CHAPTER 25 – ABC TRANSPORTERS IN MITOCHONDRIA

CHAPTER 25 – ABC TRANSPORTERS IN MITOCHONDRIA CHAPTER 25 – ABC TRANSPORTERS IN MITOCHONDRIA CHAPTER 25 – ABC TRANSPORTERS IN MITOCHONDRIA CHAPTER 25 – ABC TRANSPORTERS IN MITOCHONDRIA CHAPTER 25 – ABC TRANSPORTERS IN MITOCHONDRIA CHAPTER 25 – ABC TRANSPORTERS IN MITOCHONDRIA CHAPTER 25 – ABC TRANSPORTERS IN MITOCHONDRIA

I NTRODUCTION

Mitochondria are essential organelles of most

eukaryotic cells including fungi, invertebrates,

vertebrates and plants They perform various

processes such as oxidative phosphorylation,

the tricarboxylic acid cycle, fatty acid oxidation,

the biosynthesis of various amino acids, the

generation of iron-sulfur (Fe/S) clusters and

their insertion into apoproteins, as well as

par-tial reactions of heme biosynthesis and the urea

cycle According to the endosymbiont

hypo-thesis, virtually all of these functions have been

inherited from the bacterial ancestor of the

pres-ent-day mitochondrion, an ␣-proteobacterium

Hence, both the components and mechanisms

of the shared processes are highly related in

mitochondria and bacteria

In contrast to the aforementioned functions, reactions including membrane transport of

proteins, peptides, sugars, metabolites, vitamins

and lipids into and out of the organelle differ

quite significantly from those operating in

bac-teria For instance, the mitochondrial protein

import system involving the TOM and TIM

preprotein translocases does not exist in

bacte-ria (Neupert, 1997; Pfanner and Geissler, 2001)

Likewise, only one of the bacterial protein

export systems has been maintained in

mito-chondria, namely the Oxa1/YidC complex

(Dalbey and Kuhn, 2000) Striking differences

between mitochondria and bacteria also exist

with respect to trafficking small molecules To

facilitate this task, mitochondria contain more than 30 so-called ‘carrier’ proteins, which trans-port a variety of compounds (e.g nucleotides, di- and tricarboxylates, vitamins and amino acids) across the inner membrane (reviewed by

El Moualij et al., 1997; Nelson et al., 1998;

Palmieri et al., 2000).

No bacterial counterparts of these carrier pro-teins are known Apparently, mitochondrial car-rier proteins have replaced most of the versatile membrane transport functions performed by ATP-binding cassette (ABC) transporters of the bacterial ancestors of mitochondria In

present-day bacteria such as Escherichia coli, more than

50 members of this large protein family are found, and they are crucial for transport into and out of the bacterial cytosol (Linton and Higgins, 1998) In comparison, only a small number of ABC transporters exist in mitochon-dria Strikingly, both structural and functional evidence suggests that these mitochondrial transporters do not closely resemble any of the bacterial counterparts, but rather represent pro-teins with a role specifically adapted for eukar-yotic cells Today, we can distinguish different types of mitochondrial ABC transporters Two types belong to subclass B of the ABC trans-porter superfamily (MDR-like proteins) (Bauer

et al., 1999; Taglicht and Michaelis, 1998) and

are distinguished according to their degree of homology to the three ABC transporters present

in the yeast Saccharomyces cerevisiae, namely the

Atm1p-like proteins and the Mdl1p/Mdl2p-like

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

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

25

CHAPTER

proteins An additional type of ABC transporter,

termed CcmAB, may exist in plant

mitochon-dria, but to date only its membrane-spanning

domain (CcmB) has been identified

This review will summarize our current knowledge of mitochondrial ABC transporters

We shall first address the properties and

func-tions of the mitochondrial ABC transporters in

S cerevisiae Then, we shall introduce the ABC

transporters of mammalian cells and discuss

their (putative) functions in comparison to

those defined for the yeast proteins Finally, we

shall briefly review recent insights into plant

mitochondrial ABC transporters and their

(putative) functions

S CEREVISIAE

IDENTIFICATION OF THE FIRST MITOCHONDRIALABC TRANSPORTER,

YEASTATM1P

Based on the bacterial origin of mitochondria, Leighton and Schatz (1995) predicted the existence of ABC transporters in these organelles By using a polymerase chain reaction

TABLE25.1 MITOCHONDRIALABC TRANSPORTERS Name of ABC Chromosomal Amino acid Molecular Homologous to (Putative) transporter localization residues mass (kDa) yeast protein Function Yeast(Saccharomyces cerevisiae)

Atm1p XIII 690 78 – Maturation of

` cytosolic Fe/S

proteins, Iron homeostasis Mdl1p XII 695 76 – Peptide export

Man(Homo sapiens)

hABC7 Xq13.1–q13.3 752 83 Atm1p (47%) Maturation of

cytosolic Fe/S proteins, Iron homeostasis MTABC3 2q36 842 94 Atm1p (38%) Iron homeostasis M-ABC1 7q35–q36 718 78 Mdl2p (34%) ?

Mdl1p (32%) M-ABC2 1q42 738 79 Mdl1p (42%) ?

Mdl2p (38%)

Mouse(Mus musculus)

ABC-me – 715 77 Mdl1p (39%) Heme transport ?

Mdl2p (37%)

Plants

Sta1 (A thaliana) V 728 80 Atm1p (45%) Maturation of

cytosolic Fe/S proteins

Sta2 (A thaliana) IV 680 76 Atm1p (44%) ?

Sta3 (A thaliana) IV 678 75 Atm1p (45%) ?

CcmB (Triticum aestivum) – 206 24 E coli CcmB (27%) c-type cytochrome

(Membrane domain) biogenesis?

In all cases, a (putative) N-terminal mitochondrial presequence might be cleaved from the proteins, thus resulting in slightly shorter mature forms The highest sequence homology between the listed mammalian or plant proteins and

the Saccharomyces cerevisiae proteins is given as the fraction of identical amino acid residues in both proteins For

references see text.

(PCR) approach, they identified genes for several

of the S cerevisiae ABC transporters The first

mitochondrial representative, termed Atm1p,

was identified by virtue of an N-terminal

sequence resembling a mitochondrial targeting

signal (presequence) In a parallel genetic screen

originally intended to isolate new components

of the biogenesis of c-type cytochromes (Kranz

et al., 1998), a temperature-sensitive mutant of

the yeast ATM1 gene (Kispal et al., 1997) was

found This encodes a protein comprising 690

amino acid residues with six putative

transmem-brane segments and a C-terminal ATP-binding

domain (Table 25.1) exhibiting the characteristic

features of ABC transporter proteins Atm1p

therefore belongs to the group of ‘half

trans-porters’ It should be mentioned that no attempts

have been made so far to determine precisely

the structural mode of membrane integration of

Atm1p (or of the other mitochondrial ABC

trans-porters) Different algorithms used to predict

transmembrane helices have identified five to

six hydrophobic sequences that fulfill the criteria

for membrane integration Thus, by analogy

with classical ABC transporters (Higgins, 1992),

the Atm1p polypeptide chain may be expected

to span the membrane six times and the

func-tional protein may be a homodimer consisting of

two molecules of Atm1p (Figure 25.1).

The function of the N-terminus of Atm1p as

a mitochondrial presequence was verified by

its ability to target attached proteins to

mito-chondria (Leighton and Schatz, 1995) The

pre-cise localization of the Atm1p presequence

cleavage site is not known but, based on the consensus sequence recognized by matrix pro-cessing peptidase (MPP), it is predicted to

be after amino acid residues 25 or 41 Subcellu-lar localization of Atm1p was demonstrated

by immunostaining of cell fractions and by immunofluorescence Atm1p is localized in the mitochondrial inner membrane with the nucleotide-binding domain facing the matrix

space (Figure 25.1) We presume, as will be

developed in later sections, that Atm1p is pre-dicted to function as an exporter of compounds from the matrix to the intermembrane space

DELETION OF THE YEASTATM1GENE

Cells deficient in the ATM1 gene (strain ⌬atm1)

display a strong growth defect on rich media

containing glucose (Kispal et al., 1997; Leighton

and Schatz, 1995) and do not grow on non-fermentable carbon sources such as glycerol

The rate of growth of ⌬atm1 cells in the presence

of glucose is much slower than that of cells harboring mitochondria defective in respiration

Thus, Atm1p plays a role that goes beyond the formation of respiratory competent mitochon-dria Another phenotype resulting from the

deletion of ATM1 is a large reduction in the level

of holocytochromes (Kispal et al., 1997; Leighton

and Schatz, 1995) Immunostaining analysis showed that this is not due to the defective

biosynthesis of the apoforms of the c-type

cytochromes in ⌬atm1 cells (Kispal et al., 1997)

IMS

Ma

MIM N

Atm1p

N

ATP

ATP

N

ATP

N

Mdl1p

MOM Cytosol

Mdl2p

ATP

Figure 25.1 Model for the membrane orientation of the yeast mitochondrial ABC transporters

All three known yeast ABC transporters, Atm1p, Mdl1p and Mdl2p, share a similar membrane orientation

with the N-terminus (N) facing the matrix space, an N-terminal ATP-binding domain and a C-terminal

membrane-spanning domain with six putative transmembrane helices The drawing represents

the predicted size of the loops between the membrane segments and indicates the formation of possible

homodimers MOM, mitochondrial outer membrane; IMS, intermembrane space; MIM, mitochondrial

inner membrane; Ma, matrix.

Moreover, heme biosynthesis occurs at

wild-type rates in these cells (H Lange,

unpub-lished) Consequently, cells defective in Atm1p

seem to face a condition leading to the

degrada-tion of protein-bound heme, which can most

probably be explained by the oxidative stress

prevailing in ⌬atm1 cells (Kispal et al., 1997)

Reduced and oxidized glutathione, which are

the most important compounds required to

bal-ance the cellular redox level in yeast, are

sub-stantially increased in ⌬atm1 cells (Kispal et al.,

1997) The state of oxidative stress itself may be

a consequence of the dramatic increase in the

concentration of ‘free’ iron (i.e non-heme and

non Fe/S iron), which appears to be an early

phenotype resulting from the loss of Atm1p

function (Kispal et al., 1999) Together with the

mitochondrial matrix protein Yfh1p (frataxin),

Atm1p was the first protein for which a function

in mitochondrial iron homeostasis could be

demonstrated (Babcock et al., 1997; Foury and

Cazzalini, 1997; Kispal et al., 1997).

In some genetic backgrounds, ⌬atm1 cells lose mitochondrial DNA to yield so-called ␳0

cells (Leighton and Schatz, 1995; Senbongi

et al., 1999) This phenomenon is not an

obliga-tory consequence of the inactivation of ATM1;

for example, deletion of the gene in strain W303

does not result in ␳0 cells (Kispal et al., 1997).

Thus, loss of mitochondrial DNA may well be

an indirect consequence of the oxidative

dam-age resulting from iron overload and

impair-ment of components of the machinery involved

in mitochondrial DNA maintenance (Kaufman

et al., 2000).

ROLE OFATM1P IN THE MATURATION OF

CYTOSOLICFE/S PROTEINS

Of all these pleiotropic phenotypes associated

with ⌬atm1 cells none provide any clues

towards the understanding of the function of

the ABC transporter Initial insight into the

process in which Atm1p is involved came from

the observation that ⌬atm1 cells fail to grow

without added leucine (Kispal et al., 1999) In

yeast, leucine is synthesized from the common

leucine/valine precursor ␣-ketoisovalerate by

three specific steps catalyzed by the enzymes

␣-isopropyl malate synthase (Leu4p and Leu9p),

isopropyl malate isomerase (Leu1p) and

␤-isopropyl malate dehydrogenase (Leu2p) (see

reaction schemes in Hinnebusch, 1992; Jones

and Fink, 1982; Prohl et al., 2001) These enzymes

are compartmentalized, distributed between the

mitochondrial matrix (Leu4p and Leu9p) and the cytosol (Leu4p, Leu1p and Leu2p) (Beltzer

et al., 1988; Casalone et al., 2000; Kohlhaw, 1988a,

1988b) Measurements of individual enzymatic activities showed a quantitative deficiency of isopropyl malate isomerase (Leu1p) in ⌬atm1 cells while the other enzymes were active at

wild-type levels (Kispal et al., 1999).

What is the reason for these observations? Leu1p is a cytosolic protein that requires an Fe/S cluster, generated in this mitochon-drial matrix, for activity Leu1p closely resem-bles aconitase of the mitochondrial matrix (Kohlhaw, 1988b) However, in contrast to Leu1p, mitochondrial aconitase, another Fe/S protein, exhibits almost wild-type activity in

⌬atm1 cells, rendering a general defect in

cellu-lar Fe/S proteins unlikely (Kispal et al., 1997).

Rather, the specific defect in Leu1p indicated that Atm1p may perform a function in the maturation of extra-mitochondrial Fe/S

pro-teins (Kispal et al., 1999).

To investigate the immediate effects of Atm1p deficiency, as opposed to long-term con-sequences (see above), a yeast mutant in which

expression of the ATM1 gene was under the

control of a galactose-regulatable promoter

(Gal-ATM1 cells) was created (Kispal et al.,

1999) These cells can readily be depleted of Atm1p when grown in the absence of galactose Nevertheless, in the presence of galactose they

do not exhibit a dramatic growth defect nor do they display any of the pleiotropic phenotypes reported above (e.g cytochrome deficiency, oxidative stress) Upon depletion of Atm1p, the activity of Leu1p decreased at least 10-fold, indicating that incorporation of the Fe/S cluster into the cytosolic Leu1p apoprotein is an early consequence of Atm1p deficiency A direct func-tion of Atm1p in the assembly of the Fe/S cluster holoprotein, Leu1p, could be shown by briefly radiolabeling wild-type cells with fer-rous iron (55Fe), followed by immunoprecipita-tion of Leu1p from cell extracts using specific

antibodies (Kispal et al., 1999) The radioactive

iron associated with Leu1p served as a direct measure of the formation of the Fe/S cluster in Leu1p Cells lacking Atm1p did not incorporate any significant 55Fe radioactivity into Leu1p These results provided convincing evidence for the involvement of Atm1p in the maturation of

a cytosolic Fe/S protein

Recently, these results have been supported and extended by the analysis of another cyto-solic Fe/S protein, namely the essential protein Rli1p, which harbors an Fe/S cluster domain at

its N-terminus (G Kispal, unpublished).

Assembly of the Fe/S cluster in Rli1p also

involves the function of Atm1p, suggesting a

general role for this ABC transporter in the

bio-genesis of extra-mitochondrial Fe/S proteins

Atm1p function in cytosolic Fe/S protein

matu-ration is highly specific because no defects were

observed in Fe/S proteins localized inside

mito-chondria upon depletion of Atm1p (Kispal

et al., 1999) For a better understanding of the

distinct function of Atm1p in the maturation of

cytosolic Fe/S proteins, it is necessary to

pro-vide a brief outline of the biogenesis of Fe/S

proteins in a eukaryotic cell For a more

compre-hensive discussion of this recently discovered

process, the reader is referred to several detailed

reviews (Craig et al., 1999; Lill et al., 1999; Lill

and Kispal, 2000; Mühlenhoff and Lill, 2000)

BIOGENESIS OF EUKARYOTIC

FE/S PROTEINS

Assembly of mitochondrial Fe/S proteins

Many studies over the past four years have led

to the identification of some ten proteins of the

mitochondrial matrix, which play a role in the

formation of the Fe/S clusters and their

incor-poration into mitochondrial apoproteins (for

examples see Garland et al., 1999; Jensen and

Culotta, 2000; Kaut et al., 2000; Kim et al., 2001;

Kispal et al., 1999; Lange et al., 2000; Li et al.,

2001; Pelzer et al., 2000; Schilke et al., 1999;

Strain et al., 1998; Voisine et al., 2001) These

proteins are highly homologous to bacterial

pro-teins encoded by the isc (iron sulfur cluster)

operons (Zheng et al., 1998), and were therefore

defined as compounds of the ‘ISC assembly

machinery’ (Lill and Kispal, 2000) Even though

virtually all of these proteins have been shown

to participate in the assembly of Fe/S clusters,

comparatively little is known about the precise

roles of individual proteins or the overall

molec-ular mechanism of the pathway Nevertheless,

a number of functional studies have been

per-formed on the bacterial Isc proteins (for

exam-ples see Agar et al., 2000b, 2000c; Hoff et al.,

2000; Krebs et al., 2001; Ollagnier-de-Choudens

et al., 2001; Silberg et al., 2000; Yuvaniyama

et al., 2000; Zheng et al., 1993, 1994) Thus, the

following model combines knowledge gained

from studies on both mitochondrial and

bac-terial Isc proteins, assuming that the process

is highly similar in both environments

How-ever, it should be emphasized that we are just

beginning to understand Fe/S cluster biogene-sis, and any putative mechanistic pathways are based on rather limited experimental evidence

According to a present working model,

shown in Figure 25.2, iron, after its membrane

potential-dependent import into mitochondria

(Lange et al., 1999), binds to the two proteins,

Isu1p and Isu2p The cysteine desulfurase Nfs1p generates elemental sulfur (S0) from cysteine, which is then used to form an ‘intermediate’

Cytosol

Ala Cys

ABC transporter Atm1p

ISC assembly machinery

Extra-mitochondrial Fe/S proteins

Mitochondrial Fe/S proteins

pmf

Iron

Mitochondrion

Fe Fe

S S

Fe Fe

S S

Holo Apo

ISC export machinery

Nfs1p

Erv1p

Isu1/2p

?

Arh1p Yah1p

e ⫺ Fe

Fe

S S

Figure 25.2 Working model for the function of Atm1p in cytosolic Fe/S protein assembly in eukaryotic cells The assembly of Fe/S clusters, for both mitochondrial and cytosolic Fe/S proteins, is achieved by the ISC assembly machinery First, ferrous iron enters the mitochondrial matrix in a membrane potential (pmf)-dependent step Iron binds to the Isu proteins which provide a scaffold for the assembly of the Fe/S clusters The cysteine desulfurase, Nfs1p, generates elemental sulfur (S 0 ) from cysteine needed for Fe/S cluster formation on the Isu proteins The nascent Fe/S clusters are released from the Isu proteins upon reduction by the electron transfer chain shuttling electrons from NAD(P)H to the ferredoxin reductase Arh1p and the ferredoxin Yah1p The Fe/S clusters are then

incorporated into the apoforms of mitochondrial Fe/S proteins or exported to the cytosol, a step most likely involving Atm1p The exact nature of the substrate of Atm1p is not known yet, but a likely compound is a chelated Fe/S cluster The export process may be assisted by Erv1p, a sulfhydryl oxidase in the intermembrane space It should be noted that many of the proposed steps of this model need further experimental verification.

[2Fe-2S] cluster on Isu1p/Isu2p (Yuvaniyama

et al., 2000) This cluster may further be

modi-fied to generate a [4Fe-4S] cluster (Agar et al.,

2000a) The next steps of Fe/S cluster release

and incorporation into apoproteins have not

been defined experimentally, leaving us to

spec-ulate about the possible mechanism In vitro, the

intermediate Fe/S cluster can be released from

the Isu proteins upon the addition of reducing

agents Therefore, the ferredoxin reductase

Arh1p and the ferredoxin Yah1p may form an

electron transfer chain that provides the

reduc-ing electrons for the release of the Fe/S cluster

from the Isu proteins (Lange et al., 2000; Li

et al., 2001).

The fate of the released Fe/S cluster is unknown It may be transferred to and

incorpo-rated into the apoproteins spontaneously, or the

process may need the help of accessory proteins

It is tempting to speculate that the insertion of

the Fe/S cluster into apoproteins is a

protein-assisted reaction Stabilization of the

apopro-teins before incorporation of the Fe/S cluster

could be an obvious task of the two

mitochon-drial heat shock proteins of the Hsp70/DnaK

and Hsp40/DnaJ classes, Ssq1p and Jac1p,

respectively (Kim et al., 2001; Lutz et al., 2001;

Schilke et al., 1999; Strain et al., 1998; Voisine

et al., 2001) However, evidence for an

inter-action between the chaperones and the

apo-proteins has not, so far, been reported On the

contrary, the bacterial homologues of the two

heat shock proteins have been shown to bind to

the Isu proteins, leading to a stimulation of the

ATPase activity of the Hsp70 chaperone (Hoff

et al., 2000; Silberg et al., 2000) The mechanistic

significance of this interaction remains to be

discovered

The Isa proteins have recently been shown

to be crucial for Fe/S cluster assembly (Jensen

and Culotta, 2000; Kaut et al., 2000; Pelzer

et al., 2000) and, according to in vitro data, they

may provide the necessary scaffold for the

assembly of these Fe/S clusters (Krebs et al.,

2001; Ollagnier-de-Choudens et al., 2001) Thus,

the Isa proteins may represent an alternative to

the Isu proteins in the assembly of the Fe/S

clusters Finally, a requirement for frataxin

(yeast Yfh1p) for the normal activity of

mito-chondrial Fe/S proteins has been documented,

even though the effects of deleting the frataxin

gene were not dramatic (Foury, 1999; Rötig

et al., 1997) According to a recent study,

frataxin might play a role in the storage of iron

in mitochondria (Adamec et al., 2000) Thus,

the requirement for frataxin in Fe/S protein

maturation might well be an indirect conse-quence of the impaired delivery of iron to the Isu and Isa proteins

Maturation of extra-mitochondrial Fe/S proteins

In addition to the assembly of mitochondrial Fe/S proteins, the ISC assembly machinery also plays a crucial role in the maturation of extra-mitochondrial Fe/S proteins The currently available data suggest that the Fe/S clusters

of cytosolic Fe/S proteins are assembled in the mitochondrial matrix and, therefore, need to be exported, in some form, from mitochondria (summarized by Lill and Kispal, 2000) This contention is based on the fact that depletion

of the mitochondrial Isc components abolishes cytosolic Fe/S protein maturation Nevertheless, the molecular moiety leaving the organelle is not known at present Similarly, we are only just beginning to understand the molecular mecha-nisms underlying the export process

Since Atm1p is specifically required for the assembly of cytosolic, but not mitochondrial Fe/S proteins, it is thought to play a central role

in the release of a moiety synthesized by the ISC assembly machinery from the organelles and may be required for the assembly of cytosolic Fe/S proteins Only a few components of the so-called ‘ISC export machinery’, other than Atm1p, have been identified so far, namely Erv1p and the two homologous proteins Bat1p and Bat2p Since these proteins appear to be functionally related to Atm1p, the findings that support their involvement in Fe/S protein mat-uration in the cytosol are briefly summarized as follows

Erv1p is a component of the intermembrane space and is essential for yeast viability (Lange

et al., 2001; Lisowsky, 1992) Inactivation of

Erv1p leads to a dramatic reduction in the assembly of cytosolic Fe/S proteins Similar to what is observed when Atm1p is depleted, mitochondrial Fe/S protein assembly is not affected in Erv1p-defective cells Erv1p was found to possess sulfhydryl oxidase activity associated with the C-terminal domain of the

protein (Lee et al., 2000) Currently, the role of

this domain in Fe/S protein assembly in the cytosol is unclear Nevertheless, the localization

of Erv1p in the intermembrane space suggests that it plays a role in the export pathway subse-quent to that in which Atm1p is implicated Whether Erv1p transiently binds directly to the

transported molecule, or introduces disulfide

bonds into a component of the pathway, remains

to be determined Interestingly, the mammalian

homologue of Erv1p, termed ALR (‘augmenter

of liver regeneration’), can functionally replace

the yeast protein and thus the two proteins

appear to be orthologues All of the components

of the ISC assembly machinery and Atm1p (see

below) are conserved in mammals, suggesting

that Fe/S cluster assembly follows similar

path-ways in virtually all eukaryotes

The BAT1 gene was isolated as a high-copy

suppressor of a temperature-sensitive mutant of

ATM1 (Kispal et al., 1996) BAT1 and the highly

homologous gene BAT2 encode the

mitochon-drial and cytosolic forms of branched-chain

amino acid transaminases, respectively (Eden

et al., 1996; Kispal et al., 1996) The Bat proteins

catalyze the reversible inter-conversion of

branched-chain ␣-keto acids and amino acids

(i.e leucine, isoleucine and valine) Additionally,

they perform a second function unrelated to

amino acid synthesis This is evident from the

growth defect of ⌬bat1 ⌬bat2 cells, lacking both

BAT genes, on rich media containing glucose,

which occurs even after additional

branched-chain amino acids are added to the medium

(Kispal et al., 1996) This observation may be

explained by the participation of the Bat

proteins in the maturation of cytosolic Fe/S

pro-teins (Prohl et al., 2000; C Prohl, unpublished).

The double mutant cells show a threefold

reduc-tion in the de novo synthesis of both Leu1p and

Rli1p Fe/S proteins in the cytosol Thus, the Bat

proteins are not essential for maturation of

cytosolic Fe/S proteins, but apparently perform

an accessory function, increasing the efficiency

of the formation of holoprotein in an, as yet,

unknown way Expression of either BAT gene is

sufficient for the normal formation of cytosolic

Fe/S proteins, indicating that the specific Bat

function can be performed either in

mitochon-dria, or in the cytosol Similar to Atm1p and

Erv1p, the Bat proteins are not required for the

biogenesis of Fe/S proteins within the

mito-chondria, suggesting that they participate in the

Atm1p-mediated export pathway One possible

function may be the catalytic formation of a

com-pound required for chelation of the Fe/S cluster

(or a related compound) during export from

the mitochondria

In summary, there is ample evidence for the involvement of Atm1p in the maturation of

cytosolic Fe/S proteins, yet the molecular

details underlying its precise function have not

been unraveled so far Future progress in

understanding the roles of Atm1p, Erv1p and the Bat proteins will require the identification of the substrate for Atm1p and of any additional components of the ISC export machinery

MDL1P ANDMDL2P, TWO HOMOLOGOUS

TRANSPORTERS WITH DIFFERENT FUNCTIONS

Recently, two additional ABC transporters, termed Mdl1p and Mdl2p, have been identified

in yeast mitochondria, and were found to be homologues of the human ABCB8 and ABCB10

genes (see below) (Young et al., 2001) Like

Atm1p, these proteins are half transporters with an N-terminal membrane-spanning domain

(Figure 25.1) (Dean et al., 1994) The two

pro-teins show rather high sequence homology (46%

identical amino acid residues) and have a molecular mass of 76 kDa (Mdl1p) and 91 kDa (Mdl2p), including a putative N-terminal exten-sion serving as a mitochondrial presequence

(Table 25.1) In fact, only the N-terminus of Mdl1p resembles a canonical mitochondrial tar-geting signal, whereas the N-terminal segment

of Mdl2p does not conform to the properties of

a presequence According to biochemical frac-tionation experiments using specific antibodies, both proteins are localized in the mitochondrial inner membrane with the ABC domains facing

the matrix space (Young et al., 2001) (Figure

25.1) Thus, all three yeast mitochondrial ABC transporters appear to exhibit the same mem-brane orientation and thus are presumed to export substrates from the matrix towards the cytosol

Deletion of the MDL1 and MDL2 genes does not cause major growth defects in S cerevisiae (Dean et al., 1994) However, whilst ⌬mdl1 cells

exhibit normal growth, growth of ⌬mdl2 cells is retarded on glycerol-containing media In part, this may be explained by the finding that

⌬mdl2 cells tend to gradually lose mitochon-drial DNA (J Gerber, unpublished) Double

deletion of both MDL genes slightly exacerbates

the growth defect observed for ⌬mdl2 cells, suggesting that the proteins may perform non-overlapping functions This is supported by recent insights into the function of Mdl1p

Both Mdl1p and Mdl2p are close homologues

of the yeast a-factor pheromone receptor Ste6p

and of another ABC protein, the mammalian TAP transporter (ABCB2/ABCB3) This protein mediates the transfer of antigenic peptides after

their generation by the cytosolic proteasome to

the class I major histocompatibility complex

(MHC class I) in the endoplasmic reticulum (ER)

(see Chapter 26) Hence, it was postulated that

the Mdl proteins may facilitate the export of

peptides from the matrix to the intermembrane

space A direct test of this idea showed that

mitochondria derived from ⌬mdl1 mutant cells

displayed a 40% reduction in peptide release

(Young et al., 2001) In the assay system used,

the peptides were generated by the inner

mem-brane protease Yta10p/Yta12p (also termed

Afg3p/Rca1p) from mitochondria-encoded

membrane proteins (Arlt et al., 1996; Rep et al.,

1996) (Figure 25.3) This member of the family of

ATP-dependent AAA proteases exposes its

pro-teolytic domain in the matrix space and forms a

large hetero-oligomeric complex (for a recent

review on AAA proteases, see Langer, 2000)

The rather small decrease in peptide export

observed after deletion of MDL1 is explained by

the fact that another set of peptides generated by

the inner membrane protease Yme1p can still

leave the organelle in the absence of Mdl1p This

second mitochondrial AAA protease forms a

homo-oligomer with its proteolytic domain in

the intermembrane space (Langer, 2000) (Figure

25.3) When a double mutant ⌬mdl1 ⌬yme1 was analyzed, a 75% reduction in peptide release from the organelle was observed The length of the released peptides varied between 6 and 20 residues, strikingly similar in size to peptides transported by the TAP transporter in the

ER (Elliott, 1997; Ritz and Seliger, 2001) (see Chapter 26) The function of Mdl1p in peptide export depended on a conserved motif in the Walker A and B sites of the nucleotide-binding domain and a loop characteristic for ATPases Peptide export through Mdl1p therefore seems

to require the hydrolysis of ATP (Young et al.,

2001) For final exit from the organellar

inter-membrane space, as illustrated in Figure 25.3,

the peptides possibly pass the outer membrane with the help of mitochondrial porin or the TOM complex, both of which contain large

pores (Figure 25.3) (Künkele et al., 1998).

On the other hand, deletion of MDL2 does

not result in any alteration of peptide export from the mitochondria, suggesting that only Mdl1p mediates the release of peptides from the mitochondrial matrix These findings are nicely corroborated by the observation that Mdl1p and Mdl2p appear to be associated with different high molecular mass complexes of

IMS

Ma

Mdl1p

TOM complex Porin

Peptides

Yta10/12p Yme1p

ATP

ATP ATP

MIM MOM

Figure 25.3 Model for the function of Mdl1p in the export of peptides from the mitochondrial matrix.

Peptides generated by the inner membrane protease, Yta10p/Yta12p, are exported by the ABC transporter, Mdl1p, in an ATP-dependent fashion Another pool of proteolytic fragments is formed by the inner membrane protease Yme1p in the intermembrane space Most likely, the peptides leave the mitochondria via porin or the TOM complex, both of which contain large pores Currently, it is unknown how peptides generated by the matrix protease Pim1p (not shown) are exported from the organelles MOM, mitochondrial outer membrane; IMS, intermembrane space; MIM, mitochondrial inner membrane; Ma, matrix.

200 kDa and 300 kDa, respectively (Young et al.,

2001) This observation may argue for

homo-dimer, rather than heterohomo-dimer, association of

the Mdl proteins

Since deletion of MDL1 is not associated with

any detectable phenotype, the question arises as

to what the physiological significance of peptide

transport by Mdl1p might be Moreover,

mito-chondria can further break down the longer

peptides to amino acids or di- and tripeptides,

which can be transported independently of

Mdl1p Thus, the purpose of Mdl1p-mediated

peptide transport remains unclear, even though

the latter observation supports the finding that

Mdl1p is dispensable in yeast In vertebrates,

proteins homologous to Mdl1p (see below)

might play an important role in the transport of

antigenic peptides derived from mitochondrial

proteins for presentation on the eukaryotic cell

surface Functional complementation studies

with the mammalian homologues expressed in

the ⌬mdl1 background have not yet been

con-ducted to test this attractive hypothesis

The sequencing of the human genome has

provided us with a complete inventory of

ABC transporters in man (Klein et al., 1999;

http://www.humanabc.org) Amongst the 48

proteins with ABC domains, a few qualify as

potential mitochondrial components based on

the presence of a (putative) presequence at

their N-termini Two of these proteins, termed

ABC7 (ABCB7, according to the nomenclature

of http://www.humanabc.org) and MTABC3

(ABCB6) are homologous to the yeast Atm1p,

whereas another two proteins, namely M-ABC1

(ABCB8) and M-ABC2 (ABCB10), closely

resem-ble yeast Mdl1p and Mdl2p In mice, another

homologue of the latter subclass has been

identified and analyzed recently, the protein

ABC-me The properties and functions of these

proteins are discussed in the following sections

ABC7 ANDMTABC3, FUNCTIONAL

ORTHOLOGUES OF YEASTATM1P

The human ABC transporter ABC7 represents

the closest homologue of yeast Atm1p The

cDNA corresponding to the gene has been identified independently by several groups

(Allikmets et al., 1999; Csere et al., 1998; Mao

et al., 1998; Shimada et al., 1998) Sequencing of the entire ABC7 gene revealed 16 introns and the promoter structure (Bekri et al., 2000) At the

protein level ABC7 shares 47% amino acid

sequence identity with yeast Atm1p (Table 25.1).

Expression of the human gene in yeast has demonstrated that the human protein is the

functional orthologue of Atm1p (Allikmets

et al., 1999; Csere et al., 1998) This gene is able

to revert growth of ⌬atm1 mutant cells to almost wild-type rates and to restore normal cyto-chrome levels Furthermore, mitochondria harboring ABC7 instead of Atm1p do not accumulate iron All these observations strongly indicate that upon expression in yeast, ABC7 can replace the primary function of Atm1p in

Fe/S cluster formation (Bekri et al., 2000) These

findings further suggest that ABC7 performs a similar or identical function in the mammalian cell as that carried out by Atm1p in yeast

Mutations in the human ABC7 gene cause

X-linked sideroblastic anemia and cerebellar

ataxia (XLSA/A) (Allikmets et al., 1999; Bekri

et al., 2000) As a result of such mutations,

mito-chondria accumulate high concentrations of iron and form so-called ring sideroblasts (i.e iron-loaded ring-shaped tubules which are concen-trated around the nucleus) Thus, there exists a striking similarity in phenotypes between yeast and man upon impairment of Atm1p and ABC7 function, respectively Biochemical studies indi-cate that yeast serves as an excellent model sys-tem to study the effects of the mutations in

ABC7 When expressed in yeast, mutant ABC7

proteins, or Atm1p bearing the corresponding mutations, are functionally impaired (Allikmets

et al., 1999; Bekri et al., 2000) For instance, when

the ABC7-(E433K) mutants (mutation localized towards the matrix following TM6), or the

cor-responding ATM1-(D398K) mutants, were

expressed in ⌬atm1 yeast cells, maturation of cytosolic Fe/S proteins was twofold lower as

compared to wild-type cells (Bekri et al., 2000).

The surprisingly weak consequences of these charge exchange mutations underlines the importance of ABC7 function for a healthy cell

In fact, only slight changes to ABC7 can dramat-ically affect cellular iron homeostasis and elicit severe phenotypical consequences These obser-vations are consistent with the fact that, in yeast, Fe/S cluster formation is an indispensable

process (Lill et al., 1999) Deletion of many genes

encoding components of the ISC assembly

machinery is lethal, indicating a central role for

Fe/S proteins for life

The human protein, termed MTABC3 (Taglicht and Michaelis, 1998), represents a

second functional orthologue of yeast Atm1p

(Mitsuhashi et al., 2000) The MTABC3 gene is

encoded by human chromosome 2 (Table 25.1)

and has been mapped to within the vicinity of

the locus for lethal neonatal metabolic

syn-drome, a disorder of mitochondrial function

associated with iron metabolism Hence,

MTABC3 is a likely candidate gene for this

dis-order The homology of MTABC3 and Atm1p is

less than that of ABC7 compared with Atm1p

(38% as compared to 47% identical amino

acid residues) Nevertheless, expression of

MTABC3 in ⌬atm1 yeast cells restores growth

to wild-type levels, reverts the increase in

chondrial iron, and prevents the loss of

mito-chondrial DNA Even though the role of

MTABC3 in the biogenesis of cytosolic Fe/S

proteins has yet to be analyzed, such a function

seems likely

The relationship of the two human ortho-logues of yeast Atm1p is unclear Based on their

common role in iron homeostasis it is

con-ceivable that ABC7 and MTABC3 form a

het-erodimer in the human cell An alternative

hypothesis predicts that both genes may be

dif-ferentially expressed in human tissues The

presence of two copies of Atm1p-like proteins

may offer the possibility to fine-tune the

func-tion of the ABC transporter, as found for

numer-ous other mammalian proteins

M-ABC1 ANDM-ABC2, MAMMALIAN

HOMOLOGUES OF YEASTMDL1P AND

MDL2P

The human genome harbors four candidates

with homology to the yeast MDL genes Only

two of the encoded proteins, termed M-ABC1

and M-ABC2 (ABCB8 and ABCB10 according to

nomenclature of http://www.humanabc.org),

have been experimentally localized to

mito-chondria (Hogue et al., 1999; Zhang et al.,

2000a) Another gene product, ABCB9, has been

found in the lysosomal compartment (Zhang

et al., 2000b) Nevertheless, the protein is not a

close homologue of the vacuolar ABC

trans-porters, Ycf1p of S cerevisiae (Li et al., 1996) or

Hmt1p of Schizosaccharomyces pombe (Ortiz et al.,

1995) The fourth mammalian Mdl homologue,

ABCB5, has not yet been studied The sequence

identity between the human M-ABC1/M-ABC2

and the yeast Mdl proteins varies between 32%

and 42% (Table 25.1) Based on sequence

com-parisons, M-ABC1 may be the counterpart of Mdl2p, while M-ABC2 is more closely related to Mdl1p However, the differences in homology may be too small to infer a close functional rela-tionship

Currently, it is not known whether M-ABC1 and M-ABC2 form homo- or heterodimers in the mitochondrial inner membrane Similarly, the membrane orientation of these proteins is not yet clear, even though it is likely that it is the same as for Mdl1p and Mdl2p, with the nucleotide-binding domain facing the matrix

space (however, see Zhang et al., 2000a) No

experimental evidence has been obtained for any function of M-ABC1 and M-ABC2 in the transport of peptides out of the mitochondrial matrix, even though such a role, similar to that

of Mdl1p, seems probable (see above)

ABC-ME, A MURINE MITOCHONDRIAL

ABC TRANSPORTER WITH A FUNCTION IN HEME METABOLISM

Only one mitochondrial ABC transporter has been identified to date in mice, the protein

ABC-me (mitochondrial erythroid), and its

cellu-lar role has been investigated in some detail

(Shirihai et al., 2000) The ABC-me gene has

been isolated as one factor that is induced upon expression of the erythropoietic transcription

factor GATA-1 ABC-me is highly expressed in

erythroid tissues of embryos and adults In murine erythroleukemia (MEL) cells,

overex-pression of ABC-me strongly increased the heme concentration Conversely, ABC-me mRNA

levels are decreased by physiological concentra-tions of heme Together, these findings are con-sistent with a role for ABC-me in the trafficking

of intermediates of heme biosynthesis The heme biosynthetic steps are partitioned between the mitochondrial matrix and the cytosol, with the first reaction and the last three steps taking place in the matrix The ABC domains of ABC-me face the mitochondrial matrix and this has been taken to indicate that the protein

should function as an exporter (Shirihai et al.,

2000) ABC-me could be involved in translocat-ing either ␦-aminolevulinate or heme from the mitochondrial matrix to the cytosol The rather specific expression of ABC-me in erythroid cells may be necessary to satisfy the extraordinarily high needs for transporting heme biosynthetic metabolites across the mitochondrial inner