UBD, a downstream element of FOXP3, allows the identification of LGALS3, a new marker of human regulatory T cells

Abstract

Here, we report the identification of the ubiquitin-like gene UBD as a downstream element of FOXP3 in human activated regulatory CD4⁺CD25^hi T cells (T_reg). Retroviral transduction of UBD in human allo-reactive effector CD4⁺ T helper (T_h) cells upregulates CD25 and mediates downregulation of IL4 and IL5 expression similar to overexpression of FOXP3. Moreover, UBD impairs T_h cell proliferation without upregulation of FOXP3 and impairs calcium mobilization. In the presence of ionomycin, overexpression of UBD in T_h cells leads to the induction of IL1R2 that resemble FOXP3-transduced T_h cells and naturally derived T_reg cells. A comparison of the transcriptome of FOXP3- and UBD-transduced T_h cells with T_reg cells allowed the identification of the gene LGALS3. However, high levels of LGALS3 protein expression were observed only in human CD4⁺CD25^hi derived T_reg cells and FOXP3-transduced T_h cells, whereas little was induced in UBD-transduced T_h cells. Thus, UBD contributes to the anergic phenotype of human regulatory T cells and acts downstream in FOXP3 induced regulatory signaling pathways, including regulation of LGALS3 expression. High levels of LGALS3 expression represent a FOXP3-signature of human antigen-stimulated CD4⁺CD25^hi derived regulatory T cells.

Main

Regulatory CD4⁺CD25⁺ T (T_reg) cells have emerged as a unique population of suppressor T cells that are important for the maintenance of peripheral tolerance.^{1, 2} In humans, T_reg cells are characterized by high levels of CD25 expression (CD25^hi) and represent 1–3% of all CD4⁺ T cells in peripheral blood.³ Typically, such T_reg cells are anergic upon stimulation via their T-cell receptor and exhibit suppressive activity towards effector T-cell function.² The clinical relevance of CD4⁺CD25^hi T_reg cells has recently been illustrated in different human disease settings, that is, altered numbers and impairment of suppressor activity of T_reg cells has been demonstrated in patients with Hodgkin's lymphomas, autoimmune polyglandular syndrome type II, multiple sclerosis, myasthenia gravis, diabetes, psoriasis, patients with graft-versus-host-disease, and infectious diseases such as malaria and HIV.^{4, 5, 6, 7, 8, 9, 10, 11, 12}

Development and function of T_reg cells critically depend on the forkhead transcription factor FOXP3 that specifies the regulatory T cell lineage.¹³ As a result, humans and mice deficient in FOXP3 lack CD4⁺CD25⁺ T_reg cells and suffer from multiorgan autoimmune disease.²

T_reg cells normally develop in the thymus. However, they also may arise in a FOXP3-dependent manner from peripheral CD4⁺CD25⁻ T cells.^{14, 15} The antigen-specificity of T_reg cells is currently a matter of debate, although self-antigens most likely represent the dominating specificity.^{16, 17, 18, 19}

Mechanisms of immunosuppression by T_reg cells comprise cell-contact dependent suppression, mainly observed in vitro, and cytokine-mediated suppression of effector T cells, mainly observed in vivo.²⁰ Cytokines most commonly found are IL10 and TGF-β, which also modulate T_reg cell function in vivo and T_reg cell development from nonregulatory CD4⁺CD25⁻ T cells in vitro.^{14, 17, 21, 22} However, IL10 and TGF-β are not essential for the development and function of T_reg cells.^{23, 24}

CTLA4 also functions as an important mediator and modulator of T_reg cell activity.²⁰ This member of the CD28-family is constitutively expressed at high levels in CD4⁺CD25⁺ T_reg cells in mouse and man. In the human system, CTLA4⁺ T_reg cells display a stronger suppressive activity in vitro than CTLA4⁻ T_reg cells.²⁵ However, CTLA4 is not essential for the development and function of T_reg cells, since its loss could be compensated by increased secretion of IL10 and TGF-β by CTLA4⁻CD4⁺CD25⁺ T_reg cells.²⁶

The conversion of murine naïve nonregulatory CD4⁺CD25⁻ T cells into T_reg cells by retroviral overexpression of FOXP3 underlines the master control function of this transcription factor.^{27, 28} Consequently, the use of FOXP3 as a tool for potential clinical T_reg cell therapy has been demonstrated recently in mice. Retroviral transduction of FOXP3 in polyclonal or antigen-specific naïve CD4⁺CD25⁻ T cells generated T_reg cells that suppressed the development of inflammatory bowl disease, autoimmune gastritis, diabetes or contact allergy, and induced transplant tolerance in vivo.^{27, 29, 30, 31} Despite these encouraging results in mice, debate exists about the efficacy of FOXP3 to reprogram human naïve CD4⁺ T cells towards a regulatory phenotype.^{32, 33, 34} The most likely reason for this ambiguity arises from the fact that unlike murine naïve and effector CD4⁺ T helper cells, human nonregulatory CD4⁺ T cells do express FOXP3.³⁵

Thus, the fundamental role of FOXP3 is well established, but target genes involved in the control of the FOXP3-dependent regulatory phenotype still require defining.³⁶ However, precise knowledge of this, including considerations of the differences between FOXP3 in mice and men,³⁵ is essential for the effective manipulation of regulatory T cells as immunomodulatory therapy. Hence, we compared the genetic program of human CD4⁺CD25^hi derived T_reg cells, which were expanded in vitro by allogeneic stimulation and IL2, with that of human allo-reactive effector T_h cells, which were reprogrammed by retroviral transduction of FOXP3. Diubiquitin (UBD) was identified as a gene that is upregulated by FOXP3 overexpression in T_h cells similar to T_reg cells. UBD overexpression in T_h cells demonstrated that it contributes in part to the FOXP3-dependent regulatory phenotype without upregulation of FOXP3, and led to the identification of the β-galactoside binding protein LGALS3. High levels of LGALS3 represent a FOXP3-signature of human antigen-stimulated CD4⁺CD25^hi derived T_reg cells as it is not upregulated in nonregulatory CD4⁺CD25⁻ derived effector T_h cells despite the expression of CD25 and FOXP3.

Materials and methods

Purification and Cultivation of Human CD4⁺ T Cells

CD4⁺ T cells were prepared from peripheral blood of healthy donors by centrifugation over Ficoll-Hypaque gradients (Biochrom AG, Berlin, Germany) and MACS isolation using the CD4⁺ T cell isolation kit and AutoMACS technology (Miltenyi Biotech, Bergisch Gladbach, Germany). CD4⁺CD25⁻ T cells and CD4⁺CD25^hi T_reg cells were then isolated by cell sorting on a MoFlo (DakoCytomation, Fort Collins, CO, USA) to a purity of >98%. CD4⁺CD25^hi T_reg cells were stimulated once with plate-bound anti-CD3 (TR66, 1 μg/ml), soluble anti-CD28 (CD28.2, 1 μg/ml, BD), and 50 U/ml recombinant human IL2 (Proleukin, provided by P Wagner, Chiron Corporation, Emeryville, CA, USA), and thereafter weekly with irradiated allogeneic EBV-transformed B cells (LG2-EBV, provided by T Boon, LICR, Brussels, Belgium). CD4⁺CD25⁻ T cells were used to establish long-term, allo-reactive effector T_h cell lines against LG2-EBV cells. Culture medium was IMDM, with 10% FCS, 100 U/ml penicillin/streptomycin, and nonessential amino acids (PAA Laboratories, Linz, Austria). Human peripheral blood was obtained after informed consent according to the MHH guidelines.

Antibodies

For immunostaining PE-, FITC-, APC-, and CyChrom-conjugated Abs against CD4 (RPA-T4), CD25 (M-A251), CD28 (CD28.2), TCRαβ (T10B9.1A-31), CTLA4 (BNI3), LGALS3 (B2C10; all from BD, Bioscience, San Jose, CA, USA), and FOXP3 (PCH101, eBioscine Inc., San Diego, CA, USA) and the respective isotype controls were used. For FOXP3 staining, murine hybridoma T cell transduced with human FOXP3 or GFP were included as further control. Anti-CD3ɛ (TR66, produced from hybridoma supernatants) and anti-CD28 (CD28.2, BD) were used for T-cell stimulation.

Analysing T-Cell Functions

T-cell proliferation and suppressor activity was assessed by stimulating 3 × 10⁴ T cells in triplicate with irradiated LG2-EBV B cells or 1 μg/ml plate-bound anti-CD3 (TR66) with or without IL2, or 1 μM ionomycin (Sigma, St Louis, MO, USA), in 96-well flat-bottom microtiter plates (Nunc, Wiesbaden, Germany). For transwell experiments, T_h cells were stimulated in 96 flat-bottom plates separated by 0.2 μm-pore transwell inserts (Greiner bio-one, Frickenhausen, Germany) from the T cells above the transwell. Cells were pulsed with 1 μCi/well of [³H]thymidine after 72 h for the final 16 h. Statistical analysis was performed using two-sided Student'st-test as indicated. TCR downmodulation upon anti-CD3 stimulation (plate-bound, 1 μg/ml) was assessed by FACS analysis (FACS Calibur or FACS Scan, BD) using PE-coupled anti-TCR antibodies.

Cytokine Detection

Stimulated cytokine release of IFNγ and IL10 was detected using ELISA sets of the CytoSet series (IFNγ, IL10; Biosource, Camarillo, CA, USA) and OptEIA series (IFNγ, and IL10, BD), according to the manufacturer's instructions. Supernatants of stimulated T cells were harvested at 72 h after stimulation immediately before the [³H]thymidine pulse.

Calcium Mobilization Assays

Calcium mobilization was measured by flow cytometry and confocal microscopy. In brief, T cells were loaded with 2 μM INDO-1 (Molecular Probes, Eugene, OR, USA) for 30–45 min at 37°C, extensively washed, and loaded with 1 μg/ml anti-CD3 (TR66) for 20 min at 4°C. Samples were applied to a MoFlo for 30 to 60 s at 37°C to measure the basal intracellular calcium concentration. Cells were stimulated by crosslinking cell-bound CD3 antibodies with a 1/30 dilution of goat-anti-mouse IgG antiserum or 10 μg/ml goat-anti-mouse IgGγ (Biomol International LP, Plymouth Meeting, PA, USA). The time-dependent intracellular Ca²⁺ was measured as the ratio of calcium-bound (405 nm) to calcium-free (515 nm) INDO-1 fluorescence, analyzed using FlowJo software (Tree Star, San Carlos, CA, USA). For single cell analysis, INDO-1 labeled cells (1 μM) were stimulated in coverslip-chambers (Nunc) precoated with 10 μg/ml anti-CD3 (TR66) using confocal microscopy (BIO-Rad, Hercules, CA, USA).

Retroviral Transduction of Human Effector CD4⁺ T Cells

The cDNAs encoding human FOXP3, CTLA4, and UBD were amplified from cDNA of T_reg cells using high fidelity PFU polymerase (Promega) and specific primers (FOXP3: 5′-GAC AAG GAC CCG ATG CCC A-3′ and 5′-TCA GGG GCC AGG TGT AGG GT-3′; CTLA4: 5′-TAC ATA TCT GGG ATC AAA GC-3′ and 5′-ATT TGG GTT CCG CAT CCA-3′; UBD: 5′-TGC AGA GAT GGC TCC CAA TG-3′ and 5′-GAC CCC TGC CAA CAC CCC AT-3′). PCR products were cloned into pCR4.1 TOPO (Invitrogen, Carlsbad, CA, USA), sequenced, and inserted into a pMSCV-based retroviral vector encoding an enhanced GFP (GFP) under the control of an IRES sequence.³⁷ The amphotropic PT67 packaging cell line (provided by M Wirth, GBF) was transfected as described elsewhere.³⁷ Filtered (0.45 μm) virus-containing supernatant supplemented with 8 mg/ml sequabrene (Sigma) was applied to differentiated allo-reactive T_h cells, established as described above, at day 2 after allogeneic stimulation by centrifugation at 5000 g for 60 min at room temperature. Cells were expanded thereafter with 50 U/ml IL2, and GFP expressing cells were sorted 1 to 2 weeks later using a FACS-Vantage (BD). Sorted cells were kept for up to 3 months using the restimulation procedure described above and were tested repeatedly for stable functional, phenotypic, and transcriptional changes.

Quantitative and Semiquantitative RT-PCR

Total RNA was isolated from CD4⁺ T cells using RNAeasy (Qiagen, Hilden, Germany) or nucleospin RNA-II (Macherey Nagel, Düren, Germany). cDNA synthesis was performed using oligo-dT primers and Superscript II Reverse Transcriptase (Invitrogen). Quantitative Real-time PCR was performed on an ABI PRISM 7000 cycler (Applied Biosystems, Foster City, CA, USA) using the SYBR Green PCR kit (Stratagene, La Jolla, CA, USA), and primers specific for FOXP3 (5′-GAA CGC CAT CCG CCA CAA CCT GA-3′ and 5′-CCC TGC CCC CAC CAC CTC TGC-3′), LGALS3 (5′-CGG AGC ACC TGC ACC TGG AGT CTA-3′ and 5′-TGA AGC GTG GGT TAA AGT GGA AGG-3′), and RPS9 (5′-CGC AGG CGC AGA CGG TGG AAG C-3′ and 5′-CGA AGG GTC TCC GCG GGG TCA CAT-3′) as described.³⁷ For semiquantitative RT-PCR, three-fold dilutions of cDNA samples starting with the first dilution, normalized to the expression of RPS9 (5′-CGC AGG CGC AGA CGG TGG AAG C-3′ and 5′-CGT AGG AGA GCG CAG AGA GAA GTC A-3′) were amplified with primers for FOXP3 (5′-TCA ACG GTG GAT GCC CAC GC-3′ and 5′-TCA GGG GCC AGG TGT AGG GT-3′), UBD (5′-CCC AAT GCT TCC TGC CTC TGT G-3′ and 5′-GAC CCC TGC CAA CAC CCC AT-3′), IL4 (5′-AAC ACA ACT GAG AAG GAA ACC TTC TGC-3′ and 5′-CTC TCT CAT GAT CGT CTT TAG CCT TTC-3′), IL5 (5′-GAG GAT GCT TCT GCA TTT GAG TTT G-3′ and 5′-GTC AAT GTA TTT CTT TAT TAA GGA CAA G-3′), GATA3 (5′-ACC TGT CAG ACC ACC ACA ACC ACA C-3′ and 5′-GGA TGC CTT CCT TCT TCA TAG TCA GG-3′), IL1R2 (5′-AAA TGA CTC TGC TAG GAC GGT CCC-3′ and 5′-TCT TCC AGG GCC ACA TCG TG-3′) and LGALS3 (5′-CCC ATC TTC TGG ACA GCC AAG TGC-3′ and 5′-GCA TCA TTC ACT GCA ACC-3′).

GeneChip Assays and Microarray Data Analysis

For the screening of similarly regulated genes in T_reg versus T_h cells compared to T_hFOXP3 versus T_hGFP cells, we analyzed two independent experiments using Human Genome U133 A arrays (Affymetrix, Santa Clara, CA, USA) and MAS 5.0 software (www.affymetrix.com/support/technical/manuals.affx: Microarray Suite User's Guide, Version 5.0). Selection criteria for potential FOXP3-dependent genes in T_reg and T_hFOXP3 cells were: (1) a signal with a fold-change of more than 3 in all four comparisons, or (2) a similar increase (I, including marginal increase, MI) or decrease (D, including marginal decrease, MD) of the signal change in all four experiments. GeneChip analysis of T_hUBD compared to T_hGFP cells was performed using Human Genome U133 PLUS 2.0 arrays (Affymetrix). All data from T_hFOXP3, T_hUBD, and T_reg cells were compared to each other, according to the criteria (1) a signal fold-change of more than 2, or (2) a similar signal change as either I and MD or D and MD in all six comparisons. Selected differences of one cell type were identified by (1) a signal fold-change of less than 1, and (2) an opposite signal change or no signal change (NC) in the respective T-cell population. All microarray data have been deposited at GEO (http://www.ncbi.nlm.nih.gov/geo/; GEO accession GSE4527). All T cells used for GeneChip analysis have been established for more than 2 to 3 months.

Results

Characterization of Human CD4⁺CD25^hi Derived T_reg Cells

For the generation and differentiation of human allo-reactive T_h cells and T_reg cells, sorted CD4⁺CD25⁻ and CD4⁺CD25^hi T cells were stimulated with allogeneic EBV-transformed B cells and IL2. CD4⁺CD25^hi derived T_reg cells expanded by this protocol could be grown for a long period of time (up to 6 months), but remained anergic to anti-CD3 stimulation when IL2 was omitted (Supplementary Figure 1a). These T_reg cells consistently suppressed proliferation of established allo-reactive T_h cell lines activated by anti-CD3 (Supplementary Figure 1a) or allogeneic EBV-transformed B cells (Supplementary Figure 1b). The suppressive activity of such T_reg cells was overcome by the addition of more than 10 U/ml IL2 (Supplementary Figure 1b) and required cell-contact since it was blocked by a transwell membrane (Supplementary Figure 1c). Phenotypical analysis revealed stable expression of high levels of CD25, CD28, and intracellular CTLA4 (data not shown) as well as FOXP3 mRNA independent of their activation status (Supplementary Figure 1d). Thus, CD4⁺CD25^hi derived T_reg cells sorted and expanded by our protocol maintained high level of sustained FOXP3 expression associated with phenotypic and functional stability.

Ca²⁺ metabolism is essential for the regulation of anergy³⁸ and has recently been shown to be impaired in murine CD4⁺CD25⁺ T_reg cells analyzed ex vivo.³⁹ In contrast to the latter report, our human T_reg cells showed signs of sustained Ca²⁺-flux upon anti-CD3 stimulation. This was reflected in the prolonged time required to reach the lower plateau level, following the initial Ca²⁺-rise, and was also higher than that found for T_h cells (Supplementary Figure 2a). Consistent results were obtained at the single cell level (Supplementary Figure 2b and c).

Differences between T_reg and T_h cells were also observed in the extent of anti-CD3 induced TCR downmodulation, which was strongly impaired in T_reg cells (Supplementary Figure 2d). This finding is in line with the observation that the extent of TCR downmodulation following ligand binding correlates with the induction of effector function.⁴⁰

In summary, the T_reg cell lines established by our protocol showed all the phenotypic and functional characteristic of bona fide regulatory CD4⁺CD25^hi T_reg cells and thus can be used to establish the mechanisms of FOXP3-dependent gene regulation.

FOXP3 Overexpression in Human Effector T_h Cells Induces a Partial Regulatory Phenotype

To identify relevant FOXP3-dependent genes associated with the regulatory phenotype, we used retroviral overexpression of FOXP3 in human allo-antigen specific effector T_h cells (T_hFOXP3). Successfully transduced T_h cells were sorted for GFP+, reintroduced into culture, and tested repeatedly over a period of up to 3 months.

FOXP3-transduced T_h cells were impaired in their proliferative response upon antigen-specific stimulation using allogeneic EBV-transformed B cells or stimulation with anti-CD3, reversed by the addition of IL2 (Figure 1a). In addition, T_hFOXP3 cells acquired suppressor activity towards effector T_h cells stimulated by allogeneic EBV-transformed B cells, but this was not observed in control T_h cells transduced with GFP (T_hGFP) or CTLA4 (T_hCTLA4) (Figure 1b). Thus, FOXP3 was able to specifically reprogram differentiation of helper T cells towards a regulatory phenotype. However, compared to naturally occurring T_reg cells, this regulatory phenotype was less pronounced (Figure 1a and b).

Figure 1

FOXP3 is able to reprogram human allo-reactive effector T_h cells towards a regulatory phenotype. (a) T_reg cells and T_h cells, retrovirally transduced with FOXP3 plus GFP (T_hFOXP3), CTLA4 and GFP (T_hCTLA4), and GFP alone (T_hGFP), were stimulated with allogeneic LG2-EBV B cells or anti-CD3 without or with 100 U/ml IL2. Proliferation was assessed at day 3. (b) T_hGFP cells as in (a) were stimulated by allogeneic LG2-EBV B cells in the presence of equal numbers of indicated T cells. Only T_reg and T_hFOXP3 cells significantly suppressed T_hGFP cell proliferation; NS=not significant. Similar results were obtained with normal effector T_h cells instead of T_hGFP cells. (c) T_h cells as in (a) were tested for intracellular CTLA4 expression in flow cytometry. Shaded area: isotype control, bold line: intracellular anti-CTLA4 staining. (d) Transduced T_h cells as in (a) were tested for downmodulation of TCR expression following anti-CD3 stimulation at the times indicated. Remaining surface TCR expression was measured with anti-TCRαβ antibodies by flow cytometry (MFI: mean fluorescence intensity). (e) Sustained overexpression of FOXP3 in T_hFOXP3 compared to T_h GFP cells. Real-time RT-PCR was carried out at the times indicated after stimulation with plate-bound anti-CD3 plus IL2. (f) Intranuclear FOXP3 protein expression of T_hFOXP3 cells compared to T_reg and T_hGFP cells was measured by flow cytometry. Gates were set according to the control staining using murine hybridoma T cells transduced with GFP alone (negative control; gray shaded) or with human FOXP3 plus GFP (thin line). Percent of positive cells is indicated.

In contrast, overexpression of CTLA4 did not induce a regulatory phenotype in T_h cells, although the levels of intracellular CTLA4 expression were similar to T_reg cells (Figure 1c). A possible contribution of CTLA4 to the FOXP3-dependent transcriptional control is therefore unresolved at this point.

We then analyzed the anti-CD3 induced TCR downmodulation. Similar to T_reg cells, T_hFOXP3 cells showed an impaired downmodulation of TCR expression that was not observed in T_hGFP or T_hCTLA4 cells (Figure 1d). Interestingly, this impairment of TCR downmodulation appears to be a general effect of FOXP3 as it was even observed in murine hybridoma T cells transduced with FOXP3. However, such hybridoma T cells did not acquire an anergic phenotype, although they showed upregulation of CD25 and impaired NFAT function (data not shown). Therefore, a high level of sustained FOXP3 mRNA (Figure 1e) and protein expression in T_hFOXP3 cells (Figure 1f) induced most of the characteristics of the FOXP3-dependent regulatory phenotype.

It should be noted that in long-term cultures, T_hFOXP3 cells were able to escape the FOXP3-dependent control. Most likely, the appearance of T cells with lower expression levels of FOXP3 was responsible for this phenomenon, since such T cells also showed reduced expression of the co-transduced IRES-driven GFP. As a result, re-sorting of long-term cultured T_hFOXP3 cells for high expression of GFP re-established their regulatory phenotype (data not shown).

In addition, T_hFOXP3 cells did not reach the same high levels of CD25 expression (Figure 2a) while CD28 was present at similar levels in both types of cells (data not shown). We, therefore, analyzed the functional properties of T_hFOXP3 cells expressing different levels of CD25 (Figure 2b). CD25^hi T_hFOXP3 cells showed a more pronounced regulatory phenotype than CD25^lo T_hFOXP3 cells in terms of their proliferative impairment and suppressor function (Figure 2c and d). Again, CD25 expression decreased in long-term culture despite the presence of FOXP3 (data not shown). This suggests that the T_h cell specific regulation of CD25 expression is only transiently overcome by FOXP3 overexpression in differentiated effector T_h cells.

Figure 2

Differences in the regulatory phenotype between CD25^hi and CD25^lo T_h cells transduced with FOXP3. (a) T_h cells transduced with FOXP3 plus GFP (T_hFOXP3) and GFP alone (T_hGFP) were tested for CD25 expression and compared with T_reg and T_h cells. FOXP3 transduction never resulted in CD25 expression comparable to bona fide T_reg cells. (b) Reanalysis of T_hFOXP3 cells sorted for CD25^hi and CD25^lo expression is shown. (c) T_hFOXP3 cells sorted for CD25^hi and CD25^lo expression, T_h and T_hGFP cells were stimulated with allogeneic LG2-EBV B cells with or without 100 U/ml IL2. CD25^hi T_hFOXP3 cells are more impaired in their proliferative capacity than CD25^lo T_hFOXP3 cells. (d) T_hFOXP3 cells of (c) were added to T_h cells in equal numbers and stimulated with allogeneic LG2-EBV B cells. Suppressive activity was significantly more pronounced (^*P<0.02) in CD25^hi T_hFOXP3 than CD25^lo T_hFOXP3 cells.

Gene Expression Profiles of T_reg Cells and T_hFOXP3 Cells

The data presented above provide a biological framework for the identification of genes associated with the FOXP3-dependent regulatory phenotype. Using GeneChip analysis, we screened for genes similarly regulated in both T_reg and T_hFOXP3 cells compared to T_h and T_hGFP cells. According to our stringent selection criteria, 39 genes were defined that were similarly regulated in both regulatory T-cell populations. As expected, known markers of T_reg cells were identified, that is CD25, CTLA4, TNFRSF1B, and CCR7. In addition, we found genes that were not previously associated with FOXP3-dependent transcriptional control, for instance the ubiquitin-like gene diubiquitin (UBD) and the β-galactoside binding lectin LGALS3 (Table 1). Proteins involved in ubiquitinylation play an important role in the regulation of T-cell anergy.⁴¹ Hence, we analyzed the potential contribution of UBD to the FOXP3-dependent transcriptional control in more detail.

Table 1 Genes expressed similarly in T_hFoxp3 and T_reg cells detected by microarray analysis

UBD is Involved in the Regulation of CD4⁺ T-Cell Anergy

UBD is expressed at high levels in CD4⁺CD25^hi derived T_reg cells and consistently upregulated in FOXP3-overexpressing T_h cells compared to T_hGFP and T_hCTLA4 cells (Figure 3a). Unlike FOXP3, which is expressed at sustained high levels, UBD expression is transiently downregulated following T-cell activation using anti-CD3/-CD28 antibodies and IL2 (Figure 3a).

Figure 3

UBD contributes to the transcriptional and functional control of FOXP3. (a) T_h cells transduced as in Figure 1a were tested for UBD mRNA expression by semiquantitative RT-PCR after stimulation with plate-bound anti-CD3, soluble anti-CD28, and 100 U/ml IL2. cDNA was tested in three-fold dilutions starting with the first dilution; RPS9 was used as house-keeping control. (b) T_h cells transduced with UBD plus GFP (T_hUBD), FOXP3 plus GFP (T_hFOXP3), and GFP alone (T_hGFP) were tested for surface CD25 and CD28 expression in flow cytometry. (c) T_reg cells and T_h cells as in b were stimulated with LG2-EBV B cells alone or in the presence of 100 U/ml IL2. Proliferation was assessed at day 3. (d) The same T cells as in c were analyzed for IFNγ and IL10 cytokine secretion following stimulation with LG2-EBV B cells.

Overexpression of UBD in T_h cells (T_hUBD) via retroviral transduction induced upregulation of CD25 without affecting CD28 expression (Figure 3b). In long-term T_hUBD cell lines, significant impairment of proliferation upon TCR stimulation was observed, which was reversed by the addition of IL2 (Figure 3c). Similar to FOXP3, overexpression of UBD in effector T_h cells did not suppress IL10 and IFNγ secretion (Figure 3d). Importantly, FOXP3 expression was not upregulated in T_hUBD cells (Figure 4a and b) suggesting that UBD is a key player in the regulation of anergy in T_reg cells downstream of FOXP3.

Figure 4

UBD expression impairs T-cell proliferation independent of FOXP3. (a) Effector T_h cells transduced as described in Figure 3b were tested by semiquantitative RT-PCR as described in Figure 3a for mRNA of genes known to be regulated in FOXP3 transduced T_h cells. Note that FOXP3 is not upregulated in T_hUBD cells. (b) Expression of high levels of FOXP3 protein in T_hFOXP3 cells, but not in T_hGFP and T_hUBD cells, determined by flow cytometry as described in Figure 1f. Percent of positive cells is indicated. (c) Quantification of LGALS3 mRNA by Real-time RT-PCR in T_reg and T_hFOXP3 cells compared to T_hGFP cells. (d) Anti-CD3 induced Ca²⁺-mobilization in T_hFOXP3, T_hUBD, and T_hGFP cells was determined on a MoFlo as described in Supplementary Figure 2a. (e) T_hGFP and T_hUBD cells were tested for upregulation of IL1R2, FOXP3, and UBD mRNA. Ionomycin (Io) or vehicle (DMSO) was added. Determination was carried out after 24 h by semiquantitative RT-PCR as described in Figure 3a. (f) IL1R2 specific mRNA was determined in T_hFOXP3, T_hUBD, and T_hGFP cells of Figure 3b by semiquantitative RT-PCR. Note that no signal is obtained in T_hUBD as in T_hGFP cells. The slight discrepancy between e and f in this respect is most likely due to differences between the individual donors of T_h cells.

Like T_hFOXP3 cells, mRNA of IL5 and IL4 was significantly downregulated in T_hUBD cells (Figure 4a). Whereas GATA3, regulating T_h2 specific IL5 and IL4 cytokine expression, was unaltered (Figure 4a),⁴² LGALS3, which has been described to downregulate expression of IL4 and IL5 mRNA,⁴³ was only slightly induced (Figure 4a and c). Thus, UBD appears to be responsible for the IL4/IL5 downregulation as a downstream effector of FOXP3.

We then analyzed the influence of UBD overexpression on TCR downmodulation and Ca²⁺-mobilization. Interestingly, UBD did not impair the anti-CD3 induced TCR downmodulation observed with T_reg cells and T_hFOXP3 cells (data not shown). However, Ca²⁺-mobilization was influenced (Figure 4d). Whereas FOXP3 overexpression induced a slightly elevated steady-state phase of Ca²⁺-flux compared to T_hGFP cells, this state was lower in T_h cells overexpressing UBD (Figure 4d).

Since ionomycin is known to induce ubiuqitin-ligases associated with anergy,³⁸ we investigated the effect of ionomycin treatment on UBD expression in T_h cells. As predicted, ionomycin induced upregulation of UBD mRNA (Figure 4e), but at the same time, FOXP3 mRNA expression was downregulated despite overexpression of UBD (Figure 4e). This suggests that the ionomycin-induced anergic state is independent of FOXP3 and confirms that UBD does not induce the expression of FOXP3. Interestingly, similar to overexpression of FOXP3 in T_h cells (Figure 4f), overexpression of UBD together with ionomycin-induced sustained Ca²⁺-flux upregulated IL1R2 mRNA expression (Figure 4e). As FOXP3 overexpression leads to the upregulation of UBD and in parallel increases the levels of TCR-stimulated Ca²⁺-flux, both UBD together with an elevated Ca²⁺-flux seem to cooperatively regulate the expression of IL1R2. On the other hand, ionomycin did not improve the suppressive activity of T_h cells overexpressing UBD (data not shown), indicating that high levels of FOXP3 are required to obtain a suppressive phenotype.

High LGALS3 Protein Expression Specifies a FOXP3-Dependent Signature of Human T_reg Cells

We extended our GeneChip expression analysis of T_hFOXP3 and T_reg cells to T_hUBD cells. Regulated genes could be classified into four groups (Figure 5a). (i) Genes regulated in all three T-cell populations. These included UBD and LGALS3. (ii) Genes expressed similarly in T_hFOXP3 and T_hUBD cells only, such as CD9. (iii) Genes expressed similarly in T_hUBD and T_reg cells only, such as CBL-B. (iv) Genes expressed similarly in T_hFOXP3 and T_reg cells only, such as CCR7, FYB, and CCNG2. These data reflect the congruencies and differences between naturally derived T_reg cells, and our FOXP3 and UBD engineered effector T_h cells, respectively.

Figure 5

LGALS3 protein expression represents a FOXP3-signature of CD4⁺CD25^hi derived T_reg cells. (a) Transcriptome analysis of T_hUBD, T_hFOXP3, and T_reg cells. Four groups were defined when analyzing significantly regulated genes. T_hUBD=T_hFOXP3=T_reg: genes regulated similarly in all three regulatory types of T cells including UBD and LGALS3. T_hUBD=T_reg: genes regulated similarly in regulatory T cells and UBD transduced T_h cells. T_hUBD=T_hFOXP3: genes regulated similarly in UBD and FOXP3 transduced T_h cells. T_hFOXP3=T_reg: genes regulated similarly in regulatory T cells and FOXP3 transduced T_h cells. The signal fold change is represented as red (upregulated) or green color (downregulated) indicated below the panels. (b) T_reg and T_h cells were tested for intracellular LGALS3 expression in flow cytometry. Thin line: isotype control; bold line: anti-LGALS3 staining. (c) LGALS3 expression is not induced following T-cell stimulation. T_h and T_reg cells were tested for the expression of FOXP3, CD25, LGALS3, and CTLA4 at day 3 following allogeneic stimulation using LG2-EBV B cells and IL2. (d) Specificity of FOXP3 staining was confirmed using murine hybridoma T cells transduced with GFP alone (GFP) and human FOXP3 and GFP (FOXP3) as described in Figure 1f. Shaded area: isotype control, bold line: antibody staining. Percent of positive cells is indicated. (e) Surface staining of CD4 and CD25 on human peripheral blood lymphocytes of a healthy donor (left panel). Regions and percentage of CD4⁺ T cells expressing high, intermediary, and no CD25 expression, respectively, correspond to the colors in the right panel (CD25^hi=red, CD25^int=green, CD25^neg=blue curve), which depicts the corresponding intracellular LGALS3 staining. (f) T_h cells were stimulated by allogeneic LG2-EBV B cells in the presence of T_reg cells at the indicated ratios with or without 50 mM lactose. Proliferation was assessed at day 3.

Since upregulation of LGALS3 mRNA was common to T_reg cells and T_h cells overexpressing FOXP3 and UBD, respectively, we analyzed LGALS3 protein expression in these cells. Interestingly, high sustained levels of LGALS3 protein expression were detected only in T_reg cells (Figure 5b). Whereas significant induction of LGALS3 protein was also observed in T_hFOXP3 cells, UBD overexpression in T_h cells only minimally affected LGALS3 protein expression (Figure 5b). Contrary to mice,⁴⁴ LGALS3 protein was not significantly induced in human nonregulatory effector T cells following stimulation with allogeneic EBV-transformed B cells and IL2, although upregulation of intranuclear FOXP3, cell surface CD25, and intracellular CTLA4 protein expression was observed in such cells (Figure 5c). As LGALS3 protein expression is restricted at low levels to nonactivated human peripheral blood CD4⁺CD25⁺ T cells (Figure 5e), high level of LGALS3 protein expression represents a T_reg-specific FOXP3 signature of antigen-stimulated human CD4⁺CD25^hi-derived T_reg cells.

LGALS3 has been described to induce T-cell apoptosis in human T cells via a lactose-dependent binding to CD7 and CD29.⁴⁵ Therefore, we analyzed the suppressor function of T_reg cells in the presence or absence of 50 mM lactose, which potently impairs the apoptotic function of LGALS3.⁴⁵ But no impairment of T_reg cell suppressor function was observed suggesting that LGALS3 is not directly involved in T_reg cell suppression (Figure 5f).

Discussion

In the present study we compared the genetic program of human CD4⁺CD25^hi derived T_reg cells with that of allo-reactive effector T_h cells reprogrammed towards a partial regulatory phenotype by retroviral transduction of FOXP3. The reason for such an approach was to unmask FOXP3-dependent genes by mimicking sustained high levels of FOXP3 expression in T_h cells and to reduce the size of differentially expressed genes to the most significant set. Critical FOXP3-dependent target genes were identified.

One of the genes consistently upregulated was the ubiquitin-like gene diubiquitin (UBD, also known as FAT10). UBD has been initially cloned as a gene exclusively expressed in dendritic cells and mature B cells.⁴⁶ It was reported to be involved in protein degradation, apoptosis, and control of spindle assembly.^{47, 48} Moreover, UBD expression can be induced by IFNγ, whereas another IFNγ inducible gene, NEDD8 ultimate buster-1L (NUB1L), has been reported to accelerate the degradation of UBD.⁴⁹ A potential function of UBD in the physiology of human T_reg cells has not been addressed so far. This is most likely due to the fact that UBD expression is found at significant levels only after in vitro activation as demonstrated in this work.

Since FOXP3 is a transcriptional repressor,^{36, 50} upregulation of UBD expression should not be directly induced by FOXP3. Rather, UBD transcription should be indirectly regulated, for example, by downregulating transcriptional repressors of UBD. This is supported by preliminary experiments in which a 1.5 kb genomic DNA fragment preceding the transcriptional start of UBD was tested in a reporter system. No reporter activity was induced by FOXP3 (unpublished data).

The contribution of high levels of UBD expression to the FOXP3-dependent regulatory phenotype became apparent, as we could show that UBD overexpression in T_h cells interfered with proliferation and upregulated cell surface expression of CD25. In addition, expression of IL4 and IL5 mRNA was downregulated in T_hUBD cells comparable to T_h cells overexpressing FOXP3, although FOXP3 was not induced by UBD. In contrast, genes normally induced upon FOXP3-overexpression, for example CD28 and CTLA4, were not affected by UBD transduction. This indicates that UBD acts downstream in some of the FOXP3-induced regulatory signaling branches, and, therefore, only a partial regulatory phenotype is induced by UBD alone as it does not upregulate FOXP3. A clear confirmation of this phenomenon was obtained, when ionomycin was added to T_h cells overexpressing UBD. Only the ionomycin-mediated higher Ca²⁺-levels together with overexpression of UBD mimicked the FOXP3-induced transcriptional control of IL1R2. In line with this was the observation that UBD overexpression only minimally induced LGALS3 protein expression in T_h cells, although LGALS3 mRNA was upregulated similar to T_h cells overexpressing FOXP3.

How UBD mediates transcriptional control is questionable. It is known that E3-ubiquitin ligases such as Itch, Cbl-b, and Grail play an essential role in the induction of T-cell anergy. Their mode of action is to target key TCR signaling molecules for degradation that are localized in the cytoplasm.^{38, 51} UBD has similarly been reported to target proteins for degradation but not in the context of TCR signal transduction. However, UBD has also been found to be localized in the nucleus.⁵² Therefore, its main effect might be a transcriptional modulation, as the activity of many signal transduction pathways and transcription factors are regulated by ubiquitin-mediated proteolytic and nonproteolytic mechanisms.^{53, 54}

Whatever the mechanisms, our extended transcriptome analysis of T_reg cells compared to T_hFOXP3 and T_hUBD cells suggested that LGALS3 might play a central role for some of the overlapping functional and phenotypic changes observed in these cells.

LGALS3 represents a unique member of the family of β-galactoside-binding proteins. It is composed of a small N-terminal domain, which controls its subcellular targeting, a repetitive collagen-like region that serves as substrate for matrix metalloproteinases, and a C-terminal carbohydrate-recognition domain.⁵⁵ Expression of LGALS3 has been observed in different tissues and cells types. It displays pleiotropic biological functions depending on subcellular localization. Extracellular LGALS3 binds to cells through glycosylated receptors, thereby triggering or modulating cellular responses such as mediator release, apoptosis, cell adhesion, and migration.^{45, 56} Intracellular LGALS3 has been reported to inhibit apoptosis⁵⁷ and the cell cycle.⁵⁸ Interestingly, upregulation of LGALS3 has been observed during neoplastic transformation in several human malignancies.⁵⁹ Moreover, LGALS3 has been found in the nucleus, where it is reported to be involved in pre-mRNA splicing,⁶⁰ and the regulation of gene expression (eg cyclinD and Muc2) through the activation of specific transcription factors, including AP1, NFAT, SP1, and CREB.^{61, 62} Serine phosphorylation and dephosporylation at position 6 of LGALS3 by casein kinase 1 and protein phosphatase 1, respectively, controls the carbohydrate binding capability, nuclear localization, transcriptional modulation, and antiapoptotic function of LGALS3.^{58, 63, 64} Diversified functions of LGALS3 have also been reported for cells of the immune system. LGALS3 has been detected in monocytes and macrophages, dendritic cells, mast cells, murine T cells, and granulocytes.⁶⁵ Several important processes and functions are modulated by LGALS3 in these cells including thymocyte migration,⁶⁶ innate immunity against infections,⁶⁷ inflammation,⁶⁸ and T-cell apoptosis.^{45, 69} Owing to the non-differential expression of LGALS3 in activated murine CD8⁺ and CD4⁺ T cells,⁴⁴ it was unexpected that LGALS3 might act as a T_reg-specific marker. However, only low levels of LGALS3 protein were detectable in human peripheral blood CD4⁺CD25⁺ T cells, but sustained high levels were induced in naturally derived human CD4⁺CD25^hi T_reg cells following antigenic stimulation in vitro. Similarly, upregulation of LGALS3 protein expression was observed in FOXP3-transduced effector T_h cells reprogrammed towards a partial regulatory phenotype. This indicates that LGALS3 protein expression in fact represents a FOXP3-signature of the regulatory program.

The potential contribution of LGALS3 to the function of activated T_reg cells is speculative. The anti-apoptotic activity of LGALS3 might protect T_reg cells from apoptotic signals through CD95,⁷⁰ which is expressed at high levels on T_reg cells. Alternatively, the activation and modulation of specific transcription factors, like AP1, NFAT, SP1, and CREB,^{61, 62} might contribute to the FOXP3-dependent transcriptional control. However, more direct effects of LGALS3 on T_reg cell functions have to be considered. In murine T cells, a direct association of LGALS3 with the TCR complex, restricting TCR signaling, has been reported. This interaction dependents on N-glycosylation mediated by the enzyme beta-1,6 N-acetylglucosaminyltransferase V (Mgat5).⁷¹ Therefore, mice deficient in Mgat5 show a lowered T-cell activation threshold with enhanced recruitment of TCRs to agonist-coated beads, and actin microfilament re-organization.⁷¹ In contrast to the latter, in human CD4⁺CD25^hi derived T_reg cells, that express LGALS3, we have observed a dynamic impairment of actin remodeling compared to CD4⁺CD25⁻ T cells.⁷²

The addition of soluble LGALS3 to Jurkat T cells has been reported to trigger a sustained influx of extracellular Ca²⁺ that can be blocked by lactose.⁷³ As a sustained Ca²⁺ flux in T_reg cells following TCR stimulation was observed in our case, LGALS3 released from the cells might act in an autocrine manner and modulate the Ca²⁺ homeostasis of T_reg cells. However, similar to T cells of mice,⁴⁴ we did not detect significant cell surface bound LGALS3 in resting or activated T_reg cells. We, therefore, excluded a paracrine effect of LGALS3 in the process of suppression of antigen-induced T_h cell proliferation by T_reg cells, suggesting that LGALS3 released by T_reg cells might not reach amounts needed to induce either apoptosis or sustained Ca²⁺ flux.^{69, 73}

In general, the approach to reprogram antigen-specific effector T_h cells into therapeutic regulatory T cells by the transduction of FOXP3 is very encouraging. Clearly, however, only a partial regulatory phenotype can be induced by FOXP3 overexpression in effector T_h cells compared to CD4⁺CD25^hi derived T_reg cells. FOXP3-transduced T_h cells show less proliferative impairment and suppressor function, and less induction of LGALS3 expression. They further downregulate CD25 expression, and still secrete effector cytokines like IFNγ. Based on the observation that FOXP3 protein is expressed in human effector T_h cells, inevitable mechanisms must exist that control the overwhelming negative effects of FOXP3 in human effector CD4⁺ T cells.

Our analysis identified LGALS3 as a FOXP3-dependent signature differentiating between a regulatory and an effector program. LGALS3 is expressed at high constitutive levels only in antigen-stimulated CD4⁺CD25^hi derived T_reg cells. Moreover, we identified UBD as a downstream effector of FOXP3 involved mainly in the regulation of anergy. Thus, our analysis provides a focused starting point for further elucidation of genetic and mechanistic aspects of active immunosuppression and of differences between human and mouse regulatory T cells. This will ultimately lead to a complete understanding of human regulatory T cells.