Structure-based identification of a non-covalent thioredoxin reductase inhibitor with proven ADMET suitability

Abstract

Thioredoxin reductase 1 (TrxR1), a selenoprotein enzyme crucial for redox homeostasis in mammals, has emerged as a promising target for anticancer therapy. In this study, we present a non-covalent TrxR1 inhibitor, identified through an integrated experimental and computational approach. After identifying a plausible druggable cavity, a molecular docking-based virtual screening of over 90,000 lead-like compounds was performed. The selected compounds were evaluated for their impact on TrxR1 activity, leading to the identification of the most promising candidate, C55. The identified compound, already proven to be free from potential ADMET concerns, inhibits TrxR1 in a dose-dependent manner, with an IC₅₀ value in the micromolar range. C55′s activity was confirmed across multiple cancer cell lines, including HepG2, Huh7, MCF-7, and MDA-MB-231 cells. As a metal-free organic molecule capable of non-covalently inhibiting TrxR1, C55 represents a significant breakthrough, offering a solid foundation for hit-to-lead optimisation and the development of new anticancer drug candidates.

GRAPHICAL ABSTRACT

Introduction

Thioredoxin reductases (TrxRs) are crucial enzymes in maintaining cellular redox balance, playing a key role in protecting against oxidative stress and regulating protein function through the thioredoxin system¹. TrxRs are flavoproteins containing the 21^st encoded amino acid selenocysteine² (Sec, U), which is essential for the protein’s reducing activity. Three TrxRs are present in mammals: thioredoxin reductase 1 (TrxR1), thioredoxin reductase 2 (TrxR2), and thioredoxin reductase 3 (TrxR3). In numerous tumour types, the overexpression of cytosolic TrxR1 highlights its pivotal role in cancer cell survival, particularly in breast cancer (BCa)³⁴, and hepatocellular carcinoma (HCC) cells⁵. Unlike normal cells, cancer cells exhibit heightened dependence on TrxR1 activity due to the increased oxidative stress induced by their altered metabolic processes. Targeting TrxR1 could disrupt their ability to manage this stress, presenting a compelling strategy for therapeutic intervention. Elevated expression of TrxR1 has been associated with a poor prognosis in patients with BCa and HCC as well as an increased risk of recurrence⁴, indicating that the enzyme may play a role in tumour progression and treatment resistance⁶. Thus, TrxR1 is a promising therapeutic target, as its inhibition could reduce tumour proliferation and increase sensitivity to conventional therapies⁷.

Recent studies have therefore focused on the design of selective TrxR1 inhibitors, which could limit the ability of tumour cells to maintain their favourable redox environment, thereby increasing their vulnerability to treatment and improving outcomes in patients with tumours with high TrxR1 expression^7–10. Such therapeutic approaches could represent a new frontier in treating tumours resistant to conventional drugs. For all these reasons, the inhibition of TrxR1 has emerged as a promising strategy for cancer treatment, in recent years^11–14. In particular, the most significant and well-studied class of TrxR1 inhibitors is represented by gold-containing compounds^12,15–19. Initially recognised within the medicinal chemistry community for their efficacy in treating rheumatoid arthritis (RA) and tuberculosis^20,21, these gold-based compounds have recently been explored as anticancer agents due to their cytotoxic activity across various cancer cell types^22,23. One noteworthy example is represented by Auranofin, a gold(I)-containing compound initially approved as an antiarthritic drug, which has since been repurposed for cancer treatment^22,24. In ovarian cancer, Auranofin demonstrates cytotoxic effects linked to its ability to inhibit the antioxidant activity of TrxR1. This inhibition reduces the removal rate of peroxides, leading to the accumulation of mitochondrial H₂O₂ and subsequent cell damage^25,26. As an alternative to gold, platinum-based compounds have been employed for TrxR1 inhibition. For instance, cisplatin [cis-diamminedichloridoplatinum (II)] is a well-known platinum-based compound that has been approved for cancer treatment^27,28. Notably, cisplatin, in addition to acting on the bases of DNA²⁹ has demonstrated highly specific and irreversible inhibition of TrxR1³⁰. Another compound known to inhibit TrxR1 is arsenic trioxide³¹, which has been established as an effective treatment for acute promyelocytic leukaemia³². More recently, a rhodium(I) N-heterocyclic carbene complex has also been described to inhibit TrxR, increase intracellular ROS accumulation, and promote apoptosis in HCC cells⁵. In general, these metal-based complexes are designed to inhibit TrxR1 through the covalent modification of the catalytic Sec, displaying a high affinity towards metals³³. In a similar way, β-unsaturated carbonyl compounds known as Michael acceptors could inhibit TrxR1 by covalent interaction with the Sec³⁴. Although these molecules are undoubtedly interesting for developing new therapeutic strategies based on TrxR1 inhibition, they have several significant limitations. First, most of them contain metals, raising concerns about potential toxicity. Gold-based compounds, for instance, can cause nephrotoxicity, allergic dermatitis, and, in some cases, myelotoxicity³⁵. Platinum-based compounds, although still in use, are known to cause significant renal damage, peripheral neuropathy, hearing loss, and myelotoxicity (bone marrow suppression)^36,37. Noteworthy, most of the compounds identified so far as TrxR1 inhibitors covalently modify the protein. Identifying non-covalent inhibitors of TrxR1 could offer several advantages in terms of: i) specificity: covalent inhibitors of TrxR1 target selenocysteine sharing a similar reactivity with cysteine which is a more common amino acid in proteins, potentially leading to off-target effects; ii) pharmacokinetics properties: non-covalent inhibitors provide a more tuneable duration of action; iii) easier optimisation: non-covalent compounds can be more easily optimised through structural modifications to refine their affinity, selectivity, and bioavailability without relying on covalent interaction with a specific residue.

The present work represents a first attempt to address the limitations of the covalent inhibition of TrxR1 through an integrated computational/experimental approach. Using a workflow combining cavity mapping, molecular docking, direct binding assays, and cell-based assays, we have developed a virtual screening (VS) protocol for identifying compounds capable of non-covalently inhibiting TrxR1. Remarkably, to ensure that the selected compounds are free from potential Adsorption Distribution Metabolism Excretion and Toxicity (ADMET) issues and, therefore, suitable for rapid hit-to-lead optimisation, the screening was conducted on the Asinex Elite library, which comprises more than 90,000 compounds that have already been pre-screened against a panel of early ADMET tests. The study enabled us to identify a new compound with notable pharmacokinetic properties that inhibits TrxR1 non-covalently. The paper is discussed in the perspective of using this compound as starting point for subsequent rational optimisation studies. Notice that this study has previously made available as a preprint³⁸.

Materials and methods

Cavity mapping and data retrieval

Cavity mapping was performed on the X-ray coordinates of recombinant TrxR1 with reduced C-terminal tail (PDB code: 3EAN)³⁹. To this aim, we used the SiteMap⁴⁰ tool, available in the Schrodinger suite 2020–1. The protein was prepared using the protein preparation wizard tool available in the Schrödinger Suite^41,42. This preparation involved determining protonation states at pH 7.4, minimising energy, and removing water molecules. All the ligands were prepared using the LigPrep tool⁴³ employing the OPLS5 force field for energy minimisation⁴⁴, and a maximum of 32 stereoisomers for each ligand. Tautomeric forms at pH 7.4 were also generated.

Docking simulations

Docking simulations were performed using the GOLD docking software⁴⁵. GOLD employs a genetic algorithm for conformational sampling and defines its own performance metrics. All genetic algorithm parameters were set to auto so that GOLD was free to find the optimal parameters for the specific calculation. Building on the performed cavity mapping, the box was centred on the atom number 376 (radius = 10 Å), which is the selenium atom in the selenocysteine 498 residue. GOLD was instructed to restrict the binding site definition to concave, solvent-accessible surfaces using the LIGSITE algorithm⁴⁶. Chemscore⁴⁷ was used as a scoring function. Note that, in addition to the Chemscore value, its binding energy term was also reported, as recommended in the original paper by Eldridge et al.⁴⁸

Virtual screening

A docking-bases VS campaign was conducted on the entire Asinex Elite dataset⁴⁹, which includes 91,001 commercially available ‘lead-like’ compounds. These compounds underwent experimental screening against an initial set of ADMET tests to confirm they were free of potential ADMET issues and suitable for rapid hit-to-lead optimisation. Each compound was prepared and docked into the crystal structure, as described. The compounds were then divided into ten groups, and the docking was run in parallel on an HP Z4 workstation, a strategy that significantly optimised computational time. The compounds were then ranked based on the returned ChemScore, and the top 500 poses were selected for the next steps. The binding modes of the top 500 compounds underwent rigorous visual inspection, a crucial step in the decision-making process following a virtual screening (VS) procedure⁵⁰. 63 compounds were finally selected for further experimental testing.

Cell lines and cell culture

Human HCC HepG2 (HB-8065^TM) cells were obtained from the American Type Culture Collection (ATCC, Rockville, MD). Cells were maintained in Roswell Park Memorial Institute (RPMI) 1640 medium (MERCK, Milan, Italy) containing 10% (v/v) Foetal Bovine Serum (FBS) (Gibco-Thermo Fisher Scientific, Milano, Italy), 1 mM sodium pyruvate (MERCK), 100 units/ml penicillin (MERCK) and 100 µg/ml streptomycin (MERCK). The HCC Huh7 cell line was a gift from Professor Massimo Levrero (Sapienza University of Rome, Italy) and was maintained in a high-glucose Dulbecco’s Modified Eagle Medium (DMEM - MERCK) containing 10% FBS. The human breast cancer cell lines MCF-7 and MDA-MB-231 were obtained from ATCC (HTB-22^™ and HTB-26^™, respectively) and were maintained in high-glucose DMEM medium containing 10% FBS and antibiotics.

Cell viability assay

Cells (5 x 10³/well) were seeded in 96-well plates and, after 24 h, were exposed to the selected Asinex Elite compounds at different concentrations from 1 to 100 µM. It is worth noting that in the case of chiral compounds, a mixture of stereoisomers was tested since the supplier was not able to provide enantiomerically pure compounds. After 72 or 96 h, MTS assays were performed using the CellTiter Aqueous OneSolution kit (Promega Corporation Madison, WI, USA) according to the manufacturer’s instructions. Cell viability was expressed as a percentage of the absorbance measured in the control cells. Values were expressed as means ± Standard Deviation (SD) of at least two experiments, each performed in triplicate.

Measurement of TrxR1 activity

TrxR1 activity was assessed using the Thioredoxin Reductase Assay Kit according to the manufacturer’s instructions (Sigma-Aldrich, Milan, Italy). The TrxR1 enzyme, extracted from rat liver, was supplied with the kit (catalog number CS0170), and was provided in ready-to-use form by the manufacturer. In brief, various concentrations of SYN19823426 (C55) compound and Auranofin (Sigma-Aldrich) were incubated with TrxR1 extracted from rat liver and added to each well of 96-well microtiter plates in the presence of NAPDH and 5,5′-dithio-bis(2-nitrobenzoic acid) (DTNB) at 25 °C for 60 min. TrxR catalyses the reduction of DTNB with NADPH to 5-thio-2-nitrobenzoic acid (TNB), which produces an intense yellow colour. Therefore, TrxR1 activity was spectrophotometrically determined at 412 nm absorbance using a microplate reader (Bio-Rad Laboratories, CA, USA). Values were expressed as a percentage of activity compared to solvent alone ± SD of at least two separate experiments. Furthermore, the determination of TrxR1 activity was performed on HepG2, Huh7, MCF-7 and MDA-MB-231 whole cell lysates. Briefly, exponentially growing adherent cells were lysed by adding ice-cold Radio-Immunoprecipitation Assay (RIPA) buffer (Cell Signalling Technologies Inc., Beverly, MA, USA) containing proteases and phosphatase inhibitors (Sigma-Aldrich) for 20 min on ice. Lysates were centrifuged, and protein concentration in supernatants was quantified by a Protein assay kit (Bio-Rad). Therefore, 10 μg of proteins were used to determine TrxR1 activity in cell lysates using the Thioredoxin Reductase Assay Kit as reported above. The values were expressed as a percentage of activity compared to solvent alone ± SD of at least two separate experiments using two whole cell lysates for each cell line.

Measurement of Trx1 activity

Thioredoxin-1 (Trx1) activity was measured using the ProteoStat^™ Thioredoxin-1 assay kit (Enzo Life Sciences, Plymouth Meeting, PA, USA) according to the manufacturer’s protocol. Briefly, 10 μL of the test compound C55, at different concentrations, was added to each well of a black 96-well microplate. For control, 10 μL of Aurothiomalate working solution (provided in the kit and prepared as recommended) was dispensed into designated wells. Subsequently, 70 μL of Thioredoxin-1 Assay Master Mix and 10 μL of dithiothreitol (DTT) were added to each well. The plate was incubated for 30 min at room temperature in the dark. The reaction was stopped by adding 10 μL of the Stop Reagent working solution to each well, followed by gentle mixing for ∼30 s. Fluorescence was measured using a GloMax^® microplate reader (Promega) at an excitation wavelength of 500 nm and an emission wavelength of 603 nm.

RNA extraction and real-time PCR

Total RNA was extracted using the TRIzol method and 1 μg of RNA was used for reverse transcription to generate cDNA using the QuantiNova Reverse Transcription Kit (QIAGEN), following the instructions of the manufacturer. Quantitative Real-time PCR was analysed using the QuantiNova SYBR Green PCR Kit (QIAGEN) and QuantiTect primers (QIAGEN) specific for NQO1 (QT00050281), HO-1 (QT00092645) and β-actin (QT00095431). Data are expressed as the relative mRNA expression level of the different genes in treated cells compared with control cells. Values given are the mean ± SD of three different experiments performed in triplicate.

Intracellular ROS detection

Cells were seeded in 96-well plates and after 24 h were exposed to C55 or Auranofin. The cells were washed with phosphate-buffered saline (PBS) prior to probing with 40 μM H₂DCFDA (Invitrogen, Milan, Italy) for 1 h at 37 °C with 5% CO₂ in the dark. Excess H₂DCFDA was removed with PBS. The fluorescence intensity was measured using a GloMax^® microplate reader, with excitation and emission wavelengths of 485 nm and 530 nm, respectively. The fluorescence intensity of dichlorofluorescein (DCF) was calculated as the fluorescence intensity of treated cells relative to non-treated cells.

Western blotting

RIPA buffer (Cell Signalling Technologies Inc.) was used to obtain cell lysates, and Western blotting was performed using the methodology for the Odyssey^® infra-red imaging system (LI-COR Biosciences, NE, USA). After transfer nitrocellulose membranes were placed in Odyssey^® blocking buffer (OBB, LI-COR) diluted in tris-buffered saline (TBS) and incubated for 1 h at room temperature. Primary antibodies were diluted in OBB. Secondary antibodies conjugated to IRDye^® 800CW (LI-COR) or Alexa Fluor 680 (Molecular Probes, Invitrogen Carlsbad, CA, USA) were diluted in OBB. Membranes were scanned and analysed with an Odyssey IR scanner using Odyssey 3.0 imaging software. The relative expressions were calculated as the ratio of drug-treated samples vs. control (DMSO) and corrected using the quantified level of β-actin expression. Primary antibodies raised against γ-H2AX and β-actin were purchased from Cell Signalling Technologies and Sigma-Aldrich, respectively.

C55 stability and integrity assessment

The identity and stability of compound C55 in DMSO over one month were evaluated by Electrospray ionisation (ESI) mass spectrometry. Samples were prepared by diluting 1 µL of a 10 mM C55 solution in DMSO with 100 µL of methanol. Spectra were recorded in positive‐ion mode on an Applied Biosystems mass spectrometer equipped with a triple‐quadrupole analyser. The observed molecular‐ion peak confirmed both the integrity and the unchanged mass of C55 throughout the ageing period (see Figure S1 in the Supplementary Information).

Results and discussion

Pocket recognition

As a first step of the study, we searched for a binding site on the TrxR1 surface that could accommodate small molecules (SMs). The information regarding the protein portion responsible for the SMs accommodation is, in fact, still elusive with the few structure-based studies present in the literature that are not aligned regarding its exact localisation^51,52. The X-ray 3D structure of the protein published by Cheng et al. (PDB code: 3EAN³⁹) was used as input for Sitemap. Among the five detected cavities, P1, shown in Figure 1, was predicted to be the most promising one, as it returned the highest SiteScore (1.039) and DScore (1.075) values. The identified cavity, nestled between chains A and B of the protein and in proximity to the residue selenocysteine in position 498 (U498), holds significant promise. The computed SiteScore and DScore values, characteristic of sites with high druggability potential^40,47, underscore the potential impact of our findings on structure-based drug design approaches.

Figure 1.

Druggable cavity (P1) identified by SiteMap on the dimer of the TrxR1 protein structure (PDB code: 3EAN)³⁹. The two monomers are designated as chain A (green) and chain B (blue), while P1 is shown as a grey surface. The selenocysteine U498 of chain A is highlighted using a stick representation.

Virtual screening

To identify SMs able to target the chosen cavity, we performed a VS of a large chemical library using the software GOLD⁴⁵, available as a Cambridge Crystallographic Data Centre (CCDC) product. All the compounds were docked on the identified TrxR1 binding site (P1) employing as protein structure the one published by Cheng et al.³⁹ Based on the returned ChemScore, the top 500 compounds were selected and the corresponding top-scored poses visually inspected. We finally selected 63 initial hits for further in-vitro experimental tests (Figure 2).

Figure 2.

2D structures and Asinex title of the 63 compounds selected based on the performed VS procedure. The molecule selected as a TrxR inhibitor is highlighted with a red box, while the other molecules that showed interesting cytotoxicity values are highlighted with a black box.

Cell-based screening of asinex library compounds

Only 37 of the 63 compounds selected from the Asinex chemical library were completely soluble in DMSO at a concentration of 50 mM and therefore tested for their effect on cell cytotoxicity against human HCC and BCa cell lines. Initial screening was conducted by MTS assay for 72 h at concentrations of 10, 50 and 100 µM, and only samples found to have cytotoxic effects at a concentration ≤ 50 µM on at least 2 out of 4 cell lines were used for further analysis (Table 1).

Table 1.

IC₅₀ values of the 37 compounds against HepG2, Huh7, MCF-7 and MDA-MB-231 cells for 72h of treatment.

		IC50 (µM)
compound number	compound name	HepG2	Huh7	MCF-7	MDA-MB-231
C4	SYN22041065	96.3 ± 6.6	60.5 ± 10.2	85.5 ± 7.2	> 100
C5	SYN19990927	> 100	> 100	> 100	> 100
C18	AEM10027871	> 100	> 100	> 100	> 100
C21	LEG19990218	28.5 ± 9.8	25.7 ± 1.8	70.1 ± 3.0	51.4 ± 7.8
C22	LEG20009354	> 100	> 100	> 100	> 100
C23	LEG15341540	60.5 ± 1.8	51.7 ± 15.41	> 100	> 100
C24	LMK17221003	48.6 ± 15.3	50.2 ± 25.5	> 100	> 100
C25	LEG18687308	> 100	> 100	> 100	> 100
C26	SYN17860685	> 100	> 100	> 100	> 100
C27	ASF192797 50	21.8 ± 2.3	51.2 ± 12.5	> 100	> 100
C28	ASF19632166	> 100	> 100	> 100	> 100
C29	ADM19835767	> 100	> 100	> 100	> 100
C30	AOP18687354	> 100	> 100	> 100	> 100
C31	SYN15659685	> 100	> 100	> 100	> 100
C33	SYN15643124	> 100	> 100	> 100	> 100
C34	LEG17851037	> 100	> 100	> 100	> 100
C35	SYN17920976	> 100	> 100	> 100	> 100
C36	SYN17736818	> 100	> 100	> 100	> 100
C37	LMG18101294	> 100	> 100	> 100	> 100
C39	AEM16401255	63.4 ± 12.3	23.9 ± 12.1	81.0 ± 9.4	> 100
C40	ASF17521182	61.7 ± 24.5	78.55 ± 19.1	> 100	> 100
C41	AEM18663974	18.8 ± 14.7	> 100	48.9 ± 15.7	55.13 ± 11.1
C42	AEM18511305	45.9 ± 1.8	56.4 ± 12.5	49.2 ± 16.6	> 100
C44	ASF19865209	70.4 ± 25.5	72.9 ± 18.5	> 100	> 100
C45	LEG19971843	> 100	98.2 ± 11.1	> 100	> 100
C47	SYN13244281	54.1 ± 5.4	27.7 ± 9.0	49 ± 1.1	> 100
C49	SYN22976891	> 100	> 100	> 100	> 100
C50	LEG15394831	> 100	60.2 ± 11.0	> 100	> 100
C51	ASF20692226	> 100	73.1 ± 21.0	72.1 ± 12.5	> 100
C52	AOP20989282	> 100	> 100	> 100	> 100
C53	AOP22041441	61.1 ± 18.2	72.6 ± 20.0	> 100	> 100
C54	SYN22964503	> 100	> 100	> 100	> 100
C55	SYN19823426	29.4 ± 5.3	29.9 ± 3.6	> 100	42.7 ± 1.6
C57	LEG19412213	> 100	> 100	> 100	> 100
C58	IPE21905364	61.4 ± 12.2	24.4 ± 12.0	43.2 ± 22.0	> 100
C59	SYN20054602	> 100	> 100	> 100	> 100
C60	ASF18586547	> 100	> 100	> 100	> 100

Cell-free and cell lysates thioredoxin reductase 1 activity

Following the cell-based screening, we identified six compounds (C21, C41, C42, C47, C55, and C58) for further investigation. Our initial tests in a cell-free assay, with all compounds tested at concentrations up to 10 µM, revealed that five of the six compounds showed very low (less than 25%) or no inhibitory activity against TrxR1. However, compound C55 (SYN19823426) (Figure 3) stood out.

Figure 3.

Inhibition of TrxR1 activity by selected compounds in cell-free assay. Rat liver TrxR1 and were incubated with 10 µM of each compound and 0.5 µM of Auranofin (AUR). TrxR1 activity was assayed as described in material and methods. Values are expressed as percentage of activity compared to solvent alone (-).

We then proceeded to test the inhibitory activity of compound C55 on TrxR1 in a dose-response experiment. As shown in Figure 4(A), C55 inhibited TrxR1 activity in a dose-dependent manner, with an IC₅₀ of 4.7 µM ± 1.3. Furthermore, we extended our evaluation to protein samples extracted from HepG2, Huh7 MCF-7, and MDA-MB-231 cells. Figure 4(B) illustrates that C55′s inhibitory activity on TrxR1 varied depending on the cell type, underscoring the need for further investigation into its potential differential effects. To ensure that the measured enzymatic activity reflected TrxR1-specific activity and was not influenced by low-molecular-weight thiols, we performed the assay in the presence of Auranofin, a selective TrxR1 inhibitor. As shown in Figure S2, the marked reduction in activity upon Auranofin treatment confirmed the specificity of the assay for TrxR1. Furthermore, to investigate the specificity of the interaction between C55 and TrxR1, particularly with respect to potential interference from small molecule thiols, we conducted a thioredoxin 1 (Trx1) activity assay in the presence of the compound, since this protein is known to contains thiol groups in its active site. Aurothiomalate, a known Trx1 inhibitor, was included as a positive control. The results obtained showed that C55 did not inhibit Trx1 activity (Figure S3). These data support the idea that C55 specifically acts on the TrxR1 reductase and probably on its C-terminal Sec-containing motif.

Figure 4.

Inhibition of TrxR1 activity by C55 in vitro. A) rat liver TrxR1 and B) 10 µg of protein samples extracted from human HCC (HepG2, Huh7) or BCa (MCF-7, MDA-MB-231) cells were incubated with the indicated concentrations of C55 and then TrxR1 activity was assayed as described in material and methods. Kit inhibitor was used as positive control inhibitor for TrxR1. Values are expressed as percentage of activity compared to solvent alone ± SD of at least two separate experiments.

Effect of C55 on cell viability of HCC and BCa cell lines

Next, we examined the effect of C55 on the cell viability of HCC (HepG2, Huh7) and BCa (MCF-7, MDA-MB-231) cell lines by MTS assay. As shown in Figure 5, C55 reduced cell viability in a dose-and time-dependent manner in all cell lines. To further support the involvement of the thioredoxin system in cell viability, we evaluated the effect of Auranofin, a known TrxR1 inhibitor, on the same cell lines. As shown in Figure S4, Auranofin treatment led to a marked reduction in cell viability, consistent with the proposed role of TrxR1 inhibition in mediating the cytotoxic effects observed with C55. To assess the reversibility of TrxR1 inhibition, a washout viability assay was performed: HepG2 and MDA-MB-231 cells were incubated with the inhibitor C55 for 24 h, followed by medium replacement with fresh compound-free medium and further incubated for 48 h (Figure S5). A recovery in cell viability was observed after the washout, indicating that the inhibitory effect of C55 is reversible. In contrast to C55 treatment, Auranofin-treated cells did not recover viability upon washout, consistent with its irreversible mode of action. These results are shown in Figure S5.

Figure 5.

Effects of C55 treatment on cell viability in HCC (HepG2, Huh7) and BCa (MCF-7, MDA-MB-231) cell lines. Cells were treated with different concentrations of C55 for 72 (A) or 96 (B) hours. Cell viability was determined by using MTS assay.

However, different cell lines show different sensitivity to the compound, with MCF-7 cells showing the lowest sensitivity, as shown by the IC₅₀ values in Table 2.

Table 2.

IC₅₀ values of compound C55 against HepG2, Huh7, MCF-7, and MDA-MB-231 cells after 72 and 96h of treatment.

hours	IC50 (µM)
	HepG2	Huh7	MCF-7	MDA-MB-231
72	29.4 ± 5.3	29.9 ± 3.6	> 100	42.7 ± 1.6
96	9.3 ± 3.7	18.9 ± 8.7	> 100	29.5 ± 10.6

Effects of C55 on oxidative stress signalling

Inhibition of TrxR1 increases intracellular oxidative stress, triggering activation of the of antioxidant genes, such as NAD(P)H:quinone oxidoreductase 1 (NQO1) and haem oxygenase-1 (HO-1)^53,54. As shown in Figure 6(A), treatment with C55 for 48 h significantly upregulated the expression levels of these two genes in HepG2 and MDA-MB-231 cells, as analysed by real-time PCR.

Figure 6.

Effects of C55 on oxidative stress signalling. A) Gene expression analysis by real-time PCR in HepG2 and MDA-MB-231 cells treated for 48 h with the indicated concentrations of C55 and Auranofin (AUR). *p < 0.05, **p < 0.01, C55 or AUR versus control vehicle alone B) ROS generation in HepG2 and MDA-MB-231 cells treated with C55. Cells were untreated or treated with C55 or AUR at the indicated concentrations for 48 h, and intracellular ROS levels were evaluated using H₂DCFDA as a probe. Data are expressed as the percentage of control cells and are the means ± SD of two separate experiments, each performed in triplicate. * p < 0.05, C55 or AUR versus control vehicle alone. C) Western blotting analysis of γ-H2AX in HepG2 and MDA-MB-231 cells treated for 48 h with the indicated concentrations of C55 and AUR.

C55 treatments also increased intracellular reactive oxygen species (ROS) production, as measured by using the cell-permeable fluorescent probe H₂DCFDA (Figure 6B). Furthermore, since high levels of ROS are known to produce DNA damage, we next evaluated the effect of C55 on the expression levels of phospho-H2AX (γ-H2AX) histone, a marker of DNA damage⁵⁵ (Figure 6(C)). Treatments of HepG2 and MDA-MB-231 cells with C55 for 48 h increased γ-H2AX expression levels, indicating genotoxic stress. Auranofin produced similar effects (Figures 6(B,C)). These findings support that C55 acts as a TrxR1 inhibitor, promoting oxidative stress and DNA damage in cancer cells.

A new binding mode for TrxR1 inhibition

The experimental data, taken together, put forward compound C55 as a SM acting as non-covalent TrxR inhibitor. It is worth noting, in fact, that such a compound lacks groups that can covalently interact with selenocysteine, is metal-free, and is characterised by a novel chemotype.

Furthermore, the binding mode returned by GOLD (Figure 7), does not involve covalent interactions. Nonetheless, the (S,S) configuration of C55 shows a promising docking score (ChemScore of 36.16 KJ/mol) supported by a favourable estimated binding energy of −37.55 KJ/mol.

Figure 7.

Top-scored docking pose of C55 within the TrxR1 binding site (PDB code: 3EAN)³⁹.

The hypothesis returned by the pocket prediction phase is here confirmed, as the experimentally validated molecule is predicted to interact with the hypothesised residues, namely G499 from chain A and Y116 from chain B, with the addition of the selenium-containing residue U498, validating all the previously proposed hypotheses and in agreement with previous knowledge⁵⁶. In summary, the following clues can be derived based on the discussed data: (i) the pocket prediction algorithm correctly predicted the location of the cavity to be considered; (ii) the selenium-containing residue plays a relevant role in the binding mode, establishing a hydrogen bond involving its backbone and confirming its already known role in the inhibition of the enzyme⁵⁷; (iii) the virtual screening procedure identified 63 compounds, among which compound C55 showed an inhibitory activity with an IC₅₀ of 4.7 ± 1.3 µM; (iv) the proposed compound is devoid of pharmacokinetic hindrances by hypothesis, as it belongs to a commercial library comprising only compounds devoid of potential ADMET problems⁴⁹. This compound, to the best of our knowledge, is the first metal-free compound able to bind to TrxR with a non-covalent interaction, making it a significant starting point for further studies aimed at its optimisation and, hopefully, at the identification of unexplored molecular series capable of efficiently inhibiting such a promising anticancer target. Noteworthy, the identified initial hit has never been studied as a bioactive molecule, as evident after interrogating ChEMBL^58,59 and PubChem⁶⁰ repositories. This result is even more significant considering that the compound has already undergone experimental ADMET tests, which confirmed the absence of pharmacokinetic issues.

Conclusions

This study led to the identification of an interesting non-covalent inhibitor of Thioredoxin reductase 1 (TrxR1), a selenoprotein enzyme crucial for maintaining redox homeostasis in mammals. Through an integrated experimental and computational approach, the compound C55 was identified as a promising candidate, demonstrating dose-dependent inhibition of TrxR1 with an IC₅₀ value in the micromolar range. The identified compound, already proved to be free of ADMET-related issues, exhibited inhibitory activity across various cancer cell lines, including HepG2, Huh7, MCF-7, and MDA-MB-231. As the first metal-free organic molecule to act as a non-covalent inhibitor of TrxR1, C55 represents a significant step forward, providing a strong foundation for hit-to-lead optimisation and paving the way for the development of potent anticancer drug candidates.

Funding Statement

This work was supported grant by the Italian National Research Council (CNR) “The Bioinorganic Drugs (BIDs) joint laboratory: A multidisciplinary platform promoting new molecular targets for drug discovery” is acknowledged, and by the research project “Potentiating the Italian Capacity for Structural Biology Services in Instruct Eric” (Acronym: ITACA.SB, project n° IR0000009) within the call MUR D.D. 0003264 dated 28/12/2021 PNRR M4/C2/L3.1.1, funded by the European Union NextGenerationEU.

Author contributions statement

CRediT: Giuseppe Lamanna: Data curation, Formal analysis, Methodology, Writing – original draft; Giuseppa Augello: Data curation, Formal analysis, Writing – original draft; Luisa Ronga: Formal analysis, Writing – review & editing; Diego Tesauro: Formal analysis, Writing – review & editing; Ilaria Silvestri: Formal analysis, Writing – review & editing; Antonina Azzolina: Formal analysis, Methodology, Writing – original draft; Cervello Melchiorre: Formal analysis, Investigation, Methodology, Writing – original draft; Giuseppe Felice Mangiatordi: Conceptualization, Data curation, Formal analysis, Methodology, Supervision, Writing – original draft, Writing – review & editing; Michele Saviano: Funding acquisition, Supervision, Writing – review & editing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author/s.