Appropriate glycosylation of the fms gene product is a prerequisite for its transforming potency.

Processing inhibitors of N-linked glycans were used to determine whether correct glycosylation of the oncogene product gp140v-fms, encoded by the McDonough strain of feline sarcoma virus (SM-FeSV), is required to maintain the oncogenic properties of v-fms. SM-FeSV-transformed cells treated with the glucosidase-I inhibitors N-methyldeoxynojirimycin (MdN) or castanospermine synthesized predominantly a gp125v-fms species which had a normal half life. The molecule was transported to the plasma membrane and exhibited normal kinase activity as determined by autophosphorylation. However, although no significant change in cell morphology of the SM-FeSV-transformed cells was observed in the presence of castanospermine, growth of these cells became strictly serum-dependent. In addition, growth in soft agar was drastically retarded despite the presence of 10% calf serum, indicating that the transformed properties of the cells were altered. In contrast, swainsonine, an inhibitor of the processing alpha-mannosidase-II, had no effect. Cells transformed by the Snyder Theilen strain of FeSV were used to demonstrate that the altered proliferative properties were directly linked to the modified structure of the fms gene product. Our data suggest that the extracellular domain of gp140v-fms plays a role in regulating cell proliferation.     

Transformation by the oncogene v-fms: The effects of castanospermine on transformation-related parameters

The effects of castanospermine on various parameters associated with transformation were examined in cells expressing the viral oncogene v-fms. Fischer rat embryo (FRE) cells transformed by the oncogene v-fms and grown in the presence of castanospermine reverted to a more normal cell morphology and accumulated fms protein within the endoplasmic reticulum. Treated cells attained contact inhibition of cell growth at a much lower cell density compared to the untreated controls. No effect of castanospermine on cell growth was observed for FRE cells transformed by a different oncogene v-fgr. Castanospermine-treated SM-FRE (v-fms transformed) cells reexpressed extracellular matrix fibronectin and exhibited an extensive actin-containing cytoskeleton similar to that of normal nontransformed FRE cells. Castanospermine treatment of SM-FRE cells resulted in a sixfold decrease in [3H]deoxyglucose uptake compared to that of the nonreverted SM-FRE cells. Again, no effect was observed in FRE cells transformed by the oncogene v-fgr (GR-FRE). These results further characterize the reversion caused by castanospermine and indicate that cell surface expression coordinately controls anchorage independent growth, cell morphology, contact inhibition of growth, and hexose uptake.     

Analysis of functional domains of the v-fms-encoded protein of Susan McDonough strain feline sarcoma virus by linker insertion mutagenesis.

The Susan McDonough strain of feline sarcoma virus contains an oncogene, v-fms, which is capable of transforming fibroblasts in vitro. The mature protein product of the v-fms gene (gp140fms) is found on the surface of transformed cells; this glycoprotein has external, transmembrane, and cytoplasmic domains. To assess the functional role of these domains in transformation, we constructed a series of nine linker insertion mutations throughout the v-fms gene by using a dodecameric BamHI linker. The biological effects of these mutations on the function and intracellular localization of v-fms-encoded proteins were determined by transfecting the mutated DNA into Rat-2 cells. Most of the mutations within the external domain of the v-fms-encoded protein eliminated focus formation on Rat-2 cells; three of these mutations interfered with the glycosylation of the v-fms protein and interfered with formation of the mature gp140fms. One mutation in the external domain led to cell surface expression of v-fms protein even in the absence of complete glycosylational processing. Cell surface expression of mutated v-fms protein is probably necessary, but is not sufficient, for cell transformation since mutant v-fms protein was found on the surface of several nontransformed cell lines. Mutations that were introduced within the external domain had little effect on in vitro kinase activity, whereas mutations within the cytoplasmic domain all had strong inhibitory effects on this activity.     

Cell surface expression of v-fms-coded glycoproteins is required for transformation.

The viral oncogene v-fms encodes a transforming glycoprotein with in vitro tyrosine-specific protein kinase activity. Although most v-fms-coded molecules remain internally sequestered in transformed cells, a minor population of molecules is transported to the cell surface. An engineered deletion mutant lacking 348 base pairs of the 3.0-kilobase-pair v-fms gene encoded a polypeptide that was 15 kilodaltons smaller than the wild-type v-fms gene product. The in-frame deletion of 116 amino acids was adjacent to the transmembrane anchor peptide located near the middle of the predicted protein sequence and 432 amino acids from the carboxyl terminus. The mutant polypeptide acquired N-linked oligosaccharide chains, was proteolytically processed in a manner similar to the wild-type glycoprotein, and exhibited an associated tyrosine-specific protein kinase activity in vitro. However, the N-linked oligosaccharides of the mutant glycoprotein were not processed to complex carbohydrate chains, and the glycoprotein was not detected at the cell surface. Cells expressing high levels of the mutant glycoprotein did not undergo morphological transformation and did not form colonies in semisolid medium. The transforming activity of the v-fms gene product therefore appears to be mediated through target molecules on the plasma membrane.     

Transformation by the v-fms oncogene product: an analog of the CSF-1 receptor

The product of the c-fms proto-oncogene is related to, and possibly identical with, the receptor for the macrophage colony-stimulating factor, M-CSF (CSF-1). Unlike the product of the v-erbB oncogene, which is a truncated version of the EGF receptor, the glycoprotein encoded by the v-fms oncogene retains an intact extracellular ligand-binding domain so that cells transformed by v-fms express CSF-1 receptors at their surface. Although fibroblasts susceptible to transformation by v-fms generally produce CSF-1, v-fms-mediated transformation does not depend on an exogenous source of the growth factor, and neutralizing antibodies to CSF-1 do not affect the transformed phenotype. An alteration of the v-fms gene product at its extreme carboxyl-terminus represents the major structural difference between it and the c-fms-coded glycoprotein and may affect the tyrosine kinase activity of the v-fms-coded receptor. Consistent with this interpretation, tyrosine phosphorylation of the v-fms products in membranes was observed in the absence of CSF-1 and was not enhanced by addition of the murine growth factor. Cells transformed by v-fms have a constitutively elevated specific activity of a guanine nucleotide-dependent, phosphatidylinositol-4,5-diphosphate-specific phospholipase C. We speculate that the tyrosine kinase activity of the v-fms/c-fms gene products may be coupled to this phospholipase C, possibly through a G regulatory protein, thereby increasing phosphatidylinositol turnover and generating the intracellular second messengers diacylglycerol and inositol triphosphate.     

The amino-terminal domain of the v-fms oncogene product includes a functional signal peptide that directs synthesis of a transforming glycoprotein in the absence of feline leukemia virus gag sequences.

The nucleotide sequence of a 5' segment of the human genomic c-fms proto-oncogene suggested that recombination between feline leukemia virus and feline c-fms sequences might have occurred in a region encoding the 5' untranslated portion of c-fms mRNA. The polyprotein precursor gP180gag-fms encoded by the McDonough strain of feline sarcoma virus was therefore predicted to contain 34 v-fms-coded amino acids derived from sequences of the c-fms gene that are not ordinarily translated from the proto-oncogene mRNA. The (gP180gag-fms) polyprotein was cotranslationally cleaved near the gag-fms junction to remove its gag gene-coded portion. Determination of the amino-terminal sequence of the resulting v-fms-coded glycoprotein, gp120v-fms, showed that the site of proteolysis corresponded to a predicted signal peptidase cleavage site within the c-fms gene product. Together, these analyses suggested that the linked gag sequences may not be necessary for expression of a biologically active v-fms gene product. The gag-fms sequences of feline sarcoma virus strain McDonough and the v-fms sequences alone were inserted into a murine retroviral vector containing a neomycin resistance gene. Both constructs were biologically active when transfected into NIH 3T3 cells and produced morphologically transformed foci at equivalent efficiencies. When transfected into a cell line (psi 2) expressing complementary viral gene functions, G418-resistant (Neor) cells containing either of these vector DNAs produced high titers of transforming viruses. Analysis of proteins produced in cells containing the vector lacking gag gene sequences showed that gP180gag-fms was not synthesized, whereas normal levels of both immature gp120v-fms and mature gp140v-fms were detected. The glycoprotein was efficiently transported to the cell surface, and it retained wild-type tyrosine kinase activity. We conclude that a cryptic hydrophobic signal peptide sequence in v-fms was unmasked by gag deletion, thereby allowing the correct orientation and transport of the v-fms gene product within membranous organelles. It seems likely that the proteolytic cleavage of gP180gag-fms is mediated by signal peptidase and that the amino termini of gp140v-fms and the c-fms gene product are identical.     

Autocrine induction of major histocompatibility complex class I antigen expression results from induction of beta interferon in oncogene-transformed BALB/c-3T3 cells.

By varying growth conditions, we identified a novel mechanism of autocrine regulation of major histocompatibility complex (MHC) class I gene expression by induction of beta interferon gene expression in transformed BALB/c-3T3 cells. Low-serum conditions enhanced MHC class I antigen expression in v-rasKi- and v-mos-transformed BALB/c-3T3 cells but not in untransformed BALB/c-3T3 cells. Transformed and untransformed cells grown under standard serum conditions (10% bovine calf serum) expressed similar cell surface levels of MHC class I antigens. However, low-serum conditions (0.5% bovine calf serum) induced four- to ninefold increases in cell surface levels of MHC class I antigens in both v-rasKi- and v-mos-transformed cells but not in untransformed cells. These increases in MHC class I gene expression were seen at both the mRNA and cell surface protein levels and involved not only the heavy-chain component of the class I antigens but also beta 2 microglobulin. Beta 1 interferon mRNA and beta interferon-inducible 2',5'-oligoadenylate synthetase mRNA were induced by growth under low-serum conditions in transformed BALB/c-3T3 cells, and antibodies to beta interferon blocked the induction of MHC class I antigen expression by serum deprivation in these cells. These results demonstrate that growth under low-serum conditions leads to induction of beta interferon expression in oncogene-transformed cells which then directly mediates autocrine enhancement of MHC class I gene expression.     

Gene expression and internalization following vector adsorption to immobilized proteins: dependence on protein identity and density

BACKGROUND: Gene delivery by non-specific adsorption of non-viral vectors to protein-coated surfaces can reduce the amount of DNA required, and also increase transgene expression and the number of cells expressing the transgene. The protein on the surface mediates cell adhesion and vector immobilization, and functions to colocalize the two to enhance gene delivery. This report investigates the mechanism and specificity by which the protein coating enhances gene transfer, and determines if the protein coating targets the vector for internalization by a specific pathway. METHODS: Proteins (FBS, BSA, fibronectin, collagen I, and laminin) were dried onto culture dishes, followed by PEI/DNA complex adsorption for surface delivery. Reporter genes were employed to characterize transfection as a function of the protein identity and density. Vector immobilization was measured using radiolabeled plasmid, and internalization was quantified in the presence and absence of the endocytosis inhibitors chlorpromazine and genistein. RESULTS: Fibronectin coating yielded the greatest expression for PEI/DNA polyplexes, with maximal expression at intermediate protein densities. Expression in control studies with bolus delivery was independent of the protein identity. Substrate binding was independent of the protein identity; however, internalization was greatest on surfaces coated with fibronectin and collagen I. Inhibition of caveolae-mediated endocytosis reduced gene expression more than clathrin-mediated endocytosis. Similarly, inhibition of caveolae-mediated endocytosis significantly reduced the intracellular levels of DNA. CONCLUSIONS: Fibronectin at intermediate densities mediated the highest levels of transgene expression, potentially by targeting internalization through caveolae-mediated endocytosis. Substrate modifications, such as the identity and density of proteins, provide an opportunity for modification of biomaterials for enhancing gene expression.     

DNase X is a glycosylphosphatidylinositol-anchored membrane enzyme that provides a barrier to endocytosis-mediated transfer of a foreign gene

DNase X is the first mammalian DNase to be isolated that is homologous to DNase I. In this study, we have examined its function using a novel monoclonal antibody and showed it to be expressed on the cell surface as a glycosylphosphatidylinositolanchored membrane protein. High level expression was observed in human muscular tissues and in myotubes obtained in vitro from RD rhabdomyosarcoma cells. We observed that RD myotubes incorporated a foreign gene, lacZ, by endocytosis but that expression of the encoded coding product, beta-galactosidase, was strongly inhibited. Overexpression of DNase X inhibited endocytosis-mediated gene transfer, whereas knockdown of DNase X with small interfering RNA had the opposite effect. These results reveal that DNase X provides a cell surface barrier to endocytosis-mediated gene transfer.     

Effects of inhibitors of glycoprotein processing on the synthesis and biological activity of the erb B oncogene.

Three glycoprotein-processing inhibitors were used to resolve whether correct glycosylation was required for the oncogenic activity of erb B. The two glucosidase-I inhibitors, 1-deoxynojirimycin and 2,5-dihydroxymethyl 3,4-dihydroxypyrrolidine, arrested processing of v-erb B at the immature 68-kd form whereas, in the presence of the alpha-mannosidase-II inhibitor (swainsonine), cells synthesised an abnormally processed 70-kd form of v-erb B. Transport of incorrectly processed v-erb B to the cell surface was, however, unaffected, suggesting that correct processing is not a prerequisite for intracellular routing of v-erb B. Two systems were used to assess whether incorrectly processed erb B could maintain the transformed state. The first asked whether inhibitor treatment would release temperature-sensitive avian erythroblastosis virus (AEV) transformed erythroblasts kept at the viral permissive temperature from the erb B-induced block in differentiation, as seen when cells are normally shifted to the non-permissive temperature. The second tested the ability of AEV-transformed fibroblasts to grow in soft agar. In both systems, all three processing inhibitors did not alter the transformed phenotype suggesting that correct carbohydrate processing is not required for the transforming activity of erb B. In addition, none of the three processing inhibitors were found to have any effect on the normal maturation of bone marrow CFU-E or induced differentiation of temperature-sensitive AEV-transformed erythroblasts.     

Reassessment of the v-fms sequence: threonine phosphorylation of the COOH-terminal domain.

The v-fms oncogene product of the McDonough strain of feline sarcoma virus is a member of the receptor tyrosine kinase family. Its cellular counterpart, the c-fms product, is the receptor for colony-stimulating factor 1 (CSF-1) of macrophages. We have reanalyzed the v-fms gene by direct sequencing of a biologically active clone. An additional A nucleotide was detected in position 2810 of the published v-fms sequence. The frameshift changed the COOH-terminal sequence of the v-fms protein from -R-937-G-P-P-L-COOH to -Q-937-R-T-P-P-V-A-R-COOH. Antibodies against a synthetic peptide representing this new sequence precipitated the v-fms proteins from transformed NRK cells as well as from feline sarcoma virus (McDonough)-infected feline fibroblasts. We show by tryptic peptide mapping that threonine 939 present in the new sequence is phosphorylated by a yet unknown serine/threonine kinase in vivo. In chicken fibroblasts expressing the v-fms gene, this phosphorylation clearly depended on the addition of exogenous CSF-1. Furthermore, addition of CSF-1 appeared to activate the serine/threonine kinase, as judged by phosphorylation of the synthetic peptide QRTPPVAR.     

The host cell chaperone Hsp90 is essential for translocation of the binary Clostridium botulinum C2 toxin into the cytosol

Clostridium botulinum C2 toxin is the prototype of the binary actin-ADP-ribosylating toxins and consists of the binding component C2II and the enzyme component C2I. The activated binding component C2IIa forms heptamers, which bind to carbohydrates on the cell surface and interact with the enzyme component C2I. This toxin complex is taken up by receptor-mediated endocytosis. In acidic endosomes, heptameric C2IIa forms pores and mediates the translocation of C2I into the cytosol. We report that the heat shock protein (Hsp) 90-specific inhibitors, geldanamycin or radicicol, block intoxication of Vero cells, rat astrocytes, and HeLa cells by C2 toxin. ADP-ribosylation of actin in the cytosol of toxin-treated cells revealed that less active C2I was translocated into the cytosol after treatment with Hsp90 inhibitors. Under control conditions, C2I was localized in the cytosol of toxin-treated rat astrocytes, whereas geldanamycin blocked the cytosolic distribution of C2I. At low extracellular pH (pH 4.5), which allows the direct translocation of C2I via C2IIa heptamers across the cell membrane into the cytosol, Hsp90 inhibitors retarded intoxication by C2I. Geldanamycin did not affect toxin binding, endocytosis, and pore formation by C2IIa. The ADP-ribosyltransferase activity of C2I was not affected by Hsp90 inhibitors in vitro. The cytotoxic actions of the actin-ADP-ribosylating Clostridium perfringens iota toxin and the Rho-ADP-ribosylating C2-C3 fusion toxin was similarly blocked by Hsp90 inhibitors. In contrast, radicicol and geldanamycin had no effect on anthrax lethal toxin-induced cytotoxicity of J774-A1 macrophage-like cells or on cytotoxic effects of the glucosylating Clostridium difficile toxin B in Vero cells. The data indicate that Hsp90 is essential for the membrane translocation of ADP-ribosylating toxins delivered by C2II.     

Gene Transfer by Means of Lipo- and Polyplexes: Role of Clathrin and Caveolae-Mediated Endocytosis

In this paper we address the contribution of different endocytic pathways to the intracellular uptake and processing of differently sized latex particles and of plasmid DNA complexes by means of fluorescence microscopy and FACS analysis. By using a number of specific inhibitors of either clathrin-dependent or caveolae-dependent endocytosis we were able to discriminate between these two pathways. Latex particles smaller than 200 nm were internalized exclusively by clathrin-mediated endocytosis, whereas larger particles entered the cells via a caveolae-dependent pathway.

The route of uptake of plasmid DNA complexes appears strongly dependent on the nature of the complexes. Thus, lipoplexes containing the cationic lipid DOTAP, were exclusively internalized by a clathrin-dependent mechanism, while polyplexes prepared from the cationic polymer polyethyleneimine (PEI) were internalized in roughly equal proportions by both pathways. Upon incubation of cells with lipoplexes containing the luciferase gene abundant luciferase expression was observed, which was effectively blocked by inhibitors of clathrin-dependent endocytosis but not by inhibitors of the caveolae-dependent uptake mechanism. By contrast, luciferase transfection of the cells with polyplexes was unaffected by inhibition of clathrin-mediated endocytosis, but was nearly completely blocked by inhibitors interfering with the caveolae pathway. The results are discussed with respect to possible differences in the mechanism by which plasmid DNA is released from lipoplexes and polyplexes into the cytosol and to the role of size in the uptake and processing of the complexes. Our data suggest that improvement of non-viral gene transfection could very much benefit from controlling particle size, which would allow targeting of particle internalization via a non-degradative pathway, involving caveolae-mediated endocytosis.     

Gene transfer by means of lipo- and polyplexes: role of clathrin and caveolae-mediated endocytosis

In this paper we address the contribution of different endocytic pathways to the intracellular uptake and processing of differently sized latex particles and of plasmid DNA complexes by means of fluorescence microscopy and FACS analysis. By using a number of specific inhibitors of either clathrin-dependent or caveolae-dependent endocytosis we were able to discriminate between these two pathways. Latex particles smaller than 200 nm were internalized exclusively by clathrin-mediated endocytosis, whereas larger particles entered the cells via a caveolae-dependent pathway.The route of uptake of plasmid DNA complexes appears strongly dependent on the nature of the complexes. Thus, lipoplexes containing the cationic lipid DOTAP, were exclusively internalized by a clathrin-dependent mechanism, while polyplexes prepared from the cationic polymer polyethyleneimine (PEI) were internalized in roughly equal proportions by both pathways. Upon incubation of cells with lipoplexes containing the luciferase gene abundant luciferase expression was observed, which was effectively blocked by inhibitors of clathrin-dependent endocytosis but not by inhibitors of the caveolae-dependent uptake mechanism. By contrast, luciferase transfection of the cells with polyplexes was unaffected by inhibition of clathrin-mediated endocytosis, but was nearly completely blocked by inhibitors interfering with the caveolae pathway. The results are discussed with respect to possible differences in the mechanism by which plasmid DNA is released from lipoplexes and polyplexes into the cytosol and to the role of size in the uptake and processing of the complexes. Our data suggest that improvement of non-viral gene transfection could very much benefit from controlling particle size, which would allow targeting of particle internalization via a non-degradative pathway, involving caveolae-mediated endocytosis.     

Flat revertants derived from Kirsten murine sarcoma virus-transformed cells produce transforming growth factors

Two flat cellular revertant cell lines, F-2 and C-11, which were originally selected from the DT line of Kirsten murine sarcoma virus (Ki-MuSV)-transformed NIH/3T3 cells, were examined for the production of transforming growth factors (TGFs). The revertant cells fail to grow in semisolid medium as colonies and exhibit a markedly reduced level of tumorigenicity in nude mice, although they are known to express high levels of p21ras, the product of the Kirsten sarcoma virus oncogene, ras, and they contain a rescuable transforming virus. TGF activity associated with the transformed, revertant, and non-transformed cell lines was measured by the ability of concentrated conditioned medium (CM) from these cells to induce normal rat kidney (NRK) and NIH/3T3 cells to form colonies in semisolid agar suspension cultures and to inhibit the binding of 125I epidermal growth factor (EGF) to specific cell surface receptors. CM from the transformed DT cells and from both the F-2 and C-11 revertants contains TGF activity, in contrast to CM obtained from normal NIH/3T3 cells. Furthermore, unlike NIH/3T3 cells, neither the DT nor the revertant cells were able to bind 125I EGF. All four cell lines were able to proliferate in serum-free medium supplemented with transferrin, insulin, EGF, and Pedersen fetuin. However, in basal medium lacking these growth factors, only DT cells and, to a lesser extent, the revertant cells were able to grow. These results suggest that the F-2 and C-11 revertants fail to exhibit all of the properties associated with transformation because the series of events leading to the transformed phenotype is blocked at a point(s) distal both to the expression of the p21 ras gene product and also to the production of TGFs and that the production of TGFs may be necessary but not sufficient for maintaining the transformed state.     

A recessive cellular mutation in v-fes-transformed mink cells restores contact inhibition and anchorage-dependent growth.

A contact-inhibited revertant of mink cells transformed by the Gardner-Arnstein strain of feline sarcoma virus was isolated by fluorescence-activated sorting of cells stained with the mitochondria-specific dye rhodamine 123. The revertant cell line exhibited a decrease in its proliferative rate and saturation density and a complete loss of its capacity for anchorage-independent growth, but it remained tumorigenic when inoculated into nude mice. The revertant cells retained a rescuable Gardner-Arnstein feline sarcoma provirus, expressed high levels of the v-fes oncogene product and its associated tyrosine kinase activity, manifested elevated levels of phosphotyrosine-containing cellular proteins similar to those observed in v-fes-transformed cells, and were refractory to retransformation by retroviruses containing the v-fes, v-fms, and v-ras oncogenes. Fusion of the revertant and parental cells generated somatic cell hybrids which formed colonies in semisolid medium, indicating that the block in transformation was recessive. These data together with the observation that the revertant phenotype is unstable in continuous culture suggest that the loss of transformation is due to the presence of limiting quantities of a gene product which functions downstream of the v-fes-coded kinase in the mitogenic pathway.     

Identification and characterization of a gene and protein required for glycosylation in the yeast Golgi

The MNN2 gene of Saccharomyces cerevisiae has been cloned by complementation of the mnn2 mutant phenotype scored by a change in cell surface carbohydrate structure resulting from a lack of alpha 1----2-mannose branching in the outer chain. The gene was subcloned as a 3 kb DNA fragment that integrated at the MNN2 locus, and a gene disruption yielded the mnn2 phenotype. A lacZ-MNN2 gene fusion protein, produced in Escherichia coli, was used to raise a specific antiserum that recognized a 65 kD wild-type yeast protein. This MNN2 gene product lacks N-linked carbohydrate but appears to be an integral membrane protein. Overproduction of MNN2p does not enhance the alpha 1----2-mannosyltransferase activity of yeast cells. The results suggest that MNN2p is a Golgi-associated protein that is involved in mannoprotein sorting rather than glycosylation.     

CD99 regulates the transport of MHC class I molecules from the Golgi complex to the cell surface

The down-regulation of surface expression of MHC class I molecules has recently been reported in the CD99-deficient lymphoblastoid B cell line displaying the characteristics of Hodgkin's and Reed-Sternberg phenotype. Here, we demonstrate that the reduction of MHC class I molecules on the cell surface is primarily due to a defect in the transport from the Golgi complex to the plasma membrane. Loss of CD99 did not affect the steady-state expression levels of mRNA and protein of MHC class I molecules. In addition, the assembly of MHC class I molecules and the transport from the endoplasmic reticulum to the cis-Golgi occurred normally in the CD99-deficient cells, and no difference was detected between the CD99-deficient and the control cells in the pattern and degree of endocytosis. Instead, the CD99-deficient cells displayed the delayed transport of newly synthesized MHC class I molecules to the plasma membrane, thus causing accumulation of the molecules within the cells. The accumulated MHC class I molecules in the CD99-deficient cells were colocalized with alpha-mannosidase II and gamma-adaptin in the Golgi compartment. These results suggest that CD99 may be associated with the post-Golgi trafficking machinery by regulating the transport to the plasma membrane rather than the endocytosis of surface MHC class I molecules, providing a novel mechanism of MHC class I down-regulation for immune escape.