Transcriptional and Posttranscriptional Events Control Copper-Responsive Expression of a Rhodobacter capsulatus Multicopper Oxidase

Abstract

The copper-regulated Rhodobacter capsulatus cutO (multicopper oxidase) gene confers copper tolerance and is carried in the tricistronic orf635-cutO-cutR operon. Transcription of cutO strictly depends on the promoter upstream of orf635, as demonstrated by lacZ reporter fusions to nested promoter fragments. Remarkably, orf635 expression was not affected by copper availability, whereas cutO and cutR were expressed only in the presence of copper. Differential regulation was abolished by site-directed mutations within the orf635-cutO intergenic region, suggesting that this region encodes a copper-responsive mRNA element. Bioinformatic predictions and RNA structure probing experiments revealed an intergenic stem-loop structure as the candidate mRNA element. This is the first posttranscriptional copper response mechanism reported in bacteria.

INTRODUCTION

Copper serves as a cofactor of diverse enzymes in animals, plants, fungi, archaea, and bacteria (3, 7, 27). Cuproenzymes like amine oxidases, cytochrome oxidases, plastocyanins, nitrite reductases, tyrosinases, multicopper oxidases, laccases, and superoxide dismutases play central roles in nitrogen and carbon assimilation, respiration, photosynthesis, iron transport, and detoxification of reactive oxygen species (5, 26, 37). Above certain concentrations, however, copper is toxic. Excess copper disturbs iron-sulfur cluster formation and destabilizes iron-sulfur cofactors of dehydratases (8, 24, 25). In addition, toxicity resulting from copper-induced generation of reactive oxygen species is frequently but controversially discussed (8, 15, 24, 25, 36, 44, 46).

To cope with copper toxicity, bacteria have evolved various defense mechanisms. The enteric bacterium Escherichia coli uses different copper tolerance systems encoded by the cop, cus, cue, and pco genes (32, 36). The copA gene codes for a P-type ATPase, which translocates copper from the cytoplasm to the periplasm. The cus genes encode an efflux complex spanning both membranes, CusABC, and a periplasmic copper-binding chaperone, CusF, which act together in copper excretion from cytoplasm and periplasm. The cueO gene codes for a multicopper oxidase, which detoxifies the periplasm of excess copper by oxidizing different substrates. First, CueO oxidizes Cu(I) to the less toxic Cu(II) oxidation state (9, 42). The higher toxicity of Cu(I) possibly results from its ability to cross the cytoplasmic membrane (36, 39). Second, CueO oxidizes the iron siderophore enterobactin, thus preventing enterobactin-mediated reduction of Cu(II) to Cu(I) (16). Furthermore, oxidized enterobactin is thought to form polymers which chelate copper in the periplasm (20). In addition to the cop, cus, and cue genes, which are located on the chromosome, some highly copper-resistant strains harbor a plasmid carrying the pco gene cluster. One of these genes, pcoA, codes for another multicopper oxidase which is structurally and functionally similar to CueO (10). Besides E. coli, copper defense by multicopper oxidases has been described for many different bacteria, including Campylobacter jejuni, Myxococcus xanthus, Rhodobacter capsulatus, Salmonella enterica, Staphylococcus aureus, and Xanthomonas campestris (1, 17, 18, 38, 42, 43, 53).

In Gram-negative bacteria, copper defense genes are typically activated by CusR or CueR regulators, when the copper concentration in the environment increases (31, 33, 34, 36, 45, 54, 56). The response regulator CusR is part of the CusRS two-component system, in which the sensor kinase CusS senses periplasmic copper concentrations. CueR acts as a single protein consisting of two domains, a DNA-binding domain and a copper-binding domain, with the latter sensing the cytoplasmic copper status. Recently, a copper-dependent extracytoplasmic function (ECF) sigma factor was shown to activate P-type ATPase and multicopper oxidase genes in Myxococcus xanthus (14). In Gram-positive bacteria, copper defense genes are mostly repressed by CsoR or CopY regulators at low copper concentrations (23, 44). Like CueR, both CsoR and CopY directly bind copper in the cytoplasm. Upon increase of copper concentrations, the repressor proteins no longer bind their target promoters, thus allowing enhanced expression of copper tolerance genes.

In the present study, we examined copper-dependent cutO (multicopper oxidase gene) expression in the phototrophic alphaproteobacterium R. capsulatus. Like E. coli and other bacteria mentioned above, R. capsulatus controls multicopper oxidase expression in response to copper availability. In contrast to these bacteria, however, copper regulation in R. capsulatus is not achieved at the transcriptional level but, instead, occurs by a novel posttranscriptional mechanism.

MATERIALS AND METHODS

Bacterial strains, plasmids, oligonucleotides, and growth conditions.

Bacterial strains and plasmids are shown in Table 1. Primers for PCR amplifications are listed in Table 2. Media, antibiotic concentrations, growth conditions, and copper sensitivity assays were previously described (reference 53 and references therein).

Table 1.

Bacterial strains and plasmids

Strain or plasmid	Relevant characteristicsa	Reference or source
E. coli strains
JM83	Host for plasmid amplification	48
S17-1	Donor for biparental mating	41
R. capsulatus strains
B10S	Spontaneous Smr mutant of Rhodobacter capsulatus B10	21
CR10, CR11	cutR::[Km>], cutR::[<Km] mutants	This study
CR36-I, CR36-II	orf4974::[Km>], orf4974::[<Km] mutants	This study
CR40, CR41	orf635::[Km>], orf635::[<Km] mutants	This study
CR89, CR90	orf632::[Km>], orf632::[<Km] mutants	This study
CR99, CR100	orf633::[Km>], orf633::[<Km] mutants	This study
JW12-I, JW12-II	cutO::[Km>], cutO::[<Km] mutants	53
Plasmids
pBBR1MCS-5	Broad-host-range vector; Gmr	22
pBSL15, pBSL86	Donor plasmids for [Kmr]; Apr	2
pPHU234	Broad-host-range vector; Tcr	19
pSUP202	Narrow-host-range vector; Apr Cmr Tcr	41
pSVB10	Narrow-host-range vector; Apr	4
pUC19	Narrow-host-range vector; Kmr	55
pYP35	Donor plasmid for [lacTeT]; Apr	13
pAM39	pUC19 derivative with T7 promoter and orf635-cutO intergenic region	This study
pCR10, pCR11	pSUP202 derivatives carrying cutR::[Km>] and cutR::[<Km]	This study
pCR36-I, pCR36-II	pSUP202 derivatives carrying orf4974::[Km>] and orf4974::[<Km]	This study
pCR40, pCR41	pSVB10 derivatives carrying orf635::[Km>] and orf635::[<Km]	This study
pCR83 to pCR87	pSUP202 derivatives carrying lacZfused to orf635, orf635-cutO intergenic region, cutO, and cutR	This study
pCR89, pCR90	pSVB10 derivatives carrying orf632::[Km>] and orf632::[<Km]	This study
pCR99, pCR100	pSVB10 derivatives carrying orf633::[Km>] and orf633::[<Km]	This study
pCR209	pSUP202 derivative carrying an in-frame orf635-lacZ fusion; Gmr	This study
pCR211	pSUP202 derivative carrying an in-frame cutO-lacZ fusion; Gmr	This study
pCR213	pSUP202 derivative carrying an in-frame cutR-lacZ fusion; Gmr	This study
pJL5, pJL6, pJL8, pJL29, pJL30	pBBR1MCS-5 derivatives carrying cutO-lacZ	This study
pJL7	pBBR1MCS-5 derivative carrying orf635-lacZ	This study
pCR234, pCR235, pCR236, pCR238, pRM65, pRM74, pRM78, pRM79, pRM82, pRM85, pRM87	pJL5 derivatives with site-directed mutations in the orf635-cutO intergenic region	This study

^aAp, ampicillin; Cm, chloramphenicol; Gm, gentamicin; Km, kanamycin; Sm, streptomycin; Tc, tetracycline; >, the resistance gene reads in the same direction as the interrupted gene; <, the resistance gene reads in the opposite direction.

Table 2.

PCR primers

Primer pair	Sequence (5′-3′)a	Application
UP-CR18	GAACCGGGATCCCCAGGACGGAC	Cloning of pCR40, pCR41, pJL5
LP-CR18	GACCGTCTCGAGCACTTCCTGCGG
UP-CR18	GAACCGGGATCCCCAGGACGGAC	Cloning of pJL6
LP-CR19	CAGAAAGCCGCGGCGGGAAAGCTG
UP-CR18	GAACCGGGATCCCCAGGACGGAC	Cloning of pJL7
LP-CR20	GACAATGCTGCAGGCAGCCTTCCGG
UP-CR19	GGCTGCCTGCAGCATTGTCTGGGC	Cloning of pJL8
LP-CR18	GACCGTCTCGAGCACTTCCTGCGG
UP-CR33	CCGGATCCGGCGGGAAGCCTTGC	Cloning of pCR83
LP-CR33	GGGGAAGCTTCTGCAGGCAGCCTTC
UP-CR34	CCGGATCCGCCTCAGCCTGATCG	Cloning of pCR84
LP-CR34	GGGGAAGCTTCCGCGGCGGGAAAG
UP-CR35	CCGGATCCGATCGAGGTGAATGG	Cloning of pCR85
LP-CR35	GGGGAAGCTTCTCGAGCACTTCCTG
UP-CR36	CCGGATCCGTGCCATGCTCCTCC	Cloning of pCR86
LP-CR36	GGGGAAGCTTGGAACGTGGCGAAG
UP-CR37	CCGGATCCCAGGTGATCGCGGTG	Cloning of pCR87
LP-CR37	GGGGAAGCTTCTGCAGCGCGTCATAG
UP-CR1	AAGCTTCTGCAGATCCTGCGCCTGAAAGGCCAGA	Cloning of pCR209
LP-CR54	GGGGAAGCTTGAATCCGCATGGGAG
UP-CR34	CCGGATCCGCCTCAGCCTGATCG	Cloning of pCR211
LP-CR49	GGGGAAGCTTCGCGGCGGGAAAGC
UP-CR46	CCGGATCCCATGGCGCAGGAGATG	Cloning of pCR213
LP-CR51	GGGGAAGCTTGCTTGGTCATATTCG
UP-CoRE-C30G	CACGTTCCCATTCTTGAACGGGAGATC	Site-directed mutagenesis (pRM85)
LP-CoRE-C30G	ACCCGTTCAAGAATGGGAACGTGGCGA
UP-CoRE-d21T22T	CCACGTTCCCACTTGAACCGG	Site-directed mutagenesis (pRM78)
LP-CoRE-d21T22T	TTCAAGTGGGAACGTGGCGAAGA
UP-CoRE-G9C	CTTCCCCACGTTCCCATTC	Site-directed mutagenesis (pRM82)
LP-CoRE-G9C	GGGAACGTGGGGAAGATCA
UP-CoRE-T6VT7V	CTCTGATCVVCGCCACGTTCCCAT	Site-directed mutagenesis (pRM74, pRM79)
LP-CoRE-T6VT7V	CAAGAATGGGAACGTGGCGBBGATC
UP-CoRE-T15GT16G	CTTCGCCACGGGCCCATTCTT	Site-directed mutagenesis (pRM65)
LP-CoRE-T15GT16G	ATGGGCCCGTGGCGAAGATCA
UP-CoRE-G32TA33T	TTGAACCGTTGATCATCATGACTC	Site-directed mutagenesis (pCR235)
LP-CoRE-G32TA33T	TGATGATCAACGGTTCAAGAATGG
UP-CoRE-G32TA33G	TTGAACCGTGGATCATCATGACTC	Site-directed mutagenesis (pCR236)
LP-CoRE-G32TA33G	TGATGATCCACGGTTCAAGAATGG
UP-CR57	CACGTTCCCATTCTTGAACCGTTGATCATCATGACTC	Site-directed mutagenesis (pCR234)
LP-CR57	TGATGATCAACGGTTCAAGAATGGGAACGTGGCGA
UP-CR58	CACGTTCCCATTCTTGAACCGTGGATCATCATGACTC	Site-directed mutagenesis (pCR238)
LP-CR58	TGATGATCCACGGTTCAAGAATGGGAACGTGGCGA
UP-JL2	CGACGTTTCGGTCCGGGCCAGC	Cloning of pJL29
LP-CR19	CAGAAAGCCGCGGCGGGAAAGCTG
UP-JL3	CAGCTTCGGGGACCCAGCGGAG	Cloning of pJL30
LP-CR19	CAGAAAGCCGCGGCGGGAAAGCTG
UP-cutR-1	CATGGGCCATCTTGCGACCGGC	Cloning of pCR10, pCR11
LP-cutR-1	GTCCCCCTGCAGGTCGACGGCCGCGTCGTAGTGGGCAAAG
UP-PcutO	CCAGGGCGGCGCGGTAGAAC	Cloning of pCR99, pCR100
LP-Porf635-1	CGAAGCAGGGCGCACAGAAT
UP-rc632-1	CCTGACGGCGCTGGGATGG	Cloning of pCR89, pCR90
LP-rc632-1	TGCGGCGGATATTCGACAAAGAT
UP-rc635-cutO-1	CCGAAATTAATACGACTCACTATAGGGCTGCTTTCCCCCGAA	Cloning of pAM39
LP-rc635-cutO-1	CCGATATCCGCAGAAGCGGCCAGAAAGCCG
UP-rc4974B	AACCCGGATCTTTGCCCACTACG	Cloning of pCR36
LP-rc4974B	GCCTGCTGCCGGTTCTGTCG

^aRecognition sites of restriction enzymes are marked by underlining. A T7 promoter sequence is shown in bold.

Construction of R. capsulatus mutant strains.

Construction of insertion-knockout mutants was done essentially as described previously (reference 53 and references therein). Briefly, DNA fragments from the R. capsulatus cutO region were PCR amplified using appropriate primer pairs (Table 2) and cloned into narrow-host-range vector plasmids (Table 1). Subsequently, restriction sites shown in Fig. 1 were used to insert kanamycin resistance (Km) cassettes derived from pBSL15 or pBSL86. Conjugational transfer of resulting mutagenesis plasmids (Table 1) from E. coli S17-1 into R. capsulatus and selection for gene replacement by double crossover were carried out as described earlier (21, 28). The identities of the R. capsulatus mutants were verified by PCR (data not shown).

Fig 1.

Identification of copper tolerance genes by mutational analysis. (A) Genetic and physical map of the R. capsulatus cutO (multicopper oxidase gene; RCAP_rcc02110) region. The localizations and sizes of genes and open reading frames (orf) are given by arrows. Only restriction sites relevant for cloning of kanamycin resistance (Km) cassettes are shown. Km cassettes are not drawn to scale. Arrowheads indicate the direction of Km transcription. Designations of R. capsulatus mutants and copper susceptibilities (Cu^T, copper tolerant; Cu^S, copper sensitive) are shown next to the Km cassettes. (B) Copper tolerance of wild-type and mutant strains. Liquid cultures were plated on RCV minimal medium plates, before filter discs soaked with 2.5 mM CuSO₄ were placed on top. Inhibition zones around the discs were documented after 2 days of growth.

Construction of R. capsulatus reporter strains and β-galactosidase assays.

To construct chromosomal lacZ reporter fusions, R. capsulatus orf635, cutO, and cutR fragments of approximately 0.4 kb were PCR amplified using appropriate primer pairs (Table 2) prior to cloning into narrow-host-range plasmid pSUP202. Subsequently, the [lac TeT] cassette from plasmid pYP35 carrying the promoterless E. coli lacZ gene was added, resulting in plasmids pCR83 to pCR87 (Table 1). Cloning of the [lac TeT] cassette results in transcriptional fusions between a gene of interest and lacZ (13). To construct episomal lacZ reporter fusions, DNA fragments shown in Fig. 2 were PCR amplified using appropriate primer pairs (Table 2) prior to cloning into broad-host-range plasmid pBBR1MCS-5. Subsequently, the [lac TeT] cassette was added, resulting in plasmids pJL5 to pJL8, pJL29, and pJL30 (Table 1). Site-directed mutagenesis of plasmid pJL5 was done according to the QuikChange protocol (Stratagene, Amsterdam, The Netherlands), resulting in plasmids pCR234 to pCR236, pCR238, pRM65, pRM74, pRM78, pRM79, pRM82, pRM85, and pRM87 (Table 1). To construct translational (in-frame) lacZ reporter fusions, appropriate DNA fragments were cloned into plasmid pPHU234. Subsequently, SalI fragments from these pPHU234 derivatives were cloned into a pSUP202 derivative conferring gentamicin resistance, resulting in reporter plasmids pCR209 (orf635-lacZ), pCR211 (cutO-lacZ), and pCR213 (cutR-lacZ) (Table 1).

Fig 2.

Copper-dependent cutO expression requires two cis-regulatory elements. (A) Transcriptional lacZ reporter fusions. Plasmids carrying lacZ fusions to orf635, cutO, and cutR based on narrow-host-range plasmid pSUP202 were used to construct chromosomal R. capsulatus reporter strains. Episomal lacZ fusions are based on broad-host-range plasmid pBBR1MCS-5. The lacZ gene is not drawn to scale. (B) Expression studies. R. capsulatus reporter strains were grown in RCV minimal medium with 2 μM CuSO₄ (+Cu) or without copper addition (−Cu) prior to determination of β-galactosidase activities as described earlier (53). β-Galactosidase activities, given in Miller units (30), and standard deviations were calculated from five independent experiments for each strain.

Reporter plasmids were introduced into R. capsulatus as described earlier (21, 28). Selection for tetracycline or gentamicin resistance led to isolation of R. capsulatus reporter strains (13). R. capsulatus reporter strains were grown in R. capsulatus V (RCV) minimal medium with 2 μM CuSO₄ or without copper addition (52, 53). At late exponential phase, β-galactosidase activities were determined as described earlier (30).

In vitro transcription and RNA structure probing.

RNA for structure probing was synthesized in vitro by runoff transcription with T7 RNA polymerase using EcoRV-linearized plasmid pAM39 as the template. Transcripts were 5′ end labeled as described previously (6). RNA corresponding to 30,000 cpm was preincubated with or without CuCl (2 μM) and dithiothreitol (DTT; 2 μM), before 0.01 U RNase V (Ambion, Darmstadt, Germany), 1 U nuclease S1 (Fermentas, St. Leon-Rot, Germany), or 0.002 U RNase T1 (Ambion, Darmstadt, Germany) was added. After 5 min at 30°C, reactions were stopped by addition of 5 μl formamide loading dye as described previously (12, 49, 50). After denaturing at 95°C, cleavage products were separated on 8% polyacrylamide gels. Generation of alkaline ladders was done as described previously (6).

RESULTS

The orf635-cutO-cutR operon confers copper tolerance.

The cutO gene (RCAP_rcc02110), coding for multicopper oxidase, was previously shown to confer copper tolerance to R. capsulatus (53). To determine the roles of flanking genes in copper tolerance, insertion-knockout mutants carrying kanamycin resistance (Km) cassettes were generated (Fig. 1A). Transcription from the Km promoter is not terminated within the cassette, thus driving expression of downstream genes reading in the same direction (11, 53).

Wild-type and mutant strains were examined for growth at different copper concentrations and by filter disc assays. Under copper-limiting conditions, all mutants grew as well as the wild type, demonstrating that none of the respective genes was essential for viability under standard growth conditions (data not shown). Mutants defective for cutO were more sensitive toward copper than the wild type, thus confirming earlier studies (Fig. 1B) (53). Like cutO mutants, orf635 and cutR mutants were copper sensitive, while orf632, orf633, and orf4974 mutants were as copper tolerant as the wild type (Fig. 1B). These findings suggest that copper tolerance (cut) genes are confined to the tricistronic orf635-cutO-cutR operon, previously defined by reverse transcriptase (RT)-PCR (53).

Transcription of cutO strictly depends on the orf635 promoter.

To narrow down the cis-regulatory elements involved in copper regulation of cut genes, R. capsulatus reporter strains carrying chromosomal and episomal lacZ reporter fusions were generated (Fig. 2A). To construct transcriptional fusions, we used a promoterless lacZ gene having its own ribosomal binding site and translational initiation codon (13).

Reporter strains were grown in minimal medium with 2 μM CuSO₄ (+Cu) or without copper addition (−Cu) prior to determination of β-galactosidase activities (Fig. 2B). Results from expression studies may be summarized as follows. (i) Expression of cutO strictly depended on the orf635 promoter, as shown by examination of episomal cutO-lacZ fusions (pJL5, pJL8, pJL29, and pJL30). These results excluded the existence of a secondary promoter within the 40-bp orf635-cutO intergenic region. A 19-bp sequence deleted in pJL30 was critical for activity of the orf635 promoter. It remains speculative whether this sequence is required for RNA polymerase binding, since −35/−10 promoter sequences are not well-defined in R. capsulatus. It is worth noting that, despite several attempts, we failed to determine the orf635 transcription start site (+1) by 5′ rapid amplification of cDNA ends (RACE) experiments, suggesting that mRNA levels are rather low. (ii) Expression of cutO was significantly higher under +Cu than under −Cu conditions, thus confirming previous studies (Fig. 2B) (53). β-Galactosidase activities mediated by episomal cutO-lacZ fusions (pJL5, pJL6, pJL29) were about 10-fold higher than those based on their chromosomal counterparts (pCR84, pCR85), which was most likely due to copy-number effects. Copper-mediated regulation was comparable, showing that both chromosomal and episomal reporter fusions were equally well suited to examine cut gene expression. (iii) Similar to cutO, the cutR gene was copper regulated, as one would expect for two genes of the same operon (Fig. 2B). (iv) Expression of an episomal cutO-lacZ fusion (pJL5) was similar in wild-type and cutR mutant backgrounds (data not shown), suggesting that CutR does not act as a cut operon repressor, as suspected earlier on the basis of chromosomal cutO-lacZ fusions followed by Km cassettes (53). This discrepancy might be explained by the polarity of the cassettes not only onto cutR but also onto cutO-lacZ. As shown in Fig. 1, cutR mutants were copper sensitive, a finding also arguing against a role of CutR as a repressor. (v) In contrast to cutO and cutR, expression of the orf635 gene was copper independent. Although originating from the same promoter, orf635 and cutO expression differed regarding copper regulation, indicating posttranscriptional regulation (see below). (vi) Similar to orf635-lacZ, a lacZ fusion to the orf635-cutO intergenic region (pCR86) was not affected by copper. These findings provided first evidence that copper-dependent cutO control involved the orf635-cutO intergenic region (see below).

CutO and CutR accumulate to higher levels than Orf635.

To estimate cellular contents of Orf635, CutO, and CutR proteins, translational (in-frame) fusions between each of the three genes and the lacZ coding region were constructed (Fig. 3A). R. capsulatus reporter strains carrying chromosomal in-frame fusions were grown in minimal medium with 2 μM CuSO₄ or without copper prior to determination of β-galactosidase activities (Fig. 3B). Both translational cutO-lacZ and cutR-lacZ fusions were clearly induced upon copper addition, nicely reflecting the data obtained with transcriptional fusions (Fig. 2B). In contrast, only background β-galactosidase activities were detected in the reporter strain carrying the in-frame orf635-lacZ fusion, suggesting that only low levels of Orf635 protein are required to maintain copper tolerance.

Fig 3.

Expression of in-frame lacZ fusions to orf635, cutO, and cutR. (A) Translational (in-frame) lacZ reporter fusions. Plasmids carrying lacZ fusions to orf635, cutO, and cutR based on narrow-host-range plasmid pSUP202 were used to construct chromosomal R. capsulatus reporter strains. The lacZ gene is not drawn to scale. (B) Expression studies. R. capsulatus reporter strains were grown in RCV minimal medium with 2 μM CuSO₄ (+Cu) or without copper addition (−Cu) prior to determination of β-galactosidase activities as described earlier (53). β-Galactosidase activities, given in Miller units (30), and standard deviations were calculated from five independent experiments for each strain.

The orf635-cutO intergenic region is essential for copper-dependent expression of cutO.

The program mfold (57) predicted a stem-loop structure in the orf635-cutO intergenic mRNA (Fig. 4A). This structure consists of two stems (S-I and S-II) and two loops (L-I and L-II). S-I starts at the orf635 stop codon (UGA) and encompasses the putative Shine-Dalgarno (SD) sequence (GGAG) of cutO. The same RNA structure was predicted even if sequences from the flanking genes were added.

Fig 4.

Site-directed mutations abolishing copper control of cutO expression. (A) Structure prediction of orf635-cutO intergenic mRNA. Intergenic mRNA was predicted to form two stems (S-I and S-II) and two loops (L-I and L-II) using the program mfold (57). (B) Site-directed mutagenesis of the orf635-cutO intergenic region. To examine the role of the orf635-cutO intergenic region in copper regulation, appropriate site-directed mutations were introduced into plasmid pJL5 (cutO-lacZ). (C) Transcriptional analysis. R. capsulatus strains carrying pJL5 derivatives were grown in RCV minimal medium with 2 μM CuSO₄ (+Cu) or without copper addition (−Cu). β-Galactosidase activities, given in Miller units (30), and standard deviations were calculated from four independent experiments for each strain.

To study the role of the intergenic region in copper regulation, site-directed mutations were introduced into cutO-lacZ reporter plasmid pJL5 (Fig. 4B). A point mutation in loop L-I (pRM87) or deletion of 2 nucleotides in L-II (pRM78) altered the absolute cutO-lacZ expression levels but still permitted copper regulation. Such modulatory effects on gene expression are compatible with the flexible nature of loop regions. A double mismatch in S-II (pRM65) did not significantly affect expression, suggesting that this region might not form a stable stem structure. Interestingly, two destabilizing mutations within the putative anti-SD sequence of S-I (pRM79 and pRM74) completely abolished copper regulation of cutO-lacZ and led to high basal expression in the absence of copper. Furthermore, reducing the thermodynamic stability of S-I by opening the closing GC pair (pRM82) resulted in elevated expression under copper-limiting conditions. Like mutations in the anti-SD sequence, two destabilizing mutations within the SD sequence of S-I (pCR238 and pCR234) largely abolished copper control. It is worth emphasizing at this point that these studies were based on transcriptional lacZ fusions and therefore lacZ translation did not depend on the SD sequence of cutO. Most remarkably, compensatory anti-SD and SD sequence mutations (pCR236 and pCR235) restored copper regulation strongly, indicating that base pairing is more important than sequence conservation. These findings clearly suggested that the orf635-cutO intergenic region is essential for copper control and supported the assumption that a secondary mRNA structure upstream of cutO is involved in copper regulation.

RNA structure probing of the orf635-cutO intergenic region.

To provide evidence for the structure predicted in the intergenic orf635-cutO mRNA, it was mapped by enzymatic probing. For this purpose, the intergenic region was placed under the control of a T7 promoter to generate in vitro transcripts using T7 RNA polymerase (Fig. 5A; see Materials and Methods). Transcripts were incubated with nuclease S1, RNase V, or RNase T1, which preferably cleave 3′ of single-stranded nucleotides (loops), within paired regions (stems), and 3′ of unpaired guanines, respectively. Cleavage products were separated by polyacrylamide gel electrophoresis (Fig. 5B). Samples without RNase/nuclease addition served as controls.

Fig 5.

Identification of paired and unpaired regions in orf635-cutO intergenic mRNA. (A) In vitro transcription of the orf635-cutO intergenic region. Plasmid pAM39 served as a template for in vitro transcription of the orf635-cutO intergenic region starting at a T7 promoter (P_T7) recognized by T7 RNA polymerase. (B) RNA structure probing experiments. In vitro transcripts were incubated with nuclease S1, RNase V, or RNase T1 prior to separation of cleavage products by polyacrylamide gel electrophoresis. RNase/nuclease digestions were carried out without copper addition or in the presence of 2 μM CuCl as indicated. All samples contained 2 μM dithiothreitol (DTT). Assays without RNase/nuclease addition served as controls (co). Lane L, alkaline ladder.

The RNA structure probing results may be summarized as follows. (i) Fragmented RNA in several preparations suggested that the RNA structure is rather unstable and the backbone is accessible to spontaneous cleavage by in-line attack (35). Copper itself did not influence transcript stability. Thus, it seems unlikely that the intergenic mRNA directly binds copper. (ii) Very efficient nuclease S1 cleavage at positions 20 to 24 (for numbering, see Fig. 5A) indicated that these nucleotides are unpaired, a result compatible with predicted L-II loop formation. (iii) Nuclease S1 products at positions 42 and 43 support the prediction that the AUG start codon is unpaired, too. (iv) RNase V cuts were observed at positions 5 and 6, 11 and 12, and 14 and 15, suggesting that these nucleotides are paired. Products at positions 5 and 6 and positions 14 and 15 are located within S-I and S-II, respectively, thus strengthening the RNA structure prediction for these regions. (v) RNase T1 products were observed at positions 2, 9, 14, 26, 39, and 43. With the exception of position 39 (downstream of an unpaired uridine), all the other products are consistent with the presence of guanines in unpaired regions or at loop borders.

Taken together, RNA structure probing experiments and site-directed mutations strengthened the assumption that the intergenic mRNA between orf635 and cutO forms a stem-loop structure.

DISCUSSION

All three genes of the orf635-cutO-cutR operon are required for copper tolerance in R. capsulatus (53; this study). Orf635 (RCAP_rcc02111) appears to be specific for R. capsulatus, as no homolog was identified in any other bacterium, including the closely related Rhodobacter sphaeroides. Compared to CutO (RCAP_rcc02110) and CutR (RCAP_rcc02109), very low levels of Orf635 are apparently sufficient to maintain copper tolerance (this study). CutO most likely detoxifies the periplasm from excess copper, as described for its E. coli counterpart, CueO. Like CueO and other multicopper oxidases, CutO oxidizes several phenolic compounds and binds (at least) four copper atoms per molecule (53). CutR is a copper-binding protein, too (this study). Inductively coupled plasma-optical emission spectroscopy (ICP-OES) determined that 1.5 ± 0.2 copper atoms were bound per molecule CutR.

Multicopper oxidases play important roles in copper defense in R. capsulatus, E. coli, and many other bacteria (1, 17, 18, 38, 42, 43). Generally, synthesis of multicopper oxidases is low under copper-limiting conditions but significantly enhanced upon copper addition. The underlying regulatory mechanisms, however, differ between species. Typically, Gram-negative bacteria activate transcription of multicopper oxidase genes in the presence of copper, while Gram-positive bacteria repress transcription in the absence of copper (32, 34, 44). In contrast, transcription of R. capsulatus cutO was unaffected by copper, but instead, copper-dependent cutO expression was controlled posttranscriptionally (this study). Two distinct cis-regulatory elements, the orf635 promoter and the orf635-cutO intergenic region, were required for transcriptional and posttranscriptional control, respectively.

Our current model of cut gene regulation by copper is depicted in Fig. 6. Transcription of the orf635-cutO-cutR operon is exclusively driven by the constitutive (copper-independent) orf635 promoter. The orf635-cutO intergenic region is essential for copper control of cutO and cutR expression. Two lines of evidence suggest that orf635-cutO-cutR transcripts might be specifically degraded in the absence of copper. First, RT-PCR products of the tricistronic mRNA were detected only in the presence and not in the absence of copper (53). Second, site-directed mutations in the orf635-cutO intergenic region increased cutO-lacZ expression in the absence of copper to levels otherwise observed only in the presence of copper.

Fig 6.

Model of copper-dependent cutO expression. Copper-dependent cutO expression requires two cis-regulatory elements, the constitutive orf635 promoter (P_const), and the orf635-cutO intergenic region. RT-PCR products corresponding to orf635-cutO and cutO-cutR borders were previously identified in the presence but not in the absence of copper (53). For further details, see text.

RNA structure predictions and RNA structure probing experiments consistently suggest a distinct stem-loop structure in the orf635-cutO intergenic mRNA (57; this study). The critical element of this structure, stem S-I, consists of the cutO ribosomal binding site (SD sequence) paired with an anti-SD sequence. Interestingly, SD and anti-SD sequence mutations predicted to destabilize stem S-I abolished copper regulation of cutO expression. We used a promoterless lacZ reporter gene carrying its own SD sequence to analyze cutO-lacZ expression. Consequently, accessibility of the cutO ribosomal binding site should not have affected lacZ translation. More likely, destabilization of stem S-I by SD and anti-SD sequence mutations might have prevented mRNA degradation under copper-limiting conditions. In line with this hypothesis, compensatory SD/anti-SD sequence mutations predicted to restore base pairing and stem formation also restored copper regulation. Apparently, posttranscriptional copper control depends on secondary structures rather than the sequence context. Likewise, site-directed mutations affecting untranslated mRNA structures in Bacillus subtilis gapA, E. coli cspE, and Klebsiella pneumoniae nif genes strongly influenced mRNA stability (29, 40, 47).

We asked whether orf635-cutO intergenic mRNA by itself was sufficient for in vivo copper control. In an E. coli background, however, expression of lacZ fused to the intergenic region did not respond to copper availability (data not shown), suggesting that copper regulation required at least one factor specific for R. capsulatus. This yet unknown copper-sensing factor (factor X) might mediate mRNA degradation by binding the intergenic mRNA when copper becomes limiting (Fig. 6). The alternative mechanism, namely, copper-dependent mRNA stabilization, is less likely, because binding of intergenic mRNA by factor X would interfere with ribosome binding to the cutO SD sequence. At present, it remains unknown whether factor X is an RNA-binding protein or a small regulatory RNA (51).