594 NATURE | VOL 418 | 8 AUGUST 2002 | www.nature.com/nature

Nowadays, researchers generally suc-ceed in working out what viral proteinsdo within a few years of their discovery.

This is certainly the case for nearly all proteins from the type 1 human immuno-deficiency virus (HIV-1), whose genome is arguably the most extensively studied9,700 or so bases of genetic sequence on the planet. Nonetheless, one of the HIV-1 proteins has remained an enigma wrappedin a riddle since its discovery in the mid-1980s (refs 1, 2). From the report by Sheehyand colleagues3 on page 646 of this issue, it seems that the mystery has been solved, at least in part.

The protein in question is called Vif, for‘virion infectivity factor’: it is an accessoryprotein in HIV-1 that is also found in allother primate immunodeficiency viruses.Viral accessory proteins are generally notneeded for replication and survival; they areneither structural nor regulatory. But theyare needed under certain ‘stressful’ condi-

tions imposed on the virus by its cellularhome. Vif seems to be largely unique amongHIV-1’s accessory proteins. For example, ifcertain human immune cells — T lympho-cytes, monocytes or macrophages, the mainreservoirs for HIV-1 in vivo — are infected invitro with HIV-1 strains carrying mutant Vif,those strains produce progeny that are, to allintents and purposes, ‘dead’4. By contrast,mutation of other HIV-1 accessory proteins,such as Nef, Vpr and Vpu, leads to debilitatedbut still reproducing viruses.

It turns out that HIV-1 with a mutant Vifprotein can produce infectious virions —viral particles — from only a few selecthuman cell lines (‘producer’ cell lines)5.When infected with Vif-deficient viruses,certain producer cells, described as ‘non-permissive’, yield defective progeny virusesthat cannot infect target cells and in whichreverse transcriptase, a crucial viral enzyme,is largely inactive4,6,7. But infectious, replicat-ing virions can be produced in the absence of

vegetation changes before a new equilibriumis approached, and so Jackson et al. 3 choseshrublands that were at least 30 years old. Thelower soil-carbon stocks in shrublands thatthey measured at the wet end of the rainfallgradient indicate that inputs from woodyinvaders during the past 30–100 years wereless than losses of old soil carbon from thegrasses that previously occupied the site. Thisprocess is presumably continuing on landswhere woody encroachment is more recent.

On the dry end of the gradient, Jackson etal. found little change in soil carbon follow-ing woody encroachment, suggesting thatthere was little change in production belowground. This explanation is speculative, asJackson et al.did not directly measure inputsand losses of soil carbon, a very difficult task. Unfortunately, the authors also did notmeasure carbon stocks in woody roots (thoseof 1 cm or more in diameter), and so the sinkin deeply rooted woody shrubs is probablyan underestimate.

Woodlands, savannas, shrublands andgrasslands cover about 40% of the Earth’ssurface10, and so their potential role as car-bon sinks — or sources — is a key factor inthe global carbon budget. Measuring theeffects of woody encroachment at particularsites is one challenge; extrapolating theresults to regional or larger scales is quiteanother. Particular sites are certainly largesinks for carbon11, but the global extent ofgrassland replacement by shrubland is high-ly uncertain3,4. For a few regions, analyses ofhistorical aerial photographs or satellite datahave produced estimates for the extent ofencroachment (ref. 12; G. P. Asner, personalcommunication). Yet the results of Jackson etal.3 complicate these assessments by showingthat increases in plant biomass above groundmay be more than offset by losses of carbonfrom soil. This cancelling of gains in carbonabove ground with losses below ground hasalready been demonstrated for individualsites13, but we now have evidence of a region-al phenomenon that appears to vary some-what predictably along a climatic gradient.

What are needed are more reliableregional estimates. That will require com-prehensive assessments of the extent towhich shrubs are replacing grassland, alongwith field measurements and model simula-tions of the size and variability of plant andsoil carbon stocks across a range of climateconditions. ■

Christine L. Goodale and Eric A. Davidson are atThe Woods Hole Research Center, PO Box 296,Woods Hole, Massachusetts 02543, USA. e-mail: cgoodale@whrc.org1. Prentice, I. C. et al. in Climate Change 2001: The Scientific Basis

(eds Houghton, J. T. & Yihui, D.) 183–237 (Cambridge Univ.

Press, 2001).

2. Schimel, D. S. et al. Nature 414, 169–172 (2001).

3. Jackson, R. B., Banner, J. L, Jobbágy, E. G., Pockman, W. T. &

Wall, D. H. Nature 418, 623–626 (2002).

4. Pacala, S. W. et al. Science 292, 2316–2320 (2001).

5. Houghton, R. A. et al. Science 285, 574–578 (1999).

6. Van Auken, O. Annu. Rev. Ecol. Syst. 31, 197–215 (2000).

news and views

HIV

A tough viral nut to crackRoger J. Pomerantz

HIV cannot multiply in certain human cells unless it expresses a proteincalled Vif, the function of which has finally been revealed. It seems that itovercomes a human protein that would otherwise block viral replication.

Figure 1 How HIV-1 can overcome our cells’ defences. The crucial HIV-1 protein here is Vif. a, Normal, infectious HIV-1 can be produced in most human cell types (‘producer’ cell lines) and can infect and reproduce in most target cells. b, HIV-1 particles that lack Vif can produce infectiousprogeny only in ‘permissive’ cell types. Non-permissive cells produce non-infectious progeny. Sheehyet al.3 have found that non-permissive cell types express the protein CEM15, which thereforepresumably blocks the production of new, infectious viral particles from Vif-deficient HIV-1. Byextension, the Vif protein must be required to overcome CEM15, explaining why normal HIV-1 canproduce infectious viral particles in non-permissive cells (a). How Vif suppresses CEM15 is not known.

Vif-mutant HIV-1 (no Vif protein)

Producer Target Producer Target

CEM15 CEM15X

HIV-1 virion

Non

-per

mis

sive

Per

mis

sive

Wild-type HIV-1 a b

11.Tilman, D. et al. Ecology 81, 2680–2685 (2000).

12.Archer, S. et al. in Global Biogeochemical Cycles in the Climate

System (eds Schulze, E.-D. et al.) 115–138 (Academic,

San Diego, 2001).

13.Schlesinger, W. H. & Pilmanis, A. M. Biogeochemistry 42,

169–187 (1998).

7. Guo, L. B. & Gifford, R. M. Global Change Biol. 8, 345–360

(2002).

8. Sims, P. L. & Singh, J. H. J. Ecol. 66, 573–597 (1978).

9. Jackson, R. B. et al. Proc. Natl Acad. Sci. USA 94, 7362–7366

(1997).

10.DeFries, R. S. et al. Int. J. Remote Sensing 19, 3141–3168 (1998).

© 2002 Nature Publishing Group

Vif in ‘permissive’ cell lines. So the infectivityof Vif-deficient HIV-1 seems to depend onthe producer cell type (Fig. 1). Wild-typeHIV-1 with an intact Vif gene is not limitedin this way.

Over the past decade and longer, nu-merous laboratories have tried to explainthese puzzling observations. Studies haveshown8–10 that Vif binds to the genomic RNAof HIV-1. But it is unclear how this mightlead to the unique characteristics of Vif-defi-cient viruses. There is clearly some differencebetween the permissive and non-permissiveproducer cells, and a breakthrough came in 1998 when two groups11,12 proposed thatthere is a factor, present in some human cellsbut not in others, that can inhibit the replica-tion of the mutant HIV-1 but is overcome bythe Vif protein of the unmutated virus. Thiscellular factor seemed to alter Vif-deficientHIV-1 during late stages of its life cycle innon-permissive producer cells11,12. Sheehy etal.3 now have strong evidence that there isindeed a human protein that inhibits HIV-1but whose effects are suppressed by Vif.

Sheehy et al. looked at two geneticallyrelated cell lines, one permissive and onenon-permissive. Using a novel combinationof molecular-biological approaches, theauthors identified a protein — namedCEM15, after the cell line in which it was dis-covered — that was needed for the non-permissive cells to repress the infectivity ofthe Vif-deficient HIV-1. Sheehy et al. havedone a superb job of showing that CEM15 ispresent in all non-permissive cells but not inpermissive cells. They also expressed CEM15in normally permissive cells and found thatthey became non-permissive to the Vif-defi-cient HIV-1 (but not to the wild-type virus).

So there is an unarguable correlationbetween the production of CEM15 and thefailure of Vif-deficient HIV-1 particles toproduce infectious progeny in non-permis-sive cells (Fig. 1). Permissive cells, by con-trast, do not produce CEM15 and thereforeallow the Vif-deficient HIV-1 to replicatefreely. All of this suggests that Vif is needed to overcome the human CEM15 protein and allow effective viral replication in non-permissive human cells. The human proteinseems to represent one host mechanism —perhaps one of many13 — that can inhibit thereplication of retroviruses such as HIV-1.

This paper3 takes us a giant leap forwardin exploring one of the final frontiers of HIV-1’s molecular biology. But it’s not yet clear how CEM15 blocks the replicationof Vif-deficient HIV-1. Nor is it knownwhether this protein binds directly to Vifinside human cells infected with wild-typeHIV-1, or whether Vif operates in anotherway. Perhaps one clue lies in CEM15’s similarity to two other human proteins:APOBEC-1 (the catalytic subunit of anenzyme that ‘edits’ messenger RNA14) andphorbolin-1 (a protein induced by phorbol

news and views

NATURE | VOL 418 | 8 AUGUST 2002 | www.nature.com/nature 595

Obesity

Keeping hunger at bayMichael W. Schwartz and Gregory J. Morton

Many different hormones control our weight and appetite. The discovery ofanother hormone, which suppresses appetite for up to 12 hours, may leadto a better understanding of this complex control system.

esters — compounds that stimulate manyhuman cells). All of these proteins have azinc-coordinating motif, which is importantin cytidine deaminase enzymes (includingAPOBEC-1) in virtually all organisms. It will be interesting to see whether CEM15 isinvolved in RNA editing, especially as Vifclearly binds to viral genomic RNA8–10.

One implication of Sheehy and col-leagues’ results is that our cells have a meansof silencing HIV-1 in some situations, and at a specific stage of the viral life cycle. Thismight not be the only cellular virus-silencingtool. For instance, inactive human T lym-phocytes expressing the CD4 protein (whichHIV-1 uses to latch onto cells) are a key HIV-1 reservoir in vivo, yet are difficult toinfect with the virus in vitro. These cells also have several methods of decreasing thestability and activity of the virus after it has entered host cells but before its genomeintegrates with that of its host. The results oftwo recent studies of ‘RNA interference’ alsosuggest that this mechanism, which inhibitsgene expression, might be used by certainhost cells to decrease HIV-1 infection15–17.

Another implication is that the HIV-1 Vifprotein and its interactions with CEM15 —whether direct or indirect — provide us witha unique drug target. It might be possible todevelop small-molecule drugs that target Vifin the major cell types in which HIV-1 lurksin vivo. Another approach could be to inhibitthe aggregation of Vif proteins into multi-mers18, a process that is required for Vif tofunction in non-permissive cells. Under-standing how CEM15 inhibits HIV-1 repli-cation, and how this inhibition is overcomeby Vif, will be crucial in designing drugs tocombat this viral accessory protein.

The work of Sheehy et al.3 shows that even

the hardest viral nut can be cracked by combining the perseverance of molecularvirology laboratories with new moleculartechniques. Nevertheless, the story will sure-ly not end here. I predict that this paper willspur the hunt for cellular antiviral moleculesand will lead to a heightened understandingof the molecular mechanisms of HIV-1. Italso reveals an HIV-1 accessory protein thatis a good target for future antiviral drugs —more of which are urgently needed in bothdeveloped and developing countries. ■

Roger J. Pomerantz is at the Center for HumanVirology, Thomas Jefferson University, 1020 Locust Street, Suite 329, Philadelphia,Pennsylvania 19107, USA.e-mail: roger.j.pomerantz@mail.tju.edu

1. Fisher, A. G. et al. Science 237, 888–893 (1987).

2. Strebel, K. et al. Nature 328, 728–773 (1987).

3. Sheehy, A. M., Gaddis, N. C., Choi, J. D. & Malim, M. H. Nature

418, 646–650 (2002); advance online publication, 14 July 2002

(doi:10.1038/nature00939).

4. von Schwedler, U., Song, J., Aiken, C. & Trono, D. J. Virol. 67,

4945–4955 (1993).

5. Gabuzda, D. H. et al. J. Virol. 66, 6489–6495 (1992).

6. Goncalves, J., Korin, Y., Zack, J. & Gabuzda, D. J. Virol. 70,

8701–8709 (1996).

7. Dornadula, G., Yang, S., Pomerantz, R. J. & Zhang, H. J. Virol.

74, 2594–2602 (2000).

8. Zhang, H., Pomerantz, R. J., Dornadula, G. & Sun, Y. J. Virol.

74, 8252–8261 (2000).

9. Khan, M. A. et al. J. Virol. 75, 7252–7265 (2001).

10.Dettenhofer, M., Cen, S., Carlson, B. A., Kleinum, L. & Yu, X.-F.

J. Virol. 74, 8938–8945 (2000).

11.Madani, N. & Kabat, D. J. Virol. 72, 10251–10255 (1998).

12.Simon, J. H. M., Gaddis, N. C., Fouchier, R. A. M. & Malim,

M. H. Nature Med. 4, 1397–1400 (1998).

13.Pryciak, P. M. & Varmus, H. E. J. Virol. 66, 5959–5966 (1992).

14.Bhattacharya, S., Navaratnam, N., Morrison, J. R., Scott, J. &

Taylor, W. R. Trends Biochem. Sci. 19, 105–106 (1994).

15.Novina, C. D. et al. Nature Med. 8, 681–686 (2002).

16. Jacque, J.-M., Triques, K. & Stevenson, M. Nature 418, 435–438

(2002); advance online publication, 26 June 2002

(doi:10.1038/nature00896).

17.Pomerantz, R. J. Nature Med. 8, 659–660 (2002).

18.Yang, S. C., Sun, Y. & Zhang, H. J. Biol. Chem. 276, 4889–4893

(2001).

Spurred on both by a series of remark-able discoveries and by the emergenceof obesity as a world health problem,

research into how our bodies control ourappetite and weight continues at an unpara-lleled pace. The result is an increasingly clearinsight into the workings of a fascinating but complex neuroendocrine system, inwhich circulating hormones convey infor-mation about energy balance (the differencebetween energy intake and expenditure) tobrain pathways that control eating and energy output. The process of discovery hasbeen rather like peeling back the layers of an

onion, as new molecules involved in the sys-tem continue to be uncovered — although insome instances it is not the molecule itselfbut its role in controlling food intake that isdiscovered. Such is the case for the hormonepeptide YY3-36 (PYY3-36), as Batterham andcolleagues1 report on page 650 of this issue.

Hormones that regulate food intake can be separated into those that act rapidly to influence individual meals, and those that act more slowly to promote the stabilityof body fat stores. Long-term regulatorsinclude insulin and leptin, which are releasedinto the blood in proportion to the amount

© 2002 Nature Publishing Group