My PhD centres on the crosstalk between cells that allows glioblastoma to invade. A lot of the research I have focused on has revolved around how tumour cells actively recruit their neighbours and reprogram the microenvironment to facilitate their growth. A broader biological principle centres on this research: cells do not act in isolation, but can alter the behaviour of an entire tissue by changing the state of their environment.
I recently stumbled across this paper, which offered a very new perspective on cellular crosstalk in a completely different disease context! This paper looks at a striking form of communication whereby signals generated in the periphery trigger a cascade that reprogrammes cells in the CNS into a protective antiviral state. What perhaps caught my attention was that the cells acting as a key relay into the brain are not immune cells, but endothelial cells!
West Nile virus (WNV) is a mosquito-borne virus that, in a small subset of infected individuals, invades the central nervous system (CNS) and causes life-threatening encephalitis. While much of the field has focused on how the immune system responds once WNV enters the brain, much less is known about what happens beforehand, and whether the CNS is primed for the virus’s arrival
The authors began by comparing two infection routes in mice: injection into the footpad, modelling the natural peripheral route, and direct intracranial injection, bypassing the periphery altogether. Unsurprisingly, placing WNV directly into the brain produced more severe disease, even at a lower dose, so this comparison alone tells us little about whether peripheral sensing is protective. Interestingly, observing the brain tissue one day post footpad infection, the authors found strong antiviral transcriptional signatures, including interferon-stimulated genes. The brain seemed to be responding to something before the virus arrived?
Separating viral sensing from viral replication
During an infection, viral sensing and viral replication occur together, making it difficult to understand which is responsible for priming the brain. To investigate this, the authors used poly(I:C), a synthetic form of double-stranded RNA that mimics a viral danger signal but cannot replicate.
After injecting a panel of pathogen-associated molecular patterns, or PAMPs, (e.g PAM3CSK4, LPS, Flagellin, poly(I:C) etc.) into the footpad, the mice were then challenged with WNV into the brain for 24 hours later. A number of significant changes were seen in Poly(I:C): collectively there were lower levels of infectious virus and viral antigen in the brain, less inflammatory dysregulation, and interestingly, and a survival rate of 41%. Protection was found to be the strongest when poly(I:C) was given 24h before WNV, weaker administered at the same time, and absent if administered after WNV.
It was found that the effect was prophylactic, not therapeutic: it prepared the CNS in advance rather than treating an established infection. And when the authors depleted the NK cells or inflammatory monocytes, the effect was not eliminated, suggesting it was not the central mechanism.
Following the signal from the periphery
Was poly(I:C) therefore reaching the brain? Several tracing experiments suggested it was not; the material remained almost entirely within the footpad, and the draining lymph node. A downstream signal of peripheral sensing therefore was hypothesised. After profiling cytokines in the blood, several candidates were identified. Only blocking IFNAR1 blocked the antiviral response in the CNS, therefore suggesting that Poly(I:C) preferentially increased circulating IFN-a to act as a warning signal from the periphery to the brain.
Finding the receiver
Only half of the puzzle is actually solved. The central question remains as to which cells in the brain detect the IFN-a. Single-nucleus RNA-seq on brain tissue 2 and 16 hours after systemic IFN-a administration showed a number of different cell population responses.
At 2 hours, brain microvascular endothelial cells (BMECs) showed the strongest induction of the protective antiviral gene signature of any cell type profiled, whereas by 16 hours, this signature had spread more broadly across the CNS, particularly among microglia. The authors validated this by sequencing the RNA of isolated BMECs after footpad poly(I:C) treatment, and found that they mounted a stronger antiviral transcriptional response than whole brain tissue, driven by canonical IFN-associated transcription factors (IRF9, STAT1 and STAT2). This clearly showed that BMECs were early responders.
Proving that endothelial signalling is required
However, this response did not necessarily mean it was protective. To answer this, the authors need to remove IFNAR1 specifically from BMECs. Using an engineered AAV with the BI30 capsid, which preferentially targets endothelial cells in the brain, the authors delivered Cre recombinase to conditionally ablate IFNAR1 in BMECs. Circulating IFN-α remained comparable, indicating that systemic interferon production was unaffected.
Circulating IFN-α remained normal in these mice, confirming interferon production itself was unaffected, but the CNS antiviral response was substantially blunted, and poly(I:C) no longer protected against WNV. This showed that IFNAR1 signalling in BMECs was not simply associated with protection, but necessary for it. In other words, the blood–brain barrier is doing more than acting as a passive wall; it is actively sensing and relaying immune signals from the circulation.
What happens beyond the vessel wall?
But once BMECs sense the circulating interferon signal, how is that warning relayed deeper into the CNS?
Their single-nucleus RNA-sequencing data showed that, by 16 hours, the antiviral gene signature had become much more prominent in microglia. Further computational analysis (CellChat), predicted increased signalling from microglia to neurons through pathways associated with neuronal maintenance and survival, including neuregulin-ErbB signalling, known to be associated with protecting neurons from cell death. Microglia had also upregulated these ligands too.
This leads to a possible cascade: endothelial cells sense the signal first, and microglia may participate in a downstream neuroprotective relay. This part of the story however is inferred computationally, rather than tested functionally.
A broader antiviral warning system?
Having established a mechanism, the authors turned to a final question: was this protective circuit unique to WNV, or did it reflect a more general antiviral strategy?
The authors then tested whether this protection extended beyond WNV. Poly(I:C) pretreatment improved survival against Powassan virus, Sindbis virus and herpes simplex virus 1—spanning multiple viral families and both RNA and DNA genomes.
This breadth suggests that peripheral poly(I:C) establishes a CNS antiviral state that extends beyond WNV. Whether BMEC IFNAR1 signalling is equally necessary for protection against each of these viruses was not directly tested.
What remains unanswered?
A few limitations worth noting. Much of the mechanistic work relied on direct intracranial infection, which controls timing but doesn’t reflect how virus naturally reaches the brain. Tracing experiments couldn’t fully exclude a small amount of poly(I:C) acting directly on BMECs. The downstream signal connecting endothelial cells to the rest of the CNS also remains unclear, and the microglia to neuron pathway still needs functional testing. Finally, the work used the older NY99 WNV strain rather than currently circulating lineages, so how well this generalises to modern strains is untested.
Why this paper caught my attention…
In neuro-oncology, endothelial cells are usually discussed in terms of disruption: tumour-induced leakiness, abnormal angiogenesis, vessels co-opted to fuel tumour growth. Here, the same cell type does the opposite, acting as an active sensor of the body’s immune state rather than passive vasculature.
I find it fascinating how the underlying biology, specifically cells altering the state of their neighbours, can be hijacked to drive disease in context, and mobilised to defend it in another.