TNF-α Antibody Therapy in Combination With the T-Cell–Specific Antibody Anti-TCR Reverses the Diabetic Metabolic State in the LEW.1AR1-iddm Rat

Type 1 diabetes (T1D) is an immune cell–mediated disease that causes insulin dependence due to selective pancreatic β-cell death (1–3). The infiltrate in the pancreatic islets is composed of different immune cell types, especially macrophages and T cells (2–4). Upon activation, the immune cells release proinflammatory cytokines and other cytotoxic mediators, causing β-cell apoptosis (3,5).

Effective prevention strategies require a combination therapy (6–8) to target the proinflammatory cytokines produced in the different immune cell types. Of crucial importance is in this context the proinflammatory cytokine tumor necrosis factor-α (TNF-α). It is expressed in all immune cell types infiltrating the pancreatic islets in patients with T1D as well as in the spontaneous mouse (NOD mouse) and rat (BB, Komeda, and LEW.1AR1-iddm rats) models (3,9).

Cumulative evidence from studies in a variety of other autoimmune diseases supports the contention that the inflammatory process in the affected organs always goes along with a high TNF-α expression (9–13). Anti–TNF-α therapy is therefore an established therapeutic principle (14–16) in rheumatoid arthritis, inflammatory bowel diseases, lupus erythematosus, Sjögren syndrome, psoriasis, and Hashimoto thyroiditis (10,15,17).

Anti–TNF-α monotherapy in T1D is ineffective (18–21) and may even aggravate the disease (22,23). It is probably not surprising in an autoimmune disease such as T1D, in which the pancreatic β-cells are so extremely vulnerable (24), that these cells not only become dysfunctional but are also quickly destroyed.

As has been documented in diabetic LEW.1AR1-iddm (IDDM) rats, as well as in pancreases from patients with T1D (3), the islet-infiltrating immune cells produce and release, along with interleukin (IL)-1β, TNF-α as the dominating proinflammatory cytokine (3,5,25).

Combination therapies, at variance from monotherapies, have shown promising results in T1D (25–29) and other autoimmune diseases (11,14,15). We therefore investigated in the IDDM rat, an animal model of human T1D, which mirrors the human T1D situation most closely (3,5,30,31), whether a combination therapy of anti–TNF-α, which suppresses TNF-α release from all immune cells, with anti–T-cell receptor (TCR), which suppresses proinflammatory cytokine release from activated T cells, can be successful. Anti-TCR is an antibody against T cells that binds to an epitope of the α/β-chains of the TCR (32). It is comparable in action to anti-CD3, which binds to the ε-chain of the CD3 molecule (33). Both antibodies bind noncovalently to the TCR/CD3 complex and modulate T cells by affecting the trans-signaling pathway (32,33). Anti-TCR is used in rats because anti-CD3 is available for mice (33) and humans (33,34) but not for rats (25,35). We show in this study that this combination therapy scheme is capable of successfully reversing the diabetic metabolic state.

Congenic rats (for details see http://www.mh-hannover.de/3642.html) were bred and maintained as described (25,30). Experimental procedures were approved by the District Government of Hannover (LAVES, nos. 33-42502-05/958 and 509.6-42502-03/684).

Four experimental groups were studied. Group 1 (n = 5) comprised healthy, normoglycemic IDDM rats without therapy; group 2 (n = 12) comprised acutely diabetic IDDM rats treated for 5 consecutive days with rat-specific anti–TNF-α (5 mg/kg body weight [bw] i.v.; provided by Janssen Research & Development, Spring House, PA). Group 3 (n = 12) comprised diabetic rats treated for 5 consecutive days with anti–TNF-α and in addition with anti-TCR (0.5 mg/kg bw i.v.) (R73; AbD Serotec, Munich, Germany). Group 4 (n = 5) comprised untreated diabetic IDDM rats. All therapies were started within 1 day after T1D onset at blood glucose concentrations >10 mmol/L. Animals in groups 1 and 4 received 0.9% NaCl solution. Anti-TCR causes slight lymphocyte reduction in the periphery (<20%), which disappears within 1 week after cessation of therapy (25). Anti–TNF-α was without effect and also did not further reduce the lymphocyte count in combination therapy with anti-TCR (data not shown). The CD4-to-CD8 T-cell ratios were not affected by any of the treatments (Supplementary Table 1).

Biopsies of tissue (30 mg) were obtained from the pancreas tail of each animal under isoflurane anesthesia as previously described (36). To identify changes in the pancreatic islets, one biopsy was collected on the day of T1D manifestation, immediately before the start of therapy, and another at the end of therapy; the remaining pancreas, comprising head and tail, pancreas-draining lymph nodes, and serum, were collected 60 days after the end of therapy (25). Tissue specimens were fixed for microscopic analyses (25,36). Blood glucose concentrations were determined daily (Glucometer Elite; Bayer, Leverkusen, Germany). Serum C-peptide was analyzed with a rat-specific ELISA (Mercodia, Uppsala, Sweden) and serum cytokine protein concentrations with a multiplex immunoassay kit (Bio-Rad, Munich, Germany) (25,36).

Serial sections were stained with the avidin-biotin complex method or the double-immunofluorescence technique with primary antibodies for β-cells and immune cells (3,36). The antibodies against TNF-α and T cells recognized epitopes other than those targeted by the antibodies used for therapy. β-Cell apoptosis was quantified by TUNEL (Roche, Mannheim, Germany), as well as proliferation by a double staining with Ki67 antibody (COOH-terminal; Acris, Herford, Germany) and insulin (D3E7; AbD Serotec). At each time point, 25–40 pancreatic islets were studied (25,36). A minimum of 1,000 β-cells were counted in the proliferation and apoptosis analyses (25,36). Islet infiltration was scored as described (25). The β-cell mass, identified by insulin and GLUT2 staining on sequential sections, was calculated by the ratio of the β-cell area (mm²) to the whole pancreatic area, determined four times in distances of 100 µm each and multiplied by the pancreas weight. Analyses were performed using an Olympus BX61 microscope (25,36).

Pancreatic sections from all experimental groups were placed on three-chamber slides. The in situ RT-PCR analysis was performed on a special thermal cycler (MJ Research, Waltham, MA) (25,36). Primer sequences are provided in Supplementary Table 2.

The quality of the mRNA isolated from pancreas-draining lymph nodes was controlled with the Experion electrophoresis system (Bio-Rad). mRNA was quantified in real-time RT-PCR reactions performed with the DNA Engine Opticon fluorescence detection system (MJ Research) (36). Primer sequences are provided in Supplementary Table 3.

Results are presented as mean values ± SEM. Comparisons among the different therapy groups and the normoglycemic or diabetic control rats were analyzed with the Mann-Whitney U test and ANOVA, and correlation coefficients were calculated according to Pearson with the Prism 5 program (GraphPad Inc., San Diego, CA).

Anti–TNF-α (5 mg/kg bw, for 5 days) was administered to IDDM rats within 1 day after T1D manifestation (blood glucose >10 mmol/L), alone or combined with anti-TCR (0.5 mg/kg bw, for 5 days). Combination therapy resulted in an instantaneous and sustained return to normoglycemia within 1 day after the start of therapy in 8 of 12 rats (Fig. 1A), defined as blood glucose <10 mmol/L, compared with the nonresponding rats. Rats (n = 12) with anti–TNF-α therapy alone after T1D manifestation remained permanently hyperglycemic (>15.0 mmol/L) (Fig. 1A), and serum C-peptide levels were not preserved (Fig. 1C). In the animals responding to combination therapy, normoglycemia achieved at the end of the 5-day therapy period was further maintained for another 60 days in the absence of therapy (Fig. 1A). Thereby, combination therapy achieved glucose levels identical to those of healthy, normoglycemic control rats (5.4 ± 0.2 vs. 5.1 ± 0.1 mmol/L). Rats (n = 4) not responding to combination therapy and diabetic rats without therapy showed no return to normoglycemia and needed insulin supplementation soon after the start of therapy (Fig. 1A and B). Healthy control rats stayed normoglycemic during the entire observation period (Fig. 1B).

In the successfully treated animals with the combination therapy, serum C-peptide concentrations increased (Fig. 1C). At the end of therapy and more so 60 days later, C-peptide was significantly (P < 0.05) higher with combination therapy than with anti–TNF-α monotherapy. This continuous increase reached about two-thirds of control values (726 ± 79 vs. 1023 ± 41 pmol/L) at the end of the observation period (Fig. 1C). The untreated diabetic rats had C-peptide concentrations <20% at the time of T1D manifestation (Fig. 1C). C-peptide remained low after anti–TNF-α monotherapy and in animals nonresponsive to combination therapy (Fig. 1C).

A detailed analysis revealed interesting results. Six rats in the monotherapy group had blood glucose values at the start of therapy between 10.0 and 15.0 mmol/L (12.5 ± 0.5 mmol/L), and the other six values were above 15.0 mmol/L (16.7 ± 0.4 mmol/L) (Fig. 2A). But in none of the 12 rats did monotherapy restore β-cell mass. In the two subgroups with combination therapy, the blood glucose concentrations were also significantly (P < 0.05) different. In all rats that responded to combination therapy, blood glucose values were below 15 mmol/L (11.4 ± 0.3 mmol/L; n = 8) but were above 15 mmol/L in all nonresponding rats (16.4 ± 0.4 mmol/L; n = 4) (Fig. 2B).

The analyses of the percentage changes, as presented in Fig. 3A and B, revealed drastic increases of the rates for both parameters in the diabetic rats. At the day of T1D manifestation, the rats that were responsive to the combination therapy (8 of 12) showed significantly more than a 4-fold increase of the proliferation rate (Fig. 3A), and the apoptosis rate analyzed by TUNEL increased more than 15-fold (Fig. 3B) compared with normoglycemic control rats. In the nonresponding rats (4 of 12), the proliferation rate also increased 4-fold (Fig. 3A), but the apoptosis rate increased even more than 20-fold (Fig. 3B).

The 12 rats on anti–TNF-α monotherapy, all nonresponding irrespective of the blood glucose concentration at start of therapy (Fig. 2B), showed a 5-fold increase of the proliferation rate (Fig. 3A) at the day of T1D manifestation and a 17-fold increase of the apoptosis rate (Fig. 3B).

Immediately after the end of combination therapy, the proliferation rate was still as high as before the start of therapy in the responding rats (Fig. 3A). The apoptosis rate was strongly reduced, being close to that of healthy control animals (Fig. 3B).

In the nonresponding rats, proliferation and apoptosis rates were very low after anti–TNF-α therapy and after combination therapy due to the nearly complete β-cell loss in the islets (Fig. 3A and B).

At 60 days after the end of therapy, rats successfully treated with the combination therapy showed virtually normal proliferation and apoptosis rates, as in healthy controls (Fig. 3A and B). All β-cells were lost after anti–TNF-α monotherapy (Fig. 3A and B) and after combination therapy in the nonresponding rats (Fig. 3A and B).

Calculation of the proliferation-to-apoptosis ratios revealed an increase to values in the range of the healthy control rats only in rats responding to therapy, but the increase was persistent only after combination therapy but not anti-TCR monotherapy. Anti–TNF-α monotherapy did not even cause a transient ratio increase (Supplementary Table 4).

On the day of T1D manifestation, before the start of therapy, the infiltration score of the islets was >3 (Fig. 3C). The score was reduced to values of ∼2 at the end of combination therapy for the responding and nonresponding animals (Fig. 3C). At 60 days after the end of combination therapy, the infiltration score in the pancreases with the regenerated β-cells was reduced to <0.1, with only single nonactivated macrophages being left in the responding animals (Fig. 3C). In the nonresponding rats, islet infiltration was absent after anti–TNF-α as well as after combination therapy due to the complete loss of β-cells (Fig. 3C).

On the day of T1D manifestation, before the start of therapy, the β-cell mass of the pancreases in all rats was reduced to approximately one-third of the value in the control rats (Fig. 3D). Immediately after the end of anti–TNF-α monotherapy, the pancreatic β-cell mass was reduced even more, to <10%. But immediately after the end of successful combination therapy, the pancreatic β-cell mass had more than doubled from the value before therapy, reaching in the responding animals >80% of the value in the healthy control rats (Fig. 3D). At 60 days after the end of combination therapy, the pancreatic β-cell mass had attained normal values in these rats (Fig. 3D). In the nonresponding animals, β-cell mass was further reduced strongly already during therapy, and β-cells were completely lost 60 days thereafter (Fig. 3D).

A deeper analysis of the situation immediately before the start of therapy revealed that the percentage of insulin-positive β-cells in the pancreases was comparably high in all therapy groups (Fig. 3D), irrespective of a blood glucose value <15 mmol/L or >15 mmol/L at the start of therapy. However, the percentage of insulin-positive β-cells that were positive in addition for caspase 3 and perforin 1 was approximately three times higher when the blood glucose concentration at the start of therapy was >15 mmol/L (Table 1), indicating an accelerated β-cell death rate of the residual β-cells in the islets of the pancreases of diabetic rats with blood glucose >15 mmol/L.

The correlation coefficients between an increase of blood glucose and an increase of caspase 3 and perforin 1 were r = 0.947 and r = 0.953, respectively (both P < 0.001), for the values (n = 24) in Table 1.

On the day of T1D manifestation, a strong immune cell infiltration composed of 45% CD8 T cells, 5% CD4 T cells, 40% CD68 macrophages, and 10% of other immune cell types was observed in the islets in responding and nonresponding rats, with a concomitant loss of β-cells (Fig. 4A–D and Supplementary Fig. 1A and B, respectively) (3).

Immediately after the end of combination therapy with anti–TNF-α and anti-TCR, immune cell infiltration was still substantially reduced by >50% in responding animals compared with the situation before the start of therapy (Fig. 4A–D). Immune cell infiltration was also reduced in the nonresponding rats; however, this went along with a loss of β-cells (Supplementary Fig. 1A and B). At 60 days after the end of therapy, islet immune cell infiltration was entirely absent after combination therapy in the responding rats (Fig. 4A–D) in pancreases with a normal islet β-cell mass not different from that in healthy control animals. In pancreases of nonresponding rats with islets lacking β-cells, only occasional single nonactivated immune cells were observed (Supplementary Fig. 1A and B).

Immune cells were activated in diabetic rats. This was documented by high gene expression levels of the proinflammatory cytokines, Tnf and Il1b (Fig. 5A and B), as well as of the CD8 marker, Gzmb, in the infiltrating immune cells (Fig. 5C), and of the proapoptotic enzyme, Casp3, in the β-cells (Fig. 5D) before start of therapy. At 60 days after the end of combination therapy, an expression of these genes was virtually completely lost in the noninfiltrated islets of the responding rats (Fig. 5A-D). In both groups nonresponding to therapy, a few activated immune cells were observed, but only in the perivascular space, which was not surprising in view of the complete β-cell loss (Supplementary Fig. 1C and D).

In the animals responding to combination therapy, the serum protein concentrations of the proinflammatory cytokines TNF-α and IL-1β were as low as in healthy control animals (Fig. 6A and B). Both in animals nonresponding to monotherapy and combination therapy, TNF-α and IL-1β serum protein concentrations remained significantly (P < 0.01) increased (Fig. 6A and B). The proinflammatory cytokine interferon-γ (IFN-γ) was unchanged in all groups compared with the control groups (Fig. 6C). The increased levels of the immune cell–activating cytokine IL-2 in diabetic rats decreased only in animals responding to combination therapy with a significant (P < 0.05) reduction compared with the animals nonresponsive to the therapies (Fig. 6D). Only after successful combination therapy was the protein concentration of the anti-inflammatory cytokine IL-4 increased, reaching levels close to those of the healthy rats (Fig. 6E). After anti–TNF-α therapy, alone and combined with anti-TCR, the concentrations of the anti-inflammatory cytokine IL-10 were significantly (P < 0.01) increased, irrespective of therapy success (Fig. 6F).

The same changes of the pro- and anti-inflammatory cytokines as well as of the CD8 T-cell markers granzyme B and perforin 1 on the level of gene (Supplementary Table 5) and protein (Supplementary Table 6) expression were observed in the immune cells in the pancreas-draining lymph nodes 60 days after the end of therapy in the responding and the nonresponding rats.

This therapy combines two antibodies, anti-TCR directed against the TCR/CD3 complex, as a cornerstone compound in anti–T-cell therapy, and anti–TNF-α, the most effective therapeutic antibody in many autoimmune diseases, based on considerations of the pathophysiological understanding of the disease process of T1D in the endocrine pancreas (3). This combination provided therapy success without inclusion of other compounds such as immunomodulatory agents (i.e., anti–IL-1β, proinsulin, IL-10, fingolimod) (25,28,37).

The idea behind this therapy was that a combination of antibodies is required to reverse autoimmunity in the T1D state, because none of the antibodies is able to target all proinflammatory cytokines in the different immune cell types of the infiltrate. Therefore, monotherapies have proven not to be particularly successful (23,38–40) even with the most promising antibodies, anti-CD3 in humans and mice and in analogy anti-TCR in rats (25,27,28,41–44). We report here a successful combination therapy for T1D in the IDDM rat, an animal model of human T1D (25), using two antibodies with different target profiles.

The proinflammatory cytokine TNF-α, expressed in the infiltrating immune cells, is known to be a key player in the pathology of many autoimmune diseases, including T1D (9,12,13). However, although a monotherapy with anti–TNF-α is an established successful therapy in a number of autoimmune diseases (15–17), anti–TNF-α is not effective in T1D treatment (18–23). Anti–TNF-α neutralizes TNF-α, produced by the immune cells infiltrating the target organ during the inflammatory process (11,14,15,45), and the combination of anti–TNF-α with anti-TCR also prevents the release of IL-1β, the other critical proinflammatory cytokine for the disease process (3), from the activated T cells (25). In addition, we found that successful combination therapy also suppressed other inflammatory mediators, such as perforin 1 and granzyme B, in T cells.

The results of the current study show that anti–TNF-α therapy combined with anti-TCR, started immediately after T1D manifestation in the rats (30), was an effective strategy in reversing diabetic hyperglycemia. This was documented by normalization of glycemia, serum C-peptide, and β-cell mass. At the same time, immune cell infiltration in the islets was completely abolished. Monotherapy with anti–TNF-α was not successful in the diabetic animals, confirming earlier studies (18,46). Anti-TCR monotherapy did also not yield a sustained success under the same therapy regimen, as we reported in detail in a previous study in this rat model (25). Even the transient reduction of hyperglycemia in 3 of 12 animals was not sustained (25), a phenomenon also observed in human studies with anti-CD3 (33,34) and confirmed in the current study by the comparative calculations of the apoptosis-to-proliferation ratios under the different treatment regimens (Supplementary Table 4).

Interestingly, reversal of the T1D state was achieved only in those animals with a blood glucose concentration <15 mmol/L, whereas the one-third of the T1D animals with a blood glucose concentration >15 mmol/L at the start of combination therapy remained diabetic (Fig. 2). This observation (Fig. 2) documents strikingly how important an early start of combination therapy is, as soon as possible after diagnosis of the disease, as also concluded in recent reviews (8,23). The explanation for this observation is provided by the experiments reported in Table 1, showing that the percentage of insulin-positive β-cells, which were in addition also positive for caspase 3 and for perforin 1, was approximately three times higher at the start of therapy, when blood glucose concentrations were >15 mmol/L, than at blood glucose concentrations below this threshold (Table 1). This demonstrates that with increasing blood glucose concentrations, the percentage of dying insulin-positive β-cells increases (47). This correlation is highly significant. It also explains that at a starting blood glucose concentration >15 mmol/L, the number of the remaining healthy insulin-positive β-cells is insufficient to secure a regenerative capacity, allowing the generation of a sufficient number of new β-cells to regain normoglycemia through replication (48,49). This capacity of the β-cells to regenerate and normalize β-cell mass in rats with blood glucose <15 mmol/L could be recovered under combination therapy but not under monotherapy, convincingly emphasizing the crucial importance of the antibody combination for therapy success.

The changes in different cytokines and chemokines in the circulation, as observed under therapy in the current study, mirrored the corresponding changes in the infiltrated islets and thus represent reliable biomarkers to monitor disease progression and therapeutic success in this T1D rat model. This could not be documented so far in patients with T1D (50). In an experimental approach, such as the one used in the IDDM rat in the present and in previous studies for mono- and combination therapies (25,36), pancreatic tissue could be directly accessed for β-cell analyses by biopsies. Restitution of normoglycemia and serum C-peptide induced by β-cell mass restoration in the pancreas correlated with a normalization of the cytokine and chemokine protein expression profile in serum and pancreas-draining lymph nodes. This was documented by a reduction of the increased levels of the proinflammatory cytokines TNF-α and IL-1β and the chemokine (C-C motif) ligand-2 (MCP-1) as well as the T-cell markers perforin 1 and granzyme B to the normal range typical for the healthy state along with an increase of the anti-inflammatory cytokine IL-4.

This profile normalization in the circulation not only accompanied the loss of pancreatic islet infiltration by activated immune cells but also went along with a normalization of the cytokine and chemokine protein expression profile in the pancreas-draining lymph nodes. Thus, changes in the protein expression profile in serum are reliable biomarkers for documentation of therapy success achieved in the endocrine pancreas and in the neighboring lymph nodes.

The results can be summarized as follows:

1. A sufficient number of intact β-cells is required to allow a reversal of the diabetic state through proliferation of existing β-cells.

2. Therefore, an immediate start of therapy after T1D diagnosis is necessary to secure therapy success.

3. The proinflammatory cytokines TNF-α and IL-1β, both highly expressed in the activated immune cells of the islet infiltrate in the T1D pancreas, are targeted, thereby securing therapy success of this combination.

4. The results show convincingly that antibodies, which are not effective alone, can reverse T1D in combination therapy successfully.

Combination therapy might thus be an attractive option in the future because monotherapy prevention studies have been ineffective not only in the IDDM rat model but also in patients with T1D (30,38–40), as advocated by several scientists recently (6–8). Because the potential to optimize therapy success through dose increases of anti-CD3 and anti-TCR is limited due to T-cell reduction in the circulation, which develops under this therapy, combination with other immunomodulatory compounds might be the best solution to minimize such undesirable adverse effects.

The future will probably also see therapies with combinations of more than two agents. This could be, for example, a combination of anti-TCR/CD3 with anti–TNF-α, which both target the immune cells in the infiltrate directly, with a compound such as fingolimod (25,36), which prevents the migration of activated lymphocytes into the islets from the lymph nodes that surround the pancreas. Therapy success can thereby be maximized along with a reduction of the doses of the different compounds and a minimization of their adverse effects. This is likely to be a better option than the attempt to identify subgroups of patients with T1D who might respond, even only transiently, to anti-CD3 monotherapy (51).

Acknowledgments. The authors thank M. Jentzsch and D. Lischke (both from the Insitute of Clinical Biochemistry, Hannover Medical School, Hannover, Germany) for skillful technical assistance.

Funding. This work was supported by grants from Deutsche Forschungsgemeinschaft (JO 395/2-1) and the European Union (Collaborative Project NAIMIT in the 7th Framework Programme, Contract No. 241447).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.J. designed the study, performed experiments, analyzed and interpreted data, and wrote the manuscript. Ü.G.E. performed experiments, analyzed data, and contributed to writing the manuscript. T.A. and T.T. performed experiments. D.W. provided materials and reviewed the manuscript. S.L. designed the study, analyzed and interpreted data, and wrote the manuscript. A.J. is the guarantor of this work, and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of data analysis.