In vitro Approaches to Model Breast Tumor Complexity

2.1 Two-dimensional and three-dimensional culture

The traditional cell culture methods for studying breast cancer employ two-dimensional monolayer cultures, where cells grow flat on a surface. Two-dimensional culture is still widely used, but with advances in microfabrication now surfaces can be modified with nanostructure topographies and different levels of stiffness to mimic to some extent the physical properties of the matrix surface. These topographies (e.g., roughness, surface geometry) have the capability of providing biomimetic surfaces that have been shown to modify the morphology, proliferation and signaling, among others, of cells [26]. Similarly, changes in the mechanical properties of the ECM (e.g., stiffness) are related to increasing malignant phenotype [27], cancer progression, signaling [28, 29, 30] and drug sensitivity [31]. Despite these technological advances in 2D cultures, multiple studies have shown that cell cultures in 2D felt short to mimic cell phenotypes associated with disease progress such as cell invasion, cell function and expression of pathological markers [4, 23, 32]. In some cases, utilizing 2D culture systems has resulted in the loss of essential cell signaling pathways, hence limiting the ability to fully evaluate cell–cell and cell-ECM interactions [33]. Evidence has also shown that there are inconsistencies when comparing cell morphology, receptor expression, and polarity between cells grown in 2D and the in vivosetting [34].

In order to bridge this gap in biological complexity, multiple methods employing 3D cell culture systems have emerged and continue to be steadily improving, aiming to produce the most in vivo-like structures. Essentially, 3D models can be divided into two groups: cell aggregates (spheroids) and biomaterial constructs [35]. The most basic 3D culture models use scaffolds of synthetic (e.g., polydimethylsiloxane-PDMS, polylactic acid-PLA) and natural (e.g., collagen, Matrigel®, hydrogels) biomaterials to investigate the effect of ECM properties on cancer behavior. Spheroids have been used mostly for drug screening applications since it has been demonstrated they more closely resemble the in vivoenvironment [36]. Growing BC cells in 3D has also revealed a more realistic drug response [21, 37], cell proliferation and morphology [38], and better representation of tumor heterotypic phenotype and TME [39, 40]. For example, single-cell RNA sequencing of breast cancer spheroids have uncovered cell clusters with specific functions (e.g., proliferation, invasion) that provide evidence of the heterotypic nature and complexity of breast tumors [41]. Figure 1 below depicts the main in vitro2D and 3D culture modalities along with the most predominant co-culture models (discussed in the next subsection) to study cell crosstalk.

2.2 Co-culture

Cancer is a heterogeneous disease and even though there have been various advances in cell culture modalities, thorough comprehension of the crosstalk between cancer and non-cancer cells is still not fully understood [42]. Co-culture and multi-culture models have been long established as appropriate tools for evaluating breast cancer heterotypic interactions in vitro[6]. Co-culture refers to the culturing of two different cell lines, while multi-culture models involve three or more different cells. Historically, co-culture models have been the predominant approach in research. However, despite their ability to identify factors mediating cancer and stromal interactions, co-culture models are deficient in incorporating microenvironment structure, dimensionality, and functional response [42]. With the hopes of bridging the gap between in vitroand in vivostudies, new research has been moving away from the study of only two cell types, to studying multi-cell type systems. This type of model permits researchers to control and evaluate the influence of each cell culture component. It also allows the study of important cell–cell heterotypic signals, which would be impossible to study with a 2-cell type model [43].

There have been an increasing number of studies looking to compare tri-culture models with the more traditional mono-culture or co-culture methods. With the intention of better understanding the bone microenvironment, Pagani et al.compared a tri-culture model of osteoblasts, osteoclasts, and endothelial cells; to single and co-cultures. The results demonstrated that the behavior of the three cell types cultured together was very different from the single or the co-culture model, in terms of proliferation, activity, and viability. These results correlate with previously established data regarding their behavior in vivo[44]. Regier et al.evaluated how increased model complexity would affect gene expression. The results demonstrated that gene expression changes based on the type of model utilized; suggesting how tumor and stromal cells would respond to microenvironments of increased complexity in vivo[42]. Loy et al. investigated the effect a tri-culture model would have on angiogenesis and compared it to simpler models. The results showed that the tri-culture model promoted cell-matrix remodeling and early expression of elastic fiber-related proteins. It also reiterated the significance of multi-culture methods since culturing with fibroblasts, endothelial cells, and smooth muscle cells was required to obtain tissues with appropriate physiological-like properties [45]. All three of these studies highlight the increasing need and importance of more complex heterotypic cultures.

Co-culture models involve a cell growing arrangement, where two or more different cells are cultured with some amount of contact between them [46]. The communication between the cells may be bi-directional or multi-dimensional, and it can happen at the macro-scale or at the micro-scale [47]. The method of choice should be dependent on what is the focus of each individual study and can be grouped in: compartmentalized, conditioned media and mixed culture.

3.1 Microfluidics

Microfluidic platforms can be utilized to scale down the traditional culture modalities, yet they enable to customize the culture environments to examine more complex interactions [64]. This technology employs microsystems that allow the manipulation of small fluid volumes and control over the spatial location of cell clusters [70]. Its application to improve 3D cell culture models has been increasing since 2012, particularly in BC research [71]. In comparison to macroscopic culture, microfluidic cell culture models have several significant advantages that, when employed, lead towards better biomimetic models. Firstly, cells may be cultured in a spatially controlled environment by controlling fluid patterns and proximity across culture compartments [72, 73, 74, 75, 76, 77]. This technology permits the combination of multiple cell types and to control cell patterning, to recapitulate to some extent tissue observations. For example, microfluidic devices permit the study of angiogenesis while also allowing the study of endothelial migration and evaluation of cell response in co-culture [71, 78]. Also, microfluidics can implement continuous perfusion conditions, and controlled gradients, which are both characteristics that also resemble the cancerous in vivoenvironment more closely. Gradients are found in angiogenesis, invasion, and migration whereas perfusion is crucial in vasculature and cell extravasation as well for nutrient replenishment. Finally, microfluidic systems enable high-throughput arrays and pose lower contamination risk and reagent consumption which make them very appealing for studies with limited cell samples such as those that employ patient-derived tissues [70, 71].

Recent studies in microfluidic systems have highlighted their capability to recreate and profile some of the biological complexity of the tumor microenvironment. Such studies have revealed important information regarding the processes involved in metastasis and how the tumor microenvironment contributes. For example, single cell RNA sequencing using microfluidic devices have revealed the diversity of the breast epithelium, which sheds light about early tumorigenesis and tumor progression [79, 80]. In addition, microfluidic devices pose as an advantage to personalized medicine by aiding in the selection of appropriate pharmacologic agents. In this regard, Lanz et al.developed a 3D microfluidic device, OrganoPlate®, to be utilized for therapy selection. They showed that MDA-MB-231 (cell line isolated at MD Anderson from a pleural effusion of a 51-year old Caucasian woman) cells embedded in Matrigel® became more sensitive to the drug, thus confirming along with previous studies that drug response is tuned by the ECM. The results were promising and even though further validation is warranted, it appears to be a fine tool for pharmacologic selection and response prediction [37]. Similarly, Yildiz-Ozturk et al.studied the cytotoxicity of carnosic acid and doxorubicin on MCF-7 and MDA-MB-231 BC cell lines and demonstrated the importance of biomimicry in in vitroplatforms [20]. A breast metastatic microfluidic model was developed by Kong et al.to mimic the metastasis of circulating breast cancer cells (CBCCs) to the lung and other organs. Their microfluidic device allowed the flow of CBCCs over primary cell culture chambers, which would have been impossible with static conditions. They demonstrated that the metastatic potential of these cell lines was in concordance with animal models, providing a cost-effective and time-saving alternative [81]. Bersini et al.also developed a microfluidic co-culture model made up of metastatic BC cells, and collagen gel-embedded bone marrow-derived stem cells (hBM-MSC) lined with endothelial cells to create an osteo-conditioned microenvironment and access extravasation and micrometastases to bone tissue [59]. They found that BC receptors CXCR2 and bone-secreted chemokine CXCL5 play major roles in the extravasation process. However, due to the complexity of the design, their platform is not high throughput compatible, which adds many challenges, particularly to obtain multiple replicates in a short time. Also, in general it is important to notice that most of the organ on chip microfluidic platforms focus on the metastatic stage of the disease, leaving an evident need for research focusing on the early stages of breast cancer. Yet, some efforts are being done to overcome this gap. As an example, Choi et al.developed a compartmentalized microfluidic device that enabled co-culture of tumor spheroids and normal mammary epithelial cells in close proximity to fibroblasts, with the goal of providing a model that allows researchers to closely examine the mechanistic progression of early-stage breast ductal carcinoma in situ(DCIS) [82].

Even though microfluidic devices have given the opportunity to better replicate the tumor environment, there are still some caveats to its use. Silicone-based devices have been shown to sequester small hydrophobic molecules, which can compromise the results of some studies [70], yet researchers have been addressing this by modifying the material to make it more hydrophilic and reduce molecule sequestration [60]. Also, microfluidic devices in some cases can induce mechanical stress to the cells [83], which can modulate cell responses in an unpredictable manner, and are often limited by complicated set-ups [70], which limits their broad adoption by the scientific and clinical community. As such, simpler fabrication methods and commercial availability of customizable microscale platforms is desirable to overcome such limitations.

3.2 Bioreactors

Despite the numerous advantages of the aforementioned 3D culture methods, the duration of culture and nutrient availability can be a limitation in static cultures particularly to enable observations that occur in cells over periods of several weeks. In this case, perfusive systems, such as bioreactors, are more appropriate. A bioreactor is a canister that allows the 3D culture of cell clusters for extended periods of time. It is coupled to sensors and actuator components allowing for the controlled delivery of oxygen, nutrients and other parameters [84]. Goliwas et al.developed a perfused 3D BC surrogate model utilizing a bioreactor system that incorporated breast carcinoma epithelial cells and stromal fibroblasts into an extracellular matrix. The study demonstrated that using a bioreactor allowed for analysis of longer growth periods and a greater degree of growth when compared to solid models [85]. Bioreactors have also been utilized to study metastatic progression of breast cancer, and as potential drug development platforms for cancer treatment. Krishnan et al.utilized a compartmentalized bioreactor model, with osteoblasts and metastatic BC cells, to study the colonization of osteoblastic tissue. In their design, cultured osteoblasts were monitored over longer periods and exhibited more in vivo-like characteristics, compared to 2D cell cultures [86]. Marshall et al.developed a physiologically relevant bioreactor system that could be potentially used for pharmacologic development. Their construct was capable of supporting and perfusing larger volume, which poses as an advantage to lab-on-a-chip systems [62]. Other studies have also used bioreactors to assess drug response of BC tissue [63, 87]. Despite bioreactors being an ideal option for cultures that require long-term analysis, there are some factors that might damper their use. Membrane bioreactors may become contaminated and multilayer cell growth may cause transfer limitations [88]. Also, its complex composition and dimensionality limits their implementation in convectional labs and limits the number of experimental replicates [89].

3.3 Three-dimensional (3D) bioprinting

Another technology that has emerged in recent years and that is being applied to 3D culture technology is 3D bioprinting. Its development has been possible thanks to advances in 3D printing technology, biomaterials and tissue engineering methods. Three-dimensional (3D) bioprinting consists of printing cells together with ECM components, biomaterials and bioactive factors [90]. It has been shown that bioprinting techniques can be used to generate 3D tumor models that can better resemble the TME [90, 91]. This has been achieved as bioprinting provides the ability of controlling the spatial arrangement of cells, creating biomolecular gradients and well-organized vessel-like structures (vasculature) within a micron scale resolution [92, 93]. Therefore, bioprinted tumor models are used for angiogenesis, migration and drug development and screening studies as well as TME models [65, 94]. Although 3D bioprinting is widely used in tumor research, very few studies use bioprinted models for BC. Yet, most of these studies are focused on BC metastasis and drug resistance. A study performed by Zhou et al.evaluated the interaction between triple negative breast cancer cells (TNBC) and osteoblasts to assess metastatic progression in bone. They found that osteoblasts increased VEGF secretion and therefore, enhanced the proliferation of BC cells, while osteoblast proliferation was inhibited [58]. Bioprinted BC models have also been used for drug resistance studies. Swaminathan et al.bioprinted pre-formed MDA-MB-231 spheroids along with breast epithelial cells and vascular endothelial cells and evaluated plaxitacel chemoresistance in mono and co-culture. They demonstrated that bioprinted spheroids are more resistant to plaxitacel as it has been shown before in other studies. Yet, this resistance was decreased in co-culture with vascular endothelial cells highlighting the importance of replicating the TME complexities in vitro[95]. Another study by Duan et al.examined drug resistance using 3D bioprinted constructs of BC cells and adipose-derived mesenchymal stem cells (ADMSC). They found increased chemoresistance in BC cells cultured with ADMSC in comparison to monoculture and, thus provided a model to better understand the role of ADMSC in BC progression [66]. Likewise, Campbell et al. bioprinted MCF-7 cancer cells and showed higher resistance to Tamoxifen compared to monolayer culture, providing a more biological-like behavior [66, 96]. Despite the flexibility of 3D bioprinting systems, there are some challenges that need to be overcome to ease its application. Maintaining high viability and original phenotype is an issue in some bioprinting techniques due to exposure of cells to shear stress. Therefore, close control of bioink viscosities, extrusion rates, among other parameters, is imperative [97]. Also, lack of process standardization and guidelines pose another challenge for study comparison and reproducibility.

3.4 Organoids

The most recent 3D cell culture modality are organoids. These are 3D heterotypic in vitrotissue constructs, derived either from primary tissue or stem cells, that have the ability to mimic the in vivoorgan [98, 99]. Historically, established cancer cell lines have been widely utilized as single cell models of the cancer disease. However, their use has several drawbacks in terms of their capability to mimic the pathology of the patient. Cell lines can undergo genetic changes, losing the genetic heterogeneity of the original tumor [100]. Organoids also possess substantial similarities to cancer cell lines 3D models (spheroids) such as cell–cell and cell-matrix interactions, gradients of nutrients, oxygen and metabolites, and can be replaced from frozen supplies with ease. They are also relatively easy to handle and can be grown in infinite quantities [101]. Yet, the main characteristic of organoids is their capability to closely resemble and retain the pathology of the parental tumor over several rounds of expansion in vitro[102, 103]. They also have shown therapeutic predictability for some drugs and can be preserved as biobanks and expanded, which allows extended incubation [98, 99]. Given the number of mutational processes involved in cancer development and progression, being able to study tumorigenesis in depth is crucial. Organoids allow for organ-specific mutations to be analyzed and their whole genomes to be sequenced. Intratumor heterogeneity can also be analyzed by growing organoids from separate sections of the same tumor [100]. Another area where organoids can play a major role is drug development. Organoids appear to be much better models for identifying and testing anticancer drugs yet in a patient specific manner. For instance, studies on single cell transcriptomics of organoids have detected differences in drug sensitivity, proving that organoids maintain tumor heterogeneity, which is considered a critical aspect of tumor models [104].

Studies with BC organoids are limited, since this modality has just started to be explored. However, they have gained more popularity in the last few years. Cheung et al.used breast carcinoma organoids to understand tumor invasiveness and metastasis. They found that the heterotypic interactions between epithelial subgroups are key to collective invasion [105]. Broutier et al.was able to demonstrate that liver cancer derived organoids could be utilized for drug screening testing and identification of potential pharmacologic targets [68]. Sachs et al.demonstrated the biomimetic nature of organoids by demonstrating the reflecting histopathology of in vivotumors, as well as HER2 and hormone receptor status. Moreover, drug screening tests were consistent with patient response [69]. These promising findings suggest that organoids will be an ideal alternative model for cancer research. Nonetheless, successfully cultivating patient organoids from biopsy specimens is still a challenge mainly due to low cell recovery and heterogeneity of collected samples, and limited availability of standardized methods [103, 105].