Discrete Structural Domains Determine Differential Endoplasmic ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1994 by The American Societyfor Biochemistry and Molecular Biology, Inc

Vol. 269, No. 51, Issue of December 23, pp. 3211032119, 1994 Printed in U.S.A.

Discrete Structural Domains Determine Differential Endoplasmic Reticulum to Golgi Transit Times for Glucose Transporter Isoforms* (Received forpublication, August 17, 1994, and in revised form, September 22, 1994) Richard C. Hresko, Haruhiko Murata,Bess Adkins Marshall, and Mike Mueckler From the Department of Cell Biology and Physiology, Washington University School of Medicine, St. Louis, Missouri 63110

The rate of movement of the glucose transporter iso- tical at the amino acidlevel, these isoforms have different forms Glutl and Glut4 from the endoplasmic reticulum tissue distributions, subcellular localizations, kinetic charac(ER) to the Golgi apparatus was investigated by pulse teristics, and regulatory properties(1,2). Glutl has a ubiquilabeling and monitoring endoglycosidase H resistance tous tissue distribution and is presumably responsible for basal in mRNA-injectedxenopus oocytesand in 3T3-Ll adipo- glucose transport in many cell types. Glut4 is found exclusively cytes, a cell line that naturally expresses both transin muscle and fat andis mainly responsible for the increase in porter isoforms. Despite their high degree of sequence identity, Glutl and Glut4 exhibited dramatically differ- sugar uptake in response to insulin through its redistribution ent transit times. The tlh values for ER to Golgi transit from an intracellular compartment to thecell surface. Recently, domains that confer the intracellular targetingof for Glutl and Glut4 were e1 and 24 h, respectively, in oocytes and “5 and 20 min, respectively, in 3T3-Ll adi- Glut4 have beenidentified through the useof GlutUGlut4 chipocytes. Pulse-chase in conjunction with sucrose den- meric transporters (5-9). After some controversy, it appears sity gradient analysis revealed that the rate-limiting that the carboxyl terminus of Glut4 is mainly responsible for its step in the ER to Golgi processing of Glut4 was exit fromintracellular targeting(5-7). I t is important in these studies to the ER and not retention in an early Golgi compartment. distinguish between authentic targeting and inefficient procWe analyzed the biosynthesis of Glutl/Glut4 chimeric essing of a chimera. For example, we have shown inXenopus transporters in Xenopus oocytes in order to determine oocytes that 67% of a Glutl/Glut4 chimera containing amino whether specific domains in Glutl and Glut4 were re- acids 1-132 of Glut4 is not fully glycosylated and is sequestered sponsible for their distinct transit times. The first exo- intracellularly while the remaining 33% is fully modified and facial glycosylated loop and the cytoplasmic carboxyl- correctly targeted (7). The fact that certain chimeras poorly are terminal domain of Glut4 were crucial for its delayed processed is surprising given the high degreeof homology beexit fromthe ER. The first transmembrane, the first exotween the two transporters, however, very little is actually facial, and the cytoplasmic COOH-terminal domains of known concerningthe processing of these two isoforms. Hudson Glutl were largely responsible for Glutl’s rapid processing in the ER. Some of the chimeric transporters were et al. (10) have shown through the rate of acquisition of en50% of chimeric mol- doglycosidase H (Endo H)’ resistance that Glutl and Glut4 not fully processed. Approximately have different ER to Golgi transit times when expressed in ecules containing the cytoplasmic COOH-terminal domain of Glutl and either the first transmembrane or transfected NIH-3T3 cells. Movement from the ER to theGolgi biosynthesis and first exofacial domain of Glut4 were retained in early is generally the rate-determining step in the Golgi compartments and prevented from complete mat- transport of secretory and membrane glycoproteins (11).This uration. Normalprocessing of these chimeras was phenomenon of differential transport ratesis not unique t o the achieved by replacing the cytoplasmic COOH-terminal glucose transporters. The class I histocompatibilty antigens, domain of Glutl with that of Glut4. These data suggest H-2Kk and H-Dk, have a 4-fold difference in their rate of exit that amino acidresidues within the glycosylated exofa- from the ER despite their greater than80% sequence identity cia1 loop and the cytoplasmic COOH terminus partici- (12). However, the exact structural basis for the different tranpate in a rate-limiting step in the folding of both Glutl sit times of these two molecules is not known. and Glut4 or could act as transient ER retention signals. The rateof ER to Golgi transport is generally related to the Additionally,these results show that even chimericmol- time requiredfor folding or assembly into theproper tertiary or ecules constructed from two highly homologous prooligomeric structure, both of which usually occur in the ER teins can exhibit aberrant folding and post-translational (13-15). Misfolded and improperly assembled proteins are reprocessing. tained in theER as aggregates and thenspecifically degraded, thus providing the cell with an efficient quality control mechanism. Protein misfolding, however, is also a common feature Glutl and Glut4 are two members of the facilitative glucose in normal protein synthesis. Newly made proteinswill fold and transporter family (1, 2). The Glut proteins are N-linked gly- refold until the properconformation isattained. Molecular coproteins that have 12 membrane spanning domains based on chaperone proteinsassist in this process by specifically binding hydropathy analysis (3). Recently, the 12-helical model has to incompletely folded or unassembled proteins, preventing been confirmed forGlutl through the use of glycosylation scan- their aggregation, and allowing them time to fold properly or 65%iden- oligomerize (16). Once correctly assembled, newly synthesized ning mutagenesis (4). AlthoughGlutl and Glut4 are

* This work was supported in part by National Institutes of Health Grant DK43695 and by the Diabetes Research and TrainingCenter at Washington University Medical School. The costs of publication of this article were defrayed in part by the payment of page charges. This article must thereforebe hereby marked “aduertisernent”in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The abbreviations used are: Endo H, endoglycosidase H; BSA, bovine serum albumin;ER, endoplasmic reticulum; PAGE, polyacrylamPCR, polymeride gel electrophoresis; PBS, phosphate-buffered saline; ase chain reaction;UTPaS,uridine 5’-[a-35Slthiotriphosphate; BIP, immunoglobulin binding protein.

32110

proteins dissociate from the chaperones and exit the ER. In addition to chaperones, N-linked oligosaccharides often playa role in glycoprotein folding, oligomerization, and retention in the ER (17). Since Glutl and Glut4 are extremely homologous proteins, they are believed to possess verysimilar secondary and tertiary structures. The disparity in their ER t o Golgi transit times suggests that certain proteins that reside in theER are able to detect subtle differences in their structure. By studying the processing of Glutl and Glut4 we had a unique opportunity to identify transporter domains that have a direct influence on their respective tertiary and possibly quaternary structure. The three-dimensional structure and oligomeric state of a protein isdirectly related t o its function, subcellular targeting, and stability. The aim of the presentstudy was to determine whether specific domains in Glutl and Glut4 are responsible for their differential ER to Golgi transit times by analyzing the processing of GlutllGlut4 chimeric transporters in Xenopus oocytes. We found that the rate of movement of Glut4 from the ER was strongly influenced by its firstexofacial domainand its cytoplasmic COOH terminus. The first transmembrane, the first exofacial, and the cytoplasmic COOH-terminal domains of Glutl were predominately important in maintaining its very rapid ER to Golgi transit time. The data suggest that these domains participate in a rate-limiting step in thefolding of the facilitative transporters or may act as transient ER retention signals. EXPERIMENTALPROCEDURES Construction of Chimeric Dansporters-Human Glutl (3) and rat Glut4 (18) cDNAs were previously subclonedinto the oocyte expression vector pSP64T (19). Some of the chimeric glucose transporters were synthesized by recombinant polymerase chain reaction as described by Higuchi (20). Fragments were made using “outside” primers corresponding to the 3’- or 5’-untranslated region of the cDNAs constructed with a short multiple restriction site sequence and with ”inside”recombinant primers containing 20 nucleotides each from Glutl and Glut4. Human Glutl cDNA in pSPGT and rat Glut4 cDNA in pIRGT were used as the templates (7). Polymerase chain reactions were carried out using Vent DNApolymerase (NewEngland Biolabs, Beverly,MA) as described previously (7). Chimeric cDNA were subcloned into pSP64T in an orientation compatible with the SP6 RNA polymerasepromoter. Sequences wereverified by the Sequenase 2.0 kit (U. S. Biochemical Corp.). Spurious mutations were corrected using the Clontech Transformer Site Directed Mutagenesis kit (ClontechLaboratories, Palo Alto, CAI. Spurious mutations were observed a t a frequency of -1 in 2500 base pairs amplified. Additional chimeric transporters were constructed by ligating the amino- and carboxyl-terminal regions of two previously synthesized and sequenced transporters using endogeneous restriction sites. The sequence of the ligated junctions were then verified. Nomenclature of the Glucose Dansporter Chimeras-Chimeras contained sequences from both human Glutl and rat Glut4 transporters. The nomenclature for the chimeras have the formats #:# or #:#-# or #:#-#-#. The first number, left of the colon, denotes the isoform composition of the chimera from the amino to the carboxyl terminus. Thus, a “14”chimera begins with Glutl sequence and ends with Glut4. A”414” chimera begins and ends with Glut4 and has an internal Glutl sequence. The numberb) to the right of the colon represent the amino acid numbeds) of the transporter contributing the extreme NH,-terminal sequence at the junctionb) separating the two isofom sequences. For example, the chimera 41:289 contains the first289 aminoacids of Glut4 and the remainder is Glutl. The chimera 141:116-272 contains the following isoform sequences: Glutl 1-116, Glut4 133-287, and Glutl 272-492. The numbers 11G272 are theamino acidresidues of Glutl at each of the two isoformjunctions. Residues correspondingto Glutlwere selected in the name of the chimera since the extreme amino-terminal sequence in this particular chimera is Glutl. To help clarify the chimeric nomenclature, a schematic topological diagram for each of the chimeras is presented in Fig. 3. Synthesis of mRNA-Glutl, Glut4, or chimeric transporter mRNA were synthesized from pSP64TcDNAs linearized at a unique restriction site inthe polylinker using the MEGhcriptSP6 in vitro Transcription kit (Ambion, Austin, TX). Manufacturer’s conditions were modifiedby

the addition of 4 mM GpppG and 0.625 pCVm1 [35SlUTPaS( h e r s h a m ) and by decreasing the GTP concentration to 1 mM. Microinjection and MetabolicLabeling of Oocytes-Oocytes were surgically removedand prepared from adult female Xenopus laeuis(Nasco, Fort Atkinson, WI) as described previously(7). Stage V and VI oocytes were injected with 50 nl of Glutl, Glut4, or the appropriate chimeric transporter mRNA, incubated for 2 hat 18 “C in Barth’s modifiedsaline containing 0.5% BSAand 50 pg/ml gentamicin, and then labeled with 2 mCVm1 Tran3%-label ( E N ) in Barth’s modified saline for 1h at 18 ’C. After the 1-h pulse labeling, oocytes wereextensively washed with iced Barth’s modified saline supplemented with 5 gfliter BSA, 15 m~ cold methionine, and 2 mM cysteine and then incubated in the same media at 18 “C fora selected periodof time. At the appropriate time, 15 oocytes were transferred to 1.5 ml of homogenization buffer(10 mM HEPES, pH 7.4, 250 mM sucrose, and a mixture of protease inhibitors) and then frozen at -80 “C until intracellular membranes were prepared. The following protease inhibitors were used: 1 pg/ml leupeptin, 1 pdml antipain, 1 pg/ml benzamidine, 5 pg/ml trypsin inhibitor, 1 pdml chymostatin, 1 pg/ml pepstatin A, and 0.5 m~ phenylmethylsulfonyl fluoride. Preparation of Intracellular Membranes and Immunoprecipitationof Dansporters from Xenopus Oocytes-Intracellular membranes from 15 oocytes wereprepared as described previously(7,211. Membranes were solubilized in 80 pl of 1%SDS in phosphate-buffered saline (PBS), pH 7.4, and then diluted with 1ml of 1%Triton X-100, 1%deoxycholate,1% BSAin PBSsupplemented with the same mixture of protease inhibitors described above. Transporters were immunoprecipitated overnight at 4 “C with 1 pl of F349 rabbit antiserum raised against a synthetic peptide correspondingto the carboxyl-terminal 16 residues of rat Glut4 (22), 10 pg of the IgG fraction from rabbit antiserum F350 raised against a synthetic peptide corresponding to the carboxyl-terminal 15 residues of human Glutl (22), or 1.25 pgof the IgG fraction of a monoclonal human Glutl antibody (23, 24) (a kind gift of Dr. G. Lienhard, Dartmouth Medical School). Polyclonalantibodies were preadsorbed to 40 pl of Protein A-agarose and the monoclonal Glutl antibody was precoupled to40 pl of goat anti-mouse IgG affinity gel (Cappel,Organon Teknika Corp.,West Chester, PA). After washing the resin three times each with the immunoprecipitationbuffer and PBS, transporters were eluted with 50 pl of 1%SDS. Endoglycosidase H Digestion of Immunoprecipitated Dampoders20 pl of eluted transporter were digested with EndoglycosidaseH for 1 h at 37 ”C with the addition of 0.5 plof 3 M sodium acetate, pH 5.5, and 0.3 milliunits of enzyme (Boehringer Mannheim). Samples were then Polyacrylamide gels were extensively subjected to SDS-PAGE. destained and then radiochemically enhanced by incubation at room temperature for 15 min in 1 M salicylicacid. Autoradiograms were scanned and then analyzed by densitometry using Scantastic (Second Glance) and Image 1.32f software on a MacIntosh IIci computer. Sucrose Density Gradient Analysis-Sucrose density gradient analysis of intracellular oocyte membrane fractions were carried out as described previously by Jaunin et al. (25). Intracellular membranes from 100 oocytes in 1 ml of gradient buffer (10 n m Tris-HC1, pH 7.4, 1 mM EDTA, 250 mM sucrose) were loaded onto 12 ml of 12-50% (w/v) sucrose gradients. Gradients were centifuged for4 hat 35,000 rpmin a Beckman SW41rotor and then fractionated into 12 0.75-ml aliquots by collecting fromthe bottom of the tube. The bottom 2 ml and the top of the gradient were discarded. Half of each fraction was solubilizedwith the addition of 1 ml of 1.375%Triton X-100, 1.375%deoxycholate, 1.375%BSA in PBS, pH 7.4. Solubilizedtransporters were then immunoprecipitated overnight using the same procedure described above for the Xenopus oocytes.Transporters were eluted from the resin with SDS sample buffer and then subjected to SDS-PAGE. Metabolic Labeling of 3T3-Ll Adipocytes”3T3-Ll fibroblasts were grown to confluence and 48 h later subjected to differentiation as described previously (26). Ten days after differentiation was initiated, 3T3-Ll adipocytes were metabolicallylabeled using a slightly modified version of a procedure described by Hudson et al. (10) for labeling transfected NIH-3T3 cells. Adipocytes grown in 2.5-cm tissue culture plates were washed with PBS and then incubated for 2 h in methionineand cysteine-free Dulbecco’s modified Eagle’s medium. Cells were then incubated for 10 min in the same media supplemented with Tran3%label (0.5 mCi/ml) and 2.5% BSA,washed three times with ice-cold PBS, and then chased with Dulbecco’s modified Eagle’s medium containing 10%fetal bovine serum, 0.2 mg/ml methionine, and 0.2 mg/ml cysteine. The chase was terminated by washing with ice-cold PBSsupplemented with 0.2 mg/mlMet and 0.2mg/mlCys and the protease inhibitor mixture described above. Cells were then solubilized directly with 1% Triton X-100, 1%deoxycholate, 1%BSAin PBScontaining Met and Cys,

Domains Involved in Glucose Dansporter Processing

32112

to differences in the signal among the different chase times due amount of transporter synthesized. Since each time point was EndoH - + - + - + - + - + - + - + - + based on the amountof transporters immunoprecipitated from only 15 oocytes, variability in the amount of mRNA injected and or differences between the oocytes themselves can greatly affect the amount of transporter synthesized. This, however, had no effect on the determinationof transit times, since these 68 43 GLUT 4 results were based on the fraction of transporter (not the absolute amount) that was Endo H resistant at each chase time. In addition, the transit time was largely independent of the absolute level of synthesized transporter. Identification of Domains Responsible for the Differential ER to Golgi P a n s i t Times of Glutl and Glut4"Glutl/Glut4 chic 0 0.8 meras were used t o localize domains responsible for the differential ER toGolgi transit times. The strategy wasto identify a Q, 0.6 chimera that contained the maximum amount of Glut4 sev) v) quence and yet was stillprocessed at the same rate as Glutl, $ 0.4 GLUT 4 and vice versa. Fig. 2 illustrates the predicted transmembrane P topology of Glut4 based on the hydropathy plot of the deduced aminoacid sequence. The amino acids that are circled are found in both Glutl andGlut4. The nomenclature of the glucose LL ovL-" transporter chimerasis discussed under "Experimental Proce0 10 20 30 40 50 60 70 80 dures." To help clarify the results a schematic topological diaChase lime (hrs) gram isshown in Fig. 3 for each of the chimeras that illustrates FIG.1.ER to Golgi transit times for wild-type Glutl and Glut4. the isoform composition of that particular chimera. Glutl seOocytes injected with Glutl and Glut4mRNAs were pulse-labeled with quence is depicted by a solid black line and the Glut4 sequence 'Pran"S"labe1 for 1 h and then chased for varying periods of time. Lais shown by a dotted gray line. beled transporters were immunoprecipitated, treated without (-) and Initially two half-chimeras were analyzed using the same with (+) endoglycosidase H, and then subjected to SDS-PAGE a s described under "Experimental Procedures." The numbers on the left rep- pulse-chase and EndoHprocedure described above for the resent the mobility of molecular weight standards. Results were quan- wild-type transporters. Chimera 14:271 had a Glutl aminotitated by scanning densitometry and plotted as Fully Processedfl'otal terminal half and a Glut4 carboxyl-terminal half, while chiuersus Chase Time. Fully processed was the amount of labeled transporter found in the upper Endo H-resistant band. These results were mera 41:289 had the converse composition. Schematic topologrepresentative of 15 independent experiments. ical diagrams depicting the isoform composition of 14:271 and 41:289 are shown in Fig. 3B. The results (Fig. 4) indicate that to a the ER toGolgi transit timefor 14:271 was in between that of andtheproteaseinhibitors.Cellswerescraped,transferred Beckman Microfuge tube, and centrifuged a t 165,000 x g for 30 min. Glutl and Glut4. The t%was 5 h, and 48 h was required for Glutl and Glut4were immunoprecipitated from the high-speed super- complete Endo H resistance. In contrast, the processing patnatant using the same procedure described for the Xenopus oocytes except that samples were preincubated for 4 h a t 4 "C with 40 1.11 of tern of 41:289 was different from that of the wild types and Protein A-agarose to reducenonspecific binding. Endo H digestion was 14:271. Slightly morethan half of the molecules appeared tobe then carried out as described above. processed normally as seen by the appearance of the higher molecular weight Endo H-resistant band. The ty, from ER to RESULTS Golgi for the fraction that was normallyprocessed was 3 h. The Glutl andGlut4 ExhibitDifferent ER to Golgi P a n s i t Times other 45% of 41:289, however, was never fully glycosylated and in Xenopus Oocytes-ER to Golgi transit times of Glutl and therefore had a lower molecular weight even after 72 h of chase. Glut4 were monitored through the acquisition of Endo H re- These resultssuggested that at least half of 41:289 reached the sistance ina pulse-chase type experiment.Oocytes were micro- medial-Golgi and was fully modified, but the remaining fracinjected with either Glutl or Glut4 mRNA, pulse-labeled with tion was retained in either the ER or an early Golgi compartTran3?3-label for 1 h, and then chased for varying periods of ment(s). Because the transit timesfor both chimeras were intime by incubation in a medium containing unlabeled methio- termediate in value relative to the wild-type transit times, we nine and cysteine. Glutl and Glut4 transporters were immu- concluded that amino acid residues in both halves of the molnoprecipitated from solubilized intracellular membranes and ecule are important in determining specific the rate of exit from then tested for Endo H resistance. The results of a typical the ER for Glutl and Glut4. experiment are shown in Fig. 1. Similar to the findings of Next we analyzed the transportersby quarters startingwith Hudson et al. (10) using transfectedNIH-3T3 cells, Glutl and a Glutl backbone and changing either the first (41:132), the Glut4 had very different transit times from the ER to Golgi. second (141:116-272), the third (141:271-3601, or the fourth Immediately after the pulse, Endo H digestion resulted in a (14:359) quarter of the molecule to Glut4. A topological diagram shift in the molecular mass of both Glutl and Glut4 from 43 to of each quarter chimera is illustrated Fig. in 3C and thepulse38 kDa, indicating that both proteins were core glycosylated chase results are shown in Fig. 5. When the first quarter of and still present in the ER. However, after a 1-h chase,greater Glutl was changed to Glut4 (41:132), the processing pattern than 80% of the Glutl molecules had reached the Golgi as was almost identicalto thehalf-chimera 41:289 in that approxiindicated by the appearance of the diffuse higher molecular mately 45% of the molecules were never fully processed while mass Endo Hresistant band(54 kDa) andby 3 h essentially all the remaining55% of the labeled transporter was chasedinto a of the Glutl transporters were fully glycosylated. In contrast, higher molecular weight EndoH-resistant form. This findingis 24 h were required for 50% of the Glut4 molecules to become consistent with ourprevious subcellular fractionationand conEndo H resistant and 3 days for all the Glut4 tobe fully proc- focal immunofluorescence microscopy results that showed this essed. Note that there wassome variability in the transporter particular chimera is intracellularly sequestered (7). In con-

0

1

3

6 12 24 48 72 hrs

" " " "

" " " "

b@@'

+


32113

E N

D

COOH FIG.2. Sequence and proposed topologicalstructure of Glut4. Proposed orientation of Glut4 in the bilayer based on its homology to Glutl and on the hydropathy plot of its deduced amino acid sequence (3).The 12 membrane spanningdomains are numbered. The circled amino acids are found in both Glutl and Glut4.

trast, the rates of acquisition of Endo H resistance for 141:116- Glutl and infact the th was slightly slower than 14:271.These 272 and 141:271-360 were similar to that of Glutl. The th results indicate that itwas not possible to eliminate amino acid values were approximately 2 and 1h, respectively, and 90% of residues from contributing t o the respective isoform transit the transporters were fully glycosylated by 4 h. Pulse-chase times based on data from chimeras with processing times inresults for 14:359 were almost identical to the half-chimera termediate in value relative to the wild-type transit times. 14:271; the t% was 5 h and all transporters were completely Therefore amino acids were only considered not important in processed in 2 days. From these results we concluded that determining the respective transit times with chimeras that new chimeras unique amino acids found in the first and fourth but not the have either wild-type transitrate.Thus, second and third quartersof Glutl and Glut4 were responsible (141:271-474 and 141:271-464) were analyzed t o determine for their differential ER to Golgi transit times. In addition, even whether additional Glutl sequence at theextreme COOH terthough Glutl and Glut4 are highly homologous, it is possible to minus was necessary t o obtain Glutl-like processing (see Fig. construct chimeras that exhibit aberrant processing. In the 3E for topological models). We discovered that as more Glutl case of 41:289 and 41:132, having Glut4 sequence in the first sequence was substituted for that of Glut4 at the extreme quarter of the transporter and at least some Glutl sequence in COOH terminus of 14:271, the processing rate approached the carboxyl-terminal half prevented a fraction of the molecules Glutl. Since the ER to Golgi transit time for 141:271464 was from being fully processed. close to wild-type Glutl asshown in Fig. 6B, we concluded that Next we attempted to identify the amino acids locatedwithin the entire cytoplasmic COOH-terminal domain in addition to the last quarter of Glut4 that contribute to its slower transit amino acid residues in the first half of the molecule contributed time. We initially examined the cytoplasmic COOH-terminal to the processing time of Glutl and that residues in the fourth domain, because this is the most divergent region within the quarter other than the cytoplasmic COOH terminus did not fourth quarter. Three new chimeras were constructed that were specifically influence Glutl’s transittime. Similarly, a converse composed of all Glutl sequenceexceptfor the cytoplasmic chimera with Glut4 sequence in the first quarterand cytoplasCOOH-terminal domain(14:450), thelast 19 amino acids mic COOH-terminal tail was processed like Glut4 as discussed (14:473), or the last 11 amino acids (14:481). The topological below in Fig. 7B. diagrams are shown in Fig. 30. Quantitation of the pulse-chase Earlier pulse-chase results revealed that a fraction of certain results (Fig. 6 A ) indicate that all three new chimeras were chimeric transporters such as 41:289 and 41:132 werenot fully processed at the same rate as 14:271 and 14:359 (shown for glycosylated, suggesting that those molecules wereretained in comparison by the dashed and dotted lines, respectively). The an early biosynthetic compartment. We found that two addity, values were all 5 h and essentially all transporters were tional chimeras (41:466 and 141:33-67) had similar processing fully processed by 48 h. Given the striking similarity between phenotypes as illustrated inFig. 7A. 41:466 was entirely Glut4 the five plots, it was possible that theGlut4 processing domain except for the cytoplasmic COOH-terminal domain and in the fourth quarter of the molecule was restricted to its last11 141:33-67 was entirely Glutl except for its first exofacial doamino acids. To test this hypothesis we constructed a chimera main (see Fig. 3F for topological diagrams). All four of these (141:271-482) that had a 14:271 backbone but whose last 11 chimeras had some Glut4 sequence in the first quarter of the amino acids were changed to Glutl in anticipation that this molecule in conjunction with a Glutl COOH terminus. Since chimera would be processedlike Glutl (see Fig. 3E for topolog- the last set of chimeras (Fig. 6B) suggested that the entire ical model). Quantitation of the pulse-chase experiment (Fig. cytoplasmic COOH-terminal domain maybe important in proc6B) indicated that this chimera was not processed as fast as essing, we replaced the Glutl COOH terminus of 41:132 with

32114


FIG.3. Schematictopologicaldiagrams of theGlutl/Glut4chimeric transporters. A schematic diagram of each chimeric transporter is shown that illustrates its isoform composition. Glutl sequence is depicted by a solid black line and theGlut4 sequence is shown by a dotted gray line. The chimeric nomenclature is discussed under "Experimental Procedures.'' The chimeras are grouped (A-I)as they are presented in each of the subsequent figure panels. The values within the parentheses are the t~ values for ER to Golgi transit of each of the chimeric transporters. An asterisk signifies that only 50% or less of that particular chimeric transporter was fully processed.

D

14:-14:481

F

14:473

1

41 :466

(5h)(5h)

(5h)

I

1

141:116-451 4141:24-133-466 4141:24-83-466 4141:45-83-466

that of Glut4. Quantitation of the pulse-chase results (Fig. 7B) indicated that the chimera 414:132-467 was processed normally and at a rate almost identicalto thatof Glut4 (see Fig. 3G for topological diagram). Therefore, the cytoplasmic COOH terminus of Glut4 is necessary to insurecomplete processing of a chimera if the chimera has additional Glut4sequence within the first quarter of the molecule. This data also confirms that the processing domains of Glut4 are located within the first quarter and thecytoplasmic COOH terminus. The processing of three new chimeras was monitored to further delineate the amino-terminal Glut4 processing domain . diagrams of these chimeras areshown in (Fig. a)Topological Fig. 3H. The transit time for 1414:12-67-450, which was entirely Glutl except for the first transmembrane, the firstexofacial, and the cytoplasmic COOH-terminal domains, was almost identical to thatof Glut4. The t~ was 24 h and almost all of the transporters were fully processed by 48h. Chimera 1414:33-67-450 was all Glutl except for the first exofacial domain and the cytoplasmic COOH terminus and was also processed exactly like Glut4.The last chimera 1414:12-3-50, which is composed of Glutl sequence except for the first transmembrane andcytoplasmic COOH-terminal domains,was processed faster than Glut4 with an approximate t% of 10 h and with 90% of the transporters completely glycosylated by 48 h. These results indicate that the firstexofacial domain and the intracellular COOH terminus of Glut4 are responsible for its specific ER to Golgi transit time.

1

L

I

Next we constructed converse chimeras to determine the processing domains for Glutl (Fig. 8B).Topological diagrams of these chimeras are shown in Fig. 31. Chimera 141:116-451, which was all Glut4except for the first quarter and the cytoplasmic COOH-terminal domain, was processed a t a rate approaching wild-type Glutl with a t~ of about 1 h. The first quarter of Glutl wasdissected further by monitoring the processing of the three additional chimeras. The ta for 4141324133-466 was nearly as fast as Glutl and since this construct was identical t o 141:116-451 except for the cytoplasmic aminoterminal domain being Glut4, the NH, terminus was eliminated from contributing to the ER to Golgi transit time of Glutl. This wasconsistent with our finding that thewild-type processing times did not change when the cytoplasmic NH,terminal domains were swapped (data not shown). Chimera 4141:24-83-466 was entirely Glut4 except for the first membrane spanning domain, the first exofacial loop, and the cytoplasmic COOH terminus. This chimerawas processed slightly a t~ of 2 h. Thelast chimera,4141:45slower than Glutl with 83-466, was all Glut4 except for its first exofacial domain and cytoplasmic COOH terminus. This chimera showed the same aberrant processing described earlier whereapproximately 50% of the protein was not fully glycosylated. Similar to the other chimeras thatexhibit aberrant processing, 4141:45-83466 had a Glutl COOH terminus and some Glut4 sequence in the first quarter of the molecule. From these results we concluded that the first transmembrane, the first exofacial, and

Domains Involved in Glucose l’kansporter Processing 68

-

0

1

3

7

13

18

41:132

24 48hrs

“ ” ” ”

14:271

43-

32115

141:271-360

0 1 3 6 13244872hrs

“ ” ” ”

0 2 4 7 13182448hrs

68-

-

43--,-,aia~$illllBl

ENDOH-+-+-+-+-+-+-+-+ 68

-

Z~p.*me--o.r*

. I

41 :289

43-

1

3

6

13

24

48

““&.&*W68 I

” ” ” “

0 2 4 7 13182448hrs

72hrs

-43

” ” ” ”

0 2 4 7 13182448hrs

141:116-272

” ” ” ”

0

,-68 43

-+-+-+-+-+-+-+-+ENDOH-+-+-+-+-+

14:359

c 0 41:289 141:116-272

n

0

14359

0.4

0.2

Chase Time (hrs)

-=

FIG.4. ER to Golgi transit times of 14271 and 41:289. Oocytes

LL

0

10 20 30 40 50 60 70 80

injectedwith 14:271 and 41:289 mRNAs werepulse-labeledwith Tran3%-label for 1 h and then chased for varying periods of time. Labeled transporters were immunoprecipitated, treated without (-) and with (+) endoglycosidase H, and then subjected to SDS-PAGE a s described under “Experimental Procedures.” The nomenclature of the chimeras 14:271 and 41:289 are described under “Experimental Procedures.” Schematic topological diagrams depicting the isoform composition of 14:271 and 41:289 are shown in Fig. 3B. Results from the pulse-chase and Endo H digestion were quantitated by scanning densitometry.Theseresultswererepresentative of threeindependent experiments.

the cytoplasmic COOH-terminal domains are largely responsible for Glutl’s ER to Golgi transit time. Thus, similarregions of Glutl and Glut4 appear tobe responsible for their respective transit times. the Biosynthesis Exit from the ER isthe Rate-limiting Step in of Glutl and Glut4”Since the assay used to monitor ER to Golgi transport was Endo H resistance, which is a result of sugar modifications that occur in the medial Golgi (27), we sought to determine whether the rate-limiting step was exit from the ER rather than the rate of oligosaccharide processing per se or exit from a subsequent early Golgi compartment. Therefore, we repeated the pulse-chase experiment for Glutl and Glut4 and then fractionated purified intracellular membranes on a sucrose density gradient. Oocytes injected with Glutl and Glut4 mRNAs were pulse-labeled for 15 min and 1h, respectively, and then chased for varying periods of time. Transporters were immunoprecipitated from fractions collected from each of the gradients(Fig. 9). Immediately after the pulse (0 min chase time), Glutl and Glut4 transporters were predominantly located in the higher densityfractions of the gradient (between 1.12 and 1.15 g/ml), consistent with localization in the ER. This was confirmed by the anti-BIP Western blot shown at the bottom of Fig. 9. BIP, a 70-kDa heat-shock protein that residesexclusively in theER lumen (281, colocalized to the same region of the gradient. Transporters in thisregion were completely Endo H sensitive, which is again consistent with localization to the ER (datanot shown). With increasing time, both Glutl and Glutclabeled proteins were chasedinto a lower density region of the gradient. The most intense signal was between 1.11and 1.12 g/ml for Glutl andbetween 1.10 and 1.05 g/ml for Glut4. At longer chase times, labeled transporters increased in massfrom 43 to50 kDa whileremaining at the same average density. Endo H digestion of labeled transporters in this region of the gradientrevealed that the highermolecular weight form of the transporters was Endo H resistant while the

0

20

40

60

80

0

10 20 30 40

50

Chase Time (hrs)

FIG. 5. ER to Golgi transit timesof 41:132,141:116-272,141:271360, and 14359. Oocytes injected with 41:132, 141:116-272, 141:271360, and 14:359 mRNAs were pulse-labeled with Tran3%-label for 1 h and then chased for varying periods of time. Labeled transporters were immunoprecipitated, treated without(-) and with (+) endoglycosidase H, and then subjected toSDS-PAGE a s described under “Experimental Procedures.”Thenomenclature of thechimerasisdescribedunder “Experimental Procedures” and schematic representations of the chimeras are shown in Fig. 3C. Results from the pulse-chase and Endo H digestion were quantitated by scanning densitometry. Results of 41:132, 141:116-272,141:271-360, and 14:359 were representative of three, one, one, and two independent experiments,respectively.

lower form was still sensitive (data not shown). These results were all consistent with newly synthesized transporters leaving the ER and being fully processed in theGolgi. A comparison of the Glutl gradientsdifferent at chase timesindicated that by 1h some of the Glutl transporters hadreached the Golgi and were fully modified. By 3 h all of the Glutl molecules were completely processed. The t h from ER to Golgi is slightly longer using this method as opposed to Endo H sensitivity (Fig. 11, which is due to the difference in pulse time (15 min uersus 1h). When a 1-h pulse was used in the sucrose density gradient experiment, the tM was similar to that measured by Endo H digestion (