|Home | About | Journals | Submit | Contact Us | Français|
Many proteins, especially transporters, are thought to undergo large conformational alterations during their functional cycle. Since x-ray crystallography usually gives only the most stable conformation, other methods are needed to probe this conformational change. Site-directed disulfide cross-linking is often very useful for this purpose. We illustrate this by using the Escherichia coli AcrB, a proton-motive-force-dependent multidrug efflux transporter. Crystallographic studies of the asymmetric trimer of AcrB suggest that each protomer in the trimeric assembly goes through a cycle of conformational changes during drug export (functional rotation hypothesis). Site-directed disulfide cross-linking between those residues that come close to each other in only one stage in the cycle inactivated the transporter, showing that the conformational changes indeed occurred in vivo and that they are required for drug transport. A dsbA strain, which has a diminished activity to form disulfide bonds in the periplasm, was used to verify the conclusion by showing a restored transport activity in this strain. Furthermore, we describe “a real-time cross-linking experiment,” in which rapidly reacting, sulfhydryl-specific cross-linkers, methanethiosulfonates, inactivate the AcrB double-cysteine mutant expressed in dsbA cells instantaneously.
Disulfide cross-linking is an attractive method for probing conformational changes in proteins, as was shown more than 20 years ago (1). When this method is applied to intact cells of Escherichia coli, however, it is important to realize that the formation of disulfide bonds is facilitated by an elaborate system in the periplasm (2), whereas such a system does not exist within the membrane bilayer or in the cytosol. Thus the addition of oxidizing agents (1) or cross-linkers (3) is often necessary to cover the cysteine residues in the latter locations. It may also be possible to use genetically altered strains that are able to form disulfide bonds in the cytosol (4, 5). In the example to be described, cross-linking occurs in the periplasm, and we also describe the manipulation of the disulfide-forming machinery in this compartment.
Multidrug efflux transporters of the RND (resistance-nodulation-division) superfamily, such as AcrB of Escherichia coli, play an important role in both intrinsic and elevated multidrug resistance of Gram-negative bacteria (6). These transporters associate with two other classes of proteins, the outer membrane channel such as TolC and the periplasmic membrane fusion protein such as AcrA (6). This tripartite complex spans the entire thickness of the cell envelope, containing the inner and outer membranes as well as the periplasm between them (7) and expels, driven by a proton-motive force, the drugs directly into the external medium. This process must obviously involve large conformational alterations of AcrB.
The AcrB exists as a homotrimer in which each subunit provides 12 transmembrane helices and a large periplasmic headpiece (8). The newly elucidated asymmetric crystal structure of AcrB trimer (9–11) shows that each protomer has a conformation significantly different from that of its neighbors, presumably representing each of the steps (open access, drug binding, and drug extrusion) in the conformational change cycle for drug export and suggests a functionally rotating mechanism.
We provided biochemical evidence supporting this hypothesis by site-directed disulfide cross-linking experiments (12), taking advantage of the observation that the cyclic conformational change of AcrB involves the opening and closing of the large external cleft in the periplasmic domain. The AcrB proteins with double-Cys mutations on both sides of the cleft, where a disulfide bond would be produced only in the extrusion protomer (Fig. 1), are inactive. The fact that the inactivation is due to the formation of disulfide bonds has been supported by the restoration of transport activity in a dsbA strain, which has a defective DsbA, a main enzyme that catalyzes disulfide bond formation in the periplasm (2). However, dsbA did not restore the activity of some mutant pairs, and formation of large aggregates of AcrB suggested that cross-linking may have occurred during the assembly of the trimeric complex. In view of these potential problems relying on the activity of naturally cross-linked proteins, it is important that the cross-linking and inactivation of AcrB could be observed in “real rime” by using methanethiosulfonate (MTS) cross-linkers, which are known to act much more rapidly than the usual cross-linkers (13, 14). The addition of such cross-linkers to dsbA cells expressing double-Cys mutant AcrB proteins that are pumping out a fluorescent dye, ethidium, stops the function of the pump instantaneously. (An alternative approach is the reduction of disulfide bonds with externally added dithiothreitol, which resulted in the recovery of the transport function ). All these results show that when the external cleft in the periplasmic domain of AcrB becomes closed to allow the formation of a disulfide bridge or a cross-linked structure, then the resultant loss of flexibility in the conformation inactivates the efflux pump.
The method for introduction of Cys mutations follows the QuickChange site-directed mutagenesis protocol from Stratagene. A detailed instruction manual is available at their website.
Activity of each mutated AcrB protein can be evaluated by the drug susceptibility of the Δ acrB strain expressing each mutant protein. We use the gradient plate technique (17) (Fig. 2A and B), with a plate containing two agar layers that allow a gradual increase of the drug concentration along one horizontal axis.
This method, which relies on the formation of disulfide bonds between nearby Cys residues in the periplasm, works for many pairs of Cys residues. However, there is a possibility that disulfide bonds are formed during the formation of the multi-protein assembly, that is, in a way that has nothing to do with the conformational alterations during the functional cycle of the transporter (12). In order to exclude such a possibility, the real-time cross-linking of already assembled AcrB, described below, is very useful.
This work was supported by research grant AI-09644 from U.S. Public Health Service.
1The template plasmid should be isolated from a dam+ E. coli strain for the following removal step by DpnI digestion. The majority of the commonly used E. coli strains are dam+. We use E. coli DH5α or DH10B (Invitrogen) for plasmid preparation.
2We designed most of the primers in the length of 20–30 bases, with the calculated melting temperature of about 60°C or higher, and checked that there is no significant matching sequences in other parts of acrB.
3We chose the MTS reagents that are likely to cross the outer membrane. MTS cross-linkers with different size spacers are available from approximately 3.9 Å (MTS-1-MTS) to approximately 10.4 Å (MTS-6-MTS). We also examined MTS-5-MTS (approximately 9.1 Å spacer) and obtained similar results as MTS-2-MTS. As a non-cross-linker control reagent, we used 5-MTS whose length is similar to that of MTS-2-MTS.
4MTS-2-MTS is soluble in DMSO, but 5-MTS is not. We found that ethyl acetate (good solvent for 5-MTS) alone affected significantly the ethidium influx rate into the cells. To minimize this effect and to obtain a reasonable solubility for both MTS-2-MTS and 5-MTS, DMSO-ethyl acetate (3:1, vol/vol) was the best solvent we found so far.
5MTS reagents are sensitive to moisture and are hydrolyzed in water over a period of time. The reagents should be stored in a desiccator at −20°C and warmed up to room temperature before opening of the vial. We made up the solutions in organic solvent immediately prior to use and the solutions, kept on ice, were used within ~1.5 h.
6The protocol from Stratagene recommends the use of 1 μL (2.5 U) of PfuUltra HF DNA polymerase. We found that 0.5–0.8 μL enzyme also provided enough products for obtaining plasmids with point mutation.
7We observed that the levels of resistance of the Δ acrB cells expressing AcrB mutants from pSPORT1 derivatives often varied after storage of the transformed strains for several days at 4°C. Therefore, for these experiments we only used freshly transformed cells.
8To avoid the nonreproducible drug susceptibility patterns caused by strong overexpression of AcrB alone, acrB is expressed without isopropyl-β-D-thiogalactopyranoside (IPTG) induction.
9To calculate the relative activity of each mutated AcrB, both controls (fully functional pSCLBH and nonfunctional vector only) should grow and show large differences on the lengths of growth zone on the same plate (see Fig. 2B). We tested several AcrB substrates with gradient plates and found that cholic acid satisfied this requirement. With some compounds such as novobiocin and rhodamine 6G, it was difficult to find the appropriate concentration in the bottom layer, because the difference in susceptibility to these compounds was too large between the two controls. Thus at concentrations at which the fully active control grew well, no growth was seen for the nonfunctional control, whereas at other concentrations that gave a short growth zone for the vector only control, cells expressing the fully active transporter gave growth covering the entire length of the plate. We also adjusted the concentration of cholic acid in the lower layer to 10,000 μg/ml for AG100YB and to 16,000 or 18,000 μg/ml for AG100YBD, since the resistance levels of these strains to cholate were different.
10We streak the cells twice on the same line without refilling the loop, from high to low drug concentrations using one side of the loop and then low to high concentrations using the other side of the loop. Dry the surface of the plate well to avoid flooding of the plate.
11We assay all mutants at least four times with controls, and change the position of strains on each plate.
12Overgrowth of the preculture sometimes caused non-reproducible results in ethidium influx assay. To avoid this, the preculture was prepared from a fresh transformant colony without shaking, either at room temperature or at 30°C. We also tested the frozen glycerol stock, which was made with a culture in mid-exponential phase (OD660~0.5) from a fresh colony and stored in aliquots at −80°C, as the inoculum for the preculture. This method produced similar results until up to two weeks of storage of the stock.
13The concentrations of cell suspension and ethidium bromide can be modified depending on the conditions of experiments. Generally, higher concentrations of the cells and ethidium bromide show faster increase of the fluorescence. For our conditions (host strain and the expression levels of AcrB), the cell suspension of OD660 of 0.2 and 5 μM ethidium bromide produced good results.
14MTS reagents at these final concentrations produced the best results for these strains. MTS reagents at higher concentrations caused increased influx of ethidium even with the cells expressing the wild-type cysteineless AcrB. On the other hand, MTS-2-MTS at lower concentrations sometimes failed to cause a detectable inactivation of AcrB double-Cys mutants. We also note that a very low concentration of MTS reagents (4 μM) gave the best results for another host strain derived from BL21.