Show your cap or be packaged into HIV-1

AIDS is caused by HIV-1. HIV-1 is a member of the retrovirus family; like all retroviruses, the virus particle is roughly spherical and about 100 to 120 nm in diameter. It is assembled in the virus-producing cell from virus-coded proteins and of course, packages the viral genome, so that the necessary genetic information can be passed on to the next generation of viruses. Retroviruses are distinctive in that the genetic information in the virus particle is single-stranded RNA, but when they infect a new host cell, the RNA is copied into double-stranded DNA by reverse transcriptase, the virus-coded polymerase present in the particle. This viral DNA is subsequently inserted into the host chromosomal DNA, where it will remain for the lifetime of the cell. The DNA copy of the viral genetic information is then transcribed and translated by the cellular gene expression machinery (1). This scenario implies, in turn, that RNA copies of the genome have two distinct functions in the virus-producing cell. First, they are the messenger RNAs (mRNAs) for the virus-coded proteins, and second, they are the RNA molecules to be packaged into the assembling, nascent virus particles. In fact, full-length (∼9.7-kb) RNAs are the mRNA for the major building block of progeny virus particles (the “Gag” polyprotein) and for an extended protein containing both Gag and the virus-coded enzymes.

The mechanism for distinguishing these two functional classes of genomic RNA (gRNA) molecules has been under investigation for nearly 50 y. In 1974 (before the appearance of HIV), it was reported that treatment of murine leukemia virus (MLV)–infected cells with actinomycin D (a potent inhibitor of cellular and viral RNA synthesis) resulted in the production of particles that had all the characteristics of authentic MLV virions except that after brief exposure to the drug, the virions lacked gRNA and were not infectious; nevertheless, production of virions with the same protein profile as wild-type particles continued for many hours (2, 3). These observations led to the conclusion that MLV-infected cells have two distinct pools of gRNA: one that is short lived (i.e., RNA that is destined to be packaged) and the other, with a longer half-life, that serves as viral mRNA (3). However, how this distinction was achieved was not known. A recent study used a combination of sophisticated molecular and fluorescence techniques to demonstrate the existence of two distinct RNA pools in HIV-1–infected cells (4).

A paper in PNAS by Ding et al. (5) provides an unexpected, exciting clue to this long-standing problem. It is known that the selective packaging of viral RNA, rather than cellular RNAs, depends upon sequences in the first few hundred bases from the 5′ end of the RNA in the 5′-untranslated region (5′-UTR). This region consists of a large multibranched structure containing two stem loops, called “Transcription Activator” (TAR) and “Poly (A),” followed by two other complex stem-loop structures, designated “primer binding site” (PBS) and “Ψ” (6, 7) (Fig. 1A). It was noted many years ago that the gRNA within the virus particle is actually a dimer, with two full-length RNAs joined by a limited number of Watson–Crick base pairs (8–10). This base pairing centers on a short (∼6-base) palindromic sequence, called the dimer initiation sequence (DIS), within the 5′-UTR of the RNA (11, 12). More recently, it has become apparent from NMR, chemical probing, mutational, and single-molecule studies that the 5′-UTR can adopt two very different conformations. In one, the DIS bases are engaged in intramolecular base pairing and are thus unavailable for dimerization; in the other, the DIS is exposed, but the AUG sequence at which translation into proteins begins is unavailable because of intramolecular base pairing (13, 14). These two alternative conformers of the RNA are shown schematically in Fig. 1A.

Fig. 1.

The molecular signals for packaging and translation of HIV-1 gRNA. (A) Schematic representation of the two alternate HIV-1 5′-UTR structures. In the DIS-exposed conformer (Left), the DIS of one RNA is available to interact with a second DIS from another gRNA molecule. This allows formation of an RNA dimer (i.e., the RNA form that is packaged into newly assembled virions). Protein synthesis is not possible with this form since the initiation codon AUG is sequestered. In the DIS-sequestered conformer (Right), the AUG is exposed, thereby enabling gRNA to function as a single-stranded mRNA that directs the synthesis of viral proteins. Dimer formation cannot occur in this case. Color coding is as follows: red, TAR stem loop; green, Poly (A) stem loop; blue, bases from the 5′ end of gRNA that form base pairs with the AUG (orange) via long-distance interactions; gray, the multibranched structure that includes the PBS stem loop (upper left) and the Ψ^CES structure; and purple, DIS. *(DIS exposed) The 5′ terminus has a 7-methyl G cap followed by a single G. ***(DIS sequestered) There are three G residues downstream of the 7-methyl G cap. (B) Proposed cap sequestration model for distinguishing gRNA functioning as an mRNA from gRNA that forms a dimer and is packaged into virions. Upper shows an mRNA (blue) with a 5′ exposed cap (orange) and downstream coding sequences (gray) in an open reading frame (ORF). The CBP (CAP binding protein; red) captures the mRNA 5′ cap, and the complex is delivered to the ribosome (green), depicted with a growing polypeptide chain (yellow). In contrast, gRNA with a sequestered cap (Lower) evades capture by the CBP, and gRNA is able to form a dimer, which is packageable. Adapted from ref. 5, which is licensed under CC BY-NC-ND.

What controls the distribution of the full-length RNA molecules between these two conformational states? Recent data indicate that the key to this problem lies in the extreme 5′ end of the molecules. All of these RNAs have a special chemical tag at their 5′ end, referred to as a 5′ “cap.” The cap consists of a modified G (i.e., a 7-methyl G) attached to the next base in a unique 5′–5′ pyrophosphate linkage and is typically present at the 5′ ends of mammalian cell mRNAs (15). The presence of a 5′ cap in mRNAs promotes efficient protein synthesis by protecting mRNA from degradation and by facilitating attachment of mRNA to ribosomes through interaction with cellular proteins called initiation factors. Notably, it has recently been found that there is heterogeneity at this end of HIV-1 RNA. In all the molecules, the cap is attached to a G, the first base to be transcribed from the DNA. However, they differ at the exact site on the DNA at which transcription begins, so that some have only a single G, while others have two or three G’s downstream of the cap. For the most part, when HIV-1 gRNA functions as an mRNA, its cap is followed by three (or to a lesser extent, two) G’s [^(Cap)3G or ^(Cap)2G] (16). In contrast, some gRNA molecules have only one G downstream of the cap; these tend to be packaged into virus particles [^(Cap)1G] (16, 17).

In earlier efforts to more precisely define the 5′-UTR elements needed for packaging, it was found that an RNA 5′-UTR construct that was missing the TAR and Poly (A) stem loops and had a short GAGA tetraloop substituted for the PBS stem loop had the same activity in a packaging assay as the entire 5′-UTR structure (Fig. 1A). This short construct was named the “core encapsidation signal” (Ψ^CES) and was thought to be the minimal structure required for gRNA packaging (6, 7). However, in the PNAS paper, Ding et al. (5) used a more stringent packaging assay system that involves transfection of the complete, wild-type 5′-UTR construct together with the variant being tested. Thus, this assay measures the ability of the variant to successfully compete with the intact packaging signal for incorporation into the assembling virus particle. To their surprise, utilizing this assay, the authors found that the Ψ^CES construct had poor activity, suggesting that other elements in the 5′-UTR must also be required. In subsequent experiments, they showed that the GAGA tetraloop substitution for the PBS had no effect on activity, whereas they identified the TAR and Poly (A) stem loops as essential elements for packaging. Structural studies were critical to understanding this result. In fact, the TAR and Poly (A) stem loops are stacked coaxially in the ^(Cap)1G structure (5, 18, 19). This structural arrangement leads to the sequestration of the 5′ cap (5, 18), so that it cannot bind to a known translational initiation factor (eukaryotic translation initiation factor 4E [eIF4E]). Conversely, when the cap is exposed, as in the ^(Cap)2G or ^(Cap)3G forms, it can be captured by the translation machinery and can bind eIF4E. In this configuration, the RNA functions as an mRNA. A schematic diagram illustrating these results is shown in Fig. 1B.

Further experimental alterations of the extreme 5′ end of the RNA provided striking evidence in support of this model (5). First, the authors inserted four bases (AAGG) between the 5′-terminal G and the next base in the RNA. When this RNA is expressed in mammalian cells, it is capped; the inserted four-base linker separates the cap from the TAR stem loop. As this RNA can bind eIF4E, it is clear that its cap is exposed. While it dimerizes like the wild-type RNA in vitro and binds Gag protein like wild-type RNA, it is not packaged into virus particles. In another informative modification of the viral RNA, the authors added a 5′ extension to the Ψ^CES that has self-cleaving activity. When this RNA is expressed in cells, it is undoubtedly capped at its 5′ end, but this end, including the cap, is removed by self-cleavage. The resulting RNA, beginning with an uncapped Ψ^CES, is packaged. In contrast, a mutant that does not self-cleave is not packaged. Both of these experiments show that the presence of the cap at the 5′ end of the RNA—if it is exposed rather than sequestered—interferes with packaging.

In summary, HIV-1 RNA packaging appears to depend on a bipartite mechanism, involving both the entire Ψ^CES, as concluded earlier (6, 7), and also sequestration of the cap at the 5′ end of the RNA by the two stacked 5′ stem loops (Fig. 1B). Other data suggest that binding of the viral Gag protein to the Ψ^CES nucleates particle assembly with particularly high efficiency, explaining the selective packaging of gRNA (20). However, if the RNA has an exposed cap, this nucleation does not occur; rather, it will be used as an mRNA and will not be packaged.

With the current focus of genetic research on RNA transcriptomes, RNA structure–function studies have become an important area of contemporary molecular biology. The elegant paper by Ding et al. (5) fits this paradigm and advances our understanding of the retroviral packaging vs. translation conundrum. It will be of interest to see if other retroviruses use the capping mechanism solution evolved by HIV-1.