Antimicrobial peptides do not directly contribute to aging in Drosophila, but improve lifespan by preventing dysbiosis

Antimicrobial peptides (AMPs) are innate immune effectors found in plants and animals that show microbicidal activity in vitro. They are typically cationic and amphipathic, and disrupt microbial membranes that are more negatively charged (Broekaert et al., 1997; Christensen et al., 1988; Ludtke et al., 1996; Zasloff, 1987). Recent studies in various animal models have pointed to a role for AMPs beyond microbial infection, notably in aging and aging-related diseases (Deslouches and Di, 2017; Mookherjee et al., 2020; Semple and Dorin, 2012; Smith and Gwyer Findlay, 2022). Like other animals, genes encoding AMPs (hereafter AMP genes) are upregulated upon aging in Drosophila, and disparate studies have suggested a role for AMPs as causative agents of aging, neurodegeneration, or mitochondrial stress (Arora and Ligoxygakis, 2020; Garschall and Flatt, 2018; Hanson and Lemaitre, 2020). Here, we have leveraged a set of isogenic AMP-deficient flies to analyse the contribution of AMPs to aging in Drosophila melanogaster.

AMPs are well characterized in Drosophila for their contribution to host defence. There are currently eight families of inducible AMPs known in D. melanogaster: the antifungals Drosomycin (Drs), Baramicin (Bara) and Metchnikowin (Mtk); Cecropins (Cec) and Defensin (Def), which have both antibacterial and some antifungal activities; and Drosocin (Dro), Attacins (Att) and Diptericins (Dpt), which primarily exhibit antibacterial activity (Hanson et al., 2021; Imler and Bulet, 2005). In addition, the Drosophila genome encodes many other host defence peptide families such as Daisho (Dso) and Bomanin (Bom), for which overt antimicrobial activity in vitro has not yet been demonstrated, but functional studies have shown that they are important in vivo to resist microbial infection (Clemmons et al., 2015; Cohen et al., 2020). Although most Drosophila defence-peptide-encoding genes are strongly induced in the fat body in response to systemic infections, many show specific and complex patterns of expression in tissues such as the trachea, gut, salivary glands or reproductive tracts (Ferrandon et al., 1998; Tzou et al., 2000).

AMP genes are regulated at the transcriptional level by the Toll and Imd NF-κB signalling pathways upon systemic infection, or by the Imd pathway in local epithelia (Lemaitre and Hoffmann, 2007; Myllymäki et al., 2014). It is well established that Imd- and Toll-deficient flies show marked susceptibility to microbial infection. Until recently, the importance of immune effectors downstream of these pathways, notably AMPs, was unclear. With the advent of CRISPR gene editing, we systematically deleted seven families of AMP genes (Defensin, Cecropin, Drosocin, Attacin, Diptericin, Drosomycin and Metchnikowin) and analysed their contributions to host defence individually or collectively. We found that Drosophila AMPs are essential downstream of the Imd pathway to resist systemic infection by Gram-negative bacteria. Drosophila AMPs also contribute downstream of the Toll pathway to combat fungal and, to a lesser extent, Gram-positive bacterial infection, although Bomanins play a more prominent role against these micro-organisms (Carboni et al., 2021; Clemmons et al., 2015; Hanson et al., 2019a). Use of fly lines carrying combinations of AMP mutations revealed that they can function either additively or synergistically against some microbes, but, in some cases, AMPs exhibit striking specificity, with one peptide contributing most of the AMP-dependent defence against a specific pathogen (Hanson et al., 2019a, 2022a,b; Unckless et al., 2016). Drosophila AMPs are also important to control the fly microbiome downstream of the Imd pathway, particularly for their role in regulating Gram-negative bacteria like Acetobacter (Marra et al., 2021).

Although Drosophila AMPs were initially investigated for their contribution to host defence, AMP upregulation is also observed in non-immune contexts, including anti-tumour defence (Araki et al., 2019; Krautz et al., 2020; Parvy et al., 2019), neurodegeneration (Cao et al., 2013; Kounatidis et al., 2017; Petersen et al., 2013; Shukla et al., 2019; van Alphen et al., 2022) and aging (Glittenberg et al., 2011; Lai et al., 2007). This suggests that AMPs could have roles beyond their traditional function as microbicidal agents. Notably, transcriptomic studies revealed that high AMP expression, reflecting an increase in Imd pathway activity, is a hallmark of aging in flies (Landis et al., 2004; Pletcher et al., 2002; Seroude et al., 2002; Zerofsky et al., 2005). This is reminiscent of the situation observed in humans where a low-grade chronic inflammation is observed upon aging, termed ‘inflammaging’ (Franceschi et al., 2000, 2017; Liang and Diana, 2020). The key question now is to decipher whether this high and chronic activation of the immune response is simply a marker of aging, or whether cytotoxic immune effectors like AMPs accelerate aging-associated syndromes.

To address this question, several reports have investigated the role of the Imd pathway and AMPs in aging in Drosophila with contrasting outcomes. Most of these studies targeting Imd itself or the Imd transcription factor Relish (Rel) report that Imd pathway mutants are short-lived, (Cai et al., 2021; Kounatidis et al., 2017; Petersen et al., 2013). However, one study suggested that mutation of the Imd gene itself improved lifespan, as did fat-body-specific or whole-body AMP knockdown (Lin et al., 2018). The negative impact of Imd pathway downregulation has been associated with defects in gut homeostasis (Buchon et al., 2009), exaggerating the decline in gut resilience upon aging (Rera et al., 2012), leading to invasion of microbes into the hemolymph that drive mortality (Clark et al., 2015). Indeed, the guts of aged Relish mutant flies display precocious loss of compartmentalization, increased permeability and dysplasia (Liu et al., 2022). However, preventing the over-activation of Imd in the gut through transgene expression can improve lifespan, suggesting a careful balance of immune signalling in the gut is needed for optimal health with aging (Cai et al., 2021; Guo et al., 2014; Iatsenko et al., 2018). Some of these symptoms could be reverted when flies were raised axenically, pointing to a role of the microbiome in the precocious aging of the gut (Buchon et al., 2009; Clark et al., 2015; Iatsenko et al., 2018; Liu et al., 2022; Zhu et al., 2021). Another study reports that AMP genes are strongly induced in the head of old flies and that silencing Relish in glia can extend lifespan (Kounatidis et al., 2017). This would suggest that Imd-mediated immune responses drive aging by directly affecting brain activity. Indeed, many studies report that suppression of the Imd pathway can prevent neurodegeneration in various neurodegenerative disease models. For instance, flies mutated in the serine/threonine kinase ataxia telangiectasia mutated (ATM, also called tefu) display reduced lifespan associated with increased neuronal lesions, which can be rescued by combination with Relish mutation (Petersen et al., 2013). Relish deletion also rescues the aging-dependent neurodegeneration of Cdk5α mutant flies (Shukla et al., 2019), and interesting recent studies even showed that deletion of AMPs can improve fly survival after traumatic brain injury (Swanson et al., 2020; van Alphen et al., 2022). An effect of AMP overexpression in aging has also been proposed, albeit with conflicting results. Cao et al. (2013) reported AMP upregulation and neurodegeneration in the brains of dnr1 mutant flies, and showed that AMP overexpression in neurons was sufficient to induce neurodegeneration. Badinloo et al. (2018) found that chronic and ubiquitous AMP overexpression reduced fly lifespan alongside induction of mitochondrial stress. A recent study of the AMP gene Metchnikowin further suggests a trade-off between greater antimicrobial activity and host fitness (Perlmutter et al., 2023). These results suggest AMPs could be deleterious to host fitness, which is also supported by evolutionary studies showing that AMP deletions segregate in wild populations (Early et al., 2017; Hanson et al., 2019b). However, in direct contradiction to this idea, an earlier report found that excess expression of certain AMPs could extend lifespan (Loch et al., 2017).

Presently, we do not know how AMPs impact aging, and a direct link between AMPs and aging remains to be demonstrated. Some of the conflicting results mentioned above may arise from use of different conditions (e.g. temperature, sex, mating status, nutrition) (Camus et al., 2019; Dobson et al., 2018; Landis et al., 2015; Miquel et al., 1976), microbiome differences, which could help explain contradictory findings on the impact of germ-free conditions on lifespan (Brummel et al., 2004; Ren et al., 2007), or lack of control over genetic background, including transgene insertions (Ferreiro et al., 2017; Sasaki et al., 2021), all of which can influence longevity. In the absence of mutants, these studies have relied on the use of over-expression or RNAi to modulate AMPs – methodologies with certain limitations (Tower et al., 2017; van der Graaf et al., 2022). As a result, exactly how AMPs contribute to aging remains unclear.

Here, we leveraged a recently generated collection of AMP mutations to analyse whether these immune effectors impact aging. We found that individual AMP deletions did not markedly affect aging, with the possible exception of Defensin. However, ΔAMP14 flies lacking 14 AMP genes displayed reduced lifespan associated with microbial dysbiosis. Rearing ΔAMP14 flies in germ-free conditions significantly rescued lifespan, indicating that AMPs contribute to lifespan through their impact on the microbiome. In contrast, depleting the microbiome of Relish mutant flies did not rescue lifespan, suggesting the Imd pathway can affect lifespan independent of its regulation of AMP genes. Together, our results indicate that AMPs are likely not direct contributors to the aging process, though they have a major impact on aging through AMP-microbiome interactions. Thus, our study, using loss-of-function mutations, clarifies the role of these innate immune effectors in aging.

To address the role of AMP genes in aging, we first compared the lifespan of flies lacking eight AMPs located on chromosome II (ΔAMP Chr2 lacking Def, Dro, AttA, AttB, AttC, Mtk, DptA and DptB) to the lifespan of their DrosDel isogenic wild-type (iso w¹¹¹⁸) controls. Of note, all our mutations are in the DrosDel isogenic genetic background (unless mentioned otherwise) and are negative for the endosymbiont Wolbachia. We show the lifespans of male flies in the main figures and those of female flies in the supplementary figures. In most cases, trends were similar between the two sexes. Cases where trends differed between males and females are noted in the main text. We observed a striking effect where ΔAMP Chr2-deficient flies displayed a marked lifespan extension compared to wild-types (Fig. 1A). This first result is consistent with studies suggesting that AMPs negatively impact lifespan. However, we were surprised by the short lifespan of our iso w¹¹¹⁸ wild-type flies compared to wild-type lifespans from other aging studies, and to a second wild-type included in these experiments – Oregon R (OR-R in Fig. 1A). Drosophila can carry a number of cryptic bacterial or viral infections that affect fitness (Plus et al., 1976). Notably, infection with Drosophila Nora virus is common in lab stocks, and this virus has been shown to reduce lifespan (Habayeb et al., 2009). We thus suspected some component of the virome/microbiome of our iso w¹¹¹⁸ flies could be affecting our iso w¹¹¹⁸ wild-type but not ΔAMP Chr2 flies. We therefore cleared our flies of their virome/microbiome through bleaching, allowed microbiome recolonization by microbes in the food medium, and assessed the lifespan of the cleaned iso w¹¹¹⁸ stock (protocol in the Materials and Methods). At the same time, we screened our iso w¹¹¹⁸ flies for a panel of common contaminating viruses; specifically Drosophila sigmavirus, Drosophila A virus, Drosophila C virus and Drosophila Nora virus. We did detect Drosophila Nora virus (hereafter ‘Nora’) in our iso w¹¹¹⁸ flies, but not in our ΔAMP Chr2 mutants, nor in other stocks included in these experiments. We also detected Nora virus in four other genotypes at different times during our 5 years of study (AttC^Mi, Bom^Δ55C, Group C and OR-R; defined later in the text). In some cases, these stocks were previously Nora-negative, and so were seemingly contaminated from standard fly tipping (Fig. 1B). Strikingly, the bleaching treatment markedly improved the lifespan of all of these stocks to rival their contemporaries, such as ΔAMP Chr2 flies. The net lifespan reductions for iso w¹¹¹⁸ and OR-R were ∼39% and ∼23%, respectively, possibly suggesting genetic background effects in susceptibility to Nora virus. We conclude that our wild-type reference had artificially reduced lifespan, likely due to a cryptic Nora virus infection. Although we took care to control for genetic background effects, this stock, at the time, was not an appropriate baseline for comparison. These considerations are in line with recommendations that healthy wild-type stocks should live to median lifespans of 70-90 days (Piper and Partridge, 2018), emphasizing the importance of considering the health of our control stocks prior to making comparisons across stocks/genotypes.

After detecting this contamination in our wild-type flies and seeing the impressive deleterious effect Nora had on lifespan in our hands, we screened various lab stocks for Nora virus. We found that 20 out of 44 arbitrarily selected stocks (both from our lab and others) were Nora positive, though Nora titres were typically found only at low levels under standard rearing conditions. We also checked RNA samples collected from aged Nora-positive flies (aged >3 weeks), finding consistently high Nora titres (Fig. 1C). These stocks also often showed abdominal bloating at older ages, pre-empting mortality (M.A.H., unpublished observation), similar to bloating seen in flies infected with Drosophila C virus (Chtarbanova et al., 2014) or after systemic infection with some strains of Acetobacter bacteria (Hanson et al., 2022b). We thus suspect Nora virus contamination contributed greatly to the early mortality of our iso w¹¹¹⁸ flies in our first lifespan experiments (Fig. 1A), with a more drastic effect than previously reported (Habayeb et al., 2009). However, we will note that we did not intentionally perform Nora infection experiments, and so here we provide only correlation-based evidence.

Following these results, we screened all our fly stocks for Nora virus and bleached Nora-positive stocks before use in experiments. Subsequent longevity experiments were carried out with the following conditions: we used our standard food medium (recipe in the Materials and Methods), flipped flies three times per week, with ∼20 flies per vial, using mated males or females (sexes kept separate), and performed lifespan measurements at 25°C unless otherwise specified. As an assay for brain health, we also monitored locomotor competence during aging using the climbing pass rate assay at 5, 40, 50 and 60 days post eclosion (dpe), unless specified otherwise.

To compare lifespan, we first used a Cox mixed model commonly used in the literature. However, we found this statistical method was overly sensitive for our purposes, as even minor differences in lifespan were returned as highly significant (P<0.001). In some cases, this was driven by sex×genotype interaction effects. Because we kept males and females in separate vials, we realized that any putative sex×genotype interactions present in our Cox mixed model were indistinguishable from vial effects. This is especially important as a consideration as our AMP mutants are known to suffer dysbiosis with aging (Marra et al., 2021 and see below), compounding the impact of vial-specific microbiome stochasticity (see File S1 for further discussion). Visual inspection of survival curves suggested that, by and large, even significant Cox mixed model variation in survival largely reflected variation around the mean wild-type lifespan, which is expected when performing multiple hypothesis tests (Streiner and Norman, 2011). Thus, we paired our lifespan analyses with one-way ANOVA statistics run on median lifespans per experiment, intended to get a more stringent measure of lifespan differences by placing greater value on effects that were consistent across experiments. As such, we used one-way ANOVA P-values by default in the text to report significant lifespan differences.

Before we addressed the impact of AMP mutations on lifespan, we analysed the lifespans of mutants affected in the Toll (iso spz^rm7) or Imd (iso Rel^E20) pathways that have been backcrossed in the DrosDel background (Ferreira et al., 2014). We also compared our DrosDel iso w¹¹¹⁸ white^−/− flies with another wild-type (OR-R), which has white^+/+ red eyes. Of note, white gene deletion may affect lifespan through, for example, the role of white in the brain (Ferreiro et al., 2017), or through white-mediated regulation of intestinal stem cell proliferation (Sasaki et al., 2021). We further included non-isogenic flies with previously reported lifespan effects to show how control genotypes behave in our conditions, and set expectations for the size and consistency of lifespan effects we could reasonably observe in our hands compared to those observed in other studies. These included: methuselah mutants suggested to have exceptionally long lifespan (mth¹) (Lin et al., 1998), a dnr1 mutation that reduces lifespan previously associated with neurodegeneration and aberrant AMP induction (dnr1^2-133) (Cao et al., 2013), and ATM mutant flies with precocious neurodegeneration caused by autophagy defects (ATM⁸), also associated with Imd pathway activation (Petersen et al., 2013). mth¹, dnr1^2-133 and ATM⁸ flies were not backcrossed into the DrosDel background and had red eyes. Our experiments confirmed that ATM⁸ flies had a significantly shorter lifespan (Fig. 2A,C) associated with poor climbing competence (Fig. 2D). We also observed that our OR-R wild-type displayed a shorter lifespan compared to that of our DrosDel iso w¹¹¹⁸ wild-type. Contrary to expectation, the lifespan of mth¹ flies was not longer than that of wild-type flies (P>0.10, Fig. 2A,C). However, we did observe improved climbing competence of mth¹ flies into old age (Fig. 2D), which suggests these flies do have a form of improved fitness with aging in our hands. These experiments also confirmed a reduced lifespan in Rel^E20 flies as found by other studies (Kounatidis et al., 2017; Petersen et al., 2013), suggesting a significant contribution of the Imd pathway transcription factor Relish to aging.

We also repeated these lifespan experiments at 29°C, which represents a more stressful temperature for D. melanogaster, causing more rapid aging (Miquel et al., 1976). When we reared flies at 29°C, as expected, we observed precocious aging in terms of both lifespan and climbing pass rate compared to rates seen in flies reared at 25°C for all genotypes tested: iso w¹¹¹⁸, OR-R, dnr1^2-133, mth¹ and Rel^E20 (Fig. S2A,B). Again, mth¹ flies retained climbing competence longer than other genotypes (Fig. S2C), despite no lifespan extension effect. Of note, differences in lifespan between OR-R and iso w¹¹¹⁸ were lost when flies were raised at 29°C. The previous dnr1 mutant study showing reduced lifespan used 29°C (Cao et al., 2013), and we found that dnr1^2-133 flies generally had reduced lifespan compared to the lifespans of iso w¹¹¹⁸ and OR-R wild-type at 29°C (Fig. S2B), albeit the lifespans were only significantly different between female OR-R and dnr1^2-133 flies (P<0.05). However, at 25°C, the lifespan of dnr1^2-133 flies was not significantly reduced compared to that of OR-R flies (P>0.10).

Collectively, we observed that ATM, dnr1 and Relish deletion reduced lifespan, consistent with previous studies (Cao et al., 2013; Kounatidis et al., 2017; Petersen et al., 2013; Shaposhnikov et al., 2014). Surprisingly, we did not observe lifespan extension in methuselah mutants. However, mth¹ flies retained climbing competence far better than any other genotype assayed, indicating these flies did have a form of improved aging in our hands. Thus, we could broadly replicate the trends of findings from previous studies, although the lifespan measurements were not always repeatable in our hands.

We next analysed the lifespan of flies lacking individual or small genomic clusters of AMP genes at 25°C. These mutants, for all ‘classical’ AMPs from the seven gene families initially identified, included: a Defensin mutant (Def^SK3; referred to as Def); deletion of the four Cecropin genes CecA1, CecA2, CecB and CecC (collectively termed Cec); mutation of Dro alone (Dro^SK4); mutations of Dro, AttA and AttB (DroAttAB); single mutations of AttC, AttD, Mtk and Drs; single or combined mutations affecting the two Diptericins (Dpt^SK1), DptA (DptA^Δ822 or DptA^S69R) and DptB (DptB^KO); as well as mutants for the more recently described Toll-regulated effector genes Daisho (Dso1 and Dso2; collectively Dso), Baramicin A (BaraA1 and BaraA2; collectively BaraA) and Bomanins (Bom^Δ55C, indicated as Bom; deletions of ten Bom genes at cytogenetic map 55C).

We found no major effect on lifespan of any flies containing single AMP mutations, as survival curves tended to disperse randomly around the iso w¹¹¹⁸ lifespan curve, and median lifespans were not significantly different from those of iso w¹¹¹⁸ (Fig. 2B,C). The only exception to this trend was male Def^SK3 flies, which had noticeably reduced lifespan compared to that of iso w¹¹¹⁸ (P=0.059). In general, climbing competence was also distributed in a range similar to that of wild-type iso w¹¹¹⁸ or OR-R flies, and most mutants were not exceptionally good climbers into old age like mth¹ flies, though Mtk and Drs males had somewhat improved climbing into old age (Fig. 2D; Fig. S5). The reduction in lifespan for Def^SK3 flies was sufficiently interesting that we tested this effect using ubiquitous (Actin5C-Gal4) or glia-specific (Repo-Gal4) Def interfering RNA (Def-IR, RNAi); however, we saw, if anything, the opposite effect when silencing Def by RNA interference (RNAi or ‘-IR’) compared to the results seen with the Def^SK3 mutant (Act>Def-IR males had longer lifespans and females shorter lifespans). However, RNAi genetic controls also suggested complex genetic background effects independent of the Gal4 and RNAi constructs, confounding meaningful interpretations of those results. In general, we could not support a reduced lifespan effect of single AMP mutation with RNAi (Fig. S4).

Collectively, our study using isogenic mutants shows that individual AMP mutations do not significantly affect lifespan, with the possible exception of Def^SK3. Overall, we conclude that deleting single AMP genes has no effect on lifespan beyond levels of difference we also observed when comparing different wild types.

Drosophila possess many AMPs with possibly redundant or synergistic activities. Thus, AMPs could affect lifespan only when several genes are deleted simultaneously. After evaluating the effect on lifespan of individual AMP genes or genomic clusters of AMPs, we investigated whether combinatory loss of AMPs could impact aging. Using the same approach as described in Hanson et al. (2019a), we generated four groups of compound AMP mutants that remove different subsets of AMP gene families. Group A flies were deleted for Defensin and the four Cecropins. Group B flies were deleted for the structurally related Drosocin, Attacin and Diptericin families. Group C flies were deleted for the two antifungal Drosomycin and Metchnikowin peptide genes. Also, we introduce a new isogenic line combining the loss of two recently described Toll-regulated antifungal peptides: Group D flies that were deleted for both Baramicin A (Hanson et al., 2021) and the two Daisho genes (Cohen et al., 2020). We screened combined mutants lacking each of these AMP groups, including all the combinations of Groups A, B and C [i.e. AB, AC, BC and ABC (also known as ΔAMP14)]. We also screened flies lacking ten AMP genes, but which retain a wild-type Cecropin locus [ΔAMP10, as used previously (Carboni et al., 2021; Hanson et al., 2019a)]. In total, we screened nine AMP combinatory genotypes for lifespan and climbing effects (Fig. 3).

We found a significant effect of Group A mutations on male lifespan compared to that of iso w¹¹¹⁸, which is not unexpected given this AMP group uses the same second chromosome as the isogenic Def^SK3 mutation as above (Fig. 2B), except with the additional loss of Cecropins on the third chromosome. Comparisons between the two suggest a non-significant effect of Cecropin mutation (Def^SK3 versus Group A, P>0.10). Deleting other groups of AMPs typically resulted in non-significant effects on median lifespan compared to their iso w¹¹¹⁸ control (P>0.10). Group AB flies also had reduced lifespan compared to iso w¹¹¹⁸, but were comparable to Group A alone (Group A versus Group AB, P>0.10). Group AC and BC flies had comparable lifespans to those of iso w¹¹¹⁸ (P>0.10 in both cases), which is notable as Group AC flies have the Def^SK3 mutation, but only after an additional round of chromosome II recombination with the isogenic Mtk^R1 chromosome needed to generate the Group AC genotype. Deleting ten AMP genes, which included Defensin, also had a Group A-like lifespan (ΔAMP10 versus Group A, P>0.10). Thus, various combinations of mutants had lifespans comparable to those of either iso w¹¹¹⁸ or Def^SK3 alone. However, deletion of all 14 classic Drosophila AMP genes caused a more pronounced median lifespan reduction than the effect of Group A alone (ΔAMP14 versus Group A, male P=0.059, female P=0.002). There was minor variation in climbing competence amongst AMP mutant groups, with Group C flies (Mtk and Drs double mutants) having slightly improved climbing into old age. However, this effect was minor compared to the climbing competence of mth¹ flies into old age (Fig. 3D; File S1).

Although deletion of any single AMP gene had little impact on lifespan, deletion of all 14 classical AMP genes caused a reduction in lifespan that was significantly different from that of wild-type flies. Moreover, recombination of Def^SK3 with another mutation to produce Group AC flies yielded a lifespan comparable to that of iso w¹¹¹⁸. This result reinforces the need for careful interpretation of the Def^SK3 mutation effect of Group A. Despite this caveat, ΔAMP14 flies had significantly reduced lifespan compared to those of both wild-type flies and other AMP groups. The lifespan of ΔAMP14 flies was intermediate between those of iso w¹¹¹⁸ and Rel^E20, indicating Relish likely impacts fly lifespan through processes other than AMP regulation.

During our studies, we noticed that the food surface of the vials containing ΔAMP14 flies became discoloured and sticky upon aging, suggesting microbial proliferation (Fig. 3C, bottom). The Drosophila microbiome is found both in the gut and in the external environment due to constant ingestion and fecal deposition (Broderick et al., 2014; Pais et al., 2018; Storelli et al., 2018). We suspected this change in the fly food appearance for ΔAMP14 flies upon aging could be linked to a change in the microbiome, as we have recently described a role for AMPs in regulating Acetobacter using ΔAMP14 flies (Marra et al., 2021). Sticky and discoloured food was also observed for Rel^E20 flies, which do not express AMPs, and indeed both ΔAMP14 and Rel^E20 flies suffer increased dysbiosis over aging (Marra et al., 2021). To test if microbiome load was associated with early mortality of ΔAMP14 flies, we monitored bacterial abundance on the food medium. To do this, we emptied vials in which flies had been present for 2 days, added glass beads to these vials, shook the vials with beads for 10 s, then placed the beads on agar plates and spread vial microbes by rolling the beads over the agar for 10 s. By 40 dpe, plating of vial contents using beads confirmed far higher microbe loads in aged ΔAMP14 vials compared to those in iso w¹¹¹⁸ vials (Fig. 3C, top). Moreover, in the time period between 50-70 dpe, most major mortality events in ΔAMP14 vials were associated with sticky and discoloured bacterial food (example photo in Fig. 3C, bottom). These major mortality events are also seen as precipitous drops in ΔAMP14 survival curves shown in Fig. 3A (indicated with ‘{’), and similar trends were observed for ΔAMP10.

We conclude that ΔAMP14 flies thus suffered increased microbe load within vials with aging, agreeing with an increase in gut microbiome abundance with aging shown previously using gnotobiotic flies (Marra et al., 2021). This suggests that precocious aging observed in ΔAMP14 flies could be indirectly caused by the impact of AMPs on the microbiome.

To test whether the effect of AMP mutation on lifespan was linked to changes in the microbiome, we performed the same lifespan experiments in germ-free conditions using iso w¹¹¹⁸, ΔAMP14 and Rel^E20 flies. For this, we first bleached embryos, and then kept larvae and emerging adults on antibiotic (ABX) food media for their entire lifespan. In our hands, antibiotic-reared iso w¹¹¹⁸ flies had a similar lifespan compared to conventionally reared (CR) flies. Likewise, in our hands, Rel^E20 mutants had similar lifespan regardless of whether they were reared conventionally or in antibiotic conditions. However, rearing ΔAMP14 flies on antibiotics significantly rescued their lifespan (Fig. 4A,B; ΔAMP14 ABX versus CR, P=0.013), and a similar non-significant trend was seen in females (Fig. S7A,B). Climbing competence remained largely unchanged, although male ΔAMP14 flies showed improved climbing specifically at 50 dpe in germ-free conditions (P=0.053, Fig. 4C).

These results suggest that the short lifespan of ΔAMP14 flies primarily relies on the impact that AMPs have on the microbiome through aging. The observation that antibiotics treatment rescues ΔAMP14 but not Relish mutant lifespan suggests the short lifespan of Rel^E20 flies depends on effects other than the microbiome and independent of AMPs.

A number of studies done not only in Drosophila and in Caenorhabditis elegans, but also in vertebrates, have implicated AMPs in processes as diverse as behaviour, neurodegeneration, tumour clearance and aging. In mammals, some AMPs function as pro-inflammatory cytokines, and as such could influence these processes by disrupting homeostasis when chronically expressed (Liang and Diana, 2020; Mookherjee et al., 2020; Smith and Gwyer Findlay, 2022). However, to date there is no evidence of a cytokine role for AMPs in Drosophila (Hanson and Lemaitre, 2020). AMPs can also disrupt the membranes of aberrant host cells, which may become more negatively charged due to the exposure of phospholipids such as phosphatidylserine, a well-known ‘eat-me’ signal that promotes the phagocytosis of apoptotic cells (Fadok et al., 1992; Hakim-Mishnaevski et al., 2019; Manaka et al., 2004). Studies have now also implicated AMPs in the control of tumorous growth in vivo in Drosophila (Araki et al., 2019; Parvy et al., 2019), providing a proof of principle that AMP action can target self-tissue. Thus, AMPs could have a role in tissue maintenance with consequences on the aging process. Moreover, AMPs have a number of properties in common with neuropeptides, often being cationic and amphipathic (Augustin et al., 2017; Smith and Gwyer Findlay, 2022), alongside descriptions of AMP-like genes undergoing downregulation in the brain after exposure to pheromones (Gendron et al., 2014) or AMP-like genes modulating sleep (Sinner et al., 2021; Toda et al., 2019), memory (Barajas-azpeleta et al., 2018) and affecting behaviours (Ebrahim et al., 2021; Hanson et al., 2021; Kobler et al., 2020). Thus, it cannot be excluded that AMPs modulate neuronal activity, a process that could impact lifespan.

There was an increase of AMP gene expression observed upon aging, though it was unclear if AMPs directly affect lifespan, or if this activation was just a secondary consequence of aging. For instance, aging is accompanied by dysbiosis and gut barrier dysfunction allowing opportunistic systemic infection by microbiome bacteria (Buchon et al., 2009; Clark et al., 2015; Liu et al., 2022; Marra et al., 2021; Rera et al., 2012), which should activate immune signalling and induce AMP expression. Using flies carrying null mutations in AMP genes, we found no evidence that individual AMPs are so essential to host physiology that they have a notable impact on fly lifespan. Overall, the lifespan of flies mutated for individual genes or clusters of AMPs, Bomanins or Daishos did not differ from that of wild-type. The climbing activities of these mutants were also similar to those of wild-type flies. The only possible exception to this was the apparent lower lifespan of Def mutant males. However, conflicting results from RNAi experiments and use of Group AC flies with the Def^SK3 mutation after additional rounds of recombination suggest the somewhat shortened lifespan of Def mutant males was not linked to Def mutation itself but due to the presence of one or several cryptic mutations that were not removed during the isogenization process due to their proximity to the Def locus. Alternately, complex interactions amongst AMPs could provide a protective effect against Defensin mutation, as AMP interactions can synergize to prevent damage to host membranes (Drab and Sugihara, 2020). Overall, our study suggests that individual AMPs do not affect lifespan beyond levels of difference we also saw when comparing wild-type and/or classic aging mutant flies (e.g. mth¹), a result that contradicts other studies using RNAi or overexpression methods. However, our results do support a role for AMPs in regulating the microbiome over aging. Indeed, we recently showed that Acetobacter microbiome bacteria grow out of control in the microbiome of ΔAMP14 flies (Marra et al., 2021), and later confirmed that Diptericin B has a highly specific and important role in suppressing Acetobacter growth after systemic infection, which causes bloating similar to what we saw in Nora virus-infected flies (Hanson et al., 2022b). This phenotype of Acetobacter systemic infection could help explain why flies bloat upon infection by enteric pathogens (like Nora virus or Drosophila C virus) or upon aging, given the eventual invasion of gut microbes into the hemolymph (Clark et al., 2015; Rera et al., 2012).

We cannot exclude that additional AMP mutant combinations could reveal a stronger impact of AMPs on lifespan. Indeed, a number of immune-induced peptides that could be AMPs await description (Hanson and Lemaitre, 2020; Schlamp et al., 2021), and additional deletion of these peptides could explain the difference in lifespan between ΔAMP14 and Relish mutants. Use of other rearing conditions could also reveal a role for AMPs not found here. As part of this study, we also monitored AMP expression upon aging in four wild-type backgrounds (a lab standard w¹¹¹⁸, Exelexis, OR-R and Canton-S) by separating out dissected heads and bodies (thorax and abdomen). In our hands, we observed increased expression of AMP genes in fly bodies aged 40 days compared to expression seen in young flies, though the extent differed by genotype (Fig. S9; Table S2). However, we saw no marked AMP increase in the head with aging, including in a separate experiment following flies over multiple time points (Table S2). Indeed, when AMPs were upregulated in the head upon aging, this was often coupled with a stronger upregulation in the body, suggesting that this increase of AMP expression in the head could derive from the head-specific fat body responding like the fat body in the thorax and abdomen. We also tested if glia-specific Relish knockdown (via Repo-Gal4 and UAS-Rel-IR) could rescue lifespan, as many studies suggest a role of Relish in neurons/glia to rescue neurodegenerative syndromes (Cao et al., 2013; Kounatidis et al., 2017; Petersen et al., 2013; Shukla et al., 2019), but, if anything, we saw the opposite effect: female Repo>Rel-IR flies had shorter lifespan than genetic controls, and we found that Repo-Gal4 alone had improved climbing into old age, suggesting a genetic background effect on aging unrelated to RNAi using this driver (Fig. S8). Although Repo>Rel-IR can rescue disease phenotypes in models of neurodegeneration, taken together, our results suggest AMPs are not especially upregulated in the head upon aging (except when also upregulated in the body) and that glia-expressed Relish does not have a major deleterious role in lifespan using our standard rearing conditions. Differences in flies between research groups (local microbiome composition, including cryptic viral infections, food recipe, etc.) could account for these conflicting results. As we found by including methuselah mutant flies, improvements to healthy aging could also be consistent across research groups, but actual lifespan extension may be lab specific. These factors, and the importance of assaying multiple healthy aging metrics, should be considered when comparing our AMP mutant results to those from the larger field.

Importantly, our study suggests that AMPs collectively affect fly lifespan through their impact on the microbiome. Several studies have investigated the role of the gut microbiome on lifespan using germ-free conditions with mixed results (Brummel et al., 2004; Cai et al., 2021; Iatsenko et al., 2018; Ren et al., 2007; Shukla et al., 2021). It could be that discrepancies between studies reporting an impact of the microbiome on aging results from a complex interaction between nutrition and gut bacteria. In poor or unbalanced diets, the microbiome could have a positive impact on lifespan by extracting more nutrients from the food (Camus et al., 2019; Chaston et al., 2016; Consuegra et al., 2020; Erkosar et al., 2017; Newell and Douglas, 2014; Storelli et al., 2018; Téfit et al., 2018). In contrast, on a lab-standard nutrient-rich diet, as we have used in this study, the microbiome could have less impact. Consistent with this, we did not see major differences between germ-free and conventionally raised wild-type flies in this study. Previous studies have already revealed the role of Imd signalling in controlling microbiome load and diversity, preventing dysbiosis (Guo et al., 2014; Li et al., 2016; Liu et al., 2022; Yamashita et al., 2021). The specific microbiome of flies from a given research group could also change the impact of germ-free conditions. For instance, different Acetobacter strains have different virulence to the fly during systemic infection (Hanson et al., 2022b), which accompanies aging and intestinal barrier dysfunction (Clark et al., 2015; Rera et al., 2012). Indeed, use of ΔAMP14 flies revealed a key role of AMPs to control Acetobacter levels in the microbiome and, accordingly, ΔAMP14 flies have increased Acetobacter loads upon aging (Marra et al., 2021).

Our present study shows that the action of AMPs preserves lifespan, and that this effect is largely due to their impact on the microbiome. Thus, the impact of AMPs on lifespan is consistent with their well-established microbicidal activity. Interestingly, raising Relish mutant flies in axenic conditions did not lead to lifespan extension, indicating that Relish, and likely the Imd pathway, have a much more profound impact on host physiology independent of AMP regulation. In line with this, Relish and the Imd pathway have been implicated in neural systems (Kounatidis et al., 2017; Masuzzo et al., 2019; Shukla et al., 2019), cell competition (Meyer et al., 2014; Nandy et al., 2018), metabolism (Molaei et al., 2019; Musselman et al., 2018) and enterocyte delamination (Liu et al., 2022; Zhai et al., 2018), processes that likely impact lifespan. It is interesting to note that the food media of ΔAMP14 and Rel^E20 flies were enriched in bacteria, agreeing with elevated bacterial loads in these flies with aging as shown previously (Marra et al., 2021). External microbes colonize the Drosophila gut, and gut microbes are released into the external environment as part of excreta (Broderick et al., 2014; Newell and Douglas, 2014; Storelli et al., 2018; Winans et al., 2017). Thus, it was expected that an increased gut microbiome load in the absence of AMPs would result in high bacterial load in the fly food medium. Although we reared flies in a vial in artificial lab conditions here, it is tempting to speculate that AMPs expressed in the gut could not only shape the gut microbiome but also environmental bacteria. The increased load in the food medium could therefore rely either on AMP-mediated control of the gut microbiome or on external AMPs secreted into the food medium. This would suggest that AMP mutation can exacerbate microbiome effects from both within the fly and in the external vial environment.

During our study, we experienced a number of challenges to lifespan data interpretation. Notably, our reference wild-type and several other fly strains were infected with Drosophila Nora virus, which had a greater deleterious effect on lifespan of iso w¹¹¹⁸ compared to wild-type flies used in a previous study (Habayeb et al., 2009). Thus, we initially interpreted our early results as if AMP deletion extended lifespan to a great extent compared to that of the isogenic wild-type controls (Fig. 1A). However, our ultimate findings, including standard genotypes and different conditions, instead highlight that loss of AMPs does not extend lifespan; if anything, they show the opposite. We also note that cryptic and chronic infections common in fly stocks, such as Nora virus, represent a serious threat to aging studies. In our study, we realized this cryptic viral infection confounded our results when comparing the lifespan of our isogenic wild-type flies to the expected absolute lifespan of Drosophila according to previous recommendations (Piper and Partridge, 2018). We publish this experience, which confused years of data collection, as a cautionary note for the field of aging and immunity. Our hope is that our experience can help others avoid similar confounding factors.

In conclusion, our study reveals a key role of AMPs in the aging process, but mainly through their indirect effect on the microbiome. We cannot exclude that certain contexts could reveal an intrinsic effect of AMPs on host tissues during aging, such as conditions found in individuals that have aging-associated diseases like cancer or precocious neurodegeneration, uncommon in standard wild-types. However, here we did not find evidence of AMPs directly impacting aging in a striking way. We are still far from understanding the complex relationship between the immune system, senescence, and aging, which requires further investigation.

Drosophila stocks used in this study, including genotype descriptions, are listed in Table S3. Food media used the following recipe (per 600 ml): 3.72 g agar, 35.28 g cornmeal, 35.28 g yeast extract, 36 ml grape juice, 2.9 ml propionic acid and 15.9 ml Moldex. The antibiotic medium also contained final concentrations of 50 μg/ml ampicillin, 50 μg/ml kanamycin, 10 μg/ml tetracycline and 10 μg/ml erythromycin. Flies were flipped three times per week (Monday, Wednesday, Friday), and vials were left on their side to ameliorate the effect of the food medium stickiness on mortality with aging by allowing flies falling to the ground to drop onto plastic rather than the food surface. This precaution was taken as vial conditions differ markedly in specific immune-deficient genotypes (Marra et al., 2021).

To clear flies of Nora virus, embryos were collected from grape juice agar plates, rinsed with distilled water and left to soak in 3% bleach for 3 min. Embryos were then rinsed twice in distilled water for 1 min each. This protocol was also used to clear the microbiome of antibiotic-reared flies, whereafter we placed embryos directly on antibiotic medium for germ-free experiments.

Lifespan experiments were conducted from 2017 to 2022. Flies were allowed to emerge and mate randomly for ∼3 days prior to separating males and females. Then, groups of 20 males or 20 females (mated) were flipped three times per week (Monday, Wednesday, Friday) to measure fly lifespan in 90×15 mm polystyrene vials.

We used a Cox proportional hazard (CoxPH) mixed-effects model to initially analyse lifespan effects, with experimental replicate and biological sex as interaction terms in R version 3.6.3. In both cases, experimental replicate and sex were significant contributors to the model (P<0.001). Our impression from the initial data analysis was that the CoxPH model was overly sensitive to minor variation around the geometric mean lifespan of iso w¹¹¹⁸ control flies, exacerbated by the large sample sizes used and the many comparisons performed in our study inflating the likelihood of type I (false-positive) statistical errors. Even if minor differences in lifespan were genuine, their ultimate importance was questionable when compared to other genetic backgrounds (e.g. OR-R, mth¹), particularly given variable mutation types and transgene insertions used for different AMP mutations (i.e. point mutation, genomic deficiency, white⁺, 3×P3-EGFP or 3×P3-dsRed). We thus preferred to focus on each experiment as if that population of flies represented a single sampling. We therefore treated our data as if we had, for example, n=3 per genotype (three experiments), rather than n=120 (60 males and 60 females) per genotype across three experiments. For this purpose, we decided to use median lifespans as our primary readout. Sex-specific median lifespans were analysed using one-way ANOVA with Holm–Sidak's multiple test correction implemented in Prism v9.3.1, or one-way ANOVA with Tukey's honestly significant difference (HSD) correction in R v3.6.3.

We paired our lifespan data with the gravitaxic locomotor climbing assay to provide an independent metric of aging (Madabattula et al., 2015). This assay assesses the general locomotor competence of the fly, which is often used as a readout of neurodegeneration, but can also reflect generic aging effects (e.g. muscle weakness). Climbing pass rates were filmed for flies at 5, 40, 50 and 60 dpe ±2 days, and analysed later manually. A pass was considered if a fly climbed 5 cm within 10 s of being tapped to the bottom of the vial. Flies were transferred to a chamber made of two empty vials stacked atop each other to provide ample room to climb upwards without reaching the ceiling or disrupting other flies. Two sets of broken markings were made with permanent marker at 2.5 cm and 5 cm on the lower vial as reference points, though we ultimately report only 5 cm climbing rates given similar trends between the two. At each time point, an initial tapping down was performed to associate the flies with their new environment and encourage climbing behaviour, as we found the first repetition of this experiment often had fewer climbers than subsequent repetitions. After this initial association, three technical repeats of the climbing assay were performed, and final values represent the average of these three technical repeats. All climbing measurements were taken between 2 pm and 5 pm to ensure a consistent measurement timeframe.

This assay can be a measure of both climbing speed and/or readiness of response: for instance, older flies sometimes suffered temporary seizures after being tapped down, or might climb slower, more erratically, or simply to a lesser extent than younger counterparts. In most cases, climbing pass data mirrored lifespan data in terms of relative trends. We thus have no suspicion that AMP mutations, individually or collectively, greatly affected locomotory competence with aging. The only noteworthy exception to mirrored trends between lifespan and climbing was methuselah mutant flies, which had wild-type-like lifespans, but improved climbing competence into older age at both 25°C and 29°C.

For analysis of trends of ΔAMP14 in Fig. 4C, we fit a two-way ANOVA measuring the climbing at 40, 50 and 60 days, with an interaction term for age and germ-free conditions in Prism v9.4.1.

We monitored the microbiome of compound AMP mutants and antibiotic-reared flies by checking plated vial food contents on MRS+mannitol agar, a medium amenable to both Lactobacilli and Acetobacter. Specifically, at both ∼20 and ∼40 days, we took vials in which flies had been present for 2 days, added five glass beads to the vials, shook the glass beads in vials for 10 s, then transferred glass beads to MRS+mannitol agar plates and shook the plates for 10 s, then removed the beads and left plated microbes to grow overnight at 29°C. Vial microbiome loads were checked on Wednesdays. We chose Wednesdays rather than a precise age (e.g. exactly 20 dpe) to ensure that flies entered the plating time point after experiencing a similar regimen of flipping; i.e. in our experiments, we only plated microbes from vials in which flies had spent the past five days with ∼3 days in a vial accruing microbes over the weekend (Friday to Monday), and the next two days depositing microbes in the vial that was ultimately measured (Monday to Wednesday). This design was chosen based on demonstrations that flipping regimen drastically affects microbiome load (Arias-Rojas and Iatsenko, 2022; Blum et al., 2013; Pais et al., 2018). Beyond ∼40 days, comparisons were not equal due to onset of mortality in AMP mutants and associated drops in vial fly density.

In experiments using this method with antibiotic medium, microbes were never detected from overnight growth to monitor vial microbiome loads.

Gene expression was performed using primers listed in Table S3 with PowerUP SYBR Green Master Mix (Applied Biosystems), using the Pfaffl method of quantitative PCR (qPCR) quantification with Rp49 as the reference gene (Pfaffl, 2001). RNA was extracted using TRIzol (Invitrogen) according to manufacturer's protocol. cDNA was reverse transcribed using Takara Reverse Transcriptase.

Dissections of heads from bodies were performed in ice-cold PBS, and tubes containing pools of 20 heads or bodies were kept at −20°C until after TRIzol was added to prevent RNA degradation before sample processing.

We would like to thank Samuel Rommelaere, Jean-Philippe Boquete, Emi Nagoshi, Lukas Neukomm, Kausik Si and Anzer Khan for helpful discussion. We would also like to thank Brian McCabe, Mariann Bienz, Barry Ganetzky, Steven Wasserman, Lianne Cohen, the Vienna Drosophila Resource centre and the Bloomington Drosophila Stock Centre for fly stocks requested over the course of this research.