Linköping University | Department of Physics, Chemistry and Biology Bachelor’s thesis, 16 hp | Master of Science in Chemical Biology: Physics, Chemistry and Biology Spring term 2020 | LITH-IFM-G-EX--20/3791--SE
Characterisation of Potential Inhibitors of
Calmodulin from Plasmodium falciparum
Julia Bjers, Mohamed Hassan, Alexandra Iversen, Ebba Nordén, Filippa
Wickström, Martin Zhou
Examiner, Magdalena Svensson Supervisor, Cecilia Andrésen
Avdelning, institution
Division, Department
Department of Physics, Chemistry and Biology Linköping University
URL för elektronisk version
ISBN
ISRN: LITH-IFM-G-EX--20/3791--SE
_________________________________________________________________
Serietitel och serienummer ISSN
Title of series, numbering ______________________________
Språk Language Svenska/Swedish Engelska/English ________________ Rapporttyp Report category Licentiatavhandling Examensarbete C-uppsats D-uppsats Övrig rapport _____________ Titel Title
Characterisation of Potential Inhibitors to Calmodulin from Plasmodium falciparum
Författare
Author
Julia Bjers, Mohamed Hassan, Alexandra Iversen, Ebba Nordén, Filippa Wickström, Martin Zhou
Nyckelord
Keyword
Plasmodium falciparum, Calmodulin, Malaria, Trifluoperazine, Calmidazolium, Artemisinin, Fentanyl derivatives, Modelling, Docking, Bioinformatics
Sammanfattning
Abstract
Each year countless lives are affected and about half a million people die from malaria, a disease caused by parasites originating from the Plasmodium family. The most virulent species of the parasite is Plasmodium falciparum (P. falciparum).
Calmodulin (CaM) is a small, 148 amino acid long, highly preserved and essential protein in all eukaryotic cells. Previous studies have d etermined that CaM is important for the reproduction and invasion of P. falciparum in host cells. The primary structure of human CaM (CaMhum) and CaM from P. falciparum (CaMpf) differ in merely 16 positions, making differences in their structures and ligand affinity interesting to study. Especially since possible inhibitors of CaMpf in favor of CaMhum, in extension, could give rise to new malaria treatments.
Some antagonists, functioning as inhibitors of CaM, have already been analysed in previous studies. However, there are also compounds that have not yet been studied in regards to being possible antagonists of CaM. This study regards three known antagonists; trifluoperazine (TFP), calmidazolium (CMZ) and artemisinin (ART) and also three recently created fentanyl derivatives; 3-OH-OMe-cyclopropylfentanyl (ligand 1), 4-OH-3OMe-4F-isobutyrylfentanyl (ligand 2) and 3-OH-4-OMe-isobutyrylfentanyl (ligand 3).
Bioinformatic methods, such as modelling and docking, were used to compare the structures of CaMhum and CaMpf as well as observe the interaction of the six ligands to CaM from both species. In addition to the differences in primary structure, distinguished with ClustalW, disparities in tertiary structure were observed. Structure analysis of CaMhum and CaMpf in PyMOL disclosed a more open conformation as well as a larger, more defined, hydrophobic cleft in CaMhum compared to CaMpf. Simulated binding of the six ligands to CaM from both species, using Autodock 4.2, indicated that TFP and ART bind with higher affinity to CaMhum which is expected. Ligand 2 and ligand 3 also bound with higher affinity and facilitated stronger binding to CaMhum, which is reasonable since their docking is based on how TFP binds to CaM. However, ligand 1 as well as CMZ both bound to CaMpf with higher affinity. Despite promising results for ligand 1 and CMZ, no decisive conclusion can be made solely based on bioinformatic studies.
To gain a better understanding on the protein-ligand interactions of the six ligands to CaMhum and CaMpf, further studies using e.g. circular dichroism and fluorescence would be advantageous. Based on the results from this study, future studies on the binding of CMZ and ligand 1 to CaM as well as ligands with similar characteristics would be especially valuable. This is because they, based on the results from this study, possibly are better inhibitors of CaMpf than CaMhum and thereby could function as possible antimalarial drugs.
Datum
Date 2020-06-01
Abstract
Each year countless lives are affected and about half a million people die from malaria, a disease caused by parasites originating from the Plasmodium family. The most virulent species of the parasite is Plasmodium falciparum (P. falciparum).
Calmodulin (CaM) is a small, 148 amino acid long, highly preserved and essential protein in all eukaryotic cells. Previous studies have determined that CaM is important for the reproduction and invasion of P. falciparum in host cells. The primary structure of human CaM (CaMhum) and
CaM from P. falciparum (CaMpf) differ in merely 16 positions, making differences in their
structures and ligand affinity interesting to study. Especially since possible inhibitors of CaMpf in
favor of CaMhum, in extension, could give rise to new malaria treatments.
Some antagonists, functioning as inhibitors of CaM, have already been analysed in previous studies. However, there are also compounds that have not yet been studied in regards to being possible antagonists of CaM. This study regards three known antagonists; trifluoperazine (TFP), calmidazolium (CMZ) and artemisinin (ART) and also three recently created fentanyl
derivatives; 3-OH-4-OMe-cyclopropylfentanyl (ligand 1), 4-OH-3OMe-4F-isobutyrylfentanyl (ligand 2) and 3-OH-4-OMe-isobutyrylfentanyl (ligand 3).
Bioinformatic methods, such as modelling and docking, were used to compare the structures of CaMhum and CaMpf as well as observe the interaction of the six ligands to CaM from both species.
In addition to the differences in primary structure, distinguished with ClustalW, disparities in tertiary structure were observed. Structure analysis of CaMhum and CaMpf in PyMOL disclosed a
more open conformation as well as a larger, more defined, hydrophobic cleft in CaMhum
compared to CaMpf. Simulated binding of the six ligands to CaM from both species, using
Autodock 4.2, indicated that TFP and ART bind with higher affinity to CaMhum which is expected. Ligand 2 and ligand 3 also bound with higher affinity and facilitated stronger binding to CaMhum, which is reasonable since their docking is based on how TFP binds to CaM.
However, ligand 1 as well as CMZ both bound to CaMpf with higher affinity. Despite promising
results for ligand 1 and CMZ, no decisive conclusion can be made solely based on bioinformatic studies.
To gain a better understanding on the protein-ligand interactions of the six ligands to CaMhum and
CaMpf, further studies using e.g. circular dichroism and fluorescence would be advantageous.
Based on the results from this study, future studies on the binding of CMZ and ligand 1 to CaM as well as ligands with similar characteristics would be especially valuable. This is because they, based on the results from this study, possibly are better inhibitors of CaMpf than CaMhum and
Acronyms and abbreviations
ACT ART-based combination therapy
ApoCaM Calcium free calmodulin, closed form
ART Artemisinin
C-lobe Amino acid 76-148 in calmodulin
Ca2+ Calcium ion
CaM Calmodulin
CaMc C-lobe of CaM (amino acid 76-148)
CaMchum C-lobe of human CaM (amino acid 76-148)
CaMcpf C-lobe of CaM from P. falciparum (amino acid 76-148)
CaMhum Human CaM
CaMn N-lobe of CaM (amino acid 1-75)
CaMnhum N-lobe of human CaM (amino acid 1-75)
CaMnpf N-lobe of CaM from P. falciparum (amino acid 1-75)
CaMpf CaM from P. falciparum
CD Circular dichroism
CMZ Calmidazolium
E. coli Escherichia coli
FRET Fluorescence resonance energy transfer
GANTT Flow plan chart for project management
HoloCaM Calcium bound calmodulin, open form
IC50 Half-maximal inhibitory concentration
IEX Ion exchange chromatography
IMRAD Introduction, method, results, analysis and discussion
IPTG Isopropyl-β-D-thiogalactopyranoside Kd Dissociation constant Ki Inhibition constant Ligand 1 3-OH-4-OMe-cyclopropylfentanyl Ligand 2 4-OH-3OMe-4F-isobutyrylfentanyl Ligand 3 3-OH-4-OMe-isobutyrylfentanyl
N-lobe Amino acid 1-75 in calmodulin
OD Optical density
P. falciparum Plasmodium falciparum
PDB Protein data bank
pI Isoelectric point
PyMOL 3D visualisation system tool for molecules
RMSD Root mean square deviation
SDS-PAGE Sodium dodecyl sulphate polyacrylamide gel electrophoresis
SEC Size exclusion chromatography
SL Sesquiterpene lactone
TFP Trifluoperazine
VdW Van der Waals
Table of content
1.INTRODUCTION 1
1.1 Purpose 1
1.2 Background 1
1.2.1 Malaria 1
1.2.2 Connections between malaria and calmodulin 1
1.2.3 Inhibitors of Calmodulin 2 2.THEORY 3 2.1 Plasmodium falciparum 3 2.2 Calmodulin 3 2.2.1 Primary structure 4 2.2.2 Tertiary structure 4 2.3 Ligands 5 2.3.1 Trifluoperazine 5 2.3.2 Calmidazolium 6 2.3.3 Artemisinin 7 2.3.4 Fentanyl derivatives 8
2.4 Modelling and docking 9
2.5 Protein purification 10
2.6 Circular dichroism 11
2.7 Fluorescence 11
3.SYSTEM AND PROCESS 13
3.1 Project plan 13
3.1.1 Flow plan 13
3.1.2 Risk analysis 13
3.2 Plan for systematic follow up 16
4.MATERIAL AND METHODS 17
4.1 Literary studies 17
4.2 Bioinformatics 17
4.2.1 Material 17
4.2.2 Modelling of calmodulin 18
4.2.3 Analysing differences in structure in calmodulin from P. falciparum and human calmodulin 18 4.2.4 Conversion of the ligands into 3D-structures used for docking 18
4.2.5 Docking of ligands to calmodulin 18
5.RESULTS 20
5.2 Structure comparison of calmodulin from P. falciparum and human calmodulin 20
5.2.1 Primary structures differ in 16 positions 20
5.2.2 Differences in hydrophobicity and charge distribution affect the tertiary structure 22 5.3 Ligand docking to calmodulin from P. falciparum and human calmodulin 26 5.3.1 Three antagonists trifluoperazine, calmidazolium and artimisinine 28
5.3.2 Multiple trifluoperazine 31
5.3.3 Three fentanyl derivatives 34
6.DISCUSSION 38
6.1 Structure analysis of calmodulin from P. falciparum and human calmodulin 38
6.1.1 Differences in primary structure 38
6.1.2 Calmodulin from P. falciparum has a more closed conformation than human calmodulin 39 6.2 Ligand binding to calmodulin from P. falciparum and human calmodulin 40
6.2.1 Binding of the three antagonists 40
6.2.2 Effects of binding multiple trifluoperazine 42
6.2.3 Binding of fentanyl derivatives 43
6.2.4 Ligands for potential antimalarial drugs 44
6.3 Flow plan and risk analysis 44
6.4 Future prospects and global effects 45
7.CONCLUSIONS 47
8.ACKNOWLEDGEMENTS 48
1
1. Introduction
1.1 Purpose
The purpose of this study is to examine how holoCaM interacts with six different ligands and thereby analyse the differences in interaction between CaMhum and CaMpf. Differences in
structure between the species and interactions will be studied using bioinformatic tools.
Henceforth, when CaM is mentioned in the study, it refers to holoCaM. Three different fentanyl derivatives developed by Linköping university will be characterised along with the three known antagonists TFP, CMZ and ART.
The mosquito-borne infectious disease, malaria, is outspread in large parts of the world and P.
falciparum is the parasite that mainly contributes to the disease [1]. CaM is an essential protein
for both humans and P. falciparum and by studying the differences in interaction of inhibitors to CaMhum and CaMpf, a potential drug against malaria can be found if the ligands have higher
affinity when binding to CaMpf [2].
1.2 Background
1.2.1 Malaria
Malaria is a widespread disease in the tropical and subtropical regions of the world. The disease spreads through mosquitos that are infected by a parasite, mainly P. falciparum in Africa. Incredibly many lives are affected by the disease and a lot of people die from it each year where most deaths strike children in Africa. However, in the last years, the deaths have decreased with the aid of more effective treatment, better diagnostics and mosquito nets [1].
The malaria parasite originates from the Plasmodium family. There are five known species that infect humans; P. vivax, P. falciparum, P. knowlesi, P. ovale and P. malariae [3]. P. falciparum is the most virulent of these and causes a severe public health threat [4]. It has the ability to avoid the human immune system by cytoadherence of the microvasculature [3]. The parasite infects millions of people worldwide and causes hundreds of thousands of deaths annually [4].
1.2.2 Connections between malaria and calmodulin
CaM is a highly conserved and multifunctional protein in all eukaryotic cells. It is a small Ca2+
dependent protein containing 148 amino acids. [2]. When Ca2+ binds to CaM, a conformational
change of the protein occurs and hydrophobic regions of the protein are exposed, allowing e.g. inhibitors to bind. CaM controls several important processes, such as ion transport, apoptosis and metabolic homeostasis. Because of this, CaM is essential in all eukaryotic cells, including P.
falciparum [5]. This makes it a good target for malaria drugs. If CaMpf is inhibited, the parasite
dies and malaria will consequently be eradicated. However, if CaMhum is inhibited as well as
CaMpf, both species lose an essential protein and die. Therefore, CaMpf and CaMhum will be
2
1.2.3 Inhibitors of Calmodulin
Antagonists of CaM inhibit the activity of CaM by binding directly to the protein. To inhibit the activity of CaM the antagonists usually have three important characteristics in common; lipid solubility, ionic interactions with CaM and surtain geometric structures [6].
For antagonists to bind to the exposed hydrophobic region, they need to be lipid soluble and have hydrophobic groups so that they can interact with the hydrophobic regions of CaM. Therefore, benzene rings are common groups for antagonists of CaM. Groups that increase the lipid solubility of the benzene ring are also common in CaM antagonists. Therefore, for instance chlorine groups are commonly bound to the benzene ring of CaM antagonists [7].
The pI of both CaMhum and CaMpf is approximately 4, meaning that it is acidic and has a net
negative charge at physiological pH. The net negative charge of CaM means that ionic interactions can be formed with positively charged antagonists. Therefore, a lot of CaM antagonists are amphipathic amines with a positive charge at physiological pH [7].
However, there are multiple molecules that are both lipid soluble with hydrophobic groups and positively charged but they are still not strong CaM antagonists. This is because the geometric structure of the molecule plays an important role. The common structure of CaM antagonists consist of a large hydrophobic region, usually made up of multiple aromatic rings, and a positively charged amino group that is positioned at least four atoms away from the aromatic rings [7]. Figure 1 shows a general structure of many CaM antagonists.
Figure 1: General structure of CaM antagonists. Consists of aromatic rings and a positively charged
group (N+). The length of the carbon chain, n, is usually at least 3 carbons long. R are hydrophobic
3
2. Theory
2.1 Plasmodium falciparum
P. falciparum is known for its complex life cycle, since these pathogens use both mosquitoes and
humans as hosts for infection. The parasite initiates infection by the bite and injection of an anopheline mosquito vector. It further develops in the liver and in the erythrocytes of the human host. This is followed by several repeated steps in the erythrocytes where the blood stage
parasites replicate cyclically [3]. CaM has shown to play an essential role in the invasion and reproduction of P. falciparum in the erythrocytes of the host [8].
P. falciparum tends to replicate asexually in each reproductive cycle, whereof a minor part
converts into sexual forms, i.e. gametocytes. Conversion between the replication forms of the parasite is vital for productive transmission between the hosts. The replicative cycle in the erythrocytes runs for 48 hours and results in the production of merozoites. The merozoites are then released in the bloodstream ready to invade new red blood cells [3]. Each cycle leads to increased parasite levels of P. falciparum and often results in characteristic symptoms of malaria, e.g. fever. In addition, severe forms of malaria can cause anemia. The way the disease progresses and the severity of the parasite infection, from mild to aggressive, seems to depend on genetic factors [9].
2.2 Calmodulin
CaM has two lobes, the N-lobe and the C-lobe, and is divided into two homologous domains which are affiliated by a linker, see Figure 2. In this report, CaMn is defined as amino acid 1-75
while CaMc is defined as amino acid 76-148. The linker makes the two lobes in CaM more
flexible, which makes it possible to study them individually. Each of the two lobes consists of two helix-loop-helix motifs known as EF-hands and they can bind one Ca2+ each [5].
Figure 2: CaMhum visualised in PyMOL. The green spheres represent calcium, CaMhum colored in
4 Amino acids in position 1, 3, 5, 7, 9 and 12 in the Ca2+ binding EF-hands are responsible for Ca2+
binding. The affinity for Ca2+ is higher in EF-3 and EF-4, positioned in CaM
c, than for EF-1 and
EF-2 in CaMn [2]. Therefore, CaMc is an appropriate binding site for target peptides. CaM in its
inactive form, without Ca2+, has its hydrophobic parts turned inwards. When Ca2+ binds to
apoCaM, the changes in conformation yields an open structure. The two lobes can take
alternative conformational changes [5]. In this new open conformation, an active site is exposed and target peptides can bind to CaM. Several experimental studies confirm the hypothesis that CaM undergoes conformational changes when binding of target peptides [10].
2.2.1 Primary structure
CaMhum and CaMpf both have a molecular weight of approximately 17 kDa and a pI around 4, as
have been mentioned earlier and can be seen in Table 1. CaM from both species is characterised as a relatively small and acidic protein. The number of amino acids is 148, whereof position one is counted from alanine since the N-terminal methionine is removed post translational [11].
Table 1: Comparison of CaMhum and CaMpf with ProtParam.
CaMhum CaMpf
Nr of amino acids 148 148
Molecular weight (Da) 16703,39 16799,57
Theoretical pI 4,09 4,11
The sequences differ in 16 amino acid positions, which manually can be calculated to 89% sequence identity between CaMhum and CaMpf, see Figure 3. The high conservation observed for
the two different CaM sequences establishes the essential role of the protein in eukaryotes.
Figure 3: Multiple sequence alignment with ClustalW. “CaM_human” represents the sequence for
CaMhum and “CaM_p.falciparum“ represents the sequence for CaMpf. Stars “*” correspond to identical
amino acids, colons “:” indicate not as significant differences in sequence and blank spaces “ “ indicate significant alterations in sequence.
From the 16 differing positions, there are five alterations in CaMn, the eleven other alterations are
located in CaMc. There are alterations that affect the electrical charge, others that affect the
hydrophobicity, and those that affect both. The alterations marked as not as significant are
considered to be of less significance since the amino acids have similar properties and thereby do not affect the characteristics of the amino acids as much. However, they may still affect ligand binding [12].
2.2.2 Tertiary structure
The tertiary structure of CaM is of interest when comparing CaMhum to CaMpf since they have a
5 conformations, reactive loops and hydrophobic clefts. Target peptides usually bind to the hydrophobic cleft, making the solvent accessible area of hydrophobic regions interesting. More specifically, the surface area of methionine in the hydrophobic cleft is of great interest since studies have proven that many antagonists interact with methionine. Each lobe in CaMhum
contains four methionine and one additional is positioned in the linker between the lobes.
Together these make up nearly half of the hydrophobic surface area, which makes them probable binding targets for the natural substrates of CaM and also for antagonists [13]. In CaMpf, there are
two variations where a methionine is exchanged, M71L and M76L, see Figure 3. These alterations might still affect ligand binding properties, even though they are considered not as significant as others, due to a distinction in the degree of hydrophobicity [13,14].
2.3 Ligands
2.3.1 Trifluoperazine
The open form of CaM with four Ca2+ bound can bind different drugs, one being TFP. TFP is a
trifluoromethyl phenothiazine derivative, see structure in Figure 4, and is used as an antipsychotic drug, especially when treating mental disorders such as schizophrenia [15]. It has a molecular weight of 0,41 kDa [16].
Figure 4. Structure of TFP retrieved from PubChem. A) Represents the 2D-structure of TFP and B) represents the 3D-structure of TFP.
TFP functions as a disrupter, an inhibitor of CaM, by interfering with the association of a target protein to CaM. The active form of the CaM-TFP complex appears in three different structures with binding ratios of 1:1 (1CTR), 2:1 (1A29) and 4:1 (1LIN), see Figure 5. Binding of TFP occurs in the hydrophobic cleft of CaMc in the structures 1CTR and 1A29. The structure 1LIN
binds three TFP to CaMc and one to CaMn. Through studies, it has been proven that TFP has
higher affinity for CaMc than for CaMn. Ligands can bind to apoCaM as well. However, instead
6 Figure 5. CaM structure with PDB-id A) 1CTR, with one TFP bound B) 1A29, with two TFP bound C) 1LIN, with four TFP bound visualized in PyMOL.
Respective structures have a RMSD value of 0,7 Å, which suggests identical conformation of the backbone. However, different orientations of a specific group in TFP, the trifluoro group, have been observed in previous studies. For the 1A29 and 1LIN structures, the trifluoro grouphas been observed to be embedded in the cleft located on the C-lobe. For the 1CTR structure however, the trifluoro group seems to protrude from the cleft instead [15].
Studies regarding methionine have shown its importance regarding binding of various target proteins in the hydrophobic clefts of CaM. One of its most significant benefits is allowing targets and drugs to be recognized by CaM, especially TFP, mainly due to the manner in which its sulfur atom and methionines attract nonpolar groups. Compared to other methionines, M144 is the most variable one. It is located in one of the binding sites of TFP, CaMc, together with three other
methionines, M109, M124 and M145. M124 and M144 form a tetrad with F92 and L105, representing a motif which has contact with the bound TFP. Depending on how the trifluoro groups are orientated different conformations of M144 can occur [15].
2.3.2 Calmidazolium
CMZ is a lipophilic antagonist that inhibits the activity of CaM by mechanisms involving immediate binding to CaM. Like most CaM antagonists it is a compound containing a hydrophobic benzene-ring structure and also, in the center of CMZ a positively charged
imidazolium structure at neutral pH, as can be seen in Figure 6. Both structures are important for the ability of CMZ to bind to CaM and inhibit it [17].
7 Figure 6. Structure of CMZ retrieved from PubChem. A) Represents the 2D-structure of CMZ and B) represents the 3D-structure of CMZ.
CMZ is a large inhibitor of CaM with a molecular weight of 0,65 kDa [18]. It has previously been shown to bind to CaM through the side chains of A15, L18, F19, V35, M36, L39, M51, M71, M72 and M76 in CaMn and A88, V91, F92, V108, M109, L112, F141 and M145 in CaMc.
Thus, CMZ binds and interacts with both lobes simultaneously. The methionines of CaM that immobilizes the CMZ binding are in, or near the binding sites of CMZ [19].
CMZ is one of the strongest antagonists of CaM. It inhibits CaM 300 times better than TFP, in regards to IC50. CMZ has non-specific effects in addition to inhibiting CaM, e.g. effects on the
blockade of L-type Ca2+ channels, K+ channels and Na+ channels. Hence, if the concentration of
CMZ is high it could lead to other pharmacologic effects. CMZ is also a dominant inhibitor of Ca2+-ATPase in red blood cells since the enzyme is regulated by CaM [17]. Studies have shown
that CMZ does not only inhibit CaM through binding to the protein but also by interfering with upstreams and downstreams targets of CaM signaling [20].
2.3.3 Artemisinin
ART is a small molecule with multiple aromatic groups, see Figure 7, and a molecular weight of 0,28 kDa [21]. It is classified as an SL, a subclass of the sesquiterpenoids. The SLs originate from three isoprene units and contain a particularly reactive lactone ring [22]. These are common in many plants from the family Asteraceae. Other plant families with SLs have been identified and ART can be extracted from the plant Artemisia annua, also called sweet wormwood. SLs are an effective group of drugs that have shown to play a significant role in human therapy i.a. in the treatment of cancer, cardiovascular disease and malaria, due to their toxicity [23].
8 Figure 7. Structure of ART retrieved from PubChem. A) Represents the 2D-structure of ART and B) represents the 3D-structure of ART.
ART has, together with its derivatives, established the role as one of the most effective drugs for malaria worldwide and are selectively toxic to P. falciparum. These compounds have shown to use the mechanism of heme mediated cleavage of the endoperoxide bridge, which tend to release highly reactive carbon free radicals [24]. The quick and effective mechanism of ART is not yet well understood and contradictions occur from studying the interactions of the substance to CaMpf and CaMhum. These concerns refer to the requirement of both haem and ferrous iron for the
activation of ART and to determining the exact protein targets of activated ART [4]. Several experiments of ART have shown to react with hemin and cause oxidation of protein thiols in the presence of erythrocyte membranes. Malaria parasites tend to contain a large amount of hemin and that is a plausible reason for why ART is particularly toxic for P. falciparum [25]. Due to the high efficiency and toxicity, ART has shown a few side effects and unfortunately long-term treatment has shown increased side effects. Also, resistance has been reported from countries where ART is commonly used and P. falciparum has become resistant in several areas [26].
ART and its derivatives are powerful for malaria treatment by effectively reducing the number of
P. falciparum in the blood of malaria patients. ACT has been recommended by WHO for further
studies. ACT is a combination of ART derivatives and a partner drug that eliminates the rest of the parasites remaining. As a treatment for malaria that is already in effect, ART is an interesting ligand to study further [27].
2.3.4 Fentanyl derivatives
Fentanyl derivatives are commonly used as an alternative approach to the treatment of pain. There are not many articles published about fentanyl derivatives in regards to their ability to inhibit CaM. This is because they were recently produced by a research group at Linköping University. Their hypothesis is also that the three fentanyl derivatives are possible antagonists to CaM.
The chemical structures for ligand 1, ligand 2 and ligand 3 are very similar, see Figure 8. Ligand 1 and ligand 3 are much alike but differ in the aspect that ligand 1 contains a cyclopropyl group positioned to the left, as can be seen in A in Figure 8. Ligand 2 is quite alike ligand 3, except the ether- and hydroxyl group have switched position in the benzene rings. There is also an extra
9 fluor atom linked to the other benzene ring in ligand 2, shown by the light blue colored element sticking out from the benzene ring in B in Figure 8.
Figure 8. The fentanyl derivatives in 3D visualised in PyMOL. Shown in sticks and colored by element. A) The chemical structure for ligand 1. B) The chemical structure for ligand 2. C) The chemical structure for ligand 3.
The three fentanyl derivatives are also similar in structure to the CaM inhibitors W-7 and lubeluzole. Therefore, the binding of these molecules to CaM might be similar to the binding of the fentanyl derivatives [28]. Lubeluzole binds to CaM through the amino acids F92, L105, M109 and M12, similar to TFP. Since the fentanyl derivatives and lubeluzole are similar, these amino acids are most likely important for the fentanyl derivatives in this study as well [29].
2.4 Modelling and docking
When modelling, a suitable web-based server to use is SWISS-MODEL. It is often used when building homologous models. To be able to build a model four main steps are usually included: identification of structural templates, alignment of the template and the known target sequence, model-building and evaluation of the quality of the model. All steps can be redone until a satisfactory model has been built [30].
The computing of the SWISS-MODEL server relies on a modelling engine called ProMod3. It gathers initial information about the structure from the template structure. To settle and solve insertions and deletions, feasible candidates are searched for in a structural database. Choosing of the final candidate is done by mean force methods looking at statistical potentials. Through energy minimization clashes, small distortions in the structure and disadvantageous interactions that emerged when performing modelling can be resolved [30].
To predict binding of small molecules to desired biomacromolecular targets, the program
AutoDock 4.2 is of great use. With the assistance of a gridbox a simulation of the docking can be performed, where the ligand dock to the set of grids designed for the protein [31,32]. The gridbox can be designed manually during the AutoGrid procedure. The box is a three-dimensional grid surrounding the protein, or parts of it, and grid points to which a probe atom is placed are included in it. To the grid point, the interaction energy between the atom and the protein is conferred. AutoGrid 4 then calculates the affinity grid for every atom which is later used in the calculations performed by AutoDock 4.2 [32].
During the docking simulation, the system runs multiple times resulting in multiple docked conformations. This results in clustering of similar conformations. Clustering information along with predicted binding energies, and other information about the docking, is stored and can be
10 observed through different text and source-code editor programs. Intramolecular energies and the Ki value are also retrieved from the ligand and protein assemble. Included in the intermolecular
energies are electrostatic energies, hydrogen bonds, VdW forces, and desolvation energies [31,32]. The Ki value is used to describe the binding affinity that an inhibitor has for a protein
receptor or enzyme. Ki value is a subset class for the more common and general Kd. The
difference is that Ki value only measures the binding of an inhibitor to a protein receptor. It is
measured in micro molar and a lower value, closer to zero, indicates a better inhibitor [12]. PyMOL is a useful tool when analysing docking. It is a software that can be utilised to visualise molecular data and produce 3D-models of molecules. Multiple models can be displayed at the same time and changes in color, view, labelling, different calculations and much more can be made [33]. To study the hydrophobicity of a molecule, the Eisenberg hydrophobicity scale can be applied. In this study, deeper red indicates more hydrophobic areas and lighter red indicates more hydrophilic areas. Alignment of models can also be made and an RMSD value calculated. The RMSD value describes the average distance between atoms aligned in the structures [34]. To be able to use both AutoDock and PyMOL, each structure has to be three-dimensional. CORINA Classic is a 3D structure generator, converting 2D structures to 3D. A desired structure can be generated by either entering it as a SMILES string or by drawing the structure with the help of CORINA Classics own structure editor [35].
2.5 Protein purification
Protein purification includes a series of methods performed to isolate a desired number of proteins from a sample containing different compounds. To be able to isolate a specific protein, different purification methods need to be used to separate proteins depending on their different properties [12].
Firstly, the desired protein needs to be expressed. This can be done with the use of expression vectors where a gene encoding for the desired protein is inserted. Besides the gene that codes for the desired protein, an expression vector usually also contains a promoter and a selectable marker. Very often, the promoter is inducible, meaning that a repressor is bound to the promoter region and that an inducer needs to be added in order for the repressor to release from the
promoter region so transcription of the gene can start. The selectable marker in the vector can be used to separate recombinant vectors with the gene encoding for the desired protein from other non recombinant vectors. Usually, antibiotic resistance is used as a selectable marker [36]. If the expression vector that is used contains an inducible promoter, OD is a method that can be used to determine when the inducer needs to be added to the bacterial culture in order to get a large amount of bacteria that can express the desired protein. The growth of the cell culture can be monitored by reading the OD at 600 nm. At this wavelength, 1 OD unit corresponds to about 0,8·108 ml-1 cells [36].
Once the desired protein has been expressed, different protein purification methods can be performed. IEX is a purification method that separates molecules depending on their net surface charge. Different proteins will have different net surface charges depending on the amino acids that they are made up of. This also means that the proteins will have different pI. By using a pH below or above the pI of a molecule in the equilibration buffer, the molecule will be positively or negatively charged. If the pH is above the pI the molecule will be negatively charged and if the pH is below the pI the molecule will be positively charged. Separation can thereby be performed
11 depending on the charge of the protein, by using anion or cation exchangers that are packed into a column. Positively charged proteins will be able to bind to cation exchangers and negatively charged proteins will bind to anion exchangers. The other molecules that do not bind to the column will be eluted [37].
SEC, also called gel filtration is another purification method. This is a method used to separate proteins depending on their size. A column packed with a gel that contains porous beads is used to achieve the separation. Small sized molecules can enter the porous beads easily while larger sized molecules have difficulty entering the beads. Since the small sized molecules can enter the beads they have a larger volume to pass through before eluting out from the column. The larger molecules have a smaller volume to pass through since they move in between the porous beads instead of through them. Therefore, when running an SEC, large sized molecules are eluted first, medium sized molecules afterwards and lastly small sized molecules [12].
Another protein analysis method is SDS-PAGE which determines the purity of the protein in the presence of SDS and beta-mercaptoethanol. The principle of SDS-PAGE is to separate proteins according to molecular mass by studying the movement through a polyacrylamide gel under the influence of an electric field. A small molecule migrates through the gel more effectively than a bigger molecule [12].
2.6 Circular dichroism
CD is a biological measurement method that arises from differential absorption of left and right beams in a sample. The absorption can be expressed as ΔA = AL-AR = εcl, where AL is the left
absorption and AR is the right absorption. Δε contains the units 10-3-10-6L mol-1 cm-1. The molar
ellipticity (Θ) is used to report the results from the CD measurements. Theta and ellipticity can be interconverted, by using the formula Θ = 3298ε, degree Θ cm2L mol-1[10].
When measuring in the far UV region, which is 170-250 nm, information about the secondary structure of the protein is obtained. From the spectra obtained from the CD measurements, the amount of alpha-helices, beta-sheets or random coils can be determined depending on the appearance of the spectra. The different secondary structures have certain characteristic appearances. In the near UV region, which is 260-320 nm, the induced optical activity in
aromatic side chains is observed. From this region, residue specific information is obtained which gives knowledge of the core packing [10].
2.7 Fluorescence
When the radiation of an excited molecule emits spontaneously it can be described as
luminescence. The different classifications of luminescence depend on the mode of excitation. Fluorescence is one variant, and occurs when an excited molecule returns to its ground state via emission of the photon [38,39]. A spectrophotometer is used to detect the emission. Further, to measure the extent of fluorescence another measuring device analyses the outcome of the detector [40].
A fluorophore is a fluorescent group present in a protein and it is the fluorophore that emits the photon [39, 40]. It can exist in two ways, either extrinsic which includes dyes and radioactive probes, or intrinsic like some amino acids, more precisely tryptophan, tyrosine and
phenylalanine. When analysing protein structure it is more common to look at tryptophan and tyrosine. During irradiation, using a wavelength of 280 nm, the two amino acids are excited.
12 Tryptophan can, however, also excite at a wavelength of 295 nm. Phenylalanine is more rarely used when looking at the fluorescence, seeing that the fluorescence observed is weaker [40]. The fluorophore has certain characteristics such as intensity, where the emission wavelength is positioned, and lifetime. By plotting the fluorescence intensity against the wavelength, a fluorescence spectrum is achieved [39]. Modifications of the fluorescence intensity are
interrelated to the fluorophores environment [40]. The wavelength is kept at a certain value while running an emission monochromator. The monochromator works as a filter by choosing to only record the fluorescence intensity instead of the whole spectrum [39].
To study protein interaction, saturation binding experiments are used. The method involves retaining the concentration fixed while adding a ligand to the protein. From fluorescence measurements, Kd can be determined [40].
13
3. System and process
3.1 Project plan
3.1.1 Flow plan
The flow plan was created to get a chronological map of the project and also to enable evaluation of the project as it progressed. It was mainly expressed in an activity list and a GANTT chart. The function of the GANTT chart, that can be seen in Figure 9, was to be a visual representation of the activity list. It includes important milestones and deadlines as well as the duration of all activities. The flow plan enabled project follow up immensely and also functioned as support when decisions regarding time distribution were made.
Figure 9. The original GANTT chart. Deadlines are marked in black. The main parts of the project, such as “Project planning”, are marked in orange and have been dissected into smaller activities, such as “Make GANTT chart”, marked in yellow.
3.1.2 Risk analysis
The risk analysis is a summary of risks that might occur during the project. A high value of the risk factor indicates that the risk has a high probability of occurring and that it has major
consequences for the project. Therefore, a tactic for preventing and correcting the risk is crucial. The whole risk analysis can be seen in Table 2.
14 Table 2: Risk analysis. Shows the probability of the risk occurring, the consequences of it happening, the risk factor and the prevention and correction of occurred risks. The probability and consequences have been ranked 1-5 and the risk factor is obtained by multiplying the probability with the consequence.
Risk Probability (1-5) Consequence (1-5) Risk Factor (Probability*C onsequence) Prevention and correction Programs needed for docking do not work 3 4 12 The whole
group can test the program, and see if it does work for someone in the group. Problems with internet connection 1 5 5 If the Wi-Fi
does not work, then try to use the mobile data Choosing wrong calmodulin template 2 4 8 Prevent this by looking at the conformation and states, open/closed, and four calcium ions should be bound.
Loss of data 2 5 10 Prevent this by
saving backup data, the correction is making the experiment once again. Mistakes in ligand analysis 2 1 2 Prevent this by undergoing preparation and understanding where the ligand approximately binds to CaM. Member of the group is absent 2 4 8 Prevent this by informing the group and share
15 relevant
documents so that they can adjust. Conflicts in the group 1 3 3 Prevention by following the group contract and having good communication within the group. If a conflict occurs, the conflict should be solved with the people involved and/or the project leader. Problem with computers, computers stop working 1 5 5 Keep computers updated. Save all documentation and files in a place where they can be accessed even if the computer stops working. If the computer stops working try to find another computer or device to work on or try to do as much as possible on the phone.
Coronavirus 5 5 25 Try to prevent
the spread of the virus by working with the project from a distance. Having meetings
16 online. More bioinformatic studies and literary studies instead of physical labs.
3.2 Plan for systematic follow up
The estimated hours for each activity were included in the budget, which contains hours for the entire project. In the time report document, the work each individual had done and how much time it took was reported. The plan was followed up systematically by having meetings every morning where tasks that were completed the day before and things that were to be done on the current day were discussed. The project was also followed up by meetings every afternoon where i.a. ongoing tasks and tasks completed during the day were discussed as well as questions were resolved. In the activity list and time cost, the estimated hours were equally distributed between the members of the project and it was continuously monitored how the planned hours for each task were being followed.
17
4. Material and Methods
4.1 Literary studies
The literary studies were done by choosing a set of relevant keywords based on the topics of the study. The search database Google and Linköping University library search tool, UniSearch, were used for searching and processing material based on resource evaluation and quality review. The search was further limited by scientific texts, which included scientific reports, journal articles, textbooks, theses, etc.
Researcher, publisher and publication year was reviewed for each of the scientific articles used. Furthermore, to ensure reliable publication, the predefined quality assurance format, IMRAD, was checked. Another important quality criterion was that the text has passed the peer review, which means that other experts in the field have ruled the text to have a scientifically high standard. This requirement was covered by filling in the box for academic peer review journals when using the search databases.
The writing of the project report was initiated with background studies of malaria, CaM, ligands, modelling and docking. This was done by reading relevant articles given from supervisors and literature material from the search databases mentioned earlier. A lot of information was gathered from various scientific sources and at the same time analysed based on quality and reliability. This provided loads of complete processed material to be used. While carrying out the literary studies and writing, proofreading each other texts to maintain a breadcrumb trail and good quality was crucial.
4.2 Bioinformatics
4.2.1 Material
The model of CaMpf that was used for docking was received from Uppsala University and the
model was built from a CaM template with PDB code 1CTR (Schrödinger LLC). For CaMhum,
also with PDB code 1CTR, the crystal structure of CaMhum, which was received from Uppsala
University as well, was used for docking. A model of CaMhum was made based on the crystal
structure to confirm that the dockings performed with the crystal structure could be considered valid.
The ligands used were three known antagonists, ART, CMZ and TFP, whose structures were retrieved from PubChem, and three fentanyl derivatives, ligand 1, ligand 2 and ligand 3, whose 3D-structures were created in CORINA Classic. Programs used for modelling, docking and analysing were SWISS-Model, Autodock 4.2, PyMOL and ClustalW.
The gridboxes for CaMc, when docking all ligands except CMZ, were made based on the amino
acids that are of significance when TFP binds to CaM. Therefore, the gridbox covered the amino acids; F92, L105, M124 and M144. When docking to CaMn, a gridbox was made that included all
the methionines in CaMn. The gridbox for CMZ was made based on the amino acids that have
proven to be of significance when CMZ binds to CaM. Therefore, the gridbox covered the amino acids; A15, L18, F19, V35, M36, L39, M51, M71, M72, M76, A88, V91, F92, V108, M109, L112, F141 and M145, i.e. both lobes on CaM.
18
4.2.2 Modelling of calmodulin
A model of CaMhum was made in SWISS-model based on the crystal structure. The known target
sequence of CaMhum and the known template PDB file, i.e. the crystal structure, were entered and
a model was built based on them. The new model was downloaded as a PDB file. The purpose of making it was to analyse if there were any major differences in docking when using the crystal structure compared to the built model of CaMhum. The crystal structure does not contain
information about six amino acids located in the linker between the lobes. This is probably because of the flexibility in the linker were the amino acids are situated. The built model contains the entire amino acid sequence. To study the impact these amino acids, TFP and CMZ were docked to both the crystal structure and the built model of CaMhum. Binding of TFP and CMZ
was very similar between the two dockings where they bound to the same area, interacting with the same amino acids in both the crystal structure and the model of CaMhum. Accordingly, the six
amino acids did not seem to have an impact on ligand binding, most likely since they are positioned in the linker between CaMc and CaMn and this study focuses immensely on the
binding of ligands to the hydrophobic clefts of the two lobes. Furthermore, it was observed that none of the ligands interact with the area of the undetermined amino acids. Conclusively, the crystal structure was chosen as the receptor molecule for CaMhum in ligand binding analysis using
AutoDock.
4.2.3 Analysing differences in structure in calmodulin from P. falciparum and
human calmodulin
Through comparisons between CaMpf and CaMhum, differences in primary and tertiary structure
were analysed. By studying the crystal structure of CaMhum and the model of CaMpf, the
structures were compared. Investigations of the primary structure were made by making a sequence alignment of CaMhum and CaMpf using ClustalW and changing the output order setting
to aligned. To investigate the differences in conformation, the positions of the hydrophobic regions were determined, as was its solvent accessible area was calculated. The total solvent accessible area and the solvent accessible area of methionine in the hydrophobic cleft was calculated as well. The visual differences in tertiary structure, hydrophobicity and charge between the two species were studied in PyMOL.
4.2.4 Conversion of the ligands into 3D-structures used for docking
The fentanyl derivatives used in this study were given with skeletal formation and transformed into 3D-structures using CORINA Classic. They were retrieved by drawing the skeletal formation and downloading the finished 3D-structure as a PDB file. The 3D-structures of the known
antagonists were downloaded from PubChem as PDB files.
4.2.5 Docking of ligands to calmodulin
Docking was performed to simulate the binding of ligands to the crystal structure of CaMhum and
the model of CaMpf. The docking software that was used was AutoDock 4.2. Three different
fentanyl derivatives and three different antagonists were docked to CaMhum and CaMpf. Each
ligand, except for CMZ, was docked to CaMc and CaMn separately for the two different species.
The differences in how the ligands bind to the different lobes in CaMhum and CaMpf, were
analysed with the help of docking. In the AutoDock menu, the ligand was chosen and the root of torsion tree was detected to be able to move and rotate the ligand as wished. Next, the receptor molecule, i.e. the PDB file of the CaM models, was loaded. Everything was saved in AutoDock as a PDBQT-file. To specify where on CaM each ligand should bind, a gridbox was made. The settings were saved and the binding energies for the chosen grid map was calculated with
19 Autogrid 4. The docking was then performed with Autodock 4.2, using two different speeds. First a short run was executed to ensure that an appropriate grid box and other settings had been
chosen. If the results from the short run were pleasing, a medium run was executed. The docking was saved as a PDB-file.
When the docking was done, the results were analysed in a text editor and PyMOL. When analysing the results in a text editor, the binding pose with lower lowest binding energy and a high numbers of conformation in cluster was chosen for further studies. In addition, information about binding energies, cluster values, intermolecular energies and the dissociation constant could also be retrieved. When analysing in PyMOL, it was possible to see where the ligands had bound to CaM and analyse the interactions between ligand and CaM.
20
5. Results
5.1 Process analysis
The updated GANTT chart can be visualised in Figure 10. It was revised during the progression of the project. Most of the changes were the addition of smaller activities that were added to the activity list as the project progressed, and some changes were adjustments of dates.
Figure 10. Updated GANTT chart. Deadlines are marked in black. The main parts of the project, such as “Project planning”, are marked in blue colors and have been dissected into smaller activities, such as “Make GANTT chart”, marked in pink colors.
Docking of the ligands was due to be completed 5th of May, but the completion of the docking was delayed by two days. On the 27th of April we realised that the dockings to CaMpf and
CaMhum would take more time than scheduled since we had to redo the dockings to a correct set
of models. New models were needed since the previous models displayed extended conformations of CaM. Autodock 4.2 is not able to alter the conformation of the receptor molecule and therefore, the ligands were bound to an extended, not realistic, conformation of CaM. A new model and crystal structure with desired conformation were received from Uppsala university and used for docking. Thus, the main reason for adjusting the dates in the flow plan was the extended time spent on docking. However, even though additional time was imposed it did not cause a major delay, partly due to the result analysis being completed faster than expected as well as planned buffert time. Therefore the bioinformatic part of the project did not exceed its assigned time. Moreover, the flow plan was also adjusted in accordance to the two additional dockings that were made to ensure that the six amino acids that were not displayed in the crystal structure did not affect ligand binding. However, it did not have a major effect on the overall flow plan. Conclusively, all milestones were successfully accomplished in obedience to the revised flow plan.
5.2 Structure comparison of calmodulin from P. falciparum and human
calmodulin
5.2.1 Primary structures differ in 16 positions
Sequence alignment of CaMhum and CaMpf display 16 differences in primary structure as can be
seen in Figure 11. The length of both sequences is 148 amino acids and sequence identity between CaMhum and CaMpf corresponds to 89%.
21 Figure 11. Multiple sequence alignment with ClustalW. “CaM_human” represents the sequence for
CaMhum and “CaM_p.falciparum“ represents the sequence for CaMpf. Stars “*” correspond to identical
amino acids, colons “:” indicate not as significant differences in sequence and blank spaces “ “ indicate significant alterations in sequence. Alterations resulting in change of charge are colored green,
alterations resulting in change of hydrophobicity are colored purple and those that result in change of both charge and hydrophobicity are colored both green and purple. The alterations with no color marking do not alter the charge- and hydrophobicity characteristics of the position.
Change in charge and hydrophobicity can contribute to conformational changes and affect ligand binding. The positions where these characteristics differ between the two species have therefore been color marked in Figure 11. A more elaborate display of the alterations and in regard to those two characteristics can be seen in Table 3.
Table 3. Differences in primary structure for CaMhum and CaMpf from ClustalW. Differences are written
with residual in CaMhum before the position of the alteration, and residual in CaMpf after the position of
the variation. Change in hydrophobicity is indicated by purple color and change in charge is indicated by green color.
Variation Hydrophobicity Charge
Q3K Hydrophilic-Hydrophilic Neutral-Positive
A10S Hydrophobic-Hydrophilic Neutral-Neutral
V55I Hydrophobic-Hydrophobic Neutral-Neutral
A57T Hydrophobic-Hydrophilic Neutral-Neutral
M71L Hydrophobic-Hydrophobic Neutral-Neutral
M76L Hydrophobic-Hydrophobic Neutral-Neutral
S81T Hydrophilic-Hydrophilic Neutral-Neutral
I85L Hydrophobic-Hydrophobic Neutral-Neutral
R86I Hydrophilic-Hydrophobic Positive-Neutral
K94R Hydrophilic-Hydrophilic Positive-Positive
N97D Hydrophilic-Hydrophilic Neutral-Negative
A103D Hydrophobic-Hydrophilic Neutral-Negative
22
V136I Hydrophobic-Hydrophobic Neutral-Neutral
Q143K Hydrophilic-Hydrophilic Neutral-Positive
T146I Hydrophilic-Hydrophobic Neutral-Neutral
5.2.2 Differences in hydrophobicity and charge distribution affect the tertiary
structure
Differences in structure between CaMhum and CaMpf is observed by aligning them in PyMOL as
can be seen in Figure 12. The alignment gives an RMSD-value of 2,15 [10]. The high value indicates some structural differences between the two models. CaMhum and CaMpf have a similar
tertiary structures, although CaMhum seems to have a more open conformation and the direction
of the amino acids differ between the two models.
Figure 12. Alignment of CaMhum (magenta) and CaMpf (cyan) visualized in PyMOL. Calcium ions bound
to CaMhum are colored green and calcium ions bound to CaMpf are colored yellow.
Moreover, the distribution of hydrophobic area differs between CaMhum and CaMpf. As can be
observed in Figure 13, the unquestionably largest part of the hydrophobic area is concentrated in a hydrophobic cleft in both CaMhum and CaMpf. However, in CaMpf, the hydrophobic area is
23 Figure 13. Colored after hydrophobicity with Eisenbergs hydrophobicity scale, where deeper red
indicates more hydrophobic areas, of A) front of CaMpf B) front of CaMhum visualized in PyMOL.
Furthermore, the values in Table 4 display that the hydrophobic solvent accessible area is larger for CaMhum than for CaMpf. As a result of the differences in primary structure between CaMhum
and CaMpf, the number of hydrophobic amino acids differ between the two. The variations can be
seen in Table 3 and conclude that CaMhum contains one hydrophobic amino acid more than
CaMpf. This, in combination with the more open conformation of CaMhum, explains the difference
in hydrophobic solvent accessible area.
Table 4. Hydrophobic solvent accessible area and solvent accessible area of methionine for CaMhum and
CaMpf calculated in PyMOL.
Model Hydrophobic solvent
accessible area (Å2)
Solvent accessible area of methionine (Å2)
CaMpf 2690,83 436,40
CaMhum 2827,58 566,76
Since many ligands have proven to interact with methionine when binding to CaM, the solvent accessible area of methionine is of additional interest. As can be seen in Table 4, this area is also larger for CaMhum than for CaMpf. The more closed conformation of CaMpf is part of the reason
for this. However, this too, is furthermore explained by the observed differences in primary structure between the species, see Figure 11. The variations conclude that two methionines in CaMhum have been replaced with leucine in CaMpf making it a further cause for the different
solvent accessible areas of methionine. A visualisation of methionine in CaMpf and CaMhum can
be seen in Figure 14, where CaMhum appears to have a slightly larger area of methionine than
24 Figure 14. Methionine, colored red, in A) CaMcpf B) CaMchum C) CaMnpf D) CaMnhumand visualized in PyMOL.
Due to that the primary structures of CaMhum and CaMpf differ in 16 positions, their conformation
and charge distribution differ as well, as can be seen in Figure 15. It appears that the variations in primary structure between the two species causes CaMpf to have a more closed conformation than
25 Figure 15. Charge distribution with negatively charged amino acids in red color firebrick and positively
charged amino acids in dark blue color density. A) Left side of CaMpf B) Left side of CaMhum C) Right side
of CaMpf D) Right side of CaMhum visualized in PyMOL.
The effects of the differences in charge distribution can be particularly observed as a result of a variation in position three where glutamine in CaMhum is replaced by lysine in CaMpf. A change
of charge from neutral to positive is thus generated in that position, which is concluded in Table 3. This shift, as well as structural differences between glutamine and lysine, enables interactions in CaMpf that cannot occur in CaMhum. Specially, it appears as can be seen in Figure 16, that D2,
K3, Q8, R75 and K148 in CaMpf interact in a manner that yields a more closed conformation for
CaMpf. The two lysines and single arginine are positively charged, and their interactions are
seemingly being stabilised by the negatively charged aspartic acid and the electronegative oxygen in the glutamine. The outcome being that K148, which is located in CaMnpf and is the last amino
acid of the protein, makes contact with K3 and R75, which are located in CaMcpf. Thus, resulting
in a closing of the conformation. The interactions can be observed in Figure 16 where the five amino acids have been highlighted. Worth noticing in Figure 16 is that all of them, except for Q8, are positioned in loops, making it possible for their position in space to be somewhat altered due to the flexibility of the loops. It is, for instance, plausible to assume that the aspartic acid in position two can move slightly closer to the three positively charged amino acids to allow more apparent interactions.
26 Figure 16. Interactions of amino acids D2 (limon), K3 (hotpink), Q8 (tv_blue), R75 (purple) and K148
(tv_green) in CaMpf (cyan) visualized in PyMOL. Calcium ions bound to CaMpf are colored yellow. A)
shows the amino acids and full length CaMpf B) shows the amino acids and a zoomed in part of CaMpf. All
highlighted amino acids except for Q8 are positioned in loops making it possible for them to move slightly
due to the flexibility of the loops. K148 is positioned in CaMnpf whereas the others are positioned in
CaMcpf.
5.3 Ligand docking to calmodulin from P. falciparum and human
calmodulin
To characterize and study the interactions of six ligands to CaM from the two species, Autodock 4.2 was used to simulate their binding. Three of the ligands are previously known antagonists to CaM; TFP, CMZ and ART and three of them are newly produced fentanyl derivatives; ligand 1, ligand 2 and ligand 3. The results from docking all six ligands to both lobes of CaM from both species can be seen in Table 5. The values in the table are for one pose for each docking which had high affinity for the ligand, good values for the intermolecular energies and high number of conformation in cluster. The electrostatic energy and VdW-, hydrogen bond- and desolvation energies combined constitutes the intermolecular energy, which indicates how strong an interaction is. The Ki values convey about the binding affinity between the ligands and CaM,
which is high when the Ki-value is low. The values for CaMc values are more relevant to study
than the values for CaMn since the ligands binds to CaMc besides CMZ, which binds to both of
the lobes. The most important and significant values for each ligand and the important differences between their dockings will be presented more thoroughly below.
27 Table 5. Data from docking with ligands TFP, CMZ, ART, ligand 2, ligand 3 and ligand 1 obtained from AutoDock executed with medium length and 100 runs/docking i.e. 100 different conformations were obtained. Receptor molecule Ligand Number of conformations in cluster Lowest binding energy (kcal/mole) VdW + hydrogen bond + desolvation energy (kcal/mole) Electrostatic energy (kcal/mole) Ki (µM) CaMcpf TFP 47 -6,15 -6,35 -1,30 30,80 CaMchum TFP 31 -9,82 -7,02 -4,29 0,063 CaMnpf TFP 17 -7,66 -6,09 -3,06 2,41 CaMnhum TFP 30 -7,64 -6,50 -2,64 2,50 CaMpf CMZ 12 -6,43 -9,19 0,08 19,33 CaMhum CMZ 9 -6,01 -8,70 0,01 39,47 CaMcpf ART 83 -6,13 -5,88 -0,25 31,97
CaMchum ART 53 -7,07 -7,01 -0,05 6,61
CaMnpf ART 50 -5,47 -5,40 -0,07 97,45
CaMnhum ART 99 -6,85 -6,84 -0,00 9,60
CaMcpf Ligand 2 14 -6,04 -7,43 -1,00 37,14
CaMchum Ligand 2 15 -7,55 -8,20 -1,74 2,90
CaMnpf Ligand 2 18 -4,78 -6,14 -1,03 312,44
CaMnhum Ligand 2 18 -6,35 -8,02 -0,71 22,32
CaMcpf Ligand 3 12 -5,43 -6,97 -0,85 103,99
CaMchum Ligand 3 14 -7,18 -7,84 -1,72 5,49
CaMnpf Ligand 3 14 -4,99 -5,80 -1,57 221,71
CaMnhum Ligand 3 15 -6,57 -8,20 -0,75 15,24
CaMcpf Ligand 1 15 -6,33 -7,68 -1,03 23,07
CaMchum Ligand 1 12 -6,27 -7,96 -0,70 25,44
CaMnpf Ligand 1 11 -5,36 -6,76 -0,99 116,81
28
5.3.1 Three antagonists trifluoperazine, calmidazolium and artimisinine
TFP is an already known antagonist of CaM.It has a higher affinity to CaMc compared to CaMn,
although binding to CaMn still occurs [15]. Since previous studies have characterized how TFP
bind and interact with CaMhum it is interesting to study how TFP bind to CaMpf to gain more
knowledge for possible malaria treatments.
When docking to CaMchum, TFP interacts with the amino acids F92, L105, M109, M124 and
M144, which coincides with literature. The dockings can be seen in Figure 17. In CaMcpf TFP
roughly interacts with the same amino acids. However, the orientation of TFP differs between the two species. The binding energy received from the dockings are considerably lower for CaMchum,
close to 3 units lower. The Ki value is much lower as well, almost 500 times lower for TFP bound
to CaMchum. Moreover, even though TFP bound to CaMcpf with a higher number of conformations
in cluster, meaning a larger number of similar poses, TFP binds to CaM with a significantly lower Ki value for CaMchum which indicates a much higher binding affinity for TFP. The
orientation of TFP in the two species could be a reasonable explanation to why the values differ so markedly as well as CaMchum having a more defined hydrophobic cleft for TFP to bind to.
Figure 17. Binding of TFP to A) CaMcpf B) CaMchum C) CaMnpf D) CaMchum. CaM colored with Eisenbergs
hydrophobicity scale and TFP in sticks and spheres colored by element. When observing the docking of TFP to CaMpf, binding to CaM
npf is favored. Both the binding
energy and the intermolecular energy are lower for CaMnpf which is advantageously since it
demonstrates a stronger interaction between TFP and CaM. The Ki values also differ significantly
between the two lobes, where CaMnpf has a value twelve times lower than CaMcpf indicating
stronger binding affinity for CaMnpf. The reason why TFP have higher affinity to CaMnpf,
contrary to binding to CaMhum, might be due to the differences in some amino acids causing
CaMpf to have a more closed conformation and therefore possibly generating inconvenient
binding to CaMcpf.
Noteworthy when analysing the Ki values of the dockings is that for CaMn there is almost a
negligible difference between the two species, while for CaMc there is a significant
difference.This suggests that the difference in structure between CaMchum and CaMcpf possibly
29 CMZ is the biggest out of all the ligands studied and unlike the other three previously known antagonists it binds to both lobes of CaM. Similar to the other antagonists, it contains
hydrophobic benzene-ring structures which is important for the ability of CMZ to bind to CaM. In addition, it contains a positively charged imidazolium structure that is important for inhibiting CaM as well.
As expected, the docking of CMZ resulted in binding to CaM making contact with both lobes, as can be seen in Figure 18. Therefore it diverges from the other five ligands in this study. But still, it appears to interact with M144 and M145 in CaMchum and M71 and M72 in CaMnhum much like
the other ligands, supporting methionines important role in ligand binding. When CMZ binds to CaMpf it appears to interact more closely with the hydrophobic cleft of CaM
cpf, and not as much
with CaMnpf, namely contrary as to when CMZ binds to CaMhum. This suggests that CMZ
interacts with CaMc in greater occurrence than with CaMn, which is not surprising since that is
the case for the majority of the antagonists. Notable is that CaMnhum has an additional methionine
in position 71 and 76 while CaMnpf has two leucines in those positions, as can be seen in Table 3.
The additional methionine, which is an amino acid important in ligand binding, can be the reason as to why CMZ is positioned more towards CaMnhum than CaMchum.
Figure 18. Binding of CMZ to full length CaM showing A) full length CaMpf B) full length CaMhum C)
CaMcpf D) CaMchum E) CaMnpf F) CaMnhum. CaM colored with Eisenbergs hydrophobicity scale and CMZ
