key: cord-0008569-e6460oan
authors: Strong, James E.; Leone, Gustavo; Duncan, Roy; Sharma, Rajendra K.; Lee, Patrick W.K.
title: Biochemical and biophysical characterization of the reovirus cell attachment protein σ1: Evidence that it is a homotrimer
date: 2006-09-11
journal: Virology
DOI: 10.1016/0042-6822(91)90818-v
sha: c644e80da9d2625ae411f6a08b117d4b14684a6f
doc_id: 8569
cord_uid: e6460oan

The oligomerization state of the reovirus cell attachment protein σ1 (49K monomeric molecular weight) was determined by biochemical and biophysical means. Full-length (protein product designated A) and C-terminal truncated (protein product designated B) serotype 3 reovirus Sl mRNA transcripts synthesizedin vitro were cotranslated in a rabbit reticulocyte lysate, and the products were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under conditions which allowed for the identification of oligomeric forms of σl. A total of four oligomeric protein bands (corresponding to A(3), A(2)B(1), A(1)B(2), and B3, respectively) was consistently observed, which suggests that the protein is made up of three monomeric subunits. Biophysical characterization of purified σ1 using column filtration and sucrose gradient sedimentation analysis confirmed the highly asymmetric shape of σ1 and allowed us to determine the molecular weight of the native protein to be ∼132K (a trimer). Similar biophysical analysis on the two tryptic fragments of the σ1 [N-terminal fibrous tail (26K monomeric molecular weight) and the C-terminal globular head (23K monomeric molecular weight)] yielded molecular weights of 77K and 64K, respectively, both again corresponding to trimers. We therefore conclude that protein σ1 is a homotrimer and provide, with supportive experimental evidence, a rationale for the anomalous behavior of the oligomeric protein in SDS-polyacrylamide gels, which, coupled with chemical cross-linking studies, has in part led to the previous suggestion that σ1 might be a higher order oligomer.

Of all the proteins encoded by the double-stranded RNA genomic segments of reovirus, the e1 protein has been most thoroughly studied . Protein al is a minor outer capsid protein situated at each vertex of the outer capsid of the viral icosahedron (Lee et al., 1981 ; Furlong et at, 1988) . This protein functions as the reovirus cell attachment protein (Lee et at, 1981 ; Armstrong et al., 1984) and the viral hemagglutinin (Weiner et al., 1978 ; Yeung et al ., 1987) .

In view of the important roles played by al in reovirus infection, considerable effort has been made to probe the structure-function relationships of this protein .

The S 1 gene segment, which encodes the e1 protein, has been sequenced for all three reovirus serotypes (Nagata at al., 1984 ; Cashdollar et al., 1985 ; Bassel-Duby et al., 1985 ; Duncan at al., 1990 ; , and the deduced amino acid sequences have been analyzed to predict various structural motifs . Under electron microscopy the purified protein appears as a lollipop-shaped structure with a long fibrous tail topped with a globular head (Banerjea et al ., 1988 ; Furlong et al., 1988 ; Fraser et al., 1990) . Similar structures are also seen extending from the surface of the To whom reprint requests should be addressed . 23 virion with the head region most distal from the virion (Furlong et at, 1988) . The head corresponds to the C-terminal portion of al which is predicted, from sequence analysis, to form a complex globular structure (Bassel-Duby at a Nibert et al., 1990) and which contains the cell receptor recognition domain (Negate etal., 1987 ; Yeung eta!, 1989 ; Duncanetal., 1991) . Interestingly, treatment of a1 with trypsin cuts the protein approximately in half, essentially separating the head from the stalk of the lollipop, with the C-terminal half still fully capable of binding to cell receptors (Yeung et at, 1989) . The N-terminal half of the protein represents the fibrous stalk of the lollipop and contains a heptapeptide repeat of hydrophobic residues, suggestive of a coiled-coil structure (Bassel-Duby at at, 1985 ; Nibert et al., 1990) . This heptad repeat region was subsequently shown to be responsible for al oligomer stability and to possess intrinsic oligomerization function (Banerjea and Joklik, 1990 ; . That the N-terminus indeed represents the proximal end is also evident from the demonstration that the N-terminal one-quarter of at possesses an intrinsic virion-anchoring function (Mah at at, 1990) . The actual virion-anchoring domain is presumed to reside in the N-terminal hydrophobic tail (<20 residues) of the protein (Leone et al ., 1991 b) .

The oligomerization state of ai has been a matter of 0042-6822/91 $3 .00

Copynght cp) 1991 by Academic Press. Inc . Al, rights of repreduclion m any norm reserved . controversy . Since the al cell-binding function is manifested only by the oligomeric, but not the monomeric, form of a1 , it is important that the exact stoichiometry of the al protein be known . Based on the icosahedral distribution of al on the virion (Lee etal., 1981) and on earlier stoichiometric data (Smith at al., 1969) , it was initially suggested that two molecules of al were present on each vertex of the virion (Lee et al., 1981) . Subsequent sequence analysis was also compatible with a dimeric model of al (Bassel-Duby at at, 1985) . More recently, however, a number of studies including SDS-PAGE analysis of nondissociated protein, chemical crosslinking, sequence analysis, and electron microscopy have supported a tetrameric model for the al protein (Bassel-Duby at at, 1987 ; Fraser et at, 1990) . On the other hand, analysis of the N-terminal oligomerization domain by SDS-PAGE under nondissociating conditions suggests that it forms a three-stranded coiled coil .

In the present study, we present evidence that the al protein istrimeric . Specifically, mRNAs of full-length and various deletion mutants of the S1 gene were cotranslated in vitro to produce homo-oligomeric and hetero-oligomeric a1 complexes . After subjecting the resultant proteins to SDS-PAGE under nondissociating conditions, a total of four bands was generated, indicative of a protein composed of three monomeric units . Biophysical analyses of the oligomeric al protein and of the oligomeric N-and C-terminal tryptic fragments have provided additional evidence that these proteins are trimeric . We further demonstrate a possible reason for the large molecular weight (-200K) previously ascribed to al based on SDS-PAGE analysis under nondissociating or chemically cross-linked conditions .

The plasmid pG4T3 which contains the full-length T3 S1 gene (455 codons) cloned into pGEM-4Z (Promega) has been previously described (Leone et al., 1991a ; . Construction of a deletion mutant (d44) which lacked 44 codons at the 3'-terminus of the full-length S1 gene has also been described . The mutant dill with 142 amino acids (residues 224-365) deleted from the Ctermina'

" of -, ,ctc' by P_-,! ;,l A tion of pG4T3 with Bc/I (nucleotide 678), followed by complete digestion with Bglll (nucleotide 1104) and religation .

In vitro transcription-translation

The plasmids were linearized with Hindlll and used in in vitro transcription reactions to generate full-length or truncated S1 mRNAs . These mRNAs were translated either individually or in combination (ideally in equimolar amounts) in vitro in rabbit reticulocyte lysates according to the manufacturer's specifications (Promega) . Typically, the translation mixture contained 50-100 ng of RNA and 20 gCi of (35S]methionine in a total volume of 25 ul . Reactions were stopped after incubation for 45 min at 37° by the addition of 200,ul of phosphate-buffered saline (PBS), pH 7 .2. Protein sample buffer (5X) was then added (final 1 X concentration was 50 mMTris, pH 6 .8, 1% SDS, 2% f-mercaptoethanol, 10% glycerol, and 0 .01% bromphenol blue) and the samples were either boiled for 5 min (dissociating condition) or incubated for 30 min at 37° (nondissociating condition) prior to SDS-PAGE .

Discontinuous SDS-PAGE was performed using the protocol of Laemmli (1970) . For the hetero-oligomer experiments, 7 .5% polyacrylamide gels were used for running the nondissociated proteins and 10% gels were used for running the dissociated proteins . Gels containing (35Sjmethionine-labeled proteins were fixed and treated with DMSO-PPO, dried undervacuum, and exposed to XAR-5 film at -70°.

The protein bands from hetero-oligomer experiments were located by exposing the dried, unfixed gel to X-ray film . The bands were then excised and rehydrated with Laemmli running buffer, and the proteins electroeluted using Laemmli running buffer and ISCO electroelution cups . Protein sample buffer (5X) was added to electroeluted proteins, and the samples were boiled for 5 min prior to SDS-PAGE .

Samples of purified intact and trypsinized al (see below) were incubated in the presence of protein sample buffer (1X concentration as above) at 4 0 , 10°, 15 0 , 22°, and 37° for 30 min or boiled for 5 min prior to SDS-10% PAGE . These gels were electrophoresed at 4° in prechilled (4°) Laemmli running buffer, followed by fixing and staining by the Coomassie method of Zehr et al. (1989) .

Densitometric analyses of autoradiograms and Coomassie-stained gels were carried out using a Soft laser scanning densitometer (Biomed Instruments, Inc .) and Biomed laser software . Values for quantitation of autoradiographic bands from hetero-oligomer experiments were normalized to accommodate for the numbers of methionines present in the intact and truncated a1 proteins . Specifically, the dIll deletion mutant contains only five of the eight methionines present in the intact a1 protein and, as such, was normalized in the appropriate manner . Approximately 2 mg of purified a1, derived from a baculovirus expression vector system (R . Duncan, J . E . Strong, P . Gale, and P . W . K . Lee, unpublished data) , was dissolved in 1 ml of buffer A and applied to the CL-6B column . Eluted samples were analyzed by Bio-Rad protein assay and by SDS-PAGE .

Sephacryl 5-300 SF chromatography of a1 tryptic fragments A Sephacryl S-300 SF column (2 .5 X 95 cm) was prepared and equilibrated with buffer B [10 mM Tris buffer (pH 7 .5) and 0 .1 M NaCI] . Molecular weight standards of known Stokes radii [catalase (52 .2 A), aldolase (48 .1 A), bovine serum albumin (35 .5 A), ovalbumin (30 .5 A), chymotrypsinogen A (20 .9 A), and ribonuclease A (16 .4 A)] were prepared by dissolving 2 mg of each protein in 1 ml of buffer B and applied to the 5-300 column . Elution was carried out at 4° at a flow rate of 20 ml/hr. Eluted samples were collected and analyzed as above .

Trypsin-cleaved a1 was prepared by digesting 3 mg of purified a1 for 30 min at 37° with washed insoluble trypsin (Sigma) (3 .8 U) in 1 .2 ml PBS . After pelleting the trypsin beads, TLCK was added to the supernatant to a final concentration of 0 .2 mg/ml, and the supernatant was applied to the column .

Linear 12-ml 5-20% sucrose gradients in buffer A were used for sedimentation analysis . Approximately 1 mg of intact a1 or 2 mg of trypsin-cut a1 in 300 pl of REOVIRUS PROTEIN of IS A HOMOTRIMER 25 buffer A was applied to the gradients . Molecular weight standards of known sedimentation coefficients [catalase (11 .30 S) , aldolase (7 .35 S), bovine serum albumin (4 .30 S), ovalbumin (3 .66 S), and chymotrypsinogen A (2 .58 S)] were prepared by dissolving 1 mg of each protein in 300 µl of buffer A and applied to parallel gradients . For intact at, thesampteswere centrifuged at 38,000 rpm (Beckman SW41 rotor) for 24 hr at 4°. Trypsin-cut a1 and standards were centrifuged at 38,000 rpm for 25 hr . All samples were fractionated and analyzed by Bio-Rad protein assay . Fractions containing intact and trypsin-cut a1 proteins were subjected to SDS-PAGE to assess purity and, in the case of trypsin-cut samples, the gels were scanned with the Soft laser scanning densitometer (Biomed Instruments, Inc .) to determine the peak fraction .

Column filtration data are expressed in terms of Kim" a parameter which is defined as

where V°is the elution volume of solute ; Vo is the void volume of column ; and V, is the total bed volume of column . Siegel and Monty (1966) have demonstrated a linear relationship between elution volume (expressed in terms of K°.) and a theoretical value termed the Stokes radius for a variety of proteins . The relationship of the K,, to the Stokes radius is given by

where a is the Stokes radius, and a and S are constants related to the intrinsic properties of the gel matrix .

The Stokes radius of a protein can be measured by passing the protein through a column which has been previously calibrated with markers of known Stokes radii . After determining the sedimentation coefficient of the protein by sucrose gradient centrifugation with standards of known sedimentation values, accurate estimates of the molecular weight of the protein can be calculated using

where M is the molecular weight, a is the Stokes radius, s is the sedimentation coefficient, u is the partial specific volume (actual value used = 0 .732 cc/g), n is the viscosity of medium (value used = 1), p is the density of medium (value used = 1), and N is Avogadro's number.

(3)

Fia . 1 . Analysis of in vitro cotranslation products of full-length (FL) and 3'-truncated (d44) S1 mRNAs . Proteins were synthesized in the presence of [36S]methionine either individually or in combination as indicated (see Materials and Methods) . Synthesized proteins were mixed with protein sample buffer, incubated at 37° for 30 min, and analyzed on SDS-PAGE . The predicted compositions of the protein bands are indicated on the right where A represents the full-length product and B represents the product of the d44 transcript.

The frictional ratio (f/fo) of the protein can then be estimated using

To determine the oligomerization state of o1, we took advantage of the observation that both full-length and C-terminal truncated a1 proteins are capable of forming oligomers in an in vitro translation system and that these oligomeric forms can be resolved from each other by SDS-PAGE under nondissociating conditions . It was therefore reasoned that cotranslation of the full-length mRNA and a C-terminal truncated St mRNA should lead to the formation of a set of homo-oligomeric and hetero-oligomeric complexes as a result of random association of monomeric proteins. The oligomerization state can then be determined from the number of hetero-oligomeric species present on SDS-PAGE .

The results from one such experiment using the fulllength (FL) a1 and a truncated a1 with 44 amino acids deleted from the C-terminus (d44) are shown in Fig . 1 . Both FL and d44 were detected as single protein bands (homo-oligomers) when synthesized individually, with the d44 oligomer migrating faster than the FL oligomer . When synthesized together, four protein STRONG ET AL . bands were seen . Two of the bands migrated at the same positions as the individually synthesized FL and d44, whereas the middle two bands migrated with intermediate mobilities . The formation of four oligomeric protein species with different migration rates is consistent with the notion that a1 is composed of three subunits . Whereas the slowest and fastest migrating species represented homo-oligomers that contained only full-length (A3 in Fig. 1 ) or truncated proteins (B 3 in Fig .  1 ), the intermediate complexes contained either two full-length and one truncated (A 2B 1 in Fig. 1) or two truncated and one full-length protein (A 1 B2 in Fig . 1 ) (see below) .

The above experiment was repeated using FL and another deletion mutant (dill) lacking 142 amino acids (residues 224-365) from the C-terminal portion of at ( Fig . 2A) . Here again, cosynthesis of the full-length (A) and truncated proteins (C) produced two hetero-oligomeric complexes (A 2C1 , and A 1C2). The full-length complex was found to exist as two differently migrating species (labeled FL and FL*) . The band labeled FL represented the most dominant conformation, while FL* represented an altered, misfolded, and more compact form of the full-length oligomer, also previously noted by Banerjea and Joklik (1990) . [A comparable but somewhat diffused FL* band was also detected in the gel used in Fig . 1 , but is not shown since it migrated considerably faster than d44 (B3)] .

A number of other combinations of deletion and fulllength transcripts were also attempted . In every case, only two hetero-oligomeric complexes, in addition to the two homo-oligomeric species, were formed (data not shown) . This data strongly suggest that the oligomeric proteins are composed of three monomeric units .

To confirm this reasoning, and to dispel any suggestion that any single band might be made up of more than one homo-oligomeric or hetero-oligomeric species, each of the four bands from Fig . 2A was excised from the gel, electroeluted, boiled in SDS-containing buffer, and subjected to SDS-PAGE (Fig . 2B) . The gels were then scanned and the molar ratio of the two proteins in each complexwas determined, taking into consideration the difference in methionine content between FL and dill . As expected, bands 1 and 4 contained almost exclusively FL and dill, respectively (the minor contaminations observed were quite typical for analysis of this nature) . The ratios of FL to dill in bands 2 and 3 were found to be 2 .0:1 and 1 :2 .2, respectively, in good agreement to what would be expected for a trimeric protein . Analysis of the excised FL* band revealed that it contained only the full-length o1 protein (data not shown). Fla . 2. (A) Analysis of in vitro cotranslation products of full-length (FL) and truncated (dill) S1 mRNAs . These proteins were synthesized and analyzed as described in the legend to Fig . 1 . The predicted compositions of the protein bands are indicated on the right where A and C represent the full-length and the dill products, respectively . The FLt band represents a misfolded form of the full-length o6gomer . (8) Individual bands from (A) (labeled band 1-4) were excised, rehydrated in Laemmli running buffer, and electroeluted . After boiling the eluted proteins in protein sample buffer for 5 min, the samples were subjected to SDS-PAGE and autoradiography . The composition of each band (i .e ., the ratio of FL :dlll polypeptides) was analyzed by laser scanning densitometry and the results are shown to the right of each lane . R represents purified [ 36 S]methionine-labeled reovirus type 3 .

In order to confirm the trimeric model and analyze o1 in its native state, the physical characteristics of the intact of protein were determined using column filtration and sucrose gradient sedimentation analysis . Upon exclusion chromatography through a Sepharose CL-68 column that was calibrated with marker proteins of known Stokes radii, the al oligomer eluted as a homogenous species with an estimated Stokes radius of 63 .3 A (Fig . 3A) . Sedimentation in sucrose gradients x 10 -13 sec (Fig . 38) . The molecular weight of the oligomeric protein was then calculated to be 132K (Table  1) according to the method of Siegel and Monty (1966) [see Eq .

(3)], making of (49K monomeric molecular weight) a trimer . The highly asymmetric nature of this protein is indicated by its high frictional ratio (flfa ) of 1 .88 (an f/fe ratio of 1 .5 is consistent with a prolate ellipsoid of axial ratio of about 8:1) and supported by the observation that its apparent molecular weight by column filtration (-440K) greatly exceeds the calculated molecular weight . It was previously shown that trypsin cuts al approximately in the middle (after Arg 245 ), generating a 26K N-terminal fragment and a 23K C-terminal fragment under dissociating conditions (Yeung et at, 1989 ; . The N-terminal fragment migrates as a trimer (-80K) on SDS-PAGE under nondissociating conditions , and the C-terminal fragment also exists as an oligomeric structure, albeit less stable than the N-terminal oligomeric fragment (Duncanetat, 1991) . It was therefore of interest to use the biophysical analysis described above to determine whether these two tryptic fragments, like intact al, are also trimeric .

It was predicted that since the N-terminal tryptic fragment possesses rod-like properties, it should behave anomalously on column filtration and sedimentation analysis, whereas the C-terminal tryptic fragment, being mostly globular in nature, should behave almost ideally on both parameters . Such predictions were found to be correct . The N-terminal fragment eluted off the S-300 column with a Stokes radius of 54 .1 A (Fig .   4A ), equivalent to a protein of greaterthan 230K molecular weight, whereas by sedimentation analysis the sedimentation coefficient of 3 .37 X 10 -13 sec (Fig . 4B ) predicted a protein of less than 40K . When these values were combined in Eq . (3) (see Materials and Methods), a calculated molecular weight of 77K was obtained, which corresponds to three times the molecular weight (26K) of the monomeric subunit (Table 1) . As expected, the N-terminal tryptic fragment is highly asymmetric, with an f/fe value of 1 .92 . The C-terminal tryptic fragment behaved almost ideally since its Stokes radius, determined by S-300 column filtration to be 36 .4 A (Fig . 4A) , and its sedimentation coefficient, determined by sucrose gradient centrifugation to be 4 .16 X 10 -13 sec (Fig . 4B ), both predicted a protein STRONG ET AL .

with a molecular weight similar to that of BSA (67K) . Not surprisingly, the molecular weight of the C-terminal tryptic fragment was calculated to be 64K using Eq .

(3), again indicative of a trimer (Table 1) . The globular nature of this fragment is reflected by its calculated frictional ratio of 1 .37, a value that is comparable to that for BSA (f/fe = 1 .35) .

Effects of preelectrophoresis incubation temperature on the migration rates of intact al and al tryptic fragments

The previous suggestion that native al is a tetramer was in part based on the observation that under nondissociating conditions, at migrates at a position corresponding to -200K or higher (Bassel-Duby et at, 1987 ; Banerjea and Joklik, 1990) . We subsequently found that when trypsin-cut al is subjected to the same treatment, the N-terminal fragment migrates as an oligomer (trimer), whereas the C-terminal fragment migrates as a monomer ; also see Fig . 5 ) . The differential effects of such treatment on the two al tryptic fragments, when extrapolated to the intact protein, would be the generation of a hydra-like structure, with the stalk representing the trimeric, rodshaped N-terminal half, and the tentacles representing the three dissociated and unfolded C-terminal strands (see Fig . 6 ) . It is not difficult to envision how such a structure could manifest a retarded mobility in polyacrylamide gels .

Our present finding that the C-terminal tryptic fragment in its native state is indeed oligomeric (trimeric) has led us to reason that perhaps, under conditions which allow this fragment to remain nondissociated, intact a1 would maintain its lollipop morphology and migrate at a rate more representative of its true molecular weight . This was indeed found to be the case (Fig .  5) . By simply reducing the preelectrophoresis incubation temperature from 37° to 22° or lower, we detected a dramatic shift in the migration rate of the C-terminal tryptic fragment from the 23K (monomer) to the 55-60K (trimeric) position . The trimeric status of the N-terminal fragment remained unchanged, although a partial and minor shift in migration rate was observed . The same down-shift of temperature was found to cause intact al to shift from the position of >200K (A form) to one slightly lower than where the reovirus a proteins (-I 50K) migrated (B form), the expected position for a trimeric at protein . The absolute correlation between stabilization of the trimeric C-terminal tryptic fragment and enhanced mobility of the intact al protein has led us to safely conclude that the at A form ("hydra") migrates anomalously, resulting in the previous overestimation of its molecular weight . The somewhat hetero- FG. 4 . Biophysical analysis of the tryptic fragments of al . (A) Sephacryl S-300 column exclusion chromatography of at tryptic fragments . Approximately 3 mg of at was digested with 3 .8 U of insoluble trypsin for 30 min at 37°. Trypsin-cut at was then applied to the S-300 column previously calibrated with marker proteins . The proteins were eluted with equilibration buffer into 1-ml fractions and analyzed by Bio-Rad protein assay and SDS-PAGE (see Materials and Methods) . A calibration curve of the column was obtained by plotting (--log K a,)"2 vs Stokes radii of the markers and used to determine the Stokes radii of the N-and C-terminal al tryptic fragments . (B) Sucrose gradient sedimentation ansysis of the tryptic fragments of al . Approximately 1 mg of trypsin-cut of or marker proteins was applied to 5-20% linear sucrose gradients and centrifuged at 38,000 rpm (Beckman SW41 rotor) for 25 hr at 4°. Gradient fractions were analyzed by SDS-PAGE and laser scanning densitornetry . The sedimentation coefficients of the N-and C-terminal al tryptic fragments were determined from the calibration curve obtained by plotting the sedimentation coefficients of marker proteins vs peak fraction number .

geneous nature of the al B form is believed to reflect the partial and minor mobility shift of the N-terminal trimeric fragment upon temperature down-shift .

The oligomerization status of the reovirus cell attachment protein at has been controversial . It was initially thought that protein al, which is located at the 12 vertices of the viral icosahedron, exists as a dieter (Lee et al ., 1981 ; Bassel-Duby et al., 1985 ; Yeung et al., 1987) . Subsequent studies based on migration rate in FIG . 5 . Effects of preelectrophoresis incubation temperature on intact al and al tryptic fragments . Purified a1 and trypsin-cut al were either boiled for 5 min (B) or incubated at 37 0 , 22°, 15°, 10°, and 4° for 30 min in SDS-containing protein sample buffer . SDS-PAGE was performed at 4°, followed by fixing and Coomassie staining . R represents purified reovirus type 3 under dissociating conditions . gels, chemical cross-linking, electron microscopy, and sequence analysis have led to the suggestion that o1 is a tetramer (Bassel-Duby et at, 1987 ; Banerjea et al., 1988 ; Banerjea and Joklik, 1990 ; Fraser at al., 1990) . Recently, we demonstrated that the N-terminal tryptic fragment of a1 migrates as a trimer on SDS-PAGE under nondissociating conditions and that the N-terminal heptad repeat region possesses an intrinsic trimerization function . In the present study, we have used two independent approaches, one biochemical and the other biophysical, to demonstrate that at is trimeric and provide a rationale for the anomalous behavior of the intact protein on SDS-PAGE (under certain nondissociating conditions and not under others) which has in part led to the previous suggestion that it is a higher order oligomer .

The cosynthesis of full-length and C-terminal truncated al proteins has allowed us to identify hetero-oligomers, the number of which reflects the oligomerization status of al . If a1 were a dimer, for example, only one hetero-oligomer would be formed and, together with the two homo-oligomers, a total of three bands would be observed on SDS-PAGE . Similarly, if al were a tetramer, five complexes would be produced . The fact that four oligomeric species are consistently found for any of the combinations of full-length and truncated proteins tested is a strong indication that a1 is composed of three monomeric subunits . It deserves mention in this regard that if the assortment is indeed random, the distribution of the four oligomeric species (A 3 , 0 .8 N-fragment Fie . 6 . Schematic diagram showing the configuration and oligomerization state of intact of and ei tryptic fragments at different temperatures (in SDS-containing protein sample buffer) . The transition from the "lollipop" to "hydra" configuration for intact al is correlated with the dissociation of the trimenc C-terminal tryptic fragment (the globular head) into its monomeric subunits .

A2 B" A,B2 , and B 3 in Fig . 1 ) would be in the ratio of 1 :3:3:1 (assuming equimolar amounts of A and B were synthesized) . This has not been found to be the case . Rather, we consistently observed a preference for homotrimer over heterotrimer formation . Although we have not completely ruled out the possibility that, for whatever reason, homotrimers are assembled more efficiently or are more stable than heterotrimers, recent studies (unpublished) suggest that trimerization may not occur in solution and most likely involves a scaffold in close proximity to the site of protein synthesis . Newly synthesized monomeric a1 would be "recruited" onto this scaffold where they are assembled into trimers and subsequently released .

Results from the biophysical analysis of both intact and trypsin-cut a1 confirm the trimer model as well as the asymmetric nature of this protein . Based on sequence analysis, Bassel-Duby et al. (1985) first predicted that the al protein consists of a long fibrous tail topped with a globular head . Such predictions have subsequently been verified by electron microscopy (Banerjea et at, 1988 ; Furlong et at, 1988 ; Fraser et at, 1990) . The asymmetry of al is further indicated in the present study from its high frictional ratio and manifested by its abnormal migration on density gradient sedimentation and column filtration . We have also confirmed that the asymmetric shape is confined almost entirely to the N-terminal half of the protein since the C-terminal tryptic fragment behaved almost ideally by both parameters .

The need for the two halves of a1 to assume unique structural characteristics can readily be seen when one considers their distinct functional roles . The N-terminal one-third of a1, which is highly a-helical and contains a long heptad repeat of hydrophobic residues, possesses intrinsic oligomerization function (Banerjea and Joklik, 1990 ; Leone at at, 1991 a) and, as such, confers oligomer stability to the intact al protein . These observations suggest that the heptad repeat region is likely where trimerization of the intact al protein initiates and that trimerization of this region results in the formation of a stable extension which presumably facilitates access of the C-terminal region to the cellular receptor . More importantly, trimerization of the heptad repeat appears to be crucial for the generation of a competent virion-anchorage domain which likely resides in the N-terminal hydrophobic tail (Leone et al., 1991b) , as well as the subsequent trimerization and properfolding of the C-terminal globular head (G . Leone and P . W . K. Lee, unpublished data) . The significance of trimerization in al function is also indicated by our observation that monomeric a1 synthesized in vitro is nonfunctional (Leone et al ., 1991c) and that when expressed as a truncated protein (i.e., lacking the N-terminal oligomerization domain), the Cterminal half of a1 manifests a drastically reduced cellbinding activity (Nagata et al., 1987) . However, once formed, the C-terminal trimeric head is a compact, protease-resistant structure which is stabilized by intersubunit interactions and retains full cell-binding function even when separated from the Nterminal half of the protein (Yeung et at, 1989) . The trimeric nature of the C-terminal globular head also raises the question as to the stoichiometry of the alreceptor interaction . At present, it is not known whether each subunit possesses its own receptorbinding site or whether the three subunits contribute to a single receptor-recognition domain.

The difference in oligomer stability between the Nand the C-terminal halves of o1 is accountable for the anomalous behavior of the intact al protein in polyacrylamide gels . We have found that under conditions previously used to detect al oligomers, the C-terminal globular head dissociates into separate subunits, whereas the N-terminal fibrous portion maintains its trimeric state . The resulting structure would therefore resemble a hydra (see Fig . 6 ) whose mobility in polyacrylamide gels should be drastically retarded . However, under less stringent conditions where oligomer integrity of both the Nand the C-terminal portions is maintained, al would assume a more native structure, possibly resembling a lollipop, and should migrate at a position more representative of its true molecular weight. This has indeed been found to be the case : the "hydra" structure migrates at >200K, whereas the "lollipop" structure migrates at ^-150K, the theoretical position for a trimeric al . The dependence of mobility (in polyacrylamide gels) on al structure can also be used to explain the chemical cross-linking data previously presented as partial evidence for a tetrameric model of al (Basset-Duby et al., 1987) . One can envision how intersubunit cross-linking of the o1 protein along the N-terminal fiber, but not between the C-terminal subunits, could produce a protein which would very much resemble a hydra under dissociating conditions . Such a protein would have a retarded mobility and would therefore resemble a very large protein .

However, if the C-terminal subunits were additionally interhnked, the same protein would be more compact and therefore migrate faster . Thus, the number of bands and the apparent migration rates of these crosslinked proteins would not necessarily be a function of their actual oligomeric status, but rather a function of the conditions of cross-linking .

In consideration of our conclusion that the reovirus at protein is a trimer, it is noteworthy that a number of other viral cell attachment proteins such as the influenza hemagglutinin Wiley et al., 1981) , the VSV G protein (Doms et at, 1987 ; Kreis and Lodish, 1986) , the adenovirus fiber (Van Oostrum et al., 1987 ; Devaux et al., 1990) , the envelope glycoprotein (gpl 20) of the human immunodeficiency virus (Weiss et al., 1990) , and the coronavirus spike protein (Delmas and Laude, 1990) all appear to be trimeric . It remains to be seen whether the trimeric state reflects any unifying concepts in the virus attachment process .

Finally, it is interesting to note that Smith et al . (1969) 

Studies on reovirus receptors of L cells : Virus binding characteristics and comparison with reovirus receptors of erythrocytes Virology

High-level synthesis of biologically active reovirus protein a 1 in a mammalian expression vector system

Reovirus protein al translated in vitro, as wait as truncated derivatives of it that lack up to two-thirds of its C-terminal portion, exists as two major •etramenc molecular species that differ in electrophoretic mobility Virology

Sequence of reovirus hemagglutinin predicts a coiled-coil structure

Evidence that the a 1 protein of reovirus serotype 3 is a multimer . 1. Virol

Sequence of the Sl genes of the three serotypes of reovirus

Assembly of coronavirus spike protein into trimers and its role in epiope expression

Structure of adenovirus fibre . I . Analysis of crystals of fibre from adenovirus serotypes 2 and 5 by electron microscopy and x-ray crystallography

Role for adenosine triphosphate in regulating the assembly and transport of vesicular stomatitis virus G protein tnmers

Identification of conserved domains in the cell attachment proteins of the three serotypes of reovirus

Conformational and functional analysis of the C-terminal globular head of the reovlrus cell attachment protein

Molecular structure of the cell attachment protein of reovirus . Correlation of computer-processed electron micrographs with sequence-based predictions

01 protein of mammalian reoviruses extends from the surfaces of viral particles

Oligomenzation is essential for transport of vesicular stomatitis viral glycoproteins to the cell surface

Cleavage of structural proteins during the assembly of the head of bactenophage T4

Protein al is the reovirus cell attachment protein

a) . The N-terminal heptad repeat region of reovirus cell attachment protein o1 is responsible for of oligomer stability and possesses intrinsic oligomerlzation function Virology

b) The incorporation of reovirus cell attachment protein al into virions requires the N-terminal hydrophobic tail and the adjacent heptad repeat region

Trimerization of the reovirus cell attachment protein (al) induces conformational changes in n1 necessary for its cell binding function

The N-terminal quarter of reovirus cell attachment protein of possesses intrinsic virion-anchoring function

Molecular cloning and sequencing of the reovirus (serotype 3) S1 gene which encodes the viral cell attachment protein o1

Analysis of functional domains on reovirus cell attachment protein of using cloned S1 gene deletion mutants

Structure of the reovlrus cell-attachment protein : A model forthe domain organization of al

Determination of molecular weights and frictional ratios of proteins in impure systems by use of gel filtration and density gradient centrifugation : Application to crude preparations of sulfite and hydroxylamine reductases . Brochim

Polypeptide components of virions, top component and cores of reovirus type 3

The structure of the adenovirus capsid . Ill . Hexon packing determined from electron micrographs of capsid fragments

Identification of the gene coding for the hemagglutinin of reovirus

Ollgomenc organization of gpl20 on infectious human immunodeficiency virus type 1 particles

Structural Identification of the antibody-binding sites of Hong Kong influenza hemagglutination and their Involvement in antigenic variation

Structure of the hemagglutination membrane glycoprotein of influenza virus at 3A resolution

Purification and characterization of the reovirus cell attachment protein al

The cell attachment proteins of type 1 and type 3 reovirus are differentially susceptible to trypsin and chymotrypsin

A one step, low background Coomassie staining procedure for polyacrylamide gels