Cell, Vol.

14. 345-353,

June

1978.

Copyright

0 1978 by MIT

Complete Sequences of the Ribosome Recognition Sites in Vesicular Stomatitis Virus mRNAs: Recognition by the 40s and 80s Complexes John K. Rose Department of Biology and Center for Cancer Research Massachusetts Institute of Technology Cambridge, Massachusetts 02139

Nucleotide sequences of the ribosome-protected translation initiation sites from the vesicular stomatitis virus (VSV) M and L protein mRNAs have been determined, completing the sequences of the sites from all the VSV mRNAs. A low level of protection at two internal AUG-containing sites in the N mRNA is also described. Small homologies are evident among some of the sites, but there are no obvious features common to all the sites other than a single AUG codon. In contrast, a large homology between the VSV M mRNA site and the alfalfa mosaic virus coat mRNA site (Koper-Zwarthoff et al., 1977) is noted. This homology suggests the existence of a common ancestral gene for these two apparently unrelated viruses. For each VSV mRNA species, the smallest sites protected in either the 40s or 80s initiation complexes are identical. These sites always contained the initiation codon, but only contained the capped 5’ end in those mRNAs having the 5’ end near the initiation site. If 40s ribosomes bind to the capped 5’ end, either they do not protect it from nuclease digestion or the protection is only transitory in some VSV mRNAs. Consideration of the structures of the ribosome binding sites suggests that the differential effects of hypertonic shock on translation (Nuss and Koch, 1976) may be related to the distance between the 5’ end of the mRNA and the initiation codon.

al., 1978). Another distinguishing feature of most eucaryotic mRNAs is the presence of a methylated, capped 5’ terminus. This structure is known to increase the rate of binding of mRNA to ribosomes and the efficiency of mRNA translation (Shatkin, 1976), indicating some involvement of the 5’ end of the mRNA in ribosome recognition. Vesicular stomatitis virus (VSV) directs the synthesis of five capped mRNAs that are complementary to the single RNA strand present within the virion, and the single protein encoded by each mRNA species has been identified (Banerjee, Abraham and Colonno, 1977). It is also known that all the VSV mRNA species have equal translation efficiencies under normal conditions in vivo, although the individual mRNA species are synthesized at different rates (Villarreal, Breindl and Holland, 1976). Such detailed characterization of the system makes it especially useful for the study of mRNA structure and function. An earlier study (Rose, 1977) of ribosome binding sites from VSV mRNAs described isolation of sites protected by the 80s initiation complex, as well as the determination of the nucleotide sequences of the major protected sites from the viral mRNAs encoding the N, NS and G proteins. Completion of the sequencing of the VSV ribosome binding sites for the two additional mRNAs encoding the M and L proteins is described here. To determine which part of the VSV mRNA sequence is recognized in the 40s initiation complex, the sequences protected in this complex were also examined. Ribosome recognition of the proper translation initiation region in procaryotic mRNAs occurs first in a 30s ribosome-mRNA complex that contains initiation factors and initiator tRNA (see Lewin, 1974). The results obtained indicate that the eucaryotic 40s complex is analogous to the procaryotic 30s complex in its positioning at the initiation site.

Introduction

Results

The nucleotide sequences of translation initiation sites from several eucaryotic mRNAs have been determined within the past three years (see review in Hagenbuchle et al., 1978). A major aim of these studies has been the determination of what features of eucaryotic mRNA govern the selection of a specific site for initiation of translation. Similar studies in procaryotic systems have defined at least some of the features governing initiation site selection (Steitz and Jakes, 1975). In contrast to procaryotic mRNAs, eucaryotic mRNAs examined so far appear to have only single functional translation initiation sites. In those mRNAs which have been sequenced, this site occurs between 9 and 55 nucleotides from the 5’ terminus (Hagenbuchle et

Nucleotide Sequences of Ribosome Protected Sites from the M and L mRNAs The basic procedures used for isolation of sites in VSV mRNA protected by an 80s initiation complex were described previously (Rose, 1977) and are summarized briefly below. VSV mRNA labeled with 32P was incubated in a reticulocyte cell-free translation system in the presence of anisomycin, an inhibitor which allows the formation of a translation initiation complex, but prevents elongation (Lodish, 1971). The mRNA-ribosome complex was then digested with ribonuclease under conditions which left small mRNA fragments protected by the bound ribosomes. These bound fragments were then purified away from the digested RNA by su-

Summary

Cell 346

ii-

8% kcrylomide, Figure 1. Autoradiogram Complexes

of Two-Dimensional

Gel Electrophoretic

Separations

pH 3.5 of mRNA

Fragments

Protected

by 60s

and 405 Initiation

RNA

recovered from the ribosome-protected peaks was fractionated on separate two-dimensional gels: (A) fragments from the 60s and (B) fragments from the 40s complex. The two gels were run the same distance so that they could be compared. The lengths given at the right of the corresponding diagrams A’ and B’ below are calculated assuming a linear relationship between the distance migrated and the logarithm of the length in nucleotides using oligonucleotides of known sequence as standards. Spots with identical shading originate from the same initiator region. The origin of the second dimension is at the bottom of each photograph. Exposure was for 6 hr on Kodak NS film. Complex

crose gradient sedimentation. Unfractionated VSV mRNA was used in the binding reaction, followed by complete separation of the protected sites by two-dimensional gel electrophoresis. An autoradiogram of a typical pattern of the 80s sites obtained from total VSV mRNA is shown in Figure 1A. To determine the specific mRNA giving rise to each spot, the procedure was repeated for the individual VSV mRNA species purified by gel electrophoresis prior to ribosome binding. Table1 gives the mRNA assignments and quantitations of the individual spots. Because in some cases only partial protection from nuclease digestion occurred to one or both sides of the smallest protected fragment, a single mRNA often yielded several sizes of pro-

tected sites with partially overlapping sequences. Such related protected sites are grouped in Table 1 under the specific mRNA. The order of transcription of the VSV genes is N-NS-M-G-L (Ball and White, 1976), and the relative molar amounts of the mRNAs synthesized decrease in this order (Villarreal et al., 1976; J. Rose, unpublished data). The parallel decrease in amounts of the ribosome binding sites (Table 1) reflects only the amounts of mRNAs present. Sequence analysis of the sites from M and L mRNAs followed standard procedures (Barrell, 1971; Squires et al., 1976). Following elution from the two-dimensional gel, a portion of each site was digested with RNAase Tl (guanine-specific) or

Ribosome

Binding

Sites

in VSV mRNA

347

Table

1. Quantitation

of Protected

Sites

80s Complex mRNA N

Spot Number 1 3 4

40s Complex cm

5’ Cap

2426 4125 2252

+ + -

NS

2 5

2077 3736

M

9 10

1991 1857

G

L N (Internal)

6

Spot Number

cm

5’ Cap

1 3 4

1389 2700 1751

+ + -

9’8 15

3110 502

+ -

2 5

1390 1522

+ +

9’” 10 19 14v

1280 1067 1360

-

6 12

206 324

-

16

537

+

+ +

1533

11

720

+

11

680

+

7

1200

-

7

225

-

8

1509

-

8

327

1P 18e

148 174

Unassigned

a Only lo-20% of the sequences sites. b Spots 13 and 14 also contained c Spots 17 and 18 were not seen

in spot 9’ (40s site) were low levels reproducibly

derived

from

the M RNA site which

(120%) of sequences from the major and contained too little radioactivity

RNAase A (pyrimidine-specific), and the products of these digestions were separated by two-dimensional fingerprinting. Appropriate secondary digestions of these products were performed with RNA labeled in vivo with 32P-phosphate and RNA labeled in vitro with LI-~~P-UTP, CX-~~P-CTP or &*PGTP, so that nearest-neighbor transfer data could be obtained when needed. First consider the site from the M protein mRNA that was previously identified (Rose, 1977) as spots 9 and 10 (Figure IA). The fingerprint of a complete RNAase Tl digest of spot 10 is shown in Figure 2A (uniform label) with the sequences of the oligonucleotides indicated. Spot 9 contains all the sequences in spot 10 except the terminal oligonucleotide AUUCUCG. Table 2 gives the products of the secondary digestions of each oligonucleotide. The data were sufficient to define the sequences of all but the largest (22 nucleotide) Tl oligonucleotide, for which a single ambiguity remained (Table 2). This ambiguity was resolved by analysis of a set of partial products generated by venom phosphodiesterase digestion of the oligonucleotide. This technique has been applied previously to analysis of RNA sequence (see Rensing and Schoenmakers, 1973) and allows one to “read” an RNA sequence directly from the autoradiogram of a two-dimensional separation of the complete set of partial products because the sequential loss of nucleo-

occupies

this position

exclusively

in the 80s

N mRNA binding site. for fingerprint analysis.

tides generates mobility shifts characteristic of each nucleotide. Analysis of several oligonucleotides of known sequence indicated that the “trails” were most easily interpreted when polyethyleneimine (PEI) cellulose was substituted for DEAEcellulose in the homochromatography dimension. An autoradiogram of a separation of the venom partial products obtained from oligonucleotide M5 (c@‘P-UTP label) is shown in Figure 3. Interpretation of the sequence was straightforward because the sequence was already known except near the 3’ end, (AUUC,AUC)AUG. It is not possible to read the sequence near the 5’ end of an oligonucleotide because the absence of 5’ and 3’ terminal phosphates results in mobility near the RNA front for products ~4-5 nucleotides long. A similar pattern was obtained using a uniformly labeled oligonucleotide, and the sequence assignments substantiated by reanalysis of RNAase A secondary digestion products of each venom partial product. The product to the right of the trail (shown as x in Figure 3) was observed repeatedly and apparently resulted from a specific internal nick during digestion. This product was completely resistant to venom phosphodiesterase digestion, suggesting that it had a 3’-phosphate (perhaps 2’, 3’-cyclic). Ordering of the Tl oligonucleotides to give the sequence shown in Figure 4 was as follows. The largest RNAase A product derived from spot 10,

Cell 346

.

pH 3.5 Electrophoresis Cellulose Acetate Figure 2. Autoradiograms Oligonucleotides Derived Protected Sites

of Two-Dimensional Fingerprints of by RNAase Tl Digestion of Ribosome-

The first dimension was electrophoresis on cellulose acetate (pH 3.5) in 7 M urea followed by homochromatography on polyethyleneimine (PEl)-cellulose thin-layer plates (BarrelI, 1971; Squires et al., 1976). (A) spot 10 (M mANA site, uniform 32P label); (B) spot 11 (L mRNA site, uniform 32P label); (C) spot 7 (internal N mRNA site, &*P-GTP label); (D) spot 9 (internal N mRNA site, a-32PGTP label).

Table

2. Analysis

of Tl Oligonucleotides

RNAase AG

M2

AAG

M3

AU, 2U, 2C. G

M4

4U. 2C. AAAG

M5

AU(U)/ AU(G),

Ll

G

L2

AAG

L3

L5

U. 2C. AC.Gb by a-=P-UTP) C,AAU,C.AU,G -AU,U.U,G

L6

m’GpppAm-AC,AG

L4

the M mRNA

RNAase

A

Ml

from

and L mRNA

UT

(AAAG,AAG)AU(U), establishes the order 5’-M4-M2M3. M3 must be the 3’ end of the site because it has the 3’ nearest-neighbor G, and no free Gp is liberated in an RNAase Tl digest of the site. The RNAase A product GAGU(U) establishes the order 5’-ME&Ml, which must be at the 5’ end of the site. Next, consider the site derived from the L mRNA, spot 11 in Figure 1A. Because the L mRNA is synthesized at very low levels in vivo and in vitro, this site was not detected in the previous analysis (Rose, 1977). The RNAase Tl fingerprint of this site (uniform 32P-label) is shown in Figure 28, with the sequences of the oligonucleotides indicated. The assignment of this sequence to the L mRNA is based upon the following results. The VSV mRNAs were fractionated by electrophoresis on a formamide-polyacrylamide gel (Rose and Knipe, 1975), and each isolated mRNA band was used for ribosome protection. The largest RNA band (which contains the L mRNA) was the only band which gave a site containing the oligonucleotides shown in Figure 2B. In this fingerprint, there were also oligonucleotides from the NS and M sequences, but this result was not unexpected since aggregates of the smaller mRNAs often run with the L mRNA (J. K. Rose, unpublished results). Table 2 shows the derivation of the sequences of the individual Tl oligonucleotides from the L site. Two products of RNAase A digestion of spot 11,

Protected

Sites

CMCT/RNAase

Sequence Deduced Nearest-Neighbor

AC

and 3’

AWJ) A=(A) A, (2C, 3U)G

2AU(C), -4U, -7C

UCA(U), UUA(U), UCCA(U), UCCCAA(U), UUCA(U). UG

(AU,U)C,

AUUCUCG(G)

UC, G

WC(C), C(U) UUAAAG, CUUAAAG

UUCCUUAAAG(A)

AMC), A’JW), AUUC(A). UUAUC(C), 2-3C, AUG

UUAUCCCAAUCC(AUUC, AUC)AUG(A)

G(A) AAG(U) (Unlabeled)

(U,C,C)A,CG

UCCACG(A)

CAA,(U,C)A,UG

CAAUCAUG(G)

A&G

AUUUG(A)

B Parentheses indicate the nearest-neighbor of a particular oligonucleotide. b Lines above sequences indicate a sequence order determined from nearest-neighbor transfer data. c Sequences given were determined from nearest-neighbor transfer data using Q-~~P-UTP, w~~P-CTP 52P-labeled RNA synthesized in vivo (data not given).

m’GpppAm-ACAG(C)

and

c@~P-GTP

labels,

and from

Ribosome 349

Binding

Sites in VSV mRNA

pH 3.5 Electrophoresis,Cellulose Acetote Figure Venom

3. Autoradiogram Partial Digestion

of a Two-Dimensional Products of Oligonucleotide

Separation M5

of

The partial digestion products of oligonucleotide M5 were separated in the first dimension by electrophoresis (pH 3.5) on cellulose acetate in 7 M urea followed by homochromatography, as in Figure 2. Approximately 4000 input cpm were used, and exposure was for 36 hr on Kodak NS film.

GGAAGU(C) and GAU(U), establish overlaps which define the complete the L site shown in Figure 4.

unambiguous sequence of

Isolation of 40s Ribosome-Protected Sites To obtain VSV mRNA fragments protected by the 40s initiation complex, an inhibitor of 60s subunit addition, GMP-P(CH,)P, was included in the binding reaction in place of GTP. In the presence of this inhibitor, VSV mRNA was found distributed approximately equally between the 40s complex and the 80s complex, indicating that the action of the inhibitor was incomplete, a situation noted previously (Levin, Kyner and Acs, 1973). When the mRNA-ribosome complexes were treated with RNAase Tl prior to sucrose gradient sedimentation, about 2% of the total Cu-32P-GTP-labeled VSV mRNA was distributed about equally between the 40s and 80s peaks. The ribosome-protected RNA fragments purified from these peaks were then fractionated by two-dimensional gel electrophoresis. The autoradiograms of the gel electrophoretic separations of the 80s and 40s sites obtained in the same experiment are shown in Figures 1A and IB. Inspection of the autoradiograms indicates similar patterns of protected sites in both cases, except that some protection of larger sequences is evident in the 40s material. The analysis of the 40s

and 80s sites was as follows. A portion of the material from each spot was digested with RNAase Tl and fingerprinted as in Figure 2. Another portion was analyzed directly for the presence of the 5’ cap after complete digestion with RNAases Tl, T2 and A (Rose, 1975). From the knowledge of the fingerprints and sequences of the 80s sites, it was possible to assign the spots to specific mRNAs. Corroboration of the assignments was obtained by using purified VSV mRNAs to form the 40s complexes and then analyzing these 40s protected sites on the two-dimensional gel system. Approximate comparison of the amounts of the 40s and 80s sites can be made from the data in Table 1, which were obtained in the experiment in Figure 1. Exact comparison requires knowledge of the number of G residues in each site, and this is only known for those sites that have been sequenced. In all cases, the 40s sites corresponding in position and number to 80s sites (Figure 1) contained the same sequences, although in one 40s site, spot 9*, additional sequences from another site constituted the majority of the radioactivity present (Table 1). Note also that traces of large protected sites are present among the 80s sites. These did not all correspond in position to the large 40s sites and were present at levels insufficient for further analysis. The data on the extent of 40s and 80s ribosome protection obtained from analysis of the sites are summarized in Figure 5. Because RNAase Tl was used to generate the sites, only the positions of the target residues (G) are indicated. In cases where the sequence is not known, these positions have been estimated from the sizes of the sites. The diagram indicates the smallest protected fragments isolated from each mRNA (henceforth called the minimal sites) by unbroken lines. All larger protected fragments contained this minimal site sequence. The minimal sites always contained the initiation codon, but only in the two cases (NS and L mRNAs) where the 5’ end is close to the initiation codon did they contain the 5’ end. It should also be noted that the minimal sites are always identical for 40s and 80s protection within the same mRNA. Variation in the sizes of the minimal sites among the mRNAs may result only from the absence of RNAase Tl cleavage sites in mRNAs giving larger sites. The extent of partial protection beyond the minimal site is indicated by broken lines in Figure 5 and clearly depends upon the specific mRNA examined. First, consider the L and NS mRNAs which are the only mRNAs with the 5’ end in the minimal site. The L mRNA shows no protection beyond the minimal site, and the NS mRNA shows only a few nucleotides of protection beyond the minimal site. This partial protection is the same in the 40s and

Cell 350

M

G

(GlUUAUCCCAAUCCAUUC~AUc$ii$.AGUUCCUUAAAGAAGAUUCUCG~ YET SER SER

(GWUUCCUUGACACU~AAGUGCCUUUUGUACUUAG MET LIS

Figure 4. Complete Sequences from the VSV mRNAs

C”S

LE”

---L”S

LYS

ILE LE”

LEU

LE”

TIR

LEU

of the Ribosome-Protected

GL”

Sites

The sequences corresponding to the major ribosome protected sites from each of the indicated VSV mRNAs are shown. Regions of homology are indicated by dashed boxes, and amino acid sequences are predicted from the genetic code. The sequences of N, NS and G mRNA binding sites are from Rose (1977). In the previously published N sequence, the sequence AAAAU which occurs in the noncoding region was reported incorrectly as AAAU, and the sequence AAAAG which occurs at the 3’ end of the NS sequence was reported incorrectly as AAAG. The sequences are shown corrected here, and were verified both from venom partial sequence analysis as in Figure 3 and from reanalysis of the RNAase A-derived fragments using appropriate standards.

80s complexes. Next consider the N mRNA which has the 5’ end just outside the minimal site. In this RNA, partial protection occurs in both the 40s and 80s complexes, can extend to either or both sides of the minimal site, and does not always include the 5’ end, even in one very large 40s protected fragment (spot 15, Figure 16). The extent of partial protection of N mRNA has been quite variable in different experiments. This protection appears to be nonrandom because not all possible partial products of ribosome protection are observed. Finally, consider the M and G mRNAs in which the 5’ ends are far from the initiation site. In the M mRNA, the 40s complex gives partial protection of sites which are 8-10 nucleotides longer than those protected by the 80s complex. Because the 5’ end of the mRNA is never protected and the sequences of the larger sites have not been determined, it is not known whether this partial protection extends to the 5’ or 3’ sides. Protection of the G mRNA by 40s complexes gives two sites larger than the minimal site. The largest of these (spot 16, Figure 16) was about 76 nucleotides long and contained the 5’ end of the G mRNA, indicating a maximal distance of about 40 nucleotides between the 5’ end and the initiation codon. Protection of such a large site from G mRNA was observed only in the one experiment shown. In two other experiments, only spots 6 and 12 (Figure IB) were generated from the G mRNA.

Secondary AUG-Containing Ribosome Binding Sites Occur Two ribosome binding sites generated from total VSV mRNA were not assigned to specific VSV mRNAs in the previous study (Rose, 1977). Further analysis using purified VSV mRNAs showed that these sequences (spots 7 and 8, Figure 1) were generated from purified N mRNA, although at IO15% efficiency compared with the major N site (Table 1). In addition, a minor sequence in spot 8 originates either from the NS or M mRNAs. To determine whether these secondary sites contained initiation codons, VSV mRNA was synthesized with an w~*P-GTP label, and the sites were isolated as shown in Figure 1A. Fingerprints of Tl RNAase digests of these sites are shown in Figures 2C and 2D. Each site gave a distinct set of Tl oligonucleotides which are clearly different from each other and from the major sites in each of the five VSV mRNAs. Appropriate analysis of RNAase A digests of each Tl oligonucleotide gave the indicated sequences (Figure 2) for the 3’ termini of each oligonucleotide and showed that each site, like the major site from each mRNA, contained a single AUG initiation codon. It should also be noted that nearly quantitative binding of several large non-AUG-containing Tl oligonucleotides to the 40s subunit has been observed reproducibly. These oligonucleotides are not visible in Figure 1 B because they contain only a single pG label, but are prominent spots when uniformly labeled RNA is used. These oligonucleotides do not appear in the 80s complex, but can interfere with analysis of the 40s protected sequences. The nature of the interaction of these Tl oligonucleotides with the 40s subunit has not been studied further and may be an interaction which does not confer protection of the sequences. Discussion The sequences of the ribosome-protected translation initiation sites from VSV mRNAs (Figure 4) do not all share obvious common features. Limited homologies among three of the sites are indicated in Figure 4, but their significance is unknown. The only invariant position is a single A residue 3 nucleotides before the initiation codon, but this is not a common feature in other eucaryotic ribosome binding sites (Hagenbuchle et al., 1978). Thus a variety of initiation site sequences can result in the identical efficiencies of VSV mRNA translation in vivo (Villarreal et al., 1976). It is possible that the AUG codon nearest the 5’ end of the mRNA serves as the initiation site in all the VSV mRNAs. This situation appears to hold for other eucaryotic mRNAs as well (Hagenbuchle et

Ribosome

Binding

Sites

in VSV mRNA

351

al., 1978) if one considers only systems where there is direct evidence (ribosome binding for example) for a particular mRNA sequence serving as an initiation site. With regard to a possible role of the 3’ end of 18s ribosomal RNA in selection of the translation initiation site, Hagenbuchle et al. (1978) have noted a 5 nucleotide complementarity between a purinerich sequence in the first 20 nucleotides at the 3’ end of 18s ribosomal RNA and the noncoding region of the G mRNA (UCCUU). Related homologies are not seen in the other VSV mRNAs except minimally in the M mRNA, which has a 3 nucleotide complementarity (UCC). The significance of such limited homology cannot yet be determined. Ribosome Recognition and 5’ Cap Protection Because 5’ terminal 7-methylguanosine is known to increase the efficiency of eucaryotic mRNA translation and 40s ribosome binding (Shatkin, 1976), it has been suggested that 40s ribosomes would recognize 5’ cap structures on all eucaryotic mRNAs (Kozak and Shatkin, 1977a, 1977b). The results reported here show that the protection of VSV mRNAs by the 40s complex and the 80s complex occurs in the region surrounding the initiation codon, and only includes the 5’ end when it is situated close to the initiation codon. If the 40s complex does bind to and protect the 5’ cap structure on VSV mRNAs, this must occur only at an early transitory stage during 40s complex formation on some VSV mRNAs. Kozak and Shatkin (1977a, 1977b) have reported results on ribosome protection of reovirus mRNAs in the wheat germ system. In these experiments, they showed that the 40s complex always protects the 5’ end of the mRNA as well as the initiation codon, even when the initiation codon is as far as 32 nucleotides from the 5’ end. Protection of the 5’ end of the mRNA was not seen in the 80s complex in those mRNAs having the 5’ end distant from the initiation codon. It was suggested on the basis of these results that the initiation codon might not be recognized within the 40s complex, but might function later to position the 80s complex. The results reported here require a different conclusion-that the 40s complex and the 80s complex are positioned similarly on the mRNA, since the minimal sites protected in both complexes are identical and always contain the initiation codon, but do not always contain the 5’ end. In agreement with this conclusion, Lazarowitz and Robertson (1977) have observed inefficient protection of the 5’ end of at least one reovirus mRNA in a reticulocyte 40s initiation complex. They also report efficient protection of the 5’ end of the same mRNA in the wheat germ system. The explanation for the differences between the systems is not known.

lmtlatm

Figure 5. Extent tion Complexes

S,te Protectm

of VSV mRNA

by 405

Protection

and SOS Rlbosomes

by 40s

and SOS Initia-

The minimal site protected from RNAase Tl in each mRNA by the 40s complex or the SOS complex (as indicated) is shown by a solid line above the sequence. Dashed lines indicate the occurrence of some degree of partial protection in the region. Positions of all known G residues in each sequence are indicated, and their relative positions are noted by numbers beneath each line. In cases where the sequence is not known for a region in which partial protection occurs, the positions of G residues could be estimated from the sizes of the protected sites, and their approximate positions are given by the numbers in parentheses. Note that G residues probably also occur at some places which are not indicated and went unnoticed because cleavage at these positions was not detected.

Nuss and Koch (1976) have described experiments showing that VSV mRNAs fall into two classes with respect to their sensitivities of translation to hypertonic shock in infected cells. These experiments showed that translation of N, NS and L mRNAs was resistant to hypertonic initiation block compared with G and M mRNAs when infected cells were subjected to increasing hypertonic shock. The fact that the two sensitive mRNAs (M and G) have the 5’ end of the mRNA considerably further from the initiation site than the three resistant mRNAs (N, NS and L) suggests that the 5’ end to initiation site distance is related to the hypertonic shock effect. Inspection of the nucleotide sequences (Figure 4) reveals no other obvious structural features which might explain differential sensitivities. Hypertonic shock might inhibit a transition of the 405 complex from the 5’ end to the initiation site, if such a transition occurs. Hypertonic shock might also alter secondary structure in the untranslated regions. Substantiation of this model must come from further sequence information in other systems and more detailed information on the molecular consequences of hypertonic shock of cells. Protection of Secondary Sites Two sites other than the major site within mRNA are protected by ribosomes, although

the N at 6-

Cell 352

I

G-A-C-A-A5’ppp5’m’G VSV M protein

u U U-A-U-U-U-U-U-G5'ppp5'i'I Figure

6. Comparison

of the Nucleotide

alfalfa Sequences

mRNA

mosaic virus

MET

SER

SER

Leu

LYS

LYS

Ile

Leu

GLY

MET

SER

SER

Ser

Gin

LYS

LYS

Ala

GLY

RNA 4 (coat)

of the VSV M mRNA

and the AMV Coat mRNA

The sequences surrounding the AMV coat protein mRNA (Koper-Zwarthoff et al., 1977) initiation site and the VSV M protein site are shown with alignment by their single AUG codons. The region of perfect homology is shown by the dashed homologies are shown by lines between the sequences. Amino acids in common are shown in capital letters.

IO fold lower efficiencies than the major site. Partial sequence analysis showed that each site contained a single AUG initiation codon. It is improbable that these secondary sites are functional initiation sites in vivo because no proteins other than the five major VSV proteins have been identified, although very small proteins could have been overlooked. How ribosomes get to the internal sites is unclear. Use of degraded mRNA preparations (RNAase Tl partial digestion) in ribosome protection experiments did not give a higher yield of these sites, suggesting that the internal binding is not due to recognition of nicked molecules. In view of the recognition of internal binding sites, the assumption that the major protected sites are the major translation initiation sites requires confirmation from amino terminal sequences of the VSV proteins. These sequences are not yet available.

Heijtink, Houwing and Jaspars, 1977); both have long leader sequences preceding the initiation codon; and both encode proteins of 25,000-27,000 daltons (Knipe, Rose and Lodish, 1975; Van Beynum et al., 1977). Thus it seems probable that these genes have been derived from a common ancestral gene. The two proteins may, in fact, be involved in formation of a similar type of bacilliform shape in both viruses since VSV particles are bacilliform at least at one end. Further sequence analysis of RNAs encoding proteins of similar size and function may also reveal extensive sequence homologies among apparently unrelated viruses. It should be noted that conventional methods of sequence comparison, such as nucleic acid hybridization or RNA fingerprinting, would not be expected to reveal the level of homology seen here. Experimental

A Common Ancestral Gene for VSV M Protein and Alfalfa Mosaic Virus Coat Protein? Recent work (Koper-Zwarthoff et al., 1977) has established the complete nucleotide sequence of the first 74 5’ terminal residues of the RNA encoding the coat protein of alfalfa mosaic virus (AMV). AMV is a multicomponent bacilliform virus having four different positive-strand RNAs encapsidated in four separate bacilliform particles (Jaspars, 1974). Comparison of the AMV coat RNA sequences with that of the VSV M protein RNA reveals extensive homology. First, a 12 nucleotide sequence, C-A-UC-A-U-G-A-G-U-U-C, including the single initiation codon is identical in both RNAs (Figure 6). Such extensive homology is not likely to be random since such a sequence would be expected to occur only once in 1.7 x 10’ (4’*) nucleotides ordered randomly. Second, one half (17 out of 34) of the remaining positions are identical when the sequences are aligned by the initiation codons, and only one fourth would be expected at random. In addition to the sequence homologies, there are other structural similarities. Both mRNAs are about 800-900 nucleotides long (Rose and Knipe, 1975;

mRNA initiation box, and other

Procedures

Preparation of mRNA and Isolation of Protected Sites Preparation of 32P-labeled VSV mRNA was as described previously (Rose, 1977), except that mRNA synthesized in vitro was purified away from unincorporated triphosphate by Sephadex G50 chromatography after phenol extraction. This procedure was preferable to oligo(dT)-cellulose chromatography because it eliminated a preferential loss of the L mRNA. Ribosome binding was performed essentially as described previously (Rose and Lodish, 1976; Rose, 1977). Ribosome protection to obtain both 40s and 60s sites was as follows. Reaction mixtures (800 ~1) contained 500 ~1 of rabbit reticulocyte extract, 1 mM ATP, 0.4 mM GMP-P(Cl-&)P (Miles Laboratories), 5 mM dithiothreitol. 75 mM KCI, 1.25 mM Mg(OAck. 20 mM methionine, 25 pg of anisomycin and 5-50 pg of labeled mRNA at specific radioactivities ranging from lo8 to 5 x 10’ cpmlpg. Within the subsaturating range of RNA concentrations used, the pattern of protected sites obtained was unaffected. Anisomycin was a gift from Dr. Harvey Lodish. The mixture was incubated for 20 min at 30°C followed by a 10 min further incubation in the presence of 100 units per ml of RNAase Tl (Calbiochem). The mixture was then diluted into 2 ml of buffer (O’C) containing 50 mM NaCl, 20 mM MgCL,, 20 mM N-2-hydroxyethylpiperazine-N’-2 ethanesulfonic acid (HEPES) (pH 7.5). and layered on a 35 ml linear 15-50% (w/v) sucrose gradient in NaCI-MgCI,-HEPES buffer. Centrifugation was for 10 hr in a Beckman SW27 rotor at 26,000 rpm at 4°C. Fractionation of the protected sites by two-dimensional gel electrophoresis (De Wachter and Fiers, 1972) was as described previously (Rose, 1977).

Ribosome

Binding

Sites

in VSV mRNA

353

Nucleotide Sequence Analysis Standard procedures (Barrell. 1971; Squires et al., 1976; Rose, 1977) were used for the sequence analysis with the exception of the venom partial digests which were carried out as follows. The purified oligonucleotide (about 4000 cpm) and 20 pg carrier tRNA were incubated for 30 min at 3PC in 5 $1 of a solution containing 0.05 units of calf alkaline phosphatase (Grade I, BoehringerMannheim) in 50 mM Tris-HCI (pH 8.9) and IO mM MgQ. The sample was brought to 23°C and 0.003 units (3 +I) of venom phosphodiesterase (Boehringer Mannheim) were added. Venom phosphodiesterase was stored in the 50% glycerol solution in which it was supplied and diluted I:50 into the Tris-MgCI, buffer just prior to use. After the phosphodiesterase addition, 0.5 pl samples were withdrawn and frozen at 2 min and 5 min; 1.0 ~1 samples at 10 min and 15 min; and 2 I.LI samples at 20 min and 25 min. To freeze the samples, they were blown from a micropipette into a tube in dry ice. The pooled, frozen samples were thawed just long enough to mix in 1 ~1 of 0.1 M EDTA and were then kept frozen or lyophilized until electrophoresis. Two-dimensional fingerprinting of the venom partials was by electrophoresis (4000 V for 45 min) at pH 3.5 on a 57 cm x 3 cm cellulose acetate strip, followed by homochromatography on a 20 cm x 20 cm polyethyleneine (PEl)-cellulose thin-layer plate (Squires et al., 1976). Precise conditions to give an even distribution of venom partials may be determined for individual oligonucleotides by spotting time points of the venom digestion on PEI-cellulose thin layers and running them in the homochromatography dimension only. PEI-cellulose was preferable to DEAE-cellulose for the second dimension analysis because the spots were much tighter, and the shifts (G>A>C-U) on the PEI dimension make distinction between G and U or A and C easier than on DEAE-cellulose. Acknowledgments I am grateful to Dr. David Baltimore for his support, encouragement and valuable suggestions throughout the course of this project. I thank Drs. David Baltimore, Philip Sharp and James Flanegan for helpful suggestions in the preparation of the manuscript; Marilyn Smith for typing the manuscript: and Kahan Leong for excellent technical assistance. The work was supported by grants from the National Cancer Institute and the National Institute of Allergy and Infectious Diseases. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received

January

10. 1978;

revised

March

14,1978

References Barrell. B. G. (1971). In Procedures in Nucleic Acids Research, II, G. L. Cantoni and D. R. Davies, eds. (New York: Harper and Row), pp. 751-828. Ball, L. A. and White, 442-448.

C. N. (1978).

Banerjee, A. K., Abraham, Virol. 34, l-8. De Wachter.

Proc.

Nat. Acad.

G. and Colonno,

R. J. (1977).

R. and Fiers,

W. (1972).

Anal.

HagenbOchle. 0.. Santer. Cell 14, 551-563.

M.. Steitz.

J. A. and Mans,

Heijtink, chemistry, Jaspars, Knipe, 1011.

R. A., Houwing, in press. E. M. J. (1974). D., Rose,

C. and Jaspars, Adv.

Virus

J. K. and Lodish,

Res.

Sci. USA 73,

Biochem.

J. Gen.

49, 184-197. R. J. (1978).

E. M. J. (1973).

Bio-

19, 37-149.

H. F. (1975).

Koper-Zwarthoff, E. C., Lockard, R. E.. Bhandary. U. L. and Bol. J. F. (1977). Proc. in press.

J. Virol.

de Weerd. Nat. Acad.

75, 1004B.. RajSci. USA,

Kozak,

M. and Shatkin,

Kozak, 6908.

M. and

Lazarowitz, 7842-7849.

A. J. (1977a).

Shatkin,

S. G. and Robertson,

Levin. D. H.. Kyner, 6418-6425. Lewin, 8. (1974). 79-116. Lodish. Nuss,

H. F. (1971).

Rensing, Biochem.

Expression-l

G. (1976).

(New

U. F. E. and 33, 8-18.

J. K. (1975).

J. Biol. Chem.

J. K. (1977).

Proc.

Rose,

J. K. and Knipe,

Rose,

J. K. and Lodish, A. J. (1976).

J. Biol. Chem.252, J. Biol.

York:

Chem.

John

248,

Wiley),

pp.

J. G. G. (1973).

Eur.

J.

250, 8098-8104.

Nat. Acad. D. (1975).

6895-

17, 283-286.

Schoenmakers,

Rose,

252,

246, 7131-7138.

J. Virol.

Rose,

Shatkin,

G. (1973).

J. Biol. Chem.

D. and Koch,

J. Biol. Chem.

H. D. (1977).

D. and Acs,

Gene

J. Mol. Biol. 772, 75-96

A. J. (1977b).

Sci. USA 74, 3672-3676

J. Virol.

H. F. (1978).

15. 994-1003.

Nature

262, 32-37.

Cell 9, 845-653.

Squires, C.. Lee, F., Bertrand, K., Squires, C. L.. Bronson. and Yanofsky, C. (1978). J. Mol. Biol. 103, 351-381. Steitz, J. A. and Jakes, 4734-4738.

K. (1975).

Van Beynum, G. M. A., De Graaf, Bosch, L. (1977). Eur. J. Biochem. Villarreal, L. P., Breindl, try 15, 1663-1667.

Proc.

Nat. Acad.

J. M., Castel, 72, 63-78.

M. and Holland,

Sci.

A., Kraal,

J. J. (1976).

M. J. USA 72, B. and

Biochemis-