Genetic screen for suppression of transcriptional interference identifies a gain-of-function mutation in Pol2 termination factor Seb1

Significance

Long noncoding RNA (lncRNA) transcriptional interference with the synthesis of downstream messenger RNAs underlies regulated biological responses of yeast cells to nutrient availability. For example, expression of the phosphate (PHO) regulon in phosphate-replete fission yeast cells is tunable by genetic manipulations that favor or disfavor precocious lncRNA 3′-processing/termination. A forward genetic screen for relief of lncRNA interference with pho1 expression uncovered a mutation G476S in the RNA-binding domain of essential termination factor Seb1 that evokes precocious lncRNA termination in response to 5′-proximal poly(A) sites in a manner dependent on cleavage and polyadenylation factor CPF, termination factor Rhn1, and inositol pyrophosphate synthesis. Multiple lines of evidence point to Seb1-G476S as a unique gain-of-function mutation in a Pol2 transcription termination factor.

Abstract

The system of long noncoding RNA (lncRNA)–mediated transcriptional interference that represses fission yeast phosphate homoeostasis gene pho1 provides a sensitive readout of genetic influences on cotranscriptional 3′-processing and termination and a tool for discovery of regulators of this phase of the Pol2 transcription cycle. Here, we conducted a genetic screen for relief of transcriptional interference that unveiled a mechanism by which Pol2 termination is enhanced via a gain-of-function mutation, G476S, in the RNA-binding domain of an essential termination factor, Seb1. The genetic and physical evidence for gain-of-function is compelling: 1) seb1-G476S de-represses pho1 and tgp1, both of which are subject to lncRNA-mediated transcriptional interference; 2) seb1-G476S elicits precocious lncRNA transcription termination in response to lncRNA 5′-proximal poly(A) signals; 3) seb1-G476S derepression of pho1 is effaced by loss-of-function mutations in cleavage and polyadenylation factor (CPF) subunits and termination factor Rhn1; 4) synthetic lethality of seb1-G476S with pho1 derepressive mutants rpb1-CTD-S7A and aps1∆ is rescued by CPF/Rhn1 loss-of-function alleles; and 5) seb1-G476S elicits an upstream shift in poly(A) site preference in several messenger RNA genes. A crystal structure of the Seb1-G476S RNA-binding domain indicates potential for gain of contacts from Ser476 to RNA nucleobases. To our knowledge, this is a unique instance of a gain-of-function phenotype in a eukaryal transcription termination protein.

Termination of transcription by RNA polymerase II (Pol2) typically occurs downstream of sites of nascent RNA 3′ cleavage and polyadenylation. The prevailing “torpedo model” holds that the 5′-phosphate nascent RNA terminus generated by endonuclease cleavage permits ingress of a processive nuclear 5′ exonuclease that digests the nascent RNA and elicits termination when it “catches up” to elongating Pol2 (1, 2). Timely 3′-processing and Pol2 termination delimits the distal margins of transcription units and avoids transcription interference with the expression of neighboring genes (3). The Pol2 elongation complex and associated elongation factors, the cleavage and polyadenylation apparatus, and postcleavage termination factors together comprise a large physical and functional target for regulating 3′ end formation and termination.

Interference occurs when transcription initiating from an upstream promoter quashes initiation at a cooriented downstream promoter (4). In eukaryal genomes, in which transcription units specifying messenger RNAs (mRNAs) and long noncoding (lnc) RNAs are interspersed, there are many occasions for lncRNA synthesis to negatively impact a flanking protein-coding gene (5). This mode of control is exemplified in fission yeast, in which lncRNA interference with mRNA expression is the basis for transcriptional control of phosphate homeostasis. To wit, the fission yeast phosphate (PHO) regulon (6) comprises three phosphate acquisition genes—pho1 (cell surface acid phosphatase), pho84 (inorganic phosphate transporter), and tgp1 (glycerophosphodiester transporter)—each of which is repressed during growth in phosphate-rich medium by transcription in cis of a 5′ flanking lncRNA, these being prt(nc-pho1), prt2, and nc-tgp1, respectively (reviewed in ref. 7). lncRNA transcription traversing the PHO mRNA promoters evicts the PHO gene-activating transcription factor Pho7 from its DNA binding sites (8). The PHO genes are derepressed by genetic maneuvers that encourage precocious lncRNA 3′-processing/termination upstream of the PHO mRNA promoters, thereby allowing Pho7 binding (9–15). Conversely, the PHO genes are hyperrepressed in cells bearing loss-of-function mutations in components of the 3′-processing/termination machinery (12, 13).

It is of particular interest that lncRNA termination is influenced by the Pol2 carboxyl-terminal domain (CTD) code. The CTD of the Pol2 Rpb1 subunit consists of tandem repeats of a consensus heptapeptide sequence, Y¹S²P³T⁴S⁵P⁶S⁷, that acts as a platform to engage proteins that govern transcription and catalyze mRNA processing. The CTD primary structure, which is dynamically remodeled by serine, threonine, and tyrosine phosphorylation, conveys information about the status of the Pol2 transcription apparatus—a CTD code—that is read by CTD-interacting proteins (16–19). Replacement of all fission yeast CTD Ser7 residues with alanine results in derepression of pho1, pho84, and tgp1 expression in phosphate-replete cells, whereas replacement of all Thr4 residues with alanine causes pho1 and pho84 hyperrepression (12).

Several lines of evidence connect the impact of CTD mutations on PHO gene expression to the protein factors that direct cotranscriptional RNA 3′-processing and Pol2 termination. Cleavage and polyadenylation factor (CPF) is a 13-subunit protein assembly responsible for the cleavage and 3′-polyadenylation of nascent Pol2 transcripts that precedes termination (20). Eight of the fission yeast CPF subunits, including the cleavage endonuclease Ysh1 and the poly(A) polymerase Pla1, are essential for viability. Five of the CPF subunits are dispensable for growth: Ctf1, Ssu72 (a phosphoprotein phosphatase that acts on the CTD), Dis2 (a phosphoprotein phosphatase), Ppn1, and Swd22. Seb1 and Rhn1 are transcription termination factors that bind the phospho-CTD (21). Whereas Seb1 is essential in fission yeast, Rhn1 is not. The derepression of pho1 by the CTD-S7A mutation was negated by deletion or loss-of-function mutations of CPF subunits Ctf1, Ssu72, Ppn1, Swd22, and Dis2 and termination factor Rhn1 (12). These CPF/Rhn1 mutations per se caused hyperrepression of pho1, akin to the effects of CTD-T4A (12). The finding that CTD-T4A is synthetically lethal with ppn1∆ and swd22∆ indicates that Thr4 and the Ppn1•Swd22 module of CPF play important, albeit functionally redundant, roles in promoting Pol2 termination (12).

To understand how the CTD-T4A mutation affects interference and to potentially identify agents of prt lncRNA transcription and transcription termination, we recently undertook a genetic screen in mutagenized CTD-T4A cells for suppressors of the pho1 hyperrepressed state (22). Using a chromogenic overlay assay to gauge acid phosphatase activity of individual colonies, we obtained a collection of candidate suppressor mutants that had elevated Pho1 activity vis-à-vis the parental CTD-T4A cells. The candidates underwent several rounds of backcrossing to affirm single-gene segregation of the increased Pho1 phenotype, which was verified by quantitative assays of Pho1 activity. Analysis of 18 independent STF (suppressor of threonine four) isolates revealed, in every case, a mutation in the carboxyl-terminal pyrophosphatase domain of the enzyme Asp1 (22), a central agent of inositol pyrophosphate (IPP) metabolism. Asp1 is a bifunctional IPP kinase/pyrophosphatase that interconverts 5-IP7 and 1,5-IP8 (23, 24). We had shown that an active site mutation in the N-terminal kinase domain that abolishes IP8 synthesis results in hyperrepression of pho1 expression, while an active site mutation in the pyrophosphatase domain (that elevates IP8 levels) causes pho1 derepression in a manner that requires CPF subunits and Rhn1 (25). Transcriptome profiling of two of the new STF strains showed that all three PHO regulon genes were coordinately up-regulated vis-à-vis the parental T4A strain. Moreover, whereas these two asp1-STF alleles were lethal in a wild-type CTD background, they were viable in combination with mutations in CPF and Rhn1, in which context Pho1 was also derepressed (22). These findings implicated the Asp1 pyrophosphatase in constraining 1,5-IP8 or 1-IP7 synthesis by Asp1 kinase, without which 1-IPPs can accumulate to toxic levels that elicit precocious termination by CPF/Rhn1.

Although the results of the initial STF screen strengthened the case for IPPs as agonists of 3′-processing/termination, they did not shed light on non-IPP modulators of lncRNA-dependent transcription interference with the PHO regulon. To date, nearly all of the players in the PHO transcription interference phenomenon have been identified and characterized based on gene deletions or targeted missense mutations, most of which, perforce, involve inessential components of the 3′-processing/termination pathway(s). Other than Rpb1, we have little appreciation of the contributions of essential transcriptional/processing factors to fission yeast phosphate homeostasis.

Notwithstanding that Asp1 pyrophosphatase mutations dominated the output of the initial STF screen, we sought to maximize the utility of the STF assay and seek out additional genes that confer the STF phenotype by subjecting a larger pool of “STF positive” isolates to focused sequencing of their asp1 genes and excluding any that had acquired asp1 mutations. In this way, we obtained strain STF19, which has a single missense mutation (Gly476Ser) in the RNA-binding domain of the essential termination factor Seb1. The seb1-G476S allele derepresses pho1 (and tgp1) in a wild-type CTD background, and this effect is squelched by CPF and Rhn1 mutations and by inactivation of the Asp1 IPP kinase. The genetic and structural evidence reported here suggests that seb1-G476S is a heretofore unique gain-of-function mutation of Seb1 with respect to 3′-processing/termination.

Results

The STF Screen Identifies an Allele of Transcription Termination Factor Seb1.

Testing individual colonies of mutagenized CTD-T4A cells for cell surface acid phosphatase activity by a colorimetric colony overlay assay was performed as described previously (22). In our first round of screening, whole-genome sequencing showed that 12/12 independent STF strains had Asp1 pyrophosphatase mutations. Analysis of a second set of six independent STF isolates by focused PCR amplification of the Asp1 pyrophosphatase open reading frame (ORF) and sequencing the PCR product showed that all of them had Asp1 pyrophosphatase mutations (22). Here, by applying the asp1 gene sequencing as a pretest for new candidate STF isolates, we identified an STF strain that had a wild-type asp1⁺ locus. This strain was backcrossed twice to a T4A strain of the opposite mating type and subjected to random spore analysis, whereby populations of haploid progeny (250 to 400) were tested for Pho1 acid phosphatase activity via the overlay assay. In both backcrosses, ∼50% of the haploid progeny stained red and ∼50% were pale, signifying that the STF phenotype results from a mutation in a single gene. The T4A strain bearing this STF allele, provisionally named STF19, and the parental T4A strain were assayed quantitatively for Pho1 activity after growth at 30 °C in phosphate-replete liquid medium. STF19 T4A cells expressed 11-fold higher acid phosphatase activity than the T4A parent (Fig. 1A). STF19 T4A cells grew as well as the T4A parent on YES agar at all temperatures tested (shown in Fig. 1C and SI Appendix, Fig. S4, respectively).

Fig. 1.

The STF screen identifies an allele of transcription termination factor Seb1. (A) S. pombe strains bearing the indicated seb1 and rpb1-CTD alleles were grown to A₆₀₀ of 0.5 to 0.8 in liquid culture in YES medium at 30 °C. Cells were then harvested, washed with water, and assayed for Pho1 acid phosphatase activity by conversion of p-nitrophenylphosphate to p-nitrophenol. Activity is expressed as the ratio of A₄₁₀ (p-nitrophenol production) to A₆₀₀ (input cells). Each datum in the bar graph is the average of assays using cells from at least three independent cultures ± SEM. (B) Seb1 is composed of an N-terminal CID (colored gold) and an RRM-type RNA-binding domain (colored cyan) flanked by central and carboxyl-terminal segments (gray) that are predicted to be structurally disordered. The amino acids at the termini and the domain boundaries are indicated by number. The amino acid sequence of the Seb1 RRM domain is aligned to that of Saccharomyces cerevisiae Nrd1. Positions of side chain identity/similarity are denoted by dots above the alignment. The conserved Gly476 that was mutated to serine in the STF19 strain is highlighted in white font on a black background. Conserved amino acids that make atomic contact with RNA in Nrd1•RNA crystal structures are denoted by arrowheads: red for side chain contacts and green for main chain contacts. (C) S. pombe strains with genotypes as specified on the left were grown in liquid culture at 30 °C. Serial fivefold dilutions were spotted to YES agar and incubated at the indicated temperatures. As noted below the panels, seb1-G476S was lethal in combination with CTD alleles S7A, S5•S5A, and P6•P6A.

Paired-end Illumina sequencing of unamplified genomic DNA from the STF19 strain was performed as described (22) to achieve at least 100-fold coverage of the yeast genome. Compared to the genome of the parental T4A strain, the STF19 strain had acquired a missense mutation—Gly476Ser—in the gene encoding the essential transcription termination factor Seb1.

The G476S Mutation in Seb1.

Seb1 was originally identified in a yeast two-hybrid screen for proteins that interact with the Rpb7 subunit of fission yeast Pol2 (26). It was recognized as a homolog of the budding yeast RNA-binding protein Nrd1 and was shown to be essential for vegetative growth of fission yeast (26). The 620-aa Seb1 protein consists of a CTD interaction domain (CID; amino acids 1 to 152) and an RNA recognition module (RRM) RNA-binding domain (amino acids 388 to 540) (Fig. 1B), the structures of which have been determined (21). The central and carboxyl-terminal segments of Seb1 are predicted to be structurally disordered. The CID and RRM domain are both required for cell viability (21). The CID binds in vitro to a Ser2 phosphorylated di-heptad CTD repeat peptide with threefold higher affinity versus a Ser5 phosphorylated CTD peptide, 10-fold higher affinity versus a Ser7 phosphorylated CTD, and >40-fold higher affinity compared to an unphosphorylated CTD peptide (21). Affinity for the Thr4 phosphorylated CTD ligand was within a factor of two of that for the Ser2-PO₄ CTD, while binding to a Tyr1-PO₄ CTD peptide was similar to the unphosphorylated CTD (27). The evidence that Seb1 is an agent of 3′-processing/termination is compelling. To wit: 1) Seb1 is physically associated in vivo with the CPF complex, Rhn1, and the torpedo 5′ exoribonuclease Dhp1; 2) Seb1 is present and enriched at the 3′ end of Pol2 transcription units; 3) depletion of Seb1 or replacement of depleted Seb1 with a quadruple-alanine mutant in the RRM domain (R442A-H443A-F445A-K447A) results in Pol2 transcription read-through (RT) past normal termination sites; and 4) Seb1 depletion or substitution with the 4xAla RRM mutation favors utilization of distal poly(A) sites (21, 28).

The salient finding here is that the STF19 mutation G476S maps to the RRM domain of Seb1 (Fig. 1B). Genome-wide in vivo protein–RNA cross-linking methods have established that Seb1 recognizes a 5′-GUA trinucleotide sequence motif situated 50 to 100 nt downstream of nascent RNA cleavage/poly(A) sites (21, 28). The Seb1 RNA recognition element resembles the 5′-GUA(A/G) motif recognized by budding yeast Nrd1 (29). A primary structure alignment of the Seb1 and Nrd1 RRM domains highlights 105 positions of side chain identity/similarity (Fig. 1B). Crystal structures have been determined for the Seb1 RRM in the absence of RNA and for the Nrd1 RRM in complexes with GUA-containing RNA ligands (21, 29). The five amino acid side chains of Nrd1 RRM that make atomic contacts with the RNA (indicated by red arrowheads in Fig. 1B) are conserved perfectly in the Seb1 RRM (as Phe409, His443, Phe445, Arg472, and Arg482). Two Nrd1 amino acids within a conserved Trp-Gly-Val peptide that receive hydrogen bonds to their main chain carbonyls from the uracil-N3 (to Trp) and the adenine-N6 (to Val) of the GUA RNA element are denoted by green arrowheads in Fig. 1B. The Gly476 position that is mutated in STF19 is located within this Trp-Gly-Val motif of the RRM•RNA interface.

Structure of the Mutant Seb1-G476S RRM Domain.

Here, we produced, purified, and crystallized the RRM domain of the Seb1-G476S protein. The structure was solved at 1.4-Å resolution and refined to R/R_free of 17.1/18.8 (SI Appendix, Table S1). The Ser476 RRM domain structure is tightly superimposable on that of the wild-type Seb1 RRM, except that the Ser476 mutation elicits a 2.0- to 2.5-Å shift in the position of the 11-aa loop from Val477 to Ser485 (Fig. 2A). Attempts to cocrystallize the Seb1-G476S RRM in complex with a variety of RNA oligonucleotides containing a 5′-GUA triplet were uniformly unsuccessful. Broad crystal screens starting with a mixture of RNA and RNA ligand were undertaken, and multiple conditions were identified that did yield high-resolution diffracting crystals. However, in every case, they turned out to be in the same space group and lattice as the unliganded RRM crystals, and they were devoid of bound RNA. Inspection of these crystal lattices and superposition of the RNA-bound structure of the homologous Nrd1 RRM revealed that the tight contacts between Seb1 RRM and two symmetry-related Seb1 RRM protomers (mediated by many interprotomer pairwise side-chain hydrogen bonds and salt bridges involving Arg482-Thr509, Arg472-Asp501, Arg415-Glu455, Asn440-Glu455, and Tyr441-Lys402) would definitely occlude the RNA-binding site in Seb1 RRM. Nonetheless, some insight into the consequences of the G476S mutation could be gleaned by importing the 5′-GUA ligand from the superimposed Nrd1•RNA structure into the structure of the Seb1-G476S RRM, which suggests that the installation of Ser476 establishes new atomic contacts of Ser476-Oγ to uracil-O2 (3.7 Å), adenine-N6 (3.7 Å), and adenine-N7 (3.9 Å) moieties of the RNA ligand (Fig. 2B).

Fig. 2.

Structure of the G476S mutant version of the Seb1 RRM domain. (A) Stereo view of the tertiary structure of the G476S mutant RRM domain (colored cyan) superimposed on that of the wild-type RRM (colored gray; from Protein Data Bank [PDB] 5MDU). Mutated amino acid Ser476 is shown as a side chain stick model. The mutation elicits a 2.0- to 2.5-Å shift in the position of the 11-aa loop from Val477 to Ser485. Selected C-α atoms in the dislocated loop are connected by magenta dashed lines. (B) Stereo view of a model of Seb1-G476S RRM with a 5′-GUA ligand in the RNA-binding site. The 5′-GUA ligand (stick model with gray carbons) is from the homologous Nrd1 RRM•RNA complex structure (PDB 5O1Z) that had been superimposed on the Seb1 RRM. The putative atomic contacts of Ser476-Oγ to uracil-O2 (3.7 Å), adenine-N6 (3.7 Å), and adenine-N7 (3.9 Å) are denoted by magenta dashed lines.

Effect of the seb1-G476S Mutation in a Wild-Type rpb1-CTD Background.

We crossed the seb1-G476S T4A strain, in which the rpb1-CTD-T4A locus was flanked by a downstream nourseothricin-resistance marker, with a wild-type rpb1-CTD strain in which the rpb1⁺ allele was flanked by a G418-resistance marker. The G418/nourseothricin-resistant diploids were sporulated, and random populations of >1,000 haploid progeny were selected for the resistance markers linked to the rpb1-CTD-WT and rpb1-CTD-T4A alleles. Populations of CTD-WT cells were screened for Pho1 expression by agar overlay, with deep red color indicating presence of the seb1-G476S mutation. About half of the CTD-WT progeny backcrosses were lightly stained red (i.e., reflecting the fact that the basal level of “repressed” Pho1 activity in CTD-WT cells is several-fold higher than the hyperrepressed Pho1 state in T4A cells; Fig. 1A), and half were darker red. Sequencing of the seb1 locus in individual darker red CTD-WT haploids affirmed the presence of the seb1-G476S mutation. Note that the seb1-G476S allele per se conferred a slow growth phenotype at low temperatures vis-à-vis the wild-type strain, as gauged by colony size (Fig. 1C).

The seb1-G476S CTD-WT and seb1⁺ CTD-WT strains were grown in phosphate-replete medium at 30 °C and assayed for Pho1 acid phosphatase activity. The seb1-G476S allele increased Pho1 expression by 11-fold vis-à-vis the seb1⁺ CTD-WT control (Fig. 1A). Pho1 expression in seb1-G476S cells was sixfold higher in the CTD-WT background than in the CTD-T4A background, in keeping with the sixfold differential between the basal Pho1 activity in seb1⁺ CTD-WT versus seb1⁺ CTD-T4A cells (Fig. 1A).

To profile the effects of the seb1-G476S mutation on the fission yeast transcriptome, we performed RNA sequencing (RNA-seq) on poly(A)⁺ RNA isolated from seb1-G476S cells and a wild-type seb1⁺ control strain, both of which were wild-type with respect to the Pol2 CTD. Complementary DNAs obtained from three biological replicates (using RNA from cells grown to midlog phase in YES medium at 30 °C) were sequenced for each strain. In the datasets, 95 to 97% of the reads were mapped to Schizosaccharomyces pombe genomic loci (SI Appendix, Table S2). Read densities for individual genes were highly reproducible between biological replicates (Pearson coefficients of 0.98 to 0.99) (SI Appendix, Table S3). A cutoff of ±2-fold change in normalized transcript read level and an adjusted P value of ≤0.05 were the criteria applied to derive an initial list of differentially expressed annotated loci. We then focused on differentially expressed genes with average normalized read counts ≥100 in either the seb1-G476S or seb1⁺ strains in order to eliminate transcripts that were expressed at very low levels in vegetative cells. We thereby identified sets of 107 and 97 annotated protein-coding genes that were respectively up-regulated and down-regulated by these criteria in seb1-G476S cells. The up-regulated gene set included tgp1 (up ninefold) and pho1 (up fourfold). RNA-seq revealed no increase in seb1-G476S cells of the mRNA encoding Pho7, the transcription factor that drives pho1 and tgp1 mRNA synthesis.

seb1-G476S Genetic Interactions with CTD Mutations.

We crossed the seb1-G476S CTD-WT strain with strains bearing differentially marked rpb1-CTD alleles lacking the three other inessential CTD phospho-sites: Y1F, in which Tyr1 in all heptad repeats was replaced by Phe; S2A, in which all Ser2 positions were changed to Ala; and S7A, in which all Ser7 residues were replaced by Ala. Whereas the Y1F and S2A mutations do not affect Pho1 expression, the S7A allele results in Pho1 derepression dependent on CPF and Rhn1 (12). We also crossed seb1-G476S to strains with chimeric mutations of the three essential letters of the CTD code: the P3•P3A, S5•S5A, and P6•P6A mutant CTDs in which Pro3, Ser5, or Pro6 in every other repeat was replaced by alanine. The S5•S5A, and P6•P6A alleles derepress Pho1 under phosphate-replete conditions (13, 30). Random populations of haploid progeny were selected for the mutant CTD allele, assayed for Pho1 activity, and screened for the seb1-G476S mutation by sequencing of the seb1 locus (as described in SI Appendix, Methods). We readily obtained seb1-G476S CTD-Y1F, seb1-G476S CTD-S2A, and seb1-G476S CTD-P3•P3A double mutants at the expected frequency, and they grew well on YES agar at 30 °C (Fig. 1C). The seb1-G476S CTD-S2A strain displayed a slow growth defect at 37 °C, as seen previously for the CTD-S2A single mutant (12). By contrast, we did not recover any viable double mutants of seb1-G476S with CTD alleles S7A, S5•S5A, or P6•P6A.

To verify this result, we replaced the seb1 locus in a wild-type strain with seb1⁺ or seb1-G476S alleles marked with a 3′ flanking hygMX drug resistance cassette, which enabled us to screen a large population of random spores from the pairwise mating of seb1-G476S to the CTD-S7A, S5•S5A, and P6•P6A strains. Failure to recover any viable double mutants signified that the seb1-G476S mutation is synthetically lethal with CTD mutations S7A, S5•S5A, and P6•P6A.

Derepression of pho1 Expression by seb1-G476S Depends on CPF Subunits and Rhn1.

To test the idea that seb1-G476S elicits a gain of Seb1 function that leads to precocious termination during prt lncRNA synthesis, we performed pairwise mating of seb1-G476S cells with knockout strains lacking the Dis2, Ctf1, Ppn1, or Swd22 subunits of the CPF complex, a strain with a catalytically dead (C13S) version of the Ssu72 protein phosphatase subunit of CPF, and a strain that lacks the transcription termination factor Rhn1. Viable double-mutant haploids were recovered after sporulation of the diploids in each case. Spot tests of the double mutants and parental single mutants for growth on YES agar are shown in Fig. 3A. seb1-G476S displayed no synthetic growth defects with ctf1∆ or dis2∆. Notable findings were that seb1-G476S rescued the ts growth defect of rhn1∆ at 37 °C, while rhn1∆ rescued the cs growth defect of seb1-G476S at 20 °C (Fig. 3A), consistent with a yin-yang relationship between gain and loss of termination function in the seb1-G476S and rhn1∆ genetic backgrounds. seb1-G476S also improved the growth of ssu72-C13S cells at 37 °C (Fig. 3A). The cs phenotype of seb1-G476S was exacerbated at 25 °C by ppn1∆ and swd22∆.

Fig. 3.

Derepression of pho1 expression by seb1-G476S depends on CPF subunits and Rhn1. (A) S. pombe strains with genotypes as specified on the left were spot tested for growth on YES agar at the indicated temperatures. (B) Strains bearing the indicated seb1 alleles (wild-type or G476S) in combination with CPF subunit or Rhn1 mutations as specified were grown in liquid culture at 30 °C and assayed for acid phosphatase activity.

The single and double mutants were assayed for acid phosphatase activity during growth in phosphate-replete medium at 30 °C. The instructive findings were that the derepression of Pho1 by seb1-G476S was effaced in ppn1∆, swd22∆, and rhn1∆ cells and was severely attenuated in ssu72-C13S, ctf1∆, and dis2∆ cells (Fig. 3B). Thus, the increase in Pho1 expression in seb1-G476S cells requires CPF subunits and Rhn1, consistent with the Seb1 gain-of-function hypothesis.

Derepression of pho1 and tgp1 by seb1-G476S Depends on Upstream lncRNA Poly(A) Signals.

Plasmid-based reporter systems have been developed that recapitulate the salient features of lncRNA-mediated transcriptional interference with the pho1 and tgp1 promoters (10, 11). The prt–pho1 reporter cassette contains the tandem prt and pho1 genes spanning from 1,831 nucleotides upstream of the pho1 ORF (comprising the prt promoter and prt lncRNA) to 647 nucleotides downstream of the pho1 ORF (Fig. 4A). The prt transcription start site is located 1,147 nucleotides upstream of the pho1 transcription start site. The prt lncRNA that interferes with pho1 expression is a long RT transcript that initiates from the prt promoter, traverses the Pho7 transcription factor binding sites in the pho1 mRNA promoter, and terminates at the pho1 mRNA poly(A) site (Fig. 4A). There are two classes of short prt lncRNAs (Fig. 4A) that are cleaved and polyadenylated at sites +351 and +589 in response to two different upstream poly(A) signals (PAS). Enhancement of nascent RNA processing via the short prt lncRNA PAS elements will terminate Pol2 transcription upstream of the Pho7 binding sites in the pho1 mRNA promoter and thus alleviate transcription interference. Conversely, simultaneous nucleobase substitution mutations of the two short prt PAS elements of the prt–pho1 reporter results in hyperrepression of the flanking pho1 promoter (12). Here, we deployed the reporter system to interrogate the role of prt lncRNA 3′-cleavage/polyadenylation in pho1 derepression by seb1-G476S. prt–pho1 plasmids with wild-type or mutant PAS elements were introduced into [prt2-pho84-prt-pho1]∆ cells bearing WT or G476S seb1 alleles. Transformants were assayed for acid phosphatase activity during growth in phosphate-replete medium. Pho1 expression from the wild-type plasmid was increased sevenfold in seb1-G476S cells versus seb1-WT cells (Fig. 4B), thereby echoing the derepressive effect of seb1-G476S on pho1 expression from the chromosomal prt–pho1 locus. The crucial finding was that the derepressed level of Pho1 expression in seb1-G476S cells was strongly attenuated (10-fold) by dual PAS mutations that interdict precocious prt lncRNA 3′-processing at the two promoter-proximal sites.

Fig. 4.

Derepression of pho1 and tgp1 by seb1-G476S depends on upstream lncRNA PASs. (A) Schematic of the prt–pho1 locus in the reporter plasmid. Transcription start sites are indicated by bent blue arrows with the intervening distance indicated by the bracket. The prt lncRNAs are depicted as red wavy lines. The pho1 mRNA is a blue wavy line. (B) prt–pho1 reporter plasmids with wild-type (WT) or mutated (Mut) prt 5′-proximal polyadenylation signals PAS and PAS2 were transfected into seb1-WT pho1Δ or seb1-G476S pho1Δ cells. Plasmid-bearing cells were grown in liquid culture at 30 °C and assayed for acid phosphatase activity. (C) Schematic of the nc-tgp1–tgp1(promoter)•pho1 locus in the reporter plasmid. Transcription start sites are indicated by bent blue arrows with the intervening distance indicated by the bracket. The nc-tgp1 lncRNAs are depicted as red wavy lines. The tgp1 promoter–driven mRNA is a blue wavy line. (D) nc-tgp1–tgp1(promoter)•pho1 reporter plasmids with a WT or Mut nc-tgp1 5′-proximal polyadenylation signal were transfected into seb1-WT pho1Δ or seb1-G476S pho1Δ cells. Plasmid-bearing cells were grown in liquid culture at 30 °C and assayed for acid phosphatase activity. (E) Northern blot analysis of nc-tgp1 RNA. RNA was isolated from three independent cultures of seb1-WT pho1Δ or seb1-G476S pho1Δ cells bearing the wild-type nc-tgp1–tgp1(promoter)•pho1 reporter plasmid. The RNAs were resolved by formaldehyde agarose gel electrophoresis. The gel was stained with ethidium bromide (shown in right panel, with sizes of the stained ribosomal RNAs and transfer RNAs indicated in nucleotides) prior to transferring the resolved RNAs to a membrane, which was then hybridized to a ³²P-labeled probe complementary to nc-tgp1 nucleotides 70 to 108. An autoradiograph of the Northern blot is shown in the left panel. The species corresponding to nc-tgp1 short and full-length nc-tgp1 lncRNA are indicated.

The nc-tgp1–tgp1 reporter cassette (nc-tgp1–tgp1_prom⋅pho1) includes a fragment of the nc-tgp1–tgp1 locus, spanning 301 nucleotides upstream of the nc-tgp1 transcription initiation site (encompassing the putative nc-tgp1 promoter) and the entire 1,865-nt segment between the nc-tgp1 transcription start site and the tgp1 translation start codon (containing the tgp1 promoter and tgp1 transcription start site) fused to the pho1 ORF and its native 3′ flanking DNA (Fig. 4C). The nc-tgp1 transcription start site is located 1,823 nt upstream of the tgp1 transcription start site. The nc-tgp1 transcript that interferes with tgp1 expression is cleaved and polyadenylated at nucleotide +1636 of the nc-tgp1 transcription unit, which is located within the DNA-binding site for Pho7, the transcription factor that drives tgp1 expression (11). A short form of the nc-tgp1 lncRNA is cleaved/polyadenylated at position +504 of the nc-tgp1 transcription unit, which is 17 nt downstream of a consensus fission yeast PAS. Mutation of the PAS for the short nc-tgp1 lncRNA in the nc-tgp1–tgp1 reporter abolished derepression of the downstream tgp1 mRNA promoter in a CTD-S5A strain background (11). Here, we found that tgp1 promoter–driven acid phosphatase expression was increased 30-fold in seb1-G476S cells versus seb1-WT cells and that this derepression was effaced by mutating the promoter-proximal nc-tgp1 PAS (Fig. 4D).

RNA Analysis Confirms that seb1-G476S Elicits Precocious lncRNA Termination.

We performed Northern blot analysis of RNAs isolated from three independent cultures of seb1-WT and seb1-G476S cells bearing the nc-tgp1–tgp1 reporter plasmid (Fig. 4E). Probing the blot with a radiolabeled oligonucleotide complementary to the 5′ end of the nc-tgp1 transcript, we found that the steady-state level of the nc-tgp1-short lncRNA was increased markedly in seb1-G476S cells vis-à-vis the wild-type controls (Fig. 4E). We conclude that the G476S mutation enhanced utilization of the 5′-proximal poly(A) site and thus precocious termination of nc-tgp1 lncRNA transcription, thereby accounting for increased expression of the downstream flanking mRNA. We also conducted Northern analysis of the prt transcript from three independent cultures of seb1-WT and seb1-G476S cells bearing the prt–pho1 reporter (Fig. 5). A 5′ prt probe detected the interfering prt-pho1 RT transcript as the major lncRNA species in wild-type cells (Fig. 5B). Formation of the prt-pho1 RT transcript was almost completely squelched in seb1-G476S cells. Yet it was not the case that the decrement in the long prt-pho1 RT transcript in seb1-G476S cells was accompanied by an increase in the steady-state levels of the short prt PAS and prt PAS2 RNAs. prt transcripts can be challenging to detect because they are turned over rapidly via the two clusters of DSR (determinant of selective removal) elements encoded within the proximal segment of the transcript (the DSRs are denoted by the blue boxes in the cartoon diagram in Fig. 5A) (10). Therefore, we tested a version of the prt–pho1 reporter in which the consensus DSR sequences in the two DSR clusters are altered by base substitutions (10). A prt-probed Northern blot of RNAs isolated from two independent cultures of seb1-WT and seb1-G476S cells bearing the mutated DSR prt–pho1 reporter revealed that the seb1-G476S allele strongly skewed the distribution of prt lncRNAs toward production of the precociously terminated prt PAS2 species, at the expense of the prt-pho1 RT transcript (Fig. 5B). Quantification of the signal intensity of these two prt transcripts yielded PAS2 to RT ratios of 0.16 and 0.17 for two RNA samples from seb1-WT cells and 1.49 and 1.64 for the RNAs from seb1-G476S cells (i.e., there was an average 9.5-fold increase in apparent precocious termination directed by PAS2).

Fig. 5.

seb1-G476S elicits precocious prt lncRNA termination. (A) Schematic of the prt–pho1 locus in the reporter plasmid. Transcription start sites are indicated by bent blue arrows. Two DSR element clusters in the prt transcription unit are denoted by small blue boxes. The 5′-proximal PAS and PAS2 polyadenylation signals are denoted by triangles. The prt lncRNAs are depicted as red wavy lines. A ³²P-labeled probe complementary to prt nucleotides 160 to 202 is denoted by a black bar. The pho1 mRNA is a blue wavy line. (B) Northern blot analyses of prt lncRNA. RNA was isolated from three independent cultures of seb1-WT pho1Δ and seb1-G476S pho1Δ cells bearing the wild-type prt–pho1 reporter plasmid or from two independent cultures of seb1-WT pho1Δ and seb1-G476S pho1Δ cells bearing a prt–pho1 reporter plasmid in which both of the DSR clusters were inactivated by multiple base-substitution mutations. Photographs of the ethidium bromide–stained gels are shown in the right panels. The annealed ³²P-labeled prt probe was detected by autoradiography of the Northern blot membrane, as shown in the left panels. The species corresponding to prt PAS2 and prt–pho1 RT lncRNAs are indicated.

The results of the experiments in Figs. 3–5 collectively fortify the case for seb1-G476S as a gain-of-function mutation in Seb1 that elicits precocious lncRNA termination dependent on lncRNA PAS and cleavage/polyadenylation factors.

Interface of seb1-G476S and Inositol Pyrophosphate Metabolism.

Previous studies showed that derepression of Pho1 elicited by the Pol2 CTD-S7A mutation or a deletion of Erh1 depends on the activity of the IPP kinase Asp1 (14, 25). Asp1 is a bifunctional enzyme composed of an N-terminal IPP kinase domain that converts 5-IP7 to 1,5-IP8 and a carboxyl-terminal IPP pyrophosphatase domain that converts 1,5-IP8 back to 5-IP7 (23, 24) (Fig. 6C). The in vivo effect of an asp1∆ null allele or a kinase-dead asp1-D333A allele is to eliminate intracellular IP8 and 1-IP7 and to increase the level of 5-IP7 (23). To see if IPP status affects Pho1 derepression by the Seb1-G476S mutation, we crossed hygMX-marked seb1-G476S to asp1∆ and asp1-D333A strains. asp1∆ and asp1-D333A cells themselves display a cs growth phenotype (25). Here, we recovered viable seb1-G476S asp1∆ and seb1-G476S asp1-D333A haploid progeny that grew more slowly than wild-type at 30 °C (as gauged by colony size) but were very sick at 25 °C and 20 °C and failed to grow at 37 °C (Fig. 6A). Assays of cell surface acid phosphatase activity of cells grown at 30 °C showed that the derepression of Pho1 activity by seb1-G476S was erased in the asp1∆ and asp1-D333A backgrounds (Fig. 6B), signifying a requirement for 1-IPP synthesis and an agonist role for 1-IPPs in Seb1-mediated 3′-processing/termination during prt transcription.

Fig. 6.

Interface of seb1-G476S and IPP metabolism. (A) seb1⁺ asp1⁺ cells (WT) and cells harboring the seb1-G476S::hygMX allele in combination with mutants asp1∆, asp1-D333A (kinase-defective), or asp1-H397A (pyrophosphatase-defective) were spot tested for growth on YES agar at the indicated temperatures. seb1-G476S was lethal in the absence the Nudix-family IPP pyrophosphatase enzyme Aps1. (B) Strains bearing the indicated seb1 and asp1 alleles were grown in liquid culture at 30 °C and assayed for acid phosphatase activity. (C) Scheme of IPP metabolism by Asp1 and Aps1. Structures of 5-IP7, 1,5-IP8, and 1-IP7 are shown. Asp1 kinase converts 5-IP7 to IP8, and the Asp1 pyrophosphatase reverses this process. Aps1 pyrophosphatase converts IP8 to 1-IP7.

The fission yeast proteome includes two IPP pyrophosphatase enzymes that convert IP8 to IP7 and have demonstrated roles in repression of the PHO genes in a manner dependent on CPF and Rhn1 (25). The carboxyl-terminal IPP pyrophosphatase domain of Asp1, which belongs to the histidine acid phosphatase family, hydrolyzes the 1-pyrophosphate of IP8 to generate 5-IP7 (23, 24) (Fig. 6C). Aps1, a Nudix family enzyme, acts on the 5-pyrophosphate of IP8 to generate 1-IP7 (31, 32) (Fig. 6C). Neither pyrophosphatase activity is essential per se for fission yeast vegetative growth, i.e., the pyrophosphatase-defective asp1-H397A strain and the aps1∆ null strain grow well on YES agar at 20 to 37 °C (25). We queried epistasis of seb1-G476S with the two IPP pyrophosphatases with respect to cell growth and Pho1 expression by attempting to construct seb1-G476S asp1-H397A and seb1-G476S aps1∆ double-mutant strains via mating and sporulation. We recovered (by random spore analysis) seb1-G476S asp1-H397A progeny that grew very poorly on YES agar at all temperatures tested (Fig. 6A). The seb1-G476S and aps1∆ alleles were synthetically lethal; to wit: 1) we were unable to obtain viable double mutants after screening a large population of haploid progeny of the genetic cross; and 2) wild-type progeny and the differentially marked seb1-G476S and aps1∆ single mutants were recovered at the expected frequencies. Although seb1-G476S asp1-H397A cells grew slowly in liquid medium at 30 °C, an assay of acid phosphatase showed that the double mutant expressed threefold higher levels of Pho1 than the seb1-G476S single mutant (Fig. 6B). Pho1 expression in seb1-G476S asp1-H397A cells was ∼70% higher than that in the asp1-H397A single mutant (25), indicative of an additive effect of the Seb1-G476S and Asp1 pyrophosphatase-dead mutations on pho1 derepression.

A comparison of the presently reported RNA-seq data for seb1-G476S to the transcriptome profiles of the aps1∆ and asp1-H397A strains reported previously (25) revealed eight protein-coding RNAs that were coordinately up-regulated by >2-fold in all three genetic backgrounds (SI Appendix, Fig. S1). This set included the following: the PHO genes tgp1 and pho1; ecl3 (located adjacent to the pho84–pho1 gene cluster and coregulated with the PHO genes; 13, 22, 30); the putative phosphate acquisition gene SPAC1039.02 (encoding an extracellular 5′-nucleotidase); mbx2 (a gene likely to be subject to flanking lncRNA-mediated transcriptional interference; 25); the uracil-regulatable genes urg1 and urg2; and SPBC25B2.08 encoding a predicted polypeptide unique to S. pombe.

Rescue of seb1-G476S Synthetic Lethality by CPF/Rhn1 Mutations.

In light of the findings that the seb1-G476S mutation, which derepresses pho1, was inviable in combination with several other mutations that derepress pho1, we queried whether such synthetic lethality could be suppressed by mutations in the 3′-processing/termination machinery (i.e., CPF subunits and Rhn1). Specifically, we performed pairwise crosses of seb1-G476S and double mutants aps1∆ ctf1∆, aps1∆ ssu72-C13S, aps1∆ rhn1∆, CTD-S7A ctf1∆, or CTD-S7A ssu72-C13S and screened haploid progeny for the drug-resistance markers linked to the seb1-G476S, aps1∆, CTD-S7A, and CPF/rhn1 loci. We thereby recovered the following viable triple mutants: seb1-G476S aps1∆ ctf1∆, seb1-G476S aps1∆ ssu72-C13S, seb1-G476S aps1∆ rhn1∆, seb1-G476S CTD-S7A ctf1∆, and seb1-G476S CTD-S7A ssu72-C13S. The triple mutants were spot tested for growth on YES agar (Fig. 7A). The otherwise lethal seb1-G476S aps1∆ combination ameliorated the ts growth defect of rhn1∆ at 37 °C while conferring a strong cs phenotype at 20 and 25 °C (Fig. 7A). Rescue of the synthetic lethality of seb1-G476S aps1∆ and seb1-G476S CTD-S7A by CPF or Rhn1 loss-of-function mutations leads us to conclude that the synthetic lethality is a consequence of overzealous precocious termination of transcription of one or more fission genes that are essential for vegetative growth. The rescued triple mutants were assayed for Pho1 expression during growth in phosphate-replete medium (Fig. 7B). The level of Pho1 activity was higher in the seb1-G476S aps1∆ ctf1∆ cells than in either the seb1-G476S or aps1∆ single mutants. seb1-G476S aps1∆ ssu72-C13S and seb1-G476S aps1∆ rhn1∆ cells expressed Pho1 at a level slightly lower than the aps1∆ single mutant. The seb1-G476S CTD-S7A ctf1∆ strain had lower Pho1 activity than the CTD-S7A single mutant but still higher than wild-type cells. Pho1 was strongly hyperrepressed in seb1-G476S CTD-S7A ssu72-C13S cells to well below the wild-type level (Fig. 7B).

Fig. 7.

Rescue of seb1-G476S synthetic lethality by CPF/Rhn1 mutations. (A) Strains with genotypes specified on the left were spot tested for growth on YES agar at the indicated temperatures. The seb1-WT and seb1-G476S alleles were marked with a 3′-flanking hygMX cassette. (B) Strains with genotypes specified on the x-axis were grown in liquid culture at 30 °C and assayed for acid phosphatase activity.

seb1-G476S Elicits an Upstream Shift in Poly(A) Site Utilization in Certain mRNAs.

To explore the idea that seb1-G476S might lead to precocious 3′-processing of fission yeast mRNAs, we mined the pooled RNA-seq data from three biological replicates to identify mRNA poly(A) sites. In brief, the unmapped strand-specific RNA-seq reads were trimmed of 5′-oligo(T) tracts and then remapped to the S. pombe genome to reveal narrow peaks of sequence reads with an acute falloff at sites of cleavage and 3′ poly(A) addition. This approach accurately identified the single poly(A) site in the adh1 gene and the three poly(A) sites in the act1 gene (SI Appendix, Fig. S2) that had been mapped previously by using an RNA-seq method that focused specifically on isolated 3′-terminal fragments of poly(A)⁺ RNAs (33). The adh1 and act1 poly(A) sites were reproducible when comparing the data from the individual biological replicates (SI Appendix, Fig. S2). Comparison of the wild-type and seb1-G476S datasets highlighted poly(A) sites that were differentially utilized in seb1-G476S cells vis-à-vis wild-type cells. These entailed either 1) a switch in poly(A) site preference toward a proximal site within the mRNA transcription unit, manifest as an increased ratio of proximal to distal site utilization; or 2) cleavage at a new upstream poly(A) site in seb1-G476S cells that was not detectably utilized in wild-type cells.

For example, two distinct seb1-G476S–dependent “neo” poly(A) sites were present in the second intron of the dbp2 gene that were not evident in wild-type controls (Fig. 8A). Because there are very few total RNA-seq reads derived from the second intron in wild-type cells, the switch to the dbp2 intronic sites in seb1-G476S cells is readily appreciated in the total RNA-seq data, which shows that intronic sequences 3′ of exon 2 and preceding the neo poly(A) sites are present in the precociously terminated dbp2 transcripts (Fig. 8A). Inspection of the poly(A)-adjacent reads in wild-type and seb1-G476S cells pinpointed three closely clustered cleavage/polyadenylation sites in the 3′-UTR, preceded by an AAUAAA PAS (Fig. 8 B, Top). By contrast, the intronic neo poly(A) sites utilized in seb1-G476S cells are not preceded by any of the hexanucleotide PASs that have been reported in fission yeast (33, 34). Whereas five GUA Seb1 RNA-binding motifs are located within the 80-nt region downstream of the poly(A) site cluster in the 3′-UTR, only two GUA sites are located in the 85-nt region downstream of the second intronic neo poly(A) site (Fig. 8B).

Fig. 8.

seb1-G476S triggers utilization of an upstream intronic poly(A) site in the dbp2 transcript. (A) Strand-specific RNA-seq read densities (Top; counts/base/million averaged in a 25-nt window) or poly(A) site-adjacent reads (Bottom; counts/base/million) of wild-type or seb1-G476S cells are plotted as a function of position across the dbp2 chromosomal locus, which is depicted above the graphs with the three exons comprising the open reading frame denoted by thick bars, the two introns by thin lines, and the flanking UTRs by thin bars. The exon–intron boundaries in the RNA-seq profile are indicated by vertical red lines. The x-axis scale is shown on the bottom left. The RNA-seq read densities and poly(A) sites were determined from cumulative counts of three RNA-seq replicates. (B, Top) The DNA nucleotide sequence of the 3′-UTR is shown, starting from the dbp2 TGA stop codon (underlined). Inspection of the poly(A) adjacent reads pinpointed three closely clustered cleavage/polyadenylation sites in the 3′-UTR, denoted in bold font and green shading. The poly(A) site cluster in the 3′-UTR is preceded by an AATAAA PAS (bold red font, underlined). The DNA sequence downstream of the distal poly(A) site is shown in italics and includes five GTA Seb1-binding motifs (shaded gold). (Bottom) Shown is the DNA sequence of the segment of the second dbp2 intron surrounding the two neo poly(A) sites observed in seb1-G476S cells, neither of which is preceded by a hexanucleotide PAS. The DNA sequence downstream of the second intronic poly(A) site is shown in italics and includes two GTA Seb1-binding motifs (shaded gold). In both panels, the nucleotides adjacent to the poly(A) tails are indicated by a red triangle above the DNA sequence. Because the junction nucleotide is followed by one or two A nucleotides, the exact 3′-cleavage site within the green shaded elements is uncertain.

SI Appendix, Fig. S3 depicts 10 additional mRNAs in which the seb1-G476S allele elicits a shift in poly(A) site usage toward upstream sites within an mRNA 3′-UTR or the appearance of an upstream neo poly(A) site in the 3′-UTR. In some cases, the upstream shift in poly(A) site choice is appreciable in the total RNA-seq reads. For instance, wild-type cells make apparently equal use of two poly(A) sites in the gms1 3′-UTR, whereas seb1-G476S cells favor the use of a further upstream poly(A) site within the gms1 3′-UTR, apparently at the expense of the most distal site used in wild-type cells, which is reflected in a decrease in the RNA-seq read density over the distal 3′-UTR segment in seb1-G476S cells (SI Appendix, Fig. S3). Two upstream neo poly(A) sites in the hhp1 3′-UTR were detected in seb1-G476S cells, leading to a relative decrease in utilization of the distal poly(A) site predominant in wild-type cells and a relative decrease in RNA-seq reads over the distal hhp1 3′-UTR (SI Appendix, Fig. S3). Similar patterns of upstream shifts in poly(A) site choice and accompanying changes in distal 3′-UTR read density in seb1-G476S cells were observed for the SPCC14G10.04, ado1, and pcl1 mRNAs (SI Appendix, Fig. S3).

A Hypomorphic Loss-of-Function Seb1 Mutation Elicits Synergies Distinct from those of seb1-G476S.

The seb1-1 mutant (the first reported viable allele of seb1) was isolated from a mutagenized gene library by screening for a defect in RNA interference (RNAi)-independent pericentromeric heterochromatic silencing of a reporter gene system (35). The seb1-1 allele has seven nucleotide mutations in the coding sequence, comprising three missense changes (G76S, R442G, and I524V) and four changes that did not alter the protein sequence; notably, the coding and noncoding changes were both necessary to attain the full seb1-1 silencing defect and to confer a ts growth defect at 37 °C (35). Ensuing studies suggested the following: 1) Seb1 promotes Pol2 pausing that triggers heterochromatin formation; and 2) the seb1-1 allele diminishes this activity of Seb1 (36). In effect, seb1-1 behaves as a partial loss-of-function hypomorph, in which case the genetic interactions of seb1-1 ought to be distinct from those of seb1-G476S, a seb1 allele that we envision, based on the present studies, elicits a gain-of-function.

To address this issue, we crossed a seb1-1 strain with the various rpb1-CTD and CPF/Rhn1 mutants that had been tested for mutational synergies with seb1-G476S. With respect to the CTD mutants, seb1-1 resembled seb1-G476S in that it was synthetically lethal with the CTD-(S5•S5A) and CTD-(P6•P6A) chimeras. However, in contrast to seb1-G476S, seb1-1 was distinctively lethal in combination with CTD-S2A, barely viable in tandem with CTD-Y1F, sick at all temperatures when combined with CTD-T4A, and very sick at high temperatures in the CTD-(P3•P3A) background (SI Appendix, Fig. S4). Mutational synergies of seb1-1 with Y1F, S2A, and T4A resonate with genetic evidence that Tyr1-Ser2-Thr4 form a three-letter CTD “word” that abets termination (12). It was of particular interest that the CTD-S7A allele overcame the ts phenotype of seb1-1 and restored growth at 37 °C, while the seb1-1 allele partially rescued the cs defect of CTD-S7A (SI Appendix, Fig. S4), i.e., a hypomorphic 3′-processing/termination defect of seb1-1, which becomes growth limiting at higher temperature, is countered by the precocious termination phenotype imputed to CTD-S7A.

In its genetic interactions with CPF and Rhn1, seb1-1 differed from seb1-G476S in that seb1-1 was synthetically lethal with ctf1∆ and ssu72-C13S (SI Appendix, Fig. S4). Combining seb1-1 with swd22∆ and ppn1∆ resulted in failure to grow at 20 °C and 25 °C; seb1-1 dis2∆ cells displayed a less severe cs phenotype (SI Appendix, Fig. S4). We surmise that the residual function of the Seb1-1 mutant protein was more acutely dependent on the Ctf1 and Ssu72 components of the CPF core complex than on the three subunits that comprise the DPS module of the CPF holo-assembly. There was no apparent mutational synergy between seb1-1 and rhn1∆ with respect to growth on YES agar (SI Appendix, Fig. S4).

Discussion

The system of lncRNA-mediated transcriptional interference that underlies the repression of fission yeast phosphate homoeostasis gene pho1 has proven to be a sensitive readout of genetic influences on 3′-processing/termination and a powerful tool for discovery of agents and regulators of this step of the Pol2 transcription cycle (9, 12, 14, 25, 37). The present identification of seb1-G476S in a genetic screen for relief of transcriptional interference highlights a unique mechanism by which Pol2 termination can be enhanced via a gain-of-function mutation in an essential termination factor, Seb1. The genetic and physical evidence for gain-of-function is copious and persuasive. To wit: 1) seb1-G476S de-represses pho1 and tgp1, both of which are subject to lncRNA-mediated transcriptional interference; 2) RNA analysis shows that seb1-G476S elicits precocious lncRNA transcription termination in response to lncRNA 5′-proximal PASs; 3) seb1-G476S de-repression of pho1 is erased by loss-of-function mutations in CPF subunits and Rhn1; 4) seb1-G476S suppresses the ts growth defect of rhn1∆; and 5) seb1-G476S elicits an upstream shift in poly(A) site preference in several mRNA genes. To our knowledge, there is no prior example of a gain-of-function mutation in a eukaryal transcription termination protein. There is, however, precedent in bacteria, whereby Rho mutants were identified that were hyperactive in eliciting termination by E. coli RNA polymerase (38–40).

seb1-G476S displays a network of mechanistically informative genetic interactions with two other prominent governors of lncRNA interference with phosphate homeostasis genes: the Pol2 CTD and IPP dynamics. With respect to the CTD, seb1-G476S derepression of pho1 is reversed by a T4A mutation, which prevents installation of the Thr4-PO₄ mark and is itself hyperrepressive of the PHO regulon. Substantial prior genetic evidence implicates Thr4-PO₄ as a positive effector of 3′-processing/termination, likely by virtue of Thr-PO₄ function as a component of the CTD-binding site for Seb1 and Rhn1 (12, 27, 41). More striking is the observation that seb1-G476S derepression of pho1 is completely dependent on synthesis of IP8 by the Asp1 IPP kinase. IP8 is a metabolite agonist of 3′-processing/termination in fission yeast (25).

The mutational synergies and synthetic lethal genetic interactions of seb1-G476S make the case that the Seb1 gain-of-function in 3′-processing/termination is not restricted to the lncRNAs that control phosphate homeostasis genes (none of which are essential for fission yeast growth). seb1-G476S is synthetically very sick with the IPP pyrophosphatase-defective asp1-H397A allele, which raises the intracellular level of IP8 (23) and elicits a precocious lncRNA 3′-processing/termination phenotype similar to that of seb1-G476S (25). Combining the seb1-G476S and asp1-H397A alleles has an additive effect on the extent of pho1 derepression. seb1-G476S is synthetically lethal in the absence of another IPP pyrophosphatase, Aps1. These results show that elevated IPP concentrations and increased termination activity of Seb1-G476S synergize to affect the 3′-processing/termination of one or more essential fission yeast genes. That the lethality of seb1-G476S aps1∆ is caused by overzealous 3′-processing/termination is nicely confirmed by our findings that viability of seb1-G476S aps1∆ cells is reinstated by loss-of-function mutations of CPF subunits Ctf1 and Ssu72 and of termination factor Rhn1. In the same vein, seb1-G476S is synthetically lethal with three CTD alleles—S7A, S5•S5A, and P6•P6A—each of which derepresses pho1 in a manner dependent on CPF, Rhn1, and IP8 (12, 13, 30). The lethality of seb1-G476S CTD-S7A is overridden by Ctf1 and Ssu72 mutations, which signifies that seb1-G476S is contributing to a toxic phenotype of precocious termination. Finally, poly(A) site mapping fortifies the case for seb1-G476S as a gain-of-function in 3′-processing that favors the utilization of upstream poly(A) sites in mRNA genes that have alternative poly(A) sites or triggers the deployment of upstream poly(A) sites that were not evident in wild-type cells.

Clues to the mechanism by which seb1-G476S exerts its gain-of-function emerged from our crystal structure of the Seb1-G476S RNA-binding domain. The conformational change in the Ser476 mutant RRM vis-à-vis the wild-type RRM, entailing a 2.0- to 2.5-Å movement of the 11-aa loop from Val477 to Ser485, originates at the mutated amino acid and likely reflects the more restricted phi and psi angles available to serine (by virtue of its β-carbon) versus the less restricted Ramachandran plot unique to glycine. As yet, there has been no structure reported for Seb1 bound to RNA, and inference on that point could only be drawn by reference to the RNA complexes of the budding yeast Nrd1 RRM (29). In that regard, we could model the effect of the Ser476 mutation on RNA interaction by superimposing the shared 5′-GUA recognition element in the Nrd1 RRM structure on the Seb1-G476S RRM structure, which revealed the potential for new atomic contacts between the Ser476 and the uracil and guanine nucleobases of 5′-GUA. The predicted contacts of Ser476-Oγ are all within van der Waals distances of hydrogen-bonding atoms of the nucleobases. Taking into account that the RNA position/conformation might vary slightly between Nrd1 and Seb1, we envision that one or more of these atomic contacts will result in a gain of a hydrogen bond between Seb1 and RNA. Added van der Waals or hydrogen bond interactions might either enhance the interaction of Seb1 with consensus 5′-GUA target sites in nascent RNAs or confer interaction of new target sites that are not targets for wild-type Seb1 (e.g., 5′-GUG) by virtue of Ser476 donating a hydrogen bond to guanine-O6. We attempted to implement a gel shift assay to gauge the binding of purified wild-type and G476S RRM proteins to 5′ ³²P-labeled GUA-containing RNA oligonucleotides of varying length and flanking sequence context, but we were unable to detect a stable shifted species corresponding to an Seb1•RNA complex. The challenge now is to capture crystal structures of wild-type and mutant Seb1 RRMs in complex with RNA ligands, which will likely require a concerted effort at protein engineering to evade the favored crystal lattices that interfere with RNA binding.

Methods

The procedures for fission yeast growth and genetic manipulations, assays of cell surface acid phosphatase activity, whole genome sequencing, RNA-seq, poly(A) site mapping, Northern RNA analyses, and crystallographic structure determination are described in detail in SI Appendix. A list of fission yeast strains constructed for this study is compiled in SI Appendix, Table S4.

Data Availability

The RNA-seq data in this publication for the seb1-G476S strain and the control seb1⁺ strain have been deposited in the National Center for Biotechnology Information’s Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE168898. The Seb1 RRM-G476S crystal structure has been deposited in the PDB with identification code 7MI2. All other study data are included in the article and/or SI Appendix.