Selective Genotyping With Epistasis Can Be Utilized for a Major Quantitative Trait Locus Mapping in Hypertension in Rats

Corresponding author: Yoichi Ohno, Department of Internal Medicine, TEPCO Hospital, 9-2 Shinanomachi, Shinjuku-ku, Tokyo 160, Japan., t0799791{at}pmail.tepco.co.jp (E-mail)

Communicating editor: S. TAVARÉ

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Epistasis used to be considered an obstacle in mapping quantitative trait loci (QTL) despite its significance. Numerous epistases have proved to be involved in quantitative genetics. We established a backcross model that demonstrates a major QTL for hypertension (Ht). Seventy-eight backcrossed rats (BC), derived from spontaneously hypertensive rats (SHR) and normotensive Fischer 344 rats, showed bimodal distribution of systolic blood pressure (BP) values and a phenotypic segregation ratio consistent with 1:1. In this backcross analysis, sarco(endo)plasmic reticulum Ca²⁺-dependent ATPase (Serca) II heterozygotes showed widespread bimodality in frequency distribution of BP values and obviously demonstrated Ht. First, in genome-wide screening, Mapmaker/QTL analysis mapped Ht at a locus between D1Mgh8 and D1Mit4 near Sa in all 78 BC. The peak logarithm of the odds (LOD) score reached 5.3. Second, Serca II heterozygous and homozygous BC were analyzed separately using Mapmaker/QTL. In the 35 Serca II heterozygous BC, the peak LOD score was 3.8 at the same locus whereas it did not reach statistical significance in the 43 Serca II homozygotes. Third, to map Ht efficiently, we selected 18 Serca II heterozygous BC with 9 highest and 9 lowest BP values. In these 18 BC, the peak LOD score reached 8.1. In 17 of the 18, D1Mgh8 genotypes (homo or hetero) qualitatively cosegregated with BP phenotypes (high or low) (P < 0.0001, by chi-square analysis). In conclusion, selective genotyping with epistasis can be utilized for a major QTL mapping near Sa on chromosome 1 in SHR.

EPISTASIS (or interaction deviation) occurs when the combined effect of two or more genes on a phenotype cannot have been predicted as the sum of their separate effects (FISHER 1918 ; FRANKEL and SCHORK 1996 ). While researchers have accumulated molecular evidence of interacting genes, our knowledge of how epistatic genes influence quantitative phenotypes remains incomplete because of the complexities of studying quantitative traits (LI et al. 1997 ). Although epistasis is known to contribute a great deal to the expression of quantitative traits, quantitative trait locus (QTL) mapping efforts have scarcely dealt with the issues. It is well known that QTL that influence blood pressure (BP) in the progeny of a cross involving a hypertensive strain and a contrasting strain do not always influence BP in progeny derived from that same hypertensive strain and a different contrasting strain. This suggests that the QTL detected in the original cross may interact with alleles on the genome of the second contrasting strain (SCHORK et al. 1996 ).

We investigated the polygenic quantitative characteristics of hypertension in an experimental model of genetic hypertension, the spontaneously hypertensive rat (SHR). Tanase et al. performed a number of breeding studies almost 30 years ago using SHR and normotensive strains such as Wistar-Kyoto rats (WKY), Wistar rats, and Donryu rats (DRY) (TANASE et al. 1970 ; TANASE 1979 ). Although the individual effects of QTL appeared to be modest in F₂ and backcross analyses derived from SHR and WKY or derived from SHR and Wistar rats, F₂, backcross, and successive breedings of SHR and DRY suggested the existence of a putative single major gene locus. Bimodal distribution of the BP values in the backcrossed rats (BC) derived from SHR and DRY supported the existence of a putative single major gene locus. We have recently succeeded again in approximating a polygenic model to a monogenetic form by mating SHR to Fischer 344 normotensive rats (F344) (OHNO et al. 1997 ). Bimodality in BP values of BC demonstrated an inferred single major gene locus or a major QTL (Ht).

The regulation of intracellular Ca²⁺ ([Ca²⁺]_i) is a major factor in maintaining vascular smooth muscle tone (MORGAN 1987 ). It has been postulated that in hypertension an inherited membrane defect in vascular smooth muscle cells causes increased [Ca²⁺]_i. The sarco(endo)plasmic reticulum Ca²⁺-dependent ATPase (Serca) II gene is an important candidate in abnormal intracellular calcium homeostasis. The Serca II genotypes cosegegated with [Ca²⁺]_i and the Serca II gene locus (Serca 2) demonstrated a potential to raise BP as a minor gene (OHNO et al. 1996A , OHNO et al. 1996B ). Distribution patterns on scattergrams comparing BP and [Ca²⁺]_i appeared to differ according to the Serca II genotypes. The Serca II heterozygotes revealed two clusters and demonstrated Ht.

It has been estimated that a single monogenic trait can easily be mapped with 40 informative meioses (equivalent to 20 F₂ intercross rats) using markers at 20-cM intervals (LANDER and SCHORK 1994 ; BONYADI et al. 1997 ). Assuming that the BP difference between SHR and F344 is five environmental standard deviations and that the number of effective factors is one, the number of backcrossed progeny to genotype could be reduced to ~20 with a traditional cosegregation approach and to ~15 with interval mapping using markers at 20-cM intervals (LANDER and BOTSTEIN 1989 ). In this study, we attempted to map Ht efficiently utilizing epistasis between Ht and Serca2. First, quantitative linkage analysis was performed using Mapmaker/QTL on all 78 BC. Second, 35 Serca II heterozygotes and 43 Serca II homozygotes were analyzed separately. Third, 18 selected Serca II heterozygotes with extreme BP values were analyzed. The results of these three steps were compared with those of traditional cosegregation analysis to determine whether the homozygous (SHR) or heterozygous (F₁) genotypes cosegregated with the BP phenotypes (high or low) in the selected 18 BC. Hereafter, we will discuss how epistasis can be utilized in QTL mapping.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Animals:
SHR and F344 were maintained at the Laboratory of Animal Science and Toxicology, Sankyo Co. Ltd., Shizuoka, Japan. SHR were bred in Kyoto until the F₂₀ generation. F₁ females from SHR and F344 parents were backcrossed to male SHR. Recent genotyping data from the whole genome revealed that F344 had been mistakenly labeled as DRY in our earlier studies (OHNO et al. 1996A , OHNO et al. 1996B , OHNO et al. 1997 ). This strain is predominantly known as DRY (INNES et al. 1998 ; WHITEHEAD INSTITUTE/MIT CENTER FOR GENOME RESEARCH 1998), but is different from the former Donryu strain that Tanase et al. reported 30 years ago (TANASE et al. 1970 ; TANASE 1979 ) and from the Donryu strain that is now commercially available from Charles River Japan. The animals were housed under controlled conditions (temperature, 21 ± 1°; humidity 60 ± 10%; 12-hr-light, 12-hr-dark cycle) and fed standard rat chow and tap water ad libitum. Systolic BP was measured by the tail-cuff method (Natume KN210, Tokyo) at 10, 13, and 15 weeks of age; each pressure value was obtained by averaging at least five individual readings and the mean values of these averages were then used for statistical analysis. Systolic BP by the tail-cuff method was shown to be strongly correlated with conscious intra-arterial BP in preliminary experiments (r = 0.93, P < 0.001). Seventy-eight BC were analyzed completely.

Genotyping of the sarco(endo)plasmic reticulum Ca²⁺-dependent ATPase (Serca) II gene and satellite markers:
The Serca II gene was genotyped by restriction fragment length polymorphisms as we previously described (OHNO et al. 1996A , OHNO et al. 1996B ). Briefly, the genomic DNA was amplified by standard polymerase chain reaction (PCR). The reaction products were digested with HindIII or SauI, and fragments were separated by electrophoresis on 2% agarose gels.

A set of highly polymorphic microsatellite markers (simple sequence length polymorphisms) was selected to be spaced at 20-cM intervals, which spanned the entire genome (WHITEHEAD INSTITUTE/MIT CENTER FOR GENOME RESEARCH 1998). The molecular markers used in the current study consisted primarily of microsatellite polymorphisms typeable by PCR. The PCR primers for typing the microsatellite markers of JACOB et al. 1995 (WHITEHEAD INSTITUTE/MIT CENTER FOR GENOME RESEARCH 1998) were obtained from Research Genetics (Huntsville, AL). PCR primers for typing other microsatellites were synthesized in the ABI DNA synthesizer from previously published sequences of SERIKAWA et al. 1992 and NABIKA et al. 1997 . PCR products were electrophoresed on acrylamide gels and stained by ethidium bromide. The names of the marker loci were expressed in italics as DiMghj, DiMitj, and DiRatj (i and j express rat chromosome number and marker number, respectively) or as Pkcef11 (the protein kinase C gene locus), Sa (the Sa gene locus), Serca2 (the Serca II locus), Lsn (the leukosianine gene locus), and Enacb (the epithelial sodium channel type B gene locus; abbreviations are for names of the molecules). BC were genotyped as homozygotes (=SHR) or heterozygotes (=F₁). All markers were scored twice by independent, blinded observers.

Quantitative linkage analysis using Mapmaker/QTL:
In all BC, a major QTL mapping was analyzed by MAPMAKER/Exp 3.0 with an error detection procedure and MAPMAKER/QTL 1.1b obtained from Dr. Eric Lander (Whitehead Institute, Cambridge, MA; LANDER et al. 1987 ; LANDER and BOTSTEIN 1989 ; LINCOLN and LANDER 1992 ). To minimize -error, a P value of 0.0034 and/or a logarithm of the odds (LOD) score of 1.9 and a P value of 0.00010 and/or a LOD score of 3.3 were used as thresholds for suggestive and significant linkage, respectively, according to recently published, conservative suggestions (LANDER and KRUGLYAK 1995 ).

Classification and selection of the BC:
Serca II heterozygous BC and homozygous BC were analyzed separately by MAPMAKER/QTL 1.1b using the framework of the marker loci obtained by the MAPMAKER/Exp 3.0 in all BC.

We then divided BC into quartiles according to bimodal BP distribution. BC in the highest quartile were considered to be Ht homozygotes and may have had other minor hypertensive genes or modifying genes that interact with Ht. BC in the lowest quartile were considered to be Ht heterozygotes and may also have had minor or modifying hypotensive factors. The strategy of selective genotyping of the extreme progeny substantially increases efficiency in QTL mapping (LANDER and BOTSTEIN 1989 ). Eighteen of 35 Serca II heterozygous BC were selected: 9 with the highest BP values and 9 with the lowest. The 18 selected Serca II heterozygous BC were analyzed with Mapmaker/QTL 1.1b using the framework of the marker loci obtained by the MAPMAKER/Exp 3.0 in all BC. The same QTL analyses were performed in the other 18 selected Serca II homozygous BC using the same extreme BP criteria and in the 18 selected BC according to their BP values without epistasis.

Statistical analysis:
Data are reported as mean ±SD. Comparisons were performed by analysis of variance (ANOVA) followed by Scheffe's F-test to evaluate the differences in BP between the genotyped groups. Interaction between Ht and Serca2 in BP levels was analyzed by two-way ANOVA. Traditional segregation analysis was performed to determine whether the homozygous (SHR) or heterozygous (F₁) genotypes cosegregated with the BP phenotypes (high or low) in the selected 18 Serca II heterozygous BC at 200 microsatellite loci using chi-square analysis. When the coincidence rate of genotypes and phenotypes is >13/18 or <5/18, the probability is estimated as <0.05. These statistical computations were performed with the SPSS statistical program (SPSS, Chicago). A level of P < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Bimodality in BP in the BC:
BP was significantly higher in SHR than in F₁ rats and F344 (Fig 1). BP values showed a bimodal distribution in BC (n = 78) (Fig 2A). The bimodal distribution and the segregation ratio were consistent with the existence of a hypothetical single major gene (Ht) for hypertension. The Serca II heterozygous BC revealed widespread bimodal frequency distribution of BP values (Fig 2B). These findings suggest that epistatic interaction enhanced the effect of Ht in the Serca II heterozygotes.

View larger version (17K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Systolic blood pressure among parental strains. Values are means ±SD, SHR vs. F344, P < 0.0001, SHR vs. F1, P < 0.0001, and F1 vs. F344, P < 0.01 by ANOVA.

View larger version (17K): [in this window] [in a new window] [Download PPT slide]   Figure 2. (A) Frequency distribution of systolic blood pressure values in 78 BC. (B) Frequency distribution of systolic blood pressure values in 35 Serca II heterozygous BC. Solid bars mean the 18 selected Serca II heterozygotes and hatched bars mean the unselected Serca II heterozygotes. (C) Frequency distribution of systolic blood pressure values in 43 Serca II homozygous BC.

Quantitative mapping of Ht:
In all BC, the peak LOD score reached 5.3 between D1Mgh8 and D1Mit4 on chromosome 1 (Fig 3B). The distances from the locus of the peak LOD score were 4.0, 2.7, 3.9, and 4.9 cM apart from D1Mgh8, D1Mit4, Pkcef11, and Sa that were mapped at the same locus as Enacb, respectively (Fig 3A and Fig B). The chromosomal region that showed suggestive and significant linkage spanned a 62-cM interval and a 28-cM interval, respectively. This QTL explained 28.3% of the total variance. Systolic BP was 10 mm Hg higher in the D1Mgh8 homozygotes than in the heterozygotes (166 ± 9 vs. 156 ± 8, P < 0.0001 by ANOVA) (Fig 4A). The second-highest LOD score was observed at the locus between D2Mgh14 and D2Mgh15 on chromosome 2 in the total genome screening. However, the peak LOD score on chromosome 2 was 1.4 and did not satisfy the criteria of suggestive linkage. Therefore, the QTL near D1Mgh8 on chromosome 1 clearly proved to be Ht.

View larger version (59K): [in this window] [in a new window] [Download PPT slide]   Figure 3. (A) Mapping an inferred major gene locus (Ht) near Sa on chromosome 1. Numbers between the markers represent Kosambi distances determined by multipoint linkage analysis calculated by MAPMAKER/Exp 3.0. (B) QTL mapping in 78 BC on chromosome 1. The peak LOD score reached 5.3 between D1Mgh8 and D1Mit4. The vertical dotted and broken lines indicate the thresholds for suggestive and significant linkages of the LOD scores (1.9 and 3.3), respectively. (C) QTL mapping in 35 Serca II heterozygous BC and 43 homozygous BC. The solid and shaded lines indicate the LOD score of Serca II heterozygous BC and homozygous BC, respectively. The peak LOD score reached 3.8 between D1Mgh8 and D1Mit4 in Serca II heterozygotes. (D) QTL mapping in the 18 selected Serca II heterozygous and homozygous BC with extreme BP values and in the 18 BC with extreme BP, selected without epistasis. The solid, shaded, and dotted lines indicate the LOD score of the selected Serca II heterozygous BC, homozygous BC, and selected BC without epistasis, respectively. The peak LOD scores reached 8.1 and 9.2 between D1Mgh8 and D1Mit4 in Serca II heterozygotes and selected BC without epistasis, respectively.

View larger version (26K): [in this window] [in a new window] [Download PPT slide]   Figure 4. (A) Blood pressure difference according to the D1Mgh8 genotypes. The blood pressure level was higher in D1Mgh8 homozygotes than in D1Mgh8 heterozygotes (166 ± 9 vs. 156 ± 8, P < 0.0001). (B) Blood pressure difference according to the D1Mgh8 and Serca II genotypes. The effect of D1Mgh8 on blood pressure was enhanced as much as 15 mm Hg in the Serca II heterozygotes (168 ± 10 vs. 153 ± 9, P < 0.0001), while it was attenuated in the Serca II homozygotes (165 ± 8 vs. 159 ± 7, P < 0.16). Interaction between D1Mgh8 and Serca2 was confirmed by two-way ANOVA (P = 0.017).

In 35 Serca II heterozygous BC, the peak LOD score reached 3.8 between D1Mgh8 and D1Mit4 on chromosome 1, demonstrating significant linkage (Fig 3C). This QTL explained 42.3% of the total variance in the Serca II heterozygotes. On the other hand, in 43 Serca II homozygotes, the peak LOD score was 2.1 at the locus between Fgfrb2 and Igf2 on chromosome 1, showing suggestive linkage. The LOD score at the locus between D1Mgh8 and D1Mit4 was 1.6. Therefore, the effect of the QTL near D1Mgh8 or Sa was remarkable only in Serca II heterozygotes.

In the 18 selected Serca II heterozygous BC with extreme BP values, the peak LOD score reached 8.1 between D1Mgh8 and D1Mit4 on chromosome 1 (Fig 3D). The second-highest LOD score of 1.5 was observed on chromosome 2 and did not indicate the suggestive level. In the 18 selected Serca II homozygous BC with extreme BP values, the peak LOD score was 2.1 at the locus between D1Mit9 and D1Rat24 on chromosome 1, showing suggestive linkage. The LOD score at the locus between D1Mgh8 and D1Mit4 was 1.3. In the 18 BC with extreme BP, selected without epistasis, 15 were Serca II heterozygotes. The peak LOD score reached 9.2 between D1Mgh8 and D1Mit4 on chromosome 1. However, two other peaks were demonstrated ~20 cM and 30 cM apart from Ht.

Qualitative mapping Ht:
In 17 of the 18 selected Serca II heterozygous BC, D1Mgh8 genotypes (homo or hetero) on chromosome 1 qualitatively cosegregated with BP phenotypes (high or low; P < 0.0001, by chi-square analysis; Fig 5). The second-highest cosegregation (14/18) was observed at D2Mgh15 on chromosome 2 (P = 0.018). These results were completely consistent with those of Mapmaker/QTL.

View larger version (54K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Coincidence rate of the genotypes (homo- or hetero-) and the blood pressure phenotypes (high or low) in the 18 selected Serca II heterozygous BC. The vertical broken line indicates the significance of P = 0.05 by chi-square analysis.

Epistasis between Ht and Serca2:
Interaction between D1Mgh8 and Serca2 was confirmed by two-way ANOVA (P = 0.017). The effect of D1Mgh8 on BP was enhanced as much as 15 mm Hg in the Serca II heterozygotes (168 ± 10 vs. 153 ± 9, P < 0.0001), whereas it was attenuated in the Serca II homozygotes (165 ± 8 vs. 159 ± 7, P = 0.16) (Fig 4B). These findings suggest that Serca2 was a Ht modifier.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mapping Ht:
In this study, the QTL near D1Mgh8 or Sa on chromosome 1 exhibited the highest probability for the inferred major gene locus (Ht) in the total genome screening. The peak LOD score reached as much as 5.3 at the locus between D1Mgh8 and D1Mit4 in all BC according to QTL analysis using Mapmaker/QTL. The QTL explained 28.3% of the variance. Both in 35 Serca II heterozygous BC and in the 18 selected Serca II heterozygotes with extreme BP values, the same chromosomal region on chromosome 1 revealed the highest LOD score in the genome screening. In traditional cosegregation analysis, D1Mgh8 showed the highest coincidence rate (17/18, P < 0.0001), and only D1Mgh8 demonstrated the predicted cosegregation considered to be Ht. The second possible locus was located near D2Mgh15 on chromosome 2, with borderline significance in its coincidence rate. The systolic BP level of D1Mgh8 homozygotes was 10 mm Hg higher than that of D1Mgh8 heterozygotes. These studies provide compelling evidence that there is a major QTL near D1Mgh8 influencing susceptibility to hypertension.

Recent F₂ linkage studies using SHR have suggested that the BP QTL may exist on chromosomes 1, 2, 3, 4, 5, 7, 8, 10, 13, 16, 18, 19, 20, and X (LINDPAINTNER et al. 1993 ; SAMANI et al. 1993 , SAMANI et al. 1996 ; LODWICK et al. 1995 ; CLARK et al. 1996 ; SCHORK et al. 1996 ; INNES et al. 1998 ). However, the individual contribution of each QTL has not been clearly defined. Several recent studies have found significant linkage around Sa on chromosome 1 (LINDPAINTNER et al. 1993 ; NARA et al. 1993 ; SAMANI et al. 1993 ; INNES et al. 1998 ). INNES et al. 1998 report the results of their F₂ linkage study in mating a SHR-Melbourne substrain obtained from the National Institutes of Health to the same normotensive control used in our study. The QTL for BP between D1Mgh9 and D1Mit3 on chromosome 1 (Map-1) is the identical chromosomal region of Ht and accounts for 21.2% of the total variance in their study. Their findings are consistent with our results that the QTL near D1Mgh8 is a major candidate for Ht in the cross derived from SHR and F344. They also show another QTL for left ventricular mass independent from BP near D2Mgh15 on chromosome 2. In this locus, we found qualitative borderline significance in cosegregation with BP phenotype, whereas they detected no linkage to BP. The differences in the linkages to BP of chromosome 2 between the two studies may be related to different methods of BP measurements or to the differences between SHR substrains (CLARK et al. 1996 ). We measured BP by the tail-cuff method, which may highlight those aspects of BP control relevant to stress and the sympathetic nervous system. The region on chromosome 2 may genetically contribute to left ventricular hypertrophy through stress-related BP elevation rather than through basal nonstimulating BP regulation.

Several important candidate genes such as Sa, Pkc, and Enacb are located in the chromosomal region that contains Ht. The Sa (subtractive clone A) gene was detected by a differential hybridization technique applied to a complementary DNA library prepared from SHR from WKY (IWAI and INAGAMI 1992 ). Thus, Sa is a gene of unknown function with increased expression in the kidney of SHR. SAMANI et al. 1993 report that the Sa genotype is the main determinant of the level of Sa mRNA expression. ISHINAGA et al. 1997 observe the dissociation in Sa gene polymorphisms, Sa mRNA expression levels, and BP levels among substrains of SHR and WKY. They found that the WKY derived from the Izumo colony had an SHR-type genotype and a high level of renal mRNA expression, but did show a normal BP level. They also reported that F₂ linkage analysis mating stroke-prone SHR and WKY derived from the Izumo colony revealed a major QTL near Sa (NARA et al. 1993 ). Therefore, another gene located near Sa may be responsible for Ht. Some recent studies confirm a substantial QTL in the vicinity of Sa by establishing congenic strains (ST. LEZIN et al. 1997 ; FRANTZ et al. 1998 ; IWAI et al. 1998 ), although the exact identity of Ht remains unclear.

Epistasis between Ht and Serca2:
Epistatic interaction between Ht and Serca2 was confirmed by two-way ANOVA. The effect of D1Mgh8 on BP was enhanced in Serca II heterozygotes. Serca2 on chromosome 12 is a Ht modifier.

Genetic divergence has occurred in Serca2 of SHR and WKY substrains. The 2352nd base in the Serca II gene was adenine in SHR-Kyoto, SHR-Charles River, stroke-prone SHR, and WKY-Kyoto, but guanine in SHR-Toho and WKY-Charles River whose BP levels were slightly lower than SHR-Kyoto and WKY-Kyoto, respectively. In the crosses derived from SHR-Kyoto and WKY-Kyoto, Ht may be attenuated because Serca II genotypes may be identical.

We considered the following another reason why Ht emerged in the backcross of SHR and F344 and why it has not emerged in the cross derived from SHR and WKY. Unknown loci, specific to F344, may interact with Ht to augment its effect, or those specific to WKY may interfere with the expression of Ht. At the beginning of this study, we attempted to backcross the former DRY that Tanase used 30 years ago to SHR (TANASE 1979 ). The normotensive control rats we used were F344, not identified as the formerly used DRY, as judged from more than 50 microsatellite marker polymorphisms. Therefore, it may be crucial to mate SHR with a genetically different strain to clarify the existence of a major QTL. It is likely that some loci specific to WKY, especially derived from the National Institutes of Health, interfere with the emergence of Ht.

Selective genotyping by utilizing epistasis is an efficient method to map a major QTL:
Marker-based QTL studies are inherently inefficient at detecting epistasis (XIAO et al. 1995 ). Eshed et al. demonstrated, using nonisogenic lines, that QTL epistasis is a significant component in determining phenotypic values (ESHED and ZAMIR 1996 ). In a plant QTL study, a substantial portion of the genetic variances for complex traits that are inexplicable solely by QTL with relatively large phenotypic effects was proved due to epistasis. Moreover, "main effects" of individual QTL may be somewhat modified as a result of epistatic relationships (LI et al. 1997 ).

Detection of epistasis in QTL analysis may seem difficult; however, once epistasis is inferred, it can be a clue in isolating a major QTL from polygenic traits. The selection of the normotensive controls and the identification of epistasis between Ht and Serca2 play key roles in mapping a major QTL efficiently and accurately. We selected BC with the highest BP quartile and the lowest BP quartile that were considered Ht homozygotes and Ht heterozygotes, respectively. The efforts for genotyping can be reduced to approximately one-fourth (18/78) in ordinary QTL mapping when the known data set for the intervals of maker loci is applied throughout the genome.

In the analysis of the 18 BC with extreme BP, selected without epistasis, a major QTL was also mapped in the same chromosomal region (Fig 3D). However, the additional two peaks were observed in ~20 cM and 30 cM apart from the major QTL. It is speculated that the 18 Serca II heterozygotes selected with epistasis demonstrated a clear peak in QTL analysis because Serca II heterozygotes showed widespread bimodality in frequency distribution of BP values (Fig 2B).

Alternatively, even if we cannot utilize the exact intervals of marker loci, the results of traditional qualitative cosegregation analysis suggested that Ht be mapped near D1Mgh8. The following reasons may explain why we could detect the high significance in cosegregation. First, the widespread bimodality in BP values, especially in Serca II heterozygous BC, suggested the existence of a single major QTL. The selected Serca II heterozygous BC in the highest and in the lowest quartiles should accurately represent to be Ht homozygotes and Ht heterozygotes, respectively. Second, a set of microsatellite markers was selected to be spaced at 20-cM intervals (WHITEHEAD INSTITUTE/MIT CENTER FOR GENOME RESEARCH 1998). The possibility of double recombination between the markers is 0.01. We propose that the emergence of bimodality in quantitative traits utilizing epistasis can be a clue to dissect complex polygenic quantitative traits.

In conclusion, selective genotyping with epistasis can be utilized for a major QTL mapping in SHR. The inferred major gene locus (Ht) is mapped near Sa on chromosome 1. Serca2 on chromosome 12 is a Ht modifier. Further studies should focus on the molecular cloning of Ht and the pathophysiological mechanisms of epistasis between Ht and Serca2. In addition, an important future question is how to identify epistasis systematically in mammalian QTL analysis.

	ACKNOWLEDGMENTS

This work was supported by a research grant for the National Dairy Promotion and Research Association and by Keio Gijuku Academic Development Funds and TEPCO Hospital Funds.

Manuscript received December 28, 1998; Accepted for publication February 28, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BONYADI, M., S. A. RUSHOLME, F. M. COUSINS, H. C. SU, and C. A. BIRON et al., 1997  Mapping of a major genetic modifier of embryonic lethality in TGF beta 1 knockout mice. Nat. Genet. 15:207-211[Medline].

CLARK, J. S., B. JEFFS, A. O. DAVIDSON, W. K. LEE, and N. H. ANDERSON et al., 1996  Quantitative trait loci in genetically hypertensive rats: possible sex specificity. Hypertension 28:898-906[Abstract/Free Full Text].

ESHED, Y. and D. ZAMIR, 1996  Less-than-additive epistatic interactions of quantitative trait loci in tomato. Genetics 143:1807-1817[Abstract].

FISHER, R. A., 1918  The correlation between relatives on the supposition of Mendelian inheritance. Trans. R. Soc. Edinburgh 52:399-433.

FRANKEL, W. N. and N. J. SCHORK, 1996  Who's afraid of epistasis? Nat. Genet. 14:371-373[Medline].

FRANTZ, S. A., M. KAISER, S. M. GARDINER, D. GAUGUIER, and M. VINCENT et al., 1998  Successful isolation of a rat chromosome 1 blood pressure quantitative trait locus in reciprocal congenic strains. Hypertension 32:639-646[Abstract/Free Full Text].

INNES, B. A., M. G. MCLAUGHLIN, M. K. KAPUSCINSKI, H. J. JACOB, and S. B. HARRAP, 1998  Independent genetic susceptibility to cardiac hypertrophy in inherited hypertension. Hypertension 31:741-746[Abstract/Free Full Text].

ISHINAGA, Y., T. NABIKA, T. SHIMADA, J. HIRAOKA, and Y. NARA et al., 1997  Re-evaluation of the SA gene in spontaneously hypertensive and Wistar-Kyoto rats. Clin. Exp. Pharmacol. Physiol. 24:18-22[Medline].

IWAI, N. and T. INAGAMI, 1992  Identification of a candidate gene responsible for the high blood pressure of spontaneously hypertensive rats. J. Hypertens. 10:1155-1157[Medline].

IWAI, N., Y. TSUJITA, and M. KINOSHITA, 1998  Isolation of a chromosome 1 region that contributes to high blood pressure and salt sensitivity. Hypertension 32:636-638[Abstract/Free Full Text].

JACOB, H. J., D. M. BROWN, R. K. BUNKER, M. J. DALY, and V. J. DZAU et al., 1995  A genetic linkage map of the laboratory rat, Rattus norvegicus.. Nat. Genet. 9:63-69[Medline].

LANDER, E. S. and D. BOTSTEIN, 1989  Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121:185-199[Abstract/Free Full Text].

LANDER, E. S. and L. KRUGLYAK, 1995  Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat. Genet. 11:241-247[Medline].

LANDER, E. S. and N. J. SCHORK, 1994  Genetic dissection of complex traits. Science 265:2037-2048[Abstract/Free Full Text].

LANDER, E. S., P. GREEN, J. ABRAHAMSON, A. BARLOW, and M. J. DALY et al., 1987  MAPMAKER: an interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1:174-181[Medline].

LI, Z., S. R. PINSON, W. D. PARK, A. H. PATERSON, and J. W. STANSEL, 1997  Epistasis for three grain yield components in rice (Oryza sativa L.). Genetics 145:453-465[Abstract].

LINCOLN, S. E. and E. S. LANDER, 1992  Systematic detection of errors in genetic linkage data. Genomics 14:604-610[Medline].

LINDPAINTNER, K., P. HILBERT, D. GANTEN, G. B. NADAL, and T. INAGAMI et al., 1993  Molecular genetics of the SA-gene: cosegregation with hypertension and mapping to rat chromosome 1. J. Hypertens. 11:19-23[Medline].

LODWICK, D., M. A. KAISER, J. HARRIS, F. CUMIN, and M. VINCENT et al., 1995  Analysis of the role of angiotensinogen in spontaneous hypertension. Hypertension 25:1245-1251[Abstract/Free Full Text].

MORGAN, K. G., 1987  Calcium and vascular smooth muscle tone. Am. J. Med. 82:9-15[Medline].

NABIKA, T., T. ITO, K. KITADA, T. SERIKAWA, and T. MASHIMO et al., 1997  Comparative mapping of novel simple sequence repeat markers in a hypertension-related region on rat chromosome 1. Mamm. Genome 8:215-217[Medline].

NARA, Y., T. NABIKA, K. IKEDA, M. SAWAMURA, and M. MANO et al., 1993  Basal high blood pressure cosegregates with the loci on chromosome 1 in the F2 generation from crosses between normotensive Wistar Kyoto rats and stroke-prone spontaneously hypertensive rats. Biochem. Biophys. Res. Commun. 194:1344-1351[Medline].

OHNO, Y., K. MATSUO, H. SUZUKI, H. TANASE, and H. IKESHIMA et al., 1996a  Genotypes of sarco(endo)plasmic reticulum Ca²⁺-dependent ATPase II gene in substrains of spontaneously hypertensive rats. J. Hypertens. 14:287-291[Medline].

OHNO, Y., K. MATSUO, H. SUZUKI, H. TANASE, and T. SERIKAWA et al., 1996b  Genetic linkage of the sarco(endo)plasmic reticulum Ca²⁺-dependent ATPase II gene to intracellular Ca²⁺ concentration in the spontaneously hypertensive rat. Biochem. Biophys. Res. Commun. 227:789-793[Medline].

OHNO, Y., K. MATSUO, H. SUZUKI, H. TANASE, and T. TAKANO et al., 1997  Increased intracellular Ca²⁺ is not coinherited with an inferred major gene locus for hypertension (ht) in the spontaneously hypertensive rat. Am. J. Hypertens. 10:282-288[Medline].

SAMANI, N. J., D. LODWICK, M. VINCENT, C. DUBAY, and M. A. KAISER et al., 1993  A gene differentially expressed in the kidney of the spontaneously hypertensive rat cosegregates with increased blood pressure. J. Clin. Invest. 92:1099-1103.

SAMANI, N. J., D. GAUGUIER, M. VINCENT, M. A. KAISER, and M. T. BIHOREAU et al., 1996  Analysis of quantitative trait loci for blood pressure on rat chromosomes 2 and 13. Age-related differences in effect. Hypertension 28:1118-1122[Abstract/Free Full Text].

SCHORK, N. J., S. P. NATH, K. LINDPAINTNER, and H. J. JACOB, 1996  Extensions to quantitative trait locus mapping in experimental organisms. Hypertension 28:1104-1111.

SERIKAWA, T., T. KURAMOTO, P. HILBERT, M. MORI, and J. YAMADA et al., 1992  Rat gene mapping using PCR-analyzed microsatellites. Genetics 131:701-721[Abstract].

ST. LEZIN, E., W. LIU, J. M. WANG, N. WANG, and V. KREN et al., 1997  Genetic isolation of a chromosome 1 region affecting blood pressure in the spontaneously hypertensive rat. Hypertension 30:854-859[Abstract/Free Full Text].

TANASE, H., 1979  Genetic control of blood pressure in spontaneously hypertensive rats (SHR). Exp. Anim. 28:519-530.

TANASE, H., Y. SUZUKI, A. OOSHIMA, Y. YAMORI, and K. OKAMOTO, 1970  Genetic analysis of blood pressure in spontaneously hypertensive rats. Jpn. Circ. J. 34:1197-1212[Medline].

WHITEHEAD INSTITUTE/MIT CENTER FOR GENOME RESEARCH, 1998 Rat Genomic Mapping Project, Data Release4, February.

XIAO, J., J. LI, L. YUAN, and S. D. TANKSLEY, 1995  Dominance is the major genetic basis of heterosis in rice as revealed by QTL analysis using molecular markers. Genetics 140:745-754[Abstract].

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Ohno, Y. Articles by Saruta, T. Search for Related Content PubMed PubMed Citation Articles by Ohno, Y. Articles by Saruta, T.