Studies on Translocation of Immunoglobulins Across Intestinal Epithelium

73:1333-1339, 1977 Copyright © 1977 by the American Gastroenterological Association

Vol. 73 , No. 6 Printed in U.SA .

GASTROENTEROLOGY

STUDIES ON TRANSLOCATION OF IMMUNOGLOBULINS ACROSS INTESTINAL EPITHELIUM IV. Evidence for binding of IgA and IgM to secretory component in intestinal epithelium WILLIAM

R.

BROWN, M.D., KIYOKO IsoBE, M.D., PAuLK. NAKANE, PH.D., AND

BERNARD PACINI, M . D .

Department of Medicine (Gastroenterology) of the Denver Veterans Administration Hospital and the University of Colorado Medical Center, and the Department of Pathology of the University of Colorado Medical Center, Denver, Colorado

We conducted studies concerning the issue of whether secretory component (SC) is a specific receptor on intestinal epithelial cells for IgA and IgM. Initially, frozen sections of human intestinal mucosa were incubated with dimeric monoclonal human IgA 1 conjugated to horseradish peroxidase; adjacent sections were reacted with peroxidaseconjugated antibodies to SC. The conjugated IgA and anti-SC bound to similar sites in the epithelium, that is to basolateral margins and supranuclear cytoplasm of columnar epithelial cells, principally in gland crypts. In subsequent tests of binding specificity, binding of the dimeric IgA conjugate was inhibited by pretreating the tissues with unconjugated dimeric IgA or 198 IgM, pretreating the tissues with unconjugated antibodies to SC, or preincubating the dimeric IgA conjugate with free SC. Binding was not inhibited or only partially inhibited by pretreating the tissues with monomeric IgA or IgG, pretreating the tissues with antibodies to human or heterologous immunoglobulins, or preincubating the dimeric IgA conjugate with 118 secretory colostral IgA. The findings indicate that dimeric IgA and 198 IgM are capable of binding in vitro to specific sites on intestinal epithelial cells, most likely to SC. This supports the hypothesis that transport of these immunoglobulins into intestinal fluids involves their combination with SC in the epithelium. The mechanisms by which immunoglobulins are transported across the intestinal epithelium into the gut lumen are incompletely defined. IgA and to a limited extent IgM in intestinal fluids are combined with the antigenically distinct glycoprotein, secretory component (SC). 1- 5 Immunocytochemical studies have localized SC, IgA, and IgM to certain identical sites within or on the plasma membrane surfaces of intestinal epithelial cells. r.-9 198 IgM and dimeric IgA can combine with soluble SC in vitro. 1(}- 12 These observaReceived March 28, 1977. Accepted June 6, 1977. Address requests for reprints to: William R. Brown, M.D., Department of Medicine (lllE), Veterans Administration Hospital, 1055 Clermont Street, Denver, Colorado 80220. This study was supported in part by the Medical Research Service of the Veterans Administration; the National Foundation for Ileitis and Colitis, Inc.; Grant CA-17342 from the National Cancer Institute , United States Public Health Service, through the National Large Bowel Cancer Project; Grants AI 09109 and AM 15663 from the United States Public Health Service; and Grant DT 14 from the American Cancer Society. The authors are grateful to Evelyn Lee and Eugene Shattuck for technical assista nce, and to surgical personnel of the Denver Veterans Administration Hospital and General Rose Memorial Hospital for assistance in obtaining tissue specimens.

tions suggest that the transepithelial transport of IgA and IgM involves binding of polymeric forms of these immunoglobulins to SC in the epithelium. Some workers7· 9 • 13 have in fact proposed that SC is a specific receptor on epithelial cells for IgA and IgM. This paper presents immunohistochemical evidence that dimeric IgA and 198 IgM can combine in vitro to SC associated with epithelial cells of human intestine.

Materials and Methods Immunoglobulins. Dimeric monoclonal lgA~o prepared from human myeloma serum, was generously contributed by Dr. Brian Underdown. 10 IgG and 198 IgM were prepared in our laboratory from normal human serum as described, 14 including absorption of the lgM preparation by affinity chromatography with rabbit antihuman IgA to remove any traces of polymeric IgA. Monomeric lgA was prepared from normal human serum by zinc sulfate precipitation, Sephadex G-200 gel filtration (Pharmacia Fine Chemicals, Piscataway , N.J .), and affinity column chromatography against antihuman IgG. llS exocrine lgA and free SC were prepared from human colostrum. 6 • 15 All purified proteins were shown to be antigenically homogeneous by immunoelectrophoresis and double immunodiffusion against polyvalent antibodies to human serum and human colostrum. A portion of our SC preparation was labeled with 1251 by

1333

1334

BROWN ET AL.

the chloramine-T method.w In addition, 1251-labeled human SC was obtained from Dr. Underdown. 10 The ability of the dimeric IgA, monomeric IgA, IgM, and IgG preparations to combine with soluble free SC was tested by incubating them with radiolabeled SC as described. 10 Onetenth microgram of SC was incubated with about 5 f.J-g of immunoglobulin, and immunoglobulin-bound SC was separated from unbound SC by Sephadex G-200 column chromatography. Dimeric IgA and IgM effectively combined with labeled free SC (either Underdown's or ours), whereas monomeric lgA and lgG bound only slightly (fig. 1). These findings are consistent with previous studies on the affinity of these classes of immunoglobulins for SC. uH 2 A portion of the dim eric lgA was conjugated with horseradish peroxidase as described.,; Conjugation reduced the protein's capacity for binding radiolabeled SC by about 30%. For reaction with tissues, we tested various concentrations of the dimeric IgA conjugate diluted in bovine serum albumin (BSA, 10 mg per ml in phosphate-buffered saline, pH 7.2); 150 f.J-g per ml proved to provide satisfactory intensity of staining. Antibodies. Rabbit anti-SC , anti-IgA, anti-IgM, and horseradish peroxidase-conjugated anti-SC were those used previously.6 Fab' fragments of the anti-SC and anti-IgA globulins had been prepared by pepsin digestion." Rabbit anti-mouse IgG2b was a gift from Dr. Howard Grey, National Jewish Hospital, Denver, Colorado. Tissues. As approved by the Human Experimentation Committee of the University of Colorado Medical Center, normal small intestine was obtained from patients undergoing abdominal operations such as gastrojejunal anastomosis or jejunoileal bypass for obesity. Blocks of mucosa about 2 by 2 by 3 mm were embedded in tissue Tek II (Lab-Tek Products, Naperville, Ill.) in an aluminum cup and immediately frozen in dry ice-alcohol. Sections about 6 f.J-m thick were cut in a

Vol . 73, No . 6

cryostat, air-dried on albuminized slides, and fixed for 5 min in chilled acetone. The sections were washed in phosphatebuffered saline (pH 7.2) and treated with 10% fetal calf serum in phosphate-buffered saline for 10 min. The sections were again washed with buffer and reacted under various conditions, described below, with the peroxidase-conjugated dimeric lgA. The slides were washed in phosphate-buffered saline and stained by reaction with diaminobenzidine. 6 In certain experiments, described below , sections were reacted with various inhibitor proteins before the IgA conjugate. The sections were examined by a light microscope and photographed.

Results In initial experiments, adjacent sections of intestinal mucosa were reacted with the peroxidase-conjugated dimeric IgA or conjugated anti-SC. The conjugated antigen and antibodies were found deposited in similar epithelial sites (fig. 2). Both were distributed over the supranuclear cytoplasm and lateral and basal margins of columnar epithelial cells; both stained cells in gland crypts much more intensely than villus cells; neither was associated with goblet cell mucus. No differences in staining intensity were noted between jejunum and ileum. Control sections of human spleen, lymph nodes, and striated muscle were not specifically stained by either the conjugated dimeric IgA or anti-SC. Next, we attempted to define the conditions most favorable for binding of the IgA conjugate to intestinal epithelium. Tissue staining appeared to be as intense after incubation of sections with the conjugate at 4°C as at room temperature and did not increase after a 2-hr incubation. Therefore, incubation was routinely conducted for 2 hr at 4°C. The pH optimum for IgA binding OIMERIC lgA MONOMERIC lgA was evaluated roughly by adjusting the pH of the conjugate from pH 4 to pH 10 (a range in which 4 horseradish peroxidase is not irreversibly inactivated) before it was applied to the tissues. Staining appeared to be most prominent after incubation at a pH of about 6.5 to 7.0. These observations are in reasonably good agreement with data from Weicker and Underdown, 10 who found that optimal in vitro binding of SC to IgM occurred at a temperature of 4 to 25°C and pH 7.0 to 7.3, and that binding was maximal within 1 hr. 0 The specificity of binding of the IgA conjugate to gut lgM ::E epithelium was tested by a series of experiments. First, c.. u 5 the ability of unconjugated proteins to inhibit binding of the conjugate was tested. Twenty microliters of a 4 solution of unconjugated dimeric IgA, monomeric IgA, IgM, IgG or1% BSA were applied to intestinal sections. 3 The dimeric IgA, monomeric IgA, and IgG had been adjusted to 250 p,g per ml in 1% BSA and IgM to 500 p,g per ml. After 2 hr at 4°C, the slides were washed with phosphate-buffered saline and incubated for 2 more hours WC) with a mixture of 5 p,l of the IgA conjugate o~r-_,--~--~(150 p,g per ml) and 20 p,l of one of the unconjugated so 80 100 40 60 80 100 40 proteins. The slides were washed and reacted with ELUTION VOLUME lmll FIG. 1. Sephadex G-200 gel filtration patterns of 125 1-labeled diaminobenzidine. Results showed that the unconjusecretory component (SC) after incubation with monomeric lgA, gated dimeric IgA and IgM effectively inhibited binding dimeric IgA, 198 IgM, or lgG. The peak of the elution curve for free by the IgA conjugate, whereas IgG did not inhibit, and monomeric IgA was only minimally or not inhibitory SC alone is indicated by arrows. 0

December 1977

BINDING OF IgA AND IgM TO INTESTINAL EPITHELIUM

1335

·,_~ ... '

.. ~

~ ..

\

..

...;~

'• .

FIG. 2. A, human jejunal crypt epithelium reacted with horseradish peroxidase-conjugated rabbit antisecretory component (anti-SC). B, jejunum reacted with peroxidase-conjugated dimeric IgA. The location of darkly stained reaction product of diaminobenzidine, indicating the sites of the anti-SC or IgA dimer, are similar in the two sections. Supranuclear cytoplasm and basolateral cell margins of columnar epithelia l cells are stained. Goblet cell mucus is not stained. The large, densely stained cells in the lamina propria are endogenous peroxidase-containing cells. In control sections reacted with peroxidase alone or peroxidase-labeled nonimmune rabbit y-globulin, only endogenous peroxidase-containing cells were st ained (as in figs, 3A, 4A , 5A) ( x 300).

(fig. 3, table 1). Inhibition of binding was less successful when the conjugate was applied directly to sections after they had been reacted with unconjugated IgM or dimeric IgA than when applied in an excess of these proteins. This observation suggests relatively "loose" binding of the immunoglobulins to the tissue. In other experiments, the unconjugated dimeric IgA was reacted either with free 8C or with 118 exocrine IgA before being applied to the tissue sections. Fortyfive micrograms of free 8C or 118 IgA were incubated in the presence of 30 11-g of the conjugate in 1% B8A ovemight at 4°C. The mixtures or a control solution of dimeric IgA conjugate in B8A were applied to tissue sections and incubated for 2 hr at 4°C. The preincubation in free 8C almost completely abrogated binding of the conjugated IgA to tissues, whereas 118 IgA had no or little inhibitory effect (fig. 4, table 1). We interpreted these results as showing that soluble free 8C could

combine with the conjugated dimeric IgA and prevent its attachment to tissues, but 8C already bound to IgA (in 118 IgA) could not. In another test of specificity, the sections were reacted with various antibody preparations, then with the IgA conjugate. Unconjugated Fab' fragments of rabbit y globulin anti-8C, Fab' fragments of anti-IgA, intact anti-IgM, or intact antimouse IgG were applied to the tissues and reacted for 1 hr at room temperature. The sections were washed in phosphate-buffered saline and reacted with the conjugat ed dimeric IgA as usual. Only the anti-8C inhibited binding of the dimeric IgA (fig . 5, table 1). Altogether we studied about 15 intestinal specimens from normal subjects and 1 from a patient with the variable immunodeficiency syndrome whose epithelium contained 8C but no immunohistochemically demonstrated immunoglobulins. All specimens bound the con-

1336

BROWN ET AL.

Vol. 73, No.6

A ..

.. <

'

"f•·"?. ~

~

,_.,;;t "
.. FIG. 3. A, human jejunal epithelium reacted with unconjugated dimeric lgA, then with peroxidase-conjugated dimeric lgA. B , jejunum reacted with unconjugated monomeric lgA, then with peroxidase-conjugated dimeric lgA. The unconjugated dimeric lgA nearly completely inhibited reaction of the conjugate with the tissue, but monomeric lgA had little inhibitory effect (compare to fig. 2B ) (x300).

TABLE

1. Reaction ofperoxidase-conjugated dimeric IgA with human

intestinal epithelium Conditions

Control Pretreatmenta of tissue sections with Dimeric myeloma lgA Monomeric serum IgA 7SigG 198 IgM Preincubation oflgA conjugate with Free SC llS exocrine IgA Pretreatment of tissue sections with Fab' anti-SC Fab' anti-IgA Anti-lgM Anti-mouse lgG

jugated dimeric IgA; binding by the immunoglobulindeficient epithelium was the most intense.

Intensity of reaction

+ +++ 0 to+ +++ to++++ + + ++ 0 to +

0 ++ + + 0

+++ + ++++ ++++

a In addition to pretreatment of the tissue sections with the proteins listed, the conjugated dimeric lgA was applied to the sections in a n excess of each protein (see text, Results).

Discussion The transport of antibodies into gastrointestinal secretions has been a subject of interest for many years. In recent years, attention has focused on IgA, quantitatively the major class of immunoglobulin in intestinal fluids.1- 3 Unresolved has been the question of whether SC plays a role in transport of IgA by combining with the immunoglobulin in the epithelium, or merely combines with IgA in the gut lumen after independent secretion of these two components of exocrine liS IgA. Recent immunohistological studies from our laboratories6 and others7 • 8 supported the concept of an epithelial site of SC-IgA union by showing that the same epithelial cells that contain SC contain IgA. Furthermore, our immunoelectronmicroscopic studies6 localized SC and IgA to identical sites on the basolateral plasma

December 1977

1337

BINDING OF IgA AND IgM TO INTESTINAL EPITHELIUM

"!-'

"

41i '

f

'

FIG. 4. A, human jejunal epithelium reacted with peroxidase-conjugated dimeric IgA that had been preincubated with free secretory component (SC). B, jejunum reacted with peroxidase-conjugated dimeric lgA that had been preincubated with 118 colostral lgA (x300).

membranes, pinocytotic invaginations of the membranes, and similar-appearing cytoplasmic vesicles of columnar cells. This suggested that IgA could combine with SC initially on plasma membranes of the cells and be internalized by pinocytosis, a possibility that Brandtzaeg suggested earlier from his light microscopic observations.7 Results of the present study reinforce the concept of binding between SC and IgA occurring in the epithelium. Observations that indicated a specific epithelial reaction between lgA and SC were: (1) the distribution of peroxidase-labeled dimeric lgA when bound to the epithelium closely resembled that of SC as demonstrated by anti-SC antibodies; (2) binding of peroxidaselabeled dimeric lgA was inhibited by dimeric IgA but poorly or not inhibited by monomeric IgA or IgG; (3) binding of peroxidase-labeled dimeric IgA was abrogated by preincubation of the dimeric IgA with free SC but not with SC already bound to IgA in 118 colostral lgA; (4) time, temperature, and pH conditions for binding of IgA to the epithelium were compatible with

those for IgA-SC or IgM-SC binding in solution;10 (5) binding oflgA was abrogated by treating the epithelium with antibodies to SC but not with antibodies to human imunoglobulins known to be associated with the epithelium (lgM and lgA) or to a heterologous mouse immunoglobulin. One could question whether anti-SC antibodies interfered with IgA binding by sterically masking binding sites other than SC. This seems unlikely because Fab' fragments of anti-SC were used, and neither Fab' anti-lgA nor intactmolecules of anti-IgM inhibited IgA binding. That lgM inhibited the binding of dimeric IgA to intestinal epithelium suggests that IgM also binds to epithelial cell SC. This is consistent with the observation that IgM shares with SC and IgA certain ultrastructural locations in intestinal epithelial cells, 6 and with similarities between polymeric IgM and IgA with respect to affinity for SC. 10- 12 IgM readily binds soluble SC, 10• 12 and some studies have found IgM in parotid, breast, and intestinal secretions to be complexed with SC.4 • 5 • 9 Although studies in our laboratory recently

1338

BROWN ET AL.

Vol. 73 , No.6

A

B

.., --~~:~~-:~

'·

..

~

. L_

FIG. 5. A , human jejunal epithelium reacted with Fab' antisecretory component (anti-SC) antibodies, then with peroxidase-conjugated dimeric IgA. B , jejunum reacted with antimouse immunoglobulin, then with the lgA conjugate ( x 300)

showed that very little of the IgM in human intestinal fluid is SC-linked, SC was readily cleaved from SC-IgM complexes by incubation in intestinal fluids. 5 IgM, then, might combine with SC during transit across intestinal epithelial cells, but promptly dissociate from SC in the intestinal milieu. The observations that IgA and IgM can bind to SC in epithelial cells suggest that the cells normally contain an excess of SC that is not bound to endogenous immunoglobulin. This would be expected if epithelial cell SC is indeed a receptor for immunoglobulins.7 • 13 Also, it is consistent with the observations that free SC is secreted into external body fluids 2 and that antibodies specific for free SC reacted with gut epithelial cells. 7• 8 From our studies, a conclusion cannot be drawn as to whether IgA or IgM, during in vivo transport, bind initially to SC on plasma membranes of epithelial cells or somewhere in the cytoplasm. Logically, the membranes are the site for the initial reaction to occur, for it is there that the immunoglobulins probably first come into contact with cell-associated SC. Furthermore, antibodies specific for immunoglobulin-bound SC stained epi-

thelial cell margins/ and SC, IgA, and IgM were found at identical ultrastructural sites in epithelial cell plasma membranes. 6 This study did not address the question of what bonds might link IgA or IgM to SC in the intestinal epithelium. Other evidence indicates, though, that SC binds to dimeric IgA by disulfide bonds and noncovalent forces, whereas it binds to IgM noncovalently only_HH 2 Of interest also is the fact that the IgA 1 subclass used in this work readily binds to SC, even though IgA2 is the major subclass in external body fluidsY We wish to emphasize that our work does not exclude the possibility that some IgM or IgA reaches the intestinal lumen by routes other than that involving SC, and of course it does not explain how immunoglobulins which do not bind SC (lgE and IgG) cross the epithelium. Monomeric IgA and IgM, as well as IgE and IgG, are present in intestinal fluids. 3 • 5 • 18 Nevertheless, our studies support the hypothesis7 • 13 that SC participates in the selective transport of immunoglobulins into intestinal fluids by serving as an epithelial cell receptor for polymeric forms of IgA and IgM.

December 1977

BINDING OF IgA AND IgM TO INTESTINAL EPITHELIUM REFERENCES

1. Tomasi TB Jr, Tan EM, Solomon A, et a!: Characteristics of an immune system common to certain external secretions. J Exp Med 121:101-124, 1965 2. Tomasi TB, Grey HM: Structure and function of immunoglobulin A. Prog Allergy 16:81-213, 1972 3. Bull DM, Bienenstock J, Tomasi TB Jr: Studies on human intestinal immunoglobulin A. Gastroenterology 60:370-380, 1971 4. Thompson RA: Secretory piece linked to IgM in individuals deficient in IgA. Nature 226:946-948 , 1970 5. Richman LK, Brown WR: Characterization of IgM in human intestinal fluid (abstr). Clin Res 25:112A , 1977 6. Brown WR, Isobe Y, Nakane P: Studies on translocation of immunoglobulins across intestinal epithelium. II. Immunoelectronmicroscopic localization of immunoglobulins and secretory component in human intestinal mucosa. Gastroenterology 71:985-995 , 1976 . 7. Brandtzaeg P : Mucosal and glandular distribution of immunoglobulin components: differential localization of free and bound SC in secretory epithelial cells. J Immunol 112:1553-1559, 1974 8. Poger ME, Lamm ME: Localization of free and bound secretory component in human intestinal epithelial cells. A model for the assembly of secretory IgA. J Exp Med 139:629-642, 1974 9. Brandtzaeg P: Human secretory immunoglobulin M. An immunochemical and immunohistochemical study. Immunology 29:559-570 , 1975 10. Weicker J, Underdown BJ: A study of the association of human

11.

12.

13.

14.

15.

16.

17.

18.

1339

secretory component with IgA and IgM proteins. J Immunol 114:1337-1344, 1975 Brandtzaeg P: Human secretory immunoglobulins. VI. Association of free secretory piece with serum IgA in vitro. Immunology 21:323-332, 1971 Eskeland T, Brandtzaeg P: Does J chain mediate the combination of 19S IgM and dimeric IgA with the secretory component rather than being necessary for their polymerization? Immunochemistry 11:161-163, 1974 O'Daly JA, Craig SW, Cebra JJ: Localization of b markers, achain and SC of IgA in epithelial cells lining Lieberkuhn crypts. J Immunol 106:286-288, 1971 Brown WR, Lee EM: Radioimmunologic measurements of naturally occurring antibodies. I. Human serum antibodies reactive with Escherichia coli in gastrointestinal and immunologic disorders. J Lab Clin Med 82:125-136, 1973 Brown WR, Newcomb RW, Ishizaka K: Proteolytic degradation of exocrine and serum immunoglobulins. J Clin Invest 49:13741380, 1970 Klinman NR, Taylor RB: General methods for the study of cells and serum during the immune response: the response to dinitrophenyl in mice. Clin Exp Immunol4:473-487, 1969 Grey HM, Abel CA, Yount WJ, eta!: A subclass of human yAglobulins (yA2) which lacks the disulfide bonds linking heavy and light chains. J Exp Med 128:1223-1236, 1968 . . Brown WR, Lee EH: Studies on IgE in human mtestmal flmds. Int Arch Allergy Appl Immunol 50:87-94, 1976