Three-Dimensional Cell Culture Based on Magnetic Fields to Assemble Low-Grade Ovarian Carcinoma Cell Aggregates Containing Lymphocytes

There is a limited number of established ovarian cancer cell lines matching the low-grade serous histotype available for research purposes. Three-dimensional (3D) culture systems provide in vitro models with better tissue-like characteristics than two-dimensional (2D) systems. The goal in the study was to characterize the growth of a given low-grade serous ovarian carcinoma cell line in a 3D culture system conducted in a magnetic field. Moreover, the culture system was evaluated in respect to the assembly of malignant cell aggregates containing lymphocytes. CAISMOV24 cell line alone or mixed with human peripheral blood mononuclear cells (PBMC) were cultured using a commercially available 3D culture system designed for 24 well plates. Resulting cell aggregates revealed the intrinsic capacity of CAISMOV24 cells to assemble structures morphologically defined as papillary, and reflected molecular characteristics usually found in ovarian carcinomas. The contents of lymphocytes into co-cultured cell aggregates were significantly higher (p < 0.05) when NanoShuttle-conjugated PBMC were employed compared with non-conjugated PBMC. Moreover, lymphocyte subsets NK, T-CD4, T-CD8 and T-regulatory were successfully retrieved from co-cultured cell aggregates at 72h. Thus, the culture system allowed CAISMOV24 cell line to develop papillary-like cell aggregates containing lymphocytes.


Introduction
Epithelial ovarian cancer (EOC) is among the most lethal gynecological malignancies, ranking third as a cause of women's worldwide deaths. Surgery can cure most treated women when the malignancy is still restricted to the ovaries, but 79% of women with EOC are diagnosed at advanced stages, resulting in a poor five years survival rate of 20-25% [1][2][3]. Late diagnosis is a consequence of asymptomatic initial development and an absence of specific biomarkers for the early detection of EOC.
Abdominal bloating or pain are frequently the first symptoms of EOC, which are commonly associated with ascites and metastases beyond the ovaries [4][5][6]. Ninety percent of the ovarian malignancies are of epithelial origin and comprise four main histotypes. EOCs are further categorized based on their growth and molecular characteristics as type I or II. The serous histotype comprises 70% of all EOC, among which the type II or high-grade, accounts for two-thirds of all ovarian cancer deaths [7][8][9].
with balanced salt solution (PBS, Nutricell). Cell viability was determined by trypan blue (1% w/v in PBS) exclusion method and density adjusted to 10 6 cells/mL in RPMI-1640 supplemented medium. CAISMOV24 cells conjugated with NanoShuttles TM were seeded in 24-well ultralow-attachment plate (ULA, Cellstar ® Greiner Bio-one, Kremsmünster, Austria) at 10 5 cells and final volume of 400 µL/well. The 3D culture was achieved by incubating (37 • C and 5% CO 2 ) the plates under magnetic field, first using a bioprint drive for 3h, which was followed by a levitation drive for all culture period. This procedure promotes cells to grow as aggregates. 3D culture plate was replenished with fresh medium every 2 days until the moment of cell aggregate use.

Blood Samples
The blood samples of 7 healthy donors were collected using 9mL vacuum blood-sampling tubes containing sodium heparin (Vacuette ® , Campinas, Brazil). The peripheral blood mononuclear cells (PBMC) were isolated by gradient centrifugation, using Ficoll-Paque Plus (GE Healthcare, Uppsala, Sweden), followed by a washing procedure performed twice (centrifuged 600 g/5 min) using a PBS. Cell numbers were assessed in a Neubauer chamber using acetic acid solution (2% in PBS) and the trypan blue exclusion method to assess viability.

Three-Dimensional Co-Cultures of CAISMOV24 and Lymphocytes
PBMCs were conjugated with NanoShuttles TM by mixing them at a proportion of 20,000 viable cells to 1 µL of the nanoparticle in a conical tube. Subsequently, PBMCs suspension were subjected to three cycles of centrifugation and resuspension (30 g/5 min), by pipetting the cells up and down (50 times), without changing the medium. CAISMOV24 cells were conjugated to NanoShuttle TM as aforementioned. NanoShuttle TM conjugated PBMCs and CAISMOV24 cells were seeded at 1:5 cell ratio in 24-well ultralow-attachment plate at 1.2 × 10 5 cells and final volume of 400 µL/well. 3D culture was achieved by incubating (37 • C and 5% CO 2 ) the plates under magnetic field, first using a bioprint drive for 3h, which was followed by a levitation drive for all culture period. Co-cultured cell aggregates were either treated with IL-15 (60 ng/mL final concentration/well) daily or not. Additionally, culture plates also had wells containing PBMCs conjugated with NanoShuttel TM alone, either treated with IL-15 or not, as experiment controls. Additionally, 3D co-cultures were carried out using PBMCs labelled with carboxyfluorescein-succinimydil-ester (CFSE, Molecular Probes, Invitrogen, Burlington, ON, Canada). For cell labelling, PBMCs were incubated in 3 µM CFSE-PBS solution at a density of 2.5 × 10 5 cells/mL for 15 min at 37 • C.

Histological Analysis and Immunohistochemisry
Cell aggregates of CAISMOV24 were either submitted to cryo-sections (Cryostat CM 1850, Leica Biosystems, Wetzlar, Germany) intended for immunohistochemistry analysis or fixed in 4% formalin and routinely processed to obtain histological sections from paraffin-embedded tissue. Briefly, immunohistochemistry was carried out from a series of consecutive cryosections (4 µm) placed on silanized slides, which were fixed with acetone for 15 min, washed in PBS and incubated in appropriate dilutions of primary antibodies (PAX8 clone BC12, control no. 901-438-070919, dil. 1:100; estrogen receptor clone 1D5, control no. 901-054-081817, dil. 1:200; Progesterone receptor clone 16, control no. 903-424-020818, dil. 1:200, Biocare Medical, Pacheco, CA, USA) and secondary peroxidase conjugated-antibody, following standard procedures of the Anatomopathology Department at the University of Campinas Hospital. Additionally, cell aggregates resulting from the co-culture of CAISMOV24 cells with CFSE-labelled PBMCs were submitted to cryo-sections. Cell aggregate cuts on slides were fixed with acetone for 15 min, washed with PBS, incubated in permeabilization solution (0.0387 M Na 3 C 6 O 7 with 1% Triton X-100) for 2 min, washed again, and finally, stained with 4',6-Diamidine-2'-phenylindole dihydrochloride/5 min (DAPI, Roche Diagnostics GmbH, Mannheim, Germany). Cuts on slide were covered with mounting medium and microscope slip. Microscope examination was carried out in fluorescence microscopy (Eclipse 80i, Nikon ® , Tokyo, Japan; with LP 430 nm and 510 nm filters).

Lymphocyte Phenotyping
Lymphocytes were recovered from the co-cultured cell aggregates and phenotyped for the identification of their subsets. For this procedure, six cell aggregates obtained under the same culture conditions were grouped to be processed together for cell disaggregation by up and down pipetting in staining solution (PBS with 2% SFB and 0.2 mM EDTA). The resulting single cell suspension was washed by centrifugation, and the final cell pellet suspended with staining solution to adjust cells density to 10 6 cells/mL. A flow cytometric-based assay was conducted according to standard procedures. Briefly, the 0.3 × 10 5 cells were mixed with 50 µL of staining solution containing a mix of fluorophore-conjugated monoclonal antibodies at a 1:50 dilution: anti-CD3 APC-H7 (clone SK7), anti-CD4 PerCP-Cy5.5 (clone SK3), anti-CD25 BB515 (clone 2A3), anti-CD56 PE-Cy7 (clone B159), anti-CD127 Alexa Fluor647 (clone HIL-7R-M21) and anti-CD8 BV421 (clone 3G8) (BD Pharmingen™, San Jose, CA, USA). Cells were incubated for 30 min on ice and protected from light. After the incubation, cells were washed twice with PBS and the final pellets suspended with 300 µL PBS for acquisition on a FACS Verse flow cytometer using the FACSuite software (Becton Dickinson, San Jose, CA, USA). The FlowJo software was used for data analysis. The lymphocyte population was identified by the FSC and SSC parameters, and the FSC-Area vs. FSC-Height was used to eliminate doublets. Within the lymphocyte population, the CD3 + lymphocytes were identified by anti-CD3 APC-H7. Within the CD3 + lymphocytes, CD4 + and CD8 + populations were identified as well. Within the CD4 + population, the T-reg population was identified by plotting anti-CD25 BB515 vs. anti-CD127 Alexa Fluor647.

CAISMOV24 2D and 3D in Vitro Growth Kinetics
Cell division of CAISMOV24 cells in 2D and 3D cultures was assessed by flow cytometry. For this end, CAISMOV24 cells were labelled with violet proliferation dye 450 (VPD450, BD Horizon™, San Diego, CA, USA) according to the manufacturer's instructions, prior being cultivated under 2D or 3D culture systems. At day five of 3D culture four to six cell aggregates of CAISMOV24 were grouped to be processed together for cell disaggregation. Simultaneously, CAISMOV24 cells from 2D culture flasks were treated with trypsin/EDTA for detachment and cell disaggregation. Subsequently, 2D and 3D cell suspensions were acquired in a FACS Verse with FACS Suite software. Cell suspensions were analyzed by setting the appropriate SSC-A/FSC-A gate on tumor cells and considering the fluorescence intensities on day 0 in VPD450 labelled and unlabeled cells. The proliferation platform of FlowJo software was used for the data analysis. The proliferation index was calculated by dividing the total number of divisions by the number of cells that underwent at least one division.

Statistics and Calculations
Multi-comparison analysis of variables was performed by ANOVA followed by a post hoc multiple comparison test. The level of significance was set at p-value < 0.05. The ratio of lymphocytes retrieved from cell aggregates was calculated by dividing the number of events detected within the SSC-A/FSC-A gate of CAISMOV24 cells by the number of events within the SSC-A/FSC-A gate of lymphocytes.

3D cultures and Proliferation Assays
Three dimensional cultures of CAISMOV24 cell line was followed by phase contrast microscopy and representative images are depicted in Figure 1. The first 3 h under magnetic field promoted by the bioprint drive brought CAISMOV24 cells together in a round-shaped structure. Subsequently, under magnetic field of the levitation drive, the initial rounded structure evolved irregularly, generating regions with variable amounts of aggregated cells and spindle-like elongated structures (Figure 1b, 24 h). The final arrangement of CAISMOV24 cells was morphologically defined as papillary, as revealed by histological analysis of the cell aggregates ( Figure 1d). Histological cuts also revealed the presence of focal acinar arrangement with secreted material, pointing out that 3D culture system enabled CAISMO24 cells to evolve glandular-like structures ( Figure 1e). Finally, histological cuts of CAISMOV24 cell aggregates assessed by immunohistochemistry revealed diffuse nuclear expression of PAX8 and progesterone receptor, as well as absence of estrogen receptor expression (Figure 1f-h respectively). VPD450-stained CAISMOV24 cells assessed by flow cytometry showed that mean proliferation index of the cells maintained in 3D cultures (1.87 ± 0.15 times, n = 7) was significantly lower (p < 0.0001) than in the 2D cultures (3.14 ± 0.09 times, n = 3) (Figure 1i). Cells 2020, 9, x FOR PEER REVIEW 5 of 11 papillary, as revealed by histological analysis of the cell aggregates ( Figure 1d). Histological cuts also revealed the presence of focal acinar arrangement with secreted material, pointing out that 3D culture system enabled CAISMO24 cells to evolve glandular-like structures ( Figure 1e). Finally, histological cuts of CAISMOV24 cell aggregates assessed by immunohistochemistry revealed diffuse nuclear expression of PAX8 and progesterone receptor, as well as absence of estrogen receptor expression (Figure 1f,g and h respectively). VPD450-stained CAISMOV24 cells assessed by flow cytometry showed that mean proliferation index of the cells maintained in 3D cultures (1.87 ± 0.15 times, n = 7) was significantly lower (p < 0.0001) than in the 2D cultures (3.14 ± 0.09 times, n = 3) (Figure 1i). The mean proliferation index of CAISMOV24 cells was significant lower (p < 0.0001, t-student test) in 3D culture (n = 7 experimental repetitions) than in the 2D culture (n = 3). MPI = mean proliferation index.

CAISMOV24 Cell Aggregate Contents of Lymphocytes
Three dimensional co-cultures of CAISMOV24 cells with PBMCs (5:1 cell ratio respectively) resulted in aggregates of malignant cells containing lymphocytes. These cell aggregates evolved similarly to what was previously described for aggregates of CAISMOV24 cells alone. Moreover, PBMCs conjugated with NanoShuttle, which were maintained alone under the same 3D culture conditions for comparison purposes, showed the inability of leukocytes to generate aggregates.
The contents of lymphocytes within CAISMOV24 cell aggregates were assessed both by cell disaggregation followed by flow cytometry analysis as well as fluorescence microscopy. It was observed that co-cultures performed with NanoShuttle-conjugated PBMCs had significantly higher

CAISMOV24 Cell Aggregate Contents of Lymphocytes
Three dimensional co-cultures of CAISMOV24 cells with PBMCs (5:1 cell ratio respectively) resulted in aggregates of malignant cells containing lymphocytes. These cell aggregates evolved similarly to what was previously described for aggregates of CAISMOV24 cells alone. Moreover, PBMCs conjugated with NanoShuttle, which were maintained alone under the same 3D culture conditions for comparison purposes, showed the inability of leukocytes to generate aggregates.
The contents of lymphocytes within CAISMOV24 cell aggregates were assessed both by cell disaggregation followed by flow cytometry analysis as well as fluorescence microscopy. It was observed that co-cultures performed with NanoShuttle-conjugated PBMCs had significantly higher (p < 0.05) percentages of CD3+ lymphocytes than those with non-conjugated PBMCs (Figure 2a,b). The presence of leukocytes within CAISMOV24 cell aggregates were confirmed by fluorescence microscopy of cryo-sections obtained from cell aggregates containing CFSE-labelled PBMCs (Figure 2c). The ratio of lymphocytes retrieved from cell aggregates were approximately 20:1 (CAISMOV24: lymphocyte) 72 h after co-culture initiation, being similar between in vitro cultures that were supplemented or not with IL-15 (Figure 2d). Cells 2020, 9, x FOR PEER REVIEW 6 of 11 (p < 0.05) percentages of CD3+ lymphocytes than those with non-conjugated PBMCs (Figure 2a,b). The presence of leukocytes within CAISMOV24 cell aggregates were confirmed by fluorescence microscopy of cryo-sections obtained from cell aggregates containing CFSE-labelled PBMCs ( Figure  2c). The ratio of lymphocytes retrieved from cell aggregates were approximately 20:1 (CAISMOV24: lymphocyte) 72 h after co-culture initiation, being similar between in vitro cultures that were supplemented or not with IL-15 ( Figure 2d). Furthermore, lymphocyte subsets were further assessed. Thus, Figure 3 shows the proportions of NK (CD3 -CD56 + ) lymphocytes, as well as T lymphocytes and their subsets, T-CD4 + (CD3 + CD4 + ), T-CD8 + (CD3 + CD8 + ) and T-reg (CD4 + CD25 + CD127 -). Although, no significant differences were observed in the percentages of T-CD4 + and NK lymphocytes among the different culture conditions (Figure 3a,d), proportions of the T-CD8 + subset within the CD3 + lymphocytes decreased in CAISMOV24 cell aggregates, being significantly (p < 0.05) lower in co-cultures supplemented with IL-15 compared with PBMCs maintained in the same culture conditions (Figure 3b). In addition, the proportion of the T-reg subset within the CD4 + population was significantly higher (p < 0.05) in CAISMOV24 cell aggregates supplemented with IL-15 compared to PBMCs cultures (Figure 3c). Finally, it was observed an upregulation of the CD69 molecule on NK and CD8 lymphocytes retrieved from CAISMOV24 cell aggregates. In this context, CD69 upregulation on CD8 + lymphocytes was associated with supplementation of the culture with IL-15 (Figure 3e,f), while on NK lymphocytes it was associated with presence of CAISMOV24 cells (Figure 3g,h). Furthermore, lymphocyte subsets were further assessed. Thus, Figure 3 shows the proportions of NK (CD3 − CD56 + ) lymphocytes, as well as T lymphocytes and their subsets, T-CD4 + (CD3 + CD4 + ), T-CD8 + (CD3 + CD8 + ) and T-reg (CD4 + CD25 + CD127 − ). Although, no significant differences were observed in the percentages of T-CD4 + and NK lymphocytes among the different culture conditions (Figure 3a,d), proportions of the T-CD8 + subset within the CD3 + lymphocytes decreased in CAISMOV24 cell aggregates, being significantly (p < 0.05) lower in co-cultures supplemented with IL-15 compared with PBMCs maintained in the same culture conditions (Figure 3b). In addition, the proportion of the T-reg subset within the CD4 + population was significantly higher (p < 0.05) in CAISMOV24 cell aggregates supplemented with IL-15 compared to PBMCs cultures (Figure 3c). Finally, it was observed an upregulation of the CD69 molecule on NK and CD8 lymphocytes retrieved from CAISMOV24 cell aggregates. In this context, CD69 upregulation on CD8 + lymphocytes was associated with supplementation of the culture with IL-15 (Figure 3e,f), while on NK lymphocytes it was associated with presence of CAISMOV24 cells (Figure 3g,h).

Discussion
Studies using 3D cell cultures have increased over the past years. This fact is not only a consequence of the improved accuracy delivered by this cell culture approach, but it is also a result of the availability of new, simplified and high-throughput protocols. In this context, the present study successfully evaluated a commercially available 3D culture system to assemble low-grade serous ovarian carcinoma cell aggregates containing lymphocytes.
Frequently, 3D culture models of neoplasms, including EOC, are achieved by culturing tissue fragments, suspension of primary cells or cell lines upon a natural or synthetic polymer matrix [23][24][25][26][27]34,35]. Another way to generate 3D cultures of malignant cells is based on the use of ULA containers that prevent cell adhesion, which combined with suitable culture media and/or agitation promote the growth of epithelial cells as free cell aggregates [36][37][38]. New 3D cell culture technologies have combined magnetic field with ULA containers to promote aggregation of the cells that were previously conjugated with nanoparticles containing iron. This is the case of the n3D-Bioscieces culture system, which was successfully employed to generate cell aggregates of different cell types, including stem and primary cells from humans, as well as other human malignant cells [39][40][41].
In regard to ovarian cancer cell lines, Lee and coworkers [42] used 31 cell lines to compare a 3D culture system based on synthetic polymer matrix with the 2D culture. Heredia-Soto and coworkers [43] recently standardized a 3D culture system employing ULA containers to generate cell aggregates with 16 cell lines. These studies included OAW42 and PEO16 respectively as the only cell lines of low-grade serous histotype. Finally, Pan and coworkers [44] employed the n3D-Bioscieces system based on magnetic field to evaluate the influence of microRNA on cell aggregation of six different ovarian cancer cell lines, all of them of high-grade. Cell aggregates obtained in these studies varied in their morphology, from round to irregular-elongated, and cell compaction, from dense to loose, (e,f) CD69 upregulation on CD8+ lymphocytes is associated with supplementation of the culture with IL-15, while (g,h) on NK lymphocytes it was associated with presence of CAISMOV24 cells. Values were presented as whisker plots and medians (n = 7 experimental repetitions with different blood donors); statistical analyses were performed by ANOVA followed by Tukey's multiple comparisons test. Significant statistical differences are indicated with * (* p < 0.05, **p < 0.01 and *** p < 0.001).

Discussion
Studies using 3D cell cultures have increased over the past years. This fact is not only a consequence of the improved accuracy delivered by this cell culture approach, but it is also a result of the availability of new, simplified and high-throughput protocols. In this context, the present study successfully evaluated a commercially available 3D culture system to assemble low-grade serous ovarian carcinoma cell aggregates containing lymphocytes.
Frequently, 3D culture models of neoplasms, including EOC, are achieved by culturing tissue fragments, suspension of primary cells or cell lines upon a natural or synthetic polymer matrix [23][24][25][26][27]34,35]. Another way to generate 3D cultures of malignant cells is based on the use of ULA containers that prevent cell adhesion, which combined with suitable culture media and/or agitation promote the growth of epithelial cells as free cell aggregates [36][37][38]. New 3D cell culture technologies have combined magnetic field with ULA containers to promote aggregation of the cells that were previously conjugated with nanoparticles containing iron. This is the case of the n3D-Bioscieces culture system, which was successfully employed to generate cell aggregates of different cell types, including stem and primary cells from humans, as well as other human malignant cells [39][40][41].
In regard to ovarian cancer cell lines, Lee and coworkers [42] used 31 cell lines to compare a 3D culture system based on synthetic polymer matrix with the 2D culture. Heredia-Soto and coworkers [43] recently standardized a 3D culture system employing ULA containers to generate cell aggregates with 16 cell lines. These studies included OAW42 and PEO16 respectively as the only cell lines of low-grade serous histotype. Finally, Pan and coworkers [44] employed the n3D-Bioscieces system based on magnetic field to evaluate the influence of microRNA on cell aggregation of six different ovarian cancer cell lines, all of them of high-grade. Cell aggregates obtained in these studies varied in their morphology, from round to irregular-elongated, and cell compaction, from dense to loose, showing that the 3D culture systems did not determine alone the final shape of the cell aggregates. Thus, as stated by Lee and coworkers, 3D culture allows cell lines to reveal certain histological differentiation, even after prolonged culture in 2D [42]. Correspondingly, our results suggested that aggregates of CAISMOV24 cells were shaped not only by the magnetic field, but also by the intrinsic capacity of the growing cells to organize their final arrangement. As a result, cell aggregates observed in our 3D cultures were consistent with the cytological pattern found in peritoneal lavages or ascites from patients with ovarian cancer, which are positive for the presence of malignant cells [45]. Moreover, histological analysis showed that CAISMOV24 cell aggregates displayed papillary morphology, and molecular phenotype consistent with functional EOC cells, particularly, it was detected nuclear expression of PAX8 molecule, which is a biomarker frequently reported occurring in low-grade serous ovarian cancer [35].
Similar to what was previously reported for other EOC cell lines [42,46], we observed that CAISMOV24 cell had a lower cell proliferation index in 3D culture compared with 2D. Such a decrease in cell proliferation would most likely be a consequence of inhibition by cell contact, which happens earlier in 3D cultures than in 2D. However, whether magnetic field would play a role on cell proliferation inhibition remain to be assessed. Although, studies have stated that weak magnetic field do not produce biological effects [39,47,48], there are data suggesting that exposure to magnetic field could specifically target highly proliferative cell populations, such as malignant cells [49,50].
Differently from 3D culture employing ULA plates alone, we hypothesized that since the Bio-Assembler TM kit combines magnetic field with ULA plates, it could enable to assembly of EOC cell aggregates containing lymphocytes. Correspondingly, our results showed that co-culture of CAISMOV24 cells with NanoShuttle TM -conjugated PBMCs under magnetic field boosts significantly the contents of lymphocytes within EOC cell aggregates. Moreover, we demonstrated the feasibility of accessing different lymphocyte subsets within EOC cell aggregates, validating this 3D culture system as a useful in vitro approach to address lymphocyte interactions in EOC microenvironment.
Our results showed that cytotoxic cells (NKs and CD8 lymphocytes) were activated, and pointed the T-reg subtype as a long-term persistent lymphocyte present into the cell aggregates, in a sense similar to what have been shown in patients with EOC [51,52].

Conclusions
We conclude that the 3D culture system allowed CAISMOV24 cell line to develop papillary-like cell aggregates. The culture system also allowed retrieval of lymphocytes from cell aggregates obtained by co-culture of PBMCs and CAISMOV24 cell line. Thus, we assumed this 3D culture system suitable for the study of immune cell interactions in tumor microenvironment.