Inspired by Sea Urchins: Warburg Effect Mediated Selectivity of Novel Synthetic Non-Glycoside 1,4-Naphthoquinone-6S-Glucose Conjugates in Prostate Cancer

The phenomenon of high sugar consumption by tumor cells is known as Warburg effect. It results from a high glycolysis rate, used by tumors as preferred metabolic pathway even in aerobic conditions. Targeting the Warburg effect to specifically deliver sugar conjugated cytotoxic compounds into tumor cells is a promising approach to create new selective drugs. We designed, synthesized, and analyzed a library of novel 6-S-(1,4-naphthoquinone-2-yl)-d-glucose chimera molecules (SABs)—novel sugar conjugates of 1,4-naphthoquinone analogs of the sea urchin pigments spinochromes, which have previously shown anticancer properties. A sulfur linker (thioether bond) was used to prevent potential hydrolysis by human glycoside-unspecific enzymes. The synthesized compounds exhibited a Warburg effect mediated selectivity to human prostate cancer cells (including highly drug-resistant cell lines). Mitochondria were identified as a primary cellular target of SABs. The mechanism of action included mitochondria membrane permeabilization, followed by ROS upregulation and release of cytotoxic mitochondrial proteins (AIF and cytochrome C) to the cytoplasm, which led to the consequent caspase-9 and -3 activation, PARP cleavage, and apoptosis-like cell death. These results enable us to further clinically develop these compounds for effective Warburg effect targeting.


Introduction
Chemotherapy remains an important treatment component for the vast majority of cancer patients. However, a lack of selectivity for tumor cells over normal cells frequently results in insufficient drug concentrations in the malignant tissue, systemic toxicity, and the development of drug resistance. Therefore, inexorable efforts have been made to develop targeted therapies allowing the release of cytotoxic activity in the region of interest while sparing healthy tissue. Potential therapeutic targets include cell surface antigens, growth receptors, key regulators of cellular signal transduction, and DNA repair defects [1]. Antibody drug conjugates (ADCs) consisting of an antibody linked to a biological active payload are under intensive investigation and first drugs are available in clinics, e.g., gemtuzumab ozogamicin in acute myeloid leukemia or brentuximab vedotin in relapsed Hodgkin´s lymphoma [2]. In urothelial carcinoma, nectin-4-targeted enfortumab vedotin has been recently accepted for priority review by the FDA [3]. On a similar note, radioisotope conjugated targeting antibodies are developed for imaging and radioimmunotherapy strategies. In fact, PSMA-ligand therapy with PSMA-linked αor β-emitters are evaluated in clinical trials in prostate cancer and are applied to patients in every day s routine on individual bases [4]. Additional targeted therapies in prostate cancer include androgen-receptor (AR) targeting drugs such as abiraterone acetate, apalutamide, or enzalutamide as well as PARP-inhibitors in patients with DNA-repair defects [5].
Targeting the Warburg effect is an interesting novel treatment approach utilizing an increased sugar uptake of cancer cells to incorporate sugar conjugated cytotoxic compounds into tumor cells [6,7]. The high sugar demand of tumor cells results from the predominantly used metabolic pathway of glycolysis-even in aerobic conditions-which is energetically less efficient than oxidative phosphorylation found in healthy cells [8]. Thus, an overexpression of glycolytic enzymes in cytoplasm and glucose transporters (GLUTs) on the surface of cancer cells is frequently found in tumor cells in order to comply with the high sugar consumption [6,7].
Consequently, new compounds targeting the Warburg effect have been developed. In fact, sugar conjugates of docetaxel, busulfan, paclitaxel, and chlorambucil were found to be more selective to cancer cells in comparison with the non-conjugated "mother" molecules (reviewed in [6]). Furthermore, a glycosylated derivative of the alkylating agent ifosforamide-glufosfamide-has successfully passed clinical trials in human pancreatic, brain and CNS, lung cancer and soft tissue sarcoma [9].
More than 50% of the drugs currently available on the market are based on natural compounds and their synthetic derivatives [10]. Compared to compounds isolated from terrestrial sources, the marine-derived molecules are less studied [11,12]. At the same time, to date there are ten drugs approved for clinical use, seven of which are approved for the treatment of cancer and cancer-related conditions [13,14]. Spinochromes are an important family of polyhydroxynaphthoquinone secondary metabolites which were found in various sea urchin species. Due to their chemical diversity and promising pharmacological properties, these compounds attract the attention of scientific community as a base for development of new drugs [15,16]. Thus, spinochromes have revealed anti-allergic, anti-hypertensive, anti-diabetic, anti-oxidant, anti-inflammatory, and cardioprotective properties (reviewed in [15,16]). In general, biological effects of naphthoquinones are associated with generation of reactive oxygen species and modulation of redox signaling radical reactions [17] being either pro-or antioxidants or electrophiles [18]. Additionally, naphthazarins (5,8-dihydroxy-1,4-naphthoquinone derivatives), as well as other polymethoxylated natural and semisynthetic compounds possessing aromatic core, were reported to be potent anti-mitotic agents. Some of these compounds are capable of microtubule destabilizing and are therefore highly cytotoxic to cancer cells, representing an attractive starting point for further design as anticancer agents [19,20]. Thus, trimethyl ether of spinochrome D (tricrozarin B) inhibits a colony formation of human cancer HeLa S3 cells having IC 50 of 0.007 µg/mL [21]. Furthermore, our previous study also showed that methoxy derivatives of 1,4-naphthoquinones exhibit greater cytotoxicity compared to the corresponding unsubstituted hydroxy analogs [22].
In this study, in continuation of our research on sea urchin pigments [23][24][25], we synthesized and analyzed novel sugar conjugates of naturally derived and chemically modified hydroxyl-1,4-naphthoquinones related to the sea urchin spinochrome pigments. Recently, we identified several natural 1,4-naphthoquinones to be active in human drug-resistant prostate cancer cells [26] and were able to synthesize first conjugates with d-glucose possessing in vitro cytotoxic activity [22,26]. In the current study, we further modified the compounds bearing hydroxy-1,4-naphthoquinone scaffold and investigated their anticancer properties and the mechanism of action. Thus, to increase the selectivity of the identified natural 1,4-naphthoquinones via Warburg effect targeting, we conjugated these bioactive moieties with 6-mercaptoglucose. A glycoside bond is chemically reactive and may be easy degraded in the living system via enzyme-catalyzed hydrolysis. At the same time, thioglycosides have been reported to be more resistant to the enzyme-mediated degradation [27]. Therefore, we designed and synthesized a library of non-glycoside conjugates in order to increase stability of the target compounds under human body conditions; additionally, we introduced a novel sulfur linker (thioether bond) to prevent potential hydrolysis by the human glycoside-unspecific enzymes. It is important to note that an unsubstituted glycoside hydroxy group (at C1 position) is relevant for the stabilizing of the hydrogen bond interaction between glucose and GLUT-1 and therefore for successful uptake of the glucose conjugate via this system. In contrast, the conjugation of glucose at C6 position should have a minimal impact on the GLUT-1 mediated glucose uptake and therefore on the uptake of the synthesized compounds by the cancer cells. We were able to synthesize the new acetylated (protected) and non-acetylated (unprotected, containing free-glucose scaffold) thio-conjugates of 1,4-naphthoquinone and glucose. Human drug-resistant prostate cancer cells were chosen as the main model because of the known overexpression of GLUT-1. Here, we describe the synthesis of these new conjugates, as well as their Warburg effect-guided selective anticancer activity and mode of action.

Design and Synthesis of the 6-S-(1,4-Naphthoquinon-2-yl)-d-Glucose Chimera Molecules
In continuation of the research on synthesis of bioactive 1,4-naphthoquinones, capable of selective activity towards human drug-resistance prostate cancer cells, we designed the chimera molecules consisting of cytotoxic 1,4-naphthoquinone pharmacophore and 6-thioglucose moiety. These derivatives are expected to exhibit selective cytotoxicity to cancer cells due to Warburg effect targeting and to be more stable in human body in comparison with conventional 1,4-naphthoquinone-glucosides due to the non-glycoside bond and thioether nature of the linker.
Thus, two different synthetic approaches were used for conjugation of naphthoquinones with 6-mercaptoglucose. We applied either: (a) a substitution reaction of halogenoquinones with readily available tetra-O-acetyl-6-mercaptoglucose in a basic condition ( Figure 1A); or (b) an addition reaction of tetra-O-acetyl-6-mercaptoglucose to juglone, its acetate, or naphthazarin ( Figure 1B). This was followed by saponification of the synthesized acetylated conjugates with MeONa/MeOH ( Figure 1C).

Evaluation of Cytotoxicity and Selectivity
First, we evaluated cytotoxic activity of the synthesized compounds in human drug-resistant prostate cancer PC-3 cells and human prostate non-cancer PNT2 cells. It has been reported that GLUT-1 is a main and the most abundant expressed receptor responsible for the glucose uptake in mammalian cells [7]. The expression of GLUT-1 in PC-3 was found to be higher in comparison with PNT2 cells (Figure 2A), thus these lines were considered a suitable model for the screening of Warburg-effect targeting substances ( Figure 2B). Indeed, the selectivity of the conjugates bearing unprotected glucose residue (and therefore exhibiting increased affinity to GLUT-1) was significantly higher in comparison with acetylated derivatives ( Figure 2C). Thus, we were able to identify one conjugate, SAB-14, to have a selectivity index (SI) > 2 ( Figure 2B). For the further investigations, we selected SAB-14 and -13-the acetylated derivative revealing a~10-fold higher cytotoxicity and belonging to the same structural group as SAB-14. Both compounds were evaluated in five human prostate cancer cell lines known for different resistance profiles: LNCaP (AR-FL + and AR-V7 − , androgen-dependent, docetaxel-sensitive), 22Rv1 and VCaP (AR-FL + and AR-V7 + , androgen-independent, docetaxel-sensitive), and PC-3 and DU145 (AR-FL − and AR-V7 − , androgen-independent, docetaxel-resistant) [29,30]. In addition, five human non-cancer lines (PNT2, RWPE-1, HEK 293T, MRC-9, and HUVEC) were exposed to the derivates. In line with our previous results, both compounds were active and selective to human prostate cancer cells in comparison with non-cancer lines ( Figure 2E). Note that SAB-14, bearing unprotected glucose moiety ( Figure 2D), exhibited higher selectivity to the cancer cells ( Figure 2E).

Cytotoxicity Correlates with Glucose Uptake Rate in Prostate Cancer Cells
To examine whether synthesized compounds are able to target the Warburg effect, we accessed the glucose transporter-1 (GLUT-1) expression in the cells. Note that a mean GLUT-1 mRNA expression determined by qPCR was 1.6-fold higher in prostate cancer cells (LNCaP, 22Rv1, VCaP, PC-3, and DU145 cell lines) in comparison with human non-cancer cells (PNT2, RWPE-1, HEK 293T, MRC-9, and HUVEC cell lines), which correlated with the selectivity of the compounds ( Figure 3A). Additionally, glucose depletion from culture medium resulted in increase of cytotoxic activity of both compounds ( Figure 3B). Moreover, both compounds were able to inhibit glucose uptake in PC-3 cells, which was observed using two different detection methods ( Figure 3C,D). Taken together, our results suggest that the novel compounds are concurrently ingested by the cells via the same GLUT-1 system as glucose and thus target the Warburg effect. PC-3 cells were treated with the compounds for 24 h and then the glucose uptake was measured using 2-NBDG-based assay either in cell culture by plate reader (C) or in single cells using flow cytometry technique (D) and then normalized to cell viability (mean ± SEM; n = 3; * p < 0.05, one-way ANOVA test). Apigenin (Apig), phloretin (Plt), and cytochalasin B (Ccl-B) were used as positive controls. The viability was measured by MTS assay (B,C) or by flow cytometry using PI staining (D).

SAB-13 and -14 Induce Caspase-Dependent Apoptosis
Pro-apoptotic signs such as phosphatidylserine externalization ( Figure 4A-C) and PARP cleavage ( Figure 4D) were found in the cells following 48 h treatment with SAB-13 and -14 at the concentrations which were close to IC 50 s in the correspondent cell lines. In addition, cleavage of caspase-3 was observed ( Figure 4D). Co-treatment with pan-caspase inhibitor zVAD antagonized the cytotoxic effects of SAB-13 and -14, suggesting a caspase-dependent character of the induced apoptosis ( Figure 4B). Furthermore, the cell cycle analysis revealed DNA fragmentation, another pro-apoptotic marker, detected as sub-G1 peak ( Figure 4E). More detailed analysis of the generated data revealed a G2/M-cell cycle arrest under drug treatment ( Figure 4F). This could be at least in part explained with the observed upregulation of p21 ( Figure 4D), which may contribute both to the cycle arrest as well as to the induced apoptotic cell death [31]. one-way ANOVA test). Anisomycin (Aniso; treatment with 10 µM for 48 h) was used as a positive control. (E,F) Cell cycle analysis of 22Rv1 cells using PI staining, apoptotic cells were detected as sub-G1 population (E) (mean ± SEM; n = 3; * p < 0.05, one-way ANOVA test).
For this and further examinations, 22Rv1 cells were used as the main model. 22Rv1 cells express both AR-full length and AR-V7 [32]. The latter mediates resistance of prostate cancer to androgen receptor targeted therapies such as abiraterone or enzalutamide [33,34]. At the same time, PC-3 cells are AR-full length-and AR-V7-negative [32] and thus might be less relevant as clinical model for prostate cancer.

Effect of SAB-13 on the Proteome of Prostate Cancer Cells
To identify possible processes and molecular targets affected under the treatment, we investigated changes in proteome of 22Rv1 cells treated with SAB-13, as the most active compounds of the synthesized panel, at the concentration close to IC 50 s in the correspondent cell line. The changes in the protein expression were examined using the LC-MS/MS method in DIA (data independent acquisition) mode. In total, 2163 proteins were identified and quantified. In total, 253 proteins had 1.5 ≤ fold change ≤ 1/1.5 and were significantly regulated (p ≤ 0.05) (see Supplementary Materials Table S1). In fact, upregulation of 104 proteins and downregulation of 149 was detected following the treatment.
The proteomics data was analyzed using the Ingenuity Pathways Analysis software (IPA, QIAGEN Bioinformatics). Five top protein networks were constructed by IPA and the alteration of five kinases involved in several pro-and anti-apoptotic processes was predicted ( Figure 5A, marked with red circle) and further validated in functional assays ( Figure 5B). Thus, we showed a pronounced activation of ERK1/2, p38, and Akt kinases upon the treatment, while only slight phosphorylation of MEK1/2 and JNK1/2 kinases was observed ( Figure 5B). Thus, alteration of ERK1/2, p38, and Akt may play an important role in the cellular effect mediated by the synthesized compounds. Note that the regulation (cleavage) of PARP and caspase (caspase-3), predicted in IPA analysis ( Figure 5A), was confirmed experimentally as well ( Figure 4D).
Next, gene ontology analysis and z-score algorithm were used to further predict the effect on biological processes/target ( Figure 5C). Most of the proteins regulated under the treatment were located in cellular cytoplasm and nucleus, and were identified as enzymes ( Figure 5C). Remarkably, the activity of several upstream regulators playing an important role in prostate cancer development and progression have been predicted to be affected by SAB-13 ( Figure 5D). Among them, the following are suppressed: E2F3 (transcription factor, its activity was reported to be associated with poor prostate cancer patients treatment prognosis [35]), ADRB (β-adrenergic receptor, its signaling is involved in the development of aggressive prostate cancer [36]), BRD4 (kinase, associated with tumor growth and metastatic potential of prostate cancer [37,38]), 25s proteasome (a promising and relevant target in a number of human cancers including prostate cancer [39]), and MYC (proto-oncogene, is involved in initiation and progression of a number of human cancers [40] and drives the development of the highly aggressive neuroendocrine prostate cancer [36]) ( Figure 5D). Additionally, sirtuin and IL-8 signaling, as well as LXR/RXR and NO and ROS production in macrophages were proposed to be suppressed, while NER and EIF2 pathway, ILK signaling, and oxidative phosphorylation were predicted to be activated under the treatment ( Figure 5E). At the same time, it is important to note that predicted activation of KDM5A and XBP1 and suppression of FOXO1 and RB1 were observed. These alterations may play a possible negative role in prostate cancer as undesirable effects of the drug, which is described elsewhere. Therefore, these effects should be carefully investigated before further development of these drugs for in-human trials. Western blotting (marked by red circle). The activation was observed 2 h after the treatment.β-actin was used as a loading control. The expression phospho-kinases was normalized to the non-phospho-kinases expression levels (mean ± SEM; n = 3; * p < 0.05; one-way ANOVA test). (C) Gene ontology analysis. (D) Top upstream targets and the calculated z-score of the effect (activation/inhibition), predicted by IPA. p-values of overlap is indicated on the graph (significance: p < 0.05, Fischer s exact test). (E) Top canonical pathways, predicted by IPA. The target/process is expected to be activated if z-score > 0 (red area) or suppressed if z-score < 0 (green area) (significance: p < 0.05, Fischer s exact test). (F) Representative scheme of oxidative phosphorylation. Affected proteins discovered by proteome analysis are marked with color.
Note that we discovered the alterations of protein components of four (out of five) main protein complexes located in the inner mitochondrial membrane and involved in the oxidative phosphorylation (see Table S1). These molecules were identified as parts of Complex I (genes NDUFA4, NDUFS6, NDUFB1, and NDUFA2), Complex II (gene SDHB), Complex III (gene UQCR10), and Complex IV (genes COX6B1 and COX17). Therefore, oxidative phosphorylation was predicted to be inhibited/disrupted in SAB-13-treated cells (z-score = −1.89, p-value = 2.37 × 10 −4 ; Figure 5F).

SAB-13 and SAB-14 Induce Apoptosis via Mitochondria Targeting
Oxidative phosphorylation (or electron transport-linked phosphorylation) is a metabolic pathway of nutrients oxidation which takes place in mitochondria and ultimately results in ATP production in eukaryotic cells [41]. Thus, its disruption predicted by bioinformatical analysis of proteomics data may indicate mitochondria targeting by SAB-13 and similar compounds. SAB-13 was found to affect complexes I-IV, essential elements of oxidative phosphorylation machinery, which are located in the mitochondrial inner membrane. Consequently, we examined the effect of SAB-13 on the mitochondrial membrane potential (∆Ψ m ) at the concentration close to IC 50 . Significant and rapid loss of ∆Ψ m was detected in the prostate cancer cells following both short-(2 h) and long-term (48 h) treatment ( Figure 6A,B). Next, ROS production level was examined, which is known to be closely associated with mitochondrial integrity [42]. Indeed, a significant increase of the ROS level was observed 2 h after the treatment, simultaneously with ∆Ψ m loss ( Figure 6C,D). Moreover, pretreatment with an established antioxidant N-acetyl-L-cysteine could rescue the cells from the drug-induced apoptosis ( Figure 6E). This confirms the cytotoxic effect of ROS generation under treatment. Additionally, we were able to show that caspase-9 (but not caspase-3) is activated shortly after treatment at the concentration close to IC 50 (2 h, Figure 6F), whereas caspase-3 cleavage was observed after 48 h treatment ( Figure 6G). Finally, the examination of different subcellular fractions revealed a redistribution of such pro-apoptotic proteins as cytochrome C and AIF (apoptosis inducing factor) from mitochondria to cytoplasm ( Figure 7). This effect was accompanied by apoptosis induction, as indicated by cleaved PARP and cleaved caspase-3 as apoptotic markers (Figure 7). In conclusion, induced cancer cell apoptosis was strongly associated with targeting of mitochondria. , and the cells containing depolarized mitochondria were quantified using the Cell Quest Pro software (B) (mean ± SEM; n = 3; * p < 0.05, one-way ANOVA test). CCCP (50 µM) was used as a positive control. (C,D) Effect on ROS production in 22Rv1 cells after 2 h of treatment. Cells were stained with CM-H 2 DCFDA, treated with the drugs, harvested, and analyzed by FACS (C), and the ROS level was quantified using the Cell Quest Pro software (D) (mean ± SEM; n = 3; * p < 0.05, one-way ANOVA test). H 2 O 2 (200 µM) was used as a positive control. (E) 22Rv1 cells were pre-treated with 1 mM NaC (N-acetyl-l-cysteine) for 1 h and then co-treated with the investigated drugs for 48 h FBS-and glucose-free media. Cell viability was measured by MTT assay, treatment time was 48 h. (F,G), Western blotting analysis of cleaved caspase-3 and -9 levels in 22Rv1 cells treated with SAB-13 for 2 h (F) or 48 h (G). Anisomycin (Aniso) was used as a positive control. β-actin was used as a loading control (mean ± SEM; n = 3; one-way ANOVA test).

Figure 7.
Quantification of the proteins in mitochondrial, cytoplasmic, and total fractions. 22Rv1 cells were treated with the drugs for 48 h, the proteins were extracted and fractionated using Cell Fractionation Kit (abcam). The mitochondrial (Mito), cytoplasmatic (Cyto) or total (Total) fractions were concentrated and analyzed by Western blotting. Cells treated with carbonyl cyanide 3-chlorophenylhydrazone (CCCP; 50 µM) were used as a positive control. α-tubulin was used as a loading control, the protein expression was normalized to α-tubulin expression levels either in cytosolic fraction or in the total protein lysate (mean ± SEM; n = 3; * p < 0.05; one-way ANOVA test).

Discussion
A significant percentage of prostate cancer patients initially respond to targeted (e.g., abiraterone acetate, enzalutamide, PARP-inhibitors, and PSMA-ligand therapy) or untargeted therapies (e.g., cabazitaxel, docetaxel, and radium-223) [43]. However, the majority of patients will experience an increasing loss of sensitivity to standard medications with each additional treatment line [44]. Primary as well as secondary drug-resistance are responsible for the limited treatment success and poor prognosis of patients suffering from advanced castration resistant prostate cancer (CRPC) [45]. Thus, expression of androgen receptor transcriptional variants of AR (AR-Vs), in particular AR-V7 [33], has been identified as an important mechanism of resistance to AR-targeted therapies. In addition, side effects caused by the damage of non-malignant cells limit the application of standard therapies in a mostly older and comorbid patient population. Thus, novel therapeutics capable of overcoming drug resistance while sparing side effects are urgently needed.
Conjugation of the cytotoxic molecule and the moiety, which is responsible for selective delivery of the active substance to cancer cells may increase specificity of the drug and to reduce its side effects. In the current study we conjugated bioactive 1,4-naphthoquinone scaffold, which has been previously shown to exhibit cytotoxic activity in human cancer cell lines, with a glucose moiety via a sulfur linker at C6 position using simple and effective methods. Note that the generated non-glycoside bond simplified the chemical synthesis of the compounds. Thus, we recently reported a number of O-glycosides of 1,4-naphthoquinones to undergo decomposition during the deacetylation step [46]. Moreover, we showed that some S-glycosides can convert into linear or angular tetracycles under the same conditions [46], whereas no undesired effects or side reactions were observed during the synthesis described in the current research. Thus, we used either: (i) a substitution reaction of the halogen atoms in 1,4-naphthoquinone derivatives by a sulfur atom of 6-mercaptoglucose derivative; or (ii) an addition reaction of 6-mercaptoglucose derivative to juglone, its acetate, or naphthazarin, followed by saponification of the acetylated conjugates.
This modification resulted in a higher specificity to cancer cells due to Warburg effect targeting as well as increased water solubility and therefore bioavailability of the synthesized molecules. An exchange of the ether linker to the thioether linker most likely will provide higher stability of the compounds in the human body [27] and a greater resistance to the enzymes mediated hydrolysis. The conjugation at C6 position of glucose molecule ensures the affinity of the synthesized conjugates to GLUT-1 and therefore its successful cellular uptake via this system. Furthermore, it should provide additional resistance to glycoside-specific enzymes due to the non-glycoside nature of the bond as well as novel sulfur linker.
The mechanism of anticancer activity of the natural 1,4-naphthoquinones and other related compounds includes generation of free superoxide radicals, which results in DNA damage and may lead to p53-independent cell death [22,47,48], as well as inhibition of topoisomerase-II [47,49,50]. Consequently, the examination of the mechanisms of anticancer activity of SAB-13 and -14 revealed an induction of caspase-dependent apoptosis. Additionally, we detected a G2/M arrest of the cancer cells under drug treatment, which could be related to the observed p21 upregulation in the treated cells. Pro-apoptotic effects of the synthesized conjugates were accompanied by downregulation of several anti-apoptotic proteins. Further investigation of the drug-mediated changes in cellular proteome as well as bioinformatical analysis of the data suggested mitochondria targeting as one of the central effects of SAB-13. Indeed, we were able to validate this effect in functional assays.
Thus, SAB-13 is able to promote a generation of cytotoxic ROS and to induce a drop-down of mitochondrial membrane potential, which resulted in a release of cytotoxic mitochondrial proteins to cellular cytoplasm, caspases activation and ultimately cancer cell death (Figure 8). In addition, SAB-13 activated caspase-9 shortly after treatment and prior to caspase-3 ( Figure 8). It has been previously shown that the modification of existing drugs in order to assign them mitochondria-targeting properties (i.e., conjugation with the mitochondria-specific carrier molecule) significantly increases their activity towards cancer cells [51,52]. Moreover, these drugs were reported to be able to overcome the drug resistance to standard medications [51,52]. Despite the upregulated glycolysis (anaerobic respiration) in cancer cells in comparison to normal tissues, the oxidative phosphorylation (part of aerobic respiration machinery) was reported to be still active and even upregulated in different cancer entities [53]. Thus, oxidative phosphorylation has recently been identified as a new promising target in anticancer therapy promoting the therapeutic potential of the inhibitors of this biochemical process [41]. In fact, inhibitors of different steps of the oxidative phosphorylation cascade such as metformin, atovaquone, arsenic trioxide, carboxyamidotriazole, fenofibrate, and lonidamine have exhibited therapeutic efficacy in a number of clinical and preclinical studies for the treatment of different cancer types (reviewed in [53]). Thus, the combination of Warburg effect mediated selectivity and the simultaneous targeting of oxidative phosphorylation/mitochondria in cancer cells makes the synthesized conjugates promising agents in the therapy of the human CRPC and other cancers. Ongoing experiments will clarify if oxidative phosphorylation is a primary target of the drugs, or if the targeting of other mitochondrial components and functions cause a disruption of the oxidative phosphorylation along with other mitochondrial processes. In vivo experiments are currently in preparation to validate the above-described in vitro results.

General Chemistry (Reagents, Solvents and Equipment)
Reagents and solvents were purchased from Sigma (Taufkirchen, Germany) and Vekton (St. Petersburg, Russia). The initial quinones were purchased from Sigma or synthesized as described elsewhere. All reagents, initial quinones, and solvents were of analytical grade and used as received. The melting points were determined on a Boetius melting-point apparatus (Dresden, Germany) and are uncorrected. The 1 H and 13 C NMR spectra were recorded using Bruker Avance-300 (300 MHz), Bruker Avance III-500 HD (500 MHz), and Bruker Avance III-700 (700 MHz) spectrometers (Bruker Corporation, Bremen, Germany) using CDCl 3 and DMSO-d 6 as the solvents with the signal of the residual non-deuterated solvent as the internal reference. The progress of reaction was monitored by thin-layer chromatography (TLC) on Sorbfil plates (IMID, Krasnodar, Russia) using the following solvent systems as eluents: hexane/benzene/acetone, 3:1:1 (v/v) (System A) and benzene/ethyl acetate/methanol, 7:4:2 (System B). Individual substances were isolated and purified using crystallization, column chromatography as well as preparative TLC on silica gel (Silicagel 60, 0.040-0.063 mm, Alfa Aesar, Karlsruhe, Germany) which was preliminarily treated with hydrochloric acid (pH = 2) in order to reduce a residual adsorption of quinones, followed by drying and activation by heating at 120 • C.

Cell Lines and Culture Conditions
The human prostate cancer cell lines PC-3, DU145, 22Rv1, and LNCaP, as well as human prostate non-cancer cell lines RWPE-1 and PNT2 were purchased from ATCC (Manassas, VA, USA). The human prostate cancer cell line VCaP was purchased from ECACC (Salisbury, UK). HEK 293T (human embryonic kidney cells), MRC-9 (human fibroblast cells) and HUVEC (human umbilical vascular endothelial cells, passage 11) were a gift from Prof. Sonja Loges (University Medical Center Hamburg-Eppendorf, Hamburg, Germany). All the cells had a passage number ≤ 30. Authentification of all cell lines used was recently performed by commercial service (Multiplexion GmbH, Heidelberg, Germany).
Cells were cultured as monolayers at 37 • C in a humidified atmosphere with 5% (v/v) CO 2 in the correspondent culture medium: 10% FBS/RPMI medium (RPMI medium supplemented with Glutamax TM -I (gibco ® Life technologies TM , Paisley, UK) containing 10% fetal bovine serum (FBS, gibco ® Life technologies TM ) and 1% penicillin/streptomycin (Invitrogen)) for PNT2, LNCaP, 99.9% acetonitrile and 0.1% formic acid) buffer at a flow rate of 250 nL/min using a nano-UPLC system (Acquity, Waters Corporation, Milford, MS, USA) and full recovery autosampler vials, coupled via electrospray ionization (ESI) to a tandem mass spectrometer equipped with a quadrupole and an orbitrap (QExactive, Thermo Fisher Scientific, Bremen, Germany), using MS/MS mode in DDA (data dependent acquisition) and DIA (data independent acquisition). The peptide library necessary for DIA analysis was generated by DDA-LC-MS/MS of 1 randomly selected sample per biological triplicate. The peptides were identified with the search engine Sequest (integrated in the Proteome Discoverer software version 2.0, Thermo Fischer Scientific) using the human Uniprot protein database (EMBL; Hinxton Cambridge, UK, release December 2018). All samples were acquired in DIA mode and data extraction with the peptide library was performed using the Skyline software (version 4.2, MacCoss Lab Washington University, Seattle, WA, USA). Peptides with a dot-product (dotP) of 0.85 and above were considered for further analysis and the peptide areas were summed to protein areas. Statistical analysis (data normalization, Student s t-tests, principal component analysis, hierarchical clustering) was performed using the Perseus software (version 1.5.8, Max-Planck Institute of Biochemistry, Munich, Germany). Proteins were considered to be significantly different in abundance if a 1.5 ≤ fold change ≤ 1/1.5 and p-value ≤ 0.05 was detected (Student s t-test).

Bioinformatical Analysis
Bioinformatical analysis of the proteomics data was performed using the Ingenuity Pathways Analysis (IPA) (release 46901286, QIAGEN Bioinformatics, Venlo, The Netherlands). The list of significantly regulated proteins discovered by global proteome analysis were proceeded by IPA software to access protein interaction networks, gene ontology, relevant biological molecules, and processes affected by the drug, as previously described [58]. The shortest hypothetical pathway networks were constructed, showing the most relevant direct and indirect interactions of the regulated proteins and the proteins predicted to be involved in the interactions. z-Score algorithm was used to identify biological functions which are expected to be suppressed (z-score < 0) or activated (z-score > 0) upon treatment in cancer cells. The p-values were calculated by IPA software using the Fischer s exact test according to the software s protocols.

Analysis of Intracellular ROS Level
Intracellular ROS levels were accessed using the ROS-sensitive CM-H 2 DCFDA reagent (Cat. No. C6827, Molecular probes, Invitrogen, Eugene, OR, USA). The experiment was performed as previously described with slight modifications [59]. In brief, 22Rv1 cells (10 5 cells/well) were seeded in 12-well plates and incubated overnight. The media was changed with freshly made pre-warmed staining solution (4 µM CM-H 2 DCFDA/PBS, 0.5 mL/well) and incubated in the dark (37 • C, 5% CO 2 ) for 30 min. Then, the staining solution was exchanged with 1 mL/well of pre-warmed PBS containing the investigated compounds or H 2 O 2 (positive control) at the indicated concentrations and the cells were incubated for 2 h. Next, the cells were trypsinized, resuspended in pre-warmed PBS (200 µL/sample) and immediately analyzed by FACS according to the manufacture s protocol.

Cell Fractionation
The separation of cellular nuclear, mitochondrial, and cytosolic fractions was performed using the Cell Fractionation Kit ab109719 (abcam, Cambridge, MA, USA). 22Rv1 cells (4 × 10 6 cells/bottle) were seeded in T75 culture bottles (in 20 mL/bottle) and incubated overnight. Cells were treated for 48 h and harvested by scratching. Further procedures were performed as previously described with slight modifications [59]. Nuclear fractions were homogenized using QIAshedder kit (QUIAGEN, Hilden, Germany). Cytosolic and mitochondrial fractions were concentrated using Amicon ® Ultra-2 Centrifugal Filter device (Cat. No. UFC203024, Merck, Darmstadt, Germany).

Data and Statistical Analysis
For statistical analyses of all the datasets (apart from proteomics data), the GraphPad Prism software v. 7.05 was used (GraphPad Prism software Inc., La Jolla, CA, USA). Data are presented as mean ± SEM. The unpaired Student s t-test or one-way ANOVA followed by Dunnett s post-hoc tests were used for comparison of two or several groups, correspondently. Dunnett s post-hoc test was used to compare each of a number of a multiple treatment with a single control. All experiments were performed in in triplicates (n = 3, biological replicates) unless otherwise stated. Differences were considered to be statistically significant (*) if p < 0.05.