Structural basis for cancer immunotherapy by the first-in-class checkpoint inhibitor ipilimumab.

Rational modulation of the immune response with biologics represents one of the most promising and active areas for the realization of new therapeutic strategies. In particular, the use of function blocking monoclonal antibodies targeting checkpoint inhibitors such as CTLA-4 and PD-1 have proven to be highly effective for the systemic activation of the human immune system to treat a wide range of cancers. Ipilimumab is a fully human antibody targeting CTLA-4 that received FDA approval for the treatment of metastatic melanoma in 2011. Ipilimumab is the first-in-class immunotherapeutic for blockade of CTLA-4 and significantly benefits overall survival of patients with metastatic melanoma. Understanding the chemical and physical determinants recognized by these mAbs provides direct insight into the mechanisms of pathway blockade, the organization of the antigen-antibody complexes at the cell surface, and opportunities to further engineer affinity and selectivity. Here, we report the 3.0 Å resolution X-ray crystal structure of the complex formed by ipilimumab with its human CTLA-4 target. This structure reveals that ipilimumab contacts the front ß-sheet of CTLA-4 and intersects with the CTLA-4:Β7 recognition surface, indicating that direct steric overlap between ipilimumab and the B7 ligands is a major mechanistic contributor to ipilimumab function. The crystallographically observed binding interface was confirmed by a comprehensive cell-based binding assay against a library of CTLA-4 mutants and by direct biochemical approaches. This structure also highlights determinants responsible for the selectivity exhibited by ipilimumab toward CTLA-4 relative to the homologous and functionally related CD28.

Novel detection of DNA-alkylated adducts of antibody-drug conjugates with potentially unique preclinical and biomarker applications.

BACKGROUND: MDX-1203 is an antibody-drug conjugate (ADC) currently in clinical trials for the treatment of renal carcinoma. The active ingredient of MDX-1203 is a DNA minor groove-binding cytotoxic drug that forms a covalently linked adduct with an adenine base. Formation of this adenine adduct prevents DNA replication, thus triggering cell death. RESULTS: A method has been developed to successfully isolate, identify and quantitate the adenine adduct using LC-MS/MS. The method is highly useful to validate the mode of action of this class of ADCs. Additionally, we have demonstrated that this method could potentially be utilized to assess the efficacy of the ADC in in vitro studies by measuring the amount of adenine adduct in various cells expressing the antigen. CONCLUSION: Upon validation, this method could serve as an invaluable tool to evaluate compounds in preclinical in vivo models and in utilizing the DNA adduct as a potential biomarker.

Evidence from Raman spectroscopy that InhA, the mycobacterial enoyl reductase, modulates the conformation of the NADH cofactor to promote catalysis.

InhA, the enoyl reductase from Mycobacterium tuberculosis, catalyzes the NADH-dependent reduction of trans-2-enoyl-ACPs. In the present work, Raman spectroscopy has been used to identify catalytically relevant changes in the conformation of the nicotinamide ring that occur when NADH binds to InhA. For 4(S)-NADD, there is an 11 cm-1 decrease in the wavenumber of the C4-D stretching band (nuC-D) and a 50% decrease in the width of this band upon binding to InhA. While a similar reduction in line width is observed for the corresponding band arising from 4(R)-NADD, nuC-D for this isomer increases 34 cm-1 upon binding to InhA. These changes in nuC-D indicate that the nicotinamide ring adopts a bound conformation in which the 4(S)C-D bond is in a pseudoaxial orientation. Mutagenesis of F149, a conserved active site residue close to the cofactor, demonstrates that this enzyme-induced modulation in cofactor structure is directly linked to catalysis. In contrast to the wild-type enzyme, Raman spectra of NADD bound to F149A InhA resemble those of NADD in solution. Consequently, F149A is no longer able to optimally position the cofactor for hydride transfer, which correlates with the 30-fold decrease in kcat and 2-fold increase in D(V/KNADH) caused by this mutation. These studies thus substantiate the proposal that hydride transfer is promoted by pseudoaxial positioning of the NADH pro-4S bond, and indicate that catalysis of substrate reduction by InhA results, in part, from correct orientation of the cofactor in the ground state.

A Raman-active competitive inhibitor of OMP decarboxylase.

6-Cyanouridine 5'-phosphate was shown to act as a competitive inhibitor of yeast OMP decarboxylase, with a K(i) value of 1.1 x 10(-5)M. Upon binding by the active site of yeast OMP decarboxylase (EC 4.1.1.23), the Raman stretching frequency of the nitrile group of 6-cyanouridine 5'-phosphate decreases from 2240 to 2225 cm(-1). Based on the behavior of a model compound, 6-cyano-1,3-dimethyluracil, and on vibrational calculations, the observed change in stretching frequency is attributed to desolvation of the ligand, and distortion of the ligand in which the nitrile group moves out of the plane of the pyrimidine ring. Similar distortions may play a role in substrate activation by OMP decarboxylase, contributing to the catalytic process.

Procatalytic ligand strain. Ionization and perturbation of 8-nitroxanthine at the urate oxidase active site.

The binding of the inhibitor 8-nitroxanthine to urate oxidase has been investigated by Raman and UV-visible absorption spectroscopy. The absorption maximum of 8-nitroxanthine shifts from 380 to 400 nm upon binding to the enzyme, demonstrating that the electronic structure of the ligand is perturbed. It has been proposed that oxidation of the substrate urate by urate oxidase is facilitated by formation of the substrate dianion at the enzyme active site, and Raman spectra of urate oxidase-bound 8-nitroxanthine suggest that both the dianionic and monoanionic forms of the ligand are bound to the enzyme under conditions where in solution the monoanion is present exclusively. The C4-C5 stretching frequency appears as a relatively isolated vibrational mode in 8-nitroxanthine whose frequency shifts according to the protonation state of the purine ring. Identification of the C4-C5 stretching mode was confirmed using [4-(13)C]-8-nitroxanthine and ab initio calculation of the vibrational modes. Two peaks corresponding to the C4-C5 stretching mode were evident in spectra of enzyme-bound 8-nitroxanthine, at 1541 and 1486 cm(-)(1). The higher frequency peak was assigned to monoanionic 8-nitroxanthine, and the low-frequency peak was assigned to dianionic 8-nitroxanthine. The C4-C5 stretching frequency for free monoanionic 8-nitroxanthine was at 1545 cm(-)(1), indicating that the enzyme polarizes that bond when the ligand is bound. The C4-C5 stretching frequency in dianionic 8-nitroxanthine is also shifted by 4 cm(-)(1) to lower frequency upon binding. For 8-nitroxanthine free in solution, the C4-C5 stretching frequency shifts to lower frequency upon deprotonation, and the absorption maximum in the UV-visible spectrum shifts to higher wavelength. The spectral shifts observed upon binding of 8-nitroxanthine to urate oxidase are consistent with increased anionic character of the ligand, which is expected to promote catalysis in the reaction with the natural substrate urate. In the Raman spectra of 8-nitroxanthine bound to the F179A, F179Y, and K9M mutant proteins, the C4-C5 stretching frequency was not perturbed from its position for the unbound ligand. Both V(max) and V/K were decreased in the mutant enzymes, demonstrating a correlation between the interaction that perturbs the C4-C5 stretching frequency and the catalytic activity of the enzyme. It is suggested that hydrogen-bonding interactions that lead to precise positioning and deprotonation of the substrate are perturbed by the mutations.

Ring current effects in the active site of medium-chain Acyl-CoA dehydrogenase revealed by NMR spectroscopy.

Medium-chain acyl-CoA dehydrogenase (MCAD) catalyzes the flavin-dependent oxidation of fatty acyl-CoAs to the corresponding trans-2-enoyl-CoAs. The interaction of hexadienoyl-CoA (HD-CoA), a product analogue, with recombinant pig MCAD (pMCAD) has been studied using (13)C NMR and (1)H-(13)C HSQC spectroscopy. Upon binding to oxidized pMCAD, the chemical shifts of the C1, C2, and C3 HD carbons are shifted upfield by 12.8, 2.1, and 13.8 ppm, respectively. In addition, the (1)H chemical shift of the C3-H is also shifted upfield by 1.31 ppm while the chemical shift of the C4 HD-CoA carbon is unchanged upon binding. These changes in chemical shift are unexpected given the results of previous Raman studies which revealed that the C3=C2-C1=O HD enone fragment is polarized upon binding to MCAD such that the electron density at the C3 and C1 carbons is reduced, not increased (Pellet et al. Biochemistry 2000, 39, 13982-13992). To investigate the apparent discrepancy between the NMR and Raman data for HD-CoA bound to MCAD, (13)C NMR spectra have been obtained for HD-CoA bound to enoyl-CoA hydratase, an enzyme system that has also previously been studied using Raman spectroscopy. Significantly, binding to enoyl-CoA hydratase causes the chemical shifts of the C1 and C3 HD carbons to move downfield by 4.8 and 5.6 ppm, respectively, while the C2 resonance moves upfield by 2.2 ppm, in close agreement with the alterations in electron density at these carbons predicted from Raman spectroscopy (Bell, A. F.; Wu, J.; Feng, Y.; Tonge, P. J. Biochemistry 2001, 40, 1725-33). The large increase in shielding experienced by the C1 and C3 HD carbons in the HD-CoA/MCAD complex is proposed to arise from the ring current field from the isoalloxazine portion of the flavin cofactor. The flavin ring current, which is only present when the enzyme is placed in an external magnetic field, also explains the differences in (13)C NMR chemical shifts for acetoacetyl-CoA when bound as an enolate to MCAD and enoyl-CoA hydratase and is used to rationalize the observation that the line widths of the C1 and C3 resonances are narrower when the ligands are bound to MCAD than when they are free in the protein solution.

Probing hydrogen-bonding interactions in the active site of medium-chain acyl-CoA dehydrogenase using Raman spectroscopy.

The role of the oxyanion hole in the reaction catalyzed by pig medium-chain acyl-CoA dehydrogenase (pMCAD) has been investigated using enzyme reconstituted with 2'-deoxy-FAD. The k(cat) (18.8 +/- 0.5 s(-1)) and K(m) (2.5 +/- 0.4 microM) values for the oxidation of n-octanoyl-CoA (C(8)-CoA) by WT pMCAD recombinantly expressed in Escherichia coli are similar to those of native pMCAD isolated from pig kidney. In agreement with previous studies [Engst et al. (1999) Biochemistry 38, 257-267], reconstitution of the WT enzyme with 2'-deoxy-FAD causes a large (400-fold) decrease in k(cat) but has little effect on K(m). To investigate the molecular basis for the alterations in activity resulting from changes in hydrogen bonding between the substrate and the enzyme's oxyanion hole, the structure of the product analogue hexadienoyl-CoA (HD-CoA) bound to the 2'-deoxy-FAD-reconstituted enzyme has been probed by Raman spectroscopy. Importantly, while WT pMCAD causes a 27 cm(-1) decrease in the vibrational frequency of the HD enone band, from 1595 to 1568 cm(-1), the enone band is only shifted 10 cm(-1) upon binding HD-CoA to 2'-deoxy-FAD pMCAD. Thus, removal of the 2'-ribityl hydroxyl group results in a substantial reduction in the ability of the enzyme to polarize the ground state of the ES complex. On the basis of an analysis of a similar system, it is estimated that ground state destabilization is reduced by up to 17 kJ mol(-1), while the activation energy for the reaction is raised 15 kJ mol(-1). In addition, removal of the 2'-ribityl hydroxyl reduces the redox potential shift that is induced by HD-CoA binding from 18 to 11 kJ mol(-1). Consequently, while ligand polarization caused by hydrogen bonding in the oxyanion hole is intimately linked to substrate turnover, additional factors must be responsible for ligand-induced changes in redox potential. Finally, while replacement of the catalytic base E376 with Gln abolishes the ability of the enzyme to catalyze substrate oxidation and to catalyze the exchange of the C(8)-CoA alpha-protons with solvent deuterium, the 2'-deoxy-FAD-reconstituted enzyme catalyzes alpha-proton exchange at a rate (k(exc)) of 0.085 s(-1), which is only 4-fold slower than k(exc) for WT pMCAD (0.35 s(-1)). Thus, either the oxyanion hole plays only a minor role in stabilizing the transition state for alpha-proton exchange, in contrast to its role in substrate oxidation, or the value of k(exc) for WT pMCAD reflects a process such as exchange of the E376 COOH proton with solvent.

Ground state isomerization of a model green fluorescent protein chromophore.

The relationship between ground state cis-trans isomerization and protonation state is explored for a model green fluorescent protein chromophore, 4-hydroxybenzylidene-1,2-dimethylimidazolinone (HBDI). We find that the protonation state has only a modest effect on the free energy differences between cis and trans isomers and on the activation energies for isomerization. Specifically, the experimental free energy differences are 3.3, 8.8, and 9.6 kJ/mol for cationic, neutral, and anionic forms of HBDI, respectively, and the activation energies are 48.9, 54.8, and 54.8 kJ/mol for cationic, neutral, and anionic forms, respectively. Furthermore, these activation energies are much smaller than might be expected based on comparison with similar systems. These results suggest that there may be a sub-population of the chromophore, which is nearly equally accessible to all three protonation states, through which thermal isomerization may proceed.

Light-driven decarboxylation of wild-type green fluorescent protein.

The response of wild-type GFP to UV and visible light was investigated using steady state absorption, fluorescence, and Raman spectroscopies. As reported previously [van Thor, Nat. Struct. Biol. 2002, 9, 37-41], irradiation of GFP results in decarboxylation of E222. Here it is reported that the rate of the light-driven decarboxylation reaction strongly depends on the excitation wavelength, decreasing in the order 254 nm > 280 nm > 476 nm. The relative efficiencies of decarboxylation are explained in terms of the Kolbe-type mechanism in which the excited state of the chromophore acts as an oxidant by accepting an electron from E222. Specifically, it is proposed that 254 nm excitation populates the S2 (or higher) excited state of the chromophore, whereas 404 and 476 nm excitation populate the S1 excited state of neutral and anionic forms, respectively, and that the relative oxidizing power of the three excited states controls the rate of the decarboxylation reaction. In addition, the role of W57 in the photophysics of GFP has been probed by mutating this residue to phenylalanine. These studies reveal that while W57 does not affect decarboxylation, this residue is involved in resonance energy transfer with the chromophore, thereby partially explaining the green fluorescence observed upon UV irradiation of wild-type GFP. Finally, comparison of Raman spectra obtained from nonilluminated and decarboxylated forms of wild-type GFP has provided further vibrational band assignments for neutral and anionic forms of the chromophore within the protein. In addition, these spectra provide valuable insight into the specific interactions between the protein and the chromophore that control the optical properties of wild-type GFP.

Structure-activity studies of the inhibition of FabI, the enoyl reductase from Escherichia coli, by triclosan: kinetic analysis of mutant FabIs.

Triclosan, a common antibacterial additive used in consumer products, is an inhibitor of FabI, the enoyl reductase enzyme from type II bacterial fatty acid biosynthesis. In agreement with previous studies [Ward, W. H., Holdgate, G. A., Rowsell, S., McLean, E. G., Pauptit, R. A., Clayton, E., Nichols, W. W., Colls, J. G., Minshull, C. A., Jude, D. A., Mistry, A., Timms, D., Camble, R., Hales, N. J., Britton, C. J., and Taylor, I. W. (1999) Biochemistry 38, 12514-12525], we report here that triclosan is a slow, reversible, tight binding inhibitor of the FabI from Escherichia coli. Triclosan binds preferentially to the E.NAD(+) form of the wild-type enzyme with a K(1) value of 23 pM. In agreement with genetic selection experiments [McMurry, L. M., Oethinger, M., and Levy, S. B. (1998) Nature 394, 531-532], the affinity of triclosan for the FabI mutants G93V, M159T, and F203L is substantially reduced, binding preferentially to the E.NAD(+) forms of G93V, M159T, and F203L with K(1) values of 0.2 microM, 4 nM, and 0.9 nM, respectively. Triclosan binding to the E.NADH form of F203L can also be detected and is defined by a K(2) value of 51 nM. We have also characterized the Y156F and A197M mutants to compare and contrast the binding of triclosan to InhA, the homologous enoyl reductase from Mycobacterium tuberculosis. As observed for InhA, Y156F FabI has a decreased affinity for triclosan and the inhibitor binds to both E.NAD(+) and E.NADH forms of the enzyme with K(1) and K(2) values of 3 and 30 nM, respectively. The replacement of A197 with Met has no impact on triclosan affinity, indicating that differences in the sequence of the conserved active site loop cannot explain the 10000-fold difference in affinities of FabI and InhA for triclosan.

Stereoselectivity of enoyl-CoA hydratase results from preferential activation of one of two bound substrate conformers.

Enoyl-CoA hydratase catalyzes the hydration of trans-2-crotonyl-CoA to 3(S)- and 3(R)-hydroxybutyryl-CoA with a stereoselectivity (3(S)/3(R)) of 400,000 to 1. Importantly, Raman spectroscopy reveals that both the s-cis and s-trans conformers of the substrate analog hexadienoyl-CoA are bound to the enzyme, but that only the s-cis conformer is polarized. This selective polarization is an example of ground state strain, indicating the existence of catalytically relevant ground state destabilization arising from the selective complementarity of the enzyme toward the transition state rather than the ground state. Consequently, the stereoselectivity of the enzyme-catalyzed reaction results from the selective activation of one of two bound substrate conformers rather than from selective binding of a single conformer. These findings have important implications for inhibitor design and the role of ground state interactions in enzyme catalysis.

Synthesis and spectroscopic studies of model red fluorescent protein chromophores.

[reaction: see text]. Here we describe the synthesis and spectroscopic characterization of two compounds designed to model the chromophore in DsRed, a red fluorescent protein. Comparison with model green fluorescent protein (GFP) chromophores indicates that the additional conjugation in the DsRed models can account, in part, for the red-shifted absorption and emission properties of DsRed compared to those of GFP. In contrast to the GFP models, the DsRed models are fluorescent with quantum yields of 0.002-0.01 in CHCl3.