Cell-intrinsic and -extrinsic control of Treg-cell homeostasis and function revealed by induced CD28 deletion


  • Tea Gogishvili,

    1. Institute for Virology and Immunobiology, University of Würzburg, Würzburg, Germany
    Search for more papers by this author
  • Fred Lühder,

    1. Department of Neuroimmunology, Institute for Multiple Sclerosis Research and The Hertie Foundation, University Medical Center Göttingen, Göttingen, Germany
    Search for more papers by this author
  • Sandra Goebbels,

    1. Max Planck Institute of Experimental Medicine, Neurogenetics, Göttingen, Germany
    Search for more papers by this author
  • Sandra Beer-Hammer,

    1. Department of Pharmacology and Experimental Therapy, Institute of Experimental and Clinical Pharmacology and Toxicology, Eberhard Karls University Hospitals and Clinics, University of Tübingen, Tübingen, Germany
    2. Interfaculty Center of Pharmacogenomics and Drug Research, University of Tübingen, Tübingen, Germany
    3. Institute of Medical Microbiology and Hospital Hygiene, Clinical Centre of Heinrich Heine University, Düsseldorf, Germany
    Search for more papers by this author
  • Klaus Pfeffer,

    1. Institute of Medical Microbiology and Hospital Hygiene, Clinical Centre of Heinrich Heine University, Düsseldorf, Germany
    Search for more papers by this author
  • Thomas Hünig

    Corresponding author
    • Institute for Virology and Immunobiology, University of Würzburg, Würzburg, Germany
    Search for more papers by this author

Full Correspondence: Dr. Thomas Hünig, Institute for Virology and Immunobiology, University of Würzburg, Versbacher str. 7, 97078 Würzburg, Germany

Fax: +49-931-49-243

e-mail: huenig@vim.uni-wuerzburg.de

See accompanying Commentary by Sansom and Walker


While the requirement for CD28 and its ligands for the generation and function of “natural” (n)Treg cells is well established, it has not been possible yet to investigate cell-intrinsic effects after interrupted CD28 expression. Here, we demonstrate a selective loss of Treg cells after disruption of the CD28 gene. The decline in Treg-cell number was accompanied by reduced homeostatic proliferation, probably due to lack of costimulation during self-antigen recognition, and by impaired Treg-cell function including downregulation of CTLA-4. The decline in Treg-cell number was unaffected by thymectomy or by the presence of CD28 expressing T cells within the same animal, indicating that impairment of peripheral homeostasis and function of nTreg cells by CD28 deletion is cell-intrinsic. In contrast, downregulation of CD25, the α chain of the IL-2R, did not occur in the presence of WT T cells, indicating that its expression does not depend on CD28 signals in cis.


CD28 is the main costimulatory molecule on T cells. Its ligands B7–1 (CD80) and B7–2 (CD86) are expressed mainly on B cells, DCs, and thymic medullary epithelial cells [1], indicative of a role both during thymic T-cell selection and peripheral immune function.

Mice deficient in either CD28 or B7 have normal numbers of conventional CD4+ and CD8+ T cells but strongly reduced numbers of Treg cells in both thymus and periphery [2-4]. Thymic Treg-cell generation requires expression of CD28 on the Treg cells themselves [4-6], suggesting a need for costimulation during positive selection by self-antigens. Moreover, mice lacking components of the IL-2/IL-2R system suffer from autoimmunity due to a strict requirement for IL-2 in Treg-cell homeostasis and function [7, 8]. Finally, mixed BM radiation chimeras constructed from WT, CD28, and IL-2-deficient mice have demonstrated a dual requirement for CD28 in conventional T cells (for IL-2 production) and the Treg cells themselves [5]. While these studies had to rely on constitutively CD28-deficient mice, making it impossible to separate thymic from peripheral requirements, they were supplemented by reports in which Treg cells from WT mice were transferred into B7-deficient mice, and by experiments using in vivo ligand blockade [6]. Results obtained in such systems are, however, not conclusive because they additionally deprive CTLA-4 (CD152), the closest relative of CD28, of the same ligands [9, 10], to which it binds with much higher avidity [11] resulting in the transmission of immunosuppressive signals into and removal of these ligands from APCs [12]. In addition, the binding of recombinant CTLA-4-Ig or of B7-specific mAb to DCs during ligand blockade also induces immunosuppressive genes such as IDO in these cells [13], further complicating the issue.

In the present study, we have therefore reinvestigated the role of CD28 in the homeostasis and functional activity of peripheral Treg cells using a newly developed inducible CD28 Knock-Out (iCD28KO) mouse system in which effects on thymic maturation can be clearly separated from those on survival, phenotype, and activity in the peripheral immune system.

Results and discussion

To determine the role of CD28 in the responsiveness and homeo-stasis of peripheral T cells that had matured in the presence of this costimulator, an inducible Cre-recombinase was introduced into mice with one deleted and one “floxed” CD28 locus (Supporting Information Fig. 1A). Treatment of such inducibly CD28-deleting (iCD28KO) mice with tamoxifen (TM) for four consecutive days effectively inactivated the CD28 gene both in the thymus and in peripheral lymphoid organs as measured by flow cytometry of CD28 surface expression 1 week later (Supporting Information Fig. 1B).

Homeostasis of conventional and Treg cells after CD28 deletion

We used induced CD28 deletion to determine the importance of CD28-mediated costimulation for the maintenance of stable T-cell pools. Although the ablation of CD28 expression did not remarkably impair the absolute numbers of CD4 and CD8 T cells in the periphery (Supporting Information Fig. 1C), a drastic loss of Treg cells was observed already 1 week after induced CD28 deletion (Fig. 1A), with further decline in 3 weeks to similar frequencies as found in constitutively CD28-deficient mice [4, 6]. Furthermore, we analyzed the expression levels of Foxp3, CD25, and CTLA-4, which are decisively involved in Treg-cell lineage specification, functional activation, and execution of suppressor function, respectively. While there was no significant reduction in Foxp3 expression, CTLA-4, and CD25 were reduced by 25–30% as compared to Treg cells of oil treated mice (Fig. 1B).

Figure 1.

Induced disruption of CD28 impairs Treg-cell homeostasis and function. (A) Reduced Treg-cell frequencies in iCD28KO mice after TM-treatment (n = 6–8 mice per group per experiment). Representative flow cytometry plots of one out of three experiments are shown gated on LN CD4+ T cells of WT, CD28KO and iCD28KO mice, after 1 or 3 weeks of TM or Oil administration (left). Compiled data (each symbol represents data from an individual mouse; horizontal lines the mean). (B) Expression levels of CTLA-4, Foxp3, and CD25 on LN and spleen Treg cells of iCD28KO mice 3 weeks after CD28 disruption. Data are normalized to Oil-treated controls (n = 6–8 mice per group) and are mean + SD of one out of three independent experiments. (C) Reduced suppressive function of Treg cells in the absence of CD28. WT conventional T (Tconv) cells were cocultured with Treg cells of indicated mice at the stated concentrations and percent of suppression calculated from average cell divisions normalized to proliferation of conventional T cells. Statistics are calculated in comparison to Oil-administered groups. Data are shown as mean + SD of triplicates from one out of three experiments. (*p < 0.05, **p < 0.01, ***p < 0.001, unpaired t-test).

Treg-cell function after CD28 deletion

To investigate whether interruption of CD28 expression also affects the suppressive function of Treg cells, they were isolated from TM-treated iCD28KO mice 3 weeks after treatment and tested in vitro. The suppressive activity of Treg cells from CD28 hemizygous mice was comparable to that of WT Treg cells, whereas the interruption of CD28 expression 3 weeks earlier attenuated their suppressive capacity, with a similar degree of impairment as is seen with conventional CD28KO Treg cells (Fig. 1C). An earlier study on conventional CD28KO mice suggested that CD28 is dispensable for Treg-cell activity in vitro [14]. However, suppressive activity was measured at a Treg:conventional T cell ratio of 1:1, which yields almost complete suppression and hence is not suitable for quantitative comparison. One potential mechanism of attenuation of Treg-cell suppressive function may be the reduced expression of CTLA-4 on Treg cells upon interruption of CD28 expression (Fig. 1B) [14, 15].

Cell-intrinsic effects of induced CD28 deletion on Treg-cell number and function

In iCD28KO mice, expression of the ER-Cre recombinase fusion protein is under the control of the ubiquitously expressed Gt(ROSA)26Sor gene. Accordingly, conditional gene deletion affects the function of all cell types expressing CD28. Since besides postulated Treg-cell intrinsic effects of CD28 costimulation such as homeostatic proliferation and prevention of cell death [16-19], CD28-dependent IL-2 production by conventional CD4 T cells is crucial for Treg-cell homeostasis and function [5, 6], we wanted to identify whether the effects observed after induced CD28 deletion are due to cell intrinsic or extrinsic effects.

To address this question, CD28 deletion was induced in mixed BM chimeras generated by reconstituting lethally irradiated RAG-deficient recipient mice with BM cells from iCD28KO and Thy1.1/CD28+/– donors (Supporting Information Fig. 2A). In this setting, approximately half of the T cells maintain CD28 after TM treatment, leaving a sufficient source for IL-2 and other potentially transacting factors [5, 8]. Analysis of CD28 expression 3 weeks after TM-induced gene deletion indicated that CD28 was effectively abrogated in iCD28KO cells (Supporting Information Fig. 2B).

To investigate the CD28-dependent cell intrinsic effect on the size of the Treg-cell compartment, frequencies of CD25+Foxp3+ T cells were measured 1 and 3 weeks after induced CD28 gene deletion. In addition, before TM treatment, Treg-cell representation was identical among CD4 T cells derived from deletable and nondeletable BM (data not shown), the loss of CD28 significantly reduced the frequency of Foxp3+ Treg cells already 1 week after gene deletion (Fig. 2A). The numbers of nondeletable Treg cells in the same animals remained stable. Thus, besides dependence on IL-2 in trans, provided by conventional T cells, Treg cells require CD28 to maintain their cell numbers, presumably by costimulated recognition of self-antigens on professional APCs.

Figure 2.

Cell intrinsic effect of CD28 deletion on Treg cells. (A) Frequencies of CD25+Foxp3+ T cells 1 and 3 weeks after TM treatment in the LNs and spleen of lethally irradiated RAG-deficient recipients reconstituted with mixed BM cells from iCD28KO and Thy1.1/CD28+/− mice, depicted as mean + SD (n = 5 mice per group) of one out of two experiments. (B) Expression levels of CTLA-4, Foxp3, and CD25 within LN (upper panel) and splenic (lower panel) Treg cells of iCD28KO population after 1 and 3 weeks of CD28 deletion, normalized to levels on Treg cells of Thy1.1/CD28+/− population. iCD28KO and Thy1.1/CD28+/− populations are statistically compared. Results of one out of two experiments are shown (n = 5 mice per group per experiment). (*p < 0.05, **p < 0.01, ***p < 0.001, unpaired t-test).

We further characterized the cell-intrinsic effects of CD28 deletion on Treg-cell phenotype. In contrast to the small but significant reduction in CD25 expression observed after CD28 deletion in iCD28KO mice (Fig. 1B), we found that deletion of CD28 in these chimeric mice did not significantly affect CD25 levels (Fig. 2B). Considering the role of IL-2 in the upregulation of CD25 expression [20, 21], this difference is probably due to the availability of IL-2 in this setting from Thy1.1/CD28+/– cells. Our findings thus contradict the notion that CD28-mediated costimulation is required for the maintenance of CD25 expression in a cell-intrinsic fashion, which had been concluded from CD25 downregulation on Treg cells after B7 blockade, and from maintenance of normal CD25 levels by IL-2 loaded Treg cells transferred into CD28-deficient mice [6]. In contrast to CD25 expression, CD28 deletion, however, markedly affected the level of CTLA-4 expressed by Treg cells, which was about twofold reduced within the TM-treated cell compartment (Fig. 2B). Given the importance of CTLA-4 for Treg-cell function [12, 14, 15], the relevance of this cell-intrinsic effect of CD28-mediated costimulation is obvious, if not unexpected from the known regulation of CTLA-4 expression by CD28 [22, 23]. Finally, although the Foxp3 promoter contains a CD28 responsive element [24], we only found a slight and transient reduction in expression levels of this Treg-cell lineage specification factor.

Decline in Treg-cell numbers after CD28 deletion is independent of thymic export

In order to test the possibility that the peripheral decline in nTreg-cell numbers after termination of CD28 expression is due to an interruption of thymic generation and export of Treg cells, the conditional KO mice were thymectomized before TM-induced gene deletion. As an alternative experimental setting, total splenic and lymph node cells from iCD28KO and CD28 hemizygous congenic mice (1:1) were transferred into RAG−/− recipients that were then subjected to TM treatment.

Analysis of thymectomized mice 3 weeks after CD28 deletion showed a similar decline in peripheral Treg cells regardless of previous thymectomy (Fig. 3A). Furthermore, CD28 deletion after transfer of peripheral T cells into RAG−/− mice revealed the same degree of Treg-cell reduction in the iCD28KO derived subsets, whereas the cotransferred CD28 hemizyogous cells were unaffected by TM treatment (Fig. 3C). Together, these results indicate cell intrinsic requirement for CD28 signaling in the maintenance of the peripheral Treg-cell pool and argue against defective thymic export as the main factor behind Treg-cell decline after induced CD28 deletion.

Figure 3.

Decline in Treg-cell numbers after CD28 disruption is thymus independent and associated with reduced cell proliferation. (A, B) iCD28KO mice (n = 5–6 per group per experiment) were (sham-) thymectomized at 8–10 weeks, treated with TM or Oil 1 week later and analyzed after 3 weeks. (A) Treg-cell frequencies (mean +SD) are shown. (B) Proliferation (mean +SD) of Foxp3+ cells as a percent of Ki-67+ cells. Oil- and TM-treated groups were compared for statistically significant differences. (C, D) LN and spleen cells from iCD28KO and Thy1.1/CD28+/− mice were mixed 1:1 and 1 × 108 cells were injected to RAG−/– mice. Recipients were analyzed 3 weeks after TM treatment. (C) Treg-cell frequencies (mean +SD) and (D) proliferation (mean +SD) as percent of Ki-67 expression among Foxp3+ cells were measured in the spleen and LNs. iCD28KO and Thy1.1/CD28+/− populations are statistically compared. (A–D) Data are from one out of two independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, unpaired t-test.

Since Treg cells undergo constant turnover [25], we also analyzed the frequency of proliferating cells via the expression of the nuclear proliferation marker Ki-67. Irrespective of thymectomy, Treg cells from iCD28KO mice had a markedly (about twofold) reduced proliferation rate if CD28 expression was interrupted 3 weeks earlier (Fig. 3B). Together with the observed decline in the iCD28KO derived Treg-cell population recovered from RAG−/− recipients reconstituted with peripheral lymphoid cells of iCD28KO and CD28 hemizygous mice (Fig. 3C), the proliferative capacity of iCD28KO Foxp3+ Treg cells was similarly impaired in the lymphopenic host as observed in inducibly deleted CD28 mice (Fig. 3B and D).

As expected, conventional CD4 T cells also exhibited increased proliferation when introduced into the lymphopenic environment of RAG−/− mice (compare Supporting Information Fig. 3A and B), which was similarly reduced after interruption of CD28 expression. In keeping with the much lower turnover of conventional CD4 T cells under lymphoreplete conditions and the marginal effects of CD28 deletion on their absolute numbers (Supporting Information Fig. 1C), however, only a small reduction in proliferation was observed after interruption of CD28 expression.

Concluding remarks

In the present study, we investigated the role of CD28 costimulation in the homeostasis and functional activity of the peri-pheral T-cell compartment for the first time by interruption of this pathway in CD28 conditional KO mice. In contrast to previously described studies using the conventional KO mouse model [5, 6], the crucial advantage of this system is that thymic maturation and differentiation of Treg cells remain intact, so that the importance of CD28-mediated costimulation for turnover and function can be monitored within the mature cells in the periphery.

Our present findings consolidate the central role of CD28 within the regulatory circuit linking activation of conventional T cells and Treg cells: by driving IL-2 production in conventional T cells, it not only provides their autocrine growth signal but also allows them to sense an ongoing immune response and to respond to it [26]; by costimulating Treg-cell turnover in response to self (nTreg cells) and foreign (iTreg cells) antigens, it allows them to respond to the same activated APCs which drive the effector T-cell response; and by upregulating Treg-cell function including CTLA-4 expression, it supports termination of conventional T-cell and Treg-cell activation through the deactivation of APCs, among other suppressive mechanisms.

Materials and methods


RAG−/− and CD90.1-congenic C57BL/6 (B6) CD28+/− mice (Thy1.1/CD28+/− mice) were bred at the animal facility of the Institute for Virology and Immunobiology, University of Würzburg. All animals used for experiments were between 6 and 12 weeks of age. Experiments performed according to the Bavarian state regulations for animal experimentation and approved by the Regierung von Unterfranken as the responsible authority.

Inducible gene deletion in CD28−/floxCre+/− mice

To generate the inducible knockout mice rosa26CreER(T2) mouse strain was used, where the Cre recombinase-fused estrogen receptor (ER) is under the control of the ubiquitously expressed Gt(ROSA)26Sor gene. iCD28KO mice (CD28−/floxCre+/−) were generated by crossing CD28KOCre+/+ mice with CD28flox/flox mice. Efficient ablation of CD28 was achieved by daily forced feeding of 2.5 mg TM in sunflower oil (Sigma-Aldrich) for four consecutive days (Supporting Information Fig. 1A). Gene deletion was monitored by surface staining of CD28 on T cells (Supporting Information Fig. 1B).

Flow cytometry and antibodies used for flow cytometric analyses and for purification of T cells

The following antibodies were used for flow cytometry: CD4, CD25, Thy1.1, Ki-67, CTLA-4 (BD Pharmingen); Foxp3 (eBio-science); CD28 (BioLegend). Intracellular staining of Foxp3, Ki-67, and CTLA-4 was performed according to the manufacturer's instructions. Acquisition performed on a BD™ LSR-II and data were analyzed using FlowJo software (TreeStar Inc).

Purification of T cells for suppression assay

CD4+ T cells were negatively selected by magnetic separation (Miltenyi Biotech). PE-labeled CD25+ CD4 T cells were isolated by BD FACSAria cell sorter (purity 95–98%). WT CD4+CD25 responder T cells (5 × 104) were CFSE labeled and cultured for 3 days in U-bottomed 96-well plates with a various dilutions of purified CD4+CD25+ Treg cells from CD28KO, iCD28KO (+TM), or iCD28KO (+Oil) mice in the presence of soluble anti-CD3 (1 μg/mL) and irradiated WT splenic APCs (20 Gy; 2 × 105). Proliferation was assayed by CFSE dilution.


Anesthetized mice were ventilated by intubation and the thymus was exposed through a suprasternal transverse incision. After suturation of supplying vessels the thymus was resected.

Statistical analysis

For statistical comparison of results, we used the unpaired t-test. Values of p < 0.05 were considered statistically significant.


We thank Ulrich Hofmann and Charlotte Dienesch for thymectomy of mice and Christian Bauer for expert technical assistance. This work was supported by Deutsche Forschungsgemeinschaft through SFB-TRs 43 (to F.L.) and 52 (to T.H.), and by Bundesministerium für Bildung und Forschung through Competence Network Multiple Sclerosis (to F.L.).

Conflict of interest

The authors declare no financial or commercial conflict of interest.


inducible CD28 Knock-Out mice