Plasmodium falciparum genetic factors rather than the hosts' are likely to drive resistance to ACTs in Ghana

Background: Artemisinin-based Combination Therapy (ACT) partner drugs, currently used in Ghana are lumefantrine, amodiaquine, and piperaquine. Plasmodium falciparum isolates with reduced susceptibility to these partner drugs may affect treatment outcome. Mutations in Pfmdr1 gene is linked to reduced parasite susceptibility to amodiaquine and lumefantrine. In addition, the potency of the partner drugs in vivo depends on the metabolism by the cytochrome P450 (CYP) enzyme in the host. Mutations in the CYP2C8 and CYP3A4 genes are linked to reduced metabolism of amodiaquine and lumefantrine respectively in vitro. This study investigated the host and parasite genetic factors affecting the susceptibility of the malaria parasite to ACT partner drugs. Methods: Archived samples from 240 patients aged ≤ 9years participating in antimalarial drug resistance survey in of Ghana and given AL or AA were selected and analyzed. Polymerase chain reaction (PCR) followed by Sanger sequencing was used to determine the polymorphisms in CYP2C8, CYP3A4, and Pfmdr1 genes. Results: For CYP3A4, all had wild type alleles suggesting that the hosts are good metabolizers of lumefantrine. For CYP2C8 60% had wild type alleles, 35% heterozygous and 5% homozygous recessive alleles suggesting ecient metabolism of amodiaquine by the hosts. For Pfmdr1 gene, at codon 86, 95% were wild type (N86) and 5% mutant (Y86). For codon 184, 36% were wild type (Y184) and 64% mutant (F184) while for codons 1034, 1042 and 1246, 100% (all) were wild type. The high prevalence of N86-F184-D1246 haplotype (NFD) suggest presence of parasites with reduced susceptibility to lumefantrine and not amodiaquine. Delayed clearance was observed in individuals with mutations in the Pfmdr1 gene and not Cytochrome 450 gene. zone and 9 in the Savannah zone. Multiply synonymous mutations were observed in samples from Savannah zone. With regard to the novel non-synonymous mutations, 5 were observed in samples from the Coastal zone whilst 3 were seen in samples from the Savannah zone.

Results: For CYP3A4, all had wild type alleles suggesting that the hosts are good metabolizers of lumefantrine. For CYP2C8 60% had wild type alleles, 35% heterozygous and 5% homozygous recessive alleles suggesting e cient metabolism of amodiaquine by the hosts. For Pfmdr1 gene, at codon 86, 95% were wild type (N86) and 5% mutant (Y86). For codon 184, 36% were wild type (Y184) and 64% mutant (F184) while for codons 1034, 1042 and 1246, 100% (all) were wild type. The high prevalence of N86-F184-D1246 haplotype (NFD) suggest presence of parasites with reduced susceptibility to lumefantrine and not amodiaquine. Delayed clearance was observed in individuals with mutations in the Pfmdr1 gene and not Cytochrome 450 gene. Both synonymous and nonsynonymous mutations were observed in the Pfmdr1 at low prevalence.
Conclusion: The outcome of this study indicates that parasite's genetic factors rather than the hosts' are likely to drive resistance to ACTs in Ghana.

Background
Malaria caused by an infection of Plasmodium falciparum is one of the major causes of morbidity and mortality in sub-Saharan Africa, especially in children under 5 years and pregnant women (WHO, 2018). WHO recommends the use of a combination of a fast-acting artemisinin derivative and a relatively slowacting partner drug, for the treatment of uncomplicated malaria in disease-endemic areas (WHO, 2014).
The recommended rst line ACTs in Ghana for treating uncomplicated malaria are artesunate with amodiaquine (AA), artemether with lumefantrine (AL) or a combination of Dihydroartemisinin with piperaquine (Ministry of Health, 2009). The reason for combining the drugs (ACT) is to slow down the development of resistance to the antimalarial drugs by P. falciparum (White & Nosten, 2007). The fast acting drug quickly reduces the parasite load whilst the slow acting antimalarial gradually mob up the residue parasites. The potency of the ACTs is therefore dependent on the e cacy of both the artemisinin component and the partner drug (White & Nosten, 2007). Reduced susceptibility of parasites to the partner drugs in the ACTs can therefore potentially result in resistance to artemisinin in future as parasites that escape the fast action of the artemisinin or its derivatives will not be cleared by the partner drugs and this could allow ample time for growth and expansion of 'drug resistant parasite population' (White & Nosten, 2007).
The variations observed in the effectiveness of ACTs in malaria-endemic regions is dependent on the parasite genetic factors (Ouji et al., 2018), as well as the human genetic factors (Zanger & Schwab, 2013). For the parasite genetic factors, polymorphisms which arise due to single nucleotide changes in the Pfmdr1 gene in its coding region have been linked to differential parasite susceptibility to the ACT partner drugs, like amodiaquine (Sá et al., 2009) and lumefantrine (Sisowath et al., 2007). This makes Pfmdr1 an important likely candidate for initiating ACT partner drug resistance (Chen et al., 2010).The polymorphic Pfmdr1 alleles that are mostly found in Africa are N86Y, F184Y, and D1246Y. The P. falciparum N86-F184-D1246 haplotype (NFD haplotype) has been linked to decreased susceptibility to antimalarial drugs such as me oquine and lumefantrine. The selection of the NFD haplotype has been seen in malaria treatment using artemether+lumefantrine. The different haplotype which is the Y86-Y184-Y1246 haplotype (YYY haplotype) is associated with reduced amodiaquine susceptibility (Holmgren et al., 2007).
Differences in the genetic make-up of humans are the principal factor that de nes the level of drug availability in the blood to clear the parasites (Zanger & Schwab, 2013). The cytochrome P450 enzyme family (CYP genes) is involved in the metabolism of the different antimalarial drugs (Zanger & Schwab, 2013). Amodiaquine is mainly metabolized by CYP2C8 (Parikh et al., 2007) whiles Lumefantrine is metabolized mainly by CYP3A4 (Lee et al., 2012). Different mutations in the promoter region, introns or exons can result in different alleles of the CYP450 genes in different individuals. The metabolism of a drug or a combination of drugs could be decreased, increased or unaffected depending on the allele(s) an individual possesses (Wu, 2011). Elucidating the exact role these disparities in the genes coding for the enzymes involved in ACTs metabolism is vital for understanding the inter-individual pharmacokinetic differences observed in persons using ACTs (Ingelman-Sundberg & Rodriguez-Antona, 2005). This study investigated P. falciparum and host genetic factors that are likely to affect the e cacy of the ACT partner drugs used in Ghana.

Study Design
Blood blot lter paper used in this study were archived samples collected in 2016 from children participating in studies previously described by Abuaku and colleagues (Abuaku et al., 2016). Their study was part of routine surveillance on the therapeutic e cacy of ACT in Ghana; the e cacy of amodiaquineartesunate (AA) and artemether-lumefantrine (AL) were studied in six sentinel sites representing the forest and savannah zones of the country. Three sites representing the two ecological zones studied AA whilst the other three sites studied AL. In each site, the study was a one-arm, prospective, evaluation of the clinical, parasitological, and haematological responses to directly observed treatment with either AA or AL among children 6 months to 9 years old with uncomplicated falciparum malaria. The WHO 2009 protocol on surveillance of anti-malaria drug e cacy was used for the study with primary outcomes as prevalence of day 3 parasitaemia and clinical and parasitological cure rates on day 28.
An informed consent was obtained from each parents or guardian. A medical doctor prescribes either AA or AL to the study participants who were then followed up for 28 days.
The archived samples were selected from three sentinel sites, Navrongo, Begoro and Cape Coast which represent the three distinct eco-epidemiological zones in Ghana ( Figure 1). Begoro is located in the tropical forest ecological zone, Navrongo is in the Northern Savanna ecological zone and Cape-Coast is situated in the Coastal Savanna ecological zone. The samples (100 μl blood) were collected on Whatman 3 lter paper (Little Chalfont, UK), stored in plastic bags containing silica gels, and kept at room temperature until use.
The samples from a group of the participants referred hereafter as 'cohort 1' comprises of 120 of the recruited patients who were given AL and a second group, cohort 2, comprising of 120 patients received AA. Of the 240 study participants' archived samples analysed, 60 were selected from the Savannah zone, 90 from the Coastal zone and 90 from the Forest zone.

DNA extraction
Malaria parasite DNA was extracted from the archived blood blots lter papers using a QIAamp DNA minikit (Qiagen, Hilden, Germany) following the manufacturer's protocol. Convectional PCR (Lorenz, 2012) was performed to amplify the region of interest using published protocols (Vinayak et al., 2010;Cavaco et al., 2005;Hodel et al., 2009). PCR products were sequenced using Sanger sequencing at Macrogen, Netherlands.

Detection of Pfmdr1 polymorphisms
The regions of Pfmdr1 gene was ampli ed and sequenced to determine the presence of any mutation. The ampli cation was carried out using the protocol described by Vinayak and colleagues (Vinayak et al., 2010). The polymorphisms were analysed at codons 86 (asparagine to tyrosine), 184 (tyrosine to phenylalanine), 1034 (serine to cysteine), 1042 (asparagine to aspartic acid) and 1246 (aspartic acid to tyrosine). A polymerase chain reaction followed by Sanger sequencing was used in determining these polymorphisms. Thirty microliters aliquot of PCR products were kept on ice and shipped for sequencing at Macrogen, Netherlands.
Polymorphisms in the Pfmdr1 have been linked to differential susceptibility to amodiaquine (Sá et al., 2009) and lumefantrine (Sisowath et al., 2007). The polymorphic Pfmdr1 alleles mostly found in Africa are N86Y, F184Y, and D1246Y. The P. falciparum NFD haplotype is associated with decreased susceptibility to lumefantrine while the other haplotype, YYY, is associated with reduced susceptibility to amodiaquine (Holmgren et al., 2007).

Detection of CYP2C8 and CYP3A4 polymorphisms
The CYP2C8 polymorphisms were analysed at codon 269 (Isoleucine to phenylalanine). The CYP3A4 polymorphisms were analysed at position -392A>G of the proximal promoter region. A polymerase chain reaction followed by Sanger sequencing was used to determine the polymorphism in CYP2C8 as reported by the group of Cavaco (Cavaco et al., 2005) and CYP3A4 as described by Hodel and colleagues (Hodel et al., 2009). A PCR product of 120 bp and 717 bp represent a successful ampli cation of CYP2C8 and CYP3A4 respectively. Aliquot of the PCR products was shipped appropriately for sequencing at Macrogen, Netherlands.

Data Analysis
Data was organized using R software, SPSS software (version 20) and GraphPad Prism version 6.
Sequences data were analyzed with the BLAST program (http://blast.ncbi.nlm.nih.gov/) to determine the authenticity of the sequences. Multiple sequences were aligned with MAFFT (EMBL.EBI, Hinxton, Cambridge, UK) using the 3D7 wild-type as reference. Consensus sequence editing and single nucleotide polymorphism (SNP) detection was carried out using the CLC Main Workbench 7.9.1 (Qiagen, Aarhus, Denmark). The CYP2C8 sequences were aligned to CYP2C8 (ENSG 00000138115) as reference sequence whiles CYP3A4 sequences were aligned to CYP3A4 (ENSG 00000160868) as reference sequence from NCBI database. Genotype deviations from the Hardy-Weinberg equilibrium were also determined. The Hardy-Weinberg equilibrium determines whether or not the allele or genotype frequencies for a particular gene will remain constant from generation to generation in the absence of evolutionary in uences such as genetic drift, inbreeding, and founder effect (Wigginton et al., 2005). All tests in this study were considered statistically signi cant when p-value <0.005.
Prevalence of isolates with Pfmdr1 codons 86, 184, 1034, 1042, 1246 alleles and NFD haplotype Ninty-ve P. falciparum out of 120 clinical isolates was successfully genotyped for the Pfmdr1 gene. For Pfmdr1 genotype at codon 86, 95% (90/95) were wild type (N86) and 5% (5/95) mutant (Y86) [ Figure 3]. For codon 184, 36% (34/95) were wild type (Y184), 64% (61/95) mutant (F184) while for codons 1034, 1042 and 1246, all (100%) were wild type. There were both nonsynonymous and synonymous mutations observed at low frequencies in some of the samples analyzed (Table 1). The Pfmdr1 haplotypes observed were 57.8% (55/95) NFD, 34.7% (33/95) NYD, 6.3% (6/95) YFD and 1% (1/95) YYD. Nonsynonymous mutations that lead to change in amino acid and synonymous mutations which do not lead to a change in amino acid were both observed at low frequencies in some of the samples analyzed.It must be noted that natural selection on both synonymous and nonsynonymous mutations plays an important role in shaping levels of synonymous polymorphism. In this study, six novel synonymous mutations were observed in the Coastal zone and 9 in the Savannah zone. Multiply synonymous mutations were observed in samples from Savannah zone. With regard to the novel non-synonymous mutations, 5 were observed in samples from the Coastal zone whilst 3 were seen in samples from the Savannah zone.

Discussion
It is obvious that resistance of P. falciparum to the ACT partner drugs may lead to the gradual evolution of strains of the parasites with reduced susceptibility to the artemisinins. The failure of the partner drugs should therefore be of great concern to the National Malaria Control Program in disease endemic areas. Since both the host and the parasite genome play a role in metabolism of the ACTs, a key question often asked is: what drives parasites resistance to the ACT partner drugs? Before we attempt to address this question let us systematically examine the clinical data generated in this study, which is a subset of a bigger data published elsewhere. The clinical data indicates that 13% (16/120) of the participants treated with AL (cohort 1) still carried parasites on day 3 post-treatment compared to 4% (5/120) of those given AA (cohort 2). However all parasites were cleared by day 7 post treatment. This observation, indicate a better rate of parasite clearance with AA than AL. This observation is not surprising, as it gives credence to previous report by Abuaku and colleagues (Abuaku et al., 2016) inferring from the entire population data from which our samples set was drawn. probably explain the slight difference in e cacy between AL and AA observed in that study.
The e cacy of the partner drugs investigated in this study, amodiaquine and lumifantrine, is linked to Pfmdr1 gene which is part of the ATP-Binding Cassette (ABC) transporters (Ferreira et al., 2011). This gene encodes a transporter which is found in the digestive vacuole of the parasite (Bopp et al., 2018). The Pfmdr1 is thought to function by pumping compounds out of the parasite thus making it an important protein for antimalarial drug resistance. It must be emphasized that the true mechanistic role of the Pfmdr1 in initiating antimalarial drug resistance is poorly understood (Chen et al., 2010) although certain mutations in the gene have been associated with resistance to different antimalarial drugs (Sá et al., 2009;Sisowath et al., 2007). The exact mechanism in which mutation at the Pfmdr1 F184 confer resistance to lumefantrine whiles mutations at the Pfmdr1 Y86 and Y1246 confer resistance to amodiaquine is not well understood but have been observed to be mostly selected for during lumefantrine and amodiaquine drug pressure respectively (Sá et al., 2009;Sisowath et al., 2007).
There were high prevalence of N86, F184, and D1246 haplotypes in this study with no record of Y86, Y184 and Y1246 haplotypes. This observation is consistent with that reported by Duah and colleagues (Duah et al., (2013). The results also showed the widespread presence of these mutations in Ghana which are not ecological zonal bias. This was because there was no signi cant difference in these mutations across the different ecological zones (Supplementary Figure 1 &2).
The cytochrome P450 enzyme family (CYP genes) is a key enzyme involved in the metabolism of different antimalarial drug (Zanger & Schwab, 2013). Lumefantrine is metabolized to desbutylbenflumetol mainly by CYP3A4 (Lefevre & Thomseadn, 1999). Mutation in the gene proximal promoter region which results from a change from adenine (A) to guanine (G) at the position 392 results in CYP3A4*1B (Lamba, Lin, Schuetz, & Thummel, 2012) have been observed to have poor enzyme activity (Mutagonda et al., 2017). From the results obtained in the current study, 93 individuals were successfully genotyped for CYP3A4 of which 100% had the wild type gene. This observation suggest that lumefantrine is well metabolized in the participants. Again, delayed clearance observed in patients treated with AL were seen to have one or more mutations in the Pfmdr1 gene of the P. falciparum clinical isolates rather than mutation in the CYP3A4 gene of the individuals (Table 2). Based on these observations we are quick to infer that the parasite genetic factors could be the driving force behind drug e ciency in the children treated with AL and this could possibly be the determinant of clinical resistance to the ACT in future. However in saying this, we still tread on the side of caution since our inference could be more or less speculative due to our inability to conduct any pharmacokinetic studies of desbutyl-lumefantrine in the children in order to back our assertion. Interestingly, similar ndings have been reported by the group of Kiaco (Kiaco et al., 2017). Sixteen of the samples with delayed clearance had the CYP3A4 wild type individuals for lumefantrine metabolism and parasite Pfmdr1 mutation(s) as indicated.
The CYP2C8 is the main enzyme that metabolizes amodiaquine to desethyl amodiaquine (DEAQ) (Li et al., 2002). The wild type CYP2C8*1 and the mutant CYP2C8*2 are the most predominant in Ghana (Kudzi et al., 2009). A change from adenine (A) to thymine (T) at nucleotide position 895 on exon 5 results in the CYP2C8*2 mutant. CYP2C8*2 has been shown to be associated with decreased enzyme activity in vitro and reduced intrinsic clearance of amodiaquine (Parikh et al., 2007). From the results of the study, 94 individuals were successfully genotyped for CYP2C8 of which 60% (56/94) had wild type alleles, 35% (33/94) heterozygous and 5% (5/94) homozygous recessive alleles. This result is contrary to what has been reported by Kudzi et al., (2009). The high number of individuals with wild type CYP2C8 suggests that amodiaquine was well metabolized in the participants. It must however be emphasized that delayed clearance was observed in individuals who reported with high parasitemia (parasitemia >100,000) on day 0 and with one or more mutation(s) in the Pfmdr1 gene. These individuals had either wild type or heterozygous CYP2C8 genotype (Table 3) suggesting ample concentration of DEAQ in their plasma. Thus it was expected that their parasites should have been easily cleared. There was no delayed clearance observed in CYP2C8*2 individuals. We speculate that the CYP2C8 genotype of an individual may not alter the metabolism of the drug signi cantly, hence the plasma concentration of DEAQ may be adequate to clear the parasite. The absence of delayed clearance in CYP2C8*2 individuals can also be explained by the fact that dihydroartemisinin (DHA) which is a metabolite of artesunate in the patients clears most of the parasites and leaves only a few supposedly 'weakened parasite' residues making the presence of a suboptimal concentration of DEAQ enough to clear the parasite residue in these individuals. When Chi square test was used to determine the association between CYP2C8/ CYP3A4 and Pfmdr1 genotypes and day 2 positivity, there was no signi cant difference. For the few cases of delayed parasite clearance using AA, the lack of association between the wild type enzyme and the cases indicate that the hosts' gene-type of the enzyme couldn't be responsible for the delayed parasite clearance. Therefore this observation suggest that the parasite genetic factor among others could be responsible for the delayed clearance rather than the host genetic factors.
There were similar numbers of both non-synonymous and synonymous mutations observed at low frequencies in the coastal and forest ecological zones (Table 1). The synonymous mutations may not have any signi cant effect on the susceptibility of the parasite to the antimalarial drugs since it does not lead to change in amino acids. However, the novel non-synonymous mutations observed in this study may suggest the possible emergence of new mutations that may lead to reduced parasites susceptibility to ACTs in Ghana sooner than later.

Conclusion
All put together, observations made in this study give us enough grounds to conclude that the parasite genetic factors rather than the host's is more likely to drive resistance to the ACTs especially, the AL in Ghana. All individuals successfully genotyped for CYP3A4 were wild type, suggesting that lumefantrine is well metabolized in the participants. The high percentage of CYP2C8 wild type individuals also suggests that amodiaquine is metabolized e ciently. High prevalence of N86, F184, and D1246 suggests AL is less e cacious than AA. The outcome of this study conveys a warning that the malaria parasites are pushing towards resistance to the antimalarial drugs. Prompt monitoring of the ACTs is required. There is the need also to start thinking of other antimalarial drugs that could be used as ACT partner drugs. been explained to them in their native language and given an opportunity to ask questions that is of concern to them.

Consent for publication
Not applicable Availability of data and materials All data generated or analyzed during this study are included in this published article.

Competing interests
The authors declare that they have no competing interests Funding This research work was funded with grant offered to Peter Hodoameda by West African Centre for Cell Biology of Infectious Pathogens as part of his Mphil Fellowship.
Authors' contributions PH, NDQ, and NBQ conceived and designed the study. PH carried out the molecular genetic studies and led the drafting of the manuscript. NDQ, and NBQ also contributed to drafting of the manuscript. OCH, SM, BA, and KK gave technical support and contributed signi cantly to drafting of the manuscript. All authors read and approved the nal version of the manuscript.

Figure 1
A map of Ghana. Showing the location of the study sites used in this study: Cape Coast, Begoro and Navrongo.