Stringent Selection of Knobby Plasmodium falciparum-Infected Erythrocytes during Cytoadhesion at Febrile Temperature

Changes in the erythrocyte membrane induced by Plasmodium falciparum invasion allow cytoadhesion of infected erythrocytes (IEs) to the host endothelium, which can lead to severe complications. Binding to endothelial cell receptors (ECRs) is mainly mediated by members of the P. falciparum erythrocyte membrane protein 1 (PfEMP1) family, encoded by var genes. Malaria infection causes several common symptoms, with fever being the most apparent. In this study, the effects of febrile conditions on cytoadhesion of predominately knobless erythrocytes infected with the laboratory isolate IT4 to chondroitin-4-sulfate A (CSA), intercellular adhesion molecule 1 (ICAM-1), and CD36 were investigated. IEs enriched for binding to CSA at 40 °C exhibited significantly increased binding capacity relative to parasites enriched at 37 °C. This interaction was due to increased var2csa expression and trafficking of the corresponding PfEMP1 to the IE surface as well as to a selection of knobby IEs. Furthermore, the enrichment of IEs to ICAM-1 at 40 °C also led to selection of knobby IEs over knobless IEs, whereas enrichment on CD36 did not lead to a selection. In summary, these findings demonstrate that knobs are crucial for parasitic survival in the host, especially during fever episodes, and thus, that selection pressure on the formation of knobs could be controlled by the host.


Static Binding Assay and CSA Inhibition Assay
A static binding assay was performed to verify the binding capacity of enriched parasites. Seventy-two h before the assay, parasites were synchronized. On the same day, HBEC-5i and CHO CD36 cells were seeded on 0.1% gelatin-coated 13 mm coverslips at a density of 1.2 × 10 6 cells/mL. On the day of the assay, confluence was assessed with an inverted microscope. Ideally, the cells formed a monolayer. For inhibition assays, HBEC-5i and CHO CD36 cells were pre-incubated with 10 µg/mL soluble CSA (sCSA; Sigma-Aldrich Merck, Darmstadt, Germany) at 37 • C for 30 min. Trophozoite-stage parasites (5% parasitemia, 1% hematocrit) were resuspended in binding medium, and the suspension was added to the cell monolayers and incubated for 1 h at 37 • C or 40 • C under 5% CO 2 with gentle orbital shaking every 15 min. Subsequently, coverslips were gently washed with binding medium to remove non-bound IEs. The remaining IEs were fixed with 1% glutaraldehyde in phosphate-buffered saline (PBS; 0.14 M NaCl, 0.3 mM KCl, 8 mM NaH 2 PO 4 , 1.5 mM KH 2 PO 4 , pH 7.4) for 30 min at room temperature (RT). Fixed cells were stained with a filtered Giemsa/Weisser buffer solution (1:10) for another 30 min. The number of adherent IEs was determined by counting the number of bound IEs on 300 HBEC-5i, CHO CD36 , or CHO ICAM-1 cells under a light microscope. Assays were performed at least two times in triplicate.

DNA and RNA Purification, Library Preparation, and Transcriptome Analysis
Total DNA was isolated using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. For RNA isolation, parasites were synchronized 48 h prior to harvest. Ring-stage IEs were rapidly lysed in a 10-times higher volume of pre-warmed 37 • C TRIzol (Invitrogen, Thermo Fisher Scientific, Bremen, Germany) followed and incubated for 5 min at 37 • C. Subsequently, the samples were stored at -80 • C. RNA was isolated using a PureLink RNA Mini Kit (Thermo Fisher Scientific, Bremen, Germany) according to the manufacturer's instructions. Genomic DNA contamination was removed using the TURBO DNA-free Kit (Invitrogen, Thermo Fisher Scientific, Bremen, Germany) followed by a magnetic bead enzymatic wash using Agencourt RNAClean XP (Beckman Coulter, Krefeld, Germany). The concentration and quality of isolated RNA were assessed using an Agilent 2100 Bioanalyser System with the Agilent RNA 6000 Pico Kit (Agilent Technologies, Ratlingen, Germany). The RNA was sent to BGI (Shenzhen, China), where RNAseq was performed using the Illumina HiSeq 4000 PE100 platform (approximately 11 M PE reads per samples). Reads were trimmed and filtered using Trimmomatic [34], and aligned to IT4 genome data available at PlasmoDB [35] using RSEM [36] and Bowtie2 [37] software. Differential expression was tested using DEseq for normalization of the row reads [38]. P-values were adjusted using Holm's method.

Trypsin Assay and Western Blot
To measure surface-exposed Pf EMP1, a trypsin cleavage assay was performed using synchronized trophozoite-stage parasites (24-28 h after erythrocyte infection). IEs were isolated with magnetic cell sorting (CS columns; Miltenyi Biotec, Bergisch Gladbach, Deutschland). Subsequently, the cell count was set to 1 × 10 6 cells/µL and split into two fractions, with and without addition of 1 µg/mL trypsin for 30 min at 37 • C. Both fractions were resuspended in 10 mM HEPES buffer (pH 7.2) and lysed using three freeze-thaw cycles in liquid nitrogen. The supernatant and pellet were separated by centrifugation at 20,000× g for 10 min at 4 • C. The supernatant was removed, and the pellet containing the membrane fraction was washed twice with PBS. For western blotting, cell membranes were lysed in 2× Laemmli buffer and heated for 5 min at 95 • C. An equivalent of 1 × 10 7 cells/lane were separated on a 6% SDS gel at 400 mA and blotted onto nitrocellulose membranes. Uniform loading of the gels was confirmed by staining a second gel in parallel with Coomassie Blue. Subsequently, membranes were blocked with 5% milk in TBS (0.3 M NaCl, 20 mM Tris pure, pH 7.5) for 30 min. Primary antibodies were diluted in 2.5% dry milk/TBS as follows: mouse anti-ATS-GHI-monoclonal 1:500 (WEHI Antibody Facility, Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia); rabbit anti-spectrin, 1:2000 (Sigma-Aldrich Merck, Darmstadt, Germany); rabbit anti-KAHRP, 1:4000 (a kind gift from Prof. Brian Cooke, Monash University, Melbourne, Australia). Membranes were incubated with primary antibodies overnight at 4 • C. The secondary antibodies were diluted in 5% dry milk/TBS as follows: rabbit anti-mouse 1:5,000 (Agilent Dako, Santa Clara, CA, USA); porcine anti-rabbit 1:10,000 for spectrin and 1:20,000 for KAHRP blots (Agilent Dako, Santa Clara, CA, USA). Membranes were incubated with secondary antibodies for 2 h at RT. The chemiluminescent signal of the HRP-coupled secondary antibodies was visualized on an Amersham Hyperfilm-ECL (GE Healthcare, Freiburg, Germany). Densitometric quantification was performed with ImageJ 1.52a and Adobe Photoshop CS3 Extended (version 10.0.1).

Quantitative Real-Time PCR
For qPCR experiments, sense and antisense primers were designed to amplify 100-120 bp fragments of the target genes. The oligonucleotides used for PCR are listed in Table S1. After RNA isolation, cDNA was synthesized using SuperScript II Reverse Transcriptase (Invitrogen, Thermo Fisher Scientific, Bremen, Germany) and random hexamers (Invitrogen) at 50 • C for 1 h. The cDNA/gDNA template was mixed with SYBR Green PCR Master Mix (QuantiTect SYBR Green PCR Systems; Qiagen, Hilden, Germany) and 0.5 µM forward and reverse primer at a final volume of 10 µl. Reactions were incubated at 95 • C for 15 min, and then subjected to 40 cycles of 95 • C for 15 s and 60 • C for 1 min, and a subsequent melting step (60-95 • C). The specificity of each primer pair was confirmed after each qPCR run using melt curve analysis. The conserved ring-stage expressed gene sbp1 (skeleton-binding protein 1), and the housekeeping genes fructose-bisphosphate aldolase and arginyl-tRNA synthetase were used to normalize var gene expression (as a control, a single RNA biological sample isolated from IT4 CSA EC37 • and IT4 CSA EC40 • was used to verify qPCR results). To compare var gene expression levels in each sample, relative gene expression was calculated for each individual var gene relative to the geometric mean of the three normalizers after calculation of primer efficiency [39,40]. Statistical significance was tested using the Wilcoxon rank sum test (GraphPad Prism 7).

Cytoadhesion of Infected Erythrocytes to HBEC-5i Cells Was Mediated by VAR2CSA
After long-term cultivation of P. falciparum IT4 isolate, IEs were rarely capable of adherence to HBEC-5i cells. To identify the effects of febrile temperature on cytoadhesion of P. falciparum to HBEC-5i cells, the IT4 isolate (IT4 NE ; non-enriched) was enriched for binding to HBEC-5i cells at 37 • C and 40 • C. After 5-7 consecutive rounds of enrichment, saturation of the binding capacity was observed. To identify the var genes expressed specifically in the HBEC-5i-enriched populations, the transcriptome profiles of the parasite populations were determined by RNAseq and evaluated, focusing on var gene expression in ring-stage parasites (6-10 h post-infection) ( Figure 1A and Tables S2-S5). The IT4 NE parasite population expressed a mixture of var genes. Most parasites expressed group C IT4_var34 (79%) and group A IT4_var35 (6%), followed by IT4_var23 (3%) and IT4_var59 (2%). All other var genes were sparsely expressed, representing 1.5% or less of all var transcripts. IT4 parasites enriched for HBEC-5i binding significantly overexpressed group E IT4_var04, also known as var2csa ( Figure 1A and Tables S2-S5). At physiological as well as febrile temperatures, expression of var2csa was dominant (representing 93% and 97% of all var transcripts, respectively). While both parasite populations only expressed var2csa, the number of transcripts for var2csa nearly doubled at 40 • C (relative read count, 94134) compared with 37 • C (relative read count, 49285) ( Figure 1A and Table  S2). This result could also be confirmed with the help of qPCR (Table 1). CSA is the binding partner of VAR2CSA [16,17]. To determine whether the interaction partner on the HBEC-5i cells was, in fact, CSA or another receptor, a static cytoadhesion assay was performed using trophozoite-stage IEs and HBEC-5i cells in the absence and presence of soluble CSA (sCSA) at 37 • C. sCSA did not block binding of IEs to HBEC-5i cells because IT4 NE parasites capable of binding to CD36 adhered at similar levels when pretreated with sCSA (250 ± 101 and 199 ± 79 IEs/100 CHO CD36 cells) ( Figure 1B). However, pre-incubation of physiological and febrile temperature HEBC-5i-enriched parasite populations with sCSA completely abolished cytoadherence to HBEC-5i cells. Interestingly, the binding capacity of IT4 parasites enriched for binding to HEBC-5i cells at 40 • C (338 ± 150 bound IEs/100 HBEC-5i cells) was significantly increased compared with the binding capacity at 37 • C (70 ± 44 bound IEs/100 HBEC-5i) ( Figure 1B). Because CSA is the binding partner of IEs on HEBC-5i cells, the HEBC-5i enriched parasite populations are subsequently referred to as IT4 CSA EC37 • (enriched to HBEC-5i cells at 37 • C) and IT4 CSA EC40 • (enriched to HBEC-5i cells at 40 • C).
Microorganisms 2019, 7, x FOR PEER REVIEW 6 of 18 40 °C (relative read count, 94134) compared with 37 °C (relative read count, 49285) ( Figure 1A and Table S2). This result could also be confirmed with the help of qPCR (Table 1). CSA is the binding partner of VAR2CSA [16,17]. To determine whether the interaction partner on the HBEC-5i cells was, in fact, CSA or another receptor, a static cytoadhesion assay was performed using trophozoite-stage IEs and HBEC-5i cells in the absence and presence of soluble CSA (sCSA) at 37 °C. sCSA did not block binding of IEs to HBEC-5i cells because IT4 NE parasites capable of binding to CD36 adhered at similar levels when pretreated with sCSA (250 ± 101 and 199 ± 79 IEs/100 CHO CD36 cells) ( Figure 1B). However, pre-incubation of physiological and febrile temperature HEBC-5i-enriched parasite populations with sCSA completely abolished cytoadherence to HBEC-5i cells. Interestingly, the binding capacity of IT4 parasites enriched for binding to HEBC-5i cells at 40 °C (338 ± 150 bound IEs/100 HBEC-5i cells) was significantly increased compared with the binding capacity at 37 °C (70 ± 44 bound IEs/100 HBEC-5i) ( Figure 1B). Because CSA is the binding partner of IEs on HEBC-5i cells, the HEBC-5i enriched parasite populations are subsequently referred to as IT4 CSA EC37° (enriched to HBEC-5i cells at 37 °C) and IT4 CSA EC40° (enriched to HBEC-5i cells at 40 °C).    To determine if increased var2csa expression in IT4 CSA EC40 • relative to IT4 CSA EC37 • was solely due to the temperature increase, IT4 NE parasites were incubated once weekly (over 5 weeks) at 38.5 • C and 40 • C for 2 h without contact with HBEC-5i cells. Neither var2csa-specific upregulation nor upregulation of other selected var genes was observed (Table 1). A trypsin cleavage assay revealed that increased var2csa expression correlated with increased protein quantity of VAR2CSA (Figure 2A). The extracellular domain of Pf EMP1 molecules can be cleaved using trypsin, and the remaining intracellular acidic terminal segment (ATS) region (70-90 kDa) can be detected by western blot using an ATS-specific antibody. The amount of intracellular Pf EMP1 was 1.4-(no trypsin treatment) and 3.6-(with trypsin treatment) times higher in IT4 CSA EC40 • relative to IT4 CSA EC37 • . The IT4 CSA EC40 • population exhibited a 90 kDa protein band after trypsin digestion, but no protein could be detected in the IT4 CSA EC37 • population. To determine if surface Pf EMP1 was present at any level in the 37 • C enriched population, another western blot with 3-times higher parasite lysate load (3 × 10 7 IEs) was performed, revealing that IT4 CSA EC37 • exhibited detectable surface Pf EMP1 (Figure 2A and Figure  S1). In the sample of IT4 NE parasites containing a mixture of several Pf EMP1s, a 70 kDa protein band was detected after trypsin digestion. The reasons for the formation of a 90 kDa fragment after cleavage of VAR2CSA from the IT4 CSA EC37 • and IT4 CSA EC40 • populations, and that of a 70 kDa fragment after cleavage of Pf EMP1 from the whole IT4 NE population (Figure 2A) remain unknown. To determine if increased var2csa expression in IT4 CSA EC40° relative to IT4 CSA EC37° was solely due to the temperature increase, IT4 NE parasites were incubated once weekly (over 5 weeks) at 38.5 °C and 40 °C for 2 h without contact with HBEC-5i cells. Neither var2csa-specific upregulation nor upregulation of other selected var genes was observed (Table 1). A trypsin cleavage assay revealed that increased var2csa expression correlated with increased protein quantity of VAR2CSA ( Figure  2A). The extracellular domain of PfEMP1 molecules can be cleaved using trypsin, and the remaining intracellular acidic terminal segment (ATS) region (70-90 kDa) can be detected by western blot using an ATS-specific antibody. The amount of intracellular PfEMP1 was 1.4-(no trypsin treatment) and 3.6-(with trypsin treatment) times higher in IT4 CSA EC40° relative to IT4 CSA EC37°. The IT4 CSA EC40° population exhibited a 90 kDa protein band after trypsin digestion, but no protein could be detected in the IT4 CSA EC37° population. To determine if surface PfEMP1 was present at any level in the 37 °C enriched population, another western blot with 3-times higher parasite lysate load (3 × 10 7 IEs) was performed, revealing that IT4 CSA EC37° exhibited detectable surface PfEMP1 (Figures 2A and S1). In the sample of IT4 NE parasites containing a mixture of several PfEMP1s, a 70 kDa protein band was detected after trypsin digestion. The reasons for the formation of a 90 kDa fragment after cleavage of VAR2CSA from the IT4 CSA EC37° and IT4 CSA EC40° populations, and that of a 70 kDa fragment after cleavage of PfEMP1 from the whole IT4 NE population (Figure 2A) remain unknown.  Figure S1). Comparison of the relative amounts of protein was normalized to total protein amount using a corresponding Coomassie-stained gel ( Figure S2). (B) Western blotting for KAHRP was performed with the IE membrane fraction, using α-KAHRP and α-spectrin as controls.  Figure S1). Comparison of the relative amounts of protein was normalized to total protein amount using a corresponding Coomassie-stained gel ( Figure S2). (B) Western blotting for KAHRP was performed with the IE membrane fraction, using α-KAHRP and α-spectrin as controls.

IE Enrichment to HBEC-5i Cells at Elevated Temperature Led to Strong Selection of Knobby Infected Erythrocytes
To determine if other factors in addition to the var gene expression and amount of VAR2CSA promoted IE binding capacity, the transcriptome profiles of ring-stage IT4 CSA EC37 • and IT4 CSA EC40 • were investigated (Tables S3-S5). In total, seven of the 17 genes that expressed 5-fold or more in IT4 CSA EC40 • relative to IT4 CSA EC37 • were known to be associated with knob formation ( Table 2). Of these seven genes, four genes are located in the subtelomeric region of chromosome 2, including kahrp, which codes for the main component of knobs, KAHRP. Increased kahrp expression was verified at the protein level. KAHRP could only be detected in IT4 CSA EC40 • parasite lysates, while all other lysates were KAHRP-negative ( Figure 2B). To further determine whether KAHRP was associated with the presence of knobs in the IT4 CSA EC40 • population, electron microscopy was performed (Figure 3). In accordance with the transcriptome data and protein level observations, all non-enriched (IT4 NE , n = 318) and all 37 • C-enriched (IT4 CSA EC37 • , n = 138) parasites exhibited no obvious aggregation of electron-dense material close to the IE surface. By contrast, electron-dense truncation was observed in the parasite population enriched at 40 • C (IT4 CSA EC40 • ). All 268 quantified IEs from this treatment group exhibited knobs. To determine if occurrence of knobby IEs was promoted by heat shock alone, or whether a combination of heat shock and close proximity to HBEC-5i cells was essential, IT4 NE parasites were heat-shocked at 38.5 • C and 40 • C for 2 h weekly over a period of 5 weeks, and subsequently analyzed by electron microscopy. No knob-like formations were observed, indicating that a combination of heat shock and co-incubation with HBEC-5i cells was required to obtain knobby IEs (Figure 3). These results were also confirmed at the RNA level. For kahrp and pfemp3, as well as for five var genes including var2csa, no increase in expression was observed in the heat-shocked IT4 NE population relative to control (Table 1). To determine if cytoadherence induced knob formation at febrile temperatures, or whether the enrichment process at 40 • C led to selection of knobby IEs that could be present in very small numbers in the original parasite culture, the normalized copy numbers of kahrp and pfemp3 in IT4 CSA EC37 • and IT4 CSA EC40 • were determined. In IT4 CSA EC37 • parasites, the copy number of both genes was slightly increased (3.9-fold for kahrp and 3.3-fold for pfemp3) relative to IT4 NE . However, a strong increase in copy number was observed in the IT4 CSA EC40 • population, with a 330-fold upregulation of kahrp and a 99-fold upregulation of pfemp3 relative to IT4 NE (Table 1). Taken together, these results demonstrated that during the enrichment process on CSA at febrile temperature, knobby IEs are selected over knobless IEs.

Enrichment of IEs on CHO ICAM-1 but not CHO CD36 Cells at Febrile Temperature Led to Selection of Knobby Infected Erythrocytes
We further verified if the selection of knobby IEs was restricted only to the VAR2CSA-CSA interaction. IT4 NE parasites were enriched at 37 °C and 40 °C on transgenic CHO-745 cells that exposed ICAM-1 or CD36 on their surface (CHO ICAM-1 , CHO CD36 ). In accordance with the results of IE enrichment to CSA at 40 °C, enrichment of IEs on ICAM-1 led to selection of knobby IEs (IT4 ICAM-1 CHO ICAM-1 40°). Knobs were detected in 47% (n = 235) of the IT4 ICAM-1 CHO ICAM-1 40° population ( Figure 4A). On the other hand, enrichment of IEs to CD36 at 40 °C did not lead to selection of knobby IEs ( Figure 4B). These results could also be confirmed by measuring expression levels of kahrp and pfemp3. Only in the IT4 ICAM-1 CHO ICAM-1 40° population, and not in IT4 CD36 CHO CD36 40°, was a significant increase in expression (241-fold for kahrp and 56-fold for pfemp3) relative to control (Table 3). Increased expression of var01 and var16 indicated that the IT4 ICAM-1 CHO ICAM-1 40° population, was specifically enriched to ICAM-1, as VAR01 and VAR16 are known binding partners for ICAM-1 (Table 3) [33].

Enrichment of IEs on CHO ICAM-1 but not CHO CD36 Cells at Febrile Temperature Led to Selection of Knobby Infected Erythrocytes
We further verified if the selection of knobby IEs was restricted only to the VAR2CSA-CSA interaction. IT4 NE parasites were enriched at 37 • C and 40 • C on transgenic CHO-745 cells that exposed ICAM-1 or CD36 on their surface (CHO ICAM-1 , CHO CD36 ). In accordance with the results of IE enrichment to CSA at 40 • C, enrichment of IEs on ICAM-1 led to selection of knobby IEs (IT4 ICAM-1 CHO ICAM-1 40 • ). Knobs were detected in 47% (n = 235) of the IT4 ICAM-1 CHO ICAM-1 40 • population ( Figure 4A). On the other hand, enrichment of IEs to CD36 at 40 • C did not lead to selection of knobby IEs ( Figure 4B). These results could also be confirmed by measuring expression levels of kahrp and pfemp3. Only in the IT4 ICAM-1 CHO ICAM-1 40 • population, and not in IT4 CD36 CHO CD36 40 • , was a significant increase in expression (241-fold for kahrp and 56-fold for pfemp3) relative to control (Table 3). Increased expression of var01 and var16 indicated that the IT4 ICAM-1 CHO ICAM-1 40 • population, was specifically enriched to ICAM-1, as VAR01 and VAR16 are known binding partners for ICAM-1 ( Table 3) [33].

Discussion
A common symptom of a malaria infection is fever, but most studies of cytoadhesion have been performed at normal body temperature. The present study revealed the impact of febrile temperature on the binding phenotype of P. falciparum IEs to CSA on immortalized HBEC-5i human brain endothelial cells, as well as ICAM-1 and CD36 exposed on transgenic CHO cells.
HBEC-5i cells were derived from a human cerebral cortex and, therefore, exhibited major features of cerebral endothelial cells. Prior studies established that in addition to several receptors such as ICAM-1, VCAM-1, and VE-cadherin, CSA can also be detected on the surface of HEBC-5i cells (ATCC ® CRL-3245 ™ ). The present study demonstrated that enrichment of IEs on HEBC-5i cells produced a parasite population mainly expressing var2csa, which encodes VAR2CSA, the binding partner of CSA [7,8,47]. Binding of the enriched parasite population to HEBC-5i cells was abrogated

Discussion
A common symptom of a malaria infection is fever, but most studies of cytoadhesion have been performed at normal body temperature. The present study revealed the impact of febrile temperature on the binding phenotype of P. falciparum IEs to CSA on immortalized HBEC-5i human brain endothelial cells, as well as ICAM-1 and CD36 exposed on transgenic CHO cells.
HBEC-5i cells were derived from a human cerebral cortex and, therefore, exhibited major features of cerebral endothelial cells. Prior studies established that in addition to several receptors such as ICAM-1, VCAM-1, and VE-cadherin, CSA can also be detected on the surface of HEBC-5i cells (ATCC ® CRL-3245 ™ ). The present study demonstrated that enrichment of IEs on HEBC-5i cells produced a parasite population mainly expressing var2csa, which encodes VAR2CSA, the binding partner of CSA [7,8,47]. Binding of the enriched parasite population to HEBC-5i cells was abrogated in the presence of sCSA, demonstrating that in this context, CSA, and not another receptor, was the binding partner for CSA. These findings implied that CSA was dominant on HEBC-5i cells, and that binding of IEs to other receptors such as ICAM-1 was shielded. Endothelial cells have been demonstrated to expose high levels of CSA in prior studies [48,49]. In this context, it is also likely that the properties of CSA-mediated cytoadhesion are consistent with the initiation of obstruction in microvessels of the internal organs [48]. Additionally, VAR2CSA binds different types of tumor cells and tissues of epithelial, mesenchymal, and hematopoietic origin, which also expose CSA on their surface [21,22]. In contrast to the results described in the present study, Claessens et al. identified it4_var07 and it4_var19 as the primarily expressed var genes after enrichment of IEs to HEBC-5i cells [50]. Why enrichment to the same immortalized cell line led to different results is unclear at present. Different culture conditions could potentially influence expression of genes encoding different receptors. In connection with our study, the HBEC-5i cells are a suitable model to investigate the interaction of IEs with CSA. Interestingly, parasites enriched to HEBC-5i cells at 40 • C (IT4 CSA EC40 • ) exhibited a 2-fold upregulation of var2csa relative to parasites enriched at 37 • C (IT4 CSA EC37 • ), resulting in increased cell surface Pf EMP1 protein levels. The change in var gene expression was not solely due to heat shock, as var upregulation did not occur when parasites were only exposed to febrile temperature in the absence of HEBC-5i cells. In addition, approximately 5-times as many IEs of the IT4 CSA EC40 • population bound to the HBEC-5i cells compared with the IT4 CSA EC37 • population. These results are in accordance with previous studies, which identified heat-induced enhancement of binding capacity [28,30]. However, it is unlikely that the increased expression has such a strong influence on the binding capacity.
In addition to the amount of Pf EMP1 on IE surfaces, additional factors can affect the interaction of IEs with ECRs. Pf EMP1 proteins are anchored to parasitic structures known as knobs. Knobs are submembranous protrusions that function as tight junctions between IEs and endothelial cells [51]. Numerous studies identified KAHRP, Pf EMP3, Pf EMP2 (MESA), ring-infected erythrocyte antigen (RESA), and Pf 332 as important proteins for knob formation [23,52,53]. Alampalli et al. identified new knob constituents, including elongation factor 1 alpha, acyl-CoA synthetase, and some hypothetical proteins [46]. In addition, as mentioned above, members of the PHIST protein family also play crucial roles in binding Pf EMP1 co-migrating to knob structures [24]. Numerous genes important for knob formation, such as kahrp and pfemp3, are localized to the subtelomeric region of chromosome 2. Transcriptomic analysis of IT4 CSA EC40 • and IT4 CSA EC37 • parasites revealed that six genes located in the subtelomeric region of chromosome 2 are expressed in IT4 CSA EC40 • but not IT4 CSA EC37 • parasites, four of which are known to be associated with knob formation. In addition to kahrp and pfemp3, knob-associated heat shock protein 4 and PHISTb domain-containing RESA-like protein 1 are also present in this region [43,44]. The other differentially expressed genes within this locus are DnaJ, a putative chaperone protein, and a Plasmodium exported protein of unknown function. Referring to Alampalli et al., in P. falciparum, isolate 3D7 and acyl-CoA synthetase, a glycophorin binding protein, and several PHIST-proteins were also upregulated [46]. However, it is not clear if these genes were homologous to the genes discovered in the IT4 isolate in this study. Nevertheless, this transcriptional data provides strong supporting evidence that knob formation occurred in IT4 CSA EC40 • but not in IT4 CSA EC37 • parasites.
Because KAHRP is essential for knob formation [54], we next sought to verify the transcriptional kahrp expression data on the protein level. KAHRP protein could only be detected in IT4 CSA EC40 • parasites, consistent with the transcriptional data. Electron microscopic analysis revealed that the presence of KAHRP protein correlated with knob formation. All quantified IT4 NE and IT4 CSA EC37 • IEs were knobless, while all quantified IT4 CSA EC40 • IEs exhibited knob structures at the erythrocyte membrane.
Deletion of the subtelomeric region of chromosome 2 often occurs during long-term cultivation, but was also described for patient isolates [26]. Ruangjirachuporn and colleagues examined 60 freshly isolated P. falciparum isolates from Gambian children for the presence of knobs on the surface of IEs. They detected knobby IE in all isolates. They also found knobless IEs in 42% of the isolates tested. However, these only account for a small proportion. Out of 616 IEs examined, only 39 were knobless, while 572 had knobs. As far as we know, it is not yet clear how frequent subtelomeric deletions occur in vivo and whether and how completely the knobless IEs are sorted out by the spleen [55]. In our study was initially unclear whether knob formation was induced by the interplay of fever and cytoadhesion, or whether knob formation resulted from the selection of knobby IEs from a mixed culture consisting mainly of knobless parasites, but with a minority population of knob-forming parasites. However, the copy numbers of kahrp and pfemp3 were significantly increased in IT4 CSA EC40 • but not in IT4 CSA EC37 • compared with IT4 NE . This leads to the conclusion that long-term cultivated IT4 NE is a mixed population, in which most parasites contain the chromosomal deletion and are thus unable to form knobs, but a minority of the parasites do not contain the deletion. Binding or enrichment to CSA in the presence of increased temperature, therefore, appears to only be possible if the IEs have knobs, thereby selecting parasites without the deletion. Horrocks et al. demonstrated that parasites not expressing kahrp due to the chromosomal deletion display approximately half of the amount of Pf EMP1 on the IE surface and exhibit impaired cytoadhesion, which is consistent with the results of the present study [25].
Finally, we sought to determine if the selection of knobby parasites was restricted to CSA binding of IEs to HBEC-5i cells. As mentioned above, up to 23 ECRs have been identified as IE binding partners. Two of the most investigated receptors are ICAM-1 and CD36. ICAM-1 is a membrane-bound glycoprotein that has a central role in leukocyte trafficking, immunological synapse formation, and numerous cellular immune responses [56]. ICAM-1 is likely to be involved in malaria pathogenesis [57]. Unlike CSA, binding to ICAM-1 is mediated by multiple Pf EMP1s that contain DBLβ3 domains within the DC4-Pf EMP1 domain of group A proteins or the DBLβ5 domains of group B and C proteins, which are not present in VAR2CSA [5,33,[58][59][60][61]. The enrichment process of IT4 NE parasites to transgenic CHO ICAM-1 cells at 40 • C also led to selection of knobby IEs. This indicates that the observed selection mechanisms also apply to other IE-receptor interactions.
Over 70% of IT4 Pf EMP1 proteins contain CIDRα2-6 domains, which are involved in CD36 binding [62]. Interestingly, the enrichment of IEs to CD36 at 40 • C did not lead to selection of knobby IEs. A prior study by Tilly et al. suggested that knobs are not as decisive for CD36 binding as for ICAM-1 binding. Tilly et al. demonstrated that CD36 binding capacity does not differ between knobless and knobby IEs. Knobby IEs, however, bind significantly better to ICAM-1 [63]. In this context, it can be postulated that binding to CD36, is so strong that accumulation of Pf EMP1 molecules within the knob structure is not necessary for CD36 binding under febrile temperatures.
Fever is the most apparent symptom during malaria infection. Reducing fever in children with malaria is controversial and creates therapeutic challenges [64]. On the one hand, fever increases the risk of seizures [65]. However, treatment with the antipyretic drug paracetamol slows parasite clearance due to decreased capacity for production of tumor necrosis factor (TNF) and oxygen radicals [66]. It has also been shown that febrile temperatures increase cytoadhesion [28,30], and the ability of IEs to adhere to certain microvascular endothelial cells correlates with the severity of malaria infection [67][68][69]. On the other hand, fever inhibits parasite growth and leads to parasite death [27,70].

Conclusions
The results of the present study reveal a previously unknown effect of febrile temperature during cytoadhesion. In the presence of febrile temperatures, binding to certain ECRs can only occur if the IEs have knobs. In addition, adhesion is enhanced by increased var expression and correspondingly increased IE surface Pf EMP1. The increased cytoadhesion may, therefore, affect the clinical outcome of malaria infection. However, the mechanism that promotes binding of knobby IEs at febrile temperature is not known. One possibility is that febrile temperature causes conformational changes of Pf EMP1 molecules clustered in knobs and of the receptors, resulting in stronger binding. However, the mechanisms of enhanced adhesion at febrile temperatures requires further investigation. Despite this uncertainty, we were able to shed light on the complex interactions of the Pf EMP1-receptor with ECRs, and show how different environmental factors, such as febrile temperature, influence these interactions. In addition, our findings suggest that the selection pressure for knob formation in IEs originates from the host. Without knobs and the resultant improved cytoadhesion, parasitic survival would likely be impaired, particularly under febrile temperature.