OUP user menu

Heterologous production and characterization of bacterial nickel/cobalt permeases

Peter Hebbeln, Thomas Eitinger
DOI: http://dx.doi.org/10.1016/S0378-1097(03)00885-1 129-135 First published online: 1 January 2004


Nickel/cobalt permeases (NiCoTs, TC 2.A.52) are a rapidly growing family of structurally related membrane transporters whose members are found in Gram-negative and Gram-positive bacteria, in thermoacidophilic archaea, and in fungi. Previous studies have predicted two subclasses represented by HoxN of Ralstonia eutropha, a selective nickel transporter, and by NhlF of Rhodococcus rhodochrous, a nickel and cobalt transporter that displays a preference for the Co ion. In the present study, NiCoT genes of five Gram-negative bacteria and one Gram-positive bacterium were cloned and heterologously expressed in Escherichia coli. Based on substrate preference in metal-accumulation assays with the recombinant strains, two of the novel NiCoTs were assigned to the NhlF class. The remaining four NiCoTs belong to a yet unrecognized, third class. They transport both the nickel and the cobalt ion but have a significantly higher capacity for nickel. The observed substrate preferences correlate in many cases with the genomic localization of NiCoT genes adjacent to regions encoding nickel- or cobalt-dependent enzymes or enzymes involved in cobalamin biosynthesis. Alignment of 23 full-length NiCoT sequences and comparison with the available experimental data predict that substrate specificity of NiCoTs is an adaptation to specific transition metal requirements in various organisms from different taxa.

  • Nickel/cobalt permease family (TC 2.A.52)
  • Nickel
  • Cobalt
  • Transport
  • Cobalamin
  • Klebsiella pneumoniae
  • Novosphingobium aromaticivorans
  • Rhodopseudomonas palustris
  • Salmonella typhimurium
  • Staphylococcus aureus
  • Yersinia enterocolitica

1 Introduction

Nickel and cobalt are essential trace elements for prokaryotes and eukaryotes and involved in manifold metabolic processes (see [1,2] for a review). While the structure and mechanism of many nickel- or cobalt-containing enzymes has been investigated in great detail, only limited information is available on the transport systems that mediate nickel or cobalt uptake into cells and – for instance in plants – nickel transport among cell compartments and between cells. Nickel is transported by ABC-type systems in a number of prokaryotes and, likewise, certain ABC transporters (termed CbiMNQO) are believed to mediate cobalt uptake. The nickel/cobalt permeases (NiCoTs; TC 2.A.52), a rapidly growing family of prokaryotic and fungal secondary carriers, are another major group of nickel transporters (see [3;4;5] for a review). They are characterized by an eight-helix structure and by a number of conserved signatures mainly located in transmembrane domains. Most of these transporters are only characterized on the basis of genome sequences. Experimentally analyzed NiCoTs are those from Helicobacter pylori (NixA) [6,7], Ralstonia eutropha (HoxN) [8;9;10;11], Rhodococcus rhodochrous (NhlF) [10;11;12], Schizosaccharomyces pombe (Nic1p) [13] and Yersinia pseudotuberculosis (UreH) [14]. The substrate preferences of HoxN and NhlF have been analyzed in detail. These studies predicted that NiCoTs fall at least into two classes. Class I is represented by HoxN, a highly selective nickel transporter. NhlF is the prototype of class II. It transports both nickel and cobalt ion with a preference for cobalt [10,11]. In the present study six novel NiCoTs were analyzed experimentally. On the basis of metal-accumulation assays, two of them were assigned to class II. The remaining four permeases belong to a third class of NiCoTs and display preference for nickel. Evidence is provided that the genomic localization of NiCoT genes allows predictions on the substrate preference of the encoded permeases.

2 Materials and methods

2.1 Bacterial strains and culture conditions

NiCoT genes of the following genome sequencing strains were amplified: Klebsiella pneumoniae MGH78578, Mycobacterium tuberculosis H37Rv [15], Novosphingobium aromaticivorans F-199, Rhodopseudomonas palustris CGA009, Salmonella enterica serovar Typhimurium lysotype 2 strain SGSC1412 [16], Staphylococcus aureus NCTC8325, Thermoplasma acidophilum DSM1728 [17] and Yersinia enterocolitica 8081-L2. Y. enterocolitica was cultivated in Luria-Bertani broth (LB), N. aromaticivorans in 0.5-fold concentrated LB, and S. aureus in nutrient broth (Difco). R. palustris was grown photoorganotrophically in 50-ml glass bottles containing a mineral medium supplemented with 0.2% (w/v) yeast extract and 0.1% (w/v) sodium succinate as the electron donor in white light (100 W light bulbs). T. acidophilum was grown at 58°C and pH 2 in medium 158 as specified by the Deutsche Sammlung für Mikroorganismen und Zellkulturen GmbH (Braunschweig, Germany).

2.2 Isolation of genomic DNA

Genomic DNA of K. pneumoniae and S. typhimurium was kindly provided by S. Porwollik and M. McClelland (San Diego, CA, USA). Genomic DNA of (i) N. aromaticivorans, T. acidophilum, and Y. enterocolitica, (ii) R. palustris and (iii) S. aureus was isolated as described in [18,19,20], respectively. In the case of M. tuberculosis, cosmid pYUB412 (kindly provided by R. Brosch, Paris, France) was used for amplification of the NiCoT gene.

2.3 Amplification and cloning of NiCoT genes

NiCoT genes were amplified by various standard and ‘touch-down’ polymerase chain reaction (PCR) protocols using genomic DNAs as the template, the gene-specific primers listed in Table 1, and the proofreading thermostable DNA polymerases Platinum Pfx (Invitrogen) or Vent (New England Biolabs). The primers generated NcoI (or AflIII or PciI) sites overlapping the initiation codon of the NiCoT genes and BglII (or BamHI) sites immediately downstream of the last codon. PCR products of the expected sizes were isolated from preparative agarose gels and purified using a spin column kit (Qiagen). The identity of the isolated DNAs was verified by restriction analysis. Selected DNAs were treated with restriction endonucleases NcoI (or AflIII or PciI) and BglII (or BamHI) and inserted into a derivative of plasmid pCH675AF. This plasmid encodes the C-terminally FLAG epitope-tagged HoxN of R. eutropha H16 under control of a lac promoter and a ribosomal binding site [11]. Briefly, pCH675AF was used as the template in an inverse PCR in the presence of primers 675AF+ and 675AF− (Table 1). Treatment of the purified amplicon with NcoI and BglII resulted in a linear, hoxN-free vector with an NcoI end overlapping the initiation codon and a BglII end adjacent to the FLAG-encoding sequence. NiCoT genes were ligated into the vector and the ligation products were transformed into Escherichia coli XL1-Blue (Stratagene). The resulting plasmids were designated according to the source of the NiCoT gene, e.g. pCH675-RP which encodes RP, the NiCoT of R. palustris. Heterologous production of the FLAG epitope-tagged membrane proteins was examined by Western immunoblotting as described recently [11]. Expression of the NiCoT genes from plasmids pCH675-MT and pCH675-TA neither produced immunologically detectable material nor did it enhance metal uptake of the recombinants (not shown). Therefore, the NiCoTs from M. tuberculosis and T. acidophilum were not included in this study.

View this table:

Primers used for construction of NiCoT gene-expression plasmids

PrimerSequenceaRestriction site
  • aRestriction sites are shown in italics.

2.4 Metal accumulation assays

63Ni2+ and 57Co2+ uptake of recombinant E. coli was analyzed as previously described [8;9;10;11]. Cells were grown in LB medium supplemented with the radiolabeled ion at concentrations between 100 and 500 nM, and ampicillin (100 μg ml−1). For expression of NiCoT genes in E. coli XL1-Blue, IPTG (1 mM) was included in the medium since this strain contains a lacIq allele on an F′ episome. Where indicated, non-labeled transition metal chlorides (either Co2+, Cu2+, Mn2+, Ni2+, or Zn2+) were added to the medium as potential inhibitors of Ni and Co uptake. Cells were harvested, washed in 50 mM Tris–HCl, pH 7.5, and concentrated. Radioactivity of aliquots was quantitated in a Canberra-Packard Tri-Carb 2900 TR liquid scintillation counter. Metal accumulation is expressed as pmol mg protein−1. Background activity of the NiCoT-free parent cells was in the range of 3 pmol Ni or Co mg protein−1 at substrate concentrations of 500 nM. Each datapoint in Fig. 1 represents the mean value of double or triple assays using independent cultures grown in the same lot of medium.


Selective metal uptake. 63Ni2+ (black bars and black circles) and 57Co2+ (open bars and open circles) accumulation of recombinant E. coli expressing bacterial NiCoT genes. NiCoTs other than HoxN and NhlF are identified by the first letters of the genus and species name of their source. Left panels: Concentration-dependent accumulation. Radiolabeled ions were supplied at concentrations between 100 and 500 nM. Right panels. Effect of cold Co2+ on 63Ni2+ accumulation (black circles) and of cold Ni2+ on 57Co2+ accumulation (open circles). The concentration of labeled ions was 500 nM. Non-labeled metal chlorides were added as indicated. 57Co2+ accumulation of the HoxN-producing strain was in the range of background activity (below 3 pmol mg protein−1). Based on selectivity the permeases were grouped in class I (specific for nickel), class II (preference for cobalt) and class III (preference for nickel).

2.5 Computer methods

NiCoT genes in microbial genomes were identified by similarity searches using programs BLASTP and TBLASTN [21]. Retrieved total or partial genomic sequences were analyzed using the workbench GENESOAP written by R. Cramm [22]. The dendrogram of NiCoTs shown in Fig. 3 was generated with CLUSTALX [23] in Phylip format using the neighbor-joining method, and displayed in TREEVIEW [24].


Dendrogram of NiCoTs from bacteria, archaea, and fungi. Permeases investigated in this study are boxed. Bfu, Burkholderia fungorum; Bps, Burkholderia pseudomallei; HoxN, Ralstonia eutropha; HupN, Bradyrhizobium japonicum; KP, Klebsiella pneumoniae; Lme, Leuconostoc mesenteroides; Mavi, Mycobacterium avium; MtNicT, Mycobacterium tuberculosis; NA, Novosphingobium aromaticivorans; Ncr, Neurospora crassa; NhlF, Rhodococcus rhodochrous; Nic1p, Schizosaccharomyces pombe; NixA, Helicobacter pylori; RP, Rhodopseudomonas palustris; Rso, Ralstonia solanacearum; SA, Staphylococcus aureus; Saver, Streptomyces avermitilis; Ss, Sulfolobus solfataricus; ST, Salmonella typhimurium; TA, Thermoplasma acidophilum; YE, Yersinia enterocolitica; Yp, Yersinia pestis; Ypt, Yersinia pseudotuberculosis. The greek letters indicate subgroups of the proteobacteria.

3 Results and discussion

3.1 Metal uptake

63Ni2+ and 57Co2+ accumulation of recombinant E. coli XL1-Blue containing pCH675-KP, -NA, -RP, -SA, -ST and -YE was analyzed in metal accumulation assays during growth in complex medium. This type of assay has repeatedly been used to characterize NiCoT-mediated metal uptake [8;9;10;11,13] and has detected nickel-transport activity of NhlF, which was unrecognized in the original transport assays with buffered cell suspensions [12]. For comparison with previous data on the selectivity of NiCoTs [10,11], E. coli CC118 expressing FLAG epitope-tagged HoxN and NhlF from plasmids pCH675AF and pCH674AF, respectively, were included in the present study. The results are shown in Fig. 1. In the first series of experiments (Fig. 1, left panels) nickel and cobalt accumulation of the various recombinant strains was analyzed at substrate concentrations between 100 and 500 nM. These studies confirmed that (i) HoxN is a selective nickel transporter with very low capacity and (ii) NhlF mediates the transport of both cations. Like the latter, the six novel permeases transported both cations but displayed different preferences. Within the substrate concentration range tested, RP- and NA-mediated cobalt transport activity surpassed nickel transport activity significantly. The converse is true for SA, KP, ST and YE. Hence, along with NhlF, RP and NA were assigned to NiCoT class II, while SA, KP, ST and YE form a novel class (class III). Metal accumulation of the various recombinant strains differed considerably. At a concentration of 500 nM of the preferred substrate, cells producing HoxN contained about 35 pmol Ni mg protein−1. NA represents the other extreme. Cells containing this NiCoT accumulated between 600 and 800 pmol Co mg protein−1. These differences did not in each case correlate with the amounts of the various transporters detected in the membrane fractions (Fig. 2). High levels of YE, RP and NA were observed by Coomassie staining of electrophoretically separated membrane protein. Much less material was detectable in the case of SA (Fig. 2), although cells producing this transporter accumulated upto 500 pmol Ni mg protein−1 (Fig. 1).


Detection of recombinant NiCoTs after sodium dodecylsulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and immunoblotting. Membrane protein (approximately 65 μg for Coomassie staining and 52 μg for immunoblotting) of recombinant strains and a control strain containing an empty vector, solubilized with lithium dodecylsulfate on ice in the presence of protease inhibitors, was separated by SDS–PAGE on 10% gels. One gel was stained with Coomassie blue, the other was electroblotted onto a nitrocellulose membrane. The blot was cut in strips and stained with 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium chloride after treatment with a conjugate of monoclonal anti-FLAG M2 antibody and alkaline phosphatase (Sigma). The strips were developed for various times depending on the strength of the signals. Arrows indicate dominant bands in the stained gel, which presumably represent main portions of YE, RP and NA, respectively. Multiple bands detected by immunoblotting of NhlF, RP, SA, KP and NA may be due to incomplete denaturation by dodecylsulfate or/and partial proteolysis. C, Coomassie-stained gel; I, immunoblot.

Different substrate preferences of class II and class III NiCoTs were also obvious in inhibition experiments. In these assays, 63Ni2+ and 57Co2+ accumulation was analyzed in the presence of non-labeled metal ions as potential inhibitors. Neither 63Ni2+ nor 57Co2+ uptake was significantly affected by a 10-fold excess of copper, manganese or zinc ion (not shown). Fig. 1 (right panels) demonstrates that class II NiCoTs prefer cobalt even in the presence of a 10-fold excess of nickel. The converse is true for the class III permeases. 63Ni2+ uptake in the presence of a 10-fold excess of Co2+ surpasses 57Co2+ uptake in the absence of Ni2+.

The long-term assay used in the present study allows accurate qualitative comparisons of transport properties but is not suited for the calculation of the kinetic constants in quantitative terms. Nevertheless, some of the analyses illustrated in Fig. 1 predict that the different preferences of certain NiCoTs for Ni2+ or Co2+ ion are caused by a multistep process and are not merely the consequence of different affinities for the two metals in an initial binding reaction. YE, for example, transports three-fold more nickel than cobalt. Nevertheless, YE-mediated nickel transport is inhibited by 40% at an equimolar concentration of cobalt. Vice versa, only a slight effect of an equimolar amount of nickel on cobalt accumulation was observed. Since previous analysis of NhlF by site-directed mutagenesis suggested that Ni2+ and Co2+ are transported via the same route, it is conceivable that substrate specificity is the result of multiple coordination events during ion movement through the permeases.

Sebbane et al. [14] have reported that nickel transport mediated by the NiCoT of Y. pseudotuberculosis (designated UreH) is not significantly affected by cadmium, cobalt, copper, manganese and zinc ions, when the latter are supplied at a 30-fold excess. They concluded that this permease is selective for nickel. Our data clearly indicate that the closely related YE transports both nickel and cobalt ion, although a preference for nickel was obvious (Fig. 1). Cobalt transport by YE may be an adaptation to specific transition metal requirements of Y. enterocolitica. This bacterium, in contrast to Y. pseudotuberculosis and Y. pestis, has been shown to catalyze cobalamin synthesis and cobalamin-dependent 1,2-propanediol fermentation, encoded on a 40-kb metabolic island ([25] and M.B. Prentice, Medical Microbiology, Barts and the London Medical School, UK, personal communication).

3.2 Genomic localization of NiCoT genes

In a very recent comparative analysis of cobalamin synthesis gene clusters (cob/cbi) in prokaryotic genomes, a multitude of potential cobalt transporters belonging to various families including the NiCoT family has been predicted. Predictions were mainly based on the localization of the putative transporter genes in the vicinity of cob/cbi genes, or/and on the presence of so-called regulatory B12 elements and CBL boxes in promoter upstream regions of bacteria and archaea, respectively [5]. In this survey, NA and RP have been classified as cobalt-specific permeases. N. aromaticivorans and R. palustris contain complete sets of cobalamin biosynthesis and vitamin B12 transport (btu) genes [5]. The NA and RP genes are located adjacent to cob/cbi-btu clusters (Table 2). Our experimental data, shown in Fig. 1, demonstrate that NA and RP are class II NiCoTs and indeed have a preference for cobalt. Nevertheless, since the two permeases are capable of transporting nickel with high affinity, they cannot be designated ‘cobalt specific’.

View this table:

Correlation of the genomic localization of bacterial NiCoT genes with substrate specificity of the permeases

NiCoTSubstrate preference5′ or 3′ genomic region associated with synthesis of
RPCo>NiCoenzyme B12 (3′)
NACo>NiCoenzyme B12 (3′)
NhlFCo>NiCo-containing nitrile hydratase (5′)
YENi>CoABC-type Ni permease, urease, urea transporter (5′)
KPNi>Co[NiFe] hydrogenase (5′)
HoxNNi[NiFe] hydrogenase (5′)
SANi>CoApparently not related to Ni or Co metabolism
STNi>CoApparently not related to Ni or Co metabolism

A correlation between the genomic localization of the transporter gene and the substrate preference of the encoded NiCoT is also obvious in the cases of HoxN, NhlF, YE and KP (Table 2). HoxN is a selective nickel transporter. Its structural gene is located on megaplasmid pHG1 in R. eutropha H16 downstream of a large operon encoding two [NiFe] hydrogenases [22]. nhlF is a component of the nhl gene cluster encoding the ‘low-molecular-mass’, non-corrin cobalt-containing nitrile hydratase in Rhodochrous rhodochrous J1 [12] and correspondingly, NhlF uses cobalt as its preferred substrate. The YE gene, like its homologs in Y. pseudotuberculosis and Y. pestis, is part of a unique urea metabolism locus. The ure genes encoding the nickel-dependent urease are flanked by an ABC-type nickel transport operon on one side, and by a putative urea transporter gene and the adjacent class III NiCoT gene on the other ([14] and Table 2). KP, another class III NiCoT, is encoded downstream of [NiFe] hydrogenase genes in K. pneumoniae. The genes for SA and ST, two other class III NiCoTs, are apparently unlinked to nickel or cobalt metabolism. Taken together these findings show that the genomic localization of its gene is, in many cases, an indicator of the preferred substrate of the respective NiCoT.

3.3 Phylogenetic relationship of NiCoTs

To compare the relationship of NiCoTs, 23 full-length bacterial, archaeal and fungal sequences were aligned with CLUSTALX. A dendrogram is shown in Fig. 3. In general, the clustering of NiCoTs correlates with the phylogenetic tree of their hosts, suggesting that the permeases belong to an ancient family of proteins. An exception is NixA of the ?-proteobacterium H. pylori, which clusters with the NiCoTs of the low-GC Gram-positive bacteria S. aureus and Leuconostoc mesenteroides. In the present study, substrate specificity of six proteobacterial NiCoTs was compared (Fig. 1). These data, in combination with the dendrogram, suggest that minor sequence variations affect selectivity and transport velocity as an adaptation to the specific physiological requirements of a given organism. Differences in selectivity for Ni and Co ion seem to result from minor variations of a common theme rather than parallel evolution of two or more lines of NiCoTs with different basic properties. This conclusion is in agreement with recent experimental data demonstrating that single amino-acid replacements can affect selectivity and transport velocity of HoxN and NhlF [11].


We thank Bärbel Friedrich (Berlin) for longterm support, Edward Schwartz (Berlin) for critical reading of the manuscript and Jennifer Suhr (Berlin) for assistance during construction of pCH675-NA. The work was financially supported by the Deutsche Forschungsgemeinschaft (grant EI 374/2-1 to T.E.). We are grateful to Faith Harrison and Caroline Harwood (Iowa City, IA, USA), Mikael Skurnik (Turku, Finland), John Iandolo (Oklahoma City, OK, USA), Steffen Porwollik and Michael McClelland (San Diego, CA, USA), David Balkwill (Tallahassee, FL, USA), Margaret Romine (Richland, WA, USA), and Roland Brosch (Paris, France) for providing bacterial strains and/or total or cloned DNA thereof. Sequence data were taken from the databases maintained at the National Center for Biotechnology Information (Bethesda, MD, USA), the Wellcome Trust Sanger Institute (Cambridge, UK), the Washington University School of Medicine Genome Center (St. Louis, MO, USA), and the DOE Joint Genome Institute (Walnut Creek, CA, USA).


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
View Abstract