‘The EMBO Journal vol.18 10.11 pp.2651-2658, 1998 Co-translational trimerization of the reovirus cell attachment protein Ross Gilmore, Matthew C.Coffey, Gustavo Leone, Kevin McLure and Patrick W.K.Lee’ Depurtment of Microbiology snd Infectious Diseases, Univesity of Calgary Health Sciences Gente, Calgary, Albert, Canala TON NI ‘Conespomding author The reovirus cell attachment protein, ol, is a trimer with a ‘lollipop’ structure. Recent findings indicate that the N-terminal fibrous tail and the C-terminal globular head each possess a distinct trimerization domain, The region responsible for N-terminal trimerization (formation of a triple a-helical coiled- coil) is located at the N-terminal one-third of ol. In this study, we investigated the temporality and ATP. requirement of this trimerization event in the context, of 61 biogenesis. Jn vitro co-synthesis of the full-length, (FL) and a C-terminally truncated (444) o1 protein, revealed a preference for homotrimer over hetero trimer formation, suggesting that assembly at the N-terminus occurs co-translationally. This was cor- roborated by the observation that polysome-associated Gl chains were trimeric as well as monomeric. Trun- cated proteins (d234 and d294) with C-terminal dele- tions exceeding half the length of o1 were found to trimerize post-translationally. This trimerization did not require ATP since it proceeded normally in the presence of apyrase. In contrast, formation of stable FL of trimers was inhibited by apyrase treatment. Collectively, our data suggest that assembly of nascent Ol chains at the N-terminus is intrinsically ATP inde- pendent, and occurs co-translationally when the ribo- somes have traversed past the midpoint of the mRNA. Kewords: protein oligomerization/rabbit reticulocyte Jysate/reovirus protein G1 Introdu Significant advances have been made in recent years in the search for factors and parameters that dictate the correct folding of polypeptide chains. While the general concept that the final conformation of a protein is deter- mined by its amino acid sequence still holds true (Anfinsen, 1973), there is now overwhelming evidence that, in the call, protein folding is mediated by molecular chaperone proteins (reviewed by Georgopoulos and Welch, 1993; Hendrick and Hartl, 1993; Ellis, 1994; Hartl et al 1994). These chaperones assist protein folding in an ATP- dependent manner by preventing and reversing. hydro- phobic interactions that lead to aggregates or misfolded conformations. The initial demonstration by Beckmann 1a. (1990) that nascent polypeptides in the mammalian © Oxford Univorsity Press eytosol associate with Hsp70 has led to the hypothesis, that protein folding begins during biogenesis. Using firefly luciferase synthesized in rabbit reticulocyte lysate as a model system, Frydman et al. (1994) found that the growing polypeptide interacts, in an orderly manner, with a specific set of molecular chaperones, including Hsp70, Hsp40 and the chaperonin TCP-1 (t-complex polypeptide 1) complex (TRC), Their data suggest that Hsp70, in cooperation with Hsp40, is able to bind to very short nascent chains (~40 exposed residues), whereas TRIC interacts only with longer chains (~100-150 residues). Co-translational folding occurs once the nascent chain has, attained a length of ~250-300 residues, but the rest of the 550 residue long protein undergoes maturation folding only after release from the ribosome. That nascent luciferase peptides start to attain their native tertiary structure during protein synthesis on the ribosomes was also deduced independently by Kolb ef al, (1994) who carried out continuous monitoring of the enzymatic activity of newly synthesized luciferase in a wheat germ cell-free translation system. Using an Escherichia coli eell-rce system, Fedorov and Balwin (1995) have shown that the B-subunit of bacterial luciferase also undergoes co-translational fold ing, which presumably accounts for the rapid formation of the native structure in the cell. The demonstration that heme binds to nascent globin chains suggests that globin also folds eo-translationally (Komar et al., 1993). While the biogenetic pathway of monomeric proteins is being revealed in terms of co- and post-translational folding and of the involvement of chaperones in these processes, surprisingly litle is known about such processes for homooligomerie proteins in the eytosol. In this regard, a most fundamental question hitherto unanswered is whether these proteins assemble co-translationally or post-transla tionally. In the case of membrane-bound homooligomers, such as the influenza virus hemagglutinin and the vesicular stomatitis virus (VSV) G protein, available evidence suggests that newly synthesized full-length proteins are anchored to the endoplasmic reticulum (ER) as partially folded monomers which subsequently undergo assembly with concomitant maturation folding (reviewed by Doms et al., 1993). This post-translational assembly mechanism is unlikely to be adopted by cytosolic oligomers in view of the lack of any confining structure equivalent to the ER, For these proteins, co-translational assembly with the initial interaction occurring between nascent chains would represent a logical, if not the sole, altemative. In the present study, we test this hypothesis by examining. the initial assembly step that Teads to the generation of the trimerie reovirus cell attachment protein (protein 1), The reovirus 61 trimer is located at the 12 vertices of the icosahedral virion (Lee et ai., 1981; Furlong et al, 1988). It is highly asymmetric with an N-terminal fibrous tail thatis anchored to the virion, and a C-terminal globular 2651RGilmore et al pulse _chase A BAB — — Itrimer = — | monomer 123 4 Fig. 1, Pulse-chase analysis of full-length a (A) and a C-terminal tmuneated 1 (444) (B) synthesized in vio, Transcripts encoding these ‘wo proteins were. translated separately in rabbit reticulocyte lysate in the presence of "S]methionine for 7 min (pulse) and then subjected to ultacentrifupation at 4°C to pellet the polysomes. The superatants (pulse), which contained no translation activity, were then incuaed at 37°C for 60) min (chase). Both pulse and chase samples were analyzed by SDS-PAGE under now-dssecating conditions which allowed for te detection of timerie 01 forms stabilized via the N-terminss but ot the C-terminus (i. preincubation of samples in peotein sample butferat 97°C, followed by electrophoresis caried out at °C), ‘head that interacts with the cell receptor (Banerjea et al, 1988; Furlong et al., 1988; Yeung er al. 1989; Fraser et al, 1990; Mah et a, 1990; Duncan et al, 1991; Leone eal, 199 lab; Strong et al, 1991). Evidence from in vitro translation studies has revealed that these two structurally distinct domains are generated by independent trimerization events (Leone et al, 1992), The two domains also differ in stability, with the N-terminal trimeric Gchelical coiled-coil being significantly more stable than the C-terminal trimeric globular head (Strong es al., 1991; Tumer et al., 1992), These characteristies, together with the relative ease with which inv vitro-generated 61 folding and oligomerization intermediates can be resolved and identified using SDS-PAGE under non-dissociating condi- tions (Leone e7 al., 1992), have made Gl an attractive model system for the study of the sequence of events involved in the folding and assembly of oligomeric proteins in the cytosol. In this study, we examined the initial (N-terminal) assembly step of the 61 trimer in terms of ‘temporality (co-translational versus post-translational) and ATP requirement by analyzing full-length and various truncated G1 forms synthesized in rabbit reticulocyte lysate. Ourresulls suggest that G1 N-terminal trimerization ‘occurs co-translationally, with nascent chains interacting ‘with each other when they have traversed past the midpoint cf the polysome, and that this process is ATP independent. Results Oligomeric status of newly synthesized fulhlength ot We haye established previously an in vitro ol translation system to elucidate part of the G1 folding pathway and 2652 A B [A/B] 1 20 35545 5 Fig. 2, Analysis of ia vitro co-ranslation produets of full-length (A) and truncated (a4) (B) SU wranscripts at various Eoneentations, The two transcripts were translated separate, or mine (te Tati) and sarious dilutions ofthe mixture were trslated. Rest ‘were stopped after 60 min and samples were analyzed by SDS-PAGE ‘under not-dssociating conditions (37°C preincubation). Electr phoresis conditions were such thatthe gl was run for w further 5 h lafler the dye front had reached the botlom of the gel. Reactions in lanes and 2 contained 200 ng of RNA each. Reactions in lanes 3-3 ‘contained 1.2 ug, 400 ng and 133 ng. wespectively, of I: mature of FL an 4 RNA, All the lanes were from the same gel, However, ‘due tothe ciferent amounts of RNA wsed for traslation, different ‘exposure times were used sch that each fane shown had pprovimately the same relative protein band intensities i.e lanes 1 sn 2 were exposed for 36 hand lanes 3-5 were exposed for 12.36 And 108, respectively), have shown that the N- and C-terminal halves of 61 each possesses a distinct trimerization domain (Leone et al., 1992). To determine the timing of the initial trimerization event (at the N-terminus) relative to translation, we first ‘examined the oligomeric status of newly synthesized full- length (FL) ol, To this end, FL $1 transcripts encoding I were translated in rabbit reticulocyte lysates for 7 min Polysomes were then removed by ultracentrifugation at 4°C and o1 products in the supernatant were chased further for 60 min at 37°C. Pulse and chase samples were analyzed by SDS-PAGE under non-dissociating conditions, that allowed for the detection of stable trimeric 61 forms (stabilized via the formation of the three-stranded o-helical coiled-coil at the N-terminus). The results (Figure 1) show that, immediately after release from polysomes, 61 migrated as monomers (49 kDa molecular weight) under the conditions used (lane 1). However, a significant population could eventually be chased into the SDS-stable Uwimeric form (lane 3). Essentially the same results were obtained using a trincated 61 with 44 amino acids deleted from the C-terminus (lanes 2 and 4), We can therefore conclude that stable N-terminal trimer formation occurs posttranslationally, It was necessary to determine whether the newly released 61 polypeptides were true monomers (in which case the trimerization mentioned above would be a de facto post-translational event) or simply unstable (SDS-sens- itive) 1 trimer complexes that were formed during translation, and that were converted to the SDS-stable form upon subsequent chase. ‘To this end, approximately equimolar amounts of transcripts coding for FL (A) and the C-terminally truncated product (B) were co-translated at various, but proportionately equal, RNA concentrations, and the products analyzed by SDS-PAGE under non- dissociating conditions. It was rationalized that if theReovirus protein «1 trimerization Nondonaturing SDS-PAGE PrP eT 2 = monomer \— Peak Fractions — @i-27) Donaturing SOS-PAGE E < “ a <—— Prak Fractions — Fraction Number @t-27) Fig. 3. Detection of polysome-bound trimers. (A} Sedimentation profile of polysomes from an in vitwo G1 translation reaction. Protein of was tuansated in vitro for 10 min and the reaction mixture was subjected (0 eenfugation through a 10-43% sucrose gradient as described in Materials and methoss. The arrow indicates the direction of sedimentation. Gradient fractions were analyzed for absorbance st 254 oyn. (B) SDS-PAGE af! peak polysomal fractions. Peak fractions were subjeted to non denaiuring ot denaturing SDS-PAGE resins in protein sample bufer Jor 30 min at °C, followed by SDS-PAG Deiled for 5 min in protein sample buffer prior to SDS-PAGE at ron temperate, The kflmos lane in each gel contained Prior to sucrose eraent centlusation, newly released chains were truly monomeric, A and B 31 products would assort randomly to generate the four trimeric species Ay, AxBj, A,B) and B, in the ratio of 1:3:3:1, Furthermore, this ratio would not change with the proportionate reduction of the concentrations of the wo transcripts in the reaction mixture, If, on the other hand, trimeric structures (or structures poised to trimerize) were formed co-translationally, then the predominant species formed would be the homotrimers A, and B,, Formation of the heterotrimers AB) and A,B2 would still be possible at high transcript concentrations due to the interaction between the N-termini of nascent GI chains from adjacent transcripts. At low transcript concentrations, however, the formation of such heterotrimers would be expected to be affected drastically relative to that of homotrimers. The results of such an experiment are presented in Figure 2. Even at high transcript concentration (40 jig/ml, the distribution of the four trimeric species (A3, A2B,, AjB; ‘and By) was not in the ratio of 1:3:3:1 as would be expected had assortment occurred randomly (lane 3), ‘The observed ratio of ~I:I:1:1 suggests a preference of homoirimer over heterotrimer formation. Importantly, this 1 non-denaturing SDS-PAGE, samples were lo cared out a 4°C. For denaturing SDS-PAGE. samples were bias was amplified drastically when the amounts of the two RNA transcripts were reduced proportionately (lanes 4 and 5), These results strongly argue for the notion that nascent G1 chains are recruited predominantly co- ‘ranslationally in unstable complexes, and that the chains within these complexes are committed, upon their release from polysomes, to form stable trimers (via the Neterminas). Detection of trimeric 01 forms on polysomes In an attempt to demonstrate that o1 nascent chains indeed form trimers while on polysomes, reactions containing translated FL transcripts were subjected to suerose gradient centrifugation (Figure 3A), Fractions corresponding to the well-resolved polysomal peak were then analyzed directly by SDS-PAGE under conditions that were less stringent than those used in Figure 1 (i.e. pre-incubation of samples in protein sample buffer at 4°C rather than 37°C, followed by electrophoresis carried out at 4°C). The results (Figure 3B) show that, in addition to 61 monomers, 61 timers were also present. These trimers would represent the homotrimers whose assembly was favored in the co- 2653