A N D R E A L . C O X & R O B E R T F. S I L I C I A N O

The first paper to describe AIDS reported that patients had very few CD4+ T cells in their blood1. Depletion of this

crucial subset of immune cells is now known to be a key feature of the disease, but the mechanisms responsible for their loss have remained unclear. Particularly mysterious has been the observation that HIV­1 infection

results not only in the death of activated, productively infected CD4+ T cells (those in which the virus successfully replicates) but also in ‘bystander’ CD4+ T cells that do not seem to be infected. On page 509 of this issue, Doitsh et al.2 show that most CD4+ T cells depleted during HIV­1 infection are abortively infected cells that die through pyroptosis — a cell­death mechanism that is distinct from apoptosis and necroptosis3.

HIV­1 replication in productively infected CD4+ T cells kills them quickly, within one to two days4,5. This direct killing is apparent during acute infection, when virus levels are high and massive depletion of CD4+ T cells occurs in the gastrointestinal tract6. However, in the absence of treatment, most of the CD4+ T­cell loss associated with the infection occurs during the prolonged asymptomatic phase between the acute stage and the development of AIDS. During this period, the number of activated, productively infected CD4+ T cells is low, suggesting that the infection may promote death of quiescent (non­activated) cells.

Levels of immune activation are high in untreated HIV­1 infection, perhaps reflecting the translocation of microbial products across a compromised gastrointestinal barrier7, and it is commonly assumed that this immune activation is responsible for CD4+ T­cell loss. Perhaps the best evidence for this comes from studies of simian immunodeficiency virus infections, in which there is high virus repli­cation, but little immune activation or CD4+

that, in contrast to the tropical Pacific, north Atlantic SSTs have warmed significantly since 1979. The authors demonstrate that warmer Atlantic SSTs drive anomalous Southern Ocean winds that are consistent with the observed regional trends in Antarctic ice extent. Although forcing from the tropical Pacific dominates the variability of Antarctic winds and sea ice on interannual timescales, Atlantic forcing becomes important on decadal and longer timescales, in which Pacific SST variability is smaller.

By establishing a chain of attribution linking warming of the tropical and north Atlantic with trends in Antarctic atmos­pheric circulation and sea ice, Li and colleagues’ work helps to resolve the paradox of growing Antarctic sea­ice extent over a period when global mean temperature has increased. The researchers have also dem­onstrated that global climate models can simulate the con­nection between Atlantic SSTs and Antarctic winds. Why, then, have climate models such as those used in last year’s fifth assessment report by the Intergovernmental Panel on Climate Change been unable to reproduce the observed regional pattern of change in Ant­arctic sea ice8?

Two reasons suggest themselves. First, the recent warming of the Atlantic is the result of a combination of anthropogenic forcing

and natural internal variability of the climate system7. Only the effects of the former can be predicted in a deterministic way by climate models, with natural variability appearing as ‘noise’ in the climate­model simulations. Second, sea ice is one of the most challenging elements of the Earth system to model. The rate at which it forms or melts is controlled by the small difference between large fluxes of heat from the atmosphere and the ocean,

and its distribution is strongly influenced by winds and ocean currents. Small biases in the models’ representation of the atmosphere or ocean can thus translate into large errors in modelled sea ice.

Although accurate modelling of Antarctic sea­ice trends will require a realistic representa­tion of the processes connect­ing Atlantic SSTs and Antarctic winds, this might not be suffi­cient. Given the importance of Antarctic sea ice to the South­ern Ocean marine eco system, and its role in driving global ocean circulation by the pro­duction of ocean bottom water, understanding its behaviour and improving its representa­tion in climate models must remain a high priority for climate scientists. ■

John King is at the British Antarctic Survey, High Cross, Cambridge CB3 0ET, UK.e-mail: jcki@bas.ac.uk

1. Stroeve, J. C. et al. Clim. Change 110, 1005–1027 (2012).

2. Turner, J. et al. Geophys. Res. Lett. 36, L08502 (2009).3. Li, X., Holland, D. M., Gerber, E. P. & Yoo, C. Nature

505, 538–542 (2014).4. Yuan, X. & Martinson, D. G. Geophys. Res. Lett. 28,

3609–3612 (2001).5. Turner, J. Int. J. Climatol. 24, 1–31 (2004).6. Holland, P. R. & Kwok, R. Nature Geosci. 5, 872–875

(2012).7. Ting, M., Kushnir, Y., Seager, R. & Li, C. J. Clim. 22,

1469–1481 (2009).8. Turner, J., Bracegirdle, T. J., Phillips, T., Marshall, G. J.

& Hosking, J. S. J. Clim. 26, 1473–1484 (2013).

30° W 30° E

150° W

60° E

120° W

150° E

60° W

120°

E

90°

W 90° E

180°

0°

−0.09

−0.07

−0.05

−0.03

−0.01

0.01

0.03

0.05

0.07

0.09

Tren

d

BS

AS

RS

80° S

70° S

60° S

Figure 1 | Trend in Antarctic sea-ice coverage. The trend is expressed as the change in fractional ice coverage per decade and is calculated for the period 1979–2012. The bold lines enclose areas where the change is statistically significant at the 5% level. Ice cover has declined in the Bellingshausen Sea (BS), but compensating increases in the western Ross Sea (RS) have led to a slight overall increase in cover. These trends in ice cover are consistent with changes in winds driven by a deepening of the climatological low­pressure centre over the Amundsen Sea (AS). Li et al.3 suggest that the changes in ice cover and winds have been caused by increased temperatures in the tropical and north Atlantic. (Data: National Snow and Ice Data Center, Boulder, Colorado. Image: British Antarctic Survey.)

H I V

Not-so-innocent bystandersThe discovery that most CD4+ T cells killed during HIV infection die through a process known as pyroptosis may provide long-sought explanations for HIV-associated T-cell depletion and inflammation. See Article p.509

4 9 2 | N A T U R E | V O L 5 0 5 | 2 3 J A N U A R Y 2 0 1 4

NEWS & VIEWSRESEARCH

© 2014 Macmillan Publishers Limited. All rights reserved

T­cell depletion8. Nevertheless, the mechanis­tic link between immune activation and CD4+ T­cell depletion has remained unclear.

Doitsh and colleagues suggest that this link may lie in the manner of cell death. Using cul­tures of human cells isolated from the spleen or tonsils, they demonstrate that more than 95% of CD4+ T cells that die following HIV­1 infec­tion are quiescent cells that undergo pyrop­tosis. Only a small proportion of the dying cells were activated, productively infected CD4+ T cells undergoing apoptosis (Fig. 1). Apoptosis depends on the activation of the cell­signalling molecule caspase­3, whereas pyroptosis is triggered by inflammasome­activated caspase­1. Inflammasomes are multi­protein cytoplasmic complexes that integrate pathogen­triggered signalling pathways and then recruit and activate inflammatory cas­pase molecules. Pyroptosis results in lysis of the cell and release of the cytoplasmic con­tents into the extracellular space, and is highly inflammatory.

Productive HIV­1 infection involves the virus binding to the T­cell surface and enter­ing the cell. There, the viral RNA is reverse transcribed to DNA and integrated into the host­cell genome, resulting in replication of the virus. If this process is aborted before integra­tion and viral replication occur, the infection is termed non­productive. Doitsh and colleagues previously demonstrated9 that there is selective depletion of CD4+ T cells in which incomplete viral DNA transcripts accumulate following abortive infection. The same research group also recently identified interferon­γ­inducible protein 16 (IFI16) as the host­cell DNA sensor that triggers this cell death10.

To verify that most CD4+ T­cell depletion occurring during HIV­1 infection is medi­ated by pyroptosis, the authors treated cells with inhibitors of caspase­3 or caspase­6 (important in apoptosis), or of receptor­inter­acting protein kinase enzymes (important in necroptosis), and found that these treatments did not prevent most of the CD4+ T­cell loss. Also consistent with pyroptosis and the associ­ated release of intracellular contents into the extracellular milieu was the presence of the cytoplasmic enzyme lactate dehydrogenase in the cell­culture supernatants. In vivo evi­dence for pyroptosis came from the detection of caspase­1 in quiescent CD4+ T cells in the paracortical zone that surrounds the region of activated CD4+ T and B cells in HIV­1­infected lymph­node tissues. The authors did not detect caspase­1 in the zone of activated CD4+ T cells or in uninfected tissue.

Caspase­1 activation is known11,12 to induce secretion of the highly inflammatory cytokine proteins interleukin­1β (IL­1β) and IL­18, which contribute to inflammatory condi­tions such as atherosclerosis and metabolic syndromes11–14. Doitsh and colleagues show that IL­1β release also occurs after infection with HIV­1, and that this requires caspase­1

activation (Fig. 1). Finally, the authors show that pyroptosis induced by HIV­1 can be pre­vented with VX­765, a caspase­1 inhibitor that has previously been tested in people with chronic epilepsy and psoriasis, and found to be safe and well tolerated. VX­765 treatment inhibited caspase­1 activation, IL­1β secretion and CD4+ T­cell death in HIV­1­infected cell cultures.

These findings raise the possibility of reduc­ing immune activation and inflammation in response to chronic viral infections through caspase­1 inhibition. The research also sug­gests two new approaches to improve HIV­1 therapy: the use of antiretroviral agents that act early in the viral life cycle to block abor­tive infection, and the use of agents that inhibit caspase­1. Combination therapy with multiple classes of antiretroviral drugs is the standard of care for patients infected with HIV­1, and this therapy effectively suppresses viral replication. Suppression of caspase­1 activation may not be necessary if combination therapy prevents abortive infection as well.

Although Doitsh et al. do not report IL­18 levels in their study, this cytokine is generally produced along with IL­1β after inflamma­some activation, and elevated levels are associ­ated with inflammatory conditions13–15. Serum IL­18 levels, which are known to be high in HIV­1 infection, are reduced by anti retroviral

therapy14,15. Thus, it remains to be seen whether caspase­1 inhibitors will add to exist­ing antiretroviral therapy for the treatment of HIV­1 infection. Either way, the implication of pyroptosis in CD4+ T­cell depletion is a new explanation for this 30­year­old mystery in HIV­1 pathogenesis. ■

Andrea L. Cox and Robert F. Siliciano are in the Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland 21218, USA. R.F.S. is also at the Howard Hughes Medical Institute, Baltimore.e-mails: acox@jhmi.edu; rsiciliano@jhmi.edu

1. Gottlieb, M. S. et al. N. Engl. J. Med. 305, 1425–1431 (1981).

2. Doitsh, G. et al. Nature 505, 509–514 (2014).3. Miao, E. A. et al. Immunol. Rev. 243, 206–214 (2011).4.  Ho, D. D. et al. Nature 373, 123–126 (1995).5.  Wei, X. et al. Nature 373, 117–122 (1995).6.  Veazey, R. S. et al. Science 280, 427–431 (1998).7.  Brenchley, J. M. et al. Nature Med. 12, 1365–1371

(2006).8. Silvestri, G. et al. Immunity 8, 441–452 (2003).9. Doitsh, G. et al. Cell 143, 789–801 (2010).10. Monroe, K. M. et al. Science http://dx.doi.

org/10.1126/science.1243640 (2013).11. Latz, E. et al. Nature Rev. Immunol. 13, 397–411

(2013).12. Lamkanfi, M. et al. J. Leuk. Biol. 82, 220–225 (2007).13. Mallat, Z. et al. Circulation 104, 1598–1603 (2001).14. Iannello, A. et al. Curr. HIV Res. 8, 147–164 (2010).15. Watanabe, D. et al. Viral Immunol. 23, 619–625

(2010).

Activated CD4+ T cell HIV-1

QuiescentCD4+ T cell

a b

Caspase-3

Integration

Apoptosis5% of cell death

Reverse transcription

Viral DNA

Host DNA

Viral RNA

Caspase-1

IFI16

Abortiveinfection

Productiveinfection

Pyroptosis95% of cell death

IL-1β secretionIn�ammation

Reverse transcription

Incomplete reverse transcript

Figure 1 | CD4+ T-cell death during HIV-1 infection. a, Productive infection of a CD4+ T cell with HIV­1 involves viral entry to the cell, reverse transcription of viral RNA to DNA and integration of viral DNA into the host­cell genome. Following one or two days of viral replication, the activated, infected cell dies through apoptosis, mediated by the action of the enzyme caspase­3. Only about 5% of the CD4+ T cells that die after HIV­1 infection are activated, productively infected cells. b, Doitsh et al.2 show that most CD4+ T­cell deaths result from caspase­1­mediated pyroptosis in non­activated (quiescent) CD4+ T cells that have undergone abortive infection, during which incomplete viral DNA transcripts remain in the cells. These transcripts are sensed by the cellular DNA sensor IFI16, which leads to caspase­1 activation, resulting in the secretion of the highly inflammatory cell­signalling molecule IL­1β and pyroptosis.

2 3 J A N U A R Y 2 0 1 4 | V O L 5 0 5 | N A T U R E | 4 9 3

NEWS & VIEWS RESEARCH

© 2014 Macmillan Publishers Limited. All rights reserved