Binding of human serum proteins to P. falciparum-infected erythrocytes and its association with malaria clinical presentation

The pathogenesis of Plasmodium falciparum malaria is related to the ability of parasite ‐ infected erythrocytes (IEs) to adhere to the vascular endothelium (cytoadhesion/sequestration) or to surrounding uninfected erythrocytes (rosetting). Both processes are mediated by the expression of members of the clonally variant PfEMP1 parasite protein family on the surface of the IEs. Recent evidence obtained with laboratory-adapted clones indicates that P. falciparum can exploit human serum factors, such as IgM and α 2 ‐ macroglobulin (α 2 M), to increase the avidity of PfEMP1-mediated binding to erythrocyte receptors, as well as to evade host PfEMP1-specic immune responses. It has remained unclear whether PfEMP1 variants present in eld isolates share these characteristics, and whether they are associated with clinical malaria severity. These issues were investigated here.


Abstract Background
The pathogenesis of Plasmodium falciparum malaria is related to the ability of parasite-infected erythrocytes (IEs) to adhere to the vascular endothelium (cytoadhesion/sequestration) or to surrounding uninfected erythrocytes (rosetting). Both processes are mediated by the expression of members of the clonally variant PfEMP1 parasite protein family on the surface of the IEs. Recent evidence obtained with laboratory-adapted clones indicates that P. falciparum can exploit human serum factors, such as IgM and α 2 -macroglobulin (α 2 M), to increase the avidity of PfEMP1-mediated binding to erythrocyte receptors, as well as to evade host PfEMP1-speci c immune responses. It has remained unclear whether PfEMP1 variants present in eld isolates share these characteristics, and whether they are associated with clinical malaria severity. These issues were investigated here.

Methods
Children 1-12 years reporting with P. falciparum malaria to Hohoe Municipal Hospital, Ghana were enrolled in the study. Parasites from children with uncomplicated (UM) and severe malaria (SM) were collected. Binding of α 2 M and IgM from non-immune individuals to erythrocytes infected by P. falciparum isolates from 34 children (UM and SM) were analysed by ow cytometry. Rosetting in the presence of IgM or α 2 M was also evaluated. Experimental results were analysed according to the clinical presentation of the patients.

Results
Clinical data from 108 children classi ed as UM (n=54) and SM cases (n=54) were analysed. Prostration, severe malaria anaemia, and hyperparasitaemia were the most frequent complications. Three children were diagnosed with cerebral malaria, and one child died. Parasite isolates from UM (n=14) and SM (n=20) children were analysed. Most of the eld isolates bound non-immune IgM (33/34), whereas the α 2 M-binding was less common (23/34). Binding of both non-immune IgM and α 2 M was higher but not signi cant in IEs from children with SM than from children with UM. In combination, IgM and α 2 M supported rosette formation at levels similar to that observed in the presence of 10% human serum.

Conclusions
The results support the hypothesis that binding of non-immune IgM and/or α 2 M to IEs facilitates rosette formation and perhaps contributes to P. falciparum malaria severity.

Background
Malaria continues to be an important public health problem in the developing world. Despite intensive global efforts, the number of malaria cases worldwide has not changed for the past four years. In 2018, an estimated 228 million cases of malaria and 405,000 deaths occurred worldwide, most of them in Africa. Notably, 67% of global deaths occurred in children aged under ve years [1]. The broad spectrum of malaria-related manifestations ranges from asymptomatic parasitaemia to severe, life-threatening disease. Of the species infecting humans, Plasmodium falciparum is responsible for the vast majority of clinical cases as well as for almost all severe morbidity and mortality [1].
The pathogenesis of P. falciparum malaria is in part related to adhesion of parasite-infected erythrocytes (IEs) to the vascular endothelium (cytoadhesion/sequestration) in various tissues [2,3] or to surrounding, uninfected erythrocytes (rosetting) [4,5]. Sequestration prevents the IEs destruction in the spleen [6], but can cause tissue in ammation and organ-speci c complications [7,8]. In both, cytoadhesion and rosetting, the IEs bind to host cell membrane receptors via a diverse family of parasite-encoded protein ligands called P. falciparum erythrocyte membrane protein 1 (PfEMP1) [9,10]. These proteins are displayed on multi-protein complexes ("knobs") protruding from the IE surface [11]. Each parasite genome contains ~60 PfEMP1-encoding var genes [12,13], but only a single PfEMP1 variant at a time is expressed on the surface of a given IE due to allelic exclusion [14]. However, the parasites can switch transcription among the different var genes to evade acquired PfEMP1-speci c immunity. The set of var genes varies substantially among P. falciparum genomes, creating a vast global repertoire of PfEMP1 proteins. PfEMP1 variants mediating formation of rosettes also bind to endothelial cells via distinct receptor-ligand interactions [15]. The receptor speci city is facilitated by the PfEMP1 secondary structure, with de ned domains mediating distinct cytoadhesion phenotypes, which in turn have been associated with discrete clinical presentations [9]. This family of proteins mediate IE adhesion to a range of host receptors, including CD36 [16], intercellular adhesion molecule 1 (ICAM-1) [17,18], endothelial protein C receptor (EPCR) [18,19], gC1qR [20], and oncofoetal chondroitin sulfate (a.k.a. CSA) [21].
PfEMP1 also binds soluble plasma factors. Several PfEMP1 variants can bind IgM via the Fcµ region of the antibody rather than by the hypervariable, antigen-speci c Fab fragment [22][23][24]. This type of IgM binding is sometimes called "non-immune". Recently, we documented that non-immune IgM-binding PfEMP1 proteins are frequent in P. falciparum laboratory clones [24,25], and that a2-macroglobulin (α 2 M), another abundant serum protein, also binds to PfEMP1 [26]. Whether PfEMP1 variants present in eld isolates share these binding characteristics, and whether those features are associated with the clinical presentation of malaria are unknown. Therefore, we analysed the non-immune IgM and α 2 M binding in parasites from Ghanaian children with uncomplicated and severe malaria.

Ethical statement
The study was approved by the Noguchi Memorial Institute for Medical Research Institutional Review Board (NMIMR STC Number: STC Paper 5(1) 2013-2014) and by the Ethical Review Committee of the Ghana Health Service (026/13-14). Declaration of free willingness to participate in the study and written informed consent was obtained from parents/guardians of all study participants prior to enrolment.

Study area
This study was carried out within the frame of a broader study aimed at building malaria vaccine research capacity in Ghana (MAVARECA https://mavareca.ku.dk/). The participants were enrolled in Hohoe, a town located in the Volta Region about 220 km northeast of Accra. Malaria transmission intensity in the area is high with approximately 65 infectious bites per person per year and has two seasonal peaks, a major one in April-July and a minor one in September-November [27].
Study participants and laboratory tests Children 1-12 years of age and reporting with P. falciparum malaria to Hohoe Municipal Hospital were enrolled in the study between July-August 2016. A sample size of 96 children was calculated based on 40,092 malaria cases con rmed at the Hohoe Municipal hospital in 2014 [28], using a level of con dence of 95%, sampling error of 10%, and 50% expected prevalence for IgM or α 2 M binding. After enrolment, a project nurse and a physician completed a standardised questionnaire and performed a clinical examination. Severe malaria (SM) was de ned according to the WHO criteria [29], and children were treated with artemether-lumefantrine or quinine IV as required. Venous blood samples were taken on the day of admission to determine haemoglobin levels (Hb), ABO blood group, and for research purposes. Samples were taken daily during the hospitalisation and one week after initial presentation to assess haemoglobin levels and parasitaemia. Sickle cell Hb phenotype was determined by electrophoresis and glucose-6-phosphate dehydrogenase (G6PDH) de ciency tested by methylene blue reduction test [30].
The rest of the washed pellet was gently mixed with glycerol freezing solution, aliquoted into cryotubes, and transferred to liquid nitrogen for long-term storage. Frozen samples were shipped to the University of Copenhagen, where they were thawed by standard methods before starting short-term in vitro cultures [31]. Brie y, the parasites were placed in 2% complete culture medium at 2% haematocrit and incubated at 37% in 2% O 2 , 5% CO 2 , and 93% N 2 atmosphere. Parasitaemia was checked daily by Giemsa-stained thin smears and those with normal morphology that matured to the late-trophozoite stage were included in the study. The day before use in experiments including serum proteins, the parasites were transferred to serum-free RPMI 1640 medium (0.5% AlbuMAX II).

Rosetting assays
The rosetting of fresh isolates was assessed in the rst cycle of in vitro growth when the majority of the parasites had reached the trophozoite-late stage. After staining the parasites with 7 μg/mL of ethidium bromide for two minutes, the percentage of rosettes was assessed by counting 200 ethidium bromidestained IEs, using wet preparations and uorescence microscopy. Rosettes were de ned as IEs having two or more adhering uninfected erythrocytes.
To determine the role of non-immune IgM and a 2 M binding in rosetting, short-term in vitro cultures of thawed cryostabilates were used. Late trophozoite stage IEs at 2% haematocrit in serum-free RPMI 1640 medium were incubated 1h at 37°C in 2% O 2 , 5% CO 2 , and 93% N 2 atmosphere with 10% NHS, or with 10 nM IgM (Sigma-Aldrich, Germany), 10 nM a 2 M (Sigma-Aldrich, Germany), or both. The percentage of rosettes was assessed as described above.

Measurements of non-immune IgM and α 2 M binding to PfEMP1
Non-immune IgM and a 2 M binding to IEs was detected by ow cytometry as previously described [24,26,32]. We have previously detected binding of both proteins at 10 nM in several laboratory clones [23][24][25][26], but included two additional concentrations (1nM and 100 nM) to explore the potential binding in the eld isolates. Brie y, intact and un xed late trophozoite stage IEs puri ed by magnet-activated cell sorting (MACS) were incubated with either 1, 10, or 100 nM non-immune human IgM or a 2 M in PBS supplemented with 1% Ig-free bovine serum albumin (PBS 1% BSA). IE-bound IgM was measured with a FITC-conjugated anti-human IgM (1:150; Sigma), whereas a 2 M was determined with a combination of goat polyclonal anti-human α 2 M (1:3,000; Abcam, UK) and FITC-conjugated anti-goat IgG (1:150; Vector, UK). Non-immune IgM and a 2 M binding to IEs was quanti ed as the median uorescence intensity (MFI) in IEs labelled with 10 µg/mL Hoechst. BD LSR II ow cytometer was used for data acquisition and FlowLogic software (Inivai Technologies, Australia) for data analysis.

Statistical analysis
Data were analysed and plotted using IBM SPSS Statistics for Macintosh, Version 26.0 (IBM Corp) and GraphPad Prism version 8.0 (GraphPad Software, San Diego, California, USA), respectively. Nominal variables were analysed using descriptive statistics. The Mann-Whitney U or Friedman test and Kruskal-Wallis test followed by Dunn's multiple comparison test were used to compare two and more-than-two groups, respectively. Spearman's rank correlation (r s ) was used to assess the correlation between numeric variables. Fisher's exact test was used to compare proportions. Any p-value < 0.05 was considered statistically signi cant.
Multiple linear regression models were used to evaluate the effect of potential confounders on the relationship between percentage of rosettes or protein binding (dependent variables) and relevant independent factors. Models were adjusted by clinical category (UM and SM), age, parasitaemia, haemoglobin levels at admission, ABO blood group, and days of sickness. 1-3 days). The parasitaemia at admission and before antimalarial treatment was determined in 87 children (81%). Most (71%) presented with > 10,000 parasites/μL (median: 37,367 parasites/µL; IQR: 857-95,822 parasites/µL). Anaemia (Hb <11 g/dL) and severe anaemia (Hb <5 g/dL) were observed in 69% and 8% of the children, respectively. On day seven post-admission, 83% of the children had anaemia, but none of them had severe anaemia. Sixteen SM children received blood transfusion. Some demographic and clinical variables differed between children with UM and SM (Table 1). An additional diagnosis at admission was more frequent in SM than in UM. Acute tonsillitis, gastroenteritis, and respiratory tract infections were the most frequent in both groups. Twenty-one SM children had only one criterion for classi cation (Fig. 1). Cerebral malaria was observed in three children; it and was the only criterion in two of them. Neither severe bleeding nor pulmonary oedema were registered. One death was reported one day after enrolment, in a child with clinical shock, prostration, and hyperparasitaemia.

Rosetting ex vivo
Rosetting was assessed in 61 fresh isolates. The remaining samples (n = 47) were excluded either because the parasites failed to mature beyond ring stage in vitro or because the parasitaemia was too low for reliable assessment. Besides technical issues in the eld with samples processing, a probable explanation is that children had taken antimalarial drugs, and the parasites did not thrive overnight. Indeed, 23 out the 108 children reported taking anti-malarial drugs prior to admission (Table 1). Although most of the eld isolates formed rosettes (79%), the percentage of rosettes was low (median: 7%; range 2-64%) and not different between children with UM and SM ( Fig. 2A). However, the percentage of rosettes was negatively correlated with age (r s = -0.373; 95% CI -0.577 to -0.128; p = 0.003), even after adjusting for potential confounders such as clinical category, parasitaemia, and ABO blood group (b= -0.352; p = 0.013).

IgM-and α 2 M-binding frequencies in UM and SM
Thirty-nine frozen parasite samples were thawed and put into short-term culture, with 87% successful recovery. After one to three cycles, late trophozoite stage IEs from 34 children were puri ed and used to evaluate binding of non-immune human IgM and α 2 M by ow cytometry. Most of the eld isolates bound non-immune IgM (33/34), and the binding was dependent of the concentration (p < 0.0001; Friedman test). In contrast, the α 2 M-binding was less common than IgM (23/34; p = 0.003, Fisher's exact test), without signi cant effect of increasing the concentration (p = 0.09; Friedman test) (Fig. 3A). At 100 nM, both non-immune IgM and a2M-binding correlated positively (r s = 0.939; p < 0.001), and the association remained after adjusting for potential confounders such as clinical category, age, parasitaemia, haemoglobin levels at admission, and ABO blood group (b= 0.921; p < 0.001) (Fig. 2B).
The samples were separated into UM and SM to evaluate the association of serum protein binding with the clinical classi cation. At all concentrations, binding of non-immune IgM and α 2 M was higher in IEs from SM than from UM patients, although not signi cant at the chosen level of signi cance (Fig. 4). Nonimmune IgM binding was negatively correlated with haemoglobin levels (r s = -0.411, p = 0.016; and r s = -0.445, p = 0.009 at 10 and 100 nM, respectively). The association was not signi cant after adjusting for age.

IgM and α 2 M support rosetting in the absence of human serum
Rosetting is a highly variable phenotype, which in many cases depends on serum factors [33]. We therefore evaluated impact of non-immune IgM and α 2 M on rosetting in our parasite eld isolates after short in vitro culture of frozen samples. Rosetting did not occur or was very low in a serum-free culture medium (0.5% AlbuMAX II), and addition of 10 nM non-immune IgM to AlbuMAX medium had no effect. In contrast, the presence of 10 nM α 2 M supported rosette formation in some isolates, and in combination, IgM and α 2 M, supported rosette formation at levels similar to that observed in the presence of 10% human serum (p = 0.86, Mann-Whitney test). Further analysis according to the clinical classi cation, showed that the percentage of rosettes in the presence of both IgM and α 2 M, or NHS was higher, although not signi cantly, in parasites from SM children than from UM children (p = 0.36 and p = 0.26, respectively, Mann-Whitney test). In eld isolates from SM but not UM, the percentage of rosettes using IgM plus α 2 M or NHS was signi cantly higher than the observed in AlbuMAX medium (Fig. 5). The combined effect of IgM plus α 2 M in rosetting appeared additive under the conditions used here, rather than the synergistic effect previously reported for a laboratory clone [26].

Discussion
The pathogenesis of P. falciparum malaria is related to the ability of the IEs to adhere to the vascular endothelium [2,3] and to form rosettes [4,5,34]. Both processes are mediated by members of the PfEMP1 family proteins expressed on the IE surface [9,10,12,13]. Based on evidence obtained with parasite clones adapted to long-term in vitro culture and expressing a speci c PfEMP1 variant, we have previously hypothesised that P. falciparum exploits non-immune IgM to strengthen the low-a nity interactions of PfEMP1 with the carbohydrate receptors on uninfected erythrocytes in rosetting [23]. Moreover, we have documented that IgM-binding PfEMP1 proteins are common in each of the three laboratory clones studied (3D7, HB3, and IT4/FCR3) [24,25], although not all of them mediated rosetting [24]. Similarly, we have identi ed α 2 M as an important facilitator of rosetting that can bind at least four PfEMP1 molecules per α 2 M molecule, which might explain its effect on rosetting [26]. Remarkably, non-immune IgM and a 2 M appear to bind to the same domain in the laboratory clone HB3VAR06, where they synergistically facilitate rosetting [26].
To verify the above hypothesis, and assess its potential importance in P. falciparum malaria pathogenesis, we evaluated the binding of non-immune IgM and α 2 M to erythrocytes infected by P. falciparum eld isolates obtained from Ghanaian children with UM or SM. Likewise, the rosetting in the presence of these serum proteins was tested. Using both approaches, we found that PfEMP1 expressed on eld isolate IEs can bind both non-immune IgM and a2M, and that both proteins allow rosetting at similar levels observed in the presence of human serum. Our data suggest that binding of non-immune IgM and α 2 M is higher in SM than in UM, consistent with a role for these phenotypes in malaria pathogenesis. However, additional studies are needed to validate this hypothesis further because the differences were not statistically signi cant. This may be related to the fact that cerebral malaria was scarce in our study area, combined with the particular role for rosetting in CM [35,36].
That apart, it has to be acknowledged that most of the IgM-binding PfEMP1 proteins identi ed to date are encoded by var genes belonging to group B or Group C, which are commonly found in UM and asymptomatic infections [25]. Contrary to IgM-binding, which seems to be a frequent phenotype in laboratory clones [25], α 2 M-binding is rare and restricted to DBLe and DBLz domains (Lopez-Perez et al., unpublished data, [26]). Binding of α 2 M and non-immune IgM correlates in both eld and laboratory isolates, supporting the idea that both proteins bind to the same PfEMP1 variants. The dose-response effect observed with IgM, but not with α 2 M, it may be related to saturation. Each pentameric IgM molecule can bind two PfEMP1 molecules [23], whereas α 2 M can bind at least four [26]. The few α 2 M-binding variants to be expected among the heterogeneous PfEMP1 expressed by eld isolates might easily be saturated.
The low ex vivo percentage of rosettes observed here, and the absence of the association with SM, may be related to the low number of children with cerebral malaria or severe anaemia [35,36]. Nevertheless, the percentage of rosettes were negatively correlated with age. This is consistent with the early in life acquisition of antibodies targeting PfEMP1 variants responsible for rosetting, thus facilitating the clearance of those IEs and contributing to protective immunity to SM. Indeed, an association between higher anti-rosetting activity and age was reported in Kenyan children [37]. The differences between the eld isolates tested probably re ects the high degree of PfEMP1 diversity [38], with expression of several variants within a sample collected from a single individual in contrast with a single PfEMP1 variant in most of the laboratory clones. We did not measure the IgM or α 2 M levels in the NHS used for the rosetting assays. However, we have reported that α 2 M levels in IgM-depleted serum is too low to support rosetting alone, and requires the presence of IgM for rosetting to occur [26].

Conclusions
The binding of α 2 M to IEs was uncommon, but both non-immune IgM and α 2 M-binding were slightly higher in IEs from children with SM than those with UM, although not signi cant at the chosen level of signi cance. Moreover, both proteins, α 2 M and IgM, appear to facilitate rosetting, particularly in isolates from children with SM.   Figure 1 Distribution of severe malaria subcategories. Children were classi ed as having a single or more than one criterion. SMA: severe malaria anaemia (Hb < 5 g/dL); hyperparasitaemia (> 10%); cerebral malaria (coma with a Blantyre Coma Score of ≤ 2); multiple convulsions without coma; jaundice (plasma bilirubin >3 mg/dL); renal impairment (creatinine > 3.0 mg/mL or urea > 20 mmol/L). More than one criterion includes children with clinical shock (< 70 mm Hg) and hypoglycaemia (< 40 mg/dL).