Research Article: Covalent Binding Antibodies Suppress Advanced Glycation: On the Innate Tier of Adaptive Immunity

Date Published: July , 2009

Publisher: A.I. Gordeyev

Author(s): T. Shcheglova, S. P. Makker, A. Tramontano.



Non-enzymatic protein glycation is a source of metabolic stress that contributes to cytotoxicity and tissue damage. Hyperglycemia has been linked to elevation of advanced glycation endproducts, which mediate much of the vascular pathology leading to diabetic complications. Enhanced glycation of immunoglobulins and their accelerated vascular clearance is proposed as a natural mechanism to intercept alternative advanced glycation endproducts, thereby mitigating microvascular disease. We reported that antibodies against the glycoprotein KLH have elevated reactivity for glycopeptides from diabetic serum. These reactions are mediated by covalent binding between antibody light chains and carbonyl groups of glycated peptides. Diabetic animals that were immunized to induce reactive antibodies had attenuated diabetic nephropathy, which correlated with reduced levels of circulating and kidney-bound glycation products. Molecular analysis of antibody glycation revealed the preferential modification of light chains bearing germline-encoded lambda V regions. We previously noted that antibody fragments carrying V regions in the germline configuration are selected from a human Fv library by covalent binding to a reactive organophosphorus ester. These Fv fragments were specifically modified at light chain V region residues, which map to the combining site at the interface between light and heavy chains. These findings suggest that covalent binding is an innate property of antibodies, which may be encoded in the genome for specific physiological purposes. This hypothesis is discussed in context with current knowledge of the natural antibodies that recognize altered self molecules and the catalytic autoantibodies found in autoimmune disease.

Partial Text

The generation of an enormous diversity of antibodies in response to the multitude of possible antigens is a signature of instructive or adaptive immunity. The structural basis for adaptive immunity is expressed in the variability of the antigen binding sites displayed on antibodies and B cell receptors. Thus, antibodies are conventionally associated with the genetic recombination and accumulated mutations in their variable (V) regions that incrementally improve the complementarity between the antibody combining site and groups on the antigen. In contrast to affinity that matures gradually over time through multiple weak interactions, binding through strong forces such as a covalent bond could enable a more rapid and efficient way to capture certain antigens. Is there any case where antibodies use this form of binding and what purpose could such a binding mechanism serve?

In an alternative framework one could postulate that covalent binding antibodies might also occur naturally if this activity were advantageous to the host. We proposed that binding through a single strong interaction to an antibody would be an appropriate mechanism for the sequestration and clearance of chemically damaged proteins and cells. Such a function is increasingly recognized in studies of naturally occurring antibodies that have inherent affinity for altered structures on self (4). For example, certain IgM antibodies that compete with macrophage receptors for binding of oxidized LDL particles rely on the recognition of distinct chemical moieties such as the phosphorylcholine headgroup on oxidized phospholipids. These natural autoantibodies (nAbs) are encoded in the germline and typically lack somatic mutations (5). Armed with this “innate-like” reactivity, nAbs are believed to constitute a disposal system for continuous surveillance and elimination of altered self, or “neoantigens” shed from apoptotic cells and damaged tissues (6). The same nAbs also bind to phosphorylcholine groups on bacterial cell wall polysaccharides, thus providing a first line of defense against infections (7). This dual purpose could explain the conservation of this function in the germline repertoire. The molecular basis for the interaction of V regions of nAbs with oxidized phospholipids remains under investigation.

In order to test this hypothesis we sought to demonstrate that covalent binding antibodies could be elevated in an immune response and that these could attenuate a glycation stress by reducing the carbonyl load. Reactive immunization suggested a feasible approach. Previously, we showed that antibodies induced to a pyruvate-containing hapten-KLH conjugate could bind to antigen by recognizing only the carbonyl group in the hapten (27). However, to our surprise the anti-KLH antibodies accounted in large part for this reactivity. Antibodies to KLH bound to the reactive pyruvate in the same way as established by differential binding to the pyruvate/glycolate hapten pair (Figure 3). Covalent Schiff base formation between antibody and the pyruvate carbonyl was the most plausible explanation for this focused binding. To test whether these antibodies could also neutralize carbonyl groups on glycation products, we compared anti-KLH antibody and normal IgG in the reaction with glycated peptides from sera of diabetic rats. In fact, this assay showed that L chains of anti-KLH antibody were more reactive than L chains of normal IgG (28). The chemically reduced glycated peptides failed to form covalent adducts, indicating that carbonyl groups were necessary for the reaction (Figure 4). Proteomic analysis of the modified L chains by tandem mass spectrometry showed V region peptides derived from only two lambda L chains, even though kappa L chains comprise more than 90% of rat IgG. Remarkably, these lambda L chain peptides revealed sequences identical to the germline-encoded VL (28). However, these results also suggested that L chain reactivity was enhanced by immunization. Although mass analysis detected only peptides of unmutated germline VL, modified peptides with mutated residues might not have been identified by proteomic analysis. Alternatively, L chains with innate reactivity might be recruited in combination with somatically diversified VH domains, which provide specificity for KLH and cross-reactivity for glycated peptides.

The streptozotocin-induced diabetic rat, serving as a suitable model for diabetic nephropathy, was used to demonstrate the potential of the reactive immune response to mitigate AGE formation in vivo. Compared to diabetic animals immunized with adjuvant alone, KLH-immunized diabetic rats had significantly diminished AGE levels, which correlated with attenuated nephropathy (28). Improved renal function, as determined by glomerular morphology and proteinuria levels, was accompanied by reduced AGE staining in the renal extracellular matrix. We hypothesized that the therapeutic benefit derived from improved capture of glycation products in circulation by the reactive antibodies. This mechanism presumes that AGEs formed from antibody molecules are less cytotoxic than the alternative protein AGEs. Formation of glycated IgG in diabetic subjects and, in particular, the identification of glycated L chain in the diabetic serum (26) indicated that these molecules might play a role in AGE pathogenesis. Although modified L chains accumulated in the reaction with glycated peptides in vitro, these products were not significantly elevated in the serum of diabetic animals. Glycated IgG is cleared from circulation and taken up by kidney more efficiently than unmodified IgG (29). Furthermore, the filtration properties of low mass glycated proteins favor their selective excretion (30). Thus, native and modified L chains are found in the urine of both healthy and diabetic subjects (31, 32). Taken together, these observations are consistent with the hypothesis of enhanced clearance of selectively glycated antibodies.

In an alternative approach, we used synthetic reactive substrates as antigens to probe for antibodies capable of covalent binding. This idea followed the original concept that antibodies selected for nucleophilic reactivity could also express enzymatic activity through covalent catalysis. Irreversible covalent binding to a small organophosphorus (OP) ester was used to chemically select single-chain VH-VL fragments (scFv) from a phage display library (35). All of the selected chemically reactive scFv molecules were modified on the VL polypeptide and could be described by two canonical sequences. The more reactive clone A.17 used the DPL-5 germline VL product, which was phosphonylated at Tyr37 within the framework region FR-L2. By contrast, six other reactive clones used the DPL-3 germline VL, which reacted at Tyr32 in CDR-L1. These nucleophilic Tyr residues are conserved in the VL germline and three-dimensional models of the scFv suggest that either residue can be oriented toward the combining site. However, the models also indicate that the Tyr37 is buried at the interface between VL and VH domains in A.17, where it could be sterically inaccessible to ligand contact without large conformational motions (Figure 5). Since the library was constructed by shuffling of germline VL and VH gene segments, we did not expect natural pairs in the scFv (36). Nevertheless, the VH chains of reactive clones were represented primarily by highly homologous sequences belonging to the VH4 family. These results strongly suggested that the VH chain plays an important part in enhancing the chemical reaction at residues on the VL region. The scFv could also bind other structurally unrelated OP compounds indicating a lack of fine specificity for ligand structure (35). The A.17 scFv was also shown to have modest hydrolytic activity for peptide amides and simple carboxylic ester substrates. These Ig V regions could thus serve as primitive covalent catalysts. Most intriguing is the notion that the reactivity emerges from certain germline-encoded VL-VH pairs. Additional studies of these monoclonal Fv fragments will be of interest for understanding the origins and biochemical functions of chemically reactive Ig molecules.

IgM nAbs are predominantly produced by a population of long-lived, self-replenishing B-cells, including the B-1 and B-1a subsets. It is believed that this B cell repertoire is conserved in evolution for its contribution to host defense (5). The germline-encoded nAbs are best known for their capacity to bind conserved determinants on pathogens, referred to as pathogen-associated molecular patterns (PAMPs). More recently it was shown that nAbs also bind to altered epitopes on apoptotic cell and self proteins (37). These findings support the concept of a homeostatic function of nAbs for clearance of debris from cell death and protein decay (6, 38). Chemical structures generated by oxidation and glycoxidation of membrane phospholipids are thus the natural targets for nAb binding, presenting a case where carbonyl chemistry intersects with immune recognition. The phosphorylcholine (PC) headgroup of oxidized phospholipids and PC residues on the bacterial cell wall polysaccharides provide a common molecular determinant for the immune and homeostatic functions of nAbs (7). As a molecular receptor of PAMPs, the nAb V region can be regarded as akin to the evolutionarily conserved Toll-like receptors (TLR) of innate immunity (39). This emerging paradigm has far-reaching implications for the linkage between innate and adaptive immunity.

Polyreactivity of nAbs is defined as the promiscuous avidity of multivalent IgM molecules to disparate molecules, including intracellular proteins and nucleic acids (45). This binding capacity, which can also be manifested as autoreactivity to self, remains poorly understood in structural terms. Enhanced avidity by the recognition of repetitive structures on self molecules or membranes provides a plausible molecular mechanism for polyreactivity. Within this conceptual framework, the repetitive structures suggest another form of molecular pattern identifying the damaged cells and tissues. Thus, the high density of PC groups on oxidized phospholipids and on cell wall polysaccharides offers a common feature for nAb binding. Similarly, phosphate linkages or nucleotide sequences on DNA strands could represent the molecular pattern for DNA antibodies to recognize surface features of apoptotic cells (46, 47).

Covalent-binding antibodies obtained by immunization or reactive selection and catalytic antibodies found in autoimmune pathologies appear to have features in common with nAbs. These features include VH and VL domains in germline or near-germline configurations and the recognition of molecules presenting uncommon chemical functionality, such as reactive carbonyls and phosphorus esters. Such features may be fundamentally related in that the chemical or enzymatic function is presumably highly evolved and therefore conserved in the genome. Whereas nAbs are IgMs that rely on weak multivalent binding to molecular patterns on antigens for high avidity, chemically reactive or catalytic antibodies are also expressed as IgG that can bind antigens with high affinity. However, the latter may also react with diverse substrates through weak, noncovalent pre-reaction complexes. According to the transition state theory of catalysis the strongest binding is expressed between an enzyme and the transition state. In the covalent complex, binding is dominated by the chemical bond between a residue of the active site and a group on the ligand. The focus of the weak binding by nAbs on small, minimally altered epitopes is consistent with the recognition of substrate functionality for reactivity. Thus, chemical reactivity could provide a mechanism to translate weak non-covalent binding to strong binding or to catalytic activity. The physiological purpose of chemically reactive and catalytic antibodies must also be addressed, even as the role of nAbs in protective immunity and homeostasis is only beginning to emerge. Participation of nAbs as response elements in oxidative stress and apoptotic cell clearance suggests a housekeeping function that predates the evolution of adaptive immunity. Similar considerations may apply to the rationale for covalent-binding antibodies acting as buffers against glycation stress. The preservation of covalent reactivity or catalytic functions in adaptive responses and in autoimmunity could be attributed to biological design or to defects in immune regulation. Inducibility through adaptive immunity might offer an appropriate mechanism to invoke beneficial responses to oxidative or glycoxidative stress. On the other hand, the role or contribution of natural catalytic autoantibodies in autoimmune pathology remains obscure, and their existence could simply reflect a failure in the V gene repertoire shift on induction of IgG autoantibodies (64, 65). Designed catalytic antibodies can also be educed from the adaptive immune response by affinity maturation that is appropriate for substrate reactivity or transition state binding. To what extent do designer catalytic antibodies co-opt the functions of natural autoantibody catalysts? An important clue was provided in an earlier report indicating a high frequency of antibodies with hydrolytic activity among hybridomas sampling an autoimmune repertoire (66). Continuing studies of antibody chemical reactivity induced by immunization or discovered in the germline repertoire should provide further insights into its role in immunity or pathology and could enable technological applications of this unconventional antibody function in the future.

This work was supported by the American Diabetes Association grant 1-05-RA-136, NIH grant CA90564, and the UC Davis, Medical School – Children’s Miracle Network.