GW112, A Novel Antiapoptotic Protein That Promotes Tumor Growth Free

Apoptosis, or programmed cell death, is a fundamental process in the development and homeostatic maintenance of biological systems (1, 2). It is a tightly regulated process that is normally initiated to eliminate unwanted cells in multicellular organisms (3). Aberrations in the regulation of this process are involved in various pathological conditions such as cancer (4, 5, 6, 7, 8), neurodegenerative diseases (9, 10), autoimmune disorders (11, 12, 13), and viral infections (14, 15). Apoptosis played critical roles in carcinogenesis because it is a major hurdle that a normal mammalian cell would have to pass before it becomes carcinogenic (16). Genetic defects in the control of apoptosis allow for the expansion of neoplastic cells. It also enables them to escape immunosurveillance and treatment (chemo- or radiation therapy). Therefore, apoptosis control in cancer cells is of critical biological and clinical importance (17, 18, 19, 20).

Tumor cells encounter a wide variety of adverse growth conditions such as hypoxia, low pH, and glucose deprivation (21, 22, 23). A normal cell experiencing these stressful conditions will invoke responses such as growth arrest and activation of apoptosis. However, tumor cells can grow despite being under stress. This ability is acquired through genetic/epigenetic alterations in various genes that control several mechanisms, which include apoptosis as well as cell cycle progression. Understanding these alterations may allow us to gain insights into the process of tumor growth that may in turn lead to new targets for therapeutics development.

In our efforts to identify novel differentially expressed genes in cancer cells through the use of National Cancer Institute Cancer Genome Anatomy Project system, we came across the GW112 (Unigene ID#Hs.273321) gene by virtue of its high expression in human colon tumors. GW112 was initially cloned from human myeloblasts. However, little is known about its function or structure. It bears little homology to other known genes. There is thus far only a single published report that identified it to be a dominant mRNA species in inflamed colonic epithelium (24). Its expression was confined to the crypt epithelium. Preferential expression of GW112 in the crypt epithelium of inflamed colonic mucosa in ulcerative colitis suggests an important physiological role of GW112, although its exact function remains unknown.

We conducted a series of experiments to decipher the function of GW112. We show that GW112 is expressed in multiple human normal and malignant tissues with the highest expression in organs/tumors of the digestive system. We also identified, through the yeast two-hybrid system, that GW112 protein interacts with a known protein GRIM-19, which has been implicated in IFN/retinoic acid-induced apoptosis (25, 26, 27, 28, 29). We report that overexpression of the GW112 gene can attenuate hydrogen peroxide and IFN-β/retinoic acid-induced cellular apoptosis and apoptotic gene expression. The overexpression of GW112 also facilitated clonogenic tumor cellular survival and rapid tumor growth. Therefore, our results suggest that GW112 is an important regulator of cellular apoptosis that may be involved in tissue homeostasis and tumor development.

We have used the following cells in our studies, HEK 293, an immortalized human embryonic kidney cell line; SVEC4–10, a mouse lymphoid vein endothelial cell transformed by SV40; HeLa, a human adenocarcinoma cell line; and TRAMP-C1 cell line, a mouse prostate tumor cell line. All of the cell lines, with the exception of the last one, were obtained from American Type Culture Collection. The TRAMP-C1 cell line was obtained from Dr. Norman Greenberg of Baylor College of Medicine. The cells were cultured in DMEM (Invitrogen Inc., Carlsbad, CA) with 10% fetal bovine serum, 100 unit/ml penicillin, and 100 μg/ml streptomycin at 37°C, 5% CO_2.

For Northern blot analysis, we used commercially available eight-lane multiple tumor tissue Northern blots (MTN blots) from the Clontech Corporation (Palo Alto, CA). It contains 10 μg of total RNA samples per lane. The Northern blot analyses of GW112 expression was performed according to the manufacturer’s instruction by use of ³²P-labeled cDNA probe.

For yeast two-hybrid analysis, we used the Matchmaker Two-Hybrid System 3 and pretransformed Matchmaker kidney library (no library from tissues of the digestive system, which are more preferable, was available at the time of the experiment), both from the Clontech Corporation. The matchmaker kidney library was constructed by use of mRNA extracted from normal, whole kidney. It was built into the pACT2 plasmid and was pretransformed into yeast strain Y187, which provides for expression of library proteins fused to the 3′ terminus of GAL4 transcriptional activation domain (AD). The reverse transcription-PCR derived human GW112 cDNA (GenBank accession no. NM-006418) was cloned into the pGBKT7 vector by fusing the whole coding sequence to the 3′ terminus of the GAL4 DNA-binding domain. pGBKT7-GW112 was then transformed into the yeast strain AH109. The AH109 and Y187 yeast strains were of opposite mating types incorporating four reporter constructs, in which various transcriptional reporter genes (HIS3, ADE2, MEL1, and lacZ) were controlled by three heterologous GAL4-responsive upstream activating sequences with their corresponding promoter elements. The AH109 cells that contained the GW112-binding domain fusion protein were then mated with the Y187 cells with the cDNA-AD fusion library. The mating was performed essentially as described in the product manual provided by Clontech. The activation of transcriptional reporters was then assessed by plating the mating mixture on medium- and high-stringency synthetic dropout medium. Medium-stringency medium, lacking histidine, required activation of the HIS3 reporter to permit colony growth. The high-stringency medium, on the other hand, required activation of both HIS3 and ADE2 reporters for growth to occur. All of the plates also contained 20 μg/ml X-α-gal (Clontech Inc.), a chromogenic substrate that makes the yeast colonies blue when they had their MEL1 reporter gene activated. The positive colonies (those that appeared blue) were retested for phenotypes. The plasmids from positive colonies were then isolated after transformation into Escheria coli, and the interaction was confirmed in yeast by mating. Plasmids from different positive colonies were sequenced, and the in-frame cDNAs were selected for mammalian two-hybrid assay and in vivo coimmunoprecipitation assay.

For mammalian two-hybrid analysis, we used the Mammalian Matchmaker Two-Hybrid assay kit (BD Clontech). The human GW112 cDNA (NM-006418) was cloned into the pm vector (to derive pBD-GW112) to generate a fusion protein between the GW112 protein and the yeast GAL4 DNA binding domain. The human GRIM-19 cDNA was cloned into the pVP16 vector (to derive pAD-GRIM-19) to generate a fusion protein between GRIM-19 and the transcription AD of VP16 (VP16-AD). If there is potential interaction between GW112 and GRIM-19, the complex formed between the two fusion proteins will result in a full-fledged transcriptional factor that can activate any reporter genes that possess GAL4 binding domains. The reporter adopted in the Clontech kit was a CAT gene (G5CAT) that contains five consensus GAL4 binding site and a minimal promoter of adenovirus E1b gene. To detect potential interaction between GW112 and GRIM-19, the three plasmids, pBD-GW112, pAD-GRIM-19, and G5CAT, were cotransfected into 60–70% confluent 293 cells using the LipofectAMINE (Invitrogen Inc., Carlsbad, CA). Forty-eight to 72 h later, the cells were harvested by use of a cell lysis buffer provided in a CAT ELISA kit purchased from Roche Molecular Biology (Mannheim, Germany). The cells lysate was then centrifuged, and the supernatant was used for CAT activities assay according to manufacturer’s instruction.

For in vivo coimmunoprecipitation, the human GRIM-19 cDNA was cloned into pCMV-Tag-2B Epitope Tagging mammalian Expression Vector (Stratagene, La Jolla, CA). A FLAG epitope tag was located at the NH₂ terminal of the protein. The human GW112 cDNA (NM-006418) was engineered (through PCR) to have a hemagglutin (HA) epitope in the 3′ end and was subsequently cloned into pCMV-Tag-2B without fusion to the Flag epitope. The two plasmids were cotransfected into 293 cells at 70–80% confluency. The cells were detached and collected in 500 μl of immunoprecipitation buffer [20 mm sodium phosphate, (pH 7.5), 500 mm NaCI, 0.1% SDS, 1% NP40, 0.5% sodium deoxycholate, and 0.02% sodium azide] 48 h later. After the cells were lysed, the supernatant (400 μl) was incubated with 8 μl anti-HA antibody (1:50 dilution) overnight at 4°C. Afterward, ∼100 μl of Ultralink Immobilized protein G (Pierce, Rockford, IL) was added and gently mixed for 2 h at room temperature. The beads were washed six to eight times with 1 ml immunoprecipitation buffer and boiled in 50 μl of the SDS/sample buffer for 5–8 min. After a brief precipitation (spin down), the supernatant (10 μl) was run in a 14% SDS-PAGE gel for Western blot analysis by use of an antiflag antibody (1:1000; Stratagene Inc.).

The AdEasy system of adenovirus packaging, including plasmid pAdtrack-CMV, pAdeasy-1, and the packaging E. coli BJ5183 cells, was kindly provided by Dr. Tong-Chuan He (The Johns Hopkins University School of Medicine, Baltimore, MD; Ref. 30). The human GW112 gene was amplified by reverse transcription-PCR from total RNA of human colon tumor tissues. The 9 amino acid HA sequence was added in frame at the 3′ terminus of GW112 to facilitate Western blot analysis. It was then subcloned into the pAdtrack-CMV vector at the EcoR I/SalI sites to produce the pAdtrack-CMV/hGW112 plasmid. Packaging and production of the adenovirus to carry the human GW112 gene was achieved by use of the AdEasy system according to published protocols (30). This system allows the quick production of E1-deleted recombinant adenovirus vector. Briefly, the pAdtrack-CMV/hGW112 plasmid was linearized by PmeI and then recombined with pAdeasy-1 plasmid in recA+ bacteria BJ5183. The resultant pAdeasy containing the human GW112 gene was then transfected into low passage (<35) 293 cells after have been linearized by PacI to obtain the recombinant adenovirus vector, AdGW112. Successful adenovirus particles usually appear at days 7–12. Large-scale preparation of the virus particles was carried out following established protocols.

The in situ cell death detection kit was purchased from Roche Molecular Biology (Mannheim, Germany) and performed according to the user manual. Briefly, SVEC cells are incubated in tissue culture chambers and fixed with a freshly prepared 4% paraformaldehyde solution for 1 h at room temperature after treatments. The slides were rinsed with PBS, and endogenous peroxidase was blocked by use of 3% H₂O₂ in methanol for 10 min. After washing with PBS, the slides are permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate for 2 min on ice, then incubated with terminal deoxynucleotidyl transferase-mediated nick end labeling reaction mixture for 60 min at 37°C. After washing, converter-POD was added onto the slides and the slides were incubated for 30 min at 37°C in a humidified chamber. Around 50–100 μl 3,3′-diaminobenzidine-substrate solution was dropped to the top of the slides and analyzed under a light microscope after incubating for 10 min.

HeLa cells were cultured in 12-well plates to 50–60% confluence, and transfected with plasmids by LipofectAMINE (Invitrogen Inc.). The HeLa cells were exposed to 3000 units/ml human IFN-β (Antigenix Corporation, Huntington Station, NY) in combination with 5 μm retinoic acid (RA; Sigma) for 6 days. They were then fixed (methanol:acetic acid; 3:1) for 5 min at 4°C and washed with sterilized dH₂O three times. Subsequently, they were stained with 5 μg/ml Hoechst33342 (Molecular Probes, Eugene, OR) for 10 min at room temperature. Staining was followed by washing three times with sterilized dH₂O. Finally, the slides were mounted with mounting solution (Ref. 24; 0.588% citric acid, 0.71% Na₂HPO4, and 50% glycerol). The stained apoptotic and nonapoptotic nuclei in four randomly selected fields were counted under UV light by two independent investigators.

The anticaspase-3 and anticaspase-9 antibodies were purchased from Cell Signaling (Beverly, MA). The anti-Bcl-XL and anti-Bcl-2 antibody were purchased from BD PharMingen. After the treatments, cells were collected and lysed. About 250 ng to 2 μg of total protein was electrophoresed on a 7.5–15% SDS-PAGE gel, and then transferred to a nitrocellulose membrane. After blocking the membrane with 5% nonfat milk in(PBS +0.1% Tween 20 overnight at 4°C, the blot was incubated with primary antibody for 1 h, washed with PBS +0.1% Tween 20 three times (15 min each time), incubated with secondary antibody (IgG) conjugated with horseradish peroxidase for 1 h, and washed with PBS +0.1% Tween 20 three times. The signal was visualized with the ECL kit (Amersham, Arlington Heights, IL).

To detect the release of cytochrome c from the mitochondria to the cytosol, cells are harvested and resuspended in ice-cold lysis buffer (31) and incubated for 10 min on ice. The cell membrane was disrupted by sonicating (5 strokes at 10% duty cycle, 1 output control) by use of a commercially obtained Sonicator (Sonifer 450; VWR Scientific). The lysate was then centrifuged at 750 g for 15 min at 4°C. The supernatant was collected and concentrated by using the microconcentrator (Amicon, Beverly, MA). The pellet containing the mitochondria was dissolved in lyses buffer again and sonicated again. The supernatants were then loaded onto 13.5% SDS-PAGE gels for Western blot analysis with monoclonal anticytochrome c antibody (2 μg/ml; BD PharMingen Tech).

The apoptosis array from the Clontech Corporation was used to identify genes induced by IFN-β/RA exposure and modulation of apoptotic gene expression by overexpression of GW112 and GRIM-19. Manufacturer’s protocols were followed to carry out the hybridization and analysis.

Total RNA was isolated by TRIzol reagent (Invitrogen) and purified by acid-phenol. The reverse transcription was performed by incubating 2–3 μg total RNA, 2.5 μm random primer, and 200 units Superscript II reverse transcriptase (Invitrogen) in 20 μl reaction volume for 1 h at 42°C. PCR amplification was carried out by use of two pairs of primers in the same reaction: one amplifies the particular gene fragment and another amplifies a 1 kb GAPDH fragment that serves as an internal control. To get semiquantitative results, the PCR cycle number was adjusted to anywhere from 25 to 29 (from the usual 32–26). The PCR products were visualized on 1.2% agarose gel containing 0.5 μg/ml ethidium bromide.

To determine the intracellular localization of the GW112 protein, pCMV-GW112-HA and pCMV-GRIM19-FLAG plasmid was transfected into HeLa cells cultured in tissue culture chambers by using the LipofectAMINE (Invitrogen). The cells were fixed by 3% paraformaldehyde/2% sucrose, and then permeabilized with ice-cold 1% Triton-X-100 buffer. After blocking with 1% BSA for 30 min at room temperature, the cells were incubated with anti-HA-rhodamine and anti-FLAG-FITC antibodies (5 μg/ml) for 30 min at 37°C, washed with PBS buffer, and viewed under a Zeiss confocal fluorescence microscope.

HeLa cells at 50–60% confluency were transfected with pCMV-Tag-2B blank plasmid as a control, pCMV-GRIM-19, or pCMV-GRIM-19/pCMV-GW112 (1:1). Fifteen h after transfection, the cells were cultured with 3000 units/ml IFN-β and 5 μm RA for 6 days. About 200 trypsined cells were subsequently plated in 10-cm Petri dishes. About 10–14 days later, colonies that appeared were fixed using methanol and stained with 2% Giemsa solution. All of the colonies with >50 cells were counted. Five replicate dishes were plated for each group. The survival fraction was normalized against those of the control cells. Surviving fraction (SF) was calculated for each treatment as follows:

Plating efficiency of HeLa cells was determined by plating control nontreated cells at 100–200 cells per plate and counting emerging colonies 14 days later.

Animal care and experimental procedures were carried out in accordance with institutional guidelines. About 10⁷ cells in 50 μl PBS of Tramp-c cells infected with AdGW112 (at a multiplicity of infection of 5) or AdGFP viruses for 15 h were transplanted in the right hind limbs of syngeneic C57BL/6 mice. Over 90% of the cells were infected as evaluated by green fluorescent protein (GFP; which is constitutively expressed from the same AdGW112 vector). Each treatment group consisted of 8–10 animals. Growth curves are plotted as the mean relative tumor volume ± SE versus time after implantation. The error bars represent SE.

The quantitative analysis of apoptosis data in Fig. 4,A and Fig. 5 was carried out by use of Student’s t test. The data analysis for tumor growth (Fig. 7) was carried out by use of the nonparametric Mann-Whitney test.

The GW112 gene was identified by virtue of its high expression in colon cancer cells through analysis of Cancer Genome Anatomy Project data from the United States National Cancer Institute. An extensive database search reveals little homology between GW112 and other proteins. There is only a single published report suggesting that it is overexpressed in the crypt epithelium of inflamed colonic mucosa in ulcerative colitis (24). Because of its overexpression in colon cancer cells, we decided to characterize its function by use of a variety of molecular and cell biology approaches.

To get an idea of the expression profile of GW112 in various tissues, a survey was conducted by use of the serial analysis of gene expression database at the website of National Center for Biotechnology Information. Fig. 1,A is a summary of the results. It can be seen that GW112 is expressed in several normal tissues (prostate, breast, colon, and pancreas) but at relatively low levels. However, its expression is quite high in pancreatic cancer, stomach cancer, and colon cancer tissues. To additionally confirm the expression of GW112 in various tumors, we carried out Northern blot analysis by use of a ³²P-labeled cDNA probe and commercially available MTN blots with RNA from various human tumor tissues. Fig. 1 B shows the results. Among various tumor tissues, expression of the 2.9 kb species (the predominant species) was seen in those of digestive system origin, rectum, colon, and stomach. It was also seen in breast cancer. Interestingly, a new mRNA species of 1.7 kb, perhaps the result of alternative splicing, was observed in multiple tumors that included breast, ovary, uterus, lung, kidney, and stomach cancers. The exact significance of this new mRNA species is not clear. Only the 2.9 kb species is reported in the GenBank.

To decipher the potential function of GW112, we adopted the yeast two-hybrid system to identify the proteins that interact with GW112. The MatchMaker 3 system with a kidney library from Clontech was used. The kidney library was chosen because no library from the digestive system was available, and normal kidney tissues express low levels of the GW112 gene (data not shown). After stringent screening, 20 independent clones were found to interact with GW112 in the yeast two-hybrid assay. After sequencing these genes, we found that only 10 of the cDNAs were fused to the GAL4 AD correctly, indicating that the other clones were artifacts.

In vivo coimmunoprecipitations between GW112 and these 10 genes were subsequently carried out in mammalian cells by use of engineered genes with short protein tags because of the lack of antibodies. The GW112 gene was fused with a HA tag derived from the Influenza HA protein. The candidate genes were fused with a FLAG tag. Plasmids (with a strong, constitutively active CMV) encoding the two tagged genes were then transfected into 293 cells. Forty-eight h after transfection, the cells were lysed and immunoprecipitation/Western blot analysis was conducted. Only 4 of the 10 genes showed positive coimmunoprecipitation with GW112.

An additional test was conducted for these four proteins by use of the mammalian two-hybrid assay. The one that showed the strongest and most consistent interaction was GRIM-19, a novel apoptosis-related gene involved in retinoic acid-IFN-induced cell death (25, 26, 27, 28, 29, 32). Fig. 2 shows the coimmunoprecipitation of GW112 and GRIM-19.

GRIM-19 has been reported to be a nuclear as well as a mitochondrial protein (25, 28, 29, 32). To determine the location of GW112, we used a reporter gene approach where the GW112 gene is fused with the GFP gene at the 3′ end. The GW112-GFP fusion gene was then transfected into HeLa cells together with the reporter gene pdsRed2-Mito, which specifically locates to the mitochondria. Fig. 3,A indicates that almost all of the cytoplasmic GW112-GFP colocalizes with dsRed2-Mito, thereby confirming that cytoplasmic GW112 proteins are mostly localized in the mitochondria. Fig. 3,A also indicates that GW112 proteins exist in the nucleus as distinctive foci. To determine the location of the GRIM-19 protein, immunohistochemistry detection was carried out in cells that had been transfected with FLAG-tagged GRIM-19 gene. The result is shown in Fig. 3 B. It indicates that GRIM-19 is mostly a nuclear protein with cytoplasmic distribution, consistent with earlier reports that GRIM-19 is located both in the nucleus and in the mitochondria (25, 26, 29, 32). Therefore, the interaction of GRIM-19 and GW112 may occur both in the nucleus and in the mitochondria.

As GRIM-19 is a recognized regulator of apoptosis, we conducted two series of experiments to test whether GW112 is involved in cellular apoptotic pathway. In the first series of experiments, the influence of GW112 on apoptosis was tested in H₂O₂-treated SVEC cells. To evaluate the function of GW112 in H₂O₂-induced apoptosis, an adenovirus AdGW112 encoding the human GW112 cDNA under the control of the CMV promoter was engineered to transduce the GW112 gene into SVEC cells. The adenovirus-based approach was adopted because it was much more efficient in gene transduction than plasmid-based approaches. When this virus was used to infect SVEC cells, it mediated efficient expression of the GW112 gene, as determined by a Western blot analysis by use of an antibody to the HA tag fused to the GW112 gene (data not shown). To evaluate the effects of the GW112 expression on H₂O₂-induced cell death, SVEC cells at 50–60% confluency were infected with AdGFP (control) and AdGW112 at a multiplicity of infection of 10 (which can infect >95% of cells). After 15 h of infection, the cells were treated with 0.8 mm H₂O₂ for 5 h and followed by culturing in normal medium for another 24 h. The cells were then evaluated for viability by use of the terminal deoxynucleotidyl transferase-mediated nick end labeling assay (Fig. 4 A), which measures the amount of DNA nicks created by apoptotic enzymes. For each treatment, three Petri dishes were evaluated independently. For each dish, 2000 cells from the area with the highest rate of apoptosis were counted. The average from the three dishes was used to represent apoptotic fraction in each treatment group. About 95.8 ± 1.8% of H₂O₂ treated, AdGFP-infected cells appeared apoptotic (similar to uninfected cells). In comparison, 48.3 ± 2.3% of the AdGW112-infected cells showed signs of apoptosis, a significant attenuation in comparison to AdGFP infected cells.

We also examined the downstream pathways that might be affected by GW112 overexpression. SVEC cells at 50–60% confluency were infected with AdGFP (control vector) or AdGW112 at the multiplicity of infection of 10. Fifteen h later, >95% of the cells were infected when examined for GFP expression. The cells were then treated with 0.8 mm H₂O₂ for different lengths of time (4 and 8 h). A series of Western blot analyses were conducted to determine the status of various well-known players in the apoptotic machinery. Fig. 4 B showed the results. H₂O₂-induced cytochrome c release, a hallmark of mitochondria-mediated apoptosis (33, 34, 35), was effectively inhibited by GW112 expression, as was the activation of caspase 3 and caspase 9, which are the “executioners” of apoptosis. On the other hand, Bcl-2, a potent antiapoptotic protein, showed no change after H₂O₂ treatment in AdGFP as well as AdGW112-infected cells.

In the second series of experiments, we examined the influence of GW112 on apoptosis induced by IFN-β/RA combination. This was because previous reports had indicated that GRIM-19 was a facilitator in apoptosis induced by combined treatment of IFN-β and retinoic acid (25, 26, 27, 28, 29). Our efforts to create a GRIM-19 adenovirus failed because of the potential toxicity of GRIM-19 to host cells. Therefore, we adopted a plasmid transfection approach to transduce HeLa cells, which had shown sensitivity to combined IFN-β/RA treatment (25). HeLa cells at 50–60% confluency were transfected with pCMV-Tag-2B blank plasmid, pCMV-GRIM-19, or pCMV-GW112 +pCMV-GRIM-19 (1:1). Fifteen h later, the cells were treated with 3000 units/ml IFN-β and 5 μm RA for 6 days. To evaluate the cell death, cells were then fixed and stained with Hoechest33342. Hoechst33342 dye stains the nucleus of mammalian cells. Apoptotic cells are typically identified as those cells that possess significantly smaller, condensed, and fragmented nuclei under a fluorescence microscope (Fig. 5 A). About 42.4 ± 2.6% cells exhibited signs of apoptosis in the groups transfected with control plasmids. In comparison, ∼82.5 ± 3.6% of the cells appeared apoptotic in the GRIM-19 transfected cells, indicating that overexpression of GRIM-19 efficiently promoted RA/IFN-β induced apoptosis. Interestingly, only 20.8 ± 4.4% of the cells showed signs of apoptosis in the group cotransfected with pCMV-GW112 and pCMV-GRIM-19. These results suggest significant attenuation of GRIM-19-mediated, RA/IFN-β-induced cellular apoptosis by GW112.

To evaluate the influence of GW112 on long-term cellular survival, a clonogenic assay was used to evaluate cellular survival when HeLa cells were exposed to IFN-β/RA. As shown in Fig. 5 B, pCMV- GRIM-19-transfected cells showed a significantly lower survival (2.6 ± 2.5%; P < 0.01) rate when compared with the blank plasmid transfected cells. On other hand, in HeLa cell transfected with pCMV-GRIM-19 and pCMV-GW112, cellular survival increased to 21.5 ± 5.4% (P < 0.05). These results are consistent with attenuation of apoptosis observed with Hoechst 33342 staining.

The above results indicated to us that GW112 could clearly attenuate or block the ability of GRIM-19 to stimulate apoptosis. To additionally elucidate the molecular mechanisms of GW112 and GRIM-19 in apoptosis regulation, we screened for genes up-regulated by GRIM-19 by use of the microarray approach. Three apoptosis-related genes were identified by this approach. These genes are PIG12 (36, 37), GADD153 (38, 39, 40, 41), and c-Abl (42, 43, 44, 45, 46, 47, 48). All three of the genes have been reported to be involved in apoptosis and oxidative stress. To additionally verify the expression of these three apoptosis-related genes, total RNA from cells transfected with GW112 and/or GRIM-19 genes were extracted and reverse-transcribed. After reverse transcription, PCR amplifications were carried out by use of two pairs of primer in the same reaction: one amplifies the target gene and the other amplifies a 1-kb fragment of the control GAPDH gene. Fig. 6 shows the results of semiquantitative reverse transcription-PCR. The expression of all three of the genes was elevated by cellular exposure to retinoic acid/IFN β, indicating potential involvement of the three genes in RA/IFN-β-exposed cells. The elevation of gene expression was additionally enhanced by forced expression of the GRIM-19, which is consistent with its published role in promoting RA/IFN-β-induced cell death. Most importantly, expression of GW112 abolished the up-regulation of gene expression induced by GRIM-19 expression, suggesting its involvement in the same apoptotic pathway as the latter.

What is the role of GW112 overexpression in certain tumors? We hypothesized that GW112 is overexpressed in cancer cells to facilitate clonogenic survival and tumor growth. To evaluate the influence of GW112 on cellular survival in vivo, the TRAMP-C1 prostate cancer cells, which normally grew slowly after s.c. inoculation (49), were used to assay for their rate of tumor formation. AdGW112 and control AdGFP virus vectors were used to infect TRAMP-C1 cells (at multiplicity of infection of 5 and >90% infection rate). Twenty-four h later, the infected Tramp-C1 cells were injected s.c. into syngeneic C57/BL6 mice. As shown in Fig. 7, tumor volume increased much faster in mice injected with cells infected with AdGW112 (P < 0.01). Because AdGFP infection has no effect on tumor growth (data not shown), these results indicate that GW112 overexpression can indeed promote tumor growth in vivo.

GW112 is a gene initially cloned from human myeloblasts. There is little known about its expression profiles and function. The only report on GW112 is its preferential expression in the crypt epithelium in inflamed colonic mucosa in ulcerative colitis (24). Our results showed that GW112 is overexpressed in human tumors, especially in those of the digestive system. It is also interesting to note that different patterns of RNA expression were seen in the experiments. Whereas a single 2.9 kb RNA species was seen in some tumor tissues, an alternatively spliced 1.7-kb RNA transcript as seen in other tumor tissues (Fig. 1 B). The exact implication of the alternatively splice RNA is unclear.

Because nothing is learned from BLAST comparisons of GW112 with other sequences in GenBank, a yeast two-hybrid system was used to identify the cellular binding partners for it. The identification of GRIM-19 (Figs. 2 and 3) provides revealing information on the potential function of GW112. This is because GRIM-19 is a gene initially identified to be involved in cellular apoptosis induced by IFN-β/RA (25, 28). These studies indicated that GRIM-19 was significantly overexpressed in cells undergoing IFN-β/RA-induced apoptosis. More importantly, the expression of GRIM-19 in HeLa cells directly participated in the apoptotic process. Forced overexpression of the GRIM-19 gene significantly stimulated IFN-β/RA-induced cell death (Fig. 5). Therefore, the association of GW112 with the GRIM-19 protein suggests that GW112 may be involved in regulating apoptosis through association with GRIM-19, similar to other proteins that associate with GRIM-19 (28, 29).

A series of experiments indicated that GW112 is indeed involved regulating apoptosis. GW112 was found to attenuate much of the H₂O₂-induced apoptosis. This attenuation was accompanied by inhibition of cytochrome c release and caspase activation (Fig. 4), typical signs of mitochondria-initiated apoptosis (33, 34, 35, 50). This is consistent with the physical location of the protein, which is largely a mitochondria protein with discreet nuclear distribution (Fig. 3,A). It is worth noting that its associated protein GRIM-19 is largely a nuclear protein with mitochondrial localization (Fig. 3,B). Therefore, GRIM-19 and GW112 have distinct but overlapping localization within the nucleus and mitochondria (Fig. 3).

In addition to the suppression of oxidative stress (i.e., H₂O₂)-induced apoptosis, GW112 can also attenuate the effects of GRIM-19 expression in enhancing cellular apoptosis induced by IFN-β/RA-induced apoptosis (25, 27). However, nothing is known about the upstream activators and downstream effectors of GRIM-19 in the IFN-β/RA induced apoptotic pathway. In this study, microarray analysis identified three apoptosis-related genes Gadd153, PIG12, and c-Abl that were up-regulated during IFN-β/RA-induced apoptosis (Fig. 6). The exact significance and mechanism of the up-regulation of these genes are unknown. However, it is interesting that all three of the genes have been implicated in cellular apoptosis induced by genotoxic agents and oxidative stress (36, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48). At present nothing is known about the mechanism of GRIM-19 that up-regulates the expression of these genes and how GW112 attenuates their expression. It is possible that the expression of these genes is more downstream of GRIM-19 and their up-regulation is involved in common signal transduction pathways that are involved in apoptosis induced by diverse stress signals. However, the fact that GRIM-19 can enhance the expression of these genes, although GW112 can attenuate the expression of them, provided additional proof that GW112 and GRIM-19 are involved in common pathways of apoptosis regulation.

The antiapoptotic function of GW112 also provides a potential explanation for the overexpression of GW112 in tumors, especially those of the digestive origin. It is possible that GW112 enables tumor cells to overcome the adverse growth conditions of the tumor microenvironment, which is often hypoxic, acidic, or nutrient-deprived. Indeed, forced expression of GW112 allows prostate cancer cells that normally grow very slowly (49) to establish in a much quicker fashion (Fig. 7).

What is the normal physiological function of GW112? We still cannot answer this very important question. However, judging from a previous study that demonstrated overexpression of GW112 in the crypt epithelium of inflamed colonic mucosa in ulcerative colitis and our own data on its antiapoptotic function, we postulate that GW112 may be involved in the regulation of cellular apoptosis under inflammatory conditions in the digestive system.

Taken together, we conclude that GW112 is an important regulator of apoptosis, and its regulation roles are played through the GRIM-19 gene. Its antiapoptotic role may have significant roles in the development of cancer, especially those of digestive system.

We thank Mathew Dreher and Fan Yuan for assistance in confocal microscopy.