1. Why this study is important

Nearly all early-stage prostate cancer cases are dependent on androgen receptor (AR) signaling, making AR-targeted therapy a cornerstone of prostate cancer management. Unlike other cancer types where resistance to targeted treatment is primarily due to mutations in the drug target, prostate cancer can evade targeted treatment through lineage-switching. In this process, prostate cancer cells bypass their dependency on AR signaling by acquiring alternative phenotypes, with the neuroendocrine (NE) phenotype being the most observed. Although rare, AR^-/NE^- phenotypes have also been reported. Additionally, prostate cancer cells with different phenotypes can be observed within a single tumor, and their composition varies during cancer progression and treatment resistance. There is an urgent need for molecular landscaping studies to describe the different prostate cancer phenotypes and develop novel therapeutic targets.

2. How this research contributes to the field-directly

In our recently published paper, summarized in Figure 1, we performed a comprehensive and rigorous multi-omics study on prostate cancer samples from both humans and mice. We generated two prostate cancer single-cell datasets named HuPSA (human) and MoPSA (mouse) to describe cancer and stromal cell populations. Regarding cancer heterogeneity, we identified various AR+ prostate adenocarcinoma populations, neuroendocrine prostate cancer population, and, more importantly, two novel AR-/NE- populations named KRT7 and progenitor-like. We validated the existence of these novel populations using bulk RNAseq and immunohistochemistry. These identifications provided new diagnostic markers and laid the foundation for future mechanistic studies on these alternative phenotypes to develop new treatment strategies.

3. How this research contributes to the field-indirectly

While drafting the manuscript, we realized that the comprehensive datasets generated for our research are also valuable to others. Our research analyzed only a handful of gene expressions, but the remaining data could be a hidden gem for other researchers who are insightful and smart but may lack bioinformatics skills. To facilitate this, we launched the HuPSA&MoPSA online website, where researchers can freely browse, visualize, and download the three datasets we generated. We hope that other researchers can identify even more interesting findings using our tool.

Besides the current website, we have developed other bioinformatics websites with different focuses, as summarized in Tabel 1 below: