Bariatric Surgery Rewrites the Genetic Script of Fat Tissue, but a Stubborn Molecular Memory Remains

When a person with severe obesity undergoes metabolic bariatric surgery, the transformation is dramatic and fast: body weight falls, insulin sensitivity climbs, blood lipids normalize, and type 2 diabetes often melts away. Yet the deepest story of that recovery may not be written on the bathroom scale at all. It is written inside the fat tissue itself, in the thousands of genes that switch on and off as the body sheds its excess. A new study published in the International Journal of Obesity has now tracked that genetic script in unprecedented detail, comparing the transcriptomes of subcutaneous and visceral adipose tissue in women with severe obesity before and after surgery, and against a group of metabolically healthy lean controls. The findings reveal both a striking partial reversal of obesity-driven gene expression and a stubborn residue of molecular change that refuses to fully reset, offering a mechanistic explanation for why surgery works so well and why some metabolic dysfunction may linger even after the pounds are gone.

The research team, led by Prince Dadson and Vesa M. Olkkonen alongside senior investigator Pirjo Nuutila at the Turku PET Centre in Finland, enrolled 42 women with severe obesity scheduled for metabolic bariatric surgery and 15 sex-matched healthy women without obesity as controls. During the operation, surgeons collected excisional biopsies from two very different fat depots: subcutaneous adipose tissue, the layer just beneath the skin of the anterior abdominal wall around the umbilical region, and visceral adipose tissue, taken from the omentum deep within the abdomen. Six months after surgery, the researchers returned to collect a second subcutaneous biopsy from each patient. Total RNA was extracted from all samples using Tri-Reagent and profiled with a custom Illumina targeted RNA expression panel run on the MiSeq system. The panel was deliberately designed to capture the biological processes known to go awry in obesity, including inflammation, insulin resistance, apoptosis, lipid and fatty acid metabolism, cholesterol and bile acid metabolism, energy and glucose metabolism, inflammasome activation, and RNA processing. In total, 203 genes were analyzed, with differentially expressed genes identified using the ROTS R package, a reproducibility-optimized statistical tool, and significance controlled at a false discovery rate below 0.05 using the Benjamini-Hochberg procedure.

The first major finding concerns the fundamental divide between the two fat depots in people with severe obesity. Subcutaneous adipose tissue, the more metabolically adaptive of the two, showed significantly higher expression of genes related to adipocyte function and insulin signaling, including adiponectin, the classic insulin-sensitizing hormone produced by fat cells. It also carried elevated levels of anti-inflammatory markers such as interleukin 1 receptor antagonist and interleukin 10, genes involved in extracellular matrix remodeling like matrix metalloproteinase 9, secreted phosphoprotein 1, and the NLRP3 inflammasome, and a suite of lipid metabolism genes including AGPAT2, DGAT2, ELOVL1, ELOVL5, FADS3, lipoprotein lipase, and perilipin 1. Pre-surgery subcutaneous adipocytes were also significantly larger than their visceral counterparts, measuring 101.78 versus 91.46 micrometers on average. Visceral adipose tissue, by contrast, was a very different molecular landscape, enriched in genes driving inflammation, complement activation, stress signaling, and cell death, including interleukin-18, interleukin 1 receptor 1, PYCARD, MICA, complement component C3, FAS, TNFRSF1A, TLR4, and TRADD. Visceral fat also showed higher expression of lipid biosynthesis genes such as AGPAT9, DHCR7, FADS1, FADS2, and SCD, along with RNA processing regulators like HNRNPA1, HNRNPH1, and SRSF7, suggesting more active post-transcriptional control of gene expression in the depot most strongly linked to metabolic disease.

When the researchers compared the subcutaneous transcriptome of people with severe obesity to that of lean controls, a complex picture of stress, inflammation, and metabolic dysregulation emerged. Genes promoting chronic low-grade inflammation and immune response were upregulated in the obese subcutaneous fat, including NFKB1, a master regulator of inflammatory signaling, PYCARD, which drives inflammasome activation and caspase-1-dependent inflammatory cell death, MMP9, TNFSF10, and TNFRSF1A. Pro-apoptotic markers FAS and BAD were elevated, alongside antioxidant defenses such as SOD2 and HMOX1, indicating both tissue remodeling and active protection against oxidative damage. Cholesterol handling genes were also retuned: ABCA1, which promotes cholesterol efflux, was upregulated, as was NR1H3, the Liver X Receptor Alpha nuclear receptor involved in lipid clearance pathways. Several genes for lipid processing and turnover, including ELOVL1, AGPAT1, FADS3, and HSD17B4, were upregulated, while the expression of LIPE and PNPLA2, the two major lipases involved in triglyceride breakdown and fatty acid release, was elevated, a hallmark of insulin resistance. Crucially, the intrinsic de novo lipogenesis program was suppressed in obese subcutaneous fat, with key fatty acid synthesis genes ACACA, FASN, ACLY, SCD, FADS1, and ELOVL6 all downregulated, along with the sterol synthetic enzyme HMGCS1 and cholesterol transport genes ABCG5 and NPC1L1. The data suggest that in severe obesity, subcutaneous fat cells lose their normal capacity to synthesize and store fat from scratch, while simultaneously ramping up inflammatory signaling, oxidative stress responses, and compensatory cholesterol management efforts.

One of the most unexpected threads in the study involves RNA processing, a layer of gene regulation that operates after DNA has been transcribed into messenger RNA. In the subcutaneous fat of people with severe obesity, several regulators of post-transcriptional RNA processing were upregulated compared to controls, including CELF1, MBNL1, HNRNPD, HNRNPK, HNRNPA1, SFPQ, SRSF1, SRSF7, and SF3A1. These genes control the splicing, maturation, transport, and stability of messenger RNA molecules, effectively determining which protein variants get made from a given gene. The researchers interpret this as an adaptive retuning of gene expression under metabolic stress, a mechanism by which fat cells attempt to cope with the obesogenic environment by altering their protein output at the post-transcriptional level. Intriguingly, the splicing factor SRSF10 showed significant correlations with clinical metabolic parameters: in subcutaneous fat, its expression correlated positively with fasting glucose, fasting insulin, and HOMA-IR, a measure of insulin resistance, while in visceral fat it correlated with the pro-inflammatory cytokine MCP-1. This suggests that RNA processing regulators are not merely bystanders but may be functionally linked to the comorbidities of obesity, a finding the authors say warrants further investigation.

Six months after metabolic bariatric surgery, the subcutaneous transcriptome had shifted substantially toward a healthier state. Genes associated with normal lipogenesis and adipogenesis were upregulated compared to pre-surgery biopsies, including the lipogenic transcription factors SREBF1 in two variants, along with FASN, ELOVL6, and SCD. This indicates a partial recovery of the normal lipid storage function that was suppressed in obesity, consistent with prior studies showing that weight loss restores lipogenic messenger RNAs and improves insulin sensitivity. At the same time, several key genes related to cholesterol biosynthesis and metabolism were downregulated after weight loss, including SREBF2, SQLE, and SCP2, indicating that the elevated cholesterol synthesis seen in obese subcutaneous fat declines once the excess weight is gone. The pro-inflammatory component IL1R1 was downregulated, consistent with reduced inflammatory signaling, and SOD2, the antioxidant enzyme that had been elevated to protect against oxidative stress, was suppressed, reflecting reduced oxidative damage and improved metabolic health. Splicing factors SRSF2 and HNRNPK also showed reduced expression after surgery, suggesting a dialing back of the maladaptive RNA regulation seen in obesity. Leptin, hormone-sensitive lipase, AGPAT3, SFPQ, and TXNIP were additionally downregulated, along with ELOVL1, ADIPOR2, HSD17B4, and PPAP2A.

Cross-referencing the transcriptomic findings with the team’s earlier proteomic analysis of the same cohort revealed that two genes, FASN and SCP2, were differentially expressed at both the messenger RNA and protein levels, with FASN higher and SCP2 lower after surgery-induced weight loss, consistent with their established roles in adipose tissue lipid metabolism. Several additional genes, including SCD, LIPE, LPL, SOD2, and GPAM, were detected in the proteomic dataset, and while the protein-level differences did not reach statistical significance, SCD and LPL protein levels tended to be higher after surgery while LIPE, SOD2, and GPAM tended to be lower, directionally consistent with the messenger RNA changes identified in the present study. This convergence of transcriptomic and proteomic evidence strengthens the argument that the observed gene expression changes are functionally meaningful rather than statistical noise.

Yet the story does not end with full recovery. When the researchers compared the post-surgery subcutaneous transcriptome to that of lean controls, a substantial proportion of the obesity-associated transcriptional changes persisted. A Venn diagram analysis revealed an overlap of 17 out of 61 differentially expressed genes between the during-surgery-versus-control and after-surgery-versus-control comparisons, indicating that many obesity-driven changes in gene expression remain even after significant weight loss. Several lipid metabolism, cholesterol transport, and efflux genes remained at higher levels than controls after surgery, including ADIPOR1, ABCA1, SCARB1, AGPAT1, GPAM, LPL, ELOVL3, FADS1, SCP2, SREBF2, HMGCS1, LIPE, and PNPLA2. Stress-response and ketone body metabolism genes HSD17B4, HMOX1, and BDH1 also remained elevated. Inflammatory and immune response genes NFKB1, PYCARD, FAS, TNFRSF1A, TNFSF10, TLR4, ITGAM, CD68, BAD, IL32, SOD2, and CPPED1_1 persisted at higher levels than controls, as did RNA processing and transport genes HNRNPK, CELF1, MBNL1, SF3A1, SRSF1, SRSF7, TRA2B, and SMC1A. In contrast, ABCG5, OSTalpha, SLC10A1, and SLCO1B3 showed lower expression after surgery compared to controls. The persistence of these changes echoes the concept of an obesogenic metabolic memory in adipose tissue, a phenomenon previously demonstrated in murine models where weight loss improved most metabolic impairments but left residual issues such as glucose intolerance, hyperinsulinemia, and mild liver steatosis, with adipocytes retaining upregulated inflammatory and extracellular matrix remodeling pathways while key metabolic pathways remained downregulated.

The study has several limitations that the authors acknowledge. The cohort consists entirely of women, which limits generalizability to men and precludes sex-stratified analyses. The study population is heterogeneous, including participants with impaired fasting glucose, impaired glucose tolerance, and type 2 diabetes before and after surgery. The small sample size limited reliable comparisons between the two surgical techniques used. Post-surgery biopsies were collected while participants were still losing weight and in a catabolic state, which may have affected metabolic and transcriptomic outcomes. Visceral biopsies could only be obtained during surgery for ethical reasons, so no post-surgery visceral data were available. All participants underwent a four-week very low calorie diet before surgery, which may have introduced bias when comparing baseline profiles with controls who did not follow the diet, although the roughly 7 percent weight loss from the diet was modest compared to the 10 to 14 percent losses in prior studies that showed larger mitochondrial gene expression changes. The targeted transcriptomics approach also severely biases traditional pathway analyses, so the authors omitted pathway analysis and instead discussed the functions of robustly affected genes based on the literature characterizing their roles.

Taken together, these findings deepen the understanding of the molecular distinctions between subcutaneous and visceral adipose tissue in severe obesity and illustrate how metabolic bariatric surgery can partially reverse obesity-associated transcriptional alterations in fat. Subcutaneous fat emerges as the more metabolically adaptive depot, with enhanced storage capacity and anti-inflammatory potential, while visceral fat is characterized by inflammation, stress responses, impaired fat storage, and altered RNA regulation, changes that likely contribute to its stronger association with metabolic disease. Surgery pushes subcutaneous fat back toward a healthier transcriptional state by restoring lipogenic and adipogenic gene expression and dialing down inflammatory, cholesterol biosynthesis, and splicing-factor programs. But the residual molecular signature that persists after weight loss, including ongoing extracellular matrix remodeling, elevated inflammatory signaling, and altered RNA processing, highlights the need for targeted therapeutic strategies to more fully restore adipose tissue function and improve long-term metabolic outcomes. The results point to specific molecular pathways, from cholesterol handling to RNA splicing, that may serve as targets for future interventions aimed at completing the recovery that surgery begins.

Subject of Research: Genetics

Article Title: Adipose tissue transcriptome before and after surgery: effects on genes driving lipid metabolism, inflammation, and RNA processing

Article References: Dadson, P., Jaakkola, M. K., Sun, L., Klén, R., Suomi, T., Vaittinen, M., Anastasiou, I. A., Honka, H., Immonen, H., Badeau, R. M., Salminen, P., Helmiö, M., Roivainen, A., Pekkarinen, L., Kalliokoski, K. K., Rebelos, E., Olkkonen, V. M., & Nuutila, P. (2026). Adipose tissue transcriptome before and after surgery: effects on genes driving lipid metabolism, inflammation, and RNA processing. International Journal of Obesity. https://doi.org/10.1038/s41366-026-02243-8

Image Credits: AI Generated

DOI: 10.1038/s41366-026-02243-8

Keywords: bariatric surgery, adipose tissue, transcriptomics, obesity, lipid metabolism, inflammation, RNA processing, metabolic memory, gene expression, International Journal of Obesity, Adipose, tissue