Skip to main content
NCBI home page
Protein Science : A Publication of the Protein Society logoLink to Protein Science : A Publication of the Protein Society
. 2024 Jun 26;33(7):e5081. doi: 10.1002/pro.5081

Antigen‐binding fragments with improved crystal lattice packing and enhanced conformational flexibility at the elbow region as crystallization chaperones

Heather A Bruce 1, Alexander U Singer 1, Levi L Blazer 1, Khanh Luu 1, Lynda Ploder 1, Alevtina Pavlenco 1, Igor Kurinov 2, Jarrett J Adams 1, Sachdev S Sidhu 1,✉
PMCID: PMC11201802  PMID: 38924648

Abstract

It has been shown previously that a set of three modifications—termed S1, Crystal Kappa, and elbow—act synergistically to improve the crystallizability of an antigen‐binding fragment (Fab) framework. Here, we prepared a phage‐displayed library and performed crystallization screenings to identify additional substitutions—located near the heavy‐chain elbow region—which cooperate with the S1, Crystal Kappa, and elbow modifications to increase expression and improve crystallizability of the Fab framework even further. One substitution (K141Q) supports the signature Crystal Kappa‐mediated Fab:Fab crystal lattice packing interaction. Another substitution (E172G) improves the compatibility of the elbow modification with the Fab framework by alleviating some of the strain incurred by the shortened and bulkier elbow linker region. A third substitution (F170W) generates a split‐Fab conformation, resulting in a powerful crystal lattice packing interaction comprising the biological interaction interface between the variable heavy and light chain domains. In sum, we have used K141Q, E172G, and F170W substitutions—which complement the S1, Crystal Kappa, and elbow modifications—to generate a set of highly crystallizable Fab frameworks that can be used as chaperones to enable facile elucidation of Fab:antigen complex structures by x‐ray crystallography.

Keywords: crystal lattice packing interactions, phage display technology, protein domain swapping, surface entropy reduction, x‐ray crystallography

1. INTRODUCTION

X‐ray crystallography can be used to obtain high‐resolution structural information of protein:protein interaction interfaces. For antibody:antigen complexes, this information is highly valuable as it can be used to identify key residues in the epitope and paratope (respective binding sites in the antigen and antibody) which are important for promoting affinity and specificity, characterize the binding mode, elucidate the mechanism of action, and guide engineering of improved antibodies for therapeutic applications (Bostrom et al., 2011; Fellouse et al., 2004, 2005; Lee et al., 2017; Martyn et al., 2022; Miersch et al., 2021; Schaefer et al., 2011; Tao et al., 2019; Ye et al., 2008). Due to a flexible hinge region present within the immunoglobulin G (IgG) quaternary structure (Fellouse & Sidhu, n.d.; Boulot et al., 1988; Power & Bates, 2019; Vidarsson et al., 2014), crystallization of this full‐length form, alone or in complex with an antigen, presents a significant challenge. Consequently, the compact antigen‐binding fragment (Fab) is commonly used as it contains the antigen‐binding site, yet presents a well‐structured heterodimer of ideal size for crystallography (Fellouse & Sidhu, n.d.; Canaves et al., 2004). Nonetheless, many Fab:antigen complexes remain recalcitrant to crystallization, posing a major bottleneck in both basic and applied antibody research (Boulot et al., 1988; Chiu et al., 2019; Ereño‐Orbea et al., 2018; Ye et al., 2008).

Previously, we employed the surface entropy reduction (SER) strategy (Cooper et al., 2007; Derewenda, 2010; Derewenda & Vekilov, 2006; Garrard et al., 2001; Males & Davies, 2019) with phage‐displayed libraries (Fellouse & Sidhu, n.d.) to identify a small group of surface substitutions, collectively termed S1, which improved the crystallizability of a commonly used human Fab framework (Bruce et al., 2023). The S1 substitutions display a synergistic relationship with two other alterations—referred to as the Crystal Kappa (Lieu et al., 2020) and elbow (Bailey et al., 2018) substitutions—to facilitate crystallization of Fab:antigen complexes. Through the incorporation of S1, Crystal Kappa (C), and elbow (E) substitutions into the Fab framework (FabS1CE) (Figure 1a), the crystallization success rate was improved dramatically, and provided us with the tools to begin developing a platform for the facile crystallization of Fab:antigen complexes (Bruce et al., 2023).

FIGURE 1.

FIGURE 1

Open in a new tab

Fab library design and results. (a) FabS1CE‐F1 (PDB entry 8T7I shown as a representative structure) (Bruce et al., 2023) with the heavy chain and light chain colored light blue or gray, respectively. Constant and variable domains of the light chain (CL/VL) and heavy chain (CH/VH) are indicated. In addition, residue positions in key regions shown as spheres are colored as follows: S1 (dark blue), Crystal Kappa (green), elbow (magenta), KGP site (cyan), FPE site (orange), GLY site (yellow). (b) The principal Fab molecule heavy chain is colored light blue, while the packing Fab molecule heavy chain and light chain are colored dark blue or green, respectively. The principle Fab light chain is omitted. (Left panel) In the crystal lattice, the Crystal Kappa mediated β‐sheet stacking interaction (Lieu et al., 2020) occurs between the CH domain (from the principal Fab in this representation) and CL′ domain of the packing Fab molecule (indicated by prime symbol). (Middle panel) Due to their proximity, the Crystal Kappa substitution (green) in the packing Fab, and the elbow substitution (magenta) in the principle Fab, form a contiguous crystal lattice packing interface. Residues Phe136H, Asn137H, Gln138H, and Ile139H of the elbow region form weak intermolecular interactions with residues Thr155H′, Ser156H′, and Gly157H′ (light blue) in the CH domain of the packing Fab molecule. The KGP site, which directly follows the elbow region, is colored cyan. (Right panel) Rotating the packing Fab molecules in the crystal lattice by 180° reveals a sub‐optimal crystal lattice packing site; residues Lys141H, Gly142H, and Pro143H of the KGP site do not participate in the Fab:Fab packing interaction at the Crystal Kappa:elbow junction. The flexible side chain of Lys141H projects away from the crystal lattice packing site toward the solvent. (c) Sequences of Fab variants selected from the phage‐displayed library, The WT sequence (S1CE) is shown at the top and variant sequences (S1CE1–7) are shown below with dashes indicating identities. The number of times each variant was observed amongst 94 sequenced clones (N) is shown to the right of each sequence, followed by the protein yield from transient transfection of mammalian cell cultures, and the melting temperature (T m ) determined by differential scanning fluorimetry.

The Crystal Kappa substitution facilitates crystallization by introducing a β‐sheet stacking interaction between the CL and heavy‐chain constant (CH) domains of packing Fab molecules in the crystal lattice (Figure 1b—left panel) (Lieu et al., 2020). Meanwhile, replacing five small residues in the elbow region connecting the heavy‐chain variable (VH) and CH domains with four bulkier residues reduces Fab conformational flexibility and thus lowers the entropic cost of crystallization (Bailey et al., 2018). While these two modifications to the Fab framework play dominant and predictable roles in facilitating crystallization, we found that the S1 substitution adapts to the requirements of different crystal lattice formations by providing a more favorable packing site in the CL domain compared with the wild‐type Fab (FabWT). In this way, the S1 substitution cooperates with and complements the Crystal Kappa and elbow substitutions to enhance crystallization (Bruce et al., 2023).

Previously, we found that the Crystal Kappa β‐sheet stacking interaction also induces various weak intermolecular interactions to form nearby, between residues in the heavy‐chain elbow region of the principle Fab molecule and residues in the CH domain of the packing Fab molecule in the crystal lattice (Figure 1b, middle panel) (Bruce et al., 2023). Due to the proximity of the Crystal Kappa and elbow substitutions on adjacent packing Fab molecules, a contiguous interaction site composed of the Crystal Kappa packing site and residues at the Crystal Kappa:elbow junction forms and stabilizes this mode of Fab:Fab crystal lattice packing (Bruce et al., 2023; Lieu et al., 2020). However, as the elbow substitution was generated independently (Bailey et al., 2018) of the Crystal Kappa (Lieu et al., 2020) substitution, the intermolecular interactions at the Crystal Kappa:elbow junction are suboptimal, with several residues in prime positions not participating in the crystal lattice packing interaction (Figure 1b, right panel) and sections of the packing Fab CH domain even left unresolved from the electron density in some structures (Bruce et al., 2023). Furthermore, incorporation of the elbow substitution substantially reduces the yield of recombinant Fab protein produced from bacterial or mammalian cell cultures (Bailey et al., 2018; Bruce et al., 2023). Thus, while the Crystal Kappa and elbow modifications to the Fab framework enhance crystallization, there is still room for improvement.

Here, we used a phage‐displayed combinatorial library to explore alternative sequences in selected sites near the heavy‐chain elbow region in the FabS1CE framework with the aim to optimize the signature Fab:Fab crystal lattice packing interaction induced by the Crystal Kappa:elbow substitution combination, and consequently, to further enhance crystallizability. We identified substitutions that enhance crystallization and yield of the FabS1CE framework. Structural analyses revealed that one substitution enhanced crystal packing and another improved the compatibility of the elbow substitution with the Fab framework. Unexpectedly, a third substitution dramatically altered crystal packing by increasing the conformational flexibility at the light‐chain and heavy‐chain elbow regions.

2. RESULTS

2.1. Selection and characterization of Fab variants

We aimed to further enhance crystallizability of the FabS1CE framework by supporting the simultaneous incorporation of Crystal Kappa and elbow substitutions and improves the compatibility of the latter substitution with the Fab framework. To this end, we focused on altering three distinct sites, each composed of three residues in the CH domain (Figure 1a). We chose these sites based on their proximity to the elbow region and/or their location near the crystal lattice packing site at the Crystal Kappa:elbow junction (Bruce et al., 2023). The “KGP” site (Lys141H, Gly142H, and Pro143H) is located directly downstream of the elbow substitutions (Phe136H, Asn137H, Gln138H, and ILe139H) (Superscript “H” and “L” indicate residues in the heavy or light chain, respectively). The “FPE” site (Phe170H, Pro171H, and Glu172H) comprises a loop turn, connecting the second and third β‐sheet strands of the CH domain, and the side chains of Phe170H and Pro171H participate in a network of weak intramolecular interactions with residues in the elbow region and the VH domain (Bailey et al., 2018; Stanfield et al., 2006). Finally, residues in the “GLY” site (Gly198H, Leu199H, and Tyr200H) reside at the start of the fifth β‐sheet strand of the CH domain and pack against the FPE site.

We employed a phage‐displayed combinatorial library (Fellouse & Sidhu, n.d.) in which the nine positions across the three sites were simultaneously mutated using a “soft randomization” strategy (Frei & Lai, 2016) that allowed for ~50% wildtype (WT) sequence and 50% random substitutions at each position (Figure S1). The library was constructed in the context of a Fab (EPR‐1), which we have selected from phage‐displayed libraries and engineered to bind to the human erythropoietin receptor (EPOR) (manuscript in preparation). Although phage‐displayed antibody libraries are usually used to select for novel paratopes with desirable binding properties, we have shown that they can also be used to select for substitutions outside the paratope that do not impact antigen recognition, but rather, improve protein stability and/or yield (Barthelemy et al., 2008). Thus, although phage display cannot select directly for crystallizability, we applied it here as an initial screen to enrich for Fabs with optimized frameworks but unaltered paratopes, which could subsequently be screened directly for crystallizability.

Phage pools representing the library were subjected to binding selections for the extracellular domain (ECD) of EPOR. We isolated 94 individual clones that bound strongly to the EPOR by phage ELISA (data not shown), and DNA sequencing revealed that these represented seven unique variants (E1–E7, Figure 1c). No substitutions were observed in the GLY site, and we speculate that this region may be intolerant to changes due its close packing in the protein core (Figure 1a). In contrast, each of the seven variants contained either a F170W or E172G substitution, and the KGP site contained substitutions in most cases.

Since the FPE site packs directly against the elbow region, we reasoned that the substitutions F170W and E172G may affect the compatibility of the elbow substitution with the Fab framework. In addition, we speculated that the K141Q substitution may improve the Fab:Fab intermolecular interactions upon crystal lattice formation due to its prime location at the crystal lattice packing site at the Crystal Kappa:elbow junction, and the nature of the change from a long, charged side chain to a shorter polar side chain—in line with the SER strategy (Cooper et al., 2007; Czepas et al., 2004; Derewenda & Vekilov, 2006; Garrard et al., 2001; Longenecker et al., 2001; Males & Davies, 2019). Thus, although other substitutions in the KGP site (E5–E7) could also be explored, we focused our detailed studies on the four frameworks (E1–E4) that combined the K141Q substitution with either the E172G or F170W substitution, as follows: FabS1CE1 (E172G), FabS1CE2 (E172G/K141Q), FabS1CE3 (F170W), and FabS1CE4 (F170W/K141Q) (Figure S2).

Following expression in mammalian cells, we purified the four Fab‐EPR‐1 variants alongside the reference framework FabS1CE for comparison. The variants all exhibited striking (~10‐fold) increases in yield relative to FabS1CE, while differential scanning fluorimetry (DSF) showed a slight reduction in T m , with the F170W and E172G substitutions reducing T m by up to ~6°C or ~3°C, respectively (Figures 1c and S3). Denaturing polyacrylamide gel electrophoresis (PAGE) under non‐reducing conditions confirmed that all the Fab protein samples were highly pure, contained an interchain disulfide bond connecting the light and heavy chains as expected, and did not exhibit any degradation (Figure S4A). We conducted native PAGE in parallel to investigate the solution state of the Fab variants (Figure S4B). As with the original FabS1CE framework, the new variants FabS1CE1 and FabS1CE2 (both containing substitution E172G) ran as single species with no aggregation or oligomerization detected, and with expected relative migration positions based on their isoelectric points. FabS1CE3 and FabS1CE4 (both containing substitution F170W) ran as predominantly single species, also with their migration positions relative to the other Fabs as expected, and with no aggregation observed. However, a faint smear was observed above the main FabS1CE3 and FabS1CE4 protein bands, indicating that a minor species—likely representing a higher oligomeric state—coexists with the major species (Figure S4B).

2.2. The K141Q substitution enhances crystallizability of the FabS1CE framework

We decided to investigate the impact that the K141Q, F170W, and E172G substitutions have on crystallizability of the FabS1CE framework. To assess crystallizability, we carried out a crystallization screening of Fab‐EPR‐1 with each of the four frameworks (S1CE1–4) in addition to the original framework (S1CE) and recorded the number of conditions that generated crystal hits for each variant (Excel document in Data S1). Utilizing the commercial broad condition screen JCSG‐plus HT‐96 Eco and the PEG/ion screen PACT Premiere HT‐96 (Molecular Dimensions), we sampled a total of 192 crystallization conditions. Being already highly crystallizable (Bruce et al., 2023), the FabS1CE framework facilitated crystallization well, with 23% of conditions generating crystal hits (Figure 2a). The FabS1CE1 and FabS1CE3 frameworks, containing an E172G or F170W substitution, respectively, performed similarly well, with each variant also yielding hits in 23% of conditions. Notably, the addition of the K141Q substitution in combination with E172G (FabS1CE2) or F170W (FabS1CE4) resulted in striking increases in the number of crystal hits, with 35% or 41% of conditions generating crystals, respectively (Figure 2a). Thus, in the context of apo‐Fab‐EPR‐1, the E172G and F170W substitutions do not appear to alter the high crystallizability of the FabS1CE framework, while incorporating the K141Q substitution enhances crystallizability even further.

FIGURE 2.

FIGURE 2

Open in a new tab

Crystal screening results for Fabs and Fab:antigen complexes. Crystal hits for FabS1CE‐EPR‐1 with substitution E172G or F170W, with or without K141Q, are plotted for (a) the apo Fab‐EPR‐1 and (b) the Fab‐EPR‐1 in complex with EPOR‐ECD. The numbers of crystal hits (left y‐axis) or % of crystal hits (right y‐axis) obtained for each Fab variant (x‐axis) are shown from screening a total of 192 conditions (96 conditions each from JCSG+Eco and PACT). Putative apo Fab crystal hits (gray) in the Fab:antigen complex crystallization screening were evaluated by comparing crystal morphology for each complex drop to the corresponding apo Fab drop (see Excel document in Data S1).

2.3. The K141Q, E172G, and F170W substitutions can alter Fab quaternary structure and crystal lattice packing

To understand how substitutions K141Q, F170W, and E172G influence crystallization of the FabS1CE framework, we solved the structures of Fab‐EPR‐1 with each of the four frameworks (S1CE1–4), in addition to the reference framework (S1CE) for comparison. All structures were solved from diffraction studies on good‐sized, discrete, pickable crystals directly harvested from the broad screens, with resolutions ranging from 1.7 to 3.3 Å (Figure S5, Excel document in Data S1, and Crystallography Data Table S1).

As expected, and observed previously (Bruce et al., 2023; Lieu et al., 2020), the Crystal Kappa and S1 substitutions facilitated crystallization by mediating their signature crystal lattice packing interactions (Figure S6). While the typical Fab quaternary structure was observed in crystal structures of FabS1CE and variant FabS1CE2 (Figure 3a, i,iii), it was substantially altered in the crystal structures captured for variants FabS1CE1, FabS1CE3, and FabS1CE4 (Figure 3a, ii–vi). Specifically, the VH:VL interface forms between packing Fab molecules in the crystal lattice, as opposed to within the Fab molecule itself (Figure 3b). Notably, in each of these “split‐Fabs,” the VH:VL packing units observed in the crystal lattice structures superpose well with the Fv region in the FabS1CE structure (Table S1). Only minor structural changes in the complementary determining regions (CDRs) were detected (due to their flexibility and crystal lattice packing influences), but otherwise, the structure of the Fv region was unperturbed despite the domain swapping. The structures of the CH and CL domains were also unaltered, with the constant domain in the ASUs of all the split‐Fab structures superposing well with the constant domain of the FabS1CE structure (Table S1).

FIGURE 3.

FIGURE 3

Open in a new tab

Substitutions K141Q, F170W, and E172G alter FabS1CE‐EPR‐1 structural conformation and crystal lattice packing. (a) Crystal lattice packing arrangement of Fab‐EPR‐1 with the following frameworks: (i) FabS1CE (orthorhombic crystal system, space group P212121), (ii) FabS1CE1 (trigonal crystal system, space group P3121), (iii) FabS1CE2 (trigonal crystal system, space group P32), (iv) FabS1CE3 (trigonal crystal system, space group P3121), (v) FabS1CE3 (orthorhombic crystal system, space group P212121), and (vi) FabS1CE4 (triclinic crystal system, space group P1). The lower panel displays the asymmetric unit (ASU). Fab light and heavy chains are colored light blue or gray, respectively. The upper panel shows the ASU, in addition to the packing Fab molecules of selected symmetry mates in the crystal lattices with those light and heavy chains colored green or dark blue, respectively. Substitutions specific to each Fab variant are underscored. (b) In the FabS1CE1, FabS1CE3, and FabS1CE4 crystal lattice structures, the canonical VH:VL domain interaction occurs between packing Fab molecules in the crystal lattice as opposed to within the Fab molecule itself. The light and heavy chains of one Fab molecule are colored gray or light blue, respectively, while the light and heavy chains of selected packing Fabs are colored green or dark blue, respectively. Packing Fab molecule domains outside of the principal ASU are indicated with a prime symbol. Packing Fab molecule domains within the principal ASU are underscored. (i) In the FabS1CE1‐EPR‐1 or FabS1CE3‐EPR‐1 crystal structures (both P3121 space group), the Fab in the principal ASU interacts with a Fab in another ASU through VH:VL association in the crystal lattice to form a closed dimer arrangement (see 3aii and 3aiv). (ii) In the FabS1CE3‐EPR‐1 structure (P212121 space group), an open‐ended crystal lattice structural arrangement forms in which the principal ASU Fab molecule interacts with two different packing Fab molecules through VH:VL association, priming the formation of an interconnected lattice array (see 3av). (iii) In the FabS1CE4‐EPR‐1 structure (see 3avi), a single ASU contains two Fab molecules (colored gray/light blue or green/dark blue) that associate through their respective pairings of VH:VL domains to form a closed dimer formation, distinct from that found in the other split‐Fab structures.

2.4. The F170W substitution induces a split‐Fab conformation that can crystallize in a variety of different packing arrangements

In Fab and Fab:antigen complex structures containing the WT heavy chain elbow sequence, a so‐called “molecular ball‐and‐socket joint” forms, composed of weak intramolecular interactions between various residues in the VH and CH domains (Bailey et al., 2018; Stanfield et al., 2006). Specifically, the side chains of residues Leu12H and Thr134H in the VH domain along with Ser136H of the elbow region form the socket, while the side chains of residues Phe170H and Pro171H of the FPE site in the CH domain constitute the ball (Stanfield et al., 2006) (Figure 4a, FabC‐F1 (PDB entry 8T7G) used as a representative structure (Bruce et al., 2023)). These interactions between the CH and VH domains limit movement between the Fv and constant domain regions and stabilize the overall Fab conformation. However, as highlighted by the wide range of elbow angles (120°–220°) observed in the large pool of Fab structures (containing the WT heavy chain elbow sequence) deposited in the Protein Data Bank, significant flexibility still exists between the Fv region and constant domain region (Figure 4b, left panel) (Stanfield et al., 2006).

FIGURE 4.

FIGURE 4

Open in a new tab

Molecular details of the incorporation of substitution F170W into the FabS1CE framework. Residues in the elbow site (WT sequence or elbow substitution sequence) are colored magenta. Residues in the FPE site and “molecular socket” site (Stanfield et al., 2006) are colored orange or blue, respectively. The principle Fab heavy and light chains are colored light blue or gray, respectively, whilst the packing Fab light chain is colored green. Fab domains from a packing ASU (in FabS1CE3 structures) are indicated with a prime symbol. Packing Fab domains within the principal ASU (in FabS1CE4 structure) are underscored. (a) In Fab structures with WT elbow sequence, residue Ser136H, along with residues Leu12H and Thr134H of the VH domain, mediate weak intramolecular interactions with residues Phe170H and Pro171H of the FPE site in the CH domain (FabC‐F1 structure used for representation (Bruce et al., 2023) (PDB entry 8T7G)). (b) The elbow substitution reduces Fab conformational flexibility. (Left panel) Superposition of the Fab Fv (VH/VL) region from two representative structures that contain the WT elbow sequence (PDB entries 1BBD and 1PLG, in dark red or pink, respectively, and with elbow angles of 127.0° or 189.5°, respectively). (Right panel) Superposition of the Fv regions from two different Fab ASU molecules from a structure containing the elbow substitution (PDB entry 6AZ2, with elbow angles of 166.6° or 188.0°, respectively, colored dark blue or light blue, respectively). (c) In the FabS1CE structure, the side chains of Phe170H and Pro171H form Van der Waals interactions with the residues in the elbow region and the VH domain (Leu12H and Thr134H). (d–f) Incorporating the F170W substitution results in destabilization at the CH:VH interface; the bulky Trp170H side chain cannot be accommodated in the limited space, resulting in a split Fab conformation for structures: (d) FabS1CE3 (space group P3121), (e) FabS1CE3 (space group P212121), and (f) FabS1CE4 (space group P1).

By shortening the linker region connecting the VH and CH domains, the heavy chain elbow substitution reduces conformational flexibility (Bailey et al., 2018). In addition, the bulkier side chains of residues in the elbow substitution region further restrict mobility between the Fv and constant domains and effectively lock in the position of the elbow “joint” and reduce the attainable Fab elbow angle range (164°–186°) (Figure 4b, right panel) (Bailey et al., 2018). This effect can be observed in the FabS1CE‐EPR‐1 structure, where residues Phe170H and Pro171H of the FPE site pack tightly against residues Phe136H, Asn137H, Gln138H, and Ile139H of the elbow region and residues Leu12H and Thr134H in the VH domain. These interactions leave little room for movement and result in a Fab elbow angle of 172° (Figure 4c and Table 1) (Bailey et al., 2018; Bruce et al., 2023).

TABLE 1.

Elbow angles, space groups, PDB codes for Fab/Fab:antigen complex structures presented or used in this study.

Fab/Fab:antigen complex Mutations Crystal system Space group ASU contents ASU Fab/s elbow angle/s (°) as applicable PDB accession code
Fab‐F1 C Monoclinic C2 1 Fab 144.6 8T7G
S1CE Tetragonal P42212 1 Fab 171.8 8T7I
Fab‐EPR‐1 S1CE Orthorhombic P212121 1 Fab 172.3 8VTP
S1CE1 Trigonal P3121 1 Fab Split‐Fab 8VUA
S1CE2 Trigonal P32 2 Fabs 172.8, 172.4 8VVO
S1CE3 Orthorhombic P212121 1 Fab Split‐Fab 8VTR
Trigonal P3121 1 Fab Split‐Fab 8VU1
S1CE4 Triclinic P1 2 Fabs Split‐Fab 8VU4
Fab‐EPR‐1:EPOR‐ECD S1CE Orthorhombic P212121 1 Fab:antigen complex 173.5 8VUI
S1CE1 Orthorhombic P212121 1 Fab:antigen complex plus 1 Fab 167.8, 173.8 8VVM
S1CE2 Orthorhombic P212121 1 Fab:antigen complex plus 1 Fab 167.0, 174.0 8VUC
Open in a new tab

As observed in the crystal structures obtained with FabS1CE3 and FabS1CE4 frameworks, introducing the F170W substitution consistently resulted in a split‐Fab conformation (Figure 3a, iv–vi). This is because, when used in conjunction with the elbow substitution, the bulky side chain of Trp170H cannot be accommodated in the reduced space between the CH and VH domains, and this ultimately forces a separation of the VH and VL domains and the split‐Fab quaternary structural state (Figure 4d–f). Interestingly, the split‐Fab conformation can crystallize in various packing arrangements (Figure 3b), presumably due to the greatly enhanced flexibility at the heavy‐ and light‐chain elbow regions.

In one of the FabS1CE3 crystal structures (space group P3121, trigonal crystal system), the VH and VL domains of the ASU split‐Fab molecule interact with the VL and VH domains of a packing Fab molecule, forming a closed Fab dimer arrangement with the two newly assembled Fv regions situated >50 Å apart (Figure 3b, i). In the other FabS1CE3 structure (space group P212121, orthorhombic crystal system), an open‐ended crystal lattice arrangement emerges; the VH and VL domains of the principle Fab molecule associate with the VL and VH domains of two different packing Fab molecules, priming the lattice to form an interconnected array (Figure 3b, ii). A third crystal lattice packing arrangement for the split‐Fab conformation is observed in the FabS1CE4 crystal structure (space group P1, triclinic crystal system), where two Fab molecules present in the ASU associate through their respective VH and VL domains to form a closed dimer unit with the Fv regions positioned away from each other—distinct from the other Fab pairing arrangement observed in the FabS1CE3 crystal structure (Figure 3b, i and iii).

2.5. The K141Q substitution improves crystal lattice packing interactions at the Crystal Kappa:elbow junction

While crystal structures of FabS1CE and FabS1CE2 exhibit the same, conventional Fab quaternary structural conformation and elbow angle, the latter facilitated crystallization in a higher symmetry space group (P32, trigonal crystal system) and with two Fab molecules present in the ASU as opposed to one (Figure 3a, i and iii and Table 1). The different crystal forms are a result of the K141Q substitution because, compared with Lys141H in the FabS1CE structure, Gln141H in the FabS1CE2 structure enables more optimal crystal lattice packing interactions at the Crystal Kappa:elbow junction, with the side chain participating in several hydrogen bond and Van der Waals interactions with residues Thr155H′, Ser156H′, Gly157H′, Gly158H′, and Ser210H′ of the packing Fab CH domain (prime symbols indicate residues from a packing Fab ASU) (Figure 5a, i and ii).

FIGURE 5.

FIGURE 5

Open in a new tab

Molecular details of the impact of the K141Q substitution on crystal lattice packing at the Crystal Kappa:elbow junction. Residues in the KGP site and elbow region are colored cyan or magenta, respectively. The ASU Fab heavy chain constant (CH) and variable (VH) domains are colored light blue, whilst the packing Fab heavy chain constant domain (CH′) is colored dark blue. (a) (i) In the FabS1CE‐EPR‐1 structure (used as a representative), the side chain of residue Lys141H projects away from the crystal lattice packing site, toward the solvent. (ii) In the FabS1CE2‐EPR‐1 structure, Gln141H and Gln138H form hydrogen bonds and Van der Waals interactions with residues Thr155H′, Ser156H′, Gly157H′, Gly158H′, and Ser210H′ in the packing Fab CH domain. (b) (i) In the FabS1CE3‐EPR‐1 structure (P212121 space group structure used as a representative), the side chain of residue Lys141H is forced unfavorably away from residues in the packing Fab CH domain. Meanwhile, a small section of the packing Fab CH domain (Ser152H′ and Lys153H′) remains unresolved from the electron density (dotted line). (ii) In the FabS1CE4‐EPR‐1 structure, the side chain of Gln141H participates in hydrogen bonds and Van der Waals interactions with Gly157H′ in the packing Fab CH domain, which in turn transforms the Crystal Kappa:elbow junction into a more favorable crystal lattice packing site with an extensive interaction network, supported by interactions between residues in the packing Fab CH domain and residues in the principle Fab CH and VH domains, and elbow region (see main text for more details). (c) (i) In the FabS1CE1‐EPR‐1:EPOR‐ECD complex structure, the side chain of Lys141H is forced away from the crystal lattice packing site, and minimal intermolecular interactions form between residues in the principle Fab elbow region and packing Fab CH domain. (ii) In the FabS1CE2‐EPR‐1:EPOR‐ECD complex structure, the side chain of Gln141H forms hydrogen bond interactions with the side chain of Ser156H′ and peptide backbone amide group of Gly158H′ in the packing Fab CH domain, while elbow residue Gln138H forms interactions with Ser156H′ in the packing CH domain.

Similarly, in the FabS1CE4 structure, the side chain of Gln141H participates in crystal lattice packing interactions that are not feasible for the side chain of Lys141H in either of the FabS1CE3 structures. As a result, a more extended network of intramolecular and intermolecular interactions forms in the FabS1CE4 crystal lattice structure, mediated between Gln141H; residues Phe136H, Gln138H, and Ile139H of the elbow; Gln14H and Pro15H of the principle Fab VH domain; and Ser154H′, Thr155H′, Ser156H′, Gly157H′, Gly158H′, and Ser210H′ in the packing Fab CH domain (Figure 5b, i and ii). Altogether, the transformation in the crystal lattice packing at the Crystal Kappa:elbow junction that the K141Q substitution confers to the FabS1CE2 and FabS1CE4 frameworks (compared with FabS1CE and FabS1CE3 frameworks, respectively) results in their enhanced crystallizability, as evidenced by the number of crystal hits across the broad screens (Figure 2a).

2.6. The E172G substitution can induce a split‐Fab crystal form

In the FabS1CE structure, the side chain of Glu172H forms various Van der Waals and hydrogen bond interactions with Tyr169H, Ala192H, and Tyr200H in the CH domain (Figure 6a). This network of interactions stabilizes the structural conformation and position of the FPE loop region and supports the intramolecular interactions between residues Phe170H and Pro171H in the FPE loop, Leu12H and Thr134H in the VH domain, and Phe136H in the elbow region (Figure 6a). In contrast, in the FabS1CE1 and FabS1CE2 structures (which both contain the E172G substitution), Gly172H cannot participate in this interaction network in an equivalent way to Glu172H. Specifically, the deletion of the long, charged glutamate side chain removes the interactions between residue 172 and residues Tyr169H, Ala192H, and Tyr200H that normally stabilize the FPE loop (Figure 6a–c). In addition, the E172G substitution alters the conformation of the FPE loop region slightly, as reflected by changes of >10° to the peptide backbone φ and ψ dihedral bond angles of residues 171 and 172 in the FabS1CE1 and FabS1CE2 structures (Figure S7). The split‐Fab conformation captured in the FabS1CE1‐EPR‐1 structure (Figure 3a, ii) is a result of destabilization at the VH:CH domain interaction interface due to the altered structural conformation (and possibly increased flexibility) of this loop region through incorporation of E172G (Figure 6b). This destabilization at the VH:CH interface is evident in the slight decrease in T m determined for variants which contain the E172G substitution (Figure 1c).

FIGURE 6.

FIGURE 6

Open in a new tab

Molecular details of the incorporation of substitution E172G into the FabS1CE framework. Residues in the elbow region, FPE/FPG site in the CH domain, and “molecular socket” site (Stanfield et al., 2006) in the VH domain, are colored magenta, orange, or blue, respectively. Residues Tyr169H, Ala192H, and Tyr200H (in the CH domain) are colored blue. (a) In the FabS1CE structure, the side chain of residue Glu172H forms various hydrogen bonds and Van der Waals interactions with residues Tyr169H, Ala192H, and Tyr200H in the CH domain, which supports the position and conformation of the FPE loop region. Phe170H and Pro171H of the FPE site form weak intramolecular interactions with elbow residues, and Leu12H and Thr134H of the VH domain. (b,c) In the (b) FabS1CE1 and (c) FabS1CE2 structures, Gly172H cannot participate in the network of intramolecular interactions in an equivalent way to Glu172H. In addition, the φ and ψ dihedral angles in the peptide backbone are altered at positions 171 and 172 (see Figure S7). While a split‐Fab conformation was captured in the FabS1CE1 crystal structure, no changes to the Fab quaternary or tertiary structure occur in the FabS1CE2 crystal structure (see Figure 3a, ii and iii). Due to flexibility, the side chain of Phe136H was unresolved from the electron density in the FabS1CE1‐EPR‐1 structure.

Importantly, the E172G substitution did not result in a split‐Fab conformation in the FabS1CE2 crystal structure, and the usual Fab quaternary structure is preserved (Figure 3a, iii). Instead, the E172G substitution is accommodated in the CH domain without perturbing the Fab structure (Figures 6c and S8A,B). Thus, unlike the effects of the F170W substitution described above, the E172G substitution does not enforce a split‐Fab conformation.

2.7. The K141Q and E172G substitutions enhance crystallizability of a Fab:antigen complex

Previously, we attempted to crystallize Fab‐EPR‐1 in complex with its cognate antigen EPOR‐ECD but were unsuccessful using the FabWT framework (unpublished data). However, after establishing the highly crystallizable FabS1CE framework (Bruce et al., 2023), we were able to crystallize and solve the structure of the FabS1CE‐EPR‐1:EPOR‐ECD complex (P212121 space group, orthorhombic crystal system) from diffraction studies on a single crystal picked directly from a broad crystallization screen condition (Figure 7a and Crystallography Data Table S1).

FIGURE 7.

FIGURE 7

Open in a new tab

Substitution E172G facilitate an alternative crystal lattice packing arrangement for the Fab‐EPR‐1:EPOR‐ECD complex. Crystal lattice packing arrangement (upper panel) with symmetry mates, and asymmetric unit (lower panel) of the Fab‐EPR‐1:EPOR‐ECD complex. The ASU Fab light and heavy chains are colored light blue or gray, respectively, while packing Fab light and heavy chains are colored green or dark blue, respectively. The antigen EPOR‐ECD is colored magenta. (a) In the FabS1CE:EPOR‐ECD complex structure (orthorhombic crystal system, space group P212121), one Fab:antigen complex is present in the ASU. (b) In the FabS1CE1:EPOR‐ECD complex structure (orthorhombic crystal system, space group P212121), two Fab molecules and one antigen molecule are present in the ASU. (c) In the FabS1CE2:EPOR‐ECD complex structure (orthorhombic crystal system, space group P212121), two Fab molecules and one antigen molecule are present in the ASU. For the FabS1CE1 and FabS1CE2 complex structures, the secondary Fab molecule (i.e., unbound to the antigen) domains are underscored. (d,e) The Fv:EPOR‐ECD region of the FabS1CE‐EPR‐1:EPOR‐ECD complex structure is superposed with the corresponding region in (d) the FabS1CE1‐EPR‐1:EPOR‐ECD complex structure or (e) the FabS1CE2‐EPR‐1:EPOR‐ECD complex structure. The FabS1CE1‐EPR‐1:EPOR‐ECD and FabS1CE2:EPR‐1:EPOR‐ECD complex structures are colored cyan.

We were curious to see how the additional substitutions (K141Q, E172G, and F170W) impacted crystallization of this Fab:antigen complex. To investigate this, we performed a crystallization screening of Fab‐EPR‐1 in complex with EPOR‐ECD, with each of the five frameworks (S1CE and S1CE1–4), and the same screens (JCSG + Eco and PACT) used for the apo‐Fab variants (Figure 2b and Excel document in Data S1). The FabS1CE3 and FabS1CE4 frameworks (both containing F170W) generated putative Fab‐EPR‐1:EPOR‐ECD complex crystals, but in just one condition apiece, and neither was suitable for diffraction studies without further optimization. However, the FabS1CE1 and FabS1CE2 frameworks (both containing E172G) enhanced crystallizability of this complex, generating 9 or 12 crystal hits, respectively, compared with just 2 crystal hits with the original FabS1CE framework. Furthermore, several conditions generated putative Fab:antigen complex crystals large enough to be harvested and sent for diffraction studies (Excel document in Data S1 and Figure S5).

To further our understanding of how the E172G and K141Q substitutions enhance crystallizability of the FabS1CE framework in the context of the Fab‐EPR‐1:EPOR‐ECD complex, we solved this complex structure using frameworks FabS1CE1 and FabS1CE2 (Figure 7b,c, respectively). Importantly, superpositions of the Fv:EPOR‐ECD regions in the FabS1CE1 and FabS1CE2 complex structures with the FabS1CE complex structure revealed no discernible differences (Figure 7d,e). Further details of the paratope:epitope interaction will be presented and discussed elsewhere. Here we focus on the contributions of the E172G and K141Q substitutions to the crystallization of this complex.

Similar to the FabS1CE complex structure, the FabS1CE1 and FabS1CE2 complex structures were both solved in a P212121 space group (orthorhombic crystal system), from diffraction data collected on crystals harvested directly from the broad screening conditions (Crystallography Data Table S1). While the FabS1CE complex structure was solved with just one Fab:antigen complex in the ASU, the FabS1CE1 and FabS1CE2 complex structures were both solved with two Fab molecules and one antigen molecule present in their respective ASUs (Figure 7a–c). In the FabS1CE1 and FabS1CE2 complexes, the E172G substitution is accommodated in the CH domain and the usual Fab quaternary structure is preserved—further supporting that E172G does not enforce a split‐Fab conformation (Figures 7b,c, lower panels and S8A–C). In the ASUs of the FabS1CE1 and FabS1CE2 complex structures, the primary Fab molecule (bound to antigen) exhibited an elbow angle similar to that of the FabS1CE structure (∼174°) (Table 1). However, the secondary Fab molecule (unbound) in the ASUs of the FabS1CE1 and FabS1CE2 complex structures exhibited a significantly smaller elbow angle of ∼168° (Table 1) (Stanfield et al., 2006). Thus, for this Fab:antigen complex, incorporating the E172G substitution supports an alternative crystal lattice packing arrangement with improved crystallizability, as reflected by the increase in the number of crystal hits across the broad screen for variants FabS1CE1 and FabS1CE2 compared with FabS1CE (Figure 2b).

Notably, the FabS1CE2 framework facilitated crystallization of the complex in more conditions than the FabS1CE1 framework (Figure 2b). Once again, the inclusion of the K141Q substitution improved crystallizability (Figure 2). A comparison of the crystal lattice packing site at the Crystal Kappa:elbow junction in the FabS1CE1 and FabS1CE2 complex structures revealed that the Gln141H side chain formed several hydrogen bond and Van der Waals interactions with residues Ser156H and Gln157H in the packing Fab CH domain, whereas the Lys141H side chain projects away from the crystal lattice contact site toward the solvent (Figure 5c, i and ii). Comparable improvements in crystal lattice packing due to the K141Q substitution were observed in the structures of FabS1CE2 compared with FabS1CE (Figure 5a, i and ii) and FabS1CE4 compared with FabS1CE3 (Figure 5b, i and ii). Thus, the K141Q substitution results in an improvement in the crystal lattice packing interactions at the Crystal Kappa:elbow junction, and in this way, enhances crystallization by further supporting the Crystal Kappa mode of Fab:Fab packing.

3. DISCUSSION

We previously established that incorporating S1, Crystal Kappa, and elbow substitutions into a human Fab framework provides a striking improvement to its crystallizability (Bruce et al., 2023). However, after evaluating the crystal lattice packing site at the Crystal Kappa:elbow junction, we found that certain residues at this prime location did not participate in the crystal lattice interaction. Therefore, we speculated that crystallizability of the FabS1CE framework may be improved even further though optimization of the Crystal Kappa:elbow junction. As a further consideration, the incorporation of the elbow substitution results in a substantial reduction to Fab yield (Bailey et al., 2018; Bruce et al., 2023). Consequently, we aimed to identify substitutions that, on the one hand, improve crystal lattice packing interactions at the Crystal Kappa:elbow junction, and on the other hand, improve compatibility of the elbow substitution with the Fab framework.

To this end, we employed phage display technology to diversify three sites near the elbow region, each composed of three residues. The KGP site directly follows the elbow region and is situated at the Crystal Kappa:elbow junction. We found that a K141Q substitution enhanced crystallizability of the FabS1CE framework, as determined by an increase in the number of Fab and Fab:antigen complex crystal hits across a broad crystallization screening. Structural analyses revealed that Gln141H improves crystal lattice packing interactions in this region, and in this way, supports the Crystal Kappa‐mediated Fab:Fab packing mode as intended.

While the elbow substitution enhances Fab crystallizability by reducing conformational flexibility, it also creates a tighter interaction interface between the CH and VH domains, which generates local strain and results in significantly lower protein yield (Bailey et al., 2018; Bruce et al., 2023). Therefore, we diversified the FPE and GLY sites, which pack directly and indirectly against residues in the elbow region. While no changes were accommodated in the GLY site, all variants contained substitutions in the FPE site, and we found that the F170W and E172G substitutions each resulted in a striking enhancement to protein yield.

Structural analyses revealed that the E172G substitution results in changes to the network of intramolecular interactions in the CH domain that the Glu172H side chain participates in. Taken with the wider range of φ and ψ peptide backbone dihedral bond angles that Gly172H can access compared with Glu172H, we propose that the conformational flexibility of this loop region is increased. As the FPE‐loop region typically packs tightly against residues in the elbow substitution region, we suggest that enhancing flexibility of this loop allows the Fab framework to better accommodate the shortened and bulky elbow region nearby. We propose that incorporating the E172G substitution alleviates some of the strain in the FabS1CE framework while preserving the reduced Fab conformational flexibility conferred by the elbow substitution. In this way, the E172G substitution improves compatibility of the elbow substitution with the Fab framework, and consequently, improves the protein yield.

The E172G substitution did not affect the conventional quaternary structure of the FabS1CE framework in solution (as assessed by native and denaturing PAGE) or in three of four crystal structures in which it was present (apo‐FabS1CE2, and FabS1CE1 or FabS1CE2 in complex with EPOR‐ECD). However, incorporating the E172G substitution in the FabS1CE framework did result in some destabilization at the CH:VH interface, as reflected in a small reduction in the T m , and this facilitated separation of the VH:VL domains in one of the structures (apo‐FabS1CE1). Thus, the E172G substitution endows the FabS1CE framework with a potentially very useful and unique quality: under some crystallization conditions it can facilitate a dramatic conformational change. Although increasing flexibility of a protein is typically detrimental for crystallization, we speculate that the conformational flexibility of the split‐Fab form may prove beneficial for some Fab:antigen complex cases. While the S1 and Crystal Kappa substitutions are free to form their signature crystal lattice packing interactions, the flexibility between the Fv and constant domain regions of the split‐Fab form could allow it to adapt to crystal lattice requirements—where the canonical Fab quaternary structure cannot.

The F170W substitution also releases the strain on the FabS1CE framework incurred by the elbow substitution, which results in an increase in protein yield. Structural analyses revealed the molecular basis for this: the F170W substitution enforces a split‐Fab conformation due to unavoidable steric clashes between the bulky Trp170H side chain and residues in the elbow region and VH domain. In contrast to the E172G substitution, we predict that the F170W substitution will always result in a split‐Fab conformation when used in conjunction with the elbow substitution. Further structural and biophysical studies are required to fully understand the effects of the F170W substitution when incorporated into the FabS1CE framework, and such studies are underway. While the FabS1CE, FabS1CE1, and FabS1CE2 frameworks facilitated crystallization of the Fab‐EPR‐1:EPOR‐ECD complex, we speculate that the FabS1CE3 and FabS1CE4 frameworks (which contain the F170W substitution) may be better suited for crystallization of other Fab:antigen complexes. We intend to explore the benefits of using this split‐Fab conformation in conjunction with the S1 and Crystal Kappa substitutions for the crystallization screening of other Fab:antigen complexes.

Although rare, there is precedence in the Protein Data Bank for Fab and Fab:antigen complex crystal structures with split‐Fab conformations (Berman et al., 2000; Calarese et al., 2003; Mishra et al., 2023; Romei et al., 2024; Shahid et al., 2021). In one case, a broadly neutralizing human antibody, targeting a carbohydrate‐rich epitope on the HIV envelope glycoprotein GP120, exhibited an unusual conformation in which two Fabs were interlocked through domain swapping of their respective VH and VL domains (Calarese et al., 2003)—analogous to the split‐Fab crystal lattice packing arrangements observed in one of the split‐Fab crystal structures (P3121 space group) presented here. However, unlike the substitutions in our engineered Fab frameworks, the interlocked Fab structure in this case was the result of a rare but natural Ser to Pro substitution at position 137 in the heavy chain elbow region, which together with other changes, resulted in a Fab dimer state in solution. The relative arrangement of the interlocked Fab structure and the closeness of the Fv regions, in addition to the crystal lattice symmetry, are different from the split‐Fab formats presented here.

The crystal structure of another VH/VL domain‐swapped Fab dimer was solved more recently (Mishra et al., 2023). In this case, the authors made a serendipitous discovery during a phage‐display affinity maturation experiment. They found that deletion of residue Ser137H in the heavy chain elbow region of a human Fab framework resulted in a closed Fab dimer conformation exhibiting a “doughnut‐shaped” arrangement with the connecting Fv regions facing outward—opposed to positioned in parallel and in proximity, as found in the Fab:GP120 structure described above (Mishra et al., 2023; Shahid et al., 2021). Although predominantly a dimer state in solution, a single Fab molecule was present in the crystal structure ASU, with the CH and CL domains forming the usual interface interaction but with the VH and VL domains separated and forming interactions with the VL and VH domain of a packing Fab molecule instead (Shahid et al., 2021). Notably, the authors utilized this bivalent Fab architecture as a fiducial marker in cryogenic electron microscopy studies to solve the structure of a therapeutic antibody in complex with the T cell receptor LAG3 (Mishra et al., 2023).

In summary, the three heavy‐chain substitutions reported here can be combined with light‐chain S1 and Crystal Kappa substitutions and the heavy‐chain elbow substitution (Bailey et al., 2018; Bruce et al., 2023; Lieu et al., 2020) to further enhance the capacity for Fab frameworks to form productive crystal contacts. The K141Q substitution supports the Crystal Kappa‐mediated crystal lattice packing interaction, while the E172G substitution improves the compatibility of the elbow substitution with the Fab framework. The F170W substitution consistently generates a split‐Fab conformation and provides alternative crystal lattice packing interactions due to enhanced flexibility between the Fv and constant domains, which may be advantageous for crystallization of some complexes. Notably, all these alterations reside in the constant regions of the Fab, and we expect that any Fab of interest can converted to a highly crystallizable format by simply combining the relevant VH/VL domains with our optimized constant domains. Provided that residues in the VH and CH domain that constitute the molecular ball and socket are present (Bailey et al., 2018; Stanfield et al., 2006), we recommend combining the elbow and Crystal Kappa substitution with the K141Q and E172G substitutions to produce a Fab heavy chain with improved yield and the ability to facilitate more favorable crystal lattice packing arrangements. As shown here, this optimized heavy‐chain framework can be combined with the optimized light‐chain framework containing S1 and Crystal Kappa substitutions to greatly improve crystallizability of the entire Fab. Altogether, we anticipate that this toolkit will be broadly applicable for the crystallization of many Fab:antigen complexes and may improve Fabs as fiduciary marks for cryogenic electron microscopy to enable the elucidation of structures that are recalcitrant to standard methods.

4. MATERIALS AND METHODS

4.1. Construction and screening of the phage‐displayed library

A phage‐displayed Fab library was constructed as described previously, using a phagemid system (Fellouse & Sidhu, n.d.). Nine positions were diversified in the Fab heavy chain constant domain (Figure 1a) using a soft‐randomization strategy where ~50% of the amino acid identity was kept as the WT sequence. The phage library pool was cycled through a total of six rounds of binding selections with EPOR‐ECD protein with a C‐terminal hexahistidine tag (preparation described below) immobilized on Maxisorp Immuno plates (ThermoFisher, catalog number: 12‐565‐135). To enrich the library, the stringency was increased in each round by lowering the concentration of antigen coated and/or increasing the number of wash steps. After rounds five and six, individual phage clones were subjected to polymerase chain reaction (PCR) cycling to amplify DNA encoding the Fab CH domain, which was analyzed by DNA sequencing to decode the sequence of the CH domain.

4.2. Protein production

For expression of Fab‐EPR‐1 variants, the genes encoding the heavy‐ and light‐chains were cloned into separate vectors, suitable for mammalian expression, and purified as described (Bruce et al., 2023; Tao et al., 2019). Fabs were expressed and purified in quadruplicate to obtain a mean value of yield. The same construct and expression system was used for preparation of the EPOR‐ECD protein, but with a hexahistidine tag attached to the C‐terminus to facilitate purification by immobilized metal affinity chromatography. Mammalian cell culture expressing EPOR‐ECD‐His was supplemented with a final concentration of 5 μM Kifunensine (MedChemExpress) to inhibit mannosidase I activity. Purified EPOR‐ECD‐His was buffer exchanged into 20 mM HEPES pH 7.5, 100 mM NaCl, clarified by centrifugation, and treated with endoH (Promega, catalog number: 31985‐062) to complete deglycosylation, prior to Fab:antigen complex preparation and crystallization screening. For all Fab and antigen proteins, denaturing and native polyacrylamide gel electrophoresis (4%–15% Mini‐PROTEAN TGX Stain‐Free Precast Gels—Bio‐Rad, catalog number: 4568083) was carried out to assess purity, monodispersity, and solution state.

4.3. Differential scanning fluorimetry

SYPRO™ Orange protein stain (Thermo Fisher, Catalog number: S5692‐500UL) was added to 5 μM Fab protein in phosphate buffered saline (PBS) to perform a thermal melt of 25°C–95°C (0.5°C/30 s intervals) and determine protein melting temperature, as described (Miersch et al., 2021; Niedziela‐Majka et al., 2015). All measurements were made in quadruplicate, and their mean values presented.

4.4. Crystallization screening and crystal harvesting of apo‐Fabs and Fab:Antigen complexes

All Fab‐EPR‐1 variants were concentrated to 7 mg/mL in 20 mM HEPES pH 7.5, 100 mM NaCl, and clarified by centrifugation. The Fab‐EPR‐1:EPOR‐ECD complexes were prepared for crystallization screening at a 1:1 molar ratio (Fab:antigen), 7 mg/mL final concentration, in 20 mM HEPES pH 7.5, 100 mM NaCl. Complexes were clarified by centrifugation prior to crystallization screening.

A Mosquito Crystal robot (SPT Labtech) was used to set up sitting drop crystallization screens with protein: precipitant drops at 0.2 μL:0.2 μL, with 40 μL reservoir solution, at room temperature on 96‐well plates (Cryschem sitting drop plates, Hampton Research, catalog number: HR3‐159) and sealed with ClearSeal Film™ and Applicator (Hampton Research, catalog number: HR4‐521). For this, commercial screens JCSG‐plus HT‐96 Eco and PACT Premiere HT‐96 (Molecular Dimensions) were used. Crystallization plates were incubated at room temperature and manually checked by light microscopy every few days for 4 weeks. Crystallization conditions which generated crystal hits were recorded, along with the crystal morphology (Excel document in Data S1). Crystals of apo‐Fab‐EPR‐1 variants, and Fab‐EPR‐1:EPOR‐ECD complexes, were picked directly from the broad screen conditions (Crystallography Data Table S1), and washed with their respective crystallization reservoir solution supplemented with 25% (v/v) ethylene glycol when appropriate as a cryoprotectant, followed by flash freezing in liquid nitrogen.

4.5. Data collection and structure determination and refinement

X‐ray diffraction experiments were performed at beamline 24‐ID‐E at the Northeastern Collaborative Access Team (NECAT) at the Advance Photon Source at Argonne National Laboratory (Argonne, IL). Datasets were collected remotely from single crystals, at 100 K, using web‐based remote GUI developed by the NECAT team. Individual datasets were indexed and integrated with XDS (Kabsch, 2010), following scaling with Aimless (Collaborative Computational Project, Number 4, 1994).

Structures of the following (with their respective diffraction resolutions) were solved using molecular replacement in PHASER (McCoy et al., 2007): FabS1CE‐EPR‐1 (1.8 Å), FabS1CE1‐EPR‐1 (3.3 Å), FabS1CE2‐EPR‐1 (2.5 Å), FabS1CE3‐EPR‐1 (P3121, trigonal crystal system) (3.1 Å), FabS1CE3‐EPR‐1 (P212121, orthorhombic crystal system) (2.0 Å), FabS1CE4‐EPR‐1 (2.4 Å), FabS1CE‐EPR‐1:EPOR‐ECD (2.1 Å), FabS1CE1‐EPR‐1:EPOR‐ECD (2.9 Å), and FabS1CE2‐EPR‐1:EPOR‐ECD (3.1 Å). For FabS1CE‐EPR‐1 and FabS1CE2‐EPR‐1, the Fv and constant domain of FabS1CE‐F1 (Bruce et al., 2023) (PDB: 8T7I) were used as molecular replacement models. For all other apo‐Fab structures (i.e., with split‐Fab conformations), the constant domain, VL, and VH domains of FabS1CE‐F1 (Bruce et al., 2023) were used as molecular replacement models. For the three Fab‐EPR‐1:EPOR‐ECD complex structures, the EPOR‐ECD structure (from PDB: 1EBP), and the Fv and constant domains of FabS1CE‐F1 (Bruce et al., 2023), were used as molecular replacement models. All structures were refined in PHENIX (Afonine et al., 2012) and manually corrected in Coot (Emsley et al., 2010). Model refinement was undertaken by manual re‐modeling in Coot (Emsley et al., 2010) and automated fitting and geometry optimization with Phenix.refine (Afonine et al., 2012), with the use of TLS (Urzhumtsev et al., 2013) parameters.

4.6. Data deposition

Coordinates and structure factors have been deposited into the Protein Data Bank (Berman et al., 2000) under the following accession codes: FabS1CE‐EPR‐1 (8VTP), FabS1CE1‐EPR‐1 (8VUA), FabS1CE2‐EPR‐1 (8VVO), FabS1CE3‐EPR‐1 (8VTR) (orthorhombic crystal system), FabS1CE3‐EPR‐1 (trigonal crystal system) (8VU1), FabS1CE4‐EPR‐1 (8VU4), FabS1CE‐EPR‐1:EPOR‐ECD (8VUI), FabS1CE1‐EPR‐1:EPOR‐ECD (8VVM), and FabS1CE2‐EPR‐1:EPOR‐ECD (8VUC).

AUTHOR CONTRIBUTIONS

Heather A. Bruce: Formal analysis; software; visualization; methodology; validation; writing – review and editing; writing – original draft; investigation; conceptualization; data curation. Alexander U. Singer: Conceptualization; investigation; writing – review and editing; writing – original draft; visualization; validation; methodology; software; formal analysis; supervision; data curation. Levi L. Blazer: Supervision; project administration; writing – review and editing; investigation; methodology. Khanh Luu: Investigation; methodology; validation. Lynda Ploder: Methodology; validation; investigation. Alevtina Pavlenco: Methodology; validation; investigation. Igor Kurinov: Formal analysis; software; writing – review and editing; investigation; resources. Jarrett J. Adams: Conceptualization; writing – review and editing; methodology; project administration; supervision; data curation. Sachdev S. Sidhu: Data curation; supervision; resources; project administration; writing – review and editing; writing – original draft; funding acquisition; conceptualization.

Supporting information

Data S1. Supporting Information.

PRO-33-e5081-s002.xlsx (14.3KB, xlsx)

Data S2. Supporting Information.

PRO-33-e5081-s001.docx (2.9MB, docx)

ACKNOWLEDGMENTS

We thank Victor Pau and Daniel Mao for their technical assistance. We are grateful to Michael Suits, Greg Martyn, and Ekaterina Filippova for their helpful insights and discussions. We would like to thank Anthony Kossiakoff and Shane Atwell for their valuable comments on the manuscript. This work is based upon research conducted at the Northeastern Collaborative Access Team beamlines, which is funded by the National Institute of General Medical Sciences from the National Institutes of Health. The Eiger 16M detector on the 24‐ID‐E beam line is funded by a NIH‐ORIP HEI grant (S10OD021527). This research used resources of the Advanced Photon Source; a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE‐AC02‐06CH11357.

Bruce HA, Singer AU, Blazer LL, Luu K, Ploder L, Pavlenco A, et al. Antigen‐binding fragments with improved crystal lattice packing and enhanced conformational flexibility at the elbow region as crystallization chaperones. Protein Science. 2024;33(7):e5081. 10.1002/pro.5081

Review Editor: Jeanine Amacher

REFERENCES

  1. Afonine PV, Grosse‐Kunstleve RW, Echols N, Headd JJ, Moriarty NW, Mustyakimov M, et al. Towards automated crystallographic structure refinement with phenix.refine. Acta Crystallogr D Biol Crystallogr. 2012;68:352–367. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bailey LJ, Sheehy KM, Dominik PK, Liang WG, Rui H, Clark M, et al. Locking the elbow: improved antibody fab fragments as chaperones for structure determination. J Mol Biol. 2018;430:337–347. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Barthelemy PA, Raab H, Appleton BA, Bond CJ, Wu P, Wiesmann C, et al. Comprehensive analysis of the factors contributing to the stability and solubility of autonomous human VH domains. J Biol Chem. 2008;283:3639–3654. [DOI] [PubMed] [Google Scholar]
  4. Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, et al. The protein data bank. 2000. Available from: http://www.rcsb.org/pdb/status.html [DOI] [PMC free article] [PubMed]
  5. Bostrom J, Haber L, Koenig P, Kelley RF, Fuh G. High affinity antigen recognition of the dual specific variants of Herceptin is entropy‐driven in spite of structural plasticity. PLoS One. 2011;6:e17887. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Boulot G, Guillon V, Mariuzza RA, Poljak RJ, Riott'ot M‐M, Souchon H, et al. Crystallization of antibody fragments and their complexes with antigen. J Cryst Growth. 1988;90:213. [Google Scholar]
  7. Bruce HA, Singer AU, Filippova EV, Blazer LL, Adams JJ, Enderle L, et al. Engineered antigen‐binding fragments for enhanced crystallization of antibody:antigen complexes. Protein Sci. 2023;33:e4824. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Calarese DA, Scanlan CN, Zwick MB, Deechongkit S, Mimura Y, Kunert R, et al. Antibody domain exchange is an immunological solution to carbohydrate cluster. Science. 2003;300:2065–2071. [DOI] [PubMed] [Google Scholar]
  9. Canaves JM, Page R, Wilson IA, Stevens RC. Protein biophysical properties that correlate with crystallization success in Thermotoga maritima: maximum clustering strategy for structural genomics. J Mol Biol. 2004;344:977–991. [DOI] [PubMed] [Google Scholar]
  10. Chiu ML, Goulet DR, Teplyakov A, Gilliland GL. Antibody structure and function: the basis for engineering therapeutics. Antibodies. 2019;8:1–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Collaborative Computational Project, Number 4 . The CCP4 suite: programs for protein crystallography. Acta Crystallogr D Biol Crystallogr. 1994;D50:760–3. [DOI] [PubMed] [Google Scholar]
  12. Cooper DR, Boczek T, Grelewska K, Pinkowska M, Sikorska M, Zawadzki M, et al. Protein crystallization by surface entropy reduction: optimization of the SER strategy. Acta Crystallogr D Biol Crystallogr. 2007;63:636–645. [DOI] [PubMed] [Google Scholar]
  13. Czepas J, Devedjiev Y, Krowarsch D, Derewenda U, Otlewski J, Derewenda ZS. The impact of Lys → Arg surface mutations on the crystallization of the globular domain of RhoGDI. Acta Crystallogr D Biol Crystallogr. 2004;60:275–280. [DOI] [PubMed] [Google Scholar]
  14. Derewenda ZS. Application of protein engineering to enhance crystallizability and improve crystal properties. Acta Crystallogr D Biol Crystallogr. 2010;66:604–615. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Derewenda ZS, Vekilov PG. Entropy and surface engineering in protein crystallization. Acta Crystallogr D Biol Crystallogr. 2006;62:116–124. [DOI] [PubMed] [Google Scholar]
  16. Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of Coot. Acta Crystallogr D Biol Crystallogr. 2010;66:486–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Ereño‐Orbea J, Sicard T, Cui H, Carson J, Hermans P, Julien JP. Structural basis of enhanced crystallizability induced by a molecular chaperone for antibody antigen‐binding fragments. J Mol Biol. 2018;430:322–336. [DOI] [PubMed] [Google Scholar]
  18. Fellouse FA, Li B, Compaan DM, Peden AA, Hymowitz SG, Sidhu SS. Molecular recognition by a binary code. J Mol Biol. 2005;348:1153–1162. [DOI] [PubMed] [Google Scholar]
  19. Fellouse FA, Sidhu SS. Making antibodies in bacteria. In: Howard GC, Matthew R, editors. Kaser making and using antibodies. Boca Raton: CRC Press; 2006. p. 157–180. [Google Scholar]
  20. Fellouse FA, Wiesmann C, Sidhu SS. Synthetic antibodies from a four‐amino‐acid code: a dominant role for tyrosine in antigen recognition. Proc Natl Acad Sci U S A. 2004;101:12467–12472. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Frei JC, Lai JR. Protein and antibody engineering by phage display. Methods Enzymol. 2016;580:45–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Garrard SM, Longenecker KL, Lewis ME, Sheffield PJ, Derewenda ZS. Expression, purification, and crystallization of the RGS‐like domain from the Rho nucleotide exchange factor, PDZ‐RhoGEF, using the surface entropy reduction approach. Protein Expr Purif. 2001;21:412–416. [DOI] [PubMed] [Google Scholar]
  23. Kabsch W. XDS. Acta Crystallogr D Biol Crystallogr. 2010;66:125–132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lee HT, Lee JY, Lim H, Lee SH, Moon YJ, Pyo HJ, et al. Molecular mechanism of PD‐1/PD‐L1 blockade via anti‐PD‐L1 antibodies atezolizumab and durvalumab. Sci Rep. 2017;7:1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lieu R, Antonysamy S, Druzina Z, Ho C, Kang NR, Pustilnik A, et al. Rapid and robust antibody Fab fragment crystallization utilizing edge‐to‐edge beta‐sheet packing. PLoS One. 2020;15:e0232311. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Longenecker KL, Garrard SM, Sheffield PJ, Derewenda ZS. Biological crystallography protein crystallization by rational mutagenesis of surface residues: Lys to ala mutations promote crystallization of RhoGDI. Acta Crystallogr D Biol Crystallogr. 2001;57:679–688. [DOI] [PubMed] [Google Scholar]
  27. Males A, Davies GJ. Structural studies of a surface‐entropy reduction mutant of O‐GlcNAcase. Acta Crystallogr D Struct Biol. 2019;75:70–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Martyn GD, Veggiani G, Kusebauch U, Morrone SR, Yates BP, Singer AU, et al. Engineered SH2 domains for targeted phosphoproteomics. ACS Chem Biol. 2022;17:1472–1484. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. McCoy AJ, Grosse‐Kunstleve RW, Adams PD, Winn MD, Storoni LC, Read RJ. Phaser crystallographic software. J Appl Cryst. 2007;40:658–674. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Miersch S, Li Z, Saberianfar R, Ustav M, Brett Case J, Blazer L, et al. Tetravalent SARS‐CoV‐2 neutralizing antibodies show enhanced potency and resistance to escape mutations. J Mol Biol. 2021;433:167177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mishra AK, Shahid S, Karade SS, Agnihotri P, Kolesnikov A, Hasan SS, et al. CryoEM structure of a therapeutic antibody (favezelimab) bound to human LAG3 determined using a bivalent Fab as fiducial marker. Structure. 2023;31:1149–1157. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Niedziela‐Majka A, Kan E, Weissburg P, Mehra U, Sellers S, Sakowicz R. High‐throughput screening of formulations to optimize the thermal stability of a therapeutic monoclonal antibody. J Biomol Screen. 2015;20:552–559. [DOI] [PubMed] [Google Scholar]
  33. Power CA, Bates A. David vs. Goliath: the structure, function, and clinical prospects of antibody fragments. Antibodies. 2019;8:1–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Romei MG, Leonard B, Katz ZB, Le D, Yang Y, Day ES, et al. I‐shaped antibody engineering enables conformational tuning of biotherapeutic receptor agonists. Nat Commun. 2024;15:642. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Schaefer G, Haber L, Crocker LM, Shia S, Shao L, Dowbenko D, et al. A two‐in‐one antibody against HER3 and EGFR has superior inhibitory activity compared with monospecific antibodies. Cancer Cell. 2011;20:472–486. [DOI] [PubMed] [Google Scholar]
  36. Shahid S, Gao M, Travis Gallagher D, Pozharski E, Brinson RG, Keck ZY, et al. Crystal structure of a bivalent antibody fab fragment. J Mol Biol. 2021;433:166714. [DOI] [PubMed] [Google Scholar]
  37. Stanfield RL, Zemla A, Wilson IA, Rupp B. Antibody elbow angles are influenced by their light chain class. J Mol Biol. 2006;357:1566–1574. [DOI] [PubMed] [Google Scholar]
  38. Tao Y, Mis M, Blazer L, Jnr MU, Steinhart Z, Chidiac R, et al. Tailored tetravalent antibodies potently and specifically activate Wnt/Frizzled pathways in cells, organoids and mice. eLife. 2019;8:e46134. 10.7554/eLife.46134.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Urzhumtsev A, Afonine PV, Adams PD. TLS from fundamentals to practice. Crystallogr Rev. 2013;19:230–270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Vidarsson G, Dekkers G, Rispens T. IgG subclasses and allotypes: from structure to effector functions. Front Immunol. 2014;5:1–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Ye J‐D, Tereshko V, Frederiksen JK, Koide A, Fellouse FA, Sidhu SS, et al. Synthetic antibodies for specific recognition and crystallization of structured RNA. Proc Natl Acad Sci U S A. 2008;105(1):82–87. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1. Supporting Information.

PRO-33-e5081-s002.xlsx (14.3KB, xlsx)

Data S2. Supporting Information.

PRO-33-e5081-s001.docx (2.9MB, docx)

Articles from Protein Science : A Publication of the Protein Society are provided here courtesy of The Protein Society

RESOURCES