Molecular cloning and characterization of the RNA packaging-defective retrovirus SE21Q1b

NMR Studies of Retroviral Genome Packaging

Nearly all retroviruses selectively package two copies of their unspliced RNA genomes from a cellular milieu that contains a substantial excess of non-viral and spliced viral RNAs. Over the past four decades, combinations of genetic experiments, phylogenetic analyses, nucleotide accessibility mapping, in silico RNA structure predictions, and biophysical experiments were employed to understand how retroviral genomes are selected for packaging. Genetic studies provided early clues regarding the protein and RNA elements required for packaging, and nucleotide accessibility mapping experiments provided insights into the secondary structures of functionally important elements in the genome. Three-dimensional structural determinants of packaging were primarily derived by nuclear magnetic resonance (NMR) spectroscopy. A key advantage of NMR, relative to other methods for determining biomolecular structure (such as X-ray crystallography), is that it is well suited for studies of conformationally dynamic and heterogeneous systems—a hallmark of the retrovirus packaging machinery. Here, we review advances in understanding of the structures, dynamics, and interactions of the proteins and RNA elements involved in retroviral genome selection and packaging that are facilitated by NMR.

Orchestrating the Selection and Packaging of Genomic RNA by Retroviruses: An Ensemble of Viral and Host Factors

Infectious retrovirus particles contain two copies of unspliced viral RNA that serve as the viral genome. Unspliced retroviral RNA is transcribed in the nucleus by the host RNA polymerase II and has three potential fates: (1) it can be spliced into subgenomic messenger RNAs (mRNAs) for the translation of viral proteins; or it can remain unspliced to serve as either (2) the mRNA for the translation of Gag and Gag-Pol; or (3) the genomic RNA (gRNA) that is packaged into virions. The Gag structural protein recognizes and binds the unspliced viral RNA to select it as a genome, which is selected in preference to spliced viral RNAs and cellular RNAs. In this review, we summarize the current state of understanding about how retroviral packaging is orchestrated within the cell and explore potential new mechanisms based on recent discoveries in the field. We discuss the cis-acting elements in the unspliced viral RNA and the properties of the Gag protein that are required for their interaction. In addition, we discuss the role of host factors in influencing the fate of the newly transcribed viral RNA, current models for how retroviruses distinguish unspliced viral mRNA from viral genomic RNA, and the possible subcellular sites of genomic RNA dimerization and selection by Gag. Although this review centers primarily on the wealth of data available for the alpharetrovirus Rous sarcoma virus, in which a discrete RNA packaging sequence has been identified, we have also summarized the cis- and trans-acting factors as well as the mechanisms governing gRNA packaging of other retroviruses for comparison.

In vitro characterization of a base pairing interaction between the primer binding site and the minimal packaging signal of avian leukosis virus genomic RNA

The 5' leader region of avian sarcoma-leukosis viruses (ASLVs) folds into a series of RNA secondary structures which are involved in key steps in the viral replication cycle such as reverse transcription, dimerization and packaging of genomic RNA. The O3 stem and three stem-loops (O3SLa, O3SLb and O3SLc) form the minimal packaging signal that is located downstream of the primer binding site (PBS). The U5-PBS region contributes to packaging via a mechanism that remains unknown. In this in vitro study, we have investigated the possibility of interactions between the R-U5-PBS region and the minimal packaging signal using chemical and enzymatic probing, antisense oligonucleotides and site-directed mutagenesis. We have identified a base pairing interaction between the PBS sequence and the terminal loop of O3SLa. It was found that the PBS/O3SLa interaction was intramolecular since it occurred not only in dimeric RNA but also in monomeric RNA. This interaction probably corresponds to a pseudoknot interaction. The PBS/O3SLa interaction may be formed in vivo since the sequences are highly conserved in ASLV strains. The PBS/O3SLa interaction may regulate the processes of primer tRNA annealing, packaging and initiation of Gag translation through its involvement in leader tertiary structure. Interestingly, we found that in other retroviruses the PBS sequence can also base pair with a terminal loop of the stem-loops involved in RNA packaging.

cis-Acting elements important for retroviral RNA packaging specificity

Spleen necrosis virus (SNV) proteins can package RNA from distantly related murine leukemia virus (MLV), whereas MLV proteins cannot package SNV RNA efficiently. We used this nonreciprocal recognition to investigate regions of packaging signals that influence viral RNA encapsidation specificity. Although the MLV and SNV packaging signals (Psi and E, respectively) do not contain significant sequence homology, they both contain a pair of hairpins. This hairpin pair was previously proposed to be the core element in MLV Psi. In the present study, MLV-based vectors were generated to contain chimeric SNV/MLV packaging signals in which the hairpins were replaced with the heterologous counterpart. The interactions between these chimeras and MLV or SNV proteins were examined by virus replication and RNA analyses. SNV proteins recognized all of the chimeras, indicating that these chimeras were functional. We found that replacing the hairpin pair did not drastically alter the ability of MLV proteins to package these chimeras. These results indicate that, despite the important role of the hairpin pair in RNA packaging, it is not the major motif responsible for the ability of MLV proteins to discriminate between the MLV and SNV packaging signals. To determine the role of sequences flanking the hairpins in RNA packaging specificity, vectors with swapped flanking regions were generated and evaluated. SNV proteins packaged all of these chimeras efficiently. In contrast, MLV proteins strongly favored chimeras with the MLV 5'-flanking regions. These data indicated that MLV Gag recognizes multiple elements in the viral packaging signal, including the hairpin structure and flanking regions.

Characterization of loose and tight dimer forms of avian leukosis virus RNA

Retroviral genomes consist of two identical RNA molecules joined non-covalently near their 5'-ends. Recently, we showed that an imperfect autocomplementary sequence, located in the L3 domain, plays an essential role in avian sarcoma-leukosis virus (ASLV) RNA dimerization in vitro. This sequence can adopt a stem-loop structure and is involved in ASLV replication. Here, we found that in the absence of nucleocapsid protein, RNA transcripts of avian leukosis virus (ALV) were able to form two types of dimers in vitro that differ in their stability: a loose dimer, formed at a physiological temperature, and a tight dimer, formed at a high temperature. A mutational analysis was performed to define the features of these dimers. The results of this analysis unambiguously confirm that the two L3 stem-loops interact directly in both types of dimers. A loop-loop interaction is the main linkage in the loose dimer. In contrast, in the tight dimer, the stem and the loop of the L3 hairpin form an extended duplex. Surprisingly, we also found that the dimerization properties defined for our ALV strain (type SR-A) differ from those found in other ASLV strains.

Retrotransposition of nonviral RNAs in an avian packaging cell line

Retroviruses produced from the quail packaging cell line SE21Q1b predominantly contain cellular RNAs instead of viral RNAs. These RNAs can be reverse transcribed and integrated into the genomes of newly infected cells and are thereafter referred to as newly formed retrogenes. We investigated whether retrogene formation can occur within SE21Q1b cells themselves and whether this occurs intracellularly or via extracellular reinfection. By using packaging cell line mutants derived from the SE21Q1b provirus and selectable reporter constructs, we found that the process requires envelope glycoproteins and a retroviral packaging signal. Our results suggest that extracellular reinfection is the primary route of retrotransposition of nonviral RNAs.

A short autocomplementary sequence plays an essential role in avian sarcoma-leukosis virus RNA dimerization

Retroviral genomes consist of two identical RNA molecules joined noncovalently near their 5'-ends. Recently, two models have been proposed for RNA dimer formation on the basis of results obtained in vitro with human immunodeficiency virus type 1 RNA and Moloney murine leukemia virus RNA. It was first proposed that viral RNA dimerizes by forming an interstrand quadruple helix with purine tetrads. The second model postulates that RNA dimerization is initiated by a loop-loop interaction between the two RNA molecules. In order to better characterize the dimerization process of retroviral genomic RNA, we analyzed the in vitro dimerization of avian sarcoma-leukosis virus (ASLV) RNA using different transcripts. We determined the requirements for heterodimer formation, the thermal dissociation of RNA dimers, and the influence of antisense DNA oligonucleotides on dimer formation. Our results strongly suggest that purine tetrads are not involved in dimer formation. Data show that an autocomplementary sequence located upstream from the splice donor site and within a major packaging signal plays a crucial role in ASLV RNA dimer formation in vitro. This sequence is able to form a stem-loop structure, and phylogenetic analysis reveals that it is conserved in 28 different avian sarcoma and leukosis viruses. These results suggest that dimerization of ASLV RNA is initiated by a loop-loop interaction between two RNA molecules and provide an additional argument for the ubiquity of the dimerization process via loop-loop interaction.

The packaging phenotype of the SE21Q1b provirus is related to high proviral expression and not trans-acting factors

The avian packaging cell line SE21Q1b produces particles which encapsidate cellular RNAs. Such RNAs can be reverse transcribed by endogenous polymerase and integrated into the genomes of newly infected cells (M. Linial, Cell 49:93-102, 1987). Genomic RNA is not packaged because the packaging (psi) region of the provirus is deleted. The provirus also lacks the negative-strand primer binding site, which prevents efficient reverse transcription of randomly packaged genomic RNA. Previous work from our laboratory suggested that the trans-acting defect which allows packaging of cellular mRNA mapped to the provirus but did not map to the nucleocapsid region of the gag gene (D.J. Anderson, P. Lee, K. L. Levine, J. Sang, S. A. Shah, O. O. Yang, P. R. Shank, and M. L. Linial, J. Virol. 66:204-216, 1992). We have found, using proviral recombinants between SE21Q1b and wild-type Rous sarcoma virus, that packaging of cellular RNAs does not map to the gag gene. Rather, the propensity of SE21Q1b particles to package cellular mRNA is a function of the high level of particle production in these cells and not of any specific viral structural proteins.

Retroviral nucleocapsid domains mediate the specific recognition of genomic viral RNAs by chimeric Gag polyproteins during RNA packaging in vivo

The retroviral nucleocapsid (NC) protein is necessary for the specific encapsidation of the viral genomic RNA by the assembling virion. However, it is unclear whether NC contains the determinants for the specific recognition of the viral RNA or instead contributes nonspecific RNA contacts to strengthen a specific contact made elsewhere in the Gag polyprotein. To discriminate between these two possibilities, we have swapped the NC domains of the human immunodeficiency virus type 1 (HIV-1) and Moloney murine leukemia virus (M-MuLV), generating an HIV-1 mutant containing the M-MuLV NC domain and an M-MuLV mutant containing the HIV-1 NC domain. These mutants, as well as several others, were characterized for their abilities to encapsidate HIV-1, M-MuLV, and nonviral RNAs and to preferentially package genomic viral RNAs over spliced viral RNAs. We found that the M-MuLV NC domain mediates the specific packaging of RNAs containing the M-MuLV psi packaging element, while the HIV-1 NC domain confers an ability to package the unspliced HIV-1 RNA over spliced HIV-1 RNAs. In addition, we found that the HIV-1 mutant containing the M-MuLV NC domain exhibited a 20-fold greater ability than wild-type HIV-1 to package a nonviral RNA. These results help confirm the notion that the NC domain specifically recognizes the retroviral genomic RNA during RNA encapsidation.

Efficiency and selectivity of RNA packaging by Rous sarcoma virus Gag deletion mutants

In all retrovirus systems studied, the leader region of the RNA contains a cis-acting sequence called psi that is required for packaging the viral RNA genome. Since the pol and env genes are dispensable for formation of RNA-containing particles, the gag gene product must have an RNA binding domain(s) capable of recognizing psi. To gain information about which portion(s) of Gag is required for RNA packaging in the avian sarcoma and leukemia virus system, we utilized a series of gag deletion mutants that retain the ability to assemble virus-like particles. COS cells were cotransfected with these mutant DNAs plus a tester DNA containing psi, and incorporation of RNA into particles were measured by RNase protection. The efficiency of packaging was determined by normalization of the amount of psi+ RNA to the amount of Gag protein released in virus-like particles. Specificity of packaging was determined by comparisons of psi+ and psi- RNA in particles and in cells. The results indicate that much of the MA domain, much of the p10 domain, half of the CA domain, and the entire PR domain of Gag are unnecessary for efficient packaging. In addition, none of these deleted regions is needed for specific selection of the psi RNA. Deletions within the NC domain, as expected, reduce or eliminate both the efficiency and the specificity of packaging. Among mutants that retain the ability to package, a deletion within the CA domain (which includes the major homology region) is the least efficient. We also examined particles of the well-known packaging mutant SE21Q1b. The data suggest that the random RNA packaging behavior of this mutant is not due to a specific defect but rather is the result of the cumulative effect of many point mutations throughout the gag gene.

A base-paired structure in the avian sarcoma virus 5' leader is required for efficient encapsidation of RNA

Selective encapsidation of avian sarcoma-leukosis virus genomic RNA within virions requires recognition of a cis-acting signal (termed psi) located in the 5' leader of the RNA between the primer binding site and the splice donor site. Computer analyses indicate the potential for numerous secondary structure interactions within this region, including alternative conformations with similar free energy levels. We have constructed mutations designed to disrupt and restore potential secondary structure interactions within psi to investigate the role of these structures in RNA packaging. To test for the ability of psi mutants to package a heterologous reporter gene into virions, chimeric constructs bearing avian sarcoma virus 5' sequences fused to lacZ were transiently cotransfected with a nonpackageable helper construct into chicken embryo fibroblasts. lacZ virions produced from cotransfected cells were used to infect new cultures of chicken embryo fibroblasts, and then an in situ assay for individual cells expressing lacZ was done. Results obtained with this assay were confirmed in direct analyses of isolated virion RNA by RNase protection assays. Two mutations, predicted to disrupt a potential stem structure forming between elements located at nucleotides 160 to 167 and 227 to 234, severely inhibited packaging when either element was mutated. A construct in which these mutations were combined to restore potential base pairing between the two elements displayed a partially restored packaging phenotype. These results strongly suggest that the structure, referred to as the O3 stem, is required for efficient encapsidation of avian sarcoma virus RNA. Site-directed mutagenesis of additional sequence elements located in the O3 loop reduced packaging as measured by the indirect assay, suggesting that these sequences may also be components of the encapsidation signal. The possible implications of the O3 stem structure with regard to translation of avian sarcoma-leukosis virus short upstream open reading frames are discussed.

Retroviral RNA packaging: a review

In retroviruses, the "Gag" or core polyprotein is capable of assembling into virus particles and packaging the genomic RNA of the virus. How this protein recognizes viral RNA is not understood. Gag polyproteins contain a zinc-finger domain; mutants with changes in this domain assemble into virions, but a large fraction of these particles lack viral RNA. Thus, one crucial element in the RNA packaging mechanism is the zinc-finger domain. RNA sequences required for packaging ("packing signals") have been studied both by deletion analysis and by measuring encapsidation of nonviral mRNAs containing limited insertions of viral sequence. These experiments show that all or part of the packaging signal in viral RNA is located near the 5 end of the genome. These signals appear to be quite large, i.e., hundreds of nucleotides. Each virus particle actually contains a dimer of two identical, + strand genomic RNA molecules. The nature of the dimeric linkage is not understood. In some experimental situations (including zinc-finger mutants), only a small fraction of the particles in a virus preparation contain genomic RNA. It is striking that the genomic RNA packaged in these situations is dimeric. Because of this important observation, it is speculated that only dimers are packaged, and that the dimeric structure is an element of the packaging signal. It is also suggested that the dimers undergo a conformational change ("RNA maturation") after the virus is released from the cell, and that this change may depend upon the cleavage of the Gag polyprotein, a post-assembly event catalyzed by the virus-coded protease.

Alterations of the three short open reading frames in the Rous sarcoma virus leader RNA modulate viral replication and gene expression

The Rous sarcoma virus (RSV) leader RNA has three short open reading frames (ORF1 to ORF3) which are conserved in all avian sarcoma-leukosis retroviruses. Effects on virus propagation were determined following three types of alterations in the ORFs: (i) replacement of AUG initiation codons in order to prohibit ORF translation, (ii) alterations of the codon context around the AUG initiation codon to enhance translation of the normally silent ORF3, and (iii) elongation of the ORF coding sequences. Mutagenesis of the AUG codons for ORF1 and ORF2 (AUG1 and AUG2) singly or together delayed the onset of viral replication and cell transformation. In contrast, mutagenesis of AUG3 almost completely suppressed these viral activities. Mutagenesis of ORF3 to enhance its translation inhibited viral propagation. When the mutant ORF3 included an additional frameshift mutation which extended the ORF beyond the initiation site for the gag, gag-pol, and env proteins, host cells were initially transformed but died soon thereafter. Elongation of ORF1 from 7 to 62 codons led to the accumulation of transformation-defective virus with a delayed onset of replication. In contrast, viruses with elongation of ORF1 from 7 to 30 codons, ORF2 from 16 to 48 codons, or ORF3 from 9 to 64 codons, without any alterations in the AUG context, exhibited wild-type phenotypes. These results are consistent with a model that translation of the ORFs is necessary to facilitate virus production.

Evidence for interstrand quadruplex formation in the dimerization of human immunodeficiency virus 1 genomic RNA

Retroviruses package two homologous single-stranded RNA genomes within a gag protein-RNA complex. In mature virion particles, the two RNA strands are thought to associate primarily through direct RNA-RNA interactions, although the structural basis for this stable association is unknown. We now report that a 127-nucleotide (nt) HIV-1NL4-3 RNA fragment (positions 732-858) encompassing the 5' end of the gag gene dimerizes spontaneously under high ionic strength conditions in the absence of any protein cofactor. The HIV-1 RNA dimer is dramatically and specifically stabilized by the monovalent cation potassium. Thermal dissociation of the dimer occurs at 80 degrees C in 100 mM K+ (5 mM Mg2+) but at significantly lower temperatures in the presence of either smaller or larger monovalent cations (100 mM Li+, 40 degrees C; 100 mM Na+, 55 degrees C; 100 mM Cs+, 30 degrees C). Deletion analyses of the 3' end of the 127-nt fragment reveal that an HIV-1 RNA fragment as short as 94 nt (732-825) can dimerize spontaneously, but a further 9-base deletion of the purine-rich sequence, GGGGGAGAA from positions 817 through 825, eliminates dimerization. These experimental results support a model in which HIV-1 RNA dimerizes by forming an interstrand quadruple helix stabilized by guanine (and/or purine)-base tetrads in analogy to the well-known dimerization of telomeric DNA. We speculate that this structure may also mediate the association of genomic HIV-1 RNA in vivo, revealing how RNA itself can achieve the self-recognition required for subsequent genetic recombination.

Avian retroviral RNA encapsidation: reexamination of functional 5' RNA sequences and the role of nucleocapsid Cys-His motifs

RNA packaging signals (psi) from the 5' ends of murine and avian retroviral genomes have previously been shown to direct encapsidation of heterologous mRNA into the retroviral virion. The avian 5' packaging region has now been further characterized, and we have defined a 270-nucleotide sequence, A psi, which is sufficient to direct packaging of heterologous RNA. Identification of the A psi sequence suggests that several retroviral cis-acting sequences contained in psi+ (the primer binding site, the putative dimer linkage sequence, and the splice donor site) are dispensable for specific RNA encapsidation. Subgenomic env mRNA is not efficiently encapsidated into particles, even though the A psi sequence is present in this RNA. In contrast, spliced heterologous psi-containing RNA is packaged into virions as efficiently as unspliced species; thus splicing per se is not responsible for the failure of env mRNA to be encapsidated. We also found that an avian retroviral mutant deleted for both nucleocapsid Cys-His boxes retains the capacity to encapsidate RNA containing psi sequences, although this RNA is unstable and is thus difficult to detect in mature particles. Electron microscopy reveals that virions produced by this mutant lack a condensed core, which may allow the RNA to be accessible to nucleases.

Specificity of Rous sarcoma virus nucleocapsid protein in genomic RNA packaging

Site-directed mutagenesis has shown that the nucleocapsid (NC) protein of Rous sarcoma virus (RSV) is required for packaging and dimerization of viral RNA. However, it has not been possible to demonstrate, in vivo or in vitro, specific binding of viral RNA sequences by NC. To determine whether specific packaging of viral RNA is mediated by NC in vivo, we have constructed RSV mutants carrying sequences of Moloney murine leukemia virus (MoMuLV). Either the NC coding region alone, the psi RNA packaging sequence, or both the NC and psi sequences of MoMuLV were substituted for the corresponding regions of a full-length RSV clone to yield chimeric plasmid pAPrcMNC, pAPrc psi M, or pAPrcM psi M, respectively. In addition, a mutant of RSV in which the NC is completely deleted was tested as a control. Upon transfection, each of the chimeric mutants produced viral particles containing processed core proteins but were noninfectious. Thus, MoMuLV NC can replace RSV NC functionally in the assembly and release of mature virions but not in infectivity. Surprisingly, the full-deletion mutant showed a strong block in virus release, suggesting that NC is involved in virus assembly. Mutant PrcMNC packaged 50- to 100-fold less RSV RNA than did the wild type; in cotransfection experiments, MoMuLV RNA was preferentially packaged. This result suggests that the specific recognition of viral RNA during virus assembly involves, at least in part, the NC protein.