An unfolded CH1 domain controls the assembly and secretion of IgG antibodies

Abstract

A prerequisite for antibody secretion and function is their assembly into a defined quaternary structure, composed of two heavy and two light chains for IgG. Unassembled heavy chains are actively retained in the endoplasmic reticulum (ER). Here, we show that the C(H)1 domain of the heavy chain is intrinsically disordered in vitro, which sets it apart from other antibody domains. It folds only upon interaction with the light-chain C(L) domain. Structure formation proceeds via a trapped intermediate and can be accelerated by the ER-specific peptidyl-prolyl isomerase cyclophilin B. The molecular chaperone BiP recognizes incompletely folded states of the C(H)1 domain and competes for binding to the C(L) domain. In vivo experiments demonstrate that requirements identified for folding the C(H)1 domain in vitro, including association with a folded C(L) domain and isomerization of a conserved proline residue, are essential for antibody assembly and secretion in the cell.

Figure 1. Structural characteristics of the C_H1 domain and its assembly mechanism with the C_L domain

(A) shows a schematic representation of the IgG1 molecule. The IgG molecule is composed of two heavy chains (blue) and two light chains (green). The heavy chain consists of the V_H, C_H1, C_H2 and C_H3 domain and the light chain of the V_L and the C_L domain (listing from N- to C-terminus of each chain). The C_H2 domains of the heavy chains are glycosylated with complex biantennary oligosaccharides (depicted in grey). Each domain possesses an internal disulfide bridge (omitted for clarity) and additional disulfide bonds link the two heavy chains in the flexible hinge region. A single disulfide bond covalently connects C_H1 with the C_L domain. The two identical antigen binding sites (paratopes) are made up by the two variable domains V_H and V_L. The overall IgG molecule can be divided into two Fab fragments (composed of V_H, C_H1, V_L and C_L) and one Fc fragment (composed of two C_H2 and two C_H3 domains). (B) The isolated C_L domain (cyan) displays a typical all-β far-UV CD spectrum whereas the isolated C_H1 domain (blue) shows a random coil spectrum. To assess if C_L induces structural changes in C_H1, the spectrum of co-incubated C_L and C_H1 was recorded (green). From this spectrum and the far-UV CD spectrum of the isolated C_L domain, the spectrum of the C_H1 domain in the presence of C_L was calculated (red) which shows the characteristics of β-sheet secondary structure. (C) The affinity between C_H1 and C_L was determined by the change in the intrinsic fluorescence upon C_L induced folding of the C_H1 domain, recorded before and after a 4 h equilibration step. A one-site binding model was used to fit the data. The inset shows a representative single exponential trace observed after the addition of 1 µM C_L to 2 µM C_H1. A single exponential reaction with a very similar rate was observed by far-UV CD spectroscopy (D, red trace). The folding reaction could be accelerated by the PPIase CypB (red trace: 10 µM C_L and 10 µM C_H1 alone, blue trace: in the presence of 0.75 µM CypB). The inset shows the dependence of the slow reaction on CypB concentration. The observed rate in the absence of CypB is denoted as k₀, in the presence of CypB as k_cat. If 1µM CypB was inhibited by 2 µM cyclosporine A, no acceleration was observed (black cross). (E) Association of the C_H1 domain with a lucifer yellow labeled C_L domain was followed by the change in the lucifer yellow anisotropy signal. The observed rate constants were fitted with a linear function to yield the k_on value and the k_off value of 0.007 ±0.0002 µM⁻¹ min⁻¹ respectively 0.1 ±0.01 min⁻¹. The inset shows individual single exponential traces after the addition of 0 µM (black), 5 µM (blue), 10 µM (green) and 20 µM (red) C_H1 to 1 µM labeled C_L. (F) To assess the formation of the C_L/C_H1 interchain disulfide bridge, non-reducing SDS-PAGE experiments were carried out and the dimer band intensity was quantified. 25 µM of each domain were used. In the absence of CypB (red), a time constant of τ = 63 ±7 min was observed for the formation of covalent C_L/C_H1 dimers. In the presence of 5 µM CypB, a time constant of τ = 31 ±5 min was obtained. In (G), the overall C_H1/C_L assembly mechanism is shown. Only after formation of the internal disulfide bridge in the C_H1 domain (blue), the fast formation of a dimeric intermediate with the C_L domain (green) is observed. Subsequently, prolyl isomerization limits complete folding and formation of the interchain disulfide bridge. All measurements were carried out at 25°C in PBS.

Figure 2. NMR spectroscopic characterization of C_L induced C_H1 folding

(A) ¹⁵N-¹H HSQC spectra of the isolated C_H1 domain (red) and the assigned C_H1 domain in complex with the C_L domain (blue) are shown. In order to characterize the folding pathway of the intrinsically disordered C_H1 domain, time dependent HSQC intensities upon addition of unlabeled C_L to ¹⁵N labeled C_H1 were measured for each assigned residue at the native chemical shift position and fitted by a single exponential function. Two representative traces for Val68 (red) and Lys90 (blue) are shown in the inset. (B) Initial amplitudes for each assigned C_H1 residue were derived from the fitted exponential functions. Residues with an initial HSQC amplitude below a threshold of 25% native intensity are colored in blue and residues above the threshold in red (open bars: residues in loop regions / filled bars: residues in structured regions). Errors indicate standard deviations from single exponential fits. In (C) C_H1 residues with intensities above the threshold in the intermediate are mapped on the crystal structure of the C_H1/C_L dimer (pdb code 1Q9K). The dimerization interface between C_H1 (grey) and C_L (blue) is shown on the left with only residues of C_H1 indicated that are involved in this interaction and above the HSQC amplitude threshold. On the right, internal C_H1 residues above the 25% threshold are shown in red, the three cis proline residues in C_H1 are depicted and labeled in blue. (D) The HSQC spectra of the C_H1 Pro32Ala, Pro34Ala and Pro74Ala mutants show, that only the Pro32Ala mutant is not able to fold in the presence of C_L anymore (blue spectrum). For the other two mutants, Pro34Ala and Pro74Ala, well dispersed spectra and hence folding are observed in the presence of C_L (blue spectra). In the absence of C_L (red spectra), all three mutants show typical HSQC spectra of unfolded proteins. All measurements were carried out in PBS at 25°C except for the folding kinetics which where recorded at 12.5°C.

Figure 3. Characterization of the interaction between BiP and C_H1 in vitro

(A) The affinity between BiP and oxidized C_H1 (filled circles, straight line) as well as reduced C_H1 (open circles, dashed line) was determined by analytical HPLC experiments. The data were fitted to a one-site binding model to determine the K_d. (B) The association kinetics between 1 µM BiP and varying concentrations of oxidized C_H1 (filled circles) and reduced C_H1 (open circles) were measured to determine the rate constants of the reaction. For oxidized C_H1, k_on = 0.00026 ±0.00002 µM⁻¹ min⁻¹ and k_off = 0.0050 ±0.0002 min⁻¹ were obtained. For the reduced C_H1 domain, the corresponding values were k_on = 0.00041 ±0.00003 µM⁻¹ min⁻¹ and k_off = 0.0047 ±0.0003 min⁻¹. The left inset shows single HPLC runs of 8 µM oxidized C_H1 and 1 µM BiP after 10 min (blue) and 200 min (red) co-incubation. The right inset shows the overall observed single exponential association kinetics between 1 µM BiP and 8 µM oxidized C_H1.

Figure 4. The folding status of an antibody domain controls IgG secretion in vivo

(A) COS-1 cells were co-transfected with vectors encoding BiP and either a wild type light chain (LC_wt), a light chain containing the C_H1 domain instead of the C_L domain (LC_CH1), or a light chain containing an unfolded C_L domain (LC_CLmut). Cells were metabolically labeled for 3 h and both cell lysates (no subscript) and culture supernatants (subscript m) were immunoprecipitated with the indicated antisera (Ab). Precipitated proteins were separated by SDS-PAGE under reducing conditions and visualized by autoradiography. (B) COS-1 cells were co-transfected as in (A) except that a vector encoding a chimeric humanized mouse heavy chain was also included. Cells were labeled and analyzed as in (A). (C) COS-1 cells were co-transfected with vectors encoding BiP, a Flag-tagged truncated heavy chain consisting of only the V_H and C_H1 domains (Lee et al., 1999), and with the indicated light chain constructs (i.e., λ, wild type κ, or C_L mutant κ). Cells were metabolically labeled and both cell lysates (L) and culture supernatants (M) were immunoprecipitated with the anti-Flag antibody. Precipitated proteins were separated by SDS-PAGE under non-reducing condition (except the first lane, which included 2-ME in the sample buffer and is indicated as red) and visualized by autoradiography. Mobilities of completely reduced (ox₀), partially oxidized (ox₁), and fully oxidized (ox₂) forms of truncated heavy chain, as well as those of λ and κ light chains are indicated. The tagged forms of the κ light chain constructs co-migrate with the ox1 form of the truncated heavy chain.

Figure 5. The isomerization of a single proline residue controls the assembly and secretion of heavy chains in vivo

COS-1 cells were co-transfected with vectors encoding wild type MAK33 heavy chain (WT) or one of the three Pro to Ala mutants (P74A, P32A, or P34A) together with wild type light chain and BiP. Cells were metabolically labeled and both cell lysates (no subscript) and culture supernatants (subscript m) were immunoprecipitated with the indicated antisera (Ab). Precipitated proteins were separated by SDS-PAGE under non-reducing conditions and visualized by autoradiography. (B) The cell lysates and culture supernatants from (A) were divided in half, immunoprecipitated with the indicated antibodies, analyzed by SDS-PAGE under reducing conditions, and visualized with autoradiography. These data demonstrate that the failure of the P32A mutant to induce assembly and secretion is not because it is expressed poorly, as the signal for this mutant heavy chain is very similar to that of the wild type heavy chain.

Figure 6. A model for the overall IgG secretion control mechanism

A schematic indicating the possible pathways for the C_H1 domain (blue), its folding and assembly in association with C_L (green) and BiP (grey) is shown. C_H1 has to form its internal disulfide bridge and to be released from BiP before it can associate with C_L. In vivo, these processes are tightly coupled and thus cannot be dissected kinetically. Prior to complete folding and irreversible formation of the C_L/C_H1 interchain disulfide bridge, the proline residue 32 has to isomerize from trans to cis. The isomerization reaction can be accelerated by CyclophilinB. All rate constants were determined at 25°C.