Conserved structure of Snu13 from the highly reduced spliceosome of Cyanidioschyzon merolae

Abstract

Structural and functional analysis of proteins involved in pre‐mRNA splicing is challenging because of the complexity of the splicing machinery, known as the spliceosome. Bioinformatic, proteomic, and biochemical analyses have identified a minimal spliceosome in the red alga Cyanidioschyzon merolae. This spliceosome consists of only 40 core proteins, compared to ∼70 in S. cerevisiae (yeast) and ∼150 in humans. We report the X‐ray crystallographic analysis of C. merolae Snu13 (CmSnu13), a key component of the assembling spliceosome, and present evidence for conservation of Snu13 function in this algal splicing pathway. The near identity of CmSnu13's three‐dimensional structure to yeast and human Snu13 suggests that C. merolae should be an excellent model system for investigating the structure and function of the conserved core of the spliceosome.

Short abstract

Interactive Figure 1; Interactive Figure 2; Interactive Figure 3 | PDB Code(s): 5EWR

Introduction

Premessenger RNA (pre‐mRNA) splicing involves two transesterifications catalyzed by the spliceosome, a multi‐megaDalton‐sized particle made up of 5 small, nuclear RNAs (snRNAs) and up to 250 proteins, many of which are specifically associated with the snRNAs in small, nuclear ribonucleoprotein (snRNP) particles.1 Cyanidioschyzon merolae is a unicellular red alga that lives in geothermal springs at temperatures of 42–56°C and pH of 1.5–3.2 The genome contains only 27 introns; in contrast, the Saccharomyces cerevisiae (yeast) genome, of similar size, has more than 250.3, 4 We recently reported a computational and biochemical assessment of the C. merolae splicing factor complement that revealed a substantial reduction in core splicing proteins to only 40.5 Notably, there was no evidence for the U1 snRNA or any of its associated proteins in this organism. C. merolae therefore provides a minimal model for studying pre‐mRNA splicing, and a source of thermostable splicing proteins for crystallographic studies.6

A 13–17 kDa, U4‐associated RNA binding protein was identified as part of the U4/U6.U5 tri‐snRNP particle in humans (15.5K) and yeast (Snu13p) with a homologous protein (ribosomal protein L7) identified in archaea.7, 8, 9 We have identified the homolog of yeast Snu13p in C. merolae 5 (hereafter CmSnu13). Snu13p/15.5K has also been implicated in 2′ O‐methylation of preribosomal RNA because of its association with box C/D sno‐RNPs.10, 11, 12 It has been shown that CmSnu13 homologs recognize the kink‐turn motif, a purine‐rich asymmetrical (5 + 2) internal loop with flanking stems that bend to 65° of each other.13 Upon binding, the protein acts as a nucleation factor for RNP assembly.14, 15

Crystal structures of human, yeast, and archaeal homologs of CmSnu13 alone and bound to RNA have been reported.16, 17, 18, 19 Snu13 is an RNA‐binding protein featuring an alpha/beta sandwich fold. Interactions with both the U4 5′ stem loop and hPrp31 have been well characterized.11, 13, 14, 15, 16, 17, 18 It has also been shown that Prp3p (human 90K) and Prp4p (human 60K) associate and require Snu13p to bind to U4 snRNA prior to their own assembly into the U4 snRNP.14, 20 It is not known what proteins are associated with U4 snRNA prior to assembly of the U4/U6 di‐snRNP, but they may include the core Sm and Snu13 proteins.14, 21 In the second stage of U4/U6 snRNP assembly, Prp31 and Prp3/4 join independently.21 Notably, Prp4 and Prp31 are both missing in C. merolae. To study the role of Snu13 in C. merolae, we determined the crystal structure of free CmSnu13 and compared it to both free (yeast and human) and RNA‐bound (human) structures. The C. merolae structure is very similar to both free Snu13 and 15.5K in complex with both U4 snRNA and hPrp31; these results argue for a conserved role of the algal protein in the C. merolae splicing pathway. Importantly, the conservation of Snu13 structure also demonstrates that, despite its extreme reduction in complexity, C. merolae's spliceosome should provide a good model for the most highly conserved elements of spliceosome structure and function.

Results and Discussion

To evaluate the similarity of CmSnu13 to its homologs, we performed an amino acid alignment of CmSnu13 with diverse orthologs revealing high sequence conservation [Fig. 1(A)], suggesting functional conservation. The 57% identity between C. merolae and yeast orthologs is one of the highest levels of sequence conservation of any C. merolae splicing protein.5

Figure 1.

X‐ray structure of C. merolae Snu13. (a) Sequence alignment30 with sequences from G. sulphuraria (GsSnu13), human (15.5K), and S. cerevisiae (Snu13p), indicating sequence identity (*), conservation (:) and partial conservation (.). Secondary‐structure diagram depicting α‐helices (red) and β‐strands (yellow) derived from the structure is shown above the alignment. (b) Ribbon diagram of CmSnu13 structure depicting α‐helices (red) and β‐strands (yellow). An interactive view is available in the electronic version of the article.

Structures of Snu13 homologs have been reported alone and bound to fragments of U4 snRNA.13, 16 To determine whether similarity in primary sequence was reflected at the structural level, we solved the 2.35 Å resolution X‐ray structure of CmSnu13 [Fig. 1(B), Table 1]. To our knowledge, this is the first structure of a splicing protein from C. merolae.

Table 1.

Data Collection, Phasing, and Refinement Statistics

	CmSnu13a
Data collection
Space group	P212121
Cell dimensions
a, b, c (Å)	30.33, 57.58, 65.38
α, β, γ (°)	90, 90, 90
Wavelength (Å)	1.1271
Resolution (Å)	2.35
I/σI	21.36 (11.85)
Completeness (%)	99.77 (99.39)
Redundancy	7.7 (7.6)
Refinement
Resolution (Å)	43.21–2.35
No. reflections	158,498
R work/R free	0.166/0.220
No. of atoms
Protein	933
Water	42
B‐factors (Å2)
Protein	49.5
Water	40.0
R.M.S deviations
Bond lengths (Å)	0.016
Bond angles (°)	1.52

a Statistics for highest resolution shell (2.43–2.35 Å) are shown in parentheses.

The CmSnu13 structure is highly conserved with respect to the yeast and human homologs, with an overall root‐mean‐square deviation (rmsd) of alpha carbons of 0.014 Å among the structures shown in Figure 2(A). The role of two parts of Snu13 in splicing has been determined: a U4 snRNA‐binding region, and a hydrophobic pocket that is believed to interact with Prp3 [Fig. 2(B)]. These features are conserved in CmSnu13, and modeling of RNA from the human RNA‐protein complex into the C. merolae structure suggests that the RNA interaction could be accommodated [Fig. 2(C)]. These observations strongly support a conserved role for CmSnu13 in U4/U6 di‐snRNP structure.

Figure 2.

Structural comparison of C. merolae Snu13 with orthologs. (a) Structural overlay of ribbon diagrams of CmSnu13 (C. merolae; PDB: 5EWR; green), Snu13p (S. cerevisiae; PDB: 2ALE; cyan), 15.5K (human; PDB: 2JNB; yellow), 15.5K (human bound to U4atac RNA and Prp31; PDB: 3SIU; magenta), 15.5K (human bound to Prp31 and U4 snRNA; PDB: 2OZB; light pink). (b) Conservation of the hydrophobic pocket proposed to interact with Prp3. Shown are ribbon diagrams of CmSnu13 (C. merolae; PDB: 5EWR; green) overlaid on Snu13p (S. cerevisiae; PDB: 2ALE; cyan) highlighting the residues lining the hydrophonic pocket: F9 (Y28), P68 (P87), L69 (L88), Y78 (Y97), F80 (F99). Yeast numbering followed by algal in parentheses. (c) Detail of overlay of CmSnu13 (C. merolae; PDB: 5EWR; ribbons, green) with 15.5K bound to U4atac RNA (human; PDB: 3SIU; ribbons, magenta; RNA in stick representation). An interactive view is available in the electronic version of the article.

The Snu13 hydrophobic pocket, located within α‐helix 3 and β‐strand 3, is conserved between C. merolae and S. cerevisiae. Dobbyn et al. proposed this hydrophobic region to be a site of protein–protein interaction with either Prp3p or Prp4p.16 Our catalog of splicing factors in C. merolae revealed no Prp4p/60K homolog.5 Thus, we propose that Prp3 (90K in humans) is the most likely candidate for a protein interacting with the hydrophobic pocket of Snu13. Structurally, this interaction would mean that Prp3 bridges Stem II of the U4/U6 complex and the 5′ stem loop of U4 via Snu13. This is consistent with a recent crystal structure of Prp3 in association with portions of U4/U6.20

We compared the CmSnu13‐RNA interaction5 with reported observations for Snu13 orthologs. The dissociation constant, K _D, for for CmSnu13 binding U4 snRNA was measured using fluorescence polarization with a 5'‐fluorescein labeled RNA corresponding to U4 stem II,5 nucleotides 22–50 (Fig. 3). An increase in concentration of CmSnu13 caused an increase in fluorescence polarization, consistent with RNA‐protein binding, so we fit a Hill curve to the binding data to determine the K _D. In testing a variety of buffer conditions, neither pH, nor NaCl/MgCl₂ concentration had an effect on binding; MgCl₂ increased fluorescence anisotropy, perhaps due to the stabilization of U4's kink‐turn.23 Under optimal conditions, we measured a K _D of ∼160 nM (Fig. 3). In a control, we measured a K _D of ∼16 µM for CmSnu13‐U6 snRNA association, demonstrating that the interaction with U4 RNA is specific (Fig. 3). These observations are consistent with the conservation of CmSnu13's role in splicing via its association with U4 snRNA.

Figure 3.

Analysis of CmSnu13•RNA binding. Fluorescence polarization measured by incubation of 5′‐fluorescein‐labeled RNAs, fragments of C. merolae U4 (5′ stem loop, inset) and S. cerevisiae U6 (5′ stem loop) snRNAs, with increasing concentration of CmSnu13. Dissociation constants, K _D, were determined by fitting these data to the Hill equation. An interactive view is available in the electronic version of the article. Interactive Image 4: An interactive view is available in the electronic version of the article.

We compared our findings with those reported in the literature for homologous proteins and their cognate RNA partners (Table 2). Snu13 homologs were found to have a K _D ranging from 8 to 150 nM (Table 2). We note that these measurements were made with a variety of techniques (EMSA, FP, surface plasmon resonance, etc.) so the values may not be directly comparable.

Table 2.

Comparison of Protein‐RNA Affinities

Proteina	RNA target	Methodb	Apparent K D (nM)
15.5K24	U3 snoRNA	EMSA	30
15.5K24	U8 snoRNA	EMSA	40
15.5K24	U14 snoRNA	EMSA	8
15.5K24	U4 snRNA	EMSA	20
15.5K25	U3 snoRNA	EMSA	130 ± 13
Snu13p16	U3 snoRNA	EMSA	75
Snu13p16	U4 snRNA	EMSA	150
15.5K26	U4 5′ stem loop	SPR	27
CmSnu13b	U4 5′ stem loop	FP	160 ± 10

EMSA: electrophoretic mobility shift assay; SPR: surface plasmon resonance; FP: fluorescence polarization.

^aReferences indicated as superscript.

^bThis study.

CmSnu13 has a similar affinity for the U4 5′ stem‐loop as its homologs from humans and yeast, with a measured K _D of ∼160 nM (Table 2). This is consistent with the finding that Snu13 requires a purine‐rich (5 + 2) asymmetrical loop for optimal binding.8, 13, 14 In particular, positions 31–33 and 42–43 of the loop are conserved and positions 29 and 30 must be purines.8 The Cm U4 stem loop is identical to the human version except at position 29 (G29→A).8 The extensive conservation between C. merolae and humans, seen in both the protein and RNA structures, would suggest that U31 is sequestered in the RNA binding pocket when CmSnu13 and U4 snRNA interact.13 Furthermore, the K _D observed with respect to CmSnu13 and U4 stem loop interaction is consistent with Snu13's role as a nucleation factor for binding of other U4/U6 di‐snRNP components, such as Prp3.16

Together, the structural and functional results reported here strongly suggest a conserved role for CmSnu13 in the U4 snRNP within the C. merolae splicing pathway, and demonstrate that C. merolae's spliceosome structure is likely to be highly conserved.

Methods

Cloning, expression, and protein purification

We amplified the CmSnu13 gene from genomic DNA and subcloned it into SspI‐digested pMCGS7 for expression with an N‐terminal His tag and a TEV protease site.27 Protein was expressed in E. coli and purified by two rounds of Ni²⁺ affinity chromatography before and after cleavage with TEV protease; CmSnu13 was further purified on a Mono S column (GE Healthcare) and concentrated for crystallization/binding studies.

Fluorescence polarization

We performed binding studies with oligonucleotides (IDT) derived from the C. merolae U4 snRNA and the S. cerevisiae U6 snRNA:

Cm U4 RNA oligo: 5′ – Fluorescein – UUGCCCAGAUGAGGUUCUCCGAUGGGUAA – 3′.

Sc U6 RNA oligo: 5′ – UUCCCCUGCAUAAGGAU – Fluorescein – 3′.

We incubated CmSnu13 (0–100 µM) with 5 nM U4 or U6 RNA oligos in 100 µL binding reactions with buffer conditions varied to observe the sensitivity of the protein‐RNA interaction to pH and salt. We measured the anisotropy using a Synergy 2 Multi‐Mode reader (BioTek) with black 384‐well microplates (Nunc Thermo Scientific).

Structure determination

We concentrated the protein to 10 mg mL⁻¹ and grew crystals in 31% PEG 3350 with 100 mM sodium acetate pH 4.4. We transferred the crystals to precipitant containing 20% glycerol and froze them in liquid nitrogen. We collected diffraction data remotely from the Stanford Synchrotron Radiation Light Source, and used HKL‐200028 and CCP429 software suites for data processing. We used PHENIX22 for structure refinement, with yeast Snu13p (PDB 2ALE) as the search model for molecular replacement. We have deposited the atomic coordinates and structure factors for CmSnu13 in the Protein Data Bank with the accession code 5EWR.