TSLP Expression: Analysis with a ZsGreen TSLP Reporter Mouse

Abstract

Thymic stromal lymphopoietin (TSLP) is a type I cytokine that plays a central role in induction of allergic inflammatory responses. Its principal targets have been reported to be dendritic cells and / or CD4 T cells; epithelial cells are a principal source. We report here the development of a reporter mouse (TSLP-ZsG) in which a ZsGreen (ZsG)-encoding construct has been inserted by recombineering into a bacterial artificial chromosome (BAC) immediately at the translation initiating ATG of TSLP. The expression of ZsG by mice transgenic for the recombinant BAC appears to be a faithful surrogate for TSLP expression, particularly in keratinocytes and medullary thymic epithelial cells (mTECs). Limited ZsG and TSLP mRNA was observed in bone-marrow derived mast cells, basophils and dendritic cells. Using the TSLP-ZsG reporter mouse, we show that TNFα and IL-4/IL-13 are potent inducers of TSLP expression by keratinocytes and that local activation of Th2 and Th1 cells induces keratinocyte TSLP expression. We suggest that the capacity of TSLP to both induce Th2 differentiation and to be induced by activated Th2 cells raises the possibility that TSLP may be involved in a positive feedback loop to enhance allergic inflammatory conditions.

Introduction

Thymic stromal lymphopoietin (TSLP) is a type I cytokine that together with interleukin-7 (IL-7) plays an important role in B and T cell development (1) in mice and in T cell development in humans (2). TSLP is a critical inducer of allergic inflammatory responses (3). It shares with IL-7 the use of IL-7Rα as a receptor component but uses the TSLPR rather than the γc chain to form a signaling complex (4). It has been reported that TSLP activates Jak1 and Jak2 to cause STAT5 phosphorylation while IL-7 achieves STAT5 phosphorylation by activating Jak1 and Jak3 (5).

A large body of research has implicated TSLP as playing a major role in the induction of Th2 type immune responses and in the mediation of allergic inflammation in the skin, lung and intestine (3). There is much evidence that TSLP acts on dendritic cells that, in turn, favor Th2 differentiation when they present antigen to naïve CD4 T cells in draining lymph nodes (6,7). In particular, TSLP-treated DCs, rather than producing pro-inflammatory cytokines, express OX-40 ligand, which in turn plays a role in induction of Th2 differentiation by CD4 T cells (8). Such OX-40 ligand-stimulated Th2 cells have been reported to produce substantial amounts of TNFα and little IL-10 (6).

TSLP can also act directly on naïve CD4 T cells (9) and may aid their differentiation to Th2 cells by providing the STAT5 signals that have been shown to be essential for in vitro Th2 differentiation (10). Furthermore, TSLP can synergize with IL-33 in inducing both cytokine-dependent IL-13 and IL-5 production by Th2 cells and in driving Th2 cell proliferation (11). TSLP may also enhance IL-33-mediated expansion and IL-13-production by type 2 innate lymphoid (ILC2) cells (12), potentially contributing to allergic inflammation. The relative contribution of TSLP-activated DC, of direct action of TSLP on differentiation of naïve CD4 T cells to the Th2 phenotype and of TSLP action on differentiated Th2/ILC2 cells to sustain allergic inflammation remains to be determined.

The study of the regulation of TSLP production has been somewhat enigmatic as direct visualization of cytosolic TSLP has been difficult. In general, TSLP has been shown to be a product of epithelial cells such as skin keratinocytes (13). There is some controversy as to whether mast cells and/ or basophils are a rich source of TSLP (14). It has been proposed that papain and other cysteine proteases act as allergens because they stimulate basophils to produce TSLP (15) although it is also plausible that papain acts directly on keratinocytes and other epithelial cells to induce expression of the cytokine. Strikingly, activation of PAR2 receptors has also been implicated in TSLP induction (16), although here it is serine proteases rather than cysteine proteases that are inducers. Equally interesting is the concept that TSLP may be part of a feedback loop in which it both induces/ sustains IL-4/IL-13-producing Th2 cells and in which its production is stimulated by cytokines produced by “inflammatory” Th2 cells.

To examine these issues in greater detail, we prepared a surrogate for TSLP expression in which a ZsG construct was introduced by recombineering at the translation-initiating ATG in BAC clone RP23-256L23. Substantial amounts of 5′ and 3′ DNA flank the TSLP gene in this 183 kB BAC suggesting that many of the regulatory elements controlling TSLP expression may be present in the introduced genetic material and thus that the reporter would reflect physiologic expression of TSLP.

Materials and Methods

Mice

C57BL/6 mice were purchased from Taconic Farms. BAC transgenic mice were bred, and all animals were housed, in the National Institute of Allergy and Infectious Diseases pathogen-free animal facility and used between 8-20 weeks of age. All experiments were done under a protocol approved by the National Institute of Allergy and Infectious Diseases Animal Care and Use Committee.

Generation of TSLP/ZsG mice by recombineering

The murine BAC clone RP23-256L23, containing the TSLP locus, was obtained from the Children's Hospital Research Institute BAC library (http://bacpac.chori.org). BAC DNA was purified using the Nucleobond BAC kit (Clontech) and characterized by PCR (5′: CGACAGC ATGGGTGACTATG, 3′: GCTGTCAAAGGGTTCAGACC). Generation of a modified BAC containing the ZsG reporter at the ATG site of the TSLP gene was done using the galK recombineering technique (17 and http://recombineering.ncifcrf.gov/protocol/Protocol2_CKO_vectors.pdf). Briefly, ZsG was cloned with BAC arms corresponding to the target BAC sequences surrounding the ATG at the beginning of the TSLP gene and recombineered into the BAC. The stop codon and internal polyA from ZsG were retained in the construct to avoid any translation of the additional copies of TSLP or of a fusion protein.

Mice were generated in the SAIC/NCI/LASP facility in Frederick, MD by pronuclear microinjection of fertilized C57BL/6 mouse eggs and verified by Northern blotting. 5-10 founders were kept for screening ZsG expression. Mice used for this work contained 2-3 copies of the recombineered BAC in their genome.

Subcutaneous injection of cytokines

Mice were shaved 24-48h before any subcutaneous injection in the skin. Cytokines (IL-4, IL13, TNFα, IL1β or IFNγ) were injected twice in the skin separated by 24 hours at a concentration of 1μg/ml except where indicated.

Generation of TH1 and TH2 primed cells in vitro and adoptive transfer

Lymph nodes cells from OTII RAG1^-/- C57BL/6 mice were primed in vitro with irradiated (30GY) T cell-depleted B6 splenocytes and OVA peptide (323-339) under Th1 or Th2 conditions (18). 3-5 million primed cells were injected I.V. into syngeneic mice. Recipients were challenged twice, separated by 24 hours, s.c. in the back with 100μg OVA or lysozyme in 200 μl PBS or with PBS alone as a control and analyzed 24h later. In some experiments, 5ng LPS was added to the antigen challenges.

Preparation of bone marrow derived dendritic cells, basophils and mast cells

Bone marrow cells from TSLP-ZsG reporter mice were cultured with GM-CSF (30 ng/ml; PeproTech) in RPMI 1640 medium supplemented with 10% FBS, 2mM glutamine, 100 U/ml penicillin and streptomycin. On day 8, the suspended cells were stained with anti-CD11c-PE and anti-CD19-APC (ebioScience) and the CD11c⁺CD19^- cells (dendritic cells) were purified by cell sorting using a FACS Aria III (Becton Dickinson). After sorting, the dendritic cells were stimulated with 1μg/ml LPS (Invivogen) plus 10ng/ml IL4 (PeproTech) for 4hrs. Bone marrow cells from TSLP-ZsG reporter mice were also cultured with IL3 (50ng/ml; R&D) in RPMI 1640 medium supplemented with 10% FBS, 2mM glutamine, 100 U/ml penicillin and streptomycin. On day 9, the suspended cells were strained with anti-cKit-APC and anti-FcεR1-eF450 (ebioScience). The cKit⁺ FcεR1⁺ cells (mast cells) and cKit^- FcεR1⁺ cells (basophils) were purified by cell sorting. After sorting, basophils were stimulated with PMA (1μM) and inomycin (10ng/ml) for 3hrs. The mast cells were expanded by culture in IL-3 for 8 days and then stimulated with PMA and ionomycin.

Flow cytometry analysis

Cell surface staining was performed in PBS containing 1% FBS with different combinations of antibodies. Antibodies were purchased from BD Biosciences, E-bioscience, Biolegend, R&D, Invitrogen and Vector. Anti-pan-cytokeratin antibody was obtained from Dako and anti-TSLP antibody from Santa Cruz technology (M140). Monensin (2μM)-treated cells were stimulated with PMA (10ng/ml) and ionomycin (1μM) for 3h at 37C to induce cytokine expression. To detect intracellular cytokines, cells were fixed in PFA 4%/PBS for 15min, then washed and permeabilized in 0.5% TritonX100, 1% FBS before staining for cytokines (IL-4, 13, IFNγ). Analysis was done on a BD LSRII using FlowJo Software (Tree Star Inc.).

Preparation of thymic stromal cells for flow cytometric analysis and sorting

Thymic stromal cells were prepared using methods modified from those reported by Gray et al. (19). Following release of thymocytes by gentle teasing of the thymus, thymic fragments were digested with Collagenase/Dispase at 0.25% w/v plus DNase 1 at 0.125% w/v (Roche) in 4 sequential incubations at 37°C. Reactions were stopped by addition of FCS to 20%. For thymic epithelial cell (TEC) analysis, single cell suspensions were stained with anti-CD45.2-Pacific Blue (104; Biolegend), anti-EpCAM-PE (G8.8; eBiosciences), anti-MHC class II-APC (M5-114; Ebiosciences), and UEA-1 biotin (Vector). Dead cells were excluded with propidium iodide staining. For medullary (m) TEC sorting, enriched TEC preparations were made by discontinuous density gradient fractionation (20). Enriched TECs were stained with anti-CD45.2, anti-MHC class II, anti-EpCAM, and UEA-1. CD45.2- negative, MHC II⁺, UEA^neg (cTEC), MHC II⁺, UEA⁺ ZsG^neg (mTEC) and MHC II⁺, UEA⁺ ZsG⁺ cells were collected using a FACSAria flow cytometer (BD) and analyzed using FlowJo (TreeStar, San Carlos, CA) FACS analysis software.

RNA extraction and DNA microarray target preparation

Flow-sorted mouse cells were lysed in 600 μL of RLT buffer (Qiagen, Valencia, CA). Due to low number of ZsG+ mTEC cells (<1,000), genome copy number was measured by qPCR before RNA lysate was processed. Briefly, DNA was extracted from a 50μL aliquot of cell lysate according to the AllPrep DNA/RNA 96-well kit protocol. Absolute quantitation analysis of DNA using universal express QPCR supermix universal with premixed ROX, mouse DNA standard, and ready made 20× mouse actB primer and probe mix according to manufacturer's instructions (Life Technologies, Carlsbad, CA). The lowest cell copy number in the sample set was 654. Due to low number of cells, RNA lysate was combined with 12 μg of linear acrylamide (Life Technologies, Carlsbad, CA) to increase RNA yield from low cell count samples. RNAs were extracted using Qiagen RNeasy 96 well system according to manufacturer's recommendations except each RNA sample was treated with 27 units of DNAse I (Qiagen, Valencia, CA) for 15 minutes at room temperature during extraction to remove gDNA. RNA quality was determined using 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) and the Agilent RNA 6000 Pico kit. RNA was quantitated by RT-QPCR as described in Mackey-Lawrence et al., 2013 (21). The lowest RNA yield of 0.9 ng was from 654 cells. DNA microarray targets were prepared from extracted RNAs using WT-Ovation™ Pico RNA amplification system and FL-Ovation™ cDNA biotin module according to manufacturer's instruction (Nugen Inc., San Carlos, CA).

Hybridization, fluidics and scanning were performed according to standard Affymetrix protocols (http://www.affymetrix.com). Command Console (CC v3.1, http://www.Affymetrix.com) software was used to convert the image files to cell intensity data (cel files). All cel files, representing individual samples, were normalized by using the trimmed mean scaling method within expression console (EC v1.2, http://www.Affymetrix.com) to produce the analyzed cel files (chp files) along with the report files. The cel files were input into Partek Genomics Suite software (Partek, Inc. St. Louis, Mo., v6.6-6.12.0907) and quantile normalized to produce the principal components analysis (PCA) graph and dendrogram. An ANOVA was performed within Partek to obtain multiple test corrected p-values using the false discovery rate method (22) at the 0.05 significance level and was combined with fold change values, signal confidence (above background), and call consistency (as a percent) calculated using custom Excel templates for each comparison of interest. IPA (Ingenuity®Pathway Systems, www.ingenuity.com) was used to generate a custom pathway with selected array data overlay The array data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus (23) and are accessible through GEO series accession number GSE54343. (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE54343)

Confocal, Two-Photon Intravital Skin Microscopy and Image Analysis

All tissue samples were fixed in 1% PFA overnight at 4°C, then transferred into 30% sucrose overnight at 4°C. Tissue were embedded in OCT medium (Sakura #4583), frozen on dry ice and cut with a cryostat and kept at -80°C until staining.

All confocal images were acquired with a Leica TCS-NT/SP confocal microscope (Leica Microsystems, Exton, PA USA) using a 40× oil immersion objective. Fluorochromes were excited using an argon laser at 488nm for FITC, a krypton laser at 568nm for Alexa 568 and with a Helium-Neon laser (HeNe) at 633nm for far-red antibody detection. DAPI was excited using an Argon laser (Enterprise model 651, Coherent Inc.) at 364 nm. Detector slits were configured to minimize any crosstalk between the channels. Imaging of the ears by intravital multiphoton microscopy was acquired using an inverted LSM 510 NLO multiphoton microscope (Carl Zeiss Microimaging) as described previously (24). Images were processed using Leica TCS-NT/SP software (version 1.6.587), Imaris 7.5.1 (Bitplane AG, Zurich Switzerland), and Adobe Photoshop CS3 (Adobe systems).

Statistical Analysis

Standard error of the means was calculated by dividing the standard deviation of the population by the square root of the number of samples. Correlation coefficient (coefficient of determination, r²) in linear regression analysis was calculated in Excel spread sheets.

Results

Preparation of TSLP-ZSG Mouse

BAC clone RP23-256L23 was obtained from the Children's Hospital & Research Center, Oakland, CA. The BAC contains TSLP flanked 5′ by ∼115 kB and 3′ by ∼65 kB suggesting that it may have all of the cis-regulatory elements necessary for physiologic expression of TSLP. As described in Materials and Methods, a construct encoding ZsG was inserted at the translation-initiating ATG for TSLP using the techniques of recombineering (Fig. 1A). After purification, the recombinant BAC construct was introduced by pronuclear injection into fertilized C57BL/6 oocytes that were then transferred to pseudopregnant foster mothers. Mice were screened by Southern blotting for expression of the BAC construct and expression verified by Northern blotting for ZsG mRNA.

Figure 1.

TSLP-ZsG reporter mice.

A. Strategy for preparation of a BAC into which a ZsG construct with a polyA site and a stop codon has been introduced at the ATG of TSLP by recombineering.

B. Analysis of TSLP mRNA in ears of mice treated with ethanol or MC-903 for 1, 3, 4 and 5 days and a comparison of induction TSLP and CYP243A1, a known Vitamin D-dependent gene, after 4 days of application of ethanol or MC-903 to the ear. This experiment was performed three times with similar results.

C. Tile scans of ears of BAC transgenic mice treated with ethanol or MC-903. ZsG is shown in green and staining with anti-cytokeratin 1 in red.

D. Ethanol was applied to the left ears and MC-903 to the right ears of ZsG BAC transgenic mice for 4 or 7 days. One day later, mice were sacrificed, single cell suspensions from ear epidermis prepared and mRNA for TSLP and ZsG determined. Mean and standard error of the mean are shown. This experiment was performed twice with similar results.

E. Regression analysis of TSLP and ZsG mRNA expression in right ears of individual mice treated with MC-903 for 4 or 7 days.

Vitamin D3 Induces TSLP mRNA and ZsG Expression

As an initial test of the expression of ZsG as a surrogate for TSLP, we utilized the capacity of 1,25 dihydroxy-vitamin D3 and its non-calcemic analog MC-903 to induce keratinocyte expression of TSLP (13) and ZsG mRNA. As shown in Fig. 1B, TSLP mRNA increases over the first three to four days of application of MC-903 to the ear, reaching approximately 60-fold above levels in ears treated with ethanol.

ZsG fluorescence in the epidermal layer of the ear could be detected 2-days after application of MC-903 and reached a maximum on day 5. Examining an assemblage of a large portion of the ear sample by confocal microscopy by a montage of contiguous fields of view (a “tile” scan) revealed that ZsG expression was patchily expressed in the keratinocyte layer (Fig.1C). In some areas, keratinocytes were intensely fluorescent whereas in adjacent areas, there was little or no ZsG expression in response to vitamin D3 application.

To confirm that ZsG was faithfully reporting TSLP production, we painted the left ears of transgenic mice with ethanol and the right ears with MC-903 for 4 days (animals 1-3) or 7 days (animals 4-6) and one day later isolated cells from the epidermis with trypsin. Single cell suspensions were analyzed by RT-PCR for TSLP and ZsG mRNA. There was striking induction of both TSLP and ZsG mRNA by MC903 in the 4 and 7 day groups (Fig.1D). Furthermore, a regression analysis of TSLP and ZsG mRNA for individual animals showed a positive slope with an r² value of 0.36 (Fig. 1E) implying that ZsG mRNA expression tracked TSLP mRNA levels.

These results indicate that ZsG is a good surrogate for TSLP and that keratinocytes are the major TSLP-expressing cells in Vitamin D3-treated skin.

TSLP/ ZsG expression in mTECs

TSLP was initially described as a factor expressed in supernatants of a thymic stromal cell line that could support the growth of a pre-B cell line (25). Subsequently, it was reported that in the human thymus, TSLP was largely expressed in Hassell's corpuscles (26). Although Hassell's corpuscles are said to be poorly developed in the mouse, since we had the capacity to isolate TSLP-expressing thymocytes based on ZsG expression, we undertook to study these cells. We identified a small population of ZsG+ cells in the thymic medulla by confocal microscopy that were UEA+ (Fig. 2A). Flow cytometric analysis indicated that these ZsG+ cells constituted ∼1% of mTECs, defined by their co-expression of EPCAM, UEA and MHC class II (27, 28) (Fig. 2B). We purified cortical thymic epithelial cells (cTECs) (EPCAM+, UEA- cells) and mTECs (EPCAM+, UEA+ cells); the latter were separated into ZsG+ and ZsG- populations.

Figure 2. Expression of ZsG in thymic epithelial cells.

A. Confocal microscopy of a thymic section from a ZsG BAC transgenic donor stained for UEA expression demonstrating a cell that is ZsG+ and UEA+.

B. Flow cytometric analysis of thymic cells prepared as described in Materials and Methods from a ZsG BAC transgenic donor stained with anti-CD45, anti-EpCAM and anti-UEA-1. CD45-negative cells were analyzed for Ep-CAM expression and the EpCAM-positive cells for UEA and ZsG expression.

cTECs expressed >40 times more β5T mRNA than the ZsG- or ZsG+ mTecs (Fig. 3). β5T is a thymoproteasome subunit reported to be exclusively expressed in cTECs (29). ZsG- mTECs expressed >14 times as much AIRE mRNA than the cTECs. These relative expression results confirm our purification procedure. The ZsG+ and neg. mTECs expressed similar amounts of AIRE mRNA. Essentially all ZsG mRNA was expressed in the ZsG+ mTECs and TSLP mRNA was expressed to a substantially greater degree in the ZsG+ mTECs than the ZsG- mTECs. In a subsequent analysis of mRNA expression, we observed the involucrin, a marker for terminal epidermal differentiation (30), was ∼4-fold overexpressed in ZsG+ mTECs compared with ZsG- cells (Fig. 3).

Figure 3. Gene expression by cTECs and ZsG+ and ZsG- mTECs.

CD45-negative mTECs and ZsG-positive and ZsG-negative mTECs were isolated from the thymus of ZsG BAC transgenic mice by fluorescence activated cell sorting and relative expression of ZsG, b5t, TSLP, AIRE, CD11c and Ivl mRNA determined by real time PCR analysis. This experiment was performed once but confirmatory data was obtained in an independent microarray experiment.

To take further advantage of the ZsG marker, we purified both cTECs and ZsG+ and neg mTECs and carried out a microarray analysis. Strikingly, principal component analysis revealed that ZsG+ mTECs were substantially different from ZsG- mTECs and from cTECs (Supplementary Figure 1). Interestingly, ATG13, BCN1 and Eph receptor A1 were over expressed in ZsG+ mTECs; ATG13 has 6.8 fold ratio in ZsG+ compared to ZsG- mTECs, BCN1, 14.9, and Eph receptor A1 3.2. All three are part of an autophagy pathway that is regulated by mTOR and which, in other situations, may be involved with autophagic cell death. Indeed, as shown in Supplementary Figure 2, ATG13, BCN1 and EPH A1 are part of a pathway in which several of the members are over expressed and in which other members of the pathway that are normally inhibited when ATG13 and related genes are overexpressed are, in fact, down regulated (31). In prostatic epithelial cells and endothelial cells, but not fibroblasts, treatment with ephrin-A1 inhibits cell proliferation (32).

TNFα and IL/4/13 induce ZsG expression in keratinocytes

We then turned to the analysis of ZsG expression in the back skin or ears of mice in response to injection of various cytokines. PBS, TNFα (1 μg), IL-13 (1 μg) or IL-4 (1 μg) were injected into the shaved back skin of TSLP-ZsG transgenic mice on two successive days. Confocal microscopy carried out on skin biospies taken 24 hours later revealed that TNFα, IL-4 and IL-13 each caused induction of ZsG, with TNFα being the most striking inducer (Fig. 4A). A particularly graphic example of the degree of induction is shown by injection of TNFα or PBS into the ear of TSLP-ZsG mice twice and examination by multiphoton microscopy 24 hours later (Fig. 4B). In this case, the keratinocytes are being seen from directly above whereas in the confocal views, they are being seen transversely.

Figure 4.

Induction of ZsG expression by IL-13, IL-4 and TNFα.

A. Tile scans of back skin from ZsG BAC transgenic mice treated by two daily injections of PBS, IL-13, IL-4 or TNFα and biopsy 1 day later. This experiment was repeated at least three times with similar results.

B. Multiphton microscopy of ear skin from ZsG BAC transgenic mice treated by two daily injections of PBS or TNFα and analysis 1 day later. This experiment was done twice with similar results.

We also evaluated ZsG expression in response to TNFα by flow cytometry. Staining epidermal preparations from injected skin with anti-CD3, anti-CD11c and anti-cytokeratins 10 or 14 revealed three major cell populations. Since staining with anti-cytokeratins required fixation, which diminishes ZsG fluorescence, we repeated the staining on non-fixed cells with anti-CD3 and anti-CD11c only. Only the cells negative for both these markers expressed ZsG, implying that only the anti-cytokeratin-positive cells expressed ZsG (Fig. 5). Since cytokeratins are mainly restricted to epidermal cells (33), we conclude in mice stimulated with TNFα, ZsG expression in the skin was confined to the keratinocyte population.

Figure 5. Flow cytometric analysis of ZsG expression.

Epidermal cells were isolated from the skin of B6 mice or ZsG BAC transgenic mice that were untreated, injected with PBS or with TNFα. CD3-negative, CD11c-negative cells were analyzed for expression of ZsG. Percent ZsG+ cells and the product of percent ZsG+ cells and MFI (MFIx%+) are indicated. This experiment was performed twice with similar results.

In addition to TNFα, IL-4 and IL-13, IL-1 and LPS each induced ZsG expression upon injection into back skin when evaluated by confocal microscopy (data not shown).

Subcutaneous activation of Th2 and Th1 Cells induces ZsG Expression

The finding that TNFα and IL-4/13 induce TSLP/ZsG expression when injected into the skin implies that CD4 T cells capable of making these cytokines, if stimulated in the skin, should induce keratinocytes to express ZsG. Accordingly, we primed OT-2 cells in vitro under Th1- or Th2-inducing conditions through one or two rounds of priming. We verified that the cells had attained either a Th1 or a Th2 phenotype based on their production of IFNγ or of IL-4/13 (Fig. 6A). Both the Th1 and Th2 cells were robust producers of TNFα. After culturing the primed cells in IL-2 (Th1) or IL-2 and IL-7 (Th2), we transferred 3-5×10⁶ cells intravenously. Twenty-four hours later, we challenged the mice with 100 to 250 μg OVA or, as a control, with PBS, in the back on two successive days and evaluated keratinocyte expression of ZsG 24 hours later by confocal microscopy. Recipients of both Th1 and Th2 cells showed strong induction of ZsG expression in the keratinocyte layer in response to challenge with Ova but not HEL (Fig. 6B). PBS-injected controls showed little or no induction.

Figure 6.

Stimulated Th1 and Th2 cells induce ZsG expression.

A. Th1 and Th2 cells were prepared from OT-II cells and their expression of IL-4, IFNγ and TNFα upon stimulation with PMA and ionomycin indicated.

B. Confocal microscopy of back skin of mice that received Th1 or Th2 OT-II cells (see Results) and were challenged by injection of lysozyme or ovalbumin. This experiment was performed three times with similar results.

C. Flow cytometry of epithelial cells from the back skin of mice that received Th1 or Th2 OT-II cells and were challenged with PBS or with 50 or 250 μg of ovalbumin. This experiment was performed twice with similar results.

We also tested the contribution of IL-4/ IL-13 to the Th2 induction of ZsG expression by keratinocytes. Th2 cells were transferred into B6 mice. These animals were challenged with lysozyme or ovalbumin. Some of the ovalbumin treated mice received an intravenous injection of monoclonal anti-IL-4Rα (M1) prior to challenge. This treatment reduced but did not abolish ZsG induction implying that IL-4 or IL-13 was responsible for only some of the induction of ZsG and presumably TSLP by Th2 cells (Fig. 7).

Figure 7.

Anti-IL-4Rα partially inhibits ZsG induction by stimulated Th2 cells.

OT-II Th2 cells were transferred to ZsG transgenic mice and challenged with lysozyme or with ovalbumiun with or without M1 (anti-IL-4Rα). Tile scans of confocal micrographs and MFI are presented with calculated means and standard errors of the mean.

TSLP/ ZsG Expression in Bone Marrow Derived Dendritic Cells, Basophils and Mast Cells

While there is general agreement that epithelial cells can express TSLP when appropriately stimulated, there has been controversy about expression of TSLP by mast cells, basophils and dendritic cells (14). To study this point, we prepared bone marrow-derived mast cells, basophils and dendritic cells by culturing bone marrow cells from TSLP-ZsG and wild-type B6 mice in IL-3 (mast cells and basophils) or GM-CSF (dendritic cells). On day 9 of the IL-3 culture, both FcεRI+ c-kit+ cells, identified as mast cells (34), and FcεRI+. c-kit-, CD49+ cells, identified as basophils were present (35). When analyzed by flow cytometry, we found few if any mast cells or basophils to be ZsG+ after stimulation with PMA and ionomycin (Fig. 8). We analyzed DC on day 8 of culture; no ZsG+ cells were seen by flow cytometry (data not shown).

Figure 8.

Limited ZsG and TSLP expression by stimulated bone marrow-derived mast cells, basophils and dendritic cells.

Mast cells, basophils and dendritic cells were prepared from bone marrow of ZsG BAC transgenic mice as described in Materials and Methods. A. Mast cells (FcεR1+ DX5-) and basophils (FcεR1+ DX5+) cells were analyzed for ZsG expression in response to PMA and ionomycin by flow cytometry. This experiment was performed twice with similar results. B. Dendritic cells, basophils and mast cells were analyzed for expression of ZsG and TSLP mRNA, using 18S RNA to standardize. Dendritic cells were stimulated with LPS and IL-4. Mast cells and basophils were stimulated with PMA and ionomycin as described in Materials and Methods. Mean and std. error of the mean of three replicate cultures are shown. This experiment was performed twice.

Nonetheless, when we analyzed message expression, we did detect induction of both TSLP and ZsG mRNA in bone marrow-derived mast cells, basophils and DCs (Fig. 8). Basophils and mast cells were stimulated with PMA and ionomycin and analyzed 3 hours later; DCs were stimulated with LPS plus IL-4 for 4 hours. Based on our failure to detect ZsG protein expression in mast cells, basophils or DCs, we conclude that these three cell types, while capable of producing TSLP, do so to a much lower extent than do keratinocytes. Consistent with this, when we compared cycle number differences in the real time PCR analysis for TSLP, ZsG and 18s RNA, MC-903-stimulated keratinocytes showed an apparent greater expression of 100 to >1000 fold (Supplementary Figure 3) than did the three hematopoietic cell types. Even if 18s RNA expression is somewhat different in these various cells types, this result is nonetheless consistent with a very much lower degree of expression of both TSLP and ZsG in DCs, basophils and mast cells then in keratinocytes.

Discussion

We provide evidence that green fluorescence by cells from a BAC-transgenic mouse in which ZsG has been recombineered into the TSLP gene of BAC RP23-256L23 at the translation initiating ATG is a good reporter of TSLP expression. Vitamin D3 and MC903 were strong stimulants of ZsG expression. They modify RXR nuclear factor regulation of the TSLP promoter (36). TNFα was also a very strong stimulant of ZsG expression, presumably acting through TNF-RI expressed on keratinocytes and activating NF-κB and AP-1 (37). The mechanisms through which IL-4 and IL-13 induce TSLP expression are not clear.

Keratinocytes showed bright fluorescence upon stimulation by these stimulants injected into the ear or the skin of the back; LPS and Il-1β also modestly induced ZsG expression. Thus far, we have observed little or no fluorescence in hematopoietic cells at sites of vitamin D3 or cytokine injection.

TSLP has been reported to exert much of its function by acting on dendritic cells in the tissues and causing them to adopt a phenotype that was most effective in stimulating responding CD4 T cells to differentiate to Th2 cells (6,7). The TSLP-treated dendritic cells express OX-40 ligand and fail to make cytokines such as IL-12 that are associated with differentiation of responding CD4 T cells to a Th1 fate (8). The expression of OX-40L on DC aids their priming of normal antigen-specific naïve CD4 T cells to Th2 cells. Further, Liu and his colleagues have observed that CD4 T cells primed by OX-40L-expressing APC are more likely to produce TNFα and less likely to secrete IL-10, suggesting that they may be regarded as “inflammatory” Th2 cells (6).

However, there are at least two other options for the function of TSLP in controlling allergic inflammation. Naïve CD4 T cells express functional TSLP receptors and thus can transduce a STAT5 signal when stimulated with TSLP (5). In vitro Th2 differentiation is very dependent on STAT5 activation (10). TSLP should be capable of replacing IL-2 in such priming and thus production of TSLP in the vicinity of antigen-priming of naïve CD4 T cells could provide the STAT5 signal needed for Th2 differentiation. However, such a scenario would require a reliable source of TSLP at or near the site of priming. Naïve CD4 T cell priming occurs in the paracortex of the lymph node or in the comparable region of the spleen where little or no TSLP would be expected to be present except possibly, as reported by Sokol and Medzhitov (15), as a result of the transient recruitment of basophils to the lymph node in papain immunization and the stimulation of these cells to produce TSLP. However, IL-3 cultured basophils stimulated with PMA and ionomycin expressed limited amounts of TSLP and ZsG mRNA and we failed to detect ZsG expression by stimulated bone marrow derived basophils.

An alternate possibility is that memory/effector Th2 cells would be attracted to tissues into which antigen has been introduced. If TSLP had been induced at such sites, such as might be the case in atopic dermatitis (38), then such TSLP could act on differentiated Th2 cells, possibly together with IL-33, to drive IL-13 and IL-5 production and to enhance cell proliferation (11).

An even more intriguing possibility is that antigen-stimulation of Th2 cells in the tissues, particularly Th2 cells that produce TNFα, as has been reported to be the case for CD4 cells induced by TSLP-treated DC (6), would induce epithelial cells to produce TSLP. Indeed, we observed here that TNFα, IL-4 and IL-13 induce keratinocytes from reporter mice to become ZsG+ and others have previously reported that these cytokines can cause TSLP production by keratinocytes (39-41). We also showed that antigen-mediated activation of Th2 or Th1 cells in the skin resulted in robust induction of ZsG in the keratinocytes, indicating that in situ antigen-driven CD4 T cells activation drives TSLP induction, presumably through production of TNFα, IL-4 or IL-13 production. The mutual stimulatory properties of keratinocyte products (TSLP and IL-33) and Th2 (or possibly Th1) cell products (IL-13 in particular) could result in a chronic antigen-independent Th2-mediated allergic inflammatory response. A similar mechanism might hold for ILC-2 cells. These cells have been reported to show enhanced IL-13 production when stimulated with IL-33 and TSLP and the IL-13 they produce could contribute to a chronic allergic inflammatory response (12).

We were surprised by our failure to observe ZsG+ hematopoietic cells in skin or ear as well as in IL-3-driven bone marrow cell cultures and in GM-CSF induced dendritic cells. Indeed, Kashyap et al. (42) have recently reported that lung dendritic cells from mice challenged with huse dust mites expressed TSLP mRNA. The failure of IL-3-driven mast cells and basophils and of GM-CSF-induced DC to be ZsG+ on flow cytometry or confocal microscopy may reflect the presence of relatively small amounts of TSLP and ZsG in these cells compared to that in stimulated keratinocytes. However, in view of our failure to detect ZsG in stimulated basophils and/ or mast cells, one may ask whether these cells are physiologically important sources of TSLP. These cells could have a different role in TSLP expression, such as through the release of proteases or proinflammatory cytokines that induce TSLP production by epithelial/stromal cells.