Author response:
The following is the authors’ response to the original reviews.
Public Reviews:
Reviewer #1 (Public review):
Summary:
The extent to which P. falciparum liver stage parasites export proteins into the host cell is unclear. Most blood-stage exported proteins tested in liver stages were not exported. An exception is LISP2, which is exported in P. berghei but not P. falciparum liver stages. While the machinery for export is present in liver stages, efforts to demonstrate export have so far been mostly unsuccessful. Parasite proteins exported during the liver stage could be presented by MHC and thereby become the target of immune control, an incentive to study liver stage export and identify proteins exported during this stage. However, particularly for P. falciparum, it is very difficult to study liver stages.
This work studies LSA3 in P. falciparum blood and liver stages. The authors show that this protein is exported into the host cell in blood stages, but in liver stages, no or only very little export was detected. A disruption of LSA3 reduced liver stage load in a humanized mouse model, indicating this protein contributes to efficient development of the parasites in the liver.
The paper also studies the localization of LSA3 in blood stages and uses a known inhibitor to show that it is processed by plasmepsin 5, a protease important for protein trafficking. The work also shows that LSA3 is not needed for passage through the mosquito.
Strengths:
The main strength of this work is the use of the humanized mouse model to study liver stages of P. falciparum, which is technically challenging and requires specialized facilities. The biochemical analysis of LSA3 localization and processing by plasmepsin 5 is thorough and mostly overcame adverse issues such as a cross-reactive antibody and the negative influence of the GFP-tag on LSA3 trafficking. The mosquito stage analysis is also notable, as these kinds of studies are difficult with P. falciparum. However, there was no evidence for a function of LSA3 in mosquito stages.
We thank the reviewer for their perspective on the strengths of the study.
Weaknesses:
The cross-reactivity of the antibody, together with the co-infection strategy, prevents reliable assessment of LSA3 localization in liver stages. Despite this, it seems LSA3 is not exported in liver stages, and the paper does not bring us closer to the original goal of finding an exported liver stage protein.
While the localization analysis in blood stages is well done and thorough, the advance is somewhat limited. LSA3 may be in structures like J dots, but this hypothesis was not tested. Although parasites with a disrupted LSA3 were generated, the function of this protein was not explored. Given that a previous publication found some inhibitory effect of LSA3 antibodies on blood stage growth, a comparison of the growth of the LSA3 disruption clones with the parent would have been very welcome and easy to do. At this point, LSA3 is one more of many proteins exported in blood stages for which the function remains unclear.
It might be possible to refine some of the conclusions. The impact on liver stage development is interesting, but which phase of the liver stage is affected, and the phenotype remains largely unknown. The co-infection (WT together with LSA3 mutant) has the advantage of a direct comparison of the mutant with the control in the same liver, but complicates phenotypic analysis if the LSA3 antibody is also cross-reactive in liver stages. This issue adds a question mark to the shown localization and precludes phenotypic comparisons. The authors write that they do not know if the cross-reactive protein is expressed at that stage. But this should be immediately evident from the mixed WT/mutant infection. If all cells are positive for LSA3, there is a cross-reaction. If about half of the cells are negative, there isn't. In the latter case, the localization shown in the paper is indeed LSA3, and morphological differences between WT and LSA3 disruption could be assessed without additional experiments.
We thank the reviewer for their comments. While the LSA3-C antibody may cross-react with another parasite protein(s) in addition to binding LSA3 itself, we observed no strong evidence that this antibody localized beyond the liver-stage PVM, indicating that LSA3 is likely not targeted to the host cell compartment. We cannot exclude the possibility that a domain of LSA3 faces the hepatocyte lumen from this membrane and thus may be considered exported though follow-up studies are required (and are very challenging) to answer it. The phenotype of the NF54 DLSA3 mutant generated in this study at the blood stage was underway (by an independent lab in collaboration with us) and we are happy to disclose that the outcomes were recently published (May 2026) in an accompanying manuscript (PMID: 41135800). We completely agree that independently infected humanized mice would be helpful to address further remaining questions around the localization and temporal phenotype for LSA3 essentiality, which again will require follow up studies. In the present study, we intended to address whether LSA3 is important functionally, as this had not been reported.
Significance:
The conclusion from the paper that "our study presents just the second PEXEL protein so far identified as important for normal P. falciparum liver-stage development and confirms the hypothesized potential of exported proteins as malaria vaccine candidates" is partially misleading. Neither LISP2 nor LSA3 seems to be exported in P. falciparum liver stages, and we can't confirm the potential of vaccines with proteins exported in this stage. LSA3 is still important and may still be the target of the immune response, but based on this work, probably not due to export in liver stages.
We thank the reviewer for the comment. We would like to emphasize the possibility that proteins localized at the PVM may be considered exported ‘if’ part or all of the protein (eg, a domain) faces the host cell lumen from the hepatocyte. We have not shown this to be the case for LSA3 or LISP2 but that possibility remains open. Nonetheless, LISP2 is exported (by P. berghei liver stages) and LSA3 is exported (by P. falciparum blood stages); both are exported proteins.
Reviewer #2 (Public review):
Summary:
Immunogenic Plasmodium falciparum proteins that could be targeted to prevent parasite development in the liver are of significant interest for novel anti-malarial vaccine development. In this study, McConville et al evaluate the trafficking and functional importance of LSA3, a protein expressed in the blood and liver stages and previously shown to provide protection in immunized chimpanzees. LSA3 contains a PEXEL motif, but the authors have previously shown that this protein does not appear to be exported beyond the PVM in the liver stage (McConville et al, PNAS 2024). However, LSA3 trafficking and functional importance have not been comprehensively evaluated across stages. In the present study, the authors find that blood stage LSA3 undergoes PEXEL processing, and a portion of the protein is exported into the erythrocyte, where it localizes to punctate structures distinct from Maurer's clefts. Using a knockout mutant, LSA3 is shown to be dispensable for blood and mosquito stages but important to liver-stage development. Collectively, these results validate LSA3 as a liver-stage target and place it among several other PEXEL proteins that display differential trafficking beyond the PVM in the erythrocyte but not the hepatocyte.
Strengths:
The authors present a thorough analysis of LSA3 trafficking in the blood stage. PEXEL processing by Plasmepsin 5 is clearly demonstrated through a combination of mini LSA3-GFP reporters and Plasmepsin 5 inhibitors. Importantly, an LSA3 knockout mutant is used to show that the LSA3-C anti-sera also react with additional, unidentified parasite proteins in the blood stage. Nonetheless, comparison between the WT and KO parasites clearly indicates that a portion of LSA3 is exported into the erythrocyte, which is further supported by protease-protection assays with fractionated iRBCs. This contrasts with the liver stage, where LSA3 does not appear to traffic beyond the PVM, similar to what has been observed for other PEXEL proteins in the rodent malaria model.
This study provides the first direct analysis of LSA3 function by reverse genetics, showing this protein is important for liver stage development in chimeric human liver mice. Several PEXEL proteins in P. berghei have been shown to be exported into the host cell in the blood stage, but do not appear to cross the PVM in the liver stage. These observations reinforce that even without detectable export into the hepatocyte, PEXEL proteins play critical roles during liver stage development.
We thank the reviewer for their feedback regarding the strengths of the paper.
Weaknesses:
A previous study reported that anti-LSA3 antibodies inhibit blood-stage growth, suggesting a role for LSA3 during erythrocyte infection. While the authors carefully evaluate the LSA3 mutant in mosquito and liver stages, the impact on blood stage fitness is not tested. While the knockout shows LSA3 is not essential in the blood stage, its importance during erythrocyte infection remains unclear.
The authors previously reported that anti-LSA3-C signal in the liver stage localizes within the parasite and at the parasite periphery but is not exported into the hepatocyte. In the present study, it is shown that anti-LSA3-C reacts with other parasite proteins beyond LSA3 in the blood stage, and this may also occur in the liver stage. However, since liver-stage IFAs were only performed on samples co-infected with both WT and ∆LSA3 parasites, non-specific anti-LSA3C reactivity at this stage could not be determined, and the localization of LSA3 in the liver stage remains somewhat unclear.
We thank the reviewer for their comments. The phenotype of the NF54 DLSA3 mutant generated in this study at the blood stage was underway (by an independent lab in collaboration with us) and we are happy to disclose that the outcomes were recently published (May 2026) in an accompanying manuscript (PMID: 41135800). While the LSA3-C antibody may cross-react with another parasite protein(s) in addition to binding LSA3 itself, we observed no strong evidence that this antibody localized beyond the liver-stage PVM, indicating that LSA3 is likely not targeted to the host cell compartment. We cannot exclude the possibility that a domain of LSA3 faces the hepatocyte lumen from this membrane and thus may be considered exported though follow-up studies are required (and are very challenging) to answer it. We completely agree that independently infected humanized mice would be helpful to address further remaining questions around the localization and temporal phenotype for LSA3 essentiality, which again will require follow up studies. In the present study, we intended to address whether LSA3 is important functionally, as this had not been reported.
Reviewer #3 (Public review):
Summary:
This manuscript provides a comprehensive characterization of the Plasmodium falciparum protein LSA3, combining biochemical, genetic, and in vivo approaches. The authors convincingly demonstrate that LSA3 is expressed during liver stage infection and that disruption of the gene leads to a modest but reproducible reduction in liver stage parasite load in humanized mice.
Strengths:
Their biochemical and cell biological analysis of blood stages provides strong evidence that LSA3 is exported to the infected erythrocyte, and the detailed analysis of its PEXEL motif processing is well executed.
We thank the reviewer for their comments.
Weaknesses:
The study suggests LSA3 as one of only two known P. falciparum PEXEL proteins contributing to this stage, although there is no evidence for the export beyond the vacuolar membrane. Several key conclusions, particularly regarding antibody specificity, localization in liver stage parasites, and the interpretation of the phenotypic data, are not fully supported by the current experiments.
We understand the reviewer’s points. We agree that there is no evidence provided that LSA3 is targeted beyond the PVM; whether any of the protein faces the hepatocyte cytosol is unknown (and challenging to conduct) but this possibility remains plausible. LISP2- and LSA3deficient liver stages are less fit than parental controls and thus we stand by the conclusion that they are the two so far identified P. falciparum PEXEL proteins that are important for liver-stage development.
Recommendations for the authors:
Reviewer #1 (Recommendations for the authors):
(1) Line 163 says: "Altogether, this demonstrates that LSA3 is important but not critical for blood stage growth of P. falciparum": this is based on the cited Morita et al., 2017. However, previously LSA3 was considered dispensable based on a knock out in 3D7 (Maier et al., 2008; PMID: 18614010). Given that the authors generated a mutant for this work, it would be straightforward to test growth and clarify the importance of LSA3 in blood stages. If important, the analysis of the location and transport of LSA3 in blood stages would immediately become more relevant. Maybe the data for this is already in the paper: the number of stage V gams was similar between mutant and control (Figure 4A). If this was calculated from the total number of asexual starting parasitemia, it includes blood stage growth, and it can be assumed that there is no growth defect in the mutant in the blood stages. If the number of stage 5 gams was calculated from the number of committed schizonts/rings, nothing can be said about blood stage growth, and asexual blood stage growth should be tested in specific experiments.
We thank the reviewer for raising the function of LSA3 in blood stages and agree it was an obvious omission, though for good reason - a separate, collaborative study was underway. While this eLife preprint was in revision, our accompanying manuscript on the blood stage was published, showing the characterization of our NF54 DLSA3 mutant during blood-stage growth (PMID:41135800). The findings are now summarized and the citation included in the revised version of this preprint.
Manuscript line 105: "although, notably, functional characterization of lsa3 deletion mutants has not yet been reported to confirm an important function": at least in blood stages, it was reported to be dispensable, see above. The corresponding study (Maier et al., 2008, PMID: 18614010) could be cited in that context.
The citation of PMID18614010 and 39913589 have now been added and we thank the reviewer.
(2) Some questions central to the conclusions of this paper remain because it was unclear whether the serum did indeed detect LSA3 in the liver or not. It would be easy to check if all cells from the WT/Mutant mix experiment show LSA3 signal (this would mean it cross-reacts) or if only about half are positive (the mutants would be negative if there is no cross-reaction). This would be important to mention for Figure 5 because, at present, it is not known that what is labeled by the LSA3-C antibody in these images is (only) LSA3.
We thank the reviewer for this point and completely understand. We did check this via microscopy of liver sections co-infected with LSA3 mutant and control liver-stage parasites as we shared the reviewers line of enquiry. Unfortunately we could not detect parasites without LSA3 signal at the 5-day post-infection time point. This type of analysis does sound straightforward on paper but in reality is more challenging owing to several factors i) identifying sufficient individual parasites in an entire liver by microscopy can be challenging and variable from lobe to lobe and mouse to mouse, ii) the number of parasites required for a meaningful statistical analysis is increased due to coinfection of the liver (see Figure 4B as an illustration of this), iii) day 5 is a rather late liver-stage time point and so if there was a growth defect the defective parasites may be very small or sparse, iv) we cannot exclude that the LSA3 antibody may cross-react at the liver-stage, v) definitive conclusions are thus challenging and we feel require individual co-infections to be clear in the future. Nonetheless, the detailed qRT-PCR analyses identify a significant reduction in DLSA3 parasite liver load on day 5, indicating this protein is important for the human malaria parasite’s growth within human hepatocytes.
(3) It is also unclear which parasites were imaged in Figure 5. The text of the results states that NF54 liver stages were used, but later: "As we employed a co-infection strategy to assess the essentiality of LSA3 versus NF54 in mice, we could not perform IFAs on individually infected mice in this study to validate the specificity of LSA3-C at the liver-stage". The legend says NF54 sporozoites on day 5 post-infection were used. I suspect it was a WT/mutant mix, in which case the above applies, and in the absence of cross-reactivity, half of the cells should be LSA3-C negative. If this is not the case, the localization in the liver becomes dubious.
We apologize for the confusion and have corrected this. In Figure 5, we utilized liver sections from NF54-infected humanized mice that were stored at -80 C from a previously published study (McConville et al, PNAS 2024). Ideally, we would validate the specificity of LSA3 antibodies at the liver-stage using liver sections containing only DLSA3 parasites however the number of mice available was limited and the samples available to us also contained the Control line for qRT-PCR analyses (the co-infection strategy). As mentioned above, we couldn’t distinguish between these two strains by IFA at the time point analysed and this precluded us unequivocally validating the LSA3-C specificity in the liver-stage; however it cannot be excluded that the signal observed at the PVM is indeed LSA3. We are currently focusing research efforts on obtaining more humanised mice to answer this.
Minor:
(1) Introduction: Before the part on the PEXEL motifs, there are almost no references; please add references for all statements.
We have added references.
(2) Figure 1B is unclear regarding which part of the gene was deleted. The system used would permit a complete gene deletion, but the homology flanks seem to be within LSA3. If parts of the gene are left, the 75 kDa on the western blots might be a degradation product arising from both the truncated and the full-length protein. Please clarify in the sketch exactly where the homology flanks are, with respect to the start and stop of the gene.
The LSA3 gene was disrupted using the flanks as shown. The DHFR selection cassette comprises its own promoter and terminator such that insertion into the coding sequence completely disrupts expression of the protein thereafter, including the C-terminus within which the LSA3-C antibody binds. The new LSA3-T antibody described in our recently published accompanying manuscript that binds more N-terminally than LSA3-C also does not label the truncated protein. The original 5’ and 3’ flanks used for integration of the disrupted LSA3 allele by double cross-over recombination were then looped out into the original knockout plasmid and this was negatively selected against using exogenous 5-fluorocytidine (5-FC) via the suicide gene cassette CDUP (cytosine deaminase and uracil phosphoribosyl transferase that also contains a 5’ promoter and 3’UTR terminating element) in the construct. These features should provide clarification and have now been indicated in the figure and legend.
(3) Line 161: Replace was with were.
Corrected.
(4) Figure 2, 224: Why do the authors think LSA3 must be in the luminal leaflet of the PVM as opposed to the outer leaflet of the plasma membrane?
Several pieces of evidence combined led us to this conclusion in Figure 2B. i) if LSA3 was on the outer PVM leaflet, it would be substantially degraded in the EQT Pellet + PK fraction but a substantial population remained insensitive to PK, indicating much of the total protein pool was protected by the PVM (and possibly the parasite membrane; PM), ii) yet saponin, which leaves the PM intact, allowed PK to access and almost completely degrade LSA3 (see Saponin Pellet + PK), indicating that a substantial population of LSA3-C is located inside the boundary of the PVM, and this is membrane associated as saponin did not liberate it, rather, it remained in the Saponin Pellet before PK was added, iii) the TX-100 Super fraction confirmed LSA3 is membrane associated, as more is present in the TX-100 Super than the Saponin Super fractions, iv) if LSA3 was inside the PM, the Saponin Pellet fraction should be resistant to PK (as was the case for the cross-reactive band indicated with a red asterisk) but LSA3 (green asterisk) in the Saponin Pellet was PK sensitive. Altogether, our best conclusion from these data is that LSA3 is likely to be PVM associated with the LSA-C-binding domain facing internal to the PV, and a fraction is also exported beyond the PVM into the erythrocyte.
(5) Line 245: GFP core "derived from digestion of the reporter in the food vacuole, which confirmed it was secreted from the parasite". I wonder if the amount of GFP "core" really can be used as evidence for secretion, and its amount can be compared between experiments. Did the author quantify this for the full-length protein to get a proportion per sample?
Use of GFP core to measure defects in P. falciparum GFP reporter secretion has been described previously (for example PMID:23387285 and 35906227). The comparison the reviewer asked for is an interesting and important question: however the control would be to compare the ratio of GFP core to uncleaved in the control lanes as well, which is not possible to do since the full-length protein is digested by plasmepsin V in the native PEXEL versions of the experiments (mLSA3-GFP in the first blot, Vehicle in the second blot) leaving no full-length protein to compare to. It stands to reason that inhibition of N-terminal processing results in less protein removal from the membrane (ER or COPII vesicle or PM) resulting in less secretion out of the parasite for retrograde transport to the food vacuole with cytostomal vacuoles (analogous to plasmepsin II). In the food vacuole, the chimeras are in normal cases digested by proteases back to the GFP core that is resistant to cleavage and evident as GFP core on the immunoblots (PMID:10775264 and 14709539 and 19055692 and 20130643).
(6) Figure 3 has the word plasmid in two lanes. In Figure 3E, amend the labelling of the blots.
We apologize for the formatting error in converting the figures to PDF during the original submission and thank the reviewer for the suggestion. This has now been corrected.
(7) Lines 266/271/284: "live IFAs", live immunofluorescence assay. Does this mean an antibody was given to living parasites?
The correct term is live microscopy and this has been corrected.
(8) Does Figure 6A fit with the data in Figure 6B? It seems 6B has a milder phenotype than 6A.
We thank the reviewer for the question. Yes the data directly correspond to each other and are represented in two ways: Panel A shows the qRT-PCR raw data for liver load of each parasite strain per humanized mouse using a scientific scale on the y-axis. Panel B shows that magnitude of the DLSA3 defect as a percentage of the total liver load per mouse:
% total parasite liver load = ( strain 1 or strain 2 liver load ) x100
sum of strain 1 + strain 2 liver loads
The intent of showing both data is to convey the correct magnitude of the difference in two ways to assist the reader in understanding the true defect, both are accurate and both are statistically significant. In revision we detected mislabelling of humanized mouse 2 and 3 in the original graphs that has now been corrected and we sincerely thank the reviewer for helping us identify this error.
(9) Line 482: Please add references for this debate.
These have been added.
Reviewer #2 (Recommendations for the authors):
Major Comments:
(1) In general, the authors have taken care not to overstate conclusions from their study. Nonetheless, while not technically inaccurate, the title might misleadingly suggest LSA3 is exported in the liver stage (this was my initial impression on reading it until I looked at the data). I suggest the authors revise the title to avoid confusion by clarifying that export was only observed in the blood stage.
We sincerely appreciate the reviewer’s point. As this article was posted as a preprint that has now been cited several times, we have carefully weighed the comment and in the end decided to retain the current title for the above reason.
(2) While the ability to generate the ∆LSA3 parasites clearly shows that the protein is not essential in the blood stage, the impact on parasite fitness is never tested but simply assumed (for instance, in lines 163-164: "...this demonstrates that LSA3 is important...for blood-stage growth..."). Do the ∆LSA3 parasites have a fitness defect in the blood stage consistent with the previous GIA data that would support this claim? Since the rabbit anti-LSA3-C antibodies produced by Morita et al did not have GIA activity against the blood stage, it is possible that the GIA observed with the human and mouse antibodies might have been due to reactivity with a different protein. If ∆LSA3 does cause a fitness defect, it would be interesting to know if the endogenous GFP-tagged line, which alters protein trafficking/membrane association, also produces this effect.
We agree with the reviewer and would like to clarify that this omission was not intended to create confusion but was by design, due to a separate collaborative study that was underway to address such questions. While this eLife preprint was in revision, our accompanying manuscript on characterising NF54 DLSA3 at the blood stage was published (PMID:41135800). The findings are now summarized and the citation included in the revised version of this eLife preprint. In sum, LSA3 is not critical for erythrocyte invasion but its deletion perturbs the rate and efficiency of merozoite invasion, at the step(s) of resealing of the PVM/host cell, resulting in aberrant accole forms that protrude from the infected erythrocyte.
(2) Figure 1D: While the images are compelling and I don't doubt the claim that LSA3 is exported in the blood stage (also supported by the fractionation/Pk experiments), the authors should provide quantification of the difference in exported signal between the WT and ∆LSA3 parasites in these IFAs to rigorously support this conclusion. Also, please include details about how many independent experiments are represented by the microscopy data throughout the manuscript (Figures 1, 2, 3, and 5).
We understand the reviewer’s request and wish to indicate that the export signal was absent in all cells infected with DLSA3 that was imaged. The microscopy performed was from n=2-3 experiments except for Figure 5 which was from n=1 humanized mouse per time point in which multiple EEFs from the liver were imaged. This has been indicated in the figure legends.
(3) Careful inspection of the z-series images in Figure 5A shows that most of the LSA3-C signal seen outside the PVM (beyond the boundary delineated by EXP1) is closely associated with DAPI puncta, suggesting these are merozoites. Together with the prominent gap in the EXP1 signal, this suggests the schizont has already ruptured. Thus, anti-LSA3-C signal beyond the PV seems best explained as coming from merozoites or other material released by PV rupture, not from export across the PVM, and this should be added to the text in place of comments about localization to PV extensions or potential export (lines 358-359, 422-423).
We do appreciate the reviewer’s careful eye and caution and are in complete agreement. We have added the comment as requested.
Minor Comments:
(1) The authors may want to denote the disordered repeat region in the LSA3 schematic in Figure 1A that is mentioned in the text.
We have added the residue boundaries of the predicted domain from AlphaFold into both the schematic and the text and included a link to the LSA3 pages in PlasmoDB and
AlphaFold in the Methods section.
(2) The authors use rabbit anti-LSA3-C antibodies previously generated by Morita et al. These polyclonal antibodies were raised against a recombinant C-terminal region of LSA3 (residues 750-1433), but the schematic in Figure 1A indicates the antibodies recognize a smaller region between residues 1154-1433. Please adjust the figure accordingly, or if this is not the same antiLSA3-C antibody reported by Morita, please provide details about its production.
The figure is corrected.
(3) The authors use Alphafold to identify a region of LSA3 with similarity to the substrate binding domain of DnaK, but the data is not shown. Please include the Alphafold prediction in supplementary figures and provide information about how the predicted structural homology was determined.
We have added a link to the AlphaFold page for PF3D7_0220000 in the methods.
(4) The schematic in Figure 1B indicates that the DHFR cassette was inserted at an internal site within the lsa3 gene. If this is the case, it seems possible that an N-terminal portion of the protein is still expressed, but I was unable to find details about the boundaries of the homology flanks to determine the precise insertion site. Please clarify the knockout strategy and indicate the specific insertion site.
The LSA3 gene was disrupted using the flanks as shown. The DHFR selection cassette comprises its own promoter and terminator such that insertion into the coding sequence completely disrupts expression of the protein thereafter, including the C-terminus within which the LSA3-C antibody binds. The new LSA3-T antibody described in our recently published accompanying manuscript that binds more N-terminally than LSA3-C also does not label the truncated protein. The original 5’ and 3’ flanks used for integration of the disrupted LSA3 allele by double cross-over recombination were then looped out into the original knockout plasmid and this was negatively selected against using exogenous 5-fluorocytidine (5-FC) via the suicide gene cassette CDUP (cytosine deaminase and uracil phosphoribosyl transferase that also contains a 5’ promoter and 3’UTR terminating element) in the construct. These features should provide clarification and have now been indicated in the figure and legend.
(5) Line 162: I think this should read "antibodies that react with LSA3 were...".
Corrected.
(6) Figure 1D: The merge with the transmitted light channel is missing for the third panel in the ∆LSA3 IFAs. Also, please define the scale bar length in the legend.
Corrected.
(7) Lines 744-746: The IFA fixation panel order description (top, bottom) in the Figure 2A legend is reversed from what is shown in the actual figure. Also, please define the scale bar length.
Corrected.
(8) Lines 184-186: Since the fractionation/PK protection assays suggest most of LSA3 is in the PV, it would be interesting to know if the strong peripheral/PV signal observed in the PFA-fixed IFAs in Figure 2A is also present in the ∆LSA3 parasites, or is this non-specific?
Thank you for the suggestion. We agree this would be an interesting result to know but do not have the capacity at the present time.
(9) Lines 219-225: It is unclear to me why these results are interpreted to suggest that the majority of LSA3 is peripherally associated with the luminal leaflet of the PVM. Wouldn't an integral membrane configuration in the PVM (with the C-terminus facing the host cytosol) or PPM (with the C-terminus facing the parasite cytosol) also account for the data? Adding a carbonate extraction would help clarify this point.
Several pieces of evidence combined led us to this conclusion in Figure 2B. i) if LSA3 was on the outer PVM leaflet, it would be substantially degraded in the EQT Pellet + PK fraction but a substantial population remained insensitive to PK, indicating much of the total protein pool was protected by the PVM (and possibly the parasite membrane; PM), ii) yet saponin, which leaves the PM intact, allowed PK to access and almost completely degrade LSA3 (see Saponin Pellet + PK), indicating that a substantial population of LSA3-C is located inside the boundary of the PVM, and this is membrane associated as saponin did not liberate it, rather, it remained in the Saponin Pellet before PK was added, iii) the TX-100 Super fraction confirmed LSA3 is membrane associated, as more is present in the TX-100 Super than the Saponin Super fractions, iv) if LSA3 was inside the PM, the Saponin Pellet fraction should be resistant to PK (as was the case for the cross-reactive band indicated with a red asterisk) but LSA3 (green asterisk) in the Saponin Pellet was PK sensitive. Altogether, our best conclusion from these data is that LSA3 is likely to be PVM-associated with the LSA-C-binding domain facing internal to the PV, and a fraction is also exported beyond the PVM into the erythrocyte. If the question is whether LSA3 is an integral PVM protein, we agree that use of carbonate in the future would answer that question.
(10) Figures 3D and E: There are some problems with some of the text wrapping in these panels.
We apologise, this was a formatting issue as the manuscript was converted to PDF.
We have corrected this error.
(11) Line 422-423: In fact, the Z-sections shown in Figure 5 appear to indicate that the LSA3-C signal is predominantly located within the parasite, not at the PVM.
We do appreciate the reviewer’s careful eye and caution and are in complete agreement. We have corrected the final conclusion to be more accommodating of this.
(12) Lines 468-470: Since cross reactivity of anti-LSA3-C is substantial in the blood stage but was not defined in the liver stage by analysis of unmixed infections, how do the authors know that they were not observing ∆LSA3 parasites in their IFAs? I think what they mean here is that parasites lacking anti-LSA3-C reactivity were not observed, which is an important distinction.
The reviewer is correct and this has been corrected.
(13) Lines 478-479: The authors should also mention that the P. berghei PEXEL proteins evaluated in Fougere et al are exported in the blood stage, similar to LSA3. Moreover, other studies have shown something similar for additional endogenous PEXEL proteins or reporters in P. berghei (PMIDs 22329949, 26347246, 34956312).
We have added the additional text regarding export into the infected erythrocyte and the reference to IBIS1.
(14) Line 491: The data here don't support that LSA3 is "required" for liver stage development, only that it is important to it. Since the authors have not defined the cross-reactivity of anti-LSA3C in unmixed infections, it is not clear that ∆LSA3 parasites are arrested early in the liver stage, only that they show a reduced number of genome copies relative to the parental control.
We have amended the sentence to “required for normal liver stage development”.
(15) Line 530: I think NGF54 should be NF54.
Corrected.
Reviewer #3 (Recommendations for the authors):
(1) Antibody specificity in liver stage IFA experiments:
The specificity of the anti-LSA3 antiserum (LSA3-C) used in liver stage IFA is not fully convincing. While the KO parasites were used effectively to validate specificity in blood stages, the same is not true for liver stages.
(a) It is essential to repeat IFA with ΔLSA3 parasites in liver stage infections to determine whether the observed PVM staining is truly specific.
We appreciate the reviewer’s point, however at a cost of over $5000 per humanized mouse, we do not have the capacity to conduct this experiment at the present time. We highlight that, as the blood stage IFAs confirmed the specificity of LSA3-C for LSA3, the possibility remains open that LSA3 is specifically recognized at the PVM.
(b) If the antibody is the same polyclonal serum used in Morita et al. (2017), why did the authors not employ a monoclonal antibody, which they presumably have access to and which would provide greater specificity?
We have included new data confirming that LSA3 is exported using LSA3-T, in addition to LSA3-C.
(c) Given that rabbit antisera often show non-specific staining at the PVM in liver stage parasites, co-localization with PVM markers is not sufficient. Inclusion of the ΔLSA3 parasites in liver stage IFA is critical. It will also show whether there is any cross-reaction of the antiserum in liver stage parasites, as seen by IFA for blood stage parasites.
We thank the reviewer for their feedback.
(d) To validate the serum further, the authors should infect HC-04 cells in vitro with GFP-LSA3 parasites and stain with LSA3-C to confirm overlap between the tagged protein and the antibody signal.
We thank the reviewer for their feedback.
(e) For higher-resolution co-localization, expansion microscopy - now commonly used even in malaria research - would substantially improve the analysis.
We thank the reviewer for their feedback.
(2) The localization of LSA3 in this study differs notably from Morita et al. 2017, who reported localization to dense granules in merozoites and staining in ring-stage parasites at the PVM.
(a) The authors confirm DG localization, but they do not examine ring-stage parasites. They should include the IFA of ring stages to clarify whether they can replicate the previous findings.
We thank the reviewer for their feedback.
(b) Additionally, the differences in Western blot banding patterns between the two studies should be addressed. Do the authors have an explanation for these discrepancies?
We thank the reviewer for their feedback.
(3) The authors report a ~40% reduction in liver parasite load using qPCR, which is statistically significant. However, this phenotype is modest and should not be interpreted as showing that LSA3 is essential.
(a) Please avoid terms like "required" or "essential" and instead describe the protein as "contributing to normal development" or "influencing fitness."
We have used the term “required for normal liver stage development”.
(b) Since the authors generated liver sections, they should take advantage of these to quantify the number and size of liver stage parasites, which would help determine whether the phenotype reflects fewer infected cells or reduced parasite growth.
We did check this via microscopy of liver sections, but all mice were co-infected with LSA3 mutant and control liver-stage parasites, as we shared the reviewers line of enquiry. Unfortunately we could not detect parasites without LSA3 signal at the 5 day post infection time point. This type of analysis does sound straightforward on paper but in reality is more challenging owing to several factors i) identifying sufficient individual parasites in an entire liver by microscopy can be challenging and variable from lobe to lobe and mouse to mouse, ii) the number of parasites required for a meaningful statistical analysis is increased due to coinfection of the liver (see Figure 4B as an illustration of this), iii) day 5 is a rather late liver-stage time point and so if there was a growth defect the defective parasites may be very small or sparse, iv) we cannot exclude that the LSA3 antibody may cross-react at the liver-stage, v) definitive conclusions are thus challenging and we feel require individual co-infections to be clear in the future. Nonetheless, the detailed qRT-PCR analyses identify a significant reduction in DLSA3 parasite liver load on day 5, indicating this protein is important for the human malaria parasite’s growth within human hepatocytes.
(c) It would also be valuable to include IFA from singly infected ΔLSA3 livers (rather than co-infected), and possibly at earlier timepoints, to identify the developmental window affected.
We agree it would be valuable.
(4) The manuscript suggests that LSA3 may be exported beyond the PVM into the hepatocyte, based on a small number of peripheral puncta.
(a) This claim is not convincingly supported by the data. The punctate signals shown in Figure 5 are weak and may rather reflect PVM extensions or TVN. In fact, one punctum even overlaps with the DAPI signal (figure 5, middle panel), which raises further doubt about the localization.
We appreciate the reviewer’s careful eye and caution and have added the comment regarding DAPI.
(b) Given the lack of KO controls in these liver stage IFAs, the authors should not describe LSA3 as "exported beyond the PVM". The language should be revised to reflect that the protein localizes predominantly to the PVM, and any extra-PVM signal remains unconfirmed and could be non-specific.
(c) This is especially important given the well-known tendency of rabbit antisera to produce background PVM staining in liver stage parasites.
Corrected.
(e) In an earlier report (McConville et al, 2024, PNAS), they clearly state that LSA3 is NOT exported beyond the PVM. Actually, the staining in the previous report looks quite different from the images provided for Figure 5. The authors might wish to comment on this.
We thank the reviewer for their feedback.
Minor comments:
In some sections, the manuscript uses "exported" to refer to trafficking to the PVM. This terminology should be used more carefully and consistently, since "export" often implies translocation into the host cytosol
We understand that export involves a protein localizing within the host cell and so protrusion through the PVM may also be considered exported, however, we have not confirmed this for LSA3 in liver stages.
