Proc. Natl. Acad. Sci. USA Vol. 88, pp. 7294-7297, August 1991 Medical Sciences Human liver gene: Cloning and sequence determination of two alternatively spliced cDNAs (human liver glucokine cDNA/alternative splicing/cassette exon/ isoforms) YUKIO TANIZAWA, LASZLO 1. KORANYI, CRIS M. WELLING, AND M. ALAN PERMUTT Metabolism Division, Washington University School of Medicine, St. Louis, MO 63110 Communicated by William H. Daughaday, May 17, 1991 (receivedfor review February 20, 1991)

ABSTRACT A human liver glucokinase (ATP:D-hexose structure predicted a protein of 465 amino acids and a 6-, EC 2.7.1.1) cDNA was isolated from a molecular mass of 52 kDa; this sequence was 53% identical liver cDNA library. This cDNA (hLGLK1) appeared to be full to rat brain type I. The rat glucokinase gene was length [2548 base pairs (bp) plus additional poly(A) residues], subsequently shown to be encoded by 10 exons in 15.5 as its size was consistent with a single 2.8-kilobase (kb) glu- kilobases (kb) of DNA (11). Cloning of an islet glucokinase cokinase mRNA on Northern blot analysis of liver poly(A)+ cDNA from rat insulinoma revealed that the mRNA was at RNA. The cDNA contained an open reading frame of 1392 bp least 200 bp longer on the 5' end and that exon 1 was different, that predicted a protein of 464 amino acids and a molecular resulting in an amino terminus differing by 15 amino acids mass of 52 kDa; this protein has 97% identity to rat liver between the two tissues (13, 14). Alternative splicing of glucokinase. Fourteen residues on the amino terminus of the glucokinase mRNAs has been observed in insulinoma (13), predicted human liver glucokinase, however, differed com- liver cells (15), and pituitary cells (16) from rodents. pletely from those of the predicted rat liver enzyme and could Defects in glucokinase activity have long been suspected as be explained by alternative splicing ofa 124-bp cassette exon in contributors to the aberrant glucose metabolism of non- human cDNA. A second glucokinase cDNA (hLGLK2), missing insulin-dependent diabetes mellitus in human (for review, see the 124-bp cassette exon, was isolated by PCR amplification of ref. 9). Isolation of the rat cDNA (12, 13) gave us the human liver cDNA. The hLGLK2 cDNA contained an open opportunity to screen a human liver cDNA library and isolate reading frame of 1398 bp from an ATG codon at position 164, glucokinase cDNA clones. We now report the sequence ofan encoding a predicted protein of 466 residues, 98% identical to apparent full-length clone* and compare its predicted amino the rat enzyme, but different from the predicted protein of acid sequence to that ofrat glucokinases. In addition, another hLGLK1 cDNA by 16 amino-terminal residues. In contrast, liver glucokinase mRNA was isolated with a 124- hLGLK1 cDNA contains multiple initiator codons upstream of deletion, which predicts a protein differing by 16 amino- the predicted initiator codon at position 294 within the cassette terminal residues. exon. Translation of the two mRNAs in vitro by a reticulocyte lysate system resulted in proteins of the expected size (52 kDa) METHODS for both mRNAs; yet hLGLK2 mRNA was translated four to cDNA Library Screening, DNA Sequencing, and Data Anal- six times more efficiently. These results suggested that the ysis. A human islet cDNA library (7) was initially screened alternative splicing ofa cassette exon in hLGLK1 resulted in an with a rat-islet glucokinase cDNA (13), and a 2-kb clone mRNA with an upstream initiator codon and reduced function. (phIGLK) was isolated. This clone was used to screen a The relative biological activity of the two isoforms of human human liver oligo(dT)-primed A ZAP (Stratagene) cDNA glucokinase and their possible developmental and/or meta- library. Inserts were subcloned into Bluescript SK+ (Strat- bolic regulation remain to be determined. agene) and M13mpl8 and M13mpl9 RF DNA (BRL), and then sequenced in both strands. Questions of compressions Glucokinase, found exclusively in liver and pancreatic islet were resolved by sequencing with dITP and terminal deoxy- beta cells, is one member of a family ofhexokinases (ATP:D- , as described (17, 18). Analysis of hexose 6-phosphotransferase, EC 2.7.1.1) that appear to DNA and amino acid sequences was performed by programs have a common evolutionary origin (1-4). Glucokinase, or from DNAStar, Madison, WI. hexokinase type IV, is distinguished from the other hexoki- cDNA Synthesis and PCR Amplificatlio of Human Liver nases by (i) its low affinity for glucose, with a Km in the Glucokinase. All human tissues were obtained with institu- physiological range of plasma glucose concentration (5-15 tional approval and informed consent. RNA was extracted mM), such that glucose phosphorylation maintains a gradient (19) from human liver (D. Perlmutter, St. Louis Children's for glucose transport, (ii) by its lack of inhibition by glucose Hospital, S. Giddings, Veterans Administration Medical 6-, and (iii) by its molecular mass of50 kDa vs. 100 Center, St. Louis, and National Disease Research Inter- kDa for the other . Both tissues that express change, Philadelphia) poly(A)+ RNA was isolated, and glucokinase also express a low-affinity, high Km glucose Northern (RNA) blot analysis was done, as described (7, 17). transporter (Glut-2) (5-7), and thus these tissues play an First-strand cDNA was synthesized from 5 jig ofhuman liver important role in regulation of glucose metabolism. In the total RNA primed by (dT)1219 in a 40-tlI reaction mixture liver, the level of glucokinase activity is regulated by hor- (17), and RNA was removed with RNase H (BRL). The PCR monal and nutritional factors (8-11). was done on cDNA (5 pl) with AmpliTaq DNA Recent cloning and sequencing of a full-length rat liver (Perkin-Elmer/Cetus) in 50 pl, as described by the manu- glucokinase cDNA provided important direction for studying facturer, with 2 mM MgCl2 and 50 pmol each of oligonucle- regulation of expression of this enzyme (12). The deduced otide primers 10635, corresponding to 34-59 of hLGLK1 and primer 11133 complimentary to nucleotides The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" *The sequence reported in this paper has been deposited in the in accordance with 18 U.S.C. ยง1734 solely to indicate this fact. GenBank data base (accession no. M69051). 7294 Downloaded by guest on October 2, 2021 Medical Sciences: Tanizawa et al. Proc. NatL. Acad. Sci. USA 88 (1991) 7295

432-456. Cycle conditions were, after 3-min initial denatur- gSgeeges gecetogt gcogectes gocagcete ciatec es gccottct ccocgcgg 70 ocettoct cesogctoc ogctgtgct ggeeteogteogegoge otctctocct cc gcat 1'0 ation at 940C, 45 cycles of940C for 30 sec, and 71'C for 1 min, ctoctetto geceetce ggogtweo tigtstc seso co eccogco ccttgoctct 210 with a final extension of 720C for 9 min. .coetct ectet..etc e otoecacet etact_ cteccettooto 280 Construction of phLGLK2 and in Vitro Translation. The Nt Pro Arg Pro Arg Sor Sin Lou Pro Sin Pro Asn iS Sin Vol Clu smaller PCR-amplified glucokinase cDNA fragment (see Fig. oatotatl *ac e*ta eec ceo cc tee coo etc ceo coo eccea tee c te gag 341 Sin Ito Lou Atl Slu Ph. tn Lou Sin Clu Slu Asp Lou Lys Lys Vol Nit LAr Arg 2) was ligated to the HincII site of M13mp18 and confirmed cog ote ctg gee ga tte Coe cto Cog so sOc et aoa" gigto at ago en 396 by sequencing. To construct phLGLK2, the Xba I-Pst I Not Sin Lys Clu Not LAp Lrg Sly Lou Arg Lou Clu Thr Nis Clu Clu Ala Ser Val fragment of the phLGLK1, which contained nucleotides ate cog ag gag tgig goeg gg ets aetgOf Ace eat a 455 1-367 of hLGLK1, including the 124-bp insertion (see Re- Lys Not Lou Pro Thr Tyr Va1 lra Sa Thr Pro Stu Sty isr Glu Va1 Sly Asp Ph. sults), was replaced by theXba I-Pst I fragment from the M13 ga atg ctg C sec tac Its e tee tcc ceo go WC tea ga gtc sac tte 512 clone, missing the insertion. RNA was synthesized by using Lou isr Lou Asp Lou Sly Sly Thr Asn Phs Lrg Vol Not Lou Val Lye Val Sly Clu T3 RNA polymerase (United States Biochemical) and linear- etc tee ctg gN etg ot gg act am ttc an gtg at ets gtg a" gti gg fee S69 Sty Stu Slu Sty SIn Trp is Vat Lye Thr Lye Nis SIn Thr Tyr sr Its Pro Slu ized phLGLK1 or phLGLK2 as templates, as described (17). gt gagfl e0 g t ao gtg ag aC ma _e qe acg tc tee ate ccc gag 626 The RNA a band the on product gave single of expected size Asp Atl Not Thr Sly Thr Alt Slu Not Lou Pbs Asp Tyr 1t Ser Stu Cys Itl Ser 1.2% formaldehyde/agarose gel electrophoresis. In vitro gc goc a OCcc act got gag atg etc ttC sac tac atC tet gag toe atC tee 683 translation used 2 pg of synthetic RNA, 20 AM amino acids Asp Pb. Lou Asp Lye 01s Sln Not Lys 1is Lye Lys Lou Pro Lou Sly Ph. Thr Phs (except methionine), 40 OCi of [35S]methionine (>1000 Ci/ gac ttc ct 9g ag eact e qot ma c a a a oto ccc eto ggC ttC acC ttC 740 mmol; Amersham; 1 Ci = 37 GBq), 40 units of RNasin, and Ser Pha Pro Vol LAg Nis Clu Asp Ito Asp Lys Sly I1o Lou Lou LAn Trp Thr Lye 35 ,ul of nuclease-treated rabbit reticulocyte lysate (Promega) tee ttt cet gig am e go gac ate got ag ge ate Ctt etc m to aceagca 797 Sly Pbs Lye Ala sr Sly Ala Glu Sly Asn Asn Val Val Sly Leu Lou Arg Asp Ala in 50-1l volume. gO tte a" gcc tea asgeegm us an sat ete gig Me cit eto gs gOct 854 its Lys LrArg GSly Asp Pb. lu Not Asp Val Val Ala Net Val Asn Asp Thr Val RESULTS atc ma eg agWu c itt 9g at* got Ig gtg gec ate g sat gc acg gtg 911 Ala Thr Not lIe Ser Cy tyr Tyr Clu Asp His Sin Cys Clu Val Sly Net Its Val Cloning of Human Liver Glucokinase cDNA. A clone (ph- gee acg atg atc tee te tac tac gs gac cat Cog to ga itc gge gtg stc gt 968 LGLK1) was isolated that contained 2548 bp plus additional Sly Thr Sly Cys Asn Ala Cy tyr Not Clu Clu Not SIn Len Val Clu Lou Val Clu 3'-poly(A) residues (Fig. 1). A presumptive polyadenylyla- gg c9 ggc tgo sat gec tgo tim ag 9ga g9 a9 cg set gtg 9ga ctg gtg 9ga 1025 Sly Asp Clu Sly Arg Nt Cys Val Asn Thb Clu Tap Sly Als Phb Sly Asp Sar Sly tion signal was seen in the sequence ATTAAA (20), 17 bases 1082 upstream of the poly(A) region. An open reading frame from s sa g"a ggo Ceg a*t tgo gtc st ace 9a tgo gO see tte us gac tee ggc Clu Lou Asp Clu Phb Lou Lou Clu Yyr Asp Arg Lou Val Asp Clu Sar Ser Ala Asn ATG at position 294 to a stop codon at position 1687 encoded 9ga ctg sae gea tte ctg ctg 9ga tat gac cgo cta gig ga ga ag tet gee ac 1139 a predicted protein of 464 amino acids with an estimated Pro Sly SIn SIn Lou tyr Clu Lye Lou II. Sly Sly Lys Tyr Not Sly Clu Leu Val molecular mass of 52 kDa and a pI of 5.07. Northern blot CCC Mt cog cog cti tat g 9ag etc ata gnt ggc aag tm a*t nc gag ctg gig 1196 analysis ofliver poly(A)+ RNA revealed a single 2.8-kb RNA Arg Lou Val Lou Lou Arg Lou Val Asp Clu Len Lou Leu Phe His Cly Clu Al Ser when hybridized to 32P-labeled glucokinase cDNA (data not cgg ctt gig ctg etc agg etc gti ge gsa am cgt etc tti C musgag gee tee 1253 shown). Clu GIn Lou Arg Thr Arg Sly Ala Phe Clu Thr Ars Phe Val Ser CIn Val Clu Ser Identification of a Cassette Exon in Genomic DNA. An ag cog cti cge ma Cgo g gee ttc gag cg cgo ttc gig teg cog gtg g age9 1310 Asp Thr Sly Asp Lrg Lys Sin Ilt Tyr Asn lle Lou Ser Thr Lou Sly Lou Ars Pro overall identity of 89% was observed between human and rat sm ge99c gmc Coe a" cogetc te aS ate cti aecKs cigu ts ego ccc 1367 liver a 124 in glucokinase, except for region of bp not present Ser Thr Thr Asp Cys Asp Its Val Arg Lrg Ala CyS Clu Ser Val Sr thr ArL Ala the rat cDNA (see Discussion). This 124-bp region in the te mcc a c e c tge sac atc gti cgo cgo gee tio ga ag Sgt tct scg cgc get 1424 human cDNA occurred at thejunction of exons 1 and 2 in the Ala Nis Nt Cys Ser Ala Sly Lou Ala Sly Val lt Asin rL Net Lrg Clu Ser Lrg rat cDNA with 90% identity 3' to exon 2. The predicted geg oem ait tge teg Mgu ci gegge gtec etc am cge *tgcgtcga9 ae ego 1481 protein has no identity to the rat liver protein for the first 14 Ser Clu Asp Val Not Arg lIe Thr Val Sly Val Asp Sly Sr Val tyr Lys Lou Nis amino acids, followed by a large region of high identity. To ox gag 9s egte ait Ce stc act gti nc gitg gt mgg tee gItg tc a" ctg cm 1538 Pro Ser Ph. Lys Clu Arg Phs Nis Ala sr Val Arg Ars Leu Thr Pro Ser Cys Clu pursue this 5' difference, human glucokinase genomic clones ccc age ttc sa ga egg ttc cat gee ag gtg cgo a9 ctg Wc ccc age tge 9ga 1595 were isolated and sequenced with exon-specific oligonucle- Ite Thr Phe Ila Clu iSr Clu Clu Sly Ser Sly Ars Sly Ala Ala Lou Val Ser Ala otide primers. Based on the structure of the rat genomic atc ac ttc atc g teg gag g ggc agt gC cg 99e geg gee ctg etc teg geg 1652 glucokinase gene, the 124-bp region corresponded to a cas- Val Ala Cys Lys Lys Ala Cye Not Lou Sly SIn sette exon between exons 1 and 2 (data not shown). gt9 gee tot sag a"g gee tot atg ctg g9c cog tgag gacogggecg ceagcgcg 1710 Isolation and Sequencing of a Second Human Glucokinase ga tgeca cageccca gcCccoggc tccatggga ogt9ctce mccgigetc gegecigtw 1780 ggW9cgg gectggeett gtcaggacc sagcc9cgte ccatmc9ct Mgassesga gcgggcctit 1850 cDNA. In search for alternative forms of human liver glu- tcectcagtt tttcgt99 acaceceag ggeectsae ggggtgcgg agoeagga magagatc 1920 liver RNA was reverse transcribed to cDNA, then tgagcec cacctttet cgetggtc * tttcecag aggcgttg CtCiitCag actttpotgC 1990 cokinase, atttecacac teagaget ttiggetcc cctg ecca gtctgga aggggtgecc teitgatcet 2060 analyzed by PCR amplification with glucokinase-specific geigitgeet cittCCcts gactcatc ctotgtsgg gaggsetcc aagettg cacetgc 2130 acctgea sagggeag ccagsggcts etetcatcee agicigee attttett9c ctiggctca 2200 oligonucleotide primers chosen for exons 1 and 2 of the agaggcaag ggastgg 9gs9gctc catggaggagCgitscesg ctigosim cceesgagga 2270 human glucokinase, based on the genomic structure ofthe rat ccttttetct Cccataccat cct9got9g cttgtgittc tg9gat9gac cetccgaes gtggaagag 2330 cagawcec cagetctsg cscOagggg CCCmC"agg ggaa9aggc csgectca tettcwetc 2410 glucokinase (see Fig. 1,5'-* 3', 34-59; 3' -* 5', 456-432). For ccatageget 99ctcaggaa gaaaceeaagc geattca gcCmmcca agggoamc ccatcatatg 2480 mcatgecmccctccatgc cesacismag attgtgtggg ttttttaatt assastgtta aaagttttaa 2550 cDNA containing the cassette exon a PCR product of 423 bp 2558 was predicted, whereas without the cassette exon, a 299-bp *aaaaaaa product was predicted. Both products were seen (Fig. 2). FIG. 1. Nucleotide and predicted amino acid sequence of human Southern blot analysis and hybridization with 32P-labeled liver glucokinase 1 cDNA (hLGLK1) and enzyme. The 124 bases phLGLK1 confirmed that both PCR products were repre- deleted in liver glucokinase 2 cDNA (hLGLK2) are underlined. sentative of glucokinase mRNA. Hybridization of the blot Forward and reverse oligonucleotide primers used to amplify liver with an end-labeled oligonucleotide specific for the 124-bp cDNA by PCR as illustrated in Fig. 2 are indicated. region revealed only the larger 423-bp product (data not shown). encoding a protein of 466 residues, 98% identical to the rat The smaller PCR-amplified glucokinase cDNA fragment enzyme, but different from the predicted protein ofhLGLK1 was purified, subcloned into M13mpl8, and sequenced. The cDNA by 16 amino-terminal residues. sequence included the region of flanking primers at nucleo- In Vitro Translation of the Two Glucokinase mRNAs. Each tides 34-59 to nucleotides 432-456, but with 124 bp missing cDNA was subcloned into RNA vectors, and relative to the sequence of phLGLK1, as indicated in Fig. 1. synthetic mRNAs translated in a reticulocyte lysate cell-free Thus hLGLK2 cDNA contained a predicted open reading system resulted in proteins the predicted size for glucokinase frame of 1398 bp from an ATG codon at position 164, (52 kDa), as well as 48-kDa proteins, and less abundant Downloaded by guest on October 2, 2021 72% Medical Sciences: Tanizawa et al. Proc. NatL Acad Sci. USA 88 (1991)

1 2 A 4 E A kDa 2

97.4 -

66 -

f506, 517 -396 45 - -344 "298 36 - -220

29 - 24 -

FIG. 2. Analysis of human liver mRNA by PCR amplification FIG. 3. In vitro translation of synthetic hLGLK1 and -2 mRNAs with glucokinase-specific oligonucleotide primers (indicated in Fig. by rabbit reticulocyte lysate, as described. Aliquots (2.5 Al) of the 1). Total liver RNA was reverse transcribed, and equal aliquots of reaction mixture were subjected to SDS/10%/ PAGE (20). Fluorog- cDNA were amplified. Products were separated on a 2.00% agarose raphy was for 1 hr at -800C with Entensify (NEN). Lanes: 1, no gel and stained with ethidium bromide. Lanes: 1, control without RNA; 2, RNA transcribed from phLGLK1; 3, RNA transcribed from RNA; 2-5, human liver cDNA from four different individuals; 6, phLGLK2. Positions of molecular mass markers (Sigma) are indi- DNA size markers, 1-kb ladder (BRL; 1 A&g). cated. smaller proteins (Fig. 3). Densitometric analysis ofthe trans- identity with a comparable region (1-172 bp) ofthe rat cDNA lation product revealed that hLGLK2 mRNA was translated (Fig. 4). The next 124 bp in the human hLGLK1 cDNA have four to six times more efficiently than hLGLK1 mRNA for no homology with the rat cDNA. For human cDNAs the the predicted 52-kDa glucokinase protein. region comparable to that encoded by rat exons 2-10 has 88% sequence identity, and the 3'-untranslated region has 68% DISCUSSION identity. An isoform ofglucokinase cDNA was also described in rat We isolated a human liver glucokinase cDNA (hLGLK1) of 2548 bases liver with 151 bp ofDNA between exons 1 and 2 shown to be exclusive of poly(A) residues, which appeared to due to alternative splicing ofan additional cassette exon (15). be full length as this size was consistent with the single 2.8-kb Although insertion of the 124-bp cassette exon in the human band seen on Northern blot analysis. If the initiator codon at is start glucokinase (hLGLK1) cDNA occurred at the same site as 294 the preferred translation site (see below), this insertion in the mRNA would encode a protein of 464 amino acids and an alternatively spliced rat liver enzyme, there estimated molecular mass of 52 kDa. Of note is the fact that was no sequence identity seen between the two insertions. a fusion protein of hLGLK1 with glutathione S-, Although hLGLK1 was the only clone we isolated from the expressed in bacteria, was a functional glucokinase enzyme cDNA library of 5 x iOs plaques, hLGLK2 seemed to be (Y.T., unpublished work). A second form of glucokinase more abundant according to the result of the PCR amplifi- mRNA with a 124-nucleotide deletion was observed by PCR cation (Fig. 2). In addition, the relative amount of the two amplification of human liver cDNA. The hLGLK2 cDNA forms seemed to vary among the individuals. Nevertheless, contained a predicted open reading frame of 1398 bp from an because of the difficulty in the quantification by PCR ampli- ATG codon at position 164, encoding a predicted protein of fication, we cannot be certain about the relative abundance 466 residues, 98% identical to the rat enzyme, but different of these mRNAs. Whether nutritional and/or hormonal fac- from the predicted protein of hLGLK1 cDNA by 16 amino- tors alter levels of the two forms of human glucokinase terminal residues. This mRNA is thought to occur by alter- mRNA is also unknown. This question could be important for native splicing of a cassette exon between exons 1 and 2 in future studies, considering the potential physiological and genomic DNA. pathological consequences, were the translated forms of The rat liver glucokinase cDNA has been shown to be glucokinase different in catalytic properties or stability. encoded by 10 exons within 15.5 kb of genomic DNA (11). Rat glucokinase has a single initiator codon followed by a Comparisons ofthe nucleotide sequences ofthe 5' ends ofthe long open reading frame (21). Upstream ofthe ATG at 294 for human liver and rat glucokinase cDNAs revealed a region hLGLK1 are six other ATG codons (Fig. 1, positions 94, 164, from 30-211 bp of hLGLK1 that shared 66% sequence 170, 174, 241, and 284). Initiator ATG codons at positions 94, 5' 3, 293 42 1350 863 Human Liver I (A) FIG. 4. Comparison of human hLGLK1 and -2 and rat liver (12) glucokinase cDNAs. Heavy lines under 163 48 1350 863 the 5' regions refer to conserved regions, and the 124 Human Liver 2 |\\ \ \ (A) deletion in hLGLK2 relative to hLGLK1 is indicated by dotted lines. Boxes refer to predicted coding re- gions, and lines represent 5'- and 3'-untranslated re- gions, respectively; hatched areas represent the highly 12745 1350 835 conserved coding regions, and the other areas refer to (A) differences in amino-terminal coding regions, as de- Rat Liver scribed. Termination signals ATTAAA are indicated by closed triangles. Downloaded by guest on October 2, 2021 Medical Sciences: Tanizawa et al. Proc. Nati. Acad. Sci. USA 88 (1991) 7297

lOv 20v We thank Drs. Rick Wetsel and Harvey Colten, St. Louis Chil- HLGLK1 MPRPRSQLPQPNSQVEQILAEFQLQEED dren's Hospital, for providing the human liver cDNA library, Drs. 11111111111111 Stuart Adler and Mike Mueckler for helpful discussions and review -RATGLK MAMDTTRCGAQL-LTLVEQILAEFQLQEED ofthe manuscript, and Jeannie Wokurkafor help in preparation ofthe 1111 11 11 111111111 manuscript. Dr. Mark Magnuson provided the rat glucokinase HLGLK2 MAMDVTRSQAQTALTLVEQILAEFQLQEED cDNA, a preprint ofunpublished data, and helpful advice. This work was supported, in part, by Grant DK16746 (M.A.P.) from the FIG. 5. Comparison of the NH2 termini of human and rat liver National Institutes of Health. L.I.K. was a recipient of a Juvenile glucokinases. Diabetes Foundation Fellowship Award. Y.T. was the recipient of a Mentor Based Fellowship Award of the American Diabetes Asso- 164, 170, 174, and 284 predict small proteins that terminate ciation. before the initiation ATG at 294. Furthermore, the ATG codons at 94, 164, and 170 are potential "strong" initiator 1. Colowick, S. P. (1973) in The , ed. Boyer, P. D. codons, according to the rules ofKozak (20). Thus, hLGLK1 (Academic, New York), Vol. 9, pp. 1-48. might be a transcriptional unit encoding small upstream 2. Schwab, D. A. & Wilson, J. E. (1989) Proc. Nati. Acad. Sci. peptides, and these upstream ATGs might markedly affect USA 86, 2563-2567. the translation rate of hLGLK1 at ATG 294, which encodes 3. Middleton, R. J. (1990) Biochem. Soc. Transact. 18, 180-183. the predicted protein of 464 residues. The results of in vitro 4. Nishi, S., Susumu, S. & Bell, G. I. (1988) Biochem. Biophys. Res. Commun. 157, 937-943. translation ofsynthetic mRNAs for hLGLK1 and -2 indicated 5. Fukumoto, H., Seino, S., Imura, H., Seino, Y., Eddy, R. L., that both are functional mRNAs encoding full-length glucoki- Fukushima, Y., Byers, M. G., Shows, T. B. & Bell, G. I. nase-predicted protein of 52 kDa. The smaller (48-kDa) (1988) Proc. Natl. Acad. Sci. USA 85, 5434-5438. protein observed could represent either partially degraded 6. Thorens, B., Sarkar, H. K., Kaback, H. R. & Lodish, H. F. mRNA or protein, partially translated protein, or the prod- (1988) Cell 55, 281-290. ucts of initiation at downstream ATG codons 390 and 411, 7. Permutt, M. A., Koranyi, L., Keller, K., Lacy, P. E., Scharp, which would encode predicted proteins of48.3 and 47.3 kDa, D. W. & Mueckler, M. (1989) Proc. NatI. Acad. Sci. USA 86, 8688-8692. respectively. The translation of 52-kDa protein with hL- 8. Minderop, R. H., Hoeppner, W. & Seitz, H. J. (1987) Eur. J. GLK2 mRNA was four to six times greater (by densitometric Biochem. 164, 181-187. analysis) relative to that of hLGLK1 (Fig. 3). Because these 9. Matschinsky, F. M. (1990) Diabetes 39, 647-652. results were obtained in an in vitro system, the physiological 10. lynedjian, P. B., Pilot, P.-R., Nouspikel, T., Milburn, J. L., relevance of this finding cannot be determined. Further, a Quaade, C., Hughes, S., Ucla, C. & Newgard, C. B. (1989) full-length hLGLK2 cDNA clone was not isolated. Thus the Proc. Nati. Acad. Sci. USA 86, 7838-7842. determination of the relative level of expression of the two 11. Magnuson, M. A., Andreone, T. L., Printz, R. L., Koch, S. & Granner, D. K. (1989) Proc. Nati. Acad. Sci. USA 86, 4838- glucokinase enzymes in vivo awaits the development of 4842. specific antibodies and immunoblot analysis of liver protein. 12. Andreone, T. L., Printz, R. L., Pilkis, S. J., Magnuson, M. A. The amino acid identities between rat and the human liver & Granner, D. K. (1989) J. Biol. Chem. 264, 363-369. glucokinases were compared (Fig. 5). For hLGLK1 there is 13. Magnuson, M. A. & Shelton, K. D. (1989) J. Biol. Chem. 264, almost no identity in the first 14 amino acids with the rat, 15936-15942. whereas for hLGLK2 11/16 (69o) of the amino-terminal 14. Magnuson, M. A. (1990) Diabetes 39, 523-527. residues are identical; this is followed by an area with 97% 15. Hayzer, D. J. & lynedjian, P. G. (1990) Biochem. J. 270, amino acid identity in a region of 450-amino acid overlap. 261-263. 16. Hughes, S. D., Quaade, C., Milburn, J. L., Cassidy, L. & This comparison contrasts to 53% amino acid identity be- Newgard, C. B. (1991) J. Biol. Chem. 266, 4521-4530. tween rat glucokinase and the carboxyl-terminal region ofrat 17. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1989) in Molecular brain hexokinase. The putative glucose- and ATP-binding Cloning:A Laboratory Manual (Cold Spring Harbor Lab., Cold domains were also highly conserved (12). Spring Harbor, NY), 2nd Ed. As a consequence ofisolating two human liver glucokinase 18. Fawcett, T. W. & Bartlett, S. G. (1990) Biotechniques 9, 46- cDNAs, the contribution ofthis gene to genetic susceptibility 48. to non-insulin-dependent diabetes mellitus can now be as- 19. Chomczynski, P. & Sacchi, N. (1987) Anal. Biochem. 162, 156-159. sessed. The potential use ofalternative promoters, as well as 20. Kozak, M. (1984) Nucleic Acids Res. 12, 857-872. alternative splicing ofglucokinase mRNAs, makes the search 21. Liang, Y., Jetton, T. L., Zimmerman, E. C., Najafi, H., for defects in the glucokinase genes of diabetic subjects an Matschinsky, F. M. & Magnuson, M. A. (1991) J. Biol. Chem. interesting one. 266, 6999-7007. Downloaded by guest on October 2, 2021