THE JOURNAL 0 1942 by The American Swiety for Blochemistry end Molecular Biology, Inc.

OF B%OLOGICAL CHEM!STRY Vol .267, , No. IBsue of July 25, pp. 14629-14636,1992 Printed in U.S.A.

Processing of Proinsulin by Transfected Hepatoma (FAO) Cells* (Received for publication, November 25,1991)

Florence VollenweiderS, Jean-Claude IrmingerS, David J. Gross$, Lydia Villa-Komaroffn, and Philippe A. HalbanSII From the SLaboratoires de Recherche Louis Jeantet, Centre Medical Uniuersitaire, University of Geneva, 121 1 Geneva 4, Switzerland, the §Department of Endocrinology and Metabolism, Hadassah University Hospital, Jerusalem, Israel, and the (Department of Neurology, The Children’s Hospital and Harvard Medical School, Boston, Massachusetts 02115

Rat hepatoma (FAO) cells were stably transfected with the gene encoding either rat proinsulin E 1 (using the DOL retroviral vector) or human proinsulin (using the RSV retroviral vector). Using the DOL vector, production of insulin immunoreactive material was stimulated up to 30-fold by dexamethasone (5 x 10” M). For both proinsulins, fractional release of immu- noreactive material relative to cellular content was high, in keeping with the absence of any storage com- partment for secretory proteins in these cells. Pulse- chase experiments showed kinetics of release of newly synthesized products in keeping with release via the constitutive pathway. High performance liquid chro- matography analysis showed immunoreactivity in the medium distributed between three peaks. For rat proinsulin 11, the first coeluted with intact proinsulin; the second coeluted with des-64,65 split proinsulin (the product of endoproteolytic attack between the insulin A-chain and C-peptide followed by trimming of C- terminal basic residues by carboxypeptidase); the third (and minor peak) coeluted with native (fully processed) insulin. For human proinsulin, by contrast, the second peak coeluted with des-31,32 split proinsulin (split and trimmed at the B-chain/C-peptide junction). Analysis of cellular extracts showed intact proinsulin as the major product. The generation of the putative conver- sion intermediates and insulin was not due to proteol- ysis of proinsulin after its release but rather to an intracel lul~ event. The data suggest that proinsulin, normally processed in secretory granules and released via the regulated pathway, may also be processed, albeit less efficiently, by the constitutive pathway con- version machinery. The comparison of the sites pref- erentially cleaved in rat I1 or human proinsulin sug- gests cleavage by endoprotease($ with a preference for R/KXR/KR as substrate.

It is now clearly established that cells can secrete proteins by two distinct routes: the regulated and the constitutive pathways (1, 2). In the pancreatic B-cell, proinsulin is nor- mally targeted from the trans-Golgi to secretory granules for regulated release (3, 4). It is in the secretory granules that conversion of proinsulin to insulin occurs (5). Two endopro-

* This work was supported by Grant DK-35292 from the National Institutes of Health and by the Greenwall Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “adver- tisement’’ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

11 To whom correspondence should be addressed: Lab. Recherche Louis Jeantet, Centre Medical Universitaire, 1 rue Michel Servet, 1211 Geneva 4, Switzerland.

teases are thought to be involved, each cleaving preferentially at only one of the two sites of conversion linking the two insulin chains to C-peptide‘ (6) and probably corresponding to recently identified members of a family of mammalian proteins displaying homology with the yeast KEX2 endopro- tease (7, 8), referred to as PC3 (or PC1) (9-11) and PC2 (9, 12, 13). Expression of these enzymes is believed to be re- stricted to cells equipped with the regulated secretory pathway (14, 15).

All cells, including the pancreatic B-cell (16), are equipped with the constitutive secretory pathway, which is used not only for secretion but also for transport of integral proteins (including receptors) to the plasma membrane (1, 2). Many such proteins are synthesized as higher molecular weight precursor proproteins. Conversion of proprotein to protein arises before release (or presentation at the cell surface). The enzyme(s) responsible are other members of the mammalian KEX2-like family; the first to be positively identified was PACE (17) or furin (18). Although furin, like PC2 and PC3, cleaves at pairs of basic residues, its substrate specificity is different (14, 15), and it must operate under environmental conditions that surely differ from those encountered in ma- turing secretory granules.

There are certain circumstances in which it has been sug- gested that proinsulin is less efficiently directed toward the regulated pathway than in normal B-cells (4, 19), leading to a dispropo~io~ate amount of proinsulin diverted to the con- stitutive pathway. The question then arises as to whether any conversion of proinsulin can occur in this pathway. We have addressed this by transfecting FA0 cells, a rat hepatoma cell line that is presumed to use only the constitutive secretory pathway, with the cDNA encoding either rat preproinsulin I1 or the human preproinsulin. HPLC analysis of proinsulin- related material produced by these cells revealed that proc- essing does indeed occur, albeit to a more limited extent than in the regulated pathway.

MATERIALS AND METHODS

Transfection and Cell Culture-FA0 cells (from Dr. C. Ronald Kahn, Joslin Diabetes Center, Boston) were transfected as described previously (20,21). In brief, the cells were either transfected (calcium precipi~tion method) with pDOL-rINS (the retroviral DOL vector (22) carrying rat preproinsulin I1 cDNA) or cotransfected with pRSV- hINS (pBR322 carrying human preproinsulin cDNA driven by the RSV LTR promoter (a generous gift of Dr. H.-P. Moore, Univ. of California, Berkeley)) + pRSV-Neo. Stably transfected clones were selected using G418 (Geniticin, from GIBCO, Irvine, Scotland) at an active dose of 0.25 mg/ml. Transfected cells were grown in DMEM

’ The abbreviations used are: C-peptide, connecting peptide; DMEM, Dulbecco’s modified Eagle’s medium; Hepes, 4-(2-hydroxy- ethyl)-1-piperazineethanesulfonic acid; BSA, bovine serum albumin; HPLC, high performance liquid chromatography; IRI, immunoreae- tive insulin; LTR, long terminai repeat.

14629

14630 Proinsulin Processing in Transfected Hepatoma Cells

(GIBCO) supplemented with 10% fetal calf serum (GIBCO), 100 units/ml penicillin, 100 pg/ml streptomycin, and 450 mg/dl glucose, in a humidified atmosphere of 5% C02:95% air a t 37 "C.

Incubation of Cells and Stimulation of Immunoreactive Insulin Production by Dexamethasone-To stimulate insulin production from cells transfected with pDOL-rINS, the cells were grown to semicon- fluence in 100-mm-diameter plastic Petri dishes (3.5-4 X lo7 cells/ dish), and the medium was changed to DMEM containing 10 mM Hepes, 0.5% (w/v) bovine serum albumin (BSA), 450 mg/dl glucose, p H 7.4, supplemented with 5 X M dexamethasone. Medium was collected at set times and centrifuged (150 X g; 10 min) to pellet any cells that may have become detached during the incubation period. The cells were then extracted in 5 ml of 1 M acetic acid, 0.1% BSA.

In order to accumulate enough immunoreactive material released from cells transfected with pRSV-hINS, the medium (as above but without dexamethasone) was collected and replaced every 2 h during a total incubation period of 8 h, and the four medium samples were pooled for analysis.

Radioimmunoassay-Insulin-related products were measured by radioimmunoassay (23). Rat insulin standards (a mixture of rat insulins I and 11), human insulin standards, and guinea pig anti- porcine insulin serum were obtained from Novo-Nordisk, Bagsvaerd, Denmark, and "'I-pork insulin was obtained from Sorin Biomedica, Saluggia, Italy. Using this assay, both rat insulins are recognized equally well.' The cross-reactivity for rat proinsulin or conversion intermediates has not been established (due to lack of availability of standards). Cross-reactivity for human proinsulin or its conversion intermediates relative to native human insulin is approximately 50% on a weight basis and 75% on a molar basis (20).

Reversed Phase HPLC-Prior to injection, samples were prepuri- fied and concentrated using C,, Sep-Pak cartridges (24). Insulin, proinsulin, and conversion intermediates were separated as described in detail previously (25, 26). In brief, a Beckman (San Ramon, CA) Ultrasphere ODS 5 pm (4.6 X 250 mm) column was used attached to a Beckman System Gold HPLC apparatus. Samples were eluted using TEAP (20 mM triethylamine, 50 mM phosphoric acid, 50 mM sodium perchlorate, pH 3.0) (component A) and 90% acetonitrile, 10% water (component B). Component B was held at 34% during the first 25 min to allow for isocratic elution of insulin and then increased linearly to 36.5% over 75 min to elute proinsulin and conversion intermedi- ates. The system was standardized by injection of radiolabeled ma- terial extracted from isolated rat islets that had been labeled for 60 min with [''Hlleucine followed by a 60-min chase. The identity of each radioactive peak has been described in detail previously (25,26). Human proinsulin, conversion intermediates, and insulin, used as standards, were the generous gift of Dr. R. Chance, Eli Lilly and Co, Indianapolis, IN.

Radioactivity in each 1-ml fraction was measured by the addition of 3.5 ml of Lumaflow I1 scintillation mixture (from Lumac, Olen, Belgium). For measurement of insulin-related products by radio- immunoassay, fractions (1 ml) were collected in borosilicate glass tubes containing 100 pl of 0.5 M boric acid and 1% BSA, pH 9.3. After evaporating most acetonitrile, the fractions were lyophilized and reconstituted in 1 ml of 0.2 M glycine, 0.25% BSA, pH 8.8.

Pulse-Chase Experiments and Immunoprecipitation-Cells ex- pressing rat proinsulin I1 were preincubated with 5 X M dexa- methasone for 6 h. After washing 3 times with DMEM containing 10 mM Hepes, 450 mg/dl glucose, 0.5% BSA, but lacking methionine, the cells in each 10-cm-diameter dish were labeled for 15 min a t 37°C in 5 ml of the same medium supplemented with 1 mCi [35S]methionine (biosynthetic labeling grade from Amersham International, Amer- sham, United Kingdom). After extensive washing with medium as above but containing 0.2 mM unlabeled methionine, the cells were incubated for set postlabel (chase) times. The medium was collected and replaced after 30 and 60 min and collected at the end of the 90- min incubation, such as to be able monitor release during three "windows" (0-30, 30-60, and 60-90 rnin). Cells were extracted as described above either immediately following the pulse-label (0 min chase) or a t the end of the 90-min chase. In an additional experiment, the cells were handled as above except that the labeling time was limited to 2 min and the chase times to 2, 4, 6, 8, and 10 min.

Immunoprecipitation of insulin-related products was according to our previously published procedure (27). Samples were first prepuri- fied and concentrated by Sep-Pak (see above) and lyophilized. The samples were reconstituted in 200 p1 of 0.2 M glycine, 0.1% BSA, pH 8.8 (glycine/BSA), and 30 pl of anti-insulin serum (from Sigma) with

B. Schnetzler and P. A. Halban, unpublished data.

a total binding capacity of 300-400 ng of insulin added. After incu- bation for 1 h a t room temperature, 9 mg of protein A-Sepharose (Pharmacia, Zurich, Switzerland) in 200 p1 of glycine/BSA with 0.5% Nonidet P-40 (Fluka, Buchs, Switzerland) was added, and the samples were incubated for a further 30 min a t room temperature with frequent mixing (vortex mixer). Immune complexes bound to the protein A-Sepharose were pelleted by centrifugation (10 s a t 11,000 X g using a microcentrifuge), and the pellets were washed twice with glycine/BSA/Nonidet P-40. Finally, immunoprecipitated products were displaced from the immune complex by the addition of 200 p1 of 1 M acetic acid, 0.1% BSA. After a final centrifugation as above, the supernatants were again purified by Sep-Pak (as above) in order to remove IgG (which would otherwise saturate the column used for HPLC) from the samples. After lyophilization and reconstitution, the samples were analyzed by HPLC.

Degradation of Exogenous Labeled Proinsulin or Insulin-In order to prepare labeled rat proinsulin I1 or insulin 11, isolated rat islets were pulse-labeled for 1 h with either [3H]leucine alone or in combi- nation with [35S]methionine. After a 1 h chase, the islets were ex- tracted in 1 M HCl, 0.1% BSA, and the extract was injected onto HPLC using the system described above. Fractions eluting as insulin I1 or proinsulin I1 were pooled. After evaporating the acetonitrile, the pooled fractions were lyophilized, reconstituted in 0.1% trifluoroacetic acid, 0.1% BSA, and concentrated by Sep-Pak (see above). The specific radioactivity of the purified, concentrated material was esti- mated by dividing the radioactivity by the immunoreactivity. An aliquot of the labeled insulin or proinsulin was added to the medium bathing the transfected FA0 cells at the start of the incubation period. Samples were taken a t 1 and at 2 h of incubation, concentrated by Sep-Pak, and then analyzed by HPLC. The extent of degradation (if any) was based upon the recovery of radioactivity eluting as intact insulin or proinsulin.

RESULTS

Stimulation of Rat Proinsulinllnsulin Production by Dexamethasone

Cells stably transfected with rat proinsulin I1 (pDOL-rINS) were selected in G418. Of those clones producing insulin immunoreactive material (IRI), the highest level observed was of the order of 1 ng/106 cells released to the medium over 24 h. The murine Moloney leukemia virus LTR used to drive the insulin gene in the DOL-INS construct has been reported to be susceptible to stimulation by glucocorticoids. The trans- fected cells were therefore exposed to dexamethasone to see whether insulin output could be increased. In order to estab- lish optimum conditions for stimulation of IRI output, cells were incubated for 24 h with graded doses of dexamethasone. IRI release was stimulated at concentrations as low as lo-' M, with maximum stimulation achieved a t 5 X M (data not shown). The time course of stimulation of IRI release was next studied by incubating the cells with (or without, control) 5 X M dexamethasone and measuring IRI in the medium and in the cells at 1, 3, 6, 12, and 24 h (Fig. 1). There was a detectable stimulation of IRI release, as from 3 h, and by 24 h the accumulation of IRI in the medium bathing the cells was some 30 times greater in the presence of dexamethasone. The cellular content of IRI was also affected by dexametha- sone (Fig. 1, upper panel), reaching a maximum at 6 h, but decreasing thereafter. In keeping with cells releasing proteins only via the constitutive pathway, the cellular IRI content was low, as compared with that released, regardless of whether the cells had been incubated with dexamethasone or not (note the different scales in Fig. 1, upper and lower panels). The increase in IRI output could be attributed to a parallel increase in insulin mRNA, with the maximal effect seen at 6 h, just as for the increase in IRI content (data not shown). Preliminary experiments evaluating the effects of dexamethasone on in- sulin mRNA levels in the presence of actinomycin (an inhib- itor of transcription) or of cycloheximide (an inhibitor of translation) indicate that the increase in insulin mRNA levels

Proinsulin Processing in Transfected Hepatoma Cells 14631

1 2 3 415 6

30 t t t t t , A

1 3 6 12 24

INCUBATION TIME (hours)

FIG. 1. Stimulation by dexamethasone of immunoreactive insulin production by F A 0 cells transfected with pDOL-rINS (rat proinsulin 11). Transfected FA0 cells were incubated with (“treated”) or without (“control”) 5 X M dexamethasone for up to 24 h. At set times (as indicated), culture medium was collected, and the cells were extracted in acid prior to radioimmunoassay. A , immunoreactive insulin in cell extracts; B, immunoreactive insulin released to the medium. Note the different scales used in panels A and B.

is indeed due to a stimulation of transcription, but additional effects on mRNA stability cannot as yet be excluded. Insulin production from cells transfected with pRSV-hINS was not sensitive to dexamethasone.

Analysis of Insulin Immunoreactive Material by Reversed Phase HPLC

The anti-insulin serum used does not discriminate between proinsulin, conversion intermediates, and insulin, and it is thus not possible to distinguish between them by radio- immunoassay. This can, however, be achieved by reversed phase HPLC.

FA0 Cells Expressing Rat Insulin II-Cells were incubated in the presence of 5 X M dexamethasone for 6 h to stimulate IRI production. The medium was changed, and the cells were incubated for a further 2-h period in the continued presence of dexamethasone. The 2-h medium was collected and the cells were extracted in 1 M acetic acid, 0.1% BSA. Both the medium and the cell extracts were prepurified by passage through a CIS Sep-Pak cartridge and then injected onto HPLC. Each fraction was subjected to radioimmuno- assay and the HPLC system was standardized using authentic rat insulin 11, des-64,65 and des-31,32 split proinsulin 11, and proinsulin 11. Rat proinsulin I1 and related peptides are uniquely sensitive to oxidation during handling under acidic conditions due to the methionine residue at position 29 of the B-chain (26). The oxidized standards were therefore also injected onto HPLC. The profile of immunoreactive products released to the medium by cells transfected with pDOL-rINS is shown in Fig. 2, upper panel. According to their elution times relative to the standards, the following major peaks were identified: rat insulin 11, oxidized proinsulin 11, oxidized des-64,65 split proinsulin I1 and proinsulin 11, and des-64,65 split proinsulin 11. Des-64,65 proinsulin is the partially proc- essed conversion intermediate cleaved only between the C-

0 20 40 60 t

FRACTION NUMBER

FIG. 2. HPLC analysis of immunoreactive products released to the medium or retained within F A 0 cells transfected with pDOL-rINS (rat proinsulin 11). Transfected FA0 cells were first incubated for 6 h with 5 X lo-’ M dexamethasone to increase immu- noreactive insulin production. After changing the medium, the cells were then incubated for a further 2 h, after which the cells were extracted and both medium cell extracts analyzed by HPLC. Each fraction eluting from HPLC was subjected to radioimmunoassay. A , medium; B, cell extract. The elution times for standards injected immediately after the two samples are shown by the arrowheads: 1, oxidized rat insulin 11; 2, rat insulin 11; 3, oxidized rat proinsulin 11; 4 , oxidized des-64,65 split rat proinsulin 11; 5, rat proinsulin 11; 6, des-64,65 split rat proinsulin 11. For each product, the oxidized form arises from oxidation of MetBz9 (see Ref. 26).

peptide and the insulin A-chain. Although des-31,32 split proinsulin I1 (split between the B-chain and C-peptide) co- elutes with oxidized proinsulin 11, these two products can be resolved by complete oxidation of the sample, since oxidized des-31,32 split proinsulin I1 elutes some 10 min earlier than oxidized proinsulin 11. In contrast to culture medium, HPLC analysis of cell extracts revealed only a single major immu- noreactive peak coeluting with intact proinsulin I1 (Fig. 2, lower panel), with only a suggestion of the putative products of conversion.

When samples of culture medium were analyzed by HPLC using another well characterized elution buffer system (50 mM phosphoric acid, 100 mM sodium perchlorate, and 10 mM heptane sulfonic acid) (6, 25), the putative conversion inter- mediate eluted before intact proinsulin, again corresponding to the behavior of des-64.65 split proinsulin. The fully proc- essed insulin-like product was similarly found coeluting with rat insulin I1 standard. Since this HPLC system allows for good separation of intact insulin from mono- (or di-) arginyl insulin (the product of endoproteolytic cleavage a t both proin- sulin junctions, but without subsequent trimming of C-ter- minal basic residues by the action of a carboxypeptidase), it would appear that any endoproteolytic conversion event is followed by the action of a carboxypeptidase-like enzyme.

FA0 Cells Expressing Human Insulin-Cells were incu- bated for 8 h; after each 2-h period, the medium was collected and changed. The total (pooled) medium was prepurified by passage through CIR Sep Pak cartridges and then injected onto HPLC. Each fraction was analyzed by radioimmunoassay, and the HPLC system was standardized using authentic hu- man insulin, proinsulin, and the conversion intermediates.

The profile of immunoreactive products detected in the medium is shown in Fig. 3. According to the elution time relative to standards, three major peaks were identified hu-

14632 Proinsulin Processing in Transfected Hepatoma Cells

1 2 3 4

t t t t

- P 6 8 1 " 1

K

I" 0 20 40 60 80 I -

FRACTION NUMBER

FIG. 3. HPLC analysis of immunoreactive products released to the medium from FA0 cells transfected with pRSV-hINS (human proinsulin). Transfected cells were incubated for four successive 2-h periods. The media taken at the end of each period were pooled and then analyzed by HPLC. Each fraction eluting from HPLC was subjected to radioimmunoassay. The elution times for authentic human standards are shown by the numbered arrowheads: 1, insulin; 2, des-31,32 split proinsulin; 3, des-64,65 split proinsulin; 4, proinsulin.

man insulin, des-31,32 split human proinsulin, and human proinsulin. There was no detectable des-64,65 split human proinsulin, which unlike its rat counterpart, elutes before proinsulin using either HPLC system.

The antiserum used in the radioimmunoassay recognizes both insulin and proinsulin, but with unequal cross-reactivity. Using human proinsulin and insulin as standards, it has been estimated that the assay detects proinsulin with 50% cross- reactivity on a weight basis and with 75% cross-reactivity on a mol/mol basis relative to insulin. Similar cross-reactivity between rat insulin and proinsulin is assumed. The relative amounts of proinsulin and putative conversion intermediates shown in Figs. 2 and 3 are therefore underestimated by approximately 50% (the data having been presented on a weight basis uncorrected for cross-reactivity). The putative insulin peak is thus only a minor component.

Estimation of Degradation of Rat Proinsulin II and Insulin 11 in the Culture Medium

The generation of immunoreactive products other than proinsulin in the medium bathing the cells could have been due either to intracellular processing before release or to degradative events in the medium after the release of intact proinsulin. To distinguish between these two possibilities, cells were treated with dexamethasone and incubated for 2 h exactly as described above, except for the addition of labeled rat proinsulin I1 or insulin I1 to the medium during the 2-h incubation. The specific radioactivity of the labeled proinsulin (3750 cpm/ng of IRI) allowed for its addition without contrib- uting in a significant fashion to the total immunoreactivity in the medium. That of the labeled insulin (375 cpm/ng of IRI), however, was considerably lower. Thus, the cells released 120 ng of IRI during the 2-h incubation and 4 and 40 ng of labeled proinsulin or insulin, respectively, were added to each dish. Aliquots of medium were taken at 30, 60, and 120 min and then handled as described above. HPLC analysis showed that the labeled proinsulin was quite unaffected by the 2-h incubation with the cells, and the total proinsulin-associated radioactivity recovered from each equal volume of medium injected to HPLC was 7300, 7100, and 7300 cpm after 30,60, and 120 min. Of greater significance was the observation that only intact proinsulin I1 (and its oxidation product, which elutes some 20 min before the unoxidized form) (26) was seen eluting from HPLC (Fig. 4, left-hand panel). When labeled rat insulin 11, rather than proinsulin, was added to the me-

PROINSULIN INSULIN

1 2 3 4 5 6 3 4 5 6

3000 , I 7300 cprn

2000 - 30 mln - 1000-

E .- L

0 0 - I I .+, rC L c 7100 cpm g 2000: 60 mln

E € 1000-

u O " , 1 . f i n .

2 2000-

0 n 7300 cpm

I 120 mln

1000 -

o w . . , . h , . c . 7 0 20 40 60 80 100 0 20 40 60 80 100

FRACTION NUMBER

FIG. 4. Fate of exogenous. labeled

0 20 40 60 80 100

FRACTION NUMBER rat Droinsulin I1 or in-

sulin 11, added to themedium bathing transfected FA0 cells. Transfected FA0 cells (pDOL-rINS) were handled as described in the legend to Fig. 2, except that either labeled rat proinsulin I1 (left- handpanel) or insulin I1 (right-handpanel) was added to the medium at the start of the 2-h incubation period. Aliquots of medium were analyzed by HPLC at 30, 60, and 120 min. The total radioactivity associated with proinsulin or insulin (and their oxidized forms) is given for each time. The identification of standards (injected onto HPLC immediately after the samples) is described in the legend to Fig. 2.

dium, the results were different (Fig. 4, right-hand panel). The total radioactivity associated with insulin (or oxidized insulin) decreased with time (8900,8300, and 7600 cpm at 30, 60, and 120 min of incubation), indicating progressive degra- dation. There was, however, no generation of peaks eluting later than insulin. The degradation of insulin was not appar- ent if insulin was incubated in medium alone (i.e. without cells) (data not shown)

It is concluded from these experiments that proinsulin is not degraded by the cellular environment. In an additional series of experiments, transfected FA0 cells were exposed to dexamethasone for 6 h, but IRI release was then monitored for 2 h with the addition to the medium of a mixture of protease inhibitors consisting of 100 pM leupeptin, 20 milli- trypsin inhibitory units/ml aprotinin, 10 pM E64, 50 pM pepstatin, 10 p~ N-tosyl-L-phenylalanine chloromethyl ke- tone (each from Sigma) and 100 p~ phenylmethylsulfonyl fluoride (from Fluka, Buchs, Switzerland). HPLC analysis of products released from the cells was indistinguishable from that shown in Fig. 2 (data not shown). Since most proteases would have been inactive in the presence of the inhibitors, these data support the contention that extracellular degra- dation cannot be responsible for generation of the putative proinsulin conversion products. The generation of immuno- reactive material other than intact proinsulin must therefore be due to intracellular events. By contrast, insulin itself is subject to limited degradation in the medium, an event that will result in the underestimation of the amount of any fully processed insulin release from the transfected cells, but which does not lead to the generation of immunoreactive peaks eluting from HPLC later than insulin itself.

Proinsulin Processing in Transfected Hepatoma Cells

Kinetics of Generation of Putative Conversion Products and of Their Release (Pulse-Chase Experiments)

Measuring products by radioimmunoassay does not allow for the study of the kinetics of synthesis, conversion, and release. For this purpose, cells transfected with pDOL-rINS (after pretreatment for 6 h with dexamethasone) were labeled for 15 min and then subjected to a chase incubation. Release of labeled products was monitored between 0 and 30, 30 and 60, and 60 and 90 min. Cell extracts (taken either immediately after the pulse-label, or at the end of the 90-min chase) and medium were first prepurified by Sep-Pak. Proinsulin and related peptides were then immunoprecipitated. Much as for the HPLC analysis of immunoreactive products released from the cells, the major labeled, immunoprecipitable products coeluted with rat proinsulin 11, des-64,65 split proinsulin I1 and insulin 11, as well as their oxidized forms (Fig. 5). The relative amounts of radioactivity in each peak were essentially constant for each time period. Analysis of the cell extracts (Fig. 6) showed labeled intact proinsulin I1 as the major form eluting from HPLC, even when the cells had been extracted after 90 min of chase, with only a minor peak coeluting with des-64,65 split proinsulin 11.

The kinetics of release of labeled immunoreactive products was in keeping with the constitutive secretory pathway. As shown in Table I, 13.6% of cellular immunoprecipitable ra- dioactivity was found in the medium after only 30 min of chase, and 38.9% had been released by 90 min. The total immunoprecipitable radioactivity recovered in cells plus me- dium at 90 min was 100% of that found in the labeled cells at

1 2 3 4 5 6

300 t t t # t 0-30 rnin

0 1 0 20 40 60 80

FRACTION NUMBER I

FIG. 5. HPLC analysis of immunoprecipitable radioactive products released from pulse-labeled cells during a 90-min chase. Transfected FA0 cells (pD0L-rINS) were stimulated for 6 h with 5 X 10” M dexamethasone. After the cells had been labeled with [%]methionine for 15 min, they were incubated for 90 min in unlabeled medium (chase incubation). The chase medium was col- lected at 30, 60, and 90 min, and replaced with fresh medium a t 30 and 60 min (allowing for analysis of products released from 0 to 30, 30 to 60, and 60 to 90 rnin). Following immunoprecipitation using anti-insulin serum, samples were analyzed by HPLC, and the radio- activity in each fraction was measured. The identification of stand- ards (injected onto HPLC immediately after the samples) is described in the legend to Fig. 2.

t t t t t t 1 2 3 4 5 6

3000 I 1

14633

-I 0 20 40 60 80 100

FRACTION NUMBER

FIG. 6. HPLC analysis of immunoprecipitable radioactive products in cells after a 15-min pulse-label and at the end of a subsequent 90-min chase incubation. Transfected FA0 cells were handled as described in the legend to Fig. 5 . Some cells were extracted immediately ( 0 min) ; the remainder were extracted at the end of the 90-min chase (90 min) . Following immunoprecipitation, samples were analyzed by HPLC, and the radioactivity in each fraction was measured. The identification of standards (injected onto HPLC immediately after the samples) is described in the legend to Fig. 2.

TABLE I Fate of prelabeled proinsulin-related peptides i n F A 0 cells transfected

with pDOL-rINS (rat proinsulin II ) Cells were labeled (15 min, [%]Met), and then incubated for up

to 90 min of chase as described in the legend to Fig. 5. The cells were extracted after the pulse-label (0 min of chase) or at the end of the chase. Medium was collected and replaced a t 30 and at 60 min, such as to monitor products released from 0 to 30, 30 to 60, and 60 to 90 min. Labeled products in cells and medium were immunoprecipitated using anti-insulin serum and then analyzed by HPLC (see Fig. 5). Radioactivity eluting as insulin, proinsulin, and conversion interme- diate was summed and expressed as a percentage of that found in the cells after pulse-label. NA, not analyzed.

Labeled proinsulin-related peptides

Time Released Cells Total

Per 30 min Cumulative rnin % pulse-labeled cells

30 NA 13.6 13.6 NA 60 NA 14.6 28.2 NA 90 61.5 10.7 38.9 100.4

0 100 - - 100

the start of the chase, indicating that neither intra- nor extracellular degradation of such products occurs to a signif- icant extent.

In one additional experiment, cells were pulse-labeled for only 2 min and then chased for 2, 4, 6, 8, or 10 min. HPLC analysis of cells extracts revealed no conversion products a t any time (data not shown). These data confirm that conver- sion is a relatively late event in the secretory route.

DISCUSSION

The major findings of this study are that both rat proinsulin I1 and human proinsulin can be processed by transfected FA0 cells to products which, upon analysis by reversed phase HPLC, behave as a proinsulin conversion intermediate and fully processed insulin. This finding will be discussed in the light of what is already known about proprotein conversion in the constitutive pathway and what has been observed by

14634 Proinsulin Processing in Transfected Hepatoma Cells

others upon expression of proinsulin in cells that, like FA0 cells, are believed to secrete proteins only via this pathway.

Of the numerous secretory proteins handled by the consti- tutive secretory pathway, many are synthesized as proproteins (14, 15); the most well characterized example is proalbumin (29). The same applies to some cell surface receptors, but for such proteins, this is the exception rather than the rule, and only a handful of cases have been documented (30), including the insulin receptor (31). The processing of proprotein to the mature protein is an intracellular event, the endoprotease(s) responsible being most probably furin (or PACE) (17, 18) and/or other related members of a family of mammalian proteases displaying structural and functional homology with the yeast KEXZ gene product. Although furin, like PC2 and PC3, the endoproteases responsible for prohormone conver- sion in the regulated pathway (9-13), cleaves preferentially at pairs of basic residues (14, 15), its substrate specificity is clearly different, and indeed, a preference for pairs of basic residues preceded by another basic residue in the -4 position (ie. 4 residues removed from the site of cleavage at the carboxyl side of the pair of basic residues) has been demon- strated (14, 15, 32-34). One immediate consequence of the different substrate specificities of furin on the one hand, and PC2/PC3 on the other, is that prohormones usually converted by the latter in neuroendocrine secretory granules may or may not be converted by the constitutive pathway endopro- teases, depending upon the structural features of the sites of conversion.

The study of the production of proinsulin-related peptides following transfection of the insulin gene is often limited by a low level of expression. Indeed, basal levels of immunoreac- tive material released from the FA0 cells transfected with the cDNA encoding rat proinsulin I1 were too low to allow for any further characterization. The DOL expression vector used for transfection employs the Moloney murine leukemia virus LTR to drive the insulin gene (22). Unlike the mouse mam- mary tumor virus LTR (35) and some other murine leukemia virus LTRs (36) that have well defined glucocorticoid re- sponse elements, the Moloney murine leukemia virus LTR has not been shown as yet to have any such element (37). Despite this, it has been shown that expression of foreign genes under the control of the DOL vector LTR can be increased by dexamethasone in ,transfected hepatoma cells (28), and we confirm this observation in the present study. The striking increase in insulin mRNA levels in transfected FA0 cells following exposure to dexamethasone was accom- panied by a 30-fold increase in the release of immunoreactive products, thus achieving levels that allowed for the prelimi- nary characterization of these products. The increased im- munoreactive insulin output of the transfected cells following exposure to dexamethasone further allowed for experiments on the kinetics of release of newly synthesized products (pulse-chase protocol) in which the cells were labeled for only 15 min (and in one experiment for only 2 min), as compared with 4 h (38,39), 6 h (40), or even 15-16 h (41,42) in previous studies on constitutive cells transfected with the insulin gene.

The level of expression of the insulin gene in FA0 cells transfected pRSV-hINS was higher than that observed using pDOL-rINS. This level could not, however, be further ele- vated by treatment with dexamethasone, and in order to accumulate sufficient material for HPLC analysis, it was necessary to pool the medium from successive 2-h incubation periods.

The expression of an insulin gene in cells presumed to release proteins only through the constitutive pathway has been studied previously by several groups (38-42). As in the

present study, each of these studies concluded that proinsulin was, as expected, rapidly released after its synthesis, without any sizable storage compartment. This confirms that proin- sulin is indeed being handled by the constitutive pathway and that simply expressing a protein known to be targeted to the regulated pathway in endocrine cells is not sufficient in itself to allow for regulated release in cells normally only expressing the constitutive pathway. In contrast to the present findings, however, these earlier studies concluded that there was no conversion, or even partial conversion, of proinsulin. Al- though extracellular proinsulin conversion has not been doc- umented (other than as a result of residual activity of pro- teases used to prepare adipocytes (43)), this has been shown for formation of a C-terminal fragment from glucagon (44) and for the conversion of proneuropeptide Y by CHO cells (45). It was therefore first necessary to ensure that the puta- tive conversion products were not generated in the medium after the release of intact proinsulin. To test this, exogenous, labeled rat proinsulin I1 was added to the transfected FA0 cells. The labeled proinsulin was apparently unaffected during a 2-h incubation, since there was no loss of radioactivity from the proinsulin peak and no generation of secondary peaks. By contrast, when labeled rat insulin I1 was added to the medium surrounding the transfected cells, there was progressive deg- radation (approximately 15% over 2 h); no peaks eluting from HPLC later than insulin were generated by such degradation. Degradation of extracellular insulin by FA0 cells was ex- pected, since these cells have been shown to avidly bind and degrade insulin by a receptor-mediated pathway (46), and this will result in an underestimate of the total amount of fully processed insulin released from the transfected cells. The relative stability of proinsulin, but not of insulin, in the presence of FA0 cells can be attributed to the low affinity of the insulin receptor toward the precursor (43,471.

Our finding of conversion of a prohormone by the consti- tutive pathway machinery is not unique, since others have also found such conversion upon transfection of “constitutive” cells with the gene for a variety of prohormones (14, 15, 48). It is possible that proinsulin processing can only arise in the particular FA0 hepatoma line used in this study but not in the transformed lines used in earlier studies (COS cells (39, 40), L cells (41), CHO cells (42), and AGMK cells (3811, thereby explaining the discrepancy between our finding of proinsulin conversion and the lack of any conversion observed in these previous studies. Another possible explanation for this discrepancy could be the analytical methods employed and/or the particular insulin cDNA used.

In earlier studies on constitutive cells transfected with the insulin gene (38-42), radiolabeled peptides were analyzed by polyacrylamide gel electrophoresis under denaturing and re- ducing conditions. Under such conditions, proinsulin migrates as a relatively tight band and is quite easy to detect. Insulin, however, which will be split into its component chains, is less easy to detect since the A- and B-chains tend to run as quite broad bands even with a high percentage of acrylamide. If insulin was only a minor component of products released to the chase medium, as in the present study, it would not necessarily be possible to detect it by gel electrophoresis and fluorography. Furthermore, and perhaps of greater signifi- cance, proinsulin and conversion intermediates cannot be resolved from each other if electrophoresis is performed under reducing conditions in the presence of sodium dodecyl sulfate. Any partial conversion of proinsulin would thus have been overlooked in these studies.

The insulin gene selected for transfection is an additional and important factor. Furin, which as mentioned earlier is

Proinsulin Processing in Transfected Hepatoma Cells 14635

TABLE I1 Sequence of human and rat proinsulins at the B-chain/C-peptide and C-peptide/A-chain junctions

Basic amino acids are underlined. Note the presence of a basic residue in the -4 position (relative to the point of cleavage C-terminal to a pair of basic residues) at both junctions for rat proinsulin I, at the C-peptide/A-chain junction for rat proinsulin 11, and at the B-chain/C- peptide junction for human proinsulin, thereby providing a preferred (but not obligatory) consensus sequence for cleavage by furin.

B-chain C-peptide A-chain Rat I ...= Ser Arg Arg Glu ... & G l n Lys Arg Gly . . . . Rat I1 . . . Met Ser Arg Arg Glu . . . & Gln Lys Arg Gly . . . . Human ...& Thr Arg Arg Glu . . . Leu Gln Lys Arg Gly . . . .

one of the endoproteases believed to be implicated in the processing of proproteins in the constitutive pathway, has a rather well defined substrate specificity. The result is that not all prohormones normally converted in secretory granules by PC2/PC3 can be converted by this endoprotease (14, 15, 48-50). Of particular relevance to this discussion is the pref- erence displayed by furin for an RXR/KR consensus sequence at the site of cleavage (14, 15, 32-34), although it must be stressed that this is not an obligatory motif, since proalbumin is certainly cleaved by furin and only displays a pair of basic residues without an additional basic residue at the -4 position (39). A comparison of the cleavage sites linking the insulin chains and C-peptide of rat or human proinsulins reveals interesting differences in this context (Table 11). The pre- ferred consensus sequence for furin is displayed at the C- peptide/A-chain junction of both rat proinsulins but not human. By contrast, human and rat proinsulin I, but not rat proinsulin 11, have a potentially attractive cleavage motif at the B-chain/C-peptide junction (albeit with lysine rather than arginine at the -4 position). Based upon such gross inspection alone and without taking into account any other potentially important structural features of the proinsulin molecules un- der scrutiny (4, 19, 52), it would be predicted that rat proin- sulin I1 should be well cleaved by furin at the C-peptide/A- chain junction but only poorly at the B-chain/C-peptide junc- tion. This would lead to generation of des-64.65 split proin- sulin, which is precisely the intermediate believed to be re- leased from the transfected FA0 cells in this study. Human proinsulin, on the other hand would be predicted to be more susceptible to cleavage in the constitutive pathway at the B- chain/(=-peptide junction rather than at the C-peptide/A- chain, which is again supported by the data showing produc- tion of des-31,32 but not des-64,65 split human proinsulin in the transfected FA0 cells.

When human proinsulin was used as a substrate for prote- olysis by solubilized liver vesicles (53), there was generation of a conversion intermediate presumed to be the des-31,32 split form. There was, however, no evidence for production of fully processed insulin. It remains to be seen whether conver- sion in the constitutive pathway is the responsibility of only one or of several endoproteases. Furthermore, until the precise conversion compartment and its local environment have been established, it will be difficult to compare results obtained using solubilized vesicles (53) with those obtained in living cells as in the present study.

The kinetics of proinsulin conversion observed in the pulse- chase experiments are intriguing. The relative amounts of radiolabeled intact proinsulin, putative conversion interme- diate, and insulin were seen to be constant in the medium regardless of the time of chase (ie. 0 to 30 min, 30 to 60 min, or 60 to 90 min). Furthermore, hardly any conversion products were seen in the cell extracts even after 90 min of chase. The same was true for immunoreactive (unlabeled) products. When cells were only labeled for 2 min, followed by a chase up to 10 min, there was no suggestion of conversion. Taken together, these data suggest that conversion is a very late

event in the constitutive pathway, occurring just before re- lease. Studies on the conversion of other proproteins (33,45, 54-57), and most notably proalbumin (55-58), have led to the same conclusion. In these studies, it was similarly found that the processed forms were hard to detect within the cell but were a major (if not the sole) secreted product. The precise subcellular compartment involved in conversion in the con- stitutive pathway remains to be identified. Certainly, furin can be localized to the Golgi complex (57, 59, 60) or to secretory vesicles (51), but this does not mean that it is active within these compartments or, if so, only there. Furthermore, and as mentioned earlier, other endoproteases may be active elsewhere in the constitutive pathway. Studies depending upon the use of inhibitors to arrest the movement of propro- teins at discrete steps again suggest that the trans-Golgi, or a compartment immediately derived from it, is the site of con- version in the constitutive pathway (51, 53, 55, 61). In yeast, the KEXB protease has been localized to the Golgi complex (62) to the possible exclusion of secretory vesicles (63), and when KEXB was expressed in 3T3 cells, it was again localized to the Golgi complex (64). The unequivocal identification of the conversion compartment will depend upon the develop- ment of an appropriate cell-free system.

In conclusion, this study demonstrates for the first time that proinsulin can be converted in the constitutive pathway, albeit not as efficiently as in secretory granules. These find- ings are of importance for understanding the behavior of proinsulin if ever it is not appropriately targeted to the pancreatic B-cell secretory granules for conventional conver- sion and regulated release (4, 19), as well as for the interpre- tation of studies on proinsulin production by transfected non- B-cells (4, 19).

Acknowledgment-We thank Carole Caunes-Bourigault for expert technical assistance and Dr. Wieland Huttner for useful comments.

REFERENCES

2. Burgess, T. L. & Kelly, R. B. (1987) Annu. Reu. Cell Biol. 3 , 243-293 1. Kelly, R. B. (1985) Science 230,25-32

4. Halban, P. A. (1990) Trends Endocrinol. Metab. 1, 261-265 3. Rhodes, C. J. & Halban, P. A. (1987) J. Cell. Biol. 105,145-153

5. Orci, L., Ravazzola, M., Amherdt, M., Madsen, 0.. Vassalli, J.-D. & Perrelet,

6. Davidson, H. W., Rhodes, C. J. & Hutton, J. C. (1988) Nature 333,93-96 7. Fuller, R. S., Sterne, R. E. & Thorner, J. (1988) Annu. Reu. Physzol. 60,

8. Bathurst, I. C., Brennan, S. O., Carrell, R. W., Cousens, L. S., Brake, A. J.

9. Seidah, N. G., Gaspar, L., Mion, P., Marcinkiewicz, M., Mbikay, M. &

10. Smeekens, S. P., Avruch, A. S., LaMendola, J., Chan, S. J. & Steiner, D.

11. Korner, J., Chun, J., Harter, D. & Axel, R. (1991) Proc. Natl. Acad. Sci.

12. Smeekens, S. P. & Steiner, D. F. (1990) J. Biol. Chern. 265,2997-3000 13. Shennan, K. I. J., Smeekens, S. P., Steiner, D. F. & Docherty, K. (1991)

14. Hutton, J. C. (1990) Curr. Opin. Cell Bid. 2,1131-1142 15. Barr, P. J. (1991) Cell 6 6 , 1-3 16. Gold, G., Wieland, F. T. & Grodsky, G. M. (1988) Biochern. Biophys. Res.

17. Wise, R. J., Barr, P. J., Wong, P. A., Kiefer, M. C., Brake, A. J. & Kaufman, Cornrnun. 166,457-462

18. Van de Ven, W. J. M., Voorberg, J., Fontijn, R., Pannekoek, H., van den R. J. (1990) Proc. Natl. Acad. Sci. U. S. A. 87,9378-9382

Ouweland, A. M. W., van Duijnhoven, H. L. P., Roebroek, A. J. M. & Siezen, R. J. (1990) Mol. Biol, Rep. 14 , 265-275

A. (1985) Cell 442,671-681

345-362

& Barr, P. J. (1987) Science 235,348-350

Chrbtien, M. (1990) DNA Cell Biol. 9,415-424

F. (1991) Proc. Natl. Acad. Sei. U. S. A. 88, 340-344

U. 5. A. 88, 6834-6838

FEES Lett. 284,277-280

19. Halban, P. A. (1991) Dtabetologm 34, 767-778

14636 Proinsulin Processing in 20.

21.

22.

23.

24.

25. 26.

27. 28.

29. 30.

31.

32.

33.

34.

35.

36.

38. 37.

39. 40. 41.

Gross, D. J., Halban, P. A,, Kahn, C. R., Weir, G. C. & Villa-Komaroff, L. (1989) Proc. Natl. Acud. Sei. U. S. A. 86, 4107-4111

Gross, D. J., Villa-Komaroff, L., Kahn, C. R., Weir, G. C. & Halban, P. A. (1989) J. Biol. Chern. 264,21486-21490

Korman, A. J., Frantz, J. D., Strominger, J. L. & Mulligan, R. C. (1987) Proc. Natl. Acud. Sci. U. S. A. 84,2150-2154

Herbert, V., Lau, K.-S., Gottlieh, C. W. & Bleicher, S. S. (1965) J. Clin. Endocrinol. 25, 1375-1384

Cohen, R. M., Given, B. D., Licinio-Paixao, J., Provow, S. A,, Rue, P. A., Frank, B. H., Root, M. A,, Polonsky, K. S., Tager, H. S. & Rubenstein, A. H. (1986) Metabolism 35, 1137-1146

Sizonenko, S. V. & Halban, P. A. (1991) Biochem. J. 278,621-625 Gross, D., Skvorak, A., Hendrick, G., Weir, G., Villa-Komaroff, L. &

Halhan,. P. A. & Wollheim, C. B. (1980) J. Biol. Chem. 255,6003-6006 Hatzoglou, M., Park, E., Wynshaw-Boris, A., Kaung, H.-L. C. & Hanson,

Judab, J. D. & Quinn, P. S. (1978) Nature 271,384-385 Herz, J., Kowal, R. C., Goldstein, J. L. & Brown, M. S. (1990) EMBO J. 9,

Halhan, P. A. (1988) FEBS Lett. 241,205-208

R. W. (1988) J. Biol Chem. 263,17798-17808

1769-1 776 Hedo, J. A., Kahn, C. R., Hayashi, M., Yamada, K. M. & Kasuga, M. (1983)

Bentley, A. K., Rees, D. J. G., Rizza, C. & Brownlee, G. G. (1986) Cell 45,

- . - - - . . - J. Biol. Chem. 258,10020-10026

343-348 Foster, D. C., Sprecher, C. A., Holly, R. D., Gamhee, J. E., Walker, K. M.

Hosaka, M., Nagahama, M., Kim, W.-S., Watanahe, T., Hatsuzawa, K., & Kumar, A. A. (1990) Biochemistry 29, 347-354

Ikemizu. J., Murakami. K. & Nakavama, K. (1991) J. Biol. Chem. 266 ,

." ~~.

12127-12130

H. E. (1975) Cell 6,299-305

~.

Ringold, G. M., Yamamoto, K. R., Tomkins, G. M., Bishop, J. M. & Varmus,

Celander, D. & Haseltine, W. A. (1987) J. Virol. 61, 269-275 Overhauser, J. & Fan, H. (1985) J. Virol. 54, 133-144 Gruss, P. & Khoury, G. (1981) Proc. Nutl. Acad. Sci. U. S. A. 78, 133-137 Lomedico, P. T. (1982) Proc. Natl. Acad. Sci. U. S. A. 79,5798-5802

Moore, H.-P. H., Walker, M. D., Lee, F. & Kelly, R. B. (1983) Cell 35,531- Laub, 0. & Rutter, W. J. (1983) J. Biol. Chem. 258,6043-6050

538

Tran 42.

43.

44.

45.

46.

47. 48. 49.

50.

51. 52.

53.

55. 54.

56.

57.

58.

59.

60.

61.

62.

64. 63.

sfected Hepatoma Cells umn D. Orci L., Ravazzola, M. & Moore, H.-P. H. (1991) J. Cell. Biol.

Duckworth, W. C., Peavy, D. E., Hamel, F. G., Liepnieks, J., Brunner, M.

Blache, P., Kervran, A,, Dufour, M., Martinez, J., Le-Nguyen, D., Lotersz- R., Heiney, R. E. & Frank, B. H. (1988) Biochem. J. 255,277-284

tajn, S., Pavoine, C., Pecker, F. & Bataille, D. (1990) J. Biol. Chem. 265, 21514-21519

Sch'ellerup Wulff, B., O'Hare, M. M. T., Boel, E., Theill, L. E. & Schwartz, 'I! W. (1990) FEBS Lett. 261,101-105

Backer, J. M., Kahn, C. R. & White, M. F. (1990) J. Biol. Chem. 265, 14828-14835

Podlecki, D. A., Frank, B. H. & Olefsky, J. M. (1984) Diabetes 33, 111-118 Thomas, G., and Thorne, B. A. (1988) Annu. Reo. Physiol. 50,323-332 Hatsuzawa, K., Hosaka, M., Nakagawa, T., Nagase, M., Shoda, A., Murak-

ami, K. & Nakayama, K. (1990) J. Biol. Chem. 265, 22075-22078 Zollinger, L., Noel, G., Des Parois, L., Sales, V., Crine, P. & Boileau, G.

(1988) Mol. Cell. Endocrinol. 58,31-41 Brennan, S. 0. & Peach., R. J. (1988) FEBS Lett. 229,167-170 Weiss, M. A,, Frank, B. H., Khait, I., Pekar, A., Heiney, R., Shoelson, S.

Misumi, Y., Takami, N. & Ikehara, Y. (1984) FEBS Lett. 175,63-67 Brennan, S. 0. & Peach, R. J. (1991) J. Biol. Chem. 266,21504-21508

Oda, K., Misumi, Y. & Ikehara, Y. (1983) Eur. J. Biochem. 135,209-216 Misumi, Y., Ohkubo, M., Sohda, M., Takami, N., Oda, K. & Ikehara, Y.

Misumi, Y., Oda, K., Fujiwara, T., Takami, N., Tashiro, K. & Ikehara, Y. Redman, C. M., Avellino, G. & Yu., S. (1983) J. Biol. Chem. 258, 3446-

Bresnahan, P. A., Leduc, R., Thomas, L., Thorner, J., Gibson, H. L., Brake,

Mizuno, K., Nakamura, T. & Matsuo, H. (1989) Biochem. Biophys. Res.

Redman, C. M., Banerjee, D., Manning, C., Huang, C. Y. & Green, K.

Fuller, R. S., Brake, A. J. & Thorner, J. (1989) Science 246,482-486 Redding, K., Holcomh, C. & Fuller, R. S. (1991) J. Cell. Biol. 113 , 527-538

Germain, D., Zollinger, L., Racine, C., Gossard, F., Di nard, D., Thomas,

113,9Q7-996

E. & Neuringer, L. J. (1990) Biochemistry 29,8389-8401

(1990) Biochem. Biophys. Res. Commun. 171,236-242

(1991) J. Biol. Chem. 266, 16954-16959

3452

A. J., Barr, P. J. & Thomas, G. (1990) J. Cell. Biol. 111, 2851-2859

Commun. 164, 780-787

(1978) J. Cell. Biol. 77, 400-416

D. Y., Crine, P. & Boileau, G. (1990) Mol. Endocrinof 4 , 1572-1579