Author Response: Destabilizers of the Thymidylate Synthase Homodimer Accelerate Its Proteasomal Degradation and Inhibit Cancer Growth

Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Appendix 1 Appendix 2 Appendix 3 Appendix 4 Data availability References Decision letter Author response Article and author information Metrics Abstract Drugs that target human thymidylate synthase (hTS), a dimeric enzyme, are widely used in anticancer therapy. However, treatment with classical substrate-site-directed TS inhibitors induces over-expression of this protein and development of drug resistance. We thus pursued an alternative strategy that led us to the discovery of TS-dimer destabilizers. These compounds bind at the monomer-monomer interface and shift the dimerization equilibrium of both the recombinant and the intracellular protein toward the inactive monomers. A structural, spectroscopic, and kinetic investigation has provided evidence and quantitative information on the effects of the interaction of these small molecules with hTS. Focusing on the best among them, E7, we have shown that it inhibits hTS in cancer cells and accelerates its proteasomal degradation, thus causing a decrease in the enzyme intracellular level. E7 also showed a superior anticancer profile to fluorouracil in a mouse model of human pancreatic and ovarian cancer. Thus, over sixty years after the discovery of the first TS prodrug inhibitor, fluorouracil, E7 breaks the link between TS inhibition and enhanced expression in response, providing a strategy to fight drug-resistant cancers. Editor's evaluation Drugs that therapeutically target human thymidylate synthase are widely used to treat cancer. In this manuscript, the authors have discovered and validated an alternative strategy to target this enzyme using dimer destabilizers. By disrupting homodimerization of these proteins using a series of biophysical assays, this research team identified an analog E7 that inhibits hTS in cancer cells and demonstrates a superior profile to approved drugs targeting this pathway. The work is innovative and novel and could lead to the development of new classes of drugs targeting this key enzyme for cancer therapy. https://doi.org/10.7554/eLife.73862.sa0 Decision letter eLife's review process Introduction Small molecules able to bind at specific pockets of the monomer/monomer interface of a dimeric protein may perturb the dimeric assembly to the limit of disrupting it, and thus alter metabolic pathways associated with the cellular functions of the monomeric and dimeric protein forms (Taddia et al., 2015). Such a drastic structural change of the protein may open the way to unexpected events, such as a higher liability to intracellular degradation of the protein monomers relative to the dimers. Human thymidylate synthase (hTS) is an obligate, stable homodimer (Kd = 80 nM Genovese et al., 2010) with two active sites, each including residues from both monomers (Costi et al., 2005). As a dimeric enzyme, it provides the sole de novo pathway to deoxythymidylate (dTMP) synthesis in human cells by catalyzing the reductive methylation of deoxyuridylate (dUMP) to dTMP using methylenetetrahydrofolate (mTHF) as the one-carbon methyl donor (Costi et al., 2005; Carreras and Santi, 1995). By interacting with its own and other mRNAs, this protein regulates its own levels and those of other proteins involved in apoptotic processes, including bcl2, c-myc, and p53 (Brunn et al., 2014; Jennings and Willis, 2015). Its inhibition is usually achieved with compounds that bind at the protein active-site, competing either with the dUMP substrate, such as 5-fluorodeoxyurdine 5’-monophosphate (FdUMP), or with the folate cofactor, such as raltitrexed (RTX) and pemetrexed (PMX) (Carreras and Santi, 1995; Jennings and Willis, 2015; Figure 1A and B). However, these drugs induce cells to develop drug resistance associated with increased hTS levels which eventually leads to therapy failure (Peters, 2018). A drastic change of strategy, based on the design of new compounds with different mechanisms of action, is thus necessary (Voeller et al., 2002; Cardinale et al., 2011). Here we report the discovery of molecules that bind at the hTS monomer/monomer interface and, despite their small size, markedly shift the monomer-dimer equilibrium of the enzyme towards the inactive monomeric form. Acting as destabilizers of the hTS dimer, they not only inhibit the activity of this obligate homodimeric enzyme but also favor its intracellular degradation and, hence, decrease its level. Figure 1 with 1 supplement see all Download asset Open asset Drug target sites on hTS: the active-site (A,B) and the Y202 pocket (C–F). Drugs that are directed to the active site of hTS can mimic either the cofactor MTHF or the substrate dUMP and bind in their binding pockets. (A) hTS (in cartoon with the A and B subunits colored white and grey, respectively) inhibited by the cofactor analogue pemetrexed (PMX, in sticks, purple carbons), which occupies the cofactor binding pocket by establishing a π-π interaction with the dUMP substrate (in sticks, green carbons) (PDB ID: 1JU6). The substrate and cofactor binding sites are represented as green and orange surfaces, respectively. The catalytic cysteine, C195, is highlighted in sticks. (B) The substrate analogue 5-fluoro-2’-deoxyuridine monophosphate (FdUMP) (in sticks, purple carbons) binds inside the catalytic cavity, mainly filling the substrate binding pocket (PDB ID: 3H9K, hTS variant R163K). (C) Location of F59, Y202 and CME202 residues at the hTS dimer interface by superimposition of native hTS and hTS-C195S-Y202C double mutant structures (CME: S,S(2-hydroxyethyl)cysteine; PDB-ID: 3N5G, 4O1X). (D) Surface of the Y202 pocket (cyan) filled by F59’ from the opposite monomer (cyan sticks). The surface of Y202 is highlighted in green (PDB-ID: 3N5G). (E) Surface of the C202 pocket (cyan) filled by F59’ from the opposite monomer (cyan sticks) in the C195S-Y202C double mutant (PDB-ID: 4O1X). The surface of C202 (oxidized as CME) is highlighted in green. (F) Superimposition of the hTS-C195S-Y202C and hTS-Y202C crystal structures with the catalytic loop in hTS-Y202C (inactive conformation, in orange) and in hTS-C195S-Y202C (active conformation, in violet); notice the different orientations of the catalytic residue couples S195/SCH195 and S195’/SCH195’ (PDB-ID: 4O1X, 4O1U) (SCH: S-methyl-thio-cysteine). The one-letter code is used for the standard amino-acids, while for the chemically modified amino-acid, a conventional three-letter code is used as described above. Using a tethering approach in which sulfhydryl-containing fragments were initially identified by reaction with a cysteine residue inserted by mutation around the target site, we identified fragments anchored by disulfide bond formation at the interfacial mutant cysteine residue. We then employed molecular modeling and medicinal chemistry to modify the fragments and develop inhibitors able to bind WT hTS guided solely by their affinities for the interface region. For these small molecules, we propose a model able to quantitatively account both for the dissociation of the hTS dimer, observed spectroscopically, and for the corresponding activity inhibition. We have also shown that their ability to induce the TS-dimer dissociation in cells lies below their efficacy in inhibiting growth of colon, ovarian and pancreatic cancer cells. One of our hTS-dimer destabilizers, compound E7, induced apoptosis in cancer cells and a decrease of hTS levels due to enhanced proteasomal degradation of the enzyme monomers with respect to the dimers. Remarkably, in a mouse model of orthotopic pancreatic cancer, this dimer disrupter caused a higher reduction of cancer growth and had lower toxicity than 5-fluorouracil (5-FU), the prodrug of 5-FdUMP. Like what we observed in vitro, the molecular analysis of exported treated cancer tissues demonstrated decreased hTS protein and mRNA levels. These findings establish E7 as a new lead with an hTS dimer-disruptive ability and link the dimer-to-monomer equilibrium shift of this protein to its faster intracellular degradation. Results Choice of the ligand binding-site on the target protein surface To target the monomer-monomer interface and dissociate the hTS dimer, we applied the ‘cysteine tethering’ approach to fragment-based drug design (Erlanson et al., 2000). We investigated the crystal structures of the active and inactive hTS (PDB-IDs 1HVY and 1YPV, respectively) and carried out molecular dynamics (MD) simulations of the monomeric and dimeric forms of the enzyme to explore the monomer-monomer interface of homodimeric hTS and identify targetable binding pockets and residues suitable to be mutated to cysteine for the tethering experiments (Supplementary file 1). We selected the ‘Y202 pocket’, which accommodates the side chain of F59’ from the other monomer, as the most promising region to target (Figure 1C–F, Figure 2A–B, Figure 1—figure supplement 1; Salo-Ahen et al., 2015). Figure 2 with 1 supplement see all Download asset Open asset Disulfide library design and screening on the hTS C195S-Y202C variant. (A) Crystal structure of the hTS C195S-Y202C dimer (in the cartoon, the A and B subunits are colored pale blue and grey, respectively; the catalytic loops, in active conformation, are highlighted in cyan and pink, respectively) showing the positions of the cysteine residues (in sticks, carbon atoms are colored cyan and pink in subunits A and B, respectively). Compounds A5, A10, A20 and the reference compound N-tosyl-D-prolinecysteamine (TPC) (Appendix 2—figure 1) were selected through the tethering experiments. They are indicated close to the peptide sequence found in the mass spectrometry identification process (Supplementary file 3B and C). Their chemical structures are reported in Supplementary file 3A. (B) Sequence of the hTS C195S-Y202C variant and cysteine-containing peptides generated by trypsin digestion. Cysteine residues are displayed in red (the residue number is reported below each cysteine) and the mutated S195 (the catalytic residue) in blue. Three cysteine-containing peptides were generated by proteolytic cleavage of the hTS C195S-Y202C variant: peptide I177-R185 and peptide R176-R185 containing C180 and peptide D186-R215, containing C199, C202, and C210. Ligand (Lig)-bound cysteines and those alkylated as carbamidomethyl (Cam)-cysteines are displayed. (C) Screening cascade from the 6 million compounds from the ZINC database (Irwin and Shoichet, 2005) to the final four fragments expected to bind in the Y202 pocket. (D, E) Representative MS spectra in the hTS-binding evaluation step. Maldi (D) and ESI (E) MS spectra for the reference compound TPC (Erlanson et al., 2000). The three labeled peaks are due to hTS C195S-Y202C and to the protein bound to BME and one and two molecules of TPC. This pocket lies in the perimeter space of the intermonomer interface and is accessible in the dimeric form of the protein. It is large enough (22 Å3) to accommodate a phenyl group and has several smaller crevices around it. During the MD simulations of monomeric hTS, the side chain of Y202 moved away leaving a larger pocket and behaving like a gate. We therefore made the Y202C mutant and screened for compounds able to form a covalent disulfide bond with C202 from a library of organic disulfides. We also mutated the active site cysteine to serine to prevent the selection of compounds that bound at the active site (C195S). We made mutants with single point mutations at C195S and Y202C to test their functional and structural properties (Supplementary file 1, Figure 1—figure supplement 1) as well as the double mutant, hTS C195S-Y202C. This was used for the ligand selection according to the above-described tethering approach. As expected from removal of the catalytic C195 residue, this double mutant was inactive, but X-ray crystallography showed that it takes the active conformation of hTS (Figure 1F, Figure 2—figure supplement 1, Supplementary file 2A and B). We employed the C195S-Y202C double mutant to capture compounds that, driven by some affinity for the region near residue 202, formed a covalent disulfide bond with C202 (Supplementary file 3A). These compounds were selected by their stability towards reduction by β-mercaptoethanol (BME), and the C202-S-S-small-compound disulfides formed were identified by two mass spectrometric (MS) experiments, MALDI-TOF and ESI-QTOF (Supplementary file 3B and C). Mass-spectrometric ligand selection To identify molecules able to covalently bind near Y202, we designed a library of commercially available compounds from the ZINC database (Irwin and Shoichet, 2005; Figure 2C). We first selected 1066 disulfide compounds that were either symmetric (RSSR) or asymmetric (R1SSR2) and had drug-like or fragment-like properties. 1066 molecules were screened for their molecular properties (PCA, MW, logP, size) and 55 were evaluated with the MS techniques. Thirteen out of the 55 compounds were found to bind the protein according to the results of both MS techniques (Figure 2D and E, Appendix 1, Supplementary file 3A, B and C). Among them, seven compounds, A5, A6, A10, A15, A20, A38, and TPC, were selected based on the feasibility of their chemical modification and were further analyzed by a bottom-up MALDI-TOF and ESI-QTOF MS proteomic analysis to identify the cysteine residues they had bound (Supplementary file 3C). The 7 compounds bound either peptides (I177-R185) IIMCAWNPR (sequence A) or (R176-R185) RIIMCAWNPR (sequence B), both containing the C180 residue of the digested protein (Supplementary file 3C). Four molecules, A5, A10, A20, and TPC also bound peptide (D186-R215) DLPLMALPPSHALCQFCVVNSELSCQLYQR (sequence C) (Supplementary file 3C), containing the engineered Y202C together with the native C199 and C210. In agreement with the described computational and crystallographic data (see above), Y202C is the most exposed of these three cysteines and is thus easily available for disulfide-bond formation with the RS fragments, being therefore suitable for tethering experiments. Other reasons to select the Y202 pocket are that Y202 is not predicted to be part of the mRNA binding region and has the capability to interact with various functional groups. Hence, despite the difficulties in the identification of the precise cysteine to which the fragments could bind, all the above arguments support the choice of the Y202 pocket at the hTS interface as a suitable binding site for further structure-based studies. Therefore, we considered the fragments A5, A10, A20, and TPC for further medicinal chemistry modifications (Figure 3A). Figure 3 with 2 supplements see all Download asset Open asset From hits to dimer-destabilizing compounds. (A) Structures of the compounds identified in MS studies (in red the fragments used for the virtual screening approach to the design of the compound library without the disulfide bonds). (B) Virtual screening cascade and chemical structures of the identified compounds, B12 and B26. (C) Detail of the binding mode of compound C3 (orange sticks) with the hTS interface residues R64, Y202 and R175 (positive and negative potential regions are colored in blue and red, respectively) obtained with docking methods. (D) Stereo view of the superimposition of the calculated binding mode of C3 onto the first monomer (white cartoon) with the opposite monomer overlapping (lime cartoon). Atom color code: nitrogen – blue; oxygen – red; sulphur – yellow. (E) Results of anion exchange chromathography showing co-elution experiments of C3 with hTS dimer and monomer through anion exchange chromatography. Absorbances at 280 (blue) and 350 nm (red) of chromatographic fractions obtained by loading a 50 μM hTS and 4 μM C3 solution on an anion-exchange AEX HiPrep Q HP 16/10 mm column. (Figure 3—figure supplement 1, Figure 3—figure supplement 1—source data 1). From low-affinity compounds to TS homodimer destabilizers Starting from the four fragments identified by tethering/MS (TPC, A5, A10, A20) (Figure 3A), we performed an analogue search that yielded a dataset of 331,600 commercially available compounds (Specs dataset, https://www.specs.net) potentially able to bind non-covalently at the Y202 pocket of the monomer/monomer interface. The initial dataset was progressively reduced to 5,774 candidates using MW, structural and chemical criteria (Figure 3B). These were then docked into the Y202 pocket with the software FLAP (Cross et al., 2010) using a receptor-based pharmacophore model built from the X-ray crystal structure of hTS (PDB: 1HVY). A final set of 26 molecules (B1-B26, Supplementary file 4A) was selected to be tested for their ability to inhibit recombinant hTS and destabilize its dimeric form using, respectively, a kinetic and a spectroscopic, FRET (Förster resonance energy transfer)-based assays (Ponterini et al., 2016). For the FRET-based assay (Genovese et al., 2010), we conjugate one protein monomer with red-emitting tetramethylrhodamine (T) and the other with green-emitting fluorescein (F). Only when the monomers are close to each other, that is, combined to form a dimer, F-to-T excitation energy transfer occurs, a FRET signal is detected, and its efficiency determined. We can thus investigate the dimer/monomer equilibrium of hTS, characterize it by the corresponding dissociation equilibrium constant and evaluate the effect of a tested compound thereupon. Among the 26 compounds selected by VS, B12 and B26 (Figure 3B, Supplementary file 4A), caused low but reproducible decreases in the efficiency of FRET (ΔΦFRET = -0.13 and ΔΦFRET = -0.02, respectively), thus indicating some ability to perturb the enzyme dimeric assembly. In the docked poses, interactions were established by B12 with Y202 through the o-chloro-phenyl ring and with R175 through the p-nitro-pyridyl ring, and by B26 with Y202 through the piperidin-1-yl ring and with R175 through the quinoline ring (Figure 3C and D). Next, we designed and synthesized four sets of compounds (C-F) (Supplementary file 4) to develop B26 into biologically active compounds that exhibited the ability to both disrupt the dimer and inhibit recombinant hTS (Appendix 2—figures 2–7). The most relevant compounds were C2, C3, C4, E5, E6, and E7 (Supplementary file 4B-E). The former three, differing only for the position of the carboxylic group on the phenyl ring, featured IC50 values 5.25, 83 and 246 μM vs. 300 nM hTS (Supplementary file 4) and were selected for a quantitative mechanistic analysis of FRET and inhibition data with recombinant hTS (see the next paragraph). It was however compounds E5, E6, and E7 that proved to be the most interesting to continue our study. They featured IC50 values against 300 nM recombinant hTS of 40, 10 and 7 μM respectively, and decreased FRET efficiency by –0.24 at 50 μM (E5, E6) and –0.1 at 10 μM (E7, Supplementary file 4). Compound E7, that was the most active on cells, was determined to be a racemic mixture of two enantiomers that equilibrated with a half-life of 56 min (Appendix 3, Appendix 3—figure 1, Appendix 3—figure 2). Additional evidence of the ability of compounds C3 and E5 to bind hTS and destabilize its dimers was provided by anion-exchange (AE) chromatographic experiments. Two bands, separated by ca. 2 min, both corresponding to fractions characterized by the absorption spectrum of hTS, were found in the AE chromatogram based on absorbance at 280 nm (Figure 3—figure supplement 1A). The relative area of the second band was only about 3.5% that of the first one and decreased further (Figure 3—figure supplement 1B) when the protein was loaded together with dUMP and Raltitrexed, both at 250 μM, a concentration more than one order of magnitude larger than their dissociation constants from hTS. Because binding of hTS to its nucleotidic substrate stabilizes the active, dimeric form of the enzyme, and consistently with the expectations based on protein size and the correlated elution times, we attribute the two bands to the hTS dimers the first one, and monomers the second. In keeping with this assignment, when hTS was loaded on the AE column after a 1 hr incubation with compound C3 at three concentrations, namely 4, 40, and 80 μM, the monomer band was delayed by additional 3–4 min relative to the dimer band in the chromatograms (Figure 3—figure supplement 1A and Figure 3—figure supplement 1C–E) and its relative area increased markedly, from 0.035 to 0.08, to 0.24 and to 1.09, respectively. Co-elution, hence stable non covalent binding of the C3 inhibitor with hTS, was established by the observation that the fractions containing the hTS dimers and monomers also exhibited absorption at 350 nm (Figure 3E), a wavelength at which no absorption comes from the protein. Accumulation of compound C3 in these fractions was confirmed by identifying and quantifying it in the AE chromatographic fractions by high-resolution mass spectrometry (Supplementary file 5, Figure 3—figure supplement 2). Absorbances at 350 nm in Figure 3E indicated similar concentrations of the compound in the fractions corresponding to the protein dimers and monomers, in spite of the fact that the 280 nm dimer chromatographic band was much more intense than the monomer band. This finding suggests a somewhat larger affinity of the ligand for the latter. Consistently, parallel MS experiments in which 4 μM E5 was loaded on the AE column with hTS identified this ligand mainly in the fractions containing the protein monomers (see additional information in the Supplementary file 5). So, overall, the results synthesized here show that these small compounds co-elute with hTS, thus proving beyond doubt that they remain bound to it, possibly to the monomer more tightly than to the dimer, throughout the AE chromatographic runs. Also, they qualitatively show the ability of these ligands to de-stabilize the hTS dimers in favor of the monomers. Mechanistic analysis of dissociative inhibition The nitrothiophenyl derivative, C3 and its isomers, C2 and C4 (Supplementary file 4B), are structurally similar to compound E7 (Supplementary file 4D) and are much more soluble in water. So, they were used for quantitatively investigating the molecular mechanistic basis of their ability to both inhibit recombinant hTS and disrupt its dimer. The three isomeric compounds showed different inhibitory potencies against 300 nM hTS, with IC50 values of 5.25, 83 and 246 μM, respectively. Consistently, C3 caused the largest decrease in FRET efficiency (ΔΦFRET) at concentrations lower than 50 μM (Figure 4A, Figure 4B). Figure 4 Download asset Open asset Spectroscopic and mechanistic analysis of hTS dissociative inhibition for compounds C2, C3 and C4. (A) Emission spectra of fluorescein(F)- and tetramethylrhodamine (T)-labelled hTS (λexc = 450 nm): the decreasing emission contribution of T relative to F with increasing concentrations of C3 (0, 4.5, 9.5, 15, 25, 100 μM) parallels the decrease in the hTS dimer mole fraction. (B) Dependence of the observed FRET efficiency on the concentrations of three inhibitors (C3, black; C4, blue; C2, red); the hTS dimer concentration was 100 nM. The lines represent the best fittings of the experimental results to Equation S3 in the Appendix 4. The most significant fitting parameters are reported. (C) Cartoon, and standard chemical representations of the dissociative inhibition mechanism of hTS. Blue shapes represent enzyme monomers (M) in the active conformation bound to dUMP; red dots indicate a dissociative inhibitor (I), yellow hexagons, the folate substrate (S), green octagons, the product. (D) Dependence of FRET efficiency on the concentration of C3 with total protein concentrations (ET) of 490 nM (blue circles), 250 nM (green circles), 100 nM (red circles) and 50 nM (black circles). The horizontal segments represent the values of the efficiency in the limit of high inhibitor concentration. At ET = 50 nM and IT = 5 μM the vertical arrow indicates the evolution of the measured value of the FRET efficiency 5, 7, and 12 minu after inhibitor addition (grey and black circles). (E) Analysis of the hTS monomer-dimer equilibrium according to eq. ΦFRET = 1–0.5(ΦFRET/ET)1/2K1/2 (Genovese et al., 2010) and corresponding equilibrium constants without C3 (white circles, K=KD) and at saturating C3 concentrations (black circles, K=K’’D, ΦFRET = FRET efficiency). (F) Dependence of the initial reaction rate (v) on mTHF concentration (free substrate, [S], and total, ST) at four different C3 concentrations (from top to bottom, [I]=0, 12, 24, and 36 μM). The fitting curves represent equation v = xr2STD derived in the Appendix 4 and computed with ET (total enzyme dimer concentration)=300 nM, k=k’=0.9 s–1, k’’=0.1 s–1, K1=K2=10–5 M, K3=2 × 10–5 M, KI = 10–6 M, KI’ = KI’’=10–5 M,x = [M2]1/2 and using for xr the only acceptable root of the quadratic equation in panel 4 F (Figure 4—source data 2). Figure 4—source data 1 Spectroscopic and mechanistic analysis of hTS dissociative inhibition for compounds C2, C3 and C4. https://cdn.elifesciences.org/articles/73862/elife-73862-fig4-data1-v2.xlsx Download elife-73862-fig4-data1-v2.xlsx Figure 4—source data 2 Inhibition of hTS by compound C3. https://cdn.elifesciences.org/articles/73862/elife-73862-fig4-data2-v2.xlsx Download elife-73862-fig4-data2-v2.xlsx The dependence of ΦFRET, that in our conditions is a direct measure of the mole fraction of hTS dimers, on each inhibitor concentration (Figure 4B) was fitted according to the dimer-monomer equilibrium model sketched in Figure 4C. This was analyzed as reported in the Appendix 4. We searched for evidence of the disruptive character of compound C3 by investigating the dependence of its dose-effect curves on total protein concentration, ET (Figure 4D). As expected for a dissociative inhibition mechanism, the limit values of ΦFRET at high C3 concentrations were consistent with a 5-fold increase in the equilibrium dissociation constant for the protein bound to C3 with respect to the free protein (Figure 4E). From the common limit values of ΦFRET at high concentrations (Figure 4B), we conclude that, when saturating hTS, compounds C2-C4 have similar effects on the dimer stability. The fivefold increase in the protein-dimer dissociation equilibrium constant and the corresponding decrease in ΔG° for dimer disruption, from 40.5 to about 36.5 kJ mol–1, ΔΔG° = –4 kJ mol–1, are evidence of the dimer-destabilizing ability of compounds C2-C4. On the other hand, their different abilities to decrease the FRET efficiency at lower concentrations can be attributed to different affinities for the protein dimers, with KI’’s of 2.8x10–5, 1x10–5 and 2x10–5 M, for C2, C3, and C4, respectively. Thus, the observed overall effect of such an inhibitor results from a combination of affinity for the protein and ability to destabilize its dimeric assembly by impairing crucial attractive inter-monomer interactions. Binding of compounds C2-C4 at the Y202 pocket of the hTS monomer/monomer interface is supported by computational docking (Figure 3C and D). In the hTS dimer, the aromatic ring of the Y202 residue forms a π-π interaction with the phenyl ring of the F59’ residue from the opposite subunit. Such a ring is replaced by the phenyl ring of the benzoic acid portions of C2-C4 in their modeled complexes with hTS. Thus, these compounds mimic some of the residue side chains from the opposite monomer in the dimer, providing similar interactions in this region. This finding is supported by the higher affinities of the inhibitors for the hTS monomer (KI = 10–6 M) than for the dimer (KI’=1.7 x 10–5 M), as obtained by fitting of the FRET data in Figure 4B. The reliability of our docking virtual experiments was strengthened by their ability to explain the observed higher affinity for hTS of compound C3 relative to its isomers, C2 and C4. While the nitro-groups of all these ligands can make head-to-head hydrogen-bonding interactions with R64 (Figure 3C–D), only the para carboxylic group of compound C3 arranges properly to establish an additional H-bond with R175. Instead, the meta and ortho carboxylic groups of C2 and C4 do not have the correct orientations for interacting with this arginine residue. As for the mechanism of inhibition of the enzymatic activity of hTS by compounds C2-C4, we show that the scheme in Figure 4C can account for the dependence of the initial reaction rate (v) on the substrate concentration at different inhibitor concentrations with very reasonable fitting parameters (Figure 4F). Simple observation of the dependence of v on [C3] with total enzyme concentration (ET) 300 nM (Figure 4F) indicates a regular decrease with increasing [C3] to a limiting rate that is about 1/5 the rate in the absence of inhibitor. On the other hand, the ET-dependent limiting value of ΦFRET was about half the full one (Figure 4B). This difference in the limit behaviors of v and ΦFRET suggests that, when bound to the inhibitor, the dimeric enzyme is less catalytically active than when free. Or, in terms of the dissociative/noncompetitive model in Figure 4C, K2 and K3 are larger than K1, or k’ and/or k’’ are smaller than k. The curves that go through the kinetic data in Figure 4F are plots of eq. v = xr2STD that was obtained by fast-equilibrium solution of the scheme in Figure 4C (for the derivation of the equation and the meanings of the symbols therein, see the Appendix 4). Remarkably, the fitting parameters, provided in the caption, are consistent with the results of the FRET-based analysis of the equilibria in Figure 4C and the limit behaviors of the initial rate and of ΦFRET. More in detail: (i) k’’ is much smaller than k and k’, that is, M2I2 is almost catalytically inactive, a feature