Document resource
Background Liver metastases (LMS) are a major determinant of poor prognosis and therapeutic resistance in colorectal cancer (CRC) and pancreatic ductal adenocarcinoma (PDAC), including resistance to immunotherapy. We previously developed an in vivo gene therapy platform using lentiviral vectors (LVs) to selectively reprogram liver-resident macrophages (Kupffer cells, KCs) to secrete type I interferon (IFNα) within the metastatic niche (Kerzel et al., Cancer Cell, 2023). This approach significantly reduced liver metastatic burden in preclinical models. However, clinical translation is hindered by species-specific barriers, particularly LV-mediated transduction in human macrophages, which is extremely inefficient.Methods We exploited syngeneic murine models to investigate the immunophenotypic alterations induced upon in vivo macrophage engineering, assessing by flow cytometric, spatial and single cell trascriptomics the remodeling of the tumor microenvironment. To address human translation, we developed a novel hematochimeric mouse model harboring human KCs. Hematochimeric mice were obtained by sequential administration of LVs encoding key human cytokines and growth factors, followed by transplantation of human hematopoietic stem cells. We have also engineered the LVs to enable human macrophage-specific transgene expression while achieving high transduction efficiency.Results In murine models, in vivo macrophage engineering promoted immune activation, antigen presentation, and robust infiltration of cytotoxic CD8+ T cells. A small subset of mice exhibited resistance characterized by IL-10-driven immunosuppression, expansion of Eomes + CD4+ T cells (Tr1-like cells), and upregulation of CTLA-4. Combination of gene-based IFNa with CTLA-4 blockade synergistically expanded tumor-reactive T cells, achieving complete responses in most mice. To bridge the translational gap, novel hematochimeric mouse models faithfully recapitulated human KC identity, as confirmed by single cell-RNA sequencing, exhibiting transcriptional profiles akin to patient-derived KCs. Engineered LVs, packaging the HIV-2 accessory protein VPX, bypassed SAMHD1 restriction enabling substantial human macrophage transduction. Additionally, we developed a macrophage-specific MRC1 promoter to drive IFNα expression selectively in M2-like macrophages within LMS, minimizing systemic exposure and off-target toxicity.Conclusions We present a clinically relevant, macrophage-targeted gene therapy approach that remodels the LMS microenvironment ( figure 1). By combining vector engineering with humanized mouse models, we demonstrate the feasibility and therapeutic promise of in vivo engineering macrophages to convey IFNα to LMS. This strategy opens new avenues for treating CRC and PDAC LMS, with strong translational potential toward first-in-human applications.Abstract 1007 Figure 1Clinical translation of in vivo macrophage engineering for LMS