CzcP is a novel efflux system contributing to transition metal resistance in Cupriavidus metallidurans CH34


*E-mail; Tel. (+49) 345 5526352; Fax (+49) 345 5527010.


Cupriavidus metallidurans CH34 possesses a multitude of metal efflux systems. Here, the function of the novel PIB4-type ATPase CzcP is characterized, which belongs to the plasmid pMOL30-mediated cobalt-zinc-cadmium (Czc) resistance system. Contribution of CzcP to transition metal resistance in C. metallidurans was compared with that of three PIB2-type ATPases (CadA, ZntA, PrbA) and to other efflux proteins by construction and characterization of multiple deletion mutants. These data also yielded additional evidence for an export of metal cations from the periplasm to the outside of the cell rather than from the cytoplasm to the outside. Moreover, metal-sensitive Escherichia coli strains were functionally substituted in trans with CzcP and the three PIB2-type ATPases. Metal transport kinetics performed with inside-out vesicles identified the main substrates for these four exporters, the Km values and apparent turn-over numbers. In combination with the mutant data, transport kinetics indicated that CzcP functions as ‘resistance enhancer’: this PIB4-type ATPase exports transition metals Zn2+, Cd2+ and Co2+ much more rapidly than the three PIB2-type proteins. However, a basic resistance level has to be provided by the PIB2-type efflux pumps because CzcP may not be able to reach all different speciations of these metals in the cytoplasm.


Cupriavidus (Ralstonia, Wautersia, Alcaligenes) metallidurans is a metal-resistant β-proteobacterium of the family Burkholderiacease (Goris et al., 2001; Vandamme and Coenye, 2004; Vaneechoutte et al., 2004; von Rozycki and Nies, 2009). This bacterium survives in various mesophilic metal-contaminated environments because it possesses a variety of transition metal efflux systems, which are mainly encoded by two plasmids and the smaller of the two chromosomes (Diels and Mergeay, 1990; Mergeay, 2000; Mergeay et al., 2003; von Rozycki and Nies, 2009). Prominent among these metal resistance systems is the cobalt-zinc-cadmium resistance determinant czc on plasmid pMOL30 (Nies et al., 1987; 1989; Nies and Silver, 1989).

Metal resistance of C. metallidurans is based on metal efflux (Nies and Silver, 1989), which seems to be a two-step process. First, efflux systems that belong to the cation diffusion facilitator protein family CDF (TC 2.A.4) or the P-type ATPases (TC 3.A.3) transport surplus Zn2+, Cd2+ or Co2+ from the cytoplasm to the periplasm (Fagan and Saier, 1994; Paulsen and Saier, 1997; Busch and Saier, 2002; Saier et al., 2006). Second, RND-driven efflux systems (resistance, nodulation, cell division protein superfamily, TC 2.A.6) such as CzcCBA probably export these metals from the periplasm to the outside of the cell (Nies, 2003; 2007). This two-step process by transporters with overlapping substrate specificity may apply to export of toxic substances in Gram-negative bacteria in general (Tal and Schuldiner, 2009).

RND-driven efflux systems are multisubunit protein complexes that span the cytoplasm and the cell wall of a Gram-negative bacterium (Nies, 2003; 2007). The actual trimeric RND protein (e.g. CzcA) resides in the cytoplasmic membrane, binds the substrate and drives the transport process by using the proton motive force (Saier et al., 1994; Nies, 1995; Goldberg et al., 1999; Murakami et al., 2006; Seeger et al., 2006). It delivers the substrate through the tube-like, trimeric OMF (CzcC, outer membrane factor, TC 1.B.17) to the outside of the cell (Koronakis et al., 2000). Interaction of RND and OMF proteins is stabilized by the third component, the membrane fusion protein (CzcB, MFP, 8.A.1), which forms a trimeric (Symmons et al., 2009) or hexameric (Yum et al., 2009) ring around the contact side of RND and OMF in the periplasm (Higgins et al., 2004). MFP proteins also might have a catalytic in addition to their structural role (Bagai et al., 2008).

RND proteins belonging to different protein families are involved in export of organic substances including antibiotics, or transition metal cations (Tseng et al., 1999). All evidence obtained to date for both kinds of proteins are in agreement with the hypothesis that RND-driven protein complexes export their substrates from the periplasm to the medium (outer membrane efflux) rather than from the cytoplasm to the outside of the cell (transenvelope efflux) (Lomovskaya and Totrov, 2005; Murakami et al., 2006; Seeger et al., 2006). One piece of evidence in favour of the outer membrane efflux hypothesis comes from experiments that characterize the interplay of CDF proteins with RND-driven efflux systems in conferring metal resistance to C. metallidurans (Munkelt et al., 2004).

Cupriavidus metallidurans contains three CDF proteins that belong to three different CDF subfamilies (Nies, 2003; von Rozycki et al., 2005). CzcD, a founding member of this protein family, is part of the Czc resistance system and transports Zn2+, Cd2+ and Co2+ (Nies, 1992; Anton et al., 1999; 2004). FieF is an orthologue of the Fe2+ or Zn2+ efflux protein FieF/YiiP from Escherichia coli (Grass et al., 2005; Lu and Fu, 2007). DmeF has a broad substrate specificity with the main efflux substrates Co2+ and Ni2+ (Munkelt et al., 2004). A mutant derivative of C. metallidurans devoid of CzcD and of DmeF shows strongly decreased cobalt resistance, although the CzcCBA efflux complex is present in the cell wall (Munkelt et al., 2004): without a CDF protein to transport Co2+ first across the cytoplasm to the periplasm, further detoxification of the periplasm by CzcCBA was obviously not possible (Munkelt et al., 2004).

P-type ATPases are ATP-driven importers and exporters, predominantly of cations (Fagan and Saier, 1994). The name stems from a phosphorylated intermediate during the catalytic cycle. The best studied example is the calcium-transporting protein of the sarcoplasmic reticulum that is involved in muscle flexing (Toyoshima et al., 2004). Members of the PIB or CPx family of these P-type ATPases contain a conserved, essential proline residue, which is flanked by at least one cysteine residue N- or C-terminal to the proline. These proteins transport transition metal cations with one group (PIB1) transporting monovalent cations such as Cu+ and Ag+ and another (PIB2) divalent cations such as Zn2+, Cd2+ and Pb2+ (Nies, 2003). Both groups of proteins posses a conserved CPC motif in one of the transmembrane spans (Argüello et al., 2007). This CPC motif might be part of the metal binding site that is occupied by the cationic substrate when the enzyme-substrate complex is being formed (Gonzalez-Guerrero and Argüello, 2008). Another group of CPX-type ATPases (PIB4) contains a conserved SPC motif with the cystein residue located closer to the cytoplasm than to the outside (Argüello et al., 2007). While clear substrate profiles could be assigned to PIB1 and PIB2, function and substrate specificity of PIB4-type ATPases is enigmatic and includes Co2+, Cu+, Ca2+/heavy metals (Rutherford et al., 1999; Seigneurin-Berny et al., 2006; Moreno et al., 2008).

Cupriavidus metallidurans contains three PIB2-type ATPases: ZntA, CadA and PbrA (Nies, 2003; von Rozycki et al., 2005; von Rozycki and Nies, 2009). The contribution of these proteins to metal resistance in this bacterium has been characterized (Borremans et al., 2001; Legatzki et al., 2003a,b). The newly identified CzcP is related to those three (von Rozycki and Nies, 2009). CzcP was especially interesting because it shows sequence signatures of a Co2+-exporting protein (Rutherford et al., 1999) and therefore has been assigned to the subgroup PIB4 instead of PIB2 (Argüello et al., 2007). If CzcP is indeed a cobalt-exporting enzyme in vivo, then this would call into question both the essential contribution of the two CDF proteins CzcD and DmeF to cobalt resistance in C. metallidurans and the evidence in favour of an outer membrane efflux by metal-exporting RND-driven systems. We present evidence substantiating our hypothesis of outer membrane efflux catalysed by RND proteins and an unexpected new category of job-sharing between PIB2-type ATPases on one hand and the novel ‘fast-track’ PIB4-type ATPase CzcP on the other.


Contribution of the various metal efflux systems to cobalt, zinc and cadmium resistance in C. metallidurans

The well-characterized czc determinant on plasmid pMOL30 is organized in two divergently transcribed gene clusters (Fig. 1), one of which, the czcNICBADRSE cluster, encodes the RND-driven efflux system CzcCBA, the CDF protein CzcD, the two-component regulatory system CzcRS, the putative periplasmic proteins CzcI and CzcE, and the uncharacterized membrane-bound protein CzcN (van der Lelie et al., 1997; Anton et al., 1999; Große et al., 1999; 2004). Adjacent to this divergon is a second one and both of them, together with three additional putative genes, are flanked by insertion sequences. Thus, this complicated gene cluster could have originated from a transposon and the products of all these genes may be involved in metal resistance. The second divergon contains the gene for a PIB4-type ATPase [originally annotated as ‘cadA’, re-named here czcP because the name ‘cadA’ was already used for a PIB2-type ATPase (Legatzki et al., 2003a)].

Figure 1.

The czc determinant and its neighbours on plasmid pMOL30. The region of 70–97 kb of megaplasmid pMOL30 (accession number NC_006525) is shown. The dashed vertical lines are 1 kb distance markers. Known czc genes are coloured in red, additional genes putatively involved in transport in green, others in blue. If genes had not names, the respective ‘rme’ number of an older gene base accession is indicated. The czc region is flanked by two putative divergons that both contain insertion element-specific genes (int and rme0292, purple colour, recently re-named into pMOL30-056 and pMOL30-076 respectively). One of the two putative divergons contains in the middle the immediate czc region (rme0273-mgtC <–> czcNICBADRSE, rme0273 recently renamed into pMOL30-059). The other one contains czcJ (= rme0284, pMOL30-070) encoding a new metal-induced periplasmic protein (J. Scherer and D. H. Nies, unpubl. results), rme0285 (pMOL30-071) encoding a putative porin protein, and czcP (annotated as ‘cadA’, pMOL30-073). The latter is under the control of the CzcSR two-component regulatory system and encodes a PIB4-type ATPase.

To characterize the contribution of CzcP to metal resistance in C. metallidurans, a ΔczcP deletion mutant was constructed. Moreover, single and multiple deletion mutations were constructed in the genes zntA, cadA (both PIB2-type ATPases), czcD, dmeF, fieF (three CDF-proteins) and czcA (RND protein). The lead resistance determinant pbr that contains another PIB2-type ATPase (PbrA) on plasmid pMOL30 was completely deleted. These mutants were obtained from C. metallidurans strain AE128(pMOL30) (Mergeay et al., 1985), which harbours plasmid pMOL30 with czc, pbr and other determinants but not plasmid pMOL28 with the cnr nickel-cobalt and chr chromate resistance determinants (Table 1). Moreover, in case of the chromosomal genes zntA, cadA, dmeF and fieF, mutants were also generated from the plasmid-free strain AE104 (Mergeay et al., 1985). Metal resistance of this mutant library was determined in dose–response curves, which are parallel liquid cultures with increasing metal concentrations in non-complexing mineral salts medium (shown for one example in Fig. 2 and data not shown). For each culture and metal, the concentration that inhibits growth by half (IC50) was calculated (Fig. 2, Table 2). Moreover, the mathematical treatment of the data yielded another characteristic feature of the curves, the b-value (Table 2). Theoretically, ‘c = IC50 − 2.2·b’ is the metal ion concentration c that inhibits growth by 10% and ‘c = IC50 + 2.2·b’ that which inhibits growth by 90%. So, the steepness of the inhibition curves increases with decreasing b-value.

Table 1.  Bacterial strains and plasmids used.
NameRelevant markersReference
Bacterial strains  
 Cupriavidus metallidurans  
  AE128pMOL30Mergeay et al. (1985)
  AE104Plasmid-freeMergeay et al. (1985)
  DN178pMOL30-10, ΔczcRGroße et al. (1999)
  DN179pMOL30-11, ΔczcSGroße et al. (1999)
  DN432pMOL30-38, ΔczcALegatzki et al. (2003b)
  DN438Plasmid-free, ΔcadALegatzki et al. (2003a)
 Escherichia coli  
  W3110Wild-typeGrass et al. (2001)
  S17/1Conjugator strainSimon et al. (1983)
  GG48ΔzntA::kanΔzitB::catGrass et al. (2001)
  GR362ΔzntA::kanΔzntB::catΔzitBΔzupTΔznuABCGrass et al. (2002)
  ECA349ΔyohMΔcorAΔzupT::catKoch et al. (2007)
  BL21pLysSStratagene Europe, Amsterdam, the Netherlands
 pASK-IBA3E. coli vectorIBA GmbH, Göttingen, Germany
 pLO2 Lenz et al. (1994)
 pECD794lacZ, derivate of pLO2This study
 pCM157Contains Cre recombinaseMarx and Lidstrom (2002)
 pCM184Recombination vector cre-lox systemMarx and Lidstrom (2002)
 pBBR1-MCS-2Broad host range expression vectorKovach et al. (1995)
 pVDZ′2Broad host range expression vectorDeretic et al. (1987)
 pGEM® T-EasylacPOZ, cloning vectorPromega, Madison, WI, USA
 pDNA130pVDZ′2::czcCBAD′Nies et al. (1989)
 pECD889lacZ, sacB, derivate of pCM184Große et al. (2007)
 pECD1003Mutant lox sites, derivate of pECD889This study
 pECD799cadA in pASK-IBA3Legatzki et al. (2003a)
 pECD1014cadA2 in pASK-IBA3This study
 pECD1015cadA3 in pASK-IBA3This study
 pECD1016cadA4 pASK-IBA3This study
 pECD800zntA in pASK-IBA3Legatzki et al. (2003a)
 pECD1013zntA2 in pASK-IBA3This study
 pECD1010czcP in pASK-IBA3This study
 pECD1011czcP2 in pASK-IBA3This study
 pECD1012pbrA in pASK-IBA3This study
 pECD1009czcR in pASK-IBA3This study
 pECD850pECD885 construct for dmeF deletion by cre-lox systemThis study
 pECD888pECD889 construct for zntA deletion by cre-lox systemThis study
 pECD1020pECD885 construct for czcP deletion by cre-lox systemThis study
 pECD1023pECD1003 construct for fieF deletion by cre-lox systemThis study
 pECD1024pECD1003 construct for pbrT-pbrD deletion by cre-lox systemThis study
 pECD1025pECD1003 construct for czcD deletion by cre-lox systemThis study
 pECD1027pECD1003 construct for cadA deletion by cre-lox systemThis study
 pECD1019pCM184 construct for czcR disruptionThis study
 pECD1028czcP′ in pECD794This study
 pECD851czcE′ in pECD794This study
 pECD1017czcR in pVDZ′2This study
Figure 2.

Dose–response curve. C. metallidurans strains AE128(pMOL30) wild-type (closed circles, ●) and its mutant derivatives ΔczcP (closed squares, inline image), ΔczcA (open squares, □) and ΔczcPΔcadAΔzntAΔpbrΔdmeFΔfieFΔczcD::km (open circles, ○) were cultivated in parallel cultures with increasing Zn2+ concentrations in Tris-buffered mineral salts medium at 30°C with shaking for 20 h. Turbidity was determined at 600 nm. The data points shown are the mean values of 4–7 experiments with the deviation bars shown. The individual growth curves were used to calculate the IC50 and b-values of each curves. The mean IC50 and b-values were derived for each mutant strain and used to calculate the theoretical turbidities at the various zinc concentrations for each strain. The resulting theoretical data points were connected by the dashed lines.

Table 2.  Metal resistance of C. metallidurans mutant strains.a
Bacterial strainCodebIC50, b-value (μM)
  • a. 

    Metal resistance was determined in at least three independent dose–response curves (parallel cultures with increasing concentration of the metal cation indicated). Growth was measured as turbidity after 20 h of cultivation with shaking at 30°C in Tris-buffered mineral salts medium. For each growth curve, the IC50 was calculated and the slope ‘b’ of the curve. A value of IC50 − 2.2·b indicates a 10% growth inhibition of the cells, IC50 + 2.2·b a 90% inhibition. The cells of the table give the mean IC50 values (first in each cell, ± deviation) and the b-value (± deviation in italics).

  • b. 

    To facilitate orientation especially when multiple mutant strains are concerned, the mutant code indicates the presence of CzcA (A), ZntA (Z), CadA (C), CzcP (P), Pbr (B), CzcD (D), DmeF (M), FieF (F) or residual plasmid pMOL30-encoded factors (X) in the respective mutant strain.

AE128(pMOL30)AZCPBDMFX3362 ± 127, 307 ± 64236 ± 30, 36 ± 72145 ± 330, 374 ± 123
ΔczcPAZC_BDMFX2310 ± 136, 477 ± 86124 ± 35, 48 ± 252111 ± 317, 596 ± 475
ΔpbrAZC_BDMFX3498 ± 345, 266 ± 44279 ± 20, 72 ± 152227 ± 226, 479 ± 143
ΔzntAA_CPBDMFX3052 ± 95, 127 ± 55262 ± 15, 58 ± 172209 ± 281, 339 ± 159
ΔcadAAZ_PBDMFX3533 ± 218, 189 ± 122257 ± 24, 67 ± 282001 ± 147, 284 ± 124
ΔczcPΔzntAA_C_BDMFX2496 ± 149, 220 ± 83139 ± 15, 54 ± 202052 ± 180, 462 ± 81
ΔczcPΔpbrAZC__DMFX2687 ± 125, 186 ± 43138 ± 16, 56 ± 62107 ± 403, 526 ± 216
ΔcadAΔzntAA__PBDMFX3012 ± 416, 306 ± 218101 ± 19, 39 ± 161972 ± 132, 337 ± 103
ΔczcPΔzntAΔpbrA_C__DMFX2367 ± 103, 244 ± 97151 ± 11, 46 ± 51748 ± 172, 406 ± 165
ΔczcPΔcadAΔzntAA___BDMFX2163 ± 58, 196 ± 10235.0 ± 13.5, 7.3 ± 1.12106 ± 143, 647 ± 243
ΔcadAΔzntAΔpbrA__P_DMFX2758 ± 165, 250 ± 1437.60 ± 2.00, 4.71 ± 1.012028 ± 398, 647 ± 269
ΔczcPΔcadAΔzntAΔpbrA____DMFX2090 ± 16, 164 ± 223.50 ± 0.46, 3.53 ± 1.882256 ± 255, 752 ± 333
ΔdmeFAZCPBD_FX3666 ± 242, 326 ± 105n.d.9.10 ± 3.21, 9.26 ± 9.00
ΔdmeFΔfieFAZCPBD__X3699 ± 304, 304 ± 69286 ± 6, 53 ± 1410.2 ± 4.4, 7.0 ± 10.6
ΔdmeFΔczcDAZCPB-_FXn.d.n.d.10.1 ± 2.8, 10.1 ± 8.4
ΔczcPΔzntAΔcadAΔdmeFA___BD_FX2273 ± 174, 331 ± 61n.d.n.d.
ΔcadAΔzntAΔdmeFΔfieFA__PBD__X2937 ± 220, 293 ± 93100 ± 31, 72 ± 356.07 ± 1.89, 3.18 ± 1.76
ΔczcPΔpbrΔczcD::kmAZC___MFX2460 ± 255, 310 ± 142n.d.2095 ± 122, 212 ± 83
ΔczcPΔcadAΔzntAΔpbrΔdmeFΔfieFΔczcD::kmA_______X1273 ± 116, 260 ± 1311.52 ± 0.61, 0.80 ± 0.428.43 ± 0.71, 2.68 ± 0.53
ΔczcA_ZCPBDMFX867 ± 148, 429 ± 120132 ± 14, 42 ± 11243 ± 40, 98 ± 74
ΔczcAΔzntA__CPBDMFX438 ± 83, 402 ± 289n.d.225 ± 53, 110 ± 50
ΔczcAΔczcP_ZC_BDMFX605 ± 27, 643 ± 5892.0 ± 19.0, 19.0 ± 9.5204 ± 55, 178 ± 161
ΔczcAΔdmeF_ZCPBD_FXn.d.n.d.39.2 ± 14.8, 27.1 ± 21.3
ΔczcAΔdmeFΔczcD_ZCPB__FXn.d.n.d.17.2 ± 6.9, 12.3 ± 6.1
ΔczcAΔdmeFΔczcP_ZC_BD_FXn.d.n.d.12.3 ± 3.3, 16.3 ± 12.4
ΔczcAΔdmeFΔczcDΔczcP_ZC_B__FXn.d.n.d.14.7 ± 3.8, 4.3 ± 1.1
ΔczcAΔcadAΔzntA___PBDMFX369 ± 133, 352 ± 146,51.0 ± 9.5, 15.0 ± 10.7,253 ± 33, 122 ± 79
ΔczcAΔczcPΔpbr_ZC__DMFX564 ± 152, 417 ± 9842.2 ± 17.9, 26.0 ± 8.2234 ± 32, 78 ± 52
ΔczcAΔzntAΔcadAΔczcP::km____BDMFX253 ± 46, 278 ± 7022.8 ± 2.4, 9.2 ± 6.7n.d.
ΔczcAΔzntAΔcadAΔpbr___P_DMFX375 ± 138, 580 ± 3670.83 ± 0.14, 0.28 ± 0.15n.d.
ΔczcAΔcadAΔzntAΔdmeFΔfieF___PBD__X266 ± 136, 341 ± 11347.8 ± 5.1, 8.8 ± 1.58.90 ± 1.36, 6.42 ± 3.13
ΔczcAΔpbrΔcadAΔzntAΔdmeF fieF::km___P_D__X373 ± 89, 541 ± 23746.5 ± 2.6, 8.4 ± 3.95.06 ± 0.48, 2.26 ± 3.38
ΔczcAΔczcPΔcadAΔzntAΔdmeFΔfieF____BD__X249 ± 81, 135 ± 11622.3 ± 1.3, 4.6 ± 2.4n.d.
ΔczcAΔcadAΔzntAΔdmeFΔfieFΔczcD___PB___X232 ± 121, 420 ± 5246.5 ± 2.6, 8.4 ± 3.97.05 ± 1.33, 3.68 ± 1.55
ΔczcAΔczcPΔcadAΔzntAΔpbrΔdmeFΔfieF_____D__X317 ± 118, 477 ± 670.326 ± 0.174, 0.181 ± 0.1119.13 ± 1.77, 9.18 ± 7.61
ΔczcAΔcadAΔzntAΔpbrΔdmeFΔfieFΔczcD___P____X42.3 ± 5.2, 18.5 ± 10.70.402 ± 0.114, 0.115 ± 0.0567.38 ± 3.08, 3.24 ± 2.02
ΔczcAΔczcPΔcadAΔzntAΔdmeFΔfieFΔczcD____B___X108 ± 19, 32 ± 51.78 ± 0.64, 0.84 ± 0.286.40 ± 1.18, 3.52 ± 2.99
ΔczcAΔczcPΔcadAΔzntAΔpbrΔdmeFΔfieFΔczcD________X19.5 ± 1.3, 3.8 ± 0.50.491 ± 0.048, 0.259 ± 0.2087.73 ± 0.84, 2.75 ± 0.79
AE104_ZC___MF_1056 ± 28, 468 ± 8791.1 ± 17.0, 31.8 ± 12.7163 ± 9, 21 ± 10
ΔzntA__C___MF_481 ± 216, 831 ± 119091.2 ± 22.2, 35.3 ± 26.4179 ± 16, 30 ± 17
ΔcadA_Z____MF_1156 ± 101, 224 ± 8178.6 ± 7.9, 18.4 ± 4.6174 ± 18, 42 ± 16
ΔcadAΔdmeF_Z_____F_1026 ± 24, 175 ± 1676.8 ± 9.5, 17.2 ± 5.54.65 ± 0.79, 1.16 ± 0.68
ΔcadAΔzntA______MF_7.7 ± 0.6, 1.3 ± 0.50.193 ± 0.010, 0.048 ± 0.014184 ± 26, 53 ± 25
ΔcadAΔzntAΔfieF______M__7.30 ± 0.52, 1.70 ± 0.830.183 ± 0.015, 0.060 ± 0.01496.6 ± 12.4, 16.9 ± 10.2
ΔcadAΔzntAΔdmeF_______F_7.53 ± 0.24, 1.54 ± 0.390.207 ± 0.026, 0.054 ± 0.0165.42 ± 1.49, 1.49 ± 1.30
ΔdmeF_ZC____F_1158 ± 125, 264 ± 57129 ± 14, 65 ± 314.88 ± 0.77, 1.11 ± 0.64
ΔfieF_ZC___M__987 ± 220, 275 ± 7981.8 ± 1.7, 24.3 ± 20.592.3 ± 7.4, 12.5 ± 4.0
ΔdmeFΔfieF_ZC______1066 ± 60, 252 ± 2478.8 ± 8.6, 42.8 ± 27.75.06 ± 0.56, 0.97 ± 0.77
ΔcadAΔzntAΔdmeFΔfieF_________7.1 ± 0.7, 1.3 ± 0.80.082 ± 0.012, 0.027 ± 0.0054.92 ± 0.40, 1.48 ± 1.28

To analyse the contribution of the single resistance systems to zinc, cadmium and cobalt resistance, all the IC50 values of the mutant were compared in a pairwise manner and the increase in IC50, Q(IC50), was assigned to the system missing in one mutant and present in the other. The resulting system/Q(IC50) pairs were ranked by the Q(IC50) values and are given in Table S1. In this way, it could be evaluated if a single efflux system was always involved in a metal resistance, never or sometimes, and to what degree.


All four PIB2/4-type ATPases contributed to zinc resistance in some mutant comparisons, but not in others (Table 2, Table S1). Most important for zinc resistance was ZntA, which was responsible for a strong (up to 150-fold) increase in zinc resistance, as evidenced from the comparison of strain AE104 ΔcadA (IC50 = 1.2 mM) and strain AE104 ΔcadAΔzntA (IC50 = 7.7 μM, Table 2). In this case, ZntA was the only remaining PIB2/4-type ATPase present in the C. metallidurans cells. Deletion of zntA from strain AE104 wild-type (still containing CadA) decreased zinc resistance only 2.2-fold and its deletion in strains containing at least one other PIB2/4-type ATPase [e.g. AE128(pMOL30) wild-type, AE128(pMOL30) ΔczcP or AE128(pMOL30) ΔcadA] did not decrease zinc resistance significantly (Table 2). Thus, the other three PIB2/4-type ATPases PbrA, CzcP and CadA were able to compensate for a missing ZntA.

Deletion of cadA had only one significant effect (65-fold), evident through the comparison of AE104 mutant strains ΔzntA (IC50 = 481 μM) and ΔzntAΔcadA (IC50 = 7.7 μM, Table 2). Thus, CadA was able to compensate for a missing ZntA but only partially (IC50 of strain AE104 wild-type 1 mM). A strong effect of the deletion of pbr was observed in two instances, both times a 5.5-fold decrease in zinc resistance was noted (Table 2). This decrease occurred in AE128(pMOL30) mutant strains carrying PbrA as the only remaining PIB2-type ATPase or as the only one in addition to the PIB4-type ATPase CzcP. Thus, CadA and PbrA should be able to export Zn2+ as substitutes of ZntA.

Contribution of CzcP to zinc resistance was different from that of the other three proteins. On the one hand, when multiple mutant strains were compared, deletion of czcP had smaller effects in the various C. metallidurans single mutant backgrounds than deletion of the genes for the three PIB2-type ATPases. The strongest effect was a 2.2-fold decrease occurring when CzcP was the only remaining PIB2/4-type ATPase in an AE128(pMOL30) mutant strain (Table 2, see mutant codes ‘BX’ versus ‘X’; mutant codes were used in multiple deletion strains instead of the more complicated deletion listings). On the other hand, when the four ΔczcP, ΔzntA, Δpbr and ΔcadA single mutant derivatives of strain AE128(pMOL30) were evaluated, the ΔczcP strain was the only one with a significant decrease in zinc resistance (Table 2). Thus, the PIB4-type ATPase CzcP should also be able to transport Zn2+, more efficiently than the PIB2-type ATPases when at least one of these proteins was present but less when the PIB2-type proteins were absent: CzcP enhanced resistance that was basically provided by other systems.

Deletion of czcA for the central RND protein of the CzcCBA efflux pump always led to a strong decrease in zinc resistance (3.9- to 65-fold, Table 2 and Table S1). Therefore, loss of CzcA could not be functionally substituted by a PIB2/4-type ATPase or another system. The strongest decrease was noted when CzcCBA was the only remaining zinc efflux system in the comparison of the AE128(pMOL30) mutants with the mutant codes ‘AX’, IC50 = 1.27 ± 0.12 mM, versus ‘X’, IC50 = 19.5 ± 1.3 μM (Table 2). The smallest decrease (3.9-fold) in zinc resistance was connected to deletion of czcA from the AE128(pMOL30) ΔczcP mutant strain, indicating again a special role of CzcP in comparison to the three PIB2-type ATPases. All ΔczcA mutants, with the exception of those carrying the highest number of deletions (Table 2, mutant codes ‘PX’, ‘BX’ and ‘X’), showed very high b-values in their dose–response curves. In a few cases, the b-values were even higher than the IC50 values, which was an artifact of the algorithm used for calculation. Nevertheless, this indicated growth effects already at the lowest concentrations tested, in combination with residual growth even at concentrations well above the IC50 value.

Deletion of the czcD gene, which encodes a czc-associated CDF protein, decreased zinc resistance 16-fold when the genes for all four PIB2/4-type ATPases were already gone (Table 2, mutants with the codes ‘DX’, IC50 = 317 μM, and ‘X’, IC50 = 19 μM) or only PbrA was present (‘BDX’, IC50 = 249 μM, and ‘BX’, IC50 = 108 μM). Thus, CzcD could complement the function of the PIB2/4-type ATPases and, vice versa, a missing CzcD could also be functionally substituted by the ATPases. This indicated that CzcD and the four PIB2/4-type ATPases performed a similar function with respect to zinc export while CzcA performed a different one.

In contrast to CzcD, deletion of dmeF or fieF, encoding the other two CDF proteins of C. metallidurans, had no effect on zinc resistance [Table 2 and s1, Q(IC50) = 1.4 for all comparisons concerning DmeF and/or FieF]. Finally, there was a clear difference in zinc resistance [Q(IC50) = 2.7, Table S1] when all the AE128(pMOL30) and AE104 deletion strains were compared (mutant code ‘X’, IC50 = 19.5 ± 1.3 μM versus ‘_’, IC50 = 7.1 ± 0.7 μM, Table 2). This indicated a contribution to zinc resistance of some unknown plasmid-encoded factors (designated ‘PlsX’), e.g. copper resistance products like CupA (Taghavi et al., 2009) or periplasmic proteins of the Czc system.


All nine systems (CzcA, the four PIB2/4-type ATPases, the three CDF proteins and ‘PlsX’) contributed to cadmium resistance and cooperatively increased the IC50 3000-fold from 82 nM to 236 μM (Table 2). The four PIB2/4-type ATPases were the most important contributors with CadA having the greatest contribution. Cadmium resistance decreased 472-fold when cadA was deleted from strain AE104 ΔzntA or 43-fold when it was the last remaining PIB2-type ATPase in strain AE128(pMOL30) (Table 2, mutant code ‘CMF’ compared with ‘MF’, and ‘ACDMFX’ compared with ‘ADMFX’).

On the other hand, the other three PIB2/4-type ATPases were able to complement for a missing CadA. Deletion of cadA from strain AE104 decreased cadmium resistance only 1.2-fold but nearly 410-fold when zntA was additionally deleted. This indicated that ZntA could nearly fully (86%) complement for CadA. However, deletion of zntA did not diminish cadmium resistance in the plasmid-free strain AE104. So, CadA was sufficient to sustain the cadmium resistance level of the plasmid-free strain. In strain AE128(pMOL30), deletion of cadA did not decrease cadmium resistance and subsequent deletion of zntA only by an additional 2.5-fold. Thus, PbrA and/or CzcP could also compensate for CadA and ZntA. The strongest effect of a pbr deletion was 116-fold when pbrA was deleted from a multiple deletion strain of AE128(pMOL30) that contained only pbrA, czcP and residual plasmid genes (Table 2, comparison of mutants code ‘PBX’, IC50 = 46 μM, and ‘PX’, IC50 = 0.4 μM). Subsequent deletion of czcP from this mutant did not decease cadmium resistance any further (Table 2, ‘X’, IC50 = 0.5 μM). The strongest effect of a czcP deletion was 26-fold and observed from the mutant with the code ‘PBX’ (comparison of mutants code ‘PBX’, IC50 = 46 μM, and ‘BX’, IC50 = 1.8 μM) that contains CzcP, PbrA and ‘PlsX’. Subsequent deletion of pbr from the resulting strain containing only PbrA and ‘PlsX’ diminished cadmium resistance 3.6-fold. Again, this indicates a special role for the PIB4-type ATPase CzcP as ‘resistance enhancer’: alone, CzcP was not able to mediate cadmium resistance but it was able to enhance cadmium resistance in the presence of one of the three PIB2-type ATPases.

The contribution of CzcA to cadmium resistance was less important than the contribution of this protein to zinc resistance and also less important than that of the PIB2/4-type ATPases to cadmium resistance. The strongest effect of a czcA deletion was 3.3-fold when it was additionally deleted from a AE128(pMOL30) ΔczcPΔpbr mutant (Table 2, ΔczcPΔpbr, IC50 = 138 μM, ΔczcAΔczcPΔpbr, IC50 = 42 μM). Interestingly, deletion of czcA from the AE128(pMOL30) ΔczcP single mutant strain had only a small 1.3-fold effect (Table 2, ΔczcP, IC50 = 124 μM, ΔczcAΔczcP, IC50 = 92 μM). This again indicated a special role of the PIB4-type ATPase CzcP that seems to cooperate more closely with the function of CzcA than the three PIB2-type ATPases. Moreover, this situation indicates that there is a difference between the contribution of CzcA to cadmium and zinc resistance because deletion of czcA always decreased zinc resistance more than 3.8-fold.

The contribution of the CDF protein CzcD to cadmium resistance was reminiscent of that of CzcP. The strongest effect (12.5-fold) of a czcD deletion was from an AE128(pMOL30) multiple deletion strain carrying only CzcD, PbrA and residual plasmid genes ‘PlsX’ (Table 2, mutants codes ‘BDX’, IC50 = 22.3 μM, and ‘BX’, IC50 =  1.8 μM). On the other hand, CzcD alone was not able to increase cadmium resistance (Table 2, mutant codes ‘DX’, IC50 = 326 nM, and ‘X’, IC50 = 491 nM). Thus, and in a similar situation as found for CzcP, CzcD alone was not effective but it could enhance cadmium detoxification mediated by another system, in this case PbrA. In addition to CzcD, the other two CDF proteins DmeF and FieF also mediated some residual cadmium resistance (Table 2, AE104 mutant strains code ‘M’, IC50 = 183 nM; ‘F’, IC50 = 207 nM; and the full deletion strain code ‘_’, IC50 =  82 nM). Finally, the difference in cadmium resistance of the AE128(pMOL30) and the AE104 full deletions strains indicate a sixfold contribution of the unknown plasmid-encoded factors ‘PlsX’ to cadmium resistance (Table 2, mutants code ‘X’ and ‘_’).

So, all four PIB2/4-type ATPases contributed to zinc and cadmium resistance and, compared with other systems, their contribution to cadmium resistance was more significant than their contribution to zinc resistance. In contrast, contribution of CzcA to zinc resistance was more significant than to cadmium resistance. CzcD was involved in both processes and displayed some similarity with CzcP in its action on cadmium resistance. Both proteins acted as resistance enhancers. DmeF and FieF contributed little to cadmium and nothing to zinc resistance. This situation was completely different when the influence of these factors on cobalt resistance was analysed.


The CDF protein DmeF was the most important factor for cobalt resistance. The strongest decrease in resistance (325-fold) occurred when dmeF was deleted from an AE128(pMOL30) ΔcadAΔzntA mutant strain (Table 2, IC50 = 1.97 mM, resulting mutant strain code ‘PBDX’, IC50 = 8.9 μM). All dmeF deletions caused minimally a 6.2-fold decrease in cobalt resistance (Table S1). In the presence of CzcA, all strains containing DmeF had a mean IC50 = 2081 ± 132 μM while all strains devoid of DmeF had a mean IC50 = 8.8 ± 1.7 μM.

The second most important for cobalt resistance was CzcA that increased resistance up to 10.3-fold in AE128(pMOL30) (Table 2, comparison of the mutant ΔczcP, IC50 = 2.1 mM, and ΔczcAΔczcP, IC50 = 204 μM). In the presence of DmeF, CzcA increased cobalt resistance in the five mutant pairs analysed by a factor of 9.15 ± 0.98-fold. However, in some cases, deletion of czcA did not alter cobalt resistance or even led to its increase in strain AE128(pMOL30) (Table 2, comparison of the mutants ΔdmeF, IC50 = 9.1 μM, and ΔczcAΔdmeF, IC50 = 39.2 μM). The mean IC50 of all ΔdmeF mutant strains tested was10 ± 9 μM in the absence of CzcA and therefore not much different from 8.8 ± 1.7 μM in the presence of CzcA. In the three ΔdmeF mutant pairs analysed, CzcA did not increase cobalt resistance. Thus, CzcA was able to contribute about 10-fold to cobalt resistance but only if DmeF was present in the cells.

CzcP and CzcD were also able to contribute to cobalt resistance (3.2-fold and 2.3-fold respectively) but each only in AE128(pMOL30 ΔczcAΔdmeF). The ΔdmeF strain displayed increased cobalt resistance after introducing the czcA mutation (Table 2, comparison of the mutants ΔdmeF, IC50 = 9.1 μM, and ΔczcAΔdmeF, IC50 =  39.2 μM). Additional deletion of czcP decreased cobalt resistance to IC50 = 12.3 μM, that of czcD to IC50 =  17.2 μM, while introduction of both mutations led to an IC = 14.7 μM (Table 2). Thus, the increased cobalt resistance level that originated from deletion of czcA in a ΔdmeF mutant strain was mediated to about 50% by CzcP or CzcD, and both proteins could compensate for each other in this function. However, neither CzcP nor CzcD alone mediated a significant level of cobalt resistance (Table 2, compare multiple mutants code ‘DX’, IC50 = 9.1 μM; ‘PX’, IC50 = 7.4 μM, and the full deletion strain code ‘X’, IC50 = 7.7 μM). Again, CzcP and CzcD did not cause an increase in resistance by themselves but enhanced the performance of another system. None of the three PIB2-type ATPase was involved in cobalt resistance, since there was no effect whatsoever when one of them was deleted. However, FieF or the unknown plasmid-encoded component(s) ‘PlsX’, which increased cobalt resistance up to 1.9-fold and 1.6-fold, respectively, might be involved to some degree.

Influence of constitutively expressed czcCBA on a ΔcadAΔzntAΔfieFΔdmeF AE104 mutant strain

Plasmid pDNA130 contains czcCBAD′ on plasmid pVDZ′2 (Deretic et al., 1987; Nies et al., 1990). In this plasmid, czcCBAD′ is expressed constitutively under control of the lac promoter, which leads to an expression level of 60 000 CzcA proteins per cell while the maximum number of CzcA proteins induced by Zn2+ is ∼23 400 CzcA/cell (Legatzki et al., 2003b). The 3′ end of the czcCBAD′ fragment contained a truncated czcD′ gene, which encodes only the 200 amino acyl N-terminal and membrane-bound part of CzcD. This truncated protein has a low level of transport activity (Anton et al., 2004).

Constitutive high-level expression of czcCBAD′ increased zinc resistance of strain AE104 from IC50 = 875 ± 384 μM to IC50 = 9007 ± 340 μM (10-fold) and that of the ΔcadAΔzntAΔfieFΔdmeF mutant from IC50 = 6.8 ± 0.7 μM to IC50 = 5960 ± 1156 μM (840-fold, Table 3), which was above the resistance level of strain AE128(pMOL30) (IC50 = 3362 ± 127 μM, Table 2, Fig. 3). Plasmid pDNA130 also mediated an increase in cadmium and cobalt resistance but topped the AE128(pMOL30) resistance level only in strain AE104 and not in the quadruple mutant (Table 3). Thus, if the residual activity of the truncated CzcD′ is neglected, constitutive high-level expression of CzcCBA led to a strong increase in metal resistance despite the absence of cytoplasmic membrane efflux systems.

Table 3.  Metal resistance of functionally substituted C. metallidurans mutant strains.a
Bacterial strainPlasmidIC50, b-value (μM)
  • a. 

    Metal resistance was determined in at least three independent dose–response curves and the IC50 and b-values in italics) were calculated as in Table 2.

  • b. 

    Deretic et al. (1987).

  • c. 

    Plasmid pDNA130 is pVDZ′2::czcCBAD′(Nies et al., 1990). The growth medium contained tetracycline in the precultures to select for the presenc of the plasmids.

AE104pVDZ′2b875 ± 384, 507 ± 311114 ± 19, 27 ± 15169 ± 17, 38 ± 10
AE104pDNA130c9007 ± 340, 1172 ± 242904 ± 157, 202 ± 837045 ± 1931, 20 200 ± 25 000
ΔcadAΔzntAΔdmeFΔfieFpVDZ′26.8 ± 0.7, 3.3 ± 1.00.09 ± 0.01, 0.04 ± 0.024.10 ± 0.44, 0.97 ± 0.22
ΔcadAΔzntAΔdmeFΔfieFpDNA1305690 ± 1156, 1374 ± 4298.40 ± 0.29, 0.97 ± 0.2112.6 ± 1.2, 3.3 ± 1.2
Figure 3.

Dose–response curve of strain AE104 mutants functionally substituted in trans. C. metallidurans AE104(pVDZ′2) (open squares, □) and its ΔcadAΔzntAΔfieFΔdmeF (pVDZ′2) mutant strain (open circles, ○) were functionally substituted in trans with plasmid pDNA130 (= pVDZ′2::czcCBAD′); AE104 (closed square, inline image), mutant strain (closed circles, ●). Cells were cultivated in parallel cultures with increasing Zn2+ concentrations in Tris-buffered mineral salts medium at 30°C with shaking for 20 h. Turbidity was determined at 600 nm. The data points shown are the mean values of 4–7 experiments with the deviation bars shown. Please note the logarithmic scale of the x-axis.

Induction of czcP by transition metals

To learn more about the function of CzcP, a czcP–lacZ fusion was constructed and used to study induction of the reporter by transition metals. This construct was a transcriptional fusion that left czcP functional. Under various conditions (metals, concentrations) tested, β-galactosidase activity of the reporter gene increased as a linear function of time for more than 4 h (basic activity 10.2 ± 0.7 U mg−1 dry weight (d.w.) in the absence of metal or at time = 0, slope, e.g. 5.46 ± 0.14 U h−1 mg−1 d.w. at 100 μM Zn2+, data not shown). Thus, influence of the various metals, metal concentrations and mutations was tested after 3 h of incubation in an endpoint determination (Fig. 4).

Figure 4.

Induction of a czcP–lacZ operon fusion by metal cations.
A. The influence of of a ΔczcR mutation (open circles, ○), complementation of this deletion in trans with czcR (closed squares, inline image) and of a ΔczcS mutation (open squares, □) on induction of the czcP–lacZ operon by Zn2+. All panels show the czcP–lacZ strain without further deletions as reference (closed circles, ●).
B. The influence of different inducers. The czcP–lacZ reporter strain (closed symbols, ●, ▴, ▾) and its ΔczcR derivative (open symbols, ○, ▵, ∇) were treated with Zn2+ (circles, ●, ○), Co2+ (triangles, ▴, ▵) or Cd2+ (inverted triangles, ▾, ∇).
C. Induction of a ΔczcA czcP–lacZ strain by Zn2+ (open circles, ○), Co2+ (open triangles, ▵) or Cd2+ (open inverted triangles, ∇). Various C. metallidurans strains AE128(pMOL30) containing czcP–lacZ fusions were incubated in Tris-buffered mineral salts medium at 30°C at a cell density of 60–70 Klett units for 3 h with increasing metal cation concentrations and the specific β-galactosidase activity was determined in U mg−1 dry mass (d.w.).
Please note the difference in scale of the y-axis and the logarithmic x-axis. All data points are the mean values of at least four experiments. Deviation bars are shown in (A) but not in the (B) and (C) to avoid confusion.

Expression of czcP–lacZ was induced by Zn2+ and was under the control of the CzcSR two-component regulatory system, which is composed of the histidine kinase CzcS and the response regulator CzcR (Fig. 4). Deletion of czcR prevented zinc-dependent induction of czcP and expression of czcR in trans from a plasmid restored expression. In contrast, deletion of czcS increased the maximum expression level of czcP after induction by zinc (Fig. 4A). Cadmium was a better inducer than zinc, leading to twofold higher maximum β-galactosidase level (Fig. 4B, Table 4). Cobalt, in contrast, was less efficient as inducer than zinc (Fig. 4B, Table 4).

Table 4.  Induction parameters of C. metallidurans strain AE128(pMOL30) czcP–lacZ.a
czcA genotypeZn2+Co2+Cd2+
Amax (U mg−1 d.w.)K50 (μM)Amax (U mg−1 d.w.)K50 (μM)Amax (U mg−1 d.w.)K50 (μM)
  • a. 

    Various C. metallidurans strains AE128(pMOL30) containing czcP–lacZ fusions and the czcA wild-type gene or a ΔczcA deletion were incubated in Tris-buffered mineral salts medium at 30°C at a cell density of 60–70 Klett units for 3 h with increasing metal cation concentrations and the specific β-galactosidase activity was determined in U mg−1 dry mass (d.w.). Since the induction curves (Fig. 4) showed some resemblance to saturation curves, a Lineweaver–Burk plot was done to fit the data to the function ‘1/A = 1/Amax + (K50/Amax)·(1/c)’ with ‘A’ being the β-galactosidase activity at a given metal concentration c, ‘Amax’ the maximum β-galactosidase activity and ‘K50’ the metal cation concentration leading to half-maximum induction. The regression coefficients were between 93.6% and 99.5%.


Deletion of czcA in the czcP–lacZ strain led to strong induction of the reporter by all three metal cations, yielding maximum expression levels never reached in the presence of the czcA gene (Table 4). Moreover, 50% induction was observed at very low concentrations, in the case of Zn2+ only 66 nM, 240-fold lower than in presence of CzcA (Table 4). These data indicate a strong interference by CzcA in metal sensing by the two-component regulatory system CzcRS.

The β-galactosidase activity of the czcP–lacZ fusion in the ΔczcA mutant decreased from the high value reached after induction with Zn2+ with the function ‘A(c) =  127 − 13.2 U mg−1 ln(c)’ with ‘A(c)’ being the β-galactosidase activity at a zinc concentration c (Fig. 4C). The cadmium-specific induction curve attained similar levels to the zinc-specific curve at a higher Cd2+ concentration but decreased at higher cadmium concentrations in parallel with the zinc-specific function (Fig. 4C). This result could not be explained simply by metal cation toxicity because the curves decreased at metal concentrations, which were less than 1% of the respective IC50 values of the ΔczcA mutant strain (Table 2). Instead, the data may indicate the presence of an uncharacterized feed-back control mechanism involved in regulation of czc.

Interference between CzcA and CzcRS-dependent regulation could be explained by the assumption that CzcCBA detoxified the periplasm and the kinase CzcS-sensed periplasmic metal ion concentration. To test this assumption, interference of an efflux system that detoxifies the cytoplasm was compared with the interference by CzcA. This experiment was done with cobalt because Co2+ is mainly exported from the cytoplasm by DmeF while several efflux systems are involved in export of zinc and cadmium cations. Figure 5 compared the influence of a ΔczcA, a ΔdmeF and a ΔdmeFΔczcA mutation on cobalt-dependent induction of czcP–lacZ. Loss of DmeF increased induction of czcP–lacZ by cobalt marginally but much less than loss of CzcA. In the ΔczcA background, additional deletion of ΔdmeF increased induction of czcP–lacZ again to a small extent. To analyse whether this was a czcP-specific effect, a lacZ reporter gene fusion was also constructed with the czcE gene, which encodes a periplasmic metal-binding protein (Große et al., 2004; Zoropogui et al., 2008). The resulting czcE–lacZ fusion was also induced by cobalt, again deletion of dmeF resulted in a somewhat increased induction of czcE–lacZ by cobalt but this effect was small compared with an additional deletion of czcA. Thus, detoxification of cytoplasmic cobalt cations by DmeF interfered little with regulation of czcP and czcE, but export of cobalt by CzcCBA, probably from the periplasm to the outside, had a much stronger effect.

Figure 5.

Induction of czcP–lacZ and czcE–lacZ operon fusions by Co2+ cations.
A. The czcP–lacZ fusion without (closed circles, ●) or with a ΔdmeF deletion (open circles, ○), and similarly a czcE–lacZ fusion without (closed squares, inline image) and with a ΔdmeF deletion (open squares, □).
B. The czcP–lacZ fusion in a ΔczcA mutant (closed triangles, ▴), a ΔczcAΔdmeF double mutant (open triangles, ▵), and a ΔczcE–lacZ fusion in a ΔczcAΔdmeF double mutant (open inverted triangles, ∇). Various C. metallidurans strains AE128(pMOL30) containing czcP–lacZ fusions [circles in (A), triangles in (B), ●, ○, ▴, ▵] or czcE–lacZ fusions [squares in (A), inverted triangles in (B), inline image, □, ∇] were incubated in TMM at 30°C to a cell density of 60–70 Klett units for 3 h with increasing metal cation concentrations and the specific β-galactosidase activity was determined in U mg−1 dry mass (d.w.).
Please note the different scales of the x- and y-axis in both panels. All data points are the mean values of at least four experiments with deviation bars shown.

To demonstrate direct binding of CzcR to the czcPp promoter, CzcR was purified after expression of the gene in E. coli, and used in an electrophoretic mobility shift experiment with a DNA fragment derived from the upstream region of czcP. Two retarded complexes were visible (Fig. 6), which decreased in intensity with increasing concentrations of unlabelled DNA. In contrast, no retardation bands appeared with control DNA, a DNA fragment upstream of the atmA gene that is involved in nickel resistance in C. metallidurans (Mikolay and Nies, 2009). Thus, these data suggest that CzcR bound to the promoter fragment, maybe in two different forms.

Figure 6.

CzcR binds to the czcP promoter. A DNA fragment spanning from −390 bp upstream to +45 bp downstream of the ATG start codon of czcP1 was amplified by polymerase chain reaction and labelled with 32P by T4 polynucleotide kinase (A). As a control, a 342 bp fragment directly upstream of the start codon of the amtA gene of C. metallidurans was also amplified (Mikolay and Nies, 2009). To allow formation of protein–DNA complexes, 40 ng of labelled DNA and 400 ng of purified CzcR (A) or 40 ng of labelled DNA and increasing amounts of CzcR (B) were incubated in binding buffer for 40 min at 28°C in the absence or presence of 400 ng (10-fold), 2 μg (50-fold) or 4 μg (100-fold) of unlabelled DNA (A) respectively. Another control was performed without CzcR (A). The samples were analysed by polyacrylamide gel electrophoresis and visualized by autoradiography. The open arrow indicates the non-retarded DNA fragment, and the closed arrows two different retarded complexes.

Contribution of four PIB-type ATPases to metal resistance in E. coli strains

To characterize the transport activity of the four PIB-type ATPases from C. metallidurans in inside-out vesicles, E. coli had to be used as expression host because the yield of intact vesicles is very low when prepared from C. metallidurans cells (Nies, 1995). While the translation start point of pbrA was unambiguous due to the position of the gene in the pbr operon, two possible translation start sites existed for czcP, four for cadA and again two for zntA. To gain experimental evidence for the respective translational start sites, all possible czcP, cadA and zntA genes were cloned in pASK3 in E. coli GG48. This mutant of E. coli strain W3110 has a deletion in the genes zntA for its native zinc-exporting PIB2-type ATPase and zitB for the native zinc-exporting CDF protein.

Zinc and cadmium resistance of the resulting strains was compared with dose–response curves in Lennox medium (Table 5). All four PIB-type ATPAses increased zinc resistance of strain GG48 with the ranking 13-fold (CzcP) > 12-fold (ZntA from C. metallidurans) > 5-fold (CadA) > 4-fold (PbrA). The three PIB2-type ATPases but not the PIB4-type ATPase CzcP also enhanced cadmium resistance 21-fold (CadA) > 13-fold (ZntA) > 4-fold (PrbA). However, neither of these genes had any influence on cobalt resistance in the various strains (ΔrcnA, ΔzntA) tested using different plasmids and growth media (data not shown).

Table 5.  Heterologous expression of PIB-type ATPase genes from C. metallidurans in E. coli.a
Bacterial strainZn2+Cd2+
IC50, b-valueQ(GG48)bIC50, b-valueQ(GG48)b
  • a. 

    Various fragments that differed at the 5 ′-ends of the respective C. metallidurans genes were cloned in plasmid pASK-IBA3 under control of the tet promoter and transferred into E. coli strain GG48, which is a ΔzntAΔzitB double mutant of E. coli W3110. Metal resistance was determined in at least three independent dose–response curves and the IC50 and b-values in italics) were calculated as in Table 2. The respective genes were: zntA1 (Rmet_4594), zntA2 (a 77-amino-acyl residues, aa, N-terminal truncated form of zntA1), czcP1 (found as open reading frame in annotations before 2005), czcP2 (Rmet_5970, renamed to pMOL30-073, a 108 aa N-terminal truncated version of czcP1, previously annotated in X71400.2 as ‘cadA’), cadA2 (Rmet_2303), cadA1 (18 aa extended version of cadA2), cadA3 (14 aa truncated) and cadA4 (51 aa truncated) and pbrA (Borremans et al., 2001).

  • b. 

    Indicates the ratio of a IC50 value divided by that of strain GG48(pASK3).

  • c. 

    Not determined because the longer gene fragment did not have any effect.

W3110(pASK3)1105 ± 37, 121 ± 614.4317 ± 40, 42 ± 14109.2
GG48(pASK3)77 ± 7, 37 ± 51.02.91 ± 0.87, 1.28 ± 0.451.0
GG48(pASK3::zntA1)918 ± 42, 310 ± 25311.939.1 ± 1.3, 2.4 ± 0.213.5
GG48(pASK3::zntA2)779 ± 38, 184 ± 5110.138.5 ± 3.6, 4.0 ± 2.913.3
GG48(pASK3::czcP1)992 ± 22, 115 ± 9912.92.10 ± 0.70, 1.50 ± 0.220.7
GG48(pASK3::czcP2)415 ± 22, 69 ±
GG48(pASK3::pbrA)294 ± 11, 66 ± 73.812.3 ± 1.7, 7.4 ± 1.14.2
GG48(pASK3::cadA1)394 ± 49, 118 ± 695.160.2 ± 29.4, 57.9 ± 34.320.7
GG48(pASK3::cadA2)332 ± 26, 130 ± 284.359.1 ± 20.4, 105.3 ± 90.020.3
GG48(pASK3::cadA3)79 ± 15, 34 ± 11.02.85 ± 0.47, 1.13 ± 0.691.0
GG48(pASK3::cadA4)90 ± 11, 38 ± 71.22.63 ± 0.21, 2.43 ± 1.020.9

The longer zntA1 increased zinc resistance slightly compared with zntA2, but there was no influence on cadmium resistance (Table 5). The difference between both forms is a 77-amino-acid residue (aa) histidine-rich region at the N-terminus of ZntA1, which was Rmet_4594 as annotated. Thus, this histidine-rich region contributed to zinc resistance mediated by ZntA in E. coli but not to cadmium resistance. The gene czcP1, which did not mediate cadmium resistance in E. coli, increased zinc resistance twice as much as the shorter czcP2. CzcP2 was Rmet_5970 as annotated as ‘cadA’ (X71400.2) and supposed to start at a ‘UUG’ codon, while CzcP1 results from an extented open reading frame adding 108 amino acyl residues to the N-terminus. Of the four versions of cadA the two longer versions cadA1 and cadA2 were fully functional whereas the two shorter versions did not mediate any zinc or cadmium resistance. CadA2 was Rmet_2303 as annotated, CadA1 an 18 aa N-terminal extended version, CadA3 14 aa and CadA4 a 51 aa truncated version of CadA2 (Table 5). Thus, the longest versions of the three proteins CzcP1, CadA1 and ZntA1 were used for further characterization.

Comparison of transition metal transport by CzcP, ZntA, CadA and PbrA

Finally, transport activity of the four PIB2/4-type ATPases from C. metallidurans was compared after expression of the respective genes in E. coli and isolation of inside-out vesicles. For each protein, transport assays were performed using radioactive 57Co2+, 65Zn2+ or 109Cd2+ at various metal ion concentrations. The resulting transport velocities were plotted to yield Km and vmax values. All functions were linear and did not show any indication of cooperative substrate binding with a Hill coefficient different from 1 (plots shown for CzcP and cadmium chloride in Fig. 7, results for all data in Table 6). To calculate apparent turn-over numbers (TONs) for better comparison, the levels of expressed P-type ATPases in the vesicle preparations were estimated by quantitative Western blot analysis using the strep-tag of purified PbrA protein as standard (Fig. S1).

Figure 7.

Cadmium transport by CzcP from C. metallidurans. Inside-out vesicles were isolated from E. coli strain GR362 (ΔzntA::kanΔzntB::catΔzitBΔzupTΔznuABC) (Grass et al., 2002) expressing CzcP from C. metallidurans. Transport assays were done at 37°C and pH = 7.0 in uptake buffer using 109Cd2+-labelled solutions of cadmium chloride at various concentrations (0.1–2.5 μM CdCl2). The uptake velocities (v) were determined in nmol min−1 and mg membrane protein, and those values were plotted to calculate Km and vmax values. Shown are the linear fittings of five transport assays (different symbols). The inset is the plot of the mean velocities (deviation bars indicated) against the substrate concentration.

Table 6.  Kinetic parameters of metal cation transport by four P-type ATPases from C. metallidurans.a
P-type ATPaseZn2+Co2+Cd2+
TON (s−1)Km (μM)TON (s−1)Km (μM)TON (s−1)Km (μM)
  • a. 

    Inside-out vesicles were isolated from E. coli strains expressing P-type ATPases from C. metallidurans. E. coli strains used were GR362 (ΔzntA::kanΔzntB::catΔzitBΔzupTΔznuABC) (Grass et al., 2002) for zinc and cadmium uptake, and a ΔrcnAΔcorAΔzupT triple mutant of wild-type strain W3110 for cobalt transport. For each metal and protein, transport assays were done at 37°C at pH = 7.0 in uptake buffer using 57Co2+, 109Cd2+ or 65Zn2+-labelled solutions of the respective metal chlorides at various concentrations (0.1–2.5 μM CdCl2, 0.25–2 μM ZnCl2, 025–10 μM Co2+). The uptake velocities were determined in nmol min−1 and mg membrane protein (shown in Fig. 7 for CzcP and Cd2+, other primary data not shown), and those values were Lineweaver–Burk plotted to calculate Km and vmax values. At least five determinations per metal and transporter were done. Since the P-type ATPases were strep-tagged, the absolute amount of each protein per mg membrane protein could be determined in a quantitative Western blot using purified proteins as reference. The number of protein molecules per mg of membrane protein could be derived from that (using the molecular mass of the respective strep-tagged protein and the Avogadro number of 6.023 × 1023 mol−1), and these were used to calculate the apparent turn-over number TON as metal cation transported per second and protein.

  • n.d., none detected.

CzcP10.876 ± 0.5501.09 ± 0.720.016 ± 0.0043.11 ± 0.480.298 ± 0.0700.36 ± 0.06
PbrA0.062 ± 0.0232.33 ± 0.890n.d.0.035 ± 0.0180.55 ± 0.27
CadA10.085 ± 0.0536.22 ± 1.780n.d.0.057 ± 0.0201.38 ± 0.09
ZntA10.057 ± 0.0254.63 ± 0.420n.d.0.028 ± 0.0120.89 ± 0.12

The difference between the three PIB2-type ATPases and the PIB4-type CzcP was striking. The three PIB2-type ATPases transported zinc with similar apparent turn-over numbers, about 0.068 s−1, and Km values between 2 and 6 μM. Similarly, they transported cadmium with apparent TONs around 0.040 s−1 and Km values between 0.55 and 1.38 μM (Table 6). In contrast, zinc transport by CzcP was 13-fold faster than zinc transport by the PIB2-type ATPases and cadmium transport was sevenfold faster. Additionally, CzcP had a lower Km value for zinc and for cadmium than CadA or ZntA. Finally, CzcP was the only protein able to transport cobalt at a detectable level.

Transport of Cd2+ by CzcP was also tested in the presence of thiol compounds using a cadmium concentration corresponding to the Km value of CzcP for Cd2+ (0.36 μM, Table 6). At Cd2+:GSH molar ratios of 1:1 or 1:10, transport was not inhibited (data not shown). Cystein and mercaptoethanol supplied in a 1:10 molar ratio also had no effect (data not shown). Higher GSH concentrations, however, completely abolished cadmium transport by CzcP but also by CadA (Fig. 8).

Figure 8.

Cadmium transport by CzcP and CadA from C. metallidurans in the presence of glutathione. Inside-out vesicles were isolated from E. coli strain GR362 (ΔzntA::kanΔzntB::catΔzitBΔzupTΔznuABC) (Grass et al., 2002) expressing CzcP or CadA from C. metallidurans. Transport assays were done at 37°C and pH = 7.0 in uptake buffer containing various concentrations of glutathion and 109Cd2+-labelled solutions of cadmium chloride at the concentration corresponding to the Km value of the respective enzyme (0.36 μM for CzcP and 1.38 μM for CadA). The uptake velocities (v) were determined in nmol min−1 and mg membrane protein. Those values were divided by the velocity obtained at the lowest GSH concentration (equimolar to cadmium, 0.36 μM for CzcP and 1.38 μM for CadA). Results of two experiments, deviation bar shown.


The data presented here strengthen the hypothesis of outer membrane efflux mediated by RND-driven efflux systems in the case of metal transporters. Moreover, they provide evidence for a novel kind of functional differentiation of cytoplasmic efflux pumps beyond the level of merely different substrate ranges.

Outer membrane efflux versus transmembrane efflux, or both?

All data obtained with RND proteins such as AcrB that are involved in detoxification of organic substances are in agreement with outer membrane efflux, which is transport from the periplasm to the outside and not from the cytoplasm to the outside (Lomovskaya and Totrov, 2005). AcrB from E. coli is part of an AcrBA-TolC transmembrane protein complex. Structural and recent mutant studies suggest that the monomers of the trimeric AcrB protein can assume three different conformations (Murakami et al., 2004; 2006; Seeger et al., 2006; Takatsuka and Nikaido, 2009). The substrate binding site of a monomer is either accessible from the periplasm, or closed, or open to the interior of the protein complex. The three monomers rotate between these three conformations in such a way that no monomer is in the same state but all three states are present in the trimer (Murakami et al., 2006; Seeger et al., 2006; Takatsuka and Nikaido, 2009). Structures of metal-transporting RND proteins are difficult to obtain (Stroebel et al., 2007). Nevertheless, contribution of the CusCBA RND-driven efflux complex to copper resistance in E. coli can be fully compensated by a periplasmic copper oxidase (Grass and Rensing, 2001), and putative copper binding sites are in the periplasmic part of the protein (Franke et al., 2003). Moreover, a periplasmic copper-binding protein is able to deliver bound copper to the membrane fusion protein of the CusCBA protein complex (Bagai et al., 2008). This argues in favour of a mainly periplasmic detoxification mediated by RND proteins also in the case of metal transporters. Additional evidence for this hypothesis provided here comes from the mutant studies and the experiments that characterize induction of czcP.

The response regulator CzcR of the two-component regulatory system CzcSR bound to the czcPp (Fig. 6) and the czcNp promoters (Große et al., 1999). CzcR was essential for expression of czcP (Fig. 4) as well as czcNICBA (Große et al., 1999) from these promoters. The histidine kinase CzcS is not essential for either expression because C. metallidurans contains two additional two-component regulatory systems related to CzcSR (von Rozycki and Nies, 2009); these other two histidine kinases are able to compensate for CzcS (J. Scherer and D. H. Nies, unpubl. results). Interestingly, at least one of the three histidine kinases has to be present in the cells to allow transcription initiation from the czcNp promoter while none is needed for the czcPp promoter (J. Scherer and D. H. Nies, unpubl. results).

Induction of czcP by metals in the absence of CzcA occurs at much lower metal concentrations and to a much higher level than in the presence of the CzcCBA efflux complex (Figs 4 and 5; Table 4). Therefore, CzcCBA lowers and CzcS senses the metal concentration in the same cellular compartment. Theoretically, this compartment could be the cytoplasm or the periplasm. However, a deletion of the DmeF protein did not yield such a strong increase in inducibility as a loss of CzcA did (Fig. 5). Like the CDF protein YiiP/FieF (Chao and Fu, 2004; Wei and Fu, 2005; Lu and Fu, 2007), DmeF is located in the cytoplasmic membrane (Munkelt et al., 2004). Since there is no evidence for any contact of a CDF protein with an outer membrane protein or membrane fusion protein, which could allow transmembrane or outer membrane efflux, DmeF probably transports transport metals from the cytoplasm to the periplasm, decreasing the cytoplasmic cobalt concentration. Because DmeF did not influence czcP induction strongly and, on the other hand, loss of CzcA had a very strong effect on czcP induction, CzcA and CzcS could act on periplasmic metal cations rather than on cytoplasmic ones. This is another argument in favour of outer membrane efflux instead of transmembrane efflux as the mode of action of the CzcCBA efflux complex. Unfortunately, a direct experiment in this direction, changing the periplasmic metal binding site of the histidine kinase CzcS, cannot be easily done, as has been outlined above. Second, a small transmembrane efflux activity of CzcCBA also cannot be ruled out completely because deletion of dmeF had a small but significant effect on induction of czcPp.

A second argument in favour of outer membrane efflux comes from the mutant studies. All four PIB2/4-type ATPases (ZntA, CadA, PbrA, CzcP) and CzcD functionally substituted each other to some degree with respect to zinc and cadmium resistance (Table 2). The four P-type ATPases transported both metal cations (Table 6), as does CzcD (Anton et al., 2004). Although CzcA also transports Zn2+ and Cd2+ (Goldberg et al., 1999), CzcA cannot compensate for one of the other transport proteins. Therefore, the CzcCBA efflux complex performed a different kind of function compared with that of the other five proteins. While the five proteins decreased the cytoplasmic metal cation concentration by efflux from the cytoplasm to the periplasm, CzcCBA decreased the periplasmic metal ion concentration by efflux from there to the medium. In case of cadmium, and especially zinc resistance, high-level expression of CzcCBA resulted in a considerable increase in resistance (Table 3), because periplasmic cations may be exported again to the outside before they are able to reach uptake systems that are located in the cytoplasmic membrane. Moreover, an additional role for periplasmic efflux is to protect periplasmic enzymes and proteins at the periplasmic face of the cytoplasmic membrane.

The situation is interesting concerning cobalt. In agreement with a previous publication (Munkelt et al., 2004), DmeF and CzcA are both essential for full cobalt resistance, and neither can compensate for the other, arguing again for a basic difference in the function performed by both proteins. Nevertheless, and as predicted by Argüello (Argüello et al., 2007), CzcP is the only one of the four characterized P-type ATPases that is able to transport Co2+ (Table 6) like CoaT from the cyanobacterium Synechocystis strain PCC6803 (Rutherford et al., 1999). Contribution of CzcP and of CzcD, which is also able to transport Co2+ (Anton et al., 1999; Munkelt et al., 2004), to cobalt resistance, however, was only significant in a ΔdmeFΔczcA double mutant, which was more resistant to the metal than the ΔdmeF single mutant (Table 2). Since deletion of czcA led to an increased inducibility of genes of the czc determinant by Co2+ (Fig. 5), higher levels of CzcP and CzcD should be present in ΔczcA mutant strains. This would explain why these two proteins were responsible for half of the increase in cobalt resistance resulting from deletion of czcA in a ΔdmeF mutant background. So, although CzcD and CzcP transport cobalt cations, their contribution to cobalt resistance in neglectable in strains containing CzcA, because their synthesis level is low in these strains.

Role of CzcP

Although CadA, ZntA, PbrA and CzcP are able to functionally compensate each other with respect to cadmium and zinc resistance (Table 2) and although these four P-type ATPases are most probably decreasing the cytoplasmic metal concentration by metal cation export from this compartment to the periplasm, there nevertheless is a fine difference in function between the PIB2-type ATPases CadA, ZntA, PbrA on one hand and the PIB4-type ATPase CzcP on the other. This difference stems from three results that appear at first glance to be contradictory: (i) CzcP displayed the highest apparent turn-over number and the lowest Km value of the four proteins (Table 6) plus the strongest increase in zinc resistance in an E. coliΔzntAΔzitB mutant (Table 5); (ii) CzcP was the least important of these four when the overall contribution to zinc and cadmium resistance in C. metallidurans was concerned; (iii) however, the ΔczcP mutant was the only single mutant strain in C. metallidurans showing a clear decrease in resistance to these two cations (Table 2). Thus, CzcP seemed to enhance metal resistance but relies on the action of the PIB2-type ATPases, which have to provide a basic resistance level. This function of CzcP as ‘resistance enhancer’ is discussed below on the level of enzyme kinetics and function, different speciations of the metal cations in the cytoplasm and, finally, on the physiological level.

Enzyme kinetics and mode of action

The Km-values of all four proteins were in the low micromolar range for zinc and below this level for cadmium, which was similar to the Km = 5.1 μM Zn2+ for ZntA from E. coli (Sharma et al., 2000). The apparent turn-over numbers of all four proteins characterized here (maximum 0.876 s−1 for Zn2+ and CzcP) were much lower than that of a Ca2+-transporting protein, 120 s−1 (Ueoka-Nakanishi and Maeshima, 2000), and those of two PIB-type proteins, the zinc transporter HMA2 from Arabidopsis, 2.38 s−1, and the copper transporter CopA from the archaeon Archaeoglobus fulgidus, 3.33 s−1 (Mandal et al., 2002; Eren and Argüello, 2004). These low apparent turn-over numbers could indicate that the C. metallidurans proteins did not function properly in E. coli, because E. coli did not regain full metal resistance when one of the four genes was expressed under a strong promoter in a ΔzntAΔzitB mutant strain (Table 5). Nevertheless, the methodological error should be similar for all four proteins so that the apparent turn-over numbers are comparable among each other. Therefore, the higher apparent turn-over number of CzcP for zinc and cadmium agrees with the fact that deletion of czcP showed a significant effect on zinc and cadmium resistance while deletion of the gene for a PIB2-type ATPase did not. It does not explain, however, why this fast efflux system did not mediate a considerable level of zinc and cadmium resistance in the absence of one of the other three PIB2-type ATPases in C. metallidurans.

The three PIB2-type proteins CadA, ZntA, PbrA show the typical ‘CPC’ motif conserved in PIB2-type ATPases while CzcP and other PIB4-type ATPases contain a ‘SPC’ at the same position (Argüello et al., 2007). Besides metal binding sites in the N-terminal part of PIB-type ATPases, their CPC motif is involved in substrate-binding and transport (Liu et al., 2006). The necessity for a complete N-terminal domain has been shown here (Table 5), but binding to the N-terminal part is not a prerequisite for binding of metal cations to CPC (Liu et al., 2006). In the model system for P-type ATPases, the sarcoplasmic calcium-transporting protein (Toyoshima et al., 2004), detailed crystallographic analysis indicates a catalytic cycle composed of four steps: (i) binding of the substrate cation to a site embedded in the transmembrane part of the protein triggers (ii) binding of the ATP-magnesium complex. The CPC and SPC motifs should be part of this substrate binding site. (iii) Phosphorylation of the protein leads to release of ADP and of the substrate cation from this substrate binding site to the other side of the membrane (or more exactly to an exchange against protons). Finally, (iv) dephosphorylation closes the cycle again. Dephosphorylation was discussed as one of the rate-limiting steps of the catalytic cycle; however, the actual rate-limiting step may depend also on the temperature (Echarte et al., 2007). Since binding of the cation to the membrane-embedded site containing the CPC or SPC motif starts the reaction cycle, this step could as well be rate-limiting under physiological conditions.

Since Zn2+ and Cd2+ have a higher affinity for sulfur as first-shell ligand atom than for oxygen (Nies, 2007), both sulfur atoms of the CPC cysteines might be thermodynamically required to extract all substrate cations from their binding sites in the cytoplasm, which may be thiol complexes. CzcP with its SPC motif might only be able to reach metals as substrates that are not too tightly bound, e.g. in mono-thiol or aquo complexes. As approximated from the solubility constants (Nies, 2007), distribution of a Zn2+ from a bis–thiol complex to SPC should require +15 kJ mol−1. This may explain why one of the PIB2-type ATPases had to be present for a basic level of cadmium and zinc resistance; CzcP may not be able to reach all Zn2+ and Cd2+ cations in the cytoplasm.

On the other hand, expelling cations bound to CPC should cost more energy than releasing them from SPC, which could be the reason why CzcP had higher apparent turn-over numbers than the three PIB2-type ATPases. Agreeing to this, the serin or cystein residue of the SPC or CPC motif is located closer to the periplasm than the common cystein residue (Argüello et al., 2007). In the classical Michaelis–Menten case (Segel, 1975), three parts of an enzymatically catalysed reaction are considered: formation of the enzyme-substrate complex ES (rate constant k+1), decomposition of ES back to enzyme plus free substrate (rate constant k−1), and formation of the enzyme-product complex EP from ES (turn-over number, rate constant k2). If the exchange of the substrate cation bound to CPC or SPC against periplasmic protons is identical with the formation of EP from ES, k2 depends on the energy of the transition state according to Arrhenius equation (Housecroft and Constable, 2006), and this energetic barrier should be lower when an oxygen-transition metal bond than when a sulfur-metal bond is resolved, which leads to an increase in k2, and consequently to a higher turn-over number for CzcP compared with the PIB2-type ATPases.

Since Km equals (k−1 + k2)/k+1, Km decreases with decreasing k2, if k2 >> k−1. Since CzcP had a lower Km value but also a higher TON than the other three proteins; however, this case cannot apply. Consequently, k2 should be smaller than k−1 or in a similar range. If k2 << k−1 or k2 ≈ k−1, Km can decrease although k2 increases if k+1 increases too. So, if CzcP is able to bind some cations from their binding sites in the cytoplasm from buffer during in vitro transport experiments more rapidly than the other three P-type ATPases, there is no contradiction between an increase in the turn-over number and a decrease in the Km value.

Metal pools

The PIB2-type ATPase ZntA from E. coli displays increased transport activity in the presence of thiol compounds (Sharma et al., 2000); however, the presence of glutathione in E. coli cells is not a prerequisite for efficient zinc and cadmium detoxification by ZntA (Helbig et al., 2008a). On the contrary, 5 mM of glutathione completely abolished transport by both proteins, the SPC-containing CzcP and the CPC-containing CadA (Fig. 8), although this bacterium contains a cytoplasmic glutathione concentration in a similar range and CadA exports cadmium in vivo (Legatzki et al., 2003a; Helbig et al., 2008a) while CzcP does not (Table 5). An estimation of the sulfur pool of E. coli cells results in 56 × 106 S/cell (from 1% S present in the cellular dry mass, 0.3 pg cell−1), 51 × 106 S bound in the cellular proteins (3.6 × 106 proteins per cell, 360 aa per protein, 2.77% methionine and 1.16% cysteine per standard protein; and 1.8–2.7 × 106 S in GSH (calculated from Helbig et al., 2008b). Thus, a majority of the cellular sulfur residues are located in the methionine and cysteine residues of the cellular proteins and not in glutathione. Since protein misfolding is indeed the main reason for cadmium toxicity in E. coli (Helbig et al., 2008b), PIB2-type ATPases might be specifically required to extract Zn2+ and Cd2+ from thiol complexes on the surface of cytoplasmic proteins.

Cadmium has a higher affinity to sulfur than zinc and CzcP alone provides some zinc resistance in C. metallidurans and E. coli; however, CzcP mediates no cadmium resistance in either bacterium. Thus, these thiol-bound cations may not be available as substrates for CzcP. Formation of CdS was also discussed to be another main mechanism behind cadmium toxicity (Helbig et al., 2008b). However, transport assays with CdS and CzcP exhibited an extraordinary high background level (J. Scherer and D. H. Nies, unpubl. results), which prevented any useful experimental approach for CzcP or the other three proteins.

Physiology of CzcP action

So CadA, ZntA or PbrA may be able to bind and expell all speciations of Zn2+ and Cd2+ cations from the cytoplasm, albeit slowly, because release from CPC has a high energetic barrier, which decreases k2. These three PIB2-type ATPases PbrA, CadA and ZntA transported Zn2+ and Cd2+ with similar Km values and apparent turn-over numbers (Table 6). The prime function of ZntA in zinc resistance, of CadA in cadmium resistance (Legatzki et al., 2003a) and of PbrA in lead resistance (Borremans et al., 2001), therefore seems to be more a consequence of different expression levels in the presence of their main substrate than a result of diverse substrate affinities.

On the other hand, CzcP may reach only more loosely bound cations but obtaines a high velocity because the energetic barrier for release from SPC is lower than that from CPC, which increases k2. Therefore, CzcP can act as a ‘resistance enhancer’, as observed in the mutant studies. Like CadA and PbrA, CzcP has evolved as a paralogue of a ZntA-like protein during speciation of C. metallidurans (von Rozycki and Nies, 2009). Therefore, during evolution of this bacterium, the gene for a ZntA-like protein was duplicated twice. All four gene copies came under control of different regulatory circuits. Finally, changing the CPC motif into a SPC motif might have altered the character of CzcP from a ‘normal’ PIB2-type efflux system into a PIB4-type protein, which ejects cytoplasmic zinc rapidly into the periplasm for further transfer to the outside by CzcCBA. The CDF proteins might have a function similar to that of CzcP. This change in the metal binding site of CzcP may also have broadened its substrate specificity to include cobalt since other proteins with a SPC motif transport a variety of substrates like cobalt (Rutherford et al., 1999), copper (Seigneurin-Berny et al., 2006) as well as calcium/heavy metals (Moreno et al., 2008).

Experimental procedures

Bacterial strains and growth conditions

All strains used for experiments were derivatives of C. metallidurans strain AE128(pMOL30) and the megaplasmid-free strain AE104 (Mergeay et al., 1985). These strains, their genotypes and the plasmids used are listed in the Tables 1–3. Tris-buffered mineral salts medium (Mergeay et al., 1985) containing 2 g sodium gluconate l−1 was used to cultivate C. metallidurans strains aerobically with shaking at 30°C. Analytical grade salts of heavy metal chlorides were used to prepare 1 M stock solutions, which were sterilized by filtration. Solid Tris-buffered media contained 20 g agar l−1.

Dose–response growth curves

Dose–response growth curves for C. metallidurans were done in Tris-buffered medium and for E. coli cells in Lennox medium (Difco, Becton-Dickinson, Sparks, Maryland, USA). Genes for the P-type ATPases were cloned under the control of the tet promoter on plasmid pASK-IBA3 (IBA GmbH, Göttingen, Germany). The medium contained 200 μg anhydrotetracycline (AHT) l−1 to induce expression of these genes. Overnight cultures of C. metallidurans and 2 h precultures of E. coli strains were used to inoculate parallel cultures with increasing metal concentrations. Cells were cultivated for 20 h with shaking at 30°C (C. metallidurans) or for 16 h at 37°C (E. coli) and the optical density was determined at 600 nm. To calculate the IC50 value (metal concentration that led to turbidity reduction by half) and the corresponding b-value (measure of the slope of the sigmoidal dose–response curve), the data were adapted to the formula OD(c) = OD0/{1 + exp[(c − IC50)/b]}, which is a simplified version of a Hill-type equation as introduced by Pace and Scholtz (1997) as published (Anton et al., 2004). OD(c) is the turbidity at a given metal concentration, OD0 that at no added metal and c the metal concentration.

Induction experiments

Cupriavidus metallidurans cells with a lacZ reporter gene fusion were cultivated in Tris-buffered mineral salts medium containing 2 g l−1 sodium gluconate with shaking at 30°C. At a cell density of 60–70 Klett units, heavy metal salts were added to various final concentrations and cells were incubated with shaking for a further 3 h. The specific β-galactosidase activity was determined in permeabilized cells as published previously with 1 U defined as the activity forming 1 nmol of o-nitrophenol per min at 30°C (Nies, 1992).

Transport assays

Everted membrane vesicles were prepared essentially as previously described (Ambudkar et al., 1984) using a buffer composed of 25 mM Tris-HCl, pH 7.0, 150 mM KCl and 250 mM sucrose. E. coli strain GR362 (ΔzntA::kanΔzntB::catΔzitBΔzupTΔznuABC) was used for Zn2+ and Cd2+ transport assays, strain ECA349 (ΔyohMΔcorAΔzupT::cat) for Co2+ (Koch et al., 2007). Vesicles were stored at −80°C after flash freezing in liquid nitrogen. Transport assays were performed at 37°C in a reaction buffer consisting of 50 mM Tris-HCl, pH 7.0, 50 mM KCl and 250 mM sucrose, 50–200 μg of membrane protein ml−1, 5 mM MgCl2, and different concentrations of 65ZnCl2, 109CdCl2 or 57CoCl2. Reactions were initiated with the addition of 1 mM Na2ATP. Then 0.1 ml samples were withdrawn after 8, 13 and 18 s, filtered through nitrocellulose filters (0.22-mm pore size, Whatman, Dassel, Germany), and immediately washed with 5 ml of the same buffer containing either 10 mM ZnCl2, CdCl2 or CoCl2. The filters were dried, and the radioactivity was quantified in a liquid scintillation counter (Beckman Coulter, Krefeld, Germany). For each metal and concentration, background values were substracted, which were determined by performing the assays without MgCl2 in the reaction buffer. The background values were always small compared with the data values.

Genetic techniques

Standard molecular genetic techniques were used (Nies et al., 1987; Sambrook et al., 1989). For conjugal gene transfer, overnight cultures of donor strain E. coli S17/1 (Simon et al., 1983) and of the C. metallidurans recipient strains grown at 30°C in Tris-buffered medium were mixed (1:1) and plated onto nutrient broth agar. After 2 days, the bacteria were suspended in saline (9 g NaCl l−1), diluted, and plated onto selective media as previously described (Nies et al., 1987).

All primer pairs used were listed in Table S2. The genes czcP, czcP2 and pbrA for the P-type ATPases were amplified from total DNA of strain C. metallidurans AE128(pMOL30). The genes cadA2, cadA3, cadA4 and zntA2 were amplified from plasmid DNA of pECD799 and pECD800 respectively. The gene czcR for the response regulator CzcR was amplified from total DNA of strain C. metallidurans AE128(pMOL30). All fragments were cloned into the vector plasmid pASK-IBA3 (IBA GmbH), and sequenced.

To construct plasmid pECD1017 czcR was amplified from total DNA of strain C. metallidurans AE128(pMOL30) and cloned into vector pGEM T-Easy (Promega, Madison, WI, USA), sequenced for control, and subcloned into the broad host range vector plasmid pVDZ′2 (Deretic et al., 1987) under control of the lac promoter [which is constitutively expressed in C. metallidurans (Nies and Silver, 1989)] on this plasmid.

Plasmid pECD1003 was used to construct deletion mutants. It is a derivate of plasmid pECD889 (Große et al., 2007) and therefore a derivative of plasmid pCM184 (Marx and Lidstrom, 2002). These plasmids harbour a kanamycin resistance cassette flanked by loxP recognition sites. Plasmid pECD1003 additionally carries an exchange of 5 bp at each loxP site. Using these mutant lox sequences, multiple gene deletions within the same genome are possible without interferences and secondary recombination events (Albert et al., 1995; Suzuki et al., 2005).

Deletion mutants

Fragments of 300 bp upstream and downstream of the target gene were amplified by PCR and cloned into vector pGEM T-Easy (Promega), further cloned into plasmid pECD889 or pECD1003, and sequenced. The resulting plasmids were used in a double cross-over recombination in C. metallidurans strains to replace the respective target gene by the kanamycin resistance cassette, which was subsequently also deleted by transient introduction of cre expression plasmid pCM157 (Marx and Lidstrom, 2002). Cre recombinase is a site-specific recombinase from the phage P1 that catalysed the in vivo excision of the kanamycin resistance cassette at the loxP recognition sites. The correct deletions of the respective transporter genes were verified by Southern DNA–DNA hybridization. For construction of multiple deletion strains, these steps were repeated. The resulting mutants carried a small open reading frame instead of the wild-type gene to prevent polar effects.

Gene insertions

The gene for czcR was rapidly inactivated in the ΔczcA deletion strain DN432 by insertion mutagenesis. The central part of the czcR gene was amplified by PCR from total DNA of strain AE128(pMOL30), cloned as a NotI/MluI fragment into plasmid pCM184 and used to interrupt the coding sequence of the target gene. Second, lacZ was inserted downstream of several target genes to construct reporter operon fusions. This was done without interrupting any open reading frames downstream of the target genes to prevent polar effects. The 300–400 bp 3′ ends of the genes czcP, czcE and pbrD were amplified by PCR from total DNA of strain AE128(pMOL30) and the resulting fragments cloned into plasmid pECD794 (pLO2-lacZ). The respective operon fusion cassettes were inserted into the open reading frame of the target gene by single cross-over recombination. For construction of Plasmid pECD794, the gene lacZ including its ribosome binding site was amplified from total DNA of strain E. coli W3110. The fragment was cloned into vector pGEM T-Easy (Promega), and then into plasmid pLO2 (Lenz et al., 1994). The correct orientation of lacZ was verified by restriction analysis and sequencing.

Purification of CzcR

CzcR was heterologously expressed in E. coli as a C-terminal Strep-TagII fusion protein using the vector pECD1009. The host cells were cultivated in Lennox medium (Difco, Becton-Dickinson) with shaking at 37°C until a turbidity of 0.8 at 600 nm was reached. Expression was induced by adding 200 μg AHT l−1 of culture for 3 h and incubation was continued with shaking at 30°C. Cells were harvested by centrifugation (15 min, 7650 g, 4°C) suspended in 20 ml of buffer (100 mM Tris-HCl, pH 8.0, 150 mM NaCl), passaged three times through a French press cell (SLM Aminco, Urbana, IL, USA) at 138 kPa in the presence of 1 mM protease inhibitor phenylmethylsulfonylfluorid and DNase I (10 mg l−1). Cell debris and unbroken cells were removed by centrifugation (15 min, 23 400 g, 4°C). A one-step purification procedure was used to isolate CzcR on a StrepTactin column, as suggested by the manufacturer (IBA GmbH). Fractions containing the CzcR protein, monitored by sodium dodecyl sulfate polyacrylamide gel electrophoresis, were pooled and concentrated with Vivaspin concentrator columns (Sartorius AG, Göttingen, Germany). The protein concentration was determined by Bradford (1976) assay, using bovine serum albumin as a standard.

Quantification of the number of P-type ATPases in vesicles

The number of P-type ATPases in the inside-out vesicles was determined by quantitative Western blot analysis using the strep-tag of purified PbrA as standard. For purification of this protein, expression, induction, cell harvest and French press treatment were similar to that of CzcR. After removal of the cell debris by centrifugation (15 min, 23 400 g, 4°C), membrane fraction was isolated by ultracentrifugation (1 h, 100 000 g, 4°C). The membrane pellet was suspended in buffer (100 mM Tris-HCl, pH 8.0, 150 mM NaCl) to a final protein concentration of 10 g l−1. Membrane fraction was solubilized with 1% (w/v) SDS for 45 min on ice with stirring, and residual membrane fragments were removed by ultracentrifugation (1 h, 100 000 g, 4°C). PbrA fusion protein was purified on a StrepTactin column following instructions of the manufacturer (IBA GmbH). Fractions containing the protein were monitored and concentrated as described above. The protein concentration was determined using a Bradford (Bradford, 1976) and bicinchoninic acid protein assay (Smith et al., 1985), with bovine serum albumin as a standard. Samples of the resulting protein and of the inside-out vesicles used in transport experiments were separated on sodium dodecyl sulfate-polyacrylamide gels, blotted (SemiDry-Blot; Biometra, Göttingen, Germany) onto a polyvinylidene difluoride membrane, and incubated with a Strep-Tactin horseradish peroxidase conjugate. Blots were developed with a chromogenic substrate as described previously (Lee et al., 2002) and analysed by computer software.

Gel retardation assay

Gel retardation assay was performed essentially as previously described (Große et al., 1999). PCR product containing the czcP promoter and a control were amplified (primer pair KO1 czcP NdeI and KO2 czcP NotI, Table S2) and labelled with 32P by T4 polynucleotide kinase. Approximately 40 ng of labelled DNA and 400 ng of purified CzcR protein were mixed and incubated for 40 min at 28°C in 30 μl of binding buffer (100 mM KCl, 20 mM Tris-HCl, pH 8.0, 3 mM MgCl2, 1 mM dithiothreitol, 100 mM EDTA, 100 μg of bovine serum albumin per ml) to allow formation of the DNA–protein complex. The reaction was analysed on a 6% (w/v) polyacrylamide gel at 4°C and 35 mA for 2–3 h.


This work was supported by the Deutsche Forschungsgemeinschaft grants Ni262/3, Ni262/4 and graduate school ‘stress’. We thank Gary Sawers for critically reading the manuscript and Karola Otto and Grit Schleuder for skillful technical assistance.