News & Views | Published:

Human immunodeficiency virus

Nuclear RNA export unwound

The ways in which HIV can subvert cellular processes for its own ends seem boundless. The latest discovery — a cellular enzyme that helps to export HIV RNA from the nucleus — reveals a possible drug target.

Gene expression is a multi-step process, the first stage of which is the production of a messenger RNA transcript of a gene. That mRNA is then used as a template to produce a protein. Here, cells whose genetic material is encased in a nucleus (eukaryotic cells) face a problem: their genes are transcribed in the nucleus but proteins are made outside it, in the cytoplasm, so the mRNA must be exported.

The same stricture applies to HIV, whose genes, once incorporated into the host genetic material, are likewise transcribed in the nucleus. It is known that the viral protein Rev recruits the cellular export factor Crm1 to export several essential HIV-1 mRNAs1. Host mRNAs, by contrast, generally rely on another export factor, the Tap–Nxt1 dimer. But this is not the only requirement — a ‘remodelling’ enzyme is also needed2. For cellular mRNAs, that enzyme is thought3,4,5 to be Dbp5, but this has no role in mRNA export mediated by Rev–Crm1. Writing in Cell, however, Yedavalli et al.6 suggest that the necessary enzyme is a member of the same family.

During transcription and processing, eukaryotic mRNAs are loaded with a wide variety of nuclear proteins that regulate the export, cytoplasmic localization, translation and stability of the mRNAs2. Although some of these RNA-binding proteins work by remaining associated, at least transiently, with newly exported mRNAs, many others dissociate during or immediately after export. For most cellular mRNAs, this remodelling step is thought to be mediated, at least in part, by Dbp5, an RNA helicase belonging to the ubiquitous DEAD-box protein family3,4,5.

DEAD-box helicases use the energy released from the hydrolysis of adenosine triphosphate (ATP) to unwind RNA structures and, perhaps more importantly in this context, to dissociate RNA–protein complexes7,8. These enzymes each bear nine conserved amino-acid motifs, including the eponymous DEAD box itself — named after the abbreviations for the amino acids aspartic acid, glutamic acid and alanine — and computer analysis has used these motifs to identify numerous family members in all genomes analysed. The proteins have been proposed to participate in almost every process that involves RNA, including the splicing of precursor mRNAs, the production of the protein-synthesis machinery (ribosomes), mRNA translation, and of course nuclear mRNA export7.

Dbp5 was first genetically identified in yeast as a primarily cytoplasmic protein, enriched around the nuclear rim, that is essential for the export of cellular mRNAs from the nucleus3,4. Subsequent analysis5 of the human protein revealed a specific association with the cytoplasmic side of nuclear pore complexes (NPCs), the large protein assemblies through which molecules go in and out of the nucleus.

More recently, evidence obtained in insect cells has revealed that Dbp5 actually shuttles between the nucleus and cytoplasm and, at least in part, assembles onto mRNA–protein complexes (mRNPs) as the mRNAs are being synthesized9. Importantly, microinjection of frog eggs with a mutant form of Dbp5 that could no longer hydrolyse ATP blocked the export of cellular mRNAs but did not affect mRNA export mediated by HIV-1 Rev and its cellular cofactor Crm1 (ref. 5). Therefore, whereas export of cellular mRNAs by the canonical Tap–Nxt1 pathway depends on Dbp5, the export of very similar viral mRNA substrates through the Rev–Crm1 pathway does not.

This finding suggested two possibilities: either nuclear export mediated by Rev–Crm1 doesn't require mRNP remodelling, or it depends on a different RNA helicase. The second hypothesis has now been validated by Yedavalli and colleagues6. These authors show that overexpressing another member of the DEAD-box family, DDX3, in cultured human cells specifically enhances Rev–Crm1-dependent, but not Tap–Nxt1-dependent, nuclear mRNA export. Inhibiting DDX3 function selectively impairs the former process. Moreover, DDX3 interacts specifically with both Crm1 and Rev in vivo. Interestingly, DDX3, like Dbp5, is a nucleocytoplasmic shuttle protein that, at steady state, is predominantly localized at the cytoplasmic face of NPCs. In total, the data indicate that these two helicases, although specific for distinct RNA-export pathways, probably have a similar role in remodelling mRNPs during export (Fig. 1).

Figure 1: Escape from the nucleus: how cellular and HIV-1 messenger RNAs are exported.

a, Transcription and processing of cellular mRNAs give rise to export-competent mRNA–protein (mRNP) complexes that include a range of proteins, including the export factors Tap and Nxt1 and an RNA helicase, Dbp5. During transport through nuclear pore complexes (NPCs), the mRNP is remodelled by pre-bound and/or NPC-associated Dbp5, releasing a range of factors that then return to the nucleus. The string of A's represents the polyadenosine tail characteristic of mRNAs. b, HIV-1 mRNAs are bound by a distinct but overlapping set of nuclear proteins that includes the export factors Rev and Crm1 and, as the findings of Yedavalli et al. now suggest6, the helicase DDX3. This mRNP is also remodelled during export, in this case by pre-bound and/or NPC-associated DDX3. Ribosomes then translate the cellular and HIV-1 mRNAs into protein.

So what does DDX3, a cellular enzyme, do in uninfected cells? That remains unclear at present. Although Crm1 is essential for the export of HIV-1 mRNAs, it is not known whether it participates in the export of any cellular mRNAs. It does, however, have a well-established role in exporting proteins — as well as several non-protein-coding RNAs, including ribosomal RNAs1 — from the nucleus. Yedavalli et al. present limited data suggesting that DDX3 is not required for Crm1-mediated protein export. But it might be involved in exporting certain cellular coding or non-coding RNAs.

This issue is important, because Yedavalli et al.6 also present convincing evidence that DDX3 has an essential, and possibly rate-limiting, role in the HIV-1 life cycle. Thus they suggest that it might represent a new target for chemotherapeutic intervention. However, as cells undoubtedly did not evolve DDX3 solely for the convenience of this virus, it remains possible that inhibiting it could have unforeseen and harmful consequences. It would therefore seem to be a priority to knock down DDX3 expression in infected and uninfected T cells, the major cell type affected by HIV-1, as an initial test of the hypothesis that this helicase is indeed a worthwhile target for antiviral drug development.


  1. 1

    Cullen, B. R. J. Cell Sci. 116, 587–597 (2003).

  2. 2

    Dreyfuss, G., Kim, V. N. & Kataoka, N. Nature Rev. Mol. Cell Biol. 3, 195–205 (2002).

  3. 3

    Snay-Hodge, C. A., Colot, H. V., Goldstein, A. L. & Cole, C. N. EMBO J. 17, 2663–2676 (1998).

  4. 4

    Tseng, S. S. -I. et al. EMBO J. 17, 2651–2662 (1998).

  5. 5

    Schmitt, C. et al. EMBO J. 18, 4332–4347 (1999).

  6. 6

    Yedavalli, V. S. R. K., Neuveut, C., Chi, Y. -h., Kleiman, L. & Jeang, K. -T. Cell 119, 381–392 (2004).

  7. 7

    Rocak, S. & Linder, P. Nature Rev. Mol. Cell Biol. 5, 232–241 (2004).

  8. 8

    Schwer, B. Nature Struct. Biol. 8, 113–116 (2001).

  9. 9

    Zhao, J., Jin, S. -B., Björkroth, B., Wieslander, L. & Daneholt, B. EMBO J. 21, 1177–1187 (2002).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.