A multi-parent recombinant inbred line population of Caenorhabditis elegans enhances mapping resolution and identification of novel QTLs for complex life-history traits

Local populations of the bacterivorous nematode Caenorhabditis elegans can be genetically almost as diverse as global populations. To investigate the effect of local genetic variation on heritable traits, we developed a new recombinant inbred line (RIL) population derived from four wild isolates. The wild isolates were collected from two closely located sites in France: Orsay and Santeuil. By crossing these four genetically diverse parental isolates a population of 200 RILs was constructed. RNA-seq was used to obtain sequence polymorphisms identifying almost 9000 SNPs variable between the four genotypes with an average spacing of 11 kb, possibly doubling the mapping resolution relative to currently available RIL panels. The SNPs were used to construct a genetic map to facilitate QTL analysis. Life history traits, such as lifespan, stress resistance, developmental speed and population growth were measured in different environments. For most traits substantial variation was found, and multiple QTLs could be detected, including novel QTLs not found in previous QTL analysis, for example for lifespan or pathogen responses. This shows that recombining genetic variation across C. elegans populations that are in geographical close proximity provides ample variation for QTL mapping. Taken together, we show that RNA-seq can be used for genotyping, that using more parents than the classical two parental genotypes to construct a RIL population facilitates the detection of QTLs and that the use of wild isolates permits analysis of local adaptation and life history trade-offs.


Introduction
strategies involving multiple lines [32] or panels of wild isolates have been reported, most of the 68 work has been done on the Bristol-Hawaii derived RILs which only captures a subset of the 69 phenotypic and genetic diversity present in C. elegans. 70 Inclusion of more than two parental sources of genetic variation and alleles captures more 71 genetic variation and allows for more precise mapping and identification of potential regulatory 72 variants of complex traits [33]. An alternative to the conventional two-parental genetic mapping  Table 1). To enable crossovers 102 on Chromosome X and to generate extra crossovers, the heterozygous F2 obtained from these 103 initial crosses were further inter-crossed. To create homozygous genotypes, single worms were 104 selected from the F2 as well as from the F2 inter-cross for 6 generations of single worm inbreeding. 105 From these 383 lines, a population of 200 different multi-parental recombinant inbred lines 106 (mpRILs) was randomly picked for mRNA sequencing to obtain the genetic variation in coding 107 sequence.

161
Cross specific recombination in the mpRILs 162 The genetic map shows a highly variable frequency of recombination and introgression sizes 163 ( Figure 3A; Table 2; Supplemental Table 3 and 4). In total 1683 recombination events were found    Table 5 and 6). Correlation analysis ( Figure 5; Supplemental Table 7) across all 211 phenotypic traits showed that timing of first eggs laid were highly correlated across different food 212 conditions. This was also found for population growth, except growth on Sphingomonas. Body size 213 and developmental phenotypes were also highly correlated. This shows that these phenotypes are 214 likely to share a similar genetic architecture.  beyond the parental phenotypes on both sides, yet for the growth on the BT247 strain transgressive 242 mpRILs mostly produce better growth than the parents. Volume, length, and width phenotypes 243 were all much higher for the four parents compared to the mpRILs.

244
Together the results show that ample phenotypic variation of complex traits can be found 245 between the mpRILs and that these phenotypic differences are heritable. Genetic variation across 246 the mpRILS causal for these different functional differences is likely to have major fitness effects.     Thermomixer (Eppendorf, Hamburg, Germany). After cooling on ice for 1 minute, the standard 437 protocol was followed. To construct a genetic map from the SNPs detected in the RNA-seq data we adjusted the method 452 used in Serin & Snoek et al. 2017 [51]. For this C. elegans population we selected the SNPs by  Table 2) were used 457 directly in the SNP map of the population (Supplemental Table 3) or translated to the parental 458 origin genetic map (Supplemental Table 4). at 20 °C. After 24 hours, the locomotive activity of the worms was measured again.

513
WMicrotracker-One™ records movement as photo-beam interruptions within wells of 96-well 514 plates. The data were then processed as follows, (activity of the wells before -activity after 24 515 hours)/before = activity score. An activity score of -1 represents no movement and hence that all 516 worms were dead at the end of the treatment, and an activity score of 0 indicates an activity level 517 after 24 hours that is the same as before. This score can also generate values above zero, which 518 indicates that worms were more active after the hydrogen peroxide treatment.  535 We started with single marker mapping for each trait to find the SNP with most significant QTL 536 (Supplemental Figure 4). This SNP was used as the starting point in the forward mapping approach.

537
Forward mapping was done by selecting cofactors one by one, starting with the most significant 538 and remap with that cofactor and selecting the next most significant SNP till no more SNP was 539 present with a p < 0.001 or a maximum of 10 cofactors was reached. Then QTLs were remapped