PREreview of REL2 overexpression in theAnopheles gambiaemidgut causes major transcriptional changes but fails to induce an immune response

Summmary

In this paper from Hoermann et al., the authors engineer transgenic Anopheles gambiae mosquitoes with bloodmeal-induced overexpression of REL2. REL2 is a member of the mosquito immunodeficiency (Imd) pathway, and previous work has demonstrated that A. gambiae REL2 overexpression in another mosquito species, A. stephensi, significantly reduced Plasmodium falciparum infection. To assess the antiplasmodial activity of REL2 overexpression in A. gambiae itself, the authors used CRISPR/Cas9 to generate a markerless REL2-CP line wherein a second copy of REL2 is integrated into the carboxypeptidase (CP) locus which is upregulated upon blood feeding. Via immunofluorescence assays, the authors show that REL2 is strongly upregulated following a blood meal and shows nuclear localization as expected for its activity as a transcription factor. However, across three biological replicates there was no significant impact on the number of P. falciparum oocysts per mosquito midgut relative to WT. Unexpectedly, there was also a significant decrease in the number of REL2-CP eggs and mosquito survival over time, indicating that there were strong fitness costs associated with this transgenic line.

To assess whether these fitness costs were due to possible disruption of endogenous CP activity in the REL2-CP line, the authors generated a CP knockout mosquito line. Surprisingly, this locus was amenable to knockout and other than a slight reduction in the hatch rate of CP^KO eggs, there were no apparent fitness costs, suggesting that this was not responsible for the effects seen in the REL2-CP line. The authors also found minor effects on P. falciparum infection in the CP^KO line, including a slight reduction in the number of oocysts per midgut and a slight increase in oocyst size. To understand why REL2 overexpression had a limited impact on P. falciparum infection, the authors performed RNA-sequencing of WT and REL2-CP midguts before blood feeding and at 6 and 20 hours post-blood feeding. The authors found that some antimicrobial peptides with likely REL2 transcription factor binding sites were downregulated in the REL2-CP line, suggesting that REL2 overexpression may actually suppress its targets and provide a potential explanation for the lack of antiplasmodial activity in this line. However, mechanistic work remains to more conclusively support this hypothesis. The data in this paper largely supports the authors’ claims, but the manuscript could be improved by more consistent data analysis methods, additional validation of the RNAseq, and further interpretation of this work in the context of future genetically engineered vector control strategies.

Major points:

• One of the major findings of this paper is that REL2 overexpression does not impact P. falciparum infection. However, the authors are inconsistent in their evaluation and statistical interpretation of infection data between the REL2-CP and CP^KO lines. When assessing oocysts/midgut in the REL2-CP line, the authors evaluate each replicate separately whereas in the CP^KO line the authors pool all three replicates. In the latter, they ultimately detect a significant decrease in oocyst numbers relative to WT, but this may also be the case for REL2-CP if the replicates were pooled and would change the interpretation of the data. The authors also do not look at oocyst size in the REL2-CP line, although they do quantify oocyst diameter for the CP^KO line. The authors should still have midgut images from their infections as raw data, so I would suggest measuring the oocyst diameters from the three REL2-CP infection experiments to assess whether there is any impact on parasite size as seen in the CP^KO line.

• The authors’ primary explanation for the lack of antiplasmodial activity of the REL2-CP line is that REL2 overexpression may in fact suppress the expression of key antimicrobial peptides, including cecropins 1 and 3 and defensin 1. However, no experiments were performed to validate this hypothesis. The authors could perform knockdowns of key AMPs to define their role in P. falciparum infection. If further experiments to mechanistically support this hypothesis are outside of the scope of this paper, the authors could also perform additional analyses of their existing RNAseq data. For example, the authors could examine whether there is a conserved correlation between the number of REL2 transcription factor binding sites and gene expression. In addition to the select genes the authors have analyzed in the text, these more global analyses would provide further quantitative evidence to support their suppression hypothesis.

• The title of this paper does not accurately reflect its immune response findings. The authors’ RNAseq data shows differential gene expression of a number of immune related genes in the Rel2-CP line, the majority of which are upregulated (Fig. 5D). Although the desired P. falciparum reduction was not observed, it is inaccurate to claim that no immune response is induced. The authors should update the title to better match their data.

Minor points:

• The temporal flow of the paper is at times confusing. For example, the authors note that CP gene expression is reduced in the REL2-CP line in the text of the results section for Figure 4, but there is no data presented to demonstrate this until figure 5. The authors should more clearly reference figures when the text refers to their data, and could consider reordering their figures to ensure data is presented in an order that matches their text.

• In figure 2, images should include a scale bar and the authors should note the timepoint of images taken in figure 2B.

• Across all figures, authors should label the median/mean values for ease of interpretation by the reader.

• Oocyst counts are often not normally distributed, in which case the median should be plotted rather than the mean. Authors should consider conducting a test for normality and plotting means or medians as appropriate for these data, or just plotting medians for consistency.

• In figure 3C, the survival curve for the REL2-CP males briefly goes up around 28 days, which should not be possible. The authors should check their raw data and correct the figure as needed.

• The authors observe a significant mortality phenotype in their REL2-CP line. It is possible that the immune response varies between those mosquitoes that die quickly versus those that have a more standard lifespan, and thus only assessing parasite burden at 8 or 9 days may be obscuring an effect on parasite burden in the mosquitoes that die early. To evaluate this, the authors could quantify parasite burden at an earlier timepoint, such as ookinetes or early oocysts, when most of their mosquitoes are still alive.

• The incorrect p-value significance key is provided in the legend for figure 3A. It should be ****, p <0.0001

• Infection prevalence does not need to be its own separate graph in figure 4H. It can just be shown as a part of 4G, as was done by the authors in figure 3D. The p-value significance key (e.g. ****, p <0.0001) should also be included for relevant panels in figure 4.

• There is a typo in the “Knockout of the CP host gene reveals no significant fitness cost” text. In the second sentence, “any” should be deleted in the phrase “…which exhibited no any obvious fitness issues…”

• In the “Bloodmeal induced overexpression of REL2 causes broad transcriptional changes” section, the authors should provide a citation for the assertion “The transcriptional activity of the CP locus is known to peak at about 3h PBF…”

• Most of the RNAseq data is described in the text and supplemental tables, but this analysis is critical to the claims of the paper. The authors could consider moving some of this analysis to a main text figure, such as the GO term enrichment or key DEG analyses (either as additional panels in Figure 5 or a new figure altogether). Figure 5D is not referenced or discussed in the text, so this analysis could be moved to a supplemental figure.

• The authors should provide additional clarification on how the list of 288 genes involved in immunity was curated.

• CEC1 should be defined when it is first introduced. Currently, it is described in the text in a later paragraph.

• In the results text, authors state that DEF2-5 are either downregulated or not among DEGs (“Similarly, DEF2-5 that showed no recognizable REL2 binding sites were either downregulated or not among DEGs…”), while in the discussion the authors state that DEF2-5 were all not differentially expressed (“CEC2 and DEF2-5, all lacking REL2 binding sites in their regulatory regions, as well as GAM1, with only a single REL2 binding site, did not show any response.”). The authors should confirm which statement is accurate and update the text accordingly.

• The authors could consider adding a brief discussion of what their findings mean for future gene drive control strategies.

Competing interests

The authors declare that they have no competing interests.