A system for functional studies of the major virulence factor of malaria parasites

Curation statements for this article:
  • Curated by eLife

    eLife logo

    eLife Assessment

    This study introduces an important approach using selection linked integration (SLI) to generate Plasmodium falciparum lines expressing single, specific surface adhesins PfEMP1 variants, enabling precise study of PfEMP1 trafficking, receptor binding, and cytoadhesion. By moving the system to different parasite strains and introducing an advanced SLI2 system for additional genomic edits, this work provides compelling evidence for an innovative and rigorous platform to explore PfEMP1 biology and identify novel proteins essential for malaria pathogenesis including immune evasion.

This article has been Reviewed by the following groups

Read the full article See related articles

Abstract

PfEMP1 is a variable antigen displayed on erythrocytes infected with the malaria parasite Plasmodium falciparum . PfEMP1 mediates binding of the infected cell to the endothelium of blood vessels, a cause of severe malaria. Each parasite encodes ∼60 different PfEMP1 variants but only one is expressed at a time. Switching between variants underlies immune evasion in the host and variant-specific severity of disease. PfEMP1 is difficult to study due to expression heterogeneity between parasites which also renders genetic modification approaches ineffective. Here, we used selection linked integration (SLI) to generate parasites all expressing the same PfEMP1 variant and genome edit the expressed locus. Moving this system from the reference strain 3D7 to IT4 resulted in PfEMP1 expressor parasites with effective receptor binding capacities. We also introduce a second version of SLI (SLI2) to introduce additional genome edits. Using these systems, we study PfEMP1 trafficking, generate cell lines binding to all major endothelial receptors, survey the protein environment from functional PfEMP1 in the host cell and identify new proteins needed for PfEMP1 mediated sequestration. These findings show the usefulness of the system to study the key virulence factor of malaria parasites.

Article activity feed

  1. eLife Assessment

    This study introduces an important approach using selection linked integration (SLI) to generate Plasmodium falciparum lines expressing single, specific surface adhesins PfEMP1 variants, enabling precise study of PfEMP1 trafficking, receptor binding, and cytoadhesion. By moving the system to different parasite strains and introducing an advanced SLI2 system for additional genomic edits, this work provides compelling evidence for an innovative and rigorous platform to explore PfEMP1 biology and identify novel proteins essential for malaria pathogenesis including immune evasion.

  2. Reviewer #1 (Public review):

    One of the roadblocks in PfEMP1 research has been the challenges in manipulating var genes to incorporate markers to allow the transport of this protein to be tracked and to investigate the interactions taking place within the infected erythrocyte. In addition, the ability of Plasmodium falciparum to switch to different PfEMP1 variants during in vitro culture has complicated studies due to parasite populations drifting from the original (manipulated) var gene expression. Cronshagen et al have provided a useful system with which they demonstrate the ability to integrate a selectable drug marker into several different var genes that allows the PfEMP1 variant expression to be 'fixed'. This on its own represents a useful addition to the molecular toolbox and the range of var genes that have been modified suggests that the system will have broad application. As well as incorporating a selectable marker, the authors have also used selective linked integration (SLI) to introduce markers to track the transport of PfEMP1, investigate the route of transport, and probe interactions with PfEMP1 proteins in the infected host cell.

    What I particularly like about this paper is that the authors have not only put together what appears to be a largely robust system for further functional studies, but they have used it to produce a range of interesting findings including:

    - Co-activation of rif and var genes when in a head-to-head orientation.

    - The reduced control of expression of var genes in the 3D7-MEED parasite line.

    - More support for the PTEX transport route for PfEMP1.

    - Identification of new proteins involved in PfEMP1 interactions in the infected erythrocyte, including some required for cytoadherence.

    In most cases the experimental evidence is straightforward, and the data support the conclusions strongly. The authors have been very careful in the depth of their investigation, and where unexpected results have been obtained, they have looked carefully at why these have occurred.

    (1) In terms of incorporating a drug marker to drive mono-variant expression, the authors show that they can manipulate a range of var genes in two parasite lines (3D7 and IT4), producing around 90% expression of the targeted PfEMP1. Removal of drug selection produces the expected 'drift' in variant types being expressed. The exceptions to this are the 3D7-MEED line, which looks to be an interesting starting point to understand why this variant appears to have impaired mutually exclusive var gene expression and the EPCR-binding IT4var19 line. This latter finding was unexpected and the modified construct required several rounds of panning to produce parasites expressing the targeted PfEMP1 and bind to EPCR. The authors identified a PTP3 deficiency as the cause of the lack of PfEMP1 expression, which is an interesting finding in itself but potentially worrying for future studies. What was not clear was whether the selected IT4var19 line retained specific PfEMP1 expression once receptor panning was removed.

    (2) The transport studies using the mDHFR constructs were quite complicated to understand but were explained very clearly in the text with good logical reasoning.

    (3) By introducing a second SLI system, the authors have been able to alter other genes thought to be involved in PfEMP1 biology, particularly transport. An example of this is the inactivation of PTP1, which causes a loss of binding to CD36 and ICAM-1. It would have been helpful to have more insight into the interpretation of the IFAs as the anti-SBP1 staining in Figure 5D (PTP-TGD) looks similar to that shown in Figure 1C, which has PTP intact. The anti-EXP2 results are clearly different.

    (4) It is good to see the validation of PfEMP1 expression includes binding to several relevant receptors. The data presented use CHO-GFP as a negative control, which is relevant, but it would have been good to also see the use of receptor mAbs to indicate specific adhesion patterns. The CHO system if fine for expression validation studies, but due to the high levels of receptor expression on these cells, moving to the use of microvascular endothelial cells would be advisable. This may explain the unexpected ICAM-1 binding seen with the panned IT4var19 line.

    (5) The proxiome work is very interesting and has identified new leads for proteins interacting with PfEMP1, as well as suggesting that KAHRP is not one of these. The reduced expression seen with BirA* in position 3 is a little concerning but there appears to be sufficient expression to allow interactions to be identified with this construct. The quantitative impact of reduced expression for proxiome experiments will clearly require further work to define it.

    (6) The reduced receptor binding results from the TryThrA and EMPIC3 knockouts were very interesting, particularly as both still display PfEMP1 on the surface of the infected erythrocyte. While care needs to be taken in cross-referencing adhesion work in P. berghei and whether the machinery truly is functionally orthologous, it is a fair point to make in the discussion. The suggestion that interacting proteins may influence the "correct presentation of PfEMP1" is intriguing and I look forward to further work on this.
    Overall, the authors have produced a useful and reasonably robust system to support functional studies on PfEMP1, which may provide a platform for future studies manipulating the domain content in the exon 1 portion of var genes. They have used this system to produce a range of interesting findings and to support its use by the research community.
    Finally, a small concern. Being able to select specific var gene switches using drug markers could provide some useful starting points to understand how switching happens in P. falciparum. However, our trypanosome colleagues might remind us that forcing switches may show us some mechanisms but perhaps not all.

  3. Reviewer #2 (Public review):

    Summary

    Croshagen et al develop a range of tools based on selection-linked integration (SLI) to study PfEMP1 function in P. falciparum. PfEMP1 is encoded by a family of ~60 var genes subject to mutually exclusive expression. Switching expression between different family members can modify the binding properties of the infected erythrocyte while avoiding the adaptive immune response. Although critical to parasite survival and Malaria disease pathology, PfEMP1 proteins are difficult to study owing to their large size and variable expression between parasites within the same population. The SLI approach previously developed by this group for genetic modification of P. falciparum is employed here to selectively and stably activate the expression of target var genes at the population level. Using this strategy, the binding properties of specific PfEMP1 variants were measured for several distinct var genes with a novel semi-automated pipeline to increase throughput and reduce bias. Activation of similar var genes in both the common lab strain 3D7 and the cytoadhesion competent FCR3/IT4 strain revealed higher binding for several PfEMP1 IT4 variants with distinct receptors, indicating this strain provides a superior background for studying PfEMP1 binding. SLI also enables modifications to target var gene products to study PfEMP1 trafficking and identify interacting partners by proximity-labeling proteomics, revealing two novel exported proteins required for cytoadherence. Overall, the data demonstrate a range of SLI-based approaches for studying PfEMP1 that will be broadly useful for understanding the basis for cytoadhesion and parasite virulence.

    Comments

    (1) While the capability of SLI to actively select var gene expression was initially reported by Omelianczyk et al., the present study greatly expands the utility of this approach. Several distinct var genes are activated in two different P. falciparum strains and shown to modify the binding properties of infected RBCs to distinct endothelial receptors; development of SLI2 enables multiple SLI modifications in the same parasite line; SLI is used to modify target var genes to study PfEMP1 trafficking and determine PfEMP1 interactomes with BioID. Curiously, Omelianczyk et al activated a single var (Pf3D7_0421300) and observed elevated expression of an adjacent var arranged in a head-to-tail manner, possibly resulting from local chromatin modifications enabling expression of the neighboring gene. In contrast, the present study observed activation of neighboring genes with head-to-head but not head-to-tail arrangement, which may be the result of shared promoter regions. The reason for these differing results is unclear although it should be noted that the two studies examined different var loci.

    (2) The IT4var19 panned line that became binding-competent showed increased expression of both paralogs of ptp3 (as well as a phista and gbp), suggesting that overexpression of PTP3 may improve PfEMP1 display and binding. Interestingly, IT4 appears to be the only known P. falciparum strain (only available in PlasmoDB) that encodes more than one ptp3 gene (PfIT_140083100 and PfIT_140084700). PfIT_140084700 is almost identical to the 3D7 PTP3 (except for a ~120 residue insertion in 3D7 beginning at residue 400). In contrast, while the C-terminal region of PfIT_140083100 shows near-perfect conservation with 3D7 PTP3 beginning at residue 450, the N-terminal regions between the PEXEL and residue 450 are quite different. This may indicate the generally stronger receptor binding observed in IT4 relative to 3D7 results from increased PTP3 activity due to multiple isoforms or that specialized trafficking machinery exists for some PfEMP1 proteins.

  4. Reviewer #3 (Public review):

    Summary:

    The submission from Cronshagen and colleagues describes the application of a previously described method (selection linked integration) to the systematic study of PfEMP1 trafficking in the human malaria parasite Plasmodium falciparum. PfEMP1 is the primary virulence factor and surface antigen of infected red blood cells and is therefore a major focus of research into malaria pathogenesis. Since the discovery of the var gene family that encodes PfEMP1 in the late 1990s, there have been multiple hypotheses for how the protein is trafficked to the infected cell surface, crossing multiple membranes along the way. One difficulty in studying this process is the large size of the var gene family and the propensity of the parasites to switch which var gene is expressed, thus preventing straightforward gene modification-based strategies for tagging the expressed PfEMP1. Here the authors solve this problem by forcing the expression of a targeted var gene by fusing the PfEMP1 coding region with a drug-selectable marker separated by a skip peptide. This enabled them to generate relatively homogenous populations of parasites all expressing tagged (or otherwise modified) forms of PfEMP1 suitable for study. They then applied this method to study various aspects of PfEMP1 trafficking.

    Strengths:

    The study is very thorough, and the data are well presented. The authors used SLI to target multiple var genes, thus demonstrating the robustness of their strategy. They then perform experiments to investigate possible trafficking through PTEX, they knock out proteins thought to be involved in PfEMP1 trafficking and observe defects in cytoadherence, and they perform proximity labeling to further identify proteins potentially involved in PfEMP1 export. These are independent and complimentary approaches that together tell a very compelling story.

    Weaknesses:

    (1) When the authors targeted IT4var19, they were successful in transcriptionally activating the gene, however, they did not initially obtain cytoadherent parasites. To observe binding to ICAM-1 and EPCR, they had to perform selection using panning. This is an interesting observation and potentially provides insights into PfEMP1 surface display, folding, etc. However, it also raises questions about other instances in which cytoadherence was not observed. Would panning of these other lines have been successfully selected for cytoadherent infected cells? Did the authors attempt panning of their 3D7 lines? Given that these parasites do export PfEMP1 to the infected cell surface (Figure 1D), it is possible that panning would similarly rescue binding. Likewise, the authors knocked out PTP1, TryThrA, and EMPIC3 and detected a loss of cytoadhesion, but they did not attempt panning to see if this could rescue binding. To ensure that the lack of cytoadhesion in these cases is not serendipitous (as it was when they activated IT4var19), they should demonstrate that panning cannot rescue binding.

    (2) The authors perform a series of trafficking experiments to help discern whether PfEMP1 is trafficked through PTEX. While the results were not entirely definitive, they make a strong case for PTEX in PfEMP1 export. The authors then used BioID to obtain a proxiome for PfEMP1 and identified proteins they suggest are involved in PfEMP1 trafficking. However, it seemed that components of PTEX were missing from the list of interacting proteins. Is this surprising and does this observation shed any additional light on the possibility of PfEMP1 trafficking through PTEX? This warrants a comment or discussion.

  5. Author response:

    Public Reviews:

    Reviewer #1 (Public review):

    One of the roadblocks in PfEMP1 research has been the challenges in manipulating var genes to incorporate markers to allow the transport of this protein to be tracked and to investigate the interactions taking place within the infected erythrocyte. In addition, the ability of Plasmodium falciparum to switch to different PfEMP1 variants during in vitro culture has complicated studies due to parasite populations drifting from the original (manipulated) var gene expression. Cronshagen et al have provided a useful system with which they demonstrate the ability to integrate a selectable drug marker into several different var genes that allows the PfEMP1 variant expression to be 'fixed'. This on its own represents a useful addition to the molecular toolbox and the range of var genes that have been modified suggests that the system will have broad application. As well as incorporating a selectable marker, the authors have also used selective linked integration (SLI) to introduce markers to track the transport of PfEMP1, investigate the route of transport, and probe interactions with PfEMP1 proteins in the infected host cell.

    What I particularly like about this paper is that the authors have not only put together what appears to be a largely robust system for further functional studies, but they have used it to produce a range of interesting findings including:

    - Co-activation of rif and var genes when in a head-to-head orientation.

    - The reduced control of expression of var genes in the 3D7-MEED parasite line.

    - More support for the PTEX transport route for PfEMP1.

    - Identification of new proteins involved in PfEMP1 interactions in the infected erythrocyte, including some required for cytoadherence.

    In most cases the experimental evidence is straightforward, and the data support the conclusions strongly. The authors have been very careful in the depth of their investigation, and where unexpected results have been obtained, they have looked carefully at why these have occurred.

    (1) In terms of incorporating a drug marker to drive mono-variant expression, the authors show that they can manipulate a range of var genes in two parasite lines (3D7 and IT4), producing around 90% expression of the targeted PfEMP1. Removal of drug selection produces the expected 'drift' in variant types being expressed. The exceptions to this are the 3D7-MEED line, which looks to be an interesting starting point to understand why this variant appears to have impaired mutually exclusive var gene expression and the EPCR-binding IT4var19 line. This latter finding was unexpected and the modified construct required several rounds of panning to produce parasites expressing the targeted PfEMP1 and bind to EPCR. The authors identified a PTP3 deficiency as the cause of the lack of PfEMP1 expression, which is an interesting finding in itself but potentially worrying for future studies. What was not clear was whether the selected IT4var19 line retained specific PfEMP1 expression once receptor panning was removed.

    This is a very interesting point. We do not have systematic long-term data for the Var19 line but medium-term data. After panning the Var19 line, the binding assays were done within 3 months without additional panning. The first binding assay was 2 months after the panning and the last binding assays three weeks later. While there is inherent variation in these assays that precludes detection of smaller changes, the last assay showed the highest level of binding, giving no indication for rapid loss of the binding phenotype. Hence, we can say that the binding phenotype appears to be stable for many weeks without panning the cells again and there was no indication for a rapid loss of binding in these parasites.

    Systematic long-term experiments to assess how long the Var19 parasites retain binding would be interesting, but given that the binding-phenotype appears to remain stable over many weeks, this would only make sense if done for a much longer time (6 months or more). Due to the time needed to carry out such an experiment this would not be practical to still include into the present study. But this might be advisable if the Var19 line is used in future experiments that go over extended periods of time. We intend to include a statement in the discussion of the revised manuscript to highlight that if long-term work with this line is planned, monitoring the binding phenotype and potentially re-panning might be advisable.

    (2) The transport studies using the mDHFR constructs were quite complicated to understand but were explained very clearly in the text with good logical reasoning.

    We are aware of this being a complex issue and are glad this was nevertheless understandable.

    (3) By introducing a second SLI system, the authors have been able to alter other genes thought to be involved in PfEMP1 biology, particularly transport. An example of this is the inactivation of PTP1, which causes a loss of binding to CD36 and ICAM-1. It would have been helpful to have more insight into the interpretation of the IFAs as the anti-SBP1 staining in Figure 5D (PTP-TGD) looks similar to that shown in Figure 1C, which has PTP intact. The anti-EXP2 results are clearly different.

    We realize the description of the PTP1-TGD IFA data and that of the other TGDs was rather cursory. We intend to amend this in the revision.

    (4) It is good to see the validation of PfEMP1 expression includes binding to several relevant receptors. The data presented use CHO-GFP as a negative control, which is relevant, but it would have been good to also see the use of receptor mAbs to indicate specific adhesion patterns. The CHO system if fine for expression validation studies, but due to the high levels of receptor expression on these cells, moving to the use of microvascular endothelial cells would be advisable. This may explain the unexpected ICAM-1 binding seen with the panned IT4var19 line.

    We agree with the reviewer that it is desirable to have better binding systems for studying individual binding interactions. As the main purpose of this paper was to introduce the system and show binding, we did not move to more complicated binding systems. However, we would like to point out that the CSA binding was done on receptor alone in addition to the CSA-expressing HBEC-5i cells and was competed successfully with soluble CSA. In addition, apart from the additional ICAM1-binding of the Var19 line, all binding phenotypes were conform with expectations. We therefore hope the tools used for binding studies are acceptable at this stage of introducing the system while future work interested in specific PfEMP1 receptor interactions are advised to use better systems, ideally including also endothelial organoid models, inhibitory antibodies and possibly domain competition. We intend to add a sentence to the discussion highlighting that future work using this system to study individual receptor-interactions could benefit from using optimized binding systems.

    (5) The proxiome work is very interesting and has identified new leads for proteins interacting with PfEMP1, as well as suggesting that KAHRP is not one of these. The reduced expression seen with BirA* in position 3 is a little concerning but there appears to be sufficient expression to allow interactions to be identified with this construct. The quantitative impact of reduced expression for proxiome experiments will clearly require further work to define it.

    This is a valid point. Clearly there seems to be some impact on binding when BirA* is placed in the extracellular domain (either through reduced presentation or direct reduction of binding efficiency of the modified PfEMP1). The exact impact on the proxiome is indeed difficult to assess. However, we hope that the general coverage of proteins proximal to PfEMP1 with the 3 PfEMP1-BirA* constructs will aid in the identification of proteins involved in PfEMP1 transport and surface display as illustrated with two of the hits targeted here.

    (6) The reduced receptor binding results from the TryThrA and EMPIC3 knockouts were very interesting, particularly as both still display PfEMP1 on the surface of the infected erythrocyte. While care needs to be taken in cross-referencing adhesion work in P. berghei and whether the machinery truly is functionally orthologous, it is a fair point to make in the discussion. The suggestion that interacting proteins may influence the "correct presentation of PfEMP1" is intriguing and I look forward to further work on this.

    We hope we future work will be able to shed light on this.

    Overall, the authors have produced a useful and reasonably robust system to support functional studies on PfEMP1, which may provide a platform for future studies manipulating the domain content in the exon 1 portion of var genes. They have used this system to produce a range of interesting findings and to support its use by the research community.
    Finally, a small concern. Being able to select specific var gene switches using drug markers could provide some useful starting points to understand how switching happens in P. falciparum. However, our trypanosome colleagues might remind us that forcing switches may show us some mechanisms but perhaps not all.

    Point noted! From non-systematic data with the Var01 line that has been cultured for extended periods of time (several years), it seems other non-targeted vars remain silent in our SLI “activation” lines but how much SLI-based var-expression “fixing” tampers with the integrity of natural switching mechanisms is indeed very difficult to gage at this stage. We intend to add a statement to the manuscript that even if mutually exclusive expression is maintained, it is not certain the mechanisms controlling var expression all remain intact.

    Reviewer #2 (Public review):

    Summary

    Croshagen et al develop a range of tools based on selection-linked integration (SLI) to study PfEMP1 function in P. falciparum. PfEMP1 is encoded by a family of ~60 var genes subject to mutually exclusive expression. Switching expression between different family members can modify the binding properties of the infected erythrocyte while avoiding the adaptive immune response. Although critical to parasite survival and Malaria disease pathology, PfEMP1 proteins are difficult to study owing to their large size and variable expression between parasites within the same population. The SLI approach previously developed by this group for genetic modification of P. falciparum is employed here to selectively and stably activate the expression of target var genes at the population level. Using this strategy, the binding properties of specific PfEMP1 variants were measured for several distinct var genes with a novel semi-automated pipeline to increase throughput and reduce bias. Activation of similar var genes in both the common lab strain 3D7 and the cytoadhesion competent FCR3/IT4 strain revealed higher binding for several PfEMP1 IT4 variants with distinct receptors, indicating this strain provides a superior background for studying PfEMP1 binding. SLI also enables modifications to target var gene products to study PfEMP1 trafficking and identify interacting partners by proximity-labeling proteomics, revealing two novel exported proteins required for cytoadherence. Overall, the data demonstrate a range of SLI-based approaches for studying PfEMP1 that will be broadly useful for understanding the basis for cytoadhesion and parasite virulence.

    Comments

    (1) While the capability of SLI to actively select var gene expression was initially reported by Omelianczyk et al., the present study greatly expands the utility of this approach. Several distinct var genes are activated in two different P. falciparum strains and shown to modify the binding properties of infected RBCs to distinct endothelial receptors; development of SLI2 enables multiple SLI modifications in the same parasite line; SLI is used to modify target var genes to study PfEMP1 trafficking and determine PfEMP1 interactomes with BioID. Curiously, Omelianczyk et al activated a single var (Pf3D7_0421300) and observed elevated expression of an adjacent var arranged in a head-to-tail manner, possibly resulting from local chromatin modifications enabling expression of the neighboring gene. In contrast, the present study observed activation of neighboring genes with head-to-head but not head-to-tail arrangement, which may be the result of shared promoter regions. The reason for these differing results is unclear although it should be noted that the two studies examined different var loci.

    The point that we are looking at different loci is very valid and we realize this is not mentioned in the discussion. In the revision we intend to add this as a possible reason for this discrepancy. As stated in the discussion, the head-to-head scenario was observed before in lines obtained with panning. However, given the rather few examples where this was analyzed, it is well possible that this varies with gene locus and we will make sure that the revised version of the manuscript will be careful to highlight that it is not clear how much this observation in our work can be generalized.

    (2) The IT4var19 panned line that became binding-competent showed increased expression of both paralogs of ptp3 (as well as a phista and gbp), suggesting that overexpression of PTP3 may improve PfEMP1 display and binding. Interestingly, IT4 appears to be the only known P. falciparum strain (only available in PlasmoDB) that encodes more than one ptp3 gene (PfIT_140083100 and PfIT_140084700). PfIT_140084700 is almost identical to the 3D7 PTP3 (except for a ~120 residue insertion in 3D7 beginning at residue 400). In contrast, while the C-terminal region of PfIT_140083100 shows near-perfect conservation with 3D7 PTP3 beginning at residue 450, the N-terminal regions between the PEXEL and residue 450 are quite different. This may indicate the generally stronger receptor binding observed in IT4 relative to 3D7 results from increased PTP3 activity due to multiple isoforms or that specialized trafficking machinery exists for some PfEMP1 proteins.

    We thank the reviewer for pointing this out, it is an interesting idea that the PTP3 duplication could be a reason for the superior binding of IT4. We intend to add this point to the discussion of the revision.

    So far it seems the PTP3 issue occurred only with Var19. The thought of an extra layer of control, particularly for PfEMP1 variants that might be associated with virulence such as Var19, is very attractive. At present, the manuscript alludes to the possibility of an extra layer of control in the discussion. As var-type specificity and existence of such mechanisms in vivo are so far not known we decided not to speculate on this.

    Reviewer #3 (Public review):

    Summary:

    The submission from Cronshagen and colleagues describes the application of a previously described method (selection linked integration) to the systematic study of PfEMP1 trafficking in the human malaria parasite Plasmodium falciparum. PfEMP1 is the primary virulence factor and surface antigen of infected red blood cells and is therefore a major focus of research into malaria pathogenesis. Since the discovery of the var gene family that encodes PfEMP1 in the late 1990s, there have been multiple hypotheses for how the protein is trafficked to the infected cell surface, crossing multiple membranes along the way. One difficulty in studying this process is the large size of the var gene family and the propensity of the parasites to switch which var gene is expressed, thus preventing straightforward gene modification-based strategies for tagging the expressed PfEMP1. Here the authors solve this problem by forcing the expression of a targeted var gene by fusing the PfEMP1 coding region with a drug-selectable marker separated by a skip peptide. This enabled them to generate relatively homogenous populations of parasites all expressing tagged (or otherwise modified) forms of PfEMP1 suitable for study. They then applied this method to study various aspects of PfEMP1 trafficking.

    Strengths:

    The study is very thorough, and the data are well presented. The authors used SLI to target multiple var genes, thus demonstrating the robustness of their strategy. They then perform experiments to investigate possible trafficking through PTEX, they knock out proteins thought to be involved in PfEMP1 trafficking and observe defects in cytoadherence, and they perform proximity labeling to further identify proteins potentially involved in PfEMP1 export. These are independent and complimentary approaches that together tell a very compelling story.

    Weaknesses:

    (1) When the authors targeted IT4var19, they were successful in transcriptionally activating the gene, however, they did not initially obtain cytoadherent parasites. To observe binding to ICAM-1 and EPCR, they had to perform selection using panning. This is an interesting observation and potentially provides insights into PfEMP1 surface display, folding, etc. However, it also raises questions about other instances in which cytoadherence was not observed. Would panning of these other lines have been successfully selected for cytoadherent infected cells? Did the authors attempt panning of their 3D7 lines? Given that these parasites do export PfEMP1 to the infected cell surface (Figure 1D), it is possible that panning would similarly rescue binding. Likewise, the authors knocked out PTP1, TryThrA, and EMPIC3 and detected a loss of cytoadhesion, but they did not attempt panning to see if this could rescue binding. To ensure that the lack of cytoadhesion in these cases is not serendipitous (as it was when they activated IT4var19), they should demonstrate that panning cannot rescue binding.

    These are very important points. Indeed, we had repeatedly attempted to pan 3D7 when we failed to get the SLI-generated 3D7 PfEMP1 expressor lines to bind, but this had not been successful. After the move to IT4 which readily bound we made no further efforts to understand why 3D7 does not bind but the fact that PfEMP1 is on the surface indicates this is not a PTP3 issue. Also, as the parent 3D7 could not be panned, we assumed it is not easily fixed.

    Panning the TGD lines: we see the reasoning for conducting panning experiments with the TGD lines, but on second thought we are unsure this should be attempted. The outcome might not be easily interpretable if panning leads to increased binding and considerable follow up analyses would be needed to define what has happened. The reason for this is that at least two forces will contribute to the selection in panning experiments with TGD lines that lost binding. Firstly, panning would work against the SLI of the TGD, resulting in a tug of war between the TGD-SLI and binding: a very low frequency of parasites can be expected to loop out the TGD plasmid and would normally be eliminated during standard culturing due to the SLI drug used for the TGD. These revertant cells would bind and the panning would enrich them (hence, panning and SLI are opposed in the case of a TGD abolishing binding). It is unclear how strong such an effect can be, but this might lead to mixed populations that complicate interpretations. The second selecting force are possible compensatory changes to restore binding. These can come in two flavors: reversal of potential independent changes that may have occurred in the TGD parasites and that are in reality causing the binding loss (the concern of the reviewer) or new changes to compensate the loss of the TGD target (in case the TGD is the cause of the binding loss). As both of the TGDs in the paper show some residual binding and have VAR01 on the surface to at least some extent, it is possible that new compensatory changes might indeed occur that indirectly increase binding again. In summary, even if more binding after panning of the lines occurs, it is not clear whether this is due to a compensatory change ameliorating the TGD or reversal of an unrelated change. The impact of repeated panning against SLI is also unknown. To determine the cause, the panned TGD lines would need to be subjected to a complex and time-consuming analysis (WGS, RNASeq, possibly Maurer’s clefts IFA phenotype) to find out whether they had an unrelated chance change that was reverted or a new compensatory change that helps binding.

    The detection of VAR01 on the surface of these TGDs speaks against a PTP3 effect. While we can’t fully exclude other changes in the TGDs that might affect binding, we conducted WGS which did not show any obvious alterations that could be responsible. To fully exclude loss of ptp3 expression as the reason as seen with Var19 (something we would not have seen in the WGS if it is only due to a transcriptional change), we intend to carry out RNASeq with the two TGD lines. The third TGD mentioned by the reviewer (targeting ptp1) was a positive control of a known PfEMP1 trafficking protein, so we assume this does not need to be further validated.

    (2) The authors perform a series of trafficking experiments to help discern whether PfEMP1 is trafficked through PTEX. While the results were not entirely definitive, they make a strong case for PTEX in PfEMP1 export. The authors then used BioID to obtain a proxiome for PfEMP1 and identified proteins they suggest are involved in PfEMP1 trafficking. However, it seemed that components of PTEX were missing from the list of interacting proteins. Is this surprising and does this observation shed any additional light on the possibility of PfEMP1 trafficking through PTEX? This warrants a comment or discussion.

    This is an interesting comment and we agree we should have discussed this. A likely reason why PTEX components are not picked up as interactors is that BirA* is expected to become unfolded when it passes through the channel and in that state can’t biotinylate. Labelling likely would only be possible if PfEMP1 lingered at the PTEX translocation step before BirA* became unfolded to go through the channel which we would not expect under physiological conditions. We intend to add a sentence to the discussion why we think PTEX components would not be detected in our BioIDs even if PfEMP1 passes through it but that this might also be an argument against it passing through PTEX.