[Neurosciences/BBB] The SARS-CoV-2 spike protein alters barrier function in 2D static and 3D microfluidic in-vitro models of the human blood–brain barrier (Buzhdygan et al., Neurobiol. Dis. 2020)

Recently, a study from Ramirez and colleagues was brought to me on my attention and asked my view on it, as it was discussing SARS-CoV-2 and the blood-brain barrier (BBB). Myself dealing with manuscripts submissions (2 undergoing revisions, 2 manuscripts in preparation), I had to put it on the back burner until I could address it.
Thanks to Thanksgiving break, I can finally put my hand on and review it as it raises some good questions that several colleagues (including myself) are currently asking: What is the effect of COVID19 on the BBB? Does the neurological symptoms observed in COVID19 are linked at some form with the BBB? Can SARS-CoV-2 cross the BBB? What is the function of ACE2 at the BBB? Servio and his group decided to address that question by running this study in their lab, and after having a pre-print on BioRxiv finally found a place to have it peer-reviewed and published.

Intro and methods: In this study, Ramirez and colleagues decided to investigate how elements of the SARS-CoV-2 spike S surface proteins would impact the cell viability of primary human brain endothelial cells (hBMVECs) and compared it to hCMEC/D3 (a common cell line used as a model of the human BBB in vitro). They used various techniques including cell viability, molecular biology to assess changes in expression at mRNA and protein levels, barrier function (using the ECIS TEER and permeability to FITC-labelled 4kDa dextran), all of it in two models: one 2-dimensional (Petri dish) and one 3-dimensional (microfluidic chip). One of the caveat, and I completely understand as I have the same issue, is the use of recombinant proteins instead of a full SARS-CoV-2 virus. This is because at this point, the use of SARS-CoV-2 in the growth of virus requires a BSL-3 research laboratory. These are very uncommon and restricted to big institutions as it requires a level of biosafety drastically higher than a BSL-2 laboratory that is common place in many research universities.

Results: The first results presented is the expression of ACE2 at protein levels in small vessels of human brains tissue sections obtained by post-mortem brains from patients with a history of dementia or hypertension (Fig.1). What is interesting is what appears an up-regulation of ACE2 expression (or an increase in ACE2-positive microvessels) in dementia and hypertensive brains. This is interesting, because it would be interesting to see how those two conditions are accounting as risk factors for COVID19, or if such conditions were associated with higher incidence of neurological symptoms. What is also interesting, is that ACE2 expression is only mildly induced by TNF-alpha or exposure to recombinant spike S proteins produced in E. coli and HEK293T cells.

Fig. 1

The second figure (Fig. 2) investigated the effect of spike S protein subunits (S1, S2 and the RBD) on cell viability following 48 (A) or 72 hours (B). Saponin (a detergent) was used as a positive control to show that the method can assess live-dead cells. We can see a slight but significant increase in number of dead cells (~5%) in RBD-treated cells. The caveat is the concentration used of these protein (1 and 10nM respectively), which brings us the question of how representative such concentration matches to viral load? Is it a realistic concentration in patients on ventilator or it is just overblown? That’s something that would need to be addressed.

Fig. 2

The weird kind of data, in my opinion, is the measurement of the barrier function using the ECIS TEER and permeability. We see a dose-dependent decrease in the barrier function over 24 hours following treatment with S1, S2 or RBD proteins but one thing that bother me is the amplitude of decrease. We are measuring here less than 0.2% change in TEER compared to untreated cells. That’s low, very low to be honest (usually you are aiming at least 20-30% difference) and this is where I stood unimpressed with the data. Especially we have a conundrum here: how do you explain a minute difference in TEER (<0.2% change), while seeing seeing at least 15-20% increase in permeability to a big molecule (4kDa)? I would suspect that the ECIS TEER is not efficient, or these cells are simply too leaky to see drastic changes on the ECIS (this is supported by the use of 4kDa-dextran, which is about 10 times bigger than more common fluorescent paracellular tracers such as fluorescein and Lucifer Yellow).

Fig. 3

Figure 4 kind of reproduce the Figure 3 results, but this time in the 3-dimensional model, seems like S1 treatment resulted in a significantly higher permeability (5x or 500% increase) than the 2-dimensional model. One thing that bother me is why the author did not applied the same treatment groups (TNF-alpha, S1, RBD and S2). It is interesting also what is the rationale that explain such a drastic difference between the 2-D and 3-D models? It is I think interesting to see if it is a fluke or real, because it may give an upper hand to the later, making the microfluidic a more relevant model than the static model.
Figure 5, using flow cytometry, show that treatment with S1 or S2 (but not RBD) are somehow capable to activate hMBVECs and prime them to recruit immune cells from the periphery, as we could see an increased expression of ICAM-1 and VCAM-1 on the cell surface, such up-regulation seems to ocur as early as 4 hours and peak by 24 hours.
Finally, gene expression show that overall S1 seems to be a major driving force in gene expression, as we can see a significant increase in CCL5 and CXCL10, with CCL5 gene expression at 24 hours being peaking for all three treatment (S1, S2 and RBD). In the other hand, all three proteins were sufficient to increase gene expression of MMP3 and MMP12, two matrix metalloproteinases that could possibly induce the disruption of cell-cell junctions.

Conclusion: This is a nice little paper that brings a bit of information on SARS-CoV-2 at the BBB. Surely, spike S protein can interact and transduce some signals at the BBB which ultimately result in change in the barrier function. However, we only have fragments of it as we have to see how changes in tight junction complexes correlate with changes in the barrier function. We also have to address how relevant the concentrations used. I think 10nM is pretty high and would require some huge amount of viral load to happen, hence it would be great to compare and contrast it to viral load (I would see the use of ChadOx adenoviruses expressing spike S proteins as an alternative to the full-swing SARS-CoV-2).

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s