A novel mouse model for systemic cytokine release upon treatment with a superagonistic anti-CD28 antibody

The adaptive immune system is known to provide highly specific and effective immunity against a broad variety of pathogens due to different effector cells. The most prominent are CD4+ T-cells which differentiate after activation into distinct subsets of effector and memory cells, amongst others T helper 1 (Th1) cells. We have recently shown that mouse as well as human Th1 cells depend on T cell receptor (TCR) signals concomitant with CD28 costimulation in order to secrete interferon  (IFN) which is considered as their main effector function. Moreover, there is a class of anti-CD28 monoclonal antibodies that is able to induce T cell (re-)activation without concomitant TCR ligation. These so-called CD28-superagonists (CD28-SA) have been shown to preferentially activate and expand CD4+ Foxp3+ regulatory T (Treg) cells and thereby efficaciously conferring protection e.g. against autoimmune responses in rodents and non-human primates. Considering this beneficial effect, CD28-SA were thought to be of great impact for immunotherapeutic approaches and a humanized CD28-SA was subjected to clinical testing starting with a first-in-man trial in London in 2006. Unexpectedly, the volunteers experienced life-threatening side effects due to a cytokine release syndrome (CRS) that was unpredicted by the preclinical studies prior to the trial. Retrospectively, CD4+ memory T cells within the tissues were identified as source of pro-inflammatory cytokines released upon CD28-SA administration. This was not predicted by the preclinical testing indicating a need for more reliable and predictive animal models. Whether mouse CD4+ T cells are generally irresponsive to CD28-SA stimulation or rather the lack of a bona fide memory T cell compartment in cleanly housed specific-pathogen-free (SPF) mice is the reason why the rodent models failed to predict the risk for a CRS remained unclear. To provide SPF mice with a true pool of memory/effector T cells, we transferred in vitro differentiated TCR-transgenic OT-II Th1 cells into untreated recipient mice. Given that Treg cells suppress T cell activation after CD28- SA injection in vivo, recipients were either Treg-competent or Treg-deficient, wild type or DEREG mice, respectively. Subsequent CD28-SA administration resulted in induction of systemic pro-inflammatory cytokine release, dominated by IFN, that was observed to be much more pronounced and robust in Treg-deficient recipients. Employing a newly established in vitro system mirroring the in vivo responses to CD28-SA stimulation of Th1 cells revealed that antigen-presenting cells (APCs) amplify CD28-SAinduced IFN release by Th1 cells due to CD40/CD40L-interactions. Thus, these data are the first to show that mouse Th1 cells are indeed sensitive to CD28-SA stimulation in vivo and in vitro responding with strong IFN release accompanied by secretion of further pro-inflammatory cytokines, which is compatible with a CRS. In conclusion, this study will facilitate preclinical testing of immunomodulatory agents providing a mouse model constituting more “human-like” conditions allowing a higher degree of reliability and translationability.