Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Follicular CXCR5-expressing CD8+ T cells curtail chronic viral infection

An Erratum to this article was published on 19 October 2016

This article has been updated

Abstract

During chronic viral infection, virus-specific CD8+ T cells become exhausted, exhibit poor effector function and lose memory potential1,2,3,4. However, exhausted CD8+ T cells can still contain viral replication in chronic infections5,6,7,8,9, although the mechanism of this containment is largely unknown. Here we show that a subset of exhausted CD8+ T cells expressing the chemokine receptor CXCR5 has a critical role in the control of viral replication in mice that were chronically infected with lymphocytic choriomeningitis virus (LCMV). These CXCR5+ CD8+ T cells were able to migrate into B-cell follicles, expressed lower levels of inhibitory receptors and exhibited more potent cytotoxicity than the CXCR5 subset. Furthermore, we identified the Id2–E2A signalling axis as an important regulator of the generation of this subset. In patients with HIV, we also identified a virus-specific CXCR5+ CD8+ T-cell subset, and its number was inversely correlated with viral load. The CXCR5+ subset showed greater therapeutic potential than the CXCR5 subset when adoptively transferred to chronically infected mice, and exhibited synergistic reduction of viral load when combined with anti-PD-L1 treatment. This study defines a unique subset of exhausted CD8+ T cells that has a pivotal role in the control of viral replication during chronic viral infection.

This is a preview of subscription content, access via your institution

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

Prices may be subject to local taxes which are calculated during checkout

Figure 1: Virus-specific CXCR5+CD8+ T cells are generated during chronic infection and migrate into B-cell follicles.
Figure 2: Virus-specific CXCR5+ CD8+ T cells are less exhausted than CXCR5CD8+ T cells and control viral load during chronic infection.
Figure 3: The Id2–E2A axis is critical for the differentiation of the CXCR5+CD8+ T-cell subset during chronic viral infection.
Figure 4: CXCR5+ CD8+ T cells exhibit greater therapeutic potential than CXCR5CD8+ T cells in the control of chronic viral infection.

Similar content being viewed by others

Accession codes

Primary accessions

Gene Expression Omnibus

Data deposits

RNA-seq data has been deposited in Gene Expression Omnibus under accession number GSE74148.

Change history

References

  1. Zajac, A. J. et al. Viral immune evasion due to persistence of activated T cells without effector function. J. Exp. Med. 188, 2205–2213 (1998)

    Article  CAS  Google Scholar 

  2. Wherry, E. J. T cell exhaustion. Nat. Immunol. 12, 492–499 (2011)

    Article  CAS  Google Scholar 

  3. Gallimore, A. et al. Induction and exhaustion of lymphocytic choriomeningitis virus-specific cytotoxic T lymphocytes visualized using soluble tetrameric major histocompatibility complex class I-peptide complexes. J. Exp. Med. 187, 1383–1393 (1998)

    Article  CAS  Google Scholar 

  4. Schietinger, A. & Greenberg, P. D. Tolerance and exhaustion: defining mechanisms of T cell dysfunction. Trends Immunol. 35, 51–60 (2014)

    Article  CAS  Google Scholar 

  5. Speiser, D. E. et al. T cell differentiation in chronic infection and cancer: functional adaptation or exhaustion? Nat. Rev. Immunol. 14, 768–774 (2014)

    Article  CAS  Google Scholar 

  6. Deng, K. et al. Broad CTL response is required to clear latent HIV-1 due to dominance of escape mutations. Nature 517, 381–385 (2015)

    Article  CAS  ADS  Google Scholar 

  7. Wherry, E. J. & Kurachi, M. Molecular and cellular insights into T cell exhaustion. Nat. Rev. Immunol. 15, 486–499 (2015)

    Article  CAS  Google Scholar 

  8. Jones, R. B. & Walker, B. D. HIV-specific CD8+ T cells and HIV eradication. J. Clin. Invest. 126, 455–463 (2016)

    Article  Google Scholar 

  9. Zehn, D., Utzschneider, D. T. & Thimme, R. Immune-surveillance through exhausted effector T-cells. Curr. Opin. Virol. 16, 49–54 (2016)

    Article  CAS  Google Scholar 

  10. Cyster, J. G. et al. Follicular stromal cells and lymphocyte homing to follicles. Immunol. Rev. 176, 181–193 (2000)

    Article  CAS  Google Scholar 

  11. Crotty, S. T follicular helper cell differentiation, function, and roles in disease. Immunity 41, 529–542 (2014)

    Article  CAS  Google Scholar 

  12. Haynes, N. M. et al. Role of CXCR5 and CCR7 in follicular TH cell positioning and appearance of a programmed cell death gene-1high germinal center-associated subpopulation. J. Immunol. 179, 5099–5108 (2007)

    Article  CAS  Google Scholar 

  13. Kim, H. J., Verbinnen, B., Tang, X., Lu, L. & Cantor, H. Inhibition of follicular T-helper cells by CD8+ regulatory T cells is essential for self tolerance. Nature 467, 328–332 (2010)

    Article  CAS  ADS  Google Scholar 

  14. Kim, H. J. & Cantor, H. Regulation of self-tolerance by Qa-1-restricted CD8+ regulatory T cells. Semin. Immunol. 23, 446–452 (2011)

    Article  CAS  Google Scholar 

  15. Belle, I. & Zhuang, Y. E proteins in lymphocyte development and lymphoid diseases. Curr. Top. Dev. Biol. 110, 153–187 (2014)

    Article  CAS  Google Scholar 

  16. Miyazaki, M. et al. The opposing roles of the transcription factor E2A and its antagonist Id3 that orchestrate and enforce the naive fate of T cells. Nat. Immunol. 12, 992–1001 (2011)

    Article  CAS  Google Scholar 

  17. Shaw, L. A. et al. Id2 reinforces TH1 differentiation and inhibits E2A to repress TFH differentiation. Nat. Immunol. 17, 834–843 (2016)

    Article  CAS  Google Scholar 

  18. Vezys, V. et al. Continuous recruitment of naive T cells contributes to heterogeneity of antiviral CD8 T cells during persistent infection. J. Exp. Med. 203, 2263–2269 (2006)

    Article  CAS  Google Scholar 

  19. Quigley, M. F., Gonzalez, V. D., Granath, A., Andersson, J. & Sandberg, J. K. CXCR5+ CCR7 CD8 T cells are early effector memory cells that infiltrate tonsil B cell follicles. Eur. J. Immunol. 37, 3352–3362 (2007)

    Article  CAS  Google Scholar 

  20. Connick, E. et al. CTL fail to accumulate at sites of HIV-1 replication in lymphoid tissue. J. Immunol. 178, 6975–6983 (2007)

    Article  CAS  Google Scholar 

  21. Banga, R. et al. PD-1+ and follicular helper T cells are responsible for persistent HIV-1 transcription in treated aviremic individuals. Nat. Med. 22, 754–761 (2016)

    Article  CAS  Google Scholar 

  22. Barber, D. L. et al. Restoring function in exhausted CD8 T cells during chronic viral infection. Nature 439, 682–687 (2006)

    Article  CAS  ADS  Google Scholar 

  23. Freeman, G. J., Wherry, E. J., Ahmed, R. & Sharpe, A. H. Reinvigorating exhausted HIV-specific T cells via PD-1-PD-1 ligand blockade. J. Exp. Med. 203, 2223–2227 (2006)

    Article  CAS  Google Scholar 

  24. Velu, V. et al. Enhancing SIV-specific immunity in vivo by PD-1 blockade. Nature 458, 206–210 (2009)

    Article  CAS  ADS  Google Scholar 

  25. Fuller, M. J. et al. Immunotherapy of chronic hepatitis C virus infection with antibodies against programmed cell death-1 (PD-1). Proc. Natl Acad. Sci. USA 110, 15001–15006 (2013)

    Article  CAS  ADS  Google Scholar 

  26. Fukazawa, Y. et al. B cell follicle sanctuary permits persistent productive simian immunodeficiency virus infection in elite controllers. Nat. Med. 21, 132–139 (2015)

    Article  CAS  Google Scholar 

  27. Miles, B. & Connick, E. TFH in HIV latency and as sources of replication-competent virus. Trends Microbiol. 24, 338–344 (2016)

    Article  CAS  Google Scholar 

  28. Pauken, K. E. & Wherry, E. J. Overcoming T cell exhaustion in infection and cancer. Trends Immunol. 36, 265–276 (2015)

    Article  CAS  Google Scholar 

  29. Verdeil, G., Fuertes Marraco, S. A., Murray, T. & Speiser, D. E. From T cell “exhaustion” to anti-cancer immunity. Biochim. Biophys. Acta (2015)

  30. Kim, P. S. & Ahmed, R. Features of responding T cells in cancer and chronic infection. Curr. Opin. Immunol. 22, 223–230 (2010)

    Article  CAS  Google Scholar 

  31. Barber, D. L., Wherry, E. J. & Ahmed, R. Cutting edge: rapid in vivo killing by memory CD8 T cells. J. Immunol. 171, 27–31 (2003)

    Article  CAS  Google Scholar 

  32. Rasheed, M. A. et al. Interleukin-21 is a critical cytokine for the generation of virus-specific long-lived plasma cells. J. Virol. 87, 7737–7746 (2013)

    Article  Google Scholar 

  33. McCausland, M. M. & Crotty, S. Quantitative PCR technique for detecting lymphocytic choriomeningitis virus in vivo. J. Virol. Methods 147, 167–176 (2008)

    Article  CAS  Google Scholar 

  34. Zhong, S. et al. High-throughput illumina strand-specific RNA sequencing library preparation. Cold Spring Harb. Protoc. 2011, 940–949 (2011)

    Article  Google Scholar 

  35. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005)

    Article  CAS  ADS  Google Scholar 

  36. Zhou, X. et al. Differentiation and persistence of memory CD8+ T cells depend on T cell factor 1. Immunity 33, 229–240 (2010)

    Article  CAS  Google Scholar 

  37. Xu, L. et al. The transcription factor TCF-1 initiates the differentiation of TFH cells during acute viral infection. Nat. Immunol. 16, 991–999 (2015)

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank R. Ahmed (Emory University) for providing the P14 TCR transgenic mice, retroviral vectors and LCMV Armstrong and Cl13 viruses; Y. Zhuang (Duke University) for providing the Id2fl/fl mice; the core facility centre of Third Military Medical University for helping us with cell sorting; T. Wu (NIH) for insightful discussion. The work was supported by National Basic Research Program of China (973 program, 2013CB531500, to L.Y.; 2014CB542501 to H.Q.), the National Natural Science Foundation of China (81220108024 to Y.W.; 81471624 to L.Y.; U1202228 to J.X.; No. 81425011, 81330070 to H.Q.; No.31500733 to Q.B).

Author information

Authors and Affiliations

Authors

Contributions

L.Y. conceived the project. R.H., S.H., and C.L. performed both in vivo and in vitro experiments. Q.B., M.H., T.S., G.W., and T.N. performed RNA-seq and bioinformatics’ analysis. A.Z., Y.Y., X.Y., L.X., X.C., Y.H., P.W., K.D., Y.C., J.O., Y.L., X.Z., and H.L performed LCMV- and HIV-associated experiments; C.Z performed thymectomy; X.Z performed the reporter assay. L.Y., Y.W., H.Q., and J.X. designed the study, analysed the data and wrote the manuscript; L.Y., and Y.W. supervised the study.

Corresponding authors

Correspondence to Hai Qi, Yuzhang Wu or Lilin Ye.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Extended data figures and tables

Extended Data Figure 1 Virus-specific CXCR5+ CD8+ T cells are not apparent in acutely infected mice and in the non-lymphoid tissues of chronically infected mice and are not Qa-1-restricted.

a, CXCR5 expression in virus-activated CD8+ T cells in the spleens of Arm+-infected mice. b, CXCR5 expression in virus-activated CD8+ T cells in the lungs and livers of Cl13-infected mice. c, Helios and ICOSL expression in virus-activated CXCR5+ CD8+ T cells during Cl13 infection.

Extended Data Figure 2 Virus-specific CXCR5+CD8+ T cells are less exhausted than CXCR5CD8+ T cells on day 8 after Cl13 infection.

a, b, PD-1, Tim-3 and KLRG1 expression on virus-specific CXCR5+ and CXCR5CD8+ T cells in the spleens of Cl13-infected mice on day 8 after infection (n = 4 or 5). MFI, mean fluorescence intensity. c, Upon stimulation with the indicated peptides, the cytokine production of CXCR5+ and CXCR5CD8+ T cells in the spleens of LCMV-Cl13-infected mice was analysed on day 8 post-infection (n = 4 or 5). Data are representative of three independent experiments, and were analysed by two-tailed unpaired t-test (b, c). Error bars (b, c) denote s.e.m. *P < 0.05; **P < 0.01; ***P < 0.001. NS, not significant.

Source data

Extended Data Figure 3 Virus-specific CXCR5+CD8+ T cells localized in B-cell follicles have minimal effect on germinal centre B and TFH responses.

a, b, Equal numbers of CXCR5+ and CXCR5CD8+ T cells sorted from Cl13-infected mice were adoptively transferred into infection-matched CD8−/− mice. On day 5 after transfer, frequency and number of germinal centre B cells and TFH cells in the spleens of recipient mice were analysed (n = 3). c, Titration of LCMV-specific IgG in the serum of recipient mice (n = 3). d, The expression levels of PD-L1 and PD-L2 on cell subsets residing in the T-cell zone and in B-cell follicles (n = 4). DC, dendritic cell; FRC, fibroblast reticular cell. The data are representative of three independent experiments, and were analysed by two-tailed unpaired t-test (b–d). Error bars (b–d) denote s.e.m. *P < 0.05; **P < 0.01; ***P < 0.001. NS, not significant.

Source data

Extended Data Figure 4 The maintenance of functional CXCR5+CD8+ T cells is dependent on follicle structures.

a, Equal numbers of virus-activated CXCR5+CD8+ T and CXCR5CD8+ T cells obtained from Cl13-infected C57BL/6J (CD45.1) mice were adoptively transferred into infection-matched μMT (CD45.2) or C57BL/6 (CD45.2) (wild-type) mice. Analysis was performed on day 8 after transfer. b, c, Frequency and number of CD45.1+CXCR5+CD8+ T cells in the recipient mice (n = 3). d, e, On stimulation of peptide, surface CD107 expression and cytokine production of CD45.1+CXCR5+CD8+ T cells in the recipient mice (n = 3). f, Viral titers in the indicated tissues obtained from control wild-type and μMT mice without cell transfer and from wild-type and μMT mice receiving CXCR5+CD8+ T cell transfer (n = 3). The data are representative of three independent experiments, and were analysed by two-tailed unpaired t-test (ce). Error bars (ce) denote s.e.m. *P < 0.05; **P < 0.01; ***P < 0.001.

Source data

Extended Data Figure 5 CXCR5 expression is critical for the localization of virus-activated CD8+ T cells to B-cell follicles.

a, Set-up of splenic chimaera mice. Total splenocytes obtained from Cxcr5−/− or wild-type mice were mixed with splenocytes obtained from Cd8−/− mice and then transferred to non-lethally irradiated Cd8−/− recipients and immediately infected with Cl13. Analysis was performed on day 15 after infection. b, The localization of virus-activated CD8+ T cells in the lymph nodes was detected by confocal microscopy on day 15 after infection (blue, IgD; red, CD8; green, CD3) and follicular entry coefficiency was calculated (Cxcr5−/−, n = 15; wild-type, n = 20). Scale bar, 100 μm. c, The CD107 expression and IFN-γ secretion of wild-type and Cxcr5−/− CD8+ T cells upon peptide stimulation (n = 3). d, Viral titers in the indicated tissues from mice that received splenocytes from Cxcr5−/− or wild-type mice (n = 3). Data are representative of three independent experiments, and were analysed by two-tailed unpaired t-test (bd). Error bars (bd) denote s.e.m. *P < 0.05; **P < 0.01; ***P < 0.001.

Source data

Extended Data Figure 6 Distinct transcriptional profiles of CXCR5+ and CXCR5CD8+ T-cell populations.

a, Transcriptomic profiling of CXCR5+ and CXCR5 cell subsets. b, Gene Ontology (GO) enrichment was analysed using Gene Set Enrichment Analysis (GSEA) and significantly enriched (P value < 0.05) molecular function GO terms were shown with their enrichment scores. c, The enrichment of gene sets containing genes sharing upstream cis-regulatory motifs of transcription factor binding sites were assessed using GSEA. The transcription factor binding sites with significant enrichment (P value < 0.05) in CXCR5+CD8+ cells were listed (left). The GSEA result of the gene set including the E47 (E2A isoform) binding site (denoted as V$E47_02 in the Molecular Signatures Database version 3.0) was shown (right). d, The normalized expression levels of Id2 and E2A isoform E47 in CXCR5+ and CXCR5CD8+ cells were calculated on the basis of RNA-seq data and was expressed in reads per kilobase per million mapped reads. e, qPCR analysis of the expression levels of Id2 and E2A isoform E47 in CXCR5+ and CXCR5CD8+ cells. Data are from one experiment with two biological replicates (ad) or are representative of three independent experiments (e), and were analysed by two-tailed unpaired t-test (e). Error bars (e) denote s.e.m. **P < 0.01. NS, not significant.

Source data

Extended Data Figure 7 E2A regulates the transcription of Cxcr5 by directly binding to DNA loci.

a, Kinetic analysis of Id2 expression levels in CXCR5+ and CXCR5CD8+ T cells during Cl13 infection by qPCR (n = 3). b, Id2 mRNA expression in CD8+ T cells in the spleens of littermate control (control) and Id2−/− mice (n = 3). c, The number of CD44hiCD8+ T cells in the spleens of control and Id2−/− mice on day 25 after Cl13 infection (n = 4). d, An alignment of putative E2A-binding sites in the Cxcr5 intron. The conserved E2A-binding motif ‘CASSTG’ (or ‘GTSSAC’ on the reverse strand) is highlighted in red, and its locations relative to the transcriptional start site (TSS) of Cxcr5 are marked. e, Retroviral reporter constructs containing a wild-type or mutated Cxcr5 regulatory region and the Psv40 promoter, as well as self-inactivating mutations in the long terminal repeats (SIN), a sequence encoding Thy-1.1, and a PGK-EGFP cassette (including P-Pgk1 (a promoter of the gene encoding phosphoglycerate kinase 1) and EGFP). Arrows indicate the transcription start site and orientation, and the numbers shown above indicate the position. f, Thy-1.1 expression levels on GFP+CD8+ T cells transduced with a reporter construct containing wild-type or mutated Cxcr5 regulatory region, MFI of Thy-1.1 was normalized to GFP expression (n = 3). g, CXCR5 expression in non-transduced, E2A-overexpressing, Id2–E2A-co-overexpressing and Id2-overexpressing P14 CD8+ T cells on day 8 after Cl13 infection (n = 4). E2A refers to E47 isoform. h, PD-1 and CD107 surface expression levels and cytokine production in non-transduced P14 cells and E2A-overexpressing P14 cells (n = 4). Data are representative of three independent experiments, and were analysed by two-tailed unpaired t-test (ac, fh). Error bars (ac, fh) denote s.e.m. *P<0.05; **P<0.01; ***P<0.001. NS, not significant.

Source data

Extended Data Figure 8 Virus-activated CXCR5+CD8+ T cells are converted into CXCR5CD8+ T cells.

a, Schematic map showing the construction of CXCR5–GFP knock-in mice. b, CXCR5-staining and GFP expression in CD19+ cells and in CD44hiCD4+ T cells in CXCR5–GFP knock-in mice and from wild-type mice. c, GFP+CD44hiCD8+ T cells and GFPCD44hiCD8+ T cells were sorted from day 8 Cl13-infected CXCR5–GFP knock-in mice (CD45.2). The cells were labelled with Celltrace Violet and then transferred into infection-matched wild-type recipients (CD45.1). The presence of GFP and Celltrace Violet in the transferred cells (CD45.2) was detected on days 0, 5, and 12 after transfer. d, Id2 expression levels in GFP+ViolethiCD8+ T cells and in GFPVioletloCD8+ T cells from recipient mice receiving GFP+CD8+ T cells transfer on day 5 after transfer (n = 3). e, Surface expression of CD107 and IFN-γ production in GFP+CD8+ T cells, newly converted GFPCD8+ (GFP+/GFP) T cells and GFPCD8+ T cells (GFP, n = 4, GFP+/GFP and GFP+, n = 3). f, Equal numbers of GFP+CD8+ T cells, GFP+/GFP T cells and GFPCD8+ T cells were co-cultured with peptide-coated target cells ex vivo, respectively. Five hours later, the killing efficiency of the effector cells was analysed (n = 3). g, h, The number of CD44hiCD8+ T cells and the frequency of CXCR5+CD8+ T cells in the spleens of control mice infected on day 28 and thymectomized mice (subject to the surgery at day 21 after infection) (n = 4). i, Viral titers in the indicated tissues of control mice, mice received thymectomy and mice received CXCR5+CD8+ T cell transfer after thymectomy (control and CXCR5+ transfer, n = 3; thymectomy, n = 4). Data are representative of three independent experiments, and were analysed by two-tailed unpaired t-test (dg, i). Error bars (dg, i) denote s.e.m. *P < 0.05; **P < 0.01; ***P < 0.001. NS, not significant.

Source data

Extended Data Figure 9 The HIV-specific CXCR5+CD8+ T-cell subset is present in chronically HIV-infected patients.

a, CXCR5 expression in HIV-specific CD8+ T cells in blood of HIV-infected patients. b, The expression levels of PD-1and Tim-3 in HIV-specific CXCR5+ and CXCR5CD8+ T cells in blood of HIV-infected patients (PD-1, n = 13; Tim-3, n = 12). c, The correlation between viral copy number in serum and CXCR5+CD8+ T cell number in blood in chronic HIV-infected patients prior to anti-retroviral treatment (n = 14). d, HIV-specific (IFN-γ+) CXCR5+CD8+ T cells in lymph nodes of HIV-infected patients. e, CD8+ T-cell localization in the lymph nodes of HIV-infected patients and HIV-negative donors by confocal microscopy (green, CD20; red, CD8). Scale bar, 20μm. f, The expression levels of CD107 and perforin and cytokine production in HIV-specific CXCR5+ and CXCR5CD8+ T cells in lymph nodes of HIV-infected patients (n = 4). g, The expression levels of E2A isoform E47 and Id2 in IFN-γ+CXCR5+ and IFN-γ+CXCR5CD8+ T cells in lymph nodes of HIV-infected patients (n = 4). Data are representative of two independent experiments and analysed by two-tailed paired t-test (b, f, g). The correlation between viral load and CXCR5+CD8+ T cell number was analysed by non-parametric Spearman correlation test (c). *P < 0.05; **P < 0.01; ***P < 0.001. NS, not significant.

Source data

Extended Data Figure 10 Diagrammatic summary of the fate of CXCR5+CD8+ T cells during chronic viral infection.

During chronic viral infection, virus-specific exhausted CD8+ T cells differentiate into CXCR5+ and CXCR5 subsets governed by the Id2/–E2A axis. The CXCR5+CD8+ subset migrates into B-cell follicles, where a lesser inhibitory microenvironment prevents the rapid exhaustion and loss of effector functions of these cells. By contrast, the CXCR5 subset undergoes severe exhaustion owing to the inhibitory microenvironment outside B-cell follicles. Follicular CXCR5+CD8+ T cells eventually convert into CXCR5 cells, presumably driven by increased Id2 expression. The de novo converted CXCR5CD8+ T cells possess better cytotoxicity, hence they are capable of clearing virus-infected cells more efficiently outside of follicles when they exit B-cell follicles.

Supplementary information

Supplementary Information

This file contains the data for Supplementary Tables 1-2. (PDF 132 kb)

PowerPoint slides

Source data

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

He, R., Hou, S., Liu, C. et al. Follicular CXCR5-expressing CD8+ T cells curtail chronic viral infection. Nature 537, 412–416 (2016). https://doi.org/10.1038/nature19317

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nature19317

This article is cited by

Comments

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.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing