|Home | About | Journals | Submit | Contact Us | Français|
The five members of the inhibitor of growth (ING) gene family have garnered significant interest due to their putative roles as tumor suppressors. However, the precise role(s) of these ING proteins in regulating cell growth and tumorigenesis remains uncertain. Biochemical and molecular biological analysis has revealed that all ING members encode a PHD finger motif proposed to bind methylated histones and phosphoinosital, and all ING proteins have been found as components of large chromatin remodeling complexes that also include histone acetyl transferase (HAT) and histone deacetylase (HDAC) enzymes, suggesting a role for ING proteins in regulating gene transcription. Additionally, the results of forced overexpression studies performed in tissue culture have indicated that several of the ING proteins can interact with the p53 tumor suppressor protein and/or the nuclear factor-kappa B (NF-κB) protein complex. As these ING-associated proteins play well-established roles in numerous cell processes, including DNA repair, cell growth and survival, inflammation, and tumor suppression, several models have been proposed that ING proteins act as key regulators of cell growth not only through their ability to modify gene transcription but also through their ability to alter p53 and NF-κB activity. However, these models have yet to be substantiated by in vivo experimentation. This review summarizes what is currently known about the biological functions of the five ING genes based upon in vitro experiments and recent mouse modeling efforts, and will highlight the potential impact of INGs on the development of cancer.
Cancer is a complex genetic disease initiated by cells that have accumulated multiple mutations that ultimately bestow malignant characteristics. With rare exceptions, cancers arise from single somatic cells and their progeny. As the neoplastic cells divide, they accumulate either genetic or epigenetic changes resulting in altered phenotypes that provide various selective advantages to the cell as previously described by Hanahan and Weinberg (2000) and Ponder (2001). One key class of genes altered in cancer is the tumor suppressors. Tumor suppressor proteins have been found to regulate numerous cellular processes, including cell cycle arrest, cell senescence, DNA repair, signal transduction, and apoptosis. Reflecting this wide variety of regulatory effects, tumor suppressors include proteins that are involved in transducing external growth signals into the cell, proteins that sense or respond to genetic or metabolic insult, kinases that regulate the function of other enzymes in the nucleus or cytoplasm, proteins that can alter the cellular location or cellular levels of other regulatory proteins, and transcription factors that alter the expression of genes involved in cell growth or survival. In addition, tumor suppressors include proteins that regulate chromatin remodeling and/or modify histones to alter gene expression, including certain subunits of the ATP-dependent SWI/SNF complex, members of the CHD family of chromo-domain proteins, and more recently, members of the inhibitor of growth (ING) family of histone binding proteins.
The first member of the ING gene family was discovered through a subtractive hybridization assay between normal mammary epithelium and seven breast cancer cell lines (Garkavtsev et al., 1996a). Short cDNA sequences identified by this screen were termed genetic suppressor elements (GSE), and transfection of the “antisense” DNA sequence of these GSE into cells was found to promote cellular growth and transformation, whereas the “sense” DNA sequence inhibited growth and transformation. Sequence analysis of the gene encoding the GSE identified Inhibitor of Growth 1, or ING1. Searching of EST databases identified four other members of this gene family: ING2, ING3, ING4, and ING5, with members sharing between 32% and 76% DNA sequence homology. Comparison to genomes of other organisms revealed that the ING family is conserved from yeast to humans (He et al., 2005). Mice were shown to possess five Ing genes (Ing1–Ing5), similar to humans, whereas three ING orthologues were identified in yeast (Yng1, Yng2, and PHO23).
Human and mouse ING genes are dispersed throughout their respective genomes, as seen in Figure 1. Analysis of the genomic structure of the human ING genes revealed that most members undergo alternative splicing, with the exceptions of ING2 and ING5. The number of isoforms encoded by the ING1, ING3, and ING4 gene differs between the two species. Human ING1 was found to have five alternative splice variants, whereas mouse Ing1 encodes three variant-spliced proteins, p31Ing1a, p31Inglc, and p37Ing1b. Both human and mouse ING1 splice variants occur through alternative splicing of one of several upstream exons into a common last exon of the gene, producing a protein with a unique N-termini and a conserved C-termini. In contrast, although human ING4 encodes four splice variants, only one Ing4 transcript has been observed in mouse. The number of splice variants encoded by mouse Ing2, Ing3, and Ing5 genes is presently unknown. Several studies have examined the temporal and spatial pattern of human and mouse ING gene expression (Gunduz et al., 2002; Nouman et al., 2002b; Nagashima et al., 2003; Unoki et al., 2006; Walzak et al., 2007). All ING genes appear to be ubiquitiously expressed in fetal and adult tissues, though the relative abundance of the expression levels of the various ING genes differs between organs and developmental stages.
All ING proteins contain a plant homeodomain (PHD) at the C-terminus, a nuclear localization signal (NLS), and a unique domain with an unknown function called the novel conserved region (NCR) (Fig. 2). The PHD motif comprises approximately 60 amino acids and displays a C4HC3 signature that typically binds two Zn2+ ions (Aasland et al., 1995; Bienz, 2006). Currently it is unknown how many PHD proteins are present in man or mouse, however approximately 150 distinct PHD domain-bearing proteins have been predicted to occur in humans (Aasland et al., 1995). PHD domains closely resemble a canonical RING domain but lack the RING E2 ubiquitin ligase activity (Bienz, 2006). However, PHD domains have been implicated in chromatin remodeling, as they are often found in proteins that are known components of larger chromatin remodeling complexes, and may function by consolidating or strengthening a separate chromatin-binding activity of either the same protein or of a closely associated protein (Pascual et al., 2000; Ragvin et al., 2004; Bottomley et al., 2005; Bienz, 2006; Denslow and Wade, 2007). Recently, the PHD finger of the ING proteins has been found to bind directly to methylated histones, specifically H3K4me2 and H3K4me3 (Pena et al., 2006; Shi et al., 2006a), supporting a functional role for the PHD domain and further underscoring the link between this domain and nucleosome binding.
All ING proteins contain an NLS, and some ING proteins appear to have multiple NLS. To date, the role of the NLS has been studied extensively only for ING1 (Scott et al., 2001a; Ha et al., 2002), wherein NLS deletion results in cytoplasmic accumulation of the protein. Localization of ING proteins to the nucleus has been proposed to be critical to their function, as is evident by the observation of loss of nuclear ING1 staining in a number of cancers (Gong et al., 2005), and because deleting the entire NLS of ING4 resulted in a protein that could no longer bind p53 in co-transfection experiments (Zhang et al., 2005). Additionally, there are two copies of a putative nucleolar translocation signal (NTS) contained within the NLS of ING1, and translocation of ING1 to the nucleolus following exposure to UV light appears to be required for ING1-associated apoptosis (Scott et al., 2001a). It is currently unknown if other ING proteins have a similar NTS or can be detected in the nucleolar compartment.
Additional protein domains widely shared among the ING proteins include the leucine zipper-like (LZL) region present in ING2-5 (Soliman and Riabowol, 2007). The LZL consists of four to five conserved leucine or isoleucine residues spaced seven amino acids apart, which has the potential to form a hydrophobic face near the N-terminus of the protein. One report has indicated that the LZL motif of ING2 is required for nucleotide excision repair and induction of apoptosis (Wang et al., 2006b), though these LZL-related functions need to be substantiated by further experimentation.
Recent investigations have revealed that ING functions may be regulated at the level of ING gene expression. Analysis of mouse and human cells indicates that ING1 expression may be induced by RUNX3 but not by p53 (Zeremski et al., 1999; Cheung et al., 2000). In contrast, p53 was recently determined by chromatin immunoprecipitation (ChIP) and electrophoretic mobility shift (EMSA) assays to bind to two regions within the promoter of ING2 (Kumamoto et al., 2008). Interestingly, expression of ING2, but not ING1 (Nagashima et al., 2001) or ING3 (Nagashima et al., 2003), was also induced by the DNA damaging agents etoposide or neocarzinostatin, but not by other DNA damaging agents that also upregulate p53 activity, such as γ-irradiation or doxorubicin. Thus, the role of p53 in regulating ING2 expression needs to be analyzed further.
The activity of ING proteins may also be regulated through post-translational modifications. The p33ING1b isoform of ING1 was found to be phosphorylated on serine 126 by CDK1 under non-stress conditions or by CHK1 under DNA-damaging conditions (Garate et al., 2007). As CHK1 is normally activated by the ATM/ATR kinases, these findings suggest that p33ING1b might be a downstream target of ATM/ATR-CHK1 signaling following UV damage. In addition, phosphorylation of serine 126 was found to alter the half-life but not the sub-cellular localization of p33ING1b (Garate et al., 2007), suggesting that ser126 phosphorylation may alter p33ING1b activity by regulating its stability.
Finally. ING functions may also be modified by protein sub-cellular localization. Localization of p33ING1b appears to be regulated by 14-3-3 family members, which can tether ING1 to the cytoplasm. Binding between ING1 and 14-3-3 was dependent upon the phosphorylation status of ING1 serine residue 199 (Gong et al., 2006). Localization of ING2 to the plasma membrane was also reported to occur, and was due to interaction of the ING2 PHD domain with phosphoinositide (Gozani et al., 2003). However, clear binding preferences or consistently strong binding affinities between PHD containing proteins and specific phosphoinositides has not been observed by other groups (Bienz, 2006).
A review of the current literature suggests that ING proteins regulate a wide variety of cellular processes, including cell growth, apoptosis, DNA repair, senescence, and angiogenesis.
Several lines of evidence have indicated that ING proteins can influence cell cycle progression and are involved in cell cycle checkpoints (Campos et al., 2004a; Soliman and Riabowol, 2007). The majority of these studies have been conducted using ING1, but the fewer studies involving ING2-5 have identified similar roles for these ING proteins. In addition, suppression of ING proteins has been shown to increase cell spreading, increase cell migration, and relieve contact inhibition (Garkavtsev et al., 1996a; Kim et al., 2004; Kim, 2005; Unoki et al., 2006; Shen et al., 2007). Results from various transfection studies have suggested that most of the ING proteins are also required for proper p53 function (Campos et al., 2004a; Soliman and Riabowol, 2007), although more recent mouse modeling experiments indicate otherwise (see below). Furthermore, recent experiments have also suggested that there are p53-independent functions for the ING proteins, including regulation of the NF-κB (Garkavtsev et al., 2004) and hypoxia inducible factor (HIF) pathways (Ozer et al., 2005). Additionally, ING proteins have been found to serve as subunits in chromatin remodeling complexes (Doyon et al., 2006), indicating that ING proteins may act in the nucleus to regulate transcription (Cheung and Li, 2001; Feng et al., 2002; Berardi et al., 2004; Campos et al., 2004a; Shi and Gozani, 2005). The following is a description of the most compelling findings for each ING protein.
Overexpression of ING1 in human diploid fibroblasts resulted in a 50% increase in the number of cells found in the G0/G1 phase of the cell cycle (Garkavtsev et al., 1996a), indicating that ING1 might have a role in the G1–S transition. Conversely, antisense ING1 constructs in these cells resulted in abolition of this arrest and entry of the cells into S phase (Garkavtsev et al., 1996a). These findings were corroborated in H1299 cells by ectopically expressing p33ING1b, which resulted in a slight increase in the doubling time with fewer cells in G1, possibly due to a delay or block in S or G2/M (Tsang et al., 2003). Therefore, ING1 appears to be involved in both the G1/S and the G2/M cell cycle checkpoints. ING1 has been found to negatively regulate the expression of cyclin B1 (Takahashi et al., 2002), further supporting a possible involvement of ING1 in the G2/M cell cycle checkpoint. Additionally, ING1 expression was found to be inversely related to cyclin E expression (Ohgi et al., 2002). However, this latter finding needs to be corroborated and its significance determined with further investigation.
It has also been suggested that ING1 expression is cell cycle regulated: decreasing from G0 to G1, increasing in late G1 to become maximal during S phase, followed by a decrease in G2 (Garkavtsev and Riabowol, 1997). In addition, ectopic expression of p33ING1b can enhance the cell cycle arrest induced by doxorubicin, a topoisomerase II inhibitor that results in double-stranded DNA breaks (Tsang et al., 2003). Doxorubicin was found to decrease the population doubling time and increase the number of cells in G2/M. The enhancement of cell cycle arrest due to p33ING1b overexpression was DNA damaging agent-specific, with no enhanced cell cycle arrest observed with cisplatin or UV irradiation (Tsang et al., 2003). Antisense knockdown of ING1 was also found to promote the ability of anchorage independent growth in soft agar and increase foci formation, supporting a functional role for ING1 in preventing cell transformation (Garkavtsev et al., 1996a,b, 1998a; Takahashi et al., 2002).
Expression of ING1 increases prior to apoptosis induced by serum starvation, and ectopic ING1 expression was found to cooperate with c-Myc expression to induce apoptosis in both P19 cells and rodent fibroblasts (Helbing et al., 1997). Conversely, antisense knockdown of ING1 provided protection from apoptosis. Collectively, these results suggest that ING1 also has a role in regulating cell death. The ability of ING1 to induce apoptosis was found to depend upon an interaction with PCNA following UV irradiation (Scott et al., 2001b) and to be cell age-dependent, as only early passage fibroblasts were able to upregulate p33ING1b and undergo apoptosis following growth factor deprivation (Vieyra et al., 2002b). This study also suggests possible ING1 isoform-specific functions in apoptosis, since ectopic expression of p33ING1b, but not p47ING1a, sensitized young, but not old, fibroblasts to UV and hydrogen peroxide-induced apoptosis (Vieyra et al., 2002b). Ectopic expression of p33ING1b also sensitized cells to apoptosis induced by etoposide, taxol, and doxorubicin (Scott et al., 2001a; Takahashi et al., 2002; Vieyra et al., 2002b). However, a more recent paper appears to contradict the previous reports that overexpression of ING1 can promote apoptosis, as down-regulation of ING1 in p53-deficient glioblastoma cells sensitized these cells to cisplatin-induced apoptosis (Tallen et al., 2007). Interestingly, a very recent report found that overexpression of p33ING1b could synergize with tumor necrosis factor alpha (TNFα) treatment to induce apoptosis (Feng et al., 2006). This finding suggests an indirect role for p33ING1b in the NF-κB pathway, since NF-κB signaling protects cells from TNFα-induced apoptosis (Van Antwerp et al., 1996).
Regulation of nucleotide excision repair (NER) following UV exposure has also been linked to ING1. Overexpression of p33ING1b can enhance nucleotide excision repair (NER) of exogenously added plasmid DNA in a host-cell-reactivation assay (Cheung et al., 2001). An interaction between GADD45 and p33ING1b was also detected to support the proposed involvement of p33ING1b in NER. Mutations in the PHD domain of p33ING1b and a region found to interact with SAP30 (Sin3A Associated Protein 30), a component of Sin3A co-repressor complexes, abrogated enhancement by p33ING1b of NER in host cell reactivation assays and radioimmunoassay (Laherty et al., 1998; Campos et al., 2004b). Furthermore, the p33ING1b variant was not recruited to UV-induced DNA lesions, but enhanced NER in XPC-proficient cells possibly due to its ability to bind XPA (Kuo et al., 2007). As XPC/hHR23B acts at the first step of the NER pathway by recognizing helix-distorting DNA lesions and XPA acts to stabilize the resulting open DNA structure (Wijnhoven et al., 2007), these observations suggest that p33ING1b has an essential role in the early steps of the NER pathway, possibly by facilitating access of the nucleotide excision repair machinery to chromatin.
The ING proteins have been implicated in the regulation of the p53 pathway (Garkavtsev et al., 1998b; Nagashima et al., 2003; Shiseki et al., 2003; Pedeux et al., 2005). p53 is a tumor suppressor that responds to disruption of DNA integrity and/or perturbation of cell division in cells exposed to various forms of stress (Aylon and Oren, 2007). Following DNA damage, stress signals are transmitted to the p53 protein by a cascade of post-translational modifications that activate the p53 transcription factor and initiate or upregulate a wide variety of genes involved in cell cycle arrest, senescence, DNA repair, and apoptosis (Sherr, 1998; Sherr and Weber, 2000; Harris and Levine, 2005; Sherr, 2006; Aylon and Oren, 2007). All ING genes except ING3 have been reported to co-immunoprecipate with p53 following ectopic co-expression of ING proteins and p53, and subsequent functional studies utilizing forced overexpression of ING genes in cultured cells have indicated that ING-induced cell cycle arrest and apoptosis is compromised in p53-deficient, RKO-E6 cells, suggesting that ING proteins require functional p53 to inhibit cell growth (Berardi et al., 2004; Campos et al., 2004a; Soliman and Riabowol, 2007). Conversely, expression of antisense ING1 constructs inhibited expression of a p21WAF1-CAT construct in cultured cell that were wildtype for functional p53 (Garkavtsev et al., 1998a), suggesting that p53 requires the ING proteins to function as a transcriptional activator. Additional evidence linking ING proteins and p53 function was provided by experiments utilizing adenoviral-mediated gene transfer of p33ING1 and p53 into glioma cells. Increased apoptosis was observed in U251 and U-373MG glioma cell lines only when both genes were co-introduced (Shinoura et al., 1999), and similar results have been obtained following co-transfection of a human esophageal carcinoma cell line (Shimada et al., 2002).
Recent findings have indicated that ING1 may also regulate several components of the ARF-MDM2-p53 signaling axis. The p33ING1b protein has been proposed to compete with MDM2 for the same binding site on p53 as a means of increasing the stability and activity of p53 (Leung et al., 2002). In addition, ARF has been suggested to interact with p33ING1b in vivo to alter the sub-cellular localization of p33ING1b from the nucleus to the nucleolus (much like the effect of ARF on MDM2), and the ability of exogenous p33ING1b to induce cell cycle arrest was been found to be impaired in p19ARF-deficient mouse embryonic fibroblasts (Gonzalez et al., 2006).
There also appears to be p53-independent growth regulatory roles for the ING proteins (Takahashi et al., 2002; Tsang et al., 2003; Goeman et al., 2005; Feng et al., 2006; Wang and Li, 2006; Tallen et al., 2007). The results of microarray analysis of cells with p33ING1b knocked down by antisense constructs indicate that the cyclin B1 gene may be regulated by p33ING1b (Takahashi et al., 2002). As overexpression of p33ING1b in p53-deficient, Saos2 cells also decreased the level of cyclinB1 message, regulation of cyclinB1 expression by ING1 is likely p53-independent. Likewise, transcriptional regulation of HSP70 was also found to be regulated by p33ING1b in a p53-independent manner (Feng et al., 2006). Several p53-independent roles for p33ING1b in growth regulation and cell cycle arrest have been proposed based upon transfection studies in cell lines lacking functional p53. Overexpressing p33ING1b in H1299 lung carcinoma cells increased the doubling time of these p53-deficient cells by about 10% and enhanced a doxorubicin-induced G2/M DNA damage checkpoint (Tsang et al., 2003). Additionally, siRNA knockdown of ING1 in LN229 glioblastoma cells caused these p53-null cells to be more sensitive to cisplatin treatment and to transition faster through theG1 phase of the cell cycle (Tallen et al., 2007). These findings suggest that ING1 can function to inhibit cell growth or death independent of p53. In addition, the transcriptional silencing effect of ING1 has been proposed to be p53-independent, as ectopic expression of a Gal-p33ING1b fusion protein in p53-null, H1299 cells repressed the expression of a co-transfected reporter gene placed under transcriptional control of a modified thymidine kinase promoter (Goeman et al., 2005).
ING proteins have also been proposed to regulate NF-κB activity. NF-κB is a dimeric transcription factor composed of Rel protein family members which is sequestered in the cytoplasm by the IκB family of proteins (Moynagh, 2005a). NF-κB is activated by a number of stimuli, including components of microbial pathogens such as lipopolysaccharide (LPS), and by inflammatory cytokines (Neumann and Naumann, 2007). Activation of upstream receptors by these stimuli induces a signal transduction cascade leading degradation of the inhibitor IκB proteins and translocation of the NF-κB complex into the nucleus, where it upregulates the expression of genes involved in cellular survival and the immune response (Moynagh, 2005b). Since NF-κB regulates the expression of many genes proposed to govern apoptosis, angiogenesis, metastasis, proliferation, and tumor growth and survival (Perkins, 2004), it has been suggested that perturbation of NF-κB signaling is important in tumorigenesis. However, mutation of NF-κB itself is rarely observed in tumors. Rather, mutations in upstream NF-κB regulators have been found to result in constitutive activation of NF-κB (e.g., IκB loss of function in Hodgkin’s lymphoma) (Rayet and Gelinas, 1999). Recent microarray data has indicated that expression of the heat shock protein 70 (HSP70) is upregulated by p33ING1b expression (Feng et al., 2006). Since HSP70 can suppress the NF-κB pathway by preventing degradation of IκB (Malhotra and Wong, 2002; Ran et al., 2004; Shi et al., 2006b), these findings suggest a possible indirect role for p33ING1b in downregulating NF-κB signaling.
In addition to possible roles in p53 and NF-κB signaling, p33ING1b (but not p24ING1c) was found to associate with a chromatin remodeling complex (see Table 1). By fractionating HeLa nuclear extracts, endogenous p33ING1b protein was found as a subunit of an approximate 1–2 MDa complex (Skowyra et al., 2001). This complex also contained mSin3, SAP30, histone deacetylase 1 (HDAC1), RbAp48, and additional components of the mSin3A transcriptional co-repressor complex (Skowyra et al., 2001), and purified p33ING1b -containing complexes were found to deactylate core histones using in vitro assays. Although Sin3/HDAC is thought to function as a transcription repressor, recent evidence suggests that it might positively regulate transcription as well (Silverstein and Ekwall, 2005). Recently, p33ING1b and ING2 have been reported to recruit SIRT1 to the Sin3/HDAC complex (Binda et al., 2008). As SIRT1 negatively regulates the transcriptional repression activity of the Sin3/HDAC complex (Binda et al., 2008), this data suggests a possible role for ING proteins in transcriptional activation. Thus, p33ING1b and other ING proteins may function within the Sin3/HDAC complex to act as global regulators of gene transcription. In addition, the ING proteins might contribute to other functions ascribed to the Sin3/HDAC complex, including nucleosome remodeling, DNA methylation, N-acetylglucoseamine transferase activity, and histone methylation (Silverstein and Ekwall, 2005).
Different human ING1 isoforms have been found to associate with either HAT or HDAC activity. Overexpression of p33ING1b can induce hyperacetylation of histoneH3 andH4in vitro and in transfection assays, suggesting that ING proteins regulate HAT activity (Vieyra et al., 2002a). Conversely, overexpression of p47ING1a induces histone deacetylation, suggesting an association between ING proteins and HDAC activity (Vieyra et al., 2002a; Kataoka et al., 2003). These effects would obviously impact transcription by altering histone–DNA interactions in the promoter region of a gene or by altering the binding ability of other regulatory proteins.
Evidence has also been provided to indicate that ING1 is involved in DNA methylation. The DNA methyltransferase 1-associated protein 1 (DMAP1) has been found to physically associate with p33ING1b during S-phase at sites of pericentric heterochromatin and to correlate with methylated H3 lysine 9 (H3K9), histone deacetylation, and DNA methylation (Xin et al., 2004). Additionally, the PHD domain of all human ING proteins was found to preferentially bind di- and tri-methylated H3K4 and repress gene transcription (Pena et al., 2006; Shi et al., 2006a; Lan et al., 2007). Mutations in the PHD domain that disrupt the ability of ING1 to bind to methylated H3K4 prevent ING1 from inducing apoptosis following UV irradiation, supporting a role for ING genes in epigenetic regulation of gene expression (Pena et al., 2008). Finally, the p33ING1b protein has also been found via GST-pull-down assays to associate with ALIEN, a transcriptional co-repressor involved in gene silencing mediated by select members of nuclear hormone receptors and E2F1 (Fegers et al., 2007).
Ectopic expression of ING2 in RKO cells inhibited colony formation and induced aG1 cell cycle arrest (Nagashima et al., 2001), suggesting a role for ING2 in cell growth. Overexpression of ING2 was also found to induce replicative senescence whereas RNAi-mediated downregulation of ING2 delayed the onset of replicative senescence (Pedeux et al., 2005). However, a recent study contradicts this observation, as RNAi knockdown of ING2 was found to induce p53-independent senescence and overexpression of ING2 was found to induce a p53-dependent senescence in hTERT-immortalized human fibroblasts (Kumamoto et al., 2008). These studies suggest that ING2 has a role in senescence regulation; however, further work is needed to elucidate the mechanism of this regulation.
Consistent with a role for ING2 in cell proliferation, ING2 has recently been found to interact with SnoN and Smad2 to promote TGFβ dependent gene expression resulting in the inhibition of cell proliferation (Sarker et al., 2008). At present it is unknown if other ING family members are potential regulators of the TGFβ pathway. As TGFβ signaling is involved in cellular proliferation, cell survival, apoptosis, migration, invasion, and inflammation (Mourskaia et al., 2007), alteration of TGFβ signaling by ING proteins would be of much interest. ING2 may also have a role in etoposide-induced apoptosis (Wang et al., 2006a) and overexpression of ING2 enhanced apoptosis after irradiation of transfected cells with ultraviolet light (Wang et al., 2006b). The BCL-2 protein was upregulated following UVB exposure of MMRU melanoma cells overexpressing ING2, which also promoted the translocation of BAX to the mitochondria and subsequent alterations in the mitochondrial membrane potential. Additionally, ING2 was found to regulate Fas expression, thus suggesting a link between ING2 and both the mitochondrial:intrinsic and death receptor: extrinsic apoptotic pathways (Chin et al., 2005).
Truncation ING2 mutants lacking the leucine zipper-like (LZL) domain did not display elevated apoptosis following UV exposure (Wang et al., 2006b), suggesting that this domain is required for ING2-mediated apoptosis. RNAi-mediated knockdown of ING2 was also found to abrogate the NER capacity of melanoma cells (Wang et al., 2006a). Similar to p33ING1b, ING2 may be involved in the initial steps of NER by inducing chromatin relaxation and the recruitment of XPA to the photo-lesions in DNA. Interestingly, the NER ability of ING2 was found to require the LZL domain, even though this domain is not present in p33ING1b (Wang et al., 2006b), an ING that is also associated with NER.
Similar to p33ING1b, ING2 can also upregulate HSP70 expression, indicating that ING2 might also indirectly regulate NF-κB activity (Feng et al., 2006). ING2, like p33ING1b, was found to associate with the co-repressor ALIEN through an in vitro GST-pull-down assay (Fegers et al., 2007), to co-purify with components of the mSin3 complex, and to associate with Brg1-based SWI/SNF chromatin remodeling complexes (Table 1) (Doyon et al., 2006). Furthermore, ING2 and p33ING1b could also interact in vitro and in vivo with the RBP1 protein, a component of the mSin3A complex, and recruit SIRT1 to the mSin3A/HDAC1 complex. The recruitment of SIRT1 to this complex was found to negatively regulate the transcriptional repression activity of the mSin3A complex (Binda et al., 2008). Thus, ING2 shares many of the in vitro functional characteristics of ING1.
ING3 has been linked to regulation of the cell cycle and apoptosis. Similar to ING1, ectopic expression of ING3 in RKO cells decreased colony formation, possibly by decreasing the number of cells in S phase (Nagashima et al., 2003; Shiseki et al., 2003). ING3 overexpression also induced Fas expression, increased the cleavage of Bid, caspases-8, -9, and -3 and promoted apoptosis in UV treated cells (Wang and Li, 2006). Similar to ING2, these findings suggest that ING3 may be involved in the death-receptor/extrinsic pathway. However, unlike most other ING proteins, ING3 does not appear to play a role in NER or in replicative senescence, does not appear interact with p53, and is not involved in IR-induced cell death (Nagashima et al., 2003).
Biochemical purification of ING protein-containing complexes from HeLa cell nuclear extracts (Doyon et al., 2006) revealed that ING3 is found mainly, if not exclusively, as a subunit of the NuA4/Tip60 HAT complex (Table 1). Tip60 is an important transcriptional cofactor for p53, NF-κB, Myc, E2F1, and nuclear-receptor dependent transcriptional activation, and is involved in the cellular response to DNA damage, apoptosis, metastasis suppression, and maintenance of embryonic stem cell identity (Avvakumov and Cote, 2007; Thomas and Voss, 2007). Mice deficient for Tip60 die during embryogenesis prior to implantation (E4), and haplo-insufficiency for Tip60 results in accelerated lymphomagenesis in transgenic Eμ-Myc mice (Gorrini et al., 2007). These data suggest that ING3 might also function in these diverse processes through its association with the NuA4/Tip60 HAT complex.
Like ING3, ING4 does not appear to regulate NER. However, ectopic expression of ING4 in RKO cells was found to inhibit colony formation, likely due to a decrease in the percentage of cells in S phase (Nagashima et al., 2003; Shiseki et al., 2003), and ING4 overexpression correlated with increased Bax expression levels and upregulation of serum starvation-induced apoptosis (Zhang et al., 2004). Thus, similar to other ING proteins, ING4 has been proposed to regulate cell cycling and apoptosis. However, unlike other INGs, ING4 has been proposed to also alter angiogenesis and cell migration.
A relationship between ING4 and angiogenesis was uncovered by studies utilizing implantation of the glioblastoma cell line U87MG into the cranial windows of nude mice. Tumor cells in which ING4 levels were reduced by siRNA grew faster than control cells, and ING4-knockdown cells yielded tumors with higher vascular volume fractions (Garkavtsev et al., 2004). A reduction of ING4 expression in multiple myeloma cells also increased the expression of the pro-angiogenic molecules interleukin-8 (IL-8) and osteopontin (OPN). In addition to these in vitro observations, a correlation has also been noted between reduced expression of ING4 and increased mircovascular density in multiple myeloma patients (Colla et al., 2007).
ING4 has also been linked to cell migration, cell spreading, and contact inhibition. Ectopic expression of ING4 in cell lines both decreased cell spreading and cell migration (Unoki et al., 2006). Subsequent mass spectroscopy analysis of extracts from RKO cells overexpressing a FLAG-ING4 construct indicated that ING4 interacts with G3BP2a, COP1β, CaBP1, and several ribosomal proteins (Shen et al., 2007). ING4 was also found to co-localize in the lamellipodia of cells with Liprin α1, a protein involved in focal adhesion disassembly, which may account for the effects of ING4 upon cell spreading and migration (Serra-Pages et al., 1995). Interestingly, ING4 was also identified in a screen for genes that suppressed the loss of contact inhibition caused by MYCN overexpression (Kim et al., 2004). The ability of ING4 to suppress contact inhibition was further supported by reports indicating that ING4 expression attenuated the ability of T47D breast cancer cells to grow in soft agar (Kim et al., 2004).
ING4 has also been suggested to bind with and regulate p53, NF-κB, and HIF-1α activity. Studies utilizing forced overexpression of ING4 have indicated that ING4 can recruit p300 to the ING4-p53 complex to induce p53 acetylation on lysine-382 (Shiseki et al., 2003), although the precise effect of p53 lysine-382 acetylation on p53 activity is unclear. Forced overexpression and co-immunoprecipitation experiments performed in U87MG glioblastoma cells indicated that ING4 physically interacts with p65 (RelA), linking ING4 to the NF-κB pathway (Garkavtsev et al., 2004). Further analysis of this interaction using gel mobility shift assays indicates that ING4 can inhibit the DNA-binding activity of RelA. Additionally, knockdown of ING4 could stimulate expression of a NF-κB-dependent luciferase reporter plasmid in transfected cells. These results suggest that ING4 interacts directly with RelA to inhibit NF-κB transcriptional activity. Knockdown of ING4 by siRNA also led to elevation of the HIF-1α target genes NIP3 and AK3 under hypoxic conditions (Ozer et al., 2005). Nuclear levels of HIF-1α were unchanged in these experiments, suggesting that ING4 suppressed HIF-regulated gene expression by altering HIF-1α activity, although no direct interaction between ING4 and HIF-1α was observed. Subsequent experiments revealed that ING4 might regulate HIF-1α activity by affecting the recruitment of chromatin remodeling enzymes (Ozer et al., 2005). Additionally, ING4 has been proposed to suppress HIF-1α activity in myeloma cells by interacting with HIF prolyl hydroxylase-2 (PHD2/HPH2) (Colla et al., 2007), an oxygen sensing protein that helps to target the HIF-1α protein for degradation (Berra et al., 2006).
ING4 appears to be a component of a novel HBO1-HATcomplex (Table 1) (Doyon et al., 2006). The HBO1 protein is the catalytic subunit of two different but related HAT complexes that specifically acetylate histone H4, and HBO1 has been linked to DNA replication, S-phase progression, and transcriptional regulation (Avvakumov and Cote, 2007). In addition, HBO1 has been suggested to act as a co-repressor for the androgen receptor (Avvakumov and Cote, 2007) and NF-κB (Contzler et al., 2006). Furthermore, p53 has been found to interact with HBO1 and to down-regulate HBO1-acytyltransferase activity (Iizuka et al., 2007). While these are intriguing findings, the ability ofHBO1to function as a regulator of transcription is controversial, as the evidence is sparse and conflicting. Therefore, a role for ING4 in regulating HBO1 complex formation or contributing to HB01 function needs to be substantiated by further studies.
Biochemically, ING5 appears to be similar to ING4, as ING5 has also been found as a member of HBO1-containing HAT complex and can interact with p53 and recruit p300 to induce p53 acetylation on lysine-382 (Shiseki et al., 2003). Furthermore, ING5 appears to be a subunit of a histone H3-specific, HAT complex that included MOZ/MORF leukemic proteins (Table 1) (Doyon et al., 2006). Little is known about ING5 function, although very recent transfection experiments have indicated that ING5 reduces colony-forming efficiency, inhibits S-phase, and induces apoptosis in a p53-dependent manner (Shiseki et al., 2003). Additionally, ING5 can induce expression of the cyclin-dependent kinase inhibitor p21 (Shiseki et al., 2003), a p53-target gene. These results implicate ING5 in regulation of cell growth and p53 activity, but further studies of ING5 function are needed.
The ability of ING proteins to inhibit cell growth and their interactions with other proteins that are often dysregulated in human cancer have led to the classification of ING genes as “putative tumor suppressors.” Although analysis of ING genes in a variety of primary tumors and cell lines revealed that these genes are infrequently mutated, INGs often display reduced gene expression in tumors, and ING proteins are often mislocalized in cancerous cells. These findings categorize the ING genes as type II tumor suppressors (Sager, 1997). Additionally, some reports have noted a correlation between decreased ING expression levels and cancer progression or decreased survival rates for a number of cancer types (see below). Various mechanisms have been proposed to explain how ING gene function becomes altered in tumors, including mutations arising from gene rearrangements, loss of heterozygosity (LOH), promoter CpG hypermethylation, and protein mislocalization (Nouman et al., 2003b; Campos et al., 2004a; Gong et al., 2005). A summary of the published reports of ING gene mutations in cancers is compiled in Table 2.
Much more is known about ING1 and cancer than for the other ING family members. ING1 expression appeared to be decreased in several cancer types, and loss of heterozygosity (LOH) for ING1 has been occasionally seen in colorectal cancer, pancreatic cancer, head and neck squamous cell carcinoma (HNSCC), and breast cancer. Further evidence for a role for ING1 in tumor suppression was provided by reports showing that p33ING1b expression was greater in lower grade brain tumors than in high grade tumors, suggesting a role for ING1 in malignant progression of astrocytomas (Tallen et al., 2004). Another report on invasive carcinoma of the breast showed that reduced nuclear expression of p33ING1b correlated with larger, less differentiated and more malignant tumors that were estrogen receptor and progesterone receptor negative (Nouman et al., 2003a). Interestingly this study also found that most of these tumors had also lost functional p53. ING1 has also been proposed as a candidate tumor suppressor gene in mantle cell lymphoma (MCL), as expression of ING1 was decreased in three of five-tested MCL cell lines (Schraders et al., 2005; Ripperger et al., 2007). However, these findings have not been extended to primary tumor samples.
Studies of primary ovarian tumors using methylation-specific PCR found that the p33ING1b promoter was methylated and silenced in almost a quarter of all cases (Shen et al., 2005). This finding was confirmed using bisulfite sequencing, and methylation status statistically correlated with decreased mRNA expression. Additionally, treatment of ovarian cancer cell lines with 5′-aza-2′-deoxycytidine, a demethylating drug, caused a dosage dependent increase in ING1 expression. However, a different report utilizing cell lines and the COBRA method to detect promoter CpG methylation status found no evidence for ING1 promoter methylation in mantle cell lymphoma (Ripperger et al., 2007). Further analysis of primary tumors, ING expression levels, and promoter methylation is needed to determine the relevance of ING1 promoter inactivation in human cancer.
In contrast, no differences in ING expression were observed in recent studies of myeloid leukemia (Ito et al., 2002), or of melanoma (Stark et al., 2006), contradicting an earlier report of increased p33ING1b expression in this cancer (Campos et al., 2002). However, a link between ING1 mutation and melanoma was strengthened by a report identifying ING1 mutations in primary tumor samples. Two p33ING1b PHD domain mutations (R102L and N260S) were detected in 20% of the 46 tested melanomas (Stark et al., 2006), and either of these alterations proved as detrimental as deletion of the entire PHD domain to the enhancement of nucleotide excision repair mediated by p33ING1b in host-cell-reactivation assays and radioimmunoassays. Furthermore, those patients bearing an ING1 codon 102 or 260 mutation had a reduced 5-year survival rate (50% versus 82%) (Campos et al., 2004b). These findings highlight the importance of the PHD domain in ING1 function and tumor suppression, as loss of NER activity would likely facilitate tumorigenesis by increasing genomic instability. In agreement with this proposal, other reports have indicated the presence of ING1 mutations in the coding region for the PHD motif or the NLS in melanoma, HNSCC, esophageal squamous cell carcinoma (ESCC), breast cancer, pancreatic cancer, and in colon cancer (Zhu et al., 2005b; Pena et al., 2008).
Surprisingly, increased expression of ING1 was also observed during a study of glioblastomas. However, this overexpression resulted in the mislocalization (and presumed inactivation) of ING1 in these samples (Vieyra et al., 2003). Supporting evidence was obtained in a recent study comparing normal oral mucosa to oral squamous cell carcinoma (OSCC), finding that p33ING1b was either lost or mislocalized to the cytoplasm in the tumor cells (Zhang et al., 2008).
Chemotherapy is an important therapeutic modality for cancer, and identifying genes that participate in the response of cancer cells to these agents is critical to more effectively treat patients. Recent findings suggest that the ING genes might have a role in regulating the response of cancer cells to chemotherapeutic agents. Expression of p33ING1b was found to correlate with resistance to the antimitotic agent vincristine in brain tumor cells (Tallen et al., 2003). In contrast, camptothecin-induced cell death in melanoma cells was not dependent on p33ING1b expression (Cheung and Li, 2002). Additional studies have indicated that over-expression of p33ING1b in U2OS cells with wt p53 enhanced apoptosis induced by etoposide (Zhu et al., 2006). This enhancement of apoptosis appeared to be p53-dependedent since MG63 cells with mutant p53 did not undergo as much apoptosis following treatment. Similar results were observed using taxol in the same cell types (Garkavtsev and Boucher, 2005; Zhu et al., 2005a), suggesting that p33ING1b might be an important marker or therapeutic agent for the treatment of metastatic osteosarcoma. In contrast, down-regulation of ING1 in the p53-deficient glioblastoma cell line LN229 enhanced apoptosis following treatment with cisplatin, indicating that reduced ING1 expression may predict the sensitivity of cancer cells to chemotherapy independent of their p53 status (Tallen et al., 2007).
More recent work has begun to examine the links between ING proteins other than ING1 in tumorigenesis (see Table 2). Decreased ING2 expression (but not ING2 mutation) has been observed in lung cancer, melanoma, and in colon cancer (Shimada et al., 1998; Lu et al., 2006; Okano et al., 2006). In addition, ING2 may have a role in melanoma initiation, since reduction of nuclear ING2 has been reported in radial growth phase, vertical growth phase, and metastatic melanoma compared with dysplastic nevi (Lu et al., 2006). However, reduced expression of ING2 was not associated with tumor stage, subtype, or the 5-year survival rate of melanoma patients (Lu et al., 2006). In contrast, reduced ING2 expression was associated with tumor progression and shortened survival time in hepatocellular carcinoma (Zhang et al., 2007). These epidemiological studies suggest that ING2 loss or reduction may be important for tumor initiation and/or progression (Lu et al., 2006; Zhang et al., 2007).
Similar findings were reported for ING3 in cutaneous melanoma, with reduced ING3 expression observed in malignant melanoma compared with ING3 expression levels in dysplastic nevi (Wang et al., 2007). As with ING2, decreased nuclear ING3 was associated with poorer 5-year survival rates (Wang et al., 2007). Survival rates were 93% for strong nuclear ING3 staining and dropped to 44% for patients with negative to moderate nuclear staining. ING3 expression has also been associated with poor overall survival and tumor initiation in head and neck cancers (Gunduz et al., 2007), and a missense mutation in ING3 codon 20 has been observed in HNSCC (Gunduz et al., 2002).
ING4 also displays reduced expression levels in glioblastoma (Garkavtsev et al., 2004), HNSCC (Gunduz et al., 2005), and myeloma (Colla et al., 2007), and LOH of the ING4 locus has been observed in HNSCC (Gunduz et al., 2005) and breast cancer (Kim et al., 2004). ING4 has also been associated with tumor progression in glioblastomas (Garkavtsev et al., 2004) and myeloma (Colla et al., 2007). In both cases, decreased ING4 expression was associated with higher tumor grade and increased tumor angiogenesis. In myeloma it was also associated with increased expression of interleukin-8 and osteopontin (Colla et al., 2007). Expression of ING4 was reduced in malignant melanoma compared to dysplastic nevi, and was determined to be an independent factor for the poor prognosis of these patients (Li et al., 2008).
Currently the only report linking ING5 to cancer used 16 microsatellite markers to report LOH on the long arm of chromosome 2 in samples from patients with oral cancer (Cengiz et al., 2007). This report needs to be corroborated with further studies, especially since ING5 is only one of several genes found in this genomic region.
In summary, a number of epidemiologic studies have implicated alterations in ING expression in the development and progression of a variety of cancers, and ING proteins likely regulate the response of cancer cells to chemotherapy. However, LOH or specific point mutations within ING genes are only rarely observed, and ING mutation is not proposed as the molecular basis of a familial cancer syndrome. Thus, although ING genes have been demonstrated in vitro and in cell culture assays to impact a number of processes that regulate cell cycling, cell transformation, and apoptosis, obtaining definitive evidence that ING proteins function in vivo to suppress tumor formation has required the generation of ING mouse models
Recently, two groups have generated and analyzed Ing1-mutated mice to explore in vivo the role of this gene in cell growth and in tumorigenesis. Previous studies of mouse Ing1 revealed that this ubiquitously expressed gene utilized two promoter regions to produce three splice variants. These Ing1 spliced isoforms are translated to generate two distinct protein products (Zeremski et al., 1999). Ectopically expressing each isoform individually in transformed cell lines indicated that these two Ing1 proteins have opposing effects on p53 function. The long form (p37Ing1b) appeared to inhibit p53 function, whereas the shorter forms (p31Ing1a or p31Ing1c) has been proposed to facilitate p53 activity (Zeremski et al., 1999). However, as only the p37 protein co-immunoprecipates with p53 in forced overexpression studies, it is unclear how the shorter p31 protein might regulate p53. Kichina et al. (2006) used gene targeting in mouse ES cells to generate mice deficient for both Ing1 isoforms. Ing1-null mice were viable, approximately 20% smaller in size than their wildtype littermates, and were more prone to the effects of whole body irradiation. A subset of Ing1-null mice also developed cancer spontaneously as they aged, the most prevalent tumor type classified as follicular center B-cell lymphoma. This tumor specificity is in contrast to the spectrum of tumors that develop spontaneously in p53-deficient mice, where the most common tumors are T cell lymphomas, marginal zone B cell lymphomas, and sarcomas. Surprisingly, Ing1-deficient mouse embryonic fibroblasts (MEFs) displayed little or no changes in replicative lifespan or in cell cycling after treatment with taxol or other DNA damaging agents. Thus the link between Ing1, p53, and tumor suppression was unclear.
We have also recently employed ES cells and blastocyst injections to generate mice deficient for the predominant and longer p37Ing1b isoform of Ing1 (Coles et al., 2007). Similar to the Ing1-null mice, mice deleted specifically for p37Ing1b were viable but smaller in size, and more than half of all p37Ing1-null mice developed spontaneous follicular B cell lymphoma, suggesting that the p37Ing1b variant is responsible for B cell tumor suppression in vivo. Analysis of p37Ing1-null MEFs revealed that loss of p37Ing1increased the growth rate of MEFs, providing direct genetic evidence that endogenous Ing1 levels negatively regulate cell proliferation. However, loss of p37Ing1failed to alter other numerous cell growth characteristics normally governed by p53, including immortalization, oncogene-induced senescence, or cell growth arrest after DNA damage. In addition, upregulation of p53 activity during early embryogenesis by deletion of Mdm2 led to an embryonic lethal phenotype in mice, regardless of the presence or absence of p37Ing1. These data indicate that p37Ing1 is not required for a variety of p53 functions in cells or in mice. Furthermore, deletion of p37Ing1 also increased the proliferation of MEFs that were co-deleted for p53, demonstrating a p53-indepndent role for Ing1 in regulation of cell growth. We also examined the ability of DNA damaging agents to induce apoptosis in both MEFs and thymocytes. Although deletion of p37Ing1 did not alter levels of the pro-apoptotic, p53-responsive gene Puma, expression levels of Bax were significantly elevated in p37Ing1-null cells and correlated with increased apoptosis following DNA damage. As increased apoptosis was observed in p37Ing1-null MEFs and thymocytes regardless of p53 status, these data indicate that Ing1 has a p53-independent, pro-survival role in primary cells following DNA damage. We can conclude from these mouse studies that Ing1 is indeed a tumor suppressor that can regulate primary cell growth and apoptosis in response to DNA damage, although the data do not support a role for Ing1 in p53 functions.
Although much has been learned about the newly emerging family of ING tumor suppressors, there remain many questions regarding the functional role of ING proteins in development, in regulation of cell growth and death, and in tumorigenesis. At present, only mice deficient for Ing1 have been reported. As multiple functions for each ING have been indicated by transfection studies and by biochemical analysis, and as some of these functions appear to be shared by several ING proteins, it will likely be necessary to generate and study mice deficient for each of the Ing genes in order to dissect the unique and overlapping functions of the various ING family members. In addition, analysis of mice bearing several Ing gene mutations will also be needed to document compensation for loss of a single ING protein by other ING proteins. It is quite possible that the absence of a p53-dependent phenotype in the Ing1 mouse studies might be due to overlapping effects exerted by one or more of the other ING family members on p53.
In conclusion, ING genes appear to encode chromatin-modifying proteins that can regulate a variety of cellular processes, including tumorigenesis, and alterations in ING protein levels or localization can be found in a variety of human cancers. In addition, mice deficient for Ing1 have indicated that Ing1 suppresses the development of spontaneous follicular B cell lymphomas, although further epidemiological studies are needed to determine the significance of Ing1 loss in human lymphomagenesis. Furthermore, several in vivo mouse studies and transfection studies performed in cultured cells have indicated that there are p53-independent roles for ING proteins in the regulation of cell proliferation and apoptosis. Thus, other candidate signaling pathways involved in tumorigenesis that might be modulated by INGs need to be explored further. Although ING1 and ING4 have been proposed to regulate the NF-κB pathway, more work is needed to better dissect the mechanisms by which these ING proteins alter NF-κB activity and to determine if other ING proteins can also affect NF-κB signaling. Other important questions regarding the role of ING proteins in cell growth control involves the methods by which cells regulate these proteins under normal conditions and in cancer, as little is known about the effects of ING protein modifications on ING functions. Given that recent work has substantiated that the ING proteins function as bone fide tumor suppressors, further functional studies of the role of the ING genes in development and disease are clearly warranted.
Contract grant sponsor: National Institutes of Health;
Contract grant number: RO1CA77735.
We are grateful to Charlene Baron for assistance with the preparation of this manuscript. This work was supported by grant RO1CA77735 from the National Institutes of Health to SNJ.