Phosphoinositide-specific Phospholipase C β 1b (PI-PLCβ1b) Interactome: Affinity Purification-Mass Spectrometry Analysis of PI-PLCβ1b with Nuclear Protein*

Two isoforms of inositide-dependent phospholipase C β1 (PI-PLCβ1) are generated by alternative splicing (PLCβ1a and PLCβ1b). Both isoforms are present within the nucleus, but in contrast to PLCβ1a, the vast majority of PLCβ1b is nuclear. In mouse erythroid leukemia cells, PI-PLCβ1 is involved in the regulation of cell division and the balance between cell proliferation and differentiation. It has been demonstrated that nuclear localization is crucial for the enzymatic function of PI-PLCβ1, although the mechanism by which this nuclear import occurs has never been fully characterized. The aim of this study was to characterize both the mechanism of nuclear localization and the molecular function of nuclear PI-PLCβ1 by identifying its interactome in Friend's erythroleukemia isolated nuclei, utilizing a procedure that coupled immuno-affinity purification with tandem mass spectrometry analysis. Using this procedure, 160 proteins were demonstrated to be in association with PI-PLCβ1b, some of which have been previously characterized, such as the splicing factor SRp20 (Srsf3) and Lamin B (Lmnb1). Co-immunoprecipitation analysis of selected proteins confirmed the data obtained via mass spectrometry. Of particular interest was the identification of the nuclear import proteins Kpna2, Kpna4, Kpnb1, Ran, and Rangap1, as well as factors involved in hematological malignancies and several anti-apoptotic proteins. These data give new insight into possible mechanisms of nuclear trafficking and functioning of this critical signaling molecule.

Nuclear PI-PLC␤1 is known to be involved in specific signal transduction pathways that differ from those occurring in other cellular compartments. The role of PI-PLC␤1 has been extensively studied in the nucleus, and PI-PLC␤1 is now considered a key co-factor for several nuclear processes, including cell growth, proliferation, and differentiation.
In MEL cells, PI-PLC␤1 is involved in regulating G1/S (3) and G2/M (4) cell cycle progression. During G1/S transition, overexpression of PI-PLC␤1 results in up-regulation of the cyclin D3/cdk4 complex. This complex is responsible for the phosphorylation of retinoblastoma protein and the subsequent activation of the E2F-1 transcription factor, forcing cells out of the G1 phase of the cell cycle. In G2/M, the production of inositol-3-phosphate and diacylglycerol (DAG) from the cleavage of phosphatidylinositol-4,5-bisphosphate results in PKC␣-dependent phosphorylation of lamin B, leading to nuclear envelope disassembly and cell cycle progression. The regulation of these events falls to the activation of JNK, which translocates to the nucleus and mediates PI-PLC␤1 phosphorylation and activation and DAG production.
Indeed, nuclear PI-PLC␤1-induced signaling constitutes an autonomous lipid-dependent signaling system independent from its plasma membrane counterpart. The understanding of the nuclear protein network behind PI-PLC␤1 function might give insight as to the downstream target effectors and further clarify its nuclear signaling cascade.
To date, our group has identified few protein targets that physically associate with PI-PLC␤1 in the nucleus, but a systematic analysis of the PI-PLC␤1 protein interactome has not been undertaken previously. By means of immune-affinity purification-mass spectrometry analysis (AP-MS), we identified 160 proteins in complex with PI-PLC␤1b in the nucleus of MEL cells. Among these, we found two interactors already known to associate with PI-PLC␤1, Srsf3 and Lmnb1.
PLC␤1b precipitation. Nuclear purity was assessed based on the absence of ␤-tubulin.
Mass Spectrometry Analysis-Gel slices were washed in 100 mM ammonium bicarbonate (pH 8) and 50% acetonitrile until complete destaining and then digested with sequencing grade trypsin (Promega, Madison, WI) at 37°C. After overnight incubation, peptides were extracted sequentially three times with 50% acetonitrile and 0.1% formic acid in water. Each extraction involved 5 min of stirring followed by centrifugation and removal of the supernatant. The original supernatant and those obtained from sequential extractions were combined and completely dried down.
Elution from the analytical column was performed using a binary solvent mixture composed of 3% (v/v) acetonitrile plus 0.1% (v/v) formic acid (solvent A) and 97% (v/v) acetonitrile plus 0.1% formic acid (solvent B) at a flow rate of 0.4 l/min with the following gradient: from 4% to 30% B in 17 min, from 30% to 40% B in 3 min, from 40% to 85% B in 3 min, and 85% B for 2 min.
Data were acquired in data-dependent MS/MS mode in which, for each cycle, the three most abundant multiply charged peptides (2 ϩ to 4 ϩ ) above an absorbance threshold of 200 in the MS scan (m/z full scan acquisition range from 250 to 2450) were selected for MS/MS (m/z tandem mass spectrum acquisition range from 50 to 3200). Each peptide was selected twice and then dynamically excluded for 0.1 min.
For peptide sequence searching, the mass spectra were processed and analyzed using either Mascot v.2.2 (Matrix Science, London, UK) or X!Tandem (integrated within the Trans Proteomic Pipeline, v.4.5 Rapture, rev. 2, Institute for Systems Biology, Seattle, WA) against the mouse-derived UniProtKB-SwissProt database (April 23, 2012; total of 535,698 entries) concatenated with a decoy database constructed by randomizing all of the protein sequences present in the mouse reference database. The search parameters were as follows: one missed cleavage allowed, carbamidomethylation of cysteine as a fixed modification, oxidation of methionines as a variable modification, precursor ion mass tolerance of 10 ppm (Mascot) or 20 ppm (X!Tandem), and fragment ion tolerance of 0.12 Da.
Data Analysis-The Trans Proteomic Pipeline was used to validate MS/MS-based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95% probability as specified by the Peptide Prophet algorithm (15). Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony (16). Proteins assigned by one identified peptide were accepted only if the peptide was unique and the probability was greater than 99%. Biological replicates were analyzed within iProphet (17), and the probability scores were adjusted according to the number of replicate spectra, the number of sibling ions, the number of sibling modifications, and the number of sibling experiments. The resulting false discovery rate for these experiments was determined to be less than 1%. Gene Ontology classification was performed using the plug-in BiNGO v2.42 within Cytoscape v.2.8.2, assessing over-represented categories with a hypergeometric statistical test and Benjamini & Hochberg False Discovery Rate correction (p Ͻ 0.05). As a reference set, the whole mouse repository annotation was used.
Western Blot and Immunoprecipitation Analysis-Nuclear protein lysates were prepared as described elsewhere (3). Briefly, cells were lysed in radioimmune precipitation assay buffer containing protease and phosphatase inhibitor cocktails, calpain I and calpain II inhibitors, and benzonase. The protein concentration was determined via Bradford Protein Assay (Bio-Rad), and 60 g of nuclear lysate were separated via SDS-PAGE. For validation analysis, 800 g of nuclear proteins were used, and the immunoprecipitation was carried-out as for mass spectrometry analysis. Gels were transferred onto nitrocellulose membranes and blocked in 5% non-fat dry milk in 1X PBS containing 0.1% Tween-20 (PBST) and then incubated in primary antibody in 1% non-fat dry milk in PBST. Membranes were washed in 1X PBST and incubated for 1 h in the appropriate secondary antibody in 1% non-fat dry milk in PBST. The blots were washed three times with 1X PBST, detected using SuperSignal West Pico Reagent (Pierce, Rockford, IL), and visualized in a ChemiDoc digital image station (Bio-Rad, Hercules, CA). Antibody working dilutions are reported in supplemental Table S1.

AP-MS/MS Identification of Proteins Interacting with
PLC␤1b in the Nucleus-MEL/PI-PLC␤1b cells were obtained by infecting MEL cells with a retroviral vector containing the coding sequence for PI-PLC␤1b and selection in the presence of blasticidin. Obtained clones overexpressed full-length PI-PLC␤1b almost exclusively in the nucleus (Fig. 1A). In order to determine what proteins interact with PI-PLC␤1b in the nucleus, three independent biological experiments were conducted in which nuclei of MEL/PI-PLC␤1b cells were isolated during logarithmic growth. The PI-PLC␤1b complex was affinity purified via immunoprecipitation of MEL/PI-PLC␤1b nuclear lysates. As a control, the normal mouse IgG complex was immunoprecipitated from the same cell line. In order to control for antibody specificity and protein binding, a small fraction of the immunoprecipitates (corresponding to 200 g of nuclear lysate) was analyzed by means of SDS-PAGE and Western blotting with anti-PI-PLC␤1 (Fig. 1B). The remainder of both the immunoprecipitates (PI-PLC␤1b and mouse IgG control) was separated via SDS-PAGE and Coomassie stained (Fig. 1C). Three lanes from both MEL/pBBev and MEL/PI-PLC␤1b were completely sectioned, and each section was subjected to nano-liquid chromatography electrospray ionization MS/MS to identify the associated proteins. The majority of proteins were identified by at least two distinct peptides; however, proteins identified with a single unique peptide with a probability greater than 99% were also in-

FIG. 1. Stable overexpression of PI-PLC␤1b in MEL cells and isolation of PI-PLC␤1b complexes in the nucleus.
A, PI-PLC␤1b was cloned into the retroviral vector pBB-IRES-blast ® and infected into MEL cells. A clone that stably overexpressed PI-PLC␤1b was selected. 1 ϫ 10 7 cells were used to isolate nuclei and cytoplasm. 60 g of cellular, nuclear, and cytoplasmic lysates were separated in 8% SDS-PAGE. Immunoblotted proteins were revealed with anti-PI-PLC␤1 antibody. Lamin B and ␤-tubulin were used to assess nuclear integrity and purity, respectively. B, specific PI-PLC␤1 antibody was used to isolate potential interactors from the nuclei of MEL cells specifically overexpressing PI-PLC␤1b. A small aliquot of the immunoprecipitate was immunoblotted with anti-PI-PLC␤1 antibody to verify the specificity of the immunoprecipitation. For antibody details, refer to supplemental Table S1. mIgG, mouse IgG. C, immunoprecipitated complex was separated via SDS-PAGE and detected with Coomassie Colloidal Blue. Five independent biological replicates were performed to assess the affinity purification reproducibility. Three lanes from both MEL/pBBev and MEL/PI-PLC␤1b were completely fractionated (indicated by arrows), digested, and subjected to nanoliquid chromatography MS/MS for protein identification. cluded and are indicated with an asterisk in the tables and text. In total, 160 proteins were discovered to be in complex with PLC␤1b but not with the IgG control (Table I and supplemental Table S2). Two of the proteins identified, Srsf3 (splicing factor SRp20) and Lmnb1, have already been established as PI-PLC␤1 interactants in the nucleus (4,18).
In Vitro Binding of Identified Proteins with PI-PLC␤1b-Data obtained from mass spectrometry analysis were validated in co-immunoprecipitation experiments performed with the following proteins of interest from isolated nuclei of MEL and Ba/F3 cells: Eef1a2, Kpna2, Kpna4, Kpnb1, Lmnb1 (Ba/F3 only), Phb2, and Srsf3 (Ba/F3 only) (Fig. 2). MEL and Ba/F3 cells demonstrated equal expression of PI-PLC␤1b both when the wild-type cells were compared and when cells stably overexpressing PI-PLC␤1b were compared ( Fig. 2A). Co-immunoprecipitation is the most common method used to test whether two proteins of interest are associated; in this method, target-protein-specific antibodies are used to indirectly capture both proteins considered to be bound together. The crossed pull-down acted as a double-check on the validity of the results obtained in the mass spectrometry identifications, as PI-PLC␤1 immunoprecipitation results in the co-immunoprecipitation of protein B and targeted immunoprecipitation of protein B co-immunoprecipitates PI-PLC␤1. In MEL cells, Eef1a2, which we previously reported to be phosphorylated by PKC␤I (conventional protein kinase C, DAG-dependent) in the nuclear compartment (19), was verified by means of crossed co-immunoprecipitation as associated specifically with PI-PLC␤1b but not the IgG control (Fig. 2B). Similarly, PI-PLC␤1b was shown to interact with Phb2, a protein that shuttles between the nucleus and mitochondria, and with importins Kpna2, Kpna4, and Kpnb1, which are required for the nucleo-cytoplasmic transport of a certain set of proteins. Similar to MEL/PI-PLC␤1b cells, PI-PLC␤1b was also found in complex with Srsf3, Lmnb1, Phb2, Kpnb1, and Kpna2 in nuclei of the murine pro-B lymphoid Ba/F3 cells overexpressing PI-PLC␤1b (Fig. 2C). The interaction of PI-PLC␤1b with Eef1a2 and Kpna4 was not as evident in Ba/F3 as in MEL cells. Unfortunately, because of the antibody background it was not possible to detect bands in Western blots for Eef1a2 following immunoprecipitation with PI-PLC␤1. These data provided further support that proteins identified via AP-MS were specifically complexed with nuclear PI-PLC␤1b.
Biological Classification of Proteins Identified in Complex with Nuclear PI-PLC␤1b-Proteins identified in complex with nuclear PI-PLC␤1b were analyzed in Gene Ontology (GO) and classified according to their biological role(s). The over-representation of each GO term was determined using the plug-in BiNGO and Cytoscape, utilizing the whole mouse repository annotation as a reference dataset. GO categories that were significantly over-represented, with p values Ͻ 0.05, are shown in Fig. 3. PI-PLC␤1b was associated with proteins involved in cellular metabolic processes (31%), gene expression (19%), transport (12%), developmental processes (8%), translation (8%), RNA splicing and processing (7%), response to oxidative stress (5%), and regulation of apoptosis (4%). A minor percentage of proteins were related to the cell cycle (2%), cell proliferation (2%), regulation of (myeloid) differentiation (1%), and cell fate determination (1%) (Fig. 3A). Subdividing the gene expression categories revealed that PI-PLC␤1b specifically associated with chromatin organization proteins, chromatin silencing by methylation, and epigenetic regulation (Fig. 3B). Within the GO term "transport," proteins involved in RNA and protein transport, as well as the establishment of sub-cellular localization, were found in complex with PI-PLC␤1b. Among the proteins involved in cellular metabolic processes were proteins involved in glycolysis and oxidative phosphorylation.
PI-PLC␤1b Interactome Network-Network analysis was conducted with the open-source software String 9.0, which was used to build functional protein association networks based on data present in the database, including physical and functional associations of proteins derived from genomic, high-throughput experiments, co-expression, and literature knowledge (5,214,234 proteins from 1133 organisms; January 2013) (20). The generation of a virtual protein network for PI-PLC␤1b interactants highlighted that PI-PLC␤1b predominantly associated with protein complexes involved in protein/RNA transport, transcriptional and translational regulation (epigenetic control), splicing, response to oxidative stress, and metabolic processes (glycolysis and heme biosynthesis; Fig. 4).
Chromatin Remodeling and Gene Expression-In PI-PLC␤1b clusters, the chromatin remodeling proteins Actl6a, Smarca4 (SWI/SNF chromatin remodeling and histone acetyltransferase complex proteins), Chd4 (a central component of the nucleosome remodeling and histone deacetylase repressor complex, NuRD), and Dnajc2 (recruited at histone H2A sites; promotes the activation of polycomb genes) were identified. Proteins associated with chromatin (Hmgb1), a component of heterochromatin (Hp1bp3) and histones (H2A and H2B     Notes: Proteins identified in the nuclear PI-PLC␤1b complex. Proteins are listed in alphabetical order according to the gene name. Identified proteins are listed according to the search engine used for the analysis (proteins identified in both Mascot and X!Tandem, or proteins provided by either Mascot or X!Tandem). For proteins identified with multiple search engines (Mascot and X!Tandem), the best score is reported. The first three columns report the accession number (Acc.nb.), gene name, and description according to the UniProtKB-SwissProt database (April 2012). The next two columns, the protein probability ("Prot prob") and the percentage of coverage ("Cov %"), report data provided by iProphet analysis. The last three columns, "GO biological process," "GO molecular function," and "GO cellular component," are as annotated in the GO database and present data retrieved from the UniProtKB-SwissProt database.

PI-PLC␤1b Nuclear Interactome
(*) Indicates proteins identified with one unique single peptide; the protein probability refers to the score of the unique peptide.
( §) Indicates proteins that have been grouped together according to the Peptide Prophet algorithm, mainly because they are isoforms or homologues and thus share the peptides identified.

FIG. 2.
In vitro validation of mass spectrometry identifications. MEL and Ba/F3 cells were stably infected with the retroviral pBB-IRESblast ® vector. Clones were selected with 4 g/ml blasticidin for 5 days. PI-PLC␤1 expression was evaluated by means of Western blot analysis in both MEL and Ba/F3 cells, wild-type (pBB-ev) and overexpressing PI-PLC␤1b (pBB-PLC␤1b) (A). The association of PI-PLC␤1b with Eef1a2, Kpna2, Kpna4, Kpnb1, Lmnb1, Phb2, and Srsf3 was verified in MEL nuclei (B), Ba/F3 nuclei (C), or both by immunoprecipitating 800 g of the respective nuclear lysate as described in "Experimental Procedures." Total nuclear lysate (60 g) and the immunoprecipitates were separated via 4%-15% gradient SDS-PAGE and immunoblotted with ␣-PI-PLC␤1 or a specific antibody directed against Eef1a2, Kpna2, Kpna4, Kpnb1, Lmnb1, Phb2, or Srsf3. Detailed information about antibodies is listed in supplemental Table S1. mIgG, mouse IgG; rIgG, rabbit IgG; gIgG, goat IgG. types), were also found. In addition, proteins involved in the regulation of transcription were identified in complex with PI-PLC␤1b, including Btf3 (a general transcription factor required for transcription initiation), Cnot1 (a transcriptional repressor belonging to the CCR4-NOT complex), Dhx9 (transcriptional activator), Hnrnpab (transcriptional repressor), Hspa9 (a heat shock protein that modulates demethylation and chromatin remodeling), Mbd3 (a transcriptional repressor and component of the NuRD complex), Pelp1 (*) (a co-activator of estrogen receptor-mediated transcription), Phb2 (a mediator of transcriptional repression through recruitment of histone deacetylases), and Sub1 (a general transcription coactivator). Among the proteins cited above, Chd4, Actl6a, Smarca4, Dnajc2, and Mbd3 play a role in regulating chroma- FIG. 3. Gene Ontology of nuclear PI-PLC␤1b complex biological processes. A, biological processes involving proteins identified in complex with PI-PLC␤1b. The percentage refers to the frequency of the process in the dataset, not in the whole mouse repository annotation. The frequency represents the percentage of protein entries (by gene name) in a particular GO category relative to the respective total number of entries. B, sub-division of cellular metabolic processes (blue bars), gene expression (green bars), and transport (red bars). The overrepresentation of each sub-category was calculated as the ratio among the dataset frequency and the mouse reference annotation frequency. tin remodeling and DNA methylation, thereby acting as epigenetic regulators (21).
Ribosome Biogenesis and Protein Synthesis-Large (60S) ribosomal subunit constituent proteins (Rpl10, Rpl14, Rpl15, Rpl20, Rpl27, Rpl29, Rpl3, Rpl30, Rpl31, Rpl17, Rpl18a, FIG. 4. Network analysis of nuclear PI-PLC␤1b complex protein-protein interactions. Analysis was performed using STRING 9.0, setting neighborhood, gene fusion, co-occurrence, co-expression, experiments, databases, and text-mining as prediction methods, with a medium confidence threshold (0.04). Protein-protein connections are presented in the confidence mode (heavier connecting lines indicate more evidence/ high score for the interaction). In the box are listed proteins for which no connections were found above the score threshold. Red lines represent manually added connections based on literature evidence (4,18,21). PI-PLC␤1 binding partners Srsf3, Lmnb1, Phb2, and Eef1a are highlighted in red.
Miscellaneous-In addition to the previously mentioned proteins involved in specific processes, we found several multifunctional proteins. Among these were Apex1, which plays a central role in the oxidative stress response, DNA repair, epigenetic regulation of gene expression by DNA methylation, and regulation of transcription factors such as Fos/Jun; Npm1, which is involved in ribosome biogenesis, histone assembly, cell proliferation, and the regulation of p53; and Csnk2a1, a regulator of cell cycle progression, apoptosis, and transcription. Csnk2a1 is a Ser/Thr kinase with many known substrates, several of which were found to be in complex with PI-PLC␤1b: Anp32b (G1-S progression of the cell cycle), Ssb (pre-mRNA folding and maturation), Sptan1 (secretion), Hnrpa1, Hnrpc, Fmr1, Eif5, Gpi, and Rangap1 (22). Moreover, several glycolytic enzymes were associated with PI-PLC␤1b, including Aldoa, Pgam1, Pgam2, Tpi, and Gpi1. In addition, the oxidative-phosphorylation-associated proteins Ndufv2, Ndufa9 (*), Ndufs3, Uqcrc1, Atad1 (*), Atp5a1, Atp5f1 (*), Atp5h, and Mdh2, as well as proteins that function during reactive oxygen species metabolic processes (Prdx1, Prdx2, Prdx6 (*), Prdx4, and Ola1), were identified. Some of these enzymes are known to be present within the nucleus (Table I) and to be involved in diseases such as cancer and leukemia (23)(24)(25)(26)(27)(28). Finally, Hmbs, responsible for heme and porphyrin biosynthesis, was identified in complex with PI-PLC␤1b. DISCUSSION AP-MS has emerged as a powerful tool for studying the protein interaction network of individual proteins of interest, coupling the specificity of the pray:bait protein isolation procedure and the sensitivity of high-throughput mass spectrometry analysis. In comparison to other systems (i.e. yeast-2hybrid and in vitro tagged bait systems), the major advantage of AP-MS is the isolation of multiprotein complexes in their endogenous forms, thereby preserving the native conditions and post-translational modifications. Often, though, unfiltered datasets may present with a large number of false-positive protein interactions, which basically derive from incorrect protein identifications and antibody background (29 -32). To avoid such false-positive identifications, we performed multiple-step AP experiments from three independent biological replicates using an antibody specific to PI-PLC␤1 and the corresponding normal mouse immunoglobulin, allowing for the identifications of 160 proteins in complex with PI-PLC␤1, which included direct interactors as well as proteins with varying degrees of interconnectivity.
Nuclear localized PI-PLC␤1 has been associated with multiple cellular processes, including proliferation, survival, differentiation, and metabolism (33). In particular, nuclear PI-PLC␤1 seems to play a critical role in the self-renewal and differentiation of leukemic cells both by regulating the cell cycle during the G1-S and G2-M phases and by promoting cell differentiation in an expression-dependent manner. Based on the evidence that PI-PLC␤1 nuclear localization is critical to its function, our group is attempting to characterize the mechanism(s) by which the PI-PLC␤1 signaling network is exerted. We previously reported that in MEL cells, cyclin D3 (3), the transcription factor p45/NF-E2 (enhancer binding protein for the ␤-globin gene) (6), the antigen CD24 (involved in differentiation and hematopoiesis) (7), the splicing factor Sfrs3 (Srp20) (18), and lamin B1 (4) are affected by nuclear PI-PLC␤1, among which Srp20 and lamin B1 were found to be associated with PI-PLC␤1 in co-immunoprecipitation experiments. In recent years, the involvement of PI-PLC␤1 in the development of cancer has been proposed, as studies on myelodysplastic syndrome (MDS) and acute myeloid leukemia have demonstrated that epigenetic and genetic modifications of the PI-PLC␤1 locus occur in patients during MDS progression to acute myeloid leukemia (34). All together, these data make the need to understand the effectors and interactors of PI-PLC␤1 even more urgent, in order to precisely target the mechanisms implicated in these diseases. This study reports the identification of 160 proteins in complex with nuclear PI-PLC␤1, many of which are associated not only with established mechanisms (e.g. regulation of the cell cycle and differentiation), but also with hypothesized roles for PI-PLC␤1 (e.g. apoptosis and RNA splicing), while in addition providing some completely new insights into PI-PLC␤1's function in the nucleus (e.g. nuclear transport mechanism). PI-PLC␤1 was the first phospholipase C isoform to be described that was recruited to the nucleus following mitogenic or differentiating stimuli. It has been demonstrated that PI-PLC␤1 contains a putative non-canonical nuclear localization signal in the carboxy-terminal region (35); nevertheless, the mechanism by which translocation occurs has never been further explored. One of the more interesting findings that have emerged from this study is the association of PI-PLC␤1b with the classical import proteins Kpna2, Kpna4, Kpnab1, Ran, and Rangap1. The classical import mechanism purports that a cargo protein binds to the adapter protein importin ␣ (Kpna) via its nuclear localization signal sequence, and this promotes the association with importin ␤ (Kpnb1), forming a ternary complex. The complex then translocates to the nucleus through the binding of Kpnb1 with the pore complex proteins in a Ran-dependent process that promotes dissociation of the complex at the nucleoplasmic side (36). However, non-conventional nuclear import mechanisms have also been established (37); for example, Yagisawa reported that PI-PLC␦1, although bearing a non-canonical nuclear localization signal, is still imported into the nucleus following an increase in the levels of intracellular Ca 2ϩ via a mechanism that is only importin b1-dependent (38). Thus, the findings presented here suggest that, in contrast to PI-PLC␦1, the nuclear translocation of PI-PLC␤1 is most likely mediated via the classical mechanism, in which both alpha and beta importins are required for nuclear import. Unlike other PI-PLC isozymes, PI-PLC␤s possess a long C-terminal sequence of about 400 amino acids, containing an unusually high proportion of lysine and arginine residues, downstream of their catalytic domains (35). The two isoforms of PI-PI-PLC␤1, 1a and 1b, show 94% homology and differ only at their C-terminal ends for a sequence of 75 amino acids in the 1a isoform and 32 amino acids in the 1b isoform. Both PI-PLC␤1a and PI-PLC␤1b present with a bipartite consensus sequence for nuclear localization (K1055, K1056 separated from K1069, K1071 by a linker of 12 amino acids), but a cluster of three lysines (K1056, K1063, and K1070) was demonstrated to be essential for nuclear translocation. It was therefore suggested these three lysines were critical sites for establishing interactions that retain PI-PLC␤1 within the nucleus (2). In a recently published paper, Scarlata and colleagues report that PKC phosphorylation on S887 (in the C-terminal region of both PI-PLC␤ 1a and 1b) also regulates the subcellular distribution of PI-PLC␤1, as the lack of phosphorylation keeps the enzyme located within the nucleus (39). The sequence of 75 amino acids exclusive to PI-PLC␤1a also contains a putative nuclear export sequence, as predicted by NetNES 1.1, which likely explains the predominant presence of PI-PLC␤1a in the cytoplasm. Together these findings suggest that PI-PLC␤1b (and probably PI-PLC␤1a) enters the nucleus complexed with the alpha/beta importin system, and then PI-PLC␤1b is retained inside the nucleus through binding to negatively charged components (2), whereas PI-PLC␤1a can be exported to the cytoplasm via its nuclear export sequence, thus providing an explanation for the different sub-cellular localizations of the two isoforms of PI-PLC␤1.
Several proteins identified as nuclear PI-PLC␤1b interactors are associated with leukemic malignancies, as either prognostic markers or potential targets for therapeutic intervention. Of particular interest was the identification of two splicing factors whose genes are often mutated in MDS, Srsf2 and Sf3b1 (40,41). Srsf2 is associated with a negative prognostic impact, as patients bearing mutations in Srsf2 have significantly inferior overall survival and a more rapid and frequent progression to acute myeloid leukemia (42,43). Depending on the disease classification and how the statistical analysis was performed and applied, Sf3b1, important for anchoring the spliceosome to precursor mRNA, has been proposed to serve either as a favorable marker in MDS or as an independent prognostic factor for progression (44,45). These findings suggest that imbalances in the spliceosome machinery can have a significant role in promoting leukemogenesis. Interestingly, other proteins associated with PI-PLC␤1b serve in mRNA processing as part of the spliceosome C complex (see "Results") or as independent splicing factors, such as Srsf2, Tarbp, and Srsf3/SRp20. In mammalian cells, constituents of the pre-mRNA splicing machinery are associated with a specific nuclear sub-compartment, called speckles (46). Noteworthy, PI-PLC␤1 is also localized to nuclear speckles with its binding partner Srp20 (18) and additional proteins implicated in inositide-dependent signal transduction, including PIP kinases, PI(4,5)P 2 , DGK, PLC␦4, PI3K C2␣, and phosphatases PTEN and SHIP2, as reviewed in Ref. 47. As for the current concepts concerning factors promoting hematological malignancies, in which gene mutations, deregulated gene expression, and epigenetic changes are seen as key steps in disease pathogenesis, PI-PLC␤1 was found to be involved in each of these processes (48). The identification of proteins implicated in the epigenetic regulation of gene expression through DNA methylation as PI-PLC␤1 binders was particularly relevant. The helicase DNAbinding protein Chd4 and the methyl-CpG-binding domain protein Mbd3 were identified in complex with nuclear PI-PLC␤1b. These two proteins are components of the nucleosome remodeling and deacetylase complex, which functions as a determinant epigenetic regulator, associating with methylated DNA, in order to bring about the deacetylation and demethylation of histones (49). Deregulation of the nucleosome remodeling and deacetylase complex by oncogenes can be a cause for aberrant cell proliferation in leukemogenesis that should be further exploited for therapies.
The overexpression of PI-PLC␤1 acts as a negative regulator of erythroid-induced differentiation in both murine and human erythroleukemia cells (5). Recently, our group reported that low-risk MDS patients refractory to erythropoietin therapy presented with increased PI-PLC␤1b expression at the end of the treatment (8). In the present study, mass spectrometry analysis identified proteins involved in red cell metabolism and in the regulation of erythropoietin mRNA. Among these proteins were the poly(rC)-binding proteins Pcbp1 and Pcbp2, which are reportedly associated with the erythropoietin mRNA 3Ј-UTR, in a region required for messenger RNA stability (50). Depletion of Pcbp1 and Pcbp2 in human erythroleukemia cells (K562) was shown to decrease cell proliferation, leading to G1 arrest via the induction of p21 WAF (51).
In 2000, a study by Lee et al. (52) demonstrated that PI-PLC␤1 exerts a protective effect against oxidative-stressinduced cell death, although the mechanism was not elucidated. In accordance with these previous findings, mass spectrometry identified several proteins involved in apoptosis that were associated with nuclear PI-PLC␤1b, some specifically exerting a negative effect against apoptosis. One of these proteins, Anp32b, is an anti-apoptotic protein that acts as a decoy caspase 3 substrate, inhibiting the pro-apoptotic function of caspase 3 (53). Another protein in complex with PI-PLC␤1b Nuclear Interactome PI-PLC␤1b was Apex 1 (Ape/Ref1), a multifunctional protein that stimulates the DNA binding activity of numerous transcription factors involved in cancer promotion and progression, such as the Fos/Jun Ap1 complex, NFkB, p53, and CREB. In addition, Apex 1 is also a DNA-damage response protein. The lyase activity of Apex 1 can repair apurinic/ apyrimidinic sites of DNA following damage by reactive oxygen species. Apex1 also possesses redox activity that can control transcription factors such as Jun and Fos, and this activity is enhanced following Apex1 phosphorylation by PKC and CK2, both of which can be activated through PI-PLC␤1mediated DAG production (54). In the nucleus, DAG production leads to the activation of cPKCs and a specific isoform of CK2, Csnk2a1, which was identified in complex with PI-PLC␤1b. Moreover, our group previously determined that PI-PLC␤1 can affect cyclin D3 promoter activity during differentiation through activation of the c-jun/AP1 complex (9). As PI-PLC␤1 does not directly bind to cyclin D3 but largely affects its expression and regulation, the identification of Apex1 and Cnsk2a1 opens a new field of investigation. Npm1, another multifunctional protein identified in complex with PI-PLC␤1, exerts control over Apex1 endonuclease activity, which has been implicated in the DNA repair of peroxidedamaged cells (55). Finally, Tpt1 and Eef1a2 were recently identified as anti-apoptotic proteins. Tpt1 controls the stability of p53, and Eef1a2 inhibits apoptosis and promotes the G1-S progression of the cell cycle in human multiple myeloma (56,57). The fact that PI-PLC␤1 bound with Eef1a2, together with its previous identification as a substrate of PKC␤I (19), suggests that it can act as an intermediate effector for nuclear inositide signaling elicited by PI-PLC␤1, not only during insulin-induced differentiation, but also in the regulation of the cell cycle progression (3) and apoptosis. These data give significant insight into the molecular environment that surrounds PI-PLC␤1 and provide evidence of the interaction of nuclear PI-PLC␤1b with a number of proteins involved in nuclear import, differentiation, mRNA processing, and apoptosis, therefore hinting at multiple novel targets for therapeutic intervention in hematological malignancies.