In vivo compartmental kinetics of Plasmodium falciparum histidine-rich protein II in the blood of humans and in BALB/c mice infected with a transgenic Plasmodium berghei parasite expressing histidine-rich protein II

De-identified, human whole blood samples isolated from patients with a P. falciparum infection were collected from the Johns Hopkins Hospital Microbiology Lab. DNA was extracted from 20 µL of whole blood using the QIAamp DNA Blood Mini Kit and eluted in 50 µL of water for qPCR amplification, as described previously [31], using novel oligonucleotide primers and a fluorescent probe specific for the P. falciparum Pfs25 gene (Additional file 1: Table S1). The remaining blood was centrifuged at 5000×g for 4.5 min to separate plasma and RBC fractions. The RBCs were washed four times with 1 mL of PBS.

The PfHRP2 gene, inserted into the pET 15b vector and cloned into Escherichia coli BL 21/DE3 cells, was expressed and purified using a previously described protocol [23]. Native PfHRP2 protein was purified using the same protocol from pelleted RBCs cultured with the P. falciparum FCQ79 parasite strain provided by PATH [23]. Concentrations of purified proteins were determined using amino acid analysis completed by the Protein Structure Core at the University of Nebraska.

The full-length P. falciparum 3D7 HRP2 (PF3D7_0831800) gene was inserted into the pL1694 vector and expressed under the P. berghei HSP70 promoter, with a 2A skip peptide between PfHRP2 and GFP. The cassette was flanked with integration sites for the 230p gene in P. berghei. Second passage P. berghei (strain ANKA) parasites were purified from 5 mice at 1% parasitaemia by collecting blood via cardiac puncture. Purified, mature schizonts (5 × 10⁷) were generated in vitro and transfected with 10 µg of pL1694 + HRPII-2A-GFP linearized with Bcl1 and Sap1, as previously described [32, 33]. Transfected parasites were injected into the tail vein of 6–8-week-old female Swiss Webster mice (Taconic). Utilizing the human dihydrofolate reductase (DHFR) selection cassette, progeny were selected with pyrimethamine at 7 mg/mL in the drinking water. Blood was collected via cardiac puncture and single clones of transfected parasites were generated by diluting parasites at 1 parasite/mouse across 20 mice. Isolated clones were screened by PCR for correct genetic manipulation (Additional file 2). PfHRP2 protein expression in vivo was verified by RDT, western blot, and immunofluorescence microscopy (Fig. 3).

Eleven 8-week-old female BALB/c mice (Jackson Laboratory) were infected with approximately 500,000 to 1,000,000 PbPfHRP2 iRBCs isolated from a donor mouse. Once mice reached > 5% parasitaemia at days 4–5 post-infection, they were administered, via intraperitoneal injection, 3 doses of 50 mg/kg of artesunate at 0, 8, and 24 h and a single dose of 36 mg/kg of pyronaridine at 24 h. Artesunate (Sigma-Aldrich) was dissolved in 5% NaHCO₃ and pyronaridine tetraphosphate (Sigma-Aldrich) was dissolved in nuclease-free water. Parasite clearance time was monitored by blood films and qPCR performed as previously described [31] on DNA extracted from 20 µL of whole blood, using primers and probes specific for P. berghei ANKA 18S rRNA (Additional file 1: Table S1). Three copies of 18S rRNA/parasite was assumed to calculate parasites/µL because the primers and probes are capable of binding 3 copies of the P. berghei 18S rRNA gene on chromosomes 5, 7 and 12, as was determined by BLASTing the sequences of these primers and probes against the P. berghei ANKA genome using PlasmoDB v.41. In the P. berghei genome, there are four ribosomal units, classified into two types, A-type (includes units A and B) and S-type (includes units C and D) [34], and three of these four units have been shown to have greater than 90% sequence homology (A, C, and D) [34‐36]. 140 µL of tail blood was drawn daily for protein quantification. Drawing this volume of blood daily did not impact protein kinetics or clearance, as blood drawn only prior to treatment (day 0) and on day 5 or 7 post-treatment yielded similar PfHRP2 kinetics.

The Malaria Ag CELISA specific for PfHRP2 used to quantify PfHRP2 in murine and patient samples was performed according to the manufacturer’s instructions [37]. Absorbance was read at 450 nm using the BMG Labtech POLARstar OPTIMA plate reader. The limit of detection (LOD) was determined on each plate by averaging the absorbance of the negative controls for each blood fraction and adding 3 standard deviations to the mean. Any absorbance reading under the LOD was defined as PfHRP2 negative. The limit of quantitation (LOQ) was determined by the linear range of the standard curve on each plate. The LOQ was approximately 2 ng/mL of PfHRP2.

Plasma isolated from P. falciparum-infected patients (n = 5) at a concentration of 1000–3000 ng/mL of PfHRP2, and human serum spiked with a very high concentration of rPfHPR2 well above circulating plasma levels observed during severe infection, or a lower concentration of PfHRP2, were added to uninfected, human O⁺ erythrocytes to reconstitute whole blood at a 40% haematocrit. Additionally, 200 µL of tail blood was drawn from naïve BALB/c mice and deposited in 20 µL of 1 mg/mL heparin in PBS. Blood was centrifuged at 5000×g for 4.5 min to isolate plasma. The RBCs were washed one time with 1 mL of PBS. The mouse plasma was then spiked with the same concentrations of rPfHRP2 and added to the washed RBCs to reconstitute whole blood at 40% haematocrit. Plasma/sera samples containing no PfHRP2 protein were used as controls. All were incubated at 37 °C for at least 12 h. After incubation, samples were centrifuged to separate plasma from RBCs. RBCs were subsequently washed 4 times with 1 mL of PBS. Three individual replicates were performed.

The majority of data was analysed using GraphPad Prism (v.5). Data are graphed as mean ± SEM. PfHRP2 levels in the RBCs and plasma were extrapolated using a standard curve and linear regression analysis. Protein half-life was determined using a first-order decay equation. For the comparison between splenic and asplenic mice, data were normalized to percent of total remaining. The statistical methods used include: the Wilcoxon signed-rank test or Mann–Whitney U test to measure differences in the means of two matched or unmatched groups, first-order decay analysis to calculate half-life, log-rank survival analysis to compare clearance times, and a random-effects GLS regression model for analysing associations between tgPfHRP2 in the plasma and RBC fractions of splenic or asplenic mice. The random-effects GLS regression was used to account and adjust for within-subject correlation, using random intercept and robust standard errors. This analysis was performed with Stata version 14 (StataCorp LP, College Station, USA).

The persistence of PfHRP2 in whole blood has been documented [10‐13], but has not been extensively characterized in the blood’s individual compartments, the plasma and the RBCs. We obtained whole blood from 6 patients diagnosed with P. falciparum malaria at the hospital laboratory with parasitaemias ranging from < 0.5 to 12%, as measured by microscopy (Table 1). The whole blood was fractionated into plasma and washed RBCs for a quantitative, compartmental analysis of PfHRP2 by ELISA before and after treatment with anti-malarial combinations of artemether–lumefantrine (n = 2) or atovaquone–proguanil (n = 4) (Table 1; Fig. 1a, b). The initial ratio of RBC to plasma levels of PfHRP2 prior to treatment approximated 20-fold and this ratio tended to increase to 40-fold or greater following treatment with both anti-malarial regimens (Table 1; Fig. 1c). PfHRP2 levels in the RBC fraction only declined three-to tenfold in the days following either treatment regimen, while genomic copies of the P. falciparum Pfs25 gene decreased 20- to 100-fold (Table 1; Fig. 1a, b). All iRBCs cleared from the peripheral blood within 2–3 days post-treatment when measured by microscopy, despite parasite genomic DNA remaining detectable when measured by qPCR (Table 1). To identify the source of PfHRP2 in the blood post-parasite clearance, we performed immunofluorescence microscopy on RBCs with an antibody specific for PfHRP2 and DAPI staining to identify parasites. Remarkably, we observed PfHRP2 positive, but DNA negative, RBCs at approximately 0.1–1% of total RBCs post-treatment, specifically from days 1 to 3 and day 12 post-treatment (Table 1; Fig. 2; Additional file 3). This indicated the presence of what was likely once-infected RBCs, in which the parasite was cleared, but PfHRP2 remains within the cytoplasm of the RBC. The phenomenon was observed after treatment with both artemether–lumefantrine and atovaquone–proguanil in the small number of patients we examined (Table 1; Fig. 2). Also, in the 6 patients tested, no once-infected RBCs were observed pre-treatment (Additional file 4).

In order to fine tune the dynamics of PfHRP2 in vivo, a controlled experimental model was utilized in which a P. berghei rodent parasite (PbPfHRP2), which lacks the PfHRP2 gene family, was engineered to express PfHRP2 (tgPfHRP2) (Additional file 2). Expression and localization of the transgenic PfHRP2 protein (tgPfHRP2) was validated in the novel mouse model by western blot and immunofluorescence (Fig. 3). The presence of tgPfHRP2 in the whole blood of BALB/c mice was detected using a BinaxNOW Malaria Rapid Diagnostic Test™ that qualitatively detects PfHRP2 and pan-species aldolase (Fig. 3a). Western blot analysis confirmed the presence of tgPfHRP2 in the iRBCs (Fig. 3b). tgPfHRP2 was detected in the RBC cytosolic fraction after saponin lysis (Fig. 3b), suggesting it was exported into the iRBC cytoplasm. Immunofluorescence microscopy on PbPfHRP2 iRBCs was performed using monoclonal antibodies against PfHRP2 and Plasmodium enolase, a known parasite-localized protein, to differentiate the parasite from the iRBC cytoplasm. In this assay, tgPfHRP2 localized to both the parasite and iRBC cytoplasm (Fig. 3c). Importantly, the PfHRP2 staining observed in PbPfHRP2 iRBCs mimicked the staining of P. falciparum iRBCs (Fig. 3c). PfHRP2 was not detected in negative controls, including RBCs infected with the P. berghei ANKA wild-type strain that does not express PfHRP2 and uninfected RBCs (Fig. 3b, c).

BALB/c mice were infected with PbPfHRP2 and 5 days post-infection, reached an average parasitaemia of 10.5%, as measured by microscopy, or 675,283 parasites/µL quantified by qPCR (Fig. 4a). On day 5 post-infection, mice were treated with three doses of 50 mg/kg of artesunate (0–8–24 h post-initial treatment) and 36 mg/kg of pyronaridine (24 h post-initial treatment) to cure mice of parasites (Fig. 4a). At this peak parasitaemia prior to treatment, the average concentrations of tgPfHP2 in the plasma and RBCs were 12,512 ng/mL and 179,433 ng/mL, respectively (Fig. 4a, b). tgPfHRP2 levels were significantly higher in the RBC fraction (10–20 fold) compared to the plasma over the days post-treatment (p ≤ 0.0001, random-effects GLS regression model) (Fig. 4b, c). The half-life of tgPfHRP2 following treatment was calculated for each blood fraction using a first-order decay equation. While not significantly different, tgPfHRP2 had a slightly longer half-life in the RBCs at 13.1 h (median R² = 0.989) compared to 10.2 h in the plasma (median R² = 0.999) (p = 0.27, Wilcoxon signed-rank test) (Fig. 4d).

Median parasite clearance time measured by microscopy was 3 days post-treatment (Fig. 5a). However, the median parasite genomic DNA clearance time quantified by qPCR was significantly longer at 7 days (p  ≤ 0.0001, log-rank) (Fig. 5a). Median plasma tgPfHRP2 clearance time was 5 days while RBC clearance was significantly longer at 9 days (p ≤ 0.0001, log-rank) (Fig. 5b). Consequently, tgPfHRP2 positivity in the different blood fractions, particularly within the RBC fraction, outlasted detectable parasites by microscopy (Fig. 5a, b). Parasite genomic DNA was detected by qPCR longer than plasma tgPfHRP2, but protein in the RBCs persisted at least 1-day post genomic DNA clearance (Fig. 5a, b). tgPfHRP2 clearance time from the RBCs measured by ELISA correlated with its clearance time from whole blood of 8–9 days measured by RDT (Fig. 5c). RDT analysis also demonstrated that the Plasmodium aldolase protein was eliminated from the whole blood on day 4 post-treatment, shortly after parasites were no longer detected in peripheral blood (Fig. 5c).

The source of tgPfHRP2 in RBCs after parasite clearance was hypothesized to be once-infected RBCs, positive for PfHRP2 but parasite negative, as were observed in human patients post-treatment (Fig. 2). Immunofluorescence microscopy on mouse RBCs was performed with an antibody specific for PfHRP2 and DAPI staining to identify parasites. tgPfHRP2 was detected in iRBCs (DAPI positive) isolated pre-treatment as well as in RBCs isolated 2, 3, and 4 days post-treatment that were DAPI negative (Fig. 6a). This indicates the presence of once-infected RBCs positive for PfHRP2 in their cytoplasm in our mouse model. Importantly, PfHRP2 was not detected in naive mouse or human RBCs, indicating specific recognition of parasite derived PfHRP2 (Fig. 6). PfHRP2-positive, parasite-negative, once-infected RBCs that were observed in a human patient on days 2, 3 and 12 post-treatment with artemether–lumefantrine for a P. falciparum infection with an 11.6% initial parasitaemia (Fig. 6b), were morphologically similar to those identified in the mouse model (Fig. 6).

Slower clearance of PfHRP2 from the RBC fraction in humans and mice could be the result of plasma PfHRP2 binding to naïve RBCs in circulation. To investigate binding, partitioning of PfHRP2 was measured between naïve RBCs and plasma isolated from P. falciparum infected patients or human serum/mouse plasma spiked with recombinant PfHRP2 at average levels > 100,000 ng/mL or < 30,000 ng/mL as measured by ELISA, that were incubated together in vitro (Additional file 5: Table S2). Binding was defined as absorbance levels above those measured for RBCs incubated with naïve plasma. While protein levels quantified by ELISA were highly variable in the RBC and plasma fractions within and across replicates, the average percent of total PfHRP2 protein bound to uninfected, naïve mouse or human RBCs in vitro was extremely low, measuring below 0.5% for all plasma/serum samples tested (Additional file 5: Table S2).

Next, binding was investigated in vivo. BALB/c mice (n = 3) were injected with 100,000 ng/mL of recombinant PfHRP2 (rPfHRP2) in order to achieve high plasma protein levels and mimic plasma protein circulation and elimination observed during severe malaria infections (> 10% parasitaemia) [20, 27, 38]. ELISA was used to quantify concentration and partitioning of rPfHRP2 in the RBC and plasma fractions isolated from mice at regular intervals post-inoculation (Additional file 6: Table S3, Additional file 7). At 0.5 h post-inoculation, the average concentration of rPfHRP2 quantified in the plasma was 35,113 ng/mL (Additional file 7).

Minimal binding to uninfected RBCs was detected within the first 12 h post-inoculation, on the order of 3–13 ng/mL compared to the 900–35,000 ng/mL of rPfHRP2 in the plasma at those same time points (Additional file 6: Table S3). Consequently, the ratios of rPfHRP2 in the uninfected RBCs compared to the plasma in the first 12 h were extremely low, averaging 0.004 (Additional file 6: Table S3). The percent of the total protein (RBC + plasma) that bound to the RBCs over the time course was ≤ 1% (Additional file 6: Table S3). No binding was detected in the naïve RBCs after 24 h, despite plasma protein remaining detectable for a median of 4 days post-inoculation (Additional file 7; Additional file 6: Table S3). rPfHRP2 plasma clearance followed a first-order decay process, with a very short calculated plasma half-life of 1.9 h (median R² = 0.998) (Additional file 7).

Because RBC pitting occurs in the spleen, how splenectomy in mice alters clearance, persistence, and kinetics of tgPfHRP2 following infection with the PbPfHRP2 parasite and treatment with artesunate and pyronaridine was investigated. The parasitaemia prior to treatment (4–5 days post-infection) determined by qPCR was significantly different between the splenic (675,283 parasites/µL) and asplenic (1,429,170 parasites/µL) mice (Mann–Whitney U test, p = 0.041). To normalize the data, the percent of total parasite DNA remaining was calculated (Fig. 7a). On average, 71% of total parasite DNA cleared on day 1 post-treatment in splenic mice, but only 49% cleared in asplenic mice (Fig. 7a). Additionally, in mice with intact spleens, all parasite DNA cleared by day 7 post-treatment, but in asplenic mice, DNA was detectable at very low levels (≤ 1%) from days 12 to 14 post-treatment, leading to a significantly longer overall clearance time of 13 days (p = 0.0001, log-rank) (Fig. 7a, b). When measuring parasite clearance by microscopy, no differences between splenic and asplenic mice were observed, as both demonstrated a median clearance time of 3 days.

In asplenic mice, the concentration of tgPfHRP2 was significantly higher in the RBCs compared to the plasma over the days post-treatment (p = 0.005, random-effects GLS regression model) (Fig. 7c), which was the same trend observed in splenic mice. tgPfHRP2 elimination from the different blood fractions post-treatment also followed a first-order decay process in the mice lacking spleens. In the asplenic mice, the average tgPfHRP2 plasma half-life was 16.4 h (median R² = 0.990) and the average half-life in the RBCs was not significantly different at 13.4 h (median R² = 0.998) (p = 1.00, Wilcoxon signed-rank test). The starting tgPfHRP2 concentration in the RBC fraction of PbPfHRP2 infected mice prior to treatment was higher in the splenic group (179,432.6 ng/mL) compared to the asplenic group (144,806 ng/mL), despite a lower parasite load calculated by qPCR. Data was normalized to percent of total tgPfHRP2 remaining in the RBCs (Fig. 7d). Interestingly, when comparing percent of total protein remaining in the RBCs post-treatment for the splenic and asplenic mice, the initial tgPfHRP2 clearance rate for more than 99% of the protein was not different within the first 6 days post-treatment (Fig. 7d). Consequently, the tgPfHRP2 RBC elimination half-lives in splenic (13.1 h) and asplenic mice (13.4 h) were not significantly different (p = 1.00, Mann–Whitney U Test). However, lingering, low levels of tgPfHRP2 (< 1%) were detected in the RBCs of the asplenic mice during days 10–14 post-treatment, leading to a significantly longer overall median clearance time of 13 days (p = 0.0009, log-rank) (Fig. 7d, e). This was not observed in the splenic group, which had a median clearance time of 9 days. As in splenic mice, tgPfHRP2 persisted significantly longer in the RBCs (11–14 days) than the plasma (6–10 days) in asplenic mice (Fig. 7d). One apparent difference between these groups was that in asplenic mice, the tgPfHRP2 protein in the RBCs cleared at the same time or before parasite genomic DNA (Fig. 7b, e). In the splenic mice, at least a 1-day difference in clearance time was observed (Fig. 5).