Natural variation of chronological aging in the Saccharomyces cerevisiae species reveals diet-dependent mechanisms of life span control

Aging is a complex trait of broad scientific interest, especially because of its intrinsic link with common human diseases. Pioneering work on aging-related mechanisms has been made in Saccharomyces cerevisiae, mainly through the use of deletion collections isogenic to the S288c reference strain. In this study, using a recently published high-throughput approach, we quantified chronological life span (CLS) within a collection of 58 natural strains across seven different conditions. We observed a broad aging variability suggesting the implication of diverse genetic and environmental factors in chronological aging control. Two major Quantitative Trait Loci (QTLs) were identified within a biparental population obtained by crossing two natural isolates with contrasting aging behavior. Detection of these QTLs was dependent upon the nature and concentration of the carbon sources available for growth. In the first QTL, the RIM15 gene was identified as major regulator of aging under low glucose condition, lending further support to the importance of nutrient-sensing pathways in longevity control under calorie restriction. In the second QTL, we could show that the SER1 gene, encoding a conserved aminotransferase of the serine synthesis pathway not previously linked to aging, is causally associated with CLS regulation, especially under high glucose condition. These findings hint toward a new mechanism of life span control involving a trade-off between serine synthesis and aging, most likely through modulation of acetate and trehalose metabolism. More generally it shows that genetic linkage studies across natural strains represent a promising strategy to further unravel the molecular basis of aging.


Linkage mapping
The R/qtl work package was used to carry out Individual Segregant Analysis, considering that 556 genetic markers have been previously identified through RADseq for the entire 'sake x tecc cross' progeny. 1 The threshold at 5% significance level was determined using 1000 permutations. For Bulk Segregant Analysis, the 50 longest-and the 50 shortest-living strains were selected and grown independently, and then pooled using equal amounts of cells from each strain. Total genomic DNAs of the pools were extracted using QIAGEN Genomic-tip 20/G according to the manufacturer's instructions. Genomic Illumina sequencing libraries were prepared with a mean insert size of 280 bp. In total, eight bulks (2 bulks for each tested condition) were multiplexed and sequenced using the Illumina HiSeq2000 system with single-end 50 bp reads. The total number of reads for each sample is given in Table S8. Genetic markers were defined as the nucleotide positions that differ between the haploid YO486 and YO502 strains with a coverage of at least 10x and called by a custom next generation sequencing analysis pipeline, which includes quality control, preprocessing for reads, and mapping to the standard S. cerevisiae reference genome (UCSC release SacCer3; BWA, version: 0.7.12; SAMtools, version 0.1.18). 2,3 In total, 47,770 high-confidence SNPs were identified between the haploid YO486 and YO502 strains. The allele frequencies at each genetic marker in the pools were calculated using the pileup files generated from SAMtools. 3 The statistical and analytical framework based on smoothed G statistics (G') described by Magwene and colleagues 4 was performed with a smoothing default window size W = 33,750 kb. A QTL region was defined as a continuous run of at least 10 SNPs spanning at least a 10 kb interval, where the G' statistic exceeded the false discovery rate threshold 0.05. Peaks were identified as contiguous subregions such that G i ' = 0.95G max ', where G max ' is the site that has the largest G' within the QTL region. 5 Synonymous, nonsynonymous, non-sense and frameshift mutations within each QTL region were annotated functionally using ANNOVAR. 6

Heritability analysis
Broad-sense heritability H 2 is defined as the phenotypic variance (V P ) explained by the genetic variance (V G ). This latter parameter is partitioned as the sum of additive genetic factors (V A ), dominance effects (V D ), gene-gene interactions (V I ) and gene-environment interactions (V E ). 7,8 In this study, broad-sense heritability of the CLS phenotype was estimated as , where is the pooled phenotypic variance of the parental strains and is the pooled phenotypic variance of the segregants (progeny of the 'sake x tecc cross'). 9 By contrast, narrow-sense heritability h 2 only takes into account the effect of additive genetic factors (V A ) on phenotype variance (V P ). The h 2 was estimated using a parent-offspring regression, calculated based on the average results obtained for multiple phenotypes for the parents and for the progeny. 7 The conditions used for phenotyping the progeny and parental strains of the 'sake x tecc cross' on solid agar plates are listed in supplementary  Table S5.

Short chain fatty acid extraction, derivatization, and GC-MS measurement
This method was used to quantify acetic acid in media of aging yeast cultivations. The extraction method was based on a protocol for short chain fatty acids from Moreau et al. 10 Briefly, 20 µL of the internal standard (2-Ethylbutyric acid, c = 20 mmol/L) were added to 180 µL of medium. The samples were acidified with 10 µL of 37% hydrochloric acid, incubated for 15 min at 15 °C (Eppendorf Thermomixer). Then, 1 mL of diethyl ether was added and the samples were again vortexed for 15 min at 15 °C. The upper organic phase was separated by centrifugation (5 min, 21,000 x g and 15 °C) and 900 µL were collected in a new reaction tube. Again, 1 mL of diethyl ether were added to the medium, incubated (5 min) and separated by centrifugation. 900 µL of the organic phase were combined with the first extract. Of this preparation, 250 µl were transferred into a GC glass vial with micro insert in triplicates. For derivatization, 25 µL of N-tert-Butyldimethylsilyl-Nmethyltrifluoroacetamide with 1% tert-Butyldimethylchlorosilane (Restek) were added and the samples were incubated for a minimum of 1 hour at room temperature. For absolute quantification, an external calibration curve including all compounds of interest (Volatile Free Acid Mix, CRM46975, Sigma-Aldrich) was prepared, extracted, and derivatized as described before. GC-MS analysis was performed by using an Agilent 7890A GC coupled to an Agilent 5975C inert XL Mass Selective Detector (Agilent Technologies). A sample volume of 1 µl was injected into a Split/Splitless inlet, operating in split mode (20:1) at 270 °C. The gas chromatograph was equipped with a 30 m (I.D. 250 µm, film 0.25 µm) DB-5MS capillary column (Agilent J&W GC Column). Helium was used as carrier gas with a constant flow rate of 1.4 ml/min. The GC oven temperature was held at 80 °C for 1 min and increased to 150 °C at 10 °C/min. Then, the temperature was increased at 50 °C/min to 280 °C and held for 1.4 min. The total run time was 15 min. The transfer line temperature was set to 280 °C. The mass selective detector (MSD) was operating under electron ionization at 70 eV. The MS source was held at 230 °C and the quadrupole at 150 °C. The detector was switched off during elution of MTBSTFA. For quantification, measurements of the compounds of interest were performed in selected ion monitoring mode. All GC-MS chromatograms were processed using MetaboliteDetector, v3.020151231Ra. 11 The software package supports automatic deconvolution of all mass spectra. Compounds were annotated by retention time and mass spectrum. The data set was normalized by using the response ratio of the integrated peak area of the analyte and the integrated peak area of the internal standard.

Supplementary Tables
Supplementary  If not otherwise indicated, growth assays were carried out on solid YPD medium with 2% glucose, supplemented or not with various compounds at the indicated concentrations. The classification "Carbon utilization" indicates that glucose was added at different concentration or substituted by alternative carbon sources at the indicated concentrations.      Figures   Fig. S1 Growth variation within our natural strain collection.

Supplement
Colony growth phenotyping within our natural strain collection was performed in the 26 indicated conditions, followed by hierarchical clustering (using a centered Pearson correlation metric and average linkage mapping) (a) and Principal Component Analysis (b). YPD and YP + Glucose 2% correspond to independent replicates. The parental strains of the 'sake x tecc cross' are indicated by large dots (purple for YO486 and orange for YO502).
Colony size   Pearson coefficients (R) and the corresponding p-values (p) were calculated between CLS and specific growth rate (a) and yield of biomass (b) for each condition individually. Ascending representation of survival integrals determined for the entire progeny in the four indicated conditions. Blue lines correspond to the survival integral values (mean of 3 biological replicates) obtained for the YO486 and YO502 strains, the surrounding lighter blue areas representing the standard deviations. Green (L) and red (H) rectangles indicate the short-and long-lived strain bulks used for BSA analyses. Survival integrals were determined using an outgrowth kinetics assay for the indicated hybrid strains in SC medium containing 2% glucose and 0.5% glucose (calorie restriction). The results correspond to means ± SDs for 3 biological replicates.   a CLS profiles obtained for FY4 (orange) and FY4ser1 YO486 (violet) flask cultivations in SC medium supplemented with 2% glucose.. The aging cultivations were sampled at different time points for endoand exo-metabolome analyses. b Acetic acid was quantified by GC-MS in the extracellular medium of the FY4 (in blue) and FY4ser1 YO486 (in red) strains. c Intracellular trehalose, serine, glycine, and cysteine concentrations were also determined during aging in the two strains using LC-MS. d Schematic overview showing the metabolites quantified in this study within the known metabolic network of S. cerevisiae. The endo-metabolome data is represented by heatmaps where red and blue squares correspond to high and low amounts of each metabolite, respectively, at the different time points. Red arrows highlight reactions that may be more active in the FY4ser1 YO486 strain due to a metabolic reprogramming in response to SER1 deficiency and that could explain lower acetate accumulation, higher trehalose levels and increased CLS in this strain. The main serine synthesis pathway, starting from 3-phosphoglycerate, is also shown. The first and second reactions in this pathway are catalysed by the Ser3 (or its paralog Ser33) and Ser1 enzymes, respectively. It is the latter enzyme that is deficient in the FY4ser1 YO486 strain (as indicated by the red cross). 3-PHP: 3phosphohydroxypyruvate. In total, 5865 nuclear genes were analysed for codon usage. Data were obtained from http://wiki.yeastgenome.org/index.php/S._cerevisiae_Codon_Usage_Tables. The red arrow indicates serine. All amino acids are represented with the 1-Letter code. * designates a stop codon.