Curcumin Derivatives as Potential Mosquito Larvicidal Agents against Two Mosquito Vectors, Culex pipiens and Aedes albopictus

Vector-borne diseases have appeared or re-emerged in many Southern Europe countries making the transmission of infectious diseases by mosquitoes (vectors) one of the greatest worldwide health threats. Larvicides have been used extensively for the control of Aedes (Stegomyia) albopictus (Skuse, 1895) (Diptera: Culicidae) and Culex pipiens Linnaeus, 1758 (Diptera: Culicidae) mosquitoes in urban and semi-urban environments, causing the increasing resistance of mosquitoes to commercial insecticides. In this study, 27 curcuminoids and monocarbonyl curcumin derivatives were synthesised and evaluated as potential larvicidal agents against Cx. pipiens and Ae. albopictus. Most of the compounds were more effective against larvae of both mosquito species. Four of the tested compounds, curcumin, demethoxycurcumin, curcumin-BF2 complex and a monocarbonyl tetramethoxy curcumin derivative exhibited high activity against both species. In Cx. pipiens the recorded LC50 values were 6.0, 9.4, 5.0 and 32.5 ppm, respectively, whereas in Ae. albopictus they exhibited LC50 values of 9.2, 36.0, 5.5 and 23.6 ppm, respectively. No conclusive structure activity relationship was evident from the results and the variety of descriptors values generated in silico provided some insight to this end.


Introduction
Diseases caused from vector bites such as mosquitos, ticks and fleas have more than tripled in the United States from 2004-2016 [1]. Among them, mosquito-borne diseases (MBDs) present a major challenge and concern of global public health and safety. Travel and trade globalization, unplanned urbanization and environmental change have had a significant impact on disease transmission in recent years [2,3]. Regarding diseases caused by Aedes-borne viruses, more than one million Zika virus (ZIKV) infections and thousands of infants with birth defects were reported from 2015-2016 and similarly Dengue fever (DENV) which infects over one hundred million people in over 100 countries every year [4,5]. In 1999, the West Nile Virus (WNV) crisis across the United States with hundreds of deaths annually are some further examples of MBDs transmitted to humans usually through the bite of an infected mosquito of the genus Culex [6]. Finally, except human BF 2 . Figure 1 summarises the general structural modification of the curcumin derivatives studied herein as potential larvicidals against Ae. albopictus and Cx. pipiens.

Synthesis
The general synthetic routes to symmetric dicarbonyl analogues of curcumin 1a-i, along with the rigid cyclized pyrazole 2 and BF 2 analogues 3f-i are depicted in Scheme 1. Compounds 1a-i were prepared according to well-established curcumin synthetic protocols with slight modifications. Boron trioxide (B 2 O 3 ) was added in a mixture of tributyl borate B[(OBu) 3 ], acetylacetonate and the appropriate substituted benzaldehyde to form a complex with the acetylacetone, thus ensuring that the aldol condensation occurs only at the terminal methyl groups. The reactions took place in the absence of any organic solvent and after the addition of n-butylamine (nBuNH 2 ), the boron complexes of the final products were hydrolyzed by aqueous hydrochloric acid.
Pyrazole 2 was prepared from compound 1b using threefold excess of hydrazine hydrate in a 1:1 acetic acid/ethanol mixture as solvent. The product was obtained as a precipitated solid in satisfactory yield and high purity without any additional purification steps.
Difluoroboron curcumin derivatives 3f-i were prepared by the addition of boron trifluoride diethyletherate to the corresponding curcumins 1f-i. The formation of the complexes was evidenced by the higher frequency shifts of the β-diketone unit active methylene proton (0.33-0.50 ppm).
Non-symmetric curcuminoid 5 (known as curcumin III) was prepared from condensation of the monoarylidene curcumin analogue 4 with 4-hydroxybenzaldehyde (Scheme 2). 4 was synthesized using an excess of acetylacetone in ethyl acetate in order to avoid significant formation of the symmetric diarylidene derivative (1f).
Compounds 6a-h were obtained by Claisen-Schmidt condensation of the appropriate aromatic aldehydes and cyclohexanone or acetone using a ratio of 1:2 of ketone to aldehyde under alkaline or acidic conditions (Scheme 3). Thiophenyl and furanyl derivatives 6a,b were obtained in best purity and yields using solid potassium hydroxide, while for methoxy-substituted phenyl derivatives 6c-e, a dispersion of sodium hydroxide in ethanol was added instead. Dihydroxybenzaldehyde and vanillin derived analogues 6f-h were prepared under acidic conditions using concentrated hydrochloric acid. Finally, polyhydroxylated compound 6i was obtained by demethylation of 6e with boron tribromide in anhydrous dichloromethane (DCM). Scheme 1. Synthesis of symmetric dicarbonyl analogues of curcumin 1a-i, pyrazole derivative 2 and boron difluoride complexes 3f-i. Reagents  Similarly, to 6g and 6i, half curcuminoids (monoarylidene) 7a and 7b were prepared by Claisen-Schmidt condensation between 3,4-dihydroxybenzaldehyde and acetone (7a) and by demethylating the intermediate trimethoxy half curcuminoid following the boron tribromide protocol (7b) (Scheme 4). In both cases, excess of acetone for the Claisen-Schmidt reaction was used to minimize the production of bisarylidene molecules. Scheme 3. Synthesis of monocarbonyl analogues of curcumin 6a-h. Reagents and conditions: (i) cyclohexanone, KOH powder, EtOH, room temperature., 2 h; (ii) acetone, NaOH/EtOH 10%, room temperature, 2 h (6c-e) or acetone (or cyclopentanone), c. HCl, EtOH, room temperature, 24 h (6f-h); (iii) BBr 3 , DCM (dry), −20 • C to 0 • C to room temperature, 3 h. The purity and structure of all synthesized compounds were determined by 1 H NMR spectroscopy and elemental analyses. Novel derivative 3i was further characterized by 13 C and 2D NMR experiments.

Larvicidal Activity
Initially, all compounds were screened at a concentration of 20 ppm against Ae. albopictus and Cx. pipiens for 24 h (Figure 2). No mortality was observed in the control experiments of tap water and 2% DMSO in water. Among the tested compounds, in both species, the same 4 compounds (1f, 3f, 5 and 6d) exhibited mortality ≥ 10%, regardless of the observed difference in the overall mortality effect. For the rest of the tested compounds 11 and 6 caused <10% mortality against the larvae of Cx. pipiens and Ae. albopictus, respectively, while the other derivatives were completely inactive against larvae of both species. The results were not altered at all after a 48 h incubation time with the only exception of 3f which exhibited a very slight increase in activity. Consequently, the 4 most active compounds revealed from the 20 ppm screening experiment, namely 1f, 3f, 5 and 6d, were further evaluated for larvicidal activity using various concentrations for 24 h incubation. In that way, a dose-response bioassay was able to be drawn enabling the estimation of their respective LC 50 and LC 90 values (Table 1).
In the case of larvicidal activity against Cx. pipiens compound 3f, the BF 2 -curcumin complex, exhibited the highest activity with an LC 50 value of 5 ppm whereas curcumin itself (1f) was almost equipotent with LC 50 value of 6 (5.4-6.6) ppm. The activity of 3f was also very slightly increased upon 48 h treatment showing an LC 50 value of 4.2 against Cx. pipiens. The second curcuminoid, demethoxycurcumin, 5, was also highly active with an LC 50 value of 9.4 (8.2-10.6) ppm while the monocarbonyl curcumin analogue 6d was the least active of all. Similarly, when Ae. albopictus larvae were incubated with 3f the highest activity was observed followed by that of 1f, exhibiting LC 50 values of 5.5 (4.6-6.3) ppm and 9.2 (8.5-9.9) ppm, respectively. As before the LC 50 value of 3f was slightly smaller at the range of 4.8 ppm. The other two compounds, 5 and 6d, were less active against Ae. albopictus larvae compared to their effect against Cx. pipiens, with LC 50 values of 36 (32.6-39.5) ppm and 23.6 (21.3-26.2) ppm, respectively. Moreover, in both cases LC 90 values follow the same order of activity. a LC values are expressed in ppm (mg/L) and they are considered significantly different when 95% CL fail to overlap; b Since goodness-of-fit test is significant (p < 0.05), a heterogeneity factor is used in the calculation of confidence limits (CL).

Calculations of Molecular Descriptors
The larvicidal activity results against Ae. albopictus and Cx. pipiens clearly indicate that curcumin (1f) and their synthesized derivatives 3f, 5 and 6d possessed a remarkable activity, compared to their structural similar compounds. Based on the results obtained a better understanding of their structure-activity relationships was attempted by generating a plethora of descriptors the values of which are presented in Table 2.
There is no clear rationalization within the ALogP, whereas the active molecules seem to have a number of 5 or 6 hydrogen bond acceptors. Interestingly, the molar refractivity (MR) values of all active compounds fall within the range between 96.77 and 107.61 in addition to their very narrow range of molecular polarizability (Polar) which characterizes them ranging between 43.0 and 45.74. The almost equipotent compounds 1f and 3f seem to present a great difference in their lipophilicity values, share the same hydrogen bond donor and acceptor values and they have very close molar polarisability results.
The analysis of the calculated descriptors values revealed that the replacement of the dicarbonyl group of 1f by cyclized pyrazole ring (compound 2) has a negative contribution to the larvicidal activity. Particularly, the lipophilicity value (ALogP) of derivative 2 is increased (ALogP = 3.85) compared to the lipophilicity value of curcumin (1f) (ALogP = 3.01) and the opposite phenomenon is regarded for the electrotopological state index (E-state) value, which describes the influential molecular fragments. Especially, the E-state values for compounds 1f and 2 are 71.84 and 63.84, respectively.
In the case of compound 1g the methoxy (-OMe) and the hydroxyl (-OH) groups of curcumin (1f) have been replaced by hydrogen (-H) and amine (-NR 2 ) groups, respectively. The comparison of the calculated descriptors' values between the active compound 1f and the inactive compound 1g indicated greater values for the lipophilicity (ALogP = 3.90) and a reduction of hydrogen bond donors and acceptors. Furthermore, the Polar Surface Area (PSA) value of compound 1g, which describes the van der Waals surface area of polar nitrogen and oxygen atoms, is significantly reduced (PSA = 40.62) compared to 1f (PSA = 93.06). Comparison between curcumin (1f) and its symmetric dicarbonyl analogue 1i displayed higher values for lipophilicity (ALogP = 3.94), Molar Refractivity (MR = 123.87) and Polarizability (Polar = 53.43) for 1i. In continuation, the comparison of the physicochemical profile of the most active difluorocurcumin derivative 3f and its structural similar inactive compound 3g showed that the replacement of methoxy (-OMe) and hydroxyl (-OH) groups causes an increase in lipophilicity (ALogP), Molar Refractivity (MR) and Polarizability values and a decrease in Polar Surface Area (PSA) and Electrotopological state index (E-state) values. In the described case, the greatest difference is identified in Polar Surface Area values, which are 85.22 and 32.78 for compounds 3f and 3g, respectively. Moreover, similar physicochemical pattern is followed for compound 3i, in which the aliphatic chain contains one additional carbon double bonds at each side of the diketone moiety.
From compounds bearing mono-carbonyl substitution, compound 6d presents larvicidal activity against Ae. albopictus and Cx. pipiens. Physicochemical profile comparison of derivative 6d with its structurally similar compounds 6c, 1b, 6e and 6f indicates a notable difference in Polar Surface Area values. Indicatively, for 6d the value is 53.99 while for compound 6c, which bears only one methoxy group is 35.53. In addition, for compounds 1b, 6e and 6f an increase is regarded.
There has been, for years, a significant amount of research efforts trying to identify the most potent larvicidal extract of various sources of Curcuma spices but only very few of them isolated, identified and evaluated the activity of individual compounds. The crude methanolic extract of C. longa gave an LC 50 value against the 4th-instar larvae of Cx. pipiens pallens of 355.06 ppm [25]. This study revealed the larvicidal activity of ar-turmerone and 8-hydroxyl-ar-turmerone. The two compounds exhibited larvicidal activities against the 4th-instar larvae of Cx. pipiens pallens after 24 h of treatment with LC 50 values of 138.86 and 257.68 ppm, respectively. The petroleum ether extract of C. aromatic eliminated Cx. quinquefasciatus at an LC 50 value of 11.42 ppm [23]. Further investigation of the extract lead to the isolation of two larvicidal compounds namely 9-oxoneoprocurcumenol and neoprocurcumenol. The former exerted significant toxicity (p < 0.01) on mosquito larvae with LC 50 value of 5.81 ppm compared to the latter with 13.69 ppm. However, bearing in mind that active phytochemicals can be influenced by multiple factors, such as the type of solvent used for the extraction and the extraction process, the plant genus, the conditions under which the plants were harvested and the target mosquito species used for the tests [27] it is important to identify more synthetic derivatives of such active natural products.
In the present study, 27 synthetic analogues and derivatives of curcumin were evaluated against Cx. pipiens and Ae. albopictus larvae and only 4 of them exerted significant larvicidal activity one of them being curcumin itself, 1f. Its activity against Cx. pipiens larvae was found to have a small deviation from our previous results [26], showing an increased activity with an LC 50 value of 6 ppm compared to 19 ppm. Additionally, demethoxycurcumin, 5, had been previously found inactive against Cx. pipiens whereas an exciting LC 50 value of 9.4 ppm was exerted in this study further to the equally significant LC 50 value of 36.0 ppm against Ae. albopictus. The small discrepancies from our previous results may be related to the relative differences in the developmental stage of the mosquitoes in the two studies. It has been reported that certain biochemical changes to some target molecules from the first to the fourth instar, such as sterol carrier protein-2, acetylcholinesterase, detoxification and resistance mechanisms and others, may affect treatment results [28]. The same was observed for the methanol crude extract of Artemisia nilagirica (Clarke) with reported LC 50 values of 272.50, 311.40, 361.51 and 442.51 ppm, respectively, against the first to fourth instar larvae of An. stephensi and 300.84, 338.79, 394.69 and 470.74 ppm, respectively, against the first to fourth instar larvae of Ae. aegypti. [29] Moreover, our overall observation for the lower activity exhibited by all tested compounds against Ae. albopictus compared with Cx. pipiens may also be related to analogous interspecies variations. Similar results have been reported for the natural compound palmitic acid, extracted from Millettia pinnata (L.) seeds, which showed LC 50 values of 34.50, 42.96 and 85.61 ppm against the third instar larvae of Cx. pipiens pallens, Ae. aegypti and Ae. albopictus, respectively [30].
To our surprise any change to the curcumin skeleton eliminated larvicidal effect against both species. This was found to be the case regardless if the change was at the substitution pattern or functional group substituent or the presence of additional double bond. The only structural transformation which benefited the activity of the molecule was the formation of the BF 2 complex, 3f, which increased larval mortality even more compared to its mother compound, 1f. It is noteworthy that none of the inactive diketonic curcuminoids exerted higher activity when converted to its corresponding BF 2 complex which implies that the curcumin part dictates the degree of activity. In the case of 3f, therefore, the anticipated increased chemical, photo-and pH stability and improved solubility compared to 1f may certainly account for the increased larvicidal effect observed, especially in the case of Ae. albopictus, as well as the slightly higher activity exerted after the 48-h exposure.
Based on our previous work in which the tetrahydroxycurcumin, 1e exerted significant larvicidal effect it was part of our initial hypothesis to investigate the activity of some polyhydroxylated monocarbonyl analogues of curcumin of the type of 6a,b and 7a,b in comparison to 1d and 1e. [26]. The lack of chemical stability, the rapid degradation and poor bioavailability of curcuminoids has been strongly related to the β-diketone moiety [31,32]. In recent years, the solution to the aforementioned limitations has been intensely explored in the synthesis and application of monocarbonyl derivatives which have shown indeed improved biological and pharmacological profile [33,34]. In our case, however, neither the polyhydroxylated monocarbonyl derivatives showed any activity at all nor the diketone counterparts. On the contrary, significant larvicidal activity was recorded against both mosquito species by the tetramethoxy monocarbonyl curcumin derivative, 6d. To the best of our knowledge, only the work by Anstrom [35] has searched into similar monocarbonyl curcumin derivatives as potential mosquitocidal agents and more specifically it made an effort to relate the inhibitory potential against sterol carrier protein-2, although a clear correlation was not found. This type of monocarbonyl derivatives, due to the ease of preparation and low production cost, should be, therefore, investigated further aiming to elucidate some more mechanistic information on the mode of their larvicidal action. The structurally related chalcones have also been investigated as mosquito larvicides and the compound (E)-3-(4-bromophenyl)-1-(furan-2-yl) prop-2-en-1-one exhibited an LC 50 value of 6.66 mg/L at 24 h against Ae. aegypti larvae [36]. Additionally, 28 compounds, chalcones and some derived products, were synthesized and tested for mosquito larvicidal activity against the third instar larvae of Cx. quinquefasciatus. Four of them exhibited outstanding activity of 5-55 µM and some structure-activity relationship was derived [37]. Finally, the increased larvicidal activities of some chalcones against Ae. albopictus with LC 90 values of 5 ppm after 72 h of exposure have been related to their high activity as juvenile hormone antagonist (JHAN) [38]. From a mechanistic point of view, acetylcholinesterase (AChE) is considered as the most interesting molecular target because it is a critical nervous system enzyme responsible for synaptic transmission and is the target site for organophosphoate and carbamate insecticides [31,39]. Additionally, interference with the octopaminergic system has been related to the insecticidal activity of natural products [30]. Finally, the larvicidal mechanism of ar-turmerone has been attributed to stomach poisoning and the active sites might be the muscle and digestive tissues [40]. Further investigation is necessary and currently underway to provide some evidence on the mode of action of our compounds.
In conclusion, in our hands 4 out of 27 curcuminoids exhibited highly promising larvicidal activity against Ae. albopictus and Cx. pipiens. Further chemical derivatisation of curcumin skeleton is underway to optimize physicochemical properties and larvicidal activity in laboratory and field conditions as well as to provide further understanding of their mechanism of action.

Synthesis of Compound 2
To a solution of compound 1b (0.99 g, 2.5 mmol) in acetic acid (10 mL) and ethanol (12 mL), hydrazine hydrate (0.24 g, 7.5 mmol) was added. The solution was stirred under reflux for 24 h. Ethanol was evaporated under vacuum and the resulting solution was added to a mixture of ice and water (50 mL). The precipitate was filtered, washed with water (3 × 5 mL) and dried under vacuum and P 2 O 5 .

Synthesis of Compound 4
Boron trioxide (0.69 g, 10.0 mmol) and acetylacetone (4.12 mL, 40 mmol) were dissolved in EtOAc (30 mL) and stirred at 50 • C for 30 min. To this mixture, tributyl borate (5.4 mL, 20 mmol) and vanillin (1.52 g, 10 mmol) were added. After stirring for 15 min, n-butylamine (0.73 mL, 10.0 mmol) solution in EtOAc (5 mL) was added dropwise. The mixture was stirred at 50 • C for 24 h. Hydrochloric acid (30 mL, 1 M) was added and the mixture was stirred at 50 • C for 3 h. The mixture was extracted with ethyl acetate (3 × 30 mL). The combined organic extracts were washed with brine, dried with Na 2 SO 4 and concentrated under vacuum. The crude product was purified by flash chromatography (CHCl 3 ) and then by recrystallization (CHCl 3 and n-hexane).

Synthesis of Compounds 6f-h
To a solution of vanillin (3.04g, 20 mmol) or 3,4-dihydroxybenzaldehyde (2.76 g, 20 mmol) in EtOH (3 mL), were added acetone (0.72 mL, 10 mmol) or cyclopentanone (0.84 g and concentrated HCl (0.2 mL, 37%). The mixture was stirred at r.t. for 24 h. Upon completion of the rection, the solvent was concentrated to a volume of 0.5 mL approximately and the solution is added to ice-cold water. Aqueous KOH (1%) was added until pH 6-7 and the precipitate was filtered and washed with water (×2) and then with warm water (~60 • C). The crude product is purified by recrystallization from EtOH and water.

Synthesis of Compound 7a
To a solution of the 3,4-dihydroxybenzaldehyde (1.38 g, 10 mmol) in EtOH (2 mL), were added acetone (10 mL) and concentrated HCl (1 mL, 37%). The mixture was stirred at r.t. for 24 h. Upon completion of the reaction, the solvent was concentrated to a volume of 0.5 mL approximately and the solution is added to a small volume of ice-cold water. Aqueous KOH (1%) was added until pH 6-7 and the precipitate was filtered and washed with small amounts of water (2 × 3 mL).

Synthesis of Compound 7b
A solution of 3,4,5-trimethoxybenzaldehyde (1.96 g, 10 mmol) and NaOH 1 M (1 mL, 1 mmol) in acetone (30 mL) is stirred at r.t. for 24 h. The solution was concentrated to a volume of 10 mL and water (30 mL0 was added). The formed brown oil was separated from the aqueous solution and after drying under vacuum, it was purified by flash chromatography (CHCl 3 ). An appropriate quantity of the resulting off-white solid (118 mg, 0.5 mmol) was dissolved in anhydrous DCM (10 mL) in a 50 mL two-necked round bottom flask under N 2 . The solution was cooled to −20 • C using ice/NaCl bath. Boron tribromide (2.0 mmol) was injected carefully with a syringe. The reaction mixture was stirred for 1 h at −20 • C, then for 1 h at 0 • C and then for 1 h at rt. Upon completion of the reaction, ice-water mixture was poured into the reaction mixture and the flask was shaken for a few minutes. The crude product was extracted with EtOAc (×3). The organic solvent was evaporated under vacuum and the oily product was purified by flash chromatography (DCM/MeOH = 95:5).

Mosquito Rearing
Mosquito larvae were obtained from laboratory colonies of Ae. albopictus and Cx. p. biotype molestus, which were maintained at 25 ± 2 • C, 80% relative humidity and photoperiod of LD 16:8h in the laboratory of Benaki Phytopathological Institute, Kifissia, Greece. Wood-framed cages (33 × 33 × 33 cm 3 ) covered by a 32 × 32 mesh were used to keep adult mosquitoes of each species separately, ensuring easy access to 10% sucrose solution through a cotton wick. They have not been treated with any product prior to the tests to avoid establishment of resistance.
Aedes albopictus females were chicken blood fed by using the Hemotek membrane feeding system (Hemotek). Larvae were reared in tap water-filled cylindrical enamel pans and fed ad libitum with powdered fish food (JBL Novo Tom 10% Artemia) until the adults emerged. Plastic beakers with 100 mL water and strips of moistened filter paper were provided in the cage for oviposition. The eggs were kept wet for a few days and then placed in the pans for hatching.
Culex pipiens biotype molestus females were not blood-fed since this biotype is an autogenous biotype, i.e., female mosquitoes are able to produce their first egg-batch without a blood meal. Mosquito larvae were reared in tap-water-filled cylindrical enamel pans and were fed ad libitum with powdered fish food (JBL Novo Tom 10% Artemia).
Cages containing Cx. p. biotype molestus adults were provided with containers filled with tap water for oviposition [55,56].

Larvicidal Bioassays
The larval mortality bioassays were carried out according to the test method of larval susceptibility as suggested by the World Health Organization [56] with modifications. Stock solutions of 10% (w/v) in dimethyl sulfoxide (DMSO) were prepared for each testing material. Twenty late-third-to early-fourth-instar mosquito larvae of Ae. albopictus and Cx. pipiens were exposed to different doses of the tested materials, expressed as ppm (mg/L), under laboratory conditions. Four replicates per dose were made and treatments with only tap water and 2% water solution of DMSO were included in each bioassay as control. The number of dead larvae was recorded after 24 h of exposure and the respective mortality percentage calculated.
All compounds were initially tested at 20 ppm. For compounds exhibiting mortality ≥10% after 24 h, various concentrations were used for larvicidal activity assessment, leading to a dose-response curve to determine their respective LC 50 and LC 90 values. For compounds with mortality percentages <10% after 24 h, only the % mortality at 20 ppm was recorded [55][56][57].

Data Analysis
Data obtained from each dose-larvicidal bioassay (total mortality per milligram per liter of concentration in water) were subjected to probit analysis in which probittransformed mortality was regressed against log10-transformed dose; LC 50 , LC 90 values and slopes were generated. All analyses were conducted using the statistical package SPSS 14.0 [58].

In Silico Molecular Properties and Descriptors Predictions
The cheminformatics program Canvas [59] of Schrödinger software was utilized to generate molecular descriptors and properties for the synthesized compounds. Particularly, Canvas computes a wide range of properties, including Topological, Physicochemical and Ligfilter descriptors. In total, more than 100 descriptors were calculated for all the examined compounds. Their values were used in an effort to compare the relationship between their experimental repellent activity and their descriptors profile. For this scope, all synthesized compounds were sketched in 2D form and then were subjected to energy minimization with OPLS2005 force field, using MacroModel program [60]. Minimization was performed with Powell-Reeves Conjugate Gradient (PRCG) method, using 1000 iterations and convergence threshold of 0.001 kcal/mol Å.