Multiple Cross Mapping (MCM): A Different Recipe for Measuring QTLs and Finding Genes
R Hitzemann1,2, B Malmanger1,2, S Cooper1,2, S Coulombe1,2, C Reed1,2, K Demarest3, J Koyner3, L Cipp3, J Flint4, C Talbot5, B Rademacher1,2, K Buck1,2, J Sikela6, Y Xu6, J McCaughran Jr.3
1Department of Behavioral Neuroscience, Oregon Health & Science University, Portland OR 97201-3098
2Research Service, Veterans Affairs Medical Center, Portland, OR 97201
3Department of Psychiatry, SUNY at Stony Brook, Stony Brook, NY 11733
4Wellcome Trust Centre for Human Genetics, Roosevelt Drive, Oxford OX3 7BN, UK
5Department of Genetics, University of Leicester, Leicester LE1 7RH, UK
6Department of Pharmacology, University of Colorado Health Sciences Center, Denver, CO 80262
Previously we observed that it was possible to use data from four different diallele crosses to effectively interrogate the microsatellite map and markedly improve the resolution of QTL(s) for murine basal activity (Hitzemann et al. 2000). Rather than simply using available opportunistic samples, the current study provides some formal structure to the use of multiple cross mapping (MCM) and illustrates that MCM cannot only be used to improve QTL resolution but also as an effective tool both for finding QTL relevant genes and for investigating genetic architecture. The ingredients in the current version of MCM are as follows: four inbred mouse strains - C57BL/6J (B6), DBA/2J (D2), BALB/cJ (C) and LP/J (LP); the six diallele crosses that can be formed from these strains; a cross of the four strains (HS4), currently at G11; an eight-way cross (HS8) formed from the above four strains and the C3H/HeJ, AKR/J, A/J and CBA/J strains and currently at G44; detailed microsatellite and SNP maps; physical maps of the B6 and D2 strains; gene expression maps (Affymetrix) for the four key strains. The first example of the MCM approach illustrates the ability of the strategy to reduce the QTL interval. The phenotype is the psychomotor stimulant response to a moderate dose (1.5 g/kg) of ethanol. Previous studies (e.g. Demarest et al. 2001) established the presence of two QTLs in a B6xD2 intercross on distal chromosome 1 and the mid region of chromosome 2. The chromosome 1 QTL was also detected in the B6xC cross but not in the other four diallele crosses (B6xLP, CxLP, CxD2, LPxD2). For both the B6xC and B6xD2 crosses, the direction of the QTL was against phenotype i.e. it was the B6 allele that increased activity. Sorting the Mit microsatellite markers on the basis of whether or not the QTL was present or not, revealed a single cluster of markers at 83+2 cM; this position of the QTL was confirmed by fine mapping in HS8 animals (see Mott et al. 2000 for analytic details and Demarest et al. 2001 for a description of the HS sample). The second example of MCM illustrates the ability of the approach to interrogate polymorphisms in gene structure. For this example, we turn to the ethanol response QTL on chromosome 2. The QTL was first detected in the B6xD2 intercross and is also found in the D2xLP intercross but not in the other four intercrosses. Catalase (Cas1) was a strong candidate gene for this QTL since it has been established that increasing or decreasing brain catalase activity, increases or decreases the ethanol stimulant response. The ORF of Cas1 was sequenced in the B6, D2, C and LP strains; a single distinguishing polymorphism was found such that the B6 strain will have a threonine at position #117 while the other three strains will have alanine. The alanine to threonine polymorphism is associated with a marked decrease in catalase activity. However, on the basis of the MCM data, this polymorphism alone cannot generate the QTL. This point was confirmed in the HS8 sample; the QTL was located in a distinct linkage group from Cas1 (see Demarest et al.2001). A similar strategy was used to examine a polymorphism in the ORF of Bdnf, also a candidate within the QTL interval. At position #32, the B6 and LP strains have a leucine while the D2 and C strains have a methionine. It is not known if this polymorphism has a functional consequence but again the MCM data indicate that this polymorphism alone cannot generate the QTL. The third example of MCM illustrates the ability of the approach to interrogate gene expression data. For this example, we turn to the QTLs for basal activity on chromosome 1; at least two distinct QTLs have been identified, both of which appear to be present in the B6xD2 intercross (see e.g. Koyner et al. 2000). The most distal of these QTLs is found in the B6xC and B6xLP intercrosses but not the other three intercrosses. Brain gene expression among the four strains was assessed using the Affymetrix “A” chip (N=6/strain); 325 transcripts showed differential expression among the four strains at p < 10-6. These data were used to construct chromosome expression maps to detect the coincidence of QTLs and differential gene expression. On distal chromosome 1 and within the basal activity QTL interval , it was found that Kcnj9 (GIRK3) expression in the B6 strain is approximately 1/10 the level of that found in the other three strains. This differential pattern of expression matches the MCM data and suggests that GIRK3 is a strong candidate to be associated with the basal activity QTL. A similar conclusion was reached by Sandberg et al. (2000) when comparing the B6 and two 129 strains. Of the G-protein inwardly rectifying potassium channels found in brain, GIRK3 is normally the most prevalent and is linked to the signaling cascade of a wide variety of neurotransmitter and neuropeptide receptors.
The MCM method also provides a unique mechanism for examining genetic architecture. All of the QTLs in the six diallele crosses will be known and the question of whether or not all or only some of these are found in the four-way cross, can be examined. Importantly for such an analysis with a four-way cross and existing databases, it is possible to precisely determine the genotype.
Overall, data have been obtained showing that the MCM strategy can under certain circumstances markedly improve QTL resolution. However, the greatest value of the approach is likely to be found in using the method to interrogate structural and regulatory polymorphisms and to determine which are QTL relevant.
Demarest K, Koyner J, McCaughran J, Cipp L, Hitzemann R (2001) Further characterization and high resolution mapping of quantitative trait loci for ethanol-induced locomotor activity. Behav Gen, 31: 79-91.
Hitzemann R, Demarest K, Koyner J, Cipp L, Patel N, Rasmussen E and McCaughran J Jr. (2000) Effect of genetic background on the detection of quantitative trait loci. Pharmacol Biochem Behav 67: 767-772.
Koyner J, Demarest K, McCaughran J, Cipp L, Hitzemann R (2000)Identification and Time Dependence of Quantitative Trait Loci for Basal Locomotor Activity in the BXD Recombinant Inbred Series and a B6D2 F2 Intercross. Behav Gen 30: 159-170.
Mott R, Talbot CJ, Turri MG, Collins AC, Flint J (2000) A method for fine mapping quantitative trait loci in outbred animal stocks. Proc Natl Acad Sci 97: 12649-12654.
Sandberg R, Yasuda R, Pankratz DG, Carter TA, Del rio JA, Wodicks L, Mayford M, Lockhardt DJ, Barlow C (2000) Regional and strain-specific gene expression mapping in the adult mouse brain. Proc Nat Acad Sci (USA) 97: 11038-11043.
[This study was supported by grants from the Public Health Service - AA 11043, AA 134834 and MH 51372 and through a grant from the Veterans Affairs Research Service]