Thoughts on imaging screens on 3D cell culture models
von Michael Schell
Stem cells have been used in research for many years. However, the possibility to grow stem cells in three dimensions (3D) to build miniature organs (organoids/spheroids) that can mimic many aspects of parent tissue in a dish was somewhat game-changing. Organoids recapitulate the genetics, cellular organization, and to a certain extent the functionality of the corresponding parent organ.
Organoid models are available for researchers for diverse organs, cell types, cell environments, or disease states and can be patient-derived from human donors, too.
Figure 1: Smiling organoid - Organoid imaging and analysis was done in a 96 well plate
with Yokogawa CQ1 at 40x magnification. GFP transfection was monitored.
Growing spheroids in a U-bottom shaped low adhesion plate is one approach to understanding certain diseases and their cell-cell interactions. The cells form spheroids, three-dimensional aggregates showing a specific cellular microenvironment such as an apoptotic core with less oxygen and a mixture of proliferating and non-proliferating cells. By means of automated microscopy one can efficiently analyze these aggregates in high-throughput and receive information about the cell-cell interaction or cell-communication in this 3D model.
Why are they important and where to use them?
Many traditional 2D in-vitro compound studies have a low success rate when translating them to clinical trials. 3D culture assays are showing strong evidence of how their microenvironment is more indicative of drug treatment effects on tumors in vivo. During the last years, 3D organoid models have become more and more important. They are usable for many researchers and applications. Questions remain as to how and why 2D and 3D cultures behave differently in treatments such as anti-proliferative compounds. Previously published data have reported the utility of automated imaging and analysis of spheroids with a 96-well U-bottom low attachment plate.
The application ranges from simple tumor models to more complex co-culture systems. Organoids and spheroids are superior to flat 2D cell cultures in their physiological relevance, increased cellular complexity, and yet can be used for screening since they show high reproducibility. Because they are less expensive than breeding mice or other animal models, they are now widely used as a model for drug discovery and for drug toxicity studies.
What is needed to set up such a screen?
First and foremost you have to have your biology in place. As this is not our specialty and there are so many different models around, we won´t comment on that and focus on “everything imaging”.
And here is little way around a powerful infrastructure. Your technical stack includes an automated high-end microscope, computing, and IT infrastructure. The exact goal and readout of your screen defines the effort needed for all three of these items.
In general – and that´s what the term screening implies – your hardware needs some degree of sample throughput. Depending on the nature of your samples and the number of compounds you want to check, a “screening campaign” can range from a few hundred to hundreds of thousands of samples to image and analyse. Hence, as a minimum your hardware must accommodate SBS-standard microtiter plates, ideally not just 96well, but more dense formats as well.
At the lower end of the compound library size manual plate feeding is a valid option, but if your campaigns regularly get bigger than a few dozens of plates, a larger setup including automated plate feeding from an incubator or at least a plate stack might be an option.
A point that cannot be emphasized enough is plate quality. While many plates are suitable for cell culture, and every lab has not only their pet cells, but also their pet plates, for a screen the optical properties of the plate can make all the difference. The exact optical property needs of the plates will be defined again by the ultimate readout, and at this point it is essential to follow the microscope manufacturer´s guidelines. The rule is: thoroughly test your pet plates on the relevant imager! If they are fine, great, if not, please make the effort of moving over to recommended plate types and if needed adapt your workflow respectively.
All this being established, your objects grew nicely, your compound logistics were fine, your imaging campaign should give you excellent images and all that´s left now is running the imaging and – ideally alongside – the data evaluation. Once more, there are stark differences depending on the readout you have chosen. The amount of raw images and data to be generated, handled, stored, and processed, and therefore the required time and computing power, can vary dramatically. A small campaign on a “size and viability” screen of tumor spheroids can be easily done on a non-confocal imager, it does not require z-stacking, a plate is typically done in less than a few minutes, and the amount of data generated is typically under a few GB per plate. On the other end of the spectrum a confocal 3D-analysis of large >100µm organoids with multiple fluorescence wavelengths and the respective z-stacking will take much more time per plate (depending on the number of cameras your imager provides) and generate hundreds of GB of data, which need storing and processing.
Needless to say that the data evaluation packages coming with your hardware, their capabilities, ease of use, and power can make a difference for your overall effort and time.
Figure 2: Example for screening images and related evaluation using a heatmap for a quick trend visualisation.
In this case Organoid imaging and analysis was done in a 96 well plate with Yokogawa CQ1 at 10x magnification.
Analysis of organoid area was done with the CellPathfinder software and is represented as heatmap
with a blue red color-code representing low and high values, respectively.
But what do you need now in terms of hardware?
For many screening assays a non-confocal resolution of the microscope is sufficient, which limits the investment need not just here, but also for computing and IT infrastructure. So, if you are rather interested in a broader picture, like how many organoids are in a well or what is general the intensity of marker A versus marker B, it might be well suited to run a screening campaign on a microscope that is targeting speed and not so much the resolution.
On the other hand, if you expect tiny structural changes on the (sub)-cellular level or elucidate details of the interior of a 3D-structure such as an organoid, you may want to have a confocal microscope with a microlens-enhanced double Nipkow spinning disc at hand. Just to ensure not only the desired resolution, but also excellent brightness during the scan.
Depending on the focus of your work, and of course your budget, one or the other can be the better choice. And for pure efficiency reasons may be helpful to have access to both. Some of our customers have seen the synergies between both approaches and equipped their labs accordingly. In one of the next blogs we´ll have a closer look at what hardware options Cenibra has to offer and how such synergies can come to fruition if you have access to both these configurations.