Int J Pharm Pharm Sci, Vol 7, Issue 10, 394-398Original Article


LIGNAN DERIVATIVES POTENTIAL AS PLASMODIUM FALCIPARUM LACTATE DEHYDROGENASE INHIBITORS: MOLECULAR DOCKING APPROACH OF ANTIPLASMODIAL DRUG DESIGN

ROSMALENA1,4, VIVITRI DEWI PRASASTY2, MUHAMMAD HANAFI3, EMIL BUDIANTO4, BERNA ELYA5

1Department of Medical Chemistry, Faculty of Medicine, University of Indonesia, Jakarta, Indonesia, 2Faculty of Biotechnology, Atma Jaya Catholic University of Indonesia, Jakarta, Indonesia, 3Research Center for Chemistry, Indonesian Institute of Sciences, Puspiptek, Serpong, Indonesia, 4Department of Chemistry, Faculty of Mathematics and Natural Sciences, University of Indonesia, Depok, Indonesia, 5Faculty of Pharmacy, University of Indonesia, Depok, Indonesia
Email: rosmalena@ui.ac.id    

 Received: 14 May 2015 Revised and Accepted: 02 Sep 2015

ABSTRACT

Objectives: To investigate the lignan derivatives potential as Plasmodium falciparum Lactate Dehydrogenase (PfLDH) inhibitors by using Computer Aided Drug Design (CADD) and molecular docking approach.

Methods: In finding potential antiplasmodial, in silico approach has been utilized. Protein structure of PfLDH has been built through homology modeling. Kobamin has been used to refine the 3D PfLDH structure. Structure validation of PfLDH was done by Ramachandran Plot and ERRAT calculations. The validated PfLDH was ready for molecular docking analysis. lignan derivatives as lead compounds were designed. The pharmacophore of lignan derivatives were assessed by using Molsoft drug likeness. Both protein and Lignan derivatives were docked with Autodock Vina. The best docking score was shown by the lowest affinity energy.

Results: Homology modeling of PfLDH has been built. Moreover, PfLDH refinement and validation were importantly conducted to ensure that PfLDH structure was in good quality. According to Ramachandran Plot and Procheck analysis, PfLDH has good structure quality with 93.39% confidence value. On the other hand, lignan derivatives assessment also has been done by evaluating their physicochemical and pharmacophore properties as lead compounds. From this assessment, it showed that Aristoligol (ARG1), Aristoligone (ARG2), and Ester Asetil Aristoligol (ARG3) showed good compounds to be drug likeness by following Lipinski’s rule of five (RO5), while Ester Butiril Aristoligol (ARG4) showed poor RO5 criteria. Bioavailability of four compounds was good in body metabolism, however, ARG3 and ARG4 could not be lead like compounds due to poor lead likeness value. From molecular docking result, the most favorable binding with PfLDH was ARG4 based on its affinity energy value (-8.0 kj/mol), followed by ARG3, ARG2 and ARG1, respectively.

Conclusions: The identification of potential anti plasmodial drugs was successfully accomplished by evaluating synthetic lignan derivatives compound through physicochemical properties and molecular docking analysis. Overall, physicochemical and pharmacophore properties showed good result. Molecular docking interaction has distinct mode interactions of lignan derivatives with PfLDH. We believe that these evaluated compounds could be used as anti plasmodial drugs according to in silico evaluation results.

Keywords: PfLDH, Lignan derivatives, Molecular docking, Antiplasmodial, CADD.

© 2016 The Authors. Published by Innovare Academic Sciences Pvt Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)

INTRODUCTION

Malaria is one of the most invasive tropical diseases in the world especially in most developing countries which affecting mankind and constituting a major public health issue. It affects about 40% of the world’s population, or equal approximately 216 million individuals over tropical and subtropical areas of developing countries and causes mainly morbidity and mortality about 800,000 death worldwide [1]. Moreover, approximately one third of the world’s population is at risk for exposing the disease. The World Health Organization (WHO) always strikes to actuate campaigns for global malaria eradication [2-4]. Malaria is a global important disease and worldwide burdens due to long-term limitations of vector control and the lack of an effective vaccine, makes therapeutic antimalarial drugs development becomes the serious main strategy to eliminate malaria [5].

The causative agents which responsible to malaria disease are protozoan parasites Plasmodium sp. In particular, Plasmodium falciparum has been known as the most deadly human malaria parasite, relies on the export of virulence factors to the surface of infected erythrocytes. Over the course of its intraerythrocytic developmental cycle (IDC), the pathogenicity of Plasmodium falciparum tightly affects the fluctuation of transcript levels for hundreds of genes [6-8]. P. falciparum lactate dehydrogenase (PfLDH) has been considered as a potential molecular drug target due to this parasite's dependence on glycolysis for energy production [9, 10]. Plasmodial lactate dehydrogenase is a terminal enzyme of the glycolytic pathway, has been shown to be biochemically, immunologically and structurally distinct from the mammalian enzyme [11]. Four Plasmodium species have been well known to infect human kind, including P. falciparum, P. vivax, P. ovale, and P. malariae. Moreover, P. knowlesi, has been recently documented to cause human malarial disease in Southeast Asia countries. Most severe diseases and deaths from malaria are caused by P. falciparum. It is responsible for about 80% of all malaria cases also responsible for about 90% of deaths caused by malaria [12-15].

By understanding a protein target mechanism in the parasite population is highly important to utilize it effectively in diagnostics and therapeutics [16]. The role of LDH in P. falciparum metabolism is a major checkpoint of anaerobic glycolysis, which form lactate from pyruvate reduction. This enzyme has recently become a great deal of attention as a valid therapeutic target for malaria disease. In fact, the P. falciparum’s isoform is a key enzyme for energy generation of malarial parasites. These species mostly depend on anaerobic glycolysis for energy production, since they lack a citric acid cycle for ATP formation. Therefore, PfLDH inhibition would potentially cause mortality of P. falciparum and it would be a great target to eradicate malaria [17]. Small new bioactive molecules have been recently designed and developed to inhibit target, PfLDH. Hence, inhibition of PfLDH may constitute an efficient way to develop therapeutic potentials [18-22].

Lignan derivatives are potential renewable sources as new anti-malarial drugs, since they contain a quantity of metabolites with a great variety of structures and pharmacological activities. Traditional preparations such as macerations, extracts, steam baths, concoctions, and decoctions from plant materials, have been the main source of malarial treatment in Indonesia and other developing countries where the disease is endemic [6,7]. Phoebe declinata Ness is a plant that has potential anti-malaria due to it contains Aristoligone lignan derivatives which have anti plasmodial activity in Holostylis reniformis plant [23]. However, the assessment of lignan derivatives to PfLDH is still unknown. The aim of this research is to investigate the lignan derivatives potential as PfLDH inhibitors by using molecular docking approach through Computer Aided Drug Design (CADD). The novelty of this study is finding the potential lignan derivatives which structurally modified against PfLDH. The main emphasis of this work is to identify the most potential drug candidates against PfLDH. We use high throughput platform such as pharmacophore properties [24, 25] and Lipinski’s rule of five [26] as complementary analysis in identifying the best lead compounds of lignan derivatives.

MATERIALS AND METHODS

Protein structure preparation

The amino acid sequences of Pf LDH (Entry No. Q27743) was generated from UNIPROT protein database [27, 28]. The 3D structure of protein PfLDH was generated by web based SWISS-MODEL program. All homology-modeling methods consist of the following four steps: (i) template selection; (ii) target template alignment; (iii) model building; and (iv) model evaluation. These steps are iteratively repeated, until a satisfying model structure is determined. The SWISS-MODEL server approach can be described as rigid fragment assembly [29-32].

Protein structure refinement and validation

PfLDH 3D structure was checked by using Procheck [33] to validate its refined structural conformation. Ramachandran plot [34-37] and ERRAT [38] were used to analyze the allowed dihedral phi and psi rotation of amino acids in the protein backbones and the quality of refined 3D structure, respectively.

Lignan derivative structures preparation

All four 3D lignan derivative structures were generated by Chem Draw Ultra 12.0 [39, 40] for the molecular docking experiments and their conformational energy were minimized by using MMFF94 force field. Four molecules of lignan derivatives were designed by substituting the–R positions of lignan. These molecules are: Aristoligol (ARG1), Aristoligone (ARG2), Ester Asetil Aristoligol (ARG3), and Ester Butiril Aristoligol (ARG4), respectively. The molecule structures are depicted in fig. 1. The structures were scored based on their physicochemical properties under Chemicalize (Chem Axon) [41] and Molsoft [42] platforms. These physicochemical properties are important for developing the drug candidate in every stages from design to pre-clinical study.

Molecular docking of Pf LDH-lignan derivative complexes

The preparative protein and ligand coordinates were saved as pdb files. Molecular docking experiment is performed using Autodock Vina program (Vina, The Scripps Institute) [43]. The Autodock Tools is used to add partial charges using Gasteiger method and to arrange the polar hydrogens in the protein. The ligands are set to have flexible torsion angles at all rotatable bonds, while the protein is prepared as a rigid structure. Both protein and ligand are saved as output pdbqt files. For specific docking of ligand lignan derivatives onto the Pf LDH protein, the grid box volume was adjusted to 40x40×40 Å in the x, y and z axes, respectively, with grid-sizes have a space up to 1 Å.

The binding energy values were calculated based on the total intermolecular energies (kj/mol) including hydrogen bond energy, Van Der Waals energy, desolvation energy and electrostatic energy. On the other hand, the apropriate torsion angles of ligand are also induced as internal ligand energy. The docking program will evaluate this energy to obtain the best binding mode. The Root-Mean-Square Deviation (RMSD) which less than 2.0 Å was scored during running docking program.

RESULTS AND DISCUSSION

Protein structure of PfLDH has been built from its amino acids sequence. Kobamin was used to refine PfLDH structure, showed better folding structure. PfLDH structure validation from Ramachandran plot (fig. 1) analysis proved that four amino acid residues with non-optimal values phi and psi have been located in alpha-and pi-turns of the LDH polypeptide chains, e. g., Asn1119, Asn75, Asn38 and Asn18.

However, residues in most favored region is 87.8%, indicated a valid 3D model structure (table 1). This validation was also supported by ERRAT analysis (fig.2). On the error axis, two lines are drawn to indicate the confidence of possible reject regions that exceed the error value. The ERRAT calculation is expressed as the percentage of the protein for which the calculated error value falls below the 95% rejection limit. Good high resolution structures generally produce values around 95% or higher. For lower resolution (2.5 to 3 Å) the average overall quality factor is around 91%. In this case, PfLDH structure with 93.39% was acceptable to be used for further in silico analysis through docking interaction.

3

Fig. 1: Ramachandran Plot of PfLDH


4


Fig. 2: ERRAT analysis of PfLDH 3D structure


Table 1: Ramachandran plot summary

Plot statistics

Score

Percent

Residues in most favored regions [A,B,L] (red)

815

87.8%

Residues in additional allowed regions [a,b,l,p] (yellow)

104

11.2%

Residues in generously allowed regions [~a,~b,~l,~p] (pale yellow)

5

0.5%

Residues in disallowed regions (white)

4

0.4%

Number of non-glycine and non-proline residues

928

100%

One of lignan derivatives was drawn from natural Aristoligone. The other three derivatives were considered as synthetic structures which one–R group was synthesized to the different side chain (fig. 3). In assessing the physicochemical and pharmacophore of lignan derivatives, we evaluate the lipophilicity of derivative compounds by using Chemicalize platform. This is a crucial parameter in drug development since it impacts both properties and target affinity of drug candidates. In early drug discovery stage, accurate tools for logP prediction are highly desired. LogP value is used to predict octanol/water partition coefficients of chemical compounds. Many calculation methods were developed to assist pharmaceutical areas in drug development. In this study, four compounds have logP values ranging from 4.65 to 6.06 (table 1). In general, an orally active drug has no more than one violation according to Lipinski`s rule of five, as followed criteria: 1. No more than 5 hydrogen bond donors (the total number of nitrogen–hydrogen and oxygen–hydrogen bonds); 2. No more than 10 hydrogen bond acceptors (all nitrogen or oxygen atoms); 3. A molecular mass less than 500 daltons; 4. An octanol-water partition coefficient; 5. logP not greater than 5 [44, 45]. From pharmacopore results showed that ARG1, ARG2 and ARG3 have followed Lipinski`s rule of five, but not ARG4 due to there were found more than two violations of Lipinski criteria.

5

Fig. 3: Four Lignan derivative structures


Table 2: Physicochemical and pharmacopore properties of lignan derivatives

Compound

Molecular formula

Molecular weight

Number of HBA/HBD

pH

LogP

PSA (A2)

Lipinski’s RO5

ARG1

C22 H28 O5

372.45

 5/1

7.4

3.64

57.15

Yes

ARG2

C22 H26 O5

370.18

 5/0

7.4

3.90

53.99

Yes

ARG3

C24 H30 O6

414.20

 6/0

7.4

4.08

63.22

Yes

ARG4

C26 H34 O6

442.24

 6/0

7.4

5.22

63.22

No

Table 3: Molecular docking interaction of lignan derivatives with PfLDH summary

Compound

Structure color

Docking Affinity Energy (kj/mol)*

Bioavailability

Lead Likeness

LogSw**

ARG1

Red

-7.4

Yes

Yes

-4.45

ARG2

Blue

-7.7

Yes

Yes

-5.32

ARG3

Cyan

-7.7

Yes

No

-4.82

ARG4

Magenta

-8.0

Yes

No

-5.73

Notes: *Autodock Vina analysis, **Molsoft analysis


1

Fig. 4: PfLDH-Lignan derivatives complex through molecular docking analysis, each compounds are colored differently (Green: PfLDH structure in folding ribbon, Red: ARG1, Blue: ARG2, Cyan: ARG3, Magenta: ARG4)

Furthermore, drug likeness is also assessed relied on bioavailability, solubility and lead likeness, respectively (table 2). Drug likeness is a qualitative calculation used in drug design for how similar to drug a substance is with respect to factors like bioavailability. It is estimated from the molecular structure before the substance is even synthesized and tested.

Molecules that violate more than one of Lipinski’s rules of five may have problems with bioavailability [46]. According to Veber rule, a good oral bioavailability should have Polar Surface Area (PSA) value less than 140 Å2 or less than 12 total hydrogen bonds (acceptors plus donors). The PSA values of lignan derivatives showed within the range of 57 to 64 Å2 (table 2). Therefore, all lignan derivatives presented good bioavailability. Moreover, to have a good lead likeness, lead compounds should have LogSw (logarithm of water solubility) less than-5. From LogSw value, we could see that ARG1 and ARG3 have LogSw less than-5, while ARG2 and ARG4 have LogSw more than-5. When synthetic lead compounds are planned to be modified for improving the water solubility or lead compounds permeability purpose, a wide parameter calculation of the structures could be used. It is important to clarify the structural modifications that increase solubility will also decrease drug permeability [47-49].

2

Fig. 5: Lignan derivative modes inside the PfLDH cavity (Grey: PfLDH pocket structure in inner surface form, Red: ARG1, Blue: ARG2, Cyan: ARG3, Magenta: ARG4)

Three dimensional structure of PfLDH were built in SWISS Model web platform which its amino acids sequence was retrieved from UNIPROT, initially. Energy minimization of PfLDH protein structure was employed by using Kobamin refinement platform. included energy minimized and molecular dynamics simulations. The refined PfLDH was then used for docking interaction analysis. Autodock Vina was used for the docking studies. The docked conformation corresponding to the lowest binding energy was selected as the most favorable binding conformation. The total screened four compounds were docked into the active site of PfLDH. The docking energies of all the compounds were represented in kcal/mol. The best mode of protein-ligand complexes are shown on fig.4. Four lignan derivative compounds showed low binding energies and significant affinities with target PfLDH protein (table 3). All ligand conformers were embedded within the active site cavity of protein target. These favored ligand modes were stabilized by hydrogen bonds between the functional group from the ligands with the functional group of side chain residues of PfLDH protein (fig. 5). The best docked compounds were shown with the lowest affinity energy. From docking result, it is shown that ARG4 (-8.0 kj/mol) has the best binding interaction compared to ARG3 (-7.7 kj/mol), ARG2 (-7.7 kj/mol) and ARG1 (-7.4 kj/mol), respectively (table 3). Relied upon this study, lignan derivatives compounds might be used as leads for developing effective antimalarial drugs. However, some poor bioavailability and pharmacopore could be structurally modified in the–R group to improve their bioactivities as lead compounds.

CONCLUSION

We have accomplished our investigation in identifying potential anti plasmodial drug candidates by evaluating synthetic lignan derivatives compound through physicochemical properties and molecular docking analysis. We suggested that all studied lignan derivative compound could be further validated in vivo and in vitro studies to confirm in silico evaluation. We believe that these evaluated compounds could be used as anti plasmodial drugs.

CONFLICT OF INTERESTS

Authors have none to declare

REFERENCES

  1. Malaria Policy Advisory Committee to the WHO: conclusions and recommendations of September 2012 meeting. Malar J 2012;11:424.
  2. Burrows JN. Designing the next generation of medicines for malaria control and eradication. Malar J 2013;12:187.
  3. Breman JG, AD Brandling-Bennett. The challenge of malaria eradication in the twenty-first century: research linked to operations is the key. Vaccine 2011;29 Suppl 4:D97-103.
  4. Malaria: control vs elimination vs eradication. Lancet 2011;378:1117.
  5. Harikishore A. Small molecule plasmodium FKBP35 inhibitor as a potential antimalaria agent. Sci Rep 2013;3:2501.
  6. Rai R. Genome-wide analysis in Plasmodium falciparum reveals early and late phases of RNA polymerase II occupancy during the infectious cycle. BMC Genomics 2014;15:959.
  7. Mbengue A. Novel plasmodium falciparum maurer's clefts protein families implicated in the release of infectious merozoites. Mol Microbiol 2013;88:425-42.
  8. Ouedraogo AL. Substantial contribution of submicroscopical plasmodium falciparum gametocyte carriage to the infectious reservoir in an area of seasonal transmission. PLoS One 2009;4:e8410.
  9. Penna-Coutinho J. Antimalarial activity of potential inhibitors of Plasmodium falciparum lactate dehydrogenase enzyme selected by docking studies. PLoS One 2011;6:e21237.
  10. Kaushal DC, NA Kaushal. Diagnosis of malaria by detection of plasmodial lactate dehydrogenase with an immunodot enzyme assay. Immunol Invest 2002;31:93-106.
  11. Kaushal NA, DC Kaushal. Production and characterization of monoclonal antibodies against substrate specific loop region of Plasmodium falciparum lactate dehydrogenase. Immunol Invest 2014;43:556-71.
  12. Bouyou-Akotet MK. Impact of plasmodium falciparum infection on the frequency of moderate to severe anaemia in children below 10 y of age in gabon. Malar J 2009;8:166.
  13. Kochar DK. A prospective study on adult patients of severe malaria caused by Plasmodium falciparum, Plasmodium vivax and mixed infection from Bikaner, northwest India. J Vector Borne Dis 2014;51:200-10.
  14. Sulaiman H. Severe plasmodium falciparum infection mimicking acute myocardial infarction. Malar J 2014;13:341.
  15. Treatment with quinidine gluconate of persons with severe Plasmodium falciparum infection: discontinuation of parenteral quinine from CDC Drug Service. MMWR Recomm Rep 1991;40:21-3.
  16. Keluskar P. Plasmodium falciparum and plasmodium vivax specific lactate dehydrogenase: genetic polymorphism study from Indian isolates. Infect Genet Evol 2014;26:313-22.
  17. Granchi C. Inhibitors of lactate dehydrogenase isoforms and their therapeutic potentials. Curr Med Chem 2010;17:672-97.
  18. Megnassan E. Design of novel dihydroxynaphthoic acid inhibitors of Plasmodium falciparum lactate dehydrogenase. Med Chem 2012;8:970-84.
  19. Choi SR. Design, synthesis, and biological evaluation of Plasmodium falciparum lactate dehydrogenase inhibitors. J Med Chem 2007;50:3841-50.
  20. Choi SR. Generation of oxamic acid libraries: antimalarials and inhibitors of Plasmodium falciparum lactate dehydrogenase. J Comb Chem 2007;9:292-300.
  21. Conners R. Mapping the binding site for gossypol-like inhibitors of Plasmodium falciparum lactate dehydrogenase. Mol Biochem Parasitol 2005;142:137-48.
  22. Deck LM. Selective inhibitors of human lactate dehydrogenases and lactate dehydrogenase from the malarial parasite plasmodium falciparum. J Med Chem 1998;41:3879-87.
  23. de Andrade-Neto VF. Antiplasmodial activity of aryltetralone lignans from Holostylis reniformis. Antimicrob Agents Chemother 2007;51:2346-50.
  24. Vyas VK. Design, synthesis, pharmacological evaluation and in silico ADMET prediction of novel substituted benzimidazole derivatives as angiotensin II-AT1 receptor antagonists based on predictive 3D QSAR models. SAR QSAR Environ Res 2014;25:117-46.
  25. Zhang Y. De novo design of N-(pyridin-4-ylmethyl)aniline derivatives as KDR inhibitors: 3D-QSAR, molecular fragment replacement, protein-ligand interaction fingerprint, and ADMET prediction. Mol Divers 2012;16:787-802.
  26. Nogara PA. Virtual screening of acetylcholinesterase inhibitors using the Lipinski's rule of five and ZINC databank. Biomed Res Int 2015;1-8. doi: 10.1155/2015/870389. [Article in Press]
  27. UniProt: a hub for protein information. Nucleic Acids Res 2015;43:D204-12.
  28. Alpi E. Analysis of the tryptic search space in UniProt databases. Proteomics 2015;15:48-57.
  29. Biasini M. SWISS-MODEL: modelling protein tertiary and quaternary structure using evolutionary information. Nucleic Acids Res 2014;42:W252-8.
  30. Bordoli L. Protein structure homology modeling using SWISS-MODEL workspace. Nat Protoc 2009;4:1-13.
  31. Arnold K. The SWISS-MODEL workspace: a web-based environment for protein structure homology modelling. Bioinformatics 2006;22:195-201.
  32. Schwede T. SWISS-MODEL: An automated protein homology-modeling server. Nucleic Acids Res 2003;31:3381-5.
  33. Laskowski RA. AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. J Biomol NMR 1996;8:477-86.
  34. Carugo O, K Djinovic-Carugo. A proteomic ramachandran plot (PRplot). Amino Acids 2013;44:781-90.
  35. Gopalakrishnan K. Ramachandran plot on the web (2.0). Protein Pept Lett 2007;14:669-71.
  36. Ho BK, A Thomas, R Brasseur. Revisiting the ramachandran plot: hard-sphere repulsion, electrostatics, and H-bonding in the alpha-helix. Protein Sci 2003;12:2508-22.
  37. Kolaskar AS, S Sawant. Prediction of conformational states of amino acids using a ramachandran plot. Int J Pept Protein Res 1996;47:110-6.
  38. Colovos C, TO Yeates. Verification of protein structures: patterns of nonbonded atomic interactions. Protein Sci 1993;2;1511-9.
  39. Cousins KR. Computer review of chem draw Ultra 12.0. J Am Chem Soc 2011;133:8388.
  40. Li Z. Personal experience with four kinds of chemical structure drawing software: review on chem draw, Chem Window, ISIS/Draw, and Chem Sketch. J Chem Inf Comput Sci 2004;44:1886-90.
  41. Southan C, A Stracz. Extracting and connecting chemical structures from text sources using chemicalize. org. J Cheminf 2013;5:20.
  42. Fernandez-Recio J, M Totrov, R Abagyan. Screened charge electrostatic model in protein-protein docking simulations. Pac Symp Biocomput 2002;552-63.
  43. Trott O, AJ Olson. Auto dock vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem 2010;31:455-61.
  44. Martel S. Large, chemically diverse dataset of logP measurements for benchmarking studies. Eur J Pharm Sci 2013;48:21-9.
  45. Korinth G. Potential of the octanol-water partition coefficient (logP) to predict the dermal penetration behaviour of amphiphilic compounds in aqueous solutions. Toxicol Lett 2012;215:49-53.
  46. Smith RN. The general problem of drug bioavailability and its assessment in man. Postgrad Med J 1974;50:7-14.
  47. Savjani KT, AK Gajjar, JK Savjani. Drug solubility: importance and enhancement techniques. ISRN Pharm 2012: doi: 10.5402/2012/195727. [Article in Press]
  48. Widanapathirana LS Tale, TM Reineke. Dissolution and solubility enhancement of the highly lipophilic drug phenytoin via interaction with poly(N-isopropylacrylamide-co-vinylpyrrolidone) excipients. Mol Pharm 2015;12:2537-43.
  49. Docherty R, K Pencheva, YA Abramov. Low solubility in drug development: de-convoluting the relative importance of solvation and crystal packing. J Pharm Pharmacol 2015;67:847-56.