2.1.

gelMA Microcarrier Synthesis

The gelMA microcarriers were produced following a previously published protocol.¹³ Briefly, gelMA was first synthesized by reacting type A porcine gelatin (Sigma) and methacrylic anhydride at 0.6 mL methacrylic anhydride per gram gelatin in phosphate buffered saline (PBS) for 1 h at 50°C. The reaction was quenched with 40°C PBS then dialyzed through a 12 to 14 kDA nitrocellulose membrane against deionized water at 40°C for 7 days. Then, gelMA and photoinitiator lithium phenyl-2,4,6-trimethylbenzoylphosphinate were dissolved at a final concentration of 7.5% (w/v) and 10 mM, respectively, in PBS at 50°C. A custom polydimethylsiloxane (Sylgard 184) microfluidic device was created and bound to a 2″ × 3″ glass slide after 1 min of oxygen plasma treatment and cured overnight at 85°C. Syringe pumps (Cole Parmer) were used to flow the gelMA solution and fluorinated oil (Novec 7500) through the microfluidic device at 2 and $4\mathrm{mL}/\mathrm{h}$, respectively. The gelMA and oil droplets were then exposed to 365 nm light ($75\mathrm{mW}/{\mathrm{cm}}^2$) for 50 s to polymerize the microcarriers. The gelMA microcarriers were collected from the residual oil solution via centrifugation at 3000 $g$ for 3 min on a glycerol bed. Finally, the microcarriers were recovered from the glycerol bed and washed with PBS and stored at 4°C.

2.2.

Bone Marrow-Derived MSCs Cultured on Polystyrene Microcarriers

Bone marrow-derived human MSCs (BM-hMSCs)²⁵ transfected with red fluorescence protein (dsRed) were expanded on polystyrene microcarriers (P-221, Pall-Solohill) in a 10 mL rotating wall vessel (RWV) bioreactor (Synthecon) and fixed with neutral buffered formalin. The microcarriers have a manufacture-provided size range of 125 to $212\mu \mathrm{m}$. First, lentiviral transduction was used to make cells express dsRed (ThermoFisher, Waltham, Massachusetts). Then, BM-hMSCs were initially expanded in low-density monolayer cell culture in complete culture medium (CCM) [$\alpha$-minimum essential medium, 20% (w/v) fetal bovine serum, 2 mM L-glutamine, $100\mathrm{U}/\mathrm{mL}$ penicillin, and $100\mu \mathrm{g}/\mathrm{mL}$ streptomycin] to obtain the required cell numbers.²⁵ Finally, collagen I-coated polystyrene microspheres and BM-hMSCs were incubated at $1000\text{cells}/{\mathrm{cm}}^2$ at 37°C for 2 h in 10 mL CCM with orbital mixing at 30 revolutions per minute (RPM). Cells were fixed with 4% paraformaldehyde (PFA) and stored in PBS at a concentration of 3 mg particles/mL PBS. Prior to imaging, fixed microcarrier-cell samples were incubated with a $5\mu \mathrm{M}$ DRAQ-5 DNA buffer at 37°C for 30 min with agitation, then rinsed with PBS, and embedded in 1% agarose gel for imaging.

2.3.

ihMSCs Cultured on gelMA Microcarriers

Passage 4 ihMSCs were first expanded in a low-density monolayer cell culture in CCM to obtain the required cell numbers. As previously described elsewhere,²⁴ the ihMSCs²⁶ were cultured in an RWV bioreactor (RCCS-8DQ bioreactor, Synthecon) at 24 RPM on custom $120\pm 6.2\mu \mathrm{m}$ diameter gelMA microcarriers at $1000\text{cells}/{\mathrm{cm}}^2$.¹³ Specimens were suspended in 1 mM concentration of CellTracker Green (CTG) (Thermo Scientific) for 45 min, then fixed with 4% PFA, and stored in PBS. Prior to imaging, samples were incubated with a $5\mu \mathrm{M}$ DRAQ-5 (Thermo Scientific) DNA staining buffer at 37°C for 30 min with agitation and then rinsed with PBS. CTG permits the visualization of the cell cytoplasm and quantification of cell morphology. DRAQ-5 is used for cell nuclei visualization and direct cell enumeration.

2.4.

Light-Sheet Fluorescence Microscopy

In situ fluorescence imaging of cells attached to microcarriers was performed on a Zeiss Z.1 Lightsheet microscope (LSM) using a 20× 1.0 NA (water) detection objective lens and 10× 0.2 NA illumination objective lenses (air). BM-hMSCs were illuminated with 561 (DsRed) and 638 nm (DRAQ-5). ihMSCs were illuminated with 488 (CTG) and 638 nm (DRAQ-5). The emission filters used were 505 to 545 nm (CTG), 600 to 600 nm (DsRed), and 660+ nm (DRAQ-5). A variable zoom of 1.16× was used for an effective magnification of 23.2×, and the voxel size was $0.2\mu \mathrm{m}\times 0.2\mu \mathrm{m}\times 0.45\mu \mathrm{m}$ to satisfy Nyquist sampling requirements. Even illumination of the entire microcarrier surface was achieved using 5% to 15% power with dual-sided objective illumination with pivot scanning and online maximum intensity fusion. $Z$-stacks were acquired at 50 frames per second (FPS) using pco.edge 5.5M sCMOS cameras.

2.5.

Light-Sheet Tomography

Light-sheet tomography based on elastic scattering contrast was performed on the Z.1 LSM with the same illumination and detection optics. The laser blocking filter and emission filters were removed from the optical path between the detection objective lens and the camera to allow scattered light to enter the detection path. Samples were illuminated using the 638 nm laser at 0.1% power, which was sufficient to fill the dynamic range of the camera at 50 FPS. Imaging was performed with dual-sided objective illumination with pivot scanning and online maximum intensity fusion.

2.6.

Mie Scattering Simulations

The open-source Mie scattering software MiePlot (v4.6.21) was used to simulate the angular distribution of Mie scattering intensity.²⁷ The software accepts refractive index ($n$) inputs for the immersion medium and particle, particle size, illumination wavelength, and illumination polarization. The polystyrene particles were modeled as a sphere with a $165\mu \mathrm{m}$ diameter and $n$ equal to 1.59,²⁸ and the gelMA microcarriers were spheres with a $120\mu \mathrm{m}$ diameter and $n$ equal to 1.35.²⁴ Cells were simulated to have a diameter of $10\mu \mathrm{m}$ and $n$ equal to 1.45.^29,30 The combined water and agarose immersion media was modeled with $n$ equal to 1.34. The refractive indices of the agarose immersion media and gelMA microcarriers were measured using a handheld refractometer (KRUSS Optronic).

2.7.

Wavefront Map Simulations

The ray-tracing software ZEMAX was used to simulate the widefield detection arm of the LSM imaging through either the polystyrene or gelMA microcarrier.

The wavefront map simulation was used to calculate the microcarrier-induced wavefront error, represented by peak-valley and root mean square (RMS) error. The objective and tube lenses were simulated as paraxial lenses of focal length equal to 9 and 180 mm, respectively, to represent the 20× 1.0W Olympus objective and Olympus tube lens. The imaging medium was set with $n$ equal to 1.33 to simulate the use of the water-dipping objective lens. The cell, or object plane, is situated on the distal hemisphere of the microcarrier, which is placed at the front focal plane of the objective lens. The gelMA microcarrier was simulated as a $120\mu \mathrm{m}$ sphere of $n$ equal to 1.34; the polystyrene microcarrier was simulated as a $165\mu \mathrm{m}$ sphere of $n$ equal to 1.59; in addition, the gelMA microcarrier was simulated as $165\mu \mathrm{m}$ and the polystyrene as $120\mu \mathrm{m}$ to compare wavefront errors induced by varying the particle size.

2.8.

Data Analysis

The open-source image analysis software ImageJ/FIJI was used to generate 2D intensity projections of the 3D microcarrier fluorescence and scattering volumes. ImageJ was also used to characterize fluorescence signal attenuation with increasing depths by creating an orthogonal projection of the microcarriers and extracting intensity profiles at varying depths of the microcarrier. MATLAB 2020b was used for statistical analysis of the normalized maximum intensity depth profiles of the two microcarriers and the simulated 100% confluent microcarrier. The optical surface area per slice of a sphere was calculated by integrating the sphere surface area using the following equation:

3.1.

Mie Scattering Simulations

To assess the potential for elastic scattering-based contrast imaging of cells expanded on spherical microcarriers, Mie simulations for the two microcarriers were performed using MiePlot. The angular distributions of Mie scattering intensity for the two microcarriers and a cell were simulated using an unpolarized light source [Fig. 2(a)]. In addition, the gelMA microcarrier was simulated as a $165\mu \mathrm{m}$ diameter to match the diameter of and compare it to the polystyrene microcarrier [Fig. 2(b)].

Fig. 2

Mie scattering intensity simulations of polystyrene and gelMA microcarriers and single cells on (a) rectangular and polar plot showing the polystyrene microcarrier orthogonal scatters at much higher intensities, which inhibits the visualization of cells along the surface using LSM. (b) The gelMA microcarrier was simulated as the same diameter ($165\mu \mathrm{m}$) as the polystyrene microcarrier to decouple the scattering intensity from microcarrier size.

These simulations show that the polystyrene microcarriers scatter light much more intensely than the hydrogel microcarriers and the cells. This phenomenon is due to the refractive index mismatch of the surrounding water immersion media and the microcarrier. The Mie simulations predict the scattering intensity to be two to four orders of magnitude greater for polystyrene than gelMA microcarriers. The cells, with an overall refractive index between water and polystyrene, scatter light near the same intensity range as the hydrogel microcarriers. This phenomenon permits the visualization of cells along the entire gelMA microcarrier surface using both fluorescence and elastic scattering contrast.

3.2.

Wavefront Map Simulations

To assess the optical aberrations induced by imaging through the microcarrier, the light-sheet detection arm was simulated in ZEMAX with the distal microcarrier surface at the objective lens front focal plane. The wavefront map simulation was used to quantify the microcarrier-induced peak–valley ($P-V$) and RMS wavefront errors, which are key indicators of optical system performance. These simulations show that the gelMA microcarrier causes less optical path differences (OPD) than the polystyrene microcarrier (Fig. 3). In fact, the gelMA RMS error (0.0003 waves) is $<1/4\lambda$, which is the Rayleigh criterion that describes the threshold for diffraction-limited imaging performance [Fig. 3(a)]. Increasing the gelMA microcarrier diameter from 120 to $165\mu \mathrm{m}$ increases the RMS wavefront error to 0.0064 waves [Fig. 3(b)]; however, this is still less than the error (2.8707 waves) introduced by the $160\mu \mathrm{m}$ polystyrene microcarrier [Fig. 3(c)]. When the polystyrene microcarrier is simulated as a $120\mu \mathrm{m}$ diameter, the RMS wavefront error (1.7947 waves) is still greater than that of the 120 and $165\mu \mathrm{m}$ gelMA microcarriers. These simulations illustrate that the custom-fabricated gelMA microcarrier induces fewer optical imaging aberrations compared with commercially available polystyrene microcarriers.

Fig. 3

ZEMAX simulations of wavefront errors introduced by 120 and $165\mu \mathrm{m}$ diameter gelMA and polystyrene microcarriers. gelMA microcarriers of (a) $120\mu \mathrm{m}$ diameter introduce an RMS OPD of 0.0003 waves, and increasing the microcarrier diameter to (b) $165\mu \mathrm{m}$ introduces RMS OPD of 0.0064 waves. Polystyrene microcarriers of (c) $165\mu \mathrm{m}$ diameter introduce severe aberrations, resulting in RMS OPD of 2.8707 waves, and decreasing the microcarrier diameter to (d) $120\mu \mathrm{m}$ only slightly improves the image quality, resulting in RMS OPD of 1.7947 waves. Zoomed-in view of (e) 120 and (f) $165\mu \mathrm{m}$ gelMA microcarrier minimally aberrated wavefronts.

3.3.

Microcarrier Screening for Label-Free Elastic Scattering-Based Visualization

The polystyrene microcarrier scatters much more intensely than cells cultured on the microcarrier surface. This causes the cell signal to be obscured by the polystyrene microcarrier scattering [Fig. 4(a)]. Only cells situated on the polystyrene microcarrier equator and proximal hemisphere can be visualized using brightfield transmitted illumination [Fig. 4(b)]. This necessitates the use of fluorescence contrast to visualize and distinguish cells from the polystyrene microcarrier surface [Fig. 4(c)]. The gelMA microcarrier, conversely, scatters at intensities near or lower than cells cultured on the gelMA microcarrier surface, which permits the sub-cellular visualization of cells along the entire 3D surface using light-sheet tomography [Fig. 4(d)]. Standard deviation and sum of slices contrast enhancement projections even permit the simultaneous visualization of the microcarrier, cells, and surrounding agarose. Brightfield illumination can also be used to visualize and identify cells from the gelMA microcarrier surface [Fig. 4(e)]. Populated microcarriers can be distinguished from empty microcarriers using brightfield illumination to visualize textural features on the microcarrier surface. Fluorescence contrast similarly provides an equivalent view of cells cultured on the gelMA microcarrier surface [Fig. 4(f)]. These exemplary volumetric projections illustrate the compatibility of the hydrogel microcarrier for the label-free visualization of cells along the entire microcarrier surface.

Fig. 4

The gelMA microcarriers are superior for the label-free visualization of the entire culture surface area. Label-free intensity-based projections of the polystyrene microcarrier using (a) elastic scattering and (b) brightfield contrast provide limited visualization of cells on the microcarrier surface compared with (c) exogenous fluorescence contrast. The gelMA microcarrier permits the use of label-free (d) elastic scattering and (e) brightfield, and (f) exogenous fluorescence contrast for the visualization of cells along the entire microcarrier surface.

3.4.

Microcarrier Screening for Direct Fluorescence-Based Cell Enumeration

To evaluate the optical imaging performance of the different microcarriers, fixed MSCs attached to commercially available $165\mu \mathrm{m}$ diameter polystyrene and $120\mu \mathrm{m}$ diameter custom-fabricated hydrogel microcarriers were imaged with an LSFM. Representative data of DsRed and DRAQ-5 labeled BM-hMSCs on polystyrene and CTG and DRAQ-5 labeled ihMSCs on gelMA microcarriers are shown in Figs. 5 and 6.

Fig. 5

Polystyrene microcarriers limit visualization to the proximal microcarrier surface. Maximum (a) dsRed and (b) DRAQ-5 intensity projections of the proximal microcarrier surface. (c) dsRed and d) DRAQ-5 maximum intensity projections of the distal microcarrier surface. (e) dsRed and (f) DRAQ-5 signal intensity is (g) attenuated beyond the microcarrier equator. Line projections of fluorescence intensity in (g) correspond to dashed lines in (f).

Fig. 6

gelMA microcarriers enable high resolution visualization of the entire microcarrier surface. Maximum (a) CTG and (b) DRAQ-5 intensity projections of the proximal microcarrier surface. (c) CTG and (d) DRAQ-5 maximum intensity projections of the distal microcarrier surface. (e) CTG and (f) DRAQ-5 signal intensity is (g) maintained throughout the entire depth of the microcarrier. Line projections of fluorescence intensity in (g) correspond to dashed lines in (f).

The polystyrene microcarrier inhibits the visualization of cells on the distal half of the microcarrier equator [Figs. 5(a)–5(f)]. There are 30 cell nuclei identified via DRAQ-5 on the microcarrier equator and proximal surface, only one or two additional cell nuclei can be identified on the distal surface near the equator, and no cell nuclei are resolved in the center of the distal microcarrier hemisphere [Figs. 5(b) and 5(d)]. This is because the fluorescence signal intensity is attenuated with depth, and the ability to visualize, resolve, and enumerate cells beyond the microcarrier equator is severely hindered [Figs. 5(e) and 5(f)]. Cells on the distal half of the microcarrier are illuminated, but the polystyrene microcarrier blurs these slices in the stack and the ability to resolve individual nuclei is eliminated [Fig. 5(g)]. A scattering artifact caused by the microcarrier can be seen on the left hemisphere of the microcarrier in Figs. 5(c) and 5(d).

The hydrogel-based gelMA microcarriers, compared with commonly used polystyrene microcarriers, enable high fidelity visualization and semi-automatic direct enumeration of cells along the entire microcarrier surface [Figs. 6(a)–6(d)]. There are seven cell nuclei resolved from the DRAQ-5 contrast [Fig. 6(f)]. The fluorescence intensity is preserved along the entire depth of the microcarrier, and cell nuclei can be visualized on both the proximal and distal hemisphere of the microcarrier [Figs. 6(e)–6(g)].

3.5.

Depth-Dependent Fluorescence Signal Attenuation

To compare the depth-dependent fluorescence signal intensity attenuation caused by the microcarrier material, the maximum DRAQ-5 intensity per frame per microcarrier profile is averaged, normalized, and plotted for both microcarrier materials; in addition, the theoretical illuminated microcarrier surface area is plotted as a reference for a 100% confluent microcarrier. Only single microcarriers with two or more cells and $>20\%$ CTG confluency were included to select for moderate and highly confluent single microcarriers and to avoid including aggregated microcarriers in the analysis. The gelMA microcarriers have a more symmetrical profile than the polystyrene microcarriers, and these data show that the gelMA microcarriers cause less depth-dependent fluorescence signal intensity attenuation than the polystyrene microcarriers (Fig. 7). A two-sample Kolmogorov–Smirnov test was performed to quantitatively assess the similarity of the depth-intensity profiles between the polystyrene and gelMA microcarriers. The gelMA and polystyrene microcarriers showed a statistically significant difference from each other ($P=0.0314$). The gelMA microcarrier profile did not show statistically significant difference from the 100% confluent microcarrier ($P=0.0994$), whereas the difference between the polystyrene and 100% confluent microcarrier profiles was found to be statistically significant ($P=0.0082$). Data were considered to be significant if the $P$ values were $<0.05$. Both microcarrier material intensity profiles show a decrease in intensity and deviation from the theoretical confluent microcarrier beginning at around 55% depth; this is an interesting phenomenon that may be due to the spherical nature of the microcarriers or a sample size artifact. Mie scattering of spherical objects is well known to have both broad and sharp intensity oscillations termed wiggles and ripples, respectively, that are due to the spherical shape of the scattering object;^31,32 the depth-dependent fluorescence intensity oscillations seen here could similarly be related to the spherical nature of the microcarriers.

Fig. 7

gelMA microcarriers have superior volumetric optical imaging capabilities compared polystyrene microcarriers. Normalized average maximum intensity-depth profile of gelMA ($n=17$) and polystyrene ($n=31$) microcarriers showing the large fluorescence intensity attenuation that occurs on the distal half of the polystyrene microcarriers compared with gelMA microcarriers. A five-point sliding average filter was applied to both microcarrier profiles (solid lines). A 100% confluent microcarrier is simulated using the theoretical surface area per optical slice.