Did you encounter challenges in isolating “true” hiPSCs from the heterogeneous cell populations and did you manually select the hiPSC colonies?
Thanks for this question! Isolation of hiPSC is a major obstacle during reprogramming and in genome-editing, definitely. I have done manual selection of hiPSC colonies during reprogramming, and some colonies / clones are always better than others – so, you have to start out with more colonies than you actually end up needing. For genome-edited single-cell clones, I’ve done flow-sorting and the service iotaSciences provides. Here, it’s important to start out with a large number of initial clones from which to screen as sometimes the editing efficiency is very, vey low (only 1% in my case). Once you have your clones, you need to establish their pluripotency by different methods, but so far, I have not had any disappointing surprises where the isolated clones are morphologically hiPSC but are not pluripotent. I am aware that this is possible, but I have not had it happen to me.
Nice talk! Have you checked if by dissociating the organoid you’re loosing cellular heterogenicity and/or changing their expression profile?
Thanks a lot! We have not formally checked this, no, and it’s definitely something we need to consider. What we have seen, though, is that the cell population grown in 2D tends to get more mature than what is currently contained within the organoid (i.e., Calbindin +ve cells get more dendritic branching and express more mature markers). Whether this is due to the dissociation process or just the new 2D environment, we don’t know, but I’d guess it’s the 2D setting.
Wonderful talk. i have a naive question about the organoid formation protocol. i thought some people use FGF19 and/or SDF1 for cerebellar organoids. do you think FGF2 and inculins are sufficient to generate the cerebellar organoids?
Great - thanks for the feedback and the question! The original protocol from Muguruma et al. (2015) did use FGF19 / SDF1 to produce structurally more complex organoids. We have not really tried to replicate this and instead the Becker lab have opted to use a simplified protocol whereby insulin and FGF2 are sufficient to promote hindbrain commitment (for example, seen as expression of GBX2 in the majority of cells within the organoid). Whether FGF19 and/or SDF1 could take the organoids further in terms of maturity, I don’t know, and I am definitely interested in trying that out.
Could you explain the rationale for using the transwell model in the organoid maturation phase?
Definitely! It’s been shown for organotypic slices and different organoids that the organoid tissue survives longer on the transwells (for example, Giandomenico et al. 2019). We think that lying flatter on the Transwells makes it easier for oxygen and nutrients to diffuse throughout the organoid, and thus the centre of the organoid gets less necrotic. Therefore, you can keep the organoids in culture for longer and possibly make gains in maturity. The downside is that you lose some of the secondary structures (i.e., neuronal rosettes) on the transwells.
The Muguruma paper from 2015 describes the first protocol to generate cerebellar organoids. Your protocol seems to be based on this publications. What are the differences between the Muguruma and your protocol? Did you also see other structures except of neural rosettes in your organoids? The Muguruma paper describes a rhombic lip like structure and a neuroepithelial layer but my organoids have not recaitulated that yet. Have you seen such structes within your organoids?
Great question, and the answer is that we have not replicated the FGF19 / SDF1 phenotype with the rhombic lip-like structure in our organoids, instead we have opted to use a more simplified protocol without FGF19 or SDF1. That is one big difference between our protocols. The original Muguruma paper also used human Embryonic Stem cells (hESC) that may behave differently under different protocol conditions. We use hiPSCs and tend to go for either 3D organoids or dissociated 2D cultures, depending on what is needed in the experiment and in terms of maturity.
The morphology of the calbindin positive Purkinje neurons is rather poor. Do they increase in dendrite complexity with the incubation/maturation time?
Were the Calcium imaging assays carried out on the Calb+ cells or on the Beta3Tub neurons (presumably granule cells)?
Definitely, I agree that what we call ”Calbindin +ve Purkinje neurons” here are still very immature and require further characterisation to confirm cell type and state, but if you look at the protocol paper from the Becker lab (Watson et al. 2018), these cells do increase in dendrite complexity with time and especially when cultured in 2D. Note also that data from single-cell RNAseq experiments within the 3D organoids identify populations that express early markers of the developing Purkinje cells (Nayler et al. 2021).
Your second question is the next step in confirming the profile of these IP3-responsive cells – the current set-up does not allow for both calcium imaging and immunochemistry on the same samples, but I am working towards this. It looks like the IP3-responsive cells are in the minority in each field of view, so definitely would expect them to be either the Calbindin +ve cells or a very small subpopulation of the TUJ1 +ve cells (which may represent the granule cells).
What other cells besides Beat3Tub and Calb+ neurons do the cerebellar organoids contain? Any GFAP+/glia cells? any non neural cells?
This is an interesting question that I would really like to take further in the organoid system. In one of the previous questions, I mentioned the single-cell RNAseq paper from the Becker lab (Nayler et al. 2021) which identified 12 different cell populations within the organoid, some glial / non-neural cells. TUJ1 +ve cells are enriched, especially in the 2D system, but there are definitely others, and we’re looking at ways to promote the development of these other cells types as well.
Were the hipsc crispr-cas9 edits informed directly by patient variants, meaning edits were congruent to patients?
Yes! I had one patient-derived hiPSC line (two clones from the same patient), and I produced the same patient variant in the control hiPSC line. In addition, I produced one hotspot variant that was identified in our review of the current literature. We’re definitely interested in keeping things disease-relevant.
How to ensure the gentle separation of cells, what causes the low damage to cells?
This is an important question, and usually requires experimentation (i.e., trying out different dissociation methods). It’s looking like Neuronal isolation enzyme with Papain is currently our go-to option for dissociating the organoids into the 2D environment, but there are others that may work better for single-cell sequencing, etc. Definitely try different options before sticking to one method! Also, it’s really important to optimise the time the organoids are kept in the dissociation medium, and to work quickly (but safely) from start to finish.
Do we still need single cell isolation if we prefer to study with IPS cells?How to ensure the gentle separation of cells, what causes the low damage to cells?
Thanks for the question, and I think the answer is ‘yes’, but some leeway can be gained by using multiple hiPSC lines / clones in your experiments. When picking hiPSC colonies during reprogramming, you can’t be sure your colony comes from a single somatic cell – so use multiple clones. In terms of genome-edited hiPSCs, if you don’t do single-cell isolation, you can’t be sure you know what kind of cells you’re working with (for example, one half of your population has a mutation, the second half doesn’t). If the mutation slows down cell proliferation, you’ll lose your mutant cells over time, and your results will change from one set of experiments to the next. The same is true for genomic aberrations, etc. which might cause your results to shift over time.
How does the mechanism of picking? How does the machine do it?
For the picking process, the isoPick moves the single cell you’ve selected by gently withdrawing the fluid (containing your cell) into a specialised tip. The droplet is then deposited into your chosen collection vessel e.g. a 96-well tissue culture plate. The sheer stress cells experience during this process is <0.5 psi, which is very gentle compared to several alternative technologies out there. The isoCell also performs a similar process when harvesting colonies rather than single cells, but it also allows you to apply your dissociation reagent directly into the GRID chambers containing your chosen colonies, then move them into a collection vessel after they’ve successfully detached and dissociated.
Thank you to both speakers! I have a question for Dr. Tolonen
Why does culturing the organoid on a Transwell insert and having a liquid-air interface facilitate long-term culture of the organoid better than a suspension culture in media?
Great – good to see so much activity after the talk. I think I answered a similar question above, but it’s mainly to do with better diffusion of oxygen and nutrients, when the organoid is flatter on the transwell. I think this has been shown for organotypic slices and different organoids, ours included.
What future measures do you suggest to enhance efficacy?
Great question with lots of different answers – it depends on which segment of the project you’re referring to! If we’re talking about genome-editing, definitely work with different guide RNAs before going for the actual genome-editing and clone-picking. Unfortunately, the genomic region that I wanted to edit only had very few PAMs, so I didn’t have much choice in terms of gRNA efficiency.
In a literature I have seen some usage of TGF-B blocker to stop mesenchymal differentiation. I saw that you used just FGF2 and insulin, did you face any problems in that respect?
Sorry – I should have been clearer in my presentation of the protocol. We do also use a TGF-beta-receptor blocker SB431542. I agree it’s something you need to include in the protocol.
In general when you are growing the organoids have you faced any issues about them differentiating into different progenitors cell other than cerebellum?
That is an important question, one that we’re constantly trying to answer. The issue with most organoid work is that a lot of the markers we use to identify individual neuronal cell types are actually expressed by multiple cell populations in different parts of the brain. Therefore, you can’t be 100% sure that your cells expressing marker X are the cell population you’re actually after. Each cell type needs to be characterised by multiple markers. Keep in mind also, that organoids are still a model in progress and your cell population of interest might come with subtle differences to the cell type you’re trying to model.
Use of stem cells and organoid technology for disease modelling
Cerebellar ataxias as a group of rare disorders and their genetic aetiology
Technology advances to streamline CRISPR-Cas9 genome-engineering workflows
Single-cell handling solutions for iPSCs
Who should attend
Scientists interested in CRISPR-Cas9 genome-engineering, pluripotent stem cells, organoids and disease modelling
Scientists interested in automation solutions for single-cell handling and single-cel cloning