Mitochondrial genome variation affects fly survival when exposed to pathogenic bacteria

To study the effect of mtDNA variation on humoral innate immune responses, we utilized a panel of D. melanogaster cybrid lines [18] (referred to as “mitotypes”) and systemically infected them with DAP-type peptidoglycan containing Gram^- bacteria Providencia rettgeri and LYS-type peptidoglycan containing Gram⁺ bacteria Staphylococcus aureus. In general, females tended to be more susceptible to both bacterial species than males (Fig 1A–1B’). The cybrid lines form two haplogroups based on mtDNA sequence similarity [18] and especially haplogroup I females were more susceptible to P. rettgeri infection than males of the same haplogroup (S1A Fig). Furthermore, mtDNA mitotypes drastically affected the survival of the hosts after P. rettgeri infection, ranging from 15% (mtBS1) to 71.7% (mtKSA2) in males and from 1.7% (mtBOG1) to 33.3% (mtKSA2) in females (Fig 1A-1A’). The survival rates of S. aureus-infected mitotypes ranged from 16.7% (mtM2) to 59.3% (mtKSA2) in males and from 5% (mtBS1 and mtM2) to 63.3% (mtKSA2) in females (Fig 1B-1B’). mtDNA haplogroup -specific sexual dimorphism was also apparent in S. aureus-infected flies (S1B Fig). In general, mtKSA2 females and males had the highest survival rates in both infections when compared to the other mitotypes (S2 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. mtDNA variation affects infection outcomes after exposure to bacterial pathogens.

(A-A’) Males and females of nine cybrid lines and nuclear background line Oregon RT (ORT) were infected with Gram-negative P. rettgeri (n = 60) or with (B-B’) Gram-positive S. aureus bacteria (n = 60) (Note that mtPVM is missing from the S. aureus infection). Stars denote the overall statistical difference in the survival among all the mitotypes analysed using the Mantel-Cox log rank test. The statistical differences in the survival rate of the survival-enhancing mitotype mtKSA2 (red) in comparison to other mitotypes are shown in S2 Fig Mitotypes mtWT5A (high survival rate; purple), mtBS1 and mtORT (low survival rate; purple) were studied further alongside mtKSA2. (C-D’) Bacterial loads were measured as colony forming units (CFUs) 8 and 20 hours post infection (h p.i.) and in addition 48 h after infection with P. rettgeri and 72 h after S. aureus (n = 8). The data on CFUs was analysed with linear models with Type III Sums of Squares. (E-F’) Resistance vs. tolerance quadrants. We plotted mean (E-E’) P. rettgeri and (F-F’) S. aureus CFUs measured at 48 and 72 h p.i. against the proportion of surviving flies to summarize potential differences between resistance and tolerance among four mitotypes, focusing on mtKSA2. Quadrants (dashed lines) represent the cut-off points between different defence phenotypes determined by overall mean survival and colony forming units: i) resistant (healthy), ii) tolerant (healthy), iii) resistant (pathology) and iv) failing. ns, not significant; *, p<0.05; ** p<0.01; ***, p<0.001.

https://doi.org/10.1371/journal.pgen.1011476.g001

mtKSA2 induced increase in survival is due to tolerance rather than increased bacterial clearance

Host disease defense occurs through a combination of resistance and tolerance. Resistant flies can clear an infection, whereas tolerant flies are able to limit the infection-induced damage without clearing the pathogen [43,44]. To further study the effect of mtDNA mitotypes on infection outcomes, we measured bacterial loads (colony forming units, CFUs) in four mitotypes selected based on their survival rates after exposure to pathogenic bacteria (Fig 1A-1B’) to represent high (mtKSA2 and mtWT5A) and low (mtORT and mtBS1) survival rate lines (S2 Fig). We measured CFUs at an early-stage infection time point (8 hours post infection, h p.i.) and when the flies began to die (20 h p.i.) after exposure to P. rettgeri or S. aureus. We additionally measured CFUs at late-stage infection time points, i.e. when survival rates started to plateau, at 48 hours after P. rettgeri and 72 hours after S. aureus infection. P. rettgeri load differed among mitotypes at 8 h p.i. (Mitotype: p = 0.049) and there was a differential effect of mitotype and sex at 20 h p.i. (Mitotype x Sex: p = 0.03). S. aureus load differed between males and females (Sex: p = 0.0003) at 8 h p.i. and there was a differential effect of mitotype and sex at 20 h p.i. (Mitotype x Sex: p = 0.0006) and 72 h p.i. (p = 0.004) (Fig 1C-1D’). However, this did not directly reflect the differences seen in the survival rates. Therefore, we plotted the average CFUs quantified during late-stage infections against survival rates to better visualize average disease susceptibility among mitotypes [45,46] (Fig 1E-1F’). We found that mtKSA2 males infected with P. rettgeri appeared to be more tolerant because they had high average pathogen burden and higher survival compared to the other mitotypes (Fig 1E). Females in each of the four mitotypes had similar P. rettgeri burdens but varying survival rates, which is suggestive of variation in disease tolerance. mtORT and mtBS1 females appeared to be more susceptible to P. rettgeri infection compared with mtKSA2 and mtWT5A, illustrated by their low survival rates (Fig 1E’). In contrast, S. aureus-infected males tended to carry similar bacterial burdens but survival rates among mitotypes differed. mtKSA2 and mtWT5A males had higher survival rates than mtBS1 and mtORT males, suggesting that mtKSA2 and mtWT5A were more tolerant (Fig 1F). In S. aureus infected females, mtKSA2 had the highest survival rate with bacterial load similar to mtWT5A and mtBS1, while mtORT had both high CFUs and low survival (Fig 1F’). Taken together, the higher survival rates of mtKSA2 were generally due to increased tolerance, and not to a greater ability to clear infection.

Immune priming is mitotype-specific

To test if the increased survival of mtKSA2 upon bacterial infection could be caused by preactivation of innate immune responses we primed female and male flies with heat inactivated P. rettgeri and challenged them with live P. rettgeri 18 hours later. Our expectation was that if mtKSA2 possesses preactivated (humoral) immune response prior to infection, the line should have a weaker response to priming. We found that the effect of priming varied among the mitotypes (Mitotype x Priming: p = 0.002) (S2 Fig). Survival of mtWT5A and mtBS1 flies was higher after being primed with heat inactivated bacteria and challenged with live bacteria (S3 Fig). Conversely, mtKSA2 survival and mtORT male survival following P. rettgeri infection was not significantly affected by priming. Therefore, the protective effect of immune priming seems to depend on mitotype, and mtKSA2 flies did not respond to priming based on their survival rates.

mtKSA2 antiviral protection is less pronounced than antibacterial protection

The effect of mtDNA variation was examined upon infection with two Drosophila-specific viruses, Kallithea Virus (KV; [47]) and Drosophila C Virus (DCV; [48]). Kallithea virus (family Nudiviridae) is a large double stranded DNA virus and DCV (family Dicistroviridae) is a single stranded positive sense RNA virus. Because KV does not affect female survival [47], we only infected males and followed their survival rates for 22 days. mtORT, mtWT5A and mtBS1 showed almost identical end-point survival rates (7–14%) upon KV infection, while mtKSA2 again showed the best survival rate among the studied mitotypes, as 38% of mtKSA2 males survived the infection (Figs 2A and S4A). We followed male and female survival for 14 days after DCV infection. During the acute phase of the DCV infection, mtORT males (Fig 2B) and females (Fig 2B’) died at a faster rate than the other mitotypes, but the end-point survival percentage did not differ. However, overall mtORT was shown to be more susceptible to DCV infection than mtKSA2 (S4B-S4B’ Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. mtKSA2 effect upon viral infection is not as strong as upon bacterial infection.

(A) Survival rates of males from the four mitotype lines after infection with double stranded DNA virus Kallithea. Controls for Kallithea virus infection were injected with 1 x PBS (dashed gray lines). (B-B’) Survival of both males and females after infection with single stranded positive sense RNA virus Drosophila C virus (DCV). Controls for DCV infection were pricked with Tris-Hcl solution (dashed gray lines). Overall differences in the survival among the mitotypes were analysed using the Mantel-Cox log-rank test. Differences in comparison to survival-enhancing mitotype mtKSA2 (red) are shown in S4 Fig. (A-B’) ns not significant; * p<0.05; ** p<0.01; *** p<0.001.

https://doi.org/10.1371/journal.pgen.1011476.g002

mtDNA variation affects the transcriptome profiles of uninfected flies

We used RNA sequencing to examine the effect of mtDNA variation on the nuclear and mitochondrial transcriptomes. Bulk RNA sequencing was performed on uninfected mtKSA2 and mtORT flies as well as on the same fly lines infected with P. rettgeri, S. aureus (8 and 20 h p.i.) or DCV (3 d p.i.). Based on a principal component analysis (PCA) of the normalized read counts, female and male samples formed their own markedly distinct clusters (Fig 3A, sex explaining 82% of the variation in the data). In males, mtKSA2 and mtORT samples were separated from each other by variable degrees depending on infection type and timepoint (Fig 3A’–3A”’). In uninfected and in P. rettgeri-infected females, mitotypes showed separate clustering (S5A Fig) whereas in S. aureus and DCV infected females clustering according to mitotype and infection timepoint was less obvious (S5A’–S5A” Fig). Overall, the PCA indicates that the two mitotypes cause variation in the transcriptomes of the uninfected flies, as well as upon different infections. To emphasize the effects of mtDNA mitotypes and infection on the transcriptomes, we focused on the male transcriptome data in the present work and not on sex specific differences. The female data is available under the same GEO access number as the male data (GSE260986). The data of male log2FCs and false discovery rate (FDR) adjusted p-values are also listed in S1 Table.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Overview of RNA sequencing sample clustering and differentially expressed genes.

(A’-A”’) Principal component analysis (PCA) plots of (A) all RNA sequencing samples, females (left) and male (right) samples showing a clear sex-specific clustering pattern, (A’) P. rettgeri infected and control, (A”) S. aureus and control and (A”’) DCV infected and control mtORT and mtKSA2 cybrid male samples. Each treatment includes four biological replicates containing pools of 10 individuals. (B) Volcano plot showing differentially expressed genes in uninfected mtKSA2 vs. mtORT males. Log₂FC cut-off was set to 0.5 and FDR-adjusted p-value to 0.15. Genes in the main GO terms are shown with different colors and genes tested with RT-qPCR are labelled in the volcano plot. (C) The expression levels of common Toll and Imd pathway target AMPs in uninfected mtKSA2 and mtORT. Expression of these AMPs upon P. rettgeri, S. aureus and DCV infections is listed in S3 Table. Stars denote statistically significant fold changes (FDR< 0.15). (D-D”’) Expression levels of FucTC, AANATL3, Acp1 and CG3397 were measured with RT-qPCR in uninfected males of the four cybrids lines (n = 3, pools of 5). Data was analysed using one-way analysis of variance followed by Tukey’s HSD pairwise comparisons. Differences to mtKSA2 are indicated. ns not significant; * p<0.05; ** p<0.01; *** p<0.001.

https://doi.org/10.1371/journal.pgen.1011476.g003

We utilized Gene Ontology (GO) enrichment analyses (https://www.flymine.org/flymine; [49]) to compare the differentially expressed (DE) genes between uninfected mtKSA2 and mtORT males. Based on the enriched biological processes, a large set of the upregulated genes in mtKSA2 belonged to terms related to energy generation via mitochondrial respiration (S2 Table and Fig 3B). Other representative GO terms were associated with “cuticle development”, “muscle system process” and “glycolytic process” (S2 Table and Fig 3B). Among the genes downregulated in mtKSA2, “response to bacterium” (23 genes) was the only significantly enriched GO term (Fig 3B) which was surprising given mtKSA2 flies’ enhanced survival rates upon bacterial infections (Fig 1). Seven of these genes encode proteins with antimicrobial function (Diptericin A, Diptericin B, Defencin, Drosocin, Attacin-A, Bomanin Bicipital 1 and Baramycin), three encode for negative regulators of the Imd pathway (pirk, PGRP- SC2 and Charon) and two genes are related to stress responses (Turandot E and p38c MAP kinase). However, several of these genes are not directly connected to the core immune response (such as four genes encoding for lysozymes and a gene encoding for a myosin ATPase; S2 Table). Hence, we next examined the expression of a larger set of AMPs as well as key components of Toll and Imd pathways [26–28] in uninfected mtKSA2 and mtORT flies (Fig 3C and S3 Table). Besides the AMPs included in the significantly downregulated GO term “response to bacterium”, most of the other AMPs were either downregulated or not significantly affected in uninfected mtKSA2 when compared to mtORT (Fig 3C and S3 Table). Only Drosomycin, Metchnikowin-like and Attacin-C were significantly upregulated. Furthermore, Toll and Imd pathway components were mostly unaffected by mtDNA mitotype in the absence of infection, except for the negative regulators of the Imd pathway mentioned above. Of note, both bacterial infections and DCV infection caused a robust and consistent upregulation of the AMPs (S3 Table). This response was similar in mtKSA2 and mtORT. Likewise, P. rettgeri and S. aureus infections led to the upregulation of the upstream activators of Toll and Toll itself in mtKSA2 and mtORT. The Imd-activating PGRP-LC was upregulated in both mitotypes after infection by P. rettgeri, but not after S. aureus (except for a mild upregulation in mtORT 8 h p.i.), and Imd was upregulated by both P. rettgeri and S. aureus at 8 h p.i. Taken together, although uninfected mtKSA2 showed suppression of many genes related to NF-κB pathways, they were robustly and similarly induced in mtKSA2 and mtORT flies upon infection.

When comparing the transcriptomes of two mitotypes, one cannot conclude if the genes of interest are specifically up- or downregulated in the mitotype of interest. Hence, we measured the expression patterns of the representative genes with RT-qPCR from uninfected mtORT, mtKSA2, mtBS1 and mtWT5A flies, with a new set of male samples. We picked the gene with the highest fold change (FC) in uninfected mtKSA2 when compared to mtORT (CG3397); a highly expressed gene representing the GO term cuticle development (Acp1; Adult cuticle protein 1); and two genes that were either downregulated (FucTC; alpha1,3-fucosyltransferase C) or upregulated (AANATL3; Arylalkylamine N-acetyltransferase-like 3) in uninfected mtKSA2 in comparison to mtORT flies (Fig 3B). None of the genes showed consistent, mtKSA2 specific expression pattern, but instead FucTC (Fig 3D) and AANATL3 (Fig 3D’) were similarly expressed in mtORT and mtWT5A (highest mtDNA sequence similarity [18]), when compared to mtKSA2 and mtBS1, whereas Acp1 (Fig 3D”) and CG3397 (Fig 3D”’) had similar expression patterns in mtORT and mtBS1 (the two susceptible lines) when compared to mtWT5A and mtKSA2.

Mitotypes shape the transcriptome profiles upon infection

We performed a GO term analysis to obtain an overview of the transcriptomic profiles in mtKSA2 and mtORT when exposed to pathogens. The GO analysis showed that mtORT and mtKSA2 flies had overlapping responses to P. rettgeri and S. aureus, but there also was a large set of differentially expressed genes that were mitotype-specific (Fig 4A-4D’). Furthermore, more genes were differentially expressed at the early p.i. timepoint than later, and the enriched GO terms varied between timepoints, likely reflecting differences in the initiation phase of the immune response at 8 h p.i. and later at the 20 h p.i. Overall, the GO term analysis yielded numerous significant GO terms, full lists of which are presented in S4 Table and a manually curated overview in Fig 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Gene Ontology (GO) term analysis.

(A-B’) Number of genes and representative enriched GO terms (Biological process) up or downregulated after P. rettgeri infection in both mitotypes (intersection), in mtORT only (left) and in mtKSA2 only (right). (C-D’) Number of genes and representative GO terms up or downregulated after S. aureus infection and (E-E’) number of genes and representative GO terms up or downregulated after DCV infection. GO terms listed below the Venn diagrams denote the shared processes, terms listed above on the left denote processes specific to mtORT and on the right those specific to mtKSA2. See S4 Table for a full list of the significant GO terms in each treatment.

https://doi.org/10.1371/journal.pgen.1011476.g004

Genes upregulated in both mitotypes at 8 h and 20 h after P. rettgeri and S. aureus infection were involved in such processes as protein synthesis and immune and stress responses (Fig 4A-4B and 4C-4D). Shared downregulated genes were more variable and were involved in GO terms “cilium organization”, “cytoskeleton organization”, “protein polyglycylation” and “mitochondrial transport”, depending on infection type and timepoint (Fig 4A’–4B’ and 4C’ and 4D’). In addition, multiple developmental processes were downregulated specifically at 8 hours post P. rettgeri infection (S4 Table).

Genes related to GO terms “cell cycle” and “cell differentiation” were upregulated in mtKSA2 after P. rettgeri infection (Fig 4A and 4C). After S. aureus infection, mtKSA2-specific terms were largely related to development, including tissue and organ development (Fig 4C and 4D). Downregulated mtKSA2-related GO terms were involved in “cilium organization” and “cilium assembly” and “cuticle development” (Fig 4A’–4B’ and 4C’–4D’). Interestingly, at 20 h post P. rettgeri infection, terms related to “mitochondrial transport” and “protein localization to mitochondria” were downregulated in mtKSA2 (Fig 4B’ and S4 Table). In contrast, in mtORT mitochondria-related GO terms such as “cellular respiration” and “mitochondrial translation” were upregulated at 20 h after both infection types (Fig 4B and 4D). Many GO terms related to development were downregulated in mtORT (Fig 4A’–4B’ and 4C’–4D’).

mtKSA2 and mtORT were equally susceptible to DCV infection at the end-point survival (Fig 2B) but had surprisingly low overlap in differentially regulated genes after DCV infection when compared to their uninfected state, especially when looking at the upregulated genes (Fig 4E-4E’). Among the shared, significantly enriched upregulated processes were immune response related terms such as “response to other organisms” (Fig 4E). Genes related to mitochondrial processes such as “ATP production” and “aerobic respiration” were downregulated after DCV infection in both mitotypes (Fig 4E’). mtKSA2-specific upregulated genes did not form any significant GO terms, whereas mtORT showed upregulation of “ribosome biogenesis”, “cell migration” and multiple terms related to development (Fig 4E; S4 Table). mtKSA2 flies further downregulated genes related to mitochondrial function as well as cuticle development, and mtORT showed downregulation of several metabolic processes including carbohydrate, fatty acid and oxoacid metabolism (Fig 4E’ and S4 Table)

Overall, bacterial and viral infections induced varied responses at the gene expression level in the mtDNA mitotypes with the core immune-responsive genes equally affected in both mitotypes. Furthermore, mitochondrial processes seem to be upregulated in both mitotypes early after bacterial infection, but only in mtORT flies at the later timepoint, whereas DCV infection caused downregulation of various mitochondria related processes in both mitotypes.

OXPHOS gene expression is affected by mitotype and infection

Genes related to respiration were upregulated in uninfected mtKSA2 flies (Fig 3B), and genes belonging to multiple mitochondria-related GO terms were affected after bacterial and viral infections (Fig 4). We therefore delved deeper into the expression patterns of OXPHOS genes in uninfected and infected flies. A number of genes classified in the Drosophila database FlyBase (https://flybase.org; [50]) as OXPHOS genes, with an addition “-like (L)” in the gene name (S6 Fig), did not respond to infections in the same way as the majority of the OXPHOS genes (S6 Fig). These genes are categorized as OXPHOS genes based only on sequence similarity. They have not been experimentally linked to OXPHOS, nor do they have mammalian homologues [50]. Based on limited data availability, we speculate that they could be pseudogenes and possibly not encoding OXPHOS proteins.

We found that uninfected mtKSA2 had consistently elevated gene expression of both nuclear and mitochondrial encoded subunits of the OXPHOS complexes (Fig 5A). Next, the OXPHOS gene expression patterns were followed in mtKSA2 and mtORT flies after infection compared to their respective uninfected states. At 8 h p.i., both mitotypes showed upregulation of most OXPHOS genes, with stronger upregulation in response to P. rettgeri infection. In general, OXPHOS gene expression was higher at 8 h p.i. in mtKSA2. At 20 h p.i., mtKSA2 flies did not have upregulated OXPHOS gene expression to the same extent as mtORT, possibly because most of the mtKSA2 flies were already recovering from the infection and thus mitochondrial metabolism was returning to the baseline level. In contrast, upon DCV infection, most OXPHOS subunit genes were downregulated in both mitotypes, indicating a specific mitochondrial response induced by a viral infection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. OXPHOS gene expression is modulated by mtDNA and the infection type.

(A) The expression levels of OXPHOS complex genes. Stars denote statistically significant fold changes (FDR<0.15). (B-B’) Mitochondrial copy number was measured as copies of the mtDNA target gene 16S relative to nuclear target gene RpL32. Copy number was measured in uninfected and bacterial infected cybrid males (n = 3, pools of 5). (C-C”) Mitochondrial respiration was measured in uninfected males by measuring oxygen consumption of the OXPHOS complexes I, III and IV (n = 4, pools of 200). Data in C & D was analysed using one-way analysis of variance. ns not significant; * p<0.05; ** p<0.01; *** p<0.001.

https://doi.org/10.1371/journal.pgen.1011476.g005

Because OXPHOS gene expression differed between mitotypes, infections and time points, we examined whether mtDNA copy number varied among these treatments. For this, we collected mtORT and mtKSA2 males at 8 and 20 h p.i., as well as uninfected controls. However, we detected no differences in the mtDNA copy numbers after P. rettgeri (Fig 4B) or after S. aureus (Fig 4B’) infections between the two mitotypes. To investigate if the elevated expression of OXPHOS genes in mtKSA2 was reflected in the performance of mitochondria, we measured oxygen consumption (nmol O₂/ min/ml) of OXPHOS complexes cI, cIII and cIV in uninfected males of the four mitotypes, and did not detect significant differences in respiration between the mitotypes (Fig 4C-4C”). In conclusion, there are distinct changes in the OXPHOS gene expression patterns between mtKSA2 and mtORT flies and in how these two mitotypes respond to bacterial infection. However, these expression differences are not reflected in mtDNA copy number or respiration, and therefore the elevated OXPHOS gene expression might represent a compensatory response in mtKSA2.

ROS-related gene expression is affected by mitotype and infection, but hydrogen peroxide levels do not vary among the mitotypes

One of mitochondrial contributions to the host response against infections is the generation of ROS, mainly via OXPHOS complexes I and III [51]. mtKSA2 flies possess the OXPHOS complex III gene mt:cyt-b variant (S5 Table). Therefore, it is possible that this mitotype may affect ROS production. We compiled a list of the most common ROS-related genes, such as ROS scavenging, producing and detoxifying enzymes, and checked their expression patterns from the transcriptome data in uninfected and infected mtKSA2 and mtORT flies (Fig 6A). Mitochondrial and extracellular Superoxide dismutases (Sod2 and Sod3, respectively) that convert superoxide to hydrogen peroxide (H₂O₂) were upregulated in uninfected mtKSA2. In addition, Pxd, a peroxidase functioning in neutralizing H₂O₂ was highly upregulated in mtKSA2, whereas peroxiredoxins with the same function were not consistently up- or downregulated. There were less consistent patterns observed among the Glutathione S- transferases (Gsts), functioning in a variety of detoxifying reactions (Fig 6A). Interestingly, the ROS-producing NADPH oxidases, Nox and Dual oxidase (Duox) were upregulated in uninfected mtKSA2 flies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Expression of ROS scavenging and producing genes and hydrogen peroxide levels vary depending on mtDNA mitotype and infection type.

(A) Log2 fold changes of ROS-related gene expression in uninfected mtKSA2 flies compared to uninfected mtORT and upon bacterial and viral infections compared to uninfected flies of the same mitotype. Stars denote statistically significant fold changes (FDR< 0.15). (B-B’) Hydrogen peroxide (H₂O₂) measured from uninfected and infected male flies 8h (B) and 20h (B’) post infection. Note that the 8h and 20h timepoints were separate experiments and therefore not directly comparable to each other. Data were analyzed using linear models with a natural log transformed response where appropriate (n = 3, pools of 5 flies). Although we observed no difference among mitotypes, H₂O₂ significantly increased at 8 h p.i. with P. rettgeri (p = 0.0127). ns not significant at the level of p>0.05; * p<0.05. Error bars represent +/- one standard error of the mean.

https://doi.org/10.1371/journal.pgen.1011476.g006

In general, at 8 hours post P. rettgeri infection, we observed that many ROS-related genes were upregulated in both mitotypes. While the expression of these genes started to resemble the baseline level expression in mtKSA2 at 20 h p.i., they remained upregulated in mtORT (Fig 6A). After S. aureus infection, ROS-related genes were upregulated to a lesser extent than after infecting with P. rettgeri in both mtORT and mtKSA2, and they were mostly unaffected or downregulated upon DCV infection (Fig 6A).

Heme peroxidases exploit the reduction of H₂O₂ to catalyze various oxidative reactions. Out of the ten Drosophila heme peroxidases, Duox, Immune-regulated catalase (Irc), Peroxidase (Pxd), Peroxidasin (Pxn) and Peroxinectin-like (Pxt) were upregulated in uninfected and/or infected mtKSA2 (Fig 6A). As the upregulation of ROS producing and scavenging enzymes such as heme peroxidases might indicate an increase in the levels of ROS, we studied the effect of mitotypes and bacterial infection on the H₂O₂ levels in males from the four mitotypes. In general, H₂O₂ levels increased at 8 hours post P. rettgeri infection (Fig 6B), but although we observed increased expression of some of the ROS-related genes in mtKSA2 when compared to mtORT, this was not reflected in H₂O₂ levels in uninfected nor infected flies (Fig 6B-6B’). Taken together, both mitotypes induced expression of ROS genes as well as increased H₂O₂ after P. rettgeri infection, whereas there were less pronounced changes in gene expression and H₂O₂ level upon S. aureus infection. This suggests that ROS-related gene expression, in this context, can be used as a proxy of the degree of ROS present in the organisms. Based on these results, it is unlikely that changes in ROS production at the whole organism level are responsible for the enhanced immunocompetence in mtKSA2 flies.

mtKSA2 shows signs of an enhanced cell-mediated immune response based on transcriptome profile

We did not detect changes in the AMP expression levels that would explain the enhanced survival of mtKSA2 flies upon bacterial infections (Fig 3C and S3 Table). However, a previous study demonstrated that mtKSA2 caused the formation of melanotic nodules when introgressed into different nuclear backgrounds, in the absence of infection [21]. Melanotic nodules are hemocyte aggregates and considered one of the hallmarks of activated cell-mediated innate immunity in Drosophila [40]. Hence, it is plausible that mtKSA2 may have more plasmatocytes or hemocyte immune activation prior to infection, explaining the appearance of melanotic nodules in uninfected larvae. Phagocytosis is an ancient cell-mediated immune defense [52] and therefore, we looked for changes in the expression of scavenger receptors expressed in hemocytes (plasmatocytes), the professional phagocytes in Drosophila, in our whole fly RNA sequencing samples. We found that eater, Nimrod C1 (NimC1) and Scavenger receptor class C, type I (Sr-CI) were elevated in uninfected mtKSA2, and also following bacterial infection at some of the timepoints (Fig 7 and S6 Table). In addition to these well-characterized receptors which aid in bacterial clearance, simu (NimC4; functioning in clearance of apoptotic bodies) and some of the less well characterized Nimrod-type receptors were also elevated in both uninfected and infected mtKSA2 (Fig 7 and S6 Table). In addition, we noted increased Hemolectin (Hml) expression, a gene expressed specifically in hemocytes, in uninfected mtKSA2 flies, as well as at 8 h after P. rettgeri infection in both mitotypes. These results indicate that mtKSA2 flies either have more immune cells or they exhibit increased phagocytic receptor expression to facilitate phagocytosis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Scavenger receptors expressed in hemocytes.

Expression patterns of scavenger receptors that are expressed in plasmatocytes in the uninfected mtKSA2 compared to uninfected mtORT flies, and upon different infections (mitotype compared to its uninfected basal level). Stars denote statistically significant fold changes (FDR< 0.15).

https://doi.org/10.1371/journal.pgen.1011476.g007

mtKSA2 enhances the immune response after parasitoid wasp Leptopilina boulardi infestation

mtDNA variation, and especially mtKSA2 affected the survival rates of the flies after exposure to bacterial and viral pathogens (Figs 1 and 2). However, this was not reflected in the Toll and Imd mediated expression levels of AMPs, but instead there was evidence for increased hemocyte number in the mtKSA2 flies prior to infection (Fig 7). Furthermore, previous findings show that mtKSA2 flies exhibit melanotic nodules, a possible indication of aberrantly activated cell-mediated innate immunity [21]. Therefore, we studied the effect of mtDNA variation on the cell-mediated immune response at the larval stage. Parasitoid wasps infect Drosophila larvae and elicit a cell-mediated immune response. Melanotic nodules are reminiscent of the melanised capsule that is formed around parasitoid wasp eggs and larvae as a cellular defense mechanism to kill the intruder. We used the parasitoid wasp, Leptopilina boulardi to infect larvae of the four mitotypes and scored the efficiency of the immune response as a proportion of melanised wasp eggs and larvae found inside the fly larvae. Two rearing temperatures were used: 25°C (bacterial and viral infections were performed at this temperature) and 29°C (this temperature is commonly used for the wasp assays and when using the binary GAL4/UAS system for genetic manipulation [53]). In general, we found that fully (+) melanised wasp larvae were scarce in all four strains when reared at 25°C, ranging from 0% in mtORT and mtWT5A to 5% in mtBS1 and 9% in mtKSA2 (Fig 8A). Larvae reared at 29°C displayed an enhanced encapsulation response and mtKSA2 larvae were able to melanise (both partially and fully) more wasp larvae than the other cybrids (Fig 8A), showing that mtKSA2 enhances the cell-mediated innate immune response at the larval stage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. mtKSA2 variation modulates the cell-mediated innate immune response in larvae.

(A) The effect of mtKSA2 and three other mitotypes on cell-mediated innate immunity was studied using a parasitoid wasp infection model. The success of the immune response was categorized as (-) a completely failed response with living, unmelanised wasp larva, (-/+) a partially successful response with partially melanised larva and (+) a successful response with completely melanised and encapsulated wasp larva (n = 80). The data was analysed with logistic regression with a binomial distribution and Tukey post hoc tests were applied for pairwise comparisons. Grey letters indicate statistical differences when comparing proportion of partially or fully (LM + M) melanised wasp eggs/larvae among the strains, and black letters indicate the difference when comparing fully melanised (successful immune response) wasp eggs/larvae. The same letter indicates that there was no statistically significant difference, different letter indicates a statistical difference at the level of p < 0.05. (B-C’) Total hemocyte numbers and numbers of cells falling into the activated hemocyte gate in uninfected (B-B’) and wasp-infected (C-C’) larvae, reared either at 25°C or 29°C using a flow cytometer (n = 30). One mtKSA2 individual had a very high cell count in B’ (417 activated hemocytes) and was removed from the plot to facilitate plotting. The data on hemocyte counts was analysed using a generalized linear model with a negative binomial distribution, temperatures were analysed separately. Error bars indicate standard deviation. The same letter indicates that there was no statistically significant difference, different letters indicate a statistical difference at the level of p < 0.05. (D) The expression of a lamellocyte specific gene, PPO3, was measured in the hemocytes of mtORT and mtKSA2 male (on the left) and female (on the right) larvae reared at 29°C (n = 3) relative to His3.3B expression (2^(-dCT)). (E) Crystal cell counts were measured in larvae reared at 25°C and at 29°C. Numbers of black punctate (crystal cells) were counted from the three lowest larval segments (n = 12 for 25°C, n = 24 for 29°C). (F) Mitochondrial membrane potential was assayed as the fluorescent signal intensity of TMRM dye in hemocytes (n = 3). Data were analysed using a one-way ANOVA. (G) Reactive oxygen species (ROS) levels were measured in the hemocytes of non-infected larvae reared at 29°C using the CellROX^TM green reagent (n = 4, ~2000 hemocytes per replicate) as a percentage of hemocytes stained with CellROX^TM. The data were analysed using a logistic regression with a binomial distribution and Tukey post hoc tests were applied for pairwise comparisons.

https://doi.org/10.1371/journal.pgen.1011476.g008

mtKSA2 larvae have higher hemocyte counts pre and post infection and a stronger melanisation response

Hemocyte proliferation and lamellocyte differentiation are necessary for the successful encapsulation and melanisation of parasitoid wasp eggs and larvae. Therefore, we used flow cytometry to assess the total number of circulating hemocytes in uninfected and L. boulardi -infected larvae (S7A–S7C’ Fig). Hemocyte counts between the male and female larvae within a mitotype differed only on a few occasions (S7D–S7E’ Fig), so we pooled the hemocyte data for subsequent analyses (Fig 8B–8C’). Uninfected mtKSA2 larvae had the highest baseline hemocyte counts at both temperatures (Fig 8B). In addition, at 29°C mtKSA2 also had more hemocytes after parasitoid infection (Fig 8C).

To test if the differences in the hemocyte numbers might be due to the differences in hemocyte viability, we analysed the proportion of dead hemocytes (PI-positive) relative to total hemocyte counts. In general, there were no major differences in hemocyte viability, but mtKSA2 larvae had a slightly smaller proportion of PI-positive hemocytes than mtORT larvae in all conditions except for in uninfected larvae reared at 29°C (S7F-S7G’ Fig). Based on the flow cytometry FSC-area vs. FSC-height plot, mtKSA2 larvae also had more hemocytes that could be categorized as immune activated (lamellocytes) in both uninfected and wasp-infected larvae at 29°C compared to other cybrids (Fig 8B’ and 8C’). We measured the expression of a lamellocyte-specific gene prophenoloxidase 3 (PPO3; [38]) from the hemocytes of uninfected mtORT and mtKSA2 larvae maintained at 29°C to test for the presence of mature lamellocytes. However, there were no differences in the expression levels of PPO3 (Fig 8D).

In addition to lamellocytes, crystal cells, which are the main producers of melanin upon wounding, participate in the melanisation response of parasitoid wasp eggs and larvae. We quantified the amount of crystal cells from 3^rd instar larvae reared at 25°C and 29°C and found no differences when measured from the three lowest posterior segments from the dorsal site of the larvae (Figs 8E and S8A–S8A’). However, there were differences in the presence of crystal cells in the lymph glands: mtORT = 30%, mtWT5A = 8.3%, mtBS1 = 20.8% and mtKSA2 = 75% of the larvae (n = 12) were shown to have dark punctae in the lymph gland (S8A’ Fig).

Next, we assessed the efficacy of the melanisation response at the larval stage by dissecting larvae and bleeding the hemolymph onto a glass slide, after which we measured the strength of spontaneous melanisation (S8B Fig). mtKSA2 larval hemolymph had a significantly stronger melanisation response compared to mtORT or mtWT5A, when analysed at the endpoint of the experiment (S8B’ Fig). Therefore, we conclude that the measured hemocyte profiles are indicative of enhanced cell-mediated innate immunity in mtKSA2 flies at the larval and adult stages regardless of infection status.

mtKSA2 does not affect the mitochondrial membrane potential but increases ROS in hemocytes

In a previous study, we showed that hemocyte targeted OXPHOS gene knockdown causes mitochondrial depolarization, elevated hemocyte count, and immune activation [53]. To assess the effect of the four mtDNA mitotypes on mitochondrial function in hemocytes, we measured the mitochondrial membrane potential of hemocytes in uninfected cybrid larvae (Fig 8F). There were no significant differences in the mitochondrial membrane potential among mitotypes, but on average mtORT hemocytes tended to have a lower membrane potential. In general, higher mitochondrial membrane potential is related to elevated levels of ROS [54]. We used CellROX green probe to measure ROS from larval hemocytes reared at 29°C. First, we used hemocytes from larvae fed with the antioxidant N-acetylcysteine (NAC; 1 mg/ml of food) as a negative control, and hemocytes from wasp-infected larvae as a positive control. As expected, the highest proportion of CellROX-positive hemocytes, and also the highest CellROX-intensity, were found in the untreated wasp-infected larvae, followed by descending CellROX prevalence in untreated and uninfected, NAC-fed infected and finally NAC-fed uninfected larvae (S9A–S9C Fig). We found that the mtKSA2 larvae had the highest proportion of CellROX-positive hemocytes, while the remaining three cybrid lines showed similar ROS levels (Fig 8G). Taken together, our data on larval hemocytes suggest that mtKSA2 larvae have enhanced encapsulation efficiency as a result of increased hemocyte numbers. Further, the increased levels of ROS in mtKSA2 hemocytes may aid in killing the parasite. In addition to the enhanced melanisation response on the parasitoid eggs and larvae, the melanisation response of the larval hemolymph was also the strongest in mtKSA2.

Drosophila melanogaster cybrid strains and rearing conditions

The D. melanogaster cybrid strains were created by backcrossing virgin females of nine mtDNA genome donor strains (originally from the Bloomington Drosophila Stock Center) with the nuclear genome donor males of the Oregon-RT strain (ORT, OregonR strain from the University of Tampere, Finland). Virgin females were collected from each generation and crossed with the ORT males for >10 generations [18]. Finally, the mitotype specific SNPs or indels were sequenced from each of the cybrid lines to confirm the presence of the correct mitotype [18]. We thank Emerita Laurie S. Kaguni and Emeritus Howard T. Jacobs for gifting us the cybrid strains in 2017. Based on the mtDNA sequence variation, the ten mitotypes used in this study cluster into two haplogroups; haplogroup I mitotypes have a multicontinental origin (mtKSA2, mtPVM, mtOR, mtBOG1, mtWT5A and mtM2), whereas haplogroup II mitotypes (mtPYR2, mtLS, mtBS1 and mtVAG1) all have European origin [18].

Flies were cultured on a standard diet of cornmeal Lewis medium [91] for the bacterial and viral experiments. Larval experiments were conducted in a different laboratory, where mash potato -based food medium was used (36 g mashed potato powder, 9 g agar, 45.5 ml corn syrup, 14.5 g dry yeast, 8 g nipagin and 5 g ascorbic acid per 1 liter of water). Flies were maintained at 25°C ± 1°C in 12:12LD cycle. Some of the larval stage experiments were performed at 29°C ± 1°C, as stated in the text. Cybrid lines with the mtDNA mitotypes mtORT, mtWT5A, mtKSA2 and mtBS1 were chosen for more detailed experiments (S5 Table) and mtORT and mtKSA2 cybrids were used in RNA sequencing analysis.

Bacterial infections

1–4-day old, mated females and males of a panel of ten cybrid lines were systemically infected with LYS-type peptidoglycan containing Gram-positive bacteria Staphylococcus aureus (PIG1) or DAP-type peptidoglycan containing Gram-negative bacteria Providencia rettgeri (DSM4542). Bacteria were grown overnight in LB growth media at 37°C and the inoculum was regrown to reach the exponential growth phase based on the spectrophotometer value prior to diluting and resuspending the sample in 1 x PBS to a final concentration of OD600 = 0.1. Flies were infected by pricking with a sharp 0.15 mm tungsten needle, dipped in the bacterial solution. To control the depth of the stabbing injury, the tip of the tungsten needle was bent 90° and the flies were stabbed with the bent part of the needle. Flies were stabbed in the mesopleuron at the thorax region to minimize injuries. To monitor the original bacterial dose, four females and four males were individually homogenized in 100 μl of 1 x PBS following pricking with the bacteria dipped needle. The total homogenates were plated on nonselective LB-agar plates and the numbers of bacterial cells (colony forming units, CFUs) were counted. The OD600 = 0.1 infection dose resulted on average to 150 bacterial cells per P. rettgeri infected fly and on average 30 bacterial cells in S. aureus infections. P. rettgeri and S. aureus infections were performed with six replicate vials of 10 flies each. Flies infected with P. rettgeri or S. aureus were monitored every 2 hours to record the mortality rates until the survival curves started to plateau.

Bacterial load

CFUs were measured from flies 8 and 20 hours post P. rettgeri and S. aureus infection. CFUs were also measured 48 hours after P. rettgeri infection and 72 hours after S. aureus infection in live flies. Eight flies per sex/ strain/ treatment/ timepoint were kept in separate vials post infection. To kill the cutaneous bacteria, the flies were individually placed in 100 μl of 70% ethanol for 30–60 seconds. Ethanol was removed and 100 μl of 1 x PBS was added to the vials prior to homogenization with a plastic pestle. 10-fold dilution series were prepared in 96-well plates and 3 μl of each dilution was plated on a non-selective LB-agar plate. The original homogenate (1 fly homogenized in 100 μl 1 X PBS) was diluted 1:10–1:100000 for the CFU measurements, but in some cases even more diluted samples were required. Agar plates were incubated overnight at 29°C and CFUs were counted using a light microscope. Samples were collected directly after the infection as described above to monitor the original infection dose. Uninfected flies were homogenized as negative controls, which were obtained by sampling one female and one male from vials prior to infection.

Induction of humoral immune response prior to infection

We grew P. rettgeri overnight and heat killed the bacteria by heating it for 30 minutes at 80°C. Heat killed (HK) bacteria were pelleted down and diluted to OD 0.1. Both the undiluted and the OD 0.1 sample were plated to confirm the absence of live bacteria. 2–4 days old virgin females and males were exposed to one of the three treatments by pricking with a tungsten needle: 1) exposed to heat killed bacteria and 18 hours later pricked with 1 x PBS, 2) exposed to heat killed bacteria and 18 hours later exposed to live P. rettgeri (OD 0.1) and 3) exposed directly to live P. rettgeri. Survival was followed for 50 hours after the final treatment.

Viral infections

Drosophila C virus (DCV) is a single stranded positive sense RNA virus that belongs to the Cripavirus genus. Systemic infection with DCV was introduced to 2–5 days old, mated females and males by pricking as described for the bacterial infections. The DCV inoculum of 10⁹ viral particles was thawed on ice and diluted with 10mM Tris-Hcl pH7.2 buffer to 10⁸ viral particles. Six replicate vials of ten flies per sex were used for DCV infections and four replicate vials of ten flies were used as a mock control, pricked with 10mM Tris-Hcl buffer. Survival was monitored once a day for 14 days. Male flies were also infected with Kallithea virus, a double stranded DNA virus. Females were excluded from the experiment as Kallithea mainly affects the survival rates of males [47]. Survival rates were monitored once a day for 22 days after microinjecting 10⁸ viral particles. 5–7 replicate vials of ten males per vial were used in the infections and 5 replicate vials of controls, microinjected with 1 x PBS.

Sampling and total RNA extraction for RNA sequencing

1–4-day old virgin mtORT and mtKSA2 females and males were infected with P. rettgeri, S. aureus and DCV as described above. We collected four replicate vials of 10 females and males at 8 and 20 hours post bacterial infections, 3 days after exposure to DCV, and from age-matched untreated controls. Prior to total RNA extractions, the flies were snap frozen on dry ice and stored at -70°C. Frozen flies were homogenized in 120 μl of TRI reagent (MRC, Thermo Fisher Scientific) using a plastic pestle. Total RNA was extracted according to the manufacturer’s instructions (MRC, Fisher Scientific). RNA pellets were dissolved in 50 μl of nuclease-free water and treated with Deoxyribonuclease I according to the manufacturer’s instructions (Invitrogen). The RNA concentration and the purity of the samples were measured with Nano-Drop ND-1000 spectrophotometer (Thermo Fisher Scientific) and the samples were stored at -80°C.

RNA sequencing and data analysis

RNA sequencing and data processing was performed by Edinburgh Genomics, the Genomics and Bioinformatics core facility at the University of Edinburgh (UK). RNA sequencing libraries were prepared using Illumina TruSeq Stranded mRNA Library Prep kits and sequenced using Illumina NovaSeq. Reads were trimmed using Cutadapt (version cutadapt-1.18-venv) and trimmed for quality at the 3’ end using a quality threshold of 30 and for adaptor sequences of the TruSeq DNA kit (AGATCGGAAGAGC). After trimming, reads were required to have a minimum length of 50. D. melanogaster (BDGP_ens107), S. aureus (GCS_000013425.1), P. rettgeri (GCF_003204135) and DCV (NC_001834.1) references were used for mapping and the standard GTF-format annotation for counting. Reads were aligned to the reference genome using STAR version 2.7.3a [92] specifying paired-end reads and the option—-outSAMtype BAM Unsorted. All other parameters were left at the default parameters. Reads were assigned to features of type ‘exon’ in the input annotation grouped by gene_id in the reference genome using featureCounts version 1.5.1 [93]. Genes with biotype rRNA were removed prior to counting. featureCounts assigns counts on a fragment basis as opposed to individual reads so that fragment is counted where one or both of its reads are aligned and associated with the specified features. Strandedness was set to ‘reverse’ and specified a minimum alignment quality of 10. In addition to the counts matrix used in the downstream differential analysis, a matrix of Fragments per Kilobase of transcript per Million mapped reads (FPKM) was generated using the rpkm function of edgeR version 3.28.1 [94]. Gene lengths for the FPKM calculation were the number of bases in the exons of each gene and bases were counted once where they occurred in multiple exon annotations. Gene names and other fields were extracted from the input annotation and added to the count/ expression matrices. The raw counts table was filtered to remove genes consisting of mostly near-zero counts in different treatments and filtered on counts per million (CPM) to avoid artifacts due to library depth. If applicable, a row of the expression matrix was required to have values greater than 0.1 in at least four samples, corresponding to the smallest sample group as defined by Group once samples were removed. Reads were normalized using the weighted trimmed mean of M-values method [95], passing ‘TMM’ as the method to the calcNormFactors method in edgeR. Differential analysis, contrasting mtORT and mtKSA2 mitotypes or infected vs. uninfected animals, was carried out with edgeR (version 3.28.1) [94]. Fold changes were estimated as per the default behaviour of edgeR in order to avoid artifacts. Statistical assessment of differential expression was carried out with the quasi-likelihood (QL) F-test. The RNA sequencing data is available at NCBI’s Gene Expression Omnibus (GEO) [96,97] data repository under the accession number GSE260986. Prior to the GO term analysis, we filtered the data to exclude non-coding genes and to include protein coding genes with False discovery rate (FDR) adjusted P-value < 0.15 and log2 FC ≤ -0.5 for downregulation and ≥ 0.5 for upregulation.

Respirometry

200 uninfected, 1–3 days old male flies per replicate were collected and were left to recover from the CO₂ anaesthesia for 24 hours. Flies were quickly knocked out on ice, transferred to a chilled mortar, and crunched 50–60 times with a ceramic pestle in 500 μl of cold homogenization buffer, containing 250 mM sucrose, 2 mM EGTA and 5mM Tris/Hcl (pH 7.4). Lysates were gently brushed through a 200 μm nylon mesh to filtrate the sample. Mortar and mesh were further rinsed with 500 μl of cold homogenization buffer. Samples were kept on ice and respirometry measurements were performed immediately after removing 100 μl for measuring the protein concentration using the BCA assay kit and manufacturer’s protocol (Thermo Fisher Scientific).

Hansatech oxygen electrode and Oxygraph Plus software were used for the respirometry measurements. The oxygraph chamber was filled with 475 μl of respiration buffer, containing 120 mM KCl, 1mM EGTA, 1 mM MgCl₂, 0.2% BSA, 5 mM KH₂PO₄ and 3 mM Hepes/KOH (pH 7.2). The buffer was kept in the open chamber at 25°C with constant stirring for approximately 5 minutes until the oxygen concentration stabilized. After that, 25 μl of the lysate was added to the chamber, the chamber was closed and the oxygen concentration was measured before and after the sequential addition of OXPHOS complex I, III and IV specific substrates and inhibitors (S7 Table). The oxygen consumption rate for each complex was recorded and normalized to the sample protein content.

mtDNA copy number

mtDNA copy number was measured 8 and 20 hours after P. rettgeri and S. aureus infection and from uninfected male flies. Total DNA was isolated from pools of five flies using TRIzol™ Reagent, following the manufacturer’s protocol. Snap frozen flies were homogenized using a motorized pestle and 135 μl of Trizol reagent. DNA pellets were dissolved in 30 μl of nuclease-free water. 40 ng of total DNA was used as template in quantitative PCR using primers for the mitochondrial target gene 16S rRNA and the nuclear target gene RpL32 (S8 Table). Reactions were performed in StepOnePlus instrument (Applied Biosystems) using the Fast SYBR Green Master Mix (Applied Biosystems) under the manufacturer’s recommended conditions: 95°C for 20 s, followed by 40 cycles of 95° for 3 s and 60° for 30 s. Three biological and two technical replicates were used. Relative mtDNA copy number was calculated as 2*2^dCt, where dCt = nuclearCt - mitoCt (the threshold cycle values for the nuclear gene RpL32 and mitochondrial gene 16S).

H₂O₂ quantification

Hydrogen peroxide (H₂O₂) levels were measured from 2–5 days old uninfected males and 8 hours and 20 hours post P. rettgeri or S. aureus infection. Three biological replicates of pools of five flies were used for each treatment. Flies were homogenized with a plastic pestle in 200 μl of chilled 1 x PBS. The fly debris was pelleted by centrifuging 1 min at 12,000 x g at 4°C. Sample supernatant was used in the H₂O₂ measurements and in the protein assay. Protein samples were snap frozen with liquid nitrogen and stored at -70°C. Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit (Invitrogen, Catalog no. A22188) was used for measuring the H₂O₂ according to the manufacturer’s protocol. Amplex Red reagent reacts with H₂O₂ to produce the red-fluorescent oxidation product resorufin, which has an excitation and emission maxima of approximately 571nm and 585nm, respectively. A microplate reader (VarioskanFlash, Thermo Fisher Scientific) equipped for measuring 560 nm absorbance was used for detecting the resorufin which has a 1:1 stoichiometry with H₂O₂. Because the reaction is continuous the fluorescence was measured at multiple time points to follow the kinetics of the reaction and 45 minutes after starting the reaction was selected for the analysis. The background fluorescence was corrected by subtracting the derived values from the no-H₂O₂ control. Total protein concentration was measured from each fly homogenate using the Bio-Rad Protein Assay (BioRad) and the of H₂O₂ was normalized to the amount of mg/ml of protein.

Parasitoid wasp assay

The parasitoid wasp Leptopilina boulardi strain G486 was used as a natural model to induce the cell-mediated innate immune response in Drosophila larvae. Female flies were allowed to lay eggs for 24 hours at 25°C. The eggs were then either kept at 25°C or transferred to 29°C for further development, depending on the experiment. 2^nd instar larvae were exposed to 12–15 female wasps for two hours at RT (22°C), after which the wasps were removed, and the larvae placed back at their rearing temperature. The encapsulation response of the Drosophila larvae was checked from 3^rd instar larvae, 72 hours after the infection when the larvae were reared at 25°C and 48 hours post infection at 29°C, by dissecting the larvae individually in a drop of water under a stereomicroscope. The cellular immune response of the larvae was determined based on their ability to encapsulate and melanise the wasp egg/larva either as i) no melanisation (-): the larva failed to encapsulate the wasp egg /larva and the wasp continued its development, ii) partially melanised (-/+): the larva was able to elicit the cellular immune response against the wasp larva, which was partially melanised (encapsulated) and iii) melanised (+): the wasp larva was completely encapsulated and it was terminated as a sign of a successful cellular immune response. Three replicate crosses per genotype were made, and larvae were dissected until 50 infected larvae per replicate had been collected (150 larvae per genotype in total).

Larval hemocyte analysis

Late 3^rd instar larvae were gently washed in a drop of filtered water and individually placed in 20 μl of 1% BSA in 1 x PBS on a 12-well glass plate. The hemolymph was bled out by dissecting the larvae open from one side, along the full length of the larvae. 80 μl of 1% BSA in PBS was added to the sample and a BD Accuri C6 flow cytometer was used to analyse the hemocytes. Each cybrid line was tested on three separate occasions (on different weeks) and each time 10 larvae were dissected per strain. To assess the hemocyte counts, we utilized the fact that hemocytes can be easily separated from other particles based on their location in a forward scatter–side scatter plot ([36], S1A Fig). Dead cells were excluded from the analysis by Propidium iodide (PI) staining (1μl per 100 μl sample). Plasmatocytes and crystal cells are small and round in their inactivated state [98] and can be gated using the forward scatter area vs. forward scatter height plot, where round cells form a population detected at an 45° angle. Lamellocytes are irregularly shaped [98], which places them outside of the 45° angle in the forward scatter area vs. forward scatter height plot. These characteristics were utilized in the flow cytometry to separate plasmatocytes and crystal cells from activated immune cells appearing in circulation after an immune challenge, per Järvelä-Stölting et al. [99]. We further confirmed this gating strategy by using in vivo fluorescent hemocyte reporter lines eater-GFP (plasmatocyte reporter; [100]) and msnF9mo-mCherry (lamellocyte reporter; [101]; Fig S1C-S1C’). We followed the infection scheme described in section 2.2 to assess the hemocyte populations from wasp-infected cybrid larvae at 25°C and 29°C. Larvae were reared at 25°C or 29°C prior quantification of the crystal cells by heating 10–12 3^rd instar larvae in 100 μl of 1 x PBS at 60°C for 10 minutes. Heating induces crystal cell melanisation, making it possible to visually quantify the number of crystal cells [102]. Crystal cells were quantified from the three most posterior segments of the larvae. Larvae were placed on a drop of glycerol under a cover slip on a glass slide for imaging.

Total RNA extraction and RT-qPCR analysis of larval hemocytes

To extract total RNA from larval hemocytes, the hemocytes were collected from pools of 50 male or female third instar larvae, in ice cold 1X PBS. Samples were centrifuged for 7 minutes at 2500 g at 4°C and the hemocyte pellets were stored at -80°C. Total RNA was extracted using the Single Cell RNA Purification Kit (Norgen Biotek) including DNase I treatment (RNAse free DNase I kit, Norgen Biotek) as described in the manufacturer’s protocol. The RNA quality was checked with a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific) and each sample was diluted to 10 ng/μl and stored at -80°C. qRT-PCR reactions were performed using the iTaq Universal Sybr green One-step kit (Bio-Rad) with the following reaction mix per sample: 5 μl of iTaq universal SYBR Green reaction mix (2x), 0.125 μl of iScript reverse transcriptase (RT), 0.3 μl of forward and reverse primers each (300 nM each), 2 μl of template RNA (20 ng per reaction) and 2.275 μl of nuclease free H₂O. Each experiment included a sample without RT and without an RNA template for quality control. The reactions were run with a Bio-Rad CFX96 Real-time PCR system with a reverse transcription reaction for 10 min at 50°C, polymerase activation and DNA denaturation for 1 min at 95°C followed by 39 cycles of amplification (denaturation for 10 sec at 95°C, annealing/extension for 15 sec at 60°C) and a melt-curve analysis at 65–95°C in 0.5°C increments. Two technical and three biological replicates were included for each sample. Expression levels were measured from PPO3 (Prophenoloxidase 3) and Histone H3.3B (His3.3B) was used as a control gene (S8 Table). Efficiency of each primer pair was calculated based on a 5-fold serial dilution of the template RNA (each point assayed in duplicate), using the following formula: Efficiency (%) = (-1/10^slope– 1) x 100.

ROS in larval hemocytes

Pools of ten male and ten female late third instar larvae per strain reared at 29°C were washed three times with filtered water and then dissected in 250 μl of 1 x Schneider’s Drosophila Medium (Gibco) on ice to obtain hemolymph. The samples were incubated with 500 nM of the CellROX green reagent (molecular probes for Life Technologies) diluted in Dimethyl Sulfoxide (DMSO) for 1 hour at room temperature in the dark. After incubation, hemocytes were stained with 2.5 μl of Propidium iodide (PI) and immediately analyzed using the BD Accuri C6 flow cytometer. Each cybrid line was tested four times. CellROX green was detected using a 530/33 nm filter and PI using a 610/20 nm filter, excitation was with the 488nm laser. 2000 live hemocytes were analysed per sample, each sample contained ~10⁴ hemocytes in total. One-color controls (non-fluorescent, PI-only and CellROX-only hemocytes) were used to set the gates (S9A–S9B Fig).

Mitochondrial membrane potential

To measure the mitochondrial membrane potential in larval hemocytes, the MitoProbe TMRM Assay Kit for Flow Cytometry (Invitrogen) was used. Three replicate samples for each cybrid line were analyzed, and each sample contained hemocytes from 30 male 3^rd instar larvae. The larvae were washed, dissected in 1x PBS, and stained with 0.4 nM TMRM for 30 minutes in the dark. To distinguish between living and dead cells, SYTOX Blue Dead Cell Stain (Invitrogen) was used. The hemocytes were incubated with the stain (0.1 nM) for 15 minutes. The samples were analysed using the CytoFlex S flow cytometer (Beckman Coulter). The following channels and laser gain settings (in parentheses) were used: FSC (20), SSC (40), PE (for TMRM detection, 80), PB450 (for SYTOX detection, 40).

Hemolymph melanisation assays

Cybrid flies were kept at 25°C and vials with eggs were transferred to 29°C for egg-to-larva development. Pools of three third instar larvae were dissected per well on a 12-well slide in a drop of 1 x PBS to release the hemolymph. Carcasses were removed and the slides were imaged as soon as possible (within 10 minutes of the dissection start) using a stereomicroscope and Nikon DS-Fi2 camera. Slides were put in a container with wet tissue paper at the bottom to prevent the hemolymph samples from drying, and imaged at 15, 30, 170 and 260 minutes after the start of dissection. Samples were left in the container overnight and imaged once more the next morning (endpoint of the experiment, ~ 24 h after dissections). We conducted the experiment on two subsequent days (2 experimental blocks) and pooled the data from these experiments. Melanisation intensity was quantified using ImageJ2 (version 2.14.0/1.54f).

Statistical analyses

We used R versions 4.0.2.– 4.3.2. to analyze the data [103]. For plotting, ggplot2 [104] was utilized. To test for overall differences in survival, Log-Rank (Mantel-Cox) test was applied using the GraphPad Prism 9 software. For hazard ratios, log-rank tests were applied using the Survival package [105]. To test for differences in CFUs (Fig 1C-1D’), linear models with a natural log transformed response were fitted and the significance of the interaction effects and the main effect of the mitotype was assessed using Type III sums of squares, (Anova function available in the car package) [106]. Model fits were assessed using standard diagnostic plots. For the priming data (S3 Fig), Cox mixed effects models [105,107] were fitted to test for differences in fly survival, using hour of death as the response variable, strain and sex as dependent variables and vial as a random intercept. Flies that were dead at the end of the experiment were included as censored cases. After running the survival models, we tested whether they fulfilled the assumptions of proportional hazards over time using the cox.zph function [108]. Neither model fulfilled the assumptions of proportional hazards over time, so we used the survival regression function (survreg) to compare results. In both cases, the coxph models and exponential survival regression models gave qualitatively similar results so we present the results of the Cox mixed effect models for ease of interpretation. We analyzed differences in H₂O₂ levels using linear models with a natural log transformed response where appropriate (Fig 6). P-values were corrected for multiple testing using the p.adjust function and a Holm correction. RT-qPCR, mtDNA copy number and respiration data were analyzed by modeling mitotype as independent variable and each replicate experiment as a random effect, followed by Tukey’s HSD for pairwise comparisons. The data on melanisation of wasp eggs and larvae were analysed using a generalized linear model with binomial distribution using the lm4 package [109] with fully melanised or fully + partially melanised cases denoted as a success and non-melanised as a failure, and mitotype as an independent variable. Tukey method in the multcomp package was used for pairwise comparisons between the mitotypes [110]. The hemocyte count data were analyzed using a negative binomial generalized linear model using the MASS package [111], with mitotype as independent variable and replicate experiment as a random effect. The least square means (estimated marginal means) were analyzed for multiple comparisons using the emmeans package and the Tukey method was used for adjusting the p-value [112]. The data on the proportion of PI-positive (live) hemocytes was analysed using a generalized linear model with a binomial distribution (lme4 package) with live hemocytes as a success and dead hemocytes as a failure, mitotype as an independent variable and replicate as a random effect. For comparisons between the mitotypes, pairwise contrasts were analyzed using the emmeans package. Adobe Illustrator version 27.9 was used to edit and compile the figures. Data and code for statistical analyses are available at Zenodo (https://zenodo.org/records/14162508).