Characterization of a surrogate murine antibody to model anti-human CD3 therapies

Abstract

Fc-modified anti-human CD3ε monoclonal antibodies (mAbs) are in clinical development for the treatment of autoimmune diseases. These next generation mAbs have completed clinical trials in patients with type-1 diabetes and inflammatory bowel disease demonstrating a narrow therapeutic window. Lowered doses are ineffective, yet higher pharmacologically-active doses cause an undesirable level of adverse events. Thus, there is a critical need for a return to bench research to explore ways of improving clinical outcomes. Indeed, we recently reported that a short course of treatment affords synergy, providing long-term disease amelioration when combining anti-mouse CD3 and anti-mouse tumor necrosis factor mAbs in experimental arthritis. Such strategies may widen the window between risk and benefit; however, to more accurately assess experimentally the biology and pharmacology, reagents that mimic the current development candidates were required. Consequently, we engineered an Fc-modified anti-mouse CD3ε mAb, 2C11-Novi. Here, we report the functional characterization of 2C11-Novi demonstrating that it does not bind FcγR in vitro and elicits little cytokine release in vivo, while maintaining classical pharmacodynamic effects (CD3-TCR downregulation and T cell killing). Furthermore, we observed that oral administration of 2C11-Novi ameliorated progression of remitting-relapsing experimental autoimmune encephalitis in mice, significantly reducing the primary acute and subsequent relapse phase of the disease. With innovative approaches validated in two experimental models of human disease, 2C11-Novi represents a meaningful tool to conduct further mechanistic studies aiming at exploiting the immunoregulatory properties of Fc-modified anti-CD3 therapies via combination therapy using parenteral or oral routes of administration.

Introduction

T cells are responsible for initiating and sustaining inflammation in a wide range of diseases. As a result, strategies aimed at antagonizing pathogenic T cells have undergone intense investigation. The first therapeutic monoclonal antibody (mAb) to be approved was muromonab (Orthoclone OKT3), a murine anti-human CD3ε mAb. Muromonab potently reverses and prevents acute allograft rejection,^1,2 but it induces substantial levels of human anti-mouse antibodies³ and a severe debilitating cytokine storm.⁴ Several lines of evidence indicate that the cytokine release induced by muromonab is largely dependent on its capacity to interact with FcγR expressed on accessory cells such as monocytes/macrophages, allowing multivalent cross-linking of CD3-TCR complexes and T cell activation. In vitro, soluble muromonab does not activate T cells in the absence of FcγR-bearing cells⁵ or presence of FcγR-negative accessory cells,^6,7 and F(ab’)2 fragments of muromonab induce negligible T cell activation.⁸ In addition, in vitro and in vivo experiments conducted with various anti-CD3ε mAbs (including muromonab in humanized mice) have demonstrated that (1) variants with reduced or absent FcγR binding activity induce inferior T cell activation;^9-18 (2) fucosylation variants with increased FcγR binding activity induce superior T cell activation;¹⁹ and (3) minimal T cell activation is observed in the presence of a blocking anti-FcγR antibody²⁰ or in FcγR-deficient animals.¹⁹

Due to its efficacy, and although it provokes severe side effects, muromonab was used extensively in the field of transplantation. In addition to allograft rejection, T cells play a key role in the pathogenesis of many autoimmune diseases, including multiple sclerosis, type-1 diabetes (T1D) and inflammatory bowel disease. Hence, in the early 1990s muromonab was also administered to patients with severe multiple sclerosis,²¹ but, due to its toxicity, development for the treatment of autoimmune diseases with this anti-CD3 mAb was terminated.

To alter the risk-to-benefit ratio and thus allow assessment of CD3-directed therapies for the treatment of autoimmune diseases, humanized versions of rodent anti-human CD3ε mAbs^22-27 and a fully human mAb²⁸ were created. These mAbs have a similar engineered element built into their Fc portion, i.e., mutations introduced into the second heavy constant domain to reduce FcγR binding and cytokine release associated with an Fc-dependent mechanism. Importantly, in vitro studies using human blood cells and clinical trials in solid-organ transplant recipients showed that Fc-engineered anti-human CD3ε mAbs have immunosuppressive properties similar to that of muromonab, while they do not have its severe unwanted immune activating capacity.^{11,23-25,29-35}

Consequently, these next-generation CD3-directed therapies have been tested as mono- or add-on therapies in various autoimmune diseases. Remarkable results were obtained with teplizumab [hOKT3γ1(Ala-Ala)]^36-38 and otelixizumab (ChAglyCD3)^39,40 in Phase 1/2 clinical trials in patients with new onset type 1 diabetes mellitus (T1D). When explored at lower doses in the larger Phase 3 trials, however, the same beneficial effect was not observed globally as measured against the selected endpoints.^41,42 A similar situation emerged with visilizumab when the encouraging results from its Phase 1/2 clinical trials in steroid-resistant ulcerative colitis patients were not confirmed in the follow-up randomized placebo-controlled Phase 2/3 trial.^43,44 Once again, due to adverse events observed in the Phase 1/2 pilot trials, the dose was reduced in the subsequent Phase 2/3 trials. These results suggest that anti-CD3-based therapies have a narrow therapeutic window in which low doses are ineffective and higher pharmacologically-active doses cause intolerable levels of adverse effects.⁴⁵

Alternative strategies must be explored to better understand how to exploit the potential of anti-CD3-directed therapies. For example, the combination of anti-CD3 mAbs with other drugs, such as has been suggested for the treatment of T1D, should be examined.⁴⁶ Combinations may be the most effective way to reduce the toxicity of anti-CD3 mAbs while allowing significant therapeutic benefit to occur.⁴⁷ Another strategy proposed to widen the therapeutic window of anti-CD3 therapies consists of administering the drug orally. As shown in preclinical and clinical studies, oral anti-CD3 mAbs demonstrate potent immunomodulating properties without evoking side effects related to cytokine release or humoral immune responses.^48,49

The species specificity of CD3-directed therapies has been an obstacle to study their pharmacology in animal models of human diseases. To serve as a surrogate molecule for the anti-human CD3 mAbs currently in clinical development, we engineered an Fc-modified anti-mouse CD3 antibody, 2C11-Novi.⁵⁰ Here, we report the functional characterization of 2C11-Novi, which reproduces many of the pharmacodynamic effects of anti-CD3 therapies. In addition, we show for the first time that an intact non-FcγR binding anti-CD3 mAb is active via the oral route of administration. Taken together, these data suggest that 2C11-Novi is a relevant surrogate reagent to study Fc-modified anti-human CD3 mAb therapies and will be useful to conduct further mechanistic studies in murine models of human diseases via parenteral and oral routes of administration.

Results

Engineering and functional characterization of 2C11-Novi in vitro

The anti-mouse CD3ε, 2C11-Novi, was obtained by combining the Fv sequences of 145–2C11⁵¹ with the constant domains of the murine IgG1k, thereby producing a chimeric hamster/mouse mAb. The sequences of the Fv portion of 145–2C11 were chosen because this mAb is directed against the epsilon chain of the mouse CD3-TCR complex and has been shown to reproduce the immunosuppressive properties of both muromonab and the new generation of anti-human CD3ε mAbs.⁵² 145–2C11 is, however, known to bind mouse FcγR²⁰ and trigger cytokine release in mice,⁵³ as was similarly observed in patients treated with muromonab.⁴ Therefore, to generate an anti-mouse CD3 mAb that can be used as a surrogate for the Fc-modified anti-human CD3 therapies, the murine IgG1 was selected as the Fcγ backbone with an alanine substitution at amino acid 265 introduced into the CH2 region. The single amino acid mutation in the CH2 domain drastically reduces binding of IgG1 mAbs to FcγR.^54,55 This mAb was named 2C11-Novi.

To characterize 2C11-Novi, its biological properties were first evaluated in vitro and compared with 145–2C11. As assessed by measuring TCR expression on mouse T cells, 2C11-Novi and 145–2C11 have the same functional potency to induce the downregulation of CD3-TCR complexes (Fig. 1A). The binding capacity of the anti-CD3 mAbs to FcγR was evaluated by flow cytometry in a competition assay using the mouse macrophage cell line, RAW 264.7, which constitutively expresses CD16 (FcγRIII), CD32 (FcγRIIb) and CD64 (FcγRI). As shown in Figure 1B, the 145–2C11 hamster IgG effectively competes for binding to RAW cells with murine IgG2a, which have strong affinity for murine FcγR. 2C11-Novi, which was engineered to eliminate the binding capacity to FcγR, did not bind to RAW cells, even at concentrations largely exceeding that of the murine IgG2a antibody (15-fold excess). Next, the mitogenicity of 2C11 was compared with that of 145–2C11 using splenocytes. Cellular proliferation was measured by [3H]thymidine incorporation into DNA. T cells proliferated in response to soluble 145–2C11 in a dose-related manner but not to soluble 2C11-Novi (Fig. 1C); however, when 2C11-Novi was immobilized on cell culture plates, T cells were induced to proliferate as potently as 145–2C11 (data not shown).

Figure 1. in vitro characterization of 2C11-Novi. (A) CD3-TCR complex downregulation from the T cell surface. Mouse T cells were incubated at 37°C for 18 h with the indicated anti-CD3 mAbs prior to analysis of mouse TCR expression levels by flow cytometry. (B) Binding competition of anti-CD3 mAbs and low affinity IgG1 isotype with APC-conjugated high affinity IgG2a mAb to mouse RAW 264.7 cells. Cells were incubated with APC-IgG2a and the indicated concentrations of 145–2C11, 2C11-Novi or murine IgG1 for 30 min on ice prior analysis by flow cytometry. (C) Proliferation of mouse splenocytes in response to varying concentrations of soluble 2C11-Novi compared with 145–2C11. Cells were incubated at 37°C for three days with increasing concentrations of 2C11-Novi or 145–2C11. For the last 18 h, 3H-thymidine was added to quantify cell proliferation. The results are expressed as the mean cpm of triplicate cultures ± SEM.

These data indicate that through the Fv portion, 2C11-Novi retains the capacity to bind CD3ɛ on T cells with the same functional activity as 145–2C11 while the mutation introduced in the Fc domain of the mAb successfully abrogates binding to FcγR.

Pharmacokinetics of 2C11-Novi

The pharmacokinetics (PK) of 145–2C11 have been reported.⁹ Interestingly, chimeric variants of 145–2C11 with low affinity for FcγR have been shown to have improved PK profiles.⁹ To study the PK of 2C11-Novi, Balb/c mice were injected intravenously with a single dose of the antibody. For comparison, a single 20 μg dose of 145–2C11 was used. This dose was selected based on previously published efficacy results⁵⁶ and on the safety profile of 145–2C11, which induces signs of severe toxicity at higher doses. 2C11-Novi was administered at the same dose as 145–2C11 and the higher dose of 50 μg, which was the dose used in a previously published study involving experimental murine arthritis.⁵⁰

Single intravenous dose administration of 145–2C11 and 2C11-Novi showed good systemic exposure and distribution. The PK profiles exhibited a rapid increase after the injection, followed by a decrease in concentration (Fig. 2). The maximal concentration (Cmax) was measured 1 h post treatment (Table 1). 2C11-Novi was quantified in plasma up to day 7 and day 11 post-injection of 20 µg and 50 µg per mouse, respectively. 145–2C11 was cleared from plasma more rapidly than 2C11-Novi, i.e., at 3 d post-injection; its plasma concentration was below the lower limit of quantification of the assay. In addition, the plasma PK for single doses of 20 and 50 μg of 2C11-Novi were not dose proportionate based on AUC. This observation suggests that the clearance of 2C11-Novi was mediated by binding to the drug target, CD3ε, on T cells (Table 1). The terminal half-life of 2C11-Novi is approximately 1 d. In contrast, 145–2C11 exposure and half-life could not be determined because it was cleared too rapidly.

Figure 2. PK profiles for 2C11-Novi and 145–2C11. Mice were treated with a single dose of 145–2C11 or 2C11-Novi administered intravenously. Blood samples were collected at the indicated time points. Concentrations of biologically active anti-CD3 mAbs were determined by flow cytometry using a competition assay which measured binding inhibition of APC-conjugated 145–2C11 to T cells. Data points and bars are representative of triplicate measures ± SEM. Data are representative of two experiments with similar results.

Table 1. PK parameters for 145–2C11 and 2C11-Novi (Geomean and CV%).

	145–2C11	2C11-Novi	2C11-Novi
Dose (μg/mouse)	20	20	50
Cmax (μg/mL)	8.3 (31.7%)	8.8 (3.7%)	22.1 (4.5%)
AUC0-t last (μg.day/mL)	NA	15.7 (8.8%)	78.9 (6%)
T½ (days)	NA	1.03 (4.35%)	1.03 (13.3%)

Functional characterization of 2C11-Novi in vivo

To further characterize 2C11-Novi in vivo, various PD effects were assessed and compared with those of 145–2C11. T cell activation, cytokine levels and body weight were measured at various time points after a single injection of anti-CD3 mAb (Fig. 3). 145–2C11 induced a significant increase in the proportion of CD69⁺ (mean ± SEM at the peak, 91.2 ± 2.3% 4 h post-injection, Figure 3A) and CD25⁺ (mean ± SEM at the peak, 79.9 ± 2.8% 18 h post-injection, Figure 3B) CD4⁺ T cells in the spleen. In addition, 145–2C11 induced the release of tumor necrosis factor (TNF; mean ± SEM at the peak, 300 ± 26 pg/mL 2 h post-injection, Figure 3C), interferon-γ (IFNγ; 4642 ± 227 pg/mL 6 h post-injection, Figure 3D) and interleukin-6 (IL-6; 14240 ± 1419 pg/mL 4 h post-injection, Figure 3E) as measured in plasma. In contrast, the ability of 2C11-Novi to cause activation of T cells and induction of cytokine release was significantly reduced (Figs. 3A-E). Furthermore, body weight loss, a consequence of the cytokine release, did not occur in mice treated with 2C11-Novi (Fig. 3F), while 145–2C11 treated mice had a significant reduction in body weight at 72 and 96 h of 10.9% and 12.1%, respectively, compared with baseline (p < 0.0001).

Figure 3. T cell activation, cytokine release and body weight loss induced by 2C11-Novi and 145–2C11 in vivo. Mice were treated with a single dose of 50 µg 2C11-Novi (open squares) or 20 µg 145–2C11 (open triangles) administered intraperitoneally. (A) Frequency of CD4⁺ splenocytes that express CD69 and (B) CD25 as assessed by flow cytometry. (C) Plasma levels of TNF, (D) IFNγ and (E) IL-6 measured by multiplex analysis. (F) Body weight of the mice. Spleen and plasma samples were collected at the indicated time points. Data represent mean of five to six mice per group per time point ± SEM. Statistically significant differences were determined by unpaired Student’s two-tailed t test with *** p < 0.001. Data are representative of two experiments with similar results.

In the same study, CD3-TCR downregulation was assessed by measuring the level of surface TCR expression on CD4⁺ T cells at different time points. 2C11-Novi and 145–2C11 induced a rapid and almost complete reduction of TCR expression at the surface of CD4⁺ T cells obtained from the blood (Fig. 4A), lymph nodes (Fig. 4B) or spleen (data not shown). In contrast, the kinetics of recovery of the CD3-TCR complex was strikingly different. For 145–2C11-treated mice, TCR expression levels returned to baseline levels within 5 d post-treatment, whereas those of 2C11-Novi-treated mice remained low even after 5 d (Figs. 4A-B). Furthermore, lymph node cells isolated from mice treated with 2C11-Novi displayed a statistically significant decrease in proliferative response to the polyclonal T cell mitogen concanavalin A (ConA) compared with 145–2C11 (Fig. 4C).

Figure 4. In vivo downregulation of CD3-TCR complex,T cell unresponsiveness and lymphopenia induced by 2C11-Novi and 145–2C11. Mice were treated as described in Figure 3. TCR expression levels on (A) blood and (B) lymph node (LN) CD4⁺ T cells as assessed by flow cytometry. (C) Proliferative response of lymph node cells to Concanavalin A (Con A) assessed 24 h and 120 h after anti-CD3 mAb treatment. (D) Circulating CD4⁺ T cell count as assessed by flow cytometry using Trucount tubes. Tissue samples were collected at the indicated time points. Data represent mean of five to six mice per group per time point ± SEM. Statistically significant differences were determined by unpaired Student’s two-tailed t test with ** p < 0.01 and *** p < 0.001. Data are representative of two experiments with similar results.

In addition to CD3-TCR downregulation, anti-CD3 mAbs exert transient immunosuppressive effects via induction of lymphopenia. Therefore, T cell counts were assessed in blood after a single dose of 2C11-Novi or 145–2C11. 2C11-Novi induced a rapid and profound (96.5% average reduction of circulating CD4⁺ cells compared with baseline at 6 h post-injection), yet transient (16.1% at 5 d post-injection), lymphopenia (Fig. 4D). In comparison, 145–2C11 induced lymphopenia with similar intensity but for a more prolonged period of time (97.9% at 6 h post-injection and 51.7% at 5 d post-injection, Figure 4D).

One mechanism of action proposed to explain how anti-CD3 mAbs induce long-term immune tolerance involves the induction of T cell apoptosis and uptake by transforming growth factor (TGF)-β secreting phagocytes.⁵⁷ The capacity of 2C11-Novi and 145–2C11 to induce T cell apoptosis and TFG-β increase was evaluated. As shown in Figure 5A, 2C11-Novi induced significant cell death in lymph nodes compared with the control group. A more profound effect was observed with 145–2C11. In parallel, a less pronounced depletion of T cells was observed in animals treated with 2C11-Novi compared with 145–2C11 in the lymph nodes (Fig. 5B) and spleen (data not shown). As shown in Figure 6, 2C11-Novi induced a significant increase of TGF-β levels in plasma compared with the control group. A more pronounced increase was induced by 145–2C11.

Figure 5. In vivo T cell depletion induced by 2C11-Novi and 145–2C11. Mice were treated as described in Figure 3. (A) Cell death quantified using the Apo-TRACE™ in vivo apoptosis detection kit using cell collected from lymph nodes 24 h after anti-CD3 mAb treatment. Control mice received a single 50 µg dose of mouse IgG1κ isotype control. Symbols represent individual mice and horizontal bars represent averages. Statistical differences between groups were determined by one-way ANOVA with Bonferroni post-test with ** p < 0.01 and *** p < 0.001. (B) CD4⁺ T cell count in the lymph nodes (LN) as assessed by flow cytometry. Tissue samples were collected at the indicated time points. Data represent mean of five to six mice per group per time point ± SEM. Statistically significant differences were determined by unpaired Student’s two-tailed t test with * p < 0.05 and ** p < 0.01. Data are representative of two experiments with similar results.

Figure 6. TGF-β secretion induced by 2C11-Novi and 145–2C11. Mice were treated as described in Figure 3. Control mice received a single 50 µg dose of mouse IgG1κ isotype control. TGF-β was measured in plasma 5 d post-treatment by ELISA. Data represent mean of nine mice per group ± SEM. Statistical differences between groups were determined by one-way ANOVA with Bonferroni post-test with * p < 0.05 and ** p < 0.01. Data are representative of two experiments with similar results.

Oral immunotherapy with 2C11-Novi suppresses EAE

Promising yet controversial results have demonstrated that oral administration of anti-CD3 mAbs (145–2C11 and F(ab’)2 variant) ameliorates murine models of multiple sclerosis.⁵⁸ Therefore, we sought to confirm these findings by evaluating 2C11-Novi in remitting-relapsing EAE induced in SJL mice by immunization with PLP_139–151 peptide. 2C11-Novi or 145–2C11 administered orally (five daily injections of 5 µg mAb starting one week prior to immunization with PLP) efficiently ameliorated the progression of EAE, significantly reducing the primary acute and subsequent relapse phases of the disease (average ± SEM AUC_10–35 of 30.0 ± 7.0 and 35.5 ± 5.0 vs. 59.7 ± 6.2 respectively, p < 0.001, Figure 7).

Figure 7. Therapeutic effects of 2C11-Novi in remitting-relapsing EAE via the oral route of administration. Female SJL mice were immunized subcutaneously with 100 µg PLP_139–151 emulsified in CFA enriched with 800 μg Mycobacterium tuberculosis. Mice were also injected intraperitoneally with 400 ng of pertussis toxin on the day of immunization (day 0) to induce remitting-relapsing EAE. Mice were administered orally two days prior to PLP immunization, PBS or 5 µg of either 2C11-Novi, 145–2C11 or isotype control daily for 5 consecutive (days -6 to -2). (A) Mean score of 7 animals per dose group ± SEM from one experiment. (B) Symbols in the scatter plot represent individual mice, horizontal bars represent averages and error bars represent SEM. The statistical significance of the difference between groups was calculated from AUC for each individual mouse. Statistically significant differences were determined by t test (A) or one-way ANOVA with Bonferroni post-test (B) with * p < 0.05, ** p < 0.01 and *** denoting p < 0.001. Data are representative of two experiments with similar results.

Discussion

Recently reported results from Phase 3 studies that investigated the safety and therapeutic efficacy of modified anti-CD3 antibodies in autoimmunity indicate a narrow therapeutic window for this drug class. Therefore, further preclinical investigation is required to identify ways of widening the therapeutic windows of Fc-modified CD3-directed therapies. The challenge, however, is to have a relevant ‘surrogate’ reagent because therapeutic anti-human CD3 mAbs do not cross-react with T cells from standard laboratory species. Thus, meaningful in vivo non-clinical safety and efficacy studies with these CD3-directed therapeutic mAbs cannot be conducted in rodents or macaques. As a possible solution, the development of transgenic animals bearing the human target for the preclinical assessment of therapeutic mAbs has been considered. Indeed, de la Hera and colleagues generated mice that carry the human CD3ε chain as a transgene under the control of the CD2 promoter.⁵⁹ The caveat with these transgenic mice is that the murine CD3ε chain also must be expressed to achieve normal T cell development. Therefore, every T cell of the human CD3ε transgenic mouse expresses a 1:1 ratio of human and murine CD3ε.⁵⁹ Otelixizumab is pharmacologically active in these transgenic mice, suggesting that the human CD3ε chain can associate with the other chains of the mouse CD3-TCR complex to form a functional hybrid CD3-TCR signaling molecule.⁶⁰ In addition, these transgenic mice have been bred onto the non-obese diabetic background such that they spontaneously develop autoimmune insulin-dependent diabetes.⁶¹ Treatment of the mice with otelixizumab induced a durable disease remission dependent on transferable T cell-mediated tolerance and TGF-β.⁶¹ Therefore, the animals represent a valuable tool to conduct further preclinical studies to support new development strategies for T1D.

To investigate the utility of anti-CD3 for new treatment options in other therapeutic areas, such as inflammatory bowel disease, rheumatoid arthritis or multiple sclerosis, the use of surrogate reagents remains essential. The mAb of choice for decades has been 145–2C11, a hamster anti-mouse CD3ɛ mAb. This mAb is a good surrogate for muromonab because it reproduces its immunosuppressive properties in vivo.⁵² In addition, like muromonab, 145–2C11 binds FcγR and triggers the cytokine storm when administered parenterally.⁵³ To reduce its toxicity, F(ab’)2 fragments of 145–2C11 have been generated and used extensively,⁶² but such fragments may not be appropriate surrogate molecules for the new generations of non-FcγR binding anti-human CD3ε mAbs because of their very short half-life in vivo.

To address these issues, we generated a non-FcγR binding anti-mouse CD3ε IgG. The characteristics that we sought to reproduce in the variable regions of this mAb were the specificity for the CD3ε chain and the ability to efficiently induce CD3-TCR complex downregulation. Thus, the variable regions of the original 145–2C11 mAb were selected because they bind to the relevant chain of the CD3-TCR complex and induce its downregulation both in vitro and in vivo. Furthermore, various amino acid substitutions have been introduced into the Fc portion of the next generation of anti-CD3 mAbs to reduce or avoid FcγR-mediated toxicity. We therefore engineered the variable regions of 145–2C11 onto a mouse IgG1 Fc backbone that similarly lacks FcγR binding through mutagenesis of one amino acid in the Fc portion. In this study, we aimed at functionally demonstrating that the resulting hamster/mouse chimeric mAb, 2C11-Novi, is a better surrogate mAb for the next-generation anti-CD3 therapies, where the Fc contribution has been minimized through mutagenesis, than the parental hamster 145–2C11.

First, in vitro, we have shown that 2C11-Novi reproduces the Fc properties of Fc-mutated anti-CD3 therapies, i.e., it does not bind to FcγR expressed on macrophage or monocytic cell lines and consequently does not induce T cell proliferation (mitogenicity) in splenocytes (composed of T cells and FcγR-bearing cells) when used as a soluble mAb. In vivo, we then studied the PK of 2C11-Novi compared with that of 145–2C11. Anti-CD3 mAbs with low affinity for FcγR have been shown to have improved (i.e., longer) PK profiles.⁹ In addition, the plasma concentration of anti-CD3 mAbs has been shown to correlate with CD3-TCR levels on circulating T cells.⁶³ Consistently, the inability of 2C11-Novi to bind FcγR resulted in augmented PK and prolonged CD3-TCR downregulation. In contrast, when injected at the same dose as 2C11-Novi (data not shown), 145–2C11 was cleared faster and induced less durable CD3-TCR downregulation. Thus, the differences observed in terms of activation and cytokine release between 145–2C11 and 2C11-Novi were not due to a PK profile in which the latter was cleared faster.

We demonstrated that 2C11-Novi induces T cell activation with less intensity compared with 145–2C11 with significantly reduced cytokine release and no body weight loss. In addition, 2C11-Novi induces the classical immunosuppressive effects of anti-CD3 mAbs; transient CD3-TCR downregulation, which renders the cell unresponsive to antigenic stimulation, and lymphopenia. This is consistent with the observation that Fc-FcγR interaction is not required for these immunosuppressive effects while it contributes significantly in T cell activation and subsequent cytokine release, the cause of the unwanted side effects.^8,9,12,20

In addition to immunosuppressive effects, anti-CD3 mAbs have been shown to restore immunological tolerance, probably via activation-induced cell death (AICD) of T cells and the release of TGF-β from day 1 to day 10 post treatment.⁵⁷ TGF-β would be secreted by phagocytes digesting apoptotic T cells.⁵⁷ In turn, TGF-β would influence other cell types including the remaining T cells to induce, locally, tolerance. Indeed, post anti-CD3 administration in type 1 diabetic mice, regulatory T cells are observed to surround the pancreatic islets in mice responding to treatment.⁶⁴ Furthermore, disease remission occurs in a TGF-β dependent fashion.⁶⁵

We assessed the capacity of 2C11-Novi to induce AICD of T cells and increase systemic levels of TGF-β. 2C11-Novi was found to induce significant T cell apoptosis in secondary lymphoid tissues and to induce a significant increase of systemic levels of TGF-β. Despite its shorter PK, the capacity of 145–2C11 to bind FcγR is associated with a very potent ability to cross-link the CD3-TCR and activate T cells. Thus, compared with 2C11-Novi, 145–2C11 induces higher T cell depletion by AICD. Consistently with the model described above where phagocytes digesting apoptotic T cells are the source of the key tolerogenic cytokine, TGF-β, 145–2C11 induced higher TGF-β secretion as compared with 2C11-Novi. These results demonstrate that 2C11-Novi has immunoregulatory features.

Taken together, these data suggest that 2C11-Novi is suitable for use in assessments of the biological features of the therapeutic Fc-modified anti-human CD3 mAbs when investigating laboratory strategies using murine models of human disease. It is important to note that surrogate anti-mouse CD3 mAbs such as 2C11-Novi cannot predict the maximum tolerated dose of Fc-modified CD3-directed therapies in patients. For instance, a single dose of 50 μg (approximately 2.25 mg/kg) of 2C11-Novi activates T cells in vivo, but it does not induce the systemic release of cytokines causing side effects such as body weight reduction. In contrast, Fc-modified anti-human CD3 mAb therapies commence cytokine release and the associated syndrome in patients at doses between 0.01 and 0.1 mg/kg.^{28,36,39,43,66-68}

A hypothesis that may explain the susceptibility to invoking cytokine release involves the ratio of effector memory vs. regulatory T cells in mice vs. humans. Because laboratory mice are housed in clean and pathogen-controlled environments, their T cells have been exposed to a much narrower antigen spectrum, resulting in fewer effector memory T cells. As such, the pool of T cells to release cytokines is also diminished. Indeed, primed effector and memory T cells have an increased sensitivity to antigen stimulation compared with naive T cells,⁶⁹ and a decreased dependency on costimulation.⁷⁰ Also, as effector T cells and natural regulatory T cells compete for consumption of interleukin-2 (IL-2), it has been proposed that the depletion of IL-2 by regulatory T cells would be sufficient to suppress effector T cell responses.⁷¹ Therefore, in mice, given the scarcity of effector memory T cells, natural regulatory T cells may quickly act as IL-2 ‘sinks’ and would be able to rapidly inhibit the response induced by TCR activation. In contrast, the much larger proportion of effector memory T cells present in patients with a chronic disease, and thus a highly ‘antigen experienced’ immune system, may overcome this immunosuppressive safeguard.⁷¹

Despite some limitations, surrogate mAbs have been pivotal for deciphering the complex mode of action of anti-CD3 mAb therapies. For example, we have recently shown, using 2C11-Novi, that a short course of treatment with an anti-CD3 mAb combined with a single low dose of an anti-mouse TNF mAb efficiently depletes pathogenic T cells and is associated with long-term inhibition of established collagen-induced arthritis.⁵⁰ Furthermore, although perhaps counterintuitive because mAbs are large macromolecular proteins, oral delivery of anti-CD3 mAbs has been found to exert potent immunoregulatory functions, and in the absence of systemic exposure and cytokine release.^48,49 145–2C11 and its F(ab’)2 fragment administered orally have been shown to suppress PLP-induced EAE.⁵⁸ Using the same published treatment regimen, we wanted to confirm that intact non-FcγR binding anti-CD3ε mAbs are biologically active when administered orally. We were able to reproduce this effect with 2C11-Novi. Progression of EAE was ameliorated, with a significant reduction in both the primary acute and subsequent relapse phase of the disease. Our data confirm that anti-CD3 mAbs are pharmacologically active via the oral route of administration and indicate that 2C11-Novi can serve as a surrogate reagent for Fc-modified anti-CD3 mAb therapies to conduct mechanistic studies in murine models of human diseases via the oral route of administration.

2C11-Novi thus represents a valuable tool to conduct meaningful preclinical studies to explore further the complex mechanisms of actions of anti-CD3 mAbs alone or in combinations with other therapeutic agents with the aim of widening their therapeutic window. Furthermore, studies in additional models of human disease can be conducted via classical parenteral administration, or use of the oral route can be explored. The benefit of the latter is that there is no systemic drug exposure, and thus, no generalized immunosuppression or side effects associated with cytokine release.⁴⁸ It is therefore ideal for chronic treatment, which may be required for the treatment of autoimmune diseases in the clinic.

Materials and Methods

Animals

Studies were conducted in eight-week-old male Balb/c mice or female SJL mice (Janvier Laboratories). Animal experiments were conducted after obtaining the permission of the Swiss veterinary office for animal experimentation (authorization OVC-1015/3562/3).

Generation of 2C11-Novi

The hybridoma clone 145–2C11, which produces a hamster anti-mouse CD3ε mAb, was obtained from the American Type Culture Collection (ATCC). The sequences encoding the heavy and light variable chain regions of 145–2C11⁹ were cloned using conventional molecular techniques into two distinct plasmid vectors that encoded mouse kappa light chain and IgG1 heavy chain constant regions, respectively. The IgG1 heavy chain constant region was mutated at EU position 265 from aspartic acid to alanine by site-directed mutagenesis. The two cloning vectors were then fused into a single double gene vector. Chinese hamster ovary (CHO) cells (1.10⁷) were transfected by electroporation with 40 µg of linearized double gene vector and cultured in chemically-defined medium (CD-CHO, Invitrogen). Selective pressure was applied 24 h later and once the culture reached a density of 2.10⁶ viable cells per mL, cells were adapted to grow in suspension. Batch fermentation (10 L) was operated in a 20 L cellbag (GE Healthcare Life Sciences) using CD-CHO under selective pressure. After 10 d of culture, the supernatant was clarified by depth and sterilizing filtration (Sartorius Stedim) prior to purification by protein G affinity chromatography (Mab Select Sure resin) using an AktaPurifer system (GE Healthcare Life Sciences). The concentration of purified 2C11-Novi formulated in phosphate buffered saline (PBS) at pH 7.4, was determined by measuring its absorbance at 280 nm by spectrophotometry. The structural integrity and purity of the purified 2C11-Novi recombinant mAb (1 g) were evaluated by sodium dodecyl sulfate PAGE (SDS-PAGE), size exclusion chromatography combined with high performance liquid chromatography (SEC-HPLC) and limulus amoebocyte lysate (LAL) test (Endosafe® Endochrome-K™ LAL test kit, Charles River).

In vitro TCR downregulation

T cells from Balb/c mice were plated at 5x10⁵ cells/mL and incubated overnight at 37°C in 96-round-well plate in RPMI-1640 cell culture medium supplemented with L-glutamine, 1 mM sodium pyruvate, 50 μM β-mercaptoethanol and with 10% fetal bovine serum (cRPMI), in the presence of increasing concentrations of the indicated anti-CD3 mAbs for 18 h. Cells were then washed and resuspended in cold PBS for analysis by flow cytometry using a phycoerythrin conjugated anti-TCR mAb (clone H57–597, BD Biosciences). All flow cytometric analyses were performed using a FACSCalibur™ flow cytometer and analyzed with the CellQuest Pro software (BD Biosciences).

FcγR binding assay

The binding of anti-mouse CD3 mAb to murine FcγR was assessed by flow cytometry as previously described.^25,29 The cell surface expression of CD16 (FcγRIII)/CD32 (FcγRIIb) and CD64 (FcγRI) on the RAW-264.7 cell line (ATCC) was confirmed using the PE-conjugated 2.4G2 and X54–5/7.1 mAbs, respectively (BD Biosciences). Briefly, RAW-264.7 cells were incubated with sub-saturating concentrations (50 µg/mL) of APC-conjugated mouse IgG2a mAb (clone MOPC-173, widely used as isotype control for flow cytometric experiments, Biolegend) and increasing concentrations of 145–2C11, 2C11-Novi or mIgG1 (Clone M2, anti-FLAG® mAb, Sigma-Aldrich) mAbs for 30 min at 4°C. After incubation, the cells were washed and analyzed by flow cytometry. Data represented are the mean results from triplicate samples, and the error bars are standard deviations of the means (SEM).

T cell proliferation assays

Spleen or lymph node cells were cultured at 37°C for three days in 96-well plate in cRPMI in the presence of increasing concentrations of anti-CD3 mAb or a fixed concentration of concanavalin A (ConA, 5 μg/mL). For the last 18 h, 1 μCi/well of 3H-thymidine was added to quantify cell proliferation (expressed in counts per minute, cpm). Radioactivity uptake was quantified using a microplate scintillation counter and results are expressed as the mean cpm of triplicate cultures ± SEM.

Measurements of peripheral blood T cell counts, anti-CD3 and cytokine concentrations in plasma

The plasma concentrations of 2C11-Novi and 145–2C11 were measured by flow cytometry. A competition assay was developed which measured binding inhibition of APC-conjugated 145-C11 (BD Biosciences) to gated CD4+ T cells from murine splenocytes. The plasma levels of biologically active anti-CD3 mAbs were extrapolated from a standard curve obtained with 145–2C11 spiked in plasma at varying concentrations. The count of circulating T cells was also measured by flow cytometry. Briefly, anti-coagulated whole blood samples (50 µL) from anti-CD3-treated mice were incubated with fluorescently labeled anti-mouse CD4 (clone RM4–5) mAb in a TruCOUNT™ tube according to the manufacturer’s instructions (BD Biosciences). After incubation at room temperature in the dark for 15 min, the red blood cells were lysed with 450 µL of lysing solution and further incubated at room temperature in the dark for 10 min prior to analysis by flow cytometry. The cytokines IL-6, IFNγ and TNF were quantified in mouse plasma using the Milliplex mouse cytokine technology following the manufacturer’s instructions (Millipore). TGF-β was quantified in mouse plasma by ELISA (TGF-β1 Ready-SET-Go ELISA kit, eBioscience).

Flow cytometric analysis of T cells in vivo

Spleen and lymph nodes (popliteal, inguinal, brachial and axillary) were extracted at various time points and enzymatically digested in a solution of 1 mg/mL DNase and 2.5 mg/mL collagenase for 45 min at 37°C. Mouse leukocytes were isolated from peripheral whole blood using a solution containing ammonium chloride, which lyses red cells with minimal effect on lymphocytes. Single cell suspensions were incubated for 20 min in cold PBS with 5–10 µg/mL of the following fluorescently labeled anti-mouse mAbs: anti-CD3-PE (145–2C11), anti-CD4-FITC (RM4–5), anti-CD8a-PerCP (53–6.7), anti-CD25-APC (PC61), anti-CD69-APC (H1.2F3) or anti-TCR-PE (H57–597), all purchased from BD Biosciences. The samples contained 10 µg/mL of Mouse Fc Block™ (BD Biosciences) to reduce non-specific binding to FcγR-bearing cells. Cells were then washed and resuspended in PBS for analysis by flow cytometry. Data represent mean mfi of samples from five to six mice ± SEM.

Quantification of apoptosis in vivo

Animals were injected intraperitoneally with a single dose of 50 µg 2C11-Novi, 20 µg 145–2C11 or 50 µg mouse IgG1κ isotype control (clone 9E10, ATCC). Twenty-four hours later the mice were injected with 200 μL of Apo-TRACE™ solution (Sigma-Aldrich) via the tail vein. Two hours later, the mice were euthanized and the lymph nodes were harvested. After tissue lysis, the samples were analyzed by spectrophotometry. The fluorescence emission intensity was measured at 563 nm as recommended by the manufacturer.

Induction and evaluation of EAE

Female SJL mice were immunized subcutaneously with 200 μL of an emulsion containing 100 μg PLP_139–151 peptide (HSLGKWLGHPNKF, AnaSpec) and 800 μg of Mycobacterium tuberculosis H37Ra (Difco) in complete Freund adjuvant (CFA). Mice also received an intraperitoneal injection of 400 ng of pertussis toxin (Sigma Aldrich) on the day of immunization. Mice were administered 5 µg of 145–2C11, 2C11-Novi or mouse IgG1κ isotype control (clone 9E10, ATCC) dissolved in 200 μL of PBS by gastric intubation with a 18-gauge stainless steel feeding needle once a day for five consecutive days. Mice were immunized two days after the last antibody feeding. Individual animals were observed daily and clinical scores assessed in a blinded fashion on a 0–5 scale as follows: 0, no abnormality; 1, limp tail or hind limb weakness but not both; 2, complete tail paralysis 3, partial hind limb paralysis; 4, complete hind limb paralysis; 5, moribund state. Paralysed animals were afforded easier access to food and water. The data are reported as the mean clinical score for each group. For all EAE experiments, there were seven mice per treatment-group. The data are reported as the mean daily clinical score ± SEM.

Statistical analysis

All statistical analyses were performed using GraphPad Prism version 5.01 for Windows (GraphPad Software, www.graphpad.com). Statistically significant differences between two groups were analyzed by unpaired Student’s two-tailed t test. For experiments with multiple groups a one-way analysis of variance (ANOVA) was used. P-values were then adjusted using the Bonferroni correction to compensate for multiple testing. Statistically significant differences between groups are indicated as follows: *p < 0.05; **p < 0.01; and ***p < 0.001.

Glossary

Abbreviations:

AICD

activation-induced cell death

EAE

experimental autoimmune encephalitis

FcγR

Fc gamma receptors

IL

interleukin

mAb

monoclonal antibody

PD

pharmacodynamics

PK

pharmacokinetics

PLP

proteolipid protein

T1D

type 1 diabetes mellitus

TCR

T cell receptor

TGF-β

transforming growth factor-beta

TNF

tumor necrosis factor

Disclosure of Potential Conflicts of Interest

E.H., B.D., L.C., S.G., W.R., M.K.V. and Y.D. are shareholders in NovImmune SA. F.D., M.K.V. and Y.D. are inventors on patents filed by NovImmune SA relating to the clinical use of NI-0401/Foralumab. The other authors declare no competing interests.