The role of three-dimensional in vitro models in modelling the inflammatory microenvironment associated with obesity in breast cancer

The adipose microenvironment consists mainly of adipocytes alongside precursor adipose stem cells, fibroblasts, myofibroblasts, endothelial and immune cells. Adipocytes can adjust their number and morphology in response to energy balance via the processes involved in lipid uptake, lipolysis, and differentiation of pre-adipocytes [20]. During obesity, mature adipocytes expand in size and become hypertrophic, resulting in the secretion of inflammatory adipokines, growth factors and increased collagen production [21, 22]. The obese adipose microenvironment has been shown to favour cancer cell proliferation, promote angiogenesis and to reduce anti-cancer treatment efficacy (Fig. 1) [21, 22].

In an adipose-rich TME, cancer cells can alter the fate of adipocytes, forming cancer-associated adipocytes (CAAs) via adipocyte delipidation and atrophy [23, 24]. Reports have shown CAAs to be associated with increased production of pro-angiogenic and pro-inflammatory growth factors and cytokines such as IL-6, CXCL1 and TNF-α [25, 26], which can sustain angiogenesis and tumour progression [27]. Tumours may also benefit from CAAs as they provide fatty acids as an energy source to cancer cells. The transfer of lipids from adipocytes to cancer cells via lipolysis, has been previously reported in breast, ovarian and prostate cancer [24, 28, 29]. This lipolysis process results in the accumulation of fibroblast-like cells and a desmoplastic stroma, which suggests that some CAFs located at the adipose-tumour border might be derived from dedifferentiated adipocytes [23].

Macrophages contribute to the inflammatory state of the TME as one of the most abundant immune infiltrates [30, 31]. Greater infiltration of adipose tissue macrophages (ATMs) has been correlated with obesity, tumour size, recurrence, and development of tamoxifen resistance [32]. A characteristic of obesity is the presence of dying and necrotic adipocytes; identified by the presence of CD68-positive macrophages surrounding these dead cells, forming histological hallmarks known as crown-like structures (CLS) (Fig. 2) [33, 34]. CLS formation may result from increased levels of ATMs, contributing to both local and systemic inflammation. The presence of CLS is increased in breast AT in obese BC patients [12, 33]. However, CLS have also been reported in women with a normal BMI, who are classified as ‘metabolically obese’. For example, Iyengar and colleagues [12] identified an inflammatory state; characterised by the presence of CLS, associated with elevated aromatase levels, adipocyte hypertrophy and systemic metabolic dysfunction in women with a normal BMI. Therefore, it is still unknown whether the pro-inflammatory adipose microenvironment in obesity leads to increased CLS formation or vice versa.

Increased frequencies of CLS in individuals with obesity have been shown to negatively influence BC recurrence rates, survival, and therapy response [32, 35, 36]. CLS in obese AT has been shown to exhibit a mixed macrophage phenotype; characterised by the co-expression of surface markers that discriminate M1- such as CD11c, and M2-like macrophages such as CD206 or CD163 [37]. Furthermore, CLS localisation within the adipose microenvironment may play a role in macrophage phenotype. A recent study within our group showed that macrophages at the adipose-tumour border expressed both CD16 and CD32b, which may indicate a metabolically dysregulated macrophage phenotype [38].

CAAs and ATMs are central components of the BC niche. Nevertheless, their roles in the development of obesity-associated BC remains unclear. Understanding the mutual interaction between cancer cells, AT, and infiltrating immune cells could allow the possibility of overcoming AT-mediated therapy resistance and lead to novel therapeutic approaches.

Spheroids represent the most used liquid-based 3D tumour culture system, capable of recreating tumour characteristics that are not seen with traditional 2D cultures, such as hypoxia [44, 45]. During spheroid development, cells adhere to each other to form a spherical cell mass allowing cell–cell interactions. Spheroids can be obtained from monocultures or co-cultures with stromal cells. Spheroids are routinely cultured in low attachment plates under matrix-free conditions. However, increasing studies have furthered this model by embedding spheroids in a permeable matrix (hydrogel), allowing for interactions with the ECM to be investigated.

Most 3D models utilise a porous, permeable ECM as their main component to model the TME. The ECM plays important roles in cell adhesion, migration, invasion, and signalling. Therefore, modelling the ECM allows for preservation of the molecular and phenotypic characteristics of these multicellular structures. Multiple cell types can be embedded into a hydrogel matrix, which is polymerised via incubation at 37°C and maintained in culture over days, up to weeks. The most common hydrogels are extracted matrix proteins such as rat-tail collagen and Matrigel®. However, synthetic hydrogels, produced with polymers such as polyethylene glycol (PEG), are increasingly being used as they allow for control over matrix properties [51]. Synthetic hydrogel components can be modified to mimic ECM properties of distinct TMEs. For example, a recent study developed an adipose-derived hydrogel that supports the adipogenic differentiation of adipose-derived stem cells (ADSC’s) [52].

Hydrogels offer a tuneable system, where stiffness and matrix proteins can be monitored. Once embedded in the matrix, cells have been shown to form physiologically relevant structures, remodel matrix, and interact with each other within a 3D environment. An example of this is the generation of breast duct-like structures. In 2017, Carter et al. embedded isolated populations of primary luminal and myoepithelial cells in a hydrogel matrix, where they formed duct-like structures in 3D [53]. Therefore, hydrogels are a versatile tool used to study cell–matrix and cell–cell interactions in a 3D environment, allowing for the formation of physiological multicellular structures.

However, ECM-derived hydrogels such as collagen type I and Matrigel® are disadvantaged by their variability, limited stiffness range and structural instability caused by hydrogel shrinkage and degradation during culture. Furthermore, recovery of cells from these hydrogels for down-stream applications requires proteolytic digestion that may negatively impact cells.

Organotypic models can be modified to incorporate multiple cell types and use different matrices. A study in 2009, developed a novel organotypic model of normal breast and the BC microenvironment [60]. This was the first study to report the co-culture of three major components of the normal and malignant breast: luminal cells, myoepithelial cells, and stromal fibroblasts in a collagen matrix. This model exhibited co-unit duct-like structures recapitulating normal and DCIS breast, where myoepithelial cells localised around the luminal population. Malignant co-units displayed disruption of the basement membrane and loss of β4-integrin, characteristic of DCIS. Interestingly, addition of normal fibroblasts did not influence co-unit formation. Whereas inclusion of CAFs resulted in disruption of co-unit organisation. These results support the value of these models in dissecting normal and tumour cell behaviour to uncover mechanisms of BC progression. However, this system involved the co-culture of three cell populations, which does not fully recapitulate the breast TME.

Recent studies have incorporated adipocytes into organotypic models to better reflect obesity associated BC. Bougaret et al. developed a co-culture model of ER + BC cells with mammary adipocytes isolated from women of normal weight, overweight and obese [22]. Through this model, they demonstrated that the anti-proliferative effects of tamoxifen were no longer seen when obese mammary adipocytes were present. Similarly, in another study, co-culturing MCF-7 cells with mammary adipocytes exposed to high glucose resulted in decreased tamoxifen responsiveness of BC cells [61]. However, sensitivity was regained by inhibiting adipocyte secretion of IL-8 [22]. Additionally, Morgan et al. proposed an organotypic model of the mammary duct to investigate response to the aromatase inhibitor (AI) anastrozole in obese patients compared to those of normal weight [40]. Results showed that MCF-7-derived ducts co-cultured with obese-derived stromal cells exhibited reduced sensitivity to anastrozole compared to lean co-cultures. Interestingly, this difference was not seen in a conventional 2D culture system, highlighting the value of 3D in vitro models.

Patient-derived organoids can be grown from mechanically or enzymatically dissociated biopsies, keeping tumour cells within the same TME. Therefore, they represent an attractive alternative to investigate therapeutic compounds for BC. These systems (PDTOs) have been employed to evaluate the response to clinically relevant antitumor therapeutics such as paclitaxel, tamoxifen, trastuzumab, and combinations thereof [69, 70]. Breast cancer organoids have also been used to explore mechanisms underlying tumour cell invasion and metastasis. In 2013, Cheung and colleagues used PDTOs grown in collagen type I matrices to identify characteristics of invasive cancer cells in primary breast tumours [67]. They found that among the major BC subtypes, invasion was induced by cancer cells expressing basal epithelial genes such as cytokeratin-14 and p63 [67].

In 2020, Campaner et al., described a 3D PDTO culture system which recapitulated the histological features of four different subtypes of primary BCs [70]. Primary human epithelial cells self-organised to form complex ductal and lobular morphologies. However, PDTOs derived from a HER2-enriched carcinoma, did not show HER2 staining. In the parental tumour, 50% of cancer cells showed high HER2 staining, thus this model was not capable of maintaining all characteristics of the tumour of origin.

Organoid cultures are mainly epithelial cultures devoid of stromal cells and tumour vasculature. Therefore, further research is required into co-culturing organoids with other cell types to mimic genetic and phenotypic characteristics of the TME. Koledova et al., developed a co-culture system containing murine mammary organoids and CAF spheroids [71], though this has yet to be investigated in human cells.

At present, there are no published reports of organoid models investigating the inflammatory obese microenvironment in BC. Therefore, further studies are required looking at co-culturing organoids with other major cell types of the breast TME such as macrophages and adipocytes to investigate BC associated with obesity.

To model the inflammatory obese microenvironment in BC, previous studies have cultured primary AT as explants to develop co-culture systems. A recent study showed that extracellular vesicles isolated from obese adipose tissue explants, increased the proliferative potential of MCF-7 and MDA-MB-231 cells in vitro [75]. Interestingly, the proliferative effects of extracellular vesicles on MCF-7 cells were ERK/MAPK dependent; whereas, migration of MDA-MB-231 cells was dependent on activation of the PI3K/AKT. Similarly, an earlier study subjected breast tumour and pre-neoplastic cells to conditioned medium collected from AT from healthy individuals or BC patients [76]. Conditioned medium from tumour AT explants increased proliferation of both tumour and non-tumour breast epithelial cells and reduced adhesion of MCF-7 cells.

In 2021, Humayun et al., developed an organotypic microfluidic model to investigate cancer-vasculature interactions during cancer extravasation in TNBC [83]. This system employed a tubular endothelial vessel generated from induced pluripotent stem cell-derived endothelial cells within a collagen-fibrinogen matrix. Breast cancer cells were then injected through and cultured along the lumen of the endothelial vessel. This approach identified cancer-vascular crosstalk involving increased levels of secreted factors such as IL-6, IL-8, and MMP-3. However, this model has only looked at one element of the breast TME and future studies should focus on the use of multicellular microfluidic approaches.

Recent advances have been made utilising microfluidic systems to model BC-stroma interactions [82, 84]. Lugo-Cintron et al. developed a multicellular microfluidic organotypic model of BC cells and fibroblasts in collagen or fibronectin-rich matrixes to investigate how different ECM compositions and fibroblast compositions impacted BC migration [84]. The greatest migration was observed when TNBC cells were co-cultured with CAFs, however, culture in a fibronectin-rich matrix also demonstrated increased tumour migration. Interestingly, they show that MMP inhibitors are not effective when CAFs are present, suggesting why preclinical studies based on in vitro 2D cultures have failed. This highlights the potential of such model systems to assist clinical trials.

Berger Fridman and colleagues developed two breast TME models using a high throughput microfluidic system [85]. Their ‘pro-inflammatory’ TME model consisted of BC cells with fibroblasts, M1-like macrophages and activated T cells. In comparison, the second ‘immunosuppressive model’ contained M2-like macrophages and non-activated T cells. Results suggested that macrophages switched from an M1-like to an M2-like phenotype in the pro-inflammatory model, exhibiting increased IL-10 and CD86 expression.

Interestingly, most previous microfluidic studies have focussed on TNBC, using the invasive MDA-MB-231 cell line. The usefulness of such microfluidic models could be further improved by incorporating different cell lines that represent distinct BC subtypes, and utilising patient-derived cancer cells. Furthermore, there are currently no published reports of microfluidic models investigating the inflammatory obese microenvironment in BC, providing an avenue for further investigation.