Metabolic profiling during malaria reveals the role of the aryl hydrocarbon receptor in regulating kidney injury

Essential revisions:

Reviewer #1:

Some of the conclusions drawn are not yet fully supported by the data, and important questions remain to be addressed, which would increase the overall value of this interesting study.

1) Related to Figure 3: In the third paragraph of the subsection “Malaria is characterized by stages with unique immune, metabolic, and tissue damage events”, please explain why the plasma heme level is decreased in acute infection phase. The authors show, using metabolomics, that heme levels in the plasma of Pc infected mice during the acute phase are lower than in uninfected controls (which should have virtually 0 heme in the plasma) (Figure 1H). Later, using a formic acid method, milimolar levels of heme were detected in the plasma of Pc infected mice in the acute phase (Figures 4E, 5C, 6C). This is contradictory to data in Figure 1H, and there is a wealth of data showing that heme concentration is elevated in the plasma during peak of Plasmodium infection, presumably due to intravascular hemolysis. Were the samples used for metabolomics treated/sampled in any way that could preclude heme mass detection? Were the mass spectrometers configured/setup in a way where heme mass could be detected? Please address.

We apologize for a typo in the manuscript – the line should read “…acute Pc malaria is characterized by stable plasma heme levels…” and has been fixed. Figure 1H demonstrates that heme levels don’t decrease until late infection. Apart from our error, the reviewer also notes discrepancies in heme concentrations between figures. In Figure 1H, we report plasma heme from uninfected and infected wild-type C57BL/6 mice; levels in infected mice are statistically comparable to uninfected mice during acute infection. Figures 4E, 5C, and 6C quantify plasma heme in control and AhR^-/- mice. In control mice, plasma heme remains similar to baseline levels (Figures 4E, 6C). Only in susceptible AhR^-/- mice, AhR^Tie2Δ/Δ mice, or chimeric mice with AhR^-/- radioresistant cells does plasma heme reach the millimolar levels pointed out by the reviewer. The appropriate comparison is infected mice from Figure 1H with infected control mice from the later figures (AhR^+/- mice, AhR^fl/fl mice, and chimeric mice with AhR^+/- radioresistant cells). This comparison reveals that control or wild-type mice from Figures 1H, 4E, 5C, and 6C all have heme levels similar to uninfected mice during acute infection. We do not see a contradiction in these data.

The reviewer also mentions that heme concentration is elevated in plasma during the peak of Plasmodium infection. While plasma heme concentration surely varies by model, our data indicates that increased plasma heme is not a universal feature of malaria—indeed, acute Pc infection in C57BL/6 mice does not lead to significantly increased plasma heme. We speculate that uncomplicated malaria may not lead to increased plasma heme; rather, plasma heme increases may be a cause and/or symptom of complicated malaria. Our study demonstrates that increased plasma heme during malaria can lead to pathology and we believe that the factors underlying this increase merit further study.

2) The causal relationship between plasma heme and RBC counts warrants further investigation, and if this is not possible under the current circumstances, a more detailed Discussion could be provided.

We agree with the reviewer that the relationship between RBC density and plasma heme is ripe for exploration. RBC counts and plasma heme are certainly related, as hemolysis releases heme into plasma. The relationship is not linear, however, since plasma heme levels are affected by both heme release into plasma and the multiple processes that remove heme from plasma, such as intracellular heme metabolism by HO-1. These pathways minimize plasma heme in control AhR^+/- mice during Pc infection (Figure 4E), despite substantial decreases in RBC density relative to baseline (Figure 3C). We believe the reviewer is interested in why this relationship breaks down in AhR^-/- mice, an interest that we share.

We did not observe a large defect in heme metabolism in AhR^-/- mice, through gene expression, protein analyses, and Phz-induced heme overload (Figure 4—figure supplement 2C, D, Figure 4—figure supplement 3). As the reviewer points out later, Pc-infected AhR^-/- mice fail to upregulate Hrg1 and Slc40a1 in the liver, two heme transporters which may affect plasma heme levels (Figure 4—figure supplement 2D). On the other hand, if AhR-dependent regulation of these genes affected heme overload, Phz treatment should cause different plasma heme levels or sensitivity in AhR^+/- and AhR^-/- mice, which we did not observe (Figure 4—figure supplement 3). We did find that AhR^-/- mice have slightly decreased RBC density than controls (Figure 3C); however, we suspect that this relatively minor difference is unlikely to cause the substantially increased plasma heme (Figure 4E).

We agree with the reviewer that the relationship between plasma heme and hemolysis during malaria is critical for the field to sort out. We have added clarifying sentences to the Discussion, where we propose several models that discuss hemolysis, plasma heme, and AKI (Discussion, fifth paragraph). We have also proposed an additional model, as mentioned by reviewer #2, that our data may also support (Discussion, sixth paragraph). We agree with the reviewers that further experimentation will be required differentiate between these possibilities, but believe that these studies are outside the scope of this manuscript.

3) Related to Figure 4—figure supplement 3: Although the hemolysis model used showed no apparent difference in lethality between AhR^+/+ and AhR^-/- mice, at the PHZ dose used all the AhR^+/+ mice succumbed and this might have masked a possible pathogenic effect of AhR deletion in this experimental model. The authors could use a sub-lethal dose of PHZ to allow the AhR^+/+ or AhR^+/- mice to recover from the hemolysis and anemia (usually within day 12 post injection), monitor plasma heme, iron, RBCs count and kidney injury. If the AhR^+/- mice have a heme/iron recycling defect there should be an increase in heme accumulation in plasma, presumably explaining a similar observation during Pc infection.

The reviewer is right that the lethal dose we chose may obscure the effect of AhR loss on Phz sensitivity. However, we selected this dose because it results in plasma heme levels that are comparable to what we observe in AhR^-/- mice infected with Pc, and at a similar time scale (plasma heme increases to above 1000 μm within one day). These features, we reasoned, faithfully recapitulate the hemolysis caused by malaria. While a lower dose would cause less lethality, it would also not model this aspect of malaria. Furthermore, we monitored several factors in response to Phz treatment in addition to survival, including plasma heme and kidney function, and did not see a difference in susceptibility in any metric. Due to these reasons, and experimental challenges imposed by the pandemic, we did not perform the experiment the reviewer suggested.

4) Related to Figure 6: The authors conclude that AhR expression in the endothelium is responsible for the protective effect of AhR in Pc infection, based on bone marrow chimera and conditional AhR KO experiments, used to rule out hematopoietic (radiosensitive) cells, and to identify Tie2-expressing cells as responsible for the protection effect. Nevertheless, Tie2 also drives the expression of the Cre recombinase in subsets of yolk-derived tissue-resident macrophages (see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2938310/), which are likely radioresistant and therefore cannot be ruled out entirely. The authors should address this point. Better characterisation of the cells expressing AhR, for example in the kidney, would be useful to address where the expression of this receptor may be operational to prevent the onset of AKI.

We agree with the reviewer on this point and originally alluded to the fact that tissue-resident macrophages also fit the criteria of radioresistant Tie2-expressing cells. While future experiments to identify the Tie2-expressing radioresistant cells at play in this phenotype, we believe that such work is outside of the scope of this manuscript and should be addressed in future studies. However, we agree with the reviewer that clarification around this issue is necessary in the manuscript. We have provided more detail in the subsection “AHR is necessary in Tek-expressing cells to control parasitemia, plasma heme, and AKI during Pc infection” and throughout the Discussion.

Reviewer #2:

[…] One possibility that the authors do not seem to consider is that Ahr is necessary for a tissue protective response that protects from AKI. For example, bilirubin, biliverdin and Ahr all exhibit antioxidant properties. One possibility is that the heme metabolites are necessary to signal through Ahr to promote these antioxidant responses that protect from AKI. Similarly, this signaling pathway may be mediating anti-inflammatory responses that protect from AKI. There are likely clues to possible tissue protective responses in their RNAseq data. Is there anything in their data set to suggest such a model?

In summary, I think the data are very interesting and robust, but I would like to see the authors consider the presentation of their work and whether there is a model that better fits with the data they presented. I think there is.

We think the reviewer has raised a very insightful point. In the first model we proposed, AhR functions exclusively in pathogen control, and we have added sentences to clarify this point (Discussion, fifth paragraph). This model suggests that, in the absence of AhR, parasitemia is increased as early as 4 DPI (Figure 3A), leading to more hemolysis in AhR^-/- mice (Figure 3C). Despite equivalent plasma heme concentrations, intracellular heme concentrations could nevertheless be elevated in renal proximal tubule epithelial cells, which import and metabolize heme during malaria and are sensitive to heme toxicity (Ramos, 2019). This could explain impaired kidney function at 7 DPI (Figure 4B) despite equivalent plasma heme (Figure 4E).

In the second model we considered, AhR functions in both pathogen control and heme metabolism. We believe that our data do not support this model, as we did not detect any major deficits in heme metabolism in AhR^-/- mice. Nevertheless, we remained unable to pinpoint the source of such extremely elevated plasma heme (Discussion, sixth paragraph).

The reviewer suggests a third possibility in which AhR is required for both pathogen control and tissue protection in the kidney. This model orients elevated plasma heme as a consequence, rather than the initial cause, of impaired kidney function. Particularly tantalizing is the role of the heme metabolites bilirubin and biliverdin in activating AhR, suggesting a possible feedback loop in which AhR is activated by heme metabolism to protect from heme toxicity. As an important caveat to this model, impaired kidney function is observed prior to elevated plasma heme in AhR^-/- mice (Figure 4B, E), but not in AhR^Tie2Δ/Δ mice (Figure 6C, D) or chimeric mice with AhR^/- radioresistant cells (Figure 5, longitudinal data not shown). Whether this is biologically relevant or simply an artifact of different mouse models will have important implications, since the relative timing of heme elevation and kidney impairment is critical. While we do not believe that our RNA-seq data, a time-course analysis of wild-type livers during Pc infection, provide insight into this pathway, we agree that the role of kidney-intrinsic AhR signaling during Pc infection merits further investigation. We are grateful to the reviewer for this insight and have included further discussion of each model in the Discussion. Additionally, when we previously referred to “heme-mediated AKI” in the manuscript, we now refer to the development of high plasma heme and AKI without any causal implications.

Reviewer #3:

In this manuscript by Lissner et al., authors share the results of an extensive well-executed analysis on the role of AHR receptors in progression of malaria. The project has a large scale and a broad scope encompassing genetic, metabolic, clinical, histopathologic and cellular studies. While this generates massive amounts of data that will be useful for researchers in this field, often times authors lose focus moving from one set of experiments to another and the continuity of story gets disrupted. Authors' claim about the kidney protective role of AhR is very convincing with carefully carried out knock out, heterozygous and WT experiments. On the other hand, majority of metabolomic analyses only point out the consequences of the known fact that malaria generates hemolysis and heme is released as a by-product. Furthermore, comparisons of CM and Pc is misleading and should be avoided as I mentioned in detail in relevant section of my comments. Endothelial cell based AHR expression being critical in AHR mediated protection is the most interesting section of the paper as it finally points out the attention to a single focus so it would have been nice if authors went deeper following that lead. This paper would benefit from a rewrite connecting the loose ends into a continuous story.

Introduction, second paragraph: I recommend using WHO data to refer to the up to date global disease burden of malaria rather than a manuscript published in 2012. WHO annually releases malaria reports (world malaria report) and these may provide more up to data and more widely accepted numerical values.

We thank the reviewer for this insight and have made the change suggested.

Introduction, second paragraph: I recommend removing the vague statement: "much is known about how immune responses affect malaria outcome".

We have made the suggested change.

Introduction, fourth paragraph: Referring Plasmodium berghei as rodent malaria parasite that causes lethal cerebral malaria, while not incorrect, might be misleading. The only reference that mentions P. berghei among cited after this sentence is Brant et al. and there, it is used as P. berghei ANKA which is the sub-strain that causes Experimental Cerebral Malaria (ECM) in susceptible mice. P. berghei NK65, another sub-strain that is very closely related P. berghei ANKA never results in ECM. Please use the full strain name P. berghei ANKA.

The reviewer raises a very good point and we have made this change throughout the manuscript.

Figure 1A: While no revision is required, hemoglobin measurements instead of RBC counts is a better way to monitor anemia.

We understand the reviewer’s point and agree that hemoglobin measurements offer complementary information.

Subsection “Malaria is characterized by stages with unique immune, metabolic, and tissue damage events”, fourth paragraph and Figure 1I-J: Comparison of Pc with human CM patients is not substantiated. First of all, Pc is an uncomplicated malaria causing rodent parasite. CM on the other hands is the most lethal complication for both human and mouse (ECM) malaria. I recommend removing this data set since a more reasonable comparison would have been non-complicated human Pf vs. mouse Pc or Human CM vs. mouse Pb ANKA. Unless authors wish to offer these comparisons instead, there is no value of Pc vs CM comparison other than that they both cause hemolysis simply because of being intra RBC microorganisms that are destroying RBCs in order to spread. Pc is never used to recapitulate human CM. This parasite only recapitulates the immunity development angle of human malaria with antibodies and such. This entire paragraph and relevant figures need to be reworded. Rest of Figure 1 is informative and well carried out experiments.

We thank the reviewer for this point. We agree that uncomplicated malaria caused by Pc and CM caused by P. falciparum are considerably different types of malaria and cannot be substituted for each other. In this sense, it is surprising that the plasma metabolomes are so similar (Figure 1I)! On the other hand, Pc and Pf are closely-related species that cause infections with many similar features, such as massive RBC destruction and systemic immune activation, as the reviewer points out.

Our mouse model of Pc-infected wild-type mice is most similar to uncomplicated Pf malaria, and the reviewer correctly suggests that an uncomplicated malaria dataset would be best for our comparative analysis. However, analysis is challenged by the fact that large untargeted metabolomics datasets are relatively rare, compared to targeted analysis; and when untargeted metabolomics analysis has been employed in the past, technical limitations at the time resulted in the identification of relatively few metabolites. For example, the most detailed dataset meeting the reviewer’s specifications that we have examined focused on uncomplicated Pf- or P. Vivax infected patients (found under MTBLS664). That study identified less than a third of the molecules found in our study (190 vs. 587 metabolites), and none of the heme-related metabolites that we discuss.

Metabolomics datasets that capture the ideal type of malaria in adequate detail remain elusive. For this reason, we performed our comparative analysis using metabolomics data from CM patients, which identified 432 metabolites. The reviewer is correct that a more ideal comparison would be preferable. However, given the limited data availability and the general similarities of malaria caused by all Plasmodium species, we disagree that the comparison is inappropriate. We have added a sentence to clarify the limitations of our analysis in the subsection “Malaria is characterized by stages with unique immune, metabolic, and tissue damage event”.

Subsection “AhR ligands are more abundant during acute infection”, first paragraph: Please change to Pb ANKA (refer to my comment above).

We have made this change.

Subsection “AhR signaling is protective during Pc infection”, first paragraph: This is an interesting finding. Gender bias has always been mentioned in mouse malaria experiments but has not been properly documented. This is a useful information.

We agree with the reviewer that gender bias in malaria is a rich area for future research.

Figure 4: Authors should discuss why they only focused kidney but not the other organs. Spleen, lung and liver may show damages related to sequestration, intravascular hemolysis, side products, cytokines and anemia. This needs to be discussed.

We agree with the reviewer’s point that other organs may also be relevant to this phenotype. For that reason, we originally analyzed all tissues mentioned by the reviewer—spleen, lung, and liver. We do mention that our histological analysis of liver tissue revealed less severe damage in AhR^-/- mice (subsection “AHR signaling is protective during Pc infection”). We have now added a sentence to indicate that spleen and lung tissue showed no significant differences between genotypes.

Figures 5 and 6: Figure 5 does not add much to the story since it only says that radiosensitive cells do not contribute to the AhR mediated protection. Yet it still sets the stage for Figure 6. I would shorten that part and put more emphasis on Figure 6. I found Figure 6 to be very informative focusing the attention onto endothelial cells. The endothelial cell specific knock out really has potential to generate high quality original data, however authors cut that story short by only limiting the analyses to 4 major parameters. Extending that section with more sophisticated analyses would make the manuscript more complete.

We agree that the role of AhR in Tie2-expressing cells merits further study. Thorough characterization of this phenotype, however, requires extensive experimentation and will likely warrant an entire story in the future. We feel that additional experimentation on this point is outside the scope of this paper.

Materials and methods section: age of a mouse has a significant impact on disease progression. Therefore, age matching is critical. Authors should include a statement regarding to age matching practice between control and experimental groups, otherwise 8-12 weeks is a relatively broad range for this particular experimental model.

We appreciate the reviewer’s point. For this reason, we have always carefully matched ages. By using littermates that fall within the 8-12 week old range, we ensure that our experiments have age-matched control and experimental mice. That said, we have not seen a difference in disease severity between mice that are 8 weeks versus 12 weeks old. To clarify this point in the manuscript, we have moved all discussion of mouse age to the subsection “Cross-sectional infection”.