Malarial infection is caused by protozoan parasites belonging to the genus Plasmodium. There are five species of Plasmodium (P), P. falciparum, P. vivax, P. ovale, P. malariae and P. knowlesi. Of the five types, P. falciparum is the most dangerous and is responsible for majority of the global malarial deaths [1] [3] However, the most abundant type is P. vivax with exposure to approximately 2.5 billion humans across the globe [4] .

Plasmodium successfully thrives due to its lethal life cycle. The completion of this cycle is dependent upon two factors, the female Anopheles mosquito and humans [5] . According to Centers for Disease Control and Prevention [5] , Plasmodium life cycle can be divided into 12 stages, these stages are summarised below:

Plasmodium transmission occurs at stage 1 where an infected mosquito penetrates the human skin for a blood meal [3] [5] . Within 30 minutes, the parasite travels from the saliva of the mosquito to the host’s liver cells, where it begins to rapidly reproduce and eventually enter the blood stream to invade red blood cells, stages 2 - 6 [5] . A portion of the parasites invading the red blood cells develop into separate sex cells called gametocytes which are taken up by another mosquito during a blood meal, stage 7; this ultimately leads to the production of numerous parasitic cells in the mosquito’s salivary gland, stages 8 - 12, thus continuing the life cycle of the malarial parasite [5] .

Approved antimalarial drugs exert their effects at different stages of this life cycle with the overall aim of diminishing Plasmodium’s abundance within the human host. Antimalarial agents will be discussed in detail later.

Despite the vigorous research and development, as of 2017, there is currently no widely available antimalarial vaccine [6] . This is due to a combination of challenges such as the complex nature of Plasmodium’s life cycle, and the ever-changing antigens presented by the parasite [7] . An effective, commercially available vaccine would be a huge game changer; however, it seems like there is still a long way to go until that target is reached.

Antimalarial agents alone can reduce the incidence and prevalence of malaria, and they remain at the forefront of malaria control [8] [9] .

These drugs can be categorized into their two main therapeutic roles. Firstly, prompt use of these drugs not only treats an individual but also indirectly stops the development of gametocytes (stage 7 of the Plasmodium life cycle) hence blocking the spread of malaria. Secondly, antimalarial agents can be used as a preventative measure for people visiting or residing in endemic areas [9] . Table 1 displays the current available drugs grouped into their respective drug classes.

Although antimalarial drugs have a wide clinical application, they are surrounded by many issues that question their credibility. For many years, chloroquine was the gold standard treatment for uncomplicated malaria, however, this has drastically changed due to the global spread of chloroquine-resistant falciparum and vivax malaria [10] . Amodiaquine 3 and piperaquine 4 have similar structural similarities with chloroquine 1, as shown in Figure 1, therefore, these drugs are also subject to increasing resistance [9] .

The naturally occurring quinine 5 is the oldest antimalarial drug dating back to the seventeenth century. This drug along with other quinoline-based antimalarials such as chloroquine 1 and mefloquine 6 have widely reported adverse

psychiatric effects which includes sleep imbalances, mood changes, impulsivity, delusions, hallucinations and also mania; in extreme cases, suicidal thoughts and suicides were also reported [11] . Quinine has been known to be very hard to tolerate; therefore, its use is limited to severe malaria only [9] .

Primaquine 4 (Figure 2) was first licensed by the FDA in 1952 and has remained the only drug capable of eliminating dormant parasites residing in the host’s liver; unlike the previous quinoline-based drugs, primaquine 4 has rarely reported adverse psychiatric effects and is generally tolerable [11] . However, resistance to primaquine 4 has also been observed [12]

Pyrimethamine 8 was first introduced as a single drug, however, within a short period of time, resistance to P. falciparum and P. vivax was observed. Therefore, this drug was paired with another antifolate drug named sulfadoxine 7 which proved effective against chloroquine-resistant malaria, but again, resistance rapidly reappeared [9] . Chemical structures of both agents can be viewed in Figure 3.

More recently, lumefantrine 9 pyronaridine 10 (Figure 4) and some of the antimalarial agents mentioned previously have been partnered up with the indispensable artemisinin drugs (Figure 5) to create artemisinin-based combination therapies (ACT), see Table 2.

Artemisinins are a valuable set of tools in the battle against malaria. They have a quick onset of action and can be administered intravenously to treat individuals with severe malaria. The use of artemisinin drugs outside of ACT is highly discouraged by the WHO due to the fear of resistance. However, reports of resistance have been reported in South-east Asia [9] . The first reported case of ACT-resistance occurred 10 years ago in Cambodia and since then ACT-resistance is now predicted to cause over 116,000 deaths globally per annum and cost over $32 million per year to treat [13] .

The atovaquone 14/proguanil 15 (Figure 6) combination (Malarone®), is an effective but expensive prophylactic therapy which also has reports of resistance [9] .

Almost all antimalarial drugs have reported resistance, which results in a delayed or insufficient clearance of Plasmodium in the host’s blood; this may cause profound consequences upon the host’s wellbeing and increases the chances of transmission to another host [14] .

Despite the many attempts by humans to eradicate malaria, Plasmodium has always had the ability to mutate and navigate around the actions of antimalarial drugs. The urgency to combat resistance has increased as almost all the antimalarial drugs have recorded cases of resistance which has denounced the once indispensable agents. The race between the development of novel antimalarial drugs and the uprising of resistance has intensified, therefore, identifying new targets and developing new molecules that can interact with these targets is urgently required in order to produce novel antimalarial drugs [15] .

During the last three decades, computational methods have played a major role in drug discovery and design; in particular structure-based methods such as molecular docking has become an important technique [16] [17] .

Molecular docking studies have been routinely used to better understand current antimalarial drugs on the market. A study in 2013 found that quinoline-based drugs have better binding modes as non-competitive inhibiters than as competitive inhibiters to P. falciparum lactate dehydrogenase, a vital enzyme the parasite requires for energy [18] . This method has also paved the way for lead compounds to be considered for pharmaceutical research. A study in India found a lead compound with positive docking and pharmacophores features for leucine amino peptidase, an enzyme vital for the growth and development of P. vivax [19] .

AutoDockVina is a high speed, popular molecular docking programme which is compatible with UCSF Chimera, a molecular visualization software known for its numerous functions and capabilities [20] [21] . Both platforms will be at the heart of this study.

P. falciparum has been studied intensively due to its high mortality rate, however, P. vivax is more widely distributed around the world and brings a larger social economic burden, therefore, searching for an effective treatment against this parasite is also important [4] .

N-myristoyltransferase (NMT) is an essential enzyme found in P. vivax and humans, which catalyses the bonding of a saturated 14-carbon fatty acid named myristate to the N-terminal glycine of target proteins; this process is called myristoylation [22] . A number of essential proteins such as protein kinase 1 and glideosome found in malarial parasites need to undergo myristoylation to carry out their biological functions, therefore, by inhibiting NMT, the parasite should fail to thrive [23] . Variations between parasitic and human NMT makes it a highly promising selective target [24] .

Several published studies have aimed to discover potential inhibitors of Plasmodium vivax (Pv) NMT. A United Kingdom study in 2012 successfully discovered a hit compound named 3-butyl-4-((2-cyanoethyl)thio)-6-methoxy-2-methyl-quinoline using a high throughput screening technique involving a fluorogenic assay [24] . A similar study in the same year tested benzothiophene containing inhibitors using structure activity relationship techniques and discovered a lead compound for further development [23] .

There are several published studies involving PfNMT, however, the current lack of molecular docking studies involving PvNMT has encouraged this study to pursue this as the target protein of interest. The 3D structure of PvNMT (Figure 7).

The overall aim of this study was to prepare and evaluate a series of naphthoquinone and aryl diketone ligands for potential use as PvNMT inhibitors using computational docking techniques.

A range of ligands will be docked against PvNMT to yield an array of results. Firstly, naphthoquinone compounds will be docked to the target protein as they have been known to possess antimalarial activity [26] . There is an antimalarial drug currently on the market called atovaquone that belongs to the class of naphthoquinones, therefore, these compounds are a promising choice as ligands. Aromatic diketones, cromolyn and FDA approved antimalarial drugs will also be docked with PvNMT for comparison purposes.

The three-dimensional structure of P. vivax N-myristoyltransferase was downloaded as a PDB file from the RCSB Protein Data Bank (PDB ID: 5G22). This protein was opened in Auto Dock Tools version 1.5.6 where the co-ordinates of the binding pocket were determined using the grid box function. The figures used are shown below in Table 3.

The PDB file of the protein was opened in UCSF Chimera version 1.12 and all non-standard atoms and bonds were removed. Using the dock prep function, water molecules were deleted, partial charges were assigned and hydrogens were added to the protein. The protein was then saved in the Mol 2 format ready for the docking stage.

The naphthoquinones and aryl diketones compounds on PubChem were longlisted for the docking stage. The canonical SMILES of the compounds were

entered into UCSF Chimera using the build structure tool. The compound was then prepared by adding hydrogens using the structure editing function and moved the system towards its minimized energy using the minimize structure tool. The ligand was named “target molecule #” and saved in the Mol2 format ready for the docking stage.

On a new UCSF Chimera session, firstly, the Mol2 file of the protein was opened followed by the target moleculea. Using the structure/binding analysis tool, the AutoDockvina function was opened. The PDB format of the protein was selected as the receptor and the SMILES of the target molecule was selected as the ligand. An output file was created and the co-ordinates of the binding pocket, as shown in Table 3 was entered. All receptor and ligand options were set as true and the exhaustiveness of search was set at the maximum of 8. Also, the number of binding modes was selected as 9. Vina.exe via a local path was chosen as the executable location and docking was applied. All naphthoquinone and aryl diketone compounds were docked with P. vivax N-myristoyltransferase by following this method.

Analysis was carried out for all docking sessions on UCSF Chimera. All intra-residue hydrogen bonds were made visible by altering the hydrogen bond function on Chimera. The line width was kept at the default 4.0, however, the number of angstroms was altered to 3.0. Applying these changes displayed all hydrogen bonds (less than approximately 5 angstroms) between the protein and the 9 binding modes of the ligand. Each of the nine binding modes for all 166 ligands were analysed for any hydrogen bond(s) less than 3 angstroms. The amino acid residue involved in the hydrogen bond, its position and the number of angstroms of the bond were all recorded during the docking analysis phase.

Binding mode analysis were carried out for all docking sessions by relaunching the python files on UCSF Chimera. All intra-residue hydrogen bonds were made visible by altering the hydrogen bond function on Chimera. The line width was kept at the default 4.0, however, the number of angstroms was altered to 3.0. Applying these changes displayed all hydrogen bonds (less than approximately 5 angstroms) between the target protein and ligand across the 9 binding modes.

Angstroms is a unit of length equivalent to 0.1 nanometres and as the shorter the length of a hydrogen bond results in a stronger interaction, finding bonds with short angstroms will identify binding modes with good affinity [27] . By hovering over a hydrogen bond, the number of angstroms as well as the name and position of the amino acid involved in the bond were found (Figure 8). All hydrogen bonds across the 9 binding modes for all 166 ligands were analysed for any hydrogen bond(s) less than 3 angstroms. The amino acid residue involved in these hydrogen bonds, its position and the number of angstroms of the bond were recorded.

These data was collated and evaluated on Microsoft Excel and the short-listing process was undertaken to identify hit compounds.

As well as generating binding modes, AutoDockVina also uses a combination of knowledge-based and empirical scoring functions to predict the binding affinities of the ligands to the target protein; this is represented as the docking score [28] . These figures were obtained for the short-listed hit compounds from the View Dock window on UCSF Chimera (Figure 9). These data was collated and evaluated.

Christopher Lipinski formulated a tool named the rule of five (Ro5) to evaluate the drug-likeness of chemical compounds; he based this rule on the basis that most orally delivered drugs are small and moderately lipophilic molecules [29] . By obtaining the Ro5 parameters, the drug-likeness of the hit compounds was studied. This involved acquiring molecular weights from PubChem and cLogP values from an online software called ALOGPS 2.1. Also, the hydrogen bond donors (HBD) and hydrogen bond acceptors (HBA) for each binding mode displaying hydrogens under 3 angstroms were recorded from ViewDock.

Twelve FDA approved antimalarial drugs were also docked with PvNMT by following the same method used to dock the 166 ligands (Section 2.3). The hydrogen bond interactions and docking scores data were also obtained, recorded and evaluated by following the same methods (Sections 3.1 and 3.2). This data was used for comparative purposes.

All 166 target molecules, (labelled TM1-100 for naphthoquinones and TM101-166 for aryl diketones) were successfully docked to PvNMT. Analysis of the binding modes revealed a range in the number of hydrogen bonds below 3 angstroms (Å) between the set. Figure 10 displays this data and reveals 69 of the 166 target molecules (42%) displayed a lack of hydrogen bonds under 3Å. Therefore, the deduction can be made that these ligands have a lack of strong interactions with PvNMT.

The remaining 97 target molecules (58%) displayed at least one hydrogen bond (<3Å) across the 9 binding modes. As the number of the hydrogen bonds (<3Å) increases, the strength of the interactions also increases. Therefore, ligands with the highest number of hydrogen bonds (<3Å) were short-listed as potential hit compounds.

Figure 9 shows that the number of target molecules decreases with increasing hydrogen bonds (<3Å). This facilitated the short-listing process. Out of the 166 target molecules, six displayed more than 6 hydrogen bonds (<3Å) with PvNMT and were shortlisted as the top six hit compounds. The chemical structures of the hit compounds are shown in Figure 10.

The strengths and limitations of this study lies with the largest variable of this study, the molecular docking technique. AutoDockVina was the chosen docking software in this study, however, there are more than 60 developed docking tools available; some examples include DOCK, MOE, AutoDock and GOLD (Pagadala, Syed and Tuszynski, 2017). AutoDockVina was released in 2010 as a successor to AutoDock; the upgrade was promoted to possess twice the speed and an improved binding mode accuracy compared to its predecessor [21] .

All docking softwares including AutoDockVina are subject to the universal limitations of molecular docking. A major weakness is the accuracy of the scoring function which is one of the most important component of this computational technique. Various software’s utilise different scoring function techniques such as empirical, force-field and knowledge-based, which all carry limitations. However, several comparative studies have promoted the accuracy of AutoDockVina. A recent comparative study of docking softwares ranked AutoDockVina along with GOLD and MOE as the best tools for accurate prediction of ranking poses with the best docking scores [31] .