Mitochondria! Glycerol-3-Phosphate DehydrogenaseCloning of an Alternatively Spliced Human Islet-CellcDNA, Tissue Distribution, Physical Mapping, andIdentification of a Polymorphic Genetic MarkerJorge Ferrer, Minoru Aoki, Philip Behn, Ann Nestorowicz, Andrew Riggs, and M. Alan Permutt

Pancreatic P-cell mitochondrial glycerol-3-phosphate de-hydrogenase (mGPDH) plays a major role in glucose-induced insulin secretion. Decreased activity of thisenzyme has thus been proposed to play a role in thepathogenesis of NIDDM. Cloning of human insulinomamGPDH cDNAs disclosed the existence of two varianttranscripts with different 5' ends. Reverse transcriptionpolymerase chain reaction (PCR) confirmed the presenceof both mGPDH mRNAs in purified native human pancre-atic islets and other tissues. A major 6.5-Kb mGPDHtranscript was detected by Northern blot analysis in RNAfrom human and rat pancreatic islets, with distinctlylower levels in other human tissues, indicating that pre-viously reported high mGPDH enzymatic activity inP-cells is determined by high transcript levels. ThemGPDH gene was mapped to chromosome 2 by PCRanalysis of genomic DNA from human/rodent somaticcell hybrids, and five independent overlapping yeastartificial chromosome (YAC) clones containing the mG-PDH sequence were identified from the Centre d'Etudedu Polymorphisme Humain YAC library. Analysis ofthese YAC clones identified a highly polymorphic chro-mosome 2q21-q33 dinucleotide repeat genetic marker(D2S141) physically linked to the mGPDH gene.These studies provide the means to investigate therole of the human mGPDH gene in the pathogenesis ofNIDDM and illustrate the value of a novel strategy toidentify genetic markers for diabetes candidategenes. Diabetes 45:262-266, 1996

In pancreatic P-cells, the recognition of glucose as aninsulin secretagogue is dependent on signals gener-ated in aerobic glycolysis (1,2). Elevations of extracel-lular glucose are associated with an increase in the

ATP-to-ADP ratio, closure of ATP-sensitive K+ channels,membrane depolarization, the activation of voltage-gatedCa2+ channels, and subsequent Ca2+-dependent insulin exo-

From the Department of Medicine, Division of Endocrinology, Diabetes, andMetabolism, Washington University School of Medicine, St. Louis, Missouri.

Address correspondence and reprint requests to Dr. M. Alan Permutt, Metabo-lism Division, Washington University School of Medicine, 660 South Euclid,Campus Box 8127, St. Louis, MO 63110.

Received for publication 15 September 1995 and accepted in revised form 13November 1995.

mGPDH, mitochondrial glycerol-3-phosphate dehydrogenase; PCR, polymerasechain reaction; RT, reverse transcription; STS, sequence-tagged site; WUMS,Washington University Medical School; YAC, yeast artificial chromosome.

cytosis (1). NADH-reducing equivalents generated from theoxidation of glyceraldehyde-3-phosphate are a major sub-strate for glycolvtically derived ATP production in mitochon-drial oxidative phosphorylation (3,4). The transfer into themitochondria of electrons from cytosolic NADH is largelymediated by the glycerol phosphate shuttle (5). Flavin-adenine dinucleotide-linked mitochondrial glycerol-3-phos-phate dehydrogenase (mGPDH) is a key enzyme in thisshuttle system (5), and its catalytic activity is 40- to 70-foldhigher in pancreatic islets and insulinoma relative to othertissues, such as whole pancreas, liver, or skeletal muscle.Thus, preferential mitochondrial oxidation of utilized glu-cose is greatly favored in these cells (6). It has beenpostulated that decreased (3-cell mGPDH activity could re-sult in diminished glycolvtically derived ATP productionand hence in impaired glucose-stimulated insulin secre-tion (7,8). There is now evidence for the existence ofabnormal pancreatic islet mGPDH activity in several ani-mal models of diabetes, as well as in a limited number ofhumans with NIDDM (7-9). Thus, variations in the se-quence or level of expression of the human mGPDH genemay contribute to the pathogenesis of NIDDM.

The cDNAs encoding rat testes, liver and islets, as well ashuman cervical carcinoma cell mGPDH have been recentlyisolated (10-13). We have cloned mGPDH cDNAs from ahuman pancreatic (3-cell tumor and examined mGPDHmRNA in native human islets by reverse transcription (RT)polymerase chain reaction (PCR). The results indicate theexistence of a novel alternate 5' untranslated sequence,while a predicted translated sequence nearly identical to thatpresent in other rat and human tissues was observed. Weshow for the first time that, in accordance with previousstudies demonstrating high mGPDH activity in p-cells, theabundance of pancreatic islet mGPDH mRNA is distinctlyhigher than that observed in other tissues (6). The geneencoding mGPDH was mapped to yeast artificial chromo-some (YAC) clones that form part of a defined contig on thelong arm of chromosome 2, and a flanking polymorphicdinucleotide marker for NIDDM linkage studies was thusidentified.

RESEARCH DESIGN AND METHODSScreening of a human (J-cell tumor cDNA library and sequenceanalysis. A partial 749 bp human mGPDH cDNA fragment was isolatedwith PCR from human islet reverse transcribed cDNA, using degenerateprimers 5'-AGC(A/G/T)GG(A/T)GG(G/A/C/T)AA(A/G)TGGACCAC(A/G/

262 DIABETES, VOL. 45, FEBRUARY 1996

S. FERRER AND ASSOCIATES

C/T)TA-3' and 5'-AG(C/T)TC(A/C/T)ACCTG(A/G/C/T)CCATT(C/T)TT(A/G)TT-3' derived from the published sequence of rat testes mGPDH(Genbank/EMBL accession no. U08027). This partial cDNA fragmentwas 32P-labeled by random priming to a specific activity greater than 1X 109 cpm/(xg and used as a hybridization probe to screen 5 X 105 pfufrom a human insulinoma cDNA library cloned in \ ZAP II, essentially asdescribed (14). Sequencing was carried out on an ABI 373A automatedsequencer, and the resulting sequences were compared with the nonre-dundant nucleic acid and protein databases (National Center for Bio-technology Information) using BLASTN and BLASTX algorithms.Tissue procurement and RNA isolation. Tissue preparations thatwere enriched by density gradient centrifugation for either humanpancreatic islets or pancreatic exocrine tissue were obtained from theHuman Islet Transplantation Center, Washington University MedicalSchool (WUMS) (Dr. David Scharp). Other human tissues were obtainedthrough a rapid autopsy protocol (Drs. J. Saffitz and R. Schmidt,Department of Pathology, WUMS), or from a surgical specimen (Dr. J.Norton, Department of Surgery, WUMS). All human tissue was obtainedwith institutional approval and informed consent. Rat pancreatic isletswere obtained by a modified collagenase procedure, as previouslydescribed (15). (3TC6-F7 cells were a gift from Dr. Shimon Efrat (AlbertEinstein College of Medicine, Bronx, NY). Total RNA was extracted byhomogenization in guanidinium thiocyanate and cesium chloride gradi-ent ultracentrifugation or by a modified two-step procedure using Trizol(BRL, Bethesda, MD) reagent as recommended by the manufacturer.Northern blot and RT-PCR analysis. Northern blots prepared withtotal RNA samples and a Multiple Tissue Northern (Clontech, Palto Alto,CA) polyA+ RNA blot were hybridized with an a-32P-labeled cDNAprobe corresponding to the 3' 2.2 Kb Ncol digestion product of INS-mGPDH cDNA insert or with a (3-actin cDNA probe and washed at a finalstringency of 0.1 x sodium chloride-sodium citrate, 0.1% SDS at 65°Cbefore exposure to X-ray film for 48 h.

RT-PCR was carried out with human pancreatic islet, insulinoma,muscle, cerebellum, and liver total RNAs treated with RNAse-freeDNAse (GIBCO-BRL), extracted with phenol-chloroform, ethanol pre-cipitated, and quantified by spectrophotometry. The integrity and accu-racy of quantitation of RNA was then ascertained by ethidium bromide-formaldehyde gel electrophoresis. cDNA was synthesized from 1 |xgtotal RNA by oligo-dT priming with Superscript II reverse transcriptase(GIBCO-BRL). PCR amplification was carried out from cDNA corre-sponding to 80 ng total RNA for 30 cycles at 94°C, 52-60°C, and 72°C for1 min each step, with a final extension at 72°C for 10 min. Primers used forRT-PCR were 5'-AGACTTGCACTGAAGGTGCA-3' (M0), 5'-AGGAGAAGC-CAGATCCCAGG-3' (Ml), 5'-GGGCCGAGGCTCTGATTCTG-3' (Ml'), 5'-CACCTCCTCCAACAAGAATO3' (M2), 5'-GATTCTTGTTGGAGGAGGTG-3' (M3), 5'-TAGTGCTTCTGCTGCTGGTC-3' (M4), 5'-TATTACAGCCCA-GAGAGCAT-3' (M5), 5'-TCCCCTCTTCTCACTTCAAC-3' (M6), 5'-AGTTT-TGACAGCAGCATTTA-3' (M7), and 5'-CCTTCAAGGGGGTAAAGATTG-3'(M8), 5'-GTGGTTTGTCTTCAGCTCACCTGG-3' (M9), 5'-CCACACCTTA-AAGTTTTGGAATG-3' (M10), and 5'-TTTTTTTTTTTT(A/G/C)(A/G/C/T)-3' (Mil). PCR products were resolved on an ethidium bromide-stained 2% agarose gel and visualized under ultraviolet light.Chromosomal localization by somatic cell hybrid and mega-YAClibrary analysis. The chromosomal assignment of mGPDH was deter-mined by PCR amplification of DNA from a panel of rodent/humansomatic cell hybrids (NIGMS Mutant Cell Repository, Camden, NJ). Asequence-tagged site (STS) derived from the 3' coding and untranslatedregion of Ins-mGPDH cDNA was developed for this purpose, usingprimers M9 and M10. PCR amplification was carried out for 30 cycles at95°, 56°, and 72°C for 1 min each step, with a final extension at 72°C for10 min. Reaction products were analyzed on an ethidium bromide-stained 2% agarose gel.

Physical mapping to YACs was carried out by PCR screening of thelarge-insert Centre d'Etude du Polymorphisme Humain (CEPH) "B" YAClibrary, purchased from Research Genetics (Huntsville, AL) (16). Analiquot (1 |xl) of each 24-"block" YAC DNA pool (representing 18,432clones) was amplified in a 10-|xl reaction, using the same STS and cyclingparameters described for somatic cell hybrid analysis. A second roundof PCR screening was carried out with 28 YAC DNA pools from eachpositive block pool (representing eight plates with 768 YAC clones). Ofthe samples in the second round, 8 identified the plate containing theYAC clone, 8 pool samples identified the plate row, and 12 samplesidentified the column. To confirm the STS content of positive YACaddresses or to clarify ambiguous scores, individual clones were freshlygrown in YPD plates inoculated with single-colony frozen stocks andthen assayed for the STS by PCR. Information on positive YAC clones

5'UTR B | g |

MO M - l *

S'UTHA ^

1

1 M2_ M3 ^M4

1

MS U16 M8 M7 M9 M1O M11

I ,

Obp

5'

1000 bp 2000 bp 3000 bp

3'

FIG. 1. Schematic of INS-mGPDH, a human insulinoma mGPDH cDNAclone. The white boxed area represents the predicted coding region. The5' and 3' untranslated regions are indicated by shaded or striped boxedareas. The divergent 5' untranslated region reported in a Hela cellmGPDH cDNA (Genbank accession no. U12424) is shown as a separateshaded box joined to INS-mGPDH by a diagonal line and labeled 5' UTRB. A thin horizontal line is used to indicate the portion of the 5' segmentof INS-mGPDH believed to represent a chimeric addition of anon-mGPDH-related clone (see text). The approximate locations ofM0-M11 primers used for RT-PCR and chromosomal localization studiesare indicated with arrows. Segments amplified by RT-PCR areschematized with thick horizontal lines.

was obtained using the Whitehead Institute/MIT Center for GenomeResearch World-Wide Web servers, and further analysis of markerscontained by these clones was done using the Genome Data Base andCooperative Human Linkage Center servers.

RESULTSIsolation of a human insulinoma mGPDH cDNA. Screen-ing of 5 X 105 clones from a human insulinoma cDNA librarywith a partial mGPDH probe generated by PCR led to theisolation of 12 3,085-bp cDNA clones (INS-mGPDH). Se-quence analysis revealed an open frame encoding a 727-amino acid protein, 96% identical to that reported from ratliver and testes (11,12). During the course of this work, aHela epithelial cell line mGPDH cDNA (Genbank accessionno. U12424) was reported that contained an open-readingframe that differed in a single amino acid (R525H), resultingfrom a G to A substitution at nucleotide position 2095 ofINS-mGPDH. This amino acid change did not affect any ofthe three conserved functional domains of mGPDH that wereinitially predicted from the rat sequence, namely the calci-um-, flavin adenine dinucleotide-, and glycerol phosphate-binding sites (12). Also, compared with the Hela cell mGPDHcDNA, INS-mGPDH had a silent base change in position1910, three single nucleotide substitutions in the 3' untrans-lated region (positions 2773, 2836, and 2867), an additional 69bp in the 3' end that contained a polyadenylation signal (notshown), and completely diverged in the 509-bp 5' segmentlocated 12 bp upstream from the predicted initiator methio-nine (5' UTR "A", Fig. 1) (10). This insulinoma mGPDH 5'untranslated region had no significant homology with any ofthe two previously identified rat liver and testes 5' untrans-lated region variants, whereas 60% homology exists betweenthe human Hela and rat liver cDNA 5' untranslated regions(10-12). INS-mGPDH contained an in-frame ochre stopcodon at position - 6 from the initiator AUG, thus discardingpossible alternate mGPDH translation products derived fromthis clone.Assessment of mGPDH mRNA sequence heterogeneityby RT-PCR. The alternate 5'-untranslated regions identifiedin INS-mGPDH (Fig. 1, 5' UTR A) and Hela mGPDH (5' UTRB) cDNAs were further investigated by RT-PCR of nativehuman islet and insulinoma RNA. Forward oligonucleotideprimers (Ml, Ml') corresponding to regions located 50 or 108bp upstream from the translation initiation codon of INS-mGPDH (5' UTR A) and Hela mGPDH cDNAs (5' UTR B),respectively, and a reverse primer complementary to the

DIABETES, VOL. 45, FEBRUARY 1996 263

HUMAN ISLET-CELL MITOCHONDRIAL GPDH

•- c "5B

•5 8 o a .8 § B £._ <uu w « — -a _• ro

5'UTRB(157bp)

5' UTR A (99 bp)

FIG. 2. Tissue distribution of mGPDH alternate 5' untranslated regionsA and B. 5' UTR A and B sequences were coamplifled by RT-PCR fromfirst-strand cDNA of multiple human tissues, using primers Ml, Ml', andM2 (see Fig. 1 for approximate location), and analyzed by 2% agarosegel electrophoresis. 5' UTR A-speciflc product is 99 bp; 5' UTR B-speciflcproduct is 157 bp.

common 5' coding region (M2) revealed the two expectedamplification products in human islet, insulinoma, liver,cerebellum, and skeletal muscle cDNA. This was tested withthe two primer pair combinations separately (not shown) orcombined to simultaneously assess the presence of the twosequences in different tissues (Fig. 2).

While RT-PCR experiments proved that the 100-bp se-quence 5' of the putative translation start codon of INS-mGPDH (5' UTR A) forms part of islet-cell mGPDH mRNAs,the 383-bp segment situated in the 5'-most end of thisuntranslated region (schematized as a thin line in Fig. 1)likely represents an artifactual concatemer. BLASTN se-quence database analysis shows that this segment contains asequence that is 94-97% identical to the putative 3' end ofanonymous-expressed sequence tags isolated from differentdirectionally cloned human cDNA libraries (Genbank acces-sion nos. D20040, D29540). RT-PCR using a forward primercorresponding to this region (MO) and a reverse primercomplementary to a region in the 5' end of the mGPDHcoding sequence (M2) failed to amplify the interveningsequence predicted from the INS-mGPDH cDNA using hu-man pancreatic islet, insulinoma, or liver cDNA, while theexpected size fragment was readily amplified from a dilutedaliquot of the cDNA library (not shown).

RT-PCR analysis was also used to confirm that the codingsequence of mGPDH present in the insulinoma cDNA is alsoexpressed in native human pancreatic islets. Five primer setswere designed to amplify different segments spanning theentire coding region (Ml + M2, Ml + M4, M3 + M6,M5 + M7, and M8 -I- Mil, schematized in Fig. 1) and used forRT-PCR amplification of human islet RNA. The amplificationproducts were sized by agarose gel electrophoresis anddirectly sequenced, resulting in products identical to thoseexpected from the insulinoma cDNA (not shown). Thesefindings are thus consistent with the existence of a humanislet-cell mGPDH protein with a primary structure that isanalogous to that reported in other tissues (10-14).Tissue distribution of mGPDH. Northern blot analysis wascarried out with a 32P-labeled 2.2-Kb cDNA fragment corre-sponding to the 3' portion of INS-mGPDH. A distinct 6.5-Kbtranscript was detected in 8 |xg total RNA from human isletsafter 5 days' exposure, while RNAs of identical size, but atleast 10-fold lower in intensity, were detected in other humantissues such as brain, muscle, and heart (Fig. 3A). Very faint

6.5 Kb

4 Kb

2.4-3.1 Kb

Actin

6.5 Kb4 Kb

i_ _£>

— = 1 0

12 Kb -

6.5 Kb

4 Kb

3.1 Kb

2.4 Kb

FIG. 3. Northern blot analysis of mGPDH mRNA. A: Northern blot using8 (ig total RNA from human tissues or 5 (ig in the case of muscle,hybridized with mGPDH (top) or p-actin probes (bottom). Exposure timewas 5 days for mGPDH and 20 h for P-actin. Estimated sizes areindicated with arrows. Actin hybridization signal is 2 Kb in all tissuesexcept for skeletal muscle and heart, where the predominant mRNAcorresponds to the expected 1.6- to 1.8-Kb forms. Analogous results wereobtained in an independent experiment. B: Northern blot mGPDHanalysis of 2 u£ total RNA from rat islet and PTC6-F7 cells. Exposuretime was 2 days. C: Northern blot analysis of mGPDH in a multiplehuman tissue Northern blot containing 2 |Ag polyA+ RNA (estimated torepresent ~ 10-fold higher polyA+ RNA content than the Northern blotshown in A). Exposure time was 1 day. Hybridization of mGPDH RNAwas carried out with a partial 2.2-Kb fragment corresponding to the 3 'region of INS-mGPDH cDNA in A or with a probe generated by RT-PCRwith primers M3 + M6 and M5 + M7 in B and C.

4.4-Kb and diffuse 2.4- to 3.1-Kb bands were also seen in mosttissues. A similarly distinct 6.5-Kb band was observed in 2 fxgtotal RNA prepared from rat pancreatic islets and moreintensely in (3TC6-F7 tumor (3-cell line RNA (Fig. 3B). Hy-bridization of an mGPDH probe generated by RT-PCR with amultiple-tissue Northern blot prepared with 2 juug poly A+

RNA also disclosed the presence of a major 6.5-Kb band in alltissues, while minor bands of 12 Kb, 3.1 Kb, and 2.8 Kb couldalso be readily observed after prolonged exposures (Fig. 3C).The abundance of mGPDH transcripts among this panel ofhuman tissues, which did not include human islet RNA, washighest in skeletal muscle, where an intense 4.4-Kb transcriptwas observed in addition to those described above.Chromosomal localization of mGPDH. The gene encodingmGPDH was localized to chromosome 2 by segregationanalysis of a panel of human-rodent somatic-cell hybridscontaining different human chromosomes, using an STSdefined by primers M8 and M9 corresponding to the 3' codingand untranslated regions of INS-mGPDH cDNA (not shown).With the same PCR assay, 7 positives were identified from 24block pools from the CEPH "B" YAC library. Five of thesewere arbitrarily chosen for further analysis, and this led to

264 DIABETES, VOL. 45, FEBRUARY 1996

S. FERRER AND ASSOCIATES

WI-6265 WI-2436 GGAA-P17315 D2S1796

H 1 1 1 857e3(1010)

848b1 (1430)

851C3(890)

874a2(860)

874a12(790)

FIG. 4. Schematic showing YAC clones positive for an mGPDH STS, aUof which form part of Whitehead Institute/MIT contig WC-744. Horizontallines indicate individual YAC clones, and small vertical lines indicate theSTS markers shown above that map to each clone. The polymorphicdinucleotide repeat marker D2S141 is italicized. The sizes of each YAC(in Kb) are shown in parenthesis.

the identification of five unique mega-YAC genomic clones(Fig. 4). These YACs were then individually grown andreassayed by PCR to confirm the presence of the mGPDHSTS. The information provided by the Whitehead Institute/MIT Center for Genome Research World-Wide Web server(Data Release 7) indicated that all of these YACs form part ofa defined chromosome 2 YAC contig (WC-744) constructedby double linkage between STS. One of the 3 YAC clonespositive for mGPDH (848bl), as well as 10 others formingpart of contig WC-744, contained a chromosome 2q21-33highly polymorphic dinucleotide repeat genetic marker(D2S141), while another such marker (D2S142) has beenmapped to additional YACs that form part of contig WC-744.Based on the most conservative estimation, D2S141 is likelyto be located at <1.5 Mb from the mGPDH STS used here.

DISCUSSIONBecause the selective loss of glucose-stimulated insulinsecretion is a well-characterized defect in humans withNIDDM, the key molecules involved in (3-cell glucose sensingare logical candidates for use in investigating the pathophys-iology of this disorder (1). mGPDH is believed to be a majorcomponent in the p-cell glucose recognition apparatus, sinceit contributes to preferential oxidative metabolism of glu-cose and consequently to the generation of signals that resultin insulin secretion (4). The results of the current studyindicate that in spite of the specific role that mGPDH plays inP-cell physiology, human islet-cell RNAs encode a nearlyidentical protein to that previously characterized in a Helacell line (10). Interestingly, a novel alternate 5' untranslatedregion was identified in both native and tumor human isletcells. The existence of different 5' untranslated mGPDHmRNA regions is of potential interest because this findingsuggests the presence of alternate promoters and/or theexistence of variable translational efficiency in different celltypes or developmental stages. This could in turn provide anexplanation for the existence of tissue-specific differences inthe regulated or constitutive levels of expression of thisgene. Interestingly, thyroid hormone is known to enhancethe activity of this enzyme in liver, where basal activity islow, but not in islet cells and testes, where mGPDH activityis normally elevated (13). Although the RT-PCR coamplifica-tion experiments shown here have not detected the use of asingle variant in any of the tested tissues, purified humanislets and other tissues we have analyzed have a heteroge-neous cellular composition. Given that mGPDH mRNA ap-pears to be expressed ubiquitously, our PCR studies have notdiscarded the possibility that a single alternate 5'-untrans-

lated sequence is expressed in certain cell types, includingnative human p-cells.

A striking finding from the studies shown here is thepresence of a distinctly increased abundance of mGPDHmRNA in pancreatic islets. While no data had been previ-ously reported regarding the relative abundance of mGPDHmRNA in rat or human tissues, pancreatic islet and tumor(3-cell mGPDH catalytic activity is known to markedly ex-ceed that of a variety of other tissues, including muscle, liver,heart, and pancreas. This is in keeping with the allegedcritical role of this enzyme in the p-cell glucose-sensingapparatus (3,6). Therefore, the present Northern blot resultsshow that these tissue differences in mGPDH activity aredependent on the existence of high mRNA levels in pancre-atic islet cells. Previous studies described the presence of a2.4-Kb mRNA in testes but failed to detect this mRNA inother tissues (11). Although we did not assess RNA fromhuman testes, we consistently detected a major 6.5-Kb tran-script in a panel of human tissues and detected only aless-abundant 2.4-Kb form, using either a partial INS-mGPDHcDNA fragment or an RT-PCR fragment-generated mGPDHprobe under high-stringency hybridization conditions. It ispossible that the 2.4-Kb testes transcript observed by othersreflects a tissue-specific variation, while differences in RNAblotting-assay sensitivity may account for the lack of hybrid-ization signal in the remaining tissues that has been previ-ously reported (11). Given the well-documented existence ofdecreased mGPDH activity in multiple animal models ofdiabetes (7,8) our finding that mGPDH transcripts can bereadily detected in relatively small amounts of human islettotal RNA will be useful for studies aimed at the assessmentof the expression of this gene in islets from individuals withNIDDM.

We have now placed the mGPDH gene in a well-charac-terized region of the human genome physical map by screen-ing the CEPH "B" genome-wide YAC library, coupled withconfirmation by somatic cell hybrid chromosomal assign-ment. To this point, no form of NIDDM or other inheriteddisorder has to our knowledge been mapped to this precisechromosomal region. While an insulin-dependent diabetessusceptibility locus (IDDM 7) has been linked to a nearbygenetic marker, recent analysis with multiple regional mark-ers has refined this locus to a slightly more telomeric regionnear D2S152 (17). An important feature of the current studyis that we have successfully used a novel resource to reliablylocate a highly polymorphic microsatellite in the vicinity ofthe mGPDH gene. This marker can now be used in linkagestudies to assess the role of the mGPDH gene in the inheritedsusceptibility to NIDDM.

ACKNOWLEDGMENTSJ. F. is a recipient of a Fellowship Award from the JuvenileDiabetes Foundation International. M.A. and A.N. are recip-ients of Mentor-Based Fellowship Awards from the Ameri-can Diabetes Association. M.A. P. has a research grant fromthe National Institutes of Health (DK-16746). Genbank acces-sion no. U36310.

REFERENCES1. Meglasson MD, Matschinsky FM: Pancreatic islet glucose metabolism

and regulation of insulin secretion. Diabetes Metab Rev 2:163-214, 19862. Malaisse W, Sener A, Herchuelz A, Hutton J: Insulin release: the fuel

hypothesis. Metabolism 28:373-386, 19793. Dukes ID, Mclntyre MS, Mertz RJ, Philipson LH, Roe MW, Spencer B,

DIABETES, VOL. 45, FEBRUARY 1996 265

HUMAN ISLET-CELL MITOCHONDRIAL GPDH

Worley JF: Dependence on NADH produced during glycolysis for beta-cell glucose signaling. J Biol Chem 269:10979-10982, 1994

4. MacDonald MJ: Elusive proximal signals of beta-cells for insulin secre-tion. Diabetes 39:1461-1466, 1990

5. Stryer L: Biochemistry. 4th ed. New York, WH Freeman, 19956. MacDonald MJ: High content of mitochondrial glycerol-3-phosphate

dehydrogenase in pancreatic islets and its inhibition by diazoxide. J BiolChem 256:8287-8290, 1981

7. Ostenson CG, Abdel HS, Rasschaert J, Malaisse LF, Meuris S, Sener A,Efendic S, Malaisse WJ: Deficient activity of FAD-linked glycerophos-phate dehydrogenase in islets of GK rats. Diabetologia 36:722-726, 1993

8. Giroix MH, Rasschaert J, Bailbe D, Leclercq MV, Sener A, Portha B,Malaisse WJ: Impairment of glyceroi phosphate shuttle in islets from ratswith diabetes induced by neonatal streptozocin. Diabetes 40:227-232,1991

9. Fernandez AJ, Conget I, Rasschaert J, Sener A, Gomis R, Malaisse WJ:Enzymatic, metabolic and secretory patterns in human islets of type 2(non-insulin-dependent) diabetic patients. Diabetologia 37:177-181, 1994

10. Lehn DA, Brown LJ, Simonson GD, Moran SM, Macdonald MJ: Thesequence of a human mitochondrial glycerol-3-phosphate dehydroge-nase-encoding cDNA. Gene 150:417-418, 1994

11. Muller S, Seitz HJ: Cloning of a cDNA for the FAD-linked glycerol-3-phosphate dehydrogenase from rat liver and its regulation by thyroid

hormones. Proc Natl Acad Sci USA 91:10581-10585, 199412. Brown LJ, MacDonald MJ, Lehn DA, Moran SM: Sequence of rat

mitochondrial glycerol-3-phosphate dehydrogenase cDNA: evidence forEF-hand calcium-binding domains. J Biol Cliem 269:14363-14366, 1994

13. Macdonald MJ, Moran SM, Simonson GD: The amino acid sequence of thepancreatic islet mitochondrial glyceroi phosphate dehydrogenase is notunique and the enzyme is not thyroid or glucose responsive. ArchBiochem Biophys 319:305-308, 1995

14. Ferrer J, Nichols CG, Makhina EN, Salkoff L, Bernstein J, Gerhard D,Wasson J, Ramanadham S, Permutt MA: Pancreatic islet cells express afamily of inwardly rectifying K+ channel subunits which interact to formG-protein activated channels. J Biol Chem 270:26086-26092, 1995

15. McDaniel ML, Colca JR, Kotagal N, Lacy PE: A subcellular fractionationapproach for studying insulin release mechanisms and calcium metabo-lism in islets of Langerhans. Methods Enzymol 98:182-200, 1983

16. Cohen D, Chumakov I, Weissenbach J: A first-generation physical map ofthe human genome. Nature 366:698-701, 1993

17. Copeman JB, Cucca F, Hearne CM, Cornall RJ, Reed PW, Ronningen KS,Undlien DE, Nistico L, Buzzetti R, Tosi R, Pociot F, Nerup J, Cornells F,Barnett AH, Bain SC, Todd JA: Linkage disequilibrium mapping of a type1 diabetes susceptibility gene (IDDM7) to chromosome 2q31-q33. NatureGenet 9:80-85, 1995

266 DIABETES, VOL. 45, FEBRUARY 1996