Protection From Type 1 Diabetes in the Face of High Levels of Activated Autoaggressive Lymphocytes in a Viral Transgenic Mouse Model Crossed to the SV129 Strain

Islets of Langerhans were isolated from mice as previously described (22,29,30). In brief, the common bile duct was cannulated, and the pancreas was distended with 3 ml of Dulbecco’s modified Eagle’s medium containing 1.3 units/ml collagenase P (Boehringer Mannheim, Indianapolis, IN). Pooled pancreases were digested at 37°C for 20 min and were then disrupted by shaking. Islets were purified on Histopaque-1077 density gradients (Sigma, St. Louis, MO). The gradient was centrifuged with gradually increasing speed from 25 to 800g for 4 min then at 800g for 10 min. Islets were aspirated from the media/gradient interface, washed, and hand-picked if necessary. The islets were dispersed into single cells with 0.2% trypsin (Calbiochem, La Jolla, CA) and 10 mmol/l EDTA in Hank’s balanced salt solution (HBSS), and then they were allowed to recover in complete medium (CMRL, 7% fetal calf serum [FCS], 1% penicillin/streptomycin, and 1% glutamine) for 1 h before staining with monoclonal antibodies (mAbs). To ensure the purity of the β-cell preparation, β-cells sorted with a fluorescence-activated cell sorter (FACS) were fixed on immunohistochemistry slides and stained for intracellular insulin as previously described (31). The antibodies used were a polyclonal guinea pig anti-insulin antibody (Dako, Carpinteria, CA) as primary antibody and a biotinylated goat anti–guinea pig IgG antibody (Vector Laboratories, Burlingame, CA) as secondary antibody. Of the isolated islet cells, 56% were β-cells on average, as evidenced by positive insulin stain, and >98% of the cells in the final-gated autofluorescent population were β-cells.

Whole islets were trypsinized, and islet cells were counted. To stain class I MHCs, we used 28-14-8 (Pharmingen, La Jolla, CA), a biotinylated mouse IgG2a mAb directed against murine H-2D^b that only cross-reacts with H-2L^d and H-2D^q. Secondary antibodies were as follows: phycoerythrin-conjugated goat F(ab′)2 anti-rat IgG(H+L) (Caltag Labs, Burlingame, CA) and phycoerythrin-conjugated streptavidin (Caltag Labs). Islet cell or spleen cell suspensions were incubated for 20 min on ice with mAb diluted in Dulbecco’s phosphate-buffered saline (DPBS) containing 1% FCS. They were then washed and incubated for 15 min with a secondary antibody as necessary. Cells were finally washed and resuspended in DPBS with 1% FCS and 2.5 μg/ml propidium iodide to stain dead cells. Analysis was performed on a FACSort or FACScalibur flow cytometer (Becton Dickinson, San Jose, CA) using Cell Quest software (Becton Dickinson). β-Cells were sorted on a FACStar^plus (Becton Dickinson) based on flavin adenine dinucleotide auto-fluorescence according to the method previously described (32). Background levels were determined by staining MHC class I–deficient islets from β2m^−/− mice.

CTL activity was measured in a 5- to 6-h in vitro ⁵¹Cr release assay (33,34). Briefly, to determine the amount of MHC-restricted CTL lysis, syngeneic or allogeneic target cells were either infected with LCMV-strain ARMSTRONG (ARM) (multiplicity of infection = 1), or uninfected cells were coated with LCMV peptides GP 33-41 (KAVYNFATC), GP amino acid 276-286 (SGVENPGGYCL), and nucleoprotein (NP) 396-404 (FQPQNGQFI), all of which were H-2D^b–restricted (27). Assays used splenic lymphocytes harvested 7 days after primary LCMV infection (10⁵ pfu i.p.) at effector-to-target ratios of 50:1, 25:1, and 12:1 or CTL clones and secondary CTL lines at ratios of 10:1 and 5:1. To determine CTL activity after secondary stimulation, spleen cells harvested from mice 30–180 days after primary inoculation with 1 × 10⁵ pfu LCMV i.p. were incubated with MHC-matched, irradiated, LCMV-infected, or peptide-coated (10⁻⁵ mol/l) macrophages in the presence of T-cell growth factor (supernatant from concanavalin A–stimulated rat splenocytes) containing interleukin-2 (IL-2) and irradiated syngeneic spleen feeder cells for 5–12 days (34). The MC57 (H-2K^bD^b) and BALB/Cl7 (H-2^d) cells used as CTL targets were grown as previously reported (34). When islet cells were used as targets, whole islets were isolated from collagenase-digested pancreases of at least three mice and labeled for 1 h at 37°C in the presence of 1 mCi of ⁵⁰Cr. After washing three times in glucose-free HBSS (Gibco, Gaithersburg, MD), islets were dispersed by trypsinization, washed again three times, and then plated at 1–2 × 10⁴ cells/well in 96-well flat-bottom plates for the cytotoxicity assay. In some assays, cytokines (IFN-γ, tumor necrosis factor-α [TNF-α], and IL-1β) were added directly to the target cells at 1 μg/ml (22). Where indicated, cytotoxicity assays were carried out over an extended time period of 20 h. Precursor frequencies of LCMV-specific CTL were determined as described (28).

Generation and characterization of RIP-LCMV transgenic mice that develop type 1 diabetes after LCMV infection has been previously described (21,34). For this report, we used only RIP-GP 34-20 (H-2D^b) transgenic mice, which express viral GP from the LCMV strain ARMSTRONG only in the β-cells of their islets and not in any other organs (34). SV129 and BALB/b mice were obtained from Jackson Laboratories (Bar Harbor, ME) (14). NOD mice were obtained from the rodent breeding colony at the Scripps Research Institute. F₁ crosses were obtained in a single mating of RIP-GP mice with SV129, BALB/b, or NOD mice. Backcrosses to the C57BL/6 up to N₈ were generated from a single RIP-GPX SV129 progenitor pair. Note that RIP-LCMV-NP mice develop diabetes on both B6 (H-2^b) and BALB/c (H-2^d) backgrounds, because the CTL response to LCMV NP has both a dominant H-2^d–restricted (RPQASGVYM) and an H-2^b–restricted (FQPQNGQFI) epitope.

Virus stocks consisted of the LCMV strain ARMSTRONG (clone 53b). LCMV was plaque-purified three times on Vero cells, and stocks were prepared by a single passage on BHK-21 cells (35).

Blood samples were obtained from the retro-orbital plexus of mice, and plasma glucose concentration was determined using Accucheck III (Boehringer Mannheim). Mice with blood glucose values >300 mg/dl were considered diabetic (36).

Tissues taken for histological analysis were fixed in 10% zinc formalin and stained with hematoxylin and eosin. Immunochemical studies were carried out on 6- to 10-μm freshly frozen cryomicrotome sections using immunostaining of islets to detect expression of MHC class I and II, D^b, insulin, CD4, CD8, B220, and F4/80. Primary antibodies were applied for 1 h. These consisted of rat anti-mouse CD4 (clone RM 4-5), anti-CD8 (clone 53-6.7), anti-B220 (clone RA3 6B2), anti-F4/80 (clone A3-1), anti–MAC-1 (clone M 1/70), anti–MHC class I (clone M 1/42), and anti–class II (clone M5/114) (Pharmingen, San Diego, CA, and Boehringer Mannheim). After washing in phosphate-buffered saline, the secondary antibody (biotinylated goat anti-rat or anti-mouse IgG [Vector Laboratories]) was applied for 1 h. Color reaction was developed with sequential treatment using avidin–horseradish peroxidase conjugate (Boehringer-Mannheim, La Jolla, CA) and diaminobenzidine–hydrogen peroxide.

Double-staining for insulin and 5′-bromo-2′-deoxyuridine (BrDU) (Fig. 5) was performed by first staining freshly thawed pancreatic sections for insulin using a primary guinea pig anti insulin antibody (DACO) and biotinylated anti guinea pig secondary antibody combined with an avidin-alkaline-phosphatase conjugate and Vector-red staining kit (both from Vector Laboratories). Thereafter, staining for BrDU was performed as described using the Pharmingen BrDU staining kit, using rat anti-BrDU as primary antibody to avoid cross-reactivity with the insulin staining and a peroxidase-coupled secondary antibody with 3,3′- diaminobenzidine tetrahydrochloride as a color substrate (brown).

Staining of cell surface antigen and intracellular antigens was performed as described. Virus-specific stimulation was provided by the addition of 1 μg/μl of class I and 2 μg/μl of class II LCMV-CD4 or -CD8 peptides (27,37,38,39) in the presence of 50 units/ml recombinant human IL-2. All stimulation cultures contained 1 μg/ml brefeldin A (BFA B7651; Sigma) to block protein transport into post-Golgi compartments and to allow cytokines to accumulate within cells. Negative controls were stained with cytokine-specific phycoethrin-conjugated antibodies preincubated for 30 min at 4°C with an excess of recombinant cytokine. Cells were acquired and analyzed on a FACSort or FACScalibur flow cytometer (Beckton Dickinson, San Jose, CA) using Cell Quest software (Beckton Dickinson).

For the studies displayed in Fig. 5, groups of 10 C57BL/6 or SV129 H-2^b males were injected with 60 mg/kg body wt i.p. streptozodozin three times, which was dissolved immediately before injection in 100 mmol/l citrate buffer at pH 4.2. BrDU was provided in the drinking water at 0.8 mg/ml continuously.

Because SV129 embryonic stem cells are preferably used to generate targeted gene knockout mice, our intention was to evaluate whether the genetic background of this strain could influence the development of autoimmune disease per se. For the first experiment, RIP-LCMV-GP mice that develop diabetes dependent on H-2D^b–restricted cytotoxic CD8 lymphocytes but independent from CD4 cells (34) were crossed to various H-2D^b mouse strains, and the incidence of hyperglycemia was tracked after LCMV infection. As shown in Fig. 1, only the SV129 genetic constellation appeared to prevent disease development over up to 10 backcrosses to the original RIP-LCMV-GP C57BL/6 background (data shown for N₈ but not N₁₀). In contrast, a high degree of diabetes was observed in crosses to BALB/b and NOD H-2D^b mice. This finding was surprising because all of these mice express the H-2D^b MHC class I allele, to which the vast majority of the H-2D^b LCMV (anti-self) CTL response is restricted (27,40). In addition, because all F₁ offspring will still express 50% of the original C57BL/6 H-2D^b MHC molecules (and in N₈ backcrosses to the original C57BL/6 will express even more), the lower diabetes incidence in SV129 F₁ crosses is likely not caused by MHC-linked factors. We further investigated this issue by precisely determining autoreactive (LCMV-specific) CD8 and CD4 effector functions.

The anti-LCMV CD8 (and CD4) responses can be considered as true autoreactive lymphocytes in RIP-LCMV mice because the transgene is expressed as a pancreatic self-antigen, and mice are tolerant unless systemic LCMV infection is initiated (34). Therefore, the enumeration of these populations according to their effector functions is a very close and accurate measurement of the initial autoreactive islet-specific response in such mice, if lymphocytes can actually enter the islets, which is equivalent between RIP-GP and RIP-GPX SV129 N₈ mice (Table 1). As displayed in Fig. 2, perforin-mediated cytotoxic responses to the LCMV GP (islet antigen expressed transgenically) and NP were reduced 7 days after infection during the primary ex vivo response in SV129 mice compared with C57BL/6 RIP-LCMV mice. Interestingly, this defect was seen neither in the memory response (Fig. 2) nor in the N₈ RIP-LCMV/SV129 mice that had been backcrossed to the original RIP-LCMV-GP C57BL/6 background for eight generations but were still largely protected from diabetes (Figs. 1 and 2). Similarly, the inflammatory autoreactive cytokine response reflected in LCMV-specific IFN-γ production (Fig. 3) was also reduced in SV129 mice but not in the N₈ backcrosses (Fig. 3). Thus, because diabetes was still reduced after eight backcrosses to the original C57BL/6 genetic background (Fig. 1), but the autoreactive CTL response was not impaired compared with the original RIP-LCMV-GP mice (that developed a high degree of diabetes) (Figs. 2 and 3), other genetic, likely non–MHC-linked, factors have to account for the protection from diabetes. Along with the similarities in anti-LCMV CTL in B6- and SV129-crossed mice, viral (LCMV) clearance occurred with similar kinetics and was accomplished by days 8–10 from major organs (data not shown).

We next tested whether islets from SV129 mice might be resistant to destruction by autoreactive perforin-positive CTLs or by inflammatory cytokines in general. Previous studies have implicated both of these pathways in islet destruction (22). As shown in Fig. 4, this is not the case, and islets of SV129 mice are as susceptible as islets from C57BL/6 mice or RIP-LCMVxC57BL/6 N₈ mice (not shown) to CTL killing and cytokine-mediated death. As expected, MHC class I levels on islets from both strains were upregulated to very similar levels after LCMV infection, an event that is essential in sensitizing β-cells to CTL recognition (data not shown) (22). This observation strengthens the notion that genetic factors other than MHC class I could be responsible for protecting SV129-crossed RIP-LCMV mice from autoaggressive CTL and diabetes. We further dissected the underlying mechanism of protection.

Islets from the different inter- and backcrossed RIP-LCMV mice shown in Fig. 1 were evaluated histologically. Double-staining for insulin (red) and BrDu incorporation (brown) was performed during the prediabetic, diabetic, and nondiseased stages and is shown in Fig. 5. Interestingly, signs of β-cell regeneration, as evidenced by double-staining for BrDU as well as insulin, were only observed in the majority of SV129 crosses that had not developed autoimmune diabetes. In contrast, uninfected islets as well as islets from diabetic (Fig. 5) or nondiabetic C57BL/6 controls do not exhibit this phenotype, and β-cell regeneration is not seen. On average, 5 of 9 β-cells double-stained for insulin and BrDU were typically found per islet from the nondiabetic RIP-LCMV × SV129 (C57BL/6 N₈ [Fig. 5B] or F₁ [Fig. 5E]) mice, whereas no BrDU incorporation was observed in any of the diabetic or uninfected normoglycemic controls. Interestingly, low-dose streptozodozin treatment appeared to have a similar effect on islets of SV129 but not C57BL/6 mice, as shown in Fig. 5D. Thus, β-cells appear to generate only under attack in SV129 but not B6 mice, and increasing the islet cell mass by regeneration is likely one of the pathways preventing diabetes in mice crossed to the SV129 background. Indeed, nondiabetic but LCMV-infected RIP-LCMV-GP × SV129 C57BL/6-N₈ mice had close to normal insulin content in the pancreas (data not shown).

As shown in Fig. 6, islet infiltration in nondiabetic RIP-LCMV-GP mice crossed to the SV129 strain and back-crossed for eight generations to the C57BL/6 original genetic background is reduced, and fewer CD4 and CD8 lymphocytes are found at day 14 after the triggering LCMV infection (Table 1). It is difficult to distinguish precisely between cause and effect from such data, but the histological findings at earlier stages post–LCMV infection (Table 1) indicate that lymphocytes can still enter into the islet area of SV129-crossed transgenic mice but are present to a lesser extent inside islets in protected animals compared with diabetic non-SV129 controls (Table 1). A primary defect in lymphocytes trafficking therefore appears unlikely because similar numbers of CD4 and CD8 lymphocytes are able to reach the islets initially after LCMV infection in both mouse strains (Table 1). Furthermore, no change in the production of protective cytokines such as IL-4 or IL-10 was evident by Elispot analysis (data not shown) when comparing RIP-LCMV-GP C57BL/6 and RIP-LCMV-GP × SV129(C57BL/6-F₈) mice at day 14 post–LCMV infection, and immune deviation is consequently not the likely cause for the observed lack of diabetes in the SV129-crossed animals.

In this study, we demonstrate that genes other than the MHC class I genes can overcome the autoaggressive potential of high numbers of self-reactive CD8 (and CD4) lymphocytes. Autoimmune diabetes development is drastically reduced in the RIP-LCMV transgenic mouse model when crossed once to SV129 mice and is even reduced after eight back-crosses to the original C57BL/6 background, whereas equally high levels of autoreactive CD8 and CD4 lymphocytes are generated systemically. Islets can be killed by such CTLs and their cytokines in vitro. However, β-cells regenerate in vivo under attack in protected mice, a finding that we think accounts for at least some of the protection from diabetes in SV129 crosses that maintain sufficient insulin production. Of course, additional, still unknown factors might contribute to this protection as well. Our observation is important in illustrating that genetic factors can easily overcome the autoaggressive potential of islet antigen–specific CD8 and CD4 lymphocytes. As a consequence, one might expect that the correlation between autoreactive lymphocytes and diabetes development could possibly be low in human populations with diverse genetic backgrounds.

Previous studies using tetramers and other functional in vitro assays have shown that up to one in three lymphocytes becomes an activated CTL during acute LCMV infection in C57/BL/6 mice (39,41). This translates into a very high degree of autoreactivity in RIP-LCMV-GP mice that have a fully functional anti-LCMV T-cell repertoire and that do not delete or anergize LCMV CD8 or CD4 cells as a consequence of expressing the LCMV antigen on β-cells (34). In humans, such high frequencies of autoreactive cells are never present systemically, and it is unlikely and has never been shown that they might reach such numbers locally in the islets (4). Thus, the pathogenic potential of autoreactive lymphocytes, even if they were restricted exclusively to diabetes susceptibility MHC alleles, might be overestimated, and other factors can be at least equally as important for the pathogenesis of autoimmune diabetes. One of these factors, which is illustrated in this report, is the potential of β-cells to regenerate under attack, but others are likely also playing a role. For example, local inflammation involving the activation of APCs in the pancreas and islets might be important in providing a suitable environment for autoaggressive lymphocytes (23). We find fewer activated APCs in islets of protected RIP-GP/SV129 mice (Table 1); however, this is likely the consequence of the prevention of diabetes (Table 1). Further studies will be required to clarify this issue. Second, the susceptibility of islet cells to cytokines or CTL killing contribute to diabetes pathogenesis, but in the present study, no differences were noted between the SV129 and C57BL/6 mouse strains, ruling out this possibility as a contributing factor out (Fig. 4). Third, the degree of extravasation of lymphocytes into the pancreas and islets might differ between SV129 and C57BL/6 mouse strains. Again, in this study, it is experimentally difficult to distinguish between cause and effect, as seen with the APCs. Fewer lymphocytes are present in the islets of nondiabetic RIP-LCMV-GP × SV129 mice; however, after LCMV infection, initially the same numbers of CD4 and CD8 lymphocytes enter the islets of RIP-LCMV SV129 and C57BL/6 transgenics (Table 1). Therefore, the lower degree of infiltration in SV129 mice is more likely a consequence of maintaining sufficient islet cell mass as well as a slowly subsiding inflammatory process. Finally, because differences in cytokine profiles were not observed (Table 1), differences in effector function of autoreactive CD4 and CD8 lymphocytes are less likely. However, it cannot be excluded that hence unknown effector functions other than cell number, CTL killing, and cytokine production determine the diabetogenicity of CD4 or CD8 lymphocytes, and that these functions differ between SV129 and C57BL/6 H-2D^b mouse strains.

In summary, the degree of systemically activated autoreactive CD4 and CD8 lymphocytes does not predict the degree of autoimmune diabetes in RIP-LCMV mice, which diverges drastically between C57BL/6 and SV129-N₈ genetic backgrounds. Islet cell regeneration is likely one of the factors accounting for this difference, but other non–MHC-linked genes have possible additional effects that still need to be discovered. It is important to consider that our findings might not be generally applicable to every diabetes model crossed to the SV129, as evidenced by good induction of type 1 diabetes in SV129 mice by low-dose streptozodozin (M.G.V.H, unpublished findings) and good diabetes incidence found in other models after back-crossing to SV129 (42). However, our findings suggest being careful with the interpretation of protection from diabetes in SV129-crossed models and, most importantly, exercising caution toward using autoreactive T-cell activity as a sole correlate for disease activity in type 1 diabetes.

This work was supported by National Institutes of Health Grants R01-AI4415 and R29-DK-51091 (to M.G.V.). M.G.V. is also supported by Juvenile Diabetes Foundation International Career Development Award 296120.

This report is publication no. 13440-NP from the Department of Neuropharmacology, The Scripps Research Institute (La Jolla, CA).

We thank Diana Frye for assistance with the manuscript preparation.