I have been recently approached on social media to discuss about the recent study published by Anderson and colleagues in Annals of Neurology (https://onlinelibrary.wiley.com/doi/abs/10.1002/ana.25610), in which the authors reported the presence of residing CD8 cytotoxic T-cells in the perivascular space of brain samples from ASD patients. I wrote about it, and gave my first impressions about reading the study. After such discussion, I realized that I have an interesting review that worth being shared here on my blog.
Here is the summary of my review of this paper as I wrote it on social media. I made some corrections (mostly spelling and grammar), as well as some changes in the writing style (I wrote these comments on the spur of the moment, as I went through the paper). Also due to copyrights, I will not show any figures and tables from the study.
About the authors: The first author is Dr. Marcello DiStasio (MD/PhD) and the senior author is Dr. Matthew P. Anderson, MD/PhD. He is a clinical faculty of Harvard Medical School and member of the Department of Neurology and Pathology a Beth Israel Deaconess Medical Center, with an affiliation with Boston Children’s Hospital Intellectual and Developmental Disabilities Research Center. So we have some high profile and experts in neurodevelopment disabilities.
About the journal: The journal is published under Wiley, and is the official journal of the American Neurological Association and the Child Neurological Society. It has an impact factor of 9.49 and is in the top quartiles of neurology and neurosciences journal. Therefore, we have a study published in a very good journal that is relevant to the topic.
About the study design: The study is mostly observational and relies mostly on histological methods (tissue sections followed by staining using chemical dyes and/or antibodies targeting specific proteins). Tissue samples are freshly isolated from postmortem patients (which is a big plus compared to formalin-fixed samples, and opens up the ability to perform protein and RNA analysis if the samples are immediately treated for extraction). The sample size is pretty decent (N=25-30) with a large age spectrum and various types of ASD represented. First interesting to note, 50% of the ASD brain (N=25, not bad) have a history of seizures. Less than 30% of control brains (N=30) have history of seizures. Important thing to consider as a comorbidities and evenutally as a cofounding factor.
About the results: This is my summary of the different figures. By copyright concerns I am not showing the actual figures, but you can overlap my comments to the figures as I have separated them into sections.
Figure 1: Me being cranky, I wanted to see control brain pictures, but not avail. Also with immuno pictures, you have to be super-precautious because there is a high risk of cherry-picking a brain slice, and claim it is representative of all brain. Nevertheless lets discuss it here. On the left panel, we have an H&E staining. Pretty much a vanilla staining. Now, if you want to show a certain protein, you can do it with the right antibody and staining (DAB/peroxidase stain). It appears as the dark brown color. We can see a strong GFAP staining around the vasculature (hollow structure). Astrocytes usually line up blood vessels by forming end-feet process. S100B and ALDH1L1 are pretty standard proteins for astrocytes. However seeing GFAP expression in human astrocytes means these cells are stressed out and are reactive. This is what the quantitative bar graph is telling us. We can also see some CD8 in this perivascular space. These are the cytotoxic T-cells. I wonder what they are doing there, as they can cross the BBB only during brain injury, as microglial cells will äctivate” these endothelial cells and allow white blood cells to adhere on their surface and cross the BBB into a complicated tango dance.
Figure 2: Here there is an attempt of some matrix correlation. GFAP versus CD8 cells. And we can see there is some (expected) correlation between these two (see linear regression and R2) with ASD brains have higher rates of both compared to controls. Interestingly we see similar pattern between ASD that are genetic versus the idiopathic.
Figure 3: digs in more about the immune cells and some correlations (although the scatter of the ASD brains is less convincing here). Overall it seems we have a higher number of lymphocytes in the ASD brain compared to control, both in the white matter (WM, this is where our cables go through) and grey matter (GM, this is where our processing units are) and lepomeningeal (LM, this is our brain surface protective skin, basically the meninges, pours the CSF into the veinous blood). What seems interesting is that at young age we have lymphocytes sitting in our perivascular space, doing nothing(?) and decrease as we age (thats interesting for the non-neuroimmunologist that I am). However, these number at best slightly increase in ASD brain as we age (and I guess not convincingly enough, otherwise the authors would have reported the R2 value), or at least remain the same. With the exception of the medulla (brain stem) most blood vessels show a higher number in ASD brains versus control brains. Interestingly, cortical blood vessels being the predominant population harboring such feature. Very few NK cells to be honest on the panel.
Figure 4: Again this one makes me cringe a bit as a reviewer, because the author do not show the controls data. Yeah, I am pissed. Anyway. Again, perivascular space. Again immune cells highly present (CD3+), CD8 lymphocytes being the predominant type of T cells, in contrast not many CD4 lymphocytes (usually the T helpers, but my immunology is outdated for 20 years at least). Granzyme staining denotes the presence of natural killer (NK) cells. CD20 is a marker for B cells. What is interesting is mostly not much B-cells in either ASD and control brains, the graph more like a refried version of what we already know (higher number of immune cells and CD8 cells).
Figure 5: is a bit of a useless graph, I don’t see anything that brings us more information than before.
Figure 6: It shows us a Masson trichrome staining. It is a chemical (histological) staining aimed to make collagen fibers visible under a certain color compared to other tissues. Collagens exist in different forms (based on their alpha-fibrils) but the one you expect to see around the vasculature is Collagen Type IV (COL4A1). This one forms a basement membrane (BM) that is like a net around blood vessels. Think about standing on a trampoline. Thats it. Collagen IV forms the net that supports the BBB. Normally, you would expect the BM to be thinned out, this is the case after stroke as cells secrete protein-degrading enzymes (like matrix metalloproteinases MMP2 and MMP9) that will break it down into pieces. But here it is thickened and hhad a bigger thickness than normally. Why so? What does not mean? I dont know. The only time I have seen such things was in transgenic mice overexpression erythropoietin (EPO). These mice had a such high hematocrit that would make blood be like Aunt Jemima corn syrup. If I remember, my former PhD adviser had a collaboration with an electron microscopist and observed similar thickening. Why this is happening is a good question that is likely the next study.
Figure 7: See Figure 5 comment.
Overall thoughts, limitations and outlook: Overall, it is very interesting study, that has some methodological limitations to be noted. First, we are missing any information about the BBB integrity in general, as I wish the authors would have shown some immunofluorescence to compare changes in tight junction (TJ) proteins expression (claudin-5, occludin) in blood vessels and assess differences in TJ strands. The second problem is the lack of information about microglia activation (that would be done by Iba1 staining. The authors noted they performed it, observed a higher expression in ASD brain but decided not to show the data) and more importantly the status of endothelial cell activation (using ICAM1, VCAM1 staining). Are these residing CD8 T cells freshly migrated or just being duck sitting for a while? Is the inflammation status in the brain (and the BBB) ON or OFF at the time of autopsy? What about pro-inflammatory cytokines levels between ASD and control patients? As usual, studies like that opens much more questions that it answers. The second problem is that the study by itself fairly descriptive and observational. It would be interesting how this would compare to clinical findings done in patients. Do we see flares on the MRI indicating of the BBB opening? How does that data compares to patients with MS?
I would also be very careful on making any claim about a “leaky” or “down-regulated” BBB at this point. There is no data about the barrier integrity (TJ complexes integrity) or assessment of the barrier function (for example an MRI scan with gadolinium as contrasting agent) to support the claim. I have seen this claim floating around in non peer-reviewed articles (including in The Scientist), as a BBB expert I would not jump into that conclusion quickly until I see real data on that
This is the type of interesting study, because it opens 100 questions that incentivize to further look down the road. I hope that the authors were able to prepare RNA and protein samples for transcriptome analysis (RNAseq) and proteome analysis (2D-electrophoresis coupled with MALDI) that could help us learn more.