NOD-scid.Il2rgnull (NSG) mice are currently being used as recipients to screen for pathogenic autoreactive T cells in type 1 diabetes (T1D) patients. We questioned whether the restriction of IL-2R γ-chain (Il-2rγ)–dependent cytokine signaling only to donor cells in NSG recipients differently influenced the activities of transferred diabetogenic T cells when they were introduced as a monoclonal/oligoclonal population versus being part of a polyclonal repertoire. Unexpectedly, a significantly decreased T1D transfer by splenocytes from prediabetic NOD donors was observed in Il-2rγnull-NSG versus Il-2rγ–intact standard NOD-scid recipients. In contrast, NOD-derived monoclonal/oligoclonal TCR transgenic β cell–autoreactive T cells in either the CD8 (AI4, NY8.3) or CD4 (BDC2.5) compartments transferred disease significantly more rapidly to NSG than to NOD-scid recipients. The reduced diabetes transfer efficiency by polyclonal T cells in NSG recipients was associated with enhanced activation of regulatory T cells (Tregs) mediated by NSG myeloid APC. This enhanced suppressor activity was associated with higher levels of Treg GITR expression in the presence of NSG than NOD-scid APC. These collective results indicate NSG recipients might be efficiently employed to test the activity of T1D patient–derived β cell–autoreactive T cell clones and lines, but, when screening for pathogenic effectors within polyclonal populations, Tregs should be removed from the transfer inoculum to avoid false-negative results.

Type 1 diabetes (T1D) in both humans and NOD mice results from the autoimmune destruction of insulin-producing pancreatic β cells mediated by the combined activity of pathogenic CD4 and CD8 T cells (1, 2). Although NOD mice develop T1D through mechanisms that appear to be pathologically similar to the case in humans, this model is not perfect, as some disease interventions effective in these animals have not yet proven to be clinically translatable (3). These difficulties have prompted the development of multiple humanized mouse models that could potentially be used to assess human T cells for diabetogenic activity and to screen interventions that might attenuate such pathogenic effectors (4).

The most promising humanized mouse models are those derived from the immunodeficient NOD.Cg-Prkdcscid Il2rgnull (NSG) stock. This NOD bicongenic stock harbors the spontaneous scid mutation that eliminates mature T and B lymphocytes, and also an engineered null mutation in the Il2rg gene (IL-2 common γ-chain receptor) that ablates signaling through the IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21 cytokine receptors (4). These combined mutations, which prevent the development of functional NK cells as well as lymphocytes, in conjunction with unique features of the NOD genetic background, enable NSG mice to support engraftment with human cells and tissues far more efficiently than other immunodeficient strains (4).

In both humans and NOD mice, the primary T1D genetic risk factor is provided by various combinations of MHC (designated HLA in humans)-encoded class I and II molecules (2). For this reason, NSG mice have also been further modified to transgenically express various human T1D-associated HLA class I and class II molecules (5). In recent years, there have been several studies testing whether such NSG-HLA transgenic mouse stocks can be used to assess human T cells for diabetogenic activity. Adoptive transfer of PBMCs containing a polyclonal array of T cells from a human T1D patient carrying the HLA-A2.1 class I variant was reported to induce a leukocytic infiltration of pancreatic islets (insulitis) in NSG-HLA-A2.1 transgenic recipients (6). However, the specificity of this inflammatory response was unclear. There have been two other reports that a T1D patient–derived CD8 T cell clone or CD4 T cell lines recognizing β cell autoantigens can induce both insulitis and specific β cell death when engrafted into appropriate HLA transgenic NSG recipients (7, 8). It should be noted that, to date, transferred polyclonal or monoclonal T cells from T1D patient donors have not yet induced overt hyperglycemia in NSG recipients. Hence, whereas introduction of the inactivated Il2rg gene enables higher engraftment levels of human T cells in NSG mice compared with first-generation NOD-scid recipients, this mutation’s negative effects on cytokine receptor signaling in host APC may also limit the functional activation of potential diabetogenic effectors in the transfer inoculum. Furthermore, in NSG recipients, IL-2rγ–dependent cytokine signaling is limited to donor cells. Consequently, different outcomes might ensue if the transferred diabetogenic T cells were monoclonal or oligoclonal in nature versus being a relatively small part of a polyclonal repertoire within a PBMC inoculum also containing donor APC.

Because of the above possibilities, we assessed whether the well-known ability of total splenocytes or β cell–autoreactive T cell clones derived from standard NOD donors to transfer T1D to NOD-scid recipients was recapitulated in NSG hosts.

NOD/ShiLtDvs, NOD-scid (NOD.Emv30−/−.CB17-Prkdcscid), and NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) mice are maintained in our specific pathogen-free research colony at The Jackson Laboratory. NOG (NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac) mice were purchased from Taconic (Germantown, NY). A NOD stock transgenically expressing the TCR from the diabetogenic AI4 CD8 T cell clone, and also homozygous for an inactivated Rag1 allele (NOD.Rag1−/−.AI4αβTg/DvsJ, and in this work designated NOD-AI4), has been previously described (9). NOD stocks transgenically expressing diabetogenic TCRs derived from the NY8.3 CD8+ [NOD.Cg-Tg(TcraTcrbNY8.3)1Pesa/DvsJ] (10) or BDC2.5 CD4+ [NOD.Cg-Tg(TcraBDC2.5,TcrbBDC2.5)1Doi/DoiJ] T cell clones (11) were acquired from the type 1 diabetes resource (http://type1diabetes.jax.org/) operated at The Jackson Laboratory. A N18 backcross generation NOD stock congenically carrying the previously described Foxp3tm2(eGFP)-Tch reporter construct (12) (formal designation NOD/LtDvsJ.Cg.B6-Foxp3tm2(eGFP)-Tch/Dvs and in this work noted as NOD.Foxp3-eGFP JAX stock 25097) was generated and also typed as homozygous for NOD alleles at markers delineating all known Idd genetic loci (2). The enhanced GFP (eGFP) reporter is a knockin downstream of the Foxp3 promoter and coding sequences designed to ensure independent and functional expression of both proteins (12).

Various fluorochrome-labeled Abs specific for CD8 (53-6.72), CD4 (RM4-5), CD3 (145-2C11), TCRVα8.3 (B21.14), TCRVβ8.1.2 (KJ16), TCRVβ4 (KT4), CD44 (IM7.8.1), CD62L (MEL-14), CD25 (7D4), CD19 (ID3), CD11c (N418), CD11b (M1/70), PDCA-1 (927), CD80 (16-10A1), CD86 (GL-1), CTLA-4 (IB8), GITR (YGITR765), GITRL (YGL386), and CD70 (FR70) were purchased from BD Biosciences (San Jose, CA) or BioLegend (San Diego, CA). Anti-mouse Foxp3 (FJK-16s) was purchased from eBioscience (San Diego, CA). NRP-V7-H-2Kd tetramer-KYNKANVFL-PE was acquired from the National Institutes of Health tetramer core facility (Atlanta, GA). Anti-mouse CD25 (PC61.5) and CD4 (GK1.5) were purchased from Bioxcell (West Lebanon, NH). Dead cells were excluded by DAPI staining. Stained cells were acquired using a FACS LSRII instrument (BD Biosciences). All flow cytometric data were analyzed with FlowJo (Tree Star, Ashland, OR). In one experiment, the previously described flow cytometry–based assay (13) was used to compare the ability of residual myeloid APC from NOD-scid and NSG mice to support regulatory T cell (Treg) activity. In studies assessing their proliferative capacity by flow cytometry, Tregs (Foxp3-eGFP+) or conventional T cells (Foxp3-eGFP) were stained following the manufacturer’s recommendations with the Dye eFluor670 (eBioscience) that dilutes with each division cycle.

NOD-scid and NSG splenocytes were prepared using collagenase digestion. The cells were seeded at a density of 2.5 × 105/well in a 96-well round-bottom plate in a final volume of 200 μl culture media. Mouse rIL-2 (BioLegend) was added to a final concentration of 10 pg/ml. As one control, sodium azide (NaN3; Sigma Aldrich, St. Louis, MO) was added at a concentration of 0.1% to inhibit the receptor-mediated IL-2 uptake process (14). Another control consisted of IL-2 in media with no cells. After 3 d of incubation, remaining IL-2 in the supernatants was assayed by ELISA (mouse IL-2 ELISA set, BD OptEIA; BD Biosciences, San Diego, CA).

The indicated recipient female mice (6–10 wk old) were injected i.v. with the described combinations of 5 × 106 NOD and/or 2.5 × 106 NOD-AI4 total splenocytes (donor age 6–8 wk old). For some experiments, NSG recipients received AI4 cells alone or admixed with total or CD4 T cell–depleted splenocytes from standard NOD donors. CD4 T cells were depleted by negative selection over MACS columns (Miltenyi Biotec). Appropriate NSG recipients were injected i.p. with 50 μg depleting CD4-specific Ab (GK1.5) once weekly. Maintenance of CD4 T cell depletion was verified by flow cytometric analyses of PBL. Some experiments entailed the transfer into NSG recipients of AI4 cells alone or admixed with total or Foxp3+ Treg-depleted NOD splenocytes. eGFP reporter-expressing Tregs were depleted by flow cytometric sorting (FACSAriaII; BD Biosciences). Appropriate NSG recipients were injected i.p. with 250 μg depleting anti-CD25 Ab (PC61.5) once per week to ensure continued elimination of Tregs. Treg depletion was verified by flow cytometric analyses of PBL. CD4 or Treg-depleted splenocytes were adjusted to inject 2 × 105 CD8 T cells (equivalent quantity present in 5 × 106 total NOD splenocytes). In other experiments, NOD-scid and NSG female recipients (8–10 wk old) were injected i.v. with total splenocytes obtained from recently diabetic NOD-NY8.3 female donors. The cells were adjusted to inject 2 × 105 TCRVβ8.1.2+ CD8 T cells (NY8.3 T cells). NOD-BDC2.5 mice were used as a source for autoreactive diabetogenic CD4 T cells. Total CD4+ T cells from NOD-BDC2.5 mice were purified by negative selection using MACS columns. Both NSG and NOD-scid recipients were injected i.v. with 1.5 × 106 BDC2.5 CD4 T cells. For some experiments, 2 × 105 CD4+CD25 BDC2.5 T cells were purified using MACS columns and injected i.v. into NOD-scid or NSG mice.

Diabetes was monitored using urine glucose strips (Diastix; Bayer). Mice with two consecutive readings >3 were considered diabetic.

Data were analyzed using Prism software (GraphPad, San Diego, CA). For diabetes incidence, significance was calculated using log-rank (Mantel–Cox) test.

We compared the ability of total splenocytes from 6- to 8-wk-old NOD female donors to adoptively transfer T1D to NOD-scid and NSG recipients. As expected, NOD splenocytes efficiently transferred T1D to NOD-scid mice, but curiously induced a significantly lower level of disease in NSG recipients (Fig. 1A). In contrast, NOD-derived monoclonal CD8 T cells transgenically expressing the diabetogenic AI4 TCR (15) transferred disease significantly more rapidly to NSG than NOD-scid recipients (Fig. 1B). Thus, normally diabetogenic T cells residing within a polyclonal repertoire had a limited ability to transfer disease to NSG recipients, but a monoclonal population of such effectors efficiently did so.

FIGURE 1.

T1D transfer to NOD-scid and NSG recipients by polyclonal versus monoclonal autoreactive T cells. (A) Adoptive transfer of T1D by total NOD splenocytes (5 × 106) from prediabetic 6- to 8-wk-old female donors to NOD-scid (n = 20) and NSG (n = 20) recipients. (B) Adoptive transfer of T1D by monoclonal AI4 CD8 T cells (2.5 × 106 NOD-AI4 splenocytes) to NOD-scid (n = 27) and NSG (n = 31) mice. In both (A) and (B), T1D development rates compared by log-rank (Mantel–Cox) test (***p < 0.0001). (C and D) Frequency over time following engraftment with total NOD splenocytes of CD8 and CD4 T cells among PBL in NOD-scid and NSG recipients. (E and F) Frequency over time following engraftment with NOD-AI4 splenocytes of CD8 and CD4 T cells among PBL in NOD-scid and NSG recipients. (G) Total AI4 T cell numbers (CD8+ TCRVα8.3+) in secondary lymphoid tissues: spleen (SPL), PLN, and mesenteric lymph nodes (MLN). (H) Proportion of AI4 CD8 T cells with effector memory phenotype (CD44highCD62Llow). (G and H) Data from three independent experiments with a total n = 9 in each strain. In (C)–(H), *p < 0.05, **p < 0.001, ***p < 0.0001; ns, p > 0.05, two-way ANOVA with Bonferroni posttest correction.

FIGURE 1.

T1D transfer to NOD-scid and NSG recipients by polyclonal versus monoclonal autoreactive T cells. (A) Adoptive transfer of T1D by total NOD splenocytes (5 × 106) from prediabetic 6- to 8-wk-old female donors to NOD-scid (n = 20) and NSG (n = 20) recipients. (B) Adoptive transfer of T1D by monoclonal AI4 CD8 T cells (2.5 × 106 NOD-AI4 splenocytes) to NOD-scid (n = 27) and NSG (n = 31) mice. In both (A) and (B), T1D development rates compared by log-rank (Mantel–Cox) test (***p < 0.0001). (C and D) Frequency over time following engraftment with total NOD splenocytes of CD8 and CD4 T cells among PBL in NOD-scid and NSG recipients. (E and F) Frequency over time following engraftment with NOD-AI4 splenocytes of CD8 and CD4 T cells among PBL in NOD-scid and NSG recipients. (G) Total AI4 T cell numbers (CD8+ TCRVα8.3+) in secondary lymphoid tissues: spleen (SPL), PLN, and mesenteric lymph nodes (MLN). (H) Proportion of AI4 CD8 T cells with effector memory phenotype (CD44highCD62Llow). (G and H) Data from three independent experiments with a total n = 9 in each strain. In (C)–(H), *p < 0.05, **p < 0.001, ***p < 0.0001; ns, p > 0.05, two-way ANOVA with Bonferroni posttest correction.

Close modal

Flow cytometric analyses of PBL from both recipient types of total NOD splenocytes showed a progressive increase in CD8 T cells posttransfer, but with significantly higher proportions in NSG than NOD-scid mice (Fig. 1C). The proportion of CD4 T cells increased over time to significantly higher levels in NOD-scid than NSG recipients (Fig. 1D). These data indicated the lack of IL-2rγ expression in NSG host cells resulted in differential engraftment of donor polyclonal CD8 and CD4 T cells compared with NOD-scid recipients. However, these data also indicated that a gross engraftment deficiency of adoptively transferred NOD polyclonal T cells did not account for their suppressed induction of T1D in NSG recipients. Furthermore, both NSG and NOD-scid recipients of total NOD splenocytes exhibited equivalent levels of donor B lymphocyte engraftment (data not shown).

Flow cytometric analyses of PBL from recipients of monoclonal AI4 CD8 T cells showed a higher proportion of such effectors in NSG than NOD-scid mice (Fig. 1E). As expected, PBL from either NOD-scid or NSG recipients of monoclonal AI4 cells did not contain CD4 T cells (Fig. 1F). Similar to the case for PBL, at 2 wk posttransfer significantly higher numbers of transferred monoclonal AI4 CD8 T cells were detected in the spleens of NSG than NOD-scid recipients (Fig. 1G). In addition, analyses of activation markers on the same samples showed the proportion of effector-memory type (CD44highCD62Llow) AI4 CD8 T cells in pancreatic lymph nodes (PLN) and mesenteric lymph nodes was significantly higher in NSG than NOD-scid recipients (Fig. 1H). This greater expansion and conversion to an effector memory phenotype of transferred monoclonal AI4 T cells in NSG than NOD-scid mice most likely account for why these effectors induced T1D more aggressively in the former recipients.

The inactivated Il2rg gene in NSG host cells apparently contributes to the heightened ability of monoclonal AI4 CD8 T cells to engraft and induce T1D in this strain compared with NOD-scid recipients. Conversely, in contrast to standard NOD-scid recipients, polyclonal splenic NOD T cells could not efficiently induce T1D in NSG mice. We hypothesized these results could be explained by IL-2rγ–dependent cytokine-signaling events only taking place in donor-derived cells in NSG recipients. To initially test this possibility, we compared the in vitro ability of residual myeloid splenocytes from NSG or NOD-scid mice to capture and internalize IL-2 as a model of a common γ-chain cytokine. This assay (Fig. 2A) clearly showed an impaired ability of NSG-derived splenocytes to respond to the IL-2 common γ-chain cytokine. Based on this finding, we next tested whether the NSG environment allows some cell population present in standard NOD, but not NOD-AI4 donor splenocytes, to exert enhanced T1D-protective effects not elicited in NOD-scid recipients. This was done by determining whether the ability of monoclonal AI4 T cells to aggressively mediate T1D development in NSG recipients was attenuated by cotransfer of NOD splenocytes. This was indeed the case (Fig. 2B). We next evaluated whether the presence of some NOD splenocyte-derived cell population(s) blocked engraftment of diabetogenic AI4 CD8 T cells in NSG recipients. When cotransferred with NOD splenocytes, AI4 T cells continued to engraft at significantly higher levels in the spleens of NSG than NOD-scid recipients (Fig. 2C). However, in contrast to when they were transferred alone, following coinfusion with NOD splenocytes, AI4 T cells underwent significantly lower conversion levels to an effector memory phenotype in the PLNs of NSG than NOD-scid recipients (Fig. 2D). These collective findings indicate that when in a NSG, but not NOD-scid recipient environment, a NOD leukocyte population(s) acquires an enhanced ability to limit the cytopathic activation of diabetogenic T cells.

FIGURE 2.

The inability of NSG recipients to respond to common γ-chain cytokines is associated with an ability of NOD donor polyclonal T cells to impair T1D transfer by AI4 CD8 T cells. (A) Total splenocytes from NOD-scid or NSG mice (2.5 × 105) were cultured in vitro in the presence of 10 pg/ml IL-2 and with or without 0.1% NaN3 acting as inhibitor of cytokine translocation. After 3 d in culture, the amount of IL-2 remaining in the supernatant was quantified by ELISA. (B) NSG and NOD-scid mice were injected i.v. with a mixture of 2.5 × 106 NOD-AI4 and 5 × 106 standard NOD splenocytes. Data indicate T1D incidence from two independent experiments, total n = 20, for each strain. Disease incidence compared by log-rank (Mantel–Cox) test (***p < 0.0001). (C and D) At 2 wk posttransfer, the total number and activation phenotype of AI4 CD8+ T cells in each type of recipient (n = 3 and 4, respectively, for NOD-scid and NSG recipients) were analyzed by flow cytometry in spleen (SPL), PLN, and mesenteric lymph node (MLN). In (A), (C), and (D), bars represent mean ± SEM. *p < 0.05, ***p < 0.0001; ns, p > 0.05; two-way ANOVA with Bonferroni posttest correction.

FIGURE 2.

The inability of NSG recipients to respond to common γ-chain cytokines is associated with an ability of NOD donor polyclonal T cells to impair T1D transfer by AI4 CD8 T cells. (A) Total splenocytes from NOD-scid or NSG mice (2.5 × 105) were cultured in vitro in the presence of 10 pg/ml IL-2 and with or without 0.1% NaN3 acting as inhibitor of cytokine translocation. After 3 d in culture, the amount of IL-2 remaining in the supernatant was quantified by ELISA. (B) NSG and NOD-scid mice were injected i.v. with a mixture of 2.5 × 106 NOD-AI4 and 5 × 106 standard NOD splenocytes. Data indicate T1D incidence from two independent experiments, total n = 20, for each strain. Disease incidence compared by log-rank (Mantel–Cox) test (***p < 0.0001). (C and D) At 2 wk posttransfer, the total number and activation phenotype of AI4 CD8+ T cells in each type of recipient (n = 3 and 4, respectively, for NOD-scid and NSG recipients) were analyzed by flow cytometry in spleen (SPL), PLN, and mesenteric lymph node (MLN). In (A), (C), and (D), bars represent mean ± SEM. *p < 0.05, ***p < 0.0001; ns, p > 0.05; two-way ANOVA with Bonferroni posttest correction.

Close modal

The results described above indicated that when transferred alone into NSG, but not in NOD-scid recipients, common γ-chain cytokines produced by donor monoclonal AI4 T cells would only be able to feed back to these pathogenic effectors. We reasoned such lessened receptor-mediated competition for IL-2rγ–dependent cytokines in NSG than NOD-scid recipients could explain why, when engrafted alone, monoclonal AI4 CD8 T cells elicit a more aggressive T1D onset in the former host environment. To further test this possibility, we analyzed another NOD-scid–related strain, NOG mice encoding an alternatively mutated version of the IL-2rγ subunit that is expressed on the cell surface, but lacking a cytoplasmic domain needed to initiate cytokine signaling (5, 16). As a result, leukocytes in NOG mice can bind, but not respond, to IL-2rγ–dependent cytokines. Thus, we tested whether a possible ability of host cells to limit availability of IL-2rγ–dependent cytokines to pathogenic effectors resulted in less aggressive T1D development in NOG than NSG recipients of only monoclonal AI4 CD8 T cells. Splenic engraftment levels of adoptively transferred monoclonal AI4 T cells were significantly less in NOG than NSG recipients, but higher than in NOD-scid mice (Fig. 3A). This result might suggest that NSG-based stocks are a better choice if needed as recipients to support high expansion levels of T cell clones. However, despite this engraftment difference, AI4-mediated T1D development was comparable in both NSG and NOG and equally more aggressive than in NOD-scid recipients (Fig. 3B). Similar results were observed in an independent experiment when 10-fold lower numbers (2 × 105) of NOD-AI4 splenocytes were transferred into NSG and NOG recipients (data not shown). It should be noted that, when observed for longer times posttransfer than in the experiment shown in Fig. 3B, most NOD-scid recipients of monoclonal AI4 T cells also eventually developed T1D (see Fig. 1B). These findings utilizing NSG versus NOG recipients indicate variability in host cell–mediated competition for binding of IL-2rγ–dependent cytokines may alter the engraftment levels of monoclonal diabetogenic T cells, but does not result in differential pathogenic activation of such effectors.

FIGURE 3.

IL-2rγ–dependent receptors that can bind, but not induce cytokine signaling in host cells alter the engraftment of diabetogenic T cells without blocking their activation. Splenocytes (2.5 × 106) from NOD-AI4 TCR transgenic mice were injected i.v. into NSG (n = 13), NOG (n = 15), and NOD-scid (n = 13) recipients. (A) At 2 wk posttransfer, three mice of each recipient type were euthanized, and spleen (SPL), PLN, and mesenteric lymph nodes (MLN) were analyzed for total AI4 CD8+ T cell numbers by flow cytometry. Bars represent mean ± SEM; statistical analyses by two-way ANOVA with Bonferroni posttest correction. ns, p > 0.05; *p < 0.05, **p < 0.001. (B) Ten mice of each recipient type were assessed for T1D incidence. Log-rank (Mantel–Cox) test (***p < 0.0001).

FIGURE 3.

IL-2rγ–dependent receptors that can bind, but not induce cytokine signaling in host cells alter the engraftment of diabetogenic T cells without blocking their activation. Splenocytes (2.5 × 106) from NOD-AI4 TCR transgenic mice were injected i.v. into NSG (n = 13), NOG (n = 15), and NOD-scid (n = 13) recipients. (A) At 2 wk posttransfer, three mice of each recipient type were euthanized, and spleen (SPL), PLN, and mesenteric lymph nodes (MLN) were analyzed for total AI4 CD8+ T cell numbers by flow cytometry. Bars represent mean ± SEM; statistical analyses by two-way ANOVA with Bonferroni posttest correction. ns, p > 0.05; *p < 0.05, **p < 0.001. (B) Ten mice of each recipient type were assessed for T1D incidence. Log-rank (Mantel–Cox) test (***p < 0.0001).

Close modal

We hypothesized that the restriction in NSG recipients of IL-2rγ–dependent cytokine-signaling events to donor cells might preferentially boost the numbers or functional activity of a Treg-like subset(s) present in a transferred polyclonal, but not monoclonal effector population, to levels inhibiting T1D development. This hypothesis was based on previous findings that the survival and immunosuppressive activity of the CD4+Foxp3+ subset of Tregs are highly IL-2 dependent (17). The AI4 CD8 T cell clonotype does not require interactions with CD4 helper populations to efficiently induce T1D (15). Thus, we determined whether the capacity of total NOD splenocytes to inhibit the diabetogenic activity of cotransferred AI4 CD8 T cells in NSG recipients required CD4 T cells. Flow cytometric analyses of PBL indicated only cotransfer of total and not CD4 T cell–depleted NOD splenocytes suppressed the engraftment levels of AI4 T cells in NSG recipients (Fig. 4). Furthermore, total, but not CD4 T cell–depleted NOD splenocytes inhibited the ability of AI4 T cells to transfer T1D to NSG recipients (Fig. 5A). These results demonstrated a CD4-expressing population(s) is responsible for the ability of total NOD splenocytes to inhibit disease induced by monoclonal AI4 CD8 T cells in NSG recipients.

FIGURE 4.

CD4 T cells present in cotransferred NOD splenocytes inhibit the engraftment of diabetogenic AI4 CD8 T cell NSG recipient mice. NSG mice were injected i.v. with 2.5 × 106 NOD-AI4 splenocytes alone or admixed with either total or CD4 T cell–depleted NOD splenocytes. Recipients of CD4-depleted splenocytes were injected i.p. with 200 μg depleting CD4-specific GK1.5 Ab once per week. The total number of CD4 T cell–depleted NOD splenocytes was adjusted to inject 2 × 105 CD8+ T cells, an equivalent amount to that in 5 × 106 total NOD splenocytes. Another control group received only CD4 T cell–depleted NOD splenocytes. At various time points posttransfer, PBL from each group of NSG recipients were analyzed by flow cytometry for numbers of AI4 T cells (CD8+ TCRVα8.3+). Graph shows mean ± SEM. The statistics represent the significance of each group compared with controls receiving AI4 T cells only. ns, p > 0.05; ***p < 0.001, two way ANOVA and Bonferroni posttest.

FIGURE 4.

CD4 T cells present in cotransferred NOD splenocytes inhibit the engraftment of diabetogenic AI4 CD8 T cell NSG recipient mice. NSG mice were injected i.v. with 2.5 × 106 NOD-AI4 splenocytes alone or admixed with either total or CD4 T cell–depleted NOD splenocytes. Recipients of CD4-depleted splenocytes were injected i.p. with 200 μg depleting CD4-specific GK1.5 Ab once per week. The total number of CD4 T cell–depleted NOD splenocytes was adjusted to inject 2 × 105 CD8+ T cells, an equivalent amount to that in 5 × 106 total NOD splenocytes. Another control group received only CD4 T cell–depleted NOD splenocytes. At various time points posttransfer, PBL from each group of NSG recipients were analyzed by flow cytometry for numbers of AI4 T cells (CD8+ TCRVα8.3+). Graph shows mean ± SEM. The statistics represent the significance of each group compared with controls receiving AI4 T cells only. ns, p > 0.05; ***p < 0.001, two way ANOVA and Bonferroni posttest.

Close modal
FIGURE 5.

Polyclonal repertoire-derived CD4+Foxp3+ Tregs inhibit diabetogenic T cells in NSG recipients. (A) Diabetes incidence in NSG mice injected i.v. with 2.5 × 106 monoclonal NOD-AI4 splenocytes either alone or admixed with total or CD4 T cell–depleted NOD splenocytes. Mice receiving CD4 T cell–deficient splenocytes were injected i.p. with 200 μg depleting CD4-specific GK1.5 Ab once per week. The number of CD4-depleted NOD splenocytes was adjusted to inject 2 × 105 CD8 T cells an equivalent amount to that present in 5 × 106 total NOD splenocytes. (B) Diabetes incidence study comparing NOD.Foxp3-eGFP reporter mice with NOD/ShiLtDvs. Log-rank (Mantel–Cox) test, ns p > 0.05. (C) Spleen cells from NOD.Foxp3-eGFP reporter mice were intracellularly stained with allophycocyanin (AP)-labeled anti-Foxp3 (FJK-16s) to test whether Foxp3 and the fluorescent eGFP reporter were coexpressed in the same cells. Representative dot-plot graph gated on total CD4+ T cells shows that only Foxp3+ T cells coexpress eGFP (mean ± SEM, n = 5). (D) Newly differentiated CD4+Foxp3+ Tregs arise in NSG and NOD-scid recipients originally engrafted with reporter cell–depleted NOD.Foxp3-eGFP splenocytes. NOD.Foxp3-eGFP splenocytes depleted of reporter-positive cells by flow cytometric sorting were injected i.v. into NSG and NOD.scid mice. The number of reporter cell–depleted NOD.Foxp3-eGFP splenocytes was adjusted to inject 2 × 105 CD8 T cells. A group of NSG and NOD-scid mice receiving reporter cell–deficient NOD.Foxp3-eGFP splenocytes was also injected i.p. with 250 μg depleting CD25-specific PC61.5 Ab once per week to maintain Treg ablation (solid lines), and another group of each recipient type was kept without Ab treatment (dashed lines). At the indicated time point, PBL were analyzed by flow cytometry for proportions of eGFP reporter-expressing Tregs. Data represent the mean ± SEM (n = 5, each group). By 2 wk posttransfer, a significant increase in the proportion of CD4+Foxp3+ Tregs was observed in both types of recipients not receiving anti-CD25 Ab treatments. There were no significant differences between recipient strains in the anti-CD25–treated or untreated groups. ns, p > 0.05; ***p < 0.0001, two-way ANOVA and Bonferroni posttest. (E) NSG mice were injected i.v. with 2.5 × 106 monoclonal NOD-AI4 splenocytes either alone or admixed with total or reporter cell–depleted NOD.Foxp3-eGFP splenocytes. The number of intact or reporter cell–depleted NOD.Foxp3-eGFP splenocytes was adjusted to inject 2 × 105 CD8 T cells. Mice receiving reporter cell–deficient NOD.Foxp3-eGFP splenocytes were also injected i.p. with 250 μg depleting CD25-specific PC61.5 Ab once per week to maintain Treg ablation. (F) NSG mice were injected i.v. with total or reporter cell–depleted NOD.Foxp3-eGFP splenocytes. Numbers of the two NOD.Foxp3-eGFP splenocyte inoculums were adjusted to inject 2 × 105 CD8 T cells. Mice receiving reporter cell–deficient NOD.Foxp3-eGFP splenocytes were also injected i.p. with 250 μg depleting CD25-specific PC61.5 Ab once per week to maintain Treg ablation. The incidence of T1D in various groups was compared by log-rank (Mantel–Cox) test, ***p < 0.0001.

FIGURE 5.

Polyclonal repertoire-derived CD4+Foxp3+ Tregs inhibit diabetogenic T cells in NSG recipients. (A) Diabetes incidence in NSG mice injected i.v. with 2.5 × 106 monoclonal NOD-AI4 splenocytes either alone or admixed with total or CD4 T cell–depleted NOD splenocytes. Mice receiving CD4 T cell–deficient splenocytes were injected i.p. with 200 μg depleting CD4-specific GK1.5 Ab once per week. The number of CD4-depleted NOD splenocytes was adjusted to inject 2 × 105 CD8 T cells an equivalent amount to that present in 5 × 106 total NOD splenocytes. (B) Diabetes incidence study comparing NOD.Foxp3-eGFP reporter mice with NOD/ShiLtDvs. Log-rank (Mantel–Cox) test, ns p > 0.05. (C) Spleen cells from NOD.Foxp3-eGFP reporter mice were intracellularly stained with allophycocyanin (AP)-labeled anti-Foxp3 (FJK-16s) to test whether Foxp3 and the fluorescent eGFP reporter were coexpressed in the same cells. Representative dot-plot graph gated on total CD4+ T cells shows that only Foxp3+ T cells coexpress eGFP (mean ± SEM, n = 5). (D) Newly differentiated CD4+Foxp3+ Tregs arise in NSG and NOD-scid recipients originally engrafted with reporter cell–depleted NOD.Foxp3-eGFP splenocytes. NOD.Foxp3-eGFP splenocytes depleted of reporter-positive cells by flow cytometric sorting were injected i.v. into NSG and NOD.scid mice. The number of reporter cell–depleted NOD.Foxp3-eGFP splenocytes was adjusted to inject 2 × 105 CD8 T cells. A group of NSG and NOD-scid mice receiving reporter cell–deficient NOD.Foxp3-eGFP splenocytes was also injected i.p. with 250 μg depleting CD25-specific PC61.5 Ab once per week to maintain Treg ablation (solid lines), and another group of each recipient type was kept without Ab treatment (dashed lines). At the indicated time point, PBL were analyzed by flow cytometry for proportions of eGFP reporter-expressing Tregs. Data represent the mean ± SEM (n = 5, each group). By 2 wk posttransfer, a significant increase in the proportion of CD4+Foxp3+ Tregs was observed in both types of recipients not receiving anti-CD25 Ab treatments. There were no significant differences between recipient strains in the anti-CD25–treated or untreated groups. ns, p > 0.05; ***p < 0.0001, two-way ANOVA and Bonferroni posttest. (E) NSG mice were injected i.v. with 2.5 × 106 monoclonal NOD-AI4 splenocytes either alone or admixed with total or reporter cell–depleted NOD.Foxp3-eGFP splenocytes. The number of intact or reporter cell–depleted NOD.Foxp3-eGFP splenocytes was adjusted to inject 2 × 105 CD8 T cells. Mice receiving reporter cell–deficient NOD.Foxp3-eGFP splenocytes were also injected i.p. with 250 μg depleting CD25-specific PC61.5 Ab once per week to maintain Treg ablation. (F) NSG mice were injected i.v. with total or reporter cell–depleted NOD.Foxp3-eGFP splenocytes. Numbers of the two NOD.Foxp3-eGFP splenocyte inoculums were adjusted to inject 2 × 105 CD8 T cells. Mice receiving reporter cell–deficient NOD.Foxp3-eGFP splenocytes were also injected i.p. with 250 μg depleting CD25-specific PC61.5 Ab once per week to maintain Treg ablation. The incidence of T1D in various groups was compared by log-rank (Mantel–Cox) test, ***p < 0.0001.

Close modal

Foxp3+ Tregs are a crucial CD4-coexpressing immunological suppressor subset (17). We used a newly developed NOD mouse stock in which an eGFP reporter is specifically expressed in Tregs (NOD.Foxp3-eGFP) to determine whether this immunosuppressive population differentially engrafts NOD-scid and NSG recipients. The NOD.Foxp3-eGFP stock develops T1D at a rate similar to standard NOD/LtDvs mice and expression of Foxp3 correlated with that of eGFP (Fig. 5B, 5C). Flow cytometric analyses of CD4+Foxp3+ Treg levels at different time points postengraftment with NOD.Foxp3-eGFP splenocytes showed no numerical strain differences between NOD-scid and NSG recipients (data not shown). This indicated the lower level of T1D development in NSG versus NOD-scid recipients of polyclonal NOD T cells is not due to a greater numerical engraftment of Tregs in the former strain. However, we hypothesized that enhanced Treg function may contribute to the limited diabetogenic activity of polyclonal NOD T cells in NSG recipients. NSG recipients were infused with AI4 T cells alone or admixed with either total or reporter cell-depleted NOD.Foxp3-eGFP splenocytes. To avoid the generation of newly differentiated Foxp3+ Tregs (Fig. 5D), NSG mice engrafted with reporter cell–depleted NOD.Foxp3-eGFP splenocytes were treated with a CD25-specific Ab to maintain Treg ablation. Total, but not Treg-depleted splenocytes from NOD.Foxp3-eGFP donors blocked T1D development in NSG recipients coinfused with AI4 T cells (Fig. 5E). Furthermore, after being depleted of marked Tregs, NOD.Foxp3-eGFP splenocytes containing polyclonal T cell populations acquired an ability when transferred on their own to efficiently induce T1D in NSG recipients (Fig. 5F). Importantly, we also found the residual myeloid APC population in NSG mice supported Treg suppressive activity to a significantly greater extent than those of NOD-scid origin (Fig. 6A, 6B). In at least some circumstances, increased expression and activity of the GITR costimulatory molecule can enhance Treg function (18). Thus, we tested whether GITR expression by Tregs resident among NOD splenocytes differed when these cells were transferred alone or in combination with AI4 effectors into NSG and NOD-scid recipients. Under both transfer situations, GITR expression by NOD donor splenocyte-derived Tregs, but not conventional T cells, was significantly higher in the PLNs of NSG than NOD-scid recipients (Fig. 6C, 6D). No differences were observed for GITR ligand expression (data not shown). Thus, the induction of a higher level of GITR expression may at least partially explain why Foxp3+ Tregs exert greater suppressive activity when interacting with myeloid APC in NSG than NOD-scid recipients.

FIGURE 6.

Residual myeloid APC from NSG mice support higher levels of Treg activity than those of NOD-scid origin. CD4 T cells within splenocytes from NOD.Foxp3-eGFP mice were sorted into conventional or Treg populations based on the respective absence or presence of eGFP marker expression. The conventional CD4 T cells were then labeled with the eFlour 670 dye. The conventional CD4 and Treg populations were cocultured at the indicated ratios for 2 d in the presence of 1 μg/ml anti-CD3 (1452C11) and 2 × 105 NSG or NOD-scid splenic leukocytes as a source of APC. The percentages of proliferating conventional CD4 T cells were determined by flow cytometric analyses of eFlour 670 dilution. (A) Representative flow cytometric profiles of conventional CD4 T cell proliferation (indicated by eFlour 670 dye dilution) in the presence or absence of the indicated ratios of Foxp3+ Tregs and either NOD-scid or NSG APC. Numbers shown in each panel depict the mean percentage ± SEM of conventional CD4 T cells that remained in an undivided state (within the 0 dilution peak) under the indicated culture condition (n = 3 for each). (B) Summary of the percent suppression ± SEM (n = 3 for each culture condition) of conventional CD4 T cell responses in the presence versus the absence of Foxp3+ Tregs and either NOD-scid or NSG APC. Suppression was defined by the proportions of conventional CD4 T cells remaining in an undivided state. *Foxp3+ Treg-mediated suppression of conventional CD4 T cell proliferative responses significantly greater (p < 0.05, Student t test) in the presence of NSG than NOD-scid APC. (C) NSG or NOD-scid mice (n = 3 per group) were infused either with 5 × 106 NOD.Foxp3-eGFP splenocytes alone or (D) in combination with 2.5 × 106 NOD-AI4 splenocytes. At 10 d posttransfer, the mean fluorescence intensity (MFI) of GITR-specific Ab staining by Foxp3+ Tregs as well as CD4 and CD8 conventional T cells within the PLN of each recipient type was determined. Data represent mean ± SEM. *p < 0.05, ****p < 0.00001, two-way ANOVA and Bonferroni posttest.

FIGURE 6.

Residual myeloid APC from NSG mice support higher levels of Treg activity than those of NOD-scid origin. CD4 T cells within splenocytes from NOD.Foxp3-eGFP mice were sorted into conventional or Treg populations based on the respective absence or presence of eGFP marker expression. The conventional CD4 T cells were then labeled with the eFlour 670 dye. The conventional CD4 and Treg populations were cocultured at the indicated ratios for 2 d in the presence of 1 μg/ml anti-CD3 (1452C11) and 2 × 105 NSG or NOD-scid splenic leukocytes as a source of APC. The percentages of proliferating conventional CD4 T cells were determined by flow cytometric analyses of eFlour 670 dilution. (A) Representative flow cytometric profiles of conventional CD4 T cell proliferation (indicated by eFlour 670 dye dilution) in the presence or absence of the indicated ratios of Foxp3+ Tregs and either NOD-scid or NSG APC. Numbers shown in each panel depict the mean percentage ± SEM of conventional CD4 T cells that remained in an undivided state (within the 0 dilution peak) under the indicated culture condition (n = 3 for each). (B) Summary of the percent suppression ± SEM (n = 3 for each culture condition) of conventional CD4 T cell responses in the presence versus the absence of Foxp3+ Tregs and either NOD-scid or NSG APC. Suppression was defined by the proportions of conventional CD4 T cells remaining in an undivided state. *Foxp3+ Treg-mediated suppression of conventional CD4 T cell proliferative responses significantly greater (p < 0.05, Student t test) in the presence of NSG than NOD-scid APC. (C) NSG or NOD-scid mice (n = 3 per group) were infused either with 5 × 106 NOD.Foxp3-eGFP splenocytes alone or (D) in combination with 2.5 × 106 NOD-AI4 splenocytes. At 10 d posttransfer, the mean fluorescence intensity (MFI) of GITR-specific Ab staining by Foxp3+ Tregs as well as CD4 and CD8 conventional T cells within the PLN of each recipient type was determined. Data represent mean ± SEM. *p < 0.05, ****p < 0.00001, two-way ANOVA and Bonferroni posttest.

Close modal

We also evaluated whether the greater suppressor activity of Tregs detected in the presence of NSG than NOD-scid–derived myeloid APC could be related to a differing distribution of dendritic cell (DC) subsets in these two strains. Total numbers of splenic myeloid cells were significantly less in NSG than NOD-scid mice (Fig. 7A). After normalizing the data to 4 × 106 total splenocytes (approximate mean level in NSG mice), analyses of the most common DC subsets and macrophages (19) by flow cytometry did not detect any significant proportional differences between the stocks (Fig. 7B). Furthermore, residual myeloid cells in NSG and NOD-scid mice did not differ in baseline expression levels of T cell costimulatory (CD80, CD86) or inhibitory (PD-L1, CD70) molecules (data not shown). There were also no differences in expression of these molecules on APC from the two stocks cultured with activated T cells in vitro. These collective data indicate the greater induction of Treg activity in the presence of NSG than NOD-scid–derived myeloid cells cannot be attributed to the dominance in the former strain of any specific APC subset or to their variable expression of the analyzed T cell costimulatory or inhibitory molecules.

FIGURE 7.

Lower numbers of residual myeloid cells in NSG than NOD-scid mice do not correlate with proportional strain differences in common APC subsets. Collagenase D–digested splenocytes obtained from 7-wk-old NOD-scid and NSG mice were counted and stained with fluorochrome-conjugated Abs for flow cytometric analyses of common APC subsets. (A) Data represent the total number of myeloid residual spleen cells (mean ± SEM, n = 11 each strain). ***p < 0.0001, Student t test. (B) APC subsets from spleen of NOD-scid and NSG mice were analyzed by flow cytometry. Live single cells were gated on CD45.1+CD3CD19 myeloid cells and were subset classified based on expression of the following markers: cDC CD8+, CD11bCD11chighCD8a+; cDC CD4+, CD11b+CD11chighCD4+; cDC DN, CD11b+CD11chighCD8aCD4; Tip DC, CD11b+CD11cintCD4CD8a; pDC, CD11bCD11cint PDCA-1+; Mac, CD11b+CD11clow. Values are normalized to the ∼4 × 106 cells present in NSG spleens. Data represent mean ± SEM, n = 11 each strain. ns, p > 0.05; two-way ANOVA and Bonferroni posttest.

FIGURE 7.

Lower numbers of residual myeloid cells in NSG than NOD-scid mice do not correlate with proportional strain differences in common APC subsets. Collagenase D–digested splenocytes obtained from 7-wk-old NOD-scid and NSG mice were counted and stained with fluorochrome-conjugated Abs for flow cytometric analyses of common APC subsets. (A) Data represent the total number of myeloid residual spleen cells (mean ± SEM, n = 11 each strain). ***p < 0.0001, Student t test. (B) APC subsets from spleen of NOD-scid and NSG mice were analyzed by flow cytometry. Live single cells were gated on CD45.1+CD3CD19 myeloid cells and were subset classified based on expression of the following markers: cDC CD8+, CD11bCD11chighCD8a+; cDC CD4+, CD11b+CD11chighCD4+; cDC DN, CD11b+CD11chighCD8aCD4; Tip DC, CD11b+CD11cintCD4CD8a; pDC, CD11bCD11cint PDCA-1+; Mac, CD11b+CD11clow. Values are normalized to the ∼4 × 106 cells present in NSG spleens. Data represent mean ± SEM, n = 11 each strain. ns, p > 0.05; two-way ANOVA and Bonferroni posttest.

Close modal

To analyze whether the accelerated T1D development in NSG mice mediated by adoptively transferred monoclonal AI4 CD8 T cells was a phenomenon restricted to this clone, we tested the capacity of other NOD-derived MHC class I– and class II–restricted β cell–autoreactive TCR transgenic T cells to induce disease in such recipients. NOD-NY8.3 mice transgenically express the TCR of a potent diabetogenic CD8 T cell clone specific for a peptide derived from islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP206–214) (20). Because the used NOD-NY8.3 stock does not also carry the scid or a rag gene knockout mutation, it produces NY8.3 CD8 T cells as an oligoclonal rather than a monoclonal population. Indeed, the residual CD4 T cells present in NOD-NY8.3 mice have been reported to enhance the diabetogenic activity of the NY8.3 effectors (10). Similar to the case with monoclonal AI4 effectors, oligoclonal NY8.3 CD8 T cells transferred T1D significantly more rapidly to NSG than NOD-scid recipients (Fig. 8A). Analyses of PBL at 3 wk posttransfer in mice receiving NOD-NY8.3 splenocytes showed a greater proportion of total and tetramer-specific NY8.3 CD8+ T cells in NSG than NOD-scid recipients (Fig. 8B). Conversely, the proportion of NOD-NY8.3 donor-derived residual CD4+ T cells was higher in NOD-scid than NSG recipients (Fig. 8B). Although CD4+ Foxp3+ Tregs are phenotypically present in NOD-NY8.3 mice (∼20% of total residual CD4+ T cells in spleen), their function is clearly not sufficient to suppress the diabetogenic activity of NY8.3 effectors in the donors. The residual phenotypic Tregs present in the NY8.3 donor cell inoculum engrafted at higher levels in NOD-scid than NSG recipients (Fig. 8C). The potential importance of this engraftment difference is questionable because we also found that, when depleted of Tregs, the NY8.3 donor cell inoculum continued to more readily induce T1D in NSG than NOD-scid recipients (data not shown). However, similar to the case for monoclonal AI4 cells, the higher engraftment levels of oligoclonal NY8.3 effectors in NSG than NOD-scid recipients are most likely an important factor for the more aggressive T1D development in the former host environment.

FIGURE 8.

Oligoclonal β cell–autoreactive CD8 and CD4 T cells can induce T1D in NSG mice. Total splenocytes from recently diabetic NOD-NY8.3 females were adjusted to inject i.v. 2.5 × 105 NY8.3 CD8 T cells into 8- to 10-wk-old NOD-scid or NSG females. (A) Diabetes incidence study showing representative results from three experiments. PBL from mice receiving NY8.3 T cells were analyzed by FACS at 3 wk posttransfer, for presence of (B) CD4+, CD8+, tetramer-specific NY8.3 CD8 T cells and (C) for presence of CD4+Foxp3+ T cells. A total of 1.5 × 106 purified splenic CD4+ T cells from NOD-BDC2.5 mice was injected i.v. into NOD-scid or NSG mice. (D) Mice were monitored for diabetes development. (E) At 2 wk posttransfer, the proportion of circulating Foxp3+ Tregs was assessed by FACS in PBL isolated from mice receiving CD4+ T cells from NOD-BDC2.5 donors. (F) A total of 2 × 105 purified CD4+CD25 T cells from NOD-BDC2.5 donors was injected i.v. into NOD-scid (n = 10) or NSG (n = 10) mice; diabetes incidence is shown. Data represent mean ± SEM. **p < 0.001, ***p < 0.0001, ****p < 0.00001, nonparametric t test (Mann–Whitney).

FIGURE 8.

Oligoclonal β cell–autoreactive CD8 and CD4 T cells can induce T1D in NSG mice. Total splenocytes from recently diabetic NOD-NY8.3 females were adjusted to inject i.v. 2.5 × 105 NY8.3 CD8 T cells into 8- to 10-wk-old NOD-scid or NSG females. (A) Diabetes incidence study showing representative results from three experiments. PBL from mice receiving NY8.3 T cells were analyzed by FACS at 3 wk posttransfer, for presence of (B) CD4+, CD8+, tetramer-specific NY8.3 CD8 T cells and (C) for presence of CD4+Foxp3+ T cells. A total of 1.5 × 106 purified splenic CD4+ T cells from NOD-BDC2.5 mice was injected i.v. into NOD-scid or NSG mice. (D) Mice were monitored for diabetes development. (E) At 2 wk posttransfer, the proportion of circulating Foxp3+ Tregs was assessed by FACS in PBL isolated from mice receiving CD4+ T cells from NOD-BDC2.5 donors. (F) A total of 2 × 105 purified CD4+CD25 T cells from NOD-BDC2.5 donors was injected i.v. into NOD-scid (n = 10) or NSG (n = 10) mice; diabetes incidence is shown. Data represent mean ± SEM. **p < 0.001, ***p < 0.0001, ****p < 0.00001, nonparametric t test (Mann–Whitney).

Close modal

We also tested whether an oligoclonal population of diabetogenic CD4 T cells differed in their capacity to transfer disease to NSG and NOD-scid recipients. Indeed, when transferred as an oligoclonal population, autoreactive BDC2.5 TCR transgenic CD4+ T cells induced T1D more efficiently in NSG than NOD-scid recipients (Fig. 8D). It should be noted that NOD-BDC2.5 mice producing such CD4 effectors as an oligoclonal population have a very low T1D incidence (21, 22). By contrast, when residual nontransgenic T cells (including functional CD4+Foxp3+ Tregs) are eliminated in NOD-BDC2.5 mice by introduction of the scid mutation or a rag gene knockout, incidence is increased to 100% by 8 wk of age (22, 23). The lower incidence of T1D in NOD-scid than NSG recipients infused with oligoclonal BDC2.5 T cells was also associated with higher engraftment levels at 2 wk posttransfer by residual donor Tregs in the former host environment (Fig. 8E). Thus, we tested whether such differential engraftment and/or functional activity of cotransferred residual donor Tregs could account for the varying ability of BDC2.5 T cells introduced as an oligoclonal population to induce T1D in NSG and NOD-scid recipients. This was found to be the case because purified CD4+CD25 T cells (hence lacking Tregs) from NOD-BDC2.5 donors rapidly transferred T1D to both NSG and NOD-scid recipients (Fig. 8F). In this experiment, we did not find in either recipient type any donor CD4+ T cells expressing the BDC2.5 TCR that had converted to a Foxp3+ Treg phenotype by 2 wk posttransfer (data not shown).

Introducing an inactivated Il2rg gene into NOD-scid (designated NSG mice) greatly enhanced their ability to be engrafted with human immune cells (5). However, engraftment with various patient-derived immune cell populations has not yet induced overt T1D in NSG-based mouse strains, including HLA-humanized versions. These previous studies led us to hypothesize that, because in NSG recipients IL-2rγ–dependent cytokine-signaling events can only occur in donor-type cells, this may influence the activities of transferred diabetogenic T cells in ways that differ depending on whether such effectors were introduced as a monoclonal population versus being part of an oligoclonal or polyclonal repertoire.

To initially test the above hypothesis, we determined whether the previously well-documented ability of total standard NOD splenocytes or a monoclonal TCR transgenic β cell–autoreactive CD8 T cell clonotype (AI4) to transfer T1D to NOD-scid recipients was recapitulated in NSG hosts. As expected, adoptively transferred β cell–autoreactive AI4 monoclonal CD8 T cells elicited T1D development in NOD-scid controls, but interestingly engrafted at higher levels and induced more aggressive disease onset in NSG recipients. Unlike NSG mice, the NOD-scid–modified NOG stock expresses receptors that can bind IL-2rγ–dependent cytokines, but does not enable them to induce signaling responses. Nonetheless, despite possible differences in host cell–mediated competition for binding of IL-2rγ–dependent cytokines, T1D transfer kinetics by monoclonal AI4 T cells in NOG and NSG were comparable. Thus, diminished receptor-mediated competition by host cells for IL-2rγ–dependent cytokines that we demonstrated would occur in NSG versus NOD-scid mice is unlikely to solely account for why adoptively transferred monoclonal AI4 CD8 T cells induce a more aggressive T1D onset in the former recipients.

Diametrically opposite results were obtained when NOD diabetogenic T cells were introduced into NOD-scid and NSG recipients as a broad polyclonal repertoire. In this case, polyclonal NOD splenocytes induced a significantly lower level of T1D in NSG than NOD-scid recipients. Interestingly, cotransfer of total NOD splenocytes inhibited the ability of monoclonal AI4 T cells to induce T1D in NSG recipients. This inhibitory activity was ultimately traced to enhanced activity, but not numbers, of CD4+Foxp3+ Tregs present in polyclonal NOD splenocytes upon transfer into NSG recipients. Indeed, when first depleted of Tregs, NOD splenocytes on their own efficiently transferred T1D to NSG recipients. The question then becomes why the functional activity of Tregs resident among NOD polyclonal T cells populations only reaches a T1D-protective level when transferred into NSG recipients. Again, the most likely explanation lies in the fact that, in NSG recipients, IL-2rγ–dependent cytokine-signaling events only occur in donor cells. The survival and activities of Tregs are highly dependent on IL-2 (17). Thus, the lack of host cell–mediated receptor competition in NSG recipients, demonstrated in part by our current studies, may enhance IL-2 availability to Tregs resident among transferred NOD splenocytes and increase their functional activity to T1D inhibitory levels. This possibility is further supported by previous reports that IL-2 treatment inhibits T1D development in NOD mice by boosting Treg activity (26, 27). The lack of host cell–mediated receptor competition may also enable other IL-2rγ–dependent cytokines to contribute to enhanced donor Treg activity in NSG recipients. Indeed, a recent study indicated that, at least under some circumstances, IL-15 contributes to the accumulation of Tregs (28). Furthermore, it has been reported that IL-21–dependent signaling events are required for the normal ability of NOD APC to exert diabetogenic activity (29). Regardless of the particular cytokine(s) involved, our current results indicate that NOD APC in which IL-2rγ–dependent signaling responses do not occur may be preferentially skewed to a state promoting the maintenance of T1D-protective immune tolerance through an enhanced capacity to elicit Treg activity. Our studies did not find significant recipient strain differences in the proportions of common DC and macrophage subsets or their expression patterns of analyzed T cell costimulatory or inhibitory molecules that could account for the greater ability of NSG-derived APC to support Treg activity. However, we did find that GITR was expressed at significantly higher levels by NOD donor Tregs, but not conventional T cells, when interacting with myeloid APC in NSG than NOD-scid recipients. Previous studies (18) found at least under some conditions that increases in GITR expression and activity can enhance the suppressive activity of Tregs.

Our results utilizing the AI4 system can be extended to other diabetogenic TCR transgenic models such as NOD-NY8.3 and NOD-BDC2.5. We are aware that our results indicating oligoclonal NY8.3 CD8 T cells efficiently transfer T1D to NSG mice contradict an earlier report that such donor cells poorly induce disease in the same recipients, which was attributed to a lack of IL-15 signaling in host APC (30). Although the true basis for this discrepancy remains unknown, it might be attributable to a varying dose of transferred NY8.3 splenocytes that was not detailed in this previous study, or differing environmental conditions. Regardless, our results indicate that in order for a range of diabetogenic T cells to induce disease in NSG recipients, they must either be introduced as monoclonal or oligoclonal populations or, if part of a polyclonal repertoire, the transfer inoculum must be first purged of Tregs. Another consideration is that, whereas NK cells remain present in NOD-scid mice, they are absent in the NSG strain (5). Hence, the augmented ability of monoclonal or oligoclonal diabetogenic T cells to induce disease onset in NSG recipients is a NK cell–independent process.

In conclusion, our findings reveal some important considerations that may be applicable when contemplating the possible use of NSG mice as recipients to screen for the presence of diabetogenic T cells resident within various human donor cell inocula, or as a resource to develop means to attenuate such effectors. Our results indicate that NSG-based models may be good recipients for studies entailing the transfer of β cell–autoreactive T cell clones or lines from human donors. Other results indicate that if the goal is to maximize expansion of a transferred clonal population of diabetogenic T cells in murine recipients, then NSG may be preferable to NOG mice for this purpose. Finally, the current results indicate that, when considering the use of NSG recipients to test for the presence of diabetogenic effectors among polyclonal or even some oligoclonal populations of human T cells, Tregs should first be removed from the transfer inoculum to eliminate false-negative results.

This work was supported by National Institutes of Health Grants DK46266 and DK95735 (to D.V.S.), DK77443 (to Y.-G.C.), AI46629-13 (to L.D.S., D.L.G., and M.A.B.), DK089572 (to L.D.S. and D.L.G.), and AI112321 (to D.V.S., D.L.G., and M.A.B.); grants from the Juvenile Diabetes Research Foundation (to D.V.S. and M.P.) and American Diabetes Association (to D.V.S.); and Helmsley Charitable Trust Grants 2014PG-T1D048 (to D.V.S.) and 2012PG-T1D018 (to L.D.S., M.A.B., and D.L.G.).

Abbreviations used in this article:

DC

dendritic cell

eGFP

enhanced GFP

PLN

pancreatic lymph node

T1D

type 1 diabetes

Treg

regulatory T cell.

1
Atkinson
M. A.
,
Eisenbarth
G. S.
,
Michels
A. W.
.
2014
.
Type 1 diabetes.
Lancet
383
:
69
82
.
2
Driver
J. P.
,
Serreze
D. V.
,
Chen
Y.-G.
.
2011
.
Mouse models for the study of autoimmune type 1 diabetes: a NOD to similarities and differences to human disease.
Semin. Immunopathol.
33
:
67
87
.
3
Herold
K. C.
,
Vignali
D. A.
,
Cooke
A.
,
Bluestone
J. A.
.
2013
.
Type 1 diabetes: translating mechanistic observations into effective clinical outcomes.
Nat. Rev. Immunol.
13
:
243
256
.
4
Shultz
L. D.
,
Brehm
M. A.
,
Bavari
S.
,
Greiner
D. L.
.
2011
.
Humanized mice as a preclinical tool for infectious disease and biomedical research.
Ann. N. Y. Acad. Sci.
1245
:
50
54
.
5
Shultz
L. D.
,
Brehm
M. A.
,
Garcia-Martinez
J. V.
,
Greiner
D. L.
.
2012
.
Humanized mice for immune system investigation: progress, promise and challenges.
Nat. Rev. Immunol.
12
:
786
798
.
6
Whitfield-Larry
F.
,
Young
E. F.
,
Talmage
G.
,
Fudge
E.
,
Azam
A.
,
Patel
S.
,
Largay
J.
,
Byrd
W.
,
Buse
J.
,
Calikoglu
A. S.
, et al
.
2011
.
HLA-A2-matched peripheral blood mononuclear cells from type 1 diabetic patients, but not nondiabetic donors, transfer insulitis to NOD-scid/γc(null)/HLA-A2 transgenic mice concurrent with the expansion of islet-specific CD8+ T cells.
Diabetes
60
:
1726
1733
.
7
Unger
W. W. J.
,
Pearson
T.
,
Abreu
J. R. F.
,
Laban
S.
,
van der Slik
A. R.
,
der Kracht
S. M.
,
Kester
M. G. D.
,
Serreze
D. V.
,
Shultz
L. D.
,
Griffioen
M.
, et al
.
2012
.
Islet-specific CTL cloned from a type 1 diabetes patient cause beta-cell destruction after engraftment into HLA-A2 transgenic NOD/scid/IL2RG null mice.
PLoS One
7
:
e49213
.
8
Viehmann Milam
A. A.
,
Maher
S. E.
,
Gibson
J. A.
,
Lebastchi
J.
,
Wen
L.
,
Ruddle
N. H.
,
Herold
K. C.
,
Bothwell
A. L. M.
.
2014
.
A humanized mouse model of autoimmune insulitis.
Diabetes
63
:
1712
1724
.
9
DiLorenzo
T. P.
,
Graser
R. T.
,
Ono
T.
,
Christianson
G. J.
,
Chapman
H. D.
,
Roopenian
D. C.
,
Nathenson
S. G.
,
Serreze
D. V.
.
1998
.
Major histocompatibility complex class I-restricted T cells are required for all but the end stages of diabetes development in nonobese diabetic mice and use a prevalent T cell receptor α chain gene rearrangement.
Proc. Natl. Acad. Sci. USA
95
:
12538
12543
.
10
Verdaguer
J.
,
Schmidt
D.
,
Amrani
A.
,
Anderson
B.
,
Averill
N.
,
Santamaria
P.
.
1997
.
Spontaneous autoimmune diabetes in monoclonal T cell nonobese diabetic mice.
J. Exp. Med.
186
:
1663
1676
.
11
Haskins
K.
,
Portas
M.
,
Bradley
B.
,
Wegmann
D.
,
Lafferty
K.
.
1988
.
T-lymphocyte clone specific for pancreatic islet antigen.
Diabetes
37
:
1444
1448
.
12
Haribhai
D.
,
Lin
W.
,
Relland
L. M.
,
Truong
N.
,
Williams
C. B.
,
Chatila
T. A.
.
2007
.
Regulatory T cells dynamically control the primary immune response to foreign antigen.
J. Immunol.
178
:
2961
2972
.
13
Scheuplein
F.
,
Rissiek
B.
,
Driver
J. P.
,
Chen
Y.-G.
,
Koch-Nolte
F.
,
Serreze
D. V.
.
2010
.
A recombinant heavy chain antibody approach blocks ART2 mediated deletion of an iNKT cell population that upon activation inhibits autoimmune diabetes.
J. Autoimmun.
34
:
145
154
.
14
Bourke
E.
,
Cassetti
A.
,
Villa
A.
,
Fadlon
E.
,
Colotta
F.
,
Mantovani
A.
.
2003
.
IL-1 β scavenging by the type II IL-1 decoy receptor in human neutrophils.
J. Immunol.
170
:
5999
6005
.
15
Graser
R. T.
,
DiLorenzo
T. P.
,
Wang
F.
,
Christianson
G. J.
,
Chapman
H. D.
,
Roopenian
D. C.
,
Nathenson
S. G.
,
Serreze
D. V.
.
2000
.
Identification of a CD8 T cell that can independently mediate autoimmune diabetes development in the complete absence of CD4 T cell helper functions.
J. Immunol.
164
:
3913
3918
.
16
Ito
M.
,
Hiramatsu
H.
,
Kobayashi
K.
,
Suzue
K.
,
Kawahata
M.
,
Hioki
K.
,
Ueyama
Y.
,
Koyanagi
Y.
,
Sugamura
K.
,
Tsuji
K.
, et al
.
2002
.
NOD/SCID/γ(c)(null) mouse: an excellent recipient mouse model for engraftment of human cells.
Blood
100
:
3175
3182
.
17
Lourenço
E. V.
,
La Cava
A.
.
2011
.
Natural regulatory T cells in autoimmunity.
Autoimmunity
44
:
33
42
.
18
Walker
L. S. K.
2009
.
Regulatory T cells overturned: the effectors fight back.
Immunology
126
:
466
474
.
19
Geissmann
F.
,
Manz
M. G.
,
Jung
S.
,
Sieweke
M. H.
,
Merad
M.
,
Ley
K.
.
2010
.
Development of monocytes, macrophages, and dendritic cells.
Science
327
:
656
661
.
20
Lieberman
S. M.
,
Evans
A. M.
,
Han
B.
,
Takaki
T.
,
Vinnitskaya
Y.
,
Caldwell
J. A.
,
Serreze
D. V.
,
Shabanowitz
J.
,
Hunt
D. F.
,
Nathenson
S. G.
, et al
.
2003
.
Identification of the β cell antigen targeted by a prevalent population of pathogenic CD8+ T cells in autoimmune diabetes.
Proc. Natl. Acad. Sci. USA
100
:
8384
8388
.
21
Gonzalez
A.
,
Katz
J. D.
,
Mattei
M.-G.
,
Kikutani
H.
,
Benoist
C.
,
Mathis
D.
.
1997
.
Genetic control of diabetes progression.
Immunity
7
:
873
883
.
22
You
S.
,
Slehoffer
G.
,
Barriot
S.
,
Bach
J.-F.
,
Chatenoud
L.
.
2004
.
Unique role of CD4+CD62L+ regulatory T cells in the control of autoimmune diabetes in T cell receptor transgenic mice.
Proc. Natl. Acad. Sci. USA
101
(
Suppl. 2
):
14580
14585
.
23
Thomas
D. C.
,
Mellanby
R. J.
,
Phillips
J. M.
,
Cooke
A.
.
2007
.
An early age-related increase in the frequency of CD4+ Foxp3+ cells in BDC2.5NOD mice.
Immunology
121
:
565
576
.
24
von Herrath
M.
,
Nepom
G. T.
.
2009
.
Remodeling rodent models to mimic human type 1 diabetes.
Eur. J. Immunol.
39
:
2049
2054
.
25
Van Belle
T. L.
,
Taylor
P.
,
von Herrath
M. G.
.
2009
.
Mouse models for type 1 diabetes.
Drug Discov. Today Dis. Models
6
:
41
45
.
26
Serreze
D. V.
,
Hamaguchi
K.
,
Leiter
E. H.
.
1989
.
Immunostimulation circumvents diabetes in NOD/Lt mice.
J. Autoimmun.
2
:
759
776
.
27
Grinberg-Bleyer
Y.
,
Baeyens
A.
,
You
S.
,
Elhage
R.
,
Fourcade
G.
,
Gregoire
S.
,
Cagnard
N.
,
Carpentier
W.
,
Tang
Q.
,
Bluestone
J.
, et al
.
2010
.
IL-2 reverses established type 1 diabetes in NOD mice by a local effect on pancreatic regulatory T cells.
J. Exp. Med.
207
:
1871
1878
.
28
Raynor
J.
,
Sholl
A.
,
Plas
D. R.
,
Bouillet
P.
,
Chougnet
C. A.
,
Hildeman
D. A.
.
2013
.
IL-15 fosters age-driven regulatory T cell accrual in the face of declining IL-2 levels.
Front. Immunol.
4
:
161
. .
29
Liu
S. M.
,
Lee
D. H.
,
Sullivan
J. M.
,
Chung
D.
,
Jäger
A.
,
Shum
B. O.
,
Sarvetnick
N. E.
,
Anderson
A. C.
,
Kuchroo
V. K.
.
2011
.
Differential IL-21 signaling in APCs leads to disparate Th17 differentiation in diabetes-susceptible NOD and diabetes-resistant NOD.Idd3 mice.
J. Clin. Invest.
121
:
4303
4310
.
30
Bobbala
D.
,
Chen
X. L.
,
Leblanc
C.
,
Mayhue
M.
,
Stankova
J.
,
Tanaka
T.
,
Chen
Y. G.
,
Ilangumaran
S.
,
Ramanathan
S.
.
2012
.
Interleukin-15 plays an essential role in the pathogenesis of autoimmune diabetes in the NOD mouse.
Diabetologia
55
:
3010
3020
.

The authors have no financial conflicts of interest.