Impact of gut microbiota on the fly's germ line

Unlike vertically transmitted endosymbionts, which have broad effects on their host's germ line, the extracellular gut microbiota is transmitted horizontally and is not known to influence the germ line. Here we provide evidence supporting the influence of these gut bacteria on the germ line of Drosophila melanogaster. Removal of the gut bacteria represses oogenesis, expedites maternal-to-zygotic-transition in the offspring and unmasks hidden phenotypic variation in mutants. We further show that the main impact on oogenesis is linked to the lack of gut Acetobacter species, and we identify the Drosophila Aldehyde dehydrogenase (Aldh) gene as an apparent mediator of repressed oogenesis in Acetobacter-depleted flies. The finding of interactions between the gut microbiota and the germ line has implications for reproduction, developmental robustness and adaptation.


Supplementary Figure 5: Faster development of embryos of dechorionated parents vs. their control counterparts.
Sequence of images taken from a representative time-lapse confocal microscopy of His2Av-mRFP embryos 3 , with (Dechor.) and without (Control) removal of extracellular gut bacteria in the preceding generation. The images span development cycles [10][11][12][13] (until the onset of cellularization; indicated) and show the beginning of the cycle. Gray lines in the middle represent the whole span of embryonic development with several stages (fertilization, cellularization, hatching) indicated. The skewed blue lines symbolize the emerging and increasing time difference between the onsets of corresponding cycles in the two conditions.

RNA-seq analysis
Adaptors were removed from sequence reads using the cutadapt program 4, 5. Reads were mapped to the drosophila transcriptome (Ensembl version BDGP.25) using Bowtie2 and TopHat software 6 , then Cufflinks and Cuffmerge 6 were applied to define a list of transcripts that are comparable between all samples. Differentially expressed transcripts including fold-change and statistics were identified by applying two different methods. First Cuffdiff 6 was used directly on cufflinks output. The second method was to use the DESeq R package 7 on the bowtie2 output. Finally both methods were merged.
GO enrichments were computed using the 'David' online resource 8,9,10,11 with cutoffs for up/down regulation and FDR set to 2-fold and 0.05, respectively. Up-and downregulated gene-sets were analyzed separately.

Estimating developmental stage based on the transcriptional profile
The developmental stage of embryos at 2hr AED (with and without bacterial removal the in preceding generation) was estimated by comparing the RNA-seq mRNA data to a reference time-series transcriptional data from Lott et al. 12 . The estimation was based on the method developed by Efroni et al. 13 with a few modifications. The estimation requires identification of a subset of genes with a single peak of expression which is preceded and followed by a monotonic change along the reference time-series. This subset is used as a ruler for estimating the developmental stage of a query sample by determining, for each gene, its location along the time-series based on its measured expression in the sample. The stage of the sample as a whole is then determined by The estimations by the two schemes yielded very similar stages (no more than half a cycle difference). In this paper we used the estimation based on the first scheme and verified the main findings using the second scheme. All estimations were performed using in-house MATLAB scripts (available from the online github repository at https://github.com/elgartmi/AgeByRNA).

Inferring the Aldh network
The Aldh network of Fig. 3B was inferred from the literature using the STRING online resource 10 . STRING uses literature mining to infer connections between genes and assigns confidence to each interaction. We used it to define the Aldh network as follows: The highest confidence genes with direct connection to Aldh were determined using Aldh as an input gene and selecting the top 10 genes which are not defined only by a CG designation. Each of these genes was then re-fed into STRING to obtain secondary interactions, filtered in the same way as the genes with direct connection. To generate a connected map, STRING was queried the combined set of genes. The map was trimmed by removing secondary-interacting genes which have less than 2 interactions with the top 10 genes with direct connection to Aldh.

Measuring enzymatic activity of Aldh
Aldh activity was assayed in ovaries and larvae homogenates by monitoring the change in absorbance at 340nm in the reaction mixture as described by Moxon et al. 15 . 10 whole 3rd instar larvae or 15 dissected ovaries were homogenized, respectively, in 100l or 20l of extraction buffer described by Heinstra et al. 16 . Homogenates were allowed to stand for 15min on ice before centrifugation in a microfuge at 15,000g for 20min. The supernatants were used for subsequent assays and the operations were conducted at 4°C. Protein content was determined by the QBIT kit and used to standardize when comparing different samples. Aldh activity was measured with a TECAN device at 25 degrees Celsius. The absorbance at 340nm was recorded after addition of 15ul supernatant to 85ul of reaction buffer as described by Moxon et al. 17 containing acetaldehyde and NAD + (Sigma) supplemented with Pyrazole (to block ADH activity; Sigma) to the final concentration of 0.2M.