Engineered CD3 antibodies for immunosuppression

INTRODUCTION

Organ transplantation offers the potential to improve quality of life and prolong survival for increasing numbers of patients with organ failure, and bone marrow or peripheral stem cell transplantation may constitute the only curative approach for patients with select haematological diseases. Since the early sixties, heterologous polyclonal sera against human thymocytes or lymphocytes (ATG or ALG) are employed for immunosuppression, but significant progress in transplantation medicine was achieved with the introduction of novel oral immunosuppressive agents like calcineurin inhibitors, mycophenolate mofetil and – more recently – sirolimus. However, acute rejection episodes or high risk clinical situations, e.g. in pretransplanted patients may require even more pronounced immunosuppression. In these situations, several monoclonal antibodies against defined antigens broaden the therapeutic armentarium today [1]. Among these are antibodies against the IL-2 receptor (CD25) or CD3, and antibodies against other antigens are expected to follow along this line [2].

CD3 ANTIBODIES FOR IMMUNOSUPPRESSION

So far, OKT3 – a murine IgG2a antibody – is the only approved CD3-directed antibody [3]. Although OKT3 is amongst the most powerful immunosuppressive agents, its application is associated not only with the inherent risks of intensive immunosuppression – mainly infections and tumours – but also with the induction of human anti-mouse antibodies (HAMA) and with significant infusion-related toxicity. Improved understanding of the mechanisms of action of CD3 antibodies and of the mechanisms responsible for their toxicity have guided the generation of novel CD3 antibodies with improved therapeutic potential (Table 1). Promising results of a clinical trial with one of these constructs are presented by Meijer et al. in this issue of Clinical and Experimental Immunology [4].

Table 1.

Engineered CD3 antibodies in clinical trials

Antibody	Construct	Phase	Indication	Reference
T3/4.A	Murine IgA	II	Kidney transplantation	[4]
YTH 12·5	Aglycosylated huIgG1	I	Kidney transplantation	[19]
HuM291	Mutated huIgG2	I	Kidney transplantation	[22]
		I	Bone marrow transplantation	[21]
hOKT3γ1[Ala-Ala]	Mutated huIgG1	I	Kidney transplantation	[26]
		I/II	Psoriasis arthritis	[27]
		II/III	Type I diabetes mellitus	[28]

IMPACT OF FC RECEPTOR BINDING FOR EFFICACY AND TOXICITY OF CD3 ANTIBODIES

Different mechanisms of action have been proposed to explain the immunosuppressive activity of CD3 antibodies. For example, binding of CD3 antibodies may sensitize T cells for up-take by the RHS – leading to depletion of peripheral T cells. Furthermore, crosslinking of CD3 induces intracellular signalling, which triggers T cell anergy or apoptosis – provided that T cells do not receive a ‘second signal’ via one of several costimulatory molecules. Furthermore, CD3 antibodies were reported to induce a shift in the T cell balance from Th1 to Th2 cells [5]. The majority of these proposed mechanisms appears to be mediated by the F(ab) portion of CD3 antibodies – suggesting that the Fc region may be dispensable for their therapeutic efficacy as immunosuppressive agents. This hypothesis was supported by a clinical study with F(ab′)₂ fragments, which demonstrated immunosuppressive activity comparable to the respective whole antibody [6].

Mechanisms of toxicity triggered by CD3 antibodies have been under intensive investigation for many years. Antibody triggered activation of T cells is influenced by the F(ab) portion of the respective CD3 antibody and is therefore influenced by affinity, valency and – possibly – the targeted epitope of CD3. On the other hand, indirect mechanisms of toxicity require interactions of the antibodies’ Fc region with the patients’ immune system. Thus, systemic complement activation was correlated to clinical toxicity [7]. Additionally, several lines of evidence suggested that interactions with cellular Fc receptors significantly contribute to anti-CD3 toxicity. For example, Fc receptor-expressing by-stander cells have long been demonstrated to significantly contribute to anti-CD3 triggered cytokine release, which is timely related to clinical side-effects [8]. This cytokine release is strongly dependent on the antibody isotype. Furthermore, CD3 antibodies of murine IgG1 isotype demonstrated significant donor-dependent variation in their capacity to trigger T cell proliferation and cytokine release [9]. Subsequently, this variation lead to the identification of a bi-allelic polymorphism of the FcγRIIa receptor (either arginin or histidine in position 131) [10]. This Fc receptor polymorphism determines the efficacy of interactions with murine IgG1 (and human IgG2), and thereby side-effect intensity of murine IgG1 CD3 antibodies [11].

Recent crystallographical studies have significantly improved our understanding of the molecular interactions between IgG and its cellular receptors [12]. This improved knowledge has allowed the generation of antibody mutants with altered and specifically designed Fc receptor binding properties [13]. Considering the impact of interactions between Fc regions and Fc receptors for anti-CD3-mediated totoxicity, novel non Fc receptor-binding antibody constructs were developed, aiming to reduce toxicity, while maintaining therapeutic efficacy of ‘gold standard’ OKT3 (Table 1).

ENGINEERED CD3 ANTIBODIES WITH REDUCED TOXICITY

Assuming that Fc regions of CD3 antibodies were triggering cytotoxicity in vivo, clinical application of F(ab)₂ fragments was a logical step forward. and indeed, these fragments proved immunosuppressive in vivo and were associated with less infusion-related side-effects [6]. Interestingly, also the immunogenicity of F(ab)₂ fragments was lower than that of whole murine antibodies. However, F(ab)₂ fragments display unfourable pharmacokinetics compared to whole antibodies and are expensive to produce. The latter holds true also for monovalent CD3 antibodies, which were also evaluated as alternative for OKT3 [14].

Other approaches aimed to reduce Fc receptor binding by employing engineered Fc regions. One of these constructs is T3/4.A; a murine IgA antibody against human CD3. IgA is the most abundantly produced antibody isotype in vivo, but its therapeutic potential for immunotherapy has not been thoroughly explored [15]. Interestingly, murine IgA does not interact with the human myeloid IgA receptor (CD89). For this reason, T3/4.A was selected as a non-Fc receptor binding CD3 antibody, which was nonmitogenic in vitro. Importantly, T3/4.A proved as effective in the therapy of kidney allograft rejections as OKT3, but side-effects were less pronounced – as reported by Meijer et al. in this issue of Clinical and Experimental Immunology [4]. Although this comparison was derived from a phase II study with historical control patients, these results are encouraging. However, as expected murine IgA was immunogenic in human patients, with 14 out of 18 patients developing HAMA two weeks after a single course of T3/4.A. Although the authors observed no interference of HAMAs with therapeutic efficacy in this therapeutic schedule, HAMA induction may prohibit re-treatment with T3/4.A, and may interfere with other therapeutic or diagnostic antibody applications in these patients. Unfortunately, there is no easy solution to this problem, since chimerization/humanization of T3/4.A to human IgA would be expected to generate a potent Fc receptor-binding molecule with all its potential drawbacks.

Binding of human IgG antibodies to cellular Fcγ receptors is highly affected by their glycosylation pattern [16,17]. Therefore, an aglycosylated humanized CD3 antibody was generated by CDR-grafting on a human IgG1 backbone, in which a single amino acid substitution (Asn→Ala in position 297) reduced glycosylation. As predicted, this construct demonstrated significantly reduced Fc receptor binding and complement activation, and thus proved nonmitogenic in vitro [18]. In vivo, low toxicity and signs of immunosuppressive activity were reported from a phase I study [19].

Another approach for a nonmitogenic CD3 antibody started from the humanized CD3 antibody HuM291, which was expressed as human IgG2 isotype. Human IgG2 is documented to confer a long plasma half life in vivo, and to reduce complement activation. Furthermore, human IgG2 does not interact with the class I or III Fcγ receptors (CD64 or CD16), and binding to FcγRII receptors was reduced by two engineered mutations in the constant regions of HuM291 (234 and 237 Val→Ala) [20]. As expected, this construct was nonmitogenic in vitro. As HuM291 dissociated quickly from cell surface CD3 molecules, only minimal internalization but sustained signalling by the T cell receptor was observed. Thereby, HuM291 effectively triggered apoptosis in activated human T cells [21]. In clinical phase I studies in renal allograft or allogeneic bone marrow transplanatation patients, the majority of patients did not demonstrate measurable cytokine levels after antibody application, infusion-related toxicity was low, and immunosuppressive activity was observed [22,21].

hOKT3γ1(Ala-Ala) is a humanized IgG1 version of OKT3, in which the amino acids in position 234 and 235 have been mutated to alanine [23]. Thereby, hOKT3γ1(Ala-Ala) was reported to loose complement-activating capacity, Fcγ receptor binding and mitogenicity. Furthermore, in vitro experiments and animal studies demonstrated hOKT3γ1(Ala-Ala) to induce clonal anergy [24], and a shift from Th1 to Th2 cells [25]. A subsequent phase I study with hOKT3γ1(Ala-Ala) demonstrated efficacy similar to that of conventional OKT3 in the treatment of renal allograft rejection with markedly fewer side-effects [26]. hOKT3γ1(Ala-Ala) was also tested in patients with psoriatic arthritis [27] or type I diabetes [28]. In both patient populations, no significant cytokine release was observed, infusion-related toxicity was low and – importantly – these phase II trials suggested clinical efficacy.

PERSPECTIVE

Despite potent novel immunosuppressive agents, OKT3 is still a viable therapeutic option in steroid-refractory solid organ rejection or GvHD. Novel engineered CD3 antibody constructs promise to reduce toxicity, while retaining therapeutic efficacy of anti-CD3 therapy. Thus, CD3-directed approaches may become more widely applicable in the treatment or prophylaxis of allograft rejection or GvHD, and may also be reconsidered for severe autoimmune diseases. In addition, their application for the induction of longterm tolerance may deserve further investigation [29,30].