3D multicellular models to study the regulation and roles of acid – base transporters in breast cancer

As a result of elevated metabolic rates and net acid extrusion in the rapidly proliferating cancer cells, solid tumours are characterized by a highly acidic microenvironment, while cancer cell intracellular pH is normal or even alkaline. Two-dimensional (2D) cell monocul-tures, which have been used extensively in breast cancer research for decades, cannot precisely recapitulate the rich environment and complex processes occurring in tumours in vivo . The use of such models can consequently be misleading or non-predictive for clinical applications. Models mimicking the tumour microenvironment are particularly pivotal for studying tumour pH homeostasis, which is profoundly affected by the diffu-sion-limited conditions in the tumour. To advance the understanding of the mechanisms and consequences of dysregulated acid – base homeostasis in breast cancer, clinically relevant models that incorporate the unique microenvironment of these tumours are required. The development of three-dimensional (3D) cell cultures has provided new tools for basic research and pre-clinical approaches, allowing the culture of breast cancer cells under conditions that closely resemble tumour growth in a living organism. Here we provide an overview of the main 3D techniques relevant for breast cancer cell culture. We discuss the advantages and limitations of the classical 3D models as well as recent advances in 3D culture techniques, focusing on how these culture methods have been used to study acid – base transport in breast cancer. Finally, we outline future directions of 3D culture technology and their relevance for studies of acid – base transport.


Introduction
Breast cancer is the most frequent type of carcinoma, and the second leading cause of cancer-related death in the global female population. According to the International Agency for Research on Cancer (IARC), in 2018 alone over 2 million new breast cancer cases were diagnosed, and nearly 630 000 women died of the disease [1]. Despite the development of new early detection methods and improvements in the clinical management of the disease (e.g. targeted therapies), breast cancer remains a major public health problem [2].
The tumour microenvironment is a critical participant in breast cancer progression and therapeutic responses. However, the focus has been mainly on the impact of the stroma, i.e. the non-cancer cells in the tumour, whereas the role of the physico-chemical microenvironment in disease progression remains relatively poorly understood. Due to elevated metabolic rates, and a shift toward anaerobic glycolysis, highly proliferative tumour cells produce and extrude copious amounts of acid. Such tumours are therefore characterized by profound extracellular acidification, as low as pH 6-6.8, while intracellular pH ( pH i ) remains normal or alkaline [3,4]. We and others have demonstrated that net acid extruders such as the Na + /H + -exchanger NHE1 (SLC9A1) and the Na + ,HCO 3 cotransporter NBCn1 (SLC4A7) are frequently overexpressed or post-translationally activated in breast cancer cells and play key roles in their pH i regulation [5][6][7], three-dimensional (3D) growth [8][9][10] and cell cycle progression [11].

Version of Record published: 5 December 2019
While acidosis is a recognized hallmark of cancer with a huge clinical potential, it remains understudied in breast cancer, and ongoing research aiming at developing new treatments generally does not consider the acidic microenvironment. Clearly, to better understand the role of acid-base transport in breast cancer, we need to employ clinically relevant cellular models that include acidic extracellular pH ( pH e ) as a microenvironmental factor participating in cancer progression. It is well recognized that compared with traditional two-dimensional (2D) cell cultures, 3D models better mimic in vivo cell-cell and cell-extracellular matrix (ECM) interactions [12], thereby enabling more accurate recapitulation of in vivo gene expression [13,14] and cell polarization [15,16]. However, it is less widely recognized that such models are also essential because of the key role of dysregulated acid-base homeostasis in cancer progression.
This review focuses on classical 3D models as well as recent advances in 3D culture techniques, and their potential use in breast cancer research. We provide an overview of the commonly used 3D experimental systems and of how these culture models have been used to study acid-base transport in cancer, and we discuss future directions of 3D cell culture technology.

BOX 1: Definitions
• Co-culture: A cell culture containing two or more different types of cells.
• Spheroids: Cell lines grown at high density (generally ∼0.5-2 × 10 4 cells/ml) in suspension, forming tight aggregates often in the shape of a sphere. Multicellular spheroids can be composed of a single cell type or a co-culture of multiple cell types. • Cysts: Cell lines grown on top of or embedded into a layer of the reconstituted basement membrane, causing them to spontaneously form a polarized sphere of cells with a hollow lumen reminiscent of a duct or other tubular structures. • Organoids: Patient-or animal-derived stem cells grown in a dome of the matrix. Each organoid is derived from one or a few cells, which can self-organize and differentiate into multiple organ-specific cell types exhibiting the spatial organization and functions similar to the organ of origin. • Organotypic: Methods involving culturing of cellsusually several different cell types -in a manner that recapitulates the basic architecture and function of the tissue of origin. • Microfluidic system: Cells, including co-and 3D cultures, grown in a microfluidic device that allows precise control of biochemical and physical parameters.
For further information and references, please see the respective sections in the text.

3D culture models
Numerous methods for 3D cell culture models of cancer exist (Figures 1 and 2). Below, we describe the most important such methods and outline their use and advantages in the context of studying tumour acid-base homeostasis and its consequences for tumour development. An overview of the main 3D cell culture methods, their advantages and disadvantages is provided in Table 1.

Choices of ECM materials and scaffolds for 3D cultures
As the reciprocity between tumours and their microenvironment is instrumental in understanding the development of cancer [17], the ability to change the composition of the ECM is important when studying the interactions between tumour cells and their surrounding environment [18]. Purified ECM proteins such as collagens and laminin, as well as extracts from Engelbreth-Holm-Swarm mouse sarcoma cells [19] such as Matrigel, are inherently cytocompatible and have long been used in 3D culture [20]. Importantly, Matrigel and mammary gland have a very similar stiffness and tensional homeostasis [21]. However, large variation between batches and poorly defined composition [22] with limited ability to modulate chemical and physical properties have led to the development of a wide range of synthetic polymers (e.g. polyethylene glycol (PEG), poly(vinyl alcohol (PVA)) as well as natural biomaterials such as chitosan and alginate for 3D matrices [23]). The main advantage of matrices derived from engineered biomaterials is the ability to finely tune their properties, including matrix stiffness, pore size, degradation and the inclusion of ECM proteins, all factors demonstrated to be vital for tumour progression, invasion and metastasis [24][25][26]; see also [27] and [28] for recent reviews on biomaterials.
Another relatively recent development is the ability to produce intact decellularized matrices from in vivo tumours and use them as scaffolds for preparing new in vitro tumours [29] or for analyzing the complexity of the ECM [30]. Such studies will enhance the detailed understanding of the composition and importance of the ECM in different cancers, and guide future efforts in designing in vitro 3D models.

Spheroids
The multicellular spheroid model ( Figure 2A) is several decades old [31] (Figure 3), yet is still used extensively in cancer research. This likely reflects the many advantages this model offers in terms of providing a microenvironment mimicking that of in vivo tumours, yet still being relatively affordable and high-throughput [32]. Cells grown as spheroids not only exhibit cell-cell and cell-ECM interactions, influencing intra-and intercellular signalling, they furthermore develop chemical gradients of nutrients, oxygen, pH and waste products when spheroid diameter reaches 200-500 mm [33][34][35]. As a result, their gene expression profile differs from that of 2D cultures, reflecting clinical expression profiles more closely [32,34,36], as well as affecting growth rates and drug responses [32,36,37]. Cancer cell spheroids can be prepared using several different techniques. A simple method involves the seeding of cells into round-bottom plates with a non-adhesive coating, forcing cell-cell adhesion ( Figure 1A). This spontaneous spheroid formation is often mediated by E-and N-cadherins, however  Table 1 and text for further details.
some cell types need an ECM supplement for compact spheroid formation, often facilitated by collagen I and integrin interactions [38]. Similarly, the hanging drop method ( Figure 1B), exploits the combination of gravity and lack of attachment surfaces in a drop of medium to force spheroid formation. Finally, the spinner flask technique ( Figure 1C) uses constant stirring of a cell suspension to induce cell-cell collisions creating multiple spheroids in suspension. The observed change in anti-cancer therapy treatment response when moving from 2D to 3D cultures, has resulted in a rather large literature on spheroid drug screening assays in 3D spheroids [37,[39][40][41]. In general, treatments are expected to lose efficacy when applied in a 3D tumour microenvironment compared with 2D, at least in part reflecting a combination of reduced drug uptake and altered growth/survival signalling in the 3D setting [32,37]. The spheroid model is applicable to co-cultures with e.g. fibroblasts, adipocytes or immune cells further increasing the relevance of this model for drug screening and anti-cancer resistance evaluation [42][43][44]. Other abilities facilitating cancer progression, such as invasion, can likewise be examined using the spheroid model. Finally, by embedding spheroids into a gel mimicking the ECM ( Figure 1D), and monitoring invasion into this gel, effects of pharmaceutical or biological manipulation on metastatic potential can be evaluated [45][46][47].
In the study of acid-base homeostasis, multicellular spheroids are a very relevant model because, as discussed above, the pH environment created by the cells in the tumour is short-circuited when cells are grown in a buffered 2D monolayer. The localization, expression and activity of acid-base transporters is highly affected by the 3D structure and surrounding tumour microenvironment. For example, Andersen et al. [8] found the spatial   Figure 2B), and described a cell-type specific dependence of these transporters on 3D growth. Studies of pH i and pH e regulation in cancer cell spheroids by Swietach et al. [48] demonstrated how the carbonic anhydrase IX (CAIX) contributes to acidifying the extracellular space of spheroids, while alkalizing the intracellular space, pointing to CAIX as a potential therapeutic target ( Figure 2C). In a more recent study, the expression and inhibition of NHE1, CAIX and the vacuolar H + -ATPase (V-ATPase) were evaluated in hypoxic breast cancer spheroids, revealing a correlation between hypoxia and CAIX or NHE1 expression depending on the cell type, as well as a reduced invasion in 3D in response to CAIX inhibition [49].

Cysts
Developed as a model four decades ago [50], epithelial cell lines grown in 3D culture form growth-arrested cyst-like acini with a spherical epithelial monolayer featuring an internal apical lumen and apicobasal polarity. The cyst model has consequently been widely used for studying polarity and cell-cell and cell-ECM interactions, and the dysregulation of these processes in cancer cells [51].
To form cysts, individual cells are seeded either directly into a gel (embedded cysts) or on top of a polymerized gel (cysts on top) [52,53] (Figure 1E). Once cultured, individual acinar cells have been found to rotate in a manner similar to embryonal development. After the first mitosis cells continue to rotate and divide in a  synchronized manner as the multicellular architecture is established [54]. Blocking E-Cadherin, PAR3 or actin dynamics disrupt this synchronised rotation and the formation of polarized acini [54]. Cysts are highly amenable to confocal imaging and can also be extracted from the gel for molecular and protein analysis [52]. The morphology and expression profiles of breast cancer cell lines in 3D culture has been shown to correlate well with their invasiveness [55].
In cyst models, morphogenesis of mammary acini can be studied in detail [20,56], illuminating factors crucial to their formation [57] and to the development of malignant phenotypes [58][59][60]. Cysts can be co-cultured with tumour-derived epithelial cells as shown by Spink et al. [61], where MCF-7 cells together with MCF10A cells appear to form a luminal and myoepithelial layer respectively. In a similar experiment, Ivers et al. [62] showed that MDA-MB-231 cells engulfed and destroyed MDCK cysts when present in sufficient numbers. Gudjonsson et al. [63] co-cultured human primary luminal and myoepithelial cells to form doublelayered acini and demonstrated that normal polarity could be formed by luminal cells alone with the addition of laminin-1 to collagen gels.
The cyst model has not yet been widely used for studies of acid-base regulation. In a recent study however, NHE1 overexpression in MDCK cysts was found to cause polarity disruption and epithelial disorganization with lack of lumen formation [64] (Figure 2D). While no such studies are, to our knowledge, published so far, cysts should be amenable to live imaging of pH i and transporter activity using small-molecule or genetically encoded pH sensors, and this would be an interesting future development of this model in the context of breast cancer.

Organoids
Organoids are simplified and miniaturized organ models that mimic endogenous cell structure and interactions (Figs. 1F and 2E). According to Lancaster and Knoblich [65] an organoid exhibits the following properties: (i) is composed of multiple organ-specific cell types, (ii) recapitulates specific function of the particular organ (e.g. endocrine secretion, neural activity, contraction), (iii) its cells are spatially organized in three dimensions, similar to an organ. Organoids are derived from one or a few stem cells (e.g. embryonic stem cells, induced pluripotent stem cells, adult stem cells extracted from the target organ), which have the unique abilities of selfrenewal and differentiation [65]. The use of organoids started as a major technological breakthrough in the 1960's [66,67]. This model was seemingly forgotten for decades (Figure 3), but its popularity has significantly increased since early 2010s paralleling the widespread use of stem cell cultures and reflecting its great potential as a tool in biological research and potential clinical use [68]. Illustrating this, the organoid culture technique was named by The Scientist as one of the biggest scientific advancements of 2013 [69] and by Nature Methods as the Method of the Year 2017 [70]. To date, scientists have produced organoids derived from the brain, intestine, liver, pancreas ( Figure 2D), kidney and stomach, and many others are on the way [65].
A major advantage is that organoids can be grown from a limited amount of material, e.g. biopsies and used to develop personalized therapies. Sachs et al. have generated a biobank of 95 organoids representing major breast cancer subtypes derived from patients. These organoids have maintained the mutational signature as well as the genetic and histological profiles of their tissue of origin (e.g. over 90% of ER/PR-negative tumours gave rise to organoids with the same hormone receptor status) and were used for in vitro drug screening that was consistent with in vivo xeno-transplantations and patient response [71]. Walsh et al. have tested the potential value of patient-derived breast cancer organoids for drug screening. They found that the in vitro organoid drug response(s) to a panel of breast cancer therapeutics mirrored the result of animal studies in vivo [72]. Furthermore, exemplified by Duarte et al. [73] who studied cancer-drug resistance employing a BRCA-deficient mouse mammary tumour organoid model, organoids are also amenable to gene editing and can be used for discovery or validation of possible drug targets.
While there are, to our knowledge, no such studies published yet, we consider organoid cultures highly relevant tools for acid-base transport studies in breast cancer. Thus, contrary to conventional in vitro models, organoids preserve in vivo tumour microenvironment properties such as the environmental heterogeneity, cell-cell and cell-ECM signalling. Organoid models have the potential to advance the understanding of acid-base transport in breast cancer not only because they faithfully recapitulate and preserve cell-ECM interactions found in vivo. Organoid culture techniques allow precise control and real-time analysis of microenvironmental factors relevant to acid-base transport and tumour progressione.g. pH (using pH-sensitive fluorescent dyes for tracking cytosolic and luminal pH over time [74]), O 2 concentration (with optical sensor probes [75]) and metabolism (e.g using Seahorse XF technology [76]). Importantly, organoid cultures can be established with a very limited amount of material (e.g. from biopsies) and employing spinning bioreactors will allow for mass production of patient-derived organoids, opening for the prospect of making the analysis of dysregulated pH homeostasis a personalized medicine approach.

Microfluidic 3D systems
Microfluidic systems are increasingly being used to study cancer cell 3D cultures (Figure 3; for reviews, see [77][78][79]). Combining 3D cell culture with microfluidic systems has several advantages that are highly relevant to the study of acid-base transport. First of all, microfluidic systems with chambers of confined size for growing of spheroids can be used to make and analyze very homogenous spheroids [80,81] (Figure 1G) While mimicking the complex, superimposed gradients of the tumour microenvironment is currently beyond the limits also of this technique, microfluidic systems can nevertheless be used to precisely control gradients of, e.g. pH, lactate, and oxygen pressure and to combine multiple cell types and environments, creating an organ-on-a-chip ( Figure 1H), and monitor such gradients using e.g. optical techniques or thin-film microelectrodes [82]. The technique can also be used to co-culture cancer cells with other cell types in a 3D setting. In an elegant example of the latter, Ayuso et al. [83] used a microfluidics device to culture ductal carcinoma in situ cells inside a model mammary duct, followed by the analysis of metabolites and gene expression.
While not yet a standard method in the cancer field, the ability of using microfluidics to very precisely control microenvironmental parameters relevant to acid-base transport should be much more widely exploited. Among several other measured parameters, Weltin et al. [82] employed pH sensor electrodes based on iridium oxide and lactate biosensor electrodes based on lactate oxidase, to monitor glycolytic status in cultured human glioblastoma multiforme cells. The microfluidics technique can be combined with co-and 3D culture. Spheroids can be retrieved and analyzed off-chip, by qPCR, sequencing, and other relevant assays [81] or microfluidics can be used for selection of single cells with specific properties, followed by on-chip clonal expansion and analysis [84].

Other organotypic models
Several other organotypic models have been employed for studies of breast cancer. These include the use of decellularized matrices as a scaffold for the culture of breast cancer cells, e.g. in studies of the interaction between the specific ECM of the metastatic niche and breast cancer cell colonization [85], or in establishing models of the primary breast tumour microenvironment. Another important 3D culture technique is the in vitro maintenance of organ tissue explants ( Figure 1I, [83,84]), which allows, for instance, testing of drug candidates in a preparation in which native cell-cell interactions are maintained. To our knowledge, none of these models have so far been employed for the study of acid-base homeostasis in breast cancer, except for very brief experiments not involving in vitro growth per se [5]. To the extent that such models faithfully recapitulate the in vitro tumour and are amenable to optical analyzes and/or to mRNA and protein extraction, all of such models are likely to be of great interest in the field of pH in cancer.

Conclusions
Cell behaviour is profoundly influenced by cell-cell and cell-matrix interactions and the physico-chemical environment, making use of 3D cell culture imperative in studies of cancer. This is particularly true for studies of tumour-specific dysregulation of acid-base homeostasis, now recognized to play a key role in cancer development. Hence, increasing the use of 3D models is essential for driving further advances in understanding how this key component of the tumour microenvironment contributes to cancer progression.

Perspectives
Despite substantial advances in treatment options, breast cancer remains the leading cause of cancer-related mortality in women globally, and new therapies are urgently needed, especially targeting treatment-resistant subtypes. Breast cancers, like most other solid tumours, are characterized by increased metabolic acid production and up-regulation of acid extrusion from the cancer cells. In conjunction with poor diffusion, this creates a heterogeneous microenvironment with marked spatiotemporal gradients of extracellular acidity, which contributes to treatment resistance and favours development of a highly aggressive cancer cell phenotype.
• Most of the current understanding of the importance of the acidic microenvironment in cancer development comes from conventional 2D cell culture models, which are unable to mimic the complexity of the tumour microenvironment in the patient.
• To drive progress in this field and take the understanding of dysregulated acid-base homeostasis in tumours to a level where it can be exploited therapeutically, it is essential to more accurately reflect these conditions.
• Recent rapid advances in microfluidics techniques, 3D printing, biomaterials, and stem cell techniques have increased the repertoire and sophistication of in vitro 3D culture models mimicking the heterogeneous, acidic 3D environment of patient tumours, enabling new insights and successful development and assessment of new therapies. A more widespread use of models such as breast cancer spheroids in high-throughput testing of novel therapeutics, and patient biopsy-derived organoids as part of a personalized medicine scheme, would facilitate dissection of how tumour acidity contributes to cancer development and how this can be exploited in cancer therapeutics.

Author Contribution
The concept for this review was developed by all the authors. D.C. wrote the introduction, the section on organoids, and prepared Table 1