Easy and Reliable In Vitro BBB-on-a-Chip | Mimetas

Easy and Reliable In Vitro BBB-on-a-Chip

Easy and Reliable In Vitro BBB-on-a-Chip

This is the transcript of the webinar "Easy and Reliable In Vitro BBB-on-a-Chip". For the complete webinar, register here.

Introduction to the blood-brain barrier

The blood-brain barrier consists of several cell types. The most important are the endothelial cells, which make up the actual blood vessels, and then there are several supporting cell types, such as astrocytes and pericytes. The blood-brain barrier ensures a homeostatic environment for the brain and is essential for healthy brain function.

However, the barrier also poses a problem in drug delivery, as it prevents many therapeutic drugs from entering the brain.

BBB dysfunction in apparent in many CNS diseases. Therefore, understanding how the BBB works, will likely help us understand how these diseases work and ultimately how to treat them.

We will first go over the current limitations in modelling the BBB. Then  we will give you an introduction to the OrganoPlate technology. Next,  we will go over how to create an endothelial tubule in the OrganoPlate and  we will show you that we can assess barrier function of these tubules in real-time. Then we will move on to the co-cultures and  we will discuss transporter function, antibody transcytosis, and the development of our TEER analyzer platform. In the last part, we will show you that we are moving towards mimicking the full neurovascular unit on-a-chip.

What are the limitations with current BBB models?

One of the limitations of the currently available blood-brain-barrier models is the lack of flow. Many platforms do not incorporate flow and it has been shown that flow regulates important functions in endothelium, such as barrier strength and transport activity. Incorporating flow into your model can help you to better model the in vivo situation. Also, a lot of the current models use membranes. These are often large membranes with small pores, which reduce the sensitivity of experimental assays and make it difficult to study compound transport. Also, compounds can stick to the membrane, which makes it difficult to assess compound effects. In terms of imaging, a lot of platforms used to model the BBB suffer from low quality imaging and they make it very difficult and very laborious to assess cell morphology and results of immunostaining.

The throughput is often relatively low. It depends per platform, but very often these platforms don’t come in a throughput that is amenable to high-throughput screening. Also, when you are working with expensive reagents, such as antibodies, you want to use as little as possible. Many of the current models require large reagent volumes, which increases the costs. Last but not least, many platforms, especially microfluidic systems, require special lab equipment and imaging systems, which makes adoption of the new technology very difficult.

In vitro blood-brain-barrier in OrganoPlate®


To overcome these issues, we aimed to make a platform that can be used to make better models and also to assess compound effects on these models. We are working with the MIMETAS OrganoPlate, a microfluidic 3D tissue culture plate. Depending on the type of plate you use, there are 40 to 96 chips per plate and each of those chips can be used to culture a miniaturized blood-brain barrier model. We work with a membrane-free system, which means the platform enables direct contact between different cell types and you don’t have the problem of compounds sticking to your membrane. Also, because we don’t use the membranes with the very small pores, but we use a gel of extracellular matrix proteins, this increases the sensitivity of for instance transport assays. We do work with perfusion, but unlike many other organ-on-a-chip platforms, we don’t make use of pumps. As you can imagine, if you have many chips on one plate, it can become very complicated very quickly when you are working with pumps and tubes and such. So we do create perfusion, but we do this by means of passive leveling, which  we will come back to later.   we will show you that you can do 3D cell cultures with the OrganoPlate and also that you can do complex co-cultures of different cell types present in an organ. And what’s very nice about these plates is that the bottom of the plates is made out of coverslip-thickness glass, so very thin, high quality glass, so it ensures excellent imaging of the cultures, which is very convenient for the different assays you can do.

I’ve told you before that we don’t use membranes, so what do we use to separate model compartments? What we use is called a PhaseGuide™ (from now on "phaseguide"). A phaseguide is basically a very small pressure barrier, which is about one-fourth of the height of the microfluidic channel. When a liquid, or in our case an extracellular matrix gel, is pipetted into the gel inlet, it will flow into the gel channel and then the phaseguide will prevent it from also going into the adjacent medium channel. The video shows how the extracellular matrix gel flows into the gel channel and how the phaseguide prevents it rom going into the other channel. This is not a membrane or a physical barrier that reaches all the way up to the top of the channel, it’s just one-fourth of the height of the channel, but it’s sufficient to make sure that the gel will stay where it is. At this point you can place the OrganoPlate in the incubator and allow gelation of the extracellular matrix gel and after that you can start adding cells to the medium channel for example.   
Here you once again see the gel filling and you can see the phaseguide in the middle preventing it from flowing into the adjacent medium channel. 

After that, you can add a cell suspension to the adjacent medium channel. Then we place the plate on its side to allow the cells to sediment against the ECM gel and attach there. After that we start perfusion and the cells start to proliferate and grow all the way around the medium channel, creating a perfusable tubule, which is ideal if you want to mimic endothelial microvessels for example. As you can see we have a vessel-like structure that is perfused, a phaseguide, and an extracellular matrix gel. 

The OrganoPlate is placed on a rocker platform. The platform will switch from one side to the other, creating a bidirectional flow. There’s no need for pumps or tubes, you can just place the plate on a rocker platform and that way you can perfuse all of the chips at the same time. You can adjust flow speeds by the changing the rocker’s angle, the interval with which it switches sides, or the medium volumes in the chip’s inlets and outlets. So there are three ways to increase or decrease flow and find out what’s the best for your model.

Phase 1: Tubule seeding in the OrganoPlate®

What we first do when we want to mimic the blood-brain barrier is to start with the endothelial vessel. The video before already showed a little bit on how this vessel is created, but  we will give you a brief overview again of how this is done. Basically we have a 2-lane OrganoPlate here, in which we add an ECM gel and this ECM gel fills the gel channel of the chips. After the plate has been in the incubator, the gel has become more of a solid structure and after that we can add a cell suspension of for example brain endothelial cells to the adjacent medium channel. Then we place the OrganoPlate on its side and allow the cells to attach to the ECM gel, after which perfusion is started and the cells grow around the medium channel, forming a microvessel. 

Here, we show immunostaining of primary human brain endothelial cells, which we obtained from Cell Systems. As you can see they nicely express the adherens and tight junction proteins… so that’s CD31, ZO-1, Claudin-5, and VE-cadherin. The expression is localized at the cell-cell contacts and this indicates formation of a barrier. 

Real-time barrier integrity assay

Real time BIA in OrganoPlate

Of course it’s nice to see the junctional proteins being expressed, but we also want to look at the barrier using a functional assay. That’s why we have our unique real-time barrier integrity assay. What we do is, we grow a vessel of brain microvascular endothelial cells and add a fluorescent dye to the lumen of these vessels. Then two things can happen. Either the vessel is leaktight for the molecule that we added, in which case all of the fluorescent dye is retained in the vessel… or the vessel is leaky and the dye will leak into the ECM gel channel. We can quantify this by having our software automatically select regions of interest in each chip and measure the fluorescent signal in the medium channel, which is the channel that contains the microvessel, and the adjacent gel over time. When a vessel is leaktight, the ratio of the fluorescent signal in the two channels will remain constant over time, because all dye is retained in the medium channel and then we see a flat line over time in our quantification. In case of a leaky vessel or a cell-free control, the molecule will leak into the gel channel and the ratio will increase over time. .. So we plot this for each condition over time, and basically what you have to remember for the next few slides is that a flat line indicates that a condition is leaktight for the molecule that was used, while in increasing line indicates that there’s leakage going on.  

Comparing culture media
Perfusion vs. static culture

We’ve done this using a cell line of TY10 human brain microvascular endothelial cells and what we did is culture them either static or under perfusion for 7 days and compare the barrier function in these two conditions. The red line represents the brain microvascular endothelial cells that were grown under perfusion and you can see that this line is almost completely flat, which means its retaining almost all of the fluorescent molecules of the FITC-dextran 20 kDa dye that we used. … But when you look at the purple line, you can see that when these cells are cultured static, their barrier function is not good. It’s not much better than for cell-free controls, in which we perfused the dye through chips that don’t contain any cells, but just the ECM gel. .. So from this we can conclude that perfusion actually helped these cells very much in creating a barrier, likely because the cells are experiencing a flow through the inside of the vessel and also because there is a constant supply of nutrients and oxygen to the cells We also looked at different medium compositions because these can be very important in getting your barrier as strong as possible. We used several commercially available cell culture media and found that for these specific cells, the medium from Cell Biologics resulted in optimal barrier function. So, to summarize, we see that brain endothelial cells form a much tighter barrier when cultured under perfusion compared to static culture and also that for these cells, the Cell Biologics medium results in optimal barrier function.   

Phase 2: BBB co-cultures in the OrganoPlate

Co-cultures in OrganoPlate

Of course, the BBB doesn’t only comprise endothelial cells, there are also many other cell types involved, and fortunately the OrganoPlate gives you a lot of options to work with co-cultures. Previously we were talking about the 2-lane OrganoPlate, in which chips have two lanes or channels that can be used to culture cells in, but we can also go a little bit more complex and move to a 3-lane OrganoPlate, in which the chips have an extra channel. This channel allows for more elaborate co-cultures. There’s a bunch of options… you can culture cells in gel, you can culture cells against gel, you can make combinations and you can have flow in one channel or several channels, depending on what type of culture you want to make. So this give a lot of options for us to mimic the blood-brain barrier.  

We moved on to co-cultures of brain endothelium with astrocytes and pericytes. To do this, we used the 3-lane OrganoPlate, which has 40 chips on it. In the middle channel of each chip, we seeded an ECM gel and then we grew a vessel of brain endothelial cells in the top channel using the same method we used before. What we then did, is add mixture of astrocytes and pericytes to the bottom channel and let them attach and together with the brain microvessel, these astrocytes and pericytes complete our BBB co-culture model. 

If you look at the top left image, you see a phase contrast image of a monoculture of a TY10 brain endothelial microvessel against an ECM gel. And below that we see the same thing, but now with astrocytes and pericytes added to the bottom channel to complete the co-culture. These cells are all human immortalized cells obtained from Yamaguchi University that we used in collaboration with Biogen and I really like working with these cells. On the right you see a 3D reconstruction of the BBB model, in which the endothelial vessel is stained for tight junction marker Claudin-5 and the astrocytes and pericytes were labeled using different colors of calcein. 

This is an image that was taken of the same culture, also using confocal microscopy, and it shows a maximum projection. We are looking at the chip from the top and you can see the three cell types together.

Pgp transporter assay

What we’ve seen in the previous slides is that we can co-culture the three cell types together, and next we looked at the model’s transporter function. The first transporter we looked at is the P-gp transporter, which is one of the main efflux transporters of the blood-brain barrier. We used calcein-AM, which is taken up by live cells and converted to calcein, making the cell green-fluorescent. The P-gp transporter can efflux calcein out of the cell, thereby reducing the cell’s fluorescence. When an inhibitor is added, P-gp-mediated efflux of calcein is reduced, which results in an accumulation of green-fluorescent signal in the brain endothelial cells and we can detect this using fluorescent imaging. So what we did is we cultured our BBB co-cultures and perfused calcein-AM with or without a P-gp inhibitor through the lumen of the model. We then incubated and measured the green-fluorescent signal inside the cells at the cell-gel interface. We used two different sources of primary HBMECs, which are human brain microvascular endothelial cells, and we saw that in presence of the P-gp inhibitor, the cell’s intracellular green-fluorescent signal, which is measured in the FITC channel, is increased. From this we can conclude that the brain endothelial cells in our model show P-gp transporter function. 

GLUT-1 transport assay

A similar assay was employed to study the GLUT-1 transporter, which is an influx transporter that transports glucose into the brain. We used a fluorescent glucose-analogue that’s called 6-NBDG. The glucose analogue will be transported into the cell and make the cell fluorescent. When you inhibited the GLUT-1 transporter, less of the analogue can go in, and the cells will be less fluorescent. We perfused the glucose analogue with or without a GLUT-1 inhibitor through the lumen of our model and measured the fluorescent signal at the cell-gel interface. What we see is that in presence of the inhibitor, less fluorescence is detected in the cells, which indicates that the GLUT-1 transporter is present and functional in the model. 

Studying antibody transcytosis

We would now like to show you some data from antibody transcytosis experiments in our BBB model. In the antibody transcytosis assay, antibodies are perfused through the lumen of the model. These were either target antibodies, which were designed to cross the BBB, or control antibodies, which are not expected to cross the BBB. We then sampled from the basal compartments of the chips and a MSD analysis was performed to assess the passage of the two antibodies across the model.
The graph on the right shows the apparent permeability, so the passage, of both types of antibodies across our BBB model. What we see is that, first of all, the passage of both antibodies is approximately 50 times higher in chips that don’t contain the TY10 brain endothelial cells compared to chips that do, which indicates that the barrier is doing a very nice job in preventing antibodies from leaking out of the endothelial microvessel. Secondly, we see that in the BBB co-cultures, passage of the target antibody is approximately 2-fold higher than passage of the control antibody. Both antibodies are similar in size, so the difference in passage can likely be attributed to antibody transcytosis of the target antibody. 

Barrier integrity the fast way: TEER

Something else that we have been working on is the development of a TEER analyzer platform. TEER stands for trans-endothelial electrical resistance and is a measure barrier function of a tissue, for instance of the BBB. The TEER analyzer platform will be a faster alternative to the fluorescent barrier integrity assays I’ve shown before, as it can measure TEER in 96 chips in under a minute. The TEER analyzer is fully compatible with both 2-lane and 3-lane OrganoPlates and can be placed inside the incubator with minimal cabling. So using this device, you can measure barrier integrity the fast way.

Although we haven’t performed many TEER measurements on the BBB model yet, we have been testing the TEER analyzer extensively on our gut model, which makes use of Caco-2 gut epithelial cells. The graph on the left shows in increase in TEER, and thus an increase in barrier function, in our gut model over time. The graph on the right shows how exposure of the gut barriers to staurosporin results in a concentration-dependent decrease in barrier function. So not only is the TEER device able to measure barrier function really quickly, it can also be used to assess compound’s effects on the barrier 

What cell sources can you use?

A question that I get very often is: what cells can you use to model the BBB? There’s a lot of options out there! One option you could go for is to use all primary cells. We have experience with primary human endothelial cells and primary human astrocytes and pericytes, and this works quite well. What you can also do, is work with cell lines, because these are easy in use. The lines that I have worked with and which I’m very happy with, are the lines from Yamaguchi university, so that’s human brain endothelial cells, astrocytes, and pericytes and these are also the cells featured in our recent publication that studies antibody transcytosis across the BBB on-a-chip. What we are also looking at is to move towards iPSC-derived BBB models. The last few years, a lot of progress has been made in the generation of iPSC-derived brain endothelial cells and although working with these cells is a bit more laborious, so far the results that I see with them are very promising. 

All of these cells have their advantages and disadvantages of course, and I’ve summarized them in this table. So with the primary model we see very good tubule formation and we see expression of adherens and tight junction proteins. The barrier integrity is good, there is barrier function, but it is not as good as we see with the TY10 cell line and especially with the iPSC-derived brain endothelial cells, which show very superior barrier function. However, the ease of use of these iPSC-derived brain endothelial cells is lower, and with primary cells this is better, and with the TY10 cell line this is excellent. So there’s a lot of different options that you can go for and it depends on the readout you are looking for. If you are willing to put in the time and effort to make the iPSC-derived model, that is probably a very good choice, and if you’re looking for a faster, easier option We would say go with the primary model or the cell lines, depending on what you want to work with, but all of these cell types are compatible with the OrganoPlate. 

Phase 3: Towards a neurovascular unit on-a-chip

We have been over how to make a brain microvessel and after that we moved on co-cultures of the brain microvessel with astrocytes and pericytes. The next step of course is to further increase the complexity of the model. This slide shows some of the options you have for all of the different co-cultures. On the top, it shows a tubular structure of endothelial cells and pericytes. We can culture the endothelium with astrocytes and pericytes as we’ve seen before, but what we can also do is add neurons, to move towards mimicking the full neurovascular unit.

What you see here is a 3D reconstruction of a culture that we did using endothelial cells – in this case HUVECs, but you could also use brain endothelial cells – together with iPSC-derived astrocytes and neurons, which were seeded in Matrigel in the third channel. As you can see, the astrocytes are migrating towards the endothelial microvessel, while the neurons form a network and stay put. These are mature neurons and the culture of these three cell types together already shows a big step towards mimicking the neurovascular unit on-a-chip.


To look at the function of the neurons in our neurovascular unit model, we performed calcium imaging. A fluorescent calcium indicator is loaded into the cells are increases fluorescence upon neuronal firing. What we can see here is that the cells are very active, showing spontaneous firing inside the chips. This means that if you want to look not only at whether a compound passes the BBB, but also what this compound does to the cells in the brain when it gets there, you could do that in this model.

What can you do with the OrganoPlate?

For everyone who’s not yet familiar with organ-on-a-chip technology or with our system… I often get the question which assays are possible and how difficult it is to do these assays. To start with: the platform is compatible with high-content imaging systems, phase contrast microscopy, fluorescence microscopy, and confocal microscopy. So basically you will not need to purchase any additional microscopes, you can just work with the microscopes you already have. You can also use the OrganoPlate with other standard laboratory equipment, such as pipettes and plate readers, both absorbance and fluorescence plate readers. And it’s also compatible with automation, so especially if you want to go more towards high-throughput, that’s ideal. There’s also a lot of assays that can be done. I’ve already mentioned that you can do immunostaining, which is very simple and similar to the way you would do it for 2D cultures. But you can also isolate RNA and DNA from the cultures in the plate to assess gene expression or you can isolate protein for Western Blotting. It’s possible to do calcium imaging and of course our unique real-time barrier integrity assay. .. And as I mentioned, we can also measure TEER in the system. We can assess cell viability using an number of different assays. Furthermore, we can study antibody transcytosis and assess the function other types of transporters. And, what you can also do, is sample from the plate. For instance, if you use the 3-lane OrganoPlate, you can sample from both apical and basal compartments and use these samples for further analysis, for example using ELISA, mass spectrometry, or metabolomics 

Bring the OrganoPlate to your lab!

If you are thinking about bringing the OrganoPlate to your lab, there are some things to consider. What’s very nice about the platform is that it’s membrane free. If you are working with compounds that tend to stick to membranes, it’s good to consider this. Also, if you want high sensitivity in your assays, the membrane-free nature of the platform is beneficial. .. The imaging quality is excellent, because we employ high-quality glass and the plate is basically made for imaging, so it’s perfect for that. You can also perform live imaging, so not only immunostaining and fluorescent imaging, but you can follow your cultures live as you go. .. The OrganoPlate is commercially available, both the 2-lane and the 3-lane, so it can be shipped to you immediately and you don’t need to wait. .. Our platform incorporates flow in a very user-friendly way that doesn’t require any pumps or tubes. And the platform is compatible with standard laboratory equipment, so you don’t have to spend a lot of money on new equipment or spend a lot of time learning new techniques. The adoption of the OrganoPlate in your lab is very easy. .. Last but not least, because we are working with microfluidics, the number of cells you need per chip is very low and so is the reagent consumption, which can reduce costs if you are working with expensive materials. 

MIMETAS also has experience with mimicking other organs in the OrganoPlate. Of course you can purchase the platform and mimic the blood-brain barrier, but we also have a lot of experience with kidney models and gut models. We also have experience with liver, vasculature, and brain models. And we are also working on several disease models and cancer models. So if you have any interest in the blood-brain barrier model of in one of the other models, please don’t hesitate to contact us or place your question in the box below. 

FAQ about BBB-on-a-chip

Thank you so much for joining the webinar. I hope I was able to teach you something about the OrganoPlate and make you enthusiastic about all the options to mimic the blood-brain barrier on-a-chip. Thank you so much again and  we will now move on to answering some of the questions
The first question I got is “How many cells do you need per chip?”… Well, it kind of depends on what cell type you’d be using, but generally I’d say that for endothelial cells, you can work with about 20,000 cells per chip and for astrocytes and pericytes somewhere around 5,000 to 10,000 cells per chip, depending on how fast they proliferate. So the number of cells per chip is very low, but of course you might need to optimize a bit for your specific model. 

The second question I got is “are these chips PDMS based?” The answer is no, our chips are not made of PDMS material. That’s for several reasons. First of all, if you want to incorporate PDMS in a system with several chips, it’s quite difficult because the manufacturing process of PDMS is not ideal for high-throughput systems. And more importantly, PDMS is quite a sticky material, so compounds can stick to it and it can also absorb hydrophobic compounds… and there are methods that can reduce the stickiness of PDMS, but overall, the stickiness is just inherent to the material. We use a different kind of material, it is a proprietary polymer, so I cannot disclose what it is, but it does not absorb and it has low absorbance 

So, for the third question: the question is, do you have to use the entire plate at once? The answer is no, you don’t have to use the entire plate at once. Each chip can only be used once, but you don’t have to use all the chips at the same time. So in case you are optimizing and you don’t want to use all the 40 chips of the 3-lane OrganoPlate for example, you could use a sterile sticker, like a seal that you can also use for ELISA plates for example, and then seal all the chip you don’t use. Then you can culture your chips and do all your assays and when you are finished, you can remove the seal and use the chips that you haven’t previously used. So you can definitely do several round of experiments in one plate if you would choose to do so.


For the full webinar, you can sign up here: Easy and Reliable In Vitro BBB-on-a-Chip.


Top 5: 

Download Webinar's PDF