Genetic Basis of Variation in Ubiquitin-Proteasome System

14 The ubiquitin-proteasome system (UPS) is the cell’s primary pathway for targeted protein degra15 dation. Although the molecular mechanisms controlling UPS activity are well-characterized, we 16 have almost no knowledge of how these mechanisms are shaped by heritable genetic variation. To 17 address this limitation, we developed an approach that combines fluorescent UPS activity reporters 18 with a statistically powerful genetic mapping framework to comprehensively characterize genetic 19 influences on UPS activity in the yeast Saccharomyces cerevisiae. We applied this approach to 20 substrates of the UPS N-end rule, which relates a protein’s degradation rate to the identity of its 21 N-terminal amino acid (“N-degron”) through the Arg/N-end and Ac/N-end pathways. Genetic 22 influences on UPS activity were numerous and complex, comprising 149 loci influencing UPS ac23 tivity across the 20 N-degrons. Many loci specifically affected individual pathways or degrons and 24 multiple loci exerted divergent effects on distinct UPS pathways. One Arg/N-end pathway-specific 25 locus resulted from multiple causal variants in the promoter, open reading frame, and terminator of 26 the UBR1 E3 ubiquitin ligase gene. These variants differentially affected substrates bound by the 27 Type 1 and Type 2 recognition sites of Ubr1p. Collectively, our results provide the first systematic 28 characterization of genetic influences on UPS activity and a generalizable approach for mapping 29 genetic effects on protein degradation with high statistical power and quantitative precision. 30


Introduction
TFT was only partially stabilized in BY doa10∆ (Supplementary Figure 3), consistent with previous results 60, 63 . Specifically, when followed by a proline residue, ubiquitin is inefficiently cleaved in the 159 ubiquitin-fusion technique 33, 60 . Consequently, the proline N-end TFT simultaneously measures the 160 degradation rate of the proline N-degron and the activity of the ubiquitin-fusion degradation path-161 way 64 . Taken together, these results show that our reporters provide sensitive, pathway-specific, 162 and quantitative measurements of UPS activity.  The number and patterns of QTLs differed between the Ac/N-end and Arg/N-end pathways.

226
The Ac/N-end pathway had a significantly higher median number of QTLs per reporter than the 227 Arg/N-end pathway (9 versus 7, respectively, Wilcoxon test p = 0.021). A notable difference in the 228 patterns of QTLs between pathways was the presence of several large effect QTLs for the Arg/N- We then evaluated the extent to which individual QTLs were shared across multiple reporters.

234
To do so, we divided each chromosome into adjacent 100 kb bins. We considered a QTL to be 235 shared between reporters if the peak position for two or more QTLs were within the same bin. 236 We observed that many QTLs were unique to an individual N-degron ( Figure   tion of the His N-degron, and higher activity for Type 2 N-degrons ( Figure 4B). The QTL's effect 260 size, as measure by the allele frequency difference values, were among the highest in our set of QTLs.

262
A QTL on chromosome V was specific to the Ac/N-end pathway and was detected for 7 of 8 263 Ac/N-end reporters ( Figure 4B). For all 7 reporters, the RM allele was associated with higher UPS  A QTL on chromosome X was found only for the Asn N-degron of the Arg/N-end pathway.

277
The QTL peak is centered on the NTA1 gene, which encodes an amidase that converts N-terminal 278 Asn and Gln residues to Asp and Glu residues, respectively. This conversion is necessary for the 279 recognition and degradation of substrates with N-terminal Asn and Gln residues by the Arg/N-end 280 pathway. The RM allele of the NTA1 locus is associated with higher degradation of Asn N-degrons.

281
RM NTA1 contains two missense variants near the proton donor active site, D111E and E129G.

282
NTA1 is not affected by a local eQTL, making these missense variants strong candidate causal 283 nucleotides. More broadly, these results suggest that genetic variation can influence the full se-

Type II Reporters
Normalized UPS Activity  Table 4). We engineered BY strains carrying chimeric UBR1 alleles to assess the effects their effects. Arg/N-degrons comprise the set of amino acids with side chains too large to accommo- We showed that an Arg/N-end-specific QTL was caused by variation in the UBR1 gene, which   Table 6).

479
We then devised a general strategy to assemble TFT-containing plasmids with defined N- Coralville, IA, USA). We used the TDH3 promoter to drive expression of each TFT reporter. The

485
TDH3 promoter was PCR-amplified from Addgene plasmid #67639 (a gift from John Wyrick). 486 We used the ADH1 terminator for all TFT constructs, which we PCR amplified from Addgene    the TFT in the BY strain) and clonNAT (which selects for the NatMX cassette in the RM strain). 570 We inoculated 5 mL of YPD with freshly streaked diploid cells for overnight growth at 30°C. The 571 next day, we pelleted the cultures, washed them with sterile, ultrapure water, and resuspended the 572 cells in 5 mL of SPO liquid medium (Table 2). We sporulated the cells by incubating them at 573 room temperature with rolling for 9 days. After confirming sporulation by brightfield microscopy,  Austria) and the flowCore R package 94 . We first filtered each flow cytometry dataset to include 606 only those cells within 10% ± the forward scatter (a proxy for cell size) median. We empirically 607 determined that this gating approach captured the central peak of cells in the FSC histogram.

608
This gating approach also removed cellular debris, aggregates of multiple cells, and restricted our 609 analyses to cells of the same approximate size.

611
To better characterize differences in the degradation rate of N-end rule substrates within and 612 between our strains, we transformed our flow cytometry data as follows. We first scaled the log 2

613
TFT ratio relative to the sample with lowest degradation rate. Following this transformation, We evaluated the FDR over a LOD range of 2.5 to 10 in 0.5 LOD increments. We found that 745 a LOD value of 4.5 led to a null QTL rate of 0.0625 and an FDR of 0.507% and we used this 746 value as our significance threshold for QTL mapping. We further filtered our QTL list by excluding 747 QTLs that were not detected in each of two independent biological replicates. Replicating QTLs 748 were defined as those whose peaks were within 100 kB of each other on the same chromosome with 749 791 To create UBR1 allele swap strains, we co-transformed BY strains with 200 ng of plasmid 792 PFA0227 and 1.5 µg of UBR1 repair template. Transformants were selected and single colony pu-793 rified on synthetic complete medium lacking histidine and then patched onto solid YPD medium. 794 We tested each strain for the desired exchange of the NatMX selectable marker with a UBR1 allele 795 by patching strains onto solid YPD medium containing clonNAT. We then verified allelic exchange 796 in strains lacking ClonNAT resistance by colony PCR. We kept 8 independently-derived biological 797 replicates of each allele swap strain. To test the effects of each allele swap, we transformed a subset 798 of TFTs into our allele swap strains and characterized TFT reporter activity by flow cytometry 799 using the methods described above.