In the pharmaceuticals industry, antibodies are big business. According to a 2014 analysis by Nature Biotechnology, worldwide sales of monoclonal antibody therapeutics “reached $63 billion” in 2013 [1]. Six of the top 10 biopharmaceuticals were monoclonals, the report says—just one of which, the tumor necrosis factor-targeting anti-inflammatory Humira (adalimumab) rang up sales of $11 billion.
Naturally, pharmaceuticals companies are keen to build on that success, and their development pipelines are full of new antibodies. But many of these aren’t your garden-variety IgGs. Modern antibody therapeutics are more sophisticated than ever.
ADCs and bispecifics
Monoclonal antibodies, says Hans Peter Gerber, vice president of bioconjugates discovery and development at Pfizer’s Oncology Research Unit in Pearl River, N.Y., are among the most successful therapeutics in the oncology arena. “And they have one big advantage in the clinic: They are relatively nontoxic.”
Generally speaking, monoclonal therapeutics function by either blocking ligand-receptor interactions or agonizing or antagonizing cellular receptors.
The problem, Gerber says, is that many useful cellular receptors—those that define a rare cell population, for instance—yield no useful biological impact when antibodies bind them. Such proteins can define the cells researchers want to kill, but not simultaneously provide a built-in way to do so.
Two rapidly growing therapeutic classes address this problem.
Antibody-drug conjugates (ADCs) tether a toxic “payload” to a tumor-targeting antibody. “The premise is that an antibody can give you tremendous targeting specificity, but optimal killing requires additional firepower,” explains Richard Gregory, chief scientific officer at ImmunoGen, a company that develops ADCs. In theory, by coupling a highly toxic compound to an antibody, the payload is rendered inert until it is released in the cell.
At least 30 ADCs are in clinical development [2], and two have been approved by the U.S. Food and Drug Administration for marketing in the United States. Adcetris® (brentuximab vedotin) from Seattle Genetics, linking a CD30-targeting antibody to an auristatin-based “microtubule disrupting agent,” is approved for treatment of Hodgkin lymphoma. Genentech’s Kadcyla (ado-trastuzumab emtansine, which uses ImmunoGen’s linker and toxin) amps up the HER2 receptor-targeting anticancer therapeutic Herceptin with DM1, a microtubule-targeting cytotoxin.
An alternative antibody design is the so-called “bispecific antibody.” Naturally occurring antibodies are monospecific—they contain two arms, each of which recognizes the same antigen. Bispecific antibodies are artificial constructs designed to recognize two antigens simultaneously—sometimes on different cells. Some two dozen bispecifics are in clinical development [3]. As of December 2014, one has been approved in the United States: Amgen’s Blincyto (blinatumomab) links CD19-positive B cells and CD3-positive T cells to treat a rare form of B-cell leukemia.
Design decisions
ADCs, says Gerber, are “complex therapeutics.” Their development requires coordinating at least four variables: target biology, the antibody, the toxin and the linker that connects them.
To make a drug that is both effective and relatively nontoxic, drug designers must answer two key questions, Gerber says: How restricted is the antigen expression to the tumor, and how long does the antibody need to remain in circulation? If, for instance, the antibody’s target antigen is present on both tumor and normal cells, but the tumor cells internalize the antigen more rapidly leading to target disappearance, side effects may be minimized by limiting the drug’s half-life. On the other hand, if the antibody target is tightly restricted to tumor cells and remains expressed during treatment, drug designers can select more potent payloads and longer exposure times.
Whatever the payload, says Gregory, the highest affinity antibody isn’t always the best option. If the ADC is targeting a solid tumor that expresses high levels of antigen, the antibody may get sopped up at the tumor periphery if its affinity is too high, a phenomenon called the “binding-site barrier.” In such cases, Gregory explains, a lower-affinity antibody might be more effective, as the ADC can penetrate the tumor more deeply.
Linkers typically are of four types, says Gerber. Three are chemically sensitive to either low pH, reducing conditions or specific proteases. A fourth class of linkers is chemically stable, releasing their payloads only when the antibody itself is degraded.
The proper linker enables developers to control, among other things, how widespread the ADC’s impact will be, Gregory says. Kadcyla, for instance, kills only the cells to which it binds—with its non-cleavable linker, it releases the payload, linker and the antibody lysine residue to which they were bound as a single unit, which cannot pass from cell to cell. Sometimes, though—for instance, in certain solid tumors—it would be better for the toxin to break away from the lysine residue and diffuse from cancer cell to cancer cell, producing a bystander effect.
The final piece of the ADC puzzle is the toxin itself. The payload agents in Kadcyla and Adcetris both target microtubules, meaning they should specifically affect dividing cells. But other cellular functions also may be targeted. ImmunoGen’s IMGN779, for instance, couples a CD33-targeting antibody to DGN462, a DNA alkylating agent.
Typically, drug developers conjugate drug moieties either at cysteine or lysine residues or at a chemically receptive location engineered into the molecule. This can cause molecular heterogeneity, as toxins can conjugate to the antibody at all or only some possible locations. But rarely do ADCs have more than a handful of drug molecules per antibody, says Timothy Lowinger, chief scientific officer at Mersana Therapeutics, because as drug loading increases, pharmacokinetics, water solubility and efficacy often decrease. Seattle Genetics’ Adcetris, for instance, contains four auristatin toxins per antibody linked via cysteine residues [4]. The company’s SGN-CD33A, now in phase 1 trials, links two DNA-targeting pyrrolobenzodiazepine dimers at each of two “site-specific engineered cysteines.”
For those who want a more potent payload, there’s Mersana’s Fleximer®, a “highly biocompatible and biodegradable polymer, which is water-soluble and polyvalent,” Lowinger says. According to Lowinger, Fleximer allows the company to attach 15 or 20 drug moieties to an antibody or other delivery reagent without significantly impacting tolerability. That offers two benefits, he says. First, it allows Mersana and its partners to better target low-copy-number antigens, as more drug is delivered “per internalization event.” Similarly, it frees the company to explore less cytotoxic payloads than typical ADCs, such as vinca alkaloids, as they can be delivered at higher concentrations. “We see that by putting more copies [of vinca alkaloids] per antibody, we can get very high efficacy in a variety of in vivo models,” he says.
At Pfizer, researchers make use of patient-derived xenograft (PDX) models to test their ADCs, says Gerber [5]. A PDX, he explains, “is a human tumor [in a mouse] that has never seen any plastic or artificial environment; they only are propagated in vivo.” These tumors look and respond like human tumors and are “as close to a human tumor [as] you can get.” Using them, his team can optimize its ADC designs, as it’s impossible to accurately predict a priori which linker, for instance, will work best for a given target and indication.
“We think that patient-derived xenograft models are very predictive of what we’ll see in the clinic,” he says.
Antibody purification
Typically, says Jonathan Royce, senior product manager for antibody affinity media at GE Healthcare Life Sciences, drug developers express antibodies in eukaryotic cells and purify them on a protein A column. This is followed by cation-exchange chromatography to remove antibody aggregates and anion exchange to remove endotoxins, residual host cell proteins and other contaminants.
Why the initial protein A step? Because it is generic, Royce says. Most antibody molecules are bound by protein A, which targets the antibody Fc region. “The amount of process-development work [drug developers] need to do is minimized.”
Some antibody therapeutic designs, though, don’t include the Fc region, including single-chain variable fragment (scFv) molecules. (These designs offer certain advantages over monoclonals, such as a smaller size, which may enable them to penetrate tissues that mAbs cannot.) Historically, Royce says, researchers had to develop one-off purification procedures for each molecule, as there was no reagent equivalent to protein A that they could use instead. Now, there is.
As it turns out, Royce says, the majority of these new antibody designs include the kappa light chain in their construction. And there is an affinity reagent that is specific for kappa: protein L. GE Healthcare Life Sciences’s Capto L purification medium is based on protein L. (For lambda light-chain antibodies, the company offers LambdaFabSelect, “a camelid [antibody]-based affinity ligand with affinity for lambda light chain.”)
According to Royce, research-scale protein L reagents already exist in the market, but they have been relatively expensive for industrial applications. Capto L, he says, is comparable in price to the company’s protein A resin, making it an affordable option for developers.
How such antibody fragment-based therapeutics will fare in the marketplace is anybody’s guess. But with blockbuster antibodies already available and dozens more in development, it’s a sure bet that whatever happens, antibody therapeutics aren’t going anywhere for some time to come.
References
[1] Walsh, G, “Biopharmaceutical benchmarks 2014,” Nat Biotechnol, 32:992-1000, 2014. [PubMed ID: 25299917]
[2] Mullard, A, “Maturing antibody-drug conjugate pipeline hits 30,” Nat Rev Drug Disc, 12:329-32, 2013. [PubMed ID: 23629491]
[3] Kontermann, RE, Brinkmann, U, “Bispecific antibodies,” Drug Disc Today, March 2015. [doi: 10.1016/j.drudis.2015.02.008] [PubMed ID: 25728220]
[4] Senter, PD, Sievers, EL, “The discovery and development of brentuximab vedotin for use in relapsed Hodgkin lymphoma and systemic anaplastic large cell lymphoma,” Nat Biotechnol, 30:631-7, 2012. [PubMed ID: 22781692]
[5] Rosfjord, E, et al., “Advances in patient-derived tumor xenografts: From target identification to predicting clinical response rates in oncology,” Biochem Pharmacol, 91:135-43, 2014. [PubMed ID: 24950467]
Image: Dreamstime