Does intensity matter? Insight from divergent adipose tissue molecular responses following exercise

Adipose tissue is increasingly recognized as a dynamic, metabolically active endocrine organ that responds acutely to exercise, influencing systemic insulin sensitivity, inflammation and energy balance (James & Stanford, 2025). While endurance exercise training improves abdominal subcutaneous adipose tissue (aSAT) (Ahn et al., 2022, 2024), the early-phase molecular responses regarding how exercise prescription variables, such as intensity, shape the aSAT transcriptome remain unknown. A recently published study by Ahn et al. (2025) in The Journal of Physiology investigated the acute molecular responses of aSAT to an acute endurance exercise session in 45 physically active adults. Participants were assigned to low-intensity continuous exercise (LOW; 60 min at 30% V ̇ O 2 peak ${\dot V_{{{\mathrm{O}}_2}{\mathrm{peak}}}}$ ), moderate-intensity continuous exercise (MOD; 45 min at 65% V ̇ O 2 peak ${\dot V_{{{\mathrm{O}}_2}{\mathrm{peak}}}}$ ), or high-intensity interval exercise (HIGH; 10 × 1 min at 90% V ̇ O 2 max ${\dot V_{{{\mathrm{O}}_2}{\mathrm{max}}}}$ with 1 min active recovery), with aSAT biopsies collected prior to and 1.5 h after exercise for bulk RNA sequencing and targeted immunoblotting. HIGH preferentially upregulated genes associated with angiogenesis, protein secretion and insulin signalling, whereas MOD and LOW upregulated transcripts involved in extracellular matrix (ECM) remodelling, ribosome biogenesis and oxidative phosphorylation. Despite these exercise intensity-specific transcriptomic responses, all exercise modalities elicited similar changes in extracellular signal-regulated kinase (ERK) phosphorylation, core circadian clock gene expression and circulating cytokine levels. Cell-type deconvolution revealed heterogeneous responses across adipocyte and stromal vascular fractions. These findings characterize early, intensity-dependent transcriptional and signalling responses in human aSAT, which may represent initiating events underpinning long-term adaptations to exercise training. The distinct molecular signatures across the acute exercise bouts highlight the potential independent role of exercise intensity in shaping tissue remodelling, providing a mechanistic basis for precision and individualized exercise prescription (Fig. 1). The experimental design implemented by Ahn et al. (2025) is methodologically rigorous and conceptually innovative, providing valuable insights into how varying exercise intensities influence molecular responses in aSAT. Nonetheless, several methodological considerations warrant further discussion. First, the study employed a between-subjects design, wherein each participant completed only one of the three exercise intensity conditions. While this approach facilitates efficient group-level comparisons, gene expression in human aSAT might be susceptible to inter-individual variability influenced by factors such as genetic profile, acute nutritional status, circadian rhythms, sleep quality, and habitual physical activity. A within-subject crossover design, with sufficient washout periods, could potentially mitigate biological noise and enhance statistical power, particularly in transcriptomic analyses. Second, despite efforts to balance demographic and physiological characteristics across groups, exercise load and duration were not standardized. 60-min LOW and 45-min MOD involved continuous steady-state exercise, whereas 20-min total of HIGH consisted of intermittent high-intensity intervals, resulting in substantial variations in total energy expenditure, with MOD expending approximately 400 kcal more than LOW and HIGH (∼250 kcal each). These disparities may confound interpretation, as observed molecular response differences may reflect differences in the relative level and timing of energy turnover or metabolic stress, rather than exercise intensity alone. A notable methodological approach to address the above confounding factors is the single-leg exercise model (Bergström & Hultman, 1966), which allows for within-subject comparisons under tightly controlled workloads. By enabling each participant to serve as their own control, this model effectively reduces genetic variability introduced by inter-individual differences and enhances statistical power. Moreover, it permits precise matching of total energy expenditure between distinct exercise prescriptions, isolating the effects of variables such as intensity or contraction mode. Collectively, the single-leg exercise model offers a robust framework for isolating the independent effects of specific exercise variables such as intensity or contraction mode under matched total workload, duration or energy expenditure. Third, exercise intensity was prescribed using relative physiological markers (e.g. % V ̇ O 2 peak ${\dot V_{{{\mathrm{O}}_2}{\mathrm{peak}}}}$ ). Although widely used, these markers are influenced by individual variability and may not elicit homogeneous physiological responses across participants, even when expressed as the same percentage of each participant's maximum. A more precise method would be to prescribe intensity based on individualized metabolic thresholds, such as lactate threshold and ventilatory threshold, as recommended by a recent consensus on exercise intensity (Bishop et al., 2025). Additionally, by collecting an aSAT biopsy sample before and 1.5 h after exercise, the study may capture some of the early transcriptional changes that occur in response to exercise. However, collecting only this single aSAT sample after exercise would miss potentially important transcriptional events that may occur earlier or later than 1.5 h after exercise. Transcriptomic and proteomic responses are known to follow highly dynamic, non-linear temporal patterns, with gene expression and signalling cascades peaking at varying time points depending on the tissue and stimulus (Egan & Sharples, 2023). Accordingly, future studies should incorporate multi-time-point sampling and/or integrative multi-omics approaches to construct a more comprehensive temporal profile of human aSAT adaptations to acute exercise. Lastly, although several ideal design strategies were discussed, such as within-subject crossover designs, single-leg models, individualized intensity thresholds and serial biopsies, each presents practical feasibility challenges. Crossover designs require careful washout periods and impose repeated biopsy burdens; single-leg protocols enable precise workload control but are technically demanding and may limit generalizability; threshold-based intensity offers more precision but requires additional testing resources; and multiple time-point biopsies raise ethical, logistical and participant compliance concerns. Even with these enhancements, transient molecular responses may still be missed. Thus, future studies should balance scientific rigor with feasibility, integrating selective design elements, minimally invasive sampling and multi-omics approaches to optimize design details. One of the major strengths of this study lies in its direct sampling of aSAT biopsies, providing tissue-specific molecular data. This approach offers significantly greater accuracy and specificity compared to relying solely on circulating biomarkers, which reflect systemic signals and may not capture depot-specific adaptations. Importantly, the authors acknowledged a key limitation of bulk RNA-sequencing, namely, its inability to distinguish transcriptional contributions from heterogeneous cell populations within the analysed tissue. To address this, they implemented an advanced bioinformatic strategy, cell-type enrichment and deconvolution, using previously published single-cell RNA-sequencing reference datasets to estimate changes in cellular composition post-exercise. While this method represents a state-of-the-art computational solution, it remains an inferential tool rather than a direct measurement. The validity of its outputs depends heavily on the quality, resolution and demographic compatibility of the reference datasets used. Although the authors employed age-matched reference data, the conclusions drawn from this analysis are less robust than those derived from direct cell-type quantification techniques, such as histology or flow cytometry. Additionally, the targeted immunoblotting employed in the study was limited to a pre-selected panel of proteins. While informative for validating specific molecular pathways and hypotheses, this narrowed focus may have excluded key changes in other proteins and/or phosphorylation events. As a result, the protein-level insights presented may not fully reflect the broader landscape of post-exercise signalling in aSAT. To overcome this limitation, future investigations could adopt global phosphoproteomic profiling to achieve a more comprehensive and unbiased characterization of post-translational signalling networks influenced by acute exercise intensity (Hoffman et al., 2025). While plasma glycerol and non-esterified fatty acid concentrations were used to infer systemic lipolytic activity, these circulating markers represent the aggregate outcome of whole-body lipolysis and fatty acid uptake across all tissues, including contracting skeletal muscle. Abdominal depots are the predominant source of circulating fatty acids during lipolytic stimulation (Nielsen et al., 2004), supporting the physiological relevance of studying aSAT. However, given the localized molecular focus of this study, caution is warranted when interpreting systemic lipid markers as direct surrogates for depot-specific responses. There is currently no definitive evidence that lipolytic rates in aSAT mirror those of other depots or the systemic average during exercise. Therefore, any link drawn between circulating lipid metabolites and tissue-specific transcriptional and/or protein signalling responses might be correlative rather than causative. Lastly, Ahn et al. (2025) provide mechanistic insight into how exercise intensity modulates the aSAT response. Consistent with previous evidence that exercise modulates adipose tissue remodeling, the results show increased vasculature, an altered ECM collagen profile, and elevated abundances of ribosomal, mitochondrial, and lipogenic proteins in aSAT (Ahn et al., 2022, 2024). Meanwhile, the study's new findings demonstrate these processes have a distinct acute response based on intensity. These findings not only link early transcriptional and signalling events to the long-term structural and endocrine remodelling of aSAT during exercise training, but also underscore that exercise intensity and other prescription variables (such as volume and type) might serve as precise molecular switches that selectively engage different response pathways. This intensity-dependent molecular stratification provides a mechanistic rationale for individualized and precision-based exercise prescriptions, supporting the concept that tailored modulation of intensity might optimize aSAT response, and it provides the basis for early signalling to repeated exercise training metabolic adaptation. This study offers novel insight into how exercise intensity acutely programs the transcriptomic landscape of aSAT, revealing that different intensities elicit distinct molecular signatures. Practically, these findings suggest that exercise intensity is not merely a quantitative variable, but a biological modulator of aSAT remodelling. High-intensity exercise enhances angiogenic and inflammatory signalling, while low-to-moderate intensities favour mitochondrial, translational and extracellular matrix pathways. Despite minimal overlap in differentially expressed genes, consistent shifts in circadian gene expression across intensities point to a shared core regulatory mechanism. This divergence in transcriptomic response highlights the potential for tailoring exercise prescriptions to elicit targeted tissue adaptations, supporting a precision exercise medicine approach. Moving forward, future studies should validate these early molecular signatures through longitudinal, standardized exercise interventions, ideally using within-subject crossover designs to reduce biological variability. Multi-omics and multi-time-point sampling will be key to capturing the dynamic remodelling of aSAT. Expanding research to other depots (e.g., visceral fat) and diverse populations and linking molecular changes with clinical endpoints such as tissue perfusion, insulin sensitivity and inflammation will enhance translational relevance. In addition to intensity, investigating differences in acute aSAT responses across various exercise types, such as resistance, aerobic exercise, accumulated exercise and combined approaches (Yin et al., 2025), will be critical to delineate mode-specific adaptations and signalling pathways. In parallel, developing personalized exercise prescriptions that account for individual variability and combining exercise with nutritional and behavioural strategies may further optimize metabolic outcomes and tissue-specific adaptations. Together, these efforts can inform precision exercise approaches to improve adipose tissue function and long-term cardiometabolic health. In summary, the study demonstrates that exercise intensity might distinctly influence the early transcriptomic responses of human aSAT. These findings suggest that intensity is a critical variable in exercise prescription, with the potential to differentially influence adipose remodelling and systemic metabolic health. Future work using longitudinal, standardized study designs and follow-up mechanistic investigations will be essential to link these acute signals to long-term adaptations. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article. No competing interests declared. M.Y. conceived the study and drafted the manuscript. B.Z., K.X., and J.W. contributed to manuscript revision and provided critical feedback. All authors contributed to the interpretation of findings. All authors have read and approved the final version of the manuscript and agree to be accountable for all aspects of the work, ensuring that any questions related to its accuracy or integrity are appropriately investigated and resolved. All individuals listed as authors meet the criteria for authorship, and no one who qualifies for authorship has been omitted. M.Y. is supported by an Australian Government Research Training Program (RTP) Scholarship (doi.org/10.82133/C42F-K220). We sincerely thank Prof. Yongming Li for their valuable suggestions and revisions.