Sorting Determinants in the Transmembrane Domain of p24 Proteins*

Members of the p24 family of putative cargo receptors are proposed to contain retrograde and anterograde trafficking signals in their cytoplasmic domain to facilitate coat protein binding and cycling in the secretory pathway. We have analyzed the role of the transmembrane domain (TMD) of a p24 protein isolated from COPI-coated intra-Golgi transport vesicles. CD8-p24 chimeras were transiently expressed in COS7 cells and analyzed by immunofluorescence and pulse-chase experiments. The localization and transit of the wild-type chimera from the endoplasmic reticulum (ER) through the Golgi complex involved a glutamic acid residue and a conserved glutamine in the TMD. The TMD glutamic acid mediated the localization of the chimeras to the ER in the absence of the conserved glutamine. Efficient ER exit required the TMD glutamine and was further facilitated by a pair of phenylalanine residues in the cytoplasmic tail. TMD residues of p24 proteins may mediate the interaction with integral membrane proteins of the vesicle budding machinery to ensure p24 packaging into transport vesicles.

Members of the p24 family of putative cargo receptors are proposed to contain retrograde and anterograde trafficking signals in their cytoplasmic domain to facilitate coat protein binding and cycling in the secretory pathway. We have analyzed the role of the transmembrane domain (TMD) of a p24 protein isolated from COPI-coated intra-Golgi transport vesicles. CD8-p24 chimeras were transiently expressed in COS7 cells and analyzed by immunofluorescence and pulse-chase experiments. The localization and transit of the wild-type chimera from the endoplasmic reticulum (ER) through the Golgi complex involved a glutamic acid residue and a conserved glutamine in the TMD. The TMD glutamic acid mediated the localization of the chimeras to the ER in the absence of the conserved glutamine. Efficient ER exit required the TMD glutamine and was further facilitated by a pair of phenylalanine residues in the cytoplasmic tail. TMD residues of p24 proteins may mediate the interaction with integral membrane proteins of the vesicle budding machinery to ensure p24 packaging into transport vesicles.
Transport in the secretory pathway is mediated by a vesicular carrier mechanism that allows the cells to preserve organelle identity by selectively packaging cargo molecules destined for secretion (1,2). Trafficking between the ER 1 and the Golgi apparatus as well as intra-Golgi transport (1-4) employs coatomer, a complex of seven subunit proteins (5). The COPI coat is comprised of coatomer and the GTPase ADP-ribosylation factor (6,7). These two cytosolic proteins are sufficient to pinch off COPI-coated vesicles from Golgi membranes in a cell-free reaction (8). Anterograde-and retrograde-directed COPI-coated vesicles bud from every cisternae of the Golgi complex in vivo (9).
The sorting of secreted proteins requires adaptor molecules to link cargo to the cytoplasmic vesicle budding machinery.
Cargo receptors, possibly including p24 proteins (10 -16) and VIP36-like lectins (17)(18)(19), are proposed to move bidirectionally in the secretory pathway to carry out their function (1,2). The p24 proteins have previously been identified as components of COPI-and COPII-coated vesicles in mammals and yeast, respectively (12)(13)(14). They form a large family of type I integral membrane proteins with a characteristically short cytoplasmic tail, a predicted coiled-coil domain juxtaposed to the TMD in the luminal/exoplasmic region, and a highly variable NH 2terminal domain (10 -16, 20). Based on the analysis of yeast strains lacking the p24 members Emp24p and/or Erv25p, it was suggested that p24 proteins might be involved in the packaging of cargo molecules into vesicles as well as vesicle budding (13)(14)(15). Furthermore, yeast strains lacking Emp24p were found to be defective in the retention of ER-resident proteins (21) consistent with a role of p24 proteins in controlling the fidelity of cargo recruitment into budding vesicles.
We have previously shown that conserved phenylalanine residues in the cytoplasmic domain of p24 proteins mediate the interaction with coatomer and are likely to be involved in anterograde trafficking of p24s from the ER through the Golgi complex (12,20). As a first step toward the identification of additional components of the vesicle budding machinery we have further characterized the transport signals of p24 proteins. A charged amino acid in the p24 TMD appears to serve as a retention signal that is modulated by other TMD residues and phenylalanines in the cytoplasmic domain.

EXPERIMENTAL PROCEDURES
Materials-The OKT8 CD8 monoclonal antibody was obtained from Ortho Diagnostic Systems (Raritan, NJ), the goat anti-mouse fluorescein-conjugated secondary antibody from Molecular Probes (Eugene, OR), and Lipofectin and LipofectAMINE TM from Life Technologies, Inc. COS7 (SV40-transformed African green monkey kidney) cells were purchased from the American Type Culture Collection (Rockville, MD), protein G-agarose from Boehringer Mannheim, and [ 35 S]methionine/ cysteine from ICN.
Construction of CD8 Chimera-The CD8 chimera were constructed by the polymerase chain reaction such that the 165-amino acids of the human CD8 extracellular domain were preserved (20). Codon 166 was changed to glycine to introduce a unique ApaI restriction site, followed by a conserved proline, a stop codon, and a EcoRI site (CD8-C1). Oligonucleotides (Gene Link, Thornwood, NY) coding for the COOH-terminal 34 amino acids of chop24a (RVVLWSFFEALVLVAMTLGQIY-YLKRFFEVRRVV) and variants thereof (see Fig. 1B), preceded by an ApaI site and followed by a stop codon and EcoRI site, were annealed, subcloned into the CD8 construct, and inserted into the pECE vector (22). Sequences were verified by DNA sequencing using the Sequenase DNA sequencing kit (U. S. Biochemical Corp.).
Localization and Pulse-Chase Analysis-Transfection of COS7 cells was carried out with Lipofectin and LipofectAMINE for immunofluorescence and pulse-chase analysis, respectively, according to the manufacturer's instructions. The chimeras were analyzed 46 h after transfection. For localization by immunofluorescence, the cells were incubated in medium containing 10 g/ml cycloheximide for 2 h prior to fixation. They were permeabilized with Triton X-100 before labeling with the OKT8 CD8 monoclonal antibody (dilution of 1/50) and goat anti-mouse fluorescein-conjugated secondary antibody (dilution of 1/200) as described (17) and viewed and photographed with an Axiophot photomicroscope (Carl Zeiss, Oberkochen, Germany). For pulse-chase analysis COS7 cells were labeled for 20 min with [ 35 S]methionine/ cysteine and then incubated for 0, 15, 30, 45, 60, and 120 min in medium containing unlabeled methionine/cysteine. The cells were lysed in Triton X-100, and the CD8 chimeras were isolated by immunoprecipitation (23) with protein G-agarose before SDS-polyacrylamide gel electrophoresis (12% gel) (24).

RESULTS
The p24 family of proteins currently comprises 8 members in mammals and yeast and various orthologues and homologues in other species (Fig. 1A) (10 -16). The alignment of p24 TMDs and cytoplasmic tails showed that in addition to one absolutely conserved phenylalanine in the cytoplasmic domain (position 195 in chop24a, the p24 protein isolated from CHO cells (13)), other amino acids in the TMD are conserved. First, a glutamine residue is present in the TMDs adjacent to the TMD-cytoplasmic domain border in all p24s (position 187 in chop24a). Second, a position juxtaposed to the exoplasmic domain (position 176 in chop24a) accommodates a glutamic acid or polar residue in most p24 proteins with only some exceptions in yeast (Fig.  1A). In a helical wheel presentation, the TMD glutamic acid and glutamine (but not the strictly conserved phenylalanine in the cytoplasmic domain) line the same side of the predicted ␣-helical structure (data not shown), revealing a relatively nonhydrophobic face in the TMD of p24 proteins.
To analyze the role of the conserved residues on p24 localiza-tion and trafficking, CD8 chimeras were generated in which the CD8 TMD and cytoplasmic domain were replaced with the corresponding wild-type and mutant sequences of chop24a ( Fig. 1B) (13). The CD8 protein has been used previously to analyze cytoplasmic domain and TMD trafficking signals (25)(26)(27). CD8-chop24a chimeras were localized by immunofluorescence in transiently expressing COS7 cells (Fig. 2). When the glutamic acid residue alone (EA, panel A), in combination with the conserved glutamine (EA-QA, panel C) or phenylalanines (EA-FFAA, panel E), or all three positions (EA-QA-FFAA, panel F) were replaced with alanine, the hybrid proteins were predominantly detected at the cell surface and in a juxtanuclear area, just as the wild-type chimera had been localized (cf. the previous analysis (20) and panel G). In contrast, when the glutamine residue alone (QA, panel B) or in combination with the phenylalanines (QA-FFAA, panel D) was replaced with alanine, the chimeras were localized to the nuclear envelope and tubular-reticular structures, presumably the ER (Fig. 2). These results suggest that the presence of a glutamic acid in the TMD of p24 proteins confers localization to the early secretory pathway when the conserved TMD glutamine residue is absent.
To further corroborate this finding we performed pulse-chase experiments to measure the rate and extent of transport of the chimeras from the ER through the Golgi complex (Fig. 3A). The time required to receive O-glycans (attached to the CD8 exoplasmic domain) that were processed to the mature (sialic acid-containing) form in the medial-or trans-Golgi (28) was   FIG. 1. The p24 proteins and CD8-chop24a chimeras. A, the alignment of the COOH-terminal residues of p24 proteins was generated with the GCG program pileup (61) using the available sequences of 16 previously described p24 members (13,20), the T1/ST2-binding protein (10), and the novel entry hp24g and yp24h. The expressed sequence tags coding for hp24g and yp24h were retrieved by searching the NCBI dbest data base with the program Powerblast using tblastn (available from ftp://ncbi.nlm.nih.gov/pub/sim2/PowerBlast). The prefix a refers to Arabidopsis thaliana, h to human proteins, y to S. cerevisiae, chop24a is from CHO cells (13), gp25l is from dog (11), p23 is from rabbit (12), T1/ST2BP is from human, and Emp24p (14) and Erv25p (15) are from S. cerevisiae. The amino acid sequences of the TMD and cytoplasmic domain are indicated in single-letter code. The conserved phenylalanine/hydrophobic residues in the cytoplasmic domain, the conserved glutamine, and the frequently charged/polar residue in the TMD are boxed. GenBank/EMBL accession numbers are as follows. ap24a, Z34726; ap24b, T46519; chop24a, U26264; gp25l, X53592; p23, X98303; hp24a, X92098 (16); hp24b, T48838, R25915, T17481; hp24d, T98284; hp24e, F07445; hp24g, assembled from AA215037 and AA138141; Emp24p, X67317; yp24b, L22015; yp24c, U00059; yp24d, L22015; yp24e, X87331 and T36996; yp24f, Z48432; Erv25p, Z49810; yp24h, Z72524. B, summary of CD8-chop24a chimeras. CD8 chimeras were constructed such that the 165 amino acids of the human CD8 extracellular domain were preserved, followed by the amino acids glycine-proline and the 34 residues of the chop24a TMD and cytoplasmic domain. The wild-type chimera (wt) (20)  The localization of the wild-type and FFAA chimera have been described previously (20). Untransfected cells are shown as a control in H. The cells were incubated in medium containing 10 g/ml cycloheximide for 2 h prior to fixation. They were permeabilized with Triton X-100 before labeling with an antibody to CD8 and a secondary antibody. A set of representative cells are shown. The exposure time and printing were identical for A-H. Bar, 15 m. similar for the EA, EA-QA, and EA-FFAA mutant chimeras and comparable to the wild-type form (Fig. 3B) (cf. Ref. 20). The QA and QA-FFAA mutant hybrid proteins, however, were not significantly processed within the 2-h chase period (Fig. 3).
Charged residues in the TMD confer efficient degradation of monomeric subunits of the T cell antigen receptor in the absence of oligomerization (29 -31). But as expected from the structural context of the chop24a glutamic acid residue (i.e. TMD length and positioning) (31,32), mutant CD8-chop24a chimera retained in the ER were not rapidly degraded since more than 70% of the starting material was present after a 2-h chase for all hybrid proteins.
Unexpectedly, the EA-QA-FFAA chimera was processed to the mature form at a substantially higher rate than all other chimeras, including the wild-type CD8-chop24a form (Fig. 3B) (cf. Ref.  20). Apparently, the replacement of all three positions allowed efficient transit from the ER through the Golgi complex.
Next, we tested whether the chop24a TMD alone alters the trafficking of the reporter molecule CD8 by replacing the CD8 TMD with the respective sequence of chop24a (data not shown). Pulse-chase experiments showed that the rate of transport of the CD8-chop24a chimera was decreased relative to the CD8 wild-type protein. DISCUSSION Sorting signals in the TMD of mammalian and yeast proteins are important for their localization to the Golgi complex and the ER (27,(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45), as well as for endocytosis and apical delivery in polarized epithelial cells (46,47). Recycling of the KDEL receptor and Sec12p from the Golgi to the ER and retention of the coronavirus E1 protein in the Golgi are likely to require polar residues in the TMD of the proteins (36,38,42,48). For some proteins, trafficking determinants have been found in the TMD as well as in the cytoplasmic domain (␣2,6-sialyltransferase, Sed5p, Sec12p, and TGN38) (27,34,37,42). The chop24a member of the p24 family of putative cargo receptors (13) encodes for a glutamic acid in the TMD. This residue appears to serve as an ER retention signal which is attenuated by a conserved glutamine in the TMD and phenylalanine residues in the cytoplasmic tail (Fig. 2). The presence of all three motifs facilitates efficient ER exit and transit through the Golgi complex.
Unexpectedly, a EA-QA-FFAA CD8-chop24a hybrid protein that is devoid of all analyzed sorting determinants was delivered to the medial-/trans-Golgi at a significantly higher rate than the wild-type chimera and other hybrid proteins (Fig. 3). Apparently, the presence of phenylalanine residues in the cytoplasmic domain decreases the rate of transit from the ER through the Golgi complex when the TMD glutamic acid and glutamine are absent. Conversely, when the TMD glutamic acid and glutamine are present, the phenylalanine residues increase the rate of transport (20). Moreover, the presence of the TMD glutamine residue decreases the rate of transport when the TMD glutamic acid and the phenylalanines are replaced with alanine. Taken together, these observations suggest that the signals present in the TMD and cytoplasmic domain modulate each other. The TMD determinant appears to be transplantable since the exchange of the CD8 TMD alone with the chop24a TMD significantly decreases the rate of transport from the ER through the Golgi complex (data not shown).
For a variety of proteins the role of charged amino acids in the TMD has been analyzed in detail. In subunits of the T cell antigen receptor they are involved in quality control of heterooligomer assembly by the formation of intramembrane charge pairs (29,49) and in promoting the formation of intramembrane disulfide-linked dimers (50). Since none of the currently known p24 members contains basic residues or cysteines in the TMD, similar mechanisms for p24 oligomerization seem unlikely. Nevertheless, the recent analysis of the yeast members Emp24p and Erv25p demonstrated that they can be efficiently cross-linked to each other (15) suggesting that p24 proteins may form hetero-oligomers. It has not been investigated whether p24 proteins can also form homo-oligomers.
Several models could account for the differential localization and trafficking of the CD8-chop24a chimeras. It has recently been shown that similarly charged (glutamic acid) or polar (glutamine) TMD residues can promote homodimerization and activation of the receptor tyrosine kinase p185 neu (51). In the context of p24 proteins this could imply that the TMD is involved in oligomerization. If oligomerization of p24 proteins (and CD8-chop24a chimeras) is required for transport from the ER through the Golgi complex, then the TMD alone is apparently not sufficient for rapid transit. The FFAA chimera in which the phenylalanines in the cytoplasmic domain were re-  15,30,45,60, and 120 min in medium containing unlabeled methionine/cysteine. Then, cells were lysed in Triton X-100, and the CD8 chimeras were isolated by immunoprecipitation and analyzed by SDS-PAGE. The CD8-chop24a precursor (p) and mature (m) O-glycosylated forms (28) are indicated. Immunoprecipitation from nontransfected cells is shown as a control (Cont.). The wild-type and FFAACD8-chop24a chimeras have been analyzed previously (20). B, quantitation of the pulse-chase analysis. The amounts of the mature O-glycosylated form (28) of each chimera were determined by densitometric scanning of autoradiograms. The data points for the EA-QA-FFAA and EA-QA CD8-chop24a chimeras represent average values from two experiments, for the other hybrid proteins the analysis was carried out once. The CD8-chop24a wild-type and FFAA mutant chimeras (20) are included for comparison. placed with alanine ( Fig. 3 (20)) was transported from the ER through the Golgi complex at a significantly lower rate than the wild-type form. Thus, rapid transit requires the phenylalanine residues to aid oligomerization and/or to facilitate interaction with coatomer (20). Although this model could explain the modulatory interactions among TMD and cytoplasmic tail signals it does not account for the highly efficient transport of the EA-QA-FFAA CD8-chop24a chimera. Moreover, this premise does not provide an explanation for the apparent decreased rate of transit of CD8-chop24a chimeras that retain only one of the three positions, i.e. the TMD glutamine residue (EA-FFAA), the glutamic acid (QA-FFAA), or the phenylalanine residues in the cytoplasmic domain (EA-QA).
Precedence for an alternative model is provided from the analysis of Sec12p in yeast. Two glutamines in the Sec12p TMD were shown to be required for the recycling of escaped protein from the Golgi complex back to the ER (42). Proper Sec12p ER localization required Rer1p, a protein primarily localized to the Golgi complex (40 -42, 52). Rer1p contains four TMDs that encode for polar as well as charged residues and is proposed to function as a TMD sorting receptor. A direct interaction with Sec12p remains to be shown.
Analogous to Sec12p and Rer1p, p24 recruitment could involve not only the interaction with coatomer (20) but also the binding to other integral membrane proteins of the vesicle budding machinery. These additional protein-protein interactions could ensure the availability of essential machinery components at the budding site (e.g. v-SNAREs (53)) and might be mediated by the TMDs of the interacting proteins. Upon interaction with coatomer the p24 proteins would be released into the budding vesicle. Alternatively, efficient recruitment might require the simultaneous interaction of coatomer with both factors which could either trigger dissociation of p24 proteins for recruitment or lead to the recruitment of both components into the budding vesicle.
With respect to our analysis of the CD8-chop24a chimeras the second model predicts that the absence of any trafficking signal in the TMD and a coatomer binding motif (e.g. phenylalanines or basic residues) would allow unregulated, passive inclusion into vesicles not restricted by the limited number of coatomer binding sites. This could explain the more rapid transport of the EA-QA-FFAA versus the EA-QA form of the CD8-chop24a chimeras. Similarly, the absence of a TMD signal in a p24 protein (EA-QA-FFAA chimera versus FFAA chimera) could abolish the interaction with a negative regulatory component and effectively increase the rate of transport by bulk flow.
It is possible that hitherto unidentified factors, or members of a growing class of multispanning membrane proteins (54 -60), are involved in the regulation of p24 packaging into budding vesicles.