Fragile X syndrome, the most common heritable form of mental retardation, is caused by silencing of the FMR1 (fragile X mental retardation-1 gene). The protein product of this gene, FMRP (fragile X mental retardation protein), is thought to be involved in the translational regulation of mRNAs important for learning and memory. In mammals, there are two homologues of FMRP, namely FXR1P (fragile X-related protein 1) and FXR2P. Disruption of Fxr2 in mice produces learning and memory deficits, and Fmr1 and Fxr2 double-knockout mice have exaggerated impairments in certain neurobehavioral phenotypes relative to the single gene knockouts. This has led to the suggestion that FMR1 and FXR2 functionally overlap and that increasing the expression of FXR2P may ameliorate the symptoms of an FMRP deficiency. Interestingly, the region upstream of the FXR2 translation start site acts as a bidirectional promoter in rodents, driving transcription of an alternative transcript encoding the ABP (androgen-binding protein) [aABP (alternative ABP promoter)]. To understand the regulation of the human FXR2 gene, we cloned the evolutionarily conserved region upstream of the FXR2 translation start site and showed that it also has bidirectional promoter activity in both neuronal and muscle cells as evidenced by luciferase reporter assay studies. Alignment of the human, mouse, rat, rabbit and dog promoters reveals several highly conserved transcription factor-binding sites. Gel electrophoretic mobility-shift assays, chromatin immunoprecipitation studies and co-transfection experiments with plasmids expressing these transcription factors or dominant-negative versions of these factors showed that NF-YA (nuclear transcription factor Yα), AP2 (activator protein 2), Nrf1 (nuclear respiratory factor/α-Pal) and Sp1 (specificity protein 1) all bind to the FXR2 promoter both in vitro and in vivo and positively regulate the FXR2 promoter.
- chromatin immunoprecipitation
- fragile X mental retardation-1 gene (FMR1)
- fragile X-related gene 2 (FXR2)
- nuclear factor Y
- promoter regulation
- transcription factor
FXS (fragile X syndrome), the most common heritable form of mental retardation and the most common known cause of autism, belongs to a group of neurological diseases caused by expansion of a specific tandem repeat tract . Expansion of a CGG·CCG repeat tract in the 5′-UTR (5′-untranslated region) of the FMR1 (fragile X mental retardation-1 gene) gene to >200 repeats leads to promoter heterochromatinization [2,3]. This, in combination with difficulties in the translation of any residual mRNA produced from the expanded allele, leads to reduced levels of the FMR1 product, FMRP (fragile X mental retardation protein) . FMRP is thought to be involved in normal synaptic function and thus is obviously crucial for normal brain function. However, in addition to symptoms related to the absence of FMRP in the CNS (central nervous system), individuals with FXS also show symptoms like macroorchidism, digestive difficulties and hypotonia that point to roles for FMRP in places outside of the CNS like testes, gut and muscle.
Two autosomal homologues of FMR1, FXR1 (fragile X-related gene 1) and FXR2, have been identified . FMRP shares 86 and 70% amino acid identity with the N-terminal regions of FXR1P (Fragile X-related protein 1) and FXR2P respectively . Like FMRP, FXR1P and FXR2P are ubiquitously expressed during embryonic development and their tissue distribution overlaps with FMRP in adults . They form homo- and hetero-multimers with one another . The Fxr2 knockout mice share some behavioural similarities with Fmr1 knockout mice such as hyperactivity and impaired motor co-ordination and learning . These neurobehavioral deficits are exaggerated in Fxr2/Fmr1 double-knockout mice, indicating a co-operative contribution of Fmr1 and Fxr2 genes to these behaviours . This has led to the suggestion that FMRP and FXR2P have overlapping functions and that FXR2 may thus be capable of compensating, at least partially, for the loss of FMR1 function in FXS. This is of interest because the poor translatability of mRNAs from FXS alleles limits approaches aimed at re-activating the FMR1 gene [4,9,10]. The FXR2 gene is also interesting in its own right, since its disruption in mice produces learning and memory impairments distinct from those seen in mice lacking FMRP .
While many factors important for the regulation of the FMR1 gene have been identified [11–16], nothing is known about the regulation of the FXR2 gene. Despite the conservation in their coding sequences and the similarity in their pattern of tissue expression, the promoters of these genes are not similar at the sequence level. Substantial sequence similarity exists in the non-coding regions of the FXR2 genes in human and mouse  and this region overlaps with the aABP [alternative ABP (androgen-binding protein) promoter] that is used in rodents to transcribe the oppositely oriented ABP gene . To characterize the FXR2 promoter more fully, we compared the sequence of the human promoter with that of mouse, rat, rabbit and dog and identified a number of evolutionarily conserved regions. We demonstrated that the region upstream of the FXR2 translation start site in humans also has bidirectional promoter activity in cells derived from both brain and muscle. We identified the major transcription factors that bind to the human FXR2 promoter in vitro by gel shift assays and confirmed their in vivo association by ChIP (chromatin immunoprecipitation) assay. The contribution of these transcription factors to promoter activity was examined using co-transfections of reporter constructs with plasmids expressing wild-type and dominant-negative versions of these transcription factors.
MATERIALS AND METHODS
Chemicals and reagents
[α-32P]dCTP (6000 Ci/mmol) and [γ-32P]GTP (6000 Ci/mmol) were purchased from MP Biomedicals (Irvine, CA, U.S.A.). The reporter gene vectors (pGL3-Basic, pGL3-Control and pRL-null) and the Dual Luciferase system were purchased from Promega (Madison, WI, U.S.A.). The transfection reagents FuGENE® 6 was purchased from Roche (Indianapolis, IN, U.S.A.) and Lipofectamine™ 2000 and Cellfectin were purchased from Invitrogen (Carlsbad, CA, U.S.A.). The plasmid constructs Δ4NF-YA13m29, dn-Nrf1, pSAP2B and pcDNA6-AP2DBD were generously provided by Dr R. Mantovani (Dipartimento di Scienze Biomolecolari e Biotechnologie, Universita degli Studi di Milano, Milano, Italy), Dr R. Scarpulla (Department of Cell and Molecular Biology, Northwestern Medical School, Chicago, IL, U.S.A.), Dr M. Tainsky (Program in Molecular Biology and Genetics, Karmanos Cancer Institute and Wayne State University, Detroit, MI, U.S.A.) and Dr P. Kannan (MetroHealth Medical Center, Case Western Reserve University, Cleveland, OH, U.S.A.) respectively. Plasmids pPacSp4, pPacSp3 and pPacUSp3 that express Sp4, and the short and long isoforms of Sp3 respectively, were a gift from Professor G. Suske (Institut für Molekularbiologie und Tumorforschung, Philipps-Universität Marburg, Marburg, Germany). Plasmids pPac-Sp1 and pPac-Sp1DBD were a gift from Professor T. Johansen (Department of Biochemistry, University of Tromsø, Tromsø, Norway). The anti-Nrf1 (nuclear respiratory factor/α-Pal) antibody was a gift from Dr R. Scarpulla. The anti-AP2 (activator protein 2), anti-NF-YA (nuclear transcription factor Yα) and anti-NF-YB antibodies and the rabbit and goat pre-immune sera were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.). Anti-Sp1 (specificity protein 1) antibody was obtained from Upstate (Charlottesville, VA, U.S.A.) and Sp1 protein was purchased from Promega. Synthetic oligonucleotides were obtained from IDT (Coralville, IA, U.S.A.) and sequencing services were provided by MWG-Biotech (Highpoint, NC, U.S.A.). All other chemicals were purchased from Sigma (St. Louis, MO, U.S.A.) unless specifically noted otherwise.
Cloning of the human and rabbit FXR2/aABP promoter and construction of different plasmids
The human FXR2 proximal promoter spanning nt −898 to +214 relative to the translation start site was PCR-amplified from human genomic DNA (negative strand of Chr17: 7458362-7459473, human genome May 2004 assembly) using forward primer 1 and reverse primer 2 (Table 1) (http://genome.ucsc.edu/) [19,20]. The PCR product was digested with SphI and PstI to generate a 706 bp fragment that spanned −742 to −37 bp with respect to the translation start site and was subcloned into the promoterless pGL3-Basic vector to generate pFXR2. Several 5′ and 3′ deletion fragments of the FXR2/aABP promoter were obtained by restriction digestion and religation. The two CCAAT-box mutants and the constructs containing deletions of the CGG tracts were generated by site-directed mutagenesis. These constructs were verified by DNA sequencing. The construction of pFMR1 (aka pGL-FMR) has been described earlier .
The rabbit FXR2 promoter region was PCR-amplified using the forward primer 3 and reverse primer 4 (Table 1) and cloned (GenBank® accession no. DQ322245) into the promoterless pGL3-basic vector. The FXR2 promoter sequences for mouse (positive strand of Chr11: 69358267-69358940, mouse genome March 2005 assembly), rat (positive strand of Chr10: 5646328-56463961, rat genome June 2003 assembly) and dog (negative strand of chr5: 35496411-35497106, dog genome July 2004 assembly) were obtained from the public domain databases (http://genome.ucsc.edu/) [19,20]. The sequences were aligned using MultAlign software . The transcription factor-binding sites were analysed using MatInspector , rVista  and TESS  software.
Transient transfections and reporter gene assays
Mouse C2C12 and human SK-N-MC cells were seeded in 24-well plates containing DMEM (Dulbecco's modified Eagle's medium; Invitrogen Life Technologies, Carlsbad, CA, U.S.A.) supplemented with 10% (v/v) fetal bovine serum. The initial plating density was 0.5×105 and 2×105 cells/well respectively. Co-transfections were done with various test constructs together with pRL-null, a plasmid expressing the Renilla luciferase gene to normalize for differences in the transfection efficiency. Transfections were carried out using FuGENE™ 6 in C2C12 cells and using Lipofectamine™ 2000 in SK-N-MC cells according to the supplier's instructions. Drosophila S2 cells were plated at a density of 4×105/well in 400 μl of serum-free Schneider medium (Invitrogen) and were grown at room temperature (20 °C) overnight in 24-well plates. Cells were transfected using Cellfectin according to the manufacturer's instructions.
Luciferase activity was measured 36 h post-transfection using the Dual Luciferase Reporter assay system and a Microlumat LB 96 P luminometer (Berthold Systems, Aliquippa, PA, U.S.A.). The luciferase activity from the reporter construct in each transfection was normalized to pRL-null.
EMSAs (electrophoretic mobility-shift assays)
Nuclear extracts from the human neuronal SK-N-MC cell line were prepared using ‘NE-PER nuclear and cytosolic extraction reagents’ following the protocol provided by the manufacturer (Pierce, Rockford, IL, U.S.A.). The pFXR2 plasmid was digested with SstI and ApaI generating a 343 bp DNA fragment, fragment I (Figure 3A). This plasmid was also digested with ApaI and HindIII to generate a 367 bp DNA fragment, fragment II (Figure 3A). Fragment I and fragment II were labelled using [α-32P]dCTP and Klenow fragment. Fragment III spanning −690 to −621 bp was PCR-amplified using pFXR2 as template together with forward primer 5 and reverse primer 6 (Table 1; Figure 3A). The fragment IV spanning −655 to −596 bp was PCR-amplified using pFXR2 as template with forward primer 7 and reverse primer 8 (Table 1; Figure 3A). Fragments III and IV were labelled using [γ-32P]GTP and T4 polynucleotide kinase and purified using MicroSpin G-25 columns (Amersham Biosciences, Piscataway, NJ, U.S.A.). Binding was carried out at 20 °C in 30 μl of reaction buffer containing 25 mM Hepes (pH 7.5), 100 mM NaCl, 1 mM EDTA and 4 mM dithiothreitol with 6 fmol of labelled probe, 2 μg of protein and 1 μg of poly(dA-dT)·(dA-dT) in the presence of 1000-fold excess of specific and non-specific competitors. For antibody supershift assays, 10 μg of antibody or normal rabbit/goat pre-immune serum was used in the reactions. The purified Sp1 protein binding reactions were carried out as per the supplier's instructions. Reactions were subjected to electrophoresis on 4 or 5% polyacrylamide gels (60:1, acrylamide/bisacrylamide) containing 1.6% (v/v) glycerol in 1× Tris/glycine/EDTA buffer (50 mM Tris, 380 mM glycine and 2.1 mM EDTA, pH 8.5).
The Upstate Biotechnology (Charlottesville, VA, U.S.A.) ChIP assay kit was used according to the manufacturer's instructions with slight modifications. Briefly, 2×106 cells were cross-linked in 200 μl of DMEM with 1% formaldehyde at 37 °C for 5 min. The cells were washed in PBS and lysed in 200 μl of a buffer containing 1% SDS, 10 mM EDTA and 50 mM Tris/HCl (pH 8.0). The chromatin from lysed cells was sonicated to lengths between 200 and 1000 bp and the cell debris was removed by centrifugation. The sonicated lysates were diluted to 2 ml using ChIP dilution buffer containing 0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris/HCl (pH 8.1) and 167 mM NaCl. The lysate was precleared, an aliquot of 100 μl was removed for processing as ‘Input’, a control for the efficiency of the PCR, and the remaining 1.9 ml was incubated with the appropriate antibody or with pre-immune serum at 4 °C overnight. The immunocomplexes were recovered by Protein A–agarose, and this ‘Bound’ material was washed and eluted in 1% SDS and 0.1 M NaHCO3. The ‘Input’ sample and the ‘Bound’ samples were processed in parallel by reversing the cross-linking by incubation at 65 °C for 4 h, deproteinizing with phenol/chloroform and ethanol precipitation in the presence of Pelletpaint (EMD Biosciences, San Diego, CA, U.S.A.). The ‘Input’ material was resuspended in 100 μl of 1× TE (10 mM Tris/HCl, 1 mM EDTA, pH 8.0) and the ‘Bound’ material was resuspended in 30 μl of the 1× TE. In each FXR2 promoter and 16 centromere PCR reaction, 1 μl of ‘Input’ DNA and 4 μl of ‘Bound’ DNA was used. For the FXR2 gene, the PCR primers used for Sp1 and NF-Y ChIP analysis were forward primer 1 and reverse primer 9 (Table 1); for Nrf1 ChIP analysis, forward primer 10 and reverse primer 11 were used, and for AP2 ChIP PCR, forward primer 12 and reverse primer 2 (Table 1) were used. The centromeric region of chromosome 16 was used as a negative control . The h16CEN-F and h16CEN-R primers were used to amplify the 16 centromere region . In each case roughly 250 times more of the starting material was used for PCR of the ‘Bound’ sample than the ‘Input’ sample. ChIP experiments were repeated three times with each antibody for the FXR2 promoter and at least twice for the negative control.
RESULTS AND DISCUSSION
Sequence conservation in the FXR2 promoter
The approx. 700 bp region immediately upstream of the FXR2 translation start site shows substantial sequence conservation among various organisms (Figure 1). The sequence conservation decreases significantly upstream of this region (results not shown) due primarily to the presence of multiple order- or species-specific SINEs (short interspersed repeated elements). These elements have presumably been inserted into this region since these animals last shared a common ancestor before the mammalian radiation approx. 65 million years ago.
The human FXR2/aABP promoter region
The 706 bp region upstream of the human FXR2 translation start site has a G+C content of approx. 70%. This is consistent with the observation that many other bidirectional promoters have a higher than average G+C content. Like many known bidirectional promoters , there are no canonical TATA box or initiator elements. Like many TATA-less promoters , the human FXR2 promoter region has one CCAAT box (−656/−652 bp) and one inverted CCAAT box (−617/−613 bp). Along with two CCAAT boxes, the 706 bp human promoter region between nucleotide positions −742 to −37 contains various other potential transcription factor-binding sites (see Figure 3A) including two GC boxes (−595/−586 bp and −181/−172 bp), one Nrf1-binding site (−373/−362 bp) and two adjacent AP2-binding sites (−105/−76 bp). The human FXR2 promoter also contains two stretches of (CGG·CCG)4 repeats, one upstream of the FXR2 transcription start site and the second in the 5′-UTR. These CGG·CCG repeats have been suggested to influence both transcription  and translation . These sites are conserved in all five organisms (Figure 1).
Functional analysis of the human FXR2 promoter
The activity of the 706 bp human DNA fragment upstream of the FXR2 translation start site was examined using a luciferase reporter in transient transfection assays. For these experiments we used human SK-N-MC cells that are neuronal in origin and mouse C2C12 cells that are derived from skeletal muscle. These cells were chosen since while FXR2 is ubiquitously expressed during fetal development, significant expression in adults is restricted to a few tissues that include brain and skeletal muscle . The promoter was active in both directions in both cell lines (Figure 2) as it is in rat liver and brain where it drives production of both FXR2 mRNA and an alternative transcript encoding the ABP/SHBG (sex hormone-binding globulin) [30,31]. The human ABP gene also has two promoters and multiple ABP transcripts have been reported in human liver [32,33], but, to date, there is no evidence of any ABP transcripts that initiate in this region in this organ and it remains to be seen if this region acts as an ABP promoter in human brain. However, the fact that both the human and rat FXR2 promoters can function bidirectionally suggests that the ability to drive transcription in both directions was probably a feature of the promoter in the common ancestor of these organisms. Since many bidirectional promoters show co-ordinated expression of the two transcripts, the bidirectionality of the rodent FXR2/ABP gene may reflect the original co-regulation of these two genes. The insertion of multiple retrotransposable elements into this region since the rodent and primate lineages diverged may have disrupted the ability of the human promoter to function as an alternative promoter for the ABP gene, necessitating the recruitment of a novel promoter element closer to the human ABP open reading frame.
The strength of the promoter acting in the FXR2 direction was >50-fold higher than the vector backbone in SK-N-MC cells (results not shown) and slightly lower than that in C2C12 cells. This corresponds to approx. 35% and approx. 25% of the activity of the moderately robust SV40 (simian virus 40) early promoter/enhancer in SK-N-MC and C2C12 cells respectively (Figure 2). This activity was comparable with that seen for the FMR1 promoter under the same conditions (Figure 2). The activity of the promoter in the aABP direction was similar in C2C12 cells, but was significantly lower in SK-N-MC cells.
We generated a series of plasmid constructs that contained 5′ or 3′ deletions of this region or different point mutations and tested their activity in both SK-N-MC and C2C12 cells. Most of the mutations showed the same general trend in both cell lines (Figures 3B and 3C). The one exception was a point mutation in one of the CCAAT boxes (paABP-Δ2). This mutation had a negative effect on the aABP promoter in muscle cells but a positive effect in neuronal cells. Deletion of the conserved Nrf1-binding site had a particularly large impact on the FXR2 promoter (pFXR2-Δ2) in SK-N-MC cells (Figure 3C).
Identification of transcription factors binding to the FXR2/aABP promoter in vitro
To examine the transcription factors bound to the FXR2/aABP promoter, EMSAs were performed using SK-N-MC nuclear extracts and two different DNA fragments. The −742/−404 bp fragment I formed one DNA–protein complex with the nuclear extract from SK-N-MC cells (Figure 4A, lane 2). This complex was eliminated by an excess of either of two oligonucleotides, −689/−621 or −655/−596 bp (Figure 4A, lanes 6 and 7). The −689/−621 bp DNA fragment contains the CCAAT box and the −655/−596 bp fragment contains the inverted CCAAT box (Figure 3A). When used in EMSAs, the major protein–DNA complex formed using each of these two oligonucleotides was supershifted with antibody to NF-YB, a subunit of the trimeric NF-Y transcription factor (Figure 4A, lanes 12 and 18). An excess of unlabelled oligonucleotide containing the consensus CCAAT box eliminated the same DNA–protein complex (Figure 4A, lanes 10 and 16) and the complex was unaffected by an excess of oligonucleotide with the mutated CCAAT box (Figure 4A, lanes 11 and 17). Another DNA–protein complex was also eliminated by consensus CCAAT box (Figure 4A, lanes 10 and 16) and mutated CCAAT box oligonucleotides (Figure 4A, lanes 11 and 17). However, the same effect was seen with unrelated oligonucleotides, indicating that binding was not specific.
Fragment II (−404/−37) formed two major DNA–protein complexes (Figure 4B, lane 2). Antibody to AP2 eliminated complex 2 (Figure 4B, lane 4), while antibody to Nrf1 supershifted complex 3 (Figure 4B, lane 10). An excess of unlabelled oligonucleotide containing a consensus AP2-binding site and Nrf1-binding site eliminated complex 2 (Figure 4B, lane 3) and complex 3 respectively (Figure 4B, lane 9). Complexes 2 and 3 were unaffected by an excess of oligonucleotides with the mutated AP2 (Figure 4B, lane 7–8) and mutated Nrf1-binding site (Figure 4B, lane 13) respectively.
Purified Sp1 protein formed a DNA–protein complex with fragment I (Figure 4C, lane 2) and fragment II (Figure 4C, lane 5). Each of these complexes was competed by an excess of a consensus Sp1-binding oligonucleotide (Figure 4C, lanes 3 and 6).
Transcription factors binding to FXR2/aABP promoter in vivo
To determine if these four transcription factors are bound to the FXR2/aABP promoter in vivo, ChIP was performed using chromatin prepared from the neuronal cell line SK-N-MC. A region of the centromere of chromosome 16 was chosen as a negative control since it is known to be heterochromatinized and transcriptionally silent. While the amount of Nrf1, AP2 and NF-YA/NF-YB bound to the negative control amplicon was relatively low as expected, the amount of material immunoprecipitating with Sp1 was somewhat higher. Since Sp1 associates with histone deacetylases and can serve a repressive function, this binding may not be non-specific. Whatever the source of this binding, Sp1 binding to the FXR2 promoter region was significantly higher than this as illustrated by the significant increase in the ratio of Bound to Input DNA amplified (Figure 5). Significant enrichment of the FXR2 promoter region was also seen when antibodies to NF-YA/NF-YB or AP2 were used in the ChIP assay (Figure 5). A much smaller enrichment of the FXR2 promoter region was seen when the Nrf1 antibody was used (Figure 5).
The role of various transcription factors in regulation of the FXR2/aABP promoter
To test the role of NF-Y, Nrf1 and AP2 in the regulation of the FXR2/aABP promoter, we co-transfected our reporter constructs with dominant-negative versions of these transcription factors. In C2C12 cells both the FXR2 and the aABP promoter activities were negatively affected by the dominant-negative forms of all three transcription factors (Figure 6A). This suggests that these factors act directly or indirectly to positively regulate transcription of both promoters in these cells. However, in SK-N-MC cells the dominant-negative version of NF-YA had little or no effect on the FXR2 promoter but led to an increase in the activity of the aABP promoter. As in C2C12 cells the dominant-negative form of Nrf1 had a negative effect on the activity of both promoters.
To assess the role of Sp factors in FXR2/aABP promoter activity, we transfected the FXR2 and aABP promoter plasmids into Drosophila S2 cells, which lack endogenous Sp factors. The expression of full-length Sp factors, Sp1 or Sp3 resulted in increased promoter activity in both orientations (Figure 6C). There was no increase in FXR2 or aABP promoter activity when co-tranfections were done with plasmids expressing either the DNA-binding domains of Sp1 or Sp3 or full-length Sp4.
Nrf1 is highly expressed in brain  and it has strong homology to the Drosophila ewg (erect wing) gene that is required for normal central nervous development . NF-Y and Sp1 are ubiquitous transcription factors consistent with the ubiquitous tissue distribution of FXR2P during embryonic development. AP2 is highly expressed in the CNS and is critical for normal neuronal development . Our results thus suggest that despite the lack of extensive sequence conservation, the FMR1 and FXR2 promoters may be more similar than they appear: they are both TATA-less promoters that are regulated by Sp1, Nrf1 and AP2 [11–15] and have CGG·CCG repeats in 5′-UTR.
The knowledge of FXR2 promoter regulation gained from these studies may help us to devise ways to up-regulate this gene and thus perhaps to ameliorate the effects of the FMRP deficit in FXS.
We thank Roland Owens (NIDDK, NIH) for his careful reading of this paper. This research was supported by the Intramural Research Program of the NIDDK, NIH.
Abbreviations: ABP, androgen-binding protein; aABP, alternative ABP promoter; AP2, activator protein 2; ChIP, chromatin immunoprecipitation; CNS, central nervous system; DMEM, Dulbecco's modified Eagle's medium; EMSA, electrophoretic mobility-shift assay; FMR1, fragile X mental retardation-1 gene; FMRP, fragile X mental retardation protein; FXR1P, fragile X-related protein 1; FXR2, fragile X-related gene 2; FXS, fragile X syndrome; NF-Y, nuclear transcription factor Y; Nrf1, nuclear respiratory factor/α-Pal; Sp1, specificity protein 1; SV40, simian virus 40; UTR, untranslated region
- The Biochemical Society, London