The study team, led by researchers at Zhejiang University and The Third Affiliated Hospital of Guangzhou Medical University, analyzed ten human cervical tissue samples spanning the full pathological continuum. Single-cell RNA sequencing allowed them to profile thousands of individual cells, resolving not only which cell types were present at each disease stage but also how their gene-expression programs shifted as lesions progressed. Whole-genome bisulfite sequencing added a crucial layer of epigenetic information, revealing how DNA methylation patterns change across cell types during carcinogenesis, while spatial transcriptomic data from a public cohort confirmed that the cellular relationships observed in dissociated single-cell data hold true within intact tissue architecture.

One of the most striking findings is that disease progression is accompanied by a marked increase in NK/T cell infiltration. As normal tissue advances through LSIL and HSIL toward invasive carcinoma, immune cells of the NK and T lineages crowd increasingly into the lesion environment. This might, at first glance, seem encouraging, since cytotoxic lymphocytes are the very cells capable of destroying tumor cells. Yet the infiltration coincides with upregulation of galectin-9, an immune checkpoint ligand encoded by the LGALS9 gene. Galectin-9 is known to bind Tim-3 on T cells, a interaction that dampens antitumor immunity and drives T-cell exhaustion. In other words, the tumor microenvironment appears to respond to escalating immune pressure by deploying an immunosuppressive ligand, a classic example of adaptive immune resistance.

The epigenetic data revealed something particularly interesting about how this deployment unfolds over time. The LGALS9 promoter underwent progressive demethylation beginning at the low-grade squamous intraepithelial lesion stage, indicating that the gene was being epigenetically primed for expression long before invasive cancer appeared. However, robust transcriptional induction of LGALS9 only became prominent at the high-grade lesion stage, coinciding with activation of interferon-gamma response programs. This temporal separation between demethylation and transcriptional activation suggests a two-step mechanism: early epigenetic poising followed by inflammatory triggering. The researchers found a remarkably tight correlation between interferon-gamma activity and LGALS9 expression at the sample level, with a Pearson correlation coefficient of 0.95, indicating that the very immune cells infiltrating the lesion may be inducing the ligand that ultimately silences them.

Beyond galectin-9, trajectory analysis of the single-cell data delineated the evolution of CD8-positive T cells across disease stages and identified TFCP2 as a transcriptional regulator whose activity is associated with patient prognosis. The study also documented enrichment of LAMP3-positive mature dendritic cells in tumor tissues compared with normal controls. These dendritic cells, which emerge along a maturation trajectory from conventional cDC1 and cDC2 subsets, carried a mixture of costimulatory molecules such as CD40 and CD80 and inhibitory checkpoint molecules including CD274, IDO1, and LGALS9 itself, suggesting that even the antigen-presenting arm of the immune response becomes entangled in the checkpoint machinery as cancer develops.

Multiplex immunohistochemistry provided protein-level confirmation of the story told by the sequencing data. Staining for exhausted CD8-positive T cells, marked by the co-expression of CD3, CD8, and Tim-3, alongside staining for the epithelial marker pan-cytokeratin and galectin-9, showed that galectin-9-positive epithelial cells increase in abundance as tissue progresses from normal cervix through LSIL and HSIL to cancer. The physical co-localization of Tim-3-expressing exhausted T cells with galectin-9-expressing epithelium within the same tissue sections strengthens the argument that this ligand-receptor pair represents a functional axis of immune evasion operating during precancerous progression, not merely a correlate of advanced disease.

Perhaps the most translational portion of the work came from animal experiments. The researchers established an ectopic subcutaneous syngeneic cervical cancer model in immunocompetent mice, an experimental system in which the immune system is fully intact and therefore capable of mounting genuine antitumor responses. Blocking galectin-9 in this model reduced tumor burden, demonstrating that the protein is not simply a passive biomarker but an active contributor to tumor growth. More strikingly, combining galectin-9 blockade with an agonist antibody against GITR, a costimulatory receptor on T cells, significantly enhanced the clonal expansion and cytotoxic activity of CD8-positive T cells. This combination strategy suggests that releasing one brake on the immune system while simultaneously pressing the accelerator may produce therapeutic effects greater than either intervention alone.

The findings arrive at a moment when immune checkpoint blockade has transformed the treatment of many cancers but has delivered comparatively modest results in cervical cancer. Understanding which checkpoint pathways are active at which stages of disease could allow clinicians to intervene earlier and more precisely. The observation that LGALS9 epigenetic poising begins at the LSIL stage is particularly provocative, since low-grade lesions are common, usually regress spontaneously, and are typically managed conservatively. If reliable markers of galectin-9 activation could be incorporated into screening algorithms, they might help distinguish the minority of low-grade lesions destined for progression from those that will resolve, sparing unnecessary procedures while directing attention to lesions that truly warrant close surveillance.

The study also illustrates the growing power of integrated multi-omics approaches in cancer biology. No single technology used here could have revealed the full sequence of events. Single-cell transcriptomics exposed the cellular composition and signaling programs of each lesion stage, but only DNA methylation profiling revealed that LGALS9 had been epigenetically prepared in advance of its expression, and only spatial transcriptomics could verify that the relevant cell populations occupy adjacent territories within intact tissue. Copy number variation inference, pseudotime trajectory modeling, and regulon analysis with tools such as pySCENIC and Monocle2 added further resolution, while methylation-based deconvolution using EpiSCORE extended the key NK/T cell infiltration trend across a larger cohort of twenty-one bulk tissue samples.

Caveats remain, as they do in any early-stage translational study. The human cohort comprised ten deeply profiled samples, and the syngeneic mouse model, while immunocompetent, does not fully recapitulate HPV-driven human cervical carcinogenesis. Clinical testing of galectin-9 blockade in cervical cancer patients would need to demonstrate safety and efficacy in the neoadjuvant, recurrent, or metastatic settings where immunotherapy is currently deployed. Nevertheless, by pinpointing a checkpoint ligand whose activation is detectable during precancerous progression and whose blockade shows antitumor efficacy in vivo, the study provides both a mechanistic framework for understanding immune evasion in cervical carcinogenesis and a concrete, testable therapeutic hypothesis. For a disease that remains a leading cause of cancer death among women globally, that combination of mechanistic insight and actionable target represents a meaningful step forward.

Subject of Research: Multi-omics analysis of immune evasion during cervical carcinogenesis and galectin-9 as a candidate immunotherapeutic target

Article Title: Multi-omics analysis of cervical carcinogenesis reveals galectin-9 driven immune evasion as a candidate immunotherapeutic target

Article References: Multi-omics analysis of cervical carcinogenesis reveals galectin-9 driven immune evasion as a candidate immunotherapeutic target. (n.d.). https://doi.org/10.1186/s12935-026-04458-1

Image Credits: AI Generated

DOI: 10.1186/s12935-026-04458-1

Keywords: cervical cancer, galectin-9, LGALS9, single-cell RNA sequencing, spatial transcriptomics, DNA methylation, CD8-positive T cells, immune evasion, immunotherapy, tumor microenvironment, interferon-gamma, cervical carcinogenesis

Persistent infection with high-risk human papillomavirus, or HR-HPV, is the central engine of cervical carcinogenesis. Among the dozen or so high-risk types, HPV-16 and HPV-18 dominate the global conversation, and it is against these that prophylactic vaccines were primarily engineered. Yet the distribution of HPV types is strikingly uneven across geography. HPV-39, one of the dozen high-risk types, accounts for a comparatively larger share of infections in central China, a pattern that makes local molecular data not a luxury but a necessity. Without knowing what the virus actually looks like in a given population—which variants of its oncogenes are circulating, and how those variants differ from the reference sequences that diagnostic assays and immunological studies are typically built around—clinicians and researchers are effectively operating with an outdated map.

The research team, led by corresponding author Bing Mei of the hospital’s Department of Laboratory Medicine, approached the problem with a deliberate and technically careful methodology. The investigators began by identifying clinical samples carrying a single HPV-39 infection, as opposed to mixed infections with multiple HPV types, using HPV genotyping before any amplification step. This filtering matters: mixed infections can complicate sequencing results and obscure the true variant composition of a single viral type. Once pure HPV-39-positive samples were secured, the researchers amplified the E6 and E7 genes using polymerase chain reaction and read the resulting sequences by Sanger sequencing, the gold-standard method for confirming the exact nucleotide identity of relatively short DNA fragments.

The sequencing effort produced a rich dataset. Across the E6 and E7 sequences, the team identified a total of 33 single nucleotide variants. Of these, 13 were synonymous mutations—changes in the viral DNA that do not alter the amino acid sequence of the resulting protein—and 20 were non-synonymous mutations, meaning they actually changed the protein-building instructions. Notably, six of the non-synonymous mutations had never been recorded before, making them novel additions to the global catalogue of HPV-39 variation. The distinction between synonymous and non-synonymous changes is more than academic bookkeeping. Synonymous mutations tend to accumulate neutrally, offering a passive record of viral ancestry, whereas non-synonymous changes can reshape protein function, alter how the virus interacts with host cell machinery, or change how the immune system recognizes infected cells.

To place their findings in a broader evolutionary context, the researchers conducted phylogenetic analysis based on the combined E6 and E7 sequences. The picture that emerged was one of relative genetic conservatism: the majority of sequences—56 out of 71—showed similarity to lineage A of HPV-39, suggesting that this lineage is the predominant circulating form in central China. Lineage assignment is a useful epidemiological tool because different lineages of a given HPV type have been associated, in some studies, with different levels of oncogenic risk and different geographic distributions. The dominance of a single lineage in the region provides a baseline against which future shifts in the viral population can be measured.

The team also interrogated the sequences for evidence of selective pressure, asking whether any individual amino acid positions in E6 or E7 showed signs of being actively favored by evolution—changes that would suggest the virus is under pressure to adapt, perhaps in response to host immune defenses. The answer was negative: no positively selected sites were detected in either gene. That finding is consistent with a virus in a relatively stable evolutionary state, at least within this population and timeframe, but it does not mean the virus is standing still. Individual mutations can still matter for immune recognition even in the absence of genome-wide signatures of positive selection.

That is precisely where the study’s immunological analysis becomes compelling. The researchers predicted T-cell epitopes—short stretches of viral protein that, when displayed on the surface of infected cells by human leukocyte antigen (HLA) molecules, can be recognized by cytotoxic T lymphocytes, the immune system’s specialized killers. Substitutions within or near these epitopes can change how well they bind to HLA molecules or how they are recognized by T-cell receptors, potentially allowing infected cells to escape immune surveillance. In the E6 protein, the predicted T-cell epitopes were influenced by non-conservative substitutions at eight positions: H5Y, P7S, T19R, Q26H, E46D, A87G, L102S, and L151P. In E7, eleven positions were implicated: L8V, D14H, D14E, H50Q, D57E, R87Q, Q88E, Q88D, L89V, Q91K, and F93C. Non-conservative substitutions—those that replace one amino acid with another of very different chemical character—are particularly noteworthy because they are more likely to alter protein folding, binding interactions, or antigen presentation than conservative swaps between chemically similar residues.

Among the immunological findings, one result stands out. The team identified a predicted candidate T-cell epitope spanning residues 70 to 78 of the E6 protein, with the amino acid sequence KFYAKIREL, restricted by the HLA-C*06:02 molecule. This is the kind of precise, actionable information that underpins the design of therapeutic vaccines and adoptive T-cell therapies, which aim to train or harness a patient’s own immune system to recognize and destroy virus-infected or tumor cells. An epitope’s restriction by a specific HLA allele is critical, because HLA genes are extraordinarily polymorphic—each person carries a different combination of HLA alleles, and an epitope that is well-presented by one HLA molecule may be invisible to another. Identifying epitopes restricted by common HLA alleles in a given population is a prerequisite for developing immunotherapies that can work broadly.

The E6 and E7 proteins themselves deserve a moment of technical explanation, because their biology is central to why this study matters. In the life cycle of high-risk HPV, E6 and E7 are the two principal oncoproteins. E6 recruits a cellular complex that tags the tumor suppressor protein p53 for degradation, effectively silencing one of the cell’s most important safeguards against cancer. E7 binds to the retinoblastoma protein (pRb), releasing transcription factors that push the cell into the DNA synthesis phase of the cell cycle, creating the cellular environment the virus needs to replicate its genome. When HPV infection persists and its DNA integrates into the host genome, the expression of E6 and E7 becomes dysregulated, and their oncogenic activities accumulate. These proteins are also major targets of immune recognition, which is why sequence variation in their genes can have cascading consequences for both cancer biology and immunology.

The regional context of this work gives it particular weight. Central China has been the site of several earlier studies of HPV genetic variability, including analyses of HPV-16 E6 and E7 published in 2023 and HPV-18 E6, E7, and L1 published in 2024, as well as an examination of HPV-39 variability in Southwest China in 2021. But HPV-39 data specific to central China remained thin. By filling that gap, the Jingzhou team contributes a piece to a larger puzzle: understanding how HPV genetic diversity is distributed across populations, and whether regionally specific variants might affect the performance of HPV screening tests, which often rely on probe-based detection of particular sequence regions. Sequence variation in a probe-binding site can, in principle, reduce assay sensitivity, so cataloguing circulating variants is a practical safeguard for diagnostic reliability.

The study was supported by a key research and development plan project of Hubei Province, and it was conducted under the approval of the Ethics Committee of Jingzhou Hospital Affiliated to Yangtze University, with informed consent obtained from all participants and privacy protections in place before sample collection. The work reflects the broader push toward open-access sharing of research, with the article published under a Creative Commons license that allows free reading and distribution with attribution.

Looking ahead, the researchers position their dataset as a foundation for future investigations into HPV-39 evolution and host immune interactions. The six novel non-synonymous mutations they describe may merit functional characterization—laboratory studies that test whether these changes alter E6 or E7 protein activity, transform cell behavior in culture, or affect the binding of viral proteins to host factors. The T-cell epitope predictions, meanwhile, invite immunological validation: laboratory assays could confirm whether the KFYAKIREL epitope is genuinely presented by HLA-C*06:02 and recognized by T cells, and whether the other identified substitutions measurably reduce immune recognition. The authors also caution, implicitly, that the database of HPV-39 sequences remains sparse compared with the resources available for HPV-16 and HPV-18, so every additional regional dataset helps refine the global picture.

For a virus that infects millions and contributes to hundreds of thousands of cervical cancer deaths worldwide each year, HPV-39 has lived in the shadow of its more famous relatives. This study does not claim to overturn any paradigm, but it does something quieter and arguably more durable: it replaces guesswork with sequence data for a region where the virus matters, it documents variations that could influence immune escape, and it hands vaccine developers a concrete candidate epitope to investigate. In molecular epidemiology, that combination of surveillance, mechanism, and forward-looking application is the standard against which useful research is measured—and on that measure, this work earns its place in the growing literature on HPV diversity in China.

The broader message for public health is also worth underscoring. As cervical cancer screening and vaccination programs expand globally, understanding the genetic makeup of circulating HPV types becomes part of ensuring those programs work as intended. Variants that alter diagnostic targets, change immunodominant epitopes, or signal shifts in viral lineage composition could all influence the long-term effectiveness of prevention and treatment strategies. Studies like this one, grounded in careful sequencing of clinical samples and rigorous bioinformatic analysis, provide the granular molecular data that population-level policy ultimately depends on. The authors’ catalog of 33 variants, their lineage assignments, and their epitope predictions will now become part of the reference landscape that other researchers in China and beyond can build upon as the effort to eliminate cervical cancer enters its most demanding phase.

The distinction matters because infection with a high-risk HPV type is common, while cervical cancer is relatively uncommon among infected individuals. Most HPV infections are transient and are cleared by the immune system without causing lasting cellular damage. A smaller proportion persist, and persistent infection can create the conditions for precancerous lesions and, eventually, invasive cancer. Current screening methods—including HPV DNA testing, cytology and visual examination—have improved early detection, but they do not always identify which positive results represent a short-lived infection and which signal a transformation process already underway. This uncertainty can lead to repeated testing, unnecessary procedures or delayed attention to genuinely high-risk disease.

HPV genotyping supplies an important part of the missing information. Rather than reporting only that HPV is present, genotyping identifies the specific viral type or group. Some strains, particularly HPV16 and HPV18, are strongly associated with cervical cancer, while other high-risk types carry different levels of risk. Genotyping is therefore useful for surveillance, triage and evaluating vaccination programs. Yet the viral type alone cannot reveal whether the virus is actively altering the host cell. Two people infected with the same high-risk strain may have very different outcomes, depending on viral persistence, immune responses, viral gene expression and changes accumulating in cervical cells.

The proposed second layer of testing focuses on non-coding RNAs, or ncRNAs. These include microRNAs, long non-coding RNAs and circular RNAs, each with distinct molecular properties and functions. MicroRNAs are short RNA molecules that bind to messenger RNAs and usually reduce the production of specific proteins. Long non-coding RNAs can influence gene transcription, organize protein complexes or alter the stability and availability of other RNAs. Circular RNAs form closed loops that can be unusually resistant to degradation and may regulate gene expression by binding microRNAs or interacting with proteins. Together, these molecules create a regulatory network that can reflect how a cervical cell is responding to HPV.

The biological connection between HPV and ncRNAs is central to the review’s argument. High-risk HPV produces oncoproteins, particularly E6 and E7, that interfere with cellular safeguards governing genome stability, cell division and programmed cell death. E6 can promote the degradation or functional suppression of the tumor-suppressor protein p53, while E7 disrupts the retinoblastoma pathway, which normally restrains inappropriate entry into the cell cycle. These disturbances can reshape the cell’s transcriptional and post-transcriptional programs. In turn, altered ncRNAs may help sustain proliferation, weaken apoptosis, modify immune signaling and promote invasion—features associated with malignant progression.

Several candidate molecules illustrate how such signals might be used. The review discusses microRNAs whose abundance changes in HPV-associated cervical disease, including miR-21, miR-155, miR-375, miR-145 and miR-106b-5p. Some have been linked with enhanced growth, migration or poor prognosis, while others appear to restrain invasion or proliferation and become reduced during disease progression. The molecular effects are not interchangeable: miR-21, for example, has been associated with pathways affecting growth and migration, whereas reduced miR-375 has been linked to loss of control over transcription factor SP1. These patterns could form part of a multi-marker signature rather than serve as stand-alone diagnostic tests.

Long non-coding and circular RNAs may add information about the architecture and persistence of the cancer-associated regulatory network. HOTAIR has been associated with cervical cancer progression and has been studied in serum and vaginal discharge, while MALAT1 can be regulated by the HPV16 E7 protein and has been connected to cancer progression, metastasis, immunity and treatment resistance. Other studies summarized in the review describe HPV-related changes in lnc-FANCI-2 and circular RNAs involved in regulatory axes such as miR-1236-3p/TRIM37 and miR-1179/ABL2. Because circular RNAs are structurally stable and many ncRNAs can circulate in blood or other biological fluids, they are attractive candidates for liquid biopsy tests that might avoid or complement tissue sampling.

In practice, an integrated test could combine several forms of evidence from one sample. HPV DNA analysis would identify the viral genotype, while RNA profiling could measure a panel of host-response molecules and perhaps indicators of active oncogenic signaling. A computational model could then classify a patient’s result into a lower- or higher-risk category, guiding the interval for repeat screening or the need for colposcopy and biopsy. The same principle could potentially be adapted to self-collected vaginal samples, cervical specimens or blood-based assays, although the review does not present a clinically validated test or a new patient cohort. It describes a diagnostic framework whose usefulness depends on selecting robust markers, standardizing laboratory methods and proving that the combined signal predicts outcomes better than existing screening.

That evidence gap is the most important qualification. The authors emphasize that individual ncRNAs have shown promising associations with HPV-mediated carcinogenesis, but evidence for their combined clinical use with HPV genotyping remains limited. Biomarker levels can vary with age, inflammation, hormonal status, sampling technique, disease stage and the composition of the surrounding tissue. Results generated in cell lines or small retrospective studies may not translate directly to diverse populations or routine clinics. Prospective studies will need to follow HPV-positive individuals over time, compare transient and persistent infections, test different viral genotypes and establish thresholds that are reproducible across laboratories. The review also notes that no datasets were generated or analyzed for the study itself, because it is a narrative synthesis of existing evidence.

If validated, the strategy could make cervical screening more precise without replacing established prevention tools. HPV vaccination remains fundamental, and conventional HPV testing and cytology remain valuable for identifying people who need further assessment. The promise of ncRNA profiling is to add biological context: not only which virus is present, but how strongly it is influencing the host cell. Such information could reduce the burden of unnecessary follow-up while directing clinical attention toward infections with molecular signs of transformation. For now, the concept is best viewed as an emerging precision-screening platform rather than a ready-to-use diagnostic. Its viral-news appeal lies in a simple idea with complex biology: the future of cervical cancer detection may depend on listening simultaneously to the genome of the virus and the regulatory RNA language of the cell it infects.