chemometric studies on potential larvicidal compounds against aedes aegypti

11
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/236910787 Chemometric Studies on Potential Larvicidal Compounds Against Aedes Aegypti ARTICLE in MEDICINAL CHEMISTRY (SHĀRIQAH (UNITED ARAB EMIRATES)) · MAY 2013 Impact Factor: 1.36 · DOI: 10.2174/15734064113099990005 · Source: PubMed CITATIONS 2 READS 141 6 AUTHORS, INCLUDING: Luciana Scotti Universidade Federal da Paraíba 51 PUBLICATIONS 133 CITATIONS SEE PROFILE Marcus T Scotti Universidade Federal da Paraíba 76 PUBLICATIONS 528 CITATIONS SEE PROFILE Socrates Cavalcanti Universidade Federal de Sergipe 52 PUBLICATIONS 689 CITATIONS SEE PROFILE Francisco Jaime Bezerra Mendonça Junior Universidade Estadual da Paraíba 45 PUBLICATIONS 111 CITATIONS SEE PROFILE All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately. Available from: Marcus T Scotti Retrieved on: 04 February 2016

Upload: independent

Post on 11-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/236910787

ChemometricStudiesonPotentialLarvicidalCompoundsAgainstAedesAegypti

ARTICLEinMEDICINALCHEMISTRY(SHĀRIQAH(UNITEDARABEMIRATES))·MAY2013ImpactFactor:1.36·DOI:10.2174/15734064113099990005·Source:PubMed

CITATIONS

2

READS

141

6AUTHORS,INCLUDING:

LucianaScotti

UniversidadeFederaldaParaíba

51PUBLICATIONS133CITATIONS

SEEPROFILE

MarcusTScotti

UniversidadeFederaldaParaíba

76PUBLICATIONS528CITATIONS

SEEPROFILE

SocratesCavalcanti

UniversidadeFederaldeSergipe

52PUBLICATIONS689CITATIONS

SEEPROFILE

FranciscoJaimeBezerraMendonçaJunior

UniversidadeEstadualdaParaíba

45PUBLICATIONS111CITATIONS

SEEPROFILE

Allin-textreferencesunderlinedinbluearelinkedtopublicationsonResearchGate,

lettingyouaccessandreadthemimmediately.

Availablefrom:MarcusTScotti

Retrievedon:04February2016

Send Orders of Reprints at [email protected]

Medicinal Chemistry, 2013, 9, 000-000 1

1573-4064/13 $58.00+.00 © 2013 Bentham Science Publishers

Chemometric Studies on Potential Larvicidal Compounds Against Aedes Aegypti

Luciana Scotti*a, Marcus Tullius Scottib, Viviane Barros Silvac, Sandra Regina Lima Santosc, Sócra-tes C.H. Cavalcantic and Francisco J.B. Mendonça Juniora

aState University of Paraíba, Biological Science Department, Laboratory of Synthesis and Drug Delivery, 58070-450, João Pessoa, PB, Brazil; bFederal University of Paraíba, Department of Engineering and the Environment, Campus IV; 58297-000, Rio Tinto, PB, Brazil; cFederal University of Sergipe, Pharmacy Department, Medicinal Chemistry Labora-tory, Av. Marechal Rondon S/N, Rosa Elze, 49100-000, São Cristóvão, Sergipe, Brazil

Abstract: The mosquito Aedes aegypti (Diptera, Culicidae) is the vector of yellow and dengue fever. In this study, chemometric tools, such as, Principal Component Analysis (PCA), Consensus PCA (CPCA), and Partial Least Squares Regression (PLS), were applied to a set of fifty five active compounds against Ae. aegypti larvae, which includes terpenes, cyclic alcohols, phenolic compounds, and their synthetic derivatives. The calculations were performed using the VolSurf+ program. CPCA analysis suggests that the higher weight blocks of descriptors were SIZE/SHAPE, DRY, and H2O. The PCA was generated with 48 descriptors selected from the previous blocks. The scores plot showed good separation be-tween more and less potent compounds. The first two PCs accounted for over 60% of the data variance. The best model obtained in PLS, after validation leave-one-out, exhibited q2 = 0.679 and r2 = 0.714. External prediction model was R2 = 0.623. The independent variables having a hydrophobic profile were strongly correlated to the biological data. The inter-action maps generated with the GRID force field showed that the most active compounds exhibit more interaction with the DRY probe.

Keywords: Aedes aegypti, chemometry, CPCA, descriptors, PLS, VolSurf.

1. INTRODUCTION

Dengue is a viral disease caused by a Flavivirus transmit-ted by the mosquito Aedes aegypti (Diptera:Culicidae). Its symptoms vary from mild fever, to life-threatening severe dengue [1] which may progress to a shock syndrome [2]. The propagation of dengue is currently a public health threat, particularly in tropical and subtropical countries [3]. The disease is caused by the dengue virus, of which four sero-logically different serotypes are described (DV-1, DV-2, DV-3, and DV-4). The World Health Organization (WHO) reports that dengue is a leading cause of illness and death in the tropics and subtropics, causing a flu-like infection, which affects as many as 70 million people every year [4-6]. Since there are no effective treatments for this disease, the most effective way to control the virus outbreak is to avoid vector spreading, mainly by the use of larvicides. Organophos-phates, such as, temephos, have been used as larvicides in several countries since the 1960’s. However, resistance to pesticides [7] has guided research to find new methods in-tended to control Ae. aegypti propagation.

Several approaches have been employed to find new larvicidal candidates. Plant derived products are undoubtedly *Address correspondence to this author at the State University of Paraíba, Biological Science Department, Laboratory of Synthesis and Drug Delivery, 58070-450, João Pessoa, PB, Brazil; Tel: 55-83-3291-1805; Fax: 55-83-3291-1528; E-mail: [email protected]

the most evaluated larvicidal substances against Ae. aegypti to date. Plant extracts provide a number of secondary prod-ucts synthesized by the plants to act as natural insecticides or repellent. Phenolic acids [8], spinosyns [9], coumarins [10], essential oil monoterpenoids [11], vegetable oil [12], polya-cetylenes, phytosterols, flavonoids, sesquiterpenoids, triter-penoids [13], are within the classes of natural products with larvicidal activities.

As part of our effort to find natural products with larv-icidal activity we have evaluated a number of compounds derived from natural substances [14-16]. The essential oils of Hyptis fruticosa (Lamiaceae) Salzm., H. pectinata (La-miaceae) Poit., and Lippia gracilis (Verbenaceae) were char-acterized and evaluated against third-instar Ae. aegypti lar-vae [17]. Carvacrol was found to be the major compound in the essential oil of L. gracilis responsible for the observed activity. Additionally, other minor compounds, such as thy-mol, !-terpinene, and limonene contributed to the observed larvicidal activity [17].

These results motivated us to undertake a more detailed investigation of larvicidal compounds with the goal to study the Structure-Activity Relationships between the larvicidal activity and a set of compounds mostly found in plants [14, 15]. Subsets of aromatic, alicyclic and bicyclic compounds with a variety of substituents were previously evaluated [14-16]. The presence of lipophilic groups in aromatic rings or in hydroxyls resulted in increased potency [14]. Additionally, the presence of hydroxyls in aromatic or aliphatic rings re-

2 Medicinal Chemistry, 2013, Vol. 9, No. ?? Scotti et al.

sulted in decreased potency. Conjugated and exo double bonds in aliphatic rings appeared to increase larvicidal po-tency [15]. Moreover, replacement of double bonds by epox-ides decreased the larvicidal potency. Stereochemistry of selected compounds has as well been found to play an im-portant role on modulating the potency. Generally, lipophil-icity appears to play an important role on the larvicidal activ-ity of selected compounds [14, 15]. However, no quantitative studies have been performed to corroborate the previous findings.

Theoretical studies accomplished by applying computer-aided drug design have significantly assisted the drug dis-covery of new bioactive compounds. Technological ad-vances in the areas of structural characterization of the bio-logical macromolecules, of computational science and of molecular biology contributed to a faster and more feasible design of new molecules. Among many computational tools, chemometric studies are popular methods to model allowing the analysis of large multivariable collinear data sets [18-23].

Principal Component Analysis (PCA), Consensus PCA (CPCA), and Partial Least Squares (PLS) regression are chemometric tools used for extracting and rationalizing the information from any multivariate description of a biological system. CPCA and PCA are part of an exploratory data analysis where graphical techniques provide a maximization of insights into a data set, pointing out important variables, detecting outliers and anomalies, and developing parsimoni-ous models [22, 23].

As part of our ongoing work to study the SAR of poten-tial larvicidal agents, we investigated the molecular interac-tions of 55 selected compounds in various media (hydro-philic, hydrophobic, etc.) with the aid of chemometrics methods. These included Principal Component Analysis (PCA), Consensus PCA (CPCA), and Partial Least Squares Regression (PLS) using the VolSurf+ program. These pre-dictions provide direction with regard to the syntheses of new derivatives with improved biological activities, which can be used as insecticides or repellent with larvicidal activ-ity against the Ae. aegypti.

2. MATERIALS AND METHODS

2.1. Compounds and Larvicidal Assay

Compounds 1-5, 21-24, 27, 28, 30-33, 36, 38, 40, 41, 46, 48-55 were purchased from Sigma-Aldrich. Isopulegol (37) and neoisopulegol (43) were separated and obtained by silica gel 60 column chromatography (hexanes) from technical grade isopulegol (Dierberger – Brazil). 1,2-Carvone oxide (44) [24], RS-menthone (39) [25], limonene oxide, mixture of cis and trans (45) [26], carvacrol, thymol [27-31], and eugenol derivatives [30] were synthesized according to the literature [32]. Structures of the evaluated compounds are shown in (Fig. 1). LC50 were previously obtained by a larv-icidal assay, followed by Probit analysis of three replicates and are reported elsewhere [14,15,17].

2.2. Molecular Modelling

The three-dimensional structures were drawn using Hy-perChem 8.0 software [33] and energy-minimized employ-ing the MM+ force field without any restriction. Subse-

quently, a new geometry optimization process, based on the semi-empirical method AM1 (Austin Model 1) was per-formed [34, 35]. The optimized structures were subjected to conformational analysis using the random search method with 1,000 interactions, 100 cycles of optimization, and 10 conformers of lowest minimum energy. The selected dihe-drals were evaluated by rotation in accordance with the stan-dard (default) conditions of the program, which the number of simultaneous variations was 1 to 8. Acyclic chains were submitted to rotations from 60 to 180°, torsion rings were in the range of 30 to 120° [36, 37].

2.3. Maps

The lowest energy conformers were saved in .sdf format and imported into VolSurf+ for Windows [38, 39]. Each structure was subjected to GRID force field and interaction maps with the generation of the following "probes": H2O (light blue), O (red), N (dark blue) and DRY (green).

2.4. Chemometrics

The structures, modelled as described formerly, were used as the initial structures to calculate the molecular de-scriptors via the VolSurf+ program [38-45]. PCA, CPCA, and PLS are popular methods to model and analyze large multivariable collinear data sets. These methodologies were applied to the set of interest using the VolSurf+ software.

2.4.1. CPCA (Consensus PCA)

A preliminary exploratory analysis, CPCA, was devel-oped by considering 128 independent variables or descrip-tors. Pre-processing (autoscaling) of the data was performed and 13 blocks of descriptors were calculated.

2.4.2. PCA

With regard to the interaction of 3D structures and a GRID force field, PCA results were obtained using the de-scriptors of the high weight blocks obtained in CPCA. Forty eight molecular descriptors were selected of SHAPE, DRY and H2O blocks. The preprocessing was performed (autoscaling).

2.4.3. PLS

The training set was composed of forty-one compounds and the test set was constituted of 14 rationally selected compound (3, 7, 11, 14, 17, 23, 31, 38, 40, 45, 48, 52 and 55) [45]. The autoscaling preprocess was further applied. The PLS analysis uses the VolSurf+ descriptors as the X-block data and the larvicidal activity as the dependent vari-able. The internal validation (cross-validation test) of PLS model was performed by the leave-one-out (LOO) technique.

3. RESULTS

3.1. Experimental Larvicidal Activity

(Table 1) shows the 50% lethal concentration (LC50) and confidence interval (CI) of the investigated compounds (1–55). The LC50 values were converted to molar units and then expressed in negative logarithmic units, pLC50 (-log LC50).

Chemometric Studies on Potential Larvicidal Compounds Against Aedes Aegypti Medicinal Chemistry, 2013, Vol. 9, No. ?? 3

3.2. CPCA Thirteen blocks of descriptors with higher weights are

shown in (Fig. 2). This figure was obtained by taking into account the orthogonal properties of the CPCA, which was developed on the basis of 128 independent variables or de-scriptors.

(Table 2) represents the total variance explained using the block of descriptor with higher weight: SIZE/SHAPE, H2O and DRY. 3.3. PCA

(Table 3) shows that PC1 and PC2 accounted for more than 62% of total variance from the original data. The scores

plot exhibits satisfactory discrimination between more potent (dark grey), averaged (grey) and less potent (light grey) compounds, as presented in (Fig. 3).

3.4. PLS

The model generated in the PLS analysis, with two LVs, explained more than 72% of the total variance from the original data (Table 4). Significant statistical measures (leave-one-out cross-validation correlation coefficient, qcv

2=0.679; and regression correlation coefficient, r2=0.714) were obtained when the interactions fields were calculated by using a hydrophobic probe and two latent variables (LV) (see Fig. 4).

O O

R1

5 - R1 = OH6 - R1 = OCOCH37 - R1 = OCOCH2Cl8 - R1 = OCOCCl39 - R1 = OCOCH2CH310 - R1 = OCOPh12 - R1 = OCH2COOH

R2

38 - R2 = OH13 - R2 = OCOCH314 - R2 = OCOCH2Cl15 - R2 = OCOCCl316- R2 = OCOCH2CH317 - R2 = OCOPh18 - R2 = OCH2CH320 - R2 = OCH2COOH

2

3 4

R3

R4

11 - R3 = OH, R4 = CHO19 - R3 = CHO, R4 = OH

OH

R6

28 - R6 = OH31 - R6 = H33 - R6 = OCH3 52 - R6 = CHO

OHO

37 39

HO

43

OO O

44 45

OH

OH49

51 53

O O

50 54

OH

CHO

OCH3

55

4632

48

R7

H3COOH

34 - R7 = OCOPh35 - R7 = OH

H3CO

R5

25 - R5 = OCOCH326 - R5 = OCH2COOH29 - R5 = OCH2CH3 30 - R5 = OH42 - R5 = OCH347 - R5 = OCOCH2CH3

O(CH2)n

40 - n = 641 - n = 8

(-)-Camphene (1), (±)-camphor (2), 1,4-cineole (3), 1,8-cineole (4), carvacrol (5), carvacryl acetate (6), carvacryl chloroacetate (7), carvacryl trichloroacetate (8), carvacryl propionate (9), carvacryl benzoate (10), 2-Hydroxy-3-methyl-6-(1-methylethyl)-benzaldehyde (11), carvacrylglycolic acid (12), thymyl acetate (13), thymyl chloroacetate (14), thymyl trichloroacetate (15), thymyl propionate (16), thymyl benzoate (17), thymyl ethyl ether (18), 2-hydroxy-6-methyl-3-(1-methylethyl)-benzaldehyde (19), thymoxyacetic acid (20), 5-norbornene-2-ol (21), 5-norbornene-2,2-dimethanol (22), 5-norbornene-2-endo-3-endo-dimethanol (23), 5-norbornene-2-exo-3-exo-dimethanol (24), eugenyl acetate (25), 2-[2-methoxy-4-(2-propen-1-yl)phenoxy] acetic acid (26), borneol (27), catechol (28), 1-ethoxy-2-methoxy-4-(2-propen-1-yl)-benzene (29), eugenol (30), phenol (31), g-terpinene (32), guaiacol (33), 1-benzoate-2-methoxy-4-(3-hydroxypropyl)-phenol (34), 4-hydroxy-3-methoxy-benzenepropanol (35), isoborneol (36), isopulegol (37), thymol (38), menthone (39), nonan-2-one (40), undecan-2-one (41), 1,2-dimethoxy-4-(2-propen-1-yl)-benzene (42), neoisopulegol (43), 1,2-carvone oxide (44), limonene oxide, mixture of cis and trans (45), p-cymene (46), eugenyl propionate (47), R-carvone (48), resorcinol (49), R-limonene (50), R/S-carvone (51), salicylaldehyde (52), S-carvone (53), S-limonene (54), vanillin (55).

Fig. (1). Structures of the investigated compounds.

4 Medicinal Chemistry, 2013, Vol. 9, No. ?? Scotti et al.

Table 1. Larvicidal Activity of the Investigated Compounds

Compound Number pLC50 (CI)* Mol/L Compound Number pLC50 (CI)* Mol/L

1 2.79 (2.66 to 2.92) 29 3.40 (3.38 to 3.44)

2 2.36 (2.34 to 2.41) 30 3.35 (3.30 to 3.42)

3 2.31 (2.29 to 2.33) 31 2.69 (2.66 to 2.72)

4 2.04 (2.00 to 2.07) 32 3.39 (3.29 to 3.50)

5 3.47 (3.44 to 3.50) 33 2.84 (2.81 to 2.88)

6 3.32 (3.27 to 3.36) 34 3.28 (3.20 to 3.35)

7 3.64 (3.59 to 3.71) 35 2.05 (2.00 to 2.13)

8 3.59 (3.54 to 3.64) 36 2.41(2.46 to 2.57)

9 3.49 (3.43 to 3.56) 37 2.71 (2.68 to 2.75)

10 3.66 (3.57 to 3.72) 38 2.59 (2.38 to 2.64)

11 3.43 (3.38 to 3.49) 39 2.48 (2.44 to 2.52)

12 3.09 (3.06 to 3.12) 40 2.85 (2.80 to 2.91)

13 3.32 (3.27 to 3.37) 41 3.51 (3.43 to 3.61)

14 3.66 (3.63 to 3.69) 42 3.24 (3.22 to 3.28)

15 3.85 (3.82 to 3.90) 43 2.44 (2.39 to 2.52)

16 3.49 (3.45 to 3.53) 44 2.88 (2.85 to 2.89)

17 3.46 (3.41 to 3.49) 45 2.47 (2.43 to 2.52)

18 3.16 (2.82 to 3.79) 46 3.42 (3.38 to 3.45)

19 3.72 (3.68 to 3.75) 47 3.55 (3.50 to 3.61)

20 2.65 (2.61 to 2.69) 48 3.00 (2.94 to 3.04)

21 2.16 (2.14 to 2.18) 49 2.28 (2.24 to 2.33)

22 2.29 (2.24 to 2.34) 50 3.70 (3.64 to 3.77)

23 2.04 (1.99 to 2.09) 51 3.11 (3.09 to 3.13)

24 2.33 (2.24 to 2.42) 52 2.95 (2.90 to 3.02)

25 3.28 (3.25 to 3.32) 53 3.08 (3.06 to 3.11)

26 3.04 (2.94 to 3.09) 54 3.64 (3.62 to 3.72)

27 2.40 (2.36 to 2.46) 55 2.47 (2.41 to 2.53)

28 2.66 (2.61 to 2.70) * CI = Confidence interval.

Table 2. Explained Variance by SIZE/SHAPE, H2O, and DRY Blocks

% Explained Variance from Original Data PC

SIZE/SHAPE block H2O block DRY block

1 25.222 59.493 37.954

2 51.560 7.366 24.762

3 -3.115 2.444 8.913

4 0.511 4.658 -1.464

5 4.668 7.656 0.054

Chemometric Studies on Potential Larvicidal Compounds Against Aedes Aegypti Medicinal Chemistry, 2013, Vol. 9, No. ?? 5

Fig. (2). Plot of block weights considering PC1 and PC2. Block of descriptors: DDRY, OH2, SHAPE, Mixed descriptors (MISC), O, N1, Charge State descriptors (FU), ADME model descriptors (LOGD, LOGS), 3D pharmacophoric descriptors (TOPP), MODEL and STRUCT.

Fig. (3). Scores plot from PCA with representation of more potent (dark grey), averaged (grey) and less potent (light grey) compounds.

Fig. (4). PLS predictions found for larvicidal activity: dark grey balls represent more active compounds and light grey ball represents com-pounds that are less active compounds. The test set compounds are the small circles.

6 Medicinal Chemistry, 2013, Vol. 9, No. ?? Scotti et al.

Fig. (5). Coefficients plot: influence of descriptors in the PLS model.

Table 3. Explained Variance by PCA

PC % Explained Variance from Original Data

1 36.784

2 25.240

3 14.149

4 5.823

5 3.444

Table 4. Explained Variance by PLS

LV % Explained Variance from Original Data

1 45.160

2 27.450

3 19.910

4 7.480

5 1.538

The PLS scores plot of the resulting model is shown in (Fig. 4). The selected model provides a good discrimination between more potent (dark grey balls) and less potent (light grey balls) compounds.

The coefficient plot indicates the greater influence of de-scriptors: D1-D4, WO2-3 (Fig. 5).

The results obtained by the external predictions may be considered as acceptable. The compounds used as the test set for external prediction generated a suitable model with R2 = 0.623. The scores plot is shown in (Fig. 6).

Fig. (6). Predictive power of the test set: scores plot in the space Y CALCULATED X Y OBSERVED.

DISCUSSION

Our first study explored the computational properties contained in the orthogonal algorithm of the Consensus PCA. The method was applied to fifty-five compounds (Fig. 1) and 128 descriptors, distributed in 13 blocks. (Fig. 2) shows the weight of the blocks on the graphic and the equivalent numeric values, calculated considering PC1 and PC2. The blocks of descriptors with the highest weight are highlighted in black rectangles: H2O, DRY and SIZE/SHAPE. (Table 2) represents the variance explained by each of these blocks, in the first 5 PCs. (Table 2) shows that the variance explained by the SIZE/SHAPE block is higher than 76% and approximately 66% and 62% for the blocks H2O and DRY, respectively by the first two PCs. This first analysis guided the application of the PCA. Thus, forty-eight descriptor, contained in the block of greater weight, were selected. The first two principal components explained over than 62% of the total variance (see Table 3). This percentage value is satisfactory if compared with the findings referred in the literature [47-49]. The score plot in (Fig. 3) shows that more active compounds (dark grey balls) are tendentially located on the right side of the graph.

Chemometric Studies on Potential Larvicidal Compounds Against Aedes Aegypti Medicinal Chemistry, 2013, Vol. 9, No. ?? 7

Since multicollinearity among the descriptor variables may affect the regression analysis detrimentally, PLS is fre-quently used as the regression method in 3D QSAR (Quanti-tative structure–activity relationship) [18-21]. The PLS was applied to forty-one compounds as training set and fourteen compounds as test set (3, 7, 11, 14, 17, 23, 29, 31, 38, 40, 45, 48, 52, 55), selected following three conditions [46]:

- All compounds of the test set in the multidimensional de-scriptor space must be close to those of the training set.

-All compounds of the training set must be close to those of the test set.

-The representative points of the training set must be distrib-uted within the entire dataset.

The best model after LOO cross-validation exhibited qcv2

= 0.679 and regression correlation coefficient r2 = 0.714 in the LV1. (Table 4) shows that the model is able to explain more than 72% of the total variance, considering LV1 and LV2. The PLS demonstrate very satisfactory results, espe-cially if compared with those reached in other studies [50-52]. The score plot shows the arrangement of more active and less active objects. (Fig. 4) presents the excellent separa-tion in predicting the activities of the objects. Through the interaction maps, using the GRIG force field, the differences of interactions between the compounds and the probes H2O (light blue), DRY (green), O (red) and N (dark blue) could be compared. In (Fig. 7), we present some examples of the

differences observed between more potent (8 and 15) and less potent (21 and 22) compounds.

The interaction of molecules with biological membranes is mediated by surface properties such as shape, electrostatic forces, H-bonds and hydrophobicity. Therefore, the GRID force field was chosen to characterize potential polar and hydrophobic interaction sites around target molecules by the water (OH2), the hydrophobic (DRY), and the carbonyl oxy-gen (O) and amide nitrogen (N1) probe. The information contained in the Molecular Interaction Fields (MIF) is trans-formed into a quantitative scale by calculating the volume or the surface of the interaction contours. The VolSurf+ proce-dure is as follows: i) in the first step, the 3D molecular field is generated from the interactions of the OH2, the DRY, O and N1 probe around a target molecule; ii) the second step consists in the calculation of descriptors from the 3D maps obtained in the first step. The molecular descriptors obtained, called VolSurf+ descriptors, refer to molecular size and shape, to hydrophilic and hydrophobic regions and to the balance between them [38]. Maps generated with the GRID force field show great interactions with the O and N probes, and less interactions with the DRY probes; in less potent compounds. Hydrophobic regions are more present in more potent compounds, which leads us to believe that the hydro-phobic profile is an important factor for increasing the larv-icidal activity.

The coefficients of the descriptors selected from the best PLS model corroborate these observations (see Fig. 5). We

Fig. (7). Maps of interaction with the probes H2O (light blue), DRY (green), O (red) and N (dark blue) using GRID force field: compounds 8, 15 (more potent) and 21, 22 (less potent).

8 Medicinal Chemistry, 2013, Vol. 9, No. ?? Scotti et al.

observed positive influence of descriptors D1-D4, which are generated with hydrophobic interactions (DRY probe). Volf-surf computes these hydrophobic descriptors at eight differ-ent energy levels adapted to the usual energy range of hy-drophobic interactions (from 0.2 to 1.6 kcal/mol). GRIDa uses a probe called O (carbonylic oxygen) to generate 3D H-bond donor fields. H-bond donor regions may be defined to the molecular envelope generating attractive H-donor inter-actions. Volfsurf computes H-bond donor descriptor at six different energy levels. Two of these levels (WO2 and WO3) were observed in the present study and confirm that donor H-bonds regions influence negatively the larvicidal activity.

The external validation employed fourteen compounds as test set. The results show a good predictive power of the PLS model, with R2 = 0.623 (Fig. 6).

CONCLUSION

This work used chemometric tools to investigate a set of fifty-five compounds with activity against Aedes aegypti larvae. The results were satisfactory, with good predictive power of the PLS model. The descriptors and the maps cal-culated with GRID force field showed that hydrophobicity is strongly correlated with the larvicidal activity.

Compounds 34 and 41 exhibited reasonable larvicidal ac-tivity when compared to similar derivatives such as 35 and 40 (Table 1). The improvement in potency is herein attrib-uted to an increase in lipophilicity. However, in some cases, an increase in potency is not always followed by an increase in lipophilicity, which may be attributed to the diverse struc-tural templates encountered in this set, resulting in different mechanisms of action for each related subset. This finding encourages further theoretical and experimental researchers to continue performing studies, aiming to reveal structural groups of these compounds responsible for the larvicidal activity.

Dengue vector control is a constant social and govern-ment concern. In order to reduce the spreading of this dis-ease, public health agencies in tropical countries use larv-icides with the goal to eliminate Ae. aegypti larvae growing in breeding sites, therefore preventing virus transmission. In this work, we extract important structural features of the in-vestigated compounds, allowing future use of the method and/or directing research to the synthesis of new substances with larvicidal activity.

CONFLICT OF INTEREST

The author(s) confirm that this article content has no con-flicts of interest.

ACKNOWLEDGEMENTS

The authors are grateful to Prof. Gabriele Cruciani from Università di Perugia - Italy for the VolSurf+ license and to CNPq and UEPB (Research and Post-Graduation Incentive Program/PROPESQ-PRPGP) for financial support.

ABBREVIATIONS CPCA = Consensus PCA PCA = Principal Component

Analysis

PLS = Partial Least Squares Regression

PC = Principal Component GRID = Force field MIF = Molecular Interaction

Fields DV-1, DV-2, DV-3, and DV-4 = Serotypes of the dengue

virus WHO = World Health Organiza-

tion Ae. aegypti = Aedes aegypti SAR = Structure–Activity Rela-

tionship AM1 = Austin Model 1 H2O, DRY, O, and N = Probes of VolSurf+ pro-

gram SHAPE, DRY, and H2O = Blocks of descriptors 3D = Three-dimensional LOO = Leave-one-out technique pLC50 = !log LC50 LC50 = 50% lethal concentration LV = Latent variable D1-D4 and WO2-3 = Variables of VolSurf+

program

REFERENCES [1] WHO, 2012. Dengue and dengue haemorrhagic fever. Fact sheet

117, Geneva. [2] Balankur, M.; Valyasev, A.; Kampanar, C.; Cohen, S. Treatment of

Dengue Shock Syndrome. Bulletin of the World Health Organization, 1966, 35-75.

[3] Chiu, M.W.; Shih, H.M.; Yang, T.H.; Yang, Y.L. The type 2 dengue virus envelope protein interacts with small ubiquitin-like modifier-1 (SUMO-1) conjugating enzyme 9 (Ubc9). J. Biom. Sci., 2007, 14, 429-444.

[4] Who Dengue, available at http://www.who.int/topics/dengue/en/ ; access on 3/6/2012.

[5] Figueiredo, L.T.M. Febres hemorrágicas por vírus no Brasil. Rev. Soc. Bras. Med. Trop., 2006, 39, 203-210.

[6] Machado-Machado, E.A. Empirical mapping of suitability to den-gue fever in Mexico using species distribution modeling. Applied Geograph., 2012, 33, 82-93.

[7] Braga, I.A.; Lima, J.B.P.; Soares, S. D.; Valle, D. Aedes aegypti resistance to Temephos during 2001 in several municipalities in the states of Rio de Janeiro, Sergipe, and Alagoas, Brazil. Mem. I. Oswaldo Cruz, 2004, 99, 199-203.

[8] Vimaladevi, S.; Mahesh, A.; Dhayanithi, B.N.; Karthikeyan, N. Mosquito larvicidal efficacy of phenolic acids of seaweed Chaetomorpha antennina (Bory) Kuetz. against Aedes aegypti. Biol., 2012, 67, 212-216.

[9] Garza-Robledo, A.A.; Martinez-Perales, J.F.; Rodriguez-Castro, V.A.; Quiroz-Martinez, H. Effectiveness of spinosad and temephos for the control of mosquito larvae at a tire dump in Allende, Nuevo Leon, Mexico, J. Am. Mosq. Control. Assoc., 2011, 27, 404-407.

[10] Wang, Z.; Kim, J.R.; Wang, M.; Shu, S.; Ahn, Y.J., Larvicidal activity of Cnidium monnieri fruit coumarins and structurally related compounds against insecticide-susceptible and insecticide-

Chemometric Studies on Potential Larvicidal Compounds Against Aedes Aegypti Medicinal Chemistry, 2013, Vol. 9, No. ?? 9

resistant Culex pipiens pallens and Aedes aegypti, Pest. Manag. Sci., 2012, 68, 1041-1047.

[11] Ulzias, R.J.; Ramos, C.S.; Serafini, M.R.; Cavalcanti, S.C.H.; Alves, P.B.; Lima, G.M.; Andrade, P. H. S.; Bonjardim, L.R.; Quintans, L.J.J.; Araujo, A.A.S. Evaluation of the lethality of Porophyllum ruderale essential oil against Biomphalaria glabrata, Aedes aegypti and Artemia salina. Afr. J. Biotechnol., 2012, 11, 3169-3173.

[12] Adeleke, M.A.; Popoola, S.A.; Agbaje, W.B.; Adewale, B.; Adeoye, M.D.; Jimoh, W.A. Larvicidal efficacy of seed oils of Pterocarpus santalinoides and tropical Manihot species against Aedes aegypti and effects on aquatic fauna, Tanzan J. Health Res., 2009, 11, 250-252.

[13] Cantrell, C.L.; Pridgeon, J.W.; Fronczek, F.R.; Becnel, J.J. Structure - Activity Relationship Studies on Derivatives of Eudesmanolides from Inula helenium as toxicants against Aedes aegypti larvae and adults. Chem. Biodivers., 2010, 7, 1681-1697.

[14] Santos, S.R.L.; Silva, V.B.; Melo, M.A.; Barbosa, J.D.F.; Santos, R.L.C.; de Sousa, D.P.; Cavalcanti, S.C.H., Toxic effects on and structure-toxicity relationships of phenylpropanoids, Terpenes, and Related Compounds in Aedes aegypti Larvae. Vector-Borne Zoonot., 2010, 10, 1049-1054.

[15] Santos, S.R.L.; Melo, M.A.; Cardoso, A.V.; Santos, R.L.C.; de Sousa, D.P.; Cavalcanti, S.C.H. Structure-activity relationships of larvicidal monoterpenes and derivatives against Aedes aegypti Linn., Chemosp., 2011, 84, 150-153.

[16] Barbosa, J.D.F.; Silva, V.B.; Alves, P.B.; Gumina, G.; Santos, R.L.C.; Sousa, D.P.; Cavalcanti, S.C. H. Structure–activity relation-ships of eugenol derivatives against Aedes aegypti (Dip-tera:Culicidae) larvae. Pest Manag. Sci., 2012, 68, 1478-1483.

[17] Silva, W.J.; Doria, G.A.A.; Maia, R.T.; Nunes, R.S.; Carvalho, G.A.; Blank, A.F.; Alves, P.B.; Marcal, R.M.; Cavalcanti, S.C.H. Effects of essential oils on Aedes aegypti larvae: Alternatives to environmentally safe insecticides. Bioresource Technol., 2008, 99, 3251-3255.

[18] Beebe, K.R.; Pell, R.J.; Seasholtz, M.B. Chemometrics: A Practical Guide, Wiley & Sons: New York, 1998, p. 185.

[19] Westerhuis, J.A.; Kourti, T.; Macgregor, J.F. Analysis of multi-block and hierarchical PCA and PLS models. J. Chemometrics, 1998, 12, 301-321.

[20] Sharaf, M.A.; Illman, D.L.; Kowalski, B.R.; Chemometrics, John Wiley & Sons: New York, 1986, p. 336.

[21] Ooms, F. Molecular modeling and computer aided drug design. Examples of their applications in medicinal chemistry. Curr. Med. Chem., 2000, 7,141-158.

[22] Scotti, L.; Ferreira, E.I.; Silva, M.S.; Scotti, M.T., Chemometric Studies on Natural Products as Potential Inhibitors of the NADH Oxidase from Trypanosoma cruzi using the VolSurf approach. Molecules, 2010, 15, 7363-7377.

[23] Scotti, L.; Scotti, M.T.; Lima, E.O.; Silva, M.S.; Lima, M.C.A.; Pitta, I.R.; Moura, R.O.; Oliveira, J.G.B.; Cruz, R.M.D.; Mendonça Junior, F.J.B., Experimental methodologies and evaluations of computer-aided drug design methodologies applied to a series of 2-aminothiophene derivatives with antifungal activities. Molecules, 2012, 17, 2298-2315.

[24] Klein, E.; Ohloff, G. Der stereochemische verlauf der alkalischen epoxydation von alpha,beta-ungesattigten carbonylverbindungen der cyclischen monoterpenreihe. Tetrahedron, 1963, 19, 1091-1099.

[25] Corey, E.J.; Suggs, J.W. Pyridinium chlorochromate - Efficient reagent for oxidation of primary and secondary alcohols to carbonyl-compounds, Tetrahedron Lett., 1975, 16, 2647-2650.

[26] Thomas, A.F.; Bessiere, Y. Limonene. Nat. Prod. Rep., 1989, 6, 291-309.

[27] Ben Arfa, A.; Combes, S.; Preziosi-Belloy, L.; Gontard, N.; Chalier, P. Antimicrobial activity of carvacrol related to its chemical structure. Lett. Appl. Microbiol., 2006, 43, 149-154

[28] Grodnitzky, J.A.; Coats, J.R. Antimicrobial activity of carvacrol related to its chemical structure. J. Agric. Food Chem., 2002, 50, 4576-4580.

[29] Dolly, B.B.; Barba, F. Cathodic reduction of hydroxycarbonyl compound trichloroacetyl esters. Tetrahedron, 2003, 59, 9161-9165.

[30] Nikumbh, V.P.; Tare, V.S.; Mahulikar, P.P. Eco-friendly pest management using monoterpenoids-III: Antibacterial efficacy of carvacrol derivatives. J. Sci. Ind. Res. India, 2003, 62, 1086-1089.

[31] Coolen, H.K.A.C.; Meeuwis, J.A.M.; Vanleeuwen, P. W. N. M.; Nolte, R. J. M. Substrate Selective Catalysis by Rhodium Metallohosts. J. Am. Chem. Soc., 1995, 117, 11906-11913.

[32] Barbosa, J.D.F.; Silva, V.B.; Alves, P.B.; Gumina, G.; Santos, R.L.C.; Souza, D.P.; Cavalcanti, S. C. H. Structure–activity relationships of eugenol derivatives against Aedes aegypti (Diptera: Culicidae) larvae. Pest. Manag. Sci., 2012, 68, 1478-1483.

[33] Hyperchem Program Release 8.0 for Windows 1999-2005 Hyber-cube, Inc.: Gainesville, USA.

[34] Allinger, N.L. Hydrocarbon force-field utilizing V1 and V2 tor-sional terms. J. Am. Chem. Soc., 1977, 99, 8127-8134

[35] Dewar, M.J.S.; Zoebisch, G.; Healy, E.F.; Stewart J.J.P. The devel-opment and use of quantum-mechanical molecular-models .76. AM1 - A new general-purpose quantum-mechanical molecular-model, J. Am. Chem. Soc. 1985, 107, 3902-3909.

[36] Cohen, N.C. Guidebook on molecular modeling in drug design. Academic Press: San Diego, 1996, p. 361.

[37] Leach, A.R. Molecular Modeling: Principles and Applications. Prentice Hall: London, 2001, p. 784.

[38] Cruciani, G.; Crivori, P.; Carrupt, P.-A.; Testa, B.J. Molecular fields in quantitative structure–permeation relationships: the Vol-Surf approach. Mol. Struct., 2000, 503, 17-30.

[39] Zamora, I.; Oprea, T.; Cruciani, G.; Pastor, M.; Ungell, A-L. Sur-face descriptors for protein-ligand affinity prediction. J. Med. Chem., 2003, 46, 25-33.

[40] Crivori, P.; Cruciani, G.; Carrupt, P-A.; Testa, B. Predicting Blood-Brain Barrier Permeation from Three-Dimensional Molecular Structure. J. Med. Chem., 2000, 43, 2204-2216.

[41] Kovatcheva, A.; Golbraikh, A.; Oloff, S.; Xiao, Y-D.; Zheng, W.; Wolschann, P.; Buchbauer, G.; Tropsha, A. Combinatorial QSAR of Ambergris Fragrance Compounds. J. Chem. Inf. Comput. Sci., 2004, 44, 582-595.

[42] Oprea, T.I.; Zamora I.; Ungell A.-L. Combinatorial QSAR of Am-bergris Fragrance Compounds. J. Comb. Chem., 2002, 4, 258-266.

[43] Cruciani, G.; Pastor, M.; Guba, W. VolSurf: a new tool for the pharmacokinetic optimization of lead compounds. Eur. J. Pharm. Sci., 2000, 11, S29-S39.

[44] Cruciani, G.; Pastor, M.; Mannhold, R. Suitability of molecular descriptors for database mining: a comparative analysis. J. Med. Chem., 2002, 45, 2685-2694.

[45] Cianchetta, G.; Mannhold, R.; Cruciani, G.; Baroni, M.; Cecchetti, V.J. Chemometric studies on the bactericidal activity of quinolones via an extended VolSurf approach. J. Med. Chem., 2004, 47, 3193-3201.

[46] Golbraikh, A.; Shen, M.; Xiao, Z.; Xiao, Y.-D.; Lee, K-H.; Trop-sha, A. Rational selection of training and test sets for the develop-ment of validated QSAR models. J. Comp.-Aided Mol. De., 2003, 17, 241-253.

[47] Feio, M.J.; Dolédec, S. Integration of invertebrate traits into predic-tive models for indirect assessment of stream functional integrity: A case study in Portugal. Ecol. Indic., 2012, 15, 236-247.

[48] Vaughan, I.P.; Ormerod, S.J. Increasing the value of principal com-ponents analysis for simplifying ecological data: a case study with rivers and river birds. J. Applied Ecology, 2005, 42, 487-497.

[49] Luthuria, D.L.; Lin, L-Z.; Robbins, R.J.; Finley, J.W.; Banuelos, G.S.; Harnly, J.M. Discriminating between cultivars and treatments of broccoli using mass spectral fingerprinting and analysis of vari-ance-principal component analysis. Agric. Food Chem., 2008, 56, 9819-9827.

[50] Jean-Pierre, P.; Fiscella, K.; Freund, K.M.; Clark, J.; Darnell, J.; Holden, A.; Post, D.; Patierno, S. R.; Winters, P. C. Structural and reliability analysis of a patient satisfaction with cancer-related care measure: a multisite patient navigation research program study. Cancer, 2011, 15, 854-861.

10 Medicinal Chemistry, 2013, Vol. 9, No. ?? Scotti et al.

[51] Xuan, P.; Zhang, Y.; Tzeng, T.J.; Wan, X.; Luo, F. A quantitative structure-activity relationship (QSAR) study on glycan array data to determine the specificities of glycan-binding proteins. Glycobiol., 2011, 22, 552-560.

[52] McNally, R.C.; Akdeniz, M.B.; Calantone, R.J. New Product De-velopment Processes and New Product Profitability: Exploring the Mediating Role of Speed to Market and Product Quality. J. Prod. Innov. Manag., 2011, 28, 63-.77.

Received: December 28, 2012 Revised: May 08 2013 Accepted: May 08, 2013