Tel: +317-782-7193, Fax: +317-782-6949
Autoantibody potential of cancer therapeutic monoclonal antibodies
Article first published online: 10 NOV 2009
Copyright © 2009 UICC
International Journal of Cancer
Volume 127, Issue 2, pages 491–496, 15 July 2010
How to Cite
McIntyre, J. A. and Faulk, a. W. P. (2010), Autoantibody potential of cancer therapeutic monoclonal antibodies. Int. J. Cancer, 127: 491–496. doi: 10.1002/ijc.25038
- Issue published online: 20 MAY 2010
- Article first published online: 10 NOV 2009
- Accepted manuscript online: 10 NOV 2009 12:00AM EST
- Manuscript Accepted: 3 NOV 2009
- Manuscript Received: 14 OCT 2009
- unmasked autoantibody
We and others have reported that multiple autoantibodies are unmasked in human polyclonal antibody preparations after exposure to physiological oxidizing agents (hemin) or electromotive force. We now have asked if oxidation unmasks autoantibody reactivities in monoclonal antibodies (mAb). To do this, we have studied 9 FDA approved mAb used therapeutically, including 4 chimeric, 4 humanized and 1 chemically modified chimeric Fab that were exposed to the physiological oxidizing agent hemin at 36°C for 20 hr. These mAb were studied for autoantibody activity to phospholipids and DNA before and after oxidation with hemin and found to develop autoantibody activities after oxidation, while retaining their original specificity as measured by mAb anti-glycophorin A binding of erythrocytes, CD 19 binding to B lymphocytes and anti-HLA-A29 binding to A29-positive lymphocytes. The finding that certain mAb have the potential to unmask autoantibody activities as a consequence of exposure to physiological redox reactions in vitro gives pause to our present understanding of the immunological basis of tolerance and concern for potential autoimmune side effects in patients receiving mAb for diagnosis or treatment.
Genetic engineering is a growth strategy in the design and development of monoclonal antibodies (mAb) for clinical therapeutics, such as drug targeting in cancer chemotherapy.1, 2 However, the clinical use of mAb has been associated with serious side effects,3 presumably due to collateral damage from reactions with antigens of unintended targets.4 Inasmuch as a tenant of the scientific basis for the clinical use of mAb is that they react only with one or a small number of closely related epitopes,5 we asked if their specificities might be altered by exposure to physiological concentrations of physiological oxidants known to participate in physiological redox reactions. We posed this query due to our earlier reports of the unmasking of autoantibody activities in human blood as well as in purified IgG that had been exposed to such physiological redox reactions.6 One might then ask why such redox-reactive antibodies exist. We have proposed that redox-reactive antibodies represent elements of the body's immune defense and surveillance network.7 Recently, a similar explanation has been put forward by others.8 In the following paragraphs, we present data indicating that autoimmune activities can be unmasked in solutions of FDA approved mAb of several different specificities and immunochemistries.
Material and Methods
The experimental design and techniques used in this study are the same as those used in our earlier report of the unmasking of autoantibody activity in human IgG following oxidation with physiological oxidants such as hemin6 or by using electromotive force (EMF).9 It was shown that hemin was responsible for oxidation of the antibody because (i) none of the antibody changes occurred in the absence of hemin, (ii) none of the antibody changes occurred in the presence of hemopexin, which is a biological inhibitor of hemin, and (iii) ascorbic acid, which is an antioxidant, inhibited the hemin-mediated oxidative changes in a dose–response manner.6 Briefly, for EMF oxidation a 250 μl aliquot of mAb (250 μg/ml) in TRIS was exposed to 8 V for 10 sec. Hemin (Sigma, St. Louis, MO) was dissolved in 1.0 N NaOH (15.15 mg/ml). Each mAb (250 μg/ml) was diluted in TRIS buffer pH 7.2 and exposed to 5 μl of the hemin solution (125 μM) overnight at 36°C on a rocking platform. There were no pH changes detected in the TRIS buffer after hemin addition or after addition of an equal volume (5 μl) of 1.0 N NaOH. Both the EMF and hemin oxidized samples were diluted 1/10 before addition of 50 μl to triplicate enzyme-linked immunosorbent assay (ELISA) wells. Autoimmune antibodies in this study are defined as antiphospholipid (aPL) antibodies to cardiolipin, phosphatidylcholine, phosphatidylserine and phosphatidylethanolamine, as previously described.10 Autoantibodies were qualitatively and quantitatively identified by using an ELISA.11 All ELISA were done with TRIS buffers supplemented with 1% BSA (bovine serum albumin), or 10% ABP (adult bovine plasma) respectively, but are reported for ABP unless otherwise stated. All mAb used in this study are commercially available pharmaceutical products, the mAb to glycophorin A, notwithstanding. They were studied for the presence of autoantibodies before and after oxidation. For anti-nuclear antibodies (ANA) determinations the Immunoconcepts RELISA® ANA Test kit (Sacramento, CA) was performed according to the manufacturer's instructions.
Results and Discussion
Immunological studies of 9 commonly used therapeutic mAb that are very different in specificity and biochemistry were found to manifest autoantibody activity after incubation with hemin (Table 1). Hemin is a normal degradation product of hemoglobin,12 and thus is found at sites where erythrocytes are destroyed.13 Very little data are available on plasma concentrations of hemin, but it has been reported as 50–280 μM in patients with β-Thalassemia.14 We have determined from dose–response experiments that as little as a 28 μM hemin can yield significant unmasking of aPL from the mAb, Remicade (data not shown). In addition, evidence that the in vitro experimental conditions employed actually induced oxidation of Ig have been published,6 in which it was shown that these conditions resulted in the oxidation of antibody tyrosine to nitrotyrosine by hemin oxidation.
Only 1 (i.e., Rituxan) of the 9 mAb demonstrated aPL activity before exposure to hemin, and treatment with Rituxan has been found to have several dangerous side effects, including neutropenia and lung toxicity, both of which are associated with the presence of aPL.15–18 These results indicate that (i) aPL are identified infrequently in mAb not subjected to hemin-mediated redox reactions, and (ii) aPL have been identified in all studied mAb after being subjected to hemin-mediated redox reactions. Since it is generally accepted that aPL are associated with and/or are causally related to a variety of seemingly unrelated diseases,19 the possibility emerges that such antibodies may not be unrelated to some of the complications or side effects reported in patients given mAb for diagnostic or therapeutic purposes. This of course assumes that oxidative unmasking of mAb autoantibody activity occurs in vivo as it does in vitro, which is as yet unproven.
The reality of unmasking autoantibody activity in mAb suggests the possibility that such redox reactions also might mask or interfere with the expected reactivity of mAb, but that was not found to occur. For example, we were able to show by quantitative flow cytometry that mAb to glycophorin A, before and after oxidation demonstrated only a trivial change in mean channel shift (mcs) from 429 to 435, respectively. In addition, we found no reactivity of mAb anti-glycophorin A with any of the 4 phospholipid antigens, but significant reactivity was identified against these phospholipid antigens following oxidation by EMF, as shown in Figure 1. Also shown in Figure 1, serial absorptions with RBCs produced incremental decreases in reactivity (435–276 mcs). We have also documented by flow cytometry another example of oxidation not affecting the mAb cognate antigen by using a CD 19 mAb before and after EMF exposure as assessed on the B lymphocytes from a normal individual. Before EMF, the mAb to CD 19 showed a mcs of 82.73 and after EMF the mcs of 82.86, basically unchanged. Negative for aPL before EMF, the oxidized CD 19 sample also showed robust aPL reactivities (data not shown). In addition, hemin treatment of anti-HLA sera did not alter the HLA reactivity, and a human IgG HLA-A29 mAb was not affected by either hemin or EMF when retested on A29-positive cells.20
In an effort to diminish or circumvent serious clinical complications and undesirable side effects that result from the clinical use of mAb in patients with organ transplants, cancer or autoimmune diseases, Herculean efforts have been directed to the engineering of mAb via chimerization and humanization, resulting in FDA approval of several new therapeutic reagents.21 We have presented data in Table 1 documenting that FDA approved therapeutic mAb, regardless of their chimerization or humanization state, reveal autoantibody activities following incubation with physiological concentrations of hemin. To extend these observations, we present the autoantibody profiles for 2 commonly used therapeutic mAb. Figures 2a and 2b show the qualitative and quantitative profiles for Avastin, a humanized mAb, and ReoPro, a chimeric mAb Fab (antibody binding fragment) that is engineered to represent a light chain that contains a human constant and a mouse variable region. Both of these mAb have been reported to be associated with serious side effects22, 23 that are very similar if not identical with the vascular complications reported in patients with aPL and the aPL antibody syndrome.24 Of interest as shown in Figure 2, both of these therapeutic mAb manifest significant autoantibody activity following hemin-mediated redox reactions.
We earlier presented data showing that human polyclonal IgG subjected to hemin-mediated redox reactions unmasked the presence of many different autoantibodies in addition to aPL.6, 7 To extend this study beyond aPL, we also tested 4 of the 9 mAb in Table 1 for ANA by using a commonly employed commercial diagnostic kit. This immunoassay contains nuclear antigens such as dsDNA, histones, SSA/Ro, SSB/La, Sm, Sm/RNP, Scl-70 and Jo-1. The results revealed that all 4 mAb were ANA-negative before hemin-mediated oxidation and all 4 were ANA-positive following oxidation (Fig. 3). This finding with hemin-treated mAb is the same as our earlier findings with hemin-treated human polyclonal intravenous Ig that was studied and found to be strongly positive for ANA6 by ImmunoConcepts (Sacramento, CA), and the present experiments with mAb were performed by using their ANA testing kit. Indeed, the quantitative expression of anti-ANA activity for each of the 4 was equal to or greater than the value for the positive control included in the kit. In support of these data is a recent report showing that 11 of 15 patients with arthropathies that were treated with infliximab developed strong positive ANA titers subsequent to receiving Remicade.25
Many diseases have been shown to be associated with the presence of autoantibodies such as aPL.26 Our results support a concept that such autoantibody activity can be unmasked in vitro from commercially available and/or FDA approved mAb by physiological concentrations of physiologically available oxidants. If such reactions occur in vivo, they might be associated with the complications and side effects that result from the clinical use of mAb therapy. Another consideration of these data is that much of what we know in contemporary immunology has derived from studies that use mAb; it thus is quite challenging to find that the mono reactivity of such antibodies is subject to the redox properties of their microenvironment.27 It has long been observed that patients with autoimmune diseases are likely to have autoantibodies to several different tissues, which is referred to as, “serological overlap.”28 We suggest that this overlap could result from disease-mediated redox reactions that unmask autoantibodies when exposed above a threshold level of redox activity, whether they are polyclonal or monoclonal in nature. However, such polyreactivity does not appear to be due to an increased stickiness, for redox-produced autoantibodies have not been found to react with alloantigens. For example, hemin-treated IgG demonstrated many autoantibody activities6, 7 but did not demonstrate any anti-HLA activity.20
Regarding the redox conversion of polyclonal IgG to autoantibodies, we have found6 that this conversion is inhibited by antioxidants such as vitamin C, and propose that supplemental antioxidants might also be effective adjunct treatment for patients who are receiving therapeutic mAb. Nonetheless, it is unsettling to consider that autoantibodies traditionally have been viewed as a loss or aberration of self-tolerance.29, 30 Thus, finding that we all have the potential to produce autoantibodies as a consequence of physiological redox reactions gives pause to reconsider the immunological basis of tolerance, which is discussed by Prof. Davies30.
Although this is the first report of redox-induced autoantibody activity from FDA approved therapeutic monoclonal antibodies, other investigators have used various procedures to reveal polyreactivity of monoclonal antibodies, and some of these reports have contained data on mechanisms that might account for such promiscuous activity. Chou et al.31 have taken the interesting position that both polyclonal and monoclonal natural antibodies, which are both low titer and low affinity, have previously unrecognized specificities for oxidation-specific antigenic epitopes and can serve regulatory functions for certain enzymes and in hemostasis, reviewed by Lutz et al.32 The redox autoantibodies we described are neither natural nor are they of low titer or of low affinity. In contrast, Bobrovnik33, 34 reported that natural antibodies react with hydrophilic epitopes while polyreactive antibodies react with hydrophobic epitopes, and that monospecific antibodies can convert to polyreactive antibodies following peroxide degradation or by lipolysis of noncovalently bound lipids that maintain antibody specificity. However, the natural autoantibodies reported by Chou et al.31 and Lutz et al.32 require no lipid-altering manipulations to react with their oxidation-specific epitopes, suggesting that lipids may not be central to the unmasking of polyreactive autoantibodies.
Another approach has been taken by Dimitrov et al.35 who performed kinetic and thermodynamic analysis using a surface plasma resonance-based technique to characterize the interaction of a mouse rheumatoid factor monoclonal antibody before and after 6 M urea-induction of polyreactive antigen-binding activities. Their results showed a change in the type of noncovalent forces involved in antigen binding and an increase in the structural flexibility of the antibody paratope. Kang and Warren36 have confirmed and extended interpretation of the thermodynamic findings.
Dimitrov et al.37 have reviewed their and other studies and conclude that nondenaturing concentrations of certain protein destabilizing chemical agents, in which they include redox reactions, induce polyreactive autoantibody reactivities in monoclonal and polyclonal antibodies. They further state that the appearance of new antigen-binding specificities by a single antibody can be achieved by either an induced increase in the plasticity of the antigen-binding sites or by the use of polyreactive, low molecular weight compounds as antigen-binding cofactors. These are useful advances in building an understanding of mechanisms involved in the creation of polyreactivity, particularly urea-induced polyreactivity, but may not apply to redox-induced promiscuity. Recent reviews of autoantibodies and polyreactivity can be found in Refs.32 and38.
As far as we are aware, there are no kinetic and thermodynamic data on redox-induced polyreactive autoantibodies that support the proposed mechanisms. Also, as shown in Figure 1 above, the use of EMF-induced redox-equivalent autoantibodies9 does not utilize the antigen-binding cofactor mechanism, casting doubt on the role of a cofactor in redox-induced unmasking of autoantibody activities. In addition, redox-induced autoantibody activity in a chimeric Fab fragment (ReoPro in Table 1) indicates that none of the biological functions characteristic of the Fc fragment, such as complement fixation, are involved.
In light of the above discussion of mechanisms, we tend to favor our original proposal6 that redox-induced nitrosylation of tyrosine residues in and around the antibody hypervariable region is a likely mechanism for the masking and unmasking of autoantibody activity, for a change in nitration could produce conformational changes in the antibody binding site that masks and/or unmasks certain autoantibody activities. This is supported by the unique distribution of amino acids within the complementarity determining regions, inasmuch as the antigen contact sites contain an extraordinarily high content (∼23%) of tyrosines39 and it has been shown that nitration of tyrosine within the antibody binding site can produce an on–off switching of antibody-haptene binding similar to our masking/unmasking reaction.40 Furthermore, we have shown, by using an anti-nitrotyrosine mAb in the ELISA that a hemin-induced autoantibody is positive for nitrosylation.6 In addition, the tyrosine phosphorylation reaction is reversible,41, 42 accounting for both masking and unmasking.
In closing, it might be recalled that at least 50% of early immature B cells in bone marrow express auto reactive B-cell receptors.43 These auto reactive B cells are selected against in bone marrow and peripheral blood by deletion, anergy induction and, perhaps most importantly, by receptor editing,44 resulting in a B cell purged population where the majority of healthy people exhibit only low concentrations of auto reactive antibodies in their blood.45 Viewed in this light, it is difficult to imagine the molecular biological basis for the routine presence of numerous high titered redox-induced autoantibodies in normal blood, IVIg and even in mAb. A possible explanation of this might be that auto reactive B-cell receptors could be similarly masked to disallow recognition at auto reactive B-cell checkpoints.46 Precedence for such a mechanism derives from reports that some autoimmune diseases are associated with alterations in B-cell tolerance checkpoints.47
- 11The evolution, evaluation and interpretation of antiphospholipid antibody assays. Clin Immunol Newsl 1995; 15: 28–38., , .
- 19BoffaM-C, PietteJ-C, eds. 9th International Symposium on antiphospholipid antibodies. J Autoimmun 2000; 15: 81–271.
- 38McIntyreJA, FaulkWP, eds. Redox and autoimmunity. Autoimmun Rev 2008; 7: 515–84.