Broad expression of the FoxP3 locus in epithelial cells: A caution against an early interpretation of fatal inflammatory diseases following in vivo depletion of FoxP3-expressing cells

Abstract

Dogma that the Treg prevents catastrophic autoimmunity throughout the lifespan relies on the assumption that the FoxP3 locus is transcribed exclusively in Treg. To test the assumption, we used the Rag2^−/− and the Rag2^−/− mice with the Scurfy mutation (FoxP3^sf/y or FoxP3^sf/sf) to evaluate FoxP3 expression outside of lymphoid system. Immunohistochemistry and real-time PCR revealed FoxP3 expression in breast epithelial cells, lung respiratory epithelial cells and in prostate secretory epithelial cells, although not in liver, heart and intestine. The specificity of the assays is confirmed as the signals were ablated by the Scurfy mutation of the FoxP3 gene. Using mice with green fluorescence protein (GFP) open-reading frame knocked into the 3′ untranslated region of the FoxP3 locus, we showed that the locus is transcribed broadly in epithelial cells of multiple organs. These results refute an essential underlying assumption of the dogma and question the specificity of FoxP3-based Treg depletion.

Introduction

The forkhead transcription factor FoxP3 has been initially identified as causative mutation of a fatal autoimmune disease in mouse and human (1-4). More recently, FoxP3 was shown to be a key transcription factor for function of regulatory T cell (Treg) (5, 6) where it is expressed at levels comparable to a house-keeping gene in CD4⁺CD25⁺ Treg (5). In an attempt to test the physiological function of Treg, two groups have designed strategies to delete Tregs by treating transgenic mice expressing a diphtheria toxin receptor (DTR) under the control of the FoxP3 locus with DT (7, 8). The acute and fatal autoimmune diseases in adult and/or neonates were used to support a dramatic conclusion that the regulatory T cells prevent catastrophic autoimmunity throughout the lifespan of mice (7). For this conclusion to be valid, however, one must ascertain that Treg is the only cell type in mice that can express FoxP3, and by inference, the DTR in the transgenic mice. Unfortunately, no serious effort has been made to confirm this assumption.

We have reported expression of FoxP3 protein in thymic (9) and breast epithelial cells (10). While these data are inconsistent with the notion of Treg-exclusive expression of FoxP3, activation of the locus in these organs would in theory not affect DT-mediated fatal injury as these organs are not vital for survival of mice. In this study, we systemically investigated expression of FoxP3 by immunohistochemistry, real-time PCR and the knockin mice with a bicistronic FoxP3 locus that co-expresses the enhanced GFP (EGFP) under control of the endogenous FoxP3 promoter/enhancer elements (11). Here we report a broad expression of the FoxP3 gene in epithelial cells of multiple lineages. Our data raised a serious concern about interpretation of the fatal inflammation in DT treated FoxP3-DTR transgenic/knockin mice.

Materials and Methods

Experimental animals

WT or FoxP3^EGFP mice that express both FoxP3 and EGFP under the endogenous regulatory sequence of the FoxP3 locus have been described (11). BALB/c RAG-2^−/− and RAG-2^−/− mice with the Scurfy mutation of the FoxP3 gene have also been described (9). All mice were used at 6-8 weeks after birth.

Immunohistochemistry and immunofluorescence

Tissues from mice were fixed in 4 % paraformaldehyde, dehydrated, embedded in paraffin and sectioned according to the standard procedure. Endogenous peroxidase activity was quenched using 3% H₂O₂ for 30 min at room temperature; nonspecific binding sites in the sections were blocked using 5% normal bovine serum albumin (Sigma-Aldrich) in PBS for 1 h at room temperature. Sections of 5 μm were made and reacted with affinity-purified anti-FoxP3 polyclonal antibody (9). The Slides were washed in PBS and subsequently incubated with the second antibody, horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Santa Cruz), for 1 hour at room temperature. After being washed in PBS, slides were developed with 3, 3′-diaminobenzidine and counterstained with hematoxylin.

The FoxP3^EGFP mouse sections were incubated with rat anti-mouse K8 antibody (Troma I, Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA) over night at 4°C. This is followed by Cy3 conjugated donkey anti-rat secondary antibodies (rabbit absorbed, Millipore). The expression of GFP is visualized by Alexa 488-conjugated rabbit-anti-GFP IgG fraction (Molecular Probes) at 4°C.

Quantitative real-time PCR analysis

Real-time PCR for FoxP3 expression was performed as described previously (7). Samples were run in triplicate, and the relative expression was determined by normalizing expression of each target to the endogenous reference, hypoxanthine phosphoribosyltransferase (Hprt) transcripts. The primer sequences were as follows: FoxP3: sense 5-TACTTCAAGTTCCACAACATGCGACC-3, antisense 5-CGCACAAAGCACTTGTGCAGACTCAG-3; and Hprt: sense 5-AGCCTAAGATGAGCGCAAGT-3, antisense 5-TTACTAGGCAGATGGCCACA-3.

Results and Discussion

In order to avoid interference of FoxP3 from lymphocytes, we used RAG-2-deficient mice lacking T and B cells to analyze expression of the FoxP3 in non-lymphoid organs. As control for antibody specificity, we generated Rag2^−/− Scurfy mice that have a naturally occurring truncation mutation resulting in early termination codon. Because of the non-sense mediated decay of the mRNA, essentially no FoxP3 can be detected in the Scurfy mice (9). We performed immunohistochemical staining to detect expression of FoxP3 in various tissues using affinity purified polyclonal antibody (9). As shown in Fig. 1, in FoxP3^wtRag2−/− mice, FoxP3 was found in mammary epithelial cells, in lung bronchial epithelial cells, in prostate secretory epithelial cells. In contrast, no staining can be found in intestine (Fig. 1), kidney, liver and heart (data not shown). Since none of the tissues from the FoxP3^{sf/y or sf/sf} Rag2^−/− mice show staining, we conclude that the staining is specific for the FoxP3 protein as it is largely diminished by a FoxP3 mutation that leads to disappearance of mRNA by a nonsense-mediated decay (Fig. 2).

FIG. 1.

Immuohistochemical analysis to detect expression of FoxP3. Sections from the mammary gland, prostate, lung, and small intestine, of either Rag2^−/− BALB/c mice (top) or FoxP3^sfRag2^−/− BALB/c mice (bottom), were stained with anti-FoxP3 polyclonal antibody, and counterstained with hematoxylin. The second antibody used was horseradish peroxidase-labeled. Positive staining was found in the nuclei of lung respiratory epithelial (RE) cells, prostate secretory epithelial cells, and mammary epithelial cells. Apart from epithelial cells, no specific staining can be found in other cell types in these organs. Other organs analyzed but show no FoxP3 expression include liver, kidney, and heart (data not shown).

FIG. 2.

Identification and quantitation of the FoxP3 transcripts in nonlymphoid organs from either Rag2^−/− BALB/c mice or FoxP3^sfRag2^−/− BALB/c mice. A and B, Quantitative analysis of FoxP3 transcripts in mouse tissues by realtime PCR. A. A representative profile of FoxP3 amplification in the prostate tissue. B. The expression level of FoxP3 transcripts was expressed as fraction of Hprt. Data shown were means ± S.D of three mice and each mouse was performed triplicate experiments. C. RT-PCR analysis of FoxP3 expression in mammary gland and prostate. Lane 1, prostate from FoxP3^sf/y Rag2−/− mouse; lane 2, prostate from FoxP3^wt/yRag2^−/− mouse; lane 3, mammary gland from FoxP3^wt/wt Rag2^−/− mouse. Hprt was used as cDNA control. The molecular weight of the product is consistent with it being full length mRNA. This is confirmed by DNA sequencing of the RT-PCR product (data not shown).

In conjunction with immunohistochemistry, we investigated FoxP3 mRNA level in mouse various tissue by real-time PCR. A typical positive amplification is shown in Figure 2a, and the summary data from multiple organs are shown in Fig. 2b. These data demonstrated that FoxP3 were expressed at a significant levels in mammary gland, lung, and prostate in FoxP3^wt/yRag2^−/−, no expression can be detected in liver. Again, FoxP3 transcript was reduced by 20-500 fold in FoxP3^{sf/y or sf/sf}Rag2^−/− mice. It is worth noting that the levels of FoxP3 transcript is about 3-100 fold lower in other organs in comparison to those in the spleen. The overall lower abundance makes it harder to detect FoxP3 expression in non-lymphoid organs. In order to determine whether full length FoxP3 ORF was transcribed, we used primers that encompass the whole open-reading frame to amplify FoxP3 mRNA. The PCR products were analyzed by agarose gel electrophoresis. As shown in Fig. 2c, the molecular weight of the product is consistent with it being full length mRNA. This is confirmed by DNA sequencing of the RT-PCR product (data not shown).

In order to test whether the activation of the FoxP3 locus in the epithelial cells extends to the EGFP knocked into the 3′ untranslated region of the FoxP3 mRNA, we used anti-GFP antibodies to evaluate the GFP expression in the FoxP3^EGFP mice (11), using WT mice as control for nonspecific binding and autofluorescence. Consistent with the immunohistochemical findings, the GFP protein was detected in mammary epithelial cells, in lung bronchial epithelial cells, in prostate secretory epithelial cells of FoxP3^EGFP mice, as revealed by double staining with anti-GFP and anti-K8 antibodies. Non-specific anti-GFP binding and autofluorescence were not found in the corresponding tissue in wild-type mice (Fig. 3). We also observed no specific signal in kidney, liver, intestine, and heart in the FoxP3^EGFP mice (data not shown). It is well established that immunofluorescence with anti-GFP antibodies, a strategy used here, is more sensitive and better able to discriminate between authentic GFP expression and autofluorescence (12). Relying on direct GFP visualization may explain other's failure to observe FoxP3 locus activation in the thymic epithelial cells (13). Our previous study also showed much lower expression of FoxP3 in Rag1-deficient thymic tissue (9). Given the demonstrated function of T cells in development of thymic epithelial cells (14), it is plausible that the difficulty in detecting FoxP3 in Rag-deficient mice may be an indirect consequence of abnormal development of epithelial cells. However, it is unclear whether the different approaches explained lack of report on the GFP in non-lymphoid organs, as others have not directly addressed the issue. Since the ORF of DTR was inserted in similar fashion, it is intriguing that, in the transgenic and knockin mice used by others for Treg-deletion with DT treatments (7, 8), the epithelial cells may be among those targeted by DT-treatment. Since epithelial cells of vital organs expressed FoxP3 at high levels, such targeting, either alone or in combination with Treg depletion may contribute to lethality caused by DT-treatment.

FIG. 3.

Activation of the FoxP3 locus as revealed by GFP expression in the FoxP3^EGFP mouse. Sections from the lung, mammary gland, and prostate were stained with anti-GFP (green) and K8 (red). Tissues were obtained from 8-week old mice. Sexes of mice used are marked as either M (male) or female (F). Apart from epithelial cells, no specific staining can be found in other cell types in these organs. Corresponding tissues from WT mice were used as negative control. Additional tissues tested negative include liver, heart, kidney, and intestine (data not shown).

Finally, FoxP3 is an interesting transcription factor that binds to a large number of genes important for a number of fundamental cellular functions (10, 15-17). Our documentation of its broad expression in epithelial cells of multiple organs suggests that it may play a more broad function outside of regulatory T cells. In this regard, we reported critical roles for FoxP3 as a regulator of thymopoiesis when expressed in thymic epithelial cells and as an important tumor suppressor gene when expressed in mammary epithelial cells, likely through regulation of oncogenes such as ErbB2 and Skp2 (9, 10, 17).