- Email: [email protected]
Immunochemical characterization of anti-albumin antibodies in liver diseases
CLINICAL
IMMUNOLOGY
AND
lmmunochemical
DOINA ONICA,’
IMMUNOPATHOLOGY
26, 223-231
(1983)
Characterization of Anti-Albumin in Liver Diseases
ANA CALUGARU,
Department of Immunology,
ILCA MARGINEANU, IRENA BELASCU
GENTIANA
Antibodies
ZAMFIR,
AND
Victor Babe? Institute, Splaiul Independen!ei Buc~lzare.st.Romonicr
99. 76-701
Anti-albumin antibodies (AAA) were isolated from sera of hepatic patients and normal individuals by affinity chromatography on insolubilized glutaraldehyde-treated human albumin. Anti-albumin antibodies were found to belong to IgG and IgM classes in both normal and hepatic patients. The normal level of AAA increased in pathologic conditions, the increase recorded for IgM AAA being higher than that for IgG AAA. The dissociation rate of AAA from the radiolabeled antigen in normal and hepatic sera showed that the affinity of AAA was higher in normal sera than in sera of patients with chronic liver disease and acute viral hepatitis. Anti-albumin antibodies were fractionated into two populations (AAA, and AAA,) by a two-step chromatographic procedure. AAA, and AAA, were found different as regards their affinity for the antigen: specifically. AAA, affinity was higher than that of AAA,. The other difference between AAA, and AAA, might stand in their specificity for the haptenic and structural determinants present in the glutaraldehyde-treated albumin.
INTRODUCTION
Antibodies against glutaraldehyde-polymerized human serum albumin were described in sera of patients with various liver diseases (1, 2). Albumin binding activity mediated by antibodies was also detected in sera of normal individuals by using a radioimmunoassay (3). The anti-albumin antibodies (AAA) have shown specificity for the new antigenic determinants induced in albumin by glutaraldehyde treatment (4). The parallelism noted between the AAA titer and the severity of liver cell dysfunction, as reflected by the clinical diagnosis, showed that AAA are an immunoserological marker for liver cell alteration (5, 6). To complete the data concerning the relevance of AAA for liver disease we compared the immunochemical properties of AAA (level, immunoglobulin class distribution, affinity) in sera of patients with liver diseases (acute viral hepatitis, chronic liver diseases) and of normal individuals. MATERIALS
AND METHODS
Proteins. Human serum albumin (HSA) was purchased from Kabi, Stockholm, Sweden, and horseradish peroxidase (HRP) type II, was obtained from Sigma Chemical Company, St. Louis, Missouri. Glutaraldehyde treatment of HSA and HRP corresponding to the 8/l protein/glutaraldehyde weight ratio was performed ’ To whom correspondence tute, Splaiul Independenfei
should 99, 76201
be addressed: Department Bucharest, Romania.
of Immunology,
Victor
Babes
Insti-
223 0090- 1229/83/020223-09$0 Copyright All rights
1.50/O
0 1983 by Academic Press, Inc of reproduction in any form reserved.
224
ONICA
ET
AL.
as previously described (7). The polymerized proteins were noted as pHSA and pHRP, respectively. Radioactive glutaraldehyde-treated human albumin (1251pHSA) was obtained from HSA previously labeled with carrier-free Na1251 (100 mCiim1) (Radiochemical Centre, Amersham, U.K.) at 0.2 mCi/mg protein by the lactoperoxidase technique (8). The specific activity of 1251-pHSA was 105 cpm/pg pHSA. Sera. Normal sera were obtained from healthy blood donors (8 cases). Pathologic sera with high AAA titers were obtained from 25 patients with various liver diseases. The diagnoses based on clinical and laboratory findings were as follows: acute viral hepatitis (AVH) (14 patients), chronic hepatitis (CH) (8 patients), and liver cirrhosis (LC) (3 patients). Immunoadsorbents. Human serum albumin. pHSA, or pHRP was bound to Sepharose 4B previously activated with CNBr (9). CNBr-activated Sepharose 4B (10 g) was added to a solution of HSA, pHSA, or pHRP (100 mg) in 0.1 M NaHCO, (10 ml). The mixture was kept overnight at 4°C with gentle stirring and was washed successively with phosphate-buffered saline (PBS), 0.1 M glycine buffer, pH 1.8, 3 M KSCN, and barbital buffer, pH 8.6, p = 0.05. Detection ofAAA. Anti-albumin antibodies were detected by immunodiffusion with pHSA according to the method described by Lenkei and Ghetie (2). Radioimmunoassay of AAA was performed as described by Mihaescu et ul. (3). The AAA level was expressed by the lz5T-pHSA binding capacity of the serum, at a dilution of 1:160, i.e., the percentage of the added radioactivity bound to a protein A-containing Staphylococcus aweus (SA) suspension (strain Cowan-l). Ajjkitp chromutogruphy 011pHSA-Sepharose 4B. Pathologic serum (0.5 ml) or normal serum (1 ml) was passed through a pHSA-Sepharose 4B column (1.6 x 5.5 cm) equilibrated with barbital buffer, pH 8.6, p = 0.05, and incubated for 1 hr at room temperature. The unadsorbed material washed out with the same buffer was designated as effluent serum and was brought to 10 times the volume of the initial serum. The antibodies adsorbed on the column were eluted with 3 M KSCN and 2 ml samples were collected, dialyzed for 24 hr against several changes of PBS. and measured at 280 nm. The samples containing protein were mixed and further dialyzed against barbital buffer (24-48 hr). The antibody solution was finally adjusted with barbital buffer to 20 times the initial serum volume for binding studies, or was concentrated (Diaflo membrane PM-lo, Amicon, Lexington, Mass.) to five times the original serum volume for quantitative immunoglobulin determinations. As a control, pathologic serum was chromatographed on a HSASepharose 4B column (1.6 x 5.5 cm). Fractionation oj’AAA. Pathologic serum (0.5 ml) was passed through a pHRPSepharose 4B column (1.6 x 4.5 cm) equilibrated with barbital buffer, pH 8.6, Al. = 0.05. and incubated for 1 hr at room temperature. The effluent serum from the pHRP-Sepharose 4B column was then passed through a pHSA-Sepharose 4B column (1.6 x 5.5 cm) equilibrated with the same buffer. The antibodies adsorbed on the columns were eluted with 3 M KSCN and 2-ml samples were separately collected, dialyzed for 24 hr against several changes of PBS, and measured at 280 nm. The samples containing protein were mixed, then dialyzed against barbital buffer (24-48 hr), and finally adjusted with the same buffer to 20 times the initial
CHARACTERIZATION
OF
ANTI-ALBUMIN
ANTIBODIES
225
serum volume for binding studies or were concentrated to five times the original serum volume for quantitative immunoglobulin determinations. The antibody fractions eluted from pHRP-Sepharose 4B and from pHSA-Sepharose 4B were designated AAA, and AAA,, respectively. Radial immunodiffuusion. Quantitative immunoglobulin determinations were performed by using the technique of Mancini et al. (10). The lowest limit for IgG estimation was 10 &ml, for IgM 25 &ml, and for IgA 5 &ml. Goat serum anti-human IgG, IgM, and IgA (Cantacuzino Institute, Bucharest, Romania) were used. Binding assay. The test was performed as described by Mihaescu et al. (3). Briefly, 0.25 ml 12”I-pHSA (0.15 pg) was added to 0.25 ml of each serial dilution of hepatic serum or of purified anti-albumin antibody preparation. All dilutions were made in barbital buffer, pH 8.6, ,u = 0.05. After overnight incubation at 4”C, a SA suspension in barbital buffer (0.25 ml containing lo9 bacteria) was added; then the samples were incubated for 30 min at room temperature and centrifuged. The pellets were washed with 2 ml cold barbital buffer containing 0.1% ovalbumin and the radioactivity was counted. The correction for nonspecific binding of lzsIpHSA (about 7%) was made by using 0.1% ovalbumin in barbital buffer instead of serum or purified antibody preparation. The ABC-33 endpoint (the reciprocal of serum or purified antibody dilution at which 33% of the added ““I-pHSA was specifically bound) was calculated according to Minden and Farr (11). Dissociation assny. Dissociation of 1z51-pHSA-antibody complexes was carried out according to the procedure described by Minden and Farr (11) using SA instead of ammonium sulfate. Briefly, 0.25 ml Y-pHSA (0.15-0.5 pg) in barbital buffer, pH 8.6, p = 0.05, was added to 0.25 ml of the.appropriate dilution, in the same buffer, of hepatic serum or purified antibody preparation previously found to bind about 40% of the labeled antigen. After overnight incubation at 4”C, 0.25 ml of the unlabeled pHSA (1.5-5 Fg) was added at room temperature. At a certain time interval 0.25 ml SA suspension (10” bacteria) was added. After incubation for 30 min at room temperature the sample was centrifuged, the pellet was washed with 2 ml cold barbital buffer containing 0.1% ovalbumin, and the radioactivity was counted. The reaction with the unlabeled pHSA was stopped with SA at the following time intervals: 5 and 30 min, and 2, 4, 24, 48, and 72 hr. The percentage of i”51-pHSA still bound to the antibody was calculated according to the formula 100 x A/B, where B is the bound radioactivity in the absence of pHSA and A is the radioactivity specifically bound by the antibody at a certain time interval after addition of excess unlabeled pHSA. Using a log-linear scale, 100 x A/B was plotted against time. RESULTS [sofatim
qf‘ AAA
Anti-albumin antibodies were isolated from sera of hepatic patients by affinity chromatography on pHSA-Sepharose 4B. The AAA activity of the hepatic serum was retained by the pHSA-Sepharose 4B column so that the effluent serum no longer bound the radiolabeled pHSA, as shown by radioimmunoassay using SA.
226
ONICA
ET AL.
The protein eluted from the pHSA-Sepharose 4B column had AAA activity since it was able to bind the radiolabeled pHSA. The protein eluted from the HSASepharose 4B column, used as a control, presented only 5.8% of the AAA activity shown by the protein recovered from the pHSA-Sepharose 4B column. This value was obtained by comparing the lz51-pHSA binding activity of both protein preparations (ABC-101 (Fig. 1). The analysis of the x”51-pHSA binding curves of the isolated antibody preparation and of the initial serum showed that their ABC-33 endpoint values were different (see Fig. I). indicating a loss of AAA activity during the purification procedure. Thus the comparison of the ABC-33 endpoint values of three hepatic sera and the corresponding purified anti-albumin antibody preparations indicated that 34 t 2% of AAA activity had remained after the purification procedure. The capacity of the purified antibody to bind the radiolabeled antigen was much reduced by the concentration on Diaflo membrane PM-lo. For instance a fourfold concentration of an antibody solution changed its ABC-33 endpoint value from 160 to 36, which indicates a considerable loss of AAA activity. Itntturnoglobulitt
Class of AAA
The immunoelectrophoretic analysis showed that the purified anti-albumin antibodies belonged to the IgG and IgM classes. No reaction was seen with anti-
lx)
120
l.LO
1.K
1,160
1320
16W
1~1280
l-2560
DllUtlOn
FIG. 1. Binding of lz51-pHSA to the unfractionated serum of a patient with AVH (0) and to the purified AAA isolated by chromatography on a pHSA-Sepharose 4B column (0). The protein isolated from the hepatic serum by chromatography on a HSA-Sepharose 4B column was used as a control (A). The samples contained 0.15 pg rz51-pHSA and serial dilutions of the hepatic serum or purified antibody preparation. The volume of the antibody solution was adjusted to 20 times the original serum volume. ABC-33 represents the reciprocal dilution at which 33% of the added ?-pHSA was specifically bound.
CHARACTERIZATION
OF
ANTI-ALBUMIN
227
ANTIBODIES
human IgA serum. The immunoelectrophoretic pattern seemed to indicate that the IgG present in the purified antibody preparation was less heterogeneous than normal human IgG.
The AAA concentration of both IgG and IgM classes, in normal and hepatic sera, is presented in Table 1. The IgG AAA level in hepatic sera was about 2.5 times higher than in the normal sera, while that of the IgM AAA was 5 times higher. The proportion of immunoglobulin classes in AAA was calculated by taking into account the molar concentrations of IgG AAA and of IgM AAA. It was found that in hepatic sera 82% of AAA belonged to the IgG class and 18% to the IgM class. In normal sera IgG AAA represented 90% and IgM AAA 10%. Dissociatiott
Rates
The dissociation rates of the lz51-pHSA-antibody complexes in normal and hepatic sera after addition of an excess of unlabeled pHSA are presented in Fig. 2. The dissociation curves indicated the diversity of AAA with respect to the strength of the union between antigen and antibody in all the sera tested. The sera contained various proportions of at least two antibody populations: one of which dissociated rather freely from the lz51-pHSA during the first hours of the experiment, and another with a half dissociation time greater than 80 hr. The curves also showed the slowest lz51-pHSA dissociation rate was found in normal sera, and the most rapid rate in sera of patients with AVH. Frmtionution
qf’ AAA
Anti-albumin antibodies were fractionated into two populations, AAA, and AAA2, by successive chromatography of hepatic sera on pHRP-Sepharose 4B and pHSA-Sepharose 4B columns. It was found that AAA, remaining in the effluent serum from the pHRP-Sepharose 4B column did not react with pHRP-Sepharose 4B by rechromatography. The distribution of IgG and IgM in the AAA, and AAAB fractions is presented in
CONCENTRATION
Serum” Normal Pathologic
Number of cases 3 10
TABLE 1 OF IgG AND IgM ANTI-ALBUMIN NORMAL AND HEPATIC SERA
ANTIBODIES
IgG AAA” w&l 60.3 k 2.6 153.7 k 47.9
IN
IgM AAA” moY1 x 10W 40.2 + 1.8 102.5 2 32.0
dml 40.7 * 11.3 198.3 t 110.6
m&l
x 10-s
4.5 t 1.3 22.0 t 12.3
‘I Normal sera do not precipitate with pHSA in immunodiffusion, but bind 17% of iz51-pHSA at a serum dilution of 1: 160. Pathologic sera from patients with AVH (5 cases), CH (3 cases), and LC (2 cases) give an intense positive reaction in immunodilfnsion with pHSA (the lowest pHSA concentration reacting with undiluted serum was 20 &ml) and bind about 65% of ‘a?-pHSA at a serum dilution of 1:160. ’ The amount of IgG and IgM in purified AAA preparations, as determined by radial immunodiffnsion, is expressed in pg or mol per unit volume of the original serum. The molar concentration of IgG AAA and IgM AAA was calculated by considering the molecular weight of IgG to be 150.000 daltons and that of IgM 900,000 daltons.
ONICA
ET
Al..
FIG. 2. Dissociation rates of ““I-pHSA from complexes in normal and hepatic sera. The curves represent the median values for 5 normal cases (x). for 5 cases of chronic liver diseases (4 cases of CH and 1 case of LC) (O), and for 6 cases of AVH (0). Dissociation of the ‘%pHSA (0.5 wg) was determined after addition of unlabeled excess antigen (5 pg). Sera with high AAA titers were used.
Table 2. It can be seen that the pHRP-Sepharose 4B column retained 40-73s of the IgG AAA and almost all the IgM AAA. The dissociation curves of lz51-pHSA-antibody complexes, either with AAA, or with AAA,, are presented in Fig. 3. The data in Fig. 3 indicate that the two antibody populations were different as regards their affinity for the antigen, so that the dissociation of the radiolabeled antigen was faster from the AAAz than from the AAA, fraction. TABLE FRACTIONATION
AAA, Pathologic serum 11497” I 203Sh 13121” 13491”
bound
to pHRP-Sepharose
IgG Wml)” 183 175 83 360
U Sera of patients with AVH. * Serum of a patient with CH. ” The amount of IgG and IgM in AAA, sion. is expressed in yg per unit volume
2
OF ANTI-ALBUMIN
IgM
4B
(FLg/ml)” 331 75 104 155
and AAA, fractions. of the original serum.
ANTIBODIES
AAA,
bound
IgG (&ml)?
to pHSA-Sepharose IgM
129 105 123 130
as determined
4B
(&ml)’ 28 -
by radial
immunodiffu-
CHARACTERIZATION
OF
ANTI-ALBUMIN
229
ANTIBODIES
1
0
23
iv
30
Lo
50
bo
70
613
Trme (hwrsl
3. Dissociation rates of rz51-pHSA from complexes containing either the AAA, (0) or the AAA, fraction (0) isolated from the serum of a patient with AVH. The AAA, fraction was eluted from the pHRP-Sepharose 4B column. The AAAz fraction, not reacting with the pHRP-Sepharose 4B column, was isolated by chromatography on a pHSA-Sepharose 4B column. Dissociation of the rz51-pHSA (0.15 pg) was determined after addition of unlabeled excess antigen (1.5 pg). FIG.
DISCUSSION
Anti-albumin antibodies were isolated from sera of hepatic patients and normal individuals by afftnity chromatography on insolubilized pHSA. These antibodies seem to be very sensitive since their capacity to bind the radiolabeled antigen was considerably reduced during the purification procedure and subsequent concentration. The data confirm the earlier findings by radioimmunoelectrophoresis (2) and by indirect immunofluorescence (12) that AAA in hepatic patients belong to the IgG and IgM classes. The relative electrophoretic homogeneity of IgG AAA suggests that the in viro alteration of albumin molecules, stimulating antibody synthesis, can be represented by a limited number of new antigenic determinants, such as pyridinium structures as previously indicated (4). The isolation of AAA from normal sera confirmed the data obtained by Mihaescu et al. (3) who found AAA activity in normal sera by radioimmunoassay. Anti-albumin antibodies in normal sera were also found to belong to the IgG and IgM classes. The level of AAA increased in pathologic conditions, the increase recorded for IgM AAA was higher than that found for IgG AAA. Our preliminary data showed that the in-
230
ONICA
ET
AL
crease in IgM antibody was more marked in acute viral hepatitis than in chronic liver diseases. Since the number of sera investigated was rather small, the study should be extended to find out if the differences observed are also valid for a greater number of cases. The ability of Staplr~lococ~cus protein A (SPA) to bind immunoglobulins was used in the dissociation assay to separate the free antigen (pHSA) from the AAA-pHSA complexes. The use of SA as adsorbent for the AAA-pHSA complexes is justified by the finding that the major immunoglobulin class in AAA, in both normal and hepatic patients, was IgG (82-90%) which has a high affinity for SpA (13). Immune complexes containing IgM AAA will also bind to SA since IgM2 subclass reacts with SpA (14, 15), and IgM1 subclass, normally not reacting with SpA, might become reactive after binding to the antigen. This supposition is based on the results of Barkas and Watson (16) showing that SPA-reactive sites can be generated in chicken IgG antibodies (normally not reacting with SpA) by antigen binding. In addition Reisberg and Rossen (17) have shown that SpASepharose can be used to isolate immune complexes prepared with antibodies which do not bind to SpA in monomer form, as for instance goat IgG antibodies. The study of the dissociation rate of AAA from the radiolabeled antigen showed a difference between the normal and hepatic sera; specifically, the affinity of AAA for the antigen was highest in normal sera. This observation is in agreement with the hypothesis that AAA represent, in normal sera, physiological autoantibodies involved in the removal of altered (aged) albumin molecules from the circulation ( 18). According to this hypothesis in the presence of circulating autoantibodies the modified albumin molecules cannot reach the B cells which have receptors for the new antigenic determinants of the aged albumin since they become complexed by the corresponding autoantibodies as soon as they appear in the circulation. It was also supposed that the concentration of the modified albumin increases in pathological conditions (liver diseases) stimulating an active synthesis of AAA by the immunocompetent cells. In the presence of excess circulating modified albumin high-affinity AAA may be removed in a complexed form leaving predominantly low-affinity AAA in circulation, as seems to be the case in liver diseases. On the other hand. AAA are heterogeneous concerning the affinity for the antigen since in the same serum they can have a lower and a higher affinity. The separation of AAA into two populations (AAA, and AAA,) was possible owing to the cross-reacting antigenic determinants present on different proteins treated with glutaraldehyde, for instance peroxidase and albumin. The AAA, purified in the first chromatographic step by reaction with the modified peroxidase were of higher affinity than AAA, recovered by interaction with the modified albumin. Other possible differences between AAA, and AAA, might be related to their specificity. We can assume that the cross-reacting antigenic determinants present on the modified albumin and peroxidase are represented by the chemical groups induced in the proteins by glutaraldehyde treatment (pyridinium haptenic determinants) (7). Therefore, AAA, might recognize the haptenic determinants, while AAA, might be specific for antigenic determinants which appear during the polymerization process (structural determinants) (7) and depend on the particular structure of albumin.
CHARACTERIZATION
OF
ANTI-ALBUMIN
231
ANTIBODIES
ACKNOWLEDGMENTS The authors are grateful indebted to Mrs. Mariana
to Dr. Victor Gherie for valuable criticism Caralicea for excellent technical assistance.
and suggestions.
They
are also
REFERENCES I. 2. 3. 4. 5. 6. 7. 8. 9. 10. Il. 12. 13. 14. 15. 16. 17. 18.
Lenkei. R.. MoIa, G.. Dan, M. E., and Laky, M.. Rev. Roum. Biochim. 11, 271, 1974. Lenkei. R., and Ghetie, V.. J. Immunol. Methods, 16, 23, 1977. Mihaescu, S.. Lenkei, R.. and Ghetie, V., J. Immunol. Methods. 42, 187, 1981. Onica. D.. Margineanu, I., and Lenkei. R.. Mol. Zmmw~ol. 18, 807. 1981. Lenkei, R., Revue Roum. Med. Interne 18, 129, 1980. Lenkei, R.. Buligescu, L.. Belascu, I., Pospai. D.. and Dobre. I.. C&r. E.wp. Immlrt)o[. 43, 381. 1981. Onica, D., Lenkei. R., and Ghetie. V., Immtrnochrnzistrv 15, 687, 1978. Marchalonis. J. .I., Biochem. J. 113, 229, 1969. Fuchs. S.. and Sela, M., In “Handbook of Experimental Immunology” (D. M. Weir, Ed.), pp. 11. I ~ 11, Blackwell, Oxford, 1973. Mancini. G., Carbonara, A. 0.. and Heremans, J. F., Imt~zunochemi.ytr~ 2, 235, 1965. Minden, P., and Farr, R. S., In “Handbook of Experimental Immuncdogy” (D. M. Weir, Ed.), pp. 15. I - 15, Blackwell, Oxford, 1973. Thung, S. N.. and Gerber. M. A., Gastroenterology 80, 260, 1981. Kronvall. G., Quie, P. G., and Williams. R. C., J. Immune/. 104, 273, 1970. Harboe, M., and Foiling, I., Stand. .I. Immrmo/. 3, 471. 1974. Saltvedt, E.. and Harboe. M.. Scand. J. Immrrnol. 5, 1103, 1976. Barkas, T., and Watson, C. M. J.. Immunology 36, 557, 1979. Reisbcrg, M. A.. and Rossen. R. D., C/in. Exp. Immunol. 46, 443, 1981. Ghelie, V.. Onica, D.. Lenkei. R., and Margineanu. I.. Mrch. Aqeing De\,. 17, 27. 1981.
Received
April
29, 1982: accepted
with
revisions
July
14. 1982.
IMMUNOLOGY
AND
lmmunochemical
DOINA ONICA,’
IMMUNOPATHOLOGY
26, 223-231
(1983)
Characterization of Anti-Albumin in Liver Diseases
ANA CALUGARU,
Department of Immunology,
ILCA MARGINEANU, IRENA BELASCU
GENTIANA
Antibodies
ZAMFIR,
AND
Victor Babe? Institute, Splaiul Independen!ei Buc~lzare.st.Romonicr
99. 76-701
Anti-albumin antibodies (AAA) were isolated from sera of hepatic patients and normal individuals by affinity chromatography on insolubilized glutaraldehyde-treated human albumin. Anti-albumin antibodies were found to belong to IgG and IgM classes in both normal and hepatic patients. The normal level of AAA increased in pathologic conditions, the increase recorded for IgM AAA being higher than that for IgG AAA. The dissociation rate of AAA from the radiolabeled antigen in normal and hepatic sera showed that the affinity of AAA was higher in normal sera than in sera of patients with chronic liver disease and acute viral hepatitis. Anti-albumin antibodies were fractionated into two populations (AAA, and AAA,) by a two-step chromatographic procedure. AAA, and AAA, were found different as regards their affinity for the antigen: specifically. AAA, affinity was higher than that of AAA,. The other difference between AAA, and AAA, might stand in their specificity for the haptenic and structural determinants present in the glutaraldehyde-treated albumin.
INTRODUCTION
Antibodies against glutaraldehyde-polymerized human serum albumin were described in sera of patients with various liver diseases (1, 2). Albumin binding activity mediated by antibodies was also detected in sera of normal individuals by using a radioimmunoassay (3). The anti-albumin antibodies (AAA) have shown specificity for the new antigenic determinants induced in albumin by glutaraldehyde treatment (4). The parallelism noted between the AAA titer and the severity of liver cell dysfunction, as reflected by the clinical diagnosis, showed that AAA are an immunoserological marker for liver cell alteration (5, 6). To complete the data concerning the relevance of AAA for liver disease we compared the immunochemical properties of AAA (level, immunoglobulin class distribution, affinity) in sera of patients with liver diseases (acute viral hepatitis, chronic liver diseases) and of normal individuals. MATERIALS
AND METHODS
Proteins. Human serum albumin (HSA) was purchased from Kabi, Stockholm, Sweden, and horseradish peroxidase (HRP) type II, was obtained from Sigma Chemical Company, St. Louis, Missouri. Glutaraldehyde treatment of HSA and HRP corresponding to the 8/l protein/glutaraldehyde weight ratio was performed ’ To whom correspondence tute, Splaiul Independenfei
should 99, 76201
be addressed: Department Bucharest, Romania.
of Immunology,
Victor
Babes
Insti-
223 0090- 1229/83/020223-09$0 Copyright All rights
1.50/O
0 1983 by Academic Press, Inc of reproduction in any form reserved.
224
ONICA
ET
AL.
as previously described (7). The polymerized proteins were noted as pHSA and pHRP, respectively. Radioactive glutaraldehyde-treated human albumin (1251pHSA) was obtained from HSA previously labeled with carrier-free Na1251 (100 mCiim1) (Radiochemical Centre, Amersham, U.K.) at 0.2 mCi/mg protein by the lactoperoxidase technique (8). The specific activity of 1251-pHSA was 105 cpm/pg pHSA. Sera. Normal sera were obtained from healthy blood donors (8 cases). Pathologic sera with high AAA titers were obtained from 25 patients with various liver diseases. The diagnoses based on clinical and laboratory findings were as follows: acute viral hepatitis (AVH) (14 patients), chronic hepatitis (CH) (8 patients), and liver cirrhosis (LC) (3 patients). Immunoadsorbents. Human serum albumin. pHSA, or pHRP was bound to Sepharose 4B previously activated with CNBr (9). CNBr-activated Sepharose 4B (10 g) was added to a solution of HSA, pHSA, or pHRP (100 mg) in 0.1 M NaHCO, (10 ml). The mixture was kept overnight at 4°C with gentle stirring and was washed successively with phosphate-buffered saline (PBS), 0.1 M glycine buffer, pH 1.8, 3 M KSCN, and barbital buffer, pH 8.6, p = 0.05. Detection ofAAA. Anti-albumin antibodies were detected by immunodiffusion with pHSA according to the method described by Lenkei and Ghetie (2). Radioimmunoassay of AAA was performed as described by Mihaescu et ul. (3). The AAA level was expressed by the lz5T-pHSA binding capacity of the serum, at a dilution of 1:160, i.e., the percentage of the added radioactivity bound to a protein A-containing Staphylococcus aweus (SA) suspension (strain Cowan-l). Ajjkitp chromutogruphy 011pHSA-Sepharose 4B. Pathologic serum (0.5 ml) or normal serum (1 ml) was passed through a pHSA-Sepharose 4B column (1.6 x 5.5 cm) equilibrated with barbital buffer, pH 8.6, p = 0.05, and incubated for 1 hr at room temperature. The unadsorbed material washed out with the same buffer was designated as effluent serum and was brought to 10 times the volume of the initial serum. The antibodies adsorbed on the column were eluted with 3 M KSCN and 2 ml samples were collected, dialyzed for 24 hr against several changes of PBS. and measured at 280 nm. The samples containing protein were mixed and further dialyzed against barbital buffer (24-48 hr). The antibody solution was finally adjusted with barbital buffer to 20 times the initial serum volume for binding studies, or was concentrated (Diaflo membrane PM-lo, Amicon, Lexington, Mass.) to five times the original serum volume for quantitative immunoglobulin determinations. As a control, pathologic serum was chromatographed on a HSASepharose 4B column (1.6 x 5.5 cm). Fractionation oj’AAA. Pathologic serum (0.5 ml) was passed through a pHRPSepharose 4B column (1.6 x 4.5 cm) equilibrated with barbital buffer, pH 8.6, Al. = 0.05. and incubated for 1 hr at room temperature. The effluent serum from the pHRP-Sepharose 4B column was then passed through a pHSA-Sepharose 4B column (1.6 x 5.5 cm) equilibrated with the same buffer. The antibodies adsorbed on the columns were eluted with 3 M KSCN and 2-ml samples were separately collected, dialyzed for 24 hr against several changes of PBS, and measured at 280 nm. The samples containing protein were mixed, then dialyzed against barbital buffer (24-48 hr), and finally adjusted with the same buffer to 20 times the initial
CHARACTERIZATION
OF
ANTI-ALBUMIN
ANTIBODIES
225
serum volume for binding studies or were concentrated to five times the original serum volume for quantitative immunoglobulin determinations. The antibody fractions eluted from pHRP-Sepharose 4B and from pHSA-Sepharose 4B were designated AAA, and AAA,, respectively. Radial immunodiffuusion. Quantitative immunoglobulin determinations were performed by using the technique of Mancini et al. (10). The lowest limit for IgG estimation was 10 &ml, for IgM 25 &ml, and for IgA 5 &ml. Goat serum anti-human IgG, IgM, and IgA (Cantacuzino Institute, Bucharest, Romania) were used. Binding assay. The test was performed as described by Mihaescu et al. (3). Briefly, 0.25 ml 12”I-pHSA (0.15 pg) was added to 0.25 ml of each serial dilution of hepatic serum or of purified anti-albumin antibody preparation. All dilutions were made in barbital buffer, pH 8.6, ,u = 0.05. After overnight incubation at 4”C, a SA suspension in barbital buffer (0.25 ml containing lo9 bacteria) was added; then the samples were incubated for 30 min at room temperature and centrifuged. The pellets were washed with 2 ml cold barbital buffer containing 0.1% ovalbumin and the radioactivity was counted. The correction for nonspecific binding of lzsIpHSA (about 7%) was made by using 0.1% ovalbumin in barbital buffer instead of serum or purified antibody preparation. The ABC-33 endpoint (the reciprocal of serum or purified antibody dilution at which 33% of the added ““I-pHSA was specifically bound) was calculated according to Minden and Farr (11). Dissociation assny. Dissociation of 1z51-pHSA-antibody complexes was carried out according to the procedure described by Minden and Farr (11) using SA instead of ammonium sulfate. Briefly, 0.25 ml Y-pHSA (0.15-0.5 pg) in barbital buffer, pH 8.6, p = 0.05, was added to 0.25 ml of the.appropriate dilution, in the same buffer, of hepatic serum or purified antibody preparation previously found to bind about 40% of the labeled antigen. After overnight incubation at 4”C, 0.25 ml of the unlabeled pHSA (1.5-5 Fg) was added at room temperature. At a certain time interval 0.25 ml SA suspension (10” bacteria) was added. After incubation for 30 min at room temperature the sample was centrifuged, the pellet was washed with 2 ml cold barbital buffer containing 0.1% ovalbumin, and the radioactivity was counted. The reaction with the unlabeled pHSA was stopped with SA at the following time intervals: 5 and 30 min, and 2, 4, 24, 48, and 72 hr. The percentage of i”51-pHSA still bound to the antibody was calculated according to the formula 100 x A/B, where B is the bound radioactivity in the absence of pHSA and A is the radioactivity specifically bound by the antibody at a certain time interval after addition of excess unlabeled pHSA. Using a log-linear scale, 100 x A/B was plotted against time. RESULTS [sofatim
qf‘ AAA
Anti-albumin antibodies were isolated from sera of hepatic patients by affinity chromatography on pHSA-Sepharose 4B. The AAA activity of the hepatic serum was retained by the pHSA-Sepharose 4B column so that the effluent serum no longer bound the radiolabeled pHSA, as shown by radioimmunoassay using SA.
226
ONICA
ET AL.
The protein eluted from the pHSA-Sepharose 4B column had AAA activity since it was able to bind the radiolabeled pHSA. The protein eluted from the HSASepharose 4B column, used as a control, presented only 5.8% of the AAA activity shown by the protein recovered from the pHSA-Sepharose 4B column. This value was obtained by comparing the lz51-pHSA binding activity of both protein preparations (ABC-101 (Fig. 1). The analysis of the x”51-pHSA binding curves of the isolated antibody preparation and of the initial serum showed that their ABC-33 endpoint values were different (see Fig. I). indicating a loss of AAA activity during the purification procedure. Thus the comparison of the ABC-33 endpoint values of three hepatic sera and the corresponding purified anti-albumin antibody preparations indicated that 34 t 2% of AAA activity had remained after the purification procedure. The capacity of the purified antibody to bind the radiolabeled antigen was much reduced by the concentration on Diaflo membrane PM-lo. For instance a fourfold concentration of an antibody solution changed its ABC-33 endpoint value from 160 to 36, which indicates a considerable loss of AAA activity. Itntturnoglobulitt
Class of AAA
The immunoelectrophoretic analysis showed that the purified anti-albumin antibodies belonged to the IgG and IgM classes. No reaction was seen with anti-
lx)
120
l.LO
1.K
1,160
1320
16W
1~1280
l-2560
DllUtlOn
FIG. 1. Binding of lz51-pHSA to the unfractionated serum of a patient with AVH (0) and to the purified AAA isolated by chromatography on a pHSA-Sepharose 4B column (0). The protein isolated from the hepatic serum by chromatography on a HSA-Sepharose 4B column was used as a control (A). The samples contained 0.15 pg rz51-pHSA and serial dilutions of the hepatic serum or purified antibody preparation. The volume of the antibody solution was adjusted to 20 times the original serum volume. ABC-33 represents the reciprocal dilution at which 33% of the added ?-pHSA was specifically bound.
CHARACTERIZATION
OF
ANTI-ALBUMIN
227
ANTIBODIES
human IgA serum. The immunoelectrophoretic pattern seemed to indicate that the IgG present in the purified antibody preparation was less heterogeneous than normal human IgG.
The AAA concentration of both IgG and IgM classes, in normal and hepatic sera, is presented in Table 1. The IgG AAA level in hepatic sera was about 2.5 times higher than in the normal sera, while that of the IgM AAA was 5 times higher. The proportion of immunoglobulin classes in AAA was calculated by taking into account the molar concentrations of IgG AAA and of IgM AAA. It was found that in hepatic sera 82% of AAA belonged to the IgG class and 18% to the IgM class. In normal sera IgG AAA represented 90% and IgM AAA 10%. Dissociatiott
Rates
The dissociation rates of the lz51-pHSA-antibody complexes in normal and hepatic sera after addition of an excess of unlabeled pHSA are presented in Fig. 2. The dissociation curves indicated the diversity of AAA with respect to the strength of the union between antigen and antibody in all the sera tested. The sera contained various proportions of at least two antibody populations: one of which dissociated rather freely from the lz51-pHSA during the first hours of the experiment, and another with a half dissociation time greater than 80 hr. The curves also showed the slowest lz51-pHSA dissociation rate was found in normal sera, and the most rapid rate in sera of patients with AVH. Frmtionution
qf’ AAA
Anti-albumin antibodies were fractionated into two populations, AAA, and AAA2, by successive chromatography of hepatic sera on pHRP-Sepharose 4B and pHSA-Sepharose 4B columns. It was found that AAA, remaining in the effluent serum from the pHRP-Sepharose 4B column did not react with pHRP-Sepharose 4B by rechromatography. The distribution of IgG and IgM in the AAA, and AAAB fractions is presented in
CONCENTRATION
Serum” Normal Pathologic
Number of cases 3 10
TABLE 1 OF IgG AND IgM ANTI-ALBUMIN NORMAL AND HEPATIC SERA
ANTIBODIES
IgG AAA” w&l 60.3 k 2.6 153.7 k 47.9
IN
IgM AAA” moY1 x 10W 40.2 + 1.8 102.5 2 32.0
dml 40.7 * 11.3 198.3 t 110.6
m&l
x 10-s
4.5 t 1.3 22.0 t 12.3
‘I Normal sera do not precipitate with pHSA in immunodiffusion, but bind 17% of iz51-pHSA at a serum dilution of 1: 160. Pathologic sera from patients with AVH (5 cases), CH (3 cases), and LC (2 cases) give an intense positive reaction in immunodilfnsion with pHSA (the lowest pHSA concentration reacting with undiluted serum was 20 &ml) and bind about 65% of ‘a?-pHSA at a serum dilution of 1:160. ’ The amount of IgG and IgM in purified AAA preparations, as determined by radial immunodiffnsion, is expressed in pg or mol per unit volume of the original serum. The molar concentration of IgG AAA and IgM AAA was calculated by considering the molecular weight of IgG to be 150.000 daltons and that of IgM 900,000 daltons.
ONICA
ET
Al..
FIG. 2. Dissociation rates of ““I-pHSA from complexes in normal and hepatic sera. The curves represent the median values for 5 normal cases (x). for 5 cases of chronic liver diseases (4 cases of CH and 1 case of LC) (O), and for 6 cases of AVH (0). Dissociation of the ‘%pHSA (0.5 wg) was determined after addition of unlabeled excess antigen (5 pg). Sera with high AAA titers were used.
Table 2. It can be seen that the pHRP-Sepharose 4B column retained 40-73s of the IgG AAA and almost all the IgM AAA. The dissociation curves of lz51-pHSA-antibody complexes, either with AAA, or with AAA,, are presented in Fig. 3. The data in Fig. 3 indicate that the two antibody populations were different as regards their affinity for the antigen, so that the dissociation of the radiolabeled antigen was faster from the AAAz than from the AAA, fraction. TABLE FRACTIONATION
AAA, Pathologic serum 11497” I 203Sh 13121” 13491”
bound
to pHRP-Sepharose
IgG Wml)” 183 175 83 360
U Sera of patients with AVH. * Serum of a patient with CH. ” The amount of IgG and IgM in AAA, sion. is expressed in yg per unit volume
2
OF ANTI-ALBUMIN
IgM
4B
(FLg/ml)” 331 75 104 155
and AAA, fractions. of the original serum.
ANTIBODIES
AAA,
bound
IgG (&ml)?
to pHSA-Sepharose IgM
129 105 123 130
as determined
4B
(&ml)’ 28 -
by radial
immunodiffu-
CHARACTERIZATION
OF
ANTI-ALBUMIN
229
ANTIBODIES
1
0
23
iv
30
Lo
50
bo
70
613
Trme (hwrsl
3. Dissociation rates of rz51-pHSA from complexes containing either the AAA, (0) or the AAA, fraction (0) isolated from the serum of a patient with AVH. The AAA, fraction was eluted from the pHRP-Sepharose 4B column. The AAAz fraction, not reacting with the pHRP-Sepharose 4B column, was isolated by chromatography on a pHSA-Sepharose 4B column. Dissociation of the rz51-pHSA (0.15 pg) was determined after addition of unlabeled excess antigen (1.5 pg). FIG.
DISCUSSION
Anti-albumin antibodies were isolated from sera of hepatic patients and normal individuals by afftnity chromatography on insolubilized pHSA. These antibodies seem to be very sensitive since their capacity to bind the radiolabeled antigen was considerably reduced during the purification procedure and subsequent concentration. The data confirm the earlier findings by radioimmunoelectrophoresis (2) and by indirect immunofluorescence (12) that AAA in hepatic patients belong to the IgG and IgM classes. The relative electrophoretic homogeneity of IgG AAA suggests that the in viro alteration of albumin molecules, stimulating antibody synthesis, can be represented by a limited number of new antigenic determinants, such as pyridinium structures as previously indicated (4). The isolation of AAA from normal sera confirmed the data obtained by Mihaescu et al. (3) who found AAA activity in normal sera by radioimmunoassay. Anti-albumin antibodies in normal sera were also found to belong to the IgG and IgM classes. The level of AAA increased in pathologic conditions, the increase recorded for IgM AAA was higher than that found for IgG AAA. Our preliminary data showed that the in-
230
ONICA
ET
AL
crease in IgM antibody was more marked in acute viral hepatitis than in chronic liver diseases. Since the number of sera investigated was rather small, the study should be extended to find out if the differences observed are also valid for a greater number of cases. The ability of Staplr~lococ~cus protein A (SPA) to bind immunoglobulins was used in the dissociation assay to separate the free antigen (pHSA) from the AAA-pHSA complexes. The use of SA as adsorbent for the AAA-pHSA complexes is justified by the finding that the major immunoglobulin class in AAA, in both normal and hepatic patients, was IgG (82-90%) which has a high affinity for SpA (13). Immune complexes containing IgM AAA will also bind to SA since IgM2 subclass reacts with SpA (14, 15), and IgM1 subclass, normally not reacting with SpA, might become reactive after binding to the antigen. This supposition is based on the results of Barkas and Watson (16) showing that SPA-reactive sites can be generated in chicken IgG antibodies (normally not reacting with SpA) by antigen binding. In addition Reisberg and Rossen (17) have shown that SpASepharose can be used to isolate immune complexes prepared with antibodies which do not bind to SpA in monomer form, as for instance goat IgG antibodies. The study of the dissociation rate of AAA from the radiolabeled antigen showed a difference between the normal and hepatic sera; specifically, the affinity of AAA for the antigen was highest in normal sera. This observation is in agreement with the hypothesis that AAA represent, in normal sera, physiological autoantibodies involved in the removal of altered (aged) albumin molecules from the circulation ( 18). According to this hypothesis in the presence of circulating autoantibodies the modified albumin molecules cannot reach the B cells which have receptors for the new antigenic determinants of the aged albumin since they become complexed by the corresponding autoantibodies as soon as they appear in the circulation. It was also supposed that the concentration of the modified albumin increases in pathological conditions (liver diseases) stimulating an active synthesis of AAA by the immunocompetent cells. In the presence of excess circulating modified albumin high-affinity AAA may be removed in a complexed form leaving predominantly low-affinity AAA in circulation, as seems to be the case in liver diseases. On the other hand. AAA are heterogeneous concerning the affinity for the antigen since in the same serum they can have a lower and a higher affinity. The separation of AAA into two populations (AAA, and AAA,) was possible owing to the cross-reacting antigenic determinants present on different proteins treated with glutaraldehyde, for instance peroxidase and albumin. The AAA, purified in the first chromatographic step by reaction with the modified peroxidase were of higher affinity than AAA, recovered by interaction with the modified albumin. Other possible differences between AAA, and AAA, might be related to their specificity. We can assume that the cross-reacting antigenic determinants present on the modified albumin and peroxidase are represented by the chemical groups induced in the proteins by glutaraldehyde treatment (pyridinium haptenic determinants) (7). Therefore, AAA, might recognize the haptenic determinants, while AAA, might be specific for antigenic determinants which appear during the polymerization process (structural determinants) (7) and depend on the particular structure of albumin.
CHARACTERIZATION
OF
ANTI-ALBUMIN
231
ANTIBODIES
ACKNOWLEDGMENTS The authors are grateful indebted to Mrs. Mariana
to Dr. Victor Gherie for valuable criticism Caralicea for excellent technical assistance.
and suggestions.
They
are also
REFERENCES I. 2. 3. 4. 5. 6. 7. 8. 9. 10. Il. 12. 13. 14. 15. 16. 17. 18.
Lenkei. R.. MoIa, G.. Dan, M. E., and Laky, M.. Rev. Roum. Biochim. 11, 271, 1974. Lenkei. R., and Ghetie, V.. J. Immunol. Methods, 16, 23, 1977. Mihaescu, S.. Lenkei, R.. and Ghetie, V., J. Immunol. Methods. 42, 187, 1981. Onica. D.. Margineanu, I., and Lenkei. R.. Mol. Zmmw~ol. 18, 807. 1981. Lenkei, R., Revue Roum. Med. Interne 18, 129, 1980. Lenkei, R.. Buligescu, L.. Belascu, I., Pospai. D.. and Dobre. I.. C&r. E.wp. Immlrt)o[. 43, 381. 1981. Onica, D., Lenkei. R., and Ghetie. V., Immtrnochrnzistrv 15, 687, 1978. Marchalonis. J. .I., Biochem. J. 113, 229, 1969. Fuchs. S.. and Sela, M., In “Handbook of Experimental Immunology” (D. M. Weir, Ed.), pp. 11. I ~ 11, Blackwell, Oxford, 1973. Mancini. G., Carbonara, A. 0.. and Heremans, J. F., Imt~zunochemi.ytr~ 2, 235, 1965. Minden, P., and Farr, R. S., In “Handbook of Experimental Immuncdogy” (D. M. Weir, Ed.), pp. 15. I - 15, Blackwell, Oxford, 1973. Thung, S. N.. and Gerber. M. A., Gastroenterology 80, 260, 1981. Kronvall. G., Quie, P. G., and Williams. R. C., J. Immune/. 104, 273, 1970. Harboe, M., and Foiling, I., Stand. .I. Immrmo/. 3, 471. 1974. Saltvedt, E.. and Harboe. M.. Scand. J. Immrrnol. 5, 1103, 1976. Barkas, T., and Watson, C. M. J.. Immunology 36, 557, 1979. Reisbcrg, M. A.. and Rossen. R. D., C/in. Exp. Immunol. 46, 443, 1981. Ghelie, V.. Onica, D.. Lenkei. R., and Margineanu. I.. Mrch. Aqeing De\,. 17, 27. 1981.
Received
April
29, 1982: accepted
with
revisions
July
14. 1982.