Antibody-Drug Conjugates: Looking Ahead to an Emerging Class of Biotherapeutic

In the quest for more targeted therapies and potentially more clinically efficacious drugs, bio/pharmaceutical companies are increasing their research and product development in biologics. Although the majority of this work is focused on monoclonal antibodies (mAbs) and recombinant proteins, progress is being made in specialized drug types. Antibody–drug conjugates (ADCs), which consist of a mAb chemically linked to a small-molecule therapeutic, are a niche class of drugs that offer promise, particularly as oncology drugs. In August 2011, FDA approved Adcetris (brentuximab vedotin), codeveloped by Seattle Genetics and Millennium Pharmaceuticals (now part of Takeda Pharmaceutical), making it only the second ADC approved by FDA. With the approval of Adcetris, a drug for treating Hodgkins lymphoma and systemic anaplastic large-cell lymphoma and with a number of ADCs in clinical development, the key question is whether ADCs will be able to fill a role in biopharmaceutical development.

(COMPOSITING BY DAN WARD. IMAGES: NICK KOUDIS/INGRAM PUBLISHING/GETTY IMAGES)

ADCs at work

Adcetris consists of three parts: the chimeric IgG1 antibody cAC10, specific for human CD30, the microtubule-disrupting agent monomethyl auristatin E (MMAE), and a protease-cleavable linker that covalently attaches MMAE to cAC10 (1). Before the approval of Adcetris this year, the only other ADC approved by FDA was Mylotarg (gemtuzumab ozogamicin), approved more than 10 years ago in 2000. The drug, an anti-CD33 mAb conjugated to the cytotoxin calicheamicin, was developed by Wyeth (now part of Pfizer) and was granted accelerated approval in 2000 but was voluntarily withdrawn by Pfizer in 2010 because a required Phase III trial failed to demonstrate a survival advantage for Mylotarg plus chemotherapy compared with chemotherapy alone. Despite this setback, there are several ADCs currently in development, with more than 15 in Phase I development and several compounds from Roche and Pfizer in late-stage clinical trials. In the decade that has elapsed between the first ADC approval and the second, advances in the understanding of cancer biology, lessons learned from the development of mAbs as therapeutics, and better methods for linking small molecules to mAbs have coalesced to advance ADCs into the forefront of new therapies.

The most active area of development for this class of therapeutics has been oncology, where a mAb serves to target the therapy to cancer cells while a potent small-molecule chemotherapeutic provides the cell-killing efficacy. Both mAbs and small-molecule chemotherapeutics are used individually as cancer therapies, but an ADC is designed to overcome the limitations of each. MAbs are highly specific, but as therapeutics have demonstrated only modest efficacy and often are used in combination with a conventional chemotherapy. Chemotherapeutics are highly toxic, but nonspecific, and so suffer from poor side-effect profiles and dose-limiting toxicities. In combination, the ADC serves to keep the chemotherapuetic bound until it reaches the cancer cell, thereby limiting its ability to interact with nontargeted tissues and therefore limiting nonspecific toxicity (2).

The concept of an ADC is not a new one, but creating a clinically successful one has been challenging. For the therapeutic to work well, each of the parts—the antibody, the toxin, and the linker that holds them together—must be carefully considered.

Choosing the right antibody

In general, mAbs as therapeutics are selected to have high affinity for the targeted antigen and high selectivity. Other desirable properties in an antibody include long circulation times, immune-effector functions, and tumor-suppressing activity (2). When choosing the antigen, it is important that it be expressed at high levels in the tissue of interest to maximize the amount of ADC bound by the tumor, but at low levels elsewhere in the body to minimize off-target toxicity. Moreover, it is thought that internalization of the ADC is important for its effectiveness. Many of the chemical-linking strategies used to construct ADCs rely on conditions found inside a cell, either in the cytoplasm or in the lysosome, to release the active agent (3).

In some instances, developers have been able to leverage experience gained through the development of mAb therapies to create their ADC. Trastuzumab emtansine (T-DM1) is an ADC in Phase III, which combines trastuzumab, (Herceptin), which targets human epidermal growth factor receptor 2 (HER2) receptors in breast and stomach cancer, with a maytansine derivative DM1, a small-molecule cytotoxin that binds to tubulin to prevent microtubule formation, through a nonreducible bis-maleimido-trixyethylene glycol linker (4). Trastuzumab was developed by Genentech (now part of Roche) and was approved by FDA in 1998 for use in women with metastatic breast cancer who have tumors that overexpress the HER2 protein. The maytansine derivative DM1 and linking technology were developed by ImmunoGen. In the case of the ADC trastuzumab emtansine, developers were able to use a target that had already been validated and a well-characterized antibody with a known safety and efficacy profile as the starting point for an ADC.

Choosing the right cytotoxic small molecule

The earliest versions of ADCs used stand-alone chemotherapeutics such as doxorubicin, methotrexate, or vinca alkyloids as the cytotoxic arm of the conjugate. Clinical-trial results using these ADCs were disappointing, and it is thought that part of the problem was the relatively low potency of the toxins used (2). The newer classes of cytotoxins are at least 100-fold more potent than the older molecules, with in vitro potency against tumor cell lines of 10^–9 to 10^–11 M (5).

There are only a few major chemical classes of toxins being explored. They can be divided into two types, those that cause damage to DNA and those that interfere with tubulin polymerization. Calicheamicin, used in Mylotarg and in Pfizer's inotuzumab ozogamicin, an ADC in Phase III trials, binds to the minor groove of DNA and induces double-strand DNA breaks that result in cell death. Duocarmycins, isolated originally from Streptomyces bacteria, are DNA minor-groove binding alkylating agents (2). Fully synthetic duocarmycin derivatives are being used by the biopharmaceutical company Syntarga (acquired by the pharmaceutical company Synthon in June 2011) for ADC constructs (see sidebar).

The importance of linker technology

Microtubule disruptors are represented by two major classes: maytansinoids and auristatins. Maytansinoids are deriviatives of maytansine, a natural product originally isolated from the shrub Maytenus serrata. ImmunoGen has focused on development of this class of cytotoxic small molecules and associated linker technologies and has been devloping maytansinoid ADC compounds singularly and in partnership with other companies. In addition to trastuzumab emtansine, which is being codeveloped by Roche and ImmunoGen, another example of a maytansinoid ADC being developed by ImmunoGen is the company's IMGN901, which uses the maytansinoid DM4. Auristatins are synthetic analogs of dolostatin 10, a natural product derived from a marine mollusk, Dolabela auricularia. Like the maytansinoids, auristatins are microtubule disruptors. Millennium and Seattle Genetics' ADC Adcetris is a conjugate of an anti-CD30 mAb to monomethyl auristatin E (MMAE). Seattle Genetics focuses on the development of auristatin-conjugated ADCs, using the auristatins MMAE and monomethyl auristatin F (MMAF) and proprietary linkers.

Choosing the right linker

Developing the right linker and method of attachment is a crucial part ADC development. "Many areas around the process have improved, however, the linker strategy for ADC manufacturing and their application has certainly contributed perhaps the most in moving the field forward," says Grant Boldt, director of business development at the CMO SAFC. The creation of linkers that are stable in circulation but labile upon binding of the ADC to its target has resulted in the current generation of ADCs having better stability and lower systemic toxicity than earlier ADCs, according to Boldt. Early versions of ADCs, including Mylotarg, suffered from instability while in circulation. The linkage between the mAb and the cytotoxic small molecule were destroyed by endogenous proteases in the blood, and the premature release of the cytotoxin resulted in side-effect profiles similar to that of an unconjugated chemotherapeutic. The current generation of linkers is more resistant to degradation in the blood while still allowing release of the payload at the target. Choice of a linker is influenced by which toxin is used, as each toxin has different chemical constraints (6).

Linkers can be divided into two broad categories: cleavable and noncleavable. Cleavable linkers rely on processes inside the cell to liberate the toxin, such as reduction in the cytoplasm, exposure to acidic conditions in the lysosome, or cleavage by specific proteases within the cell. Noncleavable linkages require catabolic degradation of the conjugate for release of the cytotoxic small molecule. The released cytotoxic small molecule will retain the linker and the amino acid by which it attached to the mAb. Importantly, both classes are designed to release the cytotoxic small molecule only after the ADC has reached the interior of the cancer cell (2).

There are a limited number of chemical moities on proteins, including mAbs, that are available for chemical modification. Linkers can attach to the mAb through the amino groups of lysine residues, or by the thiol groups on cysteine residues. Attachment is a pseudorandom process: in theory, any of the targeted amino acids within the mAb, either cysteine or lysine, can be modified (3). According to Boldt, the conjugation reaction results in a heterogeneous mixture of conjugated species, but the proportion of each species in the mixture is reproducible from batch-to-batch and quantifiable.

Putting it all together

Producing the ADC requires both biologic-based and small-molecule manufacturing. "One of the biggest challenges in manufacturing ADCs is controlling all the components that go into the final conjugation step," says Boldt. "Namely, the three main components that make up an ADC (e.g., antibody, linker, and payload) are all manufactured in very different ways. For example, it is not uncommon for these components to be manufactured by synthetic chemistry and mammalian cell culture. Thus, there presents a challenge in ensuring all these components have been manufactured under cGMP, and subsequently bringing them all together to generate the final ADC under cGMP, as well."

The biologics portion of the ADC and the high-potency API require very different handling methods, and manufacturers must make sure that handling requirements for both are met. "It is imperative that manufacturers emphasize the protection of the product from workers as well as the protection of workers from the product," says Jason Brady, head of business development, conjugates and cytotoxics at the CMO Lonza. Clinical ADC manufacturing is executed in an aseptic biological manufacturing environment to protect the product from contamination, explains Brady. Once conjugated with the high-potency API (which is manufactured in a high-containment environment), the resulting ADC also is handled under high-containment conditions. The level of containment is determined by occupational exposure limits for the high-potency API and resulting ADC. The environment must provide manufacturing personnel with isolation from cytotoxic chemicals in the occupational exposure range of 5 ng/m3 of air. Also important is that facility design includes design of equipment and process contact surfaces that permit clean-in-place and steam-in-place to remove minute traces of residual drug contamination during both interbatch and product changeover cleaning, according to Brady.

Room for improvement

As ADCs advance in the clinical pipeline so does the technology to manufacture ADCs to control certain product and process conditions. "New technology that can limit the heterogeneity of ADC products is something that will be important in the future," says Brady. "ADCs made via current technologies are heterogenous mixtures. Heterogeneity can be controlled and measured by robust and reproducible manufacturing processes and proper analytics, but new technologies will likely emerge to influence and improve ADC manufacturing," he explains. Some fraction of the finished drug product consists of unconjugated antibody. The remaining portion of the finished drug product contains conjugated antibody with a variable number of the cytotoxic small molecules conjugated at different sites on the antibody. Controlling the number and location of cytotoxic molecules conjugated to the antibody is being pursued as a means to create a more uniform product and as a way of being able to explore structure–function relationships by varying the site of attachment of the cyotoxin.

One strategy for controlling the site of attachment has been developed by researchers from Genentech, a member of the Roche Group. They describe precise site-specific conjugation of human IgG1 to MMAE by replacing Ala114 at the junction of the CH1 and the variable heavy-chain domain with cysteine to create an engineered antibody called a THIOMAB. This site was chosen because it does not participate in antigen binding or effector functions. According to Jagath Reddy Junutula, senior scientist at Genentech, the process for creating a THIOMAB differs only slightly from that of a conventional mAb. The THIOMAB is subjected to partial reduction to remove cysteine and glutathione adducts. The partial reduction also breaks interchain disulfide bonds, which must be reformed by a reoxidation step. After reoxidation, the engineered cysteine residues are available for conjugation.

Genentech researchers used this process to conjugate MMAE to a THIOMAB version of an antibody against MU16, a cell-surface protein expressed in ovarian cancer cells. The THIOMAB conjugate was shown to be homogenous and to contain a single drug molecule attached to each heavy chain, for a total of two MMAE molecules per ADC. The THIOMAB–MUC16 was found to have comparable efficacy to a conventionally produced ADC and to be better tolerated in two preclinical species (7). In a subsequent study, a different cytotoxin, DM1, was conjugated to a THIOMAB version of trastuzumab. Results were similar, with the THIOMAB T–DM1 displaying comparable efficacy and better tolerability in preclinical species than its conventionally produced counterpart (8).

According to Junutula, the reoxidation step is the only thing that distinguishes manufacture of a THIOMAB drug conjugate from that of a conventional ADC. "We can make up to grams scale without any difficulty. And the results are huge—you have a homogenously conjugated cytotoxic drug to the antibody," he says.

While the THIOMAB uses the substitution of one amino acid for another to control the site of conjugation, several groups are working toward incorporating nonnatural amino acids into the mAb for to control the site of conjugation and also to provide an expanded repertoire of functional groups that could be used for linker chemistry. The biopharmaceutical company Ambrx has developed expression systems in E. coli, yeast, and Chinese hamster ovary (CHO) cells that can be used for such substitutions and which can be scaled up to volumes required for commercial manufacturing. Ambrx's expression systems contain engineered transfer RNAs that will read through a stop codon called amber, as well as engineered tRNA synthetases that will aminoacylate the orthoganal tRNA with an Ambrx nonnatural amino acid. The expression system will insert a nonnatural amino acid whenever the amber stop codon is encountered (9).

Sutro Biopharma, a provider of protein-synthesis technology, also is developing a platform for introducing nonnatural amino acids, but in a cell-free translation system that is reported to be scalable to commercial production volumes (10). The system is based on an extract of E. coli, and because it is an open system, the tRNA charged with a nonnatural amino acid can be added directly to the reaction mix as a reagent.

Looking ahead

The future of ADCs in the biopharmaceutical market will ultimately depend on their clinical success. Companies and researchers are seeking to meet that challenge by optimizing the selection of all the components in the ADC—the antibody, linker, and cytotoxin—and successfully combining manufacturing techniques for both high- potency APIs and biologics. ADCs are sometimes described as armed antibodies, and their cytotoxic components as warheads. Whether ADCs will prove to be an effective weapon against cancer or other diseases has yet to be seen as more are tested in the clinic.

References

1. FDA, "Label for Adcetris, BLA 125338," FDA Approved Drug Products: Drugs@FDA, accessed Dec. 20, 2011.

2. V.S. Goldmacher and Y.V. Kovtun, Ther. Deliv. 2 (3), 397–416 (2011).

3. F. Dosio, P. Brusa and L. Cattel, Toxins 3, 848–883 (2011).

4. H.A. Burris, Expert Opin. Biol. Ther. 11 (6), 807–819 (2011).

5. A. Beck et al., Discov. Med. 10 (53), 329–359 (2010).

6. S.V. Govindan and DM Goldenberg, Scientific World Journal 10, 2070–2089 (2010).

7. Junutula et al., Nat. Biotechnol. 26 (8), 925-932 (2008).

8. Junutula et al., Clin. Canc. Res. 16, 4769–4778 (2010).

9. A. Ritter Pharm. Tech. 35 (6), 36–39 (2011).

10. Zawada et al., Biotech. Bioeng. 108 (7), 1570–1578 (2011).