Microarray Embedding/Sectioning for Parallel Analysis of 3D Cell Spheroids

Three-dimensional cell spheroid models can be used to predict the effect of drugs and therapeutics and to model tissue development and regeneration. The utility of these models is enhanced by high throughput 3D spheroid culture technologies allowing researchers to efficiently culture numerous spheroids under varied experimental conditions. Detailed analysis of high throughput spheroid culture is much less efficient and generally limited to narrow outputs, such as metabolic viability. We describe a microarray approach that makes traditional histological embedding/sectioning/staining feasible for large 3D cell spheroid sample sets. Detailed methodology to apply this technology is provided. Analysis of the technique validates the potential for efficient histological analysis of up to 96 spheroids in parallel. By integrating high throughput 3D spheroid culture technologies with advanced immunohistochemical techniques, this approach will allow researchers to efficiently probe expression of multiple biomarkers with spatial localization within 3D structures. Quantitative comparison of staining will have improved inter- and intra-experimental reproducibility as multiple samples are collectively processed, stained, and imaged on a single slide.

The final size of spheroids was dependent on initial seeding concentration. Concentrations of 4.0x10 4 cells/mL (1600 cells/spheroid) created spheroids <200 µm that formed readily. An intermediate cell concentration of 4000 cells/spheroid formed spheroids with diameters of approximately 350-400 µm. Seeding concentrations in hanging-drops greater than 2.0x10 5 cells/mL (8000 cells/spheroid) created culture environments that failed to properly aggregate. The addition of low viscosity methylcellulose (25 cP, Sigma M6385) to culture medium appeared to help aggregation of spheroids at appropriate cell concentrations.
The 96 well GravityPLUS™ hanging drop plate (InSphero), was used to suspend cells in 40 μL droplets. A final concentration of 800,000 cell/mL resulted in spheroids with approximately 32,000 cells and a disc shape with a diameter of ~ 500 um and a height of 200 um after 3 days of culture. Media was changed after every two days by adding 20 uL of media containing 2.4 mg/mL methylcellulose to each well with a 12-channel multipipettor and then removing 20 uL of media. The process was repeated twice for each media change. The supplementation of high viscosity methylcellulose in the hanging droplet medium was a critical promoter of spheroid aggregation. The bottom chamber of the GravityPLUS™ plate contains a humidifying pad provided by InSphero as well as 7.5 mL of deionized water and 7.5 mL phosphate buffer saline (PBS). Culture plates were wrapped in paraffin during incubation to prevent media evaporation.

S.3 Spheroid Fixation
After the 72 hour time point, spheroids were transferred into GravityTRAP™ plates (InSphero) by stacking the GravityPLUS™ plate over the GravityTRAP™ plate as described in manufacturer instructions. The wells of the receiving GravityTRAP™ plate were prefilled with 70 µL of PBS and a volume of 70 µL of PBS was added to the hanging drops in the GravityPLUS™ plate so that the hanging droplet made contact with the fluid in the well below. The spheroid and some fluid from the droplet transferred into the GravityTRAP™ plate receiving well. PBS was removed from the GravityTRAP™ wells containing the spheroid according to manufacturer instructions and 2% paraformaldehyde (PFA) (Electron Microscopy Sciences) was added and allowed to incubate for 30 min -2 hours. After PFA incubation the spheroids were washed within the GravityTRAP™ plate wells and stored in PBS until they were transferred to a microarray for further fabrication.

S.4 Agarose Embedding Optimization
Agarose solution was then added to the mold and was cured as described in Methods. It was observed that high agarose concentration could result in brittle gels prone to pillar fracture or had missing pillars due to incomplete diffusion of highly viscous agarose solutions into the bottom of wells. Low agarose concentration could result in gel geometries that did not match the dimensions of the negative mold. Concentrations of 0.5%, 1%, 2%, 3%, and 4% agarose (w/v) were tested; 3% agarose (w/v) solution was determined to best ensure the complete gelation of pillars for microarray formation and limit the probability of pillar fracture during pullout. Agarose microarrays fabricated from 2.5-3% solutions added at 80-90°C were easily removed from the mold and contained the spheroids embedded in robust pillars. Spheroid location in the agarose cylinders was tangential to the top surface in height (z-plane) and at a random location in the radial/angular plane (xy plane). Figure 3F shows a microarray of 96 spheroids in the same plane during processing.

S.5 Microtome Alignment & Histological Processing
The angle of the sectioning plane relative to the plane of the spheroid microarray was analyzed and will be referred to as "tilt." The microtome (KEDEE KD-2258) used in these experiments has a head that is adjustable around the x-, y-, and z-axes, and translates forward along the z-axis as each section is taken ( Figure 3G). The desired orientation of the face of the microarray block is parallel to the xy plane. The bubble level attached to the cassette receiver was used to align the cassette about the x-and z-axes ( Figure 3G); however, alignment of tilt about the y-axis required visual alignment using an iterative process. A blank cassette was placed into the cassette receiver and the microtome blade was installed. The receiver was then adjusted to an initial setting expected to align in the xy-plane. The receiver was translated forward towards the fixed blade until contact between the blade and blank cassette was made; the blade position was verified against the blank cassette at various values along the z-axis, precisely at the intersection between the blade, and each of the two vertical edges of the orange cassette. This relationship was examined at cassette levels y = 0 (corresponding to the level of the blade; middle), y = 0.5 cm (top), and y = -0.5 cm (bottom), to assure alignment was consistent across the entire xy plane of the cassette. Alignment was considered acceptable when both vertical edges of the blank cassette met the statically locked blade at the same position during forward (z) translation in all three locations along the y-axis (top, middle, bottom of the cassette). Visual verification of contact was limited in exactness to approximately 50 µm across the width of the cassette. After successful alignment of the blank cassette, the receiver was translated away from the blade, and the cassette containing the microarray was mounted. For the sake of the study, it was assumed that all cassettes were manufactured identically. Alternatively, the blade angle was changed to zero degrees and two blank cassettes were loaded in the microtome with the positioning of the cassette mount free to rotate. The blade assembly was moved forward to press flush with the face of the cassettes. The cassette orientation locks were tightened while the cassette was flush with the blade assembly.
Then the cassette position was moved back and the blade angle was changed to 5˚ (see

Supplemental Video 3)
The paraffin sections were place in an oven at 48˚C for an hour to increase adherence to microscope slides (Tru Scientific -TruBond 380). Slides were deparaffinized, hydrated and stained with Hematoxylin and Eosin (H&E) as follows:

S.6 Theoretical Tilt Analysis
Optimally, the plane containing the spheroid array in the paraffin block will be exactly aligned to the cutting plane of the microtome (xy). Minimizing tilt (the angle between these two planes; θt, in Figure 4) is critical to obtaining single histological sections with many spheroids, each of adequate area for analysis. A schematic model ( Figure 4A Sample calculation is for a 12x8 array in the direction of 12 spheroids in the array. A calculation for a 500 µm spheroid array is used. It is assumed that the focal point of tilt angle is at the center line of the array and directly in between the 6 th and 7 th spheroid in the array ( Figure 4A). It is assumed that the spheroid array and the section line is straight and the section has zero thickness.

S.8 Supplemental Calculation 2: Estimating Area of Outer Spheroid Transferred to a Section
Based on Angle of Tilt Sample calculation is for the conditions and assumptions listed top left of the figure. First the vertical distance from center of the circle to the cutting plane is estimated. Then the length of the horizontal cross-section through that point is estimated. The radius of that distance is used to provide a rough estimate of the area transferred to a histological slide. Center focal point and array geometry is the same as in Figure 4A. Area is presented as a percentage of the area of a horizontal cross-section passing through the center of the circle.

S.9 Experimental Tilt Analysis
To determine the angles of tilt that are associated with the spheroid microarray method in practice, an experimental analysis was conducted using a blank agarose microarray embedded in a paraffin block (Figure 4D-E). The plane of a perfect theoretical spheroid array embedded in the agarose microarray was assumed to be parallel to the plane of the tips of the micropillars of the array, offset along the negative z-axis by a value equal to the average spheroid radius, so that the plane passes

S.10 Post-Processing Effects on Spheroid Size (Swelling & Contraction)
After fixation in paraformaldehyde, but before agarose infiltration, spheroids were digitally imaged and measured under bright field microscopy while immersed in PBS (Figure 5A).
Minimum and maximum spheroid diameters were used to estimate the "pre-processing maximum spheroid cross-sectional area" assuming elliptical geometries. This area was deemed the maximum possible cross-sectional area attainable by any single section from the individual spheroid, and theoretically could only be obtained should a section be taken precisely through the center of the spheroid (Figure 4A, Ɵ=0º). After histological processing and prior to H&E staining ( Figure 5B), minimum and maximum diameters were re-measured for each of ten randomly selected spheroids within a 6x4 array for all sections taken to calculate the cross-sectional area.
The largest single value from all sections was called the "post-processing maximum spheroid cross-sectional area" and was compared with the "pre-processing maximum spheroid crosssectional area" using a t-test. The null hypothesis was that processing had no effect on the mean cross-sectional areas.

S.11 Determining Spheroid Number Per Histological Section
Next, the number of spheroids successfully transferred to slides (termed "recovery") per section was quantified to determine the utility of the process. Four independent spheroid microarray blocks each containing a 6x4 array of ~ 500 µm diameter spheroids were analyzed. Spheroid recovery was quantified based on a 1/0 score for each spheroid in each well of each section taken from each microarray (n=20). Only spheroids with post-processing cross-sectional areas above the threshold of 25% of the pre-processing maximum spheroid cross-sectional area were scored as a 1, and considered successfully recovered. The procedure for determining recovery (1/0) was the following. First, fixed spheroids in PBS were imaged in wells and measured before processing.
Each spheroid was approximated as an ellipse with measurements taken for both the long and short axis of the spheroid. These dimensions were used to determine the "pre-processing average • Hole size and spacing -Hole size and spacing will determine how easily the array is filled and how many spheroids can be packed into a small area. Hole size will also effect the tendency of spheroids to be jarred loose from holes when adding agarose. Hole size and depth will affect agarose pull out (limit possibility of pillar fracture) and may affect how well water and agarose mix when the agarose is added to the assembled microarray.
• Agarose & protocol -All data presented here used UltraPure™ agarose (Invitrogen, 16500500). However, we have explored other types of agarose and some may be optimal for specific experiments. Agarose concentration is important for mixing with water in pillars, encapsulation of the spheroids, possible pillar fracture and dehydration. A specific type and concentration of agarose may be ideal for different protocols. The temperature of the agarose when it is added is also critical. Higher temperatures promote complete mixing with water in the pillars ( Figure 3D), but very high temperature could damage biological samples. In our experiments we looked at several methods of preheating the PDMS mold (briefly or water can evaporate out of the pillars) and keeping the PDMS mold warm after the liquid agarose is added. The latter was generally accomplished by keeping the PDMS mold on a hot plate or moving it to an oven for incubation.
• Dehydration & embedding protocol -Proper dehydration is critical to the success of this method. If dehydration is too fast then the agarose microarray will swell/contract and alter the in-plane geometry that is critical to success. Sufficient time and agitation and visual inspection or dimensional measurement between steps are recommended to determine if swelling/contraction is occurring. Also, the size of the agarose array must be considered.
Larger arrays are more prone to contraction/swelling due to larger penetration distances required for reagents. In our experiments, the final overnight 100% ethanol wash was performed with a freshly opened bottle. On one of our samples reported in this manuscript ( Figure 6C) there are internal pillars that appear lower the perimeter pillars. This is likely due to some contraction in the center of the microarray.
• Microtome alignment -Microtome alignment is a key component for achieving the greatest number of spheroids per section. We have presented two methods of aligning the microtome to the cutting plane, but this may be dependent on the model of microtome used. A fixed mount (non-adjustable) may be ideal for this approach.
• Plastic negative mold approach (Figure 2(#3 (2) Producing a good seal at the bottom of the mold. We found that a freshly made PDMS film was ideal for sealing. The type of plastic and surface roughness both effect how effective the seal is as well as the surface of the PDMS film. We also found that a high-quality spheroid microarray can be produced even with some slow/limited leaking occurring.
(3) Depositing agarose across the tissue cassette grid. The agarose level must extend above the tissue cassette grid to firmly attach the agarose microarray to the tissue cassette. If the microarray cavity extends outside of the edge of the tissue cassette or if the outside edge of the mold does not extend beyond the edges of the tissue cassette then agarose can leak out and it may be challenging to fill with agarose to the required height. These problems can be overcome by carefully adding agarose to the center of the tissue cassette or modifying the geometry of the mold.
• Spheroid Transfer -In our experiments were found that spheroid transfer could be more efficient if a small amount of liquid was taken up (as low at 5 µL) so that the spheroid fell to the tip of the pipette quickly. Flicking the tip of the pipette was an effective method of releasing spheroid that were stuck to the inner surface of the pipette tip so that they would fall to the tip of the pipette with gravity.