The role of the carbohydrate chains of Galβ-1,4-GlcNAcα2,6-sialyltransferase for enzyme eactivity

Biochimica et Biophysica Acta, 1202 (1993) 325-330 © 1993 Elsevier Science Publishers B.V. All rights reserved 0167-4838/93/$06.00

325

BBAPRO 34573

The role of the carbohydrate chains of Gal/3-1,4-GlcNAca2,6-sialyltransferase for enzyme activity Darren G. Fast 1, James C. Jamieson and Gwen McCaffrey 2 Department of Chemistry, University of Manitoba, Winnipeg, Manitoba (Canada) (Received 28 September 1992) (Revised manuscript received 22 January 1993)

Key words: Sialyltransferase; Golgi complex; Glycanase; Endo F; Glycoprotein; Deglycosylation

Gal/3-1,4-GlcNAca2,6-sialyltransferase (CMP-N-acetylneuraminate:fl-galactosidea2,6 sialyltransferase, EC 2.4.99.1) is a glycoprotein containing carbohydrate chains of the complex type (Jamieson, J.C. (1989) Life Sci. 43, 691-697). The carbohydrate chains may be important for controlling the expression of sialyltransferase catalytic activity during transit of the enzyme from the rough endoplasmic reticulum to the Golgi complex where it is active as a membrane bound enzyme anchored to the luminal face. To study the role of the carbohydrate chains of sialyltransferase for enzyme activity, conditions were established in which the native enzyme was deglycosylated with N-Glycanase and endo F. It was found that Glycanase removed the carbohydrate chains from native sialyltransferase, but methanol or ethanol had to be present for rapid and complete deglycosylation. Presence of methanol or ethanol were not essential for removal of carbohydrate chains with endo F. There was a correlation between the loss of catalytic activity of sialyltransferase with increased deglycosylation. After deglycosylation with Glycanase for 18 h catalytic activity was largely eliminated and there was a reduction in molecular mass of about 5 kDa compared to the untreated enzyme when examined by immunoblot analysis; this reduction was identical to that found when the denatured enzyme was deglycosylated with Glycanase. At shorter times of incubation partially deglycosylated forms of the enzyme were detected. Complete deglycosylation of native or denatured sialyltransferase with endo F could not be achieved. However, incubation with endo F for 24 h resulted in a loss of catalytic activity of about 60%. Immunoblot analysis showed the presence of three forms of the enzyme corresponding in molecular mass to the native and deglycosylated enzyme and a third form corresponding to a partially deglycosylated enzyme. Sialyltransferase was also subjected to sequential treatment with exoglycosidases. Removal of NeuAc and Gal had little effect on catalytic activity, but subsequent removal of GIcNAc resulted in a significant loss in catalytic activity suggesting that the presence of the trimannose core with GlcNAc attached is important for the expression of catalytic activity. The presence of organic solvents during deglycosylation with Glycanase may be a useful method that can be applied to other glycoproteins.

Introduction Previous studies have shown that Gal/3-1,4-GlcNAca2,6-sialyltransferase is a m e m b r a n e - b o u n d enzyme located at the luminal face of the Golgi complex [1-5]. The enzyme is a glycoprotein containing at least two N-linked carbohydrate chains which appear to be of the complex type [1]. Glycoproteins like sialyltrans-

Correspondence to: J.C. Jamieson, Department of Chemistry, University of Manitoba, Winnipeg, Manitoba, Canada R3T 2N2. 1 Present address: Department of Biochemistry, University of Alberta, Edmonton, Alberta, Canada. Present address: Department of Medicine and Biochemistry, Washington University School of Medicine, St. Louis, MO, USA. Abbreviations used: NeuAc, N-acetylneuraminic acid; CMP-NeuAc, CMP-N-acetylneuraminic acid; GlcNAc; N-acetylglucosamine; GlycanaseTM; peptide:N-glycosidase F; endo F, endo-/3-N-acetylglucosaminidase F.

ferase are initially assembled in the rough endoplasmic reticulum in a form containing high-mannose oligosaccharide chains; the high-mannose chains are then processed to complex chains during transit through the channels of the endoplasmic reticulum and Golgi complex [6]. Sialyltransferase is believed to transit to the Golgi and anchor itself to the luminal face as a result of recognition between a linker anchor on the enzyme and a site on the Golgi m e m b r a n e [7-10]. However, the enzyme is not catalytically active in the endoplasmic reticulum and only achieves full activity when the enzyme locates itself in the trans compartment of the Golgi complex [8]. One model to explain why sialyltransferase is active in the Golgi, but not in the endoplasmic reticulum, is that the presence of the carbohydrate chains are important for the catalytic activity of the enzyme. This would explain why activity is not expressed during transit of the enzyme through the channels of the endoplasmic reticulum Golgi complex

326

Materials. CMP-[4,5,6,7,8,9-14C]N-acetylneuraminic acid (247 mCi/mmol) was from New England Nuclear (Lachine, Quebec, Canada); [125I]-protein A was from Amersham (Oakville, Ontario, Canada). N-Glycanase TM, endo-/3-N-acetylglucosaminidase F and Galfll-4GlcNAca2-6 sialyltransferase isolated from rat liver were obtained from Genzyme (Boston, MA, USA). Neuraminidase, from Clostridium perfringens and /3galactosidase from jack beans were obtained from Sigma (St Louis, MO, USA), N-acetyl-/3-D-glucosaminidase from Diploccus pneumoniae was from Boehringer-Mannheim (Laval, Qu6bec, Canada). Protein molecular mass standards were from Bio-Rad (Richmond, CA, USA) and were soybean trypsin inhibitor (21.6 kDa), carbonic anhydrase (31 kDa), ovalbumin (42.699 kDa) bovine serum albumin (66.2 kDa) and phosphorylase b (97.4 kDa). Human serum albumin (purity 99%, proteinase free) was obtained as a gift from Dr. R. Janzen of the Winnipeg Rh Institute. Antiserum containing antibodies against rat hepatic a2,6-sialyltransferase was obtained as a gift from Dr. J.C. Paulson, Department of Biological Chemistry, School of Medicine, University of California (Los Angeles, CA, USA). Asialo al-acid glycoprotein was prepared as previously described [11]. Nitrocellulose membranes were obtained from Fisher Scientific. Other chemicals were of analytical grade obtained from local suppliers.

protein), 0.6 /xg human serum albumin to protect against proteinase degradation, and up to 10% methanol or ethanol (which significantly enhanced removal of carbohydrate chains from the native enzyme, see Results) in a total volume of 30 ~1. Controls in which Glycanase, methanol or ethanol were omitted were set up in parallel. Incubation was for up to 18 h at 37°C and was stopped by transferring to ice. In some experiments the above volumes were scaled up as appropriate. For removal of carbohydrate chains from the native enzyme with endo F the incubation mixture contained 0.1 M sodium acetate (pH 4.5), 1.5% Triton X-100, 10 ~1 sialyltransferase (0.2 ~g enzyme protein), 0.6 ~g human serum albumin and 30 ~1 endo F (0.15 m U / ~ l ; see Genzyme technical bulletin) in a total volume of 50 /xl; incubation was for up to 24 h. For digestion with exoglycosidases the method followed that used for endo F, except that Triton X-100 was omitted and the buffer was adjusted to pH 5.5; digestion was with 1 mU neuraminidase for 18 h followed by 2 mU of galactosidase for a further 18 h. The reaction was stopped by adjusting the pH to 7.0 with imidazole buffer used for sialyltransferase assay. Sialyltransferase activity was assayed as before; controls consisted of sialyltransferase incubated in absence of glycosidases. For digestion with N-acetylhexosaminidase sialyltransferase was treated with neuaraminidase followed by galactosidase as above and then 3 mU of N-acetylhexosaminidase were added and incubated for 18 h with a control in which the glycosidase was omitted; the reaction was stopped and sialyltransferase was assayed as above. Denatured sialyltransferase was prepared by boiling for 3 min in the appropriate buffer for glycosidase digestion in presence of a 1.2-fold excess of SDS (w/w). Sialyltransferase was assayed using asialo al-acid glycoprotein as acceptor and examined by immunoblot analysis as previously described [5]. In one experiment samples of 2 /xg native and denatured sialyltransferase were treated with Glycanase as above for 18 h and subjected to electrophoresis along with a sample of untreated sialyltransferase. The samples were transferred to nitrocellulose and stained for carbohydrate using the Glycan detection kit supplied by Boehringer-Mannheim. Where appropriate, controls were set up with the denatured enzyme which contained methanol or ethanol. 1 U of sialyltransferase activity is defined as equal to the transfer of 1 pmol of NeuAc from CMP-NeuAc to acceptor protein per h.

Digestion of sialyltransferase with endo- and exo-glycosidases. Carbohydrate chains were removed from na-

Results

until oligosaccharide processing is close to completion. As part of our studies on sialyltransferase we have now examined the effect of removing the carbohydrate chains from the native enzyme with endoglycosidases and exoglycosidases to determine how this affects enzyme activity. Evidence is presented to show that catalytic activity of sialyltransferase is dependent on the presence of the oligosaccharide chains. Results of experiments using exoglycosidases suggest that the presence of the trimannose core with N-acetylglucosamine residues attached is important for full catalytic activity. In addition, a new method is described in which deglycosylation of native sialyltransferase is much more efficient in presence of methanol or ethanol. This may prove to be applicable to other native glycoproteins to determine the importance of oligosaccharide chains for biological activity. Materials and Methods

tive and denatured sialyltransferase by digestion with Glycanase or endo F. For removal of carbohydrate chains from the native enzyme with Glycanase the incubation mixture contained 0.2 M phosphate buffer (pH 8.6), 1.2 /.tl Glycanase (0.26 U//~I, Genzyme technical bulletin), 10 /~1 sialyltransferase (0.2 ~g enzyme

Effect of endoglycosidase digestion on sialyltransferase activity It was found that incubation of native sialyltransferase with Glyeanase for up to 18 h resulted in very little loss of carbohydrate or enzyme activity unless

327 methanol or ethanol were present. Fig. 1 shows the effect of the presence of methanol or ethanol in the hydrolysis mixture on the activity of sialyltransferase when incubated for up to 4 h in presence of Glycanase; the presence of methanol in the hydrolysis mixture was more effective than ethanol in causing a reduction of sialyltransferase activity. Fig. 2 shows that after about 18 h of incubation with Glycanase and 5% methanol sialyltransferase activity was almost eliminated. Control experiments in which sialyltransferase was incubated for up to 18 h in presence of methanol and ethanol alone showed that there was little loss in enzyme activity (Figs. 1 and 2). In addition, Glycanase was not deactivated prior to assay of the native sialyltransferase since sialyltransferase would have been denatured; although the acceptor was present in large excess it could have been deglycosylated during incubation and thus cease to act as as an acceptor. To test for this a sample of asialo oq-acid glycoprotein was incubated with Glycanase under the conditions used for complete deglycosylation of sialyltransferase and then boiled to destroy the Glycanase. The oq-acid glycoprotein was recovered and used in a subsequent sialyltransferase assay along with an untreated sample of acceptor; there was no difference between the two acceptors showing that the Glycanase did not have a significant effect on the acceptor under the conditions of the experiment.

100~ 80

~6o '~ 40 2G a

6

e

Methanol/Ethanol, ?~

lb

Fig. 1. Effect on sialyltransferase activity following incubation with Glycanase for 4 h in presence of different concentrations of methanol or ethanol. Sialyitransferase was incubated with Glycanase for 4 h in presence of different concentrations of methanol or ethanol. Incubations were stopped by transferring to ice and suitable volumes were assayed for sialyitransferase. The results are from a typical experiment. The 100% value represents 6200 U of sialyltransferase activity per sample; sialyitransferase activity in control experiments in which the enzyme was incubated in absence of Glycanase, but in presence of different concentrations of methanol (11); similar results were obtained when ethanol was present (not shown); sialyltransferase activity in presence of Glycanase and different concentrations of methanol (e); sialyltransferase activity in presence of Glycanase and different concentrations of ethanol (A). Control experiments in which sialyltransferase was incubated in presence of Glycanase, but without methanol or ethanol gave similar results to those found when Glycanase was not present. Each point represents the mean for three assays and reproducibility was within + 12%.

100' 8G

Y_ 4o 20

10

15

20

Time,h

Fig. 2. Effect on sialyltransferase activity following incubation with Giycanase for different times in presence of 5% methanol. Sialyltransferase was incubated with Glycanase for up to 18 h either alone or in presence of methanol. Incubations were stopped and sialyltransferase was assayed as described in Fig. 1. The results are from a typical experiment. The 100% value represents 7500 U of sialyltransferase activityper sample; sialyltransferase activityin presence (e), or absence ( • ) of methanol. Each point represents the mean for three assays and reproducibilitywas within + 12%. The above results indicate that hydrolysis of sialyltransferase with Glycanase results in loss of enzyme activity, but that methanol or ethanol are required in the hydrolysis mixture for rapid and complete hydrolysis of oligosaccharide chains to occur. Glycanase cleaves N-linked high mannose and complex chains at the glycosylamine bond between asparagine and GIcNAc leaving a terminal aspartic acid on the polypeptide chain. This will change the charge on the protein which could contribute to the loss of enzyme activity. In order to eliminate the appearance of the charge on aspartic acid as an explanation for loss in enzyme activity the experiments described in Figs. 1 and 2 were repeated using endo F. This glycosidase can cleave between the chitobiose structure leaving a terminal GIcNAc which should have little effect on protein charge. The enzyme has a preference for high mannose and biantennary complex chains, but is inactive with tri- and tetra-antennary chains and does not work well with hybrid chains. Fig. 3 shows the effect of incubation time in presence of endo F on sialyltransferase activity. Incubation for 24 h at p H 4.5 was required for maximum reduction of enzyme activity; longer incubations did not lead to significantly reduced enzyme activities relative to the corresponding controls. Sialyltransferase retained about 90% of its activity after 24 h incubation under control conditions in absence of endo F (Fig. 3). Experiments were also conducted with endo H which cleaves high mannose chains between the chitobiose structure. Incubation of sialyltransferase in presence of this glycosidase under the conditions described above had little effect on sialyltransferase activity (not shown). Sialyltransferase was also subject to digestion with exoglycosidases. In these experiments the enzyme was treated with neuraminidase followed by galactosidase

328 lOOq

TABLE I

Effect of treatment with exoglycosidases on sialyltransferase activity

80 o~

~6o "r,

~ 4o 20 5

10 15 Time, h

20

25

Fig. 3. Effect on sialyltransferase activity following incubation with endo F for different times. Sialyltransferase was incubated alone or in presence of endo F for up to 24 h in presence of 1.5% Triton X-100. Incubations were stopped and sialyltransferase was assayed as described in Fig. 1. The results are from a typical experiment. The 100% value represents 8560 U of sialyltransferase activity per sample; sialyltransferase activity in presence (o) or absence ( • ) of endo F. Each point represents the mean for three assays and reproducibility was within + 12%.

and then assayed for catalytic activity. Assay was not carried out after neuraminidase treatment, since sialyltransferase is known to be self sialylating. In a separate set of experiments sialyltransferase was treated with N-acetylhexosaminidase after neuraminidase and galactosidase treatment. The results of these experiments are shown in Table I. Removal of NeuAc followed by Gal did not significantly affect catalytic activity compared to controls in which sialyltransferase was incu-

Controls consisted of sialyltransferase incubated at pH 5.5 for 36 h alone as for the endo F treatment for those experiments where the enzyme was incubated with neuraminidase followed by galactosidase. During this incubation the control sialyltransferase typically lost 20% of its activity. For treatment with N-acetylhexosaminidase the control was incubated for a further 18 h and lost another 15% of its activity. The results are expressed as % sialyltransferase activity remaining after incubation with glycosidases relative to the appropriate controls. Results are means from five separate experiments +S.D. Exoglycosidase

Catalytic activity remaining (% of appropriate control)

Neuraminidase + galactosidase N-Acetylhexosaminidase

89 + 5 39 _+8

bated in absence of glycosidases, but subsequent removal of GlcNAc resulted in a loss of about 60% of the catalytic activity compared to the appropriate control.

Immunoblotting of sialyltransferase The above results suggest that removal of complex chains from sialyltransferase correlates with loss in enzyme activity. Endoglycosidases act on native glycoproteins only with difficulty, but denatured glycoproteins are more sensitive to hydrolysis. In order to check that glycosidase action on the native enzyme resulted in removal of carbohydrate chains, glycosidase digestions of denatured sialyltransferase were com-

.qP

,0

4C

Fig. 4. Immunoblot analysis of sialyltransferase following deglycosylation with Glycanase and endo F. Sialyltransferase was treated with Glycanase for 18 h or with endo F for 6 h and 24 h and compared with untreated sialyltransferase by immunoblot analysis. Track 1, native sialyltransferase treated with Glycanase; track 2, denatured sialyltransferase treated with Glycanase, track 3, native sialyltransferase treated with endo F for 24 h, track 4, native sialyltransferase treated with endo F for 6 h, track 5, native sialyltransferase alone. Completely deglycosylated sialyltransferase had a molecular mass (tracks 1 and 2) of about 37 kDa and native sialyltransferase was about 42 kDa. Molecular mass values were determined as before [5] using the following markers: carbonic anhydrase (C, 31 kDa), ovalbumin (O, 42.699 kDa), bovine serum albumin (A, 66.2 kDa) and phosphorylase b (P, 97.4 kDa).

329 pared with digestions of the native enzyme by immunoblot analysis to detect molecular mass changes. Fig. 4 shows that untreated sialyltransferase was 42 kDa corresponding to the glycosylated form of the enzyme, whereas the native and denatured enzyme following treatment with Glycanase were about 37 kDa; the Glycanase-treated native and denatured enzymes did not give a positive reaction using the Glycan detection kit, whereas the native enzyme stained for carbohydrate. At shorter incubation times a third band with a mobility intermediate between the native and deglycosylated enzyme was detected; these results were similar to those shown in Fig. 4 track 3 where a partially deglycosylated form of the enzyme was found following treatment with endo F for 24 h. Treatment with endo F for 24 h showed three bands corresponding to the native and deglycosylated enzyme, as well as a band of intermediate molecular mass corresponding to a partially deglycosylated form of the enzyme, but complete deglycosylation of sialyltransferase with endo F as determined by immunoblot analysis was not achieved (Fig. 4). The molecular mass determination of samples from control experiments in which native sialyltransferase was incubated in absence of glycosidases gave values similar to the untreated enzyme. Discussion

Endoglycosidases have been used to remove oligosaccharide chains from native glycoproteins in order to determine the role played by the chains in biological activity, but the results have been variable. For example, erythropoitin was found to retain about 60% of its activity following removal of oligosaccharide chains with endoglycosidases [12], however, human chorionic gonadotropin and luteotropin lost adenylate cyclase activity and steroidogenic activities [13,14] on deglycosylation. In the present work complete deglycosylation of native sialyltransferase was observed with Glycanase and loss of oligosaccharide chains correlated with loss in enzyme activity. There was a reduction in molecular mass of about 5 kDa in samples subject to extensive digestion. This would correspond to the removal of about two complex carbohydrate chains from the enzyme. Complete deglycosylation with endo F was not achieved, but removal of oligosaccharide chains as a function of time of hydrolysis correlated with loss of activity, supporting the idea that the oligosaccharide chains of sialyltransferase are important in controlling enzyme activity. Furthermore, since endo F cleaves between the chitobiose structure the appearance of additional charge due to aspartic acid that occurs following Glycanase treatment would not appear to be an important factor affecting loss of sialyltransferase activity after deglycosylation with this endoglycosidase. Experiments using exoglycosidases suggested that the

minumum oligosaccharide structure for full expression of catalytic activity contains the trimannose core with GIcNAc residues attached although the enzyme displays reduced catalytic activity with only the trimannose-containing core present. Sequential removal of the mannose residues was difficult because of the extended incubation times that were required which affected the stability of the enzyme. It was found that the presence of methanol or ethanol caused deglycosylation of sialyltransferase to be more effective with Glycanase. The use of methanol or ethanol to enhance degycosylation of a native glycoprotein by endoglycosidases has not been reported. This is an important observation, since the use of methanol or ethanol to enhance deglycosylation of native glycoproteins with Glycanase could be taken advantage of in studies on the importance of oligosaccharide chains for biological activity in other glycoproteins. Of particular interest would be a study on deglycosylation of other sialyltransferases, such as the a2,3-sialyltransferase, involved in N-linked glycoprotein biosynthesis [4], and SAT-l, the sialyltransferase involved in ganglioside biosynthesis [14]. SAT-1 is sensitive to Glycanase, but the effect of deglycosylation on enzyme activity was not reported [15]. The role played by methanol or ethanol in deglycosylation of sialyltransferase is not clear, but it is possible that the presence of organic solvents causes partial denaturation of sialyltransferase making the enzyme more accessible to Glycanase without significantly affecting enzyme activity. Alternatively, the presence of organic solvent could alter the hydrodynamic properties of the proteins resulting in greater Glycanase reactivity. The presence of methanol has been shown to influence the folding of staphylococcal nuclease [16] and ribonuclease B [17]. Clearly, further studies are needed to determine the role of methanol and ethanol in deglycosylation of native glycoproteins by Glycanase. It is unlikely that the loss in molecular mass found following endoglycosidase treatment is due to proteinase action, since the endoglycosidases are reported to be free of proteinases and albumin was added to the system to protect the sialyltransferase against proteinase degradation. Also, in our previous work sialyltransferase was found to be resistant to proteinases, except for the action of cathepsin D which was not detectable in the hydrolysis mixtures and in the case of Glycanase would be unreactive at the pH used [5]. The results reported here indicate that at least one of the oligosaccharide chains of sialyltransferase is essential for the catalytic activity of this enzyme. Sialyltransferase attaches NeuAc to terminal positions of oligosaccharide chains of glycoproteins. It is located in an active form in the t r a n s Golgi at the end of the glycosylation pathway before glycoproteins exit the cell [1-5]. The enzyme is attached to the Golgi via a

330 9-amino-acid cytoplasmic unit which is attached to a 17-amino-acid transmembrane anchor domain and this in turn is linked to the catalytic unit by a stem or linker region consisting of about 35 amino acids [8,18]. This allows the catalytic unit which consists of the bulk of the enzyme to be exposed to glycoprotein substrates that are in transit through the lumen of the Golgi complex. In our earlier work we showed that the catalytic unit could be cleaved from the membrane bound form of the enzyme by the action of a cathepsin-D-like activity which explained how the enzyme could exist in soluble form [5]. The enzyme used in our studies consisted of the soluble purified catalytic unit which had been cleaved from the membrane anchor and stem region, but this unit is known to contain all the carbohydrate and so the results presented in this work are also valid for the membrane bound form of the enzyme as it exists in the Golgi. Although the precise structures of the oligosaccharide chains of sialyltransferase are not known, previous sudies by us and by others suggest that the chains are of the complex type [1]. This raises the interesting question about how sialyltransferase becomes sialylated as it passes through the Golgi complex and where in the Golgi the enzyme becomes catalytically active. The present work suggests that the oligosaccharide chains are important although it is not clear how complete the chains have to be before the enzyme starts to express activity. It would appear that the minimum structure for the enzyme to express full catalytic activity would have to contain the trimannose core with GIcNAc attached en route to complex chain formation by oligosaccharide processing. Also, it is not clear if both chains are needed for catalytic activity; further studies are needed to address this question.

Acknowledgements This work was supported by the Natural Sciences and Engineering Research Council of Canada (grant no. OGPOOO5394). D.F. and G.M. were supported by University of Manitoba Fellowships.

References 1 Jamieson, J.C. (1989) Life Sci. 43, 691-697. 2 Kaplan, H., Woloski, B.M.R.N.J., Hellman, M. and Jamieson, J.C. (1983) J. Biol. Chem. 258, 11505-11509. 3 Roth, J., Taatjes, D.J., Lococq, J.M., Weinstein, J. and Paulson, J.C. (1985) Cell 43, 287-295. 4 Weinstein, J., Souza-e Silva, U. and Paulson, J.C. (1982) J. Biol. Chem. 257, 13834-13844. 5 Lammers, G. and Jamieson, J.C. (1988) Biochem. J. 256, 623-631. 6 Kornfeld, S. (1987) FASEB J. 1,462-468. 7 Weinstein, J., Lee, E.U., McEntee, K., Lai, P-H. and Paulson, J.C. (1987), J. Biol. Chem. 262, 17735-17743. 8 Paulson J.C. and Colley, K.J. (1989) J. Biol. Chem. 264, 1761517618. 9 Paulson, J.C. (1989) Trends Biochem. Sci. 14, 272-276. 10 O'Hanlon, T.P., Lau, K.M., Wang, X. and Lau, J.T.Y. (1989) J. Biol. Chem. 264, 17389-17394. 11 Jamieson, J.C., Kaplan, H.A., Woloski, B.M.R.N.J., Hellman, M.A. and Ham, K. (1983) Can. J. Biochem. Cell Biol. 61, 10411048. 12 Dordal, S.D., Wang, F.F. and Goldwasser, E. (1985) Endocrinology 116, 2293-2299. 13 Channing, C.P., Sakai, C.N. and Bahl, O.P. (1978) Endocrinology 103, 341-345. 14 Kalyan, N.K. and Bahl, O.P. (1983) J. Biol. Chem. 258, 67-74. 15 Melkerson-Watson, L.J. and Sweeley, C.C. (1991) J. Biol. Chem. 266, 4448-4457. 16 Nakano, T. and Fink, A.L. (1990) J. Biol. Chem. 265, 1235612362. 17 Fink, A.L. and Painter, B. (1987) Biochemistry 26, 1665-1671. 18 Colley, K.J., Lee, E.U. and Paulson, J.C. (1992) J. Biol. Chem. 267, 7784-7793.