RESUMENLos anticuerpos anti-CD28 superagonistas inducen activación

de los linfocitos T en ausencia de estimulación del TCR. Los ensa-yos clínicos de uno de tales anticuerpos, TGN1412, han resulta-do en severas complicaciones en los voluntarios que lo recibie-ron. El análisis de las características y modo de acción de los supe-ragonistas de CD28 sugiere que dichos efectos son similares a lospreviamente observados con el uso terapeútico de anticuerposanti-CD3 y al síndrome tóxico inducido por superantígenos bac-terianos.

PALABRAS CLAVE: Célula T/ CD28/ Ensayo clínico/ Anti-cuerpo/ Superagonista.

ABSTRACTSuperagonistic anti-CD28 antibodies induce T cell activation

in the absence of TCR triggering. Clinical trials of one of such anti-bodies, TGN1412, resulted in severe adverse effects in the volun-teers receiving the drug. Analysis of characteristics and mecha-nism of action suggests that those effects are similar to those pre-viously reported for the therapeutic use of anti-CD3 antibodiesand the toxic shock syndrome triggered by bacterial superanti-gens.

KEY WORDS: T cell/ CD28/ Clinical trial/ Antibody/ Supera-gonist.

147

PanoramaInmunología

Vol. 25 / Núm 2/ Abril-Junio 2006: 147-150

TGN1412: Superagonist Dr. Jekyll and Superantigen Mr. Hyde

G. Criado

Kennedy Institute of Rheumatology, Imperial College London, London (UK).

TGN1412: SUPERAGONISTA DR. JEKYLL Y SUPERANTÍGENO MR. HYDE

Recibido: 20 Junio 2006Aceptado: 26 Junio 2006

INTRODUCTIONLast March, a phase I clinical trial resulted in severe

adverse effects for the six volunteers receiving the experimentalantibody TGN1412. As preliminary reports of the UKMedicines and Healthcare products Regulatory Agencyhave ruled out contamination of the antibody preparationas the source of the adverse reaction, the action of theantibody itself has been blamed. Since experimental workin animals did not show significant toxicity, progression toclinical testing ensued with the known results, and speculationhas arisen as to how TGN1412 could trigger such responseand as to whether it could have been predicted. Analysis

of some features of TGN1412 structure and mechanism ofaction suggest that it could behave as a superantigen, andpoint to some cautions to be kept in the future when testingsimilar reagents.

MECHANISM OF ACTION OF CD28 SUPERAGONISTSTGN1412 is a member of a class of monoclonal antibodies

termed superagonists that target the T cell co-stimulatorymolecule CD28. Treatment of T cells with anti-CD28superagonistic antibodies overcomes the need for a primaryTCR-dependent signal and results in T cell activation on its

TGN1412: SUPERAGONIST DR. JEKYLL AND SUPERANTIGEN MR. HYDE VOL. 25 NUM. 2/ 2006

148

own, hence the term «superagonist»(1). They were originallydescribed for rat CD28, and their activity has been linkedto the generation of «signalling competent» CD28 oligomersby virtue of their binding to the C’’D loop in theimmunoglobulin variable region of the molecule, far fromthe binding site of the CD28 ligands B7.1 and B7.2(2) (Fig. 1).Such «signalling competent» CD28 promotes activation andnuclear translocation of transcription factors in the absenceof overt proximal T cell signalling events characteristic ofTCR-dependent T-cell activation, like ζ phosphorylationand ZAP-70 activation. However, superagonist CD28antibodies require the TCR signalling machinery and promotethe activation of Lck, LAT, PLCY1 and PKCtheta, leadingto the suggestion that CD28 superagonists work by enhancinglow level tonic signals from the TCR(3, 4).

In the rat system, preferential expansion of Th2 andTreg cells has been shown after in vivo treatment with anti-CD28 superagonistic antibodies(5-6). Furthermore, whentested in animal models of autoimmune Th1 diseases, likeEAE and reactive arthritis, they showed a therapeuticeffect(7,8).

These promising results prompted the interest towardsthe potential benefits of similar reagents in humans.Subsequently, superagonistic antibodies to human CD28were developed that confirmed the TCR-independentactivation of nuclear factors, cytokine production and

proliferation(2). TGN1412 was developed by humanisationof the Fc binding portion of one of these antibodies thatmade it suitable for human therapy(9). Paradoxically, theonly therapeutic application of TGN1412 suggested so faris not of an immunosupressive nature, since it has beenreported to induce potent anti-tumour cytolytic activityin PBMCs from B cell chronic lymphocytic leukaemia affectedpatients(10), suggesting a predominant immunostimulatoryactivity.

LESSONS FROM THE THERAPEUTIC USE OF ANTI-CD3 AGONISTIC ANTIBODIES

T cells have long been the target of strategies aimed atpreventing unwanted immune responses. In fact, OKT3was the first monoclonal antibody used in the clinic, evenbefore its target, the TCR-CD3 complex, was identified.OKT3 promoted immunosupression and prevented transplantrejection in kidney transplant recipients(11, 12). However, italso triggered an acute cytokine release syndrome that wasprevented by corticosteroids(13). In addition, its therapeuticefficacy was limited by the development of neutralisingantibodies(14). This could be overcome by humanisation ofthe Fc portion of the antibody. On the other hand, studiesin mice showed that low affinity and non FcγR bindingantibodies avoided the cytokine release syndrome whilstmaintaining the immunosuppressive activity. Consequently,humanised non-FcγR binding versions of OKT3 and otheranti-CD3 antibodies were developed(15, 16). The combinationof both strategies has produced a safe and efficient therapyfor transplant rejection and autoimmune syndromes.Interestingly, their mechanism of action is very similar tothat reported for CD28 superagonists, as they also promotethe development of regulatory T cells(17).

Several other factors can contribute to exacerbate theseverity of the toxic syndrome in the case of the Londonvolunteers in comparison to the early anti-CD3 clinical trials.Not the least important is the immune status of the recipients:the men in London were healthy individuals with an intactimmune system while the participants in the kidney transplanttrials were already under an immunosuppressive regimeto prevent transplant rejection. Another important factoris the dose of antibody given: although the single dose of100 µg/kg used in the case of TGN1412 is within the rangeused in the therapeutic protocols currently applied withthe non-FcγR binding anti-CD3 antibodies, the latter ispreceded by a dose-escalating regime starting in the ng/kgrange.

In the development of TGN1412, consideration has beengiven to the humanisation side of the equation, but the

Figure 1. Superagonist monoclonal antibodies to CD28 like TGN1412 bindto a site in the C”-D loop of the V-like Ig extracellular domain of the disulfide-linked CD28 homodimer (dark grey). The interaction of CD28 with CD86 (B7-2) monomers or CD80 (B7-1) homodimers is monomeric and is mediated bya distant, exposed site in the F-G loop containing the sequence MYPPPY (lightgrey).

possibility of deleterious effect of agonistic antibodycrosslinking on T cells has been ignored, despite what thehistory of therapeutic treatment with agonistic anti-CD3antibodies clearly illustrates. In fact, humanisation probablyimproved FcγR binding that may have contributed to thetoxicity of TGN1412.

AFFINITY AND THE CASE OF BACTERIALSUPERANTIGENS

The likely cause of the adverse effects suffered by thevolunteers during the clinical trial of TGN1412 is a cytokinerelease syndrome triggered by overstimulation of T cells,similar to that described in the initial trials of OKT3(18). Giventhat such an effect had not been reported previously eitherin the animal model studies of CD28 superagonists or inthe preclinical testing of TGN1412 in rhesus and cynomolgusmonkeys, differences in affinity at both ends of the antibodyhave been signalled as responsible. On the antigen-bindingregion, differences surrounding the antigen binding sitebetween human and rhesus CD28 could result in reducedaffinity of TGN1412 for rhesus CD28. Likewise, on the Fc-binding region the affinity of rhesus FcγR for humanisedantibodies is likely to be lower than for endogenous antibodies.The combination of both circumstances would result indecreased interaction of the humanised antibody with rhesusCD28 and reduced T cell activation, masking a potentialtoxic effect in the human system. In the absence of experimentaldata to support the differences in affinity, this remains aworking hypothesis.

The affinity hypothesis provides an interesting parallelbetween the behaviour of CD28 superagonist and bacterialsuperantigens in humans and animal models. Bacterialenterotoxins stimulate a large fraction of peripheral T cellsin a TCR Vβ specific manner and are, therefore, termedsuperantigens(19). Just as suggested for TGN1412, bacterialsuperantigens trigger in humans a cytokine storm leadingto a lethal toxic syndrome, mainly mediated by TNFα release.In contrast, in mice the response is milder and, in order tobe lethal, they must be sensitised with galactosamine orLPS(20). Such differences in the magnitude of the responsehave been attributed to the affinity of the specific bacterialtoxins for the TCR Vβ regions on the T cells and for theMHC-II on the APCs, since they are lower for both murinemolecules than for their human counterparts(21). In addition,given that CD28 is expressed on all T cells, the responseto TGN1412 is not Vβ-restricted but targets the whole T cellpopulation.

The similarities in the behaviour of superagonisticantibodies and bacterial superantigens are not restricted to

the affinity of their interaction with their ligands, but aremanifested in the progression of the response that followsas well. As mentioned, the initial response to bacterialsuperantigens induces an acute release of cytokines. Althoughthe toxicity is mainly mediated by TNFα and IFNγ, a fullarray of cytokines is produced. However, after the initialburst, a predominant Th2 and Treg phenotype emerges(22-

25). Similarly, the reports for rat superagonistic antibodiesshowed that the predominance of Th2 and Treg cells ispreceded by a general activation of T cells, includingproduction of the prototypical Th1 cytokine, IFNγ(5).

In summary, the changes introduced to ensure theefficacy of superagonistic CD28 antibodies in humans mayhave maximised their potency to the extent of convertingTGN1412 in a superantigen-like ligand with devastatingacute effects.

ACKNOWLEDGMENTSI am grateful to Dr. JM Rojo for stimulating discussions

and helpful suggestions.GC work is funded by the Arthitis Research Campaign

(arc).

CORRESPONDENCE TO: Gabriel CriadoKennedy Institute of RheumatologyImperial College London1 Aspenlea RoadLondon W6 8LH (UK)E-mail: g.criado@imperial.ac.uk

REFERENCES1. Tacke M, Hanke G, Hanke T, Hunig T. CD28-mediated induction

of proliferation in resting T cells in vitro and in vivo withoutengagement of the T cell receptor: evidence for functionally distinctforms of CD28. Eur J Immunol 1997; 27: 239-247.

2. Luhder F, Huang Y, Dennehy KM, Guntermann C, Muller I,Winkler E, et al. Topological requirements and signaling propertiesof T cell-activating, anti-CD28 antibody superagonists. J Exp Med2003; 197: 955-966.

3. Bischof A, Hara T, Lin CH, Beyers AD, Hunig T. Autonomousinduction of proliferation, JNK and NF-kB activation in primaryresting T cells by mobilized CD28. Eur J Immunol 2000; 30: 876-882.

4. Dennehy KM, Kerstan A, Bischof A, Park JH, Na SY, Hunig T.Mitogenic signals through CD28 activate the protein kinase Ctheta-NFkB pathway in primary peripheral T cells. Int Immunol 2003;15: 655-663.

5. Rodriguez-Palmero M, Hara T, Thumbs A, Hunig T. Triggeringof T cell proliferation through CD28 induces GATA-3 and promotes

149

INMUNOLOGÍA G. CRIADO

150

T helper type 2 differentiation in vitro and in vivo. Eur J Immunol1999; 29: 3914-3924.

6. Lin CH, Hunig T. Efficient expansion of regulatory T cells in vitro andin vivo with a CD28 superagonist. Eur J Immunol 2003; 33: 626-638.

7. Beyersdorf N, Gaupp S, Balbach K, Schmidt J, Toyka KV, Lin CH,et al. Selective targeting of regulatory T cells with CD28 superagonistsallows effective therapy of experimental autoimmuneencephalomyelitis. J Exp Med 2005; 202: 445-455.

8. Rodriguez-Palmero M, Franch A, Castell M, Pelegri C, Perez-CanoFJ, Kleinschnitz C, et al. Effective treatment of adjuvant arthritiswith a stimulatory CD28-specific monoclonal antibody. J Rheumatol2006; 33: 110-118.

9. European patent application EP1600164.10. Lin CH, Kerkau T, Guntermann C, Trischler M, Beyersdorf N,

Scheuring Y, et al. Superagonistic anti-CD28 antibody TGN1412as a potential immunotherapeutic for the treatment of B CellChronic Lymphocytic Leukemia. Blood (ASH Annual MeetingAbstracts) 2004; 104: 2519.

11. Cosimi AB, Colvin RB, Burton RC, Rubin RH, Goldstein G, KungPC, et al. Use of monoclonal antibodies to T-cell subsets forimmunologic monitoring and treatment in recipients of renalallografts. N Engl J Med 1981; 305: 308-314.

12. Ortho Multicenter Transplant Study Group. A randomized clinicaltrial of OKT3 monoclonal antibody for acute rejection of cadavericrenal transplants. N Engl J Med 1985; 313: 337-342.

13. Chatenoud L, Ferran C, Legendre C, Thouard I, Merite S, ReuterA, et al. In vivo cell activation following OKT3 administration.Systemic cytokine release and modulation by corticosteroids.Transplantation 1990; 49: 697-702.

14. Bach JF and Chatenoud L. Immunologic monitoring of OrthocloneOKT3-treated patients: the problem of antimonoclonal immuneresponse. Transplant Proc 1987; 19 (suppl): 17.

15. Alegre ML, Peterson LJ, Xu D, Sattar HA, Jeyarajah DR, KowalkowskiK, et al. A non-activating "humanized" anti-CD3 monoclonal

antibody retains immunosuppressive properties in vivo.Transplantation 1994; 57: 1537-1543.

16. Cole MS, Stellrecht KE, Shi JD, Homola M, Hsu DH, Anasetti C,et al. HuM291, a humanized anti-CD3 antibody, isimmunosuppressive to T cells while exhibiting reduced mitogenicityin vitro. Transplantation 1999; 68: 563-571.

17. Chatenoud L. Anti-CD3 antibodies: towards clinical antigen-specific immunomodulation. Curr Opin Pharmacol 2004; 4: 403-407.

18. Can super-antibody drugs be tamed? Nature 2006; 440: 855.19. Kappler J, Kotzin B, Herron L, Gelfand EW, Bigler RD, Boylston

A, et al. V beta-specific stimulation of human T cells by staphylococcaltoxins. Science 1989; 244: 811-813.

20. McCormick JK, Yarwood JM, Schlievert PM. Toxic shock syndromeand bacterial superantigens: an update. Annu Rev Microbiol 2001;55: 77-104.

21. Herman A, Croteau G, Sekaly RP, Kappler J, Marrack P. HLA-DRalleles differ in their ability to present staphylococcal enterotoxinsto T cells. J Exp Med 1990; 172: 709-717.

22. Florquin S, Amraoui Z, Goldman M. Persistent production of TH2-type cytokines and polyclonal B cell activation after chronicadministration of staphylococcal enterotoxin B in mice. J Autoimmun1996; 9: 609-615.

23. Sundstedt A, Hoiden I, Rosendahl A, Kalland T, van Rooijen N,Dohlsten M. Immunoregulatory role of IL-10 during superantigen-induced hyporesponsiveness in vivo. J Immunol 1997; 158: 180-186.

24. Grundstrom S, Cederbom L, Sundstedt A, Scheipers P, Ivars F.Superantigen-induced regulatory T cells display different suppressivefunctions in the presence or absence of natural CD4+CD25+regulatory T cells in vivo. J Immunol 2003; 170: 5008-5017.

25. Feunou P, Poulin L, Habran C, Le Moine A, Goldman M, BraunMY. CD4+CD25+ and CD4+CD25- T cells act respectively asinducer and effector T suppressor cells in superantigen-inducedtolerance. J Immunol 2003; 171: 3475-3484.

TGN1412: SUPERAGONIST DR. JEKYLL AND SUPERANTIGEN MR. HYDE VOL. 25 NUM. 2/ 2006