Structure and dynamics of GAD65 in complex with an autoimmune polyendocrine syndrome type 2-associated autoantibody

Abstract

The enzyme glutamate decarboxylase (GAD) produces the neurotransmitter GABA, using pyridoxal-5'-phosphate (PLP). GAD exists as two isoforms, GAD65 and GAD67. Only GAD65 acts as a major autoantigen, frequently implicated in type 1 diabetes and other autoimmune diseases. Here we characterize the structure and dynamics of GAD65 and its interaction with the autoimmune polyendocrine syndrome type 2-associated autoantibody b96.11. Using hydrogen-deuterium exchange mass spectrometry (HDX), X-ray crystallography, cryo-electron microscopy, and computational approaches, we examine the conformational dynamics of apo- and holoGAD65 and the GAD65-autoantibody complex. HDX reveals local dynamics accompanying autoinactivation, with the catalytic loop promoting collective motions at the CTD-PLP domain interface. In the GAD65-b96.11 complex, heavy chain CDRs dominate the interaction, with a long CDRH3 bridging the GAD65 dimer via electrostatic interactions with the ₂₆₀PEVKEK₂₆₅motif. This bridging links structural elements controlling GAD65's conformational flexibility to its autoantigenicity. Thus, intrinsic dynamics, rather than sequence differences within epitopes, appear to be responsible for the contrasting autoantigenicities of GAD65 and GAD67. Our findings elucidate the structural and dynamic factors that govern the varying autoantibody reactivities of GAD65 and GAD67, offering a revised rationale for the autoimmune response to GAD65.

Fig. 1. HDX-MS analyses of GAD65.

A Representations of GAD65 sequence and dimeric structure. Left, cartoon; middle/right, orthogonal views of molecular surface. The N–terminal (NTD, blue), pyridoxal-5’-phosphate (PLP)–binding (PLP, cyan) and C-terminal (CTD, green) domains are labelled. Monomer A is coloured slightly lighter than monomer B. Within the active sites, the K396–PLP Schiff-base is shown as spheres. The catalytic loop (CL) that forms a flap over the active site of an adjacent monomer in trans is coloured red in each monomer. Further sequence annotation is shown in Supplementary Fig. 1. B Sequence coverage map of peptides used to monitor HDX in GAD65 in the presence and absence of co-factor PLP or Fab b96.11. The 77 identified peptic peptides cover 94.5% of the N-terminally truncated GAD65, with a redundancy of 3.12. Peptides with unchanged HDX between apo- and holo-forms are shown in black, and peptides with significantly reduced HDX upon both PLP and Fab b96.11 binding are shown in orange. Peptides with significantly reduced HDX solely upon PLP binding are shown in red. C Representative mass spectra of the peptide spanning residues 373–407 (m/z = 668.7, z = 6) that comprise the active site residue K396 to which PLP is covalently bound in the holo-state. D HDX profiles for representative peptides of GAD65 plotted as a function of labelling time (15 s, 1 min, 10 min, 1 h, 6 h, maximally labelled (MX) control). Grey curves illustrate the deuterium uptake for apoGAD65, red curves for holoGAD65 and orange curves for apoGAD65 in the presence of Fab b96.11, respectively. Values are the mean of three independent measurements (n = 3). Standard deviations are plotted as error bars and are in some instances too small to be visible. From left to right, HDX profiles are shown of segments of the N-terminal hinge region, the PLP-binding domain including residues spanning the ₂₆₀PEVKEK₂₆₅ sequence (H7) and residues partially covering the CL.

Fig. 2. Conformational dynamics of holoGAD65 and apoGAD65.

A Plots of the differences in the average deuterium uptake (ΔHDX) between different GAD65 states for the 77 identified peptides at the five sampled time-points. The individual peptides are arranged along the x-axis from the N- to the C-terminus of GAD65. Positive and negative values indicate reduced and increased HDX, respectively. Values represent means of independent measurements. The dotted lines mark the threshold for a significant difference in HDX, which corresponds to the 98% confidence interval based on triplicate measurements (Supplementary Table 1). Peptides covering the catalytic loop (417–435), or K396 are marked in dark red or dark blue. Peptide regions with active side residues, functional residues, or none are marked in red, blue or grey, respectively; B HDX of holoGAD65 (top) and apoGAD65 (bottom) at 15 s, 1 min, 10 min, 1 h and 6 h labelling time points are shown mapped on the model structure of apoGAD65 and holoGAD65. All HDX data are normalized to the maximally labelled control. The normalized deuterium uptake is defined by the colour code, with blue indicating low HDX and red high HDX. Regions with no HDX information are shown in grey.

Fig. 3. Differences in HDX between GAD65 states upon binding b96.11.

Plots of the differences in the average deuterium uptake (ΔHDX) between different GAD65 states for the 77 identified peptides at the five sampled time points. The individual peptides are arranged along the x-axis starting from the N- to the C-terminus. Positive and negative values indicate reduced and increased HDX, respectively. Values represent means of independent measurements. The dotted lines mark the threshold for a significant difference in HDX, which corresponds to the 98% confidence interval based on replicate measurements (Supplementary Table 1). ΔHDX of A apoGAD65 in absence and presence of Fab b96.11 and B holoGAD65 in absence and presence of Fab b96.11. The N–terminal (NTD, blue), PLP–binding (PLP, cyan) and C-terminal (CTD, green) domains are labelled. Peptides covering the ₂₆₀PEVKEK₂₆₅ motif, the CL (417–435), or K396 are marked in purple, dark red or dark blue. C HDX of apoGAD65 in absence and presence of Fab b96.11 (top) and holoGAD65 in absence and presence of Fab b96.11 (bottom) at 15 s, 1 min, 10 min, 1 h and 6 h labelling time points are shown mapped on the model structure of apoGAD65 and holoGAD65. All HDX data are normalized to the maximally labelled control. The normalized deuterium uptake is defined by the colour code, with blue indicating low HDX and red high HDX. Regions with no HDX information are shown in green.

Fig. 4. Structure of a GAD65-b96.11 complex.

A Cryo-EM map coloured according to local resolution from highest (blue) to lowest (red). Local resolution estimation of the cryo-EM map was performed with ResMap. B Structure of apoGAD65–b96.11 complex in orthogonal views. C HoloGAD65 cartoon coloured according to ΔHDX of holoGAD65 in absence and presence of Fab b96.11 HDX at 6 h labelling time point; D Close-up of interactions at the holoGAD65-b96.11 interface, with HDX colouring, showing residues in the ₂₆₀PEVKEK₂₆₅ motif on helix α7. E GAD65 molecular surface and bound b96.11. CDRs are coloured separately and labelled. F Close-up of interface showing polar interactions between GAD65 and b96.11. Helix α7/₂₆₀PEVKEK₂₆₅ shown in yellow. Residues are shown as sticks. Hydrogen-bonds are indicated by broken lines.

Fig. 5. Structural shifts upon complexation and comparison with GAD67.

A Structural alignment of GAD65-b96.11 complex (colouring as other figures) with uncomplexed GAD65 (2OKK, grey). Structures were aligned using the NTD (RMSD = 0.78 Å), showing significant rigid-body shifts of PLP and CTD, resulting in slight opening of dimer around helix α9, creating space for H1 and H3 CDRs of b96.11. B Alignment of bound b96.11 Fv region (blue=VH, wheat=VL) with free Fv (grey). Residues at the GAD-Fv interface that shift upon binding are shown as sticks. C Orthogonal representations of complex interface after alignment of GAD65-b96.11 with GAD67 (2OKJ, grey). Similarities and differences between side chains of GAD65 and GAD67 at the interface are shown as sticks and labelled (e.g., E/T denotes E in GAD65 and T in GAD67). ₂₆₀PEVKEK₂₆₅ residues in GAD65 and PEVKTK residues in GAD67 are shown in yellow and grey respectively. Other interface residues in GAD65 and GAD67 are shown in cyan/red and grey, respectively.