CHAPTER 12 – BACTERIAL MULTIDRUG RESISTANCE MEDIATED BY ABC TRANSPORTERS

CHAPTER 12 – BACTERIAL MULTIDRUG RESISTANCE MEDIATED BY ABC TRANSPORTERS CHAPTER 12 – BACTERIAL MULTIDRUG RESISTANCE MEDIATED BY ABC TRANSPORTERS CHAPTER 12 – BACTERIAL MULTIDRUG RESISTANCE MEDIATED BY ABC TRANSPORTERS CHAPTER 12 – BACTERIAL MULTIDRUG RESISTANCE MEDIATED BY ABC TRANSPORTERS CHAPTER 12 – BACTERIAL MULTIDRUG RESISTANCE MEDIATED BY ABC TRANSPORTERS CHAPTER 12 – BACTERIAL MULTIDRUG RESISTANCE MEDIATED BY ABC TRANSPORTERS CHAPTER 12 – BACTERIAL MULTIDRUG RESISTANCE MEDIATED BY ABC TRANSPORTERS

I NTRODUCTION

Microorganisms are confronted daily with

numerous environmental toxins The spectrum

of these toxins ranges from naturally produced

compounds (e.g plant alkaloids), peptides (e.g

bacteriocins) and noxious metabolic products

(e.g bile salts and fatty acids in the case of

enteric bacteria) to industrially produced

chem-icals such as organic solvents and antibiotics

Microorganisms have developed various

mech-anisms to resist the toxic effects of antimicrobial

agents, and drug-resistant pathogens are on the

rise (Cohen, 1992; Culliton, 1992; Hayes and

Wolf, 1990; Nikaido, 1994) One of the resistance

mechanisms involves the active extrusion of

antimicrobials from the cell by drug transport

systems Some transporters, such as the

tetracy-cline efflux proteins (Roberts, 1996), are

dedi-cated systems which mediate the extrusion of

a given drug or class of drugs In contrast to

these specific drug resistance (SDR) transporters,

the so-called multidrug resistance (MDR)

trans-porters can handle a wide variety of structurally

unrelated compounds On the basis of

bioener-getic and structural criteria, multidrug

trans-porters can be divided into two major classes:

(i) secondary transporters, which are driven by a

proton or sodium motive force, and (ii)

ATP-binding cassette (ABC) primary transporters,

which use the hydrolysis of ATP to fuel transport

(for a recent review, see Putman et al., 2000b).

Most bacterial multidrug transporters known

to date are secondary antiport systems that remove drugs from the cell in a coupled exchange with protons or sodium ions On the basis of size and similarities in secondary struc-ture, these transporters are classified into four major groups: the major facilitator superfamily (MFS), the small multidrug resistance (SMR) family, the resistance nodulation cell division (RND) family, and the multidrug and toxic

com-pound extrusion (MATE) family (Putman et al.,

2000b) Besides these secondary multidrug transporters, a number of ATP-dependent pri-mary drug transporters have also been

identi-fied (e.g Barrasa et al., 1995; Linton et al., 1994; Olano et al., 1995; Podlesek et al., 1995;

Rodríguez et al., 1993; Ross et al., 1990) These

primary drug transporters all belong to the ABC transporter superfamily, and most of them are SDR transporters A well-known example is

DrrAB, an SDR transporter of Streptomyces

peucetius, which confers self-resistance to its

sec-ondary metabolites daunorubicin and doxoru-bicin (Guilfoile and Hutchinson, 1991)

In the Gram-positive bacterium Lactococcus

lactis, an organism used in food manufacturing

(Figure 12.1), two distinct MDR transporters mediate resistance to toxic hydrophobic cations and antibiotics One system, designated LmrP,

is a proton/drug antiport system (Figure 12.2).

It belongs to the major facilitator superfamily, and is inhibited by ionophores that dissipate

ABC Proteins: From Bacteria to Man ISBN 0-12-352551-9

Copyright 2003 Elsevier Science Ltd All rights of reproduction in any form reserved

12

CHAPTER

the proton motive force (Bolhuis et al., 1995).

The other MDR system is an ATP-dependent

primary transporter, designated LmrA

(Figure 12.2)(van Veen et al., 1996) The role of

this chromosomally encoded primary efflux

pump in multidrug resistance was first

observed in an ethidium-resistant mutant of

L lactis subsp lactis MG1363 Ethidium efflux

in this mutant was inhibited by ortho-vanadate,

an inhibitor of ABC transporters and P-type

ATPases, but not upon dissipation of the

proton motive force (Bolhuis et al., 1994)

Iso-lated membrane vesicles and proteoliposomes,

in which purified LmrA was reconstituted,

were employed to prove that transport of

mul-tiple drugs was LmrA- and ATP-dependent

(Margolles et al., 1999; van Veen et al., 1996).

Interestingly, this lactococcal LmrA protein

was the first ABC-type multidrug transporter

identified in bacteria

Another ABC-type multidrug resistance

pump (HorA) was discovered in Lactobacillus

brevis, a major contaminant of spoiled beer

(Sami et al., 1997, 1998) This Gram-positive

lactic acid bacterium can grow in beer in spite

of the presence of antibacterial compounds (iso-␣-acids) derived from the flowers of the

hop plant Humulus lupulus L The hop resist-ance of Lb brevis is, at least in part, dependent

on the expression of the horA gene, which is

located on a 15 kb plasmid termed pRH45

(Sami et al., 1997) The role of HorA in hop

resistance was first suggested by a sponta-neous mutant lacking the pRH45 plasmid, which displayed sensitivity to the presence of hop compounds Reintroduction of pRH45 into this segregation mutant restored hop resistance

(Sami et al., 1998) These complementation

studies, as well as the heterologous expression

of the horA gene in L lactis, demonstrated that

HorA is involved in resistance to hop com-pounds Moreover, almost all lactobacilli

iso-lated as beer-spoilage strains possess horA homologues (Sami et al., 1997) In addition to

conferring hop resistance, HorA confers resist-ance to the structurally unrelated drugs

novo-biocin and ethidium bromide (Sami et al., 1997) Drug transport studies in L lactis cells and

membrane vesicles and in proteoliposomes in which purified HorA was reconstituted identi-fied this protein as a new member of the ABC family of multidrug transporters (Sakamoto

et al., 2001).

Here we summarize the existing data on the two bacterial ABC-type multidrug transporters LmrA and HorA, and analyze structural and mechanistic aspects of multidrug recognition and transport In addition, the chapter will describe how attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy

Figure 12.1 The Gram-positive lactic acid bacterium Lactococcus lactis (left picture) is used in starter

cultures for cheese production.

ATP

ADP
Pi

H

In Out

Figure 12.2 Schematic representation of two

multidrug transporters found in Lactococcus

lactis The ABC-type primary multidrug

transporter LmrA and the secondary multidrug

transporter LmrP exemplify the two major classes

of multidrug transporters found in bacteria.

Rectangles represent the transmembrane domains

of LmrA and LmrP Circles represent the

nucleotide-binding domains of LmrA.

has provided important information about

LmrA structure and the dynamic changes

occur-ring duoccur-ring its catalytic cycle

STRUCTURAL ASPECTS

All ABC transporters described so far show

a four-domain organization, which consists of

two transmembrane domains (TMDs), which

are thought to perform the transport function,

and two nucleotide-binding domains (NBDs),

which provide the energy for the transport

process (Higgins, 1992) The four domains may

be organized either in a multifunctional, single

polypeptide or as separate proteins For

exam-ple, in human P-glycoprotein (MDR1), like

many eukaryotic ABC transporters, the four

domains are found in one single polypeptide

chain arranged as TMD1-NBD1-TMD2-NBD2

As derived from the DNA sequences, bacterial LmrA is composed of 590 amino acids (calcu-lated molecular mass of 64.6 kDa) and HorA of

583 amino acids (calculated molecular mass of 64.2 kDa) Hydropathy analysis, as shown in

Figure 12.3, suggests a putative topology for both proteins of six membrane-spanning regions (putative ␣-helices) in the amino-terminal hydrophobic domain, followed by a large hydrophilic domain containing the ATP-binding

site (Sami et al., 1997; van Veen et al., 1996).

There is now experimental evidence that the membrane-spanning regions of LmrA are indeed ␣-helices (Grimard et al., 2001) Based

on the topology predictions, both the amino-terminal end and the large carboxy-amino-terminal half are located in the cytoplasm In addition to the NBD, there are two putative large

cytoplas-mic loops (Figure 12.3) (see also Chapter 11, HlyB) The predicted membrane topologies of LmrA and HorA still await experimental con-firmation The NBDs of both these bacterial transporters contain features diagnostic of an

Figure 12.3 Topology model for LmrA The LmrA protein is predicted to contain a transmembrane

domain (TMD) with six transmembrane ␣-helices, and a nucleotide-binding domain (NBD) with

the ABC signature and Walker A/B sequences A similar model is envisaged for HorA of Lb brevis.

ABC-type ATPase, such as the ABC signature

sequence, and the Walker A and B motifs

(Figure 12.4) The sequence homology between

full-length LmrA and HorA is around 53%

Sequence comparisons with other ABC

trans-porters revealed that these bacterial proteins

share significant overall sequence similarity

with members of the subfamily of multidrug

resistance P-glycoproteins, most notably the

human P-glycoprotein (MDR1) (Sami et al.,

1997; van Veen et al., 1996) For example, LmrA

and each half of MDR1 share 34% identical

residues and an additional 16% of conservative

substitutions (Figure 12.4) The ABC domain of

LmrA and the ABC1 and ABC2 domains of

MDR1 are 48% and 43% identical, respectively,

whereas the identity between the TMD of

LmrA and the amino- and carboxy-terminal

TMDs of MDR1 is 23% and 27%, respectively

The sequence conservation in the TMD of

LmrA includes particular regions (e.g the region

comprising transmembrane helices 5 and 6),

which have been implicated as being involved

in drug binding by MDR1 (Loo and Clarke,

2000) Functionally important residues in this

region of LmrA are now being identified

Interestingly, LmrA shares 28% overall sequence

identity with the lipid flippase MsbA from

Escherichia coli (Figure 12.4), the structure of

which was recently determined by X-ray

crystal-lography to a resolution of 4.5 Å (Chang and

Roth, 2001) The overall sequence similarity

between LmrA and bacterial members of other

subfamilies of the ABC transporter superfamily

is less than 28% and is mostly confined to the

hydrophilic ABC domains

In view of the general organization of ABC transporters, LmrA and HorA are considered to

be half transporters (with the two domains

arranged in TMD-NBD manner) that have to

form homodimers in order to function as full

four-domain transporters Recent studies on

LmrA provided evidence that this is indeed the

case First, two covalently linked wild-type

LmrA monomers expressed from an engineered

gene yields a functional transporter, whereas the

covalent linkage of a wild-type monomer and

an inactive mutant monomer (harboring the

K388M mutation in the Walker A region) yields

an inactive transporter (van Veen et al., 2000).

The latter covalently linked dimer had also lost

all ATPase activity, demonstrating that both

cat-alytic sites must be functional to allow ATP

hydrolysis and drug transport Second, LmrA

solubilized from membrane vesicles prepared

from LmrA-overproducing cells behaves like

a dimer on native gels (our unpublished data) Third, electron microscopy analysis of purified and reconstituted LmrA revealed small, uniform particles with a diameter of 8.5 by

5 nm, similar to those previously observed for monomeric P-glycoprotein (S Scheuring,

A Margolles, H.W van Veen, W.N Konings and

A Engel, unpublished data) Probably, the most convincing evidence for the dimeric nature of LmrA comes from co-reconstitution experi-ments into proteoliposomes of the cysteine-less wild-type LmrA and a mutant form of LmrA in

which the N-ethylmaleimide (NEM)-reactive

glycine to cysteine mutation (G386C) was

intro-duced (van Veen et al., 2000) The G386C mutant

displays wild-type transport activity but is completely inactivated upon incubation with NEM, whereas wild-type LmrA activity is not affected by NEM The transport inhibition pat-terns obtained with proteoliposomes, contain-ing different ratios of wild-type and mutant proteins, upon reaction with NEM suggest strongly that the functional unit of LmrA is a dimer and not a monomer, trimer or tetramer Taking all these data together, it is clear that the dimeric state of LmrA is a prerequisite for func-tion, and that functional crosstalk between two monomers is essential for transport

SUBSTRATE SPECIFICITY

The notion that inactivation of the secondary multidrug transporter LmrP increases drug extrusion mediated by the primary transporter LmrA points to the physiological importance

of these multidrug transporters in L lactis (Bolhuis et al., 1995) However, except for the

observation that LmrA might act as a lipid

translocase (Margolles et al., 1999), its cellular

function is still under debate The natural substrates of LmrA might be found amongst the hydrophobic compounds excreted by plants, the natural habitat of lactococci Indeed, LmrA can extrude a wide variety of amphiphilic toxic compounds, and its classification as a multidrug transporter is evident from its currently known spectrum of substrates LmrA substrates include anticancer drugs such as vinca alkaloids (vin-blastine, vincristine) and anthracyclines (dauno-mycin, doxorubicin), or cytotoxic agents such

as antimicrotubule drugs (colchicine) and DNA intercalators (ethidium bromide), or toxic peptides (valinomycin, nigericin), fluorescent membrane probes (Hoechst 33342, diphenyl-hexatriene), and fluorescent dyes such as

Figure 12.4 Amino acid sequence alignment The amino acid sequence of LmrA is shown with HorA from

Lb brevis, MsbA from E coli, and the amino- and carboxy-terminal halves of human MDR1 Residues conserved

throughout all sequences are indicated by an asterisk Residues conserved between LmrA and MDR1 are shaded

red Dashes represent residues absent in other sequences Putative transmembrane regions are boxed The Walker

A/B motifs and the ABC signature motif regions are labeled by Walker A, Walker B and ABC, respectively.

rhodamine 6G and rhodamine 123 (Margolles

et al., 1999; van Veen et al., 1996, 1998; see more

detailed discussion in Chapter 5) LmrA

modu-lators (i.e compounds that reverse

LmrA-medi-ated multidrug resistance) are also structurally

unrelated to each other and include the calcium

channel blockers verapamil and CP100-356

(analogue of verapamil), 1,4-dihydropyridines

such as nicardipine, indolizine sulfones such

as SR33557, antimalarials such as quinine

and quinidine, immunosuppressants such as

cyclosporin A, and the Rauwolfia alkaloid

reser-pine (van Veen et al., 1999) This broad drug

and modulator specificity is not only confined

to LmrA A similar range of compounds was

previously found to interact with other ABC

transporters, including yeast Pdr5p (Bauer

et al., 2000; Kolaczkowski et al., 1996) and

human P-glycoprotein (Ueda et al., 1997).

The overlapping substrate and modulator specificities of bacterial LmrA and human

P-glycoprotein reveal a functional similarity

between both proteins Expression studies of

LmrA in insect and human lung fibroblast cells

demonstrated that LmrA was indeed able to

functionally complement P-glycoprotein (van

Veen et al., 1998) Surprisingly, LmrA was

tar-geted to the plasma membrane and conferred

typical multidrug resistance on the human cells

The pharmacological characteristics of LmrA

and P-glycoprotein expressed in lung fibroblast

cells were very similar, and reversal agents of

P-glycoprotein-mediated multidrug resistance

also blocked multidrug resistance mediated by

LmrA Furthermore, the affinities of both

pro-teins for vinblastine and ATP were

indistin-guishable Finally, kinetic analysis of drug

dissociation from LmrA expressed in plasma

membranes of insect cells revealed the presence

of two allosterically coupled drug-binding sites,

indistinguishable from those of P-glycoprotein

(van Veen et al., 1998; Chapter 5) This

remark-able conservation of function between these two

ABC-type multidrug transporters implies a

com-mon overall structure and transport mechanism

L lactis is a GRAS (generally regarded as safe) organism, that is, an organism considered

to be non-pathogenic and safe to use in starter

cultures for cheese production (Figure 12.1)

(Gasser, 1994) In view of this, it is important to

know whether the substrate spectrum of LmrA

also includes clinically relevant antibiotics The

antibiotic specificity of LmrA was studied in

cytotoxicity assays, in which the antibiotic

sus-ceptibilities of E coli CS1562 cells

overexpress-ing the transporter are compared with those of

control CS1562 cells not expressing LmrA

Strain CS1562 (tolC6 :: Tn10) was used in these

assays because it is hypersensitive to drugs owing to a deficiency in the TolC protein, result-ing in an impaired barrier function of the outer

membrane (Austin et al., 1990) LmrA

expres-sion in CS1562 cells resulted in an increased resistance to 17 out of 21 clinically most used antibiotics, including broad-spectrum antibio-tics belonging to the classes of aminoglyco-sides, lincosamides, macrolides, quinolones,

streptogramins and tetracyclines (Table 12.1)

TABLE12.1 EFFECT OFLMRA

THE RELATIVE RESISTANCE TO

ANTIBIOTICS

(fold)

Aminoglycosides Gentamicin 2

␤-Lactams Ampicillin 2

Ceftazidime 3

Penicillin 4 Glycopeptides Vancomycin 1 Lincosamides Clindamycin 14 Macrolides Azithromycin 33

Clarithromycin 23 Dirithromycin 264 Erythromycin 53 Roxithromycin 35 Spiramycin 35 Quinolones Ciprofloxacin 2

Streptogramins Dalfopristin 163

Quinupristin 31

Tetracyclines Chlortetracycline 28

Demeclocycline 12 Minocycline 138 Oxytetracycline 8 Tetracycline 14 Others Chloramphenicol 11

Trimethoprim 1

aRelative resistances were determined by dividing the

IC50(the antibiotic concentration required to inhibit the growth rate by 50%) for cells harboring pGKLmrA

by the IC50for control cells harboring pGK13 For example, the latter IC50values varied between 0.3 and

2 ␮M for kanamycin, ampicillin, erythromycin, ofloxacin, dalfopristin, and minocycline Data obtained

from Putman et al (2000a) with permission.

(Putman et al., 2000a) The secondary multidrug

transporter LmrP also confers resistance to

antibiotics, although its substrate range is

smaller than that of LmrA (Putman et al., 2001).

The antibiotic specificity of HorA is currently

being analyzed The exceptionally broad

antibi-otic specificity of LmrA, the possible transfer of

the lmrA gene to other bacteria in food or the

digestive tract, and the presence of lmrA

homo-logues in pathogenic microorganisms (van Veen

and Konings, 1998) provide a serious threat to

the efficacy of valuable antibiotics It will be

interesting to find out whether P-glycoprotein is

involved in antibiotic export in human cells

Using a fluorescence quenching technique,

it has recently been demonstrated that

puri-fied and reconstituted LmrA can also transport

phospholipids (Margolles et al., 1999) In

this study, extrusion of fluorescent (C6

-NBD-labeled) phosphatidylethanolamine from the

outer leaflet of proteoliposomes by

inward-facing LmrA molecules (nucleotide-binding

domain exposed to the external surface) was

detected in the presence of ATP, with

non-hydrolyzable ATP analogues being ineffective

Phospholipid extrusion from the membrane

was inhibited by vinblastine, a high-affinity

substrate of LmrA The specificity of LmrA

with respect to lipid headgroup and acyl chain

is now being studied, possibly leading to the

identification of potential physiological lipid

substrates Several other ABC multidrug

trans-porters have also been found to possess lipid

translocation activity, including P-glycoprotein

(for a recent review, see Borst et al., 2000 and

Chapter 22of this volume)

SUBSTRATE RECOGNITION AND

TRANSPORT MODELS

Aqueous pore versus hydrophobic

vacuum cleaner and flippase models

Despite the remarkable conservation of

func-tional properties between ABC-type multidrug

transporters, there is still a considerable

contro-versy about the mechanisms by which these

proteins pump drugs from the interior of the cell

to the external medium Several transport

mod-els have been postulated for P-glycoprotein

pump function (Figure 12.5) These include (i)

the conventional aqueous pore model, in which

substrate is transported from the cytoplasm to

the exterior (Altenberg et al., 1994), (ii) the

hydrophobic vacuum cleaner model, in which

substrate is transported from the lipid bilayer

to the exterior (Raviv et al., 1990), and (iii) the

flippase model, a variation on the hydrophobic vacuum cleaner model, in which substrate is transported from the inner leaflet to the outer leaflet of the lipid bilayer, after which the sub-strate molecules will diffuse into the external medium (Higgins and Gottesman, 1992) The latter two models take into account that most drugs that interact with multidrug transporters such as P-glycoprotein and LmrA readily inter-calate into the lipid bilayer due to their high hydrophobicity and amphiphilic nature Drug extrusion from the membrane is supported by substantial experimental evidence, including the following important observations First, the non-fluorescent compound BCECF-AM (an acetoxymethyl ester derivative of 2⬘,7⬘-bis-(2-carboxyethyl)-5-(and-6-)-carboxyfluorescein)

is excreted from P-glycoprotein- and LmrA-producing cells prior to hydrolysis into the fluorescent cellular indicator BCECF by

intra-cellular esterases (Bolhuis et al., 1996; Homolya

et al., 1993) Thus, LmrA and P-glycoprotein

prevent the accumulation of the fluorescent indicator BCECF in the cytosol, despite the fact that BCECF-AM is rapidly cleaved by intracellular esterases and the resulting BCECF

is not a substrate for LmrA and P-glycoprotein

Out

In

M D R

M D R

M D

A C

Aqueous

Hydrophobic vacuum cleaner

Figure 12.5 Possible mechanisms of drug transport across the cytoplasmic membrane Drugs may be expelled from the cell by extrusion from the internal water phase to the external water phase (aqueous pore model) or by extrusion from the membrane to the exterior (hydrophobic vacuum cleaner and flippase models) Importantly, the hydrophobic vacuum cleaner model predicts that hydrophobic compounds are translocated by the MDR pump from the inner leaflet of the membrane to the external water phase, whereas the flippase model predicts extrusion from the inner leaflet to the outer leaflet of the membrane A, Drug molecules

reaching the cell rapidly insert into the outer leaflet

of the plasma membrane B, Flipping of the drug to the inner leaflet of the membrane is relatively slow and the rate-limiting step in entry C, Membrane release of drug molecules.

These observations strongly suggest that

BCECF-AM is extruded from the membrane

Second, photoaffinity analogues of substrates

of P-glycoprotein only label the two

transmem-brane domains of P-glycoprotein and not its

hydrophilic ABC domains (e.g Greenberger,

1993) Third, the affinity of binding of drugs to

purified and reconstituted P-glycoprotein is

modulated by their ability to intercalate into

the membrane (Romsicki and Sharom, 1999)

Fourth, cysteine-scanning mutagenesis, in

com-bination with reaction with the thiol-reactive

substrate dibromobimane, of all predicted

trans-membrane segments of P-glycoprotein indicates

that the drug-binding domain of P-glycoprotein

consists of residues in transmembrane segments

4, 5, 6, 10, 11 and 12 (Loo and Clarke, 2000)

Taking these data together, it seems likely

that these transporters recognize most, if not

all, of their substrates within the membrane

(hydrophobic vacuum cleaner and flippase

models) and not from the cytoplasm (aqueous

pore model) However, these observations do

not discriminate between the vacuum cleaner

model and the flippase model

Evidence for drug efflux from the inner

leaflet of the lipid bilayer to the exterior

The most convincing evidence for drug efflux

from the membrane to the aqueous phase is

provided by the kinetics of ATP-dependent

transport of TMA-DPH by LmrA (Bolhuis et al.,

1996) and of Hoechst 33342 by P-glycoprotein

(Shapiro and Ling, 1997a) The amphiphilic

character and the high lipid–water partition

coefficients result in partitioning of these

com-pounds into the lipid bilayer Conveniently,

these hydrophobic probes are strongly

fluores-cent when partitioned into the membrane but

essentially non-fluorescent in an aqueous

envi-ronment Since, therefore, the fluorescence

detected reflects the concentration of probe in

the membrane, these properties make it

possi-ble to follow fluorimetrically the partitioning

of these compounds into the lipid bilayer The

increase in fluorescence intensity due to the

partitioning of TMA-DPH into the

phospho-lipid bilayer was found to be a biphasic process

(Figure 12.6)(Bolhuis et al., 1996) This biphasic

behavior reflects the fast entry (1–2 seconds)

of TMA-DPH into the outer leaflet of the

phos-pholipid bilayer (phase 1 in Figure 12.6),

followed by a slower (several minutes)

transbi-layer movement from the outer to the inner

leaflet of the membrane (phase 2 in Figure 12.6).

When LmrA was energized in intact cells by the addition of glucose, it was observed that the initial rate of extrusion of TMA-DPH, mon-itored as a decrease in fluorescence over time, increased with an increasing concentration of TMA-DPH in the inner leaflet of the membrane

(Figure 12.6)(Bolhuis et al., 1996) The extent of

extrusion never exceeded the amount of

TMA-DPH present in the inner leaflet (Figure 12.6),

indicating that the probe cannot be extruded from the outer leaflet of the cytoplasmic mem-brane When similar experiments were done with inside-out membrane vesicles with the inner leaflet now immediately accessible to drug molecules, the situation was significantly different Upon addition of TMA-DPH to the membrane vesicle suspension, TMA-DPH rapidly intercalates into the exposed leaflet of the membrane, resulting in a maximum con-centration of TMA-DPH in this leaflet Upon energization of LmrA by the addition of ATP (the NBD of LmrA is exposed to the exterior of these vesicles), maximal rates of TMA-DPH extrusion were observed at any moment after addition of TMA-DPH and the extent of extru-sion, in contrast to intact cells, now exceeded the amount of TMA-DPH present in the inter-nal leaflet of inside-out vesicles These obser-vations strongly indicate that TMA-DPH is recognized as a substrate only after partition-ing into the normal inner leaflet of the cellular

Time (min)

1

2

Figure 12.6 Time course of the rate of energy-dependent TMA-DPH extrusion A washed cell suspension of L lactis strain MG1363 (Eth R ),

a mutant strain in which extrusion of TMA-DPH

is LmrA-dependent, was energized with 25 mM of glucose, at 5 (A), 15 (B), and 40 min (C) after the addition of 100 nM of TMA-DPH Data obtained from Bolhuis et al (1996).

membrane, and is directly transported to the

aqueous environment as observed by the

decrease in fluorescence A similar relationship

between the initial rate of transport and the

concentration of substrate in the inner leaflet of

the cellular membrane was observed for other

multidrug transporters, including secondary

transporters (Putman et al., 2000b) Thus,

sec-ondary and ABC multidrug transporters appear

to use the same mechanism of transport for

hydrophobic drugs

Hydrophobic vacuum cleaner model

versus flippase model

It is important to note that the results presented

strongly favor a vacuum cleaner mechanism of

transport by the MDR pumps and are

inconsis-tent with a flippase mechanism as proposed by

Higgins and Gottesman (1992) According to

the flippase mechanism the hydrophobic

com-pounds are translocated by the MDR pump

from the inner leaflet of the membrane to the

outer leaflet followed by diffusion into the

external medium (Figure 12.5) The

observa-tion that the fluorescence of TMA-DPH or

Hoechst 33342 falls rapidly upon energization

of LmrA or P-glycoprotein indicates that these

compounds do not stay in the lipid bilayer but

are moved into the water phase

Physiological implications of drug transport

from the inner leaflet of the membrane

This mechanism of transport of hydrophobic

drugs from the inner leaflet of the

phospho-lipid bilayer to the exterior, as illustrated in

Figure 12.7, may have several physiological

implications First, transport from the

cytoplas-mic leaflet of the membrane appears to be the

most efficient way in which MDR transporters

can prevent toxic compounds from entering

the cytoplasm As already pointed out, drug

molecules reaching the cell rapidly insert into

the outer leaflet of the membrane, but flipping

of the drug molecules from the outer to inner

leaflet is slow and the rate-limiting step in

entry (Figure 12.7) LmrA is able to transport

drug molecules from the inner leaflet back into

the external medium, counteracting the

rate-limiting step in drug entry If the transporter

were to transport drugs from the outer leaflet of

the membrane, it would probably not be able to

compete with the high rate at which drug

mol-ecules enter this leaflet Drug molmol-ecules would

‘escape’ into the inner leaflet and subsequently enter the cytoplasm Second, extrusion from the membrane may partially explain the lack of structural specificity and the consequent broad substrate range of multidrug transporters The transmembrane domains of multidrug trans-porters are thought to form a pathway across the membrane through which solutes move,

a prediction supported by structural data of

P-glycoprotein (Rosenberg et al., 2001) and MsbA

(Chang and Roth, 2001) Assuming that the translocation pathway is only accessible from the membrane, but not from the aqueous phase,

a drug molecule must be able to intercalate into the membrane in order to be recognized by the transporter Thus, the ability to intercalate into the membrane may pre-select compounds

to be transported from those which should not be transported (e.g hydrophilic cytoplasmic com-ponents) The subsequent interaction between the intercalated substrate and the transporter would be a second determinant of specificity

This would allow the transporter to have (a) rel-atively non-selective substrate-binding site(s)

NUMBER OF SUBSTRATE-BINDING SITES

Studies on the kinetics of drug dissociation have revealed the presence of two distinct, but

Membrane insertion

(fast)

Medium

Extrusion

Cytosol

Flip-flop

(slow)

Membrane release

(slow)

Figure 12.7 Proposed mechanism of LmrA-mediated TMA-DPH extrusion from a cell.

The initial rate of LmrA-dependent TMA-DPH transport correlates with the amount of TMA-DPH

in the inner membrane leaflet of whole cells, and with the amount of TMA-DPH in the outer leaflet

of inside-out vesicles Since the outer membrane leaflet of inside-out vesicles corresponds to the inner membrane leaflet of whole cells, both observations rule out the external leaflet of whole cells as a possible site of drug binding to LmrA.

allosterically linked, drug-binding sites in the

LmrA transporter (van Veen et al., 1998, 2000;

Chapter 5) The presence of these two

drug-binding sites in LmrA is strongly supported by

the finding that vinblastine equilibrium

bind-ing can best be fitted by a model in which two

vinblastine-binding sites in the LmrA

trans-porter interact cooperatively (Figure 12.8) (van

Veen et al., 2000) In this model, an initial

vin-blastine-binding event with low affinity

initi-ates a second vinblastine-binding event with

high affinity Based on the model, the

dissocia-tion constants for the two vinblastine-binding

sites were estimated to be approximately 150

and 30 nM vinblastine, respectively Moreover,

a direct determination of the

LmrA/vinblas-tine stoichiometry revealed that each

homod-imer of LmrA binds two vinblastine molecules

(van Veen et al., 2000), providing convincing

evidence for the presence of two sites

Importantly, these drug-binding sites seem to

be directly related to drug transport as shown

by the reciprocal stimulation of LmrA-mediated

vinblastine and Hoechst 33342 transport at

low drug concentrations, and reciprocal

inhibi-tion at high drug concentrainhibi-tions (van Veen

et al., 2000) Most probably, one of the

drug-binding sites interacts preferentially with vinblastine and the other preferentially with Hoechst 33342 At lower concentrations, vin-blastine binds primarily to one site and enhances transport of Hoechst 33342 bound at the other site At higher concentrations, vin-blastine is able to compete with Hoechst 33342 for binding to the same site, and therefore inhibits Hoechst 33342 transport, or vice versa Taken together, the results strongly suggest that LmrA contains at least two distinct drug-binding sites, presumably located in the TMD, with different but overlapping specificities which interact in drug transport in a positively cooperative manner Support for the presence

of at least two positively cooperative sites for drug transport in P-glycoprotein has also been presented (e.g Shapiro and Ling, 1997b; Sharom

et al., 1996) Thus, it appears that the transport

process of ABC-type multidrug transporters such as LmrA and P-glycoprotein involves two general transport-competent drug-binding sites, which may be composed of multiple drug inter-action sites to account for the wide range of compounds that are transported In addition to the transport-competent drug-binding sites, LmrA and P-glycoprotein contain regulatory sites, which may reside outside of the transport-competent drug-binding sites, to which allosteric modulators bind, but are not transported

(Martin et al., 1997; van Veen et al., 1998).

Although the topology of LmrA in the lipid membrane has been deduced from its

primary structure (Figure 12.3), its secondary

and tertiary structures are unknown since a high-resolution structure of LmrA has not yet been obtained Such a structure would supply extremely valuable information about the over-all domain organization and the interacting sites However, such a structure would also be inherently static and would not necessarily

LmrA

Control

Drug concentration (nM)

0.0 0.5 1.0 1.5

Figure 12.8 Vinblastine equilibrium binding to

LmrA Specific binding of [ 3 H]vinblastine to

inside-out membrane vesicles without LmrA

(control) or with LmrA, as a function of the free

vinblastine concentration Superimposed on the

data are the best-fit curves obtained for a

single-site binding model (hyperbolic, dotted curve)

and the cooperative two-site binding model

(sigmoidal, solid curve) Data obtained from

van Veen et al (2000) with permission from

Oxford University Press.