. The RNA was then treated with 2.5 U of Tobacco acid pyrophosphatase (Epicentre Biotechnologies, Madison, WI) for 1h to remove the 5′-cap from mRNAs. The CIP/TAP RNA was then ligated to the Oligo_Fw_RNA adapter (Table S1). The first strand of cDNA was synthesized, using primers described in Table S1, with the SuperScript^® II Reverse Transcriptase (Invitrogen), following the manufacturer’s protocol. The cDNA was then used in a PCR reaction using primers Oligo_Fw and 853_Race_Rv (Table S1) for its further cloning using the StrataClone Blunt PCR cloning kit (Stratagene), following the manufacturer’s protocol. The plasmids carrying the RACE-PCR amplifications were sequenced with the subsequent confirmation of the tstR transcriptional start site.

qRT-PCR studies

L. brevis cells were grown under aeration at 200 rpm in LAPT media in the absence or presence of 500 μM of KCN and/or 50 μM FeCl₃. The cells were harvested by centrifugation at 4 °C when the OD₆₀₀ = 0.3 (exponential phase). Total RNA was next isolated using the RiboPure-Bacteria kit (Ambion), following the manufacturer’s protocol. The cDNAs were synthesized using the iScript^™ cDNA Synthesis kit (Bio-Rad, Hercules, CA) using random hexamers, in accordance with the manufacturer’s protocol. Quantitative real-time PCR was performed in a Bio-Rad iCycler IQ apparatus, using the iQ SYBR^® Green SuperMix (Bio-Rad), following the manufacturer’s recommended procedure. The primers for tstT, tstR, and the internal control rpoD (LVIS_0756), are depicted in Table S1.

β-galactosidase assays

For the two-hybrid system, E. coli cells were grown at 37 °C with aeration in MOPS media, in the presence of either 10 or 100 μM Fe(III). Gene expression of the fusion proteins in the reporter strain was induced by the addition of 5 μM IPTG, at an OD₆₀₀ = 0.25. Culture samples were taken after 4 h, permeabilized with 0.15 % sodium dodecyl sulfate (SDS) and 1.5 % chloroform in Z-buffer (Miller, 1972).

β-galactosidase activity was assayed by following the catalytic hydrolysis of the chlorophenol red-β-D-galactopyranoside (CRPG) substrate (Sigma-Aldrich). The change in absorbance at 570 nm was read continuously using a Synergy HT 96-well plate reader (BioTek, Winooski, VT). β-galactosidase activity, expressed as arbitrary units (AU), was calculated using the slope of absorbance curve normalized with the initial cell density. The assays were performed in triplicate.

Sulfurtransferase activity assays

Thiosulfate:cyanide sulfurtransferase (TST, EC 2.8.1.1) activity was assessed as described previously (Sorbo, 1957; Westley, 1981). Briefly, the reaction mixture contained 100 mM Tris-acetate pH 8.6, 50 mM sodium thiosulfate, 50 mM KCN, and 1–10 μg of TstT or TstR proteins in a final volume of 0.5 ml. The reaction was started with the addition of cyanide, and ended after 5 min by adding 0.25 ml of 15 % formaldehyde. The characteristic red-brown color of the acidic iron-thiocyanate complex was obtained after the addition of 0.75 ml of ferric nitrate reagent (165 mM Fe(NO₃)₃·9H₂0 in 8.7 % HNO₃). The amount of thiocyanate formed was quantified at 460 nm using a UV-1700 Pharma Spec Spectrophotometer (Shimadzu, Columbia, MD). One unit was defined as the amount of enzyme required to catalyze the formation of 1 μmol of thiocyanate per min. The assays were performed in triplicate. No difference in the TST activity was detected in the presence or absence of the His₆-tag in the purified proteins. The kinetic parameters (k_0.5 and V_max) were determined by nonlinear regression of the Hill equation, using MicroCal Origin8 software (Northampton, MA).

3-mercaptopyruvate sulfurtransferase (MST, EC 2.8.1.2) activity was determined as described previously (Westrop et al., 2009). Briefly, the reaction mixture contained 100 mM Tris-HCl pH 8.4, 0.3 mM 3-mercaptopyruvic acid, 1 mM β-mercaptoethanol, 0.15 mM lead nitrate, and 1–10 μg of TstT or TstR proteins in a final volume of 1 ml. The reaction was started with the addition of lead nitrate and the change on the absorbance at 360 nm was measured after 5 and 60 min using a UV-1700 Pharma Spec Spectrophotometer (Shimadzu). The assays were performed in triplicate.