BC1 RNA motifs required for dendritic transport in vivo

BC1 RNA is a small brain specific non-protein coding RNA. It is transported from the cell body into dendrites where it is involved in the fine-tuning translational control. Due to its compactness and established secondary structure, BC1 RNA is an ideal model for investigating the motifs necessary for dendritic localization. Previously, microinjection of in vitro transcribed BC1 RNA mutants into the soma of cultured primary neurons suggested the importance of RNA motifs for dendritic targeting. These ex vivo experiments identified a single bulged nucleotide (U22) and a putative K-turn (GA motif) structure required for dendritic localization or distal transport, respectively. We generated six transgenic mouse lines (three founders each) containing neuronally expressing BC1 RNA variants on a BC1 RNA knockout mouse background. In contrast to ex vivo data, we did not find indications of reduction or abolition of dendritic BC1 RNA localization in the mutants devoid of the GA motif or the bulged nucleotide. We confirmed the ex vivo data, which showed that the triloop terminal sequence had no consequence on dendritic transport. Interestingly, changing the triloop supporting structure completely abolished dendritic localization of BC1 RNA. We propose a novel RNA motif important for dendritic transport in vivo.

BC1− /− knockout mouse line was chosen as recipient background 11 . We examined the brains of the various mice by in situ hybridization with emphasis on the dendritic fields of the hippocampi. Once more, the in vivo results deviated to some degree from the previous ex vivo data 18 .

Results
Transgenes and RNA expression. We generated six transgenic mouse lines (Tg BC1_UCU , Tg BC1_AUC , Tg BC1_AG , Tg BC1_GAU , Tg BC1_A and Tg BC1 ), which expressed mutant rat BC1 RNA as well as the rat BC1 wild-type molecule. Rat and mouse BC1 RNA are virtually identical. The genetic background for all transgenic lines was FVB/N lacking the BC1 RNA gene (BC1− /− ) 11 . We generated at least three founders for each line and examined them via in situ hybridization.
Transgene integration was confirmed by digestion with EcoRV followed by Southern blot analysis (Fig. 1). Typically, we estimated one to five transgene copies at different integration sites. We monitored the expression of each mutated RNA by Northern-blot analysis using total mouse brain RNA. BC1 RNA expression levels in Tg BC1 , Tg BC1_AUC and TG BC1_UCU were comparable to the BC1 RNA expression levels of non-transgenic wild-type mice. When compared to wild-type mice, about two thirds of the expression levels were observed in In Tg BC1_AG , about half in Tg BC1_A , and only about a quarter in Tg BC1_GAU (Fig. 2).
Radioactive in situ hybridization. For in situ hybridization (ISH) we used wild-type mice and at least three founders for each transgenic mouse line. The probe was in vitro transcribed RNA complementary to the 3′ unique part of BC1 RNA (see Materials and Methods). Control brain slices from BC1− /− animals showed no signal either using sense or antisense RNA probes (data not shown). The ISH on wild-type brain slices with the anti-sense probe showed a clear signal in the CA1-CA3 hippocampal areas including the corresponding dendritic fields (Fig. 3C). As expected, transgene-derived RNA variants were prominently expressed in the mouse brain including cortex and hippocampus. As in wild-type mice, no BC1 RNA signal was found in the dentate gyrus.  Northern-blot analysis of total brain RNA from BC1 transgenic mice. Northern blot autoradiographs of total brain RNA from BC1 transgenic mice, FVB/N wild-type and FVB/N BC1− /− mouse brains are shown in the upper panels (exposure time: ~20 h). Ethidium bromide stained tRNA loading controls are shown in the lower panels. The RNA expression signals of Tg BC1_UCU , Tg BC1_AUC and Tg BC1 are similar to BC1 RNA of FVB/N wild-type mouse. As expected, there is no signal in the lane corresponding to the BC1− /− mouse. The expression level of mutated BC1 RNA in Tg BC1_AG mice is around one third, in Tg BC1_A it is almost half and in Tg BC1_GAU mice reduced to a quarter, compared to wild-type animals. . In situ hybridization of FVB/N wild-type mouse brain. ISH of BC1 RNA in FVB/N wild-type mouse brain is shown as a control. The structure and the investigated structural motifs of BC1 RNA are shown on the left. Nucleotides corresponding to internal RNA polymerase III promoter elements box A and box B are highlighted in bold letters (A). Panel (B) depicts an autoradiography (exposure time: ~20 h). Signal (dark areas) is seen in many areas of the brain in cell bodies as well as, for example, in the dendritic fields of the CA1-CA3 area of the hippocampus. In gyrus dentatus, the signal is weak or absent as expected. Panel (C) is a dark field microscopy picture from the marked area in (B) at 200x magnification. The light spots correspond to the silver grains generated by the α -S 35 UTP radiolabeled probe. The signal in the hippocampal CA2 dendritic field indicates that BC1 RNA is transported to the distal extensions of dendrites.
Scientific RepoRts | 6:28300 | DOI: 10.1038/srep28300 Dark field microscopy depicted the CA2 field in which silver grains appearing as white spots show the hybridization along the entire length of the dendrites in BC1 wild-type mice (Fig. 3C). Tg BC1_AUC . In a previous experiment, it was suggested that the apical internal loop of BC1 RNA possibly forms a non-canonical K-turn structure (GA motif) 18,32 . The canonical K-turn motif is characterized by a three-nucleotide bulge, flanked with Watson Crick base pairs at the 5′ and non-canonical GA base pairs at the 3′ end 33,34 . It has been proposed that the GA motif may influence dendritic transport 34 and indeed, ex vivo, BC1 RNA devoid of the GA motif could only be detected in the proximal regions of dendrites 18 . Exchange of bases UAG [25][26][27] to AUC [25][26][27] was expected to eliminate the internal GA motif of BC1 RNA. The nucleotide exchange of AG [26][27] to UC 26-27 affected noncanonical GA base pairing within the predicted GA motif (Fig. 4B). The U to A mutation at position 25 could have potentially had an effect on the 3′ flanking stem and was introduced to reproduce the previously reported construct used for testing dendritic transport ex vivo 18 . This mutation could have potentially altered the apical internal loop. Additionally, the secondary structure of BC1 RNA predicts a putative pseudoknot structure involving UAG [25][26][27] and the terminal loop nucleotides CUA 37-39 (see below) 35 . Hence, the sequence change to AUC [25][26][27] would not only disturb the GA motif, but also this putative pseudoknot formation. Ex vivo experiments show a significant reduction of BC1 RNA in dendrites suggesting that this mutation might be important for dendritic targeting 18 . Our ISH experiments with Tg BC1_AUC mice revealed that mutated BC1 RNA distribution in dendrites was not markedly different from wild-type RNA (Fig. 4C,D).
GA-type targeting motif Tg BC1_UCU . In addition, we examined a BC1 RNA variant, in which the internal loop nucleotides AAG 48-51 had been replaced by UCU 48-51 in order to remove the GA motif by generating standard Watson Crick base pairing, thus extending the canonical RNA alpha helix (Fig. 5B). According to ISH of transgenic mouse brains, the mutated RNA is expressed over the entire brain in regions similar to wild-type animals (Fig. 5C). We were not able to detect a reduction in transport along the dendrites. We could show that this RNA is present in the soma of neurons and, additionally, along the entire lengths of the hippocampal dendrites (Fig. 5D).
Tg BC1_GAU . The 5′ domain of BC1 RNA forms a stem-loop structure with a three-nucleotide terminal loop. As mentioned above, a pseudoknot-structure could be predicted between the terminal loop and UAG [25][26][27] nucleotides of the apical internal loop 35 . Although in vitro chemical probing experiments did not support the formation By introducing the AUC triplet to the RNA, we altered the GA motif. This mutation has no effect on the distal dendritic transport, as can be seen in the autoradiography (exposure time: ~20 h) (Panel C). The signal is visible throughout the dendritic fields of the hippocampus. Panel (D) is a dark field microscopy picture from the marked area in (C) at 200x magnification. The light spots correspond to the silver grains generated by the α -S 35 UTP radiolabeled probe. The signal in the hippocampal CA2 dendritic field indicates that this variant of BC1 RNA is transported to the distal extensions of dendrites.
Scientific RepoRts | 6:28300 | DOI: 10.1038/srep28300 of the pseudoknot-structure, RNA folding in vivo could differ 20 . In order to test whether the sequence of the single-stranded terminal loop plays a role in dendritic transport, perhaps by interacting with another RNA region (intra-or intermolecularly), or direct binding with protein(s) important for RNA transport, we replaced the terminal loop nucleotides CUA 37-39 with GAU 37-39 . When ISH was performed on brain slides of Tg BC1_GAU mice we did not observe a reduction in the dendritic transport of mutated BC1 RNA in comparison to wild-type animals ( Fig. 6C,D). This is in agreement with ex vivo experiments 18 . Tg BC1_A . The most drastic effect observed on BC1 transport ex vivo is the removal of the bulged U-residue at position 22, resulting in the restriction of RNA to cell bodies 18 . Several attempts to generate a corresponding transgene failed, as the expression levels of the RNA were too low for meaningful analysis by in situ hybridization. In order to alternatively remove the bulge structure, we inserted an Adenosine residue after position 54. This generated a U-A base pair, extending the middle stem and thus, also removed the bulge at U 22 . In contrast to ex vivo RNA microinjection experiments, this Tg BC1_A RNA variant could clearly be detected in hippocampal dendrites of the respective mice (Fig. 7C,D). Tg BC1_AG . As mentioned in the introduction, rodent genomes contain a large number of ID repetitive elements 22,36 . In addition to other positions, members of two ID subfamilies have the following mutations: C 41 → G 41 (ID3, ID4) and G 35 → A 35 (ID4 only) in the terminal stem loop domain at the penultimate base pair abutting the triloop 20,22 . Previous ex vivo studies tested two BC1 RNA variants where this penultimate base pair is altered to noncanonical G 35 -G 41 and A 35 -G 41 , respectively. Both RNA variants were not detected in dendrites under normal conditions but only after potassium polarization 37 . We therefore opted to alter the apical stem in BC1 RNA mimicking the ID4 structure in this region. Hence, we replaced the G 35 -C 41 Watson Crick base pair with A 35 -G 41 juxtaposition (Fig. 8B). The ISH experiments on Tg BC1_AG mouse brain slices revealed, in agreement with the aforementioned ex vivo experiments, no signal in the dendritic fields of the hippocampus (Fig. 8C). Also, no silver grains were observed in hippocampal dendrites (Fig. 8D). This indicates an inability of dendritic transport of this mutant BC1 RNA in vivo.

Figure 5.
In situ hybridization of FVB/N Tg BC1_UCU mouse brain with a mutation in the apical internal loop of BC1 RNA. Wild-type BC1 RNA and the apical internal loop mutation AAG 48-50 → UCU 48-50 is shown in panels A and B. This mutation leads to a loss of the apical internal loop by building standard Watson-Crick base pairs. The apical internal loop is predicted to form a GA motif and therefore to play a role in dendritic transport. The autoradiography film (exposure time: ~20 h) (Panel C) shows the expression of mutated BC1 RNA. Panel D is a dark field microscopy picture from the marked area in (C) at 200x magnification. The light spots correspond to the silver grains generated by the α -S 35 UTP radiolabeled probe. The signal in the hippocampal CA2 dendritic field indicates that this variant of BC1 RNA is transported to the distal extensions of dendrites.

Discussion
Much is known regarding the movement of RNA between cellular compartments; this includes the transport of selected mRNAs and various other RNAs, mostly those involved in translation and its regulation from the cell bodies of neurons to their dendritic processes 38 . BC1 RNA is a preferred model to study the relationship between RNA structure and its ability to be transported due to its size (~150 nt), the availability of secondary structure and the absence of posttranscriptional modifications 19,20 . Transport competence of BC1 RNA has previously been assayed by microinjection of in vitro generated RNA into the cytoplasm of sympathetic neurons in culture 18,29,39 . These studies suggest the existence of at least two elements within the 5′ stem-loop domain of BC1 RNA; the integrity of which appear to be essential for dendritic transport. The first is a GA motif containing two noncanonical A⋅ G pairs 18,34 . This structure has been implicated to be important for transport into the distal regions of dendrites 18 . The second essential motif for dendritic transport ex vivo is a single base bulge. When the corresponding base (U 22 in the medial stem) is deleted, RNA transport is completely abolished 18 . A more genuine approach is the investigation of dendritic localization of RNA by expressing various constructs in the brains of transgenic mice (in vivo). Previously, there were conflicting ex and in vivo data concerning the ability of the 5′ stem of BC1 RNA to convey transport competence to chimeric mRNAs. Sequences identical or resembling BC1 RNA were fused to a reporter mRNA (e.g., bicoid RNA from D. melanogaster 18,33 ) or inserted into the 3′ UTR of reporter mRNAs 31,39 . The working hypothesis was that dendritic mRNAs might evolve by fortuitous insertion of repetitive ID elements harboring the 5′ stem of BC1 RNA into untranslated regions (UTRs) of an mRNA encoding gene via retroposition. As a consequence, the targeted mRNA might have acquired dendritic transport competence. Injection of reporter mRNA with or without ID element into the cytoplasm of primary culture neurons (ex vivo) appeared to confirm this idea 18,37,40 . However, analogous experiments using transgenic mice could not confirm dendritic location of reporter mRNAs with any variant of the repetitive ID domain of BC1 RNA or even full-length BC1 RNA integrated into the 3′ UTR of a reporter mRNA in either orientation 31 .
Pilot experiments showed that a rat BC1 gene fragment including flanking regulatory regions (~1.4 kb) could be specifically expressed in mouse brains. Although on Northern blots we could differentiate between the transgenic and endogenous RNAs with a short specific antisense oligodeoxynucleotide (not shown), we could not hope to achieve discrimination for in situ hybridization due to the high similarity between the rat and mouse RNA (only two nucleotide substitutions). Hence, a modified minigene was constructed replacing 20 A-residues adjacent to the 5′ stem ( Fig. 3A) with the triplet CGG. In parallel, we crossed our BC1− /− mice into the FVB/N genetic background facilitating the generation of transgenes by pronuclear injection. This eliminated any concern about possible cross-reactivity with endogenous BC1 RNA, rendering the mini-gene obsolete. We nevertheless tested the minigene along with a rat wild-type BC1 RNA transgenic construct. In situ hybridization revealed that both RNAs had a distribution in brain that was indistinguishable from endogenous BC1 RNA in wild-type mice (not shown). A major difference was that the minigene was expressed at significantly lower levels. In general, we observed that transgenes expressed BC1 RNA variants at levels that were comparable to wild-type but some expression was reduced to as low as 25% (see Fig. 2 and Table 1). Importantly, expression levels were never higher than that of endogenous BC1 RNA preempting arguments that overloading the system might lead to aberrant dendritic localization 37 . One factor for expression levels varying from construct to construct might be a correlation with copy numbers and/or loci of integration. Once more, the neuronal expression in specific brain areas of the RNAs apparently did not differ from that of wild-type mice, indicating that all necessary regulatory elements upstream from the RNA coding region are present on the microinjected 1.4 kb fragment. The different expression levels of the various constructs could also be due to alterations that affect the transcription efficiency and/or stability of the variant RNAs. In fact, in vivo expression levels sufficient for in situ hybridization could not be achieved in a U 22 deletion construct because the RNA polymerase III A-box promoter consensus sequence was affected (Fig. 3A) 21 . To avoid any consequence on promoter activity, we abrogated the bulge by inserting an A-residue on the opposite side of the stem (between C 53 and C 54 ). While RNA expression was now sufficient, it only reached about 50% of wild-type levels. The two other constructs with lower expression (Tg BC1_GAU and Tg BC1_ AG ) featured nucleotide changes that neither affect box A nor box B (for promoter boxes see Fig. 3A). Despite the lower level of expression for some of the constructs, we could readily identify the presence or absence of dendritic transport of all BC1 RNA variants in our transgenic mouse models. GA motif. The GA motif implied in conveying transport competence of RNAs in general and of BC1 RNA in particular was modified in two different ways: First, in Tg BC1_AUC , we changed the sequence of the 5′ section of the loop, thus in all likelihood eliminating the GA motif (Fig. 4A). The second (Tg BC1_UCU ) was to completely replace the apical loop harboring the GA motif by replacing all unpaired as well as noncanonical pairs with bases that form canonical Watson-Crick pairs, thus lengthening the apical stem by three base pairs to altogether fourteen  (Fig. 5A). Both variant RNAs were transported into dendrites, albeit not into proximal extensions (up to ~150 μ m) ex vivo, while transgenic mice expressing these mutant RNAs were virtually indistinguishable from wild-type mice with respect to their location in hippocampal dendritic fields (Figs 4 and 5).
Triloop sequence. Sequence alteration of the terminal triloop by exchanging all three nucleotides did not impede dendritic transport in either study 18 ruling out a significance of the triloop sequence as well as the potential formation of a pseudoknot involving the UAG at positions 25-27 and the CUA of the triloop for dendritic transport.  (Panel A,B). ID repetitive elements are abundant in rodent genomes. Ex vivo, certain ID elements (subclasses ID1 and ID2, but not ID3 or ID4) have been shown to convey dendritic transport to reporter mRNAs. We introduced one of the ID4 hallmark changes into BC1 RNA in order to mimic the ID4 structure at the apical stem. This leads to a loss of transport along the dendritic fields of the hippocampus shown on the autoradiography film (exposure time: ~20 h) (Panel C). Panel D is a dark field microscopy picture from the marked area in (C) at 200x magnification. The absence of signal in the hippocampal CA2 dendritic field indicates that this variant of BC1 RNA is not transported into dendrites.   Bulges. From ex vivo data 18 and our gradually accumulating in vivo data, we did not expect any effect from the removal of the basal internal loop (positions U 14 -C 61 , see last row of Table 1) and thus, did not generate a corresponding transgene. We rather focused on the in vivo analysis of the mutation involving the bulged U-residue at position 22. Removal of this base, thus extending the medial stem to ten bp, virtually abolished dendritic transport ex vivo (Table 1) 18 . Surprisingly, in situ hybridization signals from our construct when expressed in transgenic mice were well represented in the dendritic fields of the hippocampus (Fig. 7). For the sake of completeness, it should be mentioned that we were unable to obtain useful expression levels of the aforementioned construct in transgenic mice. Therefore, our construct for in vivo analysis was slightly different; instead of removing U 22 , we inserted an A-residue between C 53 and C 54 . Consequently, the resulting medial stem is extended by a U-A pair and is eleven instead of ten bp long, an alteration that is not expected to have a notable impact on the dendritic transport of the RNA variants.
Triloop structure. The final construct tested, was perhaps the most intriguing. In Fig. 6 we showed that the sequence of the terminal loop is interchangeable and we did not expect much of an effect by replacing the penultimate G-C base pair of the stem closing the triloop into a noncanocical A⋅ G juxtaposition. The rationale to chose this variation was based on the following observation: Ex vivo experiments using microinjection of in vitro transcribed radiolabelled tubulin reporter mRNA with various constructs inserted into the 3′ UTR into the cell bodies of primary neurons found that the 5′ BC1 RNA stem (pos. 1-74) or slight variants thereof, as represented by some ID repetitive SINE elements (subclass ID1 and ID2) impart dendritic transport behavior on the reporter mRNA 37 . In contrast, subclass ID3 and ID4 (featuring G⋅ G and A⋅ G juxtaposition at the aforementioned base pair), respectively, failed to do so under normal conditions. The authors claim that dendritic transport is restored by increased KCl concentration and suggested a model of activity dependent RNA transport in neurons 37 . In any event, although our previous study did not reveal dendritic transport with BC1 RNA or various ID elements inserted into the 3′ UTR of EGFP reporter mRNA 31 , we reproduced the corresponding variation in the penultimate base pair abutting the triloop to a transgene. Interestingly, in mice, perhaps with the exception of some transport over short distances into proximal parts of neuronal processes, there was a complete absence of in situ hybridization signal in dendritic fields of the hippocampus. This is consistent with ex vivo experiments using the corresponding unfused BC1 RNA (A 35 ⋅ G 41 ) 37 . Among all possible explanations, we would like to offer the following: Exchange of a G-C pair with an A⋅ G juxtaposition potentially changes the structure important for triloop formation. Hence, the triloop structure and not the corresponding unpaired loop bases is an important determinant for dendritic transport of BC1 RNA. From our in vivo study of dendritic transport, it appears that the two structural elements implied in the past, namely the apical GA motif and the single bulge, which have been shown ex vivo to alter the intracellular localization of BC1 RNA 18,32 , do not markedly diminish RNA dendritic transport in hippocampal neurons of transgenic mice. We are aware of the possibility that this might differ in other neuronal cell-types or as a consequence of experimental treatment 37 . Here, we identified the structure supporting the triloop as a strong candidate for conferring dendritic transportability to BC1 RNA. As shown with the transgene expressing the Tg BC1_GAU variant, the actual three-base sequence of the loop is apparently irrelevant. This is further supported by phylogenetic information: in squirrel (Sciurus carolinensis) the unpaired loop sequence is altered from CUA → CUU 41 (and unpublished data).
Despite the fact that transgenic mouse experiments are time consuming, animal welfare restricted and expensive, we felt that in vivo analysis of different BC1 RNA structures influencing dendritic targeting was an important contribution to our understanding of structure/function relationships for RNA transport.

Materials and Methods
Generation of the rat BC1 RNA transgenic mice. For transgenic mouse generation we used the Sleeping Beauty transposase system in vector pT2BH 42 . The SacI-BamHI 1.4 kb DNA fragment containing the wild-type rat BC1 RNA gene together with 5′ and 3′ flanking regions from the pBC1 plasmid 21 was cloned into EcoRV-BglII sites of the pT2BH. The resulting plasmid pT2BH_BC1 was used for further mutagenesis. Transgenic mice were generated through regular pronuclei injection in FVB/N BC1− /− embryos as described previously 31,43 . The Tg BC1_AUC and the Tg BC1_GAU mutations were generated in the pT2BH_BC1 using pairs of oligonucleotides UAG_AUC_DIR/UAG_AUC_REV and CUA_GAU_DIR/CUA_GAU_REV respectively ( Table 2). The Tg BC1_AG mutation was generated in pBC1 directly with the aid of oligonucleotides GC-AGfsp/GC-AGrsp (Table 2). Mutagenesis was performed using the PCR approach utilizing Taq and Fusion thermostable enzymes  with consecutive sequencing verification. The Tg BC1_UCU and Tg BC1_A constructs were synthesized and cloned into the pT2BH vector commercially (Blue Heron, Bothell, WA, USA).

DNA isolation and Southern blot analysis.
We performed tail biopsies from three-week-old transgenic mice. The DNA was isolated via the phenol/chloroform extraction method. The DNA was digested with EcoRV restriction enzyme (Fermentas) and separated on a 0.8% agarose gel. Following electrophoresis, the DNA was transferred by capillary blotting onto a GeneScreen hybridization membrane (Perkin Elmer). BC1 transgenic probe 800 bp DNA fragment downstream from an EcoRV site located in the 3′ flank of the BC1 RNA gene was labeled with [γ -32 P]-dCTP (PerkinElmer) and high prime DNA labeling Kit (Roche).
RNA isolation and Northern blot analysis. Total brain RNA was isolated using the Trizol reagent according to the manufacturer's manual. Ten μ g of total RNA was separated on 8% (w/v) polyacrylamide [29:1 acrylamide/bisacrylamide], 7M urea gel and electro-transferred onto a positively charged nylon membrane (Roche). Hybridization probe (BC1 unique deoxyoligonucleotide, 50 pmol, Table 2) was radioactively labeled using [γ -32 P]-ATP (PerkinElmer) and T4 polynucleotide kinase (Fermentas) according to the manufacturer's instructions. Northern blot analysis was performed as previously described 44 . To estimate expression levels of wild-type BC1 RNA and variants, we used the Advanced Data Image Analyzer software (AIDA Vers4.26.038, Raytest). Densiometry of the autoradiography film revealed expression of the mutated BC1 RNAs ranging from around one quarter (~24%) to almost equal (~90%) compared to endogenous wild-type mouse RNA (Table 1).
Radioactive in situ hybridization. Mice (aged 8 to 24 weeks) were sacrificed with CO 2 and subsequently perfused with 1x PBS media and 4% paraformaldehyde. Brains were sectioned with a vibratome (Leica) into 30 μ m coronal slices. Hybridization was performed as previously described 45 . RNA probes were generated from pMK1 plasmid as template 24 . DNA was linearized with SacI and transcribed by T3 RNA polymerase for the sense probe. The antisense probe was generated by T7 RNA polymerase from pMK1 linearized with Asp718I. In both cases, in vitro transcription was performed in the presence of α -S 35 UTP. RNA polymerases and restriction enzymes were purchased from Thermo Scientific. pMK1 sense probe: 5 ′ -gg ga ac aa aa gc ug gg ua cc AA AA AA GA CA AA AU AA CA AA AA GA CC AA AA AA AA AC AAGGUAAC UGGCACACACAACCUUUUg-3′; pMK1 antisense probe: 5 ′ -gg gc ga au ug ga gc uc AA AA GG UU GU GU GU GC CA GUUA CC UU GU UU UU UU UU GG UCUUUUUGUUAUUUUGUCUUUUUUg-3′ . Lower case letters represent transcribed vector sequences. For hybridization, we used RNA with 4.5 × 10 6 cpm and incubated in hybridization buffer at 50 °C overnight. The slices were washed once with 2x SSC and deionized formamide at a ratio of 1:1 at 50 °C and once with 2x SSC at the same temperature. Subsequently, we treated the slices with 30 μ g/ml RNase A at 40 °C and washed again in 5 liters 2x SSC at 50 °C for 3 h and continued washing in the same buffer at room temperature overnight.
The slices were mounted onto Superfrost Gold Plus (Menzel) microscopy slides, dried with increasing concentrations of 50%, 70% and 90% ethanol and exposed to autoradiography films (Kodak Biomax MR) overnight.
After developing the films the slides were dipped into autoradiography emulsion (Kodak NTB) as previously described 24 and kept at + 4 °C for 10 days. To develop the slides, we used developer D-19 and Rapid Fixer (Kodak). The brain slices were stained with cresyl violet using an autostainer (Leica). Dark field pictures were taken with an Axio imager microscope (Zeiss) at 200x magnification.
Mice. All mouse procedures were performed in compliance with the guidelines for the welfare of experimental animals issued by the Federal Government of Germany and approved by the State Agency for Nature, Environment and Consumer Protection North Rhine-Westphalia (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein Westfalen) 45 . Animals were kept in specific pathogen-free animal facilities. All breeding was conducted in a controlled (21 °C, 30-50% humidity) room with a 12:12 hour light-dark cycle. Mice were housed under non-enriched, standard conditions in individually ventilated (36 (l) × 20 (w) × 20 (h) cm) cages with up to five littermates. Pups were weaned 19-23 days after birth and females were kept separately from males 8 .