A Broad-Specificity Multidrug Efflux Pump Requiring a Pair of Homologous SMR-Type Proteins

Five currently recognized ubiquitous families of transport proteins are known to include members that are capable of functioning in multidrug resistance (MDR) (1,8; D. L. Jack and M. H. Saier, Jr., unpublished observations). Of these, the small multidrug resistance (SMR) family is unusual in that it consists of proteins with only 100 to 120 aminoacyl residues and four transmembrane α-helical spanners (see references5 and 7 for reviews). Although the subunit stoichiometry has not been defined for any member of this family, previously characterized SMR-type drug efflux pumps are thought to exist in the membrane as homo-oligomers (9, 11). Some SMR family members have not been shown to exhibit an MDR phenotype in spite of extensive effort in this direction (5). Those that have been shown to export drugs from the bacterial cell are specific for cationic drugs and are believed to translocate their substrates via a fairly hydrophobic transmembrane pathway (4).

Analysis of the Bacillus subtilis genome has revealed that this gram-positive bacterium encodes seven SMR homologues (6) (Table 1). TheEscherichia coli genome encodes four such homologues (plus a plasmid-encoded homologue), and distant SMR homologues have been detected in a variety of other bacteria as well as in archaea and eukaryotes (D. L. Jack and M. H. Saier, Jr., unpublished observations). Surprisingly, six of the B. subtilishomologues and two of the E. coli homologues are encoded from gene pairs in four distinct operons. These gene pairs areebrA and ebrB, yvdR andyvdS, and ykkC and ykkD in B. subtilis as well as b1599 and b1600 in E. coli (Table1).

One member of each B. subtilis protein pair is short (105 to 106 aminoacyl residues), while the other is longer (111 to 117 residues) (Table 1). This difference proved to be due to a partially conserved C-terminal hydrophilic extension present in the latter proteins but lacking in the former proteins. The short SMR family homologues could also be distinguished from the longer homologues of each of these protein pairs on the basis of topological features revealed by hydropathy plots (data not shown). Similar features are observed for the E. coli b1599-b1600 pair, and possibly also for the E. coli SugE-Ebr pair (Table 1). Interestingly,sugE and ebr of E. coli are chromosomally and plasmid encoded, respectively. These differences between the two members of each protein pair may provide the molecular basis for a requirement for the functional heterodimeric structure proposed here.

We initially cloned each of the seven B. subtilis genes and expressed them individually in E. coli strain DH5α. A drug resistance phenotype was not observed for any of them. We therefore initiated studies to determine if both genes in any one operon needed to be simultaneously expressed in order to observe an MDR phenotype. Results of the experiments with the ykkCD gene pair are reported below.

The B. subtilis genes ykkC and ykkDand the gene pair ykkCD were cloned into the expression vector pBAD24 (2). The procedure was as follows. (i) The targeted gene (or genes) was were amplified by PCR withPyrococcus woesei (Pwo) polymerase. ForykkC, the primers (5′ to 3′) were CATGCCATGGAATGGGGATTGGTCGTG (sense) and AAACTGCAGTTATGCCTCGCCTCCTTTTTCC (antisense); forykkD, the primers were CATGCCATGGTGCACTGGATCAGTTTATTGTG (sense) and ACGCGTCGACACCAACTGCTGAGC (antisense); forykkCD, the ykkC sense and ykkDantisense primers were used. (ii) The DNA was digested with theNcoI and SalI (ykkD; ykkCD) or NcoI and PstI (ykkC) restriction enzymes with restriction sites flanking the target gene created during the PCR. (iii) The included genes were then cloned into the pBAD24 polylinker region. (iv) The pBAD24 ligation mixture was heat shocked or electroporated into E. coli DH5α. (v) Finally, transformants were selected on the basis of ampicillin (50 μg/ml) resistance. Recombinant plasmids were checked by restriction enzyme digestion and direct sequencing. Expression of the cloned gene(s) was induced by the addition of 0.2% arabinose.

A twofold dilution series of the drugs listed in Table2 was analyzed. Drug assay plates were prepared with Luria-Bertani (LB) agar, 50 μg of ampicillin per ml, 0.2% arabinose, and a twofold series of drug concentrations (1,3) (Table 2). E. coli strain DH5α bearing the pBAD24 vector or bearing this plasmid with the gene(s) ykkC,ykkD, or ykkCD was grown overnight in LB broth with 50 μg of ampicillin per ml at 37°C with shaking (250 rpm). Subcultures were grown to an A ₆₀₀ of 0.06 optical density unit in LB broth with 50 μg of ampicillin per ml and 0.2% arabinose at 37°C with shaking. These cultures were diluted 10⁻¹, 10⁻², and 10⁻³ in LB broth, and 5-μl samples of each transformant at each dilution were plated on the above-mentioned assay plates. The plates were incubated overnight at 37°C, and drug resistance was scored after 12, 18, and 24 h of growth. The results presented in Table 2 are those performed with 0.2% arabinose present in the plates.

When both the ykkC and ykkD genes were expressed together in E. coli strain DH5α, a broad-spectrum MDR phenotype was observed (Table 2). We observed resistance to a broader range of toxic compounds than was observed for any previously studied SMR pump (5, 7). These compounds included representative cationic dyes and neutral and anionic antimicrobials (Table 2). The effects were at least 1 order of magnitude greater than the additive effect of the two individual genes, which were essentially inactive when present alone (Table 2). It seems unlikely that the effects of YkkC and YkkD are due to the activation of an endogenous E. coli MDR pump, since each protein when synthesized alone had no effect. Coexpression of the ykkC and ykkD genes led to greater than 20-times-higher MICs of many of the compounds tested and as great as 100-times-higher MICs of pyronin Y and phosphonomycin. Transport studies (not shown) revealed that expression of the ykkCD operon greatly inhibited ethidium bromide accumulation, and this effect was abolished by the addition of carbonylcyanide m-chlorophenylhydrazone (20 μM).

This report provides the first demonstration that a naturally occurring, proton motive force-dependent, secondary carrier consists of a hetero-oligomer of two or more dissimilar but homologous subunits. A heterodimer is proposed to be the actual structure. The YkkCD permease contrasts with other characterized members of the SMR family which are believed to be homo-oligomeric (7, 10). The results reported lead to a number of interesting questions. (i) Can each subunit function with just one subunit partner, or can it pair with multiple partners? (ii) Does just one of these subunits comprise the channel and determine the substrate specificity of the permease, or do both subunits participate in channel formation and substrate recognition? (iii) If multiple partners when paired are active, do the different possible combinations give rise to novel substrate specificities, or do these specificities merely reflect the specificities of the constituent subunits? (iv) Are the Sug proteins (5, 7), which have never been shown to exhibit a transport function, active only as hetero-oligomers? (v) What are the molecular determinants that allow an SMR polypeptide chain to function as a homo- or hetero-oligomer? These and other questions concerning the functionality of SMR superfamily members are currently under study in our laboratory.

Work in our laboratory was supported by NIH grants 2R01 AI14176 from the National Institute of Allergy and Infectious Diseases and 9RO1 GM55434 from the National Institute of General Medical Sciences, as well as by the M. H. Saier, Sr., Memorial Research Fund.

We thank Milda Simonaitis for her assistance in the preparation of the manuscript.

• Copyright © 2000 American Society for Microbiology