Lipedema Adipocytes in Culture: Signs of Hypertrophy, Inflammation, and Fibrosis

Abstract

Despite extensive research during the last couple of years, lipedema still appears enigmatic in respect to its pathogenesis. In our in vitro study, we have set out to further characterize lipedema adipocytes, concentrating on gene and protein expression, which might help to develop ideas explaining the excessive accumulation of adipose tissue in women with lipedema. Using 2D cultures we show that gene expression in lipedema and non-lipedema adipocytes differs significantly in terms of genes related to lipid droplet size determination, insulin signaling and glucose uptake. A pronounced hypertrophy, recognizable by a significantly increased average lipid droplet size, was visible in differentiated lipedema adipocytes grown in 3D cultures. In addition, gene and protein expression related to inflammation and fibrosis were upregulated in lipedema adipocytes compared to controls, supporting earlier reports. Taken together, results from our in vitro studies suggest that lipedema adipose cells are capable of retaining their hypertrophic nature under culture conditions and open new aspects focusing on insulin signaling and PDGFRA-mediated balancing of adipogenic versus fibrogenic differentiation of lipedema adipose tissue.

1. Introduction

According to the current state of knowledge, lipedema can be classified as a fat distribution disorder—a term usually describing the localized imbalance of subcutaneous fat deposition and affecting both overweight and normal-weight individuals.

Women who suffer from lipedema (synonymously also referred to as adiposis dolorosa or lipalgia) experience a symmetrical enlargement of their extremities, with partial involvement of the hips and buttocks at advanced stages, while their waists remain disproportionately small [1, 2, 3, 4, 5, 6]. Lipedema can occur as both a syndromic and non-syndromic disease, with a suggested but yet-undefined autosomal dominant inheritance [7, 8].

Apart from the unbalanced accumulation of subcutaneous fat on the extremities, people with lipedema show several pathological characteristics such as tenderness, pressure pain, easy bruising and depressive moods, while suffering from an inexorable, diet- and exercise-resistant weight gain [9].

Lipedema is not typically associated with metabolic complications and related cardiovascular diseases, while hypothyroidism appears to be a frequent co-morbidity of lipedema [10], particularly at higher stages.

Lipedema is found predominantly, if not exclusively, in women, the onset being associated with major hormonal changes. In this context, puberty seems to be a significant trigger for lipedema, followed by pregnancy and menopause, apart from other, less defined, causes.

The rare occurrence of lipedema in men is usually a consequence of a primary hormonal disorder [7]. However, despite the many indications that lipedema might be an estrogen-associated disease [11], scientific evidence clearly demonstrating a causal role for estrogen in the initiation or progression of lipedema is still missing.

One important key detail distinguishing lipedema from other fat distribution disorders such as Madelung’s disease (multiple symmetric lipomatosis), Dercum’s disease, or familial multiple lipomatosis, is the higher relative frequency with which it occurs [12]. While the above-mentioned diseases are particularly rare, the prevalence of lipedema is estimated up to 11% in women and post-pubertal girls [13]. However, statistic numbers concerning the prevalence of lipedema vary substantially [14]. This can be explained by the limited availability of accurate diagnostic options, the lack of molecular “markers” and probably by a long-standing ignorance of this disease. In addition, lipedema has often been misdiagnosed as obesity, lymphatic or venous insufficiency, or lipohypertrophy, which is not surprising given the phenotypic similarities and the partially shared symptomatology at advanced stages [15, 16, 17].

It needs to be emphasized that lipohypertrophy differs from lipedema in several important aspects, above all the absence of pain and the weaker (or even absent) tendency to develop edema at advanced stages. Histological and molecular characteristics to distinguish lipohypertrophy from lipedema and secondary lymphedema were recently described [18].

Since lipedema was originally described as a “syndrome characterized by fat legs and orthostatic edema” [19], the assumption that lipedema is triggered by the formation of edema was obvious. Since then, numerous studies have focused on the potential occurrence of lymphatic disturbances and vascular fragility in people with lipedema, with contradicting results (reviewed in [20, 21, 22]).

Taking together all evidence available so far, it may be postulated that edema is not integral to the initiation and early development of lipedema. However, during advanced stages, the co-occurrence of obesity and lymphatic impediment is frequently observed. Therefore, conservative treatment including decongestion therapy often achieves success in relieving the patient’s pain, but does not impact the pathogenesis of lipedema itself. Currently, liposuction appears to be the best treatment for long-term improvement [23].

The criteria used for the diagnosis of lipedema were established in the 1950s [19] and, with a few modifications introduced by Herbst [24], are still valid today. A thorough visual inspection, palpation and a detailed discussion of the patients’ histories are still the basis of daily routine diagnosis. Sonography may improve diagnosis and help to distinguish lipedema from lymphedema, particularly at advanced stages [25].

Although it is generally agreed upon that lipedema progresses through stages [26] that can develop with different dynamics, the biological mechanism(s) underlying this progression are not yet clear. Numerous studies were conducted to explore potential molecular alterations distinguishing lipedema from non-lipedema adipose tissue in vivo and in vitro (reviewed in [21, 22]). Thereby, a major focus was set on the appearance of hyperplasia and hypertrophy, two basic mechanisms underlying adipose tissue accumulation [27, 28], and on the role of low-level inflammation and fibrotic changes in the development of lipedema [29].

Concerning adipose tissue hyperplasia, in vitro studies have contradicted each other. Several articles state that lipedema preadipocytes show increased proliferative activity compared to non-lipedema cells, estimated either upon expression of gene markers such as CD90 or cyclin D1, or using cell counting and metabolic activity assays, while other studies do not support this notion [30, 31, 32, 33, 34].

Data concerning the occurrence of hypertrophy in lipedema adipose tissue are more consistent, as demonstrated by the measurements of adipocyte size in biopsied subcutaneous adipose tissue [18, 30, 35, 36]. While it appears plausible that adipocyte size is increased in lipedema adipose tissue, a more detailed analysis supporting this finding is missing. Since most of the molecular data available so far have emerged from studies using in vitro differentiated adipocytes grown from the stromal vascular fraction (SVF) of subcutaneous adipose tissue, it should be clarified, whether the suggested hypertrophy found in vivo is also evident in vitro.

The goal of this study was to further elucidate the previously suggested hypertrophy of lipedema adipocytes using 2D and 3D culture systems. Here we show for the first time that lipedema adipose cells are capable of retaining their hypertrophic nature during cultivation, which is reflected by significantly increased average size of lipid droplets (LDs) in lipedema adipocytes compared to controls. This is supported by our results from gene expression studies showing elevated expression of genes related to LD enlargement in lipedema adipocytes in vitro. In addition, we confirmed the significant upregulation of fibrosis and inflammation-related gene and protein expression in lipedema adipocytes, supporting the notion that lipedema pathology is obviously accompanied and intensified by low-level inflammation and fibrotic alterations.

2. Material and methods

2.1 Human tissue specimens

In vitro experiments were performed on adipose tissue-derived stromal cells (ASCs) isolated from lipoaspirates obtained exclusively from the thigh region of anonymized female donors during routine operations. Ex vivo experiments were performed on tissue excised from cannula incision sites directly before the operations.

Informed consent was acquired from all patients. All experimental protocols concerning human adipose tissues (biopsied material and lipoaspirate) were approved by the Ethics Committee of Salzburg, Austria.

The anonymized donors were categorized into lipedema and control donors: Obese (in the following referred to as OB) lipedema donors (n = 5, average age at operation 29.4 ± 1.5 years, BMI 36.2 ± 4.6 kg/m², non-obese (normal weight and overweight, in the following referred to as NW/OW) lipedema donors (n = 10, average age at operation 31.0 ± 6.9 years, BMI 24.4 ± 2.3 kg/m²), and controls (n = 6, average age at operation 33.4 ± 8.4 years, BMI (25.4 ± 2.1 kg/m²).

2.2 Cell isolation

SVF from lipoaspirates was isolated according to a standardized procedure. To summarize: To isolate the ASCs, 0.1% collagenase A (Sigma-Aldrich; Vienna, Austria) was dissolved in phosphate buffered saline (PBS). Collagenase solution and lipoaspirate were mixed one to one and incubated for 30 min at 37°C in a water bath under constant agitation. StemMACS MSC expansion medium (MiltenyiBiotech; Bergisch-Gladbach, Germany) was used to stop the collagenase digestion. Supernatant was discarded after centrifugation (340 g) for 10 min, and the pellet was resuspended in PBS and filtered through a 100 μm cell strainer (Falcon; Corning, NY, USA). The pellet was resuspended in culture medium after a second identical centrifugation step before seeding in culture flasks. Whenever necessary, pellets were also washed with red blood cell lysis buffer (RBC lysis buffer, Roche Diagnostics; Rotkreuz, Switzerland), centrifuged again, resuspended in culture medium, and seeded.

After isolation, the ASCs were cultured in basic medium containing high-glucose (4.5 g/L D-Glucose) DMEM with glutamine (Sigma-Aldrich), 10% fetal bovine serum (FBS, Sigma-Aldrich) and 1% antibiotic/antimycotic solution (Sigma-Aldrich). The ASCs were fed once per week and passaged 0–3 times before being used for 2D or 3D experimentation.

2.3 In vitro differentiation

Adipogenic differentiation was induced by switching out basic medium with differentiation and then maintenance medium changes every 3–4 days for 21 days. The differentiation medium ingredients were as follows: DMEM high glucose, 1 μmol/L dexamethasone (Sigma-Aldrich), 0.5 mmol/L 3-isobutyl-1-methylxanthine (IBMX) (Sigma), 10 μg/mL insulin (Sigma-Aldrich) and 100 μmol/L indomethacin (Sigma-Aldrich). The cells were cultured for 4 days in the differentiation medium and then differentiation was maintained for 3 days with DMEM high glucose with the addition of 10 μg/mL insulin, before being cultured with differentiation medium again.

After 21 days, the adipogenic differentiation was verified by Oil Red O (ORO) staining. The cell yield was fixed with 4% paraformaldehyde (PFA), stained with a 60% ORO solution (Sigma-Aldrich), and washed 3× with dH₂O to visualize the red-stained LDs.

2.4 Total RNA extraction and reverse transcription of 2D ASCs

After isolation, seeding in T25 culture flasks, and 21-day adipogenic differentiation induction after confluency, differentiated cells were harvested and homogenized in TRIzol™ Reagent (ThermoFisher Scientific; Waltham, MA, USA). RNA was extracted using the manufacture’s protocol. RNA pellets were solubilized in 20–30 μl nuclease-free H₂O (IDT) with 1 μl (1:20 dilution) of RNAase inhibitor (Sigma-Aldrich). RNA yield and quality using the 260/280 nm ratio was established with the NanoDrop 2000c Spectrophotometer (ThermoFisher Scientific).

Reverse transcription was then performed using the iScript™ cDNA synthesis kit (BIORAD; Ipswich, MA, USA) according to the manufacturer’s protocol. 1 μg of RNA was used for the cDNA synthesis.

2.5 Quantitative PCR analysis

Quantitative PCRs (qPCRs) were performed with the Luna^® Universal Probe qPCR Master Mix (New England Biolabs^®; Ipswich, MA, USA) according to the manufacturer’s instructions. A concentration of 5 ng/μl cDNA was analyzed per well. Predesigned qPCR assays were selected using Integrated DNA Technologies (IDT; Coralville, IW, USA). Table 1 lists the primers of the predesigned assays.

Function	Target gene symbol	Target gene names	ID	Primer 1 sequence; primer 2 sequence; probe
Extracellular matrix organization	COL1A1	Collagen I	NM_000088	5′-CCAGTGTCTCCCTTGGGT-3′; 5′-TGACGAGACCAAGAACTGC-3′; 5′-/56-FAM/ACCCACCGA/Zen/CCAAGAAACCACC /3IABkFQ/-3′
Receptors, growth factors, cofactors, and transporters	GLUT4 (SLC2A4)	Insulin-regulated glucose transporter	NM_001042	5′-TCCTTGCTGTGTTCTCGC-3′; 5′-CCCAGCCACGTCTCATTG-3′; 5′-/56-FAM/TGCCCCZVS/ZEN/GAAGGTGATTGAACAG/3IAVkGQ/-3′
	IGF-1R	Insulin growth like factor 1 receptor	NM_000875	5′-ATCCGCCATTCTCATGCC-3′; 5′-CCAGTCCCACAGTTGCTG-3′; 5′-/56-FAM/ACTCCTTCT/ZEN/ACGTCCTCGACAACCA/3IABkFQ/-3′
	INSR	Insulin receptor	NM_001079817	5′-AAGTGCATCCCTGAGTGTC-3′; 5′-TGGTCTTCTCGCCTTCTAGG-3′; 5′-/56-FAM/TCCCTGTCC/ZEN/CAAGGTGTGCC/3IABkFQ/-3′
	PPARGC1A	Protein alpha PPARG Coactivator 1 Alpha	NM_013261	5′-AGACGTGACCACTGACAATG-3′; 5′-CCACTGCATTCATTATAACTTAGCTG-3′; 5′-/5 g-FAM/CACCCAGG/ZEN/AGGCAGAAGAGC/3IABkGQ/-3′
Components involved in adipogenesis and lipid metabolism	ABHD5 (CGI-58)	Alpha/beta hydrolase domain-containing protein 5	NM_016006	5′-CTGCTTACTCGCTGAAGTACC-3′; 5′-TGGATTGGTCTGTCTTGATCAG-3′; 5′/56-FAM/CCCCAAGGC/ZEN/TCCACTAAAATGAGATGA/3IABkFQ/-3′
	AKR1C1	Aldo-Keto Reductase 1C1	NM_001353	5′-TCTATGCTGGTGACTGGACA-3′; 5′-GGCTTTGCTGTAGCTTGC-3′; 5′-56-FAM/CCTCTGGTT/ZEN/AAATCTCTCCTGXTTGGTG/3IABFQ7-3′
	ATGL (PLNPA2)	Adipose triglyceride lipase	NM_020376	5′-CACTGTGATGGTGTTCTTAAGC-3′: 5′-GAGCTCTCATCCAGGCCAAT-3′: 5′-/56-FAM/TGGTGGCAT/Zen/TTCAGACAACCTGC/3IABkFQ/-3′
	CIDEA	Cell Death Inducing DFFA Like Effector A	NR_036468	5′-CAATGACGAATGTTGAAGATGCT-3′; 5′-CCCACTTCATGTACCACCAA-3′; 5′-/56-FAM/ACCCACTCCT/ZEN/GAGCAGCGTGAC/3IABkFQ/-3′
	FSP27 (CIDEC)	Fat-specific protein 27	NM_022094	5′-GACACTCTGATGCTGGCA-3′; GTACTCTTCTGTCTCTACAGTTGTG-3″; 5′-56-FAM/AAGCCCTCTC/ZEN/TTCCTGGTGCTGG/3IABkFQ/-3′
	LPL	Lipoprotein lipase	NM_000237	5′-CTGACTCTGGCCGTGTG-3′; 5′-GGAATGAGGTGGCAAGTGT-3′; 5′-/56-FAM/AGAGTCTGA/ZEN/CCGCCTCCCG/3IABkFQ/-3′
	PLIN1	Perilipin 1	NM_002666	5′-CTGAAGGGCGTTACTGACAA-3′; 5′-CTGGTGGGTTGTCGATGT-3′; 5′-/56-FAM/CCAGGCTGT/ZEN/CGCTGATGGAGC/3IABkFQ/-3′
Cytokine	IL6	Interleukin-6	NM_000600	5′-GCAGATGAGTACAAAGTCCTGA-3′; 5′-TTCTGTGCCTGAGCTTC-3′; 5′-56-FAM/CAACCACAA/ZEN/ATGCCAGCCTGCT/3IABkFQ/-3′
Tissue fibrosis related markers	PDGFRA	Platelet derived growth factor receptor alpha	NM_006206	5′-GCGACAAGGTATAATGGCAGA-3′; 5′-AGACAGAAGAGAATGAGCTTGAAG-3′; 5′-/56-FAM/AGGCTACAT/Zen/CTGGGTCTGGCAC/3IABkFQ/-3′
Reference genes	TBP	TATA-Box binding protein	NM_003194	5′-GATAAGAGAGCCACGAACCAC-3′; 5′-CAAGAACTTAGCTGGAAAACCC-3′; 5′-/56-FAM/CACAGGAGC/ZRN/CAAGAGTGAAGAACAGT/3IABkFQ/-3′
	PUM1	Pumilio RNA binding family member 1	NM_014676	5′-GAGCAGCAGAGATGTATCTTCC-3′; 5′-GACCAGGACATTCACAGACAC-3′; 5′-56-FAM/TCCTTCCCT/ZEN/CTCTTCACATGGATCCT/3IABkFQ/-3′
	RPLP0	Ribosomal protein lateral stalk subunit P0	NM_053275	5′-ACATCTCCCCCTTCTCCTT-3′; 5′-CAGACAGACACTGGCAACA-3′; 5′-/56-FAM/CCTGAAGTG/ZEN/CTTGATATCACAGAGGAAACT/3IABkFQ/-3′

Table 1.

List of forward and reverse primers and probes used for qPCR (pre-designed and purchased from Integrated DNA Technologies (IDT)).

All samples were run in technical duplicates and CQ values were analyzed using qBaseplus v2.4 (Biogazelle; Zwijnaarde, Belgium). Relative gene expressions of the data were calculated by normalizing the data to the expression of three reference genes (TBP, PUM1, and RPLP0) which were stably expressed across sample conditions.

2.6 Immunofluorescence staining of cultured cells

SVF-cells were harvested and cultured as described previously. Passage 2 cells were seeded on glass coverslips and grown until confluence. Then, cells were fixed in situ with 4% PFA, and washed with PBS. After permeabilizing with 0.2% TritonX100 and blocking unspecific binding sites by preincubation with 1% Bovine Serum Albumin (BSA) in PBS + 0.05% Tween 20 (PBST) for 30 min, cells were incubated overnight at 4°C with the specific antibodies mo-anti CD68 RTU (#514H12 Bond/Leica; Wetzlar, Germany) mixed with rb-anti CD163 1:75 (#STJ110681, St. Johns; London, UK) diluted in 0.1% PBST. After 4× washing, cells were incubated with the secondary antibody mix of Alexa Fluor^® 488 Donkey anti-mouse IgG (H + L) 1:500 and Alexa Fluor^® 568 Donkey anti-rabbit IgG (H + L) 1:1500 (both Invitrogen; Waltham, MA, USA) diluted in 0.1% PBST for 30 min at room temperature. Following washing, cell nuclei were stained with DAPI (Sigma-Aldrich) for 10 min.

Cells on coverslips were mounted with CitiFluor™ (Electron Microscopy Sciences; Hatfield, PA, USA) on glass slides. For evaluation of M1 and M2 macrophage-distribution, fluorescence micrographs were taken through a 40× objective with Nikon NIS Elements B software at a Nikon Eclipse 800 microscope equipped with a Nikon DS-Ri1 camera (Nikon; Tokyo, Japan).

Micrographs from 5 to 12 fields (depending on cell density) were taken and nuclei were counted. Overlay pictures of the three channels DAPI, CD68 → green (488 nm) and CD163 → red (568 nm) were assembled with Adobe Photoshop software. CD68 alone and CD68/CD163 positive cells were counted and evaluated in percentages in comparison to DAPI positive cells.

2.7 Spheroid formation, in vitro differentiation and immunofluorescence staining

For morphologic measurements, 6000 ASCs originating from lipedema and control SVF were seeded first in hanging droplets. In short, droplets of 20 μl cell suspension were placed on the underside of a lid of a 30 mm dish, which was then immediately flipped over back on to the dish. The subsequently formed hanging droplets incubated in a 37°C CO₂ incubator for 24 h where aggregation of ASCs was induced. After spheroid formation, the spheroids were transferred to non-adhesive 24-well plates (Nunclon-Sphera; Waltham, MA, USA). Growth of ASC spheroids was monitored throughout 3 days, with daily photographic documentation using a Zeiss Axiovert™ microscope and a GryphaxR Subra camera (Jenoptik; Jena, Germany).

For differentiation experiments, aggregation of 5000 ASCs was induced as described above. After transferring to non-adhesive plates, the spheroids underwent induction of adipogenic differentiation in the same manner as the 2D cells. After 21 days of adipogenic differentiation, the spheroids were fixed 30 min with 4% PFA. After fixation, spheroids were washed 3× with PBS, followed by a 7 min incubation of 0.2% TritonX100 to permeabilize the cells for antibody penetration. To prevent non-specific binding, the spheroids were incubated in blocking buffer (0.1% BSA in PBS) for 30 min. The primary antibody (Perilipin-1 (D1D8) XP^® Rabbit mAb, Cell Signaling Technologies; Danvers, MA, USA) was incubated at a 1:100 dilution overnight at 4°C. The secondary antibody (Alexa Fluor^® 568 Donkey anti-rabbit IgG (H + L), Invitrogen) was incubated at 1:1000 for 30 min at room temperature. Between each step, cells were washed with PBS. As a last step, nuclear counterstaining was carried out by a 15 min incubation with the nuclear stain DAPI at room temperature, and then the spheroids were washed 2× with PBS, 1× with dH₂O, transferred onto glass slides and mounted with Glycergel (Dako; Carpinteria, CA, USA). The spheroids were analyzed on a confocal laser-scanning microscope (Axio ObserverZ1 attached to LSM700, Carl Zeiss, Göttingen, Germany; ×20 dry objective), using only the 568 nm excitation channel to document as many LDs as possible. Areas of the LDs were measured using ImageJ software (NIH, Maryland, USA).

2.8 Embedding of spheroids for semi-thin sectioning

Spheroids grown in hanging drops and differentiated as described above were fixed with half strength Karnovksy’s fixative for 3 h. After 3× washing in 0.1 M cacodylate buffer at pH 7.4, samples were postfixed with 1% OsO₄ in cacodylate buffer for 3 h. Following 3× cacodylate buffer washing steps, spheroids were dehydrated by a graded series of ethanol with 3 changes of 100% ethanol and 2 subsequent changes of propylene oxide. Infiltration of epoxy resin Agar 100 (Science Services; Munich, Germany) were done in 3 ascending mixtures of propylene oxide/resin followed by incubation with pure resin overnight at 4°C. Samples were transferred into silicon molds with fresh resin and polymerized at 60°C for 2 days. Semithin sections (0.5 μm) were cut on a Reichert Ultracut™ S (Reichert, Austria) which were stained with Richardson’s dye. Sections were analyzed and pictures were taken with Nikon microscope equipment (see above).

2.9 Preparation of histological sections and polarization microscopy

After biopsy, the tissues were cut into small pieces of about 3 mm³ size and immediately fixed in 4% PFA.

After PFA fixation, the samples were processed by a conventional automated paraffin-embedding process (Leica, TP 1020 paraffin embedding processor). The samples were transferred to paraffin bloc molds. 5 μm sections were cut from the paraffin blocs. The tissue sections were subsequently “baked” for 2 h at 58°C. The sections were then deparaffinized using Rotihistol™ (Roth; Karlsruhe, Germany) and rehydrated in a series of ethanol solutions with lowering dilutions, washed with dH₂O and fixed with non-permanent CitiFluor™. The unstained sections were then imaged with a 10× objective on a light microscope with polarization filters (Axioplan, Carl Zeiss; Oberkochen, Germany) using the same wavelength for each image, captured with a digital camera (AxioCam, Carl Zeiss). The images were analyzed with ImageJ software.

Analysis included 6 biological repeats for lipedema and 3 biological repeats for control tissue.

2.10 Statistics

Statistical analyses were performed using GraphPad Prism software (version 9.0.0; La Jolla, CA, USA). Shapiro-Wilk tests were done to analyze if data were normally distributed. With data with normal distribution, Welch’s t tests were performed. When the data were not normally distributed, Mann-Whitney tests were used.

3. Results and discussion

3.1 Lipogenesis and lipid droplets

In the present study, we have followed gene expression and in vitro growth of ASCs and fully differentiated adipocytes derived from the SVF of lipoaspirates from lipedema and control donors. Since we and others have shown previously that lipedema adipocytes differ from non-lipedema adipocytes by increased expression of adipogenesis-related genes (reviewed in [21, 22]), we went on to identify additional characteristics distinguishing lipedema adipocytes from their healthy counterparts.

Our gene expression study was performed using 2D cultures of lipedema and control adipocytes, while growth and LD size were studied in undifferentiated and differentiated 3D cultures generated by a conventional hanging drop method.

Here, we concentrated particularly on gene expression related to lipogenesis and the enlargement of LDs with a minor focus on inflammation and fibrosis-related gene expression. Furthermore, we addressed the question of whether the suggested hypertrophy of lipedema adipocytes described in vivo is also evident in a 3D test arrangement in vitro.

Interestingly, mRNA levels of alpha/beta hydrolase domain-containing protein 5 (ABHD5), a co-activator of adipose triglyceride lipase (ATGL) and key enzyme of cytosolic LD lipolysis [37] did not differ significantly between lipedema and non-lipedema adipocytes (Figure 1). ABHD5 plays a critical role in balancing basal and stimulated lipolysis. Under basal conditions, it stays bound to perilipin-1 (PLIN1) (described below), impeding its interaction with and activation of ATGL. Upon phosphorylation of PLIN1, ABHD5 dissociates from PLIN1 and interacts with ATGL, thereby activating the first and rate-limiting step in LD hydrolysis [38, 39]. Since ABHD5 expression was comparable in the lipedema and control group, we assume that under non-stimulated conditions, lipolysis in lipedema and non-lipedema cells is comparable, which was also reflected by the equal expression of ATGL (Figure 1).

Figure 1.

Relative gene expression after 21 days of adipogenic differentiation of lipedema (n = 9, BMI 26.7 ± 3.3 kg/m²) and controls (n = 4, BMI: 27.3 ± 3.2 kg/m²). *P < 0.05, **P < 0.01. Abbreviations: ABHD5: alpha/beta hydrolase domain-containing protein 5; PLIN1: Perilipin 1; LPL: lipoprotein lipase; SLC2A4: solute carrier family 2 member 4 gene, encoding the insulin-regulated glucose transporter GLUT4; CIDEC: cell death inducing DFFA like effector C (FSP27, fat-specific protein 27); CIDEA: cell death inducing DFFA like effector A; ATGL: adipose triglyceride lipase; PPARGC1A: peroxisome proliferator-activated receptor gamma coactivator 1-alpha; PDGFRA: platelet-derived growth factor receptor alpha; INSR: insulin receptor; IGF-1R: insulin-like growth factor 1 receptor; AKR1C1: aldo-keto reductase family 1 member C1.

We have demonstrated that mRNA levels of PLIN1 and the cell death-inducing DFFA-like effector protein C (CIDEC), also referred to as fat-specific protein 27 (FSP27), two proteins critically involved in the enlargement of LDs [40, 41, 42], are significantly upregulated in lipedema adipocytes compared to controls (Figure 1). PLIN1, a member of the PAT (perilipin, adipophilin, TIP47) family [43, 44], is robustly expressed in adipocytes of white and brown adipose tissue and associates physically with the surface of LDs. Its dual function is based on its ability to modulate lipid metabolism by blocking basal lipolysis but mediating hormone-stimulated hydrolysis of triacylglycerols [45, 46, 47]. PLIN1 binds to and activates CIDEC, a key component of LD fusion and enhancement of unilocular LDs in white adipocytes [40, 41]. CIDE proteins, currently comprising CIDEA, -B, and -C, localize at LD-LD contact sites and are involved in the directional lipid transfer between LDs, enabling LD fusion and enhancing lipid storage capacity of adipocytes [40].

As with all other members of the CIDE family, CIDEC was originally known for its apoptosis-promoting function in various cell types [48], but gradually emerged as major player in the regulation of lipolysis, storage of triglycerides and insulin signaling [49]. The antilipolytic activity of CIDEC is well documented [50]. It can directly interact with ATGL to inhibit its lipolytic activity, but may also engage in mediating the suppression of ATGL transcription [50]. Insulin significantly increases CIDEC mRNA levels, thereby facilitating the enlargement of LDs in adipocytes. This effect was abrogated upon depletion of CIDEC but not CIDEA [51].

It may be speculated that the upregulation of PLIN1 and CIDEC in lipedema adipocytes is indicative of LD enlargement contributing to the hypertrophy of lipedema adipose tissue.

We have shown that mRNA levels of the multifaceted lipoprotein lipase (LPL) and the insulin-related facilitative glucose transporter SLC2A4 (GLUT4), are significantly increased in lipedema adipocytes compared to controls (Figure 1), suggesting elevated lipogenesis and an overall increase in lipid turnover and metabolism [52, 53]. Upregulation of adipocyte LPL has also been linked to inflammation and macrophage activation in obese adipose tissue [54]. In addition, expression in preadipocytes was previously shown to correlate directly with increased intracellular lipid load [52]. This suggests that adipocyte-derived LPL is capable of managing efficient lipid storage in adipocytes without being dependent on extracellular, vascular-derived LPL activity. It will be interesting to clarify whether the adipocyte-intrinsic upregulation of LPL plays a role in lipedema adipose tissue hypertrophy.

Interestingly, mRNA levels of the insulin receptor (INSR) but not insulin-like growth factor-1 receptor (IGF-1R) are significantly elevated in lipedema adipocytes compared to controls. INSR and IGF-1R are highly homologous tyrosine kinase receptors with partly overlapping intracellular signaling pathways playing an essential role in adipose tissue development [55, 56]. However, despite their considerable structural and functional similarities, INSR and IGF-1R exert differential effects on white and brown adipose tissue differentiation [57]. As demonstrated in animal knockout experiments, development and maintenance of white adipose tissue is highly dependent on INSR while IGF-1R plays only a minor role in this respect [57]. Interestingly, INSR and IGF-1R are differentially expressed during adipogenic development. While IGF-1R levels exceed INSR levels in undifferentiated adipose cells, this situation is reversed in mature adipocytes [57, 58].

Although INSR and IGF-1R functions are primarily determined by ligand binding kinetics and intracellular tyrosine kinase activity, variations in mRNA levels could also contribute to alterations in insulin-induced cell response. In this respect, increased levels of INSR mRNA were observed in liver cells of genetically diabetic and obese mice [59]. Concerning the significantly increased expression of INSR mRNA in cultured lipedema adipocytes, additional studies are needed to determine whether this may impact insulin signaling and lipid metabolism in lipedema adipose tissue.

Results from our molecular study further reveal that platelet-derived growth factor receptor-alpha (PDGFRA) mRNA levels are significantly increased in lipedema adipocytes compared to controls. Of note, PDGFRA was shown to play a crucial role in adipose organogenesis by controlling the adipogenic and fibrogenic differentiation of precursor cells [60, 61]. Although further in-depth studies are still missing, it cannot be ruled out that PDGFRA plays a role in the development of lipedema-associated fibrosis.

Neither aldo-keto reductase 1C1 (AKR1C1), a gene involved in progesterone metabolism and upregulated in a family with non-syndromic lipedema [62], nor the peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PPARGC1A), a key regulator of energy metabolism, glucose uptake and mitochondrial biogenesis [63] showed differential expression in lipedema and non-lipedema adipocytes.

3.2 Growth and hypertrophy of 3D ASC and adipocyte cultures

In this set of experiments, we focused on the growth of spheroids originating from SVF/ASCs from NW/OW and OB lipedema donors compared to controls. No significant difference in growth and morphology was detectable between 3D aggregates grown from lipedema and control ASCs, while a slight increase in the perimeters of spheroids from lipedema ASCs was observed after 3 days of cultivation, Figure 2A and B.

Figure 2.

(A) Morphological analysis of spheroids grown from lipedema and control ASCs. Scale bar: 200 μm. (B) Perimeter was measured using ImageJ software (NIH, Maryland, USA) in NW/OW lipedema (n = 3, BMI = 21.7 ± 0.4 kg/m²) and controls (n = 3, BMI = 25.7 ± 2.5 kg/m²). NW/OW: normal weight/overweight. (C) Size of lipid droplets (LD) measured in in vitro differentiated spheroids grown from lipedema and control ASCs. NW/OW lipedema (n = 3, BMI = 23.7 ± 1.4 kg/m²), NW/OW controls (n = 3, BMI = 24.3 ± 0.4 kg/m²). (*P < 0.05). (D) Representative images of PLIN1 expression in lipedema (a) and control (b) spheroids, scale bar: 100 μm, and semi-thin sections of in vitro differentiated lipedema (c) and control (d) spheroids, scale bar: 10 μm. PLIN1: Perilipin-1. (E) Representative polarization microscopy images of biopsied lipedema and control subcutaneous adipose tissue from lateral thighs of lipedema and control donors (both BMI 24.0 kg/m²). Scale bar: 100 μm. (F) Relative gene expression of collagen 1 (COL1A1) and interleukin-6 (IL6) in adipocytes after 21 days of adipogenic differentiation. Lipedema (n = 9, BMI 26.7 ± 3.3 kg/m²). Controls (n = 4, BMI = 27.3 ± 3.2 kg/m²). Comparison of COL1A1: *P < 0.05. Comparison of IL6: **P < 0.01. G) Quantitative evaluation of M1 and M2 macrophage abundance in lipedema NW/OW (n = 4, BMI = 22.8 ± 1.0 kg/m²), lipedema OB (n = 3, BMI = 39.4 ± 2.1 kg/m²) and control (n = 2, BMI = 24.3 ± 0.5 kg/m²) cultures. NW/OW: normal weight/overweight; OB: obese. Co-expression of CD68 and CD163 is predominantly found in ASCs from obese donors. (H) Immunoreactivity of (a) CD68 and (b) CD68/CD163 positive cells in lipedema ASC cultures. Representative examples are shown.

Next, we addressed the question whether LDs of in vitro-differentiated adipocytes grown in 3D cultures differ in size between lipedema and control cells. Our results show that the average size of LDs is significantly increased in lipedema spheroids compared to controls (Figure 2C). This supports earlier findings demonstrating increased hypertrophy of adipocytes in normal weight lipedema adipose tissue compared to BMI-matched controls [18, 30, 35, 36]. Interestingly, LDs in lipedema spheroids showed a more heterogeneous picture and contained a higher number of disproportionately large droplets compared to controls (Figure 2D).

3.3 Fibrosis- and inflammation-associated markers

As evidenced by our gene expression studies, augmented deposition of collagen was detectable in histological sections from biopsied subcutaneous adipose tissue from women with lipedema compared to controls (Figure 2E). In line with this, lipedema adipocytes differ from controls by their significantly increased expression of collagen 1 (COL1A1) and interleukin-6 (IL-6) mRNA (Figure 2F), emphasizing again the inflammatory and fibrotic component of lipedema pathology.

Our results from immunostainings reveal elevated numbers of M1 macrophages (evidenced by a granulated signal) in lipedema ASCs compared to controls (Figure 2G and H). This confirms earlier reports, demonstrating increased immune cell infiltration into lipedema adipose tissue [30, 32, 35, 36, 64, 65, 66]. In our study, only low numbers of activated (M2) macrophages, characterized by the selective marker CD163 with additional expression of CD68 were present in lipedema cultured cells, while the highest percentage of M2 macrophages was detected in ASCs from OB donors. Control ASCs were devoid of activated (M2) macrophages. Again, this confirms the general notion that excessive weight gain triggers inflammatory and fibrotic mechanisms, especially in lipedema.

4. Conclusion

The significantly increased transcriptional upregulation of genes involved in the regulation of LD growth distinguishes in vitro differentiated lipedema adipocytes from their healthy counterparts. Although further experimental evidence is needed it may be speculated that the hypertrophic nature of lipedema tissue may be maintained in vitro throughout cultivation.

Increased expression of INSR, GLUT4, and LPL in lipedema adipocytes suggests involvement of insulin signaling in lipedema adipose hypertrophy.

In addition, our data support the notion that lipedema pathogenesis is accompanied by inflammatory and fibrotic conditions. Concerning lipedema-associated fibrosis, the upregulation of PDGFRA might open a new field of research, focusing on a potential imbalance of adipogenic and stromal differentiation at early stages of lipedema development. Thus, future research should concentrate on the initial stages of lipedema, when the earliest signs of adipose tissue alterations appear, and inflammation-related changes are not yet prevalent.

Acknowledgments

This work was supported by the Land Salzburg (project number 20204-WISS/240/3-2019). We thank H. Tempfer (Paracelsus Medical University of Salzburg) for valuable assistance with confocal microscopy. Dr. C. Giera (LipoClinic Dr. Heck) and Dres N. Broer and P. Heidekrueger (LIPOhelp, Salzburg) are gratefully acknowledged for providing adipose tissue samples.