Various native and hinge-modified forms of Ig with identical Ids were reacted with an anti-Id mAb, and the resultant immune complexes were analyzed by negative stain immunoelectron microscopy. Complexes were scored for their geometry (linear versus ring complexes) and size (dimer, trimer, etc.). Ring dimers are the thermodynamically most favorable configuration, unless inhibited by steric and/or flexibility constraints. We found ring dimerization to correlate with the length of the upper, but not middle or lower, hinge. In contrast, the geometry and size of complexes of those molecules lacking formal hinges were unpredictable. A hingeless IgG mutant and native IgE readily formed ring dimers. Remarkably, monomeric IgM formed more ring dimers than any of the other Igs tested, including IgG3. We also tagged the Fab arms and measured the mean Fab-Fab angles and the degree of angular variation for each type of Ig. Surprisingly, IgM proved the most flexible by this assay. In hinged Igs, there was a correlation between length of the upper hinge and Fab-Fab flexibility. In contrast, we found no correlation between the mean Fab-Fab angle in uncomplexed Igs and their ability to dimerize with anti-Id mAb. These data suggest that the physicochemical methods typically used to evaluate molecular flexibility are often of low predictive value when tested in a functional assay.

In a recent study (1), we compared the relative flexibility of the human IgG subclasses by using immunoelectron microscopy to visualize immune complexes formed with anti-idiotypic Abs (anti-Id Ab) in solution. We were able to demonstrate that differences in hinge flexibility have a profound effect on the types and sizes of soluble immune complexes formed. Moreover, we were able to dissect out one of the modes of hinge flexibility, termed the hinge-folding flexibility, which is the variation in angle between the Fab arms, for comparison between the IgG subclass. The data show that human IgG3 is more flexible than IgGs 1, 2, and 4 in all respects. In this study, we extend these findings by similarly analyzing genetically engineered forms of IgG3, in which the extended hinge has been altered, replaced, or eliminated to determine more precisely which regions of the IgG3 hinge contribute to its high flexibility. The amino acid (aa)3 composition of the native and mutant Ig hinges is given in Table I. In addition, we perform parallel flexibility analyses on human IgA2, IgM (as monomeric subunits), and IgE.

Table I.

Hinge region amino acids

Ig TypeC-terminal CH1UHMHLH\E
IgG1 VDKRV EPKSCDKTHT CPPCP APELLGGP\E 
IgG2 VDKTV ERK CCVECPPCP APPVAGP\E 
IgG3 (17-15-15-15) VDKRV ELKTPLGDTTHT CPRCP(EPKSCDTPPPCPRCP)3 APELLGGP\E 
17-15-15 VDKRV ELKTPLGDTTHT CPRCP(EPKSCDTPPPCPRCP)2 APELLGGP\E 
17-15 VDKRV ELKTPLGDTTHT CPRCP(EPKSCDTPPPCPRCP)1 APELLGGP\E 
15-15-15 VDKRV EPKS CDTPPPCPRCP(EPKSCDTPPPCPRCP)2 APELLGGP\E 
15 VDKRV EPKS CDTPPPCPRCP APELLGGP 
IgG4 VDKRV ESKYGPP CPSCP APEFLGGP\E 
HM4 (m0/231AlaAlaCysAla232) VDKRV AAA APELLGGP 
HM5 (m0/C131S) VDKRV – – APELLGGP\E 
IgA2 DVTV CPVPPPPPCC HP 
Ig TypeC-terminal CH1UHMHLH\E
IgG1 VDKRV EPKSCDKTHT CPPCP APELLGGP\E 
IgG2 VDKTV ERK CCVECPPCP APPVAGP\E 
IgG3 (17-15-15-15) VDKRV ELKTPLGDTTHT CPRCP(EPKSCDTPPPCPRCP)3 APELLGGP\E 
17-15-15 VDKRV ELKTPLGDTTHT CPRCP(EPKSCDTPPPCPRCP)2 APELLGGP\E 
17-15 VDKRV ELKTPLGDTTHT CPRCP(EPKSCDTPPPCPRCP)1 APELLGGP\E 
15-15-15 VDKRV EPKS CDTPPPCPRCP(EPKSCDTPPPCPRCP)2 APELLGGP\E 
15 VDKRV EPKS CDTPPPCPRCP APELLGGP 
IgG4 VDKRV ESKYGPP CPSCP APEFLGGP\E 
HM4 (m0/231AlaAlaCysAla232) VDKRV AAA APELLGGP 
HM5 (m0/C131S) VDKRV – – APELLGGP\E 
IgA2 DVTV CPVPPPPPCC HP 

The structural IgG3 hinge is composed of a 12-aa upper hinge (UH) stretching from the C-terminal end of CH1 to the first hinge cysteine, a 50-aa middle hinge (MH) stretching from the first to the last cysteine in the hinge, and an 8-aa lower hinge (LH) stretching from the last cysteine in the hinge to Gly237 (EU numbering) in the CH2 (2, 3, 4). The genetic IgG3 hinge, on the other hand, is coded for by four exons separated by short introns (5, 6). The first exon codes for 17 aa. Exons 2, 3, and 4 are identical, each coding for 15 aa. Thus, for convenience, the native IgG3 molecule can be described as having a 17-15-15-15 genetic hinge. The MH is rich in cysteine and proline and, by analogy to the solution and crystal forms of the IgG1 MH (7, 8), is believed to be a disulfide-bonded rodlike structure estimated by various means to be 90 to 110 Å long (1, 9, 10, 11, 12). Construction and biologic characterization of the hinge mutants used in this study have previously been described (13, 14, 15, 16). Truncated hinges were shortened by the removal of specific hinge exons and are designated 17-15-15, 17-15, 15-15-15, and 15 (17). Two additional hinge mutants were produced, one of which, HM4 (m0/231AlaAlaCysAla232), has the native genetic hinge replaced with a short but potentially flexible synthetic segment encoding the amino acids Ala-Ala-Ala-Cys-Ala (13), and the other, HM5 (m0/C131S), which has no UH or MH (14). The latter has an additional modification in which both light chains are disulfide bonded to each other as in the human Dob (18, 19) and Mcg (20) mutant IgG1 molecules.

The two human IgA subclasses have distinctly different hinge regions. IgA1, the predominantly secreted subclass, has a 26-aa hinge containing several sites for glycosylation. By analogy to the IgG hinges, it appears likely that IgA2 has a 13-aa proline-rich hinge free of glycosylation sites. Ten of the amino acids of the hinge comprise the MH, with only perhaps a single proline representing a UH and a His-Pro for a LH. Sykulev et al. (21) using spin labeling found IgA2 to be considerably less flexible than IgA1, but more flexible than IgM and IgE.

IgM and IgE, unlike IgG and IgA, do not possess a formal hinge, having instead an extra C region domain at the hinge site. However, the Fab arms of IgM and IgE are believed to exhibit some, albeit limited, flexibility at the CH1-CH2 and CH2-CH3 junctions (11, 22). Nevertheless, a variety of evidence indicates that human and murine IgE are less flexible than any of the hinged classes or subclasses (21, 23, 24). Moreover, the Fc of IgE is believed to be in a hooked conformation both in solution and when complexed with the IgE FcR (24, 25). Various physicochemical evidence suggests that the C-terminal region of the Fcε is in close proximity to the CDR region of the molecule (22, 26). Modeling suggests that the Fc is hooked so that it curls up toward the Fab arms. Anisotropy studies suggest that human IgM is moderately flexible (27, 28), but direct comparisons of monomeric IgM to IgG using combining site labels have not been reported. Electron-microscopy studies clearly show Fab-Fc flexure in IgM that appears to be both at the CH1-CH2 and CH2-CH3 junctures. The flexure is most evident when IgM is induced to adopt the staple conformation upon binding to multiple epitopes on a planar surface (4).

In this study, we compare the ability of an Id-matched set of IgM (monomeric subunits), IgE, IgA2, and six hinge-mutant forms of IgG3 to adopt the conformation necessary to form small immune complexes with an anti-Id mAb. In addition, we compare the Fab-Fab angles of uncomplexed forms of each of the Igs to gain insight into their innate hinge-folding flexibility capacity. The data reveal several unexpected observations.

Cell lines producing chimeric IgG3 wild-type and mutant molecules composed of murine VH and λ1-light chain and human γ-chains have previously been described (13, 14, 15, 16). All of the above Abs have identical specificity (for the haptens NP and NIP) and Ids. IgA2- and IgE-producing cell lines were obtained from European Center of Animal Cell Culture (Salisbury, U.K.). The construction of the IgM C575S mutant secreting only monomeric subunits has previously been described (29). The murine IgG1 anti-Id mAb, 5B5, was produced by the Hybridoma Facility (Department of Biologic Science, Florida State University, Tallahassee, FL), as previously described (1). All cells were grown in RPMI 1640 supplemented with 10% FCS.

Culture supernatants containing each of the various Igs and the anti-Id mAb were collected. Id and anti-Id Ig were purified by affinity chromatography using NIP-caproate-o-succinimide (Genosys Biotech, The Woodlands, TX)-Sepharose and Id-IgG-Sepharose columns, respectively. Specific mAbs were eluted with 0.2 M glycine sulfate, pH 2.3, and immediately neutralized with 1 M Tris base. Some of the Id Ab were also isolated from the NIP-Sepharose by 0.1 mM free NIP (Cambridge Research Biochemicals, Cambridge, U.K.). Fab fragments of anti-Id 5B5 were produced with papain-Sepharose (Pierce, Rockford, IL), as described by the manufacturer. Ab concentrations were determined by UV absorption (at 280 nm) using published extinction coefficients (30).

Immunoelectron-microscopic analyses of mAbs and immune complexes were performed by negative staining, as previously described (31, 32, 33). Whole molecule reactants were mixed at a one-to-one molar ratio (at ∼1 μg/ml each) in borate-buffered saline and incubated at room temperature for 30 min. Initially, not all of the Id/anti-Id combinations yielded >90% of the molecules in complexes (the minimum acceptable percentage), probably due to differences in size, extinction coefficient, and measurement error. Consequently, slight adjustments in the concentrations of reactants were made until the appropriate level of complex formation was achieved. Following incubation, the reactants were affixed to carbon membranes, stained with uranyl formate, and mounted on copper grids for analysis. Electron micrographs were recorded at ×50,000 or ×100,000 magnification on a JEOL CX 1200 electron microscope. Scoring of immune complexes was performed directly on the electron micrographs with a hand-held lens, and angular and hinge-length measurements were scored with an optical loupe fitted with measuring graticules (Electron Microscopy Sciences, Fort Washington, PA). At least 1000 molecules were scored for each sample of immune complexes. For angular measurements, at least 100 molecules were scored for each sample. All scoring was done blind.

The segmental flexibility of Ab molecules represents a key aspect of their functionality. As cell surface receptors, Ag-induced Ab cross-linking on, for example, B cells or mast cells, determines whether an Ag engagement translates into a signaling event. Cross-linking, in turn, depends not only on affinity, but on the ability of the Fab arms of the Ig molecules to adjust their orientation to accommodate particular orientation of any cognate epitope array. Similarly, soluble Abs such as cytotoxic and receptor-targeted Abs usually bind cell or viral surface epitopes bi- or polyvalently and generally cross-link to exert their effector function. Soluble Abs are also responsible for the neutralization of soluble Ags such as bacterial toxins through cross-linking. The biologic consequences and ultimate fate of such complexes depend on their size and Ig isotype content as they fix complement, induce ADCC or inflammation, or are cleared by Fc-mediated mechanisms (34). To fully understand Ab function, we must have an appreciation not only of the structure and function of the reactants in an immune complex, but also the geometry of the more rigid segments and the modes of segmental flexibility.

In our prior study, we investigated the relative flexibility of each of the four human IgG subclasses by direct immunoelectron-microscopic examination. The Fab arms of free molecules were visualized with anti-Id Fab fragments, and the Fab-Fab angles were measured. In addition to determining the mean angle for each IgG type, the SD from the mean was used as an indicator of the relative ability of the molecules to flex up and down. We termed this value the hinge fold flexibility function. In addition, we employed a functional assay to measure the relative flexibility. This assay consisted of mixing Id-bearing IgG molecules with an equal molar ratio of an anti-Id mAb. The resultant small soluble immune complexes were shown to approach an equilibrium for this system when incubated at room temperature for 30 min at 1 μg/ml (1). The geometry of the complexes from each Id/anti-Id combination was scored and compared. The particular negative staining protocol used in these studies is believed to give a fair representation of the types of immune complexes present in solution. In contrast to other procedures in which complexes are dried onto the supporting membrane (while being exposed to increasing concentrations of buffer and/or stain, risking dissociation or, conversely, aggregation), our preparation protocol allows molecules and complexes to spontaneously bind to the membrane while in solution (32, 34, 35).

Using this approach, we were able to show that IgG3 was, for each measurement, more flexible than the other subclasses, followed by IgG1 and, finally, IgG4 and 2. In this study, we apply the same approach to analyze additional members of this Id-matched set: IgM (as monomeric subunits), IgE, IgA2, and a series of genetically engineered IgG3 hinge mutants, in an effort to further dissect the parameter that influences the relationship between the segmental flexibility, the unique structural features of Abs, and the geometry of immune complex formation. Examples of electron micrographs showing Id/anti-Id complexes and free Fab-tagged Ig molecules are shown in Figures 1 and 2, respectively. Our previously reported data for human IgGs 1, 2, 3, and 4 (1) are included in the figures and tables for comparative purposes.

FIGURE 1.

Immunoelectron micrographs of anti-Id 5B5 in complex with IgA2, IgM, IgE, and HM5 (Dob-like). Reactants were mixed at a 1:1 ratio (1 μg/ml each) and incubated 30 min at room temperature. Bar = 200 Å.

FIGURE 1.

Immunoelectron micrographs of anti-Id 5B5 in complex with IgA2, IgM, IgE, and HM5 (Dob-like). Reactants were mixed at a 1:1 ratio (1 μg/ml each) and incubated 30 min at room temperature. Bar = 200 Å.

Close modal

At a given concentration, an equimolar mixture of mutually reactive mAbs will, for thermodynamic reasons, preferentially form ring-shaped complexes having fewest possible components. Thus, for fully flexible molecules engaging in Id/anti-Id interactions, a preponderance of ring dimers is predicted (1, 35, 37, 38, 39). Any inhibition of ring dimerization because of limitations on flexibility and/or by steric interactions will force the equilibrium to shift toward rings composed of four or, if severely restricted, six or even larger multiples of two. Uncomplexed molecules and those composed of linear chains of molecules represent either an imbalance in the Id/anti-Id ratio or unassociated molecules at equilibrium with the complexed forms.

The structure and composition of the hinge region of Ig molecules are generally considered to be most important for segmental flexibility. The amino acid sequence of the various native and mutant hinges is shown in Table I. Wild-type IgG3 has an exceptionally long structural hinge comprised of a 12-aa UH, a 50-aa MH, and an 8-aa LH. Fortuitously, the genetic hinge of IgG3 is encoded by four separate exons encoding, from N to C terminus, a 17-aa segment (of which the N-terminal 12 aa comprise the UH) and three identical 15-aa segments, thus containing a 17-15-15-15 genetic hinge. By deleting one or more of the 15-mer exons, IgG molecules have been produced in which the MH has been incrementally shortened (mutants designated 17-15-15 and 17-15) while maintaining the UH and LH. By deleting the 17-mer segment, the bulk of the MH can be left intact, but the UH is shortened to 4 aa (mutant 15-15-15). Alternatively, both the UH and MH can be shortened (mutant 15). Other variations include removal of all four hinge exons (HM5) (14), or replacement of the natural hinge with a short artificial hinge (Ala-Ala-Ala-Cys-Ala) as in HM4 (13).

As previously observed for this system, more than 90% of the Id and anti-Id molecules form complexes (Fig. 3), and the majority of these (66 to 83%) are in cyclic form. Because all Id-bearing molecules display the same Fv region and are reacting with the same anti-Id molecule, any difference in the size distribution of immune complexes between the various mutant and wild-type molecules will be the result of the structural features, in particular, the flexibility-determining regions of the Id-bearing molecules (1).

FIGURE 3.

Comparison of immune complexes formed between anti-Id, 5B5, and the various Igs and hinge mutants at the 30-min equilibrium time point.

FIGURE 3.

Comparison of immune complexes formed between anti-Id, 5B5, and the various Igs and hinge mutants at the 30-min equilibrium time point.

Close modal

A comparison between native IgG3 and hinge-modified IgG3 mutants shows that those mutants that retain the C-terminal 17-aa hinge segment (17-15-15 and 17-15) are about as capable as the wild-type IgG3 (with its 17-15-15-15 hinge) of engaging in ring dimer formation at the expense of larger ring forms (Fig. 3, Table II). In all three cases, between 41 and 56% of the molecules formed ring dimers, and 18 to 32% formed tetrameric or larger rings. In contrast, the mutants that lack the 17-aa segment (15-15-15 and 15), in which the N-terminal 15-aa segment forms a short UH of 4 aa, both show few ring dimers (3 and 7%) and a majority of ring tetramers and larger ring forms (79 and 55%, respectively). Indeed, both 15-15-15 and 15 mutants show a distribution of complexes similar to that previously reported for IgG2 and 4 (1), the least flexible IgG isotypes. The UH length of the two 17-less mutants (4 aa) falls in size between that of IgG2 (3 aa) and IgG4 (7 aa), suggesting that the relative ability of these IgG molecules to form ring dimers is correlated with the length of the UH. Previous investigators have similarly concluded that UH length and Fab arm flexibility are directly related (36, 40, 41). An alternative interpretation might be that the MH of the 15-15-15 and 15 mutants are not properly folded and that the misfolding is adversely affecting Fab-Fab flexibility in these molecules. This seems unlikely in view of the fact that we have measured the distances between the Fab arms and the Fcs in the mutants and find that they are approximately proportional to the number of amino acid contributing to the MH in each construct (Table III). These data suggest that the gross structure of the truncated hinges is rodlike as in the native IgG3.

Table II.

Comparison of Ig forms

Ig TypeaIgG1IgG2IgG3IgG417-15-1517-1515-15-1515HM4HM5IgA2IgMIgE\E
UH length (aa) 10b 12 7 (5)c 12 12 1c NA NA\E 
% ring dimer (%) 18 47 56 41 2.5 26 0.6 75 26\E 
Mean Fab-Fab angle(o) 117 127 136 128 132 133 123 119 125 106 111 105 141\E 
Hinge fold flexibility functiond ±43 ±32 ±53 ±39 ±48 ±51 ±42 ±40 ±33 ±39 ±34 ±56 ±34 
Ig TypeaIgG1IgG2IgG3IgG417-15-1517-1515-15-1515HM4HM5IgA2IgMIgE\E
UH length (aa) 10b 12 7 (5)c 12 12 1c NA NA\E 
% ring dimer (%) 18 47 56 41 2.5 26 0.6 75 26\E 
Mean Fab-Fab angle(o) 117 127 136 128 132 133 123 119 125 106 111 105 141\E 
Hinge fold flexibility functiond ±43 ±32 ±53 ±39 ±48 ±51 ±42 ±40 ±33 ±39 ±34 ±56 ±34 
a

The data for IgG1, -2, -3, and -4 have previously been published (1).

b

The UH of IgG1 is disulfide-bonded to the light chain at the fourth position, leaving 5 aa between the cysteine and the MH.

c

The two C-terminal aa of IgG4 are proline as is the single UH amino acid of IgA2.

d

The standard deviation of the mean values is taken as a measure of the degree of flexibility (hinge fold flexibility function) about the mean angle.

Table III.

Length of Ig hinges

Hinge compositionLength (Å)MH (aa)UH (aa)\E
17-15-15-15 (IgG3)a 105 50 12\E 
17-15-15 65 35 12\E 
17-15 45 20 12 
15-15-15 75 41 
Hinge compositionLength (Å)MH (aa)UH (aa)\E
17-15-15-15 (IgG3)a 105 50 12\E 
17-15-15 65 35 12\E 
17-15 45 20 12 
15-15-15 75 41 
a

IgG3 data have previously been published (1).

Comparison of the ability of the hinges of these various IgG molecules to engage in the hinge fold flexibility function shows a general concordance between the length of the UH and the degree of variation from the mean Fab-Fab angle (Tables II and IV). However, the mean Fab-Fab angles are not correlative with UH length and are presumably influenced by steric interactions of the two Fab with each other and with the Fc. Because the two Fab arms of each Id-bearing molecule in a ring dimer must be nearly parallel to each other for both CDR tips to interact with those of the anti-Id, one might have anticipated that those molecules having the smallest mean angle (i.e., nearest to the parallel) when unreacted would show the highest rate of ring dimerization. However, this was not the case.

The HM4 mutant has a short artificial hinge composed of a 3-aa UH above a MH consisting of a single Cys engaged in an interchain disulfide bond. Such a short compact hinge structure would be expected to enhance steric interactions of the hinge-proximal regions of the Fab arms with each other and with the Fc, leaving little room for the Fab arms to maneuver into the conformation needed for closure of a ring dimer. Indeed, the hinge-folding flexibility data show HM4 to be one of the least flexible (±33°) (Tables II and IV). In addition, ring dimers represented only 2.5% of the HM4/anti-Id complexes (Table II). These results also serve to confirm that essentially all of the anti-Id molecules in our preparation are hinge intact (i.e., interchain disulfide bonded). Uncoupling hinge peptides through reduction and alkylation of interchain disulfide bonds has previously been shown to foster ring dimer formation between IgGs otherwise incapable of Id/anti-Id complex formation (42).

HM5 (Dob-like) is a hingeless construct in which the light chains are covalently attached to each other rather than to the heavy chains and in which the heavy chains are not disulfide bond cross-linked due to the lack of a genetic hinge. These molecules also yielded a predominance of tetrameric and larger ring forms (45%) (Fig. 3) as seen in the electron micrograph (Fig. 1). Such a pattern is predicted for a molecule that might be expected to be severely constrained in all modes of flexibility. Consequently, we were surprised to observe that 26% of the molecules were actually able to form ring dimers, a percentage considerably higher than that of native IgG1, -2, or -4. Of the wild-type IgGs, only IgG3 produced more ring dimers (47%) (Table II, Fig. 3).

Analysis of the hinge fold flexibility for HM5 provides no obvious explanation for this unexpected result, since its value (±39°) was near the midrange of tested molecules. One possible explanation for the high level of ring dimers is that the tethered light chains hold the Fab arms of HM5 in a position that favors the conformation necessary for ring dimerization. As we and others have previously pointed out, just because a structure is not flexible, it may still engage in interactions associated with flexibility, if, by chance, the molecule is already fixed in a permissive conformation (1, 9). Although this mutant does have the smallest mean angle (105°) of those tested, this is still a long way from parallel. An interesting alternative explanation is that the L-L disulfide bond between the two Fab arms might function as a pivot point allowing a more parallel orientation of the Fab arms, as depicted in Figure 4: there is no H-H disulfide bond to prevent the lateral movement of the proximal base of the Fab arms as would be the case in all other molecules tested. Previous nanosecond polarization (42) and electron-microscopic data (41, 43) have shown that the absence of a disulfide bond to bridge the N termini of the CH2 domains greatly increases Fab-Fab flexibility by allowing the Fc halves to splay apart due to a second hinge at the CH2-CH3 junction (44). The ability of hinge-deleted IgG4 to more readily bind Ag than intact IgG4, as reported by Horgan et al. (45), can probably be similarly explained.

FIGURE 4.

Diagrammatic representation of the proposed relationship between the conformation of hinge deletion mutant HM5 in its Dob-crystal conformation (A) and in its anti-Id-binding conformation (B). To get from conformation A to B, the two Fab arms must flex upward, twist about each other 180° to end up on opposite sides, and rotate 90° axially to accommodate the L-L disulfide bridge. In the process, the CH2 domains of Fc may splay apart to accommodate the tight Fab-Fc connection. Light chains, dark; one heavy chain, stippled; the other heavy chain, white.

FIGURE 4.

Diagrammatic representation of the proposed relationship between the conformation of hinge deletion mutant HM5 in its Dob-crystal conformation (A) and in its anti-Id-binding conformation (B). To get from conformation A to B, the two Fab arms must flex upward, twist about each other 180° to end up on opposite sides, and rotate 90° axially to accommodate the L-L disulfide bridge. In the process, the CH2 domains of Fc may splay apart to accommodate the tight Fab-Fc connection. Light chains, dark; one heavy chain, stippled; the other heavy chain, white.

Close modal

Based on the ability to form ring dimeric immune complexes, IgA2 appears to be the least flexible of any of the native Ig classes, subclasses, or hinge mutants tested. Ring dimers are virtually absent (<1%), with most molecules contributing to either ring tetramers or larger ring complexes (69%) (Figs. 1 and 3, Table II). IgA2 also has a low level of hinge-folding flexibility (±34°) (Tables II and IV). From a structural standpoint, this is perhaps not too surprising. IgA2 has essentially no UH, with only a rigid Pro residue falling between the amino acid generally ascribed to CH1 and the N-terminal Cys of the MH region. It would appear that the amino acids at the CH1-hinge juncture do not provide sufficient flexure to allow the closed angle Fab-Fab orientation needed for ring dimerization. The MH of IgA2 is 8 aa long, proline rich, and thus presumably rigid. The LH is 2 aa long, and probably affords little hinge-Fc flexibility either. Thus, the IgA2 hinge is almost completely devoid of any features that would contribute to independent Fab-Fab or Fab-Fc movement. Such a short rigid hinge may be a tradeoff affording the molecule some protection from proteolytic digestion in the harsher environment following mucosal secretion (47).

IgE has no formal hinge region and might have been expected to dimerize with difficulty. Although the junctures between CH1, CH2, and CH3 may provide some hinge-like function (48), data from various anisotropic techniques show IgE to be considerably less flexible than the IgGs (23, 26, 49, 50) (reviewed in 2 . The fact that 26% of the molecules formed dimers (Figs. 1 and 3, Table II) (versus 4 and 5% for IgG2 and IgG4, respectively) suggests that IgE is either more flexible than predicted or its Fab arms are held in a conformation fairly conducive to dimerization. The general consensus is that the molecule is hook shaped, with the Fc curled back toward the Fab arms, with the Fab arms more parallel (Y-shaped) than divergent (T-shaped), forming a tripod arrangement (22, 25, 26, 51, 52). The average distance between the tips of the Fab arms and the C terminus of the Fc of murine IgE was calculated to be 71 Å, with a range of 63 to 87 Å, necessitating a fairly extreme U-shaped hooking of the Fab and Fc segments (25). When the same technique was applied to human IgG, the average N and C termini were determined to be only slightly further apart (75 Å), but with a much greater range (57–143 Å) than IgE (26). In view of other models based on small angle x-ray scattering and sedimentation studies (10) and electron microscopy studies (1, 53), such folding of the IgG1 arms seems unlikely, suggesting that IgE may not be as hooked as perceived.

Our analysis of IgE, in which the Fab arms are tagged with Fab anti-Id, does show a relatively restricted array of Fab-Fab angles (±34°) in agreement with the physicochemical data. However, we found IgE to display the most laterally protruding arms (141°, most T shaped) of any of the Igs tested (Tables II and IV). Such a strong deviation from the near parallel Fab arm orientation needed for ring dimerization is difficult to reconcile with the substantial level of observed dimerization (26%) and the various data showing restricted flexibility and the hook-shaped model of the molecule in solution (22, 23, 24, 25, 26, 49, 50, 51, 52). Perhaps the average of all of the modes of IgE segmental flexibility is low, but certain modes are more permissive. For example, the Fab arms of IgE may cross the small angle threshold (to allow the formation of ring dimers) more frequently than other Igs having shallower mean angles and a greater flexibility function (SD), but which may be otherwise inhibited, for subtle structural reasons, from crossing the threshold. Thus, flexibility, rather than a more rigid preferential orientation of the Fab arms, would be responsible for the ease with which ring dimers are formed in IgE. Investigations by others using divalent haptens show that ring dimers of murine IgE are permissible, although under strain (54).

With regard to the shape of the molecule, we propose a modification of the hooked model that fits our data, yet may still be consistent with the bulk of the energy transfer and x-ray and neutron-scattering data (22, 25, 26). Rather than having all three segments curved around toward each other, we propose that the Fab arms are pointed in opposite directions and that the Fc is hooked either toward one arm (i.e., one arm very close to the Fc terminus and the other further away) or that the Fc brings the C terminus up near the bases of the Fab arms so that both CDRs are equidistant from the Fc base (Fig. 5). This latter model has the added feature of prepositioning the Fab arms in an orientation that may be more conducive of cross-linking.

FIGURE 5.

IgE models. A, Hook-shaped model in which both Fab arms and the Fc are bent into a “U” configuration (22, 25, 26). B, Proposed model in which the Fc is hooked toward one of two laterally projecting Fab arms. C, Alternative proposed model in which the Fc is hooked toward the Fab-Fc juncture and the Fab arms are laterally projecting. Fab arms depicted above and Fc below.

FIGURE 5.

IgE models. A, Hook-shaped model in which both Fab arms and the Fc are bent into a “U” configuration (22, 25, 26). B, Proposed model in which the Fc is hooked toward one of two laterally projecting Fab arms. C, Alternative proposed model in which the Fc is hooked toward the Fab-Fc juncture and the Fab arms are laterally projecting. Fab arms depicted above and Fc below.

Close modal

It should be noted that Fcs on the IgE molecules were often difficult to discern (as though they were folded under the rest of the molecule), but that we did not observe obvious hooking of the Fc up toward the paired Fab arms when IgE was in complex or when bound to Fab anti-Id. Such details can easily be altered due to a flattening of the molecules and complexes as they adhere to the carbon membrane. A hooked Fc is presumed to be a byproduct of the need for asymmetry in the Fc to insure monovalent interaction with FcεRI (22).

The human mutant IgM (IgM C575S) molecule, secreted as monomeric subunits (29), also proved dramatically more capable of forming ring dimers than anticipated. In fact, it formed the highest percentage of ring dimers of any Ig (75%), a level far above IgG3 (47%), the Ig molecule generally viewed as being the most flexible (Fig. 1, Tables II and IV). As with IgE, the anisotropy data for IgM (including monomeric forms) show relative segmental inflexibility (27, 28). On the other hand, electron microscopic analyses reveal pronounced flexing at CH1-CH2 and CH2-CH3 switch regions in the pentameric form when binding planar Ag arrays, i.e., the molecule can bend into the so-called staple conformation (4). Our hinge-folding data show IgM to have a Fab-Fab fold flexibility (±56°) slightly greater than even IgG3 (±53°) (Tables II and IV). The mean Fab-Fab angle of IgM (106°) is considerably less than that of IgG3 (136°) and about equal to the smallest angle of all of the tested molecules (HM5 at 105°). Together, the data suggest that IgM Fab arms are capable of as broad an array of angles as IgG3, but with an average centered around a more closed angle, i.e., closer to the near parallel orientation needed for dimeric ring closure.

Perhaps the slower anisotropy decay values in the literature for IgM and IgE (2) stem from the added mass of the extra domain pair (CH2), as compared with hinge peptide segments, together with the short CH1-CH2 and CH2-CH3 peptide junctures. Either independently or together, these properties might serve to dampen the rate of motion, but not the range of motion. The observation that these two similarly constituted molecules fall near opposite ends of the Fab-Fab mean angle and hinge fold flexibility rank orders lists (Table IV) suggests that there might be considerable differences in the association of the pairing of the CH2 domains at the base of the Fab arms and in the CH2-CH3 and/or CH1-CH2 junctures.

Table IV.

Rank order of the mean Fab-Fab angles and hinge fold flexibility function

Rank Order: Mean Fab-Fab Angle (°)aRank Order: Hinge Fold Flexibility Function (SDb in °)
IgE 141 IgM ±56\E 
IgG3 136 IgG3 ±53\E 
17-15 133 17-15 ±51 
17-15-15 132 17-15-15 ±48\E 
IgG4 128 IgG1 ±43\E 
IgG2 127 15-15-15 ±42\E 
HM4 125 15 ±40\E 
15-15-15 123 IgG4 ±39\E 
15 119 HM5 ±39\E 
IgG1 117 IgA2 ±34\E 
IgA2 111 IgE ±34\E 
IgM 106 HM4 ±33\E 
HM5 105 IG2 ±32 
Rank Order: Mean Fab-Fab Angle (°)aRank Order: Hinge Fold Flexibility Function (SDb in °)
IgE 141 IgM ±56\E 
IgG3 136 IgG3 ±53\E 
17-15 133 17-15 ±51 
17-15-15 132 17-15-15 ±48\E 
IgG4 128 IgG1 ±43\E 
IgG2 127 15-15-15 ±42\E 
HM4 125 15 ±40\E 
15-15-15 123 IgG4 ±39\E 
15 119 HM5 ±39\E 
IgG1 117 IgA2 ±34\E 
IgA2 111 IgE ±34\E 
IgM 106 HM4 ±33\E 
HM5 105 IG2 ±32 
a

The data for IgG1, -2, -3, and -4 have previously been published (1).

b

The standard deviation of the mean values is taken as a measure of the degree of flexibility (hinge fold flexibility function) about the mean angle.

Various investigations have drawn a correlation between the length of the UH and flexibility (36, 40, 41). This model is appealing since it is reasoned that the MH is proline rich and bounded by cysteines, effectively excluding it from any substantial contribution to at least Fab-Fab flexibility and probably Fab-Fc flexibility as well. That the UH length-flexibility relationship may also be an oversimplification derives from the work of Schneider et al. (41), in which the composition of the CH1 domain was shown to exert a strong modulating influence on the flexibility of otherwise identical IgGs. Our data are only partly consistent with the UH length-flexibility relationship. As shown in Table II and plotted in Figure 6, it seems clear that molecules with a long unrestrained (12-aa) UH (IgG3, 17-15-15 and 17-15) are more than twice as likely to be able to form the conformationally restrictive ring dimers with anti-Id in our test system than could any of the shorter hinge forms. At the other end of the spectrum, several of the short-hinged (3–4 aa) molecules (IgG2, HM4, 15-15-15, and 15) form few dimeric rings (3 to 7%). Indeed, IgA2, the Ig with the shortest UH (a single proline residue), was virtually unable to form ring dimers (<1%).

FIGURE 6.

Plot of UH length versus percentage of the molecules in ring dimer immune complexes.

FIGURE 6.

Plot of UH length versus percentage of the molecules in ring dimer immune complexes.

Close modal

In contrast to the above pattern, several of the molecules clearly deviated from the expected. For example, IgG1 (18% ring dimer) has a 10-aa UH and might have been expected to be almost as able to dimerize as IgG3 and its 12-aa UH allies. The observation that it is less than half as capable might be attributed to the Cys residue in the middle of the peptide segment that covalently binds to the Cys at the C terminus of the light chain. This extra tether might serve to shorten the effective flexible length of the IgG1 UH to 5 aa (Asp Lys Thr His Thr). The crystal structure of this 5-aa segment shows an α-helix (7), but this structure is probably a result of crystal-packing forces (55). In comparison, the 4-aa UH molecules (15-15-15 and 15) form only 3 to 7% ring dimers, considerably less than the 18% of IgG1. We speculate that the additional N-terminal 5 aa of the UH of IgG1 (Glu Pro Lys Ser Cys), in conjunction with the C-terminal amino acid tail of the light chain to which it is disulfide bonded, also contribute some degree of flexibility of this molecule.

Another apparent exception to the pattern is IgG4, in which its 7-aa hinge allows no more ring dimer formation (5%) than the 3- or 4-aa UH molecules (3–7%). The explanation for this apparent anomaly may be that two prolines are situated just N terminus of the N-terminal Cys of the MH. Not only would these prolines serve to shorten the flexible part of the chain to 5 aa, but they could conceivably orient the protruding chains in a direction unfavorable for ring closure.

Together, our results suggest caution when using a concept as multifaceted as flexibility to predict or explain functional aspects of Ig molecules based on data from physicochemical techniques. For example, the hingeless IgG1 mutants, Dob and Mcg, were previously assumed to be rather rigid T-shaped molecules based on x-ray crystallography data (18, 20, 45, 56). In contrast, other molecules that can crystallize in two or more structurally different forms are said to be flexible. But, as we have shown, hingeless light chain-tethered IgG3 molecules must be capable of exhibiting considerable flexibility to form both ring dimers (requiring near parallel arm) and the more open tetramers and larger ring forms. Moreover, these molecules display a range of Fab-Fab angles typical of hinge-intact Ig molecules. The term flexibility, as used to interpret crystallography, really indicates only that there is more than one static permissive orientation. As another example, polarization and spin-label studies show that IgM and IgE are relatively inflexible as compared with IgG. In this context, flexibility means the ability to jiggle (rapidly change their position and/or orientation), but gives no clue as to the relative orientation or manner of motion. As we have shown, these hingeless molecules can easily form ring dimers and can display a variety of Fab-Fab angles at levels greater than some IgGs. As pointed out above, IgM and IgE Fab arms may jiggle less rapidly, but may have a greater range of motion. A clear distinction should be made between crystal-packing flexibility (static permitted orientations), dynamic flexibility (movement through space over time) of molecules in solvent and what we might call dynamic permitted orientations derived from analyzing “snapshots” of individual molecules displaying a range of forms (Fab-Fab angular measurements), or competing to obtain a sterically restrictive conformation (i.e., ring dimers).

FIGURE 2.

Immunoelectron micrographs of Fab anti-Id in complex with IgM, IgE, and HM4. A total of 2 μg of Fab was reacted with 1 μg of each Id-bearing Ig and incubated 30 min at room temperature. Bar = 200 Å.

FIGURE 2.

Immunoelectron micrographs of Fab anti-Id in complex with IgM, IgE, and HM4. A total of 2 μg of Fab was reacted with 1 μg of each Id-bearing Ig and incubated 30 min at room temperature. Bar = 200 Å.

Close modal

We thank Kimberly Riddle and Randi H. Sandin for their excellent technical assistance, and Vigdis Sorensen for constructing the IgM C575S mutant.

1

This work was supported by National Science Foundation Grant MCB-9304790 (K.H.R.).

3

Abbreviations used in this paper: aa, amino acid; CDR, complementarity-determining region; LH, lower hinge; MH, middle hinge; NIP, nitroiodophenyl; NP, nitrophenyl; UH, upper hinge.

1
Roux, K. H., L. Strelets, T. E. Michaelsen.
1997
. Molecular flexibility of human IgG subclasses.
J. Immunol.
159
:
3372
2
Nezlin, R..
1990
. Internal movement in Ig molecules.
Adv. Immunol.
48
:
1
3
Burton, D. R..
1985
. IgG: functional sites.
Mol. Immunol.
22
:
161
4
Feinstein, A., N. Richardson, M. J. Taussig.
1986
. Ig flexibility in complement activation.
Immunol. Today
7
:
169
5
Michaelsen, T. E., B. Frangione, E. C. Franklin.
1977
. Primary structure of the “hinge” region of human IgG3.
J. Biol. Chem.
252
:
883
6
Huck, S., P. Fort, D. H. Crawford, M.-P. Lefranc, G. Lefranc.
1986
. Sequence of human γ 3 heavy chain constant region: comparison with the other human C genes.
Nucleic Acids Res.
14
:
1779
7
Marquart, M., J. Deisenhofer, R. Huber.
1980
. Crystallographic refinement and atomic models of the intact Ig molecule Kol and its antigen-binding fragment at 3.0 and 1.9 Å resolution.
J. Mol. Biol.
141
:
369
8
Endo, S., Y. Arata.
1985
. Proton nuclear magnetic resonance study of human Igs G1 and their proteolytic fragments: structure of the hinge region and effects of a hinge-region deletion on internal flexibility.
Biochemistry
24
:
1561
9
Phillips, M. L., M.-H. Tao, S. L. Morrison, V. N. Schumaker.
1994
. Human/mouse chimeric mAbs with human IgG1, IgG2, IgG3 and IgG4 constant domains: electron microscopic and hydrodynamic characterization.
Mol. Immunol.
31
:
1201
10
Gregory, L., K. G. Davis, B. Sheth, J. Boyd, R. Jefferis, C. Nave, D. R. Burton.
1987
. The solution conformations of the subclasses of human IgG deduced from sedimentation and small angle x-ray scattering studies.
Mol. Immunol.
24
:
821
11
Pumphrey, R..
1986
. Computer models of the human Igs: shape and segmental flexibility.
Immunol. Today
7
:
174
12
Sjoberg, B., E. Rosenqvist, T. Michaelsen, S. Pap, R. Osterberg.
1980
. The solution shapes of IgG3 Ig and its Fch and Fc fragments.
Biochim. Biophys. Acta
625
:
10
13
Brekke, O. H., T. E. Michaelsen, R. H. Sandin, I. Sandlie.
1996
. Activation of complement by an IgG molecule without a genetic hinge: correction.
Nature
383
:
103
14
Michaelsen, T. E., O. H. Brekke, A. Aase, R. H. Sandin, B. Bremnes, I. Sandlie.
1994
. One disulfide bond in front of the second heavy chain constant region is necessary and sufficient for effector functions of human IgG3 without a genetic hinge.
Proc. Natl. Acad. Sci. USA
91
:
9243
15
Brekke, O. H., T. E. Michaelsen, R. Sandin, I. Sandlie.
1993
. Activation of complement by an IgG molecule without a genetic hinge.
Nature
363
:
628
16
Sandlie, I., A. Aase, C. Westby, T. E. Michaelsen.
1989
. Clq binding to chimeric monoclonal IgG3 Abs consisting of mouse variable regions and human constant regions with shortened hinge containing 15 to 47 amino acids.
Eur. J. Immunol.
19
:
1599
17
Michaelsen, T. E., A. Aase, C. Westby, I. Sandlie.
1990
. Enhancement of complement activation and cytolysis of human IgG3 by deletion of hinge exons.
Scand. J. Immunol.
32
:
517
18
Silverton, E. W., M. D. D. Navia.
1977
. Three-dimensional structure of an intact human Ig.
Proc. Natl. Acad. Sci. USA
74
:
5140
19
Klein, M., N. Haeffner Cavaillon, D. E. Isenman, C. Rivat, M. A. Navia, D. R. Davies, K. J. Dorrington.
1981
. Expression of biologic effector functions by IgG molecules lacking the hinge region.
Proc. Natl. Acad. Sci. USA
78
:
524
20
Guddat, L. W., J. N. Herron, A. B. Edmundson.
1993
. Three-dimensional structure of a human Ig with a hinge deletion.
Proc. Natl. Acad. Sci. USA
90
:
4271
21
Sykulev, Y. K., R. Nezlin, G. P. German, E. V. Chernokhvostova, V. V. Lavrentiev.
1984
. Structural studies of human IgA1 and IgA2 Igs tagged with two different spin labels.
Biofizika
29
:
744
22
Beavil, A. J., R. J. Young, B. J. Sutton.
1995
. Bent domain structure of recombinant human IgE-Fc in solution by x-ray and neutron scattering in conjunction with an automated curve fitting procedure.
Biochemistry
34
:
14449
23
Oi, V. T., T. M. Vuong, R. Hardy, H. Reidler, J. Dangle, L. A. Herzenberg, L. Stryer.
1984
. Correlation between segmental flexibility and effector function of Abs.
Nature
307
:
136
24
Holowka, D., T. Wensel, B. Baird.
1990
. A nano-second fluorescence depolarization study on the segmental flexibility of receptor-bound IgE.
Biochemistry
29
:
4607
25
Zheng, Y., B. Shopes, D. Holowka.
1991
. Conformations of IgE bound to its receptor FcεRI and in solution.
Biochemistry
30
:
9125
26
Zheng, Y., B. Shopes, D. Holowka, B. Baird.
1992
. Dynamic conformations compared for IgE and IgG1 in solution and bound to receptors.
Biochemistry
31
:
7446
27
Dudich, E. I., R. S. Nezlin, F. Franek.
1980
. Fluorescence polarization and spin-label study of human myeloma IgA and IgM: presence of segmental flexibility.
Mol. Immunol.
17
:
1335
28
Zagyansky, Y. A., L. A. Tumerman, A. M. Egorov.
1972
. Segmental flexibility of IgM molecules.
Immunochemistry
9
:
91
29
Sorensen, V., I. B. Rasmussen, L. Norderhaug, I. Natvig, T. E. Michaelsen, I. Sandle.
1996
. Effect of the IgM and IgA secretory tailpieces on polymerization and secretion of IgM and IgG.
J. Immunol.
156
:
2858
30
Johnstone, A., R. Thorpe.
1982
.
Immunochemistry in Practice
298
Blackwell Scientific Publications, Boston.
31
Roux, K. H..
1989
. Immunoelectron microscopy of Id-anti-Id complexes.
Methods Enzymol.
178
:
130
32
Roux, K. H..
1996
. Negative stain immunoelectron microscopic analysis of small macromolecules of immunologic significance.
Methods
10
:
247
33
Roux, K. H., N. S. Greenspan.
1994
. Monitoring the formation of soluble immune complexes composed of idiotype and anti-idiotype antibodies by electron microscopy.
Mol. Immunol.
31
:
599
34
Ravetch, J. W., R. A. Clynes.
1998
. Divergent roles for Fc receptors and complement in vivo.
Annu. Rev. Immunol.
16
:
421
35
Phillips, M. L., V. T. Oi, V. N. Schumaker.
1990
. Electron microscopic study of ring-shaped, bivalent hapten, bivalent antidansyl mAb complexes with identical variable domains but IgG1, IgG2a and IgG2b constant domains.
Mol. Immunol.
27
:
181
36
Schumaker, V. N., M. L. Phillips, D. C. Hanson.
1991
. Dynamic aspects of Ab structure.
Mol. Immunol.
28
:
1347
37
Moyle, W. R., C. Lin, R. L. Corson.
1983
. Quantitative explanation for increased affinity shown by mixtures of mAbs: importance of a circular complex.
Mol. Immunol.
20
:
439
38
Schumaker, V. N., G. W. Seegan, C. A. Smith, S. K. Ma, J. D. Rodwell, M. F. Schumaker.
1980
. The free energy of angular position of the Fab arms of IgG Ab.
Mol. Immunol.
17
:
413
39
Murphy, R. M., R. A. Chamberlin, P. Schurtenberger, C. K. Colton, M. L. Yarmush.
1990
. Size and structure of antigen-antibody complexes: thermodynamic parameters.
Biochemistry
29
:
10889
40
Dangl, J. L., T. G. Wensel, S. L. Morrison, L. Stryer, L. A. Herzenberg, V. T. Oi.
1988
. Segmental flexibility and complement fixation of genetically engineered chimeric human, rabbit and mouse Abs.
EMBO J.
7
:
1989
41
Schneider, W. P., T. G. Wensel, L. Stryer, V. T. Oi.
1988
. Genetically engineered Igs reveal structural features controlling segmental flexibility.
Proc. Natl. Acad. Sci. USA
85
:
2509
42
Roux, K. H., D. L. Tankersley.
1990
. A view of the human idiotypic repertoire: electron microscopic and immunologic analyses of spontaneous Id-anti-Id dimers in pooled human IgG.
J. Immunol.
144
:
1387
43
Chan, L. M., R. E. Cathou.
1977
. The role of the interheavy chain disulfide bond in modulating the flexibility of IgA Ab.
J. Mol. Biol.
112
:
653
44
Seegan, G. W., C. S. V. Smith.
1979
. Changes in quaternary structure of IgG upon reduction of the interheavy-chain disulfide bond.
Proc. Natl. Acad. Sci. USA
76
:
907
45
Edmundson, A. B., L. W. Guddat, R. A. Rosauer, K. N. Andersen, L. Shan, Z.-C. Fan.
1995
. Three-dimensional aspects of IgG structure and function.
Antibodies
1
:
41
46
Horgan, C., K. Brown, S. H. Pincus.
1993
. Studies on antigen binding by intact and hinge-deleted chimeric Abs.
J. Immunol.
150
:
5400
47
Plaut, A. G., J. V. Gilbert, M. S. Artenstein, J. D. Capra.
1975
.
Neisseria gonorrhoeae and Neisseria meningitidis: extracellular enzyme cleaves human IgA. Science
190
:
1103
48
Helm, B. A., Y. Ling, C. Tale.
1991
. The nature and importance of the inter-ε chain disulfide bonds in human IgE.
Eur. J. Immunol.
21
:
1543
49
Nezlin, R., Y. A. Zagyansky, A. I. Kaivarainen, D. V. Stefani.
1973
. Properties of myeloma IgE(Yu): chemical, fluorescence polarization and spin-labeled studies.
Immunochemistry
10
:
681
50
Cathou, R. E..
1978
. Solution conformation and segmental flexibility of Igs.
Contemp. Topics Immunol.
5
:
37
51
Burton, D. R..
1990
. Ab: the flexible adaptor molecule.
Trends Biochem. Sci.
15
:
64
52
Keown, M. B., R. Ghirlando, R. J. Young, A. J. Beavil, S. J. Perkins, B. J. Sutton, H. J. Gould.
1995
. Hydrodynamic studies of a complex between the Fc fragment of human IgE and a soluble fragment of the FcεRI α chain.
Proc. Natl. Acad. Sci. USA
92
:
1841
53
Phillips, M. L., M.-H. Tao, S. L. Morrison, V. N. Schumaker.
1994
. Human/mouse chimeric mAbs with human IgG1, IgG2, IgG3 and IgG4 constant domains: electron microscopic and hydrodynamic characterization.
Mol. Immunol.
31
:
1201
54
Subramanian, K., D. Holowka, B. Baird, B. Goldstein.
1996
. The Fc segment of IgE influences the kinetics of dissociation of a symmetrical bivalent ligand from cyclic dimeric complexes.
Biochemistry
35
:
5518
55
Padlan, E. A..
1994
. Anatomy of the Ab molecule.
Mol. Immunol.
31
:
169
56
Ely, K. P., P. M. Colman, E. E. Abola, A. C. Hess, D. S. Peabody, D. M. Parr, G. E. Connell, C. A. Laschinger, A. B. Edmundson.
1978
. Mobile Fc region in the Zie IgG2 cryoglobulin: comparison of crystals of the F(ab′)2 fragment and the intact Ig.
Biochemistry
17
:
820