Background

Malaria is the parasitic disease having the greatest impact on public health [1]. It is caused by different species from the Plasmodium genus, these being widely distributed throughout the world’s tropical and sub-tropical regions [2]. These parasites cause 140–300 million clinical cases and more than half a million deaths annually [3, 4]. Plasmodium falciparum is considered the most lethal species, mainly affecting vulnerable populations in sub-Saharan Africa [4]. Even though efforts were initially concentrated on controlling this species, reports of ever-increasingly severe cases caused by Plasmodium vivax [5] and the appearance of drug-resistant strains during the last few years [6, 7] has made this species a growing public health problem, affecting more than a third of the world’s population, having high prevalence in Asia and South and Central America [3, 7, 8].

Designing an anti-malarial vaccine against P. vivax (as for P. falciparum) has been focused on blocking parasite-host interactions during different parasitic stages, especially during the blood phase responsible for the disease’s clinical manifestations [9, 10]. A large amount of P. vivax antigens have been characterized to date [10, 11], however, their genetic diversity should be assessed for selecting the best antigens for vaccine development [10, 12]. Highly polymorphic antigens can provoke allele-specific immune responses leading to protection having low efficacy after vaccination. On the contrary, those having limited diversity are attractive targets for being evaluated as candidates as they avoid an allele-specific immune response [13].

Most antigens characterized to date have been merozoite proteins [10, 11], including the microneme AMA1 protein and rhoptry neck (RONs) proteins. The interaction between these proteins (specifically AMA1–RON2) has been well described in Toxoplasma gondii and P. falciparum, these being the structural basis for the tight junction (TJ), a connective ring through which a parasite enters a host cell [14–18].

The RON protein complex (characterized in P. falciparum) consists of RON2, RON4 and RON5 proteins [14, 17, 19]. Even though the mechanisms regarding function and interaction between the complex’s proteins are not clear, they are considered important targets for blocking invasion. Various studies have highlighted the potential of AMA1 and RON2 as vaccine candidates, however, current knowledge concerning the other RONs is deficient. Co-localization studies and invasion models described for Plasmodium spp and T. gondii have led to establishing RON4’s convincing participation in the TJ [15, 17, 20, 21]. Likewise, its expression in the parasite’s invasive forms [21–23], the ability to provoke an immune response in natural malarial infections [23] and the protein’s conserved nature, specifically towards the C-terminal (inferred by comparative analysis between PfRON4 and TgRON4 amino acid sequences) [24] suggest that this protein plays an important role for the parasite and could thus be evaluated as vaccine candidate.

The P. falciparum RON4 orthologue has recently been characterized in the P. vivax VCG-I strain (Vivax Colombia Guaviare-I) [22]. PvRON4 (P. vivax RON4) is encoded by a gene having around 2657 bp in this species, expressed during the last hours of the intra-erythrocyte cycle and secreted from the rhoptry neck. This consists of signal peptide sequence, a low complexity domain formed by two types of tandem repeats, a double spiral alpha helix domain and five conserved cysteines towards the C-terminal [22]; the latter region seems to be highly conserved among P. vivax and parasite species infecting monkeys [25].

Bearing RON4’s potential participation in invasion in mind and given that parasite antigen genetic diversity is an obstacle for designing a completely effective vaccine against P. vivax, this study was thus aimed at using Colombian clinical isolates for evaluating pvron4 locus genetic diversity and the evolutionary mechanisms determining its variation pattern.

Methods

Sample collection

Plasmodium vivax genomic DNA was obtained from 73 clinical isolates collected from 2007 to 2015 (2007: 10, 2008: 12, 2010: 18, and 2015: 33 samples). These came from Colombia’s Pacific coast region (Chocó and Nariño departments), Urabá/lower Cauca/southern Córdoba (Córdoba and Antioquia departments) and the Orinoquia-Amazonia region (Amazonas and Guainía departments), representing the three regions having the greatest transmissibility in Colombia [26]. More than 360,000 cases of P. vivax infection were recorded between 2007 and 2015, more than 14 % of them regarding Colombia’s Pacific coastal region, Urabá/lower Cauca/southern Córdoba 62 % while 7.5 % of P. vivax cases were recorded in the Orinoquia-Amazonia region. Malaria symptomatic patients (living in the regions described above) were diagnosed with P. vivax infection by microscopy and then invited to donate 5 mL of venous blood. Some Amazonia samples were collected and diagnosed, as has been reported elsewhere [27]. Male and female patients aged 16–64 years were invited to participate in the study. DNA was extracted and stored at −20 °C before being processed and were genotyped by PCR-RFLP from the pvmsp-3α gene.

Amplifying, cloning and sequencing the pvron4 locus

A set of primers was designed to amplify and clone pvron4 based on Sal-I genomic sequence (GenBank accession number AAKM01000005.1), sequences being as follows: pvron4 dir 5′ CACAGTGCAACCATGTCTCG 3′ (20 bp) and pvron4 rev 5′ GCAAGCTAATTTCACAAGTCTTC 3′ (23 bp) primers. Touchdown-PCR was used for amplification using the KAPA-HiFi HotStart Readymix enzyme (Kapa Biosystems) in 25 μL reactions using VCG-I strain genomic DNA as positive control. Thermal conditions were as follows: a 5 min denaturing step at 95 °C, 10 cycles consisting of 20 s at 98 °C, 15 s at 68 °C (temperature was reduced by one degree per cycle) and 1 min at 72 °C, followed by 35 cycles of 20 s at 98 °C, 15 s at 60 °C, 1 min at 72 °C and a final 5-min extension at 72 °C. PCR products were purified using an UltraClean PCR Clean-up purification kit (MOBIO).

The amplicons were ligated in pGEM T-easy cloning vector and then used for transforming Escherichia coli JM109 strain cells. Recombinant bacteria were selected by the alpha complementation method and their growth ability in the presence of ampicillin. These were confirmed by PCR using MangoTaq DNA polymerase and internal primers for the gene (intdir: 5′ TGTGGGTGGCGAGAGTG 3′ (17 bp), and intrev: 5′ ATATTTCCATTGCTGTACTAGG 3′ (22 bp), designed on Sal-I genomic sequence) using the following thermal conditions a 5 min denaturing step at 95 °C, 35 cycles of 20 s at 98 °C, 15 s at 60 °C, 1 min at 72 °C and a final 5-min extension at 72 °C. The plasmid from the two recombinant colonies per sample was extracted using an UltraClean 6 Minute Mini Plasmid Prep kit (MOBIO) and sent to be sequenced using a BigDye Terminator kit (MACROGEN, Seoul, South Korea), with universal primers SP6 Promoter Primer (Cat.# Q5011), T7 Promoter Primer (Cat.# Q5021) [28] and a set of internal primers (pvron4dseq: 5′ CACTAGAAAAGCTAAACATAAACC 3′ (24 bp), and pvron4rseq: 5′ ACTCCAATGGTCCTCAAC 3′ (18 bp) designed on the Sal-I genomic sequence) for sequencing.

Statistical analysis of pvron4 sequences

The electropherograms obtained by sequencing were assembled using CLC DNA workbench v.3 software (CLC bio, Cambridge, MA, USA). The 73 consensus sequences obtained in this study (Additional file 1), 7 reference sequences (from the Salvador-I (Sal-I, GenBank: XM_001615228.1), Mauritania-I (GenBank: AFNI01000694.1), India-VII (GenBank: AFBK01001155.1), Brazil-I (GenBank: AFMK01001544.1/AFMK01001546.1), North Korea (GenBank: AFNJ01001110.1), ctg (GenBank: AAKM01000005) and P01 (GeneDB: PVP01_0916600.1) strains) and 13 orthologous sequences (from Plasmodium cynomolgi (GeneBank: BAEJ01000746.1), Plasmodium knowlesi (GeneBank: NC_011910.1), Plasmodium inui (GeneBank: AMYR01000790.1/AMYR01000791.1), Plasmodium fragile (GeneBank: NW_012192637.1), Plasmodium coatneyi (GeneBank: CM002860.1), Plasmodium reichenowi (GeneBank: LVLA01000012.1), P. falciparum (GeneBank: XM_001347803.2), Plasmodium bergei (GeneBank: CAAI01002287.1), Plasmodium yoelii (GeneBank: AABL01000590.1), Plasmodium chabaudi (GeneBank: CAAJ01004050.1), Plasmodium vinckei (GeneBank: AMYS01000107.1), Plasmodium gaboni (GeneBank: LVLB01000012.1), and Plasmodium gallinaceum (Sanger Institute: gal28a.d000001405.Contig1/gal28a.d000000110.Contig1) were used for obtaining the deduced amino acid sequence used for multiple alignment with the MUSCLE algorithm [29]. Such alignment was manually edited to ensure correct repeat region alignment and then used for inferring DNA alignment by Pal2Nal software [30]. The T-REKS algorithm was used for identifying repeat regions [31].

DnaSP v.5 software [32] was used for calculating genetic diversity regarding Colombian sequences and P. vivax reference sequences alignment using estimators based on single nucleotide polymorphism (SNP) and sequence length (InDels). Tajima [33], Fu and Li [34], Fu [35], Fay and Wu [36] tests were used for evaluating deviations from the neutral model of molecular evolution, bearing coalescence simulations in mind for obtaining confidence intervals and their statistical significance. The repeat regions or those having gaps were not taken into account for analysis.

The Nei-Gojobori modified method [37] with MEGA v.6 software [38] was used for calculating the average number of synonymous substitutions per synonymous site (dS) and the average number of non-synonymous substitutions per non-synonymous site (dN) at intra-species level. The average amount of synonymous divergences per synonymous site (KS) and the average amount of non-synonymous divergences per non-synonymous site (KN) were calculated by modified Nei-Gojobori method with Jukes-Cantor correction [39] for determining natural selection signals throughout Plasmodium spp evolutionary history (using the P. vivax sequences, together with phylogenetically close parasites’ orthologous sequences). The differences between intra- and inter-species substitution rates were determined by Fisher’s exact test (recommended when the amount of synonymous and/or no-synonymous substitutions is fewer than ten) or the Z-test incorporated in MEGA software v6. Additionally, the McDonald–Kreitman (MK) test [40] with Jukes-Cantor correction was used for evaluating neutrality deviations using the Standard & Generalized McDonald–Kreitman Test web server [41, 42].

A sliding window was used for analysing evolutionary rate (ω = dN/dS and/or KN/KS) by evaluating the effect of selection throughout the gene. Individual sites under selection were identified by calculating synonymous and non-synonymous substitution rates per codon using SLAC, FEL, REL, IFEL [43], MEME [44], and FUBAR methods [45] in the Datamonkey online server [46]. Repeat regions or those having gaps were not taken into account for this analysis.

The random effects likelihood (REL)-branch-site method [47] was used for evaluating the existence of lineages under episodic diversifying selection in Plasmodium for the ron4 locus. The MUSCLE algorithm was used for aligning 14 orthologous protein sequences from different species from the genus; this was then used for inferring the best evolutionary model using MEGA software. Phylogeny was then inferred by using the maximum likelihood method with the JFF + G + F model. This is used as reference for analysing lineage-specific positive selection with the REL-branch-site method in the HyPhy package [48], using a DNA alignment inferred by Pal2Nal from aligning amino acids.

Effective number of codons

ENCprime [49] and DnaSP software were used for estimating the effective number of codons (ENC). This is a measurement of selective pressure at translational level, leading to protein function loss or gain [49]. This test compares the use of each codon versus a null distribution (uniform use of synonymous codons). ENC values close to 61 indicate that all synonymous codons for each amino acid are used equitably, while values close to 0 suggest bias or preferential codon use [50]. Statistical significance could be affected by gene length or recombination [50]. The codon bias index (CBI) was thus used, which takes values ranging from 0 (uniform use of synonymous codons) to 1 (maximum codon bias) [51].

Linkage disequilibrium and recombination

Linkage disequilibrium (LD) was evaluated by calculating the ZnS estimator [52]. A linear regression between this and the nucleotide distance was performed to see whether intragenic recombination occurred in pvron4. Recombination was also evaluated by ZZ estimator [53], the minimum number of recombination events (Rm) [54], the GARD algorithm [55] and the RDP v.3 software [56].

pvron4 locus differentiation and population genetic structure

The degree of genetic differentiation (or inter-population heterogeneity) in the pvron4 locus among Colombian P. vivax malaria-endemic regions was estimated by analysis of molecular variance (AMOVA) and Wright’s fixating index (FST), using the Arlequin v.3.1 software [57]. NETWORK v.5 software was used for constructing a median joining network for evaluating possible mutational pathways giving rise to pvron4 haplotypes, their distribution and frequencies. This method expresses multiple plausible evolutionary pathways as cycles, bound by vectors interpreted as extinct ancestral sequences [58].

Predicting pvron4 putative domains and antigenic potential

The Blastp algorithm from the NCBI web server was used for identifying putative domains in PvRON4 using the Sal-I sequence as reference. The B-cell epitope prediction tool available at the immune epitope database (IEDB) server was used for evaluating PvRON4s antigenic potential regarding its antigenicity [59], its hydrophobicity [60], protein solvent availability [61] and its potential linear B-cell epitopes [62]. These tests were done with two PvRON4 haplotypes differentiated by the amount of repeats: haplotype 6 (one copy of GEHGEHAEHGE) and haplotype 17 (seven copies of the repeat), to evaluate the effect of repeat number on protein antigenic behavior.

Results

pvron4 locus genetic diversity

Seventy-three sequences from the pvron4 locus obtained from the P. vivax Colombian population (24 from Orinoquia-Amazonia, 21 from the Pacific coast and 28 from Urabá/lower-Cauca/southern Córdoba) were analysed. pvron4 was initially annotated from the VCG-I strain as being a 2657 bp gene [22], however, the locus analysed from Colombian samples had a variation in length due to two tandem repeats located towards the gene’s 5′-end (Fig. 1). These repeats consisted of copies of the GTGGCGAGA nucleotide sequence encoding GES amino acids (repeated one to three times) and a longer sequence CGGAGAGCACGGTGAACACGCTGAACATGGGGAGCA encoding the GEHGEHAEHGE peptide (repeated one to seven times).

Fig. 1
figure 1

Representation of pvron4 encoding DNA sequence and sliding window for ω rate. ron4 ω values (dN/dS) within P. vivax are shown in green throughout. Likewise, ω divergence rates (KN/KS) between the species infecting primates and P. vivax are shown in blue (for the species evolutionarily closest to P. vivax) and red (most distant). A representation of the ron4 gene is given below the sliding window indicating signal peptide (blue), the repeat region (black) and the putative esterase/lipase domain (purple). The P. vivax SNP positions are shown as discontinue lines, their numbering is based on Additional file 2 alignment. The sites under purifying selection within species are represented by light green lines and dark green between species while positively selected sites between species are shown by red lines

Full size image

Few SNPs were found. Regarding the 2542 sites analysed concerning the Colombian sequences and the reference ones, the number of SNPs varied from five to eight (Table 1). The genetic diversity estimators gave low values (θw = 4.7 × 10−4 and π = 4.1 × 10−4) in the P. vivax Colombian population. Such values remained constant when comparing the Colombian sequences to the reference ones (Table 1). Likewise, the total amount of pvron4 haplotypes identified and haplotype diversity were low (Table 1), however, 15 haplotypes in the Colombian population (21 when Colombian and reference sequences are analysed together) and high diversity estimator values (Table 1) were found when analysing insertions/deletions (InDels) in pvron4. Bearing the SNPs and InDels between the pvron4 reference sequence and Colombian sequences in mind, 32 haplotypes were identified (Additional file 2).

Table 1 pvron4 genetic diversity estimators calculated from single nucleotide polymorphism and sequence length polymorphism
Full size table

Evaluating the effect of selection on the pvron4 locus

No statistically significant values were found for pvron4 when using the Tajima, Fu and Li, Fay and Wu and Fu estimators (Table 2). Likewise, the MK test did not reveal any deviations regarding neutrality (Table 3), however, the dN − dS difference (Table 3) showed that the synonymous substitution rate was higher than the non-synonymous substitution rate (p = 0.000, Fisher’s exact test), suggesting that pvron4 was under purifying selection. The sliding windows led to ω < 1 values being observed throughout the gene (Fig. 1).

Table 2 Neutrality, linkage disequilibrium and recombination tests for the pvron4 gene in the Plasmodium vivax Colombian population
Full size table
Table 3 Difference between dN − dS, KN − KS and the neutral index from MK test
Full size table

When comparing phylogenetically related species, the sliding window gave ω < 1 values, with few regions having ω ≥ 1 (positions 1172–1325, 2955–2975 and in position 2955; Fig. 1). The KN − KS difference (Table 3) gave negative values suggesting that purifying selection has played an important role in this locus' evolutionary history in the genus Plasmodium. On the other hand, four negatively selected sites (28, 112, 149, 643; Fig. 1) were found throughout the gene in P. vivax when calculating selection per codon based on maximum probability methods (SCAL, FEL, IFEL, REL, and FUBAR) and the Bayesian method (MEME), while no sites were found to be under positive selection (Fig. 1). These methods revealed 162 sites under negative selection and 21 under positive selection between species (Fig. 1).

Phylogeny was inferred from orthologous sequences from 14 Plasmodium species. This was used as reference framework for the REL-branch-site test. This method led to finding six branches (lineages) having evidence of episodic selection (Fig. 2). Three were ancestral branches (internal) while the other three (external) had given rise to Plasmodium inui, Plasmodium chabaudi and Plasmodium gaboni. The MEME method revealed codons under diversifying episodic selection.

Fig. 2
figure 2

Lineages subject to episodic diversifying selection in ron4. The branches (lineages) under episodic positive selection were identified by the Branch-site REL method. The figure shows the ω+ values, the percentage of sites (Pr [ω = ω+]) and p values. The shade of each colour on branches indicates strength of selection (red shows ω+ > 15, blue ω− ≤ 1 and grey ω = 1). The size of each colour represents the percentage of sites in the corresponding class. Branches have been classified as undergoing episodic diversifying selection by the p value corrected for multiple testing using the Holm-Bonferroni method at p < 0.05

Full size image

Effective number of codons

Given the high conservation of the pvron4 locus, deviation regarding the effective use of codons was evaluated as a means of selection at translational level. The ENC for pvron4 estimated by ENCprime (N'c = 53, scaled X 2 = 0.103) and DnaSP (ENC = 55.4, scaled X 2 = 0.159) gave values close to 61, with a CBI value of 0.162, suggesting that there was no bias regarding the effective use of codons, thereby ruling out translational selection.

Linkage disequilibrium and recombination

The ZnS estimator was used for evaluating LD between pvron4 polymorphisms, giving 0.15. Linear regression between the LD and nucleotide distance showed a reduction in LD as distance increased, thereby suggesting recombination. However, the ZZ estimator gave −0.0019 and no minimum recombination sites were detected. Likewise, GARD methods found no breakpoints nor did RDP software reveal recombination tracks.

The Plasmodium vivax Colombian population’s genetic structure regarding the pvron4 locus

An AMOVA between the three Colombian regions was calculated for evaluating pvron4 geospatial genetic diversity in Colombia, as well as Wright’s fixation index (FST) between the different populations (regions). AMOVA revealed statistically significant differences between the sub-populations within the regions (FSC = 0.06, p = 0.04). The Nariño sub-population (from the Pacific region) could be responsible for genetic differences regarding each of the other subpopulations (Additional files 3, 4). Calculating the FST index between populations (regions) gave values close to 0. There was a statistically significant difference between the Urabá/lower Cauca/southern Córdoba and Orinoquia-Amazonia regions (Table 4).

Table 4 Inter-population FST statistic for pvron4
Full size table

A median joining network was used for better understanding the evolutionary relationship between pvron4 haplotypes for describing the set of potential mutational pathways giving rise to the 32 haplotypes available for the locus (Fig. 3 and Additional file 3). The network showed that the parasite’s populations shared haplotypes, regardless of geographical region, which were related by different mutational pathways and ancestral sequences (median vectors) (Fig. 3 and Additional file 3). The most frequently occurring haplotypes were H5 (53.1 %), followed by H4 (31.2 %), H13 (25 %), H3 and H20 (12.5 % each). The presence of unique haplotypes in the Nariño (H7, H8 and H10) and Amazonas populations (H21, H22, H2) should be noted as they could be considered rare or specific regional alleles.

Fig. 3
figure 3

Median-joining network for Colombian regions. The Figure shows the pvron4 haplotypes identified from the isolates from the three regions of Colombia. Haplotypes 22 and 28 were included within haplotype 15 using the contraction star algorithm [86] for simplifying interpretation of the network. Each node is a haplotype and its size indicates its frequency. The lines connecting the haplotypes represent the different mutational paths and the median vectors are the ancestral sequences explaining the relationship and evolutionary origin

Full size image

Predicting pvron4 putative domains and antigenic potential

Analysing the PvRON4 sequence from the Sal-I strain revealed the presence of a putative domain for the esterase/lipase (pfam05057) protein superfamily between amino acids 311–425 (nucleotides 931 to 1275, numbers based on the Sal-I reference sequence, GenBank access number: XM_001615228.1) e-value 1.10e-03. This domain was located after the repeat region and was highly conserved (Fig. 1).

According to antigenicity and B-cell linear epitope prediction, there was a potentially antigenic region between positions 41–171 and 178 up to 256 for haplotype 6 and up to position 340 for haplotype 17 (Additional file 5). This agreed with hydrophobicity and solvent accessibility predictions so that the PvRON4 N-terminal region seems to be a potential immune target. By contrast, the central region (following the repeat region) and the C-terminal seemed to be less antigenic, being less solvent-exposed (Additional file 5).

Discussion

Various proteins contained within the parasite’s apical organelles seem to be crucial for host cell invasion and thus represent promising vaccine targets. RON complex proteins are among the proteins localized in the apical organelles, forming part of the TJ [17, 21, 63, 64]. This TJ plays a decisive role in parasite entry to a host cell and closure of the parasitophorous vacuole [17]. The RON4 protein located in the invasion complex is present in Phylum Apicomplexa members [15, 23, 24], suggesting that it forms part of a conserved invasion pathway. This protein has thus been described as a potential vaccine candidate.

A vaccine candidate must have several characteristics [10, 65]; one of them is to have low genetic diversity to avoid allele-specific immune responses, which could reduce vaccine efficacy. The analysis of P. falciparum laboratory strains from different geographically origins showed the pfron4 locus as being a highly conserved locus, having just one amino acid substitution [23]. Plasmodium vivax ron4 seems to have the same pattern, as analysing five reference sequences revealed low genetic diversity [25]. This study thus analysed 73 clinical isolates from the Colombian population for confirming pvron4 as a highly conserved gene in P. vivax. In spite of increasing the number of sequences analysed, pvron4 diversity remained low, the present study’s results showing that pvron4 had lower genetic diversity than in a previous report [25]. Only eight SNPs were identified in the 80 available sequences compared to 14 previously identified ones [25]. Such high number of previously reported SNPs (i.e., 14) was due to erroneous repeat region alignment. pvron4 had a similar pattern to that of other apical organelle proteins [66, 67]. pvrap1 and pvrap2 had 0.0009 to 0.001 π [67] while pvron4 had a much lower value than pvrap1, suggesting it had low genetic diversity, the locus being more conserved to date for P. vivax. As it has been suggested for pvrap1 and pvrap2 [67] the low diversity in pvron4 could be the consequence of functional/structural constraint (see below) due to the key role of this protein in parasite invasion.

Even though pvron4 was a highly conserved sequence regarding SNP occurrence, it had high polymorphism regarding size. Previous studies have identified two types of repeats towards the N-terminal of the encoded protein [22]. These repeats were reported as being imperfect copies of amino acids GGEH/SGEH/S and G/AEH. However, the analysis here performed showed that the pvron4 repeat region consisted of two types of repeats having 100 % identity; the first encoded three GES amino acids (one to three copies) and the second one GEHGEHAEHGE amino acids (one to seven copies). These repeats gave a high number of different haplotypes (alleles) in P. vivax.

Previous studies have suggested that tandem repeats could play an important role as host immune response evasion mechanism [68–71]. In this study, 21 haplotypes were identified in PvRON4 when the InDels were analysed. PvRON4 N-terminal region seems to be the most exposed protein according to solvent availability and hydrophobicity results. This region (between signal peptide and repeat region) seems to be a potential antigenic target due to this being where the largest amount of potential B-cell linear epitopes was predicted. The repeat sequences identified broadened the solvent-exposed region and the protein’s antigenic potential. The PvRON4 N-terminal region could thus be the region exposed to a host’s immune system and repeats could be acting as an immunological smokescreen. Further antigenic and immunogenic studies are needed to confirm such hypothesis. As the repeat region was highly conserved regarding sequence, it could play an important structural or functional role, as has been suggested recently for the CSP [72, 73].

While the N-terminal region might to be exposed to the immune system, the central and C-terminal regions seem to be under functional constraint. Neutrality tests (e.g., Tajima, Fu and Li) gave no statistically significant values and neutrality was thus not ruled out. If pvron4 is under neutrality then it should show high polymorphism unless there is a functional or structural constraint [74]. Given that this locus was highly conserved regarding sequence, functional/structural constraint is probable. However, selection at translational level could also be responsible for high conservation in the pvron4 sequence. Analysing regarding preferential codon use did not reveal bias regarding codon use (ENC = 53–55). In fact this value was similar to that reported for the complete genome (ENC = 52.18) [75], suggesting that high pvron4 locus conservation was not due to selection at translational level and could have been a result of strong purifying selection.

The dS rate was significantly greater than the dN rate according to Fisher’s exact test, suggesting that this locus has evolved under purifying selection. However, it is not easy to evaluate how natural selection acts in highly conserved antigens [66, 76, 77]. Previous studies have compared P. vivax sequences to phylogenetically related species to evaluate the effect of selection on parasite antigens [66, 76–78]. Sliding window analysis of ω rate gave values less than 1 towards the protein’s central region as well as towards the C-terminal. The KS was statistically greater than KN and various sites under purifying selection between species were detected in these regions, suggesting that purifying selection plays an important role during the locus’ evolution in the genus. Bearing in mind that functional regions tend to have slower evolution and are usually conserved between species [79], these results suggest that the PvRON4 C-terminal and central regions could be functionally important. The presence of conserved cysteines in the C-terminal portion (usually associated with protein–protein interaction) could be mediating the interaction between RON4 and AMA-1 and/or other RONs [16, 80], while the presence of a putative esterase/lipase domain in the protein’s central region could be involved in RON4 entry to the host cell.

Plasmodium falciparum and T. gondii studies have shown that the RON4 C-terminal region seems to play an important role in invasion [16, 24]. RON4 is located inside red blood cells (RBC), anchoring the AMA1/RON2 complex [17, 18, 24]; RON4 must thus be secreted and enter RBC during initial invasion stages by a yet-unknown mechanism. The presence of an esterase/lipase domain in the PvRON4 central region could provide a clue regarding the action mechanism. This is one of the protein’s most structured regions, being highly conserved among species and containing several sites under purifying selection, suggesting a functional/structural role. Therefore, while the PvRON4 N-terminal region seems to be associated with evasion of the immune response, the central region (containing the esterase/lipase domain) could be associated with the rupture of ester bonds in the phospholipids constituting host cell membrane. Such rupture would enable RON4 entry to RBC or hepatocyte cytoplasm. Once inside, the RON4 C-terminal region anchors RON proteins, which, in turn, enable AMA1-mediated interaction between the parasite and host cells. It can thus be hypothesized that such putative esterase/lipase domain could play a role regarding RON4 entry to a host cell, however, further functional assays are needed to confirm this.

In spite of ron4 being highly conserved between species and that purifying selection seems to be important during this locus’ evolution in Plasmodium, some sites under positive selection were identified, coinciding with the regions where ω > 1 was observed. Previous studies have shown that some antigens (regardless of their genetic diversity) have regions/codons under episodic positive selection, which could have enabled adaptation to different hosts [76, 81, 82]. The topology obtained for ron4 was similar to that obtained when analysing mitochondrial DNA [83]. The phylogenetic relationships of species infecting rodents and hominids can be seen in ron4 phylogeny, however, such relationships have not been seen for species infecting monkeys. These species have a complex evolutionary history, which includes biogeographic aspects, adaptation to new macaque hosts and even a change from monkeys to humans [81, 84]. The episodic selection observed here might thus have been a consequence of this group of parasites’ rapid diversification (in the N-terminal region and a small portion of the C-terminal region) thereby enabling RON4 to adapt from an ancestral population to new available hosts, as previously suggested [76, 81, 82, 84].

A relatively high number of haplotypes has been found in the pvron4 locus in Colombia, resulting from a combination of SNPs and tandem repeats. AMOVA analysis and median joining showed that Colombian regions shared most haplotypes and seemed to be genetically similar. However, it was observed that a 6 % of estimated variation between these regions was due to differences between the subpopulations constituting them. The FST value showed that some subpopulations might not be genetically similar; this could be associated with the presence of unique haplotypes. This agreed with studies in Colombia involving other parasite antigens [77], as well as mitochondrial DNA studies in America [85], suggesting that the parasite population in America is structured and has limited gene flow. However, since some subpopulations analysed here had limited sample size, the number of sequences must be increased for such results to be confirmed.

Conclusions

Designing a vaccine which is completely effective against the parasites causing malaria requires antigens having limited genetic diversity to avoid allele-specific immune responses. The pvron4 locus was seen to have low genetic diversity regarding SNPs but had a large amount of haplotypes due to tandem repeats located in the proteins’ N-terminal, which could be involved in evading the immune response. On the other hand, the central and C-terminal regions are highly conserved, even between species. Such regions are under purifying selection, suggesting that they are under functionally or structurally constraint. The central region has a putative esterase/lipase domain, leading to the hypothesis that this domain enables RON4 entry to host cells while the C-terminal region anchors the AMA1/RON complex. Bearing the aforementioned results in mind, PvRON4 central/C-terminal region would seem to be a promising candidate for inclusion when designing a subunit-based vaccine against P. vivax.