Address correspondence and reprint requests to Alan J. Stewart, School of Medicine, University of St Andrews, St Andrews, Fife, KY16 9TS, United Kingdom. E-mail: firstname.lastname@example.org.
Phospholipase C-η2 is a recently identified phospholipase C (PLC) implicated in the regulation of neuronal differentiation/maturation. PLCη2 activity is triggered by intracellular calcium mobilization and likely serves to amplify Ca2+ signals by stimulating further Ca2+ release from Ins(1,4,5)P3-sensitive stores. The role of PLCη2 in neuritogenesis was assessed during retinoic acid (RA)-induced Neuro2A cell differentiation. PLCη2 expression increased two-fold during a 4-day differentiation period. Stable expression of PLCη2-targetted shRNA led to a decrease in the number of differentiated cells and total length of neurites following RA-treatment. Furthermore, RA response element activation was perturbed by PLCη2 knockdown. Using a bacterial two-hybrid screen, we identified LIM domain kinase 1 (LIMK1) as a putative interaction partner of PLCη2. Immunostaining of PLCη2 revealed significant co-localization with LIMK1 in the nucleus and growing neurites in Neuro2A cells. RA-induced phosphorylation of LIMK1 and cAMP-responsive element-binding protein was reduced in PLCη2 knock-down cells. The phosphoinositide-binding properties of the PLCη2 PH domain, assessed using a FRET-based assay, revealed this domain to possess a high affinity toward PtdIns(3,4,5)P3. Immunostaining of PLCη2 together with PtdIns(3,4,5)P3 in the Neuro2A cells revealed a high degree of co-localization, indicating that PtdIns(3,4,5)P3 levels in cellular compartments are likely to be important for the spatial control of PLCη2 signaling.
Phospholipase C-η (PLCη) enzymes are a recently identified class of phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2)-hydrolyzing enzyme found in mammalian cells (Hwang et al. 2005; Nakahara et al. 2005; Stewart et al. 2005; Zhou et al. 2005). This class consists of two members, PLCη1 and PLCη2. Like other mammalian PLCs, their activity leads to the production of the second messengers Ins(1,4,5)P3 and 1,2-diacylyglycerol (DAG), which trigger the release of Ca2+ from intracellular stores and the activation of protein kinase C isoforms, respectively.
Phospholipase C-η2 is most abundant in the brain and is found within the hippocampus, habenula, olfactory bulb, cerebellum, and throughout the cerebral cortex (Nakahara et al. 2005; Kanemaru et al. 2010). Activation of PLCη2 can be triggered either by intracellular Ca2+ immobilization (Popovics et al. 2011) or by Gβγ signaling (Zhou et al. 2005, 2008). Considering its sensitivity of toward Ca2+, it is thought that PLCη2 may act synergistically with other PLCs or Ca2+ activated processes in neurons (Popovics and Stewart 2012). When transiently expressed in COS7 cells, PLCη2 associates with plasma and organelle membranes via its PH domain (Popovics et al. 2011). However, the specificity of this domain toward particular phospholipids, which is likely to control its cellular location, remains unknown.
Although a role in the amplification of PLCβ signals has recently been demonstrated for PLCη1 (Kim et al. 2011), a functional role has yet to be identified for PLCη2. Nakahara et al. previously reported that levels of PLCη2 protein increase in the brains of mice during the first few weeks after birth (Nakahara et al. 2005). Similarly, the retina expresses high levels of PLCη2 which also elevate from birth (Kanemaru et al. 2010). Its expression profile hints that this enzyme might be involved in processes linked to neuronal differentiation or maturation. In this study, we examined the importance of PLCη2 during retinoic acid (RA)-induced differentiation of Neuro2A cells. This system was chosen as it provides a well-defined cellular model of neuronal differentiation that has been used to characterize the importance of various factors in this process (Pignatelli et al. 1999; Carter et al. 2003; Bryan et al. 2006; Kouchi et al. 2011), and these cells are known to express endogenous PLCη2 (Kim et al. 2011). Stable shRNA-mediated knockdown of PLCη2 expression in these cells revealed this enzyme to be important for RA-stimulated neurite outgrowth. Reductions in PLCη2 expression in isolated clones led to reductions in RA signaling and LIM kinase-1 (LIMK1) phosphorylation. A bacterial two-hybrid screen identified LIMK1 as a putative interaction partner for PLCη2. Furthermore, it was found that PLCη2 co-localizes with LIMK1 within the nucleus and growing neurites in Neuro2A cells and that PLCη2 associates with intracellular PtdIns(3,4,5)P3 pools via its PH domain.
Materials and methods
The GFP-tagged murine PLCη2 construct incorporating residues 75–1238 was generated as described previously (Popovics et al. 2011), this region of the protein includes all functional domains (PH, EF-hand, catalytic X and Y and C2 domains). Plasmids (psi-H1) encoding control and PLCη2-targeted shRNAs were purchased from GeneCopoeia (Rockville, MD, USA). These plasmids contained the 19-mer target sequences 5′-TGACTGCAAGCTCCTCAAT-3′ or 5′-GGACCTAGTGAAATATACC-3′, which were used to generate the shRNAPLCη2-1 and shRNAPLCη2-2 cell lines, respectively. The glutathione S-transferase (GST)-tagged PLCη2 PH domain expression construct was made by inserting the corresponding cDNA fragment into pGEX6 (GE Healthcare, Buckinghamshire, UK). The same sequence was subcloned into the pcDNA3.1/NT-GFP-TOPO vector (Invitrogen, Paisley, UK) to generate a GFP-tagged PLCη2 PH domain encoding construct. The mCherry-tagged PLCδ1 PH domain expression vector was made by inserting the corresponding cDNA fragment into pmCherry-C1 (Clonetech, Saint-Germain-en-Laye, France).
Neuro2A cells were obtained from the European Collection of Cell Cultures (Salisbury, UK) and were maintained in Eagle's minimal essential medium (EMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM l-glutamine, and 50 units/mL of penicillin/streptomycin. Differentiation of Neuro2A cells was induced as previously described by Zeng and Zhou (2008). Briefly, cells were plated at a low density (100 cells/mm2) in complete EMEM which was changed to the differentiation medium the next day (EMEM with 2% FBS, 2 mM l-glutamine and 20 μM RA). Medium was renewed every day for 4 days. Micrographs showing cells at different stages of differentiation were collected by a Zeiss Axiovert 40 CFL microscope with a 10× objective (Carl Zeiss Ltd., Cambridge, UK). To assess neurite growth, four micrographs were taken by random selection from all experimental conditions, and experiments were repeated three times. All cells possessing at least one neurite with a length at least twice the cell body were accepted as differentiated. Results were expressed as a percentage of differentiated/total cell number. Neurite outgrowth was estimated based on a previously described stereological method (Lucocq 2008). Neuro2A cells were stably transfected with shRNA expressing plasmids to produce cell lines with reduced expression of PLCη2. In each case, 20 μg of DNA was transfected by electroporation using a Bio-Rad Gene Pulser (Hertfordshire, UK) at 230 V, 950 μF. Stably, transfected cells were selected 48 h after transfection by addition of media containing 3 μg/mL puromycin.
Cells were seeded onto 6-well plates and differentiated as described above. One well of cells was scraped in 50 μL radio-immunoprecipitation assay buffer containing complete protease inhibitor cocktail (Roche, Burgess Hill, UK) and cell debris was removed by centrifugation (16 000 g for 15 min). Protein concentration of the lysates was measured using the bicinchoninic acid protein assay kit (Thermo Fisher Scientific, Surrey, UK). Samples (corresponding to 10 μg total protein) were separated by NuPAGE gradient (4–12% Bis-Tris) gel electrophoresis using a MES buffer system (Life Technologies, Paisley, UK) then electro-blotted onto polyvinylidene difluoride membrane. Blots were incubated for 1 h at 20°C in blocking buffer (5% non-fat milk in TBS) before being incubated for 90 min in 1% non-fat milk and 0.5% Tween-20 in Tris-buffered saline (TBS) containing either custom-made rabbit polyclonal anti-PLCη2 antibody (1 : 5000 dilution), mouse polyclonal anti-LIMK1 or rabbit anti-phospho-LIMK1 (Thr508) antibodies (1 : 1000 dilution; Abcam, Cambridge, UK), rabbit anti-CREB or anti-phospho-CREB (Ser133) antibodies (1 : 1000 dilution; Cell Signaling Technology, Danvers, MA, USA). The custom-made antibody was raised commercially against a short peptide (SKVEEDVEAGEDSGVSRQN; EZBiolab, Westfield, IN, USA). Blots were washed three times in TBS containing 0.5% Tween before incubating with an appropriate species-specific secondary horseradish peroxidase-conjugated antibody (1 : 10 000 dilution; Thermo Fisher Scientific) in blocking buffer containing 0.5% Tween for 1.5 h at 20°C. Blots were washed three times with TBS containing 0.5% Tween and antibody-binding was detected using the SuperSignal West Dura chemiluminescent substrate (Thermo Fisher Scientific) with a LAS-3000 phosphoimager (Fujifilm, Düsseldorf, Germany). Densitometry was performed using ImageJ software (NIH, Bethesda, MD, USA).
Inositol phosphate release assays
IP-release assays were performed in triplicate as previously described (Popovics et al. 2011). Briefly, cells were seeded onto 14 cm diameter dishes and treated the next day with serum- and inositol-free Dulbecco's modified Eagle's medium (MP Biomedicals, Illkirch, France) containing 1 μCi/mL myo-D-[3H]inositol (GE Healthcare, Buckinghamshire, UK) for 24 h. On day 3, cells were trypsinized, counted, and collected by centrifugation. Cell pellets were diluted in buffer A (140 mM NaCl, 20 mM HEPES, 8 mM glucose, 4 mM KCl, 1 mM MgCl2, 1 mM CaCl2, and 1 mg/mL bovine serum albumin) to equal cell density. Thereafter, equal cell numbers (5 × 105/tube) were transferred into buffer A containing 10 mM LiCl. Tubes were incubated for 1 h before the addition of A23187 (Ca2+-ionophore) to a final concentration of 5 μM. Cells were stimulated for 3 h and reaction was terminated by the addition of a final 10 mM formic acid. This was incubated for 30 min at 4°C. Cell debris was removed by centrifugation (2 000 g for 3 min) and supernatant was transferred onto Dowex AG 1- resin (BioRad, Hemel Hempstead, UK). Resin was washed with water and 60 mM ammonium formate/5 mM sodium tetraborate. Inositol phosphates were eluted with 1 M ammonium formate/0.1 M formic acid and quantified by liquid scintillation counting.
RT-PCR and qPCR
Neuro2A cell mRNA was isolated by the Isolate RNA Mini Kit (Bioline, London, UK) according to the manufacturer's instructions. DNA contamination was removed by DNAse digestion (RQ1 RNase-Free DNase, Promega, Southampton, UK). Following this 0.5 μg of mRNA was subjected to a two-step reverse transcription reaction. The mRNA was incubated with 100 pmol oligodT18 and 0.5 mM dNTP in Diethylpyrocarbonate (DEPC)- treated water at 65°C for 5 min and then chilled on ice. RevertAid Premium Reverse Transcriptase (200 units), RiboLock RNase Inhibitor (20 units; Thermo Fisher Scientific), and 1 × RT-buffer (Fermentas, St. Leon-Rot, Germany) were added. Mixtures were incubated at 50°C for 30 min and then 60°C for 10 min. Reactions were terminated at 85°C for 5 min. The presence of RARα, RARβ, and RARγ was detected by RT-PCR using primers designed not differentiate between splice variants (sequences are shown in Table 1). PCR reactions contained 1.25 units of GoTaq polymerase (Promega), 1 × reaction buffer, 1.5 mM MgCl2, 0.2 mM dNTPs, and 250 nM of each primer. Reactions were cycled at 95°C for 15 s, 60°C for 30 s and 72°C for 1 min. The expression level of RARα in control, shRNAPLCη2-1, and shRNAPLCη2-2 cells was measured by qPCR relative to the level of the large ribosomal protein P0 mRNA. Expression was detected by Brilliant III Ultra-Fast SYBR Green mix (Agilent Technologies, Cheshire, UK) in a Rotor-Gene Q PCR machine (Qiagen, Crawley, UK). Expression levels were calculated by the ΔΔCt method.
Table 1. Primer sequences used for semi-quantitative and quantitative assessment of RAR isoform expression
RARE luciferase assay
RA response element (RARE) regulated transcriptional activity was measured by transiently transfected luciferase constructs. RARE reporter vector expressing luciferase and a CMV promoter driven luciferase containing construct were obtained from SABiosciences (Qiagen). Stable Neuro2A cells were seeded at 5 × 104 cell/well into a 96-well plate in antibiotic-free EMEM with 10% FBS. Cells were transfected with the luciferase constructs the next day using Lipofectamine 2000 (Life Technologies) in accordance with the manufacturer's instructions. The media was changed to serum- and antibiotic-free EMEM after 6 h. Cells were treated the next day with RA or dimethylsulfoxide only for 4 h and reaction and luciferase activity measured using the Dual Luciferase Reporter assay kit (Promega) with a Fluostar Optima plate reader (BMG Labtech, Ortenberg, Germany). The parameters of each measurement were a combination of a 2-s pre-measurement delay and a continuous measurement for 10 s. The assays were performed in triplicates.
Neuro2A cells were seeded onto coverslips in 6-well plates with a low cell density (100 cells/mm2). In some cases, Neuro2A cells were transfected with GFP- or mCherry-tagged constructs using Lipofectamine 2000 transfection reagent in accordance with the manufacturer's protocol. The next day, or after a 2-day or 4-day treatment with 20 μM retinoic acid, cells were fixed with 4% formaldehyde in phosphate-buffered saline (PBS) for 10 min. Nuclei were stained with 4′,6-diamidino-2-phenylindole and coverslips were mounted with Mowiol solution. For the detection of endogenous proteins, cells were fixed and permeabilized in one step with 4% formaldehyde and 0.2% Triton-X containing PBS. This solution was removed after 5 min and cells were washed with PBS three times. Coverslips were then incubated with blocking buffer containing 10% FBS in PBS for 1 h. Custom-made rabbit primary antibody (generated as described above) was used to stain endogenous PLCη2 (dilution: 1 : 100). Antibodies against LIMK1 (1 : 100 dilution; Abcam) and PtdIns(3,4,5)P3 (1 : 100 dilution; Echelon Biosciences Inc., Salt Lake City, UT, USA) were also used for immunostaining. Primary and secondary antibodies were diluted into appropriate concentrations in 5% FBS in PBS. Cells were incubated with primary antibody solution for an hour followed by three washing steps with PBS. Appropriate fluorescent species-specific secondary antibodies (Dylight 488 or 594; Jackson ImmunoResearch, West Grove, PA, USA) were diluted 1 : 300 and applied onto the coverslips for 1 h. Nuclei were stained with 4′,6-diamidino-2-phenylindole and coverslips were mounted in Mowiol solution. Slides were examined with a Zeiss Axioplan fluorescent microscope system, a DeltaVision deconvolution microscope (Applied Precision, Washington, DC, USA) or a Leica TCP SP2 multiphoton confocal microscope (Leica, Heidelberg, Germany).
Bacterial two-hybrid screen
The BacterioMatch II Two-Hybrid System (Stratagene, Cambridge, UK) was used to identify putative protein interaction partners for PLCη2 according to the manufacturer's protocol. Briefly, cDNA corresponding to the C-terminal domain of PLCη2 (residues 967-1070 of splice variant 21a/22/23) was cloned into the pBT bait plasmid and was screened against a human pancreas cDNA library (Stratagene, Cheshire, UK) encoded within the plasmid pTRG. In this system, interaction between the C-terminal domains of PLCη2 and a protein encoded by a library-derived gene product allows the transcription of His3 and Aad3 genes in the reporter E. coli strain. This permits the selection of the interaction-pairs via the addition of 3-aminotriazole (3-AT) (an inhibitor of histidine synthesis) or by streptomycin resistance, respectively. The pBT-PLCη2-C-term and pTRG-library plasmids were co-transformed into E. coli reporter cells, in accordance with the manufacturer's instructions. Initially, transformed cells were spread onto plates containing minimal media with 5 mM 3-AT in the first instance. Resultant colonies were also plated on dual-selective plates containing 5 mM 3-AT and 12.5 μg/mL streptomycin to confirm a positive interaction between bait and target proteins. Colonies were picked, plasmid DNA was isolated, and pTRG plasmid inserts were sequenced commercially (Dundee Sequencing Service, Dundee, UK). The pTRG plasmid inserts were then identified using NCBI-BLAST (Altschul et al. 1997).
FRET-based assessment of PLCη2 PH domain-phosphoinositide binding
PH domain-phosphoinositide interactions were measured using a FRET-based assay (Gray et al. 2003). GST-conjugated PLCη2 PH domain was expressed in E. coli strain BL21 and affinity-purified using Glutathione beads (Sigma, Poole, UK). Biotinylated and non-biotinylated lipids were obtained from Cell Signals, Inc. (Lexington, KY, USA). FRET assays were performed in assay buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM MgCl2, 5 mM dithiothreitol, 0.05% Chaps). PLCη2-PH (14 μL of 0.62 mg/mL) was mixed with 4 μL APC-streptavidin (2 mg/mL, Prozyme Ltd. Hayward, CA, USA), 80 μL Eu-anti-GST antibody (80 μg/mL) (Perkin-Elmer, Buckinghamshire, UK) in 2.5 mL assay buffer. From this mixture, 25 μL was added to the same volume of serially diluted (0–50 pmol in final volume) biotinylated lipids and the association of biotinylated phosphoinositide/PH domain complexes was measured in a 96-well plate (White Lumitrac 200; Greiner Ltd) in an LJL Analyst. In a separate experiment, displacement of PLCη2-PH from a pre-formed FRET complex was measured by the addition of increasing concentrations of phosphoinositides. PtdIns(4,5)P2-PLCη2 PH domain FRET complexes were formed by mixing 4 μL (2 mg/mL) Streptavidin APC, 200 pmol biotin PtdIns(4,5)P2, 14 μL (0.62 mg/mL) PH domain and 50 μL Eu-anti-GST (80 μg/mL) in 2.5 ml assay buffer. Displacing ligands were serially diluted from 200 μM in 5-fold steps in assay buffer, 25 μL added to 25 μL the FRET complex and any decrease in signal was measured as previously described (Gray et al. 2003).
Experiments were repeated independently at least three times and all assays were performed in triplicate. Data are presented as mean values ± SEM. Analyses were performed using the SigmaPlot software (Systat Software Inc, Chicago, IL, USA) with one-way anova followed by the Holm-Sidak post hoc test or Student's t-test.
PLCη2 expression increases during RA-induced Neuro2A cell differentiation
Neuro2A cells were previously described as being highly susceptible to the combination of RA-treatment, reduced serum (2% FBS), and low cell plating density (100 cells/mm2) (Zeng and Zhou 2008). Following this protocol, Neuro2A cells differentiate over a period of 4 days altering their morphology from a stem cell-like to a neuron-like shape with extended neurites and enlarged cell body (Fig. 1a). This model was used to examine PLCη2 protein levels during neuronal differentiation. The dominant expression of one splice variant was confirmed in Neuro2A cells. This was the 21a/22/23 form, previously identified by Zhou et al. (2005). Short-term treatment (4 h) with RA resulted in a 76% increase in PLCη2 protein level. Moreover, PLCη2 levels gradually increased throughout the differentiation process and were ~10-fold higher than basal on day 4 (Fig. 1b and c).
PLCη2 is involved in RA-induced Neuro2A cell differentiation
We next examined whether Neuro2A cell differentiation may be affected by reducing PLCη2 expression. Appropriate ‘knock-down’ cell lines were generated by stable expression of shRNAs targeting PLCη2 expression. Stable clones were selected by puromycin resistance following transfection of the shRNA-encoding constructs into Neuro2A cells. PLCη2 protein levels in single-cell-derived clones for were determined by Western blot. Two lines (shRNAPLCη2-1 and shRNAPLCη2-2) were chosen for further experiments. A control cell line (shRNAcontrol) produced by the stable transfection of control shRNA was also selected. The protein level of PLCη2 in each was calculated and was found to be reduced by 67% and 39% in shRNAPLCη2-1 and shRNAPLCη2-2 cell lines, respectively (Fig. 2a and b). The cellular effects of PLCη2 knockdown in Neuro2A cells during RA-stimulated differentiation were then investigated. Control and knock-down cells were treated with RA (20 μM) for 4 days and bright images were taken with a 10× objective on each day. Representative images of each cell line following 4 days of treatment are shown in Fig. 2c. The proportion of differentiated cells was significantly affected by PLCη2 knockdown with the level of reduction largely mirroring the expression level of PLCη2. The shRNAPLCη2-1 cell line, which exhibited the greatest decline in PLCη2 level, showed very little neurite growth, whereas the shRNAPLCη2-2 cells formed some short neurites. To better evaluate these results, the rate of differentiation was calculated from three independent experiments on each day (Fig. 2d). Cells were defined as differentiated if possessing at least one neurite with a length equal to or longer than twice the cell body diameter. Significant differences between the differentiation of control and knock-down cell lines were observed after 24 h of treatment. This difference gradually increased over the 4 days. On day 4, the proportion of differentiated shRNAcontrol cells was 73%, whereas only 8% and 22% of the shRNAPLCη2-1 and shRNAPLCη2-2 cells were defined as being fully differentiated. For each of the three cell lines, total neurite outgrowth (sum of neurite lengths) was also calculated based on a previously described stereological approach (Ronn et al. 2000; Lucocq 2008). Total neurite outgrowth was found to mirror the results obtained by calculating the percentages of differentiated cells (Figure S1). An inositol phosphate release assay was employed with the three cell lines following treatment with 5 μM A23187 (Ca2+ ionophore) to examine the phospholipid turnover in response to Ca2+. Both the shRNAPLCη2-1 and shRNAPLCη2-2 cells exhibited a significantly decreased inositol phosphate release relative to control cells after this treatment (Fig. 2e).
PLCη2 regulates RA receptor signaling
To gain some mechanistic information as to how PLCη2 contributes to RA-induced Neuro2A differentiation, we examined whether RA signaling is affected by reduced PLCη2 expression. The cellular effects of RA are driven by its nuclear receptor (RAR) which directly regulates gene expression upon binding of its ligand. A partner, the retinoid X receptor, with which it forms a heterodimer, is also involved in this process. The receptor–ligand complex binds to specific promoter regions of the target genes, termed as RAREs (Lane and Bailey 2005). Three variants of retinoid receptors are known to exist, RARα, RARβ, and RARγ (Chambon 1996). RT-PCR analysis performed using and RAR isoform specific primers revealed that only RARα is expressed in these cells (Fig. 3a). Quantitative PCR revealed that RARα mRNA expression did not differ significantly between the shRNAPLCη2-1, shRNAPLCη2-2 and control lines (Fig. 3b). A dual-luciferase reporter assay was employed to measure RA-induced transcriptional activity in each cell line. This included the co-transfection of two expression constructs (Fig. 3c); one containing the RARE sequence (5-AGGTCACCAGGAGGTCA-3′) followed by the firefly luciferase gene and one vector constitutively expressing luciferase to allow transfection efficiency to be assessed. Interestingly, both PLCη2 knock-down cell lines exhibited reductions in basal RA-regulated signaling compared with the control cells (Fig. 3d). Luciferase activities in shRNAPLCη2-1 and shRNAPLCη2-2 cells were reduced by 60% and 30% to that of control cells, respectively. The difference between shRNAPLCη2-1 and shRNAPLCη2-2 was significant, which is likely because of the lower level of PLCη2 in shRNAPLCη2-1. RARE activation in the RA-stimulated cells was also examined (Fig. 3e). Luciferase activity increased 101-fold in shRNAcontrol cells following treatment with RA (20 μM). Under the same conditions, shRNAPLCη2-1 and shRNAPLCη2-2 cells exhibited only 35-fold and 44-fold elevations, respectively.
LIMK1 as a putative interaction partner of PLCη2
Phospholipase C-η enzymes possess an extended C-terminal domain rich in proline, serine and threonine residues, which it is thought may provide an interaction site for other proteins in vivo (Stewart et al. 2005). To identify potential binding partners for PLCη2, a bacterial 2-hybrid screen was performed. The coding sequence corresponding to the C-terminal region of the 21a/22/23 splice variant was cloned into the pBT ‘bait’ plasmid. This particular variant was chosen as it is the shortest and with the exception of the three C-terminal amino acids derived from exon 22 (Val-Arg-Asp), the sequence is common to all variants. A pancreas cDNA library was cloned into the pTRG ‘target’ plasmid for screening. Competent E. coli reporter cells were co-transfected with ‘bait’ and ‘target’ constructs and following dual-selection (with 3-AT and streptomycin), 18 colonies were isolated and pTRG plasmids were sequenced (Figure S2). In one colony, the pTRG plasmid was found to encode the C-terminal region (residues 519-647) of LIMK1. This region incorporates part of the protein's catalytic domain but does not contain the residues implicated in substrate binding or phosphorylation activity (residues 345-514) (Manetti 2012). To establish whether PLCη2 and LIMK1 may interact within Neuro2A cells, we examined the cellular localization of both proteins. Both undifferentiated and RA-stimulated cells were stained for endogenous PLCη2 and LIMK1 by co-immunostaining (Fig. 4a and b). LIMK1 and PLCη2 were both present in neurites and were abundant in the nucleus. This is consistent with the previous observation that LIMK1 is located within nucleus in neurons (Bernard et al. 1994). The localization of LIMK1 in comparison with transiently expressed GFP-PLCη2 was also assessed in transiently transfected undifferentiated and RA-treated cells (Fig. 4c and d). Both LIMK1 and GFP-PLCη2 possessed a high degree of co-localization in the nuclei of both untreated and RA-treated cells and in the neurites of the RA-treated cells.
PLCη2 is important for LIMK1/CREB signaling
To assess whether LIMK1 activity is affected by PLCη2 expression, the phosphorylation of LIMK1 was examined in shRNAPLCη2-1 and control cells at different stages of the differentiation process (Fig. 5a and b). Phosphorylated LIMK1 (pLIMK1; Thr508) was undetectable in untreated cells. After 4 h of RA-treatment, the level of pLIMK1 increased to detectable levels in both shRNAcontrol and shRNAPLCη2-1 cells. However, pLIMK1 levels were 75.3% lower in the shRNAPLCη2-1 cells and remained lower than the shRNAcontrol for the duration of the experiment (89% and 46.7% reduction at day 2 and 4, respectively). The phosphorylation of cAMP-responsive element-binding protein (CREB) was also examined. CREB is mainly phosphorylated (and consequently activated) by the cAMP pathway (Montminy and Bilezikjian 1987). However, in differentiating neurons, CREB is also directly phosphorylated by pLIMK1 to direct transcription of genes involved in neuronal outgrowth (Yang et al. 1998). A basal level of phosphorylated CREB (pCREB; Ser133) was detectable and found to be similar in the shRNAPLCη2-1 and shRNAcontrol cells. After 4 h, CREB phosphorylation in PLCη2 knock-down cells was reduced by 30% relative to the control cells, although this was not deemed a significant effect. However, at day 2 and day 4, the pCREB levels were significantly lower than the shRNAcontrol cells (reduced 69.4% and 41.7%, respectively). These results indicate that PLCη2 is involved in LIMK1 activation and the regulation of CREB-mediated gene transcription during Neuro2A cell differentiation. LIMK1 is implicated in the regulation of actin dynamics during neurite growth (Yang et al. 1998). The distribution of F-actin was therefore examined in RA-treated control and PLCη2 knockdown Neuro2A cells by phalloidin-tetramethylrhodamine B isothiocyanate staining (Fig. 5c). The shRNA control cells contained elongated actin filaments along the neurites. The shRNAPLCη2-1 and shRNAPLCη2-2 cells exhibit a punctate distribution of F-actin and possess multiple projections forming short neurites.
PLCη2 PH domain binds preferentially to PtdIns(3,4,5)P3
The PH domain of PLCη2 is responsible for membrane association and is essential for activity (Nakahara et al. 2005; Popovics et al. 2011). The lipid-binding specificity of the PH domain is highly likely to dictate the localization of the enzyme within the cell. To determine the lipid-binding properties of the PH domain, we employed a FRET-based assay which utilized a sensor complex, composed of recombinant GST-linked PLCη2 PH domain protein (GST-PLCη2PH), an Eu-labeled anti-GST antibody, streptavidin-coupled allophycocyanin (APC), and biotinylated lipid. Using this approach, biotinylated-PtdIns(3,4,5)P3 and biotinylated-PtdIns(4,5)P2 were able to form sensor complexes in the presence of GST-PLCη2PH, each yielding a strong concentration-dependent FRET signal (Fig. 6a). Biotinylated forms of the lipids, PtdIns(3,4)P2, PtdIns(3)P, and PtdIns(4)P were unable to form FRET-active complexes. The binding characteristics of a range of non-biotinylated-phosphoinositides to GST-PLCη2PH were assessed via their ability to dissociate the biotinylated-PtdIns(4,5)P2-containing sensor complex at varying concentrations. The majority of phosphoinositides examined were able to effectively and dose-dependently compete with biotinylated-PtdIns(4,5)P2 for binding to GST–PLCη2PH (Fig. 6b). The relative IC50 value of each phosphoinositide is shown in Table 2. PtdIns(3,4,5)P3 was found to be the most effective competitor (IC50=0.83). The phosphoinositides in order of decreasing affinity are PtdIns(3,4,5)P3 > PtdIns(4,5)P2 > Ins(1,3,4,5)P4 > Ins(1,4,5)P3 > PtdIns(3,4)P3 > PtdIns(3)P. PtdIns(1,3,4)P3 was unable to compete effectively at the concentrations examined.
Table 2. Phosphoinositide binding of GST-PLCη2PH. IC50 values for various non-biotinylated-phosphoinositides based on their abilities to dissociate the biotinylated-PtdIns(4,5)P2-containing FRET sensor complex
PLCη2 co-localizes with PtdIns(3,4,5)P3 in Neuro2A cells
The phosphoinositide specificity of the PLCη2 PH domain suggests that PtdIns(3,4,5)P3 may be a critical factor in defining the cellular distribution of PLCη2. With this in mind, we examined the localization of PLCη2 and PtdIns(3,4,5)P3 in differentiated Neuro2A cells by immunostaining (Fig. 6c). Wortmannin treatment dramatically reduced PtdIns(3,4,5)P3 immunostaining in Neuro2A cells providing evidence that the anti-PtdIns(3,4,5)P3 antibody used acts specifically under the conditions employed (Figure S3). Interestingly, the highest level of PtdIns(3,4,5)P3 staining was found in the nucleus where it exhibited a high degree of co-localization with endogenous PLCη2 (and with GFP-PLCη2 in transfected cells, as shown in Figure S4). PtdIns(3,4,5)P3 was also present on the cell membrane and local elevations of both PtdIns(3,4,5)P3 and PLCη2 were detected in growing neurites. In addition, the localization of an N-terminally GFP-tagged PH domain of PLCη2 was also assessed in undifferentiated Neuro2A cells (Figure S5). PI(4,5)P2 content was also examined by the co-transfection of mCherry-tagged PLCδ1-PH domain. As expected, the GFP-PLCη2-PH exhibited a high co-localization rate with PI(3,4,5)P3 in the nucleus. In contrast, the mCherry-tagged PLCδ1-PH domain (indicative of PI(4,5)P2) mainly appeared on the plasma membrane and had a very limited co-appearance with the PH domain of PLCη2.
Neuro2A cells represent an excellent model for the study of neuronal differentiation and were used to assess the role of PLCη2 in this process. PLCη2 protein expression was found to increase in Neuro2A cells during retinoic acid induced differentiation. The involvement of PLCη2 in this process was further indicated by an observed reduction in neurite outgrowth in shRNA-mediated PLCη2 knock-down cells (shRNAPLCη2-1 and shRNAPLCη2-2 lines). Both the percentage of differentiated cells and the total neurite growth were found to be reduced. The extent of this phenotypic effect largely mirrored the degree by which PLCη2 expression was decreased in the PLCη2 knock-down cells, highly indicative of a specific role for PLCη2 in this process. Differentiating neurons are known to exhibit spontaneous Ca2+ spikes and waves (Gu and Spitzer 1995) and, because of the high Ca2+ sensitivity of PLCη2, it is possible that the enzyme plays an important role in the generation of these signals and/or their translation into downstream effects. This is supported by the observation that inositol phosphate release is reduced in PLCη2 knock-down cells, relative to control cells, following treatment with the Ca2+ ionophore, A23187. It should be acknowledged that most of the inositol phosphate release triggered by this agent seems to be directed by PLCη2 with other PLCs having little residual effect. This is likely down to that fact that PLCη enzymes are more sensitive to Ca2+ in comparison to other isotypes likely to be present (such as PLCδ enzymes). PLCη1 has been shown to amplify PLCβ signals in Neuro2A cells following Ca2+ release (Kim et al. 2011). Whether PLCη1 and PLCη2 signals are interdependent or not is still to be established but it is possible that lowering PLCη2 protein level would greatly impact upon PLCη signaling as a whole.
To determine whether PLCη2 regulates transcriptional activity we examined its influence in the control of RAR-directed gene expression. Of the three known RAR isoforms, only RARα was found to be present in Neuro2A cells. The mRNA level corresponding to expression of this isoform was found to be similar in control and PLCη2 knock-down cells. However, PLCη2 knock-down had a substantial effect on basal and RA-stimulated retinoid signaling. This likely contributes to the phenotype associated with the PLCη2 knock-down cells as neuronal differentiation is, as one would expect, associated with a substantial change in gene expression (Gurok et al. 2004). The precise mechanism(s) by which PLCη2 exerts this effect is not clear. Promoters containing RAREs have been shown to be activated by downstream regulatory element antagonist modulator, a Ca2+-effector protein (Scsucova et al. 2005). It is possible that the absence of sufficient levels of PLCη2 in the knock-down cells negatively influences cytosolic Ca2+-dynamics such that activation of downstream regulatory element antagonist modulator is compromised. PLCη2 was found to be present in the nuclei of both undifferentiated and differentiated Neuro2A cells. Other nuclear PLC enzymes are known to influence gene expression. A prime example is PLCβ1, which is present in nuclear speckles; the location of several transcription-regulating molecules (Martelli et al. 1992). PLCβ1 is involved in the regulation of c-Jun and AP1 promoter-binding in differentiating myogenic cells (Ramazzotti et al. 2008). As with cytosolic PLC enzymes, the nuclear enzymes can activate protein kinase C (PKC) (α-isoform) via the production of DAG. PKCα has been shown to phosphorylate lamin B1 to trigger the breakdown of the nuclear lamina for mitotic division in murine erythroleukemia cells (Fiume et al. 2009). The consequent action of PLC enzymes in the nucleus decrease PtdIns(4,5)P2 levels, which are important for chromatin remodeling (Zhao et al. 1998). It has also been proposed that cells may possess Ins(1,4,5)P3 receptors on the inner nuclear membrane that allow movement of Ca2+ to the nucleoplasm (Klein and Malviya 2008). Despite this, the precise roles PLC enzymes play in regulating gene expression and nuclear Ca2+ dynamics are still far from understood.
A putative interaction between PLCη2 and LIMK1 was identified using a bacterial two-hybrid screen. We attempted to confirm this interaction in Neuro2A cells using a co-immunoprecipitation approach but were unsuccessful (data not shown). Nevertheless, LIMK1 and PLCη2 do co-localize in both undifferentiated and differentiated Neuro2A cells. This implies that their distribution allows the interaction of these proteins. LIMK1 and CREB phosphorylation was significantly reduced in the PLCη2 knock-down cells, suggesting that PLCη2 regulates their activity. PLCη2 may potentially influence LIMK1 (and CREB) phosphorylation in two ways. First, it remains known that the phosphorylation of LIMK1 and CREB is regulated by Ca2+ via the Ca2+/calmodulin kinase IV pathway (Matthews et al. 1994; Takemura et al. 2009), thus phosphorylation of both proteins may be altered because of aberrations in Ca2+-signaling in the PLCη2 knock-down cells. Second, it is also possible that PLCη2 and LIMK1 interact directly in such a manner as to regulate activation of LIMK1. It is also possible that a PLCη2-LIMK1 interaction could simply serve such that PLCη2 is “on hand” to modulate LIMK1 activation by Ca2+/calmodulin kinase IV. By whichever mechanism, both PLCη2 and LIMK1 are present in growing neurites. Reduced phosphorylation of LIMK1 (as observed in the PLCη2 knock-down cells) is likely to influence phosphorylation of cofilin. Once phosphorylated, cofilin acts to alter the F-actin/G-actin ratio, a process which is important for neurite outgrowth (Yang et al. 1998, 2004). Aberrations in actin dynamics were observed by phalloidin-tetramethylrhodamine B isothiocyanate staining of shRNAPLCη2-1 and shRNAPLCη2-2 cells. In these cells, a punctate distribution of F-actin was observed with multiple projections and reduced outgrowth compared to control cells. Interestingly, Endo et al. (2007) previously found that down-regulation of LIM kinase activity in PC12 cells significantly reduces neuronal outgrowth. Although the authors did not mention it, the number of projections also appeared to be increased in these cells. Potentially, the observed phenotype of PLCη2 knock-down cells could be caused, at least in part, by an increase in cofilin activity and a consequent reduction in the polymerization rate of actin. Such effects are likely to be further compounded in PLCη2 knock-down cells by reduced expression of RARα-regulated genes involved in the differentiation process (as summarized in Fig. 7). In addition, significant staining of LIMK1 was found in the nucleus of the Neuro2A cells. Nuclear translocation LIMK1 has previously been suggested to be a process important for the control of actin dynamics (Yokoo et al. 2003). It is also possible that nuclear LIMK1 activity contributes to the regulation of genes involved in neuritogenesis.
The PH domain of PLCη2 has previously been shown to be responsible for association of the enzyme to intracellular membranes (Nakahara et al. 2005; Popovics et al. 2011). A GST-tagged PLCη2 PH domain was synthesized and its ability to bind a range of lipids and phosphoinositides was examined. The PLCη2 PH domain bound preferentially to PtdIns(3,4,5)P3. Furthermore, PLCη2 was found to co-localize with PtdIns(3,4,5)P3 in the nucleus and to some extent at the cell membrane in Neuro2A cells, suggesting that it predominantly interacts with this lipid in vivo. Neuronal cells have previously been reported to possess high levels of PtdIns(3,4,5)P3 levels in the nucleus (Neri et al. 1999; Kwon et al. 2010). This likely explains why PLCη2 is present in the nucleus of Neuro2A cells and why its cellular distribution differs in other non-neuronal cell types (Popovics et al. 2011). It is important to point out that the PLCη2 PH domain was also able to bind PtdIns(4,5)P2, although with a somewhat lower affinity. However, co-expression of GFP-PLCη2-PH with mCherry-tagged PLCδ1-PH domain in Neuro2A cells revealed the cellular locations of these proteins to be distinct. The PH domain of PLCδ1 is known to associate with cellular PtdIns(4,5)P2 very specifically (Watt et al. 2002), it is therefore unlikely that PLCη2 interacts to any great extent with this phospholipid in these cells. An interesting feature of the PtdIns(3,4,5)P3 headgroup resides in its larger size and additional negative charge compared with other phosphoinositide headgroups. This triggers its projection from the lipid bilayer which might facilitate the interaction between PtdIns(3,4,5)P3 and PH domains though little penetration of the lipid bilayer is required for docking (Chen et al. 2012). Consequently, PtdIns(3,4,5)P3-binding PLCs such as PLCη2 might represent a subgroup, which respond quicker to extracellular stimuli. Interestingly, the PLCη2 PH domain exhibited a relatively low affinity toward Ins(1,4,5)P3 as demonstrated by the high IC50 value (10.86 μM). Ins(1,4,5)P3-binding is known to regulate PLCδ1 activity, such that it competes with PtdIns(4,5)P2 to displace the PH domain (and thus the enzyme) from the plasma membrane upon its production (Hirose et al. 1999). Given the low affinity of the PLCη2 PH domain toward Ins(1,4,5)P3, it is unlikely that the hydrolyzed headgroup will be able to effectively compete with PtdIns(3,4,5)P3. Consequently, the phospholipid/phosphoinositide-binding specificity of PLCη2 likely make this enzyme particularly well-suited for the modulation of transient Ca2+-signals through sustained generation of Ins(1,4,5)P3 (and DAG) in the nucleus and in growing neurites during the differentiation process.
We gratefully acknowledge the University of St Andrews in providing a PhD studentship for Petra Popovics. We thank Dr Kevin Morgan (MRC Human Reproductive Sciences Unit, Edinburgh, UK) for cloning the pancreas cDNA library. The authors declare no conflicts of interest.