CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS

CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS CHAPTER 22 – LIPID TRANSPORT BY ABC TRANSPORTERS

I NTRODUCTION

ABC transporters, like fat, are embedded in a

lipid bilayer and some of them are good at

trans-porting lipids This could already be inferred

from the fact that classical multidrug resistance

(MDR) of cancer cells can be caused by the ABC

transporter, the P-glycoprotein (MDR1, ABCB1)

The drugs belonging to the MDR spectrum are

rather hydrophobic and MDR1 P-glycoprotein

must therefore have affinity for lipophilic

compounds and lipids The importance of ABC

transporters for lipid transport was firmly

established by the discovery in 1993 that the

human MDR3 P-glycoprotein (ABCB4;

some-times referred to as PGY3) is a dedicated

phos-phatidylcholine (PC) transporter, indispensable

for normal bile formation

Since 1993 many additional ABC trans-porters have been shown to be involved in

lipid transport, as illustrated by the overview

presented in Table 22.1 The list is undoubtedly

incomplete We know only for a fraction of the

48 human ABC transporters what their

phy-siological substrates are We expect that some

of the new ones that have recently turned up

in the human genome will also be involved in

lipid transport

It should be clear from Table 22.1 that we

use a rather broad definition of lipid

port Included are not only proteins that

trans-port indisputable lipids, such as the MDR3

P-glycoprotein (ABCB4) or ABC1 (ABCA1), but also proteins that transport acidic lipid conju-gates, such as MRP1 (ABCC1) and MRP2 (ABCC2), or ALDP (ABCD1) This serves to high-light the diverse roles of the ABC transporters in

lipid disposition The transporters listed in Table 22.1differ widely in their contribution to lipid transport Most of them, notably MDR3 P-glyco-protein, BSEP (ABCB11), MRP2 (ABCC2), ABC1 (ABCA1), ABCG5 and ABCG8, ALDP (ABCD1) and ABCR (ABCA4), are indispensable and their absence or disruption results in disease For other transporters, natural substrates are known, but transporter absence does not seem to lead to significant alterations in lipid disposition

Examples are MDR1 P-glycoprotein and MRP1

These proteins may mainly serve to defend the body against amphipathic xenotoxins This may also be the case for BCRP1 (ABCG2), which has a clearly defined protective function, but no phys-iological substrates are known yet for this drug transporter The substrate specificity and func-tion of the recently identified large transporter

ABCA2 (Vulevic et al., 2001) is also still unclear.

ABCA2 appears to be present in lysosomes, but it was also found in the endoplasmic

retic-ulum, Golgi and some peroxisomes The ABCA2

gene was overexpressed in a cell line selected for estramustine resistance, but the resistance against this drug of cells transfected with

ABCA2 cDNA constructs was only marginal.

Estramustine is a synthetic nitrogen mustard

461

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

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

22

CHAPTER

TABLE22.1 HUMAN AND RODENTABC TRANSPORTERS INVOLVED IN LIPID TRANSPORTa

1 ABC1 ABCA1 2261 mAbc1 Ubiquitous Plasma P-lipid, Tangier Hemorrhage,

TGD (12?) membrane cholesterol? disease defective

CERP

2 ABC2 ABCA2 2436 mAbc2 Brain, kidney, Lysosomal None known Estramustine – – Steroid

(12?) rAbc2 lung, heart membrane transport?

3 ABCR ABCA4 2273 mAbcr Retina Rim of outer N-retinylidene- – Stargardt As humans

RmP (12?) segment disks phosphatidyl- disease

STGD1

STGD

4 PGY1 ABCB1 1279 mMdr1a Many Plasma Glucosylceramide, Amphipathic – Drug Major defense

MDR1 (12) (Mdr3) epithelia, membrane platelet-activating drugs hypersensitivity function against

Pgp mMdr1b blood–brain (apical) factor xenotoxins

GP170 (Mdr1) barrier

(and rat homologue)

5 PGY3 ABCB4 1279 mMdr2 Liver Plasma Long-chain Some PFIC-3, Liver disease Also defense

MDR3(2) (12) rMdr2 hepatocytes membrane phosphatidylcholine amphipathic cholestasis function against

PFIC-3 (apical) drugs of pregnancy xenotoxins?

6 BSEP ABCB11 1321 mBsep Liver Plasma Bile salts Paclitaxel PFIC-2 Liver disease Drug resistance

sPGP (12?) rBsep hepatocytes membrane is low

PGY4

7 MRP1 ABCC1 1531 mMrp1 Ubiquitous Plasma LTC4 Anionic drug – Drug Also

MRP (17) rMrp1 membrane conjugates, hypersensitivity co-transports

(basolateral) GSSG, GSH drugs with GSH and

endosomes

8 MRP2 ABCC2 1545 rMrp2 Liver, Plasma Bilirubin- Anionic drug Dubin–Johnson Altered drug Also

cMOAT (17) intestine, membrane glucuronides, conjugates syndrome handlingd co-transports

kidney (apical) GSSG and GSH; drugs with GSH

acidic bile salts

9 MRP3 ABCC3 1527 rMrp3 Liver, Plasma Bile salts Anionic drug – ? Strongly

(17?) mMrp3 bile ducts, membrane conjugatese upregulated in

gut, adrenal (basolateral) cholestasis cortex

10 ALD ABCD1 745 mAld Many Peroxisomal Very long-chain – Adrenoleuko- As humans Probably

ALDP (6?) membrane saturated fatty dystrophy heterodimer

acyl-CoA with 11/12/13

11 ALDL1 ABCD2 740 rAbcd2 Many Peroxisomal As 10? ? – – As 10?

ALDR (6?) membrane

12 PMP70 ABCD3 659 mPmp70 Many Peroxisomal As 10? ? – – As 10?

PXMP1 (6?) rPmp70 membrane

13 PMP69 ABCD4 606 mP69r Many Peroxisomal As 10? ? – – As 10?

P70R (6?) membrane

PXMPIL

14 BCRP1 ABCG2 655 mBcrp1 Placenta, Plasma None known Amphipathic – Drug Major defense

MXR1 (6) gut, liver, membrane drugs hypersensitivity function against

15 ABCG5 ABCG5 651 – Liver, Plasma Plant sterols Cholesterol? Sitosterolemia ? Probably

(6?) intestine membrane? heterodimer

16 ABCG8 ABCG8 673 – Liver, Plasma Plant sterols Cholesterol? Sitosterolemia – Probably

(6?) intestine membrane? heterodimer

aSee also http://nutrigene.4t.com/humanabc.htm.

This is the website of Michael Müller, University of Wageningen, The Netherlands.

bSize in number of amino acids and topology as the most probable number of transmembrane segments.

cA dash means that no homozygous null alleles have been observed (humans, rats) or constructed (KO mice) A question mark means that no phenotype has (yet) been found.

dDecreased biliary drug clearance and increased oral drug availability; decreased biliary excretion of bilirubin glucuronides.

ePreference for glucuronosyl derivatives of drugs and steroids; does not transport GSH.

derivative of estradiol and it is therefore possible

that ABCA2 is a steroid transporter This remains

to be demonstrated, however

Many transporters listed in Table 22.1 are

discussed in detail in other chapters of this

volume The involvement of the MDR1

P-glycoprotein and MRP1 in drug transport is

discussed in Chapters 18 and 19, MRP2 (ABCC2)

in Chapter 20, MRP3 (ABCC3) in Chapter 21,

ABCA1 in Chapter 23,ABCA4 in Chapter 28,

and the peroxisomal ABC transporters in

Chapter 24 Here we concentrate on four topics:

(1) the MDR1 P-glycoprotein (ABCB1) and its

role in transporting physiological lipids; (2) the

transport of lipid analogues by ABC

trans-porters; (3) the MDR3 (ABCB4) P-glycoprotein

(the phosphatidylcholine transporter, murine

Mdr2); and (4) the role of ABC transporters in

sterol transport with an emphasis on

trans-porters not discussed in detail in other chapters

I TS R OLE IN

MDR1 P-glycoprotein substrates are rather

hydrophobic molecules (Seelig et al., 2000) and

current models for MDR1 activity propose

that the substrate is recognized within the

mem-brane In the vacuum cleaner model (Raviv et al.,

1990), the substrate molecule enters a

hydropho-bic cavity and is pumped into the extracellular

space (Bolhuis et al., 1996) In the flippase model

(Higgins and Gottesman, 1992), the substrate

enters the MDR1 P-glycoprotein from the

cyto-solic leaflet of the plasma membrane and is

sub-sequently moved into the exoplasmic leaflet

From there, it can freely diffuse into the

extra-cellular space Not surprisingly, the question

was raised as to whether MDR1 P-glycoprotein

would be capable of moving natural membrane

lipids from the cytosolic into the exoplasmic

leaflet of the plasma membrane, a process

termed flop as opposed to lipid flip in the

oppo-site direction (Devaux and Zachowski, 1994) In

this case, pumping of the substrate lipid into the

aqueous phase would be unlikely, as the change

in free energy between the monomer in aqueous

solution and the membrane form (70 kJ mol⫺1for PC) is more than the energy released from hydrolysis of an ATP molecule (30 kJ mol⫺1) (McLean and Phillips, 1984) This problem would be solved by the flippase model, in which the substrate is only moved from the cytosolic to the exoplasmic leaflet of the plasma membrane (Higgins and Gottesman, 1992) Less hydropho-bic compounds might be translocated into the exoplasmic leaflet by flippase action and readily equilibrate with the extracellular water phase One natural lipid found to be a substrate for MDR1 P-glycoprotein is platelet-activating factor (PAF) (Ernest and Bello-Reuss, 1999;

Raggers et al., 2001) This bioactive lipid is

syn-thesized by inflammatory cells upon cell activa-tion by a number of physiological stimuli Some activated cells release PAF upon specific induc-tion whereas other cells need no addiinduc-tional

stim-ulation (Prescott et al., 2000) PAF release from

these cells may be the consequence of a process that scrambles the asymmetric distribution of the bulk membrane lipids (Bratton, 1993)

Ernest and Bello-Reuss (1999) found, unex-pectedly, that PAF release from ionophore-stimulated cells is inhibited by MDR1 inhibitors

This was followed up by Raggers et al (2001),

who found that transfection of kidney epithelial

cell monolayers with human MDR1 selectively

increased PAF transport across the apical plasma membrane domain PAF transport was independent of vesicular traffic and was inhib-ited by the MDR1 P-glycoprotein inhibitors PSC833 and cyclosporin A (CsA) It is very likely

that this system mimics the in vivo situation

of constitutive PAF secretion since kidney cells possess a cholinephosphotransferase that is

spe-cific for PAF synthesis (Woodard et al., 1987),

and have a high endogenous level of MDR1 Like PAF, its structural analogue, the anti-neoplastic agent edelfosine (the di-ether PC

analogue

1-O-octadecyl-2-O-methyl-sn-glycero-3-phosphocholine) may be a substrate for translocation by MDR1 However, a report on such an activity for murine Mdr3 (also known

as Mdr1a) P-glycoprotein (Abcb1) was not consistently confirmed in subsequent studies

(Ruetz et al., 1997).

The fact that Mdr1 (Abcb1) does not

func-tionally replace Mdr2 (Abcb4) in the Mdr2

murine knockout (KO) model predicts that nat-ural long-chain PC is not a substrate for human

MDR1 (Smit et al., 1993), but a direct test must

still be done to prove this A long-chain mem-brane lipid that may be a substrate for MDR1 is sphingomyelin (SM), as one study claimed that

the MDR1 inhibitor PSC833 caused an increase

in the fraction of SM in the inner leaflet of the

plasma membrane (Bezombes et al., 1998).

Another sphingolipid that appears to be

trans-located by MDR1 is glucosylceramide From

its site of synthesis on the cytosolic surface of

the Golgi, glucosylceramide did not reach the

surface of fibroblasts derived from an Mdr1

null mutant mouse In control fibroblasts, which

express Mdr1, transport was inhibited by Mdr1/

MDR1 inhibitors Translocation of

glucosyl-ceramide across the Golgi membrane, as

measured from the synthesis of higher

glyco-lipids in the Golgi lumen, continued in the

null fibroblasts, indicating the presence of an

Mdr1-independent translocator (Raggers et al.,

unpublished results) Still, the results do not

exclude the possibility that Mdr1 can also be

active in membranes of the Golgi This has been

suggested from the observations that

transfec-tion of Madine Darby canine kidney (MDCK)

cells with human MDR1 cDNA dramatically

increased synthesis of the (lumenal) glycolipid

globotriaosylceramide, and that this increase

could be inhibited by MDR1 inhibitors (Lala

et al., 2000).

Finally, sterols have been found to act as substrates for MDR1 with variable efficiencies

Whereas dexamethasone and cortisol are

rela-tively good substrates, progesterone binds to

the substrate-binding site without being

trans-ported, and is a good inhibitor MDR1 has

also been proposed to be involved in the

intracellular trafficking of cholesterol but it is

unclear whether the effect is related to MDR1

P-glycoprotein drug transport activity (Debry

et al., 1997; Field et al., 1995; Luker et al., 1999).

Alternatively, it might be linked to

transloca-tion of glucosylceramide or another

sphingo-lipid as cholesterol preferentially interacts with

sphingolipids Whether cholesterol, the major

mammalian membrane sterol, is translocated

awaits a direct transport experiment (Barnes

et al., 1996; Wang et al., 2000).

Traditional assays for measuring the fraction of

a lipid on one side of the membrane relied on

the use of phospholipases, labeling reagents and

lipid exchange techniques, and were not well

suited as sensitive assays for protein-mediated lipid translocation (Op den Kamp, 1979;

Sillence et al., 2000) Measurements have been

greatly facilitated by the use of analogues of membrane lipids, in which one long lipid chain has been replaced by a short C5-C6 acyl chain (Seigneuret and Devaux, 1984; Sleight and Pagano, 1985) The lower hydrophobicity enhances the off-rate from the membrane, allow-ing the analogues to exchange via the aqueous phase with half-times of seconds, whereas the natural lipids need hours As a first step, the analogues can thus be easily inserted into the surface of the membrane of interest For detec-tion, the analogues are labeled on the short chain with a spin-label, a fluorescent moiety or

a radiolabel Translocation can then be moni-tored in various ways One general principle is the ‘back-exchange’ After the incubation, exo-genous bovine serum albumin (BSA) is used

to selectively extract the short-chain analogue from the outer leaflet Alternatively, the spin-labeled or fluorescent analogues in the outer leaflet can be chemically quenched (Bosch

et al., 1997; Margolles et al., 1999; Romsicki and

Sharom, 2001) The activity of a translocator can

be measured as a change in the amount of ana-logue that has been translocated One can also compare the rate of analogue uptake from an exogenous source (or the resulting accu-mulation), which is a combination of inward and outward translocation Finally, a change in the equilibrium distribution of a natural lipid across the bilayer may reflect the activity of a

translocator (Dekkers et al., 2000) Still, it must

be realized that results obtained with analogues cannot be extrapolated to the natural lipids without further experiments

The ability of ABC transporters to transport lipid analogues was first shown for the PC

translocator encoded by the mouse Mdr2 P-glycoprotein (Abcb4) and human MDR3

P-glycoprotein (ABCB4) genes Translocation

of C6-NBD-PC

(N-6[7-nitro-2,1,3-benzoxadia-zol-4-yl]-amino-hexanoyl-phosphatidylcholine) from the cytosolic leaflet to the lumen (or lumenal leaflet) of secretory vesicles isolated

from yeast transformed with Mdr2 was higher

than in vesicles from control cells (Ruetz and Gros, 1994) Translocation was ATP- and Mg2⫹ -dependent, sensitive to the inhibitor verapamil, and appeared selective for Mdr2 since Mdr3 was inactive The activity of human MDR3 towards C6-NBD-PC was confirmed in

MDR3 transfected cells, where transport of

intracellularly synthesized C6-NBD-PC to the

LIPID TRANSPORT BY ABC TRANSPORTERS 465

cell surface was measured by the BSA assay

described earlier (van Helvoort et al., 1996).

However, in contrast to the results of

fluor-escence quenching experiments (Ruetz and

Gros, 1994), the experiments on MDR1

trans-fected cells demonstrated that human MDR1

(and mouse Mdr3) possess the same capability

as human MDR3 to translocate C6-NBD-PC

Unexpectedly, MDR1 was found to translocate

a wide variety of short-chain analogues (Table

22.2) This finding was corroborated by

subse-quent work on intact cells (Bosch et al., 1997),

and in reconstituted proteoliposomes with

hamster Abcb1 (Romsicki and Sharom, 2001)

The hamster P-glycoprotein displayed even

less specificity than the human MDR1, because

in the fluorescence quenching assay of Ruetz

and Gros (1994), it transported various

ana-logues of phosphatidylserine which were not

recognized by human MDR1 in the intact cell

system Hamster Abcb1 even transported

phospholipid analogues with long acyl chains

and a phosphoethanolamine-NBD headgroup,

N-NBD-PE (Romsicki and Sharom, 2001).

These analogues may resemble the

physiologi-cal lipid N-retinylidene-PE, a presumed

sub-strate of ABCR (ABCA4), the rod outer

segment disk ABC transporter (see Chapter 28)

(Sun et al., 1999; Weng et al., 1999).

Like MDR1, human MRP1 (ABCC1) was found to transport only C6-NBD sphingolipid

analogues, but translocation by MRP1

depended on the presence of the NBD moiety

(Raggers et al., 1999) C6-NBD-PS is also

trans-ported by human and mouse MRP1/Mrp1

(Dekkers et al., 1998; Kamp and Haest, 1998).

Most convincingly, outward transport of

C6-NBD-PS across the erythrocyte membrane

was completely absent in erythrocytes from

Mrp1( ⫺/⫺) mice (Dekkers et al., 1998) The

situ-ation for PC is more complex The PCs PAF

and C16:0/C6-NBD-PC were not substrates

(Ernest and Bello-Reuss, 1999; Raggers et al.,

1999) However, in human erythrocytes, the

outward movement of C18:1/C6-NBD-PC and

C14:0/C12-NBD-PC required reduced

gluta-thione (GSH) and was strongly inhibited by

MRP1/Mrp1 inhibitors (Dekkers et al., 1998;

Kamp and Haest, 1998) One explanation for the

difference could be the difference in fatty acid at

the sn-1 position As might be expected from the

similarities in substrate specificity, ABC

trans-porters from yeasts and bacteria have also been

found to be capable of translocating lipid

ana-logues The available data are summarized in

Table 22.2

ABCB4)

MDR3/MDR2 TRANSLOCATES PHOSPHATIDYLCHOLINE

The MDR3 (ABCB4) gene for the human PC

translocator is located on chromosome 7 (q 21.1),

only 34 kb downstream of the MDR1 (ABCB1) gene (Lincke et al., 1991) A similar gene cluster

on chromosome 5 of the mouse contains the

mouse orthologue, Mdr2 (Kirschner, 1995; Raymond et al., 1990) Disruption of this gene led

to the discovery that the Mdr2 P-glycoprotein is essential for secretion of long-chain PC into bile

(Smit et al., 1993) Mdr2(⫹/⫺)-heterozygotes have no defects, but secrete only about half as

much PC as wild-type Mdr2(⫹/⫹) mice (Smit

et al., 1993) The absence of PC in the bile of Mdr2 (⫺/⫺) mice leads to a mild liver disease, because bile salt secretion is normal in these mice and the high bile salt concentrations, without accompanying PC, damage the canalicular mem-brane of the hepatocyte and the small bile ducts This causes extensive bile duct proliferation and

some hepatocyte damage (Mauad et al., 1994; Oude Elferink et al., 1997) All defects in these

mice are due to the absence of Mdr2 in the liver,

as they can be completely prevented in the KO mice by the liver-specific expression of the

human MDR3 gene under the control of an

albu-min promoter active only in the liver (Smith

et al., 1998) The murine, rat and human PC

translocator genes are also expressed at low levels in adrenal glands, skeletal and heart

mus-cle, tonsil and spleen (see Smit et al., 1994), and

the protein has been detected in murine ery-throcytes (Vermeulen, 1996) No function for the PC translocator has been found, however, in any other tissue than liver

Our present ideas about PC secretion from

hepatocytes into bile are summarized in Figure 22.1 PC secretion depends both on bile salts and

on the Mdr2 P-glycoprotein If either is lacking,

no PC secretion is detectable If Mdr2 is present

in the hepatocyte canalicular membrane, the rate of PC secretion is hyperbolically dependent

on the bile salt concentration Interestingly, PC secretion is dependent on the Mdr2 levels at all

bile salt concentrations (Figure 22.2) PC

secre-tion is higher in wild-type Mdr2(⫹/⫹) mice

than in Mdr2(⫹/⫺) heterozygotes Secretion is

TABLE22.2 TRANSPORT OF MEMBRANE LIPID ANALOGUES BY MULTIDRUG TRANSPORTERS

Speciesa Glycerophospholipids Sphingolipids SM References

PC PE PS GlcCer

H C8:0/C8:0 C8:0/C8:0 C8:1/C8:0c, C6 C8:1/C8:0c 3

not C12-NBD

H C16:0/C6-NBD not C16:0/C6-NBD not C6-NBD not C6-NBD 5

H notC16:0/C6-NBD not C16:0/C6-NBD C6-NBD not C6 C6-NBD 5

a CH, Chinese hamster; H, human; L, Lactococcus lactis; M, mouse; S, Saccharomyces cerevisiae.

bSimilar data for C12-NBD-PC and –PS and for N-NBD-diC18:1.

cContain a truncated ceramide, consisting of C8 sphingosine amide-linked to C8:0 fatty acid.

dIn addition, evidence was provided for the translocation of natural PS.

1, Ernest and Bello-Reuss (1999); 2, Raggers et al (2001); 3, van Helvoort et al (1996); 4, van Helvoort et al (1997); 5, Raggers et al (1999); 6, Romsicki and Sharom (2001);

7, Bosch et al (1997); 8, Ruetz and Gros (1994); 9, Dekkers et al (1998); 10, Kamp and Haest (1998); 11, Dekkers et al (2000); 12, Hamon et al (2000); 13, Decottignies et al (1998);

14, Margolles et al (1999).

Bile salt Cholesterol PC Other phospholipids (Glyco)sphingolipids Cytosol Canaliculus

Mdr2

Mdr2

ATP

ATP

ATP

BSEP

Bile salt (micelles)

Figure 22.1 The role of Mdr2/MDR3 P-glycoprotein in bile formation (modified from Borst et al., 2000) cBAT

is the canalicular bile acid transporter, BSEP (ABCB11) Phosphatidylcholine (PC) from the inner leaflet of the canalicular membrane is flipped by Mdr2/MDR3 (ABCB4) to the exoplasmic leaflet where it is accessible

to extraction by bile salts.

0 30 60 90 120 150

A1 0%

8%

⫹/⫺ 50%

⫹Ⲑ⫹ 100%

A63 180%

Bile salt output (nmol/min.100 g)

⫺/⫺

Figure 22.2 Relation between bile salt and phospholipid transport in mice with different expression levels of

as A63 mice (transgenic for the MDR3 gene against wild-type background; open squares) and A1 mice

amounts of the bile salt tauroursodeoxycholate, while bile was continuously collected Phospholipid secretion is hyperbolically dependent on bile secretion and the maximal phospholipid secretion capacity is strictly dependent on the expression levels of Mdr2/MDR3 The total expression levels are given in the figure, expressed as percentages of that in wild-type mice Modified from Oude Elferink et al (1998).

highest in mice transgenic for the human MDR3

gene, which have a supraphysiological PC

translocator concentration in their livers, and

very low in the A1 Mdr2(⫺/⫺) homozygote

with low MDR3 transgene expression (Smith

et al., 1998) Crawford and co-workers (1997)

have shown by electron microscopy that

adher-ent monolamellar vesicles on the outer

canali-cular membrane may represent an

intermed-iate structure in the secretion process of biliary

lipid Formation of these vesicles requires the

presence of functional Mdr2 Interestingly, the

appearance of lipoprotein X (LpX) in the blood

of cholestatic mice is also completely dependent

on functional Mdr2 (Oude Elferink et al., 1998).

LpX consists of 40–100 nm vesicles comprising

phospholipid and cholesterol with an aqueous

lumen They appear shortly after bile duct

liga-tion in wild-type mice, but not in Mdr2(⫺/⫺)

mice How LpX reaches the blood is not known

Oude Elferink et al (1998) favor a model in

which biliary vesicles continue to be formed at

the canalicular membrane after ligation and the

LpX vesicles reach the blood by transcytosis

through the hepatocyte Another possibility is

that the increased pressure in the biliary

com-partment after ligation is released from time to

time by opening of the tight junctions between

hepatocytes and a paracellular flux of bile into

the blood

(MDR3/MDR2, ABCB4) CAN ALSO

TRANSPORT DRUGS

Initial experiments on the substrate specificity

of the MDR3 PC translocator were done with

membrane vesicles from transgenic yeast

over-expressing the murine orthologue Mdr2 With

this system, Ruetz and Gros (1994) showed that

the translocator is highly specific for

phospho-lipid analogues with a choline head group and

this was confirmed by van Helvoort et al (1996)

in animal cells In addition, they found that the

protein is selective towards the PC fatty acid

moieties

All these results suggested that the PC translocator had evolved for the specific

pur-pose of transporting natural membrane PC

into bile It therefore came as a surprise that

MDR3/Mdr2 is inhibited by verapamil, a

clas-sical inhibitor of P-glycoprotein (Ruetz and

Gros, 1994; van Helvoort et al., 1996) The group

of Ueda (Kino et al., 1996) then showed that

transfection of yeast cells with an MDR3 cDNA

construct resulted in low-level resistance to the antifungal agent aureobasidin, also a substrate of MDR1 No multidrug resistance had ever been seen in animal cells transfected

with MDR3 or Mdr2 cDNA constructs, but this

is a rather insensitive assay for drug transport

Drug transport through epithelial monolayers provides a more sensitive assay, and Smith

et al (2000), using pig kidney cell monolayers

expressing MDR3, found substantial transport

of digoxin, lower rates of transport of paclitaxel, daunorubicin and vinblastine, but no significant transport of other drugs that are transported

at high rate by the drug-transporting P-glycoproteins, such as CsA and dexamethasone

Digoxin transport by MDR3 was efficiently inhibited by P-glycoprotein inhibitors, includ-ing CsA, PSC833 and verapamil To exclude the unlikely possibility that MDR3 had activated an

endogenous drug transporter, Smith et al (2000)

verified that the protein interacts with drugs by studying nucleotide trapping by MDR3 They found that the substrates paclitaxel and vinblas-tine, and the inhibitors CsA and PSC833, were able to decrease nucleotide trapping by 90% or more in concentrations similar to those used for inhibiting MDR1

As we have pointed out elsewhere (Borst

et al., 2000; Smith et al., 2000), it is puzzling that

the MDR3/Mdr2 PC translocator can bind and transport drugs in a membrane environment full of long-chain PC and that transport of PC analogues by the protein is efficiently inhibited

by drugs Whatever the explanation for this apparent paradox, it is important to

acknowl-edge that the PC translocator can be inhibited

by drugs There is now ample evidence that patients with a diminished level of MDR3 are

at risk of intrahepatic cholestasis (see below)

Some drugs might increase that risk

The possibility also remains that the MDR3

PC translocator might contribute to the MDR phenotype in some types of human cancer This was first suggested by studies of Nooter and co-workers on drug-resistant B-cell leukemias

They noted that cells with substantial MDR3

expression had diminished daunorubicin uptake

that was reversed by CsA (Herweijer et al., 1990;

Nooter et al., 1990) MDR3 expression also

corre-lated negatively with clinical outcome This was confirmed in a larger group of patients by Arai

et al (1997) Although these results are

compati-ble with a contribution of the MDR3 PC translo-cator to resistance through its ability to transport drugs, intervention studies with specific block-ers will be required to prove the point

LIPID TRANSPORT BY ABC TRANSPORTERS 469

DEFECTS IN THEMDR2/MDR3 (ABCB4)

GENE LEAD TO LIVER DISEASE IN MICE

AND HUMANS

Mdr2(⫺/⫺) mice, unable to make any PC

trans-locator, are normal at birth Only when bile flow

starts do the symptoms of a non-suppurative

inflammatory cholangitis appear, with portal

inflammation and proliferation of the bile ducts

(Mauad et al., 1994; Smit et al., 1993) At the age

of 4–6 months, the mice start to develop

multi-ple foci in their liver parenchyma, which

eventu-ally progress to tumors, often with necrosis and

hemorrhage These tumors may metastasize to

the lung (Mauad et al., 1994) The liver disease

in the Mdr2(⫺/⫺) mice is accompanied by an

increased bile flow and a strongly decreased

cholesterol and GSH secretion into bile (Smit

et al., 1993) These abnormalities are the

conse-quence of the inflammatory cholangitis caused

by the (normal) secretion of bile salts without

accompanying PC One would expect the

sever-ity of the cholangitis to be affected by the nature

of the bile salts secreted and this has been

veri-fied Thus, if the diet is supplemented with

the more hydrophobic bile salt cholate, liver

pathology worsens, while ingestion of a more

hydrophilic bile salt, ursodeoxycholate, results

in decreased liver damage (Van Nieuwkerk

et al., 1996) The severity of liver damage in the

Mdr2(⫺/⫺) mice is therefore dependent on bile

salt hydrophobicity A level of 15% of normal PC

secretion, induced by expression of a MDR3

transgene in the liver, prevented liver defects

on a standard diet in male mice and mitigated

pathology in female mice (De Vree, 1999) The

fact that female mice suffer more from defects

in the Mdr2 PC translocator than male mice

correlates with females having a more

hydro-phobic bile salt composition and, hence, increased

cytotoxicity of bile salts (Van Nieuwkerk

et al., 1997).

The Mdr2 gene in the liver is relatively

imper-vious to drastic treatments that strongly affect

other liver ABC transporters, such as bile duct

ligation, partial hepatectomy or endotoxin

treat-ment (Vos et al., 1998, 1999) However, when

mice were fed a diet supplemented with cholate,

their liver Mdr2 mRNA and PC secretion

capac-ity increased by approximately 50% (Frijters

et al., 1997) A much larger induction of the

Mdr2 gene was observed in mice exposed to

compounds that induce peroxisome

prolifera-tion (Chianale et al., 1996; Miranda et al., 1997).

Thus, a nearly sixfold increase in Mdr2 mRNA

was induced by 2,4,5-trichlorophenoxyacetic acid

and this was accompanied by a fivefold

increase in Mdr2 mRNA synthesis Biliary PC

secretion went up only twofold and it is possible that Mdr2 protein levels were less elevated than the mRNA level, or that PC supply to the pro-tein becomes rate limiting at these extreme

lev-els of Mdr2 induction Two groups (Carralla et

al., 1999; Hooiveld et al., 1999) have

indepen-dently shown that Mdr2 expression is induced

by treatment of rats with statins (inhibitors

of HMG-CoA reductase) Hooiveld et al (1999)

proposed that this induction is mediated by the sterol regulatory binding protein, SREBP, because the 5⬘ flanking region of Mdr2 contains

a potential sterol responsive element Statin treatment causes a transient induction of SREBP expression, due to the depletion of cholesterol

It is not clear what the physiological function of

increased Mdr2 expression during sterol

deple-tion would be

Whereas overexpression of the MDR3/Mdr2

PC translocator gene in liver had no obvious

deleterious effects, mice expressing an MDR3

gene under a vimentin promoter develop

cata-racts (Dunia et al., 1996) and a peripheral neuro-pathy (Smit et al., 1996) It is not clear whether

these pathological consequences are due to a redistribution of PC in membranes or only to the presence of a bulky glycoprotein in mem-branes where it does not normally belong Although it was obvious that there should be

a human counterpart of the Mdr2(⫺/⫺) mouse,

it took until 1996 before Deleuze et al (1996) found that expression of MDR3 was absent in a

patient with progressive familial intrahepatic cholestasis (PFIC) This patient belonged to a subgroup of PFIC patients characterized by

a high serum ␥-glutamyltransferase (GGT), strong bile duct proliferation, and eventual liver cirrhosis requiring liver transplantation This subgroup is now called PFIC, type 3 In sub-sequent work, 17 out of 31 patients with high GGT PFIC were found to have a mutation in

the MDR3 (ABCB4) gene (De Vree et al., 1998; Jacquemin et al., 1999, 2001) Since the gene was

not completely sequenced in all these patients, it

is not clear whether the remaining 14 patients

have as yet unidentified MDR3 mutations, or

that another gene might also be involved in this form of PFIC

Defects in the MDR3 gene do not only give rise to pediatric liver disease Jacquemin et al.

(1999) reported that the mother of a patient with PFIC type 3 and several other women from this family suffered from intrahepatic cholestasis of pregnancy These women turned