Effect of H chain V region on complement activation by immobilized immune complexes.

Activation of C by immune complexes (IC) in tissues and the inflammatory consequences are major determinants in the pathogenesis of many autoimmune disorders. To assess the factors involved in C activation by such IC, we examined the binding of C components by chimeric IgG1 antibodies bound to immobilized Ag. We previously reported that alterations in the H chain V regions can affect the binding of first component of C (C1q) and a major breakdown product of the third C component (C3b) when otherwise identical antibodies were bound to immobilized (Tyr, Glu)-Ala-Lys. To evaluate C activation of these antibodies in well defined IC, we utilized a 9-amino acid peptide conjugated to BSA as Ag. The peptide:BSA conjugate was bound similarly by the two IgG1 antibodies which differed mainly in the CDR3 regions, but also in 9 other amino acids in the H chain V region. When soluble IC were prepared with the two antibodies, they activated C similarly. However, C activation by solid phase Ag:antibody complexes differed; we found that antibody 10B bound more C1q and C3b than antibody B11 did, unless the Ag was present at high density on the plates. These data suggest that the variable region differences affect C activation by these antibody when they are bound to immobilized Ag. Furthermore, these results underscore the differences in C activation by the same antibody depending upon whether the IC are free in solution or immobilized.     

A matched set of rat/mouse chimeric antibodies. Identification and biological properties of rat H chain constant regions mu, gamma 1, gamma 2a, gamma 2b, gamma 2c, epsilon, and alpha

Rat C regions mu, gamma 1, gamma 2a, gamma 2b, gamma 2c, epsilon, and alpha have been characterized by means of chimeric antibody technology. A set of rat/mouse Ag-specific (anti-4-hydroxy-3-nitrophenacetyl) antibodies was constructed that differ only in the H chain constant region but carry identical V region and L chain, both of which are of mouse origin. All rat constant regions could be expressed and m.w. were as expected from the protein sequence. A slight variation in mobility within the IgG subclasses allowed us to establish a hierarchy for the sizes of the four gamma H chains; gamma 2b greater than gamma 1 greater than gamma 2c greater than gamma 2a. Rat IgG2c and IgG2b could be purified on both protein A and protein G while rat IgG2a could only be purified on protein G. Rat IgM and IgG2b were the most potent in C-mediated hemolysis. This was not simply a consequence of the amount of C1q bound because IgG2c bound C1q efficiently but was relatively poor in cell lysis. In ADCC using human effector and target cells, IgG2b and IgG1 were the most effective.     

The binding of complement component C3 to antibody-antigen aggregates after activation of the alternative pathway in human serum.

Preformed immune aggregates, containing antigen and either IgG (immunoglobulin G) or F(ab')2 rabbit antibody, were incubated with normal human serum under conditions allowing activation of only the alternative pathway of complement. Both the IgG and F(ab')2 immune aggregates bound C3b, the activated form of the complement component C3, in a similar manner, 2-3% of the C3 available in the serum being bound to the aggregates as C3b, and the rest remaining in the fluid phase as inactive C3b or uncleaved C3. It was found that the C3b was probably covalently bound to the IgG in the aggregates, since C3b-IgG complexes could be demonstrated on sodium dodecyl sulphate/polyacrylamide-gel electrophoresis, after repeated washing with buffers containing high salt or boiling under denaturing conditions. Incubation of the C3b-antibody-antigen aggregates in buffers known to destroy ester linkages had little effect on the C3b-IgG complexes, which suggested that C3b and IgG might be linked by an amide bond. Two main types of C3b-IgG complexes were found that had apparent mol.wts. of 360000 and 580000, corresponding to either one to two C3b molecules respectively bound to one molecule of antibody. On reduction of the C3b-IgG complexes it was found that the beta-chain, but not the alpha'-chain, of C3b was released along with all the light chain of IgG but only about half or less of the heavy chain of IgG. These results indicate that, during activation of the alternative pathway of complement by immune aggregates containing IgG antibody, the alpha'-chain of C3b may become covalently bound at one or two sites in the Fd portion of the heavy chain of IgG.     

Activation of the first component of human complement, C1, by monoclonal antibodies directed against different domains of subcomponent C1q

Two monoclonal antibodies directed against C1q, and their (Fab)2 and Fab fragments, were used to study the mechanism of C1 activation. Monoclonal antibody 2A10, an IgG2a, was digested by pepsin to yield fully immunoreactive (Fab')2. Monoclonal antibody 1H11, an IgG1, was digested by papain to yield fully immunoreactive, bivalent (Fab)2. Previously 1H11 had been shown to bind to the C1q "heads," whereas 2A10 bound to stalks. Activation of C1 was followed by the cleavage of 125I-C1s in the presence of C1 inhibitor (C1-Inh) at 37 degrees C. Spontaneous activation was minimal at inhibitor concentrations above 0.4 micron (1.3 X physiologic inhibitor concentration); all results were corrected for the spontaneous activation background. Heat-aggregated IgG activated completely in this system and was taken as 100% activation. Monoclonal antibody 2A10 caused precipitation of C1 and slow activation; neither the (Fab')2 nor the Fab' derived from 2A10-caused activation. Probably, aggregates of intact 2A10 and C1 were serving as immune complexes to activate other molecules of C1. In contrast, both 1H11 and its (Fab)2 activated completely and stoichiometrically; that is, maximal activation was achieved at a ratio of one C1q head to one antibody combining site. The monovalent Fab derived from 1H11 bound well to C1q, but no activation of C1 was observed. Thus, bivalent binding of this head-binding monoclonal is required for C1 activation, but not the presence of the antibody Fc portion. Neither 1H11 nor its (Fab)2 fragments caused C1 precipitation; however, the 1H11 did form complexes composed of two C1q cross-linked by multiple 1H11, which were visualized by electron microscopy. The presence of these dimeric complexes correlated well with activation. A model for C1 activation is proposed in which two C1q subcomponents are held together by multiple (Fab)2 bridging C1q heads. The model is roughly analogous to touching opposing pairs of fingers and thumb tips, the two hands representing the two C1q, forming a cage. C1-Inh, which probably binds to C1r through the open end of the C1 cone, is too long asymmetric to be included within the cage. Thus, according to this model, the dimers of C1 are released from the inhibitory action of C1-Inh, and activation proceeds spontaneously and rapidly at 37 degrees C.     

Development and characterization of a novel fusion protein composed of a human IgG1 heavy chain constant region and a single-chain fragment variable antibody against Venezuelan equine encephalitis virus

Murine monoclonal antibody 1A4A1 has been shown to recognize a conserved neutralizing epitope of envelope glycoprotein E2 of Venezuelan equine encephalitis virus. It is a potential candidate for development of a second generation antibody for both immunodiagnosis and immunotherapy. In order to minimize the immunogenicity of murine antibodies and to confer human immune effector functions on murine antibodies, a recombinant gene fusion was constructed. It encoded a human IgG1 heavy chain constant region and a single-chain fragment variable antibody of 1A4A1. After expression in bacteria as inclusion bodies, the recombinant antibody was purified and refolded in vitro. The recombinant soluble antibody was demonstrated to retain high antigen-binding affinity to Venezuelan equine encephalitis virus and to possess some human IgG crystallizable fragment domain functions, such as recognition by protein G and human complement C1q binding. On non-reducing and reducing gel electrophoresis analysis of proteolytic fragments of the recombinant antibody, disulfide bond formation was found in the hinge region of the antibody. From these data, it was concluded that the recombinant antibody was capable of antigen recognition, and retained several functional activities. This work forms the basis for characterization of the recombinant antibody as to efficacy in vivo.     

Physicochemical characterization of dodecylphosphocholine/palmitoyllysophosphatidic acid/myelin basic protein complexes.

The stoichiometry of dodecylphosphocholine/palmitoyllysophosphatidic acid/myelin basic protein complexes and the location of the protein in the micelles have been investigated by electron paramagnetic resonance, ultracentrifugation, small-angle X-ray scattering, 31P, 13C, and 1H nuclear magnetic resonance spectroscopy, and electron microscopy. Ultracentrifugation measurements indicated that well-defined complexes are formed by association of one protein molecule with approximately 133 detergent molecules. The spin-labels 5-, 12-, and 16-doxylstearate have been incorporated into detergent/protein aggregates. Electron paramagnetic resonance spectral parameters and 13C and 1H nuclear magnetic resonance relaxation times showed that the addition of myelin basic protein does not affect the environment and location of the labels or the organization of the micelles. Previous results suggesting that the protein lies primarily near the surface of the micelles have been confirmed by comparing 13C spectra of the detergents with and without protein with spectra of detergent/protein aggregates containing the spin labels. Electron micrographs of the complexes taken by using the freeze-fracture technique revealed the presence of particles with an estimated radius about three times the radius of the micelles measured by small-angle X-ray scattering. The structural integrity of the complexes appears to be based on intramolecular protein interactions as well as protein-detergent interactions.     

Mixed IgA-IgG aggregates as a model of immune complexes in IgA nephropathy

Patients with IgA nephropathy have circulating immune complexes containing IgA, IgG, and C3. We have mixed human IgG and IgA1 and heated them to form mixed aggregate. On sucrose density gradients IgG aggregates were 11 to 19S whereas IgA aggregates were either 11S or greater than 19S. Mixed aggregates had both an 19 and 11 S peak. The isoelectric point of aggregates with only IgG was 7 to 9 and of only IgA 4.5 to 5.5. The isoelectric point of mixed aggregates decreased as the percent IgA increased. IgG aggregates mixed with normal human serum caused 30% C3 activation (20 min, 37 degrees C) whereas IgA aggregates causes no activation. There was a linear decrease in C3 activation as the percent IgA increased. Mixed aggregates that contained either radiolabeled IgG or IgA were mixed with normal human serum (1 h, 37 degrees C) and then solubilized, reduced, and separated by 10% SDS-PAGE. Heavy m.w. bands, consistent with covalent bonding of C3b and C3bi to Ig H chain were only seen in lanes with labeled IgG. This was confirmed by Western blot analysis. A human dimeric IgA1 myeloma protein with rheumatoid factor activity was also studied. It caused 15% alternative pathway C3 activation but did not fix C3 to its H chain. Binding of aggregates (+/- C3) to E was tested. Aggregates with IgG C3 bound but IgA (+/- C3) did not. Addition of greater than 10% IgA to an IgG-C3 aggregate inhibited E binding. We conclude that IgG in mixed aggregates is the site of C3 fixation. In contrast, IgA does not fix C3 but instead lowers the isoelectric point, increases the size and inhibits binding to E. These properties would inhibit clearance and promote mesangial deposition and local C activation.     

Alteration in H chain V region affects complement activation by chimeric antibodies

Chimeric antibodies to the synthetic polypeptide (Tyr, Glu)-Ala-Lys ((T,G)-A-L) were used to examine C activation by human IgG1. Two IgG1 antibodies, which contained mouse L chains and H chains with mouse V domains and human C domains, differed only in their VH domain. Ag binding and C activation by these antibodies were analyzed by ELISA. When limiting amounts of Ag were used in the assays, the antibodies required different quantities of Ag for optimal binding, suggesting that the antibodies bind to different epitopes on the (T,G)-A-L molecule. However, when competitive inhibition assays were performed with an optimal concentration of Ag, there were no differences in relative binding affinities for (T,G)-A-L or dissociation characteristics of the antibodies. C activation was examined at optimal Ag concentration to ensure equivalent binding of two IgG1 antibodies to Ag. After combination with immobilized Ag, these two antibodies bearing different V regions exhibited marked differences in the binding of C components C1q and C3d. When present in equal amounts in the assay, antibody 10B activated C and bound more C1q and C3d than antibody B11. These results indicate that V region differences can affect C activation by IgG.     

Generation of IgG aggregates by the myeloperoxidase-hydrogen peroxide system

Incubation of native human 125I-IgG with polymorphonuclear neutrophil (PMN) peroxidase-containing granules or with purified myeloperoxidase (MPO) in the presence of H2O2 and a suitable hydrogen donor such as catechol generated large amounts of heavy IgG aggregates. Short-term incubation (15 to 60 min) of native 125I-IgG (400 microgram) with MPO-containing granules or with purified MPO (1.5 microgram) in the presence of H2O2 (0.036 to 0.36 mumol) and catechol (0.2 mumol) resulted in the generation of 8 to 100 microgram of heavy IgG aggregates (3 X 10(5) to 4 X 10(6) daltons). Aggregate formation was completely abolished by the omission of H2O2 or catechol, and by the addition of catalase, sodium azide, or cyanide. IgG aggregates were also generated with tyrosinase, tyrosine, and atmospheric oxygen. These results indicate that aggregation was due to MPO-H2O2-mediated oxidation of catechol to orthoquinone, which was deemed to be directly responsible for cross-linking by non-enzymic biochemical reactions. The IgG aggregates generated were shown to behave as typical immune complexes in that they consumed C, were detected by the solid-phase C1q and Raji cell assays, and were precipitable by monoclonal rheumatoid factor. This nonspecific oxidative protein-aggregation reaction may play an important role in the pathogenesis of tissue injury in acute and chronic inflammatory processes and in drug reactions. It could also provide an explanation for the frequent detection of circulating immune complex-like material in a large variety of acute and chronic inflammatory states.     

Isothermal titration calorimetry reveals differential binding thermodynamics of variable region-identical antibodies differing in constant region for a univalent ligand

The classical view of immunoglobulin molecules posits two functional domains defined by the variable (V) and constant (C) regions, which are responsible for antigen binding and antibody effector functions, respectively. These two domains are thought to function independently. However, several lines of evidence strongly suggest that C region domains can affect the specificity and affinity of an antibody for its antigen (Ag), independent of avidity-type effects. In this study, we used isothermal titration calorimetry to investigate the thermodynamic properties of the interactions of four V region-identical monoclonal antibodies with a univalent peptide antigen. Comparison of the binding of IgG1, IgG2a, IgG2b, and IgG3 with a 12-mer peptide mimetic of Cryptococcus neoformans polysaccharide revealed a stoichiometry of 1.9-2.0 with significant differences in thermodynamic binding parameters. Binding of this peptide to the antibodies was dominated by favorable entropy. The interaction of these antibodies with biotinylated peptides manifested greater enthalpy than for native peptides indicating that biotin labeling affected the types of Ag-Ab complexes formed. Our results provide unambiguous thermodynamic evidence for the notion that the C region can affect the interaction of the V region with an Ag.     

Immune Complexes in Cystic Fibrosis

Circulating immune complexes were detected in serum and sputum of patients with cystic fibrosis (C.F.). There were extensive deposits of immunoglobulins and complement immune complexes in several of the C.F. organs, especially the respiratory and gastrointestinal tracts, but not in the kidneys. Significant concentrations of IgG and of complement complexes could be eluted from the lungs of the C.F. patients but not from those of controls. Studies involving immunoabsorption, autoradiography, and molecular sieving through Sephadex G-200 columns identified both bovine serum albumin and staphylococcal α-haemolysin as two of the antigens present in the immune complexes. The sedimentation constant of the immune complexes was about 8S to 11S. The clinical significance of these immune complexes and the wide variety of antibodies detected in C.F. patients are discussed.     

Trapping antigen-antibody complexes within the human placenta

The study was designed to evaluate possible mechanisms for the specific depletion of antifetal antibodies within the human placenta. The principal observations were that endogenous IgG is bound to the same cells to which exogenous antigen antibody complexes attach, i.e., placental macrophages. Moreover, the IgG was demonstrated to be attached to Fcγ receptors on the placental macrophages. Endogenous IgG was completely and specifically displaced by high concentrations of heterologous nonimmune IgG. That the IgG was present as antigen-antibody complexes was evidenced by macrophage receptor affinity, IgG size profile, and C1q binding. The results strongly support the theory that antifetal antibodies are removed by macrophages in the placenta as antigen-antibody complexes in much the same manner as macrophages filter immune complexes from the blood in the liver and spleen.     

Influence of serum IgG antibodies to Chlamydia trachomatis on the functional activity of complement components

The immunoenzyme analysis and the method for the determination of IgG-containing immune complexes, carrying C1q component of the complement, were developed. In human blood sera the functional activity of components C3, complex C1r2s2, the content of C1 inhibitor and complement-activating immune complexes were determined. The comparative analysis of the activity of components C3 and C1r2s2, as well as between the content of C1 inhibitor and the activity of complex C1r2s2 for seropositive and seronegative sera, was made. Pronounced correlation for seropositive sera was observed. In addition, for seropositive sera correlation between an increase in IgG immune complexes and a drop in the functional activity of complex C1r2s2, as well as a drop in the functional activity of complex C1r2s2 and a growth in the titers of IgG antibodies to Chlamydia trachomatis, were established. The decreased functional activity of key complement components, simultaneously with the presence of complement-activating immune complexes and high titers of specific antibodies could be the diagnostic criteria of carrier state.     

Conformational changes in C1q upon binding to IgG oligomers

The interaction between C1q and immune complexes is inhibited by 1-anilino-8-naphthalenesulfonate (ANS) in the concentration range of 2-4 mM. ANS binds to Clq with a 20-fold higher affinity than to IgG [(1986) Mol. Immunol. 23, 39-44] and therefore it is possible to label only C1q with ANS in the presence of IgG. Under such conditions no inhibition is observed. Addition of monomer IgG to a solution of C1q-bound ANS did not significantly alter the fluorescence of the ANS. However when oligomeric IgG was added there was a 2-fold increase in fluorescence over the same IgG concentration range. When C1q was pretreated with diethylpyrocarbonate there was little change in the fluorescence when IgG oligomers were added to C1q:ANS solutions. These results suggest that C1q undergoes conformational changes upon binding to IgG oligomers.     

Comparative studies on biological activities of subcomponents C1q of the first component of human, bovine, mouse and guinea-pig complement

Both the haemolytic activity and the binding ability to immunoglobulin G(IgG) (Fc-binding ability) were comparatively assayed among human, bovine, mouse and guinea-pig C1q. The haemolytic activity was measured by using the sensitized sheep erythrocytes with rabbit immunoglobulin M(IgM)- or IgG-haemolysin. The Fc-binding ability was assayed by using immune complexes made of rabbit IgG-antibody against human serum albumin as well as agglutination of latex particles coated with human, bovine or rabbit IgG (IgG-latex). The specific haemolytic activity was comparable with between bovine and mouse C1q, while those of guinea pig and human C1q were significantly lower than those of the others. Only the human and mouse C1q showed significantly positive agglutinating activity of human or bovine IgG-latex. In the case of the use of rabbit IgG-latex, each of these C1q gave much weaker agglutination. On the other hand, the ability of all these C1q to bind to Fc of immune complexes specifically was almost comparable. The discrepancy in specific activities between the haemolysis and the Fc-binding ability may suggest that these two biological activities are not always correlative and that these are independent biological phenomena.     

Effector Function Activities of a Panel of Mutants of a Broadly Neutralizing Antibody against Human Immunodeficiency Virus Type 1

The human antibody immunoglobulin G1 (IgG1) b12 neutralizes a broad range of human immunodeficiency virus-type 1 (HIV-1) isolates in vitro and is able to protect against viral challenge in animal models. Neutralization of free virus, which is an antiviral activity of antibody that generally does not require the antibody Fc fragment, likely plays an important role in the protection observed. The role of Fc-mediated effector functions, which may reduce infection by inducing phagocytosis and lysis of virions and infected cells, however, is less clear. To investigate this role, we constructed a panel of IgG1 b12 mutants with point mutations in the second domain of the antibody heavy chain constant region (CH2). These mutations, as expected, did not affect gp120 binding or HIV-1 neutralization. IgG1 b12 mediated strong antibody-dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC) of HIV-1-infected cells, but these activities were reduced or abrogated for the antibody mutants. Two mutants were of particular interest. K322A showed a twofold reduction in FcgammaR binding affinity and ADCC, while C1q binding and CDC were abolished. A double mutant (L234A, L235A) did not bind either FcgammaR or C1q, and both ADCC and CDC functions were abolished. In this study, we confirmed that K322 forms part of the C1q binding site in human IgG1 and plays an important role in the molecular interactions leading to complement activation. Less expectedly, we demonstrate that the lower hinge region in human IgG1 has a strong modulating effect on C1q binding and CDC. The b12 mutants K322A and L234A, L235A are useful tools for dissecting the in vivo roles of ADCC and CDC in the anti-HIV-1 activity of neutralizing antibodies.     

Surfactant protein A binds to IgG and enhances phagocytosis of IgG-opsonized erythrocytes

Surfactant protein (SP)-A and SP-D, immunoglobulins, and complement all modulate inflammation within the lung by regulating pathogen clearance. For example, SP-A binds to and opsonizes a variety of bacteria and viruses, thereby enhancing their phagocytosis by innate immune cells such as alveolar macrophages. Immunoglobulins, which bind to antigen and facilitate Fc receptor-mediated phagocytosis, can also activate complement, a family of soluble proteins with multiple host defense functions. Previous studies showed that SP-A and complement protein C1q interact. Since complement protein C1q binds to IgG and IgM immune complexes, the hypothesis tested in this study was that SP-A, which is structurally homologous to C1q, also binds to IgG and affects its functions. SP-A binds to the Fc, rather than the Fab, region of IgG. Binding is calcium dependent but not inhibited by saccharides known to bind to SP-A's carbohydrate recognition domain. The binding of SP-A does not inhibit the formation of immune complexes or the binding of IgG to C1q. In contrast, SP-A enhances the uptake of IgG-coated erythrocytes, suggesting that SP-A might be influencing Fc receptor-mediated uptake. In summary, this study shows a novel interaction between SP-A and IgG and a functional consequence of the binding.     

Immunoglobulin G antibody bound to the C1q subcomponent of human complement exhibits segmental flexibility

The rotational dynamics of rabbit immunoglobulin G (IgG) anti-dansyl antibodies bound to the C1q subcomponent of human complement were studied by nanosecond fluorescence spectroscopy. Deconvoluted anisotropy decays of IgG-C1q mixtures were fitted to a two-exponential expression and were corrected for the effects of unbound IgG, which was determined with an analytical ultracentrifuge. Compared with the anisotropy parameters for free IgG, the pre-exponential weighting factors and the short correlation time of the C1q-bound antibody were nearly unchanged, and the long correlation time increased by only about 45 nanoseconds. These results, together with rotational diffusion calculations, indicate that the Fab arms of the C1q-bound antibody exhibited considerable flexibility. This finding may have biological relevance because it suggests that C1q can bind to the Fc segments of IgG molecules anchored in an immune complex, even though the angles between the two Fab arms of the different antibodies may vary. The results of this study also support our earlier interpretation that both the short and long correlation times of IgG principally represent flexible motions of the Fab segments.     

Interaction of C3 with antigen-antibody complexes in the process of solubilization of immune precipitates

The binding properties of activated C3 to immune complexes were studied by using solubilization phenomenon as a model system. IgG or F(ab')2 immune precipitates were solubilized by the six isolated alternative pathway proteins, and the solubilized complexes were analyzed by SDS-PAGE. As a result of solubilization, we observed some high m.w. bands. Under reducing conditions, the bands with m.w. of 150,000 and 115,000 appeared in the case of IgG and F(ab')2 complexes, respectively. Two-dimensional SDS-PAGE revealed that hydroxylamine treatment resulted in the dissociation of the 150,000-m.w. polypeptide into the C3 alpha-65 and the heavy chain of IgG. Similarly, the 115,000-m.w. polypeptide was dissociated into the C3 alpha-65 and the Fd chain. Therefore, it is likely that iC3b binds covalently to the Fd region of the heavy chain of IgG via an ester bond. Under nonreducing conditions, iC3b-IgG and iC3b-F(ab')2 complexes had apparent m.w. of 340,000 and 270,000, respectively, corresponding to one iC3b molecule bound to one antibody molecule. In addition, a considerable amount of iC3b also binds to antigen molecules via an ester bond. The findings that C3 binds to the F(ab')2 molecules and bovine serum albumin, which contain only a small amount of carbohydrate, suggest that C3 may not bind to the carbohydrate moiety of antibody molecules. Indeed, various carbohydrate molecules did not inhibit the solubilization even at high concentrations. In contrast, acetyl tyrosine having an aromatic ring and a hydroxyl group produced the best inhibition of the solubilization. Furthermore, we demonstrated that generation of C3b in the presence of 3H-tyrosine resulted in covalent binding of the tyrosine specifically to the C3 alpha' chain, indicating that the inhibition of solubilization may be due to the competition between tyrosine and immune complexes for the covalent binding of C3. Thus, it could be concluded that C3 binds covalently to the amino acid residues of antigen and antibody molecules during solubilization.     

The binding of human complement component C4 to antibody-antigen aggregates.

The binding of human complement component C4 to antibody-antigen aggregates and the nature of the interaction have been investigated. When antibody-antigen aggregates with optimal C1 bound are incubated with C4, the C4 is rapidly cleaved to C4b, but only a small fraction (1-2%) is bound to the aggregates, the rest remaining in the fluid phase as inactive C4b. It has been found that C4b and th antibody form a very stable complex, due probably to the formation of a covalent bond. On reduction of the C4b-immunoglobulin G (IgG) complex, the beta and gamma chains, but not the alpha' chain, of C4b are released together with all the light chain, but only about half of the heavy chain of IgG. The reduced aggregates contain two main higher-molecular-weight complexes, one shown by the use of radioactive components to contain both IgG and C4b and probably therefore the alpha' chain of C4b and the heavy chain of IgG, and the other only C4b and probably an alpha' chain dimer. The aggregates with bound C1 and C4b show maximal C3 convertase activity, in the presence of excess C2, when the alpha'-H chain component is in relatively highest amounts. When C4 is incubated with C1s in the absence of aggregates, up to 15% of a C4b dimer is formed, which on reduction gives an alpha' chain complex, probably a dimer. The apparent covalent interaction between C4b and IgG and between C4b and other C4b molecules cannot be inhibited by iodoacetamide and hence cannot be catalysed by transglutaminase (factor XIII). The reaction is, however, inhibited by cadaverine and putrescine and 14C-labelled putrescine is incorporated into C4, again by a strong, probably covalent, bond. It is suggested that a reactive group, possibly an acyl group, is generated when C4 is activated by C1 and that this reactive group can react with IgG, with another C4 molecule, or with water.