Overexpression of HuD, but Not of Its Truncated Form HuD I+II, Promotes GAP-43 Gene Expression and Neurite Outgrowth in PC12 Cells in the Absence of Nerve Growth Factor

Authors

  • Kim D. Anderson,

  • Melissa A. Morin,

  • Andrea Beckel-Mitchener,

  • Charlotte D. Mobarak,

  • Rachael L. Neve,

  • Henry M. Furneaux,

  • Richard Burry,

  • Nora I. Perrone-Bizzozero


  • Lippincott Williams & Wilkins, Inc., Philadelphia

  • Abbreviations used: BSA, bovine serum albumin; Elav, embryonic lethal abnormal visual; GAP, growth-associated protein; G3PD, glyceraldehyde 3-phosphate dehydrogenase; GST, glutathione S-transferase; NF-M, neurofilament medium-weight 160-kDa protein; NGF, nerve growth factor; RRM, RNA recognition motif; SDS, sodium dodecyl sulfate; TBST, Tris-buffered saline (pH 7.4) with 0.1% Triton X-100; UTR, untranslated region.

Address correspondence and reprint requests to Dr. N. I. Perrone-Bizzozero at Department of Neurosciences, University of New Mexico School of Medicine, Albuquerque, NM 87131, U.S.A. E-mail: nbizzozero@salud.unm.edu

Abstract

Abstract: We have previously shown that the RNA-binding protein HuD binds to a regulatory element in the growth-associated protein (GAP)-43 mRNA and that this interaction involves its first two RNA recognition motifs (RRMs). In this study, we investigated the functional significance of this interaction by overexpression of human HuD protein (pcHuD) or its truncated form lacking the third RRM (pcHuD I+II) in PC12 cells. Morphological analysis revealed that pcHuD cells extended short neurites containing GAP-43-positive growth cones in the absence of nerve growth factor (NGF). These processes also contained tubulin and F-actin filaments but were not stained with antibodies against neurofilament M protein. In correlation with this phenotype, pcHuD cells contained higher levels of GAP-43 without changes in levels of other NGF-induced proteins, such as SNAP-25 and tau. In mRNA decay studies, HuD stabilized the GAP-43 mRNA, whereas HuD I+II did not have any effect either on GAP-43 mRNA stability or on the levels of GAP-43 protein. Likewise, pcHuD I+II cells showed no spontaneous neurite outgrowth and deficient outgrowth in response to NGF. Our results indicate that HuD is sufficient to increase GAP-43 gene expression and neurite outgrowth in the absence of NGF and that the third RRM in the protein is critical for this function.

Growth-associated protein (GAP)-43 is expressed in neurons during the initial establishment and remodeling of neural connections (Benowitz and Routtenberg, 1997; Oestreicher et al., 1997). Overexpression of GAP-43 in adult transgenic mice induces neuronal sprouting (Aigner et al., 1995; Holtmaat et al., 1995). In contrast, depletion of GAP-43 results in growth cone collapse, abnormal axonal pathfinding, and deficient formation of topographic maps in the brain (Aigner and Caroni, 1995; Strittmatter et al., 1995; Maier et al., 1999; Zhu and Julien, 1999). Overexpression of GAP-43 in PC12 cells has been shown to enhance nerve growth factor (NGF)-induced neurite outgrowth (Yankner et al., 1990) and increase process formation in nonneuronal cells (Zuber et al., 1989; Aarts et al., 1998).

GAP-43 gene expression is regulated at both the transcriptional and posttranscriptional levels. Sequences in the promoter region are crucial for silencing GAP-43 promoter activity outside the nervous system and for controlling its activity within (Nedivi et al., 1992; Eggen et al., 1994; Vanselow et al., 1994; Chiaramello et al., 1996; Namgung et al., 1997; Weber and Skene, 1997). In addition, GAP-43 expression is posttranscriptionally regulated both in PC12 cells and in vivo (Perrone-Bizzozero et al., 1991, 1993; Cantallops and Routtenberg, 1999). This process depends on the presence of highly conserved U-rich sequences in the GAP-43 3′ untranslated region (UTR) (Kohn et al., 1996). These sequences contain recognition motifs for the binding of neuronal-specific RNA-binding proteins (Kohn et al., 1996; Irwin et al., 1997). One of these proteins was recently identified as HuD, a member of the Elav (embryonic lethal abnormal visual system) family of RNA-binding proteins first discovered in Drosophila (Campos et al., 1985; Homyk et al., 1985; Robinow et al., 1988). We recently found that HuD binds to a regulatory element in the 3′ UTR of the GAP-43 mRNA, suggesting that this protein controls the half-life of the mRNA (Chung et al., 1997; Tsai et al., 1997).

Four members of the human Elav-like family have been discovered and characterized: HuR, Hel-N1 (also known as HuB), HuC, and HuD (Szabo et al., 1991; King et al., 1994; Sakai et al., 1994; Ma et al., 1996). HuR is expressed ubiquitously, whereas the other three proteins are exclusively expressed in postmitotic neurons and neuroendocrine tumors (Dalmau et al., 1992; Gao et al., 1994; Marusich et al., 1994; Okano and Darnell, 1997; Wakamatsu and Weston, 1997). Furthermore, recent studies suggest that these proteins may induce neuronal differentiation (Wakamatsu and Weston, 1997; Akamatsu et al., 1999; Antic et al., 1999; Kasashima et al., 1999). All four members of the Elav-like family contain three RNA recognition motifs (RRMs) (Kenan et al., 1991). The first two RRMs (I and II) interact with U-rich elements in the 3′ UTR of several mRNAs (Chung et al., 1996, 1997; Ross et al., 1997), whereas RRM III interacts with long poly(A) sequences (Ma et al., 1997). These results suggest that RRM III may be important in controlling the interaction of Elav-like proteins with the poly(A) tails of mRNAs.

In this report, we investigated the function of wild-type HuD and HuD I+II, a mutant form of the protein lacking its third RRM, on GAP-43 gene expression and PC12 cell differentiation. We found that overexpression of HuD increased GAP-43 levels and induced PC12 cells to undergo process outgrowth even in the absence of NGF. These spontaneous neurites contained GAP-43-positive growth cones but were shorter than those of NGF-treated PC12 cells. Analysis of the effect of HuD on other NGF-induced proteins revealed that it did not increase the levels of the neurofilament and tau proteins that are characteristic of more mature neurites. In contrast to HuD, overexpression of HuD I+II did not result in process outgrowth even in the presence of NGF. The mutant protein failed to stabilize the GAP-43 mRNA, demonstrating that the third RRM in HuD is critical for its effect in GAP-43 gene expression and PC12 cell differentiation.

MATERIALS AND METHODS

Cloning and transfection of HuD and HuD I+II

The cDNAs for the full-length HuD protein and its truncated form HuD I+II were excised from pGEX-2 vectors with BamHI (Chung et al., 1996) and cloned into myc-tag-pGEM3Z. HuD comprised amino acids 2-373, and HuD I+II contained residues 2-216. The cDNAs were digested with HindIII and SmaI and cloned into the HindII/EcoRV sites in the pcDNA3 vector (Invitrogen, Carlsbad, CA, U.S.A.). The resulting plasmids contained the myc-tag sequence (EQKLISEEDL) followed by either the HuD or HuD I+II cDNA sequence and were called pcHuD and pcHuD I+II, respectively. The original PC12 cell line (Greene and Tischler, 1976) and two other PC12 cell clones, N21 and N36 cells (Burry and Perrone-Bizzozero, 1993), were stably transfected with either one of the HuD constructs in pcDNA3 or the vector alone by electroporation. Transfected cultures of PC12-N21, PC12-N36, and the original PC12 line were selected with G418 (500 μg/ml; Calbiochem, San Diego, CA, U.S.A.) for 2 weeks and used without further cloning.

Cell morphology

PC12 cells were grown for either 24 or 96 h in RPMI 1640 medium supplemented with 7.5% horse serum and 2.5% fetal calf serum (Sigma, St. Louis, MO, U.S.A.) as previously described (Perrone-Bizzozero et al., 1993). For NGF induction, cells were treated with 7S NGF (100 ng/ml; Sigma) for 96 h. For morphological studies, cells were grown in the absence or presence of NGF and then fixed with 4% paraformaldehyde for 30 min. Cells were stained with 0.1% Coomassie Blue (Fisher, Norcross, GA, U.S.A.) in 50% methanol/10% acetic acid to aid in neurite visualization as previously described (Perrone-Bizzozero et al., 1986). Five hundred cells were counted per condition per cell type using an Olympus CK microscope at ×400 magnification. Cells were classified as exhibiting neurites when they had one or more processes extending >20 μm in length, which constitutes the average length of the cell body. Cells were counted blindly by three independent observers, and the mean percentages of cells with neurites were determined from three separate experiments. Statistical analysis was performed using a two-tailed Student's t test.

Immunocytochemistry

For immunocytochemical experiments, cells were grown for 96 h on 12-mm-diameter circular glass coverslips (Fisher) that were pretreated with 0.1% glutaraldehyde and 0.2 mg/ml poly-L-lysine. Cells were fixed with 4% paraformaldehyde for 30 min at room temperature. Fixed cells were incubated for 20 min at room temperature in buffer containing 2% bovine serum albumin (BSA) in Tris-buffered saline (pH 7.4) with 0.1% Triton X-100 (TBST). GAP-43 was detected with a sheep polyclonal antibody (Benowitz et al., 1988) at a 1:250 dilution. HuD and HuD I+II were detected with an antibody generated against an N-terminal peptide that is uniquely present in HuD. This antibody was affinity-purified (αHuD-AfP) and used at a 1:100 dilution in BSA-TBST. The cells were incubated with the primary antibodies for 2 h at room temperature. After rinsing with TBST, cells were incubated with donkey anti-sheep—fluorescein isothiocyanate (1:50) and goat anti-rabbit—rhodamine isothiocyanate (1:50) (both from Sigma Immunochemicals, St. Louis) in BSA-TBST for 1 h in the dark. To probe for F-actin, cells previously fixed on coverslips with 4% paraformaldehyde were incubated in 1.6 μM phalloidin—fluorescein isothiocyanate (Sigma) in phosphate-buffered saline containing 1% Triton X-100 for 1 h at room temperature. Coverslips were rinsed with 1× phosphate-buffered saline and mounted onto glass slides (Becton Dickinson, Franklin Lakes, NJ, U.S.A.) with PermaFluor aqueous mounting media (Immunon, Pittsburgh, PA, U.S.A.) and dried overnight at 4°C in the dark. Photographs were taken on a Zeiss Axiovert microscope at ×400 magnification.

Western blots

For western blot analysis, cells were grown in the absence or presence of NGF for either 24 or 96 h and then harvested in 0.1% sodium dodecyl sulfate (SDS). Protein content determinations were performed for each cell type by using the method of Bradford (1976). Recombinant Elav/Hu proteins were expressed in bacteria as glutathione S-transferase (GST) fusion polypeptides or hexahistidine-tagged constructs, and 500 ng of purified proteins was used for western blots. PC12 protein extracts containing 40 μg of protein were analyzed using 4-15% SDS Tris-HCl ready gels (Bio-Rad, Hercules, CA, U.S.A.) as previously described (Perrone-Bizzozero et al., 1996). All gels were transferred to Immuno-Blot polyvinylidene difluoride membranes using a Mini Trans-Blot Module (Bio-Rad) at 100 V for 1 h at 4°C as previously described (Towbin et al., 1979). After transfer, blots were stained with Coomassie Brilliant Blue and scanned for image analysis using the PhotoAnalyst I system (FotoDyne, Hartland, WI, U.S.A.). Membranes were incubated with a sheep polyclonal antibody for GAP-43 (1:600; Benowitz et al., 1988), our αHuD-AfP (1:200), or mAb16A11 (Barami et al., 1995), and proteins were detected using a colorimetric method (nitro blue tetrazolium/3-bromo-4-chloro-5-indolyl phosphate; Sigma) as described by Sower et al. (1995). Specific protein levels in each band were corrected by the total protein loaded in the lane using the density of the Coomassie Blue staining as previously described (Perrone-Bizzozero et al., 1996). Tubulin was detected with the monoclonal anti-β-tubulin clone 2.1 (1:200; Sigma Immunochemicals), tau with a rabbit polyclonal antibody (1:750; Dako, Carpinteria, CA, U.S.A.), the neurofilament medium-weight 160-kDa protein (NF-M) with a monoclonal antibody (1:500; Boehringer Mannheim, Indianapolis, IN, U.S.A.), and SNAP-25 with a monoclonal antibody (1:150; Chemicon International, Temecula, CA, U.S.A.).

mRNA decay studies

To examine the effect of HuD or HuD I+II on GAP-43 mRNA stability, PC12-N36 cells transfected with the GAP-43 cDNA in the inducible pMEP4 vector (Tsai et al., 1997) were cotransfected with either the pcHuD or pcHuD I+II vector described above. Stable transfectants were selected with both hygromycin B (150 μg/ml; Calbiochem) and G418 (500 μg/ml). For mRNA decay studies, cells were induced for 16 h with 5 μM CdCl2. After induction the CdCl2 was removed from the cultures, and samples were harvested at 0, 1, 3, 4, and 6 h. Following RNA extraction, 15 μg of total RNA from each sample was electrophoresed on 1.1% formaldehyde-agarose gels as described by Perrone-Bizzozero et al. (1993). Membranes were probed for GAP-43 mRNA or glyceraldehyde 3-phosphate dehydrogenase (G3PD) mRNA using 32P-radiolabeled cDNA probes generated using random priming (Prime-a-Gene; Promega, Madison, WI, U.S.A.). The radioactivity associated with each band was measured using a PhosphorImager (Storm 860; Molecular Dynamics, Sunnyvale, CA, U.S.A.) and the ImageQuant software package. The levels of the GAP-43 mRNA were corrected for loading using the levels of G3PD and then expressed relative to the values at t = 0. The half-life (t1/2) of the GAP-43 mRNA was determined using the exponential function Mt = Moet, where M is the amount of mRNA at time t, Mo is the amount of mRNA at t = 0, and λ = (ln 2)/t1/2. The data were expressed as ln[Mt], and the t1/2 was calculated using linear regression analysis as previously described (Perrone-Bizzozero et al., 1993; Tsai et al., 1997).

RESULTS

None of the antibodies currently available against Elav-like proteins recognizes specific members of this family (King and Dropcho, 1996; Wakamatsu and Weston, 1997). Thus, we generated polyclonal antibodies specific for HuD using an N-terminal peptide that is uniquely present in this protein. As shown in Fig. 1A, the antibodies specifically recognized recombinant human HuD and HuD I+II proteins but did not react against any of the other neuronal Elav-like proteins) (HuC or Hel-N1). It is of interest to note that although recombinant HuD I+II had a deletion of the C-terminal domain that rendered it mAb16A11-negative (Fig. 1A), its migration in the gel was similar to that of full-length HuD. This anomalous migration was also observed in the transfected protein (see Fig. 4). Although the precise cause of this anomaly is still unknown, there are several examples of proteins with abnormal migration in SDS-polyacrylamide gels, including GAP-43, a 25-kDa protein that migrates at 43 kDa (Benowitz et al., 1987).

Figure 1.

Immunological detection of HuD with a specific anti-peptide antibody. A: Specificity of the anti-HuD antibody. Recombinant Hu proteins (500 ng) were analyzed by western blotting using our affinity-purified anti-HuD sera (αHuD-AfP) and the anti-Hu mAb16A11. αHuD-AfP recognizes GST-HuD and GST-HuD I+II but not GST-HuC or His6X-tagged Hel-N1. mAb16A11 recognizes all neuronal Hu proteins except HuD I+II. B: Distribution of HuD and GAP-43 in untransfected PC12 cells grown in the absence or presence of NGF for 96 h. Cells were fixed with 4% paraformaldehyde and probed for HuD and GAP-43 proteins as indicated in Materials and Methods. Bar = 50 μm.

Figure 4.

Analysis of levels and distribution of GAP-43 in control and NGF-treated PC12 cells transfected with HuD or HuD I+II. Immunoblots show (A) GAP-43 and (B) HuD protein levels from cells grown either in the absence or in the presence of NGF for 24 h. GAP-43 levels in pcHuD cells were significantly greater than those in PC12 or pcHuD I+II cells (p < 0.001). C: Distribution of GAP-43 in pcHuD- and pcHuD I+II-transfected cells. Both cell types were cultured for 96 h in the absence or presence of NGF, then fixed with 4% paraformaldehyde, and prepared for immunocytochemistry. GAP-43 was detected with a fluorescein isothiocyanate-conjugated secondary antibody against a sheep polyclonal antibody. Arrows point to GAP-43-positive growth cones. Bar = 50 μm.

FIG. 1.

FIG. 4.

Once we established the specificity of our antibodies, we used these to examine the distribution of HuD and GAP-43 in control and NGF-treated PC12 cells (Fig. 1B). In untransfected PC12 cells, both GAP-43 and HuD proteins were localized diffusely throughout the cytoplasm. After 4 days of culture in the presence of NGF, GAP-43 was still present in the cell body but was also intensely distributed in the neuritic processes. In contrast, HuD was primarily distributed in the cell body and in the proximal part of the processes, with little or no reactivity at the terminal part of the neurites and growth cones. Some of the Elav/Hu proteins are primarily localized to the nucleus and shuttle to the cytosol (Fan and Steitz, 1998), but others seem to be primarily cytosolic (Antic and Keene, 1998). The cytosolic localization of HuD further supports the idea that this protein may not be involved in splicing mechanisms but rather in cytosolic processes such as mRNA stability and/or translational regulation.

Overexpression of HuD induces neurite outgrowth in PC12 cells in the absence of NGF

To examine the effect of HuD on GAP-43 expression and neuronal differentiation, PC12 cells were transfected with pcDNA3 vectors containing either the complete human HuD cDNA (pcHuD) or a truncated form lacking its third RRM (pcHuD I+II). Transfection experiments were performed on the original PC12 cell line (Greene and Tischler, 1976) and the PC12-N21 clone, which shows responses to NGF similar to those of the original line (Burry and Perrone-Bizzozero, 1993). Also, to avoid selection of a particular PC12 cell phenotype, stable transfectants of both the original PC12 cell line and the N21 clone were selected without further cloning and used for all the experiments. Because initial studies demonstrated that both transfected lines showed identical phenotypic responses after HuD transfection, we chose to use primarily the PC12-N21 subclone because it represents a more homogeneous cell population.

PC12-N21 cells transfected with either pcDNA3 or pcHuD were grown for 4 days in culture, and then the percentage of cell containing processes was determined as described in Materials and Methods. As shown in Fig. 2, in the absence of NGF only pcHuD-transfected cells exhibited spontaneous neurite outgrowth and the typical cell body flattening found in NGF-treated PC12 cells, whereas untransfected cells and cells transfected with pcHuD I+II remained undifferentiated.

Figure 2.

Morphological properties of control and HuD-transfected PC12-N21 cells grown in the presence or absence of NGF. Photomicrographs show the morphology of untransfected PC12 cells and those from pcHuD- and pcHuD I+II-transfected PC12 cells grown for 96 h in either the absence or presence of 100 ng/ml NGF. Note that the morphological features of cells transfected with HuD are quite different from those transfected with HuD I+II. Bar = 50 μm. All pictures were taken at the same magnification.

FIG. 2.

Subsequent experiments examined the response of control and transfected PC12 cells to treatment with NGF. Untransfected PC12 cells and cells transfected with pcDNA3 vector alone showed minimal differentiation in the absence of NGF and responded as expected to the growth factor with increased process outgrowth (Figs. 2 and 3A). NGF-treated pcHuD cells were not different from the untreated cultures except that they exhibited longer neurites. Having determined that excess HuD was sufficient to induce neurite outgrowth in PC12 cells, we next examined whether all three RNA-binding motifs in this protein were required for this function. PC12 cells were transfected with a truncated version of the HuD protein (HuD I+II), which lacks the third RRM and thus cannot interact with the poly(A) tail. In the absence of NGF, pcHuD I+II cells showed minimal differentiation and were not different from pcDNA3-transfected cells or untransfected PC12 cells (Figs. 2 and 3A). Moreover, pcHuD I+II cells did not respond to NGF with the same extent of process outgrowth as did pcDNA3-transfected cells.

Figure 3.

Statistical analysis of the extent of neurite outgrowth in control and HuD-transfected PC12 cells. At least 500 cells were analyzed per condition and classified as exhibiting neurite outgrowth as described in Materials and Methods. Data are mean ± SD (bars) values obtained in three separate experiments. Data were analyzed using a two-tailed Student's t test. A: pcHuD cells are morphologically different from pcDNA3 (*p < 0.002) and pcHuD I+II cells (**p < 0.005). B: Neurite length analysis in control and pcHuD-transfected PC12 cells. A greater proportion of NGF-treated PC12 cells exhibit longer processes than pcHuD cells (*p < 0.05). NGF resulted in a significant increase in the proportion of pcHuD-transfected cells with processes longer than two cell body lengths (p < 0.05).

FIG. 3.

Statistical analysis of these results (Fig. 3A) indicated that overexpression of HuD significantly increased neurite outgrowth in PC12 cells in the absence of NGF (p < 0.002). In contrast, overexpression of pcHuD I+II did not promote neurite outgrowth, and the phenotype of these cells remained the same as in nonstimulated PC12 or pcDNA3 cells. Comparison of pcDNA3-transfected and untransfected PC12 cells indicated that the vector alone did not have any effect on cell differentiation in either the presence or absence of NGF. On NGF treatment, cells transfected with pcHuD I+II showed minimal process outgrowth, suggesting that the mutant form of the protein interferes with a step required for neurite formation. Comparison of the effect of HuD overexpression and NGF on neurite length (Fig. 3B) indicated that the majority of processes in pcHuD-transfected cells were significantly shorter than those induced by the growth factor and did not elongate beyond two cell body lengths after 4 days in culture. Addition of NGF to pcHuD cells caused a significant elongation of their short neurites, which reached the same process length as in NGF-treated untransfected cells after 4 days in culture (Fig. 3B). These results suggest that HuD may promote the expression of proteins involved in neurite initiation rather than in elongation. Supporting this idea, western blot analysis of these cells indicated that HuD selectively increased GAP-43 expression without affecting the expression of proteins induced after longer times of NGF exposure (see Figs. 5 and 6).

Figure 5.

Effect of HuD on expression of GAP-43 and other neuronal proteins in PC12-N21 cells. Untransfected PC12 and pcHuD- and pcHuD I+II-transfected cells were cultured for 24 h in the absence of NGF, and proteins (40 μg) were processed for western blot analysis as described in Materials and Methods. A: Immunoblots show levels of GAP-43, SNAP-25, tubulin, tau 80K, 42K, and 32K isoforms (arrowheads), and NF-M in untransfected or pcHuD- or pcHuD I+II-transfected cells. B: Densitometric quantitation of protein levels. All three main isoforms of tau were analyzed together. Data are mean ± SD (bars) values for five separate experiments. *p < 0.001, **p < 0.005.

Figure 6.

Immunochemical characterization of neurites in control and pcHuD- and pcHuD I+II-transfected PC12-N21 cells. Cells were cultured for 96 h and then processed for immunocytochemistry for β-tubulin and NF-M and for phalloidin staining for F-actin. Arrows point to F-actin-positive growth cones. Bars = 50 μm.

FIG. 5.

FIG. 6.

Levels and distribution of GAP-43 in pcHuD and pcHuD I+II cells

The expression of GAP-43 in both pcHuD- and pcHuD I+II-transfected cells was examined by western blots and immunocytochemistry (Fig. 4). As expected from their morphological properties, GAP-43 levels in nonstimulated pcHuD cells were much higher than in their untransfected counterparts, reaching the levels seen in NGF-induced control PC12 cells (Fig. 4A). In contrast, we found that pcHuD I+II cells expressed significantly less GAP-43 protein than either untransfected PC12 cells or pcHuD-transfected cells, both in the absence and in the presence of NGF. In correlation with the presence of HuD expression vectors, pcHuD and pcHuD I+II cells exhibited increased levels of HuD proteins (a twofold increase relative to untransfected PC12 cells) under basal conditions, and these levels were not further augmented in the presence of NGF (Fig. 4B).

Immunocytochemical analysis of GAP-43 expression showed that pcHuD-transfected cells contained brightly stained growth cones, similar to those found in NGF-differentiated untransfected PC12 cells (Fig. 4C). Treatment of these cells with NGF had no additional effect on the cellular localization of GAP-43 except that neurites were significantly longer. In contrast, pcHuD I+II cells showed a pattern of GAP-43 staining similar to that of undifferentiated PC12 cells. Addition of NGF to these cells did not alter the distribution of GAP-43 as the mutant protein blocked NGF-induced process outgrowth.

Selectivity of the effect of HuD on GAP-43 expression

To examine whether HuD was acting specifically on GAP-43 expression or more generally on multiple factors, we compared the levels of GAP-43 in these cells with those of other proteins that are induced with NGF such as tau, NF-M, SNAP-25, and tubulin (Fig. 5A). In contrast to GAP-43, the expression of either SNAP-25, a neuronal protein involved in synaptic release, or the cytoskeletal protein tubulin was not altered 24 h after plating pcHuD cells in the absence of NGF. The microtubule-associated protein tau and the neurofilament protein NF-M are also involved in neuronal differentiation, albeit at a later time than GAP-43 (Lee and Page, 1984). Because their mRNAs also interact with Hu proteins (Chung et al., 1997; Antic et al., 1999), we compared the levels of these proteins and GAP-43 in pcHuD cells. As shown in Fig. 5, overexpression of HuD did not increase the levels of the main splicing variants of the tau protein or those of NF-M (Fig. 5B). Comparison of the effect of pcHuD and pcHuD I+II in the expression of these proteins revealed that the overexpression of the mutant protein did not promote any changes in gene expression except for a decreased expression of NF-M (Fig. 5B).

Additional studies used immunocytochemistry to characterize the cytoskeletal composition of the neurites induced by the HuD protein. As shown in Fig. 6, neurites in pcHuD-transfected cells contained β-tubulin- and F-actin-positive growth cones, but they were not stained with an antibody against NF-M. Overexpression of HuD I+II, however, did not result in any changes in the distribution of the main cytoskeletal proteins. Because neurofilaments are markers of mature neurites (Jacobs and Stevens, 1986), these results support the idea that HuD may be involved in early events in neuronal differentiation.

Overexpression of HuD, but not HuD I+II, selectively stabilized the GAP-43 mRNA

Because HuD binds to an element in the GAP-43 3′ UTR that controls mRNA stability, subsequent studies investigated the effect of HuD and HuD I+II on the half-life of this mRNA. For these studies we used a GAP-43-deficient subclone, PC12-N36. This clone offers the advantage that it is possible to study GAP-43 mRNA stability in the absence of endogenous transcripts and without the use of transcriptional inhibitors that alter mRNA turnover (Perrone-Bizzozero et al., 1991; Kohn et al., 1996). We have previously used PC12-N36 cells to determine the half-life of the GAP-43 mRNA transcripts expressed under the control of the metallothionein IIa promoter in the expression vector pMEP-4 (Tsai et al., 1997). To examine the effect of HuD proteins in GAP-43 mRNA stability (Fig. 7), cells were cotransfected with the GAP-43 cDNA in pMEP-4 and with either pcHuD or pcHuD I+II. Transfected cells were treated with cadmium to induce GAP-43 expression, and RNAs were isolated at different times after the removal of the metal for northern blot analysis (Fig. 7A). Analysis of mRNA levels at different decay times (Fig. 7B) revealed that the GAP-43 mRNA level in control cells and HuD I+II-transfected cells followed an exponential decay. Using the equation Mt = Moet to calculate the half-life of the mRNA (see Materials and Methods), we found that HuD I+II did not have a significant effect in the stability of the GAP-43 mRNA. In contrast, expression of the full-length HuD significantly increased the half-life of the GAP-43 mRNA from 5 to 8 h (Fig. 7B). These results indicate that the interaction of the third RRM of HuD with the poly(A) tail is required for stabilization of the GAP-43 mRNA.

Figure 7.

Overexpression of full-length HuD stabilizes GAP-43 mRNA. A: Northern blot of GAP-43 and G3PD mRNAs from PC12-N36 cells cotransfected with GAP-43 in an inducible expression vector (pMEP4) and either pcHuD or pcHuD I+II. Cells were induced with cadmium for 16 h and harvested at 0, 1, 3, 4, and 6 h after removal of the metal. Control lanes (Co) show GAP-43 mRNA levels due to basal expression of the promoter in the absence of Cd. B: GAP-43 mRNA decays in transfected PC12 cells. GAP-43 mRNA levels at different times after removal of cadmium were normalized to those of G3PD. Data are mean ± SD (bars) percentages of remaining mRNA for five separate experiments. The half-life of the mRNA was calculated as described in Materials and Methods. Statistical analysis indicated that HuD but not HuD I+II significantly increased the half-life of the GAP-43 mRNA (p < 0.05).

FIG. 7.

DISCUSSION

These studies were undertaken to investigate the functional role of the neuronal-specific RNA-binding protein HuD in GAP-43 gene expression and neuronal differentiation. Here we show that overexpression of HuD is sufficient to promote neurite outgrowth in PC12 cells grown in the absence of NGF. This phenotypic change was accompanied by a selective increase in GAP-43 expression and its localization to the growth cones. This cellular phenotype was similar but not identical to that observed in NGF-induced PC12 cells, suggesting that HuD mediates only certain NGF responses. Analysis of mRNA turnover revealed that HuD delayed the degradation of the GAP-43 mRNA through a process that required the presence of the third RRM motif in the RNA-binding protein. Together, these results suggest that stabilization of the GAP-43 mRNA by HuD is a critical step in the control of GAP-43 protein expression and PC12 cell differentiation.

HuD is a member of the Elav family of RNA-binding proteins first discovered in Drosophila (Campos et al., 1985; Szabo et al., 1991). In flies, the elav gene is required for the normal development and maintenance of the nervous system (Campos et al., 1985). Likewise, in vertebrates, Hel-N1, HuC, and HuD have been implicated in promoting differentiation of neuronal cells (Barami et al., 1995; Okano and Darnell, 1997; Wakamatsu and Weston, 1997). With regard to their mechanisms of action, the Elav family of RNA-binding proteins is known to bind to AU-rich sequences within the 3′ UTR of mRNAs involved in cell growth and differentiation (Levine et al., 1993; Jain et al., 1997). HuD in particular was shown to bind to U-rich regulatory elements within the 3′ UTR of mRNAs expressed in the nervous system such as those for GAP-43, c-fos, c-myc, N-myc, and tau (Chung et al., 1997; Ross et al., 1997). These properties support the hypothesis that HuD and other Elav-like proteins play a significant role in the posttranscriptional control of gene expression during neuronal differentiation.

Although HuD binds to more than one mRNA (Chung et al., 1997), the effects of this RNA-binding protein could be made selective to a specific mRNA by a combination of different mechanisms. As described with Hel-N1, selectivity could be achieved by the localization of the mRNA or the RNA-binding protein to a specific pool of ribonuclear particles (Gao and Keene, 1996). In addition, specificity could be obtained by the interaction of the Elav-like protein or the mRNA with either other polysomal proteins or with specific cytoskeletal components (Antic and Keene, 1998). These issues are under investigation, but it is apparent that the effect of HuD during the first day in culture may be selective for GAP-43. First, overexpression of the RNA-binding protein increased the levels of GAP-43 but did not increase the levels of other neuronal proteins induced by NGF, such as SNAP-25, nor did it affect the levels of tau or NF-M, two other targets of Elav-like proteins. A similar phenomenon was observed by overexpression of the GAP-43 3′ UTR in PC12 cells, a treatment that titrates out GAP-43 mRNA-binding proteins and resulted in the selective degradation of the endogenous GAP-43 mRNA (Neve et al., 1999). Under these conditions, we found that although neurite outgrowth was blocked in the cells, the levels of other potential targets of HuD, such as those of the c-fos and tau mRNAs (Chung et al., 1997), were not affected.

Unlike GAP-43, SNAP-25, NF-M, and tau are induced during the late stages of axonal elongation (Lee and Page, 1984; Osen-Sand et al., 1993; Esmaeli-Azad et al., 1994). Consistent with this observation, we found that although both HuD and NGF treatment resulted in the outgrowth of neurites bearing GAP-43-positive growth cones, the neurites produced by HuD overexpression were generally shorter and greater in numbers that those observed with NGF treatment. In agreement with their morphological properties, histochemical analysis of the cytoskeletal elements present in the processes revealed that they were stained for F-actin and tubulin but they did not contain neurofilament proteins. As GAP-43 is involved in neurite initiation (Goslin et al., 1988; Zuber et al., 1989; Yankner et al., 1990; Aarts et al., 1998), these results further suggest that HuD plays a role in early events in neurite outgrowth. It is possible that other Elav/Hu proteins may be involved in controlling gene expression at later stages of neuronal differentiation. In agreement with this notion, HuC and HuB/Hel-N1 were recently found to induce the expression of neurofilament proteins and neuronal differentiation in neural cells in culture and in vivo (Akamatsu et al., 1999; Antic et al., 1999).

Several lines of evidence suggest that HuD induces GAP-43 gene expression in PC12 cells by promoting the stabilization of the mRNA. First, we previously showed that HuD binds to a highly conserved regulatory element within the GAP-43 3′ UTR that controls mRNA stability (Chung et al., 1997; Tsai et al., 1997). Second, we found that both GAP-43 mRNA and HuD binding was enhanced by phorbol ester treatment (Perrone-Bizzozero et al., 1993; Tsai et al., 1997). Finally and most importantly, we found that overexpression of HuD prolonged the half-life of the GAP-43 mRNA and increased GAP-43 expression even in the absence of NGF. Not only did increased levels of HuD result in increased GAP-43 mRNA stability and gene expression, but also decreased HuD expression by antisense RNA treatment was sufficient to destabilize this mRNA (C. D. Mobarak et al., manuscript submitted).

Another important observation in our study relates to the lack of function of a truncated form of HuD containing only the first two RRMs. RRMs I and II in this protein are required for binding to U-rich elements in the 3′ UTRs of GAP-43 mRNA (Chung et al., 1997), c-fos mRNA (Chung et al., 1996), and N-myc mRNA (Ross et al., 1997). In contrast, the third RRM of HuC, HuD, and HuR have been shown to interact with long poly(A) tails (Abe et al., 1996; Ma et al., 1997). The importance of the third RRM in the function of Elav-like proteins is suggested by the observation that two mutations (elavts1 and elavFliJ2) in the third RRM of the Drosophila Elav protein lead to translational truncation and the subsequent functional impairment of the mutant proteins (Samson et al., 1995). The region where these mutations occur is an 11-amino acid sequence that is identical in Drosophila Elav and RBP9 and human Elav-like proteins HuD and HuC (Samson et al., 1995). Mutant flies become flightless and uncoordinated (Homyk et al., 1980) and some die within a few days (Campos et al., 1985). However, the mechanisms underlying functional impairment of these mutant proteins have not been investigated as yet. In this study, we showed that overexpression of a truncated form of HuD, which lacks the third RRM, does not stabilize the GAP-43 mRNA, suggesting that interactions of the third RRM with the poly(A) tail may be important for controlling mRNA stability. Supporting this idea, Fan and Steitz (1998) showed that the third RRM motif in the ubiquitously expressed Elav-like protein HuR was required for stabilizing ARE (AU-rich element)-containing mRNAs such as that of c-fos. Our results demonstrate that the inability of HuD I+II to stabilize the GAP-43 mRNA has profound physiological consequences as it prevented the induction of GAP-43 expression and neurite outgrowth in response to NGF. Because these effects were observed in the presence of endogenous full-length HuD, these results further suggest that the truncated form of this protein, HuD I+II, displays a dominant-negative phenotype.

In conclusion, our results demonstrate that overexpression of HuD is sufficient to induce GAP-43 expression and neurite outgrowth in PC12 cells. In view of the role of GAP-43 in neurite outgrowth and the tight correlation between HuD's effects on GAP-43 mRNA half-life and protein levels, we propose that the effect of HuD on PC12 cell differentiation is mediated by stabilization of the GAP-43 mRNA and subsequent increases in GAP-43 gene expression. In addition to PC12 cells, overexpression of HuD was found to increase early neurite out-growth in primary neuronal cultures (K.D.A., unpublished data), suggesting that this protein displays a similar role in neurons in vivo. Our results add to an increasing number of reports indicating that, together with transcription factors, RNA-binding proteins of the Elav family are important regulators of cell differentiation (King et al., 1994; Gao and Keene, 1996; Wakamatsu and Weston, 1997; Akamatsu et al., 1999; Antic et al., 1999; Kasashima et al., 1999). Among these, HuB (Hel-N1), HuC, and HuD appear to play a critical role in the induction of neuronal-specific genes during the development of the nervous system.

Ancillary