clag9 Is Not Essential for PfEMP1 Surface Expression in Non-Cytoadherent Plasmodium falciparum Parasites with a Chromosome 9 Deletion

Background The expression of the clonally variant virulence factor PfEMP1 mediates the sequestration of Plasmodium falciparum infected erythrocytes in the host vasculature and contributes to chronic infection. Non-cytoadherent parasites with a chromosome 9 deletion lack clag9, a gene linked to cytoadhesion in previous studies. Here we present new clag9 data that challenge this view and show that surface the non-cytoadherence phenotype is linked to the expression of a non-functional PfEMP1. Methodology/Principal Findings Loss of adhesion in P. falciparum D10, a parasite line with a large chromosome 9 deletion, was investigated. Surface iodination analysis of non-cytoadherent D10 parasites and COS-7 surface expression of the CD36-binding PfEMP1 CIDR1α domain were performed and showed that these parasites express an unusual trypsin-resistant, non-functional PfEMP1 at the erythrocyte surface. However, the CIDR1α domain of this var gene expressed in COS-7 cells showed strong binding to CD36. Atomic Force Microscopy showed a slightly modified D10 knob morphology compared to adherent parasites. Trafficking of PfEMP1 and KAHRP remained functional in D10. We link the non-cytoadherence phenotype to a chromosome 9 breakage and healing event resulting in the loss of 25 subtelomeric genes including clag9. In contrast to previous studies, knockout of the clag9 gene from 3D7 did not interfere with parasite adhesion to CD36. Conclusions/Significance Our data show the surface expression of non-functional PfEMP1 in D10 strongly indicating that genes other than clag9 deleted from chromosome 9 are involved in this virulence process possibly via post-translational modifications.


Introduction
An important factor contributing to the virulence of Plasmodium falciparum is the ability of parasitized red blood cells (PRBC) to adhere to receptors such as CD36 or ICAM-1 expressed on the surface of endothelial cells, or chondroitin sulphate A (CSA) expressed on placental syncytiotrophoblasts (see [1] for a review). PRBC sequestration can lead to complete blockage of the microvasculature resulting in severe disease including cerebral malaria that may be fatal [2].
Parasite survival within the newly invaded erythrocyte depends on the synthesis and export of several components, which will be used to build the trafficking pathway and channels necessary for the import and export of essential constituents (see [3]). The variant antigen P. falciparum erythrocyte membrane protein 1 (PfEMP1) is the predominant ligand responsible for adhesion to host endothelial receptors and is essential for parasite survival and establishing chronic infection [2]. Approximately 60 var genes of the parasite's haploid genome encode for PfEMP1 and their expression occurs in a mutually exclusive manner. The var genes share a two exon structure with exon 1 coding for the extracellular highly variable adhesive region and a conserved cytoplasmic exon 2 that codes for a transmembrane region (TM) and interacts with the red blood cell cytoskeleton [4]. Exon 1 is composed of variable numbers of Duffy binding-like domains (DBL) and Cysteine-Rich Interdomain Regions (CIDR) that specifically bind to the different receptors in adhesion assays when expressed in heterologous expression systems [5][6][7].
During in vitro culture, var genes are transcribed and translated in ring stages and, despite the absence of an N-terminal signal sequence [8], PfEMP1 is exported through the endoplasmic reticulum pathway, and beyond the parasite's confines to the erythrocyte surface via the Maurer's clefts by interacting with the structural component Knob-Associated Histidine-Rich Protein (KAHRP) [9]. In general, a single var gene is expressed in a parasite, but expression can switch to another member in the absence of an immune pressure, leading to antigenic and phenotypic variation at the PRBC surface [10]. This is believed to drive escape from the host's immune response and is implicated in pathogenesis.
A recent large-scale knockout study highlighted the complexity of the interaction between parasite proteins secreted into the RBC cytoplasm and cytoadhesion [11]. In this work and other studies several proteins were identified that contribute either in loading PfEMP1 into Maurer's clefts [11] or transfer from the clefts to the erythrocyte surface [11][12][13]. Although this fascinating mechanism is being intensely studied, many cellular processes involved in trafficking of parasite proteins into the host cell remain elusive [3,14]. In this work we identified P. falciparum laboratory lines that have irreversibly lost their adhesive properties but express nonfunctional and trypsin-resistant PfEMP1 molecules on the surface of PRBC. Furthermore, to determine if loss of cytoadherence may have resulted from the absence of the cytoadherence-linked asexual gene (clag9) due to the truncation of chromosome 9 in D10, we performed targeted disruption of clag9 and show, that in contrast to previous studies [15,16], this gene is not essential for the cytoadhesion of PfEMP1.
Cultured parasites were permeabilized with 0.15% saponin [28] before extraction of genomic DNA (GenElute Mammalian Genomic DNA miniprep kit, Sigma). The full PfEMP1 gene sequence was obtained by PCR amplification and determined by gene walking (Supplementary Table 1), and is available from Genbank (accession number JF342594).
For Southern blotting genomic DNA was digested with EcoRI and SpeI (BioLabs), and processed as described [29]. Northern and Southern blots were washed with sodium saline citrate (SSC), sodium dodecyl sulfate (SDS) at 60uC and exposed with BioMax MS films and Transcreen HE (Kodak).
clag9 Knockout P. falciparum 3D7 genomic DNA was used to PCR amplify clag9 exons 1 (bp 1 to 761) and 8 (bp 3652 to 4418) (Supplementary  Table 1), the fragments were digested with SacII/SpeI and EcoRI/NcoI, and cloned into the vector pCC1 [30]. P. falciparum was transfected with this construct leading to the disruption of clag9 by allelic exchange and insertion of the hDHFR selectable marker.

Atomic Force Microscopy (AFM)
Plasmion-selected trophozoites and schizonts were fixed in 4% paraformaldehyde and 0.1% glutaraldehyde for 30 min on ice and washed twice with PBS. PRBC were immobilized on glass bottom WillCo dishes (WillCo Wells) activated using a low-pressure plasma 0 2 (Femto, Diener Electronic) [41,42]. The dishes were then covered with poly-L-lysine (Sigma) for 1 hour and rinsed with water. Experiments were conducted at room temperature on a stand alone AFM (Bioscope 2, Veeco Instruments) equipped with a Nonoscope V controller, mounted on an optical inverted microscope (Axiovert 200 M, Zeiss). Bright-field video images were acquired with a CoolSnap ES camera (Roper Scientific) driven under MetaMorph software (Molecular Devices). AFM images were acquired in Tapping Mode TM using Veeco's silicon RTESP Tip (radius of curvature ,10 nm), plane-fitted (1 st order), and processed under Vision software to get 3D representation. Height profiles of typical knobs were acquired with Veeco's integrated software and plotted using R 2.6.1. For each parasite strain three images showing the greatest dispersion in knobs sizes were selected.

Statistical Analysis
Data were analysed with MinitabH 15 using a 1-way analysis of variance (ANOVA), including Tukey's test for multiple comparisons.

P. falciparum D10 binding assays
Previous studies have shown that P. falciparum bulk-cultured parasites can be enriched for their binding to specific endothelial cell receptors [43]. C32 melanoma cells mainly express the CD36 receptor on their surface, and in lesser amounts the receptors ICAM-1 and CSA [44,45]. To determine the adhesive properties of P. falciparum D10 PRBC, we performed binding assays on C32 cells [25]. Asynchronous cultures of P. falciparum D10 and FCR3 CD36 were incubated with C32 cells, carefully washed to remove unbound PRBC, and the number of adherent PRBC determined. We observed that D10 PRBC did not bind to the melanoma cells, whereas FCR3 CD36 -infected erythrocytes were seen to adhere ( Fig. 1A and B). Furthermore, no D10 adhesion was observed (data not shown) to human umbilical vein endothelial cells (HUVEC), a line to which some non-cytoadherent parasites are able to bind [46].
As several adhesins are implicated in the cytoadherence of PRBC, we further tested the binding of PRBC on purified CD36, ICAM-1, and some receptors rarely expressed in in vitro culture such as CSA and gC1qR. We did not observe any detectable binding of D10 to the tested receptors, which is in clear contrast to FCR3 CD36 control adherence experiments ( Fig. 1C and D). Repeated panning assays (3 times) on C32 cells and CD36 receptor did not yield any increase in adherence supporting the idea that D10 has irreversibly lost the adhesion phenotype.
Characterization of the P. falciparum D10 variant PfEMP1 RT-PCR analysis using primers to amplify the conserved DBL1a domain indicated that a single type of var gene is transcribed by ring stage D10 parasites [47]. Northern blot analysis using a probe corresponding to the DBL1a domain of this var gene showed a unique major 8 kb transcript in ring stages ( Fig. 2A, lane 1). This transcript was not observed in D10 trophozoites and schizonts ( Fig. 2A, lane 2) or in any FCR3 CD36 blood stages (data not shown). Successive hybridization with a conserved ATS probe confirmed that a single major transcript is present in ring stages of D10 ( Fig. 2A, lanes 3, 4).
Next, we verified whether the var transcript was translated into the protein PfEMP1 and expressed on the surface of the PRBC.  2 and 4), or 100 mg/ml trypsin (lanes 3 and 6). TX100 ins -SDS sol extracts were probed with goat anti-ATS and alkaline phosphatase-conjugated anti-goat and revealed with NBT/BCIP. The ATS antigen is shown by an arrow and an arrowhead indicates fulllength PfEMP1. The same membrane was reacted with Hsp70 antibodies to control for loading. E. Surface immunofluorescence of FCR3 CD36 and D10. Plasmion-selected parasites were treated with trypsin prior to incubation with pooled human immune sera (PIAG) followed by goat anti-human IgG Alexa Fluor 488. Scale bars: 5 mm. doi:10.1371/journal.pone.0029039.g002 Surface iodination experiments followed by immunoprecipitation with a pool of human immune sera showed that both FCR3 CD36 and D10 displayed a protein of apparent high molecular weight of approximately 300 kDa present in the TX100 ins -SDS sol fraction. Whereas the positive control FCR3 CD36 was readily digested by 10 mg/ml of trypsin ( Fig. 2B and C, lanes 1-3), the PfEMP1 expressed by D10 was quite insensitive to digestion with even 100 mg/ml of trypsin ( Fig. 2B and C, lanes 4-6) or 100 mg/ml of chymotrypsin (data not shown). A ,75 kDa fragment corresponding to the intracellular domain of PfEMP1 was detected by Western blotting exclusively in the FCR3 CD36 trypsin-treated extracts reacted with anti-ATS antibodies (Fig. 2D lane 1-3). In contrast, these antibodies reacted with the full-length PfEMP1 of D10 treated extracts (Fig. 2D lane 4-6).
PfEMP1 localization was also confirmed by surface IFA of live PRBC following trypsin treatment. As predicted surface labelling was sensitive to trypsin in FCR3 and resistant in D10 parasites (Fig. 2E). Furthermore, cell integrity was verified by propidium iodide stanning (data not shown). These findings indicated that a trypsin-resistant PfEMP1 is expressed on the surface of D10 that has lost its adhesive properties.

Protein trafficking by P. falciparum D10
To further investigate the non-binding phenotype we analyzed by immunofluorescence whether intracellular protein trafficking was affected in D10 parasites. Several P. falciparum polypeptides are exported through the red blood cell cytoplasm in association with the Maurer's clefts, and these may be essential for the correct assembly of the cytoadherence complex at the PRBC surface [48]. PRBC double-labelled with the Maurer's clefts marker Pf332 and PfEMP1 antibodies (Fig. 3A), showed that PfEMP1 was associated with the Maurer's clefts in transit through the RBC cytoplasm. Comparable results were obtained with the FCR3 CD36 PRBC (Fig. 3A), however, the D10 Maurer's clefts labelling detected with anti-Pf332 appeared more diffuse than in FCR3 CD36 suggesting a different morphology. A similar altered Maurer's clefts architecture was observed in REX1 mutant parasites [49], one of the 25 genes lost in the chromosome 9 deletion of D10.
Furthermore, mAb89 reacted with knobs at the RBC membrane of D10-infected erythrocytes indicating that KAHRP, which interacts with the PfEMP1 ATS region, is correctly addressed to the RBC membrane (Fig. 3B). These data suggest that trafficking and export of the studied marker proteins and PfEMP1 are apparently not affected in D10 parasites.
Because PfEMP1 is anchored to the cytoskeleton of the infected erythrocyte through interactions with the knob component KAHRP, we verified if the knobs from D10 had a structure similar to parasites expressing functional PfEMP1 on their surface. The visualisation by AFM of the PRBC revealed only subtle changes in the knob morphology between D10 and FCR3 CD36 (Fig. 4A). No difference was observed in the amount of knobs and their distribution on the erythrocyte surface for these strains. The diameter and height of D10 knobs were slightly smaller than that from FCR3 CD36 (Fig. 4A and B). These results imply that although the knob morphology is marginally affected by the chromosome 9 deletion it is not causing the loss of cytoadherence of D10 parasites.

The CIDR1a domain of the PfEMP1 expressed in D10 binds to CD36
We cloned and sequenced the entire var gene transcribed in D10 (called D10 var1 ) and determined that the open reading frame of this transcript is composed of NTS-DBL1a-CIDR1a-DBL2d-CIDR2b-TM-ATS domains (Fig. 5A). The different domains of PfEMP1 can be classed according to their binding regions allowing the prediction of the corresponding adhesion receptors [5,39]. The M2 minimal domain from PfEMP1 CIDR1a is necessary and sufficient for binding to the CD36 receptor [38]. Thus, to better characterize D10 var1 , the HA tagged M2 domain was expressed on the surface of COS-7 cells (Fig. 5A). Cells were incubated with recombinant CD36 and M2-CD36 surface complexes we detected using anti-CD36 mAb-coated Dynal beads. COS-7 cells surfaceexpressing the M2 domain from D10 var1 and the positive control MC varCD36 are shown in Fig. 5B. No significant differences (P = 0.961) in binding were observed. In contrast, COS-7 cells expressing M2 domain from the negative control FCR3 var1CSA , showed significantly less coated beads on the surface (P,0.001) ( Fig. 5B and C). Transfected COS-7 cells expressing M2 domains were visualized after the binding assay by IFA staining of the HAdomain fused to M2 (green fluorescence). Our results demonstrate that the M2 domain of D10 var1 is functional when expressed outside the D10 infected erythrocyte context and indicate that D10 PRBC have lost their capacity for functional expression of parasite-mediated adhesion, maybe due either to a conformational defect, improper insertion into the RBC membrane, or missing post-translational modifications of PfEMP1. CLAG9 is not essential for binding to CD36 The site of the chromosome breakpoint leading to the deletion of the subtelomeric region was determined by PCR on the telomere repeats and the genes suspected to be located there. Based on the 3D7 genome sequence D10 has deleted around 120 kbp of the right arm of chromosome 9 [50] representing 25 genes, including clag9 (Fig. 6A), encoding a rhoptry protein [37]. CLAG9 was initially implicated in parasite adhesion to CD36 and C32 cells [16] and recently in the adhesion to placental CSA [15]. To determine its potential role in the cytoadherence phenotype, we inactivated the clag9 gene from P. falciparum 3D7. Targeted disruption of clag9 was performed by replacing exons 2 through 6 with the hDHRF selectable marker (Fig. 6B). Gene knockout and integration of the hDHFR cassette were confirmed by PCR analysis (Fig. 6C) and Southern blotting using probes for exons 6 and hDHFR (Fig. 6D-F). CLAG9 was not detected in two knockout clones (E5KO and F4KO), either by immunofluorescence ( Fig. 7A and B) or western blotting (Fig. 7C). In contrast to earlier reports implicating CLAG9 in cytoadhesion, our KO parasites bound to both C32 cells (Fig. 7D) and CD36 (Fig. 7E) at levels comparable to our control 3D7 parasites. These results show that the clag9 gene product is not directly implicated in cytoadhesion to CD36 and, therefore, in the non-cytoadherent phenotype of the D10 strain.

Discussion
P. falciparum export several proteins beyond the parasite membrane to the erythrocyte surface assuring the parasite's survival within the human host. A large number of these exported molecules have been described that are essential for, or contribute to, the trafficking of the major virulence factor PfEMP1 to the erythrocyte surface. Here we describe a novel adhesion mutant phenotype. In D10 PfEMP1 is trafficked to the surface of PRBC but is unable to bind to host adhesion receptors. Even repeated panning experiments on C32 cells or recombinant CD36 bound to plastic dishes did not yield any increase in binding, showing that no other var gene is able to restore the adhesion phenotype. Bulkcultured parasites typically contain minor populations expressing a number of different var gene members (including CD36-binders). Thus, our data indicate that in D10 non-functional PfEMP1 surface expression is implicated in the irreversible loss of cytoadherence. We also have characterized another non-cytoadherent strain, T9-96 [46] with similar results (data not shown),  suggesting that this may be a common adhesion phenotype linked to chromosome 9 deletion.
Trypsin and chymotrypsin digestion experiments of surface iodinated PRBC proteins suggest that PfEMP1 molecules expressed at the surface of D10 display an unusual conformation that protect them from the predicted protease cleavage. It may also be the reason for the loss of the expected binding to CD36. This idea is supported by the fact that recombinant CIDR1aof D10 var1 binds efficiently to CD36 when expressed by another eukaryotic cell. PfEMP1 may need 'helper' proteins acting as a support or chaperones to assume the right shape to function as a ligand. Alternatively, post-translational modifications and/or processing of the protein during export may be needed to obtain functional adhesion molecules. Further studies are required to explore this hypothesis.
Are there candidate genes deleted on chromosome 9 of D10 that could contribute to the described adherence phenotype of D10? Several studies have implicated clag9 in parasite adhesion to CD36 and CSA [15,16]. We revisited this question by irreversibly disrupting the clag9 gene by allelic replacement. The clag9 knockout in 3D7 resulted initially in low CD36 binding parasites, which can be explained by the time taken to obtain cloned knockout lines. However, after one panning step on CD36, these clag9 mutants gained an adhesion capacity similar to the reference parasite 3D7. Thus, our data demonstrate that 3D7, in contrast to previous studies, does not depend on clag9 for the PfEMP1 adhesion process. At this stage, it is difficult to explain the different set of clag9 knockout data. It cannot be ruled out that other gene deletions known to occur during in vitro culture of 3D7, or associated with the gene deletion [13], may have been picked up at the cloning step after the knockout selection. This may mask the observed phenotype of clag9 in the study reported by the Kemp laboratory [16].
clag genes form a small gene family (5 members) and two members (clag3.1 and 3.2) have recently been shown to integrate into the erythrocyte membrane and induce a new permeation pathway [51]. It is tempting to speculate that CLAG9 may accomplish a similar function, a role that is not essential during in vitro culture. Since CLAG9 apparently integrates into the erythrocyte membrane, it may interact with a particular member of the adhesion molecule family as has recently been suggested for var2CSA [15]. Further receptor binding studies of our clag9 knockout are needed to fully explore this possibility. Another candidate gene deleted on chromosome 9 is the Maurer's cleft-resident ring exported protein 1 (REX1). Its deletion results in alteration of Maurer's clefts architecture [49] and a reduction of cytoadhesion to CD36 [13]. These data point to REX1 as a plausible candidate gene for the observed adhesion phenotype in D10. We assume that additional co-deleted genes may act together with REX1 resulting in the complete loss of adhesion. Although D10 is not known to possess any of the REX proteins on chromosome 9, our data indicate that transit of proteins in this parasite strain appears functional. Indeed, labelling of PfEMP1 in D10 seems identical to that observed with the cytoadherent FCR3 CD36 . REX1 mutation resulted in the codeletion of a subtelometic region of 3D7 chromosome 2 including KAHRP [13]. In our D10 parasites this is not observed since our antibodies label KAHRP and AFM demonstrates the presence of knobs. Dixon et al. [13] observe that REX1 deletion leads to trypsin-resistant PfEMP1 and conclude that no PfEMP1 is exposed on the PRBC. In the absence of surface iodination experiments, however, it cannot be excluded that trypsin-resistant PfEMP1 is expressed on the surface. Further characterization of REX1 may unravel the function of this protein in the adhesion process.
A large-scale loss-of-function study designed to characterize candidate genes involved in interactions with PfEMP1 or with a role in knob formation was performed by the Cowman laboratory [11]. Fifty genes were functionally analysed and a number of different adhesion phenotypes were linked to proteins required for loading of PfEMP1 to Maurer's clefts and function on the PRBC surface. However, this study did not include genes deleted on chromosome 9. In addition, the surface exposure of PfEMP1 was only analysed by trypsin treatment, which releases a small trypsinresistant fragment (mainly composed of the ATS domain) between 70 and 90 kDa. Trypsin-resistant PfEMP1 was interpreted as being not exposed on the surface of PRBC. Our data using surface iodination experiments, however, show that parasites can expose non-functional trypsin-resistant PfEMP1. Since PfEMP1 mole-cules interact with KAHRP at the knob protrusions [9], we cannot rule out that the slightly modified knobs structure observed in D10 (see Fig. 4), could contribute to the complete loss of adhesion in D10. The role of the 25 genes deleted in D10 in this process will require further analysis. To answer these questions, work is in progress aiming to complement the deleted genes from chromosome 9 in D10 to establish which genes contribute to the cytoadherence phenotype.
In conclusion, we have identified mutant parasites with nonfunctional PfEMP1 surface expression. We also provide solid experimental evidence that CLAG9 does not contribute in the binding to CD36. In addition, our work illustrates that the factors leading to functional surface expression of PfEMP1 are not well elucidated and that the analysis of the genes deleted in D10 may contribute to our understanding of parasite virulence.