https://orcid.org/0000-0002-3598-3399
ABSTRACT
Camelid heavy-chain-only antibodies are a unique class of antibody that possesses only a single variable domain (termed VHH) for antigen recognition. Despite their apparent canonical mechanism of target recognition, where a single VHH domain binds a single target, an anti-caffeine VHH has been observed to possess 2:1 stoichiometry. Here, the structure of the anti-caffeine VHH/caffeine complex enabled the generation and biophysical analysis of variants that were used to better understand the role of VHH homodimerization in caffeine recognition. VHH interface mutants and caffeine analogs, which were examined to probe the mechanism of caffeine binding, suggested caffeine recognition is only possible with the VHH dimer species. Correspondingly, in the absence of caffeine, the anti-caffeine VHH was found to form a dimer with a dimerization constant comparable to that observed with VH:VL domains in conventional antibody systems, which was most stable near physiological temperature. While the VHH:VHH dimer structure (at 1.13 Å resolution) is reminiscent of conventional VH:VL heterodimers, the homodimeric VHH possesses a smaller angle of domain interaction, as well as a larger amount of apolar surface area burial. To test the general hypothesis that the short complementarity-determining region-3 (CDR3) may help drive VHH:VHH homodimerization, an anti-picloram VHH domain containing a short CDR3 was generated and characterized, which revealed it also existed as dimer species in solution. These results suggest homodimer-driven recognition may represent a more common method of VHH ligand recognition, opening opportunities for novel VHH homodimer affinity reagents and helping to guide their use in chemically induced dimerization applications.
Introduction
In the early 1990s, Hamers-Casterman et al. reported their discovery of a unique class of camelid heavy-chain-only (VHH) antibodies.Citation1 These antibodies lacked light chains and the first constant heavy-chain domain (CH1) typically found in conventional IgG antibodies. This unique architecture results in antigen recognition occurring through only a single VHH domain, as opposed to conventional antibodies that include contributions from two variable domains (VH and VL). To accommodate the absence of interactions with a light chain variable domain (VL), the VHH domain typically possesses several residue substitutions within the former VL interface (framework 2 region), including V37F/Y, G44E, L45R, and W47G (Kabat numbering), which are believed to help increase both solubility and stability of the single VHH domain.Citation2 These residues help differentiate the VHH domain from conventional VH domains. Additionally, VHH domains generally possess longer complementarity-determining region-3 (CDR3) loop lengths,Citation3,Citation4 when compared to their VH counterpart, which may serve as a surrogate for the VL domain found in conventional VH/VL heterodimers.Citation5–7
The variable domain of VHH antibodies have excellent biophysical characteristics, which make them attractive affinity reagents for life science applications, such as biological therapeutics, diagnostics and separation techniques.Citation8 Notably, VHH domains can be expressed as soluble entities, which possess affinity and specificity for their antigen that rivals conventional antibodies, despite possessing only half the binding interface.Citation5,Citation7 Their ability to achieve such high affinity for targets likely stems from a combination of the VHH’s small size and the larger repertoire of CDR loop conformations found in VHH domains, as compared to conventional antibody CDRs.Citation9 In addition, the generally longer CDR3 of VHH domains can expand the type of antigen interactions that are possible, such as CDR3-mediated penetration of enzyme active sitesCitation7 or grooves within G protein coupled receptors.Citation10
While conventional antibodies that target low molecular weight (hapten) molecules typically bind their target within the VH/VL interface, it is perhaps surprising that heavy-chain-only antibodies can bind hapten targets without a VL domain. Despite possessing only a single VH domain, anti-hapten VHH antibodies have been generated against low molecular weight targets, such as trinitrotoluene,Citation11 azo dye RR1,Citation2 and methotrexate.Citation12 With limited structural data on anti-hapten VHH complexes, less is known how VHH domains recognize small-molecule hapten targets. While VHH domains have been observed to bind haptens near the former VL interface,Citation2,Citation13,Citation14 analogous to conventional antibodies, less conventional hapten recognition has been observed in VHH domains targeting methotrexate,Citation15 the antibacterial/antifungal agent triclocarbanCitation16 and hormone stress cortisol,Citation17 where the hapten-binding pocket consists of a tunnel/cavity under CDR1.
Further supporting the notion that VHH domains possess a more diverse range of mechanisms of hapten recognition, an anti-caffeine VHH antibody,Citation18 belonging to the VHH1 family,Citation19 was observed to bind caffeine in a 2:1 stoichiometry.Citation20 Beyond serving as a potential diagnostic reagent, the anti-caffeine VHH has been explored as a chemically induced dimerization (CID) system, which can be used to control intracellular events, such as protein localization, signal pathways and transcription. Specifically, anti-caffeine VHH was used to create both transmembrane and cytosolic receptors in E. coli, whereby the anti-caffeine VHH domains controlled transcriptional regulators in a caffeine-dependent fashion.Citation21 Similarly, the anti-caffeine VHH domain was used as a CID system to control synthetic transcription factors and cell surface receptors, enabling transgene expression in designer cells to produce glucagon-like peptide 1 as a model Type-2 diabetes treatment in a mouse model.Citation22 A recent crystal structure of the 2:1 VHH/caffeine complex revealed a VHH homodimer possessing two-fold rotational symmetry around the caffeine ligand,Citation23 which provides an initial structural model of the interaction.
Here, we explore the biophysical basis of hapten recognition and the role of homodimerization for an anti-caffeine VHH. Isothermal titration calorimetry (ITC) and analytical ultracentrifugation (AUC) were used to investigate VHH mutants and caffeine analogs to decouple hapten binding from anti-caffeine VHH dimerization, which suggested VHH can form a homodimer in the absence of the caffeine ligand and that only the dimer state may bind caffeine. In addition, a 1.13 Å high-resolution x-ray structure of the caffeine-bound state was determined, allowing detailed characterization of the VHH homodimer, including comparisons to conventional VH/VL heterodimers. Notably, the structural consequences of the short CDR3 found in anti-caffeine VHH appears to expose a significantly hydrophobic surface, which likely helps drive VHH homodimerization. To test this hypothesis, an anti-picloram VHH domain containing a short CDR3 was identified, generated through over-expression, and characterized, where it was found to form a stable homodimer species. This result supports the role short CDR3 loops play in homodimerization and suggests VHH homodimers may be more common, which not only affects how such systems may be used in CID applications, but also opens opportunities for novel homodimer VHH affinity reagents.
Results
Previously, we uncovered a new mechanism of hapten recognition by an anti-caffeine VHH single-domain antibody where two identical anti-caffeine VHH domains formed an atypical 2:1 complex with its ligand, caffeine.Citation20 For such a complex, two potential models of caffeine recognition are: 1) a stepwise binding event (via a 1:1 caffeine/VHH complex), or 2) direct caffeine binding to the VHH homodimer. This original analysis of the caffeine/anti-caffeine VHH interaction using isothermal titration calorimetry and size exclusion chromatography favored a stepwise mechanism, as VHH dimerization was not observed in the absence of caffeine; however, neither model could be fully evaluated under the concentrations/conditions examined. Recently, Lesne et al. published the structure of anti-caffeine VHHCitation18 in complex with caffeine at 2.25 Å resolution,Citation23 which was deposited in the Protein Data Bank (PDB) after the initiation of our studies. Here, we report the x-ray crystal structure of the anti-caffeine VHH, generated through loop grafting (),Citation20 in complex with caffeine at 1.13 Å (, ). The structures provide molecular insight into the role of dimerization and help guide our study examining the mechanism of caffeine recognition by anti-caffeine VHH.
Figure 1. Amino acid sequences of VHH domains. VHH sequences include the anti-RNase A VHH “framework” (cAbrn05) domain,Citation5 the original anti-caffeine VHH by Ladenson et al. (Caff),Citation18 the anti-caffeine VHH produced through grafting CDR regions (Caffgraft)Citation20 examined in this study, an anti-picloram VHH (Picloram),Citation24 an anti-GFP VHH (GFP),Citation25 and an anti-HIV1 capsid protein C-terminal domain VHH (PDB ID: 2xV6). Grey highlighting: anti-caffeine residues introduced into segments of CDR loops of anti-RNase A VHH; underlined residues: anti-RNase A residues that remained post-grafting; bold italicized residues: residues mutated in this study to explore disruption of VHH homodimer.
Figure 2. Structure of caffeine/anti-caffeine VHH homodimer complex. (a) Homodimer stick structure highlighting caffine binding site and conserved water molecules within VHH:VHH interface. (b) Close up of caffeine binding site. Image is rotated 90° vertically from panel A. Y32 sidechains from both VHH domains, which are are shown in stick (foreground) and space fill (background), “sandwich” the caffine ligand. (c) Electron density map (2FoFc) of caffeine/CDR3-binding pocket with water mediated hydrogen bond. Map coutoured at 2σ. (d) Side and top view perspective of the VHH complex. VHH domains (white and gray); caffeine (magenta); interface mutations F47R (cyan), V100R (green), and Y(100B)R (orange); Kabat numbering is use for residue positions. Caffeine shown in magenta in all panels.
Table 1. Crystallography and refinement details for VHH/caffeine complex.
Thermodynamic model for caffeine recognition
The anti-caffeine VHH complex structure presented here provides a high-resolution roadmap to probe the role of dimerization in caffeine binding. A distinctive feature of the complex is the two parallel VHH domains oriented with two-fold rotational symmetry (). The VHH/VHH interface is largely composed of residues from each domain’s framework 2 region. While most of the interface surface involves hydrophobic interactions, the anti-caffeine dimer complex possesses several hydrogen bonds between the VHH domains (). CDR3 residue Y100B (Kabat numbering) faces inward toward the dimer interface, making a side chain-mediated hydrogen bond with Y100B from the second VHH domain. Y32, from CDR1, is observed in two (one from each domain) inter-domain hydrogen bonds with CDR3 residue S99. Additional hydrogen bonds are observed that are both longer and not consistently observed across all VHH dimers within the asymmetric unit. For example, a hydrogen bond is frequently observed between side chains Y58 and W103 in most VHH domains within the asymmetric unit. Overall, there is significant similarity between the two domains of the anti-caffeine homodimer within the asymmetric unit, as judged by Cα RMSD for the first and second domain, 0.33 and 0.44 Å, respectively, as well as comparing all monomer units together, Cα RMSD = 0.74 Å. (Supplemental Figure S1).
Table 2. Dimer interface hydrogen bond contacts.
Due to homodimer symmetry, the caffeine-binding pocket consists of both VHH domains contributing similar sidechain and mainchain groups from CDR3 and the C-terminal portion of CDR1. Residue Y32 (from each domain) appears to play a central role in forming the caffeine-binding site. The pyrimidine ring of caffeine is positioned approximately 3.3 Å away from each domain’s Y32, forming a pi-pi stacking interaction (). In addition, hydrogen bonds are observed between the main chain nitrogen of Y98 from each VHH domain to the caffeine’s 6 position carbonyl oxygen and the 2-position nitrogen, respectively.
Below the caffeine-binding pocket, 14 water molecules are sequestered within the interface of the VHH homodimer, with 12 occupying the same binding cavity as caffeine (). Notably, these 12 water molecules are conserved between our structure and that of Lesne et al.Citation23 Two of these waters are involved in hydrogen bonds bridging the 2-position carbonyl oxygen of caffeine and the side chains of S33 from each VHH domain (). Despite the solvent-filled cavity located below caffeine, the caffeine-binding pocket appears tightly packed, with 91% of the accessible surface area of the caffeine molecule buried within the VHH dimer. Caffeine’s three methyl substituents are poised for van der Waals interactions within primarily hydrophobic pockets. Correspondingly, these methyl groups were shown to be energetically important for caffeine recognition, based on the decrease in binding affinity for demethylated analogues ().Citation20 Notably, the N7 methyl group, which was used for conjugation of caffeine for immunization,Citation18 is located in the most solvent-exposed position (). Both Y98 residues (from each VHH domain) appear to “cap-off” caffeine in the complex and partially preclude caffeine solvent accessibility in the complex (Figure 21b); however, the Y98 side chain is noticeably the most conformationally diverse-binding site residue when comparing the four binding sites within the asymmetric unit (Supplemental Figure S2). It is interesting to note that due to steric constraints, the conformation/packing of these two tyrosine sidechains against caffeine may not be possible to the same extent when binding the original conjugated caffeine immunogen. Binding data for theophylline-7-acetic acid (), which physically resembles an N7 conjugated caffeine immunogen, displays an approximate 1.5 kcal/mol loss in affinity relative to caffeine, supporting the idea that one or both Y98 residues make favorable contributions toward caffeine binding. Alternatively, the binding penalty may also have contributions from unfavorable interactions from the presence of the carboxylic group.
Table 3. Thermodynamic data for anti-caffeine binding to xanthine analogues. Data for three arginine incorporations binding to caffeine are also shown.
The structures of the VHH/caffeine complex opened the opportunity to pursue a detailed isothermal titration calorimetry investigation into the binding thermodynamics of several caffeine analogs, as well as anti-caffeine VHH mutants to parse the mode of caffeine recognition (). To examine the possibility of a 1:1 caffeine/VHH complex, two xanthine analogues, 8-cyclopentyltheophylline and pentoxifylline, were examined because each was predicted to possess a sterically disruptive functional group directed toward one of the two VHH domains in the homodimer. As compared to caffeine analogs theophylline, paraxanthine, and theobromine, which each possess a single methyl group reduction that results in ΔΔG° penalties of 1.2 to 2.3 kcal/mol, the two “sterically disruptive” caffeine analogs possessed significant penalties toward hapten recognition (ΔΔG° = 3.6 and 5.0 kcal/mol, respectively). The more dramatic penalty is not surprising considering the expected steric clash, although the fact the compounds still bind suggests the anti-caffeine VHH complex possesses some degree of conformational plasticity. Notably, neither of the two sterically disruptive ligands was observed to bind to a single VHH, as both retained binding profiles consistent with a 2:1 binding stoichiometry.
To further explore the possibility of a 1:1 caffeine/VHH complex, site-directed mutagenesis was used to incorporate bulky/charged arginine residues in the dimer interface that should penalize (disrupt) VHH dimerization without direct influence on caffeine interactions. Three residues, which bury significant surface area in the VHH:VHH interface and, importantly, are not located in the immediate vicinity of the caffeine/VHH binding interface (), were identified for arginine substitution (F47, V100, and Y[100B]; Kabat numbering). Due to the 2-fold rotational symmetry of the VHH/VHH interface, each of the three individual arginine mutations introduce two arginine residues within the dimer interface. Each single arginine mutant VHH was characterized by ITC to determine the caffeine-binding thermodynamics. The general impact of these mutations on caffeine binding was dramatic. Both anti-caffeine VHH variants F47R and Y(100B)R displayed significantly weaker binding affinity (ΔΔG° >5 kcal/mol), while variant V100R possessed no observable heats of binding, likely indicating the complete absence of observable binding (). Based on the results of both the caffeine analogs and the dimer-disrupting arginine substitutions, disruption of dimerization resulted in disrupted ligand binding, consistent with a model that homodimerization is required for caffeine binding.
Evidence for VHH Self-Association in the absence of caffeine
To better explore the oligomeric nature of the anti-caffeine VHH, beyond the original size exclusion chromatography analysis,Citation20 AUC sedimentation velocity experiments were pursued. These experiments revealed that at 30 µM anti-caffeine VHH, the protein exists in a monomer-dimer equilibrium, but upon addition of 15 µM caffeine, the equilibrium shifts to favor the dimeric form (). Notably, the dimer interface F47R variant is monomeric in solution at 30 µM, suggesting that the Phe to Arg mutation disrupts the dimer interface as designed. Sedimentation equilibrium experiments were performed to verify the oligomerization state of the samples and to determine equilibrium assembly constants (). As suggested by the velocity data, all equilibrium data for wild-type VHH were consistent with a monomer-dimer equilibrium in solution that was modulated by the binding of caffeine. The VHH alone exhibited a monomer-dimer KD of 23.5 µM (15.4–35.7 uM), compared to a KD,obs of 848 nM (414 nM−1.6 uM) in the presence of a stoichiometric amount of caffeine (values in parenthesis represent 95% confidence limits). Thus, caffeine binding resulted in a nearly 28-fold higher observed dimerization affinity (ΔΔG°dim = −1.9 kcal/mol). Equilibrium data revealed that the F47R-VHH variant sedimented as a monomer even in the presence of 15 µM caffeine, indicating that this point mutation dramatically destabilizes the dimeric assembly. Given the velocity data indicating that F47R-VHH was completely monomeric in the absence of caffeine, this sample was not further tested by sedimentation equilibrium experiments. Notably, the ITC data for variant F47R (as well as Y[100B]R) could only be reasonably fit to a binding model when fixing the stoichiometry to a 2:1 and not 1:1 interaction, which agrees with AUC data that dimerization is necessary for caffeine recognition. These data support a linked equilibria model where VHH dimerization occurs first, followed by a second caffeine-binding event ().
Figure 3. Analytical ultracentrifugation analysis of anti-caffeine VHH dimer assembly. (a) Sedimentation coefficient distribution plots for wt. VHH alone, wt VHH with stoichiometric caffeine, and F47R VHH. Each protein was loaded at a concentration of 30 µM, and the caffeine concentration for the second sample was 15 µM. (b) Sedimentation equilibrium data for wt VHH. (c) Sedimentation equilibrium data for wt VHH with stoichiometric caffeine. (d) Sedimentation equilibrium data for F47R VHH with stoichiometric caffeine. All equilibrium experiments were conducted using protein samples at 3, 10, and 30 µM, with stoichiometric levels of caffeine (1.5, 5, and 15 µM) as required. The fits shown for each sample are the result of global analysis of all three speeds and concentrations; for clarity, only the 10 µM data at each speed are plotted.
Figure 4. Thermodynamic model and temperature profile for anti-caffeine VHH dimer dissociation. (a) Linked thermodynamic model consisting of dimerization, Kdimer, followed by caffeine binding to the VHH dimer species, Kbind. (b) Plot of ΔH°obs as a function of temperature for VHH:VHH dissociation. (c) Plot of ΔG°obs as a function of temperature for VHH:VHH dissociation based on obtained dimer dissociation thermodynamic parameters. Includes data from AUC experiments at 20° for comparison. Error bars represent 95% confidence intervals. Note: Thermodynamic terms for ΔH°obs and ΔG°obs presented in panels B and C represent the dimer dissociation reaction, which is the opposite direction of Kdimer presented in panel A.
To further explore the dimerization energetics of anti-caffeine VHH in the absence of caffeine, isothermal titration calorimetry dilution experiments were performed. While the original dilution experiments did not produce significant dissociation heats at 25°C,Citation20 measurable heats of dilution were observed when experiments were performed over a wider temperature range, including 5°C, 35°C, and 45°C (), which allowed determination of the homodimer dissociation thermodynamics (ΔG°dim, ΔH°dim, ΔS°dim, and ΔCp,dim). A large change in heat capacity for dimer dissociation was observed, ΔCp,dim = 594 kcal/mol, which clearly aided our ability to detect heats of dissociation at temperatures other than 25°C where the change in enthalpy was negligible. The ΔCp,dim, ΔH°dim, and ΔS°dim were used to calculate the free energy of dimer dissociation as a function of temperature (). Notably, at 20°C, the calorimetric dimerization dissociation constant, Kdim, was 13.1 μM, which agrees well with the analytical ultracentrifugation results (Kdim,AUC = 23.5 μM). At 25°C (temperature of the binding studies) the dimerization (VHH:VHH association) event is an entropically driven process (-TΔS°assoc = −5.2 kcal/mol), with a relatively negligible change in enthalpy (ΔH°assoc = −1.4 kcal/mol).
The entropically driven dimerization, coupled with the negative heat capacity change for dimerization are consistent with significant apolar surface area burial and solvent release upon dimerization. Notably, this is despite the noted water molecules retained in the VHH/VHH interface. The negligible ΔH°assoc value at 20°C validates the lack of measurable observed dissociation heats during the original dilutions experiments,Citation20 and informs that the favorable observed enthalpy of binding at 25°C (~-15 kcal/mol at 25°C) must originate from favorable interactions associated with caffeine recognition, as the dimerization, which occurs during the titrations, is almost entirely entropically driven at 25°C (). Finally, it is interesting to note that the temperature of maximum dimer stability is located near physiological temperature (), which may indicate physiological relevance to the dimerization-based VHH target recognition.
Structure analysis of the anti-caffeine VHH dimer/caffeine complex
To better understand this unique homodimer interaction, we examined our 1.13 Å resolution x-ray crystal structure of the anti-caffeine VHH homodimer/caffeine complex. In general, the structural organization of the anti-caffeine VHH homodimer is analogous to the heterodimer found in conventional VH/VL antibodies. In both cases, the dimerization interface of both VHH and VH domains consists of primarily hydrophobic contributions from framework two residues, with the hapten-binding pocket located between the two neighboring domains with significant contributions from CDR3. The CDR3 residues Y98, V100, and Y100B bury a total of 240 Å2 of predominantly hydrophobic surface area within the interface (). One prominent difference between the two dimer types is the angle formed by the primary axes of each domain ().
Figure 5. Structural differences between the anti-caffeine VHH homodimer and conventional VH/VL heterodimers. (a) Surface area burial heat map for residues within the anti-caffeine VHH:VHH (left) and a representative VH:VL dimer (right; PDB ID: 1Q72)Citation26. (b) Structural overlay of a VHH domain from the anti-caffeine structure with the VH domain from a murine anti-cocaine Fab (PDB entry 1Q72). The VH and VHH domain alignment is in background (gray) and the VL domain from anti-cocaine (blue) and second VHH domain from anti-caffeine VHH (red) are in foreground. Average angle between principal axes is calculated from four anti-caffeine dimers and a sampling of VH:VL structures from the PDB 1Q72, 2JB6, 2UUD, 3CFB, 3FO9).
An analysis of a random selection of anti-hapten antigen-binding fragments and single-chain variable fragments from the PDB (PDB IDS: 1Q72, 2JB6, 2UUD, 3CFB, and 3FO9) reveals an average angle of 73.8° between the primary axes (ranging from 61.7° to 77.9°), whereas the anti-caffeine dimer has a significantly smaller angle of 44.5°. Interestingly, this angle is also significantly smaller than an engineered “humanized” camelid VHH homodimer,Citation28 which has an angle of 82.4°. The average total surface area buried by VH domains in the VH/VL antibody sample set was 735 ± 62 Å2, compared to 887 ± 31 Å2 for the anti-caffeine VHH domains (). This restructuring of the antibody dimer interface allows loop residues E44, R45, and E46, which connect beta strands of framework 2, to contribute 170 ÅCitation2 of surface area burial within the dimer (). Despite the polar nature of the terminal side chain atoms of these amino acids, slightly over 50% of the buried surface area of these residues is apolar. The ~20% increase in total surface area burial within the VHH/VHH (as compared to VH/VL interfaces) includes a slight increase in the relative amount of apolar surface area buried versus the heterodimeric complexes (69 vs. 65%, respectively). Notably, the VHH/VHH interface alignment possesses a surface burial pattern that is more parallel to the principal axes, which relates to the primary axis alignment discussed above (). In addition, as opposed to VH/VL’s more centralized clustering of residues contributing greater than 20 Å2 of surface, the VHH/VHH homodimer possesses a less centralized, peripheral “o-ring” pattern of such surface buried residues (; orange/red coloring). The o-ring pattern appears in line with the centralized 14 water molecules within the VHH/VHH interface.
Table 4. Surface area burial of VH:VL and VHH:VHH antibody dimers in Å2.
Displacement of canonical VHH CDR3
When compared to VHH domains with known structure, the most significant difference observed in the anti-caffeine VHH structure is the conformation of CDR3. The anti-caffeine VHH CDR3 is 10 residues in length, which is comparable to the average CDR3 length of human VH (11.6 residues),Citation4 but significantly shorter than the average VHH CDR3 length (17.5 residues).Citation3 Such extended VHH CDR3 loops are typically observed to partially fold over what would be the VH/VL interface in conventional antibodies, consisting of framework regions 2 and 3 (FW2 and FW3). This intramolecular interaction is hypothesized to help improve solubility by compensating for the lack of interaction with a light chain. Within the anti-caffeine dimer, the CDR3 loop is displaced from the former VH/VL interface in a position analogous to that found in conventional VH domains (). The most obvious structural consequence of the reduced loop length (and more VH-like CDR3 conformation) will be reduced intramolecular interactions between CDR3 and the VHH framework. Using an anti-RNase A VHH antibody as a structural reference of a conventional VHH,Citation5 several VHH framework residues would be expected to become more exposed to solvent with a shorter CDR3, including S33, F37, R45, F47, T50, Y58, and M60. Overall, 195 Å2 of accessible surface area is predicted to be exposed by CDR3 displacement, of which 75% (147 Å2 is apolar ().
Figure 6. Stereoview image displaying the difference in CDR3 conformation between anti-caffeine VHH domain and a conventional anti-RNase a VHH. VHH domains have been superimposed and anti-RNase a framework omitted for clarity. CDR3 loop color coded: anti-caffeine VHH (red) and anti-RNaseA VHH (cyan). Framework residues that experience greater solvent exposure due to displaced CDR3 are highlighted in green sticks. Caffeine ligand is displayed in stick form at top of image (carbon-green).
The anti-caffeine VHH framework surface that is anticipated to be solvent exposed by the absence of CDR3 intramolecular interactions is observed forming intermolecular interactions across the VHH/VHH homodimer interface. A group of exposed framework residues, including F37, F47, Y58, and M60, form a band across the center of the dimer interface, burying the largely hydrophobic surface area along with the same residues on the opposite VHH (). These residues account for a total of 200 Å2 of buried surface area per VHH, of which 80% of the buried surface area is apolar. Two of the 14 sequestered water molecules are coordinated by the F47 main chain polar groups. The exposure of Y58 allows for two hydrogen bond interactions in the anti-caffeine dimer, located between the tyrosine residue’s hydroxyl group and the indole nitrogen from W103 located on the adjacent VHH domain. In addition, the CDR3 possesses two intramolecular sidechain hydrogen bonds between S99 and D101, as well as R96 and Y102.
Since our anti-caffeine VHH shares framework residues similar to a conventional, monomeric anti-RNase A VHH,Citation20,Citation29 the influence of reduced CDR3-mediated intramolecular interactions on stability can be evaluated. Thermal melt analysis of the anti-caffeine VHH revealed a reduction in Tm values by approximately 20°C at pH 7.0 (Supplemental Figure S3), suggesting anti-caffeine VHH’s shorter CDR3 loop and subsequent exposure of framework residues adversely affect VHH domain stability. Chemical stability experiments using guanidinium hydrochloride reveal ΔGu values of 7.6 ± 0.2 and 9.17 ± 0.06 kcal/mol for anti-caffeine and anti-RNase A VHH, respectively, indicating the anti-caffeine VHH was less stable by approximately 1.5 kcal/mol at 25°C (Supplemental Figure S3). Notably, the m-value for the anti-caffeine VHH (2.62 ± 0.06 kcal mol−1M−1) was smaller than that of anti-RNase A (3.37 ± 0.02 kcal mol−1M−1). Since the magnitude of m-values shows correlations with increased surface area exposure upon unfolding,Citation30 the difference is consistent with a decrease in anticipated surface area for anti-caffeine VHH unfolding. Finally, it is worth noting that, while dimerization can increase the stability of anti-caffeine VHH, less than 5% of the total VHH would be expected to be in the dimeric state during the unfolding experiments (based on the dimerization constant), so a significant increase in stability due to dimerization would not be expected under the experimental conditions.
The short CDR3 loop of the anti-caffeine VHH is a striking structural feature compared to conventional VHH domains. The apparent structural consequence of the short loop is the loss of intramolecular interactions that are typically observed between CDR3 and VHH framework residues. Consequently, the solvent exposure of this framework surface may serve to facilitate anti-caffeine VHH homodimerization. To test the hypothesis that a short CDR3 VHH loop may help drive dimerization, an anti-picloram VHH (3-1D2) domain containing a short, five residue CDR3 was identified.Citation24 The gene for the anti-picloram VHH domain was synthesized, over-expressed, and characterized by size exclusion chromatography, where it was found to form a stable homodimer species in the absence of picloram (). Based on the VHH concentration and the observation that only the dimer was observed, the equilibrium dimer dissociation constant, KD, for anti-picloram VHH is estimated as being at least 100-fold less than anti-caffeine VHH. While it cannot be ruled out that other features unique to anti-picloram may play a role in dimerization, anti-picloram VHH’s CDR3, which is even shorter than anti-caffeine VHH’s CDR3, would be even less likely to make intramolecular interactions with framework residues, resulting in solvent-exposed framework residues that may be prone to engage in intermolecular interactions upon dimer formation.
Figure 7. Normalized size exclusion profile of anti-picloram versus monomeric anti-RNase a VHH domains. Anti-RNase a (dashed pink) and anti-picloram VHH (solid blue).
Discussion
The anti-caffeine VHH antibody represents an atypical VHH/target interaction where VHH homodimerization facilitates ligand recognition. Here, structural and biophysical analyses revealed the details of the interactions between VHH domains, as well as between caffeine and the VHH homodimer. Failed attempts to isolate an anti-caffeine VHH with 1:1 binding stoichiometry, along with biophysical investigations revealing the dimerization thermodynamics in the absence of caffeine, suggests that ligand recognition requires the formation of a VHH dimer species.
There are likely to be several structural aspects of the anti-caffeine VHH that are important for dimer formation. The relatively short CDR3 exposes the largely hydrophobic region of the former VL interface (FW2 and FW3). This surface area becomes solvent exposed and capable of making new inter-domain contacts. The observed dimerization of the anti-picloram VHH, which contains an even shorter CDR3 loop, supports the role of CDR3 in homodimerization. However, a short CDR3 alone may not be enough to induce dimerization, as short VHH CDR3 loops have been observed for several other VHH antibodies where structural data was available. An anti-HIV-1 capsid VHH (unpublished work, PDB entries 2xV6 and 2xT1) and an anti-GFP VHHCitation25 have CDR3 lengths of six and seven residues, respectively (). These systems possess apparent 1:1 stoichiometry, but both complexes appear to include a significant portion of the exposed (former VL interface) framework surface, particularly FW2, in antigen recognition, suggesting that short CDR3 regions, and the resulting framework residues that are exposed to solvent, may have a wider range of functional consequences. It is also important to note that these VHH examples did not appear to be fully evaluated with a technique such as AUC, so it is possible that dimerization may occur, especially at higher concentrations. Finally, anti-caffeine VHH’s CDR3 also possesses several intramolecular polar contacts, such as S99-D101 and R96-Y102 and intermolecular contacts, such as Y100B-Y100B, which suggest sequence-specific CDR3 interactions may play an important role in stabilizing the dimeric state.
Specific framework residues may play important roles in contributing to VHH dimerization. While there are many similarities in the FW regions of anti-caffeine, picloram, GFP and HIV1 (), there are no obvious mechanisms. In many previously discovered VHH antibodies, position 47 is occupied by a glycine residue, whereas the anti-caffeine VHH domain possesses a phenylalanine, as does the anti-picloram VHH. As described by Harmsen et al.,Citation9 phenylalanine at this position is a defining characteristic of the VHH1 family of antibodies and is present to a lesser degree within other VHH families. As shown for VHH domain cAbAn33, which possessed a tryptophan residue at position 47, dimerization could be induced via the introduction of a small number of VH-mimicking mutations in framework 2.Citation28 Interestingly, this modified VHH also has a CDR3 length of 10 residues. The anti-GFP and HIV1 VHH domains possess short CDR3s with position 47 as tryptophan or leucine, respectively (). While their potential dimerization cannot be ruled out, these hydrophobic residues at position 47 for both VHH domains participate directly in antigen recognition. Consequently, position 47 exposure, due to short CDR3s, is apparently an energetically active surface that can be exploited for antigen recognition.
As the anti-caffeine VHH antibody was first generated through inoculation of a llama,Citation18 the homodimer-based hapten recognition by a camelid antibody opens the possibility that such dimer-based target recognition may occur in vivo among members of the camelid family. Along these lines, anti-caffeine VHH’s dimerization constant in the absence of caffeine (KDim of ~20 µM) is in the range of conventional VH:VL dimer dissociation constants, albeit on the weaker end.Citation31 Furthermore, the observations that the VHH homodimer exhibits maximum stability near physiological temperature may imply an evolutionary aspect to this mechanism. If this phenomenon does occur in vivo, the hinge length between the VHH and constant domains, which varies from 12 to 35 residues in heavy-chain-only antibodies,Citation32 may play a significant role in establishing correct dimer orientation and/or whether dimerization occurs. On the other hand, as with development and isolation of most VHH domains, in vitro screening is often performed as part of the post-immunization antibody selection. In the development of the original anti-caffeine VHH, phage display selection was performed to isolate the best behaving anti-caffeine VHH clones.Citation18 As such, it is possible that random mutation(s) may occur, which may also have a role in selecting functional aspects of the identified clones. Regardless of the origin, when identifying and characterizing new VHH antibodies, the possibility of functional homodimeric VHH species should not be overlooked.
On a broader scale, the anti-caffeine VHH may serve as a useful model for the development of new protein affinity reagents. One of the most apparent differences between the anti-caffeine homodimer and traditional VH:VL dimers is the burial of 14 water molecules within the VHH/VHH interface. Notably, 12 of the 14 water molecules reside near the caffeine-binding pocket. Consequently, with the low degree of packing efficiency between residues between the two VHH domains, it should be possible to exploit this solvent-occupied space to enable the binding of larger hapten targets, or conversely, to pursue interface remodeling with the goal of minimizing buried water molecules, thereby strengthening homodimerization. Such manipulations may affect how a system like anti-caffeine VHH may be used in CID applications, whether by expanding the type of potential target (trigger) molecule or by controlling (i.e., tuning) VHH dimerization in the absence of ligand. For the latter example, our interface mutants, such as F47R and Y(100B)R, provide examples of how the dimer interface can be modified to influence the strength of dimerization, while maintaining caffeine binding, to address “leaky” CID applications. Ultimately, future studies of VHH systems will be necessary to better understand the functional significance of homodimeric VHH complexes and to further explore how they may be engineered for life science applications.
Materials and methods
Mutagenesis, expression, and purification of the VHH variants
The p21-α-caff-VHH vector,Citation10 a pET-21a expression vector containing the anti-caffeine VHH gene with N-terminal His-flag-Tev- tags, served as the template to generate the three single site anti-caffeine VHH variants (F47R, V100R, and Y[100B]R) using QuikChange site-directed mutagenesis. The anti-caffeine variants were expressed and purified as described previously.Citation10 Briefly, each anti-caffeine VHH variant was expressed in BL21(DE3) cells using 1 L of LB/ampicillin media, which included a mid-log phase induction with 1 mM isopropyl-ß-D-thiogalactopyranoside (IPTG) followed by overnight incubation at 37°C. The harvested cells were sonicated, and the clarified lysate was subjected to purification on a 5 mL Histrap HP column, using a linear gradient elution. The single peak corresponding to the anti-caffeine VHH variant was further purified using a HiLoad Superdex 75 prep grade FPLC column. Finally, the His-flag-TEV tag was removed by proteolysis using TEV protease and the tag-free VHH was isolated by running the reaction over a HisTrap HP column.
A synthetic gene for the anti-picloram VHH, which was codon-optimized for E. coli, based on the published sequence,Citation24 was introduced into a pET-21a expression vector containing an N-terminal His-flag-Tev- tag. Expression and purification were performed identically to that described for anti-caffeine VHH variants with the following exceptions. The 1 L cultures were grown past mid-log phase (OD = 1.2) and subsequently subjected to a cold-shock incubation on ice for approximately 20 minutes. Next, 20% ethanol (v/v) was added along with 1 mM IPTG. The culture incubated with shaking at 20°C for 18 hours.
Crystallization and structure determination of the VHH/caffeine complex
Purified anti-caffeine VHH was concentrated to 20 mg/mL using an Amicon Ultra centrifugal concentrator with 5-kDa MWCO. A 5 mM solution of caffeine in 20 mM sodium phosphate, 150 mM NaCl buffer (pH 7.4) was then diluted into the concentrated anti-caffeine VHH solution to a final concentration of 1 mM to allow formation of the 2:1 complex. The VHH/caffeine complex was incubated for 15 minutes before running the sample on a Superdex-75 (GE Bioscience) gel filtration column. The 2:1 complex was isolated using a running buffer of 10 mM Tris buffer (pH 8.0) and 300 mM NaCl. Fractions containing the 2:1 complex were pooled and adjusted to a final concentration of 20 mg/ml.
The VHH/caffeine complex was crystallized using the hanging-drop vapor diffusion method. The drops contained 2 µL of the 20 mg/mL complex and 2 µL of the well mother liquor that contained 0.1 M BIS-TRIS (pH 5.5) and 25% w/v polyethylene glycol 3,350. Samples were incubated at 20°C and crystals appeared after 20 d. Individual crystals were isolated, briefly soaked in a cryo-solution of mother liquor supplemented with 20% glycerol and flash-frozen in liquid nitrogen.
Data to 1.13 Å were collected at Southeast Regional Collaborative Access Team (SER-CAT) 22-BM beamline at the Advanced Photon Source, Argonne National Laboratory. High- and low-resolution data sets were collected on a single crystal using 2° (5 sec exposure) and 3° (1 sec exposure) oscillation width per frame, respectively. The data sets were merged, integrated and scaled using HKL2000Citation33 and are 93.78% complete to 1.13 Å. Resulting data were reduced to 273,342 unique reflections in a P1 space group with a = 50.136, b = 50.136, c = 68.999, α = 111.62, β = 95.20, γ = 90.25.
Molecular replacement was performed using the program MOLREP and the anti-RNaseA VHH domain from PDB accession code 1BZQ (chain K) serving as the search model.Citation5 Initial rounds of molecular replacement generated a dimeric model which was used in subsequent molecular replacement searches which ultimately identified a unit cell containing four dimer complexes. Initial refinements to the data were done with the Refmac/CCP4 program suiteCitation34 and then using Phenix.Citation35 Model adjustments between refinements were performed using Coot.Citation36 The structure was deposited in the Protein Data Base (PDB ID: 8FTG). Available surface area calculations were performed using the program NACCESS.Citation37 Buried surface area calculations were carried out by comparing the surface area of each residue to an Ala-X-Ala tripeptide.
Determining Principal Axes Angles
PDB structures for anti-caffeine and VH/VL antibodies were loaded into VMD (version 1.9.1, Champaign, IL) with the principal axes calculation script installed (available at http://www.ks.uiuc.edu/Research/vmd/script_library/scripts/orient/). Principal axes of inertia were calculated, with the axis traversing the N to C terminus direction being of interest. The peptide’s center of mass and a point on the axis toward the N terminus were added to the PDB as new atoms. This process was performed for both domains of each antibody dimer. The resulting PDB file was opened with Pymol,Citation38 bonds were created between the new data points, and the dihedral angles were calculated.
Isothermal titration calorimetry
For dissociation experiments, anti-caffeine VHH was expressed and concentrated to 400 μM and dialyzed twice against 1 L phosphate-buffered saline (PBS; 20 mM sodium phosphate, 150 mM NaCl, pH 7.4). Samples were then centrifuged for 1 minute at 16,100 RCF. Concentration was determined using the extinction coefficient of ε280 nm = 30,035 M−1 cm−1, as determined by the method of Pace et al. Citation39 Isothermal titration calorimetry experiments were performed using a protein concentration between 300 and 375 μM in the syringe injected into PBS using a Microcal VP-ITC titration calorimeter (MicroCal, LLC, Northampton, MA). Trials were run three times each at 5°C, 35°C, and 45°C. Experiments followed a protocol of 27, 10 μL injections, with 330 seconds between injections and a stirring speed of 307 rpm. Dilution heats were determined by injecting PBS into PBS and were subtracted from the experimental values. Data analysis was performed using Origin and the ITC add-on. ΔH values were plotted against temperature, with a linear fit used to determine ΔCp. Using the data point closest to the linear fit as reference data, the equation ΔG°T = ΔH°R - TΔS°R + ΔCp[T – TR – T ln(T/TR)] (Equation 1) was used to simulate the ΔG° dependence on temperature. For experiments testing the binding of xanthine analogues, anti-caffeine VHH was dialyzed against 4 L PBS pH 7.4 overnight. 50 μM xanthine analogue in the syringe was injected into 10 μM VHH in the cell at 25°C. The protocol consisted of 27 injections of 10 μL each. For VHH variants F47R and V104R, an extinction coefficient of ε280 nm = 30,035 M−1 cm−1 was used, and ε280 nm = 28,545 M−1 cm−1 was used for Y106R. For all three arginine incorporations, a method of ITC for low-affinity binders introduced by Turnbull et al. Citation40 was used, in which 1 mM caffeine was diluted into 100 μM VHH. While this method requires a known binding stoichiometry, the relatively modest difference between cell and syringe concentrations allows for an observably better fit with n = 0.5 as compared to n = 1.0.
Analytical ultracentrifugation
Sedimentation velocity experiments were performed to determine the size distribution of the anti-caffeine VHH and interface variant F47R in solution (30 μM anti-caffeine VHH (WT), 30 μM F47R, and 30 μM WT +15 μM caffeine) at 20°C using a Beckman XL-I analytical ultracentrifuge. 400 μL of each protein sample and 420 μL of buffer were loaded into two-sector cells and spun at 48,000 rpm for ~16 hours. Data were collected at 284 nm using the absorbance optical system. Data were deconvoluted to determine sedimentation coefficient distributions using the c(s) analysis routine in the program SEDFIT.Citation41
Sedimentation equilibrium experiments provided information on the molecular weight and stoichiometries of the proteins and protein-caffeine complexes. Three concentrations of each sample were spun for 24 h at three speeds (27,000, 32000, and 44,000 rpm) and scanned at 294 nm, 260 nm, and 236 nm. Experiments were performed on WT (30 μM, 10 μM, and 3 μM), WT + caffeine (30 μM WT +15 μM caffeine, 10 μM +5 μM, and 3 μM +1.5 μM), and F47R + caffeine (30 μM F47R +15 μM caffeine, 10 μM +5 μM, and 3 μM +1.5 μM). Data were truncated using WinReedit and globally analyzed using WinNonlin (www.rasmb.bbri.org). Data were fitted first to a single exponential function to determine the weight-averaged molecular weight. When appropriate, data were then fitted to monomer-dimer equilibria to determine equilibrium assembly constants, as previously described.Citation42–44
Abbreviations
| AUC | = | analytical ultracentrifugation |
| CDR | = | complementarity-determining region |
| CH1 | = | first constant heavy chain domain |
| CID | = | chemically induced dimerization |
| FW | = | framework |
| IPTG | = | isopropyl-ß-D-thiogalactopyranoside |
| ITC | = | isothermal titration calorimetry |
| mL | = | milliliter |
| PBS | = | phosphate-buffered saline |
| PDB | = | Protein Data Bank |
| Tm | = | melting temperature |
| VH | = | variable heavy |
| VHH | = | variable domain from heavy-chain-only antibodies |
| VL | = | variable light |
| WT | = | anti-caffeine VHH |
Supplemental Material
Download MS Word (1.3 MB)Acknowledgments
X-ray diffraction data were collected at Southeast Regional Collaborative Access Team (SER-CAT) 22-BM beamline at the Advanced Photon Source, Argonne National Laboratory. Supporting institutions may be found at www.ser-cat.org/members.html.
Use of the Advanced Photon Source was supported by the U. S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. W-31-109-Eng-38. This work was supported by the National Science Foundation under Grant MCB-0953323 and National Institute of Health under Grant 1R15GM124607-01
Disclosure statement
No potential conflict of interest was reported by the authors.
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2023.2215363
Additional information
Funding
References
- Hamers-Casterman C, Atarhouch T, Muyldermans S, Robinson G, Hamers C, Songa EB, Bendahman N, Hamers R. Naturally occurring antibodies devoid of light chains. Nature. 1993;363(6428):446–14. PMID: 8502296. doi:10.1038/363446a0.
- Spinelli S, Tegoni M, Frenken L, van Vliet C, Cambillau C. Lateral recognition of a dye hapten by a llama VHH domain. J Mol Biol. 2001;311(1):123–29. PMID: 11469862. doi:10.1006/jmbi.2001.4856.
- Muyldermans S, Atarhouch T, Saldanha J, Barbosa JA, Hamers R. Sequence and structure of VH domain from naturally occurring camel heavy chain immunoglobulins lacking light chains. Protein Eng. 1994;7(9):1129–35. PMID: 7831284. doi:10.1093/protein/7.9.1129.
- Wu TT, Johnson G, Kabat EA. Length distribution of CDRH3 in antibodies. Proteins: Struct Funct Genet. 1993;16(1):1–7. PMID: 8497480. doi:10.1002/prot.340160102.
- Decanniere K, Desmyter A, Lauwereys M, Ghahroudi MA, Muyldermans S, Wyns L. A single-domain antibody fragment in complex with RNase A: non-canonical loop structures and nanomolar affinity using two CDR loops. Structure. 1999;7(4):361–70. PMID: 10196124. doi:10.1016/s0969-2126(99)80049-5.
- Desmyter A, Spinelli S, Payan F, Lauwereys M, Wyns L, Muyldermans S, Cambillau C. Threecamelid VHH domains in complex with porcine pancreatic α-amylase. J Biol Chem. 2002;277(26):23645–50. PMID: 11960990. doi:10.1074/jbc.M202327200.
- Desmyter A, Transue TR, Ghahroudi MA, Thi MH, Poortmans F, Hamers R, Muyldermans S, Wyns L. Crystal structure of a camel single-domain VH antibody fragment in complex with lysozyme. Nat Struct Biol. 1996;3(9):803–11. PMID: 8784355. doi:10.1038/nsb0996-803.
- Hoey RJ, Eom H, Horn JR. Structure and development of single domain antibodies as modules for therapeutics and diagnostics. Exp Biol Med (Maywood). 2019;244(17):1568–76. PMID: 31594404. doi:10.1177/1535370219881129.
- Mitchell LS, Colwell LJ. Comparative analysis of nanobody sequence and structure data. Proteins Struct Funct Bioinf. 2018;86(7):697–706. PMID: 29569425. doi:10.1002/prot.25497.
- Rasmussen SG, Choi HJ, Fung JJ, Pardon E, Casarosa P, Chae PS, Devree BT, Rosenbaum DM, Thian FS, Kobilka TS, et al. Structure of a nanobody-stabilized active state of the β2 adrenoceptor. Nature. 2011;469(7329):175–80. PMID: 21228869. doi:10.1038/nature09648.
- Anderson GP, Goldman ER. TNT detection using llama antibodies and a two-step competitive fluid array immunoassay. J Immunol Methods. 2008;339(1):47–54. PMID: 18755196. doi:10.1016/j.jim.2008.08.001.
- Alvarez-Rueda N, Behar G, Ferre V, Pugniere M, Roquet F, Gastinel L, Jacquot C, Aubry J, Baty D, Barbet J, et al. Generation of llama single-domain antibodies against methotrexate, a prototypical hapten. Mol Immunol. 2007;44(7):1680–90. PMID: 17011035. doi:10.1016/j.molimm.2006.08.007.
- Spinelli S, Frenken LG, Hermans P, Verrips T, Brown K, Tegoni M, Cambillau C. Camelid heavy-chain variable domains provide efficient combining sites to haptens. Biochemistry. 2000;39(6):1217–22. PMID: 10684599. doi:10.1021/bi991830w.
- De Genst E, Silence K, Decanniere K, Conrath K, Loris R, Kinne J, Muyldermans S, Wyns L. Molecular basis for the preferential cleft recognition by dromedary heavy-chain antibodies. Proc Natl Acad Sci U S A. 2006;103(12):4586–91. PMID: 16537393. doi:10.1073/pnas.0505379103.
- Fanning SW, Horn JR. An anti-hapten camelid antibody reveals a cryptic binding site with significant energetic contributions from a nonhypervariable loop. Protein Sci. 2011;20(7):1196–207. PMID: 21557375. doi:10.1002/pro.648.
- Tabares-da Rosa S, Wogulis LA, Wogulis MD, Gonzalez-Sapienza G, Wilson DK. Structure and specificity of several triclocarban-binding single domain camelid antibody fragments. J Mol Recognit. 2019;32(1):e2755. PMID: 30033524. doi:10.1002/jmr.2755.
- Ding L, Wang Z, Zhong P, Jiang H, Zhao Z, Zhang Y, Ren Z, Ding Y. Structural insights into the mechanism of single domain VHH antibody binding to cortisol. FEBS Lett. 2019;593:1248–56. PMID: 31049949. doi:10.1002/1873-3468.13398
- Ladenson RC, Crimmins DL, Landt Y, Ladenson JH. Isolation and characterization of a thermally stable recombinant anti-caffeine heavy-chain antibody fragment. Anal Chem. 2006;78(13):4501–08. PMID: 16808459. doi:10.1021/ac058044j.
- Harmsen MM, Ruuls RC, Nijman IJ, Niewold TA, Frenken LG, de Geus B. Llama heavy-chain V regions consist of at least four distinct subfamilies revealing novel sequence features. Mol Immunol. 2000;37(10):579–90. PMID: 11163394. doi:10.1016/s0161-5890(00)00081-x.
- Sonneson GJ, Horn JR. Hapten-induced dimerization of a single-domain VHH camelid antibody. Biochemistry. 2009;48(29):6693–95. PMID: 19572722. doi:10.1021/bi900862r.
- Chang HJ, Mayonove P, Zavala A, De Visch A, Minard P, Cohen-Gonsaud M, Bonnet J. A modular receptor platform to expand the sensing repertoire of bacteria. ACS Synth Biol. 2018;7(1):166–75. PMID: 28946740. doi:10.1021/acssynbio.7b00266.
- Bojar D, Scheller L, Hamri GC, Xie M, Fussenegger M. Caffeine-inducible gene switches controlling experimental diabetes. Nat Commun. 2018;9(1):2318. PMID: 29921872. doi:10.1038/s41467-018-04744-1.
- Lesne J, Chang HJ, De Visch A, Paloni M, Barthe P, Guichou JF, Mayonove P, Barducci A, Labesse G, Bonnet J, et al. Structural basis for chemically-induced homodimerization of a single domain antibody. null. 2019;9(1):1840. PMID: 30755682. doi:10.1038/s41598-019-38752-y.
- Yau KY, Groves MA, Li S, Sheedy C, Lee H, Tanha J, MacKenzie CR, Jermutus L, Hall JC. Selection of hapten-specific single-domain antibodies from a non-immunized llama ribosome display library. J Immunol Methods. 2003;281(1–2):161–75. PMID: 14580890. doi:10.1016/j.jim.2003.07.011.
- Kirchhofer A, Helma J, Schmidthals K, Frauer C, Cui S, Karcher A, Pellis M, Muyldermans S, Casas-Delucchi CS, Cardoso MC, et al. Modulation of protein properties in living cells using nanobodies. Nature Structural & Molecular Biology. 2010;17(1):133–38. PMID: 20010839. doi:10.1038/nsmb.1727.
- Pozharski E, Moulin A, Hewagama A, Shanafelt AB, Petsko GA, Ringe D. Diversity in hapten recognition: structural study of an anti-cocaine antibody M82G2. J Mol Biol. 2005;349(3):570–82. PMID: 15885702. doi:10.1016/j.jmb.2005.03.080.
- Williams CJ, Headd JJ, Moriarty NW, Prisant MG, Videau LL, Deis LN, Verma V, Keedy DA, Hintze BJ, Chen VB, et al. MolProbity: more and better reference data for improved all-atom structure validation. Protein Sci. 2018;27(1):293–315. PMID: 29067766. doi:10.1002/pro.3330.
- Conrath K, Vincke C, Stijlemans B, Schymkowitz J, Decanniere K, Wyns L, Muyldermans S, Loris R. Antigen binding and solubility effects upon the veneering of a camel VHH in framework-2 to mimic a VH. J Mol Biol. 2005;350(1):112–25. PMID: 15913651. doi:10.1016/j.jmb.2005.04.050.
- Koide A, Tereshko V, Uysal S, Margalef K, Kossiakoff AA, Koide S. Exploring the capacity of minimalist protein interfaces: interface energetics and affinity maturation to picomolar KD of a single-domain antibody with a flat paratope. J Mol Biol. 2007;373(4):941–53. PMID: 17888451. doi:10.1016/j.jmb.2007.08.027.
- Myers JK, Pace CN, Scholtz JM. Denaturant m values and heat capacity changes: relation to changes in accessible surface areas of protein unfolding. Protein Sci. 1995;4(10):2138–48. PMID: 8535251. doi:10.1002/pro.5560041020.
- Rothlisberger D, Honegger A, Pluckthun A. Domain interactions in the Fab fragment: a comparative evaluation of the single-chain Fv and Fab format engineered with variable domains of different stability. J Mol Biol. 2005;347(4):773–89. PMID: 15769469. doi:10.1016/j.jmb.2005.01.053.
- van der Linden R, de Geus B, Stok W, Bos W, van Wassenaar D, Verrips T, Frenken L. Induction of immune responses and molecular cloning of the heavy chain antibody repertoire of Lama glama. J Immunol Methods. 2000;240(1–2):185–95. PMID: 10854612. doi:10.1016/s0022-1759(00)00188-5.
- Otwinowski Z, Minor W. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 1997;276:307–26. PMID: 27754618. doi:10.1016/S0076-6879(97)76066-X
- Collaborative Computational Project N. The CCP4 suite: programs for protein crystallography. Acta Cryst. 1994;50(5):760–63. doi:10.1107/S0907444994003112.
- Adams PD, Afonine PV, Bunkoczi G, Chen VB, Davis IW, Echols N, Headd JJ, Hung LW, Kapral GJ, Grosse-Kunstleve RW, et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr. 2010;66(2):213–21. PMID: 20124702. doi:10.1107/S0907444909052925.
- Emsley P, Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr. 2004;60(12):2126–32. PMID: 15572765. doi:10.1107/S0907444904019158.
- Hubbard SJ, Thornton JM. NACCESS, Computer Program. Ver Vol. V2 1.1: software, University College London, “NACCESS”, Computer Program, Department of Biochemistry and Molecular Biology. 1993.
- DeLano WL The PyMOL molecular graphics system. [software] San Carlos, CA, USA: DeLano Scientific; 2002. http://www.pymol.org.
- Pace CN, Vajdos F, Fee L, Grimsley G, Gray T. How to measure and predict the molar absorption coefficient of a protein. Protein Sci. 1995;4(11):2411–23. doi:10.1002/pro.5560041120.
- Turnbull WB, Daranas AH. On the value of c: can low affinity systems be studied by isothermal titration calorimetry? J Am Chem Soc. 2003;125(48):14859–66. PMID: 14640663. doi:10.1021/ja036166s.
- Schuck P. Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and lamm equation modeling. Biophys J. 2000;78(3):1606–19. PMID: 10692345. doi:10.1016/S0006-3495(00)76713-0.
- Herr AB, Conrady DG. Thermodynamic analysis of metal ion-induced protein assembly. Methods Enzymol. 2011;488:101–21. PMID: 21195226. doi:10.1016/B978-0-12-381268-1.00005-7
- Herr AB, Ornitz DM, Sasisekharan R, Venkataraman G, Waksman G. Heparin-induced self-association of fibroblast growth factor-2. Evidence for two oligomerization processes. J Biol Chem. 1997;272(26):16382–89. PMID: 9195945. doi:10.1074/jbc.272.26.16382.
- Herr AB, White CL, Milburn C, Wu C, Bjorkman PJ. Bivalent Binding of IgA1 to FcαRI suggests a mechanism for cytokine activation of IgA phagocytosis. J Mol Biol. 2003;327(3):645–57. PMID: 12634059. doi:10.1016/s0022-2836(03)00149-9.