Overexpression of the Protein Phosphatase 2A Regulatory Subunit Bγ Promotes Neuronal Differentiation by Activating the MAP Kinase (MAPK) Cascade*

Protein serine/threonine phosphatase 2A (PP2A) is a multifunctional regulator of cellular signaling. Variable regulatory subunits associate with a core dimer of scaffolding and catalytic subunits and are postulated to dictate substrate specificity and subcellular location of the heterotrimeric PP2A holoenzyme. The role of brain-specific regulatory subunits in neuronal differentiation and signaling was investigated in the PC6-3 subline of PC12 cells. Endogenous Bβ, Bγ, and B′β protein expression was induced during nerve growth factor (NGF)-mediated neuronal differentiation. Transient expression of Bγ, but not other PP2A regulatory subunits, facilitated neurite outgrowth in the absence and presence of NGF. Tetracycline-inducible expression of Bγ caused growth arrest and neurofilament expression, further evidence that PP2A/Bγ can promote differentiation. In PC6-3 cells, but not non-neuronal cell lines, Bγ specifically promoted long lasting activation of the mitogen-activated protein (MAP) kinase cascade, a key mediator of neuronal differentiation. Pharmacological and dominant-negative inhibition and kinase assays indicate that Bγ promotes neuritogenesis by stimulating the MAP kinase cascade downstream of the TrkA NGF receptor but upstream or at the level of the B-Raf kinase. Mutational analyses demonstrate that the divergent N terminus is critical for Bγ activity. These studies implicate PP2A/Bγ as a positive regulator of MAP kinase signaling in neurons.

The phosphorylation state of key proteins is crucial in most cellular processes and depends on the precisely orchestrated balance of protein kinases and phosphatase activities. Compared with kinases, very little is known about the regulation of protein phosphatases. PP2A, 1 one of the four major classes of protein serine/threonine phosphatases in cells, is a family of abundant and ubiquitous enzymes with pleiotropic functions ranging from cell cycle regulation to synaptic plasticity (1). The predominant form of PP2A in cells has a heterotrimeric subunit structure, consisting of a core dimer of ϳ36 kDa catalytic and ϳ65 kDa scaffold subunits (subunits C and A, respectively) complexed to a third variable subunit. Variable subunits are encoded by three multigene families (B, BЈ, BЉ) and are believed to dictate substrate specificity, subcellular localization, and regulation of PP2A by phosphorylation.
There is growing evidence that incorporation of different variable subunits imparts specific cellular functions to PP2A. PP2A, containing BЈ (B56, PR61) family subunits, participates in Wnt/␤-catenin signaling, a signal transduction pathway necessary for vertebrate axis formation in early embryogenesis (2,3). BЈ subunits bind to cyclin G 1 and G 2 , suggesting that PP2A holoenzymes containing these subunits are involved in cell cycle regulation (4 -6). The BЉ subunit PR48 was first identified in a screen for proteins that interact with cdc6, a component of DNA prereplication complexes (7), and may mediate the obligatory role of PP2A in the initiation of chromosomal DNA replication (8). Another BЉ subunit, PR59, recently identified as an interaction partner for the retinoblastoma-related p107 protein (9), could be important in the rapid dephosphorylation of p107 following DNA damage (10). PP2A holoenzymes containing BЉ family subunits may thus be specialized regulators of the G 1 /S cell cycle transition.
Despite being the first PP2A regulatory subunits to be identified, roles of members of the B-type subunit family (PR55) are still largely enigmatic. In mammals, four genes (B␣-␦) code for 54 -57-kDa proteins, which are additionally diversified by alternative splicing or promoter use. B␣-␦ are more than 80% identical at the amino acid level, with the greatest clustering of divergent residues at the N terminus. B-family PP2A subunits are predicted to adopt a seven-blade ␤-propeller fold similar to the ␤-subunits of heterotrimeric G proteins (11). B␣ mediates dephosphorylation of vimentin intermediate filaments in fibroblasts (12) and has been shown to interact preferentially with microtubules and tau protein, possibly contributing to the regulation of microtubule stability by PP2A (13,14). Whereas B␣ and B␦ are widely expressed in different tissues, B␤ and B␥ proteins are detectable only in brain (15)(16)(17)(18), which suggests that B␤ and B␥ mediate specifically neuronal functions of PP2A. B␤ and B␥ expression is differentially regulated during development and maturation of the rat brain; B␤ protein and mRNA levels are high in late embryonic brain and decrease modestly after birth. In contrast, B␥ expression increases sharply postnatally to plateau at 2 weeks of age (17).
This report begins to analyze the functions of neuronal PP2A regulatory subunits in the pheochromocytoma PC12 cell line, an experimentally tractable model system of neuronal differ-entiation and neurite outgrowth (19 -22). Upon NGF treatment, PC12 cells develop into sympathetic neuron-like cells with elaborate neurites capable of generating action potentials (23). Although NGF binding to the TrkA receptor tyrosine kinase activates several signaling pathways, sustained activation of the MAP kinase cascade is both obligatory and sufficient for neurite outgrowth and differentiation (24 -26). Evidence from several laboratories supports a model in which NGF promotes persistent activation of the small GTPase Rap1. Rap1 then recruits the serine/threonine kinase B-Raf to the membrane and maintains its activity by poorly understood mechanisms ((Refs. 27-29, but also see Ref. 30). Raf family kinases phosphorylate the dual-specificity kinase MEK1, which in turn activates MAP kinases of the ERK (extracellular signal-regulated kinase) family. ERKs are broad specificity serine/threonine kinases that target cytosolic as well as nuclear substrates, including the transcription factors Myc and Elk1 (31).
Here, I show that NGF-mediated differentiation of a PC12 subline leads to expression of neuronal PP2A regulatory subunits. One of these subunits, B␥, can promote neuronal differentiation through the MAP kinase pathway by activating B-Raf.

EXPERIMENTAL PROCEDURES
Reagents-The PC6-3 cell line (32) was obtained from Henry Paulson (University of Iowa). This PC12 subline was found to display less propensity to form cell aggregates and could be transfected with higher efficiency than its parental cell line. PC6-3 cells were grown on uncoated plastic in RPMI 1640 medium containing 10% horse serum and 5% fetal bovine serum in a 5% CO 2 incubator. HEK293 and COS-M6 cells were cultured in Dulbecco's modified Eagle's medium/high glucose containing 10% fetal bovine serum. cDNAs of B subunits were isolated by reverse transcriptase-polymerase chain reaction from rat brain total RNA (Access RT-PCR kit, Promega, Madison, WI), subcloned into a pcDNA3.1 mammalian expression vector under control of the cytomegalovirus (CMV) promoter and FLAG epitope-tagged at the N terminus by PCR. The addition of the FLAG epitope did not affect the activity of B␥ in neurite outgrowth and MAP kinase assays (not shown). The B␥ 1-35␣ chimera was constructed by PCR-amplifying B␣ 5Ј and B␥ 3Ј sequences and ligating the fragments utilizing an introduced silent NheI site. Plasmids encoding FLAG-ERK2 and Myc-B-Raf were obtained from Philip Stork (Vollum Institute) and Richard Marais (Royal Cancer Hospital, London), respectively. Hemagglutinin (HA)-MEK1 wild-type and dominant-negative MEK1 K97R plasmids were provided by Jeffrey Pessin (University of Iowa). HA-tagged BЈ (B56) ␣-subunit plasmid and BЈ antibodies were obtained from David Virshup (University of Utah). PP2A/A, B␣/␦, B␤, and B␥ antibodies were from Brian Wadzinski (Vanderbilt University) (17). The following reagents were obtained commercially: antibodies to the PP2A C subunit, ERK1/2, B-Raf, protein G-agarose (Santa Cruz Biotechnology, Santa Cruz, CA); FLAG-tag antibody (M2) and its agarose conjugate (Sigma); 2.5S NGF, glutathione S-transferase (GST)-ERK2, GST-MEK1 (Upstate Biotechnology Inc., Lake Placid, NY); GST-Elk1, phospho-Ser-383 Elk1 antibody, phospho-ERK1/2 antibody (Cell Signaling Technologies, Beverly, MA); Myc-and HA-tag-directed antibodies (Covance, Princeton, NJ); U0126, K252A (Calbiochem).
Neurite Outgrowth Assays-PC6-3 cells were transfected at 20 -30% confluency with 1 g of DNA (0.8 g of PP2A subunit plasmid, 0.2 g of pEGFP-C1, a green fluorescent protein (GFP) expression vector), and 2 l of LipofectAMINE 2000 (Invitrogen) per well of a 12-well plate according to the manufacturer's instructions. After 36 -48 h, neurite outgrowth of living cells was analyzed by capturing digital images from 4 -5 randomly selected fields on an inverted epifluorescence microscope. Transmitted and mercury light sources were adjusted to superimpose phase contrast and fluorescence signals on the same image (Kodak MDS290 documentation system). Images were analyzed using NIH Image software by a second, naive experimenter, who either counted cells with neurites longer than 2-cell body diameters or measured the area of a rectangle bounding the cell body and neurites.
Generation of Tetracycline-inducible PC6-3 Cell Lines-Tetracyclineinducible, stably B␥-expressing PC6-3 cell lines were generated by two rounds of antibiotic selection essentially as described by the vendor of the T-Rex system (Invitrogen). Briefly, PC6-3 cells were transfected with the linearized vector pcDNA6/TR encoding the tetracycline repres-sor (TR) protein. After selection in blasticidine (5 g/ml), 48 clones were expanded and tested for TR function by transfection with a pcDNA5/TO/LacZ reporter plasmid and chemiluminescent ␤-galactosidase assay (Galacto-Star, Tropix, Bedford, MA). One clone (PC6-3/ TR156) displayed 25-fold induction of ␤-galactosidase activity upon treatment with doxycycline and was transfected with linearized pcDNA5/TO/FLAG-B␥, a plasmid encoding FLAG-tagged B␥ under control of a chimeric CMV promoter/tetracycline operator. Cells were selected for plasmid integration in 500 g/ml hygromycin, 2 g/ml blasticidine. Eighty-four clones were expanded and tested for inducible B␥ expression by MAP kinase reporter assays (below) and immunoblotting. Eight clones showed more than 3-fold doxycycline-inducible MAP kinase activity and a high level of B␥ expression.
[ 3 H]Thymidine Incorporation-Cells were cultured in 24-well plates for up to 4 days in the presence of 1 g/ml doxycycline, 20 ng/ml NGF, or vehicle. [ 3 H]Thymidine (2 Ci/ml) was added to the medium for 6 h followed by a wash with phosphate buffered saline and incubation with 0.5 ml/well 5% (w/v) trichloroacetic acid for 20 min at 4°C to remove unincorporated label. DNA was solubilized with 0.5 ml/well 0.1 N NaOH and scintillation-counted.
Immunoprecipitation/Kinase Assays-PC6-3 cells were cultured in 12-well plates to 40 -70% confluency and transfected with 2 l/well LipofectAMINE 2000 and 1 g/well DNA (750 ng of B␥ plasmid or empty vector plus 250 ng of epitope-tagged kinase (B-Raf, MEK1, or ERK2) plasmid). 36 -48 h after transfection and following an overnight incubation with low serum medium (1% horse serum, 0.5% fetal calf serum), cells were lysed in 150 l/well immunoprecipitation/kinase lysis buffer (1% Triton X-100, 150 mM NaCl, 20 mM Tris pH 7.5, 1 mM EDTA, 1 mM EGTA, 1 mM ␤-glycerolphosphate, 1 mM Na 3 V0 4 , 1 mM Na 4 P 2 O 7 , 1 M microcystin-LR, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 mM benzamidine) and sonicated for 2 s at low intensity with a probe tip sonicator. Debris was pelleted (20,000 ϫ g, 15 min), and kinases were immunoprecipitated from the cleared lysate with 0.5-2 g of epitope-tagged antibodies and 6 l of protein G-agarose for 4 h at 4°C. Immunoprecipitates were washed with 6 ml of immunoprecipitation/kinase lysis buffer and 1.5 ml of kinase assay buffer (25 mM Tris pH 7.5, 10 mM MgCl 2 , 2 mM dithiothreitol, 5 mM ␤-glycerolphosphate, 0.1 mM Na 3 V0 4 , 1 M microcystin-LR). Kinase assays were started by adding 25 l of assay mix to the immunoprecipitates; the mixture was incubated for 30 min at 30°C with intermittent agitation, and assays were stopped by the addition of 10 l of 4ϫ SDS-sample buffer containing 100 mM EDTA. The assay mixes contained 40 g/ml GST-Elk1 substrate and 200 M ATP in kinase assay buffer. MEK1 assays additionally included 20 g/ml GST-ERK2, and B-Raf assays included GST-ERK2 in addition to 2 g/ml GST-MEK1. Kinase assays were analyzed by SDS-PAGE and immunoblotting with phospho-Ser-383 Elk1 antibodies and chemiluminescent detection (SuperSignal, Pierce) on a Kodak Imager 440. Serially diluted samples were analyzed to ensure quantification in the linear range of detection. Phospho-Ser-383 Elk1 immunoreactivity was quantified from digital images using NIH Image software, subtracting negligible background phosphorylation from immunoprecipitates of mock-transfected cells.

Differentiation Increases Expression of Neuronal PP2A
Regulatory Subunits-PC6-3 cells are a subline of PC12 cells. Upon withdrawal of NGF, differentiated PC6-3 cells undergo rapid apoptosis even in the presence of serum, which may make them a better model of sympathetic neurons than PC12 cells (32). To investigate the expression of endogenous PP2A subunits, PC6-3 cells were treated for several days in the presence or absence of NGF and analyzed by immunoblotting with specific antibodies (17). NGF treatment resulted in robust neurite outgrowth and expression of differentiation markers such as VGF and neurofilament heavy chain ( Fig. 1 and not shown). Concomitant with differentiation, NGF induced expression of the brain-specific B␤ and B␥, as well as the brain-enriched BЈ␤ isoform of PP2A regulatory subunits. Levels of the catalytic C, the scaffolding A, the B␣ and B␦ subunit (detected by an antibody that recognizes both B subunits), and BЈ␣ and BЈ␦ remained unchanged. Thus, differentiation of PC6-3 cells is accompanied by the induction of neuronal PP2A holoenzymes.
B␥ Induces Neurite Outgrowth-To test whether neuronal PP2A isoforms are involved in the establishment of the neuro-nal phenotype, PC6-3 cells were transiently transfected with a panel of regulatory subunit expression plasmids or vector alone together with a plasmid encoding GFP to mark transfected cells. Expression was verified by immunoblotting; however, protein levels could not be compared because of a lack of antibodies that recognize all regulatory subunits (not shown). 48 h after transfection, living cells were scored for neurite outgrowth by measuring the area of a rectangle that bounded the cell body and neurites. Transfection with B␥, but not B␣ or B␤, resulted in the extension of numerous short neurites quantified as an increase of the bounding rectangle area (51 Ϯ 9%, n ϭ 11 experiments, Fig. 2 and data not shown). B␥ also enhanced neurite outgrowth in cells treated 24 h after transfection with NGF (46 Ϯ 16%, n ϭ 9 experiments), indicating that NGF and this PP2A subunit act synergistically. NGF-treated cells were also scored for neurite outgrowth by counting cells with neurites longer than twice the diameter of the soma (Fig. 2C). This method gave similar results as the bounding rectangle method. The bounding area method was used in all subsequent experiments, as it allowed quantification of early morphological changes.
B␥ Promotes Neuronal Differentiation-Neuronal differentiation of PC12 cells involves a "cell fate" choice characterized by morphological changes (i.e. neuritogenesis) but also cessation of growth, loss of chromaffin cell markers, and synthesis of neuronal proteins. Previous studies with the protein phosphatase inhibitor okadaic acid have shown that PP2A activity is required for neurite maintenance of differentiated neurons (33)(34)(35). It therefore became important to investigate whether B␥ expression causes neuronal differentiation as opposed to neurite outgrowth per se. To this end, PC6-3 cell lines were created that express FLAG epitope-tagged B␥ under control of a tetracycline-inducible CMV promoter (see "Experimental Procedures"). Several clonal isolates exhibited no detectable leak expression and responded to the addition of the tetracycline analog doxycycline to the medium with B␥ protein levels that were 5-10-fold higher than could be achieved by transient transfection of CMV-promoter driven cDNAs. In agreement with transient transfection studies, doxycycline treatment for 36 -48 h resulted in robust neurite outgrowth; however, neurites appeared somewhat shorter than in NGF-treated sister cultures (Fig. 3A). To test whether B␥ causes the growth arrest characteristic of neuronal differentiation, cells lines were incubated for 4 days with doxycycline, NGF, or vehicle alone and assayed for DNA synthesis by pulse-labeling with [ 3 H]thymidine. Cell line 275, which expresses high inducible levels of B␥, responded to doxycycline or NGF treatment with a ϳ50% decrease in [ 3 H]thymidine incorporation (Fig. 3B). In contrast, cell line 280, which was isolated in parallel but failed to induce B␥, showed decreased DNA synthesis only when cultured in NGF, demonstrating that doxycycline itself does not slow down growth. Cell line 275 was next analyzed for neuronal protein expression. Doxycycline or NGF but not vehicle treatment for 4 days induced high levels of neurofilament heavy chain protein (Fig. 3C).
B␥ must incorporate into the PP2A heterotrimer to escape degradation by the ubiquitin-proteasome pathway (11). The percentage of endogenous A and C subunits that associates with inducibly expressed B␥ was determined by quantitative immunoprecipitation. PP2A holoenzymes containing B␥ were estimated to comprise 5% of the total PP2A pool in the inducible PC6-3 line 275 (Fig. 3D). Thus, incorporation of B␥ into a minor pool of PP2A holoenzymes is sufficient for NGF-independent neuronal differentiation of PC6-3 cells.
B␥ Activates the MAP Kinase Cascade-It is well established that activation of the MAP kinase signaling cascade is both necessary and sufficient for neuronal differentiation of PC12 cells (26). This also holds for the PC12-derived PC6-3 cell line, because the MEK inhibitor U0126 blocks NGF-induced neurite outgrowth (Fig. 5B), and transfection of a constitutively active MEK1 mutant promotes neuritogenesis, growth arrest, and neurofilament expression (data not shown). Because of the pivotal role of this signaling cascade in neuronal differentiation, a possible activation of the MAP kinase signaling by B␥ was investigated. PC6-3 cells were transiently transfected with plasmids encoding different PP2A regulatory subunits or empty vector in combination with reporter plasmids that read out ERK-dependent activation of the transcription factor Elk1 (Elk1 PathDetect Trans-reporter assay, Stratagene). Transfection of B␥, but not any other regulatory subunit tested, resulted in robust activation of the MAP kinase reporter (Fig. 4A, 15.1 Ϯ 2.6-fold activation, n ϭ 13 experiments). Activation was not only subunit-but also cell line-specific, because B␥ had no effect on ERK activity in HEK293 and COS-M6 cells even though B␥ could be expressed to significantly higher levels in these non-neuronal cell lines (Fig. 4B). MAP kinase activation by B␥ in PC6-3 cells was comparable to stimulation with low (ϳ 1 ng/ml) NGF concentrations, and a combination of both treatments resulted in more than additive induction of Elk1-mediated luciferase activity (Fig. 4C). Fig. 4D illustrates this effect over a range of NGF concentrations. B␥ transfection caused a leftward shift of the NGF dose-response curve, enhancing MAP kinase signaling from 2-fold at saturating NGF concentrations to 12-fold in the absence of NGF in this experiment.
Tetracycline-inducible PC6-3 cells were used to investigate whether B␥ activates MAP kinase signaling by stimulating ERK phosphorylation of activation loop residues (Thr-202 and Tyr-204 in human ERK1). Cells treated for 24 h in the absence or presence of doxycycline to induce B␥ expression were challenged for 5 min with increasing concentrations of NGF, and ERK1/2 phosphorylation was assayed by immunoblotting lysates with a phosphorylation-state specific antibody (Fig. 4E). Paralleling the MAP kinase reporter assays in Fig. 4D, the data demonstrate that B␥ expression and NGF synergistically induce ERK phosphorylation, suggesting that PP2A isoforms

FIG. 3. Tetracycline-inducible B␥ expression causes differentiation of PC6-3 cells.
A, a polyclonal population of PC6-3 cells stably expressing FLAG-tagged B␥ from a tetracycline-inducible promoter was treated for 36 h with vehicle (control), 1 g/ml doxycycline (Dox), or 20 ng/ml NGF. B, B␥-positive (#275) and -negative (#280) clonal cell lines were treated for 4 days with vehicle, Dox, or NGF and analyzed for DNA synthesis by [ 3 H]thymidine incorporation (top, means Ϯ standard deviation of duplicate wells) and B␥ expression (bottom, FLAG epitope immunoblot). C, inducible, B␥-expressing PC6-3 cells (clone 275) were treated with doxycycline or NGF for 3 days, and extracts were immunoblotted for neurofilament heavy chain (NF-H), FLAG-B␥, and the PP2A catalytic subunit (C). D, PP2A holoenzymes containing B␥ were immunoprecipitated with FLAG-epitope antibodies from PC6-3 line 275 after induction with doxycycline for 3 days. FLAG peptide (100 g/ml) was included as a specificity control as indicated. The lysate (input) and the supernatants after immunoprecipitation (IP-supe.) as well the pellets were immunoblotted for FLAG-B␥ and endogenous A and C subunits. The percentages of total A and C subunits co-immunoprecipitating with B␥ (coIP) were determined by densitometry adjusting for 10-fold concentration in the immunoprecipitation pellet. *, significant increase over control, p Ͻ 0.01.
containing B␥ may sensitize neurons to limiting amounts of NGF.
The time course of ERK phosphorylation following induction of B␥ expression was investigated next. ERK1/2 phosphorylation was increased as early as 9 h after treatment of cells with doxycycline, even before B␥ levels reached a steady state, and remained elevated ϳ2-fold for at least 2 days in doxycyline (Fig. 4F). By 4 days, ERK1/2 phosphorylation returned to base line, despite continued high B␥ expression. In comparison, treatment of the same cell line with a saturating NGF concentration (20 ng/ml, see Fig. 4, D and E) resulted in pronounced ERK1/2 phosphorylation, which peaked at 9 h but was still above base line at 4 days after NGF treatment (Fig. 4F). These data indicate that PP2A/B␥ expression promotes a long-lasting but transient activation of the MAP kinase cascade.
B␥-induced Neurite Outgrowth Requires MAP Kinase Signaling but Not NGF Receptor Activation-The data so far demonstrate that B␥ overexpression stimulates ERK activity and or doxycycline (ϩDox, 1 g/ml) to induce B␥ expression. Cells were then stimulated for 5 min with the indicated concentrations of NGF and analyzed for ERK phosphorylation by immunoblotting total lysates for phosphorylated (pERK1, pERK2) and total ERK1/2. The inset shows a representative blot of cells treated without NGF. ERK phosphorylation was quantified by densitometry as the ratio of phosphoreactivity to total immunoreactivity normalized to control. The means Ϯ standard deviation of duplicate pERK2 determinations from a representative experiment are plotted. ERK1 phosphorylation followed an almost identical dose response. F, PC6-3 line 275 was induced to express B␥ by the addition of doxycycline (1 g/ml) or treated with NGF (20 ng/ml) for 9 -96 h. Total lysates were immunoblotted for the indicated proteins, and ERK1/2 phosphorylation was quantified as described for panel E (average of duplicate determinations from a representative experiment). Significant increases over control: *, p Ͻ 0.05, **, p Ͻ 0.005. neurite outgrowth in PC6-3 cells. Inhibitor studies were performed to address the questions whether these two phenomena are causally linked and where in the MAP kinase cascade PP2A/B␥ acts. In transient transfection assays, B␥-induced ERK activation was fully blocked by inhibiting its upstream kinase MEK1 with U0126 (20 M) (Fig. 5A) or by cotransfection of dominant-negative MEK1 (Fig. 5D). These data suggest that B␥ does not directly activate (or disinhibit) ERK kinases. MEK1 inhibition by U0126 also abrogated neurite outgrowth in the presence of B␥ (Fig. 5, B and C). This result demonstrates that MAP kinase activation is necessary for B␥-induced neurite outgrowth, strongly supporting the hypothesis that overexpression of PP2A/B␥ causes neuronal differentiation by activating this kinase cascade.
A recent report suggests that PC12 cells may sustain differentiation through an autocrine mechanism by secreting NGF (36). To address the possibility that B␥ promotes differentiation by stimulating NGF synthesis or release, PC6-3 cells were treated with the kinase inhibitor K252A at 100 nM, a concentration that is relatively selective for the NGF receptor/TrkA receptor tyrosine kinase (37). K252A preincubation completely inhibited NGF-mediated ERK activation and neurite outgrowth but had little or no effect on the same parameters in B␥-transfected cells (Fig. 5, A-C). Together, these data indicate that PP2A/B␥ acts on a signaling transducer downstream of the NGF receptor and upstream of the ERKs to promote neurite outgrowth.
B␥ Activates B-Raf-Immunoprecipitation/kinase assays were carried out to more precisely identify the site of action of PP2A/B␥ in the MAP kinase cascade. Epitope-tagged Raf, MEK, or ERK kinases were transiently co-expressed with B␥ cDNA or empty vector, immunoprecipitated with epitope-tagdirected antibodies, and assayed for kinase activity. Immunoprecipitated ERK2 was assayed for direct phosphorylation of an Elk1 GST fusion protein, whereas Raf and MEK activities were measured in coupled cascade assays in which the kinases necessary for Elk1 phosphorylation were added as recombinant, nonphosphorylated proteins (38,39). The neuronal B-Raf isoform was analyzed in this assay because it is the most abundant Raf isoform and because it mediates NGF signaling in PC12 cells (27,28). Results from these experiments are shown in Fig. 6. ERK, MEK, and B-Raf activities were elevated by B␥ co-expression. Kinase activation by B␥ was comparable to and synergistic with a 5-min treatment of cells with 1 ng/ml NGF. Thus, PP2A/B␥ activates ERK signaling at the level or upstream of B-Raf.
The Divergent N Terminus of B␥ Is Important for Its Activity-B-family regulatory subunits are more than 80% identical at the amino acid level, and divergent residues are clustered in the 20 -30 N-terminal amino acids. The divergent tail is predicted to protrude from the toroid core structure of B-family regulatory subunits and is dispensable for association of B␥ with PP2A A and C subunits (11). To test whether the B␥  6. B␥ activates B-Raf, MEK1, and ERK2 kinase activities. PC6-3 cells were cotransfected with empty vector or B␥ and one of the indicated epitope-tagged kinase constructs. After 48 h, cells were incubated for 5 min in the presence or absence of 1 ng/ml NGF and assayed for activity of the immunoprecipitated kinases by either direct (ERK2) or coupled (MEK1, B-Raf) phosphorylation of GST-Elk1. A, immunoblots of representative assays showing phosphorylated Elk1 detected by a phospho-specific antibody and immunoprecipitated kinases in the same lanes. B, quantification of kinase activities by digital image analysis (normalized means Ϯ standard deviation from two or three independent experiments as indicated). All increases are significant (p Ͻ 0.05) compared with vector-NGF control.
N terminus determines the B␥ signaling function in PC6-3 cells, the first 35 amino acids of B␥ were replaced with the corresponding residues from B␣ (Fig. 7A), a B-family subunit with wide tissue distribution that does not promote neurite outgrowth or ERK activation (Fig. 2, 4A). This chimeric PP2A subunit, B␥ 1-35␣, could be expressed to wild-type B␥ levels in PC6-3 cells and was incorporated into the PP2A holoenzyme as shown by co-immunoprecipitation with the PP2A catalytic subunit (Fig. 7B). However, when tested in ERK activation and neurite outgrowth assays, B␥ 1-35␣ activity was strongly impaired compared with wild-type (Fig. 7C), demonstrating that N-terminal residues are crucial for B␥ function in neuronal differentiation. DISCUSSION This report shows that transient and inducible expression of the neuronal PP2A regulatory subunit B␥ promotes neuronal differentiation of PC6-3 cells as evidenced by increased neuritogenesis, cessation of cell division, and neuronal protein expression. As NGF induces endogenous B␥ expression, PP2A holoenzymes containing B␥ may function in a positive feedback loop to maintain the differentiated phenotype of PC6-3 cells. Immunoblotting with an activation-state specific ERK antibody as well as luciferase reporter and kinase assays demonstrate that B␥ expression up-regulates MAP kinase activity for several days and that this up-regulation synergizes with NGF treatment. Linking the effects of B␥ on MAP kinase signaling and neuronal differentiation, ERK activation was shown to be necessary for B␥-induced neurite outgrowth. Immunoprecipitation/kinase assays implicate the neuronal B-Raf isoform as direct or indirect target of PP2A/B␥.
Complex Regulation of MAP Kinase Signaling by PP2A-The MAP kinase pathway is a key signaling cascade not only in differentiation but also in the transduction of mitogenic signals. Studies in PC12 cells suggest that the magnitude and temporal dynamics of ERK activation determine whether cells respond by increasing their rate of proliferation or by exiting the cell cycle altogether (19,22). In this context, modulators that shape the ERK activation curve are of critical importance in normal development as well as neoplastic growth.
The role of PP2A as a major negative regulator of MAP kinase signaling is well established (40). DNA tumor virus antigens, SV40 small t and polyomavirus small and middle T antigen complex with the PP2A core dimer to inhibit its activity toward MEK and ERK, resulting in cellular transformation (41)(42)(43). Dephosphorylation of Thr-183 by PP2A is the ratelimiting step in the inactivation of ERK2 after epidermal growth factor stimulation of PC12 cells (44). Ablating the single B-family regulatory subunit in Drosophila Schneider cells by RNA interference activates the MAP kinase pathway (45). PP2A A and C subunits have been shown to associate with the adaptor protein Shc to inhibit its tyrosine phosphorylation and consequent MAP kinase activation in Rat-1 fibroblasts (46).
Concentrations of okadaic acid that selectively inhibit PP2A dramatically increase ERK activity in several cell lines, including PC6-3 cells (47,48). 2 Because okadaic acid inhibits all PP2A holoenzymes, these experiments indicate that the net effect of PP2A on MAP kinase signaling is inhibitory. However, this almost certainly reflects the summation of both negative and positive effects of PP2A on multiple substrates in the MAP kinase pathway, as indicated by genetic studies of Drosophila photoreceptor development (49).
Two recent reports have documented that PP2A can act as a positive regulator of MAP kinase signaling by activating Raf. Sur-6, the single member of the Caenorhabditis elegans PP2A B-type regulatory subunit family, was identified in a screen for mutations that suppress an activated ras mutation, and epistasis experiments indicated that Sur-6 enhances MAP kinase signaling upstream of lin-45 raf in vulva organogenesis (50). A biochemical mechanism for activation of Raf by PP2A was also provided. PP2A A and C subunits were detected in a macromolecular complex with C-Raf in macrophages, and PP2A-mediated dephosphorylation of an inhibitory site, Ser-259, was shown to be necessary for full activation of the kinase (51). The results in the present report provide the third example of PP2A as a positive regulator of Raf activity and ERK signaling, as well as the first evidence for a specific function of a neuronal PP2A holoenzyme.
B␥ in Differentiation-B␥ is one of three neuronal PP2A regulatory subunits of which the expression is increased when PC6-3 cells differentiate into neurons (Fig. 1). Their high expression in adult brain (17,52) suggests that neuronal PP2A holoenzymes contribute to various physiological processes in mature neurons. Even though B␥ expression is sufficient to induce neurite outgrowth, growth arrest, and neuronal marker expression (Figs. 2 and 3), it is currently unclear whether the endogenous PP2A/B␥ enzyme plays a role in the initial differentiation response to NGF or whether B␥ becomes important after PC6-3 cells have adopted a neuronal phenotype. Similarly, because B␥ mRNA and protein is not detectably expressed in rat brain until after birth (17), by which time most cells have already committed to glial or neuronal cell fates, it is questionable whether B␥ functions in cell fate decisions of the developing central nervous system. Instead, PP2A/B␥ regulation of the MAP kinase cascade is likely to be more relevant to other functions of this pleiotropic signaling pathway in the mature brain, in e.g. synaptic plasticity (53).
Is B␥ a MAP Kinase Activator or Disinhibitor?-The data presented here suggest that PP2A/B␥ activates the MAP kinase pathway by dephosphorylating inhibitory sites on B-Raf or its activators. An alternative hypothesis is that forcibly expressed B␥ inhibits an endogenous PP2A holoenzyme that normally gates (inhibits) the MAP kinase cascade. In this scenario, B␥ would act similarly to DNA tumor virus antigens that inhibit PP2A by displacing cellular regulatory subunits. Several lines of evidence, however, argue that B␥ does not simply disinhibit MAP kinase signaling. First, ERK activation is B␥ subunit-specific (Fig. 4A) and requires the divergent N-terminal tail of B␥ (Fig. 7). Second, PP2A/B␥ stimulates MAP kinase signaling only in the neuronal PC6-3 cell line, even though much greater levels of expression (and presumably displacement of endogenous PP2A regulatory subunits) are achieved in two non-neuronal cell lines (Fig. 4B). Third, even in strongly expressing, tetracycline-inducible PC6-3 cells, B␥ is incorporated into only 5% of the cellular PP2A holoenzyme pool (Fig.  3D). These data argue for a specific role of the B␥-containing PP2A holoenzyme rather than a perturbation of other PP2A holoenzymes by B␥ expression. However, a formal proof that endogenous B␥ regulates MAP kinase signaling has to await the development of reagents to interfere with its cellular expression or function.
Mechanism of B␥ Action-Compared with protein kinases, Ser/Thr phosphatases like PP2A are relatively promiscuous enzymes in vitro, dephosphorylating serine/threonine residues in various sequence and structural contexts. This has lead to the proposition that one of the main functions of regulatory subunits is to direct PP2A holoenzymes to different parts of the cell. Differential subcellular localization of PP2A regulatory subunits has indeed been demonstrated for members of the B and BЈ subunit families (17,54,55). By subcellular fractionation of rat brain extracts, B␣, B␤, and B␦ were found to be mostly cytosolic or Triton X-100-soluble. In contrast, B␥ was predominantly found in the detergent-insoluble fraction, suggesting that PP2A/B␥ is a cytoskeletal holoenzyme (17,18). ERK/MAP kinases were first identified as microtubule-associated protein kinases (56), and all kinases of the MAP kinase cascade, including B-Raf, are heavily expressed in neuronal axons and dendrites (57)(58)(59). Consequently, B␥ may target the PP2A holoenzyme to specific cytoskeletal/membrane structures to regulate ERK signaling. This likely involves protein-protein interactions between the critically important N terminus of B␥ (Fig. 7) and yet-to-be identified anchoring proteins.
The lack of an effect of B␥ expression on MAP kinase signaling in non-neuronal cell lines (Fig. 4B) suggests that PP2A/B␥ regulates the activity of a neuronal signaling molecule. Alternatively, PP2A/B␥ may require a neuronal co-factor for regulation of a ubiquitous enzyme. Pharmacological and dominantnegative inhibitor studies restrict possible B␥ targets to molecules between the TrkA NGF receptor and the ERKs (Fig.  5). Because the most proximal kinase shown to be activated by B␥ is B-Raf (Fig. 6), an attractive hypothesis is that B␥ targets the PP2A holoenzyme to dephosphorylate inhibitory sites on this brain isoform of the Raf family of Ser/Thr kinases. Consistent with this hypothesis, B-Raf is more highly expressed in PC6-3 cells than in the non-neuronal HEK293 and COS-M6 cell lines (Fig. 4B). Regulation of Raf by phosphorylation is complex, poorly understood, and likely to be different for each of the three Raf isoforms (A-, B-, and C-Raf (60)). For instance, neuronal B-Raf is inhibited by phosphorylation at three serine/ threonine residues, only one of which is shared by the widely expressed C-Raf isoform (61). Whether PP2A/B␥ dephosphorylates inhibitory sites unique to B-Raf is currently under investigation. It is also possible that PP2A/B␥ activates or promotes the assembly of adaptor/scaffolding molecules and guanylate exchange factors that link neurotrophin receptor engagement to activation of small G-proteins, ultimately stimulating B-Raf activity. The composition of this signaling complex in PC12 cells is an area of active research (28,29), and much remains to be discovered about the role of Ser/Thr phosphorylation and the relevant phosphatases.