Design and handling of DNA sequences
The sequences used to prepare the DNA-Y-motifs YA and YB as well as the DNA linker were adapted from previous publications29,40. DNA strands were purchased either from Integrated DNA Technologies (unmodified DNA, purification: standard desalting) or Biomers (modified DNA, purification: HPLC). All DNA, apart from fluorophore-labelled strands, was diluted in 10 mM Tris (pH 8) and 1 mM EDTA (Sigma Life Science) to yield 800 μM stock solutions. Fluorophore-labelled strands were diluted in MilliQ water to yield 800 μM stock solutions. All utilized DNA sequences are listed in Supplementary Table 7. The DNA stock solutions were stored at −20 °C.
Preparation of Y-motif DNA
The DNA-Y-motifs (YA and YB) needed to form the DNA microbeads were produced via thermal annealing of the three respective single-stranded DNA strands YA-1, YA-2 and YA-3 for YA, or YB-1, YB-2 and YB-3 for YB. The strands were mixed at equimolar ratios to yield a final concentration of the resulting Y-motifs of 150 µM. In all experiments, 4 mol% of Cyanine-3 (Cy3)-labelled YB-1 strand was added to the YB mixture to allow for fluorescence microscopy of the resulting DNA microbeads. The Y-motifs were annealed in a solution containing 1× PBS (Gibco). Annealing was conducted in a thermal cycler (BioRad) by heating the samples to 85 °C for 3 min and subsequently cooling the sample to 20 °C using an increment rate of −0.1 °C s−1.
Formation of DNA microbeads
DNA microbeads were created in a templated manner after encapsulation of the gelation solution into water-in-oil droplets. To form the DNA microbeads, the annealed Y-motifs YA and YB were mixed at equimolar ratios (20 µM, 25 µM or 30 µM) in a solution containing 1× PBS. The DNA linker strand was then added to the solution in 3× excess to the Y-motifs (for example, 30 µM Y-motifs + 90 µM DNA linker). Immediately after the addition of the DNA linker, the mixture was added on top of an oil phase containing 2 wt% perfluoropolyether–polyethylene glycol (PFPE–PEG, RAN Biotechnologies) dissolved in HFE-7500 (Iolitex Ionic Liquids Technologies) at a ratio of 1:3 aqueous phase to oil phase (for example, 50 µl aqueous solution and 150 µl oil mixture) and the reaction tube with the mixture flicked with a finger 8× to create an emulsion. The resulting water-in-oil droplet emulsion was incubated at 22 °C room temperature for 72 h to ensure full gelation of the DNA microbeads. After this, the DNA microbeads were released by breaking the water-in-oil emulsion. To release the microbeads, a 1× PBS solution was added on top of the droplet emulsion. Subsequently, the emulsion was destabilized by adding the surfactant 1H,1H,2H,2H-perfluoro-1-octanol (Merck) on top of the buffer. This mix was incubated for 30 min before the resulting aqueous phase containing the DNA microbeads was taken off and transferred to a separate reaction tube. The DNA microbeads were stored at 5 °C before their use and prepared fresh for each experiment. DNA microbead components and their concentrations for all microbeads used in this study are detailed in Supplementary Table 8.
Real-time deformability cytometry
RT-DC was performed using an AcCellerator (Zellmechanik Dresden) mounted on an inverted AxioObserver microscope (Carl Zeiss AG) equipped with a 20×/0.4 Ph2 Plan-NeoFluar objective (Carl Zeiss AG). Images were acquired using a high-speed CMOS camera (MC1362, Microtron).
To measure the DNA microbeads, a suspension of microbeads (100 µl) was strained through a 20 µm EASYstrainer filter (Greiner Bio-One) and pelleted in a reaction tube by spinning them down for 2 min with a C1008-GE myFUGE mini centrifuge (Benchmark Scientific). The supernatant (80 µl) was then taken off and discarded, and the remaining pellet of DNA microbeads was resuspended in 150 µl of CellCarrierB (Zellmechanik Dresden). The resuspended microbeads were then aspirated into a 1 ml glass syringe with a PEEK tubing connector and PTFE plunger (SETonic) mounted on a syringe pump system (NemeSys, Cetoni). The DNA microbead-CellCarrierB solution was then injected into a Flic20 microfluidic chip (Zellmechanik Dresden) using PTFE tubing (S1810-12, Bola). Through a second 1 ml glass syringe, CellCarrierB was injected into the Flic20 microfluidic chip as sheath flow for the RT-DC experiment. For all samples, measurements at 0.04 µl s−1 total flow rate (ratio of sheath-to-sample flow 3:1) were run for a duration of at least 900 s each. The measurement software ShapeIn (version 2.2.2.4, Zellmechanik Dresden) was used to detect the DNA microbeads in real time. The pixel size was adjusted to 0.68 µm px−1, fitting the utilized 20×/0.4 Ph2 objective and all DNA microbeads imaged at the rear part of the flow channel ensuring regular deformation of each microbead. For each condition, triplicates were measured. Measurements of the DNA microbeads containing Wnt-surrogate were conducted in the same way, following an overnight incubation of the DNA microbeads with a Wnt-surrogate-modified DNA linker (see section ‘Formation of DNA microbeads with PC Wnt-surrogate’) and three washing steps using 1× PBS. Before the overnight incubation, the DNA microbeads were likewise filtered through a 20 µm EASYstrainer filter (Greiner Bio-One).
The same workflow was applied to dissociated medaka RO cells. In preparation for RT-DC, medaka RO were cultivated as described in ‘Generation of medaka-derived RO’ until late day 1. Forty-eight organoids per experiment were then pooled into 2 ml tubes and washed multiple times with 1× PBS. Dissociation was performed by incubation in dissociation solution (1:1 dilution of 2.5% Trypsin (Gibco, catalogue number 15090046) and 1 U ml−1 Dispase (Stemcell Technologies, catalogue number 15569185)) for 10 min under gentle shaking and occasional gentle pipetting at 28 °C. Trypsin was quenched by diluting the dissociation solution 1:2 in 50% FBS containing 1× PBS solution. Single cells were spun down at 200 × g at room temperature for 3 min, the supernatant was aspirated, and cells were resuspended in 150 µl CellCarrierB. The cells were likewise measured as triplicates (48 organoids each) resulting from independent sets of organoids for each measurement.
Following RT-DC, the utilized microfluidic chips were flushed with a fluorescein-MilliQ water solution and z-stacks of the flow channels acquired with an LSM 900 Zeiss confocal fluorescence microscope (Carl Zeiss AG). For each z-stack, the pinhole size was set to one Airy unit and a Plan-Apochromat 20×/0.8 Air M27 objective was used. The median width of each flow channel was then calculated from the z-stack using a custom Python script and the RT-DC data corrected accounting for the width of the respective flow channel.
The analysis software Shape-Out (version 2.10.0, Zellmechanik Dresden) was then used for data analysis. All samples were gated for porosity (1.0–1.2) and size (65–160 µm2). Statistical analysis based on a linear mixed model (R-lme4) as implemented in Shape-Out (version 2.10.0, Zellmechanik Dresden53), calculation of Young’s moduli, deformation and volume as well as preparation of the data for contour and violin plots were all carried out using Shape-Out (version 2.10.0, Zellmechanik Dresden). The linear mixed model was run without adjustments. P-value calculations to determine statistical significance are based on analysis of variance (ANOVA) test to correctly analyse the data as derived from RT-DC measurements53. Plots for the volume, deformation and Young’s modulus were created using OriginPro 2021, Update 6 (OriginLab Corporation).
Formation of PC DNA microbeads and quantification of DNA microbead disassembly using light
PC DNA microbeads were formed in the same way as detailed above. However, 60% of the utilized linkers contained a PC moiety in the centre of the DNA linker sequence (PC linker; for details, see Supplementary Table 7). In triplicates, five PC DNA microbeads per sample were analysed to quantify the breakdown of the DNA microbeads following exposure to 405 nm light. The microbeads were chosen to be 50 µm in diameter and imaged using 5× digital zoom. The frame time was set to 148.95 ms and the pixel size of the acquired image to 256 × 256 px. To break down the DNA microbeads, the laser power of a 405 nm confocal laser (5 mW maximum power) was set to 10% and the microbeads were continuously irradiated for 60 s, resulting in their disassembly. In addition, DNA microbeads without PC linker (five per replicate with three replicates total) were treated in the same way as above as a negative control. Analysis of the disassembly was then performed in Fiji (NIH54). For this, the mean fluorescence signal across the irradiated images was acquired and the data normalized to the first frame of each video. The data were plotted using OriginPro 2021, Update 6 (OriginLab Corporation).
Conjugation of Wnt-surrogate proteins to DNA linkers
WNT-surrogate-Fc fusion protein (Wnt-surrogate; ImmunoPrecise Antibodies; catalogue number N001, lot 5696, 6384, 7134, 7568) was dialysed against 25 mM HEPES and 500 mM NaCl buffer at pH = 8.2 using ZelluTrans/Roth Mini Dialyzer tubes MD300 (12–14 kDa, Carl Roth). Dialysis was conducted at 4 °C for 36 h with hourly buffer changes during the day and a long incubation overnight to remove Tris from the buffer solution. Modification of the Wnt-surrogate with an azide moiety was achieved using an azidobutyric-NHS ester (Lumiprobe) according to the manufacturer’s recommendations. Further, modification of the Wnt-surrogate with Alexa Fluor 647 (Wnt-AF647) was achieved by adding an NHS-modified Alexa Fluor 647 ester (AF647N-NHS, Lumiprobe) simultaneously to the azidobutyric-NHS ester in accordance with the manufacturer’s recommendations.
The resulting solution was then again dialysed against 25 mM HEPES and 500 mM NaCl buffer at pH = 8.2 in the same way as before to remove any unreacted NHS esters. DBCO-modified DNA linker strands (PC or non-PC; Supplementary Table 7) were then added to the azide-modified (azide/AF647-modified) Wnt-surrogate in a 1:1 ratio and incubated to react for 76 h, yielding a final concentration of 8 µM Wnt-surrogate-modified (Wnt-AF647-surrogate-modified) DNA linker.
Formation of DNA microbeads with PC Wnt-surrogate
After a DNA microbead suspension was passed through a 20 µm filter, 30 µl of this DNA microbead suspension was pelleted using a C1008-GE myFUGE mini centrifuge (Benchmark Scientific) for 2 min. Then, 20 µl of the supernatant was removed to leave 10 µl of the DNA microbead pellet in the reaction tube. To achieve the incorporation of DNA linker with PC Wnt-surrogate, the DNA microbead pellet was resuspended with 10 µl of PC Wnt-surrogate-modified DNA linker (8 µM), yielding a final concentration of 4 µM modified linker. The mixture was incubated overnight, after which the microbeads were washed three times using 100 µl of a 1× PBS solution to remove non-incorporated DNA linkers and proteins, yielding a final volume of 10–15 µl of modified DNA microbeads after removal of the washing solution after centrifugation. Formation of DNA microbeads with Alexa Fluor 647-labelled Wnt-surrogate was conducted in the same way using Wnt-AF647-modified DNA linkers. Note that substantially less than 1 µl of the final DNA microbead suspension is used for the microinjection of up to 50 organoids. The volume produced this way is thus sufficient for the microinjection of more than 500 organoids.
Quantification of the release of Alexa Fluor 647-modified Wnt-surrogate (Wnt-AF647) from DNA microbeads
To quantify the release of Wnt-AF647 from the DNA microbeads, the microbeads (n = 5) were illuminated with a 405 nm laser at 10% power (5 mW maximum power) and imaged for 180 s until the Wnt-AF647 signal was depleted. Irradiation of the DNA microbeads with the 405 nm laser started 20 s after the start of the imaging. The frame time was set to 148.95 ms and the pixel size of the acquired image to 256 × 256 px during imaging. The mean fluorescence signal of the Alexa Fluor 647 dye within the DNA microbeads was then measured using the circle tool in Fiji (NIH54) across all frames. All data were normalized to the mean fluorescence detected in the first frame of each video and plotted using OriginPro 2021, Update 6 (OriginLab Corporation).
Fish husbandry and maintenance
Medaka (O. latipes) stocks were maintained according to the local animal welfare standards (Tierschutzgesetz §11, Abs. 1, Nr. 1, husbandry permit AZ35-9185.64/BH, line generation permit number 35-9185.81/G-145/15 Wittbrodt). Fish are kept as closed stocks in constantly recirculating systems at 28 °C with a 14 h light/10 h dark cycle. The following medaka lines were used in this study: Cab strain as a wild type55 and Atoh7::EGFP56.
Generation of medaka-derived RO
Medaka-derived RO were generated as previously described3 with slight modifications to the procedure. In brief, medaka primary embryonic pluripotent cells were isolated from whole blastula-stage (6 h post fertilization) embryos44 and resuspended in modified differentiation media (DMEM/F12 (Dulbecco’s modified Eagle medium/Nutrient Mixture F-12, Gibco, catalogue number 21041025), 5% KSR (Gibco, catalogue number 10828028), 0.1 mM non-essential amino acids, 0.1 mM sodium pyruvate, 0.1 mM β-mercaptoethanol, 20 mM HEPES pH = 7.4, 100 U ml−1 penicillin–streptomycin). The cell suspension was seeded in densities of 1,500 cells per organoid (approximately 15 cells per µl) for standard-sized organoids and 500 cells per organoid for small organoids in 100 µl per well in a low-binding, U-bottom-shaped 96-well plate (Nunclon Sphera U-Shaped Bottom Microplate, Thermo Fisher Scientific, catalogue number 174925) and centrifuged (180 × g, 3 min at room temperature) to speed up cell aggregation. At day 1, aggregates were transferred to fresh differentiation media and Matrigel (Corning, catalogue number 356230) was added to the media for 9 h to a final concentration of 2%. From day 2 onwards, RO were kept in maturation media (DMEM/F12 supplemented with 10% FBS (Sigma-Aldrich, catalogue number 12103C), 1× N2 supplement (Gibco, catalogue number 17502048), 1 mM taurine (Sigma-Aldrich, catalogue number T8691), 20 mM HEPES pH = 7.4, 100 U ml−1 penicillin–streptomycin). For the analysis of the spatial correlation between the DNA microbeads’ position and the induced RPE differentiation, organoids were kept in differentiation media for the whole duration of organoid culture. RO thus developed less RPE after induction (alongside generally being smaller). This enabled a more precise investigation of the spatial relationship of the DNA microbead position and the emerging RPE differentiation pattern after DNA microbead-mediated Wnt-surrogate release at day 1.
RO were either derived from embryos of wild-type Cab strain only (Figs. 2b and 3, and Supplementary Fig. 12) or mixed with blastomeres of blastula-stage embryos of the Atoh7::EGFP transgenic line (outcrossed to Cab) in a 4:1 ratio. Mixing primary pluripotent embryonic stem cells from wild-type and transgenic embryos in this ratio ensured that only a fraction of retinal ganglion cells was being reported for. This facilitated the identification of qualitative differences in cell numbers and distribution within individual organoids owing to reduced clustering of reporter cells. In this way, the labelled retinal ganglion cells were used as a proxy for the overall formation of neuroretina in the organoids.
RO microinjection
For microinjection, day 1 RO were washed 3 times after 9 h of Matrigel incubation, transferred onto Parafilm (Thermo Fisher Scientific, catalogue number 13-374-10) and lined up against the edge of a square coverslip (24 × 24 mm) in differentiation media. Borosilicate micropipettes (1 mm OD × 0.58 mm ID × 100 mm L; Warner Instruments, catalogue number 30-0016) were pulled on a Flaming/Brown micropipette puller P-97 (Sutter Instruments) with the following settings: heat 505, pull 25, velocity 250, time 10, 1 cycle. The microinjection was performed with a CellTram 4m oil microinjector (Eppendorf AG) and a standard manual micromanipulator under an epifluorescence stereomicroscope (Olympus MVX10; MV PLAPO 1× objective) to visualize Cy3 fluorescently labelled DNA microbeads during microinjection. Note that all DNA microbead suspensions used for microinjection into RO were passed through a 20 µm filter before microinjection.
For UV light-triggered release of the DNA microbead’s cargo or disassembly of DNA microbeads themselves in live RO, organoids kept in 100 µl differentiation media on a culture dish were exposed for 60 s at a 1 cm distance to Leica EL6000 (100% intensity; Lamp HXP-R120W/45C VIS, power input 120 W, Osram Licht AG). Analysis of the disassembly (Supplementary Fig. 6) was then performed in Fiji (NIH54). For this, the mean fluorescence signal across a region of interest (ROI) of the DNA microbead position within the images was acquired and the data normalized to the first frame of the time-lapse imaged RO.
Wnt-surrogate release from DNA microbeads was conducted 2 h post microinjection on day 1 of RO culture, since Wnt-surrogate-mediated induction of RPE was found to be only possible on late day 1.
Embryo microinjection
Stage 20 (1 day post fertilization) embryos44 were dechorionated using hatching enzyme, washed and kept in 100 U ml−1 penicillin–streptomycin containing ERM (17 mM NaCl, 40 mM KCl, 0.27 mM CaCl2, 0.66 mM MgSO4, 17 mM HEPES). Embryos were transferred onto a 1% agarose mould57, oriented heads down for microinjection and punctured at the vegetal pole. Microinjected embryos were re-cultured on glass ware in 100 U ml−1 penicillin–streptomycin containing ERM until hatchling stage (s40 (ref. 44)) with daily assessment of their gross morphology by stereomicroscopy.
Radial diffusion analysis of Cy3-labelled DNA-Y-motif and Wnt-AF647 in small RO
For the radial diffusion analysis of the DNA-Y-motif and Wnt-AF647, the pixels with intensities above the 0.98 and 0.99 intensity quantiles in the initial images, respectively, were averaged to obtain the centre of mass positions (COM). For the Wnt-AF647, the sum projection was considered to average over a height of 30 µm.
Around the COM, the image intensities were radially averaged in azimuthal sections of 60° (Fig. 3g,f). For the Wnt-AF647, the boundary of the inclusion region in the individual sections was determined as the maximum radius with a half-maximum intensity in the Wnt-AF647 channel. For the DNA-Y-motif, the imaging plane barely touched the microinjection region and thus the inner boundary is assumed to lie at radius 0. The outer boundary in the sections was determined as the averaged boundary from manual segmentation (Wnt-AF647; Fig. 3e), or the maximum radius with an averaged half-maximum intensity as measured from the plasma membrane staining (DNA-Y-motif; Fig. 3d). For each section, the radially averaged concentration profiles between the inclusion and the organoid boundary were rescaled to the interval (0,1) and then all datasets were spatially averaged with a moving average approach with a 10 times smaller resolution as the coarsest resolution in the sections. Within these averaging intervals, the standard deviation was calculated to obtain the error bands.
Statistics and reproducibility
Statistical analysis was conducted either using a linear mixed model approach, deriving a P value using ANOVA test (according to the RT-DC workflow as published53 and implemented in the analysis software Shape-Out (version 2.10.0, Zellmechanik Dresden; for details, see section ‘Real-time deformability cytometry’)), or using two-tailed Student’s t-test with unequal variance (calculation of significant differences in Fig. 4). In all cases, P values 
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
