The copper-responsive regulator CsoR is indirectly involved in Bradyrhizobium diazoefficiens denitrification

Abstract The soybean endosymbiont Bradyrhizobium diazoefficiens harbours the complete denitrification pathway that is catalysed by a periplasmic nitrate reductase (Nap), a copper (Cu)-containing nitrite reductase (NirK), a c-type nitric oxide reductase (cNor), and a nitrous oxide reductase (Nos), encoded by the napEDABC, nirK, norCBQD, and nosRZDFYLX genes, respectively. Induction of denitrification genes requires low oxygen and nitric oxide, both signals integrated into a complex regulatory network comprised by two interconnected cascades, FixLJ–FixK2–NnrR and RegSR–NifA. Copper is a cofactor of NirK and Nos, but it has also a role in denitrification gene expression and protein synthesis. In fact, Cu limitation triggers a substantial down-regulation of nirK, norCBQD, and nosRZDFYLX gene expression under denitrifying conditions. Bradyrhizobium diazoefficiens genome possesses a gene predicted to encode a Cu-responsive repressor of the CsoR family, which is located adjacent to copA, a gene encoding a putative Cu+-ATPase transporter. To investigate the role of CsoR in the control of denitrification gene expression in response to Cu, a csoR deletion mutant was constructed in this work. Mutation of csoR did not affect the capacity of B. diazoefficiens to grow under denitrifying conditions. However, by using qRT-PCR analyses, we showed that nirK and norCBQD expression was much lower in the csoR mutant compared to wild-type levels under Cu-limiting denitrifying conditions. On the contrary, copA expression was significantly increased in the csoR mutant. The results obtained suggest that CsoR acts as a repressor of copA. Under Cu limitation, CsoR has also an indirect role in the expression of nirK and norCBQD genes.


Introduction
Rhizobia are soil bacteria with the unique ability to establish symbiosis with their host legume through the formation of nodules in their r oots, wher e they differ entiate into bacter oids, whic h ar e able to synthesize the enzyme responsible for the symbiotic dinitrogen (N 2 ) fixation, the nitrogenase (Mahmud et al. 2020 ). Given the sensitivity of this enzyme to oxygen (O 2 ), the steady-state concentr ation within r oot nodules is typicall y in the tens of nanomolar, a ppr oximatel y four orders of magnitude lo w er than equilibrium le v els in the external environment (Udvardi and Poole 2013 ). Under certain conditions, such as flooding or nitrate (NO 3 − ) excess in the rhizospher e, bacter oids can make use of other inorganic terminal electron acceptors such as NO 3 − or nitrite (NO 2 − ), which are reduced to N 2 through the denitrification pathway, producing nitric oxide (NO) and nitrous oxide (N 2 O) as intermediates (r e vie wed by Salas et al. 2021 ). At the cellular le v el, NO acts as a k e y signal molecule at low concentrations, but is a potent cytotoxic compound at high concentrations. N 2 O is the dominant nitrogenous greenhouse gas in the atmosphere that has a warming capacity 296 times greater than that of CO 2 due to its r adiativ e ca pacity, and is responsible for 6% of current global warming. Because of its long atmospheric lifetime (120-150 years), it can be converted by photolysis in the stratosphere into NO, causing ozone layer depletion (Erisman et al. 2015 ).
Bradyrhizobium diazoefficiens , the soybean endosymbiont, is considered as a model to study rhizobial denitrification because, in addition to fix N 2 , is the only species able to grow under anoxic conditions with NO 3 − as sole nitrogen sour ce b y the complete denitrification pathway, which has been deeply characterized under both free-living and symbiotic conditions (reviewed by Salas et al. 2021 ). This bacterium possesses the whole set of napEDABC , nirK , norCBQD , and nosRZDFYLX denitrification genes, which encode a periplasmic nitrate reductase system (Nap), a Cucontaining nitrite reductase (NirK), a c -type nitric oxide reductase system (cNor), and a nitrous oxide reductase (Nos), respectively. As in many other denitrifiers, the expression of these genes in B. diazoefficiens r equir es O 2 limitation and the presence of NO 3 − or another N oxide deriv ed fr om its reduction (reviewed by Torres et al. 2016, Salas et al. 2021. In this context, this bacterium detects the low O 2 signal via tw o inter connected regulatory cascades: FixLJ-FixK 2 -NnrR and RegSR-NifA (r e vie w ed b y Bueno et al. 2012, Salas et al. 2021 (Fig. 1 ). When a moderate decrease in O 2 ( ≤ 5% in the gas phase) occurs, the two-component regulatory system, FixLJ, activ ates fixK 2 tr anscription (Sciotti et al. 2003 ). Then, FixK 2 pr otein induces nap , nirK , and nos gene expression, as well as rpoN 1 and nnrR genes among others (Mesa et al. 2008. The product of the latter gene, NnrR, is the F igure 1. Regulatory netw ork of denitrification genes in B. diazoefficiens in response to O 2 and NO. The r egulatory pr otein FixK 2 is the dir ect activ ator of nap , nirK , and nos genes in response to low O 2 conditions ( ≤ 5% O 2 ). This regulator also activates nnrR , which encodes the transcription factor NnrR, necessary for induction of nor genes in response to NO. FixK 2 is also the link to the RegSR-NifA cascade responsible for activating the nif and fix genes in response to very low O2 conditions ( ≤ 0.5% O 2 ). dir ect r egulator of the norCBQD genes in r esponse to NO , Jiménez-Leiva et al. 2019. The second O 2 -r esponsiv e r egulatory cascade, RegSR-NifA, r esponds to lo w er O 2 concentrations . In this cascade , the two-component regulatory system RegSR induces the expression of nifA . When O 2 concentration substantiall y decr eases ( ≤ 0.5% in the gas phase), NifA induces the expr ession of nitr ogen-fixation genes ( nif and fix ), among others ( Fig. 1 ).
Besides O 2 and NO, copper (Cu) is an emerging candidate in the regulation of denitrification, since it is an essential cofactor in two enzymes involved in this process, NirK and NosZ. Pacheco et al. ( 2022 ) r ecentl y pr oposed that Cu r epr esents an essential factor not only for denitrifying enzymatic activities, but also for the regulation of gene expression, as well as protein synthesis and matur ation. P articularl y, the expr ession of nirK , nor , and nos genes, but not nap genes diminished under Cu-limitation in cells grown under NO 3 − -amended O 2 -limiting conditions, which suggests the involvement of a specific Cu-responsive regulator in this contr ol. In bacteria, v arious types of Cu-sensing tr anscriptional r egulators hav e been c har acterized, suc h as CopY in Esc heric hia coli and Enterococcus hirae or CsoR in Mycobacterium tuberculosis (review ed b y Rademacher and Masepohl 2012 ). In the latter, the csoR gene constitutes an operon together with other two genes: rv0968 , encoding a hypothetical protein annotated as DUF1490, and rv0969 ( ctpV ), which codes for a Cu + -ATP ase tr ansporter denoted as CtpV, pr esumabl y involv ed in Cu excr etion (Liu et al. 2007 ). CsoR and CtpV homologs have also been described in the rhizobial species Bradyrhizobium liaoningense CCNWSX0360, where their involvement in heavy metal-response regulation has been suggested. In this bacterium, CtpV is denoted as CopA (Liang et al. 2016 ).
In the present work, a gene encoding a protein from the CsoR famil y of r egulators and another r esponsible for the synthesis of the Cu + -ATP ase tr ansporter, CopA, wer e unv eiled in the B. diazoefficiens genome. To investigate the possible involvement of CsoR in the regulation of denitrification, a mutant strain was constructed by deletion of the csoR gene . T his strain was further used in growth and gene expression analyses . T he results obtained suggest that CsoR is not essential for bacterial growth, but may play an indir ect r ole in the Cu-mediated control of nirK and norC expression. Ther efor e, this work assesses a novel role of CsoR in B. diazoefficiens denitrification.

Bacterial strains and growth conditions
All the strains used in this work are listed in Table 1 . To generate the csoR deletion m utant, upstr eam (600 bp) and downstream (480 bp) DNA fr a gments flanking Bdiaspc4_RS03270 were amplified by PCR using csoR _Up_For_ Xba I/ csoR _Up_Rev_ Bam HI and csoR _Do wn_For_ Bam HI/ csoR _Do wn_Rev_ Eco RI primer pairs ( Tabl e S1, Supporting Information ). The 1080-bp fr a gment was inserted into pK18 mobsacB (Schäfer et al. 1994 ), yielding plasmid pDB4030. After corr obor ating the integrity of the cloned fr a gment by sequencing using the primers pSRKC1_F, pK18_4, and the internal primer csoR _IN_For ( Table S1, Supporting Information ), pDB4030 was conjugated with B. diazoefficiens 110 spc 4 obtaining the markerless csoR deletion m utant str ain Bd4030 (denoted as csoR throughout the manuscript) after a double recombination e v ent.
Esc heric hia coli cells were cultured in Luria-Bertani medium (Miller 1972 ) at 37 • C. The following antibiotics were added to the medium at the following concentrations ( μg ml −1 ): stre ptom ycin, 25; spectinom ycin, 25; kanam ycin, 25; and tetracycline, 10. Bradyrhizobium diazoefficiens cells were grown routinely under oxic conditions at 30 • C in PSY complete medium (Mesa et al. 2008 ) to obtain cellular mass. Subsequent experiments under microoxic conditions were carried out using a vitamin-free modified Vincent's minimal medium (BVM; Vincent 1970, Becker et al. 2004 amended with 10 mM KNO 3 (BVMN). This medium was supplemented (per litre) with 20 mM l -arabinose and 1 ml from a mineral solution originally developed by Bishop et al. ( 1976 ). Final pH was adjusted around 6.8 with 2 M NH 3 .
The specific Cu concentrations assayed in this study are extensiv el y described in P ac heco et al. ( 2022 ). Briefly, the standard Cu concentration of the BVMN medium (0.02 μM) is denoted as Cu-S. A concentration of 13 μM is r eferr ed in this manuscript as high Cu, denoted as Cu-H. Finally, the Cu-limiting medium (denoted as Cu-L) was ac hie v ed by omitting CuSO 4 ·5H 2 O from the mineral solution, and by adding 1 mM l -ascorbate (Cu(II) r educing a gent) and 10 μM bathocupr oine disulfonic acid (BCS) [Cu(I) c helating a gent] in order to decr ease Cu av ailability. In the case of Cu-L medium, glassware was treated with 0.1 M HCl and rinsed afterw ar ds with double-distilled w ater (Serventi et al. 2012 ).
For microoxic conditions, 17-ml rubber-stoppered tubes or 250 or 500-ml rubber-stoppered flasks were filled with 3 ml, 5 ml, or 100 ml, r espectiv el y, of BVMN (1:5 ratio between air and liquid phases). Then, these tubes or flasks were flushed at the starting point of the incubation with a gas mixture consisting of 2% (v/v) O 2 and 98% (v/v) N 2 , and were incubated at 30 • C with agitation at 170 rpm.

Determination of β-galactosidase activity and protein concentr a tion
β-Galactosidase activity le v els wer e anal ysed by using permeabilized cells from at least three independent cultures, assayed in triplicate for each strain and condition, as described by Cabr er a et al. ( 2016 ). Specific activity was calculated in Miller Units (Miller 1972 ). Protein concentration was estimated by using the Br adford r ea gent (Bio-Rad, CA, USA) (Br adford 1976 ).

Intracellular copper determination
Cu concentration was analysed in the cells using the Inductiv el y Coupled Plasma Optical Emission Spectrometer (ICP-OES) available at the Instrumental Technical Service of EEZ (Granada, Spain). Befor e measur ements, cell samples wer e miner alized by micro w ave digestion in the presence of the acidic HCl-HF (1/1, v/v) mixtur e. Data wer e expr essed as μg Cu mg pr otein −1 .

Quantitati v e real-time PCR analysis
Bradyrhizobium diazoefficiens cells grown for 48 h under microoxic conditions in Cu-L, Cu-S, or Cu-H BVMN medium wer e harv ested. Then, total RNA isolation and cDNA synthesis were performed according to Hauser et al. ( 2007 ), Mesa et al. ( 2008 ), and Lindemann et al. ( 2007 ). The subsequent analysis of napE , nirK , norC , and nosR expression was carried out by qRT-PCR using a QuantStudio 3 Real-Time PCR system (T hermo-Fisher Scientific , MA, USA). Primers for the PCR reactions were previously designed ( Table S 1, Supporting Information ). Each PCR reaction contained 9.5 μl of iQTM SYBR Green Supermix (Bio-Rad), 2 mM (final concentration) of individual primers, and a ppr opriate dilutions of different cDNA samples, obtaining a total volume of 19 μl per well. Reactions were run in triplicate and melting curves were generated in order to verify the specificity of each amplification. Finally, relativ e c hanges in gene expr ession wer e calculated according to the method described by Pfaffl ( 2001 ). 16S rrn expression was used as r efer ence for normalization (see primers in Table S1, Supporting Information ).

Sta tistical anal ysis
The total number of replicates appears in each Figure. Inferential statistics were performed by the application of a parametric ANOVA for unpair ed tr eatments. Next, a post hoc Tuk e y HSD test at P ≤ .05 with SPSS (v27) software was performed.

Identification of a putati v e csoR gene in B. diazoefficiens
The analysis of B. diazoefficiens 110 spc 4 genome unveiled a small open reading frame (ORF) annotated as Bdiaspc4_RS03270, encoding a CsoR-like Cu-sensing transcriptional repressor. This ORF is inv ersel y oriented and located immediatel y adjacent to copA (Bdiaspc4_RS03265), a gene encoding a P-type Cu + transporter (Fig. 2 A). In the 3 end of copA , we found nikR , whic h putativ el y encodes a nickel (Ni) transporter, and hmgL , encoding a possible h ydroxymeth ylglutaryl-CoA lyase (Fig. 2 A). This genetic arrangement is similar to that observed in B. liaoningense CCNWSX0360 (Fig. 2 B) (Liang et al. 2016 ), and M. tuberculosis H37Rv (Fig. 2 C) (Liu et al. 2007 ). As shown in Fig. 2 (B) and (C), a (G + C)-rich pseudopalindromic sequence of about 19 bp in B. liaoningense and 28 bp in M. tuberculosis was identified within the promoter region of csoR . These sequences correspond presumably to the CsoR-binding site, which has been called the CsoR box. In both species, this box encompasses inv erted r epeats coinciding with the −35 or −10 promoter elements (Fig. 2 B and C). The 6-bp inverted repetitions shown in B. diazoefficiens (Fig. 2 A) suggest the presence of a putative 17-bp CsoR box.
By using KEGG database, we found that Bdiaspc4_RS03270 encodes a protein of 91 amino acids . T he deduced amino acid sequence of this protein was aligned against other proteins from the CsoR family ( Figure S1 (Liu et al. 2007 ), and Thermus thermophilus (Sakamoto et al. 2010 ) were also included. The CsoR sequence alignment for B. diazoefficiens 110 spc 4 sho w ed 100%, 97.80%, 96.70%, and 91.21% identity with the CsoR-like protein sequence from B. diazoefficiens USDA110, B. japonicum , B. liaoningense , and R. palustris , r espectiv el y ( Table S2, Supporting Information ), suggesting that these proteins are orthologs and that the CsoR sequence from B. diazoefficiens is absolutely conserved between 110 spc 4 and USDA110 str ains. Mor eov er, the sequence of B. diazoefficiens CsoR shared identity with the sequences from other species including some denitrifiers ( Figure S1 and Table S2, Su pporting Information ). The predicted structure of the B. diazoefficiens CsoR by AlphaFold (neur osna p.ai, 15/07/2023) as homodimer, sho w ed high similarity with that from M. tuberculosis and T. thermophilus solved by Liu et al. ( 2007 ) and Sakamoto et al. ( 2010 ), r espectiv el y. The model indicates the presence of three αhelices per monomer, with compatible Cu(I) binding sites (C-H-C motif) predicted between C33 of one monomer and H58 and C62 from the other ( Figure S2, Supporting Information ). These Data expressed as μg Cu mg protein −1 , are means with standard de viation fr om at least two independent cultur es assa yed in triplicate . r esidues ar e conserv ed in the majority of the sequences analysed in Figure S1 (Supporting Information) , except for E. coli , which displays a C-H-R motif, and T. thermophilus and P. pantotrophus , with C-H-H motifs ( Figure S1, Supporting Information ). Moreover, R12, Y32, R49, and E81 proposed in other CsoR proteins necessary for protein binding to DNA (Liu et al. 2007 ) are conserved in B. diazoefficiens ( Figure S1, Supporting Information ). CsoR ortologous have been resolved as homodimers and homotetramers (Liu et al. 2007, Sakamoto et al. 2010. Other highly conserved r esidues ( Figur e S1, Supporting Information ) could be presumabl y involv ed in the inter action between CsoR monomers to originate the oligomer. Considering these results, Bdiaspc4_RS03270 has been denoted as the csoR gene henceforth in this manuscript.

Growth and gene expression analyses in a csoR strain
In order to investigate the possible role of CsoR in denitrification gene expression under Cu limitation, a B. diazoefficiens markerless deletion mutant in the csoR gene (denoted as csoR ) was constructed (see the section 'Material and Methods'). Growth of B. diazoefficiens 110 spc 4 (WT) and the csoR mutant was monitored throughout an incubation period of 6 days under microoxic conditions in Cu limitation (i.e. chelated, Cu-L), standard Cu (0.02 μM, Cu-S), or high Cu (13 μM, Cu-H) BVMN media. As Cu concentration did not affect WT growth under oxic conditions (P ac heco et al. 2022 ), Cu-S oxic growth of each strain was used as control in our experiments. It is important to note that, as shown in Fig. 3 (A), Cu-H (13 μM) is not actually toxic for B. diazoefficiens growth. In fact, Cu-H impr ov es the growth rate compared to Cu-S (Fig. 3 A), as previousl y demonstr ated (P ac heco et al. 2022 ). When B. diazoefficiens cells were grown under different Cu concentrations ranging from 13 μM to 10 mM Cu, the growth profile obtained with 1 mM was similar to that observed with Cu-S. Ho w e v er, 2 and 3 mM Cu triggered a significant delay in growth that was completely abolished in the presence of 10 mM (data not shown). Hence, Cu-H (13 μM) is simply a Cu concentration above the standard Cu level for BVMN medium (Cu-S, 0.02 μM), i.e. suitable for our r esearc h pur poses in the present work.
As shown in Fig. 3 (A), under microoxic conditions, no differences in growth behaviour wer e observ ed in the csoR in comparison to the WT, independently of the Cu concentr ation, i.e. gr owth was significantly affected in both strains by Cu limitation, compared to Cu-S or Cu-H grown cells. Consequently, CsoR is not appar entl y involv ed in B. diazoefficiens gr owth under micr ooxic conditions in response to Cu. Analyses of Cu concentration in cells r e v ealed significant differences of Cu levels between cells grown in Cu-L, Cu-S, or Cu-H media during 3-days under microoxic conditions (Fig. 3 B). As expected, Cu could not be detected in Cu-L gr own cells fr om an y of the str ains. Ho w e v er, Cu le v els incr eased to about 0.5 and 5 μg Cu mg protein −1 in WT Cu-S and Cu-H grown cells, r espectiv el y (Fig. 3 B). Inter estingl y, under Cu-S conditions, Cu concentration in the csoR mutant was significantly lo w er than that from WT cells (0.31 μg Cu mg pr otein −1 v ersus 0.47 μg Cu mg pr otein −1 , r espectiv el y). In contr ast, no significant differ ences in Cu le v els wer e observ ed in Cu-H gr own WT cells compar ed to the csoR mutant (Fig. 3 B).
Next, the influence of csoR mutation on denitrification gene expression was determined by measuring β-galactosidase activity of transcriptional lacZ fusions and performing qRT-PCR assays (Fig. 4 ). For the former a ppr oac h, csoR str ain deriv ativ es harbouring napE-lacZ , nirK-lacZ , norC-lacZ , or nosR-lacZ transcriptional fusions were firstly constructed (see Table 1 ). Then, both WT and csoR cells wer e gr own micr ooxicall y in Cu-L, Cu-S, or Cu-H BVMN media. While β-galactosidase activity was measured after 72 h of incubation, when the maximal expression levels of these fusions is r eac hed (P ac heco et al. 2022 ), for qRT-PCR experiments cells were collected after 48 h.
As shown in Fig. 4 (A), both strains, WT and csoR , displayed similar expr ession le v els for napE-lacZ fusion independentl y of the Cu condition assayed ( P > .05). Regarding nosR-lacZ expression, as pr e viousl y observ ed by P ac heco et al. ( 2022 ), β-galactosidase activity was significantly reduced in Cu-L comparing to Cu-S WT cultures (Fig. 4 B). Ho w ever, no significant differences were observed between WT and csoR strains regardless of the Cu concentration ( P > .05) (Fig. 4 B). These results were confirmed by qRT-PCR (Fig. 4 A and B), indicating that nap and nos gene expression was not affected by csoR deletion.
Regarding nirK-lacZ and norC-lacZ fusions, no differences were observ ed between str ains in Cu-L medium ( P > .05) (Fig. 4 C  and D). Ho w e v er, the r esults obtained in qR T-PCR assays sho w ed that nirK and norC expression levels were ∼37-fold and 25-fold lo w er, r espectiv el y, in csoR compar ed to WT under Cu limitation (Fig. 4 C and D). This might be an indication of a possible posttr anscriptional contr ol in the csoR m utant that affects nirK and norC mRNA stability.
Under Cu-S conditions, nirK-lacZ expr ession between str ains was not significantly different ( P > .05) (Fig. 4 C), and this result w as confirmed b y qR T-PCR, since nirK expr ession anal yses did not show a r ele v ant c hange in the WT compar ed to csoR (Fig. 4 C).
Concerning norC-lacZ expression in Cu-S medium, lo w er le v els (about 3-fold less) were observed in the csoR mutant compared to those found for the WT strain ( P < .05) (Fig. 4 D). Similarly, norC expr ession le v els obtained b y qR T-PCR in Cu-S w ere ∼29-fold lo w er in csoR compared to WT (Fig. 4 D). This significant decrease suggests that CsoR might be r ele v ant to induce maximal nor genes expr ession in r esponse to standard Cu le v els (0.02 μM) in the growth medium.
In Cu-H medium (13 μM Cu), there were no significant differences in expression for any of the transcriptional fusions assayed ( napE , nirK , norC , or nosR-lacZ ), and these results were confirmed by qRT-PCR (Fig. 4 A-D), indicating that CsoR may be not r ele v ant for B. diazoefficiens denitrification gene induction under this Cu condition.
Next, we investigated the possible involvement of CsoR in copA contr ol. Ther efor e, w e performed qR T-PCR experiments to analyse copA expression in the WT and csoR strains grown under the battery of Cu concentrations assayed in this study. As shown in Fig. 4 (E), copA expression was about 1700-fold and 3100-fold higher in the csoR str ain compar ed to the WT, when cells were grown under Cu-L or Cu-S conditions, r espectiv el y. These r esults suggest that CsoR could be involved in Cu response regulation as a repressor of copA either in Cu-L or Cu-S. When Cu concentration increased up to 13 μM (denoted as Cu-H in this work), copA expression was only about 21-fold higher in the csoR mutant compared to the WT (Fig. 4 E). This result suggests, as pr e viousl y demonstr ated in M. tuberculosis (Liu et al. 2007 ), that CsoR might be attached to the promoter region under low Cu availability, thus causing csoR oper on r epr ession. According to Liu et al. ( 2007 ), when Cu concentr ation incr eases, one Cu(I) ion is bound to eac h CsoR monomer, leading to the dissociation of the protein from the operon and allowing transcription of these genes, including CtpV, the Cu + transporter. In the case of B. diazoefficiens , it may be possible that expression of the copA gene, which encodes a putative Cu + exporter, is der epr essed in the csoR str ain gr own under Cu-L, making the cytosolic Cu concentr ation e v en mor e limiting (Fig. 5 ). Pr e vious results sho w ed a strong inhibition of nirK and norC expression by Cu limitation in WT cells (P ac heco et al. 2022 ). The qRT-PCR results from the present work (Fig 4 C and D) suggest that csoR deletion reinforces the repression of nirK and nor genes in response to Cu limitation. Mor eov er, the significant r eduction of norC expression in Cu-S (Fig. 4 D) could be explained by the increased fold-c hange v alue in copA expr ession (Fig. 4 E) in csoR compared to WT. Contrary to norC expression, csoR mutation did not affect nirK gene expression under Cu-S conditions (Fig. 4 C). This observation suggests that nor gene expression might be more sensitive to the decline in Cu concentration inside the cell than nirK gene. Supporting this suggestion, P ac heco et al. ( 2022 ) demonstrated that Cu limitation had a greater negative effect on norC gene expression than on nirK expression b y qR T-PCR analyses (33.25 versus 10.73 FC). Hence, the reduction in cytosolic Cu concentration in csoR Cu-S cells compared to WT cells (Fig. 3 B) may preclude conditions. In the WT grown under Cu limitation (A), CsoR remains attached to the csoR-copA divergon, preventing copA maximal tr anscription le v els . T he synthesized P-type ATP ase CopA pr oteins would be inactiv e because of the lo w c ytosolic Cu av ailability. In csoR gr own under Cu limitation (B), CsoR is absent and copA expression would be derepressed making the cytosolic Cu concentration even more limiting (expressed in the figure as ↑ Cu-limitation). The results obtained in this work suggest that RegR could be involved in the control of nor genes under Cu limitation. Question marks indicate nondemonstrated events. Perpendicular lines indicate negative control.
csoR cells from reaching adequate nor expression. In order to investigate the possibility that CsoR could be a direct regulator of nirK or norC genes, we analyse the promoter sequences of those genes by using the FIMO (Find Individual Motif Occurrences) tool from the MEME suite. Only in the divergent promoter region between copA and csoR , two CsoR boxes were identified with P -values of 1.01 × 10 −8 and 2.08 × 10 −8 for each gene, respectively. Ho w ever, putati ve CsoR bo xes were not present in the nirK or norC promoter regions (data not shown).
Denitrification genes (especially nor ) may be controlled by transcriptional factors able to detect transient Cu concentration shifts inside the cell. The ability of Cu to undergo redox changes, transiting from Cu(II) to Cu(I) and vice ver sa , mak es Cu an ideal cofactor for enzymes catalysing electron transfer. The major redox state of Cu in bacterial cytoplasm is Cu(I) due to the low reduction potential maintained b y lo w molecular-w eight thiols ( + 0.15 V for Cu + and −0.22 V for Cu 2 + ) (Davis and O'Halloran 2008 ). Ho w e v er, Cu(I) is a strong soft metal and can attack and destroy iron-sulfur clusters thereby releasing iron, and consequently provoking oxidativ e str ess (r e vie w ed b y Rensing and McDevitt 2013 ). T hus , we propose that Cu limitation could trigger transient changes in the cytosolic redox state that would be detected by the redox responsive RegSR system, as depicted in Fig. 5 . In fact, pr e vious findings show that, in addition to controlling nifA , RegR is also involved in B. diazoefficiens denitrification where it is r equir ed for induction of nor genes in response to anoxia and NO 3 − (Torres et al. 2014 ). Torres et al. ( 2014 ) also demonstr ated the ca pacity of RegR to bind norC pr omoter. Results fr om the pr esent work allow us to propose that Cu-limiting conditions causes a cytosolic redox change that could be detected by the RegSR system, thus RegR would bind the norC promoter and block nor expression (Fig. 5 A). This effect is strengthened in the csoR strain (Fig. 5 B). RegR might be also involved in the downregulation of nirK gene, but no evidences supporting this hypothesis have been reported yet.
To conclude, this study suggests that CsoR is not involved in nirK and nor gene regulation as a direct transcriptional factor, but it could influence indir ectl y the expression of these genes in response to the intracellular Cu concentration. Further investigation would be necessary in order to discern the potential connection between Cu homeostasis, the redox status of the cytoplasmic compartment, and regulation of denitrification.