1. Departments of Immunology and,
2. Pharmacology, and the,
3. Howard Hughes Medical Institute, University of Washington, Seattle, Washington 98195, USA
4. National Heart, Lung, and Blood Institute, NIH, 9000 Rockville Pike, Bethesda, Maryland 20892, USA
5. Present address: Amgen Corporation, Seattle, Washington 98101, USA.

Correspondence to: Alexander Y. Rudensky^1,3 Correspondence and requests for materials should be addressed to A.Y.R. (Email: aruden@u.washington.edu).

In males, Foxp3 deficiency results in fatal early-onset systemic autoimmune disease⁵. In heterozygote Foxp3^wt/null females only one-half of T cells harbours the mutant Foxp3 allele due to random X-chromosome inactivation, whereas autoimmunity is controlled by a normal T_r population expressing the Foxp3 wild-type allele. Thus, we were able to genetically mark cells actively transcribing a Foxp3^null allele, yet lacking Foxp3 protein (hereafter called T_fn for Foxp3^null-expressing T cells), through an in-frame insertion of GFP into a stop-codon-disrupted Foxp3 locus (Foxp3^gfpko) and investigate their features in mice (Fig. 1a; see also Supplementary Figs 1 and 2a). Female Foxp3^gfpko/wt mice were healthy, whereas male Foxp3^gfpko mice developed the same severity of autoimmunity as Foxp3 knockout (Foxp3^null) mice⁶, resulting in death at 4 weeks of age. Thymocyte and peripheral lymphoid organ cellularity did not differ between Foxp3^gfpko/wt and Foxp3^gfp/gfp mice, nor did the proportion of Foxp3⁺ T_r cells and Foxp3^- CD4⁺ T cells (data not shown). As our main focus was to characterize T_fn cells in healthy Foxp3^gfpko/wt mice, analysis of autoimmune male Foxp3^gfpko mice is included as Supplementary Fig. 2.

Figure 1: Phenotype and frequency of cells transcribing the Foxp3^gfpko locus in Foxp3^gfpko/wt mice.

a, We modified our previously described GFP–Foxp3 fusion protein targeting construct⁸ by inserting stop codon/frameshift cassettes downstream of GFP and also into Foxp3 exon 5. In this manner, potentially important regulatory sequences were retained without the possibility that cryptic re-initiation of Foxp3 translation would generate full-length protein. b, Flow cytometric analysis of CD4⁺CD8^-HSA^low thymocytes or CD4⁺ lymph node cells from female Foxp3^gfp/gfp or Foxp3^gfpko/wt mice. c, Frequency of GFP⁺ cells among either CD4⁺CD8^-HSA^low thymocytes (Thy, circles) or CD4⁺ lymph node cells (LN, diamonds) for individual mice. d, Flow cytometric analysis of CD4⁺ lymph node cells for the indicated markers. The four cell subsets are demarcated in b as follows: GFP^–CD25^– T_n cells (grey), GFP^–CD25⁺Foxp3^-CD4⁺ T cells (green), GFP⁺ T_r cells (red) from Foxp3^gfp/gfp mice, and GFP⁺ T_fn cells from Foxp3^gfpko/wt mice (blue). e, Ki67 staining of lymph node cells from the indicated mice. f, Proliferation of sorted T cells in response to the indicated reagents. Results are representative of three independent experiments. Data represent the mean and standard deviation for triplicate wells. g, Cytokine production by sorted cells from Foxp3^gfp/gfp (T_n, CD25⁺Foxp3^-CD4⁺, T_r cells) or Foxp3^gfpko/wt (T_fn cells) mice in response to plate-bound anti-CD3 and anti-CD28 assayed by enzyme-linked immunosorbent assay. Results are representative of three separate experiments; data represent mean and maximum/minimum for duplicate wells and colours are as described in d.

High resolution image and legend (228K)

T_fn cells constituted 1–3% of mature CD4⁺ thymocytes and peripheral CD4⁺ T cells, indicating that Foxp3 is not required to rescue T_r precursors from negative selection (Fig. 1b, c). This is consistent with a reported abundance of T-cell receptors (TCRs) characteristic of T_r cells in Foxp3^null mice⁷. As ectopic expression of Foxp3 has been shown to induce a state of hyporesponsiveness in CD4⁺ T cells³, we expected T_fn cells to appear highly activated, similar to CD25⁺Foxp3^-CD4⁺ T cells (Fig. 1b), which exhibit characteristics of activated, pro-inflammatory T cells⁸. T_fn cells, however, were small and CD62L^high, more similar to T_r cells and naive CD4⁺ cells than to CD25⁺Foxp3^-CD4⁺ T cells (Fig. 1b, d). T_fn cells expressed intermediate levels of CD25, CD44, CTLA4, GITR and ICOS in comparison to T_r cells (Fig. 1b, d; see also Supplementary Fig. 3a). Notably, reduced IL-7R expression, which is considered to be a distinguishing feature of T_r cells, was lowest on T_fn cells in comparison to the other T-cell subsets (Fig. 1d). In contrast to T_r cells, T_fn cells exhibited negligible proliferative activity in vivo based on proliferation-associated antigen Ki67 expression (Fig. 1e).

In vitro, both T_fn and T_r cells showed an anergic phenotype, which was reversed in both T_fn and T_r cells by provision of IL-2; however, proliferation was more readily restored in T_fn cells by limited CD28 co-stimulation (Fig. 1f). Thus, anergy in T_fn cells is less stable in comparison to T_r cells. T_fn cells were also similar to T_r cells—and distinct from T_n (GFP^-CD25^-CD4⁺ T cells from Foxp3^gfp/gfp mice) and CD25⁺Foxp3^-CD4⁺ T cells—in their inability to generate IL-2 and T-helper 1 and 2 (T_h1 and T_h2) cytokines (Fig. 1g). Thus, IL-2 expression may be blocked transcriptionally during T_r cell differentiation in a Foxp3-independent manner. Lack of IL-2 production and decreased IL-7R make both T_fn and T_r cells reliant on exocrine IL-2, but T_fn cells are probably less competitive due to intermediate CD25 levels. Indeed, we found that T_fn cells were hyper-responsive to elevated IL-2 levels in vivo (Supplementary Fig. 7). Together, our analyses show that T_r cell development coincides with Foxp3-independent acquisition of several key T_r cell characteristics.

Unlike T_R cells, some T_fn cells produced either IL-10 or IL-17, resembling Tr1 (IL-10-producing Foxp3^-CD4⁺ T cells⁹) and T_h17 cells, respectively (Fig. 1g; see also Supplementary Fig. 2g). Consistent with observed IL-17 production, both T_fn and CD25⁺Foxp3^-CD4⁺ T cells were found to express orphan nuclear receptor RORt (Rorc), which has been shown to both promote IL-17 and block IL-2 production in T_h17 cells¹⁰ (Supplementary Fig. 3b). Notably, Rorc transcription was reduced only fourfold in T_R relative to T_fn cells but Il17 transcription was suppressed 7,500-fold, suggesting an ability of Foxp3 to over-ride RORt activity.

The defining feature of T_r cells is their ability to suppress pro-inflammatory immune function. T_fn cells were incapable of suppressing CD4⁺CD25^- T-cell proliferation in vitro (Fig. 2a) and controlling effector T-cell expansion and ensuing splenomegaly and lymphadenopathy after adoptive co-transfer into T-cell-deficient recipient mice (Fig. 2b, c). Spleens and lymph nodes of mice that had received CD45RB^highCD25^-CD4⁺ effector T cells with T_fn cells were similar in cellularity to those of mice that had received effector T-cells alone, whereas the co-transfer of T_r cells with effector T cells significantly limited effector T-cell expansion. Similarly, effector T cells were increased in size in the presence of T_fn cells, but not T_r cells, likely reflecting increased proliferation (Fig. 2c). Lack of suppression was not due to loss of T_fn cells because donor T_fn cells competed efficiently with effector T cells during homeostatic expansion, manifested by an increase in their proportion and number (Fig. 2d).

Figure 2: Foxp3-dependent suppression and lineage stability.

a, FACS purified T_r or T_fn cells were compared for their ability to suppress CD25^-CD4⁺ T-cell proliferation in the presence of concavalin A and T-cell-depleted splenic antigen presenting cells. Data represent mean and standard deviation for triplicate wells; results are representative of four separate experiments. b–d, To examine suppressor activity in vivo, T_r or T_fn cells were co-transferred with allelically marked effector T cells into lymphopenic animals. Sorted Ly5.1⁺CD45RB^highCD25^-CD4⁺ effector T cells (T_eff, 3  10⁵) were transferred into B6.SCID mice alone or with sorted T_R or T_fn cells (10⁵). Five weeks after transfer, splenocytes were enumerated (b) and lymph node and spleen cells were analysed by flow cytometry. Cell size of the effector T-cell population (c) and abundance and GFP expression of the donor T_r or T_fn cell populations (d) are shown. Data are representative of two independent experiments. e, Quantification of Foxp3 cDNA prepared from the indicated cell populations. For peripheral cells, error bars represent mean and maximum/minimum for biological duplicates.

High resolution image and legend (129K)

T_fn cells also differed from T_r cells in their ability to retain Foxp3 transcription. Whereas 90% of donor T_r cells from Foxp3^gfp/gfp mice retained expression of GFP-tagged Foxp3 protein, two-thirds of the expanded T_fn population lost GFP expression (Fig. 2d). Thus, in the course of homeostatic proliferation, Foxp3 transcription in the absence of Foxp3 function was not efficiently retained. Furthermore, quantification of Foxp3 messenger RNA in both thymic and peripheral T_fn and T_r cells supported the model that Foxp3 promotes its own transcription (Fig. 2e). Although we cannot rule out the possibility that the premature stop codons destabilize Foxp3^gfpko mRNA, we have found that conditional deletion of Foxp3 exons 1–5 in mature T_r cells also results in an identical reduction of Foxp3 mRNA levels¹¹. Together, these findings demonstrate that stability of Foxp3 expression and suppressor activity is dependent upon Foxp3 function rather than coinciding with its expression.

To characterize the Foxp3-dependent transcriptional programme, global gene expression profiling was performed. Automated gene clustering based on similarity in fold-change values between T_n cells and CD25⁺Foxp3^-CD4⁺ T cells, T_fn cells or T_r cells revealed that a distinct set of genes with moderate fold-change values in T_fn cells was further amplified in T_r cells. These genes encoded many of the cell surface markers currently used to identify T_r cells, such as CD25, GITR and CTLA4. To incorporate this expression pattern, gene clusters were then hand-curated to generate eight clusters with 'Foxp3-amplified' genes residing in clusters T4 and P4 (Fig. 3a, b; see also Supplementary Fig. 5). For both thymocytes and peripheral cells, Foxp3-dependent genes constituted the largest cluster, reflecting a dominant role for Foxp3 in broadly effecting cellular physiology and phenotype (Fig. 3a, b; clusters T3, P3). Notably, 'Foxp3-dependent' (T3/P3) and 'Foxp3-amplified' genes (T4/P4) most faithfully retained the same expression pattern in thymic and peripheral cells, with Foxp3-amplified genes containing the highest proportion of shared genes (Supplementary Fig. 6a). These findings corroborate our functional studies in elucidating a novel facet of T_r cell differentiation whereby transcriptional and functional characteristics induced concurrently or before Foxp3 transcription are enforced by Foxp3 to become principal features of T_r cell biology.

Figure 3: Foxp3-amplified gene expression constitutes a defining characteristic of thymic and peripheral T_r cells.

a, b, Genes differentially expressed in CD25⁺Foxp3^-CD4⁺ (T₂₅), T_r and T_fn cells relative to T_n cells were identified for both thymic (a) and peripheral (b) T-cell subsets. For each cluster, values and line thicknesses represent the number of upregulated or downregulated genes, and error bars represent the full range of log₂ expression ratios. c, Expression patterns of genes that may have a role, or have been demonstrated to have a role, in T_r cell effector function. Foxp3-mediated repression of genes encoding pro-inflammatory cytokines and chemokines is also shown.

High resolution image and legend (292K)

To ascribe functional meaning to the defined gene clusters, comparisons with gene annotation databases were performed (Supplementary Fig. 6b, c), revealing that the Foxp3-dependent genes shared between thymic and peripheral data sets were significantly enriched for genes encoding cell surface and extracellular proteins. In contrast, the most enriched functional category for peripheral T_r cells was cell-cycle-associated genes (P6), demonstrating that Foxp3 rescues the inferior proliferative activity of T_fn cells in agreement with the analyses of cell size and Ki67 expression (Fig. 1d, e). Also for peripheral cells, the gene cluster specifically regulated by Foxp3 (P3) was enriched for genes involved in intercellular communication. Together, these data suggest that a second major role for Foxp3 is to adapt developing T_r cells by altering how environmental cues are integrated into cellular processes.

Because our experiments showed that suppressor function was Foxp3-dependent, we examined clusters T3 and P3 for genes defining potential T_r effector mechanisms (Fig. 3c). This analysis suggested several such mechanisms, including suppressive soluble factors^{12, 13, 14}, generation of extracellular adenosine^{15, 16, 17, 18}, and release of reactive oxygen¹⁹, and a possible, previously unanticipated role for T_r cells in regulating wound repair.

Among Foxp3-dependent genes, Pde3b (cyclic nucleotide phosphodiesterase 3B, cGMP-inhibited) was the most repressed (Fig. 4a, b). As cyclic AMP–protein kinase A pathways are well characterized attenuators of T-cell function²⁰, reduced PDE3B levels implied cAMP-mediated adaptation in T_r cells to chronic TCR and IL-2 signalling. Notably, naive T cells fail to downregulate PDE3B upon TCR engagement (Fig. 4b), and we found that Foxp3 binds a highly conserved region in the first intron of Pde3b (ref. 21). Thus, reduced PDE3B expression represents the first unique marker of T_r cells and may be considered more definitive then Foxp3 itself as it reports Foxp3 function. Hypothesizing that increased expression of PDE3B in T_r cells would be deleterious to T_r cell homeostasis, we introduced PDE3B or a catalytically inactive mutant, PDE3B(H801A)²², into T_r cells by retroviral gene transfer (Supplementary Fig. 8a, f, g). After transfer into T-cell-deficient recipients, PDE3B-expressing T_r cells were reduced in number by 60% in comparison to empty vector controls, whereas ectopic PDE3B expression in Foxp3^-CD4⁺ T cells (T_h) had no effect on their homeostasis (Fig. 4c). T_r cells expressing PDE3B(H801A) also exhibited reduced numbers, although to a lesser degree. Loss of PDE3B-transduced T_r cells was due to cell death or inefficient expansion rather than loss of T_r phenotype, because enumeration of expanded donor T_r and T_h cells based on differing Ly5 congenic markers rather than T_r phenotype revealed the same reduction of T_r cell numbers (Fig. 4c).

Figure 4: Reduced PDE3B expression permits normal T_r cell homoeostasis and T_r-specific gene expression.

a, b, Pde3b gene expression was analysed by real-time polymerase chain reaction (a) and by flow cytometry (b). Error bars represent mean and maximum/minimum for biological duplicates. Expression in GFP^–CD4⁺ and CD8⁺ gated cells after 3 days stimulation of total Foxp3^gfp/gfp splenocytes with ConA is also shown (dashed lines). T₂₅, CD25⁺Foxp3^-CD4⁺ T cells. c, Frequency in recipient mice of transduced (GFP⁺) T_r cells (Ly5.1^-CD4⁺GITR⁺) or T_n cells (Ly5.1⁺CD4⁺GITR^-), as well as the frequency of transduced Ly5.1^-CD4⁺ cells (see Supplementary Fig. 8a). Results are representative of five separate experiments. d, The effect of ectopic PDE3B (PDE3B/Vector) on expression of the listed cell-cycle-associated genes in cluster P6 was added to T_r/T_n values for the same genes (Fig. 3b) and plotted as described (Supplementary Fig. 8c). The effect of PDE3B(H801A) and PDE3B on the same genes is also shown. e, f, Proliferative responses of Pde3b^+/+ or Pde3b^-/- CD4⁺ T cells to anti-CD3 in the presence of irradiated APC (e) and to allogeneic (BALB/c) stimulator cells (f). Results are representative of two separate experiments; data represent mean and standard deviation for triplicate wells. g, h, Pre-activated CD4⁺CD25^- T cells were cultured with IL-2 alone or with the PDE3 inhibitor cilostamide or the PDE4 inhibitor rolipram as described in Supplementary Methods. Viable cell counts (g) and percentage live cells (h) are shown. Results are representative of three separate experiments.

High resolution image and legend (191K)

Transcriptional profiling of recovered cells suggested mechanistic underpinnings for reduced homeostatic fitness of PDE3B-transduced T_r cells, including reduced expression of genes encoding mitochondrial and biosynthetic proteins (Supplementary Fig. 8b). Remarkably, cell cycle genes shared by CD25⁺Foxp3^-CD4⁺ T cells and T_r cells (cluster P6) reverted to T_fn-like expression levels in PDE3B-transduced T_r cells in a manner that was largely dependent on PDE3B catalytic activity (Fig. 4d). This reversion to T_fn-like gene expression was also observed for many other Foxp3-dependent genes (Supplementary Fig. 8c). In apparent contrast to the deleterious effect of ectopic PDE3B on T_r cell homeostasis, we found no difference in the proliferative responses of PDE3B-deficient and -sufficient CD4⁺ T cells²³, and proliferation of wild-type T cells was not altered by PDE3 inhibitors (Fig. 4e, f and data not shown). However, we observed a substantial increase in viability and numbers of pre-activated T_n cells after IL-2-driven expansion in the presence of a PDE3, but not PDE4, inhibitor, suggesting that apoptosis associated with chronic IL-2-induced proliferation may be facilitated by PDE3B activity (Fig. 4g, h). Thus, we have identified Pde3b repression as a central component of Foxp3-dependent T_r cell maintenance, as its re-expression resulted in a reduction of biosynthetic processes, the attenuation of Foxp3-dependent proliferative fitness and the loss of some Foxp3-dependent gene expression.

Our study suggests that rather than initiating a de novo developmental programme in self-reactive T cells, Foxp3 takes advantage of preceding and coincidental features of T_r precursor cells probably facilitated by TCR signalling. This is accomplished through consolidation of the state of anergy and paracrine IL-2 dependence. Furthermore, Foxp3 amplifies and stabilizes expression of genes encoding cell surface or secreted molecules—like Fgl2, CD73, CD39, TRAIL or CTLA4—normally elaborated by conventional T cells upon TCR stimulation and capable of negative feedback regulation of T-cell activation 'in trans'. At the same time, Foxp3 enforces repression of TCR-activation-dependent immune response effector cytokines including IL-4, interferon-, tumour-necrosis factor-, IL-17 and IL-21. These observations might explain failure to identify a single non-redundant mechanism of suppression mediated by T_r cells. Finally, Foxp3 alters how T_r cells respond to environmental cues by modulating cell surface and signalling molecules to promote both T_r homeostasis and lineage stability. Essential adaptation to these signals results from Foxp3-dependent downregulation of PDE3B to support normal homeostasis and metabolic function and maintain a part of the T_r cell transcriptional programme. Our results suggest that, in resemblance of evolutionary processes, a lineage commitment factor may act in both an opportunistic and adaptive fashion during cellular differentiation by amplifying beneficial and correcting disabling features of precursor cells.

References

1. Fontenot, J. D. & Rudensky, A. Y. A well adapted regulatory contrivance: regulatory T cell development and the forkhead family transcription factor Foxp3. Nature Immunol. 6, 331–337 (2005) | Article |
2. van Santen, H. M., Benoist, C. & Mathis, D. Number of T reg cells that differentiate does not increase upon encounter of agonist ligand on thymic epithelial cells. J. Exp. Med. 200, 1221–1230 (2004) | Article | PubMed | ISI | ChemPort |
3. Hori, S., Nomura, T. & Sakaguchi, S. Control of regulatory T cell development by the transcription factor Foxp3. Science 299, 1057–1061 (2003) | Article | PubMed | ISI | ChemPort |
4. Khattri, R., Cox, T., Yasayko, S. A. & Ramsdell, F. An essential role for Scurfin in CD4⁺CD25⁺ T regulatory cells. Nature Immunol. 4, 337–342 (2003) | Article |
5. Wildin, R. S. & Freitas, A. IPEX and FOXP3: clinical and research perspectives. J. Autoimmun. 25, (suppl.)56–62 (2005) | Article | PubMed | ChemPort |
6. Fontenot, J. D., Gavin, M. A. & Rudensky, A. Y. Foxp3 programs the development and function of CD4⁺CD25⁺ regulatory T cells. Nature Immunol. 4, 330–336 (2003) | Article |
7. Hsieh, C. S., Zheng, Y., Liang, Y., Fontenot, J. D. & Rudensky, A. Y. An intersection between the self-reactive regulatory and nonregulatory T cell receptor repertoires. Nature Immunol. 7, 401–410 (2006) | Article |
8. Fontenot, J. D. et al. Regulatory T cell lineage specification by the forkhead transcription factor foxp3. Immunity 22, 329–341 (2005) | Article | PubMed | ISI | ChemPort |
9. Roncarolo, M. G. et al. Interleukin-10-secreting type 1 regulatory T cells in rodents and humans. Immunol. Rev. 212, 28–50 (2006) | Article | PubMed | ChemPort |
10. Ivanov, I. I. et al. The orphan nuclear receptor RORt directs the differentiation program of proinflammatory IL-17(+) T helper cells. Cell 126, 1121–1133 (2006) | Article | PubMed | ISI | ChemPort |
11. Williams, L. M. & Rudensky, A. Y. Maintenance of the Foxp3 dependent developmental program in mature regulatory T cells requires continued expression of Foxp3. Nature Immunol. (in the press).
12. Chan, C. W. et al. Soluble fibrinogen-like protein 2/fibroleukin exhibits immunosuppressive properties: suppressing T cell proliferation and inhibiting maturation of bone marrow-derived dendritic cells. J. Immunol. 170, 4036–4044 (2003) | PubMed | ISI | ChemPort |
13. Stumhofer, J. S. et al. Interleukin 27 negatively regulates the development of interleukin 17-producing T helper cells during chronic inflammation of the central nervous system. Nature Immunol. 7, 937–945 (2006) | Article |
14. Batten, M. et al. Interleukin 27 limits autoimmune encephalomyelitis by suppressing the development of interleukin 17-producing T cells. Nature Immunol. 7, 929–936 (2006) | Article |
15. Yegutkin, G. G., Henttinen, T., Samburski, S. S., Spychala, J. & Jalkanen, S. The evidence for two opposite, ATP-generating and ATP-consuming, extracellular pathways on endothelial and lymphoid cells. Biochem. J. 367, 121–128 (2002) | Article | PubMed | ISI | ChemPort |
16. Yegutkin, G. G., Samburski, S. S., Jalkanen, S. & Novak, I. ATP-consuming and ATP-generating enzymes secreted by pancreas. J. Biol. Chem. 281, 29441–29447 (2006) | Article | PubMed | ChemPort |
17. Ohta, A. & Sitkovsky, M. Role of G-protein-coupled adenosine receptors in downregulation of inflammation and protection from tissue damage. Nature 414, 916–920 (2001) | Article | PubMed | ISI | ChemPort |
18. Erdmann, A. A. et al. Activation of Th1 and Tc1 cell adenosine A2A receptors directly inhibits IL-2 secretion in vitro and IL-2-driven expansion in vivo. Blood 105, 4707–4714 (2005) | Article | PubMed | ISI | ChemPort |
19. Sheppard, F. R. et al. Structural organization of the neutrophil NADPH oxidase: phosphorylation and translocation during priming and activation. J. Leukoc. Biol. 78, 1025–1042 (2005) | Article | PubMed | ChemPort |
20. Torgersen, K. M., Vang, T., Abrahamsen, H., Yaqub, S. & Tasken, K. Molecular mechanisms for protein kinase A-mediated modulation of immune function. Cell. Signal. 14, 1–9 (2002) | Article | PubMed | ISI | ChemPort |
21. Zheng, Y. et al. Genome-wide analysis of Foxp3 target genes in developing and mature regulatory T cells. Nature doi: doi: 10.1038/nature05563 (in the press)
22. Zhang, W. & Colman, R. W. Conserved amino acids in metal-binding motifs of PDE3A are involved in substrate and inhibitor binding. Blood 95, 3380–3386 (2000) | PubMed | ChemPort |
23. Choi, Y. H. et al. Alterations in regulation of energy homeostasis in cyclic nucleotide phosphodiesterase 3B-null mice. J. Clin. Invest. 116, 3240–3251 (2006) | Article | PubMed | ChemPort |

Supplementary Information

Supplementary information accompanies this paper.

Acknowledgements

We are grateful to all members of the Rudensky laboratory for discussions. We also thank K. Forbush, L. Karpik and T. Chu for help. This work was supported by grants from the Arthritis Foundation and the Leukemia and Lymphoma Society (M.A.G.) and grants from the National Institutes of Health (A.Y.R.). A.Y.R. is a Howard Hughes Medical Institute Investigator.

Competing interests statement

The authors declare no competing financial interests.

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated.