Journal of Clinical Immunology, Vol. 11, No. 5, 1991

Relationship Between Naturally Occurring Human Antibodies to Casein and Autologous Antiidiotypic Antibodies: Implications for the Network Theory CHARLOTTE CUNNINGHAM-RUNDLES,I'3 ZHI-KUN FENG, l ZHUO ZHOU, 1 and K E N N E T H R. WOODS 2

Accepted: April 25, 1991

INTRODUCTION

Previous studies on human autotogous antiidiotypes have been based largely upon analyses of autoimmune disease. We have previously described polyclonal, naturally occurring human antoantibodies directed against antibodies with specificity toward bovine casein in the sera of IgA-deficient humans. In order to define this system more exactly we have now produced two murine monoclona] antibodies directed against bovine milk K-casein to use as clonal tools to identify specific antiidiotypes in these human sera. Kappa-casein is an important part of the casein micelle in milk and cheese; in addition to being an important immunogen for wan, K-casein is known to have conserved amino acid sequence and two antigenic epitopes. Data presented here show that the serum of up to 74% of IgA-deficient and 10% of normal humans have specific autologous antiidiotypes in their serum which bind to monoclonal antibodies directed to bovine K-casein. These human antibodies [intact or F(ab)'2] can be blocked from binding to the monoclonal anti-K-caseins by pure bovine K-casein orthe K-casein peptide fragment. In contrast to previous studies in autoimmune disease, serum levels of the autoantiidiotypes were directly proportional to the level of IgG antibody to bovine K-casein. These observations suggest that continual exposure to a ubiquitous dietary antigen may produce an antigen driven system in which stimulation of both Ab 1 and Ab2 occurs in concert.

The network theory of Jerne views the immune response as a dynamic equilibrium balanced by the reciprocal production of idiotypes and autologous antiidiotypes (1). An idiotype is a serologically defined determinant arising from structural features of the variable (V) regions of immunoglobulins; because of the uniqueness of the V region, it can serve as an immunogen and stimulate the production of antiidiotypic antibodies. In numerous animal analyses, autologous antiidiotypes have been shown to occur as a natural part of the immune response (2-4). Evidence for the reality or function of autologous antiidiotypic antibodies in human systems is much less complete, but such antibodies have been documented in certain autoimmune, infectious, and allergic diseases. For example, autologous anfiidiotypes can appear spontaneously in myasthenia gravis (5, 6), rheumatoid arthritis (7), systemic lupus erythematosus (8), mixed cryoglobulinemia (9), demyelinating neuropathy (10), autoimmune thyroidifis (11), monoclonal IgM gammopathy (12), hepatitis (13), Candida infections (14), and allergic responses to house dust mite (15) and rye antigen (16). Consistent with the view that autologous antiidiotypes may be a normal part of the immune repertoire, antibodies of this kind have been found in both nonimmunized (17) and immunized individuals (18, 19). Our work has focused on the effects of continual oral immunization with a common dietary antigen, bovine K-casein. We have previously shown that individuals who are IgA deficient have an excessive absorption of various dietary antigens into the bloodstream due to an absence of gastrointestinal

KEY WORDS: IgA deficiency; antiidiotypes; milk proteins;

gastrointestinal absorption.

2The Lindsley F. KimballResearch Institute, The New York Blood Center, New York, New York 10021. 3To whomcorrespondence shouldbe addressed. 4Mount Sinai Medical Center, One Gustave Levy Place, New York, New York 10029. 279

0271-9142D1/0900-0279506.50/0 © 1991 Plenum Publishing Corporation

280

secretory IgA (20). This results in the production of high levels of antibodies to dietary antigens such as bovine casein, circulating immune complexes containing this antigen (21), autologous antiidiotypic antibodies, such as antibodies directed to the variable regions of anticaseins (22), and cross-reactive antiidiotypic antibodies (23). Murine monoclonal antibodies (mAb) with a specificity for selected antigens have previously been successfully used as tools in the isolation of related antiidiotypic antibodies in human sera (5, I 1). In the present study, we have used two newly prepared monoclonal antibodies having specificity for bovine K-casein to demonstrate the prevalence and occurrence of specific autologous antiidiotypes in the serum of many IgA-deficient individuals and some normal subjects. MATERIALS AND METHODS Sera and Antibodies

Sera or plasma of 48 IgA-deficient patients of age range 4-62 seen at the Mount Sinai Medical Center were used in this study. IgA deficiency was determined if the serum IgA level was less than I0 mg/dl and if IgG and IgM levels were normal. Immunoglobulin (Ig) fractions of these serum or plasma samples were concentrated by precipitation by 50% saturation with (NH4)2SO 4. In some cases further purification of IgG was performed by passage over diethylaminocellulose (DEAE) (Bio-Rad, Richmond, CA) equilibrated in 0.01 M Na phosphate buffer, pH 7.4. To remove anti-K-casein antibodies from Ig preparations, samples were repeatedly passed over Sepharose-4B (Pharmacia, Piscataway, NJ) to which K-casein had been covalently attached (24). After dialysis, the absorbances at 280 nm of immunoglobulin solutions were obtained to determine the concentration of Ig (using 13.8 as extinction coefficient, El%2S0nm). Control sera were obtained from 36 normal individuals of age range 2 to 67. Pepsin-cleaved F(ab)' 2 fragments of immunoglobulin concentrates from IgA-deficient and normal sera were produced by standard methods (25). To ensure the complete removal of intact IgG and Fc fragments, absorption of these fractions by equal volumes of formalin-fixed, 10% Staphylococcus aureus Cowan I strain organisms in phosphatebuffered saline (PBS; Calbiochem, Behring, La Jolla, CA) was performed at room temperature for 1 hr. These organisms were then removed by centrifugation and the IgG content was determined. IgG

CUNNINGHAM-RUNDLES, FENG, ZHOU, AND WOODS

antibodies in the IgA-deficient and control sera directed to bovine K-casein were determined by enzyme-linked immunosorbent assay (ELISA), using in all cases I0 l~g/ml of human IgG diluted in PBS-Tween (20).

Production of Monoclonal Antibodies to Bovine K-Casein and Screening

The production of an IgG1K mAb to bovine K-casein (4.3) has been described (26). A similar antibody, also an IgGIK (B5), was produced more recently (26, 27). The hybridoma cultures which were found to produce anti K-casein antibody were cloned and culture supernatants were concentrated by 50% saturation with (NH4)2SO 4. In some experiments, mAb was purified by passage of 50% saturated (NH4)zSO4-precipitated antibody over DEAE cellulose (Bio-Rad) in 0.0I M Na phosphate buffer, pH 7.4. The fraction not adhering to DEAE cellulose was collected, dialyzed against water, and used as purified IgG antibody. For some experiments, F(ab)' z fragments of 4.3 or B5 were produced by the methods described above (25).

Test of Monoclonal Antibody Specificity

The specificity of both mAbs was tested as previously described (26). In addition to these studies, mAb 4.3 and B5 were tested to determine if these bound to dephosphorylated (28) or deglycosylated (29) K-casein or K-casein which had been reduced and alkylated. To do this, cell culture supernatants were tested for binding to microtiter plates coated with 10 ~g/ml of these antigens in 0.01 M sodium carbonate buffer, pH 9.8. For reduction and alkylation of K-casein, K-casein was dissolved at 10 mg/ml in 1.0 M Tris-HC1 buffer, pH 8.0, and made 0.02 M in dithioerythritol (Sigma, Chemical Co,, St. Louis, MO) for 90 min. The solution was then made 0.043 M in iodoacetamide (Sigma) in the cold for 30 min. Reduced and alkylated K-casein was then well dialyzed against distilled water and then lyophilized. In other experiments, K-casein was cleaved by rennin (29, 30) to provide two peptides, the NH2-terminal (residues 1-105) and the COOH-terminal (106-169) peptides. These two peptides were separated (29) and used to coat wells in an ELISA as described above to determine the specificity of mAb 4.3 and B5. Journal of Clinical Immunology, Vol. 11, No. 5, 1991

AUTOLOGOUS ANTIIDIOTYPIC ANTIBODIES IN HUMAN SERA

281

Detection of Potential Autologous Antiidiotypes in IgA-Deficient and Control Sera

Blocking of ldiotype Binding to Antiidiotype by K-Casein or K-Casein Peptides

(a) To detect antibodies in human sera which bound mAb 4.3 or B5 (potential antiidiotypes), microtiter plates were coated overnight at 4°C with mAb 4.3 or B5, 10 ~g/ml of well-dialyzed 50% saturated (NHa)zSO4-precipitated cell culture supernatant, or anion exchange-purified mAb, in 0.1 M sodium carbonate buffer, pH 9.6. After washing three times with NS-Tween, immunoglobulin fractions of sera to be investigated were reconstituted to 10 p,g/ml in PBS-Tween and incubated in coated wells for 5 hr. After washing three times with NS-Tween, alkaline phosphatase-conjugated goat anti-human IgG (-y-chain specific) (Tago), diluted t : 2500 in PBS-Tween, was added. After incubation overnight, the wells were washed, NPP solution was added, and the plates were read as usual. As controls, a mouse monoclonal IgG~ with no known specificity and Ig isolated from pooled normal mouse sera were used to coat wells and each serum was tested. Immunoglobulin fractions of human sera were produced by 50% saturated (NH4)2SO4 precipitation; the Ig content was determined by OD at 280 nm as described above. To determine if human immunoglobulins of other kinds bound to mAb 4.3 or B5, wells coated with 4.3 or B5 were exposed to myeloma proteins of the IgGl, G2, G3, and G4 subclass, of both K and "y light-chain type. For this, myeloma proteins at 10 ~g/ml (or for verification, 25 ixg/ml) in PBS--Tween were incubated in coated wells and treated as described above for other human serum immunoglobulins. (b) To test whether pepsin cleavage of the human immunoglobulins would prevent their binding to mAb 4.3 or 5B, immunoglobulin fractions from four IgAdeficient and three normal sera (previously found to bind well to 4.3 and B5) were digested with pepsin (see above). These fractions were tested on microtiter plates coated with mAb 4.3 or B5 at 25 ixg/rnl, and pepsin-treated or untreated immunoglobulin fractions were incubated (at 25 p.g/ml in PBS-Tween) in these coated wells. After overnight incubation at room temperature, the wells were washed and a 1:1 mixture of alkaline phosphatase-conjugated goat antihuman K and ~/chains (Southern Biotechnology Associates, Inc., Birmingham, AL) (each diluted 1:1000 in PBS) was added. The plates were then developed as usual, and the results for pepsin-treated fractions were compared to those for nontreated fractions.

The binding of antibodies in IgA-deficient or normal serum to mAb 4.3 or B5 was then tested in the presence of increasing amounts of K-casein, K-casein peptides, or as controls, the extraneous antigen ovalbumin (OVA) or bovine serum albumin (BSA). In these experiments, 100 Ixl of 10-1-10 -1° mM solutions of these individual antigens (diluted in PBS-Tween) were then added to 100 Ixl of 5, 10, or 25 ~g/ml solutions of the Ig prepared from each of the sera. Each serum dilution with and without the addition of K-casein, K-casein peptide, or control proteins was then tested by ELISA to determine the binding of IgG to mAb 4.3 or B5, which was used to coat microtiter plates.

Journal of Clinical Immunology, Vol. 11, No, 5, 1991

Isolation and Testing of Autologous Anti-Anti-K-Casein Antibodies (a) Potential human anti-anti-K-casein antibodies were isolated from human serum by affinity chromatography using DEAE-purified mAb 4.3 or B5 (or as a control a murine myeloma IgGlk) covalently coupled to Sepharose (24). Using separate 3-ml columns (Bio-Rad), immunoglobulin fractions of 10 IgA-deficient sera were passed over immobilized 4.3, B5, or IgGlk, and after incubation overnight, the adherent fractions were removed by 0.1 M Tris-HC1, pH 2.7. These adherent fractions were dialyzed against water and concentrated by lyophilization. (b) Immunoglobulin fractions adhering to mAb 4.3 or B5 (or the control IgGlk) were then tested by ELISA on microtiter plates coated with either mAb to determine the relative binding each displayed. For this, 100 p.1 of adherent antibodies (at 25 p~g/ml in PBS-Tween) was incubated for 5 hr at room temperature on mAb-coated plates; after washing, goat anti-human IgG alkaline phosphatase conjugate was added as usual. These adherent proteins were similarly tested for binding to the IgGlk myeloma protein. (c) To determine if the above prepared adherent immunoglobulins were of a specific IgG subclass, these proteins were used to coat wells of microtiter plates at 25 ixg/ml in 0.1 M sodium carbonate buffer, pH 9.8. After recoating with 0.1% OVA in PBS, 100 ~1 of murine mAb to IgG subclasses IgG1 (clone HP6012), IgG2 (clone HP6014), IgG 3 (clone HP6050), and IgG4 (clone HP6025) (Zymed Labo-

282

ratories, South San Francisco CA), each diluted 1:800 in PBS-Tween, was added to individual wells. (This dilution of mAb was previously found optimum in preliminary experiments done using human myeloma proteins of various IgG subclasses to coat microtiter plate wells.) After incubation for 3 hr at RT, 1:2000-diluted alkaline phosphataseconjugated goat anti-mouse IgG (Tago) was added and the plates were developed as usual.

Sodium Dodecyl Sulfate (SDS)-Gel Electrophoresis SDS-gel electrophoresis of immunoglobulin fractions from antiidiotype-positive and antiidiotypenegative IgA-deficient donors was performed with DEAE-purified immunoglobulin fractions precipitated from serum by saturation with 50% (NH4)2SO4. Vertical slab gels of 10% polyacrylamide were run under reducing and nonreducing conditions and stained with Coomassie blue (30). In some experiments, electrophoresed proteins were transferred by electrophoresis (Transblot, Bio-Rad) to nitrocellulose sheets and, after appropriate blocking (32), were incubated for 3 hr with undiluted 4.3 cell culture supernatant. After washing, these sheets were incubated with horseradish peroxidase-conjugated goat anti-mouse IgG (Tago) and then, after washing, were developed with a substrate solution containing 0.25 mg O-dianisidine (Sigma) in 0.25 ml methanol, 33 Ixl 3% H202, l ml 100 mM Tris-HCl, pH 7.4, and 9 ml H202.

Proteins Unfractionated casein was purchased from ICN; bovine serum albumin, a-lactalbulin, 13-1actoglobulin, bovine ~/-globulin (BGG), 90% pure K-casein, and ovalbumin were obtained from Sigma. Pure bovine K-casein was a gift from Dr. S. Visser, Department of Biophysical ChemistrY, Netherlands Instituut Voor Zuivelonderzoek, Ede Gld (The Netherlands). The amino acid sequence of bovine K-casein was analyzed by computer program to predict the regions of potential antigenicity from the relative hydrophobic or hydrophilic areas of consecutive sequence intervals (33); for this the published sequence data for K-casein were used (34).

Statistical Analysis The statistical significance between the amount of IgG binding to mAb 4.3 or B5 for normal sera and

CUNNINGHAM-RUNDLES, FENG, ZHOU, AND WOODS

that for IgA-deficient sera was determined by Student's t test. These same data were also examined by a nonparametric test (Wilcoxon signed-rank). To determine the correlation between the amount of IgG binding IgA-deficient or normal sera had for mAB 4.3 and for B5, these data were analyzed by Spearman rank order. Similarly, this method was used to determine whether a significant correlation existed among binding to mAb 4.3 or B5, the level of anti K-casein antibody, and the age of the patient. All statistical analyses were performed by a computer program (Stat View, Brain Power Inc., Calabasas CA). RESULTS

Binding of Human Ig to Monoclonal Antibodies The average level of IgG reactive with purified mAb 4.3 in ELISA was significantly higher for 48 IgA-deficient sera [mean absorbance at 405 n m = 0.122 (OD unit)] than for 36 normal sera (mean absorbance at 405 nm = 0.089; P -< 0.01) (Fig. 1A). Similarly, for 34 IgA-deficient sera and 32 normal controls tested for reactivity to mAb B5, IgAdeficient sera also had greater binding than did normal controls (mean absorbance at 405 nm -0.340 OD unit for IgA-deficient sera and mean absorbance at 405 n m = 0.200 for normal controls; P = 0.01) (Fig. 1B). Eleven of 48 (23%) IgAdeficient sera had levels of 4.3-reactive IgG and 25 of 34 (74%) sera had B5-reactive IgG exceeding one standard deviation from the mean for normals. However, there were examples of substantial amounts of mAb 4.3- or B5-reactive IgG in the sera of some normal individuals and virtually no detectable 4.3- or B5-reactive IgG in the sera of some who were IgA deficient. Since the values obtained by ELISA were not clearly evenly distributed for IgA-deficient or normal sera, we tested whether significant differences in IgG binding to 4.3 or B5 for IgA-deficient and normal sera could also be demonstrated by nonparametric analyses. By a Wilcoxon signed-rank test, these significant differences were verified ( P = 0.01). Neither IgA-deficient nor control sera exhibited significant binding reactivity with normal mouse immunoglobulin or with a mouse monoclonal IgGl~ myeloma protein that was unreactive with K-casein (also Figs. 1A and B). Repeated passage of IgAdeficient sera over K-casein Sepharose to remove anti-K-casein antibody did not prevent the IgG bind-

Journal of Clinical Immunology, Vol. 11, No. 5, I991

283

AUTOLOGOUS ANTIIDIOTYPIC ANTIBODIES IN HUMAN SERA

0.4

0 0

0.3

Z

ing activity of these sera to mAb 4.3 or B5. However, after passage of IgG fractions over caseinSepharose, the amount of antibody retrieved capable of binding to mAb 4.3 or B5 was reduced from that expected, possibly due to losses from immune complex formation with anti-K-casein. Pepsin treatment and testing of F(ab)'2 of IgG of four IgA-deficient and three normal sera for binding to mAbs 4.3 and B5 were also performed. The treated and untreated immunoglobulin fractions displayed similar binding capacities for each mAb fTable I). Similarly, exposure of intact IgG of five IgAdeficient sera to microtiter plates coated with F(ab)': or intact 4.3 or B5 showed similar binding capacities (data not shown).

4.3

0 El

t~ o o

z ~o

B

0.2

0 o3 m

~]0 _

0.1

On

moo ot~ mmm~l~l

o

I °°

EIOD031"I

°1 o

IgA DEFICIENT 1.0

}

NORMAL

B

Kappa-Casein or r-Casein Peptide Blocks Binding o f lg to Monoclonal Antibody 4.3 or B5

B5

o 0.8

Immunoglobulin fractions from sera of individuals with the largest amount of IgG binding to mAb 4.3 or B5 were then tested to determine if this binding could be inhibited by K-casein or its peptides. Kappa-casein at 10-1 mM could block IgG binding to 4.3 to the extent of 48, 62, 80, 88, and 91% for five donors, while control proteins (OVA and BSA) added at any concentration (10-t-10 -1° mM) had no effect (Fig. 2A). Similarly, casein at 10-2 mM could block binding of IgG to mAb B5 by 53, 80, and 82% for three sera, while BSA and OVA again displayed no blocking activity at any concentration (Fig. 2B). This blocking activity was found to be confined to the COOH-terminal peptide of K-casein, since additions of this peptide to IgG samples, but not the NHz-terminal peptide, produced complete blocking (Figs. 3A and B).

z

t,O ,,e/, 111 o

0.6

z

0c 0 m

0.4'

oo ~eoa

o o

m-do 0.2,

moo °" o

°I

o aa~]~o 7 grnO00 o== -I -

0.0

IgA DEFICIENT NORMALS Fig. 1. The relative amount of IgG antibody binding to mAb 4.3 (A) or mAb B5 (B) was determined by ELISA for IgA-deficient and normal subjects by testing the binding of antibody to DEAE-purified mAb-coated microtiter plates. The absorbance at 405 nm (OD units) indicates the relative binding for each sera. The mean -+ standard deviation is marked. Averages for binding to control wells coated with a mouse IgG myeloma (0) or immunoglobulins from normal mouse serum (O) are shown.

Table I. Binding of Pepsin-Treated Immunoglobulin to MonoclonaI Anti-K-Caseins Absorbance at 405 nm(OD units) mAb 4.3

IgA deficient 1 2 3 4 Normals 1 2 3 PBS

mAb B5

Untreated

Pepsin treated

Untreated

Pepsin treated

0.7305 0.3565 0.2680 0.5450

0.4140 0.3885 0.3295 0.3905

0.4200 0.4835 0.3210 0.3175

0.5850 0.7085 0.3635 0.2810

0.4125 0.3675 0.4340 0.0250

0.3870 0.3070 0.2900 0.0100

0.5035 0.4590 0.5270 0.0380

0.5385 0.3635 0.3975 0.0 ! 20

Journal of Clinical Immunology, Vol. 11, No. 5, 1991

284

CUNNINGHAM-RUNDLES, FENG, ZHOU, AND WOODS

100

A

4.3

A B

4.3

1.0-

H2 PEPTIDE

80

0.8'

Z

D

Z

Z

60

u..

o

0.6 .i

E

Z

0 Z

t--

OHPEPTIDE

m ~r

40

~:

o

Z

20

0.4

0.2

OVA BSA

in 1 6 ~ ~t I

I

I

1

10

I , " l m l J ="ml' ' .1.-.I . H-'ll i i.... I 6

CONCENTRATION

10 -4

OF K-CASEIN

10 -2

0.0

10 0

A 10"7 10-6 10-5 10-4 10-3 10- 2 10"1 10 0 .......

.......

I

.......

I

......

1

......

I

.......

1.o B

'1

.......

•

CONCENTRATIONmM

mM

100

B5

B5 A

i

0.8

80

B

E

60

~ ~

I

c 40

H2 PEPTIDE 0.6

u.,I 0 Z ,