A Proteome-wide Fission Yeast Interactome Reveals Network Evolution Principles from Yeasts to Human

Abstract

Here, we present FissionNet, a proteome-wide binary protein interactome for S. pombe, comprising 2,278 high-quality interactions, of which ∼ 50% were previously not reported in any species. FissionNet unravels previously unreported interactions implicated in processes such as gene silencing and pre-mRNA splicing. We developed a rigorous network comparison framework that accounts for assay sensitivity and specificity, revealing extensive species-specific network rewiring between fission yeast, budding yeast, and human. Surprisingly, although genes are better conserved between the yeasts, S. pombe interactions are significantly better conserved in human than in S. cerevisiae. Our framework also reveals that different modes of gene duplication influence the extent to which paralogous proteins are functionally repurposed. Finally, cross-species interactome mapping demonstrates that coevolution of interacting proteins is remarkably prevalent, a result with important implications for studying human disease in model organisms. Overall, FissionNet is a valuable resource for understanding protein functions and their evolution.

Figure 1. A Proteome-wide Binary Protein Interactome Map of S. pombe

(A) Network representation of FissionNet. Proteins are color-grouped based on PomBase GO slim categories. The number of FissionNet interactions per group is indicated. (B) Y2H and PCA detection rates of the PRS, NRS, FissionNet, and FissionNet hub interactions. (C) Pearson correlation coefficient (PCC) distribution of gene expression profiles of interacting and all random protein pairs. (D) Enrichment of co-localized protein pairs. (E) Enrichment of protein pairs sharing similar functions. (F) Subnetwork of Tas3 and Hhp1 in FissionNet. (G) Coimmunoprecipitation of Tas3-myc and Hhp1-HA in vivo. (H) Centromeric silencing assay of tas3Delta; and hhp1Delta; cells. A schematic of the imr1R region with the ura4⁺ reporter gene is shown. WT denotes wild-type. Data are shown as measurements + standard error (SE). * denotes significant (P<0.05); n.s. denotes not significant. See also Figure S1 and Tables S1 and S2.

Figure 1. A Proteome-wide Binary Protein Interactome Map of S. pombe

(A) Network representation of FissionNet. Proteins are color-grouped based on PomBase GO slim categories. The number of FissionNet interactions per group is indicated. (B) Y2H and PCA detection rates of the PRS, NRS, FissionNet, and FissionNet hub interactions. (C) Pearson correlation coefficient (PCC) distribution of gene expression profiles of interacting and all random protein pairs. (D) Enrichment of co-localized protein pairs. (E) Enrichment of protein pairs sharing similar functions. (F) Subnetwork of Tas3 and Hhp1 in FissionNet. (G) Coimmunoprecipitation of Tas3-myc and Hhp1-HA in vivo. (H) Centromeric silencing assay of tas3Delta; and hhp1Delta; cells. A schematic of the imr1R region with the ura4⁺ reporter gene is shown. WT denotes wild-type. Data are shown as measurements + standard error (SE). * denotes significant (P<0.05); n.s. denotes not significant. See also Figure S1 and Tables S1 and S2.

Figure 2. Atf1-Cid12 Interaction Mediates Silencing at Heat-shock Genes

(A) Subnetwork of Atf1 and Cid12 in FissionNet. (B) Coimmunoprecipitation of Atf1-myc and Cid12-HA in vivo. (C) Semi-quantitative real-time PCR (semi qRT-PCR) shows hsp16 and hsp104 transcript levels in deletion strains. (D) Y2H confirms Cid12 mutants cannot interact with Atf1, but maintain interactions with Hrr1 and Rdp1. (E) Semi qRT-PCR shows that the Cid12 mutants in cid12Delta; cells do not restore the repression of hsp16 or hsp104. (F) Centromeric silencing assay shows that Cid12 mutants retain centromeric silencing function. -RT, no reverse transcriptase. +RT, with reverse transcriptase. Act1⁺ serves as loading control. WT denotes wild-type. See also Figure S2.

Figure 3. S. pombe Protein Interactions are More Conserved in Human than in S. cerevisiae

(A) Sequence-based phylogeny dendrogram of S. pombe (S.p.), S. cerevisiae (S.c.), and human (H.s.). (B) Interaction conservation between reference-query species. (C) Sequence conservation for ortholog pairs that could be conserved between S.p.-S.c. and S.p.-H.s. (D) Interaction conservation between reference-query species for proteins that are conserved in all three species. (E) Interaction conservation in GO Slim categories with at least 50 interactions. (F) Interaction conservation among GO Slim categories that are conserved in all three species. (G) FissionNet subnetworks of Srrm1, SPAC30D11.14C, and SPAC1952.06C. (H) Global splicing profiles of deletion strains relative to wild-type. Columns represent total mRNA (T), pre-mRNA (P), and mature mRNA (M). Data are shown as measurements + SE. * denotes significant (P<0.05); n.s. denotes not significant. See also Figure S3 and Table S3.

Figure 3. S. pombe Protein Interactions are More Conserved in Human than in S. cerevisiae

(A) Sequence-based phylogeny dendrogram of S. pombe (S.p.), S. cerevisiae (S.c.), and human (H.s.). (B) Interaction conservation between reference-query species. (C) Sequence conservation for ortholog pairs that could be conserved between S.p.-S.c. and S.p.-H.s. (D) Interaction conservation between reference-query species for proteins that are conserved in all three species. (E) Interaction conservation in GO Slim categories with at least 50 interactions. (F) Interaction conservation among GO Slim categories that are conserved in all three species. (G) FissionNet subnetworks of Srrm1, SPAC30D11.14C, and SPAC1952.06C. (H) Global splicing profiles of deletion strains relative to wild-type. Columns represent total mRNA (T), pre-mRNA (P), and mature mRNA (M). Data are shown as measurements + SE. * denotes significant (P<0.05); n.s. denotes not significant. See also Figure S3 and Table S3.

Figure 4. Determinants of Interaction Conservation

(A) Interaction conservation as a function of overall protein sequence similarity. (B) Sequence similarity within protein interaction domains and other domains for interactions conserved between yeasts and human. (C) Y2H confirms the interactions of human (H.s.) DRAP1-DR1, the orthologous S. pombe (S.p.) Dpb3-Ncb2, and the cross-species interactions. (D) Crystal structure of human DR1-DRAP1. Boxed region highlights interaction domains. Gray shaded regions denote aligned interaction domain sequences. (E) Interaction conservation within and across topological clusters. (F) Interaction conservation within and across GO categories. Data are shown as measurements + SE. * denotes significant (P<0.05); n.s. denotes not significant. Abbreviations are S. pombe (S.p.), S. cerevisiae (S.c.), and human (H.s.). See also Figure S4.

Figure 5. Functional Divergence of Interactions Involving Paralogous Proteins

(A) Schematic representation of sub-functionalization and neo-functionalization. (B–C) Log odds ratios of sub-functionalization (B) for S. pombe and S. cerevisiae paralog pairs and (C) for S. cerevisiae SSD and WGD paralog pairs after correcting for divergence times. (D–E) Log odds ratios of neo-functionalization (D) for S. pombe and S. cerevisiae paralog pairs and (E) for S. cerevisiae SSD and WGD paralog pairs after correcting for divergence times. (F) Fraction of synthetic lethal pairs among SSD and WGD paralogs known or not known to share interactors. (G) Fraction of coexpressed pairs (PCC>0.4) among SSD and WGD paralogs known or not known to share interactors. Data are shown as measurements + SE. * denotes significant (P<0.05); n.s. denotes not significant. See also Figure S5.

Figure 6. Intact and Coevolved Interactions

(A) Schematic representation of conserved protein interactions that are either intact or coevolved. (B) Within- and cross-species Y2H detects coevolved interactions. (C) Fraction of S.p. interactions that are coevolved with respect to S.c. or human (H.s.). (D) Log odds ratio of co-occurrence of intact and coevolved interactions between S.p.-S.c. and S.p.-H.s. (E) Overall protein sequence similarity of S.p. proteins involved in intact or coevolved interactions. (F) Number of interactors for proteins involved in intact or coevolved interactions. Data are shown as measurements + SE. * denotes significant (P<0.05); n.s. denotes not significant. See also Figure S6 and Table S6.

Figure 7. FissionNet as a Resource for Studying Human Disease

(A) Fraction of inter-protein HGMD mutation pairs that cause the same disease in human interactions with regard to their conservation status in S. pombe and S. cerevisiae. (B) Largest connected subcomponent of FissionNet wherein all proteins have human orthologs with known germline disease or somatic cancer-associated mutations. (C) Impact of human disease mutations and a population variant on intact interactions between human and fission yeast. (D) Fraction of human SSD and WGD paralogs that cause the same disease. (E) The 2R hypothesis predicts two recent WGD events leading to the vertebrate lineage. (F) WGD can lead to more functional redundancy through targeted gene loss that maintains stoichiometric ratios of protein products. SSD leads to more neo-functionalization and sub-functionalization through alterations to initially redundant paralogs. Data are shown as measurements + SE. * denotes significant (P<0.05); n.s. denotes not significant.

Figure 7. FissionNet as a Resource for Studying Human Disease

(A) Fraction of inter-protein HGMD mutation pairs that cause the same disease in human interactions with regard to their conservation status in S. pombe and S. cerevisiae. (B) Largest connected subcomponent of FissionNet wherein all proteins have human orthologs with known germline disease or somatic cancer-associated mutations. (C) Impact of human disease mutations and a population variant on intact interactions between human and fission yeast. (D) Fraction of human SSD and WGD paralogs that cause the same disease. (E) The 2R hypothesis predicts two recent WGD events leading to the vertebrate lineage. (F) WGD can lead to more functional redundancy through targeted gene loss that maintains stoichiometric ratios of protein products. SSD leads to more neo-functionalization and sub-functionalization through alterations to initially redundant paralogs. Data are shown as measurements + SE. * denotes significant (P<0.05); n.s. denotes not significant.