Broad applicability of a streamlined ethyl cinnamate-based clearing procedure

Methods to optically clear tissues using refractive index matching have been transformative for imaging large, three-dimensional tissues. Tissue clearing has allowed long-distance mapping of axonal projections and reconstruction of entire embryos (Belle et al., 2014, 2017; Economo et al., 2016). Despite the importance of such reconstructions, the daily use of clearing agents to quantify cell populations in whole-mount preparations has seen limited use in fields such as developmental biology, organoid research or regeneration biology due to cumbersome aspects associated with each method. Aqueous-based clearing methods such as Clarity and SeeDB require long incubation times of days to weeks to complete, depending on tissue size (see Table 1). This becomes prohibitive for rapidly screening different experimental conditions. Organic solvent-based methods bypass long incubation times due to extraction of lipids and other organic material in the sample, yet usually rely on toxic components and show limited clearing or reduced preservation of fluorescence protein signal (see Table 1). We aimed to overcome these shortcomings to produce a rapid, yet effective and non-toxic, clearing protocol that preserves fluorescent protein/antibody signal. Such a method would, for example, allow the use of whole-mount imaging for genetic or chemical screenings in fluorescent protein-expressing transgenic animals or organoids, as well as whole-mount immunolabelling. Here, we describe the combination of sample dehydration in 1-propanol_pH9 followed by refractive index matching with the organic compound ethyl cinnamate (ethyl 3-phenyl-2-propenoate) as an ideal protocol for rapid, non-toxic sample preparation that preserves fluorescent proteins and antibody-conjugated fluorophores. We have named this method 2Eci (2nd generation Ethyl cinnamate based clearing method) and apply it to cerebral organoid characterization, whole-animal and whole-appendage imaging. Furthermore, an extensive protocol, including species-dependent alterations to the protocol, are provided in the supplementary Materials and Methods.

In our aim to develop an easy to use and broadly applicable method, we focused on organic-chemical based protocols as they are inherently faster compared with passive aqueous methods. Methods based on organic solvents often use toxic solutions for refractive index matching (for example BABB, a mixture of benzyl alcohol and benzyl benzoate) and result in the quenching of endogenous fluorescence. We assessed clearing efficiency and preservation of GFP fluorescence of various dehydrating agents and refractive index matching solutions using cerebral organoids that were sparsely labelled with a population of CAG:GFP-expressing cells. Human cerebral organoids are a powerful 3D culture system that reconstitutes the early development of discrete brain regions (Lancaster et al., 2013). These organoids provide a reductionist approach to understand aspects of human brain development in-vitro (Bagley et al., 2017). Uncleared cerebral organoids are highly turbid (Fig. 1A). While FluoClearBABB (Schwarz et al., 2015) clearing results in higher transparency (Fig. 1B), ethanol dehydration followed by refractive index matching using ethyl cinnamate (Eci) as previously described (Klingberg et al., 2017) efficiently cleared cerebral tissue (Fig. 1C). However, GFP fluorescence intensity, while still present, was significantly reduced, resulting in the loss of ability to detect detailed neuronal morphology such as dendrites and axons (Fig. 1F). Based on reports that dehydration using alcohols adjusted to alkaline pH levels can preserve GFP fluorescence (Schwarz et al., 2015), we assessed clearing efficiency and GFP preservation in several alcohols adjusted to pH 9 (Fig. 1D-H). We found that dehydration using methanol_ph9 results in a complete loss of specific fluorescence [1% of uncleared signal, corresponding to background autofluorescence levels, 355.7±39.39 AU (arbitrary units±s.d.)], and ethanol_ph9 dehydration resulted in the retention of a specific fluorescent signal comprising 5% (2024±96.66 AU) of the intensity observed in uncleared organoids (37991±3314 AU). For both ethanol and methanol dehydration, morphological details of GFP⁺ cells could no longer be observed (Fig. 1D-F,I). In contrast, cerebral organoids dehydrated with either 4-butanol_ph9 or 1-propanol_ph9 displayed a higher intensity of GFP signal at ∼75% (29167±5569 AU) or 50% (15542±4184 AU), respectively, of uncleared signal (Fig. 1G-H,I). To assess clearing efficiency, we examined imaging depth independently of fluorescence preservation by recording auto-fluorescence levels at 488 nm wavelength in unlabelled (GFP⁻) organoids. Methanol_ph9, ethanol_ph9 and 1-propanol_ph9 allow for autofluorescence recordings through the whole organoid (>1400 µm), while 4-butanol_ph9-mediated clearing yielded only 500 µm penetration into the organoid (Fig. 1J). We found that tissue autofluorescence levels are comparable across dehydrating agents and increased compared with unfixed control samples (Fig. S1A). We conclude that the combination of 1-propanol_ph9-mediated dehydration followed by ethyl cinnamate-mediated refractive index matching allows for efficient clearing, while preserving sufficient levels of GFP for detection. The protocol can be completed in as little as 25 h (Fig. 1K-M). This results in a method that is straightforward and broadly applicable as it only consists of three main steps: fixation, dehydration using serial concentrations of 1-propanol/PBS_ph9; and refractive index matching using ethyl cinnamate. We call this method 2Eci (2nd generation ethyl cinnamate mediated clearing).

To further validate 2Eci as an efficient method for clearing cerebral organoids, we used confocal microscopy to record z-stacks through 80-day-old cerebral organoids sparsely labelled with a population of CAG:GFP⁺-expressing cells. We found that 2Eci clearing allows for recordings throughout organoids of ∼1400 µm thickness (Fig. 2A, Movie 1), while the recording depth of uncleared cerebral organoids was approximately 100 µm into the tissue (Fig. 2B). Not only is imaging through an entire cerebral organoid possible, but detailed morphological structures can also be observed. In 80-day-old cleared cerebral organoids, neural rosettes were readily observable in toto, with neurons showing elaborate morphology engulfing neuronal rosettes (Fig. 2C). As with all dehydration-based methods, the organoids underwent tissue shrinkage, which was in the range of 30-40% reduction in diameter using 2Eci (Ertürk et al., 2012) (Fig. S1B). Despite this overall reduction in size, highly detailed morphological structures such as putative dendrites with dendritic spines and putative axons with boutons could still be observed (Fig. 2D-F).

To examine whether antibody labelling approaches are compatible with 2Eci clearing, we performed staining with nanobodies, which, owing to their small size, easily penetrate the tissue. We found that regular immunohistochemical protocols with extended wash and incubation times were sufficient to label GFP⁺ cells throughout using αGFP647 nanobodies (Fig. S1C). However, most tissue characterizations will still include conventional antibody staining approaches using primary and secondary antibodies, which are considerably bigger than nanobodies. To allow for complete tissue penetration of these antibodies, we used a modified staining protocol using elevated concentrations of Triton X-100 to 2%, as well as 20% DMSO (Zukor et al., 2010). We found consistent colocalization of GFP⁺ cells and the αGFP_568nm antibody signal throughout the entire organoid (Fig. 2G,G′, Fig. S1D,E). Taken together, these data show that 2Eci is a viable method for clearing cerebral organoids and allows for detection of GFP signal within detailed morphological structures throughout the organoid.

Not all cell types are readily labelled using transgenic approaches, and antibody labelling is essential for most tissue characterizations. We therefore assessed the broader compatibility of whole-mount antibody staining in the context of cerebral organoids. We successfully validated 14 antibodies to stain whole-mount cerebral organoids (overview provided in Table S1). These antibodies cover the development from neuronal stem cells to mature neurons. Antibody staining was performed after fixation of the cerebral organoids and prior to dehydration and refractive index matching. We found apkζ and n-Cadherin labelled the apical lumen of SOX1-, SOX2- and nestin-positive neural rosettes, which were also actively proliferating (KI67⁺) (Fig. 2H,I, Fig. S2B,D-I). We also found rosettes that express FOXG1, a known forebrain marker, colocalizing with PAX6, indicating the presence of cortical progenitors (Fig. 2J, Fig. S2J,K). From these progenitor cells, early born cortical neurons (CTIP2⁺) and late born cortical neurons (SATB2⁺) arise (Fig. 2K, Fig. S2L,M). Postmitotic neurons expressing DCX (Fig. S2N) and mature neurons expressing MAP2 (Fig. S2O) were also detectable. An extensive overview of validated primary and secondary antibodies for cerebral organoids is provided in Tables S1 and S2.

Taken together, these data show that 2Eci is a viable method for clearing intact cerebral organoids. We have furthermore validated that both endogenous fluorescence-, as well as antibody-, labelled structures can be visualized. We conclude that 2Eci provides a method to survey the overall 3D structure and tissue identity in intact cerebral organoids, providing significant advantages over previous 2D antibody characterization approaches.

Having established the conditions of 2Eci clearing and its effectiveness in clearing cerebral organoids, we set out to explore the effectiveness of 2Eci in clearing larger organisms. Using Prrx1:ER-Cre-ER; CAGGs:LP-GFP-LP-Cherry double transgenic axolotl, we indelibly labelled connective tissue throughout the limb (Logan et al., 2002; Gerber et al., 2018). These limbs can be efficiently cleared using 2Eci, resulting in the preservation of both GFP and Cherry signal (Fig. 3D-F), which can be observed throughout the entire depth of the limb (Fig. 3F′). The incomplete expression of Cherry throughout the bone is likely due to incomplete recombination. To further explore fluorophore survival upon clearing, we investigated the potential of 2Eci in clearing Brainbow2.1R-expressing tissue. Although the recent development of antigen specific fluorophores in Brainbow3 allows for antibody labelling, previous constructs in both the original Brainbow and Brainbow2 series do not have antigen specificity (Cai et al., 2013). To test fluorescent protein preservation, Axolotl Brainbow2.1R (Currie et al., 2016) was crossed to CAGGs:ER-Cre-ER-T2A-EGFP-nuc (Khattak et al., 2013) and recombination was induced followed by 2Eci clearing. We found that all Brainbow fluorophores are preserved and remain spectrally distinct (Fig. S1D), suggesting that 2Eci can also be used as a general tool for already existing Brainbow and Brainbow2 lines without having to rely on antibody labelling. A complete overview of tested fluorophores is provided in Table S3.

In addition to the preservation of endogenous fluorescent proteins, we assessed the compatibility of 2Eci with antibody labelling approaches in axolotl. For this purpose we used limbs from our Col1a2:ER-Cre-ER; CAGGs:LP-GFP-LP-mCherry reporter animals and combined this with antibody staining for Prrx1. Prrx1 is a broad marker of limb connective tissue, while Col1a2 is a known marker of a subset of Prrx1 cells, namely dermal fibroblasts and the skeletal lineage. Examining both markers concurrently should reveal the heterogeneous nature of connective tissue populations. Antibody staining of the axolotl limbs was carried out prior to dehydration and refractive index matching. Limbs were efficiently cleared and this revealed that Col1a2-expressing cells localize to the skeletal, tendon and dermis. Antibody staining for PRRX1⁺ cells is observed in dermis, peri-skeleton cells and muscle interstitium, with low intensity signal in the skeletal lineage (Fig. 4). This experiment confirms and highlights the heterogeneous nature of connective tissue. Taken together, these results show that 2Eci is compatible with not only small samples such as cerebral organoids but also with larger tissues such as axolotl limbs. More importantly, it further highlights the utility of this rapid clearing protocol combining fluorescent protein with immunofluorescence visualization to analyse the complex 3D morphologies of heterogeneous tissues.

Pigmented tissues are traditionally not removed during tissue clearing procedures. We assessed the capacity for 2Eci to clear adult zebrafish as they provide a unique challenge due to their size, scales and three types of pigment (xanthophores, melanophores and iridophores). We found that, although efficient clearing is achieved simply by increasing the time for dehydration and clearing steps (Fig. S2A-B), the silvery iridophores and black melanophores persist. Surprisingly, the yellow xanthophores are efficiently cleared. The reason underlying this difference in pigment clearing is something we currently do not understand, but can be overcome using pigmentation mutants such as Nacre (Lister et al., 1999) (Fig. S2C-D).

Xenopus laevis are also heavily pigmented, which in their specific case can be overcome by mechanical removal of the skin. After removal of the skin, the clearing of CAG-GFP Xenopus hindlimbs allows for the full 3D reconstruction of the limb (Fig. S4A, Movie 2). The strong expression of the CAG-GFP transgene in the musculature allows for the full 3D reconstruction of the limb associated musculature (Fig. S4B,C, Movie 2).

In cases where both the mechanical removal of pigmentation is impossible and pigmentation mutants are unavailable, an alternative approach is required. Axolotl embryos are heavily pigmented with melanophores. Mechanical removal of pigmentation is not an option if the overall tissue architecture needs to be preserved. To overcome this problem, we sought to assess the compatibility of chemical depigmentation approaches with the 2Eci clearing method. A combination of 3% H₂0₂ and 0.5% KOH was previously shown to efficiently clear retinal pigmented epithelium (Perez Saturnino et al., 2018). We found that this chemical depigmentation method efficiently removes pigmentation in intact axolotl embryos in as little as 20 min (Fig. 3H,I), but in doing so also destroys fluorescent protein emission. We therefore combined depigmentation with antibody staining and clearing. Indeed, when applied to double transgenic CAGGs:GFP; CAGGs:CherryNuc embryos, chemical depigmentation, antibody staining and subsequent 2Eci clearing allows for antibody labelling throughout the organism (Fig. 3J-P). This approach will allow for detailed assessment of embryonic development in a host of heavily pigmented species, without relying on tissue sectioning.

In addition to antibody labelling, whole-mount in situ hybridization (WISH) is often performed in aquatic species such as Xenopus and zebrafish. After WISH, these embryos are commonly dehydrated using methanol and subsequently cleared using BABB (Saint-Jeannet, 2017). BABB is toxic and a strong organic solvent, and has a refractive index comparable with ethyl cinnamate. We found ethyl cinnamate to provide an efficient and non-toxic alternative in clearing of axolotl embryos after WISH (Fig. S1E). We thus expect embryos from other species that are commonly cleared with BABB after WISH to also be efficiently cleared using ethyl cinnamate.

Despite being a long-established model organism, Drosophila melanogaster has remained difficult to clear. Previous efforts in clearing Drosophila were either dependent on BABB (McGurk et al., 2007) or on a highly modified CUBIC approach (Pende et al., 2018). These methods are either laborious or dependent on antibody labelling. To test whether adult Drosophila melanogaster can be cleared using 2Eci and in doing so overcome these issues, we applied 2Eci clearing to Kr-GFP transgenic larvae and adult flies, which express GFP dominantly in the fat body. Prior to dehydration, we performed bleaching and a chymotrypsin and chitinase digest to degrade parts of the exoskeleton and increase permeabilization (Manning and Doe, 2016). We found that 2Eci efficiently cleared both larvae and adult Drosophila using 6 h dehydration steps (Fig. 4A,B), while preserving GFP expression (Fig. 4C,D). For adult Drosophila, autofluorescence at 488 nm and 568 nm excitation was comparable. We therefore used the 568 nm channel to perform morphological reconstruction of Drosophila and its inner structures, and also subtracted the auto-fluorescent background of the 488 nm recording to visualize a GFP-specific signal. We find that it is possible to generate whole-fly reconstructions while retaining the cellular resolution of Kr-GFP fat body cells (Fig. 4D and Movie 3).

We additionally explored the compatibility of this Drosophila protocol with antibody staining. For this purpose, we used newly generated SBR-GFP transgenic animals that ubiquitously express a nuclear-tagged GFP, which was too weak to be observed in cleared flies (data not shown). We find that antibody staining for GFP results in specific nuclear staining, and, after clearing, allows for detailed morphological reconstruction of the adult Drosophila internal organs using conventional spinning disk microscopes (Fig. 4E). Using this label, we could observe, for example, the structure of the anterior midgut with diploid and polyploid cell populations, fat body cells and egg chamber development (Fig. 4F-I). In line with other organisms and tissues, we find that antibody staining of adult Drosophila is compatible with 2Eci clearing.

Ethyl cinnamate is a non-toxic compound and, as opposed to BABB, can be used on microscopes in multi-user microscope facilities. However, there is a current lack of commercially available lenses optimized for ethyl cinnamate both with regard to refractive index matching and immersion compatibility. Such lenses would ideally be applied in the context of high resolution light sheet microscopy. We opted instead to optimize deployment on commonly available inverted imaging platforms using low magnification (≤20×) low NA (<0.8) air objectives with long working distances. Such lenses provide a large field of view while reducing the effect of refractive index mismatching. This approach also prevents the lenses from coming into direct contact with Eci. While Eci is a non-toxic compound it is still a mild organic solvent and might attack insulation rings of objectives or imaging chambers. Thus, we set out to identify suitable commercially available mounting chambers for inverted imaging (see Table S4). From all the dishes that were tested, we identified the Ibidy μDish 35 mm Glass Bottom (81158) as compatible with ethyl cinnamate, being resistant to ethyl cinnamate for at least several months. However, we recommend storing samples in air-tight containers such as polypropylene Falcon tubes (polystyrene tubes are incompatible), as prolonged exposure to the air can result oxidation and mild declearing over time (data not shown). For high-throughput processing and imaging, we have identified the Ibidi μ-plates 96 Well Black uncoated (89621) as a compatible sample holder, which can hold several organoids or Drosophila per well and can be sealed using commonly available adhesive PCR plate seals. Image acquisition of samples in Eci, using air objectives results in a refractive index mismatch commonly referred to as a ‘fishtank effect’. This effect can be corrected for using the following formula where Z_uncorrected corresponds to the z distance set at the microscope, RI₁ to the refractive index of the immersion media of the objective [in the conditions of this study, the RI of air is 1.00 and RI₂ corresponds to the RI at the sample (in this study, ethyl cinnamate – RI=1.56)]. Samples that are difficult to orientate can be fixated in place by mounting samples in a 1% phytagel block prior to dehydration. After clearing, phytagel blocks can be mounted directly into the imaging dish using a small amount of super glue. Taken together, this mounting and imaging approach should provide a reliable method that can be easily deployed in high-throughput approaches using multi-user microscope facilities.

We conclude that 2Eci is a clearing method that makes tissue clearing available to a broader audience, as it combines the rapid and broad applicability of dehydration-based methods, and the non-toxic nature and preservation of fluorescent proteins of aqueous clearing methods. We provide detailed protocols for the clearing of a variety of tissue, including cerebral organoids, and a range of embryonic and adult species, including Drosophila, axolotl, zebrafish and Xenopus (see supplementary Materials and Methods). This method preserves fluorescent proteins while being compatible with antibody staining and in situ hybridization. 2Eci clearing was shown to be effective in a large range of species and tissues, either matching or surpassing the efficacy of BABB. Further to its broad applicability, its simple and robust approach enables its broad scale adoption in the biological sciences.

Organoids were grown as described previously (Lancaster et al., 2013) using feeder-free H9 human embryonic stem cells (hES) from WiCell with a verified normal karyotype and contamination free. Cells were cultured in mTESR1 (Cat.85850) and initial EB formation for the first 5 days was performed in mTESR1. For sparse labelling of organoids, 1-3% of the initial 9000 cells were replaced with feeder-free H9 with a CAG-GFP insertion into AAVS1 (Bagley et al., 2017) for initial EB formation.

Axolotl were maintained on a 12 h light/12 h dark cycle at 18-20°C (Khattak et al., 2014). Prior to amputation or tissue collection, animals were anaesthetized in 0.03% benzocaine and injected subcutaneously with 38 mg/kg buprenorphine. Benzocaine (0.1%) was used for terminal experiments and euthanasia of axolotl. The work was performed under an approved license from the Magistrat der Stadt Wien (GZ: 9418/2017/12). Previously published lines that were used include CAGGs:LP-EGFP-LP-Cherry (Khattak et al., 2013), CAGGs:EGFP (Sobkow et al., 2006) and CAGGs:CherryNuc (Kragl et al., 2009).

TLAB and Nacre (Lister et al., 1999) zebrafish were maintained on a 14 h light/10 h dark cycle at 28°C according to standard procedures (Westerfield, 1995). TLAB fish, generated by crossing zebrafish AB and the natural variant TL (Tübingen/Tüpfel Longfin) stocks, served as wild-type zebrafish. All fish experiments were conducted according to Austrian and European guidelines for animal research and approved by local Austrian authorities (animal protocol BMGF-76110/0017-II/B/16c/2017).

Xenopus were maintained on a 12 h light/12 h dark cycle at 20°C. Prior to amputation or tissue collection, animals were anaesthetized in 0.01% MS222 and injected subcutaneously with 38 mg/kg buprenorphine. The work is performed under an approved license from the Magistrat der Stadt Wien (GZ: 852533/2016/20).

The following Drosophila melanogaster stock was used in this study:

w; L² Pin¹/CyO, P{GAL4-Kr.C}DC3, P{UAS-GFP.S65T}DC7 (Bloomington 5194), a larval fat-body-expressing GFP reporter. The newly generated transgenic w; sbr_GFP_Precission_V5_3xFLAG [attP2]/TM3,Sb; was also used in this study.

In brief, the SBR gene was cloned from the plasmid Pacman CH322-120I06 into an AttB2 destination vector containing all the tags. This vector was then injected into recombinase-expressing embryos and confirmed for integration into AttP2 on the third chromosome. Wandering third instar larvae and young virgin females still displaying a larval fat body were used as material for clearing.

Organoids were fixed in 4% PFA for 4 h at room temperature or at 4°C overnight and transferred sequentially into a dehydration series of 30%, 50%, 70% and 2×99.7% 1-propanol (99%; Sigma, W292818; 99.7%, Sigma 279544):1×PBS solution pH adjusted to 9.0-9.5 using triethylamine (Sigma, T0886). 1-Propanol at 99.7% contained sufficient free water-derived hydrogen ions for the pH to be set using standard laboratory pH meters (PHM220 Lab pH meter; Radiometer Analytical). For comparison of dehydration agents, 1-propanol was exchanged with 4-butanol (Sigma, 471712), ethanol (Sigma, 34852-M) or methanol (Sigma, 322415), respectively. Dehydration was performed at 4°C on a gyratory rocker for at least 4 h per dehydration step in 50 ml polypropylene Falcon tubes containing 45 ml dehydration agent. Dehydration time is strongly dependent on tissue size, and while 4 h was sufficient for regular sized organoids, we recommend to extend dehydration times to at least 6 h per step for bigger organoids (4-5 mm range). After dehydration, organoids were transferred in a 50 ml tube with at least 45 ml ECi (≥98%, Sigma, W243000; 99%, Sigma, 112372) and incubated on a gyratory rocker at room temperature for at least 1 h before recording. Samples were stored in light-protected and air-sealed falcon tubes filled with 45 ml Eci. Recordings were acquired over the following days after clearing. Full details can be found in the supplementary Materials and Methods.

Immunohistochemistry was optionally performed after 4% formaldehyde fixation. In brief, organoids were rinsed and then washed in PBS for 10 min to remove residual formaldehyde. For blocking and permeabilization, organoids were shaken at room temperature in 10 ml PBS-TxDBN solution (sterile filtered solution of 10% 10×PBS, 2% TX100, 20% DMSO, 5% BSA, 0.05% NaN (all percent volume with double distilled H₂O was added to reach the final volume) for 2 days. DMSO must be added slowly and after the BSA has dissolved, otherwise protein will precipitate out. The solution is stable at 4°C for at least 1 month (PBS-TxDBN ingredients were modified from Zukor et al., 2010). Primary antibody details are in Table S1.

Organoids were stained in 2 ml Eppendorf tubes in 400 µl PBS-TxDBN at room temperature on a tube rotator for 5 days and subsequently rinsed and then washed three times with PBS-TxDBN for 2 days. Staining for secondary antibodies was performed in the same way as the primary antibody staining step (see supplementary Materials and Methods). Secondary antibody details are in Table S2. Subsequently, organoids were washed as for primary antibodies. Organoids were then fixed with 4% formaldehyde in PBS for 4 h at room temperature or at 4°C overnight. We used ten times the amount of both primary and secondary antibodies in comparison with previously used concentrations in 2D immunohistochemistry. However, we recommend the testing of any antibody in a dilution series to determine the optimal concentration. Owing to the small size of nanobodies, staining can be performed significantly faster and in milder conditions. For nanobody immunohistochemistry, we permeabilized/blotted with 1×PBS, 0.03% Tx100, 5% BSA and 0.05% NaN. Stainings were performed in 1×PBS, 0.01% Tx100, 5% BSA and 0.05% NaN, and were only performed for 2 days. Three washes were performed in PBS-T (1× PBS, 0.01% TX100) for 1 day and organoids were fixed in 4% formaldehyde for 4 h at room temperature or at 4°C overnight. Full details can be found in the supplementary Materials and Methods.

Drosophila and Drosophila larvae were used for clearing experiments. Adult Drosophila were injected with 69 nl 9% formaldehyde in PBS solution and washed after 10 min in PBS. Flies and larvae were then transferred into 30% ethanol to remove the hydrophobic fatty lipid layer from the exoskeleton for 5-10 min. Subsequently, Drosophila were briefly bleached using DanKlorix, a commercially available bleach solution (Colgate-Palmolive). To increase transparency of the exoskeleton, a CCD (chitinase-chymotrypsin-DMSO buffer) digest was performed (Manning and Doe, 2016). Drosophila were then subsequently fixed in 4% PFA in PBS for 4 h at room temperature or at 4°C overnight. Antibody staining for anti-GFP was performed identical to the organoid immunohistochemistry procedure for primary and secondary antibodies (see supplementary Materials and Methods). Primary and secondary antibody details can be found in Tables S1 and S2. Clearing was performed as for cerebral organoids; however, the incubation in ethyl cinnamate was extended to 3 or more hours and dehydration times were 6 h per step. For adult fly recordings, the GFP fluorescence (488 nm excitation, filter 525/50) as well as the autofluorescence of 561 nm (561 nm excitation, filter, 609/54) was recorded and the autofluorescence of the 561 channel was subtracted from the GFP recording. To ensure autofluorescence specificity in the subtraction process, the 561 channel was recorded with autofluorescence intensity levels below the levels of GFP autofluorescence. Alternatively, the intensity of the 561 channel was modulated to achieve levels slightly below GFP autofluorescence levels. We did not find significant differences between 488 nm/525 and 561 nm/609 autofluorescence. Additional to autofluorescence correction, the 561/609 recording was used to reconstruct morphological details of adult Drosophila. Full details can be found in the supplementary Materials and Methods.

Axolotl tissue was harvested as previously described (Roensch et al., 2013). Briefly, axolotl tissue was fixed at 4°C overnight in 1×MEMFA [0.1 M MOPS (pH 7.4), 2 mM EGTA, 1 mM MgSO₄×7H₂O and 3.7% formaldehyde], washed in PBS and cleared as for cerebral organoids; however, dehydration and ethyl cinnamate incubation steps were increased to 12 h.

Axolotl embryos were fixed at 4°C overnight in 1×MEMFA and depigmented for 20 min using PBS supplemented with 3% H₂0₂ and 0.5% KOH (Perez Saturnino et al., 2018). As depigmentation destroys endogenous fluorescence, subsequent antibody staining is essential. Staining is performed as described below. Subsequent clearing was performed identically to cerebral organoids.

Immunohistochemistry was performed optionally after MEMFA fixation. Tissue was washed in PBS at room temperature twice for 1 h, followed by three 2 h PBX (PBS+0.3% Triton X100) washes. Blocking was performed at 37°C overnight in PBX supplemented with 5% goat serum. Staining was performed at 37°C for 48 h in PBX supplemented with 5% goat serum with primary antibody. Primary antibodies used were rabbit anti-Prrx1 (1:1000) (Ocaña et al., 2017), chicken anti-GFP (1:500) (Abcam, ab13970) and rabbit anti-RFP (1:500) (Rockland, 600-401-379). Tissue was again washed and blocked, followed by staining for secondary antibodies at 37°C for 48 h in PBX supplemented with 5% goat serum with primary antibody. Secondary antibodies used were donkey anti-rabbit Alexa647 (1:1000) (Invitrogen, A-31573), donkey anti-chicken Alexa488 (1:1000) (Invitrogen, 703-545-155) and donkey anti-rabbit Alexa568 (1:1000) (Invitrogen, A-10042). Tissue was extensively washed (three times for 3 h) in PBS after staining and processed for clearing.

To label connective tissue populations in in axolotl, the newly generated lines of Col1a2:ER-Cre-ER and Prrx1:ER-Cre-ER were crossed with the already existing CAGGs:LP-EGFP-LP-Cherry (Khattak et al., 2013). To generate the Prrx1 and Col1a2 lines, the Prrx1 enhancer/promoter (Logan et al., 2002) (a kind gift from Malcolm Logan, King's College London, UK) and the Col1A2 promoter (Bou-Gharios et al., 1996) (a kind gift from George Bou-Gharious, University of Liverpool, UK) were cloned at the 5′ end of TFPnls-T2A-ERT2-Cre-ERT2 (ER-Cre-ER) cassette with flanking SceI sites. Transgenesis was performed as previously described (Khattak et al., 2014). 4-OHT treatment was carried out as described previously (Khattak et al., 2014). Briefly, 3 cm long double transgenic animals were treated with 2 μM 4-hydroxy tamoxifen (4-OHT) by bathing overnight. Tissue was collected 2 weeks post treatment.

Chromogenic in situ hybridization was performed as previously described (Cerny et al., 2004) using stage 35 axolotl. Probes for GFAP were generated as previously described (Rodrigo Albors et al., 2015). After staining and dehydration, axolotl embryos were incubated for 15-30 min in 98% ethyl cinnamate to clear the tissue.

Zebrafish were collected and fixed in 4% PFA using standard procedures (Westerfield, 1995). Clearing was performed as described for organoids; however, dehydration and ethyl cinnamate incubation steps were increased to 12 h. Additionally, the swim bladder was pierced and allowed to fill with ethyl cinnamate.

Xenopus hindlimbs were collected as for axolotl. The pigmented skin was carefully removed from the hindlimbs using forceps. Hindlimbs were cleared as described for cerebral organoids; however, dehydration and ethyl cinnamate incubation steps were increased to 12 h.

Organoid and Drosophila recordings were performed using a Yokogawa W1 spinning disk confocal microscope (VisiScope, Visitron Systems) controlled with VisiView Software (Visitron) and mounted on the Eclipse Ti-E microscope (Nikon). Recordings were performed with a 10×/0.45 CFI plan Apo Lambda, a20×/0.75 CFI plan Apo lambda or a CFI plan Apo lambda 40×/1.4 oil (Nikon) objectives with a sCMOS camera (PCO edge 4.2 m) or an EMCCD camera (Andor Ixon Ultra 888). For stitching, the stitching plug-in in FIJI (Preibisch et al., 2009) (based on ImageJ 1.51k) was used. For 3D reconstructions, the freeware Icy (Version 1.9.5.1) was used. Axolotl and Xenopus recordings were collected on an inverted Zeiss LSM780 equipped with a 10×/0.3 EC plan-neofluar objective. Zen 2.3 SP1 (black) (64 bit) was used for image acquisition and automatic stitching of images. Adult zebrafish recordings were acquired using a Zeiss Lumar stereomicroscope equipped with Spot Pursuit-XS monochrome and Spot Insight colour cameras. Image preparation was performed using FIJI (based on ImageJ 1.51k) and Inkscape 0.91 (www.inkscape.org). Full details can be found in the supplementary Materials and Methods.

One way-ANOVA and post-hoc Tukey's test were performed to determine significance between groups using GraphPad Prism 7.0D.

We acknowledge Joshua Bagley, Tzi-Yang Lin and Andrea Pauli for the contribution of tissues and cell lines, the IMBA Fly House for the generation of transgenic flies, Tobias Müller for discussions, and BioOptics for availability and maintenance of microscopes.