(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Co-chaperone involvement in knob biogenesis implicates host-derived chaperones in malaria virulence

['Mathias Diehl', 'Parasitology', 'Centre For Infectious Diseases', 'Heidelberg', 'Lena Roling', 'Biochemistry', 'Molecular Biology', 'Justus Liebig University', 'Gießen', 'Lukas Rohland']

Date: 2021-10

The pathology associated with malaria infection is largely due to the ability of infected human RBCs to adhere to a number of receptors on endothelial cells within tissues and organs. This phenomenon is driven by the export of parasite-encoded proteins to the host cell, the exact function of many of which is still unknown. Here we inactivate the function of one of these exported proteins, PFA66, a member of the J-domain protein family. Although parasites lacking this protein were still able to grow in cell culture, we observed severe defects in normal host cell modification, including aberrant morphology of surface knobs, disrupted presentation of the cytoadherence molecule PfEMP1, and a total lack of cytoadherence, despite the presence of the knob associated protein KAHRP. Complementation assays demonstrate that an intact J-domain is required for recovery to a wild-type phenotype and suggest that PFA66 functions in concert with a HSP70 to carry out host cell modification. Strikingly, this HSP70 is likely to be of host origin. ATPase assays on recombinant protein verify a functional interaction between PFA66 and residual host cell HSP70. Taken together, our data reveal a role for PFA66 in host cell modification, strongly implicate human HSP70s as being essential in this process and uncover a new KAHRP-independent molecular factor required for correct knob biogenesis.

To survive in the human body, the malaria parasite invades and lives within human red blood cells. Once within the red blood cell, the parasite renovates the host cell to its own needs. Here we have studied which factors from both parasite and host cell are required for this renovation process, and discover that human chaperone proteins, referred to as HSP70, are required. It appears that a particular parasite-derived protein, PFA66, recruits and modifies the function of the human HSP70. As this interaction between a parasite and human protein is novel and essential for parasite survival, our study identifies a potential Achilles’ Heel which may be targeted for development of new anti-malaria therapies.

Within the predicted ‘exportome’ are 19 proteins belonging to the family of J-domain proteins (JDPs, also called HSP40s), and this exported family appears to be expanded in the Laveranian clade, suggesting important functions in these parasite species [ 9 , 14 ]. In other systems, HSP40s act as cochaperones for HSP70, a protein family that lies at the heart of proteostasis and other essential cellular processes [ 15 ]. Previous studies have localised several exported members of the PfHSP40 family to various structures within the infected RBC, including RBC plasma membrane, Maurer’s clefts, knobs, and J-dots [ 13 , 16 , 17 ]. Knockout studies suggest that, although some of the exported HSP40s are essential, others can be deleted without any observable phenotype, and knockout of others results in aberrant cellular morphology, cytoadherence, and rigidity of infected RBC (iRBC), suggesting a potential role for this protein family in host cell modification and virulence characteristics [ 12 , 13 ] ( S1 Table ). One exported HSP40, PFA66 (encoded by PF3D7_0113700, formerly PFA0660w), was previously localised to J-dots, novel structures within the iRBC containing further exported parasite-encoded HSP40s, an exported parasite-encoded HSP70 (PfHSP70-X), and a number of other exported proteins [ 16 , 18 , 19 ]. An earlier medium throughput knockout study failed to generate parasites deficient in PFA66, and therefore its function in the parasite’s lifecycle remains elusive [ 12 ].

Plasmodium falciparum causes the most severe form of malaria in humans, malaria tropica, responsible for over 200 million clinical cases and 400,000 deaths per annum, mainly in children under the age of 5 and mostly in sub-Saharan Africa [ 1 ]. The pathology associated with malaria infection is largely due to the ability of infected human RBCs (red blood cells, RBC) to adhere to a number of receptors on endothelial cells within tissues and organs [ 2 ]. This cytoadherence results in reduced blood flow in the affected areas, hypoxia and (in cerebral malaria) increased intracranial pressure [ 2 , 3 ]. The phenomenon of cytoadherence results from parasite-induced host cell modification in which parasite-encoded proteins are transported to and exposed at the surface of the infected host cell, where they mediate endothelial binding and antigenic variation [ 4 – 6 ]. In addition to these surface proteins, parasites also encode, express, and export a large number of other proteins to the infected RBC [ 7 – 9 , 10 , 11 ]. Many of these proteins are specific to P. falciparum, and their function is still not well understood, partly due to limitations in reverse genetic systems [ 9 , 12 , 13 ].

A) Single turn-over ATP hydrolysis of HsHSP70 in the presence of a peptide substrate (red circles) or PFA JDS (blue circles). Note that in contrast to Daniyan et al. 2016 [ 39 ], we clearly observe a stimulation of the hydrolysis of ATP by HsHsp70 (HSPA1A) in the presence of PFA JDS B) Single turn-over ATPase rates for HsHSC70, HsHsp70 and PfHSP70-X in the absence of any substrates or JDPs (black circles), in the presence of the substrate peptide (red circles) or in the presence of PFA JDH (blue circles). Shown are individual data points and standard deviation of independent experiments (n = 3 to 8). *, p < 0.05; **, p<0.001 (ANOVA with Holm-Sidak’s multiple comparison test).

Knowing that a functional J-domain is needed to maintain normal knob morphology, we wanted to probe for potential HSP70 partners for PFA66 in the RBC. Residual human HSP70s and one parasite exported HSP70 (PfHSP70-X) are present in the RBC and it has been suggested that the parasite hijacks the RBC chaperone machinery by exporting various JDPs to facilitate remodelling of the RBC [ 16 , 18 ]. To test which of these HSP70s functionally interact with PFA66, we performed single turn-over ATPase assays with HsHSP70 (HSPA1A) and HsHSC70 (HSPA8), and the parasite encoded PfHSP70-X [ 38 ]. Previously, using steady state ATPase assays, it has been shown that PFA66 is able to stimulate the hydrolysis of ATP by PfHSP70-X and (at a lower degree) HsHSC70 but not HsHSP70 [ 39 , 40 ]. The basal ATPase activity of HSP70s is generally very low and stimulated in synergism by J-domain proteins and substrates [ 37 , 41 ]. Certain JDPs are able to provide both signals, the J-domain and the substrate, needed to synergistically stimulate the ATPase activity of HSP70s. To rule out the possibility that the lack of stimulation of HsHSP70 by PFA66 that has been observed by Daniyan et al. [ 39 ] is due to PFA66 itself not being a substrate for HsHSP70, we applied the experimental strategy of Landry and colleagues [ 42 ] and designed a minimal J-domain-substrate system by fusing a peptide derived from the human heat shock transcription factor 1 (HsHSF1, S461-Q471, referred to as substrate), that has recently been identified as binding site for HsHSC70 [ 43 ], C-terminally to the PFA66 J-domain (amino acids 81–148, resulting in PFA66 JDS ). To shield the peptide from degradation, a C-terminal strep-tag (WSHPQFEK) was added to PFA66 JDS as well. HsHSP70, HsHSC70, PfHSP70-X, and PFA66 JDS were expressed as HIS-SUMO fusions, purified to homogeneity and used to perform ATPase assays under single turn-over conditions. Under such conditions, the hydrolysis of one molecule of ATP per HSP70 molecule is monitored and other potentially rate limiting processes, such as nucleotide exchange, are neglectable. We observe that PFA66 JDS is able to stimulate the hydrolysis of ATP by HsHSC70 and PfHSP70-X, as expected, but unexpectedly also by HsHsp70 ( Fig 6A ), to significantly higher rates than the peptide alone, supporting the idea that PFA66 is able to recruit and repurpose the two major human HSP70 homologues in RBCs ( Fig 6A and 6B ).

HSP40s such as PFA66 are known interactors and regulators of HSP70s [ 14 , 15 , 37 ], and HSP70-independent functions of HSP40s are rare. The J-domain of HSP40s is crucial both for recruitment of the partner HSP70 and stimulating the ATPase activity of the HSP70 partner [ 37 ]. Having shown above that we can achieve functional complementation via episomal expression of a full-length copy of PFA66, we were interested to know whether a functional J-domain (and hence a functional interaction with a HSP70) is required for phenotypic complementation. To this end, we expressed in the ΔPFA parasite line a full-length copy of PFA66 with a H111Q amino acid replacement converting the HPD motif of the J-domain into the non-functional QPD sequence ( S6H Fig ) and assayed knob/eKnob morphology and density in comparison to the wild-type complementation parasites as a surrogate for all other phenotypic assays. Both wild-type (HPD) and mutant (QPD) fusions were expressed at a similar level, and localised to punctuate structures in the iRBC, suggestive of a correct J-dot localisation ( S6I Fig ). While expression of full-length wild-type PFA66 was able to revert the mutant knob phenotype ( Fig 2B and 2C ), similar expression of the non-functional QPD mutant failed to complement the WT phenotype ( Fig 5C and 5D ) in either density or morphology of knobs/eKnobs.

A) ΔPFA displays negligible cytoadherence and lower PfEMP1 surface exposure than CS2. CS2 and ΔPFA were assayed to test their ability to adhere to immobilised CSA in Petri dishes using microscopic counting of the cells. Cytoadhesion strength is expressed relative to CS2. Results are shown for six binding assays. B) Analysis of PfEMP1 surface exposure via flow cytometry. IRBCs were stained with DAPI and αVAR2CSA antiserum followed by a Cy3-coupled secondary antibody. ΔPFAs have lower PfEMP1 surface exposure than CS2 in six independent experiments. C, D) Expression of a PFA variant featuring a mutated HPD motif in the cell line ΔPFA [QPD::HA] does not complement reduction in knob abundance (C) and knob deformation (D) observed in ΔPFA. IRBC were purified and imaged via SEM. Knobs were then counted and grouped into three categories.

As previously mentioned, mature stage infected P. falciparum iRBCs develop additional adhesive capabilities to cells of the host, and this process underlies P. falciparum pathogenicity [ 2 ]. Cytoadherence of iRBCs to purified ligands can be assessed with a binding assay [ 21 ]. Prior to genetic manipulation we selected our parasite population for expression of the var2CSA variant of PfEMP1 by repeatedly binding them to CSA [ 22 ]. We chose to use the CS2 strain of P. falciparum as this strain stably expresses var2CSA [ 21 ]. A static cytoadherence assay against immobilised CSA demonstrated that, in direct comparison to the parental CS2 strain, ΔPFA exhibits massively reduced levels of binding ( Fig 5A ). Although all experiments were carried out on parasites that had not been maintained in culture for extended time periods (to avoid switching to another var/PfEMP1 variant), we wanted to verify that ΔPFA still expressed PfEMP1 var2csa . We therefore used flow cytometry on intact iRBCs using antisera specific against VAR2CSA (a kind gift of Benoît Gamain) to verify VAR2CSA expression and surface exposure. On RBCs infected with ΔPFA, surface expression of VAR2CSA was reduced 60% compared to wild type parasites ( Fig 5B and S6F Fig for histograms). IFA using the same antisera on fixed cells demonstrated that both CS2 and ΔPFA express VAR2CSA to similar levels, with punctuate staining distributed across the host cell. A control IFA on cells infected with 3D7 strain parasites that had not been selected for VAR2CSA expression showed only background fluorescence, verifying the specificity of the result ( S6G Fig ).

A number of lipid- or protein-dependent mechanisms can induce the initiation of curvature of biological membranes and stabilisation of the resulting structures [ 36 ]. The eKnobs we observe here extend up to 0.7 μm from the surface of the RBC and appear stable enough that we were able to observe them in live cell imaging. To understand how such structures can be generated, we treated RBCs infected with ΔPFA parasites with cytochalasin-D (cyto-D) to depolymerise actin, methyl-β-cyclodextrin (MBCD) to chelate membrane cholesterol, or the glucosidases hyaluronidase (HA) and neuraminidase (NA) to degrade RBC glycocalyx. Following treatment, cells were fixed and prepared for SEM, followed by image acquisition and analysis. Neither cyto-D nor HA/NA treatment caused a statistically significant reduction in the number of eKnobs in cells infected with ΔPFA ( S6A, S6B and S6E Fig ). Treatment with MBCD, while not causing a total reversion of eKnobs to knobs, did cause a statistically significant alteration in the observed type of eKnobs, with a decrease in the number of deformed, elongated eKnobs, and an increase in abnormal, enlarged eKnobs ( S6C and S6D Fig ).

Line scan analysis reveals that eKnobs are bounded by the RBCM. In most cases, KAHRP is found between the RBCM and the actin cytoskeleton, and the KAHRP::mCherry signal also extends into the central cavity of the eKnobs. Distribution of host actin closely follows that of WGA, apart from at the base of eKnobs where it follows a path below the KAHRP staining. Fortuitous sectioning of several membrane-bounded eKnobs extending from the RBC reveals a ring of KAHRP::mCherry staining lining the luminal face of the eKnob. Phalloidin staining was absent in this case, although it was associated with a shorter eKnob ( Fig 4 , see S5D Fig for CS2 cell line).

A, C) DNA was stained using DAPI; WGA was used to stain the RBC glycocalyx; phalloidin was used to stain actin; and RFP booster was used to label KAHRP::mCherry. B, D) Line scan profiles of fluorescence intensity (arbitrary units) along the yellow arrows shown in the fluorescent overlay. The arrows bisect KAHRP_mCherry-rich structures (likely representing knobs). These profiles demonstrate that in the vertical view, phalloidin (i.e., actin) is localised toward the cytosol from the KAHRP structures. The horizontal view shows that the KAHRP-containing structures form a ring structure. These might contain low amounts of actin but are likely filled with other material(s).

A) IFA assay of MeOH-Ac-fixed CS2 and ΔPFA using α-KAHRP antibodies reveals punctate patterns. A trend was noticed towards bigger spots in the truncation strain and verified using automated measuring via an ImageJ algorithm (See Fig 3 B, C). D) Live cell imaging of DAPI stained CS2 [KAHRP::mCherry] and ΔPFA [KAHRP::mCherry] . KAHRP::mCherry can be seen in both cell lines as punctate patterns; however, CS2 displays smaller and more dots. E) Immunogold labelling of iRBC sections in TEM using α-KAHRP antibodies. Images demonstrate label associated with normal knobs and deformed knobs in CS2 and ΔPFA, respectively. Framed areas can be seen enlarged below. F) Analysis of label density associated with the cytoplasm and area surrounding knobs. Label density is significantly higher in the area surrounding knobs than the cytoplasm for both strains. G) STED imaging of the KAHRP associated with the internal RBC cytoskeleton. For this analysis CS2 and ΔPFA iRBCs were bound to a dish and then lysed hypotonically. The cell body was then washed away, and the remaining cytoskeleton remained as it would be seen from the inside of the iRBC. These samples were then interrogated with an α-KAHRP antibody and STED imaging. G) Representative images of the KAHRP patterns observed in STED from the CS2 and ΔPFA cell line. KAHRP signals were often found to be bigger in the truncation cell line. H) Computational analysis of KAHRP signals through a self-made ImageJ tool revealed no difference in KAHRP spot numbers between both cell lines. I) Investigation of mean object size demonstrated a slight increase of KAHRP spot size in ΔPFA.

KAHRP has long been held to be a crucial knob-associated protein as parasites lacking this protein no longer form knobs [ 30 , 31 ], and KAHRP truncations show varying aberrant knob phenotypes [ 32 ]. For this reason, we investigated the localisation of KAHRP in RBCs infected with our ΔPFA cell line. IFA on fixed cells demonstrated a punctate distribution of KAHRP in cells infected with both wild type and mutant cell lines ( Fig 3A ). Automated analysis with a self-generated ImageJ plugin paired with Ilastik [ 33 ] revealed that structures labelled for KAHRP were equally numerous but statistically larger in diameter in RBCs infected with ΔPFA than with CS2 ( Fig 3B and 3C ). Localisation of other exported parasite proteins via IFA showed no dramatic difference between the cell lines ( S4 Fig ). To exclude that the KAHRP result was due to non-specific binding of antibodies, for example, we episomally expressed a KAHRP::mCherry fusion in both WT and ΔPFA parasites ( S5A Fig ). Surprisingly, considering the relatively low resolution of live cell epifluorescence microscopy, (but preserving membrane integrity and cell morphology), structures labelled with KAHRP::mCherry could be seen to emerge from the surface of RBCs infected with ΔPFA, possibly representing eKnobs ( Fig 3D , S3 – S5 Videos). Based on the previous result, we wanted to understand whether KAHRP is directly associated with eKnobs and used immunogold labelling of thin sections derived from CS2 and ΔPFA iRBCs to localise KAHRP. Although we encountered high background staining of the RBC cytosol in both cases, analysis revealed considerably more label was associated with knobs/eKnobs than with the cytosolic background ( Fig 3E and 3F ). Analysis of the location of the KAHRP label in relation to the length of the knob/eKnob or the closeness to the RBC membrane (RBCM) showed no difference between WT and mutant, or between knob/eKnob morphologies ( S5B and S5C Fig ). To gain more insight into the nature of the structures labelled with α-KAHRP antibodies, we paired membrane shearing with immunolabelling and STED (stimulated emission depletion [ 34 ]) microscopy to study the nature of the structures labelled by α-KAHRP antibodies observed above. RBCs infected with CS2 or ΔPFA were allowed to bind to a cover slip, hypotonically lysed to obtain access to the internal leaflet of the RBC plasma membrane, fixed, immunodecorated using an α-KAHRP antibody, and imaged STED microscopy. This technique allows super-resolution visualisation of KAHRP structures from the luminal side of the RBC plasma membrane and can be used to monitor the assembly of KAHRP into knobs (and in this case, eKnobs). This analysis revealed a punctuate distribution of KAHRP beneath the membrane of RBCs infected with both wild type and mutant ΔPFA cells. Although the data initially suggests that there are more KAHRP structures in membranes from ΔPFA, automated counting and measurement shows that this impression is likely to be imparted due to a higher mean size of KAHRP objects rather than an actual increase in the absolute number of objects ( Fig 3G, 3H and 3I ).

Intrigued by the apparent electron-dense core of eKnobs in TEM, we conducted electron tomography analysis on thick sections prepared from ΔPFA-iRBCs. Subsequent 3D reconstruction and surface rendering of the distinct densities in tomographic volumes allows a high-fidelity glimpse into the fine structure of eKnobs. This analysis verified the presence of electron-dense material within and at the base of eKnobs ( Fig 2E ). This material fills out the entire eKnob rather than only lining the structure, and the distribution of this material closely matched the structure of the eKnob. In some cases (3 out of 10), the electron-dense material at the base of the eKnob was connected via a thin bridge to the material inside of the eKnob ( Fig 2F , S1 Video ).

Transmission electron microscopy (TEM) on thin sections prepared from RBCs infected with either WT CS2 or ΔPFA parasites substantiated our observations. CS2 parasites produced clearly defined electron-dense knob structures of a restricted diameter and height above the RBC plasma membrane, whereas ΔPFA displayed eKnobs extending from the RBC surface into the external medium ( Fig 2D and S3 Fig ). The lumen of these eKnobs was often extremely electron dense, hinting at their molecular relation to knobs. Occasionally we observed membrane-bound structures extending from RBCs infected with CS2 parasites, but the lumen of these structures was not electron dense; therefore, these structures cannot be classed as eKnobs. The morphology of Maurer’s clefts showed the normal variation in morphology in both samples ( S3 Fig ). Several clefts appear to lie perpendicular to the host cell membrane, but this phenotype was observed in both samples ( S3 Fig ).

A) Scanning electron microscopy shows knobs on the surface of iRBCs. The mutant phenotype of ΔPFA is alleviated upon reintroduction of episomally expressed PFA66 in ΔPFA [PFA::HA] . More pictures can be found in S2 Fig . B) Quantification of knob density via ImageJ in SEM pictures (n = 20) shows significantly fewer knobs on ΔPFA iRBCs. Knob density is restored in the complementation cell line CS2ΔPFA [PFA::HA] . C) Quantification of knob morphology across all iRBCs. Knobs were grouped into three categories: small knobs, enlarged knobs, and elongated knobs. Then every knob on 20 SEM pictures of the three strains was assigned to one of these categories. Each bar represents the distribution of these knobs in the three categories across all pictures of a strain. The ΔPFA strains display an increase in deformed and enlarged knob morphologies compared to CS2 and ΔPFA [PFA::HA] . D) Internal view of the deformed knobs/eKnob via transmission electron microscopy of thin slices. More pictures can be found in S3E Fig ) Electron tomography reveals electron-dense material at the base and interior of deformed knobs/eKnobs. The marked area denotes the structure shown in F. F) 3D segmentation of discrete densities within a deformed knob/eKnob depicted with electron tomography. The example shows a severely deformed knob/eKnob. Additionally, electron dense material was detected at the base and inside of these structures.

Exported parasite proteins carry out a multitude of functions supporting the survival of P. falciparum parasites. One of these is the establishment of NPPs of the iRBC to support the uptake of essential nutrients [ 28 , 29 ]. To investigate a potential role of PFA66 in NPP activity, we used a sorbitol uptake/lysis assay, which revealed that RBC infected with ΔPFA parasites show a small but significantly reduced sorbitol-induced lysis, implying a reduced NPP capacity ( S1C Fig ). To examine whether this reduction in NPP activity affects parasite viability, growth of both cell lines was compared over four growth cycles (~8 days) using flow cytometry. Surprisingly, a slight but significant growth advantage (calculated as below 1% advantage per cycle) of the truncation cell line compared to the parental cell line was observed ( S1D Fig ).

Full-length PFA66 has previously been localised to the J-dots, and a follow up study suggested that the SBD of exported HSP40s is required for correct localisation [ 16 , 24 ]. Furthermore, the SBD of PFA66 has previously been shown to bind cholesterol, and this is also likely to be required for correct localisation and thus protein function [ 25 ]. As dPFA lacks the SBD, but still contains both an N-terminal signal peptide and a recessed PEXEL trafficking signal, we expect export of dPFA::GFP to the host cell [ 7 , 8 ]. Live cell imaging and IFA failed to detect dPFA. This is likely due to the low abundance, and dispersion of the truncated protein throughout the entire host cell. We therefore used equinatoxin (EQT) to selectively permeabilise the RBC plasma membrane and allow sub-cellular fractionation of iRBC [ 26 , 27 ]. Immunodetection using antibodies against the compartment-specific markers SERP (parasitophorous vacuole), human HSP70 (HsHSP70, RBC cytosol), aldolase (ALDO, parasite cytoplasm), and GFP reveals co-fractionation of dPFA with HsHSP70, verifying that dPFA::GFP is found in the host cell cytosol and furthermore that dPFA::GFP is found in the soluble phase and not in the membrane fraction as we have previously demonstrated for the full-length protein ( S1B Fig [ 16 ]). Taken together, these data suggests that deletion of the SBD of PFA66 leads to a non-functional protein.

A) Strategy for inactivation of PFA0660w via selection-linked integration. Expression of a neomycin resistance marker (NEO) is coupled to integration of the plasmid pSLI TGD_PFA into the genomic PFA0660w locus, leading to expression of a truncated (likely inactive) PFA66 missing its substrate binding domain (SBD). Production of PFA and NEO as separate proteins is mediated with a SKIP peptide. B) Integration PCR using gDNA extracts from the cell line ΔPFA and the parental cell line CS2 verifies integration of the plasmid pSLI TGD_PFA into the PFA0660w gene in the cell line ΔPFA. Amplification of the wild type PFA0660w locus with the primers C and D is only successful in the parental strain CS2 since integration of the plasmid dramatically increases product size. PCRs using primers spanning the junctions of the integration sites C and B for the 5’ region and A and D for the 3’ region) demonstrate disruption of PFA0660w. C) Western blot verifies truncation of PFA0660w in ΔPFA. The truncated fusion protein was detected using an α-GFP antibody, while the parasite protein SERP served as a loading control. + unskipped product, *GFP degradation product. D) Immunodetection verifies expression of HA-tagged PFA66 in the complementation cell line ΔPFA [PFA::HA] .

Genetic manipulation via single crossover was performed using a selection-linked integration targeted gene disruption strategy [ 20 ]. Integration of the plasmid would lead to the production of a GFP-tagged, truncated, non-functional PFA66 protein lacking the entire C-terminal substrate binding domain (SBD), referred to as dPFA. We transfected CS2 parasites that had previously been freshly selected for binding to chondroitin-sulphate-A (CSA) as this binding phenotype would be essential for later characterisation [ 21 , 22 ]. Plasmid integration into the PFA0660w/PF3D7_0113700 locus was verified via PCR using primers designed to yield products upon integration of the plasmid into the genome via primers spanning the integration site ( Fig 1A ). Appearance of bands representing the 5’ and 3’ integration, as well as the disappearance of bands representing the endogenous PFA0660w locus, demonstrated specific integration of the plasmid into the genome ( Fig 1B and S1A Fig ), yielding parasite line CS2ΔPFA (referred to as ΔPFA). Immunodetection using an α-GFP antibody revealed the presence of dPFA::GFP fusion at the predicted size in cell lysates derived from ΔPFA but not parental CS2 parasites (predicted MW 39 kDa, Fig 1C ). Additionally, one further higher molecular weight band was observed, likely representing a non-skipped fusion (predicted MW 68 kDa, Fig 1C + ), similar to that previously reported [ 23 ]. Although levels of the control SERP protein were similar in all samples analysed, verifying equal sample loading, the dPFA::GFP fusion appears to be present at a lower level than in parasites expressing endogenous full length PFA tagged with GFP ( Fig 1C ). Although we are unsure or the reasons for this, the higher abundance of the commonly observed GFP degradation product around 25kDa ( Fig 1C *) may suggest lower protein stability. Together, this data indicated successful integration of the vector into the PFA0660w locus, leading to the production of a truncated product dPFA lacking the substrate binding domain required for correct localisation and function [ 24 ]. To verify that any aberrant phenotypes observed were due to a lack of PFA66 and not second site events, we generated a complementation line that expressed a full-length, functional 3xHA (hemagglutinin)-tagged copy of PFA66 from an episome, under the control of the endogenous PFA66 promoter, referred to as ΔPFA [PFA::HA] . Expression of this complementation construct was verified by immunoblotting using α-HA antibodies ( Fig 1D ).

Discussion

Although it has long been recognised that malaria parasites export a substantial number of proteins to their host cell, the mature human RBC, the function of many of these proteins remains unknown [7,8,13]. P. falciparum exports a larger number of proteins to the host cell than related species, and one family that is highly represented amongst this expansion is that of J-domain proteins (HSP40s [9]). A previous medium-throughput study identified the exported HSP40 PFA66 as likely to be essential and resistant to inactivation via double-crossover integration [12]. In this study, we have used SLI-TGD [20] to generate parasites expressing a severely truncated form of PFA66. This strategy deletes the entire substrate-binding domain of the expressed protein, resulting in a non-functional truncation mutant that is incapable of carrying out its biological function, and is also present at low abundance. The resulting parasites have normal viability but exhibit severe abnormalities in host cell modification.

Parasite growth following truncation of PFA66 was slightly higher than that of wild type CS2 parasites, this effect only becoming evident after four growth cycles. A similar effect was previously observed in a ΔGEXP07 parasite line; however, the significance of this result remains unclear as the reduction in metabolic burden by the loss of only one gene/protein is likely to be negligible [44]. Lysis of iRBCs in a sorbitol uptake assay was also slightly reduced, implying either a reduction in novel permeability pathway activity, or potentially an increase in host cell stability via mechanical means. As a reduced NPP activity would be expected to lead to slower, not faster, parasite growth, we interpret this to be the result of increased robustness of the RBC plasma membrane through an unknown mechanism. Alternatively, growth in the rich media we used may not have uncovered a more obscure effect on NPP [45], and future studies may wish to address this.

The most striking result of our study was the observation that, in the ΔPFA66 cell line, normal knob biogenesis was significantly inhibited with regard to both knob density and morphology. Although earlier knockout studies have observed a reduction in knob formation, or slight alteration in knob morphology, to our knowledge this current study is the first to demonstrate such a dramatic alteration in knob structure upon inactivation of a single gene [12,44]. Indeed, so different are the structures we observe to classic knobs that we suggest calling them eKnob to distinguish them from the ‘normal’ surface extensions. eKnobs differ from knobs both in their size and length, reaching up to 0.7μm from the RBC surface. Additionally, eKnobs that split into separate branches can be observed. KAHRP, a protein known to be required for correct knob formation, can be localised to eKnobs, although its distribution (based on live cell imaging, IFA, and membrane shearing paired with STED microscopy) seems to be different from that observed in cells infected with wild type parasites. Several studies suggest that KAHRP is integrated into higher order assemblies during parasite development, eventually resulting in a ring structure underlying the knob [46,47]. Deletions in specific KAHRP domains lead to less incorporation into such structures and also appear to influence the generation of sub-knob spiral structures of unknown molecular composition [47]. The possibility exists that KAHRP, while being necessary for correct knob formation, is itself not the major structure-giving component but merely serves as a scaffold for assembly of further higher order molecular structures, which themselves generate the necessary vector force to allow membrane curvature and push the knobs above the surface of the RBC. KAHRP has been proposed to be especially vulnerable to misfolding due to its unusual amino acid composition, which would make it a likely client for chaperones / cochaperone systems [17]. Following this logic, if PFA66 is involved in the assembly of KAHRP into higher order assemblies, interruption of this process could cause knock-on misassembly of the spiral and therefore aberrant knob structures. Although we were not able to visualise spiral structures at the base of eKnobs, our observed phenotype could be explained by runaway lengthening of such a spiral structure (Fig 7). Indeed, the electron density we observed on tomograms is consistent with a high molecular weight structure as a form-giving scaffold for the formation of eKnobs. Within the resolution limit of our study, KAHRP itself appears to line the inner leaflet of the membrane-bounded eKnobs and is thus unlikely to itself be a major component of the electron dense core of the eKnobs. Host actin is known to be required for the generation of knobs [46]; however, although we could successfully visualise actin below the RBC plasma membrane, it appears to be largely excluded from eKnobs and thus cannot be responsible for maintaining the form of these structures, a view supported by the lack of action of cyto-D on eKnob structures. Similarly, enzymatic removal of the RBC glycocalyx had no effect on eKnobs, excluding a role for this in membrane shaping. Chelation of membrane cholesterol via treatment with MBCD did cause change in the morphology of eKnobs, but not a complete reversion to normal knob structures. Removal of cholesterol from biological membranes has been observed to cause an increase in membrane stiffness [48,49], and this may explain the morphological reversion upon MBCD treatment, with a stiffer membrane being more resistant to the pushing force within the eKnobs. Alternatively, removal of cholesterol and subsequent breakdown of so-called lipid rafts may interfere with the higher order organisation of membrane-bound factors involved either directly in membrane curvature or required for coordination of proteins internal to the eKnobs that generate force. Taken together, our data strongly implies that both knobs and eKnobs contain a so far unknown component, likely proteinaceous, which is required for force generation, subsequent membrane curvature, and resulting morphology. Interruption of correct assembly of this factor leads to aberrations in knob formation, eventually leading to the appearance of eKnobs. Our, and others’, data implicates KAHRP as having a role in this process, but it is unlikely to be the only structural protein involved.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. Proposed model for eKnob formation and structure. In opposition to normal knob formation in the CS2 cell line (left) runaway extension of the spiral underlying eKnobs in ΔPFA could drive their elongation (right). KAHRP is still present and associated with the inner lumen of eKnobs, PfEMP1 anchored in the eKnobs is incorrectly presented and has thus a reduced cytoadherence capacity. https://doi.org/10.1371/journal.ppat.1009969.g007

In other systems, HSP40s, through their role as regulators of HSP70 chaperone activity, have been shown to have a role in both assembly and disassembly of protein complexes [50], and it is tempting to suggest that the phenotype we observe here is due to incorrect complex assembly. Alternatively, PFA66 may be required for the correct transport of accessory proteins required for complex formation such as those proposed in a recent study [46], and thus play an indirect role in correct assembly of high-molecular weight complexes. In support of this hypothesis, PFA66 is known to associate with J-dots, highly mobile structures within the iRBC that are also known to contain a number of HSP70s [16,18,19,24]. It is also feasible that PFA66 is required for the disassembly of incorrectly folded or assembled knob protein complexes and that our knockout reveals so far unknown quality control mechanisms.

Knobs are required for correct presentation of the major virulence factor PfEMP1, and high affinity binding of such to endothelial receptors [51]. Although previous studies suggested that KAHRP and PfEMP1 formed a ‘pre-cytoadherence complex’ at the Maurer’s clefts [52], later evidence suggest strongly that PfEMP1 is incorporated only into knobs once they have been at least partly formed [46,53]. RBCs infected with ΔPFA66 parasites showed a 63% reduction in knob/eKnob density and a 10-fold increase in the frequency of abnormal knob phenotype, and we observed an almost total lack of cytoadherence in iRBC infected with ΔPFA66 parasites. This data strongly supports the view that, even in the normal knobs present, less PfEMP1 was correctly presented and could take part in cytoadherence. Flow cytometry determined an almost 60% drop in cell surface recognition of the VAR2CSA variant of PfEMP1, although IFA suggests that both cell lines express similar amounts of this protein. The total loss of cytoadherence may be the consequence of several distinct factors: a) less total surface PfEMP1, b) fewer knobs with correctly loaded PfEMP1, and c) a significant number of aberrant knobs/eKnobs. Moreover, we cannot exclude that PfEMP1, which needs to be correctly folded to bind specific receptors, does not assume the correct tertiary structure due to the lack of the necessary chaperone-cochaperone system.

As previously mentioned, HSP40s generally act in concert with members of the HSP70 family [15]. The RBC is known to contain significant amounts of residual human HSP70s [54] and additionally a parasite-encoded HSP70, PfHSP70-X [18]. We have previously demonstrated that a knockout of PfHSP70-X leads to a reduction in virulence characteristics, including cytoadherence [55]. Significantly, however, iRBCs infected with Δ70-X parasites were covered with normal knob structures at a density comparable to that of the wild type [55] and a further study even suggests that deletion or down-regulation of PfHSP70-X leads to no observable mutant phenotype [56]. Hence the mutant phenotype of RBCs infected with ΔPFA66 is distinct and significantly more dramatic than that in iRBCs infected with Δ70-X parasites. Considering the difference in phenotype between Δ70-X and ΔPFA66, here we must conclude that, if a HSP70 is involved, it is more likely to be of human rather than parasite origin, a possibility also previously raised by others [56].

The J-domain of HSP40s contains a characteristic HPD motif that is required both for binding and ATPase activation of partner HSP70s [50]. As the unlikely possibility existed that our observed phenotype was due to a HSP70-independent function of PFA66, we carried out complementation analysis with a full-length copy of PFA66 expressed from an episome, or a copy containing a H111Q mutation that renders the J-domain inactive. Despite similar expression levels and correct localisation of both fusions, only the wild type protein was able to complement the mutant knob phenotype, clearly demonstrating that recruitment of Hsp70 by PFA66 is essential for the wild type phenotype. We therefore conclude that the effects we observed upon PFA66 truncation are due to either a negative effect of the truncated protein on proper functioning of the HsHSP70 chaperone system within the host RBC or result from a lack of cochaperone activity via deletion of an essential functional domain (SBD) of the HSP40. Either way the results furthermore support HsHSP70 involvement.

To regulate the function of residual human HSP70s, the J-domain of PFA66 should be able to stimulate the ATPase activity of these proteins, however several studies have demonstrated only weak functional interactions. One potential weakness of these published studies is the failure to distinguish between chaperone-cochaperone and chaperone-substrate interactions, which may produce misleading results. To throw more light on this topic, we uncoupled these interactions and reconstituted a minimal J-domain system by fusion of a suitable substrate peptide to the J-domain derived from PFA66. Single-turnover ATPase assays revealed significant stimulation of both the parasite derived PfHSP70-X and human HSP70s compared to controls, further supporting our assertion that the mutant phenotype observed upon deletion of PFA66 is linked to the activity of residual human HSP70. We cannot formally exclude that PFA66 functionally interacts with both human and parasite HSP70 homologues and that the phenotype we observe is a combination of the effect on both chaperones, but the balance of probabilities suggests that morphological abnormalities observed in this current study are largely due to an interruption of functional PFA66-HsHSP70/HSC70 interactions.

A potential role for residual human HSP70 in host cell modification and parasite virulence has been suggested for almost 20 years [57], but to our knowledge our current study is the first to provide strong experimental evidence directly implicating human HSP70s in these processes. We wish to note that we cannot unequivocally state that our complementation construct caused reversion of all the aberrant phenotypes herein investigated, however we feel it likely that all abnormal phenotypes are linked and thus our “proof of principle” investigation of knob/eKnob phenotypes is likely to be reflected in other facets of host cell modification and can thus be seen as a proxy for such.

To conclude, in this study we show data suggesting that correct biogenesis of knobs in malaria-iRBCs is a complex process necessitating a number of proteins, the molecular identity of some of which remains enigmatic. Our data suggests that KAHRP, while obviously required for knob generation, may not directly provide a scaffold for knob structure. More importantly, our data also reveals that residual human HSP70 within the iRBC is involved in parasite-driven host cell modification processes. To our knowledge, this is the first time a host cell protein has been directly implicated in parasite virulence and protein transport processes, and this observation opens up exciting new avenues not only for intervention strategies, but also for the discovery of other novel cell biological phenomena.

[END]

[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1009969

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.