Mariana Reis', Ricardo J. Ferreira', Julianna Serly, Noelia Duarte', Ana M. Madureira', Daniel J. V. A. Santos', Joseph Molnar and Maria-Jose U. Ferreira'

glycoprotein) is one of the best studied mechanisms of MDR. This paper focuses on the inhibitory P-gp efflux activity, pharmacophore modeling and structure-activity relationships studies of sixteen macrocyclic diterpenes and polycyclic derivatives obtained from

model to screen for P-gp dependent efflux inhibitors. Most of the compounds showed potential as MDR reversal agents. Combined analysis of two different statistic algorithms, K-means clustering and Principal Component Analysis discriminated two clusters and showed a strong correlation between log were tested in combination with doxorubicin and all potentiated its activity lowering the ID50. Pharmacophore modeling allowed the definition of an aromatic moiety as an additional feature to a previous published P-gp pharmacophore, creating a new five-point pharmacophore with enhanced selectivity for the most active compounds of the present study. Docking results also show the importance of an aromatic moiety, positively identifying the most relevant residues that can be linked to an inhibitory activity increase.

Docking.

~Address correspondence to this author at the Research Institute for Medicines and Pharmaceutical Sciences (iMed.UL), Faculdade de Farmacia, Universidade de Lisboa, Av. Prof. Gama Pinto 1649-003, Lisboa, Portugal; Tel: +351217946475; Fax: +351217946470; E-mail:mjuferreira@ff.ul.pt

18'75-5992/12

The development of multidrug resistance (MDR) is one of the major obstacles in cancer therapy. The selective pressure of the therapeutic agents leads to the development of populations of tumor cells overexpressing MDR-ABC (ATP-binding cassette) trans-porters, which causes increased efflux rates of the anticancer drugs. For this reason, the intracellular drug levels are too low to induce a therapeutic effect and resistant phenotypes arise. Thus, this is considered to be one of the most common mechanisms for the

development of MDR on this type of cells (e.g. colon, kidney, breast, lung and acute myelogenous leukemia cancers). However, MDR can occur intrinsically in some cancers without previous exposure to chemotherapy [1-3]. In non-malignant cells, these efflux pumps evolved as a complex cellular defense since they are able to recognize a number of structurally different and apparently unrelated molecules and extrude them. P-glycoprotein {ABCB1, P-gp) is one of the most known and studied efflux pumps associated with MDR in cancer cells. It is responsible for causing resistance to several cytotoxic drugs such as the vinca alkaloids, podophyllotoxins and anthracycline derivatives [4-7]. Hence, research on selective and potent P-gp reversal agents that act as adjuvants, when coadministered with anticancer drugs, represent a valuable approach to avoid the chemotherapy failure [4-6, 8].

This ABC transporter is a pseudosymmetrical molecule that features four domains, two transmembranar domains and two nucleotide-binding domains. The latter is responsible for the ATP-binding and hydrolysis and consequently for the generation of conformational changes of the protein. The transmembranar domains are responsible for the transport and recognition of substrates across the membrane. The availability of high resolution structures of P-gp-ligand is essential for understanding the

molecular basis of its function and also for structure-based design studies [9, 10]. However, progress in this area has been slow, because the detailed drug-transporter interactions are still not clear, into a certain extent due to a lack of detailed structural information.

Over the last few years, X-ray structures have been determined for bacterial ABC transporters (e.g.

models were used for investigating human P-gp-ligand interactions. However, the tran smernbranar domains of these bacterial homologues share only 20o/o identity sequence with human P-gp. Meanwhile, in 2009 the structure of the first murine ABC-transporter was published, showing for the first time an ABC-transporter complexed with a ligand, In spite of having only a resolution of 3.8 A, it represents a huge step for structure-based studies on this transporter [10-12]. In this structure, the identification of three overlapping drug-binding sites {with residues Phe692 and Va1978 common to all three binding sites) is a valuable information that enables the utilization of computational docking techniques in the prediction of the binding affinity between ligands and the transporter [10, 11]. Through the evaluation of corresponding docking scores and visual examination of the poses inside the drug-binding pocket (DBP), the preferred orientation for a certain inolecule can be evaluated and, ultimately, the identification of the most favorable drug-binding sites {DBS) can lead to a better understanding of the modulatory effect that takes place on this transporter.

Classical ligand-based approaches, such as functional biological assays added to pharmacophore mapping and structure-activity relationships can also be a stepwise tool to select novel molecules with MDR reversal activity. Recently, we have proposed a 4-point pharmacophore comprising three hydrophobic and one hydrogen bond acceptor points as important features for recognition of different molecules by the P-gp drug-binding sites [13]. The design of this pharmacophore combined the structural features of macrocyclic diterpene derivatives isolated from with the majority of compounds known to possess anti-MDR activity.

In previous studies, we have evaluated several macrocyclic and rearranged diterpenes, isolated from potential as P-gp modulators on human mouse lymphoma cells. Most of these compounds showed to be very strong modulators of P-gp MDR, as well as exhibited synergistic interactions with cytotoxic drugs [14-19].

To pursue our search for effective P-gp-mediated MDR modulators and aiming to understand some structure-activity relationships, it is important to test the referred compounds in human cancer cell lines, because MDR modulators that present good inhibitory rates in these cells may have potential clinical application. The present study reports the MDR modulating activity of lathyrane, jatrophane and rearranged polycyclic jatrophane diterpenes on human colon adenocarcinoma MDR cell line (COLO 320 MDR). Moreover, drug combinations between doxorubicin and the most active compounds are also presented. Combining the experimental data with pharmacophore design, some structure-activity relationships are also reported.

2.

2.1.

Sixteen compounds Figs. {1-3) were tested for their potential as MDR modulators. The macrocyclic lathyrane-type diterpenes: latilagascenes B (1), and D {2) were isolated from Latilagascenes H (3) and I (4) were prepared through acylation reaction of latilagascene B [17]. The macrocyclic jatrophane-type diterpenes: pubescenes A (5), C (6) and D (7), euphopubescenol (8) and euphopubescene (9} were isolated &om Vahl [14, 20], Pepluanin D

rearranged polycyclic jatrophane diterpenes: 1,5,14-triacetoxy-3-benzoyloxy-segetalol and 1P,Sa,14u,17u-tetraacetoxy-3P-benzoyloxy-l SP-hydroxy-9-oxo-paraliane Euphoportlandol B L. [18, 19, 22]. All compounds were dissolved in dimethyl sulfoxide {DMSO; Sigma-Aldrich)

The human colon adenocarcinoma COLO 205 and COLO 320

MDR cell lines were cultured in RPMI 1640 medium supplemented with 10'/0 heat-inactivated fetal bovine serum, 2 mM L-glutamine, 1 mM Na-pyruvate and 100 mM Hepes. The cell lines were incubated in a humidified atmosphere (5'lo CO<, 95'to air) at 37 'C. The serm-adherent human colon cancer cells were detached with 0.25'/o

trypsin and 0.02'/0 EDTA for 5 min at 37 'C.

The antiproliferative effects of the compounds were tested in a range of increasing concentrations (0.26 — 133 pM), using COLO 320 MDR cell line as experimental model. The cells were distributed into 96-well flat bottom microtiter plates at a concentration of lx10 /mL with a final volume of 100 pL of medium per well. The different concentrations of each compound were added into duphcate wells. The plates were incubated at 37 'C for 72 h; at the end of the incubation period, 15 pL of MTT solution (from a 5 mg/mL stock) was added to each well and incubated for 4 h. Then, 100 pL of SDS (10%) solution was measured into each well and the plates were further incubated overnight at 37 'C. The cell growth was determined by measuring the optical density (OD) at 550 nm (ref. 630 nm) with a Dynatech MRX vertical beam ELISA reader. The percentage of inhibition of cell growth was determined according to this equation:

100 x 100

The cells were adjusted to a density of lx10 /mL, resuspended in serum-free RPMI 1640 medium and distributed in SOO pL aliquots. The test compounds were added at 2 and 20 pM; verapamil (positive control) at 22 pM and DMSO at 2 /o as solvent control. The samples were incubated for 10 min at room temperature, and then 10 pL {5.2 pM final concentration) of rhodamine-123 were added to the samples, followed by an incubation of 20 min at 37 'C. Finally, the samples were washed twice, resuspended in 500 pL phosphate-buffered saline (PBS) and were analyzed on a Becton Dickinson and Company FACScan flow cytometer (Franklin Lakes, USA). The histograms were evaluated regarding the mean fluorescence intensity, the standard deviation and the peak channel of 10,000 individual cells belonging to the total and the gated populations. The fluorescence activity ratio (FAR) was calculated on the basis of the measured fluorescence values (FL-1) of treated/untreated resistant cell line (COLO 320 MDR) and the sensitive cell line (COLO 205):

2.5.

A checkerboard microplate method was applied to study the effects of drug interactions between the selected diterpenes that presented a significant MDR reversal activity and doxorubicin on COLO 320 MDR cells. The dilutions of doxorubicin (14.7 — 0.1 pM) were made in a horizontal direction and the dilutions of resistance modifiers (150 — 4.7 pM) vertically in a microtiter plate to a final volume of 200 pL of medium per well. The cells were distributed into plates at a concentration of 4x10 /mL per well and were incubated for 48 h, under the standard conditions, The cell growth rate was determined after MTT staining, as it was already described above.

Drug interactions were evaluated according to Chou [23]. The effect of potentiation of doxorubicin's activity was evaluated as a fold effect.

ID50 values were obtained by best fitting the dose-dependent inhibition curves {GraphPadPrism5 program, GraphPad Inc., United States}. Statistical analysis was done by Student's t-test analysis < 0.01 was indicative of very significant difference). K-means clustering and Principal Component Analysis (PCA) were performed with Tanagra version 1.4.41 (http: //eric. univ-lyon2.fr/-ricco/tanagra/ en/tanagra.html). The K-means clustering analysis subdivides the compounds (those with the nearest mean) into groups that exhibit high degree of similarity. Principal Component Analysis simplifies the analysis by considering a smaller number of linear combinations of large number of variables. The Principal Component Analysis decomposition algorithm ensures that the first principal component explains the maximal amount of variance of the original data; the second principal component explains the maximal remaining variance in the data subject to being orthogonal to the first one [24, 25].

2.7.

Based on the pharmacophore previously published [13], an additional pharmacophoric point was searched in the most active molecules, in order to differentiate specific characteristics responsible for the activity of the compounds described in the present paper. This additional point was defined through a

alignment with emphasis on aromatic and hydrogen-bond acceptor features. The new pharmacophoric point was defined in order to increase the detection percentage of the most active molecules (FAR > 1 at 2 11,M).

18

10

»IIIH

II

3

16

18

9

16IS

II cn6

19

18

19

18

19

18

19

18

IP,5o„14c,17a-tetracetoxy-3P-

Docking calculations were performed with the murine P-glycoprotein obtained from the Protein Data Bank (ID: 3G60) [12]. The murine and human P-gp sequence share 100% identity for the residues inside the DBP. Verapamil, doxorubicin and all ligands structures were built in MOE 2010.10 [26], minimized with the MMFF94x force-field and exported as mol2 files for further docking procedures in GOLD Suite (GOLD v.5.1) [27-29]. The drug-binding pocket was defined through the selection of 33 amino acids obtained from the QZ59-RRR, QZ59-SSS and Verapamil binding sites as identified by Aller generated through a genetic algorithm and evaluated with the GoldScore fitness function, only retaining the best 10 docking poses for each molecule. Visual evaluation of the docking poses were made in MOE.

In order to investigate the potential of as MDR reversal agents on human colon adenocarcinoma cells, three sets of structurally related diterpenes that include macrocyclic lathyranes and rearranged polycyclic jatrophane derivatives were studied.

The rhodamine-123 exclusion assay was employed to assess the inhibitory effect of compounds COLO 320 MDR cells. Two concentrations (2 and 20 pM) were used in the experiments, and verapamil, a well-known modulator, was applied as positive control. Results for MDR-reversal activity are summarized in Table

accumulation (FL-1) is positively correlated with the increasing concentration of the compounds, shying a dose-dependent inhibitory effect of P-gp mediated efflux. It is considered that, when the fluorescence activity ratio (FAR) values are higher than 1.0, MDR reversal has taken place [30], which means that all the tested compounds had MDR reversal activity at 20 pM and this is also observed at 2 pM for compounds 1-4 and

To investigate the structure-activity relationships, some physico-chemical properties of compounds were submitted to statistical analysis, by combining two methods: K-means clustering and Principal Component Analysis [24, 25]. K-means clustering has defined two main clusters, I and II, which represent 64.7 and 35.7 % of the samples {Table 2}. The compounds

grouped compounds with good MDR reversal activity (FAR = 3.41 1.42 at 2 pM and 4.86 + 0.63 at 20 lLM) and higher lipophilic

characteristics (log Component Analysis also showed a strong correlation between FAR values and log reported by several authors that lipophilicity appears as general requirement for MDR related drugs, since the interactions of the compounds with the transporter most probably take place within the membrane rather than in the intracellular compartment, which is in line with the hydrophobic vacuum cleaner mechanism [10, 31, 32].

Interestingly, lathyrane-type compounds together (Table 2). From this set, latilagascene D {2) that possesses an aromatic ring at C-16 and a hydroxyl group at C-3 was the one that showed the highest inhibitory effect at the two concentrations tested. Compounds 1-4 only differ in the substitution pattern of ring

Table

COLO 205 negative control 195.125453.16

COLO 320 MDR negative control

DMSO control

307.38194.14

185.4810 ltL 192.5508.92 0.82

Lati 1 agascene 8 942.56185.70509,8720 4.01

186.41 1003.05

1329.17

511.81 4.27

Lati 1 agascene D (2)191 02502.6620

482.58 188.5

Lati 1 agascene H (3)505.9620

492.93 3.66

Lati 1 agascene 1 (4)200.21504.20 4.2220

334.56227.42527.40

Pubescene A (5)224.25 5.06517.3220

137.72222.92521.16

Pubescene C (6)371,68

0.84141.73216.75515.43

Pubescene D (7)3.04513.54217.18514.2620

503.78

Euphopubescenol 1.49252.82224.99521.2120

0.51215.71511.27

522.30Euphopubescene (9)

2.35397.87226.41

259.55185.03514.62

Pepluanin D 186.1920

2.49

Tuckeyanol A 5.041183.68509.8320

2.13516.35

Tuckeyanol 8 5.011175.83191.28527.0720

0.95223.86183.00506.12

1,5,14-triacetoxy-3-benzoyloxy-segetalol 2.15

219.35179.24509.79

Euphoportlandol 8 (14) 873.25497.3920

200.37552.91

5-acetoxy-3-benzoyloxy-segetalol (15) 147403.26191.53538.2220

0.81222.25196.41541.741P,5a,14a,17a-tetraacetoxy-3I)-benzoyloxy-15Il-hydroxy-9-oxo-paraliane (16) 2.14587.08198.45549.5320

1308.64 5.57200.59512.80Verapamfl

'FAR (fluorescence activity ratio) values were calculated by using the equation given in the Materials and Methods section; FSC: Forward scatter count of cells in the samples; 'SSC: Side scatter count of cells in the samples; FL-1: Mean fluorescence intensity of the cells.

A and from the analysis of these results it is possible to reinforce the importance of the aromatic ring at C-16 for the MDR reversal activity. Latilagascene H (3) and I {4) were obtained through acylation of the free hydroxyl groups (C-16 and C-3) of latilagascene 8 butanoyl groups at those positions, respectively. Although this structural modification did not significantly affect the inhibitory activity at 20 pM, the derivatives were more effective at 2 pM (FAR = 3.69 and 3,66, respectively) than latilagascene 8 (FAR = 1.28).

In spite of having the jatrophane scaffold, pubescene A (5) was also grouped in cluster II due to its log the other compounds with the same skeleton {log 3.39 to 5.75),

Diterpenes 6-12 were grouped in cluster I (Table 2), which was characterized by compounds with higher topological polar surface area {151.53 + 22.24), high number of hydrogen bond acceptors (11 + 2), molecular weight (611.06 + 59.98) and molecular volume (563.28 + 51.87). Regarding MDR reversal activity, this cluster

1020

Cluster I

Tuckeyanol A 168.824.51698,81 63,7312

4.01Tuckeyanol 5.01168.82 2.13 62.47

Euphoportlandol 3.72189.0513 44.79

Pubescene D (7)

Pepluanin D

3.04116.215.75511.58

538.38 44.792.98

0.51Euphopubescene 51.78157.83580.20622.71

Pubescene C (6) 116.21540.65 49.76511.58

1,5,14-triacetoxy-3-benzoyloxy-segetalol 2.15142.525.10598.69 54.7310

1II,Sa,14a, I 7u-tetraacetoxy-3P-benzoyloxy-I 5 I)-

hydroxy-9-oxo-paraliane 55.95168.82 2.14591.05656.73

49.35159.593.64Euphopubescenol (8) 1.49558.55

130.37 47,581.475-acetoxy-3-benzoyloxy-segetalol 473.26514.62

Cluster II

Lati I agascene D (2) 93.20533.78 64.36

65.43Verapamil 5.574.55454.30454.61

48.22122.29548.09Pubescene A 6.46582.69

Lati1 agascene H (3)61.1765.148.60

66.644.226S.14 3,66Lati 1 agascene I S78.29576.82

50.854.011.28Lati 1 agascene B 4.75451,41

Grouping variables: physico-chemical properties (number of hydrogen bond acceptors — HBA- and donors — HBD-, molecular weight, molecular volume, ocul/water partition coefficient — log

comprises compounds without activity at 2 pM, except for tuckeyanols A

human most of them shown very high FAR values [14-19]. The differences in the inhibitory potency of the compounds between mouse lymphoma MDR and COLO 320 MDR cells is related with different levels of P-gp expression, lower in COLO 320 MDR cells [33]. Nevertheless, comparing the former results with those presented here, it is interesting to note that the most efficient compounds on the artificially cell line were also the ones that are more efficient on COLO 320 MDR cell line. This is a very

important finding, as MDR modulators that present good inhibitory rates in human cell lines, may have potential for further studies towards clinical application.

Taking in account our previous studies and the results in the MDR reversal experiments of the present work, compounds 1-4 and

doxorubicin on COLO 320 MDR cells. Drug interactions were studied according to Chou [23]. In this method there are several basic requirements for determination of synergism or antagonism. One of them is that each drug alone should have a dose-effect relationship. If one of the two drugs has no effect by itself, then synergism/antagonism cannot be determined. Instead, potentiation or inhibition should be determined as a fold effect. In the present work, the tested compounds did not exhibit cytotoxicity by themselves {> 150 IIM; data not shown), but increased the effect of doxorubicin in combination. At the range of concentrations tested, all the assayed compounds potentiated doxorubicin's activity up to six times (Fig, 4). The IDso of doxorubicin alone was 2.68 + 0.49 pM and the ID>o of doxorubicin in combination ranged from 1.35— 0.31 pM (IDso of combinations were statistically different from

IDSQ of doxorubicin alone, P < 0.01}. Duarte same compounds in combination with epirubicin on mouse lymphoma cells and verified that all of them synergistically enhance the effect of the antitumor drug. Similarly to the MDR reversal effect, the difference on the magnitude of action may be related with a higher sensitivity of mouse lymphoma cells [33]. These findings point out that the investigated diterpenes are good MDR reversal agents and combined with drugs with neoplastic activity may enhance their effect and consequently decrease the resistance of cancer cells.

Recently there were advances towards the elucidation of the 3D structure of P-gp, this was especially important when the structure of the first mamalian ABC-transporter (murine P-gp) was published, although not in high resolution [12]. Nevertheless, there are many questions regarding the drug/transporter interactions that still need to be addressed. The interactions within the drug binding pocket account for such questions [10]. Therefore, pharmacophore modeling is an approach to study the possible ligand/receptor interactions.

The authors have already published an improved pharmacophore model for P-gp modulators based on macrocyclic diterpenes isolated from detects all of the compounds present in the current study. In order to differentiate speciftc characteristics responsible for the activity, compounds 1-16 were further evaluated. The addition of an aromatic moiety, thus creating a new five-point pharmacophore {Fig. 5), allowed to select the most active compounds (FAR > 1 at 2 pM), The aromatic moiety was defined with a 2.0 A radius, having minimum and maximum distances of 5.5 and 10.1 A from any other pharmacophoric point, in agreement with previous studies [13, 34].

This new pharmacophore detected, successfully, the most active molecules {2-5,

points located in the macrocyclic ring. However, in latilagascenes D (2) and H (3) {with lathyrane scaffold}, only one hydrophobic point remains at the macrocyclic ring. For these molecules, the HBA detection occurs at the carbonyl belonging to the ester function at C-16, one hydrophobic point in a methyl group {C-17) and another in the cyclopentane ring. The aromatic point is located in the benzoyl ring at C-15 (Fig. 6A).

In the case of the detection pattern of latilagascene I (4), the hydrogen bond acceptor was detected in the carbonyl at C-16 {as for molecules 2 and 3) or at C-3 (Fig. 6B}. Interestingly, when the detection involves the carbonyl at C-3, one of the hydrophobic points moves from the ring A to the double bond of the cinnamoyl group at C-15, reinforcing the importance of the substitution pattern at this position.

Regarding the jatrophane scaffold, the most active compound, pubescene A (5), was detected due to the existence of an acetyl group at C-15 that acts as an hydrogen bond acceptor (Fig. 7A), also corresponding to an alteration in the previously defined 4-point pharmacophore pattern. Pubescene C (6), although similar to pubescene A (5}, was not detected by this pharmacophore. An essential pre-requisite previously defined by Suzuki [34] establishes a minimum distance of 5 A between the hydrogen bond acceptor and the nearest hydrophobic point. However, in molecule 6, this distance becomes less than 5 A due to the hydrolysis of the acetyl to a hydroxyl group. Therefore, despite the existence of several other hydrogen bond acceptor groups, the 4-point pharmacophoric pattern is maintained (HBA at C-3 and three hydrophobic points in the macrocyclic ring) excluding any other possible conformation that could include the aromatic point (Fig. 7B),

The main difference between the tuckeyanols A {11) and B (12) pharmacophoric pattern is the detection of the exocyclic double bond at C-17 as an hydrophobic point in tuckeyanol B, Tuckeyanol A maintains one hydrophobic point at the macrocyclic ring, while moving another to the 2-methylbutyl substituent at C-7 and the third to the ciclopentane ring, The carbonyl group at C-3 is detected as the hydrogen bond acceptor (Fig. SA}. In tuckeyanol B a different conformation is detected, with the hydrogen bond acceptor at C-9 ketone function) and two hydrophobic points at C-16 and C-17, respectively. The third hydrophobic point remains at the rnacrocycle (Fig. SB). The main difference between the molecules is due to the fact that, in tuckeyanol B, the substituent at C-7 occupies a smaller volume allowing the acetyl group at C-5 to move closer. This forces conformation changes on the macrocyclic ring that are responsible for the alteration in the detection pattern.

5

4

pp

0

Compounds

eDoxo- 150 pM IDoxo — "5 uM BDoxo — 3;,50IiM

Fig. (4). Fold potentiation effect of combination of compounds 1-4 and 11-12 (tested in six different concentrations) with doxorubicin, in COLO 320MDR cells. ID~0 of compounds alone was greater than 150 pM and ID~0 of doxorubicin alone 2.68 pM + 0.49 (n=3). The fold effect was calculated as the ratio between the ID~0 of doxorubicin alone, and the compounds 1-4, 11-12 and doxorubicin. All the tested compounds significantly potentiated doxorubicin's action (ID~0 of combination were statistically different from ID~0 of doxorubicin alone,

the substituents of the diterpenic scaffold that contribute to the P-gp inhibitory activity. In addition, some of the most active compounds found in literature such as XR9576 [35] or LY335979 [34] are still detected. The accuracy test of this pharmacophore was assessed using the previously compiled inactive database [11], not detecting any molecule.

The analysis of the detection pattern revealed interesting data on the scaffold substituents. The previously defined 4-point pharmacophore [11] detects diterpenes due to the presence of an hydrogen bond acceptor group at C-3 with the three hydrophobic

Fig. (5). Specifications of the distances (hydrophobic) and HBA (hydrogen bond acceptor).

(hydrophobic) and HBA (hydrogen bond acceptor}.

(hydrogen bond acceptor).

HBA (hydrogen bond acceptor).

In compounds of the macrocyclic ring alters completely the detection pattern. The two detected compounds, 1,5,14-triacetoxy-3-benzoyloxy-segetalol

9-oxo-paraliane (16), share a common detection pattern with the hydrogen bond acceptor on the acetyl group at C-l, enabling the

detection of the benzoyl ring at C-3. One hydrophobic point is located in the polycyclic scaffold, with the remaining two being &equently observed at C-20 or in the acetyl group at C-5. Moreover, it is also observed additional detection patterns, in which the hydrogen bond acceptor can be located in the acetyl group at C-14 (compound

yellow. Residues from QZ59-SSS (green), QZ59-RRR (red) and verapamil (light blue) binding sites are also represented, in conjunction with residues common to all sites (magenta). Cavity surface was defined as green in lipophilic and pink in hydrophilic regions.

aromatic rings in compound 2 {one at C-15 and another in C-2) allows the detection of the HBA on carbonyl groups at C-3 or C-15, respectively, showing a synergistic effect between two aromatic and two HBA points. Docking studies also confirmed that the most active molecules dock in the same spatial region as verapamil, additionally validating the importance of the aromatic moiety of the proposed pharmacophore through the identification of an hydrophobic cavity suitable for hydrophobic and x — it stacking interactions between aromatic rings in the substituents and the sidechains of the transporter.

The authors confirm that this article content has no conflicts of interest.

This study was supported by FCT (Fundaqao para a Ciencia e a Tecnologia, Portugal), project (PTDC/QUI-QUI/099815/2008). Mariana Reis acknowledges FCT for her PhD grant (SFRH/BD/ 72915/2010).

[13

[3]

[4]

Regarding docking studies, our study shows that the diterpenoid compounds interact at the same place as verapamil (Fig. Verapamil aromatic rings dock near residues Ile336/Phe339 and Phe728/974, corresponding to the intersection between QZ59-RRR/QZ59-SSS binding sites (Fig. 9). Our docking procedure obtained consistent Gold Score values above 60 for verapamil and the most active molecules (compounds 2-4, 11 and 12), with all other less active molecules displaying smaller docking scores. The most active molecule, latilagascene D, is positioned in a similar region with the aromatic ring at C-2 in the same area as verapamil (near Phe728/974) and the ring from the cinnamoyl group at C-15 near Gln721/Phe724. Regarding tuckeyanol A, the top ranked pose is also at the intersection of the three DBS, although in a lower position when compared with verapamil (Fig. 9). However, the second ranked pose {score 58.90) also displays the aromatic ring at C-14 in a similar position as latilagascene D.

From the docking poses of the most active molecules it is possible to identify a hydrophobic cavity formed by phenylalanines (residues 332, 724, 728 and 974) and Va1978, where aromatic rings of residues lateral sidechains are found. This result is in accordance with the need for the extra pharmacophoric point, stressing the importance of an aromatic moiety in the inhibitory activity.

According to these and previous results, macrocyclic diterpenes of the jatrophane and lathyrane-type have great potential as MDR reversal agents and combined with drugs with neoplastic activity may enhance their effect and consequently decrease the resistance of cancer cells. Therefore, macrocyclic diterpenes are promising as lead compounds for the development of effective P-gp modulators.

In order to increase the selectivity of the previously published 4-point pharmacophore, an additional aromatic point was defined. This means that the presence of an aromatic moiety in the molecule is important for an increased P-gp affinity. An additional HBA connected to the oxygen at C-15 is also important, particularly in the jatrophane scaffold. In lathyrane scaffold, the presence of two

Perez-Tomas, R. Multidrug resistance: retrospect and prospects in anti-cancer drug treatment. Szakacs, G.; Paterson, J.K.; Ludwig, J.A.; Booth-Genthe, C.; Gottesman, M.M. Targeting multidrug resistance in cancer. Rat. Rev. Thomas, H.; Coley, H.M. Overcoming multidrug resistance in cancer: an update on the clinical strategy of inhibiting p-glycoprotein. Sarkadi, B.; Homolya, L.; Szakacs, G.; Varadi, A. Human Multidrug Resistance ABCB and ABCG Transporters: Participation in a Chemoimmunity Defense System. 1 179-1236.

Gottesman, M.M.; Fojo, T.; Bates, S,E. Multidrug Resistance in Cancer: Role of ATP-Dependent Transporters. 2002, 2, 48-58.

1024

l71

C21]

[22]I:9]

[23j[10]

[24]

P5]

[12j[26]

[27]

[13j

I28]

[14j

[I 5]PO]

[31][16]

P2l

[17]I33l

[181P4l

P51[19]

Baumert, C., Hilgeroth, A., Recent advances in the development of P-gp inhibitors. Ambudkar, S.V., Kimchi-Sarfaty, C.; Sauna, Z.E.; Gottesman, M.M. P-glycoprotein: from genomics to mechanism. 2003, 22, 7468-7485.

Baguley, B.C. Multiple drug resistance mechanisms in cancer.

Stouch, T,R.; Gudmundsson, O. Progress in understanding the structure — activity relationships of P-glycoprotein. Adv.

Klepsch, F.; Ecker, G.F. Impact of the recent Mouse P-Glycoprotein Structure for Structure-Based Ligand Design. Mol.

Gutmann, D.A.P,; %'ard, A.; Urbatsch„ I.L.; Chang, G.; van Veen, H.%. Understanding polyspecificity of multidrug ABC transporters: closing in on the gaps in ABCB1.

Aller, S.G.; Yu, J.; Ward, A.; Weng, Y.; Chittaboina, S.; Zhuo, R.; Harrell, P.M.; Trinh, Y.T.; Zhang, Q.; Urbatsch, I.L.; Chang, G. Structure of P-Glycoprotein Reveals a Molecular Basis for Poly-Specific Drug Binding. Ferreira, R.J.; dos Santos, D.J.V.A.; Ferreira, M.-J.U.; Guedes, R.C, Toward a Better Pharmacophore Description of P-Glycoprotein Modulators, Based on Mac rocyclic Diterpenes fromEuphorbiaSpecies. J. Chem. Valente, C.; Ferreira, M.J.; Abreu, P.M.; Gyemant, N.; Ugocsai, K.; Hohmann, J.; Molnar, J. Pubescenes, jatrophane diterpenes, from Euphorbia pubescens, with multidrug resistance reversing activity on mouse lymphoma cells. Duarte, N.; Gyemant, N.; Abreu, P,M.; Molnar, J.; Ferreira, M.J. New macrocyclic lathyrane diterpenes, from Euphorbia lagascae, as inhibitors of multidrug resistance of tumour cells. 2006, 72, 162-168. Duarte, N,; Varga, A.; Cherepnev, G.; Radics, R.; Molnar, J.; Ferreira, M.J. Apoptosis induction and modulation of P-glycoprotein mediated multidrug resistance by new macrocyclic lathyrane-type diterpenoids. Duarte, N.; Jardanhazy, A.; Molnar, J.; Hilgeroth, A.; Ferreira, M.J. Synergistic interaction between P-glycoprotein modulators and epirubicine on resistant cancer cells. 9323-9330.

Madureira, A.M.; Ferreira, M J.; Gyemant, N.; Ugocsai, K.; Ascenso, J.R.; Abreu, P.M.; Hohmann, J., Molnar, J. Rearranged jatrophane-type diterpenes from euphorbia species. Evaluation of their effects on the reversal of multidrug resistance. 2004, 70, 45-49. Madureira, A.M.; Gyemant, N.; Ascenso, J.R.; Abreu, P.M.; Molnar, J.; Ferreira, M.J, Euphoportlandols A and B, tetracylic diterpene polyesters from Euphorbia portlandica and their anti-MDR effects in cancer cells. J. stat.

Valente, C.„Pedro, M.; Ascenso, J.R.; Abreu, P.M.; Nascimento, M.S.; Ferreira, M.J. Euphopubescenol and euphopubescene, two new jatrophane polyesters, and lathyrane-type diterpenes from Euphorbia pubescens. Duarte, N.; Lage, H.; Ferreira, M.J. Three new jatrophane polyesters and antiproliferative constituents from Euphorbia tuckeyana. Ferreira, M.-J.U.; Madureira, A.M.; Ascenso, J.R. A Tetracyclic diterpene and triterpenes from Euphorbia segetalis. 1998, 49, 179-183. Chou, T.C. Drug Combination Studies and Their Synergy Quantification Using the Chou-Talalay Method. 70, 440-446. Willett, P.; Barnard, J,M,; Downs, G.M. Chemical Similarity Searching. J. Chem. Tong, W.; Welsh, %.J.; Shi„L.; Fang, H.; Perkins, R. Structure-activity relationship approaches and applications. Chem., 2003, 22, 1680-1695.

Computing Group: Montreal, Canada, 2010. Nissink, J.%, Murray, C.; Hartshorn, M.; Verdonk, M.L,; Cole, J.C.; Taylor, R A new test set for validating predictions of protein-ligand interaction. Cole, J.C.; Nissink, J.W.M.; and Virtual Screening with GOLD Discovery, editors Shoichet, B, and Alvarez, J., Taylor 4 Francis CRC Press, Boca Raton, Florida, USA, 2005. Verdonk M.L.; Cole J,C.; Hartshorn M.J.; Murray C.W,; Taylor R. Improved protein-ligand docking using GOLD. 609-623.

Voigt, B.; Coburger, C.; Monar, J.; Hilgeroth, A, Structure — activity relationships of novel N-acyloxy-1,4-dihydropyridines as P-glycoprotein inhibitors. Wiese, M.; Pajeva, I.K, Structure-activity relationships of multidrug resistance reversers.

Srinivas, E.; Murthy, J.N.; Rao, A.R.; Sastry, G.N. Recent advances in molecular modeling and medicinal chemistry aspects of phospho-glycoprotein. Engi, H.; Sakagami, H.; Kawase, M., Parecha, A.; Manvar, D.; Kothari, H.; Adlakha, P.; Shah, A.; Motohashi, N.; Ocsovszki, I.; Molnar, J. Tumour-specific cytotoxicity and MDR-reversal activity of dihydropyridines. Suzuki, T.; Fukazawa, N.; San-nohe, K.; Sato, %.; Yano, O.; Tsuruo, T. Structure — Activity Relationship of Newly Synthesized Quinoline Derivatives for Reversal of Multidrug Resistance in Cancer. J. Med. Chem., 1997, 40, 2047-2052.

Mistry, P.; Stewart, A.J., Dangerfield, W.; Okiji, S.; Liddle, C.; Bootle, D.; Plumb, J.A.; Templeton, D.; Charlton, P.

novel potent modulator, XR9576.