International Journal of Genomics

International Journal of Genomics / 2019 / Article

Research Article | Open Access

Volume 2019 |Article ID 1750291 | 13 pages | https://doi.org/10.1155/2019/1750291

Computational Identification of Metabolic Pathways of Plasmodium falciparum using the -Shortest Path Algorithm

Academic Editor: Luigi Ceci
Received26 Jul 2018
Revised28 Nov 2018
Accepted29 Jul 2019
Published01 Oct 2019

Abstract

Plasmodium falciparum, a malaria pathogen, has shown substantial resistance to treatment coupled with poor response to some vaccines thereby requiring urgent, holistic, and broad approach to prevent this endemic disease. Understanding the biology of the malaria parasite has been identified as a vital approach to overcome the threat of malaria. This study is aimed at identifying essential proteins unique to malaria parasites using a reconstructed iPfa genome-scale metabolic model (GEM) of the 3D7 strain of Plasmodium falciparum by filling gaps in the model with nineteen (19) metabolites and twenty-three (23) reactions obtained from the MetaCyc database. Twenty (20) currency metabolites were removed from the network because they have been identified to produce shortcuts that are biologically infeasible. The resulting modified iPfa GEM was a model using the -shortest path algorithm to identify possible alternative metabolic pathways in glycolysis and pentose phosphate pathways of Plasmodium falciparum. Heuristic function was introduced for the optimal performance of the algorithm. To validate the prediction, the essentiality of the reactions in the reconstructed network was evaluated using betweenness centrality measure, which was applied to every reaction within the pathways considered in this study. Thirty-two (32) essential reactions were predicted among which our method validated fourteen (14) enzymes already predicted in the literature. The enzymatic proteins that catalyze these essential reactions were checked for homology with the host genome, and two (2) showed insignificant similarity, making them possible drug targets. In conclusion, the application of the intelligent search technique to the metabolic network of P. falciparum predicts potential biologically relevant alternative pathways using graph theory-based approach.

1. Introduction

Malaria remains one of the leading global health challenges with about 216 million cases and more than 445,000 deaths recorded in 2016. According to the World Health Organization (WHO), 88% of these deaths occurred in Africa [1] and Plasmodium falciparum accounted for the majority of the cases. P. falciparum has developed resistance to all antimalarial medications including the most potent one—artemisinin [24]. Some genetic changes and metabolic alterations confer survival advantage that enables the parasite to evade drug effects thereby ensuring its survival. However, there is still a poor understanding of the processes utilized by P. falciparum to evade drug effects, which also hamper vaccine development [5]. The incomplete knowledge of the metabolic pathways of P. falciparum has also been identified as a major impediment towards the development of an effective treatment [6]; hence, the need for intensified research that is aimed at understanding better the parasite biology. Computational approach, which predicts metabolic networks and possible alternative pathways pertinent to the parasite’s survival, can identify candidate drug targets to eliminate the disease globally. Some previous studies have been done in this regard. Figure 1 explains the various paths (a), pathways (b), and complete network constructions (c) in metabolic pathways.

The identification of these undiscovered pathways within the metabolism of the malaria parasite entails enumerating not only the shortest path within the metabolic pathway but also the list of other feasible paths within a source compound and a target compound. This could be regarded as an optimization problem. The metabolic system or metabolism of a specific cell or an organism is the entire system of metabolic reactions of the cell or organism. A metabolic pathway is an associated subsystem of the metabolic system either as a representation of particular processes or characterized by functional boundaries, e.g., the system between a glucose (initial substance) and a pyruvate (final substance). The -shortest path technique enables the optimal and suboptimal metabolic pathways to be identified within the metabolic network [7].

Croes et al. [8] presented a reaction to the representation of the metabolic network in order to identify relevant pathways in a biological network. However, the path finding algorithm applied was unable to handle a large dataset. Faust et al. [9] combined a random walk-based reduction of the graph with the shortest path-based algorithm and applied on a yeast metabolic network. This approach is computationally intensive due to several runs by the shortest path algorithm used. Oyelade et al. [10] applied a colour coding algorithm to search for minimum pathways in the P. falciparum interaction network. They discovered “identified” and “unknown” genes and signal transduction pathway involved in metabolic activities of P. falciparum. However, this approach is only capable of obtaining the shortest signal pathway. Also, in another work of Oyelade et al. [11], the qualitative Petri net model for the glycolysis pathway in P. falciparum was built and analysed for its structural and quantitative properties using the Petri net theory which only give insights into the complex net behavior of the pathway. Essential reactions in the metabolic network of P. falciparum have been predicted using several computational techniques such as in silico knockout screening [12], load and choke point [13], centrality measures [14], flux balance analysis [15], and machine learning approach [16] etc.

The goal of this study is to computationally identify branching metabolic pathways of P. falciparum using the enhanced -shortest path technique with improved prediction precision. This technique was used in our previous work [17] although limited to finding only alternative paths in selected metabolic pathways. In this work, we obtained the genome-scale metabolic model of the 3D7 strain of P. falciparum from a previous study [18]. The dataset contains 325 genes and 670 metabolic reactions. The underlying architecture of the metabolic network was modelled using a reaction graph where reactions are designated as the nodes and two reactions are neighbors, if the product of a reaction is the substrate of the other. Compound graph representation is not suitable for path finding algorithms because a reaction can have more than one compound as both input and output, e.g., , thereby leading to a bunch of redundant edges. The existence of two different types of nodes in a bipartite graph makes it complex to traverse by the path finding algorithm. The choice of a reaction graph was informed by the aim of this study, which is to identify more alternative metabolic pathways, and the pathways are known to be chains of reactions and also due to the applicability of path finding algorithms which requires a single type of node to compute the shortest paths within the network. The algorithm, the -shortest path technique developed by Kadivar [19], was adapted to extract the shortest paths from a defined source node to a target node. Only two metabolic pathways were considered for the alternative path analysis in this work due to the overlapping nature of the pathways that made it difficult to identify source and target reactions relating to a specific pathway.

We set the value of of the path finding algorithm to equal five (5) because as increases, the biological relevance of the path reduces. The identification of essential enzymes in a particular network allows a possible drug target to be identified [16, 2022]. Therefore, in drug development, essential enzymes are generally recognized as perfect drug target candidates for potential new drugs and vaccines to treat and prevent diseases since their deletion from a network can compromise its integrity [21, 22]. Hence, the essentiality of reactions was carried out using the betweenness centrality measure to determine essential reactions within a reconstructed network and subsequently predicted 32 essential enzymes.

2. Materials and Methods

2.1. Reconstructing the Metabolic Network

Metabolic reaction data were obtained from Chiappino-Pepe et al.’s study [18]; also 19 metabolites and 23 reactions were obtained from the MetaCyc database [23] to fill gaps in the iPfa GEM (see appendix A for the list of reactions obtained from the MetaCyc database). In a metabolic network, there are several metabolites that are commonly involved in reactions that cause shortcuts without biological meaning when computing paths in a simple graph [24]. These metabolites are referred to as pool metabolites or currency metabolites such as proton and water. Most of them often produce biological misinterpretations due to the artifactual links between nodes.

Kim et al. [25] identified twenty-five (25) currency metabolites out of which we eliminated twenty (20) from the network before reconstructing the graph. The remaining five (5) currency metabolites identified by [16] were left in the network since their presence does not have a significant effect that could lead to shortcuts without biological meaning, and a uniform weight is assigned to all edges in the network.

2.2. The Algorithm

The algorithm was selected among the other -shortest paths because of its superior computational performance (as shown in Table 1). This algorithm requires topological sorting of the graph nodes as its input, which is only possible for a directed acyclic graph. Figure 2 shows the algorithm flow chart of the metabolic network construction.


AlgorithmTime complexity

Lazy variant of Eppstein’s algorithm (LVEA)
Eppstein’s algorithm (EA)
Feng
Yen

Given that as the set of the -shortest paths from to node and , where denotes the length of path . Let , because of topological ordering and since order () is larger than the other nodes in all paths; is a loopless path for each . Therefore if , then (stated in Algorithm 1).

1: ;
2:while(t) do
3: Update(P[next] by P[j], ─(next));
4: ;
5:end while
6: Update(P[i] by P[j], ─ (i))
7: Make min-priority queue MQ(i) by for ─(i) according to .
8:while and do
9: Move P at root of MQ(i) to P[i] and replace the moved item by its neighbour in the source list from which it came and update the nodes values traversing up in MQ(i) along the path of the moved item.
10:end while

The algorithm was modified in two significant ways:

Firstly, we eliminated the first stage (topological ordering) due to the nature of the dataset where several reactions within the dataset could be on the same level in the topological order thereby varying the results from each run of the algorithm. For instance, if reactions A, B, and C have the same topological order, then it implies that at different runs of the algorithm, reaction A could come first or reaction B or reaction C, which varies the result of the algorithm at different runs. The metabolic network of P. falciparum contains over six hundred compounds and over a thousand reactions. The inconsistency inherent in the ordering of the reactions will have severe biological implication for the eventual shortest paths generated by the algorithm. For instance, if the order of a vital reaction within the pathway is less than that of the source compound for a particular pathway, then the reaction will be omitted from the shortest paths because the algorithm starts traversing from the source node to the target node in the topological order. Secondly, the topological ordering is not applicable to networks with a cycle since metabolic networks are known to contain loops; hence, we introduced heuristic function to enhance the prediction precision of the alternative paths. Figure 3 shows the modified algorithm for metabolic network construction. The reactions in the annotated pathway were retrieved and used to guide the search of the algorithm to list biological feasible paths.

Therefore, the modified algorithm 2 is given.

1: ;
2: while do
3: Update(P[next] by P[j], ─ (next));
4: Compute Rk for P[next]
5: ;
6: end while
7: Update(P[i] by P[j],─ (i))
8: Make min-priority queue MQ(i) by P1[j] for ─ (i) according to values.
9: while and do
10: Move P at root of MQ(i) to P[i] and replace the moved item by its neighbor in the source list from which it came and update the nodes values traversing up in MQ(i) along the path of the moved item.
11: end while
2.2.1. Reward Shaping Formulation

The concept of reward shaping was also introduced to the -shortest path algorithm to improve the quality of biologically relevant alternative path prediction. Reward shaping is the addition of an extra reward signal that encodes some heuristic knowledge of the system designer or domain expert, thus encouraging the learning agent to explore parts of the state space believed to contain good solutions [2628]. In practice, the choice of reward function is intuitively selected [29] and this has been successfully applied to speed up reinforcement learning techniques in complex domains [30, 31].

The reward of each feasible path between a source reaction and target reaction is computed using the number of reactions within the path that can be found in the set of annotated reactions . An annotated pathway was obtained from the KEGG database. The justification for this model is based on the assumption that the annotated pathway for a metabolic process represents the shortest path; hence, the next shortest alternative path is most likely to have branched from the annotated path thereby containing most of the reactions in annotation. The reactions in the feasible path were compared to the annotated path to determine the number of intersecting reactions. However, the total cost of a path is a function of its length and the number of annotated reaction contained in the path. where represents the initial reward for path , represents the coefficient for reaction () which has a constant value of 1. will assume the value of 1 if is present in the set of annotated reaction for that pathway and 0 if otherwise.

In addition to the objective, obtaining biologically relevant paths is also to obtain the shortest path; hence, a penalty function is introduced which is meant to reduce the reward of paths based on their length; this is represented in where is a “balancing factor” which is an arbitrary value that is suitable to provide a penalty value that balances the positiveness of annotated reactions present in the path and the disadvantages of the path length.

The reward for a path is given as the difference between its initial reward and penalty score as shown in

In this study, we chose to be 0.5 for a fair penalty score and the adjusted cost is the difference between the number of reaction in the path and the reward. We then applied the modified algorithm to the glycolysis and pentose phosphate pathway of P. falciparum to obtain the -shortest paths for .

3. Results

We then applied the modified algorithm to the glycolysis and pentose phosphate pathway of P. falciparum to obtain the -shortest paths for . Table 2 shows the number of reactions present in a predicted path and its computed cost. The five alternative paths identified by our method for each of the target metabolic pathways were further verified manually to identify artifacts and biologically plausible paths. Two alternative paths were identified to be a feasible path for metabolic activities in the pentose phosphate pathway. One alternative path was identified in the glycolysis pathway, which could serve as an alternative path for metabolic activities in the pathway. The reactions represented in gold in Figures 4 and 5 indicate reactions identified by this study to have created alternative paths in the respective metabolic pathways. The reaction represented in blue in Figure 5 represents a nonannotated reaction identified by this study to create an alternative path glucose 6-P to glycolysis.


No. of reaction in the pathNo. of reaction in annotated pathwayReward ()Adjusted cost

Glycolysis
1983.55.5
210946
31193.57.5
41193.57.5
5121048

Pentose phosphate
18844
2994.54.5
310946
411115.55.5
511115.55.5

For the glycolysis pathway, we predicted a pathway, which could provide an alternative path for the generation of pyruvate. This metabolic reaction involves alpha-D-glucose-1-phosphate uridylyltransferase. The predicted alternative paths for the glycolysis pathway are presented in Figure 4.

The predicted alternative paths for the pentose phosphate pathway are presented in Figure 5. Two (2) additional metabolic reactions were predicted to have created an alternative path in the pathway; they are D-glyceraldehyde-3-phosphate glycolaldehyde transferase and sedoheptulose 7-phosphate 1-phosphotransferase reactions, respectively.

3.1. Essential Reaction Prediction

To verify the predicted pathways, a test of the essentiality of these genes was done on the predicted pathways. Betweenness centrality measure was applied to the reconstructed metabolic network to each reaction in the network to determine the essentiality of a particular metabolic pathway that is critical for the production of a target compound identified. The betweenness centrality method is stated as follows: where is the proportion of the shortest path linking to that contain , is the number of the shortest path that contain point as an intermediary in the shortest path from to , and is the number of the shortest path from to . where is the pair dependency which represents the degree to which a point, , must depend upon another, . where is the betweenness point centrality.

The genes that coded the enzymes responsible for the essential reactions discovered were obtained and blasted against the human genome to evaluate the existence of homology in the host. We used a centrality score of 0.25 as the level of significance because the centrality score of some of our predicted reactions that validated the gold standard set of essential enzymes and other predicted ones in literature has a centrality score of approximately 0.3. Table 3 shows the reactions with a centrality score from 0.25 and above as the predicted essential reactions. A total of 32 reactions were predicted as essential, out of which 13 have been validated in the literature.


Reaction IDReaction nameEC numberBiological processCentrality scoreReference

HBDG_cHemoglobin digestionCysteine and methionine0.521053
T_c_to_a_C00073L-Methionine transport reactionCysteine and methionine0.442105
R00896_cL-Cysteine:2-oxoglutarate aminotransferase2.6.1.1Cysteine and methionine0.426316
T_c_to_r_C006411,2-Diacyl-sn-glycerol transport reactionGlycerolipid0.409357
R02251_rAcyl-CoA:1,2-diacyl-sn-glycerol O-acyltransferase2.3.1.20Glycerolipid pathway0.324619
T_c_to_r_C006411,2-Diacyl-sn-glycerol transport reactionGlycerophospholipid pathway0.42986
R00177_cATP:L-methionine S-adenosyltransferase2.5.1.6Glycerophospholipid pathway0.264179Oyelade et al. [12]
R02037_R06868
_R06869_c
S-Adenosyl-L-methionine:(methyl)ethanolamine-phosphate N-methyltransferase2.1.1.103Glycerophospholipid pathway0.255603Choubey et al. [40], Pessi et al. [41], Witola et al. [42]
R00497_cGamma-L-glutamyl-L-cysteine:glycine ligase (ADP-forming)6.3.2.3Glycine, serine and threonine0.287619Chiappino-Pepe et al. [18], Huthmacher et al. [39]
R01061_cD-Glyceraldehyde-3-phosphate:NAD+ oxidoreductase (phosphorylating)1.2.1.12 1.2.1.59Glycolysis pathway0.552381
R01068_cD-Fructose-1,6-bisphosphate D-glyceraldehyde-3-phosphate-lyase (glycerone-phosphate-forming)4.1.2.13Glycolysis pathway0.495238Wanidworanun et al. [43]
R00769_cUTP:D-fructose-6-phosphate 1-phosphotransferase2.7.1.11Glycolysis pathway0.490476
R03427_cInositol-1,4-bisphosphate 1-phosphatase3.1.3.57Inositol phosphate pathway0.533333
R04372_cPhosphatidylinositol-3,4-bisphosphate 4-phosphatase3.1.3.66Inositol phosphate pathway0.533333
R01187_c1D-Myo-inositol 3-phosphate phosphohydrolase3.1.3.25Inositol phosphate pathway0.47619
R01802_cCDP-diacylglycerol:myo-inositol 3-phosphatidyltransferase2.7.8.11Inositol phosphate pathway0.419048Chiappino-Pepe et al. [18], Fatumo et al. [44]
R03429_cInositol-1,3,4-trisphosphate 5/6-kinase2.7.1.159Inositol phosphate pathway0.285714
R01641_cSedoheptulose-7-phosphate:D-glyceraldehyde-3-phosphate glycolaldehyde transferase2.2.1.1Pentose phosphate pathway0.328478
R00720_cInosine 5-triphosphate pyrophosphohydrolase3.6.1.8 3.6.1.19Purine metabolism0.448718
R01231_cXanthosine-5-phosphate:L-glutamine amido-ligase (AMP-forming)6.3.5.2Purine metabolism0.37833McConkey [46]
R00576_cGlutamine-pyruvate transaminase2.6.1.15Purine metabolism0.345098
R02024_c2-Deoxycytidine diphosphate:oxidized-thioredoxin 2-oxidoreductase1.17.4.1Purine metabolism0.340121Barker et al. [47], Chakrabarti et al. [48], Lytton et al. [49]
R01135_cIMP:L-aspartate ligase (GDP-forming)6.3.4.4Purine metabolism0.330982Eaazhisai et al. [50]
R00328_cGDP phosphohydrolase3.6.1.5; 3.6.1.6; 3.6.1.42Purine metabolism0.287731
R00570_cATP:CDP phosphotransferase2.7.4.6Pyrimidine0.45977Oyelade et al. [12], Chiappino-Pepe et al. [18]
R02024_c2-Deoxycytidine diphosphate:oxidized-thioredoxin 2-oxidoreductase1.17.4.1Pyrimidine0.419157Barker et al. [47], Chakrabarti et al. [48], Chiappino-Pepe et al. [18], Lytton et al. [49]
R00571_cUTP:ammonia ligase (ADP-forming)6.3.4.2Pyrimidine0.255172Chiappino-Pepe et al. [18]
R00573_cUTP:L-glutamine amido-ligase (ADP-forming)6.3.4.2Pyrimidine0.255172Chiappino-Pepe et al. [18]
R06517_cAcyl-CoA:sphingosine N-acyltransferase2.3.1.24Sphingolipid0.431917Gerold and Schwarz [32]
R02251_cAcyl-CoA:1,2-diacyl-sn-glycerol O-acyltransferase2.3.1.20Sphingolipid0.325708
R00351_mAcetyl-CoA:oxaloacetate C-acetyltransferase (thioester-hydrolysis)2.3.3.1 2.3.3.3TCA0.377739
R02569_mAcetyl-CoA:enzyme N6-(dihydrolipoyl)lysine S-acetyltransferase2.3.1.12TCA0.281995Chiappino-Pepe et al. [18]

3.2. Visualization of Predicted Pathway

Figures 6(a) and 6(b) are the visualization of the predicted glycolysis pathway where Figure 6(a) represents the unevaluated version of the particular pathway while Figure 6(b) represents the pathways of the betweenness centrality measure. The nodes with the biggest structure in Figure 6(b) represent the predicted essential reactions of the pathway. Similarly, Figures 7(a) and 7(b) are the visualization of the predicted glycolysis pathway where Figure 7(a) represents the unevaluated version of the particular pathway while Figure 7(b) represents the pathways that the betweenness centrality measure has been applied. The biggest nodes in Figure 7(b) denote the predicted essential reactions of the pathway.

4. Discussion

Several P. falciparum metabolic network reconstructions have been done previously, but this study is aimed at predicting alternative metabolic pathways for energy generation in the malaria parasite via the pentose phosphate and glycolytic reactions using the -shortest path algorithm. The essentiality of these reactions to the survival of these parasites was also evaluated in the predicted models (Table 3) and genes with no human homologs identified (Table 4). Sequence alignment was performed for all the 32 enzymes to identify the existence of homology; two of the enzymes show insignificant similarity with the host. Energy production in apicomplexans involves both the aerobic and anaerobic pathways, and this varies from species to species [32]. P. falciparum is a fast growing organism in the human erythrocyte. Therefore, in order to meet its energy need for growth and cell division, it relies on anaerobic oxidation of glucose to generate ATP. This is achieved through a series of enzyme-catalyzed breakdown of glucose to pyruvate, a metabolic process otherwise called the glycolysis or Embden–Meyerhof–Parnas pathway [33]. This process does not require oxygen to generate ATP, but results in the accumulation of pyruvate. While the main purpose of glycolysis is to generate ATP, one of its intermediates, glucose-6-phosphate, is the precursor for the pentose phosphate pathway [34]. This pathway serves an important purpose of generating NADPH and pentose sugars, which are moieties needed for the synthesis of nucleic acids and membrane lipids that are molecular raw materials needed by the parasite for cell division.


Gene nameEnzyme nameE.C. numberBiological process

PF3D7_1343000Phosphoethanolamine N-methyltransferase2.1.1.103Glycerophospholipid
PF3D7_1437200Ribonucleoside-diphosphate reductase1.17.4.1Purine/pyrimidine

Similarly, different stages of malaria parasites have been reported to have different energy requirements while P. falciparum-resistant strains have been shown to alter their genetic and metabolic pathways, in order to utilize nutrient requirements for fitness and survival in a drug environment. A recent study identified metabolic changes in mutant parasites, which shows differential transport and utilization of nutrients, as shown in computational prediction that incorporates metabolomics data from sensitive and resistant isolates [35]. Malaria parasite also lacks capability to store carbohydrates [36, 37]; it constantly requires the production of this through glucose and can utilize other carbohydrate precursors scavenged from a host such as UDP-glucose; hence, the identified alternative metabolic pathways that P. Falciparum utilizes for survival are crucial for curbing drug resistance and can be targeted as vaccine candidates to reduce malaria transmission. Variable regulation of host UTP-glucose-1-phosphate uridylyltransferase (UDP-glucose phosphorylase), which catalyzes energy-generating reactions such as galactose metabolism and glycogen synthesis, has been reported [34] thereby making it a good target. This was identified as an essential pathway in glycolysis in the present study. Other predictions from this study corroborate the findings from other studies; these are highlighted in Table 3. Amino acid deprivation can alter the growth of P. falciparum in vitro; however, it may not completely kill the parasite [38]. This regulation of nutrient requirement to cope with deficit is one of the many survival mechanisms already identified in the parasite. However, other essential amino acid pathways identified can be targeted together to achieve a detrimental effect on the parasite’s survival. Our study identified gamma-L-glutamyl-L-cysteine:glycine ligase as an essential reaction for glycine, serine, and threonine metabolism. This finding has been previously reported in other studies [18, 39].

For the cysteine and methionine pathways, hemoglobin digestion is essential for the malaria parasite; L-methionine transport reaction and L-cysteine:2-oxoglutarate aminotransferase were also predicted as essential reactions. The altered transport of nutrients across the membrane may also vary in sensitive and resistant parasites [35]; hence, vaccines or therapies that target these pathways will provide potentially lethal effects on the parasites. Other predicted essential reactions within the glycerophospholipid pathway are 1,2-diacyl-sn-glycerol transport reaction, L-methionine s-adenosyltransferase, and s-adenosyl-L-methionine (only the latter has validated predictions in the literature [4042]). In the glycolytic pathway, three reactions: D-glyceraldehyde-3-phosphate; D-fructose-1,6-bisphosphate D-glyceraldehyde-3-phosphate-lyase; and UTP:D-fructose-6-phosphate 1-phosphotransferase, were predicted as essential reactions with D-glyceraldehyde-3-phosphate-lyase validated in the literature [43]. Five reactions, namely, inositol-1,4-bisphosphate 1-phosphatase; phosphatidylinositol-3,4-bisphosphate 4-phosphatase; 1-D-myoinositol-3-phosphate phosphohydrolase; CDP-diacylglycerol:myo-inositol 3-phosphatidyltransferase; and inositol-1,3,4-trisphosphate 5/6-kinase, were the predicted essential reactions within the inositol phosphate pathway out of which CDP-diacylglycerol:myo-inositol 3-phosphatidyltransferase is validated [18, 44]. It is noteworthy that D-glyceraldehyde-3-phosphate dehydrogenase has homolog in the human genome (Table 4).

Similarly, our study predicted a metabolite—sedoheptulose-1,7-diphosphate—from the breakdown of xylulose, in the pentose phosphate pathway, which may be utilized by the parasite to scavenge for energy. Only sedoheptulose-7-phosphate:D-glyceraldehyde-3-phosphate glycolaldehyde transferase was predicted as an essential reaction within the pentose phosphate pathway. P. falciparum essentially synthesize nucleotides de novo. Hence, it cannot salvage pyrimidines from an extracellular environment. This biosynthetic pathway has been identified as a good target for malaria control [45]. For the purine and pyrimidine pathways, five and four reactions were predicted as essential and have been reported in other studies (see details in Table 3). The implications of these hypothesis are that in addition to other predicted pathways, if biologically validated through disruption of genes, encoding the enzymes for these pathways in P. falciparum will further elucidate survival and alternate energy generation pathways in the parasite. The hypothesized potential pharmacological enzyme targets essential for the parasite can be targeted to control malaria infection globally.

5. Conclusions

In this work, we have been able to predict alternative metabolic paths in the glycolysis and pentose phosphate pathways of P. falciparum. We predicted two (2) essential proteins in the glycerophospholipid, purine, pyrimidine, and glycolysis metabolic pathways (Table 4) without homology with a host. With the use of heuristic function to enhance the -shortest path algorithm, we have been able to identify potential biologically relevant paths using a computational graph-based technique which hitherto is less utilized due to its very high false positive result. Biologically targeting these candidate proteins in Plasmodium is recommended to improve understanding of the predicted alternative pathways and the precision of the graph theory-based method.

Data Availability

The datasets used in this study are available in http://lcsb-databases.epfl.ch/pathways/Gems.

Conflicts of Interest

The authors disclose no potential conflicts of interest.

Authors’ Contributions

All authors contributed to this work. Jelili Oyelade contributed to the original idea and conception. The implementation of the concept was achieved by Olufemi Aromolaran. Titilope Dokumu and Solomon Rotimi did the biological interpretation of the results. Jelili Oyelade, Itunuoluwa Isewon, and Olufemi Aromolaran wrote the manuscript. Jelili Oyelade, Itunuoluwa Isewon, and Olufemi Aromolaran were responsible for the structure and argument for the paper. Jelili Oyelade, Itunuoluwa Isewon, Efosa Uwoghiren, Olufemi Aromolaran, Titilope Dokunmu, Solomon Rotimi, Oluwadurotimi Aworunse, Olawole Obembe, and Ezekiel Adebiyi were responsible for the critical revision and approval of the final version.

Acknowledgments

This work is fully supported by the Covenant University and Covenant University Center for Research, Innovation and Discovery with Grant no. CUCRID RG 034.04.15/FS. We also thank Dr. Jelili Oyelade for the contribution from his PhD thesis.

Supplementary Materials

Table S1: reactions obtained from the MetaCyc database to fill gaps in the iPfa GEM. (Supplementary Materials)

References

  1. WHO, World Malaria Report 2016, World Health Organization, 2016.
  2. A. Chakraborty, “Emerging drug resistance in Plasmodium falciparum: a review of well-characterized drug targets for novel antimalarial chemotherapy,” Asian Pacific Journal of Tropical Disease, vol. 6, no. 7, pp. 581–588, 2016. View at: Publisher Site | Google Scholar
  3. A. F. Cowman, J. Healer, D. Marapana, and K. Marsh, “Malaria: biology and disease,” Cell, vol. 167, no. 3, pp. 610–624, 2016. View at: Publisher Site | Google Scholar
  4. World Health Organization, Antimicrobial Resistance: 2014 Global Report on Surveillance, World Health Organization, 2014.
  5. A. J. Birkett, V. S. Moorthy, C. Loucq, C. E. Chitnis, and D. C. Kaslow, “Malaria vaccine R&D in the decade of vaccines: breakthroughs, challenges and opportunities,” Vaccine, vol. 31, pp. B233–B243, 2013. View at: Publisher Site | Google Scholar
  6. W. C. Koff, D. R. Burton, P. R. Johnson et al., “Accelerating next-generation vaccine development for global disease prevention,” Science, vol. 340, no. 6136, p. 1232910, 2013. View at: Publisher Site | Google Scholar
  7. D. Eppstein, “Finding the k shortest paths,” in Proceedings 35th Annual Symposium on Foundations of Computer Science, pp. 154–165, Santa Fe, Mexico, 1994. View at: Publisher Site | Google Scholar
  8. D. Croes, F. Couche, S. J. Wodak, and J. Van Helden, “Metabolic PathFinding: inferring relevant pathways in biochemical networks,” Nucleic Acids Research, vol. 33, no. Web Server, pp. W326–W330, 2005. View at: Publisher Site | Google Scholar
  9. K. Faust, P. Dupont, J. Callut, and J. Van Helden, “Pathway discovery in metabolic networks by subgraph extraction,” Bioinformatics, vol. 26, no. 9, pp. 1211–1218, 2010. View at: Publisher Site | Google Scholar
  10. J. Oyelade, I. Ewejobi, B. Brors, R. Eils, and E. Adebiyi, “Computational identification of signalling pathways in Plasmodium falciparum,” Infection, Genetics and Evolution, vol. 11, no. 4, pp. 755–764, 2011. View at: Publisher Site | Google Scholar
  11. J. Oyelade, I. Isewon, S. Rotimi, and I. Okunoren, “Modeling of the glycolysis pathway in Plasmodium falciparum using Petri nets,” Bioinformatics and Biology Insights, vol. 10, p. BBI.S37296, 2016. View at: Publisher Site | Google Scholar
  12. J. Oyelade, I. Isewon, E. Uwoghiren, O. Aromolaran, and O. Oladipupo, “In silico knockout screening of Plasmodium falciparum reactions and prediction of novel essential reactions by analysing the metabolic network,” BioMed Research International, vol. 2018, Article ID 8985718, 11 pages, 2018. View at: Publisher Site | Google Scholar
  13. S. A. Rahman and D. Schomburg, “Observing local and global properties of metabolic pathways:“load points” and “choke points” in the metabolic networks,” Bioinformatics, vol. 22, no. 14, pp. 1767–1774, 2006. View at: Publisher Site | Google Scholar
  14. M. L. Acencio and N. Lemke, “Towards the prediction of essential genes by integration of network topology, cellular localization and biological process information,” BMC Bioinformatics, vol. 10, no. 1, 2009. View at: Publisher Site | Google Scholar
  15. S. A. Becker, A. M. Feist, M. L. Mo, G. Hannum, B. Ø. Palsson, and M. J. Herrgard, “Quantitative prediction of cellular metabolism with constraint-based models: the COBRA Toolbox,” Nature Protocols, vol. 2, no. 3, pp. 727–738, 2007. View at: Publisher Site | Google Scholar
  16. K. Plaimas, R. Eils, and R. König, “Identifying essential genes in bacterial metabolic networks with machine learning methods,” BMC Systems Biology, vol. 4, no. 1, p. 56, 2010. View at: Publisher Site | Google Scholar
  17. J. Oyelade, I. Isewon, O. Aromolaran, and E. Uwoghiren, “Computational prediction of alternative metabolic pathways of Plasmodium falciparum,” in Proceedings of the 10th International Conference on Bioinformatics and Computational Biology, BICOB 2018, 2018. View at: Google Scholar
  18. A. Chiappino-Pepe, S. Tymoshenko, M. Ataman, D. Soldati-Favre, and V. Hatzimanikatis, “Bioenergetics-based modeling of Plasmodium falciparum metabolism reveals its essential genes, nutritional requirements, and thermodynamic bottlenecks,” PLoS Computational Biology, vol. 13, no. 3, article e1005397, 2017. View at: Publisher Site | Google Scholar
  19. M. Kadivar, “A new $ O (m+ kn log overline {d}) $ algorithm to find the $ k $ shortest paths in acyclic digraphs,” Transactions on Combinatorics, vol. 5, no. 3, pp. 23–31, 2016. View at: Google Scholar
  20. I. Yeh, T. Hanekamp, S. Tsoka, P. D. Karp, and R. B. Altman, “Computational analysis of Plasmodium falciparum metabolism: organizing genomic information to facilitate drug discovery,” Genome Research, vol. 14, no. 5, pp. 917–924, 2004. View at: Publisher Site | Google Scholar
  21. M. A. Jing, Metabolic Network Based Gene Essentiality Analysis, Doctoral dissertation, 2012.
  22. C. Qin, Y. Sun, and Y. Dong, “A new method for identifying essential proteins based on network topology properties and protein complexes,” PLoS One, vol. 11, no. 8, article e0161042, 2016. View at: Publisher Site | Google Scholar
  23. R. Caspi, T. Altman, R. Billington et al., “The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of pathway/genome databases,” Nucleic Acids Research, vol. 42, no. D1, pp. D459–D471, 2014. View at: Publisher Site | Google Scholar
  24. M. Arita, “The metabolic world of Escherichia coli is not small,” Proceedings of the National Academy of Sciences, vol. 101, no. 6, pp. 1543–1547, 2004. View at: Publisher Site | Google Scholar
  25. T. Kim, K. Dreher, R. Nilo-Poyanco et al., “Patterns of metabolite changes identified from large-scale gene perturbations in Arabidopsis using a genome-scale metabolic network,” Plant Physiology, vol. 167, no. 4, pp. 1685–1698, 2015. View at: Publisher Site | Google Scholar
  26. T. Brys, A. Harutyunyan, P. Vrancx, M. E. Taylor, D. Kudenko, and A. Nowé, “Multi-objectivization of reinforcement learning problems by reward shaping,” in International Joint Conference on Neural Networks (IJCNN), pp. 2315–2322, 2014, IEEE. View at: Publisher Site | Google Scholar
  27. M. J. Mataric, “Reward functions for accelerated learning,” in Machine Learning Proceedings 1994, pp. 181–189, 1994, Elsevier. View at: Publisher Site | Google Scholar
  28. J. Randløv and P. Alstrøm, “Learning to drive a bicycle using reinforcement learning and shaping,” in ICML, vol. 98, pp. 463–471, Citeseer, 1998. View at: Google Scholar
  29. R. S. Sutton and A. G. Barto, Reinforcement learning: An introduction, vol. 1, MIT press, Cambridge, 1998.
  30. S. Devlin, D. Kudenko, and M. Grześ, “An empirical study of potential-based reward shaping and advice in complex, multi-agent systems,” Advances in Complex Systems, vol. 14, no. 02, pp. 251–278, 2011. View at: Publisher Site | Google Scholar
  31. K. Efthymiadis and D. Kudenko, “Using plan-based reward shaping to learn strategies in StarCraft: Broodwar,” in 2013 IEEE Conference on Computational Inteligence in Games (CIG), pp. 1–8, IEEE, 2013. View at: Google Scholar
  32. P. Gerold and R. T. Schwarz, “Biosynthesis of glycosphingolipids de-novo by the human malaria parasite Plasmodium falciparum,” Molecular and Biochemical Parasitology, vol. 112, no. 1, pp. 29–37, 2001. View at: Publisher Site | Google Scholar
  33. D. D. van Niekerk, G. P. Penkler, F. du Toit, and J. L. Snoep, “Targeting glycolysis in the malaria parasite Plasmodium falciparum,” The FEBS Journal, vol. 283, no. 4, pp. 634–646, 2016. View at: Publisher Site | Google Scholar
  34. Z. Bozdech and H. Ginsburg, “Data mining of the transcriptome of Plasmodium falciparum: the pentose phosphate pathway and ancillary processes,” Malaria journal, vol. 4, no. 1, p. 17, 2005. View at: Publisher Site | Google Scholar
  35. M. A. Carey, J. A. Papin, and J. L. Guler, “Novel Plasmodium falciparum metabolic network reconstruction identifies shifts associated with clinical antimalarial resistance,” BMC Genomics, vol. 18, no. 1, p. 543, 2017. View at: Publisher Site | Google Scholar
  36. M. J. Crawfod, M. J. Fraunholz, and D. S. Roos, “Part III-Energy metabolism in the apicomplexa,” in Molecular and Medical Parasitology, pp. 154–169, Academic Press, 2003. View at: Publisher Site | Google Scholar
  37. M. D. Alonso, J. Lomako, W. M. Lomako, and W. J. Whelan, “A new look at the biogenesis of glycogen,” FASEB Journal, vol. 9, no. 12, pp. 1126–1137, 1995. View at: Publisher Site | Google Scholar
  38. S. E. Babbitt, L. Altenhofen, S. A. Cobbold et al., “Plasmodium falciparum responds to amino acid starvation by entering into a hibernatory state,” Proceedings of the National Academy of Sciences, vol. 109, no. 47, pp. E3278–E3287, 2012. View at: Publisher Site | Google Scholar
  39. C. Huthmacher, A. Hoppe, S. Bulik, and H. G. Holzhütter, “Antimalarial drug targets in Plasmodium falciparum predicted by stage-specific metabolic network analysis,” BMC Systems Biology, vol. 4, no. 1, 2010. View at: Publisher Site | Google Scholar
  40. V. Choubey, P. Maity, M. Guha et al., “Inhibition of Plasmodium falciparum choline kinase by hexadecyltrimethylammonium bromide: a possible antimalarial mechanism,” Antimicrobial Agents and Chemotherapy, vol. 51, no. 2, pp. 696–706, 2007. View at: Publisher Site | Google Scholar
  41. G. Pessi, G. Kociubinski, and C. B. Mamoun, “A pathway for phosphatidylcholine biosynthesis in Plasmodium falciparum involving phosphoethanolamine methylation,” Proceedings of the National Academy of Sciences of the United States of America, vol. 101, no. 16, pp. 6206–6211, 2004. View at: Publisher Site | Google Scholar
  42. W. H. Witola, K. El Bissati, G. Pessi, C. Xie, P. D. Roepe, and C. B. Mamoun, “Disruption of the Plasmodium falciparum PfPMT gene results in a complete loss of phosphatidylcholine biosynthesis via the serine-decarboxylase-phosphoethanolamine-methyltransferase pathway and severe growth and survival defects,” Journal of Biological Chemistry, vol. 283, no. 41, pp. 27636–27643, 2008. View at: Publisher Site | Google Scholar
  43. C. Wanidworanun, R. L. Nagel, and H. L. Shear, “Antisense oligonucleotides targeting malarial aldolase inhibit the asexual erythrocytic stages of Plasmodium falciparum,” Molecular and Biochemical Parasitology, vol. 102, no. 1, pp. 91–101, 1999. View at: Publisher Site | Google Scholar
  44. S. Fatumo, K. Plaimas, J.-P. Mallm et al., “Estimating novel potential drug targets of Plasmodium falciparum by analysing the metabolic network of knock-out strains in silico,” Infection, Genetics and Evolution, vol. 9, no. 3, pp. 351–358, 2009. View at: Publisher Site | Google Scholar
  45. S. R. Krungkrai and J. Krungkrai, “Insights into the pyrimidine biosynthetic pathway of human malaria parasite Plasmodium falciparum as chemotherapeutic target,” Asian Pacific Journal of Tropical Medicine, vol. 9, no. 6, pp. 525–534, 2016. View at: Publisher Site | Google Scholar
  46. G. A. McConkey, “Plasmodium falciparum: isolation and characterisation of a gene encoding protozoan GMP synthase,” Experimental Parasitology, vol. 94, no. 1, pp. 23–32, 2000. View at: Publisher Site | Google Scholar
  47. R. H. Barker, V. Metelev, E. Rapaport, and P. Zamecnik, “Inhibition of Plasmodium falciparum malaria using antisense oligodeoxynucleotides,” Proceedings of the National Academy of Sciences, vol. 93, no. 1, pp. 514–518, 1996. View at: Publisher Site | Google Scholar
  48. D. Chakrabarti, S. M. Schuster, and R. Chakrabarti, “Cloning and characterization of subunit genes of ribonucleotide reductase, a cell-cycle-regulated enzyme, from Plasmodium falciparum,” Proceedings of the National Academy of Sciences, vol. 90, no. 24, pp. 12020–12024, 1993. View at: Publisher Site | Google Scholar
  49. S. D. Lytton, B. Mester, J. Libman, A. Shanzer, and Z. I. Cabantchik, “Mode of action of iron (III) chelators as antimalarials: II. Evidence for differential effects on parasite iron-dependent nucleic acid synthesis,” Blood, vol. 84, no. 3, pp. 910–915, 1994. View at: Google Scholar
  50. K. Eaazhisai, R. Jayalakshmi, P. Gayathri et al., “Crystal structure of fully ligated adenylosuccinate synthetase from Plasmodium falciparum,” Journal of Molecular Biology, vol. 335, no. 5, pp. 1251–1264, 2004. View at: Publisher Site | Google Scholar

Copyright © 2019 Jelili Oyelade et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


More related articles

681 Views | 399 Downloads | 2 Citations
 PDF  Download Citation  Citation
 Download other formatsMore
 Order printed copiesOrder

Related articles

We are committed to sharing findings related to COVID-19 as quickly and safely as possible. Any author submitting a COVID-19 paper should notify us at help@hindawi.com to ensure their research is fast-tracked and made available on a preprint server as soon as possible. We will be providing unlimited waivers of publication charges for accepted articles related to COVID-19. Sign up here as a reviewer to help fast-track new submissions.