In Vitro Rigidity Effect on Breast Cancer MCF-7 Three-Dimensional Microtissue Grown from Few Cells

Background: Biological tissues normally possess varying rigidities, changes in which may reflect transformation from normal to pathological state. Cancer cells within the tumor are influenced by the mechanical conditions of their microenvironment, which can drive cell fate. Methods: We present a new on-array methodology to mimic and control desired surrounding rigidity in vitro for three-dimensional (3D) breast cancer object formation and growth. 3D objects were generated from single cells within a hydrogel array, cultured under various mechanical conditions and measured at single-object resolution. Results: Alterations in development of 3D breast cancer microtissue under various rigidity conditions in vitro are revealed. Object growth rate, morphology and vital features are associated with the extent of environmental rigidity, the point in time at which embedding was performed and the initial number of seeded cells. Under stiffness that resembles tumor tissue, higher growth rate of breast cancer microtissue and specific “preinvasive” phenotype are evident. Hallmarks of this phenotype include loose morphology with unclear edges and massive peripheral cell spread, dispersed intrinsic structure and reduced expression of epithelial surface markers. Conclusions: Physical changes in cell environment without parallel changes in biochemical conditions affect growth and development of 3D cancer microtissue grown from individual cells in vitro. A stiffer environment supports 3D microtissue growth and its morphologic and functional diversity in comparison to regular low rigidity conditions.

tissue rigidity of primary breast tumors correlates with metastatic recurrence and poor patient survival [9].
Solid tumors develop in vivo, interact with and modify the local conditions, including microenvironmental stiffness, in favor of tumor progression [10,11]. Cancer cells within the tumor are intimately influenced by mechanical conditions, biochemical signals and cell-cell interactions which can change its genetic profile [12] and tumor tissue morphology [13,14], as well as driving its fate [15,16]. Culturing under altered stiffness conditions without simultaneous biochemical impact, indicates that tension can drive tumor progression through destabilization of adherent junctions that can act as cellular mechanosensors, whereas disseminated tumor cells may respond to mechanical cues differently until they re-epithelialize [9].
The last twenty years have seen in vitro research shift towards three-dimensional (3D) cell culture models which are recognized as superior to the commonly used 2D models, as they better mimic the natural structure in vivo, thereby providing a more accurate in vivo-like Organization and response to external stimuli. Today, it is believed that different tissue culture conditions, 3D as opposed to 2D, affect the physical, chemical/biochemical and biological stimuli and induce changes in cell signaling at multiple levels, such as transduction, transcription and post-transcription, cell phenotypes, growth, invasion and survival under anti-cancer drug treatment [17].
Cancer cells grown as multicellular 3D structures demonstrate the ability to quickly adapt response to extracellular stimuli, resulting in higher resistance to treatment.
Special conditions are required for 3D culturing of epithelial cells in vitro, wherein cell-cell adhesion is stronger than cell adhesion to substrates. Currently, polymer or natural hydrogels, such as agarose are the most widely used non-adherent substrates.
Agarose forms a macroporous matrix which allows rapid diffusion of molecules including macromolecules unrestricted by the gel [18].
Agarose, being nontoxic to living cells, is used for cell investigation, including cell migration study [19], bio mimetics of vasculature [20], creation of synthetic analogs of basement membrane [21], and as a half-liquid medium for cellular 3D structure formation [16,22].
Additionally, its rheological properties make it useful for modulation of matrix stiffness in vitro [23]. Finally, optical properties of agarose gel are excellent for live-cell visualization and observation due to its refractive index being comparable to water [24,25].
Manipulating the mechanical feature of microenvironments in which cancer cells interact with surrounding stiffness in 3D cell cultures remains challenging. Development of the optimal mechanical conditions in vitro can benefit both the scientific study of cancer tumor evaluation and biomedical applications, including anti-cancer drug screening, stem cell study, regenerative medicine, biomedical and tissue engineering.
In this study we propose an efficient approach to reach desired surrounding rigidity in vitro for 3D breast cancer model formation and growth. The excellent mechanical and optical Properties of agarose were used for the measurements of arrayed non-adherent non-tethered 3D objects under different mechanical conditions at single-spheroid resolution. Three-dimensional microtissues were generated from single cells in a hydrogel array, cultured under various mechanical conditions which were created by the process of agarose embedding, and measured at single-object resolution.
The study demonstrates alteration in 3D breast cancer microtissue creation, formation and development in vitro under the influence of various stiffness conditions at individual 3D object level. Significant differences that are associated with the extent of environmental rigidity, the point in time at which embedding was performed and the initial number of seeded cells, were revealed in microtissue growth rate, morphology and vital characteristics. The 3D objects initiated from less than eight cells are significantly different from those initiated by more cells and demonstrate a growth rate independent from surrounding rigidity. Under stiffness conditions that resemble tumor tissue, a higher growth rate of breast cancer microtissues and specific "pre-invasive" phenotype were evident.
In contrast, the subset with the specific "pre-invasive" phenotype has not been observed in the control culture of 3D objects grown freely under low-rigidity conditions.

Methods
The Hydrogel Microchamber Array (hMCA) was designed and fabricated as described previously [26]. Briefly, an array of square bottom pyramid shaped microchambers (MCs) was obtained from GeSiM mbH (Großerkmannsdorf, Germany) and used for production of the PDMS stamp with a negative MC array. Fabrication of the hMCA was performed in the specially modified commercial six-well glass bottom plates. Warm LMA was dripped on the surface of the plate's glass bottom and pre-heated PDMS stamp gently placed over it. The system was incubated at RT for 5-7 min for pre-gelling and pre-cooling, followed by 10 min incubation at 4°C for LMA gelation.
At the culmination of the gelation process, the PDMS stamp was peeled off, leaving agarose gel patterned with square geometry MCs heat-inactivated fetal calf serum, 100 U/mL penicillin, 100 µg/ mL streptomycin, 2% glutamine, 2% sodium pyruvate (complete medium). Cells were maintained in completely humidified atmosphere with 5% CO 2 at 37°C. Before use, the exponentially growing cells were collected by trypsinization, washed and resuspended at appropriate concentrations in fresh complete medium.

Cellular microtissue formation:
Cell suspension (50 µL, 0.1-0.3×10 6 cells/mL) was loaded onto hMCA, and then set aside to allow cellular/multicellular structure formation in each MC. Loaded MCF-7 cells were embedded in agarose gel with different w/v concentrations (1%, 2% or 3%) at 19 h or immediately after cell loading to create suitable mechanical conditions. The 3D spheroids grown either in the hydrogel layer or freely (control), each in its individual MC, were monitored for seven days. conglomerates which resulted from cell embedding. Next, the "sandwich" chip was allowed to cool for a few minutes at 4°C until the agarose gelled. After this, the fresh complete medium at RT was added into the device. In the control samples, the medium was exchanged with fresh medium (agarose 0%) at the appropriate time points and treated at 4°C, respectively. Spheroid formation and growth were prolonged until 7 days under regular conditions with medium being exchanged twice during the experiment. Image acquisition was performed continuously at each experimental step.   For histological examination, the staining by hematoxylin-eosin (H&E) dyes was performed as end-point procedures. Previously, multicellular breast cancer microtissues were fixed in situ by formalin and embedded in paraffin, with agarose "sandwich" intact.

Cellular object staining
Paraffin sections of 5 mm thickness were prepared, deparaffinized and rehydrated. Regular protocol for H&E staining was followed in the conventional manner.

Imaging system and operating software
Images were acquired using a motorized Olympus inverted  Table S1). All filters were obtained from Chroma Technology Corporation (Brattleboro, VT). A 14-bit cooled, highly sensitive ORCA II C4742-98 camera (Hamamatsu, Japan) was used for imaging. A majority of the images were taken with a ×10 magnification objective. The complete microscope system was enclosed in an incubator which provided a temperature of 37°C and humidified atmosphere containing 5% CO 2 (Life Imaging Services, Switzerland), allowing monitoring over long periods.

Mechanical properties of hydrogels
Hydrogels were made by dissolution of agarose powder (w/v) in PBS (Supplementary Data Table S2). The mechanical response of agarose hydrogels (G* complex modulus) borders on the limits [10] for pathological breast tissue. The measured complex modulus values of 1% w/v and 2% w/v agarose hydrogels are close to those measured by others for breast cancer tissue (1.7 kPa and 3.6 kPa, respectively), while those of higher concentrations of agarose rose significantly, approaching the upper limit and did not differ between 3% w/v and 6% w/v (about 10.5 kPa).

Video 1:
Arrayed format of the MCF-7 3D objects grown in the condition of 1.7 kPa introduced directly after cell seeding into hMCA is shown. Wide bright field stitch image from the single macro-well consists 720 three-dimensional objects (A) is accompanied by the enlarged representative area indicated by rectangle in the stitch image for illustration (B). Representative area of the control 3D spheroids grown in the regular condition of low rigidity (about 400 objects per macro well) is presented (C). Magnification ×10. Scale bar 0.5 mm. MCF-7 cells were loaded in the agarose chip and grown as 3D objects within the agarose layer by different stiffness for seven days. (A) Growth curves of MCF-7 microtissue population is presented as increase of microtissue SA (mm2) (ordinate) via time (hours) of growth (abscissa). First point of each graph presents SA of microtissues grown 24 h after cell seeding either freely (smooth lines) or within agarose layer (dashed line). The increase in SA of microtissues within stiffer surrounding (3.6 kPa) is presented as mean ± SD at each time point and accompanied by trendline for each graph. Growth rate was calculated as linear slope (B) of the microtissue SA increase in 19 h embedded microtissues or as exponents (C) in 0 h embedded objects. Asterisks indicate the significant differences calculated as t-test in the groups of microtissues grown from cells embedded within agarose overnight after the seeding; or as ANOVA in the groups of microtissues grown from cells embedded within agarose immediately after seeding. Microtissues were grown and measured as described in Materials and Methods. Each dot is the average growth rate (mean ± SE) calculated in the groups with equivalent initial number of cells. Growth rate (ordinate) vs number of cells seeded (abscissa) is presented. (A) Growth rate was calculated as a linear slope of SA increase for each individual microtissue grown from cells embedded in agarose 19 h after seeding (B) The linear classifier calculated by SVM algorithm distinguishes between control spheroids and microtissues grown under 3.6 kPa rigidity depending on the initial number of cells generating growth. (C) Enlarged part of the graph (A) presents the growth rate of groups of microtissues initiated by one to eight cells.

(D)
Growth rate was calculated as exponent of the SA increase for each individual microtissue grown from cells embedded in agarose immediately after seeding.  objects were observed in control freely grown spheroids (Table 1).
When seeded cells were embedded in agarose immediately after    MCF-7 cells were loaded in hMCA and embedded by agarose immediately after seeding as described in Materials and Methods. Morphology variances of the 3D multicellular microtissues were scored at end point after seven days growth. Growth rate was calculated as exponents for each individual 3D object. P/P* were calculated as ANOVA. P represents comparison within each score group of the various rigidities. P* represents comparison of rigidities between score groups: number to the left of slash -between rigidities of all score groups and number to the right of slash -between rigidities of Score groups 2, 3 and 4.
Additionally, no correlation was revealed between morphology variants and initial number of cells seeded ( Figure 4C). MCF-7 cells are a strong epithelial phenotype with Ep-CAMhigh expression which is needed as a growth-and invasion-promoting factor [28].
CXCR4 protein expression on breast cancer cells has been found to be significantly associated with lymph node metastasis and TNM stage in clinical studies [29]. Seven-day 3D microtissues were    Seven-day microtissues were grown and double stained by FDA and PI in situ. The FDA hydrolysis rate and PI positive area were calculated as described in Materials and Methods. P was calculated as ANOVA in score groups; *indicates the difference in PI positive areas in microtissues grown under the stiffer conditions.
Weak PI staining showed rare cellular death events accompanied Moreover, when FDA hydrolysis ability was analyzed in the various morphological groups, Score groups 3 and 4 3D objects grown in the stiffest surroundings (10.5 kPa) demonstrated better ability to hydrolyze FDA, as opposed to those grown in softer environments (Table 3).

Video 2:
Surprisingly, averaged TMRM FI is independent of both the environmental rigidity (P<0.7) and the point in time at which embedding was performed (P<0.1). However, when the morphology score is taken into account, a trend to lower mean FI in the Score 1 group is observed for each rigidity, while the other score groups remain similar in this parameter.

Discussion
Breast cancer development is characterized by the combined activity of the epithelium, tumor-associated vasculature, and microenvironmental cellular and extracellular components, We present an effective approach to modify the surrounding rigidity in vitro, in order to mimic the mechanical properties within primary tumor tissues and during cell invasion and expansion. Most current protocols preserve constant surroundings during formation and growth of multicellular 3D structures [31,32] in general, and in particular, during formation of 3D breast cancer spheroids [33,34].
As a rule, additives such as methocel, collagen or Matrigel, used for improving 3D spheroid formation, do not notably alter surrounding stiffness. For example, rigidity of collagen used for 3D breast cancer spheroid formation does not exceed the stiffness of normal breast tissue [7,10]. In other words, in vitro multicellular 3D breast cancer structures are developed under mechanical conditions similar to those in normal breast tissue, yet unlike the actual status in vivo.
The stiffer conditions used in this study are associated with the appearance of morphologically diverse 3D microtissue subgroups, which demonstrate different functional abilities with respect to enzymatic and mitochondrial activity. This surprising discovery can be correlated with the histological structure of primary invasive breast tumor tissue which exhibits significant intratumoral heterogeneity by morphological patterns of cancer cells [35,36]. Different morphologic type (alveolar, solid, tubular, trabecular and discrete) structures have been shown to form from transcriptionally distinct subpopulations of tumor cells [37,38]. The capacity of dissociated mice tumor cells to reconstruct their original histological structure in vitro was demonstrated previously (1959) and the exhibited histological pattern bore a basic resemblance to that of primary tumor tissue [39]. Based on this information, the Score 2 (round regular morphology with smooth edges) 3D microtissues and possibly the Score 3 (round morphology with smooth edges and single peripheral cells) 3D microtissues as well, can be interpreted as similar to alveolar structures, while the Score 4 (with "pre-invasive" phenotype) 3D microtissues better mimic solid structures with looser histology.
Perhaps the Score 1 microtissues structurally resemble discrete structures arranged by either a few cells or a single cell. So, a more rigid environment in vitro contributes to 3D microtissue growth and its structural and functional diversity resembling actual structures in the body. Additionally, it is now considered that MCF-7 cells have a low metastatic potential. However, a decrease of surface Ep-CAM expression in the 3D objects grown under stiffer conditions may suggest the switch between the strong epithelial phenotype and the less differentiated cellular phenotype [40], and together with the unaltered expression of surface CXCR4 chemokine, facilitate metastatic potential.  [15,31,41] or from primary mammary cells [14]. As a general rule, the number of cells for 3D spheroid formation is presented at concentration units, assuming a homogenous distribution of cells per device/macro-well. The easy control of spheroid size and growth rate by initial number of seeded cells is described and discussed in our previous study [42]  does a difference exist between invasive capacity of the daughter cells originating from "clone-forming" 3D objects and those originating from the 3D microtissues initiated by more than one cell? -all issues subject to further study.

Acknowledgments
This study was endowed by the Bequest of Moshe-Shimon and Judith Weisbrodt.

Conflicts of Interest
The authors declare that they have no financial conflicts of interest.

Ethical Statement
There have been no human or animal experiments carried out for this article.

a.
Elena Afrimzon initiated and led this study, carried out experiments, image analysis and assessment of the results. She performed the literature survey and prepared the manuscript.

b.
Sergei Moshkov was responsible for design of the HMC array and its production.

c.
Yana Shafran contributed to in-HMC generation and culturing of breast cancer spheroids.

d.
Ronen Pelov led the rheometric measurements of the hydrogels. Maya Freund led the histological examination of breast cancer microtissues.

e.
Yaron Hakuk was responsible for software programs and contributed to image analysis.
f. Zahavit Bar-On-Eizig was responsible for image analysis.

g.
Naomi Zurgil supervised this study and drafted the manuscript.

h.
Mordechai Deutsch supervised this study, designed and coordinated HMC array production and drafted the manuscript.