Droplet digital PCR

To design an unbiased primer set, we identified regions of identical sequence between ppw-1 and sago-2 and across the 10 strains of interest. Following Kamitaki et al. (2018), we chose primers to target both genes and probes to discriminate between ppw-1 (FAM) and sago-2 (HEX). Sequences are as follows: forward (CTTGGTACCGCTCCGCTC), reverse (GCTGATTCGGTTTGATCGTC), ppw-1 probe (AGACGAGAAATGTGGAGAGGGGAA), sago-2 probe (AGACGAGAAATGAGGAGTGGGGAA). Both probes anneal in the same location, ensuring competition between them.

Worms from strains N2, CB4856, CB4852, DL238, ECA369, EG4348, JU1088, JU1581, KR314, and QX1211 were reared under standard conditions (as above), bleached to isolate embryos, and grown to reproductive maturity. RNA was extracted with TRIzol (Invitrogen #15596026) and RNeasy columns (Qiagen #74104), following (He 2011). RNA was collected at 2 timepoints, early and middle reproductive maturity (68 ± 2 h and 90 ± 2 h after bleaching, respectively). RNA sample concentrations were quantified and standardized using a NanoDrop (Thermo Scientific), and cDNA was synthesized using the ProtoScript II First Stand cDNA Synthesis Kit (NEB #E6560S). The experiment was replicated as follows: from each experimental condition, we collected 2 RNA samples, for 2 biological replicates; within the plate, each reaction was duplicated, for 2 technical replicates; and we conducted the entire experiment twice.

Droplet digital PCR was carried out with the Bio-Rad QX200 system following the manufacturer's protocol, and results were obtained using the QuantaSoft software (Bio-Rad), via automatic thresholding followed by manual confirmation of droplet selection. All samples produced >8,000 droplets and results from all samples were retained. Concentration, given by number of copies per μL, was modeled with a quasipoisson error structure using the glm() function in R. As ppw-1 was detected at an order of magnitude higher than sago-2, we analyzed the 2 genes separately. By model selection, we identified the minimal model that best described the observed differences in concentration For the ppw-1 analysis, we dropped run date from the model, as it was not significant; for sago-2, run date contributed <1% to the total observed deviance (Supplementary Table 3) but was nevertheless significant, so it was retained. The final models were Concentration ∼ Strain*DevStage/BiolRep for ppw-1 and Concentration ∼ RunDate + Strain*DevStage/BiolRep for sago-2. To determine which strains differed in ppw-1 or sago-2 levels, we performed pairwise contrasts among strains using the TukeyHSD() function and a family-wise confidence level of 95% (only a subset of comparisons are reported in the text). To determine which strains showed differences in concentration according to developmental stage, we performed pairwise contrasts using the lsmeans() function in the package lsmeans, using a confidence level of 95% following a Bonferroni correction for multiple tests.