Design, synthesis and biological evaluation of selective survivin inhibitors

The differential distribution between cancer cells and normal adult tissues makes survivin a very attractive cancer drug target. We have previously reported a series of novel selective survivin inhibitors with the most potent compound MX106 reaching nanomolar activity in several cancer cell lines. Further optimization of the MX106 scaffold leads to the discovery of more potent and more selective survivin inhibitors. Various structural modifications were synthesized and their anticancer activities were evaluated to determine the structure activity relationships for this MX106 scaffold. In vitro anti-proliferative assays using two human melanoma cell lines showed that several new analogs have improved potency compared to MX106. Very interestingly, these new analogs generally showed significantly higher potency against P-glycoprotein overexpressed cells compared with the corresponding parental cells, suggesting that these compounds may strongly sensitize tumors that have high expressions of the P-glycoprotein drug efflux pumps. Western blotting analysis confirmed that the new MX106 analogs maintained their mechanism of actions by selectively suppressing survivin expression level among major inhibitors of apoptotic proteins and induced strong apoptosis in melanoma tumor cells.


Introduction
Survivin is the smallest (molecular weight 16.5 kDa) member of the IAP (Inhibitor of Apoptosis Protein) family [1][2] . It is a unique anti-tumor therapy target because: (a) it acts as an essential anti-apoptosis guard which protects cells from the apoptotic cascade by binding to the activated caspases and neutralizing proapoptotic receptor [3][4][5] ; (b) it is a key regulator of mitosis as a component of the chromosomal passenger complex (CPC) [6][7] ; (c) it widely participates in tumorigenesis signaling pathways such as Akt, p53, Wnt-2 and MDM2 [8][9][10] ; (d) it is highly expressed in most types of human cancer cells and embryonic tissues where rapid cell growth is needed, but has very low expression in normal adult differentiated tissues [11][12] ; (e) its expression level is closely correlated with tumor metastasis, poor disease prognosis, and high risk of chemo/radio-resistance [8,11,[13][14][15][16][17] .
However, due to the nature of survivin function (protein-protein interaction) and its cell cycle dependent expression, the existing armory of survivin inhibitors is still very limited [8,18] . Most current small molecular survivin inhibitors do not directly bind to the survivin protein, but rather interfere with survivin production (gene transcription, post-translation modification) or accelerate its degradation [8,12,[18][19][20] . Multidrug resistance (MDR) remains a major obstacle in cancer chemotherapy, accounting for more than 90% failure in clinical treatment to cancer patients. A family of ATP binding cassette (ABC) transporters, which is a class of drug efflux pumps, plays a key role in developing MDR [21] . At least 15 human ABC transporters, such as P-glycoprotein (P-gp), multidrug resistant protein1 (MRP1) and breast cancer resistant protein (BCRP) have been found to mediate MDR by inducing drug efflux [22] . Among them, P-gp has been the most extensively identified MDR in cancer cells and a broad range of chemotherapy drugs are known to be substrates transported by P-gp. As the most potent small molecule survivin inhibitor reported so far, YM155 is a survivin gene promotor inhibitor and is well known to be highly susceptible to clinically relevant MDR, including but not limited to P-gp overexpression [18][19][23][24] . It also shows potential toxicities, which calls for special dose administration arrangements. Therefore, it is important to develop novel selective survivin inhibitors that can overcome clinically relevant MDR.
Utilizing high throughput virtual screening and biological test oriented strategy, our laboratory has previously developed a platform of cell permeable small molecular survivin inhibitors, which selectively and effectively reduced survivin expression level in human melanoma and prostate cancer cells, with potent antitumor growth efficacy as validated by a human melanoma xenograft model [25][26] . UC112 is the lead compound we identified through virtual screening and in vitro biological studies. Preliminary optimization of UC112 leads to the discovery of a potent and selective survivin inhibitor, MX106. The structure of UC112 and MX106 are shown in Fig. 1. In this report, we described our efforts to further optimize the MX106 scaffolds which lead to the discovery of more potent MX106 analogs for selective survivin inhibition in tumor cells. We report that these new analogs show strong antiproliferative potency against human melanoma and colorectal cancer cells, effectively overcome P-gp mediated drug resistance, and maintain their mechanisms of action by selectively inhibiting survivin expression and inducing cancer cell apoptosis. Structure-activity relationships (SAR) are determined for these new analogs to support further development of this unique scaffold as a potential anticancer agent.

Materials and methods
General All reagents were purchased from Sigma-Aldrich Chemical Co., Alfa Aesar (Ward Hill, MA), and AK Scientific (Mountain View, CA) and were used without further purification. Routine thin layer chromatography (TLC) was performed on aluminum-backed Uniplates (Analtech, Newark, DE). NMR spectra were obtained on a Varian Inova-500 spectrometer (Agilent Technologies, Santa Clara, CA) or a Bruker Ascend 400 (Billerica, MA) spectrometer. Chemical shifts are reported as parts per million (ppm) relative to TMS in CDCl 3 . High resolution mass spectra were collected in positive detection mode on a Waters Xevo G2-S Tof instrument equipped with an electron-spray ionization (ESI) source (Milford, MA).

Synthesis
Preparation of 5-chloromethyl-8-quinolinol hydrochloride (2) A mixture of 5.84 g (40.0 mmol) of 8-quinolinol, 50 ml of concentrated hydrochloric acid, and 6.4 ml of 37% formaldehyde was treated with 0.6 g of zinc chloride and stirred for 12 hours. The mixture was filtered, washed with copious acetone and dried to give compound 2 as a yellow solid (7.2 g, 78%). 1  To a solution of substituted benzyl alcohol 3 (6 mmol) in anhydrous THF (30 mL) was added sodium hydride (60% dispersion in mineral oil, 0.72 g, 18 mmol) at 0°C. The suspension was stirred at 0°C for 30 minutes. Salt 2 (1.15 g, 5mmol) was added to the suspension. The mixture was stirred at r.t for 3 hours. Water was added to the suspension, and the mixture became homogeneous. The mixture was extracted by ethyl acetate and washed with brine, dried over anhydrous sodium sulfate and concentrated to get the crude. The crude compound was purified by flash chromatography (ethyl acetate: hexane 1:3)   General procedure for the synthesis of compounds (4a-c) An equimolar mixture of the substrates 3, paraformaldehyde, and the pyrrolidine in anhydrous ethanol (30 mL) was refluxed for 4 hours under argon. After cooling, the solvent was evaporated under reduced pressure. The crude compound was purified by flash chromatography (Dichloromethane: methanol 20: 1). Preparation of 5-((naphthalen-2-ylmethoxy)methyl)quinolin-8-ol (5) To a solution of the alcohol in anhydrous DMF (10 mL) was added sodium hydride (60% dispersion in mineral oil, 0.72 g, 18 mmol) at 0°C. The suspension was stirred at 0°C for 30 minutes. Preparation of 5-((naphthalen-2-ylmethoxy)methyl)-7-(pyrrolidin-1-ylmethyl)quinolin-8-ol (6) An equimolar mixture of the substrates 5, paraformaldehyde, and the pyrrolidine in anhydrous ethanol (30 mL) was refluxed for 4 hours under argon. After cooling, the solvent was evaporated under reduced pressure. The crude compound was purified by flash chromatography (Dichloromethane: methanol 20: 1). 1  To a solution of morpholine (6 mmol) in THF was added salt 2 (2 mmol). The mixture was refluxed for 3hr. The white precipitate was filtered, the filtrate was evaporated. The crude was purified by flash chromatography to produce 7. Preparation of 5-(morpholinomethyl)-7-(pyrrolidin-1ylmethyl)quinolin-8-ol (8) An equimolar mixture of the substrates 7, paraformaldehyde, and the pyrrolidine in anhydrous ethanol (30 mL) was refluxed for 4 hours under argon. After cooling, the solvent was evaporated under reduced pressure. The crude compound was purified by flash chromatography (Dichloromethane: methanol 20: 1). 1  Preparation of 2-chloroquinolin-8-ol (10) Quinoline-2,8-diol (5.30 g, 32.89 mmol) was suspended in anhydrous DMF (30 mL). Thionyl chloride (9.45 mL, 131.50 mmol) was added dropwise via a syringe at 0°C under argon. The resulted solution was stirred and heated to 50°C for 4 hours. Then, the reaction was quenched by pouring the solution into ice water followed by extraction with ethyl acetate (3Â50 mL), washed with brine (2Â50 mL), dried over anhydrous MgSO 4 and concentrated under reduced pressure and purified by silica gel column chromatography (eluting with CH 2 Cl 2 ) to give a pale yellow solid product, 5.0 g, 84.7% yield.

Procedure for the preparation of compound 22
2-Hydroxyquinolin-8-yl acetate (18) Quinoline-2,8-diol (5.00 g, 31.00 mmol) was dissolved in anhydrous THF (30 mL) at room temperature under argon. Acetic anhydride (10 mL) was added via a syringe. The reaction solution was stirred and heated to reflux overnight. Then, the volatile was removed under reduced pressure. The residue was purified by silica gel column chromatography (eluting with CH 2 Cl 2 /MeOH = 19/1 v/v) to give a white solid product, 5.80 g, 92.1% yield.

Synthesis of 4-((tert-butyldimethylsilyl)oxy)-1naphthaldehyde (24)
To a solution of aldehyde 15 (2 mmol) in DCM was added TBDMSCl (3 mmol) and imidazole (4 mmol). The mixture was stirred overnight. Water was added and the mixture was extracted with DCM. The organic layer was dried over anhydrous sodium sulfate. The crude was purified by flash chromatography to generate 24. Synthesis of (4-((tert-butyldimethylsilyl)oxy) naphthalen-1-yl)methanol (25) To the protected aldehyde 16 (1 mmol) in methanol was added NaBH4 (3 mmol). The mixture was stirred at room temperature for 2 hours. The solvent was evaporated and extracted with ethyl acetate. The crude was subject to flash chromatography to produce alcohol 25. 1  Synthesis of((4-((benzyloxy)methyl)naphthalen-1-yl) oxy) (tert-butyl)dimethylsilane (26) To a solution of alcohol 25 in anhydrous THF (10ml) was added sodium hydride (60% dispersion in mineral oil) at 0°C. The suspension was stirred at 0°C for 30 minutes. Benzyl bromide was added to the suspension. The mixture was stirred at r.t for 3 hours. Water was added to the suspension, the mixture became homogeneous. The mixture was extracted by ethyl acetate and washed with brine, dried over anhydrous sodium sulfate and concentrated to get the crude. The crude compound was purified by flash chromatography (ethyl acetate: hexane 1:3). 1

Synthesis of 1-(benzyloxy)-4-((benzyloxy)methyl) benzene (30)
To a solution of alcohol 29 (3 mmol) in anhydrous THF (30ml) was added sodium hydride (60% dispersion in mineral oil, 6 mmol) at 0°C. The suspension was stirred at 0°C for 30 minutes. Benzyl bromide (4 mmol) was added to the suspension. The mixture was stirred at r.t for 3 hours. Water was added to the suspension, the mixture became homogeneous. The mixture was extracted by ethyl acetate and washed with brine, dried over anhydrous sodium sulfate and concentrated to get the crude. The crude compound was purified by flash chromatography (ethyl acetate: hexane 1:3). 1

Synthesis of 4-((benzyloxy)methyl)phenol (31)
To a solution of intermediate 12 (1.0 equiv.) and NiCl 2 .6H 2 O (1.5 equiv.) in methanol (5 mL) at 0°C, NaBH 4 (3.0 equiv.) was added and stirred until complete consumption of the starting material. The progress of the reaction was monitored by TLC. After completion, the reaction mixture was quenched with methanol and stirred for another 20 minutes. It was then filtered through celite pad and the filtrate was concentrated under reduced pressure. The crude product, thus obtained, was purified by column chromatography on activated silica gel using ethyl acetate-hexane mixture as eluent to afford the pure phenol 13. To a solution of MX106 in anhydrous THF was added sodium hydride at 0°C. The mixture was added at 0°C for 30 minutes. N-Phenyl-bis (trifluoromethanesulfonimide) was added. The mixture was refluxed for 4hr. The mixture was extracted by ethyl acetate. The crude was subject to flash chromatography to generate 33. To a solution of compound 33 in methanol was added Pd/c (10%, 0.05eq), Mg (0.1 eq) and ammonium acetate (2 eq). The mixture was stirred at room temperate for 5 hours. The mixture was filtered by celite. The crude was purified to generate compound 25. To a solution of alcohol (6 mmol) in anhydrous THF (30ml) was added sodium hydride (60% dispersion in mineral oil, 0.72 g, 18 mmol) at 0°C. The suspension was stirred at 0°C for 30 minutes. Salt 2 (1.15 g, 5mmol) was added to the suspension. The mixture was stirred at r.t for 3 hours. Water was added to the suspension, the mixture became homogeneous. The mixture was extracted by ethyl acetate and washed with brine, dried over anhydrous sodium sulfate and concentrated to get the crude. The crude compound was purified by flash chromatography (ethyl acetate: hexane 1:3)

Cell culture and in vitro antiproliferative assays in human melanoma cells
Human melanoma A375 and M14 cell lines were purchased from ATCC (American Type Culture Collection, Manassas, VA, USA), and cultured in DMEM media (Mediatech, Inc., Manassas, VA) at 37°C in a humidified atmosphere containing 5% CO 2 . The culture media were supplemented with 10% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA) and 1% antibiotic-antimycotic mixture (Sigma-Aldrich, St. Louis, MO). Compounds were dissolved in dimethylsulfoxide (DMSO; Sigma-Aldrich) to make a stock solution of 10 mmol/L. Compound solutions were freshly prepared by diluting stocks with cell culture medium before use (final solution contained less than 0.5% DMSO). Five thousand cells in logarithm growing phase were seeded overnight into each well of a 96-well plate. Then the cells were continuously incubated for 48 hours with sequential diluted compound solution (3 nmol/L to 100 mmol/L, 100 mL per well) in cell culture medium. The cell viability was determined in MTS assay and IC 50 was calculated (n = 4), following similar procedures as described previously [25][26][27][28] .
Cell culture and cytotoxicity assays in human epidermoid carcinoma and colorectal cancer cells The P-glycoprotein (P-gp) overexpressing KB-C2 cell line was established from a parental human epidermoid carcinoma cell line KB-3-1, by a stepwise selection of KB-3-1 in increasing concentrations of colchicine up to 2 mg/mL [29] . SW620/Ad300, which is also a P-gp overexpressing drug resistant cell line, was established by stepwise exposure of the parental human colon cancer cell line SW620 to increasing concentrations of doxorubicin up to 300 ng/mL [30] . The SW620 and SW620/Ad300 cell lines were kindly provided by Dr. Susan E. Bates and Dr. Robert W. Robey (NCI, NIH, Bethesda, MD, USA). All the cell lines were grown in DMEM supplemented with 10% FBS and 100 unit/mL penicillin/streptomycin in a humidified incubator containing 5% CO 2 at 37°C.
The MTT colorimetric assay was used to measure the sensitivity of the cells against the synthesized compounds. The assay detects the formazan product formed from the reduction of MTT in active cells thus assesses the cell viability [31] . Cells were seeded in 96-well plates at 5,000 cells/well (KB-3-1 or KB-C2 cells) or at 7000 cells/well (SW620 or SW620/Ad300 cells) in 180 mL completed medium and cultured overnight. Then various concentrations of the compounds (20 mL) were added to the designated wells. After 72 hours continuous drug incubation, 20 mL of MTT reagent (4 mg/mL) was added to each well and the plates were incubated at 37°C for 4 hours. Subsequently, the medium was removed and 100 µL of DMSO were added to dissolve the formazan crystals in each well. The absorbance was determined at 570 nm by the accuSkan TM GO UV/Vis Microplate Spectrophotometer (Fisher Sci., Fair Lawn, NJ). The IC 50 values of each compound on each cell line were calculated from the survival curves to represent the cytotoxicity of the compounds. The fold of drug resistance was calculated by dividing the IC 50 of the P-gp overexpressing cells by that of the parental cells. Two known P-gp substrates, YM155 and paclitaxel, were used as positive controls for P-gp overexpressing cell lines. On the other hand, cisplatin, which is not a substrate of P-gp, was used as negative control drug.

Western blotting
To determine the change of protein levels of survivin and closely related IAPs, lysates of A375 or M14 melanoma cells treated by the compound solution for 24 hours were used for Western blotting analysis. Primary rabbit antibodies against IAP proteins including survivin (#2808), XIAP (#2045), cIAP1 (#7065), Livin (#5471), Cleaved PARP (#9185) and the loading control protein GAPDH (HRP Conjugate) (#3683) were purchased from Cell Signaling Technology, Inc. (Danvers, MA) and used according to manufacture instructions as reported previously. Protein lane intensities were quantified using ImageJ software (US National Institutes of Health, Bethesda, MD) and band intensities were normalized with vehicle control and GAPDH loading control.

Statistical analysis
All experiments were repeated at least three times and the differences between the mean values of two groups were determined by using two-tailed Student's t-test in cell cytotoxicity assay. The difference between the survival curves of two groups were determined by twoway ANOVA. Results were considered statistically significant when P < 0.05.

Chemistry
The general synthesis of C ring substituted UC112 analogs 4a-4c is shown in Fig. 2. The procedure is similar to the one reported in our recent findings [26] . First, salt 2 was generated, which then reacted with different substituted benzyl alcohols to form ethers 3a-3c. Then ether 3a-3c were submitted to Mannich reaction with paraformaldehyde and pyrrolidine to form compounds 4a-4c. The synthesis of compound 6 applied similar method as compound 4a-4c, which is shown in Fig. 3. Compound 8 which has morpholine ring on the C ring position was prepared as  obtained, it underwent two ways. In the first way, ether 12 went through Mannich reaction to form compound 13. In the second way, the chloro group on A ring of ether 12 is replaced with 4-methylpiperazine to generate compound 14. Compounds 15-17 are synthesized from compound 13. Compound 15 was produced through the reaction of Compound 13 with Isopropyl thiol. Compound 16 was obtained through the reaction of 4methylpiperazine with Compound 13. Replacing the pyrrolidine ring on Compound 13 with imidazole ring generated compound 17. Compound 22, which has bromo on A ring, was prepared as Fig. 6 shown. The hydroxyl group on quinoline-2,8-diol was protected by acetate group [32] . Then, the bromo group was introduced to A ring to form intermediate 19, which was converted to intermediate 20 by reacting with formaldehyde and catalytic zinc chloride in concentrated hydrochloric acid. Intermediate 20 was heated in heated in 4-isopropylbenzyl alcohol to form ether 21. Ether 21 then underwent Mannich reaction to produce compound 22. Compound 28 which replaces the nitrogen on the A ring with Carbon was synthesized as Fig. 7 shown. Aldehyde 23 was first protected by TBDMS group. Then the protected aldehyde 24 was reduced to form alcohol 25, which reacted with benzyl bromide to form ether 26. De-protection of ether 26 with TBAF generated intermediate 27. Then intermediate 27 was converted to compound 28 via Mannich reaction. Compound 32 which has no A ring was prepared as Fig. 8 shown. Alcohol 29 was first reacted with benzyl bromide to form ether 30, which was converted to intermediate 31 by removing the benzyl group on phenol position using the method reported in literature [33] . Intermediate 31 then underwent Mannich reaction to generate compound 32. The synthesis of compound 34 and 35 is shown Fig. 9. First triflate group was introduced to MX106 to form compound 33. Removal of the OTf group produced compound 34. Reduction of compound 33 with Pd/Carbon generated compound 35. Compounds 37a-b were prepared as Fig.  10 shown. First ethers 36a-b were synthesized. Then they underwent Mannich reaction to form compounds 37a-b. The amine linked analog, compound 39 was prepared as Fig. 11 shown. Salt 2 first reacted with amine to form intermediate 38, which was converted to compound 39 through Mannich reaction.

Effect of C ring modification on MX106
The in vitro activity for C ring modified MX106 analogs, compounds 4a-4c, 6 and 8 in A375 and M14 melanoma cells is shown in Table 1. Replacing the isopropyl group with an azide group (4a) or an ethynyl group (4b) resulted in lower activity than MX106 in both cell lines (1.6 µmol/L and 1.41 µmol/L vs. 0.73 µmol/L in A375; 1.4 µmol/L, and 1.38 µmol/L vs. 0.72 µmol/L in M14, for 4a, 4b, and MX106, respectively). However, introducing two substitution groups (4c) to the C-ring or replacing the substituted phenyl ring with a naphthalene ring (6) [26] .

Effect of A ring modification on MX106
As shown in Table 1, all these modifications resulted in significantly loss of activity. Introducing a chloro (13), a bromo (22), or an isopropyl thiol group (15) to  the A ring of MX106 all decreases the anti-proliferative activity compared with that of MX106. The addition of a bulky group as 4-methylpiperazine further decreased activity. Furthermore, replacing the nitrogen atom with a carbon atom on the A ring (28) or complete removal of this ring (32) caused total loss of activity. These in vitro results demonstrate that introduction of substitutions or removal of this pyridine ring of MX106 is not tolerable, and that an unsubstituted pyridine ring on the A ring positon is optimal for activity.  Effect of D ring modification on MX106 Some D ring modifications were also performed based on the A ring modified analogs. The in vitro activity for the D ring modified analogs, compounds 14 and 17 are also shown in Table 1. Compound 14 which removed D ring did not improve activity compared to MX106. Compound 17 which replaced pyrrolidine ring with imidazole ring was less active than MX106 as well. This result suggests the pyrrolidine ring in this scaffold is needed for activity.

Effect of B ring modification on MX106
Since B ring on the UC112 scaffold haven't been modified before, we made several B ring modified analogs, compound 33-35. As shown in Table 1, Compound 33 which protects the hydroxyl group on the B ring with triflate group was basically inactive, which means the free hydroxyl group is essential for activity. Compound 34, which removes the hydroxyl group on the B ring, was also inactive. This further confirms the requirement of a free hydroxyl group on B ring for activity. Compound 35 which reduced the pyridine ring to a saturated ring improved activity compared with its unsaturated form 34, but it is still less active than MX106.

Effect of linkers between B ring and C ring of MX106
Finally, we modified the linker between B ring and C ring of MX106. Three compounds 37a, 37b and 39 were  In general, this systematic modification on different parts of MX106 did not provide significant improvement to its activity, suggesting there are limited structural modifications that can be tolerated in this scaffold to maintain the efficiency for disrupting protein-protein interactions.

MDR evaluation results of MX106 analogs
In order to evaluate whether our new compounds can overcome MDR, we selected four most potent compounds, 4c, 6, 37a and 37b along with MX106, and submitted them to two pairs of cell lines for in vitro antiproliferative activity test. One pair was P-gp overexpressed SW620/Ad300 cell line and its parental, drug-sensitive cell line SW620. The other pair was P-gp overexpressed KB-C2 cell line and its parental, drugsensitive cell line KB-3-1. The IC 50 values of the five compounds on the parental cancer cell lines and their Pgp overexpressing resistant sublines were summarized in Table 2 and Table 3. From Table 2, all MX106 analogs showed significantly higher potency in drug resistant cell line SW620/Ad300 than its parental cell line SW620. The five analogs all had very small values of fold resistance. Compound 37a has a fold resistance as low as 0.07. Similarly, as shown in Table 3, MX106 analogs were more active in drug resistant cell line KB-   Fig. S1 and Supplemental Fig. S2).
As shown in Fig. S1 and Fig. S2, P-gp substrates YM155 and paclitaxel made an obvious shift to right in survival curves of KB-C2 and SW620/Ad300 in comparison with those of the parental cell lines KB-3-1 and SW620, whereas cisplatin exhibited close survival curves between parental and resistant cell lines. In contrast, P-gp overexpressing cell lines treated with our analogs had the survival curves shifting to left compared to their parental cell lines, indicating that the analogs were able to not only overcome the MDR resulting from P-gp overexpression, but also kill the resistant cells at a lower drug concentration.

Mechanism of action studies
Our previous study has established that MX106 produces its antiproliferative activity by selectively inhibiting survivin expression. To determine whether the new analogs maintain their mechanisms of action, we selected two of the most potent MX106 analog, compound 6 and 37a, and using Western blot analysis to evaluate their inhibitory profile against a panel of IAP proteins. As shown in Fig. 12, compound 6 and 37a, which showed comparatively potent cytotoxicity in tested cancer cell lines, indeed displayed selective and dosedependent survivin inhibition effects on human melanoma A375 and M14 cells after 24 hours treatment at concentrations from 2 to 20 µmol/L. At the highest concentration tested (20 µmol/L), both compound 6 and 37a effectively reduced over 80% or 90% survivin level in A375 or M14 cells, respectively. No significant inhibition effect was detected in other IAP proteins including cIAP1, XIAP, and Livin. In addition, we noticed that treatment of compound 6 or 37a significantly increased the level of cleaved PARP in both cell lines, which is an important apoptosis marker and main cleavage product of the activated caspase-3.

Discussion
In this work, we performed systematic structural modifications to our previously reported MX106 scaffold. SAR was determined through structural modifications to the A ring, B ring, C ring, D ring and the linker of MX106 scaffold. The SAR information will provide useful directions for future development. The most potent compound 37a had IC 50 value in nanomolar range in several cancer cell lines tested in this report. Compound 37a is more potent than the best compound MX106 reported previously.
The MDR evaluation result suggests that our compounds could effectively overcome P-gp mediated MDR and surprisingly induce more potent cytotoxic effects on P-gp overexpressing cells than their parental cells. A similar finding reported by Ludwig, et al. showed that a thiosemicarbazone derivative NSC73306 was more cytotoxic to isogenic KB cell lines with high P-gp expression than to the P-gp-negative KB-3-1 cell line [34] . However, the precise mechanism of the phenomenon remains unclear. Previous studies suggested that survivin plays an important role in MDR in the presence of P-gp, and that inhibition of survivin could sensitize MDR cells to anticancer drugs [35][36][37] . It was reported that in the KB cell line, human breast cancer MCF-7 cell line, and their corresponding P-gp overexpressing cell lines KBv 200 and MCF-7/ADR, neither upregulation nor downregulation of survivin could affect P-gp expression on protein level [34] , but the PI3k/Akt pathway was found to be involved in P-gp associated survivin transcription activity and affected the expression of both P-gp and survivin in the same pattern in MCF-7/ADR [36] . These suggested that survivin in MDR cancer cells may interact with P-gp indirectly. Although these findings might be cell-type specific, we could not eliminate the possibility that the compounds in our study may achieve circumvention of P-gp-mediated MDR through a similar mechanism indirectly affecting P-gp and survivin. It is also possible that by similar mechanism of action on P-gp, our new analogs could overcome MDR mediated by other ABC transporters.
We are currently working on to determine the activity of MX106 against other clinically relevant drug resistance mechanisms and will report our findings in the future. Mechanistically, Western blotting analyses confirmed that the new compounds maintained the same mode of action as MX106 by selectively suppressing survivin expression levels. The excellent activity of those new survivin inhibitors warrants further development.