A glycoprotein present in normal human tissue is characterized that is neither organ-nor tumorspecific (nonspecific crossreacting antigen) and that crossreacts (by the Ouchterlony double-diffusion technique) with the carcinoembryonic antigen. This immunological relationship indicates common determinants on the molecules of both antigens. We demonstrate that the nonspecific crossreacting antigen is not a fragment of the carcinoembryonic antigen molecule.In our first publication about the isolation of a fetal antigen from human colonic tumors (1), we mentioned two antigens of B electrophoretic mobility, the carcinoembryonic antigen (CEA) first described by Gold and Freedman (2) and an antigen present in higher amounts in intestinal carcinomas than in noncancerous mucosa that has been found to be a protein present in normal tissues. In the present paper, it will be shown that this latter antigen is neither tumor-nor organspecific. It has been named, therefore, nonspecific crossreacting antigen (NCA). The characterization of NCA and its immunological relationship to CEA is also included in this study. MATERIALS AND METHODSTumor Extracts were prepared from a pool of 15 unpreserved, primary colonic epitheliomas of the glandular type that were removed by surgery and two livers with metastases from the same malignant tumors (obtained at autopsy within 8 hr after death). In this study, only extracts prepared by precipitation of the tissue homogenates with perchloric acid were used, which have a high content of CEA and some other tissue proteins but very little serum proteins.The extraction procedure has been described (1, 3). Semipurified extracts of lung and spleen tissues were also used, as well as purified CEA (3). NCA was obtained (as a byproduct) during isolation of CEA, since it is a constant, tenacious contaminant of the different CEA-purification steps. NCA and CEA were finally separated by consecutive column filtrations on Sephadex G-200 in 0.01 M K-phosphate buffer (pH 8.2).Antisera were prepared in rabbits against semipurified extracts of colonic tumors, normal mucosa, and isolated NCA. An antiserum was also prepared in sheep against purified CEA (3, 4). Precipitation Reactions. Ouchterlony double-diffusion in agar and Scheidegger's modification of immunoelectrophoresis (5) were used, both in 1.5%o agar and 0.15 M veronal-HCl buffer (pH 8.2). RESULTS When the semipurified extract of colonic tumors was further purified by preparative electrophoresis and the cathodic fractions of the extract were analyzed with an antiserum prepared against the semipurified extract by immunoelectrophoresis, two precipitin lines of ( mobility were seen. The inner line represented CEA, and the outer line represented NCA (Fig. 1)

We studied the plasmin system with specific antisera to plasminogen, its 2 activators (urokinase-type and tissue-type) and the 2 plasmin inhibitors, alpha 2 anti-plasmin and alpha 2 macroglobulin on sections of 34 human colonic tumors by immunofluorescence. Anti-plasminogen serum showed a clear-cut reactivity at the surface of tumor cells, as it stained the contour of tumor glandular structures, foci, and isolated tumor cells. Intratumoral deposits and necrotic areas were stained as well, often strongly. Localization of plasminogen was quite different from that of fibrinogen, which was found only in peritumoral stroma, and never on tumor cells. Traces of both types of plasminogen activator were found, mainly on invasive tumor cells for urokinase type and on large tumor foci for tissue type. Images were weak and inconstant. Large amounts of both plasmin inhibitors were characterized in tumor stroma. Alpha-2 anti-plasmin was also found in intratumoral deposits and necrotic areas. It seems likely that plasminogen exudes from blood capillaries (since anti-plasminogen serum often stained the whole capillary wall), diffuses in the stroma and binds to tumor cells. Once formed, plasmin is likely to play a role in the invasion of surrounding tissues by tumor cells, in the dissociation of tumor cells from tumor glands and in the production of necrosis inside tumor areas.

Specific gene transfer into targeted tumor cells remains a critical issue for the development of systemic gene therapy protocols. With this end in view, we have tested the possibility of selectively directing genes to tumor cells through the recognition of tumor-associated antigens (TAA). This was approached in vitro on four human renal cell carcinoma (RCC) lines by means of the highly specific mouse G250 monoclonal antibody (mAb) chemically conjugated to a plasmid DNA conveying a reporter activity. This mAb directed to a TAA that is present on 95% of primary RCCs and on 60% of metastatic human RCCs was extensively characterized, including during clinical trials. Epifluorescence microscopy analysis indicated that upon specific binding to G250 TAA, G250 mAb alone or conjugated to plasmid DNA was internalized by an active endocytic process and colocalized with the transferrin concentrated in the late recycling perinuclear compartment. We also observed that both unconjugated G250 mAb or G250 mAb conjugated to plasmid DNA remained in the perinuclear region of the cells for Ն20 hours and were not rapidly translocated to lysosomes or recycled to the plasma membrane. In contrast, unconjugated plasmid DNA was not internalized. After transfection of G250 TAA-positive RCC lines with G250 mAb conjugated to a plasmid cDNA encoding mouse interleukin-2, a significant and sustained production of mouse interleukin-2 protein was detected from days 5-15 and was abrogated by inhibiting the internalization process. Altogether, our data showed that endocytosis of G250 TAA should be the basis of gene transfer to RCC, suggesting that targeting of TAA capable of internalization may be the basis of new approaches for designing alternative cancer gene therapy procedures.Key words: Renal cell carcinoma; tumor-associated antigen; endocytosis; antifection; gene therapy.M ost ongoing clinical trials in cancer gene therapy are based primarily on direct intratumoral administration of recombinant vectors and/or on ex vivo cell manipulations. 1 To date, only a few procedures permit cell-or tissue-specific targeting in vivo. This can be achieved with specific viral-based gene delivery systems 2,3 and nonviral approaches through DNA complexed to a ligand capable of binding the surface molecules expressed on the targeted cells. 4 -6 We have developed an approach named antifection that allows specific gene targeting. This technique is based on the use of antibodies (Abs) directed toward cell-specific receptors, which are chemically conjugated to plasmid DNA. This was first validated by using various reporter genes expressed in lymphoid cells, both in vitro and in vivo. 7,8 Recently, we have also reported the successful transfer of interleukin-3 (IL-3)-encoding plasmid DNA linked to anti-CD117 Abs into cultured human primary hemopoietic progenitors. 9 Although most of the ligands examined so far continue to be expressed on nontumor cells, limiting the tumor targeting, several Abs directed to tumor-associated antigen (TAA) present a widely restricted...

Twenty-three colonic adenocarcinomas were studied by immunofluorescence with antisera against components of the basement membrane (type IV collagen, laminin, and heparan sulfate-rich proteoglycan) as well as antisera against antigens of the connective tissue (type III collagen, fibronectin, and hyaluronectin). Marked alterations of the basement membranes were consistently observed on staining with each one of the first three antisera. In contrast, staining of the normal components of connective tissue was in most cases as intense in tumors as in normal colonic mucosa. Hyaluronectin, a marker of peritumoral stroma, was found to be present in 13 out of 16 tumors studied. In six metastatic lymph nodes, tumor foci were sometimes surrounded by antigens of the basement membrane. But these antigens were never found in the cytoplasm of cancer cells.

Twenty colonic adenocarcinomas were studied by immunofluorescence with antisera against components of the basement membrane: type IV collagen, laminin and heparan sulfate-rich proteoglycan, as well as antisera against antigens of the connective tissue: type-III collagen, fibronectin and hyaluronectin. Marked alterations of the basement membranes were consistently observed on staining with each one of the first three antisera. In contrast, staining of the normal components of connective tissue was in most cases as intense as in normal colonic mucosa. Hyaluronectin, a marker of peritumoral stroma, was found to be present in 12 out 15 tumors studied.