Introduction

Dopamine(DA) is an important neurotransmitter in the central nervous system. Though produced in limited cell populations throughout the mouse brain1, dopaminergic axons project widely throughout the nervous system, modulating a wide variety of circuits. Most notably, highly robust projections from the dopaminergic cells of the substantia nigra(SN) and ventral tegmental areas(VTA) to the striatum are essential for modulating behavior. Projections from VTA neurons to the nucleus accumbens have long been known to play a fundamental role in reward and are thought to be the common convergent pathway for all drugs of abuse2, while the dorsal-lateral striatum has a greater role in motor behavior3. Finally, both populations of neurons, but especially the SN, are vulnerable to genetic or environmental insults that result in their degeneration in patients with Parkinson’s disease. Thus, DA producing neurons have been a focus of intense scientific interest for decades with a deep cannon of accumulated knowledge about their morphology, projections, function, and physiology.

To enable study of translation specifically in DA producing cells of the mouse brain, we developed a transgenic mouse line expressing a ribosomal protein fused to GFP, eGFP/RPL10A, in DA neurons to permit Translating Ribosome Affinity Purification(TRAP) from these cells. Here we provide characterization of the expression of this line and validation of its ability to harvest mRNA from midbrain DA neurons. This line has now been distributed to Jackson laboratories(Stock# 030272) and should provide a resource for investigators interested in studying transcription and translation in these cells.

Results

We generated two transgenic mouse lines to target this population of neurons. We first used a bacterial artificial chromosome(BAC) containing the tyrosine hydroxylase (Th) gene, a key enzyme in the synthesis of DA and norepinephrine which has traditionally been used as a marker of these cells, by replacing the coding sequence with the eGFP/RPL10a transgene. Characterization of eGFP expression in this mouse line showed some robust expression in the regions where DA neurons were known to be found, but also ectopic expression in TH negative populations in hypothalamus, striatum, and even sparse cells in cortex. There was also labeling in a subset of Purkinje neurons in the cerebellum, though this later pattern was somewhat consistent with some prior immunohistochemical data reporting TH expression in Purkinje cells4 and suggests that the enzyme has purposes in the CNS beyond the synthesis of DA and norepinephrine. It is also possible that this discrepancy reflects some level of difference between transcription and translation of the Th gene, and is consistent with a recent report of more widespread expression of Th mRNA than protein5. Although a pilot study demonstrated that TRAP could harvest RNA from midbrain dopaminergic cells(data not shown), concerns about ectopic expression precluded further pursuit of this line. Thus, these first mice lacked specificity for DA producing neurons.

Therefore we next tested a BAC containing the Slc6a3 gene, coding for the protein commonly known as the DA transporter(DAT). Immunohistochemical characterization of this mouse line revealed robust expression of eGFP/RPL10a in midbrain DA producing neuronal populations (Fig. 1A). Colabeling with TH antibodies revealed that TH positive neurons were consistently eGFP positive in these populations (Fig. 1B). We then isolated ribosome bound RNA from adult midbrain DA neurons and measured gene expression by microarray. Independent replicates showed high reproducibility (Fig. 2A). In addition, TRAP RNA was markedly different from parallel profiles of input RNA purified from the whole midbrain dissection (Fig. 2B). Specifically, a variety of known DA neuronal markers including Th (48 fold), Slc6a3 (6.8 fold), and Ntsr1(19.9 fold), were all significantly enriched in the TRAP sample (p < 0.003, p < 0.005, p < 0.05, respectively; LIMMA, with FDR correction). We believe the relatively lower enrichment of Slc6a3 likely represents saturation of the microarray probeset for this transcript in the TRAP sample, as there is no a priori reason to assume it should be substantially less enriched than Th, and the raw intensity values for the TRAP Slc6a3 probesets are in the top 0.05% of all probesets on the array. Non-neuronal ‘negative control’ transcripts were moderately depleted at a level typical of this TRAP protocol6.

Figure 1
figure 1

Anatomical confirmation of eGFP/RPL10A expression in midbrain dopamine neurons. (A) Anti-GFP immunohistochemistry shows regional expression of eGFP/RPL10A in neurons, consistent with expression in midbrain dopamine cells. (B) Immunofluorescence colocalization confirms expression in all Th positive neurons of the midbrain and only in Th positive neurons.

Full size image
Figure 2
figure 2

Slc6a3 JD1640 TRAP line allows for reproducible and specific purification or mRNAs from midbrain dopamine neurons. (A) Illustration of the bacTRAP method. Specific cell types are driven to express a GFP tagged ribosomal protein(RPL10A) using a cell type specific promoter in engineered from a bacterial artificial chromosome (BAC). After validation of the line, midbrains can be homogenized and lysates incubated with anti-GFP coated magnetic beads to enrich for mRNA from DA neurons. (B) Two microarray experiments on independent pools of Slc6a3 JD1640 TRAP mice result in reproducible mRNA expression (Pearson r > 0.98). (C) Comparison of TRAP purified from mRNA from Slc6a3 JD1640 TRAP to total RNA of midbrain shows robust enrichment of transcripts known to be expressed in dopaminergic neurons (blue), and moderate depletion of known glial genes (red). Lines in A and B show 2-fold enrichment or depletion (Pearson r > 0.90). (D) Hierarchical clustering of a variety of TRAP microarray datasets based on all transcripts with a pSI < 0.001 in any cell type reveals midbrain dopamine neurons(blue) are most similar to other neuromodulatory cells in enriched gene expression.

Full size image

As a final validation of the new mouse reagent, we defined the set of most SN/VTA enriched transcripts (Table 1). Examination of a subset of these genes with publically available coronal in situ hybridization data confirmed high levels of expression for all in a pattern consistent with midbrain DA neurons (Fig. 3). Likewise, an analysis of enrichment of particular biological and molecular processes of the top enriched (pSI < 0.005) transcripts revealed enrichment of categories for “Ion Channel Activity” (p < 4.5E-5, Benjamini Hochberg corrected p-value), driven by transcripts such as Kcnd3, Scn3a, and Chrna6, “Dopamine Biosynthetic Process” (p < 1.3E-3), driven by transcripts for enzymes such as Th, Gch, and Ddc 7, and “Dopaminergic Neuron Differentiation” (p < 1.2E-2), driven by known master regulators such as En1, Foxa2, Nr4a2, Pitx3, and Lmx1b 8, consistent with identification of transcripts enriched in midbrain DA neurons.

Table 1 Probesets with >10 fold enrichment, p < 0.01, and pSI < 0.10e-6.
Full size table
Figure 3
figure 3

Examples of highly enriched transcripts in midbrain dopamine neurons. Shown are the first 9 genes of Table 1 with coronal in situ hybridization patterns available in the Allen Brain Atlas25. All show a regional pattern consistent with high and or/specific enrichment in midbrain dopamine neurons. Some (e.g. Anxa1) appear to only be expressed in a subset of the neurons, while others (Cpne) are highly expressed in SN/VTA but also show expression in additional populations. Images are from http://mouse.brain-map.org.

Full size image

Discussion

We report here the generation and characterization of a mouse line capable of translationally profiling midbrain dopamine-transporter expressing neurons. We show robust expression in all midbrain Th positive neurons, and confirm the ability of the mouse line to enable translational profiling. Thus, the mouse line should be useful to investigate ribosome bound transcripts in these neurons both at baseline, and in response to experimental manipulations, such as stimulation by drugs of abuse. We have also recently adapted a procedure for nuclear RNA purification from TRAP mouse lines9, so both nuclear transcription and cytoplasmic translation are in theory accessible using this line, and others have shown that the same basic approach can be applied to study the epigenetic profile of specific neuronal cell types10.

It is an interesting question as to why the Th bacTRAP line showed expression outside of TH positive cells. In the modified BAC designed to express eGFP/RPL10A in lieu of TH, the eGFP/RPL10a transgene sequence, followed by a strong polyA signal, is inserted at the translational start site of the Th gene. Thus, the transgene will co-opt the promoter/enhancers of the Th gene, due to the polyA signal but not the 3′ UTR. Thus, to the extent TH mRNA might be transcriptionally expressed, but translationally suppressed by 3′ UTR sequences in ‘ectopic’ populations, one would expect to detect eGFP/RPL10a protein in cells where TH protein is absent. Such translational regulation might also explain why many Th Cre lines, including knock-ins and transgenics, show recombination in Th negative cells11.

We also note that the new Slc6a3 TRAP line is distinct from the one recently used to successfully profile midbrain DA neurons in an MPTP Parkinson’s model12. Both lines were generated by modifying the same initial BAC, however the current line was initiated on the FVB strain and subsequently backcrossed to C57BL6/J mice, while the other line was directly generated on C57 mice from Charles River. Thus there will be modest strain differences between the two because of the different sources of C57 mice and any remaining introgressed FVB alleles in linkage with the transgene. Also, as they are separate integrants, they are expected to differ in transgene copy number and location. However, the line reported here is the exact line used to successfully profile embryonic midbrain DA neurons5. As the current line is being released by Jackson labs, we have provided this brief report to provide details of the generation and characterization of the line for future investigators, especially as minor differences between expected and actual strain or copy number could influence experimental results. Also, though they were generated independently of GENSAT, neuroanatomical data for this and other published bacTRAP lines are now also being hosted at the GENSAT website for the convenience of the field.

The translational profile of DA neurons provided here may also be of use to investigators interested in these cells. For example, it has recently been noted that midbrain DA neurons can co-release the neurotransmitter GABA13, yet do not appear to contain the traditional GABA synthesizing enzymes GAD65 and GAD6714 coded for by the genes Gad1 and Gad2. Our microarray data is consistent with this expression pattern, showing robust depletion of these two genes from the TRAP sample (e.g. Gad1 6588 arbitrary expression units in the input RNA, and only 2396 in the TRAP, with this remaining TRAP signal likely reflecting non-specific background). However, the Gad paralog GadL1 is 19 fold enriched in midbrain DA neurons (p < 0.002), though likely expressed only at low levels (absolute expression level: 140.8). Although normally thought to be involved in synthesis of Carnitine or Taurine rather than GABA15, 16, several unstudied splice isoforms are present in the UCSC genome browser. If any of the unstudied splice isoforms do produce GABA this might contribute to GABA neurotransmission from DA cells. Thus, to enable investigators to query our findings for additional insights, the complete analyzed data is provided as Supplemental Table 1.

Methods

Generation of mouse line

All procedures involving animals were approved by the Animal Studies Committee of Washington University in St. Louis and the Rockefeller University Institute Animal Care and Use Committee. All methods were carried out in accordance with relevant guidelines and regulations. A BAC RP24-269I17, was modified as described17, to insert the TRAP construct EGFP/RPL10A18 into the translation start site of the Slc6a3 gene. Recombination was confirmed by Southern blot. Modified BAC DNA was purified by CsCl centrifugation and injected into fertilized FVB oocytes as described19. Founder mice were bred to C57BL6/j mice for subsequent generations. Mice were maintained as trans-heterozygotes and transgene carrying pups were identified at each generation by genotyping tail clip DNA for GFP.

Translating Ribosome Affinity Purification

Five adult mice per replicate were euthanized with CO2, and midbrains were dissected in ice cold buffer in the presence of cyclohexamide to stall translation. TRAP was conducted as described20. Parallel input fractions were collected from each replicate as a measure of whole midbrain tissue RNA composition. RNA quantity and quality were determined with a Nanodrop 1000 spectrophotometer and Agilent 2100 Bioanalyzer with PicoChip reagents. All RINs were above 7. For each replicate, up to 10 ng of total RNA was amplified with the Affymetrix two-cycle amplification kit and hybridized to Affymetrix 430 2.0 microarrays according to the manufacturer’s instructions, and data were processed in R as described6, using GCRMA for normalization and identification of specific and enriched genes using the pSI package6 with default settings, compared to a large set of prior cell types analyzed by TRAP on the same microarray platform19, 21,22,23,24.

Anatomical analysis

For immunohistochemistry(IHC), mice were euthanized then perfused transcardially with PBS, followed by PBS with 4% paraformaldehyde. Brains were dissected, cryoprocted with 30% sucrose in PBS then processed MultiBrain Technology (NSA, NeuroScience Associates, Knoxville, TN) for DAB IHC with a 1:75,000 dilution of Goat anti-EGFP serum according to the Vectastain elite protocol (Vector Labs, Burlingame, CA). Serial sections were digitized with a Zeiss Axioskop2 microscope at 10× magnification.

For immunofluorescent studies, brains were prepared as above, frozen and sectioned to 40 uM on a crytostat, and stored in PBS with 0.1% azide until use. Sections were blocked with 5% normal donkey serum and 0.25% triton and then incubated with Chicken anti-GFP(Abcam) and Mouse anti-Th (Chemicon) followed by appropriate Alexa dye-conjugated secondary antibodies (Invitrogen, Carlsbad, CA). Images were acquired as Z stacks (2 μm sections) with a Zeiss Inverted LSM 510 confocal microscope.

For Allen Brain Atlas images of Fig. 3, we selected for presentation the first 9 available coronal in situ hybridization image sets, alphabetically, from those transcripts with > 10 fold enrichment, p < 0.01, and pSI < 0.10e-6 (Table 1).

Data Availability

Analyzed data are available as Supplemental Table 1. Raw data are available at GEO: GSE99927.