Loss of a single Zn finger, but not that of two Zn fingers, of GATA3 drives skin inflammation

2.1 Deletion of the Zn finger close to the C terminus of GATA3 leads to the suppression of type II immune response

Disruption of GATA3 activity in T cells has been induced by the CD4Cre-mediated deletion of 2 Zn fingers (D2GATA3). The Zn finger close to C terminus and adjacent basic region are essential for the DNA binding of GATA3, suggesting an essential role of this Zn finger for various functions of GATA3 (Ho et al., 2009; Nakayama et al., 2017). We have established the mouse line (D1GATA3) carrying C-terminal truncation resulting in the loss of only the C-terminal Zn finger. Both D2GATA3 and D1DATA3 mice showed impairment of CD4SP T cell development in the thymus leading to a significantly reduced number of CD4 T cells in the periphery (Figure 1). In addition, CD4⁺ CD8⁺ double positive population was detected in the periphery of both lines which was not reported by other groups.

Western blot of CD8 T cell extracts was performed to elucidate whether the mutated proteins were expressed. As shown in Figure 2, a truncated protein with a single Zn finger deletion (D1GATA3) and that with both Zn fingers deleted (D2GATA3) were detected. The expression levels of those mutated proteins were comparable to or slightly elevated to that of wild-type GATA3 protein. Those truncated proteins might have partial activity or aberrant activity. Using D2GATA3 mice, other groups reported that CD8 T cells of D2GATA3 mice are impaired in immune responses including responsiveness to Ag stimulation and the peripheral maintenance of CD8 T cells (Tai et al., 2013; Wang et al., 2013).

Topical application of vitamin D analog MC903 induces TSLP production from keratinocytes in mice. TSLP affects various immune cells in the skin to drive type II immune responses such as elevated production of IgE and infiltration of eosinophiles; therefore, it is considered a mouse atopic dermatitis (AD) model. TSLP is highly expressed in the human skin of AD patients (Ebina-Shibuya & Leonard, 2023). MC903 application to D1GATA3 mice induced skin inflammation. The serum IgE level was upregulated in MC903-administered control mice, as expected, while IgE was significantly suppressed in MC903-administered D1GATA3 mice. The amount of IgE of D1GATA3 was not affected by the administration of MC903 and was close to the basal level in the control mice (Figure 3a). Some of the p-values exceeded 0.05 (0.12) due to variability of the data in WT mice. Therefore, we cannot use the term “significant difference.” However, we believe that the data generally support our conclusions.

Eosinophil infiltration into the skin was induced in MC903-administered control mice. The eosinophil infiltration was significantly suppressed in D1GATA3 mice treated with MC903 (Figure 3b, c). Therefore, type 2 inflammatory response was inhibited in D1GATA3 mice. The p-value of the data between MC903 treated and control mice exceeded 0.05 (0.076) due to variability of the data. Therefore, we cannot use the term “significant difference.” However, we believe that the data generally support our conclusions.

2.2 D1GATA3 mice developed dermatitis

It has been reported that D2GATA3 mice develop age-dependent transient lymphadenopathy due to the expansion of CD8 T cells. Besides that, D2GATA3 mice do not develop any inflammatory diseases (Tai et al., 2013). D1GATA3 mice developed dermatitis. The onset of the dermatitis is mainly 3–6 months after birth (Figure 4a). Some mice develop the disease much later. As shown in Figure 4b, edema, erosion, and alopecia appear in the face, ears, and paws. H&E-stained section from the limb of D1GATA3 mice showed hypertrophic stratum corneum with lymphocytic infiltration in the dermis and subcutis (Figure 4c).

2.3 Innate-like T cells and Tregs in D1GATA3 and D2GATA3 mice

The percentage of the NKT cells of D1GATA3 mice was comparable to that of wild-type mice (Figure 5a). It has been reported that the NKT cell development is impaired in GATA3 deficient mice equivalent to D2GATA3 mice (Kim et al., 2006). It may suggest that D1GATA3 mutant may have some compensatory pathways for the NKT cell development.

The number of γδT cells was elevated in the spleen while there was no difference in the skin (Figure 5b). The branching point of αβ and γδT cell lineage development is in the DN stage in the thymus, therefore, the GATA3 disruption by the CD4-Cre allele does not affect the γδT cell development directly. The increase of γδT cells in the spleen of D1GATA3 mice might be caused by the inflammatory state of D1GATA3 mice.

The numbers of CD4⁺ T cells and those of Tregs in D1GATA3 and D2GATA3 mice were reduced compared to those of WT mice (Figure 5c). It has been reported that the Treg-specific disruption of GATA3 by the deletion of both Zn fingers impairs the Treg function leading to the inflammation of lung, pancreas, lacrimal glands, and salivary glands (Wang et al., 2011). Those organs were not affected in either D1GATA3 or D2GATA3 mice (data not shown). We performed CD4 T cell depletion by injecting anti-CD4 Ab into D1GATA3 mice. The skin inflammation was not induced by CD4 T cell depletion. Although the involvement of CD4 helper T cells for the induction of dermatitis cannot be eliminated, depletion of Tregs may not cause the inflammation of skin tissue.

2.4 Transcriptional profiling of CD8 T cells and CD4 CD8 double-positive T cells in spleen of D1GATA3 mice shows features of T cell exhaustion

To elucidate the difference in gene expression profiles of D1GATA3 and D2GATA3 mice, spleen T cells were isolated from the mice of 8–9 weeks of age with no dermatitis observed, and the transcriptional profiling was performed by RNA-seq. RNA was extracted from CD4, CD8, and DP T cells of D1GATA3, D2GATA3, and control mice. To figure out the differentially expressed genes (DEGs), the CD8 and DP T cells of the GATA3 mutated mice and the control mice were compared. Expression of effector genes of cytotoxic T cells such as various granzymes, perforin, and IFNγ was elevated in DP T cells of D1GATA3 and D2GATA3 mice compared to that in WT CD8 T cells, while genes expressed highly in CD4 T cells such as CD40L and ThPOK were downregulated. Thus, we consider that the gene expression profile of DP T cells is close to that of CD8 T cells and the DEGs between DP T cells and the control CD8 T cells was determined. The expression profiles of CD8 T cells and CD4 CD8 DP T cells of the D1GATA3 mice were compared with those of the CD8 T cells of WT mice. A total of 319 upregulated and 74 downregulated DEGs from CD8 T cells and 2111 upregulated and 465 downregulated DEGs from DP T cells were identified (Supplementary Table S1). Gene ontology (GO) biological process analysis of upregulated genes revealed that CD8 T cells of D1GATA3 mice include pathways related to immune response such as “inflammatory response” (p-value −13.1), “regulation of inflammatory response”(p-value −9.68) and “leukocyte migration” (p-value −8.67). GO biological process analysis of DP cells of D1GATA3 mice includes pathways related to cell cycle regulation such as “mitotic cell cycle process” (p-value −19.7) and pathways related to immune response such as “cytokine-cytokine receptor interaction” (p-value −10.1).

Unsupervised hierarchical clustering of the DEGs revealed the close relatedness of the transcriptional profiles of CD8 T cells of D1GATA3 and D2GATA3 mice as well as DP T cells of D1GATA3 and D2GATA3 mice (Figure 6a, b). However, PCA analysis indicated that the transcriptional profiles of those populations of D1GATA3 and D2GATA3 mice are related but distinct (Figure 6c).

In D2GATA3 mice, 310 upregulated and 69 downregulated DEGs from CD8 T cells, and 1359 upregulated and 805 downregulated DEGs from DP T cells were identified (Supplementary Table S1). Upregulated genes of CD8T cells of D2GATA3 mice include pathways related to immune response such as “inflammatory response” (p-value −12.1), “regulation of leukocyte migration”(p-value −8.8), and “innate immune response” (p-value −8.3). In DP T cells of D2GATA3 mice, pathways related to cell cycle regulation such as “mitotic cell cycle” (p-value −54.3) and immune response regulation such as “inflammatory response” (p-value −14.1) and “cytokine-cytokine receptor interaction” (p-value −12.4) were identified as being upregulated.

Figure 6d shows the heat map illustrating the relative expression of genes that are associated with the terminally exhausted T cells. Those include inhibitory receptors Lilrb4a, CD244, Ctla4, Pdcd1, Cd160, Lag3, Havcr2, Cd38, and Tigit as well as NK receptors, Klrc1 and Klrg1. Transcription factors upregulated in exhausted T cells include Prdm1, Eomes, Maf, Nr4a2, Crem, Tox, Tox2, Ikzf2, Egr2 Zeb2, Irf4, and Bhlhe40.

Those genes associated with the exhausted state are upregulated in CD8 T cells and DP T cells of D1GATA3 as well as D2GATA3 mice. Tox, Tox2, and Nr4a transcription factors have been reported to form a transcriptional network and to facilitate differentiation of exhausted T cells. Moreover, the exhausted cell population contains the exhausted progenitor population capable of self-renewal and required for the maintenance of the terminally exhausted cell population. The progenitor population expresses Tcf7 (Tcf1) and IL7 receptors. The expression of Tcf7 as well as IL7 receptor was downregulated in CD8 T cells and DP T cells, supporting the idea that most of the CD8 T cells and DP T cells are in the terminally exhausted state.

The upregulated genes such as inhibitory receptors of exhausted state showed more elevated expression in the DP population than the expression in the CD8 T cell population. The expression of the downregulated genes such as Il4r, Il7r, Ccr7, Sell, and Tcf7 of the exhausted state is clearly suppressed in the DP population, while in CD8 T cells, the extent of suppression was small. This might indicate that the CD8 T cell population contains cells other than the exhausted cells.

2.5 Expression of the antimicrobial peptides S100a8/S100a9 is significantly higher in CD8 T cells with D1GATA3 than D2GATA3

Antimicrobial proteins S100a8 and S100a9 are expressed abundantly in myeloid cells. In neutrophils, more than 50% of cytoplasmic proteins consist of S100a8/S100a9 complex (Striz & Trebichavsky, 2004). Monocytes, macrophages, and non-hematopoietic cells such as keratinocytes also express S100a8/S100a9. S100a8/a9 complex functions as one of danger-associated molecular patterns to promote inflammation. Toll-like receptor 4 (TLR4) and receptor for advanced glycation end products (RAGE) have been shown to act as the receptors for S100a8/S100a9 (Narumi et al., 2015; Vogl et al., 2007). Elevated expression of S100a8/a9 is associated with various inflammatory conditions such as sepsis, asthma, psoriasis, and inflammatory bowel diseases (Aoki et al., 2009; Benoit et al., 2006; Gray et al., 2008; Palmer et al., 2019; Wang et al., 2023; Wang & Zhang, 2023). S100a8/S100a9 exerts both anti-inflammatory effects as well as pro-inflammatory effects, especially in sepsis, uncontrolled systemic inflammatory response to pathogens (Wang et al., 2023).

The transcriptional profiles of the CD8 T cells and DP T cells of D1GATA3 mice were compared directly to those of D2GATA3 mice (Supplementary Table S2). The expression of S100a8/S100a9 was significantly higher in the CD8 T cells of D1GATA3 mice than that of D2GATA3 mice. S100a9, but not S100a8 was also detected as a DEG in DP T cells. The expression of S100a8/S100a9 is enhanced in psoriasis patients as well as in the mouse model of psoriasis (Benoit et al., 2006; Defrene et al., 2021; Xie et al., 2023). Therefore, the elevated expression of S100a8/S100a9 in D1GATA3 mice might be the cause of dermatitis development.

IL17A and IL17F have been reported as critical inducer cytokines for S100a8 and S100a9 in keratinocytes. RNA-seq analyses showed that the expression of IL17A in D1GATA3 CD8 T cells is higher than that in D2GATA3 CD8 T cells, although its expression level was low (Supplementary Table S3). Furthermore, the expression of IL17A in DP T cells of D1GATA3 mice was much higher than that of D2GATA3 mice in the same experiment (Supplementary Table S3). qPCR analysis of splenic CD8 T cells stimulated with anti-CD3ε Ab and anti-CD28 Ab indicated that D1GATA3 CD8 T cells produced more IL-17A mRNA than D2GATA3 CD8T cells did (Figure 7a). The p-value between the WT and D1GATA was 0.09. However, the p-value between D1GATA and D2GATA is less than 0.05, and our results are in support of the conclusion that IL-17a is induced in D1GATA3 mice. These suggest a possibility that the elevated production of IL17a in D1GATA3 mice may lead to increased A100a8/9 in keratinocytes and to dermatitis development.

In addition, the expression of granzyme A, one of the effector molecules, was higher in D1GATA3 mice than that of D2GATA3 mice. Granzyme A has been shown to play important roles in inflammation and cell death induction (Lieberman, 2010; Rawle et al., 2022). Granzyme A is another candidate gene that might drive dermatitis in D1GATA3 mice.

IFNγ was detected as DEGs in DP T cells of D1GATA3 and D2GATA3 mice (Supplementary Tables S1 and S3). The expression level was higher in D2GATA3 mice than in D1GATA3 mice. qPCR analysis of the isolated CD8 T cells was performed (Figure 7b). The expression of IFNγ of stimulated CD8 T cells was much higher in D2GATA3 mice than in D1GATA3 mice. This suggests that IFNγ may not be the critical cytokine to drive the dermatitis in D1GATA3 mice. It may also indicate that D1GATA3 mutation that retains one Zn finger has some residual activity to suppress the transcription of the IFNγ gene.

2.6 MHC class II deficiency of GATA3 mutated mice results in the exacerbation of skin inflammation

As shown in Figure 8a, D1GATA3 mice with the MHC class II deficient background (MHC class II^def) show severe alopecia of the back skin, graying hair, and erosion of ears. The onset of the inflammation is about 3 weeks with 100% penetrance. Although D2GATA3 mice do not develop dermatitis, D2GATA3 MHC class II^def mice suffered from similar severe dermatitis (Figure 8b). Histological analysis of the ear of the D1GATA3 MHC class II^def mice showed the structural breakdown of the ear tissue, thickening of the dermis, and infiltration of lymphocytes (Figure 8c). A cluster of lymphocytes was found in the sebaceous and adipose tissues around the elastic cartilage. Lymphocytic infiltrates were found in collagenous tissue, nerve tissue, and rhabdomyocytes in the underlying layer of thickened squamous epithelium of the tail (Figure 8d). CD4 T cells were almost undetectable in the periphery of both D1GATA3 and D2GATA3 MHC class II^def mice, while the number of CD8 T cells was comparable to those of D1GATA3 and D2GATA3 MHC class II sufficient mice (Figure 9). Therefore, it is inconceivable that the GATA3 mutated CD4 T cells cause the dermatitis.

RNA-seq analyses of CD8 and DP T cells of D1GATA3 MHC class II^def mice were performed. In the absence of MHC class II molecule, the numbers of DEGs of CD8 and DP T cells from D1GATA3 and D2GATA3 mice were comparable (Supplementary Table S4). GO analyses of those T cells are shown in Figure 10a–d. In all those T cell populations, pathways related to cell cycle regulation and cell proliferation are significantly enriched. Pathways related to immune regulation such as “inflammatory response” and “lymphocyte activation” appeared next to the pathways of cell cycle regulation.

Resting naive CD8 T cells are activated by Ag stimulation to become effector CD8 T cells that proliferate extensively and produce various effector molecules such as various cytokines, chemokines, chemokine receptors, caspases, perforin, and granzymes. Gene signature of effector T cells includes genes involved in cell cycle regulation such as Cdc45, Cdk1, Cdc6, Ccna2, and Chk1. The other enriched group of effector pathways includes effector molecules of activated T cells such as Gzmb, Gzma, Ccr5, Ccr2, IFNγ, Cxcr3, and Casp1.

Figure 11 shows the heat map illustrating the relative expression of genes that are associated with the effector function of T cells. CD8T cells and DP T cells of both D1GATA3 MHC class II^def and D2GATA3 MHC class II^def mice express genes induced in effector T cells including cell cycle regulatory molecules and effector molecules. Interestingly S100a8/S100a9 are extremely highly expressed in D1GATA3 MHC class II^def mice, while S100 proteins of other members such as S100a4 and S100a6 were highly expressed in D2GATA3 MHC class II^def mice (Supplementary Table S4). Severe dermatitis might be developed by those S100 proteins. In addition, many members of granzymes were also upregulated in those cells suggesting that those molecules might be the cause of the severe dermatitis.

Since the onset of the dermatitis of the mice with the MHC class II deficiency was very early, the T cells were isolated from the mice that had shown the symptoms of dermatitis. There is a possibility that the transcriptional profile of T cells might be affected by the inflammation of the skin. Indeed, the increase of cDC and granulocytes in the spleen was observed (data not shown). Therefore, the transcriptional profile of T cells might represent the combined effect of GATA3 mutation and those myeloid lineage cells.

2.7 MHC class II negatively regulates CD8 T cells through MHC class II-Lag3 axis

An important role of MHC class II molecule in the immune system is to present Ag to CD4 positive helper and regulatory T cells. Another role is to suppress Lag3 expressing cells through MHC class II—Lag3 interaction. To elucidate the role of Lag3 of GATA3 mutated T cells, administration of the blocking Ab against Lag3 into D1GATA3 mice was performed (Figure 12). To evaluate the state of activation, the expression of granzyme B was analyzed by flow cytometry. D1GATA3 mice were administered either 200 μg of anti-LAG3 Ab or the control Ab every 3 days for 2 weeks. At 4 h before the isolation of T cells, 10 μg of LPS was injected to stimulate T cells in vivo to induce granzyme B expression on the cell surface.

The blocking of the MHC class II—Lag3 axis by anti-LAG3 Ab revealed that for D1GATA3 CD8 T cells, the negative signal through Lag3 is effective in inducing granzyme B expression. This indicates that the Lag3 signal induced exhausted state to avoid extensive activation and the induction of inflammation. However, we did not observe the induction of skin inflammation in D1GATA3 mice upon Lag3 Ab injection. This could be because the period of Ab injection might not be long enough to induce inflammation. Alternatively, the inhibitory activity of the Ab may not be effective enough to induce dermatitis.

4.1 Mice

The permission to use D2GATA3 mice (gata3 gene Exon4 is floxed) was obtained from Dr. Zhu at NIAID and the mouse was distributed from Dr. Kubo at Riken. The permission to use CD4Cre mouse was obtained from Dr. Willson at The University of Washington and the mice were distributed from Dr. Kubo at Riken. The permission to use I-Aβ deficient mouse (MHC class II deficient mouse) was obtained from Dr. Mathis and distributed from Dr. Ikuta at Kyoto University. D1GATA3 mouse (floxed mouse to delete one Zn finger close to C terminus) was produced at Tokyo Metropolitan Institute of Medical Science.

All mice were maintained in specific-pathogen-free barrier facilities and used according to protocols approved by the Tokyo Metropolitan Institute of Medical Science animal care and use committees.

4.2 Topical application of MC903 to induce skin inflammation

Mice were anesthetized using intraperitoneal injection of a combination of medetomidine, midazolam, and butorphanol. MC903 (calcipotriol; Tocris Bioscience) was dissolved in 100% ethanol and topically applied on mouse ears (1 nmol in 20 μL per ear). As vehicle control, the same volume of ethanol was applied. MC903 was applied to the ear every 2 days for 2 weeks.

4.3 Elisa

Mouse sera were collected, and IgE measurement was performed by Enzyme-linked immunosorbent assay according to the manufacturer's recommendations (Biolegend, San Diego, CA).

4.4 Cell staining for flow cytometry

The following antibodies were purchased from BioLegend.

CD3ε FITC (500A2), CD8a BV421 (53-6.7), CD4 PE (GK1.5), granzyme B APC (QA18A28), γδTCR PE(UC7-13D5), NK1.1 PerCP-Cy5.5 (S17016D), foxp3 PE (MF-14).

Dead cells were discriminated by staining with Zombie aqua dye (BioLegend, San Diego CA). Intracellular staining of granzyme B was performed using the Cyto-Fast™ Fix/Perm Buffer Set (Biolegend, San Diego, CA) in accordance with the manufacturer's protocol. Flow cytometry data were acquired on a BD LSR II instrument. Data were analyzed using FlowJo software (BD).

4.5 qPCR of IL-17a and IFNγ

RNA purification and quantitative PCR. Total RNAs were isolated using TRIzol (Invitrogen). cDNAs were prepared using ReverTra Ace qPCR RT Master Mix (Toyobo). Quantitative PCR was performed on ABI7500 Real-Time PCR System.

4.6 Ab injection

A total of 200 μg of anti-LAG3 Ab (C9B7W, BioXCell, Lebanon NH) or a control rat IgG1 (BioXCell, Lebanon NH) were injected ip every 3 days for 2 weeks. A total of 10 μg LPS was injected and 4 h later, mice were sacrificed, and spleen T cells were analyzed by flow cytometry.

4.7 Statistic

Student t-test was used for statistical analysis. *p <.05, **p <.01. In some cases, the calculated p-values were presented in the figures.

4.8 RNA-seq sample preparation

NEBNext® Ultra™II RNA Library Prep Kit for lllumina (NEB) was used for the library construction. Messenger RNA was purified from total RNA using poly-T oligo-attached magnetic beads. After fragmentation, the first strand cDNA was synthesized using random hexamer primers, followed by the second strand cDNA synthesis using dTTP for non-directional library. For the non-directional library, it was ready after end repair, A-tailing, adapter ligation, size selection, amplification, and purification. Quantified libraries will be pooled and sequenced on NovaSeq 6000.

4.9 RNA-seq data analysis

The sequencing reads were mapped to the mm10 genome using STAR (Dobin et al., 2013). The number of reads falling into each gene was quantified using RSEM (Li & Dewey, 2011). The iDEP.96 software was used to detect DEGs (padj<0.01) between any pair of biological conditions (Ge et al., 2018).