ZPK inhibits PKA induced transcriptional activation by CREB ... - Nature

ã

Oncogene (1999) 18, 4474 ± 4484 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc

ZPK inhibits PKA induced transcriptional activation by CREB and blocks retinoic acid induced neuronal dierentiation Usha R Reddy*,1 Amitabha Basu4, Peter Bannerman1, Naohiko Ikegaki2, C Damodara Reddy3 and David Pleasure1 1

Department of Neurology Research, Children's Hospital of Philadelphia, University of Pennsylvania, Philadelphia, Pennsylvania, PA 19104, USA; 2Department of Oncology, Children's Hospital of Philadelphia, University of Pennsylvania, Philadelphia, Pennsylvania, PA 19104, USA; 3Department of Neuro-Oncology, Children's Hospital of Philadelphia, University of Pennsylvania, Philadelphia, Pennsylvania, PA 19104, USA; 4The Wistar Institute, Philadelphia, Pennsylvania, PA 19104, USA

Zipper Protein Kinase (ZPK) is a leucine zipper protein localized to the nucleus which exhibits serine-threonine kinase activity and is associated with the stress dependent signal transduction pathway. ZPK forms heterodimers with leucine zipper containing transcription factors such as the cyclic AMP responsive element binding protein (CREB) and Myc. Furthermore ZPK phosphorylates both Myc and CREB. Overexpression of ZPK in NTera-2 human teratocarcinoma cells results in inhibition of PKA induced transcriptional activation by CREB and prevents retinoic acid induced dierentiation of the cells to neurons. Our results suggest that ZPK sti¯es neural dierentiation of NT-2 cells partly due to its inhibitory eect on CREB function. Keywords: ZPK; CREB; NT2 dierentiation

Introduction NTera2/cl.D1(NT2), a human teratocarcinoma cell line, when properly manipulated provides an excellent model system to study retinoic acid (RA) induced dierentiation of neuronal cells (NT2-N) (Pleasure et al., 1992). Evidence suggests that neurotrophins such as brain derived nerve growth factor (BDNF) and trk receptors are upregulated during retinoic acid induced dierentiation of NT-2 cells (Cheung et al., 1996). The transcription factor cAMP response element-binding protein (CREB) is a dominant mediator of neurotrophin-induced gene expression in developing neurons. CREB and its closely related proteins, such as ATF4 and ATFa (activating transcription factors) contain the leucine zipper motif and act as transcriptional regulators that respond to changing levels of cAMP, Ca2+ and transforming growth factor b. These responses, through the cAMP mediated phosphorylation of protein kinase A (PKA) are implicated in a variety of biological responses such as neuronal excitation, long-term memory formation, neural cell proliferation and opiate tolerance (Brindle et al., 1993; Hagiwara et al., 1993; Lee et al., 1993; Martin and Kandel, 1996; Gonzalez and Montminy, 1989). CREB

functions by forming homo/heterodimers with other leucine zipper proteins and thereby bind to the consensus cAMP response element CRE to activate target gene transcription. However, other members of the CREB family, such as ATF2 and ATFa, which are involved in the transcriptional activation of the adenoviral oncoprotein E1A, are not regulated by cAMP (Hai et al., 1989; Liu and Green, 1990). We have previously reported the cloning of a novel protein, ZPK which has protein serine-threonine kinase activity and is related to the mixed lineage kinase family (Reddy et al., 1994). The mouse homolog of this gene, DLK (Holzman et al., 1994), and the rat homolog, MUK (Hirai et al., 1996), have also been isolated, and have been shown to be associated with the JNK/SAPK pathway (Fan et al., 1996; Hirai et al., 1997). We now report that ZPK phosphorylates CREB both in vitro and in vivo and forms an immunoprecipitable complex with CREB and inhibits PKA-induced transcriptional activation by CREB. Furthermore we demonstrate that ZPK, when overexpressed in human teratocarcinoma cell line NTera-2, inhibits the neuronal dierentiation induced by retinoic acid. Our results strongly suggest that this novel interaction between ZPK and CREB plays a critical role during RA induced neuronal dierentiation.

Results Speci®city of ZPK antibody To examine the role of the ZPK protein in neuronal dierentiation the speci®city of the antibodies was determined by Western blot analysis. Immunoprecipitation of the TNT reticulocyte lysate produced ZPK protein using puri®ed antibody gave a major band with an apparent molecular mass of 105 kDa corresponding to the expected molecular weight and a minor band of 120 kDa which might represent pre-processed ZPK (Figure 1A, lane 1). A negative control antibody did not react with the recombinant proteins (lane 3). We did not detect any band using vector as a negative control (lane 4). ZPK protein is localized in the nucleus

*Correspondence: UR Reddy, Room 516F, Neurology Research, Children's Hospital of Philadelphia, Philadelphia, PA 19104, USA Received 31 August 1998; revised 11 February 1999; accepted 9 March 1999

Western revealed proteins lane 1).

blot analysis using puri®ed ZPK antibodies the presence of 105 kDa, 85 kDa and 57 kDa in the lysates of NIH3T3 cells (Figure 1Ba), In order to localize the ZPK, NIH3T3 cells

ZPK inhibits transcriptional activation of CREB UR Reddy et al

were fractionated into subcellular components and Western blot analysis was carried out using ZPK antibodies. It is evident from Figure 1Bb, lane 2, that ZPK is predominantly present in the nuclear fraction from NIH3T3 cells, although minor quantities were present in the cytosolic fraction (Figure 1Bb, lane 1). In all the cellular compartments, the major band detected was approximately 85 kDa, but 105 kDa and 57 kDa protein bands were also visible. These bands may represent proteolytic processing of ZPK protein.

Attempts at detecting ZPK mRNA by Northern analysis were unsuccessful, possibly due to rapid turnover or degradation. To further con®rm the above results, we performed immunohistochemical staining with ZPK antibody of a con¯uent monolayer of NIH3T3 cells. As shown in Figure 1C, there was intense nuclear labeling and comparatively faint staining of the cytoplasm (Figure 1C). Overexpression of full length ZPK protein in NIH3T3 cells also con®rmed nuclear localization of ZPK (data not shown). Furthermore, immunohistochemical staining of sections from kidney, basal epidermal layer, keratinocytes and choroid plexus with ZPK antibody revealed that ZPK was localized in the nucleus (Reddy and Bannerman; unpublished observation). Phosphorylation of proteins by ZPK

Figure 1 (A) Detection of the ZPK protein by Western blotting. Full length ZPK cDNA was used for the synthesis of the ZPK protein using the TNT reticulocyte lysate system as detailed under Materials and methods. The protein was immunoprecipitated with anity puri®ed ZPK antibody (5 mg), (lane 1); no antibody (lane 2); anity puri®ed control rabbit IgG (5 mg), (lane 3); protein synthesized from vector alone and immunoprecipitated with ZPK antibody (lane 4). After washing the agarose pellets were boiled in sample buer under reducing conditions and examined by 6% SDS-polyacrylamide gel electrophoresis, followed by immunoblotting with ZPK antibody. The relative molecular mass markers are shown (6103). (Ba) Western blotting of cell lysates from NIH3T3 cells. Cell lysates from NIH3T3 cells were applied under reducing conditions to an 8% SDS-polyacrylamide gel, electrophoresed and then immunoblotted with anti-ZPK (lane 1), control antibody anity puri®ed rat IgG (lane 2). The immunoreactive bands are indicated by arrows. Relative molecular mass markers are shown (6103). (b). Both cytosolic and nuclear extracts from NIH3T3 cells were prepared as described in Materials and methods. Fifty mg/lane of protein was applied under reducing conditions to an 8% SDS-polyacrylamide gel and then immunoblotted. The position of a major 85 kDa ZPK protein in the 3T3 cell nuclear fraction is indicated by an arrow. Note that the nuclear fraction has abundant ZPK protein (lane 2). In addition, note the faint immunoreactive bands around 105 kDa and 57 kDa indicated by arrows. Relative molecular mass markers are indicated (6103). (C) ZPK immunoreactivity is localized to the nuclei of con¯uent NIH3T3 cells. Fluorescence analysis by confocal microscopy showing nuclear localization of ZPK immunoreactivity in NIH3T3 cells, indicated by arrows. Cells were scanned using settings at which no signal was seen in cultures labeled with control antibody (see Materials and methods). Scale bar=10 mm

As ZPK is predominantly localized to the nucleus, and contains a conserved kinase domain at the aminoterminus followed by a leucine-zipper domain, it was of interest to investigate its ability to speci®cally dimerize and phosphorylate other transcription factors. Autophosphorylation of the ZPK protein was evident when anti-ZPK immunoprecipitates from NIH3T3 cells were incubated in the presence of [g-32P]ATP (Figure 2A, lane 2). Overexpression of ZPK protein in NIH3T3 cells, followed by immunoprecipitation and autophosphorylation revealed more autophosphorylated ZPK (Figure 2B, lane 2). We next examined the ability of ZPK to phosphorylate other zipper containing proteins such as CREB and Myc (Figure 2C). The immunoprecipitated ZPK phosphorylated CREB peptide that contains the leucine zipper domain (lane 3), recombinant Myc (lane 4), but not NF-kB lacking the leucine zipper domain (lane 2). To investigate if ZPK immunoprecipitates from NIH3T3 cells can dimerize and phosphorylate CREB, we performed an in vitro kinase reaction using CREB peptide. We analysed the relative amount of CREB phosphorylation bound to protein A agarose-ZPK and in the supernatants. The results presented in Figure 2D show that phosphorylated CREB was predominantly associated with the immunoprecipitated complex containing ZPK (lane 1). Immunoprecipitation of the post-reaction supernatant with CREB antibody shows a small amount of phosphorylated CREB peptide, whereas Western blot analysis shows the presence of abundant unphosphorylated CREB peptide. These results suggest that ZPK heterodimerizes with CREB and phosphorylates CREB in vitro. Phosphorylated CREB is capable of associating with ZPK in vivo To understand if the in vitro association that we observed occurs in vivo, we labeled NIH3T3 cells with H332PO4. The cell lysates were immunoprecipitated with either anti-CREB or anti-ZPK. Phosphorylation of *43 kDa and *100 kDa proteins was observed in the immunoprecipitated pellets (Figure 3, lanes 1 and 2) that correspond to the molecular weights of CREB and ZPK respectively. We did not detect any phosphorylated CREB protein when myc antibody was used for

4475


4476

immunoprecipitation (Figure 3, lane 3). The lower panel shows a Western blot analysis of CREB content of the immunoprecipitates that corresponded to the molecular weight of 43 kDa. The 44 kDa protein band observed with myc antibody represents the heavy chain IgG. The intensity of the immunoreactive pellet precipitated with anti-ZPK was signi®cantly higher than that precipitated by anti-Myc (lane 3). When these immunocomplexes were probed with ZPK antibody we detected a very faint ZPK band when myc antibody was used for immunoprecipitation (lane 3). Overexpression of catalytically active ZPK in NIH3T3 cells To con®rm that phosphorylation of CREB is critical for dimerization with ZPK, we overexpressed ZPK and its mutant lacking the leucine zipper domain (ZPKDLZ, 1 ± 406) in NIH3T3 cells and puri®ed the recombinant proteins. The puri®ed protein was catalytically active as measured by the in vitro kinase activity. To facilitate detection and puri®cation of these proteins, we used a double epitope tag system (pSecTag) which inserts both a myc epitope and six tandem histidine residues at the C-terminus and a murine Igk-chain leader sequence at the N-terminus for protein secretion. The media in which transfected cells were grown was analysed for secreted ZPK protein by Western blot analysis. We observed that a small proportion of the total ZPK produced in NIH3T3 cells was secreted into the media and was unable to phosphorylate the substrate, histone H1 in vitro (data not shown). The recombinant ZPK was present predominantly within the cells and was puri®ed from the cell lysate using two consecutive steps of anity chromatography (see Materials and methods). As we were unable to elute ZPK bound to anti-myc-sepharose, we used the ZPK bound to sepharose beads for subsequent phosphorylation assays. These beads were catalytically active in kinase assays. Western blot analysis of the puri®ed recombinant full length ZPK (Figure 4, lane 1) and puri®ed vector protein (Figure 4, lane 2) was examined with Figure 2 In vitro phosphorylation of ZPK using cell lysates. (A) Cell lysates from NIH3T3 cells were immunoprecipitated with control antibody (anti-chicken/turkey immunoglobulins) (lane 1) and anti-ZPK antibody (lane 2). Autophosphorylation was performed as described in Materials and methods. The autophosphorylation band detected was exposed for 3 h and had an apparent molecular mass of 105 kDa. (B) Immunoprecipitation with ZPK antibody followed by autophosphorylation of ZPK kinase in NIH3T3 cells (lane 1) and ZPK transfected NIH3T3 cells as described in Materials and methods (lane 2) is shown. The autophosphorylation band detected had an apparent molecular mass of 105 kDa and was exposed for 2 min. The bottom panel shows the amount of ZPK protein present as determined by Western analysis (a-ZPK : W.B). (C) In vitro phosphorylation of various substrate proteins was performed as described in Materials and methods. Lane 1 shows the lack of detectable phosphorylation of NF-kB, which should run with a relative molecular mass of *52 kDa (personal communication, Promega Corp). Note the presence of a ZPK autophosphorylation signal indicated by an arrow. The band corresponding to a molecular mass of 35 kDa seen below 43 kDa is a nonspeci®c band. Lane 2 depicts immunoprecipitation using control rabbit IgG followed by phosphorylation in the presence of CREB peptide as substrate. In addition to the substrate proteins phosphorylated indicated by arrow-heads (CREB peptide (lane 3); N-myc/c-myc fusion protein

(lane 4)), autophosphorylation of ZPK, indicated by arrows was also detected. Relative molecular mass markers are indicated (6103). (D) In vitro phosphorylation of CREB peptide by ZPK. Cell lysates from NIH3T3 cells were incubated with ZPK antibody and immunoprecipitated (lane 1). Lane 2 represents a control lane in the absence of cell lysate, Lane 3 represents a control in the presence of protein A agarose and in the absence of cell lysate and ZPK antibody. Then in vitro kinase assay was performed as described in Materials and methods using CREB peptide (14 ± 15 kDa) as the substrate. Lanes 4, 5 and 6 correspond to unbound CREB present in the supernatant of the kinase reaction mixture of respective samples 1, 2 and 3 (i.e. the supernatant after in vitro kinase assay) and immunoprecipitation with the CREB antibody. Lane 1 shows the phosphorylated CREB peptide along with ZPK protein indicated by arrows. The bottom panel of (D) shows the corresponding samples by Western blot with anti-CREB antibody (a-CREB: W.B). The amount of unbound CREB peptide detected in lane 4, is higher when compared to the phosphorylated product seen in the lane 1. Lane 7 represents control CREB protein used for the Western blot. Relative molecular mass markers are indicated (6103)


the anity puri®ed anti-ZPK antibody (left panel), anti-myc antibody (middle panel) and control antibody (right panel). We observed an 130 kDa band which was relatively faint, and the most prominent bands appeared as a doublet with an apparent molecular mass around 85 kDa with anti-ZPK antibody. These 85 kDa bands were not detected by Western blot using anti-myc antibody. We hypothesized that puri®ed ZPK forms a homodimer, and the 85 kDa band detected with anti-ZPK is due to proteolytic processing involving the C-terminus of the recombinant ZPK. Our conclusion is supported by studies by Mata et al. (1996) who showed, using DLK, the mouse homologue of ZPK, that DLK expressed in COS7 cells did form homodimers. Role of leucine zipper domain of ZPK in CREB function The results presented in the previous section suggest that the leucine zipper domain (LZ) of ZPK is critical for the kinase activity and is involved in the dimerization with transcription factors with LZ. To evaluate the function of LZ in protein ± protein interactions we examined the phosphorylation of

CREB and Myc by recombinant ZPK. Our results con®rmed that ZPK is capable of autophosphorylation and substrate phosphorylation of CREB peptide (Figure 5A), recombinant full length CREB protein (Figure 5B) and recombinant Myc protein (Figure 5C). In these experiments, the supernatant after the kinase reaction contained very small amounts of phosphorylated CREB, whereas Western blot analysis shows that there is abundant CREB which is not associated with ZPK or phosphorylated (Figure 5Bb, lane 6). Similar results were obtained using c-Myc as the substrate (Figure 5C). ZPKDLZ, lacking a leucine zipper domain, showed a markedly diminished ability to autophosphorylate (Figure 5A, lane 2) as well as substrate phosphorylate CREB and Myc (Figure 5B and C, lanes 1 and 4). The presence of phosphorylated CREB in the unbound fraction suggests that the phosphorylated substrate can dissociate from the kinase following the enzymatic reaction. This is supported by the fact that ZPKDLZ, lacking the leucine zipper domain, does not phosphorylate eciently and thus the phosphorylated substrate in the unbound fraction is very small. These results suggest that ZPK phosphorylates CREB and myc through the interaction by the LZ domains of these proteins. CREB-ZPK heterodimer binds CREB responsive element (CRE) To examine the role of the CREB-ZPK heterodimer in DNA binding, we performed EMSA using wild type and mutant CRE oligonucleotides with nuclear extracts from NT2 cells. The speci®city of CREB binding was determined using mutant CRE oligonucleotide (Figure 6A, lane 2) and competition experiments using unlabeled CRE (Figure 6A, lanes 3 and 4). Results indicate that the CREB-ZPK complex binds the CRE

Figure 3 In vivo phosphorylation of CREB by ZPK followed by coimmunoprecipitation of CREB and ZPK. NIH3T3 cells were labeled with H332PO4. The cells were washed and lysed as described in Materials and methods. The cell lysates were immunoprecipitated with anti-CREB (lane 1), anti-ZPK (lane 2) and anti-Myc (lane 3). After immunoprecipitation the samples were run under reducing conditions on an 8% SDS ± PAGE gel. Note the prominent ZPK protein (*100 kDa) immunoprecipitated along with CREB protein (*43 kDa) when immunoprecipitated with either antibody. With the anti-Myc we detect Myc protein (*70 kDa) in addition to a faint ZPK phosphorylated band. The bands below the CREB protein correspond to nonspeci®c proteins. The bottom panels represent the corresponding samples analysed by Western blotting with anti-ZPK antibody (a-ZPK : W.B) and anti-CREB antibody (a-CREB : W.B). The signal detected in lane 3, using anti-CREB antibody corresponds to the IgG which is found to have identical mobility to that of CREB *43 kDa. The CREB protein observed in lane 2, corresponds to the phosphorylated CREB immunoprecipitated with anti-ZPK. Western analysis with anti-ZPK indicates that the intensity of ZPK immunoprecipitated with ZPK antibody is greater than when immunoprecipitated with CREB or myc (lane 2). Note that Western analysis with anti-ZPK con®rms that CREB is immunoprecipitating ZPK (lane 1)

Figure 4 Western analysis of the recombinant puri®ed ZPK protein. Puri®cation of recombinant full length ZPK and the vector proteins are described in Materials and methods. The proteins were run under reducing conditions and Western analysis with either anti-ZPK (left panel), anti-myc (middle panel) or control antibody (right panel) are depicted. The puri®ed ZPK protein reacts with both ZPK and Myc antibodies as indicated by the arrow (lane 1). The pattern of mobility of the puri®ed ZPK has an apparent molecular mass of 130 kDa and 85 kDa and is similar to the native proteins from NIH3T3 cells which migrate at 105 kDa and 85 kDa on SDS ± PAGE. The full length puri®ed synthetic ZPK protein has an additional 68 amino acids and has an apparent molecular mass of 130 kDa. We do not observe any reactivity using the vector protein using any of the antibodies (lane 2). The apparent position of the molecular mass markers are indicated (6103)

4477


4478

Figure 5 Catalytic activity of puri®ed ZPK. In vitro phosphorylation of various substrate proteins was performed as described in Materials and methods. (A) In vitro phosphorylation of CREB peptide by puri®ed ZPK. Puri®ed proteins coupled to anti-mycsepharose were subjected to in vitro phosphorylation using CREB peptide (14 ± 15 kDa) as the substrate. Lane 1 corresponds to the vector protein. We did not see any phosphorylation in the presence of ZPKDLZ protein (lane 2), whereas we detected phosphorylation of CREB peptide using the puri®ed full length ZPK protein (arrow, lane 3). Note the extensive autophosphorylation of ZPK (arrow-head). Lanes 4, 5, and 6 correspond to the unbound CREB peptide from the kinase reaction of samples 1, 2 and 3 respectively. (B) The bottom panel represents Western blot analysis with anti-CREB antibody (a-CREB : W.B) and anti-ZPK antibody (a-ZPK : W.B) of the corresponding samples. The amount of CREB peptide detected in lane 3b, corresponds to the amount of phosphorylated product bound to ZPK in lane 3a. Note the presence of equal amounts of ZPK protein in ZPKDLZ and full length ZPK (lanes 2 and 3). Relative molecular mass markers are indicated (6103). (Ba) In vitro phosphorylation of recombinant full length CREB protein by puri®ed ZPK. Puri®ed recombinant ZPK lacking the leucine-zipper domain (ZPKDLZ, lane 1), puri®ed vector coupled to anti-myc sepharose (lane 2) and puri®ed recombinant ZPK full length protein coupled to anti-myc sepharose beads (lane 3), were phosphorylated with CREB protein as substrate. In addition to the phosphorylated ZPK we detected phosphorylation of CREB in lane 3, and very low CREB phosphorylation with ZPKDLZ protein (lane 1). After the in vitro kinase assay, immunoprecipitation of unbound CREB showed a CREB phosphorylated band in lane 6 (Panel a). (B) The bottom panel represents the corresponding Western blot analysis with CREB antibody (a-CREB : W.B.). (C) In vitro phosphorylation of recombinant N-myc/c-Myc fusion protein by puri®ed ZPK. Puri®ed recombinant ZPK protein lacking the leucine-zipper domain (ZPKDLZ) (lane 1), puri®ed vector coupled to anti-myc-sepharose (lane 2, and puri®ed recombinant ZPK full length protein coupled to anti-myc-sepharose (lane 3), were phosphorylated with N-myc/c-Myc fusion protein as a substrate. A similar result was obtained as that with CREB full length protein where puri®ed full length ZPK protein phosphorylated Myc (lane 3)

(Figure 6B and C). To determine if ZPK exists as a heterodimer complex, we used antibodies speci®c to CREB and ZPK in supershift assays. As shown in Figure 6B, with increasing concentrations of ZPK antibody, the intensity of the CREB-ZPK heterodimer binding to CRE was decreased. The control antibody (anity puri®ed rabbit IgG) had no eect on the DNA binding ability of the CREB homodimer or the ZPKCREB heterodimer binding to the CRE (Figure 6C, lane 2). Thus, in the presence of ZPK speci®c antibody the binding of the ZPK-CREB heterodimer to the CRE was signi®cantly decreased whereas there is no dierence in CREB binding to the CRE (Figure 6C, lane 3). In the presence of CREB antibody the mobility of the CRE-protein complexes was almost similar to that of the CREB-ZPK heterodimer complex (Figure 6C, lane 4). ZPK blocks the Protein kinase A (PKA) mediated transcriptional activation function of CREB Since ZPK and CREB heterodimerize in vitro, we investigated the possibility of ZPK speci®cally

modulating CREB mediated transcriptional activation. We used the CREB trans-reporting system to monitor the activation of transcription of CREB in vivo (Figure 7). In these experiments we examined the ability of ZPK to phosphorylate CREB speci®c fusion trans-activator protein which binds as a dimer with GAL4 binding sites of a transiently transfected reporter gene, pFR-luciferase, and activates transcription of luciferase. The results of the luciferase assay in two dierent cell lines (NIH3T3 and NT-2) cotransfected with the reporter and expression vectors are shown in Figure 7B. It is evident from the ®gure that direct or indirect phosphorylation of CREB by PKA activates transcription of the luciferase reporter gene. Cotransfection experiments in the presence of pCB6ZPK encoding the ZPK protein reduced the transcriptional activation observed with PKA alone. In contrast, cotransfection of ZPK lacking the leucinezipper domain (ZPKDLZ) plus the PKA did not inhibit activation of CREB. A negative control using the pFC2-dbd plasmid was used to ensure the eects observed were not due to the GAL4 DNA binding domain (data not shown).


4479

Figure 6 Gel mobility shift assay showing the DNA binding of CREB and ZPK proteins. 32P labeled CRE oligonucleotide was incubated with nuclear extracts from NT2 cells (5 mg) in the absence (7) or presence of nonradioactive competitors as indicated above each lane. DNA-protein complexes were resolved on 5% native polyacrylamide gels. Lane 1 (panel A) shows binding in the absence of competitor. The binding speci®city was assessed using unlabeled CREmut (lane 2) and unlabeled CRE oligonucleotides (lanes 3 and 4). Competition of the speci®c binding (shown by arrows) was obtained with wild type CRE as a competitor. The speci®c complex is indicated by arrows whereas the star shows a nonspeci®c complex formed by extract derived proteins. Supershift analysis of proteins bound to the CRE oligonucleotide is shown in panels B and C. NT2 nuclear extracts were incubated in the absence of antibody (panel B, lane 1), or in the presence of ZPK antibody at dierent concentrations as indicated (lane 2, 3 and 4) for 30 min at room temperature followed by incubation with labeled CRE oligonucleotide for 30 min. Note the diminution in the CREB/ZPK heterodimer (arrow) with increasing ZPK antibody whereas there was no change in the CREB dimer complex. (C) Supershift analysis in the presence of control antibody showing both the CREB dimer as well as the CREB/ZPK heterodimer (lane 2), the disappearance of CREB/ZPK heterodimer in the presence of ZPK antibody (lane 3), and the supershifted complex in the presence of CREB antibody (lane 4). The CREB/ZPK-CRE complex (arrow) has a mobility very similar to that of the CREB homodimer in the presence of CREB antibody (arrowhead)

Overexpression of ZPK blocks retinoic acid-induced neuronal dierentiation of teratocarcinoma cells In order to understand the in vivo role/biological signi®cance of ZPK in neuronal dierentiation we have examined the eect of overexpression of ZPK in the human teratocarcinoma cell line NTera-2. Use of retroviral vectors to study the overexpression has distinct advantages in obtaining high levels of expression of cloned genes. The packaging cell line PA317 produced retrovirus at a titre 56105 cfu/ml. This high titre virus stock was used to infect the NT2 cells. There was no phenotypic dierence (growth or morphology) in NT2 cells transfected either with the empty vector or ZPK. Figure 8 shows NT2 cells transduced with the empty vector alone (NT2-V) or vector containing ZPK cDNA (NT2-ZPK) when

treated with RA for 5 weeks followed by replating at a lower density. NT2 cells transduced with control virus (i.e. with the vector alone) followed by retinoic acid treatment became phase bright NT2-N cells which grew as large cellular aggregates. In addition, a small number of ¯at cells which resembled undierentiated NT2 cells were also seen (Figure 8B, upper panel). Thus, more neurons could be isolated from these cellular aggregates. On the other hand, cells overexpressing ZPK following retinoic acid treatment remained as predominantly undierentiated NT2 cells (Figure 8B, lower panel). It was not possible to isolate a signi®cant number of neurons from these cultures. The results represent at least ®ve individual experiments and re¯ect the reproducibility of the inhibition of retinoic acid induced dierentiation of NT-2 cells by overexpression of ZPK. Additional evidence for the


4480

role of ZPK was obtained by monitoring differentiation using a neuronal marker such as neuro®lament

(NF-M). Western blot analysis of these cells with neuro®lament antibody which is speci®c to NF-M

Figure 7 ZPK inhibits CREB mediated transcriptional activation. (A) Schematic representation of the plasmids used for transfections into NIH3T3 and NT2 cells is shown. pFA-CMV-CREB the fusion transactivator plasmid was constructed using full length CREB cDNA (a). ZPK and ZPKDLZ are expression plasmids for ZPK and ZPK lacking the LZ domain in the pCB6 vector containing the CMV promoter (b and c). pFC-PKA is the expression plasmid expressing the catalytic subunit of protein kinase A (d); the reporter plasmid pFRLuc which contains the GAL4 binding sites upstream of a TATATA box followed by the luciferase coding sequence (e). (B) Both NIH3T3 cells and NT2 cells were transiently transfected with the reporter plasmid (pFR-Luc) in the presence of the expression plasmid pFC-PKA and other expression plasmids as indicated at the bottom of the ®gure. The luciferase assay was performed at 48 h posttransfection. The graph shows luciferase activity (relative light units/mg protein) expressed as percentage of PKA activity. Values are means+s.e.m. for three separate transfections. In case of NT2 cells transfected with ZPKDLZ expression plasmid the values represent average of two independent experiments


4481

Figure 8 Overexpression of ZPK blocks RA induced dierentiation. (A) Retroviral vector pLXSN and ZPK containing pLXSN (pLXSN-ZPK). Abbreviations: LTR, long terminal repeat; c, extended retroviral packaging signal; SV, simian virus 40 early promoter and enhancers; neo, neomycin resistance gene; pA, polyadenylation site; ZPK, zipper protein kinase. (B) The growth and dierentiation of NT2 cells were carried out as described previously (Pleasure et al., 1992). NT2-V cells expressing exclusively the vector (upper panel), or NT2-ZPK overexpressing cells (lower panel), after treatment with retinoic acid for 5 weeks followed by replating at a lower density is shown. Note the presence of phase bright cells in clumps sitting above the adherent cells in NT2-V, whereas no clumps are apparent in NT2-ZPK cells (Scale bar=10 mm). (C) Northern blotting of ZPK mRNA. Total RNA (15 mg/ lane) was isolated after treatment with RA for 5 weeks and analysed for ZPK mRNA. Lane 1 RNA from NT2-ZPK cells (NT2ZPK+RA); Lane 2 NT2-vector cells (NT2-V+RA); Lane 3 NT2 cells (NT2+RA); lane 4 human NT2-N cells (cultures are 95% neurons). Note the multiple transcripts detected for RNA derived from NT2-ZPK. The lower panel shows the corresponding ethidium bromide staining of RNA. (D) Detection of NF-M by Western blotting. Protein extracts from the above cells were applied under reducing conditions to an SDS-containing 8% polyacrylamide gel and then electroblotted. Immunoreactive NF-M was detected with anti-NF-M antibody RM 044.14. Note the absence of the upper doublet band in NT2-ZPK cells corresponding to the phosphorylated NF-M. The blot after NF-M staining was stripped and probed with anti-ZPK antibody. Note the predominant ZPK protein band in the overexpressed cells

shows a doublet band which correspond to the phosphorylated upper and unphosphorylated lower bands. The neurons obtained from the NT2-V cells express the phosphorylated neuro®lament protein (Figure 8D, lane 2) whereas in NT2-ZPK cells treated with retinoic acid the phosphorylated NF-M was absent (Figure 8D, lane 1). Northern blot analysis of these cells show abundant ZPK message in the overexpressed cells when compared to the vector control (Figure 8C, upper panel). In addition to the 3.4 kb message, there was an additional transcript of

approximately 7.0 kb possibly due to the initiation of transcription from the 5'LTR promoter. The corresponding Western blot analysis of ZPK show abundant ZPK protein in the overexpressed cells when compared to the vector control (Figure 8D). Discussion The ZPK gene encodes a protein serine threonine kinase with an apparent molecular mass of 105 kDa.


4482

The antibodies raised to this protein reacted with 105 kDa, 85 kDa and 57 kDa proteins from NIH3T3 cells. This diversity could be due to proteolytic processing. This is supported by the observation that, puri®ed ZPK fusion protein containing C-terminal myc epitope, was not recognized by the anti-myc antibody as this portion of the protein was subjected to proteolytic processing. These results support our hypothesis that the 85 kDa protein recognized by anti-ZPK antibody is the product of carboxyterminal proteolytic processing. The compartmentalization of ZPK to the nucleus could re¯ect protein targeting, the presence of an anchoring protein, or both (Mochly-Rosen, 1995). The mouse homolog of ZPK, DLK was found to be associated with both plasma membrane and cytosol fractions by subcellular fractionation studies (Mata et al., 1996). In contrast, immunostaining of intact cells always showed nuclear staining. Further studies are required to resolve this discrepancy. Our results show that the interaction of the leucine zipper domain of ZPK with other proteins appears to play an important role in its substrate speci®city. ZPK is capable of autophosphorylation and also phosphorylates other leucine zipper transcription factors such as Myc and CREB. Both autophosphorylation and phosphorylation of other leucine zipper proteins are markedly reduced when the leucine zipper domain of ZPK was deleted. These results suggest that the leucine zipper domain plays a critical role in dimerization and maintaining enzymatic activity of ZPK. In addition, we have shown that ZPK does not phosphorylate other non-leucine zipper transcription factors (e.g. NF-kB) emphasizing the requirement of LZ for the function of ZPK. The mouse and rat homolog of ZPK, DLK and MUK, have been reported to phosphorylate c-Jun (Fan et al., 1996; Hirai et al., 1996), while ZPK does not phosphorylate a c-Jun fusion protein lacking the leucine zipper domain (unpublished observation). Hirai et al. (1996) using MUK have reported that a truncated MUK lacking the C-terminal proline/glycine rich region activates JNK in vivo as eciently as intact MUK, while deletion of the leucine zipper region blocks the ability of MUK to activate JNK. Therefore it can be concluded that the kinase domain tailed by the leucine-zipper region is essential for enzymatic activity. Studies with puri®ed ZPK protein minimized the possibility of another protein kinase co-precipitating with ZPK, and these results con®rmed the studies performed with native ZPK from cell lysates. It is known that leucine zipper proteins do not interact with each other in a random fashion but that there are inherent structural features in these domains which allow speci®c interactions (Johnson and McKnight, 1989; O'Shea et al., 1989). Our results demonstrate that overexpression of ZPK inhibits RA-induced neural dierentiation of NT2 cells. Retinoic acid has been shown to aect PKA in the dierentiation of several other cell systems. An increase in PKA activity has been associated with the retinoic acid-induced dierentiation (Abraham et al., 1991). Our data shows that ZPK modulates PKA activity, suggesting that ZPK proteins dimerize with CREB to generate heterodimers and inhibit CREB transcriptional activation by PKA. Thus, ZPK negatively regulates PKA activity by heterodimerization, and

thereby regulates CREB function. Recently, another serine-threonine kinase (ZIP) has been reported, which has an N-terminal kinase domain and a C-terminal leucine-zipper domain that is necessary for homodimerization as well as heterodimerization with ATF4 (Kawai et al., 1998). In this paper we report that ZPK, a serine-threonine kinase with a leucine zipper domain, localizes in the nucleus and that overexpression of ZPK blocks the RA induced neural dierentiation of NT2 cells. We demonstrate that ZPK associates with and phosphorylates the leucine-zipper containing nuclear transcription factors, such as CREB and Myc. Further studies revealed that ZPK inhibits PKA-induced transcriptional activation by CREB. Thus, our results suggest a novel role for ZPK in CREB mediated phosphorylation pathways. These phosphorylations can proceed independent of known signaling molecules like cAMP, Ca+2 or growth factors. Materials and methods Cell culture NIH3T3 cells were grown in DMEM supplemented with 10% fetal calf serum. The growth and dierentiation of NT2 cells was carried out as described previously (Pleasure et al., 1992). Production of ZPK protein in bacteria The 5' region of ZPK cDNA corresponding to 743 bp was inserted into pMALc2 at XmnI and SalI sites. The fusion protein was puri®ed by maltose binding protein anity chromatography as described by the manufacturer (NEB). Antibody production and puri®cation The polyclonal antiserum was anity puri®ed using a ZPK fusion protein-sepharose anity column (Harlow and Lane, 1988) and tested with in vitro translated ZPK protein produced using the TNT T3 and T7-reticulocyte lysate system (Promega, Madison, WI, USA) as well as with the recombinant ZPK protein with a myc epitope tag. The anity-puri®ed antibody was used for immuno¯uorescence, immunoprecipitation and immunoblot analysis. In vitro transcription-translations In vitro transcription-translation was performed using the TNT T3 and T7-reticulocyte lysate system (Promega, Madison, WI, USA). Immunohistochemistry NIH3T3 cells were grown on coverslips and pre-®xed with 2% paraformaldehyde in PBS for 10 min at room temperature (258C). They were re®xed with cold methanol (7208C, 8 min), washed with PBS and incubated with blocking solution B comprising Minimal Essential Medium (MEM) containing 15 mM HEPES buer, 10% (v/v) FCS and 0.02% (w/v) sodium azide (block A) and 0.3% Triton X100 (v/v) for 10 min. Cells were then incubated overnight at 48C with anti-ZPK antibodies (10 mg/ml) in block B. The coverslips were washed with PBS and incubated with biotinylated sheep anti-rabbit immunoglobulins (Amersham) for 25 min. Following additional PBS washes, the coverslips were incubated with rhodamine-conjugated streptavidin for 15 min at room temperature. Then the cells were washed, post®xed with cold methanol, and mounted in Vectashield (Vector Labs).


Immuno¯uoresence and confocal microscopy Immunolabeled cell cultures were optically sectioned using a computer-interfaced, laser-scanning microscope (Leica TCS 4D), ®tted with a 488 nm/568 nm/647 nm krypton-argon laser. Immunostained cells were scanned with a 6100 oil immersion objective at the following settings: laser power=1, pinhole=100, voltage=749, voltage oset=74. Using these settings, no signal was detected in cells labeled with control antibody (rabbit antibodies to turkey immunoglobulins), while clear intracellular immunolabeling was observed in cultures incubated with rabbit antibodies to ZPK. Immunoblot anaysis Cells were lysed in a `lysis buer' consisting of 50 mM Tris (pH 7.5), 250 mM NaCl, 0.1% Triton X-100 (v/v), 0.1% deoxycholic acid (w/v), 5 mM EDTA, 1 mM PMSF, 100 mM sodium orthovanadate, 50 mM NaF, 1 mg/ml leupeptin, 1 mg/ ml pepstatin, 1 mg/ml aprotinin. Proteins were resolved by SDS ± PAGE (50 mg of protein/lane). Cytoplasmic and nuclear preparations of NIH3T3 cells (1506106 cells) were prepared as described previously (Ausubel et al., 1996). Aliquots (50 mg) of nuclear and cytosol fractions were used for immunoblotting. Samples were resolved on 8% SDS ± PAGE and then transferred electrophoretically to a nitrocellulose membrane overnight at 48C. The nitrocellulose was then incubated for 1 h at room-temperature in 5% milk. The membrane was washed and incubated for 1 h with 100 ng/ml of anti-ZPK antibody diluted in PBST (phosphate buered saline pH 7.5 and 0.05% (v/v) Tween 20) or neuro®lament antibody (RM 044.14, a kind gift from Dr Virginia Lee). After washing the membrane was incubated with horseradish peroxidase-labeled goat anti-rabbit IgG diluted 1 : 20 000 in PBST and developed using a chemiluminescence detection kit (DuPont, NEN). Retroviral expression of ZPK Maloney murine leukemia virus-based retroviral vector pLXSN (gift from Dr Osborne's Lab (Stockschlaeder et al., 1991)) was used to express the full length ZPK. ZPK cDNA was cloned into the EcoRI site of pLXSN. pLXSN-ZPK was transfected into the ecotropic packaging line PE501 (Wigler et al., 1978). Two days later, conditioned media containing the virus was used to infect PA317 amphotropic packaging cells. One day after infection, the PA317 cells were split into medium containing G418 (600 mg/ml). Genomic DNA was isolated from monoclonal populations and analysed by Southern blot to ensure the integrity of the retroviral vector sequences. The assays for the vector virus and ZPK producing virus were performed as described earlier (Miller and Buttimore, 1986). NT2 cells were transduced overnight with vector virus and ZPK virus in the presence of 8 mg/ml of Polybrene. After 48 h the cells were selected in G418 (200 mg/ ml) for about 10 days. Immunoprecipitations and in vitro kinase assays NIH3T3 cells were washed in PBS, resuspended in lysis buer, and incubated for 1 h at 48C. The lysates were clari®ed by centrifugation at 12 000 g for 20 min at 48C. Supernatants were recovered and used for immunoprecipitation. For this purpose, 2 ± 5 mg of anti-ZPK antibody was added to 1 ml of lysate and incubated for 1 h at 48C with gentle rotation. Protein A agarose (30 ml pellet) pre-washed with the same lysis buer was added to each sample and incubated for 1 h at 48C with gentle rotation. The samples were washed ®ve times with lysis buer prior to the protein kinase assay. The kinase reaction mixture (50 ml) containing 50 mM Tris-HCl (pH 7.4), 10 mM MgCl2, 1 mM DTT and 20 mCi of [g32P]ATP was incubated for 20 min at 308C in the

presence of protease inhibitors and 5 mg of one of the following substrate proteins: recombinant N-myc/c-myc fusion protein (70 kDa N-myc/c-myc fusion protein, (Ikegaki et al., 1986)); CREB-1 bZIP (10 ± 14 kDa polypeptide; Santa Cruz Biotech, CA, USA); or puri®ed recombinant full length CREB (a gift from Dr Montminy (Hagiwara et al., 1993)) whereas, for NF-kB, 2.5 mg of the protein was used in the kinase assay (p49, E382; Promega, WI, USA). Reactions were stopped by adding 50 ml of 26SDS buer. Phosphorylation of the substrate was monitored by autoradiography. Immunoprecipitation and Western blot to detect phosphorylated and nonphosphorylated CREB After in vitro phosphorylation using CREB as the substrate, the tubes were centrifuged, the supernatants removed, and immunoprecipitations were performed as described above. The phosphorylated CREB in the agarose pellet was washed ®ve times with lysis buer and subjected to SDS gel electrophoresis and Western blotting as described in the immunoblot section. CREB antibody (monoclonal mouse IgG1) which recognizes both phosphorylated and nonphosphorylated CREB was used at a 1 : 100 dilution. In vivo labeling and immunoprecipitation NIH3T3 cells were labeled with inorganic 32PO4 (0.25 mCi ± 0.75 mCi/100 mm plate; DuPont, NEN Research Products) in 7 ml of phosphate-free medium per dish for 24 h. The plates were washed with cold phosphate buered saline and the cells were lysed and immunoprecipitated as described above with anti-CREB, anti-Myc (sc-788; Santa Crutz, CA, USA) and anti-ZPK antibodies. The phosphorylated proteins were detected by SDS ± PAGE followed by autoradiography. Plasmid construction Full length ZPK was expressed using mammalian expression vector pSecTag from Invitrogen (SanDiego, CA, USA). Proteins expressed from pSecTag contain the myc epitope and six tandem histidine residues fused at the C-terminus. Full length ZPK (1 ± 850 amino acids) was subcloned into pSecTag vector B at EcoRI and XbaI sites. Similarly, a truncated ZPK lacking the leucine zipper domain (ZPKDLZ, 1 ± 406 amino acids) was expressed in the same vector. ZPK fragments, encoding 850 and 406 amino acid residues respectively, were engineered with EcoRI and XbaI restriction sites so as to maintain the appropriate reading frame, and ampli®ed by PCR. Gel-puri®ed EcoRI and XbaI digested fragments were subcloned into pSecTag vector. Expression and puri®cation of recombinant ZPK from NIH3T3 NIH3T3 cells were transfected by calcium phosphate precipitation (Wigler et al., 1978). Stably transfected cells were isolated by means of their resistance to Zeocin (600 mg/ ml). Several monoclonal populations were monitored by Northern blot and Western blot analysis for the expression of ZPK. Two clones expressing full length or truncated ZPK protein were isolated, further expanded and used for the puri®cation of the full length and truncated ZPK proteins. Similarly, a single clone expressing the empty vector sequences (82 amino acid) was isolated and used as a control. Proteins from all three clones were puri®ed using the Xpress system protein puri®cation kit from Invitrogen. The proteins obtained were dialyzed against phosphate buered saline, 50 mM HEPES buer, 10% glycerol, and protease inhibitors for 2 h and further puri®ed using anti-myc monoclonal antibody (9E10). Puri®ed monoclonal antibody speci®c for the myc epitope (EQKLISEEDL) (Evans et al., 1985) was coupled to activated CNBr-sepharose according to

4483


4484

the manufacturer's instructions (Pharmacia). The proteins eluted from the nickel column were incubated for 1 h with anti-myc-sepharose at 48C. Equal amounts of ZPKDLZ and ZPK full length proteins obtained from nickel columns were passed through an anti-myc sepharose column. The resin was washed several times with lysis buer, and the activity of ZPK enzyme was tested at each step of puri®cation by in vitro phosphorylation assay described above using histone as the substrate. An equal volume (30 ml, 50% suspension of resin) of the anti-myc-sepharose resin was used in each in vitro assay. Electrophoretic mobility shift assay (EMSA) Nuclear extracts from NT2 cells were prepared as described earlier (Schrieber et al., 1989) and incubated with a buer containing 20 mM HEPES, 50 mM KCl, 1 mM EDTA, 3 mM MgCl, 1 mM DTT, 5 ± 8% glycerol, 0.5% NP40. Synthetic 29 bp oligonucleotides containing the rat somatostatin CREB responsive element, CRE (5'-GATCTCCTTGGCTGACGTCAGAGAGAGAG-3') and mutated CRE (5'-GATCCTCTCTCTCTGTGCTGAGCCAAGGA-3') were labeled with a32P-ATP and the puri®ed probe was used in the EMSA. Five mg of protein in the binding buer was incubated with 1 mg of poly(d[I-C]) (Boehringer Mannheim) and approximately 0.1 ng of labeled oligonucleotide (20 000 c.p.m.) for 30 min at RT in a 20 ml ®nal volume. Antibody inhibition reactions were performed by incubating the protein with control antibody (anity puri®ed rabbit IgG), CREB antibody (Santa Crutz) and ZPK antibody at room temperature for 30 min and then with 32P-labeled CRE oligonucleotide for an additional 20 min. DNA-protein complexes were resolved on a 5% polyacrylamide gel in 0.256TBE buer. The gels were dried and autoradiographed.

CREB expression and luciferase assay Full length CREB cDNA was subcloned into the pFA-CMV plasmid at EcoRI and HindIII sites. The pFA-CMV plasmid is designed for convenient insertion of the activation domain sequence of the CREB transcription activator. Pathdetect, an in vivo signal transduction pathway kit, was used to measure the transcriptional activation ability of CREB (Stratagene). Studies by Sun et al. (1994), using constructs similar to ours, i.e. the GAL4-CREB fusion protein and the GAL4responsive luciferase reporter gene, have shown that the phosphorylation by Ca+2/CaM kinase II blocks the activation of CREB. NIH3T3 cells and NT2 cells were seeded at 26105 cells in 2 ml of complete media in a 6-well tissue culture dish. Transfections were performed in triplicate using the LipoTaxi reagent (Stratagene). The cells were incubated for 6 h with the DNA-lipid complex. The luciferase assay was performed after 48 h using the luciferase assay kit from Strategene. Luciferase activity was assayed as recommended by the manufacturer and the light produced was measured in a Berthoid luminometer. Abbreviations CREB, cAMP response element binding protein; ZPK, zipper protein kinase; CRE, cAMP response element; RA, retinoic acid; PKA, protein kinase A. Acknowledgements We thank Marge Williams for technical assistance with NT2 cell dierentiation and Dr Montminy for providing the recombinant full length CREB. This work was supported in part by grants from the National Institute of Health (NS25044, NS08075, NS31102) and Muscular Dystrophy Association.

References Abraham I, Sampson KE, Powers EA, Mayo JK, Ru VA and Leach KL. (1991). J. Neurol. Res., 28, 29 ± 39. Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman IG, Smith JA and Struhl K. (1996). In: Current Protocols in Molecular Biology. Vol. 3. John Wiley Sons, Inc. Brindle P, Linke S and Montminy M. (1993). Nature (Lond.,) 364, 821 ± 824. Cheung WMW, Chu AH, Leung M-F, Ip NY. (1996). Neuroreport, 7, 1204 ± 1208. Evans GI, Lewis GK, Ramsay G and Bishop VM. (1985). Mol. Cell. Biol., 5, 3610 ± 3616. Fan G, Merritt SE, Kortenjann M, Shaw PE and Holzman LB. (1996). J. Biol. Chem., 271, 24788 ± 24793. Gonzalez GA and Montminy MR. (1989). Cell, 59, 675 ± 680. Hagiwara M, Brindle P, Harootunian A, Armstrong R, River J, Vale W, Tsien R and Montminy MR. (1993). Mol. Cell. Biol., 13, 4852 ± 4859. Hai T, Liu F, Coukos WJ and Green MR. (1989). Genes Dev., 3, 2083. Harlow E and Lane D. (1988). In: Antibodies: A Laboratory Manual. Cold Spring Harbor Laboratory: Cold Spring Harbor, New York. Hirai S-I, Izawa M, Osada S-I, Spyrou G and Ohno S. (1996). Oncogene, 12, 641 ± 650. Hirai S-I, Katoh M, Terada M, Kyriakis JM, Zon LI, Rana A, Avruch J and Ohno S. (1997). J. Biol. Chem., 272, 15167 ± 15173. Holzman LB, Merritt SE and Fan G. (1994). J. Biol. Chem., 269, 30808 ± 30817.

Ikegaki N, Butovsky J and Kennett RH. (1986). Proc. Natl. Acad. Sci. USA, 83, 5929 ± 5933. Johnson PF and McKnight SL. (1989). Ann. Rev. Biochem., 58, 799 ± 839. Kawai T, Matsumoto M, Takeda K, Sanjo H and Akira S. (1998). Mol. Cell. Biol., 18, 1642 ± 1651. Lee KAW and Masson N. (1993). Biochem. Biophys. Acta, 1174, 221 ± 233. Liu F and Green MR. (1990). Cell, 61, 1217. Martin KC and Kandel ER. (1996). Neuron, 17, 567 ± 570. Mata M, Merritt SE, Fan G, Yu GG and Holzman LB. (1996). J. Biol. Chem., 271, 16888 ± 16896. Miller AD and Buttimore C. (1986). Mol. Cell. Biol., 6, 2895 ± 2902. Mochly-Rosen D. (1995). Science, 268, 247 ± 251. O'Shea EK, Rutkowski R, Staord III WF and Kim PS. (1989). Science, 245, 646 ± 648. Pleasure SJ, Page C and Lee VM-Y. (1992). J. Neurosci, 12, 1802 ± 1815. Reddy UR and Pleasure D. (1994). Biochem. Biophys. Res. Comm., 202, 613 ± 620. Schrieber E, Matthias P, Muller MM and Schaner W. (1989). Nucl. Acid Res., 17, 6419. Stockschlaeder MAR, Storb R, Osborne WRA and Miller DA. (1991). Hum. Gene. Therapy, 2, 33 ± 39. Sun P, Enslen H, Myung PS and Maurer R. (1994). Genes Dev., 8, 2527 ± 2539. Wigler M, Pellicer A, Silverstein S and Axel R. (1978). Cell, 14, 725 ± 731.