BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS ARTICLE NO.

236, 473–478 (1997)

RC976871

Human Islets of Langerhans Express Multiple Isoforms of Calcium/Calmodulin-Dependent Protein Kinase II Maria A. Breen and Stephen J. H. Ashcroft1 Nuffield Department of Clinical Biochemistry, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom

Received May 23, 1997

Previous studies have provided evidence for the presence of calcium/calmodulin-dependent protein kinase II (CaM kinase II) in rodent islets of Langerhans, and b-cell CaM kinase II activity has been correlated with insulin secretion. In this study we provide the first conclusive evidence for the expression of CaM kinase II in human islets of Langerhans and show that multiple isoforms are expressed. Screening of a human islet cDNA library resulted in the isolation of a 999bp partial cDNA clone encoding CaM kinase II. The nucleotide sequence of the islet clone showed a high degree of homology (94.8%) to the two g isoforms of CaM kinase II previously isolated from human T lymphocytes (gB and gC). In order to obtain full length sequence for the islet clone, rapid amplification of cDNA ends (RACE) was used to amplify the 3 * end of the islet clone from human islet poly A/ RNA. Two distinct g isoforms of CaM kinase II were amplified from the islet RNA. They were identified as gB and gE ; the latter is distinguished from gB by a 114bp insertion within the association domain of the cDNA. Using reverse transcriptase polymerase chain reaction (RT-PCR) we also detected in human islets of Langerhans the novel b3 isoform of CaM kinase II previously reported to be expressed in neonatal rat islets. q 1997 Academic Press

It is well established that a rise in intracellular free Ca2/ plays a key role in glucose-stimulated insulin secretion (1-3) but the cellular events linking increased Ca2/ concentration to the release of insulin are unknown. There is an increasing body of evidence to support a role for calcium/calmodulin-dependent protein kinase II (CaM kinase II) in this process. Pancreatic b-cells contain a high concentration of calmodulin (CaM) which mediates many Ca2/ regulated functions (4), and the CaM inhibitor trifluoperazine inhibits glucose-stimulated insulin release (5,6). Islets 1 To whom correspondence should be addressed. Fax: 44 1865 221834. E-mail: [email protected].

contain a calcium-dependent protein kinase that requires calmodulin for its activity (7-9). The islet CaM kinase resembles brain CaM kinase II in several ways including sensitivity to Ca2/ and calmodulin, sensitivity to inhibitors, and substrate specificity, but may not be identical (10,11). Evidence for the involvement of islet CaM kinase II in insulin secretion has been provided by inhibitor studies in which the inhibitory effect on insulin release of the diabetogenic agents alloxan (11,12), and dehydrouramil (13) was correlated with the inhibition of b-cell CaM kinase II. In the b-cell lines HIT-15 and RINm5F, insulin release has also been shown to be inhibited by the CaM kinase II inhibitor KN-62 (1-[N,O-bis(5-isoquinolinesulfonyl)-N-methylL-tyrosyl]-4-phenyl-piperazine) (14,15). These findings are however complicated by additional effects of alloxan and KN-62 on glucokinase (16) and Ca2/-channel activity respectively (17). More convincing evidence has been provided by the patch clamp technique; a specific peptide inhibitor of CaM kinase II (residues 290-309) markedly reduced insulin secretion without affecting the Ca2/ current from a single cell determined by capacitance measurements (18). The Ca2//calmodulin-dependent protein kinase of rat islets is dependent on concentrations of calcium and calmodulin present in islets which occur during stimulation of insulin release (Km for Ca2/ Å0.4mM; Km for calmodulin Å40nM) (11). Glucose and other agents which stimulate insulin secretion have been shown to activate CaM kinase II as measured by autophosphorylation of the kinase in isolated rat islets and RINm5F bcells (19,20). The extent of activation correlated closely with insulin secretion in concentration dependence and time course and was completely prevented in the rat islets by the inhibitor of glucose metabolism mannoheptulose (20). These studies provide evidence that CaM kinase II has a key role to play in glucose-stimulated insulin release. CaM kinase II is a ubiquitous serine/threonine protein kinase which has been implicated in diverse effects of hormones and neurotransmitters that utilise Ca2/ as a second messenger, ranging from muscle contraction,

473

0006-291X/97 $25.00 Copyright q 1997 by Academic Press All rights of reproduction in any form reserved.

AID

BBRC 6871

/

6932$$$981

06-26-97 11:44:34

bbrcg

AP: BBRC

Vol. 236, No. 2, 1997

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

secretion, synaptic transmission to gene expression (21,22). CaM kinase II comprises a multigene family in which each of the isoforms (a, b, g, and d) is encoded by a separate gene and further variability is provided by alternative splicing of primary transcripts. Cells can contain more than a single isoform of the kinase. Western analysis and cDNA cloning have provided evidence for the expression of the a (rat islets), b3 (rat islets), g (RINm5F), d2 (RINm5F and MIN6) and d6 (RINm5F) isoforms of CaM kinase II in rodent pancreatic islets and b-cell lines. Western blotting of rat islets indicated the presence of CaM kinase II a isoform (14,20,23), whereas the d2 and d6 isoforms have been reported in insulinoma-derived b-cell lines (24). We previously cloned a novel splice variant of the rat bisoform from a neonatal rat islet cDNA library (25). However, although rodent studies provide good working models for human systems, the genes or forms of genes expressed may vary between human islets and rodent islets and/or b-cell lines. For example, there is recent evidence that the glucose transporter GLUT-2, which was found to be expressed in rodent b-cells, is not expressed in human b-cells (26). Assessment of the importance of CaM kinase II for insulin secretion in man therefore requires determination of the isoform(s) of CaM kinase II expressed in human islets of Langerhans. MATERIALS AND METHODS Materials. [a-32P]dCTP and [a-35S]dATP were purchased from Amersham Life Sciences, Little Chalfont, UK. Sequencing was carried out using the dideoxy chain termination method with Sequenase version 2.0 kit (Amersham Life Sciences) A lgt11 cDNA library prepared from normal human islets of Langerhans was a gift from Dr. Hiroshi Sakura, Laboratory of Physiology, University of Oxford. A cDNA encoding rat CaM kinase II b3 isoform was from Dr. Virginia Urquidi, Nuffield Department of Clinical Biochemistry, University of Oxford, UK. Human islets of Langerhans were supplied by Dr. Derek Gray, Transplantation Centre, Churchill Hospital, Oxford., UK. Taq DNA polymerase was from Perkin Elmer, Warrington, UK. Other reagents and enzymes were from Promega, Southampton, UK or Sigma Chemical Co., Poole, Dorset, UK. Screening of a human islet cDNA library. A lgt11 cDNA library prepared from normal human islets of Langerhans was screened by plaque hybridisation using standard procedures (27). A cDNA encoding CaM kinase II b3 isoform cloned from a neonatal rat islet library was random primed with [a-32P]dCTP and used as a probe. Hybridisation was carried out in 50% formamide buffer overnight at 427C. Stringency washes were carried out as follows; 3 1 [ 0.51SSC, 0.1% SDS] 30 minutes at room temperature, 21 [ 0.11SSC 0.5% SDS] 15 minutes at 657C. From 11107 colonies screened 10 positives were identified, subcloned into Bluescript KS(/) and sequenced. Rapid amplification of cDNA ends (RACE). Amplification of the 3* end of the human islet cDNA clone from human poly A/ RNA was carried out using the 3 * AmpliFINDER RACE kit (Clontech Laboratories, Palo Alto, USA). Single stranded cDNA was prepared from 25ng of poly A/ RNA. A modified oligo (dT) primer containing base degeneracies at the 3* end was used to initiate the cDNA synthesis at the junction between the 3 * untranslated region of the gene and the beginning of the poly A/ tail. One twentieth of the cDNA

synthesis reaction was used as a target for a first round of PCR using a sense primer (P1) and an anchor primer. The anchor primer anneals to a sequence incorporated at the end of the molecule during the cDNA synthesis. The nucleotide sequence of P1 (5*-AAG CGC ATC ACG GCT GAC CAG GCT C-3*) corresponded to nucleotides 817-836 of the clone isolated from the human islet library. Thirty two cycles of amplification were carried out using 1 unit of Taq polymerase per reaction. Each cycle consisted of 60s denaturing at 957C, 120s annealing at 607C and 180s extension at 727C. To ensure maximum specificity a second round of PCR was performed with a nested primer (P2) and the anchor primer, using 1 ml of the primary PCR reaction as the target. P2 (5*-TCC ACG GTG GCA TCC ATG AT-3 *) corresponded to nucleotides 878-897 of the clone isolated from the human islet library. To eliminate the possibility of artefactual results, four independent RACE reactions were performed and the products cloned and sequenced. Reverse transcriptase polymerase chain reaction (RT-PCR). Poly A/ RNA from human islets was reverse transcribed using oligo (dT) and then amplified by PCR. A primer pair specific for the b3 isoform of rat islet CaM kinase II was used; the sense primer (5*-CTA GGG ATC CAT GGA GGA CGA AGA TGC CAA G-3*) corresponded to nucleotides 1131-1154 and the antisense primer (5*-AGT CGG ATC CTC TTG ATG ATT TCC TGC TTC CG-3*) to nucleotides 1413-1434.

RESULTS Cloning of CaM kinase II g from a human islet cDNA library. Screening of a human islet cDNA library resulted in the isolation of a number of positive clones. Sequencing of the 3* and 5* ends of these clones using lgt11 forward and reverse primers to prime the sequencing reactions, revealed that they all encoded CaM kinase II and were identical. The cDNA insert of one clone (LHI06) was sequenced in both directions using specific oligonucleotides to walk along the clone and was 999bp in length. Nucleotide homology searches were carried out using the GCG programs BLAST and FASTA to determine if the nucleotide sequence of the islet clone matched any known sequences. These searches revealed a high degree of homology (99.5%) to the gB and gC isoforms of CaM kinase II, previously isolated from human T lymphocytes (28). The AUG at position 54 was determined as the start of translational initiation as it was the first AUG and it was in a strong context as defined by the Kozak consensus sequence (AGCATGG) (29). The cDNA LHI06 encoded the first 315 amino acids of the g isoform of CaM kinase II, incorporating the entire regulatory and catalytic domains, plus 53 nucleotides of 5* untranslated sequence. Comparison of the nucleotide sequences of the 5* untranslated regions of the gB isoform of CaM kinase II and LHI06 clones revealed no significant sequence homology except for the Kozak consensus sequence close to the first AUG. Within the region of the cDNA that codes for amino acids there were four sequence variations. Each was a single base pair change as follows at position 121; GCC (Ala) replaced by GCT (Ala), at position 197; AAG (Lys) replaced by AAA (Lys), at position 333; ACG (Thr) replaced by ACC (Thr) and at position 336; GGA (Gly)

474

AID

BBRC 6871

/

6932$$$982

06-26-97 11:44:34

bbrcg

AP: BBRC

Vol. 236, No. 2, 1997

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

FIG. 1. Cloning of human islet CaM kinase II. Schematic representation of human gB isoform of CaM kinase II and human islet clones isolated and their relation to one another. The predicted amino acids coded for by the insertion within the association domain of RPPA1 are shown. The sequence data were obtained from four independent clones.

replaced by GGC (Gly). In all of the above cases no change in the amino acid coded for resulted from the base pair alterations. Isolation of the 3* end of human islet g CaM kinase II cDNA. As none of the clones contained a full length cDNA, RACE was used to isolate the 3* end of the human islet cDNA from poly A/ RNA. A large number of PCR products were subcloned and sequenced and revealed the expression of two g isoforms of CaM kinase II (Fig. 1). One of these, RPPA2, was 950bp in length and showed 100% homology at the amino acid level to the gB isoform of CaM kinase II previously cloned from human T Lymphocytes (28). The other, RPPA1 (1064bp), was distinct from RPPA2 in having a 114bp insertion in the region of the gene that is thought to code for the association domain of the protein. The nucleotide sequence of RPPA1 showed 99.9% homology to a partial cDNA (315bp) present in the EMBL/GenBank/DDBJ sequence databases (accession number, U32473). This short cDNA is an RT-PCR product isolated from a human gallbladder tumour cell line and has been shown to be part of a CaM kinase II designated gE (30). All the identified human isoforms of CaM kinase II differ only in insertions or deletions within the variable and/or association domain of the protein suggesting they may have originated from alternative splicing of a single gene. The insertion in gE maintains the reading frame and codes for 38 amino acids suggesting the full length gE cDNA encodes a protein of 556 amino acids with a calculated molecular mass of 76kDa. The full length sequence of gE was predicted from the overlapping clones LHI06 and RPPA1 (Fig. 1) and is shown in Figure 2. The sequences of RPPA1 and LHI06 were submitted to GenBank and assigned the accession numbers U66063 and U66064 respectively. Expression of CaM kinase II b3 in human islets of Langerhans. Control reactions designed to amplify bactin revealed that the RNA was intact and the RTPCR reactions were successful. Using primers based

FIG. 2. Nucleotide and predicted amino acid sequence of human gE isoform of CaM kinase II. The nucleotide sequence was obtained by combining the nucleotide sequence of the overlapping clones LHI06 and RPPA1. Nucleotides 1153-1267 (underlined) are present in the gE isoform but not in the gB isoform of CaM kinase II. The sequence data were obtained from 4 separate cloning experiments.

475

AID

BBRC 6871

/

6932$$$982

06-26-97 11:44:34

bbrcg

AP: BBRC

Vol. 236, No. 2, 1997

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

FIG. 3. RT-PCR of b3 isoform of CaM kinase II from human islets of Langerhans. Agarose gel electrophoresis of PCR products amplified from human islet poly A/ RNA. Lanes 1,2 human islet RNA; lanes 3,4 no target. The size of the appropriate base pair marker is indicated on the right. The data are representative of 4 similar experiments.

on the unique sequence of CaM kinase II b3, a 303bp product was amplified from human islet poly A/ RNA (Fig. 3). The nucleotide sequence of the amplified product was identical to the nucleotide sequence of the rat islet b3 isoform of CaM kinase II. DISCUSSION Non-insulin-dependent diabetes mellitus (NIDDM) is a complex metabolic disorder characterised by both inadequate insulin production in response to glucose and insulin resistance at the target organs (31). Clustering of the disease within families and the high degree of concordance among monozygotic twins implicates genetic factors in the pathology (32). The candidate gene approach has been employed in the search for NIDDM predisposition genes. Candidate genes are identified which are involved in pathways impaired in subjects with NIDDM, such as the insulin secretion pathway and association between these genes and the disease can then be investigated (33). There is an increasing amount of evidence from rodent studies that CaM kinase II has an important role to play in glucosestimulated insulin secretion (10-20) and may therefore be considered as a candidate gene for NIDDM.

CaM kinase II is a multi-functional protein kinase with broad substrate specificity and distribution (21,22). The holoenzyme is a multimer composed of 612 subunits and it is likely that both homomultimers and heteromultimers of the kinase exist (34). As previously found for multimeric ion channels, the subunit composition of the kinase may modify the properties of the holoenzyme. Alternatively spliced variants of each subunit, differing in insertions or deletions within the variable and association domains of the enzyme, are expressed in different tissues, suggesting tissue-specific functions for each isoform or combination of isoforms. Electron microscopic studies suggest that in the holoenzyme the catalytic/regulatory domains (NH2) of the subunits form spherical units which are tethered by the variable domains to a central globular core made up of the association domains (COOH) (35). Therefore inserts in the NH2 terminal half may be close to the catalytic/regulatory domain and have an effect on the Ca2//calmodulin binding or substrate specificity of the enzyme, while insertions or deletions in the COOH terminal domain would be part of the central globular core and could serve to regulate the multimer size of the kinase and its targeting to intracellular sites. In fact an insertion within the variable domain of dB CaM kinase II has been implicated in its nuclear targeting (25). Therefore in order to fully understand cellular targeting and substrate specificity of CaM kinase II in a particular cell type, it is necessary to define which isoforms are expressed there and in what ratios, and which isoforms are capable of coassembly. In this investigation we report the expression of three isoforms of CaM kinase II in human islets: two g isoforms (gB and gE) and one b isoform (b3). In contrast to rodent b-cells (14, 20, 23, 24) we could find no evidence for expression of either a or d isoforms (data not shown). However in the absence of sequence data for these isoforms of CaM kinase II in man the primers used were based on rat sequences; our negative results may simply reflect lack of sequence conservation between rodent and human CaM kinase II in the regions represented by the PCR primers used. The b3 isoform was previously cloned from a neonatal rat islet cDNA library (25). It contains a proline-rich insertion in the association domain which has been suggested to be involved in interaction with SH3-domain containing proteins. The insertion occurs at an intron/exon boundary. The 114 bp insert in the gE isoform occurs in the same relative position and may therefore occur at the corresponding intron/exon boundary of the g gene. The gE isoform was detected by RT-PCR on a cell line derived from a human gallbladder tumour and evidence for its presence in normal human gallbladder obtained by Southern blotting [34]. However its identity in the normal tissue was not confirmed by sequencing and the full length sequence of human gE has not previously been reported. In this

476

AID

BBRC 6871

/

6932$$$982

06-26-97 11:44:34

bbrcg

AP: BBRC

Vol. 236, No. 2, 1997

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

isoform the 39-amino acid insert shows homology to a motif found in integral inner-membrane proteins from bacterial transport systems (30). Again therefore the insert might be important in targeting CaM kinase II to specific subcellular structures in the b-cell. How CaM kinase II links a rise in intracellular calcium to insulin secretion is unknown, but a number of sites of action have been proposed. It has been suggested that CaM kinase II regulates insulin secretion distal to the elevation of intracellular Ca2/ (37) and that it may play a role in the interaction between the secretory granule membrane and the plasma membrane that takes place during the release of insulin by exocytosis (38). It has been proposed that CaM kinase II phosphorylates synapsin I-like protein, and is involved in the initiation of the secretary response, by releasing the secretory granules from the cytoskeleton [39]. It has also been suggested that CaM kinase II may interact with the ryanodine receptor to release calcium from the endoplasmic reticulum intracellular stores and thereby prolong the effect of the first increase in intracellular calcium (23). All or none of these ideas may prove to be correct as it is apparent from our data that several isoforms of the kinase are expressed in human islets and it is likely that different isoforms or different combinations of isoforms may mediate different functions. Since the present studies were conducted on islets of Langerhans it is now clearly important to establish which islet cell types express the individual isoforms. Determination of the ratio of these isoforms to each other, whether they co-assemble to form homomultimers or heteromers, and their subcellular localisation is required to fully understand the role of CaM kinase II in glucose-stimulated insulin secretion. Identification of the CaM kinase II genes expressed in human islets of Langerhans now provides new candidate genes in the search for the genetic elements that lead to a predisposition to NIDDM. If, as has been proposed, the defect in the insulin secretion pathway lies proximal to the point regulated by CaM kinase II, a better understanding of the site of action and the form of the gene(s) expressed may lead to development of drugs for effective control of disease or important in diagnosis.

ACKNOWLEDGMENTS We are grateful to Dr. Hiroshi Sakura, University Laboratory of Physiology, Oxford, for providing the human islet cDNA library; Dr. Derek Grey, Islet Transplant Laboratory, Churchill Hospital, Oxford, for providing the human islets of Langerhans; and Dr. Virginia Urquidi, Nuffield Department of Clinical Biochemistry, University of Oxford for the CaM kinase II b3 isoform plasmid. These studies were supported by grants from the Medical Research Council, the British Diabetic Association and the Wellcome Foundation.

REFERENCES 1. Wollheim, C. B., and Sharp, G. W. G. (1981) Physiol. Rev. 61, 914–973. 2. Malaisse, W. J., and Malaisse-Lagae, F. (1984) Experientia 40, 1068–1075. 3. Prentki, M., and Matschinsky, F. M. (1987) Physiol. Rev. 67, 1185–1248. 4. Sugden, M. C., Christie, M. R., and Ashcroft, S. J. H. (1979) FEBS Letts 105, 95–100. 5. Schubart, U. K., Fleischer, N., and Erlichman, J. (1980) J. Biol. Chem. 255, 11063–11066. 6. Gagliardino, J. J., Harrison, D. E., Christie, M. R., Gagliardino, E., and Ashcroft, S. J. H. (1980) Biochem. J. 192, 919–927. 7. Harrison, D. E., and Ashcroft, S. J. H. (1982) Biochim. Biophys. Acta 714, 313–319. 8. Landt, M., McDaniel, M. L., Bry, G. C., Kotagal, N., Colca, J. R., Lacy, P. E., and McDonald, J. M. (1982) Arch. Biochem. Biophys. 213, 148–154. 9. Colca, J. R., Brooks, C. L., Landt, M., and McDaniel, M. L. (1983) Biochem. J. 212, 819–827. 10. Mayer, P., Pfeiffer, A., and Schatz, H. (1992) Horm. Metab. Res. 24, 95. 11. Hughes, S. J., Smith, H., and Ashcroft, S. J. H. (1993) Biochem. J. 289, 795–800. 12. Colca, J. R., Kotogal, N., Brooks, C. L., Lacy, P. E., Landt, M., and McDaniel, M. L. (1983) J. Biol. Chem. 258, 7260–7263. 13. Harrison, D. E., Poje, M., Rocic, B., and Ashcroft, S. J. H. (1986) Biochem. J. 237, 191–196. 14. Niki, I., Okazaki, K., Saitoh, M., Niki, A., Tamagawa, T., Iguchi, A., and Hidaka, H. (1993) Biochem. Biophys. Res. Commun. 191, 255–261. 15. Wenham, R. M., Landt, M., Walters, S. M., Hidaka, H., and Easom, R. A. (1992) Biochem. Biophys. Res. Commun. 189, 128– 133. 16. Lenzen, S., Freytag, S., and Panten, U. (1988) Mol. Pharmacol. 34, 395–400. 17. Li, G. D., Hidaka, H., and Wollheim, C. B. (1992) Mol. Pharmacol. 42, 489–498. ¨ mma¨la¨, C., Eliasson, L., Bokvist, K., Larsson, O., Ashcroft, 18. A F. M., and Rorsman, P. (1993) J. Physiol. (Lond) 472, 665–688. 19. Norling, L. L., Colca, J. R., Kelly, P. T., McDaniel, M. L., and Landt, M. (1994) Cell Calcium 16, 137–150. 20. Wenham, R. M., Landt, M., and Easom, R. A. (1994) J. Biol. Chem. 269, 4947–4952. 21. Colbran, R. J., Schworer, C. M., Hashimoto, Y., Fong, Y. L., Rich, D. P., Smith, M. K., and Soderling, T. R. (1989) Biochem. J. 258, 313–325. 22. Braun, A. P., and Schulman, H. (1995) Annu. Rev. Physiol. 57, 417–445. 23. Takasawa, S., Ishida, A., Nata, K., Nakagawa, K., Noguchi, N., Tohgo, A., Kato, A., Yonekura, H., Fujisawa, H., and Okamoto, H. (1995) J. Biol. Chem. 270, 30257–30259. 24. Mayer, P., Mo¨hlig, M., Schatz, H., and Pfeiffer, A. (19) Biochem. J. 298, 757–758. 25. Urquidi, V., and Ashcroft, S. J. H. (1995) FEBS Lett. 358, 23– 26. 26. De Vos, A., Heimberg, H., Quartier, E., Huypens, P., Bouwens, L., Pipeleers, D., and Schuit, F. (1995) J. Clin. Invest. 96, 2489– 2495. 27. Current Protocols in Molecular Biology, Volume 1, Wiley, 6.0.36.4.7

477

AID

BBRC 6871

/

6932$$$982

06-26-97 11:44:34

bbrcg

AP: BBRC

Vol. 236, No. 2, 1997

BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

28. Nghiem, P., Saati, S. M., Martens, C. L., Gardener, P., and Schulman, H. (1993) J. Biol. Chem. 268, 5471–5479. 29. Kozak, M. (1991) J. Cell Biol. 115, 887–903. 30. Kwiatkowski, A. P., and McGill, J. M. (1995) Gastroenterology 109, 1316–1323. 31. Taylor, S. I., Accili, D., and Imai, Y. (1994) Diabetes 43, 735– 740. 32. Cerasi, E. (1992) In Insulin: Molecular Biology to Pathology (Ashcroft, F. M., Ashcroft, S. J. H., Eds.), pp. 347–380, Oxford University Press, Oxford. 33. Turner, R. C., Hattersley, A. T., Shaw, J. T., and Levy, J. C. (1995) Diabetes 44, 1–9.

34. Tobimatsu, T., and Fujisawa, H. (1989) J. Biol. Chem. 264, 17907–17912. 35. Kanaseki, T., Ikeuchi, Y., Sugiura, H., and Yamauchi, T. (1991) J. Cell. Biol. 115, 1049–1060. 36. Srinivasan, M., Edman, C. F., and Schulman, H. (1994) J. Cell. Biol. 126, 839–852. 37. Ashcroft, S. J. H. (1994) Diabetologia 37 [suppl 2., S21–S29. 38. Watkins, D. T. (1991) Diabetes 40, 1063–1068. 39. Matsumoto, K., Fukunaga, K., Miyazaki, J., Shichiri, M., and Miyamoto, E. (1995) Endocrinology 136, 3784–3793.

478

AID

BBRC 6871

/

6932$$$983

06-26-97 11:44:34

bbrcg

AP: BBRC