Folic Acid–Peptide Conjugates Combine Selective Cancer Cell Internalization with Thymidylate Synthase Dimer Interface Targeting

Drug–target interaction, cellular internalization, and target engagement should be addressed to design a lead with high chances of success in further optimization stages. Accordingly, we have designed conjugates of folic acid with anticancer peptides able to bind human thymidylate synthase (hTS) and enter cancer cells through folate receptor α (FRα) highly expressed by several cancer cells. Mechanistic analyses and molecular modeling simulations have shown that these conjugates bind the hTS monomer–monomer interface with affinities over 20 times larger than the enzyme active site. When tested on several cancer cell models, these conjugates exhibited FRα selectivity at nanomolar concentrations. A similar selectivity was observed when the conjugates were delivered in synergistic or additive combinations with anticancer agents. At variance with 5-fluorouracil and other anticancer drugs that target the hTS catalytic pocket, these conjugates do not induce overexpression of this protein and can thus help combating drug resistance associated with high hTS levels.

Cell survival of human ovarian cancer cell lines (showing low FR expression) pag. S22 treated with peptide LR and its conjugate alone or transfected into cells by peptide delivery system.

Enzymatic Assay.
hTS was purified as previously reported [1]. Enzyme solutions were thawed the day of the experiment and the enzyme concentration was determined by UV-Vis spectroscopy using  280 = 89000 M -1 cm -1 and MW=74229 Da. Thawed protein solutions were kept constantly at 4°C. In this condition, enzyme was able to reproduce normal kinetic activity values (K M dUMP = 10-12 µM, K M mTHF = 4-6 µM, k cat = ). An aliquot of recombinant hTS (2.5-3 μg) was added to 0.05 mL of assay kinetic buffer (50 mM TES, pH 7.4, 25 mM MgCl 2 , 6.5 mM HCHO, 1 mM EDTA, 75 mM β-mercaptoethanol (β-ME)) and was incubated with a peptidic inhibitor at a given concentration for 1 hour at 37°C under gentle orbital shaking. Then, the remaining reactants of the kinetic assay were added to the reaction mixture as follows: mTHF in one of the concentrations above, dUMP 0.14 mM and bi-distilled water up to 0.1 mL. As a limiting-rate substrate, the dUMP was added last. From then on, the absorbance change was monitored at 340 nm in a UV-visible microplate spectrophotometer for 3 min. The slope at the origin was determined by using the following formula: Rate = (A s -A bl ) / (t 2 -t 1 ), (where A s is the sample reading, A bl is the blank solution reading subtracted from the corresponding sample reading at each time point, t 2 -t 1 are the limits of the time S6 interval for the best fit straight line) and yielded the initial reaction rate, v. The test was repeated 3 times (Carosati et al., 2012).

S7
Superposition of FA-LR docked in TS active site with raltitrexed.

Supporting Figure SI-2.
Superposition of the docked conformation of FA-LR (cyan) with raltitrexed (yellow) co-crystallized with hTS (PDB code 1hvy). The protein is displayed in green cartoon while the ligands in capped sticks.

Molecular Modelling of FA-LR and FA-[DGln 4 ]LR at TS monomer-monomer interface.
The superimposition of the docking poses of FA-LR and FA-[DGln 4 ]LR with the structure of the LR peptide co-crystallized at the interface of hTS inactive form (PDB code 3n5e) is reported in Figure   4G and 4H, and shows a number of similarities among the predicted and experimental orientation of the peptide region.
In the case of FA-LR ( Figure 4G and SI-3a), the orientation of Leu1 is well conserved, while the sidechain of Ser2 points in the opposite direction. A very closed position to the co-crystallized one is also maintained by Cys3, still located at a disulfide bridge distance with respect to the protein Cys180, and by Gln4, whose side-chain is well superimposed with the X-ray conformation. Major adjustments are, instead, underwent by the last three residues, i.e. Tyr6, Gln7 and Arg8. A slightly different orientation is assumed by the FA-[DGln 4 ]LR conjugate ( Figure 4H and SI-3b). The side-chain of Leu1 lies at a short distance with respect to the crystallographic structure, as well as Cys3. On the contrary, the side-chain of both Ser2 and Leu5 point towards an opposite direction. Gln4 and Gln7 are shifted with respect to the X-ray orientation, while Tyr6 well resembles the original pose. Finally, as in the case of FA-LR, Arg8 assumes a completely different orientation with respect to the LR peptide. As previously underlined, in both cases the central part of the peptide region lies deeply into the crevice at the protein subunit interface while the folic moiety and the tail residues Tyr6, Gln7 and Arg8 occupy more solvent-exposed areas. The Goldscore Fitness scores for the two conjugates pose reported in Figure 4G and 4H were 79 and 89, respectively.
A detailed representation of the peptide orientation at the protein subunit interface and of the interactions made with the protein residues is reported in Supporting Figure SI

Radioligand assay
To accurately evaluate the functional FR protein levels, which is crucial for most FR-targeted therapies, we have also performed radioligand binding assays (Reddy et al., 1999). Once more, the IGROV-1 cells exhibited the highest amount of functional FR on their surface: more than 90 fmol of

Sample preparation for LC-MS/MS analysis (FA-LR compartimentalization)
IGROV-1 human ovarian cancer cells were cultured in RPMI medium supplemented with 10% fetal bovine serum (FBS) plus 20mM L-Gln at 37°C with 5% CO 2 . RPMI medium was aspirated and c) S15

Detection and quantification limits (sensitivity)
Lower limit of quantification, LLOQ, was defined as the concentration at which the quantifier transition of FA-LR yielded a signal to noise ratio of 10 and it was 0.10 µg/ml and 0.12 µg/ml in cell lysate and PBS medium, respectively. Limit of detection (LOD) was estimate based on 3:1 signal-tonoise ratio and it was 0.04 µg/ml in cell lysates and PBS.

Specificity
Typical MRM chromatograms of FA-LR and IS in cell lysate showed no interference peaks from endogenous substances at the retention times of LR and IS. The chromatograms of blank cell lysate also had no interference peaks at the retention times of FA-LR and IS. The specificity results indicate that the method is highly selective for FA-LR analysis.

Calibration curve and Linearity
Calibration curves for FA-LR were performed in cell lysates (0.10 to 8.00 µg ml -1 ) and in RPMI medium (0.12-10.00 µg ml -1 ). Five replicates were used to establish the linear calibration equation (y=mx+c) and analyzed using the ratio of analyte peak area over IS peak area after quantitative integration by MassHunter software. Linearity was measured as the coefficient of determination (R 2 ) measured for five calibration replicates. The acceptance criteria for each back-calculated standard concentration were +15% deviation from the nominal value except at LOQ, which was set at +20%.
Linearity was excellent over the respective calibration ranges with corresponding correlation coefficient > 0.99. The results are reported in Table SI-2.       nm. The extracted dye was proportional to the cellular biomass.
As it appears in Fig. SI-9, these cell lines showing low FR on their surface, were able to almost completely survive to FA-LR treatment; on the contrary, when both FA-LR and LR were transfected by means of the delivery system, which allow them to reach intracellular concentration higher than those obtained by the internalization through the few FR, cell survival was reduced by 25-50 %, depending on cell line.