FULL PAPER Physiology

Characterization of Cq3, a Quantitative Trait Locus that Controls Plasma Cholesterol and Phospholipid Levels in Mice

Jun-ichi SUTO1)

1)Department of Molecular Biology and Immunology, National Institute of Agrobiological Sciences, Tsukuba, Ibaraki 305–8634, Japan

(Received 16 May 2005/Accepted 1 December 2005)

y ABSTRACT. Cq3 was identified in C57BL/6J (B6) × KK-A F2 mice as a quantitative trait locus (QTL) that controls plasma cholesterol and phospholipid levels, and normolipidemic B6 allele was associated with increased lipids. Cq3 was statistically significant in F2-a/a, y y but not in F2-A /a; probably because the Cq3 effect was obscured by introduction of the A allele, which in itself has a strong hyperlip- idemic effect. Because the peak LOD score for Cq3 was identified near D3Mit102 (49.7 cM) on 3, linkage analyses with microsatellite markers located at 49.7 cM were performed in KK × RR F2, B6 × RR F2, and KK × CF1 F2. However, even a suggestive QTL was not identified in any of the three F2. By testing all pairs of marker loci, I found a significant interaction between Cq3 and the KK KK B6 B6 Apoa2 locus, and F2 mice with the Apoa2 /Apoa2 ; D3Mit102 /D3Mit102 genotype had significantly higher cholesterol levels than did F2 mice with other genotypes. The results showed that the ‘round-robin’ strategy was not always applicable to the search for QTL ; probably because specific -to-gene interaction limited the validity of the strategy to the utmost extent. KEY WORDS: Apoa1bp, Apoa2, cholesterol, cholesterol QTL (Cq), phospholipid. J. Vet. Med. Sci. 68(4): 303–309, 2006

Plasma lipid levels are quantitative traits that are con- between KK and RR strains [17]. In addition, Cq3 had an trolled by multiple genes under the influence of environ- effect on plasma T-CHO and phospholipid (PL) levels in y y mental stimuli. To dissect genetic basis, quantitative trait C57BL/6J × KK-A F2-a/a (n=93), but not in F2-A /a (n=97) locus (QTL) analyses have been carried out extensively dur- [19]. The aims in study are to explain the reasons why Cq3 ing the last decade, particularly on mice [19, 23]. Col- was identified only in F2-a/a, and why the ‘round-robin’ leagues and I previously carried out QTL analyses on strategy was not applicable to the case of Cq3. plasma cholesterol levels in three sets of F2 mice that were formed in a ‘round-robin’ manner from C57BL/6J, KK (- MATERIALS AND METHODS Ay), and RR strains [19–21]. We identified six QTLs on 1 (Cq1, Cq2, and Cq6), 3 (Cq3), and 9 (Cq4 Mice and genetic cross: In the present study, results from and Cq5). These QTLs have plausible candidate genes, that studies on the following four sets of F2 mice were analyzed. y y is, sterol O-acyltransferase 1 (Soat1) for Cq1, apolipopro- [B6 × KK-A F2] Inbred mouse strains KK-A (yellow hair, tein A-II (Apoa2) for Cq2 and Cq6, apolipoprotein A-I bind- AyaBBCc) and B6 (black hair, aaBBCC) were purchased ing (Apoa1bp) for Cq3, and apolipoprotein A-I from Clea Japan (Tokyo), and F2 mice (see below) were pro- (Apoa1) and apolipoprotein A-IV (Apoa4) for Cq4 and Cq5. duced as previously described [19]. Briefly, 190 F2 mice y Of these QTLs, Cq2 (identified in C57BL/6J × KK-A F2) were chosen from >200 F2 to minimize age differences colocalized with Cq6 (identified in C57BL/6J × RR F2), and between individual mice in the experiments. The 190 F2 y Cq4 (identified in C57BL/6J × KK-A F2) colocalized with comprised 93 animals with the a/a genotype at the agouti Cq5 (identified in KK × RR F2). I concluded that Apoa2 is locus (black hair, hereafter called F2-a/a, 49 males and 44 the gene causative of Cq2 and Cq6, because the KK and the females) and 97 animals with the Ay/a genotype (yellow b y RR have same Apoa2 allele that is considered to increase hair, hereafter F2-A /a, 49 males and 48 females). F2 mice plasma cholesterol level [22]. In the similar way, I con- were sacrificed at the ages of 26 to 29 weeks. cluded that Apoa4 is the gene causative of Cq4 and Cq5, [KK × RR F2] The KK strain was purchased from Clea because the only KK has unique nucleotide substitution that Japan, and the RR strain has been bred and maintained at the is considered to increase plasma cholesterol level [22]. National Institute of Agrobiological Sciences (NIAS, Thus, the ‘round-robin’ strategy is a valid method for Tsukuba, Japan). One hundred forty-five F2 females were searching QTL genes. produced and analyzed as previously described [20]. F2 In contrast to Cq2/Cq6 and Cq4/Cq5, Cq3 was identified mice were sacrificed at the age of 170 days. y only in C57BL/6J × KK-A F2 (more correctly in F2-a/a, not [B6 × RR F2] One hundred eighty-seven F2 females were y in F2-A /a, see below), and was not in KK × RR F2 and produced and analyzed as previously described [21]. F2 C57BL/6J × RR F2 [19–21]. This was at variance with the mice were sacrificed at the age of 130 ± 5 days. expectation that the QTL identified in the cross between [KK × CF1 F2] The inbred CF1 strain has been bred and y C57BL/6J and KK-A strains will be identified either in the maintained at the NIAS. Eighty-nine F2 females were pro- cross between C57BL/6J and RR strains or in the cross duced and analyzed in this study. F2 mice were sacrificed at 304 J-I. SUTO the age of 20 weeks. Stat123/Win software (Shinko Trading Co. Ltd. Publication Hereafter, I define the B6 strain as having B alleles, the Department, Tokyo) and SPSS software (SPSS for Win- CF1 strain as having C alleles, the KK strain as having K dows Release 7.5.1 J, SPSS Inc., Chicago, IL). alleles, and the RR strain as having R alleles, throughout the genome. RESULTS All mice were maintained in our laboratory throughout the experimental period under specific pathogen-free condi- For several phenotypic traits, scatter plots of the T-CHO tions, with a regular light cycle of 12 hr light/12 hr dark, and level and body weight (at the time of blood collection) in y with the temperature controlled at 22 ± 3°C. They had free parental KK-A males, KK males, B6 females, F1, and F2 are access to the diet [rodent pellet chow, CE-2 (342.2 kcal/100 shown in Figs. 1A and 1B. In general, T-CHO levels ran g, containing 4.4% crude fat), Clea Japan] and tap water. parallel with those of body weight; that is, heavier mouse y Plasma lipid measurement: In B6 × KK-A F2, plasma strains tended to have a higher T-CHO. T-CHO and body concentrations of triglyceride (TG), T-CHO, HDL-CHO, weight in KK-Ay males were higher than those in KK males. LDL+VLDL-cholesterol (LDL-CHO), free-cholesterol (F- Thus, introduction of the Ay allele simply increased the CHO), non-esterified fatty acid (NEFA), and PL were deter- amount of lipids in circulation. Similar trends were also y mined. In KK × RR F2, B6 × RR F2, and KK × CF1 F2, TG found in the comparison between F1-a/a and F1-A /a, and y and T-CHO were determined. At autopsy, after 24 hr fast- between F2-a/a and F2-A /a. Essentially, all the other lipid ing (KK × CF1 F2 was fasted for 4 hr), blood was drawn classes showed distribution patterns similar to that of T- from the heart into microtubes with heparin or EDTA as CHO [19]. anticoagulant. To separate the plasma from whole blood, Cq3 was further mapped in 93 F2-a/a mice with addi- the tubes were centrifuged at 7,000 rpm for 5 min at 4°C tional microsatellite markers to refine the 95% CI. Signifi- immediately after drawing. All samples were placed at cant QTL was obtained only for F-CHO (LOD score 4.5), –80°C until use. Plasma lipid was determined enzymati- and suggestive QTLs were identified for T-CHO (LOD cally with clinical chemical kits (Test Wako, Wako Pure score 4.2), HDL-CHO (3.8), LDL-CHO (3.0), and PL (3.6). Chemical Industries, Osaka, Japan). A peak LOD score of Cq3 for F-CHO was located at 49.7 QTL analysis: QTL analysis was carried out with the cM on chromosome 3, placing the 95% CI between 47 and Mapmaker/EXP version 3.0b and the Mapmaker/QTL ver- 54 cM (Figs. 2A). The LOD score rose abruptly from below sion 1.1b computer programs [7]. The chromosomal region a suggestive level to a significant level between D3Mit29 with a logarithm of odds (LOD) score of more than 4.3 (45.2 cM) and D3Mit102 (49.7 cM). The B allele was asso- (threshold of statistical significance at α=0.05) was recog- ciated with increased trait values in T-CHO (Table 1), and nized as indicating significant linkage, and the region with a also in F-CHO, HDL-CHO, LDL-CHO, and PL (data not LOD score between 2.8 and 4.3 was recognized as indicat- shown). Thus, Cq3 had an effect on HDL-CHO, LDL- ing suggestive linkage [8]. The α level for suggestive link- CHO, and PL, but not on TG (Table 1) and NEFA. In con- age implies the expectation that there will be one false trast, although no significant Cq3 effects were observed in y positive in a genome-wide search. Once a significant QTL F2-A /a for all traits examined (Fig. 2B), there was still a was identified, the 95% confidence interval (CI) was deter- trend that the B allele was associated with increased T- mined as an approximate indicator for postulating candidate CHO, but not with TG (Table 1). Thus, the detection of Cq3 genes. The 95% CI is defined by a 1.5 LOD decrease; there- was hindered by the presence of the Ay allele. To test for a fore, it is the equivalent of a 1.5-LOD support interval. possible statistical interaction between Ay (agouti) and Cq3 y In B6 × KK-A F2, males and females were analyzed (D3Mit102), two-way ANOVA was carried out, but there together. In order to merge data from males and females was no significant interaction between the two factors (data within each F2, trait variables for the males and females not shown). Accordingly, I next examined whether there were standardized to a mean 0 and a variance 1 by subtract- were any other loci that could modulate Cq3 action. In order ing the gender-specific mean from each individual value and to identify loci that interact with Cq3, I performed a dividing each difference by the standard deviation of its genome-wide search by testing all pairs of marker loci with respective gender. Map Manager QTX software [10]. Statistically significant The effects of gene interaction by pairs of marker loci interactions were identified between loci on chromosome 3 were analyzed with Map Manager QTX software [10]. and loci on distal (data not shown). The Apoa1bp cDNA sequencing: A part of the open reading most striking interaction was detected between D1Mit150 frame (ORF) of Apoa1bp cDNA sequences from the B6, (microsatellite marker closest to Cq2, a previously identi- CF1, KK, and RR strains were determined by direct fied T-CHO QTL) and D3Mit102 (for T-CHO, P<2.3 × sequencing of RT-PCR amplified fragments. cDNA synthe- 10–5). An interaction effect between the two loci was y sized based on total RNA from the liver of these strains detected in both F2-a/a and F2-A /a when each F2 was ana- [RNeasy QIAGEN; Takara RNA PCR kit (AMV) Version lyzed separately (data not shown); therefore, both F2 were 2.1] was amplified with a set of primers: 5’-gtcatcagccagca- merged and analyzed. Because Apoa2 was the definite can- gagtgt-3’ and 5’-gggtaagatggcaggttcag-3’. didate gene for Cq2, and because Apoa2B and Apoa2K alleles Statistics: Statistical analyses were carried out with were distinguishable by PCR-RFLP analysis [19], a direct CHOLESTEROL AND PHOSPHOLIPID QTL 305

Fig. 1. Scatter plots of plasma T-CHO (A) level and body weight (B) in parental KK-Ay males (n=10), KK males (n=10), C57BL/6J y y females (n=10), F1-a/a males (n=7), F1-a/a females (n=7), F1-A /a males (n=7), F1-A /a females (n=7), F2-a/a males (n=49), F2-a/a y y females (n=44), F2-A /a males (n=49), and F2-A /a females (n=48). Each circle represents an individual trait value. Short horizontal lines indicate mean of trait value. interaction between the Apoa2 gene and D3Mit102 was The effect of Cq3 on lipids (T-CHO and TG) was exam- examined, with the result that these loci were found to inter- ined in other F2 sets, that is, previously analyzed KK × RR –5 act significantly (for T-CHO, P<4.3 × 10 ). On the basis of F2 and newly produced KK × CF1 F2 (Table 2). For both F2, the combinations of alleles at the two loci, 190 F2 mice were microsatellite markers located at 49.7 cM on chromosome 3 divided into nine possible genotypes. One-way ANOVA were searched and genotyped. As seen in Table 2, Cq3 had showed that there was a statistically significant difference no effect on CHO as well as TG levels in all F2 sets. among the nine genotypes (P<1.5 × 10–14). In a multiple As a plausible candidate gene, Apoa1bp is known to be comparison by Scheffe’s post-hoc test, only the mice located in the mid part of chromosome 3, partial cDNA homozygous for the K allele at Apoa2 and homozygous for sequences (87% of ORF) were determined in B6, KK, RR, the B allele at D3Mit102 (Apoa2K/Apoa2K; D3Mit102B/ and CF1 (Table 3). Two successive base substitutions at D3Mit102B; that is, Cq2K/Cq2K; Cq3B/Cq3B) had signifi- nucleotides 113 (C-to-T) and 114 (A-to-G) with an amino cantly higher T-CHO levels than did mice of the other eight acid change (Thr-to-Met) were identified. The B allele was genotypes (Fig. 3A). When the four F2 subgroups (F2-a/a different from the K, R, and C alleles. In B6 × RR F2 mice, y y males, F2-a/a females, F2-A /a males, and F2-A /a females) which have the same Apoa1bp allele (as far as the above were analyzed separately, the average T-CHO levels in mice nucleotide substitutions are concerned) combination as that K K B B y of the Apoa2 /Apoa2 ; D3Mit102 /D3Mit102 genotype in B6 × KK-A F2, Cq3 again had no effect on T-CHO and was higher than in mice with other genotypes (data not TG levels, suggesting that this polymorphism was not shown), and the average T-CHO levels in mice with this responsible for Cq3 (Table 2). genotype was as follows: 179.3 mg/dl in F2-a/a males, 122.0 y mg/dl in F2-a/a females, 163.3 mg/dl in F2-A /a males, and DISCUSSION y 265.0 mg/dl in F2-A /a females (compare these to the mean T-CHO levels in each F2 subgroup shown in Fig. 1A). Several studies have identified CHO QTLs on chromo- Essentially the same analysis was carried out for TG. One- some 3. Machleder et al. [9] identified HDL-CHO QTL way ANOVA showed that there was statistically significant near D3Mit120 (28 cM) in B6 × C3H/HeJ F2, and the C3H difference among the nine genotypes (P<0.007). Although allele was associated with increased HDL-CHO levels. mice of the Apoa2K/Apoa2K; D3Mit102B/D3Mit102B geno- Mehrabian et al. [11] identified T- and HDL-CHO QTL type tended to have highest TG levels among the nine geno- near D3Mit12 (49.2cM) in B6 × CAST/Ei F2, and the B types, the statistical significance was reserved on the basis allele was associated with increased CHO levels. Shike et of the multiple comparison by Scheffe’s post-hoc test (Fig. al. [18] identified T-CHO QTL (Tcho-1) near D3Mit12 3B). (49.2 cM) in a backcross between KK × (BALB/c × KK) F1, 306 J-I. SUTO

y y y Fig. 2. LOD score plots for plasma lipid levels in C57BL/6J × KK-A /a F2-a/a mice (A) and in C57BL/6J × KK-A /a F2-A /a mice (B). Ver- tical lines in (A) represent genome-wide thresholds for suggestive linkage (LOD=2.8) and significant linkage (LOD=4.3). and the K allele was associated with increased T-CHO lev- determined Apoa1bp sequences, but did not find any els. Anunciado et al. [1] identified T-CHO QTL near sequence differences between the two strains. The present D3Mit46 (13.8 cM) in SM/J × A/J F2. This chromosome 3 study identified two successive nucleotide substitutions that locus had effects on PL and T-CHO, but not on TG, the result in an amino acid change; however, this might not be same as the effect of Cq3. Colinayo et al. [2] identified T- responsible for Cq3. There are possibilities that putative and HDL-CHO QTL near D3Mit241 (33 cM) in B6 × DBA/ nucleotide alterations in the non-coding regulatory region of 2J F2, and B allele was associated with increased CHO lev- Apoa1bp gene have an effect on transcriptional efficiency, els. Korstanje et al. [6] identified HDL-CHO QTL or the 5’- and/or 3’- untranslated region of Apoa1bp mRNA (Hdlq21) near D3Mit11 (49 cM) in SM/J × NZB/BlNJ F2. have an effect on RNA stability. Apoa1bp was by no means They also mapped one of the QTLs controlling phospholipid ruled out as a candidate, because Apoa1bp is located on transfer protein (PLTP) activity, Pltpq1, to chromosome 3 chromosome 1q21 in human, to which a familial combined near D3Mit22 (33.7 cM) [5]. PLTP has an important role in hyperlipidemia (FCHL) locus has been mapped (see below). the transfer of PL from VLDL to HDL, and in the remodel- The fact that Cq3 was identified in F2-a/a, but not in F2- ing of HDL. Several of these loci on chromosome 3 may be Ay/a, is difficult to explain. A possible explanation for this the same. Indeed, because QTL as identified by Mehrabian is that Cq3 interacts with the Ay allele; however, no signifi- et al. [11] was very close to Cq3, because the B allele was cant interaction was present between Ay (agouti) and Cq3 associated with increased CHO levels in both loci, and (D3Mit102). Another possible explanation is that there are because the B allele at both loci behaved in a recessive other loci that could modulate the action of Cq3. In fact, a mode, these QTLs are suggested to be allelic. Thus, the mid statistically significant interaction was identified between part of murine chromosome 3 is a region to which many D3Mit102 (Cq3) and Apoa2 (Cq2), although this synergism y cholesterol QTLs are frequently mapped. was found in both F2-a/a and F2-A /a. It has been reported It may be appropriate to consider that apolipoprotein and/ that APOAIBP protein interacts with APOAII protein [16]; or apolipoprotein-related genes are primary candidate genes therefore, Apoa1bp is still qualified as candidate for Cq3. for CHO QTLs. Therefore, I and others [5] examined Another possible explanation is that the Cq3 effect is Apoa1bp as a cause of relevant QTLs on chromosome 3. obscured by the introduction of the Ay allele, which in itself Korstanje et al. [5] analyzed SM/J × NZB/B1NJ F2 and has a strong hyperlipidemic effect. Elevated background CHOLESTEROL AND PHOSPHOLIPID QTL 307

y Table 1. Effects of Cq3 (D3Mit102) on T-CHO and TG in F2-a/a and F2-A /a mice Traits Mice Mean ± S.D. lipid values by D3Mit102 genotypes Nominal P value

F2-a/a KK KB BB T-CHO M+F –0.30 ± 0.77 (n=31) –0.16 ± 0.67 (n=42) 0.81 ± 1.41 (n=20) <0.00009 M 92.5 ± 22.5 (n=14) 100.7 ± 24.9 (n=25) 136.3 ± 48.9 (n=10) <0.003 F 82.8 ± 17.5 (n=17) 83.8 ± 11.9 (n=17) 102.2 ± 29.6 (n=10) <0.04 TG M+F –0.04 ± 0.97 –0.11 ± 0.69 0.28 ± 1.47 NS (P>0.3) M 123.3 ± 32.1 122.9 ± 36.3 141.8 ± 80.7 NS (P>0.5) F 111.9 ± 44.4 106.9 ± 21.9 121.1 ± 48.1 NS (P>0.6)

y F2-A /a KK KB BB T-CHO M+F –0.12 ± 0.81 (n=22) –0.06 ± 0.98 (n=51) 0.24 ± 1.17 (n=24) NS (P>0.3) M126.9 ± 35.4 (n=11) 134.5 ± 43.3 (n=25) 131.8 ± 32.7 (n=13) NS (P>0.8) F143.2 ± 39.9 (n=11) 138.7 ± 45.0 (n=26) 179.0 ± 80.6 (n=11) NS (P>0.1) TG M+F 0.08 ± 0.96 –0.05 ± 0.90 0.02 ± 1.23 NS (P>0.8) M 163.5 ± 56.2 161.3 ± 51.9 147.5 ± 57.4 NS (P>0.7) F 179.1 ± 57.1 166.0 ± 52.2 192.9 ± 87.2 NS (P>0.4) M, males; F, females; M+F, males and females (merged). M+F data are expressed as standardized values, and M and F data are expressed as mg/dl. NS, not significant.

Fig. 3. Interaction effect between Apoa2 (Cq2) and D3Mit102 (Cq3) on T-CHO (A) and TG (B). The Apoa2 allele type was determined by PCR-RFLP analysis as described previously [19]. For T-CHO, mice of the Apoa2K/Apoa2K; D3Mit102B/D3Mit102B had significantly higher T-CHO levels than did mice of the other eight genotypes (indicated by *).

y level of lipids in F2-A /a may hinder the detection of a statis- robin’ manner, cholesterol QTLs on chromosomes 1 (Cq2 tically significant difference. This Ay effect has been sug- and Cq6) and 9 (Cq4 and Cq5) were identified twice in three gested to be a metabolic consequence of obesity (Figs. 1A analyses [19–21]. I believe that this condition is crucial for and 1B). testing the candidate genes. In this respect, it was trouble- y If the QTL is identified in the cross between ‘A’ and ‘B’ some that Cq3 was identified only in B6 × KK-A (F2-a/a), strains, the QTL will necessarily be identified either in the but was not identified in KK × RR F2 and B6 × RR F2. There cross between ‘A’ and ‘C’ strains or in the cross between may be several possible explanations. First, a gender differ- ‘B’ and ‘C’ strains, and must not be identified in both ence in the effect of Cq3 might produce the above discrep- crosses [17]. Indeed, in our previous QTL studies on three ancy. Indeed, the nominal P value for T-CHO was smaller y sets of F2 mice formed from three inbred strains in a ‘round- in males than in females in B6 × KK-A (F2-a/a) (Table 1). 308 J-I. SUTO

Table 2. Effects of a locus a the 49.7 cM position on chromosome 3 on T-CHO levels in four sets of F2 mice

F2 intercrosses Marker closest Lipids Mean ± S.D. lipid values (mg/dl) for genotypes at marker Nominal (No. of mice) to Cq3 P value (Position) KK (n=28) KR (n=78) RR (n=39) KK × RR D3Mit247 T-CHO 108.6 ± 11.4 114.2 ± 12.9 114.3 ± 13.6 NS (P>0.1) (n=145) (49.7 cM) TG 144.7 ± 44.4 155.6 ± 49.5 140.0 ± 40.8 NS (P>0.1) KK (n=28) KC (n=42) CC (n=19) KK × CF1 D3Mit189 T-CHO 89.1 ± 15.5 91.8 ± 18.2 85.1 ± 15.8 NS (P>0.3) (n=89) (49.7 cM) TG 195.6 ± 66.8 202.2 ± 83.7 210.8 ± 70.2 NS (P>0.7) RR (n=44) RB (n=95) BB (n=48) C57BL/6J × RR D3Mit102 T-CHO 90.3 ± 18.2 90.4 ± 25.3 92.5 ± 21.6 NS (P>0.8) (n=187) (49.7 cM) TG 114.6 ± 58.23 113.3 ± 40.0 124.1 ± 58.5 NS (P>0.4)

In these three crosses, only F2 female mice are analyzed. Marker position is expressed as a distance in centi-Morgan from the centromere. Genotypes are expressed by abbreviated allele designation as follows: B (C57BL/6J), C (CF1), K (KK), R (RR). NS, not significant.

Table 3. Nucleotide and amino acid polymorphism in [3], and because a synteny break in the human 1q21-1q23 Apoa1bp gene region was repeatedly observed in other mammalian species Strain Nucleotides 113, 114 Amino acid as well [12], it may be possible to hypothesize that some C57BL/6J (A)CA Thr common hyperlipidemia in mammals are due to a combina- CF1 (A)TG Met tion of two independent gene loci on different chromo- KK (A)TG Met somes. RR (A)TG Met Polymorphism in Apoa1bp cDNA sequences from C57BL/ ACKNOWLEDGMENT. This work was supported in part 6J, CF1, KK, and RR strains. Two successive base by a grant from the Ministry of Agriculture, Forestry, and substitutions at nucleotides 113 (C-to-T) and 114 (A-to-G) Fisheries, Japan. were identified, and therefore Thr-to-Met amino-acid substitutions were deduced. REFERENCES

Only females were analyzed in the latter two crosses (KK × 1. Anunciad, R.V., Nishimura, M., Mori, M., Ishikawa, A., Tanaka, S., Horio, F., Ohno, T. and Namikawa, T. 2003. Quan- RR F2 and B6 × RR F2). Second, a difference in the ages of mice at the time of lipid collection may cause the above dis- titative trait locus analysis of serum insulin, triglyceride, total crepancy. The age of B6 × KK-Ay F mice was higher (195– cholesterol and phospholipids levels in the (SM/J × A/J) F2 2 mice. Exp. Anim. (Tokyo) 52: 37–42. 207 days) than that of the other F2 sets; for example, 170 2. Colinayo, V.V., Qiao, J.H., Wang, X., Krass, K.L., Schadt, E., days for KK × RR F2, 130 ± 5 days for B6 × RR F2, and 20 Lusis, A.J. and Drake, T.A. 2003. Genetic loci for diet-induced weeks for KK × CF1 F2. Third, whether Cq3 is identified or atherosclerotic lesions and plasma lipids in mice. Mamm. not may depend on the presence of a combination of differ- Genome 14: 464–471. ent Apoa2 alleles. The present results clearly demonstrated 3. Goldstein, J.L., Schrott, H.G., Hazzard, W.R., Bierman, E.L. that D3Mit102 (Cq3) significantly interacts with Apoa2 and Motulsky, A.G. 1973. Hyperlipidemia in coronary heart (Cq2). Indeed, the Apoa2 allele differed between B6 disease. II. Genetic analysis of lipid levels in 176 families and (Apoa2a) and KK (Apoa2b), whereas it did not differ delineation of a new inherited disorder, combined hyperlipi- between KK and RR (Apoa2b), or between KK and CF1 demia. J. Clin. Invest. 52: 1544–68. (Apoa2b). I consider that this is the most probable explana- 4. Huertas-Vazquez, A., del Rincon, J.P., Canizales-Quinteros, S., Riba, L., Vega-Hernandez, G., Ramirez-Jimenez, S., tion, and that specific gene-to-gene interaction limits the Auron-Gomez, M., Gomez-Perez, F.J., Aguilar-Salinas, C.A. validity of the strategy to the utmost extent. At any rate, the and Tusie-Luna, M.T. 2003. Contribution of chromosome ‘round-robin’ strategy was not always applicable to the 1q21-q23 to familial combined hyperlipidemia in Mexican search for QTL genes. families. Ann. Hum. Genet. 68: 419–427. Murine chromosomal regions containing Apoa2 (chro- 5. Korstanje, R., Albers, J.J., Wolfbauer, G., Li, R., Tu, A.Y., mosome 1) as well as Cq3 (chromosome 3) are syntenic to Churchill, G.A. and Paigen, B.J. 2004a. Quantitative trait locus human chromosome 1q21-1q23, to which a familial com- mapping of genes that regulate phospholipids transfer activity bined hyperlipidemia (FCHL) trait has been mapped in in SM/J and NZB/B1NJ inbred mice. Arterioscler. Thromb. Finnish, German, Chinese and Mexican families [4, 13–15]. Vasc. Biol. 24: 155–160. Because the FCHL phenotype is genetically heterogeneous 6. Korstanje, R., Li, R., Howard, T., Kelmenson, P., Marshall, J., Paigen, B. and Churchill, G. 2004b. Influence of sex and diet CHOLESTEROL AND PHOSPHOLIPID QTL 309

on quantitative trait loci for HDL-cholesterol levels in a SM/J demia is associated with upstream transcription factor 1 by NZB/B1NJ intercross population. J. Lipid Res. 45: 881– (USF1). Nat. Genet. 36: 371–376. 888. 15. Pajukanta, P., Nuotio, I., Terwilliger, J.D., Porkka, K.V., Yli- 7. Lander, E.S., Green, P., Abrahamson, J., Barlow, A., Daly, talo, K., Pihlajamaki, J., Suomalainen, A.J., Syvanen, A.C., M.J., Lincoln, S.E. and Newberg, L. 1987. MAPMAKER: an Lehtimaki, T., Viikari, J.S., Laakso, M., Taskinen, M.R., interactive computer package for constructing primary genetic Ehnholm, C. and Peltonen, L. 1998. Linkage of familial com- linkage maps of experimental and natural populations. Genom- bined hyperlipidaemia to chromosome 1q21–q23. Nat. Genet. ics 1: 174–181. 18: 369–373. 8. Lander, E. and Kruglyak, L. 1995. Genetic dissection of com- 16. Ritter, M., Buechler, C., Boettcher, A., Barlage, S., Schmitz- plex traits: guidelines for interpreting and reporting linkage Madry, A., Orsó, E., Bared, S.M., Schmiedeknecht, G., Baehr, results. Nat. Genet. 11: 241–247. C.H., Fricker, G. and Schmitz, G. 2002. Cloning and character- 9. Machleder, D., Ivandic, B., Welch, C., Castellani, L., Reue, K. ization of a novel apolipoprotein A-I binding protein, AI-BP, and Lusis, A.J. 1997. Complex genetic control of HDL levels secreted by cells of the kidney proximal tubules in response to in mice in response to an atherogenic diet. J. Clin. Invest. 99: HDL or Apo A-I. Genomics 79: 693–702. 1406–1419. 17. Shifman, S. and Darvasi, A. 2005. Mouse inbred strain 10. Manly, K.F., Cudmore, R.H. Jr. and Meer, J.M. 2001. Map sequence information and yin-yang crosses for quantitative Manager QTX, cross-platform software for genetic mapping. trait locus fine mapping. Genetics 169: 849–854. Mamm. Genome 12: 930–932. 18. Shike, T., Hirose, S., Kobayashi, M., Funabiki, K., Shirai, T. 11. Mehrabian, M., Castellani, L.W., Wen, P.-Z., Wong, J., Rith- and Tomino, Y. 2001. Susceptibility and negative epistatic loci aporn, T., Hama, S.Y., Hough, G.P., Johnson, D., Albers, J.J., contributing to type 2 diabetes and related phenotypes in a KK/ Mottino, G.A., Frank, J.S., Navab, M., Fogelman, A.M. and Ta mouse model. Diabetes 50: 1943–1948. Lusis, A.J. 2000. Genetic control of HDL levels and composi- 19. Suto, J., Matsuura, S., Yamanaka, H. and Sekikawa, K. 1999. tion in an interspecific mouse cross (CAST/Ei × C57BL/6J). J. Quantitative trait loci that regulate plasma lipid concentration Lipid Res. 41: 1936–1946. in hereditary obese KK and KK-Ay mice. Biochim. Biophys. 12. Murphy, W.J., Frönicke, L., O’Brien, S.J. and Stanyon, R. Acta 1453: 385–395. 2003. The origin of human chromosome 1 and its homologs in 20. Suto, J. and Sekikawa, K. 2003. Quantitative trait locus analy-

placental mammals. Genome Res. 13: 1880–1888. sis of plasma cholesterol and triglyceride levels in KK × RR F2 13. Pajukanta, P., Bodnar, J.S., Sallinen, R., Chu, M., Airaksinen, mice. Biochem. Genet. 41: 325–341. T., Xiao, Q., Castellani, L.W., Sheth, S.S., Wessman, M., Palo- 21. Suto, J., Takahashi, Y. and Sekikawa, K. 2004. Quantitative tie, A., Sinsheimer, J.S., Demant, P., Lusis, A.J. and Peltonen, trait locus analysis of plasma cholesterol and triglyceride levels

L. 2001. Fine mapping of Hyplip1 and the human homolog, a in C57BL/6J × RR F2 mice. Biochem. Genet. 42: 347–363. potential locus for FCHL. Mamm. Genome 12: 238–245. 22. Suto, J. 2005. Apolipoprotein gene polymorphisms as cause of 14. Pajukanta, P., Lilja, H.E., Sinsheimer, J.S., Cantor, R.M., cholesterol QTLs in mice. J. Vet. Med. Sci. 67: 583–589. Lusis, A.J., Gentile, M., Duan, X.J., Soro-Paavonen, A., Nauk- 23. Wang, X. and Paigen, B. 2002. Quantitative trait loci and can- karinen, J., Saarela, J., Laakso, M., Ehnholm, C., Taskinen, didate genes regulating HDL cholesterol: a murine chromo- M.R. and Peltonen, L. 2004. Familial combined hyperlipi- some Map. Arterioscler. Thromb. Vasc. Biol. 22: 1390–1401.