Ethical approval for the Ghanaian samples in the study was given by the Institutional Review Board of the Noguchi Memorial Institute for Medical Research of the University of Ghana, Accra, Ghana. Parents and guardians gave written informed consent before their children were enrolled into the study. Ethical approval for Danish blood donor samples was given by the Scientific Ethics Committee of Copenhagen and Frederiksberg, Denmark. These were individuals residing in central Copenhagen and provided written consent for a small amount of their blood stored, anonymously, and used for research purposes.

Plasmodium falciparum strain NF54 was cultured as described elsewhere 28 . Briefly, parasites were maintained in culture using 2.5% haematocrit of human blood group O + in parasite growth medium (PGM) consisting of RPMI 1640 (Lonza, USA) supplemented with 0.5% Albumax, 25 mM HEPES, 2 mM L-glutamine, 24 mM NaHCO₃, 25 μM gentamicin and 10% (v/v) heat-inactivated human blood group AB serum. Culture was maintained at 37°C in 25-sq cm flasks after gassing with a gas mixture containing 5% O₂, 5% CO₂ and 90% N₂. For the staining assays, asynchronous parasite cultures were used while successive treatment with 5% D-sorbitol 29 was used to synchronise cultures for the antibody dependent cellular inhibition assay (ADCI) assay. To obtain high-parasitaemia cultures for staining assays without driving parasites into crisis state or gametocytogenesis, cultures were double synchronized by D-sorbitol treatment and enrichment for matured stage parasites by magnetic separation (Miltenyi Biotec) after 70% Percoll treatment as described 30 . High-parasitaemia cultures were maintained at low haematocrit (0.5%) with three media changes per week.

Microscopy analysis of culture parasitaemia was performed in thin blood smears fixed with 100% methanol and stained with Giemsa (Merck Co, Germany) in a 1:10 dilution in RPMI 1640 (Lonza, USA). After 10-min incubation at room temperature, slides were washed with distilled water and dried at room temperature. Parasitaemia was estimated by counting approximately 5,000 erythrocytes under oil immersion (×100) objective lens.

Total immunoglobulin (Ig) G was purified from plasma samples as previously described 31 . The negative and positive control IgG used in the ADCI assays were pooled IgG from malaria-naïve Danish individuals (PNIG) and pooled IgG from hyper-immune African adults (PHIG), respectively 26 31 . The test samples were IgG purified from 19 Ghanaian children (age range one to 12 years) naturally exposed to malaria 32 .

To obtain the optimal concentration of CPO for staining parasites, 3 μl P. falciparum NF54 asynchronous culture at ~13% parasitaemia (microscopical estimation) was added to each well of a 96-well round-bottom plate (Nunc, Roskilde, Denmark) containing 100 μl PGM and centrifuged (Sigma 3-16 K, Germany) at 770 × g for 5 min at room temperature (rt) to harvest cells. Cell pellets were resuspended in 100 μl of PGM containing 100 nM, 20 nM, 5 nM and 2.5 nM CPO (Tokyo Chemical Industry Co, Japan), respectively, in duplicate wells and incubated in the dark at 37°C for 30 min. Cells were washed (centrifugation at 770 × g for 5 min at 5°C) twice, each with 100 μl PGM. Pellets were resuspended in 200 μl PGM for analysis by flow cytometry to identify the best CPO concentration for staining parasites.

CPO was combined with 5 μM of mitotracker red dye 26 . The double staining protocol was assessed using a P. falciparum NF54 asynchronous culture at ~4.0% parasitaemia. Here, 3 μl of parasite culture were co-stained with 100 μl of the CPO-mitotracker red mix in PGM and incubated at 37°C for 30 min in the dark. Cells were then washed twice as described above and resuspended in 200 μl PGM for counting in the flow cytometer.

To investigate the dynamics of CPO and mitotracker red dual-colour staining, P. falciparum NF54 asynchronous culture at 2% parasitaemia and 2.5% haematocrit was treated with 10 nM atovaquone/1 μM proguanil for up to 48 hours to depolarize the mitochondria membrane potential 26 33 . Parasitaemia prior to drug treatment (0 hrs) and at 24 hrs and 48 hrs post-treatment was estimated by flow cytometry using the CPO-mitotracker red dual-colour staining protocol as described above. Additionally, the same samples were analysed by confocal microscopy.

To develop the tri-colour staining protocol that enables exclusion of leucocytes from the assay, P. falciparum NF54 asynchronous culture of ~15% parasitaemia and 2.5% haematocrit was mixed with ~1% of human monocyte isolated with the EasySep™ Human Monocyte Enrichment Kit (STEMCELL™ Technologies, France). The resulting mixture was then stained with the CPO-mitotracker red dual-colour staining protocol in a 96-well round-bottom plate. Control wells contained a mixture of uRBCs and ~1% of human monocyte. Following the 30 min of CPO-mitotracker red incubation in the dark at 37°C, the plate was centrifuged at 770 × g for 5 min at 5°C to remove excess dye and cells were washed once in 100 μl PGM. Five μl of mouse anti-human CD45 mAb (clone H130), PerCP conjugate (Invitrogen™, USA) was added and cells were incubated at 4°C for 10 min resulting in the tri-colour staining protocol. Excess CD45 mAb was removed by centrifugation and the cells washed once with 100 μl PGM. Cells were resuspended in 200 μl PGM for counting in the flow cytometer.

Plasmodium falciparum NF54 asynchronous culture of ~30% parasitaemia and 2.5% haematocrit was mixed with ~26% of human monocyte isolated with the EasySep™ Human Monocyte Enrichment Kit (STEMCELL™ Technologies, France). The resulting mixture was serially diluted two-fold using erythrocytes at 2.5% haematocrit in PGM in a 96-well round-bottom plate so that the final theoretical parasite and monocyte proportions were 0.0037 and 0.0032%, respectively. Microscope slides were prepared for each well and stained with Giemsa to estimate the ‘expected’ parasitaemia by microscopy. The remaining cells were then stained using the tri-colour staining technique and analysed in the flow cytometer. Monocyte counting by microscopy was not possible after the serial dilutions therefore the ‘expected’ monocyte proportion was estimated from the dilution factor and compared with the tri-colour measurements.

Flow cytometry data acquisition was done on Beckman Coulter cytometer (Cytomics FC500 MPL) using the argon ion laser (488 nm). The detectors of forward scatter, side scatter, FL1 (for detection of DNA bound CPO signal), FL3 (for detection of mitotracker red) and FL4 (for detection of CD45-PerCP) were set in logarithmic mode. The excitation wavelength and the detector filters of CPO, mitotracker red and CD45-PerCP are, respectively, 500 nm/525 BP, 579 nm/620 BP and 488 nm/ 675 BP. The machine was set up by initially running unstained cells and cells that had been stained separately with each fluorochrome. Appropriate electronic colour compensations were adjusted between the three dyes. DNA-bound CPO florescence in FL1 was preferred to RNA-bound CPO signal in FL2 for parasite quantification to minimize the signal overlap with mitotracker red in FL3, thereby making electronic colour compensation much simpler between the two dyes. Also, DNA is fairly constant compared to RNA, which may vary widely depending on the state of the parasite. The average data acquisition speed was 500 events per sec for 1 min with a minimum of 30,000 events or more than 300 infected red blood cells (iRBCs) counted 26 and analysed with Kaluza Analysis Software (Version 1.3). In the tri-colour staining, events were first gated according to their FSC/SSC profile to exclude doublets and debris. A second gate in the dot plot SSC/FL4-CD45-PerCP was set up to exclude leucocytes (monocytes) and the CD45-PerCP negative cell population was then carried into the next level of analysis where they were displayed in SSC/FL1-CPO dot plot to identify all iRBCs (total parasitaemia). Finally, the positive events were then displayed in SSC/FL3-mitotracker red dot plot to estimate iRBCs with viable mitochondrial membrane potential and those negative for mitotracker red signal (dead and/or compromised cells).

The ADCI assays were performed as previously described 26 , with slight modifications. Briefly, about 2 × 10⁵ monocytes/well were selected by adherence to a 96-well flat-bottom culture plates (Nunc, Roskilde, Denmark) after a 2-hr incubation of two million PBMCs/well at 37°C and 5% CO₂ in monocyte medium (RPMI medium containing 5% NHS, 1% glutamine and 1% penicillin-streptomycin). Highly synchronized parasite culture, at late schizont stage of 0.5% parasitaemia and 2.5% haematocrit were added, 100 μl/well. Control and test antibodies were added to respective wells at 1 mg/ml and the total volume adjusted to 200 μl with PGM for wells containing uRBCs and blank controls. Control wells included: (i) parasite culture without monocytes; (ii) culture with monocytes but without IgG; (iii) culture with pool of naïve Danish IgG (PNIG); (iv) culture with monocytes and PNIG; (v) culture with pool of hyper immune African adult IgG (PHIG); and, (vi) culture with monocytes and PHIG. At 48 hrs and 72 hrs respectively, an additional 50 μl of PGM was added to each well and after 96 hrs the assay was stopped and the final parasitaemia determined by microscopy and flow cytometry using the novel tri-colour method. The 19 different test samples were tested on different days and the corresponding microscopy slides for each sample were also obtained for parasitaemia determination. The SGI estimating the parasite growth inhibition due to the effect of antibodies co-operating with monocytes (MN) was calculated as follows: SGI = 100 × (1 - (% parasitaemia with MN and test antibodies/% parasitaemia test antibodies)/(% parasitaemia with MN and PNIG/% parasitaemia PNIG)).

The two-way ANOVA for overall comparison and subsequent pairwise comparisons corrected with the Bonferroni method were used to test differences between the parasitaemia estimated by CPO and mitotracker red during the drug treatment experiment. Linearity in the estimation of parasitaemia and monocyte proportions, respectively, by the tri-colour staining technique was determined by Pearson’s r and linear regression. The Bland-Altman analysis was used to compare biases between SGI estimated by the tri-colour staining technique, mitotracker red and by microscopy. All statistical analyses were performed using GraphPad Prism vs.5.04 for Windows, (GraphPad Software, USA).

Plasmodium falciparum NF54 asynchronous culture at ~13.0% parasitaemia (microscopy estimate) and 2.5% haematocrit was stained with CPO at various concentrations ranging from 100 nM to 2.5 nM (Figure 1). All CPO concentrations gave a parasitaemia close to the count obtained by microscopy. Twenty (20) nM CPO was chosen for subsequent experiments since it best discriminated between iRBCs and uRBCs without measurable background staining. In addition, the clearest discrimination of the different parasite developmental stages was observed at the 20 nM CPO staining concentration (Figure 1; Panel B1, B2, B3).

CPO (20 nM) together with mitotracker red (5 μM) 26 was used to stain P. falciparum NF54 asynchronous culture at ~4.0% parasitaemia (microscopy estimate) and measured by cytometry. Applying the CPO stained uRBC control (Figure 2, Panel A1) gating to the test sample, there was a more distinct separation between uRBCs and iRBCs allowing for a much precise total parasitaemia (live and dead or compromised combined) estimation (Figure 2, Panel A2). However, when the mitotracker red-stained uRBC control (Figure 2, Panel B1) gating was applied to the test sample, the parasitaemia was underestimated (Figure 2, Panel B2) as a consequence of the poor distinction between uRBCs and early ring stages of the parasite. Thus, a strategy for estimating live parasitaemia was achieved by identifying cells that were CPO and mitotracker red double positive estimated as 3.49% (Figure 2, Panel C).

The robustness of the developed strategy to distinguish between compromised (‘crisis’ or dead) and a healthy (live) P. falciparum parasite was further investigated by both flow cytometry and confocal imaging analyses. After 24 hrs of atovaquone and proguanil treatment designed to induce compromised state parasites, the initial 2% parasitaemia dropped sharply to 0.5% by mitotracker red estimate and moderately (1.8%) by CPO estimate (Figure 3A). No further decrease in parasitaemia estimate was observed by mitotracker red staining while the CPO estimate showed a marginal decrease to 1.5% after 48 hrs (Figure 3A). The parasitaemia estimated by CPO and mitotracker red in the drug treatment assay were significantly different (p < 0.0001, Two-way ANOVA). At 24 hours and 48 hours post treatment, parasitaemia estimated by CPO was significantly higher than by mitotracker red (p < 0.0001, Bonferroni post hoc test). Confocal imaging analyses coinciding with the various sampling time for the flow cytometry indicated a drastic loss of mitotracker red signal 24 hours post treatment while CPO signal only showed a gradual decrease (Figure 3B).

The tri-colour staining technique was successfully achieved by the addition of the mouse anti-human CD45 mAb conjugated to PerCP for leucocytes quantification. Leucocytes contain both nucleic acids and mitochondria and may confound parasitaemia estimates when they are present together with iRBCs in bioassays. An initial gate (FSC/SSC), of all events was defined and doublets and debris were excluded. Secondly, a dot plot SS Log/FL4-CD45-PerCP (Figure 4, Panel A1) was made for the control sample, which contained only monocytes and uRBCs. Corresponding dot plots for SS Log/FL1-CPO (Figure 4, Panel B1) and SS Log/FL3-mitotracker red (Figure 4, Panel C1) were made for the same control well to identify where monocytes are localized in each plot. In the test sample, monocytes were first excluded from total RBCs as CD45-positive cells which were estimated to be 0.98% (Figure 4; Panel A2). Dot plots of SS Log/FL1-CPO (Figure 4; Panel B2) and SS Log/FL3-mitotracker red (Figure 4; Panel C2) without the exclusion of the monocyte population demonstrate their confounding effect in CPO and mitotracker red parasitaemia estimations, respectively. After monocyte exclusion, a dot plot of SS Log/FL1-CPO was made to identify all iRBCs (total parasitaemia) (Figure 4D). Finally, the CPO-positive events were then displayed in SS Log/FL3-mitotracker red dot plot to identify iRBCs with viable mitochondrial membrane potential (live cells) and those negative for mitotracker red signal (dead and/or compromised cells) which were estimated to be 14.02 and 0.29%, respectively (Figure 4E).

The tri-colour technique could reliably estimate leucocytes and parasites separately from a mixture containing both and the proportions were comparable to the expected. There was a strong linear relationship (r = 0.99995, slope = 1.005 ± 0.0016) between the microscopy and the tri-colour technique estimates of parasitaemia (Figure 5). Similarly, a strong linear relationship (r = 1.00, slope = 1.002 ± 0.00067) was shown for the estimation of monocyte proportion by the tri-colour method when compared with the theoretical expected proportions from the dilution (Figure 5). From the serial dilutions, the lower limit for accurate parasitaemia and monocyte detection by the tri-colour was ≥0.008 and ≥0.006%, respectively.

The tri-colour staining method developed was applied to estimate parasitaemia for the calculation of specific growth inhibition index (SGI) of IgG samples from 19 Ghanaian children on different days. The corresponding parasitaemia estimated by microscopy examination of Giemsa-stained slides was also determined and SGI calculated. The largest bias in SGI was observed when microscopy and mitotracker red alone were compared (bias = 27.62) (Figure 6A), while agreement between microscopy and the tri-colour method (bias = -1.60) was close (Figure 6B). The bias between the tri-colour method and mitotracker red alone was large (bias = 26.02). Generally, SGI estimated by microscopy and the tri-colour methods were systematically higher than the mitotracker red estimate (Figure 6A and 6C). However, microscopy based SGI estimates were slightly higher than the tri-colour estimates (Figure 6B). The tri-colour could reliably estimate the proportions of total parasitaemia that were alive or dead (and/or compromised) in the ADCI assay (Figure 6D). The percentage of live parasitaemia was estimated as mitotracker red and CPO double positive cells after the exclusion of monocytes by CD45 gating in the tri-colour. The patterns of total and live parasitaemia reflect day-to-day variability in culture growth in the assay.