Clin. exp. Inimunol. (1979) 35, 190-195.
Circulating carcinoembryonic antigen immune complexes in sera of patients with carcinomata of the gastrointestinal tract K. K A P S ( P OU I, O U - D O M I N O S & F. A. A N D E R E R Friedrich-Miescher-Laboratorium der Max-Plazuck-Gesellschaft, Tubinigen, WV. Germany
(Received 3 July 1978)
SUMMARY
The sera of patients with histologically proven carcinomata of the gastrointestinal tract were fractionated by gel filtration and the fractions assayed for the presence of free carcinoembryonic antigen (CEA) binding immunoglobulins and CEA immune complexes by radioimmuno-doublediffusion, using '25I-CEA as a marker. In eleven out of thirteen cases with disease recurrence, the presence of CEA-IgM complexes was observed, and in three out of thirteen cases the presence of CEA-IgG complexes, could be demonstrated. Free CEA-binding immunoglobulins could not be detected.
INTRODUCTION Carcinoembryonic antigen (CEA), a tumour-associated antigen first described in adenocarcinomata of the human digestive system of Gold & Freedman (1965), has become a valid parameter in post-operative management of gastrointestinal cancer. Serial determinations of CEA in the sera of patients having undergone primary resection of a tumour have been found to help detect early tumour recurrence. An important question arising from the presence of CEA in patients' sera is whether CEA may induce the format ion of specific antibodies which, at the same time, may block tumour cell destruction by immune lymphocytes. Evidence that 'blocking antibodies' of tumour-bearing individuals are antigen-antibody complexes was given by Sjogren et al. (1971), Baldwin, Price & Robins (1972) and Jose & Seshadri (1974). Experiments to demonstrate CEA-specific antibodies in the sera of patients with gastrointestinal cancer using the zirconyl phosphate gel assay (Lo Gerfo, Herter & Bennett, 1972), passive haemagglutination, immunoadsorption and immunofluorescence (Collatz, von Kleist & Burtin, 1971) were not successful, but when radioimmunoelectrophoresis was used, binding of immunoglobulins to CEA could be shown (Gold, Freedman & Gold, 1972; MacSxw een, 1975). In these cases, however, the authors did not investigate further the possible presence of circulating CEA immune complexes. Evidence for the presence of CEA immune complexes in glomerular deposits has been given for a case of colonic carcinoma with nephrotic syndrome (Costanza et al., 1973). The present study was stimulated by the simple assumption that in the presence of free CEA specifically binding serum antibodies should be found predominantly in the fraction of immune complexes which can be separated from non-bound CEA by gel filtration, as was shown with heterologous CEA immune complexes (Lo Gerfo et al., 1972; Gold & Gold, 1973). In order to obtain more detailed information on the presence of CEA immune complexes and/or free CEA-binding immunoglobulins, investigations u ere performed with the sera of patients who had undergone resection of a histologically proven adenocarcinoma of the gastrointestinal tract, but showed further disease progression. Correspondence: Dr F. A. Anderer, Friedrich- \liescher-laboratorium der Max-Planck-Gesellschaft, 7400 TUbingen
Spemannstraf3e 37-39, WV. Germany. 0099-9104 790020-0190$02.00
(c. 1978 Blackwell Scientific Publications 190
Circulating CEA immunn e complexes
191
MATERIALS AND METHODS Sera. Patients' sera were derived from bleedings taken shortly before or up to 6 months after surgery and ^ ere kept frozen at -20'C until use. Serum fractionation. Fractionation of the sera by gel filtration was performed with ultrogel AcA 22 (LKB, Bromma, Sweden) or Sephadex G200 (Pharmacia, Uppsala, Sweden) using 4 8 x 50 cm columns and 0 02M Tris/HCl pH 7 2 as elution buffer. The columns were loaded with 1-0-2 0 ml serum and eluted at 4VC collecting fractions of 3-5 ml. For the CEA assay with the CEA-Roche-RIA test kit (Roche, Basel, Switzerland) four to eight fractions were pooled and each pool concentrated in a model 8-MC Amicon micro-ultrafiltrations system (Amicon, Oosterhout, Netherlands) to give the original volume of the serum sample. Radioimmunodifusion. This was performed with sera and the pooled and concentrated column fractions, all unabsorbed as well as absorbed with blood group A and B erythrocytes (Behringwerke, Marburg, Germany). CEA-binding immunoglobulins. Precipitin patterns to demonstrate the presence of free CEA-binding immunoglobulins were developed at room temperature for 48 hr by diffusion of '25I-CEA (20,ul in the central well, containing 1 0 ng CEA with 2 x 105 dpm 1251, solely or in mixtures with 20 or 200 ng non-labelled CEA) against 20pl of patients' sera, the corresponding high molecular weight CEA fractions or the other pooled column fractions. After thorough wlashing, the agar gels were dried and autoradiographs were developed on Agfa-Gevaert Osray M 3 X-ray film. CEA immune complexes. To pros e the presence of CEA immune complexes, 30 pl of patients' sera or their corresponding high molecular weight CEA fraction were pre-incubated with 301il 1 251-CEA (containing 1 5 ng CEA with 3 x 105 dpm 1251) to allow an exchange of complex-bound CEA with the 125I-CEA marker. Portions of 20,ul of these mixtures were prediffused for 2 hr before a second set of 2001l portions was added to the same wells simultaneously with the placing of 20 ,ul of monospecific rabbit antisera against human 1gMI or IgG (Behringwerke, Marburg, Germany) into the central weells. Precipitin patterns and autoradiographs were developed as described above.
RESULTS The investigation of the sera of thirteen patients with gastrointestinal cancer shou-ed that (a) twelve sera contained a high molecular weight CEA fraction besides the fraction of normal CEA, (b) normal CEA could be dissociated from the high molecular weight CEA fraction or could be exchanged by 1 25I-CEA, (c) CEA in the high molecular weight CEA fraction was bound to IgM or/and IgG, and (d) in no case could free antibodies capable of binding to CEA be detected. Serum fractionation Fractionation of sera was performed by gel filtration with ultrogel AcA 22 and Sephadex G200. The elution diagrams as well as the CEA content of the pooled and concentrated fractions of three selected sera are given in Fig. 1. All three serum samples showed two distinct maxima of CEA-containing fractions, one corresponding to free CEA and the other exhibiting a higher molecular weight. Free CEA was clearly separated from the high molecular weight CEA fraction. The relative amount of CEA found in the high molecular weight CEA fraction varied from serum to serum; in some cases it even exceeded that of the free CEA (Fig. Ib). 125I-CEA and a mixture of anti-CEA goat serum with 125I-CEA (both CEARoche-RIA test kit) were used as column markers. The quality of the separation of the high molecular weight CEA from normal CEA depended on the ratio of column size and original volume of the serum sample as well as on the amount of CEA present, especially when Sephadex G200 was used. In twelve out of thirteen patients' sera a high molecular weight CEA fraction was detected. The sera of fourteen healthy persons did not contain a high molecular weight CEA fraction. To prove structural stability the high molecular weight CEA fractions were pooled, concentrated and rechromatographed on the same columns. The CEA activity was found only within the same elution range which corresponded fairly well with the elution range of the 1251I-CEA immune complex marker. Different results were obtained when the pooled and concentrated high molecular weight CEA fractions were kept frozen for some time. After thawing, some precipitate was very often observed and the supernatant yielded only the fraction of normal CEA when subjected to gel filtration. Dissociation of normal CEA from the high molecular weight CEA fraction could also be partially or totally achieved by treatment with 1 0 M Tris buffer pH 9-5, 1% SDS, 8 0 M urea or 0 05 M glycin/HCl buffer, pH 2-7. Acid dissociation was connected with 50-80° loss of CEA activity. Considering these findings, the high molecular weight CEA fraction that re have isolated appears to be
192
K. Kapsopoulou-Dominos L F. A. Anderer 3
0
3,to
EC
0
E
OD
DC
U
-.
c
2
E
XL
4c
0
.0E0
Id
.0
11In4
en
4
0 3 2
Fraction number
FIG. 1. Elution profiles obtained after gel filtration of three different sera of patients with gastrointestinal carcinomata. (a) and (b): Gel filtration on Sephadex G200, (c) gel filtration on ultrogel AcA 22. ( ) Absorbance at 280 nm; (............) radioactivity of the CEA immune complex and free CEA markers; (a) CEA content of pooled and concentrated fractions. Column size 4*8x 50 cm, serum sample 1P0 ml, volume of individual fraction 3-5 ml, elution buffer 0'02 M Tris/HCl pH 7-2.
different from the 370,000 Dalton CEA described by Pletsch & Goldenberg (1974), and possibly also from the 'big-CEA' found after gel filtration on Sepharose 6B (Herzog, Hendrick & Franchimont, 1976). Rather, it represents a complex of normal CEA bound to another component. This complex is stable enough for rechromatography but sensitive to freezing, possibly due to the temperature sensitivity of the structure of one or both components. Characterization of CEA immune complexes Further studies were directed at the identification of the component complexing with normal CEA. Special emphasis was put on the question of whether CEA-binding immunoglobulins were present. We selected radioimmuno-double-diffusion in PBS buffered agar gel to detect immunoglobulins which can bind with, or are bound to, CEA using 125I-CEA as a marker. To prove the presence of free CEA-binding immunoglobulins, the original sera of patients and healthy persons, their corresponding pooled and concentrated high molecular weight CEA fractions or the other pooled column fractions were directly assayed with 125I-CEA. In no case could precipitin bands be detected, though controls performed with 1-0-20 ng 125I-CEA and diluted anti-CEA goat serum under the same conditions showed positive autoradiographs of immune precipitin patterns. These results do not
Circulating CEA immune complexes
193
necessarily exclude the presence of minor amounts of free antibody which might be below the limits of sensitivity of the technique applied, i.e. less than the amount of antibody required for the formation of a precipitin band with 10 ng CEA. To prove the presence of CEA immune complexes, unabsorbed and absorbed sera and their corresponding high molecular weight CEA fractions were pre-incubated with 125I-CEA under various conditions to allow an exchange of complex-bound CEA with the '25I-CEA marker. After pre-incubation, the samples were assayed in the double-diffusion test with monospecific rabbit antisera against human IgM or IgG. In Fig. 2 immune precipitin patterns obtained with some selected sera and their corresponding high molecular weight CEA fractions are given.
.4
FIG. 2. Autoradiographs of precipitin patterns obtained after radioimmuno-double-diffusion in PBS buffered agar gel. Central well A contained anti-human IgM rabbit serum, central well B anti-human IgG rabbit serum. The sera and high molecular weight CEA fractions of wells A 1-6 and B 1-6 were pre-incubated with 125I_ CEA. A (1) serum of patient SW, high molecular weight CEA fraction; A (2) serum of a healthy person, blood group B; A (3) serum of patient GF, high molecular weight CEA fraction; A (4) serum of patient SL, high molecular weight CEA fraction; A (5) serum of patient GE, high molecular weight CEA fraction; and A (6) serum of patient GE. In the B series only the patterns of the high molecular weight CEA fractions pre-incubated with 125I-CEA are given: B (1) healthy person, blood group AB; B (2) patient GF; B (3) patient NW; B (4) patient GP; B (5) patient GP, using the high molecular weight CEA fraction after rechromatography on ultrogel AcA 22; B (6) patient SW.
Precipitin patterns with anti-human IgM were observed with eleven out of thirteen patients' sera or their corresponding high molecular weight CEA fractions. All sera and high molecular weight CEA fractions had been absorbed with human blood group A and B erythrocytes before use. Mixtures of 1251CEA with unabsorbed sera of healthy persons, as well as mixtures with the fractions corresponding to the high molecular weight CEA fractions, generally showed positive autoradiographic patterns with antiserum to human IgM. However, after absorption, all sera of healthy persons failed to yield positive autoradiographs of precipitin bands (Fig. 2; A 2), whereas the bands obtained with patients' sera (Fig. 2; A 6) or with their high molecular weight CEA fractions (Fig. 2; A 1, 3, 4 and 5) were still present. In one case, the amount of CEA present in the high molecular weight CEA fraction was too low to yield a significant precipitin band. The antiserum to human IgG did not give positive autoradiographs of precipitin bands with mixtures of 125I-CEA and sera of patients or healthy persons. However, when radioimmunodiffusion was performed with mixtures of 125I-CEA and the fractions corresponding to the high molecular weight CEA fractions, the samples derived from healthy persons did not yield positive bands (Fig. 2; B 1), but in three out ofthirteen samples obtained from patients, positive patterns were observed (Fig. 2; B 3, 4 and 6).
194
K. Kapsopoulou-Dominos & F. A. Anderer
The positive pattern ofsample B 4 (Fig. 2) was still detectable after rechromatography ofthe corresponding high molecular weight CEA fraction on a ultrogel AcA 22 column (Fig. 2; B 5). Negative results were obtained with all the other pooled column fractions of the sera of patients and healthy persons. These findings indicate that in the high molecular weight CEA fraction of patients' sera, normal CEA is bound to immunoglobulins. The detection of CEA immunoglobulin complexes depended predominantly on the conditions of pre-incubation with '25I-CEA in order to achieve optimal CEA exchange. CEA bound to IgM is readily exchanged during incubation at 370C or 4VC for 4-48 hr. However, with CEA bound to IgG we could scarcely detect an exchange with 1251-CEA at 371C; the samples had to be incubated at 40C for at least 24 hr, followed by a 2 hr incubation at 370C, to obtain more distinct precipitin patterns. Even these conditions did not always yield optimal results with one and the same high molecular weight CEA fraction.
DISCUSSION A comparison of our experimental conditions with those used for the detection of anti-CEA antibodies by other investigators (Gold et al., 1972; MacSween, 1975), suggests that it is still possible their positive findings also indicated the presence of CEA immune complexes, since in both cases the sera were preincubated with the 125I-CEA marker. Within the limits of sensitivity of our methods we failed to detect free immunoglobulins capable of binding to CEA, neither in the serum nor in the distinct serum fractions obtained after gel filtration. In the light of a number of reports of cell-mediated immune responses to human tumour associated antigens (Baldwin & Embleton, 1977), the presence of CEA immune complexes in the sera of patients with gastrointestinal carcinomata and their possible action as 'blocking antibodies' may have serious consequences for tumour progression. The group of patients studied by us had undergone primary resection of an adenocarcinoma of the gastrointestinal tract but had shown disease progression and therefore, appeared to be the most promising model system with which to investigate the presence of circulating CEA immune complexes. The fact that in the sera of most of these patients CEA immune complexes could be detected does not yet allow us to postulate a general correlation with disease progression. Tumour antigens, antibodies as well as antigen-antibody complexes, can act as specific blocking factors which impair a cell-mediated immune response against tumour cells (Hellstrom & Nepom, 1977). Thus one can expect that tumour progression might occur without circulating CEA immune complexes or antibodies. Preliminary results indicate that in some cases tumour recurrence occurred with no evidence of circulating immune complexes or antibodies, in spite of a high level of circulating free CEA. On the other hand, it is necessary to ask why antibodies reacting with CEA were formed. Are they only a result of an immunological disorder ? At present, it cannot be determined whether the formation of these antibodies was induced by circulating CEA, by cell-bound CEA or by any cross-reacting antigen different from blood group antigens. More detailed investigations, including follow-up studies with a greater number of patients, are necessary to correlate the presence of CEA immune complexes with tumour progression. Part of this work is already the subject of further studies. The authors would like to thank Dr H. J. Staab for supplying a sample of highly purified CEA and anti-CEA goat serum. REFERENCES
BALDWIN, R.W. & EMBLETON, M.J. (1977) Assessment of cell-mediated immunity to human tumor-associated antigens. Int. Rev. exp. Path. 17, 49. BALDWIN, R.W., PRICE, M.R. & ROBINS, R.A. (1972) Blocking of lymphocyte-mediated cytotoxicity for rat hepatoma cells by tumor-specific antigen-antibody complexes. Nature: New Biology, 238, 185. COLLATZ, E., VON KLEIST, S. & BURTIN, P. (1971) Further
investigations of circulating antibodies in colon cancer patients: On the autoantigenicity of the carcinoembryonic antigen. Int. J. Cancer, 8, 298. COSTANZA, M.E., PINN, V., SCHWARTZ, R.S. & NATHANSON, L. (1973) Carcinoembryonic antigen-antibody complexes in a patient with colonic carcinoma and nephrotic syndrome. N. Eng. J. Med. 289, 520. GOLD, P. & FREEDMAN, S.O. (1965) Demonstration of
Circulating CEA immune complexes tumour-specific antigen in human colonic carcinomata by tolerance and absorption techniques. J. exp. Med. 121, 439. GOLD, J.M., FREEDMAN, S.O. & GOLD, P. (1972) Human anti-CEA antibodies detected by radioimmunoassay. Nature: New Biology, 239, 60-62. GOLD, J.M. & GOLD, P. (1973) The blood group A-like site on the carcinoembryonic antigen. Cancer res. 33, 2821. HELLSTROM, J.E., HELLSTROM, I. & NEPOM, J.T. (1977) Specific blocking factors-are they important? Biochim. biophys. Acta 473, 121. HERZOG, B., HENDRICK, J.C. & FRANCHIMONT, P. (1976) Heterogeneity of carcinoembryonic antigen (CEA) in human serum. Europ. J. Cancer, 12, 657. JOSE, D.G. & SESHADRI, R. (1974) Circulating immune complexes in human neuroblastoma: direct assay and role
c
195
in blocking specific cellular immunity. Int. J. Cancer, 13, 824. Lo GERFo, P., HERTER, F.P. & BENNETT, S.J. (1972) Absence of circulating antibodies to carcinoembryonic antigen in patients with gastrointestinal malignancies. Int. 7. Cancer, 9, 344. MACSWEEN, J.M. (1975) The antigenicity of carcinoembryonic antigen in man. Int. 7. Cancer, 15, 246. PLETSCH, Q. & GOLDENBERG, D.J. (1974) Molecular size of carcino-embryonic antigen in the plasma of patients with malignant disease. 7. Nat. Cancer Inst. 53, 1201. SJOGREN, H.O., HELLSTROM, I., BANSAL, S.C. & HELLSTROM, K.E. (1971) Suggestive evidence that the 'blocking antibodies' of tumour bearing individuals may be antigenantibody complexes. Proc. Nat. Acad. Sci. (Wash.), 68, 1372.
Circulating carcinoembryonic antigen immune complexes in sera of patients with carcinomata of the gastrointestinal tract K. K A P S ( P OU I, O U - D O M I N O S & F. A. A N D E R E R Friedrich-Miescher-Laboratorium der Max-Plazuck-Gesellschaft, Tubinigen, WV. Germany
(Received 3 July 1978)
SUMMARY
The sera of patients with histologically proven carcinomata of the gastrointestinal tract were fractionated by gel filtration and the fractions assayed for the presence of free carcinoembryonic antigen (CEA) binding immunoglobulins and CEA immune complexes by radioimmuno-doublediffusion, using '25I-CEA as a marker. In eleven out of thirteen cases with disease recurrence, the presence of CEA-IgM complexes was observed, and in three out of thirteen cases the presence of CEA-IgG complexes, could be demonstrated. Free CEA-binding immunoglobulins could not be detected.
INTRODUCTION Carcinoembryonic antigen (CEA), a tumour-associated antigen first described in adenocarcinomata of the human digestive system of Gold & Freedman (1965), has become a valid parameter in post-operative management of gastrointestinal cancer. Serial determinations of CEA in the sera of patients having undergone primary resection of a tumour have been found to help detect early tumour recurrence. An important question arising from the presence of CEA in patients' sera is whether CEA may induce the format ion of specific antibodies which, at the same time, may block tumour cell destruction by immune lymphocytes. Evidence that 'blocking antibodies' of tumour-bearing individuals are antigen-antibody complexes was given by Sjogren et al. (1971), Baldwin, Price & Robins (1972) and Jose & Seshadri (1974). Experiments to demonstrate CEA-specific antibodies in the sera of patients with gastrointestinal cancer using the zirconyl phosphate gel assay (Lo Gerfo, Herter & Bennett, 1972), passive haemagglutination, immunoadsorption and immunofluorescence (Collatz, von Kleist & Burtin, 1971) were not successful, but when radioimmunoelectrophoresis was used, binding of immunoglobulins to CEA could be shown (Gold, Freedman & Gold, 1972; MacSxw een, 1975). In these cases, however, the authors did not investigate further the possible presence of circulating CEA immune complexes. Evidence for the presence of CEA immune complexes in glomerular deposits has been given for a case of colonic carcinoma with nephrotic syndrome (Costanza et al., 1973). The present study was stimulated by the simple assumption that in the presence of free CEA specifically binding serum antibodies should be found predominantly in the fraction of immune complexes which can be separated from non-bound CEA by gel filtration, as was shown with heterologous CEA immune complexes (Lo Gerfo et al., 1972; Gold & Gold, 1973). In order to obtain more detailed information on the presence of CEA immune complexes and/or free CEA-binding immunoglobulins, investigations u ere performed with the sera of patients who had undergone resection of a histologically proven adenocarcinoma of the gastrointestinal tract, but showed further disease progression. Correspondence: Dr F. A. Anderer, Friedrich- \liescher-laboratorium der Max-Planck-Gesellschaft, 7400 TUbingen
Spemannstraf3e 37-39, WV. Germany. 0099-9104 790020-0190$02.00
(c. 1978 Blackwell Scientific Publications 190
Circulating CEA immunn e complexes
191
MATERIALS AND METHODS Sera. Patients' sera were derived from bleedings taken shortly before or up to 6 months after surgery and ^ ere kept frozen at -20'C until use. Serum fractionation. Fractionation of the sera by gel filtration was performed with ultrogel AcA 22 (LKB, Bromma, Sweden) or Sephadex G200 (Pharmacia, Uppsala, Sweden) using 4 8 x 50 cm columns and 0 02M Tris/HCl pH 7 2 as elution buffer. The columns were loaded with 1-0-2 0 ml serum and eluted at 4VC collecting fractions of 3-5 ml. For the CEA assay with the CEA-Roche-RIA test kit (Roche, Basel, Switzerland) four to eight fractions were pooled and each pool concentrated in a model 8-MC Amicon micro-ultrafiltrations system (Amicon, Oosterhout, Netherlands) to give the original volume of the serum sample. Radioimmunodifusion. This was performed with sera and the pooled and concentrated column fractions, all unabsorbed as well as absorbed with blood group A and B erythrocytes (Behringwerke, Marburg, Germany). CEA-binding immunoglobulins. Precipitin patterns to demonstrate the presence of free CEA-binding immunoglobulins were developed at room temperature for 48 hr by diffusion of '25I-CEA (20,ul in the central well, containing 1 0 ng CEA with 2 x 105 dpm 1251, solely or in mixtures with 20 or 200 ng non-labelled CEA) against 20pl of patients' sera, the corresponding high molecular weight CEA fractions or the other pooled column fractions. After thorough wlashing, the agar gels were dried and autoradiographs were developed on Agfa-Gevaert Osray M 3 X-ray film. CEA immune complexes. To pros e the presence of CEA immune complexes, 30 pl of patients' sera or their corresponding high molecular weight CEA fraction were pre-incubated with 301il 1 251-CEA (containing 1 5 ng CEA with 3 x 105 dpm 1251) to allow an exchange of complex-bound CEA with the 125I-CEA marker. Portions of 20,ul of these mixtures were prediffused for 2 hr before a second set of 2001l portions was added to the same wells simultaneously with the placing of 20 ,ul of monospecific rabbit antisera against human 1gMI or IgG (Behringwerke, Marburg, Germany) into the central weells. Precipitin patterns and autoradiographs were developed as described above.
RESULTS The investigation of the sera of thirteen patients with gastrointestinal cancer shou-ed that (a) twelve sera contained a high molecular weight CEA fraction besides the fraction of normal CEA, (b) normal CEA could be dissociated from the high molecular weight CEA fraction or could be exchanged by 1 25I-CEA, (c) CEA in the high molecular weight CEA fraction was bound to IgM or/and IgG, and (d) in no case could free antibodies capable of binding to CEA be detected. Serum fractionation Fractionation of sera was performed by gel filtration with ultrogel AcA 22 and Sephadex G200. The elution diagrams as well as the CEA content of the pooled and concentrated fractions of three selected sera are given in Fig. 1. All three serum samples showed two distinct maxima of CEA-containing fractions, one corresponding to free CEA and the other exhibiting a higher molecular weight. Free CEA was clearly separated from the high molecular weight CEA fraction. The relative amount of CEA found in the high molecular weight CEA fraction varied from serum to serum; in some cases it even exceeded that of the free CEA (Fig. Ib). 125I-CEA and a mixture of anti-CEA goat serum with 125I-CEA (both CEARoche-RIA test kit) were used as column markers. The quality of the separation of the high molecular weight CEA from normal CEA depended on the ratio of column size and original volume of the serum sample as well as on the amount of CEA present, especially when Sephadex G200 was used. In twelve out of thirteen patients' sera a high molecular weight CEA fraction was detected. The sera of fourteen healthy persons did not contain a high molecular weight CEA fraction. To prove structural stability the high molecular weight CEA fractions were pooled, concentrated and rechromatographed on the same columns. The CEA activity was found only within the same elution range which corresponded fairly well with the elution range of the 1251I-CEA immune complex marker. Different results were obtained when the pooled and concentrated high molecular weight CEA fractions were kept frozen for some time. After thawing, some precipitate was very often observed and the supernatant yielded only the fraction of normal CEA when subjected to gel filtration. Dissociation of normal CEA from the high molecular weight CEA fraction could also be partially or totally achieved by treatment with 1 0 M Tris buffer pH 9-5, 1% SDS, 8 0 M urea or 0 05 M glycin/HCl buffer, pH 2-7. Acid dissociation was connected with 50-80° loss of CEA activity. Considering these findings, the high molecular weight CEA fraction that re have isolated appears to be
192
K. Kapsopoulou-Dominos L F. A. Anderer 3
0
3,to
EC
0
E
OD
DC
U
-.
c
2
E
XL
4c
0
.0E0
Id
.0
11In4
en
4
0 3 2
Fraction number
FIG. 1. Elution profiles obtained after gel filtration of three different sera of patients with gastrointestinal carcinomata. (a) and (b): Gel filtration on Sephadex G200, (c) gel filtration on ultrogel AcA 22. ( ) Absorbance at 280 nm; (............) radioactivity of the CEA immune complex and free CEA markers; (a) CEA content of pooled and concentrated fractions. Column size 4*8x 50 cm, serum sample 1P0 ml, volume of individual fraction 3-5 ml, elution buffer 0'02 M Tris/HCl pH 7-2.
different from the 370,000 Dalton CEA described by Pletsch & Goldenberg (1974), and possibly also from the 'big-CEA' found after gel filtration on Sepharose 6B (Herzog, Hendrick & Franchimont, 1976). Rather, it represents a complex of normal CEA bound to another component. This complex is stable enough for rechromatography but sensitive to freezing, possibly due to the temperature sensitivity of the structure of one or both components. Characterization of CEA immune complexes Further studies were directed at the identification of the component complexing with normal CEA. Special emphasis was put on the question of whether CEA-binding immunoglobulins were present. We selected radioimmuno-double-diffusion in PBS buffered agar gel to detect immunoglobulins which can bind with, or are bound to, CEA using 125I-CEA as a marker. To prove the presence of free CEA-binding immunoglobulins, the original sera of patients and healthy persons, their corresponding pooled and concentrated high molecular weight CEA fractions or the other pooled column fractions were directly assayed with 125I-CEA. In no case could precipitin bands be detected, though controls performed with 1-0-20 ng 125I-CEA and diluted anti-CEA goat serum under the same conditions showed positive autoradiographs of immune precipitin patterns. These results do not
Circulating CEA immune complexes
193
necessarily exclude the presence of minor amounts of free antibody which might be below the limits of sensitivity of the technique applied, i.e. less than the amount of antibody required for the formation of a precipitin band with 10 ng CEA. To prove the presence of CEA immune complexes, unabsorbed and absorbed sera and their corresponding high molecular weight CEA fractions were pre-incubated with 125I-CEA under various conditions to allow an exchange of complex-bound CEA with the '25I-CEA marker. After pre-incubation, the samples were assayed in the double-diffusion test with monospecific rabbit antisera against human IgM or IgG. In Fig. 2 immune precipitin patterns obtained with some selected sera and their corresponding high molecular weight CEA fractions are given.
.4
FIG. 2. Autoradiographs of precipitin patterns obtained after radioimmuno-double-diffusion in PBS buffered agar gel. Central well A contained anti-human IgM rabbit serum, central well B anti-human IgG rabbit serum. The sera and high molecular weight CEA fractions of wells A 1-6 and B 1-6 were pre-incubated with 125I_ CEA. A (1) serum of patient SW, high molecular weight CEA fraction; A (2) serum of a healthy person, blood group B; A (3) serum of patient GF, high molecular weight CEA fraction; A (4) serum of patient SL, high molecular weight CEA fraction; A (5) serum of patient GE, high molecular weight CEA fraction; and A (6) serum of patient GE. In the B series only the patterns of the high molecular weight CEA fractions pre-incubated with 125I-CEA are given: B (1) healthy person, blood group AB; B (2) patient GF; B (3) patient NW; B (4) patient GP; B (5) patient GP, using the high molecular weight CEA fraction after rechromatography on ultrogel AcA 22; B (6) patient SW.
Precipitin patterns with anti-human IgM were observed with eleven out of thirteen patients' sera or their corresponding high molecular weight CEA fractions. All sera and high molecular weight CEA fractions had been absorbed with human blood group A and B erythrocytes before use. Mixtures of 1251CEA with unabsorbed sera of healthy persons, as well as mixtures with the fractions corresponding to the high molecular weight CEA fractions, generally showed positive autoradiographic patterns with antiserum to human IgM. However, after absorption, all sera of healthy persons failed to yield positive autoradiographs of precipitin bands (Fig. 2; A 2), whereas the bands obtained with patients' sera (Fig. 2; A 6) or with their high molecular weight CEA fractions (Fig. 2; A 1, 3, 4 and 5) were still present. In one case, the amount of CEA present in the high molecular weight CEA fraction was too low to yield a significant precipitin band. The antiserum to human IgG did not give positive autoradiographs of precipitin bands with mixtures of 125I-CEA and sera of patients or healthy persons. However, when radioimmunodiffusion was performed with mixtures of 125I-CEA and the fractions corresponding to the high molecular weight CEA fractions, the samples derived from healthy persons did not yield positive bands (Fig. 2; B 1), but in three out ofthirteen samples obtained from patients, positive patterns were observed (Fig. 2; B 3, 4 and 6).
194
K. Kapsopoulou-Dominos & F. A. Anderer
The positive pattern ofsample B 4 (Fig. 2) was still detectable after rechromatography ofthe corresponding high molecular weight CEA fraction on a ultrogel AcA 22 column (Fig. 2; B 5). Negative results were obtained with all the other pooled column fractions of the sera of patients and healthy persons. These findings indicate that in the high molecular weight CEA fraction of patients' sera, normal CEA is bound to immunoglobulins. The detection of CEA immunoglobulin complexes depended predominantly on the conditions of pre-incubation with '25I-CEA in order to achieve optimal CEA exchange. CEA bound to IgM is readily exchanged during incubation at 370C or 4VC for 4-48 hr. However, with CEA bound to IgG we could scarcely detect an exchange with 1251-CEA at 371C; the samples had to be incubated at 40C for at least 24 hr, followed by a 2 hr incubation at 370C, to obtain more distinct precipitin patterns. Even these conditions did not always yield optimal results with one and the same high molecular weight CEA fraction.
DISCUSSION A comparison of our experimental conditions with those used for the detection of anti-CEA antibodies by other investigators (Gold et al., 1972; MacSween, 1975), suggests that it is still possible their positive findings also indicated the presence of CEA immune complexes, since in both cases the sera were preincubated with the 125I-CEA marker. Within the limits of sensitivity of our methods we failed to detect free immunoglobulins capable of binding to CEA, neither in the serum nor in the distinct serum fractions obtained after gel filtration. In the light of a number of reports of cell-mediated immune responses to human tumour associated antigens (Baldwin & Embleton, 1977), the presence of CEA immune complexes in the sera of patients with gastrointestinal carcinomata and their possible action as 'blocking antibodies' may have serious consequences for tumour progression. The group of patients studied by us had undergone primary resection of an adenocarcinoma of the gastrointestinal tract but had shown disease progression and therefore, appeared to be the most promising model system with which to investigate the presence of circulating CEA immune complexes. The fact that in the sera of most of these patients CEA immune complexes could be detected does not yet allow us to postulate a general correlation with disease progression. Tumour antigens, antibodies as well as antigen-antibody complexes, can act as specific blocking factors which impair a cell-mediated immune response against tumour cells (Hellstrom & Nepom, 1977). Thus one can expect that tumour progression might occur without circulating CEA immune complexes or antibodies. Preliminary results indicate that in some cases tumour recurrence occurred with no evidence of circulating immune complexes or antibodies, in spite of a high level of circulating free CEA. On the other hand, it is necessary to ask why antibodies reacting with CEA were formed. Are they only a result of an immunological disorder ? At present, it cannot be determined whether the formation of these antibodies was induced by circulating CEA, by cell-bound CEA or by any cross-reacting antigen different from blood group antigens. More detailed investigations, including follow-up studies with a greater number of patients, are necessary to correlate the presence of CEA immune complexes with tumour progression. Part of this work is already the subject of further studies. The authors would like to thank Dr H. J. Staab for supplying a sample of highly purified CEA and anti-CEA goat serum. REFERENCES
BALDWIN, R.W. & EMBLETON, M.J. (1977) Assessment of cell-mediated immunity to human tumor-associated antigens. Int. Rev. exp. Path. 17, 49. BALDWIN, R.W., PRICE, M.R. & ROBINS, R.A. (1972) Blocking of lymphocyte-mediated cytotoxicity for rat hepatoma cells by tumor-specific antigen-antibody complexes. Nature: New Biology, 238, 185. COLLATZ, E., VON KLEIST, S. & BURTIN, P. (1971) Further
investigations of circulating antibodies in colon cancer patients: On the autoantigenicity of the carcinoembryonic antigen. Int. J. Cancer, 8, 298. COSTANZA, M.E., PINN, V., SCHWARTZ, R.S. & NATHANSON, L. (1973) Carcinoembryonic antigen-antibody complexes in a patient with colonic carcinoma and nephrotic syndrome. N. Eng. J. Med. 289, 520. GOLD, P. & FREEDMAN, S.O. (1965) Demonstration of
Circulating CEA immune complexes tumour-specific antigen in human colonic carcinomata by tolerance and absorption techniques. J. exp. Med. 121, 439. GOLD, J.M., FREEDMAN, S.O. & GOLD, P. (1972) Human anti-CEA antibodies detected by radioimmunoassay. Nature: New Biology, 239, 60-62. GOLD, J.M. & GOLD, P. (1973) The blood group A-like site on the carcinoembryonic antigen. Cancer res. 33, 2821. HELLSTROM, J.E., HELLSTROM, I. & NEPOM, J.T. (1977) Specific blocking factors-are they important? Biochim. biophys. Acta 473, 121. HERZOG, B., HENDRICK, J.C. & FRANCHIMONT, P. (1976) Heterogeneity of carcinoembryonic antigen (CEA) in human serum. Europ. J. Cancer, 12, 657. JOSE, D.G. & SESHADRI, R. (1974) Circulating immune complexes in human neuroblastoma: direct assay and role
c
195
in blocking specific cellular immunity. Int. J. Cancer, 13, 824. Lo GERFo, P., HERTER, F.P. & BENNETT, S.J. (1972) Absence of circulating antibodies to carcinoembryonic antigen in patients with gastrointestinal malignancies. Int. 7. Cancer, 9, 344. MACSWEEN, J.M. (1975) The antigenicity of carcinoembryonic antigen in man. Int. 7. Cancer, 15, 246. PLETSCH, Q. & GOLDENBERG, D.J. (1974) Molecular size of carcino-embryonic antigen in the plasma of patients with malignant disease. 7. Nat. Cancer Inst. 53, 1201. SJOGREN, H.O., HELLSTROM, I., BANSAL, S.C. & HELLSTROM, K.E. (1971) Suggestive evidence that the 'blocking antibodies' of tumour bearing individuals may be antigenantibody complexes. Proc. Nat. Acad. Sci. (Wash.), 68, 1372.
