- Email: [email protected]
Binding of Polyreactive Antibodies to Self Versus Foreign Antigens
Immunobiol. (2002) 205, pp. 95– 107 © 2002 Urban & Fischer Verlag http://www.urbanfischer.de/journals/immunobiol
Departments of 1Surgery and 2Pathology, Duke University Medical Center, Durham, NC, USA
Binding of Polyreactive Antibodies to Self Versus Foreign Antigens WINSTON LEE1, JEFFREY G. GACA1, LINDSAY A. EDWARDS1, STEVEN J. BRAEDEHOEFT2, WILLIAM PARKER1, and R. DUANE DAVIS1 Received October 8, 2001 · Accepted in revised form November 26, 2001
Abstract Aside from their ability to bind to multiple antigens, the classic hallmark of polyreactive antibodies is their autoreactivity. Because of their ability to bind a number of common autoantigens, it has long been speculated that polyreactive antibodies are involved in the clearance of self-antigens. However, it has been demonstrated more recently that polyreactive antibodies are also capable of binding to some foreign and synthetic antigens. Although data regarding the relative reactivity of polyreactive antibodies with self versus foreign antigens is lacking, it is generally thought that both activities may play an important biological role. In this study, the relative reactivity of polyclonal human polyreactive IgM with human proteins and tissue extracts versus foreign (xenogeneic) proteins and tissue extracts was probed. The binding of affinity purified anti-ssDNA IgM from adult human serum and the binding of polyreactive IgM in human cord serum and in human adult serum were evaluated. Using competitive and direct binding assays, human polyreactive IgM were found to be generally more reactive with foreign (xenogeneic) proteins than with self or allogeneic proteins. These data shed light on the fundamental nature of polyreactive antibodies, and may provide additional insight into their putative biological roles.
Introduction Natural antibodies are those antibodies present in the blood without any known stimulus that might give rise to them. Within our natural antibody repertoire exists a subset of antibodies, known as polyreactive antibodies, which are capable of binding with low affinity to multiple structurally unrelated antigens. Antigens recognized by polyreactive antibodies include thyroglobulin, single-stranded DNA (ssDNA), actin, tubulin, and dinitrophenol, among others. (1, 2). Any given polyreactive antibody can bind to definable, restricted sets of antigens, not just to a single antigen (3–9). The human antibody repertoire first starts to develop within the fetus around the twentieth week of gestation (10). It is still unclear whether polyreactive antibodies are Abbreviations: BSA = bovine serum albumin; DNP = dinitrophenyl; EDTA = ethylenediamine tetraacetic acid; HSA = human serum albumin; ssDNA = single stranded DNA. 0171-2985/02/205/01-095 $ 15.00/0
96 · W. LEE et al. expressed at that time. However, it is known that at the time of birth, the peripheral antibody repertoire produced by the newborn consists mainly of polyreactive IgM (9). After birth, IgM remains the predominate isotype of polyreactive antibody throughout adulthood (3). Several investigators have found that polyreactive antibodies recognize the same set of antigens throughout the life of a given individual (3–9). Aside from their ability to bind to multiple antigens, the classic hallmark of polyreactive antibodies is their autoreactivity. However, polyreactive antibodies are also known to bind to a variety of foreign antigens (11–13). Because polyreactive antibodies bind to both autologous antigens and to foreign antigens, it is generally accepted that they may have a wide variety of biological functions. This notion is based in part on the idea that polyreactive antibodies bind “equally well” to foreign antigens and to autoantigens (11). However, data regarding the relative binding of polyreactive antibodies to autologous antigens versus foreign antigens are lacking. Such data might shed light on the biological role of polyreactive antibodies, and will provide additional insight into their fundamental nature. In this study, the relative reactivity of polyreactive IgM with autoantigens versus foreign antigens was probed. For this purpose, two polyclonal preparations of polyreactive IgM were utilized. First, pooled human cord serum, for which most of the IgM is polyreactive, was obtained. Second, antibodies from pooled human serum were affinity purified using immobilized ssDNA (a common antigen recognized by polyreactive antibodies). Binding of IgM in these two preparations of antibody to tissue extracts from allogeneic and xenogeneic lungs was examined. In addition, binding of IgM in the preparations to a panel of purified allogeneic and xenogeneic proteins was evaluated. Further, the binding of polyreactive antibodies in a panel of adult human sera to tissue extracts from allogeneic and xenogeneic lungs was examined. Using these approaches, polyreactive IgM were found to be generally more reactive with foreign (xenogeneic) proteins than with self proteins.
Materials and Methods Materials
Nunc Maxisorb plates were obtained from Gibbco Scientific, Inc (Coon Rapids, MN). Bovine serum albumin (fraction V), chicken serum albumin (99% pure; globulin free), Goat albumin (fraction V), guinea pig albumin (99%; fatty acid free), hamster albumin (fraction V), horse albumin (globulin free; 99% pure), human serum albumin (globulin free; 99% pure), mouse albumin (fraction V), pig albumin (globulin free; 99% pure), rabbit albumin (globulin free; 99% pure), sheep albumin (fraction V), turkey albumin (fraction V), bovine thyroglobulin, porcine thyroglobulin, affinity-isolated goat antibodies specific for human m chain, affinity-isolated goat antibodies specific for human m chain conjugated with alkaline phosphotase, purified human placental DNA, ammonium sulfate, polyoxyethylenesorbitan monolaurate (Tween20), and phosphatase substrate tablets were obtained from Sigma Chemical Co. (St. Louis, MO, USA). Human thyroglobulin (99% pure) was obtained from Golden West Biologicals Inc (Temecula, CA, USA). Sodium phosphate, monobasic monohydrate was obtained from J. T. Baker Inc. (Phillipsburg, NJ, USA). Tris hydrochloride and phosphate buffered saline were obtained from Life Technologies (Grand Island, NY, USA). Ethylenediamine tetraacetic acid (EDTA) was obtained from EM Science (Gibbstown, NJ, USA). Fibrous cellulose powder (CF11) was obtained from Whatman Inc. (Clifton, NJ, USA). Fresh frozen human plasma was purchased from the American Red Cross (Charlotte, NC, USA). Fresh human cord blood was obtained
Specificity of polyreactive antibodies · 97 from the Transfusion Service at Duke University Medical Center (Durham, NC, USA). Because of the limited amount of each sample (2 to 5 ml), samples from different individuals were pooled in order to carry out experiments. Preparation of ssDNA-cellulose.
An ssDNA-cellulose column was prepared as described previously (14). Forty grams of cellulose were washed twice by suspension in 500 ml of boiling ethanol (100%) to remove possible contamination with pyrimidine. The cellulose was then suspended sequentially in 1 l of 0.1 M NaOH, 1 l of 1mM Na3EDTA, and 1 l 10mM HCl. The cellulose was then washed with water until the wash was no longer acidic. The washed cellulose was lyophilized to dryness and stored at room temperature until use. To make single stranded human DNA, 200 mg of dried human placental DNA was dissolved in 100 ml of 10 mM K2HPO/1 mM Na3EDTA, pH 7.4 overnight at room temperature with constant stirring. The 2 mg/ml DNA solution was denatured at 100°C for 15 minutes and quenched in an ice bath. The solution was dialyzed three times for 40 minutes each time against 2 l of 20 mM TrisHCl, pH 7.4 at 4°C. Immediately following dialysis, the ssDNA was used to make the ssDNA-cellulose. To immobilize the ssDNA on the cellulose, 33 g of cellulose was added to 99 ml of the ssDNA solution in 20mM TrisHCl, pH 7.4. The resulting paste was blended gently to ensure that all cellulose was equally wet. The paste was spread out on a plate, covered with a piece of filter paper, and was left to dry at room temperature for 2 days. After drying, gentle pressure was applied to the ssDNA-cellulose in order to break it into a powder. The powdered mixture was lyophilized overnight to dryness. The dried powder was then washed once with 1 l of 10 mM Tris, 1 mM EDTA, pH 7.4 at 4°C, and then twice each time with 1 l of 10 mM Tris, pH 7.4 at 4°C. Following each wash, the DNA-cellulose was collected by sedimentation and the supernatant, which appeared cloudy due to the presence of small particles, was discarded. Finally, about 50 ml of 10 mM Tris, 1 mM EDTA, pH 7.4 was added to the ssDNA-cellulose and the slurry was flash frozen and stored at –80°C until use. Purification of anti-ssDNA antibodies.
Plasma samples from four individuals (300 ml per donor) were pooled, made 50% saturated with ammonium sulfate, and stirred for 2 hours at 4°C. The resulting mixture was centrifuged at 7,500 × g for 30 minutes at 4°C and the supernatant was discarded. After resuspending the pellets in 250 ml of 10 mM TrisHCl, pH 7.4, the mixture was spun at 7,500 × g for 30 minutes at 4°C. The supernatant was dialyzed successively with 3 l of 10 mM Tris, 150 mM NaCl, pH 7.4 (TBS) for 45 minutes, 4 l of TBS for 45 minutes, and 5 l of TBS overnight at 4°C. After thawing the frozen ssDNA-cellulose slurry in a 37°C water bath, the slurry was washed by suspension in 1 l of chilled TBS to remove fine particles. The dialyzed ammonium sulfate precipitate was mixed with the ssDNA-cellulose and gently swirled at 4°C for 2 hours. The ssDNA-cellulose was collected after sedimentation and then washed 3 times each with 1 l of chilled TBS. The ssDNA-cellulose was suspended in 150ml of 10 mM Tris, 500 mM NaCl, pH 7.4 with gentle swirling for 20 minutes at 4°C to elute anti-ssDNA antibodies. These elution conditions have been shown to effectively dissociate polyreactive antibodies from ssDNA, but not monoreactive anti-ssDNA antibodies (15). Following incubation with the elution buffer, the suspension was centrifuged at 7,500 × g for 30 minutes at 4°C. The supernatant was collected and made 10% (v/v) with glycerol, then snap-frozen in liquid nitrogen and stored at –80°C until needed. Just before use, the anti-ssDNA antibodies were thawed and diluted 1:4 (v/v) in 10 mM sodium phosphate, pH 7.4. Overall, this procedure resulted in a 0.67% yield of anti-ssDNA IgM (yield = final amount of anti-ssDNA IgM/initial amount of anti-ssDNA IgM) and a 21.2-fold increase in purity of anti-ssDNA IgM (purity = concentration of anti-ssDNA IgM/total protein concentration). Quantification of affinity purified anti-ssDNA IgM binding and serum IgM binding by ELISA.
The binding of IgM to various antigens was measured by ELISA. Briefly, 50 ml of 10 mg/ml purified protein were incubated on 96-well Nunc-Immuno Maxisorp polystyrene plates (VWR Scientific) over-
98 · W. LEE et al. night at 4°C to immobilize antigen on the plate. The plates coated with antigen were then washed 3 times each with 300 ml of PBS, and blocked for 1 hour at room temperature with a blocking solution that was varied depending on the experiment. Blocking solutions used were 0.1% human serum albumin (HSA) in PBS, 0.1% bovine serum albumin (BSA) in PBS or 0.1% bovine collagen in PBS. After blocking, the plates were washed 3 times with PBS and 50 ml of pooled human cord serum (n=20), human serum, or anti-ssDNA antibodies were added. Next, plates were incubated at room temperature for 3 hours and then washed 3 times with PBS. After washing the plate, 50 ml per well of affinity-purified alkaline phosphatase-conjugated goat antibodies specific for human m-chain in the blocking solution was added and the plate was incubated for 1 hour at room temperature. The plates were then washed 3 times with PBS, and 100 ml of developing solution consisting of 1.0 mg/ml p-nitrophenyl phosphate in 100 mM diethanolamine, 0.5mM MgCl2, and 0.2% (w/v) NaN3, pH 9.5 was added to each well. Absorbance at 405nm was determined using an EL-340 Bio Kinetics Reader (Bio-Tek Instruments, Winooski, VT, USA). Extraction of human and swine lung antigens
Human lung samples were obtained from discarded portions of transplant donor lungs that were used at Duke University Medical Center. Samples of human donor lung were obtained in accordance with the guidelines of the Duke University Institutional Review Board and with the approval of the Board. Porcine lung samples were obtained from discarded portions of donor lungs used in experimental pigto-baboon pulmonary xenotransplantation. All lungs had been perfused with a preservation solution, resulting in lung tissue without blood. Antigen from human and swine lung was extracted in the following manner: The lung samples were ground to a powder at –195°C using a BioPulverizer (Biospec, Bartlesville, OK, USA). Next, 0.5 to 1.0 g of the powder was suspended in 600 ml of PBS and the resulting suspension was filtered through four layers of gauze (Nu Gauze; Johnson and Johnson, Arlington, TX, USA) to remove large fragments of tissue. The filtrate was then centrifuged at 85 × g for 3 minutes. The supernatant was discarded and, to remove any remaining soluble proteins such as albumin or IgG, the pellets were resuspended in 600 ml of PBS and then centrifuged at 16,000 × g for 2 minutes. The supernatant was again discarded and the pellets, containing insoluble lung antigen, were utilized in competitive binding studies. To obtain soluble antigen for direct binding studies, 200 ml of the lung antigen pellet (obtained as described above) were mixed with 500 ml of PBS containing 2% TWEEN-20. The mixture was vortexed for 10 minutes and then centrifuged at 16,000 × g for 3 minutes at room temperature. The supernatant, containing the extracted lung antigen, was diluted in PBS so that the absorbance at 280 nm was 0.03, and then 50 ml/well were added to a Nunc-Immuno Maxisorp polystyrene plate. The plate was then incubated overnight at room temperature, followed by three washes each with 300 ml of PBS per well. The plate was blocked with 0.1% HSA in PBS for 2 hours at room temperature and then washed three times with 300 ml of PBS per well. Pooled human cord serum samples (50 ml), diluted appropriately in PBS, were added to each well and incubated for 3 hours at room temperature. The plate was washed, and alkaline phosphatase-conjugated goat antibodies specific for human m-chain were added as described above. The plates were washed again, developer added, and the absorbance at 405 nm determined as described above.
Results Binding of human cord serum IgM and anti-ssDNA IgM to a crude preparation of human and pig lung antigens as measured by direct ELISA
The relative reactivity of human cord serum IgM and anti-ssDNA IgM to immobilized human and pig lung antigen extracts were measured by an ELISA as described in the Methods. Both human cord serum IgM and anti-ssDNA IgM showed higher degrees of
Specificity of polyreactive antibodies · 99
Figure 1. The binding of human polyreactive IgM to human and pig lung antigens. The binding of cord serum IgM (A) and anti-ssDNA IgM (B) to immobilized antigens was measured by ELISA as described in the Methods. Human serum albumin (0.1%) was used as the blocking solution for the ELISA. The concentration of IgM in the pooled cord sera was 60 mg/ml. The experiment was performed in duplicate and the standard error is shown.
binding to the pig lung antigens than to the human lung antigens (Fig. 1). The human cord serum IgM was about 3 times more reactive toward pig lung antigens whereas the anti-ssDNA IgM was about 2 times more reactive. Binding of human cord serum IgM to pig lung antigens plateaued to some extent at a serum concentration of 50%, suggesting that the antigen was saturated or approaching saturation with antibodies at that concentration. The concentration at which the binding of anti-ssDNA antibodies plateaued was not determined.
100 · W. LEE et al. Binding of human polyreactive IgM to a crude preparation of human and pig lung antigens as measured by a competitive binding assay
Although the direct binding studies described above suggest that polyreactive antibodies are more xenoreactive than alloreactive or autoreactive, there are some limitations of direct binding studies. For example, differences in binding may represent differences in the blocking of non-specific binding to the plate. For this reason, the binding of human polyreactive antibodies to pig and human lung antigens was assessed using competitive binding assays. Specifically, the binding to DNP-albumin of IgM in adult human serum was evaluated by ELISA in the presence of human or porcine lung antigens. As shown in Figure 2, the presence of porcine lung antigens resulted in more inhibition of binding to DNP-albumin than did the presence of human lung antigens in all eight samples tested. The slight increase in binding of anti-DNP IgM in the presence of human lung antigen was not apparently due to comtamination of the lung antigen preparation with human IgM, since no binding of IgM was observed in controls lacking human serum. The slight increase in IgM binding in the presence of human lung antigens may be due to weak interactions between the human lung antigens and human polyreactive IgM, which could result in crosslinking of the IgM and thus increased cooperative binding of the IgM to the immobilized DNP. However, this idea remains to be tested.
Figure 2. The binding of human IgM to immobilized DNP-albumin in the presence or absence of human or pig lung antigens. The binding of IgM from the sera of eight normal human donors was evaluated by ELISA. In all sera tested, binding was less in the presence of pig lung antigens compared to human lung antigens, suggesting that human anti-DNP IgM binds more strongly to pig lung antigens than to human lung antigens. Bovine collagen (0.1% in PBS) was used as the blocking reagent, and the sera were diluted to 5% in PBS for the assay.
Binding of human cord serum IgM and anti-ssDNA IgM to purified albumins from various species as measured by direct ELISA
To further evaluate the reactivity of polyreactive IgM with self versus foreign antigens, the binding of human cord serum IgM and anti-ssDNA IgM to albumins from human, pig, and cow was evaluated by ELISA as described in the Methods. As a positive control, bind-
Specificity of polyreactive antibodies · 101
Figure 3. The binding of human polyreactive IgM to DNP-HSA and to human, pig, and bovine serum albumins. The binding of cord serum IgM (A) and anti-ssDNA IgM (B) to immobilized antigens was measured by ELISA as described in the Methods. Human serum albumin (0.1%) was used as the blocking solution for the ELISA. The concentration of IgM in the pooled cord sera was 60 mg/ml. The experiment was performed in duplicate and the standard error is shown.
ing to dinitrophenyl coupled HSA (DNP-HSA; a common antigen recognized by polyreactive antibodies) was also evaluated. As shown in Figure 3, both human cord serum IgM and anti-ssDNA IgM reacted most strongly with HSA-DNP, less strongly with porcine serum albumin, and least with HSA and bovine serum albumin. Wells were blocked with human serum albumin, and human serum albumin was included in the enzymeconjugated antibody as a blocker of nonspecific binding. Thus, binding to human serum albumin may have been decreased by this procedure. As with binding to tissue extracts
102 · W. LEE et al. (Fig. 1), we observed evidence of saturation of antigen using cord serum but not antissDNA antibodies under the conditions used. It was not clear whether the binding of IgM from cord serum was greater to human or to bovine serum albumin (Fig. 3a), since the dose-dependent binding overlapped at about 6% serum. One possible explanation for this observation is that the “background” binding, which might be greater at lower serum concentrations (where binding of serum components that can block “non-specific” binding to the plate is minimized), is greater in the wells coated with human serum albumin than in wells coated with bovine serum albumin. This might have occurred since the bovine albumin prep is only 96% pure whereas the HSA prep is 99% pure. Alternatively, the complex shape of the binding curves might reflect interplay between the many factors associated with binding of the heterogeneous antibodies to the plate such as antibody concentration, antibody avidity, and antigen concentration.
Figure 4. The binding of human polyreactive IgM to albumins from various species. The binding of cord serum IgM (A) and anti-ssDNA IgM (B) to immobilized albumins was measured by ELISA as described in the Methods. Bovine serum albumin (0.1%) was used as the blocking solution for the ELISA. Binding that was significantly (p < 0.05) different from the binding to HSA is indicated by the (*). The concentration of IgM in the pooled cord sera was 60 mg/ml, and the sera were diluted 1 to 10 in PBS for the experiment. The concentration of anti-ssDNA IgM was 800 ng/ml. Each point represents the average absorbance obtained from 5 wells and the standard error is shown.
Specificity of polyreactive antibodies · 103
Since it was imperative to this study not to underestimate the binding of the polyreactive IgM to human serum albumin, and since reactivity with bovine serum albumin was low (Fig. 3), bovine serum albumin rather than human serum albumin was utilized to block “nonspecific” binding in subsequent experiments. While this might cause underestimation of the binding of antibodies to some foreign albumins, subsequent results validated this approach (see below). As the next approach to evaluating the binding of polyreactive IgM to autologous versus foreign albumin, the binding of human cord serum IgM and anti-ssDNA IgM to albumins from 12 species was evaluated by ELISA. Binding to DNP-HSA was again evaluated as a positive control. As shown in Figure 4, the binding of IgM to all foreign albumins, with the exception of hamster serum albumin, was significantly greater than binding to human serum albumin for at least one source of polyreactive IgM. Binding of polyreactive IgM to chicken serum albumin was greater than the binding to human serum albumin, regardless of the source of IgM, and the binding of human cord serum IgM to dog serum albumin was slightly but significantly lower than the binding to human serum albumin. Since bovine serum albumin was used as a blocking reagent (included in the enzyme linked antibody incubation as described in the methods), it is possible that this albumin interfered with the binding of the polyreactive antibodies to some of the epitopes on the various albumins. However, even with this limitation, the results are consistent with the idea that polyreactive IgM may be more reactive with foreign antigens than with self antigens. Further, as described below, these studies were confirmed using solution phase, competitive binding studies. Binding of human polyreactive IgM to purified thyroglobulins as measured by direct ELISA
As another approach to evaluating the binding of polyreactive IgM to autologous versus foreign antigens, the binding of human cord serum IgM and anti-ssDNA IgM to porcine, bovine, and human thyroglobulin was evaluated by ELISA. Binding of polyreactive IgM to the foreign thyroglobulins was greater than binding to the human protein, regardless of the source of IgM (Fig. 5). It is known that normal adult human serum contains polyreactive antibodies that react with human thyroglobulin (16) and monoreactive antibodies that react with the carbohydrate portions of bovine and porcine thyroglobulin (1, 2). However, the monoreactive IgM that bind to the carbohydrates expressed on the bovine and porcine proteins are absent in cord sera (17, 18) and are not expected to be present in the anti-ssDNA preparation. Although speculative, it is possible that polyreactive antibodies react with the carbohydrate portions of the foreign thyroglobulins. Regardless of the epitopes recognized, these findings lend further support to the idea that polyreactive IgM are more reactive with foreign antigens than with self antigens. Binding of human polyreactive IgM to human and xenogeneic antigens as measured by a competitive binding assay
The ability of xenogeneic and allogeneic albumins in solution to inhibit binding of polyreactive antibodies to an immobilized antigen (bovine thyroglobulin) was evaluated. Bovine collagen (0.1% in PBS), which blocks plates to a similar extent as does human
104 · W. LEE et al.
Figure 5. The binding of polyreactive IgM to human, pig, and bovine thyroglobulins. The binding of cord serum IgM (A) and anti-ssDNA IgM (B) to immobilized thyroglobulins was measured by ELISA as described in the Methods. Both the human cord serum IgM and anti-ssDNA IgM showed significantly (p < 0.05) greater binding to bovine and pig thyroglobulin than to human thyroglobulin. Bovine serum albumin (0.1%) was used as the blocking solution for the ELISA. The concentration of IgM in the pooled cord sera was 60 mg/ml, and the sera were diluted 1 to 10 in PBS for the experiment. The concentration of anti-ssDNA IgM was 800 ng/ml. Each point represents the average absorbance obtained from 5 wells and the standard error is shown.
serum albumin, was used as a blocking reagent. As shown in Figure 6, chicken, guinea pig, hamster, pig and possibly horse albumins, but not dog, goat, rabbit, sheep, turkey or human albumin, inhibited binding of polyreactive IgM from human cord serum to bovine thyroglobulin. This binding pattern is identical to the pattern observed in the solid phase binding assay (Fig. 4a), suggesting that the difference in binding of polyreactive antibodies to different albumins observed using the solid phase assay is not due to putative artifacts. Interestingly, under the conditions used (5 mg/ml of albumin), no inhibition of binding of anti-ssDNA antibodies to bovine thyroglobulin was obtained using any albumin, human or xenogeneic. It is unknown whether this finding might be due to
Specificity of polyreactive antibodies · 105
Figure 6. The binding of human IgM to immobilized DNP-albumin in the presence of albumins (5 mg/ml) from various species. The binding of IgM in pooled cord serum to DNP-HSA was measured by ELISA as described in the Methods. Binding that was significantly (p < 0.05) different from the binding in the presence of HSA is indicated by the (*). The concentration of IgM in the pooled cord sera was 90 mg/ml, and the sera were diluted 1 to 10 in PBS for the experiment. Bovine collagen (0.1% in PBS) was used as the blocking reagent.
blocking of much of the binding of anti-ssDNA antibodies by bovine collagen (the blocking reagent used in the experiment). Discussion In this study, we have found that human polyreactive IgM reacted with a porcine tissue extract more so than with a human tissue extract. Similarly, human polyreactive IgM reacted either the same or more strongly with foreign (xenogeneic) serum albumins compared to human serum albumin. Further, human polyreactive IgM reacted more strongly with bovine and with porcine thyroglobulin than with human thyroglobulin. All of these observations support the idea that polyreactive IgM, both in adults and in humans, is directed against foreign antigens more so than against autoantigens. There are several possible explanations for this increased reactivity; first, foreign proteins might have more epitopes per protein than do self-proteins. Second, polyreactive antibodies might bind with higher affinity/avidity to foreign proteins than to self-proteins. Third, there may be higher concentrations of polyreactive antibodies that recognize foreign antigens than of antibodies that recognize self-antigens. Additional work using panels of monoclonal polyreactive antibodies may be needed to sort out which one(s) of these potential factors play a role in the increased reactivity of polyreactive antibodies with foreign antigens compared to self antigens.
106 · W. LEE et al. The observations described above were made using two sources of polyreactive IgM, cord serum IgM and affinity purified adult anti-ssDNA IgM. Although the tendency to react more with foreign antigens than with self-antigens was observed in both preparations, some differences in the two preparations were noted. For example, cord serum IgM reacted more strongly with horse albumin than with mouse or sheep albumin, whereas anti-ssDNA antibodies reacted more strongly with mouse or sheep albumin than with horse albumin. This might suggest that the binding of polyreactive anti-ssDNA antibodies against antigens unrelated to ssDNA is different than binding of the presumably average polyreactive antibodies present in cord serum. On the other hand, these differences may reflect differences in the relative purity of the two preparations. It is noteworthy that the similarity of various albumins to human albumin did not correlate with the binding of polyreactive antibodies against the albumin. For example, polyreactive IgM in the cord serum preparation bound to sheep serum albumin (75% identity to HSA) to a lesser extent than to pig serum albumin (76% identity to HSA). This lack of correlation was verified by a regression analysis of sequence identity to human serum albumin versus binding of polyreactive antibodies. This observation might suggest that reactivity of polyreactive antibodies with foreign albumins is dictated by a limited number of immunodominant epitopes. Acknowledgements
The authors thank KIM S. BARBER for invaluable assistance in preparing the manuscript. This work was supported by NIH grants HL50985 and HL52297 and by the FANNIE E. RIPPEL foundation.
References 1. GUILBERT, B., G. DIGHIERO, and S. AVRAMEAS. 1982. Naturally occurring antibodies against nine common antigens in human sera. J. Immunol. 128: 2779. 2. AILUS, K., and T. PALOSUO. 1995. IgM class autoantibodies in human cord serum. J. Reprod. Immunol. 29: 61. 3. LACROIX-DESMAZES, S., S. V. KAVERI, L. MOUTHON, A. AYOUBA, E. MALANCHERE, A. COUTINHO, and M. D. KAZATCHKINE. 1998. Self-reactive antibodies (natural autoantibodies) in healthy individuals. J. Immunol. Meth. 216: 117. 4. LOGTENBERG, T. 1990. Properties of Ployreactive Natural Antibodies to Self and Foreign Antigens. J. Clin. Immunol. 10: 137. 5. SIGOUNAS, G., N. KOLAITIS, E. MONELL-TORRENS, and A. L. NOTKINS. 1994. Polyreactive IgM antibodies in the circulation are masked by antigen binding. J. Clin. Immunol. 14: 375. 6. BOUVET, J. P., and G. DIGHIERO. 1998. From natural polyreactive autoantibodies to a la carte monoreactive antibodies to infectious agents: is it a small world after all? Infection & Immunity 66: 1. 7. STAHL, D., S. LACROIX-DESMAZES, L. MOUTHON, S. V. KAVERI, and M. D. KAZATCHKINE. 2000. Analysis of human self-reactive antibody repertoires by quantitative immunoblotting. J. Immunol. Meth. 240: 1. 8. LYDYARD, P. M., R. QUARTEY-PAPAFIO, B. BROKER, L. MACKENZIE, J. JOUQUAN, M. A. BLASCHEK, J. STEELE, M. PETROU, P. COLLINS, D. ISENBERG, and et al. 1990. The antibody repertoire of early human B cells. I. High frequency of autoreactivity and polyreactivity. Scand. J. Immunol. 31: 33. 9. MOUTHON, L., S. LACROIX-DESMAZES, A. NOBREGA, C. BARREAU, A. COUTINHO, and M. D. KAZATCHKINE. 1996. The self-reactive antibody repertoire of normal human serum IgM is acquired in early childhood and remains conserved throughout life. Scand. J. Immunol. 44: 243.
Specificity of polyreactive antibodies · 107 10. ANDERSSON, U., A. G. BIRD, B. S. BRITTON, and R. PALACIOS. 1981. Humoral and cellular immunity in humans studied at the cell level from birth to two years of age. Immunol. Rev. 57: 1. 11. NOTKINS, A. 2000. Polyreactive antibodies and polyreactive antigen-binding B (PAB) Cells. Curr. Top. Microbiol. Immunol. 252: 241. 12. PARKER, W., D. BRUNO, Z. E. HOLZKNECHT, and J. L. PLATT. 1994. Characterization and affinity isolation of xenoreactive human natural antibodies. J. Immunol. 153: 3791. 13. MACCHIARINI, P., R. ORIOL, A. AZIMZADEH, V. DE MONTPREVILLE, R. RIEBEN, N. BOVIN, M. MAZMANIAN, and P. DARTEVELLE. 1998. Evidence of human non-alpha-galactosyl antibodies involved in the hyperacute rejection of pig lungs and their removal by pig organ perfusion. J. Thor. and Card. Surg. 116: 831. 14. HERRICK, G., and B. ALBERTS. 1976. Purification and physical characterization of nucleic acid helix-unwinding proteins from calf thymus. J. Biol. Chem. 251: 2124. 15. KANAI, Y., and T. KUBOTA. 1989. A novel trait of naturally occurring anti-DNA antibodies: dissociation from immune complexes in neutral 0.3–0.5 M NaCl. Immunol. Lett. 22: 293. 16. THALL, A., and U. GALILI. 1990. Distribution of Gal alpha 1–3Gal beta 1–4GlcNAc residues on secreted mammalian glycoproteins (thyroglobulin, fibrinogen, and immunoglobulin G) as measured by a sensitive solid-phase radioimmunoassay. Biochem. 29: 3959. 17. PARKER, W., K. LUNDBERG-SWANSON, Z. E. HOLZKNECHT, J. LATEEF, S. A. WASHBURN, S. J. BRAEDEHOEFT, and J. L. PLATT. 1996. Isohemagglutinins and xenoreactive antibodies: members of a distinct family of natural antibodies. Hum. Immunol. 45: 94. 18. MINANOV, O. P., S. ITESCU, F. A. NEETHLING, A. S. MORGENTHAU, P. KWIATKOWSKI, D. K. COOPER, and R. E. MICHLER. 1997. Anti-Gal IgG antibodies in sera of newborn humans and baboons and its significance in pig xenotransplantation. Transplant. 63: 182. WILLIAM PARKER, Ph. D., Duke University Medical Center, Box 2605, Department of Surgery, Durham, NC 27710, Telephone: 919-681-3886, Fax: 919-681-7263, E-mail: [email protected]
Departments of 1Surgery and 2Pathology, Duke University Medical Center, Durham, NC, USA
Binding of Polyreactive Antibodies to Self Versus Foreign Antigens WINSTON LEE1, JEFFREY G. GACA1, LINDSAY A. EDWARDS1, STEVEN J. BRAEDEHOEFT2, WILLIAM PARKER1, and R. DUANE DAVIS1 Received October 8, 2001 · Accepted in revised form November 26, 2001
Abstract Aside from their ability to bind to multiple antigens, the classic hallmark of polyreactive antibodies is their autoreactivity. Because of their ability to bind a number of common autoantigens, it has long been speculated that polyreactive antibodies are involved in the clearance of self-antigens. However, it has been demonstrated more recently that polyreactive antibodies are also capable of binding to some foreign and synthetic antigens. Although data regarding the relative reactivity of polyreactive antibodies with self versus foreign antigens is lacking, it is generally thought that both activities may play an important biological role. In this study, the relative reactivity of polyclonal human polyreactive IgM with human proteins and tissue extracts versus foreign (xenogeneic) proteins and tissue extracts was probed. The binding of affinity purified anti-ssDNA IgM from adult human serum and the binding of polyreactive IgM in human cord serum and in human adult serum were evaluated. Using competitive and direct binding assays, human polyreactive IgM were found to be generally more reactive with foreign (xenogeneic) proteins than with self or allogeneic proteins. These data shed light on the fundamental nature of polyreactive antibodies, and may provide additional insight into their putative biological roles.
Introduction Natural antibodies are those antibodies present in the blood without any known stimulus that might give rise to them. Within our natural antibody repertoire exists a subset of antibodies, known as polyreactive antibodies, which are capable of binding with low affinity to multiple structurally unrelated antigens. Antigens recognized by polyreactive antibodies include thyroglobulin, single-stranded DNA (ssDNA), actin, tubulin, and dinitrophenol, among others. (1, 2). Any given polyreactive antibody can bind to definable, restricted sets of antigens, not just to a single antigen (3–9). The human antibody repertoire first starts to develop within the fetus around the twentieth week of gestation (10). It is still unclear whether polyreactive antibodies are Abbreviations: BSA = bovine serum albumin; DNP = dinitrophenyl; EDTA = ethylenediamine tetraacetic acid; HSA = human serum albumin; ssDNA = single stranded DNA. 0171-2985/02/205/01-095 $ 15.00/0
96 · W. LEE et al. expressed at that time. However, it is known that at the time of birth, the peripheral antibody repertoire produced by the newborn consists mainly of polyreactive IgM (9). After birth, IgM remains the predominate isotype of polyreactive antibody throughout adulthood (3). Several investigators have found that polyreactive antibodies recognize the same set of antigens throughout the life of a given individual (3–9). Aside from their ability to bind to multiple antigens, the classic hallmark of polyreactive antibodies is their autoreactivity. However, polyreactive antibodies are also known to bind to a variety of foreign antigens (11–13). Because polyreactive antibodies bind to both autologous antigens and to foreign antigens, it is generally accepted that they may have a wide variety of biological functions. This notion is based in part on the idea that polyreactive antibodies bind “equally well” to foreign antigens and to autoantigens (11). However, data regarding the relative binding of polyreactive antibodies to autologous antigens versus foreign antigens are lacking. Such data might shed light on the biological role of polyreactive antibodies, and will provide additional insight into their fundamental nature. In this study, the relative reactivity of polyreactive IgM with autoantigens versus foreign antigens was probed. For this purpose, two polyclonal preparations of polyreactive IgM were utilized. First, pooled human cord serum, for which most of the IgM is polyreactive, was obtained. Second, antibodies from pooled human serum were affinity purified using immobilized ssDNA (a common antigen recognized by polyreactive antibodies). Binding of IgM in these two preparations of antibody to tissue extracts from allogeneic and xenogeneic lungs was examined. In addition, binding of IgM in the preparations to a panel of purified allogeneic and xenogeneic proteins was evaluated. Further, the binding of polyreactive antibodies in a panel of adult human sera to tissue extracts from allogeneic and xenogeneic lungs was examined. Using these approaches, polyreactive IgM were found to be generally more reactive with foreign (xenogeneic) proteins than with self proteins.
Materials and Methods Materials
Nunc Maxisorb plates were obtained from Gibbco Scientific, Inc (Coon Rapids, MN). Bovine serum albumin (fraction V), chicken serum albumin (99% pure; globulin free), Goat albumin (fraction V), guinea pig albumin (99%; fatty acid free), hamster albumin (fraction V), horse albumin (globulin free; 99% pure), human serum albumin (globulin free; 99% pure), mouse albumin (fraction V), pig albumin (globulin free; 99% pure), rabbit albumin (globulin free; 99% pure), sheep albumin (fraction V), turkey albumin (fraction V), bovine thyroglobulin, porcine thyroglobulin, affinity-isolated goat antibodies specific for human m chain, affinity-isolated goat antibodies specific for human m chain conjugated with alkaline phosphotase, purified human placental DNA, ammonium sulfate, polyoxyethylenesorbitan monolaurate (Tween20), and phosphatase substrate tablets were obtained from Sigma Chemical Co. (St. Louis, MO, USA). Human thyroglobulin (99% pure) was obtained from Golden West Biologicals Inc (Temecula, CA, USA). Sodium phosphate, monobasic monohydrate was obtained from J. T. Baker Inc. (Phillipsburg, NJ, USA). Tris hydrochloride and phosphate buffered saline were obtained from Life Technologies (Grand Island, NY, USA). Ethylenediamine tetraacetic acid (EDTA) was obtained from EM Science (Gibbstown, NJ, USA). Fibrous cellulose powder (CF11) was obtained from Whatman Inc. (Clifton, NJ, USA). Fresh frozen human plasma was purchased from the American Red Cross (Charlotte, NC, USA). Fresh human cord blood was obtained
Specificity of polyreactive antibodies · 97 from the Transfusion Service at Duke University Medical Center (Durham, NC, USA). Because of the limited amount of each sample (2 to 5 ml), samples from different individuals were pooled in order to carry out experiments. Preparation of ssDNA-cellulose.
An ssDNA-cellulose column was prepared as described previously (14). Forty grams of cellulose were washed twice by suspension in 500 ml of boiling ethanol (100%) to remove possible contamination with pyrimidine. The cellulose was then suspended sequentially in 1 l of 0.1 M NaOH, 1 l of 1mM Na3EDTA, and 1 l 10mM HCl. The cellulose was then washed with water until the wash was no longer acidic. The washed cellulose was lyophilized to dryness and stored at room temperature until use. To make single stranded human DNA, 200 mg of dried human placental DNA was dissolved in 100 ml of 10 mM K2HPO/1 mM Na3EDTA, pH 7.4 overnight at room temperature with constant stirring. The 2 mg/ml DNA solution was denatured at 100°C for 15 minutes and quenched in an ice bath. The solution was dialyzed three times for 40 minutes each time against 2 l of 20 mM TrisHCl, pH 7.4 at 4°C. Immediately following dialysis, the ssDNA was used to make the ssDNA-cellulose. To immobilize the ssDNA on the cellulose, 33 g of cellulose was added to 99 ml of the ssDNA solution in 20mM TrisHCl, pH 7.4. The resulting paste was blended gently to ensure that all cellulose was equally wet. The paste was spread out on a plate, covered with a piece of filter paper, and was left to dry at room temperature for 2 days. After drying, gentle pressure was applied to the ssDNA-cellulose in order to break it into a powder. The powdered mixture was lyophilized overnight to dryness. The dried powder was then washed once with 1 l of 10 mM Tris, 1 mM EDTA, pH 7.4 at 4°C, and then twice each time with 1 l of 10 mM Tris, pH 7.4 at 4°C. Following each wash, the DNA-cellulose was collected by sedimentation and the supernatant, which appeared cloudy due to the presence of small particles, was discarded. Finally, about 50 ml of 10 mM Tris, 1 mM EDTA, pH 7.4 was added to the ssDNA-cellulose and the slurry was flash frozen and stored at –80°C until use. Purification of anti-ssDNA antibodies.
Plasma samples from four individuals (300 ml per donor) were pooled, made 50% saturated with ammonium sulfate, and stirred for 2 hours at 4°C. The resulting mixture was centrifuged at 7,500 × g for 30 minutes at 4°C and the supernatant was discarded. After resuspending the pellets in 250 ml of 10 mM TrisHCl, pH 7.4, the mixture was spun at 7,500 × g for 30 minutes at 4°C. The supernatant was dialyzed successively with 3 l of 10 mM Tris, 150 mM NaCl, pH 7.4 (TBS) for 45 minutes, 4 l of TBS for 45 minutes, and 5 l of TBS overnight at 4°C. After thawing the frozen ssDNA-cellulose slurry in a 37°C water bath, the slurry was washed by suspension in 1 l of chilled TBS to remove fine particles. The dialyzed ammonium sulfate precipitate was mixed with the ssDNA-cellulose and gently swirled at 4°C for 2 hours. The ssDNA-cellulose was collected after sedimentation and then washed 3 times each with 1 l of chilled TBS. The ssDNA-cellulose was suspended in 150ml of 10 mM Tris, 500 mM NaCl, pH 7.4 with gentle swirling for 20 minutes at 4°C to elute anti-ssDNA antibodies. These elution conditions have been shown to effectively dissociate polyreactive antibodies from ssDNA, but not monoreactive anti-ssDNA antibodies (15). Following incubation with the elution buffer, the suspension was centrifuged at 7,500 × g for 30 minutes at 4°C. The supernatant was collected and made 10% (v/v) with glycerol, then snap-frozen in liquid nitrogen and stored at –80°C until needed. Just before use, the anti-ssDNA antibodies were thawed and diluted 1:4 (v/v) in 10 mM sodium phosphate, pH 7.4. Overall, this procedure resulted in a 0.67% yield of anti-ssDNA IgM (yield = final amount of anti-ssDNA IgM/initial amount of anti-ssDNA IgM) and a 21.2-fold increase in purity of anti-ssDNA IgM (purity = concentration of anti-ssDNA IgM/total protein concentration). Quantification of affinity purified anti-ssDNA IgM binding and serum IgM binding by ELISA.
The binding of IgM to various antigens was measured by ELISA. Briefly, 50 ml of 10 mg/ml purified protein were incubated on 96-well Nunc-Immuno Maxisorp polystyrene plates (VWR Scientific) over-
98 · W. LEE et al. night at 4°C to immobilize antigen on the plate. The plates coated with antigen were then washed 3 times each with 300 ml of PBS, and blocked for 1 hour at room temperature with a blocking solution that was varied depending on the experiment. Blocking solutions used were 0.1% human serum albumin (HSA) in PBS, 0.1% bovine serum albumin (BSA) in PBS or 0.1% bovine collagen in PBS. After blocking, the plates were washed 3 times with PBS and 50 ml of pooled human cord serum (n=20), human serum, or anti-ssDNA antibodies were added. Next, plates were incubated at room temperature for 3 hours and then washed 3 times with PBS. After washing the plate, 50 ml per well of affinity-purified alkaline phosphatase-conjugated goat antibodies specific for human m-chain in the blocking solution was added and the plate was incubated for 1 hour at room temperature. The plates were then washed 3 times with PBS, and 100 ml of developing solution consisting of 1.0 mg/ml p-nitrophenyl phosphate in 100 mM diethanolamine, 0.5mM MgCl2, and 0.2% (w/v) NaN3, pH 9.5 was added to each well. Absorbance at 405nm was determined using an EL-340 Bio Kinetics Reader (Bio-Tek Instruments, Winooski, VT, USA). Extraction of human and swine lung antigens
Human lung samples were obtained from discarded portions of transplant donor lungs that were used at Duke University Medical Center. Samples of human donor lung were obtained in accordance with the guidelines of the Duke University Institutional Review Board and with the approval of the Board. Porcine lung samples were obtained from discarded portions of donor lungs used in experimental pigto-baboon pulmonary xenotransplantation. All lungs had been perfused with a preservation solution, resulting in lung tissue without blood. Antigen from human and swine lung was extracted in the following manner: The lung samples were ground to a powder at –195°C using a BioPulverizer (Biospec, Bartlesville, OK, USA). Next, 0.5 to 1.0 g of the powder was suspended in 600 ml of PBS and the resulting suspension was filtered through four layers of gauze (Nu Gauze; Johnson and Johnson, Arlington, TX, USA) to remove large fragments of tissue. The filtrate was then centrifuged at 85 × g for 3 minutes. The supernatant was discarded and, to remove any remaining soluble proteins such as albumin or IgG, the pellets were resuspended in 600 ml of PBS and then centrifuged at 16,000 × g for 2 minutes. The supernatant was again discarded and the pellets, containing insoluble lung antigen, were utilized in competitive binding studies. To obtain soluble antigen for direct binding studies, 200 ml of the lung antigen pellet (obtained as described above) were mixed with 500 ml of PBS containing 2% TWEEN-20. The mixture was vortexed for 10 minutes and then centrifuged at 16,000 × g for 3 minutes at room temperature. The supernatant, containing the extracted lung antigen, was diluted in PBS so that the absorbance at 280 nm was 0.03, and then 50 ml/well were added to a Nunc-Immuno Maxisorp polystyrene plate. The plate was then incubated overnight at room temperature, followed by three washes each with 300 ml of PBS per well. The plate was blocked with 0.1% HSA in PBS for 2 hours at room temperature and then washed three times with 300 ml of PBS per well. Pooled human cord serum samples (50 ml), diluted appropriately in PBS, were added to each well and incubated for 3 hours at room temperature. The plate was washed, and alkaline phosphatase-conjugated goat antibodies specific for human m-chain were added as described above. The plates were washed again, developer added, and the absorbance at 405 nm determined as described above.
Results Binding of human cord serum IgM and anti-ssDNA IgM to a crude preparation of human and pig lung antigens as measured by direct ELISA
The relative reactivity of human cord serum IgM and anti-ssDNA IgM to immobilized human and pig lung antigen extracts were measured by an ELISA as described in the Methods. Both human cord serum IgM and anti-ssDNA IgM showed higher degrees of
Specificity of polyreactive antibodies · 99
Figure 1. The binding of human polyreactive IgM to human and pig lung antigens. The binding of cord serum IgM (A) and anti-ssDNA IgM (B) to immobilized antigens was measured by ELISA as described in the Methods. Human serum albumin (0.1%) was used as the blocking solution for the ELISA. The concentration of IgM in the pooled cord sera was 60 mg/ml. The experiment was performed in duplicate and the standard error is shown.
binding to the pig lung antigens than to the human lung antigens (Fig. 1). The human cord serum IgM was about 3 times more reactive toward pig lung antigens whereas the anti-ssDNA IgM was about 2 times more reactive. Binding of human cord serum IgM to pig lung antigens plateaued to some extent at a serum concentration of 50%, suggesting that the antigen was saturated or approaching saturation with antibodies at that concentration. The concentration at which the binding of anti-ssDNA antibodies plateaued was not determined.
100 · W. LEE et al. Binding of human polyreactive IgM to a crude preparation of human and pig lung antigens as measured by a competitive binding assay
Although the direct binding studies described above suggest that polyreactive antibodies are more xenoreactive than alloreactive or autoreactive, there are some limitations of direct binding studies. For example, differences in binding may represent differences in the blocking of non-specific binding to the plate. For this reason, the binding of human polyreactive antibodies to pig and human lung antigens was assessed using competitive binding assays. Specifically, the binding to DNP-albumin of IgM in adult human serum was evaluated by ELISA in the presence of human or porcine lung antigens. As shown in Figure 2, the presence of porcine lung antigens resulted in more inhibition of binding to DNP-albumin than did the presence of human lung antigens in all eight samples tested. The slight increase in binding of anti-DNP IgM in the presence of human lung antigen was not apparently due to comtamination of the lung antigen preparation with human IgM, since no binding of IgM was observed in controls lacking human serum. The slight increase in IgM binding in the presence of human lung antigens may be due to weak interactions between the human lung antigens and human polyreactive IgM, which could result in crosslinking of the IgM and thus increased cooperative binding of the IgM to the immobilized DNP. However, this idea remains to be tested.
Figure 2. The binding of human IgM to immobilized DNP-albumin in the presence or absence of human or pig lung antigens. The binding of IgM from the sera of eight normal human donors was evaluated by ELISA. In all sera tested, binding was less in the presence of pig lung antigens compared to human lung antigens, suggesting that human anti-DNP IgM binds more strongly to pig lung antigens than to human lung antigens. Bovine collagen (0.1% in PBS) was used as the blocking reagent, and the sera were diluted to 5% in PBS for the assay.
Binding of human cord serum IgM and anti-ssDNA IgM to purified albumins from various species as measured by direct ELISA
To further evaluate the reactivity of polyreactive IgM with self versus foreign antigens, the binding of human cord serum IgM and anti-ssDNA IgM to albumins from human, pig, and cow was evaluated by ELISA as described in the Methods. As a positive control, bind-
Specificity of polyreactive antibodies · 101
Figure 3. The binding of human polyreactive IgM to DNP-HSA and to human, pig, and bovine serum albumins. The binding of cord serum IgM (A) and anti-ssDNA IgM (B) to immobilized antigens was measured by ELISA as described in the Methods. Human serum albumin (0.1%) was used as the blocking solution for the ELISA. The concentration of IgM in the pooled cord sera was 60 mg/ml. The experiment was performed in duplicate and the standard error is shown.
ing to dinitrophenyl coupled HSA (DNP-HSA; a common antigen recognized by polyreactive antibodies) was also evaluated. As shown in Figure 3, both human cord serum IgM and anti-ssDNA IgM reacted most strongly with HSA-DNP, less strongly with porcine serum albumin, and least with HSA and bovine serum albumin. Wells were blocked with human serum albumin, and human serum albumin was included in the enzymeconjugated antibody as a blocker of nonspecific binding. Thus, binding to human serum albumin may have been decreased by this procedure. As with binding to tissue extracts
102 · W. LEE et al. (Fig. 1), we observed evidence of saturation of antigen using cord serum but not antissDNA antibodies under the conditions used. It was not clear whether the binding of IgM from cord serum was greater to human or to bovine serum albumin (Fig. 3a), since the dose-dependent binding overlapped at about 6% serum. One possible explanation for this observation is that the “background” binding, which might be greater at lower serum concentrations (where binding of serum components that can block “non-specific” binding to the plate is minimized), is greater in the wells coated with human serum albumin than in wells coated with bovine serum albumin. This might have occurred since the bovine albumin prep is only 96% pure whereas the HSA prep is 99% pure. Alternatively, the complex shape of the binding curves might reflect interplay between the many factors associated with binding of the heterogeneous antibodies to the plate such as antibody concentration, antibody avidity, and antigen concentration.
Figure 4. The binding of human polyreactive IgM to albumins from various species. The binding of cord serum IgM (A) and anti-ssDNA IgM (B) to immobilized albumins was measured by ELISA as described in the Methods. Bovine serum albumin (0.1%) was used as the blocking solution for the ELISA. Binding that was significantly (p < 0.05) different from the binding to HSA is indicated by the (*). The concentration of IgM in the pooled cord sera was 60 mg/ml, and the sera were diluted 1 to 10 in PBS for the experiment. The concentration of anti-ssDNA IgM was 800 ng/ml. Each point represents the average absorbance obtained from 5 wells and the standard error is shown.
Specificity of polyreactive antibodies · 103
Since it was imperative to this study not to underestimate the binding of the polyreactive IgM to human serum albumin, and since reactivity with bovine serum albumin was low (Fig. 3), bovine serum albumin rather than human serum albumin was utilized to block “nonspecific” binding in subsequent experiments. While this might cause underestimation of the binding of antibodies to some foreign albumins, subsequent results validated this approach (see below). As the next approach to evaluating the binding of polyreactive IgM to autologous versus foreign albumin, the binding of human cord serum IgM and anti-ssDNA IgM to albumins from 12 species was evaluated by ELISA. Binding to DNP-HSA was again evaluated as a positive control. As shown in Figure 4, the binding of IgM to all foreign albumins, with the exception of hamster serum albumin, was significantly greater than binding to human serum albumin for at least one source of polyreactive IgM. Binding of polyreactive IgM to chicken serum albumin was greater than the binding to human serum albumin, regardless of the source of IgM, and the binding of human cord serum IgM to dog serum albumin was slightly but significantly lower than the binding to human serum albumin. Since bovine serum albumin was used as a blocking reagent (included in the enzyme linked antibody incubation as described in the methods), it is possible that this albumin interfered with the binding of the polyreactive antibodies to some of the epitopes on the various albumins. However, even with this limitation, the results are consistent with the idea that polyreactive IgM may be more reactive with foreign antigens than with self antigens. Further, as described below, these studies were confirmed using solution phase, competitive binding studies. Binding of human polyreactive IgM to purified thyroglobulins as measured by direct ELISA
As another approach to evaluating the binding of polyreactive IgM to autologous versus foreign antigens, the binding of human cord serum IgM and anti-ssDNA IgM to porcine, bovine, and human thyroglobulin was evaluated by ELISA. Binding of polyreactive IgM to the foreign thyroglobulins was greater than binding to the human protein, regardless of the source of IgM (Fig. 5). It is known that normal adult human serum contains polyreactive antibodies that react with human thyroglobulin (16) and monoreactive antibodies that react with the carbohydrate portions of bovine and porcine thyroglobulin (1, 2). However, the monoreactive IgM that bind to the carbohydrates expressed on the bovine and porcine proteins are absent in cord sera (17, 18) and are not expected to be present in the anti-ssDNA preparation. Although speculative, it is possible that polyreactive antibodies react with the carbohydrate portions of the foreign thyroglobulins. Regardless of the epitopes recognized, these findings lend further support to the idea that polyreactive IgM are more reactive with foreign antigens than with self antigens. Binding of human polyreactive IgM to human and xenogeneic antigens as measured by a competitive binding assay
The ability of xenogeneic and allogeneic albumins in solution to inhibit binding of polyreactive antibodies to an immobilized antigen (bovine thyroglobulin) was evaluated. Bovine collagen (0.1% in PBS), which blocks plates to a similar extent as does human
104 · W. LEE et al.
Figure 5. The binding of polyreactive IgM to human, pig, and bovine thyroglobulins. The binding of cord serum IgM (A) and anti-ssDNA IgM (B) to immobilized thyroglobulins was measured by ELISA as described in the Methods. Both the human cord serum IgM and anti-ssDNA IgM showed significantly (p < 0.05) greater binding to bovine and pig thyroglobulin than to human thyroglobulin. Bovine serum albumin (0.1%) was used as the blocking solution for the ELISA. The concentration of IgM in the pooled cord sera was 60 mg/ml, and the sera were diluted 1 to 10 in PBS for the experiment. The concentration of anti-ssDNA IgM was 800 ng/ml. Each point represents the average absorbance obtained from 5 wells and the standard error is shown.
serum albumin, was used as a blocking reagent. As shown in Figure 6, chicken, guinea pig, hamster, pig and possibly horse albumins, but not dog, goat, rabbit, sheep, turkey or human albumin, inhibited binding of polyreactive IgM from human cord serum to bovine thyroglobulin. This binding pattern is identical to the pattern observed in the solid phase binding assay (Fig. 4a), suggesting that the difference in binding of polyreactive antibodies to different albumins observed using the solid phase assay is not due to putative artifacts. Interestingly, under the conditions used (5 mg/ml of albumin), no inhibition of binding of anti-ssDNA antibodies to bovine thyroglobulin was obtained using any albumin, human or xenogeneic. It is unknown whether this finding might be due to
Specificity of polyreactive antibodies · 105
Figure 6. The binding of human IgM to immobilized DNP-albumin in the presence of albumins (5 mg/ml) from various species. The binding of IgM in pooled cord serum to DNP-HSA was measured by ELISA as described in the Methods. Binding that was significantly (p < 0.05) different from the binding in the presence of HSA is indicated by the (*). The concentration of IgM in the pooled cord sera was 90 mg/ml, and the sera were diluted 1 to 10 in PBS for the experiment. Bovine collagen (0.1% in PBS) was used as the blocking reagent.
blocking of much of the binding of anti-ssDNA antibodies by bovine collagen (the blocking reagent used in the experiment). Discussion In this study, we have found that human polyreactive IgM reacted with a porcine tissue extract more so than with a human tissue extract. Similarly, human polyreactive IgM reacted either the same or more strongly with foreign (xenogeneic) serum albumins compared to human serum albumin. Further, human polyreactive IgM reacted more strongly with bovine and with porcine thyroglobulin than with human thyroglobulin. All of these observations support the idea that polyreactive IgM, both in adults and in humans, is directed against foreign antigens more so than against autoantigens. There are several possible explanations for this increased reactivity; first, foreign proteins might have more epitopes per protein than do self-proteins. Second, polyreactive antibodies might bind with higher affinity/avidity to foreign proteins than to self-proteins. Third, there may be higher concentrations of polyreactive antibodies that recognize foreign antigens than of antibodies that recognize self-antigens. Additional work using panels of monoclonal polyreactive antibodies may be needed to sort out which one(s) of these potential factors play a role in the increased reactivity of polyreactive antibodies with foreign antigens compared to self antigens.
106 · W. LEE et al. The observations described above were made using two sources of polyreactive IgM, cord serum IgM and affinity purified adult anti-ssDNA IgM. Although the tendency to react more with foreign antigens than with self-antigens was observed in both preparations, some differences in the two preparations were noted. For example, cord serum IgM reacted more strongly with horse albumin than with mouse or sheep albumin, whereas anti-ssDNA antibodies reacted more strongly with mouse or sheep albumin than with horse albumin. This might suggest that the binding of polyreactive anti-ssDNA antibodies against antigens unrelated to ssDNA is different than binding of the presumably average polyreactive antibodies present in cord serum. On the other hand, these differences may reflect differences in the relative purity of the two preparations. It is noteworthy that the similarity of various albumins to human albumin did not correlate with the binding of polyreactive antibodies against the albumin. For example, polyreactive IgM in the cord serum preparation bound to sheep serum albumin (75% identity to HSA) to a lesser extent than to pig serum albumin (76% identity to HSA). This lack of correlation was verified by a regression analysis of sequence identity to human serum albumin versus binding of polyreactive antibodies. This observation might suggest that reactivity of polyreactive antibodies with foreign albumins is dictated by a limited number of immunodominant epitopes. Acknowledgements
The authors thank KIM S. BARBER for invaluable assistance in preparing the manuscript. This work was supported by NIH grants HL50985 and HL52297 and by the FANNIE E. RIPPEL foundation.
References 1. GUILBERT, B., G. DIGHIERO, and S. AVRAMEAS. 1982. Naturally occurring antibodies against nine common antigens in human sera. J. Immunol. 128: 2779. 2. AILUS, K., and T. PALOSUO. 1995. IgM class autoantibodies in human cord serum. J. Reprod. Immunol. 29: 61. 3. LACROIX-DESMAZES, S., S. V. KAVERI, L. MOUTHON, A. AYOUBA, E. MALANCHERE, A. COUTINHO, and M. D. KAZATCHKINE. 1998. Self-reactive antibodies (natural autoantibodies) in healthy individuals. J. Immunol. Meth. 216: 117. 4. LOGTENBERG, T. 1990. Properties of Ployreactive Natural Antibodies to Self and Foreign Antigens. J. Clin. Immunol. 10: 137. 5. SIGOUNAS, G., N. KOLAITIS, E. MONELL-TORRENS, and A. L. NOTKINS. 1994. Polyreactive IgM antibodies in the circulation are masked by antigen binding. J. Clin. Immunol. 14: 375. 6. BOUVET, J. P., and G. DIGHIERO. 1998. From natural polyreactive autoantibodies to a la carte monoreactive antibodies to infectious agents: is it a small world after all? Infection & Immunity 66: 1. 7. STAHL, D., S. LACROIX-DESMAZES, L. MOUTHON, S. V. KAVERI, and M. D. KAZATCHKINE. 2000. Analysis of human self-reactive antibody repertoires by quantitative immunoblotting. J. Immunol. Meth. 240: 1. 8. LYDYARD, P. M., R. QUARTEY-PAPAFIO, B. BROKER, L. MACKENZIE, J. JOUQUAN, M. A. BLASCHEK, J. STEELE, M. PETROU, P. COLLINS, D. ISENBERG, and et al. 1990. The antibody repertoire of early human B cells. I. High frequency of autoreactivity and polyreactivity. Scand. J. Immunol. 31: 33. 9. MOUTHON, L., S. LACROIX-DESMAZES, A. NOBREGA, C. BARREAU, A. COUTINHO, and M. D. KAZATCHKINE. 1996. The self-reactive antibody repertoire of normal human serum IgM is acquired in early childhood and remains conserved throughout life. Scand. J. Immunol. 44: 243.
Specificity of polyreactive antibodies · 107 10. ANDERSSON, U., A. G. BIRD, B. S. BRITTON, and R. PALACIOS. 1981. Humoral and cellular immunity in humans studied at the cell level from birth to two years of age. Immunol. Rev. 57: 1. 11. NOTKINS, A. 2000. Polyreactive antibodies and polyreactive antigen-binding B (PAB) Cells. Curr. Top. Microbiol. Immunol. 252: 241. 12. PARKER, W., D. BRUNO, Z. E. HOLZKNECHT, and J. L. PLATT. 1994. Characterization and affinity isolation of xenoreactive human natural antibodies. J. Immunol. 153: 3791. 13. MACCHIARINI, P., R. ORIOL, A. AZIMZADEH, V. DE MONTPREVILLE, R. RIEBEN, N. BOVIN, M. MAZMANIAN, and P. DARTEVELLE. 1998. Evidence of human non-alpha-galactosyl antibodies involved in the hyperacute rejection of pig lungs and their removal by pig organ perfusion. J. Thor. and Card. Surg. 116: 831. 14. HERRICK, G., and B. ALBERTS. 1976. Purification and physical characterization of nucleic acid helix-unwinding proteins from calf thymus. J. Biol. Chem. 251: 2124. 15. KANAI, Y., and T. KUBOTA. 1989. A novel trait of naturally occurring anti-DNA antibodies: dissociation from immune complexes in neutral 0.3–0.5 M NaCl. Immunol. Lett. 22: 293. 16. THALL, A., and U. GALILI. 1990. Distribution of Gal alpha 1–3Gal beta 1–4GlcNAc residues on secreted mammalian glycoproteins (thyroglobulin, fibrinogen, and immunoglobulin G) as measured by a sensitive solid-phase radioimmunoassay. Biochem. 29: 3959. 17. PARKER, W., K. LUNDBERG-SWANSON, Z. E. HOLZKNECHT, J. LATEEF, S. A. WASHBURN, S. J. BRAEDEHOEFT, and J. L. PLATT. 1996. Isohemagglutinins and xenoreactive antibodies: members of a distinct family of natural antibodies. Hum. Immunol. 45: 94. 18. MINANOV, O. P., S. ITESCU, F. A. NEETHLING, A. S. MORGENTHAU, P. KWIATKOWSKI, D. K. COOPER, and R. E. MICHLER. 1997. Anti-Gal IgG antibodies in sera of newborn humans and baboons and its significance in pig xenotransplantation. Transplant. 63: 182. WILLIAM PARKER, Ph. D., Duke University Medical Center, Box 2605, Department of Surgery, Durham, NC 27710, Telephone: 919-681-3886, Fax: 919-681-7263, E-mail: [email protected]