Overcoming Immune Escape Via Targeting Akt-Wnt/β-Catenin By Artificial-Intelligence-Powered Peptide-Producing mRNAs

Although chemoradiotherapy can be effective, chemoresistant leukemia stem cells (LSCs) within the minimal residual disease subpopulation often lead to therapy-resistant relapse. Frequent and abnormal activation of the Wnt/β-catenin and PI3K/Akt pathways occurs in a wide array of cancers, including T-acute lymphoblastic leukemia (T-ALL). These pathways are crucial for LSCs' leukemia-initiating capabilities and their ability to evade immune surveillance. Notably, phosphorylation at the C-terminus of β-catenin by Akt and PKA, including Ser552 and Ser675 has been shown to enhance the transcriptional activity of β-catenin, as well as its translocation to the nucleus, thereby activating the expression of genes that aid in tumor growth, metastasis and immune escape. The intricate relationship between the Akt-Wnt/β-catenin pathway and the mechanisms of immune escape and cancer recurrence underscores the urgent need for targeted therapeutic interventions.

Our lab previously discovered that low-dose doxorubicin (DXR) can selectively inhibit the Akt-mediated activation of β-catenin. This treatment leads to the inhibition of leukemia-initiating capacity and the downregulation of multiple immune-checkpoint genes. Additionally, we found that in patients with relapsed or refractory AML, the presence of post-chemotherapy pS552-β-catenin⁺ LSCs and their leukemia-initiating activity can be reduced with low-dose DXR treatment. However, cardiovascular toxicity associated with DXR poses a substantial restraint for its application. Additionally, though both small molecules and monoclonal antibodies have been used for treating a broad range of diseases, their use as drugs is limited by many factors. Small molecules, for instance, often pose a risk of high toxicity due to off-target binding. On the other hand, the development of monoclonal antibodies has traditionally been restricted to extracellular targets, necessitating lengthy screening processes. Despite the critical role of this pathway in leukemogenesis and LSC, there are currently no FDA-approved inhibitors targeting the Wnt/β-catenin pathway, highlighting a significant gap in our current leukemia treatment arsenal and underscoring the potential impact of developing such therapies.

To address this gap, we proposed Artificial-intelligence(AI)-powered Peptide-Producing mRNAs (APPR) to target Akt and disrupt its interaction with β-catenin. We used two pipelines to generate Akt inhibitors: 1. A β-catenin-derived peptide containing Ser552; 2. Using RFdiffusion to generate binders to the Akt kinase domain. Candidates from both pipelines were evaluated using AlphaFold and RoseTTAFold, and subsequently screened for potential immunogenicity (T-cell epitope) and toxicity in silico using multiple databases. The top candidates were then tested in vitro using reporter cell lines expressing luciferase or dGFP driven by the β-catenin enhancer TOPFlash. Afterwards, optimized cap, 5’UTR, 3’UTR, and N1-Methylpseudouridine were incorporated into the mRNA for enhanced translational efficiency and minimized immunogenicity. Lastly, AI was used to optimize the coding sequences of the selected candidates, based on codon adaptation index, minimum free energy, reducing uracil usage, and maintaining openness of 5’ Kozak region. This aimed to reduce immunogenicity and improve mRNA translation efficiency.

Through testing, the peptides significantly reduced TOPFLASH-driven luciferase and dGFP expression. Western blotting showed a significant reduction of pS552-β-catenin, pS675-β-catenin and qRT-PCR revealed decreased expression of Wnt-responsive target genes. These results provide compelling evidence of the significant inhibition of Wnt/β-catenin signaling by APPR in vitro. Furthermore, using Scl^CreER; Pten^flox/flox; Ctnnb1^ΔEx3/+ mouse model, which develops T-ALL, we found that APPR significantly reduced the LSCs and partially restored hematopoietic stem and progenitor cells, while no significant damage was observed in animal tissues. This further demonstrates the efficacy of APPR in vivo. Overall, this innovative approach aims to modulate the signaling pathway, leading to the attenuation of β-catenin's transcriptional activity. By targeting Wnt/β-catenin signaling, this approach holds the potential to influence the function of cancer stem cells and improve therapeutic responses in cancer treatment.