LurR is a regulator of the central lactate oxidation pathway in sulfate-reducing Desulfovibrio species

The central carbon/lactate utilization pathway in the model sulfate-reducing bacterium, Desulfovibrio vulgaris Hildenborough, is encoded by the highly conserved operon DVU3025-3033. Our earlier in vitro genome-wide study had suggested a network of four two-component system regulators that target this large operon; however, how these four regulators control this operon was not known. Here, we probe the regulation of the lactate utilization operon with mutant strains and DNA-protein binding assays. We show that the LurR response regulator is required for optimal growth and complete lactate utilization, and that it activates the DVU3025-3033 lactate oxidation operon as well as DVU2451, a lactate permease gene, in the presence of lactate. We show by electrophoretic mobility shift assays that LurR binds to three sites in the upstream region of DVU3025, the first gene of the operon. NrfR, a response regulator that is activated under nitrite stress, and LurR share similar binding site motifs and bind the same sites upstream of DVU3025. The DVU3025 promoter also has a binding site motif (Pho box) that is bound by PhoB, a two-component response regulator activated under phosphate limitation. The lactate utilization operon, the regulator LurR, and LurR binding sites are conserved across the order Desulfovibrionales whereas possible modulation of the lactate utilization genes by additional regulators such as NrfR and PhoB appears to be limited to D. vulgaris.


Introduction
Sulfate-reducing bacteria such as Desulfovibrio play an important role in anaerobic microbial communities in groundwater sediments and reduce toxic heavy metals such as chromium (VI) and uranium (VI) in nuclear waste-contaminated sites [1,2]. These bacteria use organic compounds such as lactate, ethanol, and pyruvate as electron donors for the reduction of sulfate [3]. Addition of lactate compounds to contaminated wells can stimulate growth of the anaerobic microbial community and promote heavy metal reduction [4]. Desulfovibrio vulgaris a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

LurR is an activator and is required for optimal growth on lactate
Although a transposon insertion was not obtained in lurR during the construction of a transposon mutant library [12], we were successful in constructing a targeted deletion of lurR (a bar-coded deletion of DVU3023, BCD1228). We compared the ΔlurR mutant with the D. vulgaris parent strain, JW710 (hereafter called WT), during growth with either lactate or pyruvate as carbon and electron donor, and sulfate as the electron acceptor. The ΔlurR strain displayed a longer lag phase during growth on lactate, and lower final cell densities (Fig 2A). The ΔlurR showed a lag in lactate consumption that corresponded to its longer lag phase (Fig 2B). The mutant also stopped consuming lactate when the cells entered stationary phase ( Fig 2B). Consequently, the mutant produced less acetate and more slowly than WT. The incomplete utilization of lactate could explain the lower final cell densities seen in the mutant. When we complemented the deletion (strain JW9388), the differences in growth lag and lactate consumption disappeared (Fig 2A and 2B). When the strains were grown on pyruvate as carbon and energy source, there were no differences in growth, pyruvate utilization, or acetate production among the three strains (Fig 2C and 2D). between Lawrence Berkeley National Laboratory and the U. S. Department of Energy. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paidup, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes.
Next, we examined the expression of por and ack genes in the ΔlurR mutant relative to the wild-type strain by RT-qPCR. Deletion of lurR reduced (~2-fold) the expression of por and ack, when the strains were grown on lactate (Fig 3). WT and ΔlurR expressed por and ack Lactate is transported inside the cell by a lactate permease (DVU3026, llp). D-(DVU3027-3028, dld-II) and L-lactate dehydrogenases (DVU3032-3033, lldGH) oxidize lactate to pyruvate [6]. Pyruvate is then oxidatively decarboxylated to acetyl-CoA via a pyruvate ferredoxin oxidoreductase (DVU3025, por) [7,8]. Acetyl-CoA is then oxidized to acetate in two steps by phosphotransacetylase (pta) and acetate kinase (ack) enzymes [8]. The operon also contains pta-N gene (DVU3031) encoding the N-terminal domain of phosphotransacetylase, whose function is unknown. B. The DVU3025-3033 lactate utilization operon and the associated lurSR two-component system are conserved across Desulfovibrio and related species (also see S1 Table). Gene numbers are indicated above the gene. Genes are color coded according to the key. Gut isolates such as D. piger have a highly reduced operon and lack lurSR. C. DAP-chip revealed a regulatory network where four response regulators-LurR, NrfR, PhoB, and DVU0539 -target the DVU3025-3033 genes (colored circles-see key) [9]. Other gene targets are shown in grey circles, and arrows indicate regulatory interactions between an RR and its target. Figure generated using Cytoscape [10]. similarly when grown on pyruvate (Fig 3). LurR also targets two putative lactate permease genes DVU2451 and DVU3284 [9]. Deletion of lurR strongly affected the expression of DVU2451, reducing it by~33-fold when grown on lactate, and by~4.5-fold when grown on pyruvate (Fig 3). Thus, our data show that LurR activates DVU3025-3033 and DVU2451, and this activation seems to be primarily in response to lactate. When we reintroduced the lurR gene into the ΔlurR strain (JW9388), expression of por, ack, and DVU2451 on lactate-sulfate were restored to WT levels. However, for unknown reasons, the complementation decreased the expression of por and ack genes during growth on pyruvate where the lurR deletion itself had no effect. Transcript levels of the second lactate permease gene, DVU3284, were unchanged in ΔlurR on both lactate and pyruvate medium.

Binding of LurR, NrfR and PhoB upstream of por
The upstream region of por has a predicted σ 54 -dependent promoter (-136 bp from start codon), three binding site motifs for LurR (at -389, -350 and -250 bp from the start codon), and one binding site for PhoB (at -183 bp from the start codon). Previously, we had validated the 16 bp binding site motif for LurR using EMSAs with wild-type and substituted sites [9]. We had also validated the PhoB binding site upstream of the phoB gene [9], and the NrfR binding sites upstream of the nrfHA genes [15].
Here, we used truncated por promoter substrates containing 0, 1, 1.5 or 3 LurR binding sites in EMSAs. LurR bound all substrates except the promoter fragments (IV and V) lacking any binding motif (Fig 4). NrfR shifted the same fragments as LurR (Fig 4). PhoB shifted all the fragments except for fragment V that does not have a PhoB binding site (Fig 4).
The consensus 16 bp NrfR binding site motif [15] is very similar to the consensus LurR binding motif (Fig 5A). To test if NrfR recognizes the LurR binding site, we performed EMSAs with NrfR and a short substrate containing the LurR binding site found upstream of por ( Fig  5B). NrfR shifted this substrate. Neither LurR nor NrfR bound an altered substrate with substitutions in the conserved bases, thus indicating that NrfR recognized the same site in the por promoter as LurR ( Fig 5B). To test if PhoB specifically recognizes the predicted site upstream of por, we used a short oligomer DNA substrate containing this predicted PhoB binding site upstream of por in an EMSA with purified His-tagged PhoB ( Fig 5C). We observed that PhoB shifted the DNA substrate, and this shift was eliminated when substitutions were made to the conserved positions within the motif.
LurR, NrfR and PhoB did not require phosphorylation for in vitro DNA binding. Twocomponent response regulators are activated by phosphorylation by the partner histidine kinase in vivo, but in vitro DNA binding often occurs without activation [9,[16][17][18][19]]. Since we lacked purified cognate histidine kinases to phosphorylate the respective RRs, we tested the addition of the small molecule phospho-donor acetyl phosphate in the binding reactions. We did not observe any effect of acetyl-phosphate on DNA binding (not shown). It is also possible that the RRs were purified with some phosphorylation since they were heterologously expressed in Escherichia coli. A high amount of PhoB was required for the EMSA reaction, and this may reflect the lack of sufficient activation.

Effect of NrfR and PhoB on gene expression
We examined por gene expression in response to nitrite stress in the wild-type and a nrfR transposon insertion mutant by exposing mid-log cultures to 2.5 mM sodium nitrite for 30 minutes [15]. However, we did not observe significant differences in por gene expression (not shown).
To observe a physiological effect of PhoB, we first confirmed experimentally using a phoB transposon insertion mutant that PhoB activates the high-affinity phosphate transport genes when D. vulgaris is subjected to phosphate limitation (S2 Fig). However, it proved challenging to assign reliable changes in por gene expression by RT-qPCR in a phoB mutant. The phoB mutant also had a growth defect (S2 Fig) that further hindered collection of adequate cell samples at comparable growth phases.

Conservation of DVU3025-3033 and lurSR
We examined 54 sequenced Desulfovibrio genomes available on the IMG (Integrated Microbial Genomes [21]) website, and other closely related species in the orders Desulfovibrionales and Desulfobacterales for the presence of orthologs of lurR and the lactate utilization operon (S1 Table). In most Desulfovibrio genomes, the lurSR genes are encoded proximal to the orthologous DVU3025-3033 operon (S1 Table). The lurSR genes are absent in the human microbiome-and rumen-associated Desulfovibrio isolates (6 sequenced genomes), and in three environmental Desulfovibrio species (D. cuneatus, D. litoralis, and D. desulfuricans DSM 642). Genomes lacking lurSR also lacked an organized lactate utilization operon. In the Desulfovibrionaceae family, the lurSR genes are also present in the five Pseudodesulfovibrio genomes, the four Halodesulfovibrio genomes, and Desulfocurvus vexinensis, but are absent in the humanassociated Bilophila and Lawsonia species (S1 Table). The lurSR genes are also represented in the Desulfomicrobiaceae (7/7 genomes), Desulfohalobiaceae (2/8 genomes) and Desulfonatronaceae (5/5 genomes) families. In the order Desulfobacterales, homologs of lurSR and the associated lactate utilization operon were found in a few genomes (S1 Table).
We searched for the binding site motif upstream of the first gene in the lactate utilization operons in the other lurR-encoding genomes. Most genomes queried had 1 to 4 binding site motifs upstream of the first gene (por in most cases, or the lactate permease gene), with the majority of the upstream regions having three binding sites as seen in D. vulgaris (S1 and S2 Tables).

Discussion
In this study, we show that the LurR regulator strongly activates the lactate permease gene DVU2451 and moderates the central carbon utilization operon DVU3025-3033 in the model sulfate reducer D. vulgaris Hildenborough. Activation by LurR depends on the presence of lactate, since deletion of lurR affected the expression of the lactate utilization operon during growth on lactate-sulfate, but not on pyruvate-sulfate. The absence of lurR also affected the expression of DVU2451 more strongly on lactate-sulfate than on pyruvate-sulfate (~8-fold difference). Even though the expression of DVU3025-3033 was reduced by only 2-fold in the absence of lurR, growth and substrate uptake were affected-compared with the WT strain, the lurR mutant had a longer lag phase accompanied by a lag in lactate uptake, and lower final cell densities resulting from incomplete lactate utilization. There was no effect on growth or substrate uptake when the strains were grown with pyruvate as the electron donor. Other studies corroborate the key role of LurR in lactate consumption. Transposon insertions in the lurSR genes of D. alaskensis G20 and D. vulgaris Miyazaki had fitness defects during growth on lactate-sulfate [13,14]. In the absence of sulfate as an electron acceptor, D. vulgaris can also grow on lactate in syntrophic association with hydrogen-consuming methanogens. Regulation by LurR may also have a bearing on the upregulation of the lactate utilization operon during coculture with Methanococcus maripaludis [22]. High levels of expression of this operon may be reflective of three genes in the operon-por, pta, and ack-being required for growth on pyruvate as well. D. vulgaris can also utilize molecular hydrogen and formate as electron sources [3], and it is possible that lurSR has evolved to express the lactate utilization operon optimally in response to electron and carbon sources. The expression of the lactate utilization operon was altered during several stress conditions such as alkaline stress [23], salt stress [24], salt adaptation [25], air exposure [26], heat shock [27], and peroxide stress [28], where transcriptomics measurements are reported. Since the lactate utilization operon has binding sites for multiple response regulators and two sigma factors, and possibly other transcription factors, the regulation of this operon could be more than a simple activation by LurR alone.
To our knowledge, LurR is both the first two-component response regulator and the first σ 54 -dependent regulator reported to activate lactate consumption. While we have not studied the sensor kinase LurS, domain predictions show that it is a large hybrid kinase with two transmembrane regions, three PAS domains, the histidine kinase domains and a C-terminal receiver domain. Reported regulators that modulate the lactate dehydrogenase and lactate permease genes include FadR family transcription factors in E. coli [29], Corynebacterium glutamicum [30], Pseudomonas aeruginosa [31], and Bacillus subtilis [32], and LysR-type regulators in Shewanella [33] and Vibrio species [34]. However, D. vulgaris has multiple lactate dehydrogenases and permeases [6,35] that likely have their own regulation. For instance, DVU2875 is a FadR family regulator LldR that is predicted to activate the L-lactate dehydrogenase and lactate permease genes in the same operon (DVU2874-3) [36]. The L-lactate dehydrogenase genes DVU1781-83 and DVU2784 are expressed at similar levels to DVU3032-3033. Expression of some of the alternate D-and L-lactate dehydrogenase genes were variably altered in response to deletions of either dld-II or lldFG [6].
DAP-chip with LurR revealed DVU3284, a third putative lactate permease, also to be a target gene [9]. However, we did not observe changes in expression of this gene under our conditions. This may be due to DVU3284's role being pertinent in stationary phase growth [37]. The lack of an endogenous control gene in stationary phase made it challenging to test expression changes in this growth phase (the rpoH and rpoD transcripts used in the exponential phase samples were differentially expressed during stationary phase). The other two permeases DVU3026 and DVU2451 are highly expressed [11], and interestingly, DVU2451 is 88% identical to DVU3026, and in a DVU3026 deletion mutant, DVU2451 expression is increased [6]; thus, they are likely serving redundant roles. The DVU2451 gene is also under the control of a sigma54-dependent promoter [9,11]. LurR has two binding sites at -401 (GTCTGCAATGTCG GAC) and -310 (GTCCATTTTTTCAGAC) upstream of DVU2451 [9].
DAP-chip studies also suggest that three other regulators targeted the lactate utilization operon. Here, we investigated the roles of two of these regulators-NrfR and PhoB. NrfR is also a σ 54 -dependent regulator, and we showed here that NrfR and LurR have very similar consensus binding sites, and that NrfR can bind to the LurR binding motif upstream of por. Thus, any regulation of the lactate utilization operon by NrfR likely utilizes the same binding sites as LurR. Although higher concentrations of NrfR were required to shift the por upstream DNA as compared with LurR in our in vitro EMSAs, it is challenging to extrapolate these observations to in vivo protein stoichiometries. In an earlier genome-wide transcriptomics study on nitrite stress in D. vulgaris, the lactate utilization genes increased by 1.5-2-fold in expression with 2.5 mM nitrite stress added at mid-log and following 4 hours of this exposure [38]. We examined gene expression changes after 30 min of nitrite exposure but did not observe any expression changes in the por gene under nitrite stress. Interestingly, few genomes other than D. vulgaris (including strains DP4 and Miyazaki) encode both lurSR and nrfSR genes. All the available host gut-associated Desulfovibrionales genomes (including Bilophila and Lawsonia strains) that lack lurSR carry the nrfSR genes [15]. Thus, possible regulation of lactate utilization by NrfR could be unique to D. vulgaris.
We confirmed the role of PhoB in responding to phosphate limitation. We showed that PhoB recognizes the predicted Pho box motif present upstream of the por gene. The location of the pho box upstream of the promoter region suggests that PhoB would be an activator for this operon (usually transcription factor binding sites for activators occur upstream of the promoter while for repressors occur downstream of the promoter [39]). Since the lactate utilization operon contains the ack and pta genes, it seems reasonable that the phosphate availability also modulates the operon. The lurSR genes of D. alaskensis G20 play a role in survival after 15 days of phosphate starvation as seen by fitness assays with transposon mutant libraries [40]. However, similar to our observations with NrfR and compounded by the growth defect in the phoB mutant, we could not get reliable expression changes of the por gene under phosphatelimiting conditions. We were also unable to find PhoB binding motifs upstream of the lactate utilization genes in lurR-encoding genomes other than the very similar D. vulgaris DP4.
Therefore, similar to NrfR, any possible regulation of the lactate utilization operon by PhoB may be limited to D. vulgaris.
Our study highlights the utility of conducting genome-wide assays to query regulator-target interactions. Studying one regulator-target gene interaction at a time may not have revealed the complexity of regulation of critical operons such as the lactate utilization regulon. Our initial genome-wide DAP-chip experiments led us to investigate and discover the cross-talk between the NrfR and LurR regulators. More recently, DAP has been combined with sequencing technologies to enable high-throughput elucidation of regulatory networks [19,41]. Our work here shows that activation of the lactate utilization operon by LurR is conserved across the Desulfovibrionales but fine-tuning of the regulation of this operon in response to other signals/stresses may be unique to each species.
Construction of D. vulgaris BCD1228 (ΔlurR) and JW9388 (ΔlurR complemented) strains. The deletion strain BCD1228 (deleted for DVU3023) was constructed similarly as other deletion strains of D. vulgaris Hildenborough [44]. Briefly, strain JW710 was grown in 20 ml of MOYLS4 [45] overnight. Then, 30 ml of MOYLPS4S3 (30 mM lactate, 30 mM pyruvate, 20 mM sulfate and 10 mM sulfite) was added to the culture and allowed to grow for 3-5 hours. Cells were harvested and transformed by electroporation with pBCD-DVU3023, a plasmid that does not replicate in D. vulgaris containing a construction that causes a marker exchange deletion of DVU3023 (ΔDVU3023::P KmR Km r -upp). Transformed cells were allowed to recover overnight in MOYLPS4S3 and were then plated in MOYLPS4S3 containing the aminoglycoside G418 (400 μg/ml). Putative transformants were screened for appropriate antibiotic phenotypes (sensitive to 100 μg spectinomycin/ml, confirming the loss of the plasmid sequences, and resistant to G418, confirming retention of Km r ). Cell lysates from isolates that passed the initial screen were used as template in a PCR reaction with primers (S3 Table) spanning both sides of the gene targeted for deletion to ensure that the correct double-homologous recombination event occurred (S3 Fig). The expected upstream product of 1057 bp from primers 1 and 2 and the downstream product of 1747 bp from primers 3 and 4 were observed, and these products were sequenced to confirm retention of the wild-type sequences flanking the deleted gene.
The complementation strain JW9388 was constructed by reintroduction of the deleted gene on a plasmid unable to replicate in D. vulgaris, pMO9380, through antibiotic selection of a double-homologous recombination event. The plasmid was constructed by the Sequence-Ligation-Independent Cloning (SLIC) method [46] with PCR products obtained with the primers (IDT, Coralville, IA) found in S3 Table and the Herculase II DNA polymerase (Stratagene, La Jolla, CA). The plasmid contained DVU3023 and the spectinomycin-resistant gene flanked by the upstream and downstream regions. It was electroporated into strain BCD1228 by previously described methods [45]. Transformed cells were recovered, plated and screened as previously described except that isolates resistant to spectinomycin were selected for further confirmation. Putative complementation isolates were confirmed by Southern blot analysis by digesting genomic DNA with EagI (New England Biolabs, Ipswich, MA) and probing with the labeled upstream PCR fragment (Prime-It RmT Random Labeling Kit, Stratagene, La Jolla, CA).
Determination of lactate, pyruvate, and acetate. WT, ΔlurR, and ΔlurR complemented strains were grown on LS4D or PS4D in 40 ml cultures at 30˚C. At specific time points, aliquots were withdrawn for OD 600 measurements on a spectrophotometer, and another 800 μl of the culture was withdrawn and stored at -20˚C. For HPLC analysis, the samples were thawed, cells were spun down, and the supernatant was centrifuged through a 10 kDa microspin filter at 14500 x g for 5 min. Samples were analyzed by an Agilent 1100 Series HPLC system (Agilent Technologies, Santa Clara, CA) equipped with an Aminex HPX-87H ion-exclusion column (300 x 7.8-mm; Bio-Rad Laboratories Inc., Hercules, CA) 9 μ with guard at 50˚C with an injection volume of 5 μl, and a flow rate of 0.6 ml/min with 4 mM H 2 SO 4 as solvent. Na-lactate, Na-pyruvate, and Na-acetate standards were used to determine unknown concentrations.
RNA extraction. Three independent experiments on different days were set up where WT, ΔlurR, and ΔlurR complemented strains were each grown in triplicate 40-ml cultures of LS4D or PS4D and 1.5 ml aliquot of cells were harvested at mid-log phase (OD 600~0 .3), frozen in liquid nitrogen and stored at -80˚C. RNA was extracted from the pellets with Agilent Total RNA mini kit following the manufacturer's protocol (Agilent Technologies, Santa Clara, CA). Genomic DNA contamination was removed by digesting the RNA samples with Turbo DNAfree DNase kit (Life Technologies, Grand Island, NY). The integrity of the RNA was examined visually by agarose gel electrophoresis. RNA was quantified spectrophotometrically on the Nanodrop ND-1000 (Thermo Fisher Scientific, Wilmington, DE).
RT-qPCR. 500 ng of RNA was reverse transcribed with the iScript RT mix (BioRad, Hercules, CA, USA). The resulting cDNA was diluted ten-fold and 2 μl was used as the template for qPCR reactions. Primers for qPCR reactions were designed using Primer3 (http://bioinfo. ut.ee/primer3/) and are in S3 Table. qPCR reactions were set up in triplicate, with each 20 μl reaction containing 1X SsoAdvanced Universal Sybr Green Supermix (BioRad, Hercules, CA), 500 nM of each primer, and 2 μl of cDNA in 96-well plates (white, BioRad). Reactions were cycled on a CFX96 Real-Time PCR Detection System (BioRad, Hercules, CA) as follows: 98˚C/2 min, and 40 cycles of 98˚C/10s, 60˚C/30 s. RNA integrity was assayed by calculating the relative quantities of the por transcript produced by primers amplifying the 3' vs. the 5' end of the gene (ratios close to 1 indicate intact RNA; all ratios were less than 1.2). Expression changes were normalized to that of two σ 70 reference genes-rpoH and rpoD.
Electrophoretic mobility shift assay. To determine the minimal promoter region of the lactate utilization operon required for binding, DNA substrates were prepared by PCR amplification with a common biotinylated reverse primer and different forward primers (See S3 Table for primer sequences). DNA substrates to test binding site motif predictions were prepared by annealing biotinylated oligonucleotides with their unlabeled complementary strands as described previously [9]. LurR (~57 kDa), NrfR (~56 kDa) and PhoB (~30 kDa) proteins were purified with a C-terminal V5-epitope and a 6X-His tag as described previously [9]. Each protein was mixed with 100 fmol of biotinylated DNA substrate in 10 mM Tris HCl pH 7.5, 50 mM KCl, 5 mM MgCl 2 , and with 1 μg/ml poly dI.dC in a total reaction volume of 20 μl and incubated at room temperature for 20 min. Electrophoresis, blotting and chemiluminescent detection were performed as described previously [9]. Final imaging of the blot was done using the Fluor Chem Q system (Protein Simple, Santa Clara, CA, USA).  Transposon insertion in phoB results in a growth defect. Growth of WT and phoB::mini-Tn5, at 5 and 10% (vol/vol) inoculum sizes on LS4D, monitored using OD 600 measurements on a Bioscreen C instrument. The growth defect of phoB::mini-Tn5 was more pronounced at lower starting cell densities. Data are average of five replicates, and error bars indicate standard deviations. B. RT-qPCR measurements of the fold change in expression of the phosphate transport gene pstS (DVU2477) normalized to reference gene rpoH in WT and phoB::mini-Tn5 under phosphate-limiting relative to phosphate-replete conditions. The two strains were grown in LS4D modified to contain 0.1 mM K 2 PO 4 and 2 mM KCl until mid-log phase. The cells were spun down gently (3000 x g for10 min), and resuspended either in LS4D with 2 mM KCl and 0 mM K 2 PO 4 (phosphatelimiting) or with 2 mM K 2 PO 4 (phosphate-replete) and allowed to grow for 1.5 h. Error bars represent the range of fold change as calculated using the standard deviations in the ΔΔC T values (n = 3).