Membrane Cofactor Protein (CD46) of Complement PROCESSING DIFFERENCES RELATED TO ALTERNATIVELY SPLICED CYTOPLASMIC DOMAINS*

Membrane cofactor protein (MCP, CD46), a widely dis- tributed regulatory protein, inhibits complement activation on host cells and serves as a measles virus recep- tor. Most cells express four isoforms (with one of two cytoplasmic tails, CYT-1 or CY”-2). Previously, we noted that MCP precursors had variable intracellular processing. Therefore, we characterized the intracellular trans- port of individual MCP isoforms. Transfectants were used for pulse-chase analyses. MCP isoforms bearing CY”-1 chased into their mature, surface forms with a half-life (t,J of 10-13 min while those with CYT-2 re- quired 35-40 min. The precursor of a tail-less mutant possessed a t,,+ of 160-166 min. Chimeras were con- structed that added both tails in opposite orientation onto the isoform (Le. CYT 1+2 or CY” 2+1). Chimera 1+2 precursor processed with a t, of 35-37 min, similar to CY”-2. Chimera 2+1 had a t, of 15-19 min, more closely resembling CY”-1. Thus, in both cases the carboxyl-ter- minal tail controlled the processing rate. Deletions were made in the beginning, middle, and carboxyl terminus of CY”-1. Deletion of the first or middle six amino acids had no effect on the processing rate. However, deletion of the terminal tetrapeptide

Membrane cofactor protein (MCP,' CD46) is a widely distributed, regulatory protein of the complement system that inhibits complement activation on host cells (reviewed in Ref. 1). MCP acts as a cofactor in concert with plasma serine protease factor I to proteolytically degrade C3b and C4b deposited on self-tissue (2-4). MCP performs this role intrinsically in that i t primarily protects the cell on which it is anchored, not bystander cells (5). MCP belongs to a family of structurally, functionally, and genetically related proteins, termed the regulators of complement activation (6). Other members of this multigene family are decay-accelerating factor (CD551, complement receptors 1 (CD35) and 2 (CD21), and two plasma proteins, C4 binding protein and factor H. Of interest, complement receptor 2 serves as the Epstein-Barr virus receptor (7-9), and recently MCP has been shown to be a measles virus receptor (10-12).
MCP is expressed on nearly every cell examined (except erythrocytes) and displays an unusual pattern on SDS-PAGE * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
in that it appears as a broad, heterogeneous doublet with a molecular mass of 58-68 and 48-56 kDa. The cDNA cloning (13) and genomic organization of MCP (14) have provided an explanation for this electrophoretic characteristic. On most hum a n cells MCP consists of a family of at least four isoforms that arise by alternative splicing of a single gene (14) (Fig. 1). The amino terminus of the extracellular portion contains four of the C3bIC4b-binding modules known as short consensus repeats (1, 15-18). Following this area is a region enriched in serines, threonines, and prolines (STP), a site of extensive 0-linked glycosylation. The STP region consists of 14 or 29 amino acids, depending on whether STP exon B (15 amino acids) is spliced out. The presence of STP-B produces the higher molecular weight protein isoforms while its absence generates the lower molecular weight forms of MCP (1,14). The STP region is followed by a tract of 12 amino acids of unknown significance that is encoded by a separate exon. A transmembrane region, intracytoplasmic anchor, and one of two alternatively spliced cytoplasmic tails (CY"-1 of 16 amino acids and CYT-2 of 2 3 amino acids) form the carboxyl terminus.
We have previously characterized two high mannose-containing pre-Golgi precursors in human cells whose processing times varied (19). We report that differences in precursor processing are directly related to the cytoplasmic tail of MCP.

EXPERIMENTAL PROCEDURES Construction of cDNA Mutants and Subcloning into Expression
Vectors-Four previously cloned MCP isoforms (14) were subcloned into the EcoRI sites of expression vectors pSFFV-neo (20) and pHPApr-1-neo (21). For controls, an isoform was subcloned in reverse orientation in both vectors.
Six modified constructs of MCP are utilized in this investigation (Fig.  2). A "tail-less" mutant consisted of the isoform MCP-BC2 in which the intracytoplasmic tail was deleted making the intracytoplasmic anchor the carboxyl terminus. The nomenclature for the MCP isoforms indicates which STP region (B or C) and which cytoplasmic tail (1 or 2) is expressed. The tail chimera designated MCP-BC2+1 added the amino acids of CYT-1 to CYT-2. The chimera designated MCP-BCl+B added CYT-2 to CYT-1. Finally, three constructs were made with deletions in the tail of MCP-BC1 of the first six (TYLTDE), the middle six (THREW), or the final four (FTSL) amino acids. Chimeras and mutants were created using polymerase chain reaction methodology in which two "partners" were generated and then ligated together. Thirtyresidue oligonucleotides were employed as amplification primers with the appropriate MCP cDNA isoform as template. The "outer" 5"amplification primer incorporated anEcoRI site joined to a common sequence in the 5'-untranslated region of MCP cDNA. The outer 3'-amplification primer for the tail chimeras consisted of a sequence complementary to MCP cDNA in the 3"untranslated domain with an added EcoRI site at its terminus. "Internal" 3'-and 5'-oligonucleotide primers were phosphorylated prior to cDNA amplification. Following the generation of the polymerase chain reaction partners, the products were blunt-end ligated. The ligated insert was restricted with EcoRI and subcloned into the expression vector. All constructs were sequenced in their entirety to verify the fidelity of created sequences.

Sugars
Some transfected cells were sorted for higher expression on an Epic 753 sorter (Coulter Corp., Hialeah, FL).
Pulse-Chase Analysis-Transfectants were starved 1 h in cysteinefree medium containing 10% dialyzed fetal calf serum. After this incubation, 100 pCi/ml L-["Slcysteine (DuPont NEN) was added for a 15min pulse. The label was removed and chased using fully supplemented medium for varying times. Cells were then washed with Dulbecco's phosphate-buffered saline (Life Technologies, Inc.) and solubilized in lysis buffer (1% Nonidet P-40, 0.05% SDS, and 1 mM phenylmethylsulfonyl fluoride in Dulbecco's phosphate-buffered saline). After incubating a t 4 "C for 20 min, debris was pelleted at 10,000 x g for 10 min. Supernatant was stored at -70°C. Immunoprecipitation, SDS-PAGE, and autoradiography were performed as described (14). Autoradiographs were scanned using a laser densitometer (Ultroscan, Pharmacia LKB Biotechnology Inc.) and software (GELscan) to analyze results.

RESULTS
Pulse-Chase Analysis of MCP Zsoforms- Ballard et al. (19) previously noted a 4-fold difference in the processing times of MCP precursors in several human cell lines. Since these lines expressed multiple isoforms of MCP, it was not possible to correlate this finding with a particular species. This issue became approachable with the molecular definition and transfection of the four commonly expressed isoforms (14).
Pulse-chase analyses were performed on stably transfected CHO and NIH-3T3 cell lines, each expressing one of the four isoforms. MCP was immunoprecipitated, characterized on SDS-PAGE, and the half-life (t,) of the precursor determined by densitometric scanning of the resulting autoradiograms ( Fig. 3 and Table I). Precursors of transfectants bearing CYT-1 were processed faster than those with CYT-2. For example, in CHO lines the t,h of the precursor for MCP-BC1 was 12-13 min while that for MCP-BC2 was 40 min. Likewise, in comparing isoform MCP-C1 with MCP-C2, the t , of the former was 10-11 min while that of the latter was 35-36 min. Similar results were obtained using the same four isoforms expressed in NIH-3T3 cell lines (Table I). These data establish that the cytoplasmic tail correlates with the difference in processing time.
Pulse-Chase Analyses of Cytoplasmic Tail Mutants and Chim-

era+"
tail-less mutant was constructed that was similar to MCP-BC1 and -BC2 in that it possessed the identical extracellular and transmembrane domains but terminated after the intracytoplasmic anchor. Pulse-chase analyses revealed that the t,h for the tail-less mutant precursor was 160-165 min (Fig.   4), a rate much slower than either MCP-BC1 (12-13 min) or -BC2 (40 m i d , suggesting that each cytoplasmic tail contained information that regulated its processing. Tail chimeras were next constructed in which both tails were placed on the carboxyl terminus of the construct in tandem, i.e. CYT-1+2 and CYT-2+1. Pulse-chase analyses revealed that the t,h of the precursor of construct MCP-BCl+B was 35-37 min (Fig. 4). Thus, the addition of CYT-2 slowed the processing time. On the other hand, the precursor of MCP-BCB+l processed with a t,h of 15-19 min showing that the addition of CYT-1 to CYT-2 accelerated the processing time of the native precursor. In both of these cases, the carboxyl-terminal sequence was a dominant factor in the processing time.
Next, mutants of MCP-BC1 were constructed in which the initial six, the middle six, or the terminal four amino acids of CYT-1 were deleted. Pulse-chase analyses of these stably transfected mutants revealed that the carboxyl terminus of the tail significantly affected processing rate. The t , of the construct deleting the first six amino acids (MCP-BCl(d1-6)) was 14-15 min (Fig. 5). The t,A2 of the construct deleting the middle six amino acids (MCP-BCl(d7-12)) was 13-14 min, and the construct deleting the last four amino acids (MCP-BCl(dl3-16)) was 30-32 min. Thus, only deletion of the last four amino acids of CYT-1 affected the t , of MCP-BC1 precursor. Also, the t,h of this mutant was now similar to that of MCP-BC2, suggesting that a positive or accelerating signal was contained within this tetrapeptide of CYT-1. DISCUSSION The goal of the present study was to examine if the cytoplasmic tails of MCP affected the processing of its precursors. We focused on differences in precursor processing rates because in a previous study Ballard et al. (19) had detected a 4-fold difference in the processing time of high mannose, pre-Golgi precursors of MCP in human cell lines. However, since multiple MCP species are expressed on human cells, it was not possible to determine which form(s) were being differentially processed. Consequently, we prepared stably transfected cell lines bearing  Table I.  Table I. the four common isoforms of MCP, employed pulse-chase methodology for analysis, and found that precursors of isoforms bearing CYT-1 chased several times faster than those possessing CYT-2 (summarized in Table I). Thus, these data clarify the earlier results and suggest that the cytoplasmic tail is responsible for this variability in MCP precursor processing time.
We next constructed a series of tail mutants and chimeras to further characterize the role of the cytoplasmic tail. If the cytoplasmic tail was deleted altogether (leaving the transmembrane domain and intracellular anchor), the processing rate slowed markedly (4-12-fold) compared with the native rate of both CYT-1 and CYT-2 isoforms, again demonstrating the importance of the cytoplasmic tail. The construction of chimeras in which both tails were present (in the order of 1+2 or 2+1) indicated that the carboxyl-terminal peptide most affected its processing rate since the chimera 1+2 possessed a processing rate similar to CYT-2 while chimera 2+1 possessed a rate more similar to CYT-1. This observation that the carboxyl-terminal amino acids had a profound effect on the processing rate was further substantiated by the finding that deletion of the first or middle six amino acids of CYT-1 had little or no effect on precursor processing, while deletion of the last four amino acids retarded its processing rate.
These findings are consistent with recent evidence that cytoplasmic domains are important determinants for the intracellular transport of certain proteins (reviewed in Ref. 23). Several studies employing either cytoplasmic tails translocated from other proteins or tail mutants (23, 24) have generally resulted in impeded processing or retention in the ER. Few, if any, studies have demonstrated an accelerated processing rate when splicing a domain onto a protein. Thus, in considering several possibilities to account for these results, we favor a hypothesis that CYT-1 contains a positive sorting signal that accelerates the processing of its precursor. Our results demonstrated that addition of CYT-1 to CYT-2 hastened its intracellular transport. Moreover, deletion of the carboxyl tetrapeptide (FTSL) of CYT-1 delayed its processing rate to that of proteins bearing CYT-2. The addition of CYT-2 to CYT-1 impeded its precursor processing rate suggesting that it "masked" the -CYT-1 (P) CK-2 I I

R Y L G R R K K K G T Y L T D E T H R E V K F T S L (P) PKC
CYT-2 -(P) Src Kinase (P) CK-2 I

Nuclear Targeting
FIG. 6. Schematic diagram of the potential phosphorylation and nuclear targeting sites on MCP cytoplasmic domains. CK-2, casein kinase 11; PKC, protein kinase C. The soflware program employed was PC/Gene (IntelliGenetics, Mountain View, CAI.
CYT-1 terminus or altered its conformation. Our results, then, suggest a specific process rather than a default pathway for transport of MCP proteins from the ER. However, alternative explanations for these results must be considered. If CYT-2 were to possess a signal for slower transport, the results would also be explained. However, it seems less likely that deletion of the last four amino acids of CYT-1 would slow its rate to nearly that of CYT-2.
We have considered several explanations to account for the cytoplasmic tail-directed differences in processing times. Following translation and entry into the ER, the carboxyl terminus of MCP is exposed on the cytoplasmic face of the ER membrane. As a result, it has ample opportunity to interact with host cell proteins (chaperones) that may direct its intracellular transport. Chaperone proteins are importantly involved in the control of protein structure, function, localization, and transport (reviewed in Refs. 25 and 26). Additionally, the cytoplasmic tails of MCP contain consensus signals for events such as phosphorylation and nuclear transport (Fig. 6). It is possible that a post-translational modification such as phosphorylation or the bidirectional transport of an MCP isoform into the nucleus (27) may have a bearing on its processing rate. These issues are being addressed utilizing MCP isoform transfectants, specific antibodies, and phosphorylation stimulators.
Another possibility is that clustering or oligomerization may affect precursor processing. Cytoplasmic domains may accelerate exit from the ER by inducing clustering of a protein at sites of transport vesicle formation (23). Additionally, it is possible that a tail signal may direct a process such as oligomerization in the ER. Many proteins undergo modifications in the ER including assembly into homo-or heterooligomers. The threedimensional structure and oligomeric state of a protein may control not only its functional properties but also its intracellular transport, cellular localization, and lifespan (28). Cytoplasmic domains have been implicated in this process for some proteins (28,29 cross-linking studies should allow us to determine whether MCP species form oligomers and if that is correlated with isoform processing rates.
Since MCP isoforms show differences in intracellular processing times, biologic advantages may be conferred as a result. For example, an isoform that can be more quickly recruited to the cell surface would be an advantage if MCP were up-regulated during complement activation at an inflammatory site. Additionally, a signaling ability mediated through a phosphorylation event could play a critical role in response to an inflammatory process. In these and other ways, a single cell that possesses multiple MCP isoforms may be protected more effectively from complement attack.