Two-Step Algorithmic Selection of Interspecies Sequence-Mismatch-Based Housekeeping Genes for Precise Gene-Expression Assessment in PBMC-Humanized NSG Mice

High Resolution Image Download MS PowerPoint Slide Accurate normalization of gene expression is crucial in humanized mouse models, in which human cells are engrafted into immunodeficient mice. However, selecting the appropriate housekeeping genes remains a challenge. We aimed to identify and validate robust housekeeping genes for gene-expression analysis for human cells residing in different tissues of human peripheral blood mononuclear cell (hPBMC)-humanized NSG mice. We employed a multifaceted approach combining computational analysis and in vivo validation. The expression of 10 candidate genes ( EEF2, EIF2S3, GDI1, MYH9, SF3B1, SPCS3, ACTB, SDHA, TBP, and TFRC ) was evaluated for stability in human PBMCs using five algorithms (Geomean, Average SD, NormFinder, GeNorm, and BestKeeper). Genes exhibiting significant interspecies variation between human and mouse cell lines were identified. These data, combined with in vivo tissue-specific expression analysis in hPBMC-humanized NSG mice exhibiting high engraftment rates (hCD45+ cells >90% in the blood), supported the development of a two-step selection algorithm. This algorithm prioritizes genes with significant interspecies differences and high expression stability, recommending the following tissue-specific housekeeping genes: ACTB, heart; EEF2, liver and spleen; SDHA, lungs and kidneys; and TFRC, hPBMCs. Although SF3B1 demonstrated high stability in the blood, its higher human Ct values in vivo warrant caution. In the liver, both EEF2 and SDHA proved to be suitable, while EEF2 demonstrated superior performance in cytokine mRNA analysis. We highlight the importance of considering species- and tissue-specific expression patterns when selecting housekeeping genes for humanized mouse models. Our tissue-specific recommendations can contribute to more accurate and reliable gene-expression analyses, ultimately enhancing the utility of these models in human disease research.