Humanized NOD/SCID/IL2rγ^null mice exhibit functionally augmented human regulatory T cells associated with enzymatic up‐regulation of H3K27me3 in comparison with humans

Summary

Humanized non‐obese diabetic/severe combined immunodeficiency/interleukin‐2 receptor‐γ‐null (NOD/SCID/IL2rγ^null) [humanized (huNSG)] mice engrafted with human hematopoietic cells have been used for investigations of the human immune system. However, the epigenetic features of the human regulatory T (T_reg) cells of huNSG mice have not been studied. The objective of this study was to clarify the characteristics of human T_reg cells in huNSG mice, especially in terms of the epigenetic aspects. We compared the populations, inhibitory molecule expression and suppressive capacity of human T_reg cells in spleens harvested from the huNSG mice 120 days after the engraftment of human umbilical cord blood CD34⁺ cells with human peripheral blood mononuclear cells (PBMCs). Histone modifications and enhancer of zeste homolog 2 (Ezh2), an H3K27 methyltransferase, of human T_reg cells were quantified in huNSG mice and human PBMCs. The effect of Ezh2 inhibitor on human T_reg cells exposed to interleukin (IL)‐6 was also compared between them. Human T_reg cells in the spleens of huNSG mice showed an increased proportion among CD4⁺ T cells, higher expressions of forkhead box protein 3 (FoxP3), cytotoxic T lymphocyte antigen 4 (CTLA‐4) and glucocorticoid‐induced tumor necrosis factor‐related protein (GITR), a higher production of IL‐10 and enhanced suppressive capacity when compared with those in human PBMCs. H3K27me3 and Ezh2 were specifically up‐regulated in human T_reg cells of huNSG mice in comparison with those of human PBMCs. The decrease in T_reg cells induced by IL‐6 exposure was attenuated in huNSG mice when compared with human PBMCs, while the difference between them was cancelled by addition of Ezh2 inhibitor. In conclusion, huNSG mice exhibit functionally augmented human T_reg cells owing to enzymatic up‐regulation of H3K27me3.

Humanized NOD/SCID/IL2rγnull (huNSG) mice engrafted with human hematopoietic cells exhibited functionally augmented human Treg cells. The higher Treg cell stability in huNSG mice depended on enhanced enhancer of zeste homolog 2 (Ezh2)‐mediated trimethylated histone H3 at lysine 27 (H3K27me3) modification in Treg cells. This is the first report to show the epigenetic features of human Treg cells in huNSG mice.

Introduction

Humanized mice are immunodeficient mice engrafted with human CD34⁺ hematopoietic cells; these mice provide a powerful tool for in‐vivo investigations of the human immune system [1, 2]. The utility of humanized mice has been widely indicated in a variety of current immunological challenges, such as human–tropic infections, autoimmunity and cancer, for elucidation of the pathological mechanisms and development of new therapeutic strategies [3].

Among several immunodeficient mice used for immune and hematopoietic reconstitution, NOD.Cg‐Prkdc^scid IL2rg^tmWjl/Sz [non‐obese diabetic/severe combined immunodeficiency/interleukin (IL)‐2 receptor‐γ‐null (NOD/SCID/IL2rγ^null; NSG] mice have been reported to lack T, B, and natural killer (NK) cells owing to a deficient common cytokine receptor γ‐chain, resulting in defects in the cytokine signaling essential for the differentiation to these cells, such as IL‐2, IL‐7 and IL‐15 signaling [4, 5]. NSG mice demonstrate more efficient engraftment of human umbilical cord blood CD34⁺ hematopoietic cells and multi‐lineage differentiation in comparison with other previously described immunodeficient mouse strains.

In terms of T cells, transplant of human CD34⁺ hematopoietic cells into NSG mice resulted in the development of human CD4⁺ and CD8⁺ T cell receptor (TCR)αβ⁺ T cells and CD4⁻CD8⁻ and CD8⁺TCRγδ⁺ T cells in recipient bone marrow and spleens [6]. These mice also exhibited the phenotypical transition from naive to effector memory T cells and human T helper type 17 (Th17) cell development, together with Th1 and Th2 cell development. Moreover, the diversity of the TCR‐β chain repertoire in humanized NSG (huNSG) mice reached 100% of the human reference samples [7, 8]. Thus, like humans, huNSG mice exhibit a variety of human T cell subsets and diversity of the TCR repertoire.

Among a variety of human T cell subsets, regulatory T (T_reg) cells expressing the transcription factor forkhead box protein P3 (FoxP3) have a critical role in the maintenance of immune homeostasis and prevention of autoimmunity [9]. Regarding cancer immunity, it is evoked effectively by suppression of T_reg cells, which has been established as a novel immunotherapeutic strategy against cancer by a variety of so‐called immune checkpoint inhibitors [10]. Recently, many studies of T_reg cells have focused upon the epigenetic modifications that play a key role in the stability of the T_reg cell lineage. Among the epigenetic factors involved in T_reg cells, histone modifications that regulate chromatin accessibility could play an essential role in T_reg cell development and function via stability of FoxP3 protein expression. For example, trimethylated histone H3 at lysine 4 (H3K4me3) induced by SET and MYND domain 3, an H3K4 histone methyltransferase, is up‐regulated in T_reg cells and positively regulates FoxP3 expression via its deposition at the FoxP3 gene locus under inducible T_reg‐skewing conditions [11]. Moreover, trimethylated histone H3 at lysine 27 (H3K27me3) induced by enhancer of zeste homolog 2 (Ezh2), an H3K27 methyltransferase of the polycomb repressor complex 2 that controls chromatin condensation, is also up‐regulated upon T_reg cell activation to maintain T_reg cell stability via indirect transcriptional regulation of FoxP3 [12, 13].

However, only a limited amount of evidence is available related to the development of human T_reg cells in huNSG mice [14, 15]. Moreover, the epigenetic modifications of human T_reg cells in huNSG mice have not been studied. Comprehensive understanding of the human immune system requires deep knowledge of T_reg cell biology. Therefore, to establish huNSG mice as a complete humanized immune system murine model, further analysis of human T_reg cells in these mice in comparison with real human references is needed.

Here, we performed analysis, especially in the epigenomic aspects, of human T_reg cells developed in huNSG mice engrafted with umbilical cord blood CD34⁺ cells in comparison with those from human references to examine the adequacy of this model as a humanized immune system murine model.

This is the first report, to our knowledge, to show the role of epigenetic modification in human T_reg cells of huNSG mice.

Materials and methods

Ethics statement

All research with human samples was performed in compliance with our institutional guidelines and the Declaration of Helsinki. Approval for this study was obtained from our local ethics committee (Tsukuba Clinical Research and Development Organization) (approval number: H24‐164). An informed consent was obtained from each subject prior to the inclusion in this study.

All experiments associated with mice were performed according to the Guide for the Care and Use of Laboratory Animals at Tsukuba University and approved by our local ethics committee (Animal Experiment Committee) (approval no.: 20‐186).

Human umbilical cord blood CD34⁺ cells

Purified human umbilical cord blood CD34⁺ cells obtained from five individuals with regular births were provided by the Cell Engineering Division of Riken BioResource Center (Tsukuba, Ibaraki, Japan).

Mice

NSG mice were obtained from the Jackson Laboratory and bred at the animal facility of the University of Tsukuba in specific pathogen‐free conditions. Four to 5‐week‐old male mice were sublethally irradiated with 1·0 Gy. Twenty‐four h after the irradiation, 1·0 × 10⁵ human umbilical cord blood CD34⁺ cells were transferred into the mice via tail vein injection.

Tissue collection from mice

One hundred and 20 days after injection with human umbilical cord blood CD34⁺ cells, the NSG mice were euthanized by use of isoflurane, and the spleen and thymus were harvested and then disrupted by grinding between frosted glass coverslips, followed by single‐cell suspensions in phosphate‐buffered saline. The spleen cells were hemolysed with 0·16 M of ammonium chloride at room temperature for 5 min.

Human peripheral blood mononuclear cells (PBMCs)

PBMCs were isolated from heparinized peripheral venous blood obtained from five human healthy volunteers by use of the Ficoll‐Hypaque gradient (GE Healthcare, Chicago, IL, USA) and washed with phosphate‐buffered saline.

Human CD4⁺ T cell isolation

Human CD4⁺ T cells were isolated from human PBMCs and spleen cells of huNSG mice, which were defined by NSG mice engrafted with human umbilical cord blood CD34⁺ cells by positive selection using an autoMACS Pro cell separator (Miltenyi Biotec, Bergisch Gladbach, Germany) after magnetic labeling of the targeted cells with anti‐human CD4 MicroBeads (Miltenyi Biotec), according to the manufacturer’s protocol. The purity of the CD4⁺ T cell populations was > 97%.

Flow cytometry

Staining for human CD3, CD4, CD8, CD25, CD27, CD45, CD127, Fc receptor‐like 3 (FcRL3), cytotoxic T lymphocyte antigen 4 (CTLA‐4), glucocorticoid‐induced tumor necrosis factor (TNF)‐related protein (GITR), murine mCD45 molecules (all from BioLegend, San Diego, CA, USA) and T cell immunoreceptor with immunoglobulin (Ig) and immunoreceptor tyrosine‐based inhibitory motif domains (TIGIT) (eBioscience, San Diego, CA, USA) was performed for 20 min using a mixture of antibodies. Intracellular staining for human FoxP3, Helios (both from BioLegend) and Ezh2 [Becton, Dickinson and Company (BD), Franklin Lakes, NJ, USA] was performed after fixation and permeabilization using a FoxP3/transcription factor staining buffer set (eBioscience), according to the protocol supplied by the manufacturer. For intracellular staining of IFN‐γ and IL‐10 (BioLegend) following stimulation, phorbol myristate acetate (50 ng/ml), ionomycin (0·5 μg/ml) and GolgiStop (eBioscience) were added during the last 6 h of each culture of CD4⁺ T cells. For intracellular staining of histone modification, after incubation with the primary antibody against H3K4me3 and H3K27me3 [both are rabbit IgG and from Cell Signaling Technology (CST), Danver, MA, USA], Alexa Fluor 488^®‐conjugated anti‐rabbit IgG (CST) was used as the secondary antibody. The samples were analysed with a fluorescence activated cell sorter (FACS)Verse flow cytometer (BD), and the data were analysed with FlowJo software (Tree Star, Inc., Ashland, OR, USA). The relative mean fluorescence intensity (MFI) ratio was calculated on the basis of the ratio of the MFI for a marker to the MFI of its isotype control.

In‐vitro suppression assay and cell culture

CD4⁺CD25⁻ and CD4⁺CD25⁻ cells were isolated using the SH800 device (Sony, Tokyo, Japan). CD4⁺CD25⁻ cells, defined as conventional T (T_conv) cells, were labeled with 10 μM CellTrace Violet by use of a CellTrace™ Cell Proliferation Kit (Invitrogen, Carlsbad, CA, USA) and then cultured with or without unlabeled CD4⁺CD25⁺ cells, defined as T_reg cells, at the indicated ratios for 96 h in the presence of Dynabeads human T‐activator CD3/CD28 (Invitrogen) in round‐bottomed 96‐well plates at a density of one bead per cell. The percentage inhibition rate on the proliferation of T_conv cells was calculated as [1‐(CellTrace Violet percentage of T_reg cells plus T_conv cells co‐culture/T_conv cells alone)] × 100%. For analysis of cytokine production in T_reg cells, the cells were collected after 72 h stimulation with Dynabeads human T‐activator CD3/CD28 for flow cytometry. For the inhibition assay of Ezh2, human CD4⁺ T cells obtained from human PBMCs and spleens of huNSG mice were cultured under stimulation with Dynabeads human T‐activator CD3/CD28 in the presence of human recombinant IL‐2 (BioLegend) (300 U/ml) with or without human recombinant IL‐6 (BioLegend) (25 ng/ml) or Ezh2 inhibitor (GSK343; Sigma‐Aldrich, St Louis, MO, USA); 5 μM) for 72 h.

Statistical analysis

All values are expressed as means ± standard deviations, unless otherwise stated. All tests were two‐sided, and results with probability values < 0·05 were considered significant. The Mann–Whitney test was used to compare two independent continuous variables. Wilcoxon’s signed‐rank test was used to compare the values of the means from two related samples.

Results

Human T cell subsets developed in huNSG mice and humans

Flow cytometric analysis revealed engraftment of human CD45⁺ cells and development of human CD4⁺ T cells and CD8⁺ T cells in spleens, as observed in the PBMCs of healthy human subjects and development of CD4⁺ single‐positive (SP) cells, CD8⁺ SP cells and CD4⁺CD8⁺ double‐positive (DP) cells in the thymuses of huNSG mice harvested on day 120 (Fig. 1a). The spleens of huNSG mice exhibited a significantly higher proportion of human CD4⁺ T cells and a lower proportion of human CD8⁺ T cells compared with the PBMCs of healthy human subjects (Fig. 1b).

Fig. 1.

Human T cell subsets and regulatory T cells develop in non‐obese diabetic/severe combined immunodeficiency/interleukin (IL)‐2 receptor‐γ‐null (NOD/SCID/IL2rγ^null) (NSG) mice and humans. (a) Representative flow cytometric data of human CD45⁺ cells and T cells in spleens and thymuses of humanized NSG (huNSG) mice harvested on day 120 and peripheral blood mononuclear cells (PBMCs) of healthy human subjects. (b) Population of human CD4⁺ and CD8⁺ T cells among CD3⁺ T cells in spleens of huNSG mice on day 120 (n = 5) and human PBMCs (n = 5). (c) Representative flow cytometric data of human CD25⁺CD127^lo cells gated on human CD45⁺CD3⁺CD4⁺ cells in spleens of huNSG mice harvested on day 120 and PBMCs of healthy human subjects (upper panel and lower panel, respectively) and those gated on human CD45⁺CD4⁺ cells (middle panel) in the thymuses of huNSG mice. (d) Representative flow cytometric data of human CD25⁺FoxP3⁺ cells gated on human CD45⁺CD3⁺CD4⁺ cells in spleens of huNSG mice harvested on day 120 and PBMCs of healthy human subjects. (e) Population of human CD25⁺FoxP3⁺ regulatory T (T_reg) cells gated on human CD45⁺CD3⁺CD4⁺ T cells in spleens of huNSG mice on day 120 (n = 5) and human PBMCs (n = 5). (f) Representative flow cytometric data of human T cell immunoreceptor with immunoglobulin (Ig) and TIGIT⁺FcFcRL3⁺ cells gated on human CD4⁺CD25⁺FoxP3⁺ T_reg cells in spleens of huNSG mice harvested on day 120 and PBMCs of healthy human subjects. (g) Representative flow cytometric data of Helios⁺ cells gated on human CD4⁺CD25⁺FoxP3⁺TIGIT⁺FcRL3⁺ T_reg cells in spleens of huNSG mice harvested on day 120 and PBMCs of healthy human subjects. (h) Population of human TIGIT⁺FcRL3⁺Helios⁺ cells gated on human CD4⁺CD25⁺FoxP3⁺ T_reg cells in spleens of huNSG mice on day 120 (n = 5) and human PBMCs (n = 5). (i) Population of human CD25⁺FoxP3⁺TIGIT⁺FcRL3⁺Helios⁺ cells gated on total human CD4⁺ T cells in spleens of huNSG mice on day 120 (n = 5) and human PBMCs (n = 5). (a,c,d,f,g) Data are representative of three independent experiments. (b,e,h,i) Mann–Whitney test. huNSG = humanized NSG mouse; 7‐AAD = 7‐amino‐actinomycin D; mCD45 = mouse CD45 = hCD45: human CD45; hCD3 = human CD3; hCD4 = human CD4; hCD8 = human CD8; hCD25 = human CD25; hFoxP3 = human forkhead box protein P3; hTIGIT = human T cell immunoreceptor with immunoglobulin and immunoreceptor tyrosine‐based inhibitory motif domains; hFcRL3 = human Fc receptor‐like 3. *P < 0.05; n.s. = not significant.

Human T_reg cells developed in huNSG mice and humans

Flow cytometric analysis revealed the fraction of human CD4⁺CD25⁺CD127^lo cells, indicating human T_reg cells by their cell surface markers in spleens and thymuses of huNSG mice harvested on day 120 as well as PBMCs of healthy human subjects (Fig. 1c). Another analysis showed the development of human CD4⁺CD25⁺FoxP3⁺ T_reg cells in spleens of huNSG mice harvested on day 120 as well as PBMCs of healthy human subjects (Fig. 1d). The proportion of human CD25⁺FoxP3⁺ T_reg cells among CD4⁺ T cells was higher in the spleens of huNSG mice than in the human PBMCs (Fig. 1e), although there were individual differences in the proportion among the umbilical cord blood donors in huNSG mice (Supporting information, Table S1, Fig. S1a). Although the proportion of human TIGIT⁺FcRL3⁺Helios⁺ cells among CD4⁺CD25⁺FoxP3⁺ T_reg cells in the spleens of huNSG mice was not significantly different from that in the human PBMCs (Fig. 1f–h), the proportion of human TIGIT⁺FcRL3⁺Helios⁺CD25⁺FoxP3⁺ T_reg cells in the total CD4⁺ T cell population was significantly higher in the spleens of huNSG mice than in the human PBMCs (Fig. 1i). The following experiments were conducted on the basis of these observations.

Expression of inhibitory molecules in human T_reg cells developed in humanized NSG mice and humans

The expressions of FoxP3, CTLA‐4 and GITR were significantly higher in human CD4⁺CD25⁺ T_reg cells from the spleens of huNSG mice than in those from human PBMCs (Fig. 2a,b). These expressions were also significantly higher in human CD4⁺CD25⁻ T_conv cells than in those from human PBMCs (Fig. 2a,b). Moreover, the population of IL‐10‐producing cells identified by IL‐10⁺IFN‐γ⁻ cells, as well as that of IFN‐γ‐producing cells identified by IL‐10⁻IFN‐γ⁺ cells after stimulation of human CD4⁺CD25⁺ cells with anti‐CD3/CD28 beads, was significantly higher in the spleens of huNSG mice than in human PBMCs (Fig. 2c–e).

Fig. 2.

Expression of inhibitory molecules in regulatory T cells of humanized non‐obese diabetic/severe combined immunodeficiency/interleukin (IL)‐2 receptor‐γ‐null (NOD/SCID/IL2rγ^null) (NSG) mice and humans. (a) Representative flow cytometric data of FoxP3, CTLA‐4 and GITR expression in human T_reg cells (red lines) and T_conv cells (blue lines) defined by hCD4⁺hCD25⁺ cells and hCD4⁺hCD25⁻ cells, respectively, from spleens of humanized NSG (huNSG) mice and human peripheral blood mononuclear cells (PBMCs). (b) Relative mean fluorescence intensity of FoxP3, CTLA‐4 and GITR in human T_reg cells from spleens of huNSG mice (n = 5) and those from human PBMCs (n = 5). (c) Representative flow cytometric data of interferon (IFN)‐γ and IL‐10 production in human T_reg cells from spleens of huNSG mice and human PBMCs after 72 h of stimulation with anti‐CD3/CD28 beads. (d) Percentage of IL‐10‐producing cells among human T_reg cells from spleens of huNSG mice (n = 3) and human PBMCs (n = 3) after 72 h of stimulation. (e) Percentage of IFN‐γ‐producing cells among human T_reg cells from spleens of huNSG mice (n = 3) and human PBMCs (n = 3) after 72 h of stimulation. (a,c) Data are representative of three independent experiments. (b) Mann–Whitney test and Wilcoxon’s signed‐rank test. (d,e) Mann–Whitney test. huNSG = humanized NSG mouse; FoxP3 = forkhead box protein P3; CTLA‐4 = cytotoxic T lymphocyte antigen 4; GITR = glucocorticoid‐induced tumor necrosis factor (TNF)‐related protein; T_reg = human regulatory T cells; T_conv = human conventional T cells; MFI = mean fluorescence intensity. *P < 0·05 by Mann–Whitney test; **P < 0·05 by Wilcoxon’s signed‐rank test; n.s. = not significant.

In‐vitro suppression assay comparing the suppressive function of human T_reg cells in humanized NSG mice and humans

The purities of human CD4⁺CD25⁺ T_reg cells and human CD4⁺CD25⁻ T_conv cells from spleens of huNSG mice or human PBMCs were 95·0% or above (Fig. 3a). Division of human CD4⁺CD25⁻ T_conv cells from spleens of huNSG mice (Fig. 3b) and human PBMCs (Fig. 3c) determined by dilution of CellTrace Violet was dose‐dependently suppressed by human CD4⁺CD25⁺ T_reg cells from the respective origin. The inhibition rate of human T_reg cells from huNSG mice was significantly higher than that from human PBMCs (Fig. 3d), although there were individual differences in the inhibition rate among the umbilical cord blood donors in huNSG mice (Supporting information, Table S1, Fig. S1b).

Fig. 3.

In‐vitro suppression assay comparing suppressive function of T_reg cells in humanized non‐obese diabetic/severe combined immunodeficiency/interleukin (IL)‐2 receptor‐γ‐null (NOD/SCID/IL2rγ^null) (NSG) mice and that in humans. (a) Representative flow cytometric data of human CD4⁺CD25⁺ (T_reg) cells and human CD4⁺CD25⁻ (T_conv) cells from spleens of huNSG mice or human peripheral blood mononuclear cells (PBMCs). (b) Proliferation of CellTrace Violet‐labeled human CD4⁺CD25⁻ (T_conv) cells from spleens of huNSG mice co‐cultured with or without human T_reg cells from spleens of huNSG mice upon 96‐h stimulation with anti‐CD3/CD28 beads. (c) Proliferation of CellTrace Violet‐labeled human T_conv cells from human PBMCs co‐cultured with or without human T_reg cells from human PBMCs upon 96‐h stimulation with anti‐CD3/CD28 beads. (d) The inhibition rates in (b) and (c). The inhibition rate of human T_reg cells from huNSG mice (n = 3) was compared with that of human PBMCs (n = 3). (b,c) Flow cytometric data are CellTrace Violet(⁺)‐gated and representative of three independent experiments. (d) Mann–Whitney test. T_conv = human conventional T cells; T_reg = human regulatory T cells; huNSG = humanized NSG mouse. *P < 0·05.

Histone modification of human T_reg cells in huNSG mice and humans

The histone modification, H3K27me3, was significantly enhanced in human T_reg cells from spleens of huNSG mice compared with those from human PBMCs (Fig. 4a,b), while another histone modification, H3K4me3, exhibited no difference between them (Fig. 4c,d). Accordingly, Ezh2 expression up‐regulated specifically in T_reg cells compared with T_conv cells was significantly higher in human T_reg cells from huNSG mice than in those from human PBMCs (Fig. 4e,f).

Fig. 4.

Histone modification of regulatory T cells in humanized non‐obese diabetic/severe combined immunodeficiency/interleukin (IL)‐2 receptor‐γ‐null (NOD/SCID/IL2rγ^null) (NSG) mice and humans. (a) Representative flow cytometric data of trimethylated H3K27me3 in human T_reg cells defined by hCD4⁺hCD25⁺ cells from spleens of huNSG mice and human peripheral blood mononuclear cells (PBMCs). (b) Relative MFI of H3K27me3 in human T_reg cells from spleens of huNSG mice (n = 5) and human PBMCs (n = 5). (c) Representative flow cytometric data of trimethylated histone H3 at lysine 4 (H3K4me3) in human T_reg cells from spleens of huNSG mice and human PBMCs. (d) Relative MFI of H3K4me3 in human T_reg cells from spleens of huNSG mice (n = 5) and human PBMCs (n = 5). (e) Representative flow cytometric data of Ezh2 in human T_reg cells and human T_conv cells defined by hCD4⁺hCD25⁻ cells from spleens and thymus of huNSG mice and human PBMCs. (f) Relative MFI of Ezh2 in human T_reg cells and T_conv cells from spleens of huNSG mice (n = 5) and human PBMCs (n = 5). (a,c,e) Data are representative of three independent experiments. (b,d) Mann–Whitney test. (f) Mann–Whitney test and Wilcoxon’s signed‐rank test. huNSG = humanized NSG mouse; H3K4me3 = trimethylated histone H3 at lysine 4; H3K27me3 = trimethylated histone H3 at lysine 27; Ezh2 = enhancer of zeste homolog 2; T_reg = human regulatory T cells; T_conv = human conventional T cells; MFI = mean fluorescence intensity. *P < 0·05 by Mann–Whitney test; **P<0·05 by Wilcoxon’s signed‐rank test; n.s. = not significant.

Stability of human T_reg cells in huNSG mice and humans

The huNSG mice exhibited a significantly higher population of FoxP3⁺ cells generated among CD4⁺Ezh2^hi cells compared with CD4⁺Ezh2^lo cells (Fig. 5a,b). The decrease in CD4⁺CD25⁺FoxP3⁺ T_reg cells under the inflammatory condition by exposure to IL‐6 in addition to IL‐2 was significantly attenuated in human CD4⁺ T cells from huNSG mice compared with those from human PBMCs (Fig. 5c,d). However, the difference between huNSG mice and human PBMCs was cancelled by the addition of Ezh2 inhibitor to IL‐6 (Fig. 5c,d).

Fig. 5.

Stability of regulatory T cells in humanized non‐obese diabetic/severe combined immunodeficiency/interleukin (IL)‐2 receptor‐γ‐null (NOD/SCID/IL2rγ^null) (NSG) mice and humans. (a) Gating strategy of Ezh2^lo cells and Ezh2^hi cells in hCD4⁺ T cells (left panel) and representative flow cytometric data of FoxP3 in Ezh2^hi cells (middle panel) and Ezh2^lo cells (right panel) from spleens of huNSG mice. (b) Population of FoxP3⁺ cells among Ezh2^hi cells and Ezh2^lo cells of hCD4⁺ T cells from spleens of huNSG mice (n = 5). (c) Representative flow cytometric data of hCD25⁺FoxP3⁺ cells gated on hCD4⁺ cells 72 h after exposure of hCD4⁺ T cells from human PBMCs and huNSG mice to IL‐2 (300 U/ml) ± IL‐6 (25 ng/ml) in the presence or absence of Ezh2 inhibitor (GSK343; 5 μM). (d) Decrease in hCD4⁺hCD25⁺FoxP3⁺ T_reg cells (%) from those after the exposure to IL‐2 and those after the exposure to IL‐2 + IL‐6 in the presence or absence of Ezh2 inhibitor (GSK343; 5 μM) were compared between human PBMCs (n = 5) and huNSG mice (n = 5). (a,c) Data are representative of three independent experiments. (b) Wilcoxon’s signed‐rank test. (d) Mann–Whitney test and Wilcoxon’s signed‐rank test. Ezh2 = enhancer of zeste homolog 2; hCD4 = human CD4; FoxP3 = forkhead box protein P3; huNSG = humanized NSG mouse; hCD25 = human CD25; T_reg = regulatory T cells; Ezh2‐I = Ezh2 inhibitor. *P < 0·05 by Mann–Whitney test; **P < 0·05 by Wilcoxon’s signed‐rank test; n.s. = not significant.

Discussion

This is the first report, to our knowledge, to show that huNSG mice established by reconstitution of human umbilical cord blood‐derived CD34⁺ cells exhibited functionally augmented human T_reg cells associated with higher stability that was obtained due to increased H3K27me3 via up‐regulation of Ezh2 in comparison with those in humans. Through our investigation, we found some critical observations concerning some aspects of the behavior of human T_reg cells induced in huNSG mice.

First, we found that huNSG mice exhibited an increased population of human CD4⁺ T cells and a higher proportion of human T_reg cells among CD4⁺ T cells in the spleens compared with human PBMCs. These observations suggested that the development of human CD4⁺CD25⁺FoxP3⁺ T_reg cells was quantitatively promoted in the primary lymphoid tissues of huNSG mice in comparison with those of adult humans. Even though the thymus of huNSG mice exhibited a lower population of human T_reg cells than did the spleen, the definite existence of human thymic T_reg cells indicated the same process of human T cell differentiation in the thymus of huNSG mice as in humans. It was also indicated by the thymic components, including human CD4⁺CD8⁺ DP cells. Moreover, human CD4⁺CD25⁺FoxP3⁺TIGIT⁺FcRL3⁺Helios⁺ T_reg cell fraction was reported to identify bona fide T_reg cells distinct from recently activated CD4⁺ T_conv cells [16]. In this regard, the bona fide T_reg cell population was also increased in huNSG mice in comparison with humans. However, the proportion of TIGIT⁺FcRL3⁺Helios⁺ cells among CD4⁺CD25⁺FoxP3⁺ T_reg cells was not different between them at baseline. In other words, there was no compositional difference in CD4⁺CD25⁺FoxP3⁺ T_reg cell population between them in this aspect. These observations could be fundamental to all the comparative analyses of the T_reg cells in our study.

Secondly, human T_reg cells of humanized NSG mice exhibited up‐regulated expression of T_reg cell‐associated inhibitory molecules such as FoxP3, CTLA‐4 and GITR and increased production of IL‐10, which is known as an essential inhibitory cytokine in T_reg cells in comparison with human PBMCs. Furthermore, although responder T cells of huNSG mice exhibited less division upon stimulation by human CD3/CD28 beads at baseline than those of humans, the inhibition rate of T_reg cells in huNSG mice, corrected by the division of responder T cells at baseline, was higher than that in human PBMCs, which was equivalent to previously reported findings under the similar condition [17]. This result suggested that human T_reg cells of huNSG mice exhibited an augmented suppressive function by themselves independently of the proliferative ability of responder T cells. Even considering that CD4⁺CD25⁺ T cell population in huNSG mice included a higher number of IFN‐γ‐producing cells, this population could have stronger suppressive capacity in comparison with human PBMCs. This qualitative enhancement could be obtained by up‐regulation of the inhibitory molecules noted above. Although there were individual differences in the development and the suppressive function of human T_reg cells in huNSG mice among the umbilical cord blood donors, they were augmented in all the huNSG mice engrafted with the CD34⁺ cells from the different donors in comparison with those in human PBMCs.

Thirdly, and more importantly, the acquisition and maintenance of the inhibitory phenotype in T_reg cells require T_reg cell‐specific epigenetic modifications [18]. Therefore, we examined the role of the epigenetic modifications in the qualitative and functional enhancement in human T_reg cells of huNSG mice.

Among a variety of epigenetic factors in T_reg cells, the specific histone modifications, H3K4me3 and H3K27me3, are well known to be involved in the stability of T_reg cells, which was proved by loss of the corresponding enzyme genes in T_reg cells [11, 13, 19]. Among these, H3K27me3 was specifically up‐regulated in human T_reg cells of huNSG mice compared with those of human PBMCs, in contrast to H3K4me3. In accordance with the change in T_reg cells, Ezh2, which catalyses the histone modification of H3K27me3, was also increased in T_reg cells from huNSG mice compared with those from humans.

H3K27me3 is known as a repressive mark correlated with gene silencing, whereas H3K4me3 is known as a permissive mark correlated with gene activation [20]. Even considering the previous report that showed that there was no H3K27me3 occupancy of the FOXP3 gene locus in T_reg cells [21], H3K27me3 might regulate T_reg cell stability indirectly. The candidate genes in which H3K27me3 deposition has such an effect by the gene silencing in T_reg cells were reported to be some of the proinflammatory genes that antagonize T_reg cell function, such as PDE3B [13].

Indeed, FoxP3 expression was strongly maintained in Ezh2‐high‐expressing CD4⁺ T cells compared with in Ezh2 low‐expressing CD4⁺ T cells in huNSG mice. This result suggested that Ezh2‐mediated H3K27me3 modification might contribute to the maintenance of FoxP3 expression in human T_reg cells of huNSG mice in vivo. Therefore, we examined the effect of Ezh2 inhibitor on the population of T_reg cells after the exposure to IL‐6 to clarify the role of Ezh2‐mediated H3K27me3 modification in T_reg cell stability under inflammatory conditions in vitro. IL‐6 is known to down‐regulate FOXP3 mRNA expression, and this down‐regulation may be partly explained by an observation that IL‐6 modifies the chromatin structure to break T_reg cell stability [22, 23]. The decrease in CD4⁺CD25⁺FoxP3⁺ T_reg cells, which means the population of so‐called ex‐T_reg cells, under the inflammatory condition by IL‐6 exposure was attenuated in huNSG mice compared with in humans. In other words, T_reg cell stability seemed to be higher in huNSG mice than in humans. In addition, Ezh2 inhibitor cancelled the difference in T_reg cell stability, which meant that the higher T_reg cell stability in huNSG mice depended upon enhanced Ezh2‐mediated H3K27me3 modification in T_reg cells (Fig. 6).

Fig. 6.

Schematic diagram of human regulatory T cells in a humanized non‐obese diabetic/severe combined immunodeficiency/interleukin (IL)‐2 receptor‐γ‐null (NOD/SCID/IL2rγ^null) (NSG) mouse depicting the role of enzymatic up‐regulation of H3K27me3. huT_reg cells in the spleens of humanized NSG (huNSG) mice showed increased population, higher expressions of FoxP3, CTLA‐4 and GITR, a higher production of IL‐10 and enhanced suppressive capacity when compared with those in human PBMCs. H3K27me3 and Ezh2 were specifically up‐regulated in human T_reg cells of huNSG mice in comparison with those of human PBMCs. The decrease in T_reg cells due to loss of FoxP3 expression induced by IL‐6 exposure was attenuated in huNSG mice when compared with human PBMCs, while the difference between them was cancelled by Ezh2 inhibitor (GSK343). Thus, huNSG mice exhibit functionally augmented human T_reg cells owing to enzymatic up‐regulation of H3K27me3. T_reg = regulatory T cell; T_conv = conventional T cell; GSK343 = Ezh2 inhibitor; Ezh2 = enhancer of zeste homolog 2; H3K27me3 = trimethylated histone H3 at lysine 27; me = methyl group; K27 = lysine 27; H3 = histone H3; FoxP3 = forkhead box protein P3; CTLA‐4 = cytotoxic T lymphocyte antigen 4; GITR = glucocorticoid‐induced tumor necrosis factor‐related protein.

Ezh2 was reported to be induced in part by CD28‐meditated co‐stimulation in T_reg cells to stabilize T_reg cell function [13] and constitutively expressed specifically in thymocytes [24]. Therefore, we hypothesized that Ezh2‐inducing mechanisms such as a CD28‐mediated co‐stimulatory signaling pathway could be constitutively strengthened, especially in the thymic microenvironment of huNSG mice. Elucidation of the key pathway involved in up‐regulation of Ezh2 by further analysis of huNSG mice is needed.

Thus, the present study provides novel insights into human T_reg cell behavior that complement the comprehensive understanding of the human immune system established in huNSG mice. When huNSG mice are used for in‐vivo analysis of the human immune response in a variety of immunological conditions, such as infection, cancer and autoimmunity, the pathological role of human T_reg cells in huNSG mice could also be analysed from another perspective on the basis of our present observations to elucidate the overall mechanism.

In conclusion, this is the first report, to our knowledge, to show that huNSG mice exhibit functionally augmented human T_reg cells associated with enzymatic up‐regulation of H3K27me3. Our observations of human T_reg cells in huNSG mice could reinforce the utility of these mice for the in‐vivo analysis of the human immune system.

Disclosures

None.

Author contributions

H. Ta., H. Ts., S. A., F. H., I. M. and T. S. designed the study. H. Ta., H. Ts., S. A. and Y. K. performed the experiments. H. Ta. and H. Ts. analysed and interpreted the data. H. Ta., H. Ts. and T. S. contributed to the first draft of the manuscript. T. S. assumed the final responsibility to submit the manuscript for publication. All authors had full access to all the data, carefully reviewed the manuscript and approved the final version.

Supporting information

Fig. S1. Individual data of development and suppressive function of human regulatory T cells in humanized NOD/SCID/IL2rγ^null (NSG) mice. (a) Individual data of proportion of human CD25⁺Foxp3⁺ regulatory T (Treg) cells among human CD4⁺ T cells in spleens of humanized NSG (huNSG) mice on day 120 (n = 5) integrated in Fig. 1e. Donor numbers correspond to those in Table S1. (b) Individual data of inhibition rates of human Treg cells from huNSG mice (n = 3) integrated in Fig. 3d. Donor numbers correspond to those in Table S1. Notes indicate conventional T cells : Treg cells ratios in the cocultures. hCD25: human CD25. hFoxp3: human forkhead box protein P3. hCD4: human CD4. Tconv: conventional T cells. Treg: regulatory T cells.

Table S1. Donor information.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.