Understanding and predicting the potency of ROS-based enzyme inhibitors, exemplified by naphthoquinones and ubiquitin specific protease-2† †Electronic supplementary information (ESI) available: All experimental procedures, analytical data for small molecules. See DOI: 10.1039/c6sc02758j Click here for additional data file.

A multidisciplinary approach, composed of organic synthesis, electrochemistry, electrocatalysis and cellular studies, for correlating the molecular features of a 1,2-naphthoquinone scaffold with its ROS generating ability.


Introduction
Reactive oxygen species (ROS) homeostasis is important for the survival and progression of both normal and cancerous cells. 1 Certain amounts of ROS are required for proper cell function, including normal metabolism and signaling, but excessive amounts lead to oxidative stress-an imbalance between the production of ROS and their elimination by molecules or enzymes with antioxidant activity. Extreme oxidative stress will certainly lead to complete cell death, as in the case of treatment of tumors by photodynamic therapy (PDT), 2 but the effect of mild conditions is much less predictable. The outcome depends very much on the primary target that will be modied by reacting with the ROS including lipids, DNA, proteins, particular enzymes, and more. 3 While many cancer cells have developed mechanisms that assist in their survival under relatively high levels of ROS, 3 they may still be vulnerable to exogenous small molecules that are known to generate ROS through redox cycling. 1 This hypothesis has been supported by several recent studies, suggesting selective targeting of cancer cells with ROS-generating small molecules as a viable approach in cancer therapy. [4][5][6] One class of cancer-relevant enzymes reported to be targeted by ROS are the cysteine proteases, whose catalytic Cys moiety has been found to undergo oxidation with consequential inhibition of their activities. 7 The thiol of the catalytic Cys moiety may be oxidized to sulfenic acid (-SOH), sulnic acid (-SO 2 H) or sulfonic acid (-SO 3 H), in a reversible manner in the rst case and irreversible for the other two (Fig. 1).
Overexpression of the ubiquitination-counteracting deubiquitinases (DUBs), a subclass of cysteine proteases, is documented in several disease states like cancer, and neurodegenerative and viral diseases. 8,9 Recent studies revealed that DUBs are susceptible to hydrogen peroxide, suggesting a potential way of regulating their cellular activity under oxidative stress (Fig. 1). [10][11][12] For example, ubiquitin specic protease 1 (USP1) is connected with DNA damage repair, whereas the brain-abundant ubiquitin C-terminal hydrolase (UCHL-1) is linked to neurodegenerative diseases. 13 DUBs are hence emerging as promising drug targets, and their targeting via a novel mechanism of inhibition has become a major goal in academia and in industry. 9 We have recently reported the ROS-susceptibility of USP1 and ubiquitin specic protease 2 (USP2) by using the orthoquinone natural product b-lapachone as a redox recycler. 14 This molecule actually progressed up to phase II clinical trials for cancer treatment, and reported mechanisms of action included delay of the S-phase checkpoint in cancer cells 15 and inhibition of NF-kB. 16 We have contributed to this eld by uncovering the effect of b-lapachone on DUBs, by demonstrating that the mechanism of inhibition by b-lapachone proceeds via ROS generation and irreversible oxidation of the catalytic Cys moiety to the sulnic acid form (Fig. 1). 14 Of particular interest is USP2, due to its association with aggressive prostate cancer and triple negative breast cancer. 17 USP2 is associated with various known substrates in cells and affects the pathways that these substrates are involved in. The best-characterized substrate of USP2 is fatty acid synthase (FAS), responsible for protection of prostate cancer cells from apoptosis. 18 The involvement of USP2 in various aspects of cancer survival leads to a great interest in the design and development of inhibitors against this DUB.
Realizing that b-lapachone is a ROS generating molecule for a dened target, e.g. USP2/1, prompted us to examine how changes in the ortho-quinone scaffold might modulate its redox potential and in turn affect its capacity to generate ROS, the consequences of DUBs inhibition and the cellular behavior of these inhibitors. Acquiring a structure/activity relationship prole and deducting the correlation with the redox properties might enable ne-tuning of potential inhibitors for therapeutic development. We now report a multidisciplinary approach, composed of organic synthesis, electrochemistry, electrocatalysis and cellular studies, for correlating the molecular features of the 1,2-naphthoquinone scaffold with its ROS generating ability. The results reveal large differences between the ROS-generating ability of ortho-vs. para-quinones, a very narrow window of redox potentials for ROS generation and an excellent relationship between ROS-generation and USP2 inhibition. Apoptosis induction by the lead compound (12) in DU145 cell lines is illustrated as well.

Results
We initiated our study by preparing a focused set of 1,2-naphthoquinone derivatives based on the bicyclic core of b-lapachone, since it is the pharmacophore unit in this drug and such simplication enables us to rapidly access the desired compounds.

(b) USP2 inhibitions
The nding that b-lapachone with its ortho-quinone moiety inhibits DUBs through ROS, prompted us to systematically investigate the effect of both para-and ortho-quinones against USP2 inhibition in addition to the synthesized ortho-quinone analogs. 22 Towards this goal, a focused collection of quinonecontaining molecules (15-24, Fig. 2) were obtained from commercial sources, which together with all the synthesized quinone derivatives described above, were tested for USP2 inhibition using our developed quenching pair assay. 23,24 Among the non-substituted derivatives, ortho-naphthoquinone 1 exhibited full inhibition at 5 mM, while only 20% inhibition was obtained for the para-naphthoquinone counterpart 15 at the same concentration. Adding a hydroxyl substituent to give 16 or para-quinone to give 17 led to marginal improvements relative to the original compound 15. 25 The comparison between b-lapachone (18) and dehydro-alapachone (19) revealed complete inhibition against USP2 at 5 mM for both, however at 1 mM b-lapachone displayed 100% inhibition whereas the activity of dehydro-a-lapachone dropped to 11%. A similar comparison with the nor-b-lapachone (20) and nor-a-lapachone derivatives (21) disclosed 100% and 68% USP2 inhibition, respectively. Taken together, these results show that 1,2-quinones are consistently more potent USP2 inhibitors than 1,4-quinones.
Armed with these new ndings, some selected anticancer drugs that are known to generate ROS [doxorubicin (22), mytomycin C (23) and menadione (24)] were screened to establish if DUBs are possible targets for them. 22 These examinations revealed that compounds 22-24 did not show appreciable inhibition against USP2, even at 5 mM concentrations.
Compounds 2-7 (Scheme 1) have different substitutions on the C4 position of ortho-naphthoquinone 1: S-alkyl groups in 2-4, amine in 5, SO 3 À in 6, and methoxy in 7. Compounds 2-4 did not exhibit measurable activity against USP2 at 1 mM, which might be attributed to oxidation of the sulde-moiety therein by the ROS. Compound 6 with its electron-withdrawing sulfonyl group did not show any inhibition at 1 mM, while compounds 5 and 7 with their electron-donating groups (-NH 2 and -OCH 3 , respectively) exhibited substantially increased activity relative to the parent compound 1. Here we observed 33% inhibition at 500 nM for 5 and nearly complete inhibitory activity at 400 nM for 7. Taken together, the methoxy substituent in 7 led to an about 12-fold increase in the activity compared to the unsubstituted naphthoquinone 1, which indicates that electrondonating groups provide a benecial effect when presented on C4.
Compounds in which a methoxy group is present on the nonquinonic ring of 1,2-naphthoquinones, at positions 5, 6 and 7 (compounds 8-10) were also prepared, however none of them displayed improved inhibitory activity at 1 mM. In contrast, compounds 12-14 which have C5-or C6-substituents in addition to the C4-OCH 3 , were potent inhibitors. In these cases, we observed 47% inhibition at 300 nM for 12, 28% at 300 nM for 13, and 32% inhibition at 500 nM for 14. 3-Hydroxy b-lapachone (25, Fig. 2) 26 exhibited 78% inhibition at 300 nM, and was the best candidate in the tricyclic class of compounds.
Having identied compound 12 as the most potent bicyclic inhibitor, its k inact was determined and found to be 3333 M À1 s À1 (Fig. 3). To verify that 12 also inhibited USP2 via the oxidation mechanism proposed for b-lapachone, the mass of the enzyme was measured before and aer treatment with compound 12. The 32 Da increase measured is in perfect agreement with the conversion of the catalytic Cys to sulnic acid ( Fig. 1, ESI †).
To understand the inuence of electron-donating and -withdrawing substituents on C4 of 1,2-naphthaquinone on the reduction potential, the CV of compounds 1 and 5-7 were examined in acetonitrile solution (Fig. 4). This study revealed that the substitution of the naphthoquinone with the electronwithdrawing SO 3 Na group (6) induced a positive shi of the reduction potential (easier to be reduced by 180 mV, Fig. 4b) while substitution with the electron-donating OCH 3 group (7) or NH 2 group (5) shied the reduction potential in the negative direction (harder to be reduced by 160 mV for 7, Fig. 4c, and by 160 mV for 5). Similar results were obtained in Tris buffer, pH 7.5. 27 CV examinations of compounds 9, 12, and 14 were also performed in acetonitrile solution and compared to those of 1 and 7. The inuence of electron-donating and -withdrawing substituents on the aromatic ring of the 1,2-naphthoquinone on the reduction potential was deduced to be considerably less than that when present on the quinone moiety. 27 Relative to 1, the reduction potential of the C4-OCH 3 compound (7) is shied by À160 mV and that of the C6-OCH 3 isomer (9) by only À50 mV. An additive effect of the substituents is obtained for the compound that contains two methoxy groups (12) whose reduction potential is shied by À220 mV. On the other hand, the shi for the C4-methoxy-C6-tosylate-1,2-naphthoquinone (14) is only À60 mV, reecting the simultaneous substitution of the 1,2-naphthoquinone building block by electron-donating and -withdrawing groups. Very similar trends were obtained for the same series of compounds, when their CV analyses were recorded in Tris buffer, pH 7.5.
Electrochemistry under an O 2 atmosphere. The abovementioned CV's were also recorded in aqueous Tris buffer saturated with oxygen to examine any electrocatalytic reduction of oxygen by menadione, b-lapachone, or dehydro-a-lapachone. A catalytic cathodic current in the presence of oxygen was obtained for all compounds, testifying that the reduced naphthoquinones catalyze the reduction of oxygen to O 2 À c, the precursor of all biologically relevant ROS. Since the chromatograms are reversible under nitrogen, the ratio between the cathodic and anodic currents (i cat /i p ) obtained under oxygen becomes a criterion for the catalytic activity. 28 This ratio was determined to be 4.7, 3.2, and 2.2 for b-lapachone, dehydro-alapachone, and menadione, respectively (Fig. 5). This difference clearly shows that b-lapachone catalyzes the reduction reaction of oxygen much more efficiently than its para-analogs. Since their redox potentials are practically identical, any biologically relevant reducing agent capable of reducing them will lead to a larger amount of ROS, in the case of b-lapachone, because of the higher i cat /i p value. The stronger inhibitory activity of b-lapachone relative to dehydro-a-lapachone and menadione hence suggests that the larger potency of the former is due to more efficient production of the enzyme-damaging ROS.
The CV's of 1, 6, 7, and 12 were also recorded under both N 2 and O 2 atmospheres (Fig. 4a-d). Compounds 7 and 12 show catalytic activities for oxygen reduction, while 1 and 6 do not. The (i cat /i p ) for all the naphthoquinones that were studied in this work are summarized in Table 1.

(d) Cell study
Having several potent bicyclic quinones in hand, we checked the ability of compounds 7, 9, 12 and 18 to induce apoptosis in DU145 prostate cancer cells, in which USP2 is overexpressed. 14 Incubation at 6 mM concentration for two hours resulted in  3.92% apoptosis for the DMSO control, 50% for b-lapachone, $51% for 12, 13% for 9 and 9% for 7 (Fig. 6).

Discussion
The ability of the para-quinone based anticancer drugs (e.g. menadione) to generate ROS and our earlier nding that blapachone with its ortho-quinone moiety inhibits DUBs through ROS induced damage to the enzyme, prompted us to systematically investigate the effect of both para-and ortho-naphthoquinones against USP2 inhibition. 22,25 Towards the above goal, a focused collection of quinone-containing molecules ( Fig. 2 and Schemes 1 and 2) were tested for USP2 inhibition. The investigations started with a comparison between the nonsubstituted ortho-naphthoquinone 1 and various para-naphthoquinones: 15-17 and the anticancer drugs 22-24. 22 The apparent superior inhibitory effect of 1 relative to these six compounds triggered efforts towards the synthesis of substituted 1,2-naphthoquinones, of which three (25, 12, and 7) were identied to be more potent USP2 inhibitors than b-lapachone.
In the search for the origin of the superiority of ortho-vs. para-naphthoquinones, both the reduction potentials (quinone/ semiquinone radical, determined under anaerobic conditions) and the electrocatalytic activity for reduction of oxygen (to O 2 À c, which undergoes spontaneous disproportionation to H 2 O 2 and O 2 ) were determined for 11 derivatives. This disclosed that in all cases of identical reduction potentials, the catalytic activity (displayed in terms of i cat /i p ) of ortho-quinones very much exceeds that of analogous para-quinones. This is apparent from the results summarized in Table 1, wherein the reduction potentials of compounds 19 and 24 are practically identical (between À0.20 and À0.24 V in aqueous pH 7.5 buffer) to those of 7, 12, 18, and 25. However the two para-quinone derivatives (19 and 24) are much less efficient O 2 reduction catalysts. The latter phenomenon is not only apparent from the lower i cat /i p ratios, but also from the difference between the voltage of maximum catalytic current and the E 1/2 values (DE in Table 1).
The data obtained regarding electrocatalytic activity serves well for addressing a reoccurring puzzle presented in many literature reports: how organic molecules that are reduced more easily 29,30 (i.e. at less negative redox potentials) than molecular oxygen can still catalyze the reduction of the latter? Under the present conditions (aqueous buffer solution of pH ¼ 7.5), the reduction potential vs. Ag/AgCl of b-lapachone under N 2 atmosphere is À0.24 V, while that of dissolved oxygen in the absence of b-lapachone is À0.53 V (À0.33 vs. NHE). 31 Still, examination of the chromatogram of b-lapachone under an oxygen atmosphere (compound 18, Fig. 5a) clearly reveals that the reduction of oxygen becomes more efficient (indicated by the larger current) and appears at a much less negative potential (maximal at À0.36 V) under these conditions. In fact, the coinciding of the voltage for maximum current in the absence and presence of oxygen clearly testies that b-lapachone acts as a true electrocatalyst. An identical type of examination for dehydro-a-lapachone (compound 19, Fig. 5b) shows that this isomer is much  less potent regarding both terms: the catalytic current (relatively low i cat /i p ) and almost no shi to lower overpotential (maximal at À0.53 V as without the catalyst). The catalytic activity of menadione (24) is even smaller. A reasonable explanation for the larger catalytic activity of ortho-quinone relative to para-quinone for reducing oxygen might be attributed to the stability of the one-electron reduction product obtained in neutral solution, a semiquinone radical. 32 The ortho-but not para-semiquinone radical intermediate may be stabilized by hydrogen bonding of the vicinal oxygen atoms and a proton, via a ve-membered ring (Scheme 3). [33][34][35][36][37] The acidity of this trapped proton should be taken into account when analyzing the reaction with oxygen, by two means: (a) it may induce an electron-coupled proton transfer to produce HO 2 c rather than ionized O 2 À c; and (b) it may facilitate the subsequent reduction to hydrogen peroxide (Scheme 3). 31,38 On the other hand, reduction of the para-quinone in neutral water solution will produce the non-stabilized semiquinone radical intermediate, which can only reduce oxygen via an electron transfer. The produced superoxide anion radical will be relatively stable regarding the second reduction to H 2 O 2 , until it reacts with a proton from the solution to produce a protonated superoxide radical. In simple words, the ortho-semiquinone radical intermediate may induce a general acid catalytic effect for the reduction of O 2 , while catalysis by the para-semiquinone radical intermediate proceeds only via specic acid catalysis.
The most interesting result of the investigations is the correlation between the redox potentials of the ortho-naphthoquinones, their electrocatalytic activity, and their ability to serve as inhibitors of USP2. The results of Table 1 clearly show that the potent inhibitors are very active catalysts for oxygen reduction and that the window of opportunity in terms of the quinone/semiquinone redox potentials is very narrow. The interpretation is that compounds that undergo reduction at potentials lower (more negative) than À0.3 V (vs. SCE, at pH 7.5) might be too short-lived to induce the bimolecular reaction with oxygen (kinetic considerations), while the reducing power of those that are reduced at potentials higher than À0.1 V is too low regarding electron transfer to oxygen (thermodynamic considerations). Even more appealing is the almost perfect correlation between the electrocatalytic activity of the orthonaphthoquinones and USP2 inhibition, which is further demonstrated in Fig. 7. The only exception is compound 5, which according to Fig. 7 and the data in Table 1 should be quite a poor inhibitor. This particular compound however contains a C4-NH 2 group which may undergo oxidation or protonation, or participate in H-bonding as both a H-donor and a H-acceptor, and these features may signicantly differ in pure aqueous and protein-containing media. These variables may affect both the inhibitory effects and electrocatalysis, which is apparently the reason for its exceptional behavior.
The examination of DU145 prostate cancer cells, in which USP2 is overexpressed, regarding induced cytotoxicity via treatment with ve selected quinones (Fig. 6) disclosed that only compound 12 was (marginally) more potent than b-lapachone (18). This result and the low potency of compound 9 are consistent with their independently acquired information regarding USP2 inhibition, redox potentials, and ROS generation. On the other hand, the same kind of rather naïve analysis would lead to the expectation that compounds 25 and 7 should also be very cytotoxic, which is clearly not the case. There are many possible reasons for that shortcoming, however these are out of the scope of the present investigations. 39 There is still no doubt that ROS generation affects the enzymatic activity of USP2, but in more realistic systems there are many more targets for those ROS and their identities might change as a function of the closeness of the particular ROS-generating molecule (naphthoquinones in the present case) to them.

Conclusions
Understanding the parameters that govern ROS generation by small molecules is crucial for the design of efficient inhibitors for biological targets. In this work, we systematically investigated the effect of substituents on the 1,2-naphthoquinone scaffold for benecial USP2 inhibition. Specically, our studies on the quinone/semiquinone redox potentials, and the electrocatalytic reduction of molecular oxygen uncovered very meaningful structure/activity relationships. The comparison of 1,2-and 1,4-naphthoquinone derivatives with identical quinone/semiquinone redox potentials revealed that the former compounds were invariably more potent enzyme inhibitors as well as better electrocatalysts. The latter feature was attributed to a hydrogen-bonding network present in the ortho-semiquinone radicals, which provides the opportunity of general Scheme 3 Possible mechanism of hydrogen peroxide generation. Fig. 7 Correlation between reduction potentials of ortho-naphthoquinones, catalytic reduction currents of oxygen, and the % USP2 inhibition at 500 nM of the ortho-naphthoquinones.
acid catalysis for the reduction of oxygen. Formation of reduced oxygen, the precursor of all ROS, was most signicant for compounds within a very narrow range of redox potentials. Optimization of all deduced variables led to the identication of a new lead compound with benecial USP2 inhibition and redox properties: the 4-methoxy-substituted 1,2-naphthoquinone (12). The obtained lead compound 12 possesses a simplied structure compared to b-lapachone, and yet exhibited potent inhibition of USP2 activity. Notably, the effect of substituents on the quinone ring is more inuential on the inhibition of USP2 compared to substitutions on the aromatic ring. In addition, we also demonstrated that the mode of inhibition of 12 is through the oxidation of a catalytic Cys moiety to its sulnic acid state and further showed that it induces apoptosis in DU145 cells. Altogether, this study uncovers an efficient strategy that may be applied in other systems that are affected by the generation of ROS.