Bispecific antibody for treating hemophilia A

Thick with potential

In 2000, Hattori learned something about a step in the clotting cascade that sparked a bold idea. A particular factor attaches to an enzyme on the platelet surface and orients it relative to its substrate with exquisite precision, thus facilitating catalysis. If F-VIII similarly repositions F-IXa and F-X, thought Hattori, maybe he could devise a F-VIII substitute.

Hattori knew coagulation and he knew antibodies, Y-shaped proteins with two identical arms, each of which normally binds the same target molecule. The distance between those binding sites is close to the distance between the F-IXa and F-X binding sites of F-VIII. Perhaps, mused Hattori, he could construct a rulebreaker antibody that clasps F-IXa with one arm and F-X with the other. Rare versions of such a bispecific antibody might arrange them exactly such that F-IXa could activate F-X.

Although audacious, his proposition held hope. Antibody-based therapeutics, which almost all employ the IgG subtype, perform well when injected under the skin, the molecules are hardy, and the immune system tends to ignore them. Furthermore, because the structures of antibodies and F-VIII are unrelated, F-VIII inhibitors shouldn’t meddle with the theoretical bispecific molecule nor should an unexpected immune response to it foil F-VIII. Hattori’s fantasy drug would therefore overcome the main problems with hemophilia A treatments. He launched the project and oversaw it, while Kitazawa and Igawa spearheaded the biology and engineering sides, respectively.

A pilot study spawned tantalizing results and revealed challenges. The most promising agent enhanced coagulation in the test tube, but worsened bleeding in animals. The investigators extracted valuable lessons from this disappointment and persevered.

Squeezing blood from a turnip

They scaled up, aiming to build a large, diverse set of antibodies whose population might include the one they wanted. After immunizing animals with either human F-IXa or F-X, the team used standard methods to grow cultured cells, each of which secreted a single—monoclonal—antibody against one of the input proteins. This procedure yielded 200 cell lines or transiently infected cells that made monoclonal antibodies against F-IXa and 200 that made monoclonal antibodies against F-X. The investigators paired antibody genes from each set to produce 40,000 types of cells that contained random combinations and had the capacity to generate not only antibodies that recognized either F-IXa or F-X, but also the sought-after bispecific IgG antibodies in which one arm bound F-IXa and the other arm bound F-X.

The team checked whether any of the 40,000 bispecific antibodies could substitute for F-VIII by incubating them with F-IXa and F-X along with a substrate of F-Xa that turns yellow upon cleavage. Antibodies that simulated F-VIII provoked a measurable color change, and such F-VIII imitators were examined further. Consistent with expectation, one arm gripped F-IXa, whereas the other gripped F-X. Dismembered molecules that bore only one arm or the other did not replace F-VIII; both arms needed to be present in a single antibody to work. These observations supported the notion that the antibody operates by uniting F-IXa and F-X. Successful prospects shortened clotting time in a test system based on human F-VIII-deficient plasma. Crucially, this result held up in the presence of inhibitors.

No matter how auspicious any such antibodies seemed, though, they would never get to the clinic unless the team could facilitate their large-scale manufacture. To achieve this goal, the researchers had to solve a formidable problem: Each cell fabricated numerous antibodies in addition to the one of interest.

In natural antibodies, two identical heavy chains and two identical light chains form the hallmark Y shape. Cells with the capacity to make a bispecific antibody, in contrast, carry genes for two different heavy chains and two different light chains. Because the heavy and light chains reassort randomly, ten combinations are possible, only one of which creates the desired antibody (Figure 2, top).

The number of options would shrink to three if a single light chain could function with both heavy chains: This situation would produce one entity with two F-IXa-binding heavy chains, a second with two F-X-binding heavy chains, and a third—the heterodimer they wanted—with one of each (Figure 2, bottom left). The team patched together portions of light-chain genes in various combinations and screened for one that worked well with both heavy chains to support F-VIII activity.

When one is greater than ten
To forge a therapeutic agent that can stand in for F-VIII, Hattori, Igawa and Kitazawa not only sifted through tens of thousands of candidate molecules, but also engineered techniques to enrich for the bispecific antibody that they sought. Two heavy chains and two light chains combine in 10 possible ways (top), only one of which produces the desired bispecific antibody. A single light chain that can collaborate with both heavy chains reduces the number of distinct entities to three (bottom left). Assuming that the two heavy chains are present in equal amounts that associate randomly in pairs, the population will contain twice the number of antibodies with mixed (blue and purple) heavy chains (heterodimers) as antibodies with either of the same (blue-blue or purple-purple) heavy chains (homodimers). Consequently, the antibody of interest composes 50% of the mixture. Adding a positive (purple) or negative (blue) charge to a particular spot encourages association of the heterodimer (bottom right).
Illustration: Heather B. McDonald

With this innovation, the bispecific antibody would still compose only 50% of the mixture (Figure 2, bottom left). To enrich the yield, the researchers deployed a trick early during development to encourage the two different heavy chains to associate with each other and discourage the two like ones from pairing up. At a spot where the chains interact, the team created a positive charge on one and a negative charge on the other. Consequently, identical chains repelled each other and opposite chains attracted (Figure 2, bottom right). This advance boosted output of the asymmetric molecule to greater than 90%.

Still, isolating the bispecific antibody from the small amount of its closely related mono-specific brethren presented a challenge. Making one heavy chain significantly more positive and the other significantly more negative resolved this issue. Antibodies that affix only to F-IXa or F-X became highly charged—one positively, the other negatively—whereas their bispecific sibling was neutral. Standard industrial chromatography that separates molecules based on charge could then purify the bispecific antibody from its unwanted relatives.

Solidifying progress

Together, these three elements paved the way toward commercial viability, but the researchers also had to optimize for eventual medicinal use. They had already rendered the lead antibody immunologically amenable to humans. Now they undertook repeated cycles of molecular engineering and evaluation to fortify effectiveness, half-life, bioavailability, and other advantageous properties.

Early candidates that had gained F-VIII activity disappeared quickly from mice, which didn’t bode well for durability in people. So the team made them stick less to molecules that steer them toward destruction. As one characteristic improved, another often declined. For instance, as the antibodies became more soluble, which was required to attain the high concentrations needed for subcutaneous administration, they lost F-VIII activity—and vice versa. Hence, each amino-acid tweak required multi-pronged scrutiny. As the researchers iteratively modified the antibodies, they gradually homed in on variants that performed better on all fronts.

To check whether the test-tube F-VIII effect would deliver in a setting that approached the human body, they mimicked hemophilia A in monkeys by neutralizing F-VIII. Then they administered their frontrunner antibody and induced injury. Relative to controls, the bispecific antibody protected the animals from internal bleeding and bruising. This result suggested that the therapy would stave off bleeds in people.

After crafting and analyzing 2400 bispecific antibodies, Hattori, Igawa, and Kitazawa identified the winner in 2010; it was later named emicizumab.

They subsequently probed the likelihood that emicizumab would adhere to physiologically acceptable behavior. For instance, they showed that its F-VIII mimetic capacity depended on exposure to the phospholipid molecules that appear on activated platelets. This observation provided confidence that emicizumab would emulate F-VIII only in appropriate locations rather than trigger clotting in the bloodstream.

Unconventional success

After initial studies established acceptable safety in humans, the drug’s efficacy was assessed. In 2016, Midori Shima at Nara Medical University and his colleagues at other Japanese institutions published results from a trial on 18 people with severe hemophilia A. Weekly subcutaneous administration of emicizumab deflated annualized bleeding rates relative to what they were in the six months before this treatment, even in individuals with inhibitors.

The next year, Shima reported data from a multi-center, international clinical investigation on 109 people with inhibitors. The therapy slashed annualized bleeding rates from 23.3 with treatment to 2.9 without, an 87% drop. Sixty-three percent of the participants had no bleeds in at least 24 weeks compared with 6% in the group that was treated only with standard therapies, not with emicuzumab.

These results, along with similar ones from a study on children under 12 years old, led to expedited approval by the U.S. Food and Drug Administration in 2017, which was extended to people without inhibitors in 2018. More than 100 countries have followed suit. The drug’s half-life is about 29 days, which has allowed many individuals to deploy a dosing frequency of once every four weeks. The real-world impact has been profound, especially on children, who can participate more fully in normal play now that mobility impairment and pain do not bedevil their joints, and the risk of a life-threatening bleed from a jungle-gym tumble has plunged. Emicizumab’s safety profile has remained favorable, with the most frequent side effect being injection-site reactions. In phase 3 clinical studies, no serious events have occurred in patients who have received emicizumab alone or in combination with F-VIII or another factor, but dangerous clots that obstruct a blood vessel have been reported in several patients when a medicine called activated prothrombin complex concentrate was repeatedly given in combination with emicizumab.

Hattori, Igawa, and Kitazawa have touched not only the 30,000 individuals across the globe who have received emicizumab (marketed as Hemlibra®).  They have also expanded the possibilities of therapeutic antibody design more widely, opening avenues toward bispecific antibodies that target diverse pathological processes. Since emicizumab arrived, 14 additional bispecifics have been approved, 11 of which aim to kill cancers. Most target T cells along with a specific molecule on the surface of lymphoma, myeloma, or other tumor cells. To date, emicizumab is the only bispecific that bypasses a molecular deficiency and restores its function. Its uniqueness underscores the scientists’ ingenuity, pluck, and dedication.

by Evelyn Strauss

Selected Publications – Kunihiro Hattori, Takehisa Kitazawa, Tomoyuki Igawa

Kitazawa T, Igawa T, Sampei Z, Muto A, Kojima T, Soeda T, Yoshihashi K, Okuyama-Nishida Y, Saito H, Tsunoda H, Suzuki T, Adachi H, Miyazaki T, Ishii S, Kamata-Sakurai M, Iida T, Harada A, Esaki K, Funaki M, Moriyama C, Tanaka E, Kikuchi Y, Wakabayashi T, Wada M, Goto M, Toyoda T, Ueyama A, Suzuki S, Haraya K, Tachibana T, Kawabe Y, Shima M, Yoshioka A, and Hattori K. (2012). A bispecific antibody to factors IXa and X restores factor VIII hemostatic activity in a hemophilia A model. Nat. Med. 18, 1570-1574.

Sampei Z, Igawa T, Soeda T, Okuyama-Nishida Y, Moriyama C, Wakabayashi T, Tanaka E, Muto A, Kojima T, Kitazawa T, Yoshihashi K, Harada A, Funaki M, Haraya K, Tachibana T, Suzuki S, Esaki K, Nabuchi Y, and Hattori K. (2013). Identification and multidimensional optimization of an asymmetric bispecific IgG antibody mimicking the function of factor VIII cofactor activity. PLoS One, 8:e57479. doi: 10.1371/journal.pone.0057479.

Muto A, Yoshihashi K, Takeda M, Kitazawa T, Soeda T, Igawa T, Sakamoto Y, Haraya K, Kawabe Y, Shima M, Yoshioka A, and Hattori K. (2014). Anti‐factor IXa/X bispecific antibody (ACE910): hemostatic potency against ongoing bleeds in a hemophilia A model and the possibility of routine supplementation. J. Thromb. Haemost. 12, 206-213.

Muto A, Yoshihashi K, Takeda M, Kitazawa T, Soeda T, Igawa T, Sampei Z, Kuramochi T, Sakamoto A, Haraya K, Adachi K, Kawabe Y, Nogami K, Shima M, and Hattori K. (2014). Anti-factor IXa/X bispecific antibody ACE910 prevents joint bleeds in a long-term primate model of acquired hemophilia A. Blood. 124, 3165-3171.

Kitazawa T, Esaki K, Tachibana T, Ishii S, Soeda T, Muto A, Kawabe Y, Igawa T, Tsunoda H, Nogami K, Shima M, and Hattori K. (2017). Factor VIIIa-mimetic cofactor activity of a bispecific antibody to factors IX/IXa and X/Xa, emicizumab, depends on its ability to bridge the antigens. Thromb. Haemost. 117, 1348-1357.

Shima M, Hanabusa H, Taki M, Matsushita T, Sato T, Fukutake K, Fukazawa N, Yoneyama K, Yoshida H, and Nogami K. (2016). Factor VIII-mimetic function of humanized bispecific antibody in hemophilia A. N. Engl. J. Med. 374, 2044-2053.