To discover novel potential mTOR kinase inhibitors, virtual high-throughput screening of the ZINC database and fragment-based design approaches have been utilized. Selected ligands were docked into the ATP binding site of mTOR, according to the earlier validated procedure (Stary et al., 2022). Before docking, compounds were prepared with LigPrep module in pH 7.4 ± 0.2 using Epic function (Schrödinger Release 2021-3; Schrödinger, LLC: New York, NY, 2021). The crystal structure of mTOR was downloaded from the PDB database (PDB code: 4JT6). Magnesium ions were added to the active site. Protein was prepared with Protein Preparation Wizard from Schrödinger Suite. Hydrogen atoms were added, and protein was ionized at pH 7.4 ± 0.2. The binding site was defined with Grid Generation module: the grid for ligand docking was set as an internal box with an edge length of 10 Å and a center corresponding to the co-crystallized ligand (PI-103). For each compound, ten poses were obtained. Analogically, compounds proposed as potential rapamycin binding site inhibitors were docked to the FRB domain (PDB code: 4DRJ). The binding site was defined based on the rapamycin position. The development of potential FRB domain ligands was based on the virtual high-throughput screening of the ZINC and Natural Product database. Obtained results were analyzed in PyMOL and Maestro. Physicochemical properties were predicted with the QikProp module from Schrodinger Suite. Based on the scoring function values, binding modes and predicted physicochemical properties of proposed ligands were sorted and analyzed. Finally, seven ligands have been selected for detailed examination and biological evaluation.

All-atom systems and input files for molecular dynamics simulations were prepared with CHARMM-GUI online server (Jo et al., 2008; Brooks et al., 2009). The protein-ligand complex was solvated with TIP3P water molecules. Sodium and chloride ions were added (0.15 M NaCl). Equilibration was performed via a one-step protocol implemented in CHARMM-GUI. Molecular dynamics simulation was carried out with NAMD 2.13 at 303.15 K. Timestep was set as 2 fs and the total time was 10 ns. Force field CHARMM36m was used for calculations. The results were analyzed in the VMD program.

ACDLAbs PhysChem Suite 14.0 software (Advanced Chemistry Development, Inc., Toronto, ON, Canada) was used for the calculation of Mw, tPSA, and HBD values. The CNS MPO score calculation method was described by Wager et al. (2016).

Before the initiation of the intended experiments, all solvents were mutually saturated at 21°C. 10 nM sodium phosphate buffer (pH 7.4) was used for logD determination. For saturation, large stock bottles containing n-octanol and 12% of water/buffer or water/buffer and 12% of n-octanol were shaken for 30 mins on a mechanical shaker and subsequently left to stand on the bench overnight until the phases separated. Approximately 0.50 mg of tested substance was added to the test vessels alongside with 1:1 volume ratio of saturated solvents (250 μL). The test vessels were shaken on the mechanical shaker (Multi Reax, Heidolph Instruments, Schwabach, Germany) for 30 min at the highest speed and then centrifuged for phase separation (16.873 × g for 5 min; Eppendorf 5418 Microcentrifuge, Prague, Czech Republic). Both organic and aqueous phases were sampled for LC-MS analysis. UHPLC-DAD-MS (UHPLC Infinity II 1290, detection by DAD and QqQ 6470; Agilent Technologies, CA, USA; separation on Arion C18 polar column 100 × 2.1 mm, 2.1 μm; Chromservis, Czech Republic) was used for determination of the exact concentration of tested substances. The total quantity of substance (in both phases) was used for logP/logD values calculation.

The spectrophotometric titration was used for dissociation constants’ determination, utilizing spectrophotometer Agilent Cary-60 (Agilent Technologies, CA, USA), pH meter WTW InoLab_IDS Multi 9430 (WTW, Czech Republic), and pH electrode SenTix^® Mic (WTW, Czech Republic). Data were analyzed by GraphPadPrism 8.0 software (GraphPad Software, CA, USA). The analyzed samples were prepared by adding 20 μL of compound stock solution (2 mg/mL) and 20 μL of 0.1 M hydrochloric acid into the cuvette with ultrapure water (total volume 3.5 mL). The samples were titrated by 0.1 M sodium hydroxide at 20°C while acquiring absorption spectra (190–500 nm) at different pH values. Absorbance at the chosen wavelength was plotted as a function of pH. The inflection point in the given plot was used for pKa value determination.

Compounds’ solubility was measured by the NEPHELOstar microplate instrument (BMG Labtech, Offenburg, Germany). Stock solutions were prepared by dissolving the tested compounds in ultrapure water or sodium phosphate buffer (pH 7.4) to obtain a concentration of 160 and 1,000 μg/mL. Each stock solution was sonicated at full power by Hielscher UP100H needle ultrasonic processor (Teltow, Germany) before loading into NEPHELOstar instrument injector A. Ultrapure water or sodium phosphate buffer was loaded into injector B. Compound was diluted in 48 wells in 0–160 or 0–1,000 μg/mL range and subsequently shook for 30 s. Each well was scanned with 80% laser power in matrix 3 × 3, beam width 2 mm, and obtained data were evaluated in GraphPad Prism 7.03 (GraphPad Software, San Diego, CA, USA) using segmental linear regression.

All cell lines used in the study were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). Human telomerase-immortalized retinal pigment epithelial cells RPE-1, glioblastoma U-87 MG, glioblastoma U373MG, glioblastoma T98G, glioblastoma A-172, breast adenocarcinoma MDA-MB-231, prostate carcinoma DU-145, and non-small cell lung cancer NCI-H1299 cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Thermo Fisher Scientific, Waltham, MA, USA) with high glucose (4.5 g/L). Primary human skin BJ fibroblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Thermo Fisher Scientific, Waltham, MA, USA) with low glucose (1 g/L). Both cell culture media were supplemented with 10% fetal bovine serum (FBS, Gibco/Thermo Fisher Scientific) and non-essential amino acids (NEAA), 100 units/mL of penicillin, and 100 μg/mL of streptomycin (Sigma-Aldrich). Cells were kept at 37°C under a 5% CO₂ atmosphere and 95% humidity.

For immunoblotting, the following primary and secondary antibodies were used: rabbit polyclonal p70 S6 kinase (#9202), rabbit monoclonal phospho-p70 S6 kinase (Thr389) (#9234), rabbit monoclonal AKT kinase (#4691), rabbit monoclonal phospho-AKT kinase (Ser473) (#9234) obtained from Cell Signaling Technology, Inc., Danvers MA, USA; p21waf1 (sc-56335) obtained from Santa Cruz Biotechnology, Inc., Dallas, TX, USA and mouse monoclonal anti-GAPDH (GTX3066) purchased from GeneTex, Inc., Irvine, CA, USA; goat anti-mouse (170-6516) or anti-rabbit IgG (H + L)-HRP conjugate (170-6515) purchased from BioRad, Hercules, CA, USA.

To obtain ionizing radiation (IR)-induced senescent cells, proliferating RPE-1 cells were exposed to a single dose of X-rays (20 Gy) using Pantak HF160 (Gulmay, Surrey, UK) X-ray instrument equipped with Pantak Seifert HF320 generator, MXR-161 X-ray tube (Comet AG, Flamatt, Switzerland), and an aluminum filter using current 1–10 mA. After irradiation, the cells were cultured for an additional 7 days until the development of the senescent phenotype (Salovska et al., 2022). To induce docetaxel-induced senescence (DIS), BJ cells were exposed to 5 nM docetaxel for 7 days, as described previously (Sapega et al., 2018). To prepare temozolomide (TMZ)-induced senescence, U87 cells were exposed to 100 μM TMZ for 14 days.

Cell proliferation was followed by the crystal violet (Feoktistova et al., 2016), resazurin assays (Anoopkumar-Dukie et al., 2005), and time-lapse microscopy.

Twenty-four hours before treatment, cells were plated in 96-well plates at densities of 2,850 cells per well for senescent RPE-1 (IR-RPE-1), 3,350 cells per well for proliferating RPE-1 and senescent BJ (DIS-BJ), and 5,000 cells per well for U373, T98, A172, U87, TMZ-U87, and proliferating BJ cells.

To determine the secretion of selected inflammatory cytokines into culture media, cell cultures were exposed to tested compounds (50 μM) for 24 h. Then the culture medium was removed, cells were washed three times with 1 mL PBS, and a fresh medium without FBS was added for 24 h. Detection of cytokines in conditioned culture media was done by Human Inflammation 11-Plex Panel: IFNγ, IL-1α, IL-1β, IL-6, IL-8, IL-10, IL-12p70, IL-27, IP-10, MCP-1, and TNFα (AimPlex Biosciences, Inc.) using a BD FACSVerseTM flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA) according to manufacturer’s protocol. Data were analyzed using FlowJo 10 software (FlowJo LLC, Ashland, OR, USA).

Cells were harvested 1 and 5 h after treatment. Cells were washed with PBS, harvested into SDS sample lysis buffer (SLB; 2% SDS, 63 mM Tris–HCl, pH 6.8, and 10% glycerol), sonicated, and centrifuged. Protein concentration was determined by BCA assay (Pierce Biotechnology, Rockford, IL, USA); samples were adjusted to equal protein amount (40 μg) with SLB containing dithiothreitol (1% final conc.) and bromophenol blue (0.02% final conc.), and separated by SDS–PAGE in Tris-glycine-SDS buffer (BioRad, 1610772). Proteins were electrotransferred onto a nitrocellulose membrane (0.45 μm NC, Amersham™, GE Healthcare Life Sciences) using wet transfer in Tris-glycine buffer (BioRad, 1810704) with 10% methanol (Sigma, 59304). After blocking with 5% non-fat milk in PBS/Tween-20, proteins were detected using specific antibodies and horseradish peroxidase (HRP)-conjugated secondary antibodies. Peroxidase activity was detected by ECL detection reagents (Thermo Fisher Scientific, Waltham, MA, USA).

The mTOR enzyme assay was used to determine the mTOR kinase activity of compound 3. Enzymatic reaction mixture was prepared as follows: mTOR enzyme (10 μg/mL, Invitrogen), ATP (10 μM), p70S6K peptide (10 μg/mL, SignalChem), and kinase buffer (50 mM HEPES, pH 7.5, 1 mM EGTA, 3 mM MnCl₂, 10 mM MgCl₂, 2 mM DTT, and 0.01% Tween-20). The reaction was performed in black half-area 96-wells at room temperature for 1 h, then stopped by adding ADP detection buffer (Transcreener ADP² FI Assay, BellBrook labs) and incubated at room temperature for 1 h. Compound concentration was 10, 20, 40, 80, and 100 μM. The final DMSO concentration was 1%. PHERAstar microplate reader was used for measuring the fluorescence intensity (excitation 575 nm, emission 620 nm).

C57BL/6 (B6) male mice, 8–12 weeks old, were obtained from the Institute of Molecular Genetics, Prague, Czech Republic. Experimental protocols were approved by the Institutional Animal Care Committee of the Institute of Molecular Genetics, Prague.

To analyze the tolerance of the new compound 3 and torkinib (as a control), the compounds were perorally administered at increasing doses (three times on days 0, 3, and 5). Compounds were dissolved in 20 μL DMSO and completed into 200 μL with corn oil.

During the experiments, mice behavior and possible apparent signs of toxicity were monitored daily. The body weights were determined on days 0, 4, and 8. On day 8, blood (25 μL) was withdrawn using a capillary pipette containing anticoagulant (heparin) from the retroorbital sinus into EDTA-containing tubes. Basic hematological parameters were measured using the Mindray BC-5300 Vet. Mice were then killed, autopsied, and abdominal organs (spleen, liver, and kidney) were removed for visual examination and imaging by a digital camera. Single-cell suspensions from the spleens were prepared by homogenization through a cell strainer (70 μm; BD Biosciences, San Jose, CA, USA). Erythrocytes were osmotically lysed using ammonium chloride-potassium lysis buffer, and the cell suspensions were washed three times in the RPMI-1640 medium and used for further analysis by flow cytometry for selected cell populations.

The cellularity of selected spleen cell populations was analyzed by flow cytometry. Cell suspensions were washed and preincubated with anti-CD16/CD32 (2.4G2) antibody to minimize non-specific binding at 4°C for 15 min following the washing step and incubation with labeled primary antibody at 4°C for 30 min. Relevant isotype controls of irrelevant specificity were used. FACS buffer (PBS, 1% FBS, 0.1% NaN₃) was used for all washing steps and analysis. The following antibodies were used for FACS analyses: BV421 anti-mouse CD45 (30-F11), PE anti-mouse CD8α (53-6.7), BV711 anti-mouse CD4 (RM4-5), APC anti-mouse Ly-6g/Ly-6C (Gr-1) (RB6-8C5), BV711 anti-mouse CD11b (M1/70). FACS analysis was performed using a flow cytometer Symphony (BD Biosciences) and analyzed using FlowJo 10 software (FlowJo LLC, Ashland, OR, USA).

Spleen, kidney, and liver tissue samples from the control and treated animals were fixed in 4% formaldehyde and embedded into paraffin. The analysis and evaluation were performed at the Czech Centre for Phenogenomics, Institute of Molecular Genetics of the CAS, Vestec, using Leica ASP6025 automatic vacuum tissue processor for tissue processing, Leica EG1150 H + C embedding station for embedding, and Leica RM2255 rotary microtome for sectioning (2 μm sections). Standard haematoxylin and eosin (HE) descriptive histopathological staining was done using Leica ST5020 + CV5030 stainer and coverslipped.

Graphs were generated using GraphPad Prism 5.04 (GraphPad Software, La Jolla, CA, USA). The data are expressed as the mean ± S.D. from three independent experiments. Statistical differences for two groups were analyzed by Student’s t-test, ^****p < 0.0001; ^***p < 0.001; ^**p < 0.01; *p < 0.05. Quantitative analysis of immunoblots was done by Image J 1.48v program with GAPDH as the internal control.

For the given computational study, two different methods have been utilized – fragment-based design and virtual high-throughput screening. Compounds were docked to the ATP binding site of mTOR (PDB code: 4JT6), according to the fully validated procedure (Stary et al., 2022). Further, their scoring function values, binding modes, and physicochemical properties were analyzed. Fragment-based design approach allowed us to obtain promising ligands, of which compound 1 (a benzimidazole derivative), its salt form – compound 2, and ligand 3 with azaindole fragment (Table 1) have been selected for biological evaluation. Within virtual screening of a part of the ZINC database, a series of small-molecule ligands with interesting binding modes in ATP catalytic cleft has been discovered, of which compound 4 has been selected for in vitro evaluation (Table 1).

Docking results showed that compound 1 exerts the following binding mode: benzimidazole ring occupied the same place as the adenine ring of ATP concurrently forming π–π stacking interaction with the side chain of Tyr2225 (Figure 2, left panel). Amine linker of compound 1 coupled with the hydroxy group of Tyr2225 by a hydrogen bond. The ionized amine group of the piperidine ring interacted with polar, negatively charged amino acids, such as Glu2190 and Asp2195 (Figure 2, left panel). Additionally, compound 1 demonstrated a less favorable score value than reference compound torkinib (−6.234 and −8.137, respectively).

Among presented compounds, ligand 3 obtained a promising binding mode as well as a docking score value (−9.580), which was more beneficial compared to that of torkinib. The top-ranked pose of compound 3 within the mTOR kinase active site is presented in Figure 2 (right panel). Herein, the azaindole fragment mimicked the adenine ring of ATP. Compared with ligand 1, compound 3 interacted with the main chain of Val2240 via hydrogen bonds, thus significantly stabilizing the ligand-protein complex. Mentioned heterocyclic moiety was able to interact with Trp2239 by π–π stacking interaction as well as with Tyr2225 via weak aromatic ring interaction. Benzamide scaffold of compound 3 was docked into a site of ribose- and phosphate-moieties binding. The 10 ns molecular dynamics (MD) simulation showed that the overall binding mode of ligand 3 was stable. Furthermore, it was observed that hydrogen bonds between azaindole moiety and Val2240 were also present in the course of the MD run. Initially, a hydrogen bond between the Val2240 oxygen atom and azaindole nitrogen atom was broken, however, this interaction was rapidly renewed (Supplementary Figure 1). Most changes in the position of compound 3 were observed for the benzamide fragment. Such finding suggests that interactions with Val2240 are crucial for binding ligand 3 into the active site of mTOR.

In silico studies revealed that ligand 4 interacted with the ATP binding site in the site of phosphate group binding. Compound 4 obtained a beneficial score: −9.068. In addition, it formed salt bridges with Glu2190, Asp2357, and ionic interactions with magnesium ions as well (not shown).

Searching for potential mTOR inhibitors acting by allosteric modulation was carried out by means of virtual high-throughput screening of ZINC and natural product database (NPDB). For biological evaluation, three potential allosteric inhibitors have been selected: compounds 5 and 6, which present a less complex structure than rapamycin, and epicatechin gallate (7) from NPDB.

Compounds 5 and 6 were located in the pocket, occupied by the triene fragment of rapamycin in its complex with mTOR, and formed specific interactions with the FRB domain of the enzyme (Figure 3; compound 5–left panel; compound 6–right panel). The methoxy group at benzylimidazolidine scaffold of compound 5 was noticed to form hydrogen bond with Ser2035. On the other hand, phenyl group was able to form aromatic interactions with the side chain of Tyr2105. Additionally, imino-imidazolidine fragment of compound 5 interacted with Asp2102 residue via the salt bridge.

The binding mode of biphenyl derivative 6 was stabilized through π–π interactions with Trp2101, Tyr2105, and Phe2039. Amide substituent formed the hydrogen bond with Asp2102 residue and water molecule 2226.

Epicatechin gallate (7) was selected from the Natural Product Database by virtual high-throughput screening. It has been observed that mentioned ligand occupied the rapamycin binding site and interacted with a water molecule and Asp2101 via hydrogen bonds. Additionally, Phe2039 formed an aromatic ring interaction with dihydroxyphenyl moiety (not shown).

To determine the effect of novel compounds on cell proliferation, normal human skin fibroblasts (BJ), non-transformed telomerase-immortalized retinal pigment epithelial cells (RPE-1), and cancerous human glioblastoma cell line U373 were exposed to different concentrations of compounds 1–7 (0, 10, 50, 80, and 100 μM) for 24 h using torkinib as a reference compound. We have decided to use only torkinib as a representative of small-molecule mTOR inhibitors for all biological experiments. Pilot tests employing crystal violet determining the residual number of attached cells (Figures 4A,C,E) and resazurin assays measuring the metabolism of cell population (Figures 4B,D,F) revealed that only compound 3 exerted a significant suppressive effect on the readout of both assays with comparable potency to torkinib in all cell types tested. Other compounds (1, 2, 4, 6, and 7) showed mild (although sometimes statistically significant) suppressive effect in dependence on the cell type used. Compound 5 showed no effect.

Next, we followed the cells exposed to compounds 1–7 and torkinib by time-lapse microscopy for 72 h (with a 1-h interval between captures) using the Incucyte SX1 platform. In concert with crystal violet and resazurin assays, only compound 3 (0, 10, 50, 80, and 100 μM) showed the effect on RPE-1 cell proliferation (Supplementary Figure 2A). Compound 7 (epicatechin gallate) showed a cytostatic effect at higher concentrations (80–100 μM). Similar effect was observed for U373 cells (Supplementary Figure 2B). Again, compound 7 showed cytostatic effect at higher concentration (100 μM).

Based on these screens, we broadened the tests of the most effective compound 3 using other glioblastoma cell lines, T98, A172, and U87. Like U373 cells, all tested glioblastoma cells, including T98 cell line characterized by higher cell death resistance compared to other glioblastoma cell lines, were significantly affected by compound 3 (see Figures 5A–F for crystal violet and resazurin assays). Further, in proliferating U87 cells, time-lapse microscopy showed that compound 3 (100 μM) caused rapid loss of cell spreading with a lack of mitotic activity and progressively occurring cell death. A similar effect was observed for torkinib (100 μM; Supplementary Figure 2C).

The induction of apoptosis by compound 3 was confirmed by annexin V staining (Figure 5G).

To conclude, compound 3, compared to the rest of the novel series of substances, showed a significant cytotoxic effect on normal and cancerous cells, including a series of glioblastoma cell lines with similar potency to reference compound torkinib.

Senescent cells are, in general, more resistant to apoptosis, which can be restored by mTOR inhibition (Laberge et al., 2015). To examine the cytotoxicity of novel compounds against non-transformed senescent cells (i.e., senolytic effect), ionizing radiation (IR)-induced and drug-induced modes of cellular senescence in RPE-1 and BJ cells, respectively, were employed (see “Materials and methods” for details). The development of senescence was confirmed by routinely used assays (an increase of senescence-associated beta-galactosidase activity, expression of cyclin-dependent kinase inhibitor p21waf1, and halt of DNA replication by EdU-incorporation assay; see Supplementary Figures 3A–G). Both IR-induced senescent RPE-1 and docetaxel (DTX)-induced senescent BJ cells were significantly sensitive to compound 3 (Figures 6A–D), however, in a non-selective manner (i.e., they were less sensitive to compound 3 than proliferating counterparts) as detected by crystal violet and resazurin assays (Supplementary Figures 4A,B). In addition, a mild but selective effect on crystal violet readout of γ-irradiated RPE-1 cells was observed for compound 5 (Supplementary Figure 4B). Reference compound torkinib exerted the non-selective effect in crystal violet and resazurin assays.

To evaluate the effect of compound 3 on the senescent cell phenotype of non-transformed cells, time-lapse microscopy of IR-senescent RPE-1 cells exposed to compound 3 (100 μM) and torkinib (100 μM) was performed for 72 h. Torkinib caused in IR-RPE-1 senescent cells rapid onset of vacuolization (after approximately 1 h of exposure) with partial loss of adhesion (decreased cells covered area) but only with a few senescent cells dying (Figure 7A). Compound 3 generated a similar phenotypic manifestation in IR-RPE-1 senescent cells as torkinib except for vacuolization. These findings indicate that both compounds induced a rather senomorphic than senolytic effect in IR-RPE-1 cells.

As cancer cells can undergo a senescent-like phenotype when exposed to subapoptotic doses of chemotherapeutics, we tested the response of compound 3 on glioblastoma cell line U87 induced to senescence by chemotherapeutics temozolomide (TMZ, 100 μM) used in the therapy of glioblastoma in clinics (Karachi et al., 2018). As shown in Figures 6E,F, compound 3 affected the readout of crystal violet and resazurin assays similarly to proliferating U87 cells. Moreover, time-lapse microscopy revealed that compound 3 induced the loss of cell adhesion followed by cell death as detected by annexin V staining (Figure 7B). Note that torkinib exerted the same effect on TMZ-U87 cells.

Together, compound 3 and reference compound torkinib affected the senescent phenotype of non-transformed cells, apparently without the induction of extensive cell death. Contrarily, both compounds induced massive cell death of temozolomide-induced glioblastoma senescent-like cells. This type of selectivity between normal and cancerous senescent cells can be desirable in adjuvant therapy regimes of cancer-related senescence induced by radiochemotherapy, as only senescent cells originating from cancer but not normal cells will be removed.

To examine whether novel compounds could inhibit mTOR kinase, BJ cells were exposed to novel compounds 1–7 (100 μM) for 1 and 5 h, and the activity of mTORC1 was detected by determining specific phosphorylation levels (threonine 389) of its substrate p70 S6 kinase (S6K), by immunoblotting with specific antibodies. As shown in Figure 8, the inhibitory effect on Thr389 S6K phosphorylation was detected only for compound 3 and torkinib. The rest of the drugs showed mild or no effect on Thr389 S6K phosphorylation. Note that the suppression of S6K phosphorylation was associated with increased electrophoretic protein mobility of S6K protein.

Thus, in correlation with the cytotoxic effect, compound 3 inhibits phosphorylation of mTORC1 substrate p70 S6 kinase indicating its inhibitory effect on mTORC1 kinase.

Due to the highest cytotoxicity and inhibitory effect on mTORC1 kinase activity, compound 3 was selected for further testing. To corroborate whether compound 3 is a direct inhibitor of mTOR kinase, we performed in vitro mTOR kinase enzyme assay (see “Materials and methods” section). As shown in Figure 9A, compound 3 inhibited mTOR kinase activity from the concentration of 10 μM with complete inhibition at 100 μM with IC₅₀ value approximately 70 μM.

To further validate the inhibition of mTOR activity by compound 3, we examined the inhibitory effect of compound 3 on S6K phosphorylation in the concentration range 0–100 μM in proliferating BJ cells by immunoblotting. The apparent effect of compound 3 on inhibition of S6K phosphorylation was revealed from the concentration 6.25 μM after 5 h-long exposure (Figure 9B).

Additionally, we tested the inhibitory effect of compound 3 on S6K phosphorylation in a series of human non-transformed and cancer-derived cell lines. For this purpose, RPE-1, glioblastoma U87, U373, T98, and A172, breast adenocarcinoma MDA-MB-231, lung cancer H1299, and prostate cancer DU-145 cell lines were exposed to compound 3 (100 μM) for 1 and 5 h and the level of mTORC1 substrate phospho-p70 S6K was assayed by immunoblotting. Notably, as shown in Figure 10, the most potent inhibitory effect of compound 3 was detected in glioblastoma U87 cells after 1 h-long exposure maintaining a reasonable level also after 5 h. Marked but transient inhibition of Thr389 S6K phosphorylation was also observed in T98, MDA-MB-231, and H1299 cells, whereas no effect was detected in RPE-1 and DU-145 cells, indicating cell type-dependent (in)sensitivity to mTORC1 inhibition by compound 3.

Furthermore, we tested the mTORC1 inhibitory activity of compound 3 in senescent cells. For this purpose, docetaxel-induced BJ, ionizing radiation-induced RPE-1, and TMZ-induced U87 senescent cells were exposed to compound 3 (100 μM) for 1 and 5 h, and the level of mTORC1 substrate phospho-p70 S6K was assayed by immunoblotting as above. The robust effect on p70 S6K phosphorylation was detected in docetaxel-induced BJ cells, whereas a less pronounced effect was revealed in temozolomide-induced U87 and ionizing radiation-induced RPE-1 cell lines.

In conclusion, compound 3 inhibitory effect on mTORC1 activity is cell type-dependent with inhibitory activity exerted in glioblastoma, breast, and lung cancer cell lines. From a series of tested tumor cells, the absence of inhibitory activity was observed only in prostate cancer cell line DU-145. Inhibitory activity of compound 3 in senescent cells was comparable to proliferating cells, only the impact on inhibition of mTORC1 activity in IR-induced senescent RPE-1 cells was more apparent than in proliferating cells.

It is generally known that only mTORC1 is acutely sensitive to rapamycin, whereas mTORC2 is inhibited just after chronic administration of the mentioned drug. More recently, it has been revealed that particularly reduced activity of mTROC2 is responsible for the adverse effects of rapamycin treatment, such as glucose intolerance or hepatic insulin resistance (Lamming, 2014). On the other hand, dual inhibition of both mTORC1 and mTORC2 complexes is proposed as a more effective treatment of cancer cells.

For this reason, we decided to test the inhibitory effect of compound 3 on phosphorylation of mTORC2 substrate AKT using several human normal, immortalized, and cancer-derived cell lines. BJ, RPE-1, U87, U373, T98, A172, MDA-MB-231, H1299, and DU-145 cells were exposed to compound 3 (100 μM) for 1 and 5 h and the level of AKT serine 473 (Ser473) phosphorylation was assayed by immunoblotting. Transient inhibition (after 1 h) of Ser473 AKT phosphorylation was observed in BJ, RPE-1, U87, U373, T98, A172, MDA-MB-231, H1299, and DU-145 cells. However, in several cell types (BJ, RPE-1, MDA-MB-231, H1299, and DU-145), the AKT phosphorylation was higher after 5 h of exposure than in control cells (Figure 11). Notably, the most potent inhibitory effect of compound 3 was revealed in glioblastoma T98 cells after 1 h, maintaining inhibitory level also after 5 h of exposure. Overall, compound 3 harbors the transient inhibitory effect on mTORC2 activity in the cell type-dependent manner.

Progressive accumulation of senescent cells can deplete the organism of functional cells required for tissue repair and regeneration, such as stem and progenitor cells. Therefore, it seems logical to remove senescent cells to solve the problem. However, the studies in different murine models showed defects in wound healing when senescent cells were eliminated (Baker et al., 2011; Demaria et al., 2014). Thus, compounds modulating the senescent secretome seem to be the better alternative, since the production of factors secreted by senescent cells termed senescence-associated secretory phenotype (SASP) is considered a primary mediator of the detrimental effects of senescent cells (Coppé et al., 2008; Wang et al., 2017). Chronic, low-grade inflammation, characterized by elevated levels of circulating cytokines and increased immune infiltration associated with inflammaging, fuels loss of resilience, and increases the risk of age-related diseases (Franceschi and Campisi, 2014; Birch and Gil, 2020).

To evaluate the effect of compound 3 on a pro-inflammatory component of SASP, we measured the level of 11 pro-inflammatory cytokines IL-1α, IL-1β, IL-6, IL-8, IL-10, IL-12p70, IL-27, MCP-1, IFNγ, TNFα, and IP-10 in culture media conditioned by three models of senescent cells (DIS-BJ, IR-RPE-1, and TMZ-U87) before and after 24 h-long incubation of senescent cells with compound 3 and torkinib (50 μM). As shown in Figure 12, the impact of both drugs on the level of inflammatory cytokines was drug- and cell type-dependent. Importantly, compound 3 decreased IL-1α, IL-8, and MCP-1 consistently in all tested senescence models. The effect of compound 3 on IL-6 was variable. The apparent decrease of IL-6 was observed in DIS-BJ and TMZ-U87 senescent cells. Notably, both compound 3 and torkinib suppressed all detectable cytokines (i.e., IL-1α, IL-1β, IL-6, IL-8, IL-27, MCP-1, and TNFα) in TMZ-induced U87 senescent cells.

To investigate in vivo tolerability of new substances, compound 3 and torkinib were repeatedly (three times) perorally administered at increasing doses (5, 25, and 50 mg/kg body weight), three times on days 0, 3, and 5. Indeed, no apparent signs of the whole body or organ toxicities and no changes in mice behavior, as well as changes in the body weights were observed after the treatments (Figure 13).

Further, the blood count analysis revealed that although there were changes between samples, no abnormal changes were detected after administration of tested compounds (3 and torkinib). The data are presented in Table 4.

Spleen cells were analyzed by flow cytometry for possible changes in the percentages of important selected immune cell populations (CD45⁺, CD4⁺, CD8⁺, and CD11b⁺/Gr-1⁺; Table 5). No significant differences were found between samples from the control and treated groups.

Histopathological analysis of the spleen, kidney, and liver tissue samples revealed pathological changes in these organs compared to treated animals. In general, they were more pronounced in the samples from the torkinib-treated mice, as compared to the compound 3-treated mice.

Analysis of the compound 3 potential toxic effects demonstrated that the pathological manifestation in the spleen appeared to be almost identical in all three different doses, with varying degrees of extramedullary haematopoiesis (Supplementary Figure 5). However, the situation was completely different in the liver and kidneys. In the liver, there was apparent just minimal evidence of damage, although there were signs of incipient and discrete centrilobular-centrozonal necrosis of hepatocytes in a dose-dependent manner (Supplementary Figure 6). The most pronounced pathological changes were evident in the kidneys (cortex and medulla), at the glomerular and tubular levels – the latter being predominant (Figures 14, 15). Again, the parenchymal damage was directly proportional to the treatment dose.

In the samples from the torkinib-treated mice, the pathological changes were more pronounced, compared to those from the compound 3-treated mice, with the exception of the spleen, in which the same minimal to moderate signs of extramedullary haematopoiesis were observed. Again, the most affected organ was the kidney in the glomeruli and the tubular system. Here, in addition to the kidneys of the substance 3-treated mice, signs of fibrosis were already evident (Figure 16). The pathological changes were again proportional to the torkinib doses used for the treatments.