JBC Papers in Press. Published on February 11, 2005 as Manuscript M414481200 RECONSTITUTION OF TWO RECOMBINANT LSM PRO- TEIN COMPLEXES REVEALS ASPECTS OF THEIR ARCHI- TECTURE, ASSEMBLY AND FUNCTION Bozidarka Zaric#, Mohamed Chami$, Hervé Rémigy$, Andreas Engel$, Kurt Ballmer-Hofer#, Fritz K. Winkler# and Christian Kambach# #Paul Scherrer Institut, Biomolecular Research, CH5232 Villigen PSI, Switzerland, and $M. E. Müller Institute for Microscopy, Biozentrum, University of Basel, Klingelbergstr. 70, CH4056 Basel, Switzerland Running Title: LSm complex architecture Correspondence to: C. Kambach, Life Sciences, OFLC 110, Paul Scherrer Institut, CH5232 Villigen PSI, Switzerland, Fax +41 (0)56 310 47 23, E-mail [email protected] D Sm and Sm-like (LSm) proteins form proteins have diversified through evolution, and o w complexes engaging in various RNA process- adopted new functionalities. LSm protein func- n lo ing events. Composition and architecture of tion in archaea is unknown. A structural and ad e the complexes determine their intracellular biochemical study on Archaeoglobus fulgidus d fro distribution, RNA targets, and function. We LSm proteins showed that they bind to RNAse P m h have reconstituted the human LSm1-7 and RNA in vivo and in vitro (6), a feature that has ttp LSm2-8 complexes from their constituent also been observed for several yeast LSm pro- ://w components in vitro. Based on the assembly teins (7). E. coli Hfq is a pleiotropic regulator of w w pathway of the canonical Sm core domain, we RNA metabolism (8). The originally identified .jb c used heterodi- and trimeric sub-complexes to canonical Sm proteins engage in pre-mRNA .o rg assemble LSm1-7 and LSm2-8. Isolated sub- splicing (9). Sm/LSm family members have b/ y complexes form ring-like higher order struc- been shown to participate in mRNA decapping g u e tures. LSm1-7 is assembled and stable in the and degradation (10;11), histone pre-mRNA s t o absence of RNA. LSm1-7 forms ring-like processing (12-14), telomere synthesis (15), n A structures very similar to LSm2-8 at the EM rRNA maturation (16;17), snoRNP assembly p level. Our in vitro reconstitution results illus- (18), pre-tRNA processing (19), and trans- ril 3 , 2 trate likely features of the LSm complex as- splicing (20;21). 0 1 9 sembly pathway. We prove the complexes to be functional both in an RNA bandshift and Sm/LSm proteins are characterized by a anin vivo cellular transport assay. bipartite sequence motif of about 80 amino acids length situated in most members at the N- The Sm and Sm-like (LSm) proteins are terminus. Recently, divergent family members a widespread protein family with members in all with additional domains have been identified kingdoms of life. Phylogenetic distribution sug- (1;4). The conserved motif translates into a fold gests Sm proteins were already present in the common to all Sm/LSm proteins. This Sm fold last universal common ancestor of all present- mediates specific Sm-Sm interaction through a day life forms, and that the family underwent an generic interface, which the various Sm protein explosive diversification with the advent of eu- family members use to build up homomeric (in karyotes (1). Archaebacteria harbour one or two prokaryotes) or heteromeric (in eukaryotes) ring- Sm/LSm genes each. E. coli Hfq and its shaped complexes. These represent the func- orthologues in other gram-negative bacteria are tional form of all Sm/LSm proteins. The com- so far the only known eubacterial LSm proteins mon fold, generic interface and ring-like mor- (2;3). In contrast, eukaryotic genomes appear to phology of Sm/LSm complexes provide a ra- contain 24 or more Sm/LSm genes (1;4). tionale for the observed large variety of RNA Thought to have originally arisen as chaperones targets bound, the diverse complex compositions mediating RNA-RNA interactions (5), Sm/LSm and functions. LSm proteins appear as building 1 Copyright 2005 by The American Society for Biochemistry and Molecular Biology, Inc. blocks for complexes whose composition and differing only in the seventh subunit (LSm8 and architecture determines their intracellular distri- LSm1, respectively, Figure 1). The LSm1-7 as- bution, interaction with RNA targets and non- sembly pathway is even less well characterized Sm effector proteins, and function. The struc- than the LSm2-8 pathway. LSm1-7 has been tural basis for the balance between interaction shown to accumulate in cytoplasmic foci to- specificity and flexibility required for assem- gether with other components of the mRNA de- bling different complexes with some subunits in capping and degradation machinery (28;29). common is unknown. These foci are apparently active sites of mRNA turnover (30), but the available data do not indi- The canonical Sm core domain com- cate whether LSm1-7 assembles in these foci or posed of the seven Sm proteins B, D , D , D , E, elsewhere in the cytoplasm, nor whether it binds 1 2 3 F, and G was demonstrated to assemble in an its mRNA targets as a preassembled complex. It ordered pathway onto a conserved, single- is unknown whether LSm protein sub-complexes stranded stretch on their target RNAs, the Sm analogous to the Sm heterodi- and trimers exist. site of the spliceosomal snRNAs U1, U2, U4, Here we show that stable, soluble human and U5 (22;23) (Figure 1). The pathway is LSm23, LSm48, and LSm567 sub-complexes marked by the RNA-free sub-complexes D D , corresponding to their paralogues SmD D , 1 2 1 2 D B, and EFG (22). The sub-complexes may SmD B, and SmEFG can be obtained by coex- 3 3 constitute stages of the assembly pathway. Sm- pression in E. coli. LSm1 and LSm4 are pro- snRNA assembly occurs in the cytoplasm. After duced from monocistronic vectors. Isolated sub- Do w hypermethylation of the snRNA moiety, the pre- complexes assemble into ring-like higher order n lo snRNPs are transported to the nucleus, where structures, underscoring the preference of eu- ad e Sthmey c moraet udroem toa infu anscsteiomnbally p iasr taic lheisg h(9ly) . r eIgnu lvaitveod, keraoryloogtoicu sL bSimnd ipnrgo tpeainrtsn etros . a sTshoec iafatec t wthitaht bhoetth- d from process involving Sm protein modification by a LSm1-7 and LSm2-8 complexes can be reconsti- http methylase (24) and numerous assembly factors tuted from these components in the absence of ://w like the survival of motor neurons (SMN) pro- RNA suggests that both species assemble in the w w tein. In vitro, the Sm core domain can be as- cytoplasm and bind to their target RNAs as pre- .jb c sembled from Sm protein sub-complexes by the assembled units. We show the recombinant .o rg addition of target snRNA in the absence of any LSm2-8 complex to be functional by in vitro b/ y auxiliary factors ((23) and Kambach et al., un- bandshift with U6 snRNA and by an in vivo cell g u published). Spliceosomal U6 snRNA differs microinjection / intracellular transport assay. es from the other snRNAs in many ways. It is t on A thought to have an entirely nuclear life cycle MATERIALS AND METHODS p (25;26), does not bear an Sm site, and does not ril 3 bind the canonical Sm proteins. However, a Cloning, expression and purification of LSm1 – , 20 1 complex built up of seven Sm-like (LSm) pro- 8 proteins and sub-complexes – LSm2, 3 and 5-8 9 teins 2 – 8 was shown to interact with the 3’ end were subcloned from a human lymphoma U937 of U6 snRNA in the nucleus (27), stabilizing U6 cDNA library (Stratagene) into a modified snRNP and the U4/U6 snRNA interaction (Fig- pUC19 vector. LSm1 and LSm4 were sub- ure 1). cloned from EST clones IMAG p998P2110673Q2 and IMAG p958A041800Q2, The Sm core domain can only assemble respectively. EST clones were obtained from onto its U snRNA target and is only stable in the the genetic resources centre, Berlin (RZPD, presence of the RNA (23). In contrast, the na- http://www.rzpd.de). LSm2/3, LSm4/8, and tive LSm2-8 complex has been shown to be sta- LSm5/6/7 polycistronic T5 expression cassettes ble in the absence of RNA (27). It is likely that were constructed by successive compatible the LSm2-8 complex is assembled in the cyto- overhang cloning using engineered BamH1/Bgl2 plasm, migrates as such to the nucleus, and there sites. The final cassettes were transferred to the binds to U6 snRNA. The LSm2-8 assembly thus pQE30 T5 expression vector (Qiagen, Basel). differs from the canonical core Sm domain path- Expression constructs bear an MRGSH -tag at 6 way. In addition to LSm2-8, a cytoplasmic the N-terminus of the first cistron, followed by a LSm1-7 complex exists that engages in mRNA tobacco etch virus (TEV) cleavage site. LSm1 decapping and degradation (10;11). The two and LSm4 were subcloned into pQE30 as mono- complexes have LSm proteins 2 to 7 in common, cistrons. SG13009[pREP4] (for LSm1, LSm2/3, 2 LSm4, and LSm5/6/7) or BLR[pREP4] (for LSm2-8 were purified by consecutive gel filtra- LSm4/8) E. coli cells were transformed with tion and anion exchange chromatographies. plasmid DNA, and plated out on selective media. LB starter cultures were grown at 30 °C O/N and Electron microscopy and image processing – 2-12l LB media inoculated the next day. Cul- Samples were diluted at 10 to 20 µg/ml. Ali- tures were grown to an OD of 0.8 at 37 °C and quots of 5 µl was stained with 1 % (w/v) uranyl 600 induced with 1mM IPTG. Induction temperature acetate after sample adsorption onto glow dis- was between 25 °C and 37 °C. Cells were har- charged 400 mesh carbon-coated grids. The mi- vested after 4-48h induction, depending on con- crographs were recorded at an accelerating volt- struct. Cell pellets were resuspended in lysis age of 100 kV and a magnification of 50,000×, buffer (20mM HEPES-Na pH 7.50, 0.5-1.0 M using a Hitachi 7000 electron microscope. All NaCl, 10mM Imidazole-Cl pH 7.50, 5mM (cid:533)- micrographs were recorded on Kodak SO-163 Mercaptoethanol), sonicated, and treated with film. Reference-free alignment was performed DNAse1. Insoluble material was removed by on manually selected particles from digitized ultracentrifugation, and supernatants purified by electron micrographs using EMAN image proc- immobilized metal ion affinity chromatography essing package (32). After multivariate statisti- (IMAC) on Ni-charged Hi-Trap chelating Sepha- cal analysis of a set of rotational and transla- rose columns (Amersham Biosciences). LSm tional invariants previously generated, a refer- proteins and sub-complexes were eluted with ence free k-means classification was performed Imidazole step gradients (60, 250 and 500mM). on the resulting footprint file. The resulting Do w If insufficiently pure, samples were subsequently classified images were then aligned and classi- n lo dialysed into 100mM NaCl buffer without Imi- fied iteratively. The class average with the best ad e dtoagzroalpeh, ya n(d1 0s0umbjMec t–e d1 Mto iNoanC el)x.c h Sanamgep lcehsr owmeare- sinig an agla ltloe rnyo. ise ratio were selected and gathered d from frozen in liquid Nitrogen in ion exchange buffer. http In some instances, the MRGSH6-tags were GPC/Static light scattering analysis – LSm sub- ://w cleaved off by TEV protease (1:100 ratio, O/N complexes were run on a Superdex 200 HR10/30 w w R.T.), and the sub-complexes purified by IMAC. gel permeation column using an ÄKTA Explorer .jb c Cloning and expression of Sm protein sub- FPLC (both Amersham Biosciences) coupled to .o rg complexes D1D2 and D3B has been described a miniDAWN static light scattering analyzer and b/ y elsewhere (31). SmEFG heterotrimer was pro- an Optilab DSP refractometer (both Wyatt g u duced from a pET15b vector (Novagen) and pu- Technology Corp., CA, USA). Data were ana- es rified via consecutive IMAC and ion exchange lyzed using Wyatt’s ASTRA 4 software. Ana- t on A chromatographies. 12% SDS-PAGE gels were lytical gel filtrations were run on a Superdex 200 p run with equivalent amounts of total cell extracts PC3.2/30 column using an Ettan HPLC (Amer- ril 3 at time of induction (T0), time of harvest (T4 – sham Biosciences). , 201 T ), soluble material, and pellet (insoluble mate- 9 48 rial). After staining with Coomassie brilliant Analytical ultracentrifugation – LSm sub- blue R250 (GERBU Biochemicals, Germany) complexes were subjected to an equilibrium and destaining, gels were scanned and the bands sedimentation run in 20 mM HEPES-Na pH 7.5, corresponding to soluble and insoluble LSm pro- 200 mM NaCl, 5 mM (cid:533)-Mercaptoethanol buffer teins integrated via densitometry using the pro- on an Optima XL-A analytical ultracentrifuge gram ImageJ 1.29x (Wayne Rasband, National (Beckman Coulter) at 12000 rpm. Data analysis Institute of Health, USA). was performed using the program DISCREEQ (50). Reconstitution and purification of LSm1-7 and LSm2-8 complexes – Individual LSm protein or Electromobility shift assays – Xenopus tropicalis sub-complex preparations were incubated in 4M U6 snRNA or Xenopus laevis U1 snRNA (a urea, 1M NaCl buffer for 2h at 37 °C and then kind gift from Iain Mattaj, EMBL Heidelberg) mixed in equimolar amounts for the assembly of were in vitro transcribed and body-labelled with the desired heptamer. The mix was incubated 32P-UTP. 20000 cpm purified U snRNA was again for 2-5h, and the sample dialyzed against incubated with 5 pmol of an equimolar mixture buffer with progressively less salt (1M and 0.5M of the Sm D D , D B, and EFG sub-complexes 1 2 3 NaCl) O/N at 4 °C. Reconstituted LSm1-7 and or LSm protein heptameric complexes in a buffer containing 20mM HEPES-Na pH 7.50, 3 300mM NaCl, 5mM MgCl , 10% Glycerol, 0.5 greatly enhanced by fusing to two N-terminal Z- 2 µl RNAsin (Promega) and 0.5 mg/ml yeast tags (Staphylococculs aureus protein A IgG- tRNA in a 10 µl assay at 30 °C for 1h, then at 37 binding domain), followed by a His -tag and a 6 °C for 1h. Samples were loaded on 6% native TEV cleavage site. This phenomenon is exem- PAGE gels and run at 4 °C for 2.5 h, 160V. plified by the ZZ-His -TEV-LSm6 purification 6 Gels were autoradiographed wet for 14-16h at - record (Figure 2d). For crystallographic and 80 °C on X-ray film. other studies, we proceeded to express LSm5, LSm6, LSm8, and the complexes LSm5/6, Cell microinjections – REF52 rat fibroblasts LSm5/7, LSm6/7, and LSm5/3 (data not shown). were grown to 60-80% confluency on coverslips In general, the solubility of a given LSm protein in Optimem™ medium with Glutamine (Gibco increased by up to 25-fold by coexpression, as BRL). LSm1-7 was labelled with Alexa555, measured by the supernatant:pellet ratio (Table LSm8 and LSm2-8 with Alexa488 succinimidyl 1). Recombinant LSm protein sub-complexes ester fluorescent dyes (Molecular Probes) ac- were purified by Ni-IMAC followed by ion ex- cording to the manufacturer’s protocol. Excess change chromatography where necessary. For dye was removed by dialysis into microinjection each polycistron, only the first cDNA bears a buffer (20mM HEPES-Na pH 7.5, 150 mM His -tag, the other LSm proteins are isolated 6 NaCl, 5 mM (cid:533)-Mercaptoethanol). Aggregates through sub-complex formation and co- were removed by centrifugation (15 min 13 purification. The complexes and single LSm krpm), and the supernatant injected into cells. proteins were purified to homogeneity, as shown Do w After incubation for 30-120 min, cells were by SDS PAGE (Figure 2a, b, c, d). Sample in- n lo fixed and visualized using an Olympus confocal tegrity is further demonstrated by the successful ad e f8l uoarcetsivcee nctera mnsipcroorst,c oWpeh. e Faot r Ganearlmys isA ogfg LluStmin2in- ctiroynsst.a l liWzaetaioknly odfi fvfararicotiunsg LcSryms taplrso tceoiunl dp rbeep aorba-- d from (WGA, SIGMA) was coinjected at a concentra- tained from LSm6 (Figure 2e) and LSm5/6/7 http tion of 2.5 mg/ml. (Figure 2f). ://w w The purified sub-complexes were characterised w RESULTS biophysically. Analytical ultracentrifugation .jbc .o Expression of canonical human Sm proteins in (AUC) and static light scattering (SLS) experi- rg E. coli from single cistron vectors gives very low ments combined with gel filtration chromatogra- by/ yields or insoluble protein. In contrast, high phy yielded molecular weights that indicate for- gu e s yields of soluble Sm proteins are obtained by mation of higher order structures (Figure 3a, b, c t o n coexpressing the SmD1D2, SmD3B and SmEFG and Table 2): The LSm2/3 heterodimer has a A p svuebc-tocorsm (p(3le1x)e, sa ndfr oCm. Kpamolbyaccisht,r ounnipcu belxisphreedss oiobn- nlyotmicianl aul lmtraocleencutrliafru gwee, itghhet LofS 2m52 /k3D oal.i gIno mtheer adnias-- ril 3, 2 servations). These correspond to the sub- tribution is bimodal at 10 µM concentration, 01 9 complexes identified in HeLa cell nuclear ex- containing a hexamer (10%) and an octamer tract (22). (87%). The MW of LSm5/6/7 is 33 kDa. Ana- lytical ultracentrifugation yields a mixture of Although some LSm proteins can be expressed individual subunits (26%), trimer (25%), more efficiently in a soluble form from monocis- hexamer (40%), and nonamer (8%) species at 16 tronic vectors than their canonical Sm protein µM concentration. Analytical gel filtration paralogues, in general yield is very low and the combined with static light scattering measure- obtained preparations tend to aggregate heavily. ments yields 85 kDa for LSm2/3 and 77 kDa for Based on our experiences with Sm protein coex- LSm5/6/7. These values reflect the heterogene- pression, and to facilitate expression and purifi- ity in oligomer distribution found by AUC. cation of LSm proteins, we initially constructed LSm5/6/7 stays intact during gel filtration and polycistronic expression vectors encoding individual subunits are not observed. Upon incu- LSm2/3, LSm4/8, and LSm5/6/7 cDNAs. These bation in up to 8M urea, the highest elution vol- heterodi- and trimers correspond to the canonical ume of LSm5/6/7 species corresponds to the SmD1D2, SmD3B, and SmEFG sub-complexes, trimer. This stands in contrast to LSm2/3, which respectively. LSm1 and LSm4 were constructed at urea concentrations of 4M and higher falls and expressed as monocistrons for the reconsti- apart to some extent into its subunits (data not tution of LSm1-7. Expression yield and solubil- shown). LSm4/8 aggregates most strongly of ity of single-cistron LSm constructs could be 4 the LSm sub-complexes and does not seem to The reconstituted complexes were then purified form oligomeric higher order structures of de- by ion exchange chromatography (Figure 5c, d) fined stoichiometry (Table 2). followed by gel filtration (Figure 5e, f). In both types of chromatography, they elute as single Our concept on sub-complex higher order struc- peaks, demonstrating sample homogeneity in ture is confirmed by negative-stain electron mi- charge and in size. The SDS PAGE gels of the croscopy. LSm2/3 (Figure 4a, overview) shows purified LSm1-7 and LSm2-8 clearly show the up as ring-shaped structures with slightly smaller presence of all seven different subunits (Figure dimensions than the 8 nm outer diameter and 2 5a, b, lanes 5 and 4, respectively). Molecular nm for the central hole that were measured for weight determination by gel filtration chroma- the native LSm2-8 complex from HeLa cell nu- tography coupled to static light scattering clear extract (27). LSm 5/6/7 shows up mainly yielded a figure of 92 kDa for LSm2-8 (Table 2). as a ring-shaped structure as well but heteroge- LSm1-7 was analysed by analytical ultracentri- neities appear in the background of the electron fugation. At 10 µM concentration, LSm1-7 is a micrographs (Figure 4b, overview). LSm4/8 mixture between heptamer and 14mer. At 20 aggregated too strongly to yield homogeneous and 50 µM, the proportion of 14mer grows (Ta- particles in electron micrographs (data not ble 2). An alternative model with the major shown). After particles classification and subse- components of sub-complex preparations quent class averaging, distinct ring particles can (LSm2/3 octamer, ~100 kDa, and LSm5/6/7 be observed in LSm2/3 galleries (Figure 4a, bot- tom) having an outer diameter of ~7 nm and a hexamer, ~66kDa) does not satisfactorily fit the Dow data. The comparatively large losses through n cavity of less than 1.5 nm. Considering the mass lo of the LSm2/3 heterodimer (25 kDa), we suggest aggregation in the LSm1-7 AUC run stem from ade that the resulting class averages correspond to utasiinnign ga no loingloym pearrst ioalfl yd ipfuferirfeinetd csoammppoles istitoilnl ctohna-t d from wocittahm tehreic ALUSCm 2m/3e a(s[uLrsemm2e/n3t]s4. ) iAn ltahcocuogrdha nthcee athpep ehare ptota mbee rf.a r mIno rec opnrcolnues itoon ,a gognrecgea tpiounri ftiheadn, http://w LSm5/6/7 preparation did not show up as ho- LSm1-7 does not fall apart into its sub- w w mogenous as LSm2/3, class averaging yielded complexes and subsequently rearranges into al- .jb ring shaped particles having a size of ~7 nm and ternate higher order structures in solution. Both c.o a cavity of 1.5 nm suggesting a hexameric ar- heptameric complexes elute from gel filtration brg/ rangement ([LSm5/6/7]2 , 2x 33kDa) based on chromatography with elution volumes corre- y g u the AUC data. Nevertheless, from the EM sponding to molecular weights between 85 and es analysis, a nonameric arrangement can not be 99 kDa (Figure 5e, 5f), in accordance with the t on A excluded. Smaller particles appear in the back- calculated masses (86 and 93 kDa, respectively). p ground (Figure 4b, arrows) and could represent It should be noted that the in vitro reconstitution ril 3 LSm5/6/7 trimers. The small size of such parti- of the canonical Sm core domain works with , 20 1 cles was not suitable for classification. good yields in native buffer, whereas LSm1-7 9 and LSm2-8 do not form in vitro under these The Sm core domain can be reconsti- conditions. tuted in vitro from recombinant Sm sub- complexes and U snRNA with good efficiency We proceeded to take electron micro- under native buffer conditions (33). The LSm2- graphs of negatively stained LSm1-7 and LSm2- 8 complex isolated from HeLa cell extract is, in 8 complexes (Figure 4c and d). Both species contrast, stable and likely to assemble in the ab- show up as ring-like shapes. In the case of sence of RNA (27). For our LSm complex in LSm1-7, as for LSm567, smaller particles ap- vitro reconstitution protocol, we required to dis- pear in the background (Figure 4C, white cir- rupt the higher order structures formed by the cles) and could represent fragments of LSm17. sub-complexes. The in vitro reconstitution proc- The LSm2-8 preparation appears more homoge- ess should then be guided by relative thermody- neous (Figure 4d). The LSm1-7 outer dimension namic stability. Reconstitution was carried out is about 7 nm. The central accumulation of stain by mixing equimolar amounts of LSm2/3, measures less than 1.5 nm. For LSm2-8, the LSm4/8 and LSm5/6/7 (for LSm2-8) or LSm4, values are 8 and 3 nm, respectively. Since in LSm1, LSm2/3, and LSm5/6/7 (for LSm1-7) contrast to the canonical core snRNP domain under semi-denaturing conditions (for details, (23), these preparations do not contain RNA, we see materials and methods) followed by dialysis. have to assume the central feature to represent a 5 cavity, or hole. The recombinant LSm2-8 archi- recombinant LSm complexes show correspond- tecture thus corresponds to its native counterpart ing subcellular distributions, we injected fluo- at the ultrastructural level (27). Remarkably, the rescently labelled LSm1-7 and LSm2-8 into rat LSm1-7 complex is very similar to LSm2-8 at fibroblasts. LSm1-7 distributes mainly in the this resolution (Figure 4, compare panels c and cytoplasm, where it accumulates in distinct spots d). Our data provide the first experimental evi- (Figure 7a). In contrast, LSm2-8 concentrates in dence that LSm1-7 assembles into a structure the cell nucleus (Figure 7b). LSm2-8 nuclear that is very similar to LSm2-8 and the canonical migration is specific, since LSm8 on its own core domain. Whether this is also true at the does not migrate to the nucleus, but leads to atomic level has to await the solution of the crys- formation of pre-apoptotic granules instead tal structure of the three complexes. (Figure 7c). Nuclear transport of LSm2-8 is an active process, since it can be transiently The native LSm2-8 complex was ini- blocked by coinjection of wheat germ agglutinin tially isolated from U4/U6 snRNP and shown to (WGA, Figure 7, compare panels d with e, (38)). bind to U6 snRNA in vitro (27). In order to test Labelling of the LSm1-7 and LSm2-8 heptamers the function of our recombinant LSm complexes does not destabilize the complexes: On gel filtra- in vitro and assess binding specificity, we per- tion, the labelled species elute at the same elu- formed electrophoretic mobility shift assays tion volume as the non-labelled heptamers (Sup- (bandshift) with U6 and U1 snRNA. Pre- plementary Figure S1a and b; the dye absorption assembled, purified LSm2-8 shifts U6 snRNA (Figure 6a, lane 5), whereas individual sub- at 495 nm coincides with the complex elution Dow profile, and data not shown), and the peaks con- n complexes LSm2/3, LSm4/8, or LSm5/6/7 do lo not (Figure 6a, lanes 2, 3, and 4). Neither does a tain all seven resident proteins in stoichiometric ade reconstituted LSm particle in which either of the amounts (Figure S1c and d). d fro m ssuinbg-lceo mLSplmex epsr o(tFeiignu r(eL S6bm, 6la, nFeisg 4u,r e5 ,6 abn,d l a6n) eo r7 a) DISCUSSION http has been left out. Leaving out LSm2/3 leads to As research in genomics and RNA proc- ://w w sample aggregation and shift into the well (lane essing progresses, ever more proteins containing w 4). LSm1-7 complex does not shift U6 snRNA the Sm/LSm motif are discovered, and new .jbc .o under the same conditions (Figure 6b, lane 3). functionalities of LSm protein complexes are rg Specificity of complex formation could further identified. Still, very little is known about LSm by/ g be demonstrated by adding an LSm2/3-specific complex assembly pathways, nor how the archi- u e s antibody to the reaction mixture. This assay tecture of the often very similar complexes de- t o n leads to a supershift (Figure 6c, lane 3). In the termines their specific function. Eukaryotic A p same assay conditions, LSm2-8 does not bind Sm/LSm proteins have a strong preference to ril 3 strongly to U1 snRNA (Figure 6d, lane 3), in form heterooligomers rather than homooli- , 2 0 contrast to a 1:1 mixture of the seven canonical gomers. Canonical Sm proteins form RNA-free 1 9 Sm proteins (D B + D D + EFG, lane 2). The heterodimers and –trimers that likely represent 3 1 2 combination Sm proteins added to U6 snRNA intermediates on the core snRNP domain assem- leads to aggregation and shift into the well (lane bly pathway. Specificity of LSm-LSm interac- 5). Increasing the stringency of the assay abol- tion impacts directly on the assembly process, ishes the slight background LSm2-8 – U1 since lack of it must be overcome by the help of snRNA interaction (panel d, lane 2) as well as cellular assembly factors. Here we have pre- the aggregation in the Sm – U6 snRNA reaction sented results that show how LSm complex self- (lane 5), but invariably the LSm2-8 – U6 snRNA assembly can be successfully carried out in vitro bandshift as well (data not shown). In summary, in the absence of such assembly factors, and re- the recombinant LSm2-8 complex shows the sults in a correct architecture and functional hep- same RNA binding characteristics as its native tameric and LSm2-8 and presumably also the counterparts, and is functional in vitro. LSm1-7 complex. Consistent with its functions in splicing LSm proteins tend to be more soluble and rRNA processing, the LSm2-8 complex has than Sm proteins when produced singly. Never- been found to localize in the nucleus (34), and in theless, providing another LSm protein as a het- the nucleolus (35). LSm1-7 accumulates in par- erologous binding partner in the same cell gen- ticular cytoplasmic features called foci (28) or erally increases solubility by a factor of up to 25. GW bodies (36;37). In order to test whether our In this way, we were able to produce soluble, 6 stable LSm2/3, LSm4/8, and LSm5/6/7 sub- dimensions (less than 1.5 nm and about 7 nm for complexes, corresponding to SmD D , SmD B, the inner and outer diameter, respectively) than 1 2 3 and SmEFG. However, the increase in solubility for LSm2-8 (see below). Since these compara- is independent of the combination, and does not tively smaller values are also found for the (hep- correlate with coexpression of assumed nearest tameric) LSm1-7 and the (octameric or neighbours in the LSm2-8 ring. We conclude a hexameric) LSm2/3 rings, one cannot conclude lower degree of LSm-LSm interaction specific- that size correlates either with the number of ity, as compared to the Sm-Sm interactions in subunits in the ring or with RNA binding charac- the core snRNP domain. The results are in line teristics. Native LSm sub-complexes probably with yeast two hybrid data indicating a greater do not bind RNA by themselves. This would fit promiscuity for LSm than Sm proteins (39;40). with our results that in contrast to LSm2-8, none The findings impact on the cellular LSm2-8 as- of the sub-complexes (or combinations thereof) sembly pathway: Lower intrinsic interaction binds U6 snRNA. Formation of higher order specificity puts a higher demand on assembly structure by the sub-complexes reflects the factors guiding productive ring assembly. In- predilection of eukaryotic LSm proteins to form deed, LSm2-8 assembly in vivo could be pro- heteromeric, rather than homomeric complexes, moted by snRNP assembly factors like SMN, as do their prokaryotic homologues. Our singly which has been demonstrated to interact with expressed LSm proteins generally form aggre- LSm4 in vitro (41;42). However, evidence that gates without defined stoichiometry (data not these interactions are also present in vivo is as shown). The ring closure is likely due to the Do w yet lacking (U. Fischer, personal communica- need to satisfy all available Sm-Sm interfaces by n lo tion). interaction with another LSm molecule. The ad e We could show the LSm2/3 and Sphmo-bSimc einletemrfeancte (p3o1s)s,e swsehsic ah ,p riof nounusnacteisdf iheydd rboy- d from LhiSgmhe5r/ 6/o7r desru br-icnogm-sphlaepxeeds hteot eroasosleigmobmlee rs inbtyo binding to a specific partner, leads to rapid ag- http negative-stain electron microscopy. From the gregation and precipitation. ://w w AUC results that indicated predominantly oc- The sub-complexes can be assembled in w tamers (tetramers of dimers) for LSm2/3, we vitro and in the absence of any RNA into LSm1- .jbc .o assume the LSm2/3 rings to represent octamers. 7 and LSm2-8 complexes. It was previously rg However, all Sm or LSm rings reported so far shown for native LSm2-8 to be stable in the ab- by/ g have either six or seven subunits, and it remains sence of its target, U6 snRNA. For LSm1-7, u e s to be proven that the generic Sm-Sm interface similar information has not yet been available. t o n defined by the D B and D D heterodimers is Human LSm1-7 accumulates in cytoplasmic foci A 3 1 2 p capable of accommodating eight subunits in a which are assumed to represent sites of mRNA ril 3 ring. Alternatively, the LSm2/3 rings could be decapping/degradation, or storage forms of the , 2 0 representing hexamers present in the LSm2/3 involved enzymes (28-30). However, it is not 1 9 preparation at low concentration, in line with known whether LSm1-7 is pre-assembled else- LSm5/6/7. Hexamer formation by an Sm sub- where in the cytoplasm, arrives at these foci as complex was previously demonstrated as a fea- an RNA-free complex, and binds to its target ture of the human EFG trimer (43). The physio- mRNAs on site. Our data suggest that LSm1-7 logical significance of this hexamer could not be indeed binds to mRNA in a pre-assembled form. demonstrated, and indeed the later establishment The reconstituted complexes elute as of the Sm core domain stoichiometry proved that single peaks from ion exchange chromatogra- the (EFG) complex is not part of the final hep- 2 phy, demonstrating sample homogeneity in tameric ring and most likely represents a storage charge and in size. Because of the great varia- form for the three proteins. Presence of the tion in pI within the LSm subunits (from 4.3 for hexamer does not preclude heptamer formation LSm8 to 10.0 for LSm4), it is thus very unlikely in vitro: Recombinant EFG preparations also that the LSm1-7 and LSm2-8 preparations con- show the hexamer, but can efficiently be recon- sist of several sub-populations, each lacking one stituted into a functional Sm core domain by the particular LSm subunit and containing two of addition of SmD D , SmD B, and U1 snRNA 1 2 3 another instead. The SDS PAGE gels of the pu- under native buffer conditions (C. Kambach, rified heptamers clearly show the presence of all unpublished). Electron micrographs of the seven different subunits. Both LSm1-7 and LSm5/6/7 sub-complex rings indicate smaller LSm2-8 preparations are homogeneous in size as 7 well: They elute as single, Gaussian peaks from solved with RNA oligonucleotides, the RNA gel filtration with elution volumes corresponding molecules mainly wrap around the rim of the to the expected molecular weights of the hep- pore, although in one case, additional binding tamers. The accuracy of molecular weight de- sites on the ring surface have been observed termination for LSm2-8 by static light scattering (48). This stands in contrast to the original con- is better than 6%. Since the smallest subunit, cept that in the core snRNP domain, the Sm site LSm6, has a weight of 9.1 kDa, representing target RNA threads through the heptamer’s pore. about 10% of the complex’s mass, the value of The concept was based on the electrostatics of 92 kDa obtained for LSm2-8 (nominal MW = 86 the core domain model and the position of con- kDa) is only compatible with a subunit number served residues assumed (and later shown) to of seven. Similarly, the AUC analysis of LSm1- bind RNA (31;49). Structural evidence to cor- 7 demonstrates the presence of a heptameric roborate this idea has so far only been obtained species in solution, which is at equilibrium with at the ultrastructural level, by cryo-electron mi- higher order oligomers, but not with smaller croscopy of the U1 snRNP (50). For LSm2-8 – complexes like those observed in the sub- U6 snRNA interaction, the binding determinant complex AUC runs. This result further illus- has been shown to be the U stretch at the 3’ end 5 trates the complex’s stability and homogeneity. of U6 snRNA (27). This target is freely accessi- Taken together, sample homogeneity and com- ble to a preassembled complex. Hence it is pos- position together with the molecular weight de- sible that the RNA threads through the LSm2-8 termination results provide strong evidence for a central cavity. However, the smaller pore di- Do w “one of each subunit” stoichiometry of the re- ameter of the recombinant LSm1-7 complex n lo combinant LSm complexes, in line with the ar- could indicate differences to LSm2-8 in RNA ad e chitecture of the canonical core Sm domain (44). bdienaddienngy. la teLdS mmR1-N7A bs.i n Adsl thtoou gthhe t h3e’ R UNTAR bsi ndo-f d from show thNate greactiovme binstaanint LSelmec2t-r8o nh asm aic rroinggr-alpikhes ing determinants for the LSm1-7 complex have http architecture with a diameter of about 8 nm. This not been characterized in detail, LSm1-7 pre- ://w sumably does not bind to the extreme 3’ end of w shape and size is highly similar to the one previ- w ously observed for the native LSm2-8 complex its target mRNAs. At least in some cases, sec- .jbc ondary structure elements found in many of its .o isolated from HeLa cell nuclear extract (8 rg nm,(27)) and core snRNP domain from the same target 3’ UTRs are likely to prevent the RNA by/ threading through the LSm1-7 hole. The estab- g source (45). The pore diameter we observed for u lished biochemical features of LSm1-7 and es the recombinant LSm2-8 complex is distinctly t o LSm2-8 – RNA interaction fit very well with the n larger than in the native LSm2-8 complexes (3 A concept that both LSm1-7 and LSm2-8 assemble p vs. 2 nm, respectively (27)). Since the recombi- in the absence of RNA, are transported to their ril 3 nant complex shows the same RNA binding site of action, and bind to their targets on site – , 20 specificity, this difference must remain unex- 1 possibly using different binding modes. Eluci- 9 plained at the present time. The LSm1-7 rings dation of the exact mode of LSm1-7 and LSm2- appear to be slightly smaller, measuring ~ 7 nm 8 – RNA interaction will have to await solution across and a pore diameter of less than 1.5 nm. of the respective crystal structures. Thus, recombinant LSm1-7 and LSm2-8 com- plexes are similar to one another and to the na- Recombinant LSm2-8 binds to U6 tive Sm/LSm complexes at this level, demon- snRNA in vitro, whereas LSm1-7 does not. The strating that LSm1-7 architecture follows the RNA binding characteristics of the two native generic Sm/LSm complex pattern. complexes are thus reflected by their engineered counterparts. However, we do not as yet possess A pore diameter of 15 Å agrees well a suitably short RNA target to demonstrate spe- with the range observed in archaebacterial LSm cific interaction with LSm1-7. Indeed the pre- protein complexes, free or bound to RNA. Dis- cise nature of the binding determinants on target tances vary from 8.8 Å for the narrowest point in mRNA for the LSm1-7 complex is currently not P. aerophilum LSm1, (46) to about 13 Å in A. known. The validity of using U6 snRNA inter- fulgidus LSm1 (6). The co-crystal structure of action as a measure for LSm2-8 function is un- hexameric E. coli Hfq with an AU G RNA oligo 5 derscored by the fact that only the integral shows that pore size increases from 12Å for the LSm2-8 complex specifically binds to U6. RNA-free hexamer to 15Å for the RNA- Leaving out a single LSm protein or one of the complex (47). In all LSm co-crystal structures 8 sub-complexes from the reconstitution procedure sult could be due to the production mode of the produces complexes incapable of binding U6 protein and time course of the experiment: Sin- snRNA. This observation holds in spite of the gly expressed, our recombinant LSm8 forms likelihood that all these mixtures will form ring- aggregates likely to mask a resident nuclear lo- shaped higher order structures, just as the sub- calization signal. The aggregates are probably complexes themselves. Ring-shaped multimers also toxic to the cells, explaining the occurrence are ubiquitous in nucleic acid binding complexes of pre-apoptotic granules. Conversely, within and other cellular processes (51-53). The ring the ~36 hours of the transfection experiment, it architecture is thought in general to generate is conceivable that the YFP-labelled LSm8 (pro- new biophysical properties on the resident pro- duced at levels only slightly higher than the en- tein subunits, and often to convey new functions dogenous protein) assembles into functional (54). The failure of the LSm sub-complexes to LSm2-8, which is then the transport substrate bind U6 snRNA shows that the ring architecture (28). On the basis of these experiments, nuclear and the presence of LSm family members in the migration of isolated LSm8 can not be ruled out, complex are not sufficient for specific interac- however. tion. This goes in line with the need for strong Our findings imply that we have to view RNA target discrimination based on the presence the specific interactions and functions of indi- or absence of a single specific subunit. vidual LSm subunits in the context of the ring Our cell microinjections of fluorescently architecture: Exposure and probably juxtaposi- D labelled LSm complexes or proteins show that tion of particular sequence elements in the sub- ow n the intracellular distribution of the recombinant units are likely to be instrumental in defining the lo a heptamers reflects the migration behaviour of complex’s interaction with its target RNA, with de d their native counterparts, implying that the in assembly factors (e.g. a presumptive nuclear im- fro m vitro reconstituted complexes are functional in port receptor for LSm2-8), and effector proteins h vivo. LSm2-8 nuclear transport is active and not (like the exonuclease Xrn1 and the decapping ttp diffusive. Fluorescent labelling of the heptamers factor Dcp1/2 in the case of LSm1-7). Our re- ://w w does not disrupt them. These observations pro- combinant LSm protein complexes represent an w vide some evidence that the transported species ideal test system to study these interactions in .jbc .o is the intact heptamer. In a transfection assay, molecular detail. Our results should contribute rg LSm8 is found to accumulate in the nucleus to the understanding of the pathway of LSm by/ g (28). In contrast, in our cell microinjection as- complex assembly and its regulation, of LSm- u e s say, LSm8 (the subunit likely to bear the nuclear RNA and LSm-protein interaction and function. t o n transport determinant of LSm2-8) fails to accu- A p mulate in the nucleus. The difference to our re- ril 3 , 2 0 1 9 FOOTNOTES The authors thank M. Steinmetz and U. Kutay for critical reading of the manuscript, Tewfik Soulimane for assistance in crystallogenesis, H.J-Schönfeld and B. Pöschl, Roche, Basel, for meas- urement of the static light scattering data, E. Kusznir and F. Müller, Roche, Basel, for running and data analysis of the analytical ultracentrifugation, and the other members of the group for fruitful dis- cussions and practical advice. This work was supported by the Swiss National Fund (SNF), grant Nr. 3100-062018. Abbreviations used in text: GPC: Gel permeation chromatography, IMAC: Immobilised metal ion affinity chromatography, IPTG: Isopropyl-thio-(cid:533)-D-galactoside, SLS: Static light scattering, EST: Expressed sequence tag, AUC: Analytical ultracentrifugation, WGA: Wheat Germ Agglutinin. FIGURE LEGENDS Fig.1. Schematic representation of the three best characterized Sm/LSm protein complexes: Sm core domain, part of the spliceosomal U1, U2, U4 snRNPs, LSm2-8 binding to the 3' end of U6 snRNA, and LSm1-7, binding to the 3' UTR (thin line) of mRNAs destined to be degraded in the cytoplasm. The latter complex has been shown to interact with other components of the mRNA decapping/ deg- radation machinery, the decapping enzyme Dcp1, the exonuclease Xrn1, and the auxiliary factor Pat1. 9 Fig. 2. SDS PAGE gels of LSm sub-complex purification protocols: (a) LSm2/3, (b) LSm4/8, (c) LSm5/6/7. (d) LSm6 (Z-tagged). Panels a-d: Lane 1: uninduced culture, lane 2: induced culture, lane 3: supernatant, lane 4: pellet. Panel d: Non-cleaved, Z-tagged LSm6 (lanes 1-6), TEV cleaved LSm6 (lane 7). IMAC: Immobilized metal ion affinity chromatography; IEX: Ion exchange chromatography; FT: Flow-through. Panels (e) and (f): crystals of LSm6 and LSm5/6/7, respectively. Fig. 3. Gel filtration chromatograms (Superdex 200 HR 10/30 column) of LSm sub-complexes LSm2/3 (a), LSm4/8 (b) and LSm5/6/7 (c). UV traces are in blue (280 nm) and red (260 nm). Fig. 4. Electron micrographs of sub-complexes LSm2/3 (a), LSm5/6/7 (b), and complexes LSm1-7 (c), and LSm2-8 (d). Scale bars are 30 nm. Class averages were created from 641, 440, 993 and 988 particles, respectively, from initial data set containing 1000 particles of LSm2/3, LSm5/6/7, LSm1-7 and LSm2-8. White circles in (b) and (c) point to smaller than average particles in the LSm5/6/7 and LSm1-7 preparations that could, in the case of LSm5/6/7, represent trimers (see text). D Fig. 5. Reconstitution of LSm1-7 (a, c, e) and LSm2-8 (b, d, f) heptamers. SDS PAGE gels show o w n input sub-complexes or subunits (panels a, b), homogeneity in charge is demonstrated by the single lo a peak elution profile from anion exchange chromatography (panels c, d), and in size by the elution pro- de d file from gel filtration chromatography (panels e, f, see text). fro m h ttp Fig. 6. RNA Bandshifts: 32P-UTP labelled, in vitro transcribed U snRNA was incubated with differ- ://w w ent LSm sub-complexes or higher order structures reconstituted from them, run on native PAGE gels, w and autoradiographed. Components present are indicated underneath each lane. Individual sub- com- .jb c plexes do not shift U6 snRNA (panel a, lanes 2-4), whereas LSm2-8 does (lane 5). LSm1-7 does not .org lead to complex formation with U6 snRNA (panel b, lane 3), nor do reconstituted particles lacking b/ y either LSm 2/3 (lane 4), LSm4/8 (lane 5), LSm5/6/7 (lane 6), or LSm6 (lane 7). Incubation of LSm2- gu e 8 with U6 snRNA in the presence of an LSm2/3-specific single chain Fv antibody leads to a complex st o shifted to higher molecular weight (panel c, lane 3, arrow). LSm2-8 does not shift U1 snRNA (panel n A dti,o lnasn, et h3e) ,S wmh eprreoates ian s1 :l1ea md itxot uargeg orefg tahteio 7n coafn Uon6i csanlR SNmA p arnodte sinhsif dt otoe st h(ela wnee l2l) (.p Uannedle rd ,t hlaen sea m5)e. Uco6n di- pril 3, 2 snRNA complex formation with LSm2-8 is unaffected (lane 6). 01 9 Fig. 7. Cell microinjections: Preassembled LSm1-7 was labelled with Alexa555 (a), LSm2-8 (b, d, e) or LSm8 alone (c) with Alexa488 fluorescent dye and injected into REF52 rat fibroblasts. Intracellu- lar distribution was monitored 30-40 min post injection by fluorescence microscopy (a-c). LSm8 in- jection led not to nuclear accumulation, but appearance of peri-nuclear granules possibly indicative of ensuing apoptosis (c). Coinjection of Wheat Germ Agglutinin (WGA) inhibited LSm2-8 nuclear transport up to ~1h (d). Inhibition was reversed upon longer incubation (2h, panel e). Fig. S1. Integrity of labelled LSm1-7 and LSm2-8 complexes: Non-labelled (a) or Alexa488-labelled, purified LSm2-8 (b) was run on an analytical gel filtration column in microinjection buffer (see Mate- rials and Methods). The 215/280 nm peaks at an elution volume of 1.47 ml correspond to the calcu- lated molecular weight of the heptamer. The peak at 2.5 ml is the solvent (ghost) peak after one col- umn volume and does not contain protein. The 495 nm trace in (b) due to the presence of the dye co- incides with the protein complex peak. Peak fractions from non-labelled (d, lane 1) and labelled (d, lane 2) LSm2-8 were analysed on SDS PAGE and contain all seven resident proteins. The analogous experiment with LSm1-7 yielded the same result (c). 10
Description: