Although dopaminergic neurons are few (e.g. 10–20,000 in the rat brain), they regulate a number of physiological, behavioral and cognitive functions, including regulation of locomotor activity, incentive behaviors, short-term and stimulus-dependent memory systems [1–3]. Dysfunction of the dopaminergic system and its targets underlie major human disorders such as Parkinson's disease, Huntington's disease, schizophrenia and addiction to drugs of abuse [4–6]. Five dopamine receptors, termed D1, D2,D3, D4 and D5, encoded by five different genes (Drd1a, Drd2, Drd3, Drd4 and Drd5 respectively) have been identified [7, 8]. The elucidation of the signaling properties of dopaminoceptive neurons is of central importance for the development of therapeutic strategies to treat dysfunctions of the dopaminergic system.

In the past years, use of the Cre/loxP system has enabled generation of cell-specific mutations In the central nervous system, efforts have generally been devoted to the generation of region-specific mutations using promoters to drive the expression of Cre recombinase in specific brain areas [9–11]. In order to target the dopaminoceptive neurons, we have generated a transgenic line that expresses Cre recombinase in dopaminoceptive neurons under the control of the dopamine D1 receptor gene (D1Cre), by using the genomic locus of a dopamine D1 receptor gene cloned in a YAC vector.

Here we examine the cellular expression of Cre recombinase in D1Cre transgenic mice by immunohistochemistry in combination with in situ hybridization, to compare it with the expression of D1 receptor mRNA and other dopaminoceptive molecular markers. The activity of the Cre was confirmed by loss of CREB, CaMKIV and GR proteins in the respective conditional mutants and by induction of beta-galactosidase activity in the R26R^D1Cre reporter mice.

A 140 kb yeast artificial chromosome (YAC) containing the entire dopamine D1 receptor gene (Drd1a) was isolated by screening a mouse genomic YAC library [12]. The YAC encompasses the entire Drd1a gene including 40 kb of upstream and 100 kb of downstream sequences. The coding sequence for the nuclear localized nlsCre [13] was introduced into the Drd1a gene by homologous recombination in yeast. To minimize the chances of altering possible regulatory elements, we avoided the use of a heterologous polyadenylation sequence and minimized the deletion of the Drd1a coding sequence. Specifically, the homology regions were chosen so that the ATG of the nlsCre would replace the ATG of the Drd1a gene and the STOP codon of the nlsCre would interrupt the Drd1a open reading frame after its last possible internal ATG.

The modified YAC was used to generate 5 independent transgenic lines (D1Cre lines A, R, S, T and U). Two lines (T and U) carried incomplete copies of the transgene as assessed by screening for the presence of the extremities of the YAC arms (data not shown). Examination of Cre expression by immunohistochemistry in the three remaining lines (A, R and S) revealed a virtually identical pattern of expression (Fig. 1). Moreover, the level of expression was in direct correlation with the copy number of the transgene (Fig. 1). Detailed analysis of the Cre expression pattern was conducted with line A which carries three copies of the YAC transgene. Expression was prominent in the major projections areas of the dopaminergic system, including striatum, nucleus accumbens, olfactory tubercles and prefrontal cortex (Fig. 2). Clear but lower expression was seen in layer VI of the cortex. In hippocampus, CA2 and scattered cells in the dentate gyrus expressed Cre (Fig. 2). Several thalamic nuclei and the ventromedian hypothalamic nucleus also expressed Cre (Fig. 2). In retina, robust staining was observed in the inner nuclear layer (Fig. 2). In order to verify that Cre is expressed in neurons expressing the D1R mRNA, we have performed in situ hybridization with a specific D1R riboprobe in combination with immunohistochemical detection of Cre expression in striatum and cortex (Fig. 3a, b).

The dynamics of Cre expression was examined during embryonic development and the postnatal period. D1R mRNA expression in the striatum has been reported at E13 in mice [14], therefore we have analyzed Cre protein at E14.5. At E14.5, Cre could not be detected in striatal anlage nor in any brain region (Fig. 4a–c and data not shown). Cre expressing cells appeared in striatal patches, the so-called striosomes at E16.5 and a only very few weakly expressing cells could be detected in the developing cortex (Fig. 4d–f). At E18.5 the number of Cre expressing cells increased in the striatum as well as in the presumptive layer VI of the cortex (Fig. 4g–i). At postnatal days 0 and 3 (P0 and P3), Cre recombinase was still predominant in the striosomal neurons and layer VI of the cortex (Fig. 5). At P6, Cre expression started to be seen between the patches. Cre expression expanded progressively until the adult stage. Transient and very weak expression was observed in the CA1 region of the hippocampus and in layer IV of the cortex (Fig. 5). These results indicate that Cre starts to be expressed, albeit at low levels, at E16 in the striosomal neurons of the striatum. At the end of the first week after birth, Cre expression is seen also in cell bodies of the matrix, recapitulating the pattern of expression described for D1R mRNA [15–17].

In order to assess the specificity and the onset of Cre activity we have crossed D1Cre transgenic mice with the R26R reporter line [18] (Figure 6). We rule out the possibility of recombination in regions outside the CNS as shown at E11 by beta-galactosidase assay in the R26R^D1Cre reporter mice (Fig. 6a). Interestingly, we were able to detect Cre activity in the tectum above the mesencephalic region (Fig. 6d). The pattern of recombination induced by Cre recombinase was also analyzed in the adult brains (Fig. 6b, c, e, f). Beta-galactosidase activity in R26R^D1Cre was induced not only in the major areas of Cre expression shown in Figure 2, but also layers in IV to VI in the cortex (Fig. 6b, f) and in the CA1 region (Fig. 6c), expressing the Cre recombinase in early postnatal development (Figure 5).

We next examined whether Cre expression was confined to a specific subpopulation of striatal neurons. In the striatum, two subtypes of spiny neurons are involved in the so called "direct-" and "indirect-pathway". The direct pathway originates from striatal neurons projecting directly to the medial globus pallidus (GPm) and the substantia nigra pars reticulata (SNr). The neurons contributing to the indirect pathway are instead connected to these areas via the basal ganglia. Although it is still a matter of debate, it has been classically thought that the direct pathway neurons express the D1 dopamine receptor and the indirect-pathway neurons express the D2 receptor (D2R) [19, 20]. In order to establish whether Cre expression is confined to a specific subpopulation, we retrogradely labeled neurons from the striatonigral pathway by stereotaxic injection of red fluorescent latex spheres into the substantia nigra pars reticulata and localized Cre within the striatum by immunofluorescence (Fig. 7a). Double labeling of neuron-specific class III beta-tubulin and Cre recombinase showed that the majority of striatal neurons express the Cre recombinase driven by the Drd1a gene (Fig. 7b). This experiment showed that Cre expression was not restricted to the retrogradely labeled neurons which correspond to neurons of the direct pathway that project directly to the substantia nigra.

To further assess the subpopulation specificity of the D1Cre transgene, we checked whether also D2R-positive cells express Cre protein in the striatum. Therefore, we performed in situ hybridization by using a D2R-specific riboprobe followed by immunohistochemistry with Cre-specific antibody (Fig. 7c). This experiment shows that D1Cre is expressed also in about 40% of D2R-expressing cells, and that only about 20% of D2R positive cells do not express Cre at detectable levels (Fig. 7d) while about 40% of the neurons express D1Cre but not D2R mRNA. Therefore, we conclude that at least 80% of the striatal neurons express D1Cre. It has to be mentioned that some studies indicate that there is only very little overlap in the number of neurons expressing both receptors [19, 21]. According to these studies we were not expecting expression of D1Cre in D2R positive cells to such an extent. However, depending on the sensitivity and resolution of the adopted technical approach, others studies reveal that there is a higher degree of overlap [22] and that possibly all dopamine receptor-containing neurons, in the striatum, express both receptor types [23, 24]. Our study does not solve this debate about the degree of overlap between D1R and D2R, however, the finding that Cre expression is detectable in both neuronal subpopulations, D1R- and/or D2R-expressing neurons, suggests that these receptors are at least partially coexpressed in a certain proportion of striatal neurons (about 40%).

The expression of D1Cre not only in D1R positive neurons (Fig. 3a), but also in D2R positive cells (Fig. 7d), makes the D1Cre line suitable for gene inactivation in virtually all dopaminoceptive neurons of the striatum. The pattern of recombination induced by Cre recombinase was verified by crossing D1Cre transgenic animals to 3 loxP lines, Creb1^loxP [25], CaMKIV^loxP [26], GR^loxP [18, 27]. Recombination was visualized by loss of CREB, CaMKIV and GR proteins in the respective mutants. Loss of CREB, CaMKIV and GR proteins was extensive (at least 90%) in the main regions of Cre expression, striatum, nucleus accumbens, olfactory tubercles, and layer VI of the cortex (Figure 8). In the hippocampus, expression of Cre in the CA2 subfield results in loss of CREB, and CaMKIV in the corresponding cells. No or very marginal loss of CREB, CaMKIV and GR was observed in CA1 and layer IV of the cortex, in spite of the transient expression of Cre during the early postnatal period. These results underscore the importance of verifying recombination in the context of the loxP modified allele of interest.

The pattern of expression obtained with the D1Cre YAC transgene was very reproducible between transgenic lines. The expression level was essentially only dependent on the copy number of the transgene. These observations illustrate well the properties of transgenes based on large DNA segments: position-independent and copy-number dependent expression of the transgene. This correlation between expression pattern and copy number usually does not apply to plasmid-based constructs [28]. The pattern of Cre expression as seen in the D1Cre transgenic animals is very close to the pattern of expression of the endogenous Drd1a gene reported in previous in situ hybridization studies [29]. Moreover, the developmental and postnatal patterns of expression match the patterns described for the endogenous Drd1a gene [15–17]. We have also performed in situ hybridization with D1R riboprobe in combination with Cre immunohistochemistry in adult cortex and striatum and found high levels of overlap (Fig. 3a, b).

Several publications have suggested that D1 and D2 receptors are segregated specifically on distinct neuronal populations of the rat striatum that projects directly to substantia nigra pars reticulata and to globus pallidus respectively [20, 30]. Some other publications have however demonstrated that D1 and D2 might actually be coexpressed [22–24]. These discordant studies clearly indicate that a fine modulation of expression levels of D1R and D2R takes place in the dopaminoceptive neurons. In our transgenic line, we observe that Cre is expressed in a majority of striatal neurons, but it remains open whether our vector, even though it contains large segments of 5'- and 3'- flanking sequences, carries all necessary elements required to achieve this fine modulation of expression levels within striatum.

A potential application of D1Cre transgene for isolation of mutated dopaminoceptive cells is represented by the generation of double transgenic mice with BAC-D1GFP and BAC-D2GFP obtained by the GENSAT BAC Transgenic project [31]. The D1Cre expression pattern matches quite closely the BAC-D1GFP pattern in the medium spiny neurons of the striatum, therefore one would expect to isolate recombined GFP positive cells. However deeper analysis by costaining is required to compare the two transgenic lines.

When recombination was verified in the context of three loxP-modified alleles, loss of the respective proteins were largely confined to the main areas of Cre expression. Thus loss of CREB, CaMKIV and GR were extensive in striatum, layer VI of the cortex and CA2 of the hippocampus (Figure 8). However, induction of beta-galactosidase activity in R26R^D1Cre reporter mice showed a more widespread pattern of recombination in layer IV of the cortex and in the CA1 of the hippocampus (Figure 6). One possible explanation for this could be that scattered recombination, as caused by weak and transient Cre expression occurring at early postnatal stages (Figure 5), is easily underestimated when followed by loss of the respective protein but very visible when revealed by beta-galactosidase staining. This may be also related to a tissue-specific difference in the stability of CREB, CaMKIV and GR proteins in CA1 and layer IV.

In view of the patterns of recombination restricted to the major dopaminoceptive regions as seen in the context of the CREB, CaMKIV and GR mutations, the D1Cre line should be useful to evaluate the contribution of these, and others, signaling molecules to the dopaminergic system. The D1Cre line, combined with specific pharmacological treatment may provide a powerful tool do genetically dissect signalling pathways underlying dopaminergic neurotransmission.

We present the generation and characterization of a transgenic mouse line expressing the Cre recombinase under the control of the D1R regulatory elements. We show that this transgenic line can be used successfully for the conditional ablation of specific genes important for biological processes involving dopamine signaling.