Post-transcriptional regulation of acetylcholinesterase mRNAs in nerve growth factor-treated PC12 cells by the RNA-binding protein HuD.

Expression of acetylcholinesterase (AChE) is greatly enhanced during neuronal differentiation, but the nature of the molecular mechanisms remains to be fully defined. In this study, we observed that nerve growth factor treatment of PC12 cells leads to a progressive increase in the expression of AChE transcripts, reaching approximately 3.5-fold by 72 h. Given that the AChE 3'-untranslated region (UTR) contains an AU-rich element, we focused on the potential role of the RNA-binding protein HuD in mediating the increase in AChE mRNA seen in differentiating neurons. Using PC12 cells engineered to stably express HuD or an antisense to HuD, our studies indicate that HuD can regulate the abundance of AChE transcripts in neuronal cells. Furthermore, transfection of a reporter construct containing the AChE 3'-UTR showed that this 3'-UTR can increase expression of the reporter gene product in cells expressing HuD but not in cells expressing the antisense. RNA gel shifts and Northwestern blots revealed an increase in the binding of several protein complexes in differentiated neurons. Immunoprecipitation experiments demonstrated that HuD can bind directly AChE transcripts. These results show the importance of post-transcriptional mechanisms in regulating AChE expression in differentiating neurons and implicate HuD as a key trans-acting factor in these events.

Expression of acetylcholinesterase (AChE) is greatly enhanced during neuronal differentiation, but the nature of the molecular mechanisms remains to be fully defined. In this study, we observed that nerve growth factor treatment of PC12 cells leads to a progressive increase in the expression of AChE transcripts, reaching ϳ3.5-fold by 72 h. Given that the AChE 3-untranslated region (UTR) contains an AU-rich element, we focused on the potential role of the RNA-binding protein HuD in mediating the increase in AChE mRNA seen in differentiating neurons. Using PC12 cells engineered to stably express HuD or an antisense to HuD, our studies indicate that HuD can regulate the abundance of AChE transcripts in neuronal cells. Furthermore, transfection of a reporter construct containing the AChE 3-UTR showed that this 3-UTR can increase expression of the reporter gene product in cells expressing HuD but not in cells expressing the antisense. RNA gel shifts and Northwestern blots revealed an increase in the binding of several protein complexes in differentiated neurons. Immunoprecipitation experiments demonstrated that HuD can bind directly AChE transcripts. These results show the importance of post-transcriptional mechanisms in regulating AChE expression in differentiating neurons and implicate HuD as a key trans-acting factor in these events.
Acetylcholinesterase (AChE) 1 is the enzyme responsible for the rapid hydrolysis of acetylcholine in the central and peripheral nervous systems (see, for review, Refs. [1][2][3][4]. The enzyme exists in multiple molecular forms that differ in their C termi-nus and mode of anchoring to subcellular structures. The different C termini of the protein are generated through alternative splicing of a single gene, and three different mature mRNAs, referred to as the T (tail), H (hydrophobic), and R (readthrough) transcripts, can be produced. The pattern of expression of these transcripts is known to be tissue-specific because, for example, the T transcript is abundantly expressed in excitable cells such as skeletal muscle and neurons, whereas the H transcript is found predominantly in the hematopoietic lineage. Two polyadenylation signals can be used in the 3Јuntranslated region (UTR) to produce a ϳ2.4or 3.2-kb transcripts. Choice of the polyadenylation signal is also tissuespecific because neurons appear to preferentially express the shorter form of the transcript whereas skeletal muscle express both species but in different amounts (5)(6)(7).
In addition to its role in cholinergic neurotransmission, converging lines of evidence indicate that it likely fulfills additional, non-catalytic functions within the nervous system (see, for review, Ref. 8). For example, Northern blot analysis and in situ hybridization experiments have revealed the presence of AChE mRNAs in non-cholinergic areas of the brain such as the cerebellum (9 -12). Moreover, AChE expression in the brain is known to precede the establishment of synaptic transmission and coincides with the period of neurite outgrowth (13,14). Finally, several studies in which AChE levels have been experimentally manipulated directly support the notion that AChE is indeed involved in neurite outgrowth (see, for example, Refs. 13-15 and 17-22).
In recent years, there have been several studies that have examined the basic molecular events that preside over AChE expression in developing and adult skeletal muscles (see, for example, Refs. [23][24][25][26][27][28][29][30]. By contrast, there is relatively little information concerning the mechanisms regulating AChE expression in neurons. In addition, of the few available reports, there also appear to be some contradictory findings. In particular, Greene and Rukenstein (31) have provided evidence indicating that differentiation of PC12 cells, which leads to an increase in AChE expression, induces an increase in AChE gene transcription. Alternatively, embryonic P19 carcinoma cells induced to differentiate into neurons via retinoic acid treatment failed to increase the transcriptional activity of the AChE gene, thereby indicating that post-transcriptional mechanisms represent key events in regulating the abundance of AChE transcripts during neuronal development (32). Given the diverse and key functions of AChE within the nervous system, it appears important to gain a more complete understanding of the molecular mechanisms that control AChE expression in neurons. Accordingly, we have initiated a series of experiments in attempts to characterize some of the molecular events in-volved in AChE expression during neuronal differentiation. Specifically, we have examined the importance of transcriptional and post-transcriptional mechanisms in the regulation of AChE during NGF-induced differentiation of PC12 cells.

EXPERIMENTAL PROCEDURES
Cell Culture-PC12 cells, a rat pheochromocytoma-derived cell line (33), were cultured on culture dishes coated with type I collagen (Sigma) in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% horse serum, 5% fetal bovine serum, and 100 units/ml penicillinstreptomycin, in a humidified chamber at 37°C containing 5% CO 2 . Stably transfected PC12 cells were maintained as described elsewhere (34). These lines were transfected with the pcDNA3 vector alone (pcDNA) or with plasmids containing the human HuD sequence in sense (pcHuD) or antisense (pDuH) orientation. All cells were plated at a density of 1-2 ϫ 10 5 cells/cm 2 and induced to differentiate by adding 100 ng/ml 7S-NGF (Sigma) to the culture medium. Culture media were changed every 72 h.
AChE Enzymatic Assay-Cultures of undifferentiated and differentiated PC12 cells (three 35-mm wells) were washed with cold phosphate-buffered saline (PBS), scraped, and homogenized on ice with a glass Kontes homogenizer in 0.5 ml of a high salt detergent buffer containing anti-proteolytic agents (10 mM Tris-HCl, pH 7.0, 10 mM EDTA, 1 M NaCl, 1% Triton X-100, 1 mg/ml bacitracin (Sigma), and 25 units/ml aprotinin (Sigma)). Following centrifugation of the homogenates (20,000 ϫ g for 15 min at 4°C), the supernatant was removed and stored immediately at Ϫ80°C. AChE activity was measured using a modified version of the spectrophotometric method of Ellman et al. (35) as described previously (36). The total amount of protein present in the extracts was determined by the bicinchoninic acid assay (BCA; Pierce).
RNA Extraction and RT-PCR-Total RNA was extracted from undifferentiated and differentiated PC12 cell cultures (three 35-mm wells) using 0.5-1 ml of TRIzol reagent (Invitrogen) according to the instructions from the manufacturer. Briefly, the cells were scraped from the plates and disrupted by vigorous pipetting, followed by addition of chloroform. The resulting solution was mixed vigorously and centrifuged (12,000 ϫ g for 15 min at 4°C). The aqueous layer was then transferred to a fresh tube and combined with an equal volume of isopropanol. The RNA was precipitated by centrifugation, and the resulting pellet was washed twice with 75% ice-cold ethanol and resuspended in RNase-free water. All samples were stored in Ϫ80°C until used.
RNA from each sample was quantified using the Amersham Biosciences Gene Quant II RNA/DNA spectrophotometer and adjusted to a final concentration of 80 ng/l. Reverse transcription of RNA was performed using 2 l of each RNA sample at 42°C for 45 min, followed by 5 min at 99°C, as previously described elsewhere (37)(38)(39). Negative controls consist of the same RT mixture in which the RNA was replaced with 2 l of RNase-free water. PCR was used to amplify cDNAs corresponding to AChE and S12 rRNA as described in detail elsewhere (23,(37)(38)(39). Primers for AChE (see Ref. 12) and S12 rRNA (used as an internal control; see Ref. 40) were synthesized based on available sequences that have been previously described, and they amplified products of 670 and 368 bp, respectively. PCR cycling parameters for AChE and S12 rRNA consisted of denaturation for 1 min at 94°C, followed by primer annealing and extension for 3 min at 70°C for AChE and primer annealing for 1 min at 54°C and extension for 2 min at 72°C for S12 rRNA, followed by a 10-min elongation step at 72°C. PCR products were visualized and quantified on ethidium bromide-stained 1.5% agarose gels. Quantitation of the labeling intensity of the PCR products was performed using the Kodak Digital Science Image Station 440 CF and related Kodak Digital Science 1D Image Analysis Software (Eastman Kodak Co.). All values obtained for AChE were corrected according to the corresponding level of S12 rRNA present in the sample.
All RT-PCR experiments aimed at determining the relative abundance of AChE transcripts were performed using cycle numbers that fell within the linear range of amplification (37)(38)(39). The cycle numbers were between 24 and 27 for AChE and 22 for S12 rRNA. RT-PCR conditions (primer concentration, input RNA, choice of RT primer, cycling conditions) were initially optimized, and these were identical for all experiments. Appropriate precautions (use of sterile filtered tips and gloves) were taken to prevent contamination of the samples and degradation of the RNA. Samples, including the negative control, were always prepared using the same RT and PCR reagents and master mixes, and were run in parallel. In all experiments, PCR products were never detected in the negative controls.
Nuclear Run-on Assays-Nuclear run-on assays were performed as described in detail elsewhere (23,41,42). Briefly, nuclei were isolated from undifferentiated and differentiated PC12 cells (two 250-ml flasks) and resuspended in a transcription buffer containing GTP, ATP, CTP, and 25 Ci of [␣-32 P]UTP. RNA was transcribed for 60 min at 30°C in the presence of an RNase inhibitor (Promega, Madison, WI). Following a 30-min RQ1 DNase I (Promega) treatment, the nascent radiolabeled RNA was extracted using TRIzol reagent (see above) and hybridized for 48 h to 10 g of linearized AChE cDNA (2 kb) immobilized on a Protran pure nitrocellulose membrane (Schleicher & Schuell). After hybridization the membranes were washed thoroughly at 42°C in a 1ϫ salinesodium citrate (SSC), 0.1% sodium dodecyl sulfate (SDS) solution and subjected to autoradiography. The intensity of the resulting signals was quantified using a STORM PhosphorImager and the accompanying ImageQuant software (Amersham Biosciences). The signals corresponding to AChE were standardized relative to the signal obtained from genomic DNA.
Reporter Constructs and Transfection Studies-The recently described 5.3-kb AChE promoter fragment termed GRAP (25) was subcloned into a LacZ reporter vector. In addition, the 3Ј-UTR from the mouse ϳ2.4-kb AChE transcript, which contains the shorter 3Ј-UTR in comparison to the ϳ3.2-kb AChE mRNA (see Introduction), was amplified by RT-PCR and first inserted into the pGL3 vector. For these experiments, we focused on the short 3Ј-UTR as opposed to the longer one because (i) previous studies showed that it appears to contain important cis-acting regulatory elements (24,43,44), and (ii) the ϳ2.4-kb AChE transcript is considerably more abundant in nervous tissues (12) including PC12 cells (45,46). The 3Ј-UTR was subsequently cloned into a luciferase reporter construct driven by the thymidine kinase promoter (phRG-TK) (Promega).
Plasmid DNA was prepared using the Mega-Prep procedure (Qiagen, Chatsworth, CA). DNA pellets were resuspended in 10 mM Tris-HCl, pH 8.5. Transfections were performed using the LipofectAMINE reagent kit (Invitrogen) according to the instructions from the manufacturer. Undifferentiated cells (1-2 ϫ 10 5 cells/cm 2 ) were transfected with 0.5 g of the appropriate reporter gene construct and 0.5 g of the constitutively expressed chloramphenicol acetyltransferase (CAT) plasmid driven by the SV40 promoter used, in this case, to control for transfection efficiency. Transfected cells were induced to differentiate 24 h later by the addition of NGF.
To determine reporter gene activity, undifferentiated and differentiated cells were washed with cold PBS and lysed in Reporter-Lysis buffer (Promega) following two cycles of freezing and thawing. The extracts were then centrifuged (15,000 ϫ g for 2 min at 4°C), and the resulting supernatants were assayed for ␤-galactosidase, luciferase, or CAT activities using available kits (Promega). The ␤-galactosidase and luciferase activities were normalized to CAT levels. Background values, obtained by transfecting promoterless LacZ and luciferase plasmids, were subtracted from the activities obtained with the reporter constructs.
In Vitro Transcription-cDNAs encoding different lengths of the AChE 3Ј-UTR were obtained by PCR amplification of the plasmid template pGL3-3Ј-UTR (see above). The PCR primers employed to amplify: 1) the full-length AChE 3Ј-UTR, 2) a truncated fragment in which the AU-rich element was absent (ϪARE), and 3) a small fragment encompassing the ARE (see Fig. 5A), were designed to include a T7 promoter. An in vitro T7 transcription system (Promega) was used to synthesize radiolabeled AChE 3Ј-UTR fragments. Briefly, the transcription reaction containing 0.5 g of PCR fragment, 5 Ci of [␣-32 P]UTP, nucleotides, RNase inhibitor, and T7 polymerase, was carried out at 37°C for 1 h. The template PCR fragments were digested with 1 unit of RQ1 DNase I (Promega) for 30 min at 37°C. The resulting radiolabeled RNA was purified on an RNase-free G-25 RNA purification column (Roche Diagnostics Corp., Indianapolis, IN). The integrity of the RNA was confirmed by gel electrophoresis. Unlabeled RNA probes were generated by the same method and used in cold competition assays.
Electrophoretic Mobility Shift Assay and Northwestern Analyses-RNA-based electrophoretic mobility shift assays (REMSAs) and Northwestern blots were performed using total protein extracts obtained from undifferentiated and differentiated cells (two 250-ml flasks). The cells were washed with cold PBS, scraped, and lysed in 300 l of homogenization buffer (0.3 M sucrose, 60 mM NaCl, 15 mM Tris, pH 8.0, 10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 10 g/l leupeptin, 10 g/l pepstatin A, 1 g/l aprotinin, pH 7.4). The samples were centrifuged (15,000 ϫ g for 15 min at 4°C), and the resulting supernatant was stored at Ϫ80°C until used. The total amount of protein present in the extracts was determined by the BCA method (see above).
Forty g of protein extract were incubated for 20 min at room temperature with 1 ϫ 10 5 cpm of 32 P-labeled AChE 3Ј-UTR fragments in 2ϫ binding buffer (20 mM Hepes, pH 7.9, 3 mM MgCl 2 , 50 mM KCl, 1 mM DTT, 5% glycerol, 0.2 g/l yeast tRNA) in a total volume of 20 l. The unbound RNA was digested with ribonuclease T1 (Calbiochem, San Diego, CA) for 20 min at 37°C, and the samples were then incubated at room temperature for 10 min with heparin (2.5 mg/ml). This mixture was separated by 4 or 6% native polyacrylamide gel electrophoresis with 0.5ϫ TBE (Tris borate-EDTA) running buffer. The gels were subsequently dried under vacuum at 80°C for 1 h and exposed to x-ray film at Ϫ70°C. Competition assays were performed by incubating a 25 M excess of cold probe with the protein extract for 10 min prior to the incubation with the radiolabeled probe. Northwestern analyses were performed according to the procedure described elsewhere (50,51). Fifty g of protein extract diluted in SDS buffer (50 mM Tris-HCl, pH 6.8, 100 mM DTT, 2% SDS, 0.1% bromphenol blue, 10% glycerol) were denatured at 100°C for 3 min and separated by 8% SDS-PAGE. After separation, proteins were electroblotted onto a polyvinylidene difluoride membrane (Schleicher & Schuell). The membrane was then incubated in renaturation buffer (15 mM Hepes, pH 7.9, 50 mM KCl, 0.1 mM MnCl 2 , 0.1 mM ZnCl 2 , 0.1 mM EDTA, 0.5 M DTT, 0.1% (w/v) Ficoll 400 D-L, 0.1% (w/v) polyvinylpyrrolidone, and 0.01% (v/v) Igepal CA-630 (a Nonidet P-40 substitute) in RNase-free water) at 4°C overnight. Following pre-hybridization at room temperature for 1 h in renaturation buffer containing 0.2 mg/ml yeast tRNA, the membrane was incubated for 4 h with 1 ϫ 10 6 cpm/ml of a probe corresponding to the radiolabeled AChE 3Ј-UTR RNA dissolved in renaturation buffer containing 0.2 mg/ml yeast tRNA and 5 mg/ml heparin at room temperature. After several 5-min washes, the membrane was put to x-ray film at Ϫ70°C. To ensure that equivalent amounts of proteins were loaded for each sample, membranes were also stained with Ponceau S (Sigma), following exposure to x-ray films.
Immunoprecipitation and AChE mRNA Analysis-The RNA-binding protein HuD was immunoprecipitated from a total protein extract as described (52). Total protein was extracted from PC12 cells stably transfected to express HuD (8 ϫ 100-mm plate). To this end, the cells were pelleted, resuspended in 0.3 ml of immunoprecipitation (IP) buffer (1% Igepal CA-630, 10 mM Tris-HCl, pH 7.5, 1% bovine serum albumin, 150 mM NaCl, 2 mM EDTA, 25 g/l pepstatin A, 2.5 g/l aprotinin, 10 units of RNase inhibitor), and sonicated (10-s pulse at 50% duty cycle and a power output of 1 using the Branson Sonifier 450). Protein samples (200 g) were incubated for 1 h at 4°C, with an affinitypurified antibody to HuD previously described (34) or with normal rabbit IgG (Jackson Immunoresearch Laboratories, West Grove, PA) in IP buffer. This reaction mixture was subsequently added to 20 l of pre-washed A/G chimera Sepharose beads (Pierce) and incubated by gentle shaking at 4°C for 1 h. The mixture was centrifuged (10,000 ϫ g for 20 s), and the supernatant was removed. After several washes with IP buffer, the RNA was extracted from the pellet using the TRIzol reagent and analyzed by RT-PCR as described above.
Western Blot-Cultured cells were washed in 1ϫ PBS; resuspended in a homogenization buffer containing 0.3 M sucrose, 60 mM NaCl, 15 mM Tris-HCl, pH 8.0, 10 mM EDTA, 0.1 mM ␤-mercaptoethanol, 0.01 mM phenylmethylsulfonyl fluoride, 0.01 mM benzamidine, 1 g of leupeptin, 10 g of pepstatin A, and 1 g of aprotinin; and sonicated (see above). Following centrifugation, the supernatant was recovered, aliquoted, and stored at Ϫ80°C. The concentration of proteins in each sample was determined using the BCA method (see above).
For Western blotting, 50 g of protein extracts were denaturated in SDS loading buffer and subjected to SDS-PAGE using a 10% gel. The proteins were then transferred onto a polyvinylidene difluoride membrane (Sigma). Following transfer, the membranes were incubated with antibodies directed against HuD (34) and revealed using a commercially available ECL kit from Pierce.
Statistical Analysis-An analysis of variance was performed to evaluate the effects of NGF-induced neuronal differentiation on AChE expression. The Fisher's Least Square Difference test was used to determine whether the differences seen between group means were significant. The level of significance was set at p Ͻ 0.05. Data are expressed as mean Ϯ S.E. throughout.

RESULTS
AChE Expression during Neuronal Differentiation-In a first series of experiments, we examined the expression of AChE during the process of NGF-induced neuronal differentiation of PC12 cells. Initially, we compared the level of cell-associated AChE between undifferentiated and 24-, 48-, and 72-h differ-entiated PC12 cells. Previous studies have demonstrated that undifferentiated PC12 cells express a basal level of AChE activity that increases upon NGF stimulation (31,53,54). In agreement with these earlier reports, Fig. 1A shows that AChE activity increases considerably during differentiation. In fact, the cell-associated activity increased significantly by ϳ2.5-fold (p Ͻ 0.01) within the first 48 h, and reached a maximal 5-fold induction (p Ͻ 0.0001) following 72 h of NGF treatment.
We next examined the impact of NGF on the relative abundance of AChE transcripts in PC12 cells. We focused on the level of the T transcript because this transcript is the predominant splice variant found in nervous tissues (12) and PC12 cells (45,46). As illustrated in Fig. 1B, AChE mRNAs could be detected in undifferentiated cells. However, treatment of PC12 cells with NGF led to a pronounced increase in the levels of AChE mRNA. These increases were highly significant, reaching more than 2-fold (p Ͻ 0.002) and 3.5-fold (p Ͻ 0.0001) by 48 and 72 h, respectively (Fig. 1C). The relative amount of S12 rRNA, which was used as an internal control for these assays, did not change during differentiation. The observed increase in transcript level was directly related to the NGF treatment and not to the length of time in culture or cell density because, in separate studies, both of these factors had no effect on AChE expression (data not shown). In addition, NGF removal from the culture medium 24 h after the initial treatment resulted in a gradual decrease in AChE mRNA levels (data not shown).
Because recent studies demonstrated the importance of transcriptional events in regulating AChE expression at the early stages of muscle differentiation (23, 30), we determined whether the increase in the abundance of AChE transcripts seen in differentiating PC12 cells could be linked to an increase in transcription. To this end, we performed transfection studies using an AChE promoter-reporter gene construct that contained a ϳ5.3-kb promoter fragment termed GRAP (24). This fragment contains, in addition to the basal promoter region, intronic elements previously shown to be important for muscle expression (23,25). We observed, using this approach, a small but significant increase (1.5-fold; p Ͻ 0.0001) in reporter gene expression during the early phases of differentiation of PC12 cells (Fig. 2). Nuclear run-on assays performed with nuclei isolated from undifferentiated and 24-h differentiated PC12 cells confirmed these findings (data not shown). This transcriptional increase parallels the initial 2-fold increase in AChE mRNA levels, which occurs during the initial phase of neuronal differentiation. However, it is important to emphasize that this increase in transcription is transient and that, therefore, it fails to account for the increase in AChE transcript level seen at later time points, i.e. 72 and 96 h. Indeed, at these later time points, transcription of the AChE gene had returned to the level seen in undifferentiated cells (data not shown).
Post-transcriptional Mechanisms Regulate AChE Transcript Levels during Neuronal Differentiation-Recent studies have demonstrated that NGF can act indirectly to stabilize neuronspecific transcripts via the ELAV-like RNA-binding protein HuD (52,55). Interestingly, examination of the sequence of the AChE 3Ј-UTR revealed the presence of a conserved element known to be recognized by HuD (see Fig. 4A). Therefore, to ascertain whether post-transcriptional mechanisms are indeed involved in regulating AChE expression during neuronal differentiation and to explore the possibility that this effect involves HuD, we next examined the expression of AChE transcripts in PC12 cells stably transfected with a HuD construct (pcHuD), with an antisense sequence to HuD (pDuH), or with the empty plasmid (pcDNA) acting in this case, as a control (34). In comparison to the control levels of HuD seen in pcDNA cells, levels of HuD in pcHuD cells are 2-3-fold higher, whereas, in pDuH cells, HuD expression is reduced by ϳ70% (34).
As observed in non-transfected PC12 cells, AChE transcripts are expressed, albeit at different levels, in these undifferentiated stable cells. Specifically, undifferentiated pcHuD cells expressed twice as much AChE transcripts as undifferentiated pcDNA cells, whereas the pDuH cells express ϳ80% less transcript than the pcDNA cells in the undifferentiated state (Fig.  3A). Interestingly, these data parallel the changes in HuD expression seen in the stable cell lines (see above).
In response to NGF, however, AChE mRNA levels showed an increase only in the PC12 cells transfected with pcDNA and pcHuD and not in the cells expressing the antisense sequence to HuD (Fig. 3). In fact, AChE transcript levels in cells transfected with pcDNA increased by a maximum of ϳ3-fold (p Ͻ 0.006) following NGF treatment, demonstrating that these cells display a pattern of AChE mRNA expression similar to that seen in non-transfected PC12. NGF treatment of pcHuD cells dramatically increased AChE transcripts to a level greater than that observed in the pcDNA cells. This increase reached more than 2.5-and 5-fold within 24 and 48 h of NGF stimulation (p Ͻ 0.02), respectively (Fig. 3B). Additionally, the maximal increase in AChE mRNA expression in pcHuD cells was reached earlier during differentiation such that, after 48 h of NGF treatment, pcHuD cells contained significantly (p Ͻ 0.006) more AChE mRNAs than pcDNA cells. By contrast, PC12 cells engineered to express an antisense sequence to HuD (pDuH cells) did not exhibit any significant changes in the relative abundance of AChE mRNA during the course of differentiation.
We next considered whether the effect of HuD occurred directly on AChE transcripts. To this end, we used an approach similar to that used recently by others (56,57). Therefore, we transfected pcDNA, pcHuD, and pDuH cells with a luciferase reporter construct in which the 3Ј-UTR of AChE transcript was inserted. This 3Ј-UTR contains the conserved ARE, which, as mentioned, is known to be a cis-acting element recognized by HuD (Fig. 4A). These experiments were performed on undifferentiated cells in attempts to minimize the potential confounding impact of an increased level of endogenous AChE caused by NGF stimulation. In comparison to cells stably transfected with pcDNA, the activity of the reporter gene was significantly higher (p Ͻ 0.05) in cells expressing HuD. Importantly, this effect was completely abolished in pDuH cells, indicating that HuD is indeed an important factor regulating AChE expression via the 3Ј-UTR.
Changes in the Pattern of RNA-Protein Interactions during Neuronal Differentiation-Based on these findings, we also examined the pattern of proteins that could bind to the AChE 3Ј-UTR by both REMSA and Northwestern analyses. For these experiments, we used three probes: one corresponding to the full-length AChE 3Ј-UTR (245 nucleotides); a truncated version of the full-length probe (90 nucleotides) in which the ARE was deleted (ϪARE), thereby eliminating the putative HuD binding region; and a smaller probe encompassing the ARE (Fig. 5A). In REMSA, four distinct protein complexes could be visualized using the full-length AChE 3Ј-UTR probe (Fig. 5B). Although the relative abundance of all four complexes appeared to in-

FIG. 4. The AChE 3-UTR increases expression of a reporter construct in PC12 cells.
A, alignment of the AChE 3Ј-UTR from mouse, rat, and human shows the presence of a conserved ARE found within an AU-rich domain (underlined) and poly(A) signal (PAS). B, undifferentiated pcDNA, pcHuD, and pDuH cells were co-transfected with a reporter construct containing the full-length AChE 3Ј-UTR and with a constitutively expressed CAT plasmid. The luciferase activity was normalized to that seen with CAT and is expressed as a percentage of the activity found in the control pcDNA cells. Symbol indicates a significant difference (*, p Ͻ 0.05; n ϭ 5 independent experiments) from control cells. shows that, indeed, HuD migrates at ϳ42 kDa (see arrow). D, example of an ethidium bromide-stained agarose gel displaying AChE PCR product obtained from an immunoprecipitate isolated from differentiated pcHuD cells using an antibody against HuD, normal serum IgG, and A/G chimera Sepharose beads (Bds). The negative control lane (water, no RNA) is shown with a Ϫ. Note the presence of a PCR product corresponding to AChE only in the immunoprecipitate obtained with the HuD antibody. crease in PC12 cells stimulated with NGF for 72 h, two, in particular, showed dramatic increases in the binding intensity. The appearance of these complexes began after 48 h of differentiation (data not shown). By contrast, the formation of these complexes could not be observed in REMSA using ϪARE probe. Specificity of the protein complexes for the AChE 3Ј-UTR was demonstrated in competition assays using a 25-fold molar excess of unlabeled, full-length AChE 3Ј-UTR probe. In addition, REMSA using the small fragment of the AChE 3Ј-UTR that encompasses the ARE (see Fig. 5A) showed that indeed a protein complex could directly interact with this specific region (Fig. 5B). This latter complex appeared to correspond to one of the larger complexes seen with the full-length probe. It is important to note that, in all these experiments, both yeast tRNA and heparin were used to eliminate the possibility of nonspecific interactions.
Northwestern analyses were also performed to confirm that the pattern of RNA-protein interactions was indeed altered following NGF stimulation. In comparison to REMSA, Northwestern analyses allow the determination of the molecular mass of specific proteins that interact with the RNA probe. In agreement with our REMSA data, several proteins appeared to interact with the AChE 3Ј-UTR (Fig. 5C). Importantly, NGF stimulation markedly increased the relative abundance of these proteins. The greatest increase appeared at the level of a protein of ϳ42 kDa. Together with our functional data (see above), the presence of this ϳ42-kDa protein (approximate mass of HuD) in PC12 cells exposed to NGF for 72 h raised the possibility that this protein may in fact correspond to HuD. Western blot analysis confirmed the molecular mass of HuD at ϳ42 kDa (Fig. 5C). In control experiments, the truncated AChE 3Ј-UTR probe, i.e. ϪARE probe, could not bind any proteins in these Northwestern assays (data not shown).
In a final series of experiments, we verified that HuD is able to interact directly and bind with the AChE 3Ј-UTR in PC12 cells. To this end, we first immunoprecipitated HuD protein from a total protein extract obtained from 48-h differentiated pcHuD cells, isolated total RNA from the immunoprecipitate, and performed RT-PCR to detect AChE transcripts. This procedure has previously been successfully used to show the interaction between other ELAV/Hu family proteins and neuronal transcripts (34,58). As shown in Fig. 5D, we were able to specifically detect the presence of AChE transcripts from the immunoprecipitate obtained using the affinity-purified anti-HuD antibody. By contrast, no AChE PCR product could be visualized in total RNA isolated from immunoprecipitates obtained using either normal rabbit IgG or A/G chimera Sepharose beads alone (34). In these assays, the identity of the AChE PCR product was confirmed by sequencing. DISCUSSION In recent years, there has been significant progress in characterizing some of the transcriptional and post-transcriptional mechanisms involved in the regulation of the AChE gene in skeletal muscle (23,25,27,30,43,59,60). By contrast, there are only a few reports available that have examined the molecular events controlling AChE expression in neurons. Given its central role in terminating neurotransmission at cholinergic synapses, its additional non-catalytic function in the control, for example, of neurite outgrowth (see Introduction), and its involvement in a variety of neurological conditions such as Alzheimer's disease, post-traumatic stress disorder (61), and brain tumors (62), it appears important to gain a better understanding of the molecular pathways ultimately controlling the expression and localization of AChE in neurons. In the present study, we have examined some of the molecular events involved in regulating expression of the AChE gene during neuronal differentiation. Our findings indicate that, although transcription provides an initial boost necessary to rapidly increase AChE mRNA levels at the onset of neuronal differentiation, post-transcriptional events involving the 3Ј-UTR and the RNAbinding protein HuD are also involved and largely account for the pronounced and sustained increase in AChE transcripts seen at later stages of differentiation.
Several years ago, a series of experiments focusing on P19 cells induced to differentiate into neurons via retinoic acid treatment led Coleman and Taylor (31) to suggest that changes in mRNA stability was a key mechanism controlling the relative abundance of AChE transcripts during neuronal differentiation. However, this initial study provided little insight into the nature of the cis-and trans-acting elements that could be involved. In this context, cis-elements contained within the 3Ј-UTR have been shown to be critical for regulating the turnover rate of pre-synthesized mRNAs in a variety of cells. In particular, the element known as the AU-rich element, which is known to exist in different classes (63,64), has received considerable attention because of its key role in mRNA metabolism. Several studies have shown that this element can either stabilize or destabilize transcripts depending on the number of sequential repeats, variations of the basic sequence, and the identity of the trans-acting factor that binds (65,66). Among the proteins known to interact with the ARE is the ELAV-like family of Hu proteins. This group of RNA-binding proteins consists of HuR, HuB (Hel-N1), HuC, and HuD, of which HuB, -C, and -D are neuron-specific and HuR is more ubiquitously expressed (65,67). Several studies from different laboratories have identified HuD as a trans-acting factor binding and stabilizing a variety of different cellular transcripts including c-Myc (68), neuroserpin (69), tau (70), and GAP-43 (34).
Because AChE contains in its 3Ј-UTR a conserved ARE, we examined in the present studies whether HuD plays a functional role in regulating AChE expression in neurons. To this end, we used a series of distinct, yet complementary, approaches to determine whether indeed HuD could affect AChE mRNA expression. Using PC12 cells stably transfected to express HuD or an antisense sequence to its mRNA, we found that the relative abundance of AChE transcripts varied with the amount of HuD expressed by these cells. Moreover, HuD expression induced a faster and greater increase of AChE mRNA levels during NGF-induced differentiation of these cells. Importantly, this differentiating effect of NGF on AChE gene expression was completely blocked in cells expressing the HuD antisense. Furthermore, we also observed that expression of HuD could increase the expression of reporter transcripts engineered to contain the AChE 3Ј-UTR. Finally, immunoprecipitation experiments revealed that, indeed, HuD can directly interact with AChE transcripts. Taken together, these results strongly implicate HuD and the AChE 3Ј-UTR in the regulation of AChE mRNA expression during neuronal differentiation.
In comparable studies, HuD has been shown recently to be capable of regulating the stability and localization of transcripts encoding GAP-43 and tau in neurons (34,55,70,71). Interestingly, these proteins are known to be essential for the process of neurite outgrowth during neuronal differentiation. Because AChE has also been implicated in the events regulating neuritogenesis (see Introduction), it appears reasonable to speculate that HuD may, in fact, directly control neurite outgrowth by regulating the abundance and localization of a variety of transcripts encoding proteins that are key to this process. In agreement with this view, several studies have demonstrated that neurite extension is completely inhibited in PC12 cells stably expressing an antisense to HuD (34, 70 -72). Such mechanism would confer HuD a central organizing role and could provide neurons with an efficient post-transcriptional regulatory pathway. In one basic scenario, neurons could therefore simply induce HuD expression to trigger the process of neurite outgrowth, as opposed to activating multiple signaling cascades that would need to culminate in the activation of the transcriptional machinery of important genes via distinct promoter elements.
Although the data presented here strongly indicate that HuD and the AChE 3Ј-UTR are indeed important elements in controlling AChE mRNA levels in differentiating neurons, our findings do not imply that these elements act alone. It is well established, for example, that the RNA-binding protein AUF1 forms a complex with several other proteins involved in transcript elongation or stability (73). Similarly, it has become increasingly apparent that the precise regulation of mRNA stability, localization, and translation depends on the presence of multiple cis-acting elements that interact with several distinct trans-acting factors (65,66). Our analysis of RNA-protein interactions using both REMSA and Northwestern blotting is consistent with this notion because we observed several complexes and proteins that could interact with the AChE 3Ј-UTR. In future studies, it will be important to characterize these factors to gain a complete understanding of the cis-and transacting factors contributing to the regulation of AChE expression in neurons.
Our experimental approach has also allowed us to examine whether differentiation of PC12 cells was accompanied by a change in the transcriptional activity of the AChE gene. Our promoter-reporter studies and nuclear run-on assays both revealed that NGF initially induces a modest but significant increase in transcription. These results suggest therefore that, during the earlier stages of neuronal differentiation, transcriptional regulation plays a key role in the initial up-regulation of AChE mRNA expression as previously suggested by Greene and Rukenstein (31). In fact, these findings fit nicely with the recent demonstration that transcription, indeed, is involved in regulating AChE expression in neuronal cells (46,74) and with the transient transcriptional induction of GAP-43 following NGF treatment of PC12 cells (75,76). Together with the observation that, during the early phases of myogenic differentiation, transcriptional changes can also account for the initial increase in AChE mRNA levels (23), these findings are therefore consistent with a model in which developmental changes in AChE mRNA expression are initially regulated at the transcriptional level with post-transcriptional mechanisms becoming more important at later time points to ensure the adequate supply of AChE to differentiating muscle and neuron.
Because differentiation of PC12 cells is dependent on the continuous presence of NGF in the culture media, it appears reasonable to argue that NGF acts indirectly through HuD to regulate AChE expression in a manner similar to that described previously for the regulation of GAP-43 (see above). In addition to directly affecting transcription of target genes (77,78), these findings indicate therefore that some of the pleiotropic effects of neurotrophins on various neuronal populations may occur through RNA-binding proteins, which in turn affect, via post-transcriptional mechanisms, the expression of key mRNAs. In this context, previous animal studies have shown, for example, that treatment of axotomized facial motoneurons and rubrospinal neurons with neurotrophins leads to an increase in the expression of both GAP-43 and AChE transcripts (16, 79 -81). Based on the foregoing discussion, we can speculate that some of these neurotrophin-induced changes in gene expression observed in injured neurons may in fact be caused by post-transcriptional mechanisms operating at the level of transcript stability, localization, and translation. Future stud-ies should therefore make an attempt at unraveling the importance of these post-transcriptional events in vivo because, ultimately, these could lead to the design of additional therapeutic strategies aimed at promoting neuronal regeneration and survival.