Endometrial cancer (EC) is the fourth most common cancer in women, accounting for roughly 7% of all female cancers and more than 68,000 new cases expected annually in the United States. In 2024, for the first time, deaths from EC exceeded those from ovarian cancer - a stark reminder that this disease is growing in urgency as a public health problem.
Unlike most cancers, endometrial cancer has been increasing in both incidence and mortality for four decades, rising roughly 1% each year. Obesity, increasing life expectancy, and poorly understood genetic or environmental factors all contribute to this trend.
EC develops through a step-by-step process: normal endometrial cells first give rise to a precancerous state called endometrioid intraepithelial neoplasia (EIN), which can then progress to invasive cancer. Understanding what molecular events trigger this earliest step is critical to developing better ways to detect and prevent the disease.
Prior research identified several genetic mutations common in EC, especially in genes like PTEN, PIK3R1, and PIK3CA. Another known mechanism is promoter DNA hypermethylation - a chemical tag that silences certain protective genes. However, these known mechanisms do not explain the full picture of how EC begins.
PAX2 is a transcription factor - a protein that controls the activity of other genes. It plays an essential role in the development of the kidney and female reproductive tract during embryogenesis. In adults, PAX2 continues to be expressed in the glandular cells of the endometrium (the lining of the uterus).
Studies in clinical pathology have consistently found that PAX2 protein is absent in more than 80% of endometrial cancers and in precancerous EIN lesions. This striking loss across so many tumors suggests that losing PAX2 may be a key early event in cancer development - but until this study, no one had established why the protein disappears or what role that loss plays.
Importantly, PAX2 mutations are extremely rare in EC (less than 1% of cases). This means the protein is not being silenced by a DNA-level error; instead, something else must be turning the gene off. Previous suggestions that abnormal DNA methylation (a chemical modification that can silence genes) was responsible lacked strong supporting evidence.
The research team used a broad set of complementary approaches to understand PAX2 loss. They analyzed tissue samples from human patients across a range of ages and cancer grades, and they used a panel of 13 human EC cell lines - lab-grown cancer cells - to perform detailed molecular experiments.
To determine whether PAX2 protein loss was due to gene silencing (the gene being turned off) rather than gene deletion or mutation, they performed RNA in situ hybridization (RNAscope) to detect PAX2 messenger RNA directly in tissue sections, and FISH (fluorescence in situ hybridization) to physically visualize the PAX2 gene location in cell nuclei. If the gene were deleted or rearranged, FISH would detect missing signals.
To investigate the chromatin (the protein-DNA structure that controls gene access), the team performed ChIP-Seq and CUT&Tag - techniques that map the precise locations of specific histone modifications across the genome. These modifications, chemical tags on the proteins that DNA wraps around, act as on/off switches for gene expression. Key marks studied included H3K27ac (active chromatin) and H3K27me3 (repressive chromatin).
The team also engineered cell lines with inducible PAX2 expression (turning PAX2 back on using a drug called doxycycline) and used CRISPR activation (CRISPRa) to reactivate PAX2 from its native chromosomal location. These tools allowed direct testing of PAX2's functional impact on cancer cell behavior, including proliferation and tumor growth in mouse models.
Using RNA detection alongside protein detection in the same tissue sections, the researchers found that PAX2 messenger RNA (the molecular instruction to make the protein) was absent in the same cells that lacked PAX2 protein. This confirmed that PAX2 is being silenced at the transcriptional level - the gene itself is being turned off, not destroyed.
FISH analysis of 12 precancerous EIN cases showed intact PAX2 gene signals in all cases, ruling out chromosomal deletions or rearrangements as the cause. Combined with the rarity of PAX2 point mutations (less than 1%), this eliminated standard genetic mechanisms and pointed firmly toward an epigenetic (non-DNA-sequence) explanation.
PAX2 loss occurred as early as the EIN precancer stage and was equally common in low-grade and high-grade EC. This distribution pattern strongly indicates that PAX2 silencing is an early initiating event that happens before - or at the very beginning of - cancer formation, rather than a later consequence of tumor progression.
Clonal patches of PAX2-negative cells (individual glands where all cells have lost PAX2) were found in normal-appearing endometrium, and this phenomenon was age-dependent: absent in women aged 18-25 but present in 12/32 women aged 44-45 (P = 0.00022). This age-linked emergence mirrors the known age-related rise in EC incidence.
The most surprising finding was the identification of the mechanism behind PAX2 silencing. The researchers ruled out DNA methylation - the mechanism that silences many other tumor suppressor genes - as the cause. Instead, they discovered that PAX2 is silenced through changes to the histone code: the chemical modifications on the proteins that DNA wraps around.
In cells where PAX2 is expressed normally, the PAX2 promoter region carries H3K27ac marks, which indicate active, open chromatin and an enhancer that drives gene expression. In PAX2-silenced cancer cells, this active mark is replaced by H3K27me3 - a repressive mark associated with "facultative heterochromatin," meaning DNA that is compacted and inaccessible to the gene-reading machinery.
This shift - active enhancer to repressive chromatin - silences PAX2 without altering the DNA sequence at all. The repressive H3K27me3 mark spread along the chromosome in a pattern described as resembling a "pearl necklace," and its extent was bounded by cohesin loops (structural elements of 3D genome organization), which limited the silencing effect to the PAX2 region and prevented neighboring genes from being accidentally shut down.
Critically, when the researchers used CRISPR activation (CRISPRa) to artificially recruit transcriptional activators to the PAX2 promoter, PAX2 expression was successfully restored in all 12 silenced cell lines. This proved that PAX2 silencing is reversible - the gene is not permanently damaged - and opens the door to therapeutic strategies aimed at reactivating PAX2.
To confirm that PAX2 loss actively promotes cancer rather than being a passive bystander, the team performed both gain-of-function and loss-of-function experiments. When PAX2 was re-expressed in PAX2-deficient EC cell lines (Ishikawa and HEC-1-A), cells showed significant suppression of proliferation and reduced ability to form colonies - key indicators of reduced cancer growth potential.
In mouse xenograft experiments (where human cancer cells are implanted into mice to form tumors), PAX2 re-expression led to slower tumor growth and smaller tumor size, while PAX2 knockdown in a cell line that normally expresses PAX2 (AN3CA) produced the opposite: faster proliferation, accelerated wound closure, and larger tumors in mice.
Cell cycle analysis confirmed that PAX2 knockdown increased the proportion of cells in S and G2/M phases - the phases associated with active DNA replication and cell division - while reducing cells in the resting G0/G1 phase. This indicates that PAX2 normally acts as a brake on the cell cycle, and losing it allows cancer cells to divide faster.
This study describes what the authors believe to be a third general mechanism of tumor suppressor inactivation in cancer - distinct from the two well-established mechanisms of DNA mutation and promoter DNA hypermethylation. The PAX2 silencing described here is driven entirely by histone modifications, without any change to the underlying DNA sequence.
This distinction matters for cancer diagnostics: standard genome sequencing approaches that look for DNA mutations or methylation changes would completely miss this type of epigenetic event. New tools that profile the histone landscape (such as ChIP-Seq or CUT&Tag) would be needed to detect it.
The 3D genomic architecture - the way DNA folds in three dimensions inside the cell nucleus - plays a key role in this process. The cohesin-mediated loop domains that organize the genome act as insulating boundaries, limiting the spread of the repressive H3K27me3 mark and preventing unintended silencing of neighboring genes. This makes the epigenetic silencing a focal, locus-specific event rather than a broad, genome-wide disruption.
The finding that more than 80% of endometrial cancers share this same initiating molecular event - the epigenetic silencing of PAX2 - is remarkable. It identifies a common molecular fingerprint across a broad range of EC and raises the possibility of using PAX2 loss as a universal early biomarker for endometrial cancer detection.
The discovery that PAX2 silencing is both an early event and reversible carries significant clinical implications. Because PAX2 loss appears before invasive cancer forms, it could potentially serve as a biomarker for early detection - identifying women at risk before cancer fully develops.
Because the PAX2 locus is not physically destroyed - only epigenetically silenced - it may be possible to design therapies that reactivate PAX2 expression, essentially reversing the initiating event in cancer. The CRISPR activation proof-of-concept demonstrated in this study supports this idea, though translating such approaches to clinical treatments remains a future challenge.
Understanding PAX2's downstream effects is also important. When PAX2 is lost, the study found that thousands of enhancers (regulatory elements that control nearby genes) are remodeled globally - a process called enhancer reprogramming - rewiring the entire transcriptional landscape of endometrial cells in ways that promote cancer growth. Identifying which gene networks are most dysregulated by PAX2 loss could reveal additional therapeutic targets.
The broader message is that cancer-initiating events can be entirely epigenetic - not encoded in the DNA sequence, but in how DNA is packaged and regulated. This challenges the prevailing mutation-centric view of cancer and highlights the importance of expanding cancer research and diagnostics to include histone-based epigenomic profiling.