Novel Splice Variants of ING4 and Their Possible Roles in the Regulation of Cell Growth and Motility*

The ING4 gene is a candidate tumor suppressor gene that functions in cell proliferation, contact inhibition, and angiogenesis. We identified three novel splice variants of ING4 with differing activities in controlling cell proliferation, cell spreading, and cell migration. ING4_v1 (the longest splice variant), originally identified as ING4, encodes an intact nuclear localization signal (NLS), whereas the other three splice variants (ING4_v2, ING4_v3, and ING4_v4) lack the full NLS, resulting in increased cytoplasmic localization of these proteins. We found that one of the three ING4 variants, ING4_v2, is expressed at the same level as the original ING4 (ING4_v1), suggesting that ING4 variants may have significant biological functions. Growth suppressive effects of the variants that have a partial NLS (ING4_v2 and ING4_v4) were attenuated by a weaker effect of the variants on p21WAF1 promoter activation. ING4_v4 lost cell spreading and migration suppressive effects; on the other hand, ING4_v2 retained a cell migration suppressive effect but lost a cell spreading suppressive effect. Therefore, ING4_v2, which localized primarily into cytoplasm, might have an important role in the regulation of cell migration. We also found that ING4_v4 played dominant-negative roles in the induction of p21WAF1 promoter activation and in the suppression of cell motility by ING4_v1. In addition, ING4 variants had different binding affinities to two cytoplasmic proteins, protein-tyrosine phosphatase, receptor type, f polypeptide (PTPRF), interacting protein (liprin), α1, and G3BP2a. Understanding the functions of the four splice variants may aid in defining their roles in human carcinogenesis.

In this study, we describe three novel splice variants of ING4, ING4_v2 (a 3-bp skip form), ING4_v3 (a 9-bp skip form), and ING4_v4 (a 12-bp skip form), besides ING4_v1 (the originally enrolled ING4, the longest form). These variants are produced from the alternative use of two splice donor sites at the end of exon 4 and two splice acceptor sites at the start of exon 5 of the ING4 gene, although one of the three variants, ING4_v4, was previously attributed to a common deletion mutation (6). The alternative RNA splicing of the ING4 pre-mRNA causes a partial loss of an NLS and affects nuclear localization of the proteins. We show that the small deletion in the ING4 protein leads to functional differences between the ING4 variants. Growth suppressive effects of the variants that have the partially missing NLS were attenuated by weaker activity on p21 WAF1 induction. In addition, ING4_v4 showed attenuated suppressive effects on cell spreading and migration compared with the original ING4 (ING4_v1). On the other hand, ING4_v2 only lost the suppressive effect on cell spreading, suggesting an important role for ING4_v2 on the regulation of cell migration. ING4_v4 played dominant-negative roles on these ING4_v1 effects. In addition, we found that ING4_v1 and ING4_v2, but not ING4_v4, have a binding affinity to Liprin ␣1 that may play a role in the regulation of focal adhesion disassembly (13,14). In addition, only ING_v1 has a binding affinity to G3BP2a, which is involved in Ras signaling, NF-B signaling, and the ubiquitin proteasome system (15)(16)(17). These differences may affect the function of ING4 splice variants.
Construction of Minigene Plasmids-All sequences for cloning were amplified by PCR using KOD-Plus (Novagen, Madison, WI) and verified by DNA sequencing. A partial ING4 gene from exon 3 to exon 6 and its point mutants were cloned into pcDNA 3.1 (ϩ) (Invitrogen) using the HindIII and XbaI site. Expression plasmids for two major ING4 splice variants (ING4_v2 and ING4_v4) were derived from the original ING4 plasmid (pFLAG-CMV-2 and pcDNA3.1/Hygro) as a template.
Transfection of Plasmids and Isolation of RNA-The plasmids were transfected into cells by Lipofectamine reagent (Invitrogen). After 24 h of transfection, cells were harvested, and total RNAs were isolated from the cells with TRIzol reagent (Invitrogen) according to the manufacturer's protocol. Then 5-10 g of the total RNAs were reverse-transcribed using the SuperScript III first-strand synthesis system (Invitrogen) according to the manufacturer's protocol.
Separation of Each ING4 Variant by PCR-Restriction Fragment Length Polymorphism -The transcribed artificial cDNAs from the series of minigenes were amplified using a forward primer at a vector-specific region (T7 promoter) between the transcriptional starting site and exon 3 of ING4 (5Ј-TAATAC-GACTCACTATAGGG-3Ј) and a reverse primer at exon 5 of ING4 (5Ј-ACTTCTTCTGGGCAGTCTTG-3Ј). A nested PCR was performed on the first PCR using a forward primer at exon 4 (5Ј-ACAAACACATTCGGCGGCTG-3Ј), combined with a reverse primer at exon 5 (5Ј-ACTTCTTCTGGGCAGTC-TTG-3Ј). One l of the gel-purified first PCR product was used as a nested PCR template. The nested PCR product was digested by HaeIII (New England Biolabs, Beverly, MA), separated by 9% polyacrylamide gel (AccuGel 19:1, National Diagnostics, Atlanta, GA), and stained with SYBR Gold (Molecular Probes, Eugene, OR).
Fractionation of Proteins-Cells were transfected with different plasmids that express FLAG-tagged ING4_v1, ING4_v2, and ING4_v4. 24 h after transfection, the cellular proteins were fractionated by the nuclear/cytosol fractionation kit (BioVision Research Products, Mountain View, CA) and the same amount of proteins from nuclear and cytoplasmic fractions was applied into a gel. Immunoblotting was performed by a standard protocol. FLAG-tagged proteins were detected by anti-FLAG antibody (Sigma). Fractionation was evaluated by a nuclear marker, anti-Lamin A/C antibody (Santa Cruz Biotechnology, Santa Cruz, CA).
Colony Formation Assay-Three cancer cell lines, RKO, U-118 MG, and U-2 OS were used for the experiments. Cells were plated on 10-cm dishes (5 ϫ 10 5 cells/dish), cultured for 12 h at 37°C, and then transfected with 5 g of pcDNA3.1/Hygro (5.6 kb) as a control, and 5.7 g of pcDNA3.1/Hygro-ING4_v1 (6.35 kb), pcDNA3.1/Hygro-ING4_v2 (6.35 kb), and pcDNA3.1/Hygro-ING4_v4 (6.35 kb). Cells were cultured in the selection medium (hygromycin B; Invitrogen) as follows: 100 g/ml for U-118 MG, 200 g/ml for U-2 OS, and 300 g/ml for RKO. After a 2-week selection, cells were fixed on the plates with 4% formaldehyde (Sigma) and stained with 0.1% crystal violet (Sigma). The area of the colonies (pixels) in each dish was calculated by Photoshop CS (Adobe, San Jose, CA). The data are shown as the average and standard deviation of three independent experiments. Statistical analysis was carried out by both Scheffé's F test and Student's t test.
Reporter Gene Assay-Cells were plated into 12-well plates and transfected with 0.1 g of pEYFP-Nuc that encodes the enhanced yellow-green variant of the Aequorea victoria green fluorescent protein gene (BD Biosciences) and with 0.5 g of WWW-Luc-p21 using Lipofectamine 2000 (Invitrogen) and with the indicated amount of pcDNA3.1/Hygro empty vector, pcDNA3.1/Hygro-ING4_v1, pcDNA3.1/Hygro-ING4_v2, or pcDNA3.1/Hygro-ING4_v4. The cells were lysed in cell culture lysis reagent (Promega, Madison, WI), and the lysates were collected to measure the p21 WAF1 promoter activity 48 h after transfection. The transcriptional efficiency was examined by detecting a signal from enhanced yellow fluorescent protein (the fluorescence excitation maximum is 513 nm and the peak of the emission spectrum is 527 nm), and the promoter activity (the peak of the emission spectrum of luciferase is 562 nm) was examined by the Bright-Glo luciferase assay system (Promega). The activities of enhanced yellow fluorescent protein and luciferase were quantified with an FLX800 microplate fluorescence reader (Bio-Tek Instruments, Winooski, VT).
Cell Spreading Assay-RKO cells transiently transfected with ING4 expression plasmids were plated on glass coverslips in medium depleted of serum. Following overnight adherence at 37°C, the starved cells were stimulated with 5% fetal bovine serum to observe the cellular response in membrane spreading. At designated time points, cells were fixed and stained on F-actin by rhodamine-phalloidin. Quantitative data of cell spreading, the percent of the cells that have filopodia and/or lamellipodia (% S), were derived from the equation % S ϭ (S/T) ϫ 100%, where S is the number of cells containing filo/lamellipodia, and T is the total number of cells counted.
Modified Boyden Chamber Migration Assay-RKO cells transiently transfected with ING4 expression plasmids were subjected to a transmembrane cell migration assay containing 8 M pore size polystyrene membranes coated with FluoroBlok materials in the migration chamber (BD Biosciences). By following the manufacturer's protocol, the transverse cells were stained with calcein AM (Molecular Probes, Eugene, OR) in the lower chamber, and the magnitude of activated fluorescence was read by fluorescence spectrometry (Victor II; PerkinElmer Life Sciences).
Immunoprecipitation-Cells were transfected with different plasmid vectors that express FLAG-tagged or untagged ING4_v1, ING4_v2, or ING4_v4, and a plasmid that expresses FLAG-tagged G3BP2a that was a generous gift from Dr. Derek Kennedy (University of Queensland, Australia). 24 h after transfection, the cells were lysed in Nonidet P-40-based lysis buffer (0.1% Nonidet P-40, 150 mM NaCl, 50 mM Tris-HCl, and protease inhibitor mixture; Calbiochem). Immunoprecipitation was performed using FLAG M2 resin (Sigma). Each precipitate was washed, and proteins were eluted in sample buffer. Immunoblotting was performed by standard protocols. FLAG-tagged ING4_v1, ING4_v2, ING4_v4, and G3BP2a were detected by rabbit anti-FLAG antibody (Sigma). Liprin ␣1 was detected by chicken antiliprin ␣1 antibody (Genway, San Diego, CA). Untagged ING4_v1, ING4_v2, and ING4_v4 were detected by anti-ING4 antibody (Rockland, Gilbertsville, PA). ING4_v2 were expressed at a high level with a similar ratio (both variants were expressed around 40 -50% among the four variants), whereas the ratio of ING4_v4 and ING4_v3 was ϳ10% and less than 10%, respectively. Because ING4_v4 was reported previously as a 12-bp common deletion mutation (6), we examined the genomic DNA sequence of the ING4 gene in H82 that was reported to have a mutation (6), and did not detect any mutations or polymorphisms (data not shown). These variations occurred only at the cDNA level, suggesting they were presumably splicing variants.

Identification of ING4 Variants by EST
Determination of Alternative Usage of Splice Donor and Acceptor Sites for the Production of the Splice Variants by Minigene Experiment-To prove our assumption, we analyzed the ING4 gene sequence and identified two potential alternative splice donor sites, D1 and D2, at the end of exon 4 and two potential alternative splice acceptor sites, A1 and A2, at the start of exon 5 (Fig. 1A). The consensus sequences for the splice donor and acceptor sites are MAGGTRAGT and YYYYYYYY-NCAGR, respectively. D1 (AAGGCAAAA) matched 67% with the consensus sequence, whereas D2 (AAG-A-GTGAGG) matched 89% with the consensus sequence but includes an extra "A" base between MAG and GTRAGT. A1 (CCTCTTC-CCTAGA) matched 92% with the consensus sequence, whereas A2 (CTTCCCTAGAAGG) matched 85% with the consensus sequence. It was predicted that ING4_v1 might be a spliced product from D1 to A1, and ING4_v2, ING4_v3, and ING4_v4 might be the spliced products from D1 to A2, D2 to A1, and D2 to A2, respectively.
To prove usage of the potential splice sites, a "minigene" approach was utilized (Fig. 1B). The plasmid contains a partial ING4 gene from exon 3 to exon 6, including the introns, a TATA box, a transcriptional starting site, and a T7 promoter just before ING4 exon 3, and a poly(A) signal just after ING4 exon 6. The expected artificial pre-mRNA should undergo alternative RNA splicing with cellular splicing machinery. Along with the wild-type minigene, we also constructed five Each minigene was transfected into cells, and total RNA was extracted from the cells 24 h after transfection for cDNA preparation by reverse transcription. We amplified the artificial cDNAs from the minigenes using a forward primer at the T7 promoter (vector specific region) and a reverse primer at exon 5 of ING4 (Fig. 1B), and reamplified it by a nested PCR using a forward primer at exon 4 and the same reverse primer. Subsequent HaeIII digestion of the nested PCR products allowed us to distinguish each variant by electrophoresis. The HaeIII restriction enzyme that recognizes the GGCC sequence creates an 82-bp fragment specific for ING4_v1, a 167-bp fragment specific for ING4_v2, a 73-bp fragment for ING4_v3, and a 70-bp fragment for ING4_v4 (Fig. 1B). Fig. 1C shows the result of the minigene experiment. For positive controls, each variantcontaining plasmid was used (Fig. 1C, lane 1-4). All four variants were made by alternative splicing from the wild-type minigene of ING4 (Fig. 1C, lane 5). As predicted, only ING4_v3 and ING4_v4 were generated from the D1 mutant (Fig. 1C, lane 6). ING4_v1 and ING4_v3 were generated from the A2 mutant (Fig. 1C,  lane 7). ING4_v1 and ING4_v2 were generated from the D2 mutant (Fig.  1C, lane 8). Only ING4_v3 (Fig. 1C,  lane 9) and ING4_v1 (lane 10) were generated from the D1A2 mutant and D2A2 mutant, respectively. This result clearly shows the alternative usage of two splice donor and acceptor sites. ING4_v4 (the 12-bp skip type) was attributed to a common deletion mutation at 379 -390 (6), although we show here that the skip region of ING4_v4 is 383-394. The variation is the result from an AAAG sequence duplication of 379 -382 and 391-394, and 391-394 is correct. Our data indicate that it is one of the three novel splice variants derived by RNA splicing from D2 to A2 (Fig. 1A).

Expression Level of the ING4 Variants in Various Tissues-Although it was reported that ING4
expression is ubiquitous in various tissues (2), the expression levels of each of the ING4 variants have not been examined. Therefore, we designed variant-specific TaqMan primer and probe sets along with a variant-nonspecific primer and probe set. First, a suitable annealing temperature for the specific amplification of each variant was determined (supplemental Fig. S1). Then the expression level of the total and each splice variant was examined. The variants were expressed ubiquitously among all tissues and cell lines we examined (supplemental Fig. S2). The expression pattern of all the variants was similar. We did not observe the expression of each variant in specific tissues or cell lines. Among the tissues that we examined, all the variants were expressed at the highest level in the brain and testis, and the next highest level in the pancreas. In the other tissues, all variants were expressed at lower levels. Among the cell lines that we examined, all the variants were expressed at the highest levels in a glioblastoma cell line, M059K, and embryonic kidney cell lines 293 and 293T. In the rest of the cell lines, all variants were expressed at lower levels.
Subcellular Localization of the ING4 Variants-So far, several putative motifs, a coiled-coil domain, two NLSs, an endoplasmic reticulum membrane retention signal, and a PHD finger motif were predicted by computer searches (supplemen- tal Fig. S3A, PSORT II Prediction and SMART). The variations of the three novel ING4 variants were located at just one of the predicted NLSs (NLS1). Because two NLSs were predicted by the computer search (PSORT II Prediction), we examined whether the predicted NLSs were functional. One study reported that NLS1 is a functional NLS (4). We hypothesized that NLS2 might also function as an NLS. We constructed FLAG-tagged NLS1, NLS2, and NLS1 ϩ NLS2 deletion mutants (FLAG-⌬NLS1, FLAG-⌬NLS2, and FLAG-⌬NLS1 ϩ 2; supplemental Fig. S3B). The expression of the mutant proteins was first determined with the anti-FLAG M2 antibody by Western blotting (supplemental Fig. S3C), and their subcellular localization was then examined by indirect immunofluorescence using the same antibody. The deletion of the NLS2 region did not affect nuclear localization of ING4, although NLS2 might alter the distribution of ING4 in the nucleus (supplemental Fig. S3D). These results indicate that NLS1 is the only ING4 NLS and suggest that sequence variations at the NLS1 region because of alternative RNA splicing could affect the localization of ING4 variants.
Subsequently, we examined the subcellular localization of two major ING4 variants, ING4_v2 (K131S,G132⌬) and ING4_v4 ( 129 KKKG 132 ⌬⌬⌬⌬). Although the two variants were expressed equally well by Western blotting assays ( Fig. 2A), they, especially ING4_v4, were localized primarily in the cytoplasm with a lesser amount in the nucleus (Fig. 2, B-D). This observation is consistent with a reported study that ING4 contained NLS deletion (ING4⌬NLS) exhibited primarily in the cytoplasm (4). These results suggest a different role for the variants in cells through different localization.

Attenuated Growth Suppression of ING4 Variants-
The overexpression of ING4 variants was conducted using two p53 wild-type cancer cell lines, U-2 OS and RKO, and a p53-mutated cell line, U-118 MG. We confirmed equal expression levels of all of the splice variants by Western blotting (data not shown). Cell growth was suppressed by ING4_v1 effectively in all cell lines we examined, but the variants, ING4_v2 and ING4_v4, that have a partial NLS were less effective (Fig.  3, A and B). We did not observe a significant p53 dependence. The ING4 variant effects on the p21 WAF1 promoter were further examined by a reporter assay. Fig. 4A shows that ING4_v1 overexpression could activate the p21 WAF1 promoter (1.5-fold when compared with the vector control), but both ING4_v2 and ING4_v4 were weaker inducers (1.2-and 0.8-fold when compared with the vector control, respectively). Equal expression levels of all of the splice variants were confirmed by Western blotting (data not shown). In addition, we found ING4_v4 could block ING4_v1-mediated p21 WAF1 induction (Fig. 4B), suggesting a dominant-negative effect of ING4_v4 on ING4_v1.
ING4 Splice Variants Lose the Capacity to Suppress Cell Spreading and Migration-Because we recently found that ING4_v1 interacts with several cytoplasmic proteins that might associate with cell migration and spreading under some conditions, 4 we examined the activity of the two major splice variants (ING4_v2 and ING4_v4) in cell spreading and cell migration assays in comparison with ING4_v1. We found that in the cell spreading assay, both ING4 variants (ING4_v2 and ING4_v4) lost the capacity to suppress actin filament polymerization and the consequent membrane spreading (Fig. 5, A-C). In the modified Boyden chamber assay, we found that ING4_v4 completely lost the capacity to suppress cell migration, whereas ING4_v2 retained the capacity (Fig. 6A). These results regarding cell migration were further confirmed by the scratch assay (supplemental Fig. S4). We confirmed equal expression levels of all of the splice variants by Western blotting (data not shown). In light of the loss-of-function characteristic of ING4_v4, which suppressed neither cell spreading nor cell migration, we performed a competition assay between ING4_v1 and ING4_v4. ING4_v4 inhibited the filopodia/lamellipodia formation by ING4_v1 (Fig. 5D) and also completely abrogated the migration suppression activity conducted by ING4_v1 (Fig. 6B), suggesting a dominant-negative regulation of ING4_v4 on ING4_v1.

ING4 Splice Variants Have Different Binding Affinity to Liprin ␣1
and G3BP2a-To explore the underlying mechanisms of ING4 variants in cell spreading and cell migration, we focused on two ING4_v1-binding proteins that we have recently determined. 4 Liprin ␣1 may play a role in the regulation of focal adhesion disassembly (13,14). G3BP2 is a member of Ras-GAP Src homology 3 domain binding protein (G3BP). Evidence to date suggests that G3BPs may modulate cell proliferation through Ras signaling, NF-B signaling, and the ubiquitin proteasome system (15)(16)(17). ING4_v1 and ING4_v2 interacted with Liprin ␣1, but ING4_v4 did not (Fig. 7A). Only ING4_v1 had a binding affinity to G3BP2a (Fig. 7B).

DISCUSSION
Alternative RNA splicing plays a major role in modulating gene function by expanding the diversity of expressed mRNA transcripts (18 -21). Alternative splicing determines cell fate in numerous contexts, such as sexual differentiation in Drosophila and apoptosis in mammals, and aberrant regulation of alternative splicing has been implicated in human diseases (22)(23)(24). Alternative RNA splicing is controlled by multiple splicing factors whose expression and status are tightly regulated in animal development, cell cycle, and cell differentiation. Although intron removal from a pre-mRNA by RNA splicing was initially thought to be controlled by intron splicing signals (25), exon sequences have been found recently to also regulate RNA splicing, polyadenylation, export, and nonsense-mediated RNA decay in addition to their coding function (26). The regulation of alternative RNA splicing by exon sequences is largely attributed to the presence of two major cis-acting elements in the regulated exons, the exonic splicing enhancer (ESE) and suppressor or silencer (26).
In this study, in addition to the original ING4 (ING4_v1) (1), we describe three novel splice variants of ING4, ING4_v2, v3, and v4, that are produced from two alternative splice donor sites at the end of exon 4 and two alternative acceptor sites at the start of exon 5. Although ING4_v4 was reported to be a common deletion-type mutation (6), we proved that the deletion was one of the four splice variants of ING4 by the minigene analysis. Interestingly, although ING5 has a similar sequence with ING4, no ESTs corresponding to ING4 splice variants have been enrolled, probably because small sequence differences between ING4 and ING5 in the region of exon 4 and exon 5 make the ING5 missing the splicing signals seen in the ING4 (supplemental Fig. S5A). These ING4 variants were expressed ubiquitously in all tissues and cell lines examined. All variants were expressed at high levels in the same tissues and cell lines, suggesting that the four splice variants are generated from the same pre-mRNA at the same ratio by cellular splicing machinery. In addition, several stimulations failed to change the composition of the variants. Serum starvation induced expression levels of all splice variants, on the other hand, serum stimulation suppressed expression levels of all splice variants in WI-38 and MRC5 normal fibroblast cells (supplemental Fig. S6A). There was no significant difference between the variants, although this observation might be very important for understanding the mechanism of cell cycle regulation by ING4. Treatment of several cell lines with DNA-damaging reagents, adriamycin and etoposide, also failed to differentially induce type-specific expression of the variants, although both adriamycin and etoposide induced expression levels of all splice variants examined in a particular cell line, U-118 MG cells, at a similar ratio (supplemental Fig. S6B). Considering the presence of putative ESE-like motifs (26) in exons 4 and 5 of the ING4 gene, we overexpressed SF2/ASF, one of the ESE-binding proteins (27)(28)(29), in cells to examine whether the protein might induce type-specific expression of the variants, and we did not observe the specific induction of any variants (data not shown). Although the exact mech-  NOVEMBER 10, 2006 • VOLUME 281 • NUMBER 45 anism that controls the modulation of each ING4 variant in mammalian cells remains to be understood, we found homologues of ING4_v1,-v2, and -v4 in the EST data base of mouse, rat, and pig and a homologue of ING4_v3 in the EST data base of pig (supplemental Fig. S5B). This conservation across species suggests their significance in biological functions.

Functional Analysis of Novel Splice Variants of ING4
Because of the relatively low expression level of ING4_v3, we focused on the two major novel splice variants, ING4_v2 and v4, for further functional analysis. We tried to analyze endogenous ING4 splice variants by making small interfering RNA but failed, because of cross-reactions between the variants. Separation of each endogenous ING4 variant by Western blotting also failed, because of the small difference of molecular weight between the splice variants. Therefore, we applied an overexpression approach to seek the functions of the splice variants. The two variants, ING4_v2 and ING4_v4, that have a partially deleted NLS were primarily distributed in the cytoplasm, whereas ING4_v1 with an intact NLS was displayed primarily in the nucleus, suggesting that the novel ING4 variants might function primarily in the cytoplasm and lose their function in the nucleus. We performed a colony formation assay and found that the variants with a partial NLS exhibited an attenuated growth suppressive effect when compared with ING4_v1, which was associated with their ability to induce p21 WAF1 promoter activity. ING4 interacts with p53 in modulating cell cycle arrest and apoptosis (1, 4), but we did not observe a p53 dependence in the colony formation assay. We also observed that ING4_v1 modestly activated p21 WAF1 promoter activity. There are several reports regarding a p53-independent mechanism of p21 WAF1 activation (30, 31); a variety of transcriptional factors, including STATs, E2Fs, AP2, C/EBPa, C/EBPb, and GAX, and other tumor suppressors, including BRCA1, transforming growth factor-␤, and Wnt-1, regulate p21 WAF1 transcription. ING4 could activate one of these p53-independent pathways as well as the p53dependent pathway under some conditions. It was reported that ING4_v1 acetylates histone H4 (9), and its PHD finger motif interacts with methylated histone H3 (10). In addition, several binding partners of ING4_v1 in nucleus have been determined (9). Therefore, the nuclear ING4_v1 may have multiple functions.
Recently, we found that ING4_v1 interacts with several cytoplasmic proteins, including Liprin ␣1 and G3BP2 by protein pulldown and subsequent mass spectrometry analysis, 4 suggesting important roles of cytoplasmic ING4_v1 similar to ING2 that have been reported to change subcellular localization dynamically between the cytoplasm and the nucleus under some conditions through its PHD finger motif that has a high identity with that of ING4 (32). Liprin ␣1 is a cytoplasmic protein necessary for focal adhesion and intracellular vesicle transport (13,14). G3BP2 has been implicated in Ras signaling, NF-B signaling, and the ubiquitin proteosome pathway (15)(16)(17). In addition, G3BP2 is overexpressed in human breast cancer tissue (33). Because we are interested in the function of Liprin ␣1 in focal adhesion and the association between the Ras signaling pathway and G3BP2, we are investigating the functions of ING4_v1 in cell migration and cell spreading. Because G3BP2 retains IB␣/NF-B complexes in the cytoplasm, ING4_v1 might also be retained in the cytoplasm by G3BP2 that is overexpressed in cancer (33).
We observed dominant-negative effects of ING4_v4 on ING4_v1 in p21 WAF1 promoter activation, cell spreading, and cell migration. Because ING4_v1 and ING4_v4 neither affect their localization when they overexpressed in same cells (supplemental Fig. S7; ING4_v1 localizes in the nucleus, whereas ING4_v4 localizes in the cytoplasm) nor make heterodimers or homodimers (data not shown), it seems that the mechanisms of the dominant-negative effects are not a physical interaction between these two variants. Cytoplasmic ING4_v4 may anchor partner nuclear proteins (9) of ING4_v1 in the cytoplasm.
ING4 has also been shown to inhibit angiogenesis though an interaction with NF-B (2) and to regulate HIF through an interaction with HPH-2/PHD2, a family member of the prolyl hydroxylases (7). NF-B shuttles between the cytoplasm and nucleus dynamically (34), and HPH-2/PHD2 is expressed both in the cytoplasm and the nucleus (7). Our report suggests that there are many chances for spatial interaction between ING4 and their partners both in the cytoplasm and nucleus. The balance among the four splice variants of ING4 and their subcellular localization may modulate the recently reported functions in controlling the cell cycle checkpoint, cell contact inhibition, and angiogenesis.