Antibody-based therapeutics are valuable tools for the treatment of many human diseases. For example, antibody-based immunotherapies targeting the leukemia antigens CD19 and CD20 have been remarkably successful against B cell malignancies, such as acute lymphoblastic leukemia (ALL) or B cell lymphomas. To extend such benefits to patients with myeloid malignancies, efforts have been made to develop therapies against the CD33 antigen, a surface protein expressed by myeloid cells and myeloid cell-derived neoplasms, including acute myeloid leukemia (AML). With gemtuzumab ozogamicin (GO), an antibody-drug conjugate targeting CD33, there is now one therapeutic approved for use in patients with AML. However, while some patients benefit from GO, many do not. Many other CD33-directed drugs have been tested but have failed in the clinic; in most cases, failures were due to insufficient clinical efficacy. Investigators in the Fred Hutch Clinical Research Division, led by former Walter Laboratory research associate Dr. Colin Godwin, explored strategies for improving CD33-directed immunotherapies. Their work, recently published in Leukemia, showed that targeting antibody-based therapies to regions of the CD33 protein that are closer to the cell membrane yielded superior efficacy against cellular targets.
Antibodies are important immune effector molecules with the capacity to tightly bind specific regions of target antigens, called epitopes. They can also function as adaptor molecules to engage other components of the immune system. Antibody-based therapeutics take advantage of the valuable properties of specificity, stability, and adaptability of these molecules, and serve as a platform for a variety of therapeutic modalities. For instance, antibodies that bind and alter the functionality of surface molecules, or those that flag target cells for immune destruction, form the basis of many therapeutics in their native forms. Further, protein engineering technology utilizes antibody domains as building blocks, linking the binding specificities of antibodies to other functional modalities. Such antibody fusion protein technologies include antibody-drug conjugates for delivering drugs to specific cellular targets (with GO being the most notable example of this type of antibody therapeutic for AML), bispecific antibodies (BsAb) for bridging interactions between two molecules (oftentimes between antigen targets and the T cell receptor), and chimeric antigen receptors (CARs) for directing T cell responses against specific antigens. Interestingly, previous work has shown that antibody-based therapeutics against several other surface proteins, including CD20, show superior efficacy when they bind membrane proximal epitopes of their target antigens. Noting that existing anti-CD33 therapeutics, including GO, T cell engaging BsAbs, and CARs, almost exclusively target epitopes located in the membrane-distal ‘V-set’ domain of the CD33 protein, the Walter Lab hypothesized that re-targeting these therapeutics to the membrane proximal ‘C2-set’ domain may be a strategy for enhancing their function against leukemic targets. “It seemed very plausible that this mechanistic principle, which has been shown for several other therapeutic targets, would hold true for CD33 as well”, said Colin Godwin, “but it has never been tested.”
To experimentally test the effect of the distance of the target epitope from the membrane, the Walter Lab generated a series of mutant CD33 proteins, with different distances between the antibody-targeted V-set domain and the cell membrane. Acknowledging its reliance on artificial antigens that are not naturally occurring, this system provides a level playing field that avoids pitfalls associated with comparing efficacies of antibodies targeting different epitopes with different affinities. The researchers expressed these mutant proteins in CD33-deleted leukemia cells, and then targeted them with V-set-directed BsAb and CAR therapies, as well as GO. Interestingly, the cytotoxic effects of GO were similar between the mutants, indicating similar levels of antibody-drug conjugate internalization between the variants. However, those cells expressing CD33 variants with shorter distances between the target epitope and the cell membrane and were killed more robustly by BsAb- or CAR-directed T cells. These data support the hypothesis that re-targeting antibody-based therapies against membrane-proximal epitopes could help improve the efficacy of CD33-directed immunotherapies. “This finding is relevant as it suggests that we will need a new generation of antibodies for CD33-targeted immunotherapy; almost all existing CD33 antibodies recognize the V-set domain, suggesting that the immune-dominant epitopes are located in the membrane-distal domain”, said Colin Godwin.
Next, the team set out to test their hypothesis in a more physiologically relevant setting by testing antibodies that bind to the membrane-proximal C2-set domain of the native CD33 protein. Since no well-characterized C2-set-targeted antibodies had yet been developed, the group raised their own C2-set-targeting monoclonal antibody by immunizing mice against a CD33 mutant lacking the V-set domain. Because, at least at the mRNA level, a splice variant of CD33 exists that lacks the V-set domain, the group was particularly interested in antibodies that recognized the C2-set domain regardless of whether the V-set domain was present or not. Such antibodies were nicknamed CD33PAN antibodies as they, theoretically, recognize all known isoforms of CD33. One of these CD33PAN antibodies, called ‘1H7’, was internalized with similar kinetics to the V-set-targeted antibody used in GO, and caused similar changes in cell surface density of CD33, indicating its suitability for delivery of payload therapies. To test the ability of this targeting specificity to direct cytotoxic T cell responses, the group constructed a BsAb utilizing the 1H7 targeting domains. In a cytotoxicity assay, the 1H7 BsAb was indeed capable of directing CD33-dependent killing of target cells. Finally, because murine antibody-based therapeutics could potentially be recognized and rejected by patients’ immune systems, the group raised a second series of C2-set-specific CD33PAN antibodies in humanized mice, engineered to express human antibody sequences. Similar to its murine counterpart, a BsAb constructed using the binding domains of the humanized antibody, called ‘1E6’, was highly potent against CD33-expressing human leukemia cells.
“These data suggest that the use of optimized therapeutics that are based on antibodies against the membrane-proximal C2-set domain could form one strategy to increase the efficacy of CD33-directed immunotherapy,” the authors write. As noted above, a second conceptual reason to pursue C2-set domain-directed CD33 antibodies as therapeutics is the existence of CD33 variants that lack the V-set domain. Thus, targeting the C2-set domain would allow universal recognition of all known naturally occurring CD33 variants. For both of these reasons, C2-set-targeting antibody-based therapeutics offer hope for increased benefit in patients with AML and other myeloid diseases. “Together, these data provide the rationale for the further preclinical development of agents that are based on binding sequences from [C2-set-specific] CD33PAN antibodies as we have initiated.”
This work was funded by the Leukemia & Lymphoma Society, the National Institutes of Health, the National Cancer Institute, the National Heart, Lung, and Blood Institute, the American Society of Clinical Oncology, the Conquer Cancer Foundation, Alex’s Lemonade Stand, and the M.J. Murdock Charitable Trust.
UW/Fred Hutch Cancer Consortium members Hans-Peter Kiem, Cameron Turtle, and Roland Walter contributed to this work.
Godwin CD, Laszlo GS, Fiorenza S, Garling EE, Phi TD, Bates OM, Correnti CE, Hoffstrom BG, Lunn MC, Humbert O, Kiem HP, Turtle CJ, Walter RB. Targeting the membrane-proximal C2-set domain of CD33 for improved CD33-directed immunotherapy. Leukemia. 2021 Feb 15. doi: 10.1038/s41375-021-01160-1. Epub ahead of print. PMID: 33589747.