Mammalian expression plasmids for wildtype mPea3 were kind gifts of Yvan de Launoit. pCDNA3 VP16 was constructed by cloning the acidic activation domain of VP16 transcription factor into the NcoI - Xba I sites of pCDNA3 vector. Pea3VP16 plasmid was constructed by cloning PCR amplification product of mouse Pea3 sequence into HindIII - XhoI sites of pCDNA3-VP16 plasmid. pCDNA3-Pea3 was constructed by cloning the PCR amplification product of mouse PEA3 sequence into the HindIII - XbaI sites of pCDNA3 vector. pEGFP-C3 was used in co-transfection experiments. Pea33-Luciferase reporter construct, where 5 × Pea3 optimal sites upstream of the TK minimal promoter drives the expression of the luciferase reporter gene.

Human Neurofilament M (NFM; promoter ID 39810) and Neurofilament L (NFL; promoter ID 41145) promoter sequences were retrieved from Cold Spring Harbor Laboratory “Transcriptional Regulatory Element Database” tool¹, and were analyzed for putative Pea3 protein binding sites with ALGGEN PROMO², a virtual laboratory to study transcription factor binding sites. Binding sites within the restricted NFM and NFL promoters were named as “ets” and sites with highest Pea3 binding affinity (with the dissimilarity rate below 7%) were selected to mutate or delete to study the trans-activation capacity of Pea3.

Cloning of wild type NFL and NFM promoters were cloned and mutations of the putative Pea3 binding regions on these promoters were introduced by using primers in Table 1. Amplification products of NFM and NFL promoters (both spanning a region from -953 to +64 for hNF-L and from -654 to +10 for hNF-M) were cloned into KpnI - HindIII sites of pGL2 - Basic vector (Promega).

Serine residues on mouse Pea3 protein that are potential targets for Proline-directed phosphorylation were mutated by two-step PCR-based site directed mutagenesis method. Essentially, forward and reverse primers were designed to replace the codon for serine with codon for either alanine or glutamic acid (Table 2). In the first step of PCR, 100 ng pcDNA3-mPea3 was mixed with the 1.25 U Pfu DNA polymerase (Promega), Pfu buffer (Promega) to final concentration 1X, dNTP mix to final concentration 200 μM each, and forward and reverse primers (at varied concentrations) with mutant codons were paired with ultimate reverse and ultimate forward primers that constraints wild type mouse Pea3 coding sequence, respectively, run at the 56°C annealing temperature. Amplicons from first step of PCR were run at 46°C for six cycles with 10× PrimeStar buffer (Takara) to final concentration 1×, PrimeStar DNA polymerase (Takara) to final concentration 1.25 U and PrimeStar (Takara) dNTP tofinal concentration 200 μM each to initiate annealing of overlapping mutant regions. To obtain the full length Pea3 gene coding sequence, second step of PCR was run at 56°C for 30 cycles with the previous reaction mix completed with ultimate forward and reverse primers and additional PrimeStar DNA polymerase and PrimeStar buffer.

Resultant PCR products were then cloned into the Hind III- XhoI sites of pcDNA3 and pCMV3-Tag-6. Bacterial transformation was done for the amplification of resultant plasmid constructs and for further validation analysis. The validation of the cloning was done by the colony PCR where positive bacterial colonies were directly used as a PCR template, then restriction enzyme digestion was done with Hind III-XhoI for further confirmation of the presence of Pea3 coding sequence. Plasmid clones from positive bacterial colonies were then purified for sequence analysis for final confirmation of presence of Pea3 sequence with desired mutation (data not shown). Pea3 phospho-mutant plasmids were further analyzed for proper gene expression and translation (data not shown).

PC12 pheochromacytoma cells were maintained in DMEM supplemented with 10% Horse serum and 5% Fetal Bovine serum in the presence of antibiotics, penicillin and streptomycin, and L-glutamine. Differentiation experiments were performed in collagen-coated plates, unless otherwise stated (Collagen Type IV, Sigma C0543). Transient transfections were carried out with Effectene reagent (Qiagen 301425), following manufacturer’s instructions. Nerve growth factor (NGF 2.5 S, Sigma N6009) was used commonly at 50-100 ng/ml. Epidermal growth factor (EGF, Sigma E4127) was used at 100 nM. Basic fibroblast growth factor (bFGF) and Insulin-like Growth Factor-1 (IGF-1) were used at final concentrations of 20 ng/ml and 100 ng/ml, respectively.

Mouse Motor Neuron (NSC-34) cell line and SH-SY5Y neuroblastoma cell line were grown in the high glucose DMEM (Gibco, 1129855) supplemented with 10% Fetal Bovine serum in the presence of penicillin, streptomycin, L-Glutamine and amphotericin B (Biological Industries, 03-033-1B) and primocin (Invivogen, ant-pm-1). Differentiation experiments were performed in collagen-coated plates (Collagen I, Gibco A10483-01). SH-SY5Y cells (10⁶ cells/dish) were treated with 0.3 μM of aphidicolin (Sigma, 10797) in serum-free medium. After 24 h, the cells were treated with 10 μM retinoic acid (RA, Sigma R2625) to induce neuronal differentiation. Cells were collected on day 0, 1, and 3 for further analysis as required.

Dorsal root ganglia (DRG) from adult rats were dissected, collected in RPMI medium with antibiotic-antimycotic and treated with collagenase (5000 U/ml) for 60 min followed by 0.25% trypsin/EDTA for 15 min. They were then transferred to RPMI medium containing 10% horse serum and 5% fetal bovine serum and gently triturated by 5-10 passages through a fire-polished Pasteur pipette. About 20 softened DRG were co-transfected with 3 μg Pea3 and 1 μg DsRed or with 1 μg DsRed alone in 100 μl Rat Neuron Nucleofector^® solution (Lonza, Austria) using the Amaxa Nucleofector (program O-03). Transfected neurons were plated onto poly-D-lysine/laminin pre-coated dishes. Cultures were maintained in RPMI medium with B27 supplement (Gibco Invitrogen) and antibiotic-antimycotic at 37°C in a humidified atmosphere with 5% CO₂. No approval for extracting tissue from dead animals was required. The DRGs used for cell culture were obtained from young adult rats immediately after exsanguinations. This procedure did not involve an animal experiment. Hence no Ethical Committee Approval was necessary.

For immunofluorescence, the cells (PC12 or NSC-34) were commonly seeded on 12-well plates. For differentiation experiments, cells were commonly co-transfected with GFP plasmid and live cells were routinely scored at 48 and 96 h after transfection under inverted microscope (Nikon ECLIPSE TE200). The expression plasmid-to-GFP ratio was commonly 3:1. Differentiation was scored as the percentage of neurite-bearing cells (at least two cell body length was taken as the cut-off point) among the GFP-expressing cell population. At least three different fields per well, and at least two different wells per transfection were scored, and average and standard deviation calculated using MS Excel software. The results were represented as the average of duplicate wells (three fields per well) in at least two independent experiments.

Total cytoplasmic RNA was prepared using commercial kits (RNAeasy kit, Qiagen, cat. no. 74104, or GeneJET RNA Purification kit, Thermo cat. no. #K0731) using manufacturer’s instructions. 1 μg (M-Mu-LV-Rtase, Roche) or 500 μg (New England BioLabs E6300L) RNA was used for each first strand cDNA synthesis reaction, as per instructions of the mentioned manufacturer, using random primers (Boehringer Mannheim). The amount of cDNA used was standardized using GAPDH and linear range determined. Typically the RT-PCR reactions were performed using 10–50 ng cDNA template in 25 μl reaction with BioTaq polymerase or 20 μl reaction with iTaq polymerase, at 50–64°C, as required, for 26–30 cycles. GAPDH primers (ADS1036/1037) were 5′ AGACAAGCTTCAGAGCCACCCGGGACC and 5′AGACTCTAGATCGGAGTCAACGGATTTGG; Pea3-specific primers (ADS1046/1047) were 5′ GACAAGCTTCGCCTACG ACTCAGATGTC and 5′ GACTCTAGAAGCTCCAATCCCTTCCTGC; Neurofilament-L primers were 5′ CAGTCTGGAGAACCTCGACC and 5′ TTCCAGGACCTTGTTCTGCT; Neurofilament-M primers were 5′AGGCATCGCACATCACGGTGGAG and GGATATTGTGATTGGGGGTCG. The products were resolved in 2.5 % Nu-Sieve agarose gels and were analyzed using QuantiOne imaging software (BioRad).

HEK293 and SH-SY5Y cells were grown in 1 g/ml DMEM (Sigma) supplemented with 10% fetal bovine serum (GIBCO-Invitrogen), and 1% antimycotic-antibiotic (GIBCO-Invitrogen). 24 h prior to transfection, cells were seeded into 24-well plates with the density of 50 thousand cells per well. Plasmid transfections of cells were performed using 1.5 μl of TransFast^TM (Invitrogen) for reporter assays with NFM and NFL promoter reporter constructs. Essentially, plasmids were transfected per well in the following amounts: 100 ng Renilla Luciferase (Promega), 200 ng reporter construct (NFM-Luc or NFL-Luc) and wild type Pea3 expression plasmid with varying concentrations as indicated in the text (usually 5, 25, 50, 100, 150, and 200 ng, completed up to 200 ng DNA in total with pCMV empty vector). After 48 h, cells were harvested and lysed for 15 min with Luciferase Cell Culture Lysis 5× Reagent (Promega) diluted to 1× with PBS. Thirty microliter of cell lysate triplicate for each sample was transferred to the white opaque luminometer microtiter 96-well plate and mixed first with the 30 μl Dual-Glo^® Luciferase Substrate to measure luciferase activity from promoter of interest and later 30 μl Dual-Glo^® Stop & Glo^® Substrate to measure luciferase activity from internal control Renilla Luciferase. Luciferase activity was measured with Luminoskan Ascent (ThermoLab Systems) for 10 s for each substrate.

Chromatin Immunoprecipitation (ChIP) assay was carried out with EZ-ChIP^TM ChIP Kit (Catalog # 17-371) according to the manufacturer’s instructions. Essentially, DNA was sheared enzymatically with the micrococcal nuclease. Some of the sheared DNA was saved as input, and rest of the sample was precipitated using 30 μl of anti-FLAG M2 affinity resin (Sigma) that was previously resuspended in Triton lysis buffer to precipitate 100 μl of Flag-tagged Pea3 protein that is bound to sheared DNA. Pea3 protein bound DNA was then eluted by reverse crosslink reaction in high salt environment enhanced with heat treatment. DNA in the samples was purified by using PureLink PCR Purification Kit (Invitrogen). Purified DNA from both input and ChIP samples were then detected by q-PCR (NFM ets-1 spanning region forward primer: 5′-AAGGGCAGGGTGAACTGGACT-3′; NFM ets-1 spanning region, reverse primer: 5′-TTCTTTAGCCTTCTACCCTCTTATCCTC- 3′). Percent Input was calculated so as to normalize qPCR data, and essentially signal intensities obtained from the ChIP samples were divided by signal intensities obtained from the Input sample.

In order to study the effects of Pea3 directly in differentiation, a chimeric protein with full length mouse Pea3 and the potent acidic activation domain of VP16 transactivator of the Herpes Simplex Virus (HSV) was constructed. The fusion protein was confirmed to activate from an artificial luciferase reporter (data not shown). When exogenous Pea3-VP16 fusion was expressed in PC12 cells, it induced neurite formation within 2 days, regardless of collagen-coating (Figure 1A). The percentage of neurite-bearing cells was significantly increase, up to around 25%, when Pea3-VP16 was transfected to PC12 cells (Figure 1B). Expression from an SRE-Luciferase reporter in PC12 cells was not greatly altered upon stimulation of PC12 cells that were serum starved for 36 h and stimulated with either EGF or NGF overnight, whereas cells transfected with exogenous mPea3 showed significant increase in luciferase reporter activity (Figure 1C), indicating that even mPea3 without any VP16 fusion was transcriptionally active in this context.

Although a constitutively active Pea3-VP16 fusion protein is useful in studying the effects of an activated Pea3, it does not represent a physiologically relevant situation. Therefore, PC12 cells were next transfected with mouse Pea3, either in the presence or absence of Epidermal Growth Factor (EGF). In contrast to Nerve Growth Factor (NGF), which results in sustained MAPK activation and neuronal differentiation, EGF is known to lead to proliferation in these cells through transient activation of the MAPK pathway (Marshall, 1995; Aksan and Kurnaz, 2003). Although EGF is not sufficient to drive differentiation of PC12 cells on its own, in combination with exogenous mPea3 it was seen to enhance long axonal projections (Figures 1D,E).

In order to further analyze whether Pea3 would have the same effect on axonal outgrowth in adult neurons, we have studied adult DRG cultures transfected with mPea3 expression vectors, in combination with EGF, bFGF, and IGF-1. It was observed that IGF-1 alone could cause a significant increase in total axonal growth (around 2000 microns), while bFGF alone could not (around only 1000 microns) in DRGs transfected with DsRed plasmid alone (Figures 2A,B). When these neurons are co-transfected with DsRed and mPea3 expression vector and then stimulated, it was observed that while stimulation of mPea3-expressing cells with bFGF did not cause a significant axonal growth, stimulation with EGF caused a nearly twofold increase (although variability was high between different DRG cultures due to transfection efficiency), and IGF-1 on mPea3-expressing cells caused a nearly threefold increase in total axonal growth (Figure 2B).

The results were similar in maximal distance growth analyses, with EGF and IGF-1 stimulations equally enhancing total growth when mPea3 was overexpressed in cells, as compared to DsRed alone (Figure 2C). In terms of number of branches, however, presence or absence of exogenous mPea3 did not have a significant effect; there was only a slight enhancement in number of branch points when mPea3-expressing cells were stimulated with EGF or IGF-1 (Figure 2D).

NSC-34 is a commercially available motor neuron cell line (produced through the fusion of embryonic mouse spinal cord cells with neuroblastoma cells); they are capable of differentiation as readily monitored by extensive axonal outgrowth, and are readily transfected, making them valuable developmental models. When these model cells were co-transfected with an empty expression vector, pCDNA3, and pEGFP, there was only background spontaneous differentiation in around 5% of the GFP-positive cells (Figure 3A), while transfection of Pea3 along with pEGFP plasmid lead to differentiation of around 15-20 % of the GFP-positive cells. pCDNA3-VP16/pEGFP co-transfection resulted in around 10% differentiation, whereas pCDNA3-Pea3VP16/pEGFP co-transfection resulted in well over 20% differentiation in GFP-positive cells (Figure 3A). Interestingly, Pea3-EN, a dominant negative form of Pea3 in fusion with the repression domain of Engrailed transcription factor, also showed some enhanced differentiation (around 15%), although this was lower than differentiation induced by either Pea3 or Pea3VP16. Representative images of NSC-34 differentiation experiments are shown in Figure 3B. While the motor neuron-derived cells normally appear round in morphology, in Pea3- or Pea3-VP16-transfected cells rather extensive axon-like neurite outgrowths of more than 2 cell body length are readily observed and scored as differentiation.

Since phosphor-specific Pea3 antibodies are not commercially available, we have generated phosphor-mutants of Pea3; particularly Serine 90 was shown to be important for transcriptional activity, as in both PC12 and SH-SY5Y cells Pea3S90A mutants showed basal level of activity on SRE-Luciferase reporter, while Pea3S90E mutants showed even enhanced activity as compared to wildtype Pea3 (data not shown). When NSC-34 cells were transfected with the wildtype mPea3 expression plasmid, differentiation was observed to increase nearly threefold, around 15%, as compared to cells transfected with empty pCDNA3 plasmid, as in the previous assay (Figure 3C; compare to Figure 3A). Exogenous expression of a transcriptionally defective Pea3-S90A mutant resulted in basal level of differentiation, whereas that of a transcriptionally active Pea3-S90E mutant resulted in differentiation to the same level as wildtype Pea3 (Figures 3C,D).

Since in experimental model systems Pea3 overexpression lead to morphological changes similar to differentiation, the next question was identifying which target genes would be responsible for this transformation. The known transcriptional targets of Pea3 mostly include matrix metalloproteases, such as MMP-2 and MMP-9, identified in the context of breast or prostate cancer models. Therefore, in an attempt to identify more neurologically relevant targets of Pea3, we have initially started investigating genes that are correlated with neuronal differentiation and whose promoters contained putative Pea3-binding motifs.

Upon initial bioinformatic analysis, two neuron-specific promoters were particularly striking with respect to their putative binding motifs: neurofilament-L and neurofilament-M. The promoter sequences for these genes were obtained through the Transcriptional Regulatory Element Database (TRED) Cold Spring Harbor Laboratory³, and thereafter analyzed by ALGGEN PROMO⁴, a virtual laboratory for the study of transcription factor binding sites in DNA sequences, to identify potential Pea3 binding motifs. Human Neurofilament-L (NF-L) promoter contained only two Pea3-specific ets motifs, with 6.6 and 3.9% dissimilarity scores (Figure 4A; it should be noted that mouse NF-L promoter contains 4 ets motifs, two with 0% and two with 1% dissimilarity scores, and rat NF-L promoter contains 3 ets motifs, with 0, 1, and 6% dissimilarities; not shown). Human NF-M promoter, on the other hand, contained four ets motifs, only one of which shared significant homology to consensus Pea3-binding motif (1% dissimilarity; Figure 4A). The other three ets motifs in the human NF-M promoter indicated 3, 4, and 6 % dissimilarities; however when mouse and rat promoters were analyzed, rat promoter was found to have 4 ets motifs, with 0.4, 5, 5, and 1.7 % dissimilarities, and mouse promoter was found to have again 4 ets motifs with 6, 5, 3, and 1.7 % dissimilarities, data not shown). The ets motifs predicted on the Neurofilament-H promoter yielded much lower similarities (not shown), hence not included in this study. We have therefore proceeded with our analyses, focusing on the only 2 ets motifs in the Neurofilament-L promoter, and the highest scoring ets motif in the Neurofilament-M promoter, in order to estimate whether these promoters indeed would respond to Pea3-induced transcriptional regulation.

Our first approach was to confirm whether Pea3/ETV4 message was indeed upregulated, albeit modestly, in human SH-SY5Y cells in response to stimulation with IGF-1 treatment, since IGF-1 caused the most prominent neurite extension in the DRG studies. To that end, we have serum starved SH-SY5Y overnight, and treated them with IGF-1 for 30 and 60 min. RT-PCR experiments confirmed a modest increase in Pea3/ETV4 expression at 60 min, parallel to a comparable increase in hNF-L expression (Figure 4B). In a complementary experiment, we have differentiated SH-SY5Y cells in the presence of Retinoic Acid (RA) and aphidicolin, as described in Materials and Methods, for 3 days (Figure 4C). Pea3 expression levels gradually increase with retinoic acid and aphidicolin treatment even at Day 1 (close to 2- and 2.5-fold in the absence and presence of treatment, respectively; this slight increase even in the absence of treatment can be due to spontaneous differentiating cells due to crowding). This increase went up to around threefold in the absence and around 3.5-fold in the presence of treatment on day 3 (Figure 4C). When NF-L levels were similarly analyzed, a modest increase (around 1.2-fold) was detected upon stimulation with retinoic acid and aphidicolin treatment on day 1, which went up to 1.7-fold in the absence and 2.2-fold in the presence of treatment on day 3. Yet, due to already high basal levels of endogenous NF-L in cell lines, the most prominent change was observed in the level of NF-M expression, with twofold increase on day 1 and around fourfold increase on day 3 of treatment (Figure 4C).

In order to analyze whether this increase in NF-L and NF-M expression was a direct consequence of transcriptional regulation by Pea3, we have initially cloned the hNF-L promoter upstream of a luciferase coding sequence, as described in Section “Materials and Methods” (Figure 5A). When SH-SY5Y cells transfected with hNFL-Luc reporter were cultured in the presence or absence of NGF, a basal level of stimulation was observed; co-expression of Pea3 in these cells in the absence of stimulation resulted in modest increase in reporter activity, whereas stimulation with NGF in the presence of exogenous Pea3 resulted in more than twofold activation when compared to Pea3 alone (Figure 5B). A similar upregulation was observed when cells were stimulated with retinoic acid in addition to exogenous Pea3 expression (Figure 5C). To confirm that the ets motifs were indeed important for this regulation, we have then constructed ets motif deletion constructs, termed NFLΔ1, NFLΔ2, and NFLΔ1Δ2, referring to deletion of ets1, deletion of ets2, or deletion of both ets1 and ets2 motifs. In the absence of exogenous Pea3 transfection to the cells, deletion of either or both ets motifs resulted in loss of basal level activation from the hNF-L promoter (Figure 5D). However when these deletion constructs were co-transfected to SH-SY5Y cells together with Pea3 expression plasmid, deletion of ets1 motif resulted in almost complete repression of reporter activity (Figure 5E), indicating that this motif is important for Pea3-induced transcription from this promoter, whereas deletion of ets2 motif resulted in only a modest decrease, if at all; strangely, deletion of both ets1 and ets2 resulted in almost restoration of basal level activity, indicating that ets2 may in fact be a repressive element of NF-L promoter.

When a similar analysis was carried out for the hNF-M promoter, which contains a single ets motif with significant homology to consensus Pea3 binding site (Figure 6A), it was observed that hNF-M-Luc reporter showed a nearly twofold activitation in response to Pea3 as compared to empty reporter, which was significantly repressed when ets motif was deleted (NFMΔ; Figure 6B). Since this highest-scoring ets motif appeared to be important for transcriptional activation, we have then analyzed Pea3 binding through ChIP assay: essentially, we have co-transfected SH-SY5Y cells with pCDNA3-Pea3-Flag expression plasmid along with either pGL2-NFM or pGL2-NFMΔ reporter plasmid, and immunoprecipitated Pea3 protein as described in Section “Materials and Methods,” thereafter amplified the ets-containing promoter region by PCR. A reaction was observed only when Pea3 and hNFM-Luc were co-transfected to the cells, but not the deletion reporter construct (Figure 6C), indicating Pea3 indeed could bind to this ets motif on the hNFM promoter.