Biochemical Characterization of a Sulfated Phosphoglycoprotein Antigen Expressed on Human Small Cell Lung Carcinoma*

A murine monoclonal antibody (mAb A23-16) was produced that recognizes a glycoprotein antigen preferentially expressed on the surface of human small cell lung carcinoma cells. This antibody is of IgG 1 isotype, has an association constant of 5 X lo7 M-*, and reacts preferentially with human small cell lung carcinoma cell lines and fresh frozen sections in enzyme-linked immunosorbent assays and immunoperoxidase assays, respectively. The antigen recognized by A23-16 is a sulfated glycoprotein with phosphorylated threonine residues. The mature 90-kDa molecule has intrachain disulfide bonds and appears to be derived from a 76- kDa precursor, that is neither sulfated nor phosphorylated, but contains N-linked oligosaccharides. Conversion of the 76-kDa precursor to the mature form is accompanied by processing of these oligosaccharides from the high mannose to the complex type, although the increase in molecular mass from 76 to 90 kDa cannot be accounted for by this modification alone. MAb A23-16 reacts with its target antigen independ- ent of the N-linked oligosaccharides, but requires intact intrachain disulfide bond(s) for reactivity. These studies on the molecular characterization of a monoclonal antibody-defined glycoprotein, preferentially expressed by small cell lung cancer, provide a basis for further structural and functional studies that may eventually

20-25% of lung cancer patients and is characterized by rapid metastatic spread and poor prognosis, although it shows some responsiveness to chemotherapy (3,4).
The use of immunological approaches to manipulate the immune response of small cell lung cancer patients offers an alternative means for therapy possibly by using monoclonal antibodies directed to SCLC-associated antigens for targeting of chemotherapeutic drugs or radionuclides. The development of hybridoma technology resulted in a number of reports describing monoclonal antibodies reacting with SCLC (5-8). However, the majority of these publications dealing with these potential glycoprotein target antigens stressed mainly the relative specificity of the monoclonal antibodies detecting them rather than an in depth biochemical characterization of their target antigens. This is in contrast to other areas of cancer research where both glycolipid and glycoprotein antigens targeted by monoclonal antibodies have been more extensively characterized by biochemical and immunochemical means as summarized in recent reviews (9)(10)(11).
In this report, we focus mainly on the biochemical characterization of a target antigen for a monoclonal antibody reactive with small cell carcinoma of lung that may be of potential use for therapy of this neoplasm.

MATERIALS AND METHODS
Cells-Tumor cell lines and lymphoblastoid cell lines were propagated in RPMI 1640 medium containing 10% fetal calf serum (FCS), 2 mM glutamine, and 25 pg/ml gentamycin. Hybridomas were cultured in Dulbecco's modified Eagle's medium supplemented with 10% FCS, 2 mM glutamine, and 25 pg/ml gentamycin, or ascites in pristaneprimed BALB/c mice.
The following human cell lines were obtained from the American Type Culture Collection (ATCC), Rockville, MD; HPB-ALL (lymphoblastoid); CALU-1 and SK-MES-1 (squamous lung carcinoma); PANC-1 (pancreatic carcinoma); and WI-38 (diploid lung fibroblast). All other cell lines used were obtained as follows: T293 (SCLC) and Tissues-Portions of fresh normal and malignant tissue were obtained from the Surgical Pathology Department of the Ida M. Green Hospital of Scripps Clinic, La Jolla, CA. Specimens were embedded in Tissue Tek Medium (Scientific Products) and frozen in blocks in isopentane at liquid nitrogen temperature. They were then stored at Production of Monoclonal Antibodie-Monoclonal antibody A23-16 was produced against the human SCLC cell line T293 by standard hybridoma technology (12,13). Details relevant to this report are as follows: BALB/c mice were injected intraperitoneally with 10' T293 -70 "C. cells at weekly intervals. Three days after the final injection, the splenocytes were fused with the M5 variant of the nonsecretor murine myeloma cell line SP2/0 and cultured in 96-well plates with 2 X lo6 murine thymocytes per ml. Hybridoma A23-16 was selected by growth in Dulbecco's modified Eagle's medium containing 10% horse serum, hypoxanthine, aminopterin, and thymidine and subcloned by limiting dilutions.
Ascites Production-BALB/c mice were primed by injecting 0.5 ml of pristane. Two weeks after priming, 10-15 X lo6 hybridoma cells were injected intraperitoneally into the mice and ascites fluid collected repeatedly every 2-3 days. Antibody Purification-Purification was achieved by precipitating the antibody from the ascites by 45% saturation with (NHd'SO4, dialyzing against 10 mM Tris-HC1, pH 8.0, and application to a DEAE-cellulose column equilibrated with this buffer (13). The antibody was eluted from the column with a linear gradient of NaCl (0-0.4 M), dialyzed against PBS, and stored at -70 "C.
Zsotyping-The isotype of the monoclonal antibody was determined to be IgG 1 by ELISA, using aliquots of diluted, affinity-purified rabbit antisera specific for different murine light chains (Southern Biotechnology Associates, Birmingham, AL) that were dried into 96well microtiter plates (Dynatech, Alexandria, VA).
ELZSA Determination of Cell Binding-Screening of monoclonal antibody by ELISA was done as described previously (14). Briefly, 5 X 10' target cells/well were plated in polyvinyl microtiter plates (Dynatech). Prior to ELISA, the dried plates were rehydrated by washing them twice with 10 mM PBS, pH 7.4, containing 0.1% Tween 20, and 0.02% Thimerosol to remove any nonspecific binding hybridoma. Supernatants were diluted 1:2 in washing buffer containing 0.1% bovine serum albumin as diluent. Diluted test supernatant (50 pl) was added to each well and plates were incubated for 1 h at 4 "C. Following three washes, 50 pl of peroxidase-conjugated goat antimouse IgG (Bio-Rad) was added to each well and the reaction mixture was again incubated for 1 h at 4 "C. After two final washes, 50 pl of substrate solution (400 pl/ml o-phenylenediamine plus 0.12% H202) was added to each well. The reaction was stopped at 15 min by addition of 25 pl of 4 N H2S04 to each well. Optical absorbance at 492 nm was measured with an ELISA plate reader.
Zmmurwperoxidase Staining of Frozen Tissues-Four micron tissue sections were mounted on gelatin-coated glass slides, air dried briefly, and tested immediately in an indirect immunoperoxidase assay (15). Briefly, after washing twice in HBSS and once in PBS, the sections were preincubated for 15 min with PBS containing 10% goat serum and 0.1% bovine serum albumin. Sections were then overlayed with appropriate diluted hybridoma supernatants and incubated for 1 h at room temperature. After two washes in HBSS and one wash in PBS, the tissue sections were overlayed with horseradish peroxidase conjugated to goat anti-mouse immunoglobulin (Bio-Rad) diluted 1:50 with immunoperoxidase dilution buffer (PBS containing 10% goat serum and 0.1% bovine serum albumin), and incubated for 1 h at room temperature. Finally, the tissue sections were incubated for 15 min at room temperature with immunoperoxidase substrate buffer (10 mM Tris, pH 7.6, containing 0.6 mg/ml 3,3'-diaminobenzidine and 0.015% H20.J after washing twice in HBSS and once in PBS. The sections were counterstained briefly in 1% methylene blue, dehydrated through graded ethanol, washed in Histoclear (National Diagnostics Somerville, NJ), mounted with Pro-Texx (Lerner Laboratories, New Haven, CT), and examined by microscopy, f'P1Orthophosphate Labeling of Cells"T293 cells were labeled as with 50 mM HEPES containing 0.1% bovine serum albumin. The described previously (16). Briefly, 2.0 X lo7 cells were washed twice cells were then propagated for 15 min in phosphate-free medium (Irvine Scientific, Santa Ana, CA) and then labeled with 0.5 mCi of [32P]orthophosphate (8 mCi/ml, carrier-free) for 3 h at 37 "C. After labeling, the cells were washed twice with cold phosphate-buffered saline containing 100 mM sodium pyrophosphate, 100 mM sodium fluoride, 4 mM EDTA, and then solubilized for 10 min on ice in RIPA lysis buffer containing 100 mM sodium pyrophosphate, 100 mM sodium fluoride, 4 mM EDTA, 2 mM phenylmethylsulfonyl fluoride. The cell lysate was cleared by ultracentrifugation at 100,000 X g for 45 min and indirect immunoprecipitation and SDS-polyacrylamide gel electrophoresis were performed as described below. Scatchard Plot Analysis-The association constant ( K A ) of mAb A23-16, Iz5I-labeled to a specific activity of 8.7 X lo3 cpm/ng by the chloramine-T procedure (17), its association constant (Ka) was determined from Scatchard plot analysis of saturation binding data (18). Briefly, T293 SCLC cells (106/0.1 ml of RPMI 1640 medium containing 2% FCS and 0.01% NaN3) were incubated for 1 h at 4 "C with increasing amounts of radiolabeled antibody (5-600 ng), diluted in the same media. Bound antibody was separated from free by a procedure involving layers of dibutyl phthalate and dinonyl phthalate (1:l) oils as described by Beaumier et al. (19). Radioactivity in each fraction corresponding to the bound antibody was determined in an LKB 1270 y-counter. The amount of nonspecific binding of mAb A23-16 was estimated by adding 100-fold excess of cold antibody to a cell suspension and subtracting bound counts/min from total counts/min. The data were subjected to linear regression analysis. Indirect Immunoprecipitation and SDS-PAGE-Purified antibodies were coupled to Sepharose CL-4B (Pharmacia LKB Biotechnology, Inc.) (10 mg of antibody/ml beads), previously activated with CNBr by the method of March et al. (20). Following two washes with 1 ml of 0.5% Tween 20, 0.1% ovalbumin in PBS (PTO buffer), 5 pl of the antibody-coupled Sepharose was incubated overnight with 0.5 ml of PTO and cell lysates containing 3-10 X lo6 cpm of trichloroacetic acid insoluble [1251]. The beads were then washed with PTO buffer and bound antigens were eluted, analyzed by SDS-PAGE as described by Laemmli (21), and visualized by fluorography according to Bonner and Lasky (22).
Cell Surface Zodination-Cell surface molecules on T293 cells were labeled with lZ5I by the lactoperoxidase method (23) using Enzymobeads (Bio-Rad). An aliquot of lo7 cells was washed three times with PBS then iodinated with 1 mCi of Iz6I according to the manufacturer's protocol. Each immunoprecipitation was performed by using 2 X lo6 cpm of trichloroacetic acid-insoluble radioiodine per immunoadsorbant.
Pulse-Chase Biosynthetic Studies-For pulse-chase studies, T293 SCLC cells were propagated for 15 min in methionine-free medium (Gibco) and then pulse-labeled for 10 min with [35S]methionine (specific activity 1295 Ci/mmol, Du Pont-New England Nuclear) at a concentration of 1.0-1.5 mCi/3.2 X lo7 cells/ml. After the removal of an aliquot of 4 X lo6 cells that constituted the zero time point, the remaining cells were washed four times with HBSS containing 4 mM unlabeled L-methionine (Sigma), resuspended in RPMI 1640 containing 10% FCS and 4 mM unlabeled methionine, and incubated at 37 "C. Aliquots were removed at the time points indicated and the cells were centrifuged and extracted in RIPA lysis buffer as described by us previously (23).
Enzymatic Digestion Studies-In some experiments, immunoprecipitates were washed, resuspended in 50 pl of 20 mM sodium citrate buffer, pH 5.5, with or without addition of 0.01 unit of Endo-H (endo-(3-N-acetylglucosaminidase H) (Miles) and incubated for 1 h at 37 "C, essentially as described previously (23,25,26). Fifty microliter of the 2 X sample buffer was then added and the samples analyzed by SDSpolyacrylamide gel electrophoresis. Immunoprecipitates were digested and analyzed likewise with Endo-F (27), kindly provided by Dr. John Elder of our institution.
Phosphoamino Acid Analysis-The phosphoamino acids were analyzed from the 32P-labeled antigen as described previously (28). Briefly, the 32P-labeled band comprising the 90-kDa antigen component was excised from a dried polyacrylamide gel and eluted with 0.1 M NHI(C03) buffer, pH 8.6, containing 0.5% SDS. The protein present in this eluate was precipitated with 20% trichloroacetic acid in the presence of bovine serum albumin (10 pg/ml) and washed with cold 80% ethanol. The protein solution was then hydrolyzed under N2 for 90 min at 110 "C. The hydrolysate was lyophilized and mixed with 5 pg each of phosphoamino acid markers. The phosphoamino acids were separated by electrophoresis at pH 3.5 (acetic acidpyridine:water, 50:5:945) on thin layer cellulose plates at 500 V for 2 h. The plate was exposed for 4 days to an x-ray film with an enhancing screen, then the plate was sprayed with ninhydrin and heated at 100 "C for 5 min to localize the phosphoamino acid markers.

RESULTS
Production, Selection, and Characterization of MA6 A23-16-Twenty-seven hybridomas that reacted in ELISA with the immunizing T293 cells, but not with the lymphoblastoid cell line L14, were selected from approximately 1000 primary hybridoma clones. From these hybridomas, 19 specifically stained fresh frozen sections of SCLC in immunoperoxidase assays. MAb A23-16 was selected for further study and subcloned after extensive analyses with ELISA cell binding assays against cell lines derived from SCLC, other solid tumor cell lines, as well as lymphoblastoid and fibroblastoid cell lines.
MAb A23-16 (IgG 1) binds to the surface of T293 SCLC cells in a saturable fashion with an association constant ( K A ) of 5 x 10' M-'. Moreover, Scatchard analysis revealed that T293 SCLC cells have 4.7 x lo5 binding sites for this antibody per cell. ELZSA Cell Binding Analysis- Table I summarizes the reaction of mAb A23-16 with various human cell lines. MAb A23-16 reacted with all SCLC cell lines tested but not with those derived from other solid tumors, with the exception of the lung adenocarcinoma cell line UCLA-P3 and the neuroblastoma cell line SK-N-RA. It is possible that the non-SCLC cell lines that reacted positively with mAb A23-16 may share a common antigen; however, immunoprecipitation analysis will be needed for confirmation. The antibody did not react with a series of lymphoblastoid cell lines of either Tor Bcell origin, human red blood cells, or WI-38 human lung fibroblasts.
Immunohistochemical Staining of Frozen Tissue Sectiom- Table I1 summarizes the reactivity of the A23-16 antibody with fresh frozen sections of tumors and normal fetal and adult tissues. MAb A23-16 reacted only with sections of SCLC and failed to stain sections of other tumors and most normal tissues. Of the normal tissues tested, sections of liver stained faintly with MAb A23-16, as did proximal and distal tubules of kidney. Normal fetal tissues that failed to react included lung, liver, kidney, colon, and spleen. Among normal adult tissues that showed no detectable immune staining were lung, pancreas, spleen, colon, and brain cortex.
Molecular Characterization of the A23-16Antigen"Indirect immunoprecipitation and Western blotting were employed for    This decrease in electrophoretic migration after chemical reduction is consistent with the presence of one or more intrachain disulfide bond(s) that stabilize a secondary structure.

Glycoprotein Antigen on
SDS-PAGE analysis was performed on unlabeled detergent extracts of T293 SCLC cells following Western blotting of proteins from SDS gels to nitrocellulose membranes. Under nonreducing conditions, mAb A23-16 reacts with the 76-kDa form of the antigen, but does not react on Western blots following chemical reduction of the samples with 2-mercaptoethanol (Fig. 1B). In addition to the 76-kDa component detected on Western blots under nonreducing conditions, mAb A23-16 also detects a molecule of 200 kDa in extracts of T293 cells, the nature of which is currently unknown.
Biosynthesis of the A23-16 Antigen-A pulse-chase labeling experiment was performed with T293 cells to assess the metabolic processing of the A23-16 antigen and determine which biosynthetic forms, if any, are recognized by mAb A23-16. Following a 10-min pulse of T293 cells with ["S]methionine and subsequent chase in unlabeled culture medium, aliquots were removed and detergent extracts prepared for immunoprecipitation and SDS-PAGE analysis. Under reducing conditions, the first molecule precipitated by A23-16 after a 10-min labeling period is one of 76 kDa, which appears to gradually give rise to the mature 90-kDa form previously found on the cell surface (Fig. 2). The kinetics with which the 76-kDa molecule disappears and is replaced by the 90-kDa form are suggestive of a precursor-product relationship. A parallel analysis of the same cell extracts on nonreducing SDS-PAGE gels indicates an initial reactive molecule of 66 kDa, that gives rise to one of 76 kDa, previously noted by surface iodination analyzed on nonreducing SDS gels.
In order to determine whether the 76-and 90-kDa molecules observed under reducing conditions contain N-linked oligosaccharides, T293 cells were metabolically labeled with [%3] methionine for 1 h in methionine-free medium, such that both components can be immunoprecipitated. Immunoadsorbants were then digested with Endo-H prior to SDS-PAGE, under conditions outlined under "Materials and Methods." According to the resulting reaction pattern shown in Fig. 3  kDa component exhibits an increased electrophoretic migration as a result of Endo-H digestion, indicating the presence of high-mannose type, N-linked oligosaccharides. In contrast, the 90-kDa molecule is not affected by Endo-H digestion. Endo-F digestion of the 90-kDa molecule, following iodination and immunoprecipitation from T293 cell extracts, reveals approximately the same drop in apparent molecular mass as that seen when the 76-kDa putative precursor is digested with Endo-H ( Figs. 1 and 3). These data further support the precursor-product relationship between these two molecules.
In order to determine whether N-linked oligosaccharides are required on the antigen molecule for the formation of the A23-16 determinant, T293 cells were pre-treated for 3 h in medium containing various concentrations of tunicamycin, then labeled for 1 h in the appropriate concentration of this drug prior to extraction with RIPA lysis buffer. Tunicamycin inhibits the addition of N-linked carbohydrates to glycoproteins by preventing the formation of the lipid-linked sugar intermediate prior to its addition to asparagine resides. Lysates of T293 cells treated only with the solvent, dimethyl sulfoxide, contain primarily the 76-kDa precursor molecule and a trace of the mature 90-kDa product when immunoprecipitated with the A23-16 antibody (Fig. 4). An increase in the concentration of tunicamycin to 1 pg/ml results in immunoprecipitation of both the 76-kDa molecule and the nonglycosylated form of 70 kDa. At tunicamycin concentrations of 3 and 9 pglml, only the nonglycosylated 70-kDa form is immunoprecipitated, suggesting that the N-linked oligosaccharides are not required for reactivity with mAb ,423-16.
Phosphorylation and Sulfation-To assess whether the molecule recognized by mAb A23-16 is phosphorylated, T293 SCLC cells were labeled with [R2P]orthophosphate for 3 h a t 37 "C, and then its detergent lysate was subjected to immunoprecipitation with mAb A23-16. Subsequently the 90-kDa molecule was found to be phosphorylated, as indicated by the autoradiogram seen in Fig. 5A. Phosphoamino acid analysis of the "2P-labeled 90-kDa antigen (Fig. 5 B ) demonstrates that this phosphorylation occurs a t threonine residue(s).
Further analyses of post-translational modification were performed to determine whether the 90-kDa component is sulfated. To this end, T293 SCLC cells were cultured in the

76-70-
Glycoprotein Antigen on H u m a n Small Cell Lung Cancer presence of "SO4, and their detergent lysates subjected to indirect immunoprecipitation with mAb A23-16, followed by SDS-PAGE analysis. It is evident from the data depicted in Fig. 5C (lane 2) that the 90-kDa form of the molecule incorporates radioactive sulfate. The nature of the sulfated moieties on the A23-16 antigen is presently unknown. In this regard, digestion of the antigen with enzymes specific for chondroitin and heparan sulfate glycosaminoglycans had no effect on its migration in SDS-PAGE (not shown).

DISCUSSION
The data presented here indicate that a murine monoclonal antibody A23-16 of the IgG 1 isotype reacts specifically with a glycoprotein antigen that is preferentially expressed on cell lines and tissues derived from small cell lung carcinoma. ELISA assays of a variety of human tumor cell lines, as well as fibroblastoid and lymphoblastoid cell lines, indicate that mAb A23-16 binds significantly to SCLC cell lines, but not to a neuroblastoma and a lung adenocarcinoma cell line. The data from immunoperoxidase assays on fresh frozen normal adult and fetal tissues, as well as a variety of tumor tissues, further support the specificity of mAb A23-16 for SCLC tissues.
The biochemical characterization of the target antigen for mAb A23-16 involved a series of studies using a combination of indirect immunoprecipitation analyses, pulse-chase biosynthetic experiments, and enzymatic digestion studies. Initial immunoprecipitation/SDS-PAGE experiments revealed that mAb A23-16 precipitates a single molecule of 90 kDa under reducing conditions and a 76-kDa molecule under nonreducing conditions (Fig. L4). This demonstrates that the native molecule contains one or more intrachain disulfide bonds, and the Western blot analysis suggests that these bonds must be intact for A23-16 reactivity, since the antibody reacted with its target antigen in Western blots only under nonreducing conditions (Fig. 1B).
An evaluation of the biosynthesis of this antigen molecule by pulse-chase studies with [3sS]methionine revealed a component of 76 kDa reacting with mAb A23-16 immediately following a 10-min pulse label and the subsequent appearance of 90 kDa within 30 min. The 90-kDa molecules labeled during the 10-min pulse remained associated with the cells for as long as 21 h (Fig. 2). These results suggest that the 90-kDa component is the mature form or biosynthetic product of this antigen that is derived from a 76-kDa precursor.
Conversion of the 76-to the 90-kDa form apparently involves the processing of N-linked oligosaccharides from the high-mannose to the complex type. Following immunoprecipitation, treatment of the respective components with Endo-H suggests that the 76-kDa precursor molecule contains Nlinked oligosaccharides of the "high-mannose'' type, whereas the 90-kDa product was unaffected by Endo-H digestion (29). Digestion of the 90-kDa molecule with Endo-F resulted in an increase in electrophoretic migration, while Endo-H digestion had no such effect, therefore, this component contains Nlinked oligosaccharide residues of the complex type. The increase in migration of the 90 kDa following Endo-F digestion was approximately the same as that observed when the 76-kDa molecule was digested with Endo-H and when the 76 kDa were analyzed in the presence of tunicamycin. These data, along with the pulse-chase data, support the hypothesis that a precursor-product relationship exists between the 76and 90-kDa molecules, since they appear to contain approximately the same amount of N-linked carbohydrate; however, the increase in apparent molecular mass from the 76-kDa putative precursor to the 90-kDa form cannot totally be accounted for by the addition and processing of N-linked oligosaccharides. Other possible explanations for this difference in molecular mass include 0-linked glycosylation, phosphorylation, and sulfation, especially since the mature, 90-kDa molecule, in contrast to the 76-kDa molecule, appears to be phosphorylated at threonine residues, as well as sulfated (Fig. 5). Although a structure-function relationship among these post-translational modifications is not immediately apparent, they are nevertheless of interest and may become better defined once more structural information is obtained on their protein backbone.
A number of tumor-associated antigens have been reported in the literature that range in molecular mass between 90 and 100 kDa, including p97, a melanoma-associated glycoprotein and the transferrin receptor that is present in a variety of tumor tissues (1 1). At present one cannot discern any apparent relationship between these molecules and the 90-kDa glycoprotein expressed on SCLC cells and specified by mAb MAb A23-16 failed to mediate either antibody-dependent cellular cytotoxicity or complement-dependent lysis of tumor cells in vitro (data not shown) and consequently appears less likely to be useful for any immunological manipulations of the host immune response to small cell lung carcinoma. However, the 90-kDa target antigen of mAb A23-16 is preferentially expressed on the surface of SCLC cells with a relatively high density of 4.7 x lo5 per cell. For this reason, and because the 90-kDa target antigen is apparently not shed, mAb A23-16 may well be useful to target chemotherapeutic drugs or radionuclides to this antigen epitope and aid in suppressing the growth of SCLC, and thereby contributing to the development of new treatment modalities for small cell carcinoma of the lung. A23-16.