Comments from curator
Phenotyping data comes from the
Supporting Information of the paper (
journal page, also on
PMC):
-
S1 Table (
pone.0170332.s003.txt): mean bacterial load in males of 36 DGRP lines, for each of four commensal bacteria.
-
S2 Table (
pone.0170332.s004.txt): ROS level in the male gut, for 9 of these lines.
Only these per-line means were published (no per-pool or per-replicate data), so the study has a summary dataset only. The lines are given as bare numbers (e.g. 208 = DGRP_208): 36 of the 40 core DGRP lines (all but DGRP_514, DGRP_705, DGRP_714 and DGRP_859). All values are from males (sex M).
Bacterial load (gnotobiotic flies): all lines were first cleared of
Wolbachia (0.03% tetracycline for seven generations, whatever their infection status), then their microbiota was restored with carcasses of untreated flies of the same line for at least four generations. Axenic flies (eggs dechorionated with bleach, reared on autoclaved food) were fed for one day on food inoculated with about 4,000 CFU of a single strain, then moved to sterile food to lay eggs: their progeny, colonised from egg to adult by this strain only, were measured. Strains:
Acetobacter tropicalis,
Lactobacillus brevis and
L. plantarum (isolated from the gut of laboratory flies) and
Enterococcus faecalis (isolated from the hemolymph of wild-caught flies). Males aged 3-5 days (25 °C, 12:12 h light:dark, Bloomington medium) were moved at dawn to sterile food, collected 6-11 h later, surface-washed in 70% ethanol and frozen. For each line and bacterium, three pools of 10 males from at least two experimental blocks were analysed (at least two independent gnotobiotic replicates). The relative load is the qPCR Ct of the single-copy fly gene
Dfd minus the Ct of a bacterium-specific gene (means of three technical replicates): a log2 scale, where higher values mean more bacterial DNA per fly DNA. The primers differ in efficiency, so the four bacteria cannot be compared in absolute terms. The published values are the
line means of the residuals of a model accounting for the experimental block (lme4), so they are centred near 0 and have no unit; they are the phenotypes of the authors' GWAS (EMMAX) for
L. plantarum,
L. brevis and
A. tropicalis.
Gut ROS: reactive oxygen species were measured with the ferrous oxidation-xylenol orange (FOX) assay (change in absorbance at 560 nm) on guts of 3-5-day-old conventionally reared (not gnotobiotic) males on standard glucose-yeast food: five guts pooled per sample, two blocks of two samples (four replicates) per line. The 9 lines were chosen from both extremes of the
L. plantarum load (ranks 1-4 and 32-36 of 36). ROS varied among lines (ANOVA, P < 0.001) but did not correlate with the bacterial load.
The summary dataset has:
-
mn_Load_Atropicalis,
mn_Load_Lbrevis,
mn_Load_Lplantarum,
mn_Load_Efaecalis: the S1 Table columns A_tropicalis, L_brevis, L_plantarum and E_faecalis. The line effect was significant for
A. tropicalis (P = 0.0005, H
2 = 0.394),
L. brevis (P = 0.0008, H
2 = 0.384) and
L. plantarum (P = 0.0007, H
2 = 0.388), but not for
E. faecalis (P = 0.462, H
2 = 0.094).
-
mn_GutROS: the S2 Table column "Relative ROS (mean)" (no unit given).
Values are uploaded exactly as published. The Spearman correlations between the four bacteria computed from the uploaded values reproduce Table 2 of the paper (6 of 6, to three decimals).
Of note:-
E. faecalis was detected in only 48 samples from 26 lines; the 10 other lines have no value (NA): DGRP_301, 307, 324, 335, 358, 360, 365, 427, 820 and 852.
-
L. brevis has no value for DGRP_307 and DGRP_486 (34 lines, as in the paper's Table 1).
- The second column of the S2 Table, "ROS model residuals (mean)", equals "Relative ROS (mean)" minus 0.014638889 (the mean of the 9 lines) for every line, so it is not uploaded: it can be derived as mn_GutROS - 0.014638889 and gives the same ranks.
Not uploaded, because no per-line values were published: the colony-forming units per gut of six
A. tropicalis-colonised lines (the text only gives the range, 47 CFU for line 306 to 3,035 CFU for line 307), the six midgut dimensions of 28 conventionally reared lines (females), and the male and female midgut lengths of 9 lines (S2 Fig only). The variants associated at FDR 0.05 (S3 Table) and the GO enrichment (S4 Table) are GWAS results, not phenotypes. The paper also compares the bacterial loads with phenotypes of other DGRPool studies, such as the nutritional indices of
study 32 (Unckless et al. 2015) and the gut immunocompetence of
study 4 (Bou Sleiman et al. 2015).