Identification of a Novel Basic Helix-Loop-Helix-PAS Factor, NXF, Reveals a Sim2 Competitive, Positive Regulatory Role in Dendritic-Cytoskeleton Modulator Drebrin Gene Expression

Isolation of NXF and plasmids.

As a screening probe, 5′-AGGGTGACAGCATCTACGACATCA-3′, found from the predicted exon database in the public domain (Ensemble; www.ensembl.org), was used to clone human NXF from a human fetal cDNA library (GIBCO-BRL) with the Gene Trapper screening system (GIBCO-BRL). Mouse and rat NXFs were cloned from mouse brain (GIBCO-BRL) and rat brain (Takara) cDNA libraries, respectively, by reverse transcription-PCR (RT-PCR), and sequencing was used to confirm that no mutations were derived from PCR. The genomic clone of mouse NXF was cloned from a 129SvJ library (Stratagene). NXF expression plasmids were constructed by introducing the NXF-coding region into pRC/RSV (Invitrogen). Sim2 and Arnt2 expression plasmids were also constructed in the same way from human Sim2- and human Arnt2-coding regions cloned by RT-PCR. The Sim2-VP16 expression plasmid was constructed by replacing the Sim2 repression domain (BamHI C-terminal fragment) with the VP16 activation domain derived from pVP16 (Clontech).

RNA blotting analysis and in situ hybridization.

Commercially available human RNA-blotted membranes (MTN blot and MTN blot IV; Clontech) were analyzed by using a cDNA probe for the full-length coding region of human NXF under highest-stringency conditions. Rat NXF cDNA (HindIII C-terminal fragment) was subcloned into pGEM (Promega), and T7 polymerase or SP6 polymerase was used to make the sense probe and the antisense probe, respectively. In situ hybridization of tissue sections from 8-week-old Sprague-Dawley rats was performed with these probes as described previously (10). For analysis of NXF expression in developing embryos, RT-PCR was performed with commercially available mouse embryo cDNAs (Multiple-Tissue cDNA; Clontech) in parallel with 4-week C57BL mouse brain cDNA.

Gal4 or VP16 fusion proteins and assays.

Gal-NXF(C), Gal-NXF(256-596), Gal-NXF(597-802), Gal-NXF(N), Gal-Sim2(N), and Gal-Clock(N) were constructed with fragments introduced into pRC/RSV-Gal4 (termed Gal4-DBD in Fig. 3). The Gal4 part was the Gal4 DNA binding domain derived from pM (Clontech). The other part in each construct was as follows: Gal-NXF(256-596), amino acids 256 to 596; Gal-NXF(597-802), amino acids 597 to 802; NXF(N), amino acids 1 to 596 (containing bHLH-PAS); Sim2(N), human N-terminal BamHI fragment (bHLH-PAS); and Clock(N), human HincII-NcoI fragment (bHLH-PAS). VP16-CP was pVP-CP (Clontech). VP16-X target plasmids (various bHLH-PAS regions) were constructed with pVP16 (Clontech): Arnt1 fragment, human N-terminal NaeI fragment (bHLH-PAS); Arnt2 fragment, N-terminal BglII fragment (bHLH-PAS). The BMAL1 and BMAL2 inserts were full length. The Sim2, Clock, and NXF fragments were the same as Sim2(N), Clock(N), and NXF(N). All factors were cloned by RT-PCR, and the lack of introduced mutations was confirmed by sequencing. A Gal4 reporter with four copies of the Gal4 binding element was constructed with two copies of the oligonucleotide 5′-CGCGTCGAGCTCGGGTCGGAGGACTGTCCTCCGACTGCTCGAGTCGAGCTCGGGTCGGAGGACTGTCCTCCGACTGCTCGAGA-3′ introduced into pGL3 (Promega). Transcriptional activity of the NXF C terminus was analyzed in the IMR32 neuroblastoma (ATCC CCL127) with a pRL control (Promega) according to the manufacturer's instructions. Essentially the same results were also obtained with 293 and HeLa cells. The interaction assay was performed in 293 cells by the mammalian two-hybrid assay with a standard procedure according to the instructions of the manufacturer (Clontech).

In vitro DNA binding assays.

Recombinant baculoviruses expressing NXF or Arnt2 were generated through homologous recombination between transfer vector pVL1392 (PharMingen) and a linearized baculovirus genome (PharMingen). pVL1392-NXF and pVL1392-Arnt2 carried the full-length coding regions. The resultant viruses were infected with Sf21 (Invitrogen) to give whole-cell lysates (15), and the in vitro DNA binding assay was performed with combinations of these as described previously (15). Cold E-box was from the Clock-responsive element (25). Labeled probes were as follows: CME (central nervous system midline element) (30), 5′-CTAGAAATTTGTACGTGCCACAGA-3′; GCGTG core, 5′-CTAGAAATTTGTGCGTGCCACAGA-3′; TCGTG core, 5′-CTAGAAATTTGTTCGTGCCACAGA-3′; and CCGTG core, 5′-CTAGAAATTTGTCCGTGCCACAGA-3′. Putative NXF-responsive elements from the NXF or Drebrin promoter as a probe were as follows: Seq1, 5′-ACAGGCAAACCACGACCATTCTGA-3′; Seq2, 5′-CATTCAGCACCACGGACAGCGACA-3′; Seq3, 5′-CTTTTCACTCCACGTTGCCTCTCT-3′; Seq4, 5′-GGATTAAAGGCGTGTGCCACCACG-3′; Seq5, 5′-GTGTGCCACCACGCCCGGCTTCCC-3′; Seq6, 5′-GGGAGGGGATCGTGGGAGAGGTTC-3′; Seq7, 5′-CGTTCGACGTCACGGGATGACGTC-3′; Seq8, 5′-ACGAGCCCCCCACGCCTGTCAGGAG-3′; Seq9, 5′-TGTGAATGAACGTGGCTGGCCCGG-3′; Seq10, 5′-GGGTGGCAGACACGAAGTCCTGGT-3′; Seq11, 5′-TGAGAGTGGGCGTGGCTGGAGAGT-3′; Seq12, 5′-GGAGCCCTGCCGTGGGAGTCTGGG-3′; Seq13, 5′-CTCTGGGAGGCGTGAGGCTTGAGA-3′; Seq14, 5′-AGGGCACAGGCGTGGAGGGGAGGA-3′; Seq15, 5′-GAAGATGTGCCGTGACACATGTTG-3′; Seq16, 5′-TTGGGGGGGACACGTGCTAAGCATG-3′. The numbering of these probes is the same in Fig. 5c and d. The mutated NXF promoter in Fig. 6a was created by site-directed mutagenesis (Quick-Change PCR kit; Stratagene). The mutated sequences (mutations are underlined) are 5′-ACAGGCAAACCTTGACCATTCTGA-3′ at Seq1, 5′-GGATTAAAGGCAAGTGCCACCACG-3′ at Seq4, and 5′-GGGAGGGGATCAAGGGAGAGGTTC-3′ at Seq6 on the NXF promoter. Mutated sequences in mutated Drebrin promoters A and B are 5′-TGTGAATGAACAAGGCTGGCCCGG-3′ at Seq9, 5′-GGAGCCCTGCCAAGGGAGTCTGGG-3′ at Seq12, 5′-GAAGATGTGCCAAGACACATGTTG-3′ at Seq15, 5′-GGGTGGCAGACTTGAAGTCCTGGT-3′ at Seq10, 5′-CTCTGGGAGGCAAGAGGCTTGAGA-3′ at Seq13, and 5′-AGGGCACAGGCAAGGAGGGGAGGA-3′ at Seq14.

Target gene identification.

The SK-N-MC neuroblastoma was transfected with NXF expression plasmids by using a Gene Pulser (Bio-Rad). From the resulting transfectant and parental cells or NXF antisense cDNA transfectants, RNAs were extracted and labeled with ³²P with an RNA labeling kit (Clontech). Each probe was applied to a nylon membrane DNA array (Atlas cDNA expression array; Clontech), and endogenous gene expression patterns were compared. To confirm the induction of genes, RT-PCR analyses were performed with PCR primers for Drebrin, Sim2, and NXF, the last primer amplifying the NXF 3′ noncoding tail sequence but not the NXF-coding sequence within the NXF expression plasmid. The 5′ upstream promoter regions (−2.7 kbp to exon 1) of NXF and Drebrin were cloned by PCR and introduced into pGL3 (Promega).

Chromatin immunoprecipitation assay.

By use of standard protocols, the SK-N-MC cells transfected with an N-terminal Flag tag-fused NXF and Arnt2 expression plasmid were maintained in neomycin selection medium for 2 months, and one stable SK-N-MC transformant cell line among the resultant neomycin-resistant colonies was selected by anti-Flag Arnt2 immunoblotting analysis. Chromatin immunoprecipitation assay was performed with anti-Flag-M2 agarose (Sigma) and a ChIP kit (Upstate Biotechnology) with PCR primer pair A (5′-GAGGCTGGCCCTGAGCGTCC-3′ and 5′-CCACTTATTCCAGGCCATTC-3′), which amplifies the NXF/Arnt2 binding region on the Drebrin promoter, and primer pair B (5′-CAGGTGGTGCTTCTGGTACC-3′ and 5′-CCGCTTTCCTCCACGGGCAG-3′), which amplifies Arnt2 gene exon 2, as the negative control primer. The buffers and procedure used were as described by the manufacturer.

Nucleotide sequence accession numbers.

The human, rat, and mouse NXF cDNA sequences and the mouse NXF genomic sequence have been deposited in GenBank under accession numbers AB049469, AB050103, AB049835, and AB054577, respectively.

Cloning of NXF, a novel bHLH-PAS factor with low sequence similarity to other vertebrate factors.

In the database of predicted exons from the human genome, we found a short partial fragment that was thought to code for a PAS domain that was characteristic of certain kinds of transcription factors. Screening of a human fetal brain cDNA library with the corresponding oligonucleotide probe generated a positive full-length clone encoding a novel bHLH-PAS domain factor, which we termed human NXF. Rat and mouse forms of NXF were also cloned and sequenced. Their sequences showed high identity to the bHLH-PAS family and were well conserved among these mammals (Fig. 1a). In homology search analyses against all known genes in gene databases, the highest degree of similarity was found with C15C8.2, a predicted Caenorhabditis elegans putative transcription factor with an unknown function (17); the similarity was only in the bHLH-PAS domain (Fig. 1b). In contrast, among the vertebrate bHLH-PAS family, only Sim2 showed very slight similarity to NXF along the overall sequence (amino acid identity, 24%); no other specific homologous gene with striking identity was found. The top three matches from homology searches in each domain showed comparable low degrees of similarity (Fig. 1c). The primary structure of NXF thus suggests that this novel factor could have a unique function distinct from those of already-known bHLH-PAS factors.

Chromosomal location and mRNA expression pattern of the NXF gene.

From screening of a 129 mouse genomic library, a lambda phage clone carrying the mouse NXF locus was obtained and completely sequenced. The NXF-coding region was divided into eight exons spanning 5 kbp within an approximately 20-kbp phage insert (Fig. 2a). This composition appears to be significantly conserved between mouse and human forms, according to human genome draft sequence data available in the public domain. With a computerized database of human genome sequences and a chromosome STS map, we identified several genome clones, such as AP001107, as containing the human NXF genome sequence, although no gene identification had been reported for the NXF region. This AP001107 genome clone contained the 54TM gene (20), the TEM1 gene (6), and the RIN1 gene (27) around the NXF locus, flanked by STS marker PYGM, D11S913, or D11S1889 (18, 19, 31, 32). Another genome clone, AP002748, which overlapped AP001107 and had the neighbor of the NXF region, contained the BBS1 gene, which is responsible for Bardet-Biedl syndrome type I (23). The NXF gene was located between D11S913 and this BBS1 locus, roughly 130 kbp distant from the latter on the human chromosome (Fig. 2b). Since all of these clustered genes and STS makers have been mapped experimentally at 11q13 on the human chromosome (19, 23), the NXF locus will be also located at 11q13.

RNA blot analysis showed a major NXF transcript present in the brain among the human adult tissues tested (Fig. 2c), with very faint signals in the skeletal muscle and kidney. Although human NXF was cloned from a fetal brain cDNA library, in contrast to embryonic expression of positive control factor such as Sim2 or NPAS3 (3), we failed to detect NXF expression in the developing mouse embryo by in situ hybridization, and we could detect NXF mRNA only in the 4-week postnatal brain by 30-cycle RT-PCR, among various developing stages of mouse embryos (at embryonic day 7 [E7], E11, E15, and E17) and 4-week postnatal brain (Fig. 2d). From 45-cycle PCR, we found that NXF started mRNA expression at E17 after Sim2 gradually reduced its expression. In situ hybridization analysis of adult rat brain revealed the dotted pattern of NXF expression in the cerebral cortex and striatum, with prominent expression in the neuronal cell layers of the hippocampus. In the cerebellum, we observed a very slight signal only at the Purkinje cell layer. These expressions appear to overlap with a neural gene Drebrin expression pattern, while they overlapped only in part (at the hippocampus) with that of Sim2 (Fig. 2e). Such a neural expression pattern of NXF implies that NXF may be involved in certain neuronal functions in the brain after birth rather than determining cell fate in the embryo.

NXF has a transcriptional activation domain and forms heterodimers with Arnt family members.

The Gal4 fusion protein with the NXF C terminus [Gal-NXF(C)] (Fig. 3a) could activate a Gal4 reporter construct that contained Gal4 DNA binding elements upstream of the promoter and a luciferase reporter gene. In addition, overexpression of the general transcriptional coactivator p300 enhanced the activity of the NXF C terminus (Fig. 3a, bars 3 and 4). p300 can associate with many unrelated transcription factors, such as nuclear hormone receptor family members, AP1, ETS-1, ETS-2, Arnt, etc. Considering these results, we concluded that NXF had a typical type of transcriptional activation domain that utilizes the general p300 coactivator mechanism as a transcription factor. This domain lay in the C-terminal last half (amino acids 597 to 802) (Fig. 3a, bars 5 and 6), while we do not identify specific sequence feature in the region.

The bHLH domains of bHLH-PAS DNA binding factors may function as heterodimers or homodimers for DNA binding, as suggested by X-ray crystallographic data of the bHLH part from, for example, the bHLH-LZ Max protein (11). To examine whether NXF could form the dimer, we performed mammalian two-hybrid assays. NXF showed specific interaction with Arnt subclass bHLH-PAS factors (Arnt1, Arnt2, and BMALI) in this assay, using a Gal4 fusion protein with the N terminal structure of NXF (amino acids 1 to 595, composed mainly of the bHLH-PAS motif) for bait and the VP16 activation domain fused with various known bHLH-PAS factors as targets (Fig. 3b, panel 1). We did not observe any NXF-NXF homodimerization. In addition, no interactions with other bHLH-PAS factors (BMAL2, Sim2, and Clock) could be observed (Fig. 3b, panel 1, bars 5 to 8). In experiments performed for confirmation, using Gal4 fusion proteins with Arnt1, Arnt2, BMAL1, or BMAL2 substituting for NXF as the bait, all Arnt subclass members except BMAL2 showed specific interaction with the VP16-NXF fusion target (data not shown). The interaction specificity in this assay was validated by control experiments showing specific associations of Sim2 with Arnt1 and Arnt2 and of Clock with BMAL1 and BMAL2, using Gal-Sim2(N) or Gal-Clock(N) as a substitute for NXF(N) (Fig. 3b, panels 2 and 3). These results indicate that NXF functions as a heterodimer with Arnt subclass members (Arnt1, Arnt2, and BMAL1), preferably with Arnt2, which is known to be abundant and restricted in brain (15).

NXF can bind and activate the Sim2-repressive CME element and its variants.

To identify potential NXF/Arnt target DNA elements, we performed a screening against randomly selected oligonucleotides with known bHLH-PAS factor-responsive sequences in the DNA binding assay. Recombinant NXF or Arnt2 as a protein source was expressed with the baculovirus vector. When we used CME (30), a Sim/Arnt-responsive DNA element, as a probe, we could detect a specific binding signal depending on both NXF and Arnt2 (Fig. 4a, lane1). The CME element is known to consist of two half sites, a Sim2 recognition half site and the flanking GTG (26) recognized by Arnts. To optimize the element for an NXF/Arnt2 complex, we systematically made a series of CME variants that were mutated one by one at the first and second positions next to the GTG Arnt binding sequence in the CME core (Fig. 4a, upper panel) and additional variants that were mutated at third position to GTG, containing parental ACGTG, GCGTG, TCGTG, or CCGTG as core sequences (Fig. 4a, lower panels). These DNA binding assays revealed a critical CGTG nucleotide, each TCGTG core-containing element as the best NXF/Arnt2 binding sequence, other tolerated NXF/Arnt2 binding elements such as GCGTG core-containing sequences (which are also another Sim2-responsive element) (26), and a weak CCGTG core sequence for NXF/Arnt2 binding. These results indicate that the sequence preference of NXF could partially overlap with that of Sim2, while most preferable element of NXF is distinct from CME, the Sim2 most preferable element. Besides the best element, as shown by confirmation experiments (Fig. 4b and c), NXF could indeed recognize CME as a direct binding element and its responsive element (Fig. 4b, lane 5). Neither NXF nor Arnt2 recognized this CME element alone. The binding signal was competed out with a nonlabeled CME probe but not with an unrelated nonlabeled probe (E-box element), supporting the binding specificity in this assay (Fig. 4b). Consistent with such a result of the DNA binding experiment, significant transcriptional induction by the NXF/Arnt2 complex was observed on four tandem copies of CME in a reporter gene assay with the IMR32 neuroblastoma cell line (Fig. 4c, bar 6). This induced activity was impaired in a dose-dependent manner by Sim2 (Fig. 4c, bars 7 and 8). Antisense Sim2 as a negative control failed to repress the NXF/Arnt2. Essentially the same results were also obtained with 293, PC12, and other cell lines tested. These results may indicate possible cross talking between an NXF positive signal and a Sim2 negative (repressive) signal in some gene transcriptional regulation.

To investigate the transcriptional activities of NXF and Sim2 on the CME variants shown in Fig. 4a, we performed reporter gene analyses with reporter constructs that have two copies of CME, the GCGTG core element, the TCGTG core element, or the CCGTG core element in the promoter. As shown in Fig. 4d, every reporter construct showed transcriptional activation dependent on Arnt2 and NXF, consistent with the DNA binding results. Moreover, Sim2-dependent transcriptional repression of the induced activity was observed with each reporter construct in a Sim2 dose-dependent manner. Since a Sim2-VP16 chimera protein, which is a strong transcriptional activator with Sim2 DNA binding specificity, was capable of activating every reporter construct (Fig. 4d, bars 9), we concluded that the repressive activity of native Sim2 observed here at least includes direct-type repression which depends on specific binding of the Sim2/Arnt2 complex to each variant element, besides functional interference which withdraws the common partner Arnt2. While the core sequence preference for NXF/Arnt2 activation and for Sim2/Arnt2 repression activity appears to be slightly different from the quantitative viewpoint, these results suggest that some group of genes activated by NXF could overlap with those repressed by Sim2.

Identification of potential NXF target genes.

To identify the target genes that respond to NXF expression, we applied DNA array technology to compare endogenous gene expression pattern in the SK-N-MC neuroblastoma host cell line with that in transfectants overexpressing NXF cDNA. In the transfectants, several neuronal function-related genes showed alteration of mRNA expression relative to parental NXF-negative SK-N-MC cells (data not shown). Interestingly, the endogenous NXF gene itself was found to be autoregulated by NXF (Fig. 5a, bottom). This is reminiscent of the Drosophila Sim gene autoregulation loop (24). Besides NXF, one of the most induced genes with NXF overexpression was Drebrin (14). RT-PCR analysis confirmed induction of the endogenous Drebrin gene by NXF overexpression in the cell (Fig. 5a, top), while the Sim2 gene for negative control was not responsive to NXF and its endogenous mRNA was constantly expressed with or without NXF (Fig. 5a, middle). Drebrin is reported to be abundant in the hippocampus of the human brain in immunocytochemical studies (13). We also confirmed the prominent Drebrin expression in the neuronal layer of the hippocampus by in situ hybridization (Fig. 2e). It is interesting that the Drebrin gene expression pattern in the brain closely overlaps with that of NXF, which is consistent with our results described here.

NXF can activate the expression of the Drebrin gene or NXF itself through each promoter, and its activity is competed with Sim2.

To determine whether NXF directly regulates the transcription of the Drebrin gene and the NXF gene itself through their promoters, we performed the reporter gene assay with a luciferase reporter construct fused with the NXF promoter (−2.7 kbp) or the Drebrin promoter (−2.7 kbp) (Fig. 5b). On both promoters, the NXF/Arnt2 complex caused significant transcriptional induction. Furthermore, Sim2-dependent transcriptional repression of the induced activity in a dose-dependent manner was observed on each promoter. The Sim2-VP16 chimera could again activate both promoters (Fig. 5b, bar 7), suggesting that the direct binding of Sim2/Arnt2 on the promoter sequence partly contributes the repression in addition to functional interference. These results indicate that both NXF/Arnt2 and Sim2 act on the gene expression directly through the promoter sequence of Drebrin or NXF itself.

The NXF gene and the Drebrin gene have NXF-activating DNA elements on their promoters.

In a search of the promoter sequences of NXF and Drebrin, putative NXF-responsive elements were found. The positions on each promoter are shown and numbered in Fig. 5c. To examine these numbered elements, we performed DNA binding analysis with each labeled element as a probe, and we found some candidate elements on both promoters (sequences 1, 4, 6, 7, 10, 13, 14, and 16) to be good NXF/Arnt2 binding elements, although this was not the case for the remainder (Fig. 5d, left). We also performed reporter gene analysis with chimera gene constructs having each DNA element (Fig. 5d, right). In this in vitro assay, some of these elements from the Drebrin or NXF gene promoter were activated by NXF/Arnt2 and repressed by Sim2, and such transcriptional activities corresponded to their binding to NXF/Arnt2.

To examine NXF/Arnt2 binding elements on the NXF and Drebrin promoter regions further, we constructed several mutated reporter plasmids. One is a mutated NXF promoter with several point mutations at the strong NXF/Arnt2 binding elements in the promoter (positions 1, 4, and 6). Another is a Drebrin promoter A that is mutated only at the false NXF/Arnt2 binding elements (positions 9, 12, and 15), which do not show a significant NXF/Arnt2 binding signal in our DNA binding experiments. The last is a Drebrin promoter B which has mutations at the strong NXF/Arnt2 binding elements (positions 10, 13 and 14). We performed additional reporter analysis with these constructs and observed dramatic decreases of NXF/Arnt2-dependent inductivity of the mutated NXF promoter and mutated Drebrin promoter B (Fig. 6a). In contrast, the Drebrin promoter A which is mutated only at the false NXF/Arnt2 binding elements showed no significant decrease of promoter activity.

Moreover, to strengthen the conclusion that the Drebrin promoter has direct NXF/Arnt2 binding elements, we performed chromatin immunoprecipitation assays. Anti-Flag antibody-Flag-fused NXF/Arnt2 protein complex could specifically pull down the Drebrin promoter element on chromatin in Flag-NXF/Arnt2-overexpressed transformants (Fig. 6b). In this analysis, control preprecipitated input (sonicated chromatin fragment) and immunoprecipitated chromatin fragment were amplified with PCR primer A for the Drebrin promoter or with PCR primer B for the negative control Arnt2 gene exon 2 region. Only primer A could amplify the chromatin that precipitated with anti-Flag antibody-Flag-fused NXF/Arnt2 protein complex, and the amplified product was competed with excess Flag peptide in immunoprecipitation, showing the specificity of the assay (Fig. 6b). These results indicate that Flag-NXF specifically associated endogenous Drebrin gene promoter region in vivo.

Conclusion.

Here we have reported the identification of novel bHLH-PAS-type transcription factor NXF. With several experiments, we concluded that the Drebrin gene and NXF itself were direct targets of NXF and that their promoter activity could be repressed by Sim2 (Fig. 7a).

An overdose of Sim2 protein due to the Sim2 locus trisomic state in DS has been proposed on the basis of the quantitative analysis of Sim2 expression in mouse models of DS (28), and this has been thought to contribute to the DS symptoms (5, 8). Drebrin has been reported to participate in synaptic plasticity (16), being engaged in morphological changes and structure-based plasticity of neurons at postsynaptic terminals in dendritic spines (14). This is a critical event in memory and information storage, although the mechanisms have yet to be completely elucidated. In DS patients' brains, where excess Sim2 causes symptoms such as impairment of learning and memory (8, 28), Drebrin is known to be remarkably reduced and even to be undetectable (29), with associated atrophy of the dendritic tree, attenuation of neuritic outgrowth, and specific dendritic spine impairment (1, 2). It is interesting that this specific and intensive decrease of Drebrin is consistent with our data described in this report.

We propose a possible mechanism of Sim2 action in the pathogenesis of DS, describing the contributions of NXF and Drebrin in the model (Fig. 7b). Sim2 would affect two points in this model, i.e., the positive feedback system of NXF gene expression and the regulatory mechanism of Drebrin expression. We have reported here that both of these were regulated and balanced by NXF and Sim2. In the DS brain, with a 1.5-fold (28) dose of Sim2 due to Sim2 gene trisomy, the positive feedback loop of the autoregulation system of NXF gene expression would be turned down exponentially, and the effect of the 1.5-fold dose of Sim2 might be amplified at this point. The resulting low level of NXF protein would be additionally competed by the excess Sim2 in DS, and that should synergistically affect the expression of certain target genes such as Drebrin.

The identification of NXF and its novel signaling with Sim2 may give the initial clues and tools to dissect some of the signaling pathways underlying the pathogenesis in DS or other neural diseases, besides contributing to the understanding of neural gene transcriptional regulation.