Video 1: The human C* spliceosome
FromCryo-EM structure of a human spliceosome activated for step 2 of splicing
- Karl Bertram1, *
- Dmitry E. Agafonov2, *
- Wen-Ti Liu1,
- Olexandr Dybkov2,
- Cindy L. Will2,
- Klaus Hartmuth2,
- Henning Urlaub3, 4,
- Berthold Kastner2,
- Holger Stark1,
- Reinhard Lührmann2,
- Journal name:
- Nature
- Volume:
- 542,
- Pages:
- 318–323
- Date published:
- DOI:
- doi:10.1038/nature21079
Streaming video does not reproduce the original video quality. Download the video using the link below to view at original quality.
This video shows a 3D EM density map of the human C* spliceosome and fit of spliceosomal RNAs and proteins.More videos from this article
Additional data
Author footnotes
These authors contributed equally to this work.
- Karl Bertram &
- Dmitry E. Agafonov
Affiliations
-
Department of Structural Dynamics, MPI for Biophysical Chemistry, Am Fassberg 11, D-37077 Göttingen, Germany
- Karl Bertram,
- Wen-Ti Liu &
- Holger Stark
-
Department of Cellular Biochemistry, MPI for Biophysical Chemistry, Am Fassberg 11, D-37077 Göttingen, Germany
- Dmitry E. Agafonov,
- Olexandr Dybkov,
- Cindy L. Will,
- Klaus Hartmuth,
- Berthold Kastner &
- Reinhard Lührmann
-
Bioanalytical Mass Spectrometry, MPI for Biophysical Chemistry, Am Fassberg 11, D-37077 Göttingen, Germany
- Henning Urlaub
-
Bioanalytics Group, Institute for Clinical Chemistry, University Medical Centre Göttingen, Robert-Koch-Straße 40, D-37075 Göttingen, Germany
- Henning Urlaub
Contributions
D.E.A. developed the purification strategy and characterized the C* complex. O.D. and H.U. analysed protein–protein crosslinking data. K.B., W.-T.L. and H.S. prepared grids and collected EM data. B.K., K.B. and K.H. carried out model building and refinement. K.B., B.K. K.H., H.S. and R.L. analysed the structure. All authors were involved in data interpretation. The manuscript was written by R.L. and C.L.W., with input from all authors. R.L., H.S., B.K. and H.U. initiated and orchestrated the project.
Competing financial interests
The authors declare no competing financial interests.
Reviewer Information Nature thanks Y. Cheng, N. Toor and J. Valcarcel for their contribution to the peer review of this work.
Author details
Karl Bertram
Search for this author in:
Dmitry E. Agafonov
Search for this author in:
Wen-Ti Liu
Search for this author in:
Olexandr Dybkov
Search for this author in:
Cindy L. Will
Search for this author in:
Klaus Hartmuth
Search for this author in:
Henning Urlaub
Search for this author in:
Berthold Kastner
Search for this author in:
Holger Stark
Search for this author in:
Reinhard Lührmann
Search for this author in:
-
Extended Data Figure 1: Compositional and functional characterization of affinity-purified, human C* complexes.Hover over figure to zoom
a, Spliceosomal assembly pathway, starting with complex B. b, c, RNA (b) and protein (c) composition of MS2 affinity-purified, human C* complexes. Only high molecular weight proteins are labelled. C* complexes contained predominantly the first-step splicing intermediates (cleaved 5′ exon and the lariat–3′-exon), and U2, U5 and U6 snRNA. d, e, Abundant C* proteins (red) identified by 2D gel electrophoresis, summarized in e. Less abundant proteins, green. Upper and lower panel: proteins >25 kDa or <25 kDa, respectively. Proteins present in substoichiometric amounts in previous human C complex preparations (for example, SF3a and SF3b proteins)49, 51 were lacking or present only in very low amounts. SRm300 does not migrate well into the second dimension of the 2D gel, but is clearly observed on a 1D gel. The step-two factors SLU7 and PRP18 were not detected, which may explain why spliceosomes are stalled after step one at pH 6.4. All abundant proteins or domains/regions thereof were modelled into the C* EM density map, except BRR2, ISY1, NY-CO-10, CCD12, PPIL3, CBP80 and CBP20 (Supplementary Table 1). As BRR2 could not be located in the C* EM density map, it is probably highly flexible in C*. f, C* complexes formed on 32P-labelled MINX pre-mRNA at pH 6.4 were purified and incubated in HeLa nuclear extract at pH 7.9 or in buffer under splicing conditions. Splicing intermediates were efficiently chased into splicing products after 15 min in the presence of nuclear extract, but not buffer alone, indicating that our purified C* complexes are functional and not dead-end complexes. g, h, MINX pre-mRNA (g) or affinity-purified C* complexes (h) were incubated with HeLa nuclear extract with or without anti-PRP16 antibodies. Efficient catalysis of step two of splicing was observed if purified C* complexes, but not pre-mRNA alone, were incubated in nuclear extract pre-incubated with PRP16 antibodies, indicating that the C* complexes are indeed stalled after the action of PRP16. All 2D analyses and in vitro splicing experiments were performed at least twice in two independent experiments.
-
Extended Data Figure 2: Cryo-EM and image processing of the human C* complex.Hover over figure to zoom
a, Typical cryo-EM raw image of human C* spliceosomes recorded with a Titan Krios (FEI Company) electron microscope at a nominal magnification of 88,000× with a Falcon II direct electron detector resulting in a pixel size of 1.59 Å per pixel. b, Several representative class averages showing different views of the C* particle after 2D classification. c, Euler angle distribution of all particle images that contributed to the final 3D map. The coordinates describe the ϕ and θ angles. Size and colour of the plotted dots indicate the number of particles at any given Euler angle. Although several sets of particle orientations dominated, an almost complete angular coverage was obtained. d, Computational sorting scheme. Imaged micrographs were first evaluated and sorted according to the quality of their Thon rings in local power spectra. Roughly 2.5 million particle images were then selected from the remaining micrographs. In a second sorting step, particle images were again discarded based on the quality of Thon rings in classified, local power spectra. Following evaluations in Fourier space, particles were then excluded according to multiple rounds of 2D classifications. The remaining 1,708,164 particle images were split into seven subsets of approximately 244,000 particles, and each subset again separated into six classes by 3D classification in RELION. One of the results is shown as an example. Images contributing to the best defined spliceosome structure (around 23%) were then merged again into one particle subset. Further rounds of 3D classification in RELION led to a refinement of particle homogeneity within the final subset. The latter, now consisting of 136,534 particles was refined to a structure at 8.4 Å resolution without masking. The final structure at 5.9 Å resolution was obtained by applying a soft mask during the final steps of the refinement process. To evaluate details at the core of the C* EM density, the unfiltered map obtained after the final 3D refinement was low-pass filtered to 4.5 Å and sharpened using a B-factor of −350 in RELION. e, Local resolution plot reveals a resolution distribution from approximately 4.5 to 10 Å with some less well-defined parts at the periphery of the complex. Higher resolution regions (in blue, up to 4.5 Å resolution) were obtained for the centrally located catalytic core of the spliceosome. f, Fourier-shell correlation function of two independently refined half data sets indicates a global resolution of 5.9 Å for the masked C* spliceosome comprising around 80% of the visible density of the whole spliceosome with respect to the unmasked density volume.
-
Extended Data Figure 3: Protein–RNA interactions in the catalytic RNP core of the C* complex.Hover over figure to zoom
a, Domain structure of the human PRP8 protein. b, c, The catalytic U2–U6 RNA core is shown docked into the active-site pocket of PRP8 (space filling model) in the human C* and S. cerevisiae C complex22. BS, branch site. d, Interaction of the N-terminal HAT repeats of SYF3 and the linker between PRP8 NTD1 and 2 with the U6 ISL in the C* complex. e, Interaction of the CDC5 Myb domain with U2/U6 helix Ia. f, Interaction of WD40 domain of PRL1 with U5 stem Ic and the loop of the U6 ISL. Not only is the docking of the U2–U6 RNA core within the active site pocket of PRP8 similar in C and C*, but also the interaction of the linker between PRP8 NTD1 and 2 with the major groove of the U6 ISL, of SYF3 with the lower stem of the U6 ISL, and of the CDC5 Myb domain with the U2/U6 helix Ia. Moreover, the WD40 domain of PRL1 interacts with U5 stem Ic and the U6 ISL loop in both organisms14, 21, 22 and α-helix 2 of CDC5 runs along the groove of the U6 ACAGA box helix (Extended Data Fig. 5b). N-terminal parts of Skip and Ad002 (amino acid positions are indicated) also contact the U6 ISL.
-
Extended Data Figure 4: Interactions of spliceosomal proteins with the U6 snRNA in the C* complex.Hover over figure to zoom
a, Overview of the location of various proteins and functionally important regions of the U6 and U2 snRNAs in the C* complex. b, c, Close up of the fit of G10/BUD31, the PRP8 NTD1, and RBM22, as well as the U6 5′ stem loop, into the EM density map of C*. d, Comparison of the domain structure of human RBM22 with that of its apparent homologues in yeast, Cwc2 and Ecm2. e, Comparison of the organization of homologous domains between RBM22, and Ecm2 and Cwc2, in the human C* and S. cerevisiae C complexes. The middle and right diagrams show different slices of the C complex EM density. Some regions of human RBM22 that are homologous to regions of the S. cerevisiae Cwc2 and Ecm2 proteins, are located near the U6 ISL in the human C* complex, similar to the situation in the S. cerevisiae C complex. For example, the N-terminal part of RBM22, comprising two zinc-finger domains that are homologous to those present in S. cerevisiae Ecm2, could be fit into a density element close to G10/BUD31, at a position similar to that in the S. cerevisiae C complex (and S. pombe ILS, not shown). The third zinc-finger domain of RBM22 (homologous to Cwc2) is located close to the 5′ stem loop of U6 snRNA. The RRM domain of RBM22 (homologous to the RRM of Cwc2) fits into a less well-defined density element close to PPIL1, aquarius, and Skip, consistent with protein crosslinking data (Supplementary Table 2). Thus, the RRM of RBM22 is located at a different position compared to that of its likely yeast homologue (Cwc2/Cwf2) in the yeast spliceosomes.
-
Extended Data Figure 5: Molecular organization of the NTC and NTC-related proteins is similar in the C* complex and S. pombe ILS.Hover over figure to zoom
a, Overview of the location of various proteins in the C* complex. The N-terminal HAT repeats of SYF1 share a large interface with the aquarius (Aqr) protein. At the interface between SYF1 and aquarius there is a density element that accommodates cyclophilin E (CypE), a spliceosomal protein not found in yeast (Fig. 1). In C*, the density accommodating aquarius is less well-defined, presumably owing to flexibility of this part of the complex. Crosslinks between ISY1 and aquarius indicate that part of ISY1 is located at the top of C* (Supplementary Table 2). The meandering path of Skip (Prp45) and Ad002 (Cwc15) around the PRL1 WD40 domain is shown. N-terminal parts of Skip and Ad002 also contact the U6 ISL (see also Extended Data Fig. 3). The position of the PRP19 helical bundle (containing SPF27, the C-terminal part of CDC5 and four copies of PRP19) is stabilized by interactions of the C-terminal end of CDC5 with the WD40 domain of the U5-40K protein, and by interactions with the cyclophilin PPIL1. The WD40 domains of PRP19 are not visualized, presumably owing to their flexibility. N or C, N or C terminus. b, Magnified view showing the fit of parts of SYF1, SYF2 and SYF3 and their neighbouring proteins into the C* EM density map. In C*, the C-terminal HAT repeats of SYF1 contact the N-terminal Myb domain of CDC5 (which in turn binds to the PRP8 RT domain). CDC5 α-helix 2 runs along the U6 ACAGA box/intron helix. c, Fit of Skip, PRL1, Ad002, SYF2 and their neighbouring proteins into the C* EM density map. Major parts of Skip (amino acid numbers shown in red) bridge the WD40 domain of PRL1 (Prp46 in S. cerevisiae) and SYF3 N-terminal HAT repeats 2–5 with the PRP8 NTD2. SYF2 (amino acid numbers in blue) contacts U2/U6 helix II. The N-terminal part of Ad002 (amino acid numbers in pink) contacts the PRL1 WD40 domain (see also panel f). d, e, Comparison of the structure/organization of SYF1 and SYF3 (shown as space filling models), aquarius and other NTC and NTC-related proteins in the human C* complex and the S. pombe ILS. The N-terminal region of CDC5, the N-terminal α-helices of SYF3, WD40 domain of PRL1, major parts of Skip and Ad002, and PPIL1 (Ppi1 in S. pombe) have nearly the same structure and position as their yeast homologues (see also panels a–c). f, N-terminal amino acids of Skip interact with PPIL1, consistent with previous biochemical and structural studies67, 68, 69.
-
Extended Data Figure 6: 3D RNA network of the human C* complex.Hover over figure to zoom
a, b, Fit of the catalytic core RNA into the 5.9 Å EM density map of C* (a) or into the EM density map low-pass filtered to 4.5 Å (b). The topography of the RNA density is consistent with formation of the catalytic triplex, as found in the C complex and ILS. That is, the Watson–Crick faces of U6 nucleotides G46 and A47 are oriented towards the Hoogsteen faces of G54 and A53, respectively, and can potentially form two base triples. The bulged U74 of the U6 ISL would be in a position to stack on G46 and may form the third triple with U6 C55 (Fig. 2). Black circle, catalytic centre. c, Overall similarity of the catalytic RNA network in the yeast C versus human C* complex. Superimposition of the core RNA elements from human C* with those from the S. cerevisiae C complex. Dark-coloured RNAs are from C*, whereas the lighter-coloured RNAs are from the C complex (PDB 5LJ3) and (PDB 5GMK). In the C* complex, the U6 ACAGA box/intron helix is tilted slightly (compared to the C complex), such that the U6 snRNA moves 5–6 Å towards the core of the complex, which is probably due to the significantly different positions of the branched intron structures in the different spliceosomal complexes. d, The conformation of U6 nucleotide A48 (U54 in yeast U6) is significantly different in the yeast C and human C* complex. Whether U6 A48 adopts this distinct conformation in other human spliceosomal complexes or whether it represents an intermediate state following PRP16 remodelling that subsequently adopts a conformation similar to U6 U54 in the yeast C complex before step two catalysis is unclear. e, U6 A48 is bound in a protein pocket comprised of amino acids from Skip and CDC5. Fit of U6 and U2 snRNA nucleotides, and amino acids of CDC5 and Skip into the C* EM density map. f, Interactions of the U5 loop 1 and PRP8 with the 5′ exon, and location of the 3′ end of the 5′ exon close to the catalytic centre (black circle).
-
Extended Data Figure 7: Location and structure of CWC22 and its interaction partner, the EJC helicase eIF4AIII, in the C* complex.Hover over figure to zoom
a, EM density fit of various proteins forming the channel in which the 5′ exon is bound. The MA3 domain of CWC22 binds the PRP8 RT/En domain and its N-terminal MIF4G domain is attached to domain 1 of SNU114. The MIF4G domain is only clearly visible in the unmasked EM density of the C* complex, and maps to UPD2 (panel a and b). The PRP8 switch loop is shown in light blue. Close to the PRP8 switch loop, there is a density element into which the approximately 20 N-terminal amino acids of SRm300 can be accommodated in a manner similar to the arrangement of the homologous sequence of Cwc21 in the C complex of S. cerevisiae (indicated in green colour). b, EM density fit of the CWC22 MIF4G domain and the RecA domains of the eIF4AIII RNA helicase in the UPD2 density of the unmasked C* model. Consistent with crosslinks (Supplementary Table 2), the RecA2 domain contacts the MIF4G domain. In the C* complex, the relative orientation of the two RecA domains is such that they adopt a partially closed conformation, in contrast to their open (inactive) conformation in the MIF4G–eIF4AIII co-crystal structure (middle panel; PDB 4C9B) and their closed conformation in the crystal structure of the EJC complex, in complex with RNA (right most panel; PDB 2XB2). Shown is the poor fit of the RecA1 domain when in the open or completely closed conformation. Protein crosslinks between eIF4AIII and an N-terminal region of SRm300 indicate that part of this region is located in the neighbouring UPD1 density element of the unmasked C* complex map, adjacent to domain 4 of SNU114 (which is not a part of UPD1), consistent with the location of two α-helical elements of Cwc21 in an equivalent position in the S. cerevisiae C complex22. We cannot locate the RS domain of SRm300, probably because it is very flexible and located at the periphery of the C* complex. c, Model of the position of an extended 5′ exon RNA (red) relative to the RecA domains of eIF4AIII in the C* complex. Extension of the 5′ exon RNA to position −26 (relative to the 5′ splice site) was performed by inserting nucleotides −26 to −18 based on the crystal structure of the isolated EJC bound to RNA and additionally modelling in exon nucleotides −17 to −13 (in an extended conformation), which we cannot clearly localize based on the EM density map. This suggests that eIF4AIII is already located at an optimal distance (that is, around 20 nucleotides upstream) from the 5′ splice site (and thus also from the exon–exon junction that is formed after step two of splicing). Tight binding of eIF4AIII to the 5′ exon probably occurs at a stage after formation of the C* complex, based on the observation that the RecA domains are not completely closed at this stage.
-
Extended Data Figure 8: Variable position of the PRP8 RH domain and α-finger in the Bact, C and C* complexes.Hover over figure to zoom
a, Comparison of the position and structure of the PRP8 RH domain (magenta ribbon diagram) and α-finger15 (purple) relative to the PRP8 RT/En domain (pink space filling model) that is oriented in the same manner in Bact, C, C* and the ILS. The position of the U2/branch site helix alone (Bact and ILS) or with the extended U2/branch site helix (as found in C and C*) is also shown. In the yeast C complex, the RH domain interacts with the 3′ part of the U2 snRNP, Cwc25 and the extended region of the U2/branch site helix21, 22, whereas in C* the RH domain interacts with extended α helices close to the tip of the En domain of PRP8 and its β-hairpin loop is inserted between the groove of the U2/branch site helix and the U6-ACAGA box/intron helix. Repositioning of the RH domain from C to C* requires not only a translational, but also rotational movement that is coordinated with repositioning of the U2/branch site helix. Asterisk, position of the β-hairpin loop of the RH domain. b, Position of the PRP8 α-finger (shown in purple in the space filling model) in the human C* complex. In the latter, the α-finger is found close to the proposed position of the catalytic Mg2+ ions (turquoise spheres). The PRP8 α-finger could potentially aid in the placement of the 3′ splice site for step two catalysis (indicated by a dashed blue line). Alternatively, it is conceivable that the α-finger must be repositioned to allow proper positioning of the 3′ splice site, which could be achieved by the interaction of a protein (for example, SLU7 or PRP18) or docking of the 3′ splice site into the catalytic centre.
-
Extended Data Figure 9: Interaction of the WD40 domain of PRP17 with the extended U2/branch site helix and U2 nucleotides linking the U2/branch site helix and U2 Sm core.Hover over figure to zoom
a, b, Interaction of the PRP17 WD40 domain with the extended U2/branch site helix, CDC5 helix 2 and CypE. The fit of the aforementioned protein domains and RNA, plus additional neighbouring proteins, into the C* EM density map is shown. Consistent with our protein crosslinks (Supplementary Table 2), loops on the top-side of the WD40 domain interact with the N-terminal part of α-helix 2 from the N-terminal region of CDC5, which in turn contacts the U6 loop between the ACAGA box and the U2/U6 helix Ia. CypE contacts SYF1, the WD40 domain of PRP17 and the U2 3′ terminal RNP domain. c, d, Fit of the extended U2/branch site helix and U2 nucleotides linking the latter with the U2 Sm core domain into the C* EM density map. Intramolecular U2 snRNA helices IIa and IIc are mutually exclusive structures70, and it was recently shown that U2 helix IIc is formed in the S. cerevisae C complex21, 22. In the human C* complex, U2 nucleotides 88–95 can be fit in a helical conformation into a helical density element located directly adjacent to the U2 Sm core. This is consistent with the possibility that U2 intermolecular helix IIc is formed (at least partially) in the C* complex, as helix IIc formation involves U2 nucleotides directly upstream of the the U2 Sm site. While the density for these U2 nucleotides is well-defined, EM density for complementary U2 nucleotides that are potentially involved in the formation of a helix are observed only at a lower threshold. This suggests that U2 helix IIc is not very stable in our C* complex, and thus may be only partially formed. EM density that would accommodate U2 intra-molecular helix IIa or IIb is not detected in the C* map, suggesting that they are unstable or not formed.
-
Extended Data Figure 10: Structural changes in the spliceosome that are probably mediated by the action of PRP16.Hover over figure to zoom
a, b, Comparison of the position/orientation of the PRP8 RH domain, U2 Sm core plus U2-A′ (S. cerevisiae Lea1) and U2-B′′ (S. cerevisiae Msl1), SYF1 and SYF3 HAT repeats, PRP17 WD40 domain and the catalytic RNA core, aligned relative to the PRP8 En/RT domain, in the S. cerevisiae C complex and the human C* complex. Black circle, catalytic centre. Asterisk, position of the β-hairpin loop of the RH domain. Structural changes probably facilitated by PRP16 include: (1) repositioning of the extended branched intron structure away from the catalytic centre together with a similar movement of the PRP17 WD40 domain; (2) a large scale movement of the entire 3′ domain of U2 snRNP and rearrangement of SYF1 and SYF3; (3) translocation of the PRP8 RH domain by approximately 35 Å (bottom) to 60 Å (tip of β-hairpin loop) such that it interacts with the branched intron structure in the C* complex; (4) rearrangement of the PRP8 α-finger; and (5) destabilization of Yju2/CCDC130 and Cwc25/CWC25.
-
Video 1: The human C* spliceosomeThis video shows a 3D EM density map of the human C* spliceosome and fit of spliceosomal RNAs and proteins.