Over the past several years, immune checkpoint inhibitors (CPI) targeting CTLA-4, PD-1 and PD-L1 have revolutionized cancer treatment. However, despite their impressive clinical efficacy, several research studies have demonstrated that the immunosuppressive tumor microenvironment is responsible for preventing many patients from deriving such benefit. In many cases, this is likely related to inadequate activation of the well-known CD28 costimulatory receptor on T cells. Both PD-1 and CTLA-4 interfere with CD28 signaling either through intracellular phosphatase recruitment and/or competition for the CD28 ligands CD80/CD86 (Hui et al. 2017; Rowshanravan et al. 2018), and blockade of PD-1, PD-L1 or CTLA-4 alone is insufficient to supply an activating signal to CD28. Loss of CD80/CD86 expression has been described as a mechanism of tumor immune escape (Zang et al. 2007). Provision of a CD28 costimulatory signal in addition to checkpoint inhibition might therefore result in more potent, clinically meaningful anti-tumor immune responses than checkpoint inhibition alone.
CD80 has actually been described to bind PD-L1 in cis, blocking the latter’s checkpoint interaction with PD-1 but preserving CD80’s ability to activate CD28 (Butte et al. 2007; Zhao et al. 2019). Since PD-L1 is also a tumor antigen, we wondered if it might be possible to modify CD80 in a way to subvert PD-L1 into a CD28 costimulator -- for cancer therapy.
To do this, we subjected CD80 to protein engineering by directed evolution (Levin et al. 2019), using selective pressures of PD-L1, CD28 and CTLA-4. This process resulted in a variant CD80 IgV domain (vIgD™) with highly improved PD-L1 affinity compared to wild-type CD80, while retaining binding to CD28 and CTLA-4. This new domain, when fused with an immunoglobulin Fc and expressed as a dimer, created the drug candidate called davoceticept, or designated ALPN-202.
This design is intended to promote T cell activation in two ways:
- When bound to PD-L1 on a tumor cell, davoceticept stimulates CD28 in trans, effectively turning PD-L1 into a CD28 activating ligand (pressing the T cell gas pedal).
- It also inhibits CTLA-4 and PD-L1 receptor-ligand interactions, acting as a dual checkpoint antagonist (removing the brakes).
In pharmacology studies, davoceticept looked very promising. In vitro, it improved T cell activation better than that observed with CPI alone – including in the challenging context of suppressive macrophages, which are thought to contribute further to the immunosuppressive TME. In in vivo tumor models, it also resulted in durable anti-tumor responses superior to that of CPI alone.
A particularly exciting aspect of davoceticept is its ability to embody multiple mechanisms of action while still being relatively simple and compact. This prompted us to understand its structure, and we were successful in doing so through x-ray crystallography of the CD80 vIgD – PD-L1 complex. Notably, a crystal structure of natural CD80-PD-L1 has not yet been reported, so this may be the field’s first insight into the structure of the CD80-PD-L1 interaction. The structure showed that davoceticept’s CD80 binds PD-L1 on one surface but binds to CD28 and CTLA-4 via a separate non-overlapping surface. These structural data aligned not only with our in vitro mechanistic data but also existing published data on this interaction (Zhao et al. 2019).
The next step is to study davoceticept in humans. Two clinical trials are ongoing, one as monotherapy (i.e., davoceticept alone) and another in combination with the anti-PD-1 therapy pembrolizumab (NCT04186637, NCT04920383). It’s exciting to see that, so far, davoceticept shows early signs in people of the activities seen in these preclinical studies (Moser et al, ASCO 2021). Hopefully one day it may prove to be an effective treatment for cancers.