conflicting candidates for cattle qtls

5
8 Mitelman, F. et al. (2004) Fusion genes and rearranged genes as a linear function of chromosome aberrations in cancer. Nat. Genet. 36, 331–334 9 Sankoff, D. et al. (2002) Chromosomal distributions of break- points in cancer, infertility, and evolution. Theor. Popul. Biol. 61, 497–501 10 Schwartz, M. et al. (2006) The molecular basis of common and rare fragile sites. Cancer Lett. 232, 13–26 11 Samonte, R.V. and Eichler, E.E. (2002) Segmental duplications and the evolution of the primate genome. Nat. Rev. Genet. 3, 65–72 12 Armengol, L. et al. (2003) Enrichment of segmental duplications in regions of breaks of synteny between the human and mouse genomes suggest their involvement in evolutionary rearrangements. Hum. Mol. Genet. 12, 2201–2208 13 Warburton, P.E. (2004) Chromosomal dynamics of human neocen- tromere formation. Chromosome Res. 12, 617–626 14 Amor, D.J. and Choo, K.H. (2002) Neocentromeres: role in human disease, evolution, and centromere study. Am. J. Hum. Genet. 71, 695–714 15 Bourque, G. et al. (2005) Comparative architectures of mammalian and chicken genomes reveal highly variable rates of genomic rearrangements across different lineages. Genome Res. 15, 98–110 16 Froenicke, L. et al. (2006) Are molecular cytogenetics and bioinfor- matics suggesting contradictory models of ancestral mammalian genomes? Genome Res. 16, 306–310 17 Glover, T.W. (2006) Common fragile sites. Cancer Lett. 232, 4–12 18 Chowdhary, B.P. et al. (1998) Emerging patterns of comparative genome organization in some mammalian species as revealed by Zoo- FISH. Genome Res 8, 577–589 0168-9525/$ - see front matter Q 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.tig.2006.04.002 Conflicting candidates for cattle QTLs Dirk-Jan de Koning Genetics and Genomics, The Roslin Institute, Roslin, Midlothian, UK, EH25 9PS Genome scans have identified quantitative trait loci (QTLs) affecting milk yield and composition in dairy cattle. For one QTL on bovine chromosome 6 (BTA6), previously fine-mapped to a 420-Kb region, mutations in two different genes (OPN and ABCG2) have been proposed as the underlying functional mutation. Comparing the arguments for each gene suggests that both mutations are equally probable. However, functional studies and/or additional populations are required to provide a definite answer. Dissection of complex traits in livestock Most genetic research in humans focuses on the genetic aspects of well-being, whereas genetic research in live- stock has been driven by an interest to increase the efficiency of food production with the aim to use functional genetic polymorphisms in marker assisted selection (MAS). A commonality between human and livestock genetic research is that the identification of point mutations underlying variation in complex traits remains a formidable task. This observation is underlined by the few functional mutations that have been identified compared with the many quantitative trait loci (QTLs) for complex traits that have been described in livestock and humans. In livestock, the successful identification of functional mutations underlying a QTL is limited to a few genes including insulin-like growth factor 2 (IGF2) in pigs (Sus scrofa) [2], Callipyge (CLPG) in sheep (Ovis aries) [3] and DGAT1 in cattle (Bos taurus), which encodes acyl-CoA diacylglycerol acyltransferase 1 [4]. Constructing transgenic models to demonstrate the functionality of candidate mutations is not an option for most livestock species. Hence evidence for functional loci needs to be provided by combining multiple sources of evidence that individually would not be convincing but jointly point to a single candidate mutation [5]. For a QTL on bovine chromosome 6, which affects milk yield, similar lines of investigation have lead to distinct conclusions regarding the candidate genes (OPN and ABCG2) [1,6] and here I compare the evidence that has been put forward for both candidates. Dairy cattle QTL and chromosome 6 Although livestock offers the opportunity of controlled breeding, it is often too expensive and time-consuming, particularly in cattle, to breed experimental populations such as crosses between extreme lines. However, because of large-scale recording of traits and the extensive use of artificial insemination (AI) with carefully controlled Glossary Linkage disequilibrium: describes a situation in which some combinations of alleles or genetic markers occur more or less frequently in a population than would be expected from a random formation of haplotypes from alleles based on their frequencies Marker assisted selection: describes the process where DNA polymorphisms are used to improve the prediction of genetic merit in animal and plant breeding. Quantitative trait locus: genetic loci or chromosomal regions that contribute to variability in complex quantitative traits, as identified by statistical analysis. Quantitative traits are typically affected by several genes and by the environment. Statistical power: a statistic that describes how effective a given experiment is to detect a certain effect. Statistical power is expressed as the proportion of tests that are expected to be significant given a certain experiment and a certain effect. Corresponding author: de Koning, D.-J. ([email protected]). Update TRENDS in Genetics Vol.22 No.6 June 2006 301 www.sciencedirect.com

Upload: dirk-jan-de-koning

Post on 13-Sep-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Conflicting candidates for cattle QTLs

Update TRENDS in Genetics Vol.22 No.6 June 2006 301

8 Mitelman, F. et al. (2004) Fusion genes and rearranged genes as alinear function of chromosome aberrations in cancer. Nat. Genet. 36,331–334

9 Sankoff, D. et al. (2002) Chromosomal distributions of break-points in cancer, infertility, and evolution. Theor. Popul. Biol. 61,497–501

10 Schwartz, M. et al. (2006) The molecular basis of common and rarefragile sites. Cancer Lett. 232, 13–26

11 Samonte, R.V. and Eichler, E.E. (2002) Segmental duplicationsand the evolution of the primate genome. Nat. Rev. Genet. 3,65–72

12 Armengol, L. et al. (2003) Enrichment of segmental duplications inregions of breaks of synteny between the human and mouse genomessuggest their involvement in evolutionary rearrangements. Hum.Mol. Genet. 12, 2201–2208

13 Warburton, P.E. (2004) Chromosomal dynamics of human neocen-tromere formation. Chromosome Res. 12, 617–626

Corresponding author: de Koning, D.-J. ([email protected]).

www.sciencedirect.com

14 Amor, D.J. and Choo, K.H. (2002) Neocentromeres: role in humandisease, evolution, and centromere study. Am. J. Hum. Genet. 71,695–714

15 Bourque, G. et al. (2005) Comparative architectures of mammalianand chicken genomes reveal highly variable rates of genomicrearrangements across different lineages. Genome Res. 15, 98–110

16 Froenicke, L. et al. (2006) Are molecular cytogenetics and bioinfor-matics suggesting contradictory models of ancestral mammaliangenomes? Genome Res. 16, 306–310

17 Glover, T.W. (2006) Common fragile sites. Cancer Lett. 232, 4–1218 Chowdhary, B.P. et al. (1998) Emerging patterns of comparative

genome organization in some mammalian species as revealed by Zoo-FISH. Genome Res 8, 577–589

0168-9525/$ - see front matter Q 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.tig.2006.04.002

Conflicting candidates for cattle QTLs

Dirk-Jan de Koning

Genetics and Genomics, The Roslin Institute, Roslin, Midlothian, UK, EH25 9PS

Genome scans have identified quantitative trait loci

(QTLs) affecting milk yield and composition in dairy

cattle. For one QTL on bovine chromosome 6 (BTA6),

previously fine-mapped to a 420-Kb region, mutations in

two different genes (OPN and ABCG2) have been

proposed as the underlying functional mutation.

Comparing the arguments for each gene suggests that

both mutations are equally probable. However,

functional studies and/or additional populations are

required to provide a definite answer.

Glossary

Linkage disequilibrium: describes a situation in which some combinations of

alleles or genetic markers occur more or less frequently in a population than

would be expected from a random formation of haplotypes from alleles based

on their frequencies

Marker assisted selection: describes the process where DNA polymorphisms

are used to improve the prediction of genetic merit in animal and plant

breeding.

Quantitative trait locus: genetic loci or chromosomal regions that contribute to

variability in complex quantitative traits, as identified by statistical analysis.

Quantitative traits are typically affected by several genes and by the

environment.

Statistical power: a statistic that describes how effective a given experiment is

to detect a certain effect. Statistical power is expressed as the proportion of

Dissection of complex traits in livestock

Most genetic research in humans focuses on the geneticaspects of well-being, whereas genetic research in live-stock has been driven by an interest to increase theefficiency of food production with the aim to use functionalgenetic polymorphisms in marker assisted selection(MAS). A commonality between human and livestockgenetic research is that the identification of pointmutations underlying variation in complex traits remainsa formidable task. This observation is underlined by thefew functional mutations that have been identifiedcompared with the many quantitative trait loci (QTLs)for complex traits that have been described in livestockand humans. In livestock, the successful identification offunctional mutations underlying a QTL is limited to a fewgenes including insulin-like growth factor 2 (IGF2) in pigs(Sus scrofa) [2], Callipyge (CLPG) in sheep (Ovis aries) [3]and DGAT1 in cattle (Bos taurus), which encodes acyl-CoAdiacylglycerol acyltransferase 1 [4]. Constructing

transgenic models to demonstrate the functionality ofcandidate mutations is not an option for most livestockspecies. Hence evidence for functional loci needs to beprovided by combining multiple sources of evidence thatindividually would not be convincing but jointly point to asingle candidate mutation [5]. For a QTL on bovinechromosome 6, which affects milk yield, similar lines ofinvestigation have lead to distinct conclusions regardingthe candidate genes (OPN and ABCG2) [1,6] and here Icompare the evidence that has been put forward forboth candidates.

Dairy cattle QTL and chromosome 6

Although livestock offers the opportunity of controlledbreeding, it is often too expensive and time-consuming,particularly in cattle, to breed experimental populationssuch as crosses between extreme lines. However, becauseof large-scale recording of traits and the extensive use ofartificial insemination (AI) with carefully controlled

tests that are expected to be significant given a certain experiment and a certain

effect.

Page 2: Conflicting candidates for cattle QTLs

Update TRENDS in Genetics Vol.22 No.6 June 2006302

pedigrees in dairy cattle breeding worldwide, QTLanalysis can be performed using existing family structureswithin the population. The methods that have beenemployed for detection and fine-mapping of QTLs indairy cattle (and other livestock) are described in Box 1.

Box 1. Methods used in cattle QTL (fine) mapping

Numerous statistical methods have been developed for QTL analysis

in livestock species, each of which incorporate the basic components

of pedigree, phenotype and genotype differently. Here I give an

overview of the methods that have been applied to map QTL in dairy

cattle populations. Each QTL mapping approach comes in many

different ‘flavours’ and the aim of this list is to outline the principles

underlying the methodology rather than provide a

detailed description.

Half-sib analyses

These approaches exploit the paternal half-sib structure that exists

because of extensive use of AI. In so called ‘daughter designs’,

marker genotypes aremeasured on AI bulls and large groups of their

daughters, whereas production traits aremeasured on the daughters

[18]. In ‘granddaughter designs’ the genotypes are gathered on older

AI bulls and moderate groups of their sons (50–200), who are also AI

bulls. The information on production traits is provided by potentially

large number of daughters from the sons (100–10 000) [18]. These

analyses only model a paternal QTL component and ignore

information from the maternal alleles or any genetic links between

half-sib families.

General pedigree linkage analysis

Methods in this category use all genetic links that are provided in the

pedigree and follow the principle that individuals that are identical

by descent (IBD) for a relevant chromosome segment and will also

be more similar for traits that are affected by that chromosome

segment [19]. These methods are applicable to any pedigree with

sufficient genetic links and are acclaimed to have greater statistical

power and precision than family-based methods.

Combined linkage and linkage disequilibrium analysis

(LDLA)Assuming that the QTL effect is a result of a mutation in a single

ancestor of the current population, individuals that carry the

mutation can be expected to share alleles identical by descent

(IBD) at, and around, the mutation. The LDLA analysis follows the

same principle as the general pedigree linkage analysis with the

addition of IBD estimates between founder animals that are not

related through the provided pedigree but are assumed to be related

through a common, unknown, ancestor in whom the mutation arose

[20]. These estimates are based on haplotypes in the founder

animals of the experimental population and assumption about

population history such as effective population size and time

since mutation.

Haplotype analysisFrom the QTL analyses, the QTL genotypes of the (grand)parents

of the QTL population can be inferred as heterozygous (Qq)

and homozygous (QQ or qq). The holy grail of fine mapping is to

find a haplotype or preferably a single mutation for which the

genotypes in the QTL parents are consistent with the inferred QTL

genotypes. This type of analysis was first demonstrated in the

fine mapping of a QTL on BTA14 [21] and is often complementary

to LDLA analysis.

Association studyIf the haplotype analysis is successful in identifying a conserved

haplotype surrounding the functional mutation or proposing a

candidate mutation, the effect of the haplotype or mutation can be

evaluated at the population level without the need for extensive

family structures. This type of analysis is also referred to as linkage

disequilibrium mapping (LD mapping).

www.sciencedirect.com

Bovine chromosome 6 (BTA6) has received considerableearly interest from the breeding community because itharbours the casein locus, which contains four closelylinked milk protein genes. Initial studies focused onassociations between casein polymorphisms and pro-duction traits [7] and, subsequently, many QTL scans indairy cattle started with BTA6 [8,9]. A recent review ofQTL studies in dairy cattle confirmed that BTA6 is one ofthe most studied chromosomes and significant QTLs formilk production traits on this chromosome are found inseveral breeds [10]. Unlike BTA14, where many studiesidentified QTLs in the region of DGAT1 affecting mainlyfat yield and fat percentage but also other milk productiontraits, on BTA6 there is evidence for multiple linked QTLsaffecting milk production traits from analysis across [10]and within breeds [11]. One of these QTLs is located in thecentre of the chromosome and seems to have its maineffect on protein and fat content (%) of milk with a lessereffect on the yields (Kg) of fat and protein.

Fine mapping of the QTL

Using data on Norwegian dairy cattle, the QTL was finemapped to an interval of 7.5 cM using combined linkageand linkage disequilibrium mapping (LDLA, Box 1) [12].In a study of Israeli dairy cattle, a QTL with similar effectsand location was mapped with a precision of 4 cM [11],which overlapped with that of the Norwegian study. Giventhe co-location of QTL and the similar effect on productiontraits, it could be postulated that both studies detected thesame QTL. Subsequently, the Norwegian group increasedthe marker density in the QTL region by typing additionalsingle nucleotide polymorphisms (SNPs) followed byLDLA analyses [13]. The haplotype analysis (Box 1)revealed six haplotypes that had a significant effect onthe protein content of milk [13]. The most common of thesehaplotypes also had the most extreme effect. Furthercharacterization of these haplotypes using sequenceinformation, radiation hybrid mapping and comparativemapping with the human genome sequence resolved theQTL to a 420-Kb haplotype region (Figure 1) [13]. Usinginformation from the comparative map with the humangenome, only six known genes were identified within theQTL region (Figure 1) [14].

From haplotype to functional mutation

On the basis of the 420-Kb interval and the assumptionthat the same functional mutation would be segregating intheir own experimental population, two groups set out toidentify the functional mutation underlying the QTL:Schnabel et al. [6] used USA dairy cattle, whereas Cohen-Zinder et al. [1] used Israeli and USA dairy cattle. Theoverlap between the two studies with regard to the USAdairy population and the similar QTL effects and locationstrongly suggest that both studies are chasing the sameQTL (see Table 1 for a comparison between the two studies).Both groups used sequence analysis to identify geneticvariation within the critical region. The ‘holy grail’ of finemapping is to find a haplotype or preferably a singlemutation for which the genotypes in the QTL parents areconsistent with the inferred QTL genotypes. Schnabel et al.singled out OPN, which encodes osteopontin, as the prime

Page 3: Conflicting candidates for cattle QTLs

TRENDS in Genetics

.

BTA6 QTL region (x10)

ILSTS093

BMS3

INRA133

ILSTS090

BM1329BMS2508

BM143BMS382

BM3026 BMS1242BMS690

OarJMF36

BL1099

BMS518BM4322BMS470BMS483

BMS360ILSTS097

BM4528RM127

BM4621RM028BM415

1518

ILSTS035OarEL03

2220CSN1S1 CSN3

ILSTS087642 BMSB4049

ALBGCBM1236

BMS2460BM4311

BMS511 ILSTS018OarJMP4

AFR227BP7

BM8124ETH8

BMC4203 OarJMP8BMS739BM9257BM9047

BM2320BL1038

OarJMP12

FAM13A1

HERC3HERC5HERC6PPM1KABCG2

PKD2SPP1(OPN)

MEPEIBSPLAP3MED28

KIAA1276HCAP-GMLR1BM143

Figure 1. The linkage map of bovine chromosome 6 (BTA6, left) and a schematic

overview of the genes in the QTL region (green region, right). The 420-Kb region is

represented in red. Bovine linkage maps (MARC97 used in this figure) can be

viewed at http://www.thearkdb.org/browser?speciesZcow. The maps were drawn

Update TRENDS in Genetics Vol.22 No.6 June 2006 303

www.sciencedirect.com

candidate among the six genes in the 420-Kb region(Figure 1) and sequenced 12.3 Kb spanning the entireOPN gene in four heterozygous and four homozygous siresto detect SNPs that would be consistent with the QTLgenotypes. By contrast, Cohen-Zinder et al. targeted partsof ten genes in and around the 420-Kb region (Figure 1),and sequenced two heterozygous sires to identify SNPsthat were subsequently evaluated in a number of experi-mental populations (Table 1). Following these analyses,Schnabel et al. proposed that an indel (insertion anddeletion) in the poly-T tract w1240 bp upstream of theOPN transcription initiation site was the causativepolymorphism. The authors propose that the motif mightrepresent a novel regulatory region. However, Cohen-Zinder et al.proposed that a missense mutation (tyrosine toserine substitution) in the fifth extracellular region of theABCG2 protein is the causal mutation underlying the QTL.OPN is a secreted glycoprotein involved in cell-matrixinteractions, whereasABCG2 is a half-transporter that hasbeen shown to transport cytostatic drugs across the plasmamembrane. The evidence that is presented to support theinvolvement of these mutations is remarkably similar. (i)Among all the SNPs that were evaluated, the genotypes ofthe putative mutation are in unique agreement with theinferred QTL genotypes of the (grand)parents of the half-sib families that were used in each study. (ii) The mutationhas a significant association at the population level, whichimplies that the mutation either represents the functionalmutation or is in strong linkage disequlibrium (LD) withthe functional mutation. (iii) Including the mutation as afixed effect in QTL or association analysis (Box 1), accountsfor the entire QTL effect. (iv) OPN and ABCG2 are the onlytwo genes (among the six in the 420-Kb region) that aredifferentially expressed between different stages of lacta-tion in the bovine mammary gland [1]. (v) Both genes havebeen shown to have a role in mammary gland processes inmouse models. (vi) Both genes have a reasonable level ofsequence (OPN) or amino acid (ABCG2) conservationamong mammals in the region of the proposedfunctional mutation.

Will the real functional mutation please stand up?

The most notable shortcoming of both studies is that theyare incomplete. The approach taken by Schnabel et al.could identify the causative mutation if the correct gene ischosen, whereas that taken by Cohen-Zinder et al. couldlead to the identification the causative mutation, if it is aprotein altering mutation but it might miss causalmutations in regulatory regions. Both studies leave alarge area of the QTL interval unexplored. Schnabel et al.acknowledge that they looked at only a fraction of the 420-Kb interval and estimate that the probability that therewill be no other SNP with the same segregation patternwithin the 420-Kb region is w0.40. This suggests that theprobability of finding at least one additional SNP with thesame segregation pattern within the 420-Kb intervalwould be w0.60, which is fairly high and a relevantdiscussion point. Cohen-Zinder et al. state that the

using MapChart. The schematic overview of the QTL region was drawn using the

results from Cohen-Zinder et al. [1].

Page 4: Conflicting candidates for cattle QTLs

Table 1. Overview of two studies that proposed a mutation for a QTL on chromosome 6

Schnabel et al. [6] Cohen-Zinder et al. [1]

Population 3147 Holstein bulls divided over 45 half-sib families;

167 sires that represent the lines that are present in an

American cattle DNA repository

Nine bulls from Israel with known QTL genotype (two

heterozygous and seven homozygous); nine bulls

from USA with known QTL genotype (one hetero-

zygous and eight homozygous); 670 daughters of

heterozygous sires (Israel); 411 Israeli bulls with

progeny testing data

Evidence from mapping QTL mapping (half-sib analysis, general pedigree

linkage analysis and LDLA, Box 1) for five milk

production traits using 38 markers and three different

statistical methods resulted in consistent QTL near

marker BM143; 95% confidence interval of 7.2 cM

QTL mapping (half-sib analyses and haplotype shar-

ing, Box 1) for loci affecting five milk production traits

described in previous papers, resulting in a confidence

interval of 4 cM spanning the QTL region [11]

Approach to detect functional

mutation

Complete sequencing of 12.3 Kb spanning OPN in four

heterozygous bulls and four homozygous bulls; any

candidate SNP should be concordant with inferred QTL

genotype of eight sequenced sires

Partial sequencing of ten known genes in and around

the 420-Kb interval in two heterozygous sires for

polymorphism detection (total of 31665 bp); sub-

sequent evaluation of these polymorphisms in the

various experimental populations

Proposed gene OPN, SPP1, Eta-1 ABCG2

Function of proposed gene OPN is a secreted glycoprotein involved in cell matrix

interactions and cellular signalling

ABCG2 is a half-transporter that has been shown to

transport xenobiotics and cytostatic drugs across the

plasma membrane; the authors propose that ABCG2

can transport cholesterol into milk

Nature of mutation Indel (OPN3907) in poly-T tract at w1240 bp upstream

of the OPN transcription initiation site; possibly novel

regulatory element?

Missense mutation: tyrosine to serine substitution in

the fifth extracellular region of ABCG2

Additional statistical evidence Analysis of four OPN SNPs in 167 sires shows that only

OPN3907 is significant; analysis of maternally inherited

OPN3907 allele shows significant effect on fat and

protein percentage; including OPN3907 genotype as a

separate effect accounts for all the variation due to the

QTL; probability that none of the undetected SNPs in

the 420-Kb interval will be concordant with segregation

status w0.40

Strongest population-wide association among all

polymorphisms that were tested; Furthermore, all

other associations were no longer significant when

accounting for the ABCG2 effect; allele frequency over

time is in agreement with changes in selection goals;

segregation status (QTL genotype) of 18 sires agrees

with genotype of candidate mutation; probability of

chance association !0.0001

Functional evidence from

cattle studies

OPN and ABCG2 are the only genes within the 420-Kb region that are differentially expressed in bovine mammary

gland [1]

Additional evidence from

other studies

Expression inmurinemammary tissue is dependant on

post-natal developmental stage; OPN is necessary for

normal mammary development (mouse model)

ABGC2 has demonstrated to be responsible for

secreting important substrate into milk in a mouse

model

Update TRENDS in Genetics Vol.22 No.6 June 2006304

probability that the SNP and the QTL co-segregate bychance is 0.00008. However, this figure ignores theprevious evidence for a QTL in this region and the levelof LD that will be present within the 420-Kb region. Bothanalyses leave sufficient room for one or more, as yetundetected, SNPs in the 420-Kb region that might be innear-complete LD with either or both putative mutations.

Arguments in favour of ABCG2 being the geneunderlying the QTL are (i) the mutation had a significanteffect in multiple sub samples of two populations; and (ii)the candidate mutation had functional implications. Thecase for OPN could have been stronger if the segregationstatus of the mutation in OPN in relation to the QTLgenotype was confirmed by Schnabel et al. for all 45 bulls inthe experimental population. The authors provide detailsof the segregation status of OPN3490 among all 45 bulls(demonstrating discrepancy between QTL genotype andSNP genotype) but curiously not for OPN3907. However,the results of Schabel et al. were available to Cohen-Zinderet al. and the latter acknowledge that they includedsequence data on the proposed OPN effect. Because OPNwas proposed first, this put the onus on Cohen-Zinder et al.to demonstrate that OPN was not the functional gene.However, they state that the region surrounding the OPNmutation is hyper variable with at least four SNPs in a20-bp region surrounding the poly-T sequence, and thattherefore this mutation cannot be the underlying

www.sciencedirect.com

functional mutation. They could have demonstrated thatthe segregation of theOPNmutation is not concordant withthe QTL genotypes in their population, but they do notmake any specific statements about the segregation ofOPN3907. Given the close links (as a result of AI) betweenHolstein cattle populations in the USA (and worldwide), itis surprising that Schabel et al. describe a limited numberof polymorphisms in the eight US bulls they analysed,whereas Cohen-Zinder et al. report that the region is hyper-variable in 18 Israeli and US bulls with known QTLgenotype. This discrepancy between their results mightresult from the difficulty in obtaining a good sequence readof a region with an indel and the risk of finding spuriousheterozygotes (R. Schnabel, personal communication).

Currently, the arguments against OPN are not convin-cing and the failure to dismiss the OPN mutation suggeststhat, in the population evaluated by Cohen-Zinder et al.,there might be perfect LD between the putativeOPN alleleand the ABCG2 allele. Therefore, based on the current,mostly statistical, evidence, the case for either OPN orABCG2 has not been conclusively proved nor dismissed.Although both studies use the reasoning of MacKay [5] tomake a joint case for their functional mutation, thestatistical arguments are not all from ‘independentsources’ and the circumstantial evidence (expression inmammary tissue, conservation among mammals, possiblefunction) is of more or less of equal merit for both studies.

Page 5: Conflicting candidates for cattle QTLs

Update TRENDS in Genetics Vol.22 No.6 June 2006 305

Concluding remarks

Although both articles present a coherent case for aputative functional mutation underlying a QTL on BTA6,there is no clear ‘winner’. It seems likely that, in thepopulations that were studied, the two candidatemutations are in strong or even complete LD. It is possiblethat the QTL is due to a polymorphism that is yet to bedetected. A third functional mutation on BTA6 (inPPARGC1A) [15] has been described, but it might underliea different QTL than the one analyzed in these studies.

Both studies depend heavily on the assumption that the420-Kb interval that was described in Norwegian cattle[13] will harbour the causative mutation underlying theirQTL, but widening the search beyond this interval shouldbe considered until this can be corroborated in anotherpopulation that is segregating for the same QTL.

An important lesson from these studies is that oneshould not ignore the extensive disequilibrium in suchpopulations. Both studies present high levels of LD thatare characteristic for dairy cattle [16]. As a result, severalcandidate mutations can be in complete LD and hencehave the same statistical evidence for being the functionalmutation. For definite proof, functional evidence, such asan effect of the mutation on expression or activity of thelocus and its product, should be provided [17], which hasbeen acknowledged by both groups.

Alternatively, one can search for populations where thecandidate mutations are not in complete LD. Given thelarge number of QTL studies that included BTA6 [10],there is ample data to test both candidate mutations. Thewide variety of breeds in which a QTL in this region hasbeen described increases the opportunity to identify apopulation in which the two candidate loci are not incomplete LD. Both groups have laid the ground work forfuture studies to determine whetherOPN,ABCG2 or an, asyet, unknown mutation is responsible for the QTL effect.

Acknowledgements

I acknowledge support from the BBSRC. Chris Haley, Robert Schnabeland two anonymous referees are gratefully acknowledged for theircomments on this article.

References

1 Cohen-Zinder, M. et al. (2005) Identification of a missense mutation inthe bovine ABCG2 gene with a major effect on the QTL on chromosome6 affecting milk yield and composition in Holstein cattle. Genome Res.15, 936–944

2 Van Laere, A.S. et al. (2003) A regulatory mutation in IGF2 causes amajor QTL effect on muscle growth in the pig. Nature 425, 832–836

3 Freking, B.A. et al. (2002) Identification of the single base change

Elsevier.com – Dynamic New Site Links S

Elsevier.com has had a makeover, inside and out.

As a world-leading publisher of scientific, technical and health

professionals to the best thinking in their fields. We offer the widest a

pollination of information, breakthroughs in research and discover

Elsevier.com.

Elsevier. Building Insights

www.sciencedirect.com

causing the callipyge muscle hypertrophy phenotype, the only knownexample of polar overdominance in mammals. Genome Res. 12,1496–1506

4 Grisart, B. et al. (2002) Positional candidate cloning of a QTL in dairycattle: identification of a missense mutation in the bovine DGAT1 genewith major effect on milk yield and composition. Genome Res. 12,222–231

5 Mackay, T.F. (2001) The genetic architecture of quantitative traits.Annu. Rev. Genet. 35, 303–339

6 Schnabel, R.D. et al. (2005) Fine-mapping milk production quan-titative trait loci on BTA6: analysis of the bovine osteopontin gene.Proc. Natl. Acad. Sci. U. S. A. 102, 6896–6901

7 Bovenhuis, H. and Weller, J.I. (1994) Mapping and analysis of dairycattle quantitative trait loci by maximum likelihood methodologyusing milk protein genes as genetic markers. Genetics 137, 267–280

8 Spelman, R.J. et al. (1996) Quantitative trait loci analysis for five milkproduction traits on chromosome six in the Dutch Holstein-Friesianpopulation. Genetics 144, 1799–1808

9 Velmala, R.J. et al. (1999) A search for quantitative trait loci for milkproduction traits on chromosome 6 in Finnish Ayrshire cattle. Anim.Genet. 30, 136–143

10 Khatkar, M.S. et al. (2004) Quantitative trait loci mapping indairy cattle: review and meta-analysis. Genet. Sel. Evol. 36,163–190

11 Ron, M. et al. (2001) Multiple quantitative trait locus analysis ofbovine chromosome 6 in the Israeli Holstein population by a daughterdesign. Genetics 159, 727–735

12 Olsen, H.G. et al. (2004) Fine mapping of milk production QTL onBTA6 by combined linkage and linkage disequilibrium analysis.J. Dairy Sci. 87, 690–698

13 Olsen, H.G. et al. (2005) Mapping of a milk production quantitativetrait locus to a 420-kb region on bovine chromosome 6. Genetics 169,275–283

14 Everts-van der Wind, A. et al. (2004) A 1463 gene cattle-humancomparative map with anchor points defined by human genomesequence coordinates. Genome Res. 14, 1424–1437

15 Weikard, R. et al. (2005) The bovine PPARGC1A gene: molecularcharacterization and association of an SNP with variation of milk fatsynthesis. Physiol. Genomics 21, 1–13

16 Farnir, F. et al. (2000) Extensive genome-wide linkage disequilibriumin cattle. Genome Res. 10, 220–227

17 Mehrabian, M. et al. (2005) Integrating genotypic and expression datain a segregating mouse population to identify 5-lipoxygenase as asusceptibility gene for obesity and bone traits. Nat. Genet. 37,1224–1233

18 Weller, J.I. et al. (1990) Power of daughter and granddaughter designsfor determining linkage between marker loci and quantitative traitloci in dairy cattle. J. Dairy Sci. 73, 2525–2537

19 Heath, S.C. (1997) Markov Chain Monte Carlo segregation andlinkage analysis for oligogenic models. Am. J. Hum. Genet. 61,748–760

20 Meuwissen, T.H. et al. (2002) Fine mapping of a quantitative traitlocus for twinning rate using combined linkage and linkagedisequilibrium mapping. Genetics 161, 373–379

21 Riquet, J. et al. (1999) Fine-mapping of quantitative trait loci byidentity by descent in outbred populations: application to milkproduction in dairy cattle. Proc. Natl. Acad. Sci. U. S. A. 96,9252–9257

0168-9525/$ - see front matter Q 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.tig.2006.04.006

cientists to New Research & Thinking

information, Elsevier is dedicated to linking researchers and

nd deepest coverage in a range of media types to enhance cross-

y, and the sharing and preservation of knowledge. Visit us at

. Breaking Boundaries