Prevention of Autoimmune Diabetes With Nonactivating Anti-CD3 Monoclonal Antibody KEVAN C. HEROLD, JEFFERY A. BLUESTONE, ANTHONY G. MONTAG, ASHU PARIH AR, AMY WIEGNER, RONALD E. GRESS, AND RAPHAEL HIRSCH
Autoreactive T cells mediate diabetes in animal models of insulin-dependent diabetes mellitus (IDDM) and are believed to cause the disease in humans. Therefore, immunotherapies directed against T cells are of particular interest for the treatment of IDDM. One candidate for such immunotherapy is anti-CD3 monoclonal antibodies (MoAbs), but clinical side effects are common with anti-CD3 treatment due to the ability of these MoAbs to activate T cells in vivo. However, F(ab')2 fragments of anti-CD3 are nonactivating and immunosuppressive. We evaluated the effects of whole anti-CD3 MoAb and F(ab')2 fragments in the setting of experimental autoimmune diabetes. Treatment with whole MoAb or F(ab')2 fragments significantly reduced the hyperglycemia induced with multiple low dosages of streptozocin (MDSDM; 232 ± 23 mg/dl, P < 0.01 and 235 ± 16 mg/dl, P < 0.01 vs. 325 ± 25 mg/dl, respectively) in male CD1 mice. Both whole MoAb and F(ab')2 fragments suppressed the development of insulitis (P < 0.001). Treatment with whole MoAb resulted in marked weight loss (10.4 ± 1.5% of total body wt), and the mice appeared ill and listless, whereas, mice treated with F(ab')2 fragments gained weight (4.9 ± 5.5% of total body wt) and appeared healthy. Treatment with whole MoAb caused activation of T cells in vivo as reflected by proliferation of freshly isolated spleen cells to recombinant interleukin-2. Depletion of T cells with whole MoAb was more pronounced than with F(ab')2 fragments, and T-cell receptor (TCR) reexpression on remaining cells occurred with F(ab')2 fragments within 48 h after F(ab')2 treatment. Despite the expression of this surface molecule, signaling by the TCR complex
From the Departments of Medicine and Pathology, the Ben May Institute, and the Committee on Immunology, The University of Chicago, Chicago, Illinois; and the Experimental Immunology Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland. Address correspondence and reprint requests to Kevan C. Herold, MD, Department of Medicine, Box 435, The University of Chicago, 5841 South Maryland Avenue, Chicago, IL 60637. Received for publication 19 June 1991 and accepted in revised form 18 November 1991.
DIABETES, VOL. 41, MARCH 1992
was blocked because T cells stimulated with anti-CD3 MoAb failed to produce lymphokines. We conclude that treatment with either whole anti-CD3 MoAb or F(ab')2 fragments suppresses the hyperglycemia and insulitis of MDSDM. The morbidity seen with whole MoAb does not occur with F(ab')2 fragments and is correlated with the failure of the latter to activate T cells in vivo. Nonactivating forms of anti-CD3 MoAbs may be a useful form of immunosuppressive treatment for incipient IDDM because they do not cause morbidity associated with activation of T cells, which occurs with whole MoAb. Diabetes 41:385-91,1992
N
umerous conventional immunotherapies have been suggested for use in the treatment of incipient insulin-dependent diabetes mellitus (IDDM) (1-3). These therapies have largely consisted of nonspecific immunosuppressive agents that incur a risk of potentially serious complications. Recently, attention has focused on developing immunotherapies that would target specific cell types involved in the autoimmune process. Anti-T-cell immunotherapy is of particular interest because autoreactive T cells mediate diabetes in animal models of the disease. For example, in the multidose streptozocin (STZ) model (MDSDM), diabetes is thought to occur after an insult to the (3-cell caused by STZ, which induces a T-cell autoimmune response that destroys p-cells (4). Diabetes in this model can be prevented by administration of anti-T-cell monoclonal antibodies (MoAbs) (5) and can be adoptively transferred with splenocytes from a diabetic animal (6). MDSDM is particularly useful as a model for the evaluation of novel immunotherapeutic interventions. Hyperglycemia and insulitis can be induced easily in a relatively short period of time in a high percentage of animals, and, importantly, treatment can be given before irreversible tissue damage or even before T cells have been sensitized to islet antigens.
385
PREVENTION OF DIABETES WITH NONACTIVATING ANTI-CD3 MoAB
and weekly injections of dPBS. In groups receiving MoAb, the first dose was administered 3-4 h before the first dose of STZ. Mice were weighed before the start of the experiment and again on days 8 and 15. Nonfasting plasma glucose levels were measured in blood obtained from the retroorbital sinus 13 days after the last dose of STZ (day 18). Pancreas histology. Mice were killed on day 19, and pancreatic tissue was harvested and fixed in 10% buffered formalin. After routine processing, 5-|xm sections were stained with hematoxylin and eosin. The slides were analyzed for presence of insulitis by a pathologist without knowledge of the identity of the sample. The presence of insulitis was scored as: 0, no evidence of insulitis; +1, mild mononuclear cell infiltrate but without loss of normal islet architecture; or +2, severe mononuclear cell infiltrate and loss of islet margins. The pancreases from at least three mice from each group were studied, and at least three levels of sections from each pancreas were analyzed. Flow cytometry analyses. Mice were killed on days 19 and 20, and single-cell suspensions of spleen or abdominal lymph nodes were prepared. (This would correspond to 24 and 48 h after the last dose of F[ab']2 anti-CD3 in mice receiving this treatment and 4 and 5 days after the last dose of whole MoAb.) Single-color flow-cytometric analyses were performed with the following fluorescein isothiocyanate (FITC)-conjugated MoAbs: anti-Leu 4 (anti-human CD3, irrelevant control; Becton Dickinson, Mountain View, CA), 145-2C11 (antimurine CD3; Boehringer Mannheim, Indianapolis, IN), RESEARCH DESIGN AND METHODS Male CD1 mice 5-9 wk old were purchased from Charles and anti-murine CD8 (Becton Dickinson). Data were River (Wilmington, MA). Mice were housed in a barrier collected on 10,000 cells with a fluorescein-activated cell sorter (FACScan) cytometer (Becton Dickinson) with facility and fed standard laboratory chow ad libitum. MoAbs. MoAb 145-2C11 is reactive with the e-chain of electronic gates placed on lymphocytes. For T-cell rethe murine CD3 complex (19). Concentrated antibody (1 ceptor modulation studies, two-color analyses were permg/ml) was obtained by growing hybridoma cells in an formed with or without preincubation of the cells with Accusyst P nitro-cell growth system (Endotronics, Min- 145-2C11. The cells were then stained with FITC-conjuneapolis, MN) and by collecting supernatant fluid. MoAb gated goat anti-hamster IgG (Kirkegaard and Perry, was purified from the supernatant on Sepharose Pro- Gaithersburg, MD) followed by biotin-conjugated antitein-A (Pharmacia, Piscataway, NJ). The antibody was Thy-1.2 (Becton Dickinson) followed by phycoerythrindialyzed into phosphate-buffered saline (dPBS) and egg white avidin (Jackson Immunoresearch, West Grove, used for treatment at a concn of 1 mg/ml. F(ab')2 PA). In these analyses, the goat anti-hamster reagent fragments were prepared by a pepsin digestion of puri- cross-reacted with mouse cells bearing IgG, but T cells fied MoAb as described previously (11). F(ab')2 frag- could be distinguished by staining positively with the ments were separated from intact MoAb by passage over anti-Thy-1.2 reagent. Analyses of these data were perSepharose Protein-A column and used for treatment of formed with Consort 30 software (Becton Dickinson). mice. Purity of F(ab')2 fragments was indicated by the Proliferation assays. To determine whether treatment absence of detectable intact MoAb by sodium dodecyl with MoAb had induced activation of T cells in vivo, some sulfate-polyacrylamide gel electrophoresis and by the mice were killed 1 day after the first dose of whole MoAb inability of the F(ab')2 fragments in soluble form to induce or F(ab')2 fragments, and single-cell suspensions of splenocytes were prepared. The spleen cells (0.4 x 106 proliferation of spleen cells in vitro. Treatment protocols. Mice were randomly assigned to cells) were placed into microwells with human recombione of four treatment groups: 1) STZ (40 mg/kg body wt nant interleukin-2 (rlL-2) in serial dilution. After 48 h in 3 i.p. in citrate buffer) on days 1-5 and weekly intraperito- culture at 37°C with 5% CO2, 1 n-Ci [ H]thymidine was neal injections of dPBS beginning on day 1 and continu- added to each microwell, and the wells were harvested 8 ing for the duration of the study, 2) STZ and weekly h later onto glass-fiber filter paper. The filters were injections of whole 145-2C11 (300 IXQ i.p. per mouse per counted in a p-counter, and the mean counts per minute injection), 3) STZ and an injection of F(ab')2 fragments of of triplicate cultures was calculated. MoAb 145-2C11 (250 ^g i.p.) on day 1 followed by 50 Lymphokine assays. Spleen cell suspensions were pre\x.g i.p. every other day, and 4) citrate buffer on days 1-5 pared on day 18 or 19. The cells were placed in culture One candidate for such T-cell immunotherapy is antiCD3 MoAb, which is effective in the suppression of organ allograft rejection in both humans (7-9) and mice (10). However, significant clinical side effects are associated with its use including pulmonary edema, CNS disturbances, hypertension, and others that appear to be the consequence of T-cell activation and cytokine release (11-15). The severity of these side effects is increased when anti-CD3 MoAb is administered in the nontransplantation setting such as to patients with malignancies (16) or IDDM (17), perhaps because the immune system of these patients has not been suppressed by previous or concomitant immunosuppressive medication. As a result, the applicability of anti-CD3 therapy to autoimmune disease has been limited and little is known regarding its effects on T-cell responses in the autoimmune setting. Nonactivating forms of anti-CD3 MoAb would presumably not induce the morbidity associated with T-cell activation and might therefore be more appropriate than whole MoAb in the therapy of autoimmune diseases. In this regard, anti-CD3 F(ab')2 fragments are capable of suppressing transplantation responses in mice without inducing T-cell activation (18). To evaluate whether antiCD3 MoAb would be effective in the setting of autoimmune diabetes, whole Ig or F(ab')2 fragments were administered to mice with MDSDM, and their effects on T-cell function, morbidity, suppression of insulitis, and hyperglycemia were compared.
386
DIABETES, VOL. 41, MARCH 1992
K.C. HEROLD AND ASSOCIATES
in 2-ml macrowells (6.0 x 106 cells/well) with 10% supernatant fluid from 145-2C11-producing hybridoma cells. After 24 h, culture supernatant fluids were harvested. IL-2 and IL-3 were measured in standard colorimetric bioassays (measuring uptake of 3-[4,5-methyl thiazol-2-yl]-2,5diphenyltetrazolium bromide), with the indicator lines CTLL-2 and FDCP-1, respectively (20,21). The units of IL-2 were calculated by comparison of the sample titer with a standard containing 1000 U human rll_-2/ml. The titer of IL-3 was compared to a supernatant fluid from the IL-3-producing cell line WEHI-3. lnterferon-7 (IFN-7) was measured in an enzyme-linked immunosorbent assay (ELISA) with previously described methods (22). The units of IFN-7 were calculated by comparing the sample titer to a standard containing 1000 U recombinant IFN-7/ml. Data analysis. Each experiment included groups of five to eight mice. For plasma glucose experiments, results of three experiments are represented together. Comparisons between groups were made with Students' nest for unpaired samples or by x2 analysis. Unless indicated otherwise, values are means ± SE. Where appropriate, data from representative experiments are presented rather than group means. RESULTS Treatment with either whole of F(ab')2 fragments of anti-CD3 MoAb suppress STZ-induced hyperglycemia and insulitis. Diabetes was induced in male CD-1 mice with five low-dose injections of STZ. Four hours before the first dose of STZ, mice were treated with either dPBS, whole anti-CD3 MoAb (300 |xg/wk per mouse), or F(ab')2 fragments of anti-CD3 MoAb (250 jxg/mouse followed by 50 |xg every other day). These dosages of MoAb were selected based on their ability to saturate T-cell receptors (TCRs) in vivo and suppress skin allograft rejection. Because of the shorter half-life of F(ab')2 fragments compared with whole MoAb, multiple doses of F(ab')2 fragments were necessary to maintain TCR modulation for the 2-wk experimental period (10,18). To evaluate the effects of these treatments on induction of diabetes, plasma glucose levels were measured 4, 7, and 13 days after completion of STZ administration (corresponding to days 9, 13, and 18). The glucose levels were significantly reduced in groups treated with either whole MoAb (232 ± 23 mg/dl, P < 0.01) or F(ab')2 fragments (235 ± 1 6 mg/dl, P250 mg/ml compared with mice treated with whole MoAb (9 of 19) or dPBS(15of22).
500-
-
1 400-
• •
300-
200-
'A •
: ;
i 1. • dPBS STZ
F(ab')2 STZ
Whole Ab + STZ
? ^ *• dPBS only
GROUP FIG. 1. Suppression of hyperglycemia induced with multiple doses of streptozocin (STZ) in CD1 mice treated with antl-CD3 monoclonal antibody (Ab). Glucose levels were measured on day 18 (13 days after last dose of STZ) in plasma from mice in each treatment group. Results of 3 experiments, each with 5 - 8 mice/group are represented together. " P < 0.001, *P< 0.01 vs. phosphate-buffered saline (dPBS) + STZ.
group were scored by a pathologist blinded to the identity of the sample (Table 1; Fig. 2). All islets from nondiabetic mice were free of insulitis as were 96% of islets from F(ab')2-treated and 91% of islets from whole MoAb-treated mice. Lesions graded as mild insulitis were found in 29% of islets from dPBS + STZ-administered mice, 4% of islets from F(ab')2 + STZ, and 7% of islets from whole MoAb + STZ-administered mice. More-severe lesions, which included loss of islet architecture and large numbers of intraislet inflammatory cells, were seen in 20% of islets from dPBS + STZ-administered mice, 2% of islets from whole MoAb + STZ, and none of the islets from F(ab')2 + STZ-administered mice. Many of the islet cells from the anti-CD3 treatment groups showed evidence of vacuolization, which we interpret to represent direct effects of the STZ. Thus, both whole MoAb and F(ab')2 fragments dramatically suppressed the development of insulitis (P< 0.001). Although not statistically significant, there was a trend towards greater protection with F(ab')2 fragments compared with whole MoAb. Treatment with whole MoAb but not F(ab')2 fragments of anti-CD3 MoAb results in morbidity and T-cell activation. Because T-cell activation-mediated toxicities have been a major limitation to the clinical use of anti-CD3 MoAb in autoimmune settings, it was of interest Because some of the elevation of the plasma glucose to compare the morbidity caused by thse two forms of levels might have been due to a direct toxic effect of STZ, treatment. As a general measure of health, mice were independent of a T-cell-mediated inflammatory re- weighed at the start of the experiments and on days 10 sponse, it was important to directly evaluate the pres- and 15 (Fig. 3). On average, mice treated with the whole ence and grade of insulitis in each group. Multiple tissue MoAb lost 10.4% of their body weight (range 2.2-17.9%, sections of the pancreases from three members of each P < 0.001 vs. whole MoAb) on day 8. These mice
DIABETES, VOL. 41, MARCH 1992
387
PREVENTION OF DIABETES WITH NONACTIVATING ANTI-CD3 MoAB
TABLE 1 Presence of insulitis in pancreases in mice in each treatment group Insulitis Group
No
Mild
Severe
Islets
Phosphate-buffered saline (dPBS) dPBS + streptozocin (STZ) F(ab')2 + STZ* Whole monoclonal antibody + STZ*
54(100) 21 (51) 24 (96) 39(91)
0 12(29) 1(4) 3(7)
0 8(20) 0 1(2)
54 41 25 43
Sections of pancreas from 3 to 4 mice/group on day 19 were scored for the presence of insulitis as described in METHODS. The data represent n islets with percentages in parentheses in each category of insulitis. Islets from each of the members of the group receiving dPBS + STZ displayed evidence of insulitis, although only 49% of the total number of islets examined had cellular infiltrates. *P < 0.001 vs. dPBS + STZ.
appeared ill and listless and lost their normal coat texture. In contrast, the F(ab')2-treated group on average gained weight during the study (mean 4.9%, range 5% loss to 8.3% gain) and appeared healthy. Mice receiving STZ only and buffers only gained 1.89 ±1.17 and 8.99 ± 1.3%, respectively, of total body weight. To determine whether this lack of morbidity with F(ab')2 fragments was correlated with failure to activate T cells, proliferative responses of freshly isolated spleen cells from treated mice were measured after stimulation with rlL-2 (Fig. 4). One of the earliest signs of T-cell activation is the expression of high-affinity IL-2 receptors (23), and it has been demonstrated previously that spleen cells from mice treated with whole anti-CD3 but not with F(ab')2 fragments proliferate to IL-2 within 24 h of injection (11). Spleen cells were isolated 24 h after treatment with antibody and cultured with dilutions of rlL-2. Spleen
cells from mice treated with F(ab')2 fragments or PBS showed little response to IL-2, whereas proliferation of spleen cells from mice treated with whole MoAb was seen at concentrations as low as 1 U/ml, indicating the presence of activated T cells in the mice treated with whole MoAb but not in mice treated with F(ab')2 fragments. Thus, both whole MoAb and F(ab% fragments suppressed insulitis, but F(ab')2 fragments did not induce the T-cell activation and morbidity associated with whole anti-CD3 MoAb. Effects of anti-CD3 MoAbs differ in respect to T-cell depletion and expression of TCR but both impair lymphokine production by T cells. To explore potential mechanisms of suppression of diabetes induced by treatment with anti-CD3 MoAb, T-cell number, phenotype, and function were analyzed on days 18 and 19. Both forms of MoAb induced depletion of T cells from spleen (Table 2). However, a greater proportion of the T cells were depleted with whole MoAb. A relatively larger proportion of CD8+ T cells were depleted as a result of F(ab')2 treatment compared with CD4+ cells, whereas both were depleted by whole MoAb. The expression of TCR on the remaining cells from
5.0-1
•^
3.0-
-t-
Autoreactive T cells mediate diabetes in animal models of insulin-dependent diabetes mellitus (IDDM) and are believed to cause the disease in humans. Therefore, immunotherapies directed against T cells are of particular interest for the treatment of IDDM. One candidate for such immunotherapy is anti-CD3 monoclonal antibodies (MoAbs), but clinical side effects are common with anti-CD3 treatment due to the ability of these MoAbs to activate T cells in vivo. However, F(ab')2 fragments of anti-CD3 are nonactivating and immunosuppressive. We evaluated the effects of whole anti-CD3 MoAb and F(ab')2 fragments in the setting of experimental autoimmune diabetes. Treatment with whole MoAb or F(ab')2 fragments significantly reduced the hyperglycemia induced with multiple low dosages of streptozocin (MDSDM; 232 ± 23 mg/dl, P < 0.01 and 235 ± 16 mg/dl, P < 0.01 vs. 325 ± 25 mg/dl, respectively) in male CD1 mice. Both whole MoAb and F(ab')2 fragments suppressed the development of insulitis (P < 0.001). Treatment with whole MoAb resulted in marked weight loss (10.4 ± 1.5% of total body wt), and the mice appeared ill and listless, whereas, mice treated with F(ab')2 fragments gained weight (4.9 ± 5.5% of total body wt) and appeared healthy. Treatment with whole MoAb caused activation of T cells in vivo as reflected by proliferation of freshly isolated spleen cells to recombinant interleukin-2. Depletion of T cells with whole MoAb was more pronounced than with F(ab')2 fragments, and T-cell receptor (TCR) reexpression on remaining cells occurred with F(ab')2 fragments within 48 h after F(ab')2 treatment. Despite the expression of this surface molecule, signaling by the TCR complex
From the Departments of Medicine and Pathology, the Ben May Institute, and the Committee on Immunology, The University of Chicago, Chicago, Illinois; and the Experimental Immunology Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland. Address correspondence and reprint requests to Kevan C. Herold, MD, Department of Medicine, Box 435, The University of Chicago, 5841 South Maryland Avenue, Chicago, IL 60637. Received for publication 19 June 1991 and accepted in revised form 18 November 1991.
DIABETES, VOL. 41, MARCH 1992
was blocked because T cells stimulated with anti-CD3 MoAb failed to produce lymphokines. We conclude that treatment with either whole anti-CD3 MoAb or F(ab')2 fragments suppresses the hyperglycemia and insulitis of MDSDM. The morbidity seen with whole MoAb does not occur with F(ab')2 fragments and is correlated with the failure of the latter to activate T cells in vivo. Nonactivating forms of anti-CD3 MoAbs may be a useful form of immunosuppressive treatment for incipient IDDM because they do not cause morbidity associated with activation of T cells, which occurs with whole MoAb. Diabetes 41:385-91,1992
N
umerous conventional immunotherapies have been suggested for use in the treatment of incipient insulin-dependent diabetes mellitus (IDDM) (1-3). These therapies have largely consisted of nonspecific immunosuppressive agents that incur a risk of potentially serious complications. Recently, attention has focused on developing immunotherapies that would target specific cell types involved in the autoimmune process. Anti-T-cell immunotherapy is of particular interest because autoreactive T cells mediate diabetes in animal models of the disease. For example, in the multidose streptozocin (STZ) model (MDSDM), diabetes is thought to occur after an insult to the (3-cell caused by STZ, which induces a T-cell autoimmune response that destroys p-cells (4). Diabetes in this model can be prevented by administration of anti-T-cell monoclonal antibodies (MoAbs) (5) and can be adoptively transferred with splenocytes from a diabetic animal (6). MDSDM is particularly useful as a model for the evaluation of novel immunotherapeutic interventions. Hyperglycemia and insulitis can be induced easily in a relatively short period of time in a high percentage of animals, and, importantly, treatment can be given before irreversible tissue damage or even before T cells have been sensitized to islet antigens.
385
PREVENTION OF DIABETES WITH NONACTIVATING ANTI-CD3 MoAB
and weekly injections of dPBS. In groups receiving MoAb, the first dose was administered 3-4 h before the first dose of STZ. Mice were weighed before the start of the experiment and again on days 8 and 15. Nonfasting plasma glucose levels were measured in blood obtained from the retroorbital sinus 13 days after the last dose of STZ (day 18). Pancreas histology. Mice were killed on day 19, and pancreatic tissue was harvested and fixed in 10% buffered formalin. After routine processing, 5-|xm sections were stained with hematoxylin and eosin. The slides were analyzed for presence of insulitis by a pathologist without knowledge of the identity of the sample. The presence of insulitis was scored as: 0, no evidence of insulitis; +1, mild mononuclear cell infiltrate but without loss of normal islet architecture; or +2, severe mononuclear cell infiltrate and loss of islet margins. The pancreases from at least three mice from each group were studied, and at least three levels of sections from each pancreas were analyzed. Flow cytometry analyses. Mice were killed on days 19 and 20, and single-cell suspensions of spleen or abdominal lymph nodes were prepared. (This would correspond to 24 and 48 h after the last dose of F[ab']2 anti-CD3 in mice receiving this treatment and 4 and 5 days after the last dose of whole MoAb.) Single-color flow-cytometric analyses were performed with the following fluorescein isothiocyanate (FITC)-conjugated MoAbs: anti-Leu 4 (anti-human CD3, irrelevant control; Becton Dickinson, Mountain View, CA), 145-2C11 (antimurine CD3; Boehringer Mannheim, Indianapolis, IN), RESEARCH DESIGN AND METHODS Male CD1 mice 5-9 wk old were purchased from Charles and anti-murine CD8 (Becton Dickinson). Data were River (Wilmington, MA). Mice were housed in a barrier collected on 10,000 cells with a fluorescein-activated cell sorter (FACScan) cytometer (Becton Dickinson) with facility and fed standard laboratory chow ad libitum. MoAbs. MoAb 145-2C11 is reactive with the e-chain of electronic gates placed on lymphocytes. For T-cell rethe murine CD3 complex (19). Concentrated antibody (1 ceptor modulation studies, two-color analyses were permg/ml) was obtained by growing hybridoma cells in an formed with or without preincubation of the cells with Accusyst P nitro-cell growth system (Endotronics, Min- 145-2C11. The cells were then stained with FITC-conjuneapolis, MN) and by collecting supernatant fluid. MoAb gated goat anti-hamster IgG (Kirkegaard and Perry, was purified from the supernatant on Sepharose Pro- Gaithersburg, MD) followed by biotin-conjugated antitein-A (Pharmacia, Piscataway, NJ). The antibody was Thy-1.2 (Becton Dickinson) followed by phycoerythrindialyzed into phosphate-buffered saline (dPBS) and egg white avidin (Jackson Immunoresearch, West Grove, used for treatment at a concn of 1 mg/ml. F(ab')2 PA). In these analyses, the goat anti-hamster reagent fragments were prepared by a pepsin digestion of puri- cross-reacted with mouse cells bearing IgG, but T cells fied MoAb as described previously (11). F(ab')2 frag- could be distinguished by staining positively with the ments were separated from intact MoAb by passage over anti-Thy-1.2 reagent. Analyses of these data were perSepharose Protein-A column and used for treatment of formed with Consort 30 software (Becton Dickinson). mice. Purity of F(ab')2 fragments was indicated by the Proliferation assays. To determine whether treatment absence of detectable intact MoAb by sodium dodecyl with MoAb had induced activation of T cells in vivo, some sulfate-polyacrylamide gel electrophoresis and by the mice were killed 1 day after the first dose of whole MoAb inability of the F(ab')2 fragments in soluble form to induce or F(ab')2 fragments, and single-cell suspensions of splenocytes were prepared. The spleen cells (0.4 x 106 proliferation of spleen cells in vitro. Treatment protocols. Mice were randomly assigned to cells) were placed into microwells with human recombione of four treatment groups: 1) STZ (40 mg/kg body wt nant interleukin-2 (rlL-2) in serial dilution. After 48 h in 3 i.p. in citrate buffer) on days 1-5 and weekly intraperito- culture at 37°C with 5% CO2, 1 n-Ci [ H]thymidine was neal injections of dPBS beginning on day 1 and continu- added to each microwell, and the wells were harvested 8 ing for the duration of the study, 2) STZ and weekly h later onto glass-fiber filter paper. The filters were injections of whole 145-2C11 (300 IXQ i.p. per mouse per counted in a p-counter, and the mean counts per minute injection), 3) STZ and an injection of F(ab')2 fragments of of triplicate cultures was calculated. MoAb 145-2C11 (250 ^g i.p.) on day 1 followed by 50 Lymphokine assays. Spleen cell suspensions were pre\x.g i.p. every other day, and 4) citrate buffer on days 1-5 pared on day 18 or 19. The cells were placed in culture One candidate for such T-cell immunotherapy is antiCD3 MoAb, which is effective in the suppression of organ allograft rejection in both humans (7-9) and mice (10). However, significant clinical side effects are associated with its use including pulmonary edema, CNS disturbances, hypertension, and others that appear to be the consequence of T-cell activation and cytokine release (11-15). The severity of these side effects is increased when anti-CD3 MoAb is administered in the nontransplantation setting such as to patients with malignancies (16) or IDDM (17), perhaps because the immune system of these patients has not been suppressed by previous or concomitant immunosuppressive medication. As a result, the applicability of anti-CD3 therapy to autoimmune disease has been limited and little is known regarding its effects on T-cell responses in the autoimmune setting. Nonactivating forms of anti-CD3 MoAb would presumably not induce the morbidity associated with T-cell activation and might therefore be more appropriate than whole MoAb in the therapy of autoimmune diseases. In this regard, anti-CD3 F(ab')2 fragments are capable of suppressing transplantation responses in mice without inducing T-cell activation (18). To evaluate whether antiCD3 MoAb would be effective in the setting of autoimmune diabetes, whole Ig or F(ab')2 fragments were administered to mice with MDSDM, and their effects on T-cell function, morbidity, suppression of insulitis, and hyperglycemia were compared.
386
DIABETES, VOL. 41, MARCH 1992
K.C. HEROLD AND ASSOCIATES
in 2-ml macrowells (6.0 x 106 cells/well) with 10% supernatant fluid from 145-2C11-producing hybridoma cells. After 24 h, culture supernatant fluids were harvested. IL-2 and IL-3 were measured in standard colorimetric bioassays (measuring uptake of 3-[4,5-methyl thiazol-2-yl]-2,5diphenyltetrazolium bromide), with the indicator lines CTLL-2 and FDCP-1, respectively (20,21). The units of IL-2 were calculated by comparison of the sample titer with a standard containing 1000 U human rll_-2/ml. The titer of IL-3 was compared to a supernatant fluid from the IL-3-producing cell line WEHI-3. lnterferon-7 (IFN-7) was measured in an enzyme-linked immunosorbent assay (ELISA) with previously described methods (22). The units of IFN-7 were calculated by comparing the sample titer to a standard containing 1000 U recombinant IFN-7/ml. Data analysis. Each experiment included groups of five to eight mice. For plasma glucose experiments, results of three experiments are represented together. Comparisons between groups were made with Students' nest for unpaired samples or by x2 analysis. Unless indicated otherwise, values are means ± SE. Where appropriate, data from representative experiments are presented rather than group means. RESULTS Treatment with either whole of F(ab')2 fragments of anti-CD3 MoAb suppress STZ-induced hyperglycemia and insulitis. Diabetes was induced in male CD-1 mice with five low-dose injections of STZ. Four hours before the first dose of STZ, mice were treated with either dPBS, whole anti-CD3 MoAb (300 |xg/wk per mouse), or F(ab')2 fragments of anti-CD3 MoAb (250 jxg/mouse followed by 50 |xg every other day). These dosages of MoAb were selected based on their ability to saturate T-cell receptors (TCRs) in vivo and suppress skin allograft rejection. Because of the shorter half-life of F(ab')2 fragments compared with whole MoAb, multiple doses of F(ab')2 fragments were necessary to maintain TCR modulation for the 2-wk experimental period (10,18). To evaluate the effects of these treatments on induction of diabetes, plasma glucose levels were measured 4, 7, and 13 days after completion of STZ administration (corresponding to days 9, 13, and 18). The glucose levels were significantly reduced in groups treated with either whole MoAb (232 ± 23 mg/dl, P < 0.01) or F(ab')2 fragments (235 ± 1 6 mg/dl, P250 mg/ml compared with mice treated with whole MoAb (9 of 19) or dPBS(15of22).
500-
-
1 400-
• •
300-
200-
'A •
: ;
i 1. • dPBS STZ
F(ab')2 STZ
Whole Ab + STZ
? ^ *• dPBS only
GROUP FIG. 1. Suppression of hyperglycemia induced with multiple doses of streptozocin (STZ) in CD1 mice treated with antl-CD3 monoclonal antibody (Ab). Glucose levels were measured on day 18 (13 days after last dose of STZ) in plasma from mice in each treatment group. Results of 3 experiments, each with 5 - 8 mice/group are represented together. " P < 0.001, *P< 0.01 vs. phosphate-buffered saline (dPBS) + STZ.
group were scored by a pathologist blinded to the identity of the sample (Table 1; Fig. 2). All islets from nondiabetic mice were free of insulitis as were 96% of islets from F(ab')2-treated and 91% of islets from whole MoAb-treated mice. Lesions graded as mild insulitis were found in 29% of islets from dPBS + STZ-administered mice, 4% of islets from F(ab')2 + STZ, and 7% of islets from whole MoAb + STZ-administered mice. More-severe lesions, which included loss of islet architecture and large numbers of intraislet inflammatory cells, were seen in 20% of islets from dPBS + STZ-administered mice, 2% of islets from whole MoAb + STZ, and none of the islets from F(ab')2 + STZ-administered mice. Many of the islet cells from the anti-CD3 treatment groups showed evidence of vacuolization, which we interpret to represent direct effects of the STZ. Thus, both whole MoAb and F(ab')2 fragments dramatically suppressed the development of insulitis (P< 0.001). Although not statistically significant, there was a trend towards greater protection with F(ab')2 fragments compared with whole MoAb. Treatment with whole MoAb but not F(ab')2 fragments of anti-CD3 MoAb results in morbidity and T-cell activation. Because T-cell activation-mediated toxicities have been a major limitation to the clinical use of anti-CD3 MoAb in autoimmune settings, it was of interest Because some of the elevation of the plasma glucose to compare the morbidity caused by thse two forms of levels might have been due to a direct toxic effect of STZ, treatment. As a general measure of health, mice were independent of a T-cell-mediated inflammatory re- weighed at the start of the experiments and on days 10 sponse, it was important to directly evaluate the pres- and 15 (Fig. 3). On average, mice treated with the whole ence and grade of insulitis in each group. Multiple tissue MoAb lost 10.4% of their body weight (range 2.2-17.9%, sections of the pancreases from three members of each P < 0.001 vs. whole MoAb) on day 8. These mice
DIABETES, VOL. 41, MARCH 1992
387
PREVENTION OF DIABETES WITH NONACTIVATING ANTI-CD3 MoAB
TABLE 1 Presence of insulitis in pancreases in mice in each treatment group Insulitis Group
No
Mild
Severe
Islets
Phosphate-buffered saline (dPBS) dPBS + streptozocin (STZ) F(ab')2 + STZ* Whole monoclonal antibody + STZ*
54(100) 21 (51) 24 (96) 39(91)
0 12(29) 1(4) 3(7)
0 8(20) 0 1(2)
54 41 25 43
Sections of pancreas from 3 to 4 mice/group on day 19 were scored for the presence of insulitis as described in METHODS. The data represent n islets with percentages in parentheses in each category of insulitis. Islets from each of the members of the group receiving dPBS + STZ displayed evidence of insulitis, although only 49% of the total number of islets examined had cellular infiltrates. *P < 0.001 vs. dPBS + STZ.
appeared ill and listless and lost their normal coat texture. In contrast, the F(ab')2-treated group on average gained weight during the study (mean 4.9%, range 5% loss to 8.3% gain) and appeared healthy. Mice receiving STZ only and buffers only gained 1.89 ±1.17 and 8.99 ± 1.3%, respectively, of total body weight. To determine whether this lack of morbidity with F(ab')2 fragments was correlated with failure to activate T cells, proliferative responses of freshly isolated spleen cells from treated mice were measured after stimulation with rlL-2 (Fig. 4). One of the earliest signs of T-cell activation is the expression of high-affinity IL-2 receptors (23), and it has been demonstrated previously that spleen cells from mice treated with whole anti-CD3 but not with F(ab')2 fragments proliferate to IL-2 within 24 h of injection (11). Spleen cells were isolated 24 h after treatment with antibody and cultured with dilutions of rlL-2. Spleen
cells from mice treated with F(ab')2 fragments or PBS showed little response to IL-2, whereas proliferation of spleen cells from mice treated with whole MoAb was seen at concentrations as low as 1 U/ml, indicating the presence of activated T cells in the mice treated with whole MoAb but not in mice treated with F(ab')2 fragments. Thus, both whole MoAb and F(ab% fragments suppressed insulitis, but F(ab')2 fragments did not induce the T-cell activation and morbidity associated with whole anti-CD3 MoAb. Effects of anti-CD3 MoAbs differ in respect to T-cell depletion and expression of TCR but both impair lymphokine production by T cells. To explore potential mechanisms of suppression of diabetes induced by treatment with anti-CD3 MoAb, T-cell number, phenotype, and function were analyzed on days 18 and 19. Both forms of MoAb induced depletion of T cells from spleen (Table 2). However, a greater proportion of the T cells were depleted with whole MoAb. A relatively larger proportion of CD8+ T cells were depleted as a result of F(ab')2 treatment compared with CD4+ cells, whereas both were depleted by whole MoAb. The expression of TCR on the remaining cells from
5.0-1
•^
3.0-
-t-
