Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Psychostimulants such as cocaine, amphetamine, and methylphenidate, and therapeutic agents such as antidepressants and drugs that are used to treat Parkinson's disease, interact with the dopamine (DA) transporter (DAT), block DA uptake, and increase its synaptic availability (for review, see Povlock and Amara, 1997). The exclusive expression of the DAT by DA neurons is dependent on activating and silencing factors in the promoter region of the DAT gene (Donovan et al., 1995). Thus, variations in coding within the promoter region, or proteins that bind to promoter regions, may control expression levels of the DAT. The DAT gene spans at least 60 kilobases of genomic sequence (Vandenbergh et al., 2000). Transcription factor binding sites in the first 8 kilobases of the 5' untranslated region of the human DAT gene have been identified, including AP, zif-268, Sp1, cyclic AMP response element binding sites, and NGF1-B response element and neuron-restrictive silencer elements (Sacchetti et al., 1999). Promoter elements including AP1, AP2, SP1, and a neuron-specific core motif are found in the first intron of the DAT gene (Kouzmenko et al., 1997). Transcription factors such as nurr1, which binds to NGF1-B response element, and Ptx3 and silencer factors, which bind to neuron-restrictive silencer elements, have been found in DA neurons. Some of these factors regulate expression of the recombinant transporter in mammalian cells (Smidt et al., 1997; Sacchetti et al., 1999). For instance, Nurr1 enhances transcriptional activity of human DAT gene constructs in cell lines that express a dopaminergic phenotype (Sacchetti et al., 2001).

These factors may be involved in the effects of chronic administration of some transporter antagonists, which increase transporter expression (Tella et al., 1997). In particular, chronic cocaine abuse causes increased DAT expression in regions of human brain (Malison et al., 1998). Furthermore, withdrawal of animals from administration of transporter blockers decreases transporter expression (Hitri and Wyatt, 1993). Linkage studies have investigated trait-specific correlations among allelic variations in noncoding portions of the human DAT gene and neuropsychiatric disorders. Linkage has been observed between attention deficit hyperactivity disorder and a polymorphism located in the 3' untranslated region of the DAT gene (Cook et al., 1995).

Extensive similarity between the mouse and human genomes facilitates linkage studies of genetic variation and trait phenotype using mouse models. Over the last two decades we have characterized two lines of mice, the C57BL/6 (B6) and DBA/2 (D2), and a panel of recombinant inbred strains (BXD RI), which have markedly different responses to behavioral tests and to drug treatments, and have extensive genetic polymorphisms. We use a large marker database for genetic polymorphisms among B6, D2, and RI mice that allows us to uncover chromosomal regions containing quantitative trait loci (QTLs) that contain alleles (genes) that influence continuously distributed or quantitative traits (phenotype) such as DAT expression. Because of their typically polygenic and polyenvironmental determination, quantitative traits are often referred to as complex traits. QTL mapping methods and results for quantitative traits relevant to substance abuse were recently reviewed by Belknap et al. (1997) and Crabbe et al. (1999). To map a QTL, its influence on the trait must be detected amid considerable "noise" from other QTLs and nongenetic sources. Using recently developed molecular techniques and statistical methods, it is possible to identify genetic variation (polymorphisms) at marker loci throughout the genome, and to map QTLs from the association of marker and trait data, including molecular and behavioral traits (Lander and Kruglyak, 1995). We have now used QTL analyses to identify chromosomal regions containing genes that play a role in regulating the expression of the DAT in neostriatum, a brain region with dense dopaminergic innervation. Correlational analysis of the data using a large BXD trait database indicates that the QTL involved in DAT expression is consistently associated with two groups of behaviors related to cocaine and amphetamine drug responses, locomotor activation, and thermal responses (hypo- or hyperthermia).

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. 3-(4-Iodophenyl) tropane-2-carboxylic acid methyl ester ([¹²⁵I]RTI-55) (2200 Ci/mmol) was purchased from PerkinElmer Life Science Products (Boston, MA). Tropolone, DA, pargyline, and most other reagents were purchased from Sigma Chemical Co. (St. Louis, MO). Mazindol was purchased from Sigma/RBI (Natick, MA). Male C57BL/6J (B6) mice, DBA/2J (D2) mice, and their RI strains (BXD) (50-60 days old) were obtained from the Portland Alcohol Research Center breeding colony. The progenitor strains and 21 inbred strains that were available in sufficient number were used in the present study. Animals were maintained on a 12-h light/dark cycle (lights on at 6:00 AM), and had access to food and water ad libitum.

Radioligand Binding to Biogenic Amine Transporters. Radioligand binding assays were conducted by minor modifications of our previously described methods (Eshleman et al., 1999). Briefly, mouse brain tissue was dissected on ice and stored at 70°C until assayed. To characterize the DAT, the neostriata from a single animal were homogenized with a Teflon-glass homogenizer in 600 volumes (original wet weight) of ice-cold sucrose (0.32 M). The suspension was centrifuged at 1,000g for 10 min at 4°C, and the resulting supernatant was centrifuged at 33,000g for 25 min at 4°C. The final pellet was resuspended in 80 volumes (original wet weight) of ice-cold sucrose (0.32 M) using a Polytron (setting 7, 10 s).

Quantitative Trait Locus Analysis of Binding Data. The BXD RI strains were developed by inbreeding from an F₂ cross between the C57BL/6J (B6) and DBA/2J (D2) progenitor inbred strains. Thus, each strain represents a unique and random "patchwork" of chromosomal segments from the B6 and D2 strains arising from crossovers, but now fixed in an inbred (homozygous) state.

Genetic Correlations between DAT Expression and Behavior in BXD Mice. A BXD psychostimulant trait database comprised of 45 traits related to sensitivity to either cocaine or methamphetamine, where at least 18 BXD strains had been tested (with at least six mice per strain), was used to determine the relationship between DAT expression and psychostimulant drug sensitivity measures. This database comprised of BXD strain means is maintained as part of the Portland Alcohol Research Center BXD database. Almost all of the 45 traits were collected as part of National Institute on Drug Abuse Research Contracts and National Institute on Drug Abuse R0-1 grants to one or more of the present authors. All quantified traits represent drug responses minus saline (control) values per strain. This correction of drug values was carried out either between groups when there was a separate saline group per strain, or within subjects, where each subject was tested for both saline and drug, allowing each subject to serve as its own control. These traits are listed in Table 2. These traits typically involved more than one dose of either cocaine (COC) or methamphetamine (MA), where each dose defined an individual trait. Each mouse received only one injection of a drug i.p., except for locomotor sensitization, where five administrations of a single dose were given per mouse, 2 days apart. The psychostimulant (MA or COC) drug sensitivity measures included hypothermia (low doses), hyperthermia (high doses), activity (either home cage, small open field, or larger open field), sensitization of activity with repeated doses, stereotypy (repetitive paw-to-mouth and chewing behavior), tremor, seizures (COC only), exophthalmos, climbing (a form of stereotypy), conditioned place preference (an index of drug reward), and two-bottle choice drinking behavior (water versus drug solution).

Multivariate Analyses. We also carried out multivariate analyses of the psychostimulant traits plus DAT density in neostriatum using the techniques of multidimensional scaling (MDS; Kruskal and Wish, 1978) and hierarchical cluster analysis using average linkage (Aldenderfer and Blashfield, 1984). The first step for either analysis was to construct a matrix by correlating all psychostimulant-related traits plus DAT density (total of 46 traits) in all pairwise combinations, for a total of 1035 unique correlations. This correlation matrix is available upon e-mail request from belknajo{at}ohsu.edu.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

DAT Characteristics in BXD Mouse Neostriatum. The binding of [¹²⁵I]RTI-55 to the DAT was characterized using a crude neostriatal membrane preparation from 21 RI strains and from B6 and D2 mice. We made 138 separate determinations for binding site density (B_max) and affinity (1/K_d), each involving neostriata from an individual mouse. For each strain, the mean of 3 to 12 independent determinations was calculated. The distribution of strain mean differences in B_max values is shown in Fig. 1. There was approximately a 3-fold difference in the density of DAT binding sites between the lowest (BXD-19, 1.7 ± 0.2 pmol/mg of protein, n = 6) and the highest BXD RI strain (BXD-16, 5.3 ± 1.1 pmol/mg of protein, n = 3). The D2 progenitor strain had 50% more binding sites (3.4 ± 0.7 pmol/mg of protein, n = 5) than the B6 strain (2.3 ± 0.2 pmol/mg of protein, n = 8) (Fig. 1). One-way analysis of variance revealed a highly significant strain difference (p < 0.001) in B_max values, but no significant difference in affinity (1/K_d; p = 0.16, N.S.). Comparing individual values for B_max and K_d resulted in a correlation coefficient of 0.2, indicating that these two DAT characteristics are not significantly associated across the RI strains. Note that for B_max values, some of the BXD RI strains score beyond the range of the two progenitor strains. This is characteristic of multilocus (polygenic) traits, but not monolocus (single gene) traits. Also, the trait is not bimodally distributed as expected for a single locus trait. Belknap et al. (1993b) have shown that a locus must account for two-thirds or more of the genetic variance before a discernibly bimodal distribution of RI strains results.

View larger version (32K): [in this window] [in a new window]   Fig. 1.   DAT characteristics in BXD recombinant inbred mice. Distribution of [125I]RTI-55 binding site density (Bmax) across 21 BXD RI strains and their progenitors, the C57BL/6J (B6) and DBA/2J (D2) strains. Each column (±S.E.M.) is based on 3 to 11 determinations per strain from individual mice (average of six determinations per strain).

Heritability of DAT Expression and Radioligand Binding Affinity in BXD Mouse Neostriatum. The marked strain differences in DAT B_max values in neostriatum indicate that this trait is substantially heritable. The heritability (h²), which indexes the proportion of the total variation in B_max values due to strain (genetic variation), was estimated by r² from a one-way analysis of variance to be 0.40, which is highly significant (p < 10⁵). The reliability of the B_max measure across strains was estimated by the split-half method. Two half-samples per strain were formed using odd-even assignments. The correlation between the two half-samples, using the Spearman-Brown formula (McNemar, 1960), was 0.88, and indicates that this trait is more than sufficiently reliable for genetic analyses. As expected, the heritability of K_d values was not significantly different from zero, and therefore no further genetic analysis was justified. For this reason, only B_max values were subjected to the QTL and other genetic analyses described below.

QTL Analysis of DAT Expression (B_max). Radioligand binding data were subjected to QTL analysis, which involved a genome-wide search for chromosomal regions that appear to influence DAT expression in neostriatum. This resulted in one very large-effect QTL and a number of much smaller provisional ones. The largest QTL effect was found on proximal chromosome 19 (8-11 cM), where a number of markers showed strong associations (Fig. 2). The confidence interval estimated by ±1.0 LOD support ranged from 8 to 11 cM: this locates the approximate 95% confidence interval for the position of the QTL. The most strongly associated marker was Pomc-ps1 (formerly Pomc2), a proopiomelanocortin pseudogene located 9 cM from the centromere, with r = 0.84, n = 20 strains, LOD = 4.7, df = 1, p = 3 × 10⁶. The correlation was positive in sign, indicating that the D2 allele is associated with higher B_max values relative to the B6 allele. The proportion of the genetic variance accounted for by this QTL can be estimated by the square of the correlation coefficient, or 0.84² = 0.70. This is likely to be an overestimate because of the phenomenon of regression toward the mean upon replication, and because the strain mean differences are not entirely due to genotype, but to a small extent are also influenced by environmental factors (Belknap, 1998). However, it is reasonable to estimate that this QTL accounts for more than half of the genetic variance.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Association of allelic variation along the length of proximal chromosome 19, with variation in the density of DAT (Bmax for [125I]RTI-55 binding) expressed in the neostriatum of BXD recombinant inbred mice indexed as logarithm of the odds in favor of the presence of a QTL (LOD; df = 1). The horizontal dotted line is the significance threshold of p = 0.0001, LOD = 3.3, df = 1.

Candidate Genes. Mouse and human genome databases were searched for potential candidate genes within the chromosome 19 QTL that could influence DAT expression. Table 1 contains a list of candidate genes that are within the QTL, including transcription factors and guanine nucleotide regulatory proteins. Genes that are found in homologous regions of the human genome are also listed.

Genetic Correlations between DAT Expression and Psychostimulant Behavior in BXD Mice. Of the 45 psychostimulant drug-related traits (most of them behaviors), eight attained p < 0.05 when their BXD RI strain means were correlated with neostriatum [¹²⁵I]RTI-55 binding B_max strain means. The correlations are given in Table 2. We would expect 0.05 × 45 = 2.25 correlations at p < 0.05 due to chance, so the eight we observed is 3.5-fold more than expected by chance. The probability that eight genetic correlations of 45 could have emerged at p < 0.05 due solely to chance is p < 0.0001 (Poisson distribution; Sokal and Rohlf, 1995). Of these eight, six are locomotor activity measures (four COC, two MA) in either home cage or open field, and two are thermal responses based on rectal probe body temperature determinations after MA administration. This strongly suggests that the density of DAT expression in the neostriatum has important influences on locomotor activity and thermal responses to psychostimulants. All eight traits are described in more detail in Table 2. The correlation of four of these traits, three activity and one thermal response trait, with the density of the DAT binding sites is shown in Fig. 3, A and B. B_max values did not significantly correlate with any trait belonging to the other groupings listed in Table 2, including sensitization, drinking traits (two-bottle choice, voluntary consumption), or conditioned place preference. This suggests that genetic variation in DAT binding site density in neostriatum is not related to genetic variation in any of these traits in the BXD set.

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Correlations between DAT binding site density ([125I]RTI-55 Bmax) and psychostimulant-induced behaviors. A, COC- and MA-induced locomotor activity in an Accuscan activity monitor. Units are distance traveled in centimeters. B, MA-induced locomotor activity in the home cage (line crossings), or MA-induced changes in body temperature (change from predrug baseline in °C). Details for the behavioral measures are included in citations in Table 2. Numbers in the graphs are BXD RI strain numbers.

View larger version (29K): [in this window] [in a new window]   Fig. 4.   Multidimensional scaling plot in two dimensions of the genetic similarities among 45 cocaine or methamphetamine behavioral traits and the density of DAT binding sites in neostriatum from the BXD recombinant inbred mice and the progenitor strains. A, each plotted point represents a single trait. The Euclidean distance between any two points, representing two psychostimulant response traits, suggests the degree of their genetic similarity. Thus, traits plotted close together are genetically more similar than are traits plotted far apart. Circles are used to indicate related behaviors. This two-dimensional plot accounted for 60% of the total variation seen in the original correlation matrix (1035 correlation coefficients) upon which this plot was based. B, three-dimensional contour plot is superimposed on the two MDS dimensions, with peaks representing a relatively high degree of aggregation of nearby points (traits) in that section of the plot, while valleys represent portions of the plot where there are no (or few) plotted points nearby. The MDS data and the x, y coordinates are the same in both A and B. The "X" shown in both plots denotes the Bmax for specific [125I]RTI-55 binding in neostriatum.

Correlation between Chromosome 19 QTL for DAT Expression and Psychostimulant Behaviors. We also carried out a QTL analysis to determine whether the chromosome 19 QTL, as indexed by the marker Pomc-ps1, had any detectable influence on any of the 45 psychostimulant traits in our BXD database other than DAT expression. When this search was conducted, a 3.5-fold greater number of psychostimulant traits (n = 8) emerged at p < 0.05 than expected by chance (p < 0.0001, Poisson distribution). [At p < 0.01, there were six traits; p < 0.00001 versus null hypothesis of chance association.] Many of the same traits associated with DAT density in neostriatum at p < 0.05 were also associated with this chromosome 19 marker. These are identified with an asterisk (p < 0.05) or double asterisk (p < 0.01) in Table 2. All are either activity or thermal response traits. Thus, the chromosome 19 QTL with a strong influence on the expression of the DAT in neostriatum may also influence the eight activity and thermal response traits, but not any trait in the other trait groupings (activity, sensitization, voluntary consumption or drinking, or tremor) listed in Table 2.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

QTL analysis is a global screening method, in that the entire genome is searched for evidence of loci influencing a trait without bias in favor of one polymorphic gene over another. In contrast, other genetic approaches such as targeted mutagenesis are more narrowly focused. One unique aspect of global unbiased screening approaches is that they are much more likely to detect modifier gene loci with important influences on the trait of interest, particularly those that would not be suspected based on what is known about the trait. Data presented above provide a good example of how a previously unknown modifier (the locus on chromosome 19) was found, and what proteins (DAT) and behaviors (activity and thermal response) are affected by the expression of the QTL.

The BXD RI strains provide a powerful tool because of the large amount of genetic, behavioral, and biochemical data that are available for each strain. The BXD RI strain mean database allowed us to carry out multivariate analyses that go well beyond the more typical bivariate statistics such as paired correlational analyses. Genome-wide searches revealed one very large effect QTL on chromosome 19 that appeared to account for somewhat over half of the genetic variance in DAT expression in the neostriatum. No other chromosomal region attained a level of statistical significance that would permit the conclusion of a definite association. However, D16Mit5 on chromosome 16 is suggestive according to the Lander and Kruglyak (1995) guidelines. Because of the high rate of false positives among markers associated only at p < 0.01 (an average of four false positives per genome-wide search; Belknap et al., 1996), smaller provisional QTLs such as that on chromosome 16 will be subject to confirmation testing in a different mouse population derived from the same progenitors to determine whether any of them can be independently supported. Despite this caveat, it is noteworthy that a provisional QTL on chromosome 7 (p = 0.005), Xmmv76, is very close to the dopamine D4 receptor gene, Drd4 (7:70). It is important to note that the findings reported here are based on RI mice from a cross between the C57BL/6 and DBA/2 inbred strains. Other inbred strain crosses could result in different findings.

Interestingly, the chromosome 19 QTL is not the site of the Dat1 gene that encodes the DAT protein, which is on chromosome 13. Since the DAT gene did not present as a QTL, there is presumably no polymorphism for Dat1 that affects binding site characteristics or expression in neostriatum in our population of mice. Furthermore, the nonsignificant heritability for K_d values across the RI strains suggests that the structure of the transporter is well conserved, or that any structural differences do not affect [¹²⁵I]RTI-55 binding site characteristics. Thus, the chromosome 19 QTL appears to be a modifier gene, which modulates the expression of Dat1 in neostriatum, and may also play a role in a number of psychostimulant effects.

In a previous study, Womer et al. (1994) reported that B6 and D2 mice have similar levels of DAT expression in both the neostriatum and the nucleus accumbens. In contrast, we found a 50% higher DAT density in D2 mice. However, a different radioligand ([³H]GBR-12935) was used in that earlier study. In addition to binding with high affinity to the DAT (Janowsky et al., 1986), [³H]GBR-12935 binds to a high capacity piperazine acceptor site, possibly cytochrome P450 2D6, which could confound strain differences in radioligand binding (Gordon et al., 1995). Interestingly, the difference in DAT expression between the progenitor strains reported here was small compared with the difference in DAT expression across all of the RI strains that were tested. This is expected of a polygenic, but not of a monogenic trait.

The difference in DAT expression across RI strains could reflect differences in the density of terminal fields, or the number of dopaminergic neurons or cell bodies that contribute to the terminal field. Support for these possibilities includes differences in the activity of tyrosine hydroxylase, the rate-limiting enzyme in dopamine biosynthesis, in BALB/cBy and C57BL/6By mice, which have been attributed to differences in the number of midbrain dopaminergic neurons (Vadasz et al., 1982).

The QTL could also play a role in endocrine system interactions with transporter expression. Ovariectomy in the rat increases DAT expression in the neostriatum and SERT expression in the hypothalamus (Attali et al., 1997). Changes in DAT expression can be reversed by administration of estradiol or estradiol plus progesterone. Differences in hormone-induced changes in DAT and SERT suggest different elements in their respective mechanisms of regulation. We do not know whether possible variations in hormones across strains of BXD male mice described here affected DAT expression.

The data also suggest that DAT expression in neostriatum is genetically related to MA- and COC-induced locomotor activation and thermal responses such as hypo- and hyperthermia, but not to other psychostimulant effects such as sensitization for activity, two-bottle choice drinking, or conditioned place preference. It is possible that DAT expression in brain regions other than the neostriatum may have different relationships with the psychostimulant groupings. For example, determination of DAT expression in mesolimbic pathways involved in drug reward might indicate an association with drinking and conditioned place preference measures, rather than activity or thermal responses.

There are a number of gene candidates that could influence trait-specific DAT expression and related behaviors. Brodkin et al. (1998) demonstrated a difference in DAT immunoreactivity in AXB, BXA RI strains derived from the A/J and C57BL/6J inbred strains, and found a correlation between FosB and DAT expression, suggesting that the early gene plays a role in differences in DAT expression across mouse strains. FosB, a splice variant of the fosB gene, has been mapped to chromosome 7:5 cM. A provisional QTL (p < 0.05) emerged in this same region in our study, but was not described here because it did not meet our arbitrary p < 0.01 criterion for reporting. Other genes and response element binding proteins that appear to affect transporter expression have also been mapped [zif-268 (chr 18:16), nurr1 (chr 2:34.5), and Ptx3 (chr3:33.8), c-fos (chr 12:40), c-jun (chr 4:45), junB (chr 8:39), and cyclic AMP response element binding (CREB) (chr 1:31); Sacchetti et al., 1999]. Provisional QTLs from the present study were mapped at about 4:54 and 12:52, in the same general regions as c-jun and c-fos. Since the 95% confidence intervals for map location of these smaller provisional QTLs is on the order of 15 to 25 cM, the c-jun and c-fos genes could be the basis for two of our provisional QTLs.

The one significant QTL on chromosome 19 does not comap with any of these genes. However, Table 1 lists a number of candidate genes that are found within the chromosome 19 QTL and that could affect DAT expression. Fxb2, a forkhead domain transcription factor gene is active during embryogenesis and could alter DAT expression during maturation. Other candidate genes include ciliary neurotrophic factor, which regulates Janus kinase 2 (also found on chromosome 19 but outside the QTL) and mediates some actions of cocaine (Berhow et al., 1996).

The previously unreported locus that accounts for half of the genetic variation in the expression of the DAT has important implications for understanding the genetic predisposition to neuropsychiatric disorders, including drug abuse, to understanding how genes interact, and to the role of regulatory factors in transporter expression. The functional DAT is required for the neurotoxin-mediated dopaminergic denervation observed in drug-induced Parkinson's disease. In addition, hyperactive behavior in drug-naïve DAT knockout mice has led to the proposal of these mice as a model of attention deficit hyperactivity disorder (Gainetdinov et al., 1999), and polymorphisms of the DAT gene are associated with attention deficit hyperactivity disorder (Gill et al., 1997). There is also a locus for bipolar disorder near the DAT gene in humans (Kelsoe et al., 1996), but this gene also falls outside of the QTL mapped here. Clearly, the DAT mediates the effects of psychostimulants, and genetic regulation of high or low DAT expression is correlated with differences in some behavioral responses to drugs (Fig. 3). Taken together with complementary approaches including genome sequencing, QTL analysis should be useful for describing the influence and interaction of genes on expression of other proteins and on associated behaviors. Congenic strains that involve the transfer of a small region of chromosome 19 from the B6 strain onto the genome of the D2 strain through repeated backcrossing will isolate the chromosome 19 QTL against a uniform genetic background, and facilitate its characterization from the molecular to the behavioral level.