Serglycin in tumor microenvironment promotes non-small cell lung cancer aggressiveness in a CD44-dependent manner.

Tumor microenvironment (TME) plays an active role in promoting tumor progression. To further understand the communication between TME and tumor cells, this study aimed at investigating the involvement of CD44, a type I cell surface receptor, in the crosstalk between tumor cells and TME. We have previously shown that chondroitin sulfate proteoglycan serglycin (SRGN), a CD44-interacting factor, was preferentially secreted by cancer-associated fibroblasts (CAFs) for promoting tumor growth in breast cancer patients. In this study, we show that SRGN is overexpressed in primary non-small cell lung cancers (NSCLCs), by both carcinoma and stromal cells. Using gain-of-function and loss-of-function approaches, we show that SRGN promotes NSCLC cell migration and invasion as well as colonization in the lung and liver in a CD44-dependent manner. SRGN induces lung cancer cell stemness, as demonstrated by its ability to enhance NSCLC cell sphere formation via Nanog induction, accompanied with increased chemoresistance and anoikis-resistance. SRGN promotes epithelial-mesenchymal transition by enhancing vimentin expression via CD44/NF-κB/claudin-1 (CLDN1) axis. In support, CLDN1 and SRGN expression are tightly linked together in primary NSCLC. Most importantly, increased expression of SRGN and/or CLDN1 predicts poor prognosis in primary lung adenocarcinomas. In summary, we demonstrate that SRGN secreted by tumor cells and stromal components in the TME promotes malignant phenotypes through interacting with tumor cell receptor CD44, suggesting that a combined therapy targeting both CD44 and its ligands in the TME may be an attractive approach for cancer therapy.

Autocrine/paracrine mechanism of interleukin-17B receptor promotes breast tumorigenesis through NF-κB-mediated antiapoptotic pathway.

Gain of function of membrane receptor was a good strategy exploited by cancer cells to benefit own growth and survival. Overexpression of HER2 has been found to serve as a target for developing trastuzumab to treat 20-25% of breast cancer. However, little or none of the other membrane receptor was found to be useful as a potential target for breast cancer treatment since then. Here, we showed that amplified signaling of interleukin-17 receptor B (IL-17RB) and its ligand IL-17B promoted tumorigenicity in breast cancer cells and impeded acinus formation in immortalized normal mammary epithelial cells. External signal transmitted through IL-17RB activated nuclear factor-κB to upregulate antiapoptotic factor Bcl-2 and induced etoposide resistance. Elevated expression of IL-17RB had a stronger correlation with poor prognosis than HER2 in breast cancer patients. Interestingly, breast cancer patients with high expression of IL-17RB and HER2 had the shortest survival rate. Depletion of IL-17RB in trastuzumab-resistant breast cancer cells significantly reduced their tumorigenic activity, suggesting that IL-17RB and HER2 have an independent role in breast carcinogenesis. Furthermore, treatment with antibodies specifically against IL-17RB or IL-17B effectively attenuated tumorigenicity of breast cancer cells. These results suggest that the amplified IL-17RB/IL-17B signaling pathways may serve as a therapeutic target for developing treatment to manage IL-17RB-associated breast cancer.

Mitochondrial genome instability resulting from SUV3 haploinsufficiency leads to tumorigenesis and shortened lifespan.

Mitochondrial dysfunction has been a hallmark of cancer. However, whether it has a causative role awaits to be elucidated. Here, using an animal model derived from inactivation of SUV3, a mitochondrial helicase, we demonstrated that mSuv3+/- mice harbored increased mitochondrial DNA (mtDNA) mutations and decreased mtDNA copy numbers, leading to tumor development in various sites and shortened lifespan. These phenotypes were transmitted maternally, indicating the etiological role of the mitochondria. Importantly, reduced SUV3 expression was observed in human breast tumor specimens compared with corresponding normal tissues in two independent cohorts. These results demonstrated for the first time that maintaining mtDNA integrity by SUV3 helicase is critical for cancer suppression.

miR-495 is upregulated by E12/E47 in breast cancer stem cells, and promotes oncogenesis and hypoxia resistance via downregulation of E-cadherin and REDD1.

MicroRNAs (miRNAs) are involved in tumorigenecity by regulating specific oncogenes and tumor suppressor genes, and their roles in breast cancer stem cells (BCSCs) are becoming apparent. Distinct from the CD44(+)/CD24(-/low) sub-population, we have isolated a novel PROCR(+)/ESA(+) BCSC sub-population. To explore miRNA-regulatory mechanisms in this sub-population, we performed miRNA expression profiling and found miR-495 as the most highly upegulated miRNA in PROCR(+)/ESA(+) cells. Coincidently, high upregulation of miR-495 was also found in CD44(+)/CD24(-/low) BCSCs, reflecting its potential importance in maintaining common BCSC properties. Ectopic expression of miR-495 in breast cancer cells promoted their colony formation in vitro and tumorigenesis in mice. miR-495 directly suppressed E-cadherin expression to promote cell invasion and inhibited REDD1 expression to enhance cell proliferation in hypoxia through post-transcriptional mechanism. miR-495 expression was directly modulated by transcription factor E12/E47, which itself is highly expressed in BCSCs. These findings reveal a novel regulatory pathway centered on miR-495 that contributes to BCSC properties and hypoxia resistance.

Application of mucin quantitative competitive reverse transcription polymerase chain reaction in assisting the diagnosis of malignant pleural effusion.

Aberrant expression of mucin genes occurs frequently in advanced cancer. Using quantitative competitive reverse transcriptase/polymerase chain reaction (QC RT-PCR), the expression of three mucin genes--MUC1 (widely expressed in epithelial cells), MUC2 (mainly expressed in intestinal epithelial cells), and MUC5AC (mainly from airway and gastric epithelial cells)--was evaluated in 112 patients with pleural effusions (including 54 cytologically positive malignant pleural effusions, 35 benign exudative pleural fluids, and 23 cytologically negative pleural effusions from cancer patients). The expression ratios of MUC1 and MUC5AC, but not MUC2 gene, were significantly higher in malignant than benign pleural fluids (p < 0.000). The cutoff value, sensitivity, and specificity of MUC1 expression ratio were: 0.126, 64.6%, and 95.7%; and were 0.028, 72.3%, and 95.7%, respectively, for MUC5AC. In combined evaluation with both MUC1 and MUC5AC, the sensitivity was 86.1% and specificity was 91.5%. The positive and negative predictive values were 93.3%, and 82.7%, respectively. We considered mucin QC RT-PCR to be a useful tool in assisting the diagnosis of malignant pleural effusion.

Luteinizing hormone up-regulates the expression of interleukin-1 beta mRNA in human granulosa-luteal cells.

PROBLEM: Previously, we observed that follicular fluid obtained from patients with premature luteinization contained elevated interleukin-1 beta (IL-1 beta) levels. In this study. we aimed to examine the effects of luteinizing hormone (LH) on IL-1 beta expression and IL-1 beta on steroidogenesis in human granulosa-luteal cells. METHOD OF STUDY: Human granulosa-luteal cells were obtained during oocyte retrieval. The cells were treated with either LH or IL-1 beta and subsequently were examined for the level of IL-1 beta transcript. The conditioned media were examined for IL-1 beta protein and steroid hormone levels. RESULTS: LH (250-500 mIU/mL) up-regulated the expression of IL-1 beta mRNA (up to a 4-fold increase over control; P<0.05) in the granulosa-luteal cells. IL-1 beta (5-50 ng/mL) increased the basal, but not LH-dependent, progesterone production from these cells in a dose-dependent manner after 96 and 144 hr of culture (P<0.05). However, an inhibitory effect of IL-1 beta on LH-dependent estradiol production was observed (up to 20% decrease, P<0.05). CONCLUSIONS: LH is capable of stimulating IL-1 beta transcript expression in human granulosa-luteal cells and may regulate ovarian steroidogenesis, at least partly through the activation of IL-1 beta.

Expression of leukemia inhibitory factor and its receptor in preimplantation embryos.

OBJECTIVE: To examine the expression of leukemia inhibitory factor (LIF) and its receptor (LIF-R) transcripts in human and murine preimplantation embryos. DESIGN: Prospective study. SETTING: University medical center. PATIENT(S): Human oocytes were obtained from patients undergoing IVF treatment. Two-cell murine embryos were obtained from ICR strain mice. INTERVENTION(S): Second-day intracytoplasmic sperm injection procedures were performed on oocytes that failed to be fertilized by IVF. Embryos were cultured to various stages and collected for study. MAIN OUTCOME MEASURE(S): The transcript levels of LIF and LIF-R in these embryos were examined and semiquantitated using single-cell reverse transcription-polymerase chain reaction methodology. RESULT(S): Leukemia inhibitory factor and LIF-R transcripts were detectable in most human preimplantation embryos (30 of 34 and 31 of 34 embryos showed LIF and LIF-R messenger RNA, respectively). There was a trend toward decreased expression of both transcripts in embryos at the four-cell stage and in embryos in which growth had been arrested for 24-48 hours. The expression of LIF and LIF-R genes in murine embryos was inconsistent. CONCLUSION(S): Preimplantation human embryos express LIF and LIF-R messenger RNA. It is suggested that LIF may be able to affect embryo development through its action at stages before implantation in an autocrine or paracrine manner.

Quantitative analysis of mRNA encoding MUC1, MUC2, and MUC5AC genes: a correlation between specific mucin gene expression and sialomucin expression in non-small cell lung cancer.

The expression of mucins is important for tumor invasiveness and metastasis. In our previous report (Am. J. Respir. Crit. Care Med. 1997; 155:1419-1427), non-small cell lung cancers bearing sialomucin expression tended to relapse earlier than those without sialomucin. However, it remained unclear whether the expression of sialomucin in lung cancer is caused by an abnormal glycosylation process or by the expression of a specific mucin gene product. To address this problem, we established a modified quantitative competitive polymerase chain reaction (QC-PCR) analysis. RNA internal standards of MUC1, MUC2, and MUC5AC non-tandem repeat sequences were constructed, and known copy numbers of mucin RNA internal standards were introduced into reverse transcription-polymerase chain reactions (RT-PCR) for each mucin gene in order to compete with native mucin gene RNA during the reaction. The RNA of Gbeta-like gene (a housekeeping gene) was used as internal control for the RNA analysis. Twenty-five lung cancer tissues (13 adenocarcinomas and 12 squamous cell carcinomas) were used for analysis. Mann-Whitney rank sum test was applied to compare the expression amounts of different mucin genes in tissues. The results revealed that adenocarcinoma expressed higher amounts of MUC5AC gene than did squamous cell carcinoma (P = 0.03). The expression amount of MUC5AC correlated positively with the expression status of sialomucin (P = 0.012). Further studies are anticipated to elucidate the underlying mechanism contributing to this phenomenon.

p53 point mutation enhanced by hepatic regeneration in aflatoxin B1-induced rat liver tumors and preneoplastic lesions.

Aflatoxin B1 (AFB1) is a well-known mutagen and carcinogen which induces human hepatocellular carcinoma (HCC). It has been found to be an important factor in inducing a high frequency of codon 249 mutation in the p53 gene. We characterized p53 mutations in specimens from preneoplastic lesions or tumors from the liver of rats induced by AFB1 with or without regeneration by partial hepatectomy treatment. PCR-SSCP and direct sequencing were used for screening and identification of p53 gene mutations in these samples. In rats treated with AFB1 with or without liver regeneration, 29% (5/17) of rats with hepatoma or neoplastic lesion had p53 mutation. No p53 mutations were found in the tumor samples from the rats without liver regeneration. However, in samples from the rats with liver regeneration, 38% (5/13) of the rats with hepatoma or neoplastic lesion were found to have p53 mutations. In one of these samples, we also observed a transversion mutation G --> T on codon 247, compared to codon 249 in humans. These findings suggest that the process of hepatocarcinogenesis induced by AFB1 does not necessarily involve p53 mutation, but mutation of the p53 gene can be enhanced by liver regeneration and that there is a possibility of a high mutability of the third base of codon 247, even though there is a small probability of detecting such a mutation in rats due to its silent mutation.

Sialomucin expression is associated with erbB-2 oncoprotein overexpression, early recurrence, and cancer death in non-small-cell lung cancer.

Mucin production, when heavily sialylated, can promote cancer cell invasion and metastasis, and modulate the immune recognition system of the host. To explore the prognostic implication of sialomucin expression in lung cancer, we studied 116 patients with non-small-cell lung cancer (NSCLC). Tumor specimens were stained immunohistochemically with monoclonal antibodies (mAbs) against mucin glycoprotein (17Q2, HMFG2, SM3), and histochemically with periodic acid-Schiff/alcian blue to differentiate neutral mucin from acid mucin, and with high-iron diamine/alcian blue to differentiate sialomucin from sulfomucin. The expression status of two established molecular prognostic factors, the p53 and erbB-2 oncoproteins, were evaluated immunohistochemically. The staining was performed on two separately archived, paraffin-embedded tumor blocks for each patient, with normal lung as a control. Correlations were subsequently made among stains and various clinicopathologic factors. All analyses were blinded, and included Kaplan-Meier survival estimates with Cox proportional hazards regression modeling. Associations were established among adenocarcinoma histotype and erbB-2 overexpression, sialomucin expression, and 17Q2 and HMFG2 immunohistochemical positivity (p < 0.05). Sialomucin expression was closely linked to erbB-2 overexpression (p = 0.01). Significant univariate predictors (p < 0.05) of recurrence and cancer death were surgical stage, p53 expression, erbB-2 overexpression, and sialomucin expression. These four factors remained as independent predictors of early recurrence (p < 0.05) after multivariate analysis. For cancer death prediction, p53 and sialomucin expression had a marginal effect. We concluded that sialomucin expression is also a poor indicator of prognosis, which is associated with erbB-2 oncoprotein overexpression, early postoperative recurrence, and cancer death in NSCLC.

Distinct p53-mediated G1/S checkpoint responses in two NIH3T3 subclone cells following treatment with DNA-damaging agents.

N3T3 and P-3T3 cells, originally isolated from a NIH3T3 cell clone on the basis of their negative and positive transformation by v-Abl, v-Src and Bcr-Abl, were previously found to show distinct cyclin activity changes following 12-O-tetradecanoylphorbol-13-acetate (TPA) treatment, which is anti-mitogenic for N-3T3 cells and mitogenic for P-3T3 cells. We have found in this study that, while the G1/S arrest and cell death induced by serum starvation and TPA treatment in N-3T3 cells did not involve p53-mediated checkpoint or apoptosis, N-3T3 and P-3T3 cells evidently responded differently in these aspects of cell cycle regulation to DNA-damaging agents, methylmethane sulfonate (MMS) and gamma-radiation. In N-3T3 cells, DNA damages elicit cell growth arrest at G1/S transition with concomitant accumulation of p53 and p53-inducible Waf1/Cip1 proteins and also signs of apoptosis such as DNA ladder patterns and apoptotic (subgenomic) peak in flow cytograph. Conversely, P-3T3 cells treated with the DNA-damaging agents showed no cell cycle interruption nor accumulation of p53 or Waf1/Cip1. However, both P-3T3 and N-3T3 cells showed the same p53 protein half-life of 40 min or less, the same wild-type p53 DNA sequence and the same co-immunoprecipitable cellular proteins in complexes with p53, suggesting that an alteration in a signal transduction pathway upstream of p53 might account for the evasion of p53-mediated G1 checkpoint in P-3T3 cells.

Mucin mRNA expression in lung adenocarcinoma cell lines and tissues.

The expression pattern of mucin genes was studied in 7 lung adenocarcinoma cell lines (CL1, CL2, CL3, NCL2, PC9, PC13, PC14) and 12 lung adenocarcinoma tissues. CL1 and PC13 are poorly differentiated cell lines with low mucin glycoprotein production. The other 5 cell lines are well differentiated and produce a higher amount of mucins. Total RNA was extracted from these cell lines. Northern blot analysis was performed by hybridization with specific antisense oligonucleotide probes recognizing mucin-specific tandem repeats of 4 mucin genes (MUC1, MUC2, MUC3, MUC4). RT-PCR was carried out to amplify the 3' and 5' nonrepetitive coding regions of MUC1 and the 5' nonrepetitive coding region of MUC2. All these cell lines expressed MUC1, MUC2, and MUC4 mRNA but in variable mounts. The poorly differentiated cell lines (CL1 and PC13) had a relatively low level of expression of MUC1, MUC2, MUC3 and MUC4. RT-PCR, with primers amplifying the MUC1 nonrepetitive coding region 5' end, 293 bp, and the 3' end, 522 bp, as well as the MUC2 nonrepetitive 5' coding region, 308 bp, revealed the presence of MUC1 and MUC2 mRNA in all the cell lines. Sequence analysis of the PCR products were very homologous, similar to previously published MUC1 and MUC2 cDNA sequences. The expression pattern of mucin genes is consistent with that of mucin glycoproteins as studied using biochemical and immunological methods. Northern blotting and RT-PCR analysis in 12 lung adenocarcinoma tissues with various grades of differentiation (6 poorly differentiated adenocarcinomas and 6 moderately to well-differentiated adenocarcinomas) showed heterogeneous expression of the 4 mucin genes in tissues without clear correlation with the differentiation grade. Therefore the clinical implications of the differential expression of the mucin genes need further investigation.

Exon 8 mutation of p53 gene associated with nodal metastasis in non-small-cell lung cancer.

The epidemiologic characteristics of lung cancer in Taiwan differ from those in other parts of the world in low male-to-female ratio, the high percentage of adenocarcinoma, and the relatively high percentage of nonsmokers who are victims. To investigate possible correlation between p53 gene alteration and the unique characteristics of lung cancer here, p53 gene status of 36 patients with primary, resected non-small-cell lung cancer (NSCLC) was studied by directly sequencing the cDNA of the p53 gene, then acquiring clinical and pathologic data to correlate p53 gene status with clinical parameters and pathologic staging. Missense mutations were present in 42% (15 of 36) of patients with NSCLC, including 42% (10 of 24) with adenocarcinomas, and 45% (five of 11) with squamous cell carcinomas. The frequency of p53 mutation was 50% in smokers and 29% in nonsmokers (p = 0.355). The mutation occurred most frequently in exon 8 (56%), and G:C to A:T transitions in non-CpG or CpG sites were the most commonly observed base changes (56%). These findings differ from the high prevalence of G to T transversion found in previous reports. The frequency of metastasis in hilar and mediastinal lymph nodes was significantly higher in tumors with p53 mutations. The association with nodal stage was strong for mutations within exon 8, but it was less apparent for mutations in other exons probably because of the small number. This study suggests that p53 gene missense is common in NSCLC in Taiwan, but smoking is probably not the sole contributing factor. More interestingly, p53 gene mutations, especially those in exon 8, may be associated with regional nodal metastasis.

Nasopharyngeal carcinoma and retinoblastoma gene expression.

BACKGROUND: The pathogenesis of nasopharyngeal carcinoma (NPC) is not well defined yet. To evaluate oncosuppressor genes in NPC, two NPC cell lines were investigated for expression of the retinoblastoma (RB) gene. EXPERIMENTAL DESIGN: We used Western blotting to identify RB protein species, and checked the RB gene conformation by Southern blotting. We also used immunohistochemistry, in situ nucleic acid hybridization, and in situ extraction of RB protein to observe RB protein in NPC culture cells. RESULTS: Both cell lines as well as sublines could synthesize normal RB proteins of 110, 113 and 114 kilodaltons. These cells showed no RB DNA rearrangement. Most interphase cells showed variable amounts of anti-RB reaction product in their nuclei when immunostained by 13 monoclonal antibodies. However, all mitotic cells contained RB protein either in the cytosol or the chromosomes, depending upon the mitotic phase. The level of RB mRNA increased slightly in mitotic cells as compared with interphase cells. Double localization of bromodeoxyuridine and RB protein in NPC cells and localization of RB protein in synchronized NPC cells in different phases of the cell cycle revealed random RB protein expression in each individual tumor cell. Co-localization of RB mRNA and RB protein in interphase cells showed a different degree of RB message expression in each cell. Low-salt hypotonic buffer could remove a fraction of RB protein from stained interphase nuclei. A similar finding with slight variations was observed in 16 other cancer cell lines. CONCLUSIONS: These data indicate that our NPC cell lines contain no obvious RB gene rearrangement, but each cancer cell may have an abnormal expression of RB mRNA and protein. This phenomenon may also be true in other cancer cell lines with a normal RB gene.

A single Cys706 to Phe substitution in the retinoblastoma protein causes the loss of binding to SV40 T antigen.

Most naturally occurring mutants of the retinoblastoma (RB) protein contain large deletions or truncations. The small cell lung carcinoma cell line H209 contains a normal-sized but unphosphorylated RB protein (Hensel et al., Cancer Res., 50: 3067-3072, 1990), which fails to form a complex with SV40 T antigen, suggesting that the RB gene of H209 may contain a subtle mutation. To define this mutation, the RB complementary DNA and genomic DNA were sequenced, revealing a point mutation in exon 21 that changed a G to a T. This results in an amino acid substitution of a Phe for Cys706. The mutant RB complementary DNA was used as a template for in vitro transcription and translation to synthesize the mutated protein. The resulting protein failed to bind to SV40 T antigen, demonstrating that a single missense mutation of the RB gene led to the complete inactivation of the ability of the RB protein to bind T antigen.

Suppression of tumorigenicity of human prostate carcinoma cells by replacing a mutated RB gene.

Introduction of a normal retinoblastoma gene (RB) into retinoblastoma cells was previously shown to suppress several aspects of their neoplastic phenotype, including tumorigenicity in nude mice, thereby directly demonstrating a cancer suppression function of RB. To explore the possibility of a similar activity in a common adult tumor, RB expression was examined in three human prostate carcinoma cell lines. One of these, DU145, contained an abnormally small protein translated from an RB messenger RNA transcript that lacked 105 nucleotides encoded by exon 21. To assess the functional consequences of this mutation, normal RB expression was restored in DU145 cells by retrovirus-mediated gene transfer. Cells that maintained stable exogenous RB expression lost their ability to form tumors in nude mice, although their growth rate in culture was apparently unaltered. These results suggest that RB inactivation can play a significant role in the genesis of a common adult neoplasm and that restoration of normal RB-encoded protein in tumors could have clinical utility.

Homozygous deletion of the retinoblastoma gene in an acute lymphoblastic leukemia (T) cell line.

Human leukemia cell lines were examined for the status of the retinoblastoma (RB) protein by immunoblotting analysis using antibodies raised against the TrpE-RB fusion protein. One of 16 cell lines examined, the T-cell acute lymphoblastic leukemia (ALL) line HSB-2, lacked the 110-Kd RB protein. Southern blot analysis of genomic DNA extracted from HSB-2 cells showed a large homozygous deletion of the RB gene, stretching from exon 18 beyond exon 27. Northern blot analysis showed multiple, abnormal RB transcripts in HSB-2. A truncated protein (72 Kd) was detected with 35S-methionine labeling but not with 32P-orthophosphate labeling of the HSB-2 cells. The genomic deletion of greater than 85 kb DNA at the RB locus (13q14) was not detectable in the karyotype of the HSB-2 cells. Among the 16 human leukemia cell lines examined for the status of the RB gene, only one, the HSB-2 line, showed an abnormal RB protein. Further study of primary leukemia and lymphoma samples appears to be warranted.

Deletion of a splice donor site ablates expression of the following exon and produces an unphosphorylated RB protein unable to bind SV40 T antigen.

Studies of mutated retinoblastoma (RB) proteins in human tumor cells potentially reveal regions of the normal RB gene product that are required for its cancer suppression function. We here characterize a mutated RB protein of Mr 104,000 (p104) from a primary small-cell lung carcinoma. Unlike normal RB protein (pp110RB), p104 was unphosphorylated and unable to bind T antigen of SV40 both in vivo and in vitro. On the other hand, nuclear localization and DNA binding activity were preserved in the mutated protein. p104 was immunoprecipitable with four separate polyclonal antibodies recognizing different epitopes of the RB polypeptide, suggesting the presence of most exons in their correct reading frame. Following reverse transcription and in vitro amplification, RB mRNA from this tumor was shown to lack nucleotides encoded by exon 16. Analysis of genomic DNA from this tumor showed that exon 16 and its flanking splice donor and acceptor sequences were present and entirely normal; however, a 43-base pair (bp) region containing the splice donor site of intron 15 was deleted instead. Exon 15 was joined directly to exon 17 during mRNA processing via a cryptic splice donor site; exon 16 was presumably skipped because the preceding mutated intron was of insufficient length (less than 80 bp) for normal RB mRNA processing. These results demonstrate that loss of a single small exon disrupts several important biochemical properties of RB protein. In addition, sequence features of the 43-bp depletion suggest involvement of a novel deletional mechanism.

C-terminal truncation of the retinoblastoma gene product leads to functional inactivation.

Mutational inactivation of the retinoblastoma (RB) gene has been implicated in the genesis of retinoblastoma, osteosarcoma, and other human tumors. Our strategy has been to characterize naturally occurring mutants from tumor cells to pinpoint potential domains of RB protein crucial for tumor suppression. We show here that osteosarcoma cell line Saos-2 contains an abnormal endogenous RB protein of 95 kDa (p95) that is located mainly in the cytoplasm. This protein was identified by antibodies recognizing several different RB epitopes, but not by one directed solely against the C terminus, suggesting C-terminal truncation. This conclusion was supported by analysis of mRNA and genomic DNA, which revealed that a transcriptionally active RB allele had a deletion of exons 21-27. In contrast to normal RB protein, this truncated protein was not phosphorylated and did not bind to the large tumor (T) antigen encoded by simian virus 40. We previously reported that introduction of normal RB protein into Saos-2 cells suppressed their neoplastic phenotype, indicating functional inactivation of their endogenous RB genes. These results provide an initial step to elucidate domains crucial to the cancer-suppression function of RB protein; its C-terminal portion is evidently important for this activity.