Evolution of increased immune defence is often limited by costs: correlated changes in other traits (viz. life-history traits) that otherwise reduce the fitness of the host organisms. Experimental evolution studies are useful for understanding the evolution of immune function, and correlated changes in other traits. We experimentally evolved replicate Drosophila melanogaster populations to better survive infection challenge with an entomopathogenic bacteria, Enterococcus faecalis. Within 35 generations of directional selection, selected populations showed marked increase in post-infection survival compared to ancestrally paired controls. We next measured various life-history traits of these populations. Our results show that the selected populations do not differ from control populations for larval development time and body weight at eclosion. No difference is also observed in case of fecundity and longevity (following the acute phase of infection), either when the flies were subjected to infection or when the flies were uninfected; although infected flies from all populations die much earlier compared to uninfected flies. Selected populations are either equally good, or occasionally better, as the control populations at surviving abiotic stressors (starvation and desiccation), although infected flies from all populations are more susceptible to stress than uninfected flies. Therefore, we conclude that (a) D. melanogaster populations can rapidly evolve to be more immune to infection with E. faecalis; (b) evolution of increased defence against E. faecalis entails no life-history cost for the hosts; and (c) evolving defence against a biotic threat (pathogen) does not make flies more susceptible to abiotic stressors.

Life-history traits, and resistance to biotic and abiotic stressors, were measured for Drosophila melanogaster flies selected for improved post-infection survival against systemic infection with an entomopathogenic, Gram-positive bacteria, Enterococcus faecalis. The populations were named the EPN populations.

1. EPN populations: history and maintenance regime

The EPN populations was derived from ancestral Blue Ridge Baseline (BRB) populations (Singh et al., 2021). The BRB populations consist of 4 replicate outbred populations, originally established by hybridizing iso-female lines which had themselves been established by wild-caught flies, from Blue Ridge, USA, in the laboratory of Prof. Daniel Promislow (Singh et al., 2015). These populations are maintained in a 14-day discrete generation cycle, with a census size of about 2800 adults in every generation. These populations are maintained under standard lab conditions (12:12 light: dark cycle, 25 °C, 60% relative humidity) on banana-jaggery-yeast food medium. Full details of maintenance of BRB populations are described in Singh et al., 2015.

From each replicate population of BRB_1-4 three populations were derived: (i) E_1-4, infected with Enterococcus faecalis; (ii) P_1-4, procedural control; and (iii) N_1-4, uninfected control. Populations having same numeral subscript shared a common recent ancestry and were treated as independent blocks (block 1, block 2 and so forth). Individual blocks were always handled together during selection and during experiments. Eggs were collected at a density of ~70 eggs per vial (25mm diameter × 90 mm height) containing 6-8 ml of standard banana-jaggery-yeast food. Ten 10 such vials were collected for each population. These vials were reared under standard laboratory (12:12 light: dark cycle, 25 °C, 60% relative humidity) conditions until 12^th day post egg collection. By 10^th day all flies eclose and had mated at least once by 12^th day (day of infection). Further handling depended on the type of population.

For E populations, on 12^th day post egg collection, every generation 200 females and 200 males were randomly picked out of total 700 flies. These flies were pricked on the dorsolateral surface of the thorax with Minutien pin (0.1mm Fine Scientific Tools, USA) dipped in bacterial suspension under light CO₂ anaesthesia (refer section 2.2 Bacterial culture, for details). After infection flies were placed inside plexiglass cage (14 cm length x 16 cm width x 13 cm height) with food in 60 mm Petri plate. About fifty percent of the infected flies die within 96 hours of infection (mortality ranged from minimum 40% to maximum 55% across different E populations across generations). Post 96 hours, fly cages were provided with oviposition food plates for 18 hours. Eggs were collected from these oviposition plates at a density of ~70 eggs per vial (as mentioned above) to start next generation. This 18-hour window, 96 hours after infection, serves as the selection window for the EPN populations. The eggs produced by the surviving flies from each population during this window is used to set up the next generation.

Similarly, for P populations, 100 females and 100 males out of total 700 flies are pricked every generation with Minutien pin dipped in sterile 10 mM MgSO₄ under light CO₂ anaesthesia. For N populations, 100 females and 100 males out of total 700 flies are sorted every generation under light CO₂ anaesthesia and transferred to the plexiglass cage. Aside from the infection treatment differences, E, P, and N populations were treated the same. Post 96 hours, eggs are collected in similar way as for E populations to start the next generation.

The E populations are initiated every generation with double the number of adults compared to P or N populations to account for the fact that about 50% of the E populations perish due to infection every generation. There is negligible mortality in P (< 2%) and N (< 1%) populations. This ensured that the census sizes of the populations remain similar on average across generations.

2. Bacterial culture

The pathogen used in this study was the Gram-positive bacterium Enterococcus faecalis (Ef, Gram-positive, grown at 37 °C, Lazzaro et al 2006). E. faecalis is an entomopathogen commonly used in eco-immunological studies in Drosophila melanogaster (Troha and Buchon 2019) and is known to establish persistent infection in flies (Chambers et al., 2019). Natural populations are known to harbour genetic variation for resistance against E. faecalis (Lazzaro et al., 2006, Chapman et al., 2020), and its virulence mechanisms and the host responses it elicits have been described in some detail (Gobert et al., 2003, Shirasu-Hiza and Schneider 2007, Ayres and Schneider 2008, Dionne and Schneider 2008, Nehme et al., 2011, Hanson et al., 2019). E. faecalis is also a common member of natural microflora of various insects (Cox and Gilmore 2007), and the pathogen isolate used in this study was originally derived from a wild-caught D. melanogaster (Lazzaro et al., 2006).

The bacterial stocks are maintained as 17% glycerol stocks frozen at -80 °C. An overnight primary culture of bacteria was set by inoculating a stab of bacterial glycerol stock in 10 ml lysogeny broth (Luria-Bertani-Miller, HiMedia) and incubating it at appropriate temperature with continuous mixing at 150 RPM (revolution per minute). Once this primary culture turned confluent (OD₆₀₀ = 1.0-1.2), a secondary culture was set up (100 microliters of primary culture in 10 ml fresh lysogeny broth) and maintained at appropriate temperature with continuous mixing until it turned confluent again. (The OD₆₀₀ for the primary and secondary cultures were monitored during their growth to ensure that the cultures were not overgrown.) This secondary culture was centrifuged, and bacterial pellets were resuspended in sterile 10 mM MgSO₄ buffer to obtain desired OD₆₀₀ for infection. This bacterial suspension was used to infect flies. Infection was done by dipping needle in the bacterial suspension or sterile 10mM MgSO₄ buffer and pricking flies on the thorax. The bacterial suspension was vortexed both prior to and during infection at regular intervals to prevent bacterial cells from settling.

For stock maintenance, E populations of EPN regime were infected with E. faecalis. Throughout the selection history of EPN, the pathogen infection dose was modulated to induce fifty percent mortality in E populations. This ensured a constant, directional selection process. The dose for fifty-percent mortality was originally determined for the ancestral BRB populations by infecting them with different doses, where we found that OD₆₀₀ = 0.8 reproducibly led to 50% (±5%) mortality. Hence, E populations were initiated by infecting the flies at this dose. During regular maintenance, the exact mortality in E populations were monitored every 3-4 generations. By generation 20, the mortality in these populations had come down to 40%, and hence the infection dose was increased to OD₆₀₀ = 1.0 from generations 21 onwards.

3. Fly standardization

Experimental eggs were collected from flies that were cultured under common environmental conditions for one generation. This was done to account for any non-genetic parental effects (Rose 1984) and flies thus generated were called standardized flies.

Experimental eggs were collected from all three populations (E, P, and N) at a density of ~70 eggs per vial and 10 such vials were established per population. The eggs were reared in the same development vials into adults, until day 12 PEL. These adults were then transferred to plexiglass cages (14x16x13 cm³) with food in Petri plates (60 mm diameter). Experimental eggs were collected from these population cages.

4. Post-infection survival and longevity

Standardized fly cages for each population (E_i, P_i, and N_i, where ‘i’ represents blocks 1-4) were provided with ad libitum yeast paste smeared on the top of the standard banana-jaggery-yeast food plate, three days prior to the egg collection. After two days, these yeasted food plates were replaced with oviposition plates for 18 hours. Eggs were collected from these oviposition plates at the density of approximately 70 eggs per vial, into 25 vials per population (E_i, P_i, and N_i), with each vial containing 8 ml of banana-jaggery-yeast food. These rearing vials were incubated at standard maintenance conditions for next 12 days. Flies generally eclose by 10^th-11^th day post-egg-laying (PEL) in the rearing vials, and mate at least once by 12^th day PEL.

On 12^th day PEL, flies from each of the E_i, P_i, and N_i populations were randomly assigned to one of the three treatments:

(a)   infected treatment: 200 females and 200 males were randomly sampled from each population and were infected with E. faecalis at OD₆₀₀ = 0.8 under light CO₂ anaesthesia;

(b)  sham-infected treatment: 100 females and 100 males were randomly sampled from each population, and sham-infected with sterile needle dipped in sterile 10 mM MgSO₄ solution; and,

(c)   uninfected treatment: 100 females and 100 males were just randomly sampled under light CO₂ anaesthesia, and not subjected to any further manipulation.

After being subjected to different treatments, the flies were housed in plexiglass cages (14 cm × 16 cm × 13 cm) having ad libitum access to banana-jaggery-yeast food provided in Petri plates (60 mm diameter). Individual blocks were experimented upon on separate days.

Survivorship of the flies was monitored every 4-6 hours for the first 96 hours after infection, and after this period, mortality in the cages was recorded once a day until the last fly died in all cages. About fifty percent of the infected fly and almost all sham-infected and uninfected flies survived post 96-hours window. Fresh food plates were provided on alternate days. Altogether, 200 flies/sex/population/block were infected with E. faecalis, 100 flies/sex/population/block were sham-infected, and 100 flies/sex/population/block were maintained as uninfected controls.

5. Fecundity and hatchability

We measured the number of eggs produced by the females (fecundity) from the EPN populations, and what proportion of these eggs produced viable larva (hatchability), to test for the effect of selection history and infection status on fecundity and hatchability. The same experimental cages that were used for assaying post-infection survival was used here for measuring fecundity and hatchability.

Fecundity was assayed after 96 hours of infection treatment (identical to the time when eggs are collected for the next generation during maintenance of the selection regime). Each of the population cages (E_i, P_i, and N_i flies either infected, sham-infected, or uninfected) were provided with an oviposition plate (60 mm diameter), containing standard banana-jaggery-yeast food, for the female flies to lay eggs on for 18 hours. After 18 hours, these plates were withdrawn, labelled with the cage identity, and stored at -20˚C for eggs to be counted later. Eggs on the surface of the food plates were counted visually using a light stereo microscope (Zeiss Stemi 2000) under 2.5X × 10X magnification. Per-female fecundity was calculated by dividing the number of eggs laid by the females during 18-hour window by the number of females alive in the respective cages at the start of the oviposition period.

Hatchability was assayed after 114 hours of infection. Following withdrawal of the oviposition plates, each of the above cages were provided with a fresh food plate (60 mm diameter) for 8 hours. From each of these plates (each coming from a single cage), three samples of 100 eggs each was picked using a moist paint brush and arranged onto the surface of three separate agar plates (90 mm Petri plates, 1.5% agar). These plates were incubated under standard laboratory maintenance conditions, and 48 hours later, the number of eggs on each agar plate that had hatched were visually counted using a light stereo microscope (Zeiss Stemi 2000) under 2.5X × 10X magnification. Hatchability was determined for each agar plate by dividing the number of eggs that hatched by the total number of eggs that were placed on the surface of that plate and was used as unit of replication. Altogether from each cage 3 × 100 eggs/treatment/population/block were scanned for hatchability.

6. Egg-to-adult development time and viability, and dry body weight

Fresh food plates with excess live yeast paste were provided to the standardised population cages (E_i, P_i, an N_i) for 48 hours. Following this, a fresh food plate with live yeast paste (yeast paste placed at the centre of the plate and with space around the circumference for females to lay eggs) was provided to each cage for 6 hours. This plate was followed up with two more similarly yeasted food plates, each for a 1-hour window. This was done to encourage the females to lay any stored eggs. After this a fresh food plate was provided to each cage for females to lay eggs on for an hour, and eggs were collected using a light stereo microscope (Zeiss Stemi 2000) under 2.5X × 10X magnification and distributed into food vials (with 8 ml of banana-jaggery-yeast food) at an exact density of 70 eggs per vial. 10 vials were set up for each population (E_i, P_i, an N_i), and blocks were handled on separate days. These vials were incubated under standard maintenance conditions, and when flies started eclosing, freshly eclosed flies were transferred to empty glass vials every 4 hours, and frozen at -20 °C for further processing; this was done till the very last pupae had eclosed. The storage vials were labelled so as to preserve parent vial, population, and block identities. The flies eclosed at each time window from each vial was later scored visually using a light stereo microscope (Zeiss Stemi 2000) under 2.5X × 10X magnification to enumerate the total number and sex of the flies eclosed in that time window. Therefore, for each vial, the data was available for the number of males and females that eclosed at each time window, and the total number of flies that eclosed out of the vial. The median development time was calculated as the time taken by half of flies of each sex to eclose (starting for the time of oviposition), and viability was calculated by dividing the total number of flies eclosed by the number of eggs seeded in the vial (which was 70 eggs). After enumeration, the flies were put back into -20 ˚C storage for further use.

Flies preserved from the development time assay were used for measuring dry body weight at eclosion. All flies eclosing out of a single parent vial were pooled together; hence there were 10 pools of flies per population per block. From each pool, 5 females and 5 males were randomly sampled and placed in 1.5 ml micro-centrifuge tubes (MCTs); females and males were placed in separate MCTs. Therefore, each vial used in the development time assay yielded one MCT with 5 males and one MCT with 5 females. These MCTs were dry heated in a hot air oven for 48 hours at 60 ˚C to eliminate all moisture. The flies were then weighed using Sartorius weighing balance (model CPA225D, least count 0.01mg).  Individual blocks were handled on separate days. Dry body weight was measured for only 3 blocks because samples of block 2 was lost in handling.

7. Starvation resistance 

Eggs were collected from standardised population cages at an approximate density of 70 eggs per vial. 20 such vials were collected for each population (E_i, P_i, an N_i), and reared under standard maintenance conditions. Adults were housed in the rearing vials until 12^th day PEL. By this time all flies were sexually mature and have mated at least once inside the rearing vials itself. On 12^th day PEL, flies from each population (E_i, P_i, an N_i) were randomly assigned to three treatments:

(a)   infected treatment: 50 females and 50 males were randomly sampled from each population, and were infected with E. faecalis at OD₆₀₀ = 0.8 under light CO₂ anaesthesia;

(b)  sham-infected treatment: 50 females and 50 males were randomly sampled from each population, and sham-infected with sterile needle dipped in sterile 10 mM MgSO₄ solution; and,

(c)   uninfected treatment: 50 females and 50 males were just randomly sampled under light CO₂ anaesthesia, and not subjected to any further manipulation.

After being subjected to treatments, the flies were housed in vials containing 2 ml 1.5% non-nutritive agar gel, at a density of 10 females (or, males) per vial. The sexes were housed separately. The presence of agar gel in the vials ensured that the flies had ad libitum access to water during the course of the starvation assay. Individual blocks were assayed upon on separate days. In total, 5 vials/sex/treatment/population/block were set up for this assay. The vials were monitored every 6-8 hours to record the number of dead flies, until the last fly perished. Surviving flies were transferred to fresh agar vials every 48 hours.

8. Desiccation resistance

A set-up identical to the starvation resistance assay was utilised to test for desiccation resistance; the only difference was that the flies, after being subjected to their respective treatments, were housed in empty vials (no food or agar gel). Additionally, 5 gm silica beads were placed in each vial (above the cotton plug; no direct contact between flies and silica), and the mouth of the vial was sealed off with parafilm tape, to eliminate moisture from the vials. Individual blocks were assayed upon on separate days. In total, 5 vials/sex/treatment/population/block were set up for this assay. The vials were monitored every 1.5 hours to record the number of dead flies, until the very last fly perished.

9. Statistical analysis

All analysis was performed using R statistical software, version 4.1.0 (R Core Team 2021). For survival analysis, mixed-effect Cox-proportional hazard models were fitted to the survival data (from post-infection survival, longevity, starvation resistance, and desiccation resistance assays) using the coxme function of the “coxme” package (Therneau 2020), and the confidence intervals for these models were calculated using confint function of the base R package. These models were subjected to Analysis of Deviance using Anova function from the “car” package. Survival curves were plotted using the ggsurvplot function of the “survminer” package (Kassambara et al 2021) after modelling the data using survfit function from the “survival” package (Therneau 2021).

Post-infection survival assay (first 96 hours following infection): We first modelled survival of all flies with a Cox model with infection treatment (uninfected, sham-infected, infected), selection history (populations N, P, and E), sex (females and males), and their pair-wise interactions as fixed factor and block as a random factor. Thereafter, we stratified the data by infection treatment, fitted a model with selection history, sex, and their interaction as fixed factor, and block as a random factor. Significance testing of all models were done using Analysis of Deviance.

Longevity assay (survival from 96 hours post-infection and onwards), starvation resistance assay, and desiccation resistance assay: We first modelled survival of all flies with a Cox model with infection treatment (uninfected, sham-infected, infected), selection history (populations N, P, and E), sex (females and males), and their pair-wise interactions as fixed factor and block as a random factor. Thereafter, we stratified the data by sex, fitted a model with selection history, infection treatment, and their interaction, and block as a random factor. Further stratifications were done wherever any interaction term had a significant effect. Significance testing of all models were done using Analysis of Deviance. 

Data from life-history traits were modelled using mixed-effects general linear models (lmer function from “lmerTest” package; Kuznetsova et al 2017) and then subjected to type-III Analysis of Variance (ANOVA; anova function from base R package) for significance tests. Pairwise comparisons wherever necessary was done using Tukey’s HSD (lsmeans function from “emmeans” package; Lenth 2021). The mixed-effects general linear models used were:

Fecundity ~ selection history + infection treatment + selection history:infection treatment + (1|block),

Hatchability ~ selection history + infection treatment + selection history:infection treatment + (1|block),

Development time ~ selection history + sex + selection history:sex + (1|block),

Viability ~ selection history + (1|block), 

Body weight ~ selection history + sex + selection history:sex + (1|block).