A human BAC clone containing hFOXP3 gene, RP11-344O14 was purchased from Life Technologies. The 5′-half of BAC DNA was removed by homologous recombination by using RedET recombinase (39, 40) (Gene Bridges) with the homologous arm encoded in the vector and the BAC DNA located 3′-downstream region of the repetitive DNA region (Figure 1A). To replace hFOXP3 gene expression with a fluorescent protein, AmCyan cDNA (from the first ATG) was inserted just downstream of the first ATG of hFOXP3 (the hFOXP3 first ATG was replaced by the AmCyan first ATG) by a homologous recombination procedure with modifications. BAC transgenic C57BL/6 mice were generated and bred by the Mouse Genetic Core at Cedars-Sinai Medical Center. Experimental protocols using the transgenic mice were approved by IACUC at Cedars-Sinai Medical Center (#3001&6528).

To generate hFOXP3 expression reporter cell lines, the same modified hFOXP3 BAC was used with the addition of a neomycin resistance gene (under control of a CMV promoter). This resistance gene was inserted into the vector sequence upstream of the 5′-end of the BAC DNA. mFoxp3 expression reporter cell lines were also generated using the mouse BAC clone RP23-267C15 (Life Technologies). mFoxp3 expression from this BAC clone was replaced by the mCherry fluorescent protein by the same procedure as described for the hFOXP3 BAC modification. The resulting hFOXP3/AmCyan and mFoxp3/mCherry modified BAC are illustrated in Figures 3A and 4A, respectively. These BAC DNAs were transfected into EL4 B02 sub-clone cells by nucleofector 4D, and transfectants were selected by culture with G418 for 2 weeks. To select AmCyan (hFOXP3) and mCherry (mFoxp3) expressing cells, these transfectants were stimulated by plate-coated anti-CD3 (5 μg/ml in PBS) and TGF-β (5 ng/ml) for 48 h and AmCyan and mCherry positive cells were sorted and cloned by limiting dilution. JE6.1 transfectants using the hFOXP3/AmCyan BAC construct were also generated by the same strategy as that used for generating the EL4 reporter cells shown in Figure S1B in Supplementary Material.

Hunan BAC DNA in hFOXP3/AmCyan transgenic mice and cell lines were analyzed by PCR using genomic DNA as well as DNA sequencing of the PCR products. PCR primer binding positions are shown in Figure 2A. Primer sequence used were Us4F: CCATTATTTGCCACCTCTTCGTGG, Us4R: AGTTCAGACTTGGTCCGGATGGT, Us3F: CAATCGCCCCTTCTTCACCT, Us3R: CCTCTGCGTTTTGTCACACG, Us2F: GCCTTACAAGAGCTCCTGAAGGAAG, Us2R: GGTGCTCCTGTATTGGGTGCATA, Us1F: CCGAAGCTTGCCAATTGCTT, Us1R: TGTGACTATTGCTGGGCTGG, hPROF: CGTGATTATCAGCGCACACACTCA, hPROR: CTGGGTACATCCCACTGTACCAGA, hCNS1F: GGTTGTCTGGTCATGTCCTTACTCC, hCNSR: ATCACACATAGGGCTTGGGGTGAC, hCNS2F: TGGTGTCGATGAAGCCCGGC, hCNS2R: CATGGAGATGATCTGTCTGGGGGTAG, hExon11F: GTACACTCAAACAACCTCAAAGCTGC, hExon11R: TGCATATGCGTGAGATACACA GGTG, Ds1F: ATGGTGGAGATCACCAGCAAGCA, Ds1R: CTGGTCGGATTTCGCAGCTCCTA, Ds2F: CTCCTATTCCTTCTACCCCAGAAGCT, Ds2R: CAGACAGTGTGATGATAAGAGCCTGG, Ds3F: TATTGTGGTGGGCAGCATAGTGGA, Ds3R: ACCGATGGAGAAGCCAATGGAGAA, Ds4F: GAAAGAAGTAGGCACAGCGGTGAG, Ds4R: GTGGATAAATGACG TGCCCATGAG, AmCyanF: ACATCCTGTCCACCGTGTTCATGT, AmCyanR: ATGGTCACGGGCTTCTTGGTCTTGT, mEN2F: GAAAGACAGAATCGATAGAACTTGG, and mEN2R: AATATGTTTTCCTATCGGGGTCTAC.

EL4 sub-clone B02 cells were cultured in IMDM with l-glutamine and 25 mM HEPES (Cellgro) and 5% FBS. Primary T-cells were cultured in RPMI 1640 with l-glutamine (Cellgro) and 10% FBS. CD4⁺CD25⁺ and CD4⁺CD25⁻ T-cells from spleen were isolated using EasySep™ mouse CD4⁺CD25⁺ regulatory T-cell Isolation kit (STEMCELL). The purity of CD4⁺CD25⁺ T-cells and CD4⁺CD25⁻ T-cells was higher than 80%. mFoxp3 expression was analyzed using a Foxp3 staining kit (eBiosciences). To generate iTreg cells, CD4⁺CD25⁻ T-cells were cultured with plate-coated anti-CD3 (KT3, 5 μg/ml in PBS), anti-CD28 (1 μg/ml), recombinant human TGF-β1 (Peprotech, 5 ng/ml), and mouse IL-2 (Peprotech, 20 ng/ml). Pharmacological inhibitors, Smad3 inhibitor (SIS3, EMD Millipore), JNK Inhibitor II (SP600125, LC laboratories), CsA (Sigma-Aldrich), STAT5 inhibitor (Calbiochem, Millipore), JAK inhibitor (R545, Rigel), and Torin1 (Millipore) were used where indicated. For FACS staining, anti-CD3, anti-CD4, anti-CD25, and anti-Foxp3 (eBioscience) were used.

Chromatin immuno-precipitation assays were performed using freshly isolated CD4⁺CD25⁻ T-cells, CD4⁺CD25⁻ T-cells treated with anti-CD3 + anti-CD28 + TGF-β, and CD4⁺CD25⁺Foxp3⁺ nTreg cells from hFOXP3 BAC transgenic mice, as well as EL4 transfectants as described previously (28). Briefly, these cells were fixed (for 10 min at room temperature in 1% formaldehyde, 4.5 mM HEPES pH 8.0, 9 mM NaCl, 0.09 mM EDTA, and 0.045 mM EGTA) and sonicated (Bioruptor) in lysis buffer (1% SDS, 10 mM EDTA, and 50 mM Tris–HCl pH 8.0) with proteinase inhibitor (Sigma-Aldrich P8340). Pre-cleared lysates were incubated overnight at 4°C with polyclonal anti-acetyl histone H4 (Millipore) or control rabbit IgG (Santa Cruz). DNA fragments were isolated from the immuno-precipitated chromatin and analyzed by real-time PCR with SsoAdvance SYBR supermix (Bio-Rad). PCR primers for ChIP used were hFOXP3ProF: CTGGCATTTCCCATCCACACATAGA, hFOXP3ProR: TGAGTGTGTGCGCTGATAATCACG, hFOXP3CNS1F: GGTTGTCTGGTCATGTCCTTACTCC hFOXP3 CNS1R: ATCACACATAGGGCTTGGGGTGAC, hFOXP3CNS2F: ACCCAAGAAGGGCCAGGTCTTC, and hFOXP3En2R: GCCGGGCTT CATCGACACCA.

To confidently exploit knowledge of foxp3 gene expression for therapeutic benefit, we sought to establish whether the human gene can be expressed in both transgenic mice and cell lines using an hFOXP3 BAC clone (RP11-344O14). To avoid any potential disruption of Treg development from an additional functional hFOXP3 gene in the transgenic mice, and to provide a sensitive reporter of hFOXP3 expression, a fluorescent protein AmCyam gene was inserted into the hFOXP3 gene in the BAC DNA by homologous recombination, resulting in the BAC construct being unable to express hFOXP3 protein (hFOXP3 expression is replaced by AmCyan) (Figure 1A). During the BAC DNA modification process, we found a long repetitive DNA region located in the 5′-half of the clone (Figure 1A). Since this repetitive sequence interfered with the modification process, we removed the 5′-half of the BAC DNA sequence. The resulting BAC, therefore, contains 52 and 59 kb DNA sequences at the 5′- and 3′-flanking regions of the hFOXP3 transcription start site, respectively. To replace hFOXP3 expression with AmCyan, cDNA encoding this fluorescent protein was inserted just downstream of the first ATG of the hFOXP3 gene (the first ATG of the hFOXP3 gene was replaced by that of the AmCyan cDNA, resulting no hFOXP3 protein expression) (Figure 1A). C57BL/6 transgenic mice were generated using the modified BAC DNA, and AmCyan (hFOXP3) expression was analyzed by FACS.

Expression of both AmCyan and endogenous mFoxp3 was detected only within CD3⁺CD4⁺ cells (but not in CD3⁻ cells nor other CD3⁺ cells such as CD3⁺CD8⁺ cells) and was restricted to CD4⁺CD25⁺ cells from the thymus and spleen of the BAC transgenic mice. No AmCyan expression could be detected in wild-type (WT) mice (Figure 1B). This tells us that the developmental pathway for mouse nTreg, can provide all the cues for BAC hFOXP3 expression. However, although cell type expression patterns of AmCyan and mFoxp3 were similar, the percentage of mFoxp3⁺ cells was higher than that expressing AmCyan⁺. Since the ratio of AmCyan⁺ compared to mFoxp3⁺ cells was similar in spleen and thymus (in Spleen, AmCyan⁺ 6.17% vs. mFoxp3⁺ 10.9% and in Thymus, AmCyan⁺ 1.99% vs. mFoxp3⁺ 3.11%), hFOXP3 induction in tTreg precursor cells appeared to be somewhat less efficient than for the endogenous gene.

To examine the performance of the BAC construct in murine iTreg, we examined the effect of TGF-β in regulating hFOXP3 expression in in vitro. CD4⁺CD25⁻ T-cells were isolated from spleens of the hFOXP3 BAC transgenic and WT mice, and cultured with anti-CD3 + anti-CD28 + IL-2 with or without TGF-β (Figure 1C). Expression of AmCyan (hFOXP3) and mFoxp3 (internal staining) was observed only in cells cultured with TGF-β, indicating that hFOXP3 gene expression is also TGF-β dependent. Yet again, however, the percentages of TGF-β-induced hFOXP3⁺ iTreg from the BAC transgenic mice (28.8%) were lower than mFoxp3⁺ iTreg generated from WT (52.0%) and the BAC transgenic mice (50.1%). Taken together, these data suggest that fundamentally, hFOXP3 and mFoxp3 genes behave similarly in the murine inductive conditions, but that hFOXP3 (AmCyan) expression is somewhat constrained. This could be due to features of the transgenic mice generated with the engineered BAC DNA construct or to transcriptional activity of gene expression for hFOXP3 being weaker than that for mFoxp3. It is also possible that the Foxp3 staining process, in some way, inhibits AmCyan intensity. We, therefore, further investigated whether the regulatory regions identified for mFoxp3 expression also function for the hFOXP3 gene. As shown in previous studies (18–30), mFoxp3 transcription is regulated by three regulatory regions located in the promoter, CNS1 and CNS2. Many histone modifications can be used to assess the accessibility of chromatin in these regions by ChIP assay using anti-acetyl-histone and anti-methyl-histone (e.g., H3K27ac, H3K27me3, H3K9me3, and H4ac). We performed ChIP assays using anti-acetyl histone H4 and human-specific PCR primer sets to compare with the previous mFoxp3 ChIP data obtained using the same antibody (28, 31). The relative positions of the promoter, human CNS1 (hCNS1), and hCNS2 are shown in Figure 1A and Figure 2A, and human specificity of the primers for ChIP assay was shown in Figure 2B. Histone H4 molecules in the hFOXP3 promoter (hPro), hCNS1, and hCNS2 were highly acetylated in both the nTreg and iTreg (CD4⁺CD25⁻ T cells stimulated with anti-CD3 + anti-CD28 + TGF-β for 48 h) enriched populations, but not in the control freshly isolated CD4⁺CD25⁻ T-cells (Figure 1D). This pattern was identical to that in mFoxp3 gene as previously shown (28, 31). Taken together, these data suggest that mFoxp3 and hFOXP3 induction can be regulated by similar mechanisms through the promoters, hCNS1 and hCNS2.

The data obtained using hFOXP3/AmCyan transgenic mice suggested that both human and mouse foxp3 gene expression can utilize identical regulatory regions in the corresponding genes. To investigate mouse and human foxp3 gene expression further, and to establish a screening system for drugs influencing hFOXP3 gene expression, we generated a hFOXP3 expression reporter cell line using hFOXP3/AmCyan BAC DNA and EL4 sub-clone B02. As we have previously shown (28, 31), both EL4 sub-clones LAF and B02 expressed mFoxp3 after stimulation with anti-CD3 ⁺ anti-CD28 + TGF-β, indicating that mFoxp3 expression in these sub-clones is induced by TCR activation and TGF-β signaling as in TGF-β induced Treg, unlike the original EL4 cell line. Furthermore, using these sub-clones, we and others have identified diverse transcription factors involved in mFoxp3 gene expression (23, 26, 28, 29, 31). Therefore, we generated hFOXP3/AmCyan BAC (Figure 3) and mFoxp3/mCherry BAC (Figure 4) transfectants using EL4 B02 cells, as hFOXP3 and mFoxp3 expression reporter systems. To select BAC transfectants, the neomycin-resistant gene was inserted at the 5′-end of the BAC DNA (Figures 3A and 4A). Although BAC transfectants were selected by G418 as neomycin resistant, many transfectants might, in principle, have lost the foxp3 gene and/or essential transcriptional regulatory regions for FoxP3 expression during the BAC DNA integration process. To select for FoxP3 expressing transfectants, these G418 resistant cells were stimulated with anti-CD3 + TGF-β for 48 h and AmCyan and mCherry positive cells were sorted and cloned by limiting dilution. During this cloning step (culturing without stimuli for about 2 weeks), stimulated cells reverted to the “resting” state and did not express these fluorescent proteins without further stimulation.

Since consistent AmCyan (hFOXP3) gene expression in the hFOXP3/AmCyan transgenic mice could be detected in appropriate cell types, it appears that mouse and human foxp3 genes exploit common regulatory mechanisms. However, reporter systems using human cell lines would, if available, enable identification of any human-specific regulatory mechanisms. To this end, we have sought human T-cell lines that regulate inducible hFOXP3 expression akin to iTreg. None have been described. However, since we can stimulate the human T cell line JE6.1 with anti-CD3 and TGF-β (41, 42), we generated transfectants using the same hFOXP3/AmCyan BAC construct (Figure 3A) with the strategy that had proven successful for EL4 B02 cells (Figure S1 in Supplementary Material). However, the expression pattern of endogenous hFOXP3 mRNA from JE6.1 was unlike iTreg. hFOXP3 mRNA was detected in non-stimulated cells, yet not significantly upregulated by stimulation (Figure S1A in Supplementary Material). Importantly, AmCyan (hFOXP3) expression from the BAC DNA in JE6.1 transfectants was similar to that of the endogenous hFOXP3 (Figure S1C in Supplementary Material), but dissimilar to iTreg and reporter cells generated using the EL4 B02 line. In short, the expression of the foxp3 gene in Treg seems not only to be determined by regulatory regions in this gene but also by other, as yet, undefined regulatory mechanisms. The result suggests although EL4 B02 line is a mouse cell line, it remains the best option for a reporter system to date, albeit not covering as putative set of human-specific regulatory mechanisms.

To examine whether expression of AmCyan (hFOXP3) and mCherry (mFoxp3) in BAC transfectants was similar to mouse primary T-cells, cloned transfectants were re-stimulated by anti-CD3 + anti-CD28 with or without TGF-β and analyzed by FACS. The expression patterns of both AmCyan (hFOXP3) (Figure 3B) and mCherry (mFoxp3) (Figure 4B) in these transfectants were similar, and induction of both AmCyan and mCherry was TGF-β dependent, just as in hFOXP3 BAC transgenic mice and WT mice (Figure 1C). However, 3 to 8% of AmCyan⁺ and mCherry⁺ cells were always detected in cultures stimulated with anti-CD3⁺ anti-CD28, but without additional TGF-β (4.5% in Figure 3B and 3.7% in Figure 4B). This residual AmCyan and mCherry induction was, however, inhibitable by addition of anti-TGF-β (Figures 3C and 4C), implicating other sources of TGF-β as responsible (e.g., TGF-β expressed by EL4 cells themselves and/or bovine TGF-β in FBS).

We, next, examined whether the three regulatory regions identified in the hFOXP3/AmCyan transgenic mice (promoter, hCNS1 and hCNS2) (Figure 1D) and WT mice (28, 31) are also involved in induction of hFOXP3 expression in the EL4 reporter cells. We assessed the “open chromatin status” of these regulatory regions by ChIP assay (Figure 3D) using anti-acetyl histone H4 and the human-specific PCR primers (Figures 2A,B). As with the hFOXP3/AmCyan transgenic mice (Figure 1D), histone H4 molecules in the hPro, hCNS1, and hCNS2 were highly acetylated only in anti-CD3 + anti-CD28 + TGF-β stimulated cells, suggesting that hFOXP3 expression in the EL4 reporter cells could also be induced through these three regulatory regions with TCR and TGF-β signaling. These data suggest that the EL4 reporter system should be useful for further studies of hFOXP3 expression and identification of potential therapeutics in modulating hFOXP3 expression.

The regulation of hFOXP3/AmCyan expression in the reporter transgenic mice and transfectants appeared similar to that previously seen in mouse primary T-cells. This indicates that the human BAC DNA in both transgenic mice and transfectants contains all relevant regulatory regions required for Treg specific hFOXP3 gene expression in murine cells. We analyzed the integrated area of the hFOXP3 BAC DNA in the transgenic mice and transfectants by PCR using human-specific primer sets (relative positions of these primer binding sites are shown in Figure 2A). As shown in Figure 2C, strong amplification bands were detected with the human-specific PCR primer sets using DNA from HeLa cells and the BAC clone (RP11-344O14) but not using DNA from WT C57BL/6 mice (only weak non-specific bands were detected with some of these primers). In the transgenic mice, a strong PCR band was detected with the Us3 primer set (binding to −40 kb from the transcription start site of the hFOXP3 gene), but not with the Us4 primer (binding to −51 kb), implicating the 5′-end of the hFOXP3 BAC DNA located between −51 and −40 kb (Figures 2A,C). Since the neomycin resistant gene is located at the 5′-end of the hFOXP3 BAC DNA (Figure 3A), the hFOXP3/AmCyan EL4 transfectant contains the 5′-end of the BAC DNA, and indeed, a strong 5′-band was also detected with the most 5′-PCR primer set (Us4) (Figures 2A,C). In both the transgenic mice and the transfectants, strong PCR bands were detected with the most 3′-PCR primer set, Ds4, indicating that both 3′-ends of the integrated DNA are located near the 3′-end of the BAC DNA (Figure 2C). Taken together with the results of Treg-specific hFOXP3/AmCyan gene expression in the transgenic mice, our data suggests that the regulatory regions required to maintain Treg-specific hFOXP3 gene expression are located within the 110-kb (−51 to +59 kb) DNA sequence area shown in Figure 2A.

We have previously shown that NF-κB cRel (in promoter) (26), Smad3 (in CNS1) (28), NFAT (in CNS1) (28), AP1 (in CNS2) (31), and STAT5 (in CNS2) (31) cooperate to regulate mFoxp3 gene expression, and that pharmacological inhibitors of the activation pathways for these transcription factors blocked mFoxp3 gene expression. In order to provide a proof of principle test that our hFOXP3 construct might serve as a tool to test drugs for induction or repression of Foxp3 transcription, we examined the contribution of these transcription factors in regulating hFOXP3 gene expression. To this end, we used these same inhibitors in the context of the hFOXP3/AmCyan expression system (Figure 5). mFoxp3 expression was also analyzed using the mFoxp3/mCherry expression system as a control. Similar to mCherry (mFoxp3) expression, AmCyan (hFOXP3) expression was blocked by the inhibitors of these transcription factors. Taken together with ChIP data shown in Figures 1D and 3D, these findings suggest that hFOXP3 gene expression is also regulated by the same transcription factors through the promoter, hCNS1 and hCNS2.

Since downmodulation of Foxp3 expression by inhibitors for NFAT, Smad3, AP1, and STAT5 has previously been described in primary iTreg (28, 29, 31), the reporter cells seem useful for identifying some drugs that can control hFOXP3 gene expression in these cells and primary T cells. To further support the potential of this system for identifying drugs able to influence hFOXP3 gene expression, we assessed a newly synthesized JAK inhibitor (R545) (43). Since mouse and human foxp3 gene expression is regulated by STAT5, the JAK inhibitor would be expected to downmodulate gene expression through blocking phosphorylation of STAT5. Indeed, expression of both foxp3 genes is strongly inhibited by R545, suggesting that these reporter systems hold promise for screening drugs modulating hFOXP3 expression.

Next, we sought evidence for drug-induced enhancement of hFOXP3 expression using this reporter system. Since Foxp3 expression is inhibited by the mTOR signaling pathway antagonizing the function of Smad3, Foxo1, and Foxo3 (44, 45), we examined the effectiveness of an inhibitor (Torin1) (44, 46, 47) of mTORC1 and mTORC2 using the stimulated (anti-CD3 + anti-CD28 + 5 ng/ml TGF-β) hFOXP3/AmCyan reporter cells with different concentrations of Torin1 (0–100 nM) (Figure 6A). AmCyan expression was induced in 40% of the reporter cells without Torin1, and increased with Torin1 in a dose-dependent manner. Since mTORC1 inhibits TGF-β-mediated Smad3 activity, the effectiveness of Torin1 (100 nM) was assessed using different dose of TGF-β (0–5 ng/ml). As shown in Figure 6B, using 100 nM of Torin1, AmCyan⁺ cells could be detected following exposure to low doses of TGF-β, and numbers increased substantially with higher doses of TGF-β. Although 39.9% of Torin1 treated cells and 7.65% of non-treated cells were detected as AmCyan⁺ cells without addition of TGF-β, induction of these cells was inhibited, as before, by anti-TGF-β (Figure 6C), indicating enhancement of hFOXP3 expression is completely TGF-β dependent.