GC Box-binding Transcription Factors Control the Neuronal Specific Transcription of the Cyclin-dependent Kinase 5 Regulator p35*

Cyclin-dependent kinase 5 (cdk5)/p35 kinase activity is highest in post-mitotic neurons of the central nervous system and is critical for development and function of the brain. The neuronal specific activity of the cdk5/p35 kinase is achieved through the regulated expression of p35 mRNA. We have identified a small 200-bp fragment of the p35 promoter that is sufficient for high levels of neuronal specific expression. Mutational analysis of this TATA-less promoter has identified a 17-bp GC-rich element, present twice, that is both required for promoter activity and sufficient for neuronal specific transcription. A GC box within the 17-bp element is critical for both promoter activity and protein-DNA complex formation. The related transcription factors Sp1, Sp3, and Sp4 constitute most of the GC box DNA binding activity in neurons. We have found that both the relative contribution of the Sp family proteins to GC box binding and the transcriptional activity of these proteins is regulated during neuronal differentiation. Thus, our data show that the GC box-binding Sp proteins contribute to the regulation of p35 expression in neurons, suggesting changes in the Sp transcription factors level and activity may contribute to cell type-specific expression of many genes in the central nervous system. The DNA elements and transcription factors regulating the spatial and temporal expression of genes in the central nervous system are critical for its development and function. Cyclin-de-pendent kinase 5 (cdk5) 1 has kinase activity that

Cyclin-dependent kinase 5 (cdk5)/p35 kinase activity is highest in post-mitotic neurons of the central nervous system and is critical for development and function of the brain. The neuronal specific activity of the cdk5/p35 kinase is achieved through the regulated expression of p35 mRNA. We have identified a small 200-bp fragment of the p35 promoter that is sufficient for high levels of neuronal specific expression. Mutational analysis of this TATA-less promoter has identified a 17-bp GC-rich element, present twice, that is both required for promoter activity and sufficient for neuronal specific transcription. A GC box within the 17-bp element is critical for both promoter activity and protein-DNA complex formation. The related transcription factors Sp1, Sp3, and Sp4 constitute most of the GC box DNA binding activity in neurons. We have found that both the relative contribution of the Sp family proteins to GC box binding and the transcriptional activity of these proteins is regulated during neuronal differentiation. Thus, our data show that the GC box-binding Sp proteins contribute to the regulation of p35 expression in neurons, suggesting changes in the Sp transcription factors level and activity may contribute to cell type-specific expression of many genes in the central nervous system.
The DNA elements and transcription factors regulating the spatial and temporal expression of genes in the central nervous system are critical for its development and function. Cyclin-dependent kinase 5 (cdk5) 1 has kinase activity that is primarily detected in the brain, although cdk5 itself has a broader distribution (1,2). The neuronal specificity of cdk5 kinase activity is achieved through its association with an obligate regulatory partner, either p35 (3)(4)(5) or p39 (6,7), whose expression patterns are spatially and temporally regulated (8 -13). p35 expression is predominant in the brain (4,5) and is highest in post-mitotic neurons of the central nervous system, with ex-pression peaking in actively migrating cells in the developing cerebral cortex (11,13). The p35 protein has a short half-life (14), and during embryogenesis there is a direct correlation between p35 mRNA levels and cdk5 kinase activity (11,13), suggesting that regulation of p35 mRNA levels is the major determinant controlling cdk5 activity during development.
The proper regulation of cdk5 kinase activity is essential for development and maintenance of the central nervous system. A gene disruption of either cdk5 or p35 leads to an abnormal development of the brain, and the cdk5 disruption is lethal (15,16). The most striking brain abnormality resulting from disruption of the p35 gene is a severe cortical lamination defect characterized by the reversed packing order of cortical neurons, suggesting that the cdk5/p35 kinase is important for neuronal migration. Dominant negative cdk5 mutants as well as antisense p35 have been found to inhibit neurite outgrowth, supporting a role for cdk5/p35 kinase in this process during neuronal differentiation (17)(18)(19). Identification of several cdk5/p35 substrates that play a role in cytoskeletal organization, such as neurofilament (20,21) and tau (22)(23)(24)(25), are consistent with the proposed role of cdk5/p35 in development. The cdk5/p35 kinase may also have roles in the adult central nervous system. For example, cdk5 appears to be involved in the regulation of exocytosis from synaptic vesicles (26), and recent evidence suggests that cdk5 modulates dopamine signaling in neurons (27,28). Furthermore, improper cdk5 activity, caused by association with a proteolytic fragment of p35, may be involved in pathogenesis of cytoskeletal abnormalities and neuronal death in neurodegenerative diseases such as Alzheimer's (25,29,30) and amyotrophic lateral sclerosis (31).
The cell type-specific, spatial and temporal expression of p35 is important for regulating the kinase activity of cdk5 during development. It has been shown recently in transgenic mice that expression of a heterologous gene regulated by a 1.2-kb fragment of the p35 5Ј region closely resembles endogenous p35 and occurs in neuronal but not glial cells (32). In this study we have analyzed the mouse p35 promoter to identify the DNA elements and transcription factors regulating p35 expression in post-mitotic neurons. We have identified a 17-bp DNA sequence that is both required and sufficient for promoter activity in neurons. Mutational analysis has shown that a GC box element within this sequence is essential for both high levels of p35 promoter activity in neurons and protein binding. Electromobility shift assays have revealed that Sp1, Sp3, and Sp4 are the major transcription factors binding to the GC box in neuronal cells. In addition, our data show that there are changes in both the ratios and transcriptional activities of the Sp proteins during neuronal differentiation that could account for enhanced activity of the GC box in neurons. Therefore, these studies suggest that the GC box-binding Sp proteins regulate the levels and the neuronal specific expression of the cdk5 activator p35.

EXPERIMENTAL PROCEDURES
Primer Extension and Northern Blot Analysis-Total RNA was isolated using Trizol reagent (Invitrogen). For primer extensions radiolabeled primers were hybridized to 20 g of RNA isolated from mouse brain (P0 and P18) or kidney (P18) at 60°C for 90 min and then reversed-transcribed at 42°C for 60 min with RNase H Ϫ reverse transcriptase (Invitrogen). The primers used were 357p 5ЈAGTCAGG-CAGGCTCCCGCGGCA3Ј and 303p 5ЈGGAATCCAACCAGGCCGCGC-A3Ј. For Northern analysis, 20 g of RNA was denatured and separated by gel electrophoresis on 6% formaldehyde, 1% agarose gels and then transferred to Hybond-N ϩ membrane (Amersham Biosciences). Hybridization with 32 P-labeled p35 or GAPDH probes was carried out using Quickhyb (Stratagene).
Cell Culture and Transfections-P19 cells were grown as described previously (33). To induce neuronal differentiation, P19 cells were aggregated and treated for 4 days with 0.3 M all-trans-retinoic acid (RA) (Sigma) dissolved in Me 2 SO and then plated onto tissue culture plates and grown in the absence of RA until harvesting (33). Control cells were treated similarly, but the RA was omitted. P19 cells were transiently transfected using LipofectAMINE reagent (Invitrogen). 1 ϫ 10 4 4-day control Me 2 SO-exposed P19 cells or 2 ϫ 10 5 4-day RA-treated P19 cells were plated into 24-well tissue culture plates, and 2 days after plating cells were co-transfected with 500 ng of luciferase reporter construct and 25 ng of the Renilla reporter construct pRL-tk. Primary cortical neurons were prepared from E18 rat cortices and plated onto laminin and poly-D-lysine-coated coverslips. Cortical neurons were transiently transfected with 1 g of luciferase reporter construct and 50 ng of pRL-tk using calcium phosphate 4 days after plating. Cells were harvested 24 h after transfection, and luciferase and Renilla assays were performed using the Dual Luciferase Assay kit (Promega). For all promoter assays the activity of the luciferase construct was normalized to the Renilla activity of the pRL-tk construct. All transfections were done in triplicate on multiple occasions.
Immunostaining-Primary cortical neurons plated onto laminin and poly-D-lysine coverslips were fixed in 4% paraformaldehyde, washed 3 times with PBS, and incubated for 30 min at room temperature in blocking buffer (10% normal goat serum, 3% bovine serum albumin, 0.1% Triton X-100, 0.1% Tween 20 in 1ϫ PBS). Cells were then washed once with PBS and incubated overnight at 4°C with anti-TuJ1 antibody (Covance Research Products). Cells were washed 3 times with PBS and incubated for 1 h at room temperature with Texas Red-conjugated sheep anti-mouse antibody (Amersham Biosciences). Following this incubation cells were washed 3 times with PBS, Hoechst-stained, washed with PBS again, and then mounted in Prolong antifade reagent (Molecular Probes).
Plasmid Constructs and Cloning-The indicated regions of the p35 promoter DNA were amplified by PCR and cloned upstream of either GFP in pEGFP1 or luciferase in pGL2. Mutations of the p35 promoter were all generated by PCR methods and verified by sequencing. The 3 ϫ 17-mer-TATA reporter construct was generated by cloning 3 copies of the 17-bp element upstream of the TATA box in E1B-luc. G5-luc, which contains 5 copies of the Gal4-binding site upstream of the luciferase reporter gene, has been described previously (35). The expression plasmids containing the Gal4 DNA-binding domain (amino acids 1-147) fused to the VP16 activation domain, AP2 activation domain, Sp1 (amino acids 83-778), Sp3 (amino acids 1-527), and Sp4 (amino acids 98 -621) have been described previously (36 -39).

RESULTS
Identification of the Mouse p35 Transcription Start Site-As a first step to investigate the regulation of p35 mRNA expression, the position of the mouse p35 promoter was identified by mapping the transcription start site. Like many neuronal genes, the DNA upstream of the mouse p35 gene is GC-rich and lacks a consensus TATA box. Primer extension analysis with multiple primers located at positions 5Ј of the ATG produced several bands specifically with RNA isolated from brain, suggesting that there were several transcription start sites located between 416 and 406 bases upstream of the ATG (Fig. 1A and data not shown). As shown in Fig. 1A, reactions using both the 303p and the 357p primers revealed a predominant start site 411 bp upstream of the ATG (Fig. 1, A and B). It is interesting to note that primer extension analysis with the primer 303p, which lies within a region previously identified as an intron A, the results of primer extension assays with primers that hybridized 357 or 303 base pairs upstream of the p35 ATG using total RNA isolated from mouse brain (P18 or P0) or kidney (P18) as a template. DNA sequencing of p35 genomic DNA was performed with the same primers. B, a schematic of the p35 promoter indicating the transcription start site. The start site (ϩ1) lies within a 26-bp region that is 100% conserved between mouse and human (indicated by the box). The position of the primers used for the primer extension assays are indicated by arrows. The reported intron is represented by the dashed line (40). The sequence below the schematic is of the conserved 26-bp region containing the transcription start site, the position of which is indicated as ϩ1. (40), indicates the presence of unspliced p35 mRNA in the brain (Fig. 1A). Intriguingly, the major start site that we have identified lies within a 26-bp region of the mouse p35 promoter that is 100% conserved with sequences 5Ј of the human p35 gene.
The p35 Promoter Is Sufficient for Expression in Primary Cortical Neurons-In transgenic mice a 1.2-kb fragment of the p35 promoter (corresponding to Ϫ895 to ϩ325 relative to our start site mapping) was found to drive expression in neuronal but not glial cells, similar to endogenous p35 (32). We examined the expression of a reporter construct containing 1.17 kb of the p35 promoter (Ϫ1116 to ϩ54) fused to GFP ( Fig. 2A) following transfection into primary cortical cells isolated from E18 rat embryos. The p35-GFP reporter construct was expressed in the primary cortical cells, and the cells expressing GFP were indeed neurons as shown by immunostaining with the neuronal cell marker TuJ1 (Fig. 2B). Thus, in a cell culture system the p35 promoter appears to be sufficient for expression in primary cortical neurons.
Transcriptional Mechanisms Regulate the Neuronal Specific Expression of p35-To address how transcriptional mechanisms were regulating the neuronal specific expression of p35 the P19 cell line was utilized. P19 cells are a mouse embryonal carcinoma cell line that can be induced to differentiate into neuronal P19 cells by aggregation and treatment with retinoic acid (41). P19 cells were chosen as a model system because the differentiated cells closely resemble neurons that are present in the mammalian central nervous system, and like normal neurons, P19 neurons are post-mitotic (41). This is an important characteristic as p35 is only expressed in post-mitotic neurons of the central nervous system (11,13). To determine whether P19 cells could be used to analyze regulation of p35 expression during neuronal differentiation, Northern blot analysis was performed using RNA isolated from undifferentiated or neuronal P19 cells. Undifferentiated P19 cells or P19 cells that had been treated with RA for 2 days, a stage in the P19 differentiation process when the cells do not yet display a neuronal phenotype, showed very low levels of p35 mRNA (Fig.  3A, 2nd and 3rd lanes). However, p35 mRNA was induced in P19 neurons (Fig. 3A, 4th lane). Thus, P19 cells appeared to be a good cell line in which to study the mechanisms regulating the neuronal specific expression of p35 mRNA.
To determine whether transcriptional mechanisms were important for regulating expression of p35 mRNA during neuronal differentiation in P19 cells, a reporter construct containing 1.17 kb of the p35 promoter (Ϫ1116 to ϩ54) fused to the luciferase gene was generated (p35-1116). The p35-1116 reporter was transiently transfected into P19 cells that were differentiated into neuronal cells with RA or control undifferentiated cells treated with Me 2 SO, and the activity of the promoter was determined by examining the relative luciferase activity. The p35-1116 luciferase reporter was Ϸ25-fold more active in the neuronal P19 cells compared with the control cells (Fig. 3B). Furthermore, the activity of the p35 promoter was not induced in P19 cells that were treated with RA for just 24 h, a time that was sufficient to activate a reporter construct dependent on a retinoic acid receptor-response element (data not shown). Thus, the activity of the p35 promoter is induced once the P19 cells have differentiated into post-mitotic neurons, which mimics the expression of endogenous p35 mRNA. This result strongly suggests that transcriptional mechanisms are important for the neuronal specific expression of p35 mRNA and provides an assay to identify the key DNA elements necessary.
Sequences between Ϫ155 and Ϫ95 Are Required for High Levels of p35 Promoter Activity in Neurons-To identify the regions of the promoter that contribute to the neuronal specific expression of p35, reporter constructs containing 5Ј-truncations of the p35 promoter fused to luciferase were generated. The activity of these reporter constructs was determined following transient transfection into control and neuronal P19 cells. Deletion of the p35 promoter from Ϫ1116 to Ϫ889 reduced the promoter activity about 2-fold in neurons, and further deletions of the promoter to Ϫ265 continued to show similar levels of activity (Fig. 3B). However, deletion of the promoter to Ϫ65 reduced promoter activity about 30-fold compared with the activity of the full-length reporter construct (Fig. 3B), suggesting that there are DNA elements between Ϫ265 and Ϫ65 important for high levels of p35 promoter activity in neurons. It is noteworthy that none of the deletion mutations increased p35 promoter activity in the undifferentiated control P19 cells. This suggests that positive regulatory elements within the p35 promoter, rather than a discreet repressive element, are responsible for cell type specificity.
To localize further the DNA elements between Ϫ265 and Ϫ65 responsible for high levels of expression in neurons, reporter constructs containing 5Ј-truncations of the promoter between Ϫ265 and Ϫ35 were generated and assayed. Reporter constructs containing deletions of the promoter up to Ϫ155 showed similar activity to the Ϫ265 construct; however, deletion up to Ϫ95 reduced promoter activity 10-fold (Fig. 3C). Reporter constructs containing deletions up to Ϫ35 did not reduce promoter activity any further. Reporter constructs containing internal deletions of the p35 promoter were also generated to confirm the importance of the promoter region identified through the 5Ј-deletion constructs. An internal deletion between Ϫ35 and Ϫ76 increased promoter activity in both control and neuronal cells about 2-fold (Fig. 3D); however, an internal deletion from Ϫ35 to Ϫ265 reduced promoter activity more than 5-fold (Fig. 3D). Thus, these data show that a small region of the p35 promoter is sufficient for high levels of activity in neuronal P19 cells and suggests that the region from Ϫ155 to Ϫ95 is required for promoter activity.
A GC-rich Repeated 17-bp Element Is Important for p35 Promoter Activity-To map further the DNA elements required for p35 promoter activity, 6-bp mutations spanning Ϫ155 to Ϫ45 were generated, and their activity was determined. None of the 6-bp mutations reduced promoter activity to the level observed in the Ϫ95 deletion (data not shown), suggesting that several elements were important for promoter activity. The mutations that reduced promoter activity the most were Ϫ133/ Ϫ128 and Ϫ124/Ϫ119 (Fig. 4A), which showed approximately a 2-fold reduction in promoter activity (data not shown). Examination of the promoter sequence revealed that these mutations were both within a 17-bp GC-rich element that is exactly repeated 50 bp downstream (Fig. 4A). We therefore considered the possibility that the two elements were functionally redundant and that in order to see a significant reduction in promoter activity both elements would need to be mutated. Hence, reporter constructs containing 11-bp mutations of the 17-bp elements individually and in combination were generated, and their activity was determined. Mutation of the 5Ј 17-bp element alone reduced promoter activity about 2-fold (Fig. 4B, m17.1) similar to the reduction seen with the 6-bp mutations 133/Ϫ128 and Ϫ124/Ϫ119 (data not shown). Although mutation of the 3Ј 17-bp element alone reduced promoter activity only slightly (Fig. 4B, m17.2), a reporter construct containing mutations within both 17-bp elements reduced promoter activity about 5-fold (Fig. 4B, m17.1 m17.2). Thus, two identical 17-bp elements within the p35 promoter appear to be required for high levels of activity in the neuronal P19 cells. Interestingly, the 3Ј 17-bp element lies within a region that is highly conserved between mouse and human p35 promoters.
By having determined that the 17-bp elements were necessary for high levels of expression in the P19 neurons, we wished to determine whether this element contributed directly to cell type specificity. Hence, the activity of reporter constructs con-taining either a consensus TATA alone or three copies of the 17-bp element cloned upstream of a consensus TATA were examined. The TATA alone showed very low levels of promoter activity in both the control cells and P19 neurons; however, the reporter construct containing the 17-bp elements showed increased promoter activity and was ϳ80-fold more active in P19 neurons (Fig. 4C). Thus, the 17-bp element appears to be sufficient for cell type-specific expression.
The 17-bp Element Is Required for p35 Promoter Activity in Primary Neurons-The 17-bp element was identified as being required for high levels of neuronal specific p35 promoter activity using the P19 cells as a model system; hence, we verified these findings in primary cortical neurons. Deletion of the p35 promoter from Ϫ155 to Ϫ65 almost eliminated activity of the promoter (Fig. 4D) indicating that this region is essential for high levels of activity, consistent with the results obtained in the P19-derived neurons. Furthermore, mutations within either of the 17-bp elements reduced promoter activity ϳ3-fold compared with the wild type Ϫ155 construct, and a reporter construct containing mutations within both 17-bp elements reduced promoter activity 30-fold, almost abolishing promoter activity (Fig. 4D). Thus, the 17-bp elements appear to be required for p35 promoter activity in primary cortical neurons. In addition, the reporter construct containing three copies of the 17-bp element upstream of a consensus TATA was 20 times more active in the primary cortical neurons compared with the FIG. 3. p35 promoter sequences between ؊155 and ؊95 are primarily responsible for high levels of activity in P19 neurons. A, neuronal P19 cells express increased levels of endogenous p35 mRNA. Northern blot analysis of total RNA was isolated from postnatal day 18 mouse brain, undifferentiated P19 cells, P19 cells treated with RA for 2 days, and P19 cells treated with RA for 4 days then further differentiated for 2 days into cultures containing phenotypically neuronal cells. The blot was hybridized with radiolabeled probes specific for p35 or GAPDH transcripts. GAPDH hybridization serves as a control for RNA loading. B, the p35 promoter from Ϫ265 to Ϫ65 is required for high levels of activity in neuronal P19 cells. Relative luciferase activity of reporter constructs containing 5Ј-truncations of the p35 promoter with the indicated end points co-transfected with the pRL-tk reporter into control or neuronal P19 cells are presented. C, a small region of the p35 promoter is sufficient for activity in neuronal P19 cells. Luciferase reporter constructs containing 5Ј-truncations of the p35 promoter with the indicated end points were co-transfected with the pRL-tk reporter into control or neuronal P19 cells. The relative luciferase activities are shown. D, internal deletions of the p35 promoter between Ϫ265 and Ϫ35 reduce p35 promoter activity in neuronal P19 cells. Relative luciferase activity of reporter constructs containing internal deletions lacking the indicated regions of the p35 promoter were co-transfected with the pRL-tk reporter into control or neuronal P19 cells. In each set of experiments the relative luciferase activity of the longest construct in the neuronal P19 cells was set at 100. The experiments were performed in triplicate; error bars indicate the S.D.
TATA alone (Fig. 4E). Thus, these data show that the 17-bp element is necessary and sufficient for promoter activity in primary cortical neurons.

A GC Box within the 17-bp Element Is Critical for p35 Promoter Activity and DNA-Protein Complex Formation in Neu-
rons-The 17-bp GC-rich element contains putative binding sites for the Egr (NGFI-A) family of transcription factors and also for GC box-binding factors (Fig. 5A). To determine the importance of these transcription factor-binding sites for the neuronal expression of p35, reporter constructs were assayed that contained 2-or 3-bp mutations that scanned the 5Ј 17-bp element (Fig. 5A). Because of the functional redundancy of the two 17-bp elements, these mutations were made in combination with an 11-bp mutation of the 3Ј 17-bp element (m17.2). The reporter constructs containing mutants m1, m2, and m6 showed similar levels of promoter activity to the control containing the wild type 17-bp element at this position (m17.2) (Fig. 5B). In contrast, the m3, m4, and m5 mutations reduced promoter activity 2-5-fold in the P19 neurons compared with the control (Fig. 5B). Importantly, when these mutant reporter constructs were assayed in primary cortical neurons, it was also found that only the m3, m4, and m5 mutants reduced promoter activity (data not shown). These data identify 9 bps within the 17-bp element that are critical for p35 promoter activity in both P19 neurons and primary cortical neurons. These data also show that mutation of several bases that disrupt the consensus Egr site (m1 and m2) do not affect p35 promoter activity, suggesting that this site is not required for the expression of p35 in P19 neurons or primary cortical cells. The putative GC box, however, does appear to be required for the neuronal expression of p35 as mutation of several bases within this site (m3, m4, and m5) reduced p35 promoter activity in neurons.
To examine the protein(s) binding to the 17-bp element, EMSAs were performed. Two major DNA-protein complexes were observed when a 36-bp radiolabeled probe, BS.1, that contains the 5Ј 17-bp element and flanking sequence was mixed with nuclear extract from the P19 cells (Fig. 5) as well as primary cortical cells (Fig. 6). The two complexes were competed using unlabeled BS.1, BS.2 (which contains the 3Ј 17-bp The relative luciferase activity of the p35-155 construct in the neuronal P19 cells was set at 100. C, the 17-bp element is sufficient for high levels of activity in neuronal P19 cells. Luciferase reporter construct containing 3 copies of the 17-bp element fused upstream of a TATA box (3 ϫ 17-mer-TATA) or a TATA box (TATA) alone were co-transfected with pRL-tk into control and neuronal P19 cells. D, the 17-bp elements are required for high levels of promoter activity in primary cortical neurons. Luciferase reporter constructs containing 11-bp mutations of the 17-bp elements in the Ϫ155 p35 promoter were co-transfected with pRL-tk into primary cortical neurons isolated from E18 rat embryos. The relative luciferase activity of the p35-155 construct was set at 100. E, the 17-bp element is sufficient for high levels of activity in primary cortical neurons. Luciferase reporter construct containing 3 copies of the 17-bp element fused upstream of a TATA box (3 ϫ 17-mer-TATA) or a TATA box (TATA) alone were co-transfected with pRL-tk into primary cortical neurons isolated from E18 rat embryos. The experiments were performed in triplicate; error bars indicate the S.D. element and flanking sequence), 17-mer (which contains the 17-bp element and only 6 bps of flanking sequence), and Sp1 (which contains a consensus GC box-binding site) oligonucleotides but not by oligonucleotides containing consensus binding sites for AP2, Egr, or NRF-1 transcription factors (Fig. 5C). Hence, these data show that the two DNA-protein complexes observed in nuclear extracts from neurons are specific for the GC box within the 17-bp element.
We next determined whether the sequences in the 17-bp element required for p35 promoter activity in neurons were also important for DNA-protein complex formation. Oligonucleotides containing the same 3-bp mutations of the 17-bp element that were used in the promoter activity assays (Fig.  5A) were used as competitors in DNA-protein binding assays. As shown in Fig. 5D, the m1 and m2 mutant oligonucleotides competed for binding to BS.1 with similar efficiency to the oligonucleotides containing the wild type 17-bp element (BS1.1 and 17-mer). In contrast the m3, m4, and m5 mutant oligonucleotides were not able to compete significantly for binding to BS.1 (Fig. 5D). Similar data were observed in competition experiments using nuclear extract from primary cortical cells (data not shown). Thus, these data reveal that protein binding to the 17-bp element correlates with p35 promoter activity and is dependent upon a GC box.  m1, m2, m3, m4, and m5) were added to the binding reaction. 43). Western blot analysis with specific antibodies against Sp1, Sp3, and Sp4 showed that all three Sp proteins were present in nuclear extract from primary cortical cells (Fig. 6B). The anti-Sp3 antibody detected three isoforms of the Sp3 protein (Fig.  6B) which has been observed previously (44). To determine whether the Sp proteins were binding to the 17-bp element in neurons, specific antibodies against the Sp proteins were added to the DNA-binding reaction using nuclear extract from primary cortical cells (Fig. 6A). Anti-Sp4 antibody specifically reduced the intensity of the slowest migrating DNA-protein FIG. 6. Sp1, Sp3, and Sp4 are components of the GC box binding activity in neurons. A, Sp1, Sp3, and Sp4 are components of the GC box binding activity in neurons. 2 g of anti-Sp1, anti-Sp3, and/or anti-Sp4 antibodies were incubated with 3 g of nuclear extract from primary cortical cells before performing DNA-binding reactions with radiolabeled BS.1. The proteins of the two DNA-protein complexes formed with the 17-bp element are indicated. B, all three Sp proteins are present in nuclear extract from primary cortical cells. Western blot analysis of 10 g of nuclear extract isolated from primary cortical cells probed with anti-Sp1, anti-Sp3, and anti-Sp4 antibodies. The position and size of the proteins recognized by the specific antibodies are indicated. C, the relative contribution of Sp protein binding to the GC box changes during P19 neuronal differentiation. 2 g of anti-Sp1, anti-Sp3, and/or anti-Sp4 antibodies was incubated with 3 g of nuclear extract from P19 neurons or control cells before performing DNA-binding reactions with radiolabeled BS.1. D, the ratio of Sp proteins in the nuclear extracts is cell type-specific. Western blot analysis of 20 g of nuclear extracts isolated from P19 neurons or control cells probed with the indicated Sp antibodies. The asterisk indicates Sp1 protein recognized from a previous probing of the blot with anti-Sp1 antibody. complex (band I) indicating that the Sp4 protein is a component of this complex. This complex was further reduced following the addition of anti-Sp1 antibody, suggesting that Sp1 also binds to the 17-bp element in neurons (band I, Fig. 6A). The faster migrating DNA-binding protein complex was completely depleted by anti-Sp3 antibody (band II, Fig. 6A). Hence, all three Sp proteins bind to the 17-bp element, although Sp3 and Sp4 are the major components of the DNA-protein complexes in primary neurons. These data suggest that the Sp transcription factors may be involved in regulating p35 promoter activity in neurons.
We next examined whether there were any changes in the protein levels or DNA binding activities of the Sp transcription factors during neuronal differentiation. Western blot analysis of nuclear extracts confirmed that all three Sp transcription factors were present in both P19 neurons and control cells. However, the relative levels of the Sp proteins changed during neuronal differentiation. Sp4 levels were significantly higher and Sp1 levels lower in the P19 neurons compared with control cells, although there were similar amounts of the three Sp3 isoforms (Fig. 6D). DNA-binding assays showed that the relative levels of Sp proteins present in the nuclear extracts correlated with the relative contribution to the DNA-protein complexes formed with the 17-bp element (Fig. 6C). Thus, there was higher levels of Sp1 binding in control cells than in P19 neurons as revealed by the relative intensity of band I, which was completely depleted by anti-Sp1 antibody in non-neuronal cells (compare band I in P19 control cells and neurons; Fig. 6C). Furthermore, antibodies against both Sp1 and Sp4 were required to deplete band I in P19 neurons but not in control cells, indicating increased contribution of Sp4 to binding to the 17-bp element in neuronal cells (band I, Fig. 6C). A similar level of the fastest migrating complex (band II) was formed in P19 neurons and control cells, and the complex was specifically depleted by the addition of anti-Sp3 antibody (band II, Fig. 6C). Hence, these data show that the relative contribution of Sp1, Sp3, and Sp4 to binding of the 17-bp element correlates with changes in protein levels during neuronal differentiation.
The Transcriptional Activity of the Sp Proteins Is Higher in P19 Neurons-To address further how the Sp factors could contribute to the expression of a neuronal specific gene, the transcriptional activity of the proteins was examined. Plasmids expressing fusions of Sp1, Sp3, and Sp4 to the Gal4 DNAbinding domain were co-transfected with a luciferase reporter regulated by five Gal4-binding sites (G5-luc). Co-transfection of the Sp1, Sp3, and Sp4 fusion proteins activated the G5-luc reporter to much higher levels in P19 neurons compared with control P19 cells (Fig. 7). In contrast the VP16 and AP2 fusions showed only modest increases in activity in the P19 neurons (Fig. 7). These data show that Sp1, Sp3, and Sp4 each have higher transcriptional activity in neurons, and hence regulated activity of these proteins may contribute to the cell type-specific activity of the p35 promoter.

DISCUSSION
In this study we have performed a detailed analysis of the mouse p35 promoter to identify the DNA elements and transcription factors that regulate expression of this cdk5 activator in post-mitotic neurons. Analysis of deletion derivatives shows that a 200-bp fragment of this TATA-less promoter is sufficient for high levels of neuronal specific activity. Our data suggest that the cell type-specific expression of p35 is not regulated by a discrete repressive element, such as the neuronal restrictive silencer element that has been described to regulate the cell type specificity of several other neuronal genes (45)(46)(47). Rather, we have found that the cell type-specific expression of p35 is determined by positive regulatory elements that are preferentially active in neurons. We have identified a 17-bp repeated element within this region of the p35 promoter that is both required and sufficient for high levels of promoter activity in P19 and primary cortical neurons. A GC box within the 17-bp element is critical for both p35 promoter activity and DNA-protein complex formation. The related transcription factors Sp1, Sp3, and Sp4 comprise most of the GC box binding activity in neurons. During neuronal differentiation of P19 cells, the relative contributions of individual Sp family proteins to the GC box binding activity as well as their transcriptional activity is regulated. Our findings suggest that changes in Sp protein ratios and activity contribute to the neuronal specific activity of the p35 promoter.
Although additional sequence elements are necessary for maximal expression of p35, our analysis has revealed a critical role for the repeated GC box in regulating p35 expression in post-mitotic neurons. Like the p35 promoter, many neuronal specific promoters are GC-rich and lack apparent TATA boxes. GC box elements are important for high level expression of many genes in neurons, such as Tau (48) and FE65 (49). Strikingly, we have found that the 17-bp element from the p35 promoter is preferentially active in neurons, and mutation of the GC box in the 17-bp element reduced both promoter activity and DNA-protein complex formation in neurons. Thus, our studies suggest that the GC box contributes to not only controlling the level of p35 expression but also for regulating the cell type specificity. GC boxes are widely distributed promoter elements that regulate genes expressed in many different cell types; thus, it may seem surprising that such a common promoter element contributes to the neuronal specific expression of p35. It should be noted, however, that in addition to important roles in basal expression of many promoters, GC box elements have also been shown to be required for induction of transcription in response to diverse stimuli. For example, GC boxes have been shown to be required for the transcriptional response to nerve growth factor in both the p21 and ␤4 nicotinic acetylcholine receptor promoters (50 -52). In addition mutation of Sp1-binding sites in the neuronal specific glycine receptor ␤ promoter was found to reduce preferentially neuronal activity (53). Thus, GC boxes and their corresponding binding proteins may play an important role in regulating both the level of expression and also the cell type specificity of many genes in the nervous system.
The Sp family of transcription factors including Sp1, Sp3, FIG. 7. The activity of the Sp transcription factors increases in the P19 neurons. The transcriptional activity of Gal4-Sp1, Gal4-Sp3, and Gal4-Sp4 is higher in P19 neurons. P19 neurons or control cells were co-transfected with a Gal4-luciferase reporter plasmid (G5-luc), expression plasmids for the indicated Gal4 fusion proteins, and the pRL-tk reporter plasmid. The fold luciferase activation represents the mean relative luciferase activity of G5-luc in the P19 neurons compared with the P19 controls. The experiments were performed in triplicate; error bars indicate the maximal and minimal fold increase in activity observed. and Sp4 are the most well characterized GC box-binding proteins. We have found that Sp1, Sp3, and Sp4 all bind to the 17-bp element in the p35 promoter. Although we cannot rule out a role for other GC box-binding transcription factors, Sp3 and Sp4 were the major components of the protein-DNA complexes observed in extracts of primary cortical cells, suggesting that these factors are of primary importance for the activity of the GC box in neurons. Sp1, Sp3, and Sp4 are highly related proteins that recognize GC boxes with similar affinities dependent on three highly conserved zinc finger motifs in the carboxyl terminus (42). In addition, each of these Sp family proteins contains two glutamine-rich regions in the amino terminus that contribute to transcriptional activation (38,54,55). Sp3, unlike Sp1 and Sp4, also contains an inhibitory domain that suppresses transcriptional activation and may contribute to active repression (54,56). Studies in cell culture as well as analysis of knockout mice support the idea that Sp1, Sp3, and Sp4 have some unique activities, although there is likely to be significant functional redundancy between these Sp family proteins (43,(57)(58)(59). Our findings suggest that regulated changes in the ratios of Sp family proteins as well as changes in the transcriptional activity of Sp1, Sp3, and Sp4 contribute to the enhanced expression of p35 in neurons.
The relative contribution of Sp family transcription factors to the GC box binding activity was observed to change during neuronal differentiation of P19 cells. In contrast to the control cells, in P19 neurons there was a higher ratio of Sp3 and Sp4 binding to the GC box compared with Sp1 binding, similar to what was observed in primary neurons. The ratios of Sp proteins present in the neuronal versus non-neuronal cells correlated well with their relative contributions to the GC box binding activity. Although Sp1 and Sp3 are ubiquitously expressed, Northern and in situ analyses have shown that Sp4 mRNA levels are highest in the central nervous system and the brain (42,43). Consistent with this, we have observed higher levels of Sp4 protein in P19 neurons compared with undifferentiated P19 cells. Gene disruption of Sp4 in mice produces a complex phenotype that includes behavioral defects, further supporting a role for Sp4 in regulating transcription of genes in the nervous system (43,58). Although the relative amount of Sp4 protein may be an important component of the GC box activity in neurons, transfection of P19 cells with an Sp4 expression construct was not sufficient to induce the p35 promoter. 2 The ratios of Sp family proteins differ between different cell types, and reminiscent of our findings in neurons, the ratio of Sp1 to Sp3 was found to change during differentiation in keratinocytes (60,61). The ratio of Sp proteins has been suggested to be important for differential gene regulation, because these proteins compete for binding to common promoter sites, although their transcriptional activity and protein-protein interactions differ (62). Thus, the relative expression of Sp family proteins in neurons may contribute to the proper regulation of the p35 promoter.
In addition to the change in ratios of the Sp family proteins, we have found that the transcriptional activity of Sp1, Sp3, and Sp4 is higher in neuronal cells. This effect is specific as the activation function of other factors such as VP16 or AP2 was not significantly altered during neuronal differentiation of P19 cells. Post-translational modifications and interactions with other factors have been shown to regulate Sp1 DNA binding and transcriptional activity (50,52,(63)(64)(65). The transcriptional properties of Sp3 are complex (44,54,66), as Sp3 can function as either a transcriptional activator (67)(68)(69) or a repressor (56, 70 -73). Interestingly, whether Sp3 functions as an activator or a repressor depends on promoter architecture (54,66,73). Our studies extend previous work (74) to clearly show that Sp3 activity is also dependent on cell type. Interestingly, the transcriptional activity of Sp1 and Sp3 is higher in PC12 cells that have been induced to differentiate with nerve growth factor (52), suggesting that the transcriptional activity of Sp1 and Sp3 is up-regulated in many types of post-mitotic neurons. Sp4 has been shown to be a transcriptional activator, although the regulation of its activity has not been described previously. Although Sp1, Sp3, and Sp4 are highly related proteins, they have distinct biological roles; hence it will be of interest to determine whether distinct or common mechanisms ensure their increased activity in neurons. Regardless of the molecular mechanisms, the increased activity of these GC box-binding proteins in post-mitotic neurons is likely to be an important component of the cell type-specific activity of the p35 promoter.
Although the repeated GC box is critical for high level expression in neurons, additional sequences in the p35 promoter are likely to contribute to the dynamic regulation of this cdk5 activator during development of the central nervous system. For example, the repeated 17-bp element in the p35 promoter contains not only the functionally important GC box, but also a consensus Egr (early growth response) (NGFI-A)-binding site. The Egr family of transcription factors has been implicated in regulating changes in gene expression following neurotransmitter stimulation or depolarization of neurons (75,76). Hence, although the predicted Egr-binding sites are not required for the expression of p35 in P19 or primary cortical neurons, the Egr transcription factors may regulate the expression of p35 in response to neuronal stimulation. Indeed, a recent study (77) suggests that an extracellular signal-regulated kinase-dependent induction of Egr1 may contribute to increased expression of p35 in response to nerve growth factor. There are also several regions of high homology between the mouse and human p35 promoters, including the 26-bp 100% conserved region encompassing the transcription start site in the mouse gene. These additional promoter elements, although not required for expression in cortical neurons, may contribute to the dynamic spatial and temporal patterns of p35 expression in the brain.
Activity of the cdk5/p35 kinase is highest in the central nervous system largely due to cell type-specific expression of p35 mRNA in post-mitotic neurons. Our studies indicate that proteins binding to two critical GC boxes regulate the levels and cell type specificity of p35 transcription. There is a pair of similarly positioned GC boxes upstream of the human p35 gene, supporting the idea that the transcription factors and mechanisms that regulate p35 expression in neurons will be conserved. The process of neuronal differentiation is complex, and it is notable that p35 is expressed in terminally differentiated, post-mitotic neurons. In this regard, it is interesting to note that several studies, including an analysis of the GC box in the p21 promoter that is important for induction during terminal differentiation in both neurons and keratinocytes (50,52,78), suggest that GC boxes and their corresponding binding proteins may play an important role in regulating transcription in post-mitotic cells. Our studies suggest that the enhanced activity of the GC box in post-mitotic neurons is achieved through cell type-specific mechanisms affecting both the relative levels of the Sp1, Sp3, and Sp4 transcription factors and their transcriptional activity. The finding that the GC box binding Sp family proteins contribute to the preferential activity of the p35 promoter in post-mitotic neurons suggests an important role for these transcription factors in regulating gene expression in the central nervous system.