Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Oct 5.
Published in final edited form as: Structure. 2023 Aug 23;31(10):1149–1157.e3. doi: 10.1016/j.str.2023.07.013

CryoEM structure of a therapeutic antibody (favezelimab) bound to human LAG3 determined using a bivalent Fab as fiducial marker

Arjun K Mishra 1,2, Salman Shahid 1,2, Sharanbasappa S Karade 1,2, Pragati Agnihotri 1,2, Alexander Kolesnikov 1,2, S Saif Hasan 3,4,5, Roy A Mariuzza 1,2,*
PMCID: PMC11197462  NIHMSID: NIHMS1925708  PMID: 37619561

SUMMARY

Lymphocyte activation gene 3 protein (LAG3) is an inhibitory receptor that is upregulated on exhausted T cells in tumors. LAG3 is a major target for cancer immunotherapy with many anti-LAG3 antibodies in clinical trials. However, there is no structural information on the epitopes recognized by these antibodies. We determined the single-particle cryoEM structure of a therapeutic antibody (favezelimab) bound to LAG3 to 3.5 Å resolution, revealing that favezelimab targets the LAG3 binding site for MHC class II, its canonical ligand. The small size of the complex between the conventional (monovalent) Fab of favezelimab and LAG3 (~100 kDa) presented a challenge for cryoEM. Accordingly, we engineered a bivalent version of Fab favezelimab that doubled the size of the Fab–LAG3 complex and conferred a highly identifiable shape to the complex that facilitated particle selection and orientation for image processing. This study establishes bivalent Fabs as new fiducial markers for cryoEM analysis of small proteins.

eTOC Blurb

LAG3 is receptor on T cells that is a major target for cancer immunotherapy using monoclonal antibodies against LAG3. However, it is unknown how these antibodies bind LAG3. Mishra et al. used cryoEM to identify the site on LAG3 targeted by a therapeutic antibody, which explains how the antibody works.

Graphical Abstract

graphic file with name nihms-1925708-f0005.jpg

INTRODUCTION

Elevated expression of immune inhibitory receptors (IRs) on exhausted T cells in the tumor microenvironment limits their anti-tumor activity.1 T cell IRs are now major therapeutic targets in cancer, with monoclonal antibodies (mAbs) that block CTLA4 and PD1/PDL1 in the clinic since 2010.2,3 While these mAbs have dramatically improved outcomes for some cancers, notably advanced melanoma, many patients do not exhibit durable responses. Therefore, there is an urgent need to identify additional targets to combine with anti-CTLA4 and anti-PD1/PDL1 mAbs to improve their efficacy.4,5 This has generated considerable interest in lymphocyte activation gene-3 (LAG3; CD223), which is the third IR to be targeted in the clinic.6 LAG3 downregulates T cell activation, proliferation, and cytokine production, rendering T cells dysfunctional.7

LAG3 is a ~55 kDa type I transmembrane glycoprotein consisting of four extracellular immunoglobulin (Ig)-like domains (D1–D4),8,9 a connecting peptide, and an intracellular region that transmits inhibitory signals to the T cell upon binding to MHC class II molecules.4,10 LAG3 shares a similar domain architecture with the T cell co-receptor CD4, which also binds MHC class II. However, the affinity of LAG3 for MHC class II is ~1000-fold greater than that of CD411. In addition to MHC class II, its primary ligand, other LAG3 binding partners include FGL1,12 α-synuclein,13 and LSECtin.14

Preclinical studies targeting LAG3 combined with PD1 have shown significant increases in tumor clearance and survival in several mouse tumor models.15,16 As a result, nearly 20 anti-LAG3 mAbs are now in clinical trials for cancer immunotherapy.4,5 Moreover, one of these mAbs, relatlimab, recently received FDA approval for treatment of unresectable or metastatic melanoma in combination with the anti-PD1 mAb nivolumab.6 Favezelimab (originally named 22D2) is an anti-LAG3 mAb developed by Merck that aims to restore T cell effector function by preventing LAG3 from binding to MHC class II.17 It binds LAG3 with high affinity (KD = 5 nM).18 Treatment of T cells with favezelimab (also called MK-4280) has been shown to increase production of cytokines (IL-2, IL-8, IFN-γ and TNF-α) and chemokines (CCL4, CCL22 and CXCL10), as well as to upregulate expression of markers associated with T cell activation (CD69, CD44, CD25, XCL1, and NFAT).19 Favezelimab combined with the anti-PD1 mAb pembrolizumab is under evaluation across multiple solid tumors and lymphomas.4,5 In a phase 1/2 trial, favezelimab demonstrated effective anti-tumor activity in patients with relapsed or refractory classical Hodgkin lymphoma whose disease had progressed on prior anti-PD1 therapy.20 A phase 3 trial for metastatic colon cancer is underway (NCT05064059).

There is a lack of structural information linking the epitopes bound by different anti-LAG3 mAbs to their effect on LAG3 activity. We previously used chimeras between human and mouse LAG3 to map the epitope recognized by favezelimab to the D1 domain,18 but could not further define the epitope. To obtain this information, we used single-particle cryoEM to determine the structure of favezelimab bound to LAG3. The relatively small size of a complex between a conventional (i.e. monovalent) Fab fragment of favezelimab and LAG3 (~100 kDa) presents a challenge for cryoEM because small targets often lack recognizable shape features that can facilitate initial image alignment at low resolution. Structure determination is further complicated by the fact that both the Fab and LAG3 have Ig domain architecture, which makes alignment and subsequent model building challenging. An additional complication is the inherent flexibility of the elbow regions linking the variable (V) and constant (C) domains of Fabs, which may reduce their ability to orient particles accurately.21 To overcome these obstacles, we engineered a dimeric bivalent version of Fab favelelimab that doubled the effective size of its complex with LAG3 and conferred a highly identifiable shape to the complex to facilitate image processing. This proof-of-concept study demonstrates the utility of bivalent Fabs as new fiducial markers for cryoEM.

RESULTS

Design and structure of dimeric Fab favezelimab

Conventional Fabs are monovalent monomers. We previously reported a method for converting the conventional Fab of an antibody (HC84.26.5D) that recognizes the E2 envelope glycoprotein of hepatitis C virus (HCV) into a bivalent dimer by deleting a single residue, VHSer113 (Kabat numbering), in the elbow region (residues 112–116) linking the VH and CH1 domains.22 This deletion resulted in formation of an interlocked domain-swapped dimer that bound two HCV E2 molecules. However, this single residue deletion was insufficient to convert monomeric Fab favezelimab to a dimer, which instead required deletion of two elbow region residues, CH114Ala and CH115Ser. We examined the oligomeric state of these proteins by size exclusion chromatography (SEC) (Figure 1A). Wild-type Fab favezelimab (Fab favezelimab(wt)) ran as a single peak of ~50 kDa, as expected. By contrast, mutated Fab favezelimab (Fab favezelimab(mut)) eluted as two separate peaks: a major peak (85%) of ~100 kDa and minor peak (15%) of ~50 kDa. This indicates that Fab favezelimab(mut) exists primarily as a dimer in solution, as was the case for mutated HC84.26.5D.22

Figure 1. Architecture of monomeric and dimeric Fab molecules.

Figure 1.

Open in a new tab

(A) Oligomeric state of Fab fragments and Fab–LAG3 complex. Superdex 200 size exclusion chromatograms of Fab favezelimab(wt) (black), Fab favezelimab(mut) with deletion of CH114Ala and CH115Ser (red), LAG3 (green), and the Fab favezelimab(mut)–LAG3 complex (yellow).

(B) Structure of monomeric Fab favezelimab(wt) showing canonical architecture and normal interactions between CH1 and CL domains. H and L chains are blue and brown, respectively.

(C) Monomer of Fab favezelimab(mut) in the asymmetric unit of the crystal showing clear separation of CH1 from its normal interactions with CL. The VH and VL domains associate in the conventional manner. The C domains are twisted with respect to the V domains compared to their standard orientation in Fab favezelimab(wt).

(D) Structure of dimeric Fab favezelimab(mut) resulting from domain swapping of CH1 and CL with the corresponding C domains of a symmetry-related Fab molecule in the crystal.

To elucidate the molecular basis for Fab favezelimab(mut) dimerization, we determined crystal structures of the wild-type and mutated Fab to 2.1 Å and 2.8 Å resolution, respectively (Table S1) (Figure 1). The electron density maps of both proteins were easily traced, except for CH1 residues 129–134 in both structures. The Fab favezelimab(wt) crystal contains one Fab molecule in the asymmetric unit. The overall architecture of Fab favezelimab(wt) is completely canonical, with paired VLCL and VHCH1 chains (Figure 1B). The Fab favezelimab(mut) crystal also contains one Fab molecule in the asymmetric unit (Figure 1C). However, the architecture of Fab favezelimab(mut) is dramatically different from that of Fab favezelimab(wt) (Figure 1D). In the Fab favezelimab(mut) asymmetric unit, the CL and CH1 domains are separated by ~35 Å, while VL and VH associate in the same manner as in Fab favezelimab(wt), as evident by superposing the VL/VH modules of the two proteins (r.m.s.d. of 0.6 Å for 220 α-carbon atoms). The only notable difference occurs in VHCDR3, where VHTrp98 at the tip of this typically flexible loop is displaced by 4.4 Å in its α-carbon position, probably due to steric clashes with a neighboring Fab favezelimab(mut) molecule in the crystal lattice (Figure S1). We do not expect this difference to affect binding affinity since alternate CDR3 loop conformation are common in comparisons of the same Fab crystallized in different space groups.

Three-dimensional domain swapping of CL and CH1 with the corresponding C domains of a symmetry-related Fab molecule in the crystal results in formation of an interlocked doughnut-shaped Fab dimer with a rectangular hole (Figure 1D). However, each CL in the dimer makes essentially the same interactions with CH1 as does CL with CH1 in monomeric Fab favezelimab (r.m.s.d. of 0.4 Å for superposition of 177 CL/CH1 α-carbon atoms).

To understand the domain movements underpinning dimerization, we used the program DynDom (http://dyndom.cmp.uea.ac.uk/dyndom/)23 to calculate the angle of rotation around the switch peptide of the favezelimab(mut) H chain (residues 110–117), which links VH and CH1, as 71° with respect to favezelimab(wt) (Figure S2A). The angle of rotation around the switch peptide of the favezelimab(mut) L chain (residues 108–110), which links VL and CL, is 113° with respect to favezelimab(wt) (Figure S2B). This larger angle of rotation results in a more extended conformation of the L chain than the H chain (Figure 1C). The more extended conformation of the favezelimab(mut) H chain compared to the favezelimab(wt) H chain is reflected in a reduction in total buried surface between VH and CH1 from 1723 Å2 for favezelimab(wt) to 782 Å2 for favezelimab(mut), as analyzed by PDBePISA (https://www.ebi.ac.uk/pdbe/pisa/ ).24 Similarly, the more extended conformation of the favezelimab(mut) L chain relative to the favezelimab(wt) L chain manifests as a reduction in total buried surface between VL and CL from 1754 Å2 for favezelimab(wt) to 814 Å2 for favezelimab(mut).

Although dimeric Fab favezelimab(mut) and Fab HC84.26.5D22 display similar overall architectures, the disposition of V and C domains is different in the two doughnut-like structures, with central holes of different shapes and dimensions: a diamond-shaped hole of 43 Å × 49 Å for favezelimab(mut) versus a rectangular-shaped hole of 33 Å × 41 Å for HC84.26.5D (Figure S3). Moreover, the angles of rotation around the switch peptides of the H and L chains of dimeric Fab HC84.26.5D with respect to its monomeric version (111° and 95°, respectively) differ significantly from the corresponding angles of rotation for Fab favezelimab(mut) (71° and 113°, respectively). The different geometries of the two domain-exchanged Fab dimers is most likely a consequence of the different elbow region deletions: VHSer113 in Fab HC84.26.5D versus CH114Ala and CH115Ser in Fab favezelimab(mut).

Antibody molecules contain a highly conserved ball-and-socket joint between VH and CH1 that plays a key role in the flexibility of V domains with respect to C domains.25 This joint is present in Fab favezelimab(wt) (Figure 2A). Residues Leu11, Thr110 and Ser112 in VH, which form the socket, pack against residues Phe146 and Pro147 in CH1, which form the ball. Although these five residues are present in Fab favezelimab(mut), they are arranged differently in the domain-exchanged dimer and do not form a canonical ball-and-socket joint (Figure 2B). Moreover, the elbow angle25 between VH and CH1 domains is 147° for Fab favezelimab(wt) compared to 217° for Fab favezelimab(mut). Conformational differences in the elbow region between VL and CL (residues 105–110) are shown in Figure 2.

Figure 2. Elbow regions in monomeric, dimeric, and LAG3-bound Fab favezelimab.

Figure 2.

Open in a new tab

(A) (top) Structure of monomeric Fab favezelimab(wt) with elbow region between VH and CH1 domains framed in black. The elbow angle is 147°. The elbow region between VL and CL domains is framed in red. (middle) Close-up view of standard ball-and-socket joint between VH and CH1 in Fab favezelimab(wt) formed by VH residues Leu11, Thr110 and Ser112 (socket) and CH1 residues Phe146 and Pro147 (ball). Also shown in magenta are CH114Ala and CH115Ser, which are deleted in Fab favezelimab(mut). (bottom) Close-up view of the residues forming the elbow region between VL and CL.

(B) (top) Structure of Fab favezelimab(mut) molecule in the crystal with elbow region between VH and CH1 framed in black. The elbow angle is 217°. The elbow region between VL and CL is framed in red. (middle) Close-up view of residues forming ball-and-socket joint in Fab favezelimab(wt) but rearranged in Fab favezelimab(mut). (bottom) Close-up view of the residues forming the elbow region between VL and CL.

(C) (top) CryoEM structure of Fab favezelimab(mut) in the complex with LAG3 (not shown) with elbow region between VH and CH1 framed in black. The elbow angle is 229°. The elbow region between VL and CL is framed in red. (middle) Close-up view of residues forming ball-and-socket joint in Fab favezelimab(wt) but rearranged in the Fab favezelimab(mut)–LAG3 complex. (bottom) Close-up view of the residues forming the elbow region between VL and CL.

Structure of the Fab favezelimab(mut)–LAG3 complex

The Fab favezelimab(mut)–LAG3 complex was purified by SEC from a mixture of Fab and LAG3 and eluted as a peak of ~220 kDa, which is consistent with a dimeric Fab with two bound LAG3 molecule of ~55 kDa each (Figure 1A). By contrast, we previously showed that the complex of Fab favezelimab(wt) and LAG3 eluted as a single peak of ~100 kDa,18 as expected for a monomeric Fab with one bound LAG3 molecule. We determined the structure of the Fab favezelimab(mut)–LAG3 complex using single-particle cryoEM at 3.5 Å resolution (Table S2) (Figure S4) (Figure 3A, C).

Figure 3: CryoEM structure of Fab favezelimab(mut)–LAG3 complex.

Figure 3:

Open in a new tab

(A) Final reconstructed cryoEM map using cryoSPARC.43

(B) Overall structure of Fab favezelimab(mut)–LAG3 complex. LAG3 monomers are yellow and green; L chains of dimeric Fab are magenta and gray; H chains are cyan and salmon. </p/>(C) EM map of the interface between LAG3 and favezelimab(mut). </p/>(D) Interactions between LAG3 and favezelimab(mut). The side chains of interacting residues are shown in stick representation with carbon atoms in yellow (LAG3), gray (VL), or salmon (VH), nitrogen atoms in blue, and oxygen atoms in red. Hydrogen bonds are indicated by black dashed lines.

In the Fab favezelimab(mut)–LAG3 complex, each Fab engages one LAG3 to form a symmetrical assembly in which the Fabs exclusively contact the membrane-distal Ig-like D1 domain of LAG3 (Figure 3B), in agreement with previous epitope mapping.18 Whereas D1 and D2 were well defined and folded into discrete Ig-like domains, little or no features were visible for D3 or D4, which we interpret as reflecting flexibility in the linker connecting D2 and D3. By contrast, in the crystal structure of human LAG3,9 all four Ig-like domains were ordered, likely due to capture of a particular LAG3 conformation by crystal lattice contacts. Superposition of D1 in the cryoEM structure onto D1 in the crystal structure gave an r.m.s.d. of 1.4 Å for 77 α-carbon atoms; superposition of the D2 domains gave an r.m.s.d. of 1.0 Å for 90 α-carbon atoms (Figure S6B). Thus, taken individually, D1 and D2 have very similar conformations in the cryoEM and crystal structures. However, superposition of the tandem D1D2 domains in the cryoEM structure onto D1D2 in the crystal structure gave a higher r.m.s.d. (3.2 Å) due to slightly different orientations of D1 relative to D2 across the D1–D2 linker (Figure S6C). The D1D2 domains contain three potential N-glycosylation sites at Asn188, Asn250, and Asn256. Electron density corresponding to one N-acetylglucosamine residue was detected at each of these sites (Figure S5A).

In the Fab favezelimab(mut)–LAG3 structure, the D2 domains of the bound LAG3 molecules are in contact, burying 486 Å2 of surface area at the D2–D2 interface to form a parallel dimer (Figure 3B). Electron density at the D2–D2 interface is shown in Figure S5B. This V-shaped dimer resembles that in the human LAG3 crystal structure9 but is somewhat wider due to a twist angle of 28° instead of 24° (Figure S7A, B). For comparison, the twist angle of the mouse LAG3 dimer is 70°,9 giving rise to an even wider V-shaped architecture (Figure S7C). Dimerization of human LAG3 in the cryoEM and crystal structures is mediated by the same cluster of hydrophobic residues (Trp184, Ile186, Phe225, and Phe227); however, these make different interactions across the D2–D2 interface in the two structures, possibly due to favezelimab binding (Figure S7A, B).

Remarkably, Fab favezelimab(mut) adopts a very different conformation in complex with LAG3 compared to its unbound conformation in the crystal (Figure 4). As calculated by DynDom,23 the angle of rotation around the switch peptide of the LAG3-bound favezelimab(mut) H chain is 49° with respect to unbound favezelimab(mut) (Figure S2A). The angle of rotation around the switch peptide of the LAG3-bound favezelimab(mut) L chain is 85° with respect to unbound favezelimab(mut) (Figure S2B). In addition, the elbow angle25 between VH and CH1 domains is 229° for LAG3-bound Fab favezelimab(mut) compared to 217° for unbound Fab favezelimab(mut) (Figure 2C). Thus, LAG3 binding induces major structural rearrangements in the bivalent Fab that probably help stabilize a dimeric form of LAG3, which is monomeric in solution (Figure 1A).

Figure 4. Conformation of unbound versus LAG3-bound Fab favezelimab(mut).

Figure 4.

Open in a new tab

(A) (left) The V domains of unbound bivalent Fab favezelimab(mut) are 61 Å apart. (right) The C domains are separated by 54 Å. The distance between V domains is between the α-carbons of VLLeu5 and VHSer60; the distance between C domains is between CLSer160 and CH1Ser176.

(B) In the LAG3-bound state, the V and C domains of Fab favezelimab(mut) are separated by 24 Å and 46 Å, respectively.

Each Fab of the Fab favezelimab(mut) dimer makes very similar interactions with LAG3 in the overall complex. Thus, superposition of the two LAG3–VL/VH sub-complexes gave an r.m.s.d. in α-carbon positions of 0.65 Å (Figure S5C). The largest divergences occur at LAG3 residues 151–157 and 161–166 and VH residues 40–45. However, these sites are away from the binding interface and there are no significant differences in LAG3–VL/VH interactions in the two sub-complexes. The Fab favezelimab(mut)–LAG3 complex buries a total solvent-accessible surface area of 877 Å2, with 16 antibody residues (6 VL and 10 VH) contacting 11 LAG3 residues (Table S3) (Figure 3D). VH contributes 607 Å2 to the buried surface on LAG3 compared to 270 Å2 by VL. Overall, VH makes 4 hydrogen bonds and 49 van der Waals contacts with LAG3 compared with 5 hydrogen bonds and 18 van der Waals contacts by VL. The main target of favezelimab is the base of a proline-rich 25-residue loop that connects the C and C’ β-strands of the D1 domain (designated the D1 loop; residues 74–98) (Figure 3D). The D1 loop is not found in CD4 and constitutes at least part of the binding site for MHC class II, as demonstrated by cell–cell adhesion assays using LAG3 mutants.26 Favezelimab interactions with LAG3 are focused on three arginine residues (Arg95, Arg97, and Arg98) located at the C-terminal base of the D1 loop. These residues were not visible in the crystal structure of LAG3,9 suggesting they were stabilized by favezelimab binding. The rest of the D1 loop was disordered in both crystal and cryoEM structures, implying flexibility.

Favezelimab forms a dense network of five hydrogen bonds with D1 loop residues Arg95, Arg97, and Arg98, all via VL: VLCDR1 Tyr27 O–Nη2 Arg98 LAG3, VLCDR1 Tyr27 Oη–O Arg98 LAG3, VLCDR1 Asp30 O–Nη2 Arg98 LAG3, VLCDR3 Thr92 O–Nη2 Arg97 LAG3, and VLCDR3 Asp94 Oδ1–Nη1 Arg95 LAG3 (Table S3) (Figure 3D). In contrast to VL, which only engages the D1 loop, VH interacts mainly with LAG3 residues in the main body of the D1 domain, notably Asp68, Ala70, Glu146, and Arg148. The structure is consistent with epitope mapping showing that favezelimab recognizes determinants both within and outside the D1 loop.18 The structure also explains the exquisite specificity of favezelimab for human, but not mouse, LAG3.17 Human and mouse LAG3 differ at antibody-contacting residues Arg(human)/Ala95(mouse), Arg/Gly97, and Arg/His148, of which Arg95 and Arg97 in the D1 loop make extensive interaction with VL (Table S3) (Figure 3D).

DISCUSSION

LAG3, an inhibitory receptor on T cells with structural similarities to the co-receptor CD4, synergizes with PD1 in suppressing anti-tumor immunity. The simultaneous blockade of PD1 and LAG3 using specific antibodies accelerates the eradication of tumors in mouse tumor models.15,16 Therefore, LAG3 is a therapeutic target in multiple clinical trials for cancer involving nearly 20 anti-LAG3 mAbs.4,5 One of these mAbs (relatlimab) recently received FDA approval for treatment of melanoma,6 while favezelimab is under investigation against various solid tumors and lymphomas.4,5,19 Both relatlimab and favezelimab target the D1 domain of LAG318 and block binding to MHC class II,17,27 its canonical ligand. Indeed, these two characteristics probably apply to most current therapeutic anti-LAG3 mAbs.18 These mAbs recognize four distinct epitopes on D1 that completely or partially overlap the MHC class II binding site, which includes the D1 loop.18,27 We previously found that replacement of the D1 loop of mouse LAG3 (which is not recognized by relatlimab) by the D1 loop of human LAG3 generated a chimera that bound relatlimab as well as human LAG3, indicating that relatlimab almost exclusively targets the D1 loop.18 In the Fab favezelimab–LAG3 structure, favezelimab targets the C-terminal base of the 25-residue D1 loop, as well as determinants outside the D1 loop. Since relatlimab and favezelimab do not compete for binding to LAG318, the relatlimab epitope must reside elsewhere on the D1 loop than at its C-terminal base. However, we cannot further localize this epitope in the absence of a relatlimab–LAG3 structure.

We previously showed that domain-swapped Fab dimers can simultaneously bind two protein ligands in solution with the same affinity as Fab monomers.22 In addition, although Fab dimers are somewhat less thermally stable than Fab monomers, dimeric Fab–ligand complexes are as stable as monomeric Fab–ligand complexes. Here we demonstrate the utility of a dimeric Fab as a fiducial marker for structure determination of LAG3, a small ~55 kDa protein, by cryoEM.

Structure determination of small proteins is of considerable interest since most proteins are smaller than 100 kDa and many are below 50 kDa. However, even with recent technical advances, such as the use of Volta phase plates, cryoEM becomes increasingly challenging as protein size decreases. Large proteins are relatively easy to identify in cryoEM micrographs and have sufficient features to determine their orientation for image alignment. By contrast, small proteins (<100 kDa) often lack recognizable shape features to facilitate particle selection and orientation for processing cryoEM images. Conventional (i.e. monomeric) Fab fragments can serve as fiducial markers by adding mass (50 kDa) to the target protein and assisting in its orientation.21,28–30 However, the intrinsic flexibility of the elbow regions linking V and C domains may reduce their ability to orient particles accurately.21 One approach to overcoming this problem involves engineering the elbow region of the H chain to lock the Fab in a defined conformation.31 More recently, target-binding camelid antibodies (nanobodies) were rigidly attached to two scaffolds: 1) the Fab of an antibody directed against the nanobody and 2) a nanobody-binding protein protein A fragment fused to maltose binding protein.32 These fiducial markers (Legobodies) of ~120 kDa overall size were used to determine cryoEM structures of two small proteins, the KDEL receptor (23 kDa) and the receptor-binding domain of SARS-CoV-2 spike protein (22 kDa). In a third approach, a Fab was rigidified by binding a nanobody to the hinge region, yielding a stable and large complex amenable for structure determination of ~50 kDa membrane proteins by cryoEM.33

We engineered a dimeric bivalent Fab that doubles the effective size of its complex with the target protein and confers a highly identifiable shape to the complex. Although the V–C elbow of the bivalent Fab retained flexibility, as shown by the markedly different conformations of its unbound and LAG3-bound forms, the entire dimeric Fab was clearly defined in EM maps of the Fab–LAG3 complex. This proof-of-concept study demonstrates the utility of bivalent Fabs as fiducial markers for cryoEM. LAG3, our target protein, comprises four Ig-like domains (D1–D4), of which only two (D1D2) were visible in the map due to flexibility in the linker connecting D2 and D3. Therefore, the effective size of each LAG3 molecule bound to the bivalent Fab is not ~55 kDa, but only ~28 kDa, which is comparable to the size of proteins analyzed using Legobodies.32

Legobodies and other nanobody-based fiducial markers employ target-binding nanobodies, which must be isolated from immunized camelids or from phage or yeast display libraries on a case-by-case basis.34,35 This can present potential challenges for broad applications of such antibody modules in structure determination. By contrast, thousands of conventional antibodies recognizing thousands of target proteins already exist, especially against proteins of biomedical interest such as LAG3. For example, the ABCD (AntiBodies Chemically Defined) Database (https://web.expasy.org/abcd/) contains 23,457 sequenced antibodies against 4,125 different targets (May 2022).

Mechanisms to explain domain swapping in proteins typically focus on the region that connects the exchanging domains (called the hinge loop), since this is the only part of domain-swapping proteins that adopts different conformations in monomeric versus oligomeric states36,37. In general, shortening a hinge loop favors domain swapping, as demonstrated here by deleting CH114Ala and CH115Ser in the hinge loop (elbow region) of the Fab favezelimab H chain. A short loop makes it difficult for a polypeptide to fold back on itself and permits the swapped portion of the chain to more readily locate partners.36,37 To engineer Fab dimers of different antibodies, we have investigated several deletions besides deletion of CH114Ala and CH115Ser. In the case of HC84.26.5D, deletion of VHSer113 induced dimerization,22 while other antibodies required deletion of VH112 and VH113 to drive domain swapping. In two cases, dimerization required deletion of three (VH112Ser–CH114Ala) or four (VH112Ser–CH115Ser) residues. However, for most antibodies, we recommend deleting VH112 and VH113 or CH114Ala and CH115Ser. Importantly, all these residues are in the elbow region of the H chain, at the opposite end of the VH domain from the antigen-binding CDR loops (Figure 2). Moreover, these residues are either invariant (CH114Ala and CH115Ser) or highly conserved (VH112 and VH113) in antibody sequences.

We identified the LAG3 epitope targeted by the therapeutic mAb favezelimab and showed that this epitope overlaps the binding site for MHC class II, thereby explaining the ability of favezelimab to prevent LAG3-mediated downregulation of T cell activation.19 We anticipate that structural information on the different epitopes recognized by other anti-LAG3 mAbs will allow structure-function analysis of their specific effects on LAG3 activity. In addition, we introduced bivalent Fabs as a new class of fiducial markers that can overcome current target size limitations of single-particle cryoEM. Because construction of bivalent Fabs is relatively easy and only requires knowing the sequence of the relevant mAb, bivalent Fabs should contribute to making structure determination of small proteins by cryoEM a routine method.

STAR METHODS

RESOURCE AVAILABILITY

Lead Contact

Roy A. Mariuzza (rmariuzz@umd.edu)

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Roy A. Mariuzza (rmariuzz@umd.edu).

Materials Availability

Plasmids generated in this study are available upon request from the lead contact.

Data and Code Availability

  • X-ray crystallographic data have been deposited in the Protein Data Bank (PDB). CryoEM data have been deposited in the PDB and the Electron Microscopy Data Bank (EMDB).

  • This paper does not report original code.

  • Any additional information required to reanalyze the data reported in this paper is available from the lead contact.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Plasmids were propagated in Escherichia coli strain DH5α. Recombinant proteins were expressed in Expi293F cells.

METHOD DETAILS

Protein expression and purification

The VL and VH sequences of the anti-hLAG3 mAb favezelimab (22D2) were obtained from the patent literature.17 To produce monovalent or bivalent Fab fragments, codon-optimized genes encoding the VH and CH1 domains of the H chain and the VL and CL domains of the L chain were synthesized chemically (GeneArt) and cloned into the mammalian expression vector pcDNA3.4-TOPO. Both chains included an N-terminal immunoglobulin κ signal sequence for secretion. A streptavidin tag (WSHPQFEK) was attached to the C-terminus of the CH1 domain for affinity purification. The gene encoding the VHCH1 region of bivalent Fab favezelimab(mut) contained a deletion of two amino acids (CH114Ala and CH115Ser) in the elbow region between VH and CH1 compared to the wild-type monovalent Fab. Both monovalent and bivalent Fab favezelimab were produced by co-transfecting equimolar amounts of H and L chain plasmids into Expi293 cells with Expifectamine (ThermoFisher). After 96 h incubation, the cells were harvested by centrifugation and supernatants dialyzed against 100 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 1 mM EDTA. Recombinant Fabs were purified using consecutive streptavidin affinity (IBA Lifesciences) and Superdex 200 (GE Healthcare) columns. Size exclusion chromatography (SEC) was carried out in 50 mM Tris-HCl (pH 8.0) and 50 mM NaCl. These materials were used for both X-ray crystallography and cryoEM. The percentage of monomeric versus dimeric Fab favezelimab(mut) in SEC was calculated by measuring peak areas in the absorbance curve. The yield of dimeric Fab favezelimab(mut) was typically ~10 mg/l culture, similar to the monomeric Fab.

A codon-optimized gene encoding the four Ig-like extracellular domains (D1–D4; residues 1–449) of human LAG3 was cloned into pcDNA3.4-TOPO with a C-terminal His6 tag for affinity purification. The plasmid was transfected into Expi293 cells with Expifectamine. After 72 h incubation, soluble LAG3 was purified from culture supernatants using sequential HisTrap excel IMAC (Cytiva) and Superdex 200 columns.

Crystallization and data collection

Crystals of monomeric and dimeric Fab favezelimab were obtained at room temperature by vapor diffusion in hanging drops. Monomeric Fab favezelimab(wt) (10 mg/ml) crystallized in 0.1 M sodium acetate dibasic/citric acid (pH 4.2), 0.2 M sodium chloride, and 20% (w/v) polyethylene glycol (PEG) 8000. Crystals of dimeric Fab favezelimab(mut) (8 mg/ml) grew in 0.1 M Tris-HCl (pH 7.5), 0.2 M sodium chloride, and 10% (w/v) PEG 8000. For data collection, crystals were cryoprotected with 20% (w/v) ethylene glycol and flash-cooled in liquid nitrogen. X-ray diffraction data were collected at beamline 23-ID-B of the Advanced Photon Source, Argonne National Laboratory. Diffraction data to 2.1 Å and 2.8 Å resolution for monomeric and dimeric Fab favezelimab, respectively, were indexed, integrated, and scaled using the program HKL3000.38 Data collection statistics are shown in Table S1.

Crystal structure determination and refinement

The structure of monomeric Fab favezelimab(wt) was solved by molecular replacement with the program Phaser39 using a CD25-binbing Fab fragment (PDB accession code 1MIM)40 as a search model. Monomeric Fab favezlimab(wt) was used as a search model to determine the structure of dimeric Fab favezelimab(mut). Manual model building and refinement were performed using Coot41 with iterative cycles of refinement using REFMAC5.42 Water molecules were added manually at 1σ contour level for 2Fo – Fc and at 3σ contour level for Fo – Fc in the electron density maps. Refinement statistics are summarized in Table S1.

Single-particle cryoEM analysis

Dimeric Fab favezelimab(mut)–LAG3 complex was purified with a Superdex 200 column from an equimolar mixture of Fab and LAG3. Complex at 1 mg/ml total protein concentration was used to prepare specimens for cryoEM characterization. Briefly, 4 μl of protein sample was applied to glow-discharged QUANTIFOIL R1.2/1.3 (300 mesh) grids. The grids were blotted for 4 s and vitrified by plunging into liquid ethane using an FEI Vitrobot Mark IV set at 4 °C and 100% humidity. Grids were screened for quality using an in-house FEI 200 kV Arctica microscope equipped with a Falcon 3EC direct electron detector (DED).

Data collection was performed with an FEI 300 kV Titan Krios microscope equipped with a Gatan K3 DED in counting mode at the National Cryo-Electron Microscopy Facility (NCEF) in Frederick, MD. A total of 5,818 movies were recorded at –1.0 to –2.5 μm defocus at a physical pixel size of 1.12 Å. An exposure time of 3.8 sec was used for each movie, with a total dose of 50 e−/Å2. CryoSPARC v443 was used to process data for the complex. The movies were subjected to patch motion correction and patch CTF estimation. Low quality micrographs were eliminated after using stringent criteria for CTF fit resolution up to 5 Å, relative ice thickness, and total full frame motion. Fab–LAG3 particles from 4,586 micrographs were then selected for particle picking using a box size of 320 Å. To generate the initial templates, particles were extracted with a binning factor of 4X and a box size of 4.48 Å (corresponding to a pixel size of 1.12 Å/pixel). The 2D class averages with prominent protein features were used to generate templates to extract Fab–LAG3 particles from the complete dataset. These particles were subsequently unbinned and subjected to iterative rounds of 2D classification to eliminate contaminants and damaged particles. Eventually, a total of 788,162 particles were extracted at full resolution, and were used for ab initio reconstruction (K=5) in C2 symmetry and an initial alignment resolution of 8 Å. The best resolved classes were then subjected to heterogeneous refinement, followed by homogenous, non-uniform, and CTF refinement of each individual class in C2 symmetry. This yielded a cryoEM map with a resolution of 3.4 Å based on the gold standard Fourier shell correlation (FSC) with a criterion of 0.143. Upon examination of the C2 symmetry map, it was evident that certain loops and side chains were absent. Additionally, the symmetry had been relaxed from C2 to C1, resulting in a resolution of 3.6 Å as determined by the gold standard FSC with a threshold of 0.143. Despite the decreased resolution, the map clearly revealed the presence of the previously missing loops and side chains in the C1 symmetry (Figure S6A). The final reconstructed Fab favezelimab(mut)–LAG3 complex comprised two LAG3 molecules showing features for D1 and D2 domains, bound to two domain-swapped Fab molecules (Figure S4).

To delineate the Fab–LAG3 interactions, local refinement was performed using a soft mask with two LAG3 and two variable regions of the favezelimab(mut) molecule with final reconstruction at 3.5 Å based on FSC curve (Figure S4). Model building was carried out by manually docking the X-ray structures of dimeric Fab favezelimab(mut) and LAG39 into the EM map using UCSF Chimera (https://www.cgl.ucsf.edu/chimera/) to generate a starting model, followed by manual rebuilding using Coot41 and refinement with Phenix real space refine (https://phenix-online.org/). Contact residues were identified with the CONTACT program44 and were defined as residues containing an atom 4.0 Å or less from a residue of the binding partner.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data collection and refinement statistics for the structures reported in this article are summarized in Table S1 and Table S2.

Supplementary Material

1

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER
Bacterial and virus strains
DH5α competent cells Invitrogen Cat#18265017
Biological samples
Chemicals, peptides, and recombinant proteins
Expi293 expression medium ThermoFisher Cat#A1435101
Expifectamine ThermoFisher Cat#A14524
Strep-TactinXT 4Flow resin IBA Lifesciences Cat#2-5010-025
HisTrap excel Cytiva Cat#17371205
Superdex 200 Cytiva Cat#28-9909-44
Tris base Sigma-Aldrich Cat#93352
Sodium chloride Sigma-Aldrich Cat#S9888
EDTA Invitrogen Cat#15575020
Sodium acetate Sigma-Aldrich Cat#241245
Polyethylene glycol (PEG) 8000 Hampton Research Cat#HR2-535
Ethylene glycol Hampton Research Cat#HR2-621
Critical commercial assays
Deposited data
Structure of favezelimab(wt) This paper PDB ID 6WKM
Structure of favezelimab(mut) This paper PDB ID 8FWH
Structure of Fab favezelimab(mut)–LAG3 complex (full refined) This paper PDB ID 8SO3; EMD ID EMD-40646
Structure of Fab favezelimab(mut)–LAG3 complex (local refined) This paper PDB ID 8SR0; EMD ID EMD-40716
Experimental models: Cell lines
Expi293F cells ThermoFisher Cat#A14527
Experimental models: Organisms/strains
Oligonucleotides
5’-CATCTGTGACCGTGGCCAGCCAAAGGGC-3’ (DSS forward) Invitrogen NA
5’-GCCCTTTGGCTGGCCACGGTCACAGATG-3’ (DSS reverse) Invitrogen NA
5’-GCCATGGATCACTGGGGCCAGGGAAATCTGTGACCGTGTCTAGCGCCAGCCAAAGGGCCCCTCTGTGTTTCCTCTGGCTCCCAG-3’ (SS deletion) Invitrogen NA
Recombinant DNA
pcDNA3.4-TOPO Invitrogen Cat#A14697
Software and algorithms
DynDom (Lee et al., 2003)23 http://dyndom.cmp.uea.ac.uk/dyndom/
PDBePISA (Krissinel and Henrick, 2007)24 https://www.ebi.ac.uk/pdbe/pisa/
HKL3000 (Minor et al., 2006)38 https://hkl-xray.com/hkl-3000
Phaser (Storoni et al., 2004)39 http://www.ccp4.ac.uk/
Coot (Emsley et al., 2010)41 http://www2.mrc-lmb.cam.ac.uk/Personal/pemsley/coot/
REFMAC5 (Murshudov et al., 2011)42 http://www.ccp4.ac.uk/
CryoSPARC v4 (Punjani et al., 2017)43 https://cryosparc.com/
UCSF Chimera NA https://www.cgl.ucsf.edu/chimera/
Phenix (Afonine et al., 2012)45 http://www.phenix-online.org/
CONTACT (CCP4, 1994)44 http://www.ccp4.ac.uk/
PyMOL Molecular Graphics System (Schrodinger, LLC) http://pymol.org/
Other
QUANTIFOIL R1.2/1.3 (300 mesh) grids Electron Microscopy Sciences Cat#Q325CR1.3
FEI Vitrobot Mark IV ThermoFisher NA
Open in a new tab

Highlights.

  • LAG3 is a major target for cancer immunotherapy using anti-LAG3 antibodies.

  • We determined the cryoEM structure of LAG3 bound to a therapeutic antibody.

  • Structure determination required engineering a bivalent Fab to increase complex size.

  • Bivalent Fabs are new fiducial markers for cryoEM analysis of small proteins.

ACKNOWLEDGMENTS

This work was supported by National Institutes of Health Grant AI144422 to R.A.M. X-ray diffraction data were collected at the GM/CA beamline at the Advanced Photon Source of Argonne National Laboratory, which is funded by the National Cancer Institute (ACB-12002) and the National Institute of General Medical Sciences (AGM-12006, P30GM138396). CryoEM data were collected at the National Cryo-EM Facility (NCEF) at the Frederick National Laboratory for Cancer Research under contract HSSN261200800001E. We thank Adam Wier (NCEF) for cryoEM data collection. We also thank Edwin Pozharski (University of Maryland) for advice on cryoEM image analysis. S.S.H. acknowledges support from University of Maryland School of Medicine (UMSOM), UMSOM Center for Biomolecular Therapeutics, and University of Maryland Marlene and Stewart Greenebaum Cancer Center.

Footnotes

DECLARATION OF INTERESTS

The authors declare no competing interests.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

REFERENCES

  • 1.Topalian SL, Drake CG, and Pardoll DM (2015). Immune checkpoint blockade: a common denominator approach to cancer therapy. Cancer Cell 13, 450–461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Hodi FS, O’Day SJ, McDermott DF, Weber RW, Sosman JA, Haanen JB, Gonzalez R, Robert C, Schadendorf D, Hassel JC, et al. (2010). Improved survival with ipilimumab in patients with metastatic melanoma. N. Engl. J. Med 363, 711–723. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Larkin J, Chiarion-Sileni V, Gonzalez R, Grob JJ, Rutkowski P, Lao CD, Cowey CL, Schadendorf D, Wagstaff J, Dummer R, et al. (2019). Five-year survival with combined nivolumab and ipilimumab in advanced melanoma. N. Engl. J. Med 381, 1535–1546. [DOI] [PubMed] [Google Scholar]
  • 4.Andrews LP, Cillo AR, Karapetyan L, Kirkwood JM, Workman CJ, and Vignali DAA (2022). Molecular pathways and mechanisms of LAG3 in cancer therapy. Clin. Cancer Res 28, 5030–5039. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Chocarro L, Blanco E, Arasanz H, Fernández-Rubio L, Bocanegra A, Echaide M, Garnica M, Ramos P, Fernández-Hinojal G, Vera R, et al. (2022). Clinical landscape of LAG-3-targeted therapy. Immunooncol. Technol 14, 100079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Tawbi HA, Schadendorf D, Lipson EJ, Ascierto PA, Matamala L, Castillo Gutiérrez E, Rutkowski P, Gogas HJ, Lao CD, et al. ; RELATIVITY-047 Investigators. (2022). Relatlimab and nivolumab versus nivolumab in untreated advanced melanoma. N. Engl. J. Med 386, 24–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Andrews LP, Yano H, and Vignali DAA (2019). Inhibitory receptors and ligands beyond PD-1, PD-L1 and CTLA-4: breakthroughs or backups. Nat. Immunol 20, 1425–1434. [DOI] [PubMed] [Google Scholar]
  • 8.Triebel F, Jitsukawa S, Baixeras E, Roman-Roman S, Genevee C, Viegas-Pequignot E, and Hercend T. (1990). LAG-3, a novel lymphocyte activation gene closely related to CD4. J. Exp. Med 171, 1393–405. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Ming Q, Celias DP, Wu C, Cole AR, Singh S, Mason C, Dong S, Tran TH, Amarasinghe GK, Ruffell B, et al. (2022). LAG3 ectodomain structure reveals functional interfaces for ligand and antibody recognition. Nat. Immunol 23, 1031–1041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Maruhashi T, Okazaki IM, Sugiura D, Takahashi S, Maeda TK, Shimizu K, and Okazaki T. (2018). LAG-3 inhibits the activation of CD4+ T cells that recognize stable pMHCII through its conformation-dependent recognition of pMHCII. Nat. Immunol 19, 1415–1426. [DOI] [PubMed] [Google Scholar]
  • 11.MacLachlan BJ, Mason GH, Greenshields-Watson A, Triebel F, Gallimore A, Cole DK, and Godkin A. (2021). Molecular characterization of HLA class II binding to the LAG-3 T cell co-inhibitory receptor. Eur. J. Immunol 51, 331–341. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Wang J, Sanmamed MF, Datar I, Su TT, Ji L, Sun J, Chen L, Chen Y, Zhu G, Yin W, et al. (2019). Fibrinogen-like protein 1 is a major immune inhibitory ligand of LAG-3. Cell 176, 334–347. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Mao X, Ou MT, Karuppagounder SS, Kam TI, Yin X, Xiong Y, Ge P, Umanah GE, Brahmachari S, Shin JH, et al. (2016). Pathological α-synuclein transmission initiated by binding lymphocyte-activation gene 3. Science 353, aah3374. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Xu F, Liu J, Liu D, Liu B, Wang M, Hu Z, Du X, Tang L, and He F. (2014). LSECtin expressed on melanoma cells promotes tumor progression by inhibiting antitumor T-cell responses. Cancer Res. 74, 3418–3428. [DOI] [PubMed] [Google Scholar]
  • 15.Anderson AC, Joller N, and Kuchroo VK (2016). Lag-3, Tim-3, and TIGIT: co-inhibitory receptors with specialized functions in immune regulation. Immunity 44, 989–1004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Woo SR, Turnis ME, Goldberg MV, Bankoti J, Selby M, Nirschl CJ, Bettini ML, Gravano DM, Vogel P, Liu CL, et al. (2012). Immune inhibitory molecules LAG-3 and PD-1 synergistically regulate T-cell function to promote tumoral immune escape. Cancer Res. 72, 917–927. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Liang L, Fayadat-Dilman L, De Waal Malefyt R, and Raghunathan G, inventors; Merck Sharp & Dohme Corp., assignee. Anti-LAG3 antibodies and antigen-binding fragments United States patent application PCT/US2015/045481, Publication No. 2016028672A1. 25 February 2016.
  • 18.Agnihotri P, Mishra AK, Agarwal P, Vignali KM, Workman CJ, Vignali DAA, and Mariuzza RA (2022). Epitope mapping of therapeutic antibodies targeting human LAG3. J. Immunol 209, 1586–1594. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Bhagwat B, Cherwinski H, Sathe M, Seghezzi W, McClanahan TK, de Waal Malefyt R, and Willingham A. (2018). Establishment of engineered cell-based assays mediating LAG3 and PD1 immune suppression enables potency measurement of blocking antibodies and assessment of signal transduction. J. Immunol. Methods 456, 7–14. [DOI] [PubMed] [Google Scholar]
  • 20.Johnson NA, Lavie D, Borchmann P, Gregory G, Francisco Herrera A, Minuk L, Vucinic V, Armand P, Avigdor A, Gasiorowski R, et al. Favezelimab (anti-LAG-3) plus pembrolizumab in patients with anti-PD-1-naïve relapsed or refractory (R/R) classical Hodgkin lymphoma (cHL): An open-label phase 1/2 study. J. Clin. Oncol 40, no. 16_suppl (June 01, 2022) 7516. [Google Scholar]
  • 21.Wu S, Avila-Sakar A, Kim J, Booth DS, Greenberg CH, Rossi A, Liao M, Li X, Alian A, Griner SL, Juge N, et al. (2012). Fabs enable single particle cryoEM studies of small proteins. Structure 20, 582–592. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Shahid S, Gao M, Gallagher DT, Pozharski E, Brinson RG, Keck ZY, Foung SKH, Fuerst TR, and Mariuzza RA (2021). Crystal structure of a bivalent antibody Fab fragment. J. Mol. Biol 433, 166714. [DOI] [PubMed] [Google Scholar]
  • 23.Lee RA, Razaz M, and Hayward S. (2003). The DynDom database of protein domain motions. Bioinformatics 19, 1290–1291. [DOI] [PubMed] [Google Scholar]
  • 24.Krissinel E, and Henrick K. (2007). Inference of macromolecular assemblies from crystalline state. J. Mol. Biol 372, 774–797. [DOI] [PubMed] [Google Scholar]
  • 25.Lesk AM, and Chothia C. (1988). Elbow motion in the immunoglobulins involves a molecular ball-and-socket joint. Nature 335, 188–190. [DOI] [PubMed] [Google Scholar]
  • 26.Huard B, Mastrangeli R, Prigent P, Bruniquel D, Donini S, El-Tayar N, Maigret B, Dréano M, and Triebel F. (1997). Characterization of the major histocompatibility complex class II binding site on LAG-3 protein. Proc. Natl. Acad. Sci. USA 94, 5744–5749. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Thudium K, Selby M, Zorn JA, Rak G, Wang XT, Bunch RT, Hogan JM, Strop P, and Korman AJ (2022). Preclinical characterization of relatlimab, a human LAG-3-blocking antibody, alone or in combination with nivolumab. Cancer Immunol. Res 10, 1175–1189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Lee Y, Wiriyasermkul P, Jin C, Quan L, Ohgaki R, Okuda S, Kusakizako T, Nishizawa T, Oda K, Ishitani R, et al. (2019). Cryo-EM structure of the human L-type amino acid transporter 1 in complex with glycoprotein CD98hc. Nat. Struct. Mol. Biol 26, 510–517. [DOI] [PubMed] [Google Scholar]
  • 29.Bloch JS, Pesciullesi G, Boilevin J, Nosol K, Irobalieva RN, Darbre T, Aebi M, Kossiakoff AA, Reymond JL, and Locher KP (2020). Structure and mechanism of the ER-based glucosyltransferase ALG6. Nature 579, 443–447. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Kim J, Tan YZ, Wicht KJ, Erramilli SK, Dhingra SK, Okombo J, Vendome J, Hagenah LM, Giacometti SI, Warren AL, et al. (2019). Structure and drug resistance of the Plasmodium falciparum transporter PfCRT. Nature 576, 315–320. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Bailey LJ, Sheehy KM, Dominik PK, Liang WG, Rui H, Clark M, Jaskolowski M, Kim Y, Deneka D, Tang WJ, et al. (2018). Locking the elbow: improved antibody Fab fragments as chaperones for structure determination. J. Mol. Biol 430, 337–347. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Wu X, and Rapoport TA (2021). Cryo-EM structure determination of small proteins by nanobody-binding scaffolds (Legobodies). Proc. Natl. Acad. Sci. USA 118, e2115001118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Bloch JS, Mukherjee S, Kowal J, Filippova EV, Niederer M, Pardon E, Steyaert J, Kossiakoff AA, and Locher KP (2021). Development of a universal nanobody-binding Fab module for fiducial-assisted cryo-EM studies of membrane proteins. Proc. Natl. Acad. Sci. USA 118, e2115435118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Zimmermann I, Egloff P, Hutter CA, Arnold FM, Stohler P, Bocquet N, Hug MN, Huber S, Siegrist M, Hetemann L, et al. (2018). Synthetic single domain antibodies for the conformational trapping of membrane proteins. Elife 7, e34317. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.McMahon C, Baier AS, Pascolutti R, Wegrecki M, Zheng S, Ong JX, Erlandson SC, Hilger D, Rasmussen SGF, Ring AM, et al. (2018). Yeast surface display platform for rapid discovery of conformationally selective nanobodies. Nat. Struct. Mol. Biol 25, 289–296. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Liu Y, and Eisenberg D. (2002). 3D domain swapping: as domains continue to swap. Protein Sci. 11, 1285–1299. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Gronenborn AM (2009). Protein acrobatics in pairs--dimerization via domain swapping. Curr. Opin. Struct 19, 39–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Minor W, Cymborowski M, Otwinowski Z, and Chruszcz M. (2006). HKL-3000: the integration of data reduction and structure solution–from diffraction images to an initial model in minutes. Acta Crystallogr. D Biol. Crystallogr 62, 859–866. [DOI] [PubMed] [Google Scholar]
  • 39.Storoni LC, McCoy AJ, and Read RJ (2004). Likelihood-enhanced fast rotation functions. Acta Crystallogr. D Biol. Crystallogr 60, 432–438. [DOI] [PubMed] [Google Scholar]
  • 40.Mikol V. (1996). Structure of the fab fragment of SDZ CHI621: a chimeric antibody against CD25. Acta Crystallogr. D Biol. Crystallogr 52, 534–542. [DOI] [PubMed] [Google Scholar]
  • 41.Emsley P, Lohkamp B, Scott WG, and Cowtan K. (2010). Features and development of Coot. Acta Crystallogr. D Biol. Crystallogr 66, 486–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Murshudov GN, Skubák P, Lebedev AA, Pannu NS, Steiner RA, Nicholls RA, Winn MD, Long F, and Vagin AA (2011). REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr. D Biol. Crystallogr 67, 355–367. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Punjani A, Rubinstein JL, Fleet DJ, and Brubaker MA (2017). cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296. [DOI] [PubMed] [Google Scholar]
  • 44.Collaborative Computational Project No. 4. The CCP4 suite: programs for protein crystallography. (1994). Acta Crystallogr. D Biol. Crystallogr 50, 240–255. [DOI] [PubMed] [Google Scholar]
  • 45.Afonine PV, Grosse-Kunstleve RW, Echols N, Headd JJ, Moriarty NW, Mustyakimov M, Terwilliger TC, Urzhumtsev A, Zwart PH, and Adams PD (2012). Towards automated crystallographic structure refinement with phenix.refine. Acta Crystallogr. D Biol. Crystallogr 68, 352–367. [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

1
NIHMS1925708-supplement-1.pdf (1.8MB, pdf)

Data Availability Statement

  • X-ray crystallographic data have been deposited in the Protein Data Bank (PDB). CryoEM data have been deposited in the PDB and the Electron Microscopy Data Bank (EMDB).

  • This paper does not report original code.

  • Any additional information required to reanalyze the data reported in this paper is available from the lead contact.

RESOURCES