CHAPTER 23 – ROLE OF ABCA1 IN CELL TURNOVER AND LIPID HOMEOSTASIS

CHAPTER 23 – ROLE OF ABCA1 IN CELL TURNOVER AND LIPID HOMEOSTASIS CHAPTER 23 – ROLE OF ABCA1 IN CELL TURNOVER AND LIPID HOMEOSTASIS CHAPTER 23 – ROLE OF ABCA1 IN CELL TURNOVER AND LIPID HOMEOSTASIS CHAPTER 23 – ROLE OF ABCA1 IN CELL TURNOVER AND LIPID HOMEOSTASIS CHAPTER 23 – ROLE OF ABCA1 IN CELL TURNOVER AND LIPID HOMEOSTASIS CHAPTER 23 – ROLE OF ABCA1 IN CELL TURNOVER AND LIPID HOMEOSTASIS CHAPTER 23 – ROLE OF ABCA1 IN CELL TURNOVER AND LIPID HOMEOSTASIS

S TRUCTURAL T RAITS OF

THE GENE AND ITS EXPRESSION

The discovery of ABC1 in 1994 stemmed from

an effort to identify novel ATP-binding cassette

(ABC) transporters in mouse macrophages,

based on the selective amplification of

consen-sus motifs in the nucleotide-binding domain

(NBD) (Luciani et al., 1994; Savary et al., 1996,

1997) Among the new genes discovered, it

soon became evident that ABC1 bore structural

features distinct from those of known

trans-porters and that it was not an isolated example

in the mammalian genome (Broccardo et al.,

1999; Dean et al., 2001) The group of

trans-porters most closely related to ABC1 has been

recently renamed the A subclass (ABC1

becom-ing ABCA1) and, to date, includes 12

trans-porters (see Chapter 3) All the ABCA genes

encode complete transporters with four

domains organized in the following fashion:

TMD1/NBD1/TMD2/NBD2 ABCA genes are

highly conserved in mammals and are present

in Drosophila melanogaster and in the nematode

Caenorhabditis elegans, but are absent from yeast

(Dean et al., 2001; Decottignies and Goffeau,

1997) In contrast to mammals, insects and

nematodes, most of the ABCA genes expressed

in Arabidopsis thaliana appear to encode

trans-porters with only two domains: TMD/NBD, so-called hemi- or half transporters

(Sanchez-Fernandez et al., 2001).

Of the 12 members of the ABCA subclass, ABCA1, ABCA2, ABCA3, ABCA4 and ABCA7 have been well characterized and identify a closely related group Two additional genes, ABCA12 and ABCA13, (M Dean, personal com-munication) also belong to this cluster but have only been partially characterized so far

The remaining five ABCA genes are clustered

on chromosome 17 in humans and, on the basis

of sequence alignments, define a subgroup dis-tinct from that defined by ABCA1, 2, 3, 4 and 7

The ABCA1– 4 and ABCA7 genes probably originated by duplications before speciation,

as suggested by their localization on different chromosomes and their mapping in syntenic regions in the human and mouse genome

(Figure 23.1) In spite of this remote evolution-ary origin, they retain a very similar exon–

intron structure exemplified by that of ABCA1

(Figure 23.2) (Allikmets et al., 1998; Azarian

et al., 1998; Broccardo et al., 2001; Kaminski

et al., 2000, 2001; Remaley et al., 1999;

Santamarina-Fojo et al., 2000; Vulevic et al.,

2001) In fact, the most divergent characteristic concerns the shrinking or expansion of inter-vening sequences This latter feature leads to genomic loci spanning more than 100 kb for ABCA1 and ABCA4 whereas ABCA2, ABCA3

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

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

23

R OLE OF ABCA1 IN C ELL

T URNOVER AND L IPID

H OMEOSTASIS

CHAPTER

q13

16p13.3

17 B

2 A-B 12.6 cM

4A5 -B3 23.1 cM

9q31 9q34

p23 p13 q12

q22

q34

p13 p12 p11 q11

q22 q23 q24

3 61.8 cM

44 cM

19p13.3

q13.4 q13.1

p12 q12 p13.1

p13.3 p32

p31 p13 q12 q25 q32 q44

p33

0 Mos B4galt1 Mup1 Ambp Tyrp1 Ifna Aldh5a1 Nppa Fv1 Gnb1

Cga 10

20 30 40 50 60 70 80 90

Car1 Fgf2 G1rb Fgg Tshb Amy1 Egf Adh3 Ptger3

10 20 30 40 50 60 70 80 90 100

10 10

0 Plg H2 Upg1 C3 Lama1 Lhcgr Ce2

Abl1 Neb Gcg Hoxd Cas1 Fmn Il1 Src a Gnas

20 20

30 30

40 40

50 50

60 60

70 70

80 80

90

10

0 Utrn Pcmt1 Myb Zfa Pfp Igf1 Kit1 Ifng Prim1

20 30 40 50 60 70 80

90 110 100 120

Figure 23.1 Schematic diagram of chromosomal mapping in the mouse or human genome of ABCA1, ABCA2, ABCA3, ABCA4 and ABCA7 Centimorgans (cM) are shown on mouse chromosomes whereas cytogenetic banding is represented on the ideograms of human chromosomes.

5 kb

1

abca1

10 kb

ABCA1

A

B

SNAP 1 and 2

Figure 23.2 A, Schematics of the 200 kb spanning the ABCA1 locus on mouse chromosome 4 as reported by Qiu et al., 2001 The exon–intron structure of the ABCA1 gene is shown in part B Color-coded boxes identify individual exons and the encoded protein domains: green, transmembrane segments; rose and light blue, N- and C-terminal extracellular loops; red, nucleotide-binding domains; orange, intervening domain with putative regulatory function The position of the starting methionine (ATG) and of the stop codon (*) are shown Figures on top indicate exon numbering ABCA2, ABCA4 and ABCA7 show a largely superposable gene structure.

(C Broccardo, unpublished results) and

ABCA7 loci span a genomic region of 20–30 kb

Most of these five ABCA genes have been cloned from different species (mouse, man, rat

and cow) and a close homologue to ABCA1 has

been identified in the chicken (Schreyer et al.,

1994) In all species, the ABCA1 coding region is

spread over 50 exons and the translational start

site is in exon 2 (Qiu et al., 2001;

Santamarina-Fojo et al., 2000) The relative relationship

between exons and protein domains is

schema-tized in Figure 23.2B and is closely conserved

in ABCA1, ABCA2, ABCA4 and ABCA7 The

exon–intron structure of ABCA3 has not

been reported so far but is likely to be similar A

recent survey of the genomic region spanning

the ABCA1 locus both in mouse and in man

failed to identify genes in the regions upstream

of ABCA1 (87 kb in the mouse and 34 kb in

man) but detected one gene in man and two

genes in the mouse, in close proximity to the

ABCA1 polyadenylation site, that are encoded

in the opposite orientation to ABCA1 itself

(Qiu et al., 2001) These were named hSNAP,

mSNAP1 and mSNAP2 hSNAP and mSNAP1

are located 9 kb and 8 kb downstream of ABCA1,

respectively In mouse, SNAP2 is located 12 kb

downstream of SNAP1

The transcriptional regulation of ABCA1 appears to be exceptionally complex, and at present, poorly understood Three clusters of transcriptional start sites for ABCA1 have been

identified The first type (class 1 in Figure 23.3),

identified in placenta, is 40 bp downstream

from a modified TATA box (Pullinger et al., 2000; Schwartz et al., 2000) Six G/C-rich

sequences, potential binding sites for Sp1 and/or SREBP, as well as AP1 and NFkB sites were identified in the same region The second

start site (class 2, Figure 23.3) is approximately

90 bp downstream of the start sites for class 1

transcripts (Santamarina-Fojo et al., 2000) A

weak TATA box is present at 32 bp 5⬘ of this start site A LXR/RXR site is present between⫺70 and⫺55 bp of this start site (that is, at ⫹19

to⫹44 relative to start site 1) (Costet et al., 2000).

Transcripts with the second start site were reported to predominate in HepG2 and THP-1 cells, and transformed human lines originating from liver and monocytic cells respectively The

third group of transcripts (class 3, Figure 23.3)

are initiated within intron 1 of the full-length gene This leads to formation of a novel first exon (exon 1a) with the loss of 28 amino acids

from ABCA1 (Cavelier et al., 2001; Singaraja

et al., 2001) This group of transcripts is initiated

Sp1, SREBP

LXR/

RXR

Exon 1,

221 bp

Exon 2

147 bp

Exon 1

303 bp

Class 1

Class 2

Sp1, SREBP

LXR/

RXR

Intron 1, 24 156 bp

Intron 1, 24 156 bp

Exon 2

147 bp

Exon 1a

136 bp

Class 3

Sp1, SREBP

LXR/

RXR

Intron 1,

2210 bp

Exon 2

147 bp

Figure 23.3 Alternative start sites for ABCA1 gene transcription Transcript classes 1–3 are defined in the

text In different analyses of the same transcript class, slight differences in length have been reported

The data shown correspond to those in the first report of each class Sp1, SREBP, LXR/RXR, consensus

binding sites for these transcription factors.

downstream of classical TATA and CAAT

sequences and a variety of potential

lipid-dependent binding sites for transcription

fac-tors Class 3 transcripts predominate in liver

tissue in mice expressing a human ABCA1 gene

construct lacking wild-type exon 1 According

to classical concepts, the basic transcription

machinery, assembled at the start site, forms an

activated complex with DNA-binding proteins

generally within 300– 400 bp upstream On this

basis, transcript classes 1–3 could each respond

to a different set of regulatory proteins The

functional effect of a given inducer (e.g

oxy-sterol, free cholesterol – FC) on the overall

ABCA1 mRNA levels would thus also depend

on the tissue-specific proportions of each

tran-script present under baseline conditions

Finally, alternative ABCA1 transcripts lack-ing part of exon 3 and all of exon 4 were

recen-tly detected in human fibroblasts, endothelial

and smooth muscle cells, and HepG2 cells

(Bellincampi et al., 2001) This variant does not

affect the promoter sequence As a result, and

predictably, induction of ABCA1 mRNA with

FC did not change the proportion of full-length

and shorter transcripts

The best-studied regulatory element control-ling ABCA1 expression is the class 2 start site

LRR/RXR (Costet et al., 2000) This is controlled

by the transcription factor PPAR gamma/delta

(Chawla et al., 2001; Oliver et al., 2001;

Venkateswaran et al., 2000) Oxysterols and

retinoic acid strongly upregulated the

expres-sion of luciferase constructs linked to such

ABCA1 type 2 promoter constructs The in vivo

relevance of the site is indicated by

upregula-tion of ABCA1-mediated lipid efflux by a

PPAR-delta agonist in monkeys (Oliver et al., 2001).

An FC-sensitive promoter region was also

iden-tified 100–200 bp upstream of the class 1 start

site (Santamarina-Fojo et al., 2000) This may be

functional for the production of class 2

tran-scripts but probably not for those in class 3

Finally, cAMP-dependent expression has been

described in transformed rodent

monocyte-derived cell lines (RAW264 and J774 cells)

but not in human-derived THP-1, CaCo-2

or HepG2 cells, or normal skin fibroblasts

(Bortnick et al., 2000) The target for

cAMP-mediated upregulation in responsive cells has

not been identified, nor the significance of this

activity established

The expression pattern of ABCA1 has been

extensively studied Early studies by in situ

RNA hybridization revealed a tight

spatio-temporal correlation between the expression of

the ABCA1 transcript and the occurrence of cell death during embryo development (Luciani and Chimini, 1996) This was further inter-preted as due to the local recruitment of macro-phages responsible for clearing the corpses of the cells committed to die The exclusive expression of ABCA1 by phagocytes in the areas of developmental cell death has recently been formally proven by the undetectability of ABCA1 transcript in these areas in PU1 null

embryos (Wood et al., 2000) These embryos,

owing to the lack of this transcription factor crucial for the differentiation of hematopoietic cell lineages, are in fact virtually devoid of

macrophages (Wood et al., 2000) The

expres-sion of ABCA1 in cells of myeloid lineage is unequivocal It has indeed been assessed in several cell lines and in primary cellular sys-tems in mouse, such as those of resident or elicited peritoneal and bone marrow derived macrophages and in humans in activated monocytes, macrophages and foam cells

(Christiansen-Weber et al., 2000; Langmann

et al., 1999; Lawn et al., 2001; Luciani et al.,

1994) Dendritic cells, whilst sharing a common precursor with monocytes in myeloid lineages, lack ABCA1 transcripts and instead express ABCA7 (C Broccardo, unpublished) ABCA1 expression by tissue macrophages can account for the detection of low/medium levels of ABCA1 transcripts in many adult tissues In addition, however, some parenchymal cells, such as liver and adrenal cells, do also express significant levels of ABCA1 in the mouse

(Luciani et al., 1994, and unpublished

observa-tions) Northern blot analysis of human tissues indicated that kidney, lung and spleen were among the major sites of ABCA1 expression

(Langmann et al., 1999) A similar tissue distri-bution in baboon was observed by in situ hybridization studies (Lawn et al., 2001) This

study also reported that although normal veins and arteries did not express ABCA1 mRNA, this was upregulated in the setting of atherosclero-sis, where widespread expression was found in macrophages within atherosclerotic lesions In contrast, the expression of ABCA1 by intestinal epithelium cells has been repeatedly suggested

but not yet formally demonstrated (Lawn et al.,

2001) A massive upregulation of ABCA1 tran-scription has been demonstrated in the uterus and the developing placenta, where the transcript has been detected in both the lab-yrinthine and decidual layers

(Christiansen-Weber et al., 2000; Hamon et al., 2000; Luciani

et al., 1994).

It is important to note that the five ABCA1-like genes show non-overlapping territories of

preferred expression This may indicate that

they exert similar functions in diverse cell

specific contexts (Broccardo et al., 1999) ABCA1,

ABCA2, ABCA3 and ABCA7 are expressed

during embryonic development as witnessed

by the detection of specific transcripts in whole

embryo RNA However, no detailed

morpho-logical assessment of the developmental

expression pattern of ABCA 2, 3 or 7 has been

reported as yet

THE PROTEIN: TOPOLOGY, ATPASE

ACTIVITY AND SUBCELLULAR

LOCALIZATION

The protein encoded by the ABCA1 gene is 2261

amino acids long in both mouse and man This

corresponds to a product 60 amino acids longer

than that of the one originally described (Costet

et al., 2000; Luciani et al., 1994; Pullinger et al.,

2000; Tanaka et al., 2001) Both the

methionines⫹1 and ⫹61 are able to support the production of a protein but only translation from the first methionine produces an active

protein (Fitzgerald et al., 2001; Wang et al., 2000).

The shorter product is retained in the endoplas-mic reticulum (Y Hamon, unpublished result), possibly as a result of improper folding

As is frequently the case for ABC transporters and more generally for other polytopic mem-brane proteins, determination of a precise topo-logical organization is extremely difficult In spite of the unambiguous identification of large blocks of hydrophobic residues, their precise distribution into the succession of individual transmembrane segments is rather uncertain on the basis of computer predictions alone

Experimental topological analyses are under-way in many laboratories Based on both direct evidence and on the analogy with ABCA4

(Bungert et al., 2001; Fitzgerald et al., 2002),

it seems reasonable, at present, to favor the

topological model shown in Figure 23.4(see also Chapter 2) This predicts two large extracellular

Mouse Human Mouse Human

Mouse Human Mouse Human

Mouse Human ABCA1 286 286 ABCA2 318 317 ABCA3 158 ABCA4 281 281 ABCA7 282 271

286 286 ABCA2 318 317 ABCA3 158 ABCA4 281 281 ABCA7 282 271

Mouse Human ABCA1 594 594 ABCA2 666 668 ABCA3 220 ABCA4 610 603 ABCA7 507 510

Mouse Human ABCA1 385 385 ABCA2 419 421 ABCA3 382 ABCA4 410 371 ABCA7 394 395

385 385 ABCA2 419 421 ABCA3 382 ABCA4 410 371 ABCA7 394 395

NBD2

Mouse Human ABCA1 508 508 ABCA2 548 548 ABCA3 458 ABCA4 521 524 ABCA7 502 496

ABCA1 508 508 ABCA2 548 548 ABCA3 458 ABCA4 521 524 ABCA7 502 496

NBD1

Extracellular loop 2 Extracellular loop 1

Figure 23.4 Model of ABCA1 membrane topology The type I or type II orientation of the N-terminus is

still controversial (see text) The length of the major extracellular loops (green) and of NBDs (yellow) in

five distinct ABCA transporters is shown for comparison.

loops between TM1 (amino acids 25–45) and 2

(starting at amino acid 640 in ABCA1) and TMS7

(amino acids 1350–1370) and 8 (starting at

amino acid 1668) A similar topology is

proba-bly shared by the other ABCA members, whose

hydrophobicity plots are largely

superimpos-able on that of ABCA1 The length of the

pre-dicted extracellular loops, however, varies

greatly among the individual transporters

ABCA2 and ABCA3 are the most divergent and

show respectively the longest and shortest

extracellular loops (Figure 23.4) Other aspects

of the topological model remain ambiguous

Thus, contradictory results have been reported

as to whether the first hydrophobic segment

(amino acids 24–48) serves as a signal peptide,

or as a signal anchor sequence In the former

situation, processing of the peptide would

lead to an externally exposed N-terminus,

commencing at position 49 (potential

cleav-age site by the algorithm SignalP V1.1 World

Wide Web Prediction Server (Nielsen et al.,

1997)) and to an asymmetrical number of

trans-membrane segments in the two halves of the

transporter Alternatively, if this segment acts

as a signal anchor, a type II orientation of the

short free N-terminus will result, and the two

halves of the transporter will show a

symmetri-cal architecture, each with six predicted

trans-membrane segments The first option is favored

by Ueda and co-workers and is based on the

inability to detect an HA epitope fused to the

N-terminus of ABCA1 by Western blotting

(Tanaka et al., 2001) In the hands of other

investigators (Fitzgerald et al., 2001), the

analy-sis of a similar EGFP/ABCA1 chimera

sug-gested, in contrast, its function as signal anchor

Moreover, Fitzgerald et al not only detected

EGFP in the final product, but also reported the

inability of the ABCA1 ‘signal peptide’ to

sup-port the secretion of rhodopsine fused to its

C-terminus Taking into account also the

for-mal biochemical evidence that in ABCA4 the

putative signal peptide is not processed (Illing

et al., 1997), we favor the hypothesis of a type

II orientation Constrained folding within the

hydrophobic membrane environment of the

very short N-terminus and/or the technical

inability to completely denature the transporter

may account for the lack of detection of the HA

epitope reported by Ueda and co-workers

(Tanaka et al., 2001) In line with that, we have

observed an inability to detect an HA epitope

inserted into the short loops separating TMSs,

that is in positions where a tight interaction

with the membrane bilayer is likely (Rigot et al.,

in preparation) In the case of ABCA4, disulfide bridging between the large extracellular loops

has been reported (Bungert et al., 2001) Similar

molecular interactions may exist in the case of ABCA1, on the basis of the observed conserva-tion of cysteine residues in the extracellular loops of both proteins

Ueda and co-workers (Tanaka et al., 2001)

showed that ABCA1 is able to bind and hydro-lyze ATP, although with low efficiency This is

in line with the presence of the two conserved Walker motifs in the NBDs and is also consis-tent with the known ATPase activity of ABCA4 (Ahn and Molday, 2000; Biswas and Biswas,

2000; Sun et al., 1999).

ABCA1 has been shown to reach the plasma membrane in a variety of transfected cell lines

(Hamon et al., 2000; Neufeld et al., 2001; Wang

et al., 2000) (Figure 23.5) The staining at the

membrane is not homogeneous but rather punctate The localization of ABCA1 in plasma membrane domains enriched in cholesterol and sphingolipids (lipid rafts or caveolae) is suggested by the partial coalescence with the membrane staining of CD14, the GPI-linked lipopolysaccharide (LPS) receptor and with GM1 distribution (O Chambenoit,

unpub-lished) It has been reported (Drobnik et al., 2002; Mendez et al., 2000), however, that in

endosomes

PM Golgi

endosomes

PM

Figure 23.5 Subcellular distribution of a transfected ABCA1/EGFP chimera as determined

by confocal microscopy The Golgi and endo-lysosomal compartments were identified

by costaining with known markers (Hamon et al., 2000) The discrete staining at the plasma membrane is clearly visible.

standard purification procedures ABCA1 does

not partition with Triton X-100 insoluble

mem-brane domains It has to be noted, however,

that the cell model chosen for this study

(immortalized skin fibroblasts) expresses few

rafts or caveolae and that also the isolation

pro-cedure used was relatively nonspecific Thus,

it seems possible that ABCA1 is preferentially

localized at the periphery of these domains,

from which it dissociates in the presence of

detergents, under standard purification

proce-dures The transfected ABCA1 protein has also

been localized to intracellular vesicles

belong-ing to the endo-lysosomal compartment and in

the Golgi stack (Hamon et al., 2000; Neufeld

et al., 2001; Wang et al., 2000) An ABCA1-specific

staining in the latter compartment is detected

also in untransfected macrophages At present

it is not known whether ABCA1 is functional

as a transporter in these intracellular

compart-ments An interesting feature associated with

the stable expression of ABCA1 in transfected

cells is a significant delay in cell growth

(Y Hamon, unpublished) This may be related

to an ability of the transporter to modulate the

composition and dynamics of the membrane,

which in turn may alter growth parameters It

is interesting to note that the forced and stable

expression of CED-7, the C elegans homologue

of ABCA1, has a similar impact on cell growth

(Y Hamon unpublished)

ABCA1 AND CELL TURNOVER:

THE CLEARANCE OF CELLS DYING BY

APOPTOSIS

Apoptosis or programmed cell death is a

genet-ically controlled and highly regulated event

responsible for cell turnover in healthy adult

tissues and of focal elimination of cells during

embryonic development (Kerr et al., 1972) The

apoptotic process itself consists of the

system-atic dismantling of the cell factory orchestrated

by the activation of caspases From a

morpho-logical standpoint, apoptosis can be easily

dis-tinguished from other forms of cell elimination

The structural changes during apoptosis take

place in two distinct steps The first involves

the generation of cell fragments still with an intact cell membrane (apoptotic bodies) This is then swiftly followed by their uptake and degradation by phagocytes, most frequently macrophages, recruited locally in large num-bers by mechanisms yet unknown Typically, a cell committed to die, rounds up and detaches from its neighbors, then it undergoes nuclear condensation and shrinkage of the cytoplasm without major morphological alteration of intracellular organelles Membrane blebs are now formed, which progressively lead to the generation of membrane-bound, compact but otherwise well-preserved cell remnants, the apoptotic bodies

Many of the morphological aspects of the apoptotic process have now been linked to pre-cise biochemical events and depend on the pro-teolytic cleavage of one or more of the molecules targeted by the effector caspases (Hengartner, 2000; Leverrier and Ridley, 2001a) A similar orchestration regulates the clearance of corpses

by phagocytes (Savill and Fadok, 2000) In physiological situations virtually no free dying cells are detectable in the body Indeed the persistence of self cells undergoing progressive disintegration is to be avoided at any cost for two main reasons Their slow removal would

be immediately harmful as a consequence of the leakage of noxious intracellular contents

It would also be expected to be dangerous in the longer term in view of the ability of cell fragments or their contents to trigger immune responses against self antigens persistently exposed to antigen-presenting cells The mole-cular circuits controlling recognition and inges-tion of corpses by phagocytes are far from being elucidated Most of the available clues come

from the model system C elegans There, genetic

dissection has highlighted the 14 genes control-ling the process of programmed cell death from the initial commitment to the final degradation inside the phagocyte Those are designated as

ced for the cell death abnormal phenotypes deriving from their mutation (Ellis et al., 1991a,

1991b) As far as the engulfment phase is con-cerned at least two genetic pathways exist in the worm that are conserved in mammals The ced-2, ced-5, ced-10 and ced-12 group of genes controls the first clearance pathway, which

cor-responds, as shown in Figure 23.6, to an

integrin-triggered signaling cascade in mammalian

phagocytes (Albert et al., 2000; Reddien and Horvitz, 2000; Tosello-Trampont et al., 2001; Wu and Horvitz, 1998a; Wu et al., 2001; Zhou et al.,

2001b) The second pathway is less defined but

includes the ABC transporter CED-7 (Wu and

Horvitz, 1998b), which bears a high sequence

similarity to ABCA1 and belongs to the

nema-tode ABCA class of transporters CED-7 works

in concert with a membrane scavenger receptor,

CED-1 (Zhou et al., 2001a), and a downstream

signaling protein, ced-6, which is also conserved

in mouse and man (Liu and Hengartner, 1998,

1999; Su et al., 2000) Apart from their clustering

in the same epistatic group, a cascade of

molec-ular interactions has not been determined so far

In the mammalian system, where

investiga-tions have mainly relied on in vitro assays, a

number of well-known phagocytic receptors

are involved in the uptake of apoptotic bodies

(Gregory, 2000; Platt et al., 1998; Ren and Savill,

1998; Savill and Fadok, 2000) Among these are

the LPS receptor, CD14, members of the family

of scavenger receptors (CD36, SRA) and of

inte-grins (␣v␤3 and ␣v␤5) Unfortunately, neither

the molecular entities any of these receptors recognize on the surface of the apoptotic prey nor what molecular modifications occur on the surface of the prey to be during apoptosis are known These are globally indicated as ACAMP (apoptotic cell-associated molecular patterns) to underline the lack of molecular data Recently two new molecules on the phagocyte surface able to engage prey ingestion have been identi-fied: a specific receptor for phosphatidylserine (PSR) and the tyrosine kinase receptor MER Both are expected to participate in the recogni-tion of the unusual amounts of phosphatidylser-ine exposed on the dying cells, at present the only available hallmark of membrane modifi-cations during apoptosis The PSR acting alone

or in concert with CD36 provides the required stereospecificity for phosphatidylserine (PS) recognition, whereas the second could act as

a receptor for gas-6 (the product of growth arrested specific gene 6), a soluble protein pre-viously implicated as a mediator of macro-phage binding to PS

A thorough overview of these receptors is beyond the scope of this chapter and is pro-vided elsewhere (Ren and Savill, 1998; Savill, 1998) It is, however, worth underlining the redundancy of surface molecules implicated in prey recognition by the macrophages; how-ever, none of them seem to be used exclusively for the engulfment This underscores the high physiological impact of the phenomenon, whose major goal is to avoid any escape of apoptotic prey from their fate

THE ROLE OFABCA1 DURING THE ENGULFMENT OF APOPTOTIC CORPSES

In mammals, the participation of an ABC trans-porter in engulfment was established in the mid-1990s by the description of an upregulation of ABCA1 transcripts in the macrophages recruited

to areas of developmental cell death (Luciani and Chimini, 1996) The functional meaning of this

upregulation was suggested by in vitro results

where an antibody-mediated block of ABCA1 function led to a reduced phagocytic perform-ance of peritoneal macrophages exclusively when the prey consisted of an apoptotic cell The subsequent molecular identification of CED-7

in the worm as an ABC transporter (Wu and Horvitz, 1998b) reinforced, by analogy, the hypothesis of an active role for ABCA1 during clearance of apoptotic cells by professional macrophages

Tyr

p130 cas

Dock 180

Crk II

rac

PI 3K

Tyr -K

avb3

?

?

SR PSr

ABC1 hced6

avb5

Elmo

Tyr-K

p130 cas

Dock 180

Crk II

rac

PI 3K

Tyr -K

avb3

?

?

SR PSr

ABC1 hced6

avb5

Elmo

Figure 23.6 Two parallel pathways are responsible

for the recognition and uptake of corpses by

phagocytes This scheme combines data from both

the nematode and mammalian systems In the left

part the pathway involving the ABC transporter

(ABC1 or CED-7), a scavenger receptor (SR⫽⫽ CD36

or CED-1) and the adaptor ced-6 is represented

Note that so far a direct molecular interaction

between the partners has not been demonstrated.

In the right section, the integrin-mediated

triggering of the signaling cascade involving

p130cas, CrkII, Dock 180, Elmo and rac1 is

represented In C elegans the triggering receptor

has not been identified but the membrane

recruitment and molecular interaction of CED-2,

CED-5, CED10 and CED-12 was demonstrated.

Membrane recruitment of tyrosine kinases of

unknown identity has also been reported (Albert

et al., 2000; Leverrier and Ridley, 2001b; Reddien

and Horvitz, 2000; Zhou et al., 2001a).

The development and combined analysis of

an in vivo loss of function and an in vitro gain of

function model (Hamon et al., 2000) allowed

us to establish unambiguously that ABCA1 is

able to promote the engulfment function of

macrophages both during embryonic

develop-ment and in adult life (Figure 23.7) However,

this is likely to be a consequence of the ability of

ABCA1 to influence the distribution of lipids on

both the transversal and lateral dimension of the

membrane (Figure 23.8; see also Figure 23.9).

Indeed the loss or gain of ABCA1 function has been directly correlated with a reduction or

an increase, respectively, in the outward flip of

PS from the inner leaflet of the plasma

mem-brane This is not an unusual activity among

ABC transporters (Higgins, 1994; Higgins and

Gottesman, 1992) It is, however, important to

stress that, in spite of clear evidence for the

par-ticipation of ABCA1 in the distribution of lipid

species across the bilayer, we cannot formally

consider PS as the sole or direct substrate of

ABCA1 Indeed the interrelationship between

the different lipid species in the environment of

the membrane is complex and highly dynamic

and we cannot ‘a priori’ exclude the possibility

that the observed movement of PS is balanced

IV

III

II

I

0 10 20 30 40 50

0 10 20 30

Particle size ( µm)

0 10 20 30 40 50

0 10 20 30

Particle size ( µm)

B

Figure 23.7 Delayed engulfment of apoptotic cells is detected during embryonic development in ABCA1 null

animals A microscopic analysis of limb buds at embryonic day 13.5 is shown A schematic representation of

the virtual section in the limb bud is shown in panel A Apoptotic cells are detected by neutral red staining

(panel B) in wild-type (upper) or ABCA1 null (lower) animals An increased number of particles of larger size

is clearly detected in the ABCA1⫺⫺/⫺ ⫺ mice In panel C, the distribution and size of particles corresponding to

apoptotic corpses in wild-type and null animals is shown The graph was derived from the microscopic

analysis of buds stained for apoptotic corpses by the TUNEL technique (Hamon et al., 2000).

ACAMP

APOPTOTIC CELL

PRR

ABCA1-induced lipid domains

Figure 23.8 Proposed model of ABCA1 function during engulfment The ABCA1-generated domains increase the efficiency of engulfment by promoting clustering, by lateral diffusion, of receptors engaged

on the phagocyte membrane in the recognition of the apoptotic prey, through modulation of the properties of the bilayer CD36 (scavenger receptor

of B class) and the PSR (PS receptor) mobilization along the lateral axis of the membrane are indicated by arrows This may allow the generation on the phagocyte surface of specific molecular arrays (PRR: pattern recognition receptors), which then efficiently detect patterns

on the surface of the apoptotic prey (ACAMP:

apoptotic cell-associated molecular patterns) (Franc et al., 1999).

by a primary flip of an as yet unknown

sub-strate We can nonetheless conclude that ABCA1

exerts an indirect control on the biophysical

properties of the membrane, which in a cascade

facilitates the engulfment Indeed in C elegans

it has been reported that the absence of CED-7

(the ABCA1 orthologue) hampers the lateral

mobility of protein involved in the recognition

of the prey It has been shown that the

redistri-bution and clustering on the phagocyte

mem-brane of the CED-1 scavenger receptor, an

event normally triggered by the contact with a

dying cell, is absent in ced-7 mutants (Zhou

et al., 2001a).

ABCA1 AND LIPID HOMEOSTASIS

Normal cellular lipid homeostasis

In normal metabolism, there is continuous

traf-fic of cell phospholipid (PL) and free cholesterol

(FC) to and from their external milieu –

inter-stitial fluids, large vessel lymph, and plasma

(Fielding et al., 1998) The extracellular

accep-tors of cell-derived lipids are mainly high

den-sity lipoproteins (HDL), whose major protein

component is apolipoprotein A-I (apo A-I)

(Frank and Marcel, 2000) Of these acceptors,

a lipid-poor fraction (normally representing

about 5% of total HDL), appears to play the

major role (Castro and Fielding, 1988) These

particles are distinguished by their

prebeta-electrophoretic mobility, which contrasts with

the alpha-mobility of the major, lipid-rich HDL

fraction In vivo, it is not clear if prebeta-HDL

originates from lipid-free apo A-I, or from a

lipoprotein precursor However, the

apopro-tein is widely available commercially and has

been used as a convenient surrogate for the, as

yet unidentified, precursor of physiological

de novo HDL formation (Hara and Yokoyama,

1992; Oram and Yokoyama, 1996)

Peripheral cells, even when quiescent, syn-thesize PL at significant rates Part of this PL is

transferred out of the cell onto lipoprotein

acceptors PL is also internalized from

extracel-lular lipoproteins and degraded by lysosomal

phospholipases (Waite, 1996) In contrast, FC is

neither synthesized nor catabolized in most

peripheral cells at rates that are significant in

comparison to those of either FC efflux or

the uptake of preformed FC from lipoproteins

(Fielding et al., 1998) As a result, most FC

leav-ing the cell has been recycled from lipoproteins,

while most PL is newly synthesized

It was recently shown that FC, internalized from extracellular lipoproteins, recycles within the cell in extra-lysosomal, weakly acidified recycling endosomes which are rich in FC, sphin-golipids and caveolin, the major structural

pro-tein of cell surface caveolae (Pol et al., 2001).

Caveolin also plays a key role in FC efflux and in returning both recycling and newly synthesized

FC to the cell surface (Fielding and Fielding,

1995; Smart et al., 1996) In contrast,

phos-phatidylcholine (PC), the major PL of the plasma membrane and probably other glycerophos-pholipids, newly synthesized in the endoplas-mic reticulum, are transported to the cell surface

by PL transfer proteins (Voelker, 1996)

In summary, while FC and PC both transfer from the cell surface to the same lipoprotein acceptor (lipid-poor HDL), their respective precursor pathways in the cell appear to differ The efflux of cellular FC, and its subsequent metabolism within the plasma compartments and catabolism in the liver, have been termed

the reverse cholesterol transport pathway (Castro

and Fielding, 1988) In this way it is possible to distinguish this flux from the equivalent but opposite ‘forward’ transport of FC, synthe-sized in the liver, to peripheral cells (Castro and Fielding, 1988)

Cellular lipid homeostasis in the context of ABCA1 deficiency

Spontaneous ABCA1 deficiency (Tangier dis-ease) is characterized by the complete absence

of alpha-migrating HDL from the plasma of affected human subjects and storage of choles-teryl esters (CE) within focal accumulations of macrophages, notably the tonsils (Assmann

et al., 2001) The low levels of HDL molecules

present in the plasma of Tangier disease patients

are almost all lipid-poor particles (Asztalos et al.,

2001) These must differ structurally from the prebeta-HDL of normal plasma, which are effec-tively converted into alpha-HDL in the presence

of lecithin:cholesterol acyltransferase (LCAT), which is decreased but not absent in Tangier

disease (Assmann et al., 2001) However, the

composition and properties of Tangier lipid-poor particles have been little investigated One intriguing recent study reported that they were ineffective as a substrate for plasma

phospho-lipid transfer protein activity (von Eckardstein et al., 1998).

Using ‘knockout’ technology, ABCA1-deficient mice have been generated and characterized