IRX1 blocks cervical cancer growth by curbing lipid synthesis enzyme ACLY

Scientists have identified a previously hidden circuit that governs how cervical cancer cells build their own fat, and the discovery may point toward new ways to starve one of the world’s most common tumors. In a study published in iScience, a team led by Xiangyu Liu, Xinwei Hou, and Yuanyuan Zhang under the supervision of Ke Wang at Tianjin Medical University Cancer Hospital reports that a developmental transcription factor called IRX1 acts as a tumor suppressor in cervical squamous cell carcinoma by directly shutting down a master metabolic enzyme, ATP-citrate lyase, or ACLY. The work also reveals an unexpected partner in this process, the nucleoside diphosphate kinase NME3, which is physically escorted into the nucleus by IRX1 to complete the repression. Together, the three molecules form what the researchers call the IRX1-NME3/ACLY axis, a regulatory node that links epigenetic silencing to the lipid-hungry growth program of cancer cells.

The significance of the finding rests on a long-standing principle in tumor biology. Cancer cells do not merely consume lipids from their surroundings; they synthesize enormous quantities of fatty acids and cholesterol from scratch, a process known as de novo lipogenesis. These lipids supply the raw material for new membranes as cells divide, fuel stores for rapid growth, and precursors for signaling molecules that remodel the tumor microenvironment. ACLY sits at the branchpoint of this pipeline. The enzyme cleaves citrate exported from the mitochondria into acetyl-CoA and oxaloacetate, effectively converting the products of carbohydrate metabolism into the two-carbon building blocks of fatty acid synthesis. Because ACLY feeds both fatty acid and cholesterol production, it has long been considered an attractive drug target, and the new study now identifies an upstream transcriptional brake on the gene that encodes it.

The story began with a simple observation: IRX1 protein levels are consistently lower in cervical cancer cell lines, including HeLa, C33A, SiHa, and CaSki, than in non-tumorigenic cervical epithelial cells. IRX1 is a member of the Iroquois homeobox family, transcription factors best known for patterning tissues during embryonic development. Prior work had already implicated the gene in other malignancies. In gastric cancer, IRX1 is frequently silenced by promoter hypermethylation, and restoring its expression suppresses proliferation and invasion. In non-small cell lung cancer, promoter methylation of IRX1 correlates with reduced patient survival, and in head and neck squamous cell carcinoma the same epigenetic lesion tracks with tumor progression. The cervical cancer study extends this pattern to a new tumor type and, crucially, explains how the silencing translates into metabolic advantage.

Functionally, the consequences of manipulating IRX1 were striking. When the researchers forced IRX1 expression in HeLa and C33A cells, proliferation measured by CCK-8, EdU incorporation, and colony formation dropped markedly, and the cells lost much of their ability to migrate through Transwell membranes or close scratch wounds. Conversely, silencing IRX1 in SiHa and CaSki cells produced the opposite phenotype, accelerating growth and movement. Western blotting of epithelial-mesenchymal transition markers provided a molecular signature for these behaviors: IRX1 overexpression raised E-cadherin while lowering N-cadherin and vimentin, shifts consistent with a less invasive, more epithelial state. Because lipid metabolism is known to feed EMT programs, these observations hinted that the transcription factor’s effects might run through metabolism rather than through classical cell-cycle genes alone.

To test that idea, the team performed RNA sequencing on IRX1-overexpressing HeLa cells and interrogated the results with Gene Ontology and gene set enrichment analysis. The differentially expressed genes clustered overwhelmingly in lipid and cholesterol metabolism pathways. Among the enzymes whose protein levels fell when IRX1 was restored were FASN, SREBP1, ACSL3, and most importantly ACLY. Oil Red O staining, which dyes intracellular lipid droplets a vivid red, showed that IRX1-overexpressing cells accumulated far fewer droplets, while IRX1 knockdown cells became visibly oilier. Direct biochemical assays confirmed the trend: triglyceride and cholesterol contents fell with IRX1 overexpression and rose when the gene was removed. The metabolites acetyl-CoA and malonyl-CoA, the committed substrates of fatty acid synthesis, followed the same pattern, placing IRX1 squarely upstream of the citrate-to-acetyl-CoA conversion step.

The mechanism of repression turned out to be strikingly direct. By cross-referencing RNA-seq data with lipid metabolism gene sets, the researchers narrowed seven candidate genes down to ACLY, which sat at the metabolic branchpoint and showed the largest expression change. Database analysis of IRX1’s binding motif predicted three potential docking sites in the ACLY promoter, located at positions -1328 to -1321, -976 to -965, and -11 to -4 relative to the transcription start site. Chromatin immunoprecipitation confirmed that IRX1 binds specifically to the first of these sites, and dual-luciferase reporter assays demonstrated that IRX1 overexpression represses a promoter fragment containing that site. When the researchers mutated the binding sequence from 5′-TGTTATCT-3′ to 5′-TGCGGCCT-3′, IRX1 lost its ability to silence the reporter entirely, proving that the physical contact between protein and DNA is required for the metabolic shutdown.

Perhaps the most novel element of the study concerns NME3, a member of the nucleoside diphosphate kinase family whose expression is prognostic in several cancers. Co-immunoprecipitation followed by mass spectrometry identified NME3 as a high-confidence IRX1-interacting protein, an association the team verified with both endogenous and tagged co-immunoprecipitation and by immunofluorescence showing nuclear co-localization. Intriguingly, IRX1 did not change how much NME3 the cells produced, and NME3 alone had no effect on ACLY promoter activity. Instead, co-expression experiments revealed that NME3 potentiates IRX1’s repression: NME3 overexpression strengthened IRX1-mediated silencing of the ACLY promoter, while its overexpression in IRX1-depleted cells attenuated repression. Nuclear-cytoplasmic fractionation and immunofluorescence showed that IRX1 overexpression drives NME3 into the nucleus, while IRX1 knockdown keeps NME3 away. In other words, IRX1 does not regulate its partner’s abundance but rather its whereabouts, recruiting NME3 to the ACLY promoter to assemble a functional co-repressor complex.

Rescue experiments established that ACLY is the load-bearing target of this axis. When the researchers re-expressed ACLY in IRX1-overexpressing HeLa cells, the metabolic brakes came off: lipid droplets, triglycerides, cholesterol, acetyl-CoA, and malonyl-CoA all rebounded toward their original levels. Symmetrically, knocking ACLY down in IRX1-depleted SiHa cells reversed the metabolite surge caused by losing IRX1. Proliferation and migration assays told the same story, with ACLY manipulation counteracting the growth and motility effects of IRX1 in both directions. The team then moved into animals, injecting control, IRX1-overexpressing, and IRX1-plus-ACLY-overexpressing HeLa cells subcutaneously into NSG mice. IRX1 overexpression slowed tumor growth, but forcing ACLY back on partially restored it, and immunohistochemistry for Ki-67 and BAX confirmed that ACLY rescue abolished the changes in proliferation and apoptosis markers induced by IRX1.

The epigenetic origin of the defect adds a translational dimension. Analysis of TCGA data through the UALCAN portal showed that the IRX1 promoter carries higher methylation beta-values in cervical tumors than in normal cervical tissue, and bisulfite sequencing PCR confirmed elevated methylation density at the promoter CpG island in cancer cell lines and clinical samples. When the researchers treated HeLa and SiHa cells with the DNA methyltransferase inhibitor 5-aza-2′-deoxycytidine, IRX1 expression returned in a dose-dependent fashion over 96 hours, while the low-methylation normal cell line End1/E6E7 responded only marginally. This suggests that the tumor suppressor is not deleted or mutated but chemically muted, and that demethylating agents might in principle reactivate the IRX1-NME3/ACLY brake in patients. Notably, the authors found that the axis operates in both HPV-positive and HPV-negative cervical cancer cell lines, indicating the mechanism is independent of viral status.

The findings arrive with appropriate caveats. The functional work relies heavily on cell culture overexpression and knockdown systems that do not fully reproduce the tumor microenvironment, the xenograft study used only five mice per group, and no independent clinical cohort validation has yet been performed. The team also did not carry out global metabolomic profiling, so other metabolic pathways controlled by IRX1 may await discovery, and the precise molecular means by which NME3 assists transcriptional repression remain unresolved. Even so, the study delivers a complete chain of evidence, from epigenetic silencing of a homeobox gene, through recruitment of a co-repressor, to direct repression of a lipogenic enzyme, and finally to measurable changes in tumor growth. If follow-up work confirms the axis in patient tumors, ACLY promoters and NME3 nuclear localization could join the growing list of metabolic biomarkers, and strategies that restore IRX1 function or mimic its grip on the ACLY promoter may offer a rational route to cutting off the fuel supply of cervical cancer.