Proteins from two strains of tick-borne encephalitis virus showed different patterns when tested against JAM-A, a human tight-junction protein associated with the blood-brain barrier. NS1 from the high-pathogenicity Hypr strain bound immobilized JAM-A in an ELISA, while NS1 from the low-pathogenicity Vs strain showed no binding under the tested conditions. In cell experiments, Hypr NS1 also showed greater overlap with JAM-A than Vs NS1. The pattern suggests a strain-specific interaction worth investigating, but it does not show that the binding disrupts the blood-brain barrier or causes neuroinvasion.
How the strains were compared
The researchers compared NS1 from Hypr and Vs using AF3 multimer modeling and in vitro binding assays. Their question was whether the high- and low-pathogenicity strains interacted differently with blood-brain barrier tight-junction proteins. The work also used cell-line imaging to examine overlap between NS1 and JAM-A signals.
Sequence comparison found 22 amino-acid differences between Hypr and Vs NS1. Nine were concentrated in the wing domain and 13 in the distal beta-ladder domain; the study reported no changes in the beta-roll or connector regions. Those locations are a clue to where strain-specific behavior might arise, but the experiments did not establish that any of the substitutions caused the functional differences.
Modeling pointed to a possible contact
AF3 modeling predicted a confident Hypr NS1-JAM-A interface, with a reported interface score of 0.7. The predicted contact involved the membrane-distal Ig-like domain of JAM-A and a hydrophobic groove on the NS1 dimer. Two residues, S142 and D236, were predicted to help form the interface, and both are among the 22 positions that differ between the strains.
Separate AF3 predictions suggested a strain-selective pattern among claudins, another group of tight-junction proteins: Vs NS1 with Claudin-5 at an interface score of 0.71, and Hypr NS1 with Claudin-10 at 0.7. Those results are computational predictions, not experimental validation of binding in vivo.
The laboratory test separated the strains
To test the JAM-A prediction, the researchers exposed immobilized JAM-A to serial dilutions of recombinant NS1. Hypr NS1 showed concentration-dependent binding. A one-site binding model produced apparent dissociation constants of 0.81 nanomolar at a starting concentration of 0.5 micrograms per milliliter, 7.98 nanomolar at 1 microgram per milliliter, and 310.38 nanomolar at 2 micrograms per milliliter. Vs NS1 showed no binding across the tested concentrations.
That spread matters when interpreting the assay. The reported apparent affinity changed with the starting NS1 concentration and may reflect factors specific to the setup, including the way the protein was immobilized and how the signal was generated. In a separate ELISA comparison, neither strain showed binding to ZO-1 at 0.5, 1, or 2 micrograms per milliliter, so no apparent dissociation constant was determined for that target.
Cell imaging found the same pattern
The protein-level pattern was echoed in cell imaging. Researchers analyzed 230 transfected cells from two independent experiments, using Manders' overlap coefficients to quantify how much NS1 signal overlapped JAM-A signal and how much JAM-A overlapped NS1. Hypr NS1 showed greater JAM-A colocalization than Vs NS1 in both A549 and hCMEC/D3 cells. Reported colocalization was 50 to 60 percent in A549 and 70 to 80 percent in hCMEC/D3, with the difference between the two cell lines reported at p = 0.0008.
Baseline TEER was higher in hCMEC/D3 monolayers than in A549 monolayers, at 83 Ω·cm² and 37 Ω·cm² respectively. But this was a model characterization: the study did not report an NS1-induced TEER change. Colocalization also cannot by itself establish direct binding or blood-brain barrier disruption in vivo.
A possible clue, not a demonstrated mechanism
The authors interpret the Hypr NS1-JAM-A result as a possible contributor to neuroinvasion, while emphasizing that its role remains unresolved. They suggest that the Vs strain may use alternative routes, an interpretation that requires further work. The paper therefore offers a lead about strain-specific engagement of junction proteins, not a demonstrated mechanism of disease.
Further in vivo validation is needed, as are experiments with native or glycosylated claudin proteins and tests of whether the 22 sequence differences alter tight-junction protein engagement. The current evidence is limited to the tested proteins, cell lines and assay conditions.
Paper data and sources
Original title: Interactions between Tick-Borne Encephalitis Virus Nonstructural Protein 1 and Blood–Brain Barrier Tight Junction Proteins: Potential Clues to Strain-Specific Neuropathogenicity
Authors: Chaopreecha J, Liburd J, Khalid S et al.
Journal/Repository: ACS omega
Status: Peer-reviewed
First online: 2026-10-09
DOI: Not available
Original paper · Full text