Materials and Reagents

Cell culture reagents
1. T-175 cm² flasks (Greiner, Cellstar, catalog number: 660175)
2. Human mesenchymal stromal cells derived from the Wharton’s jelly of the umbilical cord (hMSCs), purchased from American Type Culture Collection (ATCC) (ref #PCS-500-010)
3. α-modified Eagle’s medium (α-MEM) (Life Technologies, catalog number: 32561-029)
4. TrypLE^TM Express (Life Technologies, catalog number: 12604013)
5. Phosphate buffer saline (PBS) (Sigma-Aldrich, catalog number: D8662)
6. Fetal bovine serum (FBS) (Life Technologies, catalog number: 10500-064)
7. Penicillin- streptomycin (pen-strep) (Life Technologies, catalog number: 10378-016)
8. Triton X-100 (Sigma-Aldrich, catalog number: X100)
9. Ultra-low-melting agarose (Sigma-Aldrich, catalog number: A5030)
10. Paraformaldehyde (PFA), 16% (Alpha Aesar, catalog number: 43368)
11. Mouse anti-human CD146-Alexa Fluor 647 (clone P1-H12) (BD Biosciences, catalog number: 563619)
12. Rabbit anti-COX-2 polyclonal antibody (Abcam, catalog number: ab15191)
13. Rabbit anti-human VEGF-A monoclonal antibody (Abcam, catalog number: ab52917)
14. Mouse anti-human RUNX-2 monoclonal antibody (Abcam, catalog number: ab76956)
15. Alexa Fluor 488-conjugated goat anti-mouse IgG2a secondary antibody (Life Technologies, catalog number: A-21131)
16. Alexa Fluor 594-conjugated goat polyclonal anti-rabbit IgG secondary antibody (Life Technologies, catalog number: A-11012)
17. DAPI (Sigma-Aldrich, catalog number: 10236276001)
18. Vybrant^TM multicolor cell labeling kit (Life Technologies, catalog number: V22889)
19. hMSC culture medium (see Recipes)
20. Staining buffer (see Recipes)
21. CD146 staining solution (see Recipes)
22. Fixative solution (see Recipes)
23. Agarose solution (see Recipes)
24. Permeablization buffer (see Recipes)
25. Blocking buffer (see Recipes)
26. Primary antibodies solution (see Recipes)
27. Secondary antibodies solution (see Recipes)

Materials for microfabrication and microfluidics
1. Brass plates (5 × 5 cm)
2. Dry-film photoresists: Eternal Laminar E8020, Eternal Laminar E8013 (Eternal Materials), and Alpho NIT215 (Nichigo-Morton)
3. K₂CO₃ (Sigma-Aldrich, catalog number: P5833)
4. Poly(dimethylsiloxane) (PDMS) (Dow Corning, catalog number: SYLGARD 184)
5. 3M^TM Novec^TM 1720 Electronic Grade Coating (3M)
6. 3M^TM Fluorinert^TM Electronic Liquid FC-40 (3M)
7. PEG-di-Krytox (RAN Biotechnologies)

Equipment

Equipment for cell culture
1. CO₂ incubator (Binder, CB 170)
2. Cell sorter (flow cytometer) (BD Biosciences, FACSAria III)

Equipment for microfabrication and microfluidics
1. Office laminator (PEAK, pro PS320)
2. Ultraviolet lamp (Hamamatsu, Lightningcure)
3. Micromilling machine (Minitech Machinery, CNCMini-Mill/GX)
4. Plasma cleaner (Harric, PDC-32G)
5. 100 μl and 1 ml glass syringes (SGE, Analytical Science, Gas tight luer lock syringes)
6. 1 ml plastic syringe (Terumo, SS+01T1)
7. Syringe pumps (neMESYS Low-Pressure Syringe Pump, Cetoni GmbH)

Microscopy
1. Motorized wide-field microscope (Ti, Eclipse, Nikon), equipped with a CMOS (complementary metal-oxide semiconductor) camera (ORCA-Flash4.0, Hamamatsu), a fluorescence light-emitting diode source (Spectra X, Lumencor), and a 10× objective with a 4-mm working distance (extra-long working distance) and a 0.45 numerical aperture (NA) (Plan Apo λ, Nikon)
2. Motorized (Ti2, Nikon) confocal spinning disc microscope equipped with lasers (W1, Yokogawa) and the same camera and objective as above

Software

MATLAB (r2016a, MathWorks, Natick, MA)

Procedure

Microfabrication of the microfluidic chips
1. A detailed protocol on the fabrication of the chips can be found in Amselem et al. (2018) . The chips consist of two parts: 1) The top part comprises a flow focusing junction, a serpentine, diverging rails, and a culture chamber. The molds are fabricated using dry resins that are etched to produce the shape of the droplet generators and guiding channels by standard soft lithography. 2) The bottom of the chips consists of an array of 270 hexagonal capillary traps. The molds of the bottoms part of the chips are fabricated by micromilling the brass plates. The geometry and dimensions of the features can be found in Sart et al. (2020) (Figure 1).
  
  Figure 1. Chip design and depth of the different regions. 1. Inlet for the injection of the aqueous phase containing cells and culture medium; 2. Inlet for oil injection at the junction; 3. Inlet for oil injection to push the drops within the trapping chamber.
2. Fabricate the tops and the floors of the chips by casting poly(dimethylsiloxane) (PDMS) (a mix of 90% of base and 10% of curing agents) into the molds.
3. Place the two molds in an oven set up at 80°C for 2 h to promote the polymerization of the PDMS.
4. Extract the two parts of the chips from the molds and cut them with a scalpel.
5. Assemble the two parts of the chips by bonding the two surfaces for 40 s using a plasma cleaner.
6. Flush the chips with Novec and heat them at 110°C for three cycles to render the inner walls of the chip fluorophilic.

Cell culture, sorting, and loading in microfluidic droplets
1. Cultivate hMSCs using hMSC culture medium, from passage 2 to 7 in regular T-175 cm² flasks, and place them into a culture incubator, set up at 37°C and 5% CO₂.
2. Seed the cells into the flask at a density of 5 × 10³ cells/cm², subcultivate them every week using TrypLE, and change the medium every 2 days.
  Optionally, sort hMSCs based on their level of expression of CD146 by flow cytometry. CD146 is a marker of undifferentiated status, whose level of expression decreases upon differentiation (Sacchetti et al., 2007). To separate the CD146^dim and CD146^bright hMSCs, isolate the cells (passage 5) from the flasks, and incubate them with CD146 staining solution for 30 min. Wash the cells with straining buffer. Identify and select the hMSC population by plotting FSC-A and SSC-A signals using the flow cytometer. Eliminate cell doublets by gating the main population obtained by plotting FSC-H versus FSC-A signals. From this selected population, plot the FITC-A (or any fluorochrome for which the cells are not labeled) versus Alexa-647-A signal distribution; analyze the spread of expression of CD146 with the cytometer and sort 25% of the brightest and 25% of the dimmest cells.
3. Connect two 1 ml syringes containing a 1% RAN in FC-40 solution to inlets #3 and #2, block inlet #1, then flush the oil of each syringe at 50 μl/min to remove the air from the chip (Figure 2, Step 1).
4. Load a suspension of 6 × 10⁶ bulk- or a mix of 50:50 CD146^dim and CD146^bright sorted- hMSCs/ml in culture medium supplemented with the agarose solution into a 100 μl glass syringe.
5. Connect this syringe into inlet #1 (Figure 2, Step 2).
6. Apply a flow rate of 8-8.5 μl/min to syringe 1 and a flow rate of 11 μl/min to syringe 2. The solution containing the cells is pinched at the junction. This yields to the formation of monodispersed droplets containing a suspension of about 380 cells each (Figure 2, Step 2).
7. Apply a flow rate of 50 μl/min on syringe 3: the droplets are pushed on the rails that guide them evenly within the culture chamber. The drops are then captured within the capillary traps (Figure 2, Step 2).
8. Stop the flow of syringe #1 and apply a flow rate of 100 μl/min to syringes #2 and #3 to remove the excess of non-anchored droplets in the culture chamber (Figure 2, step 3).
9. Stop the oil flows; the cells settle down at the bottom of the drops and start clustering.
  
  Figure 2. Protocol for cell loading and droplet formation

Spheroid formation and culture in microfluidic droplets
1. Place the chips into the CO₂incubator overnight to let the cells form spheroids (Figure 3 and Video 1).
  
  Figure 3. Kinetics of MB formation in microfluidic droplets. After stopping the oil flow, the cells are allowed to settle down to the bottom of the drops, where they start clustering. Cells are monitored continuously by live imaging, while forming MBs. Scale bar = 100 μm.
  
  Video 1. Morphology of MB in 3D. The MBs in liquid drops are subjected to recirculation by applying an oil flow around them. The protocol allows the MBs to rotate on their axis. The movie demonstrates the 3D structural organization of MBs.
2. On the second day, place the chips at 4°C for 30 min to gel the agarose.
3. Flush the chips with 1 ml of pure FC-40 oil, at 80 μl/min, to dilute the surfactant.
4. Replace the oil phase with culture medium by slowly flowing the aqueous solution into the device. At this stage, the 3D aggregates are mechanically retained into the traps by the hydrogel. They can be regularly perfused with culture medium; thus, they are now ready for long-term culture while remaining fully viable (Sart et al., 2020).

Spheroid labeling and imaging within microfluidic droplets
To interrogate the structural organization within MBs:
1. Label CD146^dim- and CD146^bright- hMSCs with Vybrant DiO (green) or Vybrant DiD (red) dyes (5 μl for 1 ml culture medium) for 30 min, prior to loading them into the microfluidic drops.
  Note: CD146 protein and the Vybrant dyes are soluble in Triton-X 100 solution; thus, this protocol does not allow to combine immunostaining with the detection of the different CD146 subpopulations.
2. After spheroid formation and culture, image MBs using a fluorescent microscope (e.g., wide-field or confocal microscope equipped with large working distance objectives) (Figure 4).
  
  Figure 4. Spatial organization of CD146^dim and CD146^bright hMSCs within MBs. CD146^dim and CD146^bright are isolated by cell sorting, then labeled with Vybrant DiO (green) or Vybrant DiD (red) dyes prior to their loading into drops. Scale bar = 100 μm.
To interrogate the functional organization within MBs (e.g., the regional expression of VEGF-A, COX-2, or RUNX-2):
1. Fix the MBs by perfusing a solution of 200 µl of fixative solution, by first filling a 1 ml plastic syringe with PFA, then flowing at a flow rate of 80 μl/min.
2. Incubate the aggregates of hMSCs with the PFA solution for 30 min at room temperature.
3. Wash the culture chamber with PBS at the same flow rate (200 μl at 80 μl/min).
4. Permeabilize the MBs by perfusing at 80 μl/min with a 200 μl of permeabilization buffer, and incubate for 5 min.
5. Wash the culture chamber with PBS (200 μl at 80 μl/min).
6. Block MBs by perfusion of a blocking buffer (200 μl at 80 μl/min) and incubate for 30 min.
7. After blocking, perfuse the chamber with a solution of primary antibody (e.g., anti-COX-2 or anti-VEGF-A antibody or anti-RUNX-2; 200 μl at 80 μl/min), and incubate for 4 h.
8. Wash the primary antibody solution by perfusing PBS (200 μl at 80 μl/min).
9. Perfuse the solution of secondary antibodies (200 μl at 80 μl/min) and incubate for 1 h 30 min.
10. Wash with PBS (200 μl at 80 μl/min).
11. Image the MBs in the traps using a fluorescent microscope (Figure 5).
12. To validate the specificity of the primary antibodies, incubate the sample with the secondary antibody only, then wash with PBS. Absence of fluorescent signal validates that the primary antibody is specific to its target (i.e., VEGF-A) and that excess of antibodies is effectively washed with PBS rinsing (Sart et al., 2020).
13. To validate the absence of diffusion limitation, omit the blocking step and incubate with the secondary antibody only. Absence of diffusion limitation is demonstrated by homogeneous fluorescent signal distribution within the aggregate (Sart et al., 2020).
  
  Figure 5. Functional organization hMSCs within MBs: VEGF-A expression detected by immunostaining. Similar results were obtained for RUNX-2 and COX-2 immunostaining (Sart et al., 2020). Scale bar = 100 μm.

Data analysis

The objective of the data analysis is to provide a quantitative characterization of the structural and functional organization within mesenchymal bodies using image analysis. In the following experiments, the data are generated from about 270 replicates that have been reproduced in at least three different chips.

Quantification of the structural organization within MBs using Matlab^®
Detailed information on data analysis can be found in Sart et al. (2017) .
1. Identify the centroid of each spheroid.
2. Measure the area (A) of each spheroid.
3. Calculate R, the radius of each spheroid, defined as the square root of A/π.
4. Measure the fluorescent intensity of each pixel of Vybrant DiO (green, corresponding to CD146^dim hMSCs) or Vybrant DiD (red, corresponding to CD146^bright hMSCs) stained cells, as well as their normalized distance (r/R) to the centroid (Figures 6A and 6B).
5. Bin the values of fluorescence at specific (r/R).
6. Normalize the two fluorescent signals against the DAPI signal to take into account the spherical shape of the aggregates.
7. The relative intensity of the two different fluorescence signals indicates the relative abundance of the CD146^dimvs. CD146^bright hMSCs at different radii within the MBs (Figure 6C) (Sart et al., 2020).
  
  Figure 6. Data analysis methodology to quantify the structural organization in MBs. (A) Cell sorting yielded to the separation of CD146^dim and CD146^bright populations. (B) Position of the fluorescent signal of CD146^dim and CD146^brightlabeled with Vybrant DiO (green) or Vybrant DiD (red) identified by their radial coordinates (r/R). Scale bar = 100 µm. (C) Quantification of the distribution of the different CD146 populations within MBs, calculated based on radial coordinates. Part C is extracted from Sart et al. (2020) .

Quantification of the functional organization withing MB using Matlab^®
Detailed information on data analysis can be found in Sart et al. (2020) :
1. Identify each local maximum of DAPI-stained nuclei (Figure 7A).
2. Construct Voronoi diagrams (Chang et al., 2007) by drawing the perpendicular bisectors of the segments between each neighboring local DAPI maxima, which approximate the cell shapes inside the MBs (Figure 7A).
3. Identify the position of each layer of cells within the aggregate (Figure 7B).
4. Quantify the cellular cytoplasmic fluorescent signal (COX-2, VEGF-A, and RUNX-2) for each cell of the MBs (Figure 7C).
5. Correlate the fluorescent signal of each cell to their cell layer position within the MBs (Figure 7D). To account for the variability of the cytoplasmic signal across the entire cell (nucleus included), the fluorescence signal of a single cell was defined as the mean signal of the 10% highest pixels of the corresponding Voronoi cell (Sart et al., 2020).
6. Bin the values of fluorescence at specific cell layers.
  
  Figure 7. Data analysis methodology to quantify the functional organization in MBs: VEGF-A spatial distribution. (A) Detecting the DAPI signal enables drawing a Voronoi diagram, which allows estimating the area occupied by each cell in the aggregate. (B) The position of each cell is assigned to a layer number. (C) The fluorescent intensity in each cell defined from the Voronoi diagram is measured. (D) The fluorescent intensity of the protein of interest in each cell is correlated to the cell position (layer number) within the MBs, which allows mapping the cell function within the MBs. Similar results were obtained for RUNX-2 and COX-2 immunostaining (Sart et al., 2020).
The results show that hMSCs self-organize in a hierarchical manner: most undifferentiated MSCs are located in the core, while partially committed cells are located at the boundaries of the MBs (Figure 6). Moreover, we found that such structural organization correlated with the induction of osteo-endocrine functions (i.e., COX-2 and VEGF-A expression) (Figure 7 and Sart et al., 2020 for RUNX-2 and COX-2).

Recipes

hMSC culture medium
Mix 50 ml of FBS and 5 ml of the pen-strep (100×) with 500 ml of α-MEM basal medium, which results in a medium containing 10% (v/v) FBS and 1× pen-strep.
Staining buffer
Mix 200 μl of FBS with 9.8 ml PBS, which results in a 2% FBS (v/v) solution in PBS.
CD146 staining solution
Mix 10 μl of Alexa Fluor 647-conjugated anti-CD146 antibody stock solution with 990 μl of staining buffer, which results 1:100 diluted CD146 antibody solution.
Fixative solution
Mix 1 ml of PFA stock solution (16%, v/v) with 3 ml of PBS, which results in a 4% PFA (v/v) solution.
Agarose solution
Weight 30 mg of agarose powder and mix it with 1 ml PBS. Place this solution in a oven set at 80°C for at least one hour, in order to melt the agarose. It results in a liquid agarose solution at concentration of 3% (w/v). Prior to cell loading into drops, mix 30 μl of the the 3% (w/v) agarose solution with 70 μl of medium containing 6 × 10⁶ cells. This results in a solution containing cells at a concentration of 6 × 10⁶ cells/ml and agarose at 0.9% (w/v).
Permeablization buffer
Mix 500 μl of Triton-X100 with 10 ml PBS, which results in a permeablization buffer containing 0.5% Triton-X100 (v/v).
Blocking buffer
Mix 500 μl of FBS with 9.5 ml PBS, which results in a bolocking buffer buffer containing a 5% FBS (v/v) solution.
Primary antibodies solution
Mix 10 μl of primary antibody stock solution (anti-COX2, anti-VEGF-A, anti-RUNX-2) with 990 μl of staining buffer, which results 1:100 diluted primary antibody solution.
Secondary antibodies solution
Mix 10 μl of Alexa Fluor 594-conjugated secondary antibody stock solution and 20 μl DAPI solution at 14 μM (prepared from 5 mg DAPI resuspended in 10 ml PBS, which stock solution (1.4 mM) is then diluted at 1:100, i.e., 10 μl into 990 μl PBS) with 970 μl of staining buffer, which results in a 1:100 diluted solution of secondary antibodies containging 0.3 μM DAPI.

Acknowledgments

C. Frot is gratefully acknowledged for the help with the microfabrication, and F. Soares da Silva is gratefully acknowledged for the help in flow cytometry. We acknowledge the support of the group of Biomaterials and Microfluidics (BMCF) of the Center for Innovation and Technological Research, as well as the Center for Translational Science (CRT)–Cytometry and Biomarkers Unit of Technology and Service (CB UTechS). The original paper (Sart et al., 2020) has been published in Science Advances (doi: 10.1126/sciadv.aaw7853).

Competing interests

The authors declare that they have no competing interests.

References

Amselem, G., Sart, S. and Baroud, C. N. (2018). Universal anchored-droplet device for cellular bioassays. Methods Cell Biol 148: 177-199.
Caplan, A. I. and Correa, D. (2011). The MSC: an injury drugstore. Cell Stem Cell 9(1): 11-15.
Cesarz, Z. and Tamama, K. (2016). Spheroid Culture of Mesenchymal Stem Cells. Stem Cells Int 2016: 9176357.
Chang, H., Yang, Q. and Parvin, B. (2007). Segmentation of heterogeneous blob objects through voting and level set formulation. Pattern Recognit Lett 28(13): 1781-1787.
Dominici, M., Le Blanc, K., Mueller, I., Slaper-Cortenbach, I., Marini, F., Krause, D., Deans, R., Keating, A., Prockop, D. and Horwitz, E. (2006). Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy 8(4): 315-317.
Sacchetti, B., Funari, A., Michienzi, S., Di Cesare, S., Piersanti, S., Saggio, I., Tagliafico, E., Ferrari, S., Robey, P. G., Riminucci, M. et al. (2007). Self-renewing osteoprogenitors in bone marrow sinusoids can organize a hematopoietic microenvironment. Cell 131(2): 324-336.
Sart, S. and Agathos, S. N. (2018). Towards Three-Dimensional Dynamic Regulation and In Situ Characterization of Single Stem Cell Phenotype Using Microfluidics. Mol Biotechnol 60(11): 843-861.
Sart, S., Agathos, S. N., Li, Y. and Ma, T. (2016). Regulation of mesenchymal stem cell 3D microenvironment: From macro to microfluidic bioreactors. Biotechnol J 11(1): 43-57.
Sart, S., Tomasi, R. F., Amselem, G. and Baroud, C. N. (2017). Multiscale cytometry and regulation of 3D cell cultures on a chip. Nat Commun 8(1): 469.
Sart, S., Tomasi, R. F., Barizien, A., Amselem, G., Cumano, A. and Baroud, C. N. (2020). Mapping the structure and biological functions within mesenchymal bodies using microfluidics. Sci Adv 6(10): eaaw7853.
Sart, S., Tsai, A. C., Li, Y. and Ma, T. (2014). Three-dimensional aggregates of mesenchymal stem cells: cellular mechanisms, biological properties, and applications. Tissue Eng Part B Rev 20(5): 365-380.