ING1 Isoforms Differentially Affect Apoptosis in a Cell Age-dependent Manner¹

The ING1 candidate tumor suppressor was cloned by subtractive hybridization of human breast cancer cell line cDNAs with cDNA from normal human epithelial cells, followed by an in vivo selection assay (1). The ING1 locus maps to the terminus of chromosome 13, a site frequently associated with loss of heterozygosity in several types of cancers (2). Suppression of p33^ING1 expression promotes focus formation and growth in vitro and tumor formation in vivo, whereas ectopic overexpression of this protein blocks cell cycle progression by arresting transfected cells at G₁ of the cell cycle (1). Reduced levels of p33^ING1 have been found in breast, neuroblastoma, and glioma cancer cell lines (1) and in a proportion of primary tumors from breast (3), testis (4), esophagus (5), and lymphoid (6) tissues.

Recently, several ING1 splicing variants and ING1-like proteins have been reported (7). p33^ING1 was the first ING1 sequence cloned (1), followed by the isoforms ING1A and ING1B (8), encoding the p47^ING1a and p33^ING1b proteins, respectively. These findings led us to report that the p33^ING1 sequence initially cloned contained the majority of the p33^ING1b cDNA bound to a short portion of the p47^ING1a cDNA (8). This fusion resulted in a chimeric form of ING1 protein that was shown to be involved in senescence of HDFs³ (9) and to induce apoptosis in murine cancer cells (10). Here we refer to the naturally occurring human splicing variants as p47^ING1a and p33^ING1b, which we characterize regarding their apoptotic effects and functions in young and senescent normal HDFs.

Our recent observations (11, 12) have indicated that p33^ING1b is involved in apoptosis; however, its precise role remains uncharacterized. Furthermore, it is currently unknown whether other ING1 isoforms, such as p47^ING1a, are involved in apoptosis. Experimental evidence based on p33^ING1 suggests that up-regulation of endogenous p33^ING1b might sensitize cells to c-myc apoptosis (10) and both p53-independent (10) and p53-dependent (13) forms of apoptosis. Physical and functional interactions between p33^ING1 and the tumor suppressor and apoptotic regulator p53 have been reported (9, 14). Ectopic up-regulation of p33^ING1 has also been shown to increase the rate of transcription of reporter genes containing the p21^WAF1 promoter in a p53-sensitive fashion (14). p33^ING1 was found to interact with GADD45 and to be involved in repair of UV-mediated DNA damage (15), and we found recently that ectopic up-regulation of p33^ING1b sensitized human cells to UV-mediated apoptosis (11, 12). Paradoxically, up-regulation of endogenous p33^ING1 was also seen in senescent cells (9), which are resistant to apoptosis induced by growth factor deprivation (16). Senescent cells appeared to overexpress p33^ING1 under some conditions, whereas ectopic down-regulation of p33^ING1 with antisense constructs temporarily released these cells from this state of growth arrest (9). Finally, ING1 proteins display potential chromatin remodelling functions. We (17) and others (18, 19) have found that yeast ING1 homologues were functionally and physically linked to HATs. p33^ING1b has also been physically and functionally associated with HDACs (20), and it is worth mentioning that similar to ING1, p53 has been associated with chromatin remodeling functions (Refs. 21, 22; reviewed in Ref. 23). Together, these data seem to indicate that both the regulation of the intracellular levels of endogenous ING1 proteins and their chromatin-related functions may lead to differential biological consequences, depending on the physiological context, e.g., in young versus senescent cells.

Cell cycle regulators such as p21^WAF1 (24), p16^INK4 (25), cyclin D1 (26), and apoptotic regulators such as p53 (27) and Bcl-2 (16) are examples of proteins subject to mechanisms of age-dependent regulatory control. For example, both the DNA binding activity of the proapoptotic p53 (27, 28) and the endogenous level of the antiapoptotic Bcl-2 (16) are increased in senescent compared with young cells. Correlated with these observations, senescent but not young HDFs were found to be resistant to apoptosis induced by deprivation of growth factors (16). Because of these observations, senescence has been described not only as a nonproliferative state but also as an apoptosis-resistant state.

Because senescent and apoptotic cells were reported to up-regulate p33^ING1 (9, 10), p53 activity increases in senescent cells (27, 28) and p33^ING1 was reported to be necessary for p53 activity (13, 14), we addressed the questions of: (a) whether the endogenous levels of p33^ING1b and p47^ING1a were regulated during apoptosis in a differential manner in young and senescent HDFs; (b) whether the apoptotic effects of these different isoforms would be comparable and affected by those of p53; and (c) whether the apoptotic effects displayed by these isoforms would vary in a cell age-dependent manner. Finally, because ING1 proteins seem to be involved in chromatin-remodeling functions (17, 18, 19, 20), we asked whether ING1 proteins bound chromatin, and if so, whether this binding correlated with any apoptotic properties displayed by these proteins. By using primary HDFs, a model in which the intracellular pathways regulating apoptosis, survival, and senescence remain fully functional, we show here that the apoptotic functions of the human ING1 gene products are isoform, stimulus, and cell age dependent and are correlated with differential binding affinity to chromatin in young and senescent cells. We present evidence indicating that the expression levels of endogenous p33^ING1b are inversely correlated with its binding affinity to chromatin and directly correlated with its proapoptotic properties.

The human cell strains WI-38 (ATCC CCL-75, from embryonic lung) and Hs68 (ATCC CRL-1635, from newborn foreskin) were grown until just confluent and were induced to enter into apoptosis by serum deprivation exactly as reported (16). For UV and hydrogen peroxide (H₂O₂) experiments, we used logarithmically growing cells that were 60% confluent. UV experiments were performed as described (11) using a dose of 80 J/m². ING1 transfectants were UV treated 24 h after transfection and harvested for FACS assays 24 h after UV irradiation. For the H₂O₂ experiments, 36 h after being transfected, ING1 transfectants were incubated in regular medium containing 200 μm H₂O₂ (Sigma) for 2 h, cells were refed with fresh complete medium lacking H₂O₂ and were incubated 3 additional h before harvesting and processing for FACS assays. Cultures that were <35 MPDs for WI38 and <45 MPDs for Hs68 were considered young HDFs. Cells showing positive β-galactosidase activity and that were unable to undergo one MPD in a period of 3 weeks were considered senescent. Under the conditions used in these experiments, cells reached senescence at 56 MPDs for WI38 and 95 MPDs for Hs68.

Viability of cells was assessed weekly as reported by Wang (16) by means of the Trypan Blue (Life Technologies, Inc.) dye exclusion assay. To identify and quantitate apoptotic cells, we analyzed DNA fragmentation by TUNEL assay, visualized chromatin compaction by microscopy, and measured the proportion of cells with a sub-G₁ DNA content by laser flow cytometry (11). Annexin V kits (Roche) were used following the manufacturer’s instructions to identify early apoptotic cells, and ApopTag kits (Intergen) were used for TUNEL assays. TUNEL assays included FACS and indirect immunofluorescence microscopy studies. For microscopic visualization of cells processed for TUNEL, cells were grown on coverslips and weekly harvested as in the method of Wang (16). For microscopic visualization of chromatin compaction, cells on coverslips were fixed and stained with Hoechst 33258 dye (Sigma; Ref. 10). For analysis of DNA content by flow cytometry, cells were fixed in 70% ethanol/PBS, on ice for 1 h after, which they were subjected to analysis or were kept at −20°C for no more than 1 week. Before analysis using a Becton-Dickinson FACS scanner, ethanol was removed and cells were resuspended in PBS for 10 min, after which they were pelleted, the PBS was removed, and the cells were treated with staining solution [5 μg/ml of propidium iodide (Sigma), 1 mg/ml of RNase A (Roche) in PBS]. Analyses of flow cytometry data were done using ModFit software (Verity, Inc.). The values displayed in Fig. 4 A represent the total area of the sub-G₁ peaks modeled by this software. In transfection studies, the fraction of apoptotic cells was calculated as a percentage of total transfected cells by including parallel controls of cells transfected with GFP expression constructs.

Sample extracts were prepared as described (16), and equal amounts of proteins were run as assessed by Coomassie Blue dye staining of test gels. 15% SDS-polyacrylamide gels were run at 150 V, transferred to polyvinylidene difluoride membranes (NEN) for 2 h at 30 V and preblocked overnight at 4°C (Fig. 2) or 1 h at room temperature (Figs. 3,C and 5 B). The membranes were blotted with the primary antibodies dissolved in 0.1% Tween/PBS containing 5% nonfat milk (blocking solution). For ING1 blots, we used a 1:2 solution of CAbs (one to four; Ref. 29) in blocking solution. For Bcl-2 and p21^WAF1 blots, we used a dilution 1:1000 of Santa Cruz rabbit polyclonal antibodies (sc-7886 and sc-397, respectively). An ECL kit (Amersham) was used for chemiluminescence reactions.

RNA extraction, cDNA synthesis (Retro-Transcriptase reactions), and PCRs were performed as described previously (3). The sequences of primers used to study ING1B were 5′-CTCCATCGAGTCCCTGCCTT-3′ (forward primer) and 5′-GCCTTCTTCTTCTTGGGTGT-3′ (reverse primer). GAPDH primers were used as internal RNA loading and amplification controls (3, 30). The sequences of these GAPDH primers were 5′-CGGAGTCAACGGATTTGGTCGTAT-3′ (forward primer) and 5′-AGCCTTCTCCATGGTGGTGAAGAC-3′ (reverse primer).

[³H]Thymidine (ICN) labeling, formaldehyde fixation, preparation of samples, and chromatin immunoprecipitations were performed as in Hasan et al. (31). For ING1 immunoprecipitations, we used rabbit polyclonal anti-ING1 antibodies (29). Detection of ING1 proteins in these immunoprecipitates was performed by blotting with mouse monoclonal CAbs (29). For PCNA, cPKC, and AcH4 immunoprecipitations we used rabbit polyclonal antibodies (from Santa Cruz; sc-7907 and sc-208; and from Upstate, 06-866, respectively) and protein A-Sepharose beads (Amersham). For the comparison of chromatin immunoprecipitation studies between young and senescent cells (Fig. 6, A and B), the analysis of DNA was carried out on ING1 immunoprecipitates from non-[³H]thymidine-labeled cells using equal numbers of young and senescent Hs68 cells. DNA from immunoprecipitates was recovered by phenol:chloroform extraction, followed by ethanol precipitation (32). DNA samples were run in 1.5% agarose gels and stained with Sybergreen (Molecular Probes). For experiments in Fig. 6 C, equal amounts of [³H]thymidine-labeled cell extracts (young Hs68 cells) were used for each precipitation. Incubation of immunoprecipitates with DNase I (Sigma; 1 mg/ml) served as a control to verify that counts were precipitating with DNA.

Young HDFs were electroporated as described previously (11). The PubMed accession numbers of the ING1 isoforms used in these studies are AAF07920 for ING1A and AAF07921 for ING1B. These and the p21^WAF1 and p53 cDNAs were subcloned into pCI vector (Promega). Because in control experiments electroporation of senescent cells resulted in massive necrotic cell killing, for the comparison of ING1 apoptotic effects between young and senescent cells, we transfected with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Transfected cells were harvested 48 h after transfection for TUNEL, Annexin V, and Western blot assays, this latter to analyze the proper expression of ING1 cDNA constructs as shown in Figs. 3,C and 5 B. All experiments included a parallel control of cells transfected with GFP (Clontech) expression plasmids to determine the proportion of transfected cells.

Cells were prepared and microinjected as noted (10, 27). Microinjected cells were identified by detection of coinjected plasmids encoding GFP.

We evaluated the degree of apoptosis induced by deprivation of growth factors in the human cell strains WI-38 and Hs68 by quantitation of viable and apoptotic cells. Consistent with a previous study (16), our cell strains displayed an age-dependent sensitivity to apoptosis in response to growth factor deprivation (Fig. 1). Although our data differ quantitatively from those reported previously, senescent cells were clearly more resistant to this apoptotic stimulus than young cells as estimated by the trypan blue dye-exclusion viability assay (Fig. 1,A). This effect seemed to be cell type dependent because the percentage of viable cells for these strains of different tissue origins was found to be consistently different (Fig. 1,A). The reduced viability observed in young cells was caused by apoptosis, as assessed by TUNEL (Fig. 1, B–D).

Weekly samples of serum-starved cells processed for TUNEL and analyzed by FACS showed that this weekly decrease in the number of viable cells (Fig. 1,A) directly correlated with an increase in the staining pattern of apoptotic DNA breaks as assessed by our combined FACS/TUNEL analysis (Fig. 1,B). To ask whether necrotic effects contributed to the data of Fig. 1,B, cells were grown on coverslips and were harvested weekly before being processed for TUNEL and visualized by indirect immunofluorescence microscopy. As shown in Fig. 1, C and D upon extended serum withdrawal, only young cells displayed positive TUNEL staining, consistent with serum withdrawal inducing apoptosis and not necrotic cell death, in agreement with Wang (16). These observations were consistent with a down-regulation of the antiapoptotic protein Bcl-2 along the period of serum withdrawal only in young cells (Fig. 2,A). Conversely, the observed resistance to apoptosis displayed by senescent cells correlated with a higher and stable level of endogenous Bcl-2 in these cells (Fig. 2 A).

Because young WI-38 cells were the most sensitive of our models to apoptosis induced by deprivation of growth factors, we made use of these cells to address whether the expression of ING1 proteins varied in parallel to the induction of apoptosis, a phenomenon that we observed previously in murine cancer models (10). As shown in Fig. 2,A, we found that upon serum deprivation, p33^ING1b up-regulation correlated with an increase in the percentage of apoptotic cells and with a down-regulation of the antiapoptotic protein Bcl-2. The expression of p33^ING1b also correlated with the levels of its RNA as assessed by RT-PCR studies (Fig. 2,B), indicating that ING1B expression was regulated primarily at the transcriptional level. We were unable to detect p47^ING1a in our lysates, probably because of its low endogenous level in these cells. Fig. 2 A also confirms the senescence-related increase in p21^WAF1 levels (24).

We next asked whether the ING1B induction observed in young WI-38 cells was a process dependent on physiological context. Specifically, could senescent cells up-regulate p33^INGb in the same way as young cells after extended serum deprivation? In contrast to young cells, p33^ING1b levels did not increase in serum-starved senescent fibroblasts but seemed to decrease upon serum withdrawal (Fig. 2,C). Furthermore, we did not observe the previously reported (9) senescence-associated increase in the endogenous level of p33^ING1b in our two models (Fig. 2, A and B). In fact, low MPD HDFs expressed higher levels of p33^ING1b and ING1B RNA compared with senescent cells as assessed by Western blot (first two lanes in Fig. 2, A and C) and RT-PCR studies (Fig. 2,B). Furthermore, the expression pattern of p33^ING1b was regulated in the same age-dependent manner in our two HDF models (Fig. 2,C). Finally, this effect did not seem to depend on cell density because we were unable to find differences in the expression level of p33^ING1b between quiescent and logarithmically growing cells in the presence of serum (Fig. 2 C, Lanes 4 and 5). Together, these data led us to test whether changes in the levels of ING1 proteins (in particular p33^ING1b) could sensitize cells to, and/or trigger, apoptosis in HDFs. To address this possibility, we evaluated the effects of ectopically up-regulated ING1 proteins on the rate of apoptosis of young and senescent HDFs by transient transfection.

Because young Hs68 cells were more resistant to the induction of apoptosis by serum deprivation than WI-38s, we used HS68s as a more stringent test to evaluate whether ectopic up-regulation of different ING1 isoforms was able to trigger apoptotic cell death. As shown in Fig. 3, ING1 was able to induce apoptosis in an isoform-specific manner in these cells. Ectopic up-regulation of p33^ING1b, but not p47^ING1a, resulted in increased content of the sub-G₁ component and of DNA breaks as assessed by our FACS/TUNEL double assay (Fig. 3,A). As shown in Fig. 3,B, the effect of the p47^ING1a expression construct was similar to that of the vector, although high levels of p47^ING1a were expressed as determined by Western blotting in parallel transfections. To ask whether necrotic effects contributed to our FACS/TUNEL data, we microinjected HDFs with ING1B expression constructs and stained these cells for microscopic visualization of apoptosis-related chromatin compaction. As shown in Fig. 3, C and D, cells showing increased positive immunostaining for p33^ING1b displayed chromatin compaction consistent with p33^ING1b inducing apoptosis and not necrotic cell death.

Given the lack of apoptotic effects displayed by p47^ING1a, we asked whether p47^ING1a could act as a sensitizer rather than as a trigger of apoptosis in these cells. To test this hypothesis, we transiently transfected young Hs68 cells with expression constructs encoding ING1A or ING1B isoforms, after which we subjected these transfectants to stimuli reported previously to induce apoptosis in fibroblasts, such as UV irradiation (11, 33) and hydrogen peroxide (34). As shown in Fig. 4 A, we were unsuccessful in visualizing any apoptosis-sensitizing role for p47^ING1a in HDFs subjected to either UV irradiation or oxidative stress. Conversely, ectopic up-regulation of p33^ING1b significantly sensitized cells to these apoptotic effects, confirming the proapoptotic role of p33^ING1b as well as an isoform-dependent function for ING1 in apoptosis of young HDFs.

Because UV irradiation and hydrogen peroxide treatments are two kinds of cellular stress that increase the level of the tumor suppressor protein p53 (33, 34) and p53 was reported to be required for p33^ING1b to induce apoptosis in human glioblastoma cells (13), we asked whether cotransfection of TP53 and ING1B would increase the rate of apoptosis yielded by each of these genes alone in young Hs68 cells. As shown in Fig. 4,B, when ING1B was cotransfected with either TP53 or the p53-inducible gene WAF1, we observed an ∼3-fold increase in the percentage of apoptotic cells compared with ING1B transfectants (Fig. 4 B), confirming that functional interactions between Tp53 and p33^ING1b, and p21^WAF1 and p33^ING1b exist in our normal human cell model.

Because the RNA and protein levels of the proapoptotic isoform p33^ING1b found in senescent HDFs appeared lower than those observed for syngeneic young cells (Fig. 2) and these levels correlated with the resistance of senescent cells to enter into apoptosis (Fig. 1), we asked whether ectopic up-regulation of ING1 isoforms could induce and/or sensitize senescent cells to enter apoptosis. Different from our results from young cells (Figs. 3 and 4), we were unsuccessful in detecting a proapoptotic effect either for ING1A or ING1B in our two senescence models. As shown in Fig. 5,A, ectopic up-regulation of either p33^ING1b or p47^ING1a resulted in neither increased percentage of sub-G₁ content nor increased DNA breaks, as assessed by our FACS/TUNEL double analyses. Because these two assays are powerful tools for the analysis of late apoptotic markers such as DNA breaks and chromatin compaction (reviewed in Ref. 35) but not for early apoptotic events such as exposure of specific plasma membrane lipids (i.e., phosphatidylcholine), we tested whether these senescent ING1 transfectants displayed membrane flipping by means of the Annexin V assay (reviewed in Ref. 36). As shown in Fig. 5 B, we were also unable to detect apoptosis by this method in our two senescence models, corroborating our previous FACS/TUNEL data.

Our group and others have reported that yeast ING1 homologue proteins interact with HATs (17, 18, 19), and recently, p33^ING1b was found to bind to a HDAC (20). Because HATs and HDACs are chromatin-interacting proteins (23), we asked whether p33^ING1b bound to chromatin, and if so, whether this binding correlated with the cell age-dependent apoptotic properties displayed by p33^ING1b. As shown in Fig. 6,A, when analyzing similar numbers of young and senescent cells, the amount of chromatin immunoprecipitated by our anti-ING1 antibodies was appreciably higher in senescent cell extracts compared with young ones. This is particularly striking because the amount of endogenous p33^ING1b is considerably lower in senescent cells compared with young cells (Figs. 2 and 6,A, on the right). Such clear differential binding was not seen using anti-PKC immunoprecipitates or protein A-Sepharose bead controls. Immunoprecipitates using an antibody against a highly acetylated form of histone H4 (AcH4) showed the opposite trend, precipitating more chromatin from young cells. At similar amounts of p33^ING1b for young and senescent cells, a condition that requires approximately three times the number of senescent cells than young cells, the difference in the amount of anti-ING1-immunoprecipitated chromatin from young and senescent cell extracts was even greater (Fig. 6,B). In this case, the nonspecific binding of chromatin from senescent cell extracts now appears slightly higher using the anti-PKC and bead controls as expected. Signal was attributable to chromatin in the immunoprecipitated material, as shown by the addition of DNase to chromatin immunoprecipitation extracts from young [³H]thymidine-labeled cells. In these experiments, anti-AcH4 and anti-PCNA were used as positive chromatin immunoprecipitation controls, whereas the anti-cPKC antibodies were used as a negative control. Together, these data indicate that both the number of p33^ING1b complexes bound to chromatin (Fig. 6,A) and the binding affinity of this protein to chromatin (Fig. 6 B) are higher in senescent cells compared with young cells.

In this study, we observed an age-dependent sensitivity to the induction of apoptosis in two different strains of HDFs (Fig. 1). In these models, p33^ING1b followed an age-dependent pattern of expression that correlated with the induction of apoptosis (Fig. 2). This led us to ask whether ectopic up-regulation of p47^ING1a and/or p33^ING1b would induce and/or sensitize HDFs to apoptosis, and if so, if such an effect would depend on the physiological context, specifically, the cell passage level or in vitro cell age. We found that p33^ING1b, but not p47^ING1a, was able to trigger and sensitize HDFs to enter into apoptosis (Figs. 3 and 4,A). This effect was sensitive to that of p53 (Fig. 4,B) and age-dependent because senescent but not young cells were resistant to apoptosis induced by ectopic up-regulation of p33^ING1b (Fig. 5). Finally, we found that p33^ING1b displayed an age-dependent pattern of both expression (Fig. 2) and binding to chromatin (Fig. 6), which correlated with the proapoptotic properties of this protein (Figs. 3,4,5).

Consistent with previous observations in murine tumor models (10), we found that endogenous p33^ING1b is dramatically up-regulated upon apoptosis mediated by serum starvation (Fig. 2, A–C). Although our anti-ING1 monoclonal antibodies (29) were able to detect p47^ING1a in protein extracts from different human tissues (29) and extracts of HDFs transfected with ING1A expression constructs (Figs. 3,C and 5,B), we were unable to detect endogenous p47^ING1a in our Western blot assays (Fig. 2, A and C, and empty vector lanes in Figs. 3,C and 5,B), possibly because of the low level of this protein in our models. In contrast to what was reported previously for p33^ING1 (9), we found that the endogenous level of p33^ING1b was actually higher in young cells than in senescent cells in two different strains of fibroblasts (Fig. 2, A–C; Fig. 6, A and B). This difference in the endogenous level of p33^ING1b did not seem to depend on the density of the cells because both quiescent and logarithmically growing cells displayed similar levels of p33^ING1b (Fig. 2 C). We believe that the difference between previous and present data might be attributable to the previous lack of information regarding multiple splicing variants (1, 7, 8) combined with the lack of highly specific monoclonal antibodies (29) in the pioneering ING1 studies.

Ectopic manipulation of the protein levels of ING1 isoforms allowed us to define p33^ING1b but not p47^ING1a as a p53-sensitive, proapoptotic human ING1 isoform (Figs. 3 and 4). Although significant with respect to our controls, the magnitude of the proapoptotic effect of p33^ING1b was highly variable in our young HDF models (Fig. 3,B). On the basis of our results with senescent HDFs (Fig. 5), it is possible that variations in the passage levels of our cultures of fibroblasts may account for the dispersion of data we observed in young ING1B transfectants. Also, contrary to a previous report (13), we found that ectopic up-regulation of p33^ING1b was able to induce apoptosis in the absence of ectopic up-regulation of p53 (Fig. 3). This difference might be attributable to our use of a normal human model, in which the pathways and functions of p53 are fully functional, as opposed to the glioblastoma model used previously in which p53 is mutated (13). Alternatively, p33^ING1b may act as a sensitizer rather than as inducer of apoptosis in our models, given the cellular stress that the electroporation process might have caused. It is also possible that a functional interaction between p33^ING1b and p53 might explain the age-dependent apoptotic effect displayed by p33^ING1b in our HDF models as well as in the age-dependent onset of tumors in which ING1 and/or p53 functions are compromised. For example, similar to the case of ING1, an age-dependent activation of the p53 protein occurs (27, 28), which may be attributable to binding of p33^ING1b (14) or to acetylation by p33^ING2 (37). If the latter is the case, then perhaps overexpression of p33^ING1b stabilizes p33^ING2 in trans, an idea that is currently being explored. Finally, both the increased presence of p33^ING1b in complexes from senescent cells containing chromatin in which histone H4 acetylation is reduced and the decreased binding affinity of p33^ING1b to acetylated chromatin in young cells (Fig. 6) correlate with the age-dependent loss of sensitivity to apoptosis. This correlation is consistent with a transcriptional regulatory role for this protein.

On the basis of these and other data, we propose that one of the mechanisms by which ING1 proteins regulate apoptosis would involve chromatin-remodeling functions that may be related to their binding of chromatin. These proteins might regulate apoptosis by compacting (20) or relaxing (17, 18, 19) chromatin either to control the expression of proapoptotic and survival genes (14, 15, 36) and/or to alter the sensitivity of chromatin (17, 18, 19, 20) to nuclease degradation during the final phase of apoptosis. These ING1-mediated functions might be subject to age-dependent mechanisms of control directed to prevent induction of apoptosis in senescent but not in young cells; these mechanisms of control might be exerted at both transcriptional and posttranslational levels. For example, at the transcriptional level, the up-regulation or alternative splicing of one but not another ING1 isoform could be controlled (Fig. 2). At a posttranslational level, the chromatin-related functions of ING1 proteins (Fig. 6) could be controlled by their physical interactions with proteins such as PCNA (11), p53 (14), GADD45 (15), HATs (17, 18, 19), HDACs (20), and other chromatin-related proteins (i.e., histones), the biochemical properties of which may change during cell aging. Whether the chromatin binding (Fig. 6) and potential chromatin-remodeling properties (17, 18, 19, 20) of ING1 proteins are directly and specifically related to apoptosis regulation (Figs. 1,2,3,4,5) is currently being addressed by studying the apoptotic effects of ING1 mutant proteins whose chromatin binding and/or remodeling properties are impaired.

We thank Drs. C. Helbing, D. Bazett-Jones, E. Parr, M. Meyyappan, P. Santamaria, P. Forsyth, and A. Burette, as well as C. Andreu-Vieyra for insightful discussions; D. Ma for the ING1A and ING1B cDNA expression constructs; Dr. P. Bonnefin for the p53 and p21 cDNA expression constructs; V. Berezowski for the anti-ING1 CAbs and polyclonal antibodies; L. Robertson for assistance during FACS analysis; S. Pastyryeva for providing presenescent Hs68 cells; and H. Muzik for assistance during ³H labeling.