Proc. Nati Acad. Sci. USA Vol. 79, pp. 5235-5239, September 1982 Biochemistry Sequence studies of several alphavirus genomic RNAs in the region containing the start of the subgenomic RNA (Sindbis virus/RNA sequence determination/sequence homology/nonstructural proteins/transcription initiation) JING-HSIUNG OU, CHARLES M. RICE, LYNN DALGARNO*, ELLEN G. STRAUSS, AND JAMES H. STRAUSSt Division of Biology, California Institute ofTechnology, Pasadena, California 91125 Communicated by Ray D. Owen, June 7, 1982 ABSTRACT The alphaviruses produce two mRNAs after in- RNA contains sequences identical to the 3'-terminal one-third fection: the genomic (49S) RNA which is translated into the non- of the genomic 49S RNA. The nucleotide sequences of the ge- structural (replicase) proteins and the subgenomic (26S) RNA nomic RNA preceding and including the beginning of the 26S which serves as the mRNA for the virion structural proteins. The RNA (called the "junction region") wouldbe expected to contain sequence of the region of the genomic RNA that contains the 5' signals important in the termination of translation of the non- end of the subgenomic RNA and the 5' flanking sequences in the structural polyprotein precursor as well as sequences important genomic RNA were determined for several alphaviruses. A highly for 26S RNA transcription. conserved sequence of 21 nucleotides was found which includes By comparing several different alphavirus sequences in a the first two nucleotides of the subgenomic RNA and the 19 nu- we have found that it is to cleotides preceding it. We propose that the complement of this given region of-interest, possible sequence in the minus strand is the recognition site used by the define potentially important features involved in alphavirus viral transcriptase for initiation of transcription of 26S RNA and replication by virtue of their conservation (or lack thereof) dur- that, in general, such short recognition sequences are commonly ing evolution (4-7). Here we report the sequences of four dif- used among the RNA viruses. The COOH-terminal sequence of ferent alphavirus junction regions. Comparison of these se- the nonstructural polyprotein precursor has been deduced for quences has revealed several common features and provides each virus. These protein sequences are highly homologous and insight intothe translation and transcription ofthe 26S and 49S are followed by multiple in-phase termination codons clustered in RNAs and the evolution of these viruses. the nontranslated region ofthe 26S RNA in each case. In contrast to the proposed transcriptase recognition site, the particular trip- MATERIALS AND METHODS lets used for a given conserved amino acid have diverged markedly during evolution of these viruses. The protein homology is suffi- Preparation of Vaccinia Guanylyltransferase. Guanylyl- cient, however, for deduction of the correct coding phase of the transferase was prepared from vaccinia virus (WR strain, isolate RNA and allows the alignment of the corresponding nucleic acid 11; a gift from W. K. Joklik) by a simplified version of the sequence data from different alphaviruses without knowledge of method of Paoletti et al. (8); 2 mg ofvaccinia virus in 640 ,ul of the sequence of the entire genomes. 1 mM Tris'HCl (pH 9.0) was used for each preparation of the enzyme. After disruption ofthe virus cores, the preparation was The Alphavirus genus of the Togaviridae family includes more briefly sonicated to shear viral DNA and then centrifuged at than 20 distinct viruses, many of which are important human 136,000 x gfor 60 min. The guanylyltransferase activity present or veterinary pathogens. These viruses are transmitted in na- in the supernatant was stable for 2 weeks at 4°C. ture via arthropod vectors and have the ability to replicate in 5'-End Labeling of Alphavirus 26S RNA. The 26S RNA of awide range ofvertebrates including both avian and mammalian Sindbis virus (HR strain), Semliki Forest virus, and Middelburg hosts. Because these viruses differ in geographical distribution, virus were prepared as described (4) and decapped by ,3-elim- host range, and the pathological result of infection but are ex- ination by the method of Rose and Lodish (9). Decapped RNA tremely similar in molecular architecture and in the pattern of was then end-labeled by using conditions modified from Ahl- events comprising viral replication, they provide an ideal sys- quistet al. (10). The 100-,ul reaction mixture contained 0.16mM tem for studyingthe evolutionary relationships amongmembers ATP, 1 mM MgCl2, 1 mM dithiothreitol, 2.5 p.M [a-32P]GTP of a structurally and functionally related virus group. Viruses (410 Ci/mmol; 1 Ci = 3.7 x 1010 becquerels; Amersham), 50 are known to mutate rapidly in general (1, 2), but the wide host mM Tris HCl (pH 7.8), and 15 ,ul ofguanylyltransferase extract. range ofthis group ofviruses and the alternation in nature be- After incubation at 37°C for 15 min, the reaction was terminated tween vertebrate and invertebrate hosts means that the al- by phenol/chloroform extraction followed by ethanol precipi- phaviruses are subject to different selective pressures compared tation as described (4). RNA pellets were resuspended in 50 ,ud to viruses with more limited host ranges. of 10 mM Tris'HCl, pH 8.2/10mM NaCl/1 mM EDTA/0.1% The alphavirus genome consists ofa single plus-strand RNA NaDodSO4 and purified from the unincorporated label by gel molecule about 13,000 nucleotides long with a sedimentation filtration on a Bio-Gel A-5m column. The excluded volume frac- coefficient of 49 S (reviewed in ref. 3). This genomic RNA tions containing theRNA were pooled and the RNA was ethanol serves as the mRNA for the nonstructural proteins which func- precipitated twice with carrier RNA and resuspended in water tion as the replicase/transcriptase activities required for rep- to a final concentration of about 2,000 cpm//.l. More than 1 lication ofthe viral RNA and for transcription ofa 26S subgeno- X 105 cpm was incorporated per 30 ,ug ofRNA, which was suf- mic mRNA encoding the virion structural proteins. The 26S ficient for several experiments. The publication costs ofthis article were defrayed in part by page charge * Present address: Biochemistry Dept., The Australian National Uni- payment. This article must therefore be hereby marked "advertise- versity, P.O. Box 4, Canberra, Australia 2600. ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. t To whom reprint requests should be addressed. 5235 5236 Biochemistry: Ou et at Proc. Natl. Acad. Sci. USA 79 (1982) Direct Enwymatic Sequence Determination of the RNA from the 5' End. Partial RNase digestion ofthe 5'-end labeled -, !I RNA was performed by the method recommended by P-L Bio- '. chemicals (method E998) and the products were separated on sequencing gels. A brief description of the reaction conditions is in the legend of Fig. 1. *W ...::" Single-Stranded cDNA Synthesis and Chemical Sequence b.-r;.e O ............. Determination. The 49S RNAs of Sindbis, Middelburg, and .A..' I Semliki Forest viruses were prepared as described (4). Ross mi .00. OW River virus (T48-strain.from R. Shope) 26S and 49S RNAs were Ae,. prepared from infected BHK cells. Single-stranded cDNA to these RNAs were synthesized with reverse transcriptase and calfthymus DNA or oligo(dT) primers as described (11). Prep- aration, isolation, and chemical sequence determination of 5'- end-labeled restriction fragments from these cDNAs have been described (11). RESULTS -p In recent studies, nearly the entire 26S RNA nucleotide se- quences for Sindbis virus (5) and Semliki Forest virus (12, 13) have been determined. However, the 5' termini of these RNAs, a necessary landmark for defining the junction region, could not be localized unambiguously from these data. Thus, 4, we began by directly determining the 5'-terminal 26S RNA se- -3- quences for several' alphaviruses. The 26S RNAs of'Sindbis, 0 Middelburg, and Semliki Forest viruses were decapped by /- S- elimination, end-labeled with guanylyltransferase, and then partially digested by using endonucleases with four different base specificities. The resulting products were separated' on acrylamide gels and the sequences were determined from the ladder produced. Fig. 1 presents part of the data to illustrate 0 the Each RNA was at least and the i. technique. analyzed twice, I 1, ".... --- ., .` --dommommi. sequences obtained are shown in lowercase letters in Fig. 2. In .:.. .i general this enzymatic method gives a clean signal for the pu- rines, but the pyrimidines are often ambiguous (Fig. 1). In order to determine the sequences preceding the start of FIG. 1. Direct 5'-end sequence analysis of the 26S RNAs of Sem- 26S RNA and to clarify the direct RNA sequence analysis data, liki Forest virus (Left) and Middelburg virus (Right). About 2,000 cpm chemical of cDNA made to of end-labeled RNA was enzymatically digested at 5600 for 3-10 min sequence analysis single-strand 49S (depending on the degree of digestion preferred). Alkaline hydrolysis RNA of Sindbis, Middelburg, Semliki Forest, and Ross River was at 9000 for 6 min. G reaction: 1 1.d of RNA was added to 3 1.d of viruses was used (Fig. 2, uppercase letters). The method in- buffer I containing 1 unit of RNase T1. A>G reaction: 1 ,ulofRNA was volved analysis of Hae III fragments of the cDNA (11) by the added to 3 1d ofbuffer II containing 1 unit of RNase U2. A+U reaction: methods of Maxam and Gilbert (14). For Sindbis and Middel- 1 p1of RNAwas added to 3 p1 ofbufferIcontaining 1 unit ofPhysarum burg viruses, restriction fragments produced by Hae III diges- M RNase. U+C reaction: 1 ul of RNA was added to 3 jil of buffer m tion ofcDNA to 49S' RNA were selected for containing 1 unit of Bacillus cereus RNase.