Tronsjusion Medicine, 1992, 2, I5 I - I51

Differences in the characteristics of human monoclonal and polyclonal anti-D as revealed by immunochernical investigations: human monoclonal antibodies share specificities with natural antibodies S. J. Thorpe, S. W. Bailey, H. C. Gooi* and K. M. Thompson Division of Hoemotology, NotionolInstitute for Biologicol Stondords ond Control, Blanche h e , South Mimms, Hergordshire. * Region01 Tronsfmion Centre, Bridle Paih, b e d s and t Department of Biochemistry. Polytechnic of East London, Rornford Rood, London, U.K. Received 22 July 1991; occepted for publication 25 October 1991

SUMMARY.To determine the basis of the tissue cross-

reactions shown by some human monoclonal anti-Rh D antibodies, we have investigated the tissue reactivities of 48 further human monoclonal antibodies (mAb) against D and other Rh antigens, and compared them with those of normal and anti-D sera and immunoglobulin preparations, and affinity-purified polyclonal anti-D antibodies. Although we were unable to detect any tissue reactivities associated with the D-binding fraction of poIyclona1 antisera or prophylactic immunoglobulin, the non-erythroid cell types identified by the tissue-reactive human anti-Rh mAb of both

Although the Rh D antigen is believed to be restricted to human erythrocytes, early published reports on the distribution of Rh antigens, using anti-Rh sera, are conflicting and were confined mainly to leucocytes, sperm and secretions (Boorman-& Dodd, 1943; Mohn & Witebsky, 1948; Ashhurst et al., 1956; Jankovic & Lincoln, 1959; Mhjsky & Hraba, 1960; Levine & Celano, 1961; Mollison, 1979; Dunstan et al., 1984). Recent immunochemical studies of six human monoclonal anti-D antibodies have shown that three crossreact with human and animal tissues (Thorpe, 1989, 1990). Two IgG antibodies (FOG-] and UCHD4) were found to cross-react with smooth muscle, and the IgM antibody MAD-2 was shown to cross-react with vimentin. These observations can be accounted for as follows: either some Rh D-related epitopes occur in non-erythroid cell types, or alternatively, some human

IgM and IgG class were those recognized by antibodies present in both normal and anti-D sera. These results indicate: (a) that the tissue specificities of human antiRh mAb are similar to those of natural antibodies, and (b) that there are immunochemical differences between polyclonal and monoclonal anti-D antibodies, at least of IgG class, which may be relevant to the use of the latter in the prevention of haemolytic disease of the new-born by immune prophylaxis. Key words: human mAb, natural antibodies, Rh antigens.

monoclonal anti-D antibodies can react with structurally distinct epitopes present on the D antigen and tissue components. The latter is an important consideration of mAb intended for prophylaxis. The aims of the present study were to: (a) investigate the basis of these cross-reactions, and (b) determine whether the immunochemical characteristics of human IgG anti-D mAb are represented in polyclonal anti-D. We have therefore assessed the tissue reactivities of 48 further human mAb against D and other Rh antigens, compared them with those present in normal and antiD sera, and re-examined anti-D sera and prophylactic immunoglobulin for tissue reactivities associated with the D-binding fraction. MATERIALS A N D M E T H O D S Monoclonal anii- Rh antibodies and serum samples

Correspondence: Dr S. J. Thorpe, Division of Haernatology, National Institute for Biological Standards and Control. Blanche Lane, South Mirnrns, Potters Bar, Hertfordshire EN6 3BQ,U.K.

MAD-2 and FOG-I were kindly provided by the Bio Products Laboratory, Elstree, Hertfordshire, U.K.,

S.J . Thorpe et al.

152

and UCHD4 culture supernatant was a generous gift from Professors E. R. Huehns and D. Crawford. Forty-eight other human mAb against Rh antigens (D, E, C, e and c) were produced from immunized donors (37 IgM and 1 I IgG) as described previously (Thompson et a/., 1986, 1990; Melamed et al., 1987). Serum samples were obtained from 16 anti-D donors with either anti-D (two samples), anti-D + C (three samples) or anti-D C + E (1 1 samples) specificity. The plasma or serum titres ranged from 38 to 435 IU/ml. Normal serum samples were obtained from eight D-positive and four D-negative volunteers.

+

Assessment of tissue reactivities

Indirect immunofluorescence of cryostat sections of animal tissues (rabbit heart and kidney and rat brain) was carried out using culture supernatants containing monoclonal antibody, or anti-D sera or normal sera at dilutions of 1/5, 1/10 and 1/20, or anti-D or normal IgG immunoglobulin at 0.15-2 mg/ml, followed by fluorescein isothiocyanate-labelled anti-human IgM or anti-human IgG (Dako Ltd, High Wycombe, Bucks; Sigma Chemical Co., Poole, Dorset, U.K.) as previously described (Thorpe, 1989). Monoclonal antibody staining was also carried out on sections preincubated with, and in the presence of, the following ‘blocking’ reagents: normal rabbit serum (1/10 dilution), 0.5% (w/v) bovine serum albumin (BSA) and 0.05% (v/v) Tween 20. Animal tissues were chosen to assess tissue reactivity rather than human tissues because, as previously noted (Thorpe, 1989), indirect immunofluorescence techniques using fluoresceinlabelled anti-human Ig often result in high background fluorescence and can make accurate assessment difficult. Immunoblotting of rabbit heart tissue components separated by SDS-PAGE in 4-16% acrylamide slab gels and electrophoretically transferred to nitrocellulose was carried out as described previously (Thorpe, 1989). Fifty microlitres of normal or anti-D serum in 20 ml of PBS containing 3% (w/v) haemoglobin/track were used followed by ‘2SI-labelledmouse monoclonal anti-human IgG and anti-human IgM together. The anti-D serum samples were retested following absorption with Rh+ve (RIR,+ R2r) and Rh -ve (rr) erythrocytes (100 pl serum/2 ml packed cells). Antibody eluted from the cells with 0.1 M sodium citrate buffer, pH 4.0, and neutralized by the addition of 1 M Tris, was also tested by immunoblotting. Anti-D antibodies were similarly affinity-purified on a larger scale from a prophylactic anti-D immunoglobulin preparation (900 rng of protein/80 ml packed Rh ve erythrocytes) and concentrated using protein

+

G-sepharose. The eluate was tested for anti-D activity by haemagglutination (albumin displacement technique using 3% (v/v) group 0 Rlr and 0 rr red cell suspensions in glass tubes at a 1 : 1 ratio of antibody solution :cells), and for anti-tissue reactivities by immunofluorescence and immunoblotting. RESULTS Human anti-Rh mAb cross-react with tissue components

Using immunofluorescence, 22 of the 38 IgM anti-Rh mAb (58%) reacted with tissue components. The cell types most frequently recognized were vascular smooth muscle, ependymal cells lining the ventricles of brain, and astroglia (Bergman glia in cerebellum and/ or astrocytes), identified by 18 (82%), 17 (77%) and 15 (68%) of the tissue-reactive mAb, respectively. Other components stained by the IgM mAb included cardiac muscle, collecting ducts of kidney and cell nuclei. Tissue reactivity was associated with all anti-Rh specificities and, with one exception, the mAb reacted with more than one cell type. However, there did not appear to be any correlation between Rh antigen specificity and the distribution of tissue immunofluorescence. In contrast to the broad reactivities shown by the IgM mAb, the IgG anti-D mAb showed less frequent and more restricted patterns of tissue reactivity. Only three (19%; FOG- 1, UCHD4 and MS-26; all IgG 1) of a total of 11 IgGl and five IgG3 anti-D mAb (including those studied previously) reacted with components in the tissue panel studied, and in each case the predominant component identified was smooth muscle. Astroglia were also recognized by MS-26 (Fig. 2a and b) and, to a much lesser extent, FOG-1; ependymal cells were stained by MS-26. The tissue staining shown by the mAb was not inhibited by any of the ‘blocking’ reagents commonly used to prevent non-specific interactions or binding to Fc receptors. Cell types recognized by human anti-Rh mAb are those recognized by antibodies in normal serum

Immunohistochemistry using 12 normal serum samples showed that the tissue components identified by the anti-Rh mAb were those commonly recognized by apparently naturally occurring low titre antibodies in normal sera. All the sera contained IgM antibodies against some or all of the components recognized by the IgM antiRh mAb (Fig. 1 shows examples of the similarities in staining patterns); most sera contained IgG antibodies

Properties of monoclonal and polyclonal anti-D 153

Fig. 1. Immunofluorescence photomicrographs of cryostat sections of rabbit (a, b, i-I) and rat (c-h) tissues to show the similarities in staining patterns given by human monoclonal anti-Rh antibodies of IgM class (a, c, e, g, i and k) and natural IgM antibodies in normal human serum samples (b, d, f, h, j and I). (a and b) An anti-E mAb (MS-12) and serum ST stain collecting ducts in kidney; (c and d) an anti-c mAb (MS30) and serum TB stain ependymal cells in brain; (e and f) staining of Bergman glia (e) and cell nuclei in cerebellum by an anti-D mAb (NELP6) and serum PG; (g and h) an anti-D mAb (NELP-4)and serum TB react with Bergman glia and astrocyte processes (D.)in cerebellum; (i and j) staining of cardiac muscle by an anti-c mAb (MS-30) and serum LP; (k and I) staining of vascular smooth muscle by an anti-D mAb (NELP-6) and serum PG. The corresponding phase contrast photomicrographs of (i) and (j) are also shown (i’ and j’). Magnification: a, b, e-h x 196; c, d x 235; i, j x 3 14; k. I x 125.

against smooth muscle and over one-third contained IgG antibodies against Bergman glia and/or astrocytes, i.e. the same cell types recognized by IgG anti-D mAb (Fig. 2d and e). Although a number of other tissue components were also stained by serum IgG antibodies (e.g. endothelial cells, nuclei) these were not identified by the IgG anti-D mAb. Analysis of anti-D sera and polyclonal itnnitrnoglohrrlin for tissue reactivity associated with the D-binding fraction Using immunoftuorescence. A comparison of the

tissue reactivities in normal sera with those present in the anti-D serum samples failed to show obvious differences in the types of staining patterns or the intensity of staining of particular tissue components which might be attributable to the presence of anti-D. Similar tissue reactivities were also given by normal and anti-D prophylactic immunoglobulin preparations: at protein concentrations of 0-5-2 mg/ml, both gave high fluorescence throughout the tissue sections with accentuation of vascular smooth muscle and nuclei in brain; neither gave significant staining below 200-300 pg/ml. Identical immunofluorescence patterns were

154

S. J. Thorpe et al.

Fig. 2. Immunofluorescence photomicrographs to show staining of vessel walls, Bergman glia (e) and stellate astrocytes the IgG anti-D human mAb MS-26 (a and b) and staining of the same cell types by natural IgG antibodies in normal human serum (d and e; serum ST).Affinity-purified polyclonal IgG anti-D antibodies give no reaction with these components (c). The corresponding phase contrast photomicrograph of (c) is also shown (c'). Magnification: x 196. ( P)by

obtained using anti-D sera absorbed with Rh +ve and Rh -ve erythrocytes, and affinity-purified polyclonal anti-D antibodies from the immunoglobulin preparation gave no immunofluorescence of smooth muscle or any other tissue component (Fig. 2c), even though its haemagglutination titre was approximately six times higher than that of the tissue-reactive mAb and the immunofluorescence given by these mAb was still clearly present after diluting the supernatants to 1 in 8. Using immunoblotting. Overall, the anti-D sera reacted with a greater number of heart tissue components using immunoblotting than the normal sera. These components included those commonly recognized by the normal sera plus some additional components. However, none of this reactivity appeared to be associated with the D-binding fraction. Anti-D sera absorbed with Rh ve and Rh -ve erythrocytes gave identical immunoblotting profiles, and the anti-D eluted from the Rh +ve cells failed to give specific immunostaining (examples are shown in Fig. 3a). In contrast, the tissue reactivity shown by UCHD4 was lost following absorption with Rh +ve but not Rh

+

-ve erythrocytes but was recovered in the eluate from the Rh +ve erythrocytes (Fig. 3c). Several normal and anti-D sera reacted with tissue components of approximately 55 and 27 kDa, which may correspond to the tissue components recognized by the anti-D mAb MAD-2 and UCHD4 respectively, which in the case of MAD-2 corresponds to vimentin (Thorpe, 1989; 1990); the higher molecular weight component spanning 160-200 kDa additionally recognized by MAD-2 was indistinguishable from that also recognized by all the normal and the anti-D sera (Figs 3a and 4). The immunoblotting profiles of the normal and antiD immunoglobulin preparations were similar (225 pg protein applied): the 160-200 kDa component was clearly stained (example shown in Fig. 3b) and although the background staining was high, several other immunoreactive components were discernible, of the same approximate molecular weights as those commonly immunostained by individual sera. Affinity-purified anti-D antibodies from the same preparation showed no specific immunostaining (Fig. 3b).

Properties of monoclonal and polyclonal anti-D 155

Fig. 3. Autoradiographs showing immunoblots of electrophoretically separated components of rabbit heart homogenate to show that no specific immunostaining appears to be associated with the D-binding fraction of polyclonal immunoglobulin: (a) anti-D serum samples PR, GB, EA and CH absorbed with Rh +ve (lane I ) or Rh -ve (lane 2) cells, and the eluates from these cells (lanes 3 and 4 respectively), followed by '2SI-labelledanti-human IgG and IgM, (b) prophylactic anti-D immunoglobulin (lane 1) and the eluates from Rh +ve cells incubated with this preparation (lane 2) and a normal immunoglobulin preparation (lane 3), followed by 12sI-labelledanti-human IgG. (c) This shows positive control autoradiographs with WCHD4 unabsorbed WCHD4 (lane I), WCHD4 absorbed with Rh +ve (lane 3) and Rh -ve (lane 4) cells, and the eluates from these cells (lanes 5 and 6 respectively), followed by '2sI-labelledanti-human IgG, and this antibody alone (lane 2). The approximate molecular weights ( x lo-') of the major immunoreactive components are shown.

DISCUSSION Using immunofluorescence, the majority of human IgM mAb against Rh antigens reacted with tissue components. Although the D antigen is known to have several epitopes, the lack of a clear correlation between the Rh antigen specificity of the mAb and the distribu-

tion of tissue staining suggested that the tissue crossreactivity was not mediated by epitopes structurally related to D or other Rh antigens but through additional binding abilities of the mAb. The finding that strikingly similar tissue reactivities were commonly present in the IgM fraction of normal human sera suggests that the IgM mAb share specificities with

156

S. J. Thorpe et al.

Fig. 4. Autoradiographs showing immunoblots of electrophoretically separated components of rabbit heart homogenate with MAD-2 followed by 1251-labelled anti-human IgM, and 12 normal human serum samples (GC-AB) followed by '2SI-labelledanti-human IgM and IgG, and these latter antibodies alone (Con). The high molecular weight component recognized by MAD-2 also appears to be recognized by antibodies in normal sera (and anti-D sera, Fig. 3a). The of the major immunoreactivecomponents are shown. approximate molecular weights ( x

natural antibodies. This view is supported by recent data showing that, in common with natural antibodies, human anti-blood group IgM mAb are commonly multispecific and are able to react with a wide range of apparently unrelated auto- and foreign antigens (Thompson et af., in press). It is now well established that mAb can frequently cross-react with structurally dissimilar epitopes (reviewed by Ghosh & Campbell, 1986; Lane & Koprowski, 1982). Such interactions commonly involve multivalent antibodies with low intrinsic affinities for epitopes carried on high density, repeated sequences so the cross-reaction is amplified. This now appears to be the most likely explanation to account for the immunohistochemical cross-reactions described previously (Thorpe, 1989; 1990) and in the present study because we have subsequently shown that most of the tissue immunofluorescence shown by the anti-Rh mAb is due to reactivity with cytoskeletal components (S.J. Thorpe & S.W. Bailey, unpublished observations). Unfortunately, the IgM anti-D titres in the serum samples were too low to allow us to determine whether the immunochemical characteristics of the polyclonal IgM D-binding fraction are similar to those of the human IgM mAb. In contrast to the broad reactivities shown by the IgM mAb. the IgG anti-D mAb showed less frequent and more restricted patterns of tissue reactivity. Despite their narrower specificities, the strongest reactions

were again shown with smooth muscle and astroglia, both of which were commonly stained by IgG antibodies in normal (and anti-D) sera, which suggest that the IgG mAb tissue reactivity may also be mediated by the mAb sharing specificities with natural antibodies rather than through D-related epitopes. This supposition is supported by our failure to demonstrate any reactivity with smooth muscle, astroglia or any other component clearly associated with the D-binding fraction of polyclonal immunoglobulin, even though the haemagglutination titre of the affinity-purified polyclonal anti-D was nearly 50 times higher than the titre of the mAb at which tissue staining was still clearly present. Therefore, even if IgG anti-D molecules cross-reacting with smooth muscle constitute only 3-5% of the polyclonal anti-D fraction, they should have been detectable by immunofluorescence microscopy particularly because it has been reported that over 90% of pooled polyclonal TgG anti-D is IgG 1 (see Thomson et al., 1990), the same subclass as the tissue-reactive IgG anti-D mAb. These results provide further evidence that the D-antigen occurs only on human erythrocytes, and also show that the immunochemical characteristics of some human anti-D mAb. at least of IgG class, differ from the overall characteristics of polyclonal anti-D. Three possible explanations to account for the observable difference in tissue reactivities between

Properties of monoclonal and polyclonal anti-D

human polyclonal anti-D a n d certain monoclonal IgG anti-D antibodies are as follows. 1 A polyclonal antiserum contains a spectrum of antibody molecules derived from a number of clones and of different titres and directed against the same or different epitopes on the antigen. Any cross-reactive antibodies may be present in too low a concentration to allow detection. This obviously does not apply to monoclonal antibodies where the antibody molecules are identical and therefore may not show the same degree of overall anti-D dominance (see Lane & Koprowski, 1982). 2 Methods of human m A b production may result in or favour the selection of a restricted subset of anti-D antibodies with multispecific properties. 3 Some anti-D antibodies present in polyclonal antisera may be multispecific but these properties are masked by the binding of serum components to the relevant sites on the Ig molecules or by idiotypic interactions with other immunoglobulins.

In spite o f a concerted effort of procurement of antiD plasma, the success of anti-D prophylaxis and wider indications of its use have meant increasing demands for anti-D plasma and immunoglobulin. Human mAb against Rh D can be produced in theoretically unlimited quantities and are being evaluated for clinical use (Thomson cr al., 1990). However, as the immunological basis of anti-D prophylaxis is not understood, it is not possible to predict their clinical efficacy. It seems unlikely that the use of tissue-reactive monoclonal anti-D for prophylaxis will cause pathological problems because polyclonal immunoglobulin contains natural antibodies with similar specificities. The possibility, however, that this characteristic will adversely affect their clinical efficacy remains to be determined. ACKNOWLEDGEMENT We would like to thank Susanne Barsby for preparing the manuscript. REFERENCES Ashhurst, D.E., Bedford, D., Coombs, R.R.A. & Ronillard, L.M. (1956) Blood group antigens on human epidermal cells. Nature, 178, I 170-1 171. Boorman, K.E. & Dodd, B.E. (1943) The group-specific substances. A, B, M, N and Rh: their occurrence in tissues and body fluids. Journalof Pathology and Bacteriology, 55, 329-339. Dunstan, R.A., Simpson, M.B.& Rosse, W.F. (1984) Erythrocyte antigens on human platelets. Absence of Rh.

I57

Duffy, Kell, Kidd and Lutheran antigens. Transfusion, 29, 243-246. Ghosh, S . 8~Campbell, A.M. (1986) Multispecific monoclonal antibodies. Immunology Today, 7,217-222. Jankovic, B.D. & Lincoln, T.L. (1959) The presence of D (Rh) antigen in human leukocytes as demonstrated by the fluorescent antibody technique. Vox Sanguinis, 4, 119126. Lane, D. & Koprowski, H. (1982) Molecular recognition and the future of monoclonal antibodies. Nature, 296, 200202. Levine, P. & Celano, M.J. (1961) The question of D (Rho) antigenic sites on human spermatozoa. Vox Sanguinis, 6, 720-723. Majsky, A. & Hraba, T. (1960) Demonstration of D (Rho) agglutinogens in human spermatozoa. Fofia Biologica, 6, 342-347. Melamed, M.D., Thompson, K.M., Gibson, T. & HughesJones, N.C. (1987) Requirements for the establishment of heterohybridomas secreting monoclonal human antibody to rhesus (D) blood group antigen. Journalof Immunological Methods, 104, 245-25 1. Mohn, J.F. & Witebsky, E. (1948) The occurrence of watersoluble Rh substances in body secretions. New York State Journal of Medicine, 48, 287-290. Mollison, P.L. ( I 979) Blood Trunsfusionin Clinical Medicine. 6th edn, Blackwell Scientific Publications, Oxford. Thompson, K.. Barden, G., Sutherland, J., Beldon, I. & Melamed, M. (1990) Human monoclonal antibodies to C, c, E. e and G antigens of the Rh system. Immunology, 71, 323-327. Thomson, A.. Contreras. M., Gorick, B., Kumpel, B., Chapman, G.E., Lane, R.S., Teesdale, P.,Hughes-Jones, N.C. & Mollison, P.L. (1990) Clearance of Rh D-positive red cells with monoclonal anti-D. Lancet, 336, 1147-1 150. Thompson, K.M., Melamed, M.D.,Eagle, K., Gorick, B.D., Gibson, T., Holborn, A.M. & Hughes-Jones, N.C. (1986) Production of human monoclonal IgG and IgM antibodies with anti-D (rhesus) specificity using heterohybridomas. Immunology, 58, 157- 160. Thompson, K. M., Sutherland, J., Barden, G . , Melamed, M. D., Wright, M. G., Bailey, S. W. & Thorpe, S. J. (1992) Human monoclonal antibodies specific for blood group antigens demonstrate multispecific properties characteristic of natural autoantibodies. Immunology, 76 (I), 146- 157. Thorpe, S.J. (1989) Detection of Rh D-associated epitopes in human and animal tissues using human monoclonal antiD antibodies. British Journal of Haematology, 73, 527536. Thorpe, S.J. (1990) Reactivity of a human monoclonal antibody against Rh D with the intermediate filament protein vimentin. British Journalof Haernatology. 76, I 16120.