Saccharomyces cerevisiae Ecm2 modulates the catalytic steps of pre-mRNA splicing

Abstract

Genetic, biochemical, and structural studies have elucidated the molecular basis for spliceosome catalysis. Splicing is RNA catalyzed and the essential snRNA and protein factors are well-conserved. However, little is known about how nonessential components of the spliceosome contribute to the reaction and modulate the activities of the fundamental core machinery. Ecm2 is a nonessential yeast splicing factor that is a member of the Prp19-related complex of proteins. Cryo-electron microscopy (cryo-EM) structures have revealed that Ecm2 binds the U6 snRNA and is entangled with Cwc2, a factor previously found to promote a catalytically active conformation of the spliceosome. These structures also indicate that Ecm2 and the U2 snRNA likely form a transient interaction during 5′ splice site (SS) cleavage. We have characterized genetic interactions between ECM2 and alleles of splicing factors that alter the catalytic steps in splicing. In addition, we have studied how loss of ECM2 impacts splicing of pre-mRNAs containing nonconsensus or competing SS. Our results show that ECM2 functions during the catalytic stages of splicing. Our data are consistent with Ecm2 facilitating the formation and stabilization of the first-step catalytic site, promoting second-step catalysis, and permitting alternate 5′ SS usage. We propose that Cwc2 and Ecm2 can each fine-tune the spliceosome active site in unique ways. Their interaction network may act as a conduit through which splicing of certain pre-mRNAs, such as those containing weak or alternate splice sites, can be regulated.

INTRODUCTION

Spliceosomes are composed of small ribonucleoprotein particles (snRNPs), each containing proteins and a small nuclear RNA (U1, U2, U4, U5, or U6 snRNA), and dozens of additional protein splicing factors. Spliceosomes assemble from these factors before undergoing a number of conformational changes to form a catalytic center (activation) capable of carrying out the chemical steps of splicing (Fig. 1A): 5′ splice site (SS) cleavage (first step) and exon ligation (second step). Significant genetic, biochemical, and structural work over the past few decades has provided a wealth of information into how essential components of the splicing reaction such as the snRNAs, Prp8 protein, and DExD/H-box ATPases promote splicing (Wahl et al. 2009; Mayerle and Guthrie 2017; Kastner et al. 2019; Plaschka et al. 2019; Yan et al. 2019). In comparison, much less is known about how nonessential factors modulate the splicing reaction and interact with the core machinery.

FIGURE 1.

Structural analysis of Ecm2 during splicing. (A) Schematic of the pre-mRNA splicing pathway. ATPases tested for genetic interactions with Ecm2 are shown above the arrows of the respective steps that they promote. Spliceosome complexes containing Ecm2 are underlined in blue. (B) Cryo-EM structure and domain organization of Ecm2. U6 and U2 snRNA interacting regions are marked and shown in red and green, respectively. Locations of truncation mutants studied in panel C are indicated. Structure from 6EXN.pdb. (C) Structure of the Cwc2/Ecm2/Isy1 hub and Ecm2-RRM/U2 stem II interaction in C complex. The position of stem IIb after remodeling in C* complex has been superimposed on this structure. The U6 snRNA, Cwc2, and Ecm2 do not significantly change positions in C* complex. This figure was created using 5LJ5.pdb and 5MQ0.pdb. (D) Temperature sensitivity of ecm2Δ and truncation mutants on -ura dropout media after 3 days of growth.

Several of the nonessential splicing factors in yeast are associated with the Prp19-containing complex (NTC). Indeed, of 26 yeast proteins categorized as core-NTC components or NTC-associated, 12 can be considered as nonessential under certain growth conditions (i.e., deletion strains may possess phenotypes but nonetheless proliferate in permissive environments) (Hogg et al. 2010, 2014). Many of these proteins are well-conserved and have human splicing factor homologs. In general, the NTC is thought to stabilize catalytic conformations of the U6 snRNA and contribute to splicing fidelity (Hogg et al. 2010). Consistent with this model, cryo-EM structures and biochemical assays have shown that several nonessential NTC proteins directly interact with U6 including Cwc2, Ecm2, and Isy1 (Villa and Guthrie 2005; McGrail et al. 2009; Hogg et al. 2010; Rasche et al. 2012; Plaschka et al. 2019). Exactly how the NTC modulates RNA interactions within the spliceosome is not yet clear (Hogg et al. 2010), and it is difficult to infer potential mechanisms and the impact of mutations from cryo-EM structures alone (Mayerle and Guthrie 2017).

The NTC-associated protein Ecm2 was first isolated in a synthetic lethality screen for genetic interactors with U2 snRNA mutations (synthetic lethality with U2/slt screen) (Xu et al. 1998). This screen identified slt11/ecm2 as well as other splicing factors including Prp8 and Brr2. Additional work demonstrated genetic interactions between ecm2 and multiple components of the spliceosome but especially U2/U6 helix I and helix II mutants (Xu et al. 1998; Xu and Friesen 2001). Biochemical assays of splicing and spliceosome assembly showed that absence of Ecm2 results in loss of splicing activity at high temperatures and a block in spliceosome activation (Xu and Friesen 2001). Since it was known that U2/U6 form an intermolecular duplex during the early stages of activation (helix II) (Wassarman and Steitz 1992), it was proposed that Ecm2 functions during spliceosome activation to facilitate formation of this duplex (Xu and Friesen 2001).

Structures of Ecm2 integrated into a number of spliceosome complexes have been determined by cryo-EM (Plaschka et al. 2019). Ecm2 contains two RNA binding domains separated by a linker: an amino-terminal zinc finger motifs (ZNF) domain and a carboxy-terminal RNA recognition motif (RRM) (Fig. 1B). Unexpectedly, Ecm2 does not directly bind U2/U6 helix II. Rather the ZNF domain interacts with the U6 nucleotides (nt) 29–32, which are located between the U6 5′ stem–loop and the ACAGAGA-box/5′ SS pairing region (Fig. 1C). Ecm2 is also intertwined with another NTC-associated protein, Cwc2. Cwc2 contacts both the intronic RNA downstream from the 5′ SS and the U6 snRNA at multiple locations.

Interestingly, the carboxy-terminal RRM of Ecm2 closely approaches U2 snRNA stem IIb in the C complex spliceosome (Fig. 1C). Due to low resolution within this region, the exact molecular contacts between the RRM, U2 stem IIb, and nearby regions of Cwc2 are unclear. This interaction is likely transient: the Ecm2/U2 interaction has only been observed in structures captured just before (B* complex) and after (C complex) the first step of splicing (Galej et al. 2016; Wan et al. 2019). Large conformational changes place U2 stem IIb far away from any possible Ecm2 interaction in other structures (Fig. 1A,C; Rauhut et al. 2016; Yan et al. 2016a,b; Fica et al. 2017). Toggling of Ecm2/U2 stem IIb contacts on-or-off in different complexes resembles structural toggle switches reported for other splicing factors. These include the RNaseH domain of Prp8, the U4/U6 di-snRNA, as well as interconversion of U2 stem II itself between two mutually exclusive structures: stem IIa/b and stem IIb/c (Hilliker et al. 2007; Perriman and Ares 2007, 2010; Rodgers et al. 2016a,b; Abelson 2017; Mayerle et al. 2017). The significance of the Ecm2/stem IIb interaction has not been studied.

Human spliceosomes do not contain direct homologs of Cwc2 and Ecm2. Instead, a single protein, RBM22, binds the U6 snRNA at the corresponding positions. Based on limited sequence homology, it has previously been proposed that RBM22 represents a fusion of Cwc2 and Ecm2 (Rasche et al. 2012). This is supported by biochemical studies that show a similar function for Cwc2 and RBM22 in stabilizing the spliceosome active site (McGrail et al. 2009; Rasche et al. 2012; Hogg et al. 2014). In addition, both Cwc2 and RBM22 interact with intronic RNA at a location downstream from the 5′SS (Rasche et al. 2012; Kastner et al. 2019). Whether or not Ecm2 and RBM22 also share any conserved functions is unknown.

Here, we report genetic interactions between Ecm2 and splicing factors capable of modulating the first and second steps of splicing including the Prp2 and Prp16 ATPases, Prp8, and the U2 and U6 snRNAs. Ecm2 also exhibits genetic interactions with mutations that disrupt U2 stem II toggling, consistent with a functional interaction between the protein and U2. Genetic deletion of ECM2 changes how pre-mRNAs containing nonconsensus splice sites are processed, implicating Ecm2 in the catalytic steps of splicing in addition to its role in activation. Our results support a model in which Ecm2 has distinct functions for each catalytic step and are consistent with a proposal that several conditionally nonessential splicing factors (Ecm2, Cwc2, Isy1) function as a hub for regulating spliceosome catalysis (Hogg et al. 2010). These results have implications for the function of RBM22 in human spliceosomes as well as for how RBM22/intronic RNA interactions are formed.

RESULTS

The Ecm2 U6-binding domain is insufficient to rescue yeast growth at 37°C

Previous studies of Ecm2 reported that ecm2Δ yeast exhibited a strong temperature-sensitive (ts) phenotype with significantly reduced or no growth at temperatures above 33°C (Xu and Friesen 2001). We replicated this result by deleting ECM2 from a haploid strain of yeast and introducing plasmids containing ecm2 variants under control of their endogenous promoters. As expected, ecm2Δ yeast containing an empty plasmid grew well at permissive temperatures (16°C–30°C) but possessed a severe ts phenotype at 37°C (Fig. 1D). When we included a plasmid containing the wild type (WT) ECM2 gene, the ts phenotype was corrected, and growth was restored at 37°C.

To test if the amino-terminal, U6-binding binding domain of Ecm2 alone was capable of rescuing the ts phenotype, we used recent cryo-EM structures of yeast spliceosomes to design truncation mutants of Ecm2. Nonsense mutations were incorporated at amino acids 144, 198, 266, and 326 to allow for expression of variants containing only the U6-binding ZNF domain (Ecm2^1–143), the ZNF domain plus the inter-domain linker (Ecm2^1–197), the ZNF and a partial U2-binding, RRM domain (lacking amino acids that come nearest to U2, Ecm2^1–65), or the complete ZNF and RRM domains truncated at the last amino acid modeled into cryo-EM density but missing the carboxy-terminal lysine-rich region (Ecm2^1–325, Fig. 1D). Variants containing the U6-binding, ZNF domain but not the RRM were able to partially rescue the ts phenotype but still grew poorly at 37°C. Inclusion of the entire U2-binding, RRM (Ecm2^1–325) resulted in more significant suppression of the ts phenotype; although, cells still grew more slowly than those containing Ecm2^WT. Western blotting confirmed that all of the truncated proteins were expressed in cells; however, with the exception of Ecm2^1–325, they were expressed at lower levels than Ecm2^WT. Consequently, the inability to rescue the ts phenotype could be due to both truncation and decreased levels of the proteins. Since Ecm2^1–325 was expressed at levels comparable to Ecm2^WT, the carboxy-terminal lysine-rich region of the protein likely does play a role in suppression of the ts phenotype.

Genetic interactions between Ecm2 and the Prp2 and Prp16 ATPases

Xu and Friesen (2001) provided ample evidence that Ecm2 plays a role in spliceosome activation. We and others have previously noted that key players in the activation process such as the U2 snRNP protein Hsh155/SF3B1 and U2/U6 helix I exhibit genetic interactions with a cold-sensitive (cs) mutant of the DEAH-box ATPase Prp2 (Prp2^Q548N) (Wlodaver and Staley 2014; Carrocci et al. 2017; Kaur et al. 2020). Prp2 binds the intronic RNA downstream from the branch site and uses ATP hydrolysis to trigger release of Hsh155/SF3B1 and other U2 snRNP proteins during activation (Lardelli et al. 2010; van der Feltz and Hoskins 2019). We tested if ecm2Δ would also show a genetic interaction with Prp2^Q548N. When we combined Prp2^Q548N with ecm2Δ, we observed no growth at low or high temperatures (16°C, 23°C, or 37°C) and reduced growth at 30°C (Fig. 2A). Prp2^Q548N is synthetic lethal with ecm2Δ at low temperatures and Prp2^Q548N does not rescue the ts phenotype of ecm2Δ. This genetic interaction is consistent with Ecm2's function in promoting spliceosome activation.

FIGURE 2.

Genetic interactions between Ecm2 and spliceosomal ATPases. (A–C) Mutations in Prp2 (A, cs), Prp16 (B, cs), and Prp22 (C, cs, and ts) were combined with ecm2Δ and tested for suppression or exacerbation of temperature-dependent growth phenotypes. Yeast were plated on YPD media and imaged after 3 (23°C, 30°C, or 37°C) or 10 (16°C) days of growth.

We next tested if other spliceosome DEAH-box ATPases would also show genetic interactions with ecm2Δ or if these results were specific to Prp2^Q548N. We combined ecm2Δ with a cs mutation of the ATPase Prp16 (Prp16^R686I) or a cs and ts mutation of the ATPase Prp22 (Prp22^T637A). Prp16 uses ATP hydrolysis to promote conformational changes of the spliceosome and splicing factor release during remodeling of the active site from the first to second catalytic step (Fig. 1A; Schwer and Guthrie 1992; Semlow et al. 2016; Plaschka et al. 2019). Prp16^R686I likely impedes this transition since this mutation is rescued by alleles of Prp8 that promote exon ligation (second-step alleles, discussed below) (Query and Konarska 2004). Prp22 also uses ATP hydrolysis to promote conformational change that enables release of the spliced mRNA product from the active site (Fig. 1A; Wagner et al. 1998; Schwer 2008; Semlow et al. 2016; Plaschka et al. 2019). In this case, Prp22^T637A likely impedes mRNA release and transition of the active site out of the exon ligation conformation since this mutation is exacerbated by second-step alleles of Prp8 (Query and Konarska 2012).

When Prp16^R686I was combined with ecm2Δ, the cs phenotype of Prp16^R686I was suppressed and growth was restored at 16°C (Fig. 2B). Yeast containing both Prp16^R686I and ecm2Δ also grew at 23°C and 30°C, albeit less well than when WT alleles were present. In addition, Prp16^R686I exacerbated the ts phenotype at 37°C of ecm2Δ yeast. This is consistent with some degree of synthetic lethality between ECM2 and PRP16 at high temperatures and is consistent with a previous report of synthetic lethality between the slt11-1 and prp16-1 alleles (Xu et al. 1998). We do note, however, that since the ts phenotype resulting from ecm2Δ in this yeast strain is less severe than in others (compare growth at 37°C in Fig. 1D vs. 2B), altered expression of Prp16 could also play a role appearance or suppression of phenotypes due to ecm2Δ.

While ecm2Δ suppressed the cs phenotype of Prp16^R686I, the cs phenotype of Prp22^T637A was not suppressed (Fig. 2C). Yeast containing both Prp22^T637A and ecm2Δ grew very poorly at 37°C, and it was difficult to determine if Prp22^T637A was a weak suppressor of the ts phenotype of ecm2Δ yeast. In sum, these data strongly support genetic interactions between ecm2Δ and the Prp2 and Prp16 ATPases. Loss of Ecm2 exacerbates a cs defect in spliceosome activation caused by Prp2^Q548N and suppresses a cs defect in the first-to-second step conformational change caused by Prp16^R686I.

Genetic interactions between Ecm2 and mutations in U2 snRNA stem II

The above results are consistent with a model in which Ecm2 stabilizes the first-step conformation of the spliceosome: aiding its formation during Prp2-initiated activation and inhibiting its remodeling by Prp16. To gain further insight into Ecm2's role during these steps, we combined ecm2Δ with mutations in the stem II region of the U2 snRNA which undergo a conformational change during activation. This region of U2 includes stem IIa/c as well as stem IIb—the RNA contacted by the carboxy-terminal RRM of Ecm2 in cryo-EM structures of B* and C complex spliceosomes (Fig. 1C).

During activation, stem II undergoes a reversible conformational change from the stem IIa to the stem IIc structure, while stem IIb remains intact (Fig. 3A; van der Feltz and Hoskins 2019). 5′ SS cleavage is inhibited when formation of stem IIc is blocked by deletion of the 3′ stem (ΔCC) or destabilized by mutation (Hilliker et al. 2007; Perriman and Ares 2007). In contrast, stabilization of stem IIc with additional base pairs (IIc+) promotes the first step of splicing (Perriman and Ares 2007). Like ecm2Δ, mutations that destabilize stem IIc or disrupt an interaction that is physically mutually exclusive with stem IIa also suppress Prp16 mutants defective in remodeling the first-step spliceosome active site (Hilliker et al. 2007; Perriman and Ares 2007). We predicted that if Ecm2 is facilitating activation by assisting stem IIc formation, then deletion of ECM2 should exacerbate the phenotypes of mutants that antagonize stem IIc.

FIGURE 3.

Genetic interactions between Ecm2 and U2 stem II mutations. (A) Schematic of stem IIa/IIc toggling. Mutations which disfavor stem IIc (U2-2,4 and ΔCC; green-colored labels) are shown in the stem IIa structure. Mutations which disfavor stem IIa (G53A, IIc+; red-colored labels) are shown in the stem IIc structure. Nucleotides that are deleted in the ΔIIb mutant are shown in purple. (B) Mutations in stem II were combined with ecm2Δ and tested for suppression or exacerbation of temperature-dependent growth phenotypes. Yeast were plated on YPD media and imaged after 3 (23°C, 30°C, or 37°C) or 10 (16°C) days of growth.

The U2-2,4 and ΔCC mutations both disrupt stem IIc formation: U2-2,4 stabilizes the competing stem IIa structure while ΔCC prevents stem IIc formation entirely by deletion of the nucleotides that comprise the 3′ half of stem IIc (Fig. 3A; Perriman and Ares 2007). These mutations have little phenotypic effect by themselves in our assay. However, when combined with ecm2Δ these mutations caused synthetic lethality at 30°C and cs phenotypes at 16°C and 23°C (Fig. 3B). These results agree with our prediction that Ecm2 facilitates stem IIc formation.

This model also predicts that mutations in stem II that promote stem IIc formation may be able to suppress the ts phenotype of ecm2Δ yeast. The G53A and IIc+ mutants both favor stem IIc: G53A destabilizes the competing stem IIa structure while IIc+ extends base-pairing of IIc (Fig. 3A; Perriman and Ares 2007). These U2 mutants exhibit phenotypes even in the presence of Ecm2: both are cs while IIc+ also exhibits a modest growth defect at 30°C and 37°C. Neither mutation suppressed the ts phenotype of ecm2Δ yeast, and ecm2Δ exacerbated the cs phenotypes of both mutations. These latter results could mean that Ecm2 has additional functions in the spliceosome while stem IIa is present or that snRNA structures containing these mutations are also disruptive for growth at lower temperatures in the absence of Ecm2.

If Ecm2 functions to assist stem IIc formation during activation, it is possible that this occurs through capture of stem IIb by the carboxy-terminal RRM of Ecm2 during the B^act to B* complex transition. Stem IIb is nonessential in yeast (Ares and Igel 1990), and we tested if deletion of stem IIb (ΔIIb) resulted in a similar ts phenotype as ecm2Δ. The ΔIIb mutant yeast were not ts and exhibited minimal or no temperature-dependent phenotypes (Fig. 3B). The ts phenotype at 37°C of ecm2Δ was still observed when combined with the U2 ΔIIb mutation, and yeast containing both ecm2Δ and U2 ΔIIb grew similarly at other temperatures. This indicates that disruption of the Ecm2-RRM/stem IIb interaction is not solely responsible for the ts phenotype in ecm2Δ yeast.

Ecm2 impacts splicing of reporter pre-mRNAs containing nonconsensus SS

We next studied how Ecm2 influences splicing catalysis in vivo with the ACT1–CUP1 splicing reporter (Fig. 4A). In this assay, splicing of the ACT1–CUP1 pre-mRNA is necessary for growth of a Cu²⁺ sensitive yeast strain on Cu²⁺-containing media. The highest (Cu²⁺) at which growth is observed is proportional to the amount of spliced mRNA in the cell (Lesser and Guthrie 1993b). In the presence of an ACT1–CUP1 reporter containing consensus SS, we observed no difference in Cu²⁺ tolerance between strains with or without ECM2. We used a primer extension assay to confirm that the similar Cu²⁺ tolerance results were correlated with high splicing efficiencies for both catalytic steps in the presence or absence of Ecm2 (Supplemental Fig. S1). This indicates that splicing can still occur efficiently in the absence of Ecm2 and is consistent with lack of a significant growth phenotype in ecm2Δ strains at 30°C in our assays.

FIGURE 4.

Impact of Ecm2 on splicing of ACT1–CUP1 reporter pre-mRNAs. (A) Schematic of the ACT1–CUP1 reporter pre-mRNA with nonconsensus substitutions noted. (B) ACT1–CUP1 assay results. Representative images of yeast growth after 3 days at 30°C on agar plates made with -leu -trp dropout media containing 0 or 0.7 mM Cu²⁺ are shown above the bar graph. Each value in the graph represents the highest concentration of Cu²⁺ at which growth was observed on plates containing defined concentrations of Cu²⁺ and are representative of results obtained in replicate assays. (C) Schematic of the modified ACT1–CUP1 reporter containing a competing, cryptic 5′ SS. (D) Primer extension assay of cryptic 5′ SS usage using the reporter shown in panel C. Primer extension of the U6 snRNA was included as a control. The percentages of cryptic products (ratios of cryptic products/total products) are shown below the gel and are the averages of three replicate experiments ± the standard deviation.

We then assayed splicing in ecm2Δ yeast using ACT1–CUP1 reporters harboring substitutions at the SS. The most significant decreases in Cu²⁺ tolerance were observed using reporters containing the A3C substitution at the 5′ SS, substitutions of the branch point adenosine (A259C or A259G), or substitutions flanking the branch point (U257C and C260G) (Fig. 4B). The large impact of ecm2Δ on A3C reporter splicing was intriguing since this substitution is only limiting for the second catalytic step (Liu et al. 2007; Eysmont et al. 2019). Primer extension analysis of first- and second-step splicing products confirmed a strong defect in exon ligation for the A3C reporter in the absence of Ecm2 (Supplemental Fig. S2). Ecm2 can therefore have opposing effects on the second-step active site: it can inhibit its formation but also promote second-step catalysis on a substrate containing the A3C 5′ SS substitution. It is possible that Ecm2 has distinct functions in both spliceosome structural transitions and in each catalytic reaction.

Deletion of ECM2 improved Cu²⁺ tolerance of yeast containing the A3U or G5A 5′ SS reporters (Fig. 4B). However, analysis of first- and second-step splicing products showed similar splicing efficiencies in the presence or absence of Ecm2 (Supplemental Fig. S2). We did not study how decay of unspliced RNAs or splicing intermediates influenced these primer extension results (Liu et al. 2007). Therefore, it is difficult to conclude from the primer extension assay if ecm2Δ truly changed the splicing efficiencies for the A3U and G5A substrates.

Nonetheless, the increase in Cu²⁺ tolerance observed with the G5A mutant in ecm2Δ yeast was noteworthy since this substitution can result in use of a cryptic, upstream 5′ SS (Parker and Guthrie 1985; Kandels-Lewis and Séraphin 1993; Lesser and Guthrie 1993a). We used a modified ACT1–CUP1 reporter with competing 5′ SS to test whether or not Ecm2 changes cryptic SS usage (Fig. 4C; Supplemental Fig. S3). When Ecm2 was present in the yeast, we detected usage of both the cryptic (21 ± 1% of spliced products) and normal 5′ SS. However, in the absence of Ecm2 use of the cryptic 5′ SS was greatly reduced (7 ± 4% of spliced products, Fig. 4D). This represents at least a threefold decrease based on our lower limit of detection. Combined, our results demonstrate that Ecm2 impacts the spliceosome active site to alter splicing of pre-mRNAs with nonconsensus SS and permit the usage of an alternate, cryptic 5′ SS.

Genetic interactions between Ecm2 and the U6 snRNA first- and second-step alleles

Like ecm2Δ, a number of alleles of the U6 snRNA and Prp8 suppress Prp16 ATPase mutations, have limited impact on splicing of ACT1–CUP1 reporters harboring consensus SS, and can promote or block splicing of reporters with particular SS substitutions (McPheeters 1996; Query and Konarska 2004; Liu et al. 2007; Mayerle et al. 2017; Eysmont et al. 2019). Many of these mutants have been categorized as first- or second-step alleles since, in addition to causing ts or cs phenotypes, they promote one of the catalytic steps of splicing over the other (Fig. 5A). Since ecm2Δ and first- and second-step alleles share common features, we tested for genetic interactions between these alleles and ecm2Δ. We first examined interactions with the U6 snRNA U57C and U57A mutations which promote the first and second steps, respectively (McPheeters 1996; Liu et al. 2007).

FIGURE 5.

Genetic interactions between Ecm2 and U6 or Prp8 first- and second-step alleles. (A) Illustration of how alleles of Prp8, Prp16, and U6 function to promote the first or second step of splicing. (B) A first- or second-step allele of U6 (U57C or U57A, respectively) was combined with URA3 plasmids either lacking or coding for Ecm2 variants in ecm2Δ yeast. The strains were then tested for suppression or exacerbation of temperature-dependent growth phenotypes. Yeast were plated on -URA dropout media and imaged after 2 days of growth. (C) First- and second-step alleles of Prp8 (red and green, respectively) were combined with ecm2Δ and tested for suppression or exacerbation of temperature-dependent growth phenotypes. Yeast were plated on YPD media and imaged after 3 (23°C, 30°C, or 37°C) or 10 (16°C) days of growth.

The U6–U57A mutation had no effect on yeast growth at 16°C, 23°C, or 30°C in the absence of Ecm2 or in the presence of Ecm2^WT or Ecm2^1–143 (which contains only the U6-binding domain, Fig. 5B; Supplemental Fig. S4). The U6–U57C mutation also had no impact on growth at 16°C or 23°C but was slower growing at 30°C. U6–U57C yeast containing only the empty URA3 plasmid displayed a slower-growing phenotype compared to those containing plasmids for Ecm2^WT or Ecm2^1–143.

At higher temperatures, the impacts of the U6 mutations were more pronounced. When yeast contained the first-step, U6–U57C allele, we observed a strong synthetic sick interaction with ecm2Δ at 34°C that was partially rescued by expression of Ecm2^WT or Ecm2^1–143, with the former producing stronger rescue than the latter (Fig. 5B). At 37°C only the strain expressing Ecm2^WT was viable in the presence of U6–U57C. In contrast, we observed a weaker synthetic genetic interaction between ecm2Δ and the second-step allele, U6–U57A, at 34°C (Fig. 5B). As with U6–U57C, this interaction was more pronounced at 37°C and yeast containing U6–U57C along with the truncated Ecm2^1–143 protein grew much more slowly than yeast containing Ecm2^1–143 and U6-WT. Combined, we conclude that synthetic genetic interactions at 34°C and 37°C between Ecm2 and U6–U57C elicit stronger phenotypes than those between Ecm2 and U6–U57A.

The interactions of ecm2Δ with these U6 mutants are most similar to those of first-step alleles in other splicing factors like Prp8 (Liu et al. 2007). When U6 mutations are present, loss of Ecm2 promotes the first step of splicing and presence of Ecm2 promotes the second step. These results differ from those obtained upon deletion of the nonessential factor Isy1 (Fig. 1C): isy1Δ is synthetic lethal with U57A and suppresses the ts phenotype of U57C (Villa and Guthrie 2005). Thus, Isy1 appears to act as a first-step splicing factor when U6 is mutated, consistent with Isy1 release prior to the second step (Plaschka et al. 2019), while Ecm2 can act as a second-step factor and is consistent with its presence throughout both catalytic stages of splicing (Fig. 1A).

Genetic interactions between Ecm2 and Prp8 first- and second-step alleles

Genetic interactions between ecm2Δ and Prp2, Prp16, and U2 stem II and Ecm2-control of 5′ SS usage support a role for Ecm2 in the first step of splicing. However, genetic interactions with U6–U57 mutants and results using the A3C splicing reporter support an additional role for Ecm2 in the second step. We next tested if ecm2Δ would show genetic interactions with first- or second-step alleles of Prp8 or both. We individually combined ecm2Δ with two first-step alleles of Prp8 (Prp8^R1753K or Prp8^E1960K) or two second-step alleles (Prp8^P986T or Prp8^V1870N). For the first-step alleles, deletion of Ecm2 weakly suppressed the cs phenotype of Prp8^R1753K and strongly suppressed the cs phenotype of Prp8^E1960K (Fig. 5C). Neither Prp8 first-step allele was able to completely correct the ts phenotype of ecm2Δ; although, slightly improved growth was observed at 37°C for yeast containing Prp8^E1960K (Fig. 5C).

When ecm2Δ was combined with the second-step alleles, we observed slightly improved growth at 37°C for yeast containing Prp8^P986T. A stronger genetic interaction was observed with the Prp8^V1870N. This second-step allele exacerbated the ts phenotype of ecm2Δ, causing a strong growth defect at 30°C and no growth at 37°C (Fig. 5C). The growth defect of Prp8^V1870N was partially corrected at 30°C by combining first and second Prp8 alleles (Prp8^{V1870N,E1960K}). However, this also resulted in stronger growth defects at other temperatures.

The cs suppression we observe of the Prp8^E1960K first-step allele and ts exacerbation of the Prp8^V1870N second-step allele are consistent with ecm2Δ acting as a second-step allele and facilitating exit of the spliceosome from the first-step catalytic conformation. Both the Prp8^E1960K and Prp8^V1870N substitutions are located within Prp8's RNaseH domain. Like U2 stem II, the RNaseH domain is both highly dynamic and toggles between alternate structures (Schellenberg et al. 2013; Mayerle et al. 2017). Ecm2 may impact Prp8–RNaseH dynamics or vice versa to support the first-step reaction. This is in juxtaposition to the results obtained with the U6 mutants, which were consistent with Ecm2 having a role in the second step.

DISCUSSION

Using genetic and biochemical assays of splicing in cells, we have shown that yeast Ecm2 impacts multiple steps during the catalytic phase of splicing and that loss of Ecm2 perturbs how the spliceosome processes pre-mRNAs containing nonconsensus SS. Deletion of ECM2 results in genetic interactions with several structural switches in the spliceosome including U2 stem II, the RNaseH domain of Prp8, and the ATPases that control entry to and exit from the first step (Prp2 and Prp16, respectively). In sum, our data show that Ecm2 plays significant roles in spliceosome catalysis in addition to a function during activation (Xu and Friesen 2001).

Ecm2 modulates the catalytic steps of splicing

Our results support functions for Ecm2 during both catalytic steps in splicing. The differing genetic interactions we observe between ecm2Δ and U6 or Prp8 mutants suggest a more complicated mechanism from that of other alleles that exhibit more consistent genetic interactions (for example, a second-step allele of cef1 suppresses phenotypes of both first-step prp8 and U6 alleles) (Query and Konarska 2012). Our results could be explained by distinct and genetically separable functions for Ecm2 during the first and second catalytic steps with Prp8 mutations revealing a role in the former and U6 mutations revealing a role in the latter. Since Ecm2 has only been observed to make contact with U2 stem IIb in first-step cryo-EM structures, it is possible that this interaction contributes to Ecm2's distinct functions during each catalytic step.

Several of our observations with Ecm2 are similar to those previously reported for Cwc2 and Isy1 (Villa and Guthrie 2005; Rasche et al. 2012; Hogg et al. 2014). Cwc2, Ecm2, and Isy1 form a highly interconnected network of interactions with one another, the U6 snRNA, the intron, and a number of other splicing factors (Figs. 1C, 6A; Galej et al. 2016; Wan et al. 2016). All three proteins can suppress Prp16 mutations and have synthetic lethal interactions with active site U2/U6 helix Ia or Ib mutations (Xu et al. 1998; Villa and Guthrie 2005; Hogg et al. 2014). While deletion of Cwc2, Isy1, or Ecm2 results in distinct growth phenotypes (for example, cwc2Δ is lethal at 30°C but ecm2Δ shows minimal changes in growth at that temperature), functional spliceosomes can form in yeast in the absence of one or a subset of these factors under permissive growth conditions (Xu and Friesen 2001; Villa and Guthrie 2005; Hogg et al. 2014). In addition, loss of Ecm2, loss of Isy1, or mutation of Cwc2 results in specific splicing defects in reporter pre-mRNAs with nonconsensus SS (Fig. 4B; Villa and Guthrie 2005; Hogg et al. 2014). This implies that spliceosomes missing one of these factors possess different substrate preferences and fidelity phenotypes. This is intriguing since ecm2Δ only results in a growth defect at high temperatures and splicing of pre-mRNAs containing consensus SS remains efficient (Fig. 4B). Thus, it is possible that yeast could bias the splicing of particular pre-mRNAs with nonconsensus SS by regulating the Cwc2, Ecm2, and/or Isy1 content of spliceosomes without significantly compromising cellular splicing efficiency. This possibility has previously been proposed by Villa and Guthrie (2005), who noted that deletion of Isy1 results in reduced fidelity of 3′ SS selection.

FIGURE 6.

The Cwc2/Ecm2/Isy1 interaction network and structure of human RBM22. (A) A large number of splicing factors interact with Cwc2, Ecm2, and/or Isy1 suggesting that these proteins form a network hub for modulating spliceosome activity. In this model, regulatory signals could flow into the hub from the NTC and NTC-related proteins and outwards to the spliceosome active site consisting of the intron, Prp8, and U2/U6 snRNAs. (B) Two views of the cryo-EM structure of RBM22 from a human C complex spliceosome. Domains of RBM22 are noted and intronic RNA downstream from the 5′ SS is shown in black spacefill. Note that RBM22 completely encircles the RNA. Structure from 6EXN.pdb. (C) Hypothetical model for formation of the structure shown in panel B. The RRM domain of RBM22 could make transient contact with human U2 stem IIb to allow for docking of the intron and subsequent wrapping. Structures in panels B and C are from 5YZG.pdb. The hypothetical model in panel C was created using PyMol.

While there is some overlap in how Isy1 or Ecm2 loss or Cwc2 mutation impact splicing of nonconsensus SS, the proteins also exert unique influences of the spliceosome. For example, isy1Δ and the Cwc2^F183D mutant improve splicing of reporter pre-mRNAs containing a A302U 3′ SS, but ecm2Δ only minimally changes A302U splicing (Fig. 4B). In addition, ecm2Δ changes Cu²⁺ tolerance with the G5A reporter but this is unaffected by isy1Δ or Cwc2^F183D (Villa and Guthrie 2005; Hogg et al. 2014). Cellular splicing could thus be optimized for specific SS by independently controlling Cwc2, Ecm2, and Isy1 stoichiometry with spliceosomes.

These factors might also impact mRNA isoform production since Ecm2 additionally permits usage of an alternative 5′ SS (Fig. 4D). Interestingly, the nonessential yeast splicing factor Bud31 is required for use of an alternative 5′ SS in the SRC1 mRNA (Saha et al. 2012), and Bud31 directly contacts the U6 snRNA, Ecm2, and Cwc2 in the yeast spliceosome (Plaschka et al. 2019). Bud31 and Ecm2 could permit promiscuous 5′ SS use by similar mechanisms, although this has not yet been studied. In summary, spliceosomes may be fine-tuned by the presence or absence of nonessential splicing factors like Ecm2, and currently little is known about how the compositions of spliceosomes vary inside cells.

Consequences of a dynamic Ecm2/U2 stem II interaction during splicing

The spliceosome contains a number of proposed switches in which components toggle between one conformation or another at different stages of the reaction (Abelson 2017). The U2 snRNA contains several of these switches including a U2 stem IIa-to-IIc conformational change during activation (van der Feltz and Hoskins 2019). In addition, it has also been proposed that stem IIc switches transiently back to stem IIa between the catalytic steps of splicing before reforming IIc during the second step (Hilliker et al. 2007; Perriman and Ares 2007). This mechanism was based in part on the observation that stem II substitutions that destabilize stem IIc (or stabilize stem IIa) can suppress cs alleles of Prp16. Our observations that ecm2Δ also suppresses Prp16 cs alleles (Fig. 2) and Ecm2 contacts U2 stem IIb in C complex may provide an alternate explanation.

We propose that Prp16 indirectly disrupts the Ecm2/stem II interaction during remodeling of the spliceosome between the first and second steps. Eliminating or weakening this interaction by stem II mutation can suppress Prp16 cs alleles by destabilizing the first-step conformation. This explanation is supported by cryo-EM structural data in which a transient contact between U2 stem IIb/c and the carboxy-terminal RRM of Ecm2 is observed in first-step complexes (B* and C complexes) but not those preceding or following (B^act and C* complexes) (Galej et al. 2016; Rauhut et al. 2016; Wan et al. 2016, 2019; Yan et al. 2016a; Fica et al. 2017). While additional structural information is needed for the on-pathway intermediates during second-step active site assembly, stem IIa has not yet been observed in C* spliceosomes and accommodation of stem IIa in these complexes may be incompatible with binding of splicing factors (Prp17) and U2 snRNP interactions with Syf1 (Yan et al. 2016b; Bai et al. 2017; Fica et al. 2017, 2019; Liu et al. 2017). In light of these observations, stem IIc could remain intact throughout catalysis, and IIc-to-IIa toggling occurs later during spliceosome disassembly or U2 snRNP reassembly (Yan et al. 1998; Rodgers et al. 2016a). Regardless, further work is needed to characterize the short-lived intermediates that form during the first- to second-step transition.

The viability of ecm2Δ and stem IIbΔ strains (Fig. 3B; Ares and Igel 1990; Xu and Friesen 2001) show that the Ecm2/stem II interaction is not essential for yeast splicing. It is notable, however, that ecm2Δ exhibits synthetic lethal interactions with multiple stem II mutations. This includes, to our knowledge, the first genetic data showing synthetic lethality with the U2-2,4 mutant, which stabilizes stem IIa. Stabilization of stem IIa by U2-2,4 also results in a more significant growth defect than deletion of stem IIc (ΔCC, Fig. 3B), suggesting that failure to disrupt stem IIa is of greater consequence for the spliceosome than failure to form stem IIc in the absence of Ecm2. Therefore, these results complement previous evidence for the importance of stem IIa/IIc toggling (Hilliker et al. 2007; Perriman and Ares 2007) and highlight the need for stem IIa unwinding for efficient splicing. It is possible that the Ecm2/stem II interaction only becomes limiting for splicing when stem IIa/c toggling is disturbed or when the active site is destabilized by SS mutations.

Implications for human RBM22 and wrapped intron formation

The evolutionary histories of Cwc2, Ecm2, and RBM22 have not been studied, and it is uncertain how RBM22 may have evolved to functionally replace both proteins. Based on sequence alignments and crosslinking studies, it has been proposed that Cwc2 and RBM22 share a common function in binding U6 and interacting with the spliceosome's catalytic elements (Rasche et al. 2012). However, when fragments of the yeast and human C complex spliceosomes are aligned, RBM22 most closely mimics the interactions of Ecm2 with the U6 snRNA (Supplemental Movie S1). In terms of U6 interaction, we believe that RBM22 and Ecm2, not Cwc2, are closer structural homologs.

Both RBM22 and Cwc2 bind the intronic RNA just downstream from the 5′ SS. The RBM22/intron interaction contains an unusual and distinctive structure not observed with Cwc2. In human C and P complex spliceosomes, RBM22 completely encircles the intron (Fig. 6B; Zhan et al. 2018; Fica et al. 2019). It is unlikely that this wrapped intron structure would form by threading of the intron through RBM22. Insights from Ecm2 provide a plausible mechanism for its formation. The carboxy-terminal RRM of RBM22 could transiently interact with U2 stem IIb/c—analogous to the interaction between Ecm2 and stem II in yeast (Fig. 6C). This could open RBM22 for intron binding and enable subsequent wrapping of the intron after disruption of the RRM/stem II interaction.

Analyses of cryo-EM structures reveal that movement of RBM22 toward U2 stem II is not occluded by the presence of other splicing factors and stem IIb is within an accessible distance for the RRM, assuming structural flexibility of the linker between the ZNF and RRM domains. There is some biochemical evidence for a RBM22/U2 snRNA interaction: anti-RBM22 antibodies can immunoprecipitate (IP) small amounts of the U2 snRNA from C complex spliceosomes after proteinase K treatment and without co-IP of the U5 or U6 snRNAs—consistent with a direct interaction (Rasche et al. 2012). We note, however, that in recent structures of human pre-B^act spliceosomes RBM22 and the intron were modeled with the same intertwined topology as found in later stage complexes despite the presence of U2 stem IIa/b and a large distance between RBM22 and U2 stem II (∼150 Å) (Townsend et al. 2020). While it is possible that the limited resolution of the electron density map within the RBM22 region prevented modeling of alternate conformations, the structures nonetheless suggest that further experiments are needed to determine if intron wrapping by RBM22 occurs spontaneously (i.e, without a U2 snRNA stem II chaperone) and if an interaction between the RRM of RBM22 and stem IIb/c can occur during transitions between pre-B^act and B^act spliceosomes.

MATERIALS AND METHODS

Yeast strains and plasmids used in these studies are described in Supplemental Tables S1, S2. Yeast transformation and growth were carried out using standard techniques and media.

Genetic deletions of ECM2

Deletion of the ECM2 gene was carried out by replacement of the gene with an antibiotic resistance cassette (hygromycin or nourseothricin) by homologous recombination in the appropriate parental strain (Supplemental Table S1; Goldstein and McCusker 1999).

Cloning of ECM2 and site-directed mutagenesis

ECM2 along with 300 bp of up- and downstream DNA was amplified from yeast genomic DNA by PCR. The resulting product was digested with NotI and SalI restriction enzymes and then ligated into pRS416 (URA3 CEN6) at those same restriction sites to create plasmid pAAH1056 containing the WT ECM2 gene. Novel mutants of Ecm2 were generated using inverse polymerase chain reaction (PCR) with Phusion DNA polymerase (New England Biolabs). All plasmids were confirmed by sequencing.

Western blot analysis

A 3xFLAG epitope tag was genetically encoded at the carboxyl terminus of Ecm2 or truncation mutants by synthesis of the corresponding genes (Genewiz). Strains expressing the epitope tagged proteins were grown in -ura dropout media and 10 mL of cells were collected at an OD₆₀₀ of ∼1.0. Cellular protein was precipitated with tricholoracetic acid (TCA) and resuspended in 2× SDS-PAGE loading buffer. Proteins were separated on a 12% SDS-PAGE gel before being transferred to a nitrocellulose membrane using a semidry transfer apparatus and according to the manufacturer's directions (Pierce Power Blotter, Thermo Fisher). The blot was then stained with 0.5% w/v Ponceau S (Sigma) in 2% v/v acetic acid and imaged. The stain was removed by washing with 200 µM NaOH followed by water. The gel was blocked, incubated with HRP-linked anti-FLAG antibodies (Sigma), and washed using an iBind cassette (Invitrogen). Bands were detected using Clarity Western ECL substrate (Bio-Rad) and chemiluminescent imaging. Bands were quantified using ImageQuant TL v8.1 software (GE Healthcare) and normalized based on total protein content detected by Ponceau staining (Romero-Calvo et al. 2010).

ACT1–CUP1 copper tolerance assays

Yeast strains expressing ACT1–CUP1 reporters were grown to mid-log phase in -leu -trp dropout media to maintain selection for the plasmids, adjusted to OD₆₀₀ = 0.5 and equal volumes were spotted onto plates containing 0–2.5 mM CuSO₄ (Lesser and Guthrie 1993b; Carrocci et al. 2018). Plates were scored and imaged after 3 days of growth at 30°C.

Temperature growth assays

Yeast strains were grown to mid-log phase at permissive temperatures in YPD or -ura dropout liquid media. Cell growth was then quantified by measuring OD₆₀₀. Equal volumes of cells were diluted to an OD₆₀₀ = 0.5, were stamped onto YPD-agar plates and incubated at 23°C, 30°C or 37°C for 3 days or at 16°C for 10 days before imaging.

Primer extension

Cells were grown at 30°C in 25 mL yeast -leu -trp dropout liquid media until OD₆₀₀ reached 0.5–0.8, and 10 OD₆₀₀ units were collected by centrifugation. Total yeast RNA was isolated following the MasterPure Yeast RNA Purification Kit (Epicenter) protocol with minor changes as previously described (Carrocci et al. 2017). Primer extension was performed with IR dye conjugated probes yAC6: /5IRD700/GGCACTCATGACCTTC and yU6: /5IRD700/GAACTGCTGATCATCTCTG. purchased from Integrated DNA Technologies (Carrocci et al. 2017; Kaur et al. 2020). Gels were imaged with the Amersham IR Typhoon 5 (GE Healthcare) excitation at 685 nm, emission filter 720BP20, PMT voltage of 700V, and 100 µm pixel size. Band intensities were quantified with ImageQuant TL v8.1 (GE Healthcare).

Structural alignments and figure creation

Structural alignments of portions of human and yeast spliceosome complexes were carried out using PyMol by aligning to the U6 snRNA. Aligned structures of yeast spliceosomes were obtained from PyMOL4Spliceosome (https://github.com/mmagnus/PyMOL4Spliceosome) (Magnus et al. 2019). Figures and movies containing molecular structures were generated using Pymol (Schrödinger).

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.

COMPETING INTEREST STATEMENT

A.A.H. is a scientific advisor for Remix Therapeutics.