Effects of drug pressure and human migration on antimalarial resistance in circulating Plasmodium falciparum malaria parasites in Ecuador

Abstract Antimalarial resistance in Plasmodium falciparum is a public health problem in the fight against malaria in Ecuador. Characterizing the molecular epidemiology of drug resistance genes helps to understand the emergence and spread of resistant parasites. In this study, the effects of drug pressure and human migration on antimalarial resistance in P. falciparum were evaluated. Sixty-seven samples from northwestern Ecuador from the 2019–2021 period were analyzed. SNPs in Pfcrt , Pfdhps , Pfdhfr , Pfmdr-1 , Pfk13 and Pfaat1 were identified by Sanger sequencing and whole-genome sequencing. A comparison of the frequencies of the haplotypes was made with data from the 2013–2015 period. Also, nucleotide and haplotype diversity were calculated. The frequencies of the mutant haplotypes, CVM ET in Pfcrt and C I C N I in Pfdhfr , increased. NED F S D F Y in Pfmdr-1 was detected for the first time. While the wild-type haplotypes, SAKAA in Pfdhps and MYRIC in Pfk13 , remained dominant. Interestingly, the A16 V mutation in Pfdhfr that gives resistance to proguanil is reported in Ecuador. In conclusion, parasites resistant to chloroquine ( Pfcrt ) and pyrimethamine ( Pfdhfr ) increased in recent years, while parasites sensitive to sulfadoxine ( Pfdhps ) and artemisinin ( Pfk13 ) prevail in Ecuador. Therefore, the current treatment is still useful against P. falciparum . The frequent human migration between Ecuador and Colombia has likely contributed to the spread of resistant parasites. Keys words : Plasmodium falciparum , resistance, antimalarial, selective pressure, human migration.


Introduction
Malaria is a vector-borne infectious disease that is endemic in tropical and subtropical regions of the world 1 .Plasmodium falciparum causes the most severe malaria in humans and the highest morbidity and mortality worldwide 2 .According to the World Health Organization (WHO), in 2022, 249 million cases and 608 000 deaths by malaria occurred in the world, and 600 000 cases and an incidence of 3.6 cases per 1000 inhabitants occurred in the Americas 3 .In this region, the highest malaria incidence is attributed to P. vivax followed by P. falciparum 1,4 .
Since the approval of artemisinin-based combination therapy (ACT) against P. falciparum, several South American countries have implemented it in their treatment schemes 5,6 .Artemether (AR) and artesunate (AS) are the most common artemisinins (ARTs) used in ACTs 5 .Currently, seven of eleven countries in this region, including Ecuador, use artemether-lumefantrine (AL) + primaquine (PQ) as rst-line treatment and quinine (QN) + clindamycin (CD) as second-line treatment 5,7 .
Antimalarial resistance in P. falciparum is a public health problem that complicates malaria control in endemic regions 8,9 .Resistance is an evolutionary response to pressure exerted by drugs over parasites that allows them to survive and reproduce ( tness) during exposure to antimalarials 6 .This advantage is due to mutations that modify protein function and reduce antimalarial e cacy 10,11 .In P. falciparum, Single Nucleotide Polymorphisms (SNPs) have been identi ed in speci c genes that are related to resistance to antimalarials such as chloroquine (CQ), sulfadoxine-pyrimethamine (SP), and ARTs, among others [12][13][14] .
CQ resistance in P. falciparum originated in Southeast Asia in the late 1950s 14 .Worldwide, P. falciparum exhibits CQ resistance mediated by mutations in Pfcrt that change the con guration of the chloroquine resistance transporter (PfCRT), located in the digestive vacuole (DV), allowing drug expulsion 15,16 .High concentrations of CQ in the DV prevent the release of toxic endogenous products, which cause the death of the parasite 17 .Mutations at codons 72-76 are associated with CQ resistance 18,19 .In vitro studies show that all chloroquine-resistant parasites have the K76T mutation, essential to confer CQ resistance 15,19 .In the early 1960s, CQ resistance emerged independently in South America 14 , and CVMNT, CVMET and CVEIT haplotypes have been found to predominate on the Paci c coast and SVMNT in the Amazon region 20,21 .Speci cally, in Ecuador, CVMNT, CVMET and SVMNT have been previously reported 20,22 .In addition, other mutations such as H97Q, A220S, Q271E, N326D, C350R, I356T/L and R371I in Pfcrt contribute to different degrees of CQ resistance 15,23,24 .In South America, H97Q, A220S, N326D, S334N, I356L and R371I/T have been found 16,17,23,25 .While in Ecuador, A220S, N326D and I356L with CVMNT were identi ed in Ecu1110, a CQ resistant isolate from 1990 16,20 .
ART resistance in P. falciparum was rst reported in Southeast Asia in 2006 39 .Initially, ARTs are activated by heme and iron to then cause protein and proteasome damage, leading to prolonged cellular stress and parasite death 40 .Mutations at codons 476, 493, 539, 543 and 580 in Pfk13 are associated with ART resistance 36,40,41 .In South America, C580Y has only been found in Guyana since 2010 41,42 , while Ecuador, Colombia, Peru, Venezuela, Brazil, Suriname and French Guiana have not reported any mutation associated with ART resistance 20,42,43 .
In recent years, amino acid transporter 1 (Pfaat1) has been associated with resistance to quinolines such as QN, AQ MQ and especially CQ [44][45][46] .A study suggests that the PfAAT1 protein, located in the DV, transports amino acids out of the DV while allowing entry of quinolines 46,47 .However, the presence of mutations such as T162E changes the charge of PfAAT1, blocking drug entry into the DV, and causes resistance to CQ, QN, AQ and MQ 46 .One PfAAT1 mutation with demonstrated CQ-resistance phenotype (S258L) was found on the Paci c coast of Ecuador and Colombia 48, 49 .
In South America, P. falciparum populations exhibit varying degrees of enhanced relatedness as a result of bottlenecks caused by efforts to eliminate malaria in the region 11,21,50 .In Esmeraldas, Ecuador, a malaria outbreak was caused by a clonal expansion in 2013 21 .In addition, malaria outbreaks are in uenced by the spread of parasites through human hosts between endemic regions 21,22,51 .In the malaria-endemic countries of the Americas, 7775 imported cases were reported in 2016 and 8411 in 2017, while in Ecuador, the latest records reported 395 cases between 2015 and 2017, with 41.5% of them were due to P. falciparum 5,52 .
In 2012, Ecuador was considered one of 21 countries in the world with the potential to eliminate malaria by 2020 due to a reduction in cases since 2002 of more than 99% 2 .The decrease in malaria infections was largely due to the success of the malaria elimination programs; however, infections have increased since 2015 4 .This increase may be associated with factors such as the restructuring of the health system, climate change, indiscriminate drug use and the spread of parasites by vector and host 53,54 .Genetic surveillance can help to resolve whether the emergence of new resistant genotypes in P. falciparum are the result of de-novo mutations or the spread of parasites by human host migration.
This study focuses on the molecular epidemiology of drug resistance genes to determine the effect of drug pressure and human migration on antimalarial resistance in P. falciparum.This work aims to characterize the current epidemiological situation of resistance, contribute to the strengthening of health programs and notify the e ciency of the treatment schemes against P. falciparum in Northwest Ecuador (Fig. 1).
Whole-genome sequencing (2019-2021) P. falciparum samples underwent whole-genome sequencing (WGS) following selective whole genome ampli cation (SWGA) to boost representation of P. falciparum DNA within total DNA extracts 55 .WGS could be performed on fty-four of the sixty-seven P. falciparum samples.
WGS coverage varied across sites and samples but con rmed the mutations found by Sanger sequencing in Pfcrt (codons 72-76)  We found the A16V DHFR mutation in two samples from Tobar Donoso that have a clonal relationship (genome-wide identity by descent (IBD) > 0.99).This clone was not found within a set of 219 previously sequenced P. falciparum genomes that were sampled from the Paci c Coast from 1999 to 2017 49 .
However, it does have high relatedness (IBD = 0.71) to clonal cluster G from Carrasquilla et al (2022) 49 , which also carried the A16V DHFR variant.We examined the IBD tracts shared between these genomes and found that our new isolates and cluster G are IBD across a 0.8 Mb segment of chromosome 4, which includes the DHFR coding region.Genome wide, the extended length of the IBD segments, which in some instances cover entire chromosomes, suggests the genomes are separated by only a single recombination event.Given the temporal order of sampling, the Tobar Donoso genomes are likely the progeny of a cross between cluster G and another parasite, but we were unable to identify the second putative parent within our extended Paci c Coast data set.

Genetic diversity of drug resistance genes
The nucleotide diversity (π) and haplotype diversity (H d ) were calculated to determine the genetic diversity of Pfcrt, Pfdhps, Pfdhfr and Pfmdr- Esmeraldas (π = 0.009, H d =0.220) and no diversity was found in Tobar Donoso (Table 2).Pfk13 was excluded from the analysis because it did not exhibit polymorphisms in the molecular markers analyzed.

Discussion
The e cacy of antimalarials has contributed to the reduction of morbidity and mortality rates of malaria 57 .However, the spread and increase of antimalarial resistance contribute to the emergence of malaria outbreaks and hinder malaria elimination 24,42 .This study focused on the molecular epidemiology of drug resistance genes (Pfcrt, Pfdhps, Pfdhfr, Pfmdr-1, Pfk13 and Pfaat1) in P. falciparum in northwestern Ecuador between 2019 and 2021.We show P. falciparum CQ and pyrimethamine resistance expansion and most importantly the appearance of a mutation associated with proguanil resistance for the rst time in Ecuador.
The haplotypes found in Pfcrt (Fig. 2a) are clear evidence of strong pressure that CQ has exerted over P.
falciparum for more than ve decades in Ecuador (Fig. 4) 49,58  South America since 1950, even as prophylaxis in cooking salt 26,61 , caused a strong pressure that resulted in the emergence of CQR parasites 15,62 .Likewise, the indiscriminate use of CQ reported in several areas of the Paci c coast during the 1990s contributed to the selection of CQR parasites 63 .Besides, clonal expansions, very common in South America, may have caused the fast spread and increase of CQR parasites 21 , as well as the loss of CQ-sensitive parasites that can compete in absence of CQ pressure.
Other mutations such as H97Q, A220S, Q271E, N326D, C350S, I356T/L and R371I in Pfcrt have been associated with CQR parasites 16,17 .In this study, H97Q, A220S, N326D, S334N, I356L and R371T were found (Table 1).The presence of these mutant alleles may have contributed to CQ resistance in Ecuadorian P. falciparum because it has been suggested that at least three complementary mutations to K76T are necessary to confer high degrees of CQ resistance 18 .
Resistance to sulfadoxine and pyrimethamine (SP) appears to be more complex than CQR.Resistance mutations in Pfdhfr were found in 65% of Ecuadorian samples (Fig. 2c).In contrast, sulfadoxine resistance appears low as only one instance of a mutant Pfdhps haplotype, SGKAA, (Fig. 2b), was found among the Colombian, but not the Ecuadorian, samples.This difference may stem from the history of drug use in the region.In South America, pyrimethamine was used massively as a prophylactic against malaria in the 1950s 33 .Later, in the 1970s, pyrimethamine was co-administrated with sulfadoxine to combat CQR parasites 14 .Consequently, SP resistance has been reported in South America since 1980 26 .Speci cally, in Ecuador, SP was implemented as a second-line treatment in 1950 and, in 2005, it was used as rst-line treatment until 2006 (Fig. 4) 58 .Therefore, the emergence of resistant parasites was caused by the pressure exerted by SP for more than ve decades.However, the intermittent use of SP temporarily reduced SP pressure, which may have caused SP-sensitive parasites to continue to circulate in the region.So, the wild-type haplotypes in Pfdhps and Pfdhfr are still observed in Ecuador as in South America.Favorably, the intermittent use may have prevented SP resistance from becoming xed in the region.
Importantly, the genetic monitoring of Pfdhfr further suggests that resistance to proguanil has appeared in Ecuador.P. falciparum acquires resistance to cycloguanil, the active metabolite of proguanil via combined coding mutations for A16V and S108N/T in Pfdhfr 14,28,64 .In Colombia, A16V was detected for the rst time in combination with CNCSI and CICNI in Nariño between 2012 and 2013 31 .In this study, A16V with CNCSI was identi ed in two samples from Tobar Donoso (Table 1) and was previously detected in isolates from San Lorenzo (2016) 49 .While A16V was reported in Paci c coast parasites, the samples from San Lorenzo and Tobar Donoso are not the same clones.Instead, our IBD analysis shows that A16V was passaged onto a new genomic background through recombination 49 .Atovaquoneproguanil has been commercialized in Nariño, Colombia for many years 31 , suggesting that the emergence of A16V in Colombia has been caused by the use of atovaquone-proguanil as prophylaxis against malaria 31 .So, the presence of A16V in Tobar Donoso could be explained by recent recombination of parasites in the Paci c Coast of Colombia or Ecuador 49 .The possible spread of this mutation in Ecuador and Colombia raises an important point against the future use of proguanil in the region.
At Pfmdr-1, a shift in haplotype frequencies across time was found (Fig. 3d).While both common haplotypes have been circulating in South America for many years, the most frequent haplotype in 2019-2021 (the triple mutant NEDFSDFY) has not been sampled previously in Ecuador.Mutations in Pfmdr-1 have an important role in response to different antimalarials such as CQ, QN, AQ, MQ and LF 6,14,15 .
However, South American P. falciparum have not shown phenotypic changes in drug response even though Y184F, N1042D and D1246Y are commonly present 20,66 .An in vitro assay tested parasites from Nariño with NEDFSDFD and NEDFSDFY haplotypes, which were shown to be sensitive to AQ, MQ and LF 66 .In Esmeraldas, ESM-2013 (isolated from 2013) with Y184F and N1042D mutation showed sensitivity to QN, LF and AR 20 .Mutations in other genes located in the DV such as Pfcrt and Pfaat1 may therefore in uence resistance to these drugs 11,67 .
The prevalence of MYRIC haplotype (wild-type) in Pfk13 (Fig. 2e) shows sensitivity to ARTs in 100% of Ecuadorian P. falciparum samples.Previous studies in Ecuador, Colombia, Peru, Venezuela, Brazil, Suriname and French Guiana have not found any mutation in Pfk13 associated with ART resistance 20,42,43   .However, in Guyana, the C580Y mutation was detected in 2010 41,42 .There is, therefore, concern that C580Y could spread to other regions of South America, and molecular studies are key to monitoring for any spread or new emergence.In addition, K189T mutation in Pfk13 was identi ed in this study and recently reported in Ecuador, Colombia, Venezuela, Brazil, Guyana and French Guiana 42,49 .In Ecuador, AS was used with SP as a rst-line treatment in 2006; however, AL was already used as a second-line treatment in 2005 58 .AL has been the o cial rst-line treatment against P. falciparum since 2012 (Fig. 4) 58 .
In the last decade, resistance to quinolines such as CQ, QN, AQ and MQ has been associated with mutations in Pfaat1 [44][45][46] .The S258L mutation identi ed in this study (Table 1) shows evidence of having been under selection in the Paci c coast of Ecuador and Colombia 49 .Recently, Amambua-Ngwa et al (2023) found the same mutation segregating in Gambian parasites, and through CRISPR/Cas9 editing, validated its role in CQ resistance 48 .Likewise, in vitro study showed resistance to three compounds not yet cataloged (MMV668399, MMV011895 and MMV007224) when P380S, K238N, V185L, F230L mutations and an insertion at codon 35 were present in Pfaat1 47 .Pfaat1 is a highly conserved gene 46 , so the emergence of mutations may be related to pressure from quinolines.This pressure could have been caused mostly by CQ and QN, used historically in South America 26,68 , and AQ which was introduced in several areas of South America, including the Paci c coast, since 1949 24,59,69 .
As has been observed elsewhere in South America, our measured patterns of resistance haplotype frequency do not track contemporary drug use.Instead, the recent increase in proportions of mutant haplotypes in Ecuador such as CVMET (double mutant) in Pfcrt, CICNI (double mutant) in Pfdhfr and the appearance of NEDFSDFY (triple mutant) in Pfmdr-1 (Fig. 3) could be explained by clonal expansions caused by parasites from migrant humans and not by the pressure by CQ and SP treatments, because they have not been used against P. falciparum in the last two decades.A previous analysis of neutral microsatellites found that outbreaks that occurred in Tobar Donoso and Esmeraldas in 2019 and 2020, respectively, were mostly caused by a single parasite clone 70 .Furthermore, CVMET in Pfcrt, CICNI in Pfdhfr and NEDFSDFY in Pfmdr-1 were dominant in samples from migrant human hosts in this study (Fig. 2), and they have been found circulating in Nariño since 1999 24,26,31 .While in northwest Ecuador, CVMET and CICNI were reported in low proportions in 2013-2015 20 , and NEDFSDFY was absent, but they are now common in recent samples in this study.This suggests that recent outbreaks that occurred in Ecuador may have resulted from the spread of parasites by migrant humans from Nariño.
Previously, an important human migratory ow that contributes to the spread of parasites on the Paci c coast of Colombia has already been reported 51 .In the present study, the main routes of cross-border and intra-border human migration contributing to the spread of malaria parasites in the North coast of Ecuador and the South coast of Colombia are proposed (Fig. 5).Speci cally, the spread of resistant parasites to Ecuador could have occurred in two ways: 1) by Ecuadorians who traveled to Colombia and became infected with parasites from there, and 2) directly by infected Colombians who arrived in Ecuador.The main migration route between Nariño and San Lorenzo for several years has been by their seaports.Therefore, Colombian parasites could have reached the coast of Ecuador by the constant migratory ow along the Paci c coast due to the shell sh trade, legal and illegal mining, agriculture and illegal activities such as drug tra cking through the seaports of Nariño, San Lorenzo and Esmeraldas 71- 73 .In addition, it has recently been found that P. falciparum populations are shared between Nariño and San Lorenzo 49,50 .Currently, a new road has been opened in this area between Colombia and Ecuador, the bridge over Mataje River that connects Nariño and San Lorenzo 74 .This road will de nitely represent an important passage of malaria infected people between Ecuador and Colombia.Finally, the E15 road that connects San Lorenzo and Esmeraldas is an important corridor in the spread of malaria parasites.In addition, the presence of A16V in Pfdhfr (Table 1) in Tobar Donoso also suggests a process of human migration because this mutation has already found in San Lorenzo 49 and is very common in Nariño 31 .Possibly, the migration of human hosts occurred along of the Mira River, which comes from Nariño, and it is frequently used by Colombians to trade and transport supplies to Tobar Donoso 75 .Additionally, migration by the E10 road, which connects San Lorenzo and Carchi, could also be contributing to the spread of parasites to Tobar Donoso because it is frequently used to reach the south of Carchi province.From there third-order improvised roads and walking trails connect with Tobar Donoso.It is important to note that a helicopter from Esmeraldas transports food, medicine and soldiers to monitor the border at least once a month 75 , and could also be considered a route of entry of parasites from Esmeraldas to Tobar Donoso.
Factors such as drug pressure and host human migration in uence the emergence and spread of resistant parasites which contribute to the appearance of malaria outbreaks and complicate malaria control and elimination in Ecuador.The ndings of this study indicate that chloroquine and pyrimethamine-resistant parasites have increased in recent years, while sulfadoxine and artemisininsensitive parasites prevail in Ecuador.Therefore, the current treatment (artemether + lumefantrine) is still effective against P. falciparum.In addition, the genotypes of P. falciparum show rapid uctuations within the region, likely due to the boom-bust dynamics of clonal lineages coupled with frequent migration of human hosts across the Ecuador-Colombia border.Finally, this study shows the importance of the epidemiological characterization of resistance in P. falciparum in a region in the process of eliminating malaria: genetic surveillance of drug resistance genes is key to respond in a timely manner in case of therapeutic failure and prevent the spread of malaria.

Study area and parasite population
The study was carried out in Ecuador (South America) from 2019 to 2021, with samples collected in Esmeraldas and Carchi provinces located in the Northwest on the border with Colombia (Fig. 1).Sixtyseven blood samples were collected on Whatman lter paper from patients diagnosed with P. falciparum by microscopy or RDTs (Rapid Diagnostic Tests).Samples were collected by personnel of the MPH between 2019 and 2021 and delivered to CISeAL (Centro de Investigación para la Salud en América Latina) for molecular analysis.During this period, 702 infections by P. falciparum were registered in Northwest Ecuador 4 .Samples of local Ecuadorian cases corresponded to the counties of Esmeraldas (40.29%),San Lorenzo (7.46%) and Tobar Donoso (26.86%).The remaining 25.37% corresponded to migrants from Nariño, Colombia diagnosed in San Lorenzo.Additionally, data of P. falciparum samples (2013-2015 period) from a previous study in Ecuador were included for temporal comparisons 20 .

DNA extraction
DNA extraction of P. falciparum was performed from blood samples on Whatman lter paper stored at 4°C, using the commercial Purelink™ Genomic DNA Mini Kit, INVITROGEN following the protocol recommended by the manufacturer (Invitrogen, Life Technology, USA), and stored at -20°C for its molecular analysis.3) established in previous studies were used 20 .Ampli cations were carried out in SureCycler 8800 (Agilent Technologies, USA).Amplicon sizes were veri ed in 2% agarose gel and observed in UVP BioDoc-It digital transilluminator (UVP, CA, UK).After con rmation, ampli ed DNA was sent for puri cation and Sanger sequencing to Macrogen, South Korea.

Whole-Genome Sequencing
Fifty-four DNA samples selective whole genome ampli cation (SWGA) to boost representation of P. falciparum DNA within total DNA extracts 55 .The Broad Institute Genomics Platform then constructed NEBNext libraries and performed 150-bp paired-end sequencing on an Illumina NovaSeqSP instrument.Sequences were aligned and variants called using a standard BWA-GATK3 pipeline following the best practices established by the Pf3k consortium 76 .The WGS results were used to con rm the genotypes obtained by Sanger sequencing and to identify variants in the same genes outside the Sanger-sequenced regions.

Data Analysis
GraphPad Prism version 8.0.2 was used to calculate frequencies and make comparative graphs of Pfcrt, Pfdhps, Pfdhfr, Pfmdr-1 and Pfk13 haplotypes in each locality from the 2013-2015 and 2019-2022 periods.To compare the frequencies of haplotypes between two different time points, a Chi-square (x 2 ) test was used.Signi cance was de ned as p < 0.05.The number of polymorphic sites (S), the average number of pairwise nucleotide differences (k), nucleotide diversity (π), number of haplotypes (h) and haplotype diversity (Hd) were calculated in DnaSP software version 6.12.03.

Relatedness Analysis
We estimated relatedness all genome pairs using the program hmmIBD 77 .The algorithm estimates fractional (genome-wide) relatedness as identity by descent (IBD) and provides IBD status at all genomic coordinates included in the analysis.We used the length and position of these IBD tracts to infer relatedness at the Pfdhfr locus.The IBD analysis included both the Ecuador samples from this study and 219 additional samples from the South American Paci c Coast originally analyzed in Carrasquilla et al (2022) 49 .We excluded from the IBD analysis any genome with less than 30% of the genome covered at 5x.
The study was approved by the Research Ethics Committee on Human Beings of Ponti cia Universidad Católica del Ecuador registered in the document CEISH-571-2018 and by the Ministry of Public Health (MPH) of Ecuador registered in the document MSP-DIS-2019-044-O.
between the 2013-2015 and 2019-2021 periods in northwest Ecuador.In Pfcrt, CVMET frequency increased in San Lorenzo (from 0.22 to 0.

Table 2
Genetic diversity of molecular markers of drug resistance genes in P. falciparum.n: Number of samples, S: Number of polymorphic sites, k: Average number of pairwise nucleotide differences, π: Nucleotide diversity, h: Number of haplotypes, Hd: Haplotype diversity.