February 15, 2017 (Vol. 37, No. 4)
David D. Manning Ph.D. Associate Research Fellow AMRI
A Discussion of the Chemistry of ADCs
Antibody-drug conjugates (ADCs), also known as armed antibodies, are positioned to be a significant source of next-generation oncolytic therapies. There has been explosive growth in ADC R&D, especially since the development and subsequent marketing approval of Mylotarg (gemtuzumab ozogamicin; withdrawn from market in 2010), Adcetris (brentuximab vedotin), and Kadcyla (ado-trastuzumab emtansine). Estimates place the global ADC market at $10 billion annually after 2024 with seven to 10 new commercial ADC launches projected in the next decade.
ADCs marry the selective targeting properties of antibodies with the potency of highly cytotoxic small molecules. The antibody targets and adheres to a selected antigenic cell-surface receptor, ideally only expressed on the target cancer cell. Once an ADC binds to its target cell, the cell internalizes the ADC through endocytosis, and the cytotoxic payload is then released in the lysosomal cellular compartment to provide precise, selective delivery to cancerous cells.
The development of ADCs brings many challenges, however. Multiple disciplines across drug development must engage to successfully discover, develop, evaluate and eventually manufacture a therapeutically relevant ADC. To illustrate, large macromolecular ADCs have a complex architecture whose assembly, manufacture, and analysis presents challenges for organizations without significant experience in biological conjugation, optimization, and the development of the payload-linker (PL; a component used for antibody attachment and subsequent release of the small molecule payload).
For many companies, outsourcing this component of ADC development may make sense. In this article, we will focus on the complexities of developing ADCs with an emphasis on the nature of the PL.
ADC Anatomy and PLs
Structurally, ADCs consist of an antibody, a chemical linker, and a terminal payload (Figure 1). Multiple PL units are affixed to an antibody and typically range from about two to eight. The antibody is usually a humanized monoclonal antibody (mAb) of the IgG class, while the payload is commonly a highly cytotoxic small molecule. The linker varies in composition.
ADCs are demanding to design and develop. The PL component is deceptively simple in concept, normally represented in graphics as a mere bridge connecting a payload to the antibody. In practice, the linker of the PL is an important component of the overall performance of the ADC. For example, several classes of PLs are designed to be cleaved and subsequently release the active form of the payload once the ADC is inside the cell. However, in contrast to the deliberate intracellular frailty, scientists must design a PL that is stable in human plasma, lest the ADC prematurely release its toxic payload and drive indiscriminate cell killing. The PL must be stable for several days in human plasma, given the fact that antibodies can circulate in the bloodstream for that long. The PL is developed with this in mind.
For example, ester linkages between linker and payload are often too labile to achieve sufficient plasma stability. As a whole, PLs require highly skilled scientists to design and construct, as multiple complex and often-sensitive chemical reactions are required to attach the payload to the linker and then the payload-linker to the antibody.
Linker Chemistry—Antibody Side
Bioconjugation chemistry has historically relied on the sulfhydryl or amino functional groups found in the natural amino acids within the antibody. A typical antibody may contain 50 or more lysines and up to 12 cysteines as potential conjugation sites. The bioconjugation chemistry of these groups produces an ADC that is heterogeneous, consisting of multiple species, which can offer different biological profiles.
In this respect, the ratio of linked drug to antibody (drug antibody ratio or DAR) is a critical factor to consider when designing an ADC. Early ADC research found that high DAR was associated with increased clearance, the potential for aggregation, and increased toxicity. Therefore, antibodies with site-specific conjugation chemistries are now sought to improve ADC homogeneity.
The drive to produce homogenous ADCs has impacted PL design. For example, the non-natural amino acids p-acetyl phenylalanine (pAF) and p-azidomethyl-l-phenylalanine (pAMF) have been specifically engineered into ADCs. The acetyl group of pAF can form stable oximes with hydroxylamine-bearing PLs to create homogenous ADCs. The azide of pAMF enables chemoselective click chemistry with a corresponding PL bearing a strained alkyne.
Although homogenous ADCs are on the horizon, bioconjugation of mAb amines or mAb thiols remains a common approach. In this respect, PLs with a terminal antibody-side activated ester or maleimide are frequently encountered (Figure 2). The conjugation to mAb lysines is often conducted using N-hydroxysuccinimide esters (NHs) although these esters have moderately short half-lives (<2 hours) in aqueous media. Pentafluorophenyl esters (PFs) have improved aqueous stability and may offer an advantage for conjugation of PLs. Although cysteine-maleimide conjugates remain popular, evidence has emerged that maleimide-derived bioconjugates can undergo reversion through thiol-exchange reactions.
Linker Chemistry—Payload Side
Reactive amines and thiols have been the favored functional groups to attach the payload to the linker. The commonly employed auristatins and maytansinoids connect to the linker using these motifs. In the case of maytainsine, a thiol-bearing spacer was introduced. Currently, the pool of suitable payloads for ADCs is quite limited, with tubulin inhibitors of the auristatin (MMAE and MMAF) and maytansinoid (DM1 and DM4) class comprising 80% of the payloads in current clinical development. The paucity of diversity in both mechanism and payload identity is an opportunity for the future development of ADCs, but until new payloads are discovered, varying linker composition can modulate the properties of the ADC. New technologies that use other functional groups, such as hydroxyl, which are present on many natural products, are sought to expand the arsenal of feasible payloads.
Payload Release—Cleavable and Non-Cleavable Linkers
Linkers are categorized as cleavable or non-cleavable. Non-cleavable linkers do not fragment. Instead, proteases digest the antibody protein backbone leaving the payload-linker tagged with a terminal amino-acid residue. Kadcyla is an example of this type (Figure 3). Cleavable linkers fragment depending upon the environment within a cell.
The three common cleavage mechanisms are enzymatic, disulfide, and pH. Adcetris uses a valine-citrulline para-aminobenzyl alcohol motif as part of the linker. Amide bond cleavage of the substrate by a protease results in release of the parent payload after 1,6–elimination of the p-aminobenzyl carbamate moiety. The kinetics of the enzymatic cleavage and subsequent free payload release are important considerations during development of new PLs. In another example, Mylotarg uses a two-step cleavage method using both pH and disulfide triggers.
The growth track of ADC-based pharmaceuticals is unquestionably on the upswing. As the ADC space matures, novel antibodies, linkers, and payloads will be developed to keep pace with the goal of delivering effective medicines with improved safety. Increased pressure to create homogenous ADCs will drive new conjugation approaches. ADC discovery will catalyze the development of new chemistry and create opportunities for scientists with the expertise required to work with these entities.