CLINICAL
I~UNOL~Y
AND I~UNOPATHO~Y pp. 85-90,1992
Vol. 62, No. 1, January,
IgG Subclass Antibodies to Dietary Antigens in IgA Deficiency Quantification and Correlation with Serum IgG Subclass Levels S. HUSBY, V.-A. OXELIUS, AND S.-E. SVEHAG Institute of Medical ~ic~bia~ogy,
Odense University, Department of Pediatrics, Odense U~iversi~ Department of Pediatrics, Lund University Hospital, Sweden
Odense, Denmark, and
circulation of IgA-deficient individuals, altered IgG subclass antibody distributions to food proteins may be hypothesized. In order to study this hypothesis we determined IgG subclass antibodies to a select group of food proteins. In addition, the levels of serum IgG subclasses were determined in the IgA-deficient subjects and related to the IgG subclass antibody levels.
IgG subclassesof antibodies to the dietary antigens ovalbumin, P-lactoglobulin, casein, bovine IgM, and glycgli (a gluten component) were quantified in 20 adults and 10 children with IgA deficiency and healthy controls (21 adults and 7 children). In the IgA-deficient subjects the levels of IgG subclassesin serum were determined. Deteetabie antibody levels were observed in the majority of the subjects in IgGl and IgG4 for ~ti-ovalbumin and ~-iactoglobulin, and in IgGl, followed by IgG2, IgG3, and IgG4 for antibodies to casein, bovine IgM, and glyegli. Levelsof IgGl and IgG2 antibodies to bovine IgM were higher in the &A-deficient adults than in controls (P < 0.00005,P = 0.0007,respectively), whereas the other antibody levels did not differ significantly between the two groups. An analysis of correlation between the IgG subclassantibody levels did not provide evidence for a particular IgG subclassantibody responsepattern against different protein antigens within the single individual. Serum IgG4 levels eorrelated positively with the summed IgG4 antibody levels (S~~~‘S p = 0.673, P = 0.~51). The IgA-deficient subjects, when compared with healthy controls, did not show a particular IgG subclasspattern or restriction of antibodies to dietary antigens. D 1992 Academic press,
Hastier,
MATERIALS
Patients
AND METRODS
and Controls
Serum was obtained from 20 adults (10 males and 10 females) with IgA deficiency (serum IgA ~0.05 g/liter). Eleven suffered from asthma, chronic bronchitis, or sinuitis, 2 had collagenosis and one ataxia telangiectasia; 7 were healthy. In addition, 10 children with IgA deficiency, 6 girls and 4 boys, of whom 5 were healthy, 3 suffered from asthma and 2 from collagenosis were investigated (median age 6 years, range 2-16 years) (Table 1). The control material consisted of 21 blood donors (31-50 years of age) and 7 children. All were healthy.
Inc.
Reagents and Antisera Ovalbumin (OA; grade V, crystalline), P-lactoglobulin (BLG; crystalline A +B), casein, and gluten were purchased from Sigma (St. Louis, MO). A watersoluble digest of gluten, called glycgli, was prepared according to Douglas (10). Bovine IgM was prepared as follows (11). Bovine serum was heated to 56°C for 30 min to inactivate complement. Fat was removed with a phase separation (Frigen, Behring, Marburg, FRG). Euglobulin precipitation was performed by dialysis against 0.01 M acetate buffer, pH 5.0, for 24 hr. The preparation was centrifuged for 60 min at 10,OOOg and the supernatant was discarded. The precipitate was redissolved in 5 ml of Tris-buffered saline and chromatographed on a Sephacryl S300 column. Two distinct peaks appeared and bovine IgM was localized in the first peak, as determined with polyacrylamide gel electrophoresis. The bovine IgM preparation did not cross-react with any of
INTRODUCTION
IgA deficiency is a clinically heterogeneous condition (11, as a proportion of IgA-deficient subjects appear healthy, whereas others suffer from allergic and autoimmune diseases (2, 3). In addition, subjects with a ~on~mitant IgG subclass deficiency may experience frequent infections and reduced lung function (4). The pathogenesis of IgA deficiency may also be variable, as T cell aberrations were indicated in a minority of IgAdeficient patients (5,6). An excessive uptake of dietary antigens into the circulation has been reported (7, 8) and suggested as a pathogenic factor in IgA deficiency and common variable immunodeficiency. Higher titers of milk precipitins have been found in IgA-deficient individuals (7, 9). Secondary to the altered state of mucosal immunity and the increased uptake of dietary antigens into the 85
All
Copyright 0 1992 rights of reproduction
ooso-1229/92 $1.50 by Academic Press, Inc. in any form reserved.
86
HUSBY, OXELIUS,
AND SVEHAG
TABLE IgG Subclass Patient group IgA-deficient adults (g/liter) IgA-deficient children (%)
Levels
1
in IgA-Deficient
IgGl Median Range Median Range
10.71 7.1-21.3 145
Subjects IgG2
IgG3
4.34" 0.7-6.7 88
0.5-1.62
0.89
IgG4 0.44
0.0--l .8P
90 3 0.8-108 50-145 040u -~~ ____.-__ ~~~ ~~~ ~-~ ~~ ~~~~ Note. The median values and range are expressed in grams per liter for adults. For the children, mean values and range are expressed as the percentage (o/o)of the mean for age for healthy children (13). a One patient with IgG2 deficiency and ataxia teleangiectasia. 45-360
the mouse anti-human IgG subclass antibodies (see below). Human hIgG myelomas of the four subclasses were a gift from WHO (Professor F. Skvaril, Berne, Switzerland). A normal rabbit F(ab’), fragment and rabbit F(ab’)s anti-h human IgG were prepared as previously described (12). Monoclonal mouse antibodies against IgGl (clone No. NL-161, IgG2 (clone No. M 730131, IgG3 (clone No. ZG4), and IgG4 (clone No. RJ4) subclasses were purchased as ascites fluids from Oxoid (Bedford, UK). Alkaline phosphatase (API-labeled rabbit anti-mouse IgG from Orion (Helsinki, Finland) was absorbed with Sepharose 4B-bound human IgG before use. ELISA for IgG Subclass Antibodies Dietary Antigens
to
The assays were carried out as previously described for IgG subclass antibodies to OA, BLG, and glycgli (12, 13). Microplate wells (Immunoplate II, Nunc, Roskilde, Denmark) were coated with antigen (OA, BLG, glycgli, casein, or bovine IgM) at a concentration of 9 pg/liter of 0.1 M bicarbonate buffer, pH 9.6. Separate wells were coated with a rabbit F(ab’), anti-A antibody and, as controls, normal rabbit F(ab’), at 2.5 pgiliter bicarbonate buffer. Nonspecific binding was blocked with 0.1% human serum albumin (HSA, reinst., Behring, Marburg, FRG), a preparation specifically without human IgG and nonreacting with any secondary antibodies in ELISA. After washings, diluted sera and myeloma standards were incubated overnight. The plates were washed and incubated with monoclonal antibodies, followed by the AP anti-mouse antibody and substrate. Absorbance values at 405 nm (605~nm value subtracted as background) were read in a multichannel spectrophotometer (Immunoreader NJ2000, Nippon Intermed, Tokyo). From the myeloma standards (OD,os -, values from normal rabbit F(ab’)s-coated wells subtracted) the amount of antibody bound in micrograms per liter was estimated. The detection limit of the assays varied from 0.08 to 0.25 pg/liter of serum. All determinations were done in duplicate. Positive control sera selected to represent low and
high antibody levels, respectively, were included in each plate. For IgGl, IgG3, and IgG4, two sera were used for each subclass, whereas for the IgG2 antibody assay with a monoclonal antibody (M 73013), which we had not utilized previously, five sera were used, one for each antigen. The mean within-run coefficients of variation (CV%) were 4.7% for IgGl, 5.8% for IgG2, 3.5% for IgG3, and 7.1% for IgG4. The mean between-days CV% were 25% for IgGl, 9.7% for IgG2, 13% for IgG3, and 26% for IgG4. A near to equal number of patients and controls were tested in each run to minimize the influence of the between-days CV%. Total IgG Subclasses
Total serum IgG subclasses were determined by radial immunodiffusion with polyclonal rabbit antisera, according to Oxelius (14). Antisera were rendered specific for IgG by absorption with serum fractions or isolated myeloma proteins. Subclass specificity was ensured by absorption with isolated relevant myeloma proteins. In addition, IgG-deficient human sera were examined. Statistics
For comparison between groups, the nonparametric Mann-Whitney U test for unpaired samples was used as a two-tailed test. For correlation analysis between IgG subclass antibody levels or between the levels of total IgG subclasses in serum and IgG subclass antibodies, the nonparametric Spear-man’s test was used. The level of significance was chosen as P = 0.05. RESULTS
IgG Subclass Antibody Distribution
The patient material was separated into adults and children. The IgG subclass antibodies to OA and BLG showed a restriction to IgGl and IgG4 (Fig. l), with no significant differences between IgA-deficient subjects and controls (P > 0.19). The IgG subclass antibodies to casein and glycgli (Fig. 2) and to bovine IgM (Fig. 3) were distributed in a different way, with a predominance of IgGl, followed
IgG SUBCLASS Anti-OA
ANTIBODIES
Antl-
ant1 bodies
IqG4 19 0
87
IN IgA DEFICIENCY BLG antIbodIes
IqGl
IqGL
0
100 0
0 0
0 0
10
,”
0 o 0
8 -8-
7
10
501
w” A
B
C
0
gap-03 A
0
B
c
0
-9 A
B
c
0
+%=+%=-A
8 0
B
C
D
FIG. 1. Serum IgGl and IgG4 antibodies to ovalbumin and P-lactoglobulin. Antibody levels in IgA-deficient adults (A), in healthy adults (B), in IgA-deficient children (C), and in healthy children (D). Note log scale of the ordinate. Bars denote median values.
by IgG2 and IgG3. Only a few sera positive for the IgG4 antibody at low levels were observed. The levels of anti-casein and anti-glycgli ’ antibodies in the IgAdeficient subjects did not differ significantly from the controls (P > 0.19) in any of the subclasses (Fig. 2). However, a very clear difference between the IgAdeficient adults and the controls was observed for the IgGl (P < 0.00005) and IgG2 (P < 0.0007) anti-bovine IgM antibodies (Fig. 3). No differences were found between those IgA-deficient subjects who had a disease, mainly asthma, and those who were healthy. Correlation
between IgG Subclass Antibody
Levels
We investigated whether the antibody responses to the dietary antigens showed an IgG subclass restriction. Statistical analysis was carried out separately for both IgA-deficient subjects and controls, including and excluding subjects who were negative for both antibody populations studied at a time (double antibodynegative subjects). The Spearman’s correlation analysis tended to overestimate the statistical significance when the majority of the data pairs were zero, even when corrected for ties. For the anti-OA and anti-BLG antibodies there was a statistically insignificant tendency to respond in either the IgGl or the IgG4 subclass, with a minor proportion of the subjects reacting within both subclasses (data not shown>. IgG antibodies to casein and bovine IgM occurred mostly in the IgGl subclass. IgAdeficient and healthy subjects with high levels of IgGl antibody in addition tended to produce IgG2, IgG3, and IgG4 antibodies (data not shown). We also wanted to test whether an individual reacted preferentially with an antibody within a partic-
ular IgG subclass toward different antigens. No significant correlations were observed between the levels of IgGl, IgG2, or IgG4 antibodies to casein and bovine IgM, when the statistical analysis was corrected for subjects negative for both anti-casein and anti-IgM antibodies (data not shown). Correlation between the Levels of IgG Subclass Antibodies and Total IgG Subclasses The levels of the IgG subclass antibody to selected antigens may be related to total IgG subclass levels. Measurement of the levels of IgG subclasses was performed for the IgA-deficient subjects. A nonparametric regression analysis showed a significant correlation b e t ween the summation of the IgG4 antibody levels (expressed in micrograms per liter) and the IgG4 levels for the IgA-deficient adults (Fig. 4, Spearman~s o = 0.673, P = 0.0051). This was also the case for the IgG4 anti-OA antibody (Spearman’s p = 0.587, P = 0.0025) and the IgG4 anti-glycgli antibody (Spearman’s p = 0.633, p = 0.0013), but not for the other IgG4 antibodies (data not shown). No correlation was observed between the levels of IgGl, IgG2, or IgG3 and the antib od y 1evels within the respective subclass. DISCUSSION
This paper describes measurements of IgG subclass antibodies to five selected dietary proteins of highly varying composition and structure, namely BLG, bovine IgM, and bovine casein found in cow’s milk, OA from hen’s egg, and glycgli from gluten. The betweenassay CV% were relatively high, particularly for the IgGl and IgG4 antibody assays, but bias was reduced
88
I-IUSBY, OXELIUS, lgG
j 8
AND SVEHAG
2
igG3 Anti-casein
antlhodies
AntI-giycg~l
antlbodtes
0”
FIG. 2. Serum IgG 1,2,3, and 4 antibodies to casein and the gluten product glycgli. Antibody levels in &A-deficient adults CA), in healthy adults (B), in IgA-deficient children fC), and in healthy children CD). Note log scale of the ordinate. Bars denote median w&es.
by including an equal number of patient and control sera in each run. The occurrence of IgG subclass antibodies in the majority of the subjects studied made possible an analysis of interrelations between the antibody levels. We observed a clear tendency for the antiOA and anti-BLG antibodies to occur in either IgGl or IgG4, as opposed to the antibodies to casein, bovine IgM, and glycgli, which occurred primarily in IgGl, followed by IgG2, IgG3 and IgG4. The latter tendency may simply reflect a random activation of antibodyproducing cells determined by their frequency. We have previously described the restriction of antibodies to OA and BLG within IgGl and IgG4 in healthy adults (121, as partly confirmed by others (15). In children with coeliac disease we observed a restriction to IgGl and IgG3 for antibodies to gliadin and
glycgli, with particularly high levels after gluten challenge (13). The same IgG subclass distribution of antibody to glycgli is observed at low levels in the IgAdeficient and healthy subjects in the present study. Furthermore, a similar dist~bution was seen for antibodies to casein and bovine IgM. Several variables seem to influence the antibody isotype distribution, such as antigen dose, chronicity of exposure, epitope structure, and age of the individual (16). A significant correlation was observed between serum IgG4 and the IgG4 antibody levels (Fig. 4) for the IgA-deficient adults, probably due to the subjects’ ability to mount IgG4 responses in general rather than as a consequence of antibodies to dietary antigens con~ituting a major proportion of the total IgG4. Signifkantly raised levels of IgGl and IgG2 subclass
IgG SUBCLASS
ANTIBODIES AntI-bowne
0
IgGl
89
IN IgA DEFICIENCY IgM antIbodIes
IgG2
IgG3
IgGl
M/I
100 -
99
939" ; 0 10 -
0
8
8
:
10 -
I~UNOL~Y
AND I~UNOPATHO~Y pp. 85-90,1992
Vol. 62, No. 1, January,
IgG Subclass Antibodies to Dietary Antigens in IgA Deficiency Quantification and Correlation with Serum IgG Subclass Levels S. HUSBY, V.-A. OXELIUS, AND S.-E. SVEHAG Institute of Medical ~ic~bia~ogy,
Odense University, Department of Pediatrics, Odense U~iversi~ Department of Pediatrics, Lund University Hospital, Sweden
Odense, Denmark, and
circulation of IgA-deficient individuals, altered IgG subclass antibody distributions to food proteins may be hypothesized. In order to study this hypothesis we determined IgG subclass antibodies to a select group of food proteins. In addition, the levels of serum IgG subclasses were determined in the IgA-deficient subjects and related to the IgG subclass antibody levels.
IgG subclassesof antibodies to the dietary antigens ovalbumin, P-lactoglobulin, casein, bovine IgM, and glycgli (a gluten component) were quantified in 20 adults and 10 children with IgA deficiency and healthy controls (21 adults and 7 children). In the IgA-deficient subjects the levels of IgG subclassesin serum were determined. Deteetabie antibody levels were observed in the majority of the subjects in IgGl and IgG4 for ~ti-ovalbumin and ~-iactoglobulin, and in IgGl, followed by IgG2, IgG3, and IgG4 for antibodies to casein, bovine IgM, and glyegli. Levelsof IgGl and IgG2 antibodies to bovine IgM were higher in the &A-deficient adults than in controls (P < 0.00005,P = 0.0007,respectively), whereas the other antibody levels did not differ significantly between the two groups. An analysis of correlation between the IgG subclassantibody levels did not provide evidence for a particular IgG subclassantibody responsepattern against different protein antigens within the single individual. Serum IgG4 levels eorrelated positively with the summed IgG4 antibody levels (S~~~‘S p = 0.673, P = 0.~51). The IgA-deficient subjects, when compared with healthy controls, did not show a particular IgG subclasspattern or restriction of antibodies to dietary antigens. D 1992 Academic press,
Hastier,
MATERIALS
Patients
AND METRODS
and Controls
Serum was obtained from 20 adults (10 males and 10 females) with IgA deficiency (serum IgA ~0.05 g/liter). Eleven suffered from asthma, chronic bronchitis, or sinuitis, 2 had collagenosis and one ataxia telangiectasia; 7 were healthy. In addition, 10 children with IgA deficiency, 6 girls and 4 boys, of whom 5 were healthy, 3 suffered from asthma and 2 from collagenosis were investigated (median age 6 years, range 2-16 years) (Table 1). The control material consisted of 21 blood donors (31-50 years of age) and 7 children. All were healthy.
Inc.
Reagents and Antisera Ovalbumin (OA; grade V, crystalline), P-lactoglobulin (BLG; crystalline A +B), casein, and gluten were purchased from Sigma (St. Louis, MO). A watersoluble digest of gluten, called glycgli, was prepared according to Douglas (10). Bovine IgM was prepared as follows (11). Bovine serum was heated to 56°C for 30 min to inactivate complement. Fat was removed with a phase separation (Frigen, Behring, Marburg, FRG). Euglobulin precipitation was performed by dialysis against 0.01 M acetate buffer, pH 5.0, for 24 hr. The preparation was centrifuged for 60 min at 10,OOOg and the supernatant was discarded. The precipitate was redissolved in 5 ml of Tris-buffered saline and chromatographed on a Sephacryl S300 column. Two distinct peaks appeared and bovine IgM was localized in the first peak, as determined with polyacrylamide gel electrophoresis. The bovine IgM preparation did not cross-react with any of
INTRODUCTION
IgA deficiency is a clinically heterogeneous condition (11, as a proportion of IgA-deficient subjects appear healthy, whereas others suffer from allergic and autoimmune diseases (2, 3). In addition, subjects with a ~on~mitant IgG subclass deficiency may experience frequent infections and reduced lung function (4). The pathogenesis of IgA deficiency may also be variable, as T cell aberrations were indicated in a minority of IgAdeficient patients (5,6). An excessive uptake of dietary antigens into the circulation has been reported (7, 8) and suggested as a pathogenic factor in IgA deficiency and common variable immunodeficiency. Higher titers of milk precipitins have been found in IgA-deficient individuals (7, 9). Secondary to the altered state of mucosal immunity and the increased uptake of dietary antigens into the 85
All
Copyright 0 1992 rights of reproduction
ooso-1229/92 $1.50 by Academic Press, Inc. in any form reserved.
86
HUSBY, OXELIUS,
AND SVEHAG
TABLE IgG Subclass Patient group IgA-deficient adults (g/liter) IgA-deficient children (%)
Levels
1
in IgA-Deficient
IgGl Median Range Median Range
10.71 7.1-21.3 145
Subjects IgG2
IgG3
4.34" 0.7-6.7 88
0.5-1.62
0.89
IgG4 0.44
0.0--l .8P
90 3 0.8-108 50-145 040u -~~ ____.-__ ~~~ ~~~ ~-~ ~~ ~~~~ Note. The median values and range are expressed in grams per liter for adults. For the children, mean values and range are expressed as the percentage (o/o)of the mean for age for healthy children (13). a One patient with IgG2 deficiency and ataxia teleangiectasia. 45-360
the mouse anti-human IgG subclass antibodies (see below). Human hIgG myelomas of the four subclasses were a gift from WHO (Professor F. Skvaril, Berne, Switzerland). A normal rabbit F(ab’), fragment and rabbit F(ab’)s anti-h human IgG were prepared as previously described (12). Monoclonal mouse antibodies against IgGl (clone No. NL-161, IgG2 (clone No. M 730131, IgG3 (clone No. ZG4), and IgG4 (clone No. RJ4) subclasses were purchased as ascites fluids from Oxoid (Bedford, UK). Alkaline phosphatase (API-labeled rabbit anti-mouse IgG from Orion (Helsinki, Finland) was absorbed with Sepharose 4B-bound human IgG before use. ELISA for IgG Subclass Antibodies Dietary Antigens
to
The assays were carried out as previously described for IgG subclass antibodies to OA, BLG, and glycgli (12, 13). Microplate wells (Immunoplate II, Nunc, Roskilde, Denmark) were coated with antigen (OA, BLG, glycgli, casein, or bovine IgM) at a concentration of 9 pg/liter of 0.1 M bicarbonate buffer, pH 9.6. Separate wells were coated with a rabbit F(ab’), anti-A antibody and, as controls, normal rabbit F(ab’), at 2.5 pgiliter bicarbonate buffer. Nonspecific binding was blocked with 0.1% human serum albumin (HSA, reinst., Behring, Marburg, FRG), a preparation specifically without human IgG and nonreacting with any secondary antibodies in ELISA. After washings, diluted sera and myeloma standards were incubated overnight. The plates were washed and incubated with monoclonal antibodies, followed by the AP anti-mouse antibody and substrate. Absorbance values at 405 nm (605~nm value subtracted as background) were read in a multichannel spectrophotometer (Immunoreader NJ2000, Nippon Intermed, Tokyo). From the myeloma standards (OD,os -, values from normal rabbit F(ab’)s-coated wells subtracted) the amount of antibody bound in micrograms per liter was estimated. The detection limit of the assays varied from 0.08 to 0.25 pg/liter of serum. All determinations were done in duplicate. Positive control sera selected to represent low and
high antibody levels, respectively, were included in each plate. For IgGl, IgG3, and IgG4, two sera were used for each subclass, whereas for the IgG2 antibody assay with a monoclonal antibody (M 73013), which we had not utilized previously, five sera were used, one for each antigen. The mean within-run coefficients of variation (CV%) were 4.7% for IgGl, 5.8% for IgG2, 3.5% for IgG3, and 7.1% for IgG4. The mean between-days CV% were 25% for IgGl, 9.7% for IgG2, 13% for IgG3, and 26% for IgG4. A near to equal number of patients and controls were tested in each run to minimize the influence of the between-days CV%. Total IgG Subclasses
Total serum IgG subclasses were determined by radial immunodiffusion with polyclonal rabbit antisera, according to Oxelius (14). Antisera were rendered specific for IgG by absorption with serum fractions or isolated myeloma proteins. Subclass specificity was ensured by absorption with isolated relevant myeloma proteins. In addition, IgG-deficient human sera were examined. Statistics
For comparison between groups, the nonparametric Mann-Whitney U test for unpaired samples was used as a two-tailed test. For correlation analysis between IgG subclass antibody levels or between the levels of total IgG subclasses in serum and IgG subclass antibodies, the nonparametric Spear-man’s test was used. The level of significance was chosen as P = 0.05. RESULTS
IgG Subclass Antibody Distribution
The patient material was separated into adults and children. The IgG subclass antibodies to OA and BLG showed a restriction to IgGl and IgG4 (Fig. l), with no significant differences between IgA-deficient subjects and controls (P > 0.19). The IgG subclass antibodies to casein and glycgli (Fig. 2) and to bovine IgM (Fig. 3) were distributed in a different way, with a predominance of IgGl, followed
IgG SUBCLASS Anti-OA
ANTIBODIES
Antl-
ant1 bodies
IqG4 19 0
87
IN IgA DEFICIENCY BLG antIbodIes
IqGl
IqGL
0
100 0
0 0
0 0
10
,”
0 o 0
8 -8-
7
10
501
w” A
B
C
0
gap-03 A
0
B
c
0
-9 A
B
c
0
+%=+%=-A
8 0
B
C
D
FIG. 1. Serum IgGl and IgG4 antibodies to ovalbumin and P-lactoglobulin. Antibody levels in IgA-deficient adults (A), in healthy adults (B), in IgA-deficient children (C), and in healthy children (D). Note log scale of the ordinate. Bars denote median values.
by IgG2 and IgG3. Only a few sera positive for the IgG4 antibody at low levels were observed. The levels of anti-casein and anti-glycgli ’ antibodies in the IgAdeficient subjects did not differ significantly from the controls (P > 0.19) in any of the subclasses (Fig. 2). However, a very clear difference between the IgAdeficient adults and the controls was observed for the IgGl (P < 0.00005) and IgG2 (P < 0.0007) anti-bovine IgM antibodies (Fig. 3). No differences were found between those IgA-deficient subjects who had a disease, mainly asthma, and those who were healthy. Correlation
between IgG Subclass Antibody
Levels
We investigated whether the antibody responses to the dietary antigens showed an IgG subclass restriction. Statistical analysis was carried out separately for both IgA-deficient subjects and controls, including and excluding subjects who were negative for both antibody populations studied at a time (double antibodynegative subjects). The Spearman’s correlation analysis tended to overestimate the statistical significance when the majority of the data pairs were zero, even when corrected for ties. For the anti-OA and anti-BLG antibodies there was a statistically insignificant tendency to respond in either the IgGl or the IgG4 subclass, with a minor proportion of the subjects reacting within both subclasses (data not shown>. IgG antibodies to casein and bovine IgM occurred mostly in the IgGl subclass. IgAdeficient and healthy subjects with high levels of IgGl antibody in addition tended to produce IgG2, IgG3, and IgG4 antibodies (data not shown). We also wanted to test whether an individual reacted preferentially with an antibody within a partic-
ular IgG subclass toward different antigens. No significant correlations were observed between the levels of IgGl, IgG2, or IgG4 antibodies to casein and bovine IgM, when the statistical analysis was corrected for subjects negative for both anti-casein and anti-IgM antibodies (data not shown). Correlation between the Levels of IgG Subclass Antibodies and Total IgG Subclasses The levels of the IgG subclass antibody to selected antigens may be related to total IgG subclass levels. Measurement of the levels of IgG subclasses was performed for the IgA-deficient subjects. A nonparametric regression analysis showed a significant correlation b e t ween the summation of the IgG4 antibody levels (expressed in micrograms per liter) and the IgG4 levels for the IgA-deficient adults (Fig. 4, Spearman~s o = 0.673, P = 0.0051). This was also the case for the IgG4 anti-OA antibody (Spearman’s p = 0.587, P = 0.0025) and the IgG4 anti-glycgli antibody (Spearman’s p = 0.633, p = 0.0013), but not for the other IgG4 antibodies (data not shown). No correlation was observed between the levels of IgGl, IgG2, or IgG3 and the antib od y 1evels within the respective subclass. DISCUSSION
This paper describes measurements of IgG subclass antibodies to five selected dietary proteins of highly varying composition and structure, namely BLG, bovine IgM, and bovine casein found in cow’s milk, OA from hen’s egg, and glycgli from gluten. The betweenassay CV% were relatively high, particularly for the IgGl and IgG4 antibody assays, but bias was reduced
88
I-IUSBY, OXELIUS, lgG
j 8
AND SVEHAG
2
igG3 Anti-casein
antlhodies
AntI-giycg~l
antlbodtes
0”
FIG. 2. Serum IgG 1,2,3, and 4 antibodies to casein and the gluten product glycgli. Antibody levels in &A-deficient adults CA), in healthy adults (B), in IgA-deficient children fC), and in healthy children CD). Note log scale of the ordinate. Bars denote median w&es.
by including an equal number of patient and control sera in each run. The occurrence of IgG subclass antibodies in the majority of the subjects studied made possible an analysis of interrelations between the antibody levels. We observed a clear tendency for the antiOA and anti-BLG antibodies to occur in either IgGl or IgG4, as opposed to the antibodies to casein, bovine IgM, and glycgli, which occurred primarily in IgGl, followed by IgG2, IgG3 and IgG4. The latter tendency may simply reflect a random activation of antibodyproducing cells determined by their frequency. We have previously described the restriction of antibodies to OA and BLG within IgGl and IgG4 in healthy adults (121, as partly confirmed by others (15). In children with coeliac disease we observed a restriction to IgGl and IgG3 for antibodies to gliadin and
glycgli, with particularly high levels after gluten challenge (13). The same IgG subclass distribution of antibody to glycgli is observed at low levels in the IgAdeficient and healthy subjects in the present study. Furthermore, a similar dist~bution was seen for antibodies to casein and bovine IgM. Several variables seem to influence the antibody isotype distribution, such as antigen dose, chronicity of exposure, epitope structure, and age of the individual (16). A significant correlation was observed between serum IgG4 and the IgG4 antibody levels (Fig. 4) for the IgA-deficient adults, probably due to the subjects’ ability to mount IgG4 responses in general rather than as a consequence of antibodies to dietary antigens con~ituting a major proportion of the total IgG4. Signifkantly raised levels of IgGl and IgG2 subclass
IgG SUBCLASS
ANTIBODIES AntI-bowne
0
IgGl
89
IN IgA DEFICIENCY IgM antIbodIes
IgG2
IgG3
IgGl
M/I
100 -
99
939" ; 0 10 -
0
8
8
:
10 -
