4. Discussion
In this work blended hydrogels composed of agarose (variable weight amount from 0.125 to 0.5%) and collagen (fixed weight amount equal to 0.02%) were prepared as enabling matrices for the growth of 3D cellular structures. They combine the biomechanical properties of agarose and the bioadhesivity of collagen. The amount of collagen is 6.25-, 12.5- and 25- times lower than that of agarose, and the formation of the hydrogel is ascribed only to agarose and not to collagen. Indeed, although the self-assembling capability of collagen molecules in vitro under physiological conditions is well known, reconstituted collagen fibrils, that are held together by non-covalent interactions (hydrogen bonding, hydrophobic and electrostatic interactions), are free to slide and do not form a stable 3D network (Tian, Liu, & Li, 2016).
The characterization of the hydrogels evidenced that the agarose percentage governs the morphological, structural and mechanical features of the matrix. Reasonably, higher the agarose amount, stiffer the hydrogel results. On the other hand, the lowest agarose amount corresponds to the fastest degradation and diffusion of molecules through the matrix. The higher degree of degradation of the softest hydrogel is also associated to a higher release of collagen. Noteworthy, the collagen is simply entrapped within the hydrogel mesh and not chemically linked to the agarose backbone. Thus, it can be expected that a less compact hydrogel facilitate the release of collagen, as determined by the protein quantification assay. These findings are in accordance with the compression tests that evidenced a significant reduction of the E modulus as the agarose amount shifted from 0.5 (5.7 ± 0.5 kPa) to 0.25 (1.6 ± 0.4 kPa) up to 0.125 (0.7 ±0.2 kPa). Presumably, a higher amount of agarose allows, during the cooling down phase after synthesis of the hydrogels, the formation of more and tighter hydrogen bonds that reflect the higher structural stability observed visually because of the increase in the compression modulus.
Similarly, the entry and movement of biomolecules through the blended hydrogel looks to be associated to the agarose percentage. The bigger the biomolecule and the stronger the type of non-covalent interactions it can establish with the matrix the slower they are, and that the total amount that may reach the cells embedded into the matrix is only a fraction of the feeding solution. In this sense, the hydrogel may act as a physical barrier to the diffusive transport of specific nutrients and drugs to the tumoroid, similarly to what occurs in vivo(Monteiro, Gaspar, Ferreira, & F., 2020).
The formation and the growth of tumoroids from three breast cancer cell lines, namely MCF-7, MDA-MB-361 and MDA-MB-231 cells, were investigated upon variation of the agarose concentration of the matrix. While MCF-7 and MDA-MB-361 formed 3D structures, whose size and compactness is strictly related to the stiffness and thus to the percentage of agarose of the surrounding matrix, MDA-MB-231 cells, a triple negative breast cell line, did not, in accordance with previous findings (Iglesias et al., 2013; Tasdemir, Bossart, Li, & Zhu, 2018).
The inability of this cell line to form spontaneously tumor spheroids is probably consistent with their lack of adherens junctions.
In detail, we observed that after lading individual cells into each of the three blended hydrogels, both cell lines formed spheroids. By analysing the size and morphology of the structures, we realized that breast cancer cells prefer hydrogels with a stiffness from 1.5 to 0.7 kPa, while stiffer matrices, as those with 0.5% agarose,did not result suitable to support the growth of the spheroids. We reasoned that a different tissue-specific tropism of the cellular models probably contribute to this result. In this case, MCF-7 cells have a low metastatic potential, and are not tissue-specific, while MDA-MB-361 cells were derived from a brain metastasis. The growth of MDA-MB-361 cells on a softer hydrogel matches more closely their in vivometastatic microenvironment. This was in part confirmed when we examined the growth of a neural cell model. In A-C 0.125%-0.02% hydrogel, the neuroblastoma SH-SY5Y cells generated spheroids in a time-dependent manner and their size was bigger than when grown in the stiffer 0.25%-0.02% hydrogels (Figure 11S). As a general remark, SH-SY5Y spheroids reached larger diameters (121.7 ± 14.2 µm) than those obtained with MCF-7 and MDA-MB-361 grown in the same matrix, and this evidence could be related to the different origin of the cell line.
The Live/Dead assay and the mitochondrial staining showed that the spheroids are viable up to 14 days in both types of hydrogels. Notably, we monitored the 3D cultures up to one-month: the spheroids continued to grow in the softer hydrogel, reaching an average size of around 100 µm. On the other hand, in the 0.25%-0.02% hydrogels the spheroids stopped their growth after 2 weeks and started to senesce resulting definitely dead after 4 weeks. This different behavior might be related to the interactions between the cells and the surrounding environment as the 3D structure grows over time: the limited degradation and capability of the stiffer hydrogels to accommodate the spheroids induce their slow aging and death.
These results confirm that soft agarose-collagen hydrogels allow 3D long-term cell growth although cells derived from different tissues sense the change in stiffness of the substrate and significantly modify their behaviour. It has been shown, in this respect, that cells respond to ECM environment by regulating a plethora of transcription factors and other signals that affect cytoskeleton, cellular uptake, cell cycle that in turn determine their morphology, proliferation, differentiation, tumor invasion and metastasis (Kruger et al., 2019; Ranamukhaarachchi et al., 2019; Trappmann et al., 2012; Wang, Gong, Zhang, & Fu, 2017). Among them, it is worth mentioning the transmembrane glycoprotein E-cadherin that is expressed in epithelial cells and connect them through lateral adherent junctions. It has been demonstrated that the level of expression of cadherin represents a crucial feature in cancer progression as it is involved in the epithelial-mesenchymal transition (Wong, Fang, Chuah, Leong, & Ngai, 2018). Loss of E-cadherin expression is generally associated to the lack of intercellular contacts and to increased tumor cell invasiveness via activation of the signalling pathways that regulate metastatic progression (Vergara et al., 2016). E-cadherin expression is also associated to the formation of multicellular tumor spheroids, as already demonstrated (Cui et al., 2017). Indeed, the spheroids derived from MCF-7 and MDA-MB-361 displayed the expression of E-cadherin, as shown in Figure 7.
As a general consideration, softer hydrogels allow for the establishment of long-term culture of large and irregular breast tumor spheroids, while stiffer environment condition the growth of small and compact 3D cell strictures for shorter periods (Figure 12S).
Finally, we showed the possibility to recover the tumoroids from the hydrogel after treatment with Agarase enzyme. The yield of the recovery process is quantitative and highly reproducible in the softest gel that undergoes spontaneously to degradation, while in the matrix with 0.25% agarose only a small percentage of 3D structures can be successfully recovered for further biological and biomolecular studies.