Nuclear Phosphatidylinositol 3,4,5-Trisphosphate Interactome Uncovers an Enrichment in Nucleolar Proteins

Polyphosphoinositides (PPIns) play essential roles as lipid signaling molecules, and many of their functions have been elucidated in the cytoplasm. However, PPIns are also intranuclear where they contribute to chromatin remodeling, transcription, and mRNA splicing. The PPIn, phosphatidylinositol 3,4,5-trisphosphate (PtdIns(3,4,5)P3), has been mapped to the nucleus and nucleoli, but its role remains unclear in this subcellular compartment. To gain further insights into the nuclear functions of PtdIns(3,4,5)P3, we applied a previously developed quantitative MS-based approach to identify the targets of PtdIns(3,4,5)P3 from isolated nuclei. We identified 179 potential PtdIns(3,4,5)P3-interacting partners, and gene ontology analysis for the biological functions of this dataset revealed an enrichment in RNA processing/splicing, cytokinesis, protein folding, and DNA repair. Interestingly, about half of these interactors were common to nucleolar protein datasets, some of which had dual functions in rRNA processes and DNA repair, including poly(ADP-ribose) polymerase 1 (PARP1, now referred as ADP-ribosyltransferase 1). PARP1 was found to interact directly with PPIn via three polybasic regions in the DNA-binding domain and the linker located N-terminal of the catalytic region. PARP1 was shown to bind to PtdIns(3,4,5)P3 as well as phosphatidylinositol 3,4-bisphosphate in vitro and to colocalize with PtdIns(3,4,5)P3 in the nucleolus and with phosphatidylinositol 3,4-bisphosphate in nucleoplasmic foci. In conclusion, the PtdIns(3,4,5)P3 interactome reported here will serve as a resource to further investigate the molecular mechanisms underlying PtdIns(3,4,5)P3-mediated interactions in the nucleus and nucleolus.

While the roles and regulation of PPIn have been extensively studied in the cytoplasm, the importance of their nuclear roles are only recently becoming more apparent (6,7). The presence of PPIn as well as specific PPIn enzymes was first demonstrated in an intra-nuclear pool not associated with the nuclear envelop (8,9). The concept of PPIn metabolism and signalling occurring in the nucleus independently of the cytoplasm was reported shortly after in several studies (10)(11)(12). Consequently, with the exception of PtdIns (3,5)P2, the remaining six PPIn have been detected and/or quantified in the nucleus (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). The intra-nuclear physico-chemical state of PPIn is still unclear, but several possibilities are emerging to explain how the acyl chains can be shielded from the aqueous environment. These have been shown to be buried in the hydrophobic ligand pocket of the nuclear receptors Liver Receptor Homolog-1 and Steroidogenic Factor 1 while the inositol headgroup remains accessible for modification by PPIn enzymes (26)(27)(28). Alternatively, the presence of nuclear lipid droplets has recently been reported in a few studies (29,30) including the newly discovered nuclear lipid islets, which consist of PtdIns(4,5)P2 nuclear aggregates possibly in the form of micelles, hence accommodating the acyl chains facing inwards (31).
To identify proteins interacting specifically with PtdIns(3,4,5)P3, several interactomics studies have been performed from a variety of cell types using either cytosolic (65)(66)(67) or whole cell extract (68). To enrich for nuclear PPIn-interacting proteins, for which the least is known, we developed a PPIn quantitative interactomics approach using isolated nuclei and based on an enrichment of PPIn interactors (46). This approach led to the identification of PtdIns(4,5)P2 nuclear interacting partners involved in mRNA transcription regulation, mRNA splicing and protein folding. In this study, we have performed quantitative mass spectrometry-based PtdIns(3,4,5)P3 interactomics from isolated HeLa nuclei using the same approach (46). We identified 179 potential PtdIns (3,4,5)P3 interactors with functions highly enriched in protein folding, RNA splicing, DNA repair and cell cycle regulation. Interestingly, half of these proteins were common to the T cell nucleolome protein list (69). In this study, we focused on Poly(ADP-Ribose) Polymerase 1 (PARP1, now referred as ADP-ribosyltransferase 1, ARTD1), validated its direct interaction with PtdIns (3,4,5)P3 and showed its colocalisation in nucleoli.
In sum, this study validates our approach to identify globally PPIn-interacting proteins based in the nucleus and represents a resource for further research efforts investigating the role of PtdIns (3,4,5)P3 in these interactions.

PtdIns(3,4,5)P3 is nucleolar in HeLa cells
To extend our previous findings on the nucleolar localisation of PtdIns (3,4,5)P3 previously observed in the breast cancer cell line AU565 (25), we determined its subcellular localization in actively growing HeLa cells by immunofluorescence staining and confocal microscopy ( Figure 1). Using specific antibodies to detect PtdIns (3,4,5)P3 ((25) and Figure 1B), we observed the presence of this PPIn in the nucleolus in 74% +/-10% of asynchronous HeLa cells in either intense or diffuse foci which colocalised with the nucleolar proteins nucleolin or the transcription factor upstream binding factor (UBF) ( Figure 1D and supplementary Figure   S1A). In addition, the presence of PtdIns (3,4,5)P3 in the nucleolus was supported using the purified PH domain of the general receptor of phosphoinositides-1 (GRP1, alias cytohesin-3) conjugated to EGFP and GST as a labelling probe. The PH domain of GRP1 is well known for its affinity to PtdIns (3,4,5)P3 while the K273A point mutation disrupts this interaction (70)(71)(72)(73).
When tested by lipid overlay assay, the WT GST-EGFP-GRP1-PH demonstrated interaction with PtdIns(3,4,5)P3 but not the K273A mutant ( Figure 1C). Labelling with this probe highlighted foci within rings detected by the nucleolar protein nucleophosmin when using the WT protein in about 50% of cells ( Figure 1E-F). In contrast, the percentage of cells showing these foci was greatly reduced to 6% when using the K273A mutant ( Figure 1E-F).

The nuclear PtdIns(3,4,5)P3 interactome is enriched in nucleolar proteins
The existence of PtdIns (3,4,5)P3 in the nucleus has been reported (19,24,25,63), but so far only a few nuclear proteins have been reported to interact with PtdIns(3,4,5)P3 and knowledge of its function is limited in this cell compartment. We sought to identify the interacting partners of PtdIns (3,4,5)P3 in the nucleus using a quantitative proteomics method previously developed for the identification of nuclear PtdIns(4,5)P2 effector proteins (46) with a view to identifying nuclear processes that PtdIns(3,4,5)P3 may regulate,. Following SILAC labelling of HeLa S3 cells, nuclei were isolated and incubated with neomycin to enrich for and displace potential PPIn-binding proteins from nuclei ( (64). In addition to these proteins, 20 from our dataset had previously been identified in PtdIns (3,4,5)P3 interactomes from whole cell extracts (67, 68) ( Figure 2D and Table 1). Several proteins were also common to the nuclear PtdIns(4,5)P2 interactome ((46), Table 2). Importantly, the majority of the identified proteins are likely to be direct PtdIns(3,4,5)P3 interactions, since only a few clusters involved in proteinprotein complexes were detected using the STRING web tool (Supplementary Figure S2). We further searched for the presence of PPIn-binding domains and found only 4 proteins, including dynamin 1, 2 and 3 harbouring a PH domain with previous knowledge of PPIn interaction (75,76)) or ATP binding cassette sub family F member 1 with the less studied PDZ domain (77). In contrast, the lysine/arginine rich motif (K/R-(Xn=3-7)-K-X-K/R-K/R), which we previously reported to be enriched in PtdIns(4,5)P2-binding proteins (46), was found in 38% of PtdIns(3,4,5)P3-associated proteins, accounting for a 1.4 fold enrichment compared to proteins pulled down by control beads ( Figure 2C and Supplementary Table S1) and 1.3 fold compared to proteins annotated to the nucleus (nucleome). For a better understanding of the biological processes of these proteins, they were mapped to the Gene Ontology (GO) database for biological processes and an enrichment test was performed using the PANTHER 14.1 web tool (2019-03-12 release, (78,79)). The biological processes that were over-represented by >5 fold are shown in Figure 2E. In particular, RNA splicing/processing, protein folding, cytokinesis and DNA repair were functions particularly enriched in the PtdIns (

PtdIns(3,4,5)P3 co-localizes with PARP1 in nucleoli
PARP1, a chromatin-associated protein previously reported to be abundant in the nucleolus (81,82) and which harbours one K/R motif, was identified as a specific PIP3 interacting protein with a PtdIns(3,4,5)P3/control SILAC ratio of 2.5 (Table 1 and   Supplementary Table S1). We first biochemically validated the direct interaction of PARP1 with PPIn by lipid overlay assay using phospholipid-immobilized strips and GST-PARP1 recombinant protein ( Figure 3A). PARP1 was found to interact with most PPIns as well as phosphatidic acid and phosphatidylserine but not with other glycerophospholipids or sphingolipid. In contrast, GST alone showed no interaction. We validated these results using a

Discussion
Evidence of the presence of PPIn in the nucleus together with the kinases responsible for their synthesis is now well established (34,38,(83)(84)(85). Interestingly, they are found in RNArich membrane-less compartments, such as the nuclear speckles and nucleolus in particular for PtdIns(4,5)P2 (20)(21)(22)53)   were indeed annotated to other compartments than the nucleus in that study, hence masking potential nuclear effector proteins. Except for a few proteins known to be engaged in proteinprotein complexes, the remaining proteins identified in this study are likely to be direct  (91). Although dynamin family members are GTPase considered to be localised on membranes and microtubules, dynamin-2 and -3 were identified in nucleolar proteomes (69,92). The function of dynamin in the nucleolus has not been investigated so far. IQGAP1 binds to PtdIns(3,4,5)P3 via an atypical PPIn binding domain lined with basic residues, with a distinct fold to most known domains (74). Although IQGAP1 has clear roles in the cytoplasm, it was also reported to accumulate in the nucleus at the G1/S phase of the cell cycle (93). Still how PPIn binding affects its nuclear role is unknown.
The majority of the PtdIns(3,4,5)P3-interactors identified in this study are characterised by the presence of at least one polybasic motif shown previously to serve as PPIn interaction sites via electrostatic interactions in other nuclear proteins (25,41,43,45,(48)(49)(50)(51)(52) or of basic patches, as found in nucleophosmin, ALY and OGT (42,44,64). This finding is consistent with the enrichment of such motifs in the nuclear PtdIns(4,5)P2 interactome that we have previously reported (46).  (13). The identity of these foci has not been investigated but appear to be distinct to PtdIns(4,5)P2-positive sites which localises to nuclear speckles (21) but not with PARP1. Knowledge of the synthesis route of this PPIn in the nucleus is limited but was shown in one study to be produced by the 5-phosphatase, SHIP2, by dephosphorylating PtdIns(3,4,5)P3 in vascular smooth muscle cells (99). SHIP2 (99) or in its phosphorylated form on serine 132 (100) was found in nuclear speckles in different cells. Alternatively, the class II PI3K, PI3KC2α, known to produce PtdIns(3,4)P2 by phosphorylating PtdIns4P, was also reported to localise in nuclear foci (101). and UBF (82,103,104). Altogether, these studies suggest that the organisation of proteins and lipids within the nucleolus is affected by the active transcription of rRNA. Interestingly, both PARP1 and nucleophosmin are also histone H1 interacting proteins (105), which is emerging to play important roles in the structure and integrity of the nucleolus (106). PARP1-dependent PARylation of histone H1 has been shown to remove this histone from the chromatin, hence causing it to relax (107). Nucleophosmin binds to histone H1.5 and has a silencing effect on this linker histone (105). A link between histone H1 and PPIn has previously been reported, demonstrating its interaction with PtdIns(4,5)P2 via its C-terminal region (57). binders. This resource is amenable for further biochemical and functional characterisation assessing how the array of nuclear, and in particular nucleolar, functions these interactions can regulate.

Materials and methods
Cell culture and SILAC labelling:  µl PPIn-conjugated bead slurry.

Proteomics
In-gel digestion: In-gel trypsin digestion was performed as described (108)  which were searched as variable modifications. Using a reversed decoy database, false discovery rate (FDR) was less than 1%. Only proteins identified with at least 2 peptides (and including at least 1 unique peptide) and common to the two replicate runs were kept.

Bioinformatic analyses
For the K/R polybasic motifs search, an in lab Linux shell script was used to first download the sequences of the PtdIns(3,4,5)P3 pulled down proteins from Uniprot (curl https://www.uniprot.org/uniprot/) (using the curl tool) and search for the (K/R-(X3-7)-K-X-K/R-K/R) motif was then carried out using the grep tool.
For the enrichment analyses, the identified Uniprot entries were statistically compared to those of the human genome restricted to entries annotated to the nucleus compartment (GO:0050789) using PANTHER classification system version 13.1 (78,79 STRING analysis of all PtdIns(3,4,5)P3-binding protein entries was based upon experimental prediction methods and a confidence score > 0.9.

Immunofluorescence staining and microscopy
HeLa cells grown on 12 mm coverslips were fixed with 3.7 % paraformaldehyde for 10 min and washed twice with PBS and then permeabilised with 0.25 % Triton X-100 in PBS for 10 min at room temperature. Cells were blocked for 1 h with 5% goat serum in PBS-0.1% Triton.
Primary antibody (diluted in blocking buffer) incubation was performed overnight at 4°C followed by secondary antibody conjugated to Alexa-488 or Alexa-594 incubation for 1 h at room temperature. Washes were performed with PBS-T (0.05% Tween20), between each antibody incubation. Nucleic acid staining was performed by 15 min incubation with Hoechst 33342 diluted 1:1,000 in PBS. For antibody dilutions, see the supplementary Table S2. For cell labelling using the recombinant EGFP-GRP1-PH protein, cells were permeabilised with 0.1% Triton X-100 in PBS and blocked in 3% fatty-acid free BSA and 0.05% Triton-X100 in PBS for 1 h at RT. This was followed by incubation with 40 µg/ml of the probe in 1% fatty-acid free BSA and 0.05% Triton-X100 in PBS for 2 h at RT. Images were acquired with a Leica DMI6000B fluorescence microscope using x40 or x100 objectives or Leica TCS SP5 confocal laser scanning microscope using a 63x/1.4 oil immersion lens. Images were processed with a Leica application suite V 4.0.

SDS-PAGE and Western Immunoblotting
Proteins were resolved by SDS-PAGE and then transferred to nitrocellulose membranes. The membrane was then blocked with 7% milk in PBS-T (PBS pH 7.4, 0.05 % Tween-20) for 1 hour at room temperature before incubation with primary antibodies overnight at 4°C (for  antibody dilutions see the supplementary Table S3). After washing with PBS-T, the membrane was incubated with HRP conjugated secondary antibodies for 1 hour at room temperature. The enhanced chemiluminescence (ThermoFisher Scientific) was added and the Chemidoc XRS+ imaging system from Bio-Rad was used for visualization.

Recombinant protein expression and purification
The was as described previously (46).

ID
Proteins pulled down by PtdIns (3,4,5)P3 and annotated to the DNA repair enriched process, identified with at least 2 peptides, with heavy/light log2 ratios >0.5, are indicated in this table.