Multiplex, DNase-free one-step reverse transcription PCR for Plasmodium 18S rRNA and spliced gametocyte-specific mRNAs

Background Plasmodium gametocytes are sexual stages transmitted to female Anopheles mosquitoes. While Plasmodium parasites can be differentiated microscopically on Giemsa-stained blood smears, molecular methods are increasingly used because of their increased sensitivity. Molecular detection of gametocytes requires methods that discriminate between asexual and sexual stage parasites. Commonly tested gametocyte-specific mRNAs are pfs25 and pfs230 detected by reverse transcription polymerase chain reaction (RT-PCR). However, detection of these unspliced mRNA targets requires preceding DNase treatment of nucleic acids to eliminate co-purified genomic DNA. If gametocyte-specific, spliced mRNAs could be identified, DNase treatment could be eliminated and one-step multiplexed molecular methods utilized. Results Expression data was used to identify highly-expressed mRNAs in mature gametocytes that were also low in antisense RNA expression in non-gametocyte stages. After testing numerous candidate mRNAs, the spliced female Pf3D7_0630000 mRNA was selected as a Plasmodium falciparum gametocyte-specific biomarker compatible with Plasmodium 18S rRNA RT-PCR. This mRNA was only detected in samples containing mature gametocytes and was absent in those containing only asexual stage parasites or uninfected human blood. PF3D7_0630000 RT-PCR detected gametocytes across a wide range of parasite densities in both spiked and clinical samples and agreed with pfs25 RT-PCR, the gold standard for RT-PCR-based gametocyte detection. PF3D7_0630000 multiplexed with Plasmodium 18S rRNA RT-PCR was more sensitive than other spliced mRNA targets for one-step RT-PCR gametocyte detection. Conclusions Because the spliced target does not require DNase treatment, the PF3D7_0630000 assay can be multiplexed with Plasmodium 18S rRNA for direct one-step detection of gametocytes from whole human blood. Electronic supplementary material The online version of this article (doi:10.1186/s12936-017-1863-3) contains supplementary material, which is available to authorized users.

detection strategies should ideally be able to achieve this level of analytical sensitivity.
Gametocytes can be identified by light microscopy of Giemsa-stained thick and thin blood smears but, like all microscopic methods for Plasmodium parasites, only to a density of ~5000-20,000 parasites/mL (5-20/μL) by thick blood smear [10]. Molecular methods are more sensitive and include qualitative or quantitative RT-PCR [3,[11][12][13][14] and NASBA [9,15,16]. mRNA-based methods are used to differentiate gametocytes from asexual stages by detecting stage-specific mRNAs. The most common gametocyte mRNA targets are pfs25 [17,18] and pfs230 [18][19][20] for P. falciparum and pvs25 for P. vivax [17]. These well-studied targets are all derived from unspliced mRNAs, so a DNase treatment step is required to destroy genomic DNA prior to RT. When manual DNase treatment is performed, there is partial loss of sample material and increased risk for sample cross-contamination due to added handling steps. DNase treatment also makes the process more time consuming. For detection of asexual stage parasites, some laboratories already perform one-step multiplex Plasmodium 18S rRNA RT-PCR directly from extracted whole blood without DNase treatment [21,22]. In these assays, DNase treatment is not required because Plasmodium 18S rRNAs are more than three orders of magnitude more abundant than the coding 18S rDNA genes [21,22]. Plasmodium 18S rRNAs are developmentally regulated between sexual and asexual stages [23][24][25] but gametocytes express both S-(sexual) and A-(asexual)-type 18S rRNAs [18]. Because of this expression, 18S rRNAs alone cannot be used to differentiate gametocytes from asexual stage parasites. Spliced mRNAs that were highly expressed in gametocytes and showed nearly absent antisense RNA expression in the asexual stage were hypothesized to be ideal targets for multiplexing with the Plasmodium 18S rRNA assay for one-step RT-PCR Plasmodium detection. Although three spliced gametocyteexpressed mRNAs have been reported as RT-PCR targets [26,27], it was unknown whether these targets were suitable for multiplexing with the 18S rRNA assay.
Here, bioinformatic methods were used to search for spliced, gametocyte-specific mRNA targets compatible with one-step Plasmodium 18S rRNA RT-PCR methods. After testing candidate targets and identifying several gametocyte-specific spliced mRNAs, the PF3D7_0630000 mRNA was selected as a P. falciparum-specific spliced mRNA for RT-PCR without DNase treatment. This novel mRNA target was then tested against pfs25 and a known spliced mRNA using clinical samples.

Methods
Bioinformatics mRNA expression fold-change from Lopez-Barragan et al. [28] was evaluated. Fold-change data was summed for asexual or gametocyte stages, sorted and filtered to remove mRNAs with total asexual expression ≥0 or total gametocyte expression ≤0 (all relative to ring stage expression). Single exon genes were eliminated as were those lacking data in http://www.plasmodb.org and those representing redundant mRNA isoforms. Fold change data in [28] were based on RNASeq reads per kilobase of exon per million (RPKM). Since RPKM can originate from sense or antisense transcripts, the strand-specific fragments per kilobase of exon per million (FPKM) [28] were also inspected, and genes with schizont antisense FPKM equal to 50-100% of maximum were eliminated as were those with minimum asexual antisense FPKM within twofold of the maximum gametocyte antisense FPKM values. Remaining genes were sorted on stage V gametocyte expression, and the most highly expressed genes were evaluated for suitable intron-spanning splice RT-PCR designs. RPKM and FPKM data deposited in plasmodb.org from [28] for selected genes is in Additional file 1.

Plasmodium falciparum cultures
Asexual stage Plasmodium parasites were cultured as described [21]. Gametocytes were cultured as described [29]. Purity of cultures was confirmed by microscopy. Some asexual stage cultures were further confirmed gametocyte-free by pfs25 RT-PCR.

Human clinical samples
Leftover clinical whole blood specimens (50 μL) were preserved in 2 mL of NucliSens Lysis Buffer (bioMérieux, Marcy-l'Étoile, France). These samples were used under a protocol approved by the University of Washington Institutional Review Board (IRB) (protocol 47026, S. Murphy). Samples were included from a clinical trial of a novel chemical entity DSM265 approved under Fred Hutchinson Cancer Research Center IRB (protocol 8408, J. Kublin). Samples from a field study in Uganda were collected from consenting study participants or with the assent of children and consenting legal care givers within the context of studies approved by the relevant IRBs. Specimens from Nagongera, Tororo District, Uganda were collected under a study approved by the University of California San Francisco IRB 11-05995 and Uganda IRB 2011-0167. In Uganda, EDTA whole blood samples were collected from a previously described surveillance cohort [30]; all participants provided informed consent. Samples from Uganda were collected from May to October 2015.

Nucleic acid extraction
Nucleic acids were extracted on a bioMérieux EasyMAG [21] or an Abbott m2000sp as described [22]. Briefly, on the bioMérieux instrument, 1 mL of the 2.05 mL lysed sample (50 µL whole blood plus 2 mL NucliSENS lysis buffer) was extracted for total nucleic acids and eluted into 40 µL of elution buffer. On the Abbott m2000sp instrument, 1 mL of the 2.05 mL lysed sample was extracted preferentially for RNA and eluted into 53 μL of elution buffer.

Statistics
To evaluate unlysed sample stability, an unpaired t test was used to compare mean C T values for 18S rRNA and gametocyte-specific markers and the difference in these C T values. For sensitivity analyses, exact confidence intervals were calculated [32]. Statistical significance for both was P < 0.05.

Target selection
One-thousand one-hundred twenty-nine genes evaluated previously by RNAseq [28] were filtered to remove those with increased asexual expression (sum of asexual fold change ≥0) and decreased gametocyte expression (sum of stage II and V gametocyte fold change ≤0). This resulted in 372 genes with gametocyte stage II and/or V expression and an absence of asexual-stage expression. The most well-known gametocyte target pfs25 (PF3D7_1031000) was one of the most highly expressed gametocyte-specific genes in this set (Additional file 1). Six genes were eliminated because they lacked data in plasmodb.org and all single exon genes were also eliminated. Two-hundred genes contained two or more exons. Since antisense transcripts are produced for many Plasmodium genes [28] and since such transcripts can lead to false positive results in primer-specific RT-PCRs [33][34][35][36], antisense strand-specific FPKM data [28] were evaluated next. Genes with low or absent antisense transcript expression in non-gametocyte stages were retained (48 genes), sorted on stage V gametocyte expression and evaluated for intron-spanning primer/probe designs compatible with the Plasmodium 18S rRNA RT-PCR cycling temperatures. RT-PCR designs for 14 genes were ultimately selected using this strategy as were designs for four additional genes overexpressed in stage II/V gametocytes but not identified by published fold-change data in [28]; published RT-PCRs for pfs25 and three reported multi-exon gametocyte-specific genes [26,27] were also included (Additional file 1). Based on recently published data [37], two genes were male-specific (PF3D7_1477700, PF3D7_1438800), one showed mixed expression (PF3D7_1020100) and the rest were female-specific (Additional file 1).

RT-PCR with LCGreen intercalating dye for candidate targets
LCGreen RT-PCR was performed on 27 target regions of the 18 novel genes of interest using total nucleic acids obtained from either cultured mature P. falciparum gametocytes or from cultured synchronized ringstage asexual parasites (Additional file 2). Although these RT-PCR assays do not provide absolute quantification, the difference (Δ) between C T values for the 18S rRNA and the mRNA of interest provide an estimate of spliced mRNA abundance. In cultured samples enriched for mature gametocytes, gametocyte-specific spliced mRNAs were detected 6.9-25.2 cycles later than Plasmodium 18S rRNA (ΔC T Gam mRNA-18SrRNA ). In cultured samples containing only asexual-stage parasites, the ΔC T iRBC mRNA-18SrRNA was typically larger than the aforementioned ΔC T Gam mRNA-18SrRNA for each candidate mRNA, as expected for gametocyte-expressed mRNAs (Additional file 2). mRNAs with the smallest ΔC T Gam mRNA-18SrRNA and the largest ΔC T iRBC mRNA-18SrRNA were considered to represent the best balance between high and specific gametocyte expression. Gel electrophoresis was used to evaluate some RT-PCR products. For example, for PF3D7_0630000, a band of the expected size (141 bp) was only generated in the presence of gametocytes by RT-PCR (Fig. 1a). The probe for this target was designed to span an intron (Fig. 1b).

PF3D7_0630000-specific qRT-PCR
Nucleic acids from a dilution series of mature gametocytes (Day 14 of culture) were tested using RT-PCRs for Plasmodium 18S rRNA, PF3D7_0630000 and pfs25 to generate a standard curve and evaluate assay efficiency and linearity Pf3D7_1214500 (S1)  Table 2). Plasmodium 18S rRNA qRT-PCR was able to quantitatively detect as few as 30 parasites/mL of whole blood, consistent with the assay limit of detection of 20 parasites/mL; the C T curve generated for the most dilute sample was below the assay limit of quantification.

PF3D7_0630000 mRNA stability
To assess spliced mRNA target stability, intact gametocytes in whole blood were either preserved in lysis buffer immediately or held at room temperature for 3-7 days before processing into lysis buffer. For all samples (nominal densities 3 × 10 2 -3 × 10 5 gametocytes/mL), mRNA (pfs25 and PF3D7_0630000) degradation was observed at 3 and 7 days of storage, whereas statistically significant 18S rRNA degradation was not observed until 7 days of storage (Fig. 4). PF3D7_0630000 degraded more rapidly than pfs25 at 3 days but this difference was not statistically different at 7 days. These data suggest, while samples for 18S rRNA testing can be stored for several days prior to sample processing, samples for gametocyte-specific pfs25 or PF3D7_0630000 mRNA testing should be processed immediately following collection.

Kinetics of PF3D7_0630000 mRNA expression
To study the time course of PF3D7_0630000 expression, samples were taken from gametocyte cultures during Days 6-14 of culture and compared to expression of the Day 6 sample. This approach was used because on the sixth day of culture, a mix of residual asexual stages and immature gametocytes present, but after the tenth day of culture, most of the parasites were mature gametocytes with ring, trophozoite and schizont stages no longer observed. Unlike asexual-stage parasites that do not express the PF3D7_0630000 spliced mRNA, gametocytes express both pfs25 and PF3D7_630000 mRNAs from by the sixth day of culture. Expression of both targets increased as the cultures matured with pfs25 expression exceeding that of PF3D7_0630000 (Fig. 5). At peak expression (day 14), pfs25 was increased by >2000-fold compared to day 6 of culture while PF3D7_0630000 increased >25-fold.

Detection of gametocytaemia in a volunteer in a controlled human malaria infection trial
To determine if the novel spliced gametocyte-specific mRNA RT-PCR could detect low level gametocytaemia from clinical specimens, samples were first evaluated from a clinical trial in a Seattle-based controlled human malaria infection trial evaluating the prophylactic activity of the drug DSM265. This placebo-controlled prophylactic drug trial was blinded at the time of this manuscript, and it was thus unknown whether subjects received the active compound or the placebo control. Subjects were challenged with 3200 PfSPZ (Sanaria, Inc., Rockville, MD) on Day 0. Final details of the ongoing DSM265 study will be published in a separate manuscript at a later date and do not affect the data described below. Plasmodium infection status was monitored by daily 18S rRNA RT-PCR. One subject developed a late onset infection with 18S rRNA qRT-PCR positivity starting on Day 23 (Fig. 6). The subject was symptomatic (grade 1-2 nausea, vomiting, headache) during this time and was thick blood smear positive on Day 25. The subject was treated on Day 25 with atovaquoneproguanil with rapid conversion to negative blood smears and resolution of symptoms. Plasmodium 18S rRNA concentrations dropped following treatment but did not reach undetectable levels. On Day 33 post-challenge, resurgence in Plasmodium 18S rRNA was noted although the subject was asymptomatic. Persistent asymptomatic positivity for Plasmodium 18S rRNA continued from days 33-42 postchallenge. Neither pfs25 nor PF3D7_0630000 mRNAs were detected during the symptomatic period but both were detected during the second wave of asymptomatic parasitaemia (Fig. 6). The patient was subsequently treated orally with 45 mg primaquine and was negative for all biomarkers at 56 days post-challenge.
The PF3D7_1477700 and PF3D7_1438800 markers [27] could not be successfully multiplexed with the pan-Plasmodium 18S rRNA RT-PCR because false positive results were obtained for asexual samples at annealing temperatures of 50 °C (standard for 18S rRNA RT-PCR) and 60 °C (as originally reported for these gametocyte markers Dilution series of gametocytes tested by RT-PCR for 18S rRNA, Pf3D7_0630000 and pfs25. a RT-PCR was performed with hydrolysis probes on nucleic acids from enriched gametocyte cultures diluted into whole human blood across the range of parasite densities indicated in the figure (3 × 10 7 gametocyte/mL blood to 3 gametocytes/mL blood). Samples for 18S rRNA and PF3D7_0630000 were not treated with DNase; samples for pfs25 RT-PCR were DNase treated as required for an unspliced target. DNase-treated 18S rRNA C T values were comparable to non-DNase-treated C T s indicating that little to no 18S rRNA was lost in the DNase treatment (data not shown). Duplicates shown with Y-axis = ΔRn (change in fluorescence) from the Abbott m2000rt instrument. Labels indicate estimated parasite densities in parasites/mL (see Table 2 for log 10 values used in panel b. b Linear regression was performed using nominal log 10 transformed parasite densities and mean C T s (error bars 95% confidence intervals for available duplicate samples) to generate standard curves for the enriched gametocyte material [27]). False positive results for the PF3D7_1477700 and PF3D7_1438800 markers were also obtained when testing singleplex reactions against asexual stage parasites (data not shown). PF3D7_0630000 and pfs25, the spliced PF3D7_0816800 target were positive only in the presence of gametocytes (Table 3).
Quantification of gametocytes in field samples based on pfs25 resulted in estimated densities 1.18 log 10 parasites/ mL (95% CI 0.89-1.46 log 10 parasites/mL) lower on average than the paired estimates made with PF3D7_0630000. Lower pfs25-derived gametocyte quantification could not be explained by overall loss of total nucleic acids during DNase treatment since 35/39 evaluable paired samples (positive by pfs25 and PF3D7_0630000) actually showed earlier 18S rRNA C T s in the pfs25 sample compared to the PF3D7_0630000 sample (Additional file 3).

Discussion
Plasmodium gametocytes can be detected with mRNAbased molecular methods at sub-microscopic densities. Commonly targeted gametocyte-specific P. falciparum mRNAs include those made from single exon genes pfs25 and pfs230 [17][18][19][20], which necessitates DNase treatment prior to RT-PCR. Here, based on published RNAseq data [28], 18 gametocyte-specific, multi-exon mRNAs were identified and evaluated as DNase-free gametocyte targets to determine if they could be multiplexed with an existing highly sensitive Plasmodium 18S rRNA qRT-PCR. The bioinformatics strategy employed here selected for genes that were highly expressed in mature gametocytes and showed near-zero sense or antisense expression in asexual stages. In wet-lab testing, two female gametocyte-specific mRNAs showed suitable multiplex RT-PCR target characteristics: PF3D7_0630000 and PF3D7_0514500. PF3D7_0630000 encodes a CPW-WPC protein likely expressed in ookinetes, though the coding mRNAs are first expressed and post-transcriptionally regulated in gametocytes [38]. Since the PF3D7_0630000 mRNA accumulates in mature female gametocytes as the parasite awaits ookinete formation, this target may be an ideal gametocyte marker in human blood. PF3D7_0514500 (alias PFE0725C) encodes a six-exon conserved membrane protein of unknown function and was noted to be a member of the sexual development gene cluster in a previous full-genome high-density oligonucleotide microarray study [39]. Interestingly, PF3D7_0630000 was absent from the same gene cluster study. Since PF3D7_0630000 was the most promising target, it was intended to be compared against RT-PCR assays for pfs25 and for three other known spliced gametocyte-expressed mRNAs: PF3D7_0816800, PF3D7_1477700 and PF3D7_1438800. However, two of the intended comparator targets could not be multiplexed  with the 18S rRNA RT-PCR assay and were therefore not studied against PF3D7_0630000. These targets produced positive results when tested against microscopically-pure asexual stage cultures (that were also pfs25-negative).
The extraction method and RT-PCR mastermix used here differed from that originally reported for these markers [27], which may account for discrepancies. In addition, unlike the other targets studied in this project, the PF3D7_1477700 and PF3D7_1438800 mRNAs were both male gametocyte-specific [37], which could have led to false positives in asexual samples if male gametocytes (or less apparent exflagellated forms) were present at minuscule concentrations in asexual cultures; such forms would also be pfs25-negative. While PF3D7_1477700 and PF3D7_1438800 mRNAs may be suitable for multiplexing with 18S rRNA in another extraction/mastermix system, this possibility was not further evaluated here. The PF3D7_0630000 multiplex RT-PCR was comparably sensitive to pfs25 when tested against a samples from asymptomatic Ugandan subjects. Given this performance, PF3D7_0630000 RT-PCR will be able to detect gametocytes at densities that contribute to transmission, although this should be studied in future in prospective studies. Interestingly, PF3D7_0630000-derived gametocyte density estimates were 1.18 log 10 higher on average than pfs25-based estimates, which likely reflects degradation of pfs25 mRNA (but not more robust 18S rRNA) during the DNase treatment step. Detection of pfs25 mRNA by QT-NASBA (which does not require DNase treatment) was reportedly more sensitive than pfs25 RT-PCR, and the authors postulated that the DNase treatment step could have reduced the sensitivity of pfs25 RT-PCR [12]. Thus, it is likely that the increased sensitivity of pfs25 RT-PCR (due to its increased expression) is somewhat offset by increased degradation at the DNase treatment step.
By eliminating DNase treatment, gametocyte-specific RT-PCR targets can be directly incorporated into multiplex RT-PCR assays from total nucleic acids. This importance of this workflow improvement may be significant. Although 'on-column' DNase treatment is available for some manual RNA purification kits, this was not an option for the platform used here (Abbott m2000sp/rt [22]. Elimination of DNase treatment would be advantageous. In epidemiological studies, simplified sample processing that minimizes hands-on time and eliminates manual steps are desirable since extra steps serve to increase false positive and negative results through crosscontamination, target degradation and other processing errors. The use of spliced gametocyte-specific mRNAs such as those identified here (PF3D7_0630000 and PF3D7_0514500) or those reported previously [26,27] offer the possibility of this sort of simplified testing.
Other spliced mRNA targets beyond these may also be suitable for RT-PCR. Similarly, spliced mRNA RT-PCR may also be useful for detecting asexual stage-specific spliced mRNAs such as the ring-specific transcript from the two-exon PF3D7_0501300 (PFE0065w) gene previously used as an asexual parasite marker [27]. Multiplex assays that include the Plasmodium 18S rRNA, a spliced gametocyte-specific mRNA and a spliced ring/asexualspecific mRNA could eventually provide for one-step P. falciparum infection monitoring that would differentiate between potentially symptomatic and asymptomatic infections. The PF3D7_0630000-specific spliced marker identified here may be useful in future gametocyte screening studies.