Transcriptional Regulation of the Human Neuropeptide Y Gene by Nerve Growth Factor*

Neuropeptide Y (NPY) is the most ubiquitously ex- pressed peptide in the mammalian nervous system. Transcription of the NPY gene in PC12 cells is regulated by a number of agents, including the neurotrophic peptide nerve growth factor (NGF). In this paper, we define the cis-acting promoter elements which respond to NGF and characterize the trans-acting factors which interact with these sequences. The NGF-responsive elements of the NPY gene lie between nucleotides -87 and -36. At least four proteins interact with this promoter region. One of these proteins interacts with a CT-rich sequence centered at position -51, which closely abuts a binding site for transcription factor AP-2 centered at position -63. Two newly characterized factors bind between po- sitions -87 and -58. These proteins are expressed in a tissue-specific manner and, together with the other binding activities, modulate the transcriptional activity of the NPY gene. These results suggest that the con-certed interplay of these proteins, in response to NGF, increases the transcriptional activity of the NPY gene. Nerve growth factor (NGF)l is the best characterized member of a family of neurotrophins. These molecules promote the growth and differentiation of distinct neuronal populations. NGF is required for the survival of sympathetic neurons in the peripheral nervous system and cholinergic neurons in the cen-tral nervous system (1). The actions of NGF have been exten-sively studied in rat pheochromocytoma cells (PC12), a popu-lation of cells derived from neoplastic adrenal chromaffin cells a bipotent developmental (2). following modifications. Fifteen micrograms of NPY/CAT fusion constructions plus 5 pg of pRSVLacZ, con- taining the coding sequence for P-galactosidase under the control of the Rous sarcoma virus promoter, were precipitated per 100-mm dish at approximately 60% confluence. The plates were incubated in a 4% CO, environment for 16-24 h and glycerol shocked for 2 min using DMEM containing 20% glycerol. Subsequently, they were rinsed twice with DMEM, refed with growth media plus or minus 100 ng/ml NGF, and allowed to recover for 48 h. The cells were harvested by scraping, pel-leted, and cellular extract was prepared by three freeze-thaw cycles (dry id37 "C). Aliquots of the extracts used for CAT analyses were heated for 15 min at 65 "C to remove endogenous deacetylating activities (16). CAT assays were performed as described by Gorman et al. (17) or by Nielson et al. (18). P-Galactosidase activity was assayed by monitoring the conversion of o-nitrophenyl-P-D-galactopyranoside to galactose and o-nitrophenyl at A4,0nm. CAT activities for the NGF-treated samples were normalized for transfection efficiency by multiplying by the ratio of the sample's Pgalactosidase activity divided by the groups average P-galactosidase activity. Samples not treated with NGF were similarly analyzed. This normalization was necessary because NGF treatment increased the expression directed by pRSVLacZ approximately 2-fold.

pressed peptide in the mammalian nervous system. Transcription of the NPY gene in PC12 cells is regulated by a number of agents, including the neurotrophic peptide nerve growth factor (NGF). In this paper, we define the cis-acting promoter elements which respond to NGF and characterize the trans-acting factors which interact with these sequences. The NGF-responsive elements of the NPY gene lie between nucleotides -87 and -36. At least four proteins interact with this promoter region. One of these proteins interacts with a CT-rich sequence centered at position -51, which closely abuts a binding site for transcription factor AP-2 centered at position -63. Two newly characterized factors bind between positions -87 and -58. These proteins are expressed in a tissue-specific manner and, together with the other binding activities, modulate the transcriptional activity of the NPY gene. These results suggest that the concerted interplay of these proteins, in response to NGF, increases the transcriptional activity of the NPY gene.
Nerve growth factor (NGF)l is the best characterized member of a family of neurotrophins. These molecules promote the growth and differentiation of distinct neuronal populations. NGF is required for the survival of sympathetic neurons in the peripheral nervous system and cholinergic neurons in the central nervous system (1). The actions of NGF have been extensively studied in rat pheochromocytoma cells (PC12), a population of cells derived from neoplastic adrenal chromaffin cells which display a bipotent developmental potential (2). These cells closely resemble their non-tumor counterparts and, when treated with glucocorticoids, continue to develop into adrenal chromaffin-like cells. However, a divergent developmental fate occurs upon exposure to NGF. Over a period of a few days, the NGF-treated cells stop dividing and mature into cells that exhibit characteristics reminiscent of sympathetic neurons, including neurite outgrowth and electrical excitability (3).
Part of the biochemical basis underlying the differentiation into these neuronal-like cells has been elucidated and shown to involve a cascade of events initiated by NGF. Characterization of the NGF receptor has shown that it is composed of two disparate proteins, one of which is the product of the trk protooncogene (4,5). When NGF binds to its receptor, trk's intrinsic tyrosine kinase activity leads to the activation of several signal-* This work was supported by National Institutes of Health NINDS Grant NSZ8496. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Neuropeptide Y is one of the genes whose transcription is increased in the intermediate wave of NGF-stimulated gene expression (10). This peptide hormoneheurotransmitter is expressed in a wide variety of neuronal tissues, including sympathetic neurons and cells of the adrenal medulla (11). PC12 cells produce very low levels of NPY mRNA prior to treatment with NGF. Previous studies have shown that treatment of these cells with NGF causes a maximal 60-fold increase in the steady state levels of NPY mRNA within 24 h. Increased transcriptional activity accounts for only 2-7-fold of this increase, occurring within 6-8 h of treatment (12). The remainder of this response is presumably due to increased mRNA stability. Dual control of NPY mRNA levels is also manifested by LA-N-5 and LA-N-1 neuroblastoma cells upon treatment with 12-0-tetradecanoylphorbol-12-acetate and forskolin.2 Increasing the stability of the mRNA allows for rapid increases in NPY production. This may be particularly important under conditions of intense neuronal stimulation. However, since many different stimuli can impinge upon a single neuron, it is likely that different signaling pathways may selectively regulate transcription while others regulate mRNA stability.
NPY's predominantly neuronal expression pattern and NGF responsiveness make it a good candidate for elucidating the biochemical mechanisms responsible for the development of a neuronal phenotype. We have undertaken a study to determine the mechanisms responsible for the NGF elicited increase in the transcription of the NPY gene in PC12 cells. We report the characterization of cis-acting sequences and their corresponding trans-acting factors which participate in this developmental response. These results are a first step toward a better understanding of how differential gene expression occurs during neuronal development.
MATERIALS AND METHODS Construction of Recombinant Plasmids-NPYICAT chimeras were made by fusing varying amounts of the NPY gene 5'-flanking sequence to the coding sequences for chloramphenicol acetyltransferase as described previously (13). The number following the A indicates the end of the 5'-flanking sequence with respect to the transcriptional start start (designated +1) which was determined by in vitro transcription analyses and primer extension analyses (13,14). In addition to the specified amounts of 5"flanking sequence, these constructs contained the start of transcription and 51 base pairs of 5'-untranslated leader sequence.
Plasmids for analysis of individual NPY gene elements were constructed with the minimal Herpes simplex virus thymidine kinase (TK) promoter (-105 to +51) directing expression of the CAT gene (TKCat, Ref. 15). NPY elements (-87/-47) were created by annealing comple-C. Minth-Worby, manuscript in preparation. mentary synthetic oligonucleotides to generate BarnHI and HindIII ends. These double-stranded molecules were ligated into the TKCat vector digested with BarnHI and HindIII. The TWNPY fusions containing specific mutations were constructed as described above using the following mutated oligonucleotides (top strand).
Cell Culture and Dansient Dansfection Analyses-PC12 cells (kindly donated by Dr. Gordon Guroff, National Institutes of Health) were maintained in Dulbecco's modified Eagle's medium (DMEM) plus 10% horse serum and 5% fetal calf serum in a humidified 8% CO, atmosphere. The cells were routinely divided 1:4 onto collagen-coated dishes prior to transfection. Only cell passages 6 through 16 were used in these experiments because older cells appeared to have an altered phenotype and a diminished response to NGF (generously provided by Dr. Steve Sabol, National Institutes of Health).
Transient transfections were done using the Transfinity kit (Life Technologies, Inc.) with the following modifications. Fifteen micrograms of NPY/CAT fusion constructions plus 5 pg of pRSVLacZ, containing the coding sequence for P-galactosidase under the control of the Rous sarcoma virus promoter, were precipitated per 100-mm dish at approximately 60% confluence. The plates were incubated in a 4% CO, environment for 16-24 h and glycerol shocked for 2 min using DMEM containing 20% glycerol. Subsequently, they were rinsed twice with DMEM, refed with growth media plus or minus 100 ng/ml NGF, and allowed to recover for 48 h. The cells were harvested by scraping, pelleted, and cellular extract was prepared by three freeze-thaw cycles (dry i d 3 7 "C). Aliquots of the extracts used for CAT analyses were heated for 15 min a t 65 "C to remove endogenous deacetylating activities (16). CAT assays were performed as described by Gorman et al. . P-Galactosidase activity was assayed by monitoring the conversion of o-nitrophenyl-P-D-galactopyranoside to galactose and o-nitrophenyl a t A4,0nm. CAT activities for the NGF-treated samples were normalized for transfection efficiency by multiplying by the ratio of the sample's Pgalactosidase activity divided by the groups average P-galactosidase activity. Samples not treated with NGF were similarly analyzed. This normalization was necessary because NGF treatment increased the expression directed by pRSVLacZ approximately 2-fold.
Nuclear Extract Preparation and Gel Retardation Analyses-PC12 nuclear extracts from nonstimulated and NGF-stimulated cells were prepared as described by Dignam et al. (19). Protein concentrations were determined using the BCAprotein assay kit (Pierce Chemical Co.).

FIG. 2. Gel retardation analysis of the NPY promoter between
The Spl-like binding site (CT box) is boxed, and nucleotide changes -66 and -36.A, the sequence of the NPY promoter between -65 to -36. present in the mutated oligonucleotide listed to the right are noted above. Putative AP-2 binding sites are underlined. B, lanes 1 and 10 represent the gel retardation pattern obtained with PC12 extract and the oligonucleotide -651-36 when no competitor DNA is present. Lane 2 is the pattern obtained from extract prepared from PC12 cells which were treated with 100 ng/ml NGF for 8 h prior to extract preparation. Competitor DNAs (0.1 pg) present in lanes 3-9 are listed above the lanes in which they appear. Lunes 11 and 12 represent the pattern generated when recombinant purified proteins AP-2 and Spl are allowed to interact with the -65 to -36 oligonucleotide. The three major complexes formed with PC12 cell extract are designated S l S 3 .
Purified recombinant Spl and AP-2 were purchased from Promega, Madison, WI.
Methylation Interference Footprinting-Methylation interference footprinting was performed with a double-stranded oligonucleotide spanning the -87 to -58 region of the human NPY gene. Oligonucleotides for each strand were separately labeled with [y-32PlATP using T4 polynucleotide kinase. Following inactivation of the kinase by heating for 5 min a t 90 "C, the labeled oligonucleotides were annealed to their unlabeled counterparts, and double-stranded complexes were purified after electrophoresis in nondenaturing 10% polyacrylamide gels (acrylamide:bisacrylamide, 50:l). Methylation interference footprinting was performed essentially as described by Andrisani et al. (20). Cleavage fragments resulting from piperidine treatment of the methylated DNA were loaded onto denaturing 20% polyacrylamide sequencing gels.
Maxam and Gilbert G-and G + A-specific reactions were included for orientation. Gels were exposed 24-48 h a t -80 "C using an intensifying screen.
Southwestern Analysis-Nuclear extracts were prepared as de- Filters were prehybridized overnight in the hybridization solution minus the labeled oligonucleotides, then hybridized for 3 h a t room temperature with labeled oligonucleotides. This was followed by four 30min washes a t room temperature in the hybridization buffer minus probe. Filters were exposed to film overnight a t -80 "C with an intensifying screen.
Oligonucleotides used in the construction of the TKCat fusion constructions were described above.

Determination of the Human NPY Gene's BI-Flanking Sequences Important for the Danscriptional Increase Elicited by
NGF-To define the elements important for NGF regulation, plasmid constructions containing varying amounts of the human NPY gene's 5'-flanking sequences fused to the bacterial reporter gene CAT were transiently introduced into the PC12 cell line. Analysis of the 5"flanking sequences spanning -796 to -30 indicated that sequences between -246 and -30 were sufficient to increase CAT activity 5-fold in response to NGF unless indicated. treatment (Fig. 1). These results are consistent with the NGFinduced transcriptional increase (2-7-fold) measured for the endogenous NPY gene by nuclear runoff analyses (12). Finer dissection of the region between -246 and -30 demonstrated that as little as 63 base pairs 5' to the transcriptional start site were capable of increasing CAT activity when the cells were exposed to NGF (Fig. 1). The addition of sequences between -83 and -63 increased the basal activity of the promoter as well as maximizing the NGF response. The increases for the A83 and the A63 constructs were consistently 3-and 2-fold, respectively. The A30 construct which was previously shown to be incapable of correctly initiating transcription (13) did not consistently respond to NGF. This experiment defined the NGF response element to be contained within the sequence spanning -83

NGF-induced NPY Gene Danscription
Gel Retardation Analysis-The sequence between -87 and -36 was subjected to gel retardation analyses to determine the number and nature of interacting proteins present in PC12 cell crude nuclear extracts. Due to the complexity of these interactions, this region was divided into three overlapping doublestranded oligonucleotides, -65 to -36, -75 to -46, and -87 to -58 (Figs. 2 A , 3A, and 4 A , respectively). Computer analysis (MacVector 3.5, International Biotechnologies, Inc., New Haven, CT) of these sequences for potential DNA-protein interactions revealed two putative AP-2 binding sites centered around positions -61 and -45 (Fig. 2 A ) . Previous analysis of to -30.
the region between -65 and -36 carried out in a human neuroblastoma cell line, LA-N-5, demonstrated that the CT-rich sequence designated the CT box in Fig. 2A was necessary for basal level expression. The associated protein binding activity had a specificity similar to that of transcription factor Spl (13). It is noteworthy in this regard that transcription factor AP-2 also binds to GC-rich sequences. Gel retardation analysis of this region using PC12 cell nuclear extracts is shown in Fig. 2B. Three major complexes were observed with no apparent NGFdependent differences (Fig. 2B, compare lanes 1 and 10 with 2 1. Analogous to the LA-N-5 cell experiments, the slowest mobility bands, S1 and S2 are efficiently competed by addition of an excess of unlabeled oligonucleotide -651-36 (Fig. 2B, lane 3) and an oligonucleotide containing two high affinity Spl binding sites (Fig. 2B, lane 4 ) . A GC-rich AP-2 binding oligonucleotide is also able to partially compete these complexes (Fig. 2B, lane  5). Competitors which do not contain the CT sequence or have mutations in this binding site are not able to compete for S1 and S2 binding (Fig. 2B, lanes 6, 7, and 9). The -751-46 oligonucleotide, containing the CT binding site near the 3' end (Fig.   a), displays somewhat reduced competition (Fig. 2B, lane 8).
This sequence contains not only the CT box but also a high affinity AP-2 binding site centered a t position -61. Therefore, its reduced ability to compete the S1 complex could be due either to the noncentral location of the CT binding site or to blockage of CT box protein binding to -751-46 by an AP-2-like factor which is bound. However, the inability of -75/-46M1 to compete suggests that the CT box is the sequence responsible for competition by this oligonucleotide (Fig. 2B, lane 9). Since the oligonucleotide representing the AP-2 consensus binding site is able to partially compete the S1 and S2 complexes (Fig.  2B, lane 5), we sought to determine whether AP-2 as well as Spl-like molecules are able to interact with this sequence. When purified Spl and AP-2 are used in the gel retardation analysis along with the -65 to -36 oligonucleotide, AP-2 is not able to specifically interact with this oligonucleotide (compare Fig. 2B, lane 11 with Fig. 3C, lane 4 ) and Spl is able to interact only marginally (Fig. 2B, lane 12). Furthermore, the Spl complex runs at a slower mobility than complex S1, suggesting that complex S1 is not formed by Spl. Complex S3 probably represents a nonspecific binding reaction, because it is inefficiently competed by all competitors tested. The sequences of the oligonucleotide -75/-46 and selected mutants are shown in Fig. 3A. Both the CT box binding element and an AP-2 consensus binding site are located in this sequence. Three major complexes, designated Cl-C3, are formed when this labeled oligonucleotide is incubated in the presence of PC12 nuclear extracts (Fig. 3B, lane 1 ).Although not shown, extracts prepared from cells treated with NGF display no discernible differences in complex formation. Competition with an excess of unlabeled -75/-46 reduces the intensity of Cl-C3 (Fig. 3B, lane 2 ) as does competition with consensus binding sites for AP-2, Spl, and the CT box containing oligonucleotide -65/-36 (Fig. 3B, lanes 3, 4, and 6). All oligonucleotides compete C3 to a similar degree, suggesting that it is generated by a nonspecific interaction. The Spl consensus oligonucleotide and -651-36 compete the most efficiently for complex C1 and C2 formation. Oligonucleotides which do not contain a CT sequence (-87/-58) or have a mutated version of this sequence (-65/-36M, -75/-46M1) are no longer able to efficiently compete (Fig. 3B, lanes 5, 7, and 8). Oligonucleotide -75/-46M2, which contains mutations in the AP-2 binding site, retains partial competitive ability (Fig. 3B, lane 9). Therefore, similar to the overlapping sequence (-65 to -361, -751-46 is capable of forming a DNA-protein complex involving nucleotides in the CT box. Additionally, it is capable of interacting with AP-2. The ability of this sequence to interact with purified AP-2 and Spl was determined as shown in Fig. 3C. The AP-2 binding site located in the center of the sequence binds purified AP-2 strongly while interacting only marginally with Spl. (Fig. 3C,   lanes 1 and 2). The strength of the AP-2 binding to the NF'Y sequence in comparison with the consensus oligonucleotide, defined by its interaction with the hMT-IIA gene, can be seen by comparing lanes 1 and 4. When added in concert these proteins do not form a supershifted complex (Fig. 3C, lane 3 ) . Since the identity of the CT box binding protein is currently unknown, it is not possible to ascertain whether it and AP-2 will act in concert or will bind in a mutually exclusive manner to these promoter sequences.
The sequence of the NPY promoter between -87 and -58, as well as defined binding sites and mutant oligonucleotides, are depicted in Fig. 4A. In addition this region contains putative NF1-CAAT and Ap-1 binding sites centered around -76 and -68, respectively. Gel retardation analyses of this sequence resulted in the formation of three complexes designated A-C (Fig. 4B, lanes 1,2, and 17). As noted previously, treatment of the PC12 cells with NGF 8 h prior to extract preparation did not change the gel retardation pattern obtained for this oligonucleotide (compare lanes 1 and 17 with 2). The most slowly migrating complex, designated A, is most easily competed with increasing amounts of unlabeled oligonucleotide -871-58, followed closely by complex B (Fig. 4B, lanes 3-5). Oligonucleotides containing mutations in the putative CAAT-and Ap-1-like sequences, -871-58M2 and -871-58M1, respectively, preferentially compete these two complexes (Fig. 4B, lanes 6 and 7). This suggests that distinct complexes form on this palindromic region. A battery of oligonucleotides were tested for their ability to compete specifically with complex A and B (Fig. 4B, lanes  8-16). Only the TRH and fibronectin CREs were able to compete complex A (Fig. 4B, lanes 12 and 13), respectively. However, when purified CREB was used in place of the PC12 extract, it was unable to bind, indicating that the binding activities generated by this oligonucleotide do not involve CREB (data not shown). Complex C is not efficiently competed by any of the oligonucleotides. It requires 300 ng of the AP-2 consensus oligonucleotide for partial competition and is therefore categorized as nonspecific (Fig. 4B, lane 10).
Due to the novel nature of the protein interactions occurring on this portion of the NPY promoter, additional experiments were performed to further characterize these binding activities.
Localization of the Protein Binding Sites between -87 and -58-The methylation interference assay was performed to identify the G residues of the -871-58 NPY promoter fragment that come into direct contact with the proteins forming complexes A-C. A pattern of reduced contact with methylated G residues was apparent only for complex A (Fig. 5 A ) . Three G residues on the coding strand at positions -73, -71, and -69 appeared to be involved in complex formation. On the noncoding strand the intensity of 2 G residues were reduced at positions -72 and -67 (summarized in Fig. 5B). No G residues were observed to be critical for complex B or C binding. It is possible that the binding sites for these complexes are too close to the ends of the oligonucleotide for this type of analysis or that they do not contact the DNA through analyzable G residues.
To determine the importance of the identified G residues in complex formation, specific bases were mutated within the -87 to -58 sequence (Fig. 4A) and subjected to gel retardation analysis (Fig. 5C). The gel retardation pattern produced by the oligonucleotide -871-58 without the addition of competitors is shown in lane 7 preceded by the addition of increasing amounts of unlabeled oligonucleotide (Fig. 5C, lanes 6 and 5 ) , respectively. Competitors -871-58M3 and -871-58M4, which contain mutations in the binding site defined by the methylation interference assay, are unable to effectively compete complex A (Fig.  5C, lanes 3 and 4 ) . However, complex B is differentially competed by these two sequences. The M3 but not the M4 oligonucleotide is able to partially compete this complex (Fig. 5C,  lane 3 versus 4). This evidence suggests that these complexes are generated by different proteins. Furthermore, M2, which contains mutations in the CAAT-like sequence, very efficiently competes only complex A. Conversely, M1, which contains mutations within complex A's binding site, efficiently competes only complex B.
These complexes also display tissue specificity. The gel retardation pattern in lane 8 (Fig. 5C) was generated by incubation of labeled -871-58 with HeLa cell nuclear extracts. Neither complex A nor B is formed to any appreciable extent.
Southwestern Analysis-In order to further characterize the factors binding to the sequences between -87 and -58, a Southwestern analysis was performed. The ability of this sequence to specifically interact with proteins was tested in HeLa, LA-N-5, and PC12 extracts. LA-N-5 cells are human neuroblastoma cells which highly express NPY (13). Using this technique, three bands were visualized (Fig. 6). The highest molecular weight band (A) was nonspecific. All labeled oligonucleotides produced this signal (data not shown). Band B, however, appears to be specific in that it is completely competed by cold self but not by NPY mutant oligonucleotides -871-58M1 and -871 -58M2 nor oligonucleotides representing the consensus sequences for AP-1, Spl, or NF1-CAAT. The apparent molecular mass of band B is approximately 40 kDa. It is also present in LA-Nd cell extract but not in HeLa extract, consistent with the inability of the -871-58 oligonucleotide to produce the characteristic pattern in the gel retardation analysis when HeLa extract is used. The fastest mobility band, C, is not consistently produced and is not characterized further in this report.
DNase Z Footprint Analysis-The complexity of the DNA binding activities defined by the gel retardation analyses suggested the possibility of multiple DNA-protein interactions occurring within the sequence between -87 and -46. In order to further define the combinatorial nature of these interactions, the fragment spanning -143 to +51 was labeled on the coding strand, incubated with PC12 cell nuclear extracts, and treated with DNase I (Fig. 7A). The DNA ladder obtained by treating this labeled fragment with DNase I in the absence of any extract is represented in lanes 1, 7, and 14. Consistent with the gel retardation analyses, treatment of the cells with NGF 8 h prior to extract preparation did not alter the footprinting pattern (compare lanes 2 7A, lanes 4 and 6). Mutation of the CT box in -651-36M abolishes this competition (Fig. 7A, lane 5 ) . GC-rich oligonucleotides containing consensus binding sites for AP-2, Spl, and oligonucleotides NPY -981-71 and NPY -1321-106 also partially compete this footprint (Fig. 7A, lanes 9 and 11-13). FP2 is only marginally competed by the aforementioned oligonucleotides and is not competed by unlabeled oligonucleotide -871 -58 (Fig. 7A, lane 8). The inability of oligonucleotides -871-58 and -981-71 to compete FP2 is perplexing. It is possible that more unlabeled oligonucleotide may be necessary to compete this footprint. Oligonucleotide -132/-106 is also not able to compete FP3 efficiently at these concentrations. Finally, purified AP-2 strongly footprints the NPY promoterbetween -75 and -47 which overlaps FP1 and FP2 generated by the PC12 extract (Fig. 7B, lane 3). The AP-2 footprint, however, does not match the footprint generated from nuclear extract, indicating a more complex interaction than AP-2 binding alone. Purified Spl is not able to footprint this fragment providing further evidence that the protein which interacts with the CT box is not S p l (Fig. 7B, lane 2). Furthermore, when added in concert with AP-2, S p l does not significantly alter the footprint of this region (Fig. 7B, lane 5). AP-1 is also not able to interact with these sequences and does not change the footprint generated by AP-2 (Fig. 7B, lanes 4 and 6).
Mutational Analysis of the NGF Response Element-Due to the low magnitude of the response produced by NPY fusion constructs A 8 3 and A63, coupled with the variability inherent in transient transfection analyses, it was desirable to place these sequences in front of a heterologous promoter to determine if they were capable of eliciting a transcriptional increase in a different context. When these sequences were placed 5' to the minimal TK promoter, they enabled this promoter to be more efficiently expressed in PC12 cells and conferred NGF inducibility. The region between -87 and -36 designated TK -871-36 produced around a 2.5-fold increase in CAT activity when treated with NGF (Fig. 8). This is consistent with the fold increases observed for the A83 and A63 constructs. Since the aforementioned experiments have shown that the sequence of the NPY promoter between -87 and -36 contains the binding sites for at least four proteins, we sought to ascertain which of these interactions were important for the NGF response. The binding sites designated A B , CT box, and AP-2 were mutated in the context of the NPY/TKCat fusion construct (Fig. 8). The mutations generated were shown previously to disrupt protein binding in the gel retardation analyses. When these constructs were transiently introduced into PC12 cells, the basal level of expression decreased to approximately that of TKCat (Fig. 8). Moreover, none of the individually mutated constructs were capable of responding to NGF to the level of the wild type construct. This suggests that the NGF response may be generated by the interaction of all these proteins on the NPY promoter. DISCUSSION The binding of NGF to its receptor on PC12 cells activates a cellular program leading to the differentiation of these cells into sympathetic-like neurons. Early events in the NGF cascade increase the transcription of proto-oncogenes such as c-fos and c-myc as well as that of other putative transcriptional regulatory factors (9). These proteins function in the control of other genes which ultimately serve to initiate the differentiation program. We have determined the sequences involved and partially characterized the DNA-binding proteins responsible for increasing NPY mRNA at the transcriptional level during this developmental process.
Transient transfection analyses determined that the NPY gene sequences between positions -87 and -36 were capable of increasing the activity of NPYICAT reporter when transfected PC12 cells were exposed to NGF. These same sequences were also able to confer approximately a 2.5-fold increase in the ability of the TK promoter to respond to NGF. We have therefore designated this sequence as the NGF-responsive element (NGF-RE) of the NPY gene.
Several proteins are capable of interacting with the NPY gene's NGF-RE (Fig. 9). The CT box-binding protein displays the same type of binding specificity as the transcriptional activator, Spl, and potentially represents a member of this rapidly growing family. Spl is a transcription factor whose consensus binding site is defined as the GC box and whose DNA binding domain consists of structures known as zinc fingers (24). Additionally, other proteins such as Sp2 and Sp3, which bind to GT box consensus sequences, have been characterized recently (26). These proteins were isolated based on the premise that their DNA binding domains would contain sequence homology to the Spl DNA binding domain. Additional GC sequence binding proteins include NGFI-A, NGFI-C, Krox-20, and the Wilm's tumor gene product (6,271. The proteins NGFI-(A-C) are of interest as they were isolated as early response genes of the NGF cascade. Since purified S p l is not able to footprint the CT box, it is likely that this binding activity is either another one of these family members or represents a novel addition to this family.
Immediately upstream of the CT-rich sequence (NPY 5'flanking sequences -65 to -57) is a consensus AP-2 binding site (Fig. 9). The AP-2 protein binds GC-rich recognition sequences present in the cis-regulatory regions of the SV40 enhancer, HTLV-1 enhancer, as well as the human metallothionein IIa, proenkephalin, and mouse major histocompatibility complex h-2K genes (25,28,29). It is also promiscuous in its binding to GC-rich consensus binding sites for S p l and NF1 (30). AP-2 is developmentally regulated and displays selected tissue specificity (31,321. It has also been demonstrated to mediate induction of the hMT-IIA enhancer by combination of the protein kinase C and CAMP signal-transduction pathways (25). The AP-2 protein binds as a dimer to sequences with the consensus 5'-GCCCCAGGC-3'. This sequence differs by only one nucleotide from the NPY sequence between -65 and -57. The effects of NGF on the expression of AP-2 in differentiating PC12 cells is not known, and we were unable t o demonstrate any differences in binding intensities of the AP-2 complex as measured by the gel retardation analysis in extracts prepared from NGF treated cells uersus nontreated cells. Since, the pattern ofAP-2 expression in neural crest cells and the developing nervous system indicates that it may play a major role in establishing the peripheral nervous system and its connection with the central nervous system, it would be consistent for AP-2 to play a

NGF-induced NPY Gene 7kanscription
NGF-RE The DNA-protein interactions occurring between -87 and -58, as illustrated by complexes B and A (Fig. 9), are not known to be represented by previously characterized DNA binding activities. Addition of these sequences to NPYICAT constructs increases the transcription observed in PC12 cells whether or not the cells have been exposed to NGF. This suggests that this sequence binds a positive regulator(s) whose activity is further increased via the NGF cascade. The purification and characterization of these proteins will provide important information concerning their role in modulating the promoter activity of the NPY gene and the ability of the NGF cascade to further increase this activity. The availability and potential modifications of these proteins as modulated by the NGF cascade are likely to determine the overall activity of the NPY gene during the differentiation process occurring in PC12 cells. However, no differences in gel retardation patterns nor footprinting patterns were apparent between extracts prepared from NGFtreated cells uersus non-NGF-treated cells. It is possible that protein modifications may not be apparent under these circumstances and just as likely that any quantitative differences would be muted by the artificial concentration of proteins during nuclear extract preparation. It will, therefore, be of interest to determine the identity of these proteins and to study not only their interactions with the NPY promoter but also their potential ability to interact with one another and the other proteins which bind to proximal NPY promoter sequences.