lnt. J . Cancer: 47, 163-169 (1991) 0 1991 Wiley-Liss , Inc .
Publication of the International Union Against Cancer Publication de I'Union lnternationale Contre le Cancer
NORMAL TISSUE REACTIVITY OF FOUR ANTI-TUMOR MONOCLONAL ANTIBODIES OF CLINICAL INTEREST Rhona STEIN,David M. GOLDENBERG and M. Jules MATTES' Centerfor Molecular Medicine and Immunology at the University of Medicine and Dentistry of New Jersey, 1 Bruce Street, Newark, NJ 07103, USA. Normal tissue reactivity on frozen sections was examined with 4 monoclonal antibodies (MAbs) that were reported previously to be negative or weakly reactive with normal tissues, and strongly reactive with some types of carcinoma. All 4 antibodies reacted strongly with certain normal epithelial cells. The antibodies tested include 2 antibodies to ovarian cancer, MOv18 and MOvl9, one antibody to breast cancer, H23, and one antibody reactive with a range of carcinomas, 672.3. MOVIE, MOv19 and H23 reacted with many normal glandular and ductal epithelial cells, while 672.3 reacted most strongly with secretions of the small intestine epithelium and the suprabasal squamous epithelial cells of the esophagus. Since the tissue distribution of MOvl9 was very similar to that of another antibody described previously, MW207, these two antibodies were compared by competitive binding inhibition, and found to recognize the same epitope. Our data emphasize the importance of repeated, independent tests of antibody specificity.
A major factor in selecting monoclonal antibodies (MAbs) for tumor immunotherapy is the specificity of the antibody, since there is a clear advantage in using MAbs that are nonreactive or weakly reactive with normal tissues. Antigen distribution in normal tissues and tumors is generally determined by immunohistology on frozen or paraffin-embedded specimens, since this procedure allows for a comparison of a wide range of specimens, even though the results may not necessarily reflect antigen that is accessible in vivo to MAb injected parenterally. Our efforts to generate tumor-specific MAbs (Mattes et al., 1984, 1985, 1987; Stein and Goldenberg, 1988) have resulted in MAbs that react with tumors, but also react equally strongly with a considerable number of normal tissues. Since we have focused primarily on carcinomas, the antigens identified are epithelial differentiation antigens. While many other investigators have had similar results, there are a significant number of reports of more specific MAbs, including MOv18 and MOv19 for ovarian carcinoma (Miotti et al., 1987), H23 for breast carcinoma (Keydar et al., 1989), and B72.3 for a variety of carcinomas (Thor et al., 1986). The first 3 of these MAbs were reported to be non-reactive with all normal tissues tested, while B72.3 was reported to be weakly reactive with a small number of normal epithelial cells. In order to select the optimal MAbs for immunotherapy experiments, we have tested these 4 MAbs, and here present some of our results. Our conclusion is that all of these MAbs react much more extensively with normal tissues than has been appreciated. These data emphasize the importance of repeated independent investigation of MAb specificity. The antigens recognized by the MAbs used in this study have been described previously in detail. B72.3 recognizes a mucin-like antigen (Johnson et al., 1986), and the determinant recognized has been identified as sialyl-Tn, a core region structure of 0-linked carbohydrate chains (Gold and Mattes, 1988; Kjeldsen et al., 1988). MOv 18 and MOv 19 recognize different determinants on a glycoprotein of MW 38-40 kDa (Miotti et al., 1987). H23 recognizes a 68 kDa glycoprotein present in a preparation described as purified human mammary-tumor virus, isolated from breast-cancer cell line T47D (Keydar et al., 1989).
MATERIAL AND METHODS
Specimens and antibodies Normal human tissues were obtained from surgical specimens or from autopsies performed 6 1 8 hr after death. The surgical specimens included the uterus, cervix and Fallopian tube. The autopsy specimens were obtained primarily from 4 autopsies. Tissues were frozen in 2-methylbutane cooled to - 70°C with dry ice and stored at - 70°C. Cryostat sections, 8-10 p,m thick, were dried thoroughly under a hair dryer and stored at -70°C. Hybridomas B72.3 (an IgG,) and H23 (an IgG,) were obtained from ATCC (Rockville, MD, HB 8108 and HB 8630, respectively). The cells were grown in tissue culture and i.p. in Pristane-primed BALB/c mice, obtained from Charles River (Wilmington, MA). MOv18 (an IgG,) and MOv19 (an IgG,,) ascites fluid were generously provided by Dr. M.I. Colnaghi (Istituto Nazionale per lo Studio e la Cura dei Tumori, Milan, Italy). MH99, an IgG,, Ab to epithelial surface antigen (ESA), which is the same antigen recognized by the Ab 17-1A, was described previously (Thampoe et al., 1988) and was included as a positive control to show cellsurface staining of most epithelial cells. MW207 and MA103 were also described previously (Mattes et al., 1987; Ong and Mattes, 1989). Immunohistology The avidin-biotin complex method (Vector, Burlingame, CA) was used, which has been previously described in detail (Ong and Mattes, 1989). Duplicate sections were tested with each MAb in every experiment. Sections were fixed for 10 min with 95% ethanol, and treated to destroy endogenous peroxidase activity with 0.3% H,O, in methanol, before application of the MAb. H23 was applied as undiluted culture supernatant, and the other MAbs were applied as 1500 dilutions of hybridoma ascites fluid. Negative control MAbs, tested on every specimen, included Ag8 ascites (ATCC TIB 9, an IgG,) and 3B18 culture supernatant (ATCC HB 8654, an IgG,). Although we did not include an IgG,, negative control, in many previous experiments (Mattes er al., 1984, 1985) non-reactive IgG,, MAbs were tested, together with IgG,s, against many normal tissues, and no subclass-dependent variation in background staining was detected. The IgG concentration in ascites fluids was determined by radial immunodiffusion (Williams and Chase, 1971), using a 1:200 dilution of goat anti-mouse IgG (401210, Calbiochem, San Diego, CA) in agarose, and purified MAbs of the appropriate subclasses, prepared as described by Ong and Mattes (1989), as standards. The concentration of IgG in undiluted ascites was 2.3, 3.2, 5.6, 3.1 and 8.1 mg/ml for MAbs MOv-18, MOv-19, B72.3, MH99 and Ag8 respectively, so the 1500 dilution used had 4.6-16.2 p,g/ml, with the negative control MAb having the highest concentration. Sections were counterstained with hematoxylin. 'To whom correspondence and reprint requests should be addressed. Received: April 19, 1990 and in revised form September 1, 1990.
164
STEIN ET AL
Competitive binding inhibition This assay was described in detail by Mattes et al. (1989). Briefly, MW207 was purified, labelled with 1251, and tested for binding to cells of the lung-carcinoma cell line SK-LC- 1 in the presence or absence of various unlabeled MAbs. RESULTS
A summary of immunohistology results on frozen sections is presented in Table I and examples are shown in Figures 1 and 2. In most cases, multiple specimens of each tissue type were tested and consistent results were obtained, with some exceptions that are discussed below. All MAbs reacted exclusively with epithelial cells and/or epithelial secretions, connective tissue being negative. The only exception to this was reactivity of B72.3 with interstitial (Leydig) cells in the testis. Each of the MAbs reacted with a distinct subset of epithelial cells. Except as noted in the following summary, positive MAbs reacted with most or all epithelial cells in the designated tissue. MOv19 reacted in the pancreas strongly with ducts, with weaker staining of acinar cells (Fig. 2); in the cervix and uterus with only the luminal edge and secretions of glandular cells; in the bronchus with glandular cells; in the lung with epithelial cells and/or secretions (Fig. 2); in the prostate with only 5-20% of epithelial cells, which could be ejaculatory ducts; in the skin, with sweat glands only; in the esophagus, with glandular and ductal epithelial cells only; and in several other tissues, predominantly with the luminal border of epithelial cells. MOv18 was similar to MOv19 but considerably weaker (Fig. 2): all tissues stained by MOv18 were stained similarly by MOv19, but a number of tissues negative for MOv18 were reactive with MOv19. There was no indication that MOv18 TABLE I - REACTIVITY OF MAbs WITH CRYOSTAT SECTIONS OF NORMAL HUMAN TISSUES 872 3
Tissue
MOv18 ~ _ _MOvlY
~~
Esophagus Stomach Duodenum
(3)’
Jejunum
(2) (2) (2)
Ileum
(4)
Colon
(2)
Trachea Bronchus Lung Pancreas Salivary gland Liver Skin
-
++ + I ++ ++ +-+ + + + ++ ++
(1)
(2)
(4) (3) (3) (2) (3)
Thyroid (1) Prostate (2) Fallopian tube (3) Uterus Endocervix Exocervix
(3) (2)
Breast Ureter
(2)
Heart
(2)
Spleen LvmDh node
(4) (4) (21
+-+
(1)
+
-
(1)
Testis
n23
++ + -
‘Number of specimens tested. All specimens were from autopsies except the Fallopian tube, utems, endocervix and exocervix. which were surgical specimens.-21n all cases except the testis, only epithelial cells were stained. More information on the identity of stained cells is provided in the text. In all cases except the bronchus, as indicated, consistent results were obtained with the multiple specimens of the same tissue type. + indicates dark staining; indicates relatively light but definite staining.-’Of 2 specimens, one was positive and the other negative. See text for further discussion.
+
+
had a pattern of reactivity distinct from that of MOv19, as opposed to simply reacting more weakly. H23 reacted in the lung with epithelial cells and/or secretions (Fig. 2); in the pancreas with acinar cells; in the salivary gland heterogeneously with 1G20% of epithelial cells; in the prostate with only 5-20% of epithelial cells, which could be ejaculatory ducts; in the uterus, cervix and Fallopian tubes predominantly with the luminal edge and secretions of glandular cells; in the skin, with sebaceous glands only (Fig. 1); and in the esophagus, with suprabasal squamous epithelial cells. B72.3 reacted heterogeneously with the stomach, in which 1G20% of glandular epithelial cells were positive; in the salivary gland, in which 5-10% of epithelial cells were positive; and in the uterus, where a subset of glands and secretions was positive. It reacted predominantly with secretions, and possibly with the luminal border of epithelial cells, in the duodenum, jejunum (Fig. l ) , ileum and cervix. It also reacted with interstitial (Leydig) cells in the testis, and with the suprabasal squamous epithelial cells of the esophagus (Fig. 1). The tissue distribution of MOv19 described here is similar to that of a MAb we described previously, MW207 (Mattes et al., 1987), and the molecular weights of the antigens recognized, 3 8 4 0 kDa, are the same. Therefore, we performed a competitive binding inhibition assay with radiolabelled MW207, which in fact showed that the epitopes recognized by MOv19 and MW207 are the same (Fig. 3). MOv18 produced partial inhibition of binding of MW207 (Fig. 3), which is consistent with results of Miotti et al. (1987) that MOv18 and MOv19 recognize different determinants on the same molecule. Since this study is concerned with the reliability of immunohistology data, and since variation between specimens may be a significant variable, it is of interest to compare results with the different specimens. Of the 20 tissue types for which more than one specimen was tested, results were consistent in the great majority of cases. However, 14 discrepancies were initially found, in which some specimens were positive while other specimens of the same tissue type were negative. The reason for the discordant results in these 14 cases was investigated further. First, the assays were repeated, to determine whether experimental variability could explain the results. In fact, 7 specimens previously found to be negative were subsequently determined to be positive, thereby eliminating the discrepancy between specimens in these 7 cases. Such falsenegative results can probably be attributed primarily to 2 factors. First, in a few assays there was generally weak staining with all MAbs, which might have been due to any of a number of technical problems. Second, in some cases there was unusually high background staining, which can obscure positive results; the reason for sporadically high background has not been determined. One other discrepancy can be attributed to a lack of sebaceous glands in 1 specimen of skin. For the remaining 6 discrepancies, the positive specimen was re-tested and found consistently positive, just as the negative specimen was consistently negative, so the difference could not be attributed to assay variability. In 3 of these cases, a very small fraction of cells were stained in the positive specimen,
Publication of the International Union Against Cancer Publication de I'Union lnternationale Contre le Cancer
NORMAL TISSUE REACTIVITY OF FOUR ANTI-TUMOR MONOCLONAL ANTIBODIES OF CLINICAL INTEREST Rhona STEIN,David M. GOLDENBERG and M. Jules MATTES' Centerfor Molecular Medicine and Immunology at the University of Medicine and Dentistry of New Jersey, 1 Bruce Street, Newark, NJ 07103, USA. Normal tissue reactivity on frozen sections was examined with 4 monoclonal antibodies (MAbs) that were reported previously to be negative or weakly reactive with normal tissues, and strongly reactive with some types of carcinoma. All 4 antibodies reacted strongly with certain normal epithelial cells. The antibodies tested include 2 antibodies to ovarian cancer, MOv18 and MOvl9, one antibody to breast cancer, H23, and one antibody reactive with a range of carcinomas, 672.3. MOVIE, MOv19 and H23 reacted with many normal glandular and ductal epithelial cells, while 672.3 reacted most strongly with secretions of the small intestine epithelium and the suprabasal squamous epithelial cells of the esophagus. Since the tissue distribution of MOvl9 was very similar to that of another antibody described previously, MW207, these two antibodies were compared by competitive binding inhibition, and found to recognize the same epitope. Our data emphasize the importance of repeated, independent tests of antibody specificity.
A major factor in selecting monoclonal antibodies (MAbs) for tumor immunotherapy is the specificity of the antibody, since there is a clear advantage in using MAbs that are nonreactive or weakly reactive with normal tissues. Antigen distribution in normal tissues and tumors is generally determined by immunohistology on frozen or paraffin-embedded specimens, since this procedure allows for a comparison of a wide range of specimens, even though the results may not necessarily reflect antigen that is accessible in vivo to MAb injected parenterally. Our efforts to generate tumor-specific MAbs (Mattes et al., 1984, 1985, 1987; Stein and Goldenberg, 1988) have resulted in MAbs that react with tumors, but also react equally strongly with a considerable number of normal tissues. Since we have focused primarily on carcinomas, the antigens identified are epithelial differentiation antigens. While many other investigators have had similar results, there are a significant number of reports of more specific MAbs, including MOv18 and MOv19 for ovarian carcinoma (Miotti et al., 1987), H23 for breast carcinoma (Keydar et al., 1989), and B72.3 for a variety of carcinomas (Thor et al., 1986). The first 3 of these MAbs were reported to be non-reactive with all normal tissues tested, while B72.3 was reported to be weakly reactive with a small number of normal epithelial cells. In order to select the optimal MAbs for immunotherapy experiments, we have tested these 4 MAbs, and here present some of our results. Our conclusion is that all of these MAbs react much more extensively with normal tissues than has been appreciated. These data emphasize the importance of repeated independent investigation of MAb specificity. The antigens recognized by the MAbs used in this study have been described previously in detail. B72.3 recognizes a mucin-like antigen (Johnson et al., 1986), and the determinant recognized has been identified as sialyl-Tn, a core region structure of 0-linked carbohydrate chains (Gold and Mattes, 1988; Kjeldsen et al., 1988). MOv 18 and MOv 19 recognize different determinants on a glycoprotein of MW 38-40 kDa (Miotti et al., 1987). H23 recognizes a 68 kDa glycoprotein present in a preparation described as purified human mammary-tumor virus, isolated from breast-cancer cell line T47D (Keydar et al., 1989).
MATERIAL AND METHODS
Specimens and antibodies Normal human tissues were obtained from surgical specimens or from autopsies performed 6 1 8 hr after death. The surgical specimens included the uterus, cervix and Fallopian tube. The autopsy specimens were obtained primarily from 4 autopsies. Tissues were frozen in 2-methylbutane cooled to - 70°C with dry ice and stored at - 70°C. Cryostat sections, 8-10 p,m thick, were dried thoroughly under a hair dryer and stored at -70°C. Hybridomas B72.3 (an IgG,) and H23 (an IgG,) were obtained from ATCC (Rockville, MD, HB 8108 and HB 8630, respectively). The cells were grown in tissue culture and i.p. in Pristane-primed BALB/c mice, obtained from Charles River (Wilmington, MA). MOv18 (an IgG,) and MOv19 (an IgG,,) ascites fluid were generously provided by Dr. M.I. Colnaghi (Istituto Nazionale per lo Studio e la Cura dei Tumori, Milan, Italy). MH99, an IgG,, Ab to epithelial surface antigen (ESA), which is the same antigen recognized by the Ab 17-1A, was described previously (Thampoe et al., 1988) and was included as a positive control to show cellsurface staining of most epithelial cells. MW207 and MA103 were also described previously (Mattes et al., 1987; Ong and Mattes, 1989). Immunohistology The avidin-biotin complex method (Vector, Burlingame, CA) was used, which has been previously described in detail (Ong and Mattes, 1989). Duplicate sections were tested with each MAb in every experiment. Sections were fixed for 10 min with 95% ethanol, and treated to destroy endogenous peroxidase activity with 0.3% H,O, in methanol, before application of the MAb. H23 was applied as undiluted culture supernatant, and the other MAbs were applied as 1500 dilutions of hybridoma ascites fluid. Negative control MAbs, tested on every specimen, included Ag8 ascites (ATCC TIB 9, an IgG,) and 3B18 culture supernatant (ATCC HB 8654, an IgG,). Although we did not include an IgG,, negative control, in many previous experiments (Mattes er al., 1984, 1985) non-reactive IgG,, MAbs were tested, together with IgG,s, against many normal tissues, and no subclass-dependent variation in background staining was detected. The IgG concentration in ascites fluids was determined by radial immunodiffusion (Williams and Chase, 1971), using a 1:200 dilution of goat anti-mouse IgG (401210, Calbiochem, San Diego, CA) in agarose, and purified MAbs of the appropriate subclasses, prepared as described by Ong and Mattes (1989), as standards. The concentration of IgG in undiluted ascites was 2.3, 3.2, 5.6, 3.1 and 8.1 mg/ml for MAbs MOv-18, MOv-19, B72.3, MH99 and Ag8 respectively, so the 1500 dilution used had 4.6-16.2 p,g/ml, with the negative control MAb having the highest concentration. Sections were counterstained with hematoxylin. 'To whom correspondence and reprint requests should be addressed. Received: April 19, 1990 and in revised form September 1, 1990.
164
STEIN ET AL
Competitive binding inhibition This assay was described in detail by Mattes et al. (1989). Briefly, MW207 was purified, labelled with 1251, and tested for binding to cells of the lung-carcinoma cell line SK-LC- 1 in the presence or absence of various unlabeled MAbs. RESULTS
A summary of immunohistology results on frozen sections is presented in Table I and examples are shown in Figures 1 and 2. In most cases, multiple specimens of each tissue type were tested and consistent results were obtained, with some exceptions that are discussed below. All MAbs reacted exclusively with epithelial cells and/or epithelial secretions, connective tissue being negative. The only exception to this was reactivity of B72.3 with interstitial (Leydig) cells in the testis. Each of the MAbs reacted with a distinct subset of epithelial cells. Except as noted in the following summary, positive MAbs reacted with most or all epithelial cells in the designated tissue. MOv19 reacted in the pancreas strongly with ducts, with weaker staining of acinar cells (Fig. 2); in the cervix and uterus with only the luminal edge and secretions of glandular cells; in the bronchus with glandular cells; in the lung with epithelial cells and/or secretions (Fig. 2); in the prostate with only 5-20% of epithelial cells, which could be ejaculatory ducts; in the skin, with sweat glands only; in the esophagus, with glandular and ductal epithelial cells only; and in several other tissues, predominantly with the luminal border of epithelial cells. MOv18 was similar to MOv19 but considerably weaker (Fig. 2): all tissues stained by MOv18 were stained similarly by MOv19, but a number of tissues negative for MOv18 were reactive with MOv19. There was no indication that MOv18 TABLE I - REACTIVITY OF MAbs WITH CRYOSTAT SECTIONS OF NORMAL HUMAN TISSUES 872 3
Tissue
MOv18 ~ _ _MOvlY
~~
Esophagus Stomach Duodenum
(3)’
Jejunum
(2) (2) (2)
Ileum
(4)
Colon
(2)
Trachea Bronchus Lung Pancreas Salivary gland Liver Skin
-
++ + I ++ ++ +-+ + + + ++ ++
(1)
(2)
(4) (3) (3) (2) (3)
Thyroid (1) Prostate (2) Fallopian tube (3) Uterus Endocervix Exocervix
(3) (2)
Breast Ureter
(2)
Heart
(2)
Spleen LvmDh node
(4) (4) (21
+-+
(1)
+
-
(1)
Testis
n23
++ + -
‘Number of specimens tested. All specimens were from autopsies except the Fallopian tube, utems, endocervix and exocervix. which were surgical specimens.-21n all cases except the testis, only epithelial cells were stained. More information on the identity of stained cells is provided in the text. In all cases except the bronchus, as indicated, consistent results were obtained with the multiple specimens of the same tissue type. + indicates dark staining; indicates relatively light but definite staining.-’Of 2 specimens, one was positive and the other negative. See text for further discussion.
+
+
had a pattern of reactivity distinct from that of MOv19, as opposed to simply reacting more weakly. H23 reacted in the lung with epithelial cells and/or secretions (Fig. 2); in the pancreas with acinar cells; in the salivary gland heterogeneously with 1G20% of epithelial cells; in the prostate with only 5-20% of epithelial cells, which could be ejaculatory ducts; in the uterus, cervix and Fallopian tubes predominantly with the luminal edge and secretions of glandular cells; in the skin, with sebaceous glands only (Fig. 1); and in the esophagus, with suprabasal squamous epithelial cells. B72.3 reacted heterogeneously with the stomach, in which 1G20% of glandular epithelial cells were positive; in the salivary gland, in which 5-10% of epithelial cells were positive; and in the uterus, where a subset of glands and secretions was positive. It reacted predominantly with secretions, and possibly with the luminal border of epithelial cells, in the duodenum, jejunum (Fig. l ) , ileum and cervix. It also reacted with interstitial (Leydig) cells in the testis, and with the suprabasal squamous epithelial cells of the esophagus (Fig. 1). The tissue distribution of MOv19 described here is similar to that of a MAb we described previously, MW207 (Mattes et al., 1987), and the molecular weights of the antigens recognized, 3 8 4 0 kDa, are the same. Therefore, we performed a competitive binding inhibition assay with radiolabelled MW207, which in fact showed that the epitopes recognized by MOv19 and MW207 are the same (Fig. 3). MOv18 produced partial inhibition of binding of MW207 (Fig. 3), which is consistent with results of Miotti et al. (1987) that MOv18 and MOv19 recognize different determinants on the same molecule. Since this study is concerned with the reliability of immunohistology data, and since variation between specimens may be a significant variable, it is of interest to compare results with the different specimens. Of the 20 tissue types for which more than one specimen was tested, results were consistent in the great majority of cases. However, 14 discrepancies were initially found, in which some specimens were positive while other specimens of the same tissue type were negative. The reason for the discordant results in these 14 cases was investigated further. First, the assays were repeated, to determine whether experimental variability could explain the results. In fact, 7 specimens previously found to be negative were subsequently determined to be positive, thereby eliminating the discrepancy between specimens in these 7 cases. Such falsenegative results can probably be attributed primarily to 2 factors. First, in a few assays there was generally weak staining with all MAbs, which might have been due to any of a number of technical problems. Second, in some cases there was unusually high background staining, which can obscure positive results; the reason for sporadically high background has not been determined. One other discrepancy can be attributed to a lack of sebaceous glands in 1 specimen of skin. For the remaining 6 discrepancies, the positive specimen was re-tested and found consistently positive, just as the negative specimen was consistently negative, so the difference could not be attributed to assay variability. In 3 of these cases, a very small fraction of cells were stained in the positive specimen,
