Antibody interfaces revealed through structural mining

Abstract

Antibodies are fundamental effectors of humoral immunity, and have become a highly successful class of therapeutics. There is increasing evidence that antibodies utilize transient homotypic interactions to enhance function, and elucidation of such interactions can provide insights into their biology and new opportunities for their optimization as drugs. Yet the transitory nature of weak interactions makes them difficult to investigate. Capitalizing on their rich structural data and high conservation, we have characterized all the ways that antibody fragment antigen-binding (Fab) regions interact crystallographically. This approach led to the discovery of previously unrealized interfaces between antibodies. While diverse interactions exist, β-sheet dimers and variable-constant elbow dimers are recurrent motifs. Disulfide engineering enabled interactions to be trapped and investigated structurally and functionally, providing experimental validation of the interfaces and illustrating their potential for optimization. This work provides first insight into previously undiscovered oligomeric interactions between antibodies, and enables new opportunities for their biotherapeutic optimization.

Keywords: Antibody; Cluster; Crystallography; Homotypic; Interface; Oligomer; Protein data bank; Structure.

Graphical abstract

Fig. 1

Antibody Fab contacts and Fab-Fab interfaces revealed from informatic PDB analysis. (a) Computational flow diagram for analyzing inter-Fab interfaces. Fab X-ray structures were universally renumbered and the asymmetric unit was expanded to 30 Å². Symmetric oligomers were extracted, and all interacting residue pairs were identified using a distance cutoff of either 4 or 6 Å between any two inter-Fab residues. Similarity between a pair of interfaces was calculated as the Jaccard index or weighted Jaccard index between the sets of interface residue positions within each interface. Hierarchical clustering was performed based on the similarities calculated to result in a dendrogram of Fab interfaces. (b) Recurrent packing Fab-Fab interfaces throughout the collective Fab PDB in order of decreasing prevalence. Prevalence reflects an incidence measure of each interface that is unbiased by the presence of multiple structures of the same Fab sequence in the PDB. The inset provides the nonredundant (NR) size and prevalence values for the 6-most prevalent clusters. (c) Percentage of PDBs within each of the 42 most prevalent clusters that include antigen in the structure (red, left axis), and number of distinct space groups among the PDBs within each cluster (blue, right axis). The plot shows that the lack of cluster bias due the presence of antigen or crystallographic lattice. For example, the six most prevalent interfaces were observed in structures where Fab/antigen complexes make up 63 %, 57 %, 72 %, 41 %, 59 %, and 41 % of the cluster respectively, and that crystallized in 30, 17, 9, 11, 13, and 10 different space groups respectively. (d) Mean buried surface area and (e) number of H-bonds for the 42 most prevalent interfaces, in order of decreasing prevalence. Error bars represent standard deviations. Upper and lower dashed lines correspond to averages for obligate and transient protein complexes respectively . Data in (b-e) are vertically aligned and thus there is correspondence with labeled columns in (b). Numeric values for prevalence and PISA results are provided in Supplementary Table S1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Representative structures of the 42 most prevalent Fab-Fab interfaces. Domains are colored as follows: VH (light blue), CH1 (dark blue), VL (pink), CL (dark red). The labels for each interface include the cluster rank based on prevalence, interface name based on structurally central residue (Kabat numbering for Fv region, EU number for constant regions), nonredundant prevalence percentage, and mean buried surface area. Interfaces are ranked numerically based on prevalence, with equivalent prevalence being designated with arbitrary alphabetic qualifiers; for example, there are three interfaces that occur in 17 PDB entries for prevalence = 1.7 %, ranked as 9a, 9b, and 9c. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

β-sheet dimers are commonly observed Fab oligomers. (a) Structural similarity between sheet-extended Fab oligomers that interact as CH1-CH1 homodimers (top left), CH1-CL heterodimers (top right), CL-CL homodimers (bottom left), VH-VH homodimers (bottom middle), and VL-VL homodimers (bottom right). Domains are colored as follows: VH (light blue), CH1 (dark blue), VL (pink), CL (dark red). The labels for each interface include the cluster rank based on prevalence, interface name based on structurally central residue, nonredundant prevalence percentage, and mean buried surface area. (b) Fab β-sheet dimers are structurally similar across cluster members. Interface residues of 10 cluster members were superimposed for Fab oligomers within the CH1-211 homodimer (left), CL-211 heterodimer (middle), and CL-205 homodimer (right). RMSD values were calculated that reflect alignment of Cα atoms within stacked CH1 and/or CL domains for the 10 representative cluster members depicted. For example, for CH1-211 both CH1 domains of one structure were together aligned to both CH1 domains of another structure. Calculations were run in Pymol without an outlier cutoff. The reported RMSD is an average of the matrix produced from aligning 10 structures to each other (CH1-211: CH1-CH1, 1.6 Å; CL-211: CH1-CL, 0.9 Å; CL-205: CL-CL, 2.3 Å). (c) Constant domain superposition of CH1-CH1 homodimer with CH1-CL heterodimer (left) and CL-CL homodimer (right). β-sheet residues at the Fab-Fab dimer interface were superimposed for the structural representatives of clusters CH1-211 and CL-211 (left), or CH1-211 and CL-205 (right). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Interface profiles of the six most prevalent clusters. For all interfaces, the upper left-most pie chart provides the sequence distribution of cluster members based on species: human (blue, H), mouse (red, M), and other (e.g., rat, rabbit, etc., green). The lower left-most pie chart provides the sequence distribution of cluster members based on light chain type: kappa (blue, κ), and lambda (red, λ). The four right-most pie charts provide the sequence distribution of human VH (H VH), human VL (H VL), mouse VH (M VH), and mouse VL (M VL) subgroups. The labels above the pie charts in the upper left profile (CH1-211) are not repeated in the other charts for visual simplicity. The %ID plot on the right provides the intra (red) versus inter (black) cluster sequence identity for both the entire variable region (Fv) as well as only those residues at the interface as shown in the sequence logo and Supplementary Table S2. For the Fv, these values reflect the mean pairwise identities for all VH and VL sequences within the cluster (intra) versus the mean pairwise identities between each member of the cluster and all other members of all other clusters (inter). A similar comparison is made for each interface, where %ID reflects the mean pairwise identity of the residues at the interface of a given cluster aligned with those same residues for each member of the cluster (intra) versus all other members of all other clusters (inter). The sequence logo at the bottom provides weighted sequence composition at interface residues for members of the designated cluster, with numbering according to Kabat and EU conventions for the Fv and constant regions, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5

Fab dimers can be trapped with engineered disulfides. (a) Representative SEC chromatograms of select cysteine variant and WT anti-Her2 Fabs after affinity chromatography. Peaks are labeled as monomer (M), dimer (D), and higher-order (HO). (b) Summary of monomer (green diamonds), dimer (blue circles), and higher order (red squares) species for all cysteine variants by SEC post-affinity chromatography. Data represent percentage of species based on integration of SEC peaks. Multiple symbols for a given cluster represent multiple designed cysteine variants. The identities of all variants and associated numeric SEC data are provided in Supplementary Tables S3-S5. (c) Correlation between expression and in vitro coupling data for disulfide-trapped dimers. (d) Summary of data from in vitro coupling of selected cysteine variant Fabs. The data reflect the % monomer (green diamonds), dimer (blue circles), and higher-order (red squares) species after assembly for the designed and mismatched (MM) variants run as negative controls. For MM variants the pairings of HCs and LCs with single cysteine mutations were scrambled in four separate combinations. The identities of all variants and associated numeric data are provided in Supplementary Table S6. (e) OX40 receptor activation by variant and WT Fab and full-length IgG antibodies based on an NFκB luciferase reporter assay. VH-16 represents Fab cysteine variant VH(S113C) / CH1(G178C) and CH1-176 represents Fab cysteine variant VH(P14C) / CL(D151C). Fold receptor activation represents normalized RLU relative to cells alone. For the IgG1 versions, +XL and -XL indicate the presence or absence of secondary crosslinker antibody, respectively. RGY represents a triple Fc variant that promotes IgG hexamer formation. No crosslinker was used for the RGY version or the Fab monomer or dimer versions. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6

Disulfide-trapped Fab dimers affirm conformations of PDB mining. All Fabs comprise the variable region of the anti-Her2 antibody trastuzumab. Fab structures represent the β-sheet-extended interface cluster CL-205 [variant CL(S202C) / CL(S208C)] (a) and the elbow region clusters CH1-207 [variant CH1(S119C) / CH1(G122C)] (b) and VL-108 [variant VL(G16C) / CL(D170C)] (c). Fab domain colors match those in Fig. 2, with the informatically-mined structure at 0% transparency (left) and the experimentally-determined structure at 30% transparency (right). Sulfur atoms participating in disulfide bonds are shown in yellow. Overlays of the interface (middle) for (a) and (c) were configured by aligning the α carbons of the Fab dimers from each structure, whereas the overlay for (b) used a single Fab for alignment (see Results). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)