ASLV Packaging Sequences

Initial evidence for the role of specific sequences in the packaging of viral RNA came from the analysis of naturally occurring deletion mutants of ASLV such as SE21Q1b and TK15 (Linial et al. 1978; Kawai and Koyama 1984). The region deleted in SE21Q1b begins just upstream of the primer-binding site (PBS) and extends to approximately 100 nucleotides upstream of the start of the gag gene (Shank and Linial 1980; Anderson et al. 1992); the similar but smaller deletion in TK15 does not affect the PBS sequence (Kawai and Koyama 1984; Nishizawa et al. 1985). These mutants are defective in packaging genomic RNA. TK15 is able to package spliced viral RNA at low efficiency, as complementary DNA copies of v-src mRNA are found integrated after infection with this virus (Koyama et al. 1984; Nishizawa et al. 1987). Taken together, these results point to the importance of the 5′region of the genome for the efficient packaging of genomic RNA. However, the sequence missing in TK15 is not sufficient for efficient packaging as shown by the inefficient packaging of spliced RNAs from wild-type virus that retain this region (Stacey 1979; Svoboda et al. 1986). A 115-nucleotide-long noncoding sequence present near the 3′end of the genome has also been implicated as an important determinant that significantly enhances the efficiency of packaging (Sorge et al. 1983; Aronoff and Linial 1991). Recent experiments also imply a role for this region in transport of unspliced RNA from the nucleus (Chapter 6.

Sequences sufficient to direct specific packaging of ASLV RNA appear to lie largely outside of the regions encoding the viral proteins. A vector containing only 5′- and 3′-noncoding regions, along with the first 51 codons of gag and the 3′half of env, is efficiently packaged (Norton and Coffin 1985; Stoker and Bissell 1988), and a virus-host chimeric transcript with viral sequences derived from only the 320 nucleotides upstream of the splice donor site is also packaged (Aronoff and Linial 1991; Swain and Coffin 1992). A region as small as 270 nucleotides, extending from the downstream boundary of the PBS to the splice donor site (six codons past the gag initiation codon) is able to confer specific packaging of a heterologous transcript into ASLV particles with, however, reduced efficiency (Aronoff and Linial 1991; Aronoff et al. 1993). An as yet unresolved dilemma is that the sequences identified in these studies all reside on both genomic RNA and the subgenomic spliced viral mRNAs (for env and v-src), yet only the full-length RNA is efficiently packaged. It is possible that the sequences which comprise the packaging signal assume different conformations in these different contexts, although there is no information on this point.

Initial attempts have been made to implicate specific structural features of the RNA in packaging. The importance of a putative stem-loop structure that lies within the deletion-sensitive region of the 5′-noncoding sequence (Katz et al. 1986) has been examined. Mutations in either side of the stem that disrupt base pairing significantly reduce RNA packaging, and compensatory mutations to restore the stem partially restore packaging (Knight et al. 1994), suggesting both the presence of this stem structure and a role for the structure in packaging.

MLV and SNV Packaging Sequences

In the mammalian type-C viruses, a sequence required for efficient packaging lies downstream from the splice donor site, thus promoting packaging of genomic RNA but not subgenomic RNA. Deletions in the MLV and SNV genomes that remove most of the region between the splice donor site and the gag start site produce viruses that are defective for packaging their own genomes but are still able to make particles that can encapsidate other genomes (Watanabe and Temin 1982; Mann et al. 1983; Sorge et al. 1984). If the deleted sequence is moved to the U3 region of the viral genome, packaging of genomic RNA is maintained and the subgenomic env mRNA is also packaged (Mann and Baltimore 1985). An 800-nucleotide region of the MLV genome starting just downstream from the splice donor and extending into gag is sufficient to direct the efficient packaging of a heterologous transcript (Adam and Miller 1988; Dornburg and Temin 1990).

In the context of the MLV genome, a deletion within the 5′half of the U5 region blocks packaging of viral RNA (Murphy and Goff 1989), although these sequences are not needed to direct the packaging of a heterologous transcript (Adam and Miller 1988). Perhaps the U5 deletion disrupts a normal feature of secondary structure so that it now folds in a manner that is disruptive to the packaging signal. The region of the MLV genome that is sufficient to direct packaging can be divided into at least two subdomains, since viable mutants with deletions within this sequence have been made (Schwartzberg et al. 1983).

Important determinants of secondary structure have been identified within the packaging signal region. Two hairpins found between the splice donor site and the gag initiation codon are widely conserved among the MLV-like viruses (Konings et al. 1992). The loops of these two hairpins have the sequence GACG, a motif that is frequently found in four-base loops (tetraloops) and is able to form additional hydrogen bonds between bases in the loop, adding stability to the hairpin (Cheong et al. 1990; Heus and Pardi 1991). Chemical and enzymatic probing of MLV RNA has provided evidence for the existence of hairpins with the tetraloop sequence at or near the ends of the hairpins (Alford et al. 1991; Tounekti et al. 1992). Mutational analysis of the SNV double-hairpin structure also indicates the importance of these structures for packaging (Yang and Temin 1994). The secondary structures that form the tetraloops are not altered during RNA dimerization in vitro, suggesting that these structural elements may not participate in the formation of the dimer linkage structure (see below) (Tounekti et al. 1992).

HIV-1 Packaging Sequences

Deletions within the approximately 50 nucleotides between the splice donor site and the gag initiation codon reduce the incorporation of HIV-1 RNA into budding virions, although the magnitude of the reported effect varies (Lever et al. 1989; Aldovini and Young 1990; Clavel and Orenstein 1990; Luban and Goff 1994). Another region, within the first 40 nucleotides of gag, has also been found to be important for efficient RNA packaging (Luban and Goff 1994), as well as sequences in U5 (Kim et al. 1994; Vicenzi et al. 1994). HIV-1 vectors with internal deletions of most of the structural genes are packaged into virions (Poznansky et al. 1991; Shimada et al. 1991; Buchschacher and Panganiban 1992; Richardson et al. 1993). Such studies have argued for a role of at least a portion of the gag gene in enhancing the efficiency of packaging (Parolin et al. 1994). However, it has been difficult to define a sequence from the 5′region of the genome that is sufficient to direct the packaging of a heterologous sequence (Berkowitz et al. 1995b), although inclusion of a sequence extending to the 5′end of the genome is beneficial (McBride et al. 1997). Features of secondary structure that may play a part in packaging have been identified (Harrison and Lever 1992; Baudin et al. 1993; Hayashi et al. 1993; Rizvi and Panganiban 1993; Sakaguchi et al. 1993; Luban and Goff 1994; Clever et al. 1995). Mutations within two hairpins that flank the splice donor site reduce the level of RNA packaging (Berkhout and van Wamel 1996; McBride and Panganiban 1996, 1997; Paillart et al. 1996; Clever and Parslow 1997; Laughrea et al. 1997).

Other Retroviruses

The packaging signal of BLV is discontinuous. The major determinant of RNA packaging lies in the 5′-untranslated region, between the PBS and the start of gag, and a secondary determinant lies in the 5′end of gag (Mansky et al. 1995).

A 624-nucleotide-long fragment of the M-PMV genome, spanning a region from just downstream from the PBS into gag, is necessary to mediate packaging of M-PMV RNA into virions (Vile et al. 1992). A model for features of secondary structure in this region has been proposed, with elements of the structure identified as common among diverse retroviruses (Harrison et al. 1995). M-PMV particles lacking the cognate viral RNA can package RNA from other type-D retroviruses, but they do not incorporate RNA from mammalian type-C retroviruses (Vile et al. 1992). Similarly, mammalian type-C retroviruses engineered to make particles without their cognate RNA will not package RNA from a type-D retrovirus (Takeuchi et al. 1992).

Identification of Genomic RNA for Packaging

Full-length viral RNA present in the cell can be encapsidated into virions as genomic RNA or serve as message for the synthesis of the Gag and Gag-Pro-Pol precursor proteins. The partitioning of RNA for these two functions is likely to be regulated to ensure the availability of both the RNA and protein components of new viral particles. However, the basis for this presumptive regulation is unknown. Furthermore, a subset of the full-length viral RNA remains intact, but a portion of it undergoes splicing to generate subgenomic RNAs (see Chapter 6. Although spliced viral RNAs are not efficiently packaged, the act of splicing itself does not preclude packaging. Furthermore, the splice signals that generate subgenomic viral RNA are protected from the splicing machinery during the sequestration of genomic RNA; heterologous splice sites incorporated into viral RNA are not. Thus, it is possible to place unspliced versions of heterologous sequences into a retroviral vector and obtain back as virion RNA spliced versions of the inserted sequences (Shimotohno and Temin 1982; Sorge and Hughes 1982; Cepko et al. 1984).

Evidence for distinct pools of genomic RNA and of gag and gag-pro-pol mRNA comes from studies in which the RNA synthesis inhibitor actinomycin D was used to block viral RNA synthesis. Under these conditions, viral particles are released from infected cells, but these particles are devoid of genomic RNA. One interpretation of this result is that the pool of full-length viral RNA which is utilized as mRNA for the synthesis of Gag and Gag-Pro-Pol is distinct from the pool of full-length viral RNA which is packaged into virions and that translated RNA cannot be reclaimed to be packaged as virion RNA (Levin et al. 1974; Levin and Rosenak 1976). It has also been proposed that Gag competes with ribosomes in packaging versus translation (Sonstegard and Hackett 1996). In this scheme, the intracellular concentration of Gag determines the fate of genomic RNA. A clear lack of a requirement to function as gag mRNA as part of the packaging pathway is seen with retroviruses that carry cell-derived oncogenes as fusions with truncated gag genes and with retroviral vectors that can be mutated to remove any potential to serve as gag mRNA while still being efficiently packaged (Bender et al. 1987; see Chapter 9.

Packaging of RNA seems somehow to be associated with dimerization, but there is conflicting evidence whether full-length RNA exists in the form of a stable dimer before it is incorporated into virions (Méric and Spahr 1986; Stewart et al. 1990; Darlix et al. 1992). Newly budded RNA is in an immature dimer form that matures to a more stable form after budding (see below). It is possible that some type of initial dimerization occurs inside of the cell and that this event defines the pool of genomic RNA precursors. It has been argued for MLV (Fu and Rein 1993) that the presence of only dimeric RNA in virions formed under conditions of limiting viral RNA (e.g., in the presence of actinomycin D [Messer et al. 1981] or with an NC mutant [Méric and Goff 1989]) provides evidence that intracellular dimerization may precede and even direct RNA packaging. In another example, deletion of the packaging signal results in a dramatic decrease in the efficiency of packaging viral RNA, but the residual RNA that is packaged is still in a high-molecular-weight complex (Tchenio and Heidmann 1995). However, under different circumstances (see below), both monomeric and dimeric RNAs have been detected in virions. Thus, a causal relationship between dimerization and packaging has not yet been clearly established.

Viral particles are capable of packaging RNAs that are somewhat longer than the length of wild-type virus, but there are upper limits on RNA size. The genome of Rous sarcoma virus (RSV), with its v-src insert in addition to the replicative genes, is 9.3 kb in length, about 1.9 kb larger than its avian leukemia virus (ALV) parent. The examination of packaged readthrough RNAs in ASLV virions (RNAs that failed to undergo cleavage at the LTR-encoded polyadenylation signal) showed that RNAs up to 11.2 kb can be found in virus (Herman and Coffin 1987), although the infectivity of virus carrying these larger RNAs is not known. The ASLV-based vector RCAS remains replication-competent with modified genomes up to 10.5 kb in length (S.H. Hughes, pers. comm.). On the basis of the replication properties of vectors with inserts, the packaging limit of SNV has been estimated to be 10.0 kb (Gelinas and Temin 1986). Using a similar approach, it was found that the HIV-1 genome is near the packaging size limit since an insert of as little as 700 nucleotides is deleterious to viral replication (Terwilliger et al. 1989a). The capacity to package large genomes varies among retroviruses, as seen by the fact that the genome size of some retroviruses is greater than 12 kb (see Chapter 2. Although the possibility of a lower size limit for packaging has not been systematically examined, very small RNAs can be efficiently packaged—as small as 2 kb in the case of SNV (Embretson and Temin 1987) and 1.4 kb in the case of ASLV (Sorge et al. 1983). Thus, there appears to be more flexibility in packaging smaller RNAs than larger ones. Some evidence also exists that four rather than two copies of small RNAs can be packaged per virion (Sakalian et al. 1994).

Viral Proteins Involved in Packaging Genomic RNA

Genomic RNA is present in virions made with Gag as the only protein constituent (Shields et al. 1978; Sakalian et al. 1994). This implies that the gag gene product is the only protein required for the selective packaging of viral RNA into virions. Attempts to identify the role of viral proteins in the selective uptake of full-length RNA have focused on two areas: demonstration of RNA-binding domains within the Gag protein (and its processed products) and demonstration of specificity in binding to viral RNA in vitro or in the uptake of viral RNA into virions. Mutants that fail to process Gag still specifically package viral RNA (Shields et al. 1978; Crawford and Goff 1985; Oertle and Spahr 1990; Stewart et al. 1990; Fu and Rein 1993). It is likely that the initial steps of selective RNA recognition take place between the full-length Gag precursor and full-length viral RNA.

Most of the attention to RNA-binding activity has been focused on the NC protein which is bound to RNA in the virion and usually carries one or two Cys-His boxes resembling zinc fingers (Chapter 2. Evidence that the NC protein is involved in RNA packaging comes from mutagenesis studies. Mutations at a number of positions within the Cys-His box result in virus that is defective for replication. Several phenotypes have been reported for these mutants. In most cases, there is a decrease in virion-associated genomic RNA, suggesting that the NC protein and the Cys-His box are involved in RNA packaging. This phenotype has been seen with NC mutants of ASLV (Méric and Spahr 1986; Méric et al. 1988; Dupraz et al. 1990; Sakalian et al. 1994), MLV (Gorelick et al. 1988; Méric and Goff 1989; Housset et al. 1993), and HIV-1 (Aldovini and Young 1990; Gorelick et al. 1990; Dorfman et al. 1993). In two of the viruses that contain two Cys-His boxes, ASLV and HIV-1, the two sequence motifs are not equivalent in that one motif cannot replace the other (Bowles et al. 1993; Gorelick et al. 1993). Although most of the mutagenesis studies of NC have focused on the Cys-His boxes, a role for basic residues in NC in RNA packaging and/or viral infectivity has also been demonstrated (Fu et al. 1988; Housset et al. 1993; Rein et al. 1994a; Poon et al. 1996). Another phenotype is seen with a subset of NC mutants, where virion RNA levels are significantly higher than the residual infectivity (Méric and Spahr 1986; Méric et al. 1988; Méric and Goff 1989; Poon et al. 1996). These mutants provide evidence that the NC protein has a role in the replication cycle beyond packaging RNA, perhaps in annealing the tRNA primer (Prats et al. 1988) or during viral DNA synthesis (Lapadat-Tapolsky et al. 1993).

Demonstrating the basis of specificity of viral RNA selection has been difficult to approach experimentally, and the results obtained thus far do not provide a single answer to the question of the mechanistic basis of specificity. Many studies have noted the ability of NC to bind nonspecifically to RNA (see, e.g., Karpel et al. 1987; You and McHenry 1993; Khan and Giedroc 1994; and references therein). Nevertheless, specific interactions between the HIV-1 NC protein and regions of viral RNA spanning the packaging signal and the 5′end of gag have been detected with several different assays (Darlix et al. 1990; Berkowitz et al. 1993; Sakaguchi et al. 1993; Berkowitz and Goff 1994; Dannull et al. 1994). Specific interactions have also been measured between full-length Gag protein and viral RNA (Luban and Goff 1991; Berkowitz et al. 1993; Berkowitz and Goff 1994; Dannull et al. 1994; Clever et al. 1995; Geigenmuller and Linial 1996). With both Gag and NC, mutations within NC have shown the importance of both the Cys-His motifs and the basic amino acids (Berkowitz et al. 1993; Dorfman et al. 1993; Berkowitz and Goff 1994; Dannull et al. 1994; Mizuno et al. 1996). Furthermore, evidence for specific RNA packaging into virions directed by the NC protein was obtained with an ASLV mutant in which its own NC protein was replaced with the MLV NC protein, resulting in partially altered packaging specificity (Dupraz and Spahr 1992). Similar chimeras involving the placement of NC in a heterologous Gag context, in this case using HIV-1 and MLV, have shown a direct role for NC in the specificity of viral RNA selection in packaging (Berkowitz et al. 1995a; Zhang and Barklis 1995)

Other regions of the Gag precursor outside of NC have also been implicated in the specific packaging of viral RNA. A role for the MA protein in the selective binding of viral RNA has been reported for BLV (Katoh et al. 1991, 1993), although MA from ASLV has been shown to bind to RNA nonspecifically (Steeg and Vogt 1990 and references therein). The deletion of a portion of the ASLV CA domain reduces the efficiency of RNA packaging (Sakalian et al. 1994), and a role for the HIV-1 CA protein in enhancing the specificity of NC RNA binding was suggested by an in vitro assay (Geigenmuller and Linial 1996). Thus, although much evidence suggests that NC has a central role in the selective packaging of viral RNA, there are indications that other protein determinants may also be involved.

Packaging of tRNA and Other RNAs

Retroviral particles contain a variety of cellular RNAs, most of which are presumed to be packaged fortuitously during virion assembly. A specific cellular tRNA has the critical role of primer for the synthesis of the first strand of viral DNA (see Chapter 4. To accomplish this, the tRNA must be brought into the virion during assembly of the viral particle, and then the 3′end of the tRNA must be annealed to a complementary site near the 5′end of the viral genome. Virions also contain a pool of free tRNA molecules that represents a distinct subset of the tRNA pool in the cell and is enriched for the primer tRNA.

Two key observations provide insight into how the primer tRNA is enriched in the virion. First, the total tRNA pool in the virion is unchanged in particles that are either devoid of genomic RNA (Levin and Seidman 1979; Peters and Hu 1980) or contain genomic RNA deleted in the PBS sequence (Jiang et al. 1993). Thus, the primer tRNA is not directed into the virion as the result of being bound to viral RNA. Second, the free primer tRNA is not enriched in virions made by a virus with a mutation in the RT-coding domain (Sawyer and Hanafusa 1979; Peters and Hu 1980; Levin and Seidman 1981; Mak et al. 1994). This implies a role for RT in concentrating the primer tRNA in the virion. tRNAs found in the virion in the absence of a functional RT appear to represent the total population of cellular tRNAs rather than the primer-enriched subset of tRNAs characteristic of wild-type viral particles. It has also been shown that the HIV-1 Gag-Pro-Pol precursor can direct the enrichment of its tRNA^Lys3 primer into virions (Mak et al. 1994). Taken together, the evidence strongly suggests that RT, most likely as part of the Gag-Pro-Pol precursor, is responsible for bringing the primer tRNA into the virion such that it is overrepresented in the virion tRNA pool.

The absence of RT in the virion does not, however, preclude placement of the primer tRNA on the PBS sequence of genomic RNA in the virion. In the case of MLV, particles made in the absence of RT still contain tRNA^Pro annealed to the virion RNA (Levin and Seidman 1981), and this tRNA is localized to the PBS (Fu et al. 1997). It appears that this abundant cellular tRNA is fortuitously packaged at a sufficiently high rate to allow occupancy of the PBS even in the absence of RT. Thus, although RT is not required for localizing the primer tRNA to the virion or placing it on the genome, it does increase the concentration of the primer tRNA in the virion. The concentration of the primer tRNA in the virion by RT may be essential when the primer tRNA is a rare species in the cell. Two examples of this relationship can be seen with ASLV. First, ASLV can be forced to utilize an alternative tRNA by replacement of the PBS sequence; in this case, the stability of the mutant is correlated with the cellular abundance of the tRNA (Whitcomb et al. 1995). Second, in contrast to the case with MLV, the abundance of the normal primer in the absence of RT is too low to allow significant occupancy of the PBS (Fu et al. 1997).

In several cases, the basis of specificity in the selection of the primer tRNA for inclusion into the virion has been shown to reside in RT by its ability to bind the primer tRNA. In the ASLV genus, the tRNA pool in the virion is enriched for the primer tRNA, tRNA^Trp, and this tRNA binds selectively to the ASLV RT (Panet et al. 1975; Cordell et al. 1979). A specific interaction between the HIV-1 primer tRNA, tRNA^Lys3, and the HIV-1 RT has also been observed (Barat et al. 1989; Sallafranque-Andreola et al. 1989). It has been suggested that the basis of this specificity lies in part in recognition of modified bases (Barat et al. 1991), although other workers have failed to detect a specific interaction in vitro (Delahunty et al. 1994).

In contrast, no specificity has been demonstrated in the interaction between the MLV RT and the MLV tRNA primer tRNA^Pro (Haseltine et al. 1977; Panet and Berliner 1978). It is likely that the absence of a detectable interaction represents a lower affinity compared to the RT-tRNA interaction in the avian system, since the primer tRNA^Pro is still enriched in MLV particles (Peters et al. 1977; Levin and Seidman 1981). However, other tRNAs can serve as primer in the MLV system if the PBS sequence on the viral genome is altered to be complementary to another tRNA (Lund et al. 1993), suggesting that a sufficient variety of tRNAs is incorporated to allow utilization of alternative tRNA primers as dictated by the PBS sequence. Such a change in the primer appears to have occurred in nature since some of the endogenous retroviral genomes in the mouse genome have a PBS sequence complementary to tRNA^Gln (Nikbakht et al. 1985), and this PBS is functional since it can be rescued as part of a replication-competent, recombinant MLV (Colicelli and Goff 1986). HIV-1 can also utilize different tRNAs as primers when the sequence of the PBS is changed, although this leads to inefficient replication and eventual reversion to tRNA^Lys3 (Das et al. 1995; Wakefield et al. 1995). As noted above, ASLV can use different tRNAs as primer directed by a mutated PBS sequence, but these mutants revert to the wild-type tRNA^Trp (Whitcomb et al. 1995).

Other RNAs have been found to be associated with retroviral particles, although the issue of fortuitous contamination is always a concern in evaluating the significance of the observed association. Abundant RNAs like ribosomal RNAs and 7S RNA have been detected in association with virions (Bishop et al. 1970a,b). In addition, several examples of cellular mRNAs have been detected in virions. Globin mRNA has been isolated at a low level from Friend leukemia virus (Ikawa et al. 1974), and mRNAs for β-actin and GAPDH can be detected in RSV (Adkins and Hunter 1981; Aronoff and Linial 1991). Another indication that cellular RNAs can be fortuitously packaged comes from the observation that a retrovirus can mediate the transfer of a dominant selectable marker between cells via an RNA intermediate, albeit at very low efficiency (Dornburg and Temin 1988, 1990; Hajjar and Linial 1993). The packaging of subgenomic viral mRNAs may also represent the fortuitous packaging of random RNAs, except that in this case, the presence of the cis-acting replication signals (PBS, polypurine tract [PPT], and R) allows for efficient reverse transcription (Koyama et al. 1984; Svoboda et al. 1986; Nishizawa et al. 1987). It is possible that viral particles require the presence of RNA to bud and replace viral RNA with other cellular RNAs if the former is not available. However, this does not appear to be the case for virions made with NC mutants that fail to package viral RNA. In this case, virions are made but without evidence of at least one abundant cellular mRNA (actin) (Poon et al. 1996).