Epigenetic alterations of AKT1 orchestrate a metabolic reprogramming in advanced lipedema: translational insights from an integrated multi-omics study

Abstract

Background

lipedema is a chronic, progressive adipose disorder predominantly affecting women, characterized by painful, symmetrical subcutaneous fat accumulation, and typically resistant to lifestyle interventions. The pathophysiology of advanced-stage lipedema remains poorly defined, and no validated biomarkers or targeted therapies are currently available.

Methods

in this observational study, we applied a comprehensive multi-omics approach to dissect the molecular and metabolic alterations underlying late-stage lipedema.

Results

Genome-wide DNA methylation profiling identified over 5,000 differentially methylated CpG sites affecting genes involved in receptor tyrosine kinase signaling, phospho-metabolism, and immune pathways. Transcriptomic analysis revealed profound downregulation of mitochondrial functions, including oxidative phosphorylation, the TCA cycle, and fatty acid β-oxidation, alongside disruption of the sirtuin pathway and extracellular matrix remodeling. Integrative analysis pinpointed AKT1 as a central regulatory node: its promoter region was hypomethylated, correlating with increased gene expression and protein phosphorylation. Metabolomic profiling confirmed AKT1-linked metabolic dysregulation, including altered levels of L-arginine, NADP+, ATP, guanosine, glycerol, and glutamate, indicating impaired redox balance and energy metabolism. Trans-omic network analysis positioned AKT1 at the intersection of multiple dysregulated pathways, suggesting its key role in advanced-stage lipedema.

Conclusions

the consistent enhancing of AKT pathway signaling across omic layers highlights its potential not only as a biomarker for disease stratification but also as a putative druggable target for therapeutic intervention. These findings offer new mechanistic insights into lipedema pathophysiology and provide a rationale for future personalized treatment strategies guided by AKT1-centric molecular profiling.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07726-w.

Highlights

Epigenetic profiling reveals AKT1 promotes hypomethylation and overexpression in advanced-stage lipedema.

Transcriptomic analysis shows mitochondrial dysfunction, with downregulation of OXPHOS, TCA cycle, and β-oxidation genes.

Untargeted metabolomics identifies AKT1-associated alterations in redox balance, amino acid metabolism, and energy production.

Trans-omic network analysis positions AKT1 as a central regulatory hub linking epigenetic, transcriptional, and metabolic changes.

Integrated multi-omics profiling supports AKT1 as a candidate biomarker and therapeutic target in advanced-stage lipedema.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07726-w.

Background

Lipedema is classified as a chronic disorder of loose connective tissue, predominantly affecting women and characterized by marked alterations in adipose tissue structure and consistency in the affected regions [1, 2]. Core clinical features include pain, swelling, and a tendency to bruise easily, typically involving the arms and legs [3]. In its advanced stages, lipedema can lead to severe impairment of mobility and significant reductions in quality of life [1]. In the 11th revision of the International Classification of Diseases (ICD-11), the World Health Organization designated lipedema as “Lipoedema” under the code EF02.2 [4]. Although common, the condition is frequently misdiagnosed, with an estimated prevalence of up to ~ 10% in women [5–7]. In its 2023 research roadmap, the Lipedema Foundation highlighted the “lack of validated, clearly druggable targets” as a critical unmet need and called for urgent scientific action [8].

Liposuction remains the most widely used intervention for lipedema, offering effective plastic correction, although it may negatively impact adipose tissue integrity over time [9, 10]. Understanding of the disease aetiology remains limited, partly due to the involvement of multiple cell and tissue types and the heterogeneous presentation across clinical stages [11, 12]. Disease onset is commonly associated with puberty and pregnancy, suggesting a prominent role of sex hormones such as estrogen and cyclical changes in intestinal permeability [13–15]. A hereditary component has also been proposed, with a likely autosomal dominant transmission pattern exhibiting sex-specific differences [16]. The progressive nature of lipedema has led to its classification into three clinical stages [17]. In stage 1, the skin surface appears normal, while the hypodermis is enlarged. In stage 2, the skin becomes uneven, and larger, non-encapsulated fat deposits develop. In stage 3, patients exhibit pronounced deformities, particularly of the thighs, due to massive adipose protrusions. Although lymphedema may co-occur at any disease stage, it is most observed at stage 3. Allen and Hines hypothesized that fluid accumulation results from the low resistance of expanded adipose tissue to hydrostatic pressure from capillaries to the interstitium [18].

Adipokines are key regulators of biological processes such as inflammation and fibrosis, and their secretion is influenced by age, body mass index (BMI), and menopausal status—factors that may contribute to lipedema development [19]. Among them, adiponectin and leptin are the most studied; their levels typically correlate negatively and positively with BMI, respectively [20]. While adiponectin functions as an insulin sensitizer with anti-inflammatory and anti-fibrotic effects, leptin promotes energy homeostasis but also exerts pro-inflammatory and pro-fibrotic actions [21].

Recently, Straub et al. conducted a pioneering multi-omics investigation of lipedema adipose tissue, integrating transcriptomics, proteomics, and epigenomics to identify disease-specific molecular signatures in early-stage patients [22]. Their study provided foundational evidence supporting the distinct molecular identity of lipedema. However, their analysis focused primarily on early disease stages and did not functionally resolve upstream regulatory drivers or their metabolic implications, particularly in advanced lipedema [22]. The pathophysiology of advanced-stage lipedema remains poorly characterized, with no validated biomarkers or therapeutic targets currently available for clinical use [23]. In this observational study, we applied an integrative multi-omics approach to explore the molecular and metabolic landscape of late-stage lipedema in female patients [24]. By combining genome-wide DNA methylation profiling, transcriptomic analysis of subcutaneous adipose tissue, and untargeted NMR-based metabolomics, we aimed to uncover defining molecular signatures of disease progression. We performed a cross-platform analysis to identify epigenetically regulated pathways and metabolite profiles associated with lipedema severity. Differentially methylated positions and gene expression data were used to draw information concerning functional enriched pathways via annotation analyses that were integrated with metabolomic findings to reveal coordinated alterations in key bioenergetic and amino acid metabolic signaling [25, 26]. Our goal was to identify key molecular drivers underlying the transcriptomic and metabolic reprogramming in advanced-stage lipedema, with a particular focus on epigenetic mechanisms and their potential as therapeutic targets.

Methods

Population study

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki. The study protocol was reviewed and approved by the ethics committee of the Land Salzburg. All study participants gave written informed consent and were informed about the study (EK Nr: 1023/2024). The study is designed to be an experimental non-therapeutic biomedical study investigating the molecular basis of lipedema. All operative interventions to collect tissue sample were conducted in the clinic Dr. Papp & Papp in Salzburg, Austria.

Description of two cohorts and criteria

The study included two cohorts, each consisting of relatively equal groups of female patients, a cohort of lipedema patients and a healthy control group.

The inclusion criteria for the study participants were explicitly chosen:

• 19 to 70 years old;

• Women with Lipedema stage III (cohort 1) and without lipedema who are receiving an alternative surgical procedure, as liposuction and biopsy (cohort 2).

The exclusion criteria for the study participants were:

• Patients under 18 years of age or did not consent to the study or wanted to withdraw at any given time of the study;

• Female with other lipid disorder;

• Use of any immunosuppressive or corticosteroid pharmacotherapy.

The conventional surgical procedure employed to remove lipedema tissue is liposuction from the extremities (arms and legs), with reductive plastic surgery of the extremities being utilised in exceptional cases of severe lipedema. Tissue samples from non-lipedema patients were collected from other types of reductive surgical operations from body regions such as breast (breast reduction surgery) or abdomen (abdominoplastic surgery).

Tissue sampling

The period of sample collection was about 12 months from 03/24 until 03/25.

Subcutaneous adipose tissue samples were collected from female patients grouped into two cohorts, one lipedema cohort and a control-cohort. The tissue samples were taken during operations and placed directly into sterile Eppendorf plastic cryogenic vials or aspiration syringes, which were at once flash-frozen in a container of liquid nitrogen and transferred to a − 80 °C freezer. In an Eppendorf Cyrocube© freezer, the samples were stored for a couple of weeks until transported to the laboratory facilities on dry ice.

DNA extraction and Genome-wide methylation array

Genomic DNA was extracted from approximately 100 mg of adipose tissue (n = 8; 4 lipedema and 4 non-lipedema controls) using a combination of mechanical disruption and commercial kits. Briefly, frozen tissue samples were pulverized in liquid nitrogen using a mortar and pestle, followed by DNA isolation with the Genomic DNA Isolation Kit (Norgen Biotek, Cat. 24700), according to the manufacturer’s instructions. The extracted DNA was subsequently cleaned and concentrated using the DNA Clean & Concentrator-5 kit (Zymo Research, Cat. D4003) following the recommended protocol. DNA purity was assessed using a NanoDrop™ 2000/2000c spectrophotometer (Thermo Fisher Scientific), while DNA concentration was determined with the Qubit™ Fluorometer (Life Technologies, Monza, Italy) using the Quant-iT™ dsDNA High Sensitivity Assay Kit. For each sample, 500 ng of genomic DNA was subjected to bisulfite conversion using the EZ DNA Methylation-Gold™ Kit (Zymo Research), as previously described [23]. Subsequently, 250 ng of bisulfite-converted DNA was hybridized to the Infinium MethylationEPIC v2.0 BeadChip array (Illumina, Cat. 20087706), which interrogates approximately 930,000 unique methylation sites. BeadChips were scanned with the Illumina iScan platform. The genome-wide methylation array data are available in the EBI ArrayExpress database (http://www.ebi.ac.uk/arrayexpress) with accession number E-MTAB-15,365.

Protein extraction, Western blot and antibodies

Total proteins extraction was performed as described in An et al. [24]. Protein concentrations were determined using Bradford assay and their expression was analysed by western blotting. Briefly, protein sample extracts were denatured and separated on 10% polyacrylamide, 0.1% SDS (SDS-PAGE), and transferred onto a nitrocellulose blotting membrane (GE Healthcare, Milan, Italy). Following blocking with 5% nonfat dry milk in TBST buffer (0.01 m Tris-HCl, pH 8.0, 0.15 m NaCl, and 0.1% Tween 20), membranes were immunoblotted with different primary antibodies. In details, the antibodies used for western blot experiments were Pan-Akt Polyclonal Antibody (E-AB-30471, Elabscience), Phospho-Pan-Akt (Ser473) Polyclonal Antibody (E-AB-20802, Elabscience) and anti-α-Tubulin (sc-32293, Santa Cruz Biotechnology). Then, primary Abs were detected by horseradish peroxidase-conjugated secondary Abs (GE Healthcare) and revealed by chemiluminescence and autoradiography. Densitometry was performed by ImageJ software analysis [25].

RNA isolation and RNA sequencing analysis

Total RNA was isolated from the same subset of adipose tissue samples used for genome-wide DNA methylation analysis (n = 8; 4 lipedema and 4 non-lipedema controls). Tissues were snap-frozen in liquid nitrogen at once after collection and stored at − 80 °C until further processing. Prior to RNA extraction, samples (~ 150 mg each) were cryopulverized using the CP02 cryoPREP^® Automated Dry Pulverizer (Covaris). RNA was extracted using TRIzol™ Reagent (Invitrogen, #15596026) according to the manufacturer’s protocol. Briefly, 1 ml of TRIzol was added to ~ 150 mg of pulverized adipose tissue, followed by homogenization and phase separation. RNA was quantified with Qubit 2.0 fluorimeter using Qubit RNA HS assay kit (Termo Fisher Scientifc, USA), and the assessment of nucleic acids integrity (RNA Integrity Number) was performed with Agilent 4150 TapeStation System (Agilent Technologies, USA). Indexed libraries were prepared starting from 200 ng total RNA according to TruSeq Stranded Total RNA Library Prep Gold (Cat. 20020599, Illumina, San Diego, California, USA) and sequenced on the Novaseq 6000s4 v 1.5 platform (Illumina Inc.) using 2 × 100 bp paired end mode as previously described [26, 27]. RNA sequencing primary data have been deposited in the EBI ArrayExpress database (http://www.ebi.ac.uk/arrayexpress) with accession number E-MTAB-15,370.

Metabolic extraction and NMR metabolomics

Adipose tissue samples were collected according to standard operating procedures (SOPs) [28]. Metabolic extraction was carried out using a two-step protocol. To each tissue sample collected from the limbs, weighing between 40 and 60 mg, 1 mL of H2O and 2 mL of cold methanol were added with the aim of halting enzymatic metabolism (quenching). Subsequently, the samples were subjected to vortexing to ensure uniform dispersion of the material and were then sonicated for 15 min, utilising an on/off mode under temperature-controlled conditions of 3 °C [28]. The amplitude was set at 23% (referred to as 23 A.U. in the ultrasonicator operating system) to facilitate complete cell lysis and tissue homogenization. Following this, 4 mL of methyl-t-butyl ether (MTBE) was added initially. After maintaining the samples on dry ice for one hour, 2 mL of water was added to promote the separation of polar and apolar phases [28, 29]. Subsequently, all samples were subjected to centrifugation at 1500 rpm for 15 min at 4 °C, facilitating a clear delineation of the two phases: the upper (lipophilic) phase, which contains the apolar metabolites, and the lower (hydrophilic) phase, which comprises the polar metabolites of interest. Three millilitres were taken from each phase, and the extraction was repeated on the original samples that still contained the pellet, adding four millilitres of MTBE and two millilitres of H₂O. The samples were mixed by vortexing and then subjected to centrifugation at 1000 rpm for 5 min at 4 °C to re-establish separation of the two phases. Afterwards, 1 ml was taken from each phase and added to the respective phases collected during the previous extraction. After removing the solvents from the polar phases using an SP Genevac EZ-2 series 4.0 centrifugal evaporator, the lyophilised extracts were resuspended in 500 µL of a tissue-specific phosphate buffer containing 50 mM Na₂HPO₄, 1 mM 2,2,3,3-d₄-trimethylsilyl-propionic acid, sodium salt (TSP-d4), 50 µL of D2O, and 2 mM sodium azide (NaN3) as an antibacterial agent. The resultant samples were ultimately transferred into 5-mm NMR tubes for spectra acquisition utilising ¹H-NMR spectroscopy. TSP-d4 at a concentration of 0.1% in D2O served as an internal standard for the alignment and quantification of NMR signals [30].

NMR spectra acquisition and assignment

¹H NOESY spectra acquisition was conducted with a spectral width of 12 ppm, 20,000 data points, presaturation during the relaxation delay, and a mixing time for water suppression along with gradient spoilage, a 5-second relaxation delay, and a mixing time of 10 ms [31–33]. Topspin version 3.0 (Bruker Biospin) was utilised for spectrometer control and data processing. Analysis of the NMR spectra was conducted using a targeted metabolomic approach. Consequently, each metabolite was identified prior to statistical analysis with Chenomx NMR-Suite v8.0 software (Chenomx NMR suite, v8.0, Edmonton, AB, Canada), which combines advanced analysis tools with a compound library. Quantitative analysis of the NMR spectra was conducted using NMRProcFlow, and the obtained data matrix was subjected to statistical analysis [34].

NMR statistical analysis

Partial least squares discriminant analysis (PLS-DA) was performed on the normalised dataset using MetaboAnalyst 6.0 (http://www.metaboanalyst.ca/). The model derived from the supervised methodology underwent validation through the 10-fold cross-validation technique, considering the Q2 and R2 indices as well as accuracy. The contribution of individual variables to the separation of clusters was assessed using Variable Importance Projection (VIP), with only those metabolites exhibiting VIP values greater than 1.3 being significant [35, 36].

Univariate analysis was performed on the adipose tissue metabolomic profiles using the T-test and Fold Change, with the results displayed in a Volcano plot, setting the threshold at p-value < 0.05 and fold change ± 1 [37]. The hierarchical clustering analysis was performed using MetaboAnalyst 6.0, taking into account the median metabolomic profile to provide a comprehensive depiction of the quantitative variations in metabolites and to ascertain the similarities between the adipose tissue metabolomic profiles obtained from various anatomical sites. Heatmaps were generated from normalized data, average group concentration, and Euclidean distances [38]. The meta-analysis, conducted with Metaboanalyst 6.0, employed integrated principal component analysis (iPCA) using both the sites of sampling for pathological tissue (leg, arm) and healthy tissue (abdomen and breast) as discriminatory variables, in addition to the presence or absence of lipedema [39]. Pathway Topology analysis was carried out using MetPa. Pathways with more than two Hits, i.e. metabolites belonging to the biochemical pathway and a p-value of less than 0.05, were deemed significant. To assess the impact of individual pathways on the examined clusters, the Pathway Impact (PI) value was calculated. This was achieved by combining path centrality and enrichment results [40].

Bioinformatics analysis

Infinium MethylationEPIC v2.0 analysis was performed using Chip Analysis Methylation Pipeline (ChAMP) package [41, 42]. The analysis was performed by comparing 4 lipedema tissues with 4 control tissues. Preprocessing, normalization and filtering. Raw IDAT files were imported in R using champ.load with default settings. During data loading, ChAMP applies standard sample- and probe-level QC filters, including removal of probes failing the detection p-value criterion (i.e., probes with detection p-value > 0.01 are excluded; detPcut = 0.01). Samples with an excessive fraction of failed probes were discarded using the default cutoff (SampleCutoff = 0.1), and probe filtering was then applied to the retained samples. Probes with low beadcount (< 3 beads) in at least 5% of samples were removed (default settings). Non-CpG probes were excluded. SNP-affected probes were filtered using the recommended annotations described by Zhou et al., and multi-hit/cross-reactive probes were removed based on published lists implemented in ChAMP (e.g., Nordlund et al.). Probes mapping to sex chromosomes (X and Y) were excluded (default; filter XY).

Differential methylation analysis was performed within ChAMP and p-values were adjusted for multiple testing using the Benjamini–Hochberg procedure. CpGs were considered differentially methylated if they met both an adjusted p-value ≤ 0.1 and an effect-size threshold |Δβ| ≥ 0.15. The |Δβ| cutoff was used to prioritize changes more likely to be biologically meaningful and is commonly adopted in array-based methylation studies; in our dataset, it also corresponded approximately to the first quartile of the |Δβ| distribution.

Genomic annotation of CpGs was performed using the information available in Infinium MethylationEPIC v2.0 manifest file. In detail, transcriptional start site (TSS)200 refers to CpGs between 0 and 200 bases upstream of the TSS; TSS1500 refers to CpGs between 200 and 1500 bases upstream of the TSS; 5’UTR refers to those GpGs within the 5’ untranslated region, between the TSS and the ATG start site; gene body refers to CpGs between the ATG and stop codon, regardless of the presence of introns and exons. Promoter region includes TSS1500, TSS200, 5’UTR, and 1st exon regions. Functional annotation analyses were performed using Enrichr tool [43].

For RNA-Seq data analysis was performed as previously described [26].

In brief, raw sequencing reads were assessed for quality using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/), and adapter sequences were trimmed using Trimmomatic [44]. Reads were aligned to the human genome reference (GRCh38.p13, Gencode Release 41) using STAR v2.7.10a with default parameters [45]. Transcript quantification was carried out using FeatureCounts, and differential expression analysis was performed using DESeq2 [46, 47]. The analysis compared lipedema versus non-lipedema patients. Transcripts were considered differentially expressed if they showed an absolute fold change (|FC|) ≥ 1.5 and an adjusted p-value ≤ 0.05 (Benjamini–Hochberg correction). Functional enrichment analysis was conducted using Ingenuity Pathway Analysis (IPA, QIAGEN) and Gene Set Enrichment Analysis (GSEA) [48].

Statistical methods

All statistical analyses were performed using R software (version 4.0.2). Data are presented as mean ± standard deviation (SD). Statistical significance was assessed using an unpaired two-sample t-test, unless otherwise specified. For genome-wide methylation array, RNA-Seq analyses and metabolic profiling, specific statistical approaches and thresholds are described in the corresponding Methods sections. For batch/variance associated to methylation and RNA-seq experiments, we performed sample-level QC using per-sample distribution plots and sample–sample correlation matrices to identify outliers and gross technical shifts. For metabolomics, multivariate models were validated by 10-fold cross-validation, and sampling-site effects were assessed using iPCA including site as a factor.

Results

Characteristics of the participants

The anthropometric parameters of participants and the collected tissue regions are presented in Table 1. The average age of the subjects as a collective group was 42.29 (± 13.07) years. The average body weight and BMI were 69.24 (± 9.43) and 24.38 (± 2.98), respectively. Concerning lipedema, all patients were at stage 3.

Table 1.

Characteristics of the participants and anatomical regions of collected tissue samples. Data are presented as mean ± standard deviation (SD) or percentage

Variable	Lipedema group n = 12 (± SD)	No-lipedema group n = 9 (± SD)
Age	41.6 (± 11.3)	43 (± 15.8)
BMI (kg/m2)	24.4 (± 2.6)	24.3 (± 3.6)
Body-region-tissue
Abdomen	-	4 (44.4%)
Arms	6 (50%)	-
Breast	-	5 (55.6%)
Upper leg	6 (50%)	-

All samples were subjected to multi-omics analyses according to the experimental workflow depicted in Fig. 1. We integrated epigenetics, transcriptomics, and metabolomics profiling to garter deep insight lipedema pathophysiology and development.

Fig. 1.

Experimental plan. Schematic overview of the experimental workflow involving multi-omics profiling in lipedema patients and control tissues

Genome-wide methylation profiling highlighted a dysregulated receptor tyrosine kinase and phospho-metabolism signaling pathways in advanced stage lipedema patients

Genome-wide methylation profiling highlighted a dysregulated receptor tyrosine kinase and phospho-metabolism signaling pathways in advanced stage lipedema patients. Epigenetic modifications, able to alter gene expression without changing the underlying DNA sequence, are increasingly recognized to play essential roles in the development and progression of several diseases including possibly lipedema. These modifications, such as DNA methylation, can influence how genes involved in fat metabolism, inflammation, and other relevant pathways are deregulated and thus influence disease initiation and progression. To gain insight into the epigenetic mechanisms underlying lipedema pathogenesis, a genome-wide DNA methylation analysis was performed on subcutaneous adipose tissue samples collected from from non-affected adipose tissue from no-lipedema group two cohorts of patients. QC analyses (MDS/clustering and β-value distributions) did not identify outlier samples or gross technical shifts (Supplementary Fig. S2A, S2B). By using high density DNA methylation array with single CpG site resolution we were able to detect a total 5,484 CpGs plotted in the heatmap reported in Fig. 2A. According to the methylation beta value that clearly distinguished between the two groups, with a statistically significant difference in the methylation level (Δβ ≤-0.15 and Δβ ≥ 0.15; padj ≤ 0.1) between lipedema and healthy tissue; in particular, as reported in the volcano plot in Fig. 2B, we found 3,343 CpG sites are hypermethylated, whereas 2,140 are hypomethylated, respectively (Supplementary Table 1). This analysis revealed a clear pattern of differential methylation across the genome with greatest piking in promotor region and last methylation in the 3′ UTRs (untranslated regions) and gene bodies as shown in Fig. 1C, despite the overall pattern resulted quite similar for both Hyper- and Hypo-methylated lipedema specific CpGs. Given the impact of methylation changes observed we further functionally annotated these profiles to gather insight into deregulated pathways affected by observed methylation changes. As reported in Fig. 2D we interestingly found that methylation profiling significantly impacted receptor tyrosine kinase and phospho-metabolism signaling pathways with also a concomitant action on T-cell activation–related pathways and PD-1 signaling, highlighting the impact that DNA methylation deregulation could have on cell communication and other vital cellular processes like growth, differentiation and metabolism, but also on immune responses, as already hypothesized for the genesis an progression of this disease.

Fig. 2.

Genome-wide DNA methylation profiling in lipedema patients. A) Hierarchical clustering heatmaps of methylation levels CpG sites in healthy control and lipedema expressed as β-value. Color gradient from dark blue (β-value = 1) to white (β-value = 0) represents hypermethylation to hypomethylation. B) Volcano plot depicting differentially methylated CpGs between control and lipedema tissues (|ΔB|≥ 0.15 and padj < 0.1). C) Pie chart showing the distribution of differentially methylated CpGs (Hypermethylated; upper panel and hypomethylated; lower panel) on different element of the transcriptional unit. D) Table showing statistically enriched pathway according to “Enrichr” functional annotation tool considering genes harbouring differentially methylated CpGs. Both nominal p-values and Benjamini–Hochberg adjusted p-values are reported. Pathway results are interpreted as exploratory

Transcriptome profiling of advanced stage lipedema patients confirmed mitochondrial and metabolic disfunctions and highlighted Sirtuin signaling alteration

Together with DNA methylation profile, transcriptome reprogramming has emerged to have a significant on impact cellular behaviour and to contribute to disease development or progression. Thus, the understanding these changes is crucial for the developing of new diagnostic strategy and potential therapeutic approaches. Considering these premises, we employed total-RNA sequencing for gene expression profiling comparing lipedema and healthy patient tissues. Sample-level QC (count distributions and correlation/PCA) did not indicate outliers or obvious batch-driven clustering (Supplementary Fig. S3A-B). As shown in Fig. 3A, we observed a marked deregulation of gene expression program in lipedema samples compared to control tissues that highlighted 668 up- and 266 down-regulated genes, respectively (Supplementary Table 2). Transcriptome deregulation functional annotation (Fig. 3B and C) revealed an impact on oxidative phosphorylation mitochondrial protein degradation as well as TCA cycle and Fatty acid beta oxidation in coherence with already published data [22]. However, our results highlighted also an up-regulation of estrogen signaling as well as an impairment of extracellular matrix organization and sirtuin signaling pathway. This resulted particularly interesting considering possible additional epigenetic influence of lipedema progression.

Fig. 3.

Lipedema transcriptome profiling. A) Volcano plot depicting differentially expressed transcripts between control and lipedema tissues (|FC|) ≥ 1.5 and an adjusted p-value ≤ 0.05. B) Dot plot chart showing statistically enriched pathway according to IPA functional annotation tool according to activation Z-score. C) Statistically significant functions highlighted by Gene Set Enrichment Analysis (GSEA) in lipedema transcriptome

Multi-omics data integration revealed the RAC (Rho family) -alpha serine/threonine-protein kinase AKT1 alteration as novel biomarker for lipedema patients

Multiomics offers a comprehensive perspective on biology, driving breakthroughs in human disease research. By integrating multiple genomics layers like transcriptomics and epigenetics, this approach enables a deeper understanding of gene expression, regulation, and protein activity. In this context we employed an integrative analysis of epigenetic (DNA-methylation) and transcriptomics deregulation in advanced stage lipedema tissues. Results obtained indicate that concerning deregulated pathways, obtained from this integration, there is a major influence on Metabolic pathways, in general, and on Pyruvate metabolism, in particular, that resulted the top enriched one (Fig. 4A). We then restricted our focus to transcriptional charges couplet to concomitant and opposite methylation alterations as shown in Fig. 4B (Supplementary Table 3). These further analyses revealed that among deregulated transcripts the RAC (Rho family)-alpha serine/threonine-protein kinase AKT1 resulted particularly interesting that results significantly increased in lipedema vs. controls (DESeq2: log2FC = 0.765; FC ≈ 1.70; padj = 0.0125). To further strengthen the directionality of the integrative signal, we performed an additional enrichment analysis restricted to genes showing promoter-associated hypomethylation together with increased expression in RNA-seq. This focused subset yielded a highly significant enrichment of metabolic and mitochondrial programs, including metabolism, aerobic respiration and respiratory electron transport and additional pathways related to mitochondrial protein turnover and Complex IV assembly (Supplementary Figure S5). Notably, this analysis also highlighted growth/immune signaling modules linked to AKT biology, including CD28-dependent PI3K–Akt signaling, providing further support for AKT1 as a convergent node emerging from coordinated epigenetic and transcriptional alterations.

Fig. 4.

Multi-Omics profiling integration in lipedema. A) Dot plot chart showing statistically enriched pathway according to “ShinyGO” functional annotation tool considering all differentially methylated CpG and the corresponding differentially expressed transcript obtained as described in methods sections. B) Heatmap depicting differentially expressed CpGs and transcripts between control and lipedema tissues showing opposite behaviours C) Lollipop graph depicting differentially expressed CpGs harboured in AKT1 transcription unit D) Western blot (upper) and box plot (lower) showing AKT1 protein and phosphoprotein levels comparing healthy control and lipedema patients. Band intensities were quantified by densitometry and normalized to α-tubulin (loading control). The box blot shows the average of band intensity relative to control or lipedema

Given the established role of AKT1in regulating cell growth and survival and its involvement in metabolic control (including glucose handling) [52], we investigated whether epigenetic changes could contribute to AKT1 abundance. Consistently, we found a strong hypo-methylated region in the 5’ UTR of the protein able to influence its abundance in lipoedema patients (Fig. 4C). To validate our finding, we performed an analysis of protein expression levels comparing healthy controls and disease samples. Results obtained and showed in Fig. 4D clearly demonstrate an AKT1 and phospho-AKT1 increase in lipedema tissues highlighting this enzyme as a new possible biomarker for this pathology to translate to the clinical management of patients affected by this disease. Notably, the p-AKT/total AKT ratio does not show a corresponding increase, suggesting that the higher p-AKT signal is mainly driven by increased AKT1 abundance; nevertheless, the concurrent changes in AKT1 expression and Ser473 phosphorylation are consistent with altered AKT pathway signaling.

Trans-omic integration highlights AKT1-linked metabolic signatures in advanced lipedema

Given the broad metabolic impact of the phosphoinositide 3-kinase (PI3K)-AKT signaling pathway, whose aberrant activation can modulate multiple metabolic processes through direct and indirect mechanisms, we next investigated the metabolic alterations in lipedema using untargeted metabolomics [49, 50]. The adipose tissue samples were subjected to a double-phase extraction, both polar and apolar. Samples containing the polar phase were prepared by adding deuterium-enriched phosphate buffer, and one-dimensional ¹H NOESY spectra were acquired using a 600 MHz NMR spectrometer. NMR spectra were analyzed using Chenomx software, resulting in the identification of 56 metabolites (Supplementary Fig. S5A). An analogous procedure was followed for the adipose tissue of the healthy subjects. Accordingly, we obtained two data matrices: one containing metabolite concentrations for each lipedema patient (LP) and the other containing metabolite concentrations for each healthy control (HC) subject. The data matrices were compiled and analysed using both univariate and multivariate statistical methods on the MetaboAnalyst 6.0 online platform. Supervised Partial Least Squares Discriminant Analysis (PLS-DA) revealed that the tissue metabolomic profiles of patients with lipedema are significantly distinct from those of the control group (PC’1 accuracy: 1.0; Q2: 0.95) (Supplementary Fig. S6B). Variable Importance Projection (VIP) analysis identified abnormal metabolite levels in LP adipose tissues: (i) increased concentrations of several amino acids, including lysine, N6-acetyl-lysine, leucine, serine, and aspartate; (ii) changes in energetic metabolites, with upregulation of methylmalonate and malate, and a decrease in glucose; (iii) additionally, higher methylmalonate levels and lower sn-glycerol-3-phosphocholine levels were observed (Fig. 5A, left panel). The univariate approach utilising the Volcano plot (Supplementary Fig. S6C), together with the heatmap representation displaying the initial 25 discriminative variables (Fig. 5A right panel), confirmed the energetic dysregulation caused by higher concentrations of NADP+, malate, and lower concentrations of succinate and glucose. Additionally, it corroborated the amino acid dysregulation, evidencing a downregulation of valine and an upregulation of serine, leucine, and lysine in the pathological tissue. Subsequently, an analysis of dysregulated pathways was carried out utilising the Metaboanalyst 6.0 software with the MetPa tool. The findings indicate that the onset of lipedema contributes to dysregulation of energy metabolism, involving changes in the metabolic pathways. Pathways analysis identified dysregulation in bioenergetics, indicating disruptions in glyoxylate and dicarboxylate metabolism, pyruvate metabolism, and the citrate cycle (TCA). Additionally, amino acid dysregulation depends on changes in arginine biosynthesis and the metabolism of alanine, aspartate, glutamate, glycine, serine, threonine, cysteine, and methionine. We also highlighted a dysregulation in BCAA metabolism. Lipedema tissue is notably marked by an alteration in a single carbon pool through folate pathways, suggesting a disruption in vitamin B synthesis and homocysteine regulation. Lastly, the identification of an imbalance in the redox environment was recognised because of the upregulation of glutathione metabolism (Fig. S5D and E). To investigate the functional crosstalk between gene expression and metabolic alterations in advanced lipedema, we performed a trans-omic network analysis by integrating untargeted metabolomics data with bulk RNA sequencing profiles derived from adipose tissue biopsies. This approach aimed to uncover coordinated molecular interactions linking metabolic dysfunction with transcriptional remodeling in the diseased tissue. Initial pathway-level interrogation of the metabolomics dataset using the MetaboAnalyst 6.0 platform and the MetPa enrichment tool revealed prominent alterations in bioenergetic circuits, including the TCA cycle, glycolysis/gluconeogenesis, pyruvate metabolism, and fatty acid elongation (Fig. 5B). These pathways exhibited high statistical significance and impact scores, underscoring a generalized disruption in central carbon metabolism in lipedema. A quantitative summary of the most significantly enriched metabolic routes, along with matched metabolite counts and corresponding false discovery rates (FDR) (Fig. 5C). The enrichment of pathways associated with mitochondrial respiration and redox balance further supported a scenario of increased energetic and oxidative burden in lipedema tissue. To mechanistically link these metabolic alterations with gene expression dynamics, we constructed a trans-omic network model integrating differentially expressed genes (pink nodes) and metabolites (blue squares) through known biochemical associations and literature-curated interactions (Fig. 5D). This network pinpointed AKT1, a serine/threonine kinase, as a central transcriptional hub significantly overexpressed in lipedema and exhibiting strong connectivity to altered metabolites such as NADP⁺, glutamate, glycerol, arginine, and guanosine. Further analysis revealed that AKT1 is also highly associated with key metabolic nodes, including adenine nucleotides (ATP, AMP), glutamic acid, and intermediates of the TCA cycle, suggesting a rewiring of the metabolic landscape toward increased energy demand, redox activity, and utilization of carbon and nitrogen substrates. These patterns are consistent with a model of metabolic reprogramming driven by chronic stress, mitochondrial dysfunction, or compensatory responses during adipose tissue remodeling in late stage lipedema. Notably, these findings align with AKT1’s well-established role in regulating glycolysis, glucose uptake, and TCA cycle flux via downstream effectors such as mTOR, GSK3, and FOXO1 [51–53]. Together, these results reinforce the concept that AKT1 acts as a key integrator of metabolic stress and transcriptional reprogramming in lipedema, orchestrating alterations in energy production, amino acid metabolism, and antioxidant defence.

Fig. 5.

Metabolomic profiling and data integration. A) The VIP score plot displays the most relevant metabolites for distinguishing between groups, arranged according to their statistical contribution to the PLS-DA model (VIP > 1.3 is considered relevant) (left panel). Heatmap show the top 25 altered metabolites in lipedema tissue are compared to healthy control adipose tissue. The colour of each section corresponds to the concentration value of each metabolite, calculated through a normalised concentration matrix (red indicating upregulated; blue indicating downregulated) (right panel). B) The analysis of metabolic pathways was conducted utilising MetaboAnalyst 6.0 with the application of the MetPa tool. Bubble color indicates pathway significance (p-value/-log₁₀(p)), and bubble size is proportional to pathway impact derived from topology analysis, with large red circles signifying highly altered pathways. Prominent dysregulated routes include the TCA cycle, glycolysis/gluconeogenesis, pyruvate metabolism, and fatty acid elongation. C) The table presents the discriminant biochemical pathways associated with the metabolic profiles of pathological and control adipose tissue. “Hits” refer to the number of matched metabolites, while the False Discovery Rate (FDR) reflects the expected proportion of false positives. Pathway impact values combine topological centrality and enrichment scores, with higher values indicating stronger influence within the metabolic network. D) Trans-omic network analysis integrates transcriptomic and metabolomic data to reveal key interactions centered around overexpressed AKT1. The network connects AKT1 to several differentially regulated metabolites including NADP⁺, glutamate, glycerol, arginine, and guanosine. Blue and green arrows indicate directionality of change (up = light blue or down = green), supporting a model of AKT1-driven metabolic rewiring involving redox imbalance, amino acid metabolism, and altered energy flux in lipedema adipose tissue

Discussion

In this study, we employed an integrative multi-omics approach to delineate the molecular and metabolic landscape of advanced-stage lipedema, a disabling adipose tissue disorder characterized by chronic pain, inflammation, and progressive fat deposition. By combining genome-wide DNA methylation profiling, transcriptomic analysis, and untargeted metabolomics in subcutaneous adipose tissue, we found a regulatory role for AKT1, linking epigenetic deregulation to mitochondrial dysfunction and metabolic reprogramming. These findings offer novel mechanistic insight into late-stage lipedema pathophysiology and highlight a potential molecular target for disease stratification and therapeutic intervention. Our data show extensive epigenetic remodeling in lipedema adipose tissue, with over 5,000 differentially methylated CpG sites affecting pathways related to receptor tyrosine kinase signaling, immune regulation, and phospho-metabolism. This aligns with previous findings suggesting chronic low-grade inflammation, vascular dysregulation, and hormonal sensitivity as core features of lipedema pathogenesis [54–56]. Notably, we identified a hypomethylated regulatory region in the promoter of AKT1, corresponding with its overexpression and increased phosphorylation, suggesting that epigenetic modulation may contribute to the observed AKT1 signaling changes. Transcriptomic profiling further revealed marked downregulation of mitochondrial genes involved in oxidative phosphorylation (OXPHOS), fatty acid β-oxidation, and the TCA cycle. These alterations suggest impaired mitochondrial bioenergetics, consistent with previous reports showing reduced mitochondrial activity and altered lipid metabolism in lipedema adipose tissue [57, 58]. In line with these findings, Straub et al. demonstrated reduced expression of mitochondrial proteins in early-stage lipedema using proteomic profiling, though they did not identify upstream regulatory nodes or link the observed changes to DNA methylation [22]. Our study extends these observations by proving a mechanistic connection between epigenetic deregulation and mitochondrial dysfunction in the context of advanced disease. Metabolomic profiling supported these transcriptomic and epigenomic changes, revealing altered levels of several metabolites associated with AKT1-driven pathways, including arginine, NADP, glycerol, guanosine, and glutamate. These metabolites are critical regulators of redox homeostasis, energy balance, and amino acid metabolism—key processes that are frequently disrupted in chronic metabolic diseases. A previous study by Al-Ghadban et al. [8] also reported elevated oxidative stress and inflammation in lipedema tissue, and our metabolomic data support the existence of a dysregulated redox environment potentially sustained by aberrant AKT1 activation. Together, these findings suggest that AKT1 may serve as a central metabolic integrator that drives adipose tissue dysfunction and disease chronicity in lipedema. The functional convergence of multi-omic layers on AKT1 is particularly notable given its central role in regulating insulin signaling, nutrient sensing, and cell survival [49, 59]. In the context of obesity, AKT1 associated signaling alterations is known to promote adipocyte hypertrophy, lipid storage, and macrophage recruitment, but its involvement in non-obesity-related disorders such as lipedema has not been previously characterized [60–62]. Our findings suggest that AKT1 may represent a shared pathological mechanism across distinct adipose tissue disorders, with lipedema presenting a unique epigenetically driven pattern of activation. Compared to the study by Straub et al., which analyzed early-stage disease using transcriptomics, proteomics, and cytokine profiling, our approach emphasizes functional integration across omics layers and focuses on advanced-stage pathology [22]. While Straub’s study provided valuable insight into early molecular alterations, it did not resolve mechanistic regulators or explore metabolic outputs. By incorporating DNA methylation and NMR-based metabolomics in a tissue-specific context, our study uncovers a regulatory axis that may contribute to the transition from early to advanced disease stages. Importantly, the convergence of methylation, expression, and metabolite changes on AKT1 provides a strong rationale for its future evaluation as a biomarker and druggable target. Among the strengths of our study are the use of clinically and histologically validated tissue samples from patients with stage III lipedema, the simultaneous profiling of methylome, transcriptome, and metabolome in the same samples, and the application of integrated bioinformatic tools for cross-layer analysis. This comprehensive design enables a systems-level understanding of disease progression and identifies molecular candidates with translational relevance.

Translational implications

This study offers critical translational insights into the molecular pathogenesis of advanced-stage lipedema, highlighting AKT1 as a central integrator of epigenetic, transcriptomic, and metabolic dysregulation. Our multi-omics approach uncovered a hypomethylated regulatory region in the AKT1 promoter, leading to increased gene expression and protein phosphorylation. These findings suggest that increased AKT1 expression is driven by stable epigenetic changes, rather than somatic mutations, underscoring its potential as a modifiable molecular target. Given AKT1’s established role in insulin signaling, mitochondrial function, and energy metabolism, its upregulation may explain key pathological features of lipedema, including impaired oxidative phosphorylation, chronic inflammation, and altered adipocyte metabolism. Moreover, the identification of AKT1-linked metabolic signatures, such as changes in NADP+, ATP, glutamate, and amino acid profiles, supports the development of metabolomics-based biomarkers for non-invasive disease monitoring. From a systems biology perspective, these findings position AKT1 at the intersection of signaling networks that govern adipose tissue homeostasis and disease progression. Therapeutically, our results suggest that pharmacological modulation of AKT1 or its downstream effectors may represent a novel strategy for intervention in late-stage lipedema. Importantly, AKT inhibitors are already in clinical development for other metabolic and oncologic disorders, potentially accelerating drug repurpose efforts. Overall, our integrative analysis not only elucidates a previously unrecognized regulatory axis in lipedema but also provides a mechanistic framework for biomarker discovery and targeted therapy development, aligning with the core translational goals of precision medicine.

Conclusion

To our knowledge, this is the first multi-omics study in advanced-stage lipedema to identify an association between epigenetic regulation of AKT1 and coordinated metabolic reprogramming, providing mechanistic hypotheses and translational candidates. Our data provide compelling evidence that AKT1 functions as a epigenetically regulated node in advanced lipedema, integrating signals across molecular and metabolic layers. These findings expand current understanding of lipedema pathophysiology and open new avenues for biomarker discovery and mechanism-based therapeutic development. Moreover, our study highlights the utility of multi-omics integration in identifying disease-driving mechanisms in underexplored adipose tissue disorders. Taken together, these findings position AKT1 not only as a molecular hallmark of advanced lipedema but also as a promising therapeutic target that warrants further validation in preclinical models and pharmacological intervention studies.

Limitations of the study

However, several limitations should be acknowledged. First, the sample size was modest due to the clinical challenges associated with obtaining large adipose tissue biopsies from late-stage patients. Depot heterogeneity remains a potential confounder (extremity lipedema vs. abdomen/breast controls); we assessed site effects in metabolomics (iPCA/heatmap), but site-matched cohorts will be required for definitive attribution. While the molecular alterations identified were robust and internally consistent, validation in larger, independent cohorts will be important to confirm our findings and assess inter-individual variability. Second, functional experiments (e.g., AKT1 knockdown or overexpression) were not performed; therefore, further mechanistic studies are required to establish a causal role for AKT1 in mediating the observed phenotypes. Third, although we focused exclusively on female patients in line with the epidemiology of lipedema, future studies should consider potential sex-specific differences and include appropriate controls for hormonal status, menopausal state, and body mass index (BMI). In addition, while our untargeted metabolomics approach provided a comprehensive overview of metabolic alterations, more refined techniques, such as targeted fluxomics or lipidomics, may yield deeper insights into lipid processing and redox cycling. That said, it is noteworthy that portions of our transcriptomic and metabolomic profiles are consistent with previously published findings [22], providing an independent layer of validation. Moreover, the role of AKT1 and its associated pathways was supported by orthogonal evidence from integrated omics layers, underpinned by robust statistical power. Finally, AKT1 dysregulation and phosphorylation were independently confirmed through validation assays.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AKT1

RAC-alpha serine/threonine-protein kinase

BMI

Body Mass Index

CpG

Cytosine-phosphate-Guanine dinucleotide

DESeq2

Differential Expression analysis based on the Negative Binomial distribution

DNA

Deoxyribonucleic Acid

DMSO

Dimethyl Sulfoxide

FDR

False Discovery Rate

FC

Fold Change

GSEA

Gene Set Enrichment Analysis

GO

Gene Ontology

GRCh38

Genome Reference Consortium Human Build 38

IPA

Ingenuity Pathway Analysis

KEGG

Kyoto Encyclopedia of Genes and Genomes

LP

Lipedema Patient

HC

Healthy Control

MTBE

Methyl-t-butyl ether

NMR

Nuclear Magnetic Resonance

PLS-DA

Partial Least Squares Discriminant Analysis

RIN

RNA Integrity Number

RNA

Ribonucleic Acid

RNA-Seq

RNA Sequencing

SD

Standard Deviation

SOP

Standard Operating Procedure

STAR

Spliced Transcripts Alignment to a Reference

TCA

Tricarboxylic Acid (cycle)

TSS

Transcription Start Site

TSP-d4

2,2,3,3-d4-trimethylsilyl-propionic acid sodium salt

UTR

Untranslated Region

VIP

Variable Importance in Projection

Author contributions

Biagio Santella: Writing-original draft, Methodology, Investigation, Data collection; Annamaria Salvati: Writing-original draft, Methodology, Investigation, Formal analysis. Alexander Papp: Patient recruitment, Data collection, Sampling tissues; Annamaria D’Ursi: Writing-review & editing, Supervision; Monica Mingo: Methodology & Data collection; Christoph Pulai: Patient recruitment, Data collection, Sampling tissues; Carmen Marino: Writing-original draft, Methodology, Formal analysis; Domenico Memoli: Investigation, Formal analysis, Data curation; Luca Rastrelli: Writing-review & editing; Maria D’Elia: Writing-review & editing; Giovanni Nassa: Writing-review & editing, Supervision, Conceptualization, Funding acquisition; Luigi Schiavo: Writing-review & editing, Supervision, Conceptualization, Funding acquisition.

Funding

Work supported by: Italian Ministry of University and Research PNRR-MUR NextGenerationEU PRIN 2022, cod. 202282CMEA - CUP: D53D23007790001 and PNRR-MUR NextGenerationEU PRIN-PNRR 2022 cod. P2022N28FJ - CUP: D53D23016530001 (to G.N.); University of Salerno, Fondi FARB (to G. N.) and National Center 5 “National Biodiversity Future Center” (identification code CN00000033), theme “Biodiversity”, funded under the National Recovery and Resilience Plan (PNRR)—Mission 4, Component 2 “From Research to Business” Investment 1.4 “Strengtheningresearch structures and creation of “national R&D champions” in some Key Enabling Technologies”,funded by the European Union—Next-Generation EU.

Data availability

The data supporting the findings of this study are available within the main text and from the corresponding authors upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki. The study was reviewed and approved by the ethics committee of the Land Salzburg. All study participants gave written informed consent and were informed about the study (EK Nr: 1023/2024).

Consent for publication and Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper