Mechanism of assembly of snRNP cores assisted by ICln and the SMN complex in fission yeast

Summary

The spliceosomal snRNP cores, each comprised of a snRNA and a seven-membered Sm ring (D1/D2/F/E/G/D3/B), are assembled by twelve chaperoning proteins in human. However, only six assembly-assisting proteins, ICln and the SMN complex (SMN/Gemin2/Gemin6-8), have been found in Schizosaccharomyces pombe (Sp). Here, we used recombinant proteins to reconstitute the chaperone machinery and investigated the roles of these proteins systematically. We found that, like the human system, the assembly in S. pombe requires ICln and the SMN complex sequentially. However, there are several significant differences. For instance, h_F/E/G forms heterohexamers and heterotrimers, while Sp_F/E/G only forms heterohexamers; h_Gemin2 alone can bind D1/D2/F/E/G, but Sp_Gemin2 cannot. Moreover, we found that Sp_Gemin2 is essential using genetic approaches. These mechanistic studies reveal that these six proteins are necessary and sufficient for Sm core assembly at the molecular level, and enrich our understanding of the chaperone systems in species variation and evolution.

Graphical abstract

Highlights

• •Systematically study the molecular mechanism of snRNP core assembly in S. pombe
• •ICln and the 5-membered SMN complex are essential and sufficient for the assembly
• •Similarity and difference between the vertebrate and S. pombe systems are compared
• •Enrich understanding of the chaperone system in species variations and evolution

Protein; Molecular genetics; Molecular interaction; Experimental systems for structural biology; Model organism

Introduction

Splicing is a fundamental biochemical process in all eukaryotes to remove non-coding segments (introns) within a gene and piece together remaining coding regions (exons) from original transcripts to generate mature message RNAs. This process requires gigantic and dynamic machinery, the spliceosome, which consists of five small nuclear RNAs (snRNAs) and more than a hundred of proteins. The five snRNAs interact with many proteins to form five small nuclear ribonucleoprotein particles (snRNPs), which are the main players in the spliceosome for splicing catalysis.^1,2 According to their protein components and biogenesis, the snRNPs can be classified into two categories: Sm-class and Lsm-class snRNPs. The Sm-class snRNPs include U1, U2, U4, and U5 snRNPs for the major spliceosome (and U11, U12, U4atac, and U5 for the minor spliceosome) and have a common doughnut-shaped architecture composed of seven Sm proteins, D1/D2/F/E/G/D3/B, in this order, around a specific RNA segment, PuAUUUNUGPu (Pu = purine), termed Sm site. The Lsm-class snRNPs include only U6 and U6atac snRNPs, and have a similar doughnut-shaped heteroheptamer, but it consists of a different set of Sm-like proteins, Lsm2-8.^1,2 The biogenesis of the Sm-class snRNPs is more complicated and is mediated by a number of assisting proteins.^2,3 After being transcribed in the nucleus, snRNAs are transported into the cytoplasm, where seven Sm proteins are assembled onto snRNAs to form the common core of snRNPs, which further enables the hypermethylation of the 5′-ends of snRNAs. Both the core formation and hypermethylation serve as the signals for the import of snRNP cores into the nucleus, where additional snRNP-specific proteins are added before they mature and join the splicing process. It will be the main topic in this study. In contrast, the biogenesis of Lsm-class snRNPs is simple, exclusively inside the nucleus, and is believed to require no assistance of assembly chaperones, and therefore it will not be discussed further.

The Sm core formation is a simple and spontaneous process in vitro as demonstrated in early studies. It can be completed simply by mixing the three preformed stable Sm subcomplexes, D1/D2, F/E/G, and D3/B, with a snRNA, or even an oligonucleotide containing just the nonameric Sm site.^4,5 During the assembly, there is a stable intermediate state comprised of D1/D2 and F/E/G and RNA, termed as Sm subcore.⁴ However, Sm core formation inside cells is much more complicated, for in high eukaryotes the number of assisting proteins is larger than that of the major actors. Taking the most studied human cells as an example, 12 proteins, integrated into two complexes—the protein arginine methyltransferase 5 (PRMT5) complex (including 3 proteins: PRMT5, WD45, and pICln) and survival motor neuron (SMN) complex (including 9 proteins: SMN, Gemin2-8, and Unrip)—are involved in assembling the seven Sm proteins on one snRNA.^2,6 In addition, many of these assisting proteins play essential roles for cellular viability. For example, disruption studies of pICln gene demonstrated that pICln is essential for cellular and early embryonic viability in vertebrates,⁷ and either SMN or Gemin2 gene knockout studies caused early embryonic death in vertebrates too.^8,9 Furthermore, the deficient expression of SMN protein also causes spinal muscular atrophy (SMA) in human, a devastating motor neuron degenerative disease, with incident frequency of 1 in 10,000.^10,11 How these trans-acting proteins play their roles has attracted many researchers over the world since the discovery of the role of SMN in snRNP biogenesis in the middle of 1990s.^12,13

More than two decades of efforts have generated many insights into the mechanism of the chaperones-assisted snRNP core assembly in high eukaryotes. The PRMT5 and SMN complexes perform their chaperoning roles in consecutive phases. In the first phase, the PRMT5/WD45 complex methylates the C-terminal arginine residues of SmD3, SmB, and SmD1, which is believed to bind the tudor domain of the SMN protein and enhance the interactions between Sm proteins and SMN.¹⁴ pICln recruits SmD1/D2 and SmF/E/G to form a doughnut-shaped 6S complex, which pre-arranges the five Sm proteins in the correct order and simultaneously prevents the binding of any RNAs.^6,15,16 In addition, pICln also binds SmD3/B.^6,16 In the second phase, the SMN complex accepts SmD1/D2/F/E/G (5Sm) and SmD3/B and releases pICln.^6,16 Gemin2 is a central player for it is not only the acceptor of 5Sm,¹⁷ but also enhances the specificity of RNA binding to 5Sm and mediates the release of the SMN complex from Sm cores.¹⁸ In addition to the tudor domain in the middle, SMN has two highly conserved regions: an N-terminal Gemin2-binding helix, which binds the C-terminal helical domain of Gemin2,^17,19 and a C-terminal self-oligomerized YG box,²⁰ which interacts with Gemin8.²¹ Gemin8 further interacts with Gemin6/7 and Unrip.^21,22,23 Gemin3 contains a DEAD-box domain and is believed to be an RNA helicase.²⁴ Gemin4 usually forms a complex with Gemin3, but its role is unknown.²⁵ Gemin5 is the component to initially bind pre-snRNAs and deliver them to the rest of the SMN complex for assembly into the Sm core.²⁶ Although it had been considered as the main factor determining the RNA specificity in Sm core assembly previously,^27,28,29,30 recent discovery of the role of Gemin2 in RNA selection challenged this idea.¹⁸ Moreover, Gemin5 also plays roles in other processes.³¹

Although Sm cores are highly conserved in all eukaryotes, the chaperone systems are not equally conserved among different species. For example, many species other than vertebrates lack homologs of all or some of Gemin3-8 and Unrip from the SMN complex.^32,33 Although SMN is highly conserved in many eukaryotes, there is no SMN homolog found in the unicellular baker yeast Saccharomyces cerevisiae^32,33 and even the multicellular plant Arabidopsis.³⁴ In addition, despite their high conservation, some chaperones are essential in some eukaryotic species, but are nonessential in others. For example, Gemin2 seems the most highly conserved member in the SMN complex, but the genes of Gemin2 homolog in S. cerevisiae and in Arabidopsis are nonessential.^34,35,36 Another case is pICln: although it is essential for cellular viability in mouse models,⁷ its homologous protein in the unicellular fission yeast Schizosaccharomyces pombe is nonessential.³⁷ Furthermore, some conserved chaperones have significant difference in their sequences among different species. For example, although the orthologous protein of SMN in S. pombe, spSMN (152 residues), had been characterized and found to be essential for viability,^38,39,40 it is much shorter than the human SMN (294 residues) and does not contain in its sequence the important tudor domain, in which several point mutations cause SMA in humans. Therefore, mechanistic studies of the chaperones-assisted Sm core assembly on other species than the vertebrates are of equal importance, for they would generate insights into the variations of the assembly mechanisms and the evolution of the assembly chaperones.

S. pombe is a commonly used unicellular eukaryote model system with easy genetic manipulation. SMN homolog in this species has been genetically and biochemically characterized.^38,39,40 Moreover, the homologous protein of Gemin2 in S. pombe has also been characterized to interact with the N terminus of SMN;³⁸ however, whether Sp_Gemin2 is essential and whether it interacts with Sm proteins have remained elusive. A homolog of pICln in S. pombe, ICln, has also been studied to play an important role in optimal production of spliceosomal snRNPs and efficient splicing in vivo, but the mechanism of ICln in the processing had not been deeply analyzed.³⁷ Recently, homologs of Gemin6-8 were found in S. pombe to interact with SMN and genetic studies showed that they are essential for viability.⁴¹ In addition, interactions of these five members, SMN, Gemin2, and Gemin6-8, of the SMN complex and the oligomeric states of the SMN complex had been characterized.⁴¹ However, these studies had not integrated these many pieces into Sm core assembly for mechanistic characterization and for comparison with the human system, which is currently the only relatively well-characterized chaperone system in metazoans.

In this study, we systematically studied the interactions among three Sm subcomplexes and between them and all these assisting proteins, including ICln and the five members of the SMN complex, in the process of Sm core assembly using reconstituted approaches and also genetically studied the essentiality of Gemin2. These studies filled the above-mentioned gaps and integrated all the information into the molecular mechanism of Sm core assembly, and would enrich our understanding of this assembly chaperone system in species variation and evolution.

Results

Interactions among seven S. pombe Sm proteins

Since Sm core formation is the main event in this study and Sm proteins in S. pombe have not been well characterized biochemically, we first studied the interactions between Sm proteins and complexation among them. With the knowledge of human Sm proteins forming three subcomplexes, SmD1/D2, SmF/E/G, and SmD3/B, we constructed the coding sequences of the corresponding S. pombe Sm proteins in single plasmids and co-expressed them. As expected, D1/D2, F/E/G, and D3/B could be expressed well and purified together from different chromatographic techniques, indicating that each of them does form a stable subcomplex (Figures 1A–1F). We next tested the interactions between each pair of Sm subcomplexes using Ni-beads pull-down assay followed by SDS-PAGE and Coomassie brilliant blue (CBB) staining. D1/HT-D2 pulled down F/E/G significantly (Figure 1A), but not D3/B (Figure 1B). In addition, HT-F/HT-E/G did not pull down D3/B (Figure 1C). This indicates that only D1/D2 and F/E/G have strong interaction. Reversing the HT-tag cleavage on Sm proteins to perform pull-down assay also showed the same results (data not shown).

Figure 1.

Oligomerization states of 3 Sm subcomplexes from S. pombe and their interactions

(A–C) Interaction between each pair of S. pombe Sm subcomplexes, D1/D2 and F/E/G (A), D1/D2 and D3/B (B), and F/E/G and D3/B (C), by pull-down assay. In each pair, there is one Sm subcomplex containing His-TEV (HT-) tag and the other with HT-tag removed. Each pair of Sm subcomplexes were incubated at room temperature for 30 min and pulled-down by Ni-beads, with each Sm subcomplex alone as control. M: markers; S: supernatant; P: pellet.

(D–H) Oligomerization states of D1/D2 (D), F/E/G (E), and D3/B (F), and the interaction between D1/D2 and F/E/G (G), and the three Sm subcomplexes (H) were tested with GFC (left panels) and SDS-PAGE/CBB staining (right panels).

(I) Summary of oligomerization states and interactions of the three Sm subcomplexes. Thick lines with arrows indicate the preferred reaction directions.

We further used gel filtration chromatography (GFC) to study the oligomeric state of each of the three Sm subcomplexes and the complexation among them. As expected, either D1/D2 (Figure 1D, peak at 15.23 mL, calculated apparent molecular weight [MW] = 33.9 kDa) or D3/B (Figure 1F, peak at 16.57 mL, calculated apparent MW = 18.7 kDa) was eluted as a single dimer. Surprisingly, F/E/G was exclusively eluted at 13.66 mL (Figure 1E), corresponding to apparent MW of 68.3 kDa, indicating that it is completely a hexamer. In contrast, although human SmF/E/G has been known to form a hexamer,⁴ it still has a small, but significant fraction of trimer when characterized by GFC (10%–40% of total SmF/E/G).^6,18 The mixture of D1/D2 and F/E/G was eluted with two peaks at positions similar to those of individual F/E/G and D1/D2 (Figure 1G), and the SDS-PAGE/CBB staining showed that the bands of D1/D2 and F/E/G had little overlap. This indicates that although there is a stronger interaction between D1/D2 and F/E/G than the other pairs, they could not form a stable D1/D2/F/E/G pentamer or higher oligomers. This observation is also significantly distinct from the human version of D1/D2 and F/E/G, which could form a significant fraction (∼50%) of D1/D2/F/E/G pentamer.¹⁸ When all the three Sm subcomplexes were incubated and analyzed by GFC, they were eluted with three peaks at positions similar to those of the individual subcomplexes (Figure 1H), indicating that there is little cooperativity among them to form a larger oligomeric form. The previously described observations are summarized in Figure 1I.

SMN enhances the binding of Gemin2 with SmD1/D2/F/E/G

S. pombe Gemin2 had been identified as Yip1p together with S. pombe SMN as Yab1p in 2000.³⁸ However, when S. pombe SMN was characterized to be essential for viability in 3 independent studies,^38,39,40 whether S. pombe Gemin2 is essential has been unknown. In addition, although characterization of the SMN complex in S. pombe has recently been performed, the study was focused on the identification of Gemin6-8, interactions among the member proteins and oligomeric state of the entire SMN complex, and no interactions between Sm subcomplexes and the SMN complex, especially Gemin2, have been studied.⁴¹ With the interactions between human Gemin2 and Sm proteins as a reference,¹⁷ we examined the interactions of Gemin2 or Gemin2 bound by SMNp (p: peptide, residues 1–35), the Gemin2-interacting segment of SMN, with D1/D2 and F/E/G alone or together using pull-down assay. Gemin2 alone pulled down D1/D2 weakly (4.1%), but F/E/G (7.4%) and both together (15.1% D1/D2 and 9.2% F/E/G bound) a little more significantly (Figure 2A). Compared with Gemin2 alone, Gemin2/SMNp pulled down F/E/G similarly (14.1%, 1.9-fold increase), but D1/D2 (19.8%, 4.8-fold increase) and both together (59.0% D1/D2 and 53.5% F/E/G bound, 3.9- and 5.8-fold increases, respectively) more significantly (Figure 2B). This indicates that SMNp binds to the C-terminal D1/D2-binding domain, enhancing the domain’s interaction with D1/D2. We further determined the complexation of Gemin2 alone or Gemin2/SMNp with these two Sm subcomplexes using GFC. The mixture of Gemin2, D1/D2, and F/E/G was eluted with two peaks at positions similar to those of F/E/G and Gemin2 alone and the fractions analyzed by SDS-PAGE/staining showed that F/E/G was eluted before Gemin2 while D1/D2 spanned both peaks (Figure 2C), indicating that Gemin2 alone cannot form a stable complex with D1/D2 and F/E/G. This is significantly distinct from the corresponding human proteins, which can form a stable Gemin2/D1/D2/F/E/G heterohexamer from GFC (unpublished observation). In contrast, the mixture of Gemin2/SMNp, D1/D2, and F/E/G was eluted as a single peak at position of 12.70 mL, which is larger than any of the input components and has an apparent MW of 104.9 kDa (Figure 2D), indicating that a stable Gemin2/SMNp/D1/D2/F/E/G heterohaptamer can form. These observations are in accordance with the previous NMR and SAXS characterization of the C-terminal domain (CTD, residues 95–280) of human Gemin2 alone and in complex with human SMN (26–51), which indicated a conformational alteration of Gemin2-CTD in presence and absence of SMN (26–51).¹⁹ However, this is the first observation that in the absence of the Gemin2-binding domain (Ge2BD) of SMN, the interaction between Gemin2 and D1/D2 was reduced, and the assembly of Gemin2 and D1/D2/F/E/G is almost completely lost. Based on the structure of human SMN_Ge2BD/Gemin2/D1/D2/F/E/G,¹⁷ we made a structural model of S. pombe version of the heteroheptameric complex (Figure S3). In the model, the binding sites of SMN_Ge2BD and D1/D2 on the surface of Gemin2 do not overlap, and therefore, the binding of SMN to Gemin2 allosterically enhances the binding between Gemin2 and D1/D2/F/E/G, emphasizing SMN’s active role in the assembly of the latter. The results are summarized in Figure 2E.

Figure 2.

SMNp enhances Gemin2’s binding to D1/D2/F/E/G

(A and B) binding assay of S. pombe Gemin2 (A) and Gemin2/SMNp (B) to D1/D2 and F/E/G alone and together by Ni-beads pull-down. HT-SMNp was invisible on SDS-PAGE/staining due to its small size for both (B) and (D). M: markers; S: supernatant; P: pellet. The bound D1/D2 and F/E/G percentages were calculated on the basis of the percentages of the loading samples [P: 10%; S: 2% in (A) and 3% in (B)] and the intensities of the bands which were analyzed using the ImageJ software.

(C and D) Complexation assays. Equimolar amount of Gemin2 (C) or Gemin2/SMNp (D) was mixed with D1/D2 and F/E/G and the mixture was subjected to GFC (left panels). The elute fractions (red bar) were analyzed by SDS-PAGE/CBB staining (right panels). The GFC traces of the input components, D1/D2, F/E/G, Gemin2, and Gemin2/SMNp, were included in the same corresponding plots for comparison.

(E) Cartoon model of the allosteric enhancement of SMNp on Gemin2’s binding to D1/D2/F/E/G. Thick lines with arrows indicate the preferred reaction directions.

Effect of Gemin2/SMNp on RNA selection and Sm core release

In our previous study, we used human Gemin2/SMN_Ge2BD to study RNA selection and release of the SMN complex in Sm core assembly and found a negative cooperativity between Gemin2 and RNA in their binding to D1/D2/F/E/G (5Sm).¹⁸ Gemin2 binds outside the horseshoe-shaped 5Sm and constricts the D1-G opening narrow, which exerts RNA selectivity and prefers cognate RNAs, which contain both the Sm site and 3′-stem loop (SL), to assemble; once a cognate RNA is bound inside 5Sm, it widens the conformation of 5Sm, resulting in the release of Gemin2/SMN.¹⁸ So, we tested whether the same mechanism occurs in the unicellular S. pombe system. As in the study of human version,¹⁸ we synthesized U4-snRNA and three derivatives, U4ΔSm (UUUUU in the Sm site was replaced by CCCCC), U4-3’ss (3′-SL was mutated to a single-stranded 2^nd structure), and U4-3′Δ (deletion of 3′SL), and tested their assemblies into the Gemin2/SMNp/5Sm complex (hereafter, we will refer to it as the 7S complex for brevity as in the human case¹⁸). To make a better analysis of component change during the assembly, we used a two-dimensional separation approach, GFC plus SDS-PAGE/staining, which we had used in characterization of the corresponding human system.¹⁸ As a control, we first analyzed the assembly of these RNAs with D1/D2 and F/E/G. Unexpectedly, all the four RNAs incubated with D1/D2 and F/E/G were eluted earlier than RNAs alone (Figures S4B–S4E), although U4ΔSm was supposed to be a negative control. This was due to unavoidable non-specific interactions between D1/D2 and RNA (Figure S4A), which we could not eliminate although we had made several trials including addition of more salt, detergent, heparin, etc., in the buffer. Luckily, by the 2^nd dimensional SDS-PAGE/silver-staining analysis, we observed small amount of F/E/G bands in the RNA-proteins peak fractions in U4ΔSm assembly (Figure S4B), but stoichiometric amounts of F/E/G and D1/D2 bands in the assemblies of U4, U4-3’ss, and U4-3′Δ (Figures S4C–S4E), indicating that the latter three RNAs assembled into D1/D2/F/E/G specifically to form Sm subcores. We next analyzed the assembly of these four RNAs with the pre-constituted 7S complex. The presumed negative control RNA, U4ΔSm, still was eluted earlier than RNA alone, and the peak fractions contained both D1/D2 and F/E/G (Figure S5A), made it hard to conclude for the preferences of assemblies of the other three RNAs (Figures S5B–S5D). Considering that the join of D3/B to Sm subcore facilitates the release of Gemin2/SMN in human case, we tested whether it is also the case in S. pombe system. Compared with U4ΔSm, which still had significant amount of D1/D2 bound non-specifically in the eluted peak (Figure 3A), in U4 assembly (Figure 3B), U4 was eluted together with all the Sm proteins (lanes 3–6), more Gemin2/SMNp was eluted at the position of Gemin2/SMNp alone (∼14.0 mL, lanes 11–12), and less D3/B eluted (16.0–17.5 mL, lanes 15–17). This indicates that for cognate snRNAs, D3/B joining to complete the assembly of Sm core does cause the release of Gemin2/SMNp, which is consistent with our previous conclusion drawn from human case.¹⁸

Figure 3.

U-rich snRNA binding to Gemin2/SMNp/D1/D2/F/E/G and D3/B to form Sm core and release Gemin2/SMNp

(A and B) U4ΔSm (A) or U4 snRNA (B) was mixed with preformed Gemin2/SMNp/D1/D2/F/E/G complex (7S) and D3/B (left panels) and the mixture was subjected to GFC (middle panels) and SDS-PAGE/silver-staining analysis (right panels). For each, one representative result from two independent experiments is shown. The red bar indicated the fractions containing RNA; the orange bar indicated the elute fractions of Gemin2/SMNp alone; the blue bar indicated the elute fractions of the preformed 7S complex; the purple bar indicated the elute fractions of D3/B alone. The red arrow head points to the peak of RNA alone and the blue to the peak of the 7S complex.

Double gene deletion in S. pombe shows that Gemin2 is essential

After the previously described Sm protein binding and RNA assembly studies, we can see that S. pombe Gemin2 is a critical protein in the pathway of Sm core assembly and it should be essential for cell viability. However, since the first genetic knockout study of the SMN gene in S. pombe in 2000,^38,39,40 there has been no conclusive study on the knockout of Gemin2 gene. We searched the PomBase (www.pombase.org) and found that there are two copies of gene coding the Gemin2 protein, yip11 (SPAC19B12.12c) and yip12 (SPAPB17E12.02), located in the same chromosome I, on different strands, separated by about 3.6M bp. Both genes have the same DNA sequence on the coding regions and the sequence identity extends over 800 bp upstream and over 100 bp downstream. Although genome-wide gene deletions had been carried out in S. pombe using a traditional homologous recombination method in 2010, the study showed that the cell phenotype of Gemin2-coding gene deletion, either yip11Δ or yip12Δ, is viable.⁴² So, we speculated that for true characterization of Gemin2 essentiality, both yip11 and yip12 genes should be deleted in the genome and the traditional homologous recombination method might not be efficient for this purpose.

We took a Cas9-assisted homologous recombination approach to achieve the double deletion of both yip11 and yip12 genes (Figure 4A). We first transformed a plasmid, pFA6a-URA3-spGe2, into haploid S. pombe cells. The plasmid contained an Ura3 gene for selection and S. pombe Gemin2 complementary DNA, in which a silent site mutation was introduced at the potential sequence targeted by Cas9-sgRNA to avoid cleavage at it. After the transformed strain was established, we transformed these cells with a second plasmid, pCAS-sgRNA, which would express Cas9 and a specific sgRNA targeting an identical site in both yip11 and yip12 genes, and a mixture of 2 PCR products (1:1) which contained the Nat1 gene flanked by 183 bp upstream and 255 bp downstream sequences of yip11 and yip12, respectively. Cas9-assisted cleavage at yip11 and yip12 genes facilitated homologous recombination. The selected strains were verified by PCRs (Figure 4B) and sequencing (data not shown). Finally, the selected cells were placed on a culture medium containing 5-fluorolactic acid (5-FOA) for viability assay with wild-type cells as a control (Figure 4C). The S. pombe cells containing yip11Δ/yip12Δ and pFA6a-URA3-spGe2 could not grow on FOA-containing plate. This demonstrated that Gemin2 is essential for cell viability, in consistent with the previously characterized biochemical role of Gemin2.

Figure 4.

Deletion of two genes expressing Gemin2 in S. pombe indicates that Gemin2 is essential

(A) The process of knockout of two Gemin2-expressing genes, yip11 and yip12, in S. pombe. In step 1, transformation of a Gemin2-expressing plasmid, the plasmid contained the Gemin2-coding cDNA which also had a silent mutation inside the sequence to avoid sgRNA-targeting in step 3. In step 2, the transformed DNAs included a plasmid containing a specific sgRNA targeting the same sites in both yip11 and yip12 genes and a mixture of 2 PCR products (1:1) which contained the Nat1-coding gene flanked by 183 bp upstream and 255 bp downstream sequences of yip11 and yip12, respectively. In step 3, Cas9-assisted recombination of yip11 and yip12 with the Nat1-coding gene could automatically occur. The genetic state could be analyzed by 2 PCR assays (bottom-right). In step 4, the S. pombe cells were placed on FOA-containing medium. FOA forces the exclusion of the Ura3-containing plasmid, which also containing the only Gemin2-expressing gene, therefore, if cells could not survive, it means that Gemin2 is essential and vice versa.

(B) The genetic state of the yip11/yip12 double-knockout S. pombe cells were confirmed by 2 PCR assays and further by sequencing.

(C) Check the essentiality of Gemin2. S. pombe wild-type cells and the strain containing yip11Δ/yip12Δ and pFA6a-URA3-Ge2 were spotted in serial dilutions on the FOA-containing plate for essentiality test (bottom panel) and on the normal plate as a control (top panel).

Interactions between ICln and Sm subcomplexes

In metazoans, pICln serves as a critical chaperon in the first phase of Sm core assembly before the SMN complex takes over the task⁶ and plays an essential role in cell survival.⁷ However, the homolog in S. pombe, ICln, is nonessential although it is required for optimal growth of cells and associated with optimal production of the spliceosomal snRNPs.³⁷ Interaction of ICln with single Sm protein had been characterized and its interactions with SmD1 and SmD3 had been found.³⁷ However, interactions of ICln with the Sm subcomplexes have not been studied. So, we tested the interactions and complexation of ICln with the three Sm subcomplexes. Initial expression and purification study of full-length ICln showed that it is easy to degrade (data not shown). Taking the previous studies of metazoan pICln^6,15 as references, we turned to a C-terminal truncated ICln (residues 1–160), which removed the nonessential flexible C terminus (see the structural model in Figure S2A and we still refer to it as ICln for brevity). The purified ICln on SDS-PAGE/staining analysis appeared as two bands: at ∼20 kDa, which was expected, and at ∼ 40 kDa, which looked as a dimer (Figure 5A). However, ICln was eluted from GFC as a single peak at 15.75 mL (corresponding to apparent MW of 26.9 kDa), indicating that it is a monomer (Figure 5C). Further study showed that the appearance of dimeric band on SDS-PAGE is influenced by protein concentrations and salt in the buffer (Figure S6). Using Ni-beads pull-down assay, we observed that ICln interacted with D1/D2 (100% bound), but not with F/E/G (0.5% bound), yet when D1/D2 and F/E/G were used together, ICln could pull down all of them (95.0% D1/D2 and 71.7% F/E/G) (Figure 5A). These phenomena were confirmed by GFC complexation assay (Figures 5C–5E). ICln forms a stable complex with D1/D2 (Figure 5C), but no complex with F/E/G (Figure 5D). The mixture of ICln, D1/D2, and F/E/G was eluted as a single peak at 13.63 mL (Figure 5E), corresponding to MW of 69.2 kDa, indicating that the complex is a stable, compact hexamer (termed 6S as in the metazoan case⁶). In addition, pull-down assay also showed that ICln has a strong interaction with D3/B (88.9% bound) (Figure 5B) and GFC complexation assay confirmed that ICln and D3/B form a stable heterotrimer (Figure 5F, peak at 14.58 mL, apparent MW of 45.3 kDa). The results are summarized in Figure 5G, in which, significantly, although F/E/G does not interact strongly with either D1/D2 (weaker than human case) or ICln, ICln helps D1/D2 to recruit F/E/G to form a stable complex. Considering the doughnut-shaped structure of fly pICln (Δ90–125 and H144A) complexed with human SmD1/D2/F/E/G,¹⁵ the double interfaces of ICln/D1/D2 to contact F/E/G may shift the equilibrium to the formation of 6S.

Figure 5.

Interaction between S. pombe ICln and 3 Sm subcomplexes

(A and B) binding assay of S. pombe ICln to D1/D2 and F/E/G alone and together (A) and to D3/B (B) by Ni-beads pull-down. Purified HT-ICln showed an expected band at about 20 kDa and an additional band at about 40 kDa (labeled as dimer) on SDS-PAGE/staining. The dimerization state was further analyzed (see Figure S6). M: markers; S: supernatant; P: pellet. The bound D1/D2, F/E/G, and D3/B percentages were calculated on the basis of the percentages of the loading samples (P: 12%; S: 2.4%) and the intensities of the bands which were analyzed using the ImageJ software.

(C–F) Complexation assays. Equimolar amount of ICln was mixed with D1/D2(C), F/E/G(D), D1/D2, and F/E/G together (E) or D3/B (F), and the mixture was subjected to GFC (left panels). The elute fractions (red bar) were analyzed by SDS-PAGE/CBB staining (right panels). For each, one representative result from two independent experiments is shown. The GFC traces of the input components, ICln, D1/D2, F/E/G, and D3/B, were included in the same corresponding plots for comparison.

(G) Cartoon model of ICln interactions. Thick lines with arrows indicate the preferred reaction directions.

The SMN holo-complex efficiently displaces ICln

Because ICln in the stable 6S complex blocks the binding of RNA to 5Sm, 6S has to transfer to the SMN complex to release ICln. Therefore, we tested which members of the SMN complex perform this role, using reconstituted Gemin2/SMNp, Gemin2/SMN (Ge2S), and Gemin2/6/7/8/SMN (Ge2678S). The reconstituted Ge2S was eluted at 8.96 mL, much earlier than Gemin2/SMNp (13.7–14.0 mL), with apparent MW of 556 kDa (Figures 6A and F), much larger than the MW of a single heterodimer. It is due to the oligomerization of SMN at its C-terminal YG box^41,43 as well as a long flexible unstructured sequence in the middle (See the structural model of SMN in Figure S2B). The reconstituted Ge2678S was eluted at 9.18 mL, with apparent MW of 504 kDa (Figures 6B and 6G), indicating that the join of Gemin6/7/8 does not increase the volume of the complex, rather makes it more compact. The GFC behaviors of both complexes were similar to those reported recently.⁴¹ Due to the similar size of ICln to SMN and Gemin8, we removed all HT-tags from the previously described three reconstituted complexes and kept HT-tag on ICln to identify it on SDS-PAGE by western blot (Figure 6C). In addition, HT-tag was also kept on D2, F, and E for reconstitution of the 6S complex (Figure 6C).

Figure 6.

Full-component SMN complex causes release of ICln

(A) Reconstitution of HT-tag-removed Gemin2/SMN complex (Ge2S) analyzed by GFC. The GFC trace was showed in blue in panel (F) and the peak fractions were analyzed by SDS-PAGE/CBB-staining.

(B) Reconstitution of HT-tag-removed Ge2/Ge6/Ge7/Ge8/SMN complex (Ge2678S) analyzed by GFC. The GFC trace was showed in blue in panel (H) and the peak fractions (P1 in blue) were analyzed by SDS-PAGE/CBB-staining. The components for reconstitution, Ge2, SMN, and Ge6-8 (Input), all with HT-tag cleaved, were separately purified from E. coli.

(C) The HT-tagged components, including ICln, D2, (F) and (E), used for constitution of the 6S complex were checked by CBB staining and anti-His antibody.

(D and E) SMNp/Ge2 and 6S form a larger complex without ICln release. Purified SMNp/Ge2 and 6S were mixed and subjected to GFC (D) and SDS-PAGE/CBB-staining analysis (E). The GFC traces of SMNp/Ge2 and 6S were included in the plot for comparison.

(F and G) Ge2S binds 6S and causes ICln to release inefficiently. Purified Ge2S and 6S were mixed and incubated, and the mixture was subjected to GFC (F) and the elute peak fractions were subjected to SDS-PAGE followed by CBB-staining (G, top) and anti-His immunoblot analysis (G, bottom). The GFC traces of Ge2S and 6S were included in the plot for comparison.

(H and I) Full-component SMN complex, Ge2678S, binds 6S and cause ICln to release efficiently. Purified Ge2678S and 6S were mixed and incubated, and the mixture was subjected to GFC (H) and the elute peak fractions were subjected to SDS-PAGE followed by CBB-staining and anti-His immunoblot analysis (I). The GFC traces of Ge2678S and 6S were included in the plot for comparison. The entire fractions of peaks P2 and P3 from the GFC analysis of the mixture of Ge2S + 6S (F and G) and Ge2678S + 6S (H and I) are shown in Figure S7. Asterisks indicate a small amount of impurities carried from purified SMN. Pound signs indicate trace amount of potential dimer band from D2 detected in western blot.

The mixture of Gemin2/SMNp and 6S was eluted as a single peak at 13.75 mL, a little earlier than Gemin2/SMNp alone and similar to 6S alone (Figures 6D and 6E), corresponding to an apparent MW of 65.6 kDa. Because the peak position was close to those of both the input complexes, it was possible that the two input complexes did not form a complex. To clarify it, we took the peak fraction to run through ion-exchange chromatography (IEC), a Mono Q column, and found that there still was a single peak eluted (Figures S7A–S7B). As in the SDS-PAGE analysis of the GFC peak, the IEC peak contained all the components, indicating that Gemin2/SMNp accepted D1/D2/F/E/G but could not displace ICln. When we tested the mixture of Gemin2/SMN and 6S by GFC, there were three peaks eluted, 8.96, 13.69, and 15.69 mL, corresponding to the peak positions of Gemin2/SMN, 6S, and ICln, respectively (Figures 6F–6G, also Figures S7C and S7D). The appearance of ICln alone indicated that Gemin2/SMN is able to displace ICln. Examination of the peak components at the position of Gemin2/SMN showed that small percentage of ICln and D1/D2/F/E/G was bound to Gemin2/SMN, which means that the displacement of ICln was inefficient. Moreover, it is quite surprising that Gemin2/SMN did not bind D1/D2/F/E/G as much as Gemin2/SMNp did (compare Figures 6E and 6G), indicating that the full-length SMN plays some negative effects on the binding of D1/D2/F/E/G to Gemin2. When we examined the effect of Ge2678S, the holo-SMN complex in S. pombe known so far, on 6S by GFC, we also observed three elute peaks at positions similar to those of the two input complexes and ICln (Figures 6H–6I, also Figures S7E and S7F). The separate ICln was eluted indicating the displacement of it from D1/D2/F/E/G. The earliest elute peak fractions contained all the Sm proteins at much higher percentage than those in the Ge2S case. More significantly, they did not contain any ICln indicating an efficient displacement of ICln from 5Sm. We concluded that only when all the five members are present can the SMN complex efficiently accept 5Sm and displace ICln from 6S.

Discussion

In this study, we used reconstituted S. pombe proteins and protein complexes and in vitro transcribed RNAs to biochemically examine the steps of Sm core assembly mediated by ICln and the SMN complex and also genetically investigated the essentiality of Gemin2 in S. pombe. This study fills many unanswered gaps and provides an overall mechanism for the entire process of chaperones-assisted Sm core assembly in unicellular eukaryotes for the first time. In addition, this study also provides an opportunity for comparison with the corresponding human system, the only relatively well-studied system among metazoans. The assembly mechanism is summarized in Figure 7 and it requires two phases, mediated by ICln and the SMN complex sequentially, which is similar to the human system. Nevertheless, there are significant variations at several steps between them. We will discuss these processes in the following section, with an emphasis on comparison with the human system. Although the three Sm subcomplexes are stable, they adopt different oligomeric states. While both D1/D2 and D3/B are single heterodimers, F/E/G is a dimer of heterotrimer (Figures 1 and 7). Compared with human F/E/G, which coexist in both hexameric and trimeric states, although the proportion of the latter is small,^6,18 S. pombe F/E/G has almost no heterotrimer detected, indicating that the hexameric state is more stable and little heterotrimer is dissociated from it (Figure 7A). This also explains why there was no further formation of the heteropentamer D1/D2/F/E/G in GFC compared to the human system (Figure 1). Although ICln does not interact with F/E/G, it can form a stable heterotrimer with D1/D2 (Figure 7B); the ICln/D1/D2 complex may provide two interfaces to interact with F/E/G to form the 6S complex and facilitate the assembly of 5Sm (Figure 7C). ICln interacts with D3/B to form a stable heterotrimer (Figure 7I), which may just serve as a storage of D3/B or have little beneficial effect in Sm core assembly. Although the SMN holo-complex consists of 5 member proteins, it is still Gemin2 that binds 5Sm (Figure 7D). However, Gemin2 alone cannot bind 5Sm stably, which is different from human Gemin2. In addition to its binding effect on Gemin2, SMN_Ge2BD also significantly enhances the binding of Gemin2 to 5Sm and the 6S complex, which reveals the active role of SMN in binding to 5Sm. However, Gemin2/SMN_Ge2BD cannot displace ICln. Even Gemin2/SMN cannot efficiently replace ICln. Meanwhile, Gemin2/SMN causes a loss of binding between 5Sm and Gemin2 in an unidentified way. The holo-form of the SMN complex can displace ICln completely from 5Sm (Figure 7E), which reveals the importance of Gemin6-8 and the C-terminal YG box of SMN. Although there is no direct structural evidence, the opening of the horseshoe-shaped 5Sm bound by Gemin2 in the 5Sm-SMN complex after ICln release should be narrow, as snRNA binding into 5Sm and the join of D3/B cause the release of Gemin2/SMNp from the Sm core, which is consistent with the human system. Therefore, the binding of a cognate RNA into 5Sm should widen the opening of 5Sm (Figure 7F), which in turn reduces the affinity between Gemin2 and 5Sm and triggers the dissociation of the Sm subcore from the SMN complex (Figure 7G). Because both formation and breakup of interactions with 5Sm are involved in this reaction (treating steps f and g as one reaction), neither direction would be preferred. However, since the Sm core is a very stable complex, the binding of D3/B to the Sm subcore should be a spontaneous process (Figure 7H), which can couple with the previous reaction and drive it to form the Sm core. In addition, since there are more interactions formed in this reaction than in the binding of ICln to D3/B, the Sm subcore can successfully compete with ICln to dissociate D3/B from ICln (Figure 7I). Therefore, with the assistance of ICln and the five-membered SMN complex, the Sm core can be efficiently assembled and the assembly chaperones can be recycled. Hence, at the molecular level, the six assisting proteins in S. pombe are both essential and sufficient for Sm core assembly. This is significantly different from the human system, in which additional proteins (PRMT5/WD45, Gemin3/4/5, and Unrip) are important and the process also requires ATP as demonstrated in the experiments using cell extracts.^44,45 In the human system, the methylation of RG-rich tails of SmD1, SmD3, and SmB by PRMT5/WD45 might be important for interactions with the tudor domain in the middle of SMN, which is absent from S. pombe SMN. In human, Gemin5 has been demonstrated to bind to snRNAs^{26,27,28,29,30} and also interact with the others in the SMN complex,^23,46,47 Gemin3 is an RNA helicase and contains an ATPase domain,²⁴ which might use ATP to transfer the Gemin5-bound snRNAs to the SMN complex. However, there were no homologous proteins co-precipitated from the five-membered SMN complex.⁴¹

Figure 7.

The mechanism model of Sm core assembly assisted by ICln and the SMN complex in S. pombe

The thickness of the arrow lines indicates the preferred direction. The complexes with square brackets are unstable or transient. The SMN protein is showed in dimeric form, with two sets of Gemin6-8 attached on its C-terminal YG box domain. See details in the main context.

Although the essentiality of SMN had been clarified in 2000, whether Gemin2, the SMN-interacting protein identified also in 2000, is essential in S. pombe had remained unclear. It is surprising that there are two Gemin2-coding genes, yip11 and yip12, in the genome of S. pombe. Although genome-wide gene deletion study had been performed, in the documented database, either yip11Δ or yip12Δ show no abnormal phenotype.⁴² Considering that both genes are identical in their coding sequences and flanking sequences for homologous recombination, it is hard to deduce whether the documented phenotype was from deletion of both genes or of just a single gene. What makes it even more confusing are the facts that although Gemin2 in many species such as mice and fruit flies is essential,^9,48 deletion of Gemin2 homologs in some other species such as S. cerevisiae and Arabidopsis had been reported to have little effect on survival.^34,35,36 To clarify whether Gemin2 is essential, we made double gene deletion study and demonstrated that Gemin2 is essential for S. pombe. This fills the last gap in the genetic characterization of the SMN complex. Together with previous genetic studies,^{37,38,39,40,41} of the six known proteins assisting in Sm core assembly in S. pombe, all the five members of the SMN complex are essential and ICln is nonessential but required for optimal growth of S. pombe. Our biochemical studies also provide the molecular basis for the essentiality of these assisting proteins. The root of their essentiality arises from the inability of the three Sm subcomplexes to further form any stable complex among them. ICln is largely needed for optimal cell growth because it can form a stable heterotrimer with D1/D2 and the ICln/D1/D2 complex can provide two interfaces to interact with F/E/G to form the 6S complex in competition with F/E/G dimerization. So, ICln facilitates the association of D1/D2 and F/E/G to assemble into the same order as they are in the final Sm core. However, in the same time, they form a closed ring and the location of ICln blocks the entry of RNA, which requires a mechanism to displace ICln. Here comes the SMN complex to play this role. Gemin2, with SMN_Ge2BD bound, can bind the peripheral surface of 5Sm. However, only SMN and Gemin2 are unable to displace ICln from 5Sm (Figure 6). Gemin6-8 can bind to the C-terminal YG box of SMN and facilitate the dissociation of ICln from 5Sm (Figure 6). This explains why the entire SMN complex is essential for cell viability. Consistent with it, recent in vivo replacement study of the C-terminal YG box of SMN by artificial oligomers, which although mimicked its oligomeric states, would not recruit Gemin6/7/8, failed to complement smn1 gene deletion in S. pombe.⁴⁹ Possibly because Gemin2, with SMN_Ge2BD bound, can bind D1/D2/F/E/G (Figure 2), this makes ICln nonessential. However, because the interaction of Gemin2 with either D1/D2 or F/E/G alone is not substantially strong, it is rational that inside the cells crowded with millions of proteins, it is inefficient for Gemin2 to bind both D1/D2 and F/E/G. This explains why ICln, although nonessential, is needed for optimal growth of cells. In line with this molecular mechanism of snRNP core assembly assisted by these chaperones, ICln disruption decreases the levels of U-rich snRNPs and produces differential pre-mRNA splicing defects in fission yeast³⁷; Gemin8 defect in S. pombe by temperature-degron construction under non-permissive temperature displays reduced levels of Sm-class snRNPs as well as reduced splicing⁴¹; Gemin2 overexpression in S. pombe causes retention of Sm proteins in the cytoplasm and becomes toxic.⁵⁰ All these observations support that the essential or important roles of these chaperones in S. pombe are through snRNP biogenesis and splicing.

Besides the human and S. pombe systems, the homologs of the snRNP core assembly chaperones from other species have also been searched and studied. In the PRMT5 complex, pICln seems the most conserved protein in almost all species. The deletion of pICln causes embryonic death in mice,⁷ while the decrease of its expression causes SMA-like phenotypes in zebrafish and Drosophila.^51,52 Even in the species such as S. pombe and Arabidopsis, where pICln homologs are nonessential, the disruption of their genes causes pre-mRNA splicing changes.^37,53 PRMT5 homologs have also been identified in almost all species.⁵⁴ However, because they are involved in various biological processes by providing symmetrical dimethylation of broader substrates, such as histones, methylation of the tails of certain Sm proteins in some species seems less important than in vertebrates. For example, Sm protein methylation is dispensable for snRNP assembly in Drosophila melanogaster.⁵⁵ As for the components of the SMN complex, SMN, Gemin2, Gemin3, and Gemin6-8 seem highly conserved in metazoans with the exemption of Diptera, in which the homologs of Gemin6-8 had not been found in early studies.^32,33 However, recent studies suggest that the homologs of Gemin6-8 exist in Drosophila melanogaster,⁵⁶ although it is still controversial and needs further verification.⁵⁷ Moreover, Gemin5 is also highly conserved in metazoans and its homologs are not found in Nematoda only.^32,33 The conservations of Gemin4 and Unrip are less than the other members: the homologs of Gemin4 had been found only in Deuterotomia, and the homologs of Unrip had not been found in many species of Ecdysozoa.^32,33 The number of homologs of the SMN complex in fungi and plants seems the least conserved, and in early studies, only SMN and Gemin2 have been found in most of them.^32,33 Research in recent years has updated some of these components, such as the discovery of SMN not present in Arabidopsis,³⁴ but the discovery of Gemin6/7/8 homologs in S. pombe.⁴¹

Compared with 12 assisting proteins in total in the vertebrate system and more than six in metazoans, the system in S. pombe contains only six proteins, ICln, SMN, Gemin2, and Gemin6-8. As discussed previously, our studies demonstrated that they are not only essential but also sufficient for Sm core assembly in S. pombe. Therefore, the system in S. pombe is indeed a miniature chaperone system for Sm core assembly. Considering the high conservation of the C-terminal YG box of SMN and its role to recruit Gemin6/7/8 to efficiently displace ICln, we support the idea that the homologs of Gemin6-8 exist in Drosophila⁵⁶ and predict that there may be Gemin6/7/8-homologous proteins in many species^32,33 in which the SMN protein has been identified as essential, even though Gemin6/7/8 homologous proteins have not yet been identified. All the other six proteins in vertebrate systems might have joined later in the evolution of the chaperone system, playing additional regulatory roles.

How the SMN complex displaces ICln remains unclear. It may be hard to determine the entire SMN complex structure because there are substantial long flexible linkers between the N-terminal Ge2BD and the YG box in SMN, and many flexible segments in Gemin8 (Figure S2). In addition, the replacement of ICln by the SMN complex seems a dynamic process and it may be hard to capture. Nonetheless, because of its miniaturization, the chaperone system in S. pombe may be more suitable for further structural and mechanical investigation than the human system. Combining easy genetic manipulation, the system in S. pombe may also be used for characterization of many mutations that cause SMA in human SMN.

Besides the species containing Sm core assembly chaperone proteins like S. pombe, there are also some species, such as S. cerevisiae^32,33 and Arabidopsis thaliana³⁴ in which SMN was reported to be absent and even the homologous proteins of Gemin2 are nonessential. In these species, there might be difference in the interactions between Sm subcomplexes and/or the structure and function of Gemin2. We note that Gemin2 in these species (341 residues in S. cerevisiae, and 515 residues in A. thaliana) are quite longer than those of Gemin2 in human (280 residues) and S. pombe (235 residues) and they might play a role like the entire SMN complex. We are also undertaking the biochemical and structural characterizations of these systems. Like this study, studies of these chaperone systems in different species other than vertebrates would enrich our understanding of the mechanism and evolution of Sm core assembly.

Limitations of the study

In this study, due to the unexpected non-specific interaction between RNAs and Sp_SmD1/D2, no conclusion has been drawn on the preference of different types of RNAs, including the RNAs containing the Sm site followed by a single-stranded sequence (i.e., U4-3’ss) and the RNAs containing just the Sm site at 3′-end (i.e., U4-3′Δ), being assembled into Sm cores in S. pombe.

Moreover, the molecular mechanism how the SMN complex (Gemin2/SMN/Gemin6-8) displaces ICln from the 6S complex has not been revealed.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
HRP-conjugated His-tag monoclonal antibody	Proteintech	Cat# HRP-66005; RRID: AB_2857904
Bacterial and virus strains
DH5α	Beijing Genesand Biotech Co.,Ltd	Cat#SCC12
BL21 (DE3)	Beijing Genesand Biotech Co.,Ltd	Cat#SEC14
Chemicals, peptides, and recombinant proteins
Nourseothricin Sulfate	Solarbio	Cat# N9210
G418 sulfate	BBI	Cat# A600958
5-fluorolactic acid (5-FOA)	Thermo Fisher Scientific	Cat# R0812
Ampicillin sodium	BBI	Cat# A610029
Kanamycin sulfate	BBI	Cat# A600286
Streptomycin sulfate	BBI	Cat# A610494
Isopropyl β-D-thiogalactopyranoside(IPTG)	BBI	Cat# A600168
MES monohydrate	BBI	Cat# A610341
Tris (hydroxymethyl) aminomethane	BBI	Cat# A610195
Imidazole	BBI	Cat# A600277
Sodium dodecyl sulfate(SDS)	BBI	Cat# A600485
PVDF Transfer Membrane	Solarbio	Cat# BSP0161
Critical commercial assays
Monclone Single Assembly Cloning Mix	Monad Biotech	Cat#MC40201
T7 high yield RNA Transcription kit	Vazyme	Cat#TR101
ECL Supers Signal West Pico PLUS	Beyotime Biotechnology	Cat#P0018AS
Protein Marker	Thermo Fisher Scientific	Cat#26617
DNA Marker	Thermo Fisher Scientific	Cat#SM0331
Experimental models: Organisms/strains
The haploid strain 503 of Schizosaccharomyces pombe	Gift of Dr. Cong Liu (Sichuan University)	N/A
Oligonucleotides
U4ΔSm	In this paper	N/A
U4	In this paper	N/A
U4-3’ss	In this paper	N/A
U4-3’Δ	In this paper	N/A
Software and algorithms
CHOPCHOP	http://chopchop.cbu.uib.no/	yip11; yip12
SWISS-MODEL	https://swissmodel.expasy.org/	N/A
Pombe UniProt database	https://www.uniprot.org/proteomes/UP000002485	5,117 entries, (2012/11)
Rosetta	https://www.rosettacommons.org/	N/A
ImageJ	https://imagej.nih.gov/ij/	N/A
Other
Ni-NTA agarose beads	Changzhou Smart-Lifesciences Biotechnology Co.,Ltd.	Cat#SA004
Superdex 200 Increase 10/300 GL	Cytiva	Cat#28990944
Q Beads 6FF	Changzhou Smart-Lifesciences Biotechnology Co.,Ltd.	Cat#SI001
SP Beads 6FF	Changzhou Smart-Lifesciences Biotechnology Co.,Ltd.	Cat#SI003

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Rundong Zhang (rundongzhang@scu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Experimental model and study participant details

Cell lines and culture

The haploid strain 503 (h+ ade6-704 leu1-32 ura4-D18 his3-D1) of Schizosaccharomyces pombe was obtained from Dr. Cong Liu (Sichuan University, China), and cultured in YES or EMM2 medium at 30°C. The yip11/yip12 knockout strain cells were cultured in EMM2 medium containing 100 μg/mL nourseothricin (NTC), 50 μg/mL G418, 0.25 g/L each of histidine, adenine, lysine and leucine, but no uracil (EMM2-ΔURA).

Method details

Construction of plasmids for protein expression in E. coli

The genes of seven Sm proteins, ICln and the five member proteins of the SMN complex were all cloned from complementary DNAs (cDNAs) which were reverse transcribed from the S. pombe (Sp) RNA extracts. For the seven Sm proteins, they were constructed into 3 plasmids. Full-length SmD1 and SmD2 were constructed in the vector pCDFDuet with a tag sequence coding hexameric histidine followed by Tobacco Etch Virus (TEV) cleavage site (His₆-TEV-tag or HT-tag) at the N-terminal end of SmD2 (pCDFDuet-HT-SpD2-SpD1). Full-length SmF, SmE and SmG were constructed in the vector pCDF as two versions: with His₆-TEV-tag at the N-terminal ends of both SmF and SmE (pCDF-HT-SpF-HT-SpE-SpG) and with His₆-TEV-tag at the N-terminal end of only SmE (pCDF-SpF-HT-SpE-SpG). For SmD3/B, the C-terminal truncated SmB (residues 1-93) and SmD3 (residues 1-83), which keep the Sm fold and remove the nonessential C-terminus for better expression, were used and constructed in the vector pCDFDuet with His₆-TEV-tag at the N-terminal end of SmB (pCDFDuet-HT-SpB-SpD3). For the Sm core assembly assisting proteins, The C-terminal truncated ICln (residues 1-160), which removes the nonessential flexible C-terminus for better expression, was constructed in pCDF vector with His₆-TEV-tag at the N-terminus (pCDF-HT-SpICln). Full-length Gemin2 was constructed in pCDF vector with His₆-TEV-tag at the N-terminus (pCDF-HT-SpGe2) for single protein expression. Full-length Gemin2 was also constructed with the N-terminal Gemin2-binding segment of SMN (residues 1-35, SMNp, p: peptide) in the single vector pCDFDuet for co-expression (pCDFDuet-HT-SpSMNp-SpGe2), in which His₆-TEV-tag was only fused in the N-terminus of SMNp. Full-length SMN was constructed in the vector pRSF with His₆-TEV-tag fused at its N-terminus (pRSF-SpSMN). Full-length Gemin6, Gemin7 and Gemin8 were constructed in the vector pCDF with His₆-TEV-tag at the N-terminus of Gemin8 (pCDF-HT-SpGe8-SpGe6-SpGe7). In the plasmids, each gene was controlled under a single T7 promoter and has a ribosomal binding site. The insert fragments and the vector fragments were linked by a homologous recombination method, seamless cloning, via Monclone Single Assembly Cloning Mix (Monad Biotech, China). All the constructs were verified by DNA sequencing.

Protein expression and purification

The recombinant proteins and protein complexes were all expressed using BL21 (DE3) competitive cells. The transformed cells were cultured in Luria-Bertani (LB) medium supplemented with appropriate antibiotics (50 μg/mL), at 37°C and 200 rpm, until the OD_600nm reached 1.0. 0.25 mM IPTG was added to induce protein expression and the cells continued to grow at 16°C and 200 rpm for 18 h. Cells were collected by centrifugation and cell pellet was suspended in lysis buffer (50 mM Tris-HCl, 250 mM NaCl, 20 mM imidazole, 5% glycerol, pH 8.0). The cell suspension was either snap frozen in liquid nitrogen and stored at - 20°C for use later or directly used for purification.

The cells were broken by passing a high-pressure cracker (700 Bar, 4°C, 5 min) and the supernatant was collected after centrifugation. Proteins or protein complexes were purified generally by Ni-column first, ion-exchange chromatography second and gel filtration chromatography (GFC) at last. If the removal of His₆-TEV-tag was needed, TEV protease cleavage was performed after Ni-column purification, and the tag-removed proteins were collected by passing Ni-column again at low concentration of imidazole in the buffer. At each step, the quality and quantity of proteins were monitored by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and Coomassie brilliant blue (CBB) staining.

Protein-protein interaction by pull-down assay

In some cases such as for interactions between each pair of Sm subcomplexes, about 50 μg of Ni-NTA agarose beads were incubated with equal molar amount of HT-tagged protein (or complex), tag-removed protein (or complex) or both in binding buffer (50 mM Tris-HCl, 250 mM NaCl, 20 mM imidazole, 5% glycerol, 0.01% (v/v) NP-40, and protease inhibitors, pH 8.0) at 4°C for 1h. The supernatants were collected after centrifugation (600×g, 3 min), and the beads were then washed with 1 mL binding buffer, three times. After washing, the proteins bound on beads were resolved by SDS-PAGE and CBB staining. In other cases in which one protein was tested to interact with several other proteins, such as for interactions between HT-tagged Gemin2 with SmD1/D2, SmF/E/G and both, the first step took a variation while the remaining steps kept the same. The variation is, Ni-NTA agarose beads were incubated with HT-tagged protein first, and then aliquoted into different tubes, in which different tag-removed proteins were added for interaction.

Protein complexation by GFC

For protein complexation test, the purified proteins or complexes was mixed at equimolar ratio and incubated at 4°C for at least 1 h. After the mixture was centrifuged at 13,000×g at 4°C for 15 min to remove any debris, the supernatant passed an analytical gel filtration column, Superdex 200 Increase 10/300 GL (GE healthcare, USA), in the running buffer (25 mM HEPES, 250 mM NaCl, 1 mM EDTA, pH 7.5). The eluted fractions were collected at each 0.5 mL and analyzed by SDS-PAGE and staining. The apparent molecular weights (MWs) of the protein complexes were calculated on the basis of the GFC protein standards (Bio-Rad, USA) as the control (Figure S1). The theoretical and apparent MWs of the proteins and complexes in this study are listed in Table S1.

Western blotting

The protein components separated by SDS-PAGE were transferred to the polyvinylidene fluoride membrane (PVDF) for western blotting. The membrane was washed by 1×TBST buffer (25 mM Tris-HCl, 140 mM NaCl, 3 mM KCl, 0.1% Tween 20, pH7.4) for three times, and then incubated with 1×TBST containing nonfat milk at room temperature for 2h. After washing, HRP-conjugated His-tag monoclonal antibody (proteinTech Group, USA) followed by enhanced chemiluminescence (ECL) was used for detection of His-tagged protein.

In vitro RNA production and purification

Because only the Sm site nonanucleotide sequence and its 3′-stem-loop secondary structure (instead of the sequence) of snRNAs are important for assembly into Sm cores¹⁸ and these components are highly conserved in eukaryotes, human U4-snRNA and their derivatives were used in this study (Their sequences are in Table S2). All RNAs were produced by in vitro transcription using T7 high yield RNA Transcription kit (Vazyme, China). The DNA templates were made by PCR. Transcribed RNAs were purified by GFC (Superdex 200 Increase 10/300 GL) in the running buffer containing 20 mM Tris-HCl, 250 mM NaCl, 2 mM MgCl₂ and 1 mM EDTA, pH 8.0. The qualities of these RNAs were further checked on agarose gel electrophoresis.

In vitro RNA–protein complex assembly assay

RNA-protein complex assembly assays were performed by incubating purified protein complexes with various RNAs in final volume of 400 μL in assembly buffer containing 20 mM Tris-HCl, 250 mM NaCl, 2 mM MgCl₂, 1 mM EDTA, pH 8.0, with their amounts described in detail in Table S3. RNAs were pre-incubated at 65°C for 10 min followed by cool-down in room temperature before mixing with proteins. After incubation at 37°C for 40 min, the samples were collected at 15,000 rpm for 20 min in a table centrifuge and applied into superdex 200 Increase 10/300 GL GFC via a 500 μL sample loop. The eluted fractions were collected at each 0.5 mL, resolved by SDS-PAGE and silver staining.

In vivo double gene knockout of Gemin2 in S. pombe

The haploid strain 503 (h+ ade6-704 leu1-32 ura4-D18 his3-D1) of Schizosaccharomyces pombe (gift from Dr. Cong Liu) was used for knockout study of the Gemin2-expressing genes yip11 (SPAC19B12.12c) and yip12 (SPAPB17E12.02) by Cas9-assisted homologous recombination (Figure 4A). The S. pombe cells preserved in glycerol resuscitated and cultured in YES or EMM2 medium at 30°C.

For possible no growth phenotype of double deletion of yip11 and yip12 genes, a plasmid expressing spGemin2 under the control of nmt1 promoter, pFA6a-URA3-SpGe2 (see next section), was transformed into the S. pombe cells. In addition, to make more efficient recombination, a second plasmid, pCAS-sgRNA (see next section), which contains Cas9 expressing gene and sgRNA targeting the same site of both yip11 and yip12 genes was constructed. The donor DNA fragments (Table S4) were made by PCR and contained the nourseothricin acetyltransferase 1(Nat1) gene flanked by 183 and 255 nt long sequences homologous to the 5′ and 3′ sides of yip11 and yip12 respectively. Both pCAS-sgRNA and donor DNA fragments were transformed into the above S. pombe stain containing pFA6a-URA3-SpGe2. The transformed cells were placed on the EMM2 medium containing 100 μg/mL nourseothricin (NTC), 50 μg/mL G418, 0.25 g/L each of histidine, adenine, lysine and leucine, but no uracil (EMM2-ΔURA), for selection. The primer pairs P1-F/P1-R and P2-F/P2-R (Table S5) were used in PCR reaction to verify the genotype (Figure 4A). The PCR products were also sequenced for further verification. In order to evaluate whether Gemin2 is essential for S. pombe cell survival, the verified transformant was cultured in EMM2-ΔURA medium and spotted in serial 1:10 dilutions on the EMM2+FOA selection plate (EMM2 medium containing 0.5 mg/mL each of 5-fluorolactic acid (5-FOA), 0.25 g/L each of histidine, adenine, lysine and leucine, and 0.0625 g/L of uracil) for essentiality test and on the EMM2-ΔURA plate as a control. The plasmid pFA6a-URA3-SpGe2 in S. pombe cells would be excluded by the presence of FOA.

Construction of the plasmids for gene deletion in S. pombe

For construction of the plasmid pFA6a-URA3-Ge2, we first used PCR to produce DNA fragments containing the major parts, including the yeast autonomous replication sequence ARS1 (primer pair ARS1-F/R), URA3 gene (primer pair URA3-F/R) and SpGemin2 gene (primer pair in-SpGe2-F/R). The corresponding primer sequences are listed in Table S5. The fragments were linked to the pFA6a vector fragment by a homologous recombination method, seamless cloning, via Monclone Single Assembly Cloning Mix (Monad Biotech, China). For construction of the plasmid pCAS-sgRNA, we used PCR to generate the pCAS vector fragment (Addgene #60847) and the sgRNA insert fragment. The two fragments were linked by seamless cloning method too. We made three pCAS-sgRNA plasmids each containing a different sgRNA sequence to maximize the successful chance of Cas9 cleavage of the target gene in genome. The three pairs of primers for sgRNA fragment are gRNA1-F/R, gRNA2-F/R and gRNA3-F/R (Table S5). All the plasmids were verified by DNA sequencing.

Structure models of the proteins and complexes

The model of each single protein (Figure S2) was directly downloaded from AlphaFold protein structure database (https://alphafold.ebi.ac.uk/).⁵⁸ For construction of the structure model of the S. pombe SMNp/Gemin2/SmD1/D2/F/E/G complex (Figure S3), we used the structures of SmD1, SmD2, SmF, SmE and SmG from an S. pombe spliceosome structure (PDB code: 3JB9), the SMN (residues 11-35) structure from PDB code 7BB3, and Gemin2 (residues 46-235) from SWISS-MODEL (https://swissmodel.expasy.org/). We then aligned all these proteins and domains against the corresponding ones in the human 7S structure (PDB code: 5XJL). Finally we made an extensive optimization of the complex using the Rosetta program.⁵⁹

Published: August 12, 2023

Data and code availability

• •The structural models of the six assembly chaperone proteins in S. pombe can be found in AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/) and the structural model of S. pombe Gemin2/SMNp/SmD1/D2/F/E/G (7S) complex is in supplementary data. All other data reported in this paper will be shared by the lead contact upon request.
• •This study did not generate original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.