Characterization of the bovine PRKAG3 gene: structure, polymorphism, and alternative transcripts

Characterization of the bovine PRKAG3 gene: structure,polymorphism, and alternative transcripts

Matthieu Roux,1 Angelique Nizou,1 Lionel Forestier,1 Ahmed Ouali,2 Hubert Leveziel,1

Valerie Amarger1

1Faculte des Sciences et Techniques, Unite de Genetique Moleculaire Animale, Unite Mixte de Recherches 1061 Institut national de larecherche agronomique (INRA)/Universite de Limoges, 123 av Albert Thomas, 87060 Limoges Cedex, France2Qualite des Produits Animaux, Institut national de la recherche agronomique (INRA) de Clermont Ferrand�Theix,63122 Saint Genes Champanelle, France

Received: 13 July 2005 / Accepted: 24 August 2005

Abstract

The bovine PRKAG3 gene encodes the AMPK c3subunit, one isoform of the regulatory c subunit ofthe AMP-activated protein kinase (AMPK). TheAMPK plays a major role in the regulation of energymetabolism and mutations affecting the genesencoding the c subunits have been shown to influ-ence AMPK activity. The c3 subunit is involved inthe regulation of AMPK activity in skeletal muscleand strongly inflences glycogen metabolism. Glyco-gen content in muscle is correlated to meat qualityin livestock because it influences postmortem mat-uration process and ultimate pH. Naturally occur-ring mutations in the porcine PRKAG3 gene highlyaffect meat quality by influencing glycogen contentbefore slaughter. We present the characterization ofthe bovine PRKAG3 gene and a polymorphismanalysis in three cattle breeds. Thirty-two SNPswere identified among which 13 are in the codingregion, one is in the 3¢ UTR, and 18 are in the in-trons. Five of them change an amino acid in thePRKAG3 protein sequence. Allelic frequencies weredetermined in the three breeds considered, and mu-tant alleles affecting the coding sequence are foundat a very low frequency. Alternative splicing siteswere identified at two positions of the gene, intro-ducing heterogeneity in the population of proteinstranslated from the gene.

Introduction

In the field of meat-producing animals, the influenceof genetics on meat quality is now well recognizedbut still poorly understood. Domestic animals havebeen subjected to intensive selection since thebeginning of domestication, particularly concerningtraits related to growth, development, body compo-sition, reproduction, behavior, and resistance todiseases (Andersson and Georges 2004). In cattle,tenderness is the primary quality attribute for con-sumer acceptance of meat, followed by juiciness andflavor. Extensive efforts are being made to controland improve these qualities but are still impaired bythe fact that these characters are subjected to manygenetic and environmental factors and the elucida-tion of the mechanisms involved requires extensivestudies. Breeding selection has certainly driven theaccumulation of natural mutations responsible forphenotypic differences. Finding these mutations andelucidating how they control a character is the majorobjective in farm animal genome research and in theunderstanding of complex phenotypic traits.

Skeletal muscle metabolism is a very complexparameter which has a major influence on meatquality because it influences both the structure andthe biochemical characteristics of the muscle. It isclosely linked to the energy intake and use and thusto some environmental factors that can be more orless controlled. However, muscle metabolism has astrong genetic determinism which is still poorlyunderstood because of its complexity in both thenumber of genes involved and the close interactionbetween them.

Previously, the AMP-activated protein kinase(AMPK) has been pointed out as one of the mainactors in the regulation of intracellular energy

Sequence data for this article have been deposited in the GenBanklibrary under accession numbers DQ082732�DQ082736.

Correspondence to: Valerie Amarger; E-mail: [email protected]

DOI: 10.1007/s00335-005-0093-0 � Volume 17, 83�92 (2006) � � Springer Science+Business Media, Inc. 2006 83

metabolism (Carling 2004). It is considered a cellularfuel gauge because it is activated by a drop in theenergy status. The effect of this activation is toswitch off energy-using pathways and switch onenergy-generating pathways, thus helping to restorethe energy balance within the cell. Numerousmechanisms of AMPK action on lipid and carbohy-drate metabolism have been proposed (Ferre et al.2003; Hardie et al. 2003). AMPK is a heterotrimericenzyme complex comprising a catalytic a subunitand regulatory b and c subunits. Seven differentisoforms (a1, a2, b1, b2, c1, c2, and c3), each encodedby a different gene, have been characterized so farand all the combinations (12 in total) are possible.The different combinations depend on the tissuetype and they have different levels of activity (Sta-pleton et al. 1996; Thornton et al. 1998; Kemp et al.1999; Cheung et al. 2000). There are now severalstudies suggesting that naturally occuring mutationsin AMPK subunits could cause significant physio-logic effects. However, most of these mutations wereidentified, so far, in c subunits only. Mutations inthe human c2 subunit are associated with cases ofcardiac hypertrophy in several unrelated families(Blair et al. 2001), as well as Wolff�Parkin-son�White syndrome (Gollob et al. 2001). A char-acteristic of these patients is an accumulation ofglycogen in the heart (Arad et al. 2002). This seemsto be a particular trait shared with the cases ofmutations in the c3 subunit but with a differenttissue localization. Indeed, the RN� mutation(R225Q) in the c3 subunit of the AMPK is respon-sible for an increase in the glycogen level in skeletalmuscles in pigs (Milan et al. 2000). The effect ofmutations in the c3 subunit on skeletal muscleglycogen content and overall muscle biochemicalcontent is strengthened by the fact that other naturalmutations (for instance, V224I) occurring in pigsare associated with a decrease in glycogen content aswell as lactate content and glycolytic potential,suggesting a ‘‘one gene�several polymorphisms�diverse phenotypes’’ model (Ciobanu et al. 2001).The implication of the c3 subunit in AMPK-depen-dent glucose uptake in skeletal muscle and glycogenresynthesis after exercise was demonstrated by usingAMPK c3 knockout mice and transgenic miceoverexpressing the mutated (225Q) c3 subunit(Barnes et al. 2004). An AMPK activator failed toincrease glucose uptake in AMPK c3 knockout mice,and glycogen resynthesis after exercise was impairedin AMPK c3 knockout mice and markedly enhancedin transgenic mutant mice. The effect on metabo-lism appears to be even larger since a transgenicmutant seems to be protected against triglycerideaccumulation and insulin resistance in skeletal

muscle when placed under a high fat diet (Barnes etal. 2004), suggesting an influence of the mutation onAMPK activity on downstream targets involved inthese mechanims.

Knowledge of muscle biochemical characteris-tics related to meat quality in cattle is still quitelimited. However, it is now well established thatthere is an important variability in muscle biologywhich is directly associated with downstreamparameters like postmortem meat maturation, driploss, cooking loss, tenderness, juiciness, and taste.Because of the large effect of AMPK c3 subunitmutations on muscle glycogen content and pH ob-served in pig and the close association between gly-cogen metabolism, pH, and meat quality in cattle(Immonen et al. 2000), we wondered how much thevariability in these characters could be related tomutations in the bovine gene encoding the AMPK c3subunit. We decided to assess the variability of thegene at the sequence and expression levels. We havecloned and sequenced the PRKAG3 gene in cattleand screened it for polymorphisms inside severalcattle breeds. We found evidence of a large sequencevariability of the gene and alternative splicing eventscausing additional diversity among the resultingproteins. These alternative splicing events appear tobe specific to the bovine gene and might be related toan adaptation to the particular ruminant metabo-lism.

Materials and methods

Genomic DNA extraction. Genomic DNA was ex-tracted from blood samples as previously described(Rouzaud et al. 2000) or from muscle samples usingthe FastDNA kit (Q.Biogene) and the Fast PrepTM

FP120 apparatus (Thermo Savant).

cDNA and genomic DNA amplification andsequencing. Total RNA was isolated from muscle(longissimus thoracis or rectus abdominalis) sam-ples, previously frozen in liquid nitrogen immedi-ately after death, using Trizol Reagent (Invitrogen).One microgram of RNA was reverse transcribedwith oligo-dT primer and Superscript II RT (Invitro-gen). Several primer pairs were designed on ex-pressed sequence tag (EST) sequences and used toamplify from cDNA or total genomic DNA. PCRreactions were carried out in a total volume of 20 llcontaining 20 ng DNA, 0.2 mM dNTPs, 1.5 mMMgCl2, 10 pmol of each primer, UptiTherm DNApolymerase, and reaction buffer (Interchim). Thecycling conditions included an initial incubation at94�C for 2 min followed by 35 cycles comprising 30sec at 94�C, 30 sec at 58�C, and 1 min 30 sec at 72�C.

84 M. ROUX ET AL.: TRANSCRIPTS AND POLYMORPHISMS OF THE BOVINE PRKAG3 GENE

In some cases, touchdown PCR conditions wereused. In these cases, the annealing temperature wasdecreased 1�C per cycle for the first seven cycles andthen maintained constant for the next cycles.

PCR products were purified using the QIAquickPCR purification kit (Qiagen) and directly sequencedusing Big Dye Terminator chemistry on a ABI PrismCycle sequencing 310 (Applied Biosystems). Theexon�intron boundaries were established by com-paring genomic and cDNA sequences. cDNA 3¢ endwas determined by rapid amplification of cDNAends (RACE) polymerase chain reaction (PCR) usingthe SMART RACE cDNA Amplification kit(Ozyme) and following the manufacturer�s instruc-tions. Total RNA was used for first-strand cDNAsynthesis. The 3¢-RACE products were cloned intopGEM-T vector (Promega) and sequenced.

BAC subcloning and sequencing. A BAC clonecontaining the PRKAG3 gene was obtained fromthe INRA bovine bacterial artificial chromosome(BAC) library (Eggen et al. 2001) by PCR screeningusing two pairs of primers amplifying two regionssituated at both ends of the gene. BAC DNA wasdigested by several restriction enzymes (EcoRI,HindIII, BamHI) and a suitable fragment was de-tected by Southern blot using a probe correspondingto PRKAG3 exon 1. The fragment was then gelpurified, subcloned into pUC18 vector and se-quenced.

Screening for polymorphism and mutationdetection. Screening for polymorphism was firstdone by Single Strand Conformation Polymorphism(SSCP) analysis with the GenePhor ElectrophoresisUnit (Amersham Pharmacia Biotech). PCR productswere half diluted in denaturing loading dye (95%formamide, 0.025% bromophenol blue, 0.025% xy-lene cyanol). The SSCP analysis was performed onGeneGel SSCP (Amersham Pharmacia Biotech) withSSCP electrode buffer A. PCR products harboringdifferent SSCP profiles were sequenced. Sequenceswere aligned using the SequencherTM software (GeneCodes Corporation).

SNP genotyping. Unrelated animals from threecommon breeds were genotyped: Charolais (40individuals) and Limousin (40 individuals) are twoFrench beef breeds and Holstein (77 individuals) is adairy breed. Genotyping of SNPs at positions 1428,3451, 3718, 3869, 4649, and 5851 was performed by aTaqMan� 5¢ allelic discrimination assay (AppliedBiosystems). The primers and TaqMan probe se-quences were designed using the assay-by-designservice from Applied Biosystems and the probes

were labeled with the fluorescent dyes VIC andFAM, respectively. PCR was carried out followingmanufacturer�s instructions with 5 ng of genomicDNA in a total reaction volume of 20 ll using thefollowing amplification protocol: denaturation at95�C for 10 min, followed by 40 cycles of denatur-ation at 92�C for 15 sec, and annealing and extensionat 60�C for 1 min. After PCR, the genotype of eachsample was automatically attributed by measuringthe allele-specific fluorescence on the ABI Prism7900 Sequence Detection System, using the SDS 2.1software for allele discrimination (Applied Biosys-tems).

Amplifluor assay (Pickering et al. 2002) was usedto genotype SNP at position 4695. Sequence primerswere designed using the Amplifluor Assay Architectsoftware (http://www.assayarchitect.com). Briefly,the primer sequences were [Tail-1-FAM]-5¢-GAAGGT GAC CAA GTT CAT GCT GGC CGT GGTGCT GGA AAC (allele-specific forward primer 1),[Tail-2-JOE]-5¢-GAA GGT CGG AGT CAA CGGATT CTT GGC CGT GGT GCT GGA AAT (allele-specific forward primer 2) and 5¢-CCG GTC CACAAA TAT GTC CA (reverse primer). Genotypingwas performed as recommended by the manufacturer(Serologicals Corp.). The Amplifluor reagent systemincludes the two Universal Amplifluor primers [la-beled with the fluorophores FAM (fluorescein) andJOE (6-carboxy-4¢,5¢-dichloro-2¢,7¢-dimethoxyfluo-rescein)], 10· PCR buffer, and deoxynucleoside tri-phosphates]. PCR was carried out in a total reactionvolume of 10 ll. Amplification protocol comprisedtwo cycling units: 96�C, 10 min; (95�C, 15 sec;56�64�C, 10 sec; 72�C, 15 sec) · 20 cycles; (95�C, 15sec; 56�C, 30 sec; 72�C, 40 sec) · 22 cycles; 72�C, 3min, and the resulting fluorescence signal was scoredon an ABI Prism 7900HT Sequence Detection System(Applied Biosystems).

mRNA quantification of alternative transcriptsusing real-time PCR. Total RNA was isolated frommuscle samples using RNeasy maxi kit (Qiagen).One microgram of RNA was reverse transcribedusing the High-Capacity cDNA Archive kit (AppliedBiosystems). Quantification of PRKAG3 cDNA frombovine tissues was performed using real-time PCRwith MGB gene expression assay-by-design probes(Applied Biosystems) and TaqMan chemistry on theABI Prism 7900HT Sequence Detection System(Applied Biosystems) in 20-ll reactions following themanufacturer�s instructions. The primers and probeswere as follows: 5¢�3¢: GCGCACTCGCTCATGGA(forward primer), CGGGTCCCCCAAAGCT (re-verse primer), FAM-CACTGTGTGGGACCCC(probe) for the short transcript and TGGAGCACG

M. ROUX ET AL.: TRANSCRIPTS AND POLYMORPHISMS OF THE BOVINE PRKAG3 GENE 85

CACTGTGT (forward primer), CGGGTCCCCCAAAGCT (reverse primer), FAM-CTCTTCTCCACACAGACCC (probe) for the long transcript. Allsamples were analyzed in triplicates, and the datawere normalized using the 18S ribosomal gene as aninternal control.

Results

Characterization of the bovine PRKAG3 gene. Bo-vine ESTs corresponding to the PRKAG3 gene wereidentified with BLAST searches using the humanand pig cDNA sequences (GenBank accession Nos.AW427435, BI775360, AW314499, BF601364,BF604927) and used to design primers. Overlappinggenomic fragments covering the gene from exon 1 to3¢ untranslated region (UTR) were amplified fromtotal bovine genomic DNA. cDNA sequences wereobtained from muscle cDNA and used to determinethe exon�intron organization. The 3¢-UTR sequencewas determined by RACE PCR. The promoter regionwas subcloned and sequenced from a bovine BACclone containing the gene. The bovine PRKAG3 genehas a total length of 8048 bp including part of thepromoter region (Fig. 1A). It is organized into 13coding exons and one 3¢ untranslated exon. The en-coded protein displays 84%, 83%, and 81% sequenceidentity with the human, pig, and mouse PRKAG3protein, respectively.

Screening for polymorphisms. The PRKAG3gene was screened for polymorphisms by using geno-mic DNA from three bovine breeds: Limousin, Cha-rolais, and Holstein. PCR fragments covering most ofthe gene were analyzed using the SSCP (single strandconformation polymorphism) technique, and frag-ments harboring various electrophoresis profiles werethen sequenced. A total of 32 single nucleotide poly-morphisms (SNPs) were identified, among which 13were in the coding region, one was in the 3¢ UTR, and18 were in the introns (Fig. 1A). Five of these poly-morphisms induce a change in the amino acid se-quence of the PRKAG3 protein, two of them (A121Sand W147S) in the N terminal part of the protein, two(R256W and D351Y) in the interval between CBS do-mains (CBS1 and 2, CBS2 and 3, respectively), and one(T366M) inside the CBS3 domain (Fig. 1B). All fivemutations induce a change in the amino acid subclass(hydrophilic, hydrophobic, acidic, or basic). We havedeveloped genotyping tests for 12 SNPs among whichare the five responsible for an amino acid change. TheSNPs 1428 A/G, 3451 C/T, 3718 C/T, 3869 G/C, 4649G/T, and 5851 G/T were genotyped using the TaqMansystem (Applied Biosystems), 4695 C/T using theAmplifluor technology (Chemicon International).Five closely linked SNPs (2260 G/T, 2339 G/C, 2343G/A, 2547 T/G, and 2643 T/C) were genotyped bysequencing. Allelic frequencies were determined in apopulation including three common French breeds,Charolais, Limousin, and Holstein. All SNPs were not

C/G

-756

C/T

-801

G/T

-254

7

G/T

-464

9

T/C

-264

3T

/C-3

315

G/A

-352

8A

/G-3

631

A/G

-142

8

A/G

-366

7G

/A-3

753

A/G

-380

8

G/A

-395

4

G/T

-226

0

G/C

-233

9G

/A-2

343

C/T

3451

C/T

-371

8

G/C

-386

9

G/C

-479

4G

/A-4

850

G/A

-565

9

G/A

-678

1

T/C

-571

7G

/T-5

851

C/T

-449

6

C/T

-458

5C

/T-4

695

T/C

-660

5

C/T

-473

8

A/G

-660

8

T/C

-666

8

* * * * * A

B

1 2 3 4 5 6 7 8 9 10 11 12 13 14

36 40 156 404 82 59 46 55 127 166 38 147 232 >793bpATG TGA

CBS1 CBS2 CBS3 CBS4 * * * * *

NH2 COOH

DQ082736: 8048 bp

Fig. 1. Genomic structure and polymorphisms of the bovine PRKAG3 gene. (A) Boxes indicate exons and lines connectingthe boxes indicate introns. The length of exons (bp) are shown under the gene. SNPs are indicated according to theirposition in the sequence (DQ082736). (B) Structure of the protein encoded by the PRKAG3 gene. * = SNPs responsible foran amino acid substitution and position on the protein structure.


genotyped in all individuals. Allelic frequencies for 21SNPs that were genotyped on 60 or more individualsare listed in Table 1 (in addition to the 12 SNPs pre-viously cited, the others were genotyped by directsequencing of PCR products). The first strikingobservation is that most mutant alleles have a lowallelic frequency, independent of the breed. Only se-ven SNPs have a mutant allele with a frequencygreater than 0.2 in at least two breeds. Some mutantalleles are encountered in one breed only. This isparticularly the case for SNPs inducing an amino acidsubstitution, the mutation D351Y is found only inCharolais and T366M only in Limousin and at a verylow frequency (0.02). However, because of the loca-tion of the mutation, inside the CBS3 domain, andbecause it changes a hydrophilic to a hydrophobicamino acid, this mutation may influence the confor-mation and the activity of the protein. The only aminoacid substitution with a relatively high allelic fre-quency (>0.2), A121S, was previously identified byMcKay et al. (2003).

Haplotype analysis. Twelve SNPs for which thegenotypes were available for animals from the threebreeds were used for haplotype determination using theprogram PHASE v2.1.1 (Stephens et al. 2001). Nineteendifferent haplotypes (Fig. 2A) were identified amongwhich were a very common one (frequency = 0.54), asecond common one (frequency = 0.17), and a third

common one (frequency = 0.07) and four haplotypeswere encountered 9�13 times, and 12 haplotypes wereencountered at least once. The common haplotypes arethe same in the three breeds considered. Three groupscan be made from the three most common alleles, withseveral rare alleles differing from them by one, two, orthree mutations. Most rare alleles seem to derive fromthe common one, suggesting that mutations occurredindependently and were then more or less spread.Among the seven haplotypes that are observed onlyonce, the phase is confirmed for six of them becausethey are found together with a frequent or highly fre-quent haplotype. If we consider only the 5 amino acidsubstitutions for the haplotype determination (Fig. 2B),two groups are observed as derived from the first andthe second common haplotypes.

Alternative splicing of the bovine PRKAG3gene. In the process of cloning and sequencing thePRKAG3 cDNA from skeletal muscle total RNA todetermine the exon�intron organization of the gene,we found evidence of the presence of two sites ofalternative splicing, at the 5¢ end of exons 2 and 10(Fig. 3). We observed transcripts with an exon 2 har-boring 18 additional nucleotides coming from the endof intron 1 and transcripts with an exon 10 showingthree additional nucleotides from the end of intron 9.A closer examination of the intron sequence showsthe presence of an alternative acceptor site (with the

Table 1. Allelic frequencies of the 21 SNPs that were genotyped on 60 or more individuals

E (exon) or I (intron)Mutant allele frequency

SNPanumber / nucleotideposition Amino acid change

Charolais(n = 30�40)

Limousin(n = 30�40)

Holstein(n = 77)

1428 A/G I2 / 254 0.40 0.40 0.282260 G/T E4 / 129 A121S 0.28 0.20 0.132339 G/C E4 / 208 W147S 0.02 0.04 02343 G/A E4 / 212 0.14 0.13 0.152547 G/T I4 / 12 0.27 0.20 0.132643 T/C I4 / 108 0.44 0.43 0.273451 C/T E5 / 78 0.13 0.09 0.133528 G/A I5 / 73 0 0.07 Nd3631 A/G I5 / 176 0.06 0.07 Nd3667 A/G I5 / 212 0.06 0.07 Nd3718 C/T E6 / 48 R256W 0.02 0.06 0.043753 G/A I6 / 24 0.06 0.07 Nd3808 A/G I6 / 55 0.36 0.34 Nd3869 G/C E7 / 39 0.18 0.07 0.144496 C/T I9 / 27 0.17 0 Nd4585 C/T I9 / 116 0.01 0.14 Nd4649 G/T E10 / 46 D351Y 0.17 0 04695 C/T E10 / 92 T366M 0 0.03 04738 C/T E10 / 135 0.31 0.25 Nd4794 G/C I10 / 25 0.41 0.40 Nd5851 G/T I10 / 1082 0.01 0.02 0.13aSNP names are given according to their position in the sequence DQ082736, and the common and mutant allele.Nd = not done.


consensus nucleotides AG) located 20 or 5 nucleo-tides upstream from the normal splice site for exons 2and 10, respectively (Fig. 3). Cloning and sequencingof several cDNA clones from several individuals showevidence of the use of both splice sites at both posi-tions for all samples considered. Moreover, all thecombinations were found, with transcripts contain-ing both exons in their short form, both exons in theirlong form, or alternately one or the other. This meansthat four different transcripts are produced from eachallele, potentially leading to four different proteins.

So far there is no evidence of such a mechanismfor the PRKAG3 gene in human, mouse or pig. Welooked for the presence of the alternative splice sitesin the corresponding regions for these three species(Fig. 3C). In the case of intron 1/exon 2, the AGdinucleotide is conserved at the same position inhuman and pig. In mouse, the AG dinucleotide ispresent but there is a 2-nucleotide gap that wouldcreate a frameshift in the case of the utilization ofthis splice site, whereas, in human, the splice site isalso present and the 18 nucleotides might be insertedin the reading frame. We looked for the presence inEST databases of eventual transcripts with the samestructure of the alternative transcript we observed incattle, i.e., containing exon 1 plus the additional 18bp of intron 1, but we could not find any in human,mouse, or pig. The 18 bp are fully conserved betweenpig and cattle, but the three nucleotides situatedupstream of the putative splice site are different,which might prevent the splicing phenomenon. In-deed, no alternative transcript was detected by RT-PCR using pig skeletal muscle cDNA (data notshown). In the case of intron 9/exon 10, the AGnucleotide is found in cattle only (Fig. 3C).

The consequence of using these alternativesplice sites on the protein is the presence of sixadditional amino acids in the N-terminal region andone amino acid in the interval between CBS2 andCBS3 domains (Fig. 3B). The function of the N-ter-minal region is unknown; it does not harbor anyknown protein motif and it varies a lot in size andsequence between the different c isoforms, suggest-ing that it may confer some isoform specificity. Atleast part of it should be involved in the formation ofthe AMPK complex by binding the b subunit.

Quantitative expression of the alternativetranscripts. The relative expression level of thelong and short transcripts (with or without the last18 nucleotides from the end of intron 1, respec-tively) was measured using semiquantitative RT-PCR on skeletal muscle (longissimus thoracis)from ten different individuals harboring severalhaplotype combinations (Fig. 4). A three- to sev-enfold higher expression level is observed for theshort transcript compared with the long transcript.In all cases, the short transcript shows a signifi-cantly higher expression level compared with thelong transcript, even though global expressionlevels vary from one individual to another. Wetried to develop a semiquantitative PCR method tomeasure the relative amount of trancripts usingalternative splice sites at the end of intron 9, butwe failed to obtain specific signals using TaqManprobes (data not shown).

A Haplotypes (n / 296 Chr)

A G G G G T C C G G C G 159 A G G G G T C C G T C G 13

A G G G G T C C C G C G 2# A G G G G T C T G G C G 11 A G G G G T T C G G C G 1

A G G G G C C C G G C G 3 A G C G G T C C G G C G 3

G G G G G T C C G G C G 1 G G G G G C C C G G C G 9

A G G A G C C C G G C G 1 G G G A G T C C G G C G 1

A T C G T T C C G G C G 1* A T G G T C C C G G C G 1 A G G A G C C C G T C G 1

G T G G T C C C G G C G 51 G T G G T T C C G G C G 2

G G G A G C T C C G C T 22 G G G A G C T C C G C G 12 G G G A G C T C C G T T 2

B Haplotypes (n / 296 Chr)

G G C G C 210 G C C G C 3 G G C T 14 C G G C G T 2 G G T G C 11

T G C G C 55 T C C G C 1*

Fig. 2. Haplotypes description and frequency as deter-mined by the program PHASE v2.1.1 using the genotypesfrom 148 individuals (77 Hoslteins, 35 Limousins, and 36Charolais). (A) The considered SNPs are 1428 A/G, 2260G/T, 2339G/C, 2343 G/A, 2547 G/T, 2643 T/C, 3451 C/T,3718 C/T, 3869 G/C, 4649 G/T, 4695 C/T, and 5851 G/T.(B) The considered SNPs are 2260G/T, 2339 G/C, 3718 C/T, 4649 G/T, and 4695 C/T. Major haplotypes are boxed.For each group the nucleotides differing from the majorhaplotypes are in bold/gray (# = one individual is homo-zygous for this haplotype, * = phase not sure).


Discussion

We determined the genomic sequence and structureof the bovine PRKAG3 gene. This structure is thesame as the human PRKAG3 gene but differs fromthe mouse and pig genes, which do not have an in-tron in the 3¢ UTR (Amarger et al. 2003). Thrity-twoSNPs were identified on a total of 8048 bp, whichmeans one SNP every 250 bp on average. In the hu-man genome, SNPs are now estimated to occur every300 bp on average (Nelson et al. 2004), but this fre-quency is highly variable among regions and codingregions are particularly stable, most probably be-cause of the strong selection pressure to which theyare submitted. In cattle, several genes have beenscreened for SNPs and they reveal a comparable oreven higher variation frequency (Konfortov et al.1999; Heaton et al. 2001). Some parts of them can beconsidered hypermutable regions. However, SNPsearches on a large set of genes in cattle reveals thatvariation frequency varies a lot from one gene to

another (V. Amarger, unpublished), suggesting thatthe selection pressure is different between genesand/or regions of the genome.

Several mutant alleles we identified have a lowallelic frequency (<1%) in the population we tested.This is particularly true for the SNPs that induce anamino acid substitution. This is not surprising andthe same phenomenon was observed in human geneswhere SNPs in coding regions, referred to as cSNPS,and particularly nonsynonymous cSNPs, occur lessoften but also have lower minor allele frequency(Cargill et al. 1999; Halushka et al. 1999). This sug-gests a strong selection pressure against amino acid-altering changes, especially deleterious alleles.Among the amino acid substitutions, two are ob-served in one breed only, D351Y in Charolais andT366M in Limousin, the latter at a very low fre-quency, and one (W147S) only in Limousin andCharolais. The fact that these mutant alleles are veryrare suggests that they might be recent mutationsthat have not been spread so far. Moreover, for these

1

2

3 4 5 6 7 8 9

10

10

11 12 13 14

2

--TPSWSS-- --RTLLPR--

--QRTLLPR----SLFSTQTPSWSS--

mRNA

amino-acid sequence

amino-acid sequence

CBS1 CBS2 CBS3 CBS4 NH2 COOH

A

B

end intron 1 start exon 2

Bovine tgttctccccacctcttcag tca ctc ttc tcc aca cag ACC CCC TCC TGG AGC APig cgtcttccccaccccaaaag tca ctc ttc tcc aca cag ACT CCC TCC TGG AGC AHuman cactttccccactccttcag aaa ctc ttc tcc cca cag ACC CCT TCC TGG AGC AMouse tagcattttcacttc---ag aca ttc tct --c cca aag ACC CTG ACC TGG AGC C

end intron 9 start exon 10

Bovine ttcaacccaagcag CGC ACC CTG CTGPig cccaacccaaccag GGC ACC CTG CTGHuman ccctaaccatccag GGT TCC CTG CTGMouse cccatcctaa-cag GGT GCC CTG TTG

C

Fig. 3. Alternative splice sites in the bovine PRKAG3 gene. (A) The structure of the mRNA is represented with twodifferent splice sites at the 5¢ end of exons 2 and 10. The two possible amino acid sequences resulting from the translationof this part of the mRNA are shown above and below the mRNA. (B) Structure of the protein encoded by the PRKAG3gene. Arrows indicate the position of the modifications in the amino acid sequence as a result of these alternative splicingevents. (C) Sequence alignment of the two regions of the PRKAG3 gene containing alternative splice sites in cattle. Intronsequences are in lower case and exon sequences are in upper case with codon position indicated. The conventional splicesites are in bold and underlined; the alternative splice sites are in bold, italic, and underlined.


three amino acid substitutions, the common allele(R256, D351, and T366) is conserved with the hu-man, mouse, and pig reference sequences. On theother hand, the mutant S121 allele is surprisinglyencountered in pig and human, suggesting that thecommon allele in cattle (A121) is, in fact, probablythe mutant one. The amino acid encountered at thecorresponding position in mouse is a proline but thisregion is not well conserved between species. For theW147S substitution, the common allele W147 isfound in human and mouse and the mutant S147 inpig, resulting from the same mutation and suggest-ing an ancestral origin. Breed-specific rare alleles canvery well be mutations emerging from a recentselection pressure on important characters, espe-cially in such breeds (Limousin and Charolais) thatare selected on the best performance for meat pro-duction traits, whereas the Holstein breed is selectedfor milk production traits. Most association studiesusing SNPs usually eliminate low-frequency vari-ants because of their poor level of informativity.However, considering the organization of the breed-ing programs in cattle, a rare allele might spreadquite fast in the population if it is found in a sireselected for reproduction. That is what happened forthe RN- mutation in pig, which probably originatesfrom a single individual (Andersson 2003). Theidentification of such mutations and the character-ization of their effect on the phenotype might thenbe helpful for the orientation of breeding programs.

The influence of the amino acid substitutions onthe activity of the protein are still to be determined.Two substitutions (R256W and D351Y) are situated

in the interval between CBS domains (CBS1 and 2and CBS2 and 3, respectively) and one (T366M) in-side the CBS3 domain in a region that is highlyconserved between cattle, human, mouse, pig, Dro-sophila, and even yeast (Snf4), all of them harboringthe T366. CBS domains invariably occur as tandempairs and are known to bind one molecule of AMP orATP in various proteins (Scott et al. 2004). The threec subunit isoforms of AMPK are unique amongeukaryotic proteins in having four tandem domains,thus forming allosteric binding sites for two mole-cules of AMP or ATP. Pathogenic mutations knownso far in CBS domains tend to occur in positionslikely to be involved in the binding of the adenosine-containing ligand; this is probably the reason fortheir strong effect (Scott et al. 2004). This might bethe case for the T366M substitution, but functionalstudies would be required to confirm or disprove thishypothesis. On the other hand, one could think thatmutations affecting other sites of the protein, forinstance, regions between CBS domains, could havea milder effect if they influence the conformation ofthe domains with respect to each other, leading to achange in the affinity of the binding domains.

Eight of 13 SNPs situated in the coding sequenceare silent mutations that affect the third codon po-sition. These mutations do not change the proteinsequence but they might influence the translationefficiency if they introduce rare codons. Indeed, co-don use in mammals is biased and correlates withthe GC content, the preferred codons being the onesending with G or C (Marin et al. 1989; Porter 1995).In our case, five SNPs that affect the third codon

short transcript

long transcript

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

mR

NA

(ar

bitr

ary

unit)

L502 L507 L508 L515 L516 L521 L531 C601 C603 C607

individuals(L: Limousin, C: Charolais)

Fig. 4. Expression of the long andthe short (with or without the last18 nucleotides from intron 1)transcripts assessed by real-timePCR analysis. The values arehomogenized to an endogenouscontrol (18S ribosomal RNA) andexpressed as means ± SEM. Theresults were compiled from threedifferent experiments for tenanimals.


position (2343 G/A, 4738 C/T, 6605 T/C, 6608 A/Gand 6668 T/C) lead to the mutation of a commoncodon to a rare codon or vice versa. Further studieswould be required to establish if these mutationsshow any correlation with the translation efficiencyof the gene.

We observed alternative splicing events poten-tially introducing some heterogeneity in the popu-lation of AMPK c3 subunit population. This was notobserved in any species before and its functionnalsignificance in cattle remains to be determined.Quantitative analysis of the expression of the shortand long transcripts harboring or not the last 18nucleotides from intron 1 reveals that the shorttranscript is about three to seven times more ex-pressed than the long transcript. This might reveal adifference in the affinity of the splicing machineryfor one of the two splicing sites or in the stability ofthe mRNAs. Alternative splicing is a way used byeukaryotic genomes to increase diversity at theprotein level without increasing the number of genesand is now recognized as a regulatory process, con-tributing to biological complexity (Lareau et al.2004). The AMPK system is already extremely di-verse and heterogeneous because of the existence ofseveral isoforms which expression varies from onetissue to another. The reasons for the existence ofsuch a high diversity are not known but they arelikely to be linked to the large spectrum of actionsand targets on which the AMPK is acting. One canassume that because of the vital importance of theAMPK system in the maintenance of balance inenergy metabolism, diversity at the gene and proteinlevels is assumed to avoid any kind of ‘‘accident’’that might occur and infer with the protein activity.The existence of new sites of alternative splicing inthe bovine PRKAG3 gene might be a response to ademand for additional diversity of the AMPK systemto adapt to the energy metabolism that is specific inruminants.

In conclusion, we found evidence of a largevariability in the PRKAG3 gene in cattle. Associa-tion studies between polymorphisms and pheno-typic traits related to meat quality are now neededin order to establish if any allele influences AMPKactivity in muscle and, therefore, meat characteris-tics. Indeed, AMPK has been shown to stronglyinfluence muscle glycogen content and glycolysis inpostmortem muscle (Du et al. 2005). In addition tothis high level of polymorphism, we show that thecomplexity of the AMPK system is increased incattle by the presence of alternatively spliced tran-scripts of the PRKAG3 gene that are susceptible tocreate a subfamily of isoforms of the AMPK c sub-unit.

Acknowledgments

The authors thank all the people who provided DNAand tissue samples. The Charolais and Limousinindividuals were collected as a part of the GemqualEuropean Project (QLRT-CT2000-00147). MatthieuRoux was supported by a INRA/Region Limousingrant.

References

1. Amarger V, Erlandsson R, Pielberg G, Jeon JT, An-dersson L (2003) Comparative sequence analysis of thePRKAG3 region between human and pig: evolution ofrepetitive sequences and potential new exons. Cyto-genet Genome Res 102, 163�172

2. Andersson L (2003) Identification and characterizationof AMPK gamma3 mutations in the pig. Biochem SocTrans 31, 232�235

3. Andersson L, Georges M (2004) Domestic-animal ge-nomics: deciphering the genetics of complex traits.Nat Rev Genet 5, 202�212

4. Arad M, Benson DW, Perez-Atayde AR, McKenna WJ,Sparks EA, et al. (2002) Constitutively active AMPkinase mutations cause glycogen storage diseasemimicking hypertrophic cardiomyopathy. J Clin In-vest 109: 357�362

5. Barnes BR, Marklund S, Steiler TL, Walter M, HjalmG, et al. (2004) The 5¢-AMP-activated protein kinasegamma3 isoform has a key role in carbohydrate andlipid metabolism in glycolytic skeletal muscle. J BiolChem 279, 38441�38447

6. Blair E, Redwood C, Ashrafian H, Oliveira M, Brox-holme J, et al. (2001) Mutations in the gamma(2) sub-unit of AMP-activated protein kinase cause familialhypertrophic cardiomyopathy: evidence for the centralrole of energy compromise in disease pathogenesis.Hum Mol Genet 10, 1215�1220

7. Cargill M, Altshuler D, Ireland J, Sklar P, Ardlie K,et al. (1999) Characterization of single-nucleotidepolymorphisms in coding regions of human genes. NatGenet 22, 231�238

8. Carling D (2004) The AMP-activated protein kinasecascade—a unifying system for energy control. TrendsBiochem Sci 29, 18�24

9. Cheung PC, Salt IP, Davies SP, Hardie DG, Carling D(2000) Characterization of AMP-activated protein ki-nase gamma-subunit isoforms and their role in AMPbinding. Biochem J 346(Pt 3), 659�669

10. Ciobanu D, Bastiaansen J, Malek M, Helm J, WoollardJ, et al. (2001) Evidence for new alleles in the proteinkinase adenosine monophosphate-activated gamma(3)-subunit gene associated with low glycogen content inpig skeletal muscle and improved meat quality.Genetics 159, 1151�1162

11. Du M, Shen QW, Zhu MJ (2005) Role of beta-adreno-ceptor signaling and AMP-activated protein kinase inglycolysis of postmortem skeletal muscle. J Agric FoodChem 53, 3235�3239


12. Eggen A, Gautier M, Billaut A, Petit E, Hayes H, et al.(2001) Construction and characterization of a bovineBAC library with four genome-equivalent coverage.Genet Sel Evol 33, 543�548

13. Ferre P, Azzout-Marniche D, Foufelle F (2003) AMP-activated protein kinase and hepatic genes involved inglucose metabolism. Biochem Soc Trans 31: 220�223

14. Gollob MH, Green MS, Tang AS, Gollob T, Karibe A,et al. (2001) Identification of a gene responsible forfamilial Wolff-Parkinson-White syndrome. N Engl JMed 344, 1823�1831

15. Halushka MK, Fan JB, Bentley K, Hsie L, Shen N, et al.(1999) Patterns of single-nucleotide polymorphisms incandidate genes for blood-pressure homeostasis. NatGenet 22, 239�247

16. Hardie DG, Scott JW, Pan DA, Hudson ER (2003)Management of cellular energy by the AMP-activatedprotein kinase system. FEBS Lett 546, 113�120

17. Heaton MP, Grosse WM, Kappes SM, Keele JW, Chit-ko-McKown CG, et al. (2001) Estimation of DNA se-quence diversity in bovine cytokine genes. MammGenome 12: 32�37

18. Immonen K, Ruusunen M, Hissa K, Puolanne E (2000)Bovine muscle glycogen concentration in relation tofinishing diet, slaughter and ultimate pH. Meat Sci 55,25�31

19. Kemp BE, Mitchelhill KI, Stapleton D, Michell BJ,Chen ZP, et al. (1999) Dealing with energy demand:the AMP-activated protein kinase. Trends Biochem Sci24, 22�25

20. Konfortov BA, Licence VE, Miller JR (1999) Re-sequencing of DNA from a diverse panel of cattle re-veals a high level of polymorphism in both intron andexon. Mamm Genome 10, 1142�1145

21. Lareau LF, Green RE, Bhatnagar RS, Brenner SE (2004)The evolving roles of alternative splicing. Curr OpinStruct Biol 14, 273�282

22. Marin A, Bertranpetit J, Oliver JL, Medina JR (1989)Variation in G + C-content and codon choice: differ-ences among synonymous codon groups in vertebrategenes. Nucleic Acids Res 17, 6181�6189

23. McKay SD, White SN, Kata SR, Loan R, Womack JE(2003) The bovine 5¢ AMPK gene family: mapping andsingle nucleotide polymorphism detection. MammGenome 14, 853�858

24. Milan D, Jeon JT, Looft C, Amarger V, Robic A, et al.(2000) A mutation in PRKAG3 associated with excessglycogen content in pig skeletal muscle. Science 288,1248�1251

25. Nelson MR, Marnellos G, Kammerer S, Hoyal CR, ShiMM, et al. (2004) Large-scale validation of singlenucleotide polymorphisms in gene regions. GenomeRes 14, 1664�1668

26. Pickering J, Bamford A, Godbole V, Briggs J, ScozzafavaG, et al. (2002) Integration of DNA ligation and rollingcircle amplification for the homogeneous, end-pointdetection of single nucleotide polymorphisms. Nu-cleic Acids Res 30, e60

27. Porter TD (1995) Correlation between codon usage,regional genomic nucleotide composition, and aminoacid composition in the cytochrome P-450 genesuperfamily. Biochim Biophys Acta 1261, 394�400

28. Rouzaud F, Martin J, Gallet PF, Delourme D, (2000) Afirst genotyping assay of French cattle breeds based ona new allele of the extension gene encoding the mel-anocortin-1 receptor (Mc1r). Genet Sel Evol 32, 511�520

29. Scott JW, Hawley SA, Green KA, Anis M, Stewart G,et al. (2004) CBS domains form energy-sensing mod-ules whose binding of adenosine ligands is disruptedby disease mutations. J Clin Invest 113, 274�284

30. Stapleton D, Mitchelhill KI, Gao G, Widmer J, MichellBJ, et al. (1996) Mammalian AMP-activated proteinkinase subfamily. J Biol Chem 271, 611�614

31. Stephens JC, Schneider JA, Tanguay DA, Choi J,Acharya T, et al. (2001) Haplotype variation and link-age disequilibrium in 313 human genes. Science 293,489�493

32. Thornton C, Snowden MA, Carling D (1998) Identifi-cation of a novel AMP-activated protein kinase betasubunit isoform that is highly expressed in skeletalmuscle. J Biol Chem 273, 12443�12450


Documents

Characterization of the bovine PRKAG3 gene: structure, polymorphism, and alternative transcripts