Granzyme B is dispensable in the development of diabetes in non-obese diabetic mice

Abstract

Pancreatic beta cell destruction in type 1 diabetes is mediated by cytotoxic CD8(+) T lymphoctyes (CTL). Granzyme B is an effector molecule used by CTL to kill target cells. We previously showed that granzyme B-deficient allogeneic CTL inefficiently killed pancreatic islets in vitro. We generated granzyme B-deficient non-obese diabetic (NOD) mice to test whether granzyme B is an important effector molecule in spontaneous type 1 diabetes. Granzyme B-deficient islet antigen-specific CD8(+) T cells had impaired homing into islets of young mice. Insulitis was reduced in granzyme B-deficient mice at 70 days of age (insulitis score 0.043±0.019 in granzyme B-deficient versus 0.139±0.034 in wild-type NOD mice p<0.05), but was similar to wild-type at 100 and 150 days of age. We observed a reduced frequency of CD3(+)CD8(+) T cells in the islets and peripheral lymphoid tissues of granzyme B-deficient mice (p<0.005 and p<0.0001 respectively), but there was no difference in cell proportions in the thymus. Antigen-specific CTL developed normally in granzyme B-deficient mice, and were able to kill NOD islet target cells as efficiently as wild-type CTL in vitro. The incidence of spontaneous diabetes in granzyme B-deficient mice was the same as wild-type NOD mice. We observed a delayed onset of diabetes in granzyme B-deficient CD8-dependent NOD8.3 mice (median onset 102.5 days in granzyme B-deficient versus 57.50 days in wild-type NOD8.3 mice), which may be due to the delayed onset of insulitis or inefficient priming at an earlier age in this accelerated model of diabetes. Our data indicate that granzyme B is dispensable for beta cell destruction in type 1 diabetes, but is required for efficient early activation of CTL.

Figure 1. Reduced proportion of CD8⁺ T cells in granzyme B-deficient pancreatic lymph nodes.

(A) Total cell counts in ILN, PLN or spleens (n = 4–7, age 70–100 days) from NOD or granzyme B^−/− (GrzB) mice. ns = not significant. (B) Pancreatic lymph node cells from 70–100 day old NOD or granzyme B^−/− (GrzB) mice were stained with antibodies to CD3 (n = 18), CD4 and CD8 (n = 13). CD3⁺ T cells were gated on CD4 and CD8 to determine the percentage of cells. No difference was observed in the overall percentage of CD3⁺ T cells. The percentage of CD8⁺ T cells was significantly reduced in granzyme B^−/− mice compared to NOD (***p = <0.0001), and the percentage of CD4⁺ T cells was increased (***p = <0.0001). Data analyzed by unpaired Student’s t test. (C) Absolute numbers of cells were determined. There was no difference in total CD3⁺ T cells in the PLN of 70–100 day old NOD or granzyme B^−/− mice (n = 7 mice). The number of CD4⁺ T cells was increased (n = 7 mice, *p = 0.0442) and the number of CD8⁺ T cells was reduced (n = 7 mice, *p = 0.0421) in the PLN of granzyme B^−/− mice. ns = not significant. Data analyzed by unpaired Student’s t test. (D) Percentage of CD4⁺CD8⁻ and CD8⁺CD4⁻ T cells in CD3 gated thymocytes in 70 day old female NOD or granzyme B^−/− mice (n = 6 mice performed over 3 independent experiments). ns = not significant. (E) The percentage of CD8⁺ T cells expressing CD44 was determined by flow cytometry at 100 days of age (n = 5). ns = not significant. (F) BrdU incorporation into thymic CD4⁺CD8⁻ and CD8⁺CD4⁻ T cells, and CD4⁺ and CD8⁺ T cells from PLN of 4-week-old NOD and granzyme B^−/− mice (n = 3 mice). Percentage of CD8⁺BrdU⁺ cells in PLN was significantly reduced in granzyme B^−/− compared to NOD (**p = 0.0012). ns = not significant. Data analyzed by unpaired Student’s t test. Horizontal bars in A-F show mean±SEM.

Figure 2. Antigen specific proliferation of granzyme B-deficient CD8⁺ T cells is reduced.

(A, B) NOD8.3, GrzB^−/−NOD8.3 or Pfp^−/−NOD8.3 splenic CD8⁺ T cells were labeled with CFSE and transferred into 70- or 100-day old female NOD or granzyme B^−/− (GrzB) mice. Six days later ILN and PLN were harvested and analyzed for the proliferation of CFSE labeled cells. Data represent mean±SEM of n = 5–10 recipients for NOD8.3 or GrzB^−/−NOD8.3, n = 4 for Pfp^−/−NOD8.3, performed as individual mice over at least three independent experiments. Statistical significance, *p = 0.0183 (NOD8.3 vs GrzB^−/−NOD8.3 in granzyme B^−/− PLN) and **p = 0.0315 (GrzB^−/−NOD8.3 in NOD vs granzyme B^−/− PLN). ns = not significant. Data analyzed by unpaired Student’s t test. (C) Dot plots showing proliferation of CFSE labeled NOD8.3 (left panels) or GrzB^−/−NOD8.3 (right panels) T cells in the islets of 70 day-old female NOD (top panels) or granzyme B^−/− (bottom panels) mice. The box in each dot plot shows the number of CD8⁺CFSE⁺ cells that has migrated into the islets. Data are representative of n = 3 experiments. (D) Percentage of CFSE labeled CD8⁺ NOD8.3 or GrzB^−/−NOD8.3 T cells that have undergone division in the islets of 70 day old NOD or granzyme B^−/− recipients (n = 3). No significant difference was observed.

Figure 3. Granzyme B deficient NOD mice have reduced insulitis at 70 days of age, which is not maintained at 100 days.

(A) Pancreata from female NOD and granzyme B^−/− (GrzB) mice were harvested at 70 days of age (n = 10, total islets scored: NOD = 1004 islets, Granzyme B^−/− = 1443 islets), 100 days of age (n = 5–7, total islets scored: NOD = 572 islets, Granzyme B^−/− = 548 islets) and 150 days of age (n = 10, total islets scored: NOD = 688 islets, Granzyme B^−/− = 1079 islets) and scored for insulitis on frozen sections stained for insulin. The data are the mean of the percentage of islets with each score. (B) For each mouse the insulitis score was calculated as described in material and methods, and data are the mean±SEM insulitis score. Insulitis is significantly reduced in 70 day old granzyme B^−/− mice compared to NOD (*p = 0.0274), but not at 100 days or 150 days. Data analyzed by unpaired Student’s t test. ns = not significant. (C) Flow cytometric analysis of islets from female NOD and granzyme B^−/− (GrzB) mice at 70 days of age. Live cells were gated using propidium iodide exclusion. CD45 is the proportion of live cells expressing CD45. CD3 is the proportion of live CD45⁺ cells expressing CD3. The proportion of CD4⁺ and CD8⁺ cells was determined as the percentage of live CD45⁺ and CD3⁺ cells. The percentage of CD45⁺ cells (*p = 0.0461), CD3⁺ cells (**p = 0.0087) and CD8⁺ cells (*p = 0.0141) but not CD4⁺ cells was significantly reduced in granzyme B^−/− mice. Mean±SEM for n = 10–15 NOD mice and n = 5–9 GrzB mice is shown. 150 islets were used for each sample. Data analyzed by unpaired Student’s t test. (D) The proportion of live CD45⁺ cells in the islets from 70 day old female mice that were CD11c⁺ or F4/80⁺. The percentage of CD11c⁺ cells (*p = 0.0286), but not F4/80⁺ cells (p = 0.5189) were increased in granzyme B^−/− mice. Data analyzed by unpaired Student’s t test.

Figure 4. Diabetes is unchanged in granzyme B-deficient NOD mice but is delayed in granzyme B-deficient NOD8.3 mice.

(A) The proportion of CD8⁺IGRP-tetramer⁺ T cells in freshly isolated islets from 100 day old female NOD or granzyme B^−/− mice was determined by flow cytometry (n = 3). No significant difference was observed. (B) Activated NOD8.3 or GrzB^−/−NOD8.3 CD8⁺ T cells were cultured for 16 hours with IGRP pulsed and ⁵¹Cr labeled NOD islets at an effector to target ratio of 20∶1. Data are mean±SEM of three independent experiments. No significant difference was observed. (C) A cohort of female NOD (n = 35) and granzyme B^−/− mice (n = 19) mice were observed for 300 days for the incidence of spontaneous diabetes. No difference was observed between the groups. Data analyzed by Log-rank (Mantel-Cox) Test. (D) Pancreata from NOD8.3 or GrzB^−/−NOD8.3 mice were harvested at 50 days of age for insulitis scoring on insulin antibody-stained frozen sections. Total islets scored: NOD8.3 = 596 islets, GrzB^−/− NOD8.3 = 508 islets. (E) For each mouse the insulitis score was calculated using the method as described in material and methods, and data are the mean±SEM insulitis score. Data analyzed by unpaired Student’s t test. ns = not significant. (F) Incidence of diabetes was studied in cohorts of NOD8.3 (n = 12) and GrzB^−/−NOD8.3 (n = 12) mice. The development of diabetes in GrzB^−/−NOD8.3 mice was significantly delayed compared to NOD8.3 (*p = 0.0399). Data analyzed by Log-rank (Mantel-Cox) test.