Primary cytomegalovirus (CMV) infection and reactivation of persistent CMV are associated with significant morbidity and mortality in immunocompromised individuals. Although recovery from CMV disease is correlated with the development of CMV-specific cytotoxic T lymphocytes (CTL), the major viral target antigens to which the response is directed are ill-defined, though they may comprise viral structural elements. We now identify the CMV matrix protein pp65 as a significant target antigen for CD8+ class I major histocompatibility complex (MHC)-restricted CMV-specific CTL derived from the peripheral blood of four of five latently infected individuals. CMV-specific CTL recognition of pp65 on target cells occurs prior to the onset of viral gene expression and persists throughout the duration of the replicative cycle. Recognition in the absence of viral gene expression suggests that abundant viral protein enters the normal trafficking pathway upon viral penetration and is readily made available to MHC molecules for presentation at the cell surface. Thus pp65 specific CTL may represent an important effector population for early control and limitation of CMV infection and disease. The observation that CMV-specific CTL can be induced in vitro using peptide fragments derived from pp65 supports the future use and manipulation of this and similar effector populations in a clinical setting.

A direct method for the determination of N-linked glycosylation sites in highly glycosylated proteins is described. Carcinoembryonic antigen (CEA) and a nonspecific crossreacting antigen (NCA) were chemically degycosylated, and peptide maps were prepared by reverse-phase HPLC. The peptides were sequenced on a gas-phase microsequencer, and glycosylation sites were identified as the phenylthiohydantoin derivative of N-acetylglucosaminylasparagine. The sequences were confirmed by fast atom bombardment mass spectrometry. Highly homologous, extended amino-terminal sequences were determined for CEA and two NCAs, NCA-95 and NCA-55. Cysteine (6) and NCA~7) are identical but differ from the CEA sequence (15) in having alanine rather than valine at position 21. The further structural analysis of CEA has been hampered by the high degree of glycosylation, which renders it extremely resistant to proteolytic cleavage. Shively et al. (16) digested CEA with trypsin in the presence of Triton X-100; however, extended sequences of the resulting peptides were not obtained, probably due to the presence of carbohydrate in these peptides. CEA has also been deglycosylated with anhydrous HF (17). Preliminary studies showed that deglycosylated CEA was more amenable than native CEA to proteolytic cleavage and subsequent sequence analysis (18).In the present study, we describe a general method for the structural analysis of highly glycosylated proteins and apply this method to CEA, NCA-95, and NCA-55. This method, which includes chemical deglycosylation, peptide mapping by reverse-phase HPLC, gas-phase microsequence analysis, and fast atom bombardment (FAB) mass spectrometry (MS), allowed the direct determination of numerous N-linked glycosylation sites in CEA and NCA-95 peptides. We also report highly homologous, extended amino-terminal sequences for CEA, NCA-95, and NCA-55 and homologous cysteine-containing sequences for CEA and NCA-95. The CEA cysteine-containing sequences show internal homology, suggesting that CEA evolved by a series of gene duplication events. In addition, CEA cysteine-containing sequences are homologous to sequences found in several members of the immnunoglobulin supergene family (19). MATERIALS AND METHODSPurification of Antigens. CEA, NCA-95, and NCA-55 were isolated from liver metastases of colon tumors and purified as described for CEA (20). Final purifications of NCA-95 and NCA-55 were achieved by reverse-phase HPLC (R.J.P. and J.E.S., unpublished data).Deglycosylation and Peptide Mapping. CEA and NCA-95 were chemically deglycosylated with trifluoromethanesulfonic acid (TFMSA):anisole, 2:1 (vol/vol), as described (21) and desalted by dialyzing against 10% (vol/vol) aqueous Abbreviations: CEA, carcinoembryonic antigen; NCA, nonspecific crossreacting antigen; FAB, fast atom bombardment; >PhNCS, phenylthiohydantoin; GlcNAc, N-acetylglucosamine; Asn(OlcNAc), N-acetylglucosamine attached to the side chain of asparagine; Asn(GlcNAc)>PhNCS, phenylthiohydantoin derivative of Asn(GlcNAc); N-CAM, neural cell-a...

Abstract. Recent results showing that a singlefibronectin gene can give rise to several different mRNAs by alternative splicing have offered an explanation for fibronectin polymorphism. Here we report on monoclonal antibodies that show specificity for a fibronectin segment (ED) that can be included or omitted from the molecule depending on the pattern of splicing of the mRNA precursors. Using these monoclonals, we have quantitatively analyzed the expression of the ED sequence in human fibronectin from different sources. The results demonstrated that, at the protein level, the ED segment is not expressed in plasma fibronectin and that, in fibronectin from the tissue culture medium of tumor-derived or simian virus-40-transformed human cells, the percentage of fibronectin molecules containing the ED segment is about 10 times higher than in fibronectin from normal human fibroblasts. These results suggest that in malignant cells the mechanisms that regulate the splicing of mRNA precursors are altered. F IBRONECT1NS (FNs) ~ are high-molecular-mass, adhesive glycoproteins present in the soluble form in plasma and other body fluids and in insoluble form in the extracellular matrices and basement membranes. FN molecules act as bridges between the cell surface and extracellular material. In fact, the FN molecules contain a cellbinding site and binding sites for collagen, heparin, gangliosides, and fibrin. Because of their multiple interactions, FNs play an important role in diverse biological phenomena, including cell adhesion, cell migration, hemostasis and thrombosis, wound healing and the ability to induce a more normal phenotype in transformed cells (for reviews on distribution, structure, and biological functions, see references 1,7,8,18,24).It has been demonstrated that FN polymorphism may be at least partially due to alternative splicing schemes in two regions (ED and IIICS) because as many as 10 different mRNAs may originate from the primary transcript of a single gene (9, 12-15, 25, 26, 29) localized on chromosome 2 (11, 34). In fact, Schwarzbauer et al. (25) have shown that an antiserum specific for the rat fibronectin IIICS sequence recognizes the larger subunit of rat plasma FN (plFN), but not the smaller one.Here we report the characterization of mAbs for the ED 1. Abbreviations used in this paper: cFN and plFN, fibronectin from cultured cell media and plasma, respectively. fragment of fibronectin. Using these mAbs in a quantitative assay, we have demonstrated that this sequence is not present in plFN and that tumor-derived and SV-40-transformed human cells release a population of FN molecules in which the percentage of subunits containing the ED sequence is about 10 times higher than in the FN released by normal human fibroblasts.

ABSTRACT

Carcinoembryonic antigen (CEA) is a glycoprotein important as a tumor, marker for colonic cancer.Immunological and biochemical studies have shown it to be closely related to a number of other glycoproteins, which together make up a gene family. We have cloned a member of this gene family by using long oligonucleotide probes (42-54 nucleotides) based on our protein sequence data for CEA and NCA (nonspecific cross-reacting antigen) and on human codon usage. The clone obtained (X39.2) hybridizes with six probes and has a 15-kilobase insert. The 5' end of the gene is contained within a 2700-base-pair EcoRI fragment, which hybridizes with five of the six synthetic probes. Sequencing of the 5' end region revealed the location and structure of one exon and two putative intron boundaries. The exon encodes part of the leader sequence and the NH2-terminal 107 amino acids of NCA. Southern blot analysis of human normal and tumor DNA, using as probes two X39.2 fragments that contain coding sequences, suggests the existence of 9-11 genes for the CEA family. One of the restriction fragments described here has been used by