Identification and characterization of a coronavirus packaging signal.

Previously, a mouse hepatitis virus (MHV) genomic sequence necessary for defective interfering (DI) RNA packaging into MHV particles (packaging signal) was mapped to within a region of 1,480 nucleotides in the MHV polymerase gene by comparison of two DI RNAs. One of these, DIssF, is 3.6 kb in size and exhibits efficient packaging, whereas the other, DIssE, which is 2.3 kb, does not. For more precise mapping, a series of mutant DIssF RNAs with deletions within this 1,480-nucleotide region were constructed. After transfection of in vitro-synthesized mutant DI RNA in MHV-infected cells, the virus product was passaged several times. The efficiency of DI RNA packaging into MHV virions was then estimated by viral homologous interference activity and by analysis of intracellular virus-specific RNAs and virion RNA. The results indicated that an area of 190 nucleotides was necessary for packaging. A computer-generated secondary structural analysis of the A59 and JHM strains of MHV demonstrated that within this 190-nucleotide region a stable stem-loop of 69 nucleotides was common between the two viruses. A DIssE-derived DI DNA which had these 69 nucleotides inserted into the DIssE sequence demonstrated efficient DI RNA packaging. Site-directed mutagenic analysis showed that of these 69 nucleotides, the minimum sequence of the packaging signal was 61 nucleotides and that destruction of the secondary structure abolished packaging ability. These studies demonstrated that an MHV packaging signal was present within the 61 nucleotides, which are located on MHV genomic RNA 1,381 to 1,441 nucleotides upstream of the 3' end of gene 1.     

Characterization of a murine coronavirus defective interfering RNA internal cis-acting replication signal.

The mouse hepatitis virus (MHV) sequences required for replication of the JHM strain of MHV defective interfering (DI) RNA consist of three discontinuous genomic regions: about 0.47 kb from both terminal sequences and a 0.13-kb internal region present at about 0.9 kb from the 5' end of the DI genome. In this study, we investigated the role of the internal 0.13-kb region in MHV RNA replication. Overall sequences of the 0.13-kb regions from various MHV strains were similar to each other, with nucleotide substitutions in some strains; MHV-A59 was exceptional, with three nucleotide deletions. Computer-based secondary-structure analysis of the 0.13-kb region in the positive strand revealed that most of the MHV strains formed the same or a similar main stem-loop structure, whereas only MHV-A59 formed a smaller main stem-loop structure. The RNA secondary structures in the negative strands were much less uniform among the MHV strains. A series of DI RNAs that contained MHV-JHM-derived 5'- and 3'-terminal sequences plus internal 0.13-kb regions derived from various MHV strains were constructed. Most of these DI RNAs replicated in MHV-infected cells, except that MRP-A59, with a 0.13-kb region derived from MHV-A59, failed to replicate. Interestingly, replication of MRP-A59 was temperature dependent; it occurred at 39.5 degrees C but not at 37 or 35 degrees C, whereas a DI RNA with an MHV-JHM-derived 0.13-kb region replicated at all three temperatures. At 37 degrees C, synthesis of MRP-A59 negative-strand RNA was detected in MHV-infected and MRP-A59 RNA-transfected cells. Another DI RNA with the internal 0.13-kb region deleted also synthesized negative-strand RNA in MHV-infected cells. MRP-A59-transfected cells were shifted from 39.5 to 37 degrees C at 5.5 h postinfection, a time when most MHV negative-strand RNAs have already accumulated; after the shift, MRP-A59 positive-strand RNA synthesis ceased. The minimum sequence required for maintenance of the positive-strand major stem-loop structure and biological function of the MHV-JHM 0.13-kb region was about 57 nucleotides. Function was lost in the 50-nucleotide sequence that formed a positive-strand stem-loop structure identical to that of MHV-A59. These studies suggested that the RNA structure made by the internal sequence was important for positive-strand MHV RNA synthesis.     

Analysis of efficiently packaged defective interfering RNAs of murine coronavirus: localization of a possible RNA-packaging signal.

We have previously shown that most of the defective interfering (DI) RNA of mouse hepatitis virus (MHV) are not packaged into virions. We have now identified, after 21 serial undiluted passages of MHV, a small DI RNA, DIssF, which is efficiently packaged into virions. The DIssF RNA replicated at a high efficiency on its transfection into the helper virus-infected cells. The virus released from the transfected cells interfered strongly with mRNA synthesis and growth of helper virus. cDNA cloning and sequence analysis of DIssF RNA revealed that it is 3.6 kb and consists of sequences derived from five discontinuous regions of the genome of the nondefective virus. The first four regions (domains I to IV) from the 5' end are derived from gene 1, which presumably encodes the RNA polymerase of the nondefective virus. The entire domain I (859 nucleotides) and the first 750 nucleotides of domain II are also present in a previously characterized DI RNA, DIssE, which is not efficiently packaged into virions. Furthermore, the junction between these two domains is identical between the two DI RNAs. The remaining 77 nucleotides at the 3' end of domain II and all of domains III (655 nucleotides) and IV (770 nucleotides) are not present in DIssE RNA. These four domains are derived from gene 1. In contrast, the 3'-most domain (domain V, 447 nucleotides) is derived from the 3' end of the genomic RNA and is also present in DIssE. The comparison of primary sequences and packaging properties between DIsse and DIssF RNAs suggested that domains III and IV and part of the 3' end of domain II contain the packaging signal for MHV RNA. This conclusion was confirmed by inserting these DIssF-unique sequences into a DIssE cDNA construct; the in vitro-transcribed RNA from this hybrid construct was efficiently packaged into virion particles. DIssF RNA also contains an open reading frame, which begins from domain I and ends at the 5'-end 20 bases of domain III. In vitro translation of DIssF RNA and metabolic labeling of the virus-infected cells showed that this open reading frame is indeed translated into a 75-kDa protein. The structures of both DIssE and DIssF RNAs suggest that a protein-encoding capability is a common characteristic of MHV DI RNA.     

Specific binding of host cellular proteins to multiple sites within the 3' end of mouse hepatitis virus genomic RNA.

The initial step in mouse hepatitis virus (MHV) RNA replication is the synthesis of negative-strand RNA from a positive-strand genomic RNA template. Our approach to begin studying MHV RNA replication is to identify the cis-acting signals for RNA synthesis and the proteins which recognize these signals at the 3' end of genomic RNA of MHV. To determine whether host cellular and/or viral proteins interact with the 3' end of the coronavirus genome, an RNase T1 protection/gel mobility shift electrophoresis assay was used to examine cytoplasmic extracts from mock- and MHV-JHM-infected 17Cl-1 murine cells for the ability to form complexes with defined regions of the genomic RNA. We demonstrated the specific binding of host cell proteins to multiple sites within the 3' end of MHV-JHM genomic RNA. By using a set of RNA probes with deletions at either the 5' or 3' end or both ends, two distinct binding sites were located. The first protein-binding element was mapped in the 3'-most 42 nucleotides of the genomic RNA [3' (+42) RNA], and the second element was mapped within an 86-nucleotide sequence encompassing nucleotides 171 to 85 from the 3' end of the genome (171-85 RNA). A single potential stem-loop structure is predicted for the 3' (+)42 RNA, and two stem-loop structures are predicted for the 171-85 RNA. Proteins interacting with these two elements were identified by UV-induced covalent cross-linking to labeled RNAs followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis. The RNA-protein complex formed with the 3'-most 42 nucleotides contains approximately five host polypeptides, a highly labeled protein of 120 kDa and four minor species with sizes of 103, 81, 70, and 55 kDa. The second protein-binding element, contained within a probe representing nucleotides 487 to 85 from the 3' end of the genome, also appears to bind five host polypeptides, 142, 120, 100, 55, and 33 kDa in size, with the 120-kDa protein being the most abundant. The RNA-protein complexes observed with MHV-infected cells in both RNase protection/gel mobility shift and UV cross-linking assays were identical to those observed with uninfected cells. The possible involvement of the interaction of host proteins with the viral genome during MHV replication is discussed.     

Sequence and structural elements at the 3' terminus of bovine viral diarrhea virus genomic RNA: functional role during RNA replication

Bovine viral diarrhea virus (BVDV), a member of the genus Pestivirus in the family Flaviviridae, has a positive-stranded RNA genome consisting of a single open reading frame and untranslated regions (UTRs) at the 5' and 3' ends. Computer modeling suggested the 3' UTR comprised single-stranded regions as well as stem-loop structures-features that were suspected of being essentially implicated in the viral RNA replication pathway. Employing a subgenomic BVDV RNA (DI9c) that was shown to function as an autonomous RNA replicon (S.-E. Behrens, C. W. Grassmann, H. J. Thiel, G. Meyers, and N. Tautz, J. Virol. 72:2364-2372, 1998) the goal of this study was to determine the RNA secondary structure of the 3' UTR by experimental means and to investigate the significance of defined RNA motifs for the RNA replication pathway. Enzymatic and chemical structure probing revealed mainly the conserved terminal part (termed 3'C) of the DI9c 3' UTR containing distinctive RNA motifs, i.e., a stable stem-loop, SL I, near the RNA 3' terminus and a considerably less stable stem-loop, SL II, that forms the 5' portion of 3'C. SL I and SL II are separated by a long single-stranded intervening sequence, denoted SS. The 3'-terminal four C residues of the viral RNA were confirmed to be single stranded as well. Other intramolecular interactions, e.g., with upstream DI9c RNA sequences, were not detected under the experimental conditions used. Mutagenesis of the DI9c RNA demonstrated that the SL I and SS motifs do indeed play essential roles during RNA replication. Abolition of RNA stems, which ought to maintain the overall folding of SL I, as well as substitution of certain single-stranded nucleotides located in the SS region or SL I loop region, gave rise to DI9c derivatives unable to replicate. Conversely, SL I stems comprising compensatory base exchanges turned out to support replication, but mostly to a lower degree than the original structure. Surprisingly, replacement of a number of residues, although they were previously defined as constituents of a highly conserved stretch of sequence of the SS motif, had little effect on the replication ability of DI9c. In summary, these results indicate that RNA structure as well as sequence elements harbored within the 3'C region of the BVDV 3' UTR create a common cis-acting element of the replication process. The data further point at possible interaction sites of host and/or viral proteins and thus provide valuable information for future experiments intended to identify and characterize these factors.     

Functional equivalence of common and unique sequences in the 3' untranslated regions of alfalfa mosaic virus RNAs 1, 2, and 3.

The 3' untranslated regions (UTRs) of alfalfa mosaic virus (AMV) RNAs 1, 2, and 3 consist of a common 3'-terminal sequence of 145 nucleotides (nt) and upstream sequences of 18 to 34 nt that are unique for each RNA. The common sequence can be folded into five stem-loop structures, A to E, despite the occurrence of 22 nt differences between the three RNAs in this region. Exchange of the common sequences or full-length UTRs between the three genomic RNAs did not affect the replication of these RNAs in vivo, indicating that the UTRs are functionally equivalent. Mutations that disturbed base pairing in the stem of hairpin E reduced or abolished RNA replication, whereas compensating mutations restored RNA replication. In vitro, the 3' UTRs of the three RNAs were recognized with similar efficiencies by the AMV RNA-dependent RNA polymerase (RdRp). A deletion analysis of template RNAs indicated that a 3'-terminal sequence of 127 nt in each of the three AMV RNAs was not sufficient for recognition by the RdRp. Previously, it has been shown that this 127-nt sequence is sufficient for coat protein binding. Apparently, sequences required for recognition of AMV RNAs by the RdRp are longer than sequences required for CP binding.     

Deletion mapping of a mouse hepatitis virus defective interfering RNA reveals the requirement of an internal and discontiguous sequence for replication.

All of the defective interfering (DI) RNAs of mouse hepatitis virus (MHV) contain both the 5' and 3' ends of the viral genomic RNA, which presumably include the cis sequences required for RNA replication. To define the replication signal of MHV RNA, we have used a vaccinia virus-T7 polymerase-transcribed MHV DI RNA to study the effects of sequence deletion on DI RNA replication. Following infection of susceptible cells with a recombinant vaccinia virus expressing T7 RNA polymerase, various cDNA clones derived from a DI RNA (DIssF) of the JHM strain of MHV, which is a 3.5-kb naturally occurring DI RNA, behind a T7 promoter were transfected. On superinfection with a helper MHV, the ability of various DI RNAs to replicate was determined. Serial deletions from the middle of the RNA toward both the 5' and 3' ends demonstrated that 859 nucleotides from the 5' end and 436 nucleotides from the 3' end of the MHV RNA genome were necessary for RNA replication. Surprisingly, an additional stretch of 135 nucleotides located at 3.1 to 3.3 kb from the 5' end of the genome was also required. This stretch is discontiguous from the 5'-end cis replication signal and is present in all of the naturally occurring DI RNAs studied so far. The requirement for a long stretch of 5'- and 3'-end sequences predicts that the subgenomic MHV mRNAs cannot replicate. The efficiency of RNA replication varied with different cDNA constructs, suggesting possible interaction between different regions of DI RNA. The identification of MHV RNA replication signals allowed the construction of an MHV DI-based expression vector, which can express foreign genes, such as the chloramphenicol acetyltransferase gene.     

Comparison of SIV and HIV-1 Genomic RNA Structures Reveals Impact of Sequence Evolution on Conserved and Non-Conserved Structural Motifs

RNA secondary structure plays a central role in the replication and metabolism of all RNA viruses, including retroviruses like HIV-1. However, structures with known function represent only a fraction of the secondary structure reported for HIV-1_NL4-3. One tool to assess the importance of RNA structures is to examine their conservation over evolutionary time. To this end, we used SHAPE to model the secondary structure of a second primate lentiviral genome, SIVmac239, which shares only 50% sequence identity at the nucleotide level with HIV-1_NL4-3. Only about half of the paired nucleotides are paired in both genomic RNAs and, across the genome, just 71 base pairs form with the same pairing partner in both genomes. On average the RNA secondary structure is thus evolving at a much faster rate than the sequence. Structure at the Gag-Pro-Pol frameshift site is maintained but in a significantly altered form, while the impact of selection for maintaining a protein binding interaction can be seen in the conservation of pairing partners in the small RRE stems where Rev binds. Structures that are conserved between SIVmac239 and HIV-1_NL4-3 also occur at the 5′ polyadenylation sequence, in the plus strand primer sites, PPT and cPPT, and in the stem-loop structure that includes the first splice acceptor site. The two genomes are adenosine-rich and cytidine-poor. The structured regions are enriched in guanosines, while unpaired regions are enriched in adenosines, and functionaly important structures have stronger base pairing than nonconserved structures. We conclude that much of the secondary structure is the result of fortuitous pairing in a metastable state that reforms during sequence evolution. However, secondary structure elements with important function are stabilized by higher guanosine content that allows regions of structure to persist as sequence evolution proceeds, and, within the confines of selective pressure, allows structures to evolve.     

The 5'-terminal region of the Aichi virus genome encodes cis-acting replication elements required for positive- and negative-strand RNA synthesis

Aichi virus is a member of the family Picornaviridae. It has already been shown that three stem-loop structures (SL-A, SL-B, and SL-C, from the 5' end) formed at the 5' end of the genome are critical elements for viral RNA replication. In this study, we further characterized the 5'-terminal cis-acting replication elements. We found that an additional structural element, a pseudoknot structure, is formed through base-pairing interaction between the loop segment of SL-B (nucleotides [nt] 57 to 60) and a sequence downstream of SL-C (nt 112 to 115) and showed that the formation of this pseudoknot is critical for viral RNA replication. Mapping of the 5'-terminal sequence of the Aichi virus genome required for RNA replication using a series of Aichi virus-encephalomyocarditis virus chimera replicons indicated that the 5'-end 115 nucleotides including the pseudoknot structure are the minimum requirement for RNA replication. Using the cell-free translation-replication system, we examined the abilities of viral RNAs with a lethal mutation in the 5'-terminal structural elements to synthesize negative- and positive-strand RNAs. The results showed that the formation of three stem-loops and the pseudoknot structure at the 5' end of the genome is required for negative-strand RNA synthesis. In addition, specific nucleotide sequences in the stem of SL-A or its complementary sequences at the 3' end of the negative-strand were shown to be critical for the initiation of positive-strand RNA synthesis but not for that of negative-strand synthesis. Thus, the 5' end of the Aichi virus genome encodes elements important for not only negative-strand synthesis but also positive-strand synthesis.     

An internally located RNA hairpin enhances replication of Tomato bushy stunt virus RNAs

Defective interfering (DI) RNAs of Tomato bushy stunt virus (TBSV), a plus-sense RNA virus, comprise four conserved noncontiguous regions (I through IV) derived from the viral genome. Region III, a 70-nucleotide-long sequence corresponding to a genomic segment located 378 nucleotides upstream of the 3' terminus of the genome, has been found to enhance DI RNA accumulation by approximately 10-fold in an orientation-independent manner (D. Ray and K. A. White, Virology 256:162-171, 1999). In this study, a more detailed structure-function analysis of region III was conducted. RNA secondary-structure analyses indicated that region III contains stem-loop structures in both plus and minus strands. Through deletion analyses of a DI RNA, a primary determinant of region III activity was mapped to the 5'-proximal 35-nucleotide segment. Compensatory-type mutational analyses showed that a stem-loop structure in the minus strand of this subregion was required for enhanced DI RNA replication. The same stem-loop structure was also found to function in a position-independent manner in a DI RNA (albeit at reduced levels) and to be important for efficient accumulation within the context of the TBSV genome. Taken together, these observations suggest that the 5'-proximal segment of region III is a modular RNA replication element that functions primarily through the formation of an RNA hairpin structure in the minus strand.     

Human cytomegalovirus UL84 interacts with an RNA stem-loop sequence found within the RNA/DNA hybrid region of oriLyt

Human cytomegalovirus (HCMV) lytic DNA replication is initiated at the complex cis-acting oriLyt region, which spans nearly 3 kb. DNA synthesis requires six core proteins together with UL84 and IE2. Previously, two essential regions were identified within oriLyt. Essential region I (nucleotides [nt] 92209 to 92573) can be replaced with the constitutively active simian virus 40 promoter, which in turn eliminates the requirement for IE2 in the origin-dependent transient-replication assay. Essential region II (nt 92979 to 93513) contains two elements of interest: an RNA/DNA hybrid domain and an inverted repeat sequence capable of forming a stem-loop structure. Our studies now reveal for the first time that UL84 interacts with a stem-loop RNA oligonucleotide in vitro, and although UL84 interacted with other nucleic acid substrates, a specific interaction occurred only with the RNA stem-loop. Increasing concentrations of purified UL84 produced a remarkable downward-staircase pattern, which is not due to a nuclease activity but is dependent upon the presence of secondary structures, suggesting that UL84 modifies the conformation of the RNA substrate. Cross-linking experiments show that UL84 possibly changes the conformation of the RNA substrate. The addition of purified IE2 to the in vitro binding reaction did not affect binding to the stem-loop structure. Chromatin immunoprecipitation assays performed using infected cells and purified virus show that UL84 is bound to oriLyt in a region adjacent to the RNA/DNA hybrid and the stem-loop structure. These results solidify UL84 as the potential initiator of HCMV DNA replication through a unique interaction with a conserved RNA stem-loop structure within oriLyt.     

The Bovine Leukemia Virus Encapsidation Signal Is Composed of RNA Secondary Structures

The encapsidation signal of bovine leukemia virus (BLV) was previously shown by deletion analysis to be discontinuous and to extend into the 5′ end of the gag gene (L. Mansky et al., J. Virol. 69:3282–3289, 1995). The global minimum-energy optimal folding for the entire BLV RNA, including the previously mapped primary and secondary encapsidation signal regions, was analyzed. Two stable stem-loop structures (located just downstream of the gag start codon) were predicted within the primary signal region, and one stable stem-loop structure (in the gag gene) was predicted in the secondary signal region. Based on these predicted structures, we introduced a series of mutations into the primary and secondary encapsidation signals in order to explore the sequence and structural information contained within these regions. The replication efficiency and levels of cytoplasmic and virion RNA were analyzed for these mutants. Mutations that disrupted either or both of the predicted stem-loop structures of the primary signal reduced the replication efficiency by factors of 7 and 40, respectively; similar reductions in RNA encapsidation efficiency were observed. The mutant with both stem-loop structures disrupted had a phenotype similar to that of a mutant containing a deletion of the entire primary signal region. Mutations that disrupted the predicted stem-loop structure of the secondary signal led to similar reductions (factors of 4 to 6) in both the replication and RNA encapsidation efficiencies. The introduction of compensatory mutations into mutants from both the primary and secondary signal regions, which restored the predicted stem-loop structures, led to levels of replication and RNA encapsidation comparable to those of virus containing the wild-type encapsidation signal. Replacement of the BLV RNA region containing the primary and secondary encapsidation signals with a similar region from human T-cell leukemia virus (HTLV) type 1 or type 2 led to virus replication at three-quarters or one-fifth of the level of the parental virus, respectively. The results from both the compensatory mutants and BLV-HTLV chimeras indicate that the encapsidation sequences are recognized largely by their secondary or tertiary structures.     

Structural elements of the 3'-terminal coat protein binding site in alfalfa mosaic virus RNAs.

The 3'-terminal of the three genomic RNAs of alfalfa mosaic virus (AIMV) and ilarviruses contain a number of AUGC-motifs separated by hairpin structures. Binding of coat protein (CP) to such elements in the RNAs is required to initiate infection of these viruses. Determinants for CP binding in the 3'-terminal 39 nucleotides (nt) of AIMV RNA 3 were analyzed by band-shift assays. From the 5'- to 3'-end this 39 nt sequence contains AUGC-motif 3, stem-loop structure 2 (STLP2), AUGC-motif 2, stem-loop structure 1 (STLP1) and AUGC-motif 1. A mutational analysis showed that all three AUGC-motifs were involved in CP binding. Mutation of the A- and U-residues of motifs 1 or 3 had no effect on CP binding but similar mutations in motif 2 abolished CP binding. A mutational analysis of the stem of STLP1 and STLP2 confirmed the importance of these hairpins for CP binding. Randomization of the sequence of the stems and loops of STLP1 and STLP2 had no effect on CP binding as long as the secondary structure was maintained. This indicates that the two hairpins are not involved in sequence-specific interactions with CP. They may function in a secondary structure-specific interaction with CP and/or in the assembly of the AUGC-motifs in a configuration required for CP binding.     

Interaction between polypeptide 3ABC and the 5'-terminal structural elements of the genome of Aichi virus: implication for negative-strand RNA synthesis

Secondary structural elements at the 5' end of picornavirus genomic RNA function as cis-acting replication elements and are known to interact specifically with viral P3 proteins in several picornaviruses. In poliovirus, ribonucleoprotein complex formation at the 5' end of the genome is required for negative-strand synthesis. We have previously shown that the 5'-end 115 nucleotides of the Aichi virus genome, which are predicted to fold into two stem-loops (SL-A and SL-C) and one pseudoknot (PK-B), act as a cis-acting replication element and that correct folding of these structures is required for negative-strand synthesis. In this study, we investigated the interaction between the 5'-terminal 120 nucleotides of the genome and the P3 proteins, 3AB, 3ABC, 3C, and 3CD, by gel shift assay and Northwestern analysis. The results showed that 3ABC and 3CD bound to the 5'-terminal region specifically. The binding of 3ABC was observed on both assays, while that of 3CD was detected only on Northwestern analysis. No binding of 3AB or 3C was observed. Binding assays using mutant RNAs demonstrated that disruption of the base pairings of the stem of SL-A and one of the two stem segments of PK-B (stem-B1) abolished the 3ABC binding. In addition, the specific nucleotide sequence of stem-B1 was responsible for the efficient 3ABC binding. These results suggest that the interaction of 3ABC with the 5'-terminal region of the genome is involved in negative-strand synthesis. On the other hand, the ability of 3CD to interact with the 5'-terminal region did not correlate with the RNA replication ability.     

Secondary structure determination of the conserved 98-base sequence at the 3' terminus of hepatitis C virus genome RNA.

The RNA genome of hepatitis C virus (HCV) terminates with a highly conserved 98-base sequence. Enzymatic and chemical approaches were used to define the secondary structure of this 3'-terminal element in RNA transcribed in vitro from cloned cDNA. Both approaches yielded data consistent with a stable stem-loop structure within the 3'-terminal 46 bases. In contrast, the 5' 52 nucleotides of this 98-base element appear to be less ordered and may exist in multiple conformations. Under the experimental conditions tested, interaction between the 3' 98 bases and upstream HCV sequences was not detected. These data provide valuable information for future experiments aimed at identifying host and/or viral proteins which interact with this highly conserved RNA element.     

Specific interaction of hepatitis C virus protease/helicase NS3 with the 3'-terminal sequences of viral positive- and negative-strand RNA

The hepatitis C virus (HCV)-encoded protease/helicase NS3 is likely to be involved in viral RNA replication. We have expressed and purified recombinant NS3 (protease and helicase domains) and Delta pNS3 (helicase domain only) and examined their abilities to interact with the 3'-terminal sequence of both positive and negative strands of HCV RNA. These regions of RNA were chosen because initiation of RNA synthesis is likely to occur at or near the 3' untranslated region (UTR). The results presented here demonstrate that NS3 (and Delta pNS3) interacts efficiently and specifically with the 3'-terminal sequences of both positive- and negative-strand RNA but not with the corresponding complementary 5'-terminal RNA sequences. The interaction of NS3 with the 3'-terminal negative strand [called 3'(-) UTR(127)] was specific in that only homologous (and not heterologous) RNA competed efficiently in the binding reaction. A predicted stem-loop structure present at the 3' terminus (nucleotides 5 to 20 from the 3' end) of the negative-strand RNA appears to be important for NS3 binding to the negative-strand UTR. Deletion of the stem-loop structure almost totally impaired NS3 (and Delta pNS3) binding. Additional mutagenesis showed that three G-C pairs within the stem were critical for helicase-RNA interaction. The data presented here also suggested that both a double-stranded structure and the 3'-proximal guanosine residues in the stem were important determinants of protein binding. In contrast to the relatively stringent requirement for 3'(-) UTR binding, specific interaction of NS3 (or Delta pNS3) with the 3'-terminal sequences of the positive-strand RNA [3'(+) UTR] appears to require the entire 3'(+) UTR of HCV. Deletion of either the 98-nucleotide 3'-terminal conserved region or the 5' half sequence containing the variable region and the poly(U) and/or poly(UC) stretch significantly impaired RNA-protein interaction. The implication of NS3 binding to the 3'-terminal sequences of viral positive- and negative-strand RNA in viral replication is discussed.