An Experimental Human Blood-Stage Model for Studying Plasmodium malariae Infection

this study was to develop an induced blood-stage malaria (IBSM) model of P. malariae to study parasite biology, diagnostic assays, and treatment. Methods. This clinical trial involved 2 healthy subjects who were intravenously inoculated with cryopreserved P. malariae – infected erythrocytes. Subjects were treated with artemether-lumefantrine after development of clinical symptoms. Prior to antimalarial therapy, mosquito-feeding assays were performed to investigate transmission, and blood samples were collected for rapid diagnostic testing and parasite transcription profiling. Serial blood samples were collected for biomarker analysis. Results. Both subjects experienced symptoms and signs typical of early malaria. Parasitemia was detected 7 days after inoculation, and parasite concentrations increased until antimalarial treatment was initiated 25 and 21 days after inoculation for subjects 1 and 2 respectively (peak parasitemia levels, 174 182 and 50 291 parasites/mL, respectively). The parasite clearance half-life following artemether-lumefantrine treatment was 6.7 hours. Mosquito transmission was observed for 1 subject, while in vivo parasite transcription and biomarkers were successfully profiled. Conclusions. An IBSM model of P. malariae has been successfully developed and may be used to study the biology of, diagnostic testing for, and treatment of this neglected malaria species.

information for the development of diagnostic tests, vaccines, and drugs. While reliable VIS models have been developed for P. falciparum and P. vivax, none have been developed for P. malariae. The aim of this study was to develop an induced blood-stage malaria model (IBSM) with a wild isolate of P. malariae.

Study Design
This was a clinical trial involving 2 sequential cohorts of 1 subject each (clinical trials registration ACTRN12617000048381). The study was approved by the QIMR Berghofer Medical Research Institute Human Research Ethics Committee and conducted in accordance with the Declaration of Helsinki. Subjects provided written informed consent prior to participation in the study. The primary objective was to determine the safety and infectivity of the P. malariae isolate in healthy subjects following inoculation with blood-stage parasites. The secondary objectives were to define parasite growth curves and clearance profiles after administration of the antimalarial drug artemether-lumefantrine and to investigate the infectivity of parasites to vector mosquitoes. The exploratory objectives were to evaluate the in vivo pattern of gene transcription of P. malariae, to establish host and parasite biomarker dynamics during early stage infection, and to evaluate a commercially available rapid diagnostic test (in cohort 2 only).

Study Subjects
Healthy adults aged 18-55 years were eligible for participation. Subjects were required to be malaria naive, blood group A or AB, and Rh(D) positive. A full list of the inclusion/exclusion criteria for this study is included in the Supplementary Materials.

P. malariae Inoculum
The P. malariae isolate (HMPBS-Pm) used in this study was derived from blood donated by a traveler returning from Guinea, Africa, who presented to the Royal Brisbane and Women's Hospital with symptomatic, polymerase chain reaction (PCR)-confirmed P. malariae infection. The genome of this isolate has been sequenced and has previously been compared to the P. malariae reference genome [8]. The production of parasite inoculum banks from donor blood specimens has been previously described for P. falciparum and P. vivax [17,18].

Study Procedures
After intravenous injection of parasites on study day 0, subjects were monitored by daily telephone calls for the first 5 days. From day 6 onward, subjects visited the study site daily for clinical monitoring of adverse events (AEs), signs and symptoms of malaria parasite infection, and blood sampling for measurement of the parasitemia level.
Antimalarial treatment with artemether-lumefantrine was instituted following development of clinical symptoms consistent with malaria parasite infection or if the parasitemia level exceeded 200 000 parasites/mL, a safety threshold based on previous IBSM studies [19]. Prior to treatment, mosquito feeding assays were performed, and blood samples were collected for transcriptomic analysis, biomarker studies, and rapid diagnostic tests. Treatment consisted of 6 doses (administered 12 hours apart) of artemether-lumefantrine (Riamet, Novartis Pharmaceuticals Australia). Safety assessments and blood sampling to monitor the parasitemia level were conducted twice daily until parasite clearance and then at protocol-specified visits until 3 months after treatment, to monitor for any signs of recrudescence.

Safety and Parasitemia Assessment
Safety monitoring of subjects included recording adverse events, measuring vital signs, physical examination, electrocardiography (ECG), and measuring clinical laboratory parameters (via hematologic studies; clinical chemistry analyses, including creatinine studies; and urinalysis, including urinary protein studies).
The parasitemia level was measured using quantitative PCR targeting the gene encoding P. malariae 18S ribosomal RNA (rRNA), as previously described [17]. Search of the P. malariae genome confirmed that this Plasmodium species carries 2 copies of this gene target.

Calculation of Parasite Reduction Ratio (PRR)
The PRR and corresponding parasite clearance half-life were estimated using the slope of the optimal fit for the log-linear relationship of the decay in parasitemia level following artemether-lumefantrine treatment, as previously described [20].

Mosquito Feeding Assays
Transmission of gametocytes to Anopheles stephensi mosquitoes was evaluated using a direct skin feeding assay (DFA) and a direct membrane feeding assay (DMFA) as previously described for P. falciparum [21]. Mosquito midguts were dissected and tested for transmission positivity, using light microscopy with mercurochrome staining and also using quantitative PCR targeting the gene encoding P. malariae 18S rRNA.

Biomarker Dynamics Assessment
P. falciparum histidine-rich protein 2 (HRP2), pan-genus lactate dehydrogenase (LDH), P. vivax LDH, and human C-reactive protein (CRP) were quantified in blood samples collected at various time points, using the recently developed Q-plex array enzyme-linked immunosorbent assay kit (Quansys Biosciences, Logan, UT) [22]. Untransformed quantitative parasitemia and quantitative biomarker measurements were correlated using a Spearman correlation.

Assessment of Rapid Diagnostic Testing
Rapid diagnostic testing was performed using the lateral flow antigen detection test CareStart Malaria HRP-2/pLDH (Pf/pan) Combo Test (Apacor, Berkshire, United Kingdom). Blood specimens collected at the peak parasitemia level prior to treatment were tested as per manufacturer guidelines.

Assessment of P. malariae Transcription Profile
A 20-mL blood sample collected at the peak parasitemia level prior to treatment was depleted of leukocytes by using a Plasmodipur push column, and infected erythrocytes were concentrated using a Percoll gradient [23]. RNA was extracted and complementary DNA synthesized using a SMART-seq2 kit (Clonetech/Takara) with 32 cycles of amplification and PhiX bacteriophage as a control. The library was sequenced on a MiSeq (150-bp reads) and a HiSeq 2500 Illumina machine (125-bp paired-end reads). Sequence reads from the subject were excluded by alignment against the H. sapiens HS19 reference genome, using BWA as part of an automatic pipeline at the Wellcome Sanger Institute. Parasite messenger RNA reads were mapped against the P. malariae reference genome PmUG01 [8] from geneDB.org, using STAR default settings [24]. Visualization was performed in BamView [25] after transforming the read file into the BAM format by using SAMtools [26]. P. malariae reads were compared to existing P. falciparum data to describe the pattern of gene transcription and determine the developmental stage of parasites. Correlation to the P. falciparum data set [27,28] from PlasmoDB was performed in R, using the corrgram function. To enable the comparison, 1:1 orthologs between P. falciparum and P. malariae were determined by orthoMCL [29]. To better assess the more variable subtelomeric gene families, the genomic reads of the inoculum isolate HMPBS-Pm (accession ERS567899) were assembled using MaSuRCA (default parameters) [30]. Genome annotation was performed with Companion [31], using P. malariae UG01 as a reference (parameter Augustus ab initio set to 0.2).

Subjects
A total of 8 subjects were screened for eligibility, and 2 subjects were enrolled in the study (1 in each cohort). Subject 1 and subject 2 were healthy white men aged 24 and 18 years, respectively.

Safety
A total of 35 AEs were reported (Table 1), the majority of which (30 of 35 [85.7%]) were attributed to P. malariae infection. The remaining AEs were either attributed to direct skin feeding with mosquitoes (2 of 35) or considered to be unrelated to study procedures (3 of 35). Most AEs (27 of 35 [77.2%]) were mild in severity, with the remaining AEs moderate in severity. None of the AEs reported in this study were classified as serious. Both subjects experienced chills, fatigue, headache, malaise, myalgia, and nausea during the study. Fever and transient moderate neutropenia were also recorded for both subjects and were consistent with malaria parasite infection.

Parasitemia
Parasitemia was first detected 7 days after inoculation, and the level increased at a similar rate in both subjects until antimalarial treatment was initiated 25 and 21 days after inoculation for subject 1 and subject 2, respectively ( Figure 1). Parasitemia levels at the time of treatment were 140 890 and 50 291 parasites/mL, respectively. Clearance of parasitemia occurred following artemether-lumefantrine treatment in both subjects, with no recrudescence observed up to 3 months after treatment. Furthermore, there was no evidence of persisting gametocytemia, as would be indicated by persistent detection of P. malariae DNA. The peak parasitemia level recorded for subject 1 was 174 182 parasites/ mL, which occurred 4 hours after treatment. The parasitemia level at the time of treatment was the highest level recorded for

Transmission to Mosquitoes
DFAs and DMFAs were performed just prior to antimalarial treatment (25 and 21 days after inoculation for subject 1 and subject 2 respectively). The average feeding rate of mosquitoes was high (minimum, 95.1%), and the mortality rate was <26% for all assays (Supplementary Table 1). Transmission of parasites to mosquitoes was detected for subject 1, with 1 of 34 mosquitoes (2.9%) becoming infected following the DFA. Light microscopy confirmed that the infected mosquito contained a single oocyst. No transmission was observed with the DFA for subject 2 or with the DMFA for either subject.

Biomarker Dynamics and Rapid Diagnostic Testing
Serial blood sampling for evaluation of biomarkers of malaria parasite infection was performed for subject 2. Quantitative analyses revealed that HRP2 and P. vivax pLDH remained undetectable throughout the course of parasitemia. Pan-genus pLDH first became detectable on day 13 after inoculation, at a parasitemia level of 137 parasites/mL ( Figure 2). The pan-genus pLDH level was strongly correlated with the parasitemia level (R = 0.78; 95% CI, .52-.91), although there was an evident time lag in biomarker response. CRP levels increased with parasitemia levels, although the relationship was weaker (R = 0.13; 95% CI, −.32-.53; Figure 2). The time lag in biomarker response was more pronounced for CRP; peak values occurred 36 hours after initiation of treatment, and values remained elevated after parasite clearance. A CareStart Malaria HRP-2/pLDH (Pf/pan) rapid diagnostic test performed on day 21 after inoculation, at a parasitemia level of 50 291 parasites/mL, had a negative result.

Transcriptome Analysis
A blood specimen collected from subject 1 at the time of the peak pretreatment parasitemia level (140 890 parasites/mL) underwent RNA sequencing (RNA-Seq). Extraction of the RNA generated 100 ng to 3.3 µg of total complementary DNA for high-throughput sequencing. The raw reads can be found under the sample accession number ERS1699009. The accession numbers of the sequencing reads for the Illumina MiSeq and HiSeq 2500 are ERX2597961 and ERR2690580, respectively. A large proportion of sequenced reads corresponded to the host (51.2%), which were excluded automatically by the Wellcome Sanger Institute sequencing pipeline, or to the PhiX control (25.6%); 5.8% of sequenced reads were mapped against P. malariae. Of these, approximately 75% corresponded to rRNA. While the mRNA yield of P. malariae was low (1.45%), deep sequencing (>260 million reads) enabled approximately 3.8 million informative parasite-derived reads to be obtained that clearly mapped to the P. malariae genome. This represents a theoretical coverage of 33 times all coding sequences. If normalized coverage (in reads per kilobase of transcript, per million reads mapped [RPKM]) detection thresholds of 5 or 10 are assumed, expression was detected for 948-1463 genes ( Figure 3).
The RPKM raw reads count per gene for the 15 genes with highest expression are shown in Supplementary Table 2. Two of these (PmUG01_14016600 and PmUG01_14018200) encode gamete antigen 27/25. The P. malariae genome includes 20 apparently intact copies of this gene, compared with a single copy in other examined Plasmodium species [8]. Of these 20 copies, only 3 were highly expressed, with comparatively low expression in others (Figure 4).
Parasite development stages within the sample were determined by comparing gene expression profiles against 2 timecourse data sets from P. falciparum (Supplementary Figure  1). Correlations were low (0.12-0.37) but peaked against the 32-hour intraerythrocytic form (0.35) and gametocyte II (0.37).
The structure of expressed genes could be clearly resolved by visual inspection in which mRNA reads were compared to PmUG01 reference genome predictions ( Figure 3A). Expression signals with peaks in the 5′ untranslated region (UTR) or 5′ intergenic regions were seen in coverage for several highly expressed genes ( Figure 3B-D). In some cases, discrete coverage distal to an existing gene prediction was indicative of a novel noncoding RNA ( Figure 3B). In other cases, the 5′ UTR coverage bias appeared to reflect overamplification during library production ( Figure  3C). No evidence of a long UTR or novel noncoding RNA was found in the regions of higher expression when comparing the P. malariae genome PmUG01_05020900 to published data sets on PlasmoDB for P. falciparum orthologs (PF3D7_0825100; Figure 3D). RNA-Seq data were also used to correct the intronexon structure of predicted genes in the current annotation of the reference genome ( Figure 3E). It was possible to perform an exploratory analysis of expression of predicted exported proteins  that may contribute to parasite virulence. The P. malariae genome published by Rutledge et al [8] reported 2 new tandemly organized multigene families (fam-l and fam-m) that share structural similarity with a P. falciparum merozoite surface protein, encoded by Rh5 (PF3D7_0424100), that is essential for erythrocyte invasion. We were unable to find evidence for expression of any fam-l and fam-m pairs, even when RNA-Seq data were matched to an assembled and annotated genome from the study isolate. In contrast, evidence of expression (RPKM, >10) was found for other subtelomeric gene family members, of which 22 encoded putatively exported proteins, including 8 PHIST proteins and 1 STP1.
Ten pir genes also appeared to be expressed, but all had a RPKM value of < 5, based on the PmGN01 annotation.

DISCUSSION
Inoculation of healthy subjects with the P. malariae HMPBS-Pm clinical isolate resulted in a safe and reproducible course of parasitemia that was well tolerated. Specifically, there was no evidence of renal impairment or late recurrent infection in the 3 months following treatment. The overall nature, frequency, and severity of AEs was similar to that reported following experimental inoculation with blood stages of other Plasmodium species, where mild-to-moderate symptoms attributable to malaria were observed with increasing parasitemia levels [17,18]. Significantly, no AEs were classified as severe or serious, and no biochemical abnormalities were noted. Parasite growth following inoculation was slower for P. malariae, compared with that reported for P. falciparum or P. vivax [17,18], consistent with the slower 72-hour intraerythrocytic life cycle and smaller number of merozoite schizogony per cycle characteristic of the species [32]. The pyrogenic threshold of P. malariae infection in this study was lower than that reported for historical case series from the malariotherapy era, where the median parasitemia at first fever was reported to range from 997 800 to 2 664 000 parasites/mL [19]. This may be due to differences in interindividual infection tolerability, the strain of P. malariae, or increased subject monitoring in modern clinical trials.
Parasite clearance following administration of artemether-lumefantrine was slower than that estimated for species with a shorter 48-hour intraerythrocytic life cycle. The parasite clearance half-life following artemether-lumefantrine treatment was 6.7 hours (95% CI, 5.91-7.74) in this study. As a comparison, a parasite clearance half-life of >5.0 hours is used to indicate artemisinin resistance in P. falciparum [33]. Nonetheless, standard-dose artemether-lumefantrine was effective, without evidence of recrudescence at follow-up 3 months after treatment. However, there are reports of treatment failure in the literature [9,13]. Postulated mechanisms of a refractory response of P. malariae infection to artemether-lumefantrine include a relative reduction in drug exposure due to a prolonged parasite life cycle or delayed hepatic schizogony, parasite dormancy and reduced susceptibility to treatment, an undescribed hypnozoite stage, and genetic subpopulations with intrinsic drug resistance [11,34]. Treatment success in this study may be related to a relatively low parasitemia level at the time of treatment or to the susceptibility of this P. malariae strain. In the absence of a hepatic stage of infection in the IBSM model, it is not possible to assess the effect of delayed hepatic schizogony or as-yet-undescribed hypnozoite forms.
The successful transmission of P. malariae to mosquitoes confirms gametocytogenesis in the model. While only a single transmission episode occurred, the low rate of transmission is consistent with that for other Plasmodium species at low levels of parasitemia/gametocytemia and validates the model as a tool for the evaluation of parasite transmission [21]. Relatively little is known regarding the dynamics of P. malariae gametocyte commitment and transmission [19]. Significantly, P. malariae coinfection has been implicated in enhanced P. falciparum gametocytogenesis, suggesting that this species could potentiate transmission of more-virulent disease [35][36][37]. Such interactions are exceedingly difficult to confirm in the field, making this model attractive to characterizing parasite biology and developing transmission-blocking interventions. Pan-genus pLDH was the only Plasmodium-specific biomarker detected during the course of P. malariae infection, and its levels were positively correlated with parasitemia levels. Results of rapid diagnostic testing targeting pan-genus pLDH were, however, negative, despite the presence of symptoms consistent with malaria for 72 hours prior to testing. This is consistent with previous reports of poor assay performance and an approximately 30% detection rate [10,38,39]. Although pan-genus pLDH appears to have the appropriate characteristics for a screening test, with a proportional increase with the parasitemia level and a relatively prompt clearance after treatment, the lack of sensitive rapid diagnostic testing for this Plasmodium species represents a challenge to case detection, true estimates of disease burden, and elimination efforts. An increased CRP level with an increasing parasitemia level reflects a nonspecific host response but is consistent with the correlation between this biomarker and disease severity in other Plasmodium species [40]. P. malariae infection in our study demonstrated different biomarker dynamics to those described in P. falciparum infection, with a relatively delayed lag period in pLDH positivity after the first observation of parasitemia, as well as a more pronounced CRP signal [22]. While a larger sample is required to confirm these findings, such interspecies differences highlight the need for potentially different methods of detection. The high-throughput loop-mediated isothermal amplification assay has been demonstrated to be an effective and sensitive alternative detection method for other neglected Plasmodium species not readily identified by current rapid diagnostic tests, although application in P. malariae cases is limited, and field applications may be limited by cost [41].
Although exploratory assessment of the P. malariae transcription profile in vivo isolated only a limited amount of mRNA for analysis, this proof of concept offers a number of prospects for further study of parasite biology and enabled the reference genome annotation for P. malariae to be refined. We did not find evidence of expression among the pir genes, which have been shown to have a major role in virulence in some other species [42,43], or among the novel transmembrane protein families (fam-m or fam-l). Instead, other exported proteins, including PHISTs, were identified, suggesting these proteins may play a more prominent role in this species. Serial assessment of parasite transcription during the course of infection would also be informative and indicate whether antigenic variation or adhesion molecule upregulation are parasite strategies permitting the prolonged, latent infections that have been previously described [4]. The identification of putative gametocyte gene transcripts may enable the development of a targeted reverse transcription PCR probe for more-sensitive analysis of gametocyte dynamics, as has been developed for both P. falciparum and P. vivax [44].
In conclusion, we have outlined an experimental system of P. malariae infection that offers the potential to study this neglected tropical pathogen and subsequently accelerate the development of diagnostic tests, drugs, and vaccines. The logistic challenges and financial costs of field-efficacy studies are even greater for P. malariae than for the more virulent malaria parasite species, owing to relatively lower case numbers, challenges in case identification, and the relative frequency of coinfection. Nonetheless, the global effort to eradicate malaria necessitates an understanding of all infective species. The ability to undertake experimental study of P. malariae malaria in a dedicated clinical trial setting represents a significant resource, enhancing safety and enabling the design and conduct of well-controlled research.