As cells occupying the largest portion of adipose tissue, adipocytes engage in typical functions related to the metabolic and endocrine roles of adipose tissue. In response to energy flux or external stress signals, adipocytes maintain metabolic homeostasis or participate in other regulatory pathways, such as energy storage and consumption, appetite regulation, insulin sensitivity, and inflammatory responses, through various endocrine factors. In addition, continuous exposure to excessive nutritional and stressful environmental changes leads to metabolic diseases due to adipocyte dysfunction, such as insulin resistance and dyslipidemia. Obesity, which is caused by continuous and excessive energy influx in adipose tissue, is typically accompanied by hypertrophy and hyperplasia of adipocytes, resulting in dysfunction of the adipocytes themselves and further dysfunction of surrounding cells. Moreover, the dysfunction of these adipocytes leads to aggravation of the disease through abnormal interactions with other diseased tissues. A recently identified example is tumorigenesis, which has the potential to increase due to obesity. The International Agency for Research on Cancer has reported that higher amounts of body fat are associated with an increased risk of a number of cancers [12].

In addition to changes in size and functionality of adipocytes, adipocytes are highly plastic cells [13]. Adipocytes can be converted into other types of cells according to the physiological/pathological conditions of the individual [14,15]. The most common event is the dedifferentiation of adipocytes that revert to progenitor- and fibroblast-like cell types. This phenomenon is commonly observed in obesity-induced fibrosis of adipose tissue [14,15], tissue regeneration processes [16,17], and interactions with certain tumors [13]. Furthermore, it has been reported that dedifferentiated adipocytes can induce angiogenesis and increase mature vasculature density [18]. Although high plasticity makes it difficult to utilize in in vitro experiments of primary adipocytes, research on dedifferentiation observed in various pathological processes and its accompanying signaling pathways will greatly contribute to understanding related diseases and developing therapeutics.

Therefore, adipose tissue is an organ that actively interacts with itself and other tissues, and research on its function is important for understanding various fat-related diseases as well as basic physiological phenomena such as fat and glucose metabolism and endocrine signal pathways. In the medical field, in vitro modeling is required to improve the understanding of basic research to elucidate the unknown functions of adipose tissue and various diseases related to adipose tissue. A technology that can enhance the functionality of differentiated adipocytes like in vivo conditions or maintain the phenotype of primary adipocytes for a long time in in vitro culture is required. Adipocytes are mainly cultured in 3 ways. The most used cell source for research on adipose tissue is adipogenic differentiated adipocytes [19,20]. Various types of progenitor cells (cell lines, mesenchymal stem cells [MSCs], or the stromal vascular fraction [SVF]) are used to artificially differentiate cells. However, differentiated adipocytes are not morphologically and functionally like in vivo adipocytes, and it is difficult to maintain a long-term phenotype. Another culture method is ex vivo tissue culture, or the culture of primary adipocytes extracted from adipose tissue [20]. However, it is also difficult to maintain fat cells for a long time without dedifferentiation. Moreover, the lack of co-culture technology for interactions between various cells is still an obstacle to studies using adipocytes.

The SVF, which can be obtained from adipose tissue, contains heterogeneous cell populations such as mesenchymal progenitor/stem cells, preadipocytes, endothelial cells, pericytes, T cells, and M2 macrophages [21]. Representative resident cells include immune cells. These immune cells are involved in the systemic immune system through the secretion of cytokines while interacting with external signals or surrounding cells [22]. In particular, resident microphages in adipose tissue play a key role in maintaining metabolic homeostasis and insulin sensitivity. In healthy adipose tissue, anti-inflammatory immune cells (typically M2 macrophages, eosinophils, and regulatory T cells) suppress excessive immune responses and induce healthy cell reactivity through proper adipokine balance. However, when exposed to a stressful environment caused by excessive energy accumulation or abnormal cellular responses, inflammatory immune cells (typically M1 macrophages, CD4 and CD8 T-cells, activated natural killer cells, or neutrophils) are activated and collect in adipose tissue [23]. The composition and functionality of the cells become more inflammatory. Remodeling of these immune cells can lead to insulin resistance, which can lead to more serious secondary diseases [24].

The difficulty in reproducing the complex and systemic reactivity has led to the use of laboratory animals for the study of adipose tissue. In vitro models are being developed to replace in vivo experiments and to conduct more relevant research. To model the pathological state of adipose tissue in vitro, interactions between adipocytes and immune cells are required. In many studies, immune cells activated through inflammation-inducing substances such as lipopolysaccharide or pro-inflammatory cytokines are usually used to induce an inflammatory response of adipocytes in vitro and reproduce insulin resistance [25-27]. However, the inflammatory response caused by a single artificial stimulus has limitations in mimicking the dynamic inflammatory response of in vivo adipose tissue.

Adipose tissue, a representative endocrine organ, has a well-developed vascular structure that facilitates interactions with other tissues [28]. The development of blood vessels in adipose tissue is not only for the maintenance of adipose tissue, which can grow up to 10 times its original size for energy storage, but also for the acceptance and release of various bioactive molecules such as adipokines, cytokines, nutrients, oxygen and metabolic wastes, and other cells such as immune cells and progenitor cells [29]. In addition, blood vessels contain perivascular/endothelial progenitors, which can act as a niche for adipocytes [30]. Therefore, blood vessels are structurally and functionally transformed according to the systemic metabolic activity and disease state of adipose tissue. Although the mechanisms underlying the changes of blood vessels in fat are still elusive, they can be triggered by several dysfunctional modulations, such as hypoxic signals from an out-of-balance energy state and unusual expansion of adipose tissue, pro-inflammatory signals from immune cells, and fibrosis of the ECM [31,32]. In obesity, the expansion of adipose tissue causes a lack of vascular structure, which induces hypoxic signals. Induction of inflammation due to hypoxia causes the collection and accumulation of pro-inflammatory immune cells and dysfunction of surrounding cells [32]. A variety of dysregulated factors stimulate dysfunction of vascular endothelial cells. In addition, excess free fatty acids released from obese adipocytes can induce inflammatory responses in endothelial cells and lead to dysfunction [33].

For a physiological and pathological understanding of adipose tissue, the role of vascular endothelial cells is important. In addition, proper integration of vascular tissue is required for long-term maintenance of adipose tissue in vitro models. For this purpose, various vascularized adipose tissues have been prepared. As an attempt, vascularized adipose organoids were fabricated through in vitro differentiation of cells using SVF containing various adipose-derived cells [34,35]. As another example, the integration of microfluidics enabled improved co-culture of adipocytes and vascular endothelial cells, which allowed the development of an organ-on-a chip that reproduced highly vascularized adipose tissue [36]. These various attempts reproduced the interactions between adipocytes and vascular endothelial cells in vitro and showed potential as more relevant in vitro models.

In adipose tissue, fibroblasts play an important role in maintaining adipose tissue homeostasis. Fibroblasts, especially fibroblast-specific protein-1-positive fibroblasts, resist obesity induced by a high-calorie diet and cause adipose tissue loss [37]. This is achieved by fibroblasts regulating the turnover of adipocytes and preadipocytes, preventing maintenance of the pool of preadipocytes and regulating their differentiation into adipocytes. Fibroblasts also induce the expression of transcription factors and related sub-markers for adipogenic differentiation of adipose stromal/progenitor cells (ASCs), thus facilitating differentiation even at low densities [38]. Fibroblast differentiated from ASCs also play a major role in the wound healing process [39]. ASCs in adipose tissue induce differentiation into fibroblasts, which are major effector cells for wound healing, and secrete various ECM and growth factors. In addition, adipocytes are converted into fibroblasts through lipolysis in adipocytes around the wound, and they are involved in skin repair [17]. In an obese environment, it is known that preadipocytes with a myofibroblasts phenotype accumulate ECM and upregulate profibrotic cytokines, along with the upregulation of pro-inflammatory factors of fibroblast-like preadipocytes in adipose tissue [40]. Therefore, fibroblasts and myofibroblasts found in adipose tissue cause the onset of metabolic diseases with the development of obesity.

Collectively, fibroblasts not only maintain adipose tissue homeostasis by regulating the turnover of adipocytes, but also the surrounding microenvironment, particularly pathological factors such as obesity, which promotes the differentiation of precursors such as adipocytes or ASCs into fibroblasts and participates in inflammatory signaling by increasing the collection and proliferation of fibroblasts. The various functionalities of fibroblasts can be effectively utilized in in vitro experiments on adipocytes. In particular, fibroblasts could facilitate the differentiation of preadipocytes into adipocytes in vitro and induce a more in vivo-like phenotype. In addition, since fibroblasts are closely involved in the proliferation, differentiation, and maintenance of adipocytes, they are a factor to consider for the successful reproduction of adipose tissue.

The ECM is a non-cellular component found in all tissues. It is an extensive 3D network composed of polymeric proteins such as collagen and elastin, proteoglycans including glycosaminoglycans (GAGs), and glycoproteins such as laminin and fibronectin [41]. Collagen is a major component of the ECM that occupies a significant portion of adipose tissue. It serves to support mechanical stress caused by external force or the expansion of adipocytes. Moreover, it is significantly involved in the adhesion, proliferation, and survival of various cells. In the normal ECM, the types and amounts of collagen are properly balanced, but in disease states such as obesity, fibrosis and dysfunction occur due to the accumulation of various types of collagens, such as collagens type I, III, V, and VI [9]. These changes increase the strength of the adipose tissue, limiting its ability to expand. In particular, an increase in type VI collagen and its cleavage product, endotrophin, can be observed in obese adipose tissue, which is involved in metabolic dysfunction such as insulin resistance and even tumorigenesis [4].

In obesity, along with the accumulation of collagen, the number of proteoglycans composed of a core-protein and multiple GAGs chains, which are characterized by a high degree of sulfation and serve as charge-barriers in the ECM [42], also increases [43]. Proteoglycans have multiple ligands, through which they interact with a variety of cytokines, cell surface receptors, adhesion proteins, and other ECM proteins. Various proteoglycans such as lumican, perlecan, decorin, aggrecan, beta-glycan, and biglycan are abundantly present in adipose tissue, are involved with several soluble factors, and take part in cell adhesion, metabolism, and the immune response. As in the case with collagen, it is known that metabolic disorders such as diabetes are closely related to the composition and content of proteoglycans [43,44].

Multiple tissue receptors, including integrin [45], CD44 [46] and CD36 [47], bind to various ECM components, which are implicated in several physiological pathways. These receptors, which enable cell-cell and cell-ECM interactions, are regulated by the surrounding microenvironment to respond appropriately to the maintenance of tissue homeostasis. For example, integrins are known to be involved in the development of insulin resistance. In the obese state, integrin decreases the expression of Glut4, thereby reducing insulin sensitivity and reducing the phosphorylation of AKT, which is involved in glucose uptake [48].

Taken together, the ECM is dynamically regulated by various cells and signals, showing specificity according to the pathological condition of each tissue. Therefore, an analysis of the ECM in various conditions can represent the inflammatory state of an individual. In obesity, adipose tissue greatly expands in volume by storing excess energy, resulting in a change in the ECM. The phenomenon is due to several factors mentioned above. Comprehensive inflammatory signals caused by obesity consistently lead to low levels of systemic inflammation and further exacerbation of ECM fibrosis, contributing to the development of serious secondary diseases such as diabetes, cardiovascular disease, and tumors.

Spheroids are cellular aggregates and one of the most versatile ways to culture cells in 3D (Fig. 2A). They are better able to recapitulate not only the in vivo morphology, but also cell-cell interactions and tissue architecture, which makes them more physiologically relevant and predictive than 2D models. The formation of scaffold-free spheroids relies on ECM secreted from the cells themselves, which encourages preliminary aggregation through the process of self-assembly. Since the ECM secretion and junctional formation of adipocytes are lower than those of preadipocytes, forming adipocyte aggregates is generally less efficient than forming preadipocyte spheroids. Thus, in several studies, white adipose tissue has been modeled by forming preadipocyte spheroids, followed by subsequent differentiation into adipocytes.

Several technical methods, including the hanging drop, liquid overlay, and magnetic levitation techniques, have been used to generate adipose spheroids. The hanging drop culture comprises a small droplet of cell culture medium suspended by gravity, allowing cellular aggregates growth in 3D. Immortalized human and mouse preadipocytes, as well as primary preadipocytes (SVF and adipose-derived stem cells [ADSCs]) have been incorporated into hanging drop cultures and successfully differentiated into adipocytes [77-80]. In a recent study, Klingelhutz et al. [78] developed adipocyte spheroids by placing preadipocytes in drops on the inside cover of a Petri plate, transferring the spheroids to ultra-low-attachment wells, and providing differentiation cues in the medium (Fig. 3A). The spheroids exhibited unilocular lipid droplets seen in mature adipocytes reaching a volume of ~0.1 cm³ and showed less adiponectin secretion than 2D culture. The adiponectin secretion of adipose spheroids was significantly reduced when exposed to the environmental toxins (indoxyl sulfate and PCB126) for 30 days, which was not achievable in 2D culture, raising the possibility of using 3D adipocyte spheroids as a reliable system to study long-term effects of drugs or toxins [78]. Adipocyte spheroid systems have been similarly developed using a liquid overlay technique, also known as static suspension culture, which is based on a non-adhesive surface that prevents cellular adhesion by coating a culture plate with agarose [81] or elastin-like polypeptide-polyethyleneimine [79]. Magnetic levitation was also utilized to form ADSC-based adipocyte spheroids by mixing the ADSCs with magnetic particles (poly-lysine-based magnetic nanoparticles [80]) and subjecting them to magnetic force during culture to make cells stay levitated against gravity.

Scaffold-free adipose spheroids based on patient-derived ADSCs can be utilized to study adipose tissue diseases because of the dependence of cell adherence on autocrine ECM molecules rather than on artificial substrates, giving rise to realistic patterning of diseased cell differentiation and cell-ECM interactions. Al-Ghadban et al. [81] developed patient-derived ADSC spheroids for modeling lipedema, a connective tissue disorder characterized by fibrosis, hypertrophic adipocyte, inflammation, and leaky blood. Spheroid cultures of ADSCs extracted from lipedema patients and healthy individuals were differentiated into adipocytes. The spheroids derived from lipedema patients showed higher matrix metalloproteinases (MMP11) expression than healthy spheroids, which may explain the increase in fibrosis in lipedema adipose tissue [81]. However, they did not show a higher adipogenic differentiation potential related to hypertrophic adipocytes, as has been documented in lipedema patients, possibly because of low diffusion of nutrients and oxygen throughout the entire spheroid due to the absence of vasculature.

To overcome this drawback, several attempts have been made to generate a vascularized adipose spheroid. 3T3-L1 preadipocytes aggregated with murine endothelial cells (bEND.3) showed the formation of a vascular-like network concomitantly with lipogenesis in perivascular cells upon adipogenesis induction [80]. Spheroid culture of SVF also displayed a vascular-like structure without using exogenous endothelial cells [80]. However, the absence of biomaterial-based scaffolds makes it challenging to create functional vasculature due to the lack of angiogenic signals. Muller et al. [34] solved this problem by developing co-differentiation methods for SVF adipocytes/endothelial cells. The incorporation of SVF spheroids in Matrigel and 4 days of endothelial growth induction in EGM2 medium provided the angiogenic signals to form a highly branched vasculature (Fig. 3B). Subsequent differentiation of perivascular cells using an optimized adipogenic cocktail (without IBMX, rosiglitazone, and indomethacin) successfully preserved the vascular network and induced adipogenesis in a spheroid. The adipose spheroids generated using the optimized method showed much longer and highly branched vascular networks than those generated using the traditional method. These models can serve as useful tools to study interactions between adipocytes and endothelial cells and offer a possibility of long-term maintenance of adipose spheroids by supplying nutrients and oxygen to the sphere center via the vascular network.

Recently, the presence of SVF immune cells in adipose spheroids from SVF was demonstrated by Taylor et al. [82]. As the SVF is composed of a heterogeneous assortment of cells, even including immune cells, it was predicted that adipose organoids generated from SVF might contain immune cell populations. The researchers found that 20%-30% of differentiated adipose spheroid cells were CD45+ immune cells, and of the immune cell populations, macrophages (CD11b+F4/80+) accounted for 60% to 70% [82]. This finding implies that resident immune cells can be maintained in 3D spheroids, even in the absence of M-CSF supplementation in the culture medium, whereas SVF macrophages could not be maintained in 2D culture [82]. This study focused efforts on the evaluation of changes in the lipidome of spheroid in response to dietary interventions, rather than investigating the interaction between adipocytes and immune cells; however, it suggests that scaffold-free adipose spheroids could be utilized to model many pathological conditions associated with immune responses in adipose tissue.

Three-dimensional porous or permeable biomaterials promote the recapitulation of the structure and functionality of native tissues through cell-to-ECM and cell-to-cell interactions (Fig. 2B). For decades, biomaterial scaffolds such as collagen [83,84], gelatin (Fig. 3D) [85,86], polymer mixtures [87-89] and decellularized ECM [90,91] demonstrated roles in the improvement of adipogenic differentiation efficiency or the maintenance of mature adipocytes. Researchers have recently started to use this strategy for the long-term culture of mature adipocytes in vitro. Louis et al. [83] developed a collagen microfiber-based mature adipocyte culture method, which ensured the maintenance of unilocular mature adipocytes for up to 2 weeks, exhibiting higher cell viability and perilipin expression than 2D-cultured adipocytes. Another study [89] suggested a biofunctionalized hydrogel with components of adipose ECM as a relevant in vitro model for adipose tissue (Fig. 3E). To provide the adipocytes with spatial and chemical cues for effective maturation, they developed a cross-linked hyaluronic acid scaffold containing key adipose ECM components. The 3D cultured preadipocytes in the fabricated scaffolds showed improved in vivo-like morphology and metabolism. However, these strategies are generally employed for adipose tissue regeneration; thus, the development of 3D scaffolds has been focused on the enhancement of in vivo stability and the functional lifetime, supporting adipocytes inside the body.

A transwell system is a cell-culture device that provides independent access to both sides of a permeable plastic membrane (Fig. 2C). Harms et al. [92] recently established a new primary adipocyte culture method using transwells, named membrane mature adipocytes aggregate cultures (MAAC). Culturing freshly isolated mature adipocytes underneath the permeable membrane of the transwell system enhanced the maintenance of the integrity and functionality of adipocytes for up to 2 weeks without dedifferentiation (Fig. 3C). Considering that functional maintenance of mature adipocytes is highly required to study long-term drug responses in vitro, the MAAC-based culture system exhibits great potential to serve as a promising drug test platform to treat metabolic diseases. The MAAC method allowed the co-culture of adipocytes with macrophages in a different compartment, which is also advantageous for inducing pathological conditions such as inflammation-associated metabolic disorders. This improved platform can be an experimental tool capable of increasing the in vitro accessibility of primary adipocytes for various applications, such as the pathologic modeling of obesity or drug screening. However, there is still an insufficient understanding of the multi-cellular interactions and dynamic ECM remodeling observed in obese adipose tissue.

Over the last decade, 3D printing emerged as a promising technology to generate the adipose tissue construct for tissue regeneration, due to its efficient control of spatial positioning of multiple cells using various bio-inks (Fig. 2D) [63,64,66-68,93,94]. Cho et al. [66] developed a flexible adipose-vascular tissue assembly composed of a synthetic polymer-based substructure and decellularized ECM bio-inks using planar 3D printing (Fig. 3F). The formation of vascular structures was induced by printing human umbilical vein endothelial cells with vascular tissue-derived ECM bio-ink; meanwhile, the functional maintenance of adipocytes was successfully achieved using adipose tissue-derived ECM bio-ink. To enhance the long-term in vivo retention efficiency, a polycaprolactone module holder was additionally used, resulting in maintenance of its original shape over 7 days. Another study reported that MSCs bioprinted using GelMA bio-ink were robustly differentiated into the adipogenic lineage (Fig. 3G) [68]. The researchers demonstrated that the pore size of the scaffold determined the differentiation and infiltration efficiency, highlighting the importance of the physiochemical properties of bio-ink. This approach can be exploited to develop adipose tissue co-culture platforms to model pathological conditions in vitro [26,65]. Crosstalk between adipocytes and breast cancer cells was recapitulated using 3D bioprinting technology [65]. The spatially controlled deposition of adipocytes and breast cancer cells enabled the generation of direct or indirect contact between the 2 types of cells, showing how adipocytes contribute to cancer aggressiveness. Insulin resistance was also modeled by patterning ADSCs and activated macrophages as a platform to study drug sensitivity [26]. Significant limitations of 3D bioprinting remain in terms of pattern resolution; however, its high flexibility and versatility would offer great advantages in creating complex adipose tissue models. These findings suggest that 3D bioprinting technology may be suitable for the regeneration studies with artificial adipose tissue through improved vascular co-culture and structural simulation. These results also show the possibility of pathological modeling through the mixing of various materials. However, even with this advanced technology, some obstacles remain. Non-cell-friendly printing processes and selection-limiting types of bio-inks and their properties are known as limitations of 3D bioprinting, which are still being improved. In addition, unlike the various evaluations of adipose tissue performed on other platforms, the in vitro maturation of printed artificial tissues and extensive functional evaluation thereof are lacking in 3D bioprinting technology.

A microphysiological system (MPS) is an integrated 3D culture platform designed to recapitulate the key organ features (Fig. 2E). This technology presents advantages over other in vitro systems by mimicking the precisely controlled physical, chemical, and biological microenvironment of organs. Because adipose tissue is a dynamic organ composed of a variety of cell populations, which actively interact with other organs by secreting various adipokines, MPS technology has been suggested as a promising tool for emulating its complex physiology [36,76,95-97] and pathology [25,70-72,74,75,98,99].

Rogal et al. [76] developed a two-channel adipose tissue MPS (white adipose tissue-on-a-chip) to solve the issue of low maintenance of mature adipocyte culture. The mature adipocytes cultured within a microfluidic channel were functionally maintained for 36 days while embedded in collagen hydrogel due to the efficient nourishment from vasculature-like perfusion. Isoproterenol-stimulated lipolysis was also observed in the device, showing its potential as a platform for pharmaceutical development and testing (Fig. 3H). Even though media perfusion enabled the long-term maintenance of mature adipocytes in the device, functional regulation of adipocytes through crosstalk with vascular endothelial cells was not achievable in the study. Yang et al. [36] developed a vascularized adipose tissue MPS using adipocytes and endothelial cells. To accomplish adipogenic differentiation and vascularization under the same media conditions, the composition of cell culture media was extensively optimized in the microfluidic device. The adipose tissue generated in the optimized media revealed the formation of an interconnected vascular network (Fig. 3I); however, the cellular function of the vascularized adipose tissue platform was not fully investigated. For disease modeling, MPS has been used to recapitulate the complex pathophysiological features of adipose tissue in disease conditions. An insulin-resistant adipose tissue model was developed using a two-channel microfluidic device, where mature adipocytes were cultured in a microchannel interfaced with a media perfusion chamber [72]. Insulin resistance was efficiently induced by treating mature adipocytes with tumor necrosis factor-alpha, as confirmed through Akt phosphorylation, Glut4 expression, and glucose uptake. Furthermore, the therapeutic efficacy of rosiglitazone, an antidiabetic drug, was evaluated in an MPS device, demonstrating its potential for the study of metabolic diseases and the discovery of new therapeutics. Similarly, Liu et al. [74] reproduced the insulin resistance of adipocytes by culturing adipocytes and immune cells together in adjacent spaces separated by a porous barrier. The insulin sensitivity of the resulting micro-artificial adipose tissue was verified by observing glucose uptake in the presence of insulin and changes in adipokine expression. The currently developed adipose tissue MPS systems have demonstrated the benefits of microfluidic systems, including the long-term cultivation of adipocytes without a loss of viability by supplying medium through flow, and dynamic communication with different types of cells through the different chambers. Although various types of microfluidic devices have been proposed for adipose tissue-related studies, there are still difficulties in inducing cell-ECM interactions or complex in vivo-like inflammatory reactions.