The research team used 5xFAD mice, a widely studied transgenic model that carries five human Alzheimer’s disease mutations and progressively accumulates amyloid-beta plaques in the brain, as a proxy for an amyloid-associated brain state. Four-month-old female 5xFAD mice and age-matched wild-type controls were divided into two exposure groups: one received only routine handling, while the other underwent combined isoflurane anesthesia and abdominal surgery, a standard experimental paradigm for modeling the perioperative experience. Twenty-four hours later, a time point chosen to capture the acute molecular aftermath rather than the immediate chaos of the operation itself, the researchers dissected the hippocampus, the seahorse-shaped structure essential for memory formation that is also among the first brain regions damaged in Alzheimer’s disease, and performed bulk RNA sequencing to read out the activity of every gene in that tissue.

The experimental design followed a factorial structure, with three animals in each of the four genotype-by-exposure cells, and the team applied DESeq2, a widely used statistical framework for count-based gene expression data, to estimate three separate effects: the effect of genotype, the effect of anesthesia and surgery exposure, and, most importantly, the genotype-by-exposure interaction. This interaction term is the statistical heart of the study. It asks whether the transcriptional response to surgery differs between the two genotypes in direction, not merely in magnitude. When the researchers examined each genotype separately, the wild-type mice showed 20 genes significantly associated with exposure at a false discovery rate below 0.05, while the 5xFAD mice showed only 3. On its face, that might suggest the Alzheimer-model brain responds less to surgery. But the interaction analysis told a strikingly different story.

Testing the interaction directly, the team identified 53 genes whose response to anesthesia and surgery differed between genotypes at false discovery rate below 0.05, with 37 of those also showing an interaction effect size of at least one log2 unit, meaning the exposure effect was at least doubled in one genotype relative to the other. Remarkably, all 53 of these genes had opposite-signed exposure-effect estimates in wild-type versus 5xFAD mice: genes that went up after surgery in one genotype went down in the other, and vice versa. In other words, the Alzheimer-like brain did not simply mount a weaker or stronger version of the surgical response; it mounted a directionally inverted one. The authors are careful to note that separate within-genotype significance was not required for these genes, which is precisely why the interaction model, rather than the two separate contrasts, was the appropriate lens.

Because small exploratory cohorts are vulnerable to the influence of any single animal, the researchers subjected their findings to an unusually thorough battery of sensitivity analyses. They re-ran the model twelve times, each time leaving out one sample, and found that all 53 interaction genes retained their direction of effect in every leave-one-out fit. They repeated the analysis with edgeR quasi-likelihood modeling, a second independent statistical framework, and again all 53 genes kept their direction, although none reached genome-wide false discovery rate significance in edgeR, a candid acknowledgment that gene-level significance in this dataset is method-dependent. Cook’s-distance diagnostics, which flag individual data points with disproportionate influence on a regression, showed that no gene exceeded the cutoff. The directional finding, in short, was robust; the gene-by-gene statistical significance was not, and the authors say so plainly.

To move from individual genes to biological meaning, the team performed gene-set enrichment analysis using the Mouse Hallmark collection of curated pathway definitions, ranking genes by their interaction statistics rather than by any single genotype’s effect. This revealed 10 Hallmark pathways significantly enriched at false discovery rate below 0.05, and the pattern they formed was coherent and provocative. On the positive side of the interaction, meaning programs amplified in the 5xFAD response relative to wild type, sat inflammatory signaling pathways, including the TNF-alpha/NF-kappaB axis, a canonical pro-inflammatory cascade, and interferon programs, the antiviral-style immune signaling increasingly implicated in neurodegeneration. On the negative side, meaning programs blunted in the 5xFAD response, sat oxidative phosphorylation, the mitochondrial energy-generating machinery, and lipid and metabolic programs. All 10 pathways retained their direction and significance when the analysis was repeated with edgeR-ranked statistics.

That immune-up, energy-down signature is a pattern that will look familiar to anyone who follows Alzheimer’s research. Chronic neuroinflammation and mitochondrial dysfunction are two of the most consistently reported features of the diseased brain, and the study suggests that anesthesia and surgery may push an amyloid-burdened hippocampus further along both of these axes while a healthy hippocampus responds with a different, less inflammatory and less metabolically suppressed program. The authors also scored the expression of four marker-based cell-type signatures to ask whether the interaction effects could be explained by shifts in the relative abundance or activity of specific cell populations such as neurons, astrocytes, or microglia, but none of the four scores showed a significant interaction, leaving the cellular source of the reconfiguration an open question for future single-cell work.

The team also checked their findings against two public mouse transcriptomic datasets for external context, and here the results were sobering: none of the highlighted genes replicated at false discovery rate below 0.05 in those external datasets. This is not entirely surprising, since public datasets rarely match the precise combination of genotype, age, sex, exposure type, and post-exposure time point used here, but it underscores the exploratory nature of the work. The authors are equally explicit about the limits of interpretation in the other direction: the study does not establish a perioperative neurocognitive disorder phenotype, because no behavioral testing was performed; it cannot distinguish the effects of anesthesia from those of surgery, because the two were always combined; and it does not validate any biomarker. With three animals per group, the findings are hypothesis-generating by design, not definitive.

What the study does deliver is a well-guarded hypothesis with an unusually honest statistical presentation. By framing the question as an interaction rather than a pair of separate contrasts, the researchers were able to detect a phenomenon that the conventional approach would have missed entirely: a qualitative reorganization, rather than a quantitative amplification, of the hippocampal transcriptional response to the perioperative period in an Alzheimer-susceptible brain. The robustness of the pathway-level immune-metabolic pattern, which survived every sensitivity check the authors threw at it, contrasts instructively with the fragility of individual gene significance, a lesson in where the signal in small transcriptomic studies is most likely to live. The authors call for larger, longitudinal, behaviorally phenotyped and experimentally validated studies to determine whether this molecular reconfiguration translates into the cognitive vulnerability that clinicians observe in the operating room, and whether protecting the hippocampus from that inverted inflammatory-metabolic shift could one day shield vulnerable patients from postoperative cognitive decline.

Subject of Research: Genotype-dependent hippocampal transcriptional responses to anesthesia and surgery in 5xFAD Alzheimer-model mice

Article Title: Genotype-dependent reconfiguration of hippocampal transcriptional responses to anesthesia/surgery at 24 h in 5xFAD mice: an exploratory interaction-based analysis

Article References: Wang, S., Song, J., Sun, S., Xiao, R., Zhang, Z., Zhang, J., & Yao, J. (2026). Genotype-dependent reconfiguration of hippocampal transcriptional responses to anesthesia/surgery at 24 h in 5xFAD mice: an exploratory interaction-based analysis. BMC Genomics. https://doi.org/10.1186/s12864-026-13390-7

Image Credits: AI Generated

DOI: 10.1186/s12864-026-13390-7

Keywords: Alzheimer's disease, 5xFAD mice, anesthesia, surgery, perioperative neurocognitive disorders, hippocampus, RNA sequencing, gene expression, genotype-by-exposure interaction, gene-set enrichment analysis, neuroinflammation, oxidative phosphorylation

The eleven hallmarks trace their lineage to a landmark 2013 review in Cell by Carlos López-Otín and colleagues, which was expanded in 2023 to include disabled macroautophagy, chronic inflammation, dysbiosis and splicing dysregulation alongside the original nine categories such as genomic instability, telomere attrition, loss of proteostasis, epigenetic alterations, mitochondrial dysfunction and cellular senescence. The framework has been enormously influential, but it has also exposed a practical problem: the genes assigned to each hallmark live in different databases with different curation standards, and no single resource allowed researchers to ask, in one place, which genes are shared across hallmarks, which are tissue-specific, or which hallmark processes are enriched in a given disease. AgingHallmarksDB was designed to close exactly that gap.

Technically, the platform integrates gene sets from seven sources, including the Human Ageing Genomic Resources, the Aging Atlas, the Digital Ageing Atlas, Open Genes, AgingReG and CSGene, harmonizing identifiers through the HGNC gene nomenclature resources. From the merged compendium of 3,111 genes, the team derived a high-confidence consensus set of 1,089 genes supported by multiple databases, giving users a choice between broad coverage and stricter evidence. Each gene is annotated with tissue expression from the Human Protein Atlas, cell-type class information, and links to exosomal cargo catalogued in ExoRBase, reflecting the growing recognition that extracellular vesicles carry aging-related signals between tissues. Protein-protein interactions and transcriptional regulatory relationships, drawn from resources such as the HuRI reference interactome, OmniPath and Reactome, complete the molecular wiring diagram.

The analytical heart of the platform is a pair of standard but powerful statistical tools: over-representation analysis (ORA) and gene set enrichment analysis (GSEA). ORA asks whether a user-supplied list of genes contains more members of a particular aging hallmark than would be expected by chance, using a false discovery rate correction to control for multiple testing. GSEA, by contrast, takes a ranked list, for example every gene in a transcriptome ordered by how much its expression changed under some condition, and asks whether hallmark genes cluster disproportionately at the top or bottom of the ranking. Together these methods let researchers translate raw gene lists and expression datasets into statements about which aging processes are actually at work.

One of the first things the team did with their own resource was to quantify how interconnected the hallmarks really are. Measuring gene overlaps and Jaccard similarity, a metric that scores the size of the shared gene set relative to the union of two sets, they found substantial crosstalk between hallmark categories, confirming what many biologists have suspected: the hallmarks are not independent modules but overlapping facets of a shared underlying biology. Network analysis reinforced the point. When hallmark gene pairs were mapped onto the human protein-protein interaction network, the pairs turned out to be directly connected far more often than chance would predict, suggesting that hallmark genes physically cooperate in molecular machines.

Topology analysis of that interaction network also surfaced something potentially more consequential: a set of key age-associated genes that act as central hubs across multiple aging hallmarks. In network language, hubs are nodes with disproportionately many connections, and they often represent proteins whose perturbation has outsized effects on the system. The identification of such multi-hallmark hub genes provides a systems-level shortlist for experimentalists interested in interventions that might influence several aging processes at once, a long-standing goal in the search for geroprotective therapies.

To demonstrate the platform’s disease relevance, the team performed hallmark enrichment analyses on gene sets associated with aging-related and non-aging-related diseases. The contrast was striking. Diseases known to be age-associated showed extensive enrichment across the hallmarks, while developmental and congenital diseases, which are not aging-related, showed only limited enrichment, an internal validation that the hallmark gene sets capture something specific about aging biology rather than disease generally. Within individual diseases, the analyses recovered biologically meaningful associations: loss of proteostasis emerged as a highly significant hallmark in Alzheimer’s disease, consistent with the well-documented accumulation of misfolded and aggregated proteins in the Alzheimer’s brain, and chronic inflammation ranked as a top hallmark in atherosclerosis, echoing a large body of work implicating inflammatory pathways in plaque formation and rupture.

These disease-hallmark associations were then stress-tested with network proximity analysis, a technique from network medicine that measures how close two sets of genes sit within the human interactome. If the hallmark genes for a given process and the genes associated with a given disease occupy nearby positions in the interaction network, that offers evidence of shared molecular machinery beyond simple gene overlap. The proximity results supported the enrichment findings, adding an independent line of evidence that the disease-hallmark links recovered by AgingHallmarksDB reflect real biological relationships. The code for this analysis is publicly available through the project’s GitHub repository, and the database itself is freely accessible at cb.imsc.res.in/aginghallmarksdb.

Perhaps the most topical demonstration involves air pollution. The team applied GSEA to a skin transcriptome associated with exposure to PM2.5, the fine particulate matter under 2.5 micrometers in diameter that is a major component of urban air pollution and an established driver of skin injury and premature skin aging. The analysis found significant enrichment of aging hallmarks in the polluted-skin gene expression signature, with epigenetic alterations standing out as particularly significant. That result aligns with earlier experimental work showing that particulate matter induces senescence in skin keratinocytes through oxidative stress-dependent epigenetic modifications, and it shows how the database can reframe an environmental exposure dataset in the language of aging biology, a capability with obvious implications for dermatology, toxicology and the growing market for interventions against extrinsic skin aging.

The broader significance of AgingHallmarksDB lies in what it makes routine. A researcher with a gene list from a CRISPR screen, a proteomics experiment or a patient cohort can now, in minutes, ask which hallmarks of aging are over-represented, which of their genes sit in the high-confidence consensus set, in which tissues and cell types those genes act, and how far their disease genes lie from hallmark genes in the interactome. The authors position the resource as a framework for investigating hallmark-associated functionality and for potential therapeutic discovery in aging and longevity, and the timing is apt: the field has been actively debating its foundational principles, with recent surveys revealing both broad engagement with the hallmark framework and genuine disagreement about its boundaries. Tools that make the framework computationally explicit, testable and comparable across datasets are exactly what such debates need. As populations worldwide age and the World Health Organization pushes healthy-aging agendas, resources that connect genes, hallmarks, tissues and diseases into one navigable map are likely to become standard infrastructure for the next generation of geroscience.

Subject of Research: An integrated database resource for systems-level analysis of aging hallmarks and associated genes

Article Title: An integrated resource for systems-level analysis of aging hallmarks and associated genes

Article References: Tiwari, R., Balaji, M., Chivukula, N., Sil, P., & Samal, A. (2026). An integrated resource for systems-level analysis of aging hallmarks and associated genes. Biogerontology, 27(5), Article 173. https://doi.org/10.1007/s10522-026-10520-2

Image Credits: AI Generated

DOI: 10.1007/s10522-026-10520-2

Keywords: aging hallmarks, AgingHallmarksDB, biogerontology, gene set enrichment analysis, over-representation analysis, protein-protein interaction network, network medicine, cellular senescence, epigenetic alterations, longevity, PM2.5, disease enrichment

The smoke tree is a striking shrub best known for the wispy, smoke-like plumes that give it its common name, but its heartwood has long attracted attention from natural products chemists. Heartwood extracts of C. coggygria are rich in flavonoids, a sprawling family of plant secondary metabolites that includes many compounds with documented anticancer activity in laboratory settings. Earlier work had established that butin, butein, and sulfuretin exhibit cytotoxic effects against leukemia cells, yet the molecular mechanisms underlying these effects in K562 cells—a widely used model of chronic myeloid leukemia—remained poorly defined. The new study set out to close that gap by treating the cells with the purified compounds as well as with crude heartwood extracts prepared with ethanol, water, and a methylene chloride/methanol mixture.

The experimental design combined classical pharmacology with modern genomics. Flow cytometry was used to determine how the treatments altered cell cycle distribution and whether they triggered the activation of apoptosis, the programmed cell death pathway that many anticancer therapies seek to engage. To pin down the mode of cell death, the researchers employed caspase inhibition, a technique that reveals whether cell death depends on the caspase enzymes that execute apoptosis. The centerpiece of the study, however, was the transcriptomics: three independent mRNA sequencing experiments were performed on treated K562 cells, and the resulting data were subjected to differential gene expression analysis followed by pathway-level analyses using the Reactome and KEGG databases through gene set enrichment analysis, or GSEA. Network and pathway-cluster analyses then tied the individual enriched pathways together into coherent biological stories.

The headline finding is that all three flavonoids push leukemia cells in a broadly similar direction—toward growth suppression, a metabolic downshift, and the activation of adaptive stress-response networks—but each compound leaves its own distinctive fingerprint on the transcriptome. Butein emerged as the most transcriptionally aggressive of the three, inducing 809 differentially expressed genes. Butin followed with 439 differentially expressed genes, and sulfuretin produced 299. Despite these differences in breadth, the team identified 134 differentially expressed genes shared across all three treatments, suggesting a common core response to this family of flavonoids that may underlie their shared antileukemic activity.

Butin’s transcriptomic signature tells a story of systematic shutdown. Treatment with this compound produced a transcriptional shift marked by the suppression of RNA processing, mRNA splicing, translation initiation, and mitochondrial bioenergetics. In practical terms, the compound appears to choke off the leukemia cell’s ability to read out its genetic information, mature its messenger RNAs, and manufacture new proteins, while simultaneously starving the cell’s power plants. The enrichment analysis pointed specifically to downregulation of oxidative phosphorylation, the OXPHOS system that generates most of a cell’s ATP, and of the tricarboxylic acid cycle, or TCA cycle, which feeds that system. For rapidly dividing cancer cells, which depend heavily on coordinated protein synthesis and mitochondrial metabolism, this dual hit represents a serious metabolic and biosynthetic crisis.

Butein, the most potent transcriptional perturber, induced a strikingly coordinated downregulation of many of the same programs. Mitochondrial respiration, RNA maturation, mRNA maturation, and mRNA transport were all suppressed after butein treatment, indicating that the compound disrupts both the energy supply and the information-processing pipeline of the leukemic cell. The overlap between butin and butein is notable from a chemical biology standpoint: both compounds share a core flavonoid scaffold, and their convergent effects on mitochondrial and RNA-related pathways suggest that these processes may be particularly vulnerable targets for this class of molecules. At the same time, the fact that butein altered nearly twice as many genes as butin hints at additional, compound-specific activities that remain to be fully characterized.

Sulfuretin stood apart. Instead of the broad biosynthetic and metabolic shutdown seen with its two chemical cousins, sulfuretin displayed a distinct pattern characterized by positive enrichment of stress, inflammatory, immune, and proteasome-related pathways. In other words, treated cells mounted what looks like an adaptive stress response, ramping up programs associated with inflammation and protein quality control. Alongside this activation, the analysis revealed negative enrichment of RB1/E2F-associated G1/S transition programs—a reference to the molecular circuitry that governs the transition from the G1 phase of the cell cycle into DNA synthesis. Suppressing this program suggests that sulfuretin may hold leukemia cells at the G1 checkpoint, preventing them from committing to another round of cell division.

These findings matter because they move the field beyond the simple observation that a plant compound kills cancer cells in a dish. By resolving the transcriptomic architecture of the response, the study offers mechanistic hypotheses that can be tested directly: whether the compounds induce reactive oxygen species through mitochondrial dysfunction, whether the proteasome activation seen with sulfuretin reflects a genuine unfolded protein burden, and whether the shared 134-gene core response contains biomarkers that predict sensitivity to flavonoid treatment. The authors also emphasize that the framework they have built is not limited to leukemia. Because the transcriptomic signatures point to fundamental processes—protein synthesis, mitochondrial respiration, cell cycle control—their approach can be translated to other types of malignancy, where the same compounds might exploit different vulnerabilities depending on the tissue of origin.

The study also illustrates the power of pathway-level analysis over simple gene lists. Differential expression analysis alone can identify thousands of changed transcripts, but GSEA against curated databases such as Reactome and KEGG, combined with network and pathway-cluster analyses, revealed that the changed genes were not scattered at random. Instead, they converged on coherent biological modules—RNA processing, translation initiation, OXPHOS, TCA cycle, G1/S transition, proteasome function—that together explain how a flavonoid treatment can translate into cytotoxicity. The team’s careful handling of technical variation, documented in supplementary analyses showing sample clustering before and after batch correction, adds confidence that the observed signatures reflect biology rather than experimental noise.

For now, the work remains at the level of cell culture, and the usual caveats apply: concentrations that are cytotoxic in a dish do not automatically translate into safe and effective drugs, and the pharmacokinetics of flavonoids in the human body are notoriously complex. But the study delivers something that earlier cytotoxicity screens could not—a transcriptomic framework for further investigation of the antileukemic properties of butin, butein, and sulfuretin. As interest in natural products as sources of anticancer lead compounds continues to grow, the smoke tree’s heartwood chemistry has just become considerably more interesting, and the road from Balkan shrub to molecular medicine has gained a well-marked first milestone.

Subject of Research: Transcriptomic effects of Cotinus coggygria flavonoids on K562 myeloid leukemia cells

Article Title: Transcriptome analysis of K562 myeloid leukemia cells treated with butin, butein, and sulfuretin from the heartwood of Cotinus coggygria Scop.

Article References: Transcriptome analysis of K562 myeloid leukemia cells treated with butin, butein, and sulfuretin from the heartwood of Cotinus coggygria Scop.. (n.d.). https://doi.org/10.1186/s12906-026-05570-5

Image Credits: AI Generated

DOI: 10.1186/s12906-026-05570-5

Keywords: Cotinus coggygria, flavonoids, butin, butein, sulfuretin, K562 leukemia cells, transcriptome analysis, RNA sequencing, gene set enrichment analysis, apoptosis, mitochondrial respiration, natural products

The research team, led by Pernilla Wikström and Anders Bergh, focused on a concept they call tumor-instructed normal tissue, or TINT. The idea is that tumors are not passive invaders but active communicators, secreting signals that reprogram neighboring cells to create a supportive environment for growth and invasion. To test how far this instruction reaches and what it depends on, the researchers analyzed gene expression in tissue samples from 44 men with unifocal prostate cancer who had undergone surgical removal of the prostate. For each patient, they sampled the tumor itself, benign tissue taken close to the tumor, and benign tissue taken at a distance from it, all preserved as formalin-fixed paraffin-embedded samples and profiled using Clariom D microarrays.

The results were striking. The transcriptome — the complete set of RNA messages produced by the cells — in tumor-adjacent benign tissue differed significantly from that in cancer-free biopsy control tissue. Even more revealing, the expression patterns in this benign tissue tracked three distinct variables: the distance from the tumor, the tumor’s grade as judged by its aggressiveness under the microscope, and the tumor’s molecular subtype. In other words, a pathologist looking at seemingly normal tissue could, in principle, read out molecular information about a tumor they might not even be sampling directly.

To dissect what these molecular changes meant functionally, the team used gene set enrichment analysis, a computational method that tests whether predefined groups of genes — so-called hallmark gene sets representing known biological processes — are coordinately upregulated or downregulated in a given sample. They compared hallmark pathway activity in both tumor and TINT tissue across tumor grade, contrasting high-grade with low-grade cancers, and across molecular subtype, contrasting Luminal B with Luminal A tumors, a classification borrowed from systems that categorize prostate tumors by their cellular lineage and proliferative character.

A clear pattern emerged. Hallmarks related to cell proliferation were enriched in both the tumor and the adjacent benign tissue of high-grade and Luminal B tumors, and the researchers confirmed this at the protein level: immunohistochemical staining for Ki67, a well-established marker of cell division, was increased in the benign tissue near aggressive tumors. Signs of increased metabolism also appeared in both tumor and TINT tissue associated with high grade and the Luminal B subtype. This suggests that the metabolic reprogramming long recognized as a hallmark of cancer cells extends into the surrounding normal tissue, which may be co-opted to support the tumor’s energy demands.

The immune landscape told a different and more nuanced story. Rather than a simple gradient fading with distance, certain immune response programs were uniquely enriched in the benign tissue of high-grade tumors and, separately, in the benign tissue of Luminal A tumors. Critically, these immune signatures were more pronounced in tissue taken close to the tumor than in tissue taken farther away, indicating that the tumor’s immunological influence on its surroundings is both subtype-specific and spatially graded. This spatial dimension matters, because it implies that the tumor actively instructs its immediate microenvironment rather than merely reflecting a whole-organ field effect.

Perhaps the most clinically provocative part of the study involved biopsies. The researchers compared gene expression in tumor-negative biopsy samples from men who were later diagnosed with high-grade prostate cancer — samples that had been declared cancer-free at the time — with tumor-negative biopsies from men who remained cancer-free, and with biopsies that did contain cancer. The later-cancer group showed enrichment of metabolic gene programs compared with the controls, even though no clear proliferative or immunological differences were detectable. This hints that subtle molecular changes in apparently cancer-free tissue might foreshadow the presence or future emergence of aggressive disease elsewhere in the prostate.

That possibility points toward a genuinely new class of biomarkers. Today, prostate cancer diagnosis relies on needle biopsies that sample only a tiny fraction of the gland, and a negative biopsy does not rule out cancer hiding elsewhere. If the molecular state of benign tissue in a biopsy can indicate that a high-grade tumor is present in unsampled regions, pathologists could gain a powerful adjunct to conventional assessment. The study’s authors suggest that TINT alterations hold potential as diagnostic and prognostic biomarkers, and the grade- and subtype-specific signatures they identified could eventually help stratify patients without requiring the tumor itself to be captured by the biopsy needle.

The work also reframes how scientists think about tumor-host interactions. Much of cancer research focuses on the tumor cell itself or on the stromal and immune cells immediately within the tumor mass. The TINT concept pushes the boundary outward, treating the entire neighborhood of the tumor as a dynamically negotiated territory. The finding that proliferation, metabolism, and immune programs in benign tissue vary systematically with tumor proximity, grade, and subtype suggests a structured dialogue between cancer and host tissue — one that could be interrupted. If specific signaling pathways mediate the instruction of normal tissue, blocking them might deprive tumors of support from their surroundings, opening new avenues for therapeutic intervention.

Caveats remain, as they do in any study of this kind. The cohort comprised 44 men with unifocal disease treated at a single institution, and the biopsy analysis, while suggestive, identified metabolic enrichment rather than definitive predictive signatures. The authors themselves frame the findings as a basis for further exploration rather than a finished clinical tool. Still, the message is compelling: the benign tissue around a tumor is not a silent bystander but a molecular archive of the tumor’s identity and behavior. Learning to read that archive could change how prostate cancer is detected, graded, and ultimately treated, turning the tissue that pathologists currently dismiss as normal into one of the most informative samples in the gland.

Subject of Research: Gene expression changes in tumor-adjacent benign prostate tissue in relation to tumor proximity, grade, and molecular subtype

Article Title: Gene expression profiles in benign prostate tissue reflect tumor proximity, grade, and molecular subtype

Article References: Wikström, P., Freyhult, E., Semenas, J., Lundholm, M., Josefsson, A., Bergström, S. H., & Bergh, A. (2026). Gene expression profiles in benign prostate tissue reflect tumor proximity, grade, and molecular subtype. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-09035-8

Image Credits: AI Generated

DOI: 10.1186/s12967-026-09035-8

Keywords: prostate cancer, tumor-instructed normal tissue, TINT, gene expression profiling, transcriptomics, tumor microenvironment, molecular subtype, Luminal A, Luminal B, biomarkers, Ki67, gene set enrichment analysis

The research team, led by Hedda Fjell Scheel of the Norwegian University of Life Sciences together with colleagues at Oslo University Hospital and the University of Oslo, took advantage of a publicly available dataset of whole transcriptome sequencing data from canine mammary tumors. After quality control and the removal of animals with incomplete survival records, the final cohort comprised 146 dogs aged between two and nineteen years, representing twenty different breeds. Thirty-eight of these dogs had a recorded time of death after surgery, ranging from one day to 709 days with a median of 205.5 days, while the remaining 108 were right censored, meaning they were still alive at their last follow-up. This substantial censoring proportion, roughly 74 percent of the cohort, posed a genuine statistical challenge that shaped every subsequent analytical decision.

The investigators built three competing Cox proportional hazards survival models. The first, a clinical model, used only variables measurable through standard veterinary workup. The second, a gene expression model, relied exclusively on standardized, pre-filtered transcriptomic features. The third combined both data types. Because the gene expression matrix contained 18,579 genes, vastly more predictors than the 38 survival events available, the team applied careful feature pre-filtering: they retained the 1,000 most variable genes by mean absolute deviation, removed highly correlated pairs above a threshold of 0.95, and selected genes significant in univariate Cox models at a false discovery rate below 0.1. This left forty candidate genes, none of which belonged to the PAM50 subtyping panel used in human breast cancer. Regularized elastic net regression with cross-validated tuning of the alpha and lambda parameters then selected the most informative features for the gene-based models.

Model performance was evaluated with Uno’s C-index, a censoring-robust concordance measure in which 0.5 represents random prediction and 1.0 perfect discrimination, and with time-dependent ROC-AUC at six months, one year, and two years after surgery. Across 100 iterations of Monte Carlo cross-validation with stratified train-test splits of 4:1, the results were strikingly clear. The clinical model achieved a median C-index of 0.646, marginally ahead of the gene expression model at 0.635 and the combined model at 0.621, though none of these differences reached statistical significance. More tellingly, at the clinically critical timepoints of six months and one year after surgery, the clinical model significantly outperformed both transcriptomic models, with a median ROC-AUC of 0.811 at six months compared with 0.727 for the gene expression model and 0.768 for the combined model.

What emerged as decisive were three humble, readily obtainable clinical variables. Stepwise regression favored a multivariate model containing age, lymphatic invasion assessed by tumor histology, and estrogen receptor status evaluated by immunohistochemistry. Lymphatic invasion exerted the largest effect on the hazard of death, confirming its previously reported prognostic power, while estrogen receptor positivity was associated with improved survival, mirroring the pattern well established in human breast cancer. Notably, histopathological grade and molecular subtype, both significant in univariate analyses, dropped out of the multivariate model, likely because they were strongly correlated with lymphatic invasion and estrogen receptor negativity and therefore added no independent information. The final clinical model explained 17.8 percent of the overall variation in survival.

The molecular subtyping analysis itself yielded biologically meaningful results. Using the single-sample MPAM50 classifier, every tumor in the cohort was assigned to either the luminal A or the basal-like subtype, with luminal A accounting for 58.2 percent of samples. Subtype was significantly associated with age, breed, lymphatic invasion, grade, and estrogen receptor status. Basal-like tumors were more frequently grade 2 or 3, more often showed lymphatic vessel invasion, and were overrepresented among estrogen receptor negative tumors, with 69.6 percent of the 23 receptor-negative tumors classified as basal-like. Dogs with luminal A tumors were on average one year younger, and Maltese dogs were more likely to harbor luminal A tumors. These findings indicate that the canine PAM50 subtypes capture tumor biology beyond what estrogen receptor status alone reveals, echoing their prognostic significance in human patients.

Perhaps the most intriguing result came when the researchers used the best-performing clinical model to assign each dog a risk score and split the cohort at the median into high-risk and low-risk groups of 73 dogs each. Survival differed significantly between the groups, with a log-rank p-value of 0.0036. Differential gene expression analysis then uncovered 433 genes significantly differentially expressed between the groups, 279 upregulated in high-risk tumors and 154 in low-risk tumors. Gene set enrichment analysis using the Hallmark collection from the Molecular Signatures Database revealed twenty-four significantly enriched pathways. High-risk tumors were enriched for signatures of proliferation, including E2F targets and the G2M checkpoint, and for metabolic programs such as oxidative phosphorylation and glycolysis, while low-risk tumors showed enrichment of immune-related pathways, including TNF alpha signaling, IL6-JAK-STAT signaling, inflammatory response, and interferon alpha response.

That immune activity marks the low-risk group is a finding with real translational resonance. In human breast cancer, both innate and adaptive immune components are associated with recurrence-free survival, and immune gene expression carries prognostic weight particularly in hormone receptor negative tumors. The canine data suggest that the immune system plays an equally central role in anti-tumor response across species, reinforcing the value of dogs as a comparative model for human breast cancer research. The two diseases share spontaneous onset, histological subtypes, underlying genetic alterations, and gene expression changes, yet molecular markers that transformed human breast cancer management, from PAM50 subtyping to gene expression signatures like MammaPrint, have never been incorporated into routine canine staging. This study represents one of the most systematic attempts to close that gap.

Why, then, did gene expression fail to improve prediction? The authors offer several candid explanations. Adding thousands of transcriptomic features to a small, heavily censored dataset may introduce more noise than signal, a phenomenon documented in large-scale benchmark studies showing that multi-omics survival models are widely sensitive to noise. Linear Cox models may also miss non-linear interactions between genomic and clinical features. Moreover, with only 38 events, statistical power is inherently limited, and a larger number of events might reveal differences that this cohort cannot detect. None of the models exceeded a time-independent C-index of 0.65, which, while comfortably above random, falls short of what would be needed for clinical implementation, though comparable multi-omics studies in humans rarely exceed 0.7 either.

The practical message for veterinary oncology is nonetheless concrete. Estrogen receptor status measured by immunohistochemistry and lymphatic invasion assessed by routine histology could be valuable additions to current TNM staging, potentially refining survival predictions and informing decisions such as whether ovariohysterectomy at the time of tumor removal might benefit a particular patient, since dogs with high serum estradiol and receptor-positive tumors appear more likely to benefit from that intervention. The authors also emphasize the limitations of using overall survival rather than cancer-specific survival in an elderly cohort, and they call for future studies to record cause of death, complete TNM parameters, and richer genomic data. Alternative modeling approaches such as boosting methods and random survival forests may yet extract more from transcriptomic data. For now, the study stands as a sobering and instructive reminder that in the era of big data, the humble microscope slide and the immunohistochemistry stain still hold their ground.

Subject of Research: Machine learning-based identification of prognostic factors for survival in canine mammary gland tumors

Article Title: Unraveling prognostic factors in canine mammary gland tumors using machine learning

Article References: Scheel, H. F., Zobolas, J., Lien, T. G., Lingaas, F., & Bergholtz, H. (2025). Unraveling prognostic factors in canine mammary gland tumors using machine learning. Veterinary Oncology, 2(1), Article 16. https://doi.org/10.1186/s44356-025-00030-7

Image Credits: AI Generated

DOI: 10.1186/s44356-025-00030-7

Keywords: canine mammary tumors, machine learning, Cox regression, gene expression, estrogen receptor, lymphatic invasion, survival prediction, TNM staging, PAM50 subtyping, gene set enrichment analysis, veterinary oncology, precision medicine

The core problem the researchers set out to solve is a bottleneck familiar to anyone working in targeted protein degradation. Proteomics-based approaches can, in principle, reveal every protein whose abundance changes when a compound is applied, but they are expensive and slow, making them impractical as a first-pass screen for large chemical libraries. Target-based biochemical assays such as AlphaLISA or TR-FRET require purified proteins and, crucially, prior knowledge of which target-E3 pair to test—an assumption that defeats the purpose of hunting for glues against unknown or classically undruggable substrates like transcription factors and scaffold proteins. Phenotypic screens, meanwhile, are biologically rich but cannot easily distinguish a true degrader from an ordinary inhibitor, forcing laborious downstream mechanistic deconvolution. What the field needed was a cheap, fast, and mechanistically informative primary readout.

The team’s answer was to make the transcriptome itself the universal readout. Gene expression signatures are high-dimensional proxies for cellular state, a principle demonstrated at scale by the L1000 platform and made affordable by DRUG-seq, which skips RNA purification and uses direct cell lysis with multiplexed library preparation. The researchers employed an optimized version, DRUG-seq2, which they report offers superior sensitivity for low-abundance regulatory transcripts—a critical feature when the signals of interest come from immature, weak molecular glues. In their workflow, Hep3B liver cancer cells were seeded in 96-well plates, treated with compounds at a uniform concentration of 1 micromolar for 24 hours, and processed through the miniaturized sequencing protocol. The library comprised 117 compounds: 11 commercial molecular glues serving as validated benchmarks and 106 in-house candidates synthesized by two collaborative research groups, all rationally designed to modulate the cereblon (CRBN) E3 ligase.

A deceptively simple but essential innovation lies in how the baseline is defined. Because DRUG-seq2 relies on direct lysis, technical noise can swamp the subtle transcriptional ripples produced by nascent glues. Before any analysis, the team performed systematic pairwise Pearson correlation analysis across all DMSO vehicle wells on each plate, excluding any replicate with a correlation coefficient below 0.95. By constructing the reference transcriptome solely from high-consensus control wells, they established a robust, self-consistent zero-point that maximizes statistical power to resolve low-magnitude perturbations. Downstream, differential expression analysis with DESeq2—flagging genes with absolute log2 fold change greater than 1 and adjusted p-value below 0.05—was paired with gene set enrichment analysis (GSEA) built from six curated collections, including Hallmark, Reactome, KEGG, Gene Ontology Biological Process, C6, and the C3 transcription factor target set. This two-layer design means even compounds with few or zero individual differentially expressed genes can still betray coordinated pathway-level activity.

The mechanistic heart of the platform is a genetic benchmark: an isogenic pair of Hep3B cell lines, one wild-type and one in which CRBN has been knocked out via CRISPR-Cas9. The researchers established a stable Cas9-expressing parent line, introduced single-guide RNAs targeting CRBN, and isolated single-cell clones. Clone 3 emerged as the definitive knockout after rigorous validation. Treatment with CC-90009, a clinical molecular glue that degrades the translation termination factor GSPT1, provided the proof: in wild-type cells the drug rapidly and almost completely eliminated GSPT1 protein, while in the knockout the effect vanished entirely, along with the drug’s cytotoxic potency, which showed a pronounced right-shift in the IC50 curve. Any compound whose transcriptional signature collapses in the knockout is therefore flagged as CRBN-dependent—a hallmark of a genuine glue—whereas signals that persist likely reflect off-target toxicity, conventional inhibition, or hijacking of a different E3 ligase such as VHL or DCAF16.

Screening the full library revealed a strikingly polarized activity spectrum. Roughly 20 percent of the compounds were transcriptionally inert, producing zero differentially expressed genes under the test conditions. At the other extreme, the high-activity cluster was dominated by known glues: SJ3149, which targets CK1α, elicited the most profound response, followed by lenalidomide, BMS-986397, and a MYC degrader. For most of these, transcriptional activity was almost completely abolished in the CRBN knockout, exactly as expected for classic glues. Encouragingly, one in-house compound, LC-02-047-P1, landed among the potent performers, marking it as a promising novel candidate. Conversely, compounds such as LC-02-105 and LC-02-033 displayed a paradoxical knockout-amplified phenotype, indicating CRBN-independent mechanisms and allowing the team to filter out molecules whose intrinsic scaffold toxicity overwhelmed any degradation function.

The pathway-level analysis added mechanistic texture that raw gene counts alone could not provide. Potent GSPT1 degraders such as CC-90009 upregulated regulators of the integrated stress response—DDIT3 (CHOP), GDF15, TRIB3, DDIT4, and PPP1R15A—consistent with the biology of translation termination failure: ribosomes stall at stop codons, collide, and trigger the ZAKα-mediated ribotoxic stress response, which in turn activates ATF4 and suppresses the translational apparatus to mitigate proteotoxic stress. GSEA of CC-90009 confirmed enrichment of integrated stress response, proteasome, and apoptosis pathways alongside downregulation of translation machinery. LC-02-047-P1, by contrast, showed metabolic rather than stress-related regulatory characteristics, hinting that its degraded target might be an upstream switch governing multiple metabolic pathways. The enrichment layer also rescued latent hits: LC-02-051 performed modestly in the differential expression analysis but ranked near the top in pathway enrichment, and its signal was completely blocked by CRBN knockout—suggesting a genuine but weak glue worth optimizing. Even CC-92480 yielded an instructive anomaly, showing slightly expanded enrichment in knockout cells, which the authors attribute to the absence of its high-affinity target IKZF1 in Hep3B, pushing the compound toward non-canonical substrates or generic xenobiotic stress.

To translate these multidimensional data into decisions, the researchers built a scatterplot integrating transcriptional potency (differential gene counts) with functional breadth (enriched pathway counts) across both wild-type and knockout conditions, then stratified the library into four tiers. Tier one comprises potent glues with high wild-type activity that nearly vanishes upon CRBN loss—the highest priorities for optimization. Tier two captures intermediate and latent leads whose coordinated functional signals suggest seed scaffolds needing structural refinement. Tier three contains active but E3-independent perturbators, likely conventional inhibitors, and tier four collects inert compounds. The spatial logic is intuitive: strong glues such as SJ3149, lenalidomide, CC-90009, and LC-02-047-P1 occupy the upper-right quadrant and collapse toward the origin in the knockout, while pharmacologically inert molecules cluster near the origin in both conditions and target-independent agents stubbornly refuse to move. The thresholds, the authors note, can be customized for different libraries.

Proteomic cross-validation demonstrated that the transcriptomic tiers predict real degradation events. Tandem mass tag quantitative proteomics of LC-02-047-P1-treated Hep3B cells revealed marked downregulation of CSNK1A1 (CK1α), and western blotting across Hep3B, Huh7, and HepG2 hepatocellular carcinoma lines confirmed near-complete, selective CK1α degradation with no effect on GSPT1. A decisive control sealed the case: cells engineered to express a non-degradable GSPT1 mutant abolished CC-90009’s activity but left LC-02-047-P1’s CK1α degradation untouched, proving the candidate acts entirely independently of GSPT1. The weaker tier-two compound LC-02-051 produced only minor proteomic perturbation, with TPM3 changes failing statistical validation—consistent with its latent classification. On scalability, the economics are compelling: the platform costs less than 100 RMB (roughly 14 US dollars) per sample, about 2 percent of standard TMT proteomics, with an end-to-end turnaround of three to four weeks even beyond 1,000 compounds.

The authors are candid about limitations. The single 24-hour window captures an integrated cellular state—primary degradation consequences blended with secondary cascades and stress responses—rather than the immediate transcriptional footprints of neosubstrate depletion, and future iterations will incorporate time-resolved profiling. Transcriptomic signatures also remain an indirect proxy for protein-level events, so candidate leads must ultimately be confirmed by quantitative proteomics and structural biology demonstrating direct ternary complex formation. The choice of screening cell line matters too, since E3 and substrate abundance dictate sensitivity; the authors suggest abundance-matched models, such as VHL-high or DCAF15-high backgrounds, for targeted campaigns. Even so, the GPS platform fills a genuine gap, bridging high-volume chemical screening and low-throughput mechanistic validation, and offering the field a practical roadmap for converting molecular glue discovery from a lucky accident into an engineering discipline.

Subject of Research: High-throughput transcriptomic screening for the discovery of CRBN-dependent molecular glue degraders

Article Title: A scalable, high-throughput glue perturbation screening platform for molecular glue discovery

Article References: Huang, D., Shuang, Z., Chen, L., Ouyang, H., Yang, P., Huang, L., Lu, W., Shi, M., Ding, X., Jiang, B., & Wu, W. (2026). A scalable, high-throughput glue perturbation screening platform for molecular glue discovery. iScience, 29(10), Article 117600. https://doi.org/10.1016/j.isci.2026.117600

Image Credits: AI Generated

DOI: 10.1016/j.isci.2026.117600

Keywords: molecular glue degraders, targeted protein degradation, CRBN, DRUG-seq2, transcriptomics, gene set enrichment analysis, CRISPR knockout, E3 ubiquitin ligase, proteomics, drug discovery, GSPT1, CK1α