Neuronal HuD gene encoding a mRNA stability regulator is transcriptionally repressed by thyroid hormone


  • Ana Cuadrado,

    1. Instituto de Investigaciones Biomédicas ‘Alberto Sols’, Consejo Superior de Investigaciones Científicas-Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author
  • Cristina Navarro-Yubero,

    1. Instituto de Investigaciones Biomédicas ‘Alberto Sols’, Consejo Superior de Investigaciones Científicas-Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author
  • Henry Furneaux,

    1. Department of Physiology, University of Connecticut Health Center, Farmington, Connecticut, USA
    Search for more papers by this author
  • Alberto Muñoz

    1. Instituto de Investigaciones Biomédicas ‘Alberto Sols’, Consejo Superior de Investigaciones Científicas-Universidad Autónoma de Madrid, Madrid, Spain
    Search for more papers by this author

Address correspondence and reprint requests to Professor Alberto Muñoz, Instituto de Investigaciones Biomédicas, Arturo Duperier, 4, E-28029 Madrid, Spain. E-mail:


Many genes governed by thyroid hormone (T3) lack binding sites for its receptor (TR) and are thought to be post-transcriptionally regulated by T3. Here we demonstrate that the HuD gene, which encodes a neurone-specific protein that binds to mRNA and modulates its stability, is regulated by T3. HuD RNA and protein expression were strongly up-regulated in specific areas of the hypothyroid rat brain, and reduced by T3 in rat PC12 and mouse N2a cells containing appropriate TR levels. Furthermore, T3 inhibited the transcription of HuD in run-on assays. Finally, HuD protein bound with high affinity to two sequences in acetylcholinesterase mRNA, and ectopic HuD expression increased its abundance in N2a cells. This is the first report of a gene encoding an mRNA stability regulator that is under T3 control. The results suggest that HuD may mediate some T3 effects by altering the half-life of mRNAs for acetylcholinesterase and other genes.

Abbreviations used

bovine serum albumin




glyceraldehyde-3-phosphate dehydrogenase




KH-type splicing regulatory protein


neuroblastoma apoptosis-related RNA-binding protein


nerve growth factor

Nova-1, PTB

polypyrimidine tract binding protein 1


phosphate-buffered saline


rexinoid receptor




thyroid receptor




untranslated region

Thyroid hormone (tri-iodothyronine; T3) regulates brain development and physiology by controlling gene expression (Dussault and Ruel 1987; Anderson and Mariash 2002; Bernal 2002). The classical mechanism of action of T3 is the regulation of gene transcription by binding to specific nuclear receptors (TRα1, TRβ1, TRβ2 and TRβ3) which interact with specific nucleotide sequences (thyroid response elements) (Bernal and Guadaño-Ferraz 1998; Zhang and Lazar 2000). T3 is believed to regulate many genes; for instance, it controls around 11% of those expressed in the liver (Feng et al. 2000; Flores-Morales et al. 2002). However, only 58% of them appear to be direct targets that respond rapidly to T3; many others are late responding genes that may be regulated indirectly (Flores-Morales et al. 2002). In line with this, many well known T3 target genes do not contain thyroid response elements (Bernal and Guadaño-Ferraz 1998). Several studies have reported post-transcriptional regulatory effects of T3 on the mRNA half-life of a number of genes including those for malic enzyme, 3-hydroxy-3-methylglutaryl CoA reductase, thyrotropin β subunit, rexinoid receptor (RXR)γ, acetylcholinesterase, uncoupling protein-1 and Tα1 α-tubulin (Back et al. 1986; Simonet and Ness 1989; Mano et al. 1994; Leedman et al. 1995; Puymirat et al. 1995; Guerra et al. 1996; Lorenzo et al. 2002). T3 has also been implicated in pre-mRNA splicing (Aniello et al. 1991; Gerrelli et al. 1994; Cuadrado et al. 2002a) as well as in translational and post-translational processes (Silva and Rudas 1990; Dace et al. 2000). However, the molecular mechanisms underlying the post-transcriptional gene regulatory actions of T3 effects remain largely unknown.

We have described the regulation by T3 of the splicing regulators suppressor-of-white-apricot (SWAP) (Cuadrado et al. 1999) and musashi-1, the latter of which encodes an RNA-binding protein that affects tau pre-mRNA splicing (Cuadrado et al. 2002a). These findings support the hypothesis that T3 controls post-transcriptional processes through the regulation of genes encoding RNA-binding proteins.

Several neural RNA-binding proteins are known, including Nova-1, PTB (polypyrimidine tract binding protein 1), KSRP (KH-type splicing regulatory protein), NAPOR, (neuroblastoma apoptosis-related RNA-binding protein) and the Musashi (Msi) and Hu/Elav families (Grabowski 1998; Zhang et al. 2002). The Elav (Embryonal lethal abnormal vision) gene is essential for the appropriate development and maintenance of the nervous system in Drosophila (Jiménez and Campos-Ortega 1979; Robinow and White 1988). Four Hu/Elav genes have been identified in mammals: HuR (HuA in rodents), Hel-N1 (HuB in rodents), HuC and HuD. HuC and HuD are specifically expressed in neuronal tissues and neuroendocrine tumours, HuB is expressed in neurones, testes and ovaries, and HuR is ubiquitous (Szabo et al. 1991; Antic and Keene 1997; Wakamatsu and Weston 1997).

The Hu/Elav proteins are pleiotropic regulators of gene expression in mammalian cells (Vakalopoulou et al. 1991; Zhang et al. 1993; Ma et al. 1996; Wilusz et al. 2001). They mainly regulate mRNA stability, although additional effects on RNA transport and translation have also been proposed (Antic and Keene 1998; Antic et al. 1999; Gallouzi and Steitz 2001). Hu/Elav proteins can stabilize or destabilize specific mRNAs perhaps by regulating the access of degrading enzymes or other proteins to RNA substrates (Ford et al. 1999; Brennan and Steitz 2001). HuD has three RNA-binding domains with affinity for the AU-rich elements and the poly(A) tail present in the 3′-untranslated region (UTR) of many mRNAs (Wang and Tanaka Hall 2001; Manohar et al. 2002). HuD binds to mRNAs and regulates the expression of genes involved in CNS differentiation such as tau and GAP-43 (Aranda-Abreu et al. 1999), and also of oncogenes, angiogenic factors, cytokines, cell cycle regulators and transcriptional repressors (King et al. 1994; Chung et al. 1997; Manohar et al. 2002). Overexpression of HuD induces neuronal differentiation of PC12 and P19 cells (Mobarak et al. 2000), probably by modulating the stability of mRNA.

Here we demonstrate that T3 regulates HuD expression in the rat brain and in cultured neuronal cells. Run-on assays showed that T3 reduces the transcription rate of the HuD gene. Our results also show that HuD protein binds with high affinity to acetylcholinesterase mRNA, suggesting that HuD may mediate some of the post-transcriptional regulatory actions of T3.

Experimental procedures


Wistar rats were raised in our animal facilities. The maintenance and handling of the animals were as recommended by the European Communities Council Directive of November 24 1986 (86/609/EEC). Embryonic and neonatal hypothyroidism were induced as described (Alvarez-Dolado et al. 2000). 2-Mercapto-1-methylimidazole (0.02%; Sigma, St Louis, MO, USA) and potassium perchlorate (1%, Merck Darmstadt, Germany) were administered in the drinking water of the dams from the 9th day after conception until the animals were killed. The results were based on the analysis of sections from at least three animals per experimental group. Hypothyroid animals showed the characteristic arrest of bodyweight increase [around 25% on postnatal day 15 (P15), 60% on P25].

In situ hybridization

Under deep pentobarbital anaesthesia, rats were perfused through the heart with cold 4% p-formaldehyde in 0.1 m sodium phosphate (pH 7.4). The brains were quickly removed, post-fixed in 4% p-formaldehyde in 0.1 m sodium phosphate, pH 7.4, and cryoprotected in 4% p-formaldehyde containing 30% sucrose (w/v) in 0.1 m sodium phosphate, pH 7.4, at 4°C. Subsequently, coronal sections 25 µm thick were cut using a cryostat. In situ hybridization on floating sections was performed as described (Cuadrado et al. 1999). Hyperfilm β-MAX films (Amersham Pharmacia, Biotech Europe, Barcelona, Spain), were exposed for 15–21 days, developed with Kodak D19 Kodak Industrie, Chalon s/Sao^ne, France and fixed.


Immunohistochemistry was performed as described previously (Cuadrado et al. 2002b). Sections were incubated with mouse monoclonal 16A11 anti-HuD/HuC antibody (1 : 200; Marusich et al. 1994; donated by Dr Marusich, University of Oregon, and purchased from Molecular Probes Inc., Eugene, OR, USA) overnight at 4°C and then a pre-adsorbed biotinylated secondary rat anti-mouse antibody (1 : 200; Vector Laboratories, Burlingame, CA, USA) for 1 h at room temperature (25°C), followed by immunocomplex detection with the ABC reagent (Elite kit; Vector Laboratories). Anatomical abbreviations follow Paxinos and Watson (1998).

Cell cultures

Rat PC12 phaeochromocytoma cells and their derivatives PC12 + TRα1 and PC12 + v-erbA cells (Muñoz et al. 1993), which express exogenous TRα1 and v-erbA genes respectively, were grown in Dulbecco's modified Eagle's medium supplemented with 10% horse serum, 5% fetal calf serum and 2 mm glutamine (all from Gibco-BRL, Rockville, MD, USA). Mouse N2a + TRβ neuroblastoma cells (Lebel et al. 1994) that express exogenous TRβ1 gene were grown in Dulbecco's modified Eagle's medium–HEPES, pH 7.2, medium supplemented with 10% fetal calf serum and 2 mm glutamine. T3, nerve growth factor (NGF) and cycloheximide (CHX) were purchased from Sigma.

Western blotting

Cell protein extracts were prepared following the Dignam C method (Sambrook et al. 1989). Aliquots were electrophoresed in 12% polyacrylamide gels and transferred to nylon polyvinylidene difluoride membranes. The filters were washed, blocked with 5% skimmed milk in phosphate-buffered saline (PBS) containing 0.1% Tween-20, and incubated overnight at 4°C with the mouse monoclonal 19F12 anti-HuR antibody (1 : 2000 dilution). Blots were washed three times for 15 min in PBS, 0.1% Tween-20 and incubated with horseradish peroxidase-conjugated anti-rabbit antibody for 1 h at room temperature. Blots were developed by a peroxidase reaction using the ECL detection system (Amersham Biosciences).

RNA extraction and northern analysis

To prepare total RNA from PC12 or N2a cells we used standard methods (Sambrook et al. 1989). RNAs were fractionated in formaldehyde agarose gels and blotted on to nylon membranes (Nytran; Schleicher and Schuell, Dassel, Germany). As control for RNA loading, the filters were stained with 0.02% methylene blue in 0.3 m sodium acetate. Radioactive probes were prepared by the random priming procedure using Ready-to-go kit (Amersham Biosciences).

Nuclear run-on transcription assays

We followed the procedure described by López-Carballo et al. (2002). Briefly, cells were lysed at 4°C in 20 mm Tris-HCl, pH 8.0, 0.3 m sucrose, 60 mm KCl, 15 mm NaCl, 2 mm EDTA, 0.5 mm EGTA, 0.5 mm spermidine, 0.15 mm spermine, 0.5 mmβ-mercaptoethanol and 0.1% Nonidet P-40. After washing in cold buffer B [50 mm HEPES-NaOH, pH 8.0, 5 mm MgCl2, 0.5 mm dithiothreitol, 1 µg/mL bovine serum albumin (BSA), 25% glycerol] nuclei were pelleted, resuspended in the same buffer and stored at − 70°C in aliquots of 5 × 106. Nytran N45 membranes containing 1 µg purified cDNA fragments of HuD, β-actin or glyceraldehyde-3-phosphate dehydrogenase (GAPDH), or 5 µg of linearized empty vector pBS SK+ vector were prepared with the aid of a dot blot apparatus (Minifold II; Schleicher and Schuell). Transcription reactions were set up with 50 µL transcription mix (50 mm HEPES-NaOH, pH 8.0, 2 mm MgCl2, 2 mm MnCl2, 1 µg/mL BSA, 300 mm NH4Cl); 5 µL 10 mm CTP, ATP and GTP mix; 100 µCi [α-32P]UTP (Amersham Biosciences; 800 Ci/mmol); 40 U Rnasin (Promega, Madison, WI, USA); and 50 µL of nuclei (5 × 106) in buffer B. After incubation for 20 min at 30°C, nuclear DNA was digested with 10 U RQ1 Dnase (Promega). Purification of labelled RNA and hybridization of membranes were as described (López-Carballo et al. 2002).

Preparation of labelled RNA transcripts

DNA templates for acetylcholinesterase transcripts were synthesized by the PCR using the following oligonucleotides. For transcript A′: 5′ GTAATATACGACTCACTATAGGGCCCTGCATACACCTTCCC 3′ and 5′ TCCAGTATTGATGAGAGC 3′. For transcript B: 5′ GTAATATACGACTCACTATAGGGCCTGGTGGGTGTGGTGAAGG 3′ and 5′ AACAGTTTATTGGCGGCC 3′. For transcript A′: 5′ GTAATATACGACTCACTATAGGGCCCTGCATACACCTTCCC 3′ and 5′ CCATAGATCCAGATGAGG 3′. For transcript B′: 5′ GTAATATACGACTCACTATAGGGCTCTTAATGTGTGGACACC 3′ and 5′ CCAGCACTCACAGTGGCC 3′. For transcript C′: 5′ GTAATATACGACTCACTATAGGGCGGCTGTCCTCCAGAGTGG 3′ and 5′ TCCAGTATTGATGAGAGC 3′. For transcript D′: 5′ GTAATATACGACTCACTATAGGGCCTGGTGGGTGTGGTGAAGG 3′ and 5′ CCAAGTCAGTGTGGAGGC 3′. For transcript E′: 5′ GTAATATACGACTCACTATAGGGCCCTACATCTTTGAACACC 3′ and 5′ CAGAAGGCGCAGGTCTGG 3′. For transcript F′: 5′ GTAATATACGACTCACTATAGGGCAAGCCCTTAGAGGTGCGG 3′ and 5′ AACAGTTTATTGGCGGCC 3′. The c-fos 3′-UTR sequence used was that described by Chung et al. (1996). For the unrelated transcript corresponding to the 3′-UTR of tenascin-C: 5′ GTAATATACGACTCACTATAGGGCCACGCTCAACTGGACTGC and 5′ GGCTGTTGTTGCTATGGC 3′. All templates were gel-purified. RNA transcripts were synthesized using T7 RNA polymerase (Promega) and purified as described (Ma et al. 1996).

Purification of glutathione-S-transferase (GST)-HuD protein

An overnight culture of Escherichia coli BL 21 transformed with pGST-HuD plasmid (Ma et al. 1996) was diluted 1 : 50 in Luria broth (LB) medium. At an optical density at 600 of 0.4, the culture was induced with isopropyl-β-d-thiogalactopyranoside (0.1 mm) at 30°C. Four hours later, the cells were spun down and resuspended in 10 mL buffer A (50 mm Tris, pH 8.0, 50 mm NaCl, 1 mm EDTA). The cells were lysed by the addition of lysozyme (0.2 mg/mL) and Triton X-100 (1%). The lysate was centrifuged at 12 000 g for 30 min, and the resulting supernatant was collected and passed through 19- and 23-G needles several times. It was then incubated with GST-Sepharose beads for 1 h at 4°C, centrifuged at 720 g at 4°C, and washed five times in PBS. The purified GST protein was eluted with 10 mm glutathione in 50 mm Tris-HCl, pH 8.0, dialysed overnight in PBS containing 10% glycerol, and then pooled and stored at − 70°C. GST protein concentration was determined by comparison with a standard BSA curve in an acrylamide gel stained with Coomassie Brilliant Blue.

Agarose RNA gel-shift assay

Reaction mixtures (20 µL) contained 50 mm Tris, pH 7.0, 0.25 mg/mL tRNA, 0.25 mg/mL BSA, 20 fmol labelled RNA and protein as indicated. Mixtures were incubated at 37°C for 10 min. Following incubation, 4 µL dye mixture (50% glycerol, 0.1% bromophenol blue, 0.1% xylene cyanol) was added, and 25% of the reaction mixture was immediately loaded on a 1% agarose gel in TAE buffer pH 8.0 (40 mm Tris acetate, 1 mm EDTA). Gels were electrophoresed at 40 V for 3 h and then dried on DE81 paper (Whatman, Maidstone, UK) with a backing of gel drying paper (Hudson City Paper, West Caldwell, NJ, USA). XAR5 films (Kodak Industrie) were exposed for 4–6 h at − 70°C.

Rnase T1 selection assay

Reaction mixtures (20 µL) contained 50 mm Tris, pH 7.0, 0.25 mg/mL tRNA, 0.25 mg/mL BSA, 10–20 fmol labelled RNA (100 000–600 000 cpm) and purified GST-HuD (75 ng). After 10 min incubation at 37°C, Rnase T1 (5 U; Calbiochem, La Jolla, CA, USA) was added to each reaction and incubated at 37°C for a further 10 min. The mixtures were diluted 1 : 6 with buffer FBB (20 mm Tris-HCl, pH 7.0, 0.05 mg/mL tRNA) and filtered through nitrocellulose (BA 85; Schleicher and Schuell). After washing the nitrocellulose twice in FBB, bound HuD–RNA complex was extracted with phenol/chloroform and concentrated by ethanol precipitation. The resulting RNA was dissolved in formamide stop buffer (Gibco-BRL) and denatured at 65°C for 2 min. Samples were analysed by 10% polyacrylamide/50% urea gel electrophoresis. The gel was fixed with 1 : 1 : 8 acetic acid : methanol : water and dried. XAR5 films were exposed overnight at − 70°C.

Ectopic HuD expression

Four 10-cm dishes of N2a cells were independently tranfected with 2 µg pCEFL-AU5HuD plasmid. As control, cells were transfected with empty pCEFL-AU5 vector. Upon selection with 500 µg/mL G418, tranfected cells were screened for tagged HuD protein expression by western blotting using an anti-AU5 antibody (Babco, Berkeley, CA, USA; 1 : 1000). As control, we measured β-actin using an appropriate antibody (sc-1615, 1 : 1000; Santa Cruz Biotechnology, Santa Cruz, CA, USA).


HuD expression is up-regulated by hypothyroidism in the rat brain

The observation that T3 affects the half-life of mRNAs and the HuD protein controls the stability of many important mRNAs prompted us to test whether T3 regulates the HuD gene in the developing rat brain. By in situ hybridization we detected maximal HuD expression at P5, as reported before (Okano and Darnell 1997) (Fig. 1). At P5–P20, thyroid deficiency caused an up-regulation of HuD RNA levels in the retrosplenial cortex and layer V in the cerebral cortex, but not in other brain areas where HuD expression is also high, such as in layers II–IV and VIb or the piriform cortex (Fig. 1). To examine whether hypothyroidism also altered expression of the HuD protein we performed immunohistochemical studies with an antiserum (16A11) that detects HuD and HuC in rat brain (Wakamatsu and Weston 1997). In agreement with the RNA data, HuD protein expression was higher at P20 in hypothyroid animals than in controls in cortex layer V and retrosplenial cortex (Fig. 2a), two areas in which HuD is the only expressed member of the HuD/Elav family (Okano and Darnell 1997). The strongest staining was found in layer V pyramidal neurones, mainly in the cell bodies (Fig. 2b). No differences were found in other regions of high HuD expression such as the piriform cortex or thalamus.

Figure 1.

Expression of HuD RNA is up-regulated by hypothyroidism in the developing rat brain. Analysis by in situ hybridization of HuD RNA levels in coronal sections of brains of control (a, c, e, g, i) and hypothyroid (b, d, f, h, j) animals. At P5, higher HuD RNA levels were found in the cortex layer V (V) and retrosplenial cortex (RSCx) of hypothyroid animals (arrowheads). At P15, the differences in layer V persisted (arrowheads). At P20, overall HuD expression decreased throughout the brain but hypothyroidism still caused abnormally high levels in the RSCx and cortex layer V (arrowheads). Scale bar (in a) 1.0 mm for (a), (b), (g) and (h); 0.5 mm for (c), (d), (e), (f), (i) and (j). Pir, piriform cortex.

Figure 2.

Hypothyroidism causes abnormal HuD protein expression in pyramidal neurones of cortex layer V. (a) Analysis by immunohistochemistry of HuD levels in coronal sections of brains of control and hypothyroid P20 rats. Pir, piriform cortex; Th, thalamus. Scale bar 750 µm. (b) Upper panels: low magnification (20 ×) of cerebral sections of cortex layers IV–VI from control and hypothyroid animals showing intensely stained layer V cells; lower panels: high magnification (40 ×) of layer V pyramidal neurones (arrows).

T3 inhibits HuD expression in PC12 and N2a cells

To examine the effect of T3 on the HuD gene we first used rat phaeochromocytoma PC12 cells. T3 induced a threefold to fourfold decrease in the expression of both HuD RNAs (4.2 and 3.7 kb) in cells containing appropriate TR levels (PC12 + TRα1) (Fig. 3a), but not in parental PC12 cells that contain very low TR levels or in cells expressing the v-erbA oncogene, which encodes a non-hormone-binding variant of TRα1 (PC12 + v-erbA) (Muñoz et al. 1993). T3 repressed HuD RNA expression at physiological concentrations (Fig. 3b). The inhibitory action of T3 on HuD RNA expression was detected 2 h after addition and was maximal at 2 days (Fig. 4a). The lack of suitable specific anti-HuD antibodies precluded a western blotting study. The effect of T3 on HuD was specific, as the expression of HuR, another member of the Elav/Hu family, was unaffected (Fig. 4b).

Figure 3.

T3 inhibits HuD RNA expression in PC12 cells. (a) Northern analysis of the effect of T3 on HuD RNA levels in parental PC12, PC12 + TRα1 and PC12 + v-erbA cells. Cells were treated or not with 150 nm T3 for 2 days as indicated. Some 20 μg total RNA was loaded per lane. Sizes of the two HuD mRNAs are indicated. (b) Dose curve of HuD RNA inhibition in PC12 + TRα1 cells by T3. Cells were treated for 48 h with the indicated T3 concentrations.

Figure 4.

Time course, specificity and biological relevance of the inhibition of HuD expression by T3. (a) Northern analysis of the kinetics of HuD RNA inhibition by T3. PC12 + TRα1 cells were treated or not with T3 (150 nm) for the indicated periods. Some 20 μg total RNA was loaded per lane. Sizes of the two HuD mRNAs are indicated. Right panel shows the quantification of 3.7-kb HuD RNA levels. Mean ± SD values obtained in three experiments are shown. (b) T3 does not change HuR expression. Western analysis of HuR expression in PC12 + TRα1 cells treated or not with T3 (150 nm) for the indicated times. (c) T3 overrides the effect of NGF on HuD expression. Northern analysis showing that the induction of HuD RNA by NGF is reverted by concomitant treatment with T3. PC12 + TRα1 cells were treated with NGF (50 ng/mL), T3 (150 nm), or their combination for the indicated times. Conditions were as above. (d) T3 inhibits HuD RNA expression in N2a + TRβ cells. Northern analysis of 3.7-kb HuD RNA levels in cells treated for the indicated times with 150 nm T3. Some 20 μg total RNA was loaded per lane. The percentage inhibition with respect to the level in control untreated cells is shown.

The effects of NGF on HuD expression are controversial (Steller et al. 1996; Dobashi et al. 1998). We consistently found that NGF increased (by about 70%) the levels of HuD RNA in PC12 + TRα1 cells (Fig. 4c). Combined treatment with NGF and T3 reduced HuD RNA expression to a level below that of untreated cells (Fig. 4c), thus emphasizing the functional relevance of T3 action and its dominance over NGF.

To consolidate this finding and to test whether the effect of T3 is receptor isoform specific we studied the action of T3 on the HuD gene in mouse neuroblastoma N2a cells expressing TRβ (N2a + TRβ) (Lebel et al. 1994). T3 also inhibited HuD RNA expression in these cells (Fig. 4d), indicating that both TRα1 and TRβ mediate the effect of T3 on HuD.

HuD gene transcription is repressed by T3

To examine the mechanism underlying the inhibitory action of T3 on HuD expression, run-on assays were done. T3 inhibited (60%) the rate of transcription of the HuD gene as early as 4 h after addition (Fig. 5a). The same result was obtained at 8 h of treatment (not shown). In contrast, no effect was seen on genes encoding GAPDH or β-actin which were used as a control. Empty pBS SK+ vector was included as a negative control. Next, we studied whether the repressive effect of T3 on HuD required protein synthesis de novo. To this end, the effect of CHX, an inhibitor of translation, was determined. In northern blot analysis, CHX did not block the inhibitory action of T3 (Fig. 5b). These results confirm that T3 has a direct regulatory effect on HuD gene transcription.

Figure 5.

T3 represses HuD gene transcription. (a) Run-on assay showing that T3 reduces the transcription rate of HuD. Nuclei from PC12 + TRα1 cells treated or not with T3 (150 nm) for 4 or 8 h (not shown) were subjected to an in vitro transcription reaction. As controls we used β-actin and GAPDH genes, and also the empty pBS SK+ vector. Right panel shows the quantification of HuD and β-actin transcription normalized to that of GAPDH. (b) Northern blot showing that CHX does not block the inhibition of HuD RNA level by T3 in PC12 + TRα1 cells. Cells were pretreated with CHX (8 µg/mL) 30 min before addition of T3 (150 nm). Single treatments with either T3 or CHX were done as controls. RNAs were obtained 8 h after T3 addition. The 3.7-kb HuD transcript is shown.

HuD protein binds with high affinity to two regions of acetylcholinesterase RNA

To investigate the biological significance of the regulation of HuD gene by T3 we studied whether HuD protein could bind to acetylcholinesterase RNA, whose stability is regulated by T3 in N2a + TRβ cells (Puymirat et al. 1995). Two binding sites were found in T1 Rnase digestion assays, one 35 nucleotides long in the 5′-half of the coding sequence and the other 21 nucleotides long in a region abutting the end of the coding sequence and the 3′-UTR of the RNA (Fig. 6a). These results were confirmed by agarose gel-shift assays using purified GST-HuD fusion protein. Incubation of in vitro-labelled transcripts A and B with GST-HuD, but not with GST, led to the detection of protein–RNA complexes (Fig. 6b). The c-fos 3′-UTR (Ma et al. 1996) and an unrelated sequence from the tenascin-C mRNA were used as positive and negative controls respectively.

Figure 6.

HuD binds to acetylcholinesterase mRNA. (a) Rnase T1 analysis of HuD–acetylcholinesterase mRNA binding in vitro. The structure of acetylcholinesterase mRNAs is shown at the top. GST-HuD or GST proteins (200 nm) were incubated with 32P-labelled transcripts A or B (10 fmol, 200 000–400 000 cpm) of acetylcholinesterase. Total T1 digests of both transcripts (T1A, T1B) are shown. HuD protein binds to two sites in the acetylcholinesterase mRNA. Arrows indicate HuD-bound RNA fragments. (b) Agarose gel-shift assays showing binding of HuD to acetylcholinesterase A and B transcripts. The c-fos 3′-UTR transcript and a RNA of unrelated sequence were used as positive and negative controls respectively. 32P-labelled transcripts (20 fmol, 400 000–600 000 cpm) were incubated with the indicated concentration of GST or GST-HuD protein. After 10 min incubation mixtures were resolved on 1% agarose gels. Retarded bands are indicated.

To identify the sequences bound by HuD in acetylcholinesterase RNA, T1 Rnase selection assays were done using six partially overlapping transcripts (A′–F′) covering the entire RNA (Fig. 7a). HuD–acetylcholinesterase RNA complexes were allowed to form and then subjected to digestion with T1 Rnase. RNA fragments bound to HuD were isolated by adsorption of complexes to nitrocellulose followed by elution with phenol–chloroform. In agreement with results of the gel-shift assays, a single 35-nucleotide fragment of transcript A′ and another 21-nucleotide fragment of transcript F′ were detected (Fig. 7b). Likewise, no binding was found when GST protein was used as control. Taking into account the sizes of the fragments and the sequence specificity of T1 Rnase (cleaves after a G), their sequences were readily located in the acetylcholinesterase RNA (Fig. 7c; underlined). The 35-nucleotide fragment in transcript A′ is near the initiator AUG triplet and contains a CUnC sequence. The 21-nucleotide fragment in transcript F′ is an AU-rich sequence located in the 3′-UTR that shows a high level of homology with other HuD-binding sequences (Chen and Shyu 1995; Chung et al. 1996; Ma et al. 1997; Manohar et al. 2002).

Figure 7.

Localization of HuD binding sites in acetylcholinesterase mRNA. (a) Scheme of acetylcholinesterase mRNA and the six transcripts (A′–F′) used. (b) HuD binds to one site in the coding sequence (transcript A′) and another site in the 3′-UTR (transcript F′) of acetylcholinesterase mRNA. Rnase T1 analysis of HuD–acetylcholinesterase RNA complexes. GST-HuD or GST proteins (75 ng) were incubated with 32P-labelled acetylcholinesterase transcripts (20 fmol, 400 000–600 000 cpm) at 37°C for 10 min. Rnase T1 (5 U) was then added to reaction mixtures before they were filtered through nitrocellulose. Protein-bound RNA fragments were extracted and resolved on 12% denaturing polyacrylamide gels. Rnase T1-undigested fragments are indicated. T1 digests of each transcript are shown (T1A′, T1B′, T1C′, T1D′, T1E′, T1F′). (c) Sequence of the acetylcholinesterase mRNA showing the precise location of the HuD-binding sites (underlined). Initiator AUG and stop UGA codons are shown in bold.

Ectopic HuD expression regulates the abundance of acetylcholinesterase RNA

To study whether HuD modulates the abundance of acetylcholinesterase RNA we stably expressed an ectopic AU5-HuD protein in N2a cells. Four independent pools of transfected cells were obtained upon appropriate antibiotic selection. Western analysis using an anti-AU5 antibody demonstrated the expression of exogenous tagged HuD protein (Fig. 8a). Northern analysis showed that the cellular content of acetylcholinesterase RNA in all four HuD-expressing pools was higher (2–3.5-fold) than in cells transfected with empty vector and correlated with the level of AU5-HuD protein (Fig. 8b).

Figure 8.

Stably ectopic expression of HuD increases acetylcholinesterase mRNA abundance. (a) Western analysis (WB) of exogenous tagged HuD protein expression in N2a cells. Protein extracts (40 µg) of four G418-selected pools of cells transfected independently with an HuD expression vector (pCEFL-AU5HuD) and of cells transfected with the empty vector (pCEFL-AU5) were analysed using an anti-AU5 antibody. Expression of β-actin protein was also analysed as control. (b) Northern analysis (NB) of acetylcholinesterase (AchE) mRNA abundance in cells expressing HuD ectopically. Some 10 μg poly(A)+ RNA was loaded per lane. The 18S ribosomal RNA was stained with methylene blue. The increase in acetylcholinesterase mRNA content estimated upon normalization with 18S rRNA level is shown.


Our data show that hypothyrodism induces an increase in the expression of HuD RNA and protein in certain areas of the rat brain. This effect occurs during the early postnatal period, when HuD expression (Okano and Darnell 1997) and the number and occupancy of TRs (Ferreiro et al. 1990) are maximal. Furthermore, we show that expression of the HuD gene is under direct T3 control in cultured neuronal cells. This is the first demonstration of a direct regulation by T3 of a gene involved in the control of mRNA stability, and perhaps transport and translation capacity.

Our data indicate that HuD is under direct T3 control. Since HuD regulates the expression of many genes by post-transcriptional mechanisms, this amplifies the range of the T3 gene regulatory network. Other genes encoding RNA-binding proteins such as msi-1 and SWAP that regulate splicing (Cuadrado et al. 1999, 2002a) or NAT-1 that encodes a translational repressor (Shah et al. 1998) are also modulated by T3.

The developing CNS is rich in post-transcriptional regulatory processes, in particular the regulation of the half-life of mRNAs (Grabowski 1998). Regulation of mRNA stability is currently considered to be a major control point in gene expression (Guhaniyogi and Brewer 2001; Fan et al. 2002). Several cis-acting elements play a role in mRNA stability. Best characterized among them are the AU-rich element and the poly(A) tail located in the 3′-UTR of many mRNAs (Chen and Shyu 1995; Wilusz et al. 2001). Other stability determinants have been identified within the coding regions or the 5′-UTRs of some genes, but the proteins binding to them are mostly unknown (Wellington et al. 1993; Yeilding and Lee 1997). The variety of HuD RNA targets includes several that code for proteins with distinct functions of which two, GAP-43 and Tau, are neurone specific and have roles in neuronal differentiation and function (Chung et al. 1996; Aranda-Abreu et al. 1999; Mobarak et al. 2000). Emphasizing the biological significance of our results, we have characterized acetylcholinesterase RNA as a new target of HuD. Acetylcholinesterase is involved in synaptic transmission and, importantly, is regulated at the RNA stability level by T3 in the same N2a cells (Puymirat et al. 1995) as those in which we found HuD expression to be under T3 control.

The complex pattern of expression of the various Hu/Elav proteins in the rat brain (Okano and Darnell 1997) suggests a region-specific regulation of their target RNAs. Given the wide range of RNA substrates known for Hu proteins, in areas such as cortex layer V and retrosplenial cortex where it is the only member of the Hu/Elav family expressed, HuD may mediate post-transcriptional regulatory actions of T3. The partially overlapping patterns of expression of hu/Elav genes and of their substrates, and the lack of appropriate specific antibodies make it difficult to ascertain whether other members are under thyroid control in vivo, an interesting possibility that remains unexplored. In PC12 cells, however, HuR expression is unchanged by T3.

HuD is involved in the initiation of neurites and early neuronal differentiation of PC12 cells (Dobashi et al. 1998; Anderson et al. 2000). We have reproduced the increase in HuD expression by NGF described in some studies (Dobashi et al. 1998) but not in others (Steller et al. 1996), which also links HuD to differentiation in this cell type. T3, an inducer of neuronal differentiation (Dussault and Ruel 1987; Anderson and Mariash 2002), however, inhibits HuD expression. Gene repression is a common control mechanism during development, and it is also a major action of T3 in brain and liver (Anderson and Mariash 2002; Flores-Morales et al. 2002). Moreover, neuronal differentiation is a complex process that results from the coordinated function of many proteins. Furthermore, hypothyroidism alters the expression of multiple genes which impairs neuronal differentiation. The effect of HuD may be different in cultured PC12 cells and in cortex layer V neurones of hypothyroid animals as a result of the interaction with other proteins or differences in localization.

Two HuD-binding sites have been characterized in acetylcholinesterase mRNA. Remarkably, one of the binding sites of HuD in acetylcholinesterase RNA is a consensus AU-rich sequence whereas the other is a CU-rich sequence located quite 5′ in the coding sequence, around 40 nucleotides downstream of the initiator AUG. Elav/Hu proteins bind preferentially to AU-rich sequences in the 3′-UTR, although examples of CU-rich sequences bound by HuD have also been described (Joseph et al. 1998; references therein). Furthermore, RNA sequences at the 5′-UTR are postulated to be involved in RNA translation control (Stripecke et al. 1994; Millard et al. 2000), and CU-rich sequences are targets of other RNA-binding proteins such as PTB (Grabowski 1998). The role of this sequence might therefore be related to regulation of stability, splicing, export or translation of acetylcholinesterase RNA by HuD, perhaps in cooperation with other proteins. We also demonstrated that stable expression of exogenous HuD protein regulates the abundance of acetylcholinesterase mRNA in N2a cells. This result agrees with a recent report that showed a similar effect of HuD in PC12 cells (Deschenes-Furry et al. 2003). T3 has been noted to increase acetylcholinesterase mRNA stability in N2a cells (Puymirat et al. 1995). Our data show that T3 inhibits HuD expression, and that this protein can augment acetylcholinesterase mRNA abundance. Several protein complexes bind acetylcholinesterase RNA in differentiated neurones (Deschenes-Furry et al. 2003). Therefore, it can be concluded that HuD contributes but is not the unique factor controlling stability of this transcript, and that T3 action at this level must involve other proteins.

In summary, our data demonstrate that T3 represses HuD gene transcription in cultured neurones, which may explain the up-regulation of HuD RNA and protein levels in certain areas of the hypothyroid rat brain. In addition, we show that acetylcholinesterase RNA is a new substrate of HuD protein at least in vitro. Our data indicate that some of the actions of T3 are mediated by genes involved in post-transcriptional mechanisms, such as HuD.


We thank Drs M. F. Marusich and J. Puymirat for providing the Mab16A11 the N2a + TRβ cells respectively, and Professor B. Vennström for critical reading of the manuscript. We are also grateful to Dr D. Barettino for his help with the run-on assays and to Robin Rycroft for his valuable assistance in the preparation of the English manuscript. CN-Y was supported by a predoctoral fellowship from Fondo de Investigaciones Sanitarias of Ministerio de Sanidad y Consumo of Spain. This work was supported by Grant SAF2001-2291 from Ministerio de Ciencia y Tecnología of Spain.