Expression level of valosin-containing protein (VCP) as a prognostic marker for gingival squamous cell carcinoma.

BACKGROUND: Valosin-containing protein (VCP) is associated with anti-apoptotic function and metastasis via activation of the nuclear factor-kappaB signaling pathway. In the present study, association of VCP expression with prognosis of gingival squamous cell carcinoma (GSCC) was examined. PATIENTS AND METHODS: VCP expression in 74 patients with GSCC (34 males and 40 females) with ages ranging from 42 to 85 (median 66) years was evaluated by immunohistochemistry, in which staining intensity in tumor cells was categorized as either weaker (level 1) or equal to/stronger (level 2) than that in the endothelial cells. RESULTS: Twenty-four (32.4%) cases showed level 1 and 50 (67.6%) level 2 VCP expression. Patients with level 1 GSCC showed a significantly better 5-year survival rate than those with level 2 GSCC (5-year overall survival: 100% versus 84.9%, P < 0.05). Multivariate analysis revealed VCP expression level, lymph node metastasis and pT(TNM) to be independent factors for overall survival. Patients with GSCC at stages I and II showed favorable prognosis regardless of VCP expression status, whereas at stages III and IV, patients with level 1 VCP expression showed better survival rates than those with level 2 expression. CONCLUSION: Prognostic significance of VCP expression level in GSCC was demonstrated.

Expression of dentin matrix protein 1 (DMP1) during fracture healing.

Dentin matrix protein 1 (DMP1) is one of the acidic phosphorylated extracellular matrix proteins called the SIBLING (small integrin-binding ligand, N-linked glycoproteins) family. Recent studies showed that DMP1 is expressed in the mineralized tissues and suggested that DMP1 is involved in the mineralization. We investigated the precise localization of DMP1 messenger RNA (mRNA) and protein during fracture healing. In situ hybridization demonstrated that DMP1 mRNA was strongly expressed in preosteocytes and osteocytes in the bony callus during intramembranous and endochondral ossification while DMP1 mRNA was not detected in osteoblasts and chondrocytes. During endochondral ossification, however, a low number of DMP1-expressing cells were identified in the cluster of hypertrophic chondrocytes. However, these DMP1-expressing cells were not hypertrophic and were likely to be osteoblast-lineage cells, which were embedded in the matrix of bone or cartilage, because type I collagen-expressing cells and invasion of capillary vessels were observed in the same area. Northern blot, in situ hybridization, and immunohistochemical analyses showed that DMP1 mRNA and protein expressions were increased until day 14 postfracture, when bony callus was formed, and then declined to a lower level during remodeling of the bony callus. Therefore, DMP1 is likely to play an important role in the mineralization of the bony callus.

mRNA expression and protein localization of dentin matrix protein 1 during dental root formation.

Dentin matrix protein 1 (DMP1) is an acidic phosphoprotein. DMP1 was initially detected in dentin and later in other mineralized tissues including cementum and bone, but the DMP1 expression pattern in tooth is still controversial. To determine the precise localization of DMP1 messenger RNA (mRNA) and the protein in the tooth, we performed in situ hybridization and immunohistochemical analyses using rat molars and incisors during various stages of root formation. During root dentin formation of molars, DMP1 mRNA was detected in root odontoblasts in parallel with mineralization of the dentin. However, the level of DMP1 mRNA expression in root odontoblasts decreased near the coronal part and was absent in coronal odontoblasts. DMP1 protein was localized along dentinal tubules and their branches in mineralized root dentin, and the distribution of DMP1 shifted from the end of dentinal tubules to the base of the tubules as dentin formation progressed. During the formation of the acellular cementum, DMP1 mRNA was detected in cementoblasts lining the acellular cementum where its protein was localized. During the formation of the cellular cementum, DMP1 mRNA was detected in cementocytes embedded in the cellular cementum but not in cementoblasts, and its protein was localized in the pericellular cementum of cementocytes including their processes. During dentin formation of incisors, DMP1 mRNA was detected in odontoblasts on the cementum-related dentin, where its protein was localized along dentinal tubules near the mineralization front. The localization of DMP1 mRNA and protein in dentin and cementum was related to their mineralization, suggesting that one of the functions of DMP1 may be involved in the mineralization of dentin and cementum during root formation.

Dentin matrix protein 1 is predominantly expressed in chicken and rat osteocytes but not in osteoblasts.

Although osteocytes are the most abundant cells in bone, little is known about their function, and no specific marker protein for osteocytes has been described. Dentin matrix protein 1 (DMP1) is an acidic phosphoprotein expressed in tooth organ and bone. Our previous work showed that in the chicken, which is not capable of forming tooth, DMPI messenger RNA (mRNA) is highly expressed in bone by Northern blot analysis. To clarify the significance of DMP1 expression in bone, the expression of DMP1 mRNA and its protein was examined in the chicken and rat. In the chicken, DMPI mRNA was detected only in bone tissues and was localized in osteocytes and preosteocytes but not in osteoblasts. Similarly, in the rat, DMPI mRNA was predominantly expressed in osteocytes and preosteocytes in bone matrix but not in osteoblasts located at the bone surface. Antiserum was raised against the peptide from rat DMP1, and the localization of DMP1 was examined by immunohistochemistry. In the development of bone, DMP1 was first detected in newly formed bone matrix after osteoblastic cells had been embedded within it. After the appearance of typical osteocytes, DMP1 was localized in the pericellular bone matrix of osteocytes, including their processes. These data show that DMP1 is a bone matrix protein specifically expressed in osteocytes and preosteocytes and suggest that DMP1 plays a role in bone homeostasis because of its high calcium ion-binding capacity.

Induction of adenocarcinoma containing myoepithelial cells in rat submandibular gland by 7,12-dimethylbenz(a)anthracene.

In an attempt to induce adenocarcinoma containing myoepithelial cells (MECs) in the rat submandibular gland, we injected 7,12-dimethylbenz(a)anthracene (DMBA) dissolved in acetone into the glands of rat pups at the age of 10 days. In both male and female pups, the glands, including their developing terminal secretory units, contained far greater numbers of cells positive for proliferating cell nuclear antigen (PCNA) than did adult glands. A single administration of 1% DMBA (0.05 ml/130 g b.w.) did not produce adenocarcinoma, but did induce occasional sarcomas, such as rhabdomyosarcoma and fibrosarcoma, in 2 months. Most glands regenerated with minimal scar formation. Microscopically, these glands were atypical in that they contained increased numbers of PCNA-positive cells, underdeveloped granular ducts, and striated ducts surrounded by MECs positive for alpha smooth muscle actin (alphaSMA). Though these features were also observed in the regenerated glands after acetone injection, the number of PCNA-positive cells was relatively high in the glands of DMBA-treated females, especially in the terminal secretory unit. The second DMBA injection at 10 weeks of age produced adenocarcinoma made up of alphaSMA-positive MECs and keratin 19-positive duct cells. Such MEC-associated adenocarcinoma was induced in the glands of more than half the female but not the male animals. Replacement of either of the double DMBA treatments with acetone, or DMBA treatment, single or double, of adult glands did not produce adenocarcinoma, but did produce sarcoma and squamous cell carcinoma. These results suggest that (1) at least two genetic mutations are necessary for induction of adenocarcinoma with MECs in the rat submandibular gland, (2) the mutation is efficiently introduced to pup glands whose terminal secretory units exhibit extreme proliferative activity, and (3) the second mutation is difficult to introduce in male glands, whose proliferative activity is relatively low, and/or transformed cells need some female hormone after the mutation to propagate.

Cloning and characterization of the human ameloblastin gene.

We isolated the full-length human ameloblastin (AMBN) cDNA clone using reverse transcription-polymerase chain reaction (RT-PCR) methods. Sequence analysis of the AMBN cDNA revealed an open reading frame of 1341bp encoding a 447-amino-acid protein. Comparison with pig, cattle, rat, and mouse AMBN sequences showed a high amino acid sequence similarity and led to the identification of a novel 78bp (26 amino acids) insert resulting from internal sequence duplication. By DNA analysis of a human genomic clones, the AMBN gene was shown to consist of 13 exons and a novel 78bp segment, which proved to comprise two small exons. Human ameloblastomas express AMBN transcripts that contain some mutations.

Keratin 14 immunoreactive cells in pleomorphic adenomas and adenoid cystic carcinomas of salivary glands.

Our recent study of developing myoepithelial cells (MECs) in rat salivary glands demonstrated that developing MECs begin to express alpha-smooth muscle actin (alphaSMA) first and, thereafter, keratin 14. Therefore, it is unlikely that duct basal cells expressing keratin 14 alone are immature or undifferentiated MECs. In this study we carried out immunohistochemistry of pleomorphic adenomas and adenoid cystic carcinomas including normal salivary glands using monoclonal antibodies to keratin 14, smooth muscle proteins and keratin 19. The smooth muscle proteins examined included alphaSMA, h-caldesmon and h1-calponin; h1-calponin was observed in keratinocytes and nerve fibers, indicating that the protein is not specific to smooth muscle, whereas alphaSMA and h-caldesmon turned out to be highly specific markers for smooth muscle cells in normal tissues. In normal glands, MECs were positive for both keratin 14 and smooth muscle proteins (alphaSMA and h-caldesmon). Non-MEC cells were essentially devoid of smooth muscle proteins. Non-MEC duct basal cells expressed keratin 14 with or without keratin 19, and luminal cells keratin 19 with or without keratin 14. This suggests that the keratin 14-positive, smooth muscle proteins-negative duct basal cells are luminal cell progenitors. Luminal cells in tubular structures of both tumors were positive for keratin 19 with or without keratin 14. Nonluminal peripheral cells of pleomorphic adenomas were mostly positive for keratin 14, and a small fraction of them expressed smooth muscle proteins. Conversely, peripheral cells of adenoid cystic carcinomas were mostly positive for smooth muscle proteins, and some of them expressed keratin 14. These results strongly suggest (1) that the luminal cell progenitors transform into major constituents of pleomorphic adenoma cells with keratin 14 but not smooth muscle proteins, and (2) that the peripheral cells of adenoid cystic carcinoma are derived from undifferentiated MECs. Solid structures of pleomorphic adenomas were formed by proliferation of the peripheral cells. MECs were observed only occasionally in the periphery. Solid and cribriform structures of adenoid cystic carcinomas were formed by proliferation of the luminal cells. MECs were observed in the periphery and around the pseudocyst.

Small cell undifferentiated carcinoma of the submandibular gland: immunohistochemical evidence of myoepithelial, basal and luminal cell features.

A primary small cell undifferentiated carcinoma of the submandibular gland is reported. Histological studies revealed that the major part of this tumor was composed of cells slightly larger (10-14 microm) than lymphocytes. These tumor cells showed myoepithelial-cell differentiation, which was confirmed by the immunohistochemical and ultrastructural findings. Furthermore, some of them showed luminal-cell and basal-cell differentiation immunohistochemically. However, there was no evidence of neuroendocrine differentiation. These findings demonstrated that the tumor had the features of all the salivary ductal components (myoepithelial, basal, and luminal cells) and supported that the tumor might arise from the salivary duct. Furthermore, it supports the hypothesis of multipotential stem cells as the origin for small cell undifferentiated carcinomas in salivary glands.

Immunohistochemistry of myoepithelial cells during development of the rat salivary glands.

Using a battery of monoclonal antibodies specific for rat proteins, immunohistochemistry was carried out on the developing myoepithelial cells (MECs) of the rat major salivary glands. The proteins examined were alpha-smooth muscle actin (alphaSMA), h1-calponin (calponin), keratin 14 (K14), beta subunit of S-100 protein (S-100beta), vimentin and glial fibrillary acidic protein (GFAP). The MECs exhibited immunoreactivity for alphaSMA, calponin and K14, but not that for S-100beta, vimentin and GFAP. Immunoreactivity for alphaSMA appeared in the MECs from the time when the microfilaments were initially deposited in these cells, i.e., at 20 days in utero in the sublingual and submandibular glands and at birth in the parotid gland. Calponin immunoreactivity was seen 1 day earlier than alphaSMA. The appearance was almost at the same time as the onset of the MEC differentiation in each gland. A small number of the MECs expressed weak K14 immunoreactivity from the time when the acinus-intercalated duct structure was established, i.e., at 21 days in utero in the sublingual gland, at 5 days after birth in the perotid gland and after 5 weeks post-natally in the submandibular gland. In addition, K14 immunoreactivity was observed in the basal cells of the striated and excretory ducts. The first appearance of K14 in these cells again coincided with the emergence of the duct system in each gland, i.e., at 20 days in utero in the sublingual gland, at 21 days in utero in the submandibular gland and at 3 days after birth in the parotid gland. Finally, the MECs in all the glands were found to redistribute as the acini matured. As the acini grew rapidly during the weaning period in the parotid and the sublingual glands, the MECs ceased to surround the acini. Thereafter, they disappeared from the acini in the parotid gland, whereas they reappeared in the sublingual gland. In the submandibular gland, the MECs were confined to the terminal tubules until 4 weeks after birth. Thereafter, the acini were established and invested by the MECs. In conclusion, immunohistochemistry of calponin and alphaSMA is a useful tool for identification of the MEC during its earliest differentiation, which has hitherto been possible only electron microscopically. In addition, it is suggested that the MEC is heterogeneous and the functionally differentiated MEC appears after weaning around acini of the mucous and seromucous glands.

Reduced expression of cyclin-dependent kinase inhibitor p27(Kip1) in oral malignant tumors.

p27(Kip1), a cyclin-dependent kinase (CDK) inhibitor, blocks progression from the G(1) to S phase by binding cyclin E-CDK2 and inhibiting their activities. We studied the expression of p27 in oral tumors by immunohistochemistry to determine whether lack of p27 plays a role in the development and progression of oral cancer. Reduced expression of p27 was detected in 86% of the squamous cell carcinomas and 95% of the mucoepidermoid carcinomas, respectively, while p27 expression was well preserved in the pleomorphic adenomas. The expression of p27 showed an inverse correlation with the expression of cyclin E in the squamous cell carcinomas and mucoepidermoid carcinomas. However, there was no relationship between clinicopathological parameters and p27 expression. These results suggest that the reduction of p27 protein may confer the development of oral squamous cell carcinoma and mucoepidermoid carcinoma partly through the increased expression of cyclin E.

Tumor necrosis factor alpha-mediated tumor regression by the in vivo transfer of genes into the artery that leads to tumors.

We report that tumor necrosis factor (TNF) alpha induced a strong antitumor immune reaction when it was produced in arteries leading to tumors by gene transfer in vivo. We used a mouse model carrying a sarcoma-180 tumor in the right footpad and injected the fusogenic liposomes encapsulating the human TNF-alpha gene into the right femoral artery. Under this condition, human TNF-alpha was detected only in the artery leading to the tumor and in the tumor. There was a significant regression in tumor growth when the TNF-alpha gene was delivered into the right femoral artery, with 4 of 11 mice completely cured. No regression was observed when the TNF-alpha gene was delivered into the left femoral artery or into the tumor or when the luciferase gene was administered. Tumor regression was inhibited by the injection of anti-TNF-alpha, anti-CD4, or anti-CD8 monoclonal antibody, and CD8+ T cells accumulated in the tumors of TNF-alpha-treated mice. These results suggest that TNF-alpha expressed locally in the arteries leading to tumors efficiently suppresses tumor growth through reinforcement of an antitumor immune reaction. The significance of this phenomenon for cancer gene therapy was discussed.

Immunohistochemistry of carbonic anhydrase isozymes VI and II during development of the rat salivary glands.

Secreted carbonic anhydrase (isozyme VI; CA VI) was localized by immunohistochemistry in the developing postnatal rat submandibular and parotid glands using a specific monoclonal antibody to the rat enzyme. CA VI immunostaining was not detectable in the glands before birth. In the submandibular gland, granular immunostaining for CA VI was detectable in several terminal tubule cells of 1-day-old rats. At 1 week, the CA VI-positive cells were located at the periphery of the terminal tubules and appeared to be budding off the tubules. These cellular buds gradually increased, and, by 4 weeks, formed acini. CA VI was also detected in the duct lumen from day 1. The immunostaining in the parotid gland was detected sporadically in the acinar cells at 2 or 3 weeks. By 4 weeks, when the gland was almost indistinguishable from the adult one, the number of positive acinar cells had increased. Their number, however, was far smaller than in the adult gland, and the enzyme could not be detected in the duct lumen. CA II was also localized using specific antibodies to the rat isozyme. CA II was detectable in the inter- and intralobular striated ducts at 2 weeks after birth in the submandibular gland and at 3 weeks in the parotid gland. These results suggest that CA VI is secreted into saliva from soon after birth and that CA II appears in parallel with the functional maturation of the ducts. In addition, CA II was transiently expressed by the cellular buds of the submandibular gland at 2 and 3 weeks.

Clear cell odontogenic tumour: a case with induction of dentin-like structures?

A case of clear cell odontogenic tumour, which occurred centrally in the mandible of a 56-year-old Japanese woman, is reported with its histochemical, immunohistochemical and ultrastructural findings. Histologically, the tumour nests were composed of large glycogen-rich clear cells and small non-clear polygonal cells and were separated by thin mature fibrous connective tissue septae. Immunohistochemically, both types of tumour cells showed positive expression of various cytokeratins, in particular cytokeratin 19, and of epithelial membrane antigen. Eosinophilic hyaline deposits and possible dentin-like structures were occasionally formed in contact with the epithelial nests and are regarded as indicative of the epithelial-mesenchymal inductive capacity of this tumour. The aggressive nature of the present tumour was assumed through its invasive growth pattern and occasional mitotic figures. Although it was diagnosed as clear cell odontogenic tumour according to the present WHO classification, the patient must be followed carefully because of its probable malignant nature.

Ectopic expression of N-acetylglucosaminyltransferase III in transgenic hepatocytes disrupts apolipoprotein B secretion and induces aberrant cellular morphology with lipid storage.

N-Acetylglucosaminyltransferase III (GnT-III) produces "bisecting-GlcNAc" and regulates the branching of N-glycans. GnT-III activity is elevated during hepatocarcinogenesis, which is in contrast to the undetectable level found in normal hepatocytes. To determine the biological significance of GnT-III in hepatocytes, transgenic mice that specifically express GnT-III in the liver were established and characterized. The transgenic hepatocytes had a swollen oval-like morphology, with many lipid droplets. Apolipoprotein B, which contained increased level of bisecting-GlcNAc accumulated in the transgenic hepatocytes. In the transgenic serum, triglycerides, the beta- and pre-beta-lipoprotein fractions, and apolipoprotein B100 were significantly decreased, compared with levels in nontransgenic serum. These abnormal phenotypes were more prominent in the mice with more copies of the transgene and a resulting high GnT-III activity. We demonstrate that aberrant glycosylation, as the direct result of the formation of bisecting-GlcNAc, disrupts the function of apolipoprotein B, leading to the generation of fatty liver. This observation suggests a novel mechanism for the pathogenesis of fatty liver.

Immunohistochemical detection of prostaglandins E2, F2 alpha, and 6-keto-prostaglandin F1 alpha in experimentally induced periapical inflammatory lesions in rats.

The immunohistochemical localization of prostaglandin (PG) E2, PGF2 alpha, and 6-keto-PGF1 alpha (a stable metabolite of PGI2) was demonstrated in rat periapical inflammatory lesions induced by opening the pulp chamber. Two wk postoperatively, suppurative periapical lesions were formed, and active bone resorption was seen surrounding these lesions. Immunohistochemical examination showed that macrophages infiltrating in inflammatory tissue were positively stained for the examined PGs. In some lesions, wherein acute inflammatory changes subsided and proliferation of fibroblasts started, the fibroblasts were positively stained for 6-keto-PGF1 alpha. Osteocytes and osteoblasts were also positive for 6-keto-PGF1 alpha not only in experimental animals, but also in untreated animals. However the staining intensity of the PG in these cells was higher in periapical lesions than in normal condition. These findings suggested that the cellular sources of the PGs in the periapical lesions are mainly macrophages and fibroblasts, and that the PGs produced by these cells, and possibly osteoblast and osteocytes, may contribute to the osteolytic resorption of periapical lesions.

Immunohistochemical demonstration of prostaglandins E2, F2 alpha, and 6-keto-prostaglandin F1 alpha in rat dental pulp with experimentally induced inflammation.

Using formalin-fixed and EDTA-decalcified cryostat sections, the immunohistochemical localization of prostaglandin (PG) E2, PGF2 alpha, and 6-keto-PGF1 alpha (a stable metabolite of PGI2) was examined in normal rat and inflamed dental pulp. Inflammation was induced by opening the pulp chamber. There was no immunoreactivity for prostaglandins in normal dental pulp, whereas positivities for PGE2, PGF2 alpha, and 6-keto-PGF1 alpha were demonstrated in the cytoplasm of macrophages and endothelial cells in the inflamed dental pulp. In addition to these cells, numerous pulp cells and odontoblasts existing in the inflamed pulp and its apical noninflamed area also were intensely stained for PGF2 alpha. Such an area with positive cells gradually extended in an apical direction with the progression of inflammation. These findings suggested that PG production from these host cells is involved in development of inflammation of rat dental pulp.

Immunohistochemical localization of carbonic anhydrase isozyme II in rat incisor epithelial cells at various stages of amelogenesis.

Carbonic anhydrase II (CAII) was purified from erythrocytes of male Sprague-Dawley rats, and its localization in rat maxillary incisor epithelial cells at various stages of amelogenesis was studied by means of immunoperoxidase staining using a rat CAII-specific monoclonal antibody. In the most apical portion of the incisor, some CAII immunoreactivity was localized in the outer or inner dental epithelium near the apical loop (i.e., the multiple layer of the outer dental epithelium and the posterior portion of ameloblasts facing the pulp). Immunoreactivity disappeared largely during the presecretory and secretory stages. CAII immunoreactivity appeared suddenly in ameloblasts during the transitional stage between enamel secretion and maturation. Immunoreactivity became intense in both ameloblasts and papillary cells during enamel maturation; the intracellular distribution of CAII was in the cytosol. The CAII signal in these cells was constant until the end of the maturation stage. These findings support the notion that the ameloblasts and papillary cells change into ion transport epithelial cells from the secretory to the maturation stage and that CAII in these cells plays an important role in the regulation of pH.

Transfection of wild-type TP53 induces differentiation in human gingival carcinoma cells.

We investigated the effects of transfection of wild-type TP53 on the growth properties of a human gingival carcinoma cell line, KOSC-3, in which the TP53 gene is mutated at codon 248 and overexpressed. The wild-type TP53 expression plasmid, pCDM8-p53/neo and the control plasmid, pCDM8/neo, were each stably transfected into KOSC-3 cells by using the calcium phosphate method. The number of G418-resistant colonies from wild-type TP53-transfected cells was approximately half that from plasmid controls. Exogenous wild-type TP53 transcripts were identified in four of the 20 G418-resistant clones analysed by reverse transcription PCR. Although the growth rates of the wild-type TP53+ clones did not drastically change during log phase, their saturation density was significantly reduced. The wild-type TP53+ cells were morphologically flat and enlarged when cultured in vitro, and were less able to form colonies in soft agar. In nude mice, the wild-type TP53+ clones formed subcutaneous tumours with conspicuous keratinisation and notable cell death that was not manifested in the parental and plasmid control cells. These findings indicate that the wild-type TP53 gene, even when it coexists with a mutated form, may function as a growth suppressor and differentiation inducer under restricted conditions in gingival squamous cell carcinoma.

Immunohistochemical observations on a possible ameloblastic fibro-odontoma.

An ameloblastic fibro-odontoma which occurred in the mandible of a 3-year-old Japanese boy is reported together with immunohistochemical findings. Histologically, the tumor consisted of an ameloblastic fibroma-like area and some typical complex odontoma-like areas. The epithelial islands in the ameloblastic fibroma-like area showed different developmental stages of the epithelial-connective tissue interface. Immunohistochemical examination revealed that all epithelial components in the ameloblastic fibroma-like area showed expression of CK 8, CKs 13, 16, CK 14, CK 18 and CK 19, and coexpression of these cytokeratins and vimentin. These findings suggest that even the epithelial component without obvious epithelial-mesenchymal induction showed the final cell differentiation of the enamel organ with the potential for epithelial-mesenchymal induction.

Bisecting N-acetylglucosamine on K562 cells suppresses natural killer cytotoxicity and promotes spleen colonization.

beta 1-4 N-acetylglucosaminyltransferase (GnT-III) catalyzes the formation of bisecting N-acetylglucosamine (GlcNAc) in the biosynthesis of N-linked oligosaccharides. To examine the effect of bisecting GlcNAc on the natural killer (NK) cytotoxicity, the GnT-111 gene was introduced into NK-sensitive K562 cells that have no detectable GnT-III activity. We obtained three clones stably expressing high GnT-III (positive transfectants). Introduction of the GnT-III gene resulted in an increase of bisecting GlcNAc and a decrease of external sialic acid as well as tri- and tetraantennary sugars, as judged by flow cytometry. Compared to controls, the NK cytotoxicity was completely blocked against positive transfectants. The binding of effector cells to positive transfectants was also decreased. After s.c. injection into nude mice, positive transfectants produced spleen colonization, although no spleen lesions were formed by control cells. In nude mice depleted of NK cells by anti-asialo GM1 antibody, both positive transfectants and controls produced spleen colonization equally. These results indicate that K562 cells expressing GnT-III are resistant to NK cytotoxicity, resulting in spleen colonization in nude mice.