Partial and transient modulation of the CD3–T-cell receptor complex, elicited by low-dose regimens of monoclonal anti-CD3, is sufficient to induce disease remission in non-obese diabetic mice

Abstract

It has been established that a total of 250 μg of monoclonal anti-mouse CD3 F(ab′)₂ fragments, administered daily (50 μg per dose), induces remission of diabetes in the non-obese diabetic (NOD) mouse model of autoimmune diabetes by preventing β cells from undergoing further autoimmune attack. We evaluated lower-dose regimens of monoclonal anti-CD3 F(ab′)₂ in diabetic NOD mice for their efficacy and associated pharmacodynamic (PD) effects, including CD3–T-cell receptor (TCR) complex modulation, complete blood counts and proportions of circulating CD4⁺, CD8⁺ and CD4⁺ FoxP3⁺ T cells. Four doses of 2 μg (total dose 8 μg) induced 53% remission of diabetes, similarly to the 250 μg dose regimen, whereas four doses of 1 μg induced only 16% remission. While the 250 μg dose regimen produced nearly complete and sustained modulation of the CD3 –TCR complex, lower doses, spaced 3 days apart, which induced similar remission rates, elicited patterns of transient and partial modulation. In treated mice, the proportions of circulating CD4⁺ and CD8⁺ T cells decreased, whereas the proportions of CD4⁺ FoxP3⁺ T cells increased; these effects were transient. Mice with greater residual β-cell function, estimated using blood glucose and C-peptide levels at the initiation of treatment, were more likely to enter remission than mice with more advanced disease. Thus, lower doses of monoclonal anti-CD3 that produced only partial and transient modulation of the CD3–TCR complex induced remission rates comparable to higher doses of monoclonal anti-CD3. Accordingly, in a clinical setting, lower-dose regimens may be efficacious and may also improve the safety profile of therapy with monoclonal anti-CD3, potentially including reductions in cytokine release-related syndromes and maintenance of pathogen-specific immunosurveillance during treatment.

Introduction

Extensive preclinical and clinical experience supports the rationale for treatment of patients with new-onset autoimmune type 1 diabetes with monoclonal antibodies (mAbs) raised against CD3 (monoclonal anti-CD3). Monoclonal anti-CD3 appear to arrest ongoing disease by down-regulating or clearing pathogenic T cells from the pancreatic islets and promoting long-term T-cell-mediated active tolerance, probably by up-regulating or inducing T-regulatory (Treg) cells that can prevent further autoimmune attack.^1–4 The potential efficacy of monoclonal anti-CD3 therapy for type 1 diabetes was first demonstrated in the non-obese diabetic (NOD) mouse model of spontaneous autoimmune diabetes and in the transgenic rat insulin promoter-lymphocytic choriomeningitis virus glycoprotein (RIP-LCMV-GP) mouse model of virus-induced autoimmune diabetes, where tolerance to pancreatic islets and durable remission were induced.^1,5,6 Fc-intact monoclonal anti-mouse CD3 was used in initial murine studies, but induced severe morbidity and mortality as a result of cytokine-release syndrome, mediated through engagement of the Fc receptor (FcR).^7–9 It was subsequently demonstrated that FcR engagement is not required for efficacy because F(ab′)₂ fragments of the mAb, which lack the Fc region, induced disease remission without systemic cytokine release.^1,9,10

Based on this preclinical evidence, minimization of FcR binding has been a priority in the development of partially or fully humanized monoclonal anti-CD3. Otelixizumab, a chimeric humanized monoclonal anti-CD3, has been modified by amino acid substitution to eliminate the N-linked glycosylation site in the CH2 domain of the Fc region, resulting in significantly reduced FcR binding and the inability to fix complement.^11–15 Cytokine release in subjects administered otelixizumab is significantly reduced compared with cytokine release in subjects administered OKT3, an Fc-intact monoclonal anti-CD3.^13,14 In a Phase 2 trial conducted by the Belgian Diabetes Registry (BDR), subjects with new-onset type 1 diabetes who received a single 6-day course of otelixizumab (total dose 48–64 mg) had significantly greater endogenous insulin production than subjects who received placebo, and this effect was durable for at least 48 months.^14,16 Preliminary clinical activity in new-onset type 1 diabetes has also been demonstrated with teplizumab, another Fc-modified monoclonal anti-CD3.¹⁷

Upon the administration of monoclonal anti-CD3, antibody rapidly binds the CD3 molecule and is internalized, resulting in modulation of the CD3–T-cell receptor (TCR) complex. Loss of CD3–TCR complex expression is reversible, as it recycles back to the surface after clearance of the antibody. Binding and subsequent modulation of the CD3–TCR complex by monoclonal anti-CD3 is considered to be pharmacodynamically important and is routinely assessed in clinical studies evaluating monoclonal anti-CD3 therapies. This pharmacodynamic (PD) effect potentially impacts the mechanism of action of monoclonal anti-CD3 in at least two ways: (i) temporarily blocking antigen binding; and (ii) delivering a partial agonist signal, which may induce anergy of autoreactive T cells while allowing for the expansion of Treg cells (reviewed in^2,18).

In the Phase 2 BDR study of otelixizumab, profound and sustained modulation of the CD3–TCR complex occurred on the first day of dosing and persisted through the 6-day dosing period.¹⁴ In the mouse, there are limited data evaluating dose responses with monoclonal anti-mouse CD3 F(ab′)₂ or examining modulation of the CD3–TCR complex during treatment and its potential correlation with efficacy. We performed dose-ranging studies in diabetic NOD mice to determine the minimum effective dose of monoclonal anti-CD3 F(ab′)₂. CD3–TCR complex-modulation patterns elicited during antibody administration were assessed to determine whether nearly complete and sustained modulation is required for efficacy of monoclonal anti-CD3 therapy. We demonstrated that doses resulting in partial and transient modulation of the CD3–TCR complex are sufficient to induce remission in diabetic NOD mice, such that doses more than 30-fold less than the originally published 250 μg regimen resulted in similar rates of remission. We also examined PD effects on lymphocyte counts and circulating T-cell subsets, demonstrating that the efficacy of treatment with monoclonal anti-CD3 is associated not only with PD changes anticipated based on the mechanism of action of the mAb, but also with residual β-cell function at the time of treatment.

Materials and methods

Mice

BALB/c mice (Harlan, Boston, MA) were used in Study A. Female NOD/ShiLtJ mice (Jackson, Bar Harbor, ME) were used in Study B; NOD/ShiLtJ mice were bred at Tolerx under pathogen-free conditions for use in Study C.

Antibodies

Hamster monoclonal anti-(mouse CD3) (clone 145-2C11; ATCC) was purified using protein G affinity chromatography (GE Healthcare, Piscataway, NJ) and formulated in Dulbecco’s phosphate-buffered saline (PBS). Monoclonal anti-CD3 F(ab′)₂ fragments were generated by digestion with pepsin (Sigma, St Louis, MO) for 17 hr at 37° in acetic acid, pH 4·0. The reaction was quenched with 2 m Tris and dialysed against PBS overnight at 2–8°. F(ab′)₂ fragments were further purified by size-exclusion chromatography. Purity was assessed by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and found to be 90% of total integrated density with no intact antibody. The F(ab′)₂ preparation included ≤ 3 endotoxin units/ml, as measured using the Pyrotell gel-clot assay (Associates of Cape Cod, East Falmouth, MA).

Treatment with monoclonal anti-CD3

In Study A, BALB/c mice were dosed with the following regimens: five doses of 50 μg every 24 hr (total dose 250 μg); four doses of 25 μg every 72 hr (total dose 100 μg); four doses of 5 μg every 72 hr (total dose 20 μg); four doses of 2 μg every 72 hr (total dose 8 μg); and four doses of 1 μg every 72 hr (total dose 4 μg). In Study B, NOD/ShiLtJ mice were administered the following dose regimens: five doses of 50 μg every 24 hr (total dose 250 μg); four doses of 25 μg every 72 hr (total dose 100 μg); three doses of 25 μg every 72 hr (total dose 75 μg); four doses of 5 μg every 72 hr (total dose 20 μg); and three doses of 5 μg every 72 hr (total dose 15 μg). In Study C, NOD/ShiLtJ mice were administered the following dose regimens: three doses of 5 μg every 72 hr (total dose 15 μg); four doses of 2 μg every 72 hr (total dose 8 μg); and four doses of 1 μg every 72 hr (total dose 4 μg). Each study also included a vehicle (PBS) control. All doses were delivered intraperitoneally (i.p.).

In Studies B and C, blood glucose levels were measured twice weekly in female NOD/ShiLtJ mice. Mice with two consecutive blood glucose level readings of > 250 mg/dl were considered to have new-onset diabetes and were enrolled in the study such that variation in age at disease onset was represented equally across dose regimens. After treatment, the blood glucose level was measured weekly. Remission was defined as a return to normal glycaemia in the absence of exogenous insulin.

Immunogenicity assay

An enzyme-linked immunosorbent assay (ELISA)-based assay was developed to determine whether an immunogenic response towards the monoclonal anti-CD3 F(ab′)₂ had been induced in mice treated with monoclonal anti-CD3 F(ab′)₂. Maxisorp 98-well plates (Nunc, Rochester, NY) were coated with monoclonal anti-CD3 F(ab′)₂. Following incubation with mouse serum from treated mice, mouse antibodies specific for monoclonal anti-CD3 F(ab′)₂ were detected using donkey anti-mouse IgG (H + L) (horseradish peroxidase-conjugated, minimally cross-reactive to hamster species; Jackson ImmunoResearch, West Grove, PA). ELISAs were developed using o-phenyl diamine dihydrochloride (OPD) substrate (Sigma) in sodium citrate buffer, pH 5, plus H₂O₂. H₂SO₄ (12·5%) was used to stop the OPD reaction, and plates were read at 490 nm using Softmax™ Pro software (MDS Analytical Technologies, Sunnyvale, CA).

Flow cytometry analysis

Modulation of the CD3–TCR complex in peripheral blood was analyzed by flow cytometry 2 and 24 hr after each dose when mice were dosed every 24 hr and 2 and 72 hr after each dose when mice were dosed every 72 hr. Following red blood cell lysis, cells were stained using murine antibodies to CD3 (145-2C11), CD4 (RM4-5), CD8 (53-6.7) and TCR-β (H57-597) (BD Biosciences, San Jose, CA). Molecules of equivalent soluble fluorochrome (MESF) values were generated using Quantam™ fluorescein isothiocyanate (FITC) MESF microspheres as per the manufacturer’s instructions (Bangs Laboratories, Fisher, IN). FoxP3 expression was evaluated using a FoxP3 staining kit (NRRF30 clone; eBioscience, San Diego, CA), as per the manufacturer’s instructions. Fluorescent cells were analyzed by flow cytometry using FACScalibur (BD Biosciences).

Analysis of C-peptide levels in serum

In Study B, serum was collected before and after treatment and analyzed for the murine C-peptide I content by ELISA, according to the manufacturer’s instructions (ALPCO, Salem, NH).

Pancreatic histology

In Study B, pancreata were fixed in formalin, processed and embedded in paraffin. Sections of 4–5 μm in thickness were stained with haematoxylin and eosin. Islet inflammation was evaluated using light microscopy by a board-certified veterinary pathologist (Charles River Laboratories, Wilmington, MA). Peri-insulitis inflammation was scored as: 0 = normal (no leucocytes); 1 = minimal (< 5 leucocytes in any islet); 2 = mild (6–20 leucocytes in the ‘most severe’ islet); 3 = moderate (21–50 leucocytes in the ‘most severe’ islet); 4 = marked (> 50 leucocytes in the ‘most severe’ islet); or 5 = severe (> 50 leucocytes in > 1 islet).

Statistical methods

MESF values were analyzed using repeated-measures analysis of variance (anova), with treatment and time as factors. Lymphocyte count data were analyzed by one-way anova. Pairwise treatment group comparisons for these analyses were carried out using the corresponding t-tests. Fisher’s exact test was used for pairwise treatment group comparisons of proportion data. Exploratory comparisons between post-treatment remission and diabetic groups were made using t-tests (quantitative data), Fisher’s exact test (proportion data), or the chi-square test (categorical data). P-values were not adjusted for multiple comparisons.

Results

Modified dose regimens result in transient and partial modulation of the CD3–TCR complex

In previous preclinical studies, a regimen of 250 μg of monoclonal anti-CD3 F(ab′)₂, 50 μg per day for 5 consecutive days [50 μg (5×/24 hr)], resulted in a 67% remission rate in new-onset diabetic NOD mice,¹⁰ but there are limited data evaluating the PD effects during dosing. Given that it has been previously demonstrated that the biological effects of the antibody are similar in NOD and non-autoimmune mice,^7,9,10,19 we elected to first examine the PD effects of monoclonal anti-CD3 F(ab′)₂ on modulation of the CD3–TCR complex in BALB/c mice in Study A. TCR expression on peripheral blood CD4⁺ and CD8⁺ lymphocytes was analyzed 2 and 24 hr after each dose. The resulting patterns of TCR expression on both CD4⁺ and CD8⁺ lymphocytes were equivalent; therefore, only CD4⁺ lymphocytes are shown in Fig. 1. In the first segment, the well-established dose regimen of 50 μg (5×/24 hr) of monoclonal anti-CD3 F(ab′)₂ was evaluated. Expression of the CD3–TCR complex was reduced 2 hr after the first dose and remained almost completely down-regulated before the second dose. These low levels of expression of the CD3–TCR complex were sustained throughout dosing (Fig. 1a), similar to the pattern observed in the BDR clinical trial where high-dose regimens of otelixizumab were evaluated.¹⁴ Expression of the CD3–TCR complex was partially restored within 72 hr following the end of dosing and returned to baseline within 10 days of the last dose.

Figure 1.

Modulation of the CD3–T-cell receptor (TCR) complex on circulating T cells during treatment with monoclonal anti-CD3 F(ab′)₂ in Study A. Mean (± standard deviation) TCR expression levels are presented as molecules of equivalent soluble fluorochrome (MESF) units. (a) BALB/c mice (n= 3 per dose regimen) received 50 μg of monoclonal anti-CD3 F(ab′)₂ each day for 5 consecutive days. TCR expression was evaluated on circulating CD4⁺ lymphocytes, 2 hr (post-dose) and 24 hr (pre-dose) after each dose. (b) BALB/c (n= 3 per dose regimen) mice received four doses of 25, 5, 2, or 1 μg of monoclonal anti-CD3 F(ab′)₂ or vehicle control every 72 hr. TCR expression was evaluated on circulating CD4⁺ lymphocytes 2 hr (post-dose) and 72 hr (pre-dose) after each dose. At the pre-dose 4 and post-dose 4 time-points, differences in TCR expression levels between the 1 and 2 μg dose regimens were significant; P < 0·05 and P < 0·01 respectively.

Because the 50 μg (5×/24 hr) dose regimen resulted in nearly complete and sustained modulation of the CD3–TCR complex, we were interested in developing and evaluating dose regimens that would elicit a partial and transient pattern of modulation. First, lower doses of monoclonal anti-CD3 F(ab′)₂ were evaluated. TCR expression was measured in BALB/c mice administered five doses of 25, 5, 2, or 1 μg of monoclonal anti-CD3 F(ab′)₂, 24 hr apart. The 25 μg (5×/24 hr) dose regimen resulted in profound and sustained modulation of the CD3–TCR complex, similar to the 50 μg (5×/24 hr) dose regimen (data not shown). Lower doses produced dose-dependent reductions in modulation of the CD3–TCR complex, but a sustained level of modulation was observed in all dose regimens (data not shown). This suggested that to achieve a pattern of transient modulation of the CD3–TCR complex, it would be necessary to space the doses further apart. We next determined how soon after dosing the surface expression of the CD3–TCR complex returned to baseline levels in the mouse. After a single 25 μg dose of monoclonal anti-CD3 mAb F(ab′)₂, expression of the CD3–TCR complex was markedly down-regulated at 24 hr; showed signs of recovery, but was still significantly down-regulated at 48 hr; and recovered to near-baseline values at 72 hr (data not shown).

In the second segment of Study A, a range of doses of monoclonal anti-CD3 F(ab′)₂ (1, 2, 5 and 25 μg) was administered four times, 72 hr apart, given that a fifth dose resulted in anti-drug antibodies in three out of six mice (detected using an ELISA-based assay). The mice did not develop any adverse events associated with immunogenicity to the monoclonal anti-CD3 F(ab′)₂. The 72 hr dose regimen resulted in transient, and sometimes partial, modulation of the CD3–TCR complex that was clearly dose-dependent (Fig. 1b). The 5 and 25 μg (4×/72 hr) dose regimens produced ‘saw-tooth’ patterns, where expression of the CD3–TCR complex was quickly down-regulated after each dose but returned to near predose values before the subsequent dose. With each successive dose, the level of modulation of the CD3–TCR complex increased. In the 2 and 1 μg (4×/72 hr) dose regimens, the extent of modulation was considerably less than in other dose regimens and was clearly discernable only after the fourth dose (Fig. 1b). After the fourth dose, the difference in the percentage of modulation of the CD3–TCR complex between the 2 and 1 μg (4×/72 hr) dose regimens was significant (30·3% versus 19·7% modulation, P < 0·01). Furthermore, in all dose regimens, there was a transient decrease in lymphocyte numbers in the peripheral blood during and shortly after dosing (Fig. 2), consistent with what has been observed in the spleens of both NOD and non-autoimmune mice administered monoclonal anti-CD3 F(ab′)₂.^9,10,19 This observation of lymphopenia during dosing could be the result of either depletion of a subset of lymphocytes or retrafficking of monoclonal anti-CD3 F(ab′)₂-bound lymphocytes from the peripheral blood.

Figure 2.

Lymphocyte counts during treatment with monoclonal anti-CD3 F(ab′)₂ in Study A. Complete blood counts were performed 2 hr after administration of the last dose. The lymphocyte count (K/μl) is the mean from three to five mice per dose regimen; error bars represent the standard deviation. All dose regimens, with the exception of the 25 μg dose regimen, were significantly different from the vehicle group (*P < 0·05). There was no significant difference between the 1 and 2 μg dose-regimens.

Lower doses of monoclonal anti-CD3 F(ab′)₂ are efficacious in new-onset diabetic NOD mice

In Study B, the effectiveness of the various dose regimens in inducing remission of diabetes was investigated in new-onset diabetic NOD mice. In order to evaluate whether a shorter duration of modulation of the CD3–TCR complex or a lower cumulative dose affects efficacy, Study B also included groups given only three doses. Animals were randomly enrolled into one of five monoclonal anti-CD3 F(ab′)₂ dose regimens – 50 μg (5×/24 hr), 25 μg (4×/72 hr), 25 μg (3×/72 hr), 5 μg (4×/72 hr), or 5 μg (3×/72 hr) – or placebo. The 25 and 5 μg doses were chosen based on the results of Study A, in which expression of the CD3/TCR complex, 24 hr after dose 4, was approximately 12% and 50% of baseline, respectively. No animals in the placebo group entered remission during the 12-week observation of blood glucose levels. In all dose regimens, approximately half of the mice (44–60%) had long-term remission (Table 1). There was no statistically significant difference in remission rates between the various dose regimens. The well-established 50 μg (5×/24 hr) dose regimen resulted in 56% of the mice being in remission for 12 weeks, which is similar to the originally published 67% remission rate.¹⁰ There was no apparent relationship between the dose and the rate of remission. As in previous studies,¹⁰ the majority of mice in all dose regimens that entered remission did so 1–2 weeks after treatment and all remained in remission for the 12 weeks of follow-up.

Table 1.

Remission rates of new-onset diabetic female non-obese diabetic (NOD)/ShiLtJ mice treated with monoclonal anti-CD3 F(ab′)₂ fragments

Dose regimen (μg)	Total dose (mg/kg)	Total dose as a multiple of 250 μg	Per cent (number) experiencing remission	Duration of remission
Study B
50 μg × 5/24 hr (250)	250 μg (0·0125)	1	56 (9/16)	12 weeks1
25 μg × 4/72 hr (100)	100 μg (0·005)	1/2·5	50 (8/16)	12 weeks1
25 μg × 3/72 hr (75)	75 μg (0·0038)	1/3·3	50 (8/16)	12 weeks1
5 μg × 4/72 hr (20)	20 μg (0·001)	1/12·5	60 (9/15)	12 weeks1
5 μg × 3/72 hr (15)	15 μg (0·00075)	1/16·6	44 (7/16)	12 weeks1
Placebo	–	–	0 (0/15)	N/A
Study C
5 μg × 3/72 hr (15)	15 μg (0·00075)	1/16·6	63 (5/8)	12 weeks
2 μg × 4/72 hr (8)	8 μg (0·0004)	1/31·3	53 (10/19)*	12–24 weeks
1 μg × 4/72 hr (4)	4 μg (0·0002)	1/62·5	16 (3/19)*	12–24 weeks
Placebo	–	–	0 (0/13)	N/A

¹After 12 weeks of remission, mice were killed for histological analysis.

In Study B, mice were killed 12 weeks after treatment. In Studies B and C, remission rates were determined 12 weeks post-treatment. In Study C, mice in remission were not killed and the range of durability was assessed up to 24 weeks post-treatment for the 2 and 1 μg dose regimens. Differences in remission rates between the 50, 25, 5 and 2 μg dose regimens in both Studies B and C were not significant; however, the difference between the 2 and 1 μg dose regimens was significant (*P < 0·05).

N/A, not applicable.

Study B demonstrated that a total dose as low as 15 μg resulted in long-term remission of diabetes in NOD mice. In Study C, lower doses were examined to determine the minimum effective dose with the 72 hr dose regimen. Also, antibody-treated mice in Study C were followed for at least 12 weeks after treatment to determine the durability of remission and up to 24 weeks after treatment in the lowest dose regimens. The lowest dose regimen from Study B, 5 μg (3×/72 hr), was repeated, and two lower dose regimens, 2 μg (4×/72 hr) and 1 μg (4×/72 hr), were added. The 5 μg (3×/72 hr) and 2 μg (4×/72 hr) dose regimens had remission rates of 63% and 53%, respectively, similar to the higher dose regimens in Study B. Again, there was no statistically significant difference in remission rates between the 5 μg (3×/72 hr) and 2 μg (4×/72 hr) dose regimens in Study C, or the various dose regimens in Study B. As in the higher dose regimens in Study B, these mice entered remission 1–2 weeks after treatment and the remission was long-lasting, up to 24 weeks of follow-up. However, at the 1 μg (4×/72 hr) dose regimen, the remission rate dropped to 16% and this reduction was significantly different compared with the 2 μg (4×/72 hr) dose regimen (P < 0·05). Yet, for mice that did enter remission, the remission was long-term (up to 24 weeks). Thus, the minimum effective dose of monoclonal anti-CD3 F(ab′)₂ for the 4×/72 hr dose regimen is ≥ 1 μg.

In both Studies B and C, partial remission was observed in one or two mice within each dose regimen, such that normal glycaemia was detected in these mice for a transient period ranging from 3 to 11 weeks post-treatment. Thereafter, the blood glucose levels rose quickly and were sustained at levels of ≥ 250 mg/dl. There was no correlation between dose and the numbers of mice exhibiting partial remission. Overall, all of the mice that entered remission did so within 1–2 weeks after treatment, consistent with previous studies,¹⁰ and the majority of remissions observed were durable for at least the 12-week observation period.

Treatment with monoclonal anti-CD3 F(ab′)₂ alters the proportions of T-cell subsets

In addition to modulation of the CD3–TCR complex, the PD parameters routinely assessed in clinical studies of otelixizumab include changes in various immune-cell subsets such as CD4⁺, CD8⁺ and CD4⁺ FoxP3⁺ T cells. Because we wanted to mirror the PD parameters routinely collected in clinical situations, we specifically elected to evaluate similar flow-cytometric PD parameters in the peripheral blood of mice from Studies B and C. In Studies B and C, the proportions of CD4⁺, CD8⁺ and CD4⁺ FoxP3⁺ T cells were assessed before dosing and again within 24 hr of the last dose. We elected to use the CD4⁺ FoxP3⁺ phenotype to identify Treg cells in the periphery, given that FoxP3 expression directly correlates with Treg-cell function, regardless of the CD25 expression levels²⁰ and because CD25 is also found on activated CD4⁺ T cells. In Study B, T-cell subsets were also evaluated at the 12-week end-point. We first compared T-cell subset proportions between two groups: (i) placebo and (ii) all mice that received antibody in Studies B and C. At the time of the last dose, the mice that received monoclonal anti-CD3 F(ab′)₂ had significantly lower percentages of CD4⁺ T cells (placebo: 60·6% ± 3·3%, treated: 31·6% ± 2·4%, P < 0·001) and CD8⁺ T cells (placebo: 19·2% ± 1·2%; treated: 10·7% ± 0·6%; P < 0·001) in peripheral blood (Fig. 3a). However, there was no significant alteration in the CD4⁺ : CD8⁺ T-cell ratio when comparing the placebo group with the monoclonal anti-CD3 F(ab′)₂-treated group as a whole. By contrast, the percentage of CD4⁺ T cells in peripheral blood that were FoxP3⁺ (i.e. Treg cells) was markedly higher in the monoclonal anti-CD3 F(ab′)₂-treated mice (23·0% ± 1·4%) compared with placebo mice (8·1% ± 1·0%, P < 0·001).

Figure 3.

Evaluation of lymphocyte populations in the peripheral blood of mice treated with CD3 monoclonal antibody (mAb) F(ab′)₂ fragments (1, 2, 5, 25, or 50 μg) in Studies B and C. The proportions of CD4⁺, CD8⁺ and CD4⁺ FoxP3⁺ T cells, measured by flow cytometry, in peripheral blood within 24 hr of the last antibody dose. (a) Mean (± standard error of the mean) proportions of T-cell subsets in the antibody-treatment groups (all dose regimens combined (n= 45–51) versus the placebo group (n= 9). (b) Proportions of T-cell subsets in each group for mice that entered remission versus mice that remained diabetic (n= 2–9 per group).

Given the transient decline in total lymphocyte numbers in the peripheral blood, and the increased percentage of CD4⁺ FoxP3⁺ T-cells at the end of dosing, we hypothesized that CD4⁺ FoxP3⁺ T cells were either selectively maintained or expanded as a result of treatment with monoclonal anti-CD3 F(ab′)₂. At the 12-week end-point, flow cytometric analysis of peripheral blood showed that CD4⁺ and CD8⁺ T-cell populations had significantly recovered but remained below baseline levels, and that the CD4⁺ FoxP3⁺ T-cell population had diminished (from elevated post-dosing levels) to slightly above baseline levels (Table 2). While significant changes in the proportion of various T-cell subsets in peripheral blood were detected during the dosing period, long-term follow-up of peripheral blood PD parameters did not reveal any long-term changes. Potential differences in the T-cell compartments sequestered at the site of inflammation (e.g. the pancreas) were not assessed.

Table 2.

Assessment of T-cell subset populations in peripheral blood of mice treated with monoclonal anti-CD3 F(ab′)₂ fragments (5, 25, or 50 μg) in Study B

Assessment time	Treatment	CD4+ (percentage of lymphocytes)	CD8+ (percentage of lymphocytes)	CD4+ FoxP3+ (percentage of CD4+ cells)
Pretreatment	Monoclonal anti-CD3 F(ab′)2 (n= 6)	47·9 (± 3·1)	14·2 (± 1·6)	6·7 (± 0·4)
Last dose	50 μg/24 hr (n= 6)	15·4 (± 2·0)	9·0 (± 1·4)	40·3 (± 5·8)
	25 μg/72 hr (n= 11)	25·5 (± 2·9)	10·1 (± 1·5)	24·3 (± 3·4)
	5 μg/72 hr (n= 10)	16·8 (± 3·1)	8·9 (± 1·4)	19·3 (± 2·0)
	Placebo (n= 5)	54·1 (± 3·4)	19·8 (± 0·8)	6·8 (± 0·6)
12-week study end-point	50 μg/24 hr (n= 8)	32·9 (± 2·5)	11·8 (± 2·6)	9·0 (± 2·8)
	25 μg/72 hr (n= 6)	28·7 (± 3·1)	14·8 (± 2·1)	10·0 (± 1·7)
	5 μg/72 hr (n= 8)	30·2 (± 2·7)	11·8 (± 4·3)	8·9 (± 1·6)

The proportion of CD4⁺, CD8⁺ and CD4⁺ FoxP3⁺ T cells (mean ± standard error of the mean) was measured by flow cytometry in peripheral blood before treatment within 24 hr of the last dose and at the 12-week study end-point.

The PD parameters observed at completion of dosing were also analyzed according to the monoclonal anti-CD3 F(ab′)₂ dose regimen and whether the mice had entered remission or remained diabetic after treatment. Reductions in the proportions of CD4⁺ and CD8⁺ T cells, and increases in the proportions of CD4⁺ FoxP3⁺ T cells tended to be greater at higher doses (Fig. 3b). Also, at the higher doses, reductions in CD4⁺ T-cell proportions were greater than that observed in CD8⁺ T cells, resulting in a temporary decrease in the CD4⁺ : CD8⁺ T-cell ratio. At the 12-week end-point, the CD4⁺ : CD8⁺ T-cell ratio returned to baseline, as both CD4⁺ and CD8⁺ T-cell populations had significantly recovered (Table 2). At the lower, but still efficacious, doses, a decrease in the CD4⁺ : CD8⁺ T-cell ratio was not observed.

Ultimately, unlike the modulation patterns of the CD3–TCR complex that were elicited by varying doses of monoclonal anti-CD3 F(ab′)₂ (Fig. 1b), a strictly dose-dependent relationship for the alterations in proportions of T-cell subsets was not observed. Furthermore, within each dose regimen, proportions of circulating CD4⁺, CD8⁺ and CD4⁺ FoxP3⁺ T cells at completion of dosing were similar in responder and non-responder mice. However, it is possible that at local sites of inflammation, such as the pancreas and pancreatic lymph nodes, there may be significant differences between responder and non-responder mice in the proportions of these T-cell populations.

Responder mice have greater residual β-cell function at the initiation of treatment

To investigate why some diabetic mice responded to therapy while others did not, even when they had experienced similar changes in PD parameters, the pretreatment level of β-cell function was evaluated by measuring blood glucose and random serum C-peptide levels. As shown in Fig. 4a, pretreatment blood glucose values were significantly lower in mice that entered remission than in those that remained diabetic [mean ± standard error of the mean: remission 383 ± 9·3 mg/dl, diabetic 441 ± 14·2 mg/dl, P < 0·005] (Fig. 4a). This suggests that mice which had a higher level of residual β-cell function at study entry were more likely to respond to treatment. Similarly, the remission group had higher random serum C-peptide levels than the diabetic group, but this difference was not statistically significant (Fig. 4b). These data suggest that efficacy of treatment may be related to baseline β-cell function. At the end of the 12-week follow-up period, C-peptide levels were significantly higher in the remission group than in the diabetic group (Fig. 4b).

Figure 4.

Estimation of the β-cell mass of non-obese diabetic (NOD)/ShiLtJ mice before and after treatment with CD3 monoclonal antibody (mAb) F(ab′)₂ fragments, and histological analyses of pancreata from treated mice that were either in remission or remained diabetic at the end of the study. (a) Comparison of blood glucose measurements before initiation of antibody treatment for treated mice that were either in remission (n= 47) or remained diabetic (n= 32) at the end of the study (mean ± standard error of the mean). (b) Comparison of serum C-peptide levels before (n= 4–5) and 12 weeks after (n= 8–9) antibody treatment in treated mice. (c) Representative photograph of peri-insulitis of islet from a mouse treated with a 5 μg (4×/72 hr) dose that was in remission at the 12-week study end-point. (d) Peri-insulitis scores (PIS) of islets in pancreatic sections from Study B at the 12-week study assessment (diabetic, n= 19; remission, n= 36).

At the 12-week assessment in Study B, histological sections of pancreas were prepared and evaluated for islet content and the presence of leucocytes within the islets. Eighty-one per cent of pancreatic sections from mice that entered remission contained islets (n= 43), whereas 74% of pancreatic sections from treated mice that remained diabetic contained islets (n= 27). In the placebo group, only 71% of pancreatic sections contained islets (n= 14). While these differences were not statistically significant, probably because of the limited number of sections analyzed, the data suggest that the pancreata of non-responders were likely to have fewer preserved islets. Leucocytes present within the islets consisted almost entirely of lymphocytes that were always found at the islet periphery (Fig. 4c), rather than infiltrating throughout the islet, as observed during destructive intra-insulitis. This pattern of peri-insulitis is commonly observed in diabetic mice that have undergone some type of immune therapy.^1,6,21,22 Interestingly, of the mice treated with anti-CD3 F(ab′)₂, those that entered remission had markedly higher scores for peri-insulitis than mice which remained diabetic (Fig. 4d). This suggests that the lymphocytes present in peri-insulitis either are not destructive or are being held at bay by some regulatory mechanism.

Discussion

In this study, dose-ranging experiments were performed in new-onset diabetic NOD mice to determine if low-dose regimens of monoclonal anti-CD3 F(ab′)₂ were efficacious and to examine potential PD effects associated with remission. It had previously been established that a daily dose regimen of 50 μg of monoclonal anti-CD3 F(ab′)₂ for five doses (250 μg total) resulted in high rates of remission.^4,10 We observed that, with this dose regimen, nearly complete modulation of the CD3–TCR complex occurred after the first dose and was sustained throughout the dosing period in peripheral blood. By lowering the dose of monoclonal anti-CD3 F(ab′)₂ and modifying the dose regimen, we were able to achieve a pattern of transient and partial modulation of the CD3–TCR complex during dosing that was as efficacious as the higher doses previously established in the literature. Changes in PD parameters in the peripheral blood of mice treated with monoclonal anti-CD3 F(ab′)₂, such as a transient decrease in lymphocyte counts, a decrease in the percentage of CD4⁺ and CD8⁺ T cells, and a marked increase in the proportion of CD4⁺ FoxP3⁺ T cells, were present at all dose regimens tested. Moreover, these PD effects were similar in responders and non-responders, indicating that the drug was active in all treated mice. Instead, our data suggest that mice which had successfully responded to treatment with monoclonal anti-CD3 F(ab′)₂ had better residual β-cell function at initiation of treatment.

Overall, we provided the first preclinical evidence that lower doses of a monoclonal anti-CD3 F(ab′)₂ are as effective in new-onset diabetic NOD mice as the higher doses previously established in the literature. Furthermore, the PD effects we observed during treatment with low-dose anti-CD3 F(ab′)₂ suggest a non-deletional mechanism of action where activated effector T cells that direct the pathogenic autoimmune response are down-regulated, while local Treg cells that prevent further immune attack are up-regulated in order to achieve long-term clinical stabilization and/or immunologic remission after a short course of therapy.

In a Phase 2 clinical study carried out by the BDR, new-onset type 1 diabetic subjects treated with high doses of otelixizumab had profound and sustained modulation of the CD3–TCR complex throughout the dosing period.¹⁴ Otelixizumab-treated subjects had improved β-cell function compared with placebo for as long as 18 months after dosing¹⁴ and the follow-up data showed a significant decrease in insulin use up to 48 months after dosing.^14,16 Tolerx has explored modifications of the high dose regimen of otelixizumab used in the BDR study to optimize safety and tolerability, specifically investigating regimens that result in lower and less sustained levels of modulation of the CD3–TCR complex. These optimized otelixizumab dose regimens are associated with a transient pattern of modulation of the CD3–TCR complex (Fig. 5) and are very similar to what we describe in this study with the 72 hr dose regimen in mice (Fig. 1b). One of these optimized otelixizumab dose regimens is currently being studied in a Phase 3 pivotal clinical trial (DEFEND). The safety advantages of lower doses of monoclonal anti-CD3 are numerous, including greatly reduced cytokine release, sustained Epstein–Barr virus (EBV) immunosurveillance and the lack of immunogenicity, which would allow for repeat dosing, if required. Interestingly, preliminary clinical studies with teplizumab, another Fc-modified monoclonal anti-CD3, suggest that higher doses do not improve efficacy and are associated with an increase in adverse events.²³

Figure 5.

CD3–T-cell receptor (TCR) complex modulation on circulating T cells during treatment with monoclonal anti-CD3 in a clinical study. Subjects (n= 16) were dosed with an 8-day regimen of otelixizumab. TCR expression on circulating CD4⁺ T cells was assessed using flow cytometry pre-infusion, at end of infusion (EOI), and 2 hr after EOI. Mean (± standard deviation) TCR-αβ expression levels are presented as molecules of equivalent soluble fluorochrome (MESF) units.

In this study we demonstrated that dose regimens of monoclonal anti-CD3 F(ab′)₂, featuring low doses 3 days apart, elicited patterns of transient and partial modulation of the CD3–TCR complex and resulted in remission rates comparable to the higher doses previously established in the NOD mouse model. Furthermore, even at low doses, remission was durable. A total dose of 8 μg resulted in 53% long-term remission for up to 24 weeks after treatment. This is comparable to the 56% remission in the 250 μg total dose regimen, despite the difference of > 30-fold in dose. It has been reported that single high doses [one dose of 18–50 μg of anti-CD3 mAb F(ab′)₂] produce similarly high remission rates; however, the mice that responded favourably to such treatment were within a very limited glycaemia range (300–349 mg/dl) at the start of treatment, making a direct comparison with our data difficult.²⁴

Various PD parameters were evaluated in mice that received monoclonal anti-CD3 F(ab′)₂. Modulation of the CD3–TCR complex on peripheral T cells was dose-dependent. Interestingly, as little as 30% modulation of the CD3–TCR complex, elicited by the 2 μg (4×/72 hr) dose regimen, was sufficient to induce high rates of durable remission in new-onset diabetic NOD mice. The difference in the level of modulation of the CD3–TCR complex between the 2 μg (4×/72 hr) dose regimen and the less effective dose regimen of 1 μg (4×/72 hr) was not large –∼30% versus 20%– but it was statistically significant. We estimate that the 2 μg (4×/72 hr) dose regimen results in having antibody occupy as little as one-fifth of the total number of CD3 molecules in the mouse. Overall, this work demonstrated that in the NOD mouse model: (i) sustained modulation of the CD3–TCR complex during the dosing period was not required for efficacy and remission can occur at lower doses that produce only transient modulation of the CD3–TCR complex, and (ii) partial modulation of the CD3–TCR complex on circulating lymphocytes was sufficient to induce remission.

By the end of dosing, there were transient decreases in lymphocyte counts in the peripheral blood, similar to that observed in clinical studies with otelixizumab, but they were not strictly dose dependent.¹⁴ Also, at the end of dosing, there were reductions in the percentages of CD4⁺ and CD8⁺ T cells, and a marked increase in the proportion of CD4⁺ FoxP3⁺ T cells in the peripheral blood. Similar changes have been observed in new-onset type 1 diabetic subjects administered otelixizumab.¹⁴ In NOD mice, the altered proportions of T-cell subsets were not strictly dose dependent, although they tended to be more marked at higher doses. Given that similar PD effects occurred in both mice that entered remission and in those that remained diabetic, these PD parameters alone could not be used to predict response to monoclonal anti-CD3 F(ab′)₂ treatment in NOD mice. It is likely that an optimal amount of PD activity (including modulation of the CD3–TCR complex, transient loss of circulating lymphocytes, and/or alterations in T-cell subsets) is necessary, but not sufficient, for efficacy, and that efficacy will also be dependent on the level of β-cell mass and/or function before treatment.

The hypothesis that efficacy of treatment with monoclonal anti-CD3 is correlated with residual β-cell status is supported by the observation that mice with better residual β-cell function, as measured by blood glucose and serum C-peptide levels, were more likely to respond to treatment. It is also supported by earlier studies in which NOD mice that remained diabetic after treatment with monoclonal anti-CD3 F(ab′)₂ were restored to full metabolic control with syngeneic islet transplantation.¹ These observations are consistent with findings in the Phase 2 BDR study, where increases in endogenous insulin production were most pronounced in otelixizumab-treated subjects with initial residual β-cell function at or above the 50th percentile.¹⁴

Overall, our results demonstrate that low, subimmunogenic doses of monoclonal anti-CD3 F(ab′)₂, which result in transient and partial modulation of the CD3–TCR complex, are sufficient to induce high rates of remission in new-onset diabetic NOD mice. While the autoimmune component of type 1 diabetes may be sufficiently resolved following therapy with monoclonal anti-CD3, glycaemic control and functional remission of disease probably depend upon the level of residual β-cell function at the time of treatment. Successfully translating therapy with monoclonal anti-CD3 mAb into a clinical situation may therefore depend not only upon identifying dosing strategies that minimize adverse effects while maximizing efficacy, but also upon identifying the window of treatment during which patients are most likely to respond favorably to treatment.

Glossary

Abbreviations:

BDR

Belgian Diabetes Registry

ELISA

enzyme-linked immunosorbent assay

FcR

Fc receptor

mAb

monoclonal antibody

MESF

molecules of equivalent soluble fluorochrome

NOD

non-obese diabetic

PBS

phosphate-buffered saline

PD

pharmacodynamic

TCR

T-cell receptor

Treg

T regulatory

Disclosures

Devangi S. Mehta, Rudy A. Christmas and Michael Rosenzweig are employees of Tolerx, Inc. Herman Waldmann is a co-founder of Tolerx, Inc. and is a member of the Board of Directors.