Microfabricated disk technology: rapid scale up in midbrain organoid generation

By providing a three-dimensional in vitro culture system with key features of the substantia nigra region in the brain, 3D neuronal organoids derived from human induced pluripotent stem cells (iPSCs) provide living neuronal tissue resembling the midbrain region of the brain. However, a major limitation of conventional brain organoid culture is that it is often labor-intensive, requiring highly specialized personnel for moderate throughput. Additionally, the methods published for long-term cultures require time-consuming maintenance to generate brain organoids in large numbers. With the increasing need for human midbrain organoids (hMOs) to better understand and model Parkinson’s disease (PD) in a dish, there is a need to implement new workflows and methods to both generate and maintain hMOs, while minimizing batch to batch variation. In this study, we developed a method with microfabricated disks to scale up the generation of hMOs. This opens up the possibility to generate larger numbers of hMOs, in a manner that minimizes the amount of labor required, while decreasing variability and maintaining the viability of these hMOs over time. Taken together, producing hMOs in this manner opens up the potential for these to be used to further PD studies.


Introduction
Parkinson's disease (PD) is a neurodegenerative disorder, affecting more than 1% of the population over 65 years of age [1]. Classical hallmarks of PD are the loss of dopaminergic (DA) neurons in the substantia nigra pars compacta, accompanied by the presence of neuronal inclusions termed "Lewy bodies". To date, treatment of PD is limited to symptom management, without any cure to halt the progression of the disease [2]. Thus, it is imperative we better refine the models we use in fundamental research to better understand the pathophysiology of PD and for developing more effective therapeutic strategies. The brain is the most complex organ in the human body, rendering fundamental research and drug development for neurodegenerative disease extremely challenging.
Moreover, the inaccessibility of neuronal tissue from the brain has led to an overreliance on immortalized cancer cells or animal models to capture the architecture and interconnected nature of the brain. Yet, this approach is limited, and often intrinsic differences exist between human neurons and cell types from other species making it difficult to interpret findings. When researchers can access human brain tissue, it is often from autopsies with extensive cell loss, or from surgical samples, meaning that material is limited and often linked to a given disease [3,4]. Thus, to truly understand the brain, we need to access human neuronal tissue in quantities sufficient for largescale discovery work.
With the discovery of the Yamanaka factors in 2006 [5], the ability to generate induced pluripotent stem cells (iPSC) from the blood of any individual opened up infinite possibilities to generate customizable and physiological human cell types; and miniaturized organs in a dish. In particular, we can now generate human neurons at levels that had proved elusive. With their self-renewal abilities and potential to be differentiated into disease-relevant cells, iPSCs provide a unique platform to study PD within disease relevant human cell types [6][7][8][9][10][11][12][13][14], without the difficulties of obtaining neurons from a human surgical sample. Building on these discoveries, novel brain organoid methodologies emerged in 2013 and 2014 from the work of Madeline Lancaster and Juergen A. Knoblich [15,16]. Since these seminal discoveries, 3D neuronal structures have been used to model early brain development, neurodevelopmental and neurodegenerative disorders, to develop treatments against Zika virus [17][18][19][20][21][22][23], and for modelling COVID-19 infection [24][25][26][27]. Yet often the methods to generate brain organoids are quite labor intensive, costly and time consuming, limiting their widespread use where large amounts of cells or tissue are required. Often, research groups lack the means to make these organoids at levels needed for translational or fundamental discovery purposes [28].
High throughput screening is also an application which requires a large number of organoids of a consistent quality, and a number of recent publications have pointed the way towards incorporating brain organoids into small molecule screens. In 2021, JC Park and colleagues performed a small molecule screen with 1,300 cerebral organoids generated in AggreWell TM plates, from the iPSCs of healthy participants or sporadic Alzheimer's disease patients [29]. These plates could then be transferred into 96-well plates for screening purposes. Along these same lines, Renner and others developed an automated workflow to produce thousands of brain organoids in 96-well plates (midbrains and forebrains), from cell seeding to maintenance and analysis using robotic liquid handlers [30]. Such standardized methods reduce the variability of manual handling and demonstrate that methodologies can be applied to grow human brain organoids at large enough numbers when needed.
In our group, with the need for more than 500 human midbrain organoids (hMOs) per batch, we developed and refined our earlier process [7] of producing hMOs to ensure large numbers could be generated from batch to batch with reproducible quality. Our objective was to ameliorate hMO production, making it robust, consistent and scalable, to levels where organoids can be provided for multiple uses across a given batch. In this manuscript we describe our standardized methods to produce hMOs derived from several iPSC lines, with the use of a microfabricated embryoid body disk device (EB disk). We show that large numbers of hMOs can be simultaneously generated, similar in size, morphology, and cellular number. In addition, we describe our workflow for maintaining these hMOs in long-term culture and provide details on our microscopy analysis tools to quantify the area of hMOs and to investigate the 3D structure and total cell numbers of hMOs. With the advent of iPSCs [5] and 3D brain organoids [3, 6-15, 17-21, 31-43], the opportunity is now available to all researchers to work with living neuronal tissue that more closely resembles what we could obtain from the same region in the human brain.

Generation of hMOs using eNUVIO EB-Disks
2.1.1. List of material, reagents and equipment for generation of hMOs using eNUVIO EB-Disks A method to generate hMOs using ultra-low attachment 96-well plates was previously described [7]. Background information regarding the different media used to culture hMOs are described in [7]. The media and biochemicals used to generate the results are listed in Table 1. Consumables and equipment used are shown in Table   2.   (Fig. 1).
Details about the EB-Disk device, timeline, patterning factors and results expected are described in section 3. Results.

Day 0: Seeding of iPSCs
• Start with a 10 cm dish of human iPSCs at 80% confluency. There should be less than 5% differentiated cells. All areas of differentiation must be removed before proceeding to the next steps (Fig. 1C).
• Dissociate cells with Accutase: remove medium used to wash cells and add 5 mL of Accutase to the dish. Incubate at 37°C for 3 min. The colonies should have detached. Then add 5 mL of DMEM/F-12 + GlutaMAX TM -I + Antibiotic-Antimycotic to stop the enzymatic reaction.
• Transfer cells to a 15 mL tube, then wash the dish with 5 mL of DMEM/F-12 + GlutaMAX TM -I + Antibiotic-Antimycotic to collect the remaining detached cells.
Transfer the wash medium to the 15 mL tube.
• Remove supernatant without disturbing the pellet.
• Add 1 mL of neuronal induction media and resuspend the pellet up and down, using a P1000 pipette with a 1000 µl tip and a 200 µl tip combined.
• Count cells and evaluate the cell viability with trypan blue, which is typically around 95%.
• Prepare a mix of 9.5 million cells in 1.5 mL of neuronal induction media with ROCK inhibitor.
• Add slowly the 1.5 mL cell suspension on top of the EB-Disk 948 and close the 60 mm dish.
• Place the dish into the centrifuge adapter and centrifuge for 10 min at 1200 rpm.
• Remove the dish from the centrifuge adapter and complete with 5.5 mL of neuronal induction media with ROCK inhibitor. Note: the complementary media should be added slowly on the side of the dish to avoid disturbing the aggregates formed in the microwells. The media should fill the dish and properly cover the surface of the EB-Disk.
• Place the 60 mm dish into a 150 mm dish before moving it into the incubator.
Three 60 mm dishes can be placed into each 150 mm dish. This step prevents incubator contamination from media spills.
Critical: The quality of iPSCs is crucial to successfully generate hMOs. Thus, mycoplasma tests should be performed routinely, we suggest once a month, to confirm the absence of contamination before using the iPSCs to generate hMOs. Moreover, colonies should not present with any sign of differentiation (5% differentiation is the maximum) prior to enzymatic detachment (Fig. 1C). Additionally, to guarantee the high viability of the iPSC single cell suspension, the cell detachment step should be monitored carefully and stopped once all colonies are detached. Excessive incubation with Accutase will decrease cell viability and interfere with EBs formation.

Day 2: Neuronal induction medium change
Observation: EBs should have formed properly with a defined smooth border (Fig.   2B).
• To change the media, aspirate out the liquid from the side of the dish. Ideally, insert the end of the tip between the EB-Disk and the border of the dish. Do not aspirate from the top of the microwells to avoid suction of the EBs.
• Delicately pour 7 mL of neuronal induction media without ROCK inhibitor on the side of the dish ( Table 3). Note: pouring the liquid on top of the disk will cause EBs to lift.
• Place back the dish in the incubator.
Critical: Considering the ultra-low attachment surface of the disk it is important to not disturb the EBs that form within the wells. Dishes should be handled with care, and not be tilted at any time during the media change. With this in mind, the dish should be kept flat in the biosafety cabinet throughout this step.
• To change the media, aspirate out the liquid from the side of the dish. Ideally insert the end of the tip between the disk and the border of the dish. Do not aspirate on top of the microwells to avoid aggregate suction.
• Delicately pour 7mL of midbrain patterning medium ( Table 3) on the side of the dish to avoid lifting the EBs up.
• Place back the dish in the incubator.
Critical: Considering the ultra-low attachment surface of the disk it is important to not disturb the EBs formed within the wells. Dishes should be handled with care, and not tilted at any time during media change, as described in previous step. Thus, the dish should be kept flat in the biosafety cabinet for this step.

Day 7: Embedding and tissue induction medium change
Observation: hMOs should have grown and present with some neuroectodermal bud extrusions ( Fig. 2B-triangles). At this step hMOs attain approximately 500 µm in diameter (Fig. 2D).
Note: Matrigel® reduced growth factor should be thawed slowly at 4°C. We recommend putting a bottle of Matrigel® at 4°C on Day 6 (d6). Place 5 mL pipettes or 1 mL tips at 4°C at the same time.
• Remove the media by aspirating out the liquid from the side of the dish. Ideally insert the end of the tip between the EB-Disk and the border of the dish. Do not aspirate from the top of the microwells to avoid suction of the aggregates.
• Add 1.5 mL of Matrigel® reduced growth factor with cold tips or a cold pipet on top of the EB-disk.
• Place the dish back into the incubator for 30min.
• After 30 min, under the biosafety cabinet, add 7 mL of tissue induction medium ( Table 3) from the side of the dish. Place the dish back into the incubator.
Critical: 1) Considering the ultra-low attachment surface of the disk, it is important to not disturb the hMOs formed within the wells. Dishes should be handled with care, and not be tilted at any time during media change. As a result, the dish should be kept flat in the biosafety cabinet for this step. 2) Matrigel® reduced growth factor is temperature sensitive and gels when the environment is warmer than 4°C, thus it is critical to always keep it on ice when removing it from the fridge, and to manipulate it only with cold tips or cold pipets to prevent it from gelling.

Day 8: Transfer into spinner flask bioreactors and final differentiation medium change
Observation: hMOs will have grown in size and present with more neuroectodermal buds extruding (Fig. 2B-arrows).
Note: Metal forceps must be autoclaved before performing this step. In addition, bioreactors must be assembled and autoclaved according to manufacturer's instructions.
• Using autoclaved metal forceps, delicately transfer the EB-Disk to a 150 mm dish containing 10 mL of final differentiation medium ( Table 3). Place the disk with the microwells facing downwards.
• Gently agitate the disk with the metal forceps, a number of embedded hMOs should float out into the medium.
• Transfer hMOs with a 10 mL pipet broken at the tip (Fig. S1A), to an autoclaved 500 mL spinning bioreactor containing 300 mL of final differentiation media.
• Repeat the two previous steps until no more hMOs remain in the disk or the media.
• Place the bioreactor on a magnetic stirrer set at 42 rpm inside a 37°C incubator with 5% CO2. At this stage, hMOs can be observed floating in the media.
• Loosen the lateral caps to ensure proper air-gas exchange.
Critical: Matrigel® reduced growth factor embedding provides a scaffold for hMOs to develop at early stages. Use care when transferring these hMOs into bioreactors to preserve the scaffold and not to damage it. The use of 10 mL pipettes, broken at the extremity under the biosafety cabinet, in sterile conditions, is required (Fig. S1A).
Alternatively, you could use 25 mL and 50 mL pipettes as they have wider tips.

Suggestion:
In this protocol we did not describe the reuse of the EB disks, but it is possible to rejuvenate the disks by following the manufacturer's guidelines: https://enuvio.com/tele/2021/02/EB-Disk-User-Guide-v2.1.pdf.

Day 15 and following weeks: Weekly maintenance of the spinner flask bioreactors
Observation: hMOs have grown to full size within the bioreactors (Fig. 3A).
• Close the lateral caps before removing the bioreactors from the incubator.
• To change the media, aspirate out the media from one lateral cap. Be careful to not aspirate the hMOs. Leave 100 mL of media on top of hMOs before pouring in 400 mL of fresh final differentiation media.
• Place the bioreactors back on the magnetic stirrer set at 42 rpm inside the incubator. You should see the hMOs floating in the media with a fluid motion.
• Loosen the lateral caps to ensure a proper air-gas exchange.
Critical: Volume of media should be increased up to 500 mL in bioreactors when hMOs become bigger and turn the media yellow after one week. After media change, make sure the bioreactors are properly replaced on the cross marks, on the magnetic stirrer, to ensure the proper motion of the media spin within the flask. A media spin disturbance will lead to the disintegration of the hMOs. Perform mycoplasma test every month to ensure the quality of the material prior analytic experiments.  were set up according to optimal PMT sensitivity using the peak 2 (Spherotech) voltration technique described previously by Maeker and Trotter [46]. All data was acquired on the Attune NxT. Finally, all data generated was analyzed using FlowJo (Version 10.6) (Becton-Dickinson Biosciences).

Fixation and immunofluorescence staining of hMO cryosections
For detailed protocols, please refer to our previous publications [7,47]. Briefly, hMOs were fixed and cryosectioned according to our published methods [7]. Relative gene expression levels were analyzed using the Comparative CT Method (ΔΔCT method). The results were normalized to the GAPDH expression. The relative quantification (RQ) was estimated according to the ∆∆Ct methods [48].

Tissue clearing and imaging on hMOs
Despite the small size of hMOs, the three-dimensional (3D) imaging of organoids is challenging due to the intrinsic opacity of the biological tissues. Tissue clearing techniques make biological samples transparent by reducing light scattering and absorption throughout by removing lipids from cell membranes and reducing water content [49]. Here we use an optimized version of the water-based clearing protocol, Clear, Unobstructed Brain Imaging Cocktails (CUBIC) [50,51] described by Gómez-Gaviro et al. [52]. CUBIC is a non-toxic and straightforward to implement protocol that facilitates antibody penetration during immunolabeling, reduces fluorescence quenching and ensures high tissue transparency for optimal 3D imaging of hMOs.

List of material, reagents and equipment for tissue clearing and imaging
The composition and methods to prepare CUBIC reagent 1 and reagent 2 are described in Gómez-Gaviro et al. [52] and Table 4.

3D image processing and analysis
3D images were processed, and volumes quantified using the free software tool Fiji [53]. For each channel, the background of the images was subtracted with a rolling ball radius =50.0 pixels. The images were binarized with a "default" method, and the volumes quantified with the voxel counter plugin. Data was analyzed and graphically presented using GraphPad Prism (version 8.2.1).
Note: hMOs accumulate spontaneous dark neuromelanin granules (Fig. 4E) that will create a local black shadow during imaging.

Results
We derived hMOs from different iPSC lines using classical midbrain patterning factors [14,[54][55][56][57][58][59][60], following our previously published timeline [7] (Fig. 1A). To  (Fig 1B). We observed that the condition of the iPSCs is a primary determinant for successfully generating high-quality hMOs. iPSC colonies were maintained daily and passaged in order to present with minimal areas of differentiated cells/cells that had lost their pluripotency (Fig. 1C). We used several disks in parallel across different lines, synchronized in confluency, to generate batches with different conditions. The aggregates in each disk were then transferred into a specific bioreactor and placed on a magnetic stirrer housed within an incubator. Each magnetic stirrer can hold up to nine bioreactors (Fig. 1D).
The generation process described in this manuscript, to scale-up the production of hMOs, was based on the use of a microfabricated device or "EB-Disk" detailed in Fig. 2A. For the production of batches with several hundred hMOs, we used EB-Disks 948 containing 948 microwells ( Fig. 2A, left and top middle pictures) pre-coated with an ultra-low attachment coating, composed of 5% Pluronic acid F-127, and with defined dimensions as depicted on the lateral sectioned disk. The diameter of each disk was 52 mm and could be fitted into a 60 mm round culture dish. The volume of each microwell was 1 mm 3 while the microwell diameter at the base was 900 μm. The disks were made from PDMS, a flexible material, permitting easy handling of the platform for hMO production, while being biologically compatible with the generation of neuronal tissue [61,62].
We observed the progressive appearance of aggregates within the disk from the seeding day (d0) until their transfer into bioreactors (d8). Indeed, after 48 hours in the disk, EBs formed and presented with smooth edges (Fig. 2B). They grew and started to present with neuroectodermal buds after 3 days in midbrain patterning media (Fig.   2B, d7, triangles). The embedding on d7 led to more pronounced bud extrusions 24 hours later (Fig. 2B, d8, arrows), before the EBs were transferred into spinning flasks bioreactors to promote the growth of the organoids (Fig. 1D). We quantified the area of EBs and hMOs, derived from five distinct iPSC lines, including two Parkinson's disease lines, and compared their size from d2 to d8 within the disks, using an inhouse CellProfiler pipeline that created colored masks to visually verify the area to be quantified (Fig. 2C). Our analysis demonstrated that EBs from all lines grew in size from d2 to d8 within the disks, to attain a diameter of ~500-600 μm (Fig. 2D). The area of the hMOs generated at d8 from the disks were similar across all lines (Fig. 2D).
After 10 days in spinning flasks, we counted hMOs from two different backgrounds, and counted 650 and 714 hMOs, respectively, in each of the bioreactors (Fig. 3A).
The yield of material obtained from a 948-well EB-Disk was approximately ~65-70%.
We noted the fusion of some hMOs that accounted for numbers being lower than our starting number of 948, as well as some hMOs being lodged in the bottom of the spinner impeller. Within each cell line, we also noted a homogenous size for the organoids when we collected and fixed our hMOs at 100 days (Fig. 3B).
To assess the cellular composition of the hMOs, we performed flow cytometry analysis with hMOs from our control line that was dissociated into single cell suspensions after 50 days in culture. The numbers of cells per hMO was on average ~106,674+/-SEM 7401 cells/hMO across two batches, and the viability was ~81.3% ( Fig. 4A). Additionally, to assess the diversity of cell populations within the hMOs, we stained cryosections for S100b, MAP2 and TH, markers for astrocytes, neurons and a catecholaminergic neuron/rate-limiting enzyme in dopamine biosynthesis, respectively (Fig. 4B). Higher magnification revealed typical astrocytes, neurons and dopaminergic neurons morphologies (Fig. 4C). In 50-day old hMOs, in addition to immunocytochemistry, we utilized quantitative PCR (qPCR) to detect several transcripts known to be expressed in midbrain DNs, including nuclear receptor related 1 protein (Nurr1), LIM homeobox transcription factor 1-beta and alpha (LMX1A, LMX1B), TH, catechol-O-methyltransferase (COMT) and dopa decarboxylase (DDC).
As expected, the expression of these markers was absent in the parental iPSCs (Fig.   4D, red heatmap). Conversely, OCT3/4 and NANOG, transcripts known to be expressed in iPSC were only detected in iPSCs and absent from the hMOs samples ( Fig. 4D, blue heatmap). Additionally, after 100 days in final differentiation media, hMOs started to develop and accumulate spontaneous neuromelanin granules (Fig.   4E). This observation was previously confirmed to be a feature of functional dopaminergic neurons [7,14,63].
Next, to further validate the midbrain identity of our hMOs and the reproducibility of our method, we cleared and stained four 6-month-old hMOs from our control line for TH and MAP2. Consistent with published findings [7,9,14], we observed the presence of dopaminergic neurons stained for TH within the hMOs (Fig. 5,

Conclusion
In this manuscript, we describe a novel protocol applying a microfabricated EB-Disk device for the generate large batches of homogenous hMOs across multiple different iPSC lines. We describe in detail the critical points to successfully achieve hMO generation. For instance, the quality of iPSC colonies is the first critical aspect in generating high quality hMOs [7]. Our protocol presents the following advantages to users. Working with this method, they will be able to: 1) generate hundreds of hMOs for a given batch all from a single iPSC in one workflow; 2) avoid direct contact with EBs and minimize manual handling of hMO cultures; 3) lower reagent volumes required, by reducing the amount of media used for the generation and maintenance of hMOs, and also the quantity of Matrigel® reduced growth factor required for the embedding step, compared to previously described protocols [7,14,15].; and 4) reduce the overall cost of hMO production, by reducing the media changes to once a week, compared to three times a week as in our previous methods [7]. To note, this method does not require any robotic or automation system as described in other publications [30], and therefore can be used across the majority of laboratories, once they have access to an incubator, a magnetic stir plate and several glass bioreactors, equipment that can be easily purchased.
Upscaling has been a focus of the organoid field and different methods are being explored to achieve a higher throughput. Spinning bioreactors are used to grow large numbers of organoids and have the benefit of providing improved oxygen and nutrient distribution to the organoids in long term culture [15]. Regarding brain organoids, spinning bioreactors have been used by several groups [15,36,64], and while they require a large amount of media and space, maintenance is less time consuming and does not require manipulation to the same extent as multi-well plates.
Qian and colleagues [17] developed miniaturized bioreactors, reducing considerably the amount of media needed. However, this required 3D printing and a complex assembly, and does not reduce the need for manual handling of the culture vessels.
Other types of organoids can also be created using adapted protocols with the EB-Disk to model diseases that affect other areas of the brain, or even other organs, opening up a number of new possibilities for these microfabricated devices. Depending on the organoid type, stem cells can be cultured in a different vessel prior to transfer into spinner flask bioreactors or can be directly seeded into them. Groups with an interest in tissue transplantation and regenerative medicine successfully upscaled their organoid production for liver or pancreas, by directly seeding cells in the flasks which permits an increase in the size of their organoids and the number of organoids in culture at a time [65,66].