Targeting the cancer neoantigens p53 and KRAS with TCR mimic antibodies

Statement of Significance

Recent publications describe TCR mimic antibodies targeting the mutation-associated neoantigens, such as peptide/human leukocyte antigen class I complexes, which are derived from the intracellular antigens p53 or KRAS and are displayed on the cell surface. They have promising effects in the form of diabodies and might provide “off-the-shelf” protein immunotherapy for cancer treatment.

MAIN TEXT

Mutated antigens, like mutated tumor suppresser genes TP53, and oncogenes KRAS, HRAS, and NRAS, are attractive targets because they are specific for malignant cells. However, these proteins are intracellular antigens, which are generally difficult to target. The peptide containing the mutation (neopeptide) derived from the intracellular proteins can be processed and presented by the human leukocyte antigen (HLA) or major histocompatibility complex (MHC) class I on the cancer cell surface, which become neoantigens and thus make them feasible for immunotherapies [1]. While most mutations are not shared among many patients, the common hotspot mutations, such as TP53, RAS, and BRAF, called “public neoantigens,” are attractive because they are shared by many patients [1]. Antibodies targeting those peptide/HLA complexes are called T cell receptor (TCR) mimic (TCRm) antibodies. Recent studies by Hsiue et al. [2] and Douglass et al. [3] describe the development of TCRm antibodies that recognize the mutation-associated neoantigens derived from TP53 or KRAS. These agents have been converted to bispecific antibodies to target diverse cancers with these mutations.

The presence of the neoantigens on the tumor cell surface is the foundation to develop TCRm antibodies. Hsiue et al. and Douglass et al. have adopted a similar strategy to identify the neoantigens. They used computational modeling on the NetMHCpan 4.0 server to predict the binding of mutated p53 and RAS peptides to different HLA class I molecules. For instance, the p53^R175H was predicted to bind to HLA-A^*02:01 at 5177.6 nM, while the wild-type peptide binds at 7121.5 nM. To validate the prediction, they used a mass spectrometry (MS)–based method to quantify the complex on the cells. This method used an antibody targeting HLA molecules to do the immunoprecipitation and enrich the HLA-binding peptides. The peptides were separated from the HLA molecules and were then analyzed by multidimensional chromatography-based MS. Hsiue et al. showed that only the p53^R175H peptide was detected by MS analysis when the full length p53^R175H or p53^WT was co-expressed with HLA-A^*02:01 in COS7 cells. Furthermore, they detected several tumor cell lines and found there were 2.4, 1.3, and 1.5 copies of p53^R175H/HLA-A^*02:01 molecules per cell on the cell surfaces of KMS26, KLE, and TYK-nu. Douglass et al. showed peptides from KRAS^G12V or KRAS^Q61L could be presented by HLA-A^*03:01 or HLA-A^*01:01 on human cancer cell lines at single digit copies per cell. This direct evidence from both studies confirms peptide/HLAs are present at a very low level on human cancer cells. There are >20,000 HLA class I alleles in different human populations, and HLA-A^*02 is the most common allele among different populations [4]. If the targeted neoantigen has a common HLA allele like HLA-A^*02, it might potentially benefit more people than other alleles do.

The confirmed complexes were generated as monomers to be used in phage display to screen naive human antibody libraries. TCRm antibodies specific to those monomers were discovered in both studies (H2 for TP53^R175H and V2 for KRAS^G12V). The H2 has moderate affinity to its corresponding complex with a K_D of 86 nM, while V2 has a strong affinity to its complex with a K_D of 0.28 nM. Both antibodies have a higher affinity than that of typical for T cell receptors. Affinity plays an important role in the antibody efficacy. If the antibody has high affinity (IC₅₀ < 10 nM), it might eradicate the tumor. Otherwise, the tumor might relapse easily if the affinity is intermediate or low [5]. In terms of construction of T cell recruiting bispecific antibodies, a potent anti-CD3 antibody is important to activate endogenous T cells in a tumor antigen-dependent manner [6]. In both studies, the TCRm antibodies were fused with various anti-CD3 single chain Fv (scFv) antibodies, including UCHT1 (K_D: 1.8 nM), UCHT1v9 (K_D: 4.7 nM), L2K-07 (K_D: 110 nM), OKT3 (K_D: 2730 nM), and hXR32 (K_D: 21.2 nM), to form a bispecific antibody or single chain diabody (scDb) (Fig. 1A). Eventually, UCHT1 with the highest affinity can activate T cells at the lowest antigen concentration and was selected at the CD3 engaging arm of the scDb. Although the cancer cells expressed <10 copies of the neoantigen, the scDbs have been shown to activate T cells in vitro efficiently to lyse the tumor cells in both studies. In Hsiue’s study, the H2-scDb can activate human T cells to inhibit the xenograft growth of multiple myeloma (KMS26) in mice, which have very low copies of the p53^R175H/HLA complex. Even though the tumor inhibition in mice was not super impressive (from around 10⁸ to 10⁶ of bioluminescence signal), the inhibition specificity was validated. If the TP53 was knocked out by CRISPR in KMS26 or no T cells were injected together with the scDb, the tumor inhibition was abolished. Both studies provided the evidence of inhibition activity by the bispecific antibodies for the intracellular antigens, which were previously considered as a non-druggable target and have a very low presentation frequency on the cell surface [1].

Figure 1.

The mechanism of tumor cell killing by the scDb and T cells and the binding of the H2 antibody to the complex. (A) The tumor cells expressed the peptide/HLA complex as the neoantigen on the surface. The neoantigen can be targeted by TCRm antibody (scFv) fused with CD3-specific antibody (scFv) as a scDb. The CD3-bispecific antibody can activate T cells to release cytotoxic proteins to kill the tumor cells. (B) The crystal structure of the H2-Fab to p53^R175H/HLA-A^*02:01 (PDB ID 6W51) was visualized by ChimeraX (UCSF). The five residues on the peptide and six on the HLA-A^*02:01, which have main bonds, are shown.

A major challenge in immunotherapy with antibodies is off-target toxicity harming normal cells because of cross-reactivity to similar peptides. Both studies used different ways to assess the cross-reactivity to other peptides, including testing on different cells lines, BLAST to human peptidome, or peptide scanning mutagenesis, similar to the X-scanning mutagenesis that is used to access the binding specificity [7]. Briefly, each amino acid of the target peptide is systematically substituted with the other common 19 amino acids to generate libraries of variant peptides that have just one amino acid difference from the original peptide. The results showed that the TCRm antibodies and scDbs can specifically recognize mutated peptides and not the corresponding wild-type or related mutated peptides. Furthermore, Hsiue et al. has structural studies by x-ray crystallography which provides evidence that the H2-Fab binds only to mutated p53^R175H/HLA-A^*02:01 (Fig. 1B). All the CDRs of H2-Fab are involved in the binding. There are 79 contacts between H2-Fab and HLA-A^*02:01 of the complex and 36 contacts between H2-Fab and the C-terminus of the peptide. It is clear that H2-Fab binds both HLA and the peptide, however, they did not show any data about the cross-reactivity to different HLA alleles which might be beneficial for even more diverse populations. Overall, the binding of the H2-Fab to the mutated peptide is specific. However, they cannot rule out the possibility of off-tumor reactivity since the diversity of peptides in humans is enormous. Thorough and formal toxicity testing will be needed in nonhuman primate and eventually in human clinical trials.

TCR gene therapy or TCRm antibody-based therapeutics, including chimeric antigen receptor (CAR) T-cell therapy, have been developed to target several intracellular antigens like Wilms tumor 1, New York esophageal squamous cell carcinoma-1 (NY-ESO-1), or alpha-fetoprotein (AFP) [4]. Those cell therapies have been shown to be effective and useful for some patients, but they need patients’ autologous cells, and the production of engineered cells is complicated. This kind of therapy is highly individualized and costly. So far, TCRm antibody-based therapeutics targeting WT-1 and AFP have been advanced to clinical trial stages, while many other TCRm antibodies failed to advance further [4]. In both studies of Hsuie et al. and Douglass et al., they engineered TCRm antibodies into bispecific antibodies, which are protein-based therapies. This strategy has several advantages over cell-based therapies. Most importantly, the antibodies can be produced like antibody therapeutics by using established protocols, which is “off-the-shelf.” They can be manufactured at a large-scale and are less expensive. In addition, they both chose the format of CD3 BsAbs, which is well developed in the field with many different forms [8]. In both studies, they found the scDb format had the best activity. Considering that tumor cells have very low expression of the neoantigens, those scDbs have an exceptional activity to react to the target antigen. The reasons maybe the unique cage-like structure of the binding mode to the pHLA, which resulted in an optimized immunological synapse, or the high affinity of the anti-CD3 scFv (UCHT1) used. Although the antibodies in the scDb format looks promising, it will be interesting to explore the efficacy of those antibodies in the format of antibody-drug conjugates, immunotoxins, and CAR-T.

Although studies of Hsuie et al. and Douglass et al. showed promising activity for both scDbs against cancer, there are several issues to consider before further advancement into clinical trials. First is the limitation of HLA type for both antibodies. For H2-scDb, it targets p53^R175H/HLA-A^*02:01, and V2-scDb targets KRAS^G12V/HLA-A^*03:01. Thus, these antibodies can only be utilized for patients having a specific subset of HLA class I. Development of TCRm antibodies with less restriction to a particular HLA type will certainly be beneficial for more patients. The second issue is the short half-life of scDbs since scDbs are small molecules, which are rapidly cleared from the blood in mice and humans. Protein engineering of the scDbs, like addition of immunoglobulin G fragment crystallizable domain [9], might be helpful to increase the plasma half-life and may further increase the efficacy in vivo. Since scDbs have CD3 that activates T cells to lyse the tumor cells, another issue is that T cells might be dysfunctional under the tumor microenvironment as a result of the expression of inhibitory immune checkpoints like program death protein 1 (PD-1) and cytotoxic T lymphocyte-associated protein 4 (CTLA-4) [10]. Thus, the combination of the scDbs with blockers of immune checkpoints might be helpful to facilitate the T cell-killing activity.

In summary, both studies by Hsiue et al. and Douglass et al. developed TCRm antibodies in the scDb format targeting specific neoantigens to treat cancer. These protein-based immunotherapies can treat patients from off-the-shelf production, which can be complementary to current cell-based therapies. Nevertheless, further preclinical and clinical studies will be needed to validate this strategy for treating cancer. Furthermore, development of TCRm antibodies targeting the complex derived from intracellular proteins can expand the antigen repertoire and can potentially make many “undruggable” targets, like TP53 and KRAS, accessible to immunotherapy.

FUNDING

This work is supported by the Intramural Research Program of National Institutes of Health, NCI CCR Antibody Engineering Program [ZIC BC 011891].

DATA AVAILABILITY

The data included in this News and views are available upon request from the corresponding author.

CONFLICT OF INTEREST STATEMENT

M.H. is the Editor-in-Chief of the journal and is blinded from reviewing or making decisions on the manuscript. The authors do not claim other conflicts of interest to this work.