Discovery of lirafugratinib (RLY-4008), a highly selective irreversible small-molecule inhibitor of FGFR2

Fibroblast growth factor receptor (FGFR) kinase inhibitors have been shown to be effective in the treatment of intrahepatic cholangiocarcinoma and other advanced solid tumors harboring FGFR2 alterations, but the toxicity of these drugs frequently leads to dose reduction or interruption of treatment such that maximum efficacy cannot be achieved. The most common adverse effects are hyperphosphatemia caused by FGFR1 inhibition and diarrhea due to FGFR4 inhibition, as current therapies are not selective among the FGFRs. Designing selective inhibitors has proved difficult with conventional approaches because the orthosteric sites of FGFR family members are observed to be highly similar in X-ray structures. In this study, aided by analysis of protein dynamics, we designed a selective, covalent FGFR2 inhibitor. In a key initial step, analysis of long-timescale molecular dynamics simulations of the FGFR1 and FGFR2 kinase domains allowed us to identify differential motion in their P-loops, which are located adjacent to the orthosteric site. Using this insight, we were able to design orthosteric binders that selectively and covalently engage the P-loop of FGFR2. Our drug discovery efforts culminated in the development of lirafugratinib (RLY-4008), a covalent inhibitor of FGFR2 that shows substantial selectivity over FGFR1 (~250-fold) and FGFR4 (~5,000-fold) in vitro, causes tumor regression in multiple FGFR2-altered human xenograft models, and was recently demonstrated to be efficacious in the clinic at doses that do not induce clinically significant hyperphosphatemia or diarrhea.

RLY-4008, the First Highly Selective FGFR2 Inhibitor with Activity across FGFR2 Alterations and Resistance Mutations

Oncogenic activation of fibroblast growth factor receptor 2 (FGFR2) drives multiple cancers and represents a broad therapeutic opportunity, yet selective targeting of FGFR2 has not been achieved. Although the clinical efficacy of pan-FGFR inhibitors (pan-FGFRi) validates FGFR2 driver status in FGFR2 fusion-positive intrahepatic cholangiocarcinoma, their benefit is limited by incomplete target coverage due to FGFR1- and FGFR4-mediated toxicities (hyperphosphatemia and diarrhea, respectively) and the emergence of FGFR2 resistance mutations. RLY-4008 is a highly selective, irreversible FGFR2 inhibitor designed to overcome these limitations. In vitro, RLY-4008 demonstrates >250- and >5,000-fold selectivity over FGFR1 and FGFR4, respectively, and targets primary alterations and resistance mutations. In vivo, RLY-4008 induces regression in multiple xenograft models-including models with FGFR2 resistance mutations that drive clinical progression on current pan-FGFRi-while sparing FGFR1 and FGFR4. In early clinical testing, RLY-4008 induced responses without clinically significant off-isoform FGFR toxicities, confirming the broad therapeutic potential of selective FGFR2 targeting.

SIGNIFICANCE

Patients with FGFR2-driven cancers derive limited benefit from pan-FGFRi due to multiple FGFR1-4-mediated toxicities and acquired FGFR2 resistance mutations. RLY-4008 is a highly selective FGFR2 inhibitor that targets primary alterations and resistance mutations and induces tumor regression while sparing other FGFRs, suggesting it may have broad therapeutic potential. See related commentary by Tripathi et al., p. 1964. This article is featured in Selected Articles from This Issue, p. 1949.

The Necessary Nitrogen Atom: A Versatile High-Impact Design Element for Multiparameter Optimization

There is a continued desire in biomedical research to reduce the number and duration of design cycles required to optimize lead compounds into high-quality chemical probes or safe and efficacious drug candidates. The insightful application of impactful molecular design elements is one approach toward achieving this goal. The replacement of a CH group with a N atom in aromatic and heteroaromatic ring systems can have many important effects on molecular and physicochemical properties and intra- and intermolecular interactions that can translate to improved pharmacological profiles. In this Perspective, the "necessary nitrogen atom" is shown to be a versatile high-impact design element for multiparameter optimization, wherein ≥10-, 100-, or 1000-fold improvement in a variety of key pharmacological parameters can be realized.

DOCK 6: Impact of new features and current docking performance

This manuscript presents the latest algorithmic and methodological developments to the structure-based design program DOCK 6.7 focused on an updated internal energy function, new anchor selection control, enhanced minimization options, a footprint similarity scoring function, a symmetry-corrected root-mean-square deviation algorithm, a database filter, and docking forensic tools. An important strategy during development involved use of three orthogonal metrics for assessment and validation: pose reproduction over a large database of 1043 protein-ligand complexes (SB2012 test set), cross-docking to 24 drug-target protein families, and database enrichment using large active and decoy datasets (Directory of Useful Decoys [DUD]-E test set) for five important proteins including HIV protease and IGF-1R. Relative to earlier versions, a key outcome of the work is a significant increase in pose reproduction success in going from DOCK 4.0.2 (51.4%) → 5.4 (65.2%) → 6.7 (73.3%) as a result of significant decreases in failure arising from both sampling 24.1% → 13.6% → 9.1% and scoring 24.4% → 21.1% → 17.5%. Companion cross-docking and enrichment studies with the new version highlight other strengths and remaining areas for improvement, especially for systems containing metal ions. The source code for DOCK 6.7 is available for download and free for academic users at http://dock.compbio.ucsf.edu/.

Selective photocrosslinking of functional ligands to antibodies via the conserved nucleotide binding site

The conserved nucleotide binding site (NBS), found in the Fab variable domain of all antibody isotypes, remains a not-so-widely known and under-utilized site. Here, we describe a UV photocrosslinking method (UV-NBS) that utilizes the NBS for site-specific covalent functionalization of antibodies, while preserving antibody activity. We identified a small molecule, indole-3-butyric acid (IBA), which has affinity for the NBS (K(d) = 1-8 μM) and can be photocrosslinked to antibodies upon UV energy exposure. By synthesizing their IBA conjugated versions, we have successfully photocrosslinked various types of functional ligands to antibodies at the NBS, including affinity tags (biotin), fluorescent molecules (FITC), peptides (iRGD), and chemotherapeutics (paclitaxel). An optimal UV exposure of 1-2 J/cm(2) yielded the most efficient photocrosslinking and resulted in 1-2 conjugations per antibody, while preserving the antigen binding activity and Fc related functions. Analysis of the photocrosslinked conjugates using western blotting, mass spectrometry, and computational docking simulations demonstrated that the photocrosslinking specifically takes place at the Y/F42 residue in framework region 2 of the antibody light chain. Taken together, the UV-NBS method provides a practical, site-specific, and chemically efficient method to functionalize antibodies with significant implications in diagnostic and therapeutic settings.

Design of a heterobivalent ligand to inhibit IgE clustering on mast cells

We describe the design, synthesis, and characterization of a heterobivalent ligand (HBL) system that competitively inhibits allergen binding to mast cell bound IgE antibody, thereby inhibiting mast cell degranulation. HBLs are composed of a hapten conjugated to a nucleotide analog allowing simultaneous targeting of the antigen-binding site as well the "unconventional nucleotide binding site" on IgE Fab domains. Simultaneous bivalent binding to both sites provides HBLs with over 100-fold enhancement both in avidity for IgE(DNP) (K(d) = 0.33 μM) and in inhibition of allergen binding to IgE(DNP) (IC(50) = 0.45 μM) than the monovalent hapten (K(d)(mono) = 41 μM; IC(50)(mono) = 55.4 μM, respectively). In cellular assays, HBL2 effectively inhibits mast cell degranulation (IC(50) = 15 μM), whereas no inhibition is detected by the monovalent hapten. In conclusion, this study establishes the use of multivalency in a novel HBL design to inhibit mast cell degranulation.

Dependence of avidity on linker length for a bivalent ligand-bivalent receptor model system

This paper describes a synthetic dimer of carbonic anhydrase, and a series of bivalent sulfonamide ligands with different lengths (25 to 69 Å between the ends of the fully extended ligands), as a model system to use in examining the binding of bivalent antibodies to antigens. Assays based on analytical ultracentrifugation and fluorescence binding indicate that this system forms cyclic, noncovalent complexes with a stoichiometry of one bivalent ligand to one dimer. This dimer binds the series of bivalent ligands with low picomolar avidities (K(d)(avidity) = 3-40 pM). A structurally analogous monovalent ligand binds to one active site of the dimer with K(d)(mono) = 16 nM. The bivalent association is thus significantly stronger (K(d)(mono)/K(d)(avidity) ranging from ~500 to 5000 unitless) than the monovalent association. We infer from these results, and by comparison of these results to previous studies, that bivalency in antibodies can lead to associations much tighter than monovalent associations (although the observed bivalent association is much weaker than predicted from the simplest level of theory: predicted K(d)(avidity) of ~0.002 pM and K(d)(mono)/K(d)(avidity) ~ 8 × 10(6) unitless).

Fluoroalkyl and alkyl chains have similar hydrophobicities in binding to the "hydrophobic wall" of carbonic anhydrase

The hydrophobic effect, the free-energetically favorable association of nonpolar solutes in water, makes a dominant contribution to binding of many systems of ligands and proteins. The objective of this study was to examine the hydrophobic effect in biomolecular recognition using two chemically different but structurally similar hydrophobic groups, aliphatic hydrocarbons and aliphatic fluorocarbons, and to determine whether the hydrophobicity of the two groups could be distinguished by thermodynamic and biostructural analysis. This paper uses isothermal titration calorimetry (ITC) to examine the thermodynamics of binding of benzenesulfonamides substituted in the para position with alkyl and fluoroalkyl chains (H(2)NSO(2)C(6)H(4)-CONHCH(2)(CX(2))(n)CX(3), n = 0-4, X = H, F) to human carbonic anhydrase II (HCA II). Both alkyl and fluoroalkyl substituents contribute favorably to the enthalpy and the entropy of binding; these contributions increase as the length of chain of the hydrophobic substituent increases. Crystallography of the protein-ligand complexes indicates that the benzenesulfonamide groups of all ligands examined bind with similar geometry, that the tail groups associate with the hydrophobic wall of HCA II (which is made up of the side chains of residues Phe131, Val135, Pro202, and Leu204), and that the structure of the protein is indistinguishable for all but one of the complexes (the longest member of the fluoroalkyl series). Analysis of the thermodynamics of binding as a function of structure is compatible with the hypothesis that hydrophobic binding of both alkyl and fluoroalkyl chains to hydrophobic surface of carbonic anhydrase is due primarily to the release of nonoptimally hydrogen-bonded water molecules that hydrate the binding cavity (including the hydrophobic wall) of HCA II and to the release of water molecules that surround the hydrophobic chain of the ligands. This study defines the balance of enthalpic and entropic contributions to the hydrophobic effect in this representative system of protein and ligand: hydrophobic interactions, here, seem to comprise approximately equal contributions from enthalpy (plausibly from strengthening networks of hydrogen bonds among molecules of water) and entropy (from release of water from configurationally restricted positions).

Preclinical pharmacokinetic/pharmacodynamic models to predict synergistic effects of co-administered anti-cancer agents

PURPOSE

Pharmacokinetic/pharmacodynamic (PK/PD) models have been shown to be useful in predicting tumor growth rates in mouse xenografts. We applied novel PK/PD models to the published anticancer combination therapies of tumor growth inhibition to simulate synergistic changes in tumor growth rates. The parameters from the PK/PD model were further used to estimate clinical doses of the combination.

METHODS

A PK/PD model was built that linked the dosing regimen of a compound to the inhibition of tumor growth in mouse xenograft models. Two subsequent PK/PD models were developed to simulate the published tumor growth profiles of combination treatments. Model I predicts the tumor growth curve assuming that the effect of two anticancer drugs, AZD7762 and irinotecan, is synergistic when given in combination. Model II predicts the tumor growth curve assuming that the effect of co-administering flavopiridol and irinotecan is maximally synergistic when dosed at an optimal interval.

RESULTS

Model I was able to account for the synergistic effects of AZD7762 following the administration of irinotecan. When Model II was applied to the antitumor activity of irinotecan and flavopiridol combination therapy, the modeling was able to reproduce the optimal dosing interval between administrations of the compounds. Furthermore, Model II was able to estimate the biologically active dose of flavopiridol recommended for phase II studies.

CONCLUSIONS

The timing of clinical combination therapy doses is often selected empirically. PK/PD models provide a theoretical structure useful in the design of the optimal clinical dose, frequency of administration and the optimal timing of administration between anticancer agents to maximize tumor suppression.

A synthetic trivalent hapten that aggregates anti-2,4-DNP IgG into bicyclic trimers

This paper describes the synthesis of the trivalent hapten molecule 1, containing three 2,4-dinitrophenyl (2,4-DNP) groups, and the use of this molecule to aggregate three molecules of anti-2,4-DNP IgG into a complex with 3:2 stoichiometry (IgG312). The equilibrium product IgG312 was generated in approximately 90% yield upon mixing IgG and 1; during incubation, thermodynamically unstable, high-molecular-weight aggregates (>104 nm in diameter) form first and convert subsequently to IgG312. The thermodynamics and the kinetics of the formation of aggregates were studied using size-exclusion high-performance liquid chromatography (SE-HPLC), dynamic light scattering (DLS), and analytical ultracentrifugation (AUC). An analytical model based on multiple species in equilibrium was developed and used to interpret the SE-HPLC data. The aggregate IgG312 was more stable thermodynamically and kinetically than monomeric aggregates of this IgG with monomeric derivatives of 2,4-DNP; this stability suggests potential applications of these aggregates in biotechnology.

Thermodynamic parameters for the association of fluorinated benzenesulfonamides with bovine carbonic anhydrase II

This paper describes a calorimetric study of the association of a series of seven fluorinated benzenesulfonamide ligands (C6H(n)F(5-n)SO2NH2) with bovine carbonic anhydrase II (BCA). Quantitative structure-activity relationships between the free energy, enthalpy, and entropy of binding and pKa and log P of the ligands allowed the evaluation of the thermodynamic parameters in terms of the two independent effects of fluorination on the ligand: its electrostatic potential and its hydrophobicity. The parameters were partitioned to the three different structural interactions between the ligand and BCA: the Zn(II) cofactor-sulfonamide bond (approximately 65% of the free energy of binding), the hydrogen bonds between the ligand and BCA (approximately 10%), and the contacts between the phenyl ring of the ligand and BCA (approximately 25%). Calorimetry revealed that all of the ligands studied bind in a 1:1 stoichiometry with BCA; this result was confirmed by 19F NMR spectroscopy and X-ray crystallography (for complexes with human carbonic anhydrase II).

Development and validation of a modular, extensible docking program: DOCK 5

We report on the development and validation of a new version of DOCK. The algorithm has been rewritten in a modular format, which allows for easy implementation of new scoring functions, sampling methods and analysis tools. We validated the sampling algorithm with a test set of 114 protein-ligand complexes. Using an optimized parameter set, we are able to reproduce the crystal ligand pose to within 2 A of the crystal structure for 79% of the test cases using our rigid ligand docking algorithm with an average run time of 1 min per complex and for 72% of the test cases using our flexible ligand docking algorithm with an average run time of 5 min per complex. Finally, we perform an analysis of the docking failures in the test set and determine that the sampling algorithm is generally sufficient for the binding pose prediction problem for up to 7 rotatable bonds; i.e. 99% of the rigid ligand docking cases and 95% of the flexible ligand docking cases are sampled successfully. We point out that success rates could be improved through more advanced modeling of the receptor prior to docking and through improvement of the force field parameters, particularly for structures containing metal-based cofactors.

Increasing the net charge and decreasing the hydrophobicity of bovine carbonic anhydrase decreases the rate of denaturation with sodium dodecyl sulfate

This study compares the rate of denaturation with sodium dodecyl sulfate (SDS) of the individual rungs of protein charge ladders generated by acylation of the lysine epsilon-NH3+ groups of bovine carbonic anhydrase II (BCA). Each acylation decreases the number of positively charged groups, increases the net negative charge, and increases the hydrophobic surface area of BCA. This study reports the kinetics of denaturation in solutions containing SDS of the protein charge ladders generated with acetic and hexanoic anhydrides; plotting these rates of denaturation as a function of the number of modifications yields a U-shaped curve. The proteins with an intermediate number of modifications are the most stable to denaturation by SDS. There are four competing interactions-two resulting from the change in electrostatics and two resulting from the change in exposed hydrophobic surface area-that determine how a modification affects the stability of a rung of a charge ladder of BCA to denaturation with SDS. A model based on assumptions about how these interactions affect the folded and transition states has been developed and fits the experimental results. Modeling indicates that for each additional acylation, the magnitude of the change in the activation energy of denaturation (DeltaDeltaG(double dagger)) due to changes in the electrostatics is much larger than the change in DeltaDeltaG(double dagger) due to changes in the hydrophobicity, but the intermolecular and intramolecular electrostatic effects are opposite in sign. At the high numbers of acylations, hydrophobic interactions cause the hexanoyl-modified BCA to denature nearly three orders of magnitude more rapidly than the acetyl-modified BCA.

Eliminating positively charged lysine epsilon-NH3+ groups on the surface of carbonic anhydrase has no significant influence on its folding from sodium dodecyl sulfate

This study compares the folding of two polypeptides--bovine carbonic anhydrase (BCA) and peracetylated BCA (BCA-Ac(18))--having the same sequence of amino acids but differing by 18 formal units of charge, from a solution containing denaturing concentrations of sodium dodecyl sulfate (SDS). Acetylation of BCA with acetic anhydride converts all 18 lysine-epsilon-NH(3)(+) groups to lysine-epsilon-NHCOCH(3) groups and generates BCA-Ac(18). Both BCA and BCA-Ac(18) are catalytically active, and circular dichroism spectroscopy (CD) suggests that they have similar secondary and tertiary structures. SDS at concentrations above approximately 10 mM denatured both proteins. When the SDS was removed by dialysis, both proteins were regenerated in native form. This study suggests that large differences in the net charge of the polypeptide have no significant influence on the structure, the ability to refold, or the rate of refolding of this protein from solutions containing SDS. This study reinforces the idea that charged residues on the surface of BCA do not guide protein folding and raises the broader question of why proteins have charged residues on their surface, outside of the region of the active site.

Unraveling the interface of signal recognition particle and its receptor by using chemical cross-linking and tandem mass spectrometry

Among the methods used to unravel protein interaction surfaces, chemical cross-linking followed by identification of the cross-linked peptides by mass spectrometry has proven especially useful in dynamic and complex systems. During the signal recognition particle (SRP)-dependent targeting of proteins to the bacterial plasma membrane, the specific interaction between Ffh (the protein component of SRP) and FtsY (the SRP receptor) is known to be essential for the efficiency and fidelity of this process. In this work, we studied the Escherichia coli and Thermus aquaticus Ffh.FtsY complexes by using chemical cross-linking and tandem mass spectrometry to identify nine intermolecular cross-linked peptides. This information was used in conjunction with a previously undescribed model-building approach that combines geometric restraint optimization with macromolecular docking. The resulting model of the Ffh.FtsY complex is in good agreement with the crystal structure solved shortly thereafter. Intriguingly, four of the cross-linked pairs involve the M domain of Ffh, which is absent from the crystal structure, providing previously undocumented experimental evidence that the M domain is positioned in close proximity to the Ffh.FtsY interface in the complex.

Structure of tRNA pseudouridine synthase TruB and its RNA complex: RNA recognition through a combination of rigid docking and induced fit

RNA pseudouridine synthase, TruB, catalyzes pseudouridine formation at U55 in tRNA. This posttranscriptional modification is almost universally conserved and occurs in the T arm of most tRNAs. We determined the crystal structure of Escherichia coli TruB apo enzyme, as well as the structure of Thermotoga maritima TruB in complex with RNA. Comparison of the RNA-free and -bound forms of TruB reveals that this enzyme undergoes significant conformational changes on binding to its substrate. These conformational changes include the ordering of the "thumb loop," which binds right into the RNA hairpin loop, and a 10 degree hinge movement of the C-terminal domain. Along with the result of docking experiments performed on apo TruB, we conclude that TruB recognizes its RNA substrate through a combination of rigid docking and induced fit, with TruB first rigidly binding to its target and then maximizing the interaction by induced fit.