Cancer-associated adipocytes: emerging supporters in breast cancer

Physiologically, adipocytes that constitute breast tissue and encapsulate around the mammary gland, can maintain normal breast morphology and the energy balance. These mature adipocytes are mainly white adipocytes (WAT), with round shape, large unilocular lipid droplets and endocrine functions [12, 13]. CAAs mainly locate at the invasive front of breast tumors, namely adjacent to BC cells. Notably, compared with mature adipocytes, CAAs possess smaller cell sizes, irregular shapes, and small/dispersed lipid droplets [14, 15]. Moreover, CAAs exhibit a modified phenotype characterized by the loss of lipid content, decrease in late adipocyte differentiation markers and overexpression of inflammatory cytokines and proteases [14]. In terms of adipokines, secretions of leptin, adiponectin, interleukin (IL)-6, chemokine ligand 2 (CCL2), chemokine ligand 5 (CCL5) and other cytokines in CAAs are more aberrant than that in mature adipocytes [5, 16]. Most importantly, compared with the distant mature adipocytes, CAAs could gain a more vicious phenotype, and enhance BC growth and metastasis, by a less physical barrier and more aggressive secretome of adipokines (as described in Table 1).

Adipocytes can secrete more than 600 kinds of metabolites, hormones and cytokines, collectively known as adipokines. These adipokines are involved in the regulation of multiple life activities, including fat distribution, insulin secretion, energy consumption, inflammatory reaction [5]. Adipokine-mediated crosstalk network between adipocytes and BC cells favors the proliferation, survival and metastasis of BC [17]. Here, we especially focus on the emerging roles of CAA-derived leptin, adiponectin, IL-6, CCL2, CCL5 and other key adipokines in this reciprocal crosstalk. Uncovering the functions and the mechanisms of CAA-derived adipokines are essential to understand the BC behavior and the establishment of accurate interventions.

Metabolic reprogramming, generally categorized as an emerging hallmark of cancer, is hijacked by BC to meet bioenergetic and biosynthetic demands, the redox balance maintenance, and cell proliferation [64]. The metabolic processes of BC and normal breast tissue are quite different in the aspect of glycolysis, tricarboxylic acid cycle, amino acids, nucleotides and/or lipid metabolism [65]. The nutrient transfer is an underlying mechanism in cancer-stromal cell crosstalk. Glutamine, the most abundant amino acid in the plasma, is the chief nutrient available to tumors, and is of great importance in tumor cell metabolism [66]. Recent studies have recognized that stromal cells could generate glutamine to promote cancer growth [67, 68]. It has been reported that adipocytes secreted glutamine and rescued pancreatic cancer cell proliferation in the absence of glutamine in vitro, suggesting that glutamine transfer is a potential mechanism in adipocyte-induced pancreatic cancer cell proliferation [68]. Glutaminolysis enables cancer cells to reduce NADP⁺ to NADPH, a reaction that is catalyzed by malic enzymes. Besides, NADPH is a required electron donor for reductive steps in lipid synthesis, nucleotide metabolism, and maintenance of reduced glutathione (GSH) [69]. CAAs play a functional role in tumor development through secreted adipocyte-derived factors and exosomes, and through metabolic symbiosis, in which malignant cells absorb lactic acid, fatty acids, and glutamine produced by adjacent adipocytes [70]. Therefore, it might be inferred that the metabolic reprogramming of glutaminolysis induced by CAAs could promote BC cells to regulate the redox state in order to favor proliferation.

Cancer cells prefer to produce adenosine triphosphate (ATP) by glycolysis, which is a less efficient pathway named as the “Warburg effect” compared to oxidative phosphorylation. The “reverse Warburg effect” is observed when cancer cells utilize the energy generated from stromal cells in the TME. In this theory, cancer cells can absorb free fatty acids (FFAs) and glycerol from interstitial adipocytes as energy sources [8, 71]. Adipocytes store triglycerides through adipogenesis, and produce diacylglycerols, monoacylglycerols and FFAs through lipolysis, thus regulating the energy balance of the entire organism [72]. Cancer-associated fibroblasts (CAFs), which are the main cellular members of the TME undergo metabolic reprogramming associated with the reverse Warburg effect [69]. Cancer cells create a “pseudo-hypoxic” microenvironment for CAFs. As HIF-1α promotes glycolysis and provides cancer cells with lactate and glutamate, increased reactive oxygen species (ROS) production in cancer cells induces the uptake of intermediate metabolites of the tricarboxylic acid (TCA) cycle in mitochondria indirectly. Cancer cell-derived ROS could down-regulate the expression of caveolin 1 (CAV1) in CAFs, and the loss of CAV1 also leads to increased ROS levels, thereby stabilizing HIF-1α [69]. The metabolic symbiosis between CAFs and epithelial cancer cells requires each other to express a different monocarboxylate transporter (MCT) isoform. MCT1 conduces to uptake the lactate produced by CAV1-deficient CAFs expressing MCT4, a marker of aerobic glycolysis and lactate efflux. Cancer cells then utilize the lactate to synthesize pyruvate which is an intermediate metabolite in the TCA cycle, while the acidic extracellular space rich in lactate, in turn, results in the formation of pseudo-hypoxic [73]. A previous study indicated that adipocytes adjacent to BC cells underwent phenotypic changes first marked by decreased adipocyte size and lipid content (CAAs), next by acquiring a fibroblast-like morphology (adipocyte-derived fibroblasts, ADFs), still with small lipid droplets. Then, ADFs exhibited increased migratory and invasive abilities and were able to join the center of the tumor, and these cells no longer expressed adipose markers nor exhibited lipid droplets, indistinguishable from other CAFs cell population [74]. Moreover, in addition to that CAAs possess a similar functionality with CAFs, both CAAs and CAFs participate in the complex and dynamic metabolic reprogramming to shape the TME in favorable conditions.

The continuous interaction between BC and adipocytes results in metabolic competition and symbiosis, leading to oncogenic-driven metabolic reprogramming of cancer cells and neighboring adipocytes. FFAs could be released over time from lipid droplets through an adipose triglyceride lipase (ATGL) dependent lipolytic pathway. ATGL was expressed in human BC and was up-regulated by the contact with adipocytes, thus boosting BC progression [75]. Through adipocyte-BC cell co-cultivation, Balaban verified that adipocyte-derived FFAs transferred into BC cells and drove fatty acid metabolism via increased CPT1A and electron transport chain complex protein levels, resulted in fueling BC proliferation and migration [76]. Lipid originated from tumor-surrounding adipocytes could be transferred into BC cells. Utilization of adipocyte-derived lipids and enhanced intracellular trafficking of fatty acids contributed to BC progression, by increased expression of the key lipase ATGL and intracellular fatty acid trafficking protein fatty acid-binding protein 5 (FABP5) [77]. Tumor lipid metabolism is regulated not only by genetic and epigenetic changes in the tumor cells but also by the availability of lipids provided by tumor-surrounding adipocytes. Zaoui et al. showed that mammary adipocytes potentiated the invasiveness of BC cells and the number of lipid droplets in adipocyte cytoplasm, which were mediated by metabolic reprogramming via CD36-mediated exogenous fatty acid uptake [78]. Wang et al. reported the function of the JAK/STAT3 pathway in regulating lipid metabolism, once inhibiting JAK/STAT3 would lead to the block of diverse lipid metabolic gene expression, including CPT1B that encoded the critical enzyme for FAO [25]. Mammary adipocyte-derived leptin could also down-regulate CD8⁺ T cell effector functions through activating STAT3-FAO and inhibiting glycolysis in promoting obesity-associated breast tumorigenesis [27].

In addition to BC cell reprogramming induced by CAAs, adipocytes also undergo reprogramming to become CAAs under the stimulation of BC cells in reverse. Cytokines or exosomal compositions (proteins, miRNAs, lncRNAs and circRNAs) are probably exploited by BC cells as a sort of “signal” to convert the cells in TME such as adipocytes and CAFs into a pro-tumor niche. For example, Wu et al. confirmed that exosomes containing miRNA-144 and miRNA-126 were highly secreted from BC cells co-cultured with adipocytes, leading to promoted metastasis by inducing beige/brown differentiation and reprogramming the metabolism in surrounding adipocytes in vivo [79]. They indicated a new mechanism that tumor-adipocyte interaction reprogramed systemic energy metabolism to facilitate tumor progression [79]. Lee et al. also confirmed the importance of miRNA-regulatory mechanisms of the transition into inflammatory CAAs in BC microenvironment [41]. In this study, mmu-miR-5112 showed the highest expression in the preadipocyte state and the expression was suppressed during the process of adipocyte differentiation. Mmu-miR-5112 mediated the regulation of IL-6 in CAAs, which presented de-differentiated and inflammatory natures in BC TME [41]. This miRNA-based regulatory mechanism involved acquiring the inflammatory phenotypes of CAAs. In addition, using specific shRNA, anti-IL-8 antibody, or reparixin, Al-Khalaf et al. found that the IL-8 signaling inhibition suppressed the active features of CAAs features, including non-cell-autonomous tumor-promoting activities both on breast luminal cells and in orthotopic tumor xenografts in mice [58]. It provided a clear indication that IL-8 was a critical factor in the activation of breast CAAs. Wei et al. reported that plasminogen activator inhibitor type 1 (PAI-1), which had been implicated as a pro-tumorigenic agent in cancers, particularly in cancer metastasis, was required to activate the expression of the intracellular enzyme PLOD2 in CAAs [80]. Furthermore, PLOD2-producing CAAs could remodel collagen alignment during crosstalk with BC cells and further promoted BC metastasis [80]. Therefore, these studies have indicated that there is a complex metabolic network favoring malignant progression that is established between BC cells and adipocytes at the tumor invasive front. The interaction entails a vicious circle, wherein BC cells reprogram adipocytes, which in turn promote tumor progression. Specific factors, such as inflammatory factors, lncRNAs, exosomal compositions and adrenomedullin, have been demonstrated to be involved in this process.

Adipose tissue is surrounded by a basement membrane, which is mainly composed of collagen IV, laminin and fibronectin. Components released by adipocytes possess the capability of remodeling ECM and shaping the BC course [8]. ECM adaptation is essential for the progression and invasion of BC, which could be possibly altered by adipocytes. Interestingly, in an accessible 3-dimensional (3D) co-culture system, Pallegar et al. demonstrated that adipocytes promoted a mesenchymal-to-epithelial transition (MET)-like change in mesenchymal TNBC cells, emphasizing the reciprocal interaction between high adiposity and BC metastasis with adipocyte abundance [81].

Adipocytes in BC can affect the synthesis of matrix proteins and the activation of matrix-related enzymes. Adipocytes can secrete multiple components orchestrating for ECM mechanical support. An immunostaining study of the clinical samples represented that the metastatic human BC tumors had higher levels of matrix metalloproteinase (MMP)-2, compared with non-metastatic tumor tissues; whereas adipocytes around metastatic BC tumors had higher levels of IGFBP-2 than that of non-metastatic sites [55]. Furthermore, co-culture media of mature adipocytes and MCF-7 BC cells showed remarkably up-regulated MMP-2 and enhanced invasion ability of MCF-7 cells [55]. Juárez-Cruz et al. proved that adipocyte-derived leptin promoted the secretion of ECM remodelers, MMP-2 and MMP-9 in a FAK and Src-dependent manner, thereby degrading collagen IV and promoting the rupture of basement membranes [23]. The results suggested that leptin promoted BC into a more aggressive invasive phenotype. Furthermore, another study showed that MMP-2 and MMP-9 suppression could result in the inhibition of BC invasion via emodin by down-regulating the level of CCL5 from adipocytes. It was meaningful to antagonize the secretion of CCL5 from adipocytes for inhibiting BC metastasis. Similarly, adipocytes cultivated with cancer cells exhibited an altered phenotype including MMP-11 overexpression, and played a key role in the acquired pro-invasive effect of tumor cells [14, 82].

Adipocytes also played a vital role in tumor growth at early stages through secretion and processing of collagen VI [83]. Collagen VI, a soluble ECM protein abundantly expressed in adipocytes and enriched in BC lesions, is a key protein for BC ECM. By secreting and processing collagen VI, adipocytes were essential in defining the ECM environment for normal and tumor-derived ductal epithelial cells and contributed significantly to tumor growth [83]. Park et al. investigated that endotrophin, which was derived from collagen VI cleaved by MMP-11, showed a strong stimulating effect on the growth, angiogenesis, and tissue fibrosis of BC [84]. Besides, endotrophin acted as a stimulator to promote the EMT through enhanced TGF-β signaling, contributing to aggressive and high metastatic BC growth [84].

A collagen-dense ECM can potently interact with hormonal signals to drive invasion and metastasis of ER-positive BC [85]. PLOD2 participated in collagen synthesis. Accordingly, the research of He et al. clarified that adipocyte-derived IL-6 and leptin could promote PLOD2 expression by activating the JAK/STAT3, thus facilitating the hardness and metastasis of BC [24]. By constructing a 3D collagen invasion model, Wei et al. showed that CAAs remodeled collagen alignment concerning of PAI-1 or PLOD2 up-regulation during crosstalk with BC cells in vitro and in vivo, further boosting BC metastasis [80]. It is worth noting that the increased matrix stiffness, which can be sensed by many cell types, is a feature of most solid tumors [86]. Yes-associated protein (YAP) function is essential to establish and maintain the tumor-promoting function of CAFs, including stromal sclerosis, invasion and angiogenesis. The high level of YAP in the matrix is maintained by the positive feedback between CAF-driven matrix hardening and mechanical conduction, leading to YAP activation [87]. It has been reported that the Hippo pathway can be activated by stromal stiffness in solid tumor tissues, and central to this signaling is a kinase cascade leading from the tumor suppressor Hippo to the oncogenic Yki (YAP/TAZ in mammals), which is a transcriptional coactivator of target genes involved in cell proliferation and survival [88]. In addition, YAP is activated in CAFs in response to mechanical stress, perturbation of the actin cytoskeleton and ECM stiffness [86, 87]. Besides, the increased expression of YAP in preadipocytes is correlated with the repression of adipogenic differentiation processes [89]. Wang et al. conducted a transcriptomic analysis to find that mesenchymal stromal/stem cell (MSC)-differentiated adipocyte exosomes activated the Hippo signaling pathway involved in two key downstream Hippo proteins YAP and TAZ in MCF-7 cells, while the Hippo pathway blockade could lead to reduced tumor growth [90]. This result mechanistically emphasized that the Hippo signaling pathway was partially responsible for the tumor-promoting effects of MSC-differentiated adipocyte exosomes [90]. Therefore, YAP-dependent matrix stiffening driven by CAFs or YAP activation induced by adipocytes, might lead to pro-tumorigenic YAP activation in BC cells in close proximity to the stiff matrix.

Therefore, these studies demonstrate that adipocyte-derived factors exert roles in ECM architecture remodeling for facilitating the BC cell migration and motility. Abrogating CAA-dependent ECM re-organization of BC will decrease the directional migration of cancer cells within the matrices, thus leading to favorable outcomes.

miRNAs are a unique class of endogenous small, short single-stranded, and non-coding RNA molecules. The most important function of miRNAs is to negatively regulate numerous mRNAs by silencing their target transcripts. Accumulating evidence indicates that the expression of miRNAs during BC progression is aberrantly altered while several miRNAs are recognized to be crucial regulators of adipose microenvironment [91, 92]. Rajarajan et al. found that in vitro co-culture of BC cells with mature adipocytes increased proliferation, migration, and invasive phenotype of BC cells, and 98 miRNAs were differentially expressed in BC cells by using small RNA-sequencing analysis [93]. Among them, miR-3184-5p and miR-181c-3p were found to be the most up-regulated and down-regulated miRNAs, and direct targets were FOXP4 and PPARα, respectively. It was hypothesized by Lee et al. that mmu-miR-5112 treated-adipocytes showed up-regulation of IL-6 by targeting Cpeb1, proposing a miRNA-based regulatory mechanism underlying the process of acquiring inflammatory phenotypes in CAAs [41]. Wu et al. revealed that tumor cells co-cultivated with mature adipocytes exhibited an aggressive phenotype through inducing EMT, while BC cell-secreted miR-155 promoted beige/brown differentiation and remodeled metabolism in resident adipocytes [94]. Interactions between adipocytes and BC cells stimulated cytokine production and drove src/sox2/mir-302b-mediated malignant progression [95]. Besides, miR-141 and miR-146b-5p were two important tumor suppressor miRNAs in BC, and mediated the p16-negative regulation of leptin in adipocytes [96]. In summary, miRNAs are key factors and mediators in intercellular communication. In the progress and metastasis of BC, the miRNA-based regulatory mechanism is closely related to the adipocyte-secreted factors.

The peritumoral adipocytes induce an immune remodeling in favor of cancer aggressiveness and progression, mainly including cells of monocytes, macrophages, neutrophils and T cells [97, 98]. Crown-like structures (CLSs), composed of macrophages encapsulating dead or dying adipocytes, are a histologic hallmark of the pro-inflammatory process. In this process, adipose tissue contributes to the increased risk and worse prognosis of BC in obese, postmenopausal patients as well as some normal-weight women [99]. CD68⁺ and/or CD163⁺ TAM infiltration, coupled with CLSs, is present in adipose tissue nearby the BC lesion, and is associated with various clinicopathologic parameters of BC [17]. Besides, the adipocyte-derived chemokines, such as CCL2, recruited macrophages to the TME and further triggered BC development [48, 84, 100, 101]. In addition, adipokines are able to induce angiogenesis in driving tumor progression through macrophages. Several adipokines including leptin, IL-6, and TNF-α are capable of increasing VEGFA expression in THP-1 macrophages co-cultured with adipocytes. Compared with THP-1 cells cultured alone, the medium of THP-1 cells previously exposed to human adipocytes stimulated endothelial tube formation more significantly [102]. Therefore, revealing the underlying mechanisms of adipocyte-driven immune cells and immune cascades is necessary to understand the role of adipokines in TME.