An examination of the structural and biological properties of three intravenous immunoglobulin preparations.

Three commercial preparations of immunoglobulin G prepared for administration by the i.v. route were tested for their physical integrity and in vitro biological activity. Size exclusion chromatography by HPLC in native and denaturing buffers together with SDS-PAGE analysis were used to determine whether covalent-bond cleavage had occurred as a result of procedures used in their preparation. C1 complement binding assays and measurements of competitive binding to an Fc receptor-bearing promonocyte cell line U937 were used to assess whether such changes had altered the capacity of these preparations to engage biological effector functions. A purified IgG1 myeloma protein was used as a reference standard. WinRho, an unmodified IgG, consisted almost wholly of monomeric IgG by HPLC size exclusion and showed no evidence of proteolytic fragments in denaturing buffers or on SDS-PAGE. Sandoglobulin, a product treated at pH 4 with pepsin, contained about 10% dimeric protein and, as revealed under denaturing conditions, about 2% fragments. Relative affinity of binding to U937 cells was similar to WinRho. C1 binding by Sandoglobulin showed normal activity with 50% inhibition at 2.8 nM. Gamimune, modified by partial reduction and alkylation, contained about 15% dimers. Between 20 and 30% of the preparation retained covalent interchain disulfides. Binding to U937 cells was two-fold weaker than the other preparations and binding to C1 was also diminished and modified. This accords well with previous reports of the deleterious effect of reduction and alkylation on Fc function.

The interaction of the Facb fragment of rabbit anti-sheep red cell IgG with guinea pig macrophages, and human monocytes and granulocytes.

Yasmeen et al. (J. Immun. 110, 1706-1709, 1973) have previously reported on the binding requirements of the guinea pig peritoneal macrophage Fc receptor. The C gamma 3 domain fragments of human IgG1, in contrast to the C gamma 2 domain fragment, were able to bind to these macrophages, as demonstrated by both direct and indirect rosette tests. We now report that we have been unable to show binding by the C gamma 2-bearing rabbit Facb fragment to either peritoneal or alveolar macrophages of the guinea pig. This evidence is therefore in agreement with the hypothesis proposed by Yasmeen et al. (1973) that the C gamma 2 homology region does not contribute directly to the binding requirements for this cell type. The same protein, rabbit anti-sheep erythrocyte Facb, when coated on sheep erythrocytes, did not form rosettes with human granulocytes, but did form some rosettes with human monocytes.

Human IgG1 and its Fc fragment bind with different affinities to the Fc receptors on the human U937, HL-60 and ML-1 cell lines.

Measurements were made of the binding of human monomeric 125I-IgG1 and 125I-Fc to U937 cells at room temp. Analyses of the binding data showed that these cells possessed a single class of receptor (FcR) for Fc or IgG and, although both ligands were found to bind to the same number of sites per cell, Fc was found to bind with about twice the affinity of IgG. At 20 degrees C estimates of the forward rate constants and the dissociation rate constants for IgG and Fe were 1.13 and 3.65 X 10(7) M-1 min-1 and 0.33 and 0.57 X 10(-2) min-1 respectively. Independent determinations of the association constants (Ka) under the same experimental conditions gave values of 0.98 X 10(9) M-1 for IgG and 3.1 X 10(9) M-1 for Fc. Thus the Fc fragment of IgG appears to bind to U937 FcR at 3-4 times the rate of IgG and to dissociate at about twice the rate, resulting in higher values of Ka for the Fc-FcR than for the IgG-FcR interaction. Also, in competitive-binding experiments and in EA rosette inhibition assay the Fc fragment was consistently found to be more efficient in FcR binding than IgG. Similar results were obtained using HL-60 and ML-1 cells which possess FcR like those on U937 cells and with IgG1 and Fc prepared from other myelomas. IgG and Fe which had undergone mild reduction and alkylation bound to the same number of FcR per U937 cell as the non-reduced ligands but the affinity of binding was diminished to a similar degree with both ligands, suggesting that the major effect of cleavage of the interchain disulfide bonds on cytophilic binding is due to alteration of the native quaternary relationships of the C gamma 2 and C gamma 3 domains.

The effect of fragment B of staphylococcal protein A on the binding of rabbit IgG to human granulocytes and monocytes.

The effect of fragment B of staphylococcal protein A on the binding of rabbit IgG to human granulocytes has been examined. When rabbit IgG sensitized sheep erythrocytes (EA) were preincubated with increasing amounts of fragment B, a dose dependent inhibition of rosette formation between the human granulocytes and the treated EA was observed. In contrast, rosette formation between human monocytes and fragment B treated EA was not impaired. These results confirm previous findings which revealed differences in the specificities of the human Fc receptors of these cells. Rabbit IgG was used in place of human IgG for the preparation of the sensitized erythrocytes since human IgG coated human erythrocytes were agglutinated by fragment B, indicating the possibility of a secondary binding site for human IgG in this fragment.

Functional affinity constants of subfragments of immunoglobulin G for Clq.

The functional affinity constants for Clq of subfragments if IgG1 representing the C gamma 2 (c gamma 2III) region or the whole C gamma 3 region of Fc (pFc'), have been measured by examining the ability of these fragments to inhibit the interaction between radiolabelled Clq and glutaraldehyde-treated human erythrocytes or aggregated human IgG. The value of the functional affinity constant for the C gamma 2III fragment was the same as that for Fc and that determined previously for monomeric IgG, indicating that all the elements necessary for Clq binding are contained in a single C gamma 2 domain. The pFc' fragment was inactive but a more degraded trypsin fragment from this region, at C gamma 3, showed the same affinity of binding for Clq as the C gamma 2III and Fc. These results confirm earlier findings that it is not combination of residues in the C gamma 2 which bind Clq which is responsible for their activity but their accessibility.

The structure and binding characteristics of serum amyloid protein (9.5S alpha 1-glycoprotein).

No difference have been observed in the properties of amyloid P-component (AP) and its serum counterpart (SAP) which might account for the deposition of the former in amyloidosis. Purified by nonselective techniques, preparations of AP and SAP were shown to have similar molecular weights and peptide composition, identical morphology in the electron microscope (EM) and to be antigenically indistinguishable. Both proteins were soluble in EDTA but readily precipitated in the presence of calcium ions, forming characteristic two-dimensional arrays in the EM. In serum however, SAP was not aggregated and sedimented at 9.5S in Ca2+ as did the purified protein in EDTA. Precipitation of purified SAP by calcium could be prevented by pretreatment with acid hydrolysates of agarose or SP Sephadex, matrices for which SAP has a calcium-dependent affinity. It is proposed that SAP circulates in combination with a low molecular weight natural ligand which prevents its aggregation. While the identity of natural ligand for SAP is as yet unknown, it is likely to resemble the glycosidic subunits in agarose.

The C1q receptor site on human immunoglobulin G.

Of the many functional properties of the immunoglobulin G (IgG) molecule, only antigen binding and the interaction with the C1q component of complement have been shown to be uniquely associated with the individual compact domains which make up this immunoglobulin. The chemical and biological evidence for the exclusive association of the C1q-binding site with the CH2 domain is reviewed, affirming that the site probably is centered on the residues 279-295 which are located on the outside of the molecule and contain, in particular, three positively charged residues which are thought to be vital to the interaction. An alternative site (residues 316-338), having similar exposure and charge characteristics, is discussed and arguments are presented indicating why the former is currently favoured.

Electron microscopy of serum amyloid protein in the presence of calcium; alternative forms of assembly of pentagonal molecules in two-dimensional lattices.

The P component is a protein believed to be of serum origin commonly found in the highly ordered proteinaceous tissue deposits called amyloid. Both the protein recovered from the tissue and its serum counterpart have an identical characteristic appearance in the electron microscope, consisting of pentagonal plates which are often assembled as columns of stacked discs. These images, seen in the absence of calcium, are replaced by three more complex assemblies when low concentrations of calcium are present. One of the structures is crystalline (III), the other two appear to be planar lattices, one having a distinguishable structure with threefold symmetry (1). The structural elements of the other lattice (II) which takes the form of a cylinder are less distinct. It is suggested that the regular arrays which P component forms in the presence of calcium may be the basic framework on which amyloid deposits form.

The identification of a previously unrecognized subcomponent of the first component of complement.

The use of an affinity chromatography method designed to isolate C1 from serum has led to the discovery of a novel plasma protein, II-P2, associated with C1. The persistent Ca++-dependent association of II-P2 with C1 subcomponents following euglobulin precipitation, affinity chromatography on Sepharose-IgG, and density gradient ultracentrifugation indicates that II-P2 might be a C1 subcomponent. Using purified preparations of II-P2 it was found that a) II-P2 was retained on Sepharose-IgG through a Ca++-dependent link with C1q,b) II-P2 enhanced the C1 activity of mixtures of C1s and C1q in a dose-dependent fashion, c) II-P2 bound firmly to EAC1q4 cells and enhanced their C1s-binding ability. Fractionation of C1 by DEAE-Cellulose chromatography under the conditions that led to the original identification of C1q, C1r, and C1s resulted in recovery of II-P2 in the fractions containing C1r. The evidence presented confirms that II-P2 is a C1 subcomponent (C1t).

The macromolecular structure of the first component of complement.

The binding of C1 to IgG and the interactions between C1 subcomponents have been studied by affinity chromatography of serum C1 and purified C1 proteins on Sepharose-IgG. Affinity chromatography of serum on Sepharose-IgG resulted in the binding of C1; subsequent washing with EDTA removed only C1s and C1t. Affinity chromatography of serum-EDTA on Sepharose-IgG resulted in binding of only C1q and C1r. Affinity chromatography of serum on Sepharose-tryptophan-modified-IgG resulted in the binding of only C1r and C1s. By the use of purified C1 proteins and Sepharose-IgG in binding studies it was confirmed that both C1q and C1r bind independently to sites on IgG and hold C1s and C1t by Ca++-dependent bonds. Measurements of the hemolytic activity of various combinations of C1 subcomponents confirmed the data obtained by the affinity binding studies. Both C1t and C1r independently enhanced the C1 activity of C1q-C1s mixtures; maximal activity required all four subcomponents. Sucrose gradient ultracentrifugation of mixtures of C1 proteins showed formation of the following complexes: C1qs (12S), C1qrs (15S), C1qst (18-23S), and C1qrst (19S). The evidence suggests that the spatial sequence of the components of the Sepharose-IgG-Serum C1 complex is: Sepharose-IgG: C1q: C1t: C1s: C1r: IgG-Sepharose. The probable physiologic significance of this model is discussed.

The use of diafiltration for the measurement of dialysable ligand binding to macromolecules: an examination of the theoretical parameters using concanavalin A in an Amicon cell.

The technique of diafiltration as a means of determining the characteristics of small ligand binding by proteins has been evaluated by measuring the binding of 45Ca to concanavalin A. Deposition of protein on the diafiltration membrane was found to be an important source of error but which could be eliminated by presaturation of the membrane with polylysine. The association constant for concanavalin A and calcium was 1.8 x 10(4) M-1, with a value of n = 2.2 atoms of calcium being bound per dimer of concanavalin A. The concentration of protein relative to the dissociation constant is also an important variable, and it was found that the method of diafiltration is only useful when protein concentrations greater than the value of the dissociation constant are used.

Structural requirements of immunoglobulin G for binding to the Fc gamma receptors of the human tumor cell lines U937, HL-60, ML-1, and K562.

Fc receptors (FcR) on U937, HL-60, ML-1, and K562 cells were characterized by using competitive binding assays and EA rosette inhibition studies. Like human monocytes, U937, HL-60, and ML-1 cells bound monomeric human IgG Fc with high affinity and mildly reduced and alkylated human IgG and Fc with a somewhat diminished affinity. In contrast, K562 cells had a much lower affinity for monomeric human IgG or Fc. Concentrations of these proteins as high as 10(-5) M were needed to completely block EA rosette formation. There was no binding of reduced and alkylated IgG and Fc. We assessed the influence of various segments of IgG on FcR interactions by using human pFc', rabbit Facb, mouse IgG2b-IgG2a hybrid Ig, and also studied the effect of reduction of the interchain disulfide bonds. The FcR on all four cell types bound rabbit IgG but not rabbit Facb or human pFc', suggesting that rabbit C gamma 3 domains are required for FcR interaction and that isolated human C gamma 3 domains do not have a human FcR binding site. Murine IgG2a, but not IgG2b, was cytophilic for U937, HL-60, and ML-1 cells. Binding studies with the use of several mouse myeloma variant Ig molecules having hybrid gamma 2b-gamma 2a heavy chains showed that variants with a complete gamma 2a Fc region bound to these FcR-like IgG2a, whereas those having gamma 2a sequences only in the C gamma 3 region and in a short adjacent segment of the C gamma 2 region behaved like IgG2b and did not bind. These results suggested that additional murine gamma 2a sequences are required for FcR binding. Interestingly, the Fc fragments from murine proteins with a complete gamma 2a Fc region bound only to a limited extent. These fragments are shorter than the human IgG1 Fc fragments, and they lack that segment of the hinge region that includes the interchain disulfide bonds. Cleavage of the interchain disulfide bonds of murine Ig having a complete gamma 2a Fc region diminished binding to a similar extent as that observed with human IgG. Together, these findings provide additional insight into the roles of the hinge, C gamma 2, and C gamma 3 regions of human, rabbit, and mouse IgG in their interaction with the FcR of human tumor cells.

The structure and function of immunoglobulin domains: studies with beta-2-microglobulin on the role of the intrachain disulfide bond.

Beta-Microglobulin, a low-molecular-weight protein structurally related to the homology regions of immunoglobulins, has been used to study the role of the intrachain disulfide bond in the unfolding of immunoglobulin domains. The intact protein could be reversibly unfolded in guanidine hydrochloride, as judged by circular dichroism and optical rotation. Similarly, reoxidation of the reduced protein, during transfer from high concentrations of guanidine to neutral aqueous buffer, yielded a product with spectral characteristics typical of the native protein. However, if the free SH groups were prevented from reoxidizing either by chemical modification or by holding them in the reduced state, the molecule appeared to be in the randomly coiled state even under conditions where the intact protein is in the native conformation, judged on the basis of chiroptical measurements. The complement-fixing activity exhibited by native beta-2-microglobulin was retained by the reduced and alkylated derivative, suggesting that the site may be formed from a linear array of amino acids. We suggest a model for the folding of beta-2-microglobulin (and immunoglobulin domains) in which one of the early folding events results in disulfide bond formation, the latter being an obligatory step for continued folding to the native state.

A plasma protein fractionation procedure for use in studies of protein metabolism.

A two-step method for the separation of five different plasma proteins on a preparative scale, which is capable of being extended to allow the separation of other plasma proteins, is described. The proteins separated were fibrinogen, two alpha(1)-glycoproteins, albumin and transferrin. The alpha(1)-glycoproteins were characterized in terms of electrophoretic mobility, ultracentrifugal and immunological characteristics. By using this method, it was shown that a single sample of plasma could be fractionated to yield purified proteins in sufficient quantity to simultaneously measure the synthesis of the two alpha(1)-glycoproteins, albumin and transferrin in the rat with McFarlane's technique (McFarlane, 1963; Reeve et al., 1963; McFarlane et al., 1965).

The structure and function of immunoglobulin domains. II. The importance of interchain disulfide bonds and the possible role of molecular flexibility in the interaction between immunoglobulin G and complement.

The observation that reduction of the inter-chain disulfides in rabbit antibody destroys its ability to interact with complement was confirmed and shown to be true also of human meyloma IgG1 subclass proteins. In the latter case a C1-binding assay was used. Further studies indicated that it was the interheavy chain disulfides which were essential for complement-binding activity: Non-covalently reassembled IgG (LHHL) was devoid of C1-fixing activity whereas molecules formed from covalently linked heavy chain dimers, and reduced and alkylated light chains (ie., LH-HL) were as active as the parent intact IgG. Fc fragments from IgG1 bound C1 and this activity was insensitive to the presence or absence of intact interchain disulfides. These bonds therefore are neither directly involved in C1 binding nor essential for the integrity of the binding site. We have also shown that although IgG4 does not bind C1, Fc fragments derived from this subclass fix C1 with an affinity comparable to that of the corresponding fragment from IgG1. These data suggest that quaternary interaction with other regions of the molecule (ie., Fab) may modulate the activity of the C1-binding site.

The relationship of a serum protein, C1t, to a common nonfibrillar constituent of amyloid (P component) as revealed by immunohistochemical studies.

C1t, a serum protein isolated by affinity chromatography on Sepharose bears a remarkable ultrastructural and physiocochemical resemblance to P component, a common constituent of amyloid. This study provides further evidence for their similarity by the demonstration of immunologic identity and by the presence of C1t in amyloid deposits of various types using immunoperoxidase and immunofluorescent techniques. In addition, subcomponents of C1, as well as C3, C4, C5, and properdin, were demonstrated to a limited extent. The possible role of C1t/P component in amyloidogenesis is discussed in the light of recent advances in our knowledge of the nature of amyloid substance.

C1 binding by murine IgM. The effect of a Pro-to-Ser exchange at residue 436 of the mu-chain.

We have examined a defect in complement activation in a mutant trinitrophenyl-binding pentameric murine monoclonal IgM which has serine replacing the proline normally found at position 436 in the protein. The mutant protein showed equivalent hapten binding but a 100-fold decreased ability to initiate complement-dependent lysis of trinitrophenyl-coupled erythrocytes at physiological ionic strength (mu = 0.15). C4b deposition mediated by the mutant protein was impaired to a similar degree. C1 bound by the mutant protein showed C1s to C1-s conversion, suggesting normal activation. When measured at reduced ionic strength (mu = 0.06), the C1 and C1q binding affinity of the mutant protein was approximately one-half that of the wild type. However, the C1 binding affinity of the mutant protein showed a greater dependence upon ionic strength such that at physiological ionic strength we estimate a 50-fold lower C1 binding affinity for the mutant molecule. Kinetic studies suggested that this difference in affinity was largely attributable to differences in association rates. In addition, a fixed proportion of the mutant molecules showed no C1 binding. We conclude that the defect in complement activation occurs at the level of C1 binding. Our data support a role for the C mu 3 domain (residues 340-440) in C1 binding by IgM.

A functional, thioester-containing alpha 2-macroglobulin homologue isolated from the hemolymph of the American lobster (Homarus americanus).

An alpha 2-macroglobulin-like protease inhibitor was isolated from the cell-free hemolymph of the american lobster (Homarus americanus) by ion-exchange chromatography and gel filtration. Whereas the undissociated molecule has a molecular weight of 342,000 as determined by ultracentrifugation studies, under reducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis, the protein has a subunit molecular weight of 180,000. On the basis of this and other evidence, we conclude that the lobster protein is a dimer consisting of two disulfide-bonded monomers. The purified protein inhibits proteolytic enzymes but protects the esterolytic activity of trypsin toward low molecular weight substrates from inactivation by soybean trypsin inhibitor. The methylamine sensitivity of this activity suggests the presence of an internal thioester bond. This was confirmed by the covalent incorporation of [14C]methylamine, by the formation of Mr 55,000 and 125,000 autolytic cleavage fragments in sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and, more directly, by the amino acid sequence of a tryptic peptide containing the putative thioester region. Whereas the N-terminal amino acid sequence (22 residues) of the protein revealed an overall identity of only 18% when compared with the human protein, the sequence of the thioester-containing peptide was highly conserved, both with respect to human alpha 2-macroglobulin and to other proteins having a thioester bond. The protein showed the "slow to fast" conformational change typical in alpha 2-macroglobulins in nondenaturing gel electrophoresis after treatment with trypsin, but not after incubation with methylamine.