Real-time PCR detection of mixed Plasmodium ovale curtisi and wallikeri species infections in human and mosquito hosts

Plasmodium ovale curtisi (Poc) and Plasmodium ovale wallikeri (Pow) represent distinct non-recombining malaria species that are increasing in prevalence in sub-Saharan Africa. Though they circulate sympatrically, co-infection within human and mosquito hosts has rarely been described. Separate 18S rRNA real-time PCR assays that detect Poc and Pow were modified to allow species determination in parallel under identical cycling conditions. The lower limit of detection was 0.6 plasmid copies/μL (95% CI 0.4-1.6) for Poc and 4.5 plasmid copies/μL (95% CI( 2.7- 18) for Pow, or 0.1 and 0.8 parasites/μL, respectively, assuming 6 copies of 18s rRNA per genome. However, the assays showed cross-reactivity at concentrations greater than 103 plasmid copies/μL (roughly 200 parasites/μL). Mock mixtures were used to establish criteria for classifying mixed Poc/Pow infections that prevented false-positive detection while maintaining sensitive detection of the minority ovale species down to 10° copies/μL (<1 parasite/μL). When the modified real-time PCR assays were applied to field-collected blood samples from Tanzania and Cameroon, species identification by real-time PCR was concordant with nested PCR, but additionally detected two mixed Poc/Pow infections where nested PCR detected a single Po species. When real-time PCR was applied to 14 oocyst-positive Anopheles midguts saved from mosquitoes fed on P. ovate-infected persons, mixed Poc/Pow infections were detected in 11 (79%). Based on these results, 8/9 P. ovate carriers transmitted both P. ovate species to mosquitoes, though both Po species could only be detected in the blood of two carriers. The described real-time PCR approach can be used to identify the natural occurrence of mixed Poc/Pow infections in human and mosquito hosts and reveals that such co-infections and co-transmission are likely more common than appreciated.

blood-fed mosquitoes.This suggests that Poc and Pow co-infect the same hosts more frequently than 50 previously realized.

INTRODUCTION
Plasmodium ovale curtisi (Poc) and Plasmodium ovale wallikeri (Pow), long recognized simply as Plasmodium ovale due to their similar morphology under the microscope, in fact represent two distinct, non-recombining malaria species [1,2].P. ovale (Po, inclusive of both species) has more commonly been reported from West Africa [3], but its prevalence based on recent PCR surveys appears to be increasing in East Africa [4][5][6][7][8].There is some evidence that the two species differ in their latency period or relapse periodicity, as well as their presentation in travelers [3,[9][10][11][12].Still, the extent to which Poc and Pow differ in their biology, epidemiology, and clinical manifestations in Africa, where they are most frequently found, remains an active area of research [13,14].Interestingly, though both P. ovale species circulate sympatrically in time and space, mixed infections of the two species have rarely been described [3,13,15,16].Since the existence of mixed Plasmodium infections of other species is well-documented, we suspected this was due to technical limitations rather than biological constraints.We adapted previously published nested and real-time PCR (qPCR) assays for Po species detection to establish criteria for defining when both Poc and Pow are present, then applied these assays to Po-positive isolates from Africa.Our optimized protocol for running parallel Poc and Pow species-specific qPCRs detected mixed Po species infection in a smaller proportion of blood samples from Tanzania and Cameroon.However, this percentage was much higher in Po-positive mosquito midgut samples derived from mosquito feeding studies in Tanzania, raising the possibility that mixed Poc/Pow infections may be more common than currently appreciated.

METHODS
Previously published molecular assays for the detection of P. ovale species were reviewed (Table 1).Based on initial testing, a nested PCR (nPCR) assay from Calderaro, et.al [17] and real-time qPCR assays from Perandin, et al. [18] and Calderaro, et.al [19], all targeting the small subunit RNA gene (18S rRNA), were selected for further assay development and comparison.Samples that were successfully sequenced were of higher parasitemia (median density 23 vs 9 parasites/uL, p = 0.007)

Modifications to the nested PCR assay
The nPCR assay published by Calderaro, et al. [1] was performed as described using HotStarTaq (Qiagen).A Poc 18S rRNA plasmid (MRA-180, BEI Resources) and genomic DNA from a sequence-confirmed Pow clinical sample from Tanzania were used as positive controls.When PCR products were not visualized on the 1% agarose gel using the standard conditions, the nPCR was repeated with adjustments made to cycle number, annealing temperatures, and input DNA volume to increase yield.The number of cycles was increased from 35 cycles in both rounds (70 total cycles) to as high as 40 cycles in round 1 and 45 cycles in round 2 (85 total cycles).Annealing temperatures were dropped in the second round species-specific PCR from 60℃ and 58℃ for Poc and Pow, respectively, to as low as 58℃ and 56℃, respectively.Finally, the input DNA volume was increased from 5 μL up to 10 μL for the round 1 PCR, but maintained at 5 μL for the round 2 PCR.A subset of nPCR products underwent Sanger sequencing for confirmation of species designations.

Modifications to real-time PCR assays
The sensitivity of the real-time PCR assays designed by Perandin, et.al [2] and Calderaro, et.al [3] to detect Poc and Pow was tested using dilutions of plasmids containing the small subunit ribosomal RNA gene (18S rRNA) specific to each species.The Poc 18S rRNA plasmid was obtained from BEI resources (MRA-180; GenBank: AF145337, 1,100 bp insert).The Pow plasmid control was created by cloning the first round nested PCR product of a Sanger-sequenced Pow clinical sample from Tanzania (MqTZ-0123) into a Topo 2.1kb vector using One Shot Top10 competent cells (Thermofisher Scientific).Real-time PCR was carried using FastStart Universal Probe Master mix (ROX,Roche) and published primer concentrations.
To maximize detection sensitivity of both species, both assays were run in parallel to 50 cycles, instead of the originally published 45 and 55 cycles.
A common annealing temperature of 52.8℃ was chosen for yielding similar Ct thresholds for detection of the same plasmid copy concentrations.At this lower annealing temperature, the assays could not be multiplexed into a duplex assay without detecting both species in each run regardless of the species-specific plasmid used, likely owing to the two real-time PCR assays having identical reverse primers and forward primers that differ by only two nucleotides, in addition to probes that also differ by only two nucleotides.Thus, species-specific assays were run side-by-side in separate reactions under the same conditions.

Real-time PCR of mono-and mixed species samples
The sensitivity of the modified qPCR assays for their respective 18S targets was tested for plasmid concentrations ranging from 10 5 down to 10 -1 plasmid copies/L.A limit of detection was calculated for each assay using a Probit analysis [24].To determine how the assays would perform for detecting samples with both species present, mock plasmid DNA control mixtures were created with Poc and Pow concentrations at a lower range (10 2 to 10 -1 plasmid copies/L) in ratios of 1:1, 1:2, 1:5, and 1:10 to simulate clinical samples.Each mixture was run 10 times using the qPCR assays, with the mean Ct value from the 10 runs reported.Based on these data, a classification scheme was developed for identifying mixed species infections (Poc and Pow).

P. ovale mixed species detection in clinical blood and mosquito samples
DNA extracted from blood samples from Cameroon and Tanzania previously identified as P. ovale-positive based on an 18S qPCR that detects both P. ovale species [23] were used to compare the performance of the selected nPCR and modified qPCR assays.The proportion of samples that amplified in each assay and assay concordance for P. ovale species identification were examined, including whether there was detection of mixed species infection.Additionally, real-time PCR detection of P. ovale species was performed on DNA extracted from P. ovale-positive mosquito midgut samples.These were obtained from mosquito feeding assays performed on P. ovale-carriers in Tanzania using colony-reared Anopheles gambiae IFAKARA strain [8], including both direct skin feeding and membrane feeding assays.Mosquito midguts that were dissected and oocyst-positive by microscopy at day 8 post-blood feeding were stored in either ethanol or DNA/RNA shield (Zymo Research) then subjected to DNAzol-based DNA extraction (Invitrogen).Extracted midgut DNA was amplified using Plasmodium genus-specific primers in a conventional PCR as a first round reaction [25].A second round P. ovale qPCR was performed on a 1:50 dilution of the first round product.Samples found to be positive were selected for further P. ovale species identification.Dried blood spot or leukodepleted blood samples obtained at the time of mosquito feeding were also subjected to realtime PCR species detection to compare the presence and type of P. ovale species detected in the blood vs. mosquito midgut samples.

Real-time PCR detection of P. ovale curtisi and wallikeri in mono-and mixed infections
Run individually, both modified real-time PCR assays consistently detected down to 10 0 18S plasmid copies/L of their respective P. ovale species, or the equivalent of 1-2 parasites/L assuming 5-8 copies of 18S rRNA per genome (Table 2).Since distilled water negative controls also sometimes demonstrated late cycle amplification for the Poc assay (mean Ct = 49.4,range 49.2-49.9),Ct thresholds for positivity were set at Ct <49 for both species.Using these Ct thresholds, the limit of detection of the Poc assay based on Probit analysis was slightly lower than that of the Pow assay (0.6 plasmid copies/L (95% CI 0.4-1.6) vs. 4.5 plasmid copies/L (95% CI 2.7-17.7)for Poc and Pow, respectively).At higher concentrations of Poc and Pow plasmids, on the order of 10 3 copies/L, cross-reactivity between species was observed (Table 2).The cross-reactivity between assays at higher parasite densities (10 3 copies/L or >100-200 parasites/L) could lead to misidentification of secondary species when none is present.The opposite may occur at lower parasite densities.To understand how the qPCR assays would perform for detecting mixed P. ovale species infections at lower densities, mock plasmid mixtures were created to mimic different Poc and Pow ratios of species at different concentrations (Table 3).Both species remained detectable in the 1:1 mixtures down to their previous limits of detection.In fact, the sensitivity of the Pow assay appears "boosted" in the presence of Poc at the lowest concentration (10 -1 copies/L), likely due to low-level cross-species reactivity though none was observed at this level in the mono-species reactions.In the 1:2 and 1:5 mixtures, detection of the minor species was preserved when using a Ct threshold for positivity of <49.However, detection sensitivity of Pow as the minor species was compromised once the ratio reached 1:10.Table 3. Performance of species-specific real-time PCR detection in mock mixtures of P. ovale curtisi and wallikeri 18S plasmid controls.Mixtures were created either in equal proportions (A) or at ratios of 1:2, 1:5, and 1:10 (B).The number of positive runs and mean Ct of the positive qPCR runs for each species-specific assay is shown for 18S plasmid concentrations ranging from 10 2 to 10 -1 copies/L.A Ct value <49 was used to determine a positive run.Based on the performance of the qPCR assays in mono-and mixed infections, an algorithm was developed to prevent false-positive detection of a secondary species within a mixed Poc/Pow infection.This is more likely to happen if the majority species is abundant (Ct <39), leading to a cross-reactive false positive result for the other species that is indistinguishable from a small concentration of the minor species (Ct >44) (Table 2).The proposed classification system (Figure 1) does not indicate the relative abundance of the two species, but simply whether both are present.This approach maintains 100% specificity for both species, 97% sensitivity (203/210) for Poc, and 92% sensitivity (194/210) for Pow when applied to all mixtures down to 10 -1 L concentration.

Performance of real-time PCR in clinical samples
Thirty-seven clinical blood samples from Cameroon and Tanzania that previously tested positive for P. ovale using a pan-ovale species 18S rRNA real-time PCR [23] were used to compare the performance of a published nested PCR [17] and the adapted real-time PCR (qPCR) classification algorithm.Results were successfully obtained for 70% (26/37) of samples using qPCR versus 35% (13/37) of samples using published nested PCR conditions.When nested PCR was repeated with a greater number of cycles (up to 85 cycles across two rounds) and/or increased template DNA volume (up to 10 L from 5 L), 85% (23/27) could be successfully amplified (Figure 2, Table 4).Identified species was confirmed by Sanger sequencing of nPCR products obtained from leukodepleted blood (LDB) samples.4.
The second round products of the nested PCR assay were run on a 1% agarose gel for Poc (top) and Pow (bottom) detection.Fragment lengths of 787-789 bp and 782 bp were expected for Poc and Pow, respectively.The results of the nested PCR assay on these human clinical blood samples, determined by the presence of a band on the depicted gel, are displayed in Table 4. Good concordance was observed among the 19 samples successfully amplified in both nested PCR and real-time PCR assays.There was 100% agreement with regard to identification of the major P. ovale species present.However, real-time PCR additionally detected the presence of a second minor species in 2/20 samples that was not identified by nPCR (Table 4).Overall, among 28 unique P. ovale-infected individuals with blood samples with species determined by real-time PCR or nested PCR listed in Table 4, 54% (15/28) were infected with Poc, 36% (10/28) were infected with Pow, and 11% (3/28) were identified as harboring mixed Poc/Pow species infections.13 Mosquito-based xenodiagnosis revealed a much higher prevalence of mixed Poc/Pow infections than that identified from human clinical blood samples.Seventeen oocyst-positive mosquito midguts obtained from mosquito feeds performed on 9 P. ovale infected persons in Tanzania, that had previously tested qPCRpositive for P. ovale, were available for species analysis.Fourteen of 17 (82%) midgut DNA samples successfully amplified in the P. ovale species qPCR assays, of which 11/14 (79%) were positive for both Poc and Pow using the classification scheme outlined in Figure 1 (Table 5).Of 9 P. ovale carriers, only 2 had mixed P. ovale infections that could be detected in the blood at the time of mosquito feeding, but mosquitobased xenodiagnosis revealed that all but one (8/9) harbored both Poc and Pow and transmitted both species to mosquitoes.(Table 5).[3,13,15,16,20,23].By modifying Poc-and Pow-specific 18S rRNA real-time PCR assays and developing a classification algorithm to detect mixed Poc/Pow infections that avoids false-positive detection due to cross-reactivity, we show that mixed Poc/Pow infections occur naturally (~11% in our initial blood survey) and may be much more common than anticipated (89% by mosquito-based xenodiagnosis in our small sample).Given their sympatric distribution, co-transmission of both Poc and Pow species within the same Anopheles mosquitoes is not unexpected.Frequent co-transmission means that the two species have ample opportunity to recombine within mosquitoes.Yet a species barrier appears to be firmly established, likely due to prior distinct evolutionary pathways before their present co-existence in human hosts [1,2,9].The higher frequency of Poc/Pow mixed infections we discovered among Po-infected mosquitoes compared to blood infection was a surprise.Xenodiagnosis has previously been suggested to be the most sensitive method for detecting human blood-stage infection, often detecting subpatent infections [28][29][30] and, in our experience, sometimes detecting parasitemias circulating just below the limit of detection of PCR [31].Genetic diversity undetected by blood sampling can be revealed through mosquito sampling [32][33][34][35] and attests to the sampling efficiency of mosquitoes and transmission efficiency of gametocytes.While our results could be explained by a simultaneous outbreak of both Po species, the mosquito feeding assays depicted in Table 5 spanned two years of data collection.Rather, those exposed to one Po species may be more likely to also be exposed to the other Po species, with malaria exposure concentrated in a small proportion of the population who then serve as a reservoir [37].The role of relapse, in which persons once exposed remain latently infected in the liver, could also contribute to unexpected high rates of co-infection.Po species inoculated separately could relapse in unison when conditions are right.Larger molecular studies in other settings and including wildcaught mosquitoes are needed to verify our findings.This work is not without limitations.First, our experiments to determine analytical sensitivity and develop a species classification scheme used plasmid controls, and we did not specifically test their robustness within blood or mosquito samples.Our finding that co-infection by both species was more common in mosquitoes than in human blood samples was surprising.However, we expect lower assay sensitivity in mosquito samples due to low parasite burdens and to PCR inhibition rather than enhanced cross-reactivity or compromised diagnostic specificity.Second, our classification scheme is expected to underestimate mixed Poc/Pow infections when the majority species is at high density.Third, given the slightly greater sensitivity of the Poc real-time PCR assay over the Pow assay, we cannot draw firm conclusions about the relative prevalence of Poc and Pow in our surveys, aside from finding that both were represented in blood and mosquito samples.Fourth, though we showed excellent concordance of the real-time PCR assays with nested PCR and Sanger sequencing, we did not attempt to sequence our predominantly-mixed midgut samples to verify the presence of both species due to the bioinformatic challenges of detecting mixed Poc/Pow infections by targeted sequencing [23].Finally, some of our midgut PCR results involved high Ct values and might reflect contamination of samples, but the three-year time span of sample collection and the detection of Poc single species midguts (as well as midguts that did not amplify) make this less likely.
In conclusion, the real-time PCR approach described here represents an efficient method for detecting mixed Poc/Pow infections in both human clinical blood samples and mosquito midguts.Mixed Poc/Pow infections were commonly detected in mosquito midguts, and were also detected, albeit to a lesser degree, in both human dried blood spots and leukocyte-depleted blood samples.This suggests that the extent of mixed Poc/Pow infection may be greater than previously appreciated.Issues with cross-reactivity remain with the real-time PCR assays, which would best be resolved by using separate species-specific gene targets that would allow development of primer and probe sets with no potential for cross-reactivity [38].Much remains to be learned about Poc and Pow epidemiology in sub-Saharan Africa, including how they may be evolving in the face of malaria control efforts designed to target P. falciparum.Recognition of the limitation of current assays for detecting co-infection and continued development of better, more facile diagnostics will improve our ability to understand whether and how these two species potentially differ in their epidemiology, biology, and clinical manifestations.Our findings suggest that the degree to which these closely related but sympatric species cocirculate within their human and mosquito hosts may be underappreciated.4.

Figure 1 .
Algorithm to determine the species of P. ovale malaria using 18S real-time PCR, whether as a single or mixed species infection.

Figure 2 .
Gel electrophoresis showing nested PCR results for select samples from Table

Figure 1 .
Figure1.Algorithm to determine the species of P. ovale malaria using 18S real-time PCR, whether as a single or mixed species infection.Figure 2. Gel electrophoresis showing nested PCR results for select samples from Table4.The second round products of the nested PCR assay were run on a 1% agarose gel for Poc (top) and Pow (bottom) detection.Base pair sizes of 787-789 and 782 were expected for Poc and Pow, respectively.The results of the nested PCR assay on these human clinical blood samples, determined by the presence of a band on the depicted gel, are displayed in Table4.

Figure 2 .
Figure1.Algorithm to determine the species of P. ovale malaria using 18S real-time PCR, whether as a single or mixed species infection.Figure 2. Gel electrophoresis showing nested PCR results for select samples from Table4.The second round products of the nested PCR assay were run on a 1% agarose gel for Poc (top) and Pow (bottom) detection.Base pair sizes of 787-789 and 782 were expected for Poc and Pow, respectively.The results of the nested PCR assay on these human clinical blood samples, determined by the presence of a band on the depicted gel, are displayed in Table4.

Table 2 . Limit of detection and cross-reactivity of real-time PCR assays targeting P. ovale curtisi and wallikeri.
The mean Ct of the positive qPCR runs for each assay is shown for 18S plasmid concentrations ranging from 10 5 to 10 -1 copies/L.

Table 4 . Comparison of P. ovale species nested PCR and real-time PCR assays in field samples from Cameroon and Tanzania.
Poc = P.o.curtisi; Pow = P.o.wallikeri.Samples not successfully amplified in species assays are indicated with -.
a Nested PCR required increase in amplification cycles and/or increased volume of template DNA b Confirmed by Sanger sequencing of nPCR product Note: Five samples did not amplify in either the nested PCR or real-time PCR assay.