QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Data Supplement All Versions of this Article: jpet.106.103572v1 319/1/299    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map 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 HighWire Citing Articles via Google Scholar Google Scholar Articles by Bennett, B. Articles by Johnson, T. E. Search for Related Content PubMed PubMed Citation Articles by Bennett, B. Articles by Johnson, T. E.

BEHAVIORAL PHARMACOLOGY

Confirmation and Fine Mapping of Ethanol Sensitivity Quantitative Trait Loci, and Candidate Gene Testing in the LXS Recombinant Inbred Mice

Institute for Behavioral Genetics, Boulder, Colorado (B.B., P.C.-L., T.E.J.); and Department of Pharmacology, University of Colorado at Denver and Health Sciences Center at Fitzsimons, Aurora, Colorado (N.R.Z.)

Received for publication February 27, 2006
Accepted June 26, 2006.

	Abstract

Top Abstract Materials and Methods Results Discussion References

 
In previous studies, we have mapped quantitative trait loci (QTLs) for hypnotic sensitivity to ethanol using a small recombinant inbred (RI) panel and a large F₂ backcross. Alcohol sensitivity is a major predictor of long-term risk for alcoholism. We remapped hypnotic sensitivity using a new set of 75 RI strains, the LXS, derived from Inbred Long Sleep and Inbred Short Sleep strains. We expected to improve mapping resolution in the QTL regions and to identify novel QTLs for loss of the righting reflex due to ethanol. We used three common mapping algorithms (R/qtl, QTL Cartographer, and WebQTL) to map QTLs in the LXS, and we compared the results. Most mapping studies use only a single algorithm, an approach that may result in failure to identify minor QTLs. We confirmed most of our previously reported QTLs, although one major QTL from earlier work (Lore2) failed to replicate, possibly because it represented multiple linked genes separated by recombination in the RI strains. We also report narrowed confidence intervals, based on mapping with a new genetic resource of more than 4000 polymorphic single-nucleotide polymorphism markers. These narrowed confidence intervals will facilitate candidate gene identification and assessment of overlap with human regions specifying risk for alcoholism. Finally, we present an approach for using these RI strains to assess evidence for candidate genes in the narrowed intervals, and we apply this method to a strong candidate, the serotonin transporter.

In previous work, we mapped QTLs for sensitivity to an intoxicating dose of ethanol [loss of righting due to ethanol (LORE), Lores 1-5], using a small panel of 25 RI strains (LSXSS) derived from the noninbred Long Sleep (LS) and Short Sleep (SS) (Markel et al., 1996) mice, and subsequently, a large F₂ intercross between ILS and ISS mice of more than 1000 mice (Markel et al., 1997). The latter cross confirmed four of the putative QTLs from the LSXSS and identified three novel regions. We confirmed and captured four of the major QTLs for this trait, on chromosomes 1, 2, 11, and 15, in reciprocal congenic strains (Bennett et al., 2002b), and we narrowed the interval surrounding the QTL on two chromosomes to less than 12 megabases (Mb) (Bennett et al., 2002a).

There are compelling reasons to use a large RI set such as the recently created LXS set of 75 strains (Williams et al., 2004) for genetic mapping (Belknap, 1998; Williams et al., 2004; Chesler et al., 2005): 1) a 4-fold map expansion provides improved map resolution; 2) trait heritabilities are higher in RI than in F₂ or backcross animals from equivalent crosses, allowing the mapping of traits with lower narrow-sense heritability (h²); 3) genotyping is done only once; and 4) phenotypic data from many different laboratories and experiments can be combined to derive genetic correlations and assess gene-environment (GXE) interaction. The study of GXE interactions is uniquely suited to RI sets and is likely to be enormously important in understanding and dissecting complex diseases (Churchill et al., 2004).

Haughey et al. (2005) used the LXS, along with binding assays and molecular evidence from ILS and ISS mice, to identify the norepinephrine transporter (NET) gene on chr 8 as a candidate for a LORE QTL. Here, we present new haplotype data further supporting this candidate.

The serotonin transporter (SERT) is an attractive candidate for ethanol sensitivity in both mice and humans. In a human population, a polymorphism in SERT resulted in a GXE interaction such that stressful events elicited more serious responses in one genotype (Caspi et al., 2003). Mice with one or two null alleles show more anxiety and greater increases in stress hormones following stressful stimuli (Murphy et al., 2001). The gene is located at the peak of the Lore4 QTL on chr 11 (Markel et al., 1997) at 76.7 Mb. The SERT inhibitor fluoxetine has differential effects in ILS and ISS mice on MK-801-induced activity, indicating that 5-hydroxytryptamine (serotonin) affects activity induced by N-methyl-D-aspartate receptor blockade in these mice (Hanania et al., 2002). An inverse relationship between ethanol consumption and serotonin level has been observed in mice (Kelai et al., 2003), and treatment with selective serotonin reuptake inhibitors has been reported to decrease drinking in alcohol-dependent humans (Lejoyeux, 1996). Here, we tested and ultimately rejected SERT as a candidate for Lore4.

All LORE testing in the LXS panel, reported here, was done in three independent cohorts over a 1-year period. This experimental design was initially developed to spread out specific environmental effects on phenotypic variability to reduce the impact stemming from a single cohort. Nonetheless, environmental effects were large, and some variability in QTLs among cohorts occurred. Here, we report QTLs based on analyses of the individual cohorts and the combined data set. Although the multiple cohort design did not minimize the effect of nonspecific environmental variation, it did allow us to use the second cohort to estimate effect size more accurately (Bennett and Carosone-Link, 2006). Several QTLs that replicated across cohorts, and previous studies, attained high combined significance levels, confirming them as important genomic regions for follow-up work in gene identification.

	Materials and Methods

Top Abstract Materials and Methods Results Discussion References

 
Animals. The LXS RI panel was generated in the specific pathogen-free (SPF) facility at the Institute for Behavioral Genetics (Boulder, CO) from crosses between the ILS and ISS mice (Bennett et al., 2002b; Williams et al., 2004). The number of consecutive filial matings currently ranges from F₂₇ to F₃₃, with 75 strains still extant. At 25 days of age, mice were weaned, tail clipped, and transferred to male-only or female-only cages, two to five per cage. Mice were maintained on a 12-h light/dark cycle and were given food and water ad libitum (Breeder Diet, 9% fat; Harlan, Indianapolis, IN). All procedures followed guidelines developed by the National Institutes of Health and the Institutional Animal Care and Use Committee.

Loss of Righting Reflex Due to Ethanol. All testing was conducted in the SPF facility. Mice were tested for alcohol sensitivity first at 55 to 65 days (trial 1) and again 7 days later (trial 2) by i.p. injection of a 4.1-g/kg dose of ethanol [20% (w/v) solution in saline]. Blood ethanol concentration (BEC) at awakening was determined using a spectrophotometric assay (Smolen and Smolen, 1989). All testing was done between 9:00 AM and 2:00 PM, during the light cycle. If the mouse still had not lost the ability to right itself, after 10 min, the injection was considered faulty, and a retest was done 1 week later. Approximately 10% of all mice injected failed to lose the righting response on one occasion; this is generally attributed to a misplaced or leaky injection and is a typical occurrence for i.p. injections (Markel et al., 1995; Crabbe et al., 2005). Blood ethanol concentration was determined in 66% of these animals 10 min postinjection to corroborate this conclusion; these data are presented here. Duration of LORE was determined using the method of McClearn and Kakihana (1981), modified as reported by Markel et al. (1997). In brief, righting response was lost when the mouse could not right itself three times within 1 min. All mapping was done using the mean LORE duration from trials 1 and 2. ISS and ILS mice were injected as controls for environmental variability in LORE.

[³H]Citalopram Binding to SERT. Mice were sacrificed by cervical dislocation. Brain regions of interest (prefrontal and remaining cortex, cerebellum, hippocampus, nucleus accumbens, amygdala, ventral midbrain, and caudate) were dissected from ILS and ISS mice (n = 6-8/strain) and frozen at -80°C. On the day of assay, tissues were homogenized in 30 mM sodium phosphate buffer, pH 7.4, containing 0.32 M sucrose and centrifuged at 20,000g at 4°C for 20 min. The membrane pellets were resuspended in the phosphate-sucrose buffer and incubated in a volume of 0.25 ml with [³H]citalopram (PerkinElmer Life and Analytical Sciences, Boston, MA), unlabeled citalopram, and fluoxetine (Sigma-Aldrich, St. Louis, MO) at room temperature for 1 h. Indirect saturation curves were generated in "remaining cortex" and cerebellum using 4.7 nM [³H]citalopram; concentrations of unlabeled citalopram ranging from 0.03 nM to 1 µM, and 80 µg of protein. Nonspecific binding was defined with 10 µM fluoxetine. In the other brain regions, specific binding was measured with a single, 2-fold higher concentration of [³H]citalopram (9.5 nM ± 10 µM fluoxetine) and 75 µg of protein. The assays were terminated by rapid vacuum filtration over GF/B filters (Brandel Inc., Gaithersburg, MD) and three washes with ice-cold sodium phosphate buffer. The retained radioactivity was measured by liquid scintillation spectrometry. Proteins were determined by the method of Bradford (1976) using bovine serum albumin as the standard. The maximal number of binding sites (B_MAX) and affinity (K_D) were determined from the saturation curves with nonlinear fitting (GraphPad Software Inc., San Diego, CA).

Heritability Determination and Statistical Analyses. Narrow-sense heritability (h²) assesses the proportion of the phenotypic variance (V_P), due to additive genetic variance (V_A). h² is easily determined in an RI study as the variance of the strain means, relative to the total variance, and was calculated separately for sex and cohort by adjusting r² from a one-way analysis of variance for varying sample sizes (Belknap et al., 1996). The 95% confidence limits on h² were determined using the Moriguti-Bulmer procedure for approximating confidence limits (Sokal and Rohlf, 2000). Statistical analyses were done using SPSS for Windows, version 12.0 (SPSS Inc., Chicago, IL).

QTL Mapping and Haplotype Analysis. QTL mapping was done in several stages, to identify loci acting individually and QTLs that interacted, either additively or epistatically, to affect each phenotype. Initial analysis was done with MapManager QTX, version 19 (Manly and Olson, 1999; Manly and Cudmore, 2001; http://www.mapmanager.org/mmQTX.html), to produce a genetic map that was then used as the input file for R/qtl. The genetic map was based on genotype determined previously for the LXS strains (Williams et al., 2004; http://www.genenetwork.org/genotypes/LXS.geno). Strain means for all phenotypes were analyzed using R/qtl (Broman et al., 2003; http://www.biostat.jhsph.edu/~kbroman/qtl/), as described in Bennett et al. (2005). R/qtl was also used to assess sex specificity of QTL regions by mapping on the mean phenotypic difference between males and females in each strain (K. Broman, personal communication). Further mapping was done using a new genetic resource (Wellcome-CTC Mouse Strain SNP Genotype set). This genotype set consists of genotypes for 480 strains, including the full LXS panel, and 13,370 successful SNP assays that are mapped to build 34 of the mouse genome; 4834 of these are polymorphic between ILS and ISS mice (http://www.well.ox.ac.uk/mouse/INBREDS). This genetic map was used in R/qtl and WebQTL (Wang et al., 2003; http://www.webqtl.org/home.html).

Finally, we used several mapping strategies available in Windows QTL Cartographer (Wang et al., 2005; http://statgen.ncsu.edu/qtlcart/WQTLCart.htm). Composite interval mapping adds background loci to simple interval mapping (we specified all other significant and suggestive QTL regions) to remove their effects on the target QTL. We also used the multiple trait analysis option of Cartographer, because the individual LORE phenotypes (sex- and cohort-specific) were correlated traits (Table 1). Accounting for this correlation among traits increases power and reduces sampling variance (Jiang and Zeng, 1995).

We mapped in each of three cohorts individually and in the combined data set. This combined data set was also used to investigate cohort as a main effect, using a General Linear Model in SPSS version 12.0 (SPSS Inc.), with the phenotype as dependent variable and cohort, sex, and strain as independent variables. A combined p value (Sokal and Rohlf, 2000) for each QTL identified in multiple cohorts was also determined.

A haplotype is a group of markers retained as a block. We used SNP markers identified in the Wellcome-CTC Mouse Strain SNP Genotype set, for which ILS, ISS, and the LXS RI mice were genotyped, at an average density of 45 kb, to generate coarse haplotypes for ILS and ISS mice. Much higher resolution is possible using the SNPs for A/J (A), C57BL/6J (B6), and DBA/2J (D2) (the three progenitors to ILS and ISS strains for which complete sequence is available) compiled from Celera Genomics (Rockville, MD), the Perlegen/National Institute of Environmental Health Sciences resequencing project, the Wellcome-CTC SNP Project, and the Mouse Phenome Database (http://aretha.jax.org/pub-cgi/phenome/mpdcgi?rtn=docs/home) archived at GeneNetwork (http://www.genenetwork.org/cgi-bin/beta/snpBrowser.py) to determine strain-specific haplotypes. The latter is the denser map (averaging 1 SNP/4 bp), detailing all SNPs identified by these sequencing efforts; however, our strains of interest are not genotyped for this density. We used the Wellcome SNP set to create a skeleton of markers from the strains fully covered at GeneNetwork. ILS and ISS strains were placed on this map wherever genotypes permitted, and their haplotypes were inferred from comparison with strains with matching flanking patterns. To error-check data base entries, SNP genotypes from the Wellcome set, Mouse Phenome Database, and GeneNetwork were compared. All haplotypes were clearly consistent (data not shown), although not all data bases contain all SNPs.

	Results

Top Abstract Materials and Methods Results Discussion References

 
LORE in the LXS. In total, 952 mice in three cohorts were tested for LORE (Table 1). The third cohort included only 206 mice from 57 strains, providing low power for mapping QTLs; consequently, no individual analyses of this cohort are reported, although we did use these data in pooled analyses. Correlations among all three cohorts, along with heritabilities, are given in Table 1, and they indicate good phenotypic correspondence in all three cohorts. At the time testing began, there were 77 strains extant; three have since been lost. There was no transgression, that is, no strain means for LORE surpass either of the progenitor strains (Fig. 1). This is an optimal situation for genetic mapping of the selected trait (Liu, 1997). LORE in strain 35 (µ = 20.5) was as low as that in ISS (µ = 22.5; p = 0.84), and ILS mice had the longest duration of LORE (µ = 185), significantly greater than the next highest strain (µ = 149; one-tailed p = 0.04). Heritability of LORE, based on the RI strain means in the combined data set, was 57%. Heritability values by sex and cohort all clustered close to this value (Table 1).

View larger version (35K): [in this window] [in a new window]   Fig. 1. Strain distribution for duration of loss of righting postethanol (LORE) in minutes in the LXS and the two parental strains. Data from all cohorts, and both sexes, were combined, because not all strains were tested in each cohort. ILS and ISS strains are shown with striped columns. Error bars represent standard error of the mean.

For all strains but ISS, failure to lose the righting response within 10 min was most likely due to a faulty injection. Mean BEC for these mice was 323 mg/100 ml (Fig. 2A), statistically lower (p < 0.001) than in mice that did lose the righting response (µ = 357 mg/100 ml), although this latter value is BEC at awakening, which will be somewhat lower than BEC at loss of righting. Consequently, data from animals failing to lose the righting response were not used in determining strain mean for LORE. In ISS, a different picture emerged. In a subset of these mice, low BEC undoubtedly followed a misplaced injection (Fig. 2B, columns on the left; µ = 199 mg/100 ml). In a second group, BEC was significantly higher (µ = 417 mg/100 ml; p < 0.001), sufficient to cause LORE in LS (Smolen and Smolen, 1989) and most inbred strains (Crabbe et al., 2005), indicating that selection for resistance to the sedative effect of ethanol in the short sleep mice resulted in a floor effect.

View larger version (21K): [in this window] [in a new window]   Fig. 2. BEC (in milligrams per 100 milliliters) in mice that failed to lose the righting response. A, less sensitive, non-ISS strains were pooled. Mean BEC was significantly less than BEC at awakening in mice in which LORE occurred. B, resistant ISS strain was analyzed separately. Two groups emerged: those consistent with a faulty injection (left), which had significantly lower BEC than those apparently resistant (right) to the sedative effect of the administered dose (4.1 g/kg).

Sex was a significant main effect influencing LORE (p < 0.001). Males had a longer LORE duration (µ = 81.3 ± 1.59 min) than females (µ = 75.3 ± 1.59 min), which agrees well with previous literature. Because this effect was consistently seen in all strains, there was no significant sex by strain interaction (p = 0.23). It is notable that within-cohort correlations always exceeded between-cohort correlations, although most correlations were significant at or below p < 10^-7 (Table 1).

Identifying LORE QTLs. Because within-cohort correlations were larger than between-cohort correlations, cohorts 1 and 2 were initially analyzed separately; subsequently, all data were pooled over cohorts. The QTL region on chr 1 was identified in both cohorts (Table 2). Regions on chr 3, 8, 14, and 18 emerged in one cohort but not the other (Table 2). Pooling over cohorts identified all QTL regions albeit with slightly lower LODs. The number of strains tested showed a suggestive correlation with LOD score (r = 0.34; p = 0.09). No sex-specific QTLs were identified.

All mapping programs gave very similar findings for QTLs (Fig. 3, R/qtl (A) and QTL Cartographer (B); Supplemental Fig. 1, WebQTL; and Supplemental Fig. 2, QTL Cartographer). Figure 3 illustrates discrepancies among the mapping programs. For example, R/qtl (Fig. 3A) identified the chr 3 QTL as significant (p = 0.05), whereas this region is suggestive in QTL Cartographer (Fig. 3B). Likewise, regions on chr 14 and 18 are suggestive in R/qtl (Fig. 3A) but not in QTL Cartographer (Fig. 3B).

View larger version (32K): [in this window] [in a new window]   Fig. 3. Interval mapping results for LORE from R/qtl (A) and QTL Cartographer (B), sexes and cohorts combined.

Replication across Mapping Programs. We compared suggestive (p < 0.63) and significant (p < 0.05) QTLs from interval analysis from all three mapping packages (Table 3; full graphics from WebQTL and multiple trait mapping analyses from QTL Cartographer are provided in Supplemental Figs. 1 and 2, respectively). Sexes were analyzed separately, pooling data over cohorts 1 and 2, because the data are archived in this format in WebQTL. The QTLs on chr 1, 3, and 14 replicated in all algorithms, although significance levels differed. There was a surprising range of significance levels. For example, regions on both chr 1 (males) and 3 (females) approached significance in R/qtl and Cartographer but not in WebQTL (Table 3). However, perusal of the LODs partially resolves the dilemma; all LODs are quite close but the reported p values differ. This disparity is likely due to the fact that WebQTL reports only suggestive, significant, or highly significant cut-offs, whereas in the other programs it is possible to obtain a more precise probability. In addition, WebQTL defaults to 2000 permutations for estimating significance cut-offs, whereas in other programs 1000 permutations were done. In R/qtl and Cartographer, there was little difference between the significance cut-offs obtained using 1000 or 2000 permutations. Despite this variation, general trends over packages were very similar, with the largest disparity in WebQTL. For example, the chr 1 QTL in males had lower LOD scores than females in all programs. For chr 3, the direction of the difference was reversed, with males having higher LODs than females. The peak position and support interval for QTLs on chr 1 and 3 were extremely similar among all mapping programs, although the chr 14 region varied considerably. A suggestive region on chr 18 was identified in males only in R/qtl and suggested in both sexes by a significant GXE interaction determined by multiple trait mapping in Cartographer (Supplemental Fig. 2). A suggestive region on chr 19 was identified by both WebQTL and Cartographer. In all analyses of variance, strain contributed significantly (p < 0.001) to LORE. Although two to three different investigators tested LORE in each cohort, this variable did not have a significant effect on LORE for either sex.

1-LOD Intervals Are Reduced in the LXS. The 1-LOD support intervals in the RI panel (LXS columns) were reduced relative to those obtained in the F₂ map (Table 2), with the exception of the chr 1 region. (And as noted above, the determination of this interval in the F₂ is suspect.) These intervals were reduced even further when the denser SNP map, archived on the WebQTL site, was used (Table 4). On average, using the denser SNP map reduced the 1-LOD support interval 4-fold, relative to the microsatellite map. The range in absolute values was quite remarkable: the chr 14 region showed a support interval of 2.9 Mb in males (down from 40.8 Mb using microsatellites), whereas the chr 3 region, 18.3 Mb by microsatellite mapping, was reduced to approximately 9 Mb using the SNP map.

Composite Interval and GXE Mapping. Controlling for all suggestive and significant background QTLs reported in Table 2 (composite interval mapping in QTL Cartographer) resulted in a large increase in the significance of the QTLs on chr 1 and 3 in both sexes, such that both of these regions showed a highly significant probability of containing a QTL for LORE (p < 0.001 for females and p < 0.01 for males). Although a number of other QTL regions were suggestive by this method, none of them surpassed the significance cut-off determined by 1000 permutations. Including independent sex- and cohort-specific LORE phenotypes in a multiple trait mapping model corroborated both the chr 1 region (p < 0.01) and the chr 8 region (p < 0.05) at significant levels. The multiple trait mapping analysis also provides a likelihood ratio statistic for GXE interaction, because the same phenotype was measured in different environments (i.e., cohorts and sexes). Significant GXE interaction was seen on chr 4, 8, 10, 11, and 15, indicating genomic areas that affected LORE in specific environments (Supplemental Fig. 2).

For each cohort, a model was developed including interacting loci that together accounted for a sizable portion of the phenotypic variance (Table 5, top). These loci were identified from a two-way scan of strain means and locus genotypes in R/qtl and selected on the basis of permutation tests for joint LODs. None of these interactions exceeded the permutation cut-off for epistasis, indicating only additive effects among the loci, which were the same in each cohort (Table 5, bottom). For cohort 1, these three loci, on chr 1, 3, and 8, acting additively, accounted for more than 40% of the phenotypic variance, whereas for cohort 2, this value was 30%.

Replication of LORE QTLs. We asked whether previously identified QTLs from our initial LSXSS RI (27 strains; Markel et al., 1996) or the subsequent ILSXISS F₂ intercross (Markel et al., 1997) would replicate in this panel. Four of the five regions identified in the LXS-replicated regions identified in the LSXSS panel (chr 1, 3, and 18; Table 2) and the ILSXISS F₂ (chr 1, 8, and 18; Table 2). Each replication gave significant combined p values, using Fisher's test (Table 2) (Sokal and Rohlf, 2000), although the chr 1 and 3 regions were more significant by several orders of magnitude. Two QTLs identified from the F₂ genome scan, on chromosomes 11 and 15, replicated in a multiple trait model (which entered mean values for LORE, by sex and in each cohort, as separate independent variables), with a significant GXE interaction. For the chr 11 region, there was a strong GXE interaction (LOD = 2.6), stemming from the fluctuation of the additive effect value, with both sexes in cohort 1 showing an opposite trend from mice in cohort 2. This can be seen in Supplemental Fig. 2. The chr 15 region shows a more complicated picture, which could be due to a similar phenomenon producing a GXE interaction (LOD = 2.2); break up of the QTL into multiple, linked QTLs (as suggested by analysis of interval-specific congenic strains; Bennett et al., 2002a); or a combination of these factors.

SERT Testing. A preliminary analysis in the LXS showed that the 46 strains with an ILS region spanning the serotonin transporter gene (Slc6a4) on chr 11, near the peak of a QTL identified in an F₂ mapping study (Markel et al., 1997), had longer LORE (µ = 82.7) than the 28 strains with ISS genotype (µ = 74.8; one-tailed p = 0.1). Genotypes of microsatellites near the gene were polymorphic between ILS and ISS (data not shown), suggesting Slc6a4 may have been polymorphic and thus a potential candidate for LORE.

The following experiments were done to test the hypothesis that Slc6a4 was a candidate gene underlying some of the difference in LORE between ILS and ISS. Specific binding of [³H]citalopram, a highly selective SERT inhibitor (Hyttel, 1994), to crude membranes prepared from brain regions of interest, was used to determine whether there were any regional differences in the number of SERTs between the ILS and ISS mice. Full saturation curves in cerebral cortex (minus prefrontal cortex) and cerebellum showed a single binding site with high affinity (K_D values of 3 nM) and B_MAX values of 500 fmol/mg protein, but no strain differences. Likewise, specific binding of 9.5 nM [³H]citalopram to membranes prepared from other brain regions showed the regional differences in levels of SERTs (ventral midbrain, 500 fmol/mg protein; nucleus accumbens, 480 fmol/mg protein; amygdala, 480 fmol/mg protein; dorsal striatum, 360 fmol/mg protein; hippocampus, 320 fmol/mg protein; and prefrontal cortex, 230 fmol/mg protein) but no differences between the ILS and ISS mice.

Haplotype analysis suggested that ILS and ISS are not polymorphic through Slc6a4 (boldface in Table 6). Although ILS and ISS were not genotyped in the gene itself, they and all other strains but B6 share the same haplotype for almost 600 kb upstream of the gene. In the three strains completely sequenced (A, B6, and D2), which were ancestral to ILS and ISS, there are two distinct haplotypes over the 107 SNPs in the SERT gene. The B6 is different from the other two strains. The divergence of B6 both upstream (10 SNPs) and in the gene suggests that ILS and ISS do not possess this haplotype but instead share a nonpolymorphic haplotype with the other strains, thus ruling out Slc6a4 as a candidate.

Haplotype analysis of 78 SNPs in the NET gene (Slc6a2) revealed substantial polymorphism (Fig. 4). ILS, D2, and A share one haplotype. B6 has a different haplotype upstream, and in the proximal region of the gene. Through the distal portion of the gene (exon 4 and 11), all strains but ISS share a common haplotype. ISS is B6-like in the proximal region of the gene, but it has a unique SNP pattern in exons 4 and 11.

View larger version (11K): [in this window] [in a new window]   Fig. 4. Three haplotype blocks in Slc6a2 (NET gene). ILS and ISS are polymorphic for all SNPs typed in the gene and 150 kb upstream. ILS, D2, and A share one haplotype, which differs from the B6 haplotype in the proximal region of the gene. ISS is similar to B6 in exon 1 and in upstream and downstream regions. In exons 4 and 11, a unique haplotype is seen in ISS.

	Discussion

Top Abstract Materials and Methods Results Discussion References

Mapping using the combined data resulted in replication of virtually all QTLs from the individual cohort analyses, with slightly lower LOD scores (Table 2) and effect sizes (Bennett and Carosone-Link, 2006). Environmental effects specific to a single cohort were averaged over the unaffected mice in the other cohorts, decreasing the significance of cohort-specific QTLs (Table 2). Thus, this approach may provide more robust identification of QTLs operating across most conditions. A preferred approach involves resampling the pooled data, randomly assigning all values to one of two cohorts: the first cohort for estimating QTL location, and the second cohort for estimating effect size (Melchinger et al., 1998). All but one of the QTLs identified in the original data set replicated in the resampling design with higher LODs and larger effect sizes (Bennett and Carosone-Link, 2006).

Although QTL regions on chr 1 and 3 were significant for one sex but only suggestive for the other sex, there were no completely sex-specific QTLs. This result is not surprising, given the consistent difference between LORE in males and females, reported in a previous study (Markel et al., 1995). This differential sensitivity is due, in part, to differences in body fat and aqueous compartments between the sexes, which alter ethanol distribution (Goldstein, 1983), and slightly higher elimination rates in females (Owens et al., 2002).

The QTL regions on chr 14, 18, and 19 illustrate two issues relating to nonreplicability, which are important to recognize for follow-up confirmation studies. Although the chr 14 region was identified in cohorts 1 and 3 (Table 2), and in all three mapping packages (in at least one sex), the support intervals do not overlap. This ambiguity makes it difficult to pursue this QTL, despite its overlap with a human QTL for ethanol sensitivity (Schuckit et al., 2005). The chr 18 and 19 regions illustrate a different problem. The chr 18 region was identified by R/qtl in cohort 1, and in previous mapping studies (Table 2) as well as by Cartographer. However, WebQTL did not find this region even suggestive, highlighting the importance of using multiple approaches for mapping. R/qtl did not pick up the suggestive region on chr 19, identified in females by WebQTL and Cartographer.

All but one (on chr 14) of the LORE QTLs reported here replicated regions identified in previous mapping populations (Table 2). This replication, assessed by overlapping 1-LOD support intervals, allowed us to combine p values for each QTL. For the regions on chr 1 and 3, these p values were highly significant (p < 0.001). Composite interval mapping also identified these two QTLs as significant, as did the additive model shown in Table 5. In both cohorts analyzed separately, the same three loci, on chr 1, 3, and 8, explained much of the phenotypic variance. One of these intervals, on chr 1, also contains one of 15 genes with significant differences in expression between ILS and ISS in cerebellum (MacLaren et al., 2006) and ventral tegmentum (M. Miles, personal communication). X-ray repair complementing defective repair in Chinese hamster cells (Xrcc5) has a 2.5-fold higher expression in ISS strain and a human ortholog in a genomic region linked to ethanol sensitivity in humans. A study using data from more than 700 individuals collected by the Collaborative Study on the Genetics of Alcoholism (http://www.niaaa.nih.gov/ResearchInformation/ExtramuralResearch/SharedResources/projcoga.htm) identified several markers on chromosome 2 linked to sensitivity in this population. XRCC5, the human ortholog of Xrcc5, maps within 2 Mb of the marker with the highest linkage score (Schuckit et al., 2001).

Both F₂ mapping (Markel et al., 1997) and interval-specific congenic recombinant strain (ISCR) confirmation (Bennett et al., 2002a) identified a slightly more proximal region on chr 1 than did the LXS. Of six ISCR, four carried a proximal region of ILS on an ISS background, whereas two carried the more telomeric ILS region, identified in the LXS. All ISCR showed the same pattern, of increasing LORE; however, larger sample size in the four proximal strains provided stronger support for this region. The LXS results reported here suggest a second, linked QTL for LORE. The abundance of support for the chr 1 and 3 regions: replication in all cohorts, and all mapping programs, as well as significant LOD scores by composite interval mapping, multiple trait mapping for chr 1, and differential expression for Xrcc5, favor these two as worthwhile of follow-up study.

It is clearly desirable to use the dense SNP map (archived on the Wellcome Trust site at http://zeon.well.ox.ac.uk/rmottbin/strains.cgi). LOD scores were somewhat higher (Table 3), but the main advantage to using the SNP genotypes is the large (on average, 72%) reduction in the confidence interval surrounding the QTL. This reduction will facilitate congenic (Bennett et al., 2002b) and ISCR strain (Bennett et al., 2002a) construction.

Two of the previously confirmed QTLs for LORE, on chr 11 and 15 (Markel et al., 1997; Bennett et al., 2002b), were identified in the LXS only in a multiple trait model, which gains increased power from correlations among individual traits. The significant GXE interactions for these regions indicate their specificity to as yet unknown environmental effects. An RI panel of 75 strains can reliably detect QTLs accounting for 10% (Valdar et al., 2003) to 20% of V_G (Belknap, 1998); thus, it is not surprising that not all QTLs replicated in all mapping populations.

The QTL on chr 8 was initially identified in a large F₂ intercross (Lore3; Markel et al., 1997) and replicated in two of the three LXS cohorts (Table 2) by all three mapping packages. The 1.5 LOD support interval included the NET gene (Slc6a2), at 92.2 Mb. Knockout mice lacking dopamine -hydroxylase cannot synthesize norepinephrine (NE), and they are hypersensitive to the sedative effects of ethanol (Weinshenker et al., 2000), making it a reasonable candidate for Lore3. This effect is blocked by acute replacement of central NE. ILS and ISS strains differ in a number of NET characteristics, including Slc6a2 haplotypes, [³H]NE uptake, NET binding, and mRNA levels, which are consistently 30 to 50% lower in ILS strain (Haughey et al., 2005). NET genotype significantly (p = 0.04) affected LORE in the LXS RI strains, explaining 5.6% of the phenotypic variance, with strains ILS in the NET region sleeping an average of 14 min longer than strains with an ISS genotype. This relatively small effect is likely the reason that Lore3 was the only QTL region from the F₂ mapping that failed to confirm in reciprocal congenic strains (Bennett et al., 2002b).

The SERT gene is located at the peak of the Lore4 QTL, and numerous studies have implicated serotonin in ethanol-related behaviors. A suggestive difference in LORE in the LXS further supported SERT as a candidate. The initial SNP analysis suggested that ILS and ISS strains were polymorphic through the gene region, but more recent data argue against this conclusion. The denser haplotype (Table 6) suggests that there are no ILS/ISS polymorphisms in the gene; however, because as no SNPs in the gene were typed in these strains, this conclusion is inferred based on flanking regions. This and the lack of binding differences rule against SERT as a candidate. NET remains as a strong candidate, based on the numerous polymorphisms between ILS and ISS in the gene, particularly in the unique haplotype seen in exons 4 and 11.

LORE constitutes a mouse model with strong face validity to a major risk factor for alcoholism in humans (Schuckit, 2000). Our mapping results in the LXS RI panel provide additional support for the heritable nature of this trait in mice and replicated many previously identified QTLs. Several of these QLTs have now emerged as candidates for intense follow-up to pursue the underlying gene(s) based on their replicability, significance, and interval reduction.

	Acknowledgements

 
We thank our colleagues at Institute for Behavioral Genetics for animal care and testing. Dr. Karl Broman patiently and on numerous occasions gave BB very thorough instruction on application and interpretation of R/qtl. Dr. Shengchu Wang kindly provided assistance with QTL Cartographer for Windows. Drs. Jonathon Flint and Richard Mott at the Wellcome Trust Centre for Human Genetics (Wellcome-CTC Mouse Strain SNP Genotype set) coordinated the SNP analysis of the LXS RI. We are indebted to Gaynor A. Larson for performing the SERT studies.

	Footnotes

 
This work was supported by National Institutes of Health Grants R01 AA11984 and AA008940 (to T.E.J.) and AA014666 and DA015663 (supporting the SPF), the Ellison Foundation for Medical Research (to T.E.J.), and by funds from the University of Colorado Denver Health Sciences Center (5 K05 DA015050-04) (to N.R.Z.).

Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.

doi:10.1124/jpet.106.103572.

ABBREVIATIONS: ILS, Inbred Long Sleep; ISS, Inbred Short Sleep; QTL, quantitative trait loci; LORE, loss of the righting reflex due to ethanol; Lore, QTL for LORE; RI, recombinant inbred; LS, Long Sleep; SS, Short Sleep; LXS, RI panel derived from ILS x ISS cross; GXE, gene-environment; NET, norepinephrine transporter; chr, chromosome; SERT, serotonin transporter; MK-801, 5H-dibenzo[a,d]cyclohepten-5,10-imine (dizocilpine maleate); SPF, specific pathogen-free; BEC, blood ethanol concentration; SNP, single-nucleotide polymorphism; kb, kilobase; LOD, logarithm of the odds; Xrcc5, X-ray repair complementing defective repair in Chinese hamster cells; ISCR, interval-specific congenic recombinant; NE, norepinephrine.

The online version of this article (available at http://jpet.aspetjournals.org) contains supplemental material.

Address correspondence to: Dr. Beth Bennett, Institute for Behavioral Genetics, 447 UCB, Boulder, CO 80309-0354. E-mail: bennettb{at}colorado.edu

	References

Top Abstract Materials and Methods Results Discussion References

 

Beavis WD (1994) The power and deceit of QTL experiments: lessons from comparative QTL studies. 49th Annual Corn & Sorghum Industry Research Conference (Johnston, IA ed); 1994 Dec 7-8, Chicago, IL. pp 250-266, American Seed Trade Associates, Chicago, IL.
Belknap JK (1998) Effect of within-strain sample size on QTL detection and mapping using recombinant inbred mouse strains. Behav Genet 28: 29-38.[CrossRef][Medline]
Belknap JK, Mitchell SR, O'Toole LA, Helms ML, and Crabbe JC (1996) Type I and type II error rates for quantitative trait loci (QTLl) mapping studies using recombinant inbred mouse strains. Behav Genet 26: 149-160.[CrossRef][Medline]
Bennett B, Beeson M, Gordon L, Carosone-Link P, and Johnson TE (2002a) Genetic dissection of quantitative trait loci specifying sedative/hypnotic sensitivity to ethanol: mapping with interval-specific congenic recombinant lines. Alcohol Clin Exp Res 26: 1615-1624.[CrossRef][Medline]
Bennett B, Beeson M, Gordon L, and Johnson TE (2002b) Reciprocal congenics defining individual quantitative trait loci for sedative/hypnotic sensitivity to ethanol. Alcohol Clin Exp Res 26: 149-157.[CrossRef][Medline]
Bennett B and Carosone-Link P (2006) Replication of small effect QTLs for behavioral traits: estimation of effect size from independent cohorts. Genes Brain Behav 5: 404-412.[CrossRef][Medline]
Bennett B, Carosone-Link PJ, Lu L, Chesler EJ, and Johnson TE (2005) Genetics of body weight in the LXS recombinant inbred mouse strains. Mamm Genome 16: 764-774.[CrossRef][Medline]
Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72: 248-254.[CrossRef][Medline]
Broman KW, Wu H, Sen S, and Churchill GA (2003) R/qtl: QTL mapping in experimental crosses. Bioinfomatics 19: 889-890.
Caspi A, Sugden K, Moffitt TE, Taylor A, Craig IW, Harrington H, McClay J, Mill J, Martin J, Braithwaite A, et al. (2003) Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Science (Wash DC) 301: 386-389.[Abstract/Free Full Text]
Chesler EJ, Lu L, Shou S, Qu Y, Gu J, Wang J, Hsu HC, Mountz JD, Baldwin NE, Langston MA, et al. (2005) Complex trait analysis of gene expression uncovers polygenic and pleiotropic networks that modulate nervous system function. Nat Genet 37: 233-242.[CrossRef][Medline]
Churchill GA, Airey DC, Allayee H, Angel JM, Attie AD, Beatty J, Beavis WD, Belknap JK, Bennett B, Berrettini W, et al. (2004) The Collaborative Cross, a community resource for the genetic analysis of complex traits. Nat Genet 36: 1133-1137.[CrossRef][Medline]
Crabbe JC, Metten P, Cameron AJ, and Wahlsten D (2005) An analysis of the genetics of alcohol intoxication in inbred mice. Neurosci Biobehav Rev 28: 785-802.[CrossRef][Medline]
DeFries JC, Wilson JR, Erwin VG, and Petersen DR (1989) LSxSS recombinant inbred strains of mice: initial characterization. Alcohol Clin Exp Res 13: 196-200.[CrossRef][Medline]
Goldstein DB (1983) Pharmacology of Alcohol, Oxford University Press, New York.
Hanania T, McCreary AC, Haughey HM, Salaz DO, and Zahniser NR (2002) MK-801- and ethanol-induced activity in inbred long-sleep and short-sleep mice: dopamine and serotonin systems. Eur J Pharmacol 457: 125-135.[CrossRef][Medline]
Haughey HM, Kaiser AL, Johnson TE, Bennett B, Sikela JM, and Zahniser NR (2005) Norepinephrine transporter: a candidate gene for initial ethanol sensitivity in inbred long-sleep and short-sleep mice. Alcohol Clin Exp Res 29: 1759-1768.[CrossRef][Medline]
Hyttel J (1994) Pharmacological characterization of selective serotonin reuptake inhibitors (SSRIs). Int J Psychopharmacol 9 (Suppl 1): 19-26.
Jiang C and Zeng ZB (1995) Multiple trait analysis of genetic mapping for quantitative trait loci. Genetics 140: 1111-1127.[Abstract]
Kelai S, Aissi F, Lesch KP, Cohen-Salmon C, Hamon M, and Lanfumey L (2003) Alcohol intake after serotonin transporter inactivation in mice. Alcohol Alcohol 38: 386-389.[Abstract/Free Full Text]
Lejoyeux M (1996) Use of serotonin (5-hydroxytryptamine) reuptake inhibitors in the treatment of alcoholism. Alcohol Alcohol Suppl 1: 69-75.[Medline]
Liu BH (1997) Statistical Genomics: Linkage, Mapping, and QTL Analysis (Stern RB and Didier D eds) p 59, CRC Press, Inc., Boca Raton, FL.
MacLaren EJ, Bennett B, Johnson TE, and Sikela JM (2006) Expression profiling identifies novel candidate genes for ethanol sensitivity QTLs. Mamm Genome 17: 147-156.[CrossRef][Medline]
Manly KF and Cudmore RH (2001) Map manager QTX, cross-platform software for genetic mapping. Mamm Genome 12: 930-932.[CrossRef][Medline]
Manly KF and Olson JM (1999) Overview of QTL mapping software and introduction to map manager QT. Mamm Genome 10: 327-334.[CrossRef][Medline]
Markel PD, Bennett B, Beeson M, Gordon L, and Johnson TE (1997) Confirmation of quantitative trait loci for ethanol sensitivity in long-sleep and short-sleep mice. Genome Res 7: 92-99.[Abstract/Free Full Text]
Markel PD, DeFries JC, and Johnson TE (1995) Use of repeated measures in an analysis of ethanol-induced loss of righting reflex in inbred long-sleep and short-sleep mice. Alcohol Clin Exp Res 19: 299-304.[CrossRef][Medline]
Markel PD, Fulker DW, Bennett B, Corley RP, DeFries JC, Erwin VG, and Johnson TE (1996) Quantitative trait loci for ethanol sensitivity in the LSxSS recombinant inbred strains: interval mapping. Behav Genet 26: 447-458.[CrossRef][Medline]
McClearn GE and Kakihana R (1981) Selective breeding for ethanol sensitivity: short-sleep and long-sleep mice, in Development of Animal Models as Pharmacogenetic Tools (McClearn GE, Deitrich RA and Erwin VG eds) pp 81-113, U.S. Government Printing Office, Washington, DC.
Melchinger AE, Utz HF, and Schön CC (1998) Quantitative trait locus (QTL) mapping using different testers and independent population samples in maize reveals low power of QTL detection and large bias in estimates of QTL effects. Genetics 149: 383-403.[Abstract/Free Full Text]
Murphy DL, Li Q, Engel S, Wichems C, Anderews A, Lesch K-P, and Uhl G (2001) Genetic perspectives on the serotonin transporter. Brain Res Bull 56: 487-494.[CrossRef][Medline]
Owens J, Bennett B, and Johnson TE (2002) Possible pleiotropic effects of genes specifying sedative/hypnotic sensitivity to ethanol on other alcohol-related traits. Alcohol Clin Exp Res 26: 1461-1467.[CrossRef][Medline]
Schuckit MA, Edenberg HJ, Kalmijn J, Flury L, Smith TL, Reich T, Bierut L, Goate A, and Foroud T (2001) A genome-wide search for genes that relate to a low level of response to alcohol. Alcohol Clin Exp Res 25: 323-329.[CrossRef][Medline]
Schuckit MA (2000) Genetics of the risk for alcoholism. Am J Addict 9: 103-112.[CrossRef][Medline]
Schuckit MA, Wilhelmsen K, Smith TL, Feiler HS, Lind P, Lange LA, and Kalmijn J (2005) Autosomal linkage analysis for the level of response to alcohol. Alcohol Clin Exp Res 29: 1976-1982.[CrossRef][Medline]
Smolen A and Smolen TN (1989) Blood and brain ethanol concentrations during absorption and distribution in long-sleep and short-sleep mice. Alcohol 6: 33-38.[CrossRef][Medline]
Smolen A, Marks MJ, Smolen TN, and Collins AC (1986) Dose and route of administration alter the relative elimination of ethanol by Long-sleep and Short-sleep mice. Alcohol Clin Exp Res 10: 198-204.[Medline]
Sokal R and Rohlf F (2000) Biometry, W.H. Freeman and Company, New York.
Valdar WS, Flint J, and Mott R (2003) QTL fine-mapping with recombinant-inbred heterogeneous stocks and in vitro heterogeneous stocks. Mamm Genome 14: 830-838.[CrossRef][Medline]
Wang J, Williams RW, and Manly KF (2003) WebQTL: Web-based complex trait analysis. Neuroinformatics 1: 299-308.[CrossRef][Medline]
Wang S, Basten CJ, and Zeng ZB (2005) Windows QTL Cartographer 2.5, Department of Statistics, North Carolina State University, Raleigh, NC.
Weinshenker D, Rust NC, Miller NS, and Palmiter RD (2000) Ethanol-associated behaviors of mice lacking norepinephrine. J Neurosci 3157-3164.
Williams RW, Bennett B, Lu L, Gu J, DeFries JC, Carosone-Link P, Rikke BA, Belknap JK, and Johnson TE (2004) Genetic structure of the LXS panel of recombinant inbred mouse strains: a powerful resource for complex trait analysis. Mamm Genome 15: 637-647.[CrossRef][Medline]

This article has been cited by other articles:

				J. C Crabbe Neurogenetic studies of alcohol addiction Phil Trans R Soc B, October 12, 2008; 363(1507): 3201 - 3211. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Data Supplement All Versions of this Article: jpet.106.103572v1 319/1/299    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map 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 HighWire Citing Articles via Google Scholar Google Scholar Articles by Bennett, B. Articles by Johnson, T. E. Search for Related Content PubMed PubMed Citation Articles by Bennett, B. Articles by Johnson, T. E.