The N-terminal Hydrophobic Domain of P450c21 Is Required for ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Vol. 268, No. 20, Issue of July 15, pp. 14682-14686,1!393 Printed in U.S.A.

8 1993 by The American Society for Biochemistry and Molecular Biology, Inc.

The N-terminal Hydrophobic Domain of P450c21 Is Required for Membrane Insertion and Enzyme Stability* (Received for publication, January 22, 1993, and in revised form, March 16, 1993)

Li-Chung HSU, Meng-Chun Hu, Hsu-Chen Cheng, Juh-Chin Lu, and Bon-chu ChungS From the Institute of Molecular Biology, Academia Sinica, Nankang, Taipei, Taiwan 11529, Republic of China

Microsomal cytochromes P-450are known to be integrated into smooth endoplasmic reticulum through their hydrophobic sequences located at the N termini. The length requirement of the membrane insertion signal was determined by the generation of six plasmids encoding mutant P450c21 that lacked various portions of the N-terminal hydrophobic domains. When they were transcribed and translated in vitro in the presence of endoplasmicreticulum membranes, mutant protein lacking more than a third of the first hydrophobic domain gradually lost the ability to insert into the membrane and stayed mostly in the soluble fraction when the first N-terminal hydrophobicdomain was removed. The steady-state amountof the truncated proteins was progressively reduced in parallel to the extent of their N-terminal deletions, due to their fast degradation. This process was accompanied by a decrease in the enzymatic activity. Therefore, the first hydrophobic domain of P450c21 not only serves as a membrane targeting and anchoring domain, but it is also important for the in vivoprotein stability.

based on the three-dimensional structure of a distantly related protein, bacterial P450cam (11).A major problem in using this comparison is that mammalian P-450sare integral membrane proteins, whereas P450cam exists in the soluble form. On the basis of sequence alignments among 34 cytochromes P-450, Nelson and Strobe1 (12) predicted that microsomal P450s are each integrated into the membranethrough two hydrophobic stretches at theN terminus, whereas the rest of the protein probably assumes a structure similar to that of P450cam. Yet P450b and P450-1A1 were shown in vitro by Monier et al. (13) and Sakaguchi et al. (14), respectively, to be attached to the microsomal membrane through a single short hydrophobic segment at the N terminus. On the other hand, P450 2E1 protein with deletion of the first 21 amino acids is nevertheless tightly associated with the bacterial membrane when expressed in Escherichia coli (15). To better characterize the role of the first hydrophobic domain for membrane insertion, we produced a seriesof mutant P450c21 proteins with increasing lengths of truncation from the N terminus and characterized the properties of these deletion mutants. Our results support the notion of Monier et al. (13) and Sakaguchi et al. (14) that the firsthydrophobic segment, followed by a few positive charges, is sufficient to target the Steroid 21-hydroxylase (P450c21) catalyzes the conversion protein into the microsomal membrane in the correct orienof progesterone and 17a-hydroxyprogesterone into deoxycor- tation. In addition, we find that the extent of membrane ticosterone and 11-deoxycortisol, respectively. Deficiency in integration depends on the length of the first hydrophobic 21-hydroxylase results in impaired steroid synthesis and ac- segment. At least two-thirds of the segment is required for counts for more than 90% of a common genetic disease, efficient integration. Moreover, proteins missing 6 residues or congenital adrenal hyperplasia (1, 2). The enzyme deficiency more at theN terminus arerapidly degraded in uiuo. This Ncan be attributed to mutations in the CYP2lB gene that terminal hydrophobic domain thus apparently also contribencodes P450c21 (2). These mutations include gene deletions, utes to the structuralrequirements for maintaining the prosplicing errors, and point mutations (3-5). Many of the point tein’s stability. mutations are located in the coding region of the CYPZlB MATERIALSANDMETHODS gene, resulting in formation of proteins with impaired enzyPlasmid Construction-To create N-terminal deletions, the plasmatic activities (6-9). For example, the Asn-172 and Trp-356 mutant enzymes have severely impaired activities and are mid pc21 that contains the full-length human P450c21 cDNA (16) associated with the simple virilizing form of the disease. The was linearized at the SalI site located 10 nucleotides upstream from the initiation codon and then sequentially digested by Bal-31 and Leu-281 and Leu-30 mutant enzymes have slightly higher EcoRI. The shortened DNA fragments were cloned into a vector residual activities and are associated with the milder, non- (pB3’c21) under the control of a T3 promoter downstream from the classical type of the disease (10). efficient translational start signal of black beetle virus RNA1 (17). and A23) with the correct reading The structure-function relationship of P450c21, as well as Five clones (A4, A7,AlO,A12, of other eukaryotic cytochromes P-450, is not well understood frame and initiation codon were identified by sequencing. These plasmids were subcloned into a pCDBam vector with the SV40 early because of the apparent difficulty in producing protein crys- promoter for expression in mammalian cells (18). The fragment tals for three-dimensional structure determination. Many of between the 53rd codon and the 3’-region of the P450c21 cDNA was the tertiary structure predictions of mammalian P-450s are generated by polymerase chain reaction while creating an ATG codon

* This work was supported by Grant NSC80-0412-B-001-10from the Academia Sinica and National Science Council, Republic of China. 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. $ T o whom all correspondence and reprint requests should be addressed.

at its 5’-end. This fragment was cloned into pB3’c21 to form A52. The c21145~-lactamase41~~ fusion clone was generated by replacing the SspI/XmnI fragment of pBR322 (containing codons 1-41 of the /3-lactamase gene) with the PuuII/SmaI fragment of pc21 that has the T3promoter and codons 1-35 of P45Oc21. In Vitro Protein Analysis-The in vitro transcription, translation, and translocation experiments were performed by using kits purchased from Promega Biotec (Madison, WI). For membrane integration tests, translationswere performed in the presence of dogpancreas

14682

14683

Membrane Integrationof P450c21 microsome membranes (PromegaBiotec) according to manufacturer’s suggestions. Translation products were incubated in 0.1 M sodium carbonate, pH 11.5, on ice for 30 min before centrifugation at 100,000 rpm in aBeckman TLA 100.2 rotorfor 15 min. The pellet was dissolved directly in gel loading buffer for electrophoresis, whereas the supernatant was precipitated in 10% cold trichloroacetic acid on ice for 30 min followed bycentrifugation. The pellet was washed with cold acetone, dried completely, and redissolved in gel loading buffer for electrophoresis. For proteinase K sensitivity assay, translation products were treated with 0.1 mg/ml proteinase K in 10 mM TrisHC1, pH 8.0, 10 mM CaC12a t room temperature for 30 min. Digestion was stopped in phenylmethylsulfonyl fluoride (1 mg/ml) and gel loading dye before electrophoresis. Transfection, Enzymatic Assay, Immunobbtting Immunoprecipitation, and Pulse-Chose Experiments-For each mutant plasmid, Rat1 or COS-1 cells in 6-cm plates were transiently transfected with 5 pgof the pCD21 deletion plasmid and 5 pg Rous sarcoma virus-(3galactosidase for transfection efficiency control. 21-Hydroxylase activity was determined by incubating [“C] 17a-hydroxyprogesterone with the cells for 1 h followedby thin layer chromatography as previously described (18).For determination of expression levels, cell extracts were divided into two parts for (3-galactosidase assay and immunoblotting analysis. (3-Galactosidase assay was performed according to the instructions provided by Pharmacia (Uppsala, Sweden). Immunoblotting was performed according to an established procedure except that 35S-labeled protein A was used to label the proteinbands (18). For pulse-chase experiments, cells 48h after transfection were incubated for 1 h with 250 pCi/ml [35S]methionine, followed by a chase in complete, nonradioactive Dulbecco’s modified Eagle’s medium for the indicated time periods. Cells were then lysed, reacted with antisera against recombinant P450c21 (17). and precipitated with formalin-treated Staphylococcus aureus. The immunoprecipitate was washed sequentially with the lysis buffer, lysis buffer with 1 M NaCI, lysis buffer without 1 M NaCl, and finally with 10 mM Tris-HCI, pH 7.4, before dissolving in sample loading buffer for electrophoresis. The protein bands were first visualized by autoradiography, followed by quantitation using a PhosphoImager machine (Molecular Dynamics, Sunnyvale, CA). RESULTS

Characterization of the in Vitro Translation System-To test the topology of P450c21 in terms of its integration into the membrane, a coupled i n uitro protein translation/translocation system was used. Human P450c21 cDNA was cloned downstream from the bacteriophage T3 promoter for the in uitro synthesis of P450c21 RNA followed by protein translation in rabbit reticulocyte lysate in the presence of dog pancreas microsomes (Fig. 1). A major band of about 50 kDa corresponding to P450c21 was synthesized (lanes 1 and 3 ) . In addition, there were some minor bands that could be due to internalinitiation or incomplete proteinsynthesisinthis system. Since the mature P450c21 had the same size as the nascent peptide (comparing lanes 1 and 3 ) , the membranetargeting signal peptide was not cleaved after protein translocation. The full-length P450c21 band was digested by proteinase K even when membrane was present (lanes 2 and 4 ) , indicating that P450c21 was located on the cytoplasmic side of the membrane. These results show that P450c21, similar to other microsomal P-450s (13, 14, 19), is synthesized as a 50-kDa protein with a noncleavable targeting signal and is located on thecytoplasmic side of the membrane. The major P450c21 band disappeared upon proteinase K digestion. However, two faint bands of 52 and 30 kDa could be observed after the digestion (marked by dots in lane 4 ) . The 30-kDa band might be the degradation product derived from P450c21. This finding suggested that a portion (30 kDa) of P450c21 might either be embedded in the membrane or form a compact structure thatis more resistant to proteinase K digestion. The band that migrated more slowly than the major 50-kDa band was probably synthesized from the low levels of inside-out vesicles in the microsome preparation (data notshown) that did notaffect our data interpretation.

P-Lactamase

pBC21 Memb.-

prot. K

“110”

- + + + - +

Wa)

- 84 -

- - + + - + - +

T S P

L 47

- 33 - 24 -161

2

3

4

5

6

7

8

91011

FIG. 1. In vitro translation of P450c21 and @-lactamase. P450c21 (lanes 1-4) and (3-lactamase (lanes 5-11) were translated in rabbit reticulocyte lysate in the presence (lanes 3, 4, and 7-11) or absence (lanes 1 , 2 , 5 , and 6 ) of dog pancreas microsomes, digested with proteinase K (lanes 2, 4 , 6, and 8), or subjected to alkaline extraction followed by centrifugation (lanes IO and 11). The translation products were separated by SDS-polyacrylamide gel electrophoresis. The arrow points to P450c21. Two dots denote proteinase K-resistant peptides. The top and bottom arrowheads point to the precursor and processed forms of (3-lactamase, respectively. T,total translation product; S , supernatant; P, pellet. A1 0 A12 A23 A52 S M_S- M- S M- S”M S- M S M S M

c21 A7A4

” ” ” -

kDa -200

- 97.4 - 69

c21j

- 46

1 2 3 4 5 6 7 8 9 10 1 1 121314

FIG.2. Membrane attachment assay of the P450c21 deletion mutants. The in uitro translation products from the series of P450c21 deletion mutants were subjected to alkaline extraction and centrifugation. Proteins recovered from thesupernatant ( S ) and pellet ( M )were electrophoresed in a 12% gel. The arrow indicates the P450c21 protein.

As a control, 0-lactamase was also synthesized in the same system. Its secretory peptide was cleaved after translocation, and itwas in thesoluble fraction but protected from proteinase K digestion; thus it must be .inside the lumen of endoplasmic reticulum (Fig. 1, lanes 5-11). When P450c21 protein was extracted by alkaline sodium carbonate followed by centrifugation to separate the membrane-bound form from soluble proteins, most of the protein was in themembrane fraction, indicating that themembrane integration process was very efficient in this system (Fig. 2, lanes I and 2).

MembraneIntegration of theN-terminal Deletion Mutants-The first one or two hydrophobic domains of P-450s are thought to be important for membrane integration (1214). P450c21 has two hydrophobic stretches at theN terminus covering amino acids 1-22 and 29-53. To find out how many N-terminal amino acids are required for P450c21 membrane insertion, serial deletions of the P450c21 cDNA were performed to create mutants lacking codons 2-4,2-7,2-10,212, 2-23, or 2-52, respectively. Mutant proteins were synthesized in uitro in the presence of microsomal membranes and were followed by alkaline extraction to test for their ability to integrate into the membrane (Fig. 2). The mutantproteins missing the 2nd to the 4th amino acids (A4) or the 2nd to the 7th aminoacids (A7) were still integrated into themembrane

Integration Membrane

14684

fraction asefficiently as thefull-length P450c21, whereas the one lacking residues 2-10 (A10) was equally distributed into the soluble portion and the membrane fraction. About 70% of the A12 and A23 proteins were soluble, as quantified by counting the radioactivities in the protein bands in Fig. 2. This result indicated that the lack of a third of the first hydrophobic domain(asin clone A7) did notdestroythe ability of the protein to enter the membrane. In other words, two-thirds of the first hydrophobic domain were enough to confer membrane integration. However, deletion of a half of it (as inclone AlO) started to affect membrane insertion. Since this result indicated that the first hydrophobic domain of P450c21 was required for membrane integration, we tested its ability to target proteins to membranes by replacing the secretorysignal peptide of the p-lactamaseby amino acids 1-35 of P450c21. This hybrid protein was found associated with thepelletable fraction and on the cytoplasmic sideof the membrane (Fig. 3), whereas the normal p-lactamase was in the soluble fraction in the lumen(Fig. 1, lanes 5-11). Therefore, amino acids 1-35 of P450c21 contained the membranetargeting signal that can direct a heterologous protein into the membrane in the same orientation as that of P450c21. It appears that the N-terminalhydrophobic domains in all microsomalcytochromesP-450, such as P450c21 (Figs. 1-3), P450b (13), and P450 1Al (14), have similar properties in targeting the protein to the microsomal membrane in vitro. Production of the N-terminal Deletion Mutants-In the in vitro translation system, it appeared that less protein was detected when the N-terminal deletion became longer. For the A52 mutant protein, very little of it could be detected although much more translation mixture was loaded on the gel (Fig. 2, lanes 13 and 14). T o study whether this phenomenon alsooccurs in vivo, truncated P450c21 proteins were produced in COS-1 or Rat-1cells after being transfected with variousdeletionplasmidssubcloned into a mammalian expression vector pCDBam(18).Rat-1 cells were used because 21-hydroxylase activity can be detected when these cells are transfected with theP450c21 gene (16), and they also contain less cross-reactingmaterialthan COS-1 cells to P450c21 antiserum. The amountsof mutant proteins produced in the transfected cells were determined by immunoblotting analysis. As shown in Fig. 4, the steady-state level of the mutant protein decreased progressively as the extent of N-terminal deletion increased. A4 produced the highest amount of proProt. K

- +

of P450c21 %c21: 0 4 16 16 2523

68100

-21

1 2 3 4 5 6 7 8 FIG. 4. Immunoblot analysis of P450c21 proteins produced by deletion mutants. After transfection of the wild-type or deletion plasmids, about one-third of the COS-1 cell extract was electrophoresed in 10% polyacrylamide gels. After transfer tonitrocellulose, the protein was reacted with anti-P450c21 antiserum, followed by ”Slabeled protein A, and autoradiographed. Lune 1 refers to cell extract transfected with no DNA. Lunes 2-8 refer to cell extracts transfected with deletion plasmids as marked on thetop of each lane. Lune 8 was cell extract with full-length c21 DNA. All the band intensities were quantified by a PhosphoImager and normalized. The relative intensity of each band in percentage was shown above each lane.

I

2

O

7

c21 A 4 A 7 A l O A 1 2 A 2 3 A 5 2 FIG. 5. 21-Hydroxylase activity assay. [“C]17a-hydroxyprogesterone (1 p M ) was added to the cells in 6-cm plates 48 h after cotransfection with wild-type or P450c21 deletion plasmids in mammalian expression vectors together with Rous sarcoma virus-8-galactosidase. The steroid products were extracted 1 h later and separated by thin layer chromatography. 21-Hydroxylase activities were calculated after normalization with P-galactosidase activities. Five independent transfections into COS-1 and one transfection into Rat-1 cells were performed, and mean enzymatic activity of each mutant relative to the wild-type activity was calculated with standard deviations shown.

tein; A7, AlO, A12,and A23 produced similarlylower amounts; and A52 produced an almostundetectable amount. Thisresult “”69kDa was different from our previous observation using mutantsof -46kDa P450c21 with single substitutions, in which the amounts of mutant proteinswere about the same as that of the wild type c (18). b-3OkDa Enzymatic Activity of N-terminal Deletion Mutants-The enzymatic activities of the deletion mutant proteins were measured in intact cells after transient transfection (Fig. 5). While the A4 protein retained about 70% of the wild-type activity, other mutants had less than 10% of the wild-type 1 2 3 4 M activity. Althoughthis decrease in enzymatic activity parallels the reduction in the level of proteins in Rat-1 cells (Fig. 4), c21 fi-Lactamase the specific activity of each mutant protein appeared to be lower when the total activities were divided by the amountof each mutant protein. FIG. 3. Membrane targeting property of amino acids 1-35 Stability of the N-terminal Deletion Mutants-Since all the of P450c21. Hybrid protein ~ 2 1 ~ - ~ ~ ~ - 1 a c t a was m a stranslated e~~-~~ deletion mutants have been cloned into the same vector and in vitro inthe presence of membrane (lane 1 ) and subjected to proteinase K digestion (lane 2) or alkaline extraction followed by shared the same transcriptional and translational initiation centrifugation to be separated into the supernatant (S, lane 3 ) and sites, itis likely that thedifferences in the steady-statelevels pellet (P,lane 4 ) forms. M is a size marker. of the mutant proteins are due to differences in the rates of

Integration Membrane their in uiuo degradation. Therefore, deletion mutants were pulse-labeled aftertransienttransfectionintoRat-1 cells, chased for different periods, and detectedby immunoprecipitation to measure the rates of degradation. As shown in the top panel of Fig. 6, full-length c21 protein produced in Rat-1 cells was readilyvisible after a 1-h pulseperiod and remained detectablethrough a 24-h chase. The control transfection without P450c21 plasmid produced only cross-reacting materials without a specific c21 band (lane 1 ) . In the cells transfected with A7, a protein band thatwas detected immediately after the pulse period gradually decreased in intensity a t a 1and 2-h chase. It eventually disappeared after about 16 h. This result indicated that A7 was less stable than the fulllength protein. The amountof the remaining proteina t each time point, as quantified by a PhosphoImager, is plotted in Fig. 7A. The full-length protein is apparently very stable, havingahalf-life of more than 20 h (Fig. 7B).Withan increasing length of deletion, the half-lives of the proteins decreased. The AlO, A12, and A23 proteins had a half-life of about only 2 h. The half-life of the A52 protein was too short to be measured accurately.

f c21 ]CRM

16 6

2

1

0" - Time(h)

" " "T"

-A7 CRM

7

6

5

4

3

2

dl A4

A7 x

A10 A12

a A 2 3

0

30

20

10

Time (h)

(B)

cl

Integral membrane proteins canbe inserted into cell membranes with various topologies. Some proteins span the membrane many times, whereas others span the membrane only once (20). Integral membrane proteins have been classified into three kinds (21); class I proteins contain an N-terminal cleavablesignalsequence,class I1 proteins contain a noncleavable signal sequence with NeytCero topology, and class I11 proteins also have a noncleavable signal sequencebut assume the NexoCcytorientation. The orientation of the transmembrane segment isinfluenced greatly bythe charges presenta t both ends of the transmembrane segment (22, 23). P-450s usually have a net positive charge at the C-terminal end of the transmembrane domain with class I11 topology (13, 14, 23). Introducing positive charges into the N-terminal endof P450 2C2 causes translocation of the protein into the endoplasmic reticulum lumen (24, 25). Our results show the class I11 topology for the integration of P450c21 into the endo-

1

14685

3

DISCUSSION

.24

of P450c21

1

FIG. 6. Immunoprecipitationof pulse-labeled c 2 1 or c21A7 protein chased for different times. Rat-1 cells in 10-cm plates were transiently transfected withpCDc21 (top panel), A7 (bottom panel), or no plasmid (lane I of bothpanek), separated into six 6-cm plates, and pulsed for 1 hwith[3sS]methionine. P450c21 proteins were chasedfor 0, 1, 2, 6, 16, or 24 h (lanes 2-7, respectively), immunoprecipitated, andelectrophoresed in polyacrylamide gels. The arrow indicatesfull-lengthP450c21 or A7 protein; CRM refers to cross-reacting material.

2

5 10

z

0

l

l

¶

m

z n

0 1 A 4 A7 AlOA12A23 FIG. 7. Half-lives of deletion mutants. A , the intensity of the bands corresponding to wild-type or deletion mutants in Fig. 6 was quantified by a PhosphoImager and plotted against time on a semilogarithmic scale by using a linear regression program. The R2values are larger than 0.7 in all cases. B, the half-life of each protein was calculated from panel A and shown uersus the length of the deletion.

plasmic reticulum membrane.The first hydrophobic domain of P450c21 at the N terminus is about 22 amino acidsin length. After deletion of about a third of it (as in A7), the protein was still tightly integrated into the membrane. Deleting about one-half of the first domain (as in AlO) resulted in equal distribution of the protein into the membrane-bound in 623) and the free form. Loss of the entire domain (as resulted in loss of most of the membrane integration. These in vitro results indicate that the membrane-spanning property of P450c21 depends on the length of its first hydrophobic domain of 23 amino acids. The participation of the second hydrophobic stretch from amino acid 29 to 53 asa membrane targeting signal appears unlikely or dependent on the first stretch, because A23, which retained the second hydrophobic stretch, did not integrate into the membrane. Lower levels of proteins were detected when P450c21 with N-terminal deletions were expressed in COS-1 orRat-1 cells. In contrast, P45017a is produced at the same steady-state level in COS-1 cells when it was missing 17 amino acids at the N terminus (17). P450 2E1, P450c7, and P45017a with N-terminal deletions are also produced at high levels and remain active when expressedin E. coli (15, 27-29). This discrepancy may represent differences in protein structures among members of the cytochrome P-450 family. The truncated proteins lost enzymatic activities in parallel to their fast degradation. Since enzymatic activity is determined by protein structure, mutant proteins must have perturbed structures. The N-terminal hydrophobic domain,when inserted into the membrane, probably does not form a separate domain apart from the bulk of the protein. Rather, it probably interacts with the rest of the protein to hold the overall structure of P-450 together. Absence of the N-terminal

Membrane Integration of P450c21

14686

domain of the structural component would lead to partial unfolding of the protein and subsequent fastdegradation. The mechanism of P-450 degradation is not well understood. P450 2E1 is suggested to be degraded through two pathways (30); one is the slow autophagosomal/lysosomal pathway for the bulk of the protein, whereas the other is a hormone- and substrate-regulatedpathway via a specific proteolytic system in the endoplasmic reticulum. Our data suggested that the wild-type P450c21 was degraded in a monophasic pattern with a half-life of about 24h. Truncated P450c21 proteins were degraded much faster but followed the similar monophasic pattern. Correct folding of aprotein should be important for its stability, as denatured proteins are known to be degraded more rapidly than thenative forms (26). A single amino acid mutation of P4502C13 shows altered protein stability (31). Presumably a structural change in the protein triggers the degradation pathway. The precise mechanisms for P-450 protein degradation still await further investigation. Acknowledgments-We thank Drs. Victor Guzov, C. C. Wang, and Cathy Fletcher for reading the manuscript. REFERENCES 1. Miller, W. L., and Levine, L. S. (1987) J. Pediutr. 1 1 1 , l - 1 7 2. Miller, W. L., and Morel, Y. (1989) Annu. Reu. Genet. 23,371-393 3. White, P. C., Vitek, A., Dupont, B., and New, M.I. (1988) Proc. Natl. Acad. Sci. U. S. A. 86,4436-4440 4. Hi ashi, Y., Tanae, A,, Inoue, H., Hiromasa, T., and Fujii-Kuriyama, Y. fl988) Proc. Natl. Acad. Sei. U. S. A. 86, 7486-7490 5. Wedell, A., Ritzen, E. M., Haglund-Stengler, B., and Luthman, H.(1992)

Proc. Natl. Acad. Sci. U. S. A. 8 9 , 7232-7236 6. Amor, M., Parker, K. L., Globeman, H.,New,M. I., and White, P. C. (1988) Proc. Natl. Acad. Sci. U.S. A. 86,1600-1604 7. Chiou, S.-H., Hu, M.-C., and Chung, B. (1990) J. Biol. Chem. 2 6 6 , 35493552 8. Wu, D.-A., and Chung, B.-c. (1991) J. Clin. Inuest. 88,519-523 9. Tusie-Luna, M.-T., Speiser, P. W., Dumic, M., New, M. I., and White, P. C. (1991) Mol. Endocrid. 6,685-692 10. Speiser, P. W., Dupont, J., Zhu, D., Bue eleisen, M., Tusie-Luna, M.-T., Lesser, M., New, M. I., and White, P. (1992) J. Clin. Inuest. 9 0 , 584595 11. Poulos, T. L., Finzel, B.C., and Howard, A. J. (1987) J. Mol. Biol. 196,

8.

68-700

12. Nelson, D. R., and Strobel, H. W. (1988) J. BioL Chem. 263,6038-6050 13. Monier, S., Van Luc, P., Kreibich, G., Sabatini, D. D., and Adesnik, M. (1988) J. Cell Biol. 107,457-470 14. Sakaguchi, M., Mihara, K., and Sato, R. (1987) EMBO J. 6,2425-2431 15. Larson, J. R., Coon, M. J., and Porter, T. D. (1991) J. Biol. Chem. 2 6 6 , 7321-7324 16. Ricketts, M. H.,Chiao, E., Hu,M.-C., and Chung, B. (1992) Biochem. Biophys. Res. Commun. 186,426-431 17. Dasmahapatra, B., Rozhon, E. J., and Schwartz, J. (1987) Nucleic Acids Res. 16,3933 18. Hu, M.-C., and Chung B.-c. (1990) Mol. Endocrinol. 4,893-898 19. Sakaguchi, M.,Mihda, K., and Sato, R. (1984) Proc. Natl. Acad. Sci. U. S. A. 81,3361-3364 20. Wickner, W. T., and Lodish, H. F. (1985) Science 230,400-407 21. von Heijne, G. (1988) Biochim. Bio h s Acta 947,307-333 22. Boyd, D., and Beckwith, J. (1990) 6e$62,1031-1033 23. Hartman, E.,Ra oport T A , and Lodish, H. F. (1989) Proc. Natl. Acad. Scr. U. S. A. 88,5788-5i90 24. Szczesna-Skorupa, E., Browne, N., Mead, D., and Kemper, B. (1988) Proc. Natl. Aead. Sei. U. S. A. 86, 738-742 25. Szczesna-Skorupa, E., and Kern er, B (1989) J Cell Biol. 108,1237-1243 26. Pakula, A. A,, and Sauer, R. T. rl989jAnnu. Reu. Genet. 23,289-310 27. Clark, B. J., and Waterman, M.R. (1991) J. BWL Chem. 266,5898-5904 28. Barnes, H. J.! Wada, A,, Imai, T., Sa ara Y and Waterman, M. R. (1992) J. Baste Clm. Physml. Pharmacol. f , (dup"1) 13 14 29. Li, Y. C., and Chian , J. Y. L (1991) J. Brof & e ..; 266,19186-19191 30. Eliasson, E., Mkrtckian, S., and Ingelman-Sundberg, M. (1992) J. Biol. Chem. 267,15765-15769 31. Faletto, M. B., Linko, P., and Goldstein, J. A. (1992) J. Biol. Chem. 2 6 7 , 2032-2037