Energetics and Topology of CzcA, a Cation/Proton Antiporter of the Resistance-Nodulation-Cell Division Protein Family*

The membrane-bound CzcA protein, a member of the resistance-nodulation-cell division (RND) permease superfamily, is part of the CzcCB2A complex that mediates heavy metal resistance in Ralstonia sp. CH34 by an active cation efflux mechanism driven by cation/proton antiport. CzcA was purified to homogeneity after expression in Escherichia coli, reconstituted into proteoliposomes, and the kinetics of heavy metal transport by CzcA was determined. CzcA is composed of 12 transmembrane α-helices and two large periplasmic domains. Two conserved aspartate and a glutamate residue in one of these transmembrane spans are essential for heavy metal resistance and proton/cation antiport but not for facilitated diffusion of cations. Generalization of the resulting model for the function of CzcA as a two-channel pump might help to explain the functions of other RND proteins in bacteria and eukaryotes.

Multiple drug resistant bacteria poses a threat to man's fight against infectious diseases. Some multiple drug resistance systems may detoxify their substrates by transport across the complete cell wall of Gram-negative bacteria, across cytoplasmic membrane, periplasm, and outer membrane. These assumed transenvelope transporters are composed of a pump protein that energizes the transport, in addition to a membrane fusion and an outer membrane-associated protein (1,2). The pump protein may be an ATP-binding cassette transporter (3,4), a transporter of the major facilitator superfamily (5), or a resistance-nodulation-cell division (RND) 1 protein (4,6,7). The archetype of the RND permease superfamily family is CzcA from the Gram-negative bacterium Ralstonia sp. CH34 (formerly Alcaligenes eutrophus strain CH34) (8 -12).
This bacterium contains at least seven heavy metal resistance determinants, located either on the bacterial chromosome or on one of the two indigenous plasmids pMOL28 (163 kilobase pairs) and pMOL30 (238 kb) (8,(13)(14)(15)(16). One of them, the czc-determinant of plasmid pMOL30, mediates inducible resistance to millimolar concentrations of Co 2ϩ , Zn 2ϩ , and Cd 2ϩ in strain CH34 (8,17). The products of the genes czcA, czcB, and czcC form a membrane-bound protein complex catalyzing an energy-dependent efflux of these three metal cations (9,11), probably across the complete envelope. The mechanism of action of CzcCB 2 A is that of a proton/cation antiporter, and the K m values of the efflux system for the substrate heavy metal cations are also in the millimolar range (10).
Although indirect evidence led to the assumption that CzcA is the central cation/proton antiporter of the CzcCB 2 A complex (10), this has not been shown directly. This paper demonstrates that CzcA is a cation/proton antiporter, and develops the model of CzcA as a two-channel pump based on topology studies and the function of CzcA mutant proteins. This model sheds some light on other RND proteins involved in multiple drug resistance of bacteria or with previously unknown functions in mammals. 2

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Growth Conditions-Ralstonia sp. strain AE104 (14) is a metal-sensitive, plasmid-free derivative of strain CH34. Escherichia coli K38(pGP1-2) (19) was used for expression of czcCBAD derivatives under control of the phage T7 promoter as described (20). Tris-buffered mineral salts medium (14) containing 2 g/liter sodium gluconate was used for testing metal resistance and growth of Ralstonia. E. coli was cultivated in Luria broth (21). Analytical grade salts of CdCl 2 ⅐H 2 O, ZnCl 2 , and CoCl 2 ⅐6 H 2 O were used to prepare 1 M stock solutions, which were sterilized by filtration. Solid Tris-buffered medium contained 2 g/liter agar. Minimal inhibitory concentrations were determined as described (14) using Tris-buffered mineral salts medium. Protein concentrations were determined using the Bradford method (22), unless otherwise stated.
Genetic Techniques-Standard molecular genetic techniques were used (8,21). Transformation of E. coli strains was conducted as described previously (8). For expression of czcCBAD derivatives under control of the lac promoter in Ralstonia strain AE104, the plasmid pT7-5-derivatives containing the various czc constructs were cut with EcoRI and XbaI and cloned into the broad host range plasmid pVDZЈ2 (23). The pVDZЈ2-derivatives were transformed into E. coli S17-1 (24) and transferred into Ralstonia AE104 by conjugation as described (8). For reporter gene fusions, fusion vector pECD500 (11) and E. coli CC118 were used (25). All fusions were done immediately downstream of an arginine or lysine residue of CzcA. Activity of alkaline phosphatase (25) was determined in triplicate as published previously.
Mutations in the czcA gene were constructed using PCR by an over-preps, Promega, Madison, WI), each pair of PCR fragments was used as a template in another PCR reaction with the NheI and MunI primers, which both contained additional BamHI sites at their ends to facilitate cloning. The resulting DNA fragment was purified, digested with BamHI, and cloned into pUC19 (27). All final PCR products were sequenced to verify the exchange and to check for PCR-mediated unwanted changes. To facilitate the exchange of the wild type NheI-MunI fragment against the mutated fragments, the respective fragment from plasmid pECD110 was first deleted and a kanamycin resistance gene was inserted instead leading to plasmid pECD451. This was done as published (11). Next, the mutated NheI-MunI fragments were inserted into pECD451 instead of the kanamycin resistance gene. The resulting plasmids were verified by restriction endonuclease digestion using several enzymes, and correct expression of the mutant proteins was demonstrated by T7 expression (19).
Purification of CzcA-The czcA gene was PCR-amplified as an EcoRI-BamHI fragment from plasmid pECD110 (9) and cloned into the vector pASK3 (Institut fü r Bioanalytik, Göttingen, Germany) in E. coli BL21 (Stratagene Europe, Amsterdam, Netherlands). The resulting plasmid pECD559 was verified by DNA sequencing. For each expression, the bacterial strain was freshly transformed. E. coli BL21(pECD559) was cultivated for 16 h at 30°C and diluted into 1 liter of fresh Luria broth (21). The cultures were cultivated with shaking at 30°C until the optical density at 600 nm reached 1.0. CzcA production was induced with 100 g of anhydrotetracyclin/liter of medium. The cells were cultivated for an additional 3 h with shaking at 30°C and harvested by centrifugation. The pellet was washed and suspended in 10 ml of buffer W (100 mM Tris-HCl buffer (pH 8.0), 1 mM EDTA), treated twice with the French press (SLM Aminco, SOPRA GmbH, Germany; 1,000 psi) in the presence of protease inhibitor (1 mM phenylmethylsulfonyl fluoride) and DNase (10 g/liter), and cell debris was removed by centrifugation for 15 min at 20,000 ϫ g. The membrane fraction of the resulting supernatant was harvested by ultracentrifugation (1.5 h, 100,000 ϫ g) and suspended in buffer W at 10 g of protein/liter (28,29). CzcA was solubilized from membranes of the E. coli host using 1 g of n-dodecylmaltoside/g of membrane protein and 3.5 g of L-␣-phosphatidylcholine, ␤-linoyl-␥-palmitoyl/liter (stirred for 30 min at 23°C), and membrane debris was removed (30 min, 100,000 ϫ g). CzcA was purified on a strep-tactin-Sepharose column (bed volume 1 ml) equilibrated with buffer W containing 0.1 mM n-dodecyl-maltoside and 0.2 g of phospholipid/liter. After the column was washed with 12 bed volumes buffer W containing 0.1 mM n-dodecyl-maltoside and 0.2 g of phospholipid/liter, CzcA was eluted using buffer W containing 0.1 mM n-dodecyl-maltoside, 0.2 g of phospholipid/liter, and 2.0 mM desthiobiotin. Aliquots from the different steps of the purification procedure were loaded on a SDS gel (30).
Reconstitution of CzcA-Soybean L-␣-phosphatidylcholine (type II-S, 17% phosphatidylcholine) was suspended in Tris buffer (20 mM Tris-HCl, pH 7.0, 2 mM dithiothreitol, ␤-D-octylglucoside (15 g/liter)) to yield a lipid concentration of 10 g/liter. Subsequently, ␤-D-octylglucoside was removed by dialysis against 20 mM Tris-HCl, pH 7.0, 2 mM dithiothreitol, and the resulting liposomes were frozen in liquid nitrogen and stored at Ϫ80°C. For reconstitution, the liposomes were thawed and extruded through a 400-nm filter (31). Triton X-100 (0.45% w/v) was added, and total solubilization of lipids was determined by measuring changes in adsorbance at 540 nm (32,33). Detergent-destabilized liposomes were mixed with purified CzcA in a 100:1 ratio (w/w) and incubated at room temperature under gentle agitation for 15 min. Detergent was removed by adding Bio-Beads SM-2 at a final concentration of 80 mg/ml (34). After 30 min of incubation, Bio-Beads were removed by filtration on glass silk, and incubation was continued with fresh Bio-Beads for an additional 30 min at 23°C, and overnight at 4°C. The turbid proteoliposome suspension was dialyzed two times against 20 mM Tris-HCl, pH 7.0, 2 mM dithiothreitol at 4°C, and concentrated by centrifugation at 300,000 ϫ g for 45 min. For NH 4 Cl loading, the proteoliposomes (pellet) were suspended in 4 ml of 20 mM Tris-HCl buffer, pH 7.0 (containing 0.5 M NH 4 Cl) and incubated on ice for 30 min, followed by a second ultracentrifugation at 300,000 ϫ g for 45 min. The pellet was suspended in 20 mM Tris-HCl buffer, pH 7.0 (containing 0.5 M NH 4 Cl). Light scattering indicated a mean diameter of the resulting proteoliposomes of about 500 nm. This calculates to a liposome volume of 65.4 attoliter, an outer surface area of 785,000 nm 2 and an inner surface area of 736,000 nm 2 . With a surface area of 0.23 nm 2 /phospholipid (35), each liposome contained 6.61⅐10 6 phospholipids/liposome. With 1 g of CzcA (116,611 g/mol)/100 g of phospholipids (734 g/mol), about 417 CzcA proteins should have been present in each proteoliposome. The loaded proteoliposomes were frozen in liquid nitrogen and stored at Ϫ80°C. Acridine Orange Fluorescence Quenching-As published (10, 36), 1 l of proteoliposomes were diluted into 2 ml of buffer C (10 mM Tris-HCl (pH 8.0), 0.5 M choline chloride, 5 mM MgCl 2 ) containing 2 M acridine orange (3,6 bis-dimethylaminoacridine). Acridine orange fluorescence was measured using an excitation wavelength of 490 nm and an emission wavelength of 530 nm (SFM25 spectrofluorometer, Kontron, Zü rich, Switzerland) in stirred cuvettes at 24°C.
Uptake Experiments-Cation uptake experiments using the filtration method were performed as described (17,36) with some modifications. The NH 4 Cl-containing proteoliposomes were diluted into Trischoline buffer (0.5 M choline chloride, 20 mM Tris (pH 9.0)) to a final volume of 30 l. After 1 min (5 min in inhibitor experiments), cation uptake was started by the addition of the radioactive cations 65 Zn 2ϩ , 57 Co 2ϩ , or 109 Cd 2ϩ (Amersham Pharmacia Biotech, Braunschweig, Germany), and the reaction mixture was incubated at 30°C. Samples (5 l) were filtered through membrane filters (pore size 0.45 m, Schleicher & Schuell) and rinsed with 0.3 ml of Tris-choline buffer containing 10 mM EDTA and 10 mM Mg 2ϩ . The radioactivity that remained on the membrane filter was determined with a scintillation counter (LS6500, Beckman, Mü nchen, Germany). Control liposomes without CzcA were prepared using the same amounts of phospholipids but no CzcA. To calculate a "mol zinc/mol CzcA" value with these negative controls, the amount of zinc accumulated by control liposomes was divided by the CzcA content of the CzcA-containing proteoliposomes in the parallel experiment, which was 0.5 g/sample. With 417 CzcA molecules/proteoliposome and an internal volume of 65.4 attoliter/proteoliposome, 1 Zn 2ϩ /CzcA transported into the proteoliposome corresponds to an internal zinc concentration (c i ) of 10 M.

RESULTS
Purification of CzcA and Its Active Reconstitution into Proteoliposomes-CzcA was purified to homogeneity (Fig. 1). The first amino acids of CzcA were determined as MFE as expected from DNA sequence analysis (9). CzcA was reconstituted into proteoliposomes, which were incubated in 100 mM Tris-HCl, pH 5.0. At 1 mM Zn 2ϩ , CzcA proteoliposomes accumulated about 70 mol of Zn 2ϩ /mol of CzcA (c i ϭ 0.7 mM) more within the first 15 s than control liposomes ( Fig. 2A). In the following 2 min, another 100 mol of Zn 2ϩ /CzcA (final c i ϭ 1.7 mM) were accumulated, whereas binding of zinc by control liposomes did not increase with time. The rapid transport of zinc by CzcA proteoliposomes could be inhibited by 100 M of the protonophore FCCP ( Fig. 2A) or using Tris buffer, pH 5.0, instead of pH 7.0 for dilution (data not shown). Thus, this rapid transport was probably driven by the zinc concentration gradient across the proteoliposome membrane. Increased amounts of CzcA proteoliposomes led to increased metal transport, but increased amounts of control liposomes did not (Fig. 2B). Thus, CzcA was functional, catalyzed a proton-independent fast facilitated diffusion of zinc and a proton-dependent slower transport of zinc into the proteoliposomes.
To yield a stronger proton gradient, the proteoliposomes were loaded with 0.5 M NH 4 Cl; diffusion of NH 3 out of the liposome leaves protons inside and generates a proton gradient. A stable gradient for at least 20 min was shown with acridine orange fluorescence quenching experiments (data not shown). Ammonium chloride-charged CzcA proteoliposomes accumulated Zn 2ϩ rapidly (30-fold, Fig. 2C), whereas control liposomes did not. CzcA-dependent uptake of zinc into ammonium-charged proteoliposomes could be partially inhibited with 100 M FCCP or 500 M carbonylcyanide m-chlorophenylhydrazone, but neither with 10 M FCCP nor 50 M carbonylcyanide m-chlorophenylhydrazone (data not shown). The high concentration of uncoupler needed in these experiments might be the result of the very high concentration of membranes used in the experiments (Fig. 2).
Kinetics of Cation Transport by CzcA-Velocity of zinc uptake was two-fold higher when 3 mM zinc was used as the substrate instead of 1 mM, however, nearly no uptake was measured with 0.5 mM zinc (Fig. 2D). The velocity of zinc transport by CzcA followed a sigmoidal substrate saturation curve (Fig. 3A). A V max value of 385 s Ϫ1 (Table I) was calculated from the velocities at 10 and 5 mM zinc, and this velocity was used for a Hill plot (Fig. 3B) yielding a cooperativity constant n ϭ 2 and a K 50 of 6.6 mM (Table I). Concentration higher than 10 mM gave no significant uptake, probably due to the toxic effect of zinc on the integrity of the proteoliposomes (data not shown).
Cobalt transport by CzcA was much slower than zinc transport. To measure any significant transport, 1.5 l of proteoliposomes instead of 0.3 l had to be used, and the cobalt concentration had to be raised to 10 mM (Fig. 4A). The substrate saturation of cobalt transport by CzcA was again sigmoidal (Fig. 3A), and, due to a toxic effect, no uptake was detectable at concentrations higher than 50 mM. Using the velocity measured at 50 mM, a Hill plot was performed (Fig. 3B) yielding n ϭ 2 and a K 50 of 18.5 mM (Table I). Cadmium transport by CzcA was even slower than cobalt uptake (Fig. 4B). No uptake could be detected at concentrations lower than 1 mM or higher than 5 mM. In this narrow range of substrate concentration (Fig. 3A), a Lineweaver-Burk plot was linear and yielded a V max of 28 s Ϫ1 and a K m of 7.7 mM (Table I).
Topology of CzcA and Aminoacyl Residues Essential for Cation/Proton Antiport-Computer predictions (data not shown) indicated 12 hydrophobic peaks in the amino acid sequence of CzcA, which might resemble membrane-spanning ␣-helices (TMHs), and two large hydrophilic regions between TMH I/II and TMH VII/VIII, respectively. The specific activities of CzcA::PhoA translational fusions (Table II) gave evidence for (i) a cytoplasmic location of both termini of CzcA, (ii) a periplasmic location of both large hydrophilic domains, (iii) the presence of TMHs I (between N terminus and large hydrophilic domain 1), II, III, IV, VII, VIII, IX, XI and XII. No fusion could be isolated between TMHs IX and X, and the low specific activity of the fusion between TMHs V and VI (position 475) gave no evidence for a periplasmic location of this region (Table II).
The highly conserved (Fig. 5) aminoacyl residues Asp-402, Asp-408, and Glu-415 of CzcA were mutated to D402N, D408N, E415Q, and E415D. The predicted structure of CzcA leaves as possible metal-binding residues 3 histidine residues and 1 cysteine residue between TMH IV/V, these were mutated to C417S, H423R, H427R, and H439R. None of the mutations in the possible metal binding site yielded any decrease in metal resistance, neither on solid medium (Table III) nor in liquid culture (data not shown). Thus, the residues Cys-417, His-423, His-427, and His-439 alone are not essential for transport of any of the three CzcA substrate cations Co 2ϩ , Zn 2ϩ , or Cd 2ϩ . In contrast, mutation to Asn or Gln of the negatively charged aminoacyl residues in the middle of the transmembrane ␣-helix  A and B), or proteoliposomes charged with 0.5 M NH 4 Cl were diluted into choline buffer, pH 9.0 (panels C and D). After 1 or 5 min, 65 Zn 2ϩ was added. 5-l samples (containing 0.5 g of CzcA each) were filtrated, washed, and used to calculate the mols of zinc transported into the proteoliposomes per mol of CzcA (molecular mass of CzcA-strep-tag, 116,611 Da). Control liposomes did not contain CzcA but were prepared in a parallel fashion, contained the same amount of phospholipids, and were adjusted to a concentration giving the same acridine orange fluorescence quench level as CzcA proteoliposomes. To calculate a "mol of zinc/mol of CzcA" value with these negative controls, the amount of zinc accumulated by control liposomes was divided by the CzcA content of the CzcA-containing proteoliposomes in the parallel experiment, which was 0.5 g/sample. Panel A, the uptake of CzcA proteoliposomes (q) in the Tris system was compared with CzcA-containing proteoliposomes in the presence of 100 M FCCP (f). After 10-fold dilution, 1 mM of zinc was added after 1 min. The control (E) were liposomes without CzcA. Panel B, uptake of zinc (1 mM) by CzcA-containing proteoliposomes diluted 10-fold (q) or 6-fold (f). Control liposomes were diluted 10-(E), 6-(f), or 2.5-fold (⌬). Please note that these data are given in "nmol of zinc/ sample" to show the differences between the various amounts of proteoliposomes and liposomes. Panel C, CzcA-containing proteoliposomes (q) and control liposomes in the ammonium chloride/choline system (E). Panel D, uptake of 3 (q), 1 (f), or 0.5 mM of zinc (OE). The values obtained for the liposome control were subtracted from the CzcA proteoliposome values to yield a zero point. IV led to a complete loss of metal resistance to each of the three heavy metals (Table III), and not the slightest residual resistance could be observed in liquid culture (data not shown).
Although the E415Q mutant CzcA protein was not functional, the E415D protein gave full resistance on solid media (Table III). However, in liquid culture in the presence of 2.5 mM Co 2ϩ , 2.5 mM Zn 2ϩ , or 1 mM Cd 2ϩ , an AE104 derivative with the E415D mutation in czcA grew slower than the respective wild type strain (data not shown). Thus, the small change in the position of the carboxyl group of aminoacyl residue 415 of the CzcA protein diminished the efficiency of heavy metal transport in the respective mutant strain.
CzcA as a Two-channel Pump-The mutant proteins with defects in the aspartate residues, CzcA-D402N (pECD557) and CzcA-D408N (pECD558), were purified like the CzcA wild type protein and reconstituted into proteoliposomes. When they were compared with the wild type protein, all three proteins displayed the rapid facilitated diffusion of zinc (Fig. 6A), but the mutant proteins were no longer able to catalyze the slower proton/zinc antiport. To analyze, if the zinc bound in the rapid reaction was transported into the inside of the proteoliposomes or were bound at the outside, proteoliposomes with the wild type (data not shown) and the mutant proteins (Fig. 6B) were incubated for 2 min with 1 mM 65 Zn 2ϩ , and then an additional 10 mM of nonradioactive zinc was added. In all cases, CzcAcontaining proteoliposomes accumulated more zinc in this isotope competition experiment, although the amount of zinc bound by the control liposomes decreased. Transport of Co 2ϩ and Cd 2ϩ by wild type and both mutant proteins was not The velocity of metal cation uptake was determined for all three metals. For zinc and cobalt, the substrate saturation curve was sigmoidal. For zinc, the V max was determined using the velocities at 5 and 10 mM; for cobalt, it was the velocity at 50 mM. Higher concentrations of either metal inactivated the proteoliposomes. With this V max value, a Hill plot using the equation ln(V/V max Ϫ V) ϭ n ⅐ ln S Ϫ n ⅐ ln K 50 was performed yielding the resulting values. For cadmium, concentration higher than 5 mM inactivated the proteoliposomes, and no significant transport could be measured with concentrations lower than 1 mM. Within this range of concentrations, the Lineweaver-Burk plot was linear and yielded the listed K m and V max values.   5. Conserved amino acid residues in TMH IV of the RND proteins. Multiple alignments of all CzcA-related protein sequences in the current data bases were performed (data not shown). These alignments grouped the RND proteins into a heavy metal subfamily with a CzcA core and some proteins clustering around the CzcA core, and into a multiple drug resistance subfamily with an Acr core and some sequences clustering around this core. The figure gives the conserved TMH IV of CzcA (Swiss-Prot accession number spP13511), other divalent heavy metal cation transporters like CnrA and NccA (spP94177, spQ48815, spQ44586, spP37972), possible monovalent heavy metal cation transporters like SilA and HelA_ECOLI (gi4155474, gn-PIDd1011351, gi4206630, gi2314494, spP38054), an outgrouping sequence from a cyanobacterium, SLL0142 (gnPIDd1019295), the Acr outgroup (gnPIDd1010740, gnPIDd1018802, gi2313726), the Acr core (spP24181, spP31224, spP37637, spP52002, gi953226, gi1408202, gn-PIDe256815, gi2109271), and another outgrouping sequence from a cyanobacterium, SLR0454 (gnPIDd1019544). Conserved aspartate and glutamate residues are shown in bold letters. different (data not shown). Thus, the mutant proteins were still able to transport metal cations into proteoliposomes but were unable to catalyze a proton/cation antiport.

DISCUSSION
The roughly determined structure of CzcA as N terminus/ TMH I/periplasmic domain 1/TMHs II-VII/periplasmic domain 2/TMHs VIII-XII/cytoplasmic C terminus fits into the model of a "6ϩ6 spanner transporter" (37). If two independent channels exist in CzcA, they may have different functions. The highly conserved "DDE" motif in one of these channel was essential for CzcA function in vivo and for proton/cation antiport in vitro but not for facilitated diffusion of cations. Thus, the DDE channel could form a charge-relays system (38) with the DDE residues required for proton transport, whereas the second channel may be a cation channel. Therefore, the best model to describe the function of CzcA, so far, is that of a two-channel pump.
In a working model, binding of Zn 2ϩ to a cytoplasmic metalbinding site of CzcA may trigger proton transport across the proton channel into the cytoplasm. This could create negative charges in the large periplasmic domains of CzcA. According to calculations using Fick's first law, diffusion of Zn 2ϩ through CzcA should be far too slow to explain the observed turnover number of CzcA, thus, these negative charges plus the bound cation could generate an electrical field that drives the cation through the cation channel to the periplasm. Finally, the cation should be exchanged for protons from the periplasm.
Other RND Proteins-Asp-402 is not conserved (Fig. 5) in the SilA protein (39) involved in efflux of Ag ϩ , but Asp-408 and Glu-415 are conserved, which may be explained by a 1H ϩ /1Ag ϩ ratio of transport by this system in contrast with a ratio of 2 H ϩ / 1 Zn 2ϩ for CzcA (10). Therefore, RND proteins with a DE motif like SilA may be involved in transport of a monovalent cation, and this has been demonstrated for YbdE from E. coli, which detoxifies Ag ϩ . 3 Surprisingly, the Acr-like RND proteins (HAE1 family) (7) contain another conserved DDE motif in the respective region after TMH IV (Table II, Fig. 5); however, in contrast to CzcA, Asp-402 is not conserved, but another Asp at the (CzcA-) position Asp-409.
In Acr proteins, the substrates do not have to carry any charge, increased hydrophobicity even enhances transport (40). Thus, although the conserved proton channel suggests that Acr-like RND proteins are driven by the proton gradient, an electrical field is unlikely to drive export of substances like toluene, hexadekan, or ␤-lactams with highly hydrophobic side chains (41)(42)(43). In Acr-like proteins, protonation/deprotonation may switch a periplasmic substrate-binding site between a hydrophilic and a hydrophobic character. If a hydrophobic substrate is bound to this site, it may be driven off into the periplasm when the site becomes hydrophilic as a consequence of protonation.
Some other bacterial proteins are related to RND proteins or even members of this family. 2 SecD is involved in protein export and seems to catalyze the release of the transported preprotein portion from the Sec pore. This reaction is also proton-dependent (44,45). Because SecD is related to the proposed proton channel of RND proteins (7), the SecD result supports the reaction mechanism proposed above. Moreover, some proteins in eukaryotes may also be RND members, 2 therefore, they may catalyze comparable reactions. The Niemann-Pick disease C1-protein NPC1_MAN seems to be a lysosomal cholesterol transporter (46 -48), and should be proton-driven. Members of the patched gene regulator family are involved in a signal transduction chain of the hedgehog developmental signal (18, 49 -51); these proteins should be protondriven transporters too. The homology to NPC1_MAN suggests a function as a cholesterol transporter. Because covalently bound cholesterol is essential for the activity of the hedgehog protein, release and uptake of the hedgehog-bound cholesterol might be required to set back the hedgehog signal.
Thus, the data obtained with the purified and reconstituted CzcA protein might be of help to understand the function of other RND-related proteins involved in multiple drug resistance and protein export in bacteria, and of proteins involved in cholesterol metabolism, development and cancer generation in mammals.
FIG. 6. Zinc transport by CzcA mutant proteins. Panel A shows the uptake of 1 mM 65 Zn into proteoliposomes with CzcA wild type protein (q) and the D402N (f) or D408N (OE) mutant proteins. Panel B shows an isotope competition experiment: proteoliposomes with the mutant proteins D402N (f) and D408N (OE) or control liposomes (E) were incubated with 1 mM 65 Zn 2ϩ . After 2 min (arrows), 10 mM of nonradioactive Zn 2ϩ was added.