Genetics Principles And Analysis

[Demetrius Dade]

Thankyou for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

According to whether gene deletion causes inviability or not, genes can be classified as either essential genes or non-essential genes. Essential genes are considered the foundation for life and gene essentiality is regarded as a key criterion when selecting drug targets for combating pathogens and cancer cells1,2. However, essentiality is not a static gene property. In recent years, it has been shown in yeast3 and human cell lines4,5,6 that gene essentiality can vary between genetic backgrounds. Thus, to fully grasp the underpinnings of life and to improve drug target selection, it is imperative to understand which genes can undergo essentiality change and how essentiality change can happen. In particular, it is of both fundamental and practical values to investigate which and how essential genes in a well-defined genetic background can lose essentiality and be converted to non-essential genes.

BOE interactions that bypass the 38 chrII-L essential genes. Genes are represented as nodes, and BOE interactions are represented as directed edges pointing from suppressor genes (small nodes) to query essential genes (large nodes). Node colors represent the indicated gene functional categories and edge colors represent the indicated suppressor types

Thus, we conclude that hidden behind the seemingly identical inviability phenotype of essential gene deletions are real differences in gene importance, which manifest as two observable gene properties: rapidity of lethality upon gene disruption and bypassability. Consistent with the idea that gene importance is a key underlying determinant of bypassability, we found that bypassability no longer exhibited statistically significant correlations with evolutionary rate, species distribution, and codon optimality when we controlled for gene importance by considering only genes with slow spore lethality (Supplementary Figs. 2h-j).

We also examined the relationship between bypassability and the interspecific variation of gene essentiality by focusing on the 124 query genes that have a one-to-one ortholog in S. cerevisiae. Strikingly, among this subset of query genes, 77% (24/31) of the bypassable genes have a non-essential ortholog in S. cerevisiae; for the non-bypassable genes this percentage is only 14% (13/93) (Fig. 3g). Thus, bypassability and differential essentiality between these two species are strongly correlated. Examined from a different angle, 37 of these 124 genes have a non-essential ortholog in S. cerevisiae, and 65% (24/37) of them can be converted into non-essential genes in S. pombe by BOE suppressors. This is remarkable because it means that monogenic changes can eliminate much of the differences in essentiality that have accumulated over the approximately 500 million years since these two species diverged25. Interestingly, the correlation between bypassability and differential essentiality remained highly significant after gene importance was controlled for (Supplementary Fig. 2k). In other words, there appears to be a particularly intimate relationship between bypassable essentiality and evolutionary variation of essentiality. It follows that essentiality bypass may be a common cause of essentiality changes during evolution.

All seven query genes that function in mitochondrial translation are bypassable and share a common set of 12 BOE suppressors (Fig. 2 and Supplementary Fig. 3a). Because a failure to express mtDNA-encoded genes is equivalent, in consequence, to mtDNA loss, we hypothesized that these suppressors may also render mtDNA dispensable. Indeed, mtDNA loss can be readily induced in strains carrying any one of these suppressors but not in a wild-type control strain (Supplementary Fig. 3b). Thus, BOE analysis led to the identification of genes whose alteration can convert S. pombe into a petite-positive species.

One mtDNA-bypassing C-BOE suppressor, atp3-R282C, is a mutation in the gene encoding the gamma subunit of mitochondrial F1-ATPase. Mutations in this gene have been shown to render Kluyveromyces lactis, another petite-negative yeast, and the protist Trypanosoma brucei, tolerant of mtDNA loss29,30. Such mutations are believed to increase the ATP hydrolysis capacity of F1-ATPase and thereby allow the mitochondrial inner membrane potential to be maintained in the absence of mtDNA.

Another mtDNA-bypassing C-BOE suppressor, mts4-S412F, is a mutation in the gene encoding a 19S proteasome subunit. Probably not by coincidence, 6 of the 10 mtDNA-bypassing OP-BOE suppressor genes encode either proteasome subunits or proteasome-associated proteins (Supplementary Fig. 3a), suggesting that proteasome alteration is a common mechanism of mtDNA bypass in S. pombe, even though this mode of mtDNA bypass has not been reported before in any petite-negative species. Thus, through BOE analysis we uncovered a previous unknown link between the proteasome and mtDNA dispensability.

Interestingly, overexpression of Dicer (Dcr1), a ribonuclease known to be a limiting factor of the S. pombe RNAi pathway31, can also bypass mtDNA (Supplementary Fig. 3b), suggesting that upregulating the RNAi pathway promotes cell survival against mtDNA loss.

The shared bypassability of mitochondrial translation genes suggested that bypassability is associated with particular functional modules, and the most common functional modules are protein complexes. Upon examining protein complexes containing at least two subunits encoded by chrII-L query genes, we found that the constituent subunits of a given essential protein complex indeed tend to be either all bypassable or all non-bypassable (Fig. 4a and Supplementary Data 3). In three follow-up analyses, we found that when a complex contains one known bypassable subunit, the other subunits turned out to be bypassable as well (Supplementary Figs. 3c-e). Thus, subunits belonging to the same protein complex tend to share bypassability and complex membership can be used to predict gene bypassability if the bypassability of one complex member is known.

To see whether protein complex bypassability can be predicted a priori, we broadly surveyed complex features and identified 9 features significantly correlated with complex bypassability (Supplementary Figs. 4a, b). Using these features to perform hierarchical clustering of 127 essential protein complexes in S. pombe (Fig. 4b and Supplementary Data 4), we found that all five of the bypassable complexes (names in dark blue in Fig. 4b) defined by our BOE analysis fell into a cluster containing 17 complexes (blue branches in the dendrogram in Fig. 4b). We noted that this same cluster includes five additional complexes (names in light blue in Fig. 4b) whose bypassability is known or expected. Therefore, we predicted that the other complexes in this cluster, including the THO complex (name in black in Fig. 4b), are likely to be bypassable. To test this prediction, we performed a T-BOE screen for suppressors of tho2, which encodes an essential subunit of the THO complex, and found that tho2 is indeed bypassable and that its deletion mutant can be rescued by deleting git1, cyr1, or pka1, three genes acting in the cAMP-protein kinase A signaling pathway (Supplementary Fig. 4c).

In cases where subunits of the same protein complex do not share bypassability or BOE suppressors, new insights on the functional differences between subunits can be gained. Upon surveying the protein complexes with two or more subunits analyzed in our study, we identified three types of non-uniformity between subunits of the same complex (Supplementary Fig. 5).

Second, bypassable subunits belonging to the same complex occasionally do not share the same BOE suppressors. The histone deacetylase Clr6 occurs in two complexes (Clr6 complex I and Clr6 complex II)35,36. Clr6 and four complex I-specific subunits (Pst1, Pst3, Sds3, and Rxt3) are essential, whereas none of the complex II-specific subunits are essential. Thus, it has been assumed that the essential function of Clr6 is that of complex I35. Clr6 and the complex I-specific subunit Pst3 were shown by our BOE analysis to be bypassable, but they do not share any BOE suppressors (Fig. 2). clr6 deletion can be rescued by mutations disrupting either the AMPK complex or the CCR4-NOT complex, while pst3 deletion can be rescued by deleting Clr6 complex II genes (pst2 or cph2). Our follow-up analyses showed that the other three complex I-specific essential subunits, Pst1, Sds3, and Rxt3, can also be bypassed by pst2 deletion (Supplementary Fig. 5b).

These results suggest that the essentiality of Clr6 complex I is at least partly due to its role in antagonizing complex II. Interestingly, simultaneously deleting all three paralogous pst genes (pst1, pst2, and pst3) resulted in lethality (Supplementary Fig. 5b), suggesting that besides their functions in counteracting complex II, the two complex I components Pst1 and Pst3 redundantly contribute to a growth-promoting function. This function is probably Clr6-dependent, as the lethality of pst1 pst2 pst3 triple deletion can be rescued by a BOE suppressor of clr6 (Supplementary Fig. 5b). Thus, a more intricate than expected relationship among the components of Clr6 complexes was revealed by our BOE analysis. In another example, we found that two of the three essential subunits of the NURS complex37, Pir2 and Red5, are bypassable but share no common suppressor (Fig. 2). Follow-up analysis showed that the third essential subunit of this complex, Mtl1, is also bypassable by a suppressor of red5 (Supplementary Fig. 5c), indicating that Mtl1 and Red5 may have a closer relationship with each other than with Pir2.

3a8082e126