INTRODUCTION

TOP ABSTRACT INTRODUCTION THE MECHANISM OF EXCHANGE... INHERITED AMINOACIDURIAS HSHATs LSHATs CELL PHYSIOLOGY OF THE... CELL PHYSIOLOGY OF THE... REFERENCES

SIX FAMILIES OF AMINO ACID transporters for the cell plasma membrane have been described in mammals (reviewed in Refs. 85, 121, 123), one of which has a heteromeric structure. These heteromeric amino acid transporters (HATs) are composed of a heavy subunit (rBAT or 4F2hc) and the corresponding light subunit, linked by a disulfide bridge (Table 1 and Fig. 1). These transporters were identified after the cloning of rBAT (also named NBAT and D2) from kidney by their functional expression in oocytes (9, 171, 186). Amino acid transport in oocytes was also induced by the expression of the rBAT-homologous protein, the heavy chain of the surface antigen 4F2 (4F2hc; 4F2 is also referred to as CD98) (8, 187). These two heavy subunits of HAT (HSHATs) are type II membrane glycoproteins with a single transmembrane (TM) domain, an intracellular NH₂ terminus, and an extracellular domain that shows significant homology with bacterial -glucosidases. rBAT is mainly expressed in the epithelial cells of the kidney proximal tubule and of the small intestine, where it is located in the brush border. In contrast, 4F2hc is ubiquitous, with a basolateral location in epithelial cells. The first studies of rBAT and 4F2hc have been reviewed extensively (121).

View larger version (62K): [in this window] [in a new window]   Fig. 1.   Schematic representation of the heteromeric amino acid transporters (HAT). The heavy subunit of HAT (HSHAT; dark gray) is linked by a disulfide bridge to the corresponding light subunit of HAT (LSHAT; light gray). The cysteine residues involved in this bond (S-S) are located extracellularly, just after the transmembrane (TM) domain of HSHAT and in the proposed extracellular loop 2 (EL2) of LSHAT. HSHAT is a type II membrane glycoprotein with a bulky extracellular domain that is homologous to bacterial glucosidases. LSHATs are nonglycosylated proteins with 12 putative TM domains. Loops and TM domains are not drawn to scale.

Studies of covalent inactivation by mercurial agents demonstrated that 4F2hc needs an accompanying subunit(s) to express transport activity (47; reviewed in Ref. 122). The first light subunits of HAT (LSHATs) were identified in 1998 (LAT-1, y⁺LAT-1, and y⁺LAT-2) (79, 101, 128, 133, 175). Since then, four more mammalian LSHAT members have been cloned xCT, LAT-2, asc-1 and b^0,+AT (BAT1 in Ref. 27) (7, 18, 27, 52, 47a, 112, 127, 131, 137, 138, 141, 149, 150, 154). The LSHAT members have recently been reviewed (40, 118, 178, 179). Their general characteristics are the following. First, LSHATs are not N-glycosylated and are highly hydrophobic, with 12 putative TMs (Fig. 1). The lack of N-glycosylation of LAT-1 after translation in vitro has been demonstrated (79). The highly hydrophobic character of LSHATs (molecular mass ~50 kDa) results in an anomalous mobility in SDS-PAGE, compatible with an apparent molecular mass of 35-40 kDa. Second, LSHATs are linked to the corresponding HSHAT by a disulfide bridge. For this reason, HATs are also named glycoprotein-associated amino acid transporters (gpaATs) (178). The intervening cysteine residues are located in the putative extracellular loop II of LSHATs and a few residues toward the COOH terminus from the single putative TM of HSHATs (Fig. 1). Evidence for the formation of these heterodimers (~125 kDa) in heterologous expression systems has been obtained for 4F2hc (~85 kDa) with LAT-1, LAT-2, or y⁺LAT-1 (101, 110, 128, 129, 141, 175) and for rBAT (~94 kDa) with b^0,+AT (Fernández E, Chillarón J, and Palacín M, unpublished observations), and by coimmunoprecipitation of LAT-1 and 4F2hc (100) and b^0,+AT and rBAT (Fernández E, Chillarón J, and Palacín M, unpublished observations) from naturally occurring tissues. Third, LSHAT members need coexpression with the corresponding HSHAT to reach the plasma membrane in heterologous expression systems [LAT-1, LAT-2, asc-1, y⁺LAT-1, and xCT with 4F2hc (7, 101, 109, 110, 129, 131), and b^0,+AT with rBAT (47a)]. Fourth, LSHATs confer specific amino acid transport activity to the heteromeric complex (Table 1). Coexpression of 4F2hc with LAT-1 induces a variant of system L (sodium-dependent transport of neutral amino acids with a large side chain) (79, 101), with LAT-2 induces another variant of system L for neutral amino acids of any size (131, 141, 154), with asc-1 induces system asc (sodium-independent transport for neutral amino acids of small side chain) (52, 112), with xCT induces system x (sodium-independent transport for cystine and anionic amino acids) (7, 149, 150), and with y⁺LAT-1 or y⁺LAT-2 induces system y⁺L (sodium-independent transport of dibasic amino acids and sodium-dependent transport for neutral amino acids) (78, 128, 175). In contrast to the above-mentioned LSHAT members, b^0,+AT induces with rBAT, but not with 4F2hc, system b^0,+ (sodium-independent transport for dibasic amino acids and neutral amino acids, including cystine) (27, 47a, 104, 127). Fifth, all the amino acid transport activities associated with the LSHATs behave like amino acid exchangers: systems induced by LAT-1, LAT-2, y⁺LAT-1, y⁺LAT-2, and xCT, and system b^0,+ (reviewed in Ref. 12) show a tight coupling exchange of substrates (7, 20, 101, 104, 128, 131, 149, 154), whereas asc-1 might have some unidirectional flux of substrates in addition to the exchange transport (52).

	THE MECHANISM OF EXCHANGE OF HATs

TOP ABSTRACT INTRODUCTION THE MECHANISM OF EXCHANGE... INHERITED AMINOACIDURIAS HSHATs LSHATs CELL PHYSIOLOGY OF THE... CELL PHYSIOLOGY OF THE... REFERENCES

The amino acid exchange activity of HATs was evidenced before the identification of LSHATs (reviewed in Ref. 124). The seminal observations were described simultaneously and independently by two groups (21, 34), who reported outward positive currents associated with the heteroexchange of neutral (efflux) and dibasic amino acids (influx) via system b^0,+ in whole or cut-open oocytes expressing rBAT. Further studies demonstrated that system b^0,+/rBAT acts as a tertiary active transporter, mediating the electrogenic exchange of dibasic amino acids (influx) for neutral amino acids (efflux) with a stoichiometry of 1:1 (30). This exchange has also been demonstrated in the apical plasma membrane of the proximal tubular cell model of opossum kidney (OK) cells in rBAT-antisense experiments (105) and in chicken brush-border jejunum (174).

System y⁺L induced by 4F2hc in oocytes also behaves as an electroneutral and asymmetric amino acid exchanger (30): it mediates the efflux of dibasic amino acids and the influx of neutral amino acids plus sodium. The transport of sodium via 4F2hc/y⁺LAT-1-induced system y⁺L has recently been demonstrated in oocytes (78). Similarly, the amino acid transport activity x exchanges the ionic form of cystine for glutamate with a 1:1 stoichiometry (6). This mechanism of exchange has also been demonstrated in oocytes coexpressing 4F2hc and xCT (7, 149). As mentioned above, the exchange of substrates has been demonstrated for all the LSHATs cloned when coexpressed with the corresponding HSHAT.

A functional model for rBAT-induced system b^0,+ exchange activity was proposed by Coady et al. (33). These authors observed in rabbit rBAT-expressing, cut-open oocytes that aminoisobutyric acid (AIB) induced amino acid currents across system b^0,+ without being transported itself, thus suggesting variable stoichiometry of exchange. To explain these results, a "double-gated" pore model with a binding site accessible at each side of the membrane was proposed.

Very recently, Torras-Llort et al. (174) studied system b^0,+ in chicken brush-border jejunum. In these vesicles, accessibility to both sides of the plasma membrane allowed kinetic and simulation analysis of the system b^0,+ amino acid exchanger. The results were compatible with a sequential mechanism, which implies the formation of a ternary complex (i.e., the transporter bound to a substrate at each side of the membrane). In contrast, the results ruled out a "ping-pong" mechanism (i.e., binding of a substrate on one side of the membrane and then translocation and release of the substrate on the other). The study did not distinguish between ordered and random binding of substrates to the transporter in the sequential mechanism, but the estimated dissociation constants for extracellular (extravesicular or external) or intracellular (intravesicular or internal) substrates suggest that the binding affinity for the extracellular amino acid is higher than for the intracellular substrate. An ordered mechanism, in which the free transporter binds first to the external amino acid and then to the internal one, may account for these results. However, because the binding affinity for the internal amino acid is high (in the micromolar range), the results could also be explained by a random mechanism with a preferential route (i.e., preference for the binding of the external amino acid first). Such preferential behavior might be due to the negative membrane potential, which would favor the binding of cationic amino acids to the transporter from the external side rather than from the internal one. The "double-gated" pore model was not supported in the chicken brush-border jejunum studies because interaction of AIB with system b^0,+ was not substantiated (174).

The results in chicken brush-border jejunum are compatible with a double-exchange pathway with alternating access (Fig. 2). A similar model was proposed by Dierks et al. (42) for the mitochondrial aspartate/glutamate antiporter, which includes two functional "subunits" or pathways with binding sites alternating at each membrane domain in which the translocation step is under membrane potential control. The functional oligomeric structure of rBAT/b^0,+AT system b^0,+ is unknown, as it is for the other HATs. Interestingly, Western blot analysis revealed a high-molecular-mass complex (~250 kDa), in addition to the ~125-kDa heterodimer, for rBAT (120, 181) and for b^0,+AT (Fernández E, Chillarón J, and Palacín M, unpublished observations) in kidney brush-border preparations under nonreducing conditions. This suggests that system b^0,+ might be a heterotetramer comprising two heterodimers of rBAT and b^0,+AT linked by disulfide bridges. In this scenario, each heterodimer would represent a single pathway of transport with alternating accessibility, and the two heterodimers together would represent the double-exchange pathway. Structural and functional (i.e., with dominant negative mutants) studies to define the functional structural unit of the heteromeric amino acid transporters are presently in progress.

View larger version (7K): [in this window] [in a new window]   Fig. 2.   Model of sequential exchange with a double-transport pathway with alternating access for the heteromeric amino acid transporters. In this model, the transporters are composed of 2 transport pathways. Left to right: 1) binding of a substrate to the outside face ; higher affinity than the inside face) increases affinity for substrates in the inside face ; 2) binding of a substrate inside results in the formation of a ternary complex; 3) transporter changes accessibility to each of the two transport pathways; 4) dissociation of the substrate outside reduces the affinity of the substrate inside; 5) dissociation of the substrate inside. This is the simplest model that explains the kinetics results obtained for system b0,+ in chicken brush-border jejunum by Torras-Llort et al. (174).

	INHERITED AMINOACIDURIAS

TOP ABSTRACT INTRODUCTION THE MECHANISM OF EXCHANGE... INHERITED AMINOACIDURIAS HSHATs LSHATs CELL PHYSIOLOGY OF THE... CELL PHYSIOLOGY OF THE... REFERENCES

The transport characteristics of two of the LSHAT-associated transport systems are relevant to inherited aminoacidurias cystinuria and lysinuric protein intolerance (LPI; see Transepithelial transport of amino acids). First, system b^0,+ (induced by rBAT and b^0,+AT) acts as a tertiary active mechanism of renal reabsorption and intestinal absorption of dibasic amino acids and cystine; it mediates the electrogenic exchange of dibasic amino acids (influx) for neutral amino acids (efflux). Second, system y⁺L (induced by 4F2hc and y⁺LAT-1) mediates the electroneutral exchange of dibasic amino acids (lysine, arginine, and ornithine; efflux) for neutral amino acids plus sodium (30, 40, 78). It is assumed that this transport system allows the efflux of dibasic amino acids against the membrane potential in many cell types, particularly in the basolateral domain of epithelial cells. The role of rBAT, b^0,+AT, and y⁺LAT-1 in cystinuria and LPI has recently been reviewed (119).

Cystinuria

Presently, we classify two types of cystinuria types: type I (MIM 220100) and non-type I (MIM 600918) (124). Type I heterozygotes are silent, whereas in non-type I heterozygotes there is a variable degree of urinary hyperexcretion of cystine and dibasic amino acids. Patients with a mixed type, inheriting type I and non-type I alleles from either parent, have also been described (57). Type I cystinuria represents >60% of the cases of the disease.

The amino acid transport activity associated with rBAT (system b^0,+) and the expression of rBAT in the brush border of the renal epithelial cells of the proximal tubule and of the small intestine pointed to the rBAT gene (SLC3A1) as a candidate for cystinuria. Mutational and linkage studies demonstrated that mutations in SLC3A1 cause type I cystinuria (23, 24, 54). Over 60 distinct rBAT mutations have been described, including nonsense, missense, splice site, and frameshift mutations, as well as long deletions; mutation M467T is the main type I cystinuria allele found worldwide in 38 nonrelated chromosomes (reviewed in Ref. 124). Cystinuria resembling type I, due to mutations in canine SLC3A1, has been reported in Newfoundland dogs (61).

The gene causing non-type I cystinuria was assigned by linkage analysis to the 19q12-13.1 region (11, 166, 184). In 1999, the non-type I cystinuria gene was identified as SLC7A9 (47a). SLC7A9 was a positional candidate gene for non-type I cystinuria because it has the proper chromosomal location, rBAT-associated amino acid transport activity (system b^0,+), and tissue expression (mainly in kidney and small intestine, but also in pancreas and liver). The protein product encoded by SLC7A9 was termed b^0,+AT (for b^0,+ amino acid transporter).

SLC7A9 is the main, if not the only, non-type I cystinuria gene. In fact, after an exhaustive screening of the open reading frame of SLC7A9 by the International Cystinuria Consortium (49a), 35 distinct mutations were found, accounting for 79% of the carrier chromosomes in 61 non-type I patients, mutation G105R being the main non-type I cystinuria allele (25%). The unexplained alleles might be due to mutations outside the open reading frame of SLC3A1, although other gene(s) might be involved in non-type I cystinuria.

All the data discussed so far strongly indicate that rBAT and b^0,+AT are subunits of the amino acid transporter b^0,+. However, this view is challenged by the finding that rBAT has a gradient of expression along the kidney proximal tubule: segment S3 > S2 > S1 (27, 53, 80, 127, 130, 136), whereas b^0,+AT has the opposite gradient of expression: S1 > S2 > S3 (27, 104, 127, 136) [S1 being the small initial part of the proximal convoluted tubule (PCT), S2 the rest of PCT plus the cortical proximal straight tubule (PST), and S3 the terminal part of the PST located in the outer stripe of the outer medulla]. Then, an additional LSHAT for rBAT or HSHAT for b^0,+AT might be available. Identification of these proteins might help us to understand the molecular bases of cystinuria and why mutations in SLC3A1 are completely recessive, whereas mutations in SLC7A9 are incompletely recessive (for a detailed discussion, see Refs. 47a and 49a).

LPI

The LPI locus was assigned to chromosome 14q11.2 by linkage analysis (92, 93). The LPI gene was later identified by candidate positional cloning (14, 176) after identification of y⁺LAT-1 (SLC7A7 gene) (175). y⁺LAT-1 is expressed in target tissues of LPI such as kidney, lung, and small intestine, among others, and induces system y⁺L transport activity when coexpressed with 4F2hc (128, 175). The defective system y⁺L transport in LPI is restricted to intestine, kidney, and probably liver and lung. In fact, system y⁺L is not altered in LPI erythrocytes or fibroblasts (15, 38), most probably because of the expression of the y⁺L transporter isoform y⁺LAT-2 in these cells.

Twenty-five distinct LPI-associated mutations spread along the entire SLC7A7 gene have been identified in 96 LPI patients; only 3 alleles remain to be explained (reviewed in Ref. 119). All Finnish LPI patients share the same founder mutation: a splice site mutation (IVS6-2 AT) that creates a frameshift after Val²⁹⁸ within putative extracellular loop 4 and a premature stop-codon 9 amino acid residues thereafter (109, 176). A genotype-to-phenotype correlation cannot be established in LPI because of extensive clinical variability associated with the same genotype (reviewed in Ref. 119). Then, other factors in addition to mutations of SLC7A7might have a role in the pathogenesis and clinical manifestations of LPI.

	HSHATs

TOP ABSTRACT INTRODUCTION THE MECHANISM OF EXCHANGE... INHERITED AMINOACIDURIAS HSHATs LSHATs CELL PHYSIOLOGY OF THE... CELL PHYSIOLOGY OF THE... REFERENCES

Six sequences are available for mammalian rBAT (human, rat, mouse, rabbit, dog, and a partial sequence for the American opossum), sharing 69-89% amino acid identity and five sequences for vertebrate 4F2hc (human, rat, mouse, Chinese hamster, and zebrafish), sharing 41-89% amino acid identity (see GenBank accession nos. in the legend for Table 1). The rBAT protein (685 amino acid residues for the human counterpart) is longer than the 4F2hc protein (529 for the human counterpart), and they share ~25% amino acid sequence identity (Fig. 3). N-glycosylation was shown for both proteins (reviewed in Ref. 121).

View larger version (99K): [in this window] [in a new window]   Fig. 3.   Multialignment of HSHAT proteins and oligo-1,6-glucosidase (O1,6G). Human rBAT and 4F2hc, two hypothetical proteins from Caenorhabditis elegans (TrEMBL accession nos. O45298 and Q9XVU3;), one hypothetical protein from Drosophila melanogaster (Q9VHX9), and O1,6G from Bacillus cereus (GenBank accession no. X53507) are shown. The alignment was obtained initially with tcoffee (113), and problematic parts (between A4 and A8) were aligned by threading with FUGUE (155), and finally by alignment of predicted secondary structures with Jpred (37). The principal secondary structural elements (PDB file 1UOK) of O1,6G are indicated at the top of each sequence (, -helix; , extended strand) in the 3 protein domains [TIM barrel (A); protruding domain (B); antiparallel -barrel (C)], and the sequence is underlined. The secondary structures in the rest of the sequences were also predicted (solid underline) or not (dotted underline) with Jpred (37). Amino acid residues conserved in O1,6G are shown shaded in gray. Residues affected by cystinuria-specific missense mutations in human rBAT are shown in white over black. The cysteine residue conserved in the HSHAT family is boxed. The putative TM domain in HSHAT proteins is underlined in bold. The asterisks indicate amino acid residues in O1,6G suspected of being involved in the catalysis of the enzyme. The motif QPDLN at the COOH-terminal end of domain B of O1,6G is indicated.

Membrane topology algorithms predicted that both proteins would be type II membrane glycoproteins, with an NH₂ terminus inside the cell, a single TM, and a bulky COOH terminus located outside the cell (reviewed in Ref. 121). The cysteine residue participating in the disulfide bridge with the LSHATs is four to five amino acids away from the TM, toward the COOH terminus (Fig. 3). In contrast to this view, Mosckovitz and co-workers (107), on the basis of accessibility studies with various antibodies, proposed that rBAT contains at least four TMs, with NH₂ and COOH termini located intracellularly. Recently, however, Fenczik and co-workers (49) showed with HA-tag constructs that the NH₂ terminus of 4F2hc is intracellular whereas the COOH terminus is extracellular (49), as expected for a type II membrane glycoprotein. A further argument in favor of this structure is the homology of the HSHAT bulky COOH-terminal domain with -amylases (9, 134, 171, 172, 186). Indeed, this HSHAT domain shows some homology only with insect maltase (-glucosidase) and maltase-like precursors (35-40% amino acid identity) and with bacterial -glucosidases (~30% amino acid identity).

This -amylase family consists of a large group of starch hydrolases and related enzymes, comprising ~20 different enzyme specificities, and is presently known as glycosyl hydrolase family 13 (75). The members have a similar architecture, with a catalytic (/)8-barrel or TIM-barrel (domain A), interrupted by a small calcium-binding subdomain (domain B) protruding between the third -strand (A3) and the third -helix (A3), and a COOH-terminal domain (domain C) with an antiparallel -barrel structure. Major differences in amino acid sequence among the -amylase family members occur within domain B. Janecek et al. (75) clustered the -amylase members in five groups with >50% sequence identity for domain B and suggested that it varies with enzyme specificity. The group defined by Bacillus cereus oligo1,6-glucosidase (O1,6G) also includes the rBAT proteins. Domain B of O1,6G has a complex topology not shared by other structurally solved -amylases, which, in turn, have a common - architecture known as a two-layer sandwich (CATH; protein structure classification: http://www. biochem.ucl.ac.uk/bsm/cath_new/).

The three-dimensional structure of O1,6G has been refined at 2.0-Å resolution (185). The secondary structural elements of this enzyme are indicated in Fig. 3, together with an alignment with human rBAT and 4F2hc, one putative HSHAT from Drosophila melanogaster and two from Caenorhabditis elegans. Because of the different levels of sequence similarity for each domain (A, B, and C) between HSHATs and O1,6G, a global alignment has first been performed for the corresponding glycosyl hydrolase family, and, finally, ambiguous regions have been refined by combining secondary structure predictions and threading techniques (see legend for Fig. 3 for details). Sequence homology between HSHATs and O1,6G starts with two contiguous tryptophan residues a few amino acid residues away from the cysteine residue involved in the formation of the disulfide bridge with LSHATs. Domain A of O1,6G is highly conserved in HSHATs. Indeed, the secondary structure elements of the (/)8 barrel, i.e., A1, A1, A2, A2, A3, A3, A4, A8, and A8, are easily identified by sequence homology in human rBAT and 4F2hc and in the putative HSHATs from D. melanogaster and C. elegans (Fig. 4). In contrast, sequence homology with O1,6G is poor for HASATs in the region between 4 and 7. Then, threading and secondary structure prediction have been used to assign the secondary structure elements, i.e., A4, A5, A5, A6, A7, and A7, of the (/)8 barrel in the protein sequence of most of the vertebrate and invertebrate HSHATs (Fig. 3). This strongly suggests that the structure of the bulky COOH-terminal domain of HSHATs corresponds to a TIM-barrel.

View larger version (58K): [in this window] [in a new window]   Fig. 4.   Membrane topology model of human b0,+AT. The 22 cystinuria-specific mutations affecting single amino acid residues are indicated. The boxed mutations are those associated with a severe phenotype (see text and Table 2 for details). Mutations indicated in bold italics are those associated with a mild phenotype. The rest of the mutations (smaller font) have an unknown or ambiguous phenotype. Conserved amino acid residues in all the LSHAT subfamily members (31 transporters, excluding the mammalian orphan transporters) are indicated inside filled circles or squares. Amino acid residues with small side chain (i.e., Gly, Ser, or Ala) in all of these transporters are indicated in gray circles or squares. The extracellular (EL) and intracellular (IL) loops are numbered. SH, the cysteine residue involved in the disulfide bridge with rBAT; N- and C-, NH2 and COOH terminus, respectively. Amino acid residues within squares are those with small side chain (Sm) in the transmembrane SmxxxSm motifs completely or highly conserved in LSHATs. Thus, with the 31 sequences available within the LSHAT subfamily (excluding the mammalian orphan members), the first motif in TM I and the motifs in TM VI and VIII are present in all of the sequences, the second motif in TM I is only absent in LAT-1 transporters, and the motif in TM VII is only absent in SPRM1. Of the 31 LSHAT subfamily members used to construct this figure, some correspond to D. melanogaster and C. elegans hypothetical proteins. The sequences of some of these proteins may be erroneous. Thus T32479 has an extremely long IL4, in T28818 the TM VIII is missing, T21445 and CG12317 are truncated sequences lacking the TM XII and the COOH terminus, and in CG9413 the NH2 terminus and the first part of TM I are missing.

Three catalytic residues (D199, E255, and D329 in O1,6G) and two residues for substrate binding (H103 and H328) within the TIM-barrel constitute the active site of family 13 glycosyl hydrolases (reviewed in Ref. 185). All these residues are in the COOH-terminal face of the TIM-barrel and, together with the protruding domain B, constitute the active site cleft. Of these TIM-barrel residues of O1,6G, only D199 in the A4 region, D329 between A7 and A7', and H105 before domain B can be identified unambiguously in the mammalian rBAT proteins, but none of them in the vertebrate 4F2hc proteins (Fig. 3). Similarly, none of the three invertebrate putative HSHAT proteins share all the active site residues of O1,6G (Fig. 3). In agreement with this, Wells and co-workers (187) did not observe -glucosidase activity for 4F2hc after expression in Xenopus laevis oocytes.

Janecek and co-workers (75) reported that domain B, including the COOH-terminal motif QPDLN (residues 167-171 in O1,6G), is conserved in the rBAT proteins but not in the 4F2hc proteins. Indeed, there is complete absence of domain B for 4F2hc and the putative HSHAT O45298 from C. elegans (Fig. 3). In contrast, rBAT and Q9XVU3 from C. elegans and Q9VHX9 from D. melanogaster contain a domain B with the structural features of this domain in O1,6G, including a complete or partial conservation of the COOH-terminal motif. This suggests that O45298 is the C. elegans ortholog of 4F2hc and that Q9XVU3 and Q9VHX9 are the C. elegans and D. melanogaster orthologs of rBAT, respectively. A search through the entire human genome and the genome of D. melanogaster and C. elegans revealed no other putative HSHAT proteins.

The COOH-terminal domain (domain C) of -amylases corresponds to a -barrel structure of eight antiparallel -strands folded in double Greek key motifs, which is distorted in the sixth strand C6 (185). Sequence alignment, threading, and secondary structure prediction fit the entire O1,6G domain C into the COOH-terminal region of the vertebrate and invertebrate HSHATs, with the exception of C6, which is not clearly predicted in these proteins (Fig. 3). The function of domain C, located far from the active site of O1,6G, is unknown. Finally, the last 30 amino acid residues of the rBAT proteins do not align with the -amylases.

Structure-Function Relationship Studies

Deora and co-workers (39) studied a series of COOH-terminal deletions of rat rBAT. Surprisingly, expression of these truncated proteins in oocytes yielded an unusual bimodal pattern of the induction of amino acid transport activity. Thus minimal COOH-terminal truncations (658-683, which eliminates the COOH-terminal tail, and 615-683, which eliminates the last four -strands of the domain B and the COOH-terminal tail; Fig. 3) abolished transport activity. The next mutants in the series (588-683, elimination from the last 6 -strands of domain B; Fig. 3) induced amino acid transport almost like that of the complete rBAT and with the characteristics of rBAT/system b^0,+. Further deletions (566-683, elimination from all of domain C, and 508-683, elimination from the last -helix of the TIM-barrel; Fig. 3) abolished amino acid transport induction. There is no obvious reason for the discrepancy of the amino acid transport induction by 508-683 rat rBAT (39) and 511-685 human rBAT (102). Deora and co-workers (39) studied further the transport-active deletion 588-683. The cysteine residue at position 111 in rBAT forms part of the disulfide bridge with the corresponding LSHAT. A mutation to the serine of this residue (C111S) renders a protein that induces 70% of the amino acid transport induced by wild-type rBAT. A notable difference is that the C111S mutant in 588-683 rBAT completely abolished its transport activity. This suggests the following. First, the formation of the disulfide bridge with the corresponding LSHAT is not necessary for the functional association with rBAT. This has also been demonstrated for 4F2hc alone and 4F2hc/y⁺LAT-1- or LAT-1-induced transport (47, 129, 175). Thus other interactions beside the disulfide bridge keep the functional transport complex intact. Second, lacking C111 (rat rBAT numbering), the 588-683/C111S mutant cannot form a stable complex with the endogenous LSHAT. Thus, in the absence of the disulfide bridge the integrity of domain B and the COOH-terminal tail is necessary for a functional transport complex. Third, the COOH terminus holds rBAT in an active conformation, perhaps by providing the sites for interaction with other rBAT regions or with the light subunit.

The above-mentioned experiments with truncated versions of rBAT could also be interpreted as the result of interactions, depending on the different COOH-terminal deletions, with different oocyte LSHATs. Indeed, Peter and co-workers (126) analyzed the mutations of the three conserved cysteine residues located in the COOH-terminal tail of rBAT (C664, C671, and C683 in rat rBAT; Fig. 4): replacement of C664 by alanine eliminates the functional interaction of rBAT with the putative endogenous b^0,+AT-type subunit and keeps (or causes) the functional interaction with a putative endogenous y⁺LAT-type subunit. Bröer and co-workers (19) studied COOH-terminal deletions of 4F2hc coexpressed in oocytes with LAT-1, LAT-2, or y⁺LAT-2. Surprisingly, association of these LSHATs requires different domains. Thus trafficking to the plasma membrane and induction of LAT-1/system L transport activity require only the NH₂-terminal tail, the TM domain of 4F2hc, and 30 extracellular amino acid residues, including the disulfide bridge-forming cysteine residue and the first -strand of domain B (Fig. 3) (the longest deletion studied in this work). In contrast, functional recognition of LAT-2 and y⁺LAT-2 needs the complete extracellular domain of 4F2hc. This suggests that the 4F2hc protein has different interaction sites for its associated light chains. In this study, all truncated versions of 4F2hc delayed the trafficking of LAT-1 to the plasma membrane. Moreover, this trafficking was more severely affected by truncations involving part of the extracellular glucosidase-like domain of 4F2hc than by those that eliminate it almost completely. Some short COOH-terminal truncations resulted in large aggregates that might be responsible for the severe trafficking defect. Finally, the more severe defects in LAT-1 recognition occurred when the last 70 amino acids were removed (i.e., the shortest deletion studied, which eliminates the last 6 -strands of domain B and the COOH-terminal tail; Fig. 3). This is reminiscent of the study by Deora et al. (39) of truncated versions of rBAT: the activity lost when only small parts of the COOH-terminal domain of the HSHAT are removed can be regained by larger COOH-terminal deletions. This suggests that the COOH-terminal tail of HSHATs plays a role in the proper folding of rBAT and/or in its interaction with the corresponding LSHAT.

In addition to the amino acid transport function of HSHATs, 4F2hc has been related to integrin function (see CELL PHYSIOLOGY OF THE CD98 COMPLEX). Fenczik and co-workers (49) have examined which domains of 4F2hc play a role in amino acid transport and in regulation of integrins function. By constructing chimeras with 4F2hc and the type II TM protein CD69, the authors showed that the NH₂ terminal and the TM domain of 4F2hc are required for its effects on integrin function, whereas the extracellular glucosidase-like domain is required for the stimulation of LAT-1 amino acid transport. This study together with that mentioned above on truncated versions of 4F2hc (19) point to multiple interactions, both at the NH₂-terminal tail and the TM domain, and at the extracellular glucosidase-like domain, between 4F2hc and LAT-1. Thus the NH₂-terminal tail and the TM domain are sufficient for the functional interaction of 4F2hc with LAT-1, but replacement of the extracellular domain by another prevents this functional interaction.

All but one of the rBAT missense mutations described in type I cystinuria (Fig. 3) are located in the extracellular glucosidase-like domain, which is consistent with its proposed role in amino acid transport. The most obvious role of HSHATs in amino acid transport is to help the trafficking of LSHATs to the plasma membrane. In agreement with this, for several cystinuria-specific rBAT mutations a trafficking defect to the plasma membrane has been substantiated (M467T, M467K) or suggested (T216M, S217R) (31, 143). On the other hand, some cystinuria-specific rBAT mutations affecting transport properties of system b^0,+ are presently under study, suggesting a participation of rBAT in the transport mechanism of the holotransporter. Reconstitution studies of HATs will be needed to demonstrate whether isolated LSHATs display amino acid transport activity and to identify the role of HSHATs in the mechanism of transport.

Another unsolved question is why the extracellular domain of HSHATs resembles that of glucosidases but without apparent catalytic activity. One could envisage that the noncatalytic glucosidase-like domain of HSHATs might hold extracellular glucidic structures to locate HATs properly in the plasma membrane. This question should be solved by purification and testing of glucid binding to the extracellular domain of HSHATs. Homology modeling of HSHATs with -glucosidases is in progress, but it could be hampered by the low identity of parts of domain A with crystallized -amylases and the lack of domain B in 4F2hc proteins. Purification and crystallization may be needed to establish the structure of the bulky extracellular region of HSHATs.

	LSHATs

TOP ABSTRACT INTRODUCTION THE MECHANISM OF EXCHANGE... INHERITED AMINOACIDURIAS HSHATs LSHATs CELL PHYSIOLOGY OF THE... CELL PHYSIOLOGY OF THE... REFERENCES

There are 23 reported LSHATs with ascribed amino acid transport function: 22 vertebrate sequences that correspond to 7 LSHAT paralogs and the Schistosoma mansonii SPRM1 protein (see the beginning of this study and legend for Table 1 for details). Amino acid sequence identity among these LSHATs ranges between 39 and 70% for different paralogs and between 85 and 98% for different mammalian orthologs. In addition, two other orphan LSHAT cDNAs have been cloned in mice (LSHAT-8 and -9) and humans (LSHAT-8) (Bassi MT, Gasol E, Zorzano A, Palacín M, and Borsani G, unpublished observations). Identity of the mammalian orphan LSHATs with the LSHATs with known transport functions drops to 23-29%.

LSHATs belong to the large superfamily of APC transporters (>175 transporters; for amino acids, polyamines, and organocations). Jack and co-workers (74) clustered this superfamily into 10 families. One of these is the LAT family (TC 2.A.3.8), which received its name from the first LSHAT identified (LAT-1; Refs. 79 and 101) and clusters the above-mentioned vertebrate LSHAT members and SPRM1, the yeast high-affinity methionine permease MUP1 (U40316) and the hypothetical yeast protein MUP3 (protein GenBank accession no. P38734) (72), and several hypothetical proteins from C. elegans and D. melanogaster. Jack et al. (74) also identified a signature sequence specific to the LAT family, G[WFY][DNFS]X[LIV][NH][FYT][LIVAGS][TALIV] [EGPS]E[LIVM]X[NDE]PX[RK][NT][LIVM][PK], where X represents any residue. This signature is located in the third putative intracellular loop of the topology model shown in Fig. 4 (see below for discussion of this model).

There is a key structural feature of LSHAT members, the conserved cysteine residue in the putative extracellular loop 2 that participates in the disulfide bridge with the corresponding HSHAT (Fig. 4), which was first functionally identified by Pfeiffer and co-workers (129). This cysteine residue is present in 34 LSHATs: 25 vertebrate sequences and SPRM1 (see above), D. melanogaster minidisks (AF139834), and 4 D. melanogaster (GenBank accession nos. CG1617, CG12317, CG6070, and CG9413) and 3 C. elegans (protein GenBank accession nos. T21445, T32479, and T28818) hypothetical proteins. In contrast, this residue is not conserved in the other LAT family members: yeast MUP1 and MUP3 and in 5 C. elegans (T15226, T16854, T24837, T32821, T31554) hypothetical proteins. We propose to cluster the former group of LAT family members (i.e., those with the conserved cysteine residue) in the LSHAT subfamily. Sequence analysis (173) revealed a specific signature sequence for this subfamily (34 transporters) located between transmembrane (TM) domain I and IL1 (see Fig. 4): [IVFL] G[SAT]GIF[VI][STA]P(X₂₆)[GS][AST][LYVIF][CSAV] [YFSN][AS]E[LI][GSA](X₅)SG[GA]X[YW]X[YF], where X represents any residue.

Phylogenetic analysis (173) of the LSHAT subfamily revealed a nodal relationship of CG12317 and CG9413 from D. melanogaster with LAT-1 and b^0,+AT transporters, respectively, suggesting that these hypothetical proteins correspond to the orthologs of these transporters. In contrast, the rest of the hypothetical proteins from C. elegans and D. melanogaster cluster in two nodes (T32479 and T21445 from C. elegans and CG1607 from D. melanogaster; T28818 from C. elegans and minidisks and CG6070 from D. melanogaster) together with transporters for neutral amino acids of the LSHAT subfamily (i.e., LAT-1, LAT-2, and asc-1). This sequence analysis is therefore not enough to ascribe the amino acid transport function to these C. elegans and D. melanogaster LSHAT members.

Membrane topology predictions revealed that the transporters of the APC family display 10, 12, or 14 TMs (74). Most of the APC families (8 families and the bacterial transporters of the CAT family) display 12 TMs. Figure 4 shows the 12-TM model for human b^0,+AT. This model is based on TM-HMM algorithms (164) with the multialignment of the 34 transporters of the LSHAT subfamily. To our knowledge, four studies deal with the experimental determination of the topology of APC transporters: LysP [amino acid transporter (AAT) family, specific for bacteria (46)]; PotE [basic amino acid/ polyamine antiporter (APA) family, specific for bacteria] (83); AroP (AAT family) (35); and the B subunit of glyceraldehyde-3-phosphate dehydrogenase (GABP; AAT family) (63). All of these studies used the strategy of transporter-monitor enzyme hybrids as topological sensors. The four studies obtained results compatible with the 12-TM model with the NH₂ and COOH termini located intracellularly. There is agreement on the location of the amino acids tested in these transporters and the membrane topology model for LSHATs shown in Fig. 4 for b^0,+AT. In contrast, Isnard et al. (72) proposed a 13-TM model for MUP1 and MUP3. Sequence alignment between these two yeast transporters and LSHAT transporters could be traced in their whole sequences, and TM VIII in the 13-TM model does not show enough hydrophobicity in any of the sequences. This strongly supports the 12-TM model for the LSHAT transporters, but direct experimental evidence is lacking.

Structure-Function Relationship Studies

The location of the 22 cystinuria-specific mutations affecting single amino acid residues of b^0,+AT is shown in Fig. 3. Seventeen mutations involve amino acid residues within the putative TM domains of the protein and the rest within the putative intracellular loops. Thus none of the mutations is located within the putative extracellular loops of b^0,+AT. This is also the case for the seven missense LPI mutations mentioned above. Similarly, all the transport activity-relevant residues identified in PotE are located in the loops and TM facing the cytoplasmic site of the transporter (82, 83). It has also been shown that the main functional amino acids of the lactose/H⁺ symporter (76, 132) and the metal tetracycline/H⁺ antiporter (192) are located on the cytoplasmic side of the protein. Kashiwagi and co-workers (82) interpreted this asymmetry as the structural basis to provide a quick response of the transporter to any change in cellular substrate concentration.

Six missense b^0,+AT mutations (A70V, V170M, A182T, A354T, G105R, and R333W) have been tested for function after coexpression with rBAT in HeLa cells (49a). Some of these mutations (V170M, A354T, G105R, and R333W) cause a complete or almost complete loss of function (<10% residual transport activity), whereas the others (A70V and A182T) have only a partial effect (>50% residual activity) (Table 2). It remains to be determined whether these b^0,+AT mutations affect trafficking to the plasma membrane or whether they inactivate the transporter.

Urinary excretion of cystine and the three dibasic amino acids (arginine, lysine, and ornithine) in heterozygotes bearing the major missense SLC7A9 mutations (G105R, V170M, A182T, R333W) showed mutation-specific severity (49a). Thus A182T heterozygotes are associated with a mild phenotype (urine cystine and dibasic amino acid levels within 5 times the average of the urinary excretion values in controls). In contrast, heterozygotes bearing the other common mutations show higher urinary excretion values (8-18 times higher than controls). This parallels the degree of the transport defect associated with these mutations (Table 2). If this correlation is extended to uncommon cystinuria-specific b^0,+AT mutations (Table 2), several implications may be extracted (49a). First, b^0,+AT and y⁺LAT-1 mutations affecting highly conserved amino acid residues within the LSHAT subfamily result in severe amino acid transport defects and/or severe urinary phenotypes in heterozygotes, whereas mutations in nonconserved residues are associated with mild phenotypes. The only exception is the mild urinary excretion phenotype found in two of the three G63R heterozygotes analyzed; individual variability might explain the behavior of these heterozygotes (49a). Second, four b^0,+AT or y⁺LAT-1 mutations are located within putative TM and involve residues with a small side chain (Gly, Ala, or Ser) in all the LSHAT transporters. These four mutations are associated with a severe phenotype. In contrast, mutations A70V, A126T, and A182T, which involve residues in putative TM with different side chain sizes in other LSHAT transporters, are associated with mild phenotypes.

The relevance of small-side chain residues (Gly, Ala, or Ser) in contact regions of -helix TMs has recently been highlighted. Thus highly conserved residues with a small side chain (Gly or Ala) are present in the contact regions of the TM -helices of aquaporin-1 (AQP1) (108). Moreover, the GlyxxxGly motif (where x represents any residue and Gly can be replaced by Ser) has been reported to be a framework for the TM helix-helix association (142), and in this sense, helixes 3 and 6 of AQP1 contain a GlyxxxGly motif (where Gly can be replaced by Ala). Similarly, SmxxxSm motifs (where Sm stands for residues with a small side chain: Gly, Ala, or Ser) are present in the LSHAT subfamily in TM I, VI, VII, and VII (Fig. 4). Mutation G259R in b^0,+AT and mutation G54V in y⁺LAT-1 involve SmxxxSm helix-helix association motifs in TM VII and I, respectively. G259R is associated with a severe urinary phenotype in heterozygotes, and G54V is associated with a dramatic loss of transport function (49a, 109). This suggests that residues with small side chains, which are conserved in TMs of LSHAT transporters, may participate in their TM helix-helix associations.

	CELL PHYSIOLOGY OF THE CD98 COMPLEX

TOP ABSTRACT INTRODUCTION THE MECHANISM OF EXCHANGE... INHERITED AMINOACIDURIAS HSHATs LSHATs CELL PHYSIOLOGY OF THE... CELL PHYSIOLOGY OF THE... REFERENCES

In this section, we will use the name CD98 to refer to the 4F2 complex and CD98hc for the 4F2hc, to maintain the nomenclature as used in the papers cited. The light chains are termed LSHATs, as before.

As stated above, LSHATs are believed to mediate amino acid transport itself due to their polytopic structure, whereas the type II protein CD98hc seems to act as a guidance molecule for the light chains on their way to the plasma membrane (110). However, localization and maturation of CD98hc in the absence of light chains are not well studied. At least in oocytes and L cells, CD98hc is expressed at the plasma membrane without a light chain (101, 172). Overexpression of CD98hc leads to its expression, apparently as a monomer, at the plasma membrane (47, 49). However, no study has measured the amount of plasma membrane free CD98hc in the normal in vivo situation (for instance, in activated T cells, proximal tubule cells, intestinal epithelial cells, etc.), which is important to an understanding of some of the data that will be discussed below.

CD98 was identified in the 1980s as an early activation antigen of T and B cells (60, 99, 134). Since those studies, many different functions besides amino acid transport have been ascribed to the complex, from cellular activation and division (41, 167, 191) to cell adhesion, fusion, and differentiation (114, 117, 170, 182). A recent careful review from Devés and Boyd (40) offered the first systematic overview of the huge amount of data dealing with these apparently disconnected functions. In this section, we will focus on the recent work performed in the fields of integrin activation, viral and cell fusion and differentiation, T cell activation, and oncogenesis.

Integrin Activation

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Complementation of dominant supression by CD98hc. A: expression cloning strategy used a Chinese hamster ovary cell line stably expressing a chimaeric integrin () constitutively activated (i.e., in a conformation that binds ligand L with high affinity). The activation is thought to occur through the binding to cytoplasmic integrin tails of a putative "integrin activating complex" (IAC). B: overexpression of a TAC- tail chimera at the plasma membrane (1) leads to a dose-dependent suppression of integrin activation, probably by titration of the IAC. C: overexpression of the CD98hc (2) leads to the complementation of the dominant supression caused by the TAC- chimera. The authors showed that CD98hc is able to bind -integrin tails (49, 194), which might be a possible mechanism for this complementation. Stoichiometry of Tac- chimeras and Tac--CD98hc has not been determined experimentally. Adapted from Ref. 48.

The same laboratory went further in the characterization of CD98hc-mediated CODS (194). They found that solubilized CD98 binds in vitro to _1A- and ₃-integrin tails (but not to _1D or ₇). This binding correlated with the lack of CD98hc-mediated CODS caused by _1D- or ₇-integrin cytoplasmic tails. _1A-Cytoplasmic tail deletion mutants that still bound CD98hc were no longer able to mediate dominant suppression (194), suggesting that the mechanism of dominant suppression by integrin cytoplasmic tails is due to mechanisms other than titration of endogenous CD98hc molecules. The role of the different domains of the CD98hc molecule in CODS, integrin cytoplasmic tail binding, and amino acid transport induction was also investigated by using chimeric constructs (see HSHATS) (49). The cytoplasmics tail and TM and extracellular domains of CD98hc and CD69 (another type II membrane protein) were exchanged. The cytoplasmic and the TM of CD98hc were necessary and sufficient for CODS or integrin binding, whereas the extracellular domain was dispensable for these functions but essential for amino acid transport. Overexpressed wild-type and cysteine mutants of CD98hc (including the one that partially impairs amino acid transport and light chain association) were localized in monomeric form at the plasma membrane and mediated CODS and bound integrin cytoplasmic tails. Interestingly, the authors pointed out that coclustering of CD98 molecules and integrins might also help to localize amino acid transport in defined regions of the plasma membrane. For instance, _1A-integrin is basolateral in epithelial cells (193), where CD98 is also found.

Together, these studies showed that CD98hc binds some, but not all, -integrin tails and can perform CODS independently of amino acid transport and LSHAT. It should be noted, however, that CODS depends on the overexpression of CD98hc molecules that most likely act by titration of excess -integrin tails, exerting the dominant suppressive effect. Whether integrin activation per se (not CODS or -integrin tail binding) requires association between CD98hc and LSHAT proteins is a matter of speculation (see Oncogenic Potential of CD98).

Viral and Cell Fusion and Differentiation

In the early 1990s, Ito and colleagues (73) isolated monoclonal antibodies (MAbs) that enhanced cell fusion of the Newcastle disease virus. These MAbs immunoprecipitated either CD98hc or the  ₃-integrin subunit (114), named fusion regulatory protein-1 and -2, respectively. Since then, the aforementioned authors have extended their studies to paramyxovirus-induced syncitium cell formation and human immunodeficiency virus entry into monocyte/macrophage cells (115, 117), isolating new anti-CD98hc MAbs that are able to either induce or inhibit those processes. The fact that different anti-CD98hc MAbs have opposing effects on fusion events suggests that the MAbs can induce or fix conformations of the extracellular domain of CD98hc, competent or not to transduce cell fusion signals into the cell. Moreover, in regard to the work of Ginsberg and co-workers (48, 49, 194), it is worth mentioning that fusion induced by anti CD98hc mAbs was blocked by fibronectin and anti-₁-integrin antibodies, indicating functional and/or physical interactions between the integrin system and CD98hc.

The role of CD98hc was further investigated by the generation of stable cell lines expressing a chimera in which the cytoplasmic domain of human CD98hc was replaced by the cytoplasmic domain of the hemaglutinin-neuraminidase from the human parainfluenza virus type 2 and the mutant C330S [which does not impair either disulfide linking to the LSHATs or amino acid transport (129)]. Both mutants suppressed the cell fusion-enhancing activity of anti-CD98hc MAbs (117). The mechanisms of these dominant-negative effects are not known. If LSHATs are required for cell fusion, the mutant CD98hc may titrate them in a way not competent for fusion. The conformation of the CD98hc extracellular domain may also change to that of a fusion-incompetent molecule. The binding to the MAbs, however, was not affected. Titration of -integrin cytoplasmic tails (194) by the C330S mutant could also explain the dominant-negative effect, but this mechanism cannot be invoked for the chimera.

More recently, Ito's group (62) has concentrated its efforts on the osteoclast differentiation pathway. Anti-CD98hc MAbs induced homotypic cell aggregation and multinucleated giant cell formation of monocytes without any other fusogen (62). These polykaryocytes displayed several (but not all) of the exclusive features of the osteoclasts. Two recent reports have begun to delineate monocyte-to-osteoclast signaling pathways elicited by the anti-CD98hc MAbs. The first (103) highlights the importance of induction of c-src expression and activation by the MAbs. This protein kinase is widely believed to play a role in cell differentiation (17). Targeted disruption of the c-src gene causes a form of osteopetrosis whereby osteoclasts are present but inactive (165), indicating that c-src is involved in osteoclastogenesis. Transcription of c-src is dependent on Sp1, which is also upregulated by anti-CD98hc MAbs. The use of a panel of inhibitors suggested the involvement of a tyrosine kinase-Ras-Map kinase-Sp1 pathway in the anti-CD98hc MAb induction of c-src in monocytes. As expected, an anti-CD98hc MAb that inhibited polykaryocyte formation (in the presence of an anti-CD98hc active MAb) also suppressed Sp1 and c-src expression.

The second report links the two known routes of osteoclastogenesis, the CD98-mediated pathway and that mediated by the osteoclast differentiation factor (ODF; a member of the cytokine family) (106) by showing that the latter is suppressed by an inhibitory anti-CD98hc MAb and that the former is inhibited by osteoclast inhibitory factor (a secreted member of the tumor necrosis factor receptor family), the classic inhibitor of the ODF-mediated pathway. The expression of the ODF receptor increases on incubation with the active anti-CD98hc MAbs.

The authors did not investigate the role of the integrin system in osteoclast differentiation. One might speculate that similar results to those seen in viral fusion would have been observed, suggesting the intimate relationship between CD98 and integrins. In this sense, Suga and colleagues (168) reported recently that crosslinking of CD98 by MAbs mediated cell aggregation and adhesion of lymphocytes, most likely by increasing the avidity of the _L2- integrin for intercellular adhesion molecule. The activation was dependent on phosphatidylinositol (PI)3-kinase and on the persistent activation of the Ras-related small GTPase Rap1. Moreover, in vitro fertilization of murine eggs (which express CD98 on the surface) is inhibited by anti-CD98hc antibodies. Fertilization is also inhibited by the recombinant soluble disintegrin domain of A disintegrin and metalloprotease-3 protein (10, 195), which seems to interact with ₁-integrins on the egg surface. This interaction is also inhibited by anti-CD98 antibodies (170).

It becomes evident from the above results that the anti-CD98hc MAbs could somehow mimic natural ligands for this protein, involved in cell fusion and adhesion processes. CD98hc is a possible receptor for galectin-3 (44), a 26-kDa -galactoside binding protein of the galectin family (65). This protein is secreted by monocytes/macrophages (151) and epithelial cells (96) and may have a role in cell-cycle control, prevention of T cell apoptosis, activation of several cell types, including lymphocytes and monocytes/macrophages, and as a mediator of cell-cell and cell-extracellular matrix adhesion (67, 86, 97). Very recently, galectin-3 has been shown to be a chemoattractant for monocytes and macrophages (144) and to induce uptake of extracellular calcium in T cells (43). More studies are needed to define the roles of galectin-3-CD98hc interactions and to identify other possible ligands of the CD98 complex.

T Cell Activation

Oncogenic Potential of CD98