Several viruses can cross barriers, infect the central nervous system, and ultimately lead to life-threatening diseases such as encephalitis and meningitis. Our lab has long been interested in studying these neurotropic flaviviruses and their interplay with host response through type I interferon pathway in mouse primary cells and mouse brain. We, as well as many others, had characterized virus infection in the brain by looking through immunostaining of infected brain sections or viral nucleic amplification from dissected brain tissue. However, the spatial information is deduced so much that questions like how the virus spreads in the brain, what neurocircuits are infected, and how this infection might have contributed to brain pathology remain largely unanswered in the field of virology.

Tissue clearing is a recently emerging technology to make brain become transparent. It is widely used in the field of neuroscience to visualize whole rodent brains. Applying this method to our work, we can immunostain viral proteins and visualize their distribution in the whole brain using 3D optical imaging techniques, such as optical projection tomography (OPT) and light-sheet fluorescence microscopy (LSFM). This enables us to ‘see’ virus distribution in the whole brain and determine which parts of the brain are susceptible to the infection.

However, the limitation of these techniques is the limited anatomical information since the anatomical outline is only acquired based on tissue autofluorescence. As the brain is highly compartmentalized and can be subdivided into hundreds of areas, without the trained eyes of neuroscientists (which we are not), we are unable to specify the exact areas or neurocircuits that are susceptible to the infection. To overcome this limitation, we teamed up with neuroscientists who suggested a brilliant idea to overlay our OPT brain image on top of the magnetic resonance imaging (MRI) brain image. MRI provides impeccable details in the tissue, making it widely used for anatomical identification of the brain. Initially, we attempted to overlay our OPT brains on the existing MRI templates, which usually image the brain with an intact skull. We found that the tissue process for optical clearing causes significant shrinkage, leading to misalignment of our brain to the existing templates. Therefore, we decided to create our own MRI template customized for optical clearing brains, called the OCUM brain template, and create the atlas to annotate different anatomical areas of the mouse brain (Willekens et al, BioRxiv 2022). At this point, we were able to map virus infection in the entire brain and identify and quantify the affecting areas or neurocircuits with anatomical precision.

After the methods has been established, we were ready to dive deep into virology. As we are interested in studying neurotropic flavivirus and its interplay with host defense, we used Langat virus (LGTV), a Biosafety level-2 model for a Biosafety level 4 Tick-borne encephalitis virus to investigate how the virus spread in the brain of a normal (wildtype; WT) and innate immune compromised (lacking type I interferon response; Ifnar^-/-) mice. We observed specific localization of the virus to the grey matter of sensory neurons in WT, and to our surprise, the virus did not spread all over the brain of ifnar^-/- mice which lack one of the most fundamental antiviral defense mechanisms. Instead, we observed the expansion into the white matter of the those neurocircuits seen in WT, rostral migratory stream, meninges, and choroid plexus. This observation indicates that type I interferon (IFN-I) plays an essential role in restricting virus infection in these areas. In the grey matter of some areas, especially in the auditory circuit, there were higher infections in the WT than Ifnar^-/-, suggesting that this pattern may not be due to the loss of the defense mechanism in general, but might be the differences in the susceptible target cells.

We then further investigated the susceptible cell type (aka. the tropism of infection) using confocal and electron microscopy. We found that choroid plexus and meninges were infected only in ifnar^-/- mice and astrocytes were refractory in both genotypes, indicating that IFN-I plays an important role in restricting virus infection in choroid plexus and meninges while astrocytes use other mechanisms to defend against the infection. To our surprise, we observed the susceptible cells in the cerebral cortex shifted from neurons in WT to microglia in Ifnar^-/- mice. It is worth noting that this alteration of tropism has not been seen in our previous in vitro infection of primary neurons and microglia monocultures. 

To unravel the underlying mechanisms of this tropism shift, we employed single-nucleus sequencing of the cerebral cortex of the infected and uninfected WT and Ifnar^-/- mice. We initially investigated the fundamental differences between these two genotypes in an absence of infection and found no significant differences in gene expression, indicating that the baseline expression of these two genotypes is unlikely contributed to the tropism shift. When analyzing the infected brains, we did not anticipate being able to differentiate the infected population from nuclei isolations since the nature of LGTV replication is on the endoplasmic reticulum (ER) membrane. However, we were thrilled to observe the enrichment of viral RNA in neurons and pericytes of WT and microglia and choroid plexus of ifnar^-/- mice. Our result is unlikely due to RNA contamination but likely due to incomplete removal of the ER as the enrichment of LGTV reads corresponded with the observation from confocal microscopy, emphasizing that our result has been cross-validated between different experimental platforms. To our knowledge, this is the first study to report the differentiation of infected and bystander nuclei from viruses that replicate in the cytoplasm.

Regarding the molecular mechanisms of microglia susceptibility, we found the upregulation of IFN-I response to infection in WT mice as anticipated. Additionally, we identified the infiltration of peripheral macrophages, activation of residential microglia, and the induction of IFN-II from infiltrating CD8+ NK cells in response to virus infection in WT mice. On the other hand, we observed no infiltration of CD8+ NK cells and minimal infiltration of macrophage in Ifnar^-/-. Microglia in Ifnar^-/- mice remained inactivated in the absence of IFN-I and II responses, and only low IFN-I independent responses was detected. Collectively, these infiltrating cells create an inflammatory milieu in response to the virus and prevent the microglia cells from becoming infected in WT, whereas inactivated and unresponsive microglia in the Ifnar^-/- brain provide an easy target for virus infection.

Taken together, these collective efforts and marrying of techniques among interdisciplinary fields, transcriptomics, imaging, neurosciences, and virology, coupled with several surprises, the highs, and the lows, have shaped our study to provide unprecedented insight into virus infection and pathogenesis with exquisite details of spatial information in the brain. Our whole tissue imaging to transcriptomics multimodalities can serve as a platform for other pathogen-brain infection studies, ultimately advancing our comprehension of infection and brain pathogenesis.

Reference:

Willekens, S. M. A. et al. An MR-based brain template and atlas for optical projection tomography and light sheet fluorescence microscopy. BioRxiv., https://doi.org/10.1101/2022.11.14.516420 (2022).