Variation in ubiquitin system genes creates substrate-specific effects on proteasomal protein degradation

  1. Mahlon A Collins  Is a corresponding author
  2. Gemechu Mekonnen
  3. Frank Wolfgang Albert  Is a corresponding author
  1. Department of Genetics, Cell Biology, and Development, University of Minnesota, United States

Peer review process

This article was accepted for publication as part of eLife's original publishing model.

History

  1. Version of Record published
  2. Accepted Manuscript published
  3. Accepted
  4. Received
  5. Preprint posted

Decision letter

  1. Magnus Nordborg
    Reviewing Editor; Gregor Mendel Institute, Austria
  2. David Ron
    Senior Editor; University of Cambridge, United Kingdom
  3. Magnus Nordborg
    Reviewer; Gregor Mendel Institute, Austria

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Variation in Ubiquitin System Genes Creates Substrate-Specific Effects on Proteasomal Protein Degradation" for consideration by eLife. Your article has been reviewed by 4 peer reviewers, , including Magnus Nordborg as the Reviewing Editor and Reviewer #1 and the evaluation has been overseen by David Ron as the Senior Editor.

The reviewers have discussed their reviews with one another and agreed this work is elegant, and absolutely deserves to be published. The Reviewing Editor has drafted this to help you prepare a revised submission – please also see the individual reviews for further suggestions for improvements.

There is an overall need to quantify several statements (see individual reviews). This should be straightforward and requires no additional experiments.

In addition, we would like to encourage you to put a bit more thought into the evolutionary analysis. For example, the issue of pleiotropy could be discussed further (as noted by Reviewer 1), and so could the persistent shifts in effect between the strains (Reviewer 4). It may be worth considering Hunter Fraser's use of sign tests to detect selection, e.g.,

https://www.pnas.org/doi/10.1073/pnas.0912245107 in this context.

Reviewer #1 (Recommendations for the authors):

I trust more mechanistic reviewers to comment on your experiments. From my point of view, I have only a few suggestions for improvements:

1) You argue (in connection with Figure 2)1 that "many QTLs were found only for individual pathways", but this is surely sensitive to significance thresholds. A more sophisticated analysis of pleiotropy would be in order.

2) You never quantify how much of the total variation your QTL explains.

3) You never quantify how much of each QTL your identified causal SNPs explain.

4) When dissecting your UBR1 alleles, would be nice to pay homage to this classic:

Laurie-Ahlberg, C. C., and L. F. Stam. 1987. "Use of P-Element-Mediated Transformation to Identify the Molecular Basis of Naturally Occurring Variants Affecting Adh Expression in Drosophila melanogaster." Genetics 115 (1): 129-40.

Reviewer #2 (Recommendations for the authors):

The only thing I would have appreciated is the validation of some of their findings using orthogonal approaches. There are a lot of readily established biochemical assays they can use to support their claims.

Reviewer #3 (Recommendations for the authors):

1. Figure 1 was super helpful in explaining the question and the methods used in this paper.

2. In this sentence, I am not sure that the second portion follows from the first: "These results are consistent with our observation that RM had higher UPS activity for 15 of 20 N-degrons (Figure 1D, Supplementary Table 1), suggesting that the QTLs we have mapped underlie a substantial portion of the heritable UPS activity difference between BY and RM." Maybe you missed the majority of the genetic variation that influences UPS activity, but you got lucky and the portion you detected just happened to be consistent with the overall trend. I think the authors need a different type of analysis to draw a conclusion about whether they sampled "a substantial portion" of heritable differences or whether there is missing heritability here. Can the authors use their data to calculate the narrow sense heritability and then report how much of that heritability is explained by the detected variants?

3. Perhaps reconsider using the term N-degron in the abstract. It's a bit specific. The intro and figures do a great job at defining an N-degron.

4. I did not know that each of the 20 amino acids is dealt with by one of two UPS systems. This was eventually clear in figure 1B where it seems Trp and Met N-degrons are degraded by different systems. Perhaps make this clearer in the text where N-degrons are described and defined.

5. I was mildly confused by the "italics vs plain" statement in figure 3D. Is it intended for only this panel or for the whole figure, or for the whole paper?

6. Some of the focus of the manuscript might be shifted away from general "straw man" questions, like whether this trait is mendelian and controlled by large effect variants (intro lines 55 – 66). The discovery of smaller effect variants is presented as a major finding, but it seems obvious that these exist. Perhaps instead focus on a more quantitative analysis of the power of the current study. How much heritability is explained by previously known genes or variants, and how much additional heritability is explained by the previously unknown genes/variants detected here? Even if a heritability analysis is not feasible, it think a shift in focus from a qualitative statement – we detect small effect variants – to a more quantitative or nuanced statement, would be appropriate.

Reviewer #4 (Recommendations for the authors):

I'm interested in the overall pattern that BY seems to have systematically lower UPS activity than RM. BY carries a rare allele at 3 out of the 4 examples in Figure S5, which hints that perhaps there has been an adaptation of BY for lower UPS. It would be interesting to explore this hypothesis further, or at least to discuss it. Has there been an overall adaptation for BY to be less transcriptionally active, coupled with lower degradation rates? Perhaps this might be explored using the previous data sets on mRNA in these lines – presumably, changes in transcription under this model could be either cis or trans. (That analysis is distinct from the analysis reported at L312, which focuses on specific trans-effects of the UPS variants.) And perhaps the authors may have other ideas as well to explore the evolutionary questions.

The paper reports 149 loci, but if I am reading this correctly it looks like this may double-count shared loci. It would be nice to discuss a bit more about likely sharing (ie pleiotropy) of the hits. (It's also a bit hard to see from 2B which of these are likely shared.)

L143: it should be possible to make this statement quantitative.

Para 303, 312: It seems likely that you may be looking for an effect that is small at individual genes but widespread. I would suggest looking more carefully at the overall distribution: eg in the protein data is there an upward shift in the mean? You could also use a more sophisticated method like ashR to study the distribution of changes. For Figure 5 you could also consider adding information about global patterns in means and correlations, which are difficult to read from these plots.

https://doi.org/10.7554/eLife.79570.sa1

Author response

Reviewer #1 (Recommendations for the authors):

I trust more mechanistic reviewers to comment on your experiments. From my point of view, I have only a few suggestions for improvements:

1) You argue (in connection with Figure 2)1 that "many QTLs were found only for individual pathways", but this is surely sensitive to significance thresholds. A more sophisticated analysis of pleiotropy would be in order.

The revised manuscript includes an expanded analysis of pleiotropy and how the significance threshold used for QTL detection influences QTL pathway specificity. The high degree of QTL pathway specificity we observed was robust to these additional analyses.

As described in the revised manuscript (pg. 6, para. 3, line 180), we considered a QTL's effect common to multiple N-degrons when QTL peaks for distinct N-degrons were within 100 kb of each other and had the same direction of effect. Using these criteria, the 149 QTLs detected for the 20 N-degrons correspond to 35 distinct QTL regions that each affect between 1 and 11 N-degrons. At the LOD score threshold of 4.5 used in the manuscript, 23 of these 35 QTL regions specifically affected N-degrons from an individual pathway in the N-end Rule, with 11 regions affecting only Ac/N-degrons and 12 regions affecting only Arg/N-degrons. Five of the 23 distinct, pathway-specific QTL regions exclusively affected individual N-degrons. The remaining 12 of the 35 distinct QTL regions have the same direction of effect for at least one reporter each from the Arg/N-end and Ac/N-end pathways. The revised manuscript includes additional text in the Results section (pg. 6, para. 3) discussing pleiotropy among QTLs and a new table (Figure 2—figure supplement 2), which provides detailed information on the 35 distinct QTL regions, including their pathway- and N-degron specificity.

To understand the influence of significance threshold on these results, we analyzed how relaxing the LOD score threshold affected the number of pathway- or N-degron-specific QTL regions. The results of this analysis are summarized in Author response table 1. Supplementary file 2 shows the LOD score and allele frequency difference traces for all 20 N-degrons at the 23 N-end Rule pathway- or N-degron-specific QTL regions at multiple significance thresholds. Author response table 2 and Supplementary file 2 provide detailed information on how the N-end Rule pathway- and N-degron-specificity of each of the 35 distinct QTL regions is influenced by LOD score threshold.

Author response table 1
Influence of LOD score significance threshold on the fraction of N-end rule pathway- and N-degron-specific QTLs.
LOD 4.5LOD 3.5LOD 2.5
N-end Rule Pathway-specific QTLs23 / 35 (66%)19 / 35 (54%)16 / 35 (46%)
N-degron-specific QTLs5 / 35 (14%)3 / 35 (9%)3 / 35 (9%)
Author response table 2
Shared QTL Regions and N-degrons Affected.
NQTLchrQTL_CI_leftQTL_peakQTL_CI_rightLODRM_AFDN. N-degrons AffectedN-degrons AffectedPathway-Specific (LOD 4.5)Pathway-Specific (LOD 3.5)Pathway-Specific (LOD 2.5)N-degron-Specific (LOD 4.5)N-degron-Specific (LOD 3.5)N-degron-Specific (LOD 2.5)
1chr_1_a15021394005212245-0.3257Ala, Cys, Gly, Pro, Ser, Thr, ValAc/N-endAc/N-endAc/N-endnonono
2chr_2_a247345051551657393411.10.133Arg, Lys, PheArg/N-endArg/N-endnononono
3chr_2_b25349705856306400108.90.1125Ala, Arg, Asp, Lys, Metnononononono
4chr_4_a429500876621165257.3-0.1084Gln, Glu, Gly, Hisnononononono
5chr_4_b427317532730036260017.50.1522Trp, TyrArg/N-endArg/N-endnononono
6chr_4_c437681742555046356714.30.1566Ala, Phe, Pro, Ser, Thr, Trpnononononono
7chr_4_d43922754982005570008.2-0.1072Asn, AspArg/N-endArg/N-endArg/N-endnonono
8chr_5_a535140037213639625037.50.2487Ala, Cys, Gly, Pro, Ser, Thr, ValAc/N-endAc/N-endAc/N-endnonono
13chr_7_a7420756277510322512.5-0.1552Asn, Thrnononononono
9chr_7_b79817813277917227914.3-0.1567Ala, Cys, Gly, Pro, Ser, Thr, ValAc/N-endnonononono
10chr_7_c7389941418641451318180.16711Ala, Asn, Cys, Gly, Phe, Pro, Ser, Thr, Trp, Tyr, Valnononononono
11chr_7_d784130086970089993415.2-0.1573Arg, Asn, AspArg/N-endArg/N-endArg/N-endnonono
12chr_7_e7856120870980883830107.50.4095His, Leu, Phe, Trp, TyrArg/N-endArg/N-endArg/N-endnonono
15chr_8_a850150980501273006-0.1011TyrArg/N-endArg/N-endArg/N-endTyrTyrTyr
14chr_8_b81188751484001936758.40.1042Asp, LysArg/N-endArg/N-endArg/N-endnonono
17chr_9_a9945501307751680005.10.0992His, LysArg/N-endArg/N-endArg/N-endnonono
16chr_9_b926764129149231541723.50.1956Ala, Gly, Pro, Ser, Thr, ValAc/N-endAc/N-endAc/N-endnonono
18chr_10_a1032390034548237118621.90.18211Ala, Arg, Cys, Gln, Gly, Lys, Phe, Pro, Ser, Trp, Tyrnononononono
19chr_10_b1058977061566064160025.10.1925Ala, Arg, Asn, Cys, Glnnononononono
21chr_11_a11118750156450173600160.1521HisArg/N-endArg/N-endArg/N-endHisnono
20chr_11_b112910303397903887508.4-0.1065Ala, Arg, Cys, Phe, Thrnononononono
22chr_12_a121572001970502268508.20.1262Ala, ProAc/N-endnonononono
23chr_12_b1264441465724167472342.1-0.2511Ala, Cys, Gly, Met, Phe, Pro, Ser, Thr, Trp, Tyr, Valnononononono
24chr_12_c126396376741257026509.70.1084Asp, Gly, His, IleArg/N-endArg/N-endArg/N-endnonono
25chr_13_a13024650589509.6-0.1621AlaAc/N-endAc/N-endnoAlanono
27chr_13_b1331800565257767514.90.1452His, PheArg/N-endArg/N-endArg/N-endnonono
26chr_13_c1326535029616733158416.50.176Arg, Asn, Asp, Glu, His, LysArg/N-endnonononono
29chr_14_a1445073146539448291289.3-0.3428His, Leu, Lys, Met, Phe, Pro, Trp, Tyrnononononono
28chr_14_b1444972547085050402586.70.3362Asp, Thrnononononono
31chr_15_a1513362516386218483120.2-0.1838Arg, Asn, Asp, Glu, His, Lys, Met, Phenononononono
30chr_15_b1534285038805043130010.10.1163Ala, Pro, ThrAc/N-endnonononono
32chr_15_c1552520056152559177510.60.1282Ser, ThrAc/N-endAc/N-endAc/N-endnonono
33chr_15_d155185505697505946508.8-0.0961CysAc/N-endAc/N-endAc/N-endCysCysCys
34chr_16_a1616603019483022207011.60.1215Ala, Gly, Pro, Thr, ValAc/N-endAc/N-endAc/N-endnonono
35chr_16_b1637505040335042895024.5-0.221ProAc/N-endAc/N-endAc/N-endProProPro
  1. Abbreviations: "chr": chromosome, "CI": confidence interval, "RM_AFD": RM allele frequency difference (high – low UPS activity pools)

Generally, altering the significance threshold does not affect the conclusions that at least half of the detected QTLs are specific to an individual N-end Rule pathway and that approximately 10% of QTLs are specific to individual N-degrons. In the revised Results section (pg. 6, para. 3, line 18), the qualitative statement mentioned by the reviewer is replaced with the quantitative information in Author response table 1 for the 4.5 LOD threshold column.

We note that N-end Rule pathway-specificity for 3 QTL regions (IVb, IXa, and XVc) is subject to the following considerations that may not be immediately obvious from reviewing Supplementary file 2.

The Arg/N-end-specific chromosome IVb and IXa QTL regions overlap the leftmost shoulders of QTLs detected with Ac/N-degrons, creating the appearance that these QTLs are not Arg/N-degron-specific when plotted as in Supplementary file 2. However, the peaks of these Arg/N-degron and Ac/N-degron QTLs are not within 100 kb and their confidence intervals do not overlap. We therefore interpret their effects as pathway-specific.

The Ac/N-degron-specific chromosome XVc QTL contains peaks from both glutamate Arg/N-degron replicates. However, the direction of effect of these two peaks is not consistent between replicates (the only such instance in our dataset). Because we could not unambiguously assign a direction of effect for the glutamate N-degron at this region, we do not include it in our set of QTLs and the chromosome XVc region is deemed to be Ac/N-degron-specific. In the revised manuscript, we report the chromosome IVb, IXa, and XVc QTL regions as pathway-specific.

We also note that the pathway-specific QTL regions displayed on pages 7 and 8 (QTL regions containing UBR1) as well as 20 and 21 of Supplementary file 2 are highly overlapping. However, these overlapping regions have opposing directions of effect on the sets of reporters they affect. We, therefore, continue to report these QTL regions as distinct in the revised manuscript, an interpretation supported by our fine-mapping data (see e.g., Figure 3C).

2) You never quantify how much of the total variation your QTL explains.

Our bulk segregant analysis QTL mapping method is based on comparing allele frequencies obtained from whole-genome sequencing pools of cells with extreme phenotypes. The genotypes of individual segregants, which would be needed for calculating explained variance, are not ascertained using this approach. Thus, it is not readily possible to calculate the proportion of variance explained by our QTLs.

3) You never quantify how much of each QTL your identified causal SNPs explain.

Author response table 3 displays the variance in ubiquitin-proteasome system (UPS) activity explained by each tested causal gene allele and variant.

Author response table 3
Variance Explained by Causal Alleles and Variants.
GeneAlleleN-degronVariance Explained
DOA10K1012NThr0.04
DOA10Q410EThr0.075
DOA10RM_fullThr0.911
DOA10Y1186FThr0.597
DOA10K1012NGly0.608
DOA10Q410EGly0.74
DOA10RM_fullGly0.879
DOA10Y1186FGly0.77
NTA1D111EAsn0.204
NTA1E129GAsn0.764
NTA1RM_fullAsn0.789
NTA1RM_prAsn-0.014
UBC6D229GAla0.717
UBC6RM_fullAla0.666
UBC6RM_prAla-0.027
UBC6RM_termAla0.183
UBC6D229GThr0.762
UBC6RM_fullThr0.928
UBC6RM_prThr0.008
UBC6RM_termThr0.006
UBR1RM_causalPhe0.9
UBR1RM_fullPhe0.917
UBR1RM_nonPhe-0.019
UBR1RM_prPhe0.818
UBR1RM_causalTrp0.937
UBR1RM_fullTrp0.975
UBR1RM_nonTrp0.059
UBR1RM_prTrp0.934
UBR1RM_fullAsn0.438
UBR1RM_ORFAsn0.134
UBR1RM_prAsn-0.012
UBR1RM_termAsn0.064
UBR1RM_fullAsp0.403
UBR1RM_ORFAsp0.101
UBR1RM_prAsp0.097
UBR1RM_termAsp0.048
UBR1RM_fullPhe0.866
UBR1RM_ORFPhe0.343
UBR1RM_prPhe0.316
UBR1RM_termPhe-0.031
UBR1RM_fullTrp0.964
UBR1RM_ORFTrp0.813
UBR1RM_prTrp0.928
UBR1RM_termTrp0.028

We note that these estimates are obtained from experiments in near-isogenic strains that differ only at the tested causal gene allele or variant. The fraction of variance explained is thus inflated relative to what would be observed in the segregant populations used for QTL mapping and should not be used to estimate the variance explained by a QTL region. Therefore, we do not include these estimates in the revised manuscript.

4) When dissecting your UBR1 alleles, would be nice to pay homage to this classic: Laurie-Ahlberg, C. C., and L. F. Stam. 1987. "Use of P-Element-Mediated Transformation to Identify the Molecular Basis of Naturally Occurring Variants Affecting Adh Expression in Drosophila melanogaster." Genetics 115 (1): 129-40.

We thank the reviewer for the suggestion to include this important work on identifying causal variants for enzyme activity in highly polymorphic genomic regions. The work appears as reference 62 (pg. 8, para 3, line 218) in the revised manuscript.

Reviewer #2 (Recommendations for the authors): The only thing I would have appreciated is the validation of some of their findings using orthogonal approaches. There are a lot of readily established biochemical assays they can use to support their claims.

Previously published theoretical and empirical observations have demonstrated that tandem fluorescent timers (TFTs) provide precise and sensitive measures of protein degradation kinetics. In particular, the TFT system has been extensively used to measure differences in the degradation rate of UPS N-end Rule substrates (Khmelinskii et al., 2012, Khmelinskii and Knop, 2014, Kats et al., 2018). Given the well-established validity of the TFT system for measuring N-end Rule activity and the comparatively lower precision, sensitivity, and throughput of conventional biochemical measurements of protein degradation (in particular, pulse-chase Western blotting and cycloheximide chase analysis [Kong et al., 2021]), we argue that further experimentation is not needed to establish the claims made in our work.

Reviewer #3 (Recommendations for the authors):

1. Figure 1 was super helpful in explaining the question and the methods used in this paper.

Thank you!

2. In this sentence, I am not sure that the second portion follows from the first: "These results are consistent with our observation that RM had higher UPS activity for 15 of 20 N-degrons (Figure 1D, Supplementary Table 1), suggesting that the QTLs we have mapped underlie a substantial portion of the heritable UPS activity difference between BY and RM." Maybe you missed the majority of the genetic variation that influences UPS activity, but you got lucky and the portion you detected just happened to be consistent with the overall trend. I think the authors need a different type of analysis to draw a conclusion about whether they sampled "a substantial portion" of heritable differences or whether there is missing heritability here. Can the authors use their data to calculate the narrow sense heritability and then report how much of that heritability is explained by the detected variants?

As noted in our response to Reviewer 1, because of the pooled nature of our genetic mapping method, it is not readily possible to calculate heritability from our QTL mapping data. Author response table 3 presents the variance explained by the tested causal alleles and variants. However, as noted in our response to Reviewer 1, these estimates are inflated because they are derived from near-isogenic cell populations that differ only at the tested alleles / variants. Given these considerations, the second clause in the sentence mentioned by the reviewer has been removed from the revised manuscript.

3. Perhaps reconsider using the term N-degron in the abstract. It's a bit specific. The intro and figures do a great job at defining an N-degron.

The term “N-degron” has been removed from the abstract and replaced with more general terms.

4. I did not know that each of the 20 amino acids is dealt with by one of two UPS systems. This was eventually clear in figure 1B where it seems Trp and Met N-degrons are degraded by different systems. Perhaps make this clearer in the text where N-degrons are described and defined.

We thank the reviewer for this suggestion to improve the manuscript’s clarity. The revised introduction indicates that the N-end Rule contains two distinct targeting complexes when introducing the N-end Rule (pg. 3, para. 2, line 105) and references Figure 1, which illustrates the two pathways of the N-end rule.

5. I was mildly confused by the "italics vs plain" statement in figure 3D. Is it intended for only this panel or for the whole figure, or for the whole paper?

The “italics vs. plain” statement is intended only for panels 3D and 4B/F/J. To improve clarity, we have added additional annotation to these panels.

6. Some of the focus of the manuscript might be shifted away from general "straw man" questions, like whether this trait is mendelian and controlled by large effect variants (intro lines 55 – 66). The discovery of smaller effect variants is presented as a major finding, but it seems obvious that these exist. Perhaps instead focus on a more quantitative analysis of the power of the current study. How much heritability is explained by previously known genes or variants, and how much additional heritability is explained by the previously unknown genes/variants detected here? Even if a heritability analysis is not feasible, it think a shift in focus from a qualitative statement – we detect small effect variants – to a more quantitative or nuanced statement, would be appropriate.

The revised manuscript provides a more nuanced introduction to the genetics of UPS activity, in particular, emphasizing the expectation that UPS activity, like most traits, is genetically complex. We agree with the reviewer that the relative contributions of individual QTLs to the heritability of UPS activity would be an interesting and informative analysis. However, as noted in our response to Reviewer 1, it is not readily feasible to perform such an analysis. Instead, as suggested by the reviewer, several qualitative statements have been replaced by quantitative descriptions. In particular, the qualitative statement mentioned by the reviewer is replaced with a quantitative statement regarding QTL effect sizes (pg. 6, para. 1, line 164).

Reviewer #4 (Recommendations for the authors):

Specific comments:

I'm interested in the overall pattern that BY seems to have systematically lower UPS activity than RM. BY carries a rare allele at 3 out of the 4 examples in Figure S5, which hints that perhaps there has been an adaptation of BY for lower UPS. It would be interesting to explore this hypothesis further, or at least to discuss it. Has there been an overall adaptation for BY to be less transcriptionally active, coupled with lower degradation rates? Perhaps this might be explored using the previous data sets on mRNA in these lines – presumably, changes in transcription under this model could be either cis or trans. (That analysis is distinct from the analysis reported at L312, which focuses on specific trans-effects of the UPS variants.) And perhaps the authors may have other ideas as well to explore the evolutionary questions.

We thank the reviewer for these interesting suggestions, which we have addressed with a series of new analyses. In brief, we did not detect evidence for lineage-specific selection on UPS gene expression using eQTL data or on individual N-end reporters.

To explore whether the consistent differences in UPS activity between the BY and RM strains might reflect adaptive changes in these lineages, we performed several additional analyses. We first applied the sign test (Fraser et al., 2010; https://doi.org/10.1073/pnas.0912245107) to a recent comprehensive BY / RM eQTL mapping dataset, which became available after the original sign test publication. This newer eQTL dataset comprises 36,498 eQTLs mapped for 5,643 genes in a panel of 1,000 recombinant offspring from the BY / RM cross (Albert et al., 2018; https://doi.org/10.7554/eLife.35471). The results of this analysis are presented in Author response table 4. We performed the analysis at 11 different LOD thresholds (ranging from 2.5 to 50) to examine the influence of QTL effect size on the results. Across all genes, we do not find evidence for lineage-specific selection except at LOD thresholds of 40 and 45. Given the large fraction of eQTLs that are excluded at these high thresholds (94.6% and 95.3%), the large fraction of genes excluded (70.5% and 73.8%), and especially the marginally significant p-values (0.038 and 0.026) obtained at these two thresholds, we conclude that there is, at best, limited evidence from the sign test for lineage-specific selection on overall mRNA transcript abundance in the BY / RM cross. Future work is needed to reconcile discrepancies in the results of the sign test as obtained in different eQTL datasets from the same cross.

Author response table 4
Results of the Sign Test for Lineage-Specific Selection Applied to All BY / RM eQTLs.
LODn_eQTLsn_genesn_pairsreinf_BY_upreinf_RM_upoppos_BY_upoppos_RM_upexcess_reinforcing_pairschi_sq_p
2.43649856432845627757952509–140.818
51969453722081449577698357190.701
109379451411242253204041754.630.934
15612536867451542022791102.240.992
20445230764801081281796518.20.431
253458260933673951274120.60.276
30278822242305465852622.60.145
352310191217844486818200.144
40196316641403541531124.30.0375
4517061474110283342722.90.0261
501475129985212731617.90.0569

Abbreviations: “LOD”: LOD score threshold for calling cis / trans eQTL pairs, "n_eQTLs": number of eQTLs, "n_genes": number of genes with an eQTL, "n_pairs": number of cis / trans eQTL pairs, "reinf_BY_up": number of cis / trans eQTLs where the BY allele of the cis and trans eQTLs increases expression (reinforcing pairs), "reinf_RM_up": number of cis / trans eQTLs where the RM allele of the cis and trans eQTLs increases expression (reinforcing pairs), "oppos_BY_up": number of cis / trans eQTLs where the BY allele of the cis eQTL increases expression and the RM allele of the trans eQTL increases expression (opposing pairs), "oppos_RM_up": number of cis / trans eQTLs where the RM allele of the cis eQTL increases expression and the BY allele of the trans eQTL increases expression (opposing pairs), "excess_reinforcing_pairs", the number of excess reinforcing eQTL pairs calculated as in Fraser et al., 2010, "chi_sq_p": p-value of the chi-square test for enrichment of reinforcing pairs.

As in the original manuscript sign test manuscript, we performed a gene ontology (GO) enrichment analysis on the sets of reinforcing cis / trans eQTL pairs from the set of eQTLs to determine whether such pairs are enriched for UPS genes. We were able to replicate the previously-described enrichment for genes of the ergosterol biosynthesis pathway (Fraser et al., 2010, Author response table 5). However, there was no enrichment for ubiquitin system or proteasome GO terms in the sets of reinforcing eQTL pairs at any of the tested LOD score thresholds (Author response table 5).

Author response table 5
Results of Gene Ontology Enrichment of All cis / trans eQTL pairs.
GOBPIDp_valueOddsRatioExpCountCountTermcategoryLOD
GO:00321972.8E-05315.0630transposition; RNA-mediatedall_reinforcing2.426606803
GO:00150740.000463.19.8720DNA integrationall_reinforcing2.426606803
GO:00064870.000772.711.622protein N-linked glycosylationall_reinforcing2.426606803
GO:00465130.001210.72.227ceramide biosynthetic processall_reinforcing2.426606803
GO:00064580.00293.36.1713'de novo' protein foldingall_reinforcing2.426606803
GO:00189040.0037Inf0.994ether metabolic processall_reinforcing2.426606803
GO:00511310.00399.21.976chaperone-mediated protein complex assemblyall_reinforcing2.426606803
GO:19011350.00451.492.08114carbohydrate derivative metabolic processall_reinforcing2.426606803
GO:00700850.00631.821.2332glycosylationall_reinforcing2.426606803
GO:00610770.00683.93.959chaperone-mediated protein foldingall_reinforcing2.426606803
GO:00066720.0075.42.727ceramide metabolic processall_reinforcing2.426606803
GO:00091010.0071.819.7530glycoprotein biosynthetic processall_reinforcing2.426606803
GO:00162260.00893.15.4311iron-sulfur cluster assemblyall_reinforcing2.426606803
GO:00310700.00936.12.226intronic snoRNA processingall_reinforcing2.426606803
GO:00349650.00936.12.226intronic box C/D snoRNA processingall_reinforcing2.426606803
GO:00321972.8E-05315.0630transposition; RNA-mediatedall_reinforcing2.426606803
GO:00150740.000463.19.8720DNA integrationall_reinforcing2.426606803
GO:00064870.000772.711.622protein N-linked glycosylationall_reinforcing2.426606803
GO:00465130.001210.72.227ceramide biosynthetic processall_reinforcing2.426606803
GO:00064580.00293.36.1713'de novo' protein foldingall_reinforcing2.426606803
GO:00189040.0037Inf0.994ether metabolic processall_reinforcing2.426606803
GO:00511310.00399.21.976chaperone-mediated protein complex assemblyall_reinforcing2.426606803
GO:19011350.00451.492.08114carbohydrate derivative metabolic processall_reinforcing2.426606803
GO:00700850.00631.821.2332glycosylationall_reinforcing2.426606803
GO:00610770.00683.93.959chaperone-mediated protein foldingall_reinforcing2.426606803
GO:00066720.0075.42.727ceramide metabolic processall_reinforcing2.426606803
GO:00091010.0071.819.7530glycoprotein biosynthetic processall_reinforcing2.426606803
GO:00162260.00893.15.4311iron-sulfur cluster assemblyall_reinforcing2.426606803
GO:00310700.00936.12.226intronic snoRNA processingall_reinforcing2.426606803
GO:00349650.00936.12.226intronic box C/D snoRNA processingall_reinforcing2.426606803
GO:00321974E-063.511.5127transposition; RNA-mediatedall_reinforcing5
GO:00150740.000113.77.4818DNA integrationall_reinforcing5
GO:00064580.0013.94.7912'de novo' protein foldingall_reinforcing5
GO:00610770.00115.53.079chaperone-mediated protein foldingall_reinforcing5
GO:00064870.00132.78.8218protein N-linked glycosylationall_reinforcing5
GO:00189040.0013Inf0.774ether metabolic processall_reinforcing5
GO:00465130.00248.51.736ceramide biosynthetic processall_reinforcing5
GO:00705250.00248.51.736tRNA threonylcarbamoyladenosine metabolic processall_reinforcing5
GO:00511310.003810.61.345chaperone-mediated protein complex assemblyall_reinforcing5
GO:00322590.00451.821.2933methylationall_reinforcing5
GO:00461650.00522.211.3220alcohol biosynthetic processall_reinforcing5
GO:00442810.00631.3150.75177small molecule metabolic processall_reinforcing5
GO:00066620.007Inf0.583glycerol ether metabolic processall_reinforcing5
GO:00029490.007Inf0.583tRNA threonylcarbamoyladenosine modificationall_reinforcing5
GO:00332150.007Inf0.583iron assimilation by reduction and transportall_reinforcing5
GO:00461310.00833.83.268pyrimidine ribonucleoside metabolic processall_reinforcing5
GO:00195090.00867.11.535L-methionine salvage from methylthioadenosineall_reinforcing5
GO:00066940.00862.48.0615steroid biosynthetic processall_reinforcing5
GO:00066960.00892.75.9512ergosterol biosynthetic processall_reinforcing5
GO:00066720.00945.12.116ceramide metabolic processall_reinforcing5
GO:00198560.00945.12.116pyrimidine nucleobase biosynthetic processall_reinforcing5
GO:19026520.00982.56.7113secondary alcohol metabolic processall_reinforcing5
GO:00321971.5E-064.56.4420transposition; RNA-mediatedall_reinforcing10
GO:00150747.7E-054.34.6214DNA integrationall_reinforcing10
GO:00189040.00022Inf0.494ether metabolic processall_reinforcing10
GO:00066620.0018Inf0.363glycerol ether metabolic processall_reinforcing10
GO:00062780.00212.76.815RNA-dependent DNA biosynthetic processall_reinforcing10
GO:19020470.00229.11.095polyamine transmembrane transportall_reinforcing10
GO:00461650.00412.66.5614alcohol biosynthetic processall_reinforcing10
GO:00436050.00569.70.854cellular amide catabolic processall_reinforcing10
GO:00195090.00569.70.854L-methionine salvage from methylthioadenosineall_reinforcing10
GO:00724880.0064.91.826ammonium transmembrane transportall_reinforcing10
GO:00158460.00646.11.345polyamine transportall_reinforcing10
GO:00068330.006521.80.493water transportall_reinforcing10
GO:00420440.006521.80.493fluid transportall_reinforcing10
GO:19026520.0072.94.2510secondary alcohol metabolic processall_reinforcing10
GO:00020980.0073.92.437tRNA wobble uridine modificationall_reinforcing10
GO:00064000.0082.37.0514tRNA modificationall_reinforcing10
GO:00090670.00862.55.7112aspartate family amino acid biosynthetic processall_reinforcing10
GO:00066960.009233.779ergosterol biosynthetic processall_reinforcing10
GO:00066940.00962.65.111steroid biosynthetic processall_reinforcing10
GO:00342200.00961.625.2737ion transmembrane transportall_reinforcing10
GO:00322590.011.911.6620methylationall_reinforcing10
GO:00150743.3E-0553.7113DNA integrationall_reinforcing15
GO:00321976.2E-0544.9815transposition; RNA-mediatedall_reinforcing15
GO:00461659.3E-0544.5914alcohol biosynthetic processall_reinforcing15
GO:00062780.000313.55.0714RNA-dependent DNA biosynthetic processall_reinforcing15
GO:19026520.00113.93.3210secondary alcohol metabolic processall_reinforcing15
GO:00195090.001118.70.594L-methionine salvage from methylthioadenosineall_reinforcing15
GO:00705250.001118.70.594tRNA threonylcarbamoyladenosine metabolic processall_reinforcing15
GO:00066960.001642.939ergosterol biosynthetic processall_reinforcing15
GO:00090860.001642.939methionine biosynthetic processall_reinforcing15
GO:00712670.002512.50.684L-methionine salvageall_reinforcing15
GO:00431020.002512.50.684amino acid salvageall_reinforcing15
GO:00436050.002512.50.684cellular amide catabolic processall_reinforcing15
GO:00161280.00333.53.229phytosteroid metabolic processall_reinforcing15
GO:00068110.00341.727.8142ion transportall_reinforcing15
GO:00709000.0034280.393mitochondrial tRNA modificationall_reinforcing15
GO:19008640.0034280.393mitochondrial RNA modificationall_reinforcing15
GO:00066940.00493410steroid biosynthetic processall_reinforcing15
GO:00090660.0052.65.9513aspartate family amino acid metabolic processall_reinforcing15
GO:00441070.00513.33.429cellular alcohol metabolic processall_reinforcing15
GO:00161250.00632.74.7811sterol metabolic processall_reinforcing15
GO:19020470.00757.50.884polyamine transmembrane transportall_reinforcing15
GO:00065310.0079140.493aspartate metabolic processall_reinforcing15
GO:00906460.0079140.493mitochondrial tRNA processingall_reinforcing15
GO:00158040.00825.21.375neutral amino acid transportall_reinforcing15
GO:00162260.00825.21.375iron-sulfur cluster assemblyall_reinforcing15
GO:00986550.00931.912.2921cation transmembrane transportall_reinforcing15
GO:00158400.0095Inf0.22urea transportall_reinforcing15
GO:00343110.0095Inf0.22diol metabolic processall_reinforcing15
GO:00343120.0095Inf0.22diol biosynthetic processall_reinforcing15
GO:00197550.0095Inf0.22one-carbon compound transportall_reinforcing15
GO:00428830.0095Inf0.22cysteine transportall_reinforcing15
GO:00905020.00962.27.1214RNA phosphodiester bond hydrolysis; endonucleolyticall_reinforcing15
GO:00150741.8E-066.82.8713DNA integrationall_reinforcing20
GO:00321972.8E-065.43.8815transposition; RNA-mediatedall_reinforcing20
GO:00062784.4E-054.63.7313RNA-dependent DNA biosynthetic processall_reinforcing20
GO:00705250.0004724.20.474tRNA threonylcarbamoyladenosine metabolic processall_reinforcing20
GO:00905020.00063.34.7313RNA phosphodiester bond hydrolysis; endonucleolyticall_reinforcing20
GO:00346540.00171.641.4459nucleobase-containing compound biosynthetic processall_reinforcing20
GO:00065510.001736.10.313leucine metabolic processall_reinforcing20
GO:00709000.001736.10.313mitochondrial tRNA modificationall_reinforcing20
GO:19008640.001736.10.313mitochondrial RNA modificationall_reinforcing20
GO:00158040.00217.61.015neutral amino acid transportall_reinforcing20
GO:00162260.00217.61.015iron-sulfur cluster assemblyall_reinforcing20
GO:00013020.00293.92.568replicative cell agingall_reinforcing20
GO:00442490.00391.591.27111cellular biosynthetic processall_reinforcing20
GO:19015760.0041.592.28112organic substance biosynthetic processall_reinforcing20
GO:00906460.0041180.393mitochondrial tRNA processingall_reinforcing20
GO:00090830.0041180.393branched-chain amino acid catabolic processall_reinforcing20
GO:00063100.0052.28.6917DNA recombinationall_reinforcing20
GO:00158400.006Inf0.162urea transportall_reinforcing20
GO:00197550.006Inf0.162one-carbon compound transportall_reinforcing20
GO:00428830.006Inf0.162cysteine transportall_reinforcing20
GO:00436050.0077120.473cellular amide catabolic processall_reinforcing20
GO:00195090.0077120.473L-methionine salvage from methylthioadenosineall_reinforcing20
GO:00708800.0077120.473fungal-type cell wall beta-glucan biosynthetic processall_reinforcing20
GO:00708790.0077120.473fungal-type cell wall beta-glucan metabolic processall_reinforcing20
GO:00442830.00981.721.1932small molecule biosynthetic processall_reinforcing20
GO:00150741.7E-067.42.412DNA integrationall_reinforcing25
GO:00062781.7E-055.62.9212RNA-dependent DNA biosynthetic processall_reinforcing25
GO:00321972.7E-055.23.0512transposition; RNA-mediatedall_reinforcing25
GO:00905029.7E-054.53.4412RNA phosphodiester bond hydrolysis; endonucleolyticall_reinforcing25
GO:00346540.00141.829.5445nucleobase-containing compound biosynthetic processall_reinforcing25
GO:00013020.0024.71.887replicative cell agingall_reinforcing25
GO:00442490.0031.664.6182cellular biosynthetic processall_reinforcing25
GO:00158040.00398.40.714neutral amino acid transportall_reinforcing25
GO:00063100.0042.56.4314DNA recombinationall_reinforcing25
GO:19015760.0041.665.1982organic substance biosynthetic processall_reinforcing25
GO:00158400.0042Inf0.132urea transportall_reinforcing25
GO:00197550.0042Inf0.132one-carbon compound transportall_reinforcing25
GO:00705250.004614.60.393tRNA threonylcarbamoyladenosine metabolic processall_reinforcing25
GO:00075680.00633.32.868agingall_reinforcing25
GO:00903050.00872.36.313nucleic acid phosphodiester bond hydrolysisall_reinforcing25
GO:00066960.00933.91.886ergosterol biosynthetic processall_reinforcing25
GO:00150742E-079.31.9912DNA integrationall_reinforcing30
GO:00062781.6E-067.22.3712RNA-dependent DNA biosynthetic processall_reinforcing30
GO:00321973.5E-066.62.5312transposition; RNA-mediatedall_reinforcing30
GO:00905024.5E-066.42.5812RNA phosphodiester bond hydrolysis; endonucleolyticall_reinforcing30
GO:00063100.000283.44.9514DNA recombinationall_reinforcing30
GO:00903050.001934.6812nucleic acid phosphodiester bond hydrolysisall_reinforcing30
GO:19020470.004513.50.383polyamine transmembrane transportall_reinforcing30
GO:00442490.00461.74559cellular biosynthetic processall_reinforcing30
GO:00346540.00471.820.5932nucleobase-containing compound biosynthetic processall_reinforcing30
GO:19015760.00591.645.4359organic substance biosynthetic processall_reinforcing30
GO:00158460.00710.80.433polyamine transportall_reinforcing30
GO:19030080.00924.61.345organelle disassemblyall_reinforcing30
GO:00150743.7E-0811.21.7412DNA integrationall_reinforcing35
GO:00321974.5E-078.42.1212transposition; RNA-mediatedall_reinforcing35
GO:00062784.5E-078.42.1212RNA-dependent DNA biosynthetic processall_reinforcing35
GO:00905025.9E-078.12.1712RNA phosphodiester bond hydrolysis; endonucleolyticall_reinforcing35
GO:00063100.0001244.0513DNA recombinationall_reinforcing35
GO:00903050.0002143.7112nucleic acid phosphodiester bond hydrolysisall_reinforcing35
GO:00442490.000622.134.6750cellular biosynthetic processall_reinforcing35
GO:00346540.000772.315.6228nucleobase-containing compound biosynthetic processall_reinforcing35
GO:19015760.0008235.0150organic substance biosynthetic processall_reinforcing35
GO:19020470.003315.30.343polyamine transmembrane transportall_reinforcing35
GO:00158460.005112.20.393polyamine transportall_reinforcing35
GO:00434570.006740.30.142regulation of cellular respirationall_reinforcing35
GO:00066960.00954.51.355ergosterol biosynthetic processall_reinforcing35
GO:00150741.8E-0812.21.6412DNA integrationall_reinforcing40
GO:00905021.7E-079.41.9612RNA phosphodiester bond hydrolysis; endonucleolyticall_reinforcing40
GO:00321972.2E-079.12.0112transposition; RNA-mediatedall_reinforcing40
GO:00062782.2E-079.12.0112RNA-dependent DNA biosynthetic processall_reinforcing40
GO:00063103.1E-054.83.6113DNA recombinationall_reinforcing40
GO:00903056.7E-054.73.3312nucleic acid phosphodiester bond hydrolysisall_reinforcing40
GO:00158460.001721.70.273polyamine transportall_reinforcing40
GO:19020470.001721.70.273polyamine transmembrane transportall_reinforcing40
GO:00442490.0027228.8741cellular biosynthetic processall_reinforcing40
GO:00346540.00532.112.8822nucleobase-containing compound biosynthetic processall_reinforcing40
GO:00086100.00633.13.479lipid biosynthetic processall_reinforcing40
GO:00066960.006451.235ergosterol biosynthetic processall_reinforcing40
GO:00158040.00879.30.463neutral amino acid transportall_reinforcing40
GO:00150746.3E-0812.31.511DNA integrationall_reinforcing45
GO:00905023.7E-079.91.711RNA phosphodiester bond hydrolysis; endonucleolyticall_reinforcing45
GO:00062783.7E-079.91.711RNA-dependent DNA biosynthetic processall_reinforcing45
GO:00321976.3E-079.31.811transposition; RNA-mediatedall_reinforcing45
GO:00063102.8E-055.33.112DNA recombinationall_reinforcing45
GO:00903057.6E-055.12.911nucleic acid phosphodiester bond hydrolysisall_reinforcing45
GO:00442490.000392.522.936cellular biosynthetic processall_reinforcing45
GO:19015760.00052.523.236organic substance biosynthetic processall_reinforcing45
GO:19013600.00182.220.732organic cyclic compound metabolic processall_reinforcing45
GO:00346540.00242.41019nucleobase-containing compound biosynthetic processall_reinforcing45
GO:00066960.00296.215ergosterol biosynthetic processall_reinforcing45
GO:00161280.00485.41.25phytosteroid metabolic processall_reinforcing45
GO:00441070.00565.21.25cellular alcohol metabolic processall_reinforcing45
GO:19026520.00565.21.25secondary alcohol metabolic processall_reinforcing45
GO:00551140.00752.47.114oxidation-reduction processall_reinforcing45
GO:00066940.00754.81.35steroid biosynthetic processall_reinforcing45
GO:00461650.00874.61.35alcohol biosynthetic processall_reinforcing45
GO:00066440.00994.41.45phospholipid metabolic processall_reinforcing45
GO:00150742.5E-058.81.328DNA integrationall_reinforcing50
GO:00062785.6E-057.71.478RNA-dependent DNA biosynthetic processall_reinforcing50
GO:00905026.8E-057.51.58RNA phosphodiester bond hydrolysis; endonucleolyticall_reinforcing50
GO:00321970.000126.81.618transposition; RNA-mediatedall_reinforcing50
GO:00063100.000524.72.499DNA recombinationall_reinforcing50
GO:00066960.00118.10.845ergosterol biosynthetic processall_reinforcing50
GO:00161280.0026.90.955phytosteroid metabolic processall_reinforcing50
GO:00903050.0024.22.428nucleic acid phosphodiester bond hydrolysisall_reinforcing50
GO:00441070.00236.60.995cellular alcohol metabolic processall_reinforcing50
GO:19026520.00236.60.995secondary alcohol metabolic processall_reinforcing50
GO:00066940.003361.065steroid biosynthetic processall_reinforcing50
GO:00461650.00385.81.15alcohol biosynthetic processall_reinforcing50
GO:00161250.00754.81.285sterol metabolic processall_reinforcing50
GO:19013620.00942.48.2115organic cyclic compound biosynthetic processall_reinforcing50

Abbreviations: "GOBPID": gene ontology biological process ID, "ExpCount": expected number of genes for a given GOBPID, "LOD": LOD score threshold for including a gene with a cis / trans eQTL pair.

Because the results of our GO enrichment could be affected by the reference gene set (the list of all genes with at least one cis and one trans eQTL), we devised a complementary strategy to test for lineage-specific selection on UPS gene expression. Using the same set of BY / RM eQTLs, we applied the sign test to the sets of UPS genes, proteasome genes, ubiquitin system genes, E3 ligases, and proteasome chaperone genes. We did not detect lineage-specific selection in any of these gene sets (Author response table 6).

Author response table 6
Results of the Sign Test for Lineage-Specific Selection Applied to UPS Gene BY / RM eQTLs.
LODn_eQTLsn_genesn_pairsreinf_by_upreinf_rm_upoppos_by_upoppos_rm_upexcess_reinforcing_pairschi_sq_pgene_set
2.411281869516302920–4.210.815all_UPS_genes
5587176721323231301all_UPS_genes
102501422921287–4.410.428all_UPS_genes
15162105191873–2.740.619all_UPS_genes
2011585100442–3.20.472all_UPS_genes
25907280332-30.475all_UPS_genes
30635650221–1.61all_UPS_genes
35524740121-21all_UPS_genes
403734301200NaNall_UPS_genes
453230201100NaNall_UPS_genes
2.424233143326–0.8571proteasome_genes
514632123225–1.331proteasome_genes
106230400040NaNproteasome_genes
153219200020NaNproteasome_genes
201814100010NaNproteasome_genes
25119100010NaNproteasome_genes
2.48291457613232515-40.812ubiquitin_system_genes
541013757101820901ubiquitin_system_genes
101731052421174-11ubiquitin_system_genes
1511880171772–1.651ubiquitin_system_genes
20886580341-21ubiquitin_system_genes
25735860231-21ubiquitin_system_genes
30544840121-21ubiquitin_system_genes
35444030021–2.670.324ubiquitin_system_genes
403129200200NaNubiquitin_system_genes
452625100100NaNubiquitin_system_genes
2.46181115611161910-11E3_ligase_genes
5300103398121452.670.909E3_ligase_genes
10129791527421.61E3_ligase_genes
158356914401.781E3_ligase_genes
206446502300NaNE3_ligase_genes
254939301200NaNE3_ligase_genes
303733301200NaNE3_ligase_genes
353028200200NaNE3_ligase_genes
402220200200NaNE3_ligase_genes
451817100100NaNE3_ligase_genes
2.4579604200NaNproteasome_chaperone_genes
5318403100NaNproteasome_chaperone_genes
10158201100NaNproteasome_chaperone_genes
15127101000NaNproteasome_chaperone_genes
2096101000NaNproteasome_chaperone_genes
2565101000NaNproteasome_chaperone_genes
3054101000NaNproteasome_chaperone_genes
3543101000NaNproteasome_chaperone_genes
4043101000NaNproteasome_chaperone_genes
4543101000NaNproteasome_chaperone_genes

Abbreviations: “LOD”: LOD score threshold for calling cis / trans eQTL pairs, "n_eQTLs": number of eQTLs, "n_genes": number of genes with an eQTL, "n_pairs": number of cis / trans eQTL pairs, "reinf_BY_up": number of cis / trans eQTLs where the BY allele of the cis and trans eQTLs increases expression (reinforcing pairs), "reinf_RM_up": number of cis / trans eQTLs where the RM allele of the cis and trans eQTLs increases expression (reinforcing pairs), "oppos_BY_up": number of cis / trans eQTLs where the BY allele of the cis eQTL increases expression and the RM allele of the trans eQTL increases expression (opposing pairs), "oppos_RM_up": number of cis / trans eQTLs where the RM allele of the cis eQTL increases expression and the BY allele of the trans eQTL increases expression (opposing pairs), "excess reinforcing pairs" the number of excess reinforcing pairs, calculated as in Fraser et al., 2010, "chi_sq_p": p-value of the chi-square test for enrichment of reinforcing pairs.

Although we did not detect lineage-specific selection on global or UPS gene mRNA abundance, we note that these analyses do not detect lineage-specific selection on factors other than transcript abundance. For example, lineage-specific selection may occur through causal missense variants, such as the ones we identified here. For example, the DOA10 and NTA1 genes each contain multiple causal missense variants that alter UPS activity, but neither gene's transcript abundance is affected by a local eQTL. Integration of the effects of missense variants with those that alter gene expression in selection tests remains an interesting avenue for future work.

We note that the significant excess of UPS QTLs at which the RM allele increases UPS activity (pg. 6, para. 2, line 170) provides evidence that N-end Rule activity could have been subject to lineage-specific selection. To test whether this result was due to a strong enrichment of RM alleles at certain individual reporters, we applied this analysis to the sets of QTLs obtained for the 20 individual N-degrons. We did not detect an enrichment of any of the individual N-degrons, suggesting that the enrichment of RM QTLs that increase UPS activity is a general effect (Author response table 7).

Author response table 7
Results of Binomial Enrichment Test Applied to QTLs for Individual N-degrons.
N-degronN_QTLsN_RM_upN_BY_upp_value
Ala151050.30
Arg7431.00
Asn7341.00
Asp8530.73
Cys9451.00
Gln3211.00
Glu4221.00
Gly9541.00
His9630.51
Ile1101.00
Leu2111.00
Lys7520.45
Met4130.63
Phe10640.75
Pro13850.58
Ser9630.51
Thr12840.39
Trp6420.69
Tyr7431.00
Val7431.00

Abbreviations: "N_QTLs": Number of QTLs for the indicated N-degron, "N_RM_up": Number of QTLs where the RM allele increases UPS activity for the indicated N-degron, "N_BY_up": Number of QTLs where the BY allele increases UPS activity for the indicated N-degron. "p_value": p-value of the binomial test for the set of QTLs for the indicated N-degron.

The paper reports 149 loci, but if I am reading this correctly it looks like this may double-count shared loci. It would be nice to discuss a bit more about likely sharing (ie pleiotropy) of the hits. (It's also a bit hard to see from 2B which of these are likely shared.)

The reviewer is correct. As described in our response to Reviewer 1, the revised manuscript now indicates that the 149 instances of QTL detection correspond to 35 distinct QTL regions (pg. 6, para. 3, line 185). Figure 2—figure supplement 2 provides detailed information on these regions, including which reporters each region affects. The revised manuscript includes an extended discussion of pleiotropy among the set of N-end Rule QTLs (pg. 6, para. 3).

L143: it should be possible to make this statement quantitative.

As noted in our response to Reviewer 1, it is not readily possible to calculate the amount of variance explained by each QTL due to the pooled nature of our bulk segregant analysis QTL mapping method. Accordingly, we have removed the statement mentioned by the reviewer from the revised Results section.

Para 303, 312: It seems likely that you may be looking for an effect that is small at individual genes but widespread. I would suggest looking more carefully at the overall distribution: eg in the protein data is there an upward shift in the mean? You could also use a more sophisticated method like ashR to study the distribution of changes.

We appreciate the reviewer’s suggestion, which we suspect may have been prompted by our wording of the following sentence from the Results section (page 23, paragraph 2, lines 307-310 of the original manuscript):

“This result is consistent with recent observations that suggest that altering UBR1 expression exerts broad effects on protein degradation or related processes controlling protein abundance and that protein sequences, rather than function or subcellular localization, are the primary determinants of degradation rates.”

Our intent was to convey that substrates of E3 ligases such as Ubr1 are more likely to share sequence features than they are to share a function or subcellular localization. In other words, E3 ligases typically target sets of proteins that are functionally diverse but that share common sequence features, e.g., an N-degron. The wording of this sentence may have unintentionally suggested that the causal UBR1 promoter variant should affect the abundance of many proteins. However, E3 ligases influence the abundance of distinct sets of dozens to several hundred proteins (Kong et al., 2021, Christiano et al., 2020). The moderate effect of the causal UBR1 promoter variant on UBR1 expression is, therefore, not expected to create widespread effects on the proteome but instead to affect a small subset of Ubr1 substrates and related proteins. We have revised this section to more clearly articulate these ideas. We have also re-analysed our data as described below.

The causal UBR1 variant significantly altered the abundance of 39 of 3,047 detected proteins at a 0.1 false discovery rate (FDR) threshold. Following the reviewer’s suggestion, we computed the overall median log2 fold change and found that it was -0.012 for all 3,047 detected proteins (that is, very close to zero) and 0.37 for the set of 39 differentially abundant proteins (that is, an average increase for proteins with significantly different abundance). The number of differentially abundant proteins, the average upward shift in log2 fold change for differentially abundant proteins, and the significant fraction of differentially abundant proteins exhibiting increased abundance (28 / 39, 72%, binomial p = 9.5e-3) are all consistent with the causal variant’s moderate effect on UBR1 expression.

The causal variant also did not have widespread effects on the transcriptome (median log2 fold change = -0.0024 [again, very close to zero]). As reported in our initial submission, 78 genes were differentially expressed at an FDR of 0.1. For these genes, the overall effect of the causal BY allele, which decreases UBR1 expression, was to decrease transcript abundance (median log2 fold change = -0.18, 60 / 78 decreased expression [77%, binomial p = 2e-6]).

Following the reviewer’s suggestion, we used ashr to explore how using the false sign rate (FSR, Stephens, 2017; https://doi.org/10.1093/biostatistics/kxw041) to call differentially expressed genes influenced our analysis of the causal UBR1 variant’s effects. Using an FSR threshold of 0.1, we detected 86 genes with altered mRNA transcript abundance. For these genes, the BY allele tended to decrease transcript abundance (median log2 fold change = -0.2, 65 / 86 [76%, binomial p = 2e-6]). All differentially expressed genes detected using the FDR were also detected using the FSR. Eight additional differentially expressed genes were detected using the FSR, but not the FDR. Each of these eight genes had lower absolute log2 fold changes than the 78 differentially expressed genes detected using the FDR. Therefore, we conclude that while the FSR is slightly more permissive, the FDR and FSR both capture the same global patterns of effects in our RNA-seq data.

In contrast to these consistent results in RNA-seq, ashr produced different results than the FDR when applied to our proteomics data. As reported in our initial submission, a 0.1 FDR threshold applied to abundance differences reported by the Proteome Discoverer software results in 39 differentially abundant proteins. The derived BY UBR1 allele increased the abundance of 28 of these 39 (72%, binomial p = 9.5e-3) differentially abundant proteins. At a 0.1 FSR threshold, we identified 56 differentially abundant proteins, with the BY allele increasing the abundance of 13 / 56 (23%, binomial p = 7.3e-5).This difference arises from the fact that there are considerable discrepancies among the sets of differentially abundant proteins called by FDR and FSR. We note that of the 39 differentially abundant proteins reported by Proteome Discoverer, 12 were estimated to have large absolute fold changes (>= 0.5, the 99th percentile observed in our data) but were not reported as significant by FSR (Author response image 1). Notably, these 12 genes included the known Ubr1-regulated proteins Tma10 and Adh2 (Kong et al., 2021, Christiano et al., 2020). For 8 of these 12 proteins, the BY allele increased protein abundance. These 12 proteins had relatively high standard errors and, as expected, their fold changes were considerably lower following the adaptive shrinkage that is part of ashr (Author response image 1). Second, ashr identified an additional 48 differentially abundant proteins. The BY allele decreases the abundance of 41 of these 48 proteins (Author response image 1).

Given that ashr strongly depends on correctly estimated standard errors, its application to mass-spectrometry data with a small sample size could be potentially problematic in a manner that does not seem to apply to RNA-seq, which, particularly at the very high sequencing depth used here, produces more accurate estimates of the standard error. We have therefore elected to continue to use the FDR applied to p-values reported by Proteome Discoverer to call differentially expressed genes at the protein and RNA levels. We note that our general conclusion that the causal UBR1 promoter variant affects the expression of dozens of genes at the protein and RNA levels remains unchanged irrespective of the method used to call differentially expressed genes.

Author response image 1

For Figure 5 you could also consider adding information about global patterns in means and correlations, which are difficult to read from these plots.

The revised Figure 5 contains the median log2 fold changes for our proteomics and RNA-seq data (pg. 14). The revised Results section reports the correlation in log2 fold change for genes detected in both our proteomics (pg. 13, para. 4, line 362) and RNA-seq data (pg. 13, para. 5, line 377).

https://doi.org/10.7554/eLife.79570.sa2

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Mahlon A Collins
  2. Gemechu Mekonnen
  3. Frank Wolfgang Albert
(2022)
Variation in ubiquitin system genes creates substrate-specific effects on proteasomal protein degradation
eLife 11:e79570.
https://doi.org/10.7554/eLife.79570

Share this article

https://doi.org/10.7554/eLife.79570