The ability to predict when an individual will die can be extremely useful for many research problems in aging. A technique for predicting death in the model organism, Drosophila melanogaster , has been proposed which relies on an increase in the permeability of the fly intestinal system, allowing dyes from the diet to permeate the body of the fly shortly before death. In this study we sought to verify this claim in a large cohort study using different populations of D. melanogaster and different dyes. We found that only about 50% of the individuals showed a visible distribution of dye before death. This number did not vary substantially with the dye used. Most flies that did turn a blue color before death did so within 24 hours of death. There was also a measurable effect of the dye on the fly mean longevity. These results would tend to limit the utility of this method depending on the application the method was intended for.

Populations

Five large independent populations of Drosophila melanogaster were used in this experiment. Two of these populations, ACO and CO, are large, outbred populations that have been maintained on different age-at-reproduction schedules for hundreds of generations. The ACO population was maintained on 9 day discrete generation cycles. The CO population was kept on 28 day discrete generation cycles. The remaining populations, S93, A4 3852 and Canton S (CAS), were inbred lines raised on three week cycles in the Long lab at the University of California, Irvine. All populations were raised in identical conditions of temperature, food, cultures and density for three generations prior to these experiments . 

Mortality Assay

Adult, 14 day old (from egg) flies were knocked out with CO₂ gas and placed into individual plastic straws about 4 inches in length and capped with plastic pipette tips on both ends (Fig. 1). During anesthetization, a steady supply of CO₂was flowing through a semi-porous plate. The flies were placed on the plate and separated by gender and each fly was gently swept into the plastic straw using a fine painters brush.  An equal number of females and males were used per population. Food was provided to each fly at one end of the straw. Each fly was transferred to a new straw with new food and new pipette tips every 3 days to maintain a clean environment. The straw length and girth permitted individuals to fly from one end to the other. 

The process of transferring the flies, as well as daily checking of the flies, required a light tapping of the fly into the pipette tip. Cohorts of about 56 adult flies, equal numbers of males and females from each of the five populations were exposed to either control food or food with one of six dyes (table 1) added to their food. Substantial replication was used. Thus, the original dye, SPS Alfachem, was replicated in 5 different populations, and each population was replicated in 6 different dye environments. The use of different FDA FD&C Blue dye #1’s permitted us to determine if the development of the SMURF phenotype was sensitive to the particular dye used. By using a combination of different populations of D. melanogaster,which varied in levels of inbreeding, we could determine if the development of the SMURF phenotype was limited to inbred populations.  

The flies were exposed to the blue dyes from day 14 (from egg) continuously to their death. Each fly was individually checked underneath a microscope and light to see if it had become a ‘smurf’. Smurf status required that the entire body changed to any variation of a blue color. This was an important distinction as all the Drosophilaflies fed food with a blue dye would have visible blue coloring in only the gut portion when they weren’t a Smurf.  Some of the dyes resulted in a slight variation in blue color in the Smurfs. Every day under a microscope with a light we looked for any change of color in the fly thorax, head and abdomen. If the fly was any shade of blue in all three sections, it was marked as a Smurf and was then checked daily to see when it died. We did not limit our observations to individual sections of the fly, such as only the thorax, for our evaluation of when a fly became a Smurf.

Tapping

We did the tapping experiment to see if the physical disruption, the process of tapping the fly into the pipette tip, affected the mean longevity and lifespan of the fly. A total of 164 ACO flies were chosen for this assay – 83 males and 81 females. The 164 flies were placed into regular food straws with no dye. A total of 84 flies (42 male and 42 female) were tapped 5 times daily, mimicking the checking that occurred in the original experiment, and the other 81 (41 males and 39 females) flies were not tapped. The flies were transferred to new straws, with fresh food and new pipette tips every 3 days. Each fly was checked daily for movement and if no movement was detected, the fly was classified as deceased on that day. Only ACO flies were used as the purpose of the Tapping experiment was to see if our methods for checking for Smurf flies would affect the mean longevity of the fly. 

Food & Dyes

Flies were provided with banana-molasses food with one of the dyes added. The control flies received only banana molasses food in their respective straws. The recipe for the banana molasses food used in the lab, as well as the experiment, can be found in the Supplemental Portion. Food with dye was prepared by mixing 2.5 grams of each dye to create a 100 ml solution of the banana molasses food mixed with the dye (2.5% wt/vol). Food was always prepared the day before it was needed and stored in a refrigerator until it was used. The dyes were kept separate and carefully handled so no cross-contamination occurred during the preparation and food blending process. 

Statistical Analysis 

To analyze the effects of dye, sex and population on longevity we let y_ijklbe the age at death of the lth individual of sex-i(i=1 (female), 2 (male)), treatment-j (j=1,,..,7 (see table 1 of paper, 7=control)), and population-k(k=1,..,5 (see table 1 of paper)), whered_s=0 if s=1, and 1 otherwise, e_ijklis an error term assumed to have normal distribution with mean 0 and variance s². An initial test showed no significant differences between sexes so the final model tested did not include the bparameter. These tests were run with R (version 3.4.3, R Core team, 2017) and the lmfunction. Pairwise tests with Bonferroni corrections for simultaneous tests were conducted with the R emmeansfunction. 

            At the time of death each fly was classified according to their sex, population, treatment, and Smurf status (blue: yes or no). Using hierarchical log-linear models (loglmfunction in the R MASS package) we tested in succession whether sex, treatment, and population would have an effect on Smurf status at the time of death. 

A t-test was performed on the Tapping Experiment results, comparing the mean longevity of the tapped flies versus the non-tapped flies to see if the mechanical disruption would affect their mean longevity.