Research ArticleAutoimmunity

CD4+ T Cells from IPEX Patients Convert into Functional and Stable Regulatory T Cells by FOXP3 Gene Transfer

See allHide authors and affiliations

Science Translational Medicine  11 Dec 2013:
Vol. 5, Issue 215, pp. 215ra174
DOI: 10.1126/scitranslmed.3007320


In humans, mutations in the gene encoding for forkhead box P3 (FOXP3), a critically important transcription factor for CD4+CD25+ regulatory T (Treg) cell function, lead to a life-threatening systemic poly-autoimmune disease, known as immune dysregulation, polyendocrinopathy, enteropathy, X-linked (IPEX) syndrome. Severe autoimmunity results from the inborn dysfunction and instability of FOXP3-mutated Treg cells. Hematopoietic stem cell transplantation is the only current curative option for affected patients. We show here that when CD4+ T cells are converted into Treg cells after lentivirus-mediated FOXP3 gene transfer, the resulting CD4FOXP3 T cell population displays stable phenotype and suppressive function, especially when naïve T cells are converted. We further demonstrate that CD4FOXP3 T cells are stable in inflammatory conditions not only in vitro but also in vivo in a model of xenogeneic graft-versus-host disease. We therefore applied this FOXP3 gene transfer strategy for the development of a Treg cell–based therapeutic approach to restore tolerance in IPEX syndrome. IPEX-derived CD4FOXP3 T cells mirrored Treg cells from healthy donors in terms of cellular markers, anergic phenotype, cytokine production, and suppressive function. These findings pave the way for the treatment of IPEX patients by adoptive cell therapy with genetically engineered Treg cells and are seminal for future potential application in patients with autoimmune disorders of different origin.


Primary immunodeficiency disorders (PIDs) are rare genetic diseases of the immune system primarily characterized by recurrent infections and often associated with autoimmune manifestations and malignancies (1). Among those, immune dysregulation, polyendocrinopathy, enteropathy, X-linked (IPEX) syndrome is the only immunodeficiency characterized by the loss of function of thymus-derived CD4+CD25+ regulatory T (tTreg) cells (2), which are devoted to the control of immune responses to self and foreign antigens (Ags) (3). In IPEX patients, several mutations disrupting the gene encoding for the Treg-specific transcription factor forkhead box P3 (FOXP3) have been identified as responsible for disease development (4). Although FOXP3 mutations do not prevent thymic development and the emergence of peripheral tTreg cells (5, 6), FOXP3-mutated Treg-like cells do not gain suppressive ability (79) but are rather prone to acquire effector T (Teff) cell functions under inflammatory stimuli (6). In IPEX syndrome, loss of peripheral regulation results in the development of a devastating multiorgan autoimmunity, mainly characterized by early-onset life-threatening enteropathy, type 1 diabetes (T1D), and eczema (10, 11). The disease can be fatal in early infancy, if not promptly diagnosed and treated with hematopoietic stem cell transplantation (HSCT), the only curative treatment so far available (10). Data from murine models demonstrated that partial bone marrow transplant (BMT) or adoptive Treg cell transfer is sufficient to control the development of the scurfy phenotype (12), suggesting that the presence of few functional Treg cells could be sufficient to control the development of autoimmunity in IPEX syndrome.

The recent advances in gene therapy technologies have greatly improved our capability of genetically reprogramming mammalian cells (13). Gene therapy with engineered stem cells has become a safe and effective therapeutic strategy alternative to BMT for monogenic PIDs (1416). Gene transfer–based therapy with selected lymphocyte subsets has also proven to be feasible and safe in the past for adenosine deaminase–deficient severe combined immune deficiency (ADA-SCID) (16) and more recently for the treatment of cancer (17).

Because of the great potential of Treg cells as modulators of immunity, Treg-based therapies with freshly isolated or expanded Treg cells have been translated in clinical practice with the aim of preventing graft-versus-host disease (GVHD) in patients undergoing allogeneic HSCT (1821), inhibiting rejection in solid organ transplantation, and controlling autoimmunity in patients with T1D (22). The recently completed trials paved the way to wider application of Treg cells as therapeutic agents in autoimmune diseases (23), although many questions, such as Ag specificity, cell dose, and stability, are issues that need to be addressed.

We and others previously demonstrated that ectopic overexpression of FOXP3 in conventional CD4+ T cells from healthy donors (HDs) can produce potent suppressor cells in vitro, with low proliferative potential and reduced cytokine production (2, 2426), indicating that Treg cells can be de novo–generated by gene transfer approaches. Whether the latter CD4FOXP3 Treg-like cells maintain their features when challenged by an inflammatory environment, making them suitable for application in clinical protocols, has not yet been clarified.

On the basis of these premises, we explored the possibility to restore immune regulation in FOXP3-mutated T cells by wild-type FOXP3 gene transfer, with the ultimate goal to develop it as a therapy suitable for IPEX patients. Here, we demonstrate the feasibility of conversion of IPEX conventional CD4+ T cells into fully functional Treg-like cells by lentiviral vector (LV)–mediated overexpression of wild-type FOXP3. The resulting transduced CD4FOXP3 Treg cells have potent suppressor function both in vitro and in vivo, because they are able to control xenogeneic GVHD (xeno-GVHD), and display enhanced stability when generated from naïve T cells. We believe that the clinical translation of this approach could be beneficial not only for IPEX patients but also for patients with pathologies that stem from insufficient Treg cell function or number.


CD4FOXP3 T cells produce low levels of cytokines

We generated CD4FOXP3 T cells by transduction of conventional CD4+ T cells isolated from the peripheral blood of HDs with an LV encoding for full-length FOXP3 and ΔNGFR under the control of a ubiquitous promoter (LV-EF1α-FOXP3). As control, T cells were transduced with an LV encoding for ΔNGFR only (control LV). As previously described, transduction of T cells with LV-EF1α-FOXP3 allows generation of potent suppressor cells (hereafter indicated as CD4FOXP3 T cells) (24). Because T cell plasticity represents a relevant safety issue for the development of Treg-based cellular therapies, we tested the stability of CD4FOXP3 T cells exposed to an inflammatory environment.

We first evaluated the cytokine production profile of CD4FOXP3 T cells generated from HD conventional T cells. Both T helper 1 (TH1) [interferon-γ (IFN-γ) and interleukin-2 (IL-2)] and TH2 (IL-5) cytokine production was drastically reduced (a minimum of five donors for each cytokine were tested, CD4NGFR versus CD4FOXP3, P < 0.05 for all cytokines) (Fig. 1A and table S1). Production of IL-22, a TH17-related cytokine recently described as a direct target of FOXP3 (27), was also reduced in CD4FOXP3 T cells [n = 4, P = not significant (NS)], whereas IL-17A production was highly variable (n = 14) (Fig. 1A and table S1).

Fig. 1. Cytokine production profile of CD4FOXP3 T cells.

(A to C) Transduced and untransduced (UT) cells of HDs were stimulated with immobilized anti-CD3 and soluble anti-CD28 monoclonal antibodies (mAbs) in the absence (A) or presence (B and C) of inflammatory cytokines (IL-1β and IL-6). Culture supernatants were collected after 24 hours (for IL-2 determination), 72 hours (for all other cytokines), and 7 days (C), and the concentrations of IFN-γ, IL-5, IL-17A, IL-22, and IL-2 were determined by enzyme-linked immunosorbent assay (ELISA). Mean values ± SEM are plotted in the graphs. A minimum of four HDs per cytokine were analyzed in (A) and (B). Values plotted in (C) represent the means ± SEM of three HDs. Statistical analysis was performed by nonparametric Mann-Whitney U test. Dark gray, untransduced; light gray, CD4NGFR; white, CD4FOXP3.

The cytokine production profile of CD4FOXP3 T cells was further confirmed by analysis of the expression of T cell lineage–related transcription factors Tbet, FOXP3, and RORC2. CD4FOXP3 T cells expressed high FOXP3 and RORC2 mRNA, in line with residual IL-17 production by CD4FOXP3 T cells generated from total CD4+ T cells, whereas Tbet (TBX21) expression was suppressed compared to untransduced and control CD4NGFR T cells after anti-CD3/CD28 mAb activation (fig. S1).

Upon activation in the presence of high doses of IL-6 and IL-1β suppression of TH1 (IFN-γ) and TH2 (IL-5), cytokine production in CD4FOXP3 T cells was preserved (a minimum of eight donors were tested, CD4NGFR versus CD4FOXP3, P < 0.05). Production of IL-22 was also reduced in CD4FOXP3 T cells from HDs (n = 4, P = NS). On the contrary, production of IL-17 was preserved, but it was not significantly increased upon culture in the presence of TH17-polarizing cytokines (n = 8) (Fig. 1B and table S2). Furthermore, production of IFN-γ and IL-5 was stably low even in prolonged cultures (n = 3) (Fig. 1C).

CD4FOXP3 T cells are suppressive in vivo in a model of xeno-GVHD

To test the suppressive capacity of CD4FOXP3 cells in vivo and assess the stability of CD4FOXP3 T cell function and phenotype in an inflammatory environment, we developed a humanized murine model of xeno-GVHD using NSG mice (Fig. 2) and preconditioned non-obese diabetic–severe combined immunodeficient (NOD-SCID) mice (fig. S2) as recipients of human T cells. Lethal GVHD was induced by injection of allogeneic CD4+ T cells (herein called Teff), as previously described (28, 29). CD4FOXP3 or control CD4NGFR T cells from HDs were co-injected with Teff cells at a 1:1 ratio. To further assess the ability of CD4FOXP3 cells to inhibit expansion of preactivated Teff cells, we also tested suppressor cells in a “late transfer protocol,” in which CD4FOXP3 cells were injected in NSG mice 6 days after the transfer of Teff cells. Injection of control Teff cells in NSG mice induced xeno-GVHD development in 90% of animals (Fig. 2A) and in all animals in NOD-SCID mice (fig. S2). Co-injection of CD4NGFR T cells did not protect animals from disease, whereas CD4FOXP3 T cell injection allowed the survival of 75 and 71% of mice when injected at day 0 or later, at day 6, respectively (Fig. 2A). In line with the survival results, injection with CD4NGFR T cells and Teff cells did not inhibit weight loss in the recipients, whereas weight was better maintained in animals receiving CD4FOXP3 T cells, either at day 0 or at day 6 (Fig. 2B). Cotransfer experiments in NOD-SCID hosts gave similar results, with 50% survival in mice receiving Teff cells together with CD4FOXP3 T cells (fig. S2).

Fig. 2. In vivo function of CD4FOXP3 T cells.

Preconditioned NSG mice were infused intraperitoneally with 2 × 106 CD4+ T cells (Teff) with or without transduced T cells generated from either HDs at 1:1 ratio. (A and B) Mice survival (A) and weight loss (B) were followed for 6 weeks after cell injection. (C) Frequency of human CD45+ cells (human chimerism) over the total circulating lymphocytes. (D) Frequency of transduced cells (measured as %ΔNGFR+ cells within human CD45+CD4+ T cells) in the peripheral blood of injected mice.

Human chimerism was measured as the percentage of circulating human cells. Despite the absence of xeno-GVHD in mice receiving CD4FOXP3 T cells, all animals developed substantial chimerism in peripheral blood with kinetics similar to control mice, suggesting that CD4FOXP3 cells did not hamper the engraftment of Teff cells but rather controlled their expansion or effector function (Fig. 2C). In NOD-SCID animals, protected mice maintained low chimerism in peripheral blood (fig. S2).

On the basis of cell surface expression of the ΔNGFR marker gene, transduced cells were tracked in the peripheral blood of treated mice. Results showed that these cells did not expand in vivo but rather declined with time as the Teff cells peaked (Fig. 2D). In line with their low proliferative capacity, when CD4FOXP3 T cells were injected in immunodeficient mice in the absence of Teff cells (n = 3), they did not induce GVHD (fig. S3).

The stability of CD4FOXP3 T cell phenotype was further endorsed in the in vivo late transfer experiments upon exposure to the proinflammatory environment induced by irradiation and xenoreactive Teff cells (Fig. 3). As depicted in Fig. 3A, transduced cells were isolated from secondary lymphoid organs of NSG mice 7 days after adoptive cell transfer. A large percentage of the recovered CD4FOXP3 cells maintained high CTLA4 and CD25 expression, whereas CD4NGFR cells were FOXP3 and expressed low CD25 and CTLA4 (Fig. 3B and fig. S4A). All recovered transduced cells produced low amounts of both TH1 and TH17 cytokines (fig. S4B).

Fig. 3. In vivo stability of CD4FOXP3 T cells.

(A) The diagram outlines the time points and the procedure followed for the in vivo transfer into NSG mice of CD4+ Teff cells and transduced cells. On day 13 (D13), 7 days after injection of transduced T cells, mice were sacrificed and lymphocytes were recovered from spleens. (B) The expression of FOXP3 (right panels), CTLA4 (middle panels), and CD25 (left panels) by transduced human T cells recovered from the spleens was determined. Upper dot plots show one representative mouse of the CD4NGFR (n = 4) and CD4FOXP3 (n = 7) groups. Gating strategy is reported in fig. S4A. Mean values ± SEM for each marker are plotted in the lower graphs and were as follows: CD4FOXP3 (n = 7): FOXP3, 48 ± 6%; CTLA4, 40 ± 5%; CD25, 61 ± 4%; CD4NGFR (n = 4): FOXP3, 3 ± 1%; CTLA4, 14 ± 2%; CD25, 26 ± 2%. Gray bars, CD4NGFR; white bars, CD4FOXP3.

Overall, these data indicate that CD4FOXP3 T cells have a stable phenotype and that the presence of an inflammatory environment is not sufficient to alter their functional properties and phenotype, making them suitable for in vivo applications.

CD4FOXP3 T cells generated by naïve T cell transduction show superior stability compared with memory-derived CD4FOXP3 T cells

We hypothesized that the expression of IL-17 by CD4FOXP3 T cells (Fig. 1A) could derive from memory T cells present in the starting CD4+ T cell population. Therefore, we generated LV-EF1α-FOXP3 Treg-like cells starting from CD4+CD25CD45RO+ memory or CD45RO naïve T cells. We evaluated FOXP3 expression of transduced cells in steady-state conditions and upon activation with or without inflammatory cytokines (namely, IL-6). Results showed that Treg-like cells generated from naïve CD4+ T cells (naïve CD4FOXP3) had stable expression of the transgene under all conditions tested. On the contrary, those generated from Ag-experienced T cells (memory CD4FOXP3) displayed FOXP3 expression superimposable to naïve CD4FOXP3 only in steady-state conditions, but, upon stimulation, the percentage of FOXP3+ T cells decreased, especially in the presence of inflammatory cytokines (Fig. 4A). The transduction efficiency and the vector copies per genome (CpG) were similar in both naïve and memory CD4FOXP3 (Fig. 4B), indicating that FOXP3 instability cannot be attributed to differences in transduction efficiency.

Fig. 4. Superior stability of naïve CD4FOXP3 T cells compared with memory CD4FOXP3 T cells.

(A) Expression of FOXP3 in CD4FOXP3 T cells generated from naïve (left panel) and memory (right panel) CD4+ T cells of HDs unstimulated (white dots) or 72 hours of activation with anti-CD3/CD28 mAbs in the presence (dark dots) or absence (gray dots) of IL-6. Results are shown as means ± SEM (n = 4). Percent FOXP3+ T cells (mean ± SEM) were as follows: naïve CD4FOXP3: unstimulated, 77 ± 1%; anti-CD3/CD28 mAbs (72 hours), 77 ± 3%; anti-CD3/CD28 mAbs + IL-6 (72 hours), 77 ± 2%; n = 4; memory CD4FOXP3: unstimulated, 84 ± 5%; anti-CD3/CD28 mAbs (72 hours), 78 ± 5%; anti-CD3/CD28 mAbs + IL-6 (72 hours), 70 ± 9%. (B) Vector copy number per genome (left axis) was assessed in both naïve (white bars) and memory (dashed bars) CD4FOXP3 cell cultures 4 weeks after transduction (n = 3). Mean CpG ± SEM were as follows: naïve CD4FOXP3, 1.8 ± 0.1 (n = 3); memory CD4FOXP3, 1.6 ± 0.4 (n = 3). Transduction efficiency 5 days after transduction (right axis) is also plotted (dots, n = 5). Mean percent ΔNGFR+ cells (±SEM) at day +5 were as follows: naïve CD4FOXP3, 43 ± 6% (n = 6); memory CD4FOXP3, 42 ± 3% (n = 4). (C) CD4+ responder T cells (R) were stimulated with soluble anti-CD3 mAb (1 μg/ml) in the presence of Ag-presenting cells (APCs) with or without transduced T cells at the indicated ratio of responder to CD4FOXP3. Average percent inhibition of proliferation of the responder cells by naïve (white bars) or memory (dashed bars) CD4FOXP3 T cells for each experiment, run in triplicate wells, is plotted in the left panel. Naïve and memory CD4FOXP3 T cells were tested for their ability to proliferate in response to APCs plus anti-CD3 mAb (1 μg/ml). Bars in the right panel indicate average counts per minute (cpm) for each experiment, run in triplicate wells. Error bars indicate SEM. (D and E) The cytokine production profile of untransduced (dark gray bars), CD4NGFR (light gray bars), and CD4FOXP3 (white bars) generated from naïve (filled bars) and memory (dashed bars) T cells was assessed by intracytoplasmic staining after activation with 12-O-tetradecanoylphorbol 13-acetate (TPA) and ionomycin (D) and as cytokine release in 72-hour cell culture supernatants upon anti-CD3/CD28 mAbs activation (E).

The instability of FOXP3 expression in memory CD4FOXP3 cells resulted in weaker suppressive function (at responder/CD4FOXP3 ratio of 4:1). Suppressive ability was reduced by 2.2-fold (P = 0.03, n = 5), and proliferative capacity was increased by 3.6-fold (n = 4), confirming that acquisition of Treg functions is dependent on stable FOXP3 expression and that memory T cells are more resistant to Treg conversion by LV-mediated gene transfer of FOXP3 (Fig. 4C).

IL-17A was barely present in naïve T cell cultures (Fig. 4D and table S3, n = 4; Fig. 4E and table S3, n = 7). In naïve CD4FOXP3, IFN-γ production was strongly inhibited in terms of both frequency of IFN-γ–producing cells (n = 4) (Fig. 4D and table S3) and cytokine released in the supernatant (n = 5, P = NS) (Fig. 4E and table S3). In contrast, control cultures from memory cells contained higher frequency of IL-17A–producing cells and released more IL-17A in culture supernatants compared to control naïve T cell cultures (Fig. 4, D and E, and table S3; n = 4). Transduction of memory T cells with LV-EF1α-FOXP3 decreased the frequency of IFN-γ–producing cells compared to control cultures, whereas that of IL-17–producing cells was not reduced (n = 4, Fig. 4D and table S3). However, unlike in naïve cells, overexpression of FOXP3 in memory T cells did not inhibit IL-17 or IFN-γ release in supernatants (n = 4), whereas IL-5 production was stably reduced in both naïve and memory CD4FOXP3 T cells (naïve: n = 6, P = 0.04; memory: n = 4) (Fig. 4E and table S3). Overall, these results indicate that FOXP3-mediated Treg conversion of conventional CD4+ T cells is more efficient when targeted to naïve rather than memory T cells and that the variability in IL-17 production observed in CD4FOXP3 T cells is most likely due to the presence of memory T cells in the starting population.

FOXP3-mutated CD4+ T cells can be efficiently transduced with LV-EFIα-FOXP3

Because our in vitro and in vivo data indicate that CD4FOXP3 T cells are functional and stable in inflammatory environment, thus being potentially suitable for application in clinically relevant settings, we tested whether Treg-like CD4FOXP3 T cells could be generated also from IPEX conventional T cells. Peripheral CD4+ T cells from five patients with different mutations in the FOXP3 gene (see fig. S5 for details) were transduced with either LV-EF1α-FOXP3 vector or control LV. T cells from patients were transduced with the same efficiency as those isolated from HDs (n = 16) (Fig. 5A). The vector content in purified ΔNGFR+ cell populations was also comparable between patients’ and HDs’ transduced cells (Fig. 5B).

Fig. 5. Transduction of IPEX-derived CD4+ T cells with LV-EFIα-FOXP3.

(A) Transduction efficiencies of LV-EF1α-FOXP3 (CD4FOXP3, white bars) and control vectors (CD4NGFR, gray bars) assessed by flow cytometry 5 days after infection of CD4+ T cells isolated from HDs and IPEX patients. Middle lines represent the median. Percent ΔNGFR+ cell median and range were as follows: for patients’ cells: CD4FOXP3, 55% (49 to 66); CD4NGFR, 62 (50 to 87); for HDs: CD4FOXP3, 58% (34 to 68); CD4NGFR, 74% (52 to 89). (B) Vector copy number per genome was assessed in both CD4FOXP3 (white bars) and CD4NGFR (gray bars) cell cultures using NGFR+ purified cells 4 weeks after infection. Bars indicate mean ± SEM. CpG means ± SEM were as follows: for patient-derived cells: CD4NGFR, 2.2 ± 0.7; CD4FOXP3, 1.3 ± 0.3; n = 3; for HDs: CD4NGFR, 2.5 ± 0.3; CD4FOXP3, 1.7 ± 0.1; n = 5. (C and D) FOXP3 expression in transduced cells was detected by flow cytometry on purified ΔNGFR+ cells 4 weeks after infection. The percentage of FOXP3+ cells in untransduced (dark gray), CD4NGFR (light gray), and CD4FOXP3 (white) samples from HDs (left panel) and patients (right panel) is depicted in the box plots in (C). Middle lines represent the median. Representative dot plots from one HD and one patient are shown in (D).

In Fig. 5, C and D, CD4FOXP3-transduced T cells from patients expressed FOXP3 (see also table S4) and maintained FOXP3 expression even in resting phase (at least 14 days after stimulation) when expression of the endogenous activation-induced FOXP3 is down-regulated in both untransduced and control LV–transduced cells. Thus, transduction with LV-EF1α-FOXP3 induces high and persistent FOXP3 expression in FOXP3-mutated conventional T cells.

IPEX-derived CD4FOXP3 T cells acquire Treg-like phenotype, cytokine profile, and function

CD4FOXP3 T cells differentiated from IPEX patients acquired Treg-like phenotype, with high expression of IL-2Rα chain (CD25) and CTLA4, low expression of IL-7Rα chain (CD127), and up-regulation of the Ikaros family transcription factor Helios, known to be highly expressed in human and mouse tTreg cells (30, 31) (Fig. 6A and table S4). Several Treg-associated molecules, such as CCR4, CD39, CD49d, and granzyme A, were not differentially expressed in CD4FOXP3 compared to control transduced cells in both patients and HDs (fig. S6), suggesting that either the expression of these molecules is not controlled by FOXP3 or their expression is lost during in vitro culture. Notably, we observed that LV-mediated transfer of FOXP3 in T cells from HDs induced the up-regulation of cell surface activation markers, such as HLADR, CD69, and CD71, and inhibited granzyme B expression (fig. S7 and table S4).

Fig. 6. Phenotype and cytokine profile of IPEX CD4FOXP3 T cells.

(A) Expression of Treg-related markers (CD25, CD127, CTLA4, and Helios) in untransduced (dark gray), CD4NGFR (light gray), and CD4FOXP3 (white) T cells was determined by flow cytometry 4 weeks after transduction. The percentage of marker-positive cells is plotted in the graphs. Analysis was gated on CD4+ΔNGFR+ cells for transduced cells and on CD4+ cells for untransduced cells. Middle lines represent the median. Statistical analysis was performed using nonparametric Mann-Whitney U test for groups with n ≥ 5: CD25: HDs, n = 15; patients, n = 5; CD127: HDs, n = 9; patients, n = 3; CTLA4: HDs, n = 5; patients, n = 3; Helios: HDs, n = 5; patients, n = 2. (B) Transduced and untransduced cells from IPEX patients were stimulated with immobilized anti-CD3 and soluble anti-CD28 mAbs. Culture supernatants were collected after 24 hours (for IL-2 determination) and 72 hours (for all other cytokines), and the concentrations of IFN-γ, IL-5, IL-17A, and IL-2 were determined by ELISA. Mean values ± SEM are plotted in the graphs. A minimum of three patients’ samples per cytokine were analyzed. Dark gray, untransduced; light gray, CD4NGFR; white, CD4FOXP3.

Similar to CD4FOXP3 T cells from HDs, both TH1 (IFN-γ and IL-2) and TH2 (IL-5) cytokine production by patients’ CD4FOXP3 cells was drastically reduced (three to four patients for each cytokine were tested), whereas IL-17A production by CD4FOXP3 T cells was highly variable (n = 4) (Fig. 6B and table S1).

Patients’ CD4FOXP3 T cells acquire Treg-like functional properties, including in vitro–suppressive capacity and hyporesponsiveness. Upon in vitro coculture with allogeneic CD4+ cells, used as responders, CD4FOXP3 cells displayed potent in vitro–suppressive function (Fig. 7A). Patients’ CD4FOXP3 cells suppressed by an average of 74 ± 5% (mean ± SEM) at a responder/suppressor ratio of 1:1 and maintained potent suppressive ability even at lower ratios (69 ± 7% at a ratio of 2:1 and 54 ± 8% at a ratio of 4:1) (Fig. 7A, right panel). Similarly, conventional CD4+ cells from HDs acquired in vitro–suppressive ability upon transduction with LV-EF1α-FOXP3 (78 ± 4%, mean percent suppression ± SEM, at a ratio of 1:1; 71 ± 6%, at a ratio of 2:1; 59 ± 6%, at a ratio of 4:1) (Fig. 7A, left panel).

Fig. 7. In vitro function of CD4FOXP3 T cells.

(A) CD4+ responder T cells were stimulated with soluble anti-CD3 mAb (1 μg/ml) in the presence of APCs with or without transduced T cells at the indicated ratio of responder (R) to suppressor (S). Average percent inhibition of proliferation of the responder cells alone for each experiment run in triplicate wells is plotted in the graphs. Error bars indicate SEM. (B) Transduced T cells were tested for their ability to proliferate in response to APCs plus anti-CD3 mAb (1 μg/ml). Bars indicate average cpm for each experiment, run in triplicate wells. Error bars represent SEM. Numbers in the plots indicate percent inhibition of proliferation of CD4FOXP3 versus CD4NGFR. CD4NGFR versus CD4FOXP3: in HDs: P = 0.01, n = 11; in patients: P = NS. Statistical analysis was performed with nonparametric Mann-Whitney U test.

IPEX-derived CD4FOXP3 T cells also became hyporesponsive to CD3-mediated stimulation. Their proliferation was inhibited (on average 78% for patient-derived cells, P = NS, and 73% for HDs, P = 0.01) compared to control transduced CD4NGFR T cells, in line with the anergic phenotype of tTreg cells (Fig. 7B).

The suppressive ability of CD4FOXP3 T cells generated from peripheral total CD4+ T cells from one IPEX patient (patient 21) was also tested in vivo in a model of xeno-GVHD, as described above. Late transfer of IPEX CD4FOXP3 T cells efficiently protected animals from lethal GVHD (67% survival), whereas transfer of CD4NGFR T cells did not (Fig. 8A). Protected mice, despite development of detectable human chimerism in peripheral blood, better maintained weight (Fig. 8, B and C).

Fig. 8. In vivo function of IPEX CD4FOXP3 T cells.

Preconditioned NSG mice were infused intraperitoneally with 2 × 106 CD4+ T cells (Teff). Six days after transduced T cells generated from patient 21 were infused at 1:1 ratio. (A and B) Mice survival (A) and weight loss (B) were followed for 6 weeks after cell injection. (C) Frequency of human CD45+ cells over the total circulating lymphocytes. (D) Frequency of transduced cells (measured as %ΔNGFR+ cells within human CD45+CD4+ T cells) in the peripheral blood of injected mice.

Overall, these data demonstrate that CD4FOXP3 T cells generated from FOXP3-mutated CD4+ T cells are fully functional, both in vitro and in vivo.


Here, we provide evidence that the de novo–generated CD4FOXP3 T cells have stable function, especially if generated from naïve T cells, and remain regulatory in inflammatory conditions. We further show that FOXP3-mutated CD4+ T cells isolated from the peripheral blood of IPEX patients can convert into functional Treg cells using lentiviral-mediated FOXP3 gene transfer. The resulting CD4FOXP3 T cell population acquires Treg-like phenotype and function in vitro and in vivo, regardless of the underlying mutation. Thus, we propose that the CD4FOXP3 T cell product has the potential to control autoimmunity and inflammation in patients with IPEX syndrome.

Once thought to be a stable subset of CD4+ T cells, there is now compelling evidence that Treg cells are plastic (32). Therefore, one major concern regarding the use of in vitro–generated Treg cells for immunotherapy is the risk of their in vivo conversion into Teff cells and consequent loss of suppressive ability. This phenomenon is particularly relevant in the context of IPEX syndrome, in which loss of functional FOXP3 results in the generation of impaired and unstable tTreg cells, which, under inflammatory stimuli, convert in potentially autoreactive IL-17–producing cells (6). We show that CD4FOXP3 T cells maintain Treg properties even when challenged by inflammatory stimuli, both in vitro and in vivo, although IL-17A production is not completely suppressed. However, IL-17 detection by Treg could be considered as a hallmark of their functional specialization, rather than lack of stability, because IL-17 is produced by a subset of highly suppressive human Treg cells that express CCR6, a chemokine receptor preferentially used for recruiting TH17 cells to the site of inflammation (33).

A distinctive feature of specialized versus plastic Treg cells remains the stable expression of FOXP3, and therefore, by that means, CD4FOXP3 T cells can be considered stable in vitro–induced Treg cells, despite the associated production of IL-17. On the same motif, our results point out that, unlike memory CD4FOXP3 T cells, naïve-generated CD4FOXP3 T cells have constant FOXP3 expression and therefore are stronger suppressors and better maintain stability. Although Ag-specific Treg cells have been shown to be more efficient suppressors (34), Ag specificity may not be an absolute requirement for Treg cell therapy, because Treg cells exert their function also in an Ag-independent manner. However, the naïve T cell compartment potentially comprises self-Ag–specific T cells (35), and, in the case of IPEX syndrome and many other autoimmune diseases, the knowledge of the type of Ags triggering the pathology is limited, making it difficult to select relevant auto-Ag–specific cells for Treg conversion. Still, the possibility remains that infusion of polyclonal Treg cells would interfere with immune responses to pathogens. Although this has not been the case up to now in patients receiving Treg therapy (1821), studies aimed at ruling out this possible drawback are ongoing.

We demonstrate that CD4FOXP3 T cells maintain their function also in vivo, when co-injected with the effector cells or after the xeno-reactions have started, suggesting a certain degree of regulation also on an ongoing reaction. However, the xeno-GVHD model only allows short-term evaluation of the survival of human cells and gives limited indications of cell-dose requirement. Therefore, questions on the in vivo life span of in vitro–manipulated CD4FOXP3 T cells cannot be addressed in this model. Data from recent clinical trials in which ex vivo–expanded Treg cells were infused to control GVHD after HSCT showed that these cells are detectable in vivo for 2 weeks after injection (18), and in one patient, increase in circulating Treg cells was present 6 months after transplant (20). In addition, results from ADA-SCID trial, in which gene-modified T cells persisted for many years in vivo (15), are encouraging, together with the notion that T cells with stem cell–like properties exist within the CD45RO T cell subset (36).

In tTreg cells, FOXP3 expression is essential for determining a Treg cell–specific gene expression profile, characterized, for example, by repression of IL-2 transcription (37). However, among the different tTreg-specific molecules, some appear to be expressed regardless of FOXP3, such as Eos and Helios, whose enhancement is mainly established by a Treg cell–specific CpG hypomethylation at their locus (38). Here, we show that LV-mediated overexpression of wild-type FOXP3 induces an overall reprogramming of surface and intracellular molecules in T cells. As a consequence, CD4FOXP3 T cells closely resemble activated tTreg cells, because they acquire a Treg-like modulation of FOXP3-dependent or FOXP3-independent markers such as CD25, CTLA4, CD127, and Helios (31, 38). We also report low expression of granzyme B (39) and up-regulation of activation molecules, such as HLADR, a marker of terminally differentiated tTreg cells (40), and CD69, associated with fully functional activated Treg cells (41).

Notably, LV-FOXP3–dependent phenotypic and functional reprogramming occurs in patients’ CD4+ T cells as efficiently as in those of healthy subjects, thus indicating that coexpression of mutated forms of FOXP3 in patients’ T cells does not hamper conversion in Treg-like cells. The use of a strong promoter, as EF1α, driving wild-type FOXP3 expression, ensures high transgene expression, overcoming the risk of interference by endogenous mutated molecules.

We believe that our results pave the way for the development of alternative or complementary therapies for IPEX syndrome, which would be an ideal model for adoptive transfer of genetically modified autologous Treg cells. Indeed, evidences from studies in healthy female carriers of FOXP3 mutations and IPEX transplanted patients have demonstrated that tTreg cells expressing the wild-type FOXP3 allele have a selective advantage over those expressing the one mutated and that a small number of wild-type Treg cells can control the disease (4244). This suggests that infusion of few functional CD4FOXP3 T cells could be sufficient to control autoimmunity in patients. Whether multiple infusions would be required or whether polyclonal CD4FOXP3 would be sufficient to generate “infectious tolerance” after the first infusion remains to be addressed. Similar approaches could be applied in several other more frequent pathologic conditions, including autoimmune diseases associated with defective Treg cell number and/or function and chronic inflammatory diseases.


Study design

The overall objective of the present research was to set the basis for the development of cell/gene transfer–based therapies for IPEX syndrome, and to this aim, we (i) assessed whether functional Treg-like cells can be generated from FOXP3-mutated CD4+ T cells by LV-mediated overexpression of wild-type FOXP3, (ii) investigated the functional stability of CD4FOXP3 Treg-like cells in inflammatory conditions and from naïve and memory cells, and (iii) tested their regulatory potential in vivo in murine models.

Patients affected with IPEX syndrome (n = 5) were selected on the basis of the presence of typical clinical manifestations and detection of FOXP3 gene mutation. Patients’ mutations and Treg cell phenotype are reported in fig. S5. The clinical details of patients 9, 11, and 20 are reported elsewhere (6, 10). Patients 19 and 21 have not been previously reported. T cells isolated from peripheral blood of an HD were transduced and tested in parallel to those of each patient. Additional HDs were tested for in vitro experiments (n = 16 in total). At the end of the CD4FOXP3 differentiation protocol, purity of the transduced cultures (measured as %ΔNGFR+ cells) and FOXP3 expression were tested. Cultures with purity <80% were either repurified or excluded from subsequent functional analysis. Peripheral blood was obtained from patients and HDs upon informed consent in accordance with the ethical committee of San Raffaele Scientific Institute approval (protocol TIGET 02).

All animal experiments were designed with a commitment to minimizing the numbers of mice and suffering. For each experiment, we used groups of n ≥ 3 mice. Those mice dying within 2 weeks after irradiation were excluded from analysis in all groups. Experiments in which the control group developed human chimerism below 10% were also excluded from analysis. The experimental protocol was approved by the Institutional Animal Care and Use Committee of the San Raffaele Scientific Institute.

Cell purification and T cell culture

Peripheral blood mononuclear cells (PBMCs) were purified over Ficoll-Hypaque gradients. For transduction experiments, either total, naïve (CD25CD45RO), or memory (CD25CD45RO+) CD4+ T cells were used. CD4+, CD25, and CD45RO T cells were isolated by negative selection (Miltenyi Biotec). Purified populations were more than 90% pure. All cultures were performed in X-VIVO 15 (Lonza), 5% human serum (Lonza), and penicillin/streptomycin (EuroClone).

LV production and T cell transduction

All LVs were produced by transient four-plasmid transfection of human embryonic kidney (HEK) 293T cells, as previously reported (45). FOXP3-expressing and control vectors are third-generation LVs, previously described in (24). For transduction, T lymphocytes were preactivated for 18 hours in complete medium plus recombinant human IL-2 (rhIL-2) and rhIL-7 in the presence of soluble anti-CD3 mAb (1 μg/ml, clone OKT3, Janssen-Cilag) and allogeneic CD3-depleted PBMCs (APCs), irradiated 6000 rads, at a 1:5 ratio of T cells to APCs. Cells were then infected with LV supernatant at a multiplicity of infection of 20 in the presence of polybrene (8 μg/ml) (Sigma-Aldrich). ΔNGFR+ transduced cells were purified after 9 days by magnetic bead selection (Miltenyi Biotec) and expanded with rhIL-2 and rhIL-15. T cells underwent two rounds of expansion, and they were tested for phenotype and function at least 12 days after activation, after overnight resting with low-dose rhIL-2.

Nucleic acid extraction and real-time polymerase chain reaction

Genomic DNA was extracted 4 weeks after transduction with QIAamp DNA Blood Mini Kit (Qiagen) following the manufacturer’s instructions. We quantified vector copies per genome by quantitative polymerase chain reaction (q-PCR) using 100 ng of template DNA, as previously described (46). RNA was extracted with RNeasy purification kits (Qiagen) and reverse-transcribed with High Capacity cDNA Reverse Transcription Kit (Applied Biosystems), following the manufacturer’s instruction. RORC2, Tbet, and FOXP3 mRNAs were quantified with Assay-on-Demand RT-PCR kits (Applied Biosystems). Relative expression was calculated with the ΔCt method after normalization to hypoxanthine-guanine phosphoribosyltransferase (HPRT) expression. The fold increase versus untransduced cells was plotted in the graphs.

Proliferation and suppression assays

In proliferation assays, 0.5 × 105 T cells per well were activated with irradiated APCs (T cell/APC ratio of 1:1) plus soluble anti-CD3 mAb (1 μg/ml) (OKT3). To evaluate the suppressive activity, allogeneic CD4+ T cells were activated with irradiated APCs (T cell/APC ratio of 1:1) plus soluble anti-CD3 mAb (1 μg/ml) in the presence or absence of different doses of CD4FOXP3 cells or control T cells. After 72 hours of culture, cells were pulsed for 16 hours with [3H]thymidine (1 μCi per well) (Amersham Biosciences).

Cytokine production

T cells (106/ml) were activated with plate-bound anti-CD3 (10 μg/ml) plus soluble anti-CD28 (1 μg/ml) mAbs (BD Pharmingen). Supernatants were collected after 24 hours for IL-2 detection and 72 hours for all other cytokines. The presence of cytokines in culture supernatants was evaluated by ELISA. Intracellular cytokines upon activation with TPA (10 ng/ml) and ionomycin (500 ng/ml) (Calbiochem) were detected by flow cytometry, as described (47). Phycoerythrin (PE)–anti–human IL-17A and fluorescein isothiocyanate (FITC)–anti–human IFN-γ mAbs from eBioscience and BD Pharmingen, respectively, were used.

Flow cytometry

Anti-CD4, anti-CD25, anti-HLADR, anti-CD69, anti-CD71, anti-CCR4 (all from BD Pharmingen), anti-CD49d (ImmunoTools), anti-CD39 (Miltenyi), and anti-CD127 (eBioscience) were used for surface staining. Intracellular staining with anti-CTLA4–biotin plus streptavidin PE or PE-Cy7 (BD Pharmingen), anti-FOXP3 (clone 259D) and anti-Helios (BioLegend), and anti-granzyme A (BD Pharmingen)/granzyme B (Invitrogen) was performed after fixation and permeabilization of cells with eBioscience FOXP3 staining solutions, following the manufacturer’s instructions. Samples were acquired on BD FACSCanto and analyzed with FCS Express Pro Software version 3 (De Novo Software).

Graft-versus-host disease model

Eight-week-old female NOD-SCID mice were injected intraperitoneally with 1 mg of blocking anti-mouse CD122 (IL-2Rβ) mAb to neutralize residual natural killer cell activity at day −1. The antibody was produced from the TMβ-1 hybridoma (provided to A.B. by T. Tanaka, Osaka University, Japan). At day 0, both NOD-SCID and NSG mice received total body irradiation with a single dose of 350 cGy (γ-irradiation from a linear accelerator) and were immediately infused with negatively selected CD4+ T cells (5 × 106 or 2 × 106 cells per mouse in NOD-SCID or NSG, respectively) either alone or with equal numbers of CD4FOXP3 or CD4NGFR cells. When late transfer protocol was used, transduced cells were infused at day 6. Survival and weight loss were monitored at least three times per week, and bleeding of the mice was performed once a week, as previously described (28).

Statistical analysis

Results are presented either as median and range or as mean ± SEM, as indicated. Analyses were performed with nonparametric Mann-Whitney U test. P < 0.05 was considered significant. Statistical analysis was applied to groups of five or more independent results.


Fig. S1. Expression of T cell lineage–related transcription factors by CD4FOXP3 T cells.

Fig. S2. In vivo function of CD4FOXP3 T cells—NOD-SCID model.

Fig. S3. CD4FOXP3 T cells do not induce xeno-GVHD.

Fig. S4. In vivo stability of CD4FOXP3 T cells.

Fig. S5. IPEX patients’ mutations and Treg cell phenotype.

Fig. S6. Treg-related markers in CD4FOXP3 T cells.

Fig. S7. Expression of activation markers by HDs’ transduced T cells.

Table S1. Cytokine production profile of HDs’ and patients’ transduced T cells.

Table S2. Cytokine production profile of HDs’ transduced T cells upon activation in the presence of inflammatory cytokines.

Table S3. Cytokine production profile of naïve and memory transduced T cells from HDs.

Table S4. Treg and activation marker expression by HDs’ and patients’ transduced T cells.


  1. Acknowledgments: We thank F. Barzaghi and all the members of the Italian Study Group of IPEX ( We thank D. Coviello and M. Cecconi (Galliera Hospital, Genova, Italy) for genetic analyses. We also thank S. Gregori and B. Camisa (San Raffaele Scientific Institute) for help in setting the xeno-GVHD model and T. Tanaka (Osaka University, Japan) for provision of the TMβ-1 hybridoma. We are grateful to M. Levings and F. Santoni de Sio for scientific discussion. We are indebted to the following colleagues who provided biological samples and clinical information of their IPEX patients: F. Locatelli and B. Lucarelli (Bambin Gesù Paediatric Hospital, Rome, Italy), L. Guidi (Università Cattolica del Sacro Cuore, Rome, Italy), M. Cipolli (Cystic Fibrosis Centre, Verona, Italy), A. Tommasini (Burlo Garofalo Hospital, Trieste, Italy), and F. Gurakan (Ankara, Turkey). Last, we thank the patients and their families for their participation in our studies. Funding: This work was supported by grants to R.B. from the Italian Telethon Foundation (Tele10A4), the Italian Ministry of Health (RF-2009-1485896), and the Seventh Framework Project of the European Community (Cell-PID). Author contributions: L.P. optimized the transduction protocol, performed in vitro experiments, analyzed the data, and wrote the manuscript; E.R.M. and G.F. performed in vivo experiments; C.S. performed in vitro experiments; A.B. established the xeno-GVHD model at our Institute; L.N. and M.G.R. critically discussed the data; R.B. designed the study, supervised the experiments, and revised the manuscript; all authors contributed to the critical discussion of the data. Competing interests: The authors declare that they have no competing interests. Bidirectional LVs used in this study are patented in the application “Lentiviral vectors carrying synthetic bi-directional promoters and uses thereof,” file number US8501464 and EP1616012.
View Abstract

Navigate This Article