NMDA induces post-transcriptional regulation of alpha2-guanylyl-cyclase-subunit expression in cerebellar granule cells.

Activation of N-methyl-D-aspartate (NMDA) glutamate receptors commonly affects gene expression in different neurons. We reported previously that chronic treatment of rat cerebellar granule cells with NMDA (24 hours) upregulates the expression of mRNA encoding the α2 subunit of the nitric-oxide-sensitive guanylyl cyclase. However, the molecular mechanisms involved in this process remained to be elucidated. Here, we have performed mRNA-decay experiments using the transcriptional inhibitor actinomycin D, providing evidence that the half-life of α2 mRNA is significantly prolonged in cells exposed to NMDA. The role of the 3′ untranslated region of the α2 transcripts in NMDA-induced mRNA stabilisation was examined and an association between the RNA-binding proteins AUF1 and ELAV-like protein 1 (HuR/HuA), and endogenous α2 mRNA was demonstrated in vivo, as revealed by coimmunoprecipitation experiments with specific antibodies against AUF1 and HuR. Further studies indicated that stimulation of the NMDA receptor induces a downregulation in AUF1 levels stabilising the α2 mRNA transcripts. These events are triggered through a mechanism that depends on formation of nitric oxide, and on the subsequent activation of guanylyl cyclase and cGMP dependent protein kinases.


Introduction
Cerebellar granule cells express both Ca 2+ permeable Nmethyl-D-aspartate (NMDA) and non-NMDA glutamate receptors (Garthwaite et al., 1986;Meier and Jorgensen, 1986;Pearce et al., 1987). Indeed, stimulation of NMDA receptors in these cells is essential for survival and differentiation during their development, and it also plays a fundamental role in neuroplasticity (Kleinschmidt et al., 1987;Balazs et al., 1988;Bessho et al., 1994;Contestabile, 2000). Formation of nitric oxide (NO) is one consequence of NMDA receptor stimulation (Garthwaite and Boulton, 1995) and because neuronal NO synthase (nNOS) is anchored to NMDA receptors through the postsynaptic density protein 95 (PSD-95), Ca 2+ entry through these receptors activates nNOS (Brenman et al., 1996;Craven and Bredt, 1998;Sheng and Pak, 2000). Hence, it has been proposed that NO acts as a messenger during development and synaptic plasticity, as well as in the cell-death process that can also be initiated through NMDA-receptor activation (Dawson et al., 1993;Bredt and Snyder, 1994;Holscher, 1997;Contestabile, 2000). What is more, the activity of this receptor dynamically regulates nNOS expression in the cerebellum and in cerebellar granule cells (Baader and Schilling, 1996;Virgili et al., 1998), and mRNA expression of NO-activated guanylyl cyclase subunits in cultured granule cells, the mRNA of the ␣ 2 subunit being the most strongly regulated (Jurado et al., 2003).
It is thought that the expression of nNOS in developing cerebellar granule cells is regulated by excitatory neurotransmission and that Ca 2+ is an important signal transduction molecule involved in this regulatory process (Baader and Schilling, 1996). Nevertheless, the mechanism by which NMDA augments the expression of the ␣ 2 subunit mRNA in granule cells remains to be elucidated.
Post-transcriptional gene regulation, particularly events affecting mRNA stability, are emerging as crucial factors in the regulation of gene expression (Sachs, 1993). Although the mechanisms determining mRNA turnover are poorly understood, they are generally believed to involve the recognition of specific sequences by RNA-binding proteins. The best characterised RNA sequences involved in stabilising mRNA are the AUUUA-motifs (AU-rich elements, AREs). These elements are usually found in the 3Ј untranslated regions (3Ј UTR) of mRNAs, such as those encoding cytokines and growth factors (Chen and Shyu, 1995;Xu et al., 1997). To date, such sequences have been found in many different mRNAs, including several components of the NO-cGMP pathway, the activities of which are known to be regulated by altering mRNA stability, e.g. ␣ 1 and ␤ 1 subunits of guanylyl cyclase in rat aorta (Klöss et al., 2003;Klöss et al., 2004), NO synthase II in human mesangial cells (Pérez-Sala et al., 2001), and in human intestinal epithelium (DLD1) cells (Rodríguez-Pascual Activation of N-methyl-D-aspartate (NMDA) glutamate receptors commonly affects gene expression in different neurons. We reported previously that chronic treatment of rat cerebellar granule cells with NMDA (24 hours) upregulates the expression of mRNA encoding the ␣ ␣ 2 subunit of the nitric-oxide-sensitive guanylyl cyclase. However, the molecular mechanisms involved in this process remained to be elucidated. Here, we have performed mRNA-decay experiments using the transcriptional inhibitor actinomycin D, providing evidence that the half-life of ␣ ␣ 2 mRNA is significantly prolonged in cells exposed to NMDA. The role of the 3Ј Ј untranslated region of the ␣ ␣ 2 transcripts in NMDA-induced mRNA stabilisation was examined and an association between the RNA-binding proteins AUF1 and ELAV-like protein 1 (HuR/HuA), and endogenous ␣ ␣ 2 mRNA was demonstrated in vivo, as revealed by coimmunoprecipitation experiments with specific antibodies against AUF1 and HuR. Further studies indicated that stimulation of the NMDA receptor induces a downregulation in AUF1 levels stabilising the ␣ ␣ 2 mRNA transcripts. These events are triggered through a mechanism that depends on formation of nitric oxide, and on the subsequent activation of guanylyl cyclase and cGMP dependent protein kinases. et al., 2000). The stability of a particular mRNA species is controlled by specific interactions between its structural elements and RNA-binding proteins (Guhaniyogi and Brewer, 2001). The most extensively studied RNA-binding proteins that stabilise or destabilise different mRNAs are the AUF1 proteins and the ELAV (embryonic-lethal abnormal vision)like family that is comprised of four members: HuD (ElrD), HuC (ElrC, and ple21), He1-N1 (HuB, Mel-N1 and ElrB) and HuR (HuA and ErlA) (Fan and Steitz, 1998;Peng et al., 1998;Akamatsu et al., 1999;Bhattacharya et al., 1999;Lal et al., 2004).
The aim of the present study was to characterise the molecular mechanisms underlying the upregulation of ␣ 2 guanylyl cyclase subunit mRNA following stimulation of NMDA receptors in cultured cerebellar granule cells. We observed that the 3Ј UTR of ␣ 2 mRNA is very long and that it contains several AREs that are targets of elav family members and AUF1 proteins. Furthermore, we show that exposure to NMDA increases ␣ 2 mRNA stability by decreasing the levels of AUF1 proteins through a mechanism that requires NO and cGMP synthesis, as well as the activation of a cGMPdependent protein kinase.

Results
NMDA treatment increases the stability of mRNA transcripts encoding the ␣ 2 subunit of guanylyl cyclase in cerebellar granule cells We measured the amount of mRNA encoding the ␣ 2 subunit of guanylyl cyclase in granule cells when stimulated by NMDA. A time-dependent increase in the accumulation of ␣ 2 guanylyl cyclase mRNA transcripts was observed in the presence of NMDA. Indeed, the amount of mRNA progressively augmented during the exposure to NMDA until reaching a maximum at 24 hours of treatment. From this point onwards, continued exposure to NMDA produced a decrease in the levels of ␣ 2 mRNA (Fig. 1).
These results raised the question whether the upregulation of ␣ 2 mRNA upon exposure to NMDA depends on transcription or translation. To address this, we analysed the effect of NMDA in the presence of actinomycin D or cycloheximide, which inhibit transcription and translation, respectively. When granule cells that had been 7 days in vitro (7-DIV) were incubated in the presence of 50 M NMDA for 24 hours, the amount of ␣ 2 mRNA increased threefold ( Fig.  2A). Likewise, when translation was blocked by the addition of 5 g/ml cycloheximide, the amount of ␣ 2 mRNA also increased approximately threefold, although the subsequent Fig. 1. Time-course of the upregulation of guanylyl cyclase ␣ 2 mRNA induced by NMDA in cerebellar granule cells. NMDA 100 M was added to the culture medium and the total RNA extracted at the time points indicated. The RNA was reverse transcribed using random hexamers and the amount of ␣ 2 mRNA and 18S rRNA was determined by quantitative PCR using specific primers and probes for both. All the results were normalised against the 18S rRNA values and represent the mean ± s.e.m. of three experiments performed in triplicate. *P<0.001, significant difference from control values.   addition of NMDA did not produce a further increase in transcript levels. However, none of these treatments modified the total amount of RNA (control cells 10.65±0.95 g/5ϫ10 6 cells; NMDA-treated cells 9.90±0.63 g/5ϫ10 6 cells; cycloheximide-treated cells 9.4±1.26 g/5ϫ10 6 cells). Whereas the inhibition of transcription with actinomycin D led to an approximately eightfold decrease in the amount of ␣ 2 mRNA, NMDA was still capable of producing a threefold increase in these levels (Fig. 2B). These results suggest that the effect of NMDA does not depend on translation and augments the abundance of ␣ 2 mRNA by decreasing its degradation rate because the accumulation of a particular mRNA species is the result of the equilibrium between the rate of synthesis and the rate of degradation. Moreover, it has been shown that inhibition of protein synthesis increases AREs-containing mRNA stability in different systems (Altus et al., 1987;Oliveira and McCarthy, 1995;Ross, 1995;Guhaniyogi and Brewer, 2001) and, here, we observed an increase in ␣ 2 mRNA after cycloheximide treatment. To determine whether the stability of ␣ 2 mRNA is affected by NMDA, control and NMDA-treated cells were incubated in the presence of actinomycin D for different periods of time and the amount of ␣ 2 transcripts was measured.
In control cells, the amount of ␣ 2 mRNA dropped to 50% after a 5-hour treatment with actinomycin D (Fig. 2C). However, 58% of the initial amount of ␣ 2 mRNA remained after an 8-hour actinomycin D addition in NMDA-treated cells. As such, it appeared that NMDA might indeed increase the level of ␣ 2 mRNA by decreasing the degradation rate of this transcript.
␣ 2 mRNA has a long 3Ј UTR that contains several AU-rich sequences It is becoming clear that determinants in the 3Ј UTR of mRNA transcripts control aspects of their degradation. In different mRNAs, AREs present in the 3Ј UTR influence the stability of the mRNA transcripts, because the proteins that bind to these elements interact with, and stabilise or destabilise them. The 3Ј UTR of the rat ␣ 2 subunit of guanylyl cyclase had not been sequenced previously. Therefore, the 3Ј flanking region of sGC ␣ 2 mRNA from cerebellar granule cells was cloned and sequenced, in order to characterise the sGC 3Ј UTR in ␣ 2 transcripts. An oligonucleotide was designed into the coding region (F1; Fig. 3A), which, when used in combination with the oligonucleotide (R1), amplified a fragment of 843 bp. This fragment was amplified from mRNA because it was not detected in the absence of reverse transcriptase and, thus, was not amplified from contaminating DNA (Fig. 3B). The combination of a second pair of primers, Forward 2 (F2) and Reverse 2 (R2; Fig. 3A), amplified a 1176 bp fragment of the mRNA (Fig. 3B). The PCR products were sequenced and subcloned into the vector pCR ® II, and their identity was confirmed by restriction analysis (results not shown). The two fragments amplified were overlapping, the complete fragment corresponding to the 3Ј UTR of ␣ 2 mRNA being 1873 bp long Journal of Cell Science 119 (8) (GenBank accession number AY795577). However, we believe the complete 3Ј UTR is 100-150 bp longer because the PCR fragment obtained by using the Forward 2 and an oligo-dT primer is larger than the fragment obtained with the F2-R2 primer pair (Fig. 3B). Although we did not clone or sequence these additional 100-150 bp, it is clear that the sGC ␣ 2 mRNA has a 3Ј UTR of about 2000 bp and bears many AU-rich motifs that could be targets for the binding of proteins that stabilise or destabilise the mRNA.
Binding of synthetic RNA probes to the ␣ 2 3Ј UTR in nuclear extracts from granule cells The 3Ј UTR fragments we had isolated were employed to synthesise radiolabelled RNA probes to perform RNA-protein binding assays. These binding assays were carried out on nuclear or cytosolic extracts from granule cells, with RNA probes generated from both the 843 bp fragment (probe I) and the 1176 bp fragment (probe II). Protein-RNA complexes were detected in these assays when probe I was incubated with nuclear extracts (from Fig. 4A, lanes 3 and 5; Fig. 4B, lanes 2 and 3), but not when this RNA probe was incubated with the same amount of cytosolic protein (Fig. 4A, lanes 2 and 4); nor were labelled complexes observed when the experiment was performed in the presence of competing unlabelled probe (lane 4, Fig. 4B). In contrast to, we did not observe the formation of RNA-protein complexes when the same experiments were performed with probe II (Fig. 4C). These results can be explained by the concentration of most of the AU-rich elements in the sequence corresponding to probe I (as it is shown in Fig.  3).
Binding of Hu and AUF1 to endogenous mRNA transcripts encoding the ␣ 2 subunit of guanylyl cyclase Of the proteins that influence mRNA-turnover through their recognition of and binding to AREs, the most extensively studied are the adenosine-uridine binding factor AUF1 and members of the ELAV-like family of proteins that are also known as Hu proteins (HuA and/or R, HuB, HuC, HuD). To determine whether the AUF1 or Hu proteins bind to ␣ 2 mRNA in vivo, these proteins were immunoprecipitated from cell extracts wih specific antibodies, and the presence of ␣ 2 mRNA in the purified complexes was investigated. As a control of the non-specific association of mRNAs during immunoprecipitation (IP), parallel incubations were performed with IgG. In the latter, little or no material was observed after 35 cycles of PCR amplification. Control 18S rRNA was used to monitor the background of RNA-binding during IP. Again, no amplification products were observed when reverse transcriptase was omitted. The enrichment of transcripts in each sample was calculated by quantitative real-time reverse transcriptase (RT)-PCR. To quantify the amount of 18S rRNA and ␣ 2 mRNA in each sample, the c T (threshold cycle) values were compared with standard curves obtained from serial dilutions of a RT product and with the specific 18S and ␣ 2 primers and probes. The relative amount of ␣ 2 mRNA in each sample was always considered respective to the amount of 18S rRNA, which was virtually equivalent in all the samples. In AUF1 IPs from nuclear extracts there was an 18-fold enrichment of ␣ 2 mRNA compared to a twofold enrichment in the AUF1 IPs from the cytosolic fraction, and the nuclear HuR IPs (Fig. 4D). By contrast, there was no enrichment of ␣ 2 mRNA in the cytosolic HuR IP. These results show that both AUF1 proteins and HuR proteins bind to ␣ 2 mRNA in vivo, and these protein-mRNA complexes are enriched in the nuclear fraction. Although the ␣ 2 transcripts were associated in complexes immunoprecipitated with both antibodies, ␣ 2 mRNA was more abundant in the AUF1 protein complex. This led us to ask whether the levels of these proteins are affected by NMDA treatment, thereby explaining the effect of NMDA on ␣ 2 mRNA stability and accumulation.
NMDA treatment leads to a decrease in the nuclear levels of AUF1 AUF1 levels were analysed in total, nuclear and cytosolic extracts obtained from both control and NMDA-treated cells.
In total extracts, all four AUF1 isoforms were detected (molecular mass: 37, 40, 42 and 45 kDa), and their levels Fig. 4. Interaction of the RNA probes from the 3Ј UTR of the rat ␣ 2 subunit of guanylyl cyclase with nuclear or cytosolic proteins extracted from granule cells. (A) A radiolabelled, in vitro transcribed RNA fragment corresponding to the 843 bp fragment obtained with F1 and R1 primers (see Fig. 3) was incubated with of cytosolic (Cyt) or nuclear (Nucl) protein extracts from control or NMDA-treated cells. After binding, RNA-protein complexes were treated with T1 RNase and resolved by non-denaturing gel electrophoresis as described in the Materials and Methods. Controls were performed without the protein extract (NP). (B) Radiolabelled, in vitro transcribed RNA fragments corresponding to that of described in panel A was incubated with nuclear (Nucl) protein extracts from NMDA-treated cells or control cells, in the presence or absence of the non-labelled probe (NLP). (C) Radiolabelled, in vitro transcribed RNA fragment corresponding to the 1176 bp fragment obtained with the F2 and R2 primers (see Fig. 3) was used in a similar experiment to that described in A. (D) The presence of endogenous ␣ 2 mRNA was assayed in the IP material from cytosolic or nuclear extracts obtained with anti-HuR or anti-AUF1 by real-time RT-PCR. The results are the mean ± s.e.m. of two experiments. *P<0.01 and + P<0.01, significant difference from control values (IgG). decreased after NMDA treatment (with 50 or 100 M) (Fig.  5A). When the presence of these proteins was analysed in nuclear and cytosolic extracts they were, in most experiments, clearly detected in nuclei but not in the cytosol. In some experiments AUF1 proteins were detected in the cytosolic extracts but always at very low levels (Fig. 5B). In the majority Journal of Cell Science 119 (8) of the experiments the 40 kDa isoform was not detected, not even in nuclear extracts where the quantity of the other isoforms was very high. In cells treated with NMDA, the amount of AUF1 proteins decreased in both the nuclei and the whole-cell extract by 50%, and this reduction might be responsible for the increased stability of ␣ 2 mRNA in response to NMDA. The level of the nuclear protein lamin B1 and the level of GAPDH, used as loading control, did not change after NMDA treatment (Fig. 5B).
NMDA decreases the AUF1 levels and increases ␣ 2 mRNA levels by activating the NO-cGMP pathway It is well known that, the activation of NMDA receptor permits the entry of Na + and Ca 2+ , and increases in intracellular Ca 2+ activate a wide range of intracellular cascades. One such cascade leads to the synthesis of NO and, thus, we analysed whether NMDA stimulation leads to an increase in cytosolic Ca 2+ and in NO synthesis in our cell preparations. When 7-DIV cells were loaded with the Ca 2+ indicator Fluo 4-AM and then stimulated with NMDA, an increase in intracellular Ca 2+ was observed (Fig. 6A). This increase in intracellular Ca 2+ was accompanied by an increase in NO accumulation (Fig. 6B) that was impaired by the broad-spectrum NOS inhibitor L-NAME.
Thus, NMDA appears to stimulate an increase in endogenous NO synthesis. The specificity of the assay was confirmed by adding the nitric oxide donor diethylamino-diazenolate-2oxido (DEA/NO) that overcame the inhibition of L-NAME. The endogenous NO produced upon exposure to NMDA, stimulated guanylyl cyclase to synthesise cGMP, as observed in most cells in response to NMDA (Fig. 6C). Indeed, treatment with the cyclic nucleotide phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX) increased cGMP in the soma of granule cells and there was a big increase in cGMP after exposure to NMDA both in the soma and in the fibres of the granule cells. The synthesised cGMP could activate cGMP-dependent protein kinases in 7-DIV granules cells because in culture, these cells have been shown to express both cGMPdependent protein kinases I and II (cGKs), (Jurado et al., 2004). Hence, it was no surprise that NMDA  stimulation lead to a 2.6-fold increase in VASP phosphorylation, one of the best-characterised endogenous substrates of these enzymes. This increase in VASP phosphorylation was prevented in the presence of a specific inhibitor of these kinases (Fig. 6D). NMDA stimulation also increased CREB phosphorylation, which was also prevented by Rp-8-pCPT-cGMPS (Fig. 6E). Thus, it appears that NMDA triggers the activation of the NO-cGMP pathway in cultured cerebellar granule cells.
To determine whether these signalling events are involved in the NMDA-elicited reduction of AUF1 and the increased accumulation of ␣ 2 mRNA transcripts, the effect of NMDA on these phenomena was analysed in the presence of inhibitors of this pathway. NMDA treatment produced a 50% reduction in AUF1 protein accumulation but this effect was prevented by ODQ (10 M) and KT5823 (1 M), inhibitors of guanylyl cyclase and cGMP-dependent-protein kinases, respectively (Fig. 7A). Furthermore, the effect of NMDA was mimicked by Sp-8-Br-PET-cGMPS (50 M), a cGMP analogue that activates cGKs. KT5823 not only prevented the decrease in AUF1 protein caused by NMDA but also increased AUF1 levels in non-stimulated cells (Fig. 7B), indicating that basal cGK activity is controlling AUF1 levels. Since the downregulation of AUF1 levels by NMDA involves a mechanism that depends on activation of the NO-cGMP pathway, we determined whether these compounds also affected the NMDA-elicited upregulation of ␣ 2 mRNA. Indeed, the presence of the guanylyl cyclase inhibitor ODQ or of the cGK inhibitor KT5823 impaired accumulation of ␣ 2 mRNA transcripts induced by the NMDA receptor, as did the NMDA receptor antagonist MK801 (Fig. 7C). When the effect of these compounds was analysed in cells that had not been not stimulated with NMDA, ODQ produced a slight but not statistically significant decrease in ␣ 2 mRNA. However, in cells incubated with KT5823 the ␣ 2 mRNA amount was reduced by 40%. Conversely, in cells treated with Sp-8-Br-PET-cGMPS ␣ 2 mRNA increased 2.75-fold.

Discussion
Treatment with NMDA increased the accumulation of mRNA encoding different subunits of NO-sensitive guanylyl cyclase, the transcript encoding the ␣ 2 subunit being the most affected (Jurado et al., 2003). Whereas different mechanisms have been proposed to regulate the expression of the ␣ 1 and ␤ 1 subunits of guanylyl cyclase in different cell types (Ferrero et al., 2000;Ferrero and Torres, 2002;Klöss et al., 2003;Klöss et al., 2004), no studies on the regulation of ␣ 2 expression have been carried out to date. This is surprising, given that this subunit seems to be responsible for the synaptic localisation of guanylyl cyclase (Russwurm et al., 2001), which it is highly expressed in different areas of the central nervous system (Gibb and Garthwaite, 2001;Mergia et al., 2003), and that its expression profile coincides with neuronal differentiation (Jurado et al., 2003). Thus, understanding the mechanisms that regulate its expression may help to clarify its role in neuronal development and synaptogenesis (Polleux et al., 2000;Wagenen and Rehder, 2001;Ding et al., 2005). In this study, we have shown that the mRNA transcripts that encode the ␣ 2 subunit of NO-sensitive guanylyl cyclase have a 3Ј UTR of at least 1873 bp. Furthermore, we identified several AREs in this region that are targeted by proteins of the elav family and AUF1 proteins, involved in stabilising/destabilising mRNA. This structural characteristic has previously not been studied in mRNAs encoding the ␣ 2 subunit, although it has already been shown that mRNAs encoding the ␣ 1 and ␤ 1 subunits have a long 3Ј UTR to which stabilising proteins bind (Klöss et al., 2003;Klöss et al., 2004). Furthermore, our results demonstrate that expression of the ␣ 2 subunit of NO-sensitive guanylyl cyclase is regulated in granule cells by a mechanism that affects mRNA stability. The regulation of mRNA decay is an important means to control gene expression (Guhaniyogi and Brewer, 2001) and it has been seen to apply to several components of the NO-cGMP pathway Pérez-Sala et al., 2001;Klöss et al., 2003;Klöss et al., 2004). Although we cannot rule out an increase in ␣ 2 mRNA transcription in the presence of NMDA, the almost 3.5-fold increase in ␣ 2 mRNA accumulation induced by NMDA in the presence of actinomycin D suggests that the rate of mRNA degradation is lowered by NMDA. It was somewhat surprising that cycloheximide, long recognised for its ability to inhibit protein synthesis, produced a threefold increase in the levels of ␣ 2 mRNA, and that NMDA did not produce any additional effect under these conditions. These results suggest that both cycloheximide and NMDA share the mechanism by which they increase the levels of ␣ 2 mRNA, and imply that these mRNA transcripts are more stable under these conditions. This is further supported by the fact that the inhibition of protein synthesis increases mRNA stability in several different systems (Altus et al., 1987;Oliveira and McCarthy, 1995;Ross, 1995), mainly in relation to mRNA transcripts that contain AU-rich elements (AREs) (Guhaniyogi and Brewer, 2001) such as ␣ 2 mRNA. Cycloheximide is thought to interact with some aspects of mRNA translation itself, elements that are coupled to the activation of ARE-mRNA degradation (Savant-Bhonsale and Cleveland, 1992).
To date, there are three ARE-binding proteins that reportedly are involved in regulating rapid mRNA decay in vivo: HuR (Ma et al., 1996), tristetraprolin (TTP) (Carballo et al., 2000), and the poly(U)-binding and degradation factor AUF1 (also known as hnRNP D) (Brewer, 1991;Wagner et al., 1998). We found that nuclear protein extracts from granule cells bound in vitro to RNA probes that corresponded to the ␣ 2 mRNA 3Ј UTR. Although these experiments did not offer any information regarding the identity of the proteins bound to this RNA, we know that they are concentrated in the nucleus. Moreover, even though this technique does not allow the intensity of the band corresponding to the RNA-protein complexes to be accurately quantified, there was a clear reduction in the intensity of the complex when the RNA probe was incubated with nuclear extracts from NMDA-treated cell.
Hence, it appears that NMDA treatment causes a decrease in the amount of the protein that associates with this RNA or it favours their displacement. In conjunction, the identification of ␣ 2 mRNA transcripts in complexes immunoprecipitated with an anti AUF1 antibody show that these destabilising factors do indeed bind to the ␣ 2 mRNA. The amount of the four AUF1 isoforms bound is reduced by about 50% in the cells incubated with NMDA. As a consequence, one would expect AREcontaining mRNAs to be stabilised -as demonstrated for the ␤-adrenergic receptor mRNA whose stability is controlled by alterations in the levels of AUF1 (Pendel et al., 1996). Immunoprecipitation studies also provided evidence for the binding of Hu proteins to ␣ 2 mRNA, which was present in complexes immunoprecipitated with an anti-HuR antibody (which also recognises HuB, HuC, HuD as well as HuR). Hence, these ␣ 2 mRNA transcripts can bind both stabilising and destabilising proteins. It has been shown that the binding of HuR and AUF1 to a target transcript may occur in a competitive fashion (Lal et al., 2004). Thus, we cannot rule out the possibility that the decrease in the amount of AUF1 leads to an increase in the formation of Hu-mRNA complexes. The binding of HuR to RNA probes synthesised from the 3Ј UTR of ␣ 1 and ␤ 1 mRNAs has been previously observed in smooth muscle cells (Klöss et al., 2003;Klöss et al., 2004). Hence, the mechanism controlling the decay of guanylyl cyclase subunit mRNA may be widespread to regulate guanylyl cyclase activity and might explain the accumulation of ␣ 1 , ␤ 1 and ␣ 2 mRNA induced by NMDA already observed by us (Jurado et Journal of Cell Science 119 (8) al., 2003). Moreover, the reduction in AUF1 protein levels may affect the stability of many different mRNAs and might provide a means by which NMDA regulates gene expression.
Exposure to NMDA provokes a decrease in the levels of AUF1 and an increase in the accumulation of ␣ 2 mRNA by a mechanism that depends on synthesis of cGMP and cGK activation. This mechanism is tonically active in control cells, and NMDA stimulates it. This conclusion is supported by the fact that, both the increase in ␣ 2 mRNA and the decrease in AUF1 levels caused by NMDA treatment were effectively prevented by pre-treatment with specific inhibitors of the NOsensitive guanylyl cyclase and of cGK and, because cGK inhibition in non-NMDA stimulated cells caused an increase in AUF1 levels, an a reduction in the amount of ␣ 2 mRNA. Moreover, pre-treatment with the cGMP analogue Sp-8-Br-PET-cGMPS that activates cGK, also caused a reduction in the levels of AUF1 and an increase in the amount of ␣ 2 mRNA comparable with that induced by NMDA. We demonstrated that stimulation with NMDA produced an increase in NO, which in turn activated cGMP synthesis by guanylyl cyclase. It appeared that the produced cGMP binds to and activates cGK, because the phosphorylation of two well-known substrates of this enzymatic activity (VASP and CREB) increased. Although the mechanism by which cGK regulates AUF1 levels remains unclear, one possibility is that cGK suppresses the expression of AUF1 proteins by phosphorylation of CREB or of another transcription factor. Indeed, the influence of PKG on gene expression has been demonstrated in different cell types, including neurons (Gudi et al., 1999;Fujiwara et al., 2002;L'Etoile et al., 2002;Smolenski et al., 2004;Pilz and Broderick, 2005), and in some cases this activity is clearly mediated by CREB phosphorylation (Ciani et al., 2002;Chen et al., 2003;Chan et al., 2004). Based on the data presented here, we hypothesise that NMDA stimulation decreases the amount of AUF1 in granule cells and, as a consequence, stabilises different mRNAs that are targeted by these proteins.

Cell culture
All procedures relating to the care of animals were conducted according to institutional ethical guidelines for animal experiments and following the regulations of the European Council Directive (86/609/EEC). Primary dissociated cerebellar cultures were established using cerebellar tissue from 7-day-old (P7) male or female Wistar albino rat pups, according to the methods described previously (Baptista et al., 1994;Brewer, 1995) with minor modifications (Jurado et al., 2003). Cerebellar cells were diluted in Neurobasal A supplemented with B27 (GIBCO BRL, Uxbridge, UK), 20 mM KCl, 0.5 mM glutamine and the stabilised antibiotic antimycotic solution (SIGMA). The cells were seeded onto coverslips, 96-well, or 6-well tissue culture plates pre-coated with poly-L-lysine at a density of 100ϫ10 5 cell/coverslip, 200ϫ10 5 cell/well or 5ϫ10 6 cells/well, respectively. The cultures were maintained in a humidified incubator at 5% CO 2 , 37°C and after 24 hours in culture, 10 M cytosine-␤-D-arabinofuranoside (final concentration) was added to each well to inhibit the proliferation of non-neuronal cells. This culture medium is optimised for neuronal survival and minimal glial proliferation, and more than 95% of the cells expressed neuronal markers after 48 hours of culture. The cells were maintained in culture and the experiments were performed after 7 days in vitro (7-DIV).

RNA isolation and quantification
After subjecting 7-DIV cells to the appropriate treatment, total RNA was extracted from cells using the RNeasy kit (Quiagen GmbH, Hilden, Germany) and the RNA was quantified using the RiboGreen TM RNA Quantification Kit (Molecular Probes Invitrogen, Barcelona, Spain) as previously described (Ferrero and Torres, 2002).

Reverse transcriptase (RT)-PCR and real-time RT-PCR reactions
RT-PCR reactions were performed in two steps as described previously (Jurado et al., 2003). First strand cDNAs were synthesised in RT buffer containing 5.5 mM MgCl 2 , 500 M per dNTP, 2.5 M random hexamers, 0.4 U/ml RNAse inhibitor and 3.125 U/ml MultiScribe TM RT (Applied Biosystems). The reactions were performed in a final volume of 50 l containing 1 g of RNA and an incubation step of 10 minutes at 25°C to maximise primer-RNA-template binding. The reverse transcription reaction was performed at 48°C for 30 minutes and reverse transcriptase was inactivated before the PCR reactions by heating the samples at 95°C for 5 minutes. The specific PCR primers and the probe for the guanylyl cyclase ␣ 2 subunit were designed using the published sequence (accession number AF109963 and NM_023956) (Koglin and Behrends, 2000) with the help of the primer express software package (Applied Biosystems, Madrid, Spain). Forward primer, position 828 5Ј-TCT GCA GAC CAT TCC AAC AAA G-3Ј; reverse primer, position 941 5Ј-TCC TCA CCA AAC CTC TCT TGA ATT-3Ј; TaqMan ® probe, position 907 5Ј-(FAM)-TCA GTC CGA GTA CAT TTG CAG TAC ACC GAA-(TAMRA)-3Ј. 18S rRNA was used as the endogenous control. To amplify a 200 bp fragment, we used a commercial mixture of primers and TaqMan probe labelled with VIC and TAMRA at the 5Ј and 3Ј end, respectively (Applied Biosystems). PCR reactions were followed in an ABI PRISM 7700 Sequence Detection System (Applied Biosystems) with TaqMan Gold PCR reagents. The reaction mixture contained TaqMan PCR buffer, 5.5 mM MgCl 2 , 200 M dATP, 200 M dCTP, 200 M dGTP, 400 M dUTP, 0.025 U/l AmpliTaq Gold DNA polymerase, 0.01 U/l AmpErase UNG, 300 nM of each primer and 300 nM of TaqMan probe. Reactions were performed with an initial incubation at 50°C for two minutes followed by 10 minutes at 95°C for AmpliTaq Gold activation and 40 cycles (melting 95°C for 15 seconds, annealing and extension 60°C for 1 minute). Fluorescence was determined at each step of every cycle. Quantifications were always normalised using endogenous control 18S rRNA to check for variability in the initial concentrations, the quality of the total RNA and the conversion efficiency of the reverse transcription reaction.

mRNA decay assay
The decay rate of ␣ 2 mRNA was measured in an actinomycin D time-course assay. Briefly, transcription was inhibited in control or NMDA-treated cells by adding actinomycin D (10 g/ml) to the culture medium, and RNA samples were harvested at selected time points thereafter. The relative abundance of ␣ 2 mRNA was determined by quantitative RT-PCR as described before.

Intracellular Ca 2+ measurements
Increases in intracellular Ca 2+ were determined using the FLUOstar OPTIMA fluorescence microplate reader equipped with two injectors (BMG LabTechnologies, Offenburg, Germany). Granular neurons maintained in culture for 7-DIV were loaded with 4 M Fluo-4 acetoxy methylester (Fluo-4 AM) in Locke's buffer for 45 minutes at 37°C, and they were subsequently washed twice with fresh Locke's buffer before further incubating for 15 minutes. All experiments were performed at 37°C and the data collected at 0.5 or 1 seconds intervals. The initial volume in each well was 48 l of Mg 2+ -free Locke's buffer. Fluorescence was excited using a 485 nm filter and read through a 520 nm long-pass emission filter. To measure Ca 2+ responses in cells, the baseline fluorescence was first monitored for 5 seconds, then cells were stimulated with 100 M NMDA (final concentration) and fluorescence monitored for up to 35 seconds. At the end of each measurement, 0.1% Triton X-100 (final concentration) was added to each well to permeate cells and the maximal fluorescence in each well measured (F max ). Finally, EGTA-Tris was added and the minimal fluorescence (F min ) in each well was measured. To calibrate the response the following equation was used to determine the intracellular free Ca 2+ concentration [Ca 2+ ] i =k d [F-F min ]/[F max -F], using K d =345 nM for Fluo-4. Each well was calibrated using its own F max and F min readings.

NO measurements
NO was measured using the FLUOstar OPTIMA fluorescence microplate reader equipped with two injectors (BMG LabTechnologies, Offenburg, Germany). Cells plated on 96-well polylysine-coated plates and maintained in culture for 7-DIV were loaded with 10 M DAF-FM diacetate for 45 minutes at room temperature in Locke's buffer. Fluorescence was recorded using 495 nm excitation and 515 nm emission filters. Basal fluorescence (F 0 ) was first monitored for 5 seconds, before NMDA 100 M (final concentration) was injected and the fluorescence recorded for an additional 25 seconds (data were collected at 1 second intervals). After this time period, DEA/NO (~80 M final concentration) was added and the fluorescence monitored for an additional 10 seconds. Incubation with L-NAME was performed during DAF-FM loading. F/F 0 was calculated for each time point and presented versus time. Although DAF-FM has been shown to be more photostable than DAF-2 it should bear in mind that the fluorescence of the NO adduct of DAF-2 is enhanced by Ca 2+ and light, making it difficult to compare different measurements performed in different conditions (Broillet et al., 2001).

Immunocytochemistry
Cerebellar granule cells were plated on polylysine-coated coverslips at a density of 200,000 cell/coverslip. Cells were stimulated with 100 M NMDA in Locke's buffer for 2 minutes and then fixed for 15 minutes in 4% paraformaldehyde freshly prepared in 0.1 M phosphate buffer pH 7.4. The cells were then washed twice with TBS, (5 minutes each), and the coverslips were incubated with blocking buffer (TBS containing 0.2% Triton X-100, 10% normal donkey serum). The cells were washed twice more for 5 minutes with TBS and then incubated for 24 hours at 4°C with the primary sheep anti-formaldehyde-fixed cGMP antibody (1:4000; kindly provided by Jan de Vente) diluted in TBS containing 1% normal donkey serum and 0.2% Triton X-100. After washing three times (5 minutes each) with TBS containing 0.1% Triton X-100, the coverslips were incubated with Cy TM 2-conjugated affinitypurified donkey anti-sheep IgG (H+L; 1:200 in TBS; Jackson ImmunoResearch, West Grove, PA). After several washes in TBS, the coverslips were mounted with prolong antifade (Molecular Probes) and the cells were viewed with a Nikon Diaphot microscope equipped with a 40ϫ1.3 N.A oil immersion fluor objective, a mercury lamp light source and fluorescein Nikon filter sets.

Fractionation of granule cells and mobility-shift assays
Whole-cell lysates from control or NMDA-treated granule cells (100 M, 24 hours) were prepared by scraping PBS-washed cells in ice-cold lysis buffer (20 mM Hepes, pH 7.9, 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 0.5% NP-40, containing protease inhibitors: 1 mM AEBSF, 0.8 M aprotinin, 50 M Bestatin, 15 M E-64; 20 M leupeptin and 10 M pepstatin A). The cell debris was cleared from the cell lysates by centrifugation at 14,000 g for 5 minutes at 4°C in a Heraeus Microfuge, and the protein concentration in the supernatants determined by the microBradford assay (Bio-Rad). Nuclear and cytoplasmic fractions were prepared according to a procedure described previously (Schreiber et al., 1989). Briefly, cells were washed once with ice-cold PBS and scraped in the same medium. Upon centrifugation they were resuspended in 10 mM Hepes pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol (DTT), 1 mM AEBSF, 0.8 M aprotinin, 50 M Bestatin, 15 M E-64; 20 M leupeptin and 10 M pepstatin A. After 15 minutes incubation on ice, Nonidet P-40 was added to a final concentration of 0.5% and cells were vortexed for 10 seconds before pelleting the nuclei by centrifugation for 5 minutes at 14,000 g. After removing the supernatant (cytoplasmic fraction) the nuclear pellets were extracted by shaking on ice for 45 minutes in 50 l of 20 mM Hepes pH 7.9, 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, and protease inhibitors as above. After centrifugation at 14,000 g for 5 minutes, the protein content was determined using the Bradford method (Bio-Rad) and subjected to electrophoresis or stored at -70°C.
The 3Ј UTR of the rat guanylyl cyclase ␣ 2 subunit mRNA was cloned by RT-PCR as follows. Total RNA was extracted from 7-DIV cells and reverse transcribed with Oligo-dT and MultiScribe TM RT (Applied Biosystems) in RT buffer containing 5.5 mM MgCl 2 , 500 mM per dNTP, 2.5 M OligodT, 0.4 U/ml RNase inhibitor and 3.125 U/ml MultiScribe TM RT. Reactions were performed in a final volume of 50 l containing 1 g RNA with an incubation step of 10 minutes at 25°C, 30 minutes at 48°C and finally 5 minutes at 95°C to inactivate the reverse transcriptase before the PCR reaction. The following oligodeoxynucleotides were used to amplify two different regions of the ␣ 2 subunit 3Ј-UTR: 5Ј-CTCCGGGAAACTAGCCTCTG-AGA-3Ј (upstream) and 5Ј-ATACAGAAAAAATGGAATTTTCTAGTAGAAC-3Ј (downstream) used to amplify a 843 bp fragment and 5Ј-TTAGTGATTAG-TGCTTGAGCTGAGA-3Ј (upstream) and 5Ј-CCTGGGTATTCCTAGGCAGCTT-3Ј (downstream) to amplify a 1176 bp fragment. The PCR reaction mixture contained TaqMan PCR buffer, 5.5 mM MgCl 2, 200 M dATP, 200 M dCTP, 200 M dGTP, 400 M dUTP, 0.025 U/l AmpliTaq Gold ® DNA polymerase, 300 nM of each primer, 5 l of cDNA from the previous RT reaction. The PCR reactions were performed in a final volume of 50 l with an initial incubation at 95°C for 10 minutes to activate the AmpliTaq Gold, and over 40 cycles (melting 60 seconds at 94°C, followed by 30 seconds at 62°C for annealing and 90 seconds at 70 °C for elongation) with a final elongation of 7 minutes at 68°C. The PCR products were sequenced (3730 DNA Analyzer, Applied Biosystems, Madrid, Spain), and verified by restriction mapping and sequencing.
The RNA probes were transcribed as follows. Two micrograms of DNA (linearised or PCR fragments) were added to transcription buffer, which consisted of 40 mM Tris-HCl pH 7.9, 6 mM MgCl 2 , 10 mM DTT, 2 mM spermidine, 10 mM NaCl, 1 mM ATP, CTP and GT, 10 U RNasin (Promega, Madison, WI), 2 l of [␣ 32 P]UTP to give a final reaction volume of 20 l, and 20 U T7 RNA polymerase (Promega, Madison, WI). The reactions were carried out for 1 hour at 37°C. Three units of DNase I (Promega, Madison, WI) were added to the samples and incubated for an additional 45 minutes at 37°C. The radiolabelled RNA probes were purified in Sephadex G-50 Quick Spin Columns (Roche) and the radiolabelled transcripts were analysed by electrophoresis to estimate the yield and specific activity. The radioactivity incorporated into the transcripts was usually higher than 80% and the specific activity ranged from 0.2 to 0.5 Ci/pmol.
For the binding experiments, nuclear or cytosolic extracts (2 g per lane) prepared from control or NMDA-treated cells were incubated in 10 mM Tris-HCl pH 7.4, 50 mM NaCl, 1 mM MgCl 2 , 0.5 mM EDTA, 5 mM DTT, 5% glycerol, and containing 0.2 mg/ml of tRNA (Pharmacia) as competitor with approximately 0.3 pmol (65,000 cpm) of the respective probe. After 15 minutes at room temperature, 6 U of RNase T1 were added for 20 minutes to digest the unbound RNA. RNAprotein complexes were resolved by non-denaturing 5% polyacrylamide gel in 0.25ϫ Trisborate-EDTA buffer and visualised by autoradiography.
To analyse the RNA in the IP complexes, the beads were incubated with 100 l NT2 buffer containing 20 U RNase-free DNase I (Sigma) for 15 minutes at 30 °C. They were then washed with NT2 buffer and further incubated in 100 l NT2 buffer containing 0.1% SDS and 50 g proteinase K (15 minutes, 55°C). The RNA was extracted using the RNeasy mini Kit and then, RT-PCR and real-time RT-PCR were performed. The RNA isolated was reverse transcribed using random hexamers and the resulting material was used for PCR amplification using gene-specific primer pairs as described before. As a negative control, 50% of the RNA was incubated without MultiScribe TM RT. Control 18S rRNA, which is not a target of either RNA binding protein (RBP), was used to monitor the background levels of RNA in the IP complexes.

Statistical analysis
The results are expressed as means ± s.e.m. of three or more experiments. The data were analysed by one-way ANOVA followed by Bonferroni's test (confidential interval 95%). The differences between the means were considered statistically significant when P<0.05. This study was financed by a grant from the Ministerio de Educación y Ciencia (BFI2003-00731). S.J. was supported by a fellowship from the same source. We thank J. de Vente for providing us with the anti-cGMP antibody and M. Sefton for linguistic assistance.