CHAPTER 26 – ASSOCIATED WITH ANTIGEN PROCESSING AND LOADING COMPLEX ESSENTIAL FOR CELLULAR IMMUNE RESPONSE

CHAPTER 26 – ASSOCIATED WITH ANTIGEN PROCESSING AND LOADING COMPLEX ESSENTIAL FOR CELLULAR IMMUNE RESPONSE CHAPTER 26 – ASSOCIATED WITH ANTIGEN PROCESSING AND LOADING COMPLEX ESSENTIAL FOR CELLULAR IMMUNE RESPONSE CHAPTER 26 – ASSOCIATED WITH ANTIGEN PROCESSING AND LOADING COMPLEX ESSENTIAL FOR CELLULAR IMMUNE RESPONSE CHAPTER 26 – ASSOCIATED WITH ANTIGEN PROCESSING AND LOADING COMPLEX ESSENTIAL FOR CELLULAR IMMUNE RESPONSE CHAPTER 26 – ASSOCIATED WITH ANTIGEN PROCESSING AND LOADING COMPLEX ESSENTIAL FOR CELLULAR IMMUNE RESPONSE

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

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

2

FUNCTIONAL CLASSIFICATION OF

CHAPTER

MHC C LASS I A NTIGEN

During evolution, the adaptive immune system

has developed to protect the organism against

pathogens This system consists of three

inter-related branches of defense, depending on

where the first step of elimination of foreign

antigens occurs The humoral system is

respon-sible for recognition and elimination of intact

pathogens such as viruses or bacteria in the

extracellular space via antibodies These are

produced by B-lymphocytes and, subsequently,

they activate the complement system The

cel-lular immune system is subdivided into two

components In the first, endogenous proteins

are degraded in the cytosol by the proteasome

and the resulting peptides are transported into

the endoplasmic reticulum (ER), where they

bind to major histocompatibility complex (MHC)

class I molecules In the second, exogenous

proteins are degraded after internalization in

specialized late endosomal or pre-lysosomal

compartments which contain MHC class II

molecules The antigenic peptides are directly

loaded onto the MHC class II complexes

Both peptide-loaded MHC class I and class II

molecules are transported on the cell surface and recognized by cytotoxic and helper T-lymphocytes, respectively The focus of this chapter is on the MHC class I pathway

In the non-infected state, MHC class I com-plexes are presented on the cell surface by binding peptides derived from normal cellular proteins Cytotoxic T-lymphocytes (CTLs) are not activated by this chronic presentation of self-peptides, because T-cells with the ability

to respond to these molecules are eliminated during thymus development The pathways leading to the generation of peptides, their binding to MHC molecules, and their sub-sequent expression on the cell surface are called antigen processing and presentation

(Figure 26.1A) During viral infection or malignant transfor-mation, a set of ‘non-self’ peptides of the target cell is presented to the CTLs, which recognize the MHC class I molecule as a self-component loaded with a peptide derived from non-self proteins and which eliminate these cells (for

review, see Ljunggren et al., 1990; Townsend

et al., 1989) Interference with this antigen

pres-entation pathway is an effective method for pathogens to dodge the immune response

MHC class I molecules are composed of a polymorphic heavy chain (␣-chain), which is encoded within the MHC locus, and an invariant

26

CHAPTER

non-MHC encoded subunit, ␤2-microglobulin

(for review, see Ljunggren et al., 1990;

Townsend et al., 1989) The assembly of the

dif-ferent subunits proceeds in the ER by a folding

process that is synchronized in time and space

by various chaperones After co-translational translocation, the ␣-chain of the MHC class I molecule associates with the chaperone protein,

CTL

A

B

MHC class I molecule

proteasome

peptides protein

tapasin

calreticulin calnexin translocon

ribosome TAP

ERp57 β 2 m MHC

heavy chain

ER

Golgi PM

heavy chain α3

α3

α2

β 2 m

α2 α1

α1 peptide

Figure 26.1 A, Antigen processing pathway via MHC class I In the ER, the MHC class I heavy chain

associates with the chaperone calnexin After assembly with ␤2 -microglobulin (␤2 m) and the thiol reductase ERp57, calnexin is replaced by calreticulin Subsequently, tapasin mediates the association with TAP, forming a macromolecular transport and loading complex for antigenic peptides Peptides are generated within the ubiquitin–proteasome pathway in the cytosol After peptide loading, MHC class I molecules are released from the TAP complex and traffic via the Golgi apparatus and the trans-Golgi network to the cell surface There CTLs recognize the antigenic peptides in complex with MHC class I molecules B, Structure of the peptide–MHC complex Side and top view of the MHC class I molecules HLA-A2 with bound antigenic peptide derived from tax protein of human T-cell lymphotropic virus (Madden et al., 1993) The peptide epitope (LLFGYPVYV, yellow, residues are numbered) binds in a groove of the heavy chain of HLA-A2 (blue), which is formed by two ␣-helices on the rim and eight ␤-strands at the bottom of the ␣1 and ␣2 domain Peptide, heavy chain and ␤2 -microglobulin (green) form a stable complex with a half-life of several days The groove fits peptides with a length of eight to ten amino acids Peptides are fixed via their free amino and carboxy termini as well as via side-chain interactions at position two or three and the C-terminus Side-chains, which point out of the groove, are monitored by the T-cell receptor.

calnexin, and forms a dimer with ␤2

-micro-globulin (for review, see Pamer and Cresswell,

1998) ␤2-Microglobulin interacts extensively

with domains of the ␣-chain and, consequently,

the correct folding of the ␣-chain is dependent

on the dimerization with ␤2-microglobulin

Calnexin is then replaced by another chaperone,

calreticulin, and the thiol reductase ERp57 then

associates with the complex For the binding of

peptides to the MHC class I heterodimer, a

macromolecular loading complex, together with

tapasin and an ATP-binding cassette (ABC)

het-erodimeric protein known as the transporter

associated with antigen processing (TAP)

(ABCB2/ABCB3), is formed Tapasin is an

ER-resident type I glycoprotein that mediates the

efficient interaction of TAP and class I molecules

(Sadasivan et al., 1996) The peptide loading

onto the MHC class I heterodimer stabilizes the

molecule and it is released from the assembly

complex for transport to the plasma membrane

via the Golgi apparatus and the trans-Golgi

net-work (Figure 26.1B).

Normal cellular proteins, as well as viral proteins or proteins that are artificially

intro-duced into the cytosol (Moore et al., 1988;

Yewdell et al., 1988), are degraded by an

extralysosomal pathway (Morrison et al.,

1986) The major sources for antigenic

pep-tides are proteins cleaved by a

proteasome-based mechanism, which degrades unfolded

or ubiquitinated proteins in the cytoplasmic

compartment (Rock and Goldberg, 1999; York

et al., 1999) One form of the proteasome is a

20S (700 kDa) cylindrical particle, consisting

of 28 subunits arranged in four heptameric

rings The outer rings are composed of seven

␣ subunits with regulatory and structural

functions, while the inner rings consist of

seven ␤ subunits containing the catalytic sites

(Baumeister et al., 1998) The 26S proteasome

(1500 kDa) is associated with additional

sub-units, which have a regulatory function

Approximately one-third of newly

synthe-sized proteins are degraded by proteasomes

into peptides with a size distribution of 3–30

amino acid residues (Kisselev et al., 1999;

Schubert et al., 2000; Turner and Varshavsky,

2000) The optimal size is 6–11 residues,

which overlaps with the size of antigenic

pep-tides (8–11 residues) bound to MHC class I

molecules (Kisselev et al., 1999).

The transport of the peptides generated in the cytosol into the ER lumen is executed by

TAP (ABCB2/ABCB3) It has been established

that any defect in TAP severely impairs antigen

presentation Since peptide binding is neces-sary for stabilization of the MHC complex, a reduced or abolished transport activity of TAP results in reduced cell surface expression of MHC class I molecules Thus, TAP function is essential for antigen presentation and, conse-quently, inhibition of TAP function is an effective strategy for pathogens to avoid immune surveil-lance, leading to chronic or latent infections

During the last few years, the understanding

of the function of TAP has increased signifi-cantly Disturbance of peptide delivery to the MHC class I complex is associated with various human diseases from tumor development to infections In this chapter, the current knowl-edge of the mechanisms enabling transport of peptides from the cytosol into the lumen of the

ER for antigen presentation is summarized In addition, TAP serves as an important model system to aid in understanding the topic of multisubstrate specificity, transport mecha-nism and inhibition of function by natural inhibitors

(ABCB2/ABCB3)

GENOMIC ORGANIZATION AND REGULATION OFTAP

Some years ago, it was observed that some tumor cell lines exhibit a low cell surface expres-sion of MHC class I molecules and are deficient

in antigen presentation However, at low tem-peratures, the expression of the MHC class I

␣-chain and ␤2-microglobulin could be restored

to normal levels (Ljunggren et al., 1990;

Townsend et al., 1989) The defect was located

in the MHC locus and it was concluded that

a gene or genes were involved in peptide load-ing of the class I molecules In the followload-ing years, four groups independently discovered candidate genes for proteins that were impli-cated in the transport of peptides from the cytosol into the lumen of the ER (Deverson

et al., 1990; Monaco et al., 1990; Spies et al., 1990;

Trowsdale et al., 1990) Since then, these human,

mouse and rat genes have been renamed as the

transporter associated with antigen processing,

TAP (TAP1/ABCB2 for RING4, PSF1, mtp1 and

HAM1; TAP2/ABCB3 for RING11, PSF2, mtp2

and HAM2) Transfection of defective cell lines

with TAP1 (ABCB2) and/or TAP2 (ABCB3)

cDNAs restores MHC class I surface expression

and antigen presentation These findings

indi-cate that MHC class I molecules are stabilized

by binding peptides and that the majority of

peptides are transported by TAP from the

cytosol into the ER lumen

The human TAP genes are located on

chro-mosome 6 band p21.3 in the MHC II locus

They are 8–12 kb in size and consist of 11 exons

each (Hanson and Trowsdale, 1991) The TAP1

2.5 kb mRNA encodes a protein of 748 amino

acids, while the TAP2 2.8 kb mRNA encodes a

protein of 686 amino acids Sequence alignments

of the coding region from human to horned

shark, the most distant vertebrate class

dis-playing an adaptive immune system, exhibit

the expected phylogenetic differences For

example, human TAP1 shares 98.8% amino

acid homology with the gorilla TAP1 protein,

69.2% with the hamster protein, and

approxi-mately 43% with the horned shark protein

The homology between TAP1 and TAP2 is

approximately 35% in all species examined

thus far and the two proteins share a similar

predicted membrane topology It is likely that

these genes evolved from a common ancestral

gene by gene duplication prior to the

develop-ment of the adaptive immune system in jawed

vertebrates

The human TAP genes contain putative

GC-rich elements (Sp1-binding sites) in their

5⬘-flanking sequences, but no TATA box motifs

(Beck et al., 1992) It was shown by mutagenesis

that the Sp1-binding sites are necessary for

the basal promoter activity of TAP1 (Wright

et al., 1995) In addition, several other motifs

induce TAP1 promoter activity such as

inter-feron (IFN)-␥- and p53-responsive elements

(Zhu et al., 1999) Interestingly, the TAP1 gene is

coordinately regulated by a bi-directional

pro-moter with the divergently transcribed LMP2

gene (Israel et al., 1989) LMP2 encodes the

alternative ␤-type proteasomal subunit, which

is important for differential processing of

epi-topes by constitutive and immunoproteasomes

(Gaczynska et al., 1994; Toes et al., 2001) Both

genes are stimulated by tumor necrosis factor

(TNF)-␣ The induced expression of TAP1 and

LMP2 concordantly with upregulated MHC

class I genes suggests a link between generation

of peptides and expression levels of the trans-porter

Expression of the MHC class I molecules cor-relates with CTL function and can be increased

by cytokines such as IFNs (for review, see Früh

and Yang, 1999) TAP1 and TAP2 mRNA and

protein levels are rapidly upregulated by IFN-␥, whereas MHC class I ␣-chains and cell surface

expression increase more slowly (Ma et al.,

1997) A similar enhancement of TAP1 was

observed by in vitro treatment of tumor samples

with TNF-␣ (Nagy et al., 1998) In contrast to these cytokines, interleukin-10 has a reverse effect on TAP expression and reduces TAP1 and

TAP2 levels (Salazar-Onfray et al., 1997).

In addition to the interference of TAP func-tion by cytokines, some other mechanisms are known to regulate TAP activity In certain breast cancer cell lines, TAP expression was observed to be dependent on the cell cycle, and

the overall amounts of TAP mRNAs were lower

than in normal breast epithelial cells (Alpan

et al., 1996) Tumor cells may evade host tumor

surveillance by mutations that inhibit TAP function Because more than 50% of human tumors have no functional p53, the influence of

p53 on TAP1 levels was examined (Zhu et al., 1999) Overexpression of p53 increased TAP1

mRNA and protein levels and, subsequently, MHC class I cell surface expression The authors suggested that a non-functional p53 cannot induce TAP following genotoxic stress Thus, p53 may act as a tumor suppressor

by inducing TAP and thereby tumor sur-veillance

STRUCTURAL ORGANIZATION OFTAP

Like MDR1 (ABCB1) and MDR3 (ABCB4), TAP1 and TAP2 belong to subfamily B of the ABC superfamily Each TAP protein consists of one ATP-binding domain (nucleotide-binding domain: NBD) and one hydrophobic region of 10 (TAP1) or 9 (TAP2) transmembrane (TM) helices The homology with other ‘half-size’ transporters indicated that a functional TAP complex con-sisted of either a homodimer of TAP1 or TAP2, or

a TAP1/TAP2 heterodimer Heterologous coex-pression of TAP1 and/or TAP2 in yeast and insect cells demonstrated that TAP is active as a heterodimer and that no additional factors of the adaptive immune system are needed for

TAP function (Meyer et al., 1994; Urlinger et al.,

1997) Moreover, immunoprecipitation with antibodies directed against TAP1 co-precipitate

TAP1 and TAP2 (Kelly et al., 1992) The TAP

complex is located in the ER as shown by

immu-noelectron and immunofluorescence microscopy

(Kleijmeer et al., 1992; Meyer et al., 1994) Both

proteins lack an NH2-terminal signal sequence

for ER targeting The complex is retarded in the

ER by an internal signal sequence Recent

stud-ies with truncated proteins indicate that ER

retention of both TAP1 and TAP2 is achieved

by multiple signals in the transmembrane

regions (Vos et al., 1999) TAP1 has three

pre-dicted glycosylation sites, two facing the cytosol

and one placed in an ER loop, which is likely

to be too short for glycosylation Consistent

with these predictions, it was found that both

proteins are predominantly non-glycosylated

(Meyer et al., 1994) A minor subpopulation of

TAP has been reported to be N-glycosylated,

but this may consist of misfolded protein (Russ

et al., 1995).

Hydrophobicity analysis predicts that each TAP protein contains 6–10 TM helices

depend-ing on the algorithm used (Figure 26.2A) (Elliott,

1997; Gileadi and Higgins, 1997; Nijenhuis and

Hämmerling, 1996; Tampé et al., 1997) A core

domain of six TM-spanning helices, which is

found in all other ABC transporters, may

possi-bly serve to align the translocation pore, and the

sequence similarity increases from TM1 through

to TM6 By sequence alignments, the first 175

and the first 140 NH2-terminal amino acid

residues of TAP1 and TAP2, respectively, which

are encoded by exon 1, show no corresponding

domains in related ABC transporters It is

assumed that these hydrophobic regions

con-tain an additional four and three TM helices,

respectively, as extensions and might be

neces-sary for specialization or assembly of the TAP

complex (Tampé et al., 1997) To clarify the

topology of the TAP complex, it may be useful

to construct cysteine-less mutants of TAP1 and

TAP2, which are functionally active Single

cys-teines can then be reintroduced in predicted

loop regions and the accessibility checked by

thiol-specific reagents

Linked to the hydrophobic domains are the NBDs containing the conserved Walker A and

B motifs and the ‘C’ transport family signature

sequence located in the cytosol According to

this model, the complex contains large

cyto-solic loops, but only a small part passes into the

lumen of the ER (Tampé et al., 1997) It was

pro-posed that these TM-spanning domains are

arranged in a head–head/tail–tail orientation

(Vos et al., 2000), but this alignment contradicts

established models of other transporters such

as P-glycoprotein (ABCB1) (Loo and Clarke,

1995, 2001a)

The peptide-binding site is shared by both subunits as was shown by peptide photo-crosslinking and binding experiments (Androlewicz and Cresswell, 1994; Androlewicz

et al., 1993; van Endert et al., 1994) Digestion

of TAP after photo-crosslinking and subse-quent immunoprecipitation with antibodies directed against different epitopes of TAP

TAP1 TAP2

TM1

TM1

TM5

peptide-binding site TM6

TM2

TM3 TM4 TM5

TM6 TM3

TM2

TM4

TMD

A

B

ER N

N

cytosol

peptides

N

B

B

Figure 26.2 A, Structural organization of the TAP transporter Membrane topology was predicted based on sequence alignments with other ABC transporters including MDR1 and hydrophobicity analysis The translocation pore is framed by 2⫻⫻ 6 transmembrane helices from TAP1 and TAP2 (blue cylinders) N-terminal regions of TAP1 and TAP2 have no counterparts in other ABC proteins They putatively contain four and three transmembrane helices, respectively (orange cylinders) The yellow circle encompasses the highly conserved NBD with the Walker A (P loop) and B motifs (red bars) and the C-loop (blue bar) The cytosolic loops following TM6 and within TM5 and TM6 of both subunits delineate the potential peptide-binding site B, Arrangement of the transmembrane helices The transmembrane helices (light blue for TAP1 and dark blue for TAP2) are organized according to the model for MDR1 (Loo and Clarke, 2001a).

provided more detailed insight into the

peptide-binding region of this transporter (Nijenhuis

and Hämmerling, 1996) The cytosolic loops

between TM4 and TM5 and a COOH-terminal

stretch of approximately 15 amino acids

fol-lowing TM6 of TAP1 and TAP2 participate in

peptide binding (Figure 26.2A and B) Deletion

of some of these potential binding sites of TAP1

resulted in loss of transporter function (Ritz

et al., 2001) According to one topological model

of TAP (Abele and Tampé, 1999; Tampé et al.,

1997), these regions should all be located in the

cytosol A different model was proposed by

Vos et al (1999), which is in contrast to the

established membrane topology of other ABC

transporters (Loo and Clarke, 1995) These

authors could find no evidence for membrane

integration of the two hydrophobic regions

adja-cent to the NBDs This could be a misleading

experimental result derived from singularly

expressed, non-functional deletion constructs

of TAP1 and TAP2

HOMOLOGUES OFTAP

As mentioned previously, sequence alignments

show that both TAP proteins belong to

subfam-ily B of the ABC transporter superfamsubfam-ily

Mem-bers of this subfamily may be ‘full’ transporters

like P-glycoprotein/MDR1 (ABCB1) and MDR3

(ABCB4), or half-size transporters like TAP1

(ABCB2), TAP2 (ABCB3) and ABCB9 These

transporters translocate a variety of molecules

across different biological membranes, e.g

steroids and hydrophobic compounds by

P-glycoprotein, phosphatidylcholine by MDR3/

ABCB4 (see Chapter 22), possibly

iron/glu-tathione complexes by ABCB6 and ABCB7

(Chapter 25), and monovalent bile salts by BSEP

(ABCB11, sPgp) For other members of this

subfamily, the substrates are unknown at

pres-ent Sequence alignments of the NBDs and a

phylogenetic tree of the members of subfamily

B reveals that TAP1 and TAP2 are most related

to ABCB9, a half-size transporter of unknown

function (Zhang et al., 2000) (Figure 26.3) The

three genes may have arisen by duplication from

an ancestral gene Because of the close

relation-ship with TAP1 and TAP2, it is likely that

ABCB9 may act as a peptide transporter but

this remains to be established At the moment,

no partner protein for ABCB9 is known;

there-fore, the functional complex may be a homo- or

a heterodimer The next closest relatives to the

TAP proteins are ABCB8 and ABCB10, two

mitochondrial ABC transporters which, due to their relation to the yeast transporter MDL1, putatively transport peptides (see Chapter 25)

SUBSTRATE SELECTION AND SPECIFICITY

OFTAP

The first data concerning the character of pep-tides that are transported by the TAP complex was obtained by trapping peptides in the ER via glycosylation or by binding to MHC class I

molecules (Androlewicz et al., 1993; Neefjes

et al., 1993) Peptides with a length of 8–16

amino acids were found to have equal affinity

for TAP (van Endert et al., 1994), but are most

efficiently translocated into the ER when they

are 8–12 residues long (Koopmann et al., 1996).

Moreover, free NH2- and COOH-termini are

pre-requisites for transport (Momburg et al., 1994; Schumacher et al., 1994a; Uebel et al., 1997)

By screening combinatorial peptide libraries, the contribution of each amino acid to the

MDR subfamily (drugs, lipids) sPgp(N) ABCB5 MDR3(N) MDR1(N) 0.1

MDL1

ABCB10 M-ABC2 ABCB8 M-ABC1 ABCB9 TAP2

TAP1

ABCB7 subfamilyATM1

(Me-glutathione, peptides)

TAP subfamily (peptides)

MDL1 subfamily (peptides)

ABCB6 MTABC3

sPgp(C)

MDR1(C)

MDR3(C)

Figure 26.3 Phylogenetic analysis of TAP homologues The NBDs of all members of the subfamily B of human ABC transporters including the TAP homologue MDL1 in yeast (S cerevisiae) are aligned by ClustalW The member of this family most closely related to TAP is ABCB9 Also, the putative peptide transporters ABCB6, ABCB7, ABCB8 and ABCB10, all located in the inner membranes of mitochondria, are close relatives The yeast mitochondrial peptide transporter MDL1

is included for comparison (gray) The other members of subfamily B transport hydrophobic drugs and lipids (indicated in red) (N) and (C) refer to NBDs of the N- or C-terminal half or full-size transporters.

stabilization of peptide binding to TAP was

analyzed (Uebel et al., 1997) A randomized

peptide mixture with one defined residue was

compared with a totally randomized peptide

library, and the influence of each amino acid on

the affinity for TAP was determined The peptide

with the best binding characteristics showed a

45-fold higher affinity for TAP than a totally

randomized peptide mixture

The effect of each amino acid was found to

be critically dependent on its position in the

peptide (Figure 26.4) Thus, the first three NH2

-terminal residues and the COOH terminal

amino acid were most important for substrate

specificity TAP displayed preferences for pep-tides with Lys, Asn and Arg in the first, Arg in the second, and Trp and Tyr in the third posi-tion at the NH2-terminus The most profound differences were observed for peptide residues

at the COOH-terminus The highest affinity for TAP binding was found for peptides with hydrophobic or basic amino acids (Phe, Leu, Tyr or Arg) in this position It is interesting that these residues at the COOH-terminus are also advantageous for binding to MHC class I mole-cules Moreover, none of the disfavored amino acids served as a preferred anchor for MHC class I binding Thus, it is speculated that the recognition and binding principles of TAP and MHC class I molecules coevolved

The contribution of the peptide backbone

to the substrate specificity of TAP was deter-mined by using peptides of different length by exchanging each residue with its D-enantiomer

D-Amino acids in positions 1–3 and the COOH-terminal position resulted in a markedly reduced affinity for TAP Thus, contact between a pep-tide and TAP seems to occur via the peppep-tide backbone, the amino acid side-chains and the free NH2- and COOH-termini, which is fixed

by hydrogen bonding (Uebel et al., 1997) The

residues in the center of the peptide between positions 1–3 and the COOH-terminal amino acid seem to have only a little or no effect on the substrate specificity of TAP This binding prop-erty can explain how larger peptides can bulge out of the binding pocket and how large amino acid side-chains and even fluorescence labels can

be accommodated (Neumann and Tampé,

1999; Uebel et al., 1995) Interestingly, these

residues are responsible for the detection of the MHC class I-bound peptide by the T-cell recep-tor Therefore, by binding at the termini, TAP transports peptides with maximal diversity in the center of the peptide (positions 5–8), where T-cell recognition occurs Therefore, a coevolu-tion of the genes involved in antigen presenta-tion seems likely to have taken place in order to optimize the antigen processing and recogni-tion machinery (Uebel and Tampé, 1999)

Polymorphisms in TAP have been found in human, mouse and rat by sequence analysis and restriction length polymorphism analysis

(Daniel et al., 1997; Momburg et al., 1994;

Powis et al., 1992; Schumacher et al., 1994a,

1994b) Although polymorphisms can contribute

to immune diversity, no effect of the amino acid changes on the substrate specificity for human and mouse TAP was observed However, a rat

TAP polymorphism has a significant influence

Specificity

Position

Affinity for TAP G

Promiscuity

in sequence and length

T-cell recognition

HLA restricted

Contact sites (TAP1/2)

H H H

1 2 3 4 5 6 7 8 9

⫹N

⫺8 kJ

0 kJ

8 kJ

Specificity

Diversity

A D

F G H I K L M N P Q R S T V W Y

1 2 3 C

Figure 26.4 Specificity of TAP By using

combinatorial peptide libraries and statistical

analysis, human TAP was found to be most specific

for the three N-terminal and C-terminal residues

(Uebel et al., 1997) Favored amino acids (K, N and

R in the first, R in the second and W and Y in the

third position) are shown in blue (negative ⌬⌬⌬

values) in the middle panel At the C-terminus F, L,

R or Y are preferred for binding Disfavored amino

acids are shown in red (positive ⌬⌬⌬⌬G values) In

the lower panel a model of the peptide-binding site

to TAP is shown with the variable region of the

peptide labeled in yellow.

on peptide selectivity (Powis et al., 1992).

Human TAP and rat TAP from the RT1astrain

were found to prefer peptides with hydrophobic

or basic residues at the COOH-terminus, while

mouse TAP and rat TAP from the strain RT1u

favored peptides with hydrophobic

COOH-terminal residues The TAP complexes from

RT1aand RT1ustrains differ by the exchange of

25 amino acids in TAP2: 23 in the

transmem-brane domain (TMD) and two in the NBD

In addition to gene polymorphisms, altered substrate specificity can be achieved by

alter-native splicing Recently, a variant of the human

TAP2 protein, called TAP2iso, was described

(Yan et al., 1999) This splice variant lacks exon

11 comprising a part of the coding region and

the original 3⬘-untranslated sequence Instead,

it contains a previously unidentified exon 12

Expression of TAP2iso mRNA was found to

be coincident with TAP2 mRNA in several cell

lines Interestingly, heterodimers of TAP1 and

TAP2iso more resemble mouse and rat RT1u

TAP, with preferences for peptides with

hydrophobic COOH-termini Thus, alternative

splicing may be a strategy for the organism to

acquire broadened epitope diversity How the

variable COOH-terminus of the TAP2 NBD

affects TAP substrate specificity remains an

open and intriguing question

TRANSPORT MECHANISM OFTAP

The transport mechanism of TAP is the subject

of intensive investigation because of its

impor-tant role in immune recognition Translocation

into the lumen of the ER is a multistep process,

consisting of binding of the peptide to TAP,

iso-merization of the complex, and transport

(Figure 26.5) (review by Abele and Tampé,

1999) Peptide interaction with the binding site

shared by both TAP subunits is

ATP-independ-ent and follows a monophasic 1:1 Langmuir

adsorption model (Uebel et al., 1995) In direct

binding or competition assays, no evidence for

a second interaction site was found However,

it cannot be excluded that a second binding

site with a very low affinity, or with a similar

affinity, exists Real-time kinetic analysis of

the peptide binding with environmentally

sen-sitive fluorescence labeled peptides revealed

that this process could be subdivided into two

steps (Neumann and Tampé, 1999) The

pep-tide binding occurs in a fast bimolecular

associ-ation step and determines the specificity of

TAP; subsequently, a slow isomerization of the

TAP complex takes place It is proposed that the conformational change of the molecule triggers ATP hydrolysis and, thereby, peptide transport into the lumen of the ER The isomer-ization of the complex also affects its lateral mobility as analyzed by fluorescence recovery

after photobleaching (Reits et al., 2000) This

increases when TAP is inactive and decreases during peptide translocation, as was shown

in studies with TAP1 tagged with green fluo-rescent protein (GFP) at its cytosolic COOH-terminus However, owing to the presence of endogenous TAP, the activity of the GFP-tagged complex could not be unequivocally established

Peptide

TAP 1

TAP 1

TAP 2

TAP 1 TAP 2

TAP 1 TAP 2

TAP 2

ATP

ATP

ATP

ADP ADP

Figure 26.5 Working model of the translocation mechanism by TAP In the ground state, ATP interacts primarily with the TAP1 subunit, whereas TAP2 most probably contains pre-bound ADP in an occluded state (red) (Alberts et al., 2001; Karttunen

et al 2001; Lapinski et al., 2001; Saveanu et al.,

2001) High-affinity peptide binding occurs in a fast reaction followed by a slow isomerization of the TAP complex, promoting allosteric coupling between the two NBDs (Neumann and Tampé, 1999) Peptide binding to TAP triggers ATP hydrolysis and subsequent translocation of the solute (Gorbulev

et al., 2001) ATP hydrolysis and peptide binding are

tightly coupled The release of inorganic phosphate and subsequently ADP at TAP1 might catalyze nucleotide exchange at TAP2 ATP hydrolysis at TAP2 finally closes the transport cycle by restoration of the high-affinity peptide-binding pocket The maximal turnover rate of the transport cycle was determined to be around 5 ATP per second (Gorbulev et al., 2001).

The transport of peptides from the cytosol to the ER lumen strictly requires hydrolysis and

not merely binding of ATP, UTP, CTP or GTP

(Androlewicz et al., 1993) Non-hydrolyzable

analogues of ATP, nucleotide depletion by

apyrase, or competition with ADP, completely

abrogate peptide translocation (Meyer et al.,

1994; Neefjes et al., 1993; Shepherd et al., 1993).

Evidence of the binding of nucleotides to the

NBDs was obtained by crosslinking

experi-ments with 8-azido-ATP (Müller et al., 1994;

Russ et al., 1995) Nucleoside tri- and

diphos-phates can compete for binding and have

simi-lar affinities for TAP Thus, for example, ADP

inhibits peptide translocation By developing

an enrichment and reconstitution protocol for

TAP, it was possible to restore the function

in proteoliposomes and to examine the

spe-cific ATPase activity (Gorbulev et al., 2001).

Nucleotide hydrolysis was found to be strictly

dependent on binding of peptides and on

crosstalk with the peptide-binding and the

translocation sites The strict correlation between

peptide binding and stimulation of ATP

hydro-lysis may be a strategy to avoid ‘wasting’ ATP

without transport of peptides A further

indica-tion of the tight coupling between ATP

hydro-lysis and peptide transport is the observation

that sterically restricted peptides, which cannot

be transported by TAP, do not induce ATP

hydrolysis Maximal ATPase activity of TAP

was found to be independent of substrate

affin-ity, because peptides with different KDvalues

for the transporter exhibited the same Vmax

val-ues (Gorbulev et al., 2001) The two NBDs of the

functional TAP complex can both interact with

ATP, even if TAP1 or TAP2 are expressed

sepa-rately (Müller et al., 1994; Wang et al., 1994).

However, alone, the NBDs are unable to

hydrolyze ATP Thus, communication between

the NBDs and TMDs leading to a conformational

change of the NBDs by peptide binding to TAP

seems to be a requirement to activate ATPase

function Furthermore, TAP function is

depend-ent on the presence of both NBDs since

disrup-tion of one NBD leads to loss of transport (Chen

et al., 1996) Thus, it has been speculated that

hydrolysis at one NBD is necessary for the

begin-ning of the transport, whereas hydrolysis at the

second NBD completes the cycle and may

pro-mote the reconversion of the peptide binding

site to the initial state (Abele and Tampé, 1999)

But one question remains: are the two NBDs equal in function or do they have distinct

functional properties? To address this point,

several groups have introduced mutations in the

Walker A and/or B motifs in the NBDs of TAP1 and TAP2 and examined the effects on trans-port function Lysine mutations in the Walker A sequences affecting nucleotide binding/hydro-lysis by TAP1 or TAP2 suggest that each NBD

plays a distinct functional role (Karttunen et al., 2001; Lapinski et al., 2001; Saveanu et al., 2001).

Even if the data concerning nucleotide and peptide binding from the different groups are

in part contradictory, it seems that nucleotide binding to TAP2 maintains a peptide-receptive TAP conformation, and that TAP2-mediated ATP hydrolysis is essential for translocation

The functional consequences of mutations in the Walker A motifs of TAP1 and TAP2 have led to speculations that ATP hydrolysis at TAP1 initiates the transport cycle and is a require-ment for binding of ATP to the NBD of TAP2

(Alberts et al., 2001) Hydrolysis by TAP2 might

then complete the cycle by restoring the peptide-binding site The use of chimeric proteins consisting of the TAP1 membrane-spanning domain and the TAP2 NBD, and vice versa, indicate that both membrane-spanning domains

of TAP1 and TAP2 are necessary, but that neither NBD encompasses signals unique for peptide

binding (Arora et al., 2001) These observations

are in agreement with earlier data demonstrating that the TM domains of both TAP1 and TAP2 are needed to form the peptide-binding pocket

The NBD-switched complexes are all transport

competent (Arora et al., 2001) Even TAP

com-plexes with two identical NBDs are able to translocate peptides, although with a lower effi-ciency The two NBDs of TAP1 and TAP2 appear

to possess different functional properties Fur-ther studies will elucidate the exact roles of each NBD within the transport cycle

TAP not only delivers peptides necessary for antigen presentation into the ER lumen, but

is also part of a large macromolecular loading complex which is critical for MHC class I

maturation A number of proteins have been

identified which play an important role in this

complex (for review, see Cresswell et al., 1999).

At least three proteins are involved in the

assembly of peptide-loaded TAP and MHC

class I heterodimers: the chaperone calreticulin,

the thiol reductase ERp57, and the glycoprotein

tapasin (TAP-associated glycoprotein)

Cal-reticulin acts as a chaperone to ensure proper

folding (Sadasivan et al., 1996), whereas ERp57

probably supports the correct formation of

disulfide bridges (Lindquist et al., 1998), and

tapasin mediates the efficient interaction of TAP

and the MHC class I molecules and stabilizes

the loading complex (Ayalon et al., 1998;

Ortmann et al., 1997) The stoichiometry of this

complex was determined to be four MHC class

I heterodimers associated with four tapasins to

one TAP heterodimer (Ortmann et al., 1997) In

the absence of tapasin, the assembly in the ER

is impaired and MHC class I antigen

presenta-tion decreases However, in a tapasin-deficient

cell line, the class I cell surface expression and

function is restored by a truncated soluble

tapasin lacking the transmembrane region, even

if the remaining cytosolic tail is not linked to

TAP (Lehner et al., 1998) The physical

associa-tion of TAP and class I molecules leads to the

assumption that the peptides are directly loaded

from TAP onto the MHC class I complex

How-ever, application of anti-peptide antibodies

inhibits peptide binding to class I molecules

(Hilton et al., 2001) Therefore, these authors

suggested that most TAP-transported peptides

diffuse through the ER lumen before being

loaded onto MHC class I molecules The binding

of the peptides is necessary for the dissociation

of TAP–MHC class I complexes and is

depend-ent on conformational signals from TAP in an

ATP-dependent manner (Cresswell et al., 1999;

Knittler et al., 1999) Recent observations point

to a more pronounced conformational role of

TAP1 in the dynamic activity of the loading

complex (Alberts et al., 2001) The release of the

MHC class I molecules for transport to the cell

surface is synchronized with peptide binding

and peptide translocation by TAP

In non-infected cells, MHC class I molecules

are stably expressed on the cell surface

present-ing peptides derived from intracellular proteins

Presenting peptides from viral proteins enables T-cells to recognize and eliminate infected cells

To propagate in the presence of an active immune system, the virus must develop a strat-egy to avoid an immune response One mecha-nism is to interfere with the antigen presentation pathway at different stages (for review see Ploegh, 1998) Many steps are susceptible to viral disturbances, such as the generation of peptides

(Gilbert et al., 1996; Levitskaya et al., 1997), the

export of class I molecules to the cell surface

(Früh et al., 1999; Hengel et al., 1999), and the

transport of peptides by TAP Here, we will

focus on the latter point (Figure 26.6).

Human cytomegalovirus (HCMV) encodes several proteins inhibiting cell surface expres-sion of MHC class I molecules (for review, see Hengel and Koszinowski, 1997) One of these proteins is known to interfere with intracellular peptide transport: gpUS6 gpUS6 is an ER-resi-dent glycoprotein that probably binds to the

ER luminal part of the TAP complex (Ahn et al., 1997; Hengel et al., 1997; Lehner et al., 1997).

The association of TAP with tapasin, calreticulin and MHC class I molecules seems to be unaf-fected by gpUS6, as is the binding of peptides

to TAP Recent data point to a binding of gpUS6

to TAP, which results in stabilization of a TAP1 conformation that is unable to bind ATP (Hewitt

et al., 2001; Kyritsis et al., 2001) Consequently, the

energy for peptide transport is lacking It is possi-ble to override this inhibition by overexpression

of TAP, for example by induction with IFN-␥ Herpes simplex virus type I (HSV-1) has evolved a completely different strategy to avoid immune recognition This virus encodes the

herpes simplex virus human cytomegalovirus

ER

TAP1 TAP2 TAP1 TAP2 cytosol

ICP47

US6

Figure 26.6 Inhibition of TAP function by viral proteins The herpes simplex virus-encoded protein ICP47 blocks the TAP-mediated peptide transport

by binding to the cytosolic part of the TAP complex (left side), whereas the human cytomegalovirus protein US6 inhibits TAP function by binding to the ER-luminal side (right side) The association of TAP with tapasin and MHC class I molecules seems to be unaffected by both proteins.