Structure-function analysis of the Yhc1 subunit of yeast U1 snRNP and genetic interactions of Yhc1 with Mud2, Nam8, Mud1, Tgs1, U1 snRNA, SmD3 and Prp28

Yhc1 and U1C are homologous essential subunits of the yeast and human U1 snRNP, respectively, that are implicated in the establishment and stability of the complex of U1 bound to the pre-mRNA 5′ splice site (5′SS). Here, we conducted a mutational analysis of Yhc1, guided by the U1C NMR structure and low-resolution crystal structure of human U1 snRNP. The N-terminal 170-amino acid segment of the 231-amino acid Yhc1 polypeptide sufficed for vegetative growth. Although changing the zinc-binding residue Cys6 to alanine was lethal, alanines at zinc-binding residues Cys9, His24 and His30 were not. Benign alanine substitutions at conserved surface residues elicited mutational synergies with other splicing components. YHC1-R21A was synthetically lethal in the absence of Mud2 and synthetically sick in the absence of Nam8, Mud1 and Tgs1 or in the presence of variant U1 snRNAs. YHC1 alleles K28A, Y12A, T14A, K22A and H15A displayed a progressively narrower range of synergies. R21A and K28A bypassed the essentiality of DEAD-box protein Prp28, suggesting that they affected U1•5′SS complex stability. Yhc1 Arg21 fortifies the U1•5′SS complex via contacts with SmD3 residues Glu37/Asp38, mutations of which synergized with mud2Δ and bypassed prp28Δ. YHC1-(1-170) was synthetically lethal with mutations of all components interrogated, with the exception of Nam8.


INTRODUCTION
Yeast pre-mRNA splicing (1)(2)(3) begins with the formation of a complex comprising the U1 snRNP bound at the intron 5 0 splice site (5 0 SS; 5 0 -GUAUGU) and the Msl5Mud2 heterodimer engaged at the intron branchpoint (BP; 5 0 -UACUAAC). Bridging interactions between the U1 snRNP and Msl5Mud2 stabilize the complex and prepare a scaffold for recruitment of the U2 snRNP to the branchpoint. The U1 snRNP is ultimately ejected from the pre-mRNAU1U2-containing spliceosome when the U5U4U6 tri-snRNP complex joins en route to forming a pre-mRNAU2U5U6 spliceosome. Dissociation of U1 snRNP is thought to be triggered by the DEAD-box protein Prp28 (4,5), acting to disrupt the short U1:5 0 SS RNA duplex or remodel protein-RNA contacts at the 5 0 SS (or both).
This is not to say that the many dispensable elements of the yeast U1 snRNP are functionally irrelevant. Rather, traditional and genome-wide genetic analyses have highlighted a network of genetically buffered functions during early spliceosome assembly, embracing the U1 snRNP, the Msl5Mud2 branchpoint-binding protein, the TMG cap and the Cbc2Sto1 nuclear m 7 G cap-binding complex (7,(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34). This network was defined by the numerous instances in which null alleles of inessential players, or benign mutations in essential factors, elicited synthetic lethal phenotypes when combined with other benign mutations in the splicing machinery. Such genetic interactions among actors in a common pathway meet an operational definition of redundancy, which does not necessitate that the synthetic interactor proteins or RNA elements perform the same task, but rather suggests that spliceosome assembly can be accomplished or stabilized via different sub-pathways.
Although mutational synergies have played a decisive role in identifying many of the components of the yeast spliceosome (7,18,19,27,29), the interpretation of synthetic phenotypes can be elusive, especially when the atomic structures or biochemical activities of the genetically interacting factors are unknown. However, when structures are available, they can be exploited to program mutations with specific functional defects and then systematically test an allelic series for synthetic genetic interactions with other spliceosome components or splicing factors. This has been applied to the m 7 G-cap binding pocket of yeast Cbc2 (34), guided by the crystal structure of the homologous human CBCm 7 G-cap complex (35,36), and to the branchpoint RNA-binding site of yeast Msl5 (25,26), directed by the NMR structure of the human homolog SF1 bound to an RNA containing the yeast branchpoint consensus sequence (37).
In the present study, we extend this approach to interrogate structure-function relations and genetic interactions of the essential Yhc1 subunit of yeast U1 snRNP. Yhc1 is a 231-amino acid (aa) polypeptide ( Figure 1). Although the N-terminal 40-aa segment of Yhc1 is homologous to the N-terminus of the 159-aa human U1C (27/40 positions of side chain identity or similarity; Figure 2B), the respective C-terminal segments have little or no apparent primary structure similarity (6,38). By serial deletions, we define the distal margin of a minimal functional Yhc1 protein. We then exploit the NMR structure of the N-terminal domain of human U1C (38) (Figure 2A) to direct mutations to the Cys6-Cys9-His24-His30 zinc binding site and to conserved surface residues that we considered candidates to interact with RNA or protein components of the spliceosome. We identify benign mutations of Yhc1 that display allelespecific synergies with other mutations in the spliceosome, especially the absence of Mud2, Nam8, Mud1, and the TMG cap as well as specific perturbations of the 5 0 end of U1 snRNA. We also identify novel Yhc1 mutations that bypass the essentiality of Prp28. We interpret our findings in light of the 5.5 Å crystal structure of the human U1 snRNP that includes the N-terminal domain of U1C (10). Our results add Yhc1 to an extensive reticulum of intramolecular and intermolecular genetic interactions of the U1 snRNP.

Yeast strains and tests of YHC1 function in vivo
To develop a plasmid shuffle assay for gauging mutational effects on Yhc1 function, we generated a yhc1Á strain that relies for viability on maintenance of the wild-type YHC1 gene on a CEN URA3 plasmid, p316-YHC1. In brief, we first replaced the YHC1 locus from position +1 to +688 with a hygMX cassette in the BY4743 diploid strain and then transformed the heterozygous diploid with p316-YHC1. The diploid was sporulated, asci were dissected and haploid yhc1Á Ura + progeny were recovered. The yhc1Á [p316-YHC1] cells were resistant to hygromycin and unable to grow on medium containing 0.75 mg/ml 5-fluoroorotic acid (FOA). To assay the function of wild-type and mutated YHC1 alleles, yhc1Á [p316-YHC1] cells were transfected with CEN HIS3 YHC1 plasmids. Individual His + transformants were selected and streaked on agar medium containing FOA. The plates were incubated at 20, 30 or 37 C, and mutants that failed to form macroscopic colonies at any temperatures after 8 d were deemed lethal. Individual FOA-resistant colonies with viable YHC1 alleles were grown to mid-log phase in YPD broth and adjusted to the same A 600 values. Aliquots (3 ml) of serial 10-fold dilutions were spotted to YPD agar plates, which were then incubated at temperatures ranging from 18 to 37 C.

C-terminal truncations define a minimized functional Yhc1 domain
The size differences and lack of amino acid sequence similarity between S. cerevisiae Yhc1 and human U1C distal to the N-terminal zinc-binding domain raised the prospect that the yeast and human C-termini might have diverged in parallel with the differences in U1 snRNA size and the number of U1-specific protein subunits of the respective U1 snRNPs. By contrast, an alignment of the 231-aa S. cerevisiae Yhc1 to the homologous 189-aa polypeptide from the yeast Kluyveromyces lactis highlighted, in addition to the strong conservation of the N-terminus (35/40 positions of side chain identity or similarity), a modest conservation throughout the distal segment of the respective proteins (64/159 positions of side chain identity or similarity), punctuated by two peptide inserts and a 19-aa C-terminal tail unique to S. cerevisiae Yhc1 ( Figure 1). We used this alignment to guide serial C-terminal truncations of Yhc1, at sites denoted by reverse arrows in Figure 1. The wild-type and truncated alleles were placed on CEN HIS3 plasmids under the control of the native YHC1 promoter and tested by plasmid shuffle for complementation of a yhc1Á p[CEN URA3 YHC1] strain. The resulting YHC1-(1-212), YHC1-(1-189) and YHC1-(1-170) strains were viable after FOA selection and grew as well as wild-type YHC1 cells on YPD agar at all temperatures tested, as gauged by colony size (Figure 1). The YHC1-(1-150) strain was viable but sick at all temperatures ( Figure 1). The YHC1-(1-133), YHC1-(1-116), YHC1-(1-87) and YHC1-(1-65) alleles were lethal, i.e. they failed to complement growth of the yhc1Á p[CEN URA3 YHC1] test strain on FOA at any temperature tested. We conclude that (i) the C-terminal 61 aa are dispensable for Yhc1 function in vivo; (ii) the segment from aa 151-170 is needed for optimal Yhc1 function; and (iii) the segment from aa 134-151 is essential for cell viability.

Effects of mutating the zinc-binding site of Yhc1
The NMR structure of the N-terminal segment of human U1C (38) revealed a globular fold nucleated by a Cys6-Cys9-His24-His30 zinc coordination complex that is conserved in yeast Yhc1 (Figure 2A and B). Mutating human U1C Cys6 or Cys9 to serine was reported to abolish the binding of U1C to the human U1 snRNP, as did mutating His24 or His30 to glutamine (39). To probe the contributions of the four zinc-binding side chains to yeast Yhc1 activity in vivo, we mutated them individually to alanine in the context of full-length Yhc1. The C6A, C9A, H24A and H30A alleles were tested by plasmid shuffle for yhc1Á complementation. Although C6A was and His30 (shown as stick models). Other conserved side chains of U1C that were subjected to mutational analysis in Yhc1 are shown as stick models and labeled in regular font. Non-identical side chains that were mutated are denoted in italics: the label specifies the side chain in human U1C, with the yeast Yhc1 side chain in parenthesis. (B) Alignment of the primary structures of the N-terminal domains of S. cerevisiae (Sce) Yhc1 and human (Hsa) U1C. Positions of side chain identity/similarity are indicated by above the alignment. The a helices are depicted below the alignment as cylinders. The 13 residues of Yhc1 that were subjected to alanine scanning in the present study are highlighted in gold shading. (C) Mutational analysis of the zinc binding site. The C6A, C9A, H24A and H30A mutants were tested for yhc1Á complementation by plasmid shuffle. C6A was lethal. Viable FOA-resistant C9A, H24A and H30A strains were spot-tested for growth on YPD agar at the temperatures specified in parallel with the isogenic wild-type YHC1 strain. unconditionally lethal, we were surprised to find that C9A, H24A and H30A cells were viable. The H24A strain was relatively healthy, growing on YPD agar at all temperatures with colony size slightly smaller than wild-type ( Figure 2C). The C9A and H30A strains grew well at low temperatures (18-20 C) but did not thrive at high temperatures (30-37 C) ( Figure 2C). These results show that Yhc1 function in vivo can be sustained when Cys6 and any two of the other three zinc-binding residues are present. It is conceivable that, when Cys9, His24 or His30 are replaced by alanine, a water can substitute for the missing amino acid sulfur or nitrogen atom in the tetrahedral zinc coordination complex, albeit with some loss of stability/activity, as reflected in the slower growth or temperature sensitive (ts) growth of the respective alanine mutants.

Effects of alanine mutations of conserved surface residues in the N-terminal module
The N-terminal domain of Yhc1/U1C is imputed to make physical contacts with the 5 0 leader sequence of U1 snRNA, the pre-mRNA sequence in the vicinity of the 5 0 SS, and with other protein components of the U1 snRNP (6,10,12,40,41,42). In our view, the likely candidates to mediate these RNA and protein contacts are to be found within the set of conserved basic and hydrophilic side chains that emanate from the surface of the Nterminal domain (Figure 2A). Therefore, we targeted nine surface-exposed residues for alanine scanning: Arg3, Tyr12, Thr14, His15, Arg21, Lys22, Asn29, Lys28 and Asp36 (Figure 2A and B). The YHC1-Ala alleles were tested for yhc1Á complementation by plasmid shuffle and all of them were found to retain biological activity; each of the YHC1-Ala strains grew as well as wild-type YHC1 on YPD agar at all temperatures tested (Supplementary Figure S1).
Synthetic genetic interactions of Yhc1-Ala mutants with Mud2, Nam8, Mud1 and Tgs1 To survey genetic interactions of the apparently fully active Yhc1-Ala mutants, we constructed a series of yeast strains in which the genes encoding inessential splicing factors Mud2, Nam8 or Mud1 were deleted in the yhc1Á p[CEN URA3 YHC1] background, thereby allowing tests of synergy by plasmid shuffle. Although the ablation of Mud2 has no effect on yeast vegetative growth by itself, mud2Á caused synthetic lethality in combination with the YHC1-R21A allele ( Figure 3A). mud2Á elicited strong 'synthetic sick' phenotypes in combination with YHC1 mutants K28A and Y12A, manifested as no growth at 30-37 C and minimal growth at 25 C ( Figure 3A). YHC1-T14A had a cold-sensitive (cs) growth defect in the mud2Á background. The YHC1 mutants R3A, H15A, K22A, N29A and D36A displayed no synergy with mud2Á.
Loss of U1 snRNP subunit Nam8 has no effect on yeast vegetative growth, yet nam8Á caused a strong synthetic growth defect in combination with YHC1 mutant R21A allele, whereby the YHC1-R21A nam8Á strain formed pinpoint colonies at 37 and 34 C and failed to grow at 30 C. By contrast, none of the other eight YHC1-Ala mutants displayed a growth defect in the nam8Á background ( Figure 3B).
Two of the YHC1-Ala mutants had conditional synthetic defects in a yeast strain that lacked the Mud1 subunit of the U1 snRNP. YHC1-R21A mud1Á cells were ts at 37 C and cs at 18 C. YHC1-Y12A cells were ts at 37 C. The seven other Ala mutants grew normally in the mud1Á background (Supplementary Figure S2).
Tgs1 is the enzyme responsible for conversion of m 7 G RNA caps to the m 2,2,7 G trimethylguanosine caps that are a signature feature of the U1, U2, U4 and U5 snRNAs (17,43,44). Yeast tgs1Á cells are viable and grow well at 30-37 C, but they fail to grow at 18-20 C. Here, we constructed a tgs1Á yhc1D p[CEN URA3 YHC1] strain to test synergies between the absence of a TMG cap and the Yhc1-Ala mutants. We found that the YHC1-R21A tgs1Á strain was very sick, forming pinpoint colonies at 37 and 34 C and failing to thrive at 30 C (Figure 4). The YHC1-Y12A tgs1Á strain had a similar strong synthetic phenotype ( Figure 4). The T14A and K28A alleles elicited mild synthetic effects in the tgs1Á background ( Figure 4).

Synthetic genetic interactions of Yhc1-Ala mutants with alterations of the U1 snRNA
The base-pairing interaction between the 5 0 leader sequence 3 ACUUAC 8 of yeast U1 snRNA and the consensus 5 0 SS of yeast pre-mRNAs is depicted in Figure 5. Recently, we reported that mutating U1 snRNA nucleobase U 5 to C had no effect on yeast vegetative growth, signifying that a Py 5 :U mispair in the U15 0 SS RNA hybrid suffices for seemingly normal U1 snRNA activity (25). Yet, we observed strong genetic interactions of U1-U 5 C with mud2Á (unconditional synthetic lethality) and nam8Á (slow growth at 37 and 34 C as well as failure to grow at 30, 25 and 20 C) (25). Here we tested for genetic interactions of U1-U 5 C with the Yhc1-Ala mutants, by constructing a U1Á yhc1Á p[CEN URA3 U1 YHC1] strain, co-transforming it with YHC1-Ala and U1-U 5 C plasmids, screening for growth on FOA and spot testing the FOA-resistant YHC1-Ala U1-U 5 C strains for growth on YPD agar ( Figure 5). The YHC1-R21A U1-U 5 C strain was uniquely sick, forming pinpoint colonies at 37, 34 and 30 C, and failing to thrive at 25 and 20 C. The YHC1-Y12A U1-U 5 C strain displayed a mild slow growth phenotype at all temperatures.
Extending the 5 0 leader sequence by adding 25 nt upstream of the native 5 0 end (as depicted in Supplementary Figure S3) had no effect per se on yeast vegetative growth. However, the U1+25 allele was synthetic lethal with mud2Á and synthetic sick with nam8Á and tgs1Á (25). Here we examined the effects of otherwise benign Yhc1-Ala mutants in the U1+25 background. Again, the YHC1-R21A allele was uniquely synthetically sick with U1+25 (Supplementary Figure S3).
The 5 0 AUACUUACCU 10 single-stranded segment of U1 snRNA precedes the folded U1 RNA tertiary structure that initiates at nucleotide U 11 (i.e. a four-helix junction depicted in Figure 6C). The 8 CCU 10 segment connecting the 5 0 SS complementary motif to helix 1 interacts with the U1C subunit in the U1 snRNP (6,10). We showed previously that the spacing between the 5 0 SS complementary motif and helix 1 is important, insofar as insertions of 5 or more nucleotides between C 8 and C 9 of yeast U1 snRNA were uniformly lethal (25). By contrast, insertion of a single G nucleotide, as in the [+1] allele in Figure 6C, was viable. However, the [+1] insertion was lethal in combination with nam8Á, mud1Á, tgs1Á and mud2Á (25). Here we found that the Yhc1 mutants R21A, K22A and K28A displayed severe synthetic sick phenotypes in a U1-[+1] background ( Figure 6A). The T14A U1-[+1] and H15A U1-[+1] strains had a milder synthetic defect, growing slowly at 25-34 C ( Figure 6A).
Helix 1 is conserved with respect to its presence and position in yeast and human U1 snRNAs. To probe the effects of increasing the length of yeast helix 1, the 9 CU 10 dinucleotide was changed to 9 UA 10 , thereby extending the base-pairing potential to the sequence at the 3 0 end of U1 RNA, such that helix 1 might span 8 bp instead of 6 bp (Supplementary Figure S4). This U1 allele, named H1L ('helix 1 long'), supported normal yeast growth but was synthetic lethal with mud2Á and synthetic sick with nam8Á, mud1Á and tgs1Á (25). Of the collection of Yhc1-Ala mutants, R21A was uniquely synthetic sick in the U1-H1L background (Supplementary Figure S4).
The strongest genetic interactions of the Yhc1-Ala mutants with U1 snRNA were seen when the spacing between the 5 0 SS complementary motif and helix 1 was shortened by deletion of the U 10 nucleotide, as depicted in Figure 6C. The U1-DU 10 single mutant grows normally at 30-34 C but does not thrive at cold temperatures ( Figure 6B). We reported previously that U1-DU 10 was synthetic lethal with nam8Á, mud1Á, tgs1Á and mud2Á (25). Here, we found that Yhc1 mutants T14A, R21A and K28A were synthetic lethal with U1-DU 10 , whereas Y12A, H15A, K22A and N29A displayed severe synthetic sickness in the U1-DU 10 background ( Figure 6B). However, on the contrary, the YHC1-D36A allele elicited a gainof-function effect in the U1-DU 10 background, fully suppressing the U1-DU 10 cs growth defect ( Figure 6C, compare WT and D36A).

Bypass of the essentiality of Prp28 by Yhc1 R21A and K28A mutations
Yeast Prp28 is an essential pre-mRNA splicing factor and a member of the DEAD-box family of nucleic acid-dependent NTPases (45,46). Prp28 is implicated genetically in displacing the U1 snRNP from the 5 0 SS during the transition from a pre-mRNAU1U2-containing spliceosome to a pre-mRNAU2U4U6 spliceosome (4,5,47). The strongest evidence for this idea emerges from the findings that the essentiality of Prp28 for vegetative growth can be bypassed by mutations in the essential U1 snRNP subunits Yhc1, Prp42 and Snu71, and by specific U1 snRNA mutations located within and flanking the segment that base pairs with the intron 5 0 SS (4,25,47). These results highlight that the need for Prp28 during U1 snRNP ejection from the early spliceosome is alleviated by certain alterations that are presumed to weaken U15 0 SS contacts.
Yhc1 figures prominently in the U1 snRNP-Prp28 dynamic, insofar as Chen et al. (4) identified in a genetic screen two independent mutations of the same amino acid in Yhc1 (Leu13 changed to either Phe or Ser) that allowed growth of a prp28Á strain. They proceeded to test all 19 possible substitutions at Leu13 and found that (i) any amino acid at position 13 sufficed for complementation of yhc1Á and (ii) 13 of the amino acids allowed some level of growth of prp28Á (4). They proposed that the conserved Leu13 residue 'makes specific contact' to stabilize the U15 0 SS duplex (4). The subsequent report of the NMR structure of human U1C (38) revealed that the Leu13 side chain points into the globular core of the N-terminal zinc-binding domain, where it makes van der Waals contacts via Cd2 to the His24 side chain that coordinates the zinc ion (Figure 2A). The structure indicates that Leu13 is not in a position to interact with another protein or RNA, making it likely that the Prp28 bypass elicited by Leu13 mutations is an indirect effect of a conformational change in the Yhc1 N-terminal domain.
We envisioned a scenario in which subtraction of surface-exposed side chains of the Yhc1 N-terminal domain might weaken the U15 0 SS contacts and thereby bypass the requirement for Prp28. To put this idea to the test, we constructed a yhc1Á prp28Á p[CEN URA3 PRP28 YHC1] strain, transfected it with wild-type YHC1 and each of the nine YHC1-Ala plasmids and then screened the transformants for growth on FOA, i.e. survival in the absence of PRP28. Two YHC1 mutant alleles passed this test: R21A and K28A. The viable yhc1Á prp28Á cells bearing these bypass alleles were   tested for growth on YPD agar at 20, 25, 30, 34 and 37 C (Figure 7). R21A was the better of the Prp28 bypass suppressors, insofar as YHC1-R21A prp28Á cells grew about as well as wild-type cells at 37, 34 and 30 C, as gauged by colony size (Figure 7). That they grew slowly at 25 C and did not thrive at 20 C signified that the R21A bypass effect is cs (Figure 7). This result suggests that higher temperatures help destabilize the weakened U15 0 SS pairing in YHC1-R21A cells and thereby permit U1 snRNP ejection without the assistance of Prp28. However, when the pairing is more stable at lower temperatures, Prp28 is still required. The YHC1-K28A allele allowed Prp28 bypass at 37 and 34 C, but was less effective at 30 C and ineffective at 25 C (Figure 7).

Interactions of Yhc1 Arg21 with the SmD3 subunit of the Sm ring
Inspection of the low-resolution U1 snRNP structure containing the U1C N-terminal domain (with the higher-resolution U1C fold superimposed) indicates that the Yhc1/U1C Arg21 residue is disposed on the protein surface opposite from the face that contacts the intron 5 0 SS or the 5 0 segment of U1 snRNA ( Figure 8A). Rather, the view in Figure 8B reveals that Arg21 is a likely component of the interface of Yhc1/U1C with the SmD3 subunit of the Sm protein ring. Although atomic contacts cannot be surmised from the structures, the proximity of the basic Arg21 residue to human SmD3 residues His11, Glu36 and Asp37 (H, E and D in Figure 8B) hints that Arg21 might participate in ionic and/or hydrogen bonding interactions with the corresponding side chains (Asn12, Glu37 and Asp38 in yeast SmD3) at the putative Yhc1-SmD3 interface of the yeast U1 snRNP. Thus, we speculated that perturbing the Yhc1-SmD3 interaction accounts for the widespread mutational synergies of YHC1-R21A, whereby splicing depends on Mud2, Mud1, the TMG cap and so forth, because R21A is a hypomorph that affects the formation and/or stability of the U1pre-mRNA[Msl5Mud2] early spliceosome intermediate. This view is consistent with our finding that R21A bypasses the lethality of prp28Á, presumably because the R21A mutation destabilizes the U15 0 SS complex in the later spliceosome intermediate (pre-mRNAU1U2U5U4U6) that is the putative substrate for Prp28-mediated U1 snRNP ejection.
To evaluate this scenario, we established a plasmid shuffle assay to gauge the effects of mutations N12A and E37A-D38A in the putative interface of yeast SmD3 with Yhc1 Arg21. smd3Á cells bearing SMD3-N12A and SMD3-E37A-D38A plasmids were viable and grew as well as smd3Á SMD3 cells on YPD agar at all temperatures tested ( Figure 9A). We then queried the SMD3 alleles for their ability to complement smd3Á mud2Á and smd3Á mud1Á strains. Although no synergies were evident for SMD3-N12A, the E37A-D38A allele was severely synthetically sick in the absence of Mud2, resulting in only microscopic growth on YPD agar at 25 C and 18 C and no growth at higher temperatures ( Figure 9B). Also, the E37A-D38A allele elicited a tight ts growth defect at 37 C in the mud1Á background ( Figure 9B). Thus, loss of the SmD3 Glu37 and Asp38 moieties, one or both of which is poised to make a salt bridge to Yhc1 Arg21, recapitulated several of the synthetic genetic interactions described for R21A, fortifying our hypothesis that the SmD3-Yhc1 interface is pertinent to the genetics of Yhc1 Arg21.
As a further test of this model, we asked whether the SMD3-E37A-D38A allele could bypass prp28Á and found that it did, allowing for growth of SMD3-E37A-D38A prp28Á cells at 37 and 34 C ( Figure 9C). The SMD3-E37A-D38A prp28Á strain grew slowly at 30 C and did not thrive at 25 C, reaffirming that prp28Á bypass mutants are generally cs. The SMD3-N12A allele did not bypass prp28Á. We infer that contacts between Yhc1 Arg21 and SmD3 Glu37 and/or Asp38 stabilize the U1 snRNP5 0 SS complex.

Yhc1-D36A exacerbates cs Prp28 mutations
The finding that the YHC1-D36A allele suppressed the cs growth defect of U1-DU 10 ( Figure 6C) suggested that the D36A change might stabilize interactions of the U1 snRNA or snRNP with the 5 0 SS. If this is the case, the stability-enhancing YHC1-D36A allele ought to synergize with cs mutants of Prp28, which are presumed to be defective in dissociating the U1 snRNP from the spliceosome at low temperatures, when the U15 0 SS base-pairing interactions are more stable, but which can function at higher temperatures that destabilize the U15 0 SS duplex. As a part of a separate effort to chart the structure-function relations of yeast Prp28 by alanine scanning, we identified two mutations that conferred cs growth: K227A and R499A. Lys227 is located within the Prp28 P-loop motif (GSGKT) and is imputed to contact the ATP band g-phosphates. Arg499 is predicted, based on the crystal structure of the homologous DEAD-box protein Vasa (48), to make salt bridges to the Glu342 and Asp344 side chains of the Prp28 DEAD-box. The restrictive temperatures for the PRP28 K227A and R499A strains were 25 and 20 C, respectively ( Figure 10 and data not shown). However, in the YHC1-D36A background, the restrictive temperatures were shifted upward, such that YHC1-D36A PRP28-K227A cells did not grow at 30 C and YHC1-D36A PRP28-R499A cells did not thrive at 25 C ( Figure 10). These results are consistent with the idea that Yhc1-D36A enhances U15 0 SS stability. Asp36 is located within the long a helix that follows the globular N-terminal domain of the U1C protein ( Figure 8B). Interactions of Asp36 cannot be surmised from the structure.

Genetic interactions of Yhc1 C-terminal truncations
In light of the allele-specific mutational synergies involving the N-terminal domain of Yhc1, we proceeded to survey genetic interactions of the benign Yhc1 C-terminal truncations. The results (compiled in Figure  11, top panel) disclosed an informative hierarchy of synthetic mutational effects. YHC1-(1-212) was uniquely lethal in combination with U1 snRNA mutant DU 10 but had no effect on growth in combination with mud1Á, tgs1Á or mud2Á or with U1 snRNA mutants U 5 C, +25 or H1L. The YHC1-(1-212) U1-[+1] strain displayed a mild synthetic defect compared with YHC1 U1-[+1] (scored as + + growth). We conclude that the otherwise dispensable C-terminal segment of Yhc1 from aa 213 to 231 becomes essential under a very specific genetic circumstance.
At the other extreme, the YHC1-(1-170) allele encoding the shortest of the biologically active Yhc1 proteins was lethal in combination with mud1Á, tgs1Á and mud2Á and with U1 snRNA mutants U 5 C, +25, H1L, [+1] and DU 10 ( Figure 11). Yet, YHC1-(1-170) was entirely benign in the nam8Á background (+ + + growth; see Figure 11, bottom panel). This result diverged sharply from the strong synergy of nam8Á with YHC1-R21A ( Figure 3B) and suggested that the N and C domains of Yhc1 have distinctly buffered functions during pre-mRNA splicing.
The intermediate truncation allele YHC1-(1-189) displayed a spectrum of mutational synergies in different genetic backgrounds. Because YHC1-(1-189) mud1Á cells grew about as well as YHC1 mud1Á cells at all temperatures tested (Figure 11, bottom panel), whereas YHC1-(1-170) mud1Á cells were inviable, we surmise that the Yhc1 segment from aa 171 to 189 is genetically redundant with the Mud1 subunit of the U1 snRNP. We found that YHC1-(1-189) mud2Á cells grew well at lower temperatures (18-25 C) but were tightly ts and failed to thrive at 34-37 C (Figure 11, bottom panel). In comparison with the lethality of mud2Á to YHC1-(170) and the normal growth of YHC1 mud2Á cells, this result signifies that the Yhc1 segment from aa 171 to 189 compensates for the absence of Mud2 at low temperatures, whereas the Yhc1 segment from aa 190 to 212 is required to buffer Mud2 loss at high temperatures. The YHC1-(1-189) tgs1Á strain displayed a mild synthetic defect compared with the YHC1 tgs1Á, seen as smaller colony size at 34-37 C (not shown) and scored as + + growth ( Figure 11). Given that tgs1Á was lethal with YHC1-(1-170), we invoke genetic redundancy of Yhc1-(171-189) with the TMG cap.
On the severe end of the synergy spectrum, YHC1-(1-189) was lethal when combined with U1 snRNA allele U1-[+1]. Thus, the Yhc1 segment from aa 190 to 212 is uniquely essential in this genetic background, among those tested. When combined with other U1 snRNA variants, YHC1-(1-189) displayed a ts growth defect with U1-H1L ( Figure 10, bottom panel) and a cs slow growth defect with U1+25, although thriving with U1-U 5 C. Collectively, our findings indicate that the Finally, we tested the three active Yhc1 C-terminal truncation mutants for their ability to bypass prp28Á and found that none was able to do so (not shown). Thus, the prp28Á bypass effect appears to be specific to mutations in the globular N-terminal domain of Yhc1.

DISCUSSION
The present study comprises a mutational analysis of the essential U1 snRNP subunit Yhc1, guided in large part by the NMR structure of the homologous U1C subunit of the human U1 snRNP. Deletion analysis and alanine scanning revealed a surprising (to us) high tolerance for subtraction of Yhc1 structural elements, whereby removal of the C-terminal 61 aa had no impact on yeast growth nor did alanine substitutions for any of the nine conserved and surface-exposed amino acids in the Yhc1 N-terminal domain. Even the signature zinc-binding site in the globular N-terminal domain tolerated non-lethal alanine changes at three of the four zinc ligands.
The low-resolution crystal structure of human U1 snRNP containing the N-terminal domain of the U1C subunit (10) revealed that the globular fold nucleated by the tetrahedral zinc complex (rendered as a light green surface model in Figure 8A) is in proximity to the conserved single-stranded U1 snRNA 5 0 leader sequence 1 AUACUUACCU 10 . In the crystal, the U1 leader is annealed to another RNA strand from an adjacent U1 RNA to form a short duplex that mimics the base-paired segment formed by the conserved U1 snRNA 3 ACUUAC 8 motif (depicted as blue spheres for phosphorus atoms in Figure 8A) and the 5 0 SS of the pre-mRNA intron (rendered as beige spheres for phosphorus atoms in Figure 11. Synthetic genetic interactions of Yhc1 C-terminal truncations. (Top panel) Plasmid shuffle assays were used to test whether the truncated YHC1 alleles were functional in the genetic backgrounds specified in the left column. Synthetic lethality was indicated by failure to form colonies on FOA agar after 8 d at 20, 30 and 37 C. Cultures of viable FOA-resistant cells were grown in YPD broth at 30 C and the growth phenotypes were assessed by spotting serial 10-fold dilutions. Exemplary spot tests are shown in the bottom panel. Growth is tabulated in the top panel as follows: + + + denotes growth similar to the pertinent strains bearing wild-type YHC1; + + indicates smaller colony size than the strain bearing wild-type YHC1; ts signifies a temperature-sensitive growth defect; cs indicates a cold-sensitive growth defect. Figure 8A). In Figure 8, we superimposed the NMR structure of the U1C globular fold on the low-resolution envelope for U1C in the U1 snRNP crystal structure (pdb 3CW1, which does not include atomic models of the amino acid side chains or RNA nucleotides). This model implicates several of the conserved side chains mutated presently as potential contact points to the intron 5 0 SS, the U1 snRNA TMG cap or the H1 helix. Apparently, none of these putative interactions is essential per se for U1 snRNP function.

Novel genetic interactions of the Yhc1 N-domain mutations
Deeper insights emerged from a survey of the collection of fully active Yhc1 mutants for synthetic loss of function phenotypes in nine different genetic contexts: in the absence of the inessential U1 snRNP subunits Nam8 and Mud1; in the absence of the TMG cap on U1 snRNA (and other snRNAs); in the absence of the Mud2 subunit of the Msl5Mud2 branchpoint binding protein complex and in the presence of five different biologically active U1 snRNAs with mutations in or flanking the U1 segment that base-pairs with the 5 0 SS of the pre-mRNA intron. The mutational synergies were many and notable for a hierarchy of severity and allele specificity, as summarized in Supplementary Figure S5.
The R21A mutation in Yhc1 elicited the most severe and widespread effects, being lethal in the absence of Mud2, severely sick in the absence of Nam8 or the TMG cap and mildly sick in the absence of Mud1. The R21A lesion also conferred lethality of severe sickness in the presence of all five U1 snRNA variants tested. We inferred from the superimposed U1 snRNP and U1C structures that Yhc1 Arg21 contacts the SmD3 Glu37 and/or Asp38 side chains, and that loss of these contacts might account for the synthetic growth defects of YHC1-R21A and its ability to bypass prp28Á. Consistent with this hypothesis, we found that mutating SmD3 Glu37 and Asp38 to alanine had no effect on growth per se but mimicked several of the phenotypes seen for YHC1-R21A, including prp28Á bypass, near-lethality with mud2Á and failure to grow at 37 C in the mud1Á background. The genetics implicates the Yhc1(R21)SmD3 interface in stabilizing the U1pre-mRNA[Msl5Mud2] early spliceosome intermediate.
The next most consequential lesion in Yhc1, with respect to synergies, was K28A, which was lethal with U1-DU 10 and severely sick with U1-[+1] and mud2Á (Supplementary Figure S5). The equivalent Arg28 side chain in human U1C is poised to interact with the pre-mRNA intron 5 0 SS:U1 snRNA duplex ( Figure 8A). We attribute the various synthetic phenotypes of YHC1-K28A, and its ability to bypass the lethality of prp28Á, to the loss of a Yhc1RNA interaction that in turn affects the stability of the U15 0 SS complex.
Alanine substitutions for Tyr12 or Thr14 caused strong synthetic growth defects in diverse genetic backgrounds (Supplementary Figure S5). Tyr12 and Thr14 in human U1C are located near the human U1 snRNA (Figure 8A), at the junction of the H1 helix and the duplex stem of the stem-loop 3 (SL3) secondary structure (10). It makes sense that loss of Tyr12 and Thr14 is deleterious in the U1-UD 10 background, where the distance between the 3 ACUUAC 8 motif that pairs with the 5 0 SS and the start of the H1 helix is shortened. The synergy of Y12A and T14A with mud2D suggests that these lesions also impact cross-intron interactions of the U1 snRNP with the branchpointbinding protein.
The YHC1-H15A allele was severely sick in only one of the genetic backgrounds tested, U1-DU 10 , and was mildly sick with U1- [+1] in which an extra nucleotide is inserted between C 8 and C 9 (Supplementary Figure S5). These results resonate with the crystal structure of the U1 snRNP and the aligned U1C N-domain, which places His15 near the short U1 snRNA segment that connects the 3 ACUUAC 8 motif to the start of the H1 helix. The structures suggest that His15 contacts nucleotide 11 of the U1 snRNA (G 11 in human U1; U 11 in yeast U1).

Synergies of Yhc1 C-domain mutations
The C-terminal half of U1C is not present in the crystal structure of the human U1 snRNP. Lack of conservation of this segment in U1C and yeast Yhc1 hints that they might have distinct functions dictated by the significant differences in the RNA and protein composition of the human and yeast U1 snRNPs. By conducting an analysis of the effects of C-terminal truncations of yeast Yhc1, we traced a wide network of genetic interactions between inessential segments of the Yhc1 C-terminal domain and all other components of the early spliceosome that were interrogated here, with the notable exception of Nam8. The nam8Á allele is itself synthetic lethal or severely sick in the absence of Mud1, Mud2 or TMG caps, or when combined with benign mutations in U1 snRNA or an essential splicing factor (17,21,25,26,(32)(33)(34)49), including, as shown here, with the N-terminal R21A mutation in Yhc1. The fact that nam8Á has no effect when combined with the YHC1-(1-170), and that nam8Á and YHC1-(1-170) have a similar spectrum of synergies with other spliceosome components, suggest that they are epistatic, i.e. Nam8 and the Yhc1 C-terminal domain act in the same (genetically buffered) pathway during spliceosome assembly.