Whole-mount Retinal Organoid Visualization with Cellular Resolution

Summary

This protocol combines optical clearing and immunolabeling for full-volume confocal imaging of whole-mount retinal organoids. It preserves 3D structure, enabling detailed visualization of key neuronal pathways, improving the study of retinal maturation, spatial organization, and its potential applications in disease modeling and personalized medicine.

Abstract

Retinal organoids derived from human induced pluripotent stem cells are intricate 3D structures that mimic the human retina, offering a powerful platform for studying retinal development, disease mechanisms, and potential therapeutic strategies. Moreover, as they are derived from patients, they are becoming increasingly popular as they hold great promise as a tool for personalized medicine. Unlike conventional 2D cell cultures, retinal organoids preserve the native 3D architecture of the retina, allowing for a more realistic representation and enabling more physiologically relevant studies. However, their structural complexity, high cellular density, and diverse composition present significant challenges for characterization.

To address these challenges and enhance our understanding of retinal organoid maturation while preserving the 3D context, we combined optical clearing methods with immunolabeling to visualize the entire structure of whole-mount organoids with confocal microscopy. For this, we employed a clearing method compatible with low- and high-numerical-aperture objectives, facilitating full-volume imaging and capturing certain regions of interest with cellular resolution. Using this approach, we identified the morphology and distribution in 3D of the three main neuron paths responsible for the visual information transmission: cone and rod photoreceptors, bipolar and ganglion cells. Our findings shed more light on the visualization techniques to address the spatial organization of retinal cells within the organoid.

Introduction

Retinal organoids

Retinal organoids are complex 3D structures that mimic the human retina, offering a valuable platform for studying retinal development, disease mechanisms, and potential therapeutic strategies¹. They maintain the natural 3D architecture of the retina, demonstrate transcriptomic fidelity and functional competence, reproducing phototransduction and synaptic connectivity². Additionally, they closely mimic in vivo development^3,4, allowing for a more realistic representation as compared to flat cell cultures. In previous studies, Isla-Magrané et al. established a two-step protocol to obtain multiocular organoids derived from human induced pluripotent stem cells (hiPSC) to model cellular features of the human eye in development^5,6. Due to the scarcity of fetal tissue, human eye development research has been restricted to anatomical studies.

The retinal organoids obtained with this protocol exhibited proper layering and contained all major retinal cell subtypes, such as photoreceptors, bipolar cells, ganglion cells, amacrine and horizontal cells, and macroglia, along with key morphological features and correct spatial organization and connections. Organoid models, although a potential solution, exhibit significant variability, making it difficult to achieve consistent results. Therefore, there is a pressing need to employ a variety of advanced technologies in more reliable human models. To date, these organoids have only been studied using histological sections, as their complex cellular composition, density, and size hinder light penetration, resulting in poor optical imaging performance in deeper layers, thus limiting comprehensive analysis.

Whole-mount imaging

Traditional histological techniques, which involve sectioning and staining of fixed tissues, can provide detailed information about specific regions but fail to capture the overall architecture and spatial relationships within the entire organoid. To overcome this limitation, whole-mount imaging emerges as a valuable tool for studying retinal organoids. Whole-mount imaging allows visualization of the entire organoid in its intact 3D form, preserving the spatial distribution of cell types and extracellular matrix components. This approach is particularly beneficial for investigating the cellular composition within and between organoids, providing a more complete understanding of their developmental processes and functional organization. However, due to their cellular density and compaction, retinal organoids are opaque (Figure 1, uncleared). The inherent opaqueness of samples causes light scattering, which hinders detailed visualization of internal structures using microscopy. Here, optical clearing techniques emerge as a powerful tool.

Optical clearing techniques

Optical clearing refers to physicochemical treatments that make thick biological samples transparent by homogenizing the refractive index (RI) gradients of the tissue to match those of the clearing agent and removing light-absorbing substances⁷. These methods dramatically reduce light scattering and increase transparency. This transformation allows us to see through the entire retinal organoid, enabling comprehensive analysis of its intricate cellular organization and morphology through whole-mount imaging techniques. In practice, this is achieved by impregnating the sample with a high RI substance and removing lipids and pigments.

Different clearing methods are available, differing in various aspects (e.g., degree of transparency, compatibility with microscopic techniques, specimens, and fluorescent molecules, feasibility, duration), but none stands out as the best common protocol⁸. They are classified into four major groups: (1) simple immersion, (2) hydrophobic of organic solvent-based, (3) hyperhydration-based, and (4) tissue transforming- or hydrogel embedding-based, as nicely summarized by Avilov et al⁷. All methods include a step for the RI matching, usually by dipping the sample in a solution that matches the average RI of the tissue. In hydrophobic methods, lipids and water are removed by dipping the sample in an increasing organic solvent concentration (such as methanol or propanol).

The presented protocol is an optimized method that integrates optical clearing with immunolabeling to facilitate whole-mount 3D visualization of retinal organoids with confocal microscopy, enabling full-volume imaging while preserving structural integrity. This approach allows for the identification of key retinal cell types within the organoid, providing researchers with a powerful tool to investigate retinal organoid maturation and spatial organization in a physiologically relevant context.

Protocol

1. Preparation of custom-made fluorescent secondary antibodies

NOTE: This approach serves as an alternative to the use of commercial fluorescent antibodies (see the discussion for the advantages of preparing fluorescent secondary antibodies). Moreover, this protocol can be used for conjugating fluorophores to primary antibodies in case direct immunofluorescence is to be performed and the fluorescent primary antibody is not commercially available. Controlling the fluorophore-to-antibody ratio by adjusting the amount of fluorophore in the labeling reaction produces brighter conjugated antibodies. However, custom-made secondary antibodies are less stable, and the concentration of the resulting conjugate is 10x lower than that of the commercial fluorescent antibodies. This protocol is adapted from the work described by Bálint et al⁹.

1. Dissolve 1 mg of the fluorophore in anhydrous DMSO and aliquot into tubes for a final 0.02 mg of fluorophore per tube. Remove all DMSO using a speed vacuum concentrator, running for 1 h and 30 min. Store aliquots at -20 ºC protected from light. For Alexa Fluor 405, ultrapure H₂O is the appropriate solvent for making aliquots.
2. For performing the labeling reaction, dissolve one aliquot of the fluorophore in 1-10 µL of DMSO (or ultrapure H₂O for Alexa Fluor 405).
   NOTE: The exact amount of DMSO is determined through previous control conjugation tests. These tests are performed when a new batch of a fluorophore is to be used (or when running out of a previous fluorophore batch), or when a new fluorophore is used for the first time in the conjugation. The control conjugation test involves carrying out labelling reactions with varying amounts of fluorophore to determine the optimal fluorophore-to-antibody ratio in the conjugate. In our case, the optimal ratio typically falls between 3 µL and 5 µL.
3. Incubate 50 µL of secondary IgG (or primary), 6 µL of 1 M NaHCO₃, and 1-5 µL of fluorophore for 40 min at room temperature (RT) on a rocking platform, protected from light.
4. While the reaction is progressing, prepare the purification size exclusion columns by removing its lids and letting the buffer pass through. Equilibrate the column (one for each labeling reaction) by running 3 x 2-3 mL of PBS through the column.In case the last equilibration step finishes before the incubation ends, put the lids back to avoid them drying up and wait until the labeling reaction finishes.
   NOTE: Purification size exclusion columns work by gravity. Make sure that the column does not dry up at any moment, or the purification reaction will not work properly.
5. When the reaction finishes (step 1.3), add 140 µL of PBS to the labeling reaction to make the volume approximately 200 µL and vortex it. Add this solution to the center of the column, allow the sample to enter the column, and after the last drop has eluted, push by adding 550 µL of PBS. When the liquid stops falling, elute by adding 300 µL of PBS and collect in a 1.5 mL microcentrifuge tube.
6. Calculate the antibody concentration and labeling ratios by measuring the absorbance of the sample at 280 nm and the absorbance of the fluorophores used. Apply the Beer-Lambert Law to determine the concentration (in molarity) as follows:
   
   Where A stands for the absorbance, ε for the molar extinction coefficient, and CF₂₈₀ is a measure of the absorbance that the specific fluorophore has at 280 nm that serves as a correction factor. Then, the concentration of the antibody is:
   
   And the labeling ratio is:
   
7. Store the labeled antibodies at 4 ºC, protected from light, for up to 6 months.

2. Immunofluorescence labeling

1. Fix the organoids with 4% paraformaldehyde (PFA) at RT for 45 min.
   NOTE: OPTIONAL: Incubate the organoids with an antigen-retrieval solution (steps 2.2 and 2.3) to enhance the immunoreactivity of the antigens.
2. Prepare 40 mL of antigen-retrieval solution by mixing 0.117 mg of sodium citrate tribasic dihydrate in ultrapure H₂O and adjust to pH 9.
3. Incubate the organoids with the antigen-retrieval solution with mild shaking (30 rpm) at 60 °C for 1 h.
4. Permeabilize the organoids with PBST (PBS with 1% Triton X-100) with mild shaking at RT for 4 h.
5. Block the organoids with blocking solution (2% bovine serum albumin (BSA) with 0.1% Triton X-100) at RT for overnight (or over 1 day).
   NOTE: Primary and secondary antibody concentrations depend on the particular antibody. For immunofluorescence of whole-mount samples, we use a higher concentration in comparison to sectioned samples. We incubate the primary antibody at 10 µg/mL which, in most cases, corresponds to 1:100 from the stock (Table 1). We incubate our custom-made secondary antibodies at 20-50 µg/mL.
6. Incubate the organoids with the primary antibodies diluted in washing solution (0.1% BSA with 0.1% Triton X-100) at 4 °C for 2 days with mild shaking.
7. Wash the organoids for 3 x 15 min in washing solution with mild shaking at RT.
8. Incubate the organoids with the secondary antibodies diluted in washing solution at 4 °C for 2 days with mild shaking.
9. Wash the organoids for 3 x 15 min in washing solution with mild shaking at RT.
   NOTE: After immunolabeling, fluorescent dyes of choice can be incubated. We normally use fluorescent nuclei markers, such as DAPI (1:1,000) and DRAQ5 (1:500), or phalloidin (1:100) for visualizing the actin filament network.
10. Incubate the organoids with the fluorescent dyes diluted in washing solution at RT for 1 h with mild shaking.
11. Wash the organoids for 3 x 15 min in washing solution with mild shaking at RT.

3. Optical clearing with FluoClear BABB

NOTE: 1-Propanol might compromise plastic containers, particularly for long-term exposure. To be safe, use glass containers.

1. Dehydration step
   1. Prepare 1-propanol solutions in ultrapure H₂O at the given percentages (15%, 30%, 45%, 60%, 75%, 90%) and then adjust to pH 9.5 with trimethylamine.
      CAUTION: Triethylamine is toxic if inhaled. Work under a fume hood. Solutions can be reused if stored properly (well-sealed container). Adjust the pH before usage.
   2. Incubate the samples with the increasing gradient of 1-propanol solutions (15%, 30%, 45%, 60%, 75%, 90%) for 2 h each at 30 °C with mild shaking.
   3. Incubate the samples with 100% 1-propanol overnight at 30 °C with mild shaking.
2. Clearing step
   CAUTION: BABB is a potent solvent mixture that is known for its ability to dissolve or significantly damage many plastics. To be safe, use glass containers with compatible lids and work inside a fume hood.
   1. Prepare the BABB by mixing Benzyl alcohol (BA) and Benzyl benzoate (BB) at a 1:2 ratio.
   2. Immerse the samples in BABB at RT overnight.
   3. Before imaging, refresh the BABB.
      NOTE: For ~1 mm diameter organoids, BABB induces a clearing effect that is evident in tens of minutes. Depending on the size and nature of the sample, the degree of clearing could be different and/or require longer exposure time to the clearing solution. We recommend acquiring images before and after BABB immersion to improve the clearing protocol.
   4. Store BABB cleared samples with BABB solution for the long term at 4 °C. Protect from light to preserve the fluorescence.

4. Whole-mount image acquisition

1. Using a glass pipette, position the organoid in a glass-bottom Petri dish with a drop of BABB. Make sure that the sample contacts the surface of the coverglass.
   CAUTION: Take care that the drop of BABB containing the sample does not come into contact with the plastic edges of the dish, or it will melt the plastic, compromising the sample and the quality of the imaging.
2. Using an inverted confocal laser-scanning microscope equipped with low and high magnification objectives, acquire Z-stack images to gain the full 3D insight of the sample with cellular detail.
   NOTE: For this study, a 10x numerical aperture (NA) 0.4 air objective and a 63x NA 1.4 objective were used, with a step size of 3.5 µm and 1 µm, respectively.

5. Image processing

1. Recalibrate the step size of the Z-stack acquisition considering the refractive index (RI) mismatch between the clearing solution and the immersion media.
   1. Calculate the recalibrated step size as follows:
      
      For a sample embedded in BABB, imaged with an air objective with a 3.5 µm step size, it will be as follows:
      
   2. Using ImageJ, change the voxel depth (click on Image | Properties...) for the recalculated step size.
2. Create image projections of the Z-stack.
   1. Using ImageJ, inspect the Z-stack in depth (click on Image | Stacks | Orthogonal views). XY, XZ, and YZ views of the retinal organoid will appear.
   2. Save the desired regions of interest (click on File | Save as...).
3. Create an animation of the 3D render of the Z-stack using processing software (see the Table of Materials).
   1. Select the channels to be visualized by enabling their checkboxes.
   2. Adjust the Z-planes to visualize the interior of the sample (enable Clipping in the checkbox).
   3. Record the animation using the Movie editor (enable Movie editor in the checkbox). Drag and drop, and zoom in and out directly using the mouse. Click Add to include a different clip for the animation. If adjusting the Z-planes, enable the Clip planes option (click on Settings... | check Clip planes box).

Results

Figure 1: Comparison of three clearing methods. Left: Brightfield images of (A) uncleared and (B-D) cleared samples. Right: Z-stack with xz and yz orthogonal views of 90 DIV retinal organoids stained with the nuclear marker DRAQ5 (yellow) acquired with a confocal laser-scanning microscope equipped with a 10x objective of air (RI = 1.00). (A) Uncleared retinal organoid (N = 3). PFA-fixed retinal organoids are not transparent enough for 3D visualization. (B) Fructose/Glycerol-cleared organoid (N = 3). (C) ECi-cleared organoid (N = 3). (D) FluoClear BABB-cleared organoid (N = 3). Scale bars = 100 µm. Abbreviations: BF = brightfield; CLSM = confocal laser-scanning microscope; RI = refractive index; PFA = paraformaldehyde; ECi = ethyl cinnamate; BABB = benzyl alcohol/benzyl benzoate; DIV = days in vitro. Please click here to view a larger version of this figure.

Comparison of optical clearing methods for immunofluorescence studies in neuroretinal organoids

Here, we present the results of three clearing methods to demonstrate the efficacy of FluoClear BABB for rendering retinal organoids transparent. In Figure 1, we observe retinal organoids cleared with Fructose/Glycerol (Figure 1B), ECi (Figure 1C), and FluoClear BABB (Figure 1D). The clearing methods tested were chosen because they all have been proven^10,11 to be compatible with immunofluorescence staining and preserve fluorescence from endogenous proteins (e.g., GFP present in GCaMPs).

Fructose/Glycerol¹² is a non-toxic, simple, immersion clearing protocol that does not require dehydration, significantly reducing the processing time compared to other methods. It has been reported to be efficient for transparentizing organoids derived from various tissues: airways, kidney, liver, and human breast cancer¹².

ECi (Ethyl Cinnamate)¹¹ is an FDA-approved food flavor and additive for cosmetics^13,14, is considered harmless, and has proven to be an excellent clearing reagent for mammalian tissues. It is known for its rapid clearing capability and minimal tissue distortion. It is classified as a hydrophobic clearing method, but in comparison to BABB-based solvents, it is non-toxic and preserves fluorescence from fluorescent proteins longer (for up to 14 days). As is common in most clearing protocols, ECi requires a prior dehydration step. For optimal results, this step is achieved by dipping the sample in solutions with progressively increasing alcohol concentrations. Because ECi is not harmful, once the sample is cleared it can be manipulated safely and stored in plastic containers. However, the dehydration steps are performed using 1-propanol, an organic solvent that must be manipulated under a fume hood and stored in glass containers.

FluoClear BABB¹⁰ is a variant form of a hydrophobic protocol based on benzyl alcohol/benzyl benzoate (BABB) as the refractive index-matching solution. It is known to be very toxic and corrosive, so it needs to be handled with care. FluoClear BABB has been described to be an improved version that preserves GFP and RFP better than previous versions and allows the use of either 1-propanol or tert-butanol as dehydration agents.

To assess the performance of the three clearing methods, we imaged Z-planes with a confocal microscope equipped with a 10x NA0.4 air objective, providing a large field of view (FoV) of 2.5 x 2.5 x 2.56 mm to capture the whole volume of the retinal organoid. We used retinal organoids at an intermediate stage of maturation (90 days in vitro (DIV)) that contain all retinal neurons (ganglion cells, bipolar cells, and photoreceptors), achieving the cellular density and compaction and thus being opaque. As explained in the introduction, the optical clearing is achieved when RI is matched all over the object to be imaged. Thus, the immersion media used for imaging will impact drastically the transparency achieved. Table 2 details the RI of immersion media and the clearing solutions tested.

	Solution	Refractive index
Immersion media	Air	1
	Water	1.33
	Glycerol	1.45
	Oil	1.51
Clearing solutions	Fructose/Glycerol	1.468
	Eci	1.558
	FluoClear BABB	1.56

Table 2: Refractive index (RI) of the immersion media most used for light microscopy and the optical clearing methodologies tested for clearing retinal organoids. Abbreviation: RI = refractive index.

Comparing the resulting confocal images (Figure 1, right panel), fructose/glycerol treatment resulted in the least improvement in transparency, limiting visualization of the retinal layers and preventing visualization of the organoid core. ECi achieved better transparency, allowing the visualization of the retinal layers, but still limited visualization of the retinal organoid core due to light scattering that hindered fluorescence detection from the inner parts of the sample. FluoClear BABB provided the greatest improvement in transparency, enabling clear visualization of both the core and cortex of the retinal organoid at higher resolution with similar fluorescence intensity.

It is worth mentioning that both ECi and FluoClear BABB protocols induced notable sample shrinkage, as illustrated by the corresponding brightfield images of each cleared retinal organoid (Figure 1, left panel). Sample shrinkage is due to the previous dehydration steps that remove water, and that is the reason why the Fructose/Glycerol protocol does not shrink the sample. This shrinkage, however, presents a unique opportunity for deeper exploration. Although it compresses the overall size of the organoid and potentially underestimates internal distances, it also increases compactness. This isotropic compactness, relatively uniform in all directions, allows us to leverage high magnification objectives with superior resolving power (due to their larger NA) that would otherwise only image the outermost cell layer due to their shorter working distances (WD). For instance, a 10x NA 0.4 air objective offers a WD of 2.53 mm, whereas a 63x NA 1.4 oil objective with significantly better resolution has a WD of only 0.14 mm. By utilizing BABB-cleared, more compact organoids, we can effectively overcome this limitation and achieve high-resolution imaging of deeper structures, such as the ganglion cell layer located within the organoid's interior (Figure 2).

Figure 2: 3D-reconstructions of a retinal organoid (200 DIV) cleared with FluoClear BABB imaged with a laser scanning-confocal microscope. The microscope was equipped with a (A) 10x NA 0.4 air objective and (B) a 63x NA 1.5 oil objective. The clearing and the imaging allow for visualization of the entire organoid and at the same time, image with fine detail some regions of interest. DRAQ5 (nuclei), opsin B+GR (cones), and TUJ1 (neurons). Abbreviations: DIV = days in vitro; BABB = benzyl alcohol/benzyl benzoate; NA = numerical aperture. Please click here to view a larger version of this figure.

In conclusion, our choice for clearing was FluoClear BABB as it offered superior visualization of the entire retinal organoid with high resolution and signal to noise ratio. Importantly, BABB-induced shrinkage, compacting the organoid isotropically. This compactness allowed us to leverage high magnification objectives for detailed exploration of deeper structures. Finally, BABB's compatibility with oil immersion objectives ensured minimal light scattering and high-quality imaging across magnifications.

Identification and spatiotemporal distribution of retinal cells within the organoids over maturation

Following the methodology here detailed, and employing whole-mount imaging on organoids immunolabeled with the specific antibodies detailed in Table 1, we studied how the neuroretina is shaped from organoids at different points of maturation, from early developmental stages (40 DIV) to more mature stages (up to 250 DIV), exhibiting a well-defined 3D structure. This approach allowed us to visualize the formation and organization of various retinal components described in this section.

Antibody	Specificity	Targets	Location in Retina	[   ]final (dilution)	Host	#RRDI
Anti-GFP	GFP	GFP	GCaMP6s-positive cells	10 µg/mL (1:1,000)	Ch	AB_300798
Chx10	Ventromedial hypothalamus homeobox 2 (Vsx2)	Transcription factor	Neuroblasts and bipolar cells	10 µg/mL (1:2)	Ms	AB_10842442
Opsin B	Opsin blue	Outer segment of blue-cones	Blue-cones	10 µg/mL (1:100)	Rb	AB_177457
Opsin RG	Opsin green and red	Outer segment of green- and red- cones	Green- and red- cones	10 µg/mL (1:100)	Ch	AB_11213279
RECOV	Recoverin	Calcium-binding protein	Photoreceptors	10 µg/mL (1:100)	Rb	AB_2253622
RHO (RET-P1)	Rhodopsin	Outer segment of rods	Rods	63 µg/mL (1:100)	Ms	AB_260838
TUJ1	Neuron-specific class III beta-tubulin	Beta-III tubulin isoform	Neurons	50 µg/mL (1:200)	Ms	AB_2315514

Table 1: Primary antibodies tested in retinal organoids from different stages of maturation. Abbreviations: Chicken = Ch; Mouse = Ms; Rabbit = Rb.

The immunolabeling with different markers allowed the identification of the cells conforming to this layered structure. TUJ1 a neuron-specific class III beta-tubulin (Table 1), typically used in sectioned samples to label retinal ganglion cells (RGCs)¹⁵ due to its abundance. TUJ1 is present in all retinal neurons¹⁶ and from early stages of neural differentiation¹⁷. TUJ1 expression was found in all maturation stages at the cortex of the organoid, allowing for the follow-up of the organization of retinal cells in layers. In Figure 3, we can observe that at early stages of development (40 DIV), TUJ1 is uniformly distributed in a single, thin and uniform layer. From 90 DIV, the apical region of the organoid becomes denser and more compact, as it populated with many more cells that accumulate in the apical region. At 170 DIV, retinal organoid acquired a 3D-stratified retinal structure with a discrete apical layer. At 200 DIV, we can see that there are other cell bodies extending long projections, which are much less numerous, found deeper inside the organoid. Finally, at 250 DIV, the structure in three nuclear layers is evident, corresponding to: (1) the outer nuclear layer (ONL) conformed by mature photoreceptors, (2) the inner nuclear layer (INL) mainly populated by bipolar cells and (3) the ganglion cell layer (GCL) formed by RGCs, as indicated in Figure 3 (bottom right).

Figure 3: Neuron organization upon retinal organoids maturation in a layer-stratified structure. Z-stack projections of confocal images of FluoClear BABB-cleared retinal organoids stained with TUJ1. At 40 DIV, there are no layered structures formed. From 170 DIV, the neuroretina stratifies into layers. The retinal nuclear layers are indicated: ONL, INL, and GCL (yellow arrowheads). Scale bars = 25 µm (left) and 10 µm (right). Abbreviations: BABB = benzyl alcohol/benzyl benzoate; DIV = days in vitro; ONL = outer nuclear layer, INL = inner nuclear layer; GCL = ganglion cell layer. Please click here to view a larger version of this figure.

With this methodology, we could also report the assembly of bundles of axons. As seen in the magnified regions (Figure 4), we observed bundles of fibers extending from the inner core of the organoid towards the periphery, projecting towards regions that are far apart (up to 1 mm). Interestingly, the formation of these fibers occurred across different stages of maturation. As the organoid grows, the fibers thicken, get more complex, and collect more cell projections (Figure 4 [250 DIV] and Supplemental Video S1). RGCs are the first cells to differentiate during retinal development, and thus, they can arrange their projections from very early stages of development. RGCs may initiate neurite extension even before establishing their final position within the organoid. Our findings align with previous reports¹⁸ that demonstrate the formation of an optic nerve-like structure in co-cultures of retinal and brain organoids.

Figure 4: Assembling of neuronal projections at various stages of maturation. The fibers extend from the inner planes of the retinal organoid towards the periphery. The formation of these fibers occurred across different stages of maturation, from 40 DIV to the oldest maturation point studied (250 DIV). Scale bars = 25 µm (top row) and 100 µm (bottom row). Abbreviations: DIV = days in vitro. Please click here to view a larger version of this figure.

At the apical site of retinal organoids, some photoreceptors mature into rods, which contain rhodopsin in the outer segments to detect monochromatic vision in dim light conditions, while others mature into cones, which express blue, green, and red opsins and are responsible for color and high-acuity vision in bright light conditions. Figure 5 (top) illustrates the population of cone photoreceptors within the retinal organoid that expressed blue, and/or green/red opsins (OPSIN B and GR, Table 1). These cells exhibit an elongated morphology with the different opsins distributed throughout the cell body, excluding the nucleus. Notably, the outer tip of these cells appears brighter, potentially resembling the outer segment of the photoreceptor containing the discs. To visualize both blue and green/red cones using a single channel for fluorescence detection, secondary antibodies against blue-opsin (OPSIN B, rabbit) and green/red opsin (OPSIN GR, chicken) were tagged with the same fluorophore, optimizing channel use and streamlining the imaging process while avoiding spectral overlap. A subset of photoreceptor cells in Figure 5 (bottom) expressed rhodopsin (RHO, Table 1 and Supplemental Video S2), indicating unequivocally their identity as rods. These rod cells share the previously described elongated morphology, with the nucleus positioned centrally and the opsin-rich cell body oriented towards the periphery of the organoid.

Figure 5: Retinal organoids from later stages of development. Top: mature cones, bottom: rods. Z-projection of confocal images of FluoClear BABB-cleared retinal organoids stained with TUJ1 (neurons), DRAQ5 (nuclei), opsin-B+GR (cones), and RHO (rods). Schematic drawings are shown for reference. Scale bars = 100 µm (left) and 10 µm (right). Abbreviations: DIV = days in vitro; BABB = benzyl alcohol/benzyl benzoate. Please click here to view a larger version of this figure.

The bipolar cells and their spatiotemporal distribution were identified by tracking the Chx10 expression (also known as Vsx2) over 250 DIV of maturation (Figure 6). Chx10 (Table 1) is a transcription factor critical for progenitor cell proliferation and bipolar cell determination in the developing retina^19,20. As a homeobox gene, it is expressed at early stages of development in neuroblasts, and in the mature retina, it is found exclusively in bipolar cells. Figure 6 demonstrates the distribution of Chx10-positive cells within the developing organoid. Initially, numerous Chx10-positive cells are observed extending from the neuroretinal periphery towards the center. As the organoid matures, these cells become less abundant and relocate to the INL, where bipolar cells reside.

Figure 6: Chx10 expression over 250 DIV of maturation. Z-projection confocal images of a FluoClear BABB-cleared retinal organoid stained with DRAQ5 (nuclei) and Chx10 (neuroblasts and mature bipolar cells). Scale bar = 100 µm. Abbreviations: DIV = days in vitro; BABB = benzyl alcohol/benzyl benzoate. Please click here to view a larger version of this figure.

Finally, we demonstrate that the endogenous GFP from GCaMP calcium markers is preserved after clearing with this methodology. In Figure 7, we can observe the neuroretina of 70 DIV retinal organoids infected with Ad5-CMV-GCaMP6s and immunolabeled with anti-GFP (Table 1) to enhance fluorescence after long-term fixation.

Figure 7: Endogenous GCaMP preserved in the FluoClear BABB clearing method. A 70 DIV retinal organoid infected with Ad5-CMV-GCaMP6s immunolabeled with anti-GFP (GCaMP-expressing cells), Chx10 (neuroblasts and early-stage bipolar cells), and DRAQ5 (nuclei). Scale bars = 100 µm (left) and 25 µm (right). Abbreviations: BABB = benzyl alcohol/benzyl benzoate; DIV = days in vitro. Please click here to view a larger version of this figure.

Supplemental Video S1: 3D-reconstruction of a retinal organoid presenting an assemblage of neuronal projections. Z-projection of confocal images of FluoClear BABB-cleared retinal organoids at 250 DIV stained with TUJ1 (neurons) represented in purple and DRAQ5 (nuclei) in yellow. Abbreviations: BABB = benzyl alcohol/benzyl benzoate; DIV = days in vitro. Please click here to download this video.

Supplemental Video S2: 3D-reconstruction of the rods aligned in the surface of a retinal organoid at late stages of development (200 DIV). Z-projection of confocal images of FluoClear BABB-cleared retinal organoids at 200 DIV stained with RECOV (photoreceptors) represented in green and RHO (rods) in purple. Abbreviations: DIV = days in vitro; BABB = benzyl alcohol/benzyl benzoate. Please click here to download this video.

Discussion

The protocol described here facilitates the study of the spatiotemporal distribution of the cells forming the retinal organoid at different stages of maturation, preserving its natural 3D architecture, a critical aspect for understanding these models in the context of future personalized medicine. It provides a detailed view of the retinal architecture at the level of distinct layers and individual cell types, enabling visualization of subcellular structures while preserving the global 3D context.

The alternative to whole-mount imaging is postprocessing the sample for performing imaging on conventional sections (paraffined or cryopreserved). While sectioning provides valuable information about the localization of proteins and structures, this data is restricted to a single plane, limiting the understanding of spatial cellular connections. As a result, features such as the proportion of cones and rods or the formation of axonal bundles in specific regions may not be accurately represented, complicating the interpretation in three dimensions. Retinal organoids are highly heterogeneous, with different regions potentially containing different cell types and/or exhibiting varying degrees of complexity and differentiation. Whole-mount imaging solves this problem as it allows analyzing heterogeneity gradients globally without relying on the selection of a subset of sections that might not represent the entirety of the sample.

Light-sheet fluorescence microscopy (LSFM)^21,22,23 could be used as an alternative to laser-scanning confocal microscopy. In LSFM, the specimen is illuminated with a thin sheet of light of a few microns thick, and all the fluorescence generated from the illuminated optical section is detected using a camera. LSFM is a widefield-based detection technique that represents a highly efficient strategy for fast visualization of large volumes (up to centimeter-scale samples) with minimal photobleaching and phototoxicity effects. However, organoids may exhibit significant spherical aberration because light passing through the curved surfaces is refracted differently depending on the angle of incidence. In these experiments, this distortion produced a loss of focus, leading to blurring and compromising the resolution of deeper structures. For imaging our retinal organoids (approximately 1 mm in diameter), laser-scanning confocal microscopy proved to be the most effective approach.

This protocol is tailored to our experimental requirements. It can be used as a starting guide for researchers venturing into whole-mount imaging of optically cleared retinal organoids and should be adapted if those differ. In this section, the critical steps, modifications, and limitations of the methodology are described in detail.

A critical step is the decision on the clearing protocol. Many criteria need to be considered. First, consider the typology of the sample. Clearing bones is not the same as clearing brain tissue, just as clearing retinal organoids differs from clearing gastric organoids, as each requires a distinct approach. To visualize cleared organoids of a different type, it is essential to review the literature to identify protocols that have been optimized specifically for the sample of interest. Fructose/glycerol is a fast and non-toxic clearing method that has been proven to be useful for clearing organoids derived from different tissues (human airway, colon, kidney, liver, and breast tumor organoids, and mouse mammary gland organoids)¹² but was not suitable for making human retinal organoids transparent. To the best of our knowledge, no study reports fructose/glycerol application to neural tissue/organoid. Therefore, we conclude that this clearing method is not appropriate for clarifying neural tissues. Further investigation is warranted to elucidate the underlying reasons for this incompatibility.

Another critical factor to consider is the FoV required, as it directly impacts the imaging strategy and the choice of optical components. Large FoVs are typically achieved with low magnification objectives, which are often designed for use without immersion media (e.g., air, RI = 1.00). These objectives experience a larger RI mismatch when imaging BABB-cleared samples (RI = 1.56), leading to shifts in depth focusing that must be corrected during post-processing. In contrast, oil immersion objectives (RI = 1.51) present less RI mismatch with clearing solutions like BABB (RI = 1.56) and, therefore, do not require recalibration.

Simple immersion methods usually do not achieve high transparency and are not suitable for acquiring high-quality images of whole-mount samples. Tissue clearing protocols influence effective resolution via swelling or shrinkage of the samples. Depending on the goal of the experiment, either swelling or shrinkage can be beneficial. Protocols that expand the sample can enhance the resolution of fine details in the same optical system incapable of showing the same details before. Shrinkage can be desired to visualize large samples that otherwise would not fit into the sample chamber or adjust to the objective working distances. However, because shrinkage compresses the sample and may distort internal distances, this technique is not suitable for dimensional assessment or precise measurement of intercellular distances.

Another important factor to be considered is compatibility with fluorescent proteins. Mostly organic solvent-based protocols quench fluorescent proteins as they require high temperatures or degrade the fluorescent proteins while embedding with RI-matching solutions. Fortunately, most of the methods are compatible with immunolabeling techniques. However, the compatibility of each epitope should be confirmed experimentally.

Moreover, penetration of large IgG molecules through tissue can be obstructed, decreasing staining efficiency and limiting visualization of the targeted structure to the sample surface. Staining whole-mount specimens with antibodies might take several days (or even weeks), depending on the size and the nature of the sample, substantially increasing the duration of the protocol.

Lastly, compatibility with the optical instrument is mandatory. Most organic solvent-based methods (except ECi) are very corrosive and not chemically compatible with commercial sample chambers and immersion objectives. Furthermore, some clearing solutions have an RI that does not match the range required by the available objectives (usually water).

There are some steps in the protocol that need special attention. FluoClear BABB, as a hydrophobic clearing method, requires a previous dehydration step. The pH plays an important role in preserving the fluorescence of endogenous proteins such as GFP. Then, it is crucial to ensure that the alcohol solutions have the pH properly adjusted. Dehydration is key for complete clearing. Optimal results are achieved by incubating the samples with an increasing gradient of alcohol solutions.

The size of the sample plays an important role in allowing antibodies to reach their specific targets. Antibodies might fail to penetrate homogeneously and be retained in the outside of your sample. If this occurs, consider switching to another type of detergent (Tween-20 or Saponin), increasing the concentration and incubation time of the detergent and the antibodies. If these adjustments are insufficient, alternative strategies can improve antibody permeabilization. For whole-mount immunolabelling of big specimens (centimeter-scale), researchers have found that the SDS can facilitate deeper antibody penetration²⁴. SDS induces a partial loss of the tertiary structure of antibodies, effectively linearizing the protein and allowing it to penetrate deeper into the tissue. Then, washing out the SDS restores the antibodies' antigen-binding properties, enabling more uniform staining of thicker tissues and reducing the commonly observed gradient of stronger labeling on the surface of the sample.

Preparing your own fluorescent secondary antibodies offers several advantages, as it provides greater flexibility in immunolabeling design, allowing for the selection of the specific fluorophore-secondary antibody combination, increasing the number of colors that can be used in the same sample. Additionally, the fluorophore-to-antibody ratio can be tuned, so they are generally brighter. However, some aspects must be considered. Custom-made secondary antibodies are generally less stable than the commercially available ones and should be used fresh. Most retain their ligand affinity for 3-6 months at 4 °C, but stability needs to be experimentally determined, as some combinations could degrade more quickly. Moreover, some antibody-fluorophore pairings can alter the affinity of the antibody for its epitope, resulting in partial or complete nonspecific labeling. To ensure reliability, it is advisable to first test custom-made secondary antibodies using controls, such as cells in culture, before applying them to experimental specimens.

This optimized protocol integrates optical clearing with immunolabeling to facilitate confocal imaging of whole-mount retinal organoids. It preserves structural integrity and enables the identification of key retinal cell types within the organoid. Thus, the protocol provides researchers with a powerful tool to investigate retinal organoid maturation and spatial organization.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
10x NA0.40 DRY objective	Leica	HC PL APO CS2	-
63x NA1.40 OIL objective	Leica	HC PL APO CS2	-
anti-GFP antibody	Abcam	ab13970	-
Benzyl alcohol (BA)	Sigma Aldrich	13162	-
Benzyl benzoate (BB)	Sigma Aldrich	W213802	-
bottom-glass petri dish	MatTek	P35G-1.5-14-C	-
Bovine Serum Albumine (BSA)	Sigma Aldrich	A7030	-
Chx10 antibody	Sta Cruz Biotech	sc-365519	-
Cytiva Illustra Nap-5 gel filtration columns, Sephadex G-25	Sigma Aldrich	17-0853-02	-
DAPI	Sigma Aldrich	D9542	-
DMSO	Sigma Aldrich	D8418	-
DRAQ5	Thermofisher	62251	-
Dye (carboxylic acid, succinimidyl ester)	Invitrogen or Abberior	-	-
ImageJ/Fiji	NIH	v1.50i	-
Inverted confocal laser-scanning microscope	Leica microsystems	TCS SP8 STED 3X-FALCON	-
LAS X 3D processing software	Leica microsystems	-	-
mQ H2O	-	-	Ultrapure water from the Synergy water purification system
NaHCO3	Acros Organics	424270010	-
Non-labelled secondary IgG	Jackson Immunoresearch	-	-
Opsin blue antibody	Millipore	AB5407	-
Opsin RG antibody	Millipore	AB5745	-
Paraformaldehyde (PFA) 4%	Thermo scientific	J199943-K2	-
PBS Solution	Gibco	18912-014	-
Phalloidin labelling probes	Thermofisher	-	-
Phosphate-Buffered Saline (PBS)	Gibco	10010023	-
RECOV antibody	Millipore	AB5585-I	-
RHO (RET-P1) antibody	Sigma Aldrich	O4886	-
Sodium citrate tribasic dihydrate	Sigma Aldrich	C8532	-
Triton X-100	Sigma Aldrich	T8787	-
TUJ1 antibody	Covance	MMS-435P	-