What turns on Foxp3?Harald von Boehmer & Jens Nolting

Inducible regulatory T cells respond to TGF-β by upregulating Foxp3 expression. Tone and colleagues identify an enhancer site in Foxp3 that binds transcription factors Smad3 and NFAT, suggesting a means by which TGF-β regulates Foxp3 expression.

Harald von Boehmer and Jens Nolting are with the

Dana-Farber Cancer Institute, Harvard Medical

School, Boston, Massachusetts 02115, USA.

e-mail: harald_von_boehmer@dfci.harvard.edu

interleukin 10 and interleukin 2. Those obser-vations initiated a search for Foxp3 enhancers, as Foxp3 promoter activity was strong in Foxp3+ cells in the presence of an SV40 enhancer. The search resulted in the identification of an evo-lutionarily conserved enhancer region (posi-tions +1988 to +2738) with NFAT and Smad binding sites. The authors found that binding of NFAT and Smad3 to the enhancer resulted in considerable enhancement of Foxp3 promoter activity. Synergistic action of NFAT as well as Smad3 was required for histone H4 acetylation; the authors also found H4 acetylation of the conserved enhancer region in ex vivo Foxp3+ Treg cells but not Foxp3– T cells. Of interest, binding of Smad3 to the enhancer occurred only at early stages of TGF-β-dependent Foxp3 induction, and it seemed to be dispens-able at later stages, whereas NFAT binding was required throughout the entire 24-hour period of observation.

At first glance, these results seem to be somewhat at odds with the idea that a defi-

ciency in TGF-β signaling has no effect on the intrathymic generation of Treg cells. It could be argued, however, that activins or bone morphogenetic factors of the TGF-β ligand family can still work intrathymically, even in the presence of a dominant negative TGF-β receptor II, to phosphorylate Smad proteins. Furthermore, the apparent requirement for TGF-β signaling to maintain peripheral Treg cells3,4 seems to be not very compatible with the idea that Smad3 is required only at early stages of Foxp3 upregulation. Here it could be argued that the maintenance of Treg cells requires TGF-β for reasons other than upregulation of Foxp3. Also, the continual requirement for binding of NFAT to the Foxp3 enhancer10 could be in conflict with the idea that NFAT binds to Foxp3 protein, which con-ceivably could interfere with its binding to the Foxp3 enhancer.

Despite those reservations, the authors would like to conclude that binding of NFAT as well as Smad3 to the Foxp3 enhancer is mandatory for

The Foxp3 transcription factor, through its expression in regulatory T cells (Treg

cells), makes an essential contribution to the prevention of autoimmune disease1. For this reason, the modalities involved in regulating Foxp3 expression in thymic and peripheral Treg cells are of considerable interest. In this issue of Nature Immunology, Tone et al. report that binding of the transcription factors NFAT and Smad3 to a Foxp3 enhancer is important in enhancing Foxp3 promoter activity and con-clude that the synergistic action of NFAT and Smad3 is mandatory for Foxp3 expression2.

Foxp3-expressing Treg cells can be generated by antigenic stimulation intrathymically and extrathymically. The intrathymic generation can proceed in the absence of transforming growth factor-β (TGF-β)3 and in the presence of a dominant negative TGF-β receptor II (ref. 4); that is, by TGF-β-independent mechanisms. In contrast, TGF-β signaling is needed to main-tain normal Treg cell numbers in peripheral lymphoid tissue3. The conversion of naive T cells into Treg cells in peripheral lymphoid tissue is enhanced by TGF-β signaling in the converting cells5,6.

Relatively little is known about the factors involved in Foxp3 expression. Binding of tran-scription factors Sp1, NFAT, AP1 (ref. 7) and STAT5 (ref. 8) to the Foxp3 promoter has been reported2. So far, there has been no clue as to how TGF-β signaling may contribute to Foxp3 expression, as binding of Smad proteins to the Foxp3 promoter has not been noted. Indeed, Tone et al. report that the promoter does not contain a consensus sequence for Smad bind-ing2. Using a luciferase assay, the authors report that the Foxp3 promoter has only weak activ-ity in Foxp3+ mouse lymphoblast EL4 T cells generated by T cell receptor and CD28 stimula-tion in the presence of TGF-β in vitro2,9. In fact, the promoter binds Sp1 and NFAT only weakly relative to the binding of these transcription factors to the promoters of the genes encoding

Foxp3

NFAT AP1 Sp1 STAT5 NFAT Smad3 STAT5 CREB-ATF

Foxp3 promoter Enhancer 1 Enhancer 2 ATG

Exon 1

Foxp3-promoter Enhancer 1 Enhancer 2

TGF-β

TGF-β receptor

a

b

Smad2,3

Smad2,3

Cytoplasm

Nucleus

P

Smad3 P

Smad4 NFAT

NFAT NFAT STAT5 STAT5 AP1 Sp1 CREB-ATF Foxp3

TCR

Figure 1 Activation of Foxp3 enhancer 1. (a) The Foxp3 promoter and enhancer, with consensus binding sites for transcription factors NFAT, AP1, Sp1, STAT5, Smad3 and CREB-ATF. (b) Binding of transcription factors resulting from antigenic T cell stimulation and TGF-β signaling. TCR, T cell antigen receptor.

Kat

ie R

is-V

icar

i

NATURE IMMUNOLOGY VOLUME 9 NUMBER 2 FEBRUARY 2008 121

NEWS AND V IEWS©

2008

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y

Foxp3 expression. The fact that such binding causes H4 acetylation and that such acetyla-tion is only found in Foxp3+ cells constitutes a good argument for this claim. However, these results do not rule out that there is a Smad3-independent and therefore TGF-β ligand–independent pathway of Foxp3 induc-tion. Despite this cautionary note, the authors seem to have established a pathway of TGF-β-dependent induction of Foxp3 expression. It has been known for some time that in vitro TGF-

β-dependent induction does not yield a very stable phenotype of Foxp3+ Treg cells and thus other factors may contribute in vivo as well. At this point, the results of Tone et al.2 raise the question of whether there are Smad3-inde-pendent pathways of Foxp3 induction, which presumably can be tested by a ‘knock-in’ Foxp3 allele with a mutated Smad3 binding site.

1. Brunkow, M.E. et al. Nat. Genet. 27, 68–73 (2001).2. Tone, M. et al. Nat. Immunol. 9, 194–202 (2008).

3. Marie, J.C., Letterio, J.J., Gavin, M. & Rudensky, A.Y. J. Exp. Med. 201, 1061–1067 (2005).

4. Huber, S. et al. J. Immunol. 173, 6526–6531 (2004).

5. Kretschmer, K. et al. Nat. Immunol. 6, 1219–1227 (2005).

6. von Boehmer, H. J. Exp. Med. 204, 1737–1739 (2007).

7. Burchill, M.A., Yang, J., Vogtenhuber, C., Blazar, B.R. & Farrar, M.A. J. Immunol. 178, 280–290 (2007).

8. Kim, H.P. & Leonard, W.J. J. Exp. Med. 204, 1543–1551 (2007).

9. Chen, W. et al. J. Exp. Med. 198, 1875–1886 (2003).

10. Wu, Y. et al. Cell 126, 375–387 (2006).

CD160 and BTLA: LIGHTs out for CD4+ T cellsJonathan Kaye

The T cell costimulatory protein LIGHT and coinhbitory protein BTLA share a common ligand, HVEM. Now CD160 is also shown to bind HVEM and deliver a potent inhibitory signal to CD4 T cells.

Jonathan Kaye is in the Department of Immunology,

IMM-4, The Scripps Research Institute, La Jolla,

California 92037, USA.

e-mail: jkaye@scripps.edu

Although the T cell antigen receptor (TCR) is responsible for imparting specificity

to the initial activation of the cell, it is the assortment of additional cell surface proteins acting in a non–antigen-specific way that is central to the regulation of both the quan-titative and qualitative aspects of the cell’s response. Such modifiers can act in either a positive or a negative way in conjunction with TCR-mediated activation and thus have been called ‘costimulators’ or ‘coinhibitors’, respectively. Because T cells recognize foreign antigen in the form of peptides bound to self major histocompatibility complex (MHC) molecules on the surfaces of other cells, the T cell immune response can be ‘fine tuned’ on two levels: by the specific array of coinhibi-tory and costimulatory proteins expressed on the T cell surface and by the specific array of ligands for these receptors expressed by the antigen-presenting cell. In this issue of Nature Immunology, Freeman and colleagues add CD160 to the list of coinhibitory mol-ecules and identify an unexpected dimension of an already complex network of competing activating and inhibitory signals received by T cells through a single ligand1.

The monoclonal antibody BY55 was derived from mice immunized with a human leukemic natural killer (NK) cell line in an attempt to develop markers for cytotoxic cells2. Almost a decade ago, the gene was

identified that encodes the protein now des-ignated CD160 (ref. 3), which is recognized by BY55. Expression of CD160, a member of the immunoglobulin ‘superfamily’ of pro-teins, is reported to be expressed by NK cells, NKT cells, intraepithelial T cells, γδ TCR+ T cells, and memory-phenotype, activated and effector CD8+ T cells. In terms of func-tion, work has centered on the role of CD160 in enhancing NK or CD8 T cell activation. Such effects have been attributed to the abil-ity of CD160 to bind classical and nonclas-sical MHC class I molecules, although with apparent low affinity, requiring clustering of MHC class I molecules or overexpression of CD160 or MHC class I for detection of the interaction.

The findings reported here1 stem from the production of additional monoclonal antibodies to CD160 and a reevaluation of the pattern of CD160 expression on human peripheral blood lymphocytes. Thus, as with CD8+ T cells, the authors find that a minor subset of CD4+ T cells expresses CD160. The CD160+ CD4+ T cells include subsets of T cells with a memory or activated phe-notype. Unexpectedly, however, they find that coimmobilization of anti-CD160 with an activating mix of monoclonal antibodies to CD3 and CD28 (targeting the TCR com-plex and the main costimulatory pathway) profoundly inhibits total CD4+ T cell pro-liferation and cytokine production.

Paradoxically, the proliferation of even purified CD160– CD4+ T cells was inhib-ited by antibody to CD160 (anti-CD160), although a higher concentration of inhibi-tory antibody was required. The investiga-

tors attribute this unexpected result to the upregulation of CD160 during T cell activa-tion. However, this may not be the whole story. Although they did not find substantial staining with anti-CD160 on bulk CD4+ T cells until 3 days after activation, much ear-lier events, such as induction of interleukin 2 mRNA and CD3ζ phosphorylation, were inhibited by anti-CD160. The interpretation of these results is complicated somewhat by the use of total CD4 T cells in these experi-ments. Thus, the contribution by subsets of CD160+ CD4+ T cells needs to be clarified, particularly as activated- and memory-phe-notype cells could account for a dispro-portionate fraction of the total response. Nevertheless, whether a very small amount of CD160 on the naive CD4+ T cell surface is nonetheless sufficient to deliver an inhibi-tory signal must also be considered. If con-firmed, this raises a strong note of caution for the reliance on flow cytometry as the sole means of determining patterns of expression of proteins and, by extension, susceptibility to biological effects.

What is the basis for the CD160 inhibition of T cell activation? Using a global micro-array analysis approach, these investigators show that CD160-mediated coinhibition inhibits the expression of many mediators of T cell activation, including cytokines, cytokine receptors, and nutrient transporters such as the amino acid exchanger SLC7A5. Notably, these results are reminiscent of sim-ilar properties of another coinhibitory cell surface protein, B and T lymphocyte attenu-ator (BTLA)4. Thus, high expression of CD160 on a subset of CD4+ T cells, delayed

122 VOLUME 9 NUMBER 2 FEBRUARY 2008 NATURE IMMUNOLOGY

NEWS AND V IEWS©

2008

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y