Ethics statement

Animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at Georgia State University under protocol A25015 and conducted at ABSL-1.

Pharmacokinetics (PK)

Female 4- to 5-weeks-old Syrian Golden Hamsters were obtained from Charles River and drug administration was performed via oral gavage (1 ml per 100 g body weight). For the single-dose PK study, subjects (n = 8 per group) were dosed once with either 2 or 10 mg/kg of 4′-FIU (EIDD-2749, provided by the Emory Institute of Drug Development). Five consecutive daily doses were administered in the multi-dose PK study. Blood was collected at 8 time points (between 30 minutes to 24hours) after dosing in both studies and tissues (brain and lung) from 4 subjects per group were collected at 3 and 24 hours post dosing. Calibration curve ranges were established using plasma and tissue blanks and ranged from 10 to 10,000 ng/ml for 4’-FlU in plasma, and 50–20,000 ng/g of tissue for 4’-FlU-TP. Quality control samples (3 concentrations) were run before and after analyzing a sample set using plasma and tissue blanks. Uridine 5’-triphosphate (UTP) was monitored as a control for sample integrity in tissues. Measured levels of UTP were approximately 3-times lower in the single-dose hamster groups compared to the repeated-dose and vehicle-dosed hamster groups.

Ethics for studies with infectious agents

Animal experiments involving infectious NiV were conducted at animal biosafety level 4 (ABSL-4) at The University of Texas Medical Branch under The Institutional Animal Care and Use Committee (IACUC) protocol 2205033 and performed in accredited facilities, ensuring compliance with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC).

Virus and cells

A recombinant Nipah virus that is a molecular copy of the SPB200401066 isolate (accession no. AY988601) from the 2004 outbreak in Rajbari district in Bangladesh [31], and encoding Gaussia luciferase (Gluc) and enhanced GFP reporter genes, was generated (rNiV-B Gluc) using the same strategy as the recombinant Malaysia prototype strain previously rescued [32] and sequenced for identity confirmation. The rNiV-B Gluc was then propagated in Vero E6 cells (CRL1586, ATCC, Gaithersburg, Maryland) to make a stock and was titrated by plaque assay technique using Vero (CCL81, ATCC, Gaithersburg, Maryland) [32]. The rNiV-B Gluc harboring mutations identified in this study on N, P, or L gene were generated from rNiV-B Gluc backbone by site-directed mutagenesis through in-fusion cloning and rescued by reverse genetics as previously described [32]. Whole plasmid sequences were verified (Plasmidsaurus) prior rescue and virus stocks were made from the first passage in Vero cells. Wild-type NiV-B (SPB200401066 isolate, accession no. AY988601) was also used in this study and stock was similarly prepared. For in vitro antiviral testing, both wild-type and recombinant viruses were used, while only rNiV-B Gluc was utilized in in vivo studies. All infectious work was conducted at BSL-4 at UTMB.

Assessing 4′-FIU efficacy in vitro by plaque reduction assay

Vero cells (CCL81, ATCC, Gaithersburg, Maryland) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and seeded at 2.5 × 10⁵ cells/well in 12-well plate format the day prior to infection. Virus inoculation was done with 100μl of wild-type NiV-B for 1 hour at 37°C, after which 1 mL of overlay was added (1% methylcellulose: 2X MEM, containing 2% FBS). Note that virus dose was adjusted to yield approximately 40 plaques. Plates were then incubated for 2 days at 37°C in a 5% CO₂ incubator to allow plaque formation. Cells were then fixed with 10% buffered formalin with crystal violet for 30 minutes to visualize plaques. 4′-FIU was added during infection and supplemented into the overlay when assessing drug efficacy. The percent plaque reduction was calculated using the virus control (no drug in overlay) as reference. The effective concentration required to reduce plaques by 50% (EC₅₀) was determined by plotting percent inhibition against 4′-FIU concentrations.

Assessing 4′-FIU efficacy in vitro through a luminescence reporter assay

Vero cells (CCL81, ATCC, Gaithersburg, Maryland) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and seeded at 10 x10³ cells/well in a 96-well plate the day prior to infection. A low infective dose of rNiV-B Gluc (MOI 0.001) was then added along with 4′-FIU where appropriate (total 100μl) and plates were incubated for 2 days at 37°C in a 5% CO₂ incubator. Luminescence was detected after adding 50μl of Pierce Gaussia Luciferase Glow substrate (Pierce, Rockford, Illinois) per reaction using a Cytation 7 reader (Agilent, Santa Clara, California) at BSL-4. Percentage replication was calculated using the virus control (no drug in medium, biological triplicates for S1 Fig) as reference. This assay was also used to concomitantly compare viral replication of parental rNiV-B Gluc and 3 rNiV-B Gluc harboring a single mutation on the N, P or L gene on distinct 96-well plates using biological quadruplicates (8 biological replicates per plates for each virus and mock controls). The effective concentration required to reduce the luminescence signal by 50% (EC₅₀) was determined by plotting percent inhibition against 4′-FIU concentrations.

Assessing in vivo efficacy of 4′-FIU in the Syrian hamster model for rNiV-B Gluc

Male and female 5- to 6-week-old Syrian Golden Hamsters (Envigo, Indianapolis, Indiana) were evenly divided into 2 cohorts of six groups each (n = 5 male and 5 female per group). Hamsters were anesthetized using isoflurane prior to each manipulation, including rNiV-B Gluc challenge, administration of 4′-FIU, or sampling (blood and swabs). A challenge dose of 10⁵ particle forming units (PFU) per animal was administered intraperitoneally (IP, 100μl per animal) on study day 0 where appropriate. Backtitration was performed to determine delivered infective dose. The 4′-FIU treatment was 10 or 25 mg/kg and administered once daily orally (PO, 100μl per animal) starting either immediately or 8–10 hours (Delayed) post-challenge. Animals were then dosed with 4′-FIU daily for a total of 7–28 days, depending on the cohort. Matching vehicle formulation was made with 10mM sodium citrate buffer, 0.5% Tween 80 and administered using the same method and treatment schedule in mock groups.

Two treatment cohorts were run, a Short treatment Protocol (n = 60) and an Extended treatment Protocol (n = 60). In cohort 1 study (short treatment protocol), two control (virus and mock) groups received the vehicle, with only the virus group challenged with virus. The remaining four infected groups received 10 or 25 mg/kg 4′-FIU for 7 consecutive days either immediately or 8–10 hours (delayed) post-challenge. Based on the results from this cohort, the dose of drug and duration of treatment was adjusted in the cohort 2 study (extended treatment protocol). Specifically, four groups received virus and 10 mg/kg 4′-FIU either immediately or 8–10 hours (delayed) post-challenge once daily but for a total of 28 days or for 21 consecutive days. The two last groups received the vehicle but only one was challenged with virus, similar to the cohort 1 study.

Hamsters were monitored for weight loss daily for the first 10 days or during peak disease, and then every 2^nd or 3^rd day until end of study. Weight change was calculated from baseline weight at day 0. Daily health checks were conducted until study termination (up to day 28 for cohort 1 study, and days 34 and 51 for cohort 2 study, respectively) for monitoring any clinical signs of disease. Note that animals were monitored for a minimum for 14 days after termination of treatment. Animals were scored 1 if appearing healthy, 2 if lethargic, 3 if scoring 2 plus exhibiting ruffled fur, hunched posture, orbital tightening, rapid or labored respiration, or 15% weight loss. Animals showing 20% or more weight loss, paralysis, ataxia, head tilt, or seizures (score of 4) were immediately euthanized, and terminal sampling (blood and tissues) was performed.

Blood and swabs collection, and luminescence signal detection

Whole blood was collected in BD vacutainer serum collection tubes (BD, Franklin Lakes, New Jersey) to recover serum after centrifugation (8,000 g for 10 minutes). Oral and nasal swabs were collected using, respectively, Puritan PurFlock (REF 25–3317-U, Puritan Medical Products Company, Guilford, Maine) and Microbrush applicators (REF MUO400, Young innovations, Grafton, Wisconsin) and then stored in screw top vials containing 250ul PBS (Gibco, Grand Island, New York). Luminescence was detected with a Cytation 7 reader (Agilent, Santa Clara, California) using 20µl of sample and 50µl of Pierce Gaussia Luciferase Glow substrate (Pierce, Rockford, Illinois) per reaction in 96-well plates.

Tissue processing and virus titration

Brain, lung, and liver tissues were collected after euthanasia and were homogenized in PBS (Gibco, Grand Island, New York) using a TissueLyser (Qiagen, Hilden, Germany) and stainless-steel beads, as previously described [33]. The liquid phase was recovered and then processed, along with terminal sera, for titration by plaque assay on Vero cells as previously described [33].

Luminescence Reduction Neutralization Test (LRNT)

Terminal sera from extended treatment protocol cohort were heat inactivated at 56°C for 30min. Starting from a 5-fold dilution, sera were 7 times 2- and 3-fold serially diluted in microplate format using DMEM supplemented with 2% FBS. An infective dose (MOI of 0.0005) of rNiV-B Gluc was added (1:1 volume) where appropriate, and plates were incubated at 37°C for 1h under periodic agitation on plate shaker (400rpm). Sera-virus mixtures were then transferred into microplates pre-seeded with Vero cells (CCL81; 20,000 cells/well; DMEM with 2% FBS) and further incubated for 48h at 37°C in a 5% CO₂ incubator. Luminescence signal was detected as described in the above section “Assessing 4′-FIU efficacy in vitro through a luminescence reporter assay”. Note that an automated reagent dispenser (ThermoFisher scientific, St. Louis, Missouri) was used to accurately dispense virus inoculum, Vero cells suspension, and Gluc substrate. Each sera from 4′-FIU-treated animals was evaluated in duplicates in distinct microplates. Each plate included virus (8 replicates), mock (4 replicates), and neutralization (4 replicates, mouse anti-Nipah G [nAH1.3], Absolute Antibody) controls. Values were normalized to virus and mock controls for each plate, and percentage replication was calculated. Curve fitting was done from plotting average percent inhibition against 4′-FIU concentrations and the effective concentration required to reduce the luminescence signal by 50% (LRNT₅₀) was determined.

Tissue histology

Brain, lung, and liver tissues were collected after euthanasia for histopathology ad immunohistochemistry (IHC) analyses. Briefly, tissues were first fixed in 10% neutral buffered formalin for 24 hours and for an additional 48 hours with fresh buffered formalin in a clean container in BSL-4 following approved protocols before removal and processing of specimens under BSL-2 conditions. Tissues were embedded with paraffin prior sectioning (5 µm thickness). Sections were stained with hematoxylin and eosin for histopathological analyses. Others were processed for detection of NiV antigen using an anti-nucleocapsid antibody provided by Dr. C. Broder (Uniformed Services University, Bethesda, Maryland) or commercially available (HL1436, ThermoFisher Scientific, St. Louis, Missouri), an HRP-conjugated secondary antibody (ab7090, Abcam, Eugene, Oregon), and chromogenic substrate (NovaRED, VectorLabs, Newark, California). Review of slides and grading of lesions were performed by a veterinary pathologist.

Multiplex immunoassay

Sera received a gamma irradiation treatment (5Mrad dose) on dry ice to inactivate virus and were then handled at BSL-2. A Hamster Cytokine Panel 1 Luminex 9-plex (Ampersand Biosciences, Lake Clear, New York) was used to detect interleukin (IL)-10, IL-2, IL-4, IL-6, interferon (IFN)-γ, monocyte chemoattractant protein 1 (MCP-1), macrophage inflammatory proteins (MIP)-1α, tumor necrosis factor alpha (TNFα), and vascular endothelial growth factor (VEGF) in sera following manufacturer recommendations. Samples were then run on a Bio-Plex 200 system and concentration of each analyte in sera was determined by plotting data against the corresponding standard curve.

Sequencing of rNiV-B Gluc genome from stock and tissues

Virus stock and tissue specimens were homogenized in respectively TRIzol LS or TRIzol Reagent (Invitrogen, Carlsbad, California), as previously described [33] and RNA was extracted using Direct-zol RNA Miniprep kits (Zymo Research, Irvine, California). Detection of virus genome by quantitative RT-PCR assay was then performed using QuantiFast RT-PCR mix (Qiagen, Germantown, Maryland) and a set of primers (IDT, San Jose, California) and probe (TIB MOLBIOL, Adelphia, New Jersey) targeting the P gene as previously described [34]. Virus-positive tissues, mock controls, and virus stock were then submitted for viral whole-genome sequencing. Libraries were generated by converting the extracted RNA to double-stranded cDNA, which was subsequently fragmented, end-repaired, indexed, purified, and amplified using QIAseq xHYB Microbial Hyb&Lib Kit A (Qiagen, Germantown, Maryland). Library fragment sizes were estimated using the TapeStation 4200 D1000 (Agilent, Santa Clara, California), and concentrations were measured by Qubit 4 Fluorometer with the Qubit dsDNA HS Assay Kit (Invitrogen, Carlsbad, California). Lung, mock-infected, and virus stock libraries were shotgun sequenced using pre-capture libraries on Illumina NextSeq 2000 using a 2x150bp read format. Due to the high cellular content of brain specimens, these libraries were sequenced after affinity enrichment of Nipah virus sequences by hybridization capture. Hybridization capture was performed using the QIAseq xHYB Microbial Hyb&Lib Kit A with a custom Qiagen hybridization panel containing probes synthesized based on Nipah virus Malaysia (AF212302) [35] and Bangladesh (AY988601) [31] containing GLuc-P2A-GFP reporters [32]. The enriched libraries were further amplified and purified via bead clean-up. Library fragment sizes were estimated as described above and sequenced on the same instruments as lung samples. Raw reads were trimmed and quality filtered using fastp (v0.23.4) [36]. The QC-filtered and trimmed reads from virus stock were randomly subsampled to 1 million pairs using seqtk (v1.4) (https://github.com/lh3/seqtk). The consensus sequence of virus stock was generated by majority voting using Geneious Prime with NiV-B genome [31] containing GLuc-P2A-GFP reporters. Variants were identified using RAVA workflow (https://github.com/greninger-lab/RAVA_Pipeline/tree/publications_2025_01_01) [36,37]. The virus stock consensus sequence was used as the reference for variant calling. Shotgun libraries were analyzed in paired-end mode, whereas capture libraries were analyzed in single-end mode. Raw sequencing data is publicly available in NCBI BioProject PRJNA1214874.

Assessing viral fitness of rNiV-B Gluc viruses in the brain

Human primary astrocytes (HA, #1800, ScienCell, Carlsbad, California) were cultured using the recommended astrocyte medium and protocol, per manufacturer instructions. HA were seeded at 10⁵ cells/well in 24-well plate format the day prior to infection. Virus inoculation (MOI 0.01) was done with 100μl of inoculum for 1 hour at 37°C. Cells were rinsed with 200μl PBS twice prior adding 500μl of HA medium for the assay. Supernatants were collected and titrated by plaque assay as described above. Kinetic was performed with biological triplicates.

Statistical analysis

Log-rank test was used for comparison of survival curves and Chi-square test was used to compare mortality rates. One-way ANOVA followed by Tukey’s multiple comparisons test was used to compare weight loss, luminescence signal, and immune response between groups. Kruskal-Wallis test followed by Dunn’s multiple comparisons test was used when data were not normally distributed. Point-biserial correlation was used to determine the relationship between weight or seroconversion level with survival outcome. Fisher’s exact test was used to compare frequency of clinical symptoms as well as infectious virus in sera or tissues between groups (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). Comparison of curves from rNiV-B Gluc datasets in vitro was done using the Extra sum-of-squares F test to test whether one curve fits for the replication of all viruses.

In vitro efficacy of 4′-fluorouridine (4′-FlU, EIDD-2749) against NiV

4′-FlU ribonucleoside analog previously showed antiviral activity against NiV-M (SPB199901924 prototype isolate, and isogenic recombinant reporter NiVs) with EC₅₀ ranging from 0.1 to 2.9 μM, depending on the assays applied [28,29]. To characterize the activity of 4′-FlU against a predominant pathogenic henipavirus genotype and validate antiviral activity prior to in vivo studies, we tested 4’-FlU against rNiV-B Gluc and its corresponding wild-type 2004 Bangladesh isolate (NiV-B). Virus (MOI 0.001) was added to cells along with 4′-FIU and plates were further incubated for 48h prior to luminescence acquisition or plaque count. 4′-FlU EC₅₀ values were 2.00 and 3.66 μM for, respectively, rNiV-B Glucand NiV-B (S1 Fig), which was comparable to previous results and prompted testing of this drug candidate in animal studies.

Pharmacokinetics studies in Syrian Golden hamster

The pharmacokinetic properties of 4′-FlU have not previously been evaluated in hamsters. We administered 4′-FlU at 2 or 10 mg/kg of body weight to two groups of hamsters in an oral single-dose PK study (Fig 1A). Plasma concentrations of 4′-FlU were determined up to 24 hours post-dose (Fig 1B). 4’-FlU was rapidly absorbed following oral administration, reaching its maximum observed concentration (C_max) by the time of first collection (30 minutes). There was a dose-dependent increase in C_max from 1,700–5,190 ng/ml between the 2 and 10 mg/kg groups, respectively. Concentration of free 4′-FlU in plasma then rapidly declined up to four hours post-dose, suggesting distribution to tissues. Plasma concentrations were relatively stable from 4 hours onwards with concentrations of 42 and 84 ng/ml at 24 hours post-dose in the 2 or 10 mg/kg group, respectively. Terminal half-lives were 16.6 and 35.4 hours for the 2 and 10 mg/kg groups, respectively. Brain tissue samples were collected at 3 and 24 hours post-dose to address whether 4’-FlU crosses the blood-brain barrier in Syrian golden hamsters and determine CNS exposure of its pharmacologically active metabolite, 4’-FlU-TP (Fig 1C). There was a dose-dependent increase in 4’-FlU-TP concentrations in the brain at three hours post-dose, indicating dose-dependency of drug tissue levels. The 10 mg/kg dose level was selected for a subsequent multi-dose PK study (Fig 1D). Plasma dynamics of 4’-FlU were similar after repeated dosing at 10 mg/kg/day compared to those after a single dose (Fig 1E). Plasma C_max reached 6,010 ng/mL after repeated dosing, but trough concentrations were approximately one order of magnitude higher, indicating that substantial accumulation occurred. Note that no evidence of 4’-FlU toxicity was noted during the pharmacokinetics studies.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. 4’-Fluorouridine (4’-FIU) exposure in hamster following a single and multi-oral dose.

(A) Timeline illustration of the pharmacokinetics study conducted after 4’-FIU oral gavage (single dose). Female hamsters (n = 8) received either 2 or 10 mg/kg 4’-FIU. Blood samples were collected as indicated prior final blood and tissue harvest at 24 hours post dosing. (B-C) 4’-FIU concentration in plasma and brain tissues over time. Note that the bioactive triphosphate form of 4’-Fluorouridine (4’-FIU-TP) is detected in the tissue. (D) Timeline illustration of the 4’-FIU oral gavage multi dose treatment prior to the pharmacokinetics study. Hamsters (n = 8) received 5 doses of 10 mg/kg 4’-FIU treatment prior serial blood collection as in illustration A. (E) 4’-FIU concentration in plasma over time. Data are presented as median with error or mean with standard deviation. Panel A and D partially created with BioRender. Plemper, R. (2026) https://BioRender.com/6ri40jk.

https://doi.org/10.1371/journal.ppat.1014093.g001

Proof-of-concept study of 7-day treatment with 4′-FlU against NiV (cohort 1 study: short treatment protocol)

Considering its broad-spectrum activity against several members of the Hareavirales and Mononegavirales order [21,25,28,29,38], we first aimed to test anti-NiV efficacy when 4’-FlU was administered per oral once daily (q.d.) at 10 or 25 mg/kg for 7 days, starting either immediately post infection (pi) with a lethal rNiV-B Gluc inoculum or delayed 8–10 hpi (Fig 2A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Efficacy of orally dosed 4’-FIU against an intraperitoneal rNiV-B Gluc challenge in the Syrian hamster model (Cohort 1).

Five groups of hamsters (n = 10 per group, male and female evenly distributed) received each a targeted 10⁵ PFU dose of rNiV-B Gluc through the intraperitoneal route and were then either immediately or 8-10 hours later (delayed) treated daily and for 7 days with vehicle, or 10 to 25 mg/kg 4’-FIU. One additional group was used as negative control (A). Created in BioRender. Pearson, M. (2026) https://BioRender.com/983h0ga. All six groups were further monitored for 21 days to assess for (B) survival as well as (C, D) clinical signs of disease. Note that some data sets are either (B) aligned or (C, D) staggered for improving visualization of values from each subject over time. The yellow shade (B, C, D, and in legend) indicates days of treatment. The Log-rank test was applied for comparison of survival curves and the Chi-square test was used to compare mortality rates. Asterisks ****) indicate statistical differences of p < 0.0001 between groups.

https://doi.org/10.1371/journal.ppat.1014093.g002

The short treatment protocol delayed disease onset: Five groups of hamsters (n = 10 per group, equal sex) were inoculated with a target dose of 10⁵ PFU/animal using the intraperitoneal (IP) route. Four groups received oral 4′-FlU treatment thereafter. Delivered dose was 2.5 × 10⁴ PFU and all subjects from the virus group (no treatment) succumbed to disease by day 6 (Fig 2B), which equated to a median survival of 5 days. Negligeable to mild weight loss occurred in 90% of animals prior to death (-7.9% range; -4.9% median) (S2A Fig) and was sometimes accompanied by respiratory and neurological signs of disease including labored breathing, paralysis, and tremor requiring euthanasia (Fig 2C,2D). Animals of the mock-infection group did not exhibit any signs of disease and survived until end of study (day 28) (Fig 2B) with an average body weight gain of 17% (S2A Fig). Treatment with 4′-FlU significantly (p < 0.0001) altered disease progression in all treatment arms and none of the animals in any treatment group died while receiving 4’-FlU. Time to death was increased by more than 13 days (median time between 18.5 to 21 days) compared to the virus control group (Fig 2B–2D). However, starting 4–8 days after treatment end, animals in the treatment group succumbed with fatality rates statistically indistinguishable from that of the vehicle control group (total fatality rate 34 of 40 animals, 85%, p = 0.19). Notably, excluding weight loss criteria, no sign of disease was observed up to day 12 in 4′-FlU-treated groups. After treatment end, clinical signs of NiV infection were identical in nature and severity in animals that had received 4′-FlU and succumbed to disease similar to those from the vehicle-treated control group (Fig 2C,2D), suggesting that 4′-FlU delayed disease onset. In addition, weight loss was generally greater and seen in more animals during the 7-day treatment period (p < 0.001 or 0.01, S2B Fig) and up to day 12 (S2C and S2D Fig) when comparing the 25 versus 10 mg/kg cohorts or compared to the mock group, overall suggesting reduced tolerability of the higher dose. Inclusion of drug-only controls would be necessary to confirm this observation

4′-FlU suppressed virus shedding and replication during treatment: Collection of nasal and oral swabs from each subject (Fig 3A,3B) allowed near real-time detection of virus replication during the first 10 days through monitoring luminescence intensity [32,39]. Consistent with the lack of clinical signs (excluding body weight loss) during that time frame, luciferase signals from swabs collected from 4′-FlU-treated hamsters did not increase and were comparable to the background signal of mock-infected animals, indicating efficient suppression of NiV replication. In contrast, at least 2 nasal (p > 0.05) and all oral (p < 0.0001) swabs from the vehicle control group showed elevated luciferase activity, concomitant with disease onset on day 4. Similar results were obtained from retro-orbital bleeds (Fig 3C), altogether revealing complete suppression of virus shedding and replication during treatment and at least up to 10 days post challenge. At study end, after treatment had been discontinued, reporter signals in sera from terminal bleeds of most (86%) animals that had received 4′-FlU and ultimately required euthanasia and from untreated animals were comparable (Fig 3C), indicating viremia.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Efficacy of 4’-FIU at controlling rNiV-B Gluc shedding and dissemination in the Syrian hamster model (Cohort 1).

(A, B) Nasal and oral shedding of virus was assessed during treatment (day 2, 4, and 6) and up to day 10 post challenge through detection of luminescence from the Gaussia luciferase activity. (C) Viremia was assessed through the same method as in A and B. Data are shown as mean with replicates in A, B, and C. Open symbols in C correspond to values from survivors from the challenge.

https://doi.org/10.1371/journal.ppat.1014093.g003

Consistent with viremia, the proportion of animals that harbored infectious particles in lung and liver at some point in the study was similar across the different groups (S2E and S2F Fig). However, virus emerged significantly later in 4’-FlU-treated animals, and the four treatment groups contained less subjects with virus in the brain (p > 0.05) compared to the vehicle group (S2G Fig), indicating that 4′-FlU interfered with viral host invasion.

Proof-of-concept study of extended treatment with 4′-FlU against NiV (cohort 2 study)

Based on the observation that 4’-FIU prevented disease progression in the active treatment phase, we next asked whether a prolonged treatment course will facilitate viral clearance, resulting in full recovery of animals. Since we did not observe significant differences in efficacy between the different dose groups in the cohort 1 study, we selected only the 10 mg/kg dose level for this extended treatment course.

The extended treatment courses improved survival: Five groups of hamsters were inoculated IP with a target dose of 10⁵ PFU per animal (actual dose delivered was 5.3 × 10⁴ PFU) and four groups received oral 4′-FlU treatment thereafter (Fig 4A). Compound was administered per oral for a total of 28 days with an unplanned 4-day gap at days 14–17, or for 21 consecutive days starting either immediately after infection or 8–10 hpi. A mock control group remained uninfected, and a reference group received vehicle only. All animals in the vehicle control group again succumbed rapidly to infection (median survival 5 days) (Fig 4B), showed rapid body weight loss (90% of animals) (S3A Fig), and developed lethargy progressing to respiratory and neurological endpoints (Fig 4C). Hamsters in the mock group remained healthy until end of study (day 51) and gained on average 36% from original weight (S3A Fig). Compared to the 7-day treatment regimen applied before, the extended treatment courses significantly improved survival (Fig 4B). When monitoring animals for an additional 20 days after completion of a 28-day treatment regimen, survival rates ranged from 60 to 80% compared to 0% in the virus control group (p < 0.01 or 0.001). Similarly, 60% survival was observed in animals monitored for 14 days after a 21-day continuous treatment course compared to 0% in virus control group (p < 0.01). Total study length for the two 21-day treatment groups was restricted to 34 days to probe for infectious particles in tissues of phenotypically healthy survivors (score of 1), or to monitor for potential delayed neurological symptoms. Disease progression varied notably between the different treatment groups (p < 0.05) (Fig 4C,4D). Mean time-to-death ranged from 40.5 to 42 days and from 31.5 to 32.5 days in groups receiving treatment for 28 or 21 days, respectively. All treated animals appeared healthy up to day 31 and 24 in these groups without body weight loss. These results indicate that both survival and delay in progression of disease are 4′-FlU treatment-dependent. Weight gain by the end of dosing in the 28-days treatment groups tended to be slightly delayed (p < 0.05) compared to mock groups (S3B Fig). Body weight on the last day of treatment and overall outcome were not correlated (point-biserial correlation; r ranging from -0.38 to 0.30). However, body weight gain of survivors between end of treatment and termination of study was steady in all groups and comparable to that of the mock group (S3A, S3C, and S3D Fig). Seroconversion to infection was assessed using samples collected at time of animal meeting criteria for euthanasia or end of study (S3E Fig). An LRNT₅₀ higher than 10 (reciprocal dilution) could be calculated for most 4’-FlU-treated animals and tended to be lower (p < 0.01) in those from the 21- (range: 176, mean:43) versus 28- (range: 548, mean:138) day treatment regimen. However, seroconversion levels did not correlate with survival to infection (point-biserial correlation; r = -0.02).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Effect of prolonged 4’-FIU treatment against an intraperitoneal rNiV-B Gluc challenge in the Syrian hamster model (Cohort 2).

Four groups of hamsters (n = 10 per group, male and female evenly distributed) received each a targeted 10⁵ PFU dose of rNiV-B Gluc through the intraperitoneal route and were then dose similarly as in cohort 1 for 28 days total or 21 consecutive days with 10 mg/kg 4’-FIU. Two additional groups were used as virus and negative controls, similar to cohort 1 (A). Created in BioRender. Pearson, M. (2026) https://BioRender.com/983h0ga. Animals were monitored up to day 34 or 51 after challenge to assess for (B) survival and (C, D) clinical signs of disease. Note that some data sets are either (B) aligned or (C, D) staggered for improving visualization of values from each subject over time. The yellow shade (C, D, and legend) indicates days of treatment whereas arrowheads (B) indicate treatment cessation. The Log-rank test was applied for comparison of survival curves, and the Chi-square test was used to compare mortality rates. Asterisks (*, ****) indicate statistical differences of respectively p < 0.05, and 0.0001 between groups.

https://doi.org/10.1371/journal.ppat.1014093.g004

The 28-day treatment course suppressed virus shedding, and viremia, as well as reduced incidence of respiratory disease: Luminescence signals in nasal swabs were elevated only in animals of the vehicle-treated control group at day 4 (10%) or when moribund (100%, p < 0.0001), whereas signals in all 4′-FlU-treated animals remained at background level, similar to those seen in mock-infected animals (Fig 5A). Luciferase activity in oral swabs from animals of the vehicle control group was up to two orders of magnitude higher than those in 4’-FlU-treated animals and the mock control group (p < 0.01) (Fig 5B). These data suggest that extended treatment with 4′-FlU efficiently suppressed virus shedding, especially in animals that received the compound for 28 days. Luminescence values in blood samples were high in all vehicle animals but remained at background level in the 4’-FlU cohorts except for three hamsters in the 21 days dosing groups, which showed high values at terminal sampling (Fig 5C). In line with this, infectious virus was retrieved in terminal blood of 86% of vehicle animals while it was undetectable in 95% of terminal samples from 4’-FlU-treated animals (S4A Fig). These results were also consistent with our observation that fewer subjects in the 4′-FlU-treated versus vehicle groups contained infectious virus in brain (approaching significance in 2 groups), lung (p < 0.05 to 0.0001) and liver (p < 0.01 to 0.001) tissues (S4B, S4C, and S4D Fig). Although the incidence of neurological symptoms was similar between groups, fewer respiratory clinical signs were present in groups receiving 4′-FlU for 28 (p < 0.0001) or 21 days (p > 0.05) (S1 Table), suggesting that prolonged treatment, especially the 28-day regimen, reduced viremia and NiV pneumonia in the hamster model.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Effect of prolonged 4’-FIU treatment on rNiV-B Gluc shedding and dissemination in the Syrian hamster model (Cohort 2).

(A, B) Nasal and oral shedding of virus was assessed during treatment (day 1, and 19 or 28) and at the end of study through detection of luminescence from the Gaussia luciferase activity. (C) Viremia was assessed through the same method as in A and B. Data are shown as mean with replicates in A, B, and C. Open symbols in A, B, and C correspond to values from subjects succumbing to disease. One way ANOVA followed by Tukey’s multiple comparisons test was used to compare shedding and viremia data between groups. Asterisks (*, **, ****) indicate statistical differences of respectively p < 0.05, 0.01, or 0.0001 between groups. “ns” for not significant.

https://doi.org/10.1371/journal.ppat.1014093.g005

The 28-day treatment course reduced NiV-induced lung tissue pathology: Lung lobes from seven animals in the vehicle group showed NiV antigen in type II pneumocytes, and syncytia were frequently noted in endothelial cells of the pulmonary vasculature (Fig 6A and S2 Table), as opposed to lung sections derived from mock-infected hamsters (Fig 6B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Virus nucleocapsid antigen distribution in the lung and associated lesions following rNiV-B Gluc challenge.

(A) Virus nucleocapsid staining (dark red signal) in the alveolar area including in type II pneumocytes and endothelial cells of the small vasculature. (B) No specific virus nucleocapsid staining is observed in the alveolar space of mock tissues. (C) Hematoxylin and eosin staining showing virus-induced lung lesions including syncytia formation, necrosis, and hemorrhages. (D) Hematoxylin and eosin staining of a mock-infected untreated animal showing no remarkable lung lesions.

https://doi.org/10.1371/journal.ppat.1014093.g006

Histopathological analysis revealed that all subjects of the vehicle control group experienced either infiltration of mononuclear cells or developed necrotic and hemorrhagic lesions in the perivascular region and interstitium (Fig 6C and S2 Table). Mock-infected animals remained histopathologically unremarkable (Fig 6D). Lung edema or fibrinous exudate were furthermore consistently noted in animals in the vehicle control group (S2 Table). In contrast, lungs of 65% of animals in the 28-day 4′-FlU treatment groups were unremarkable, 30% showed mild interstitial increase of inflammatory cells, and only 5% presented with edema (S2 Table). Lung histopathology of 21-day treated animals did not fully correlate with disease outcome. Lesions were more frequent (68%) and presented predominantly more advanced (nodules, necrosis and edema) in hamsters with additional clinical signs (S2 Table), underscoring that a longer treatment course reduced NiV-induced lung pathology.

Brain lesions did not correlate with neurological clinical signs or with presence of viral genome: Brain lesions in the vehicle group could only be detected based on focal demyelination in the thalamus of two subjects (28%) (S2 Table), including one animal presenting with paralysis, despite the presence of NiV antigen in endothelial cells and neurons in the striatum and hypothalamus (Fig 7A,7B and S2 Table). These animals also lacked neurological symptoms. Brain lesions were reported in 30% and 40% of, respectively, the 28- and 21-day 4′-FlU treatment groups and included mild to moderate increases in mononuclear cells in the perivascular region in pons (Fig 7C and S2 Table), focal necrosis in the thalamus (Fig 7D and S2 Table), mild to moderate meningitis in the cortex (Fig 7E and S2 Table), and focal demyelination in the thalamus (S2 Table). Brain histopathology did not consistently correlate with neurological signs including paralysis, tremor, and head tilt. No remarkable lesions were noted in the corresponding sections of mock-infected animals (Fig 7F and S2 Table). Viral genomic RNA could be retrieved from one brain of a survivor (28-day treatment group, S2 Table) that presented without neurological signs or brain lesions.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Virus nucleocapsid antigen distribution in the brain and associated lesions following rNiV-B Gluc challenge.

(A, B) Virus nucleocapsid staining (dark red signal) in the hypothalamus, thalamus, and pons including in neurons, and endothelial cells. Hematoxylin and eosin staining showing virus-induced brain lesions including (C) mononuclear cells infiltrations, (D) focal necrosis, and (E) meningitis in the cortex. (F) Hematoxylin and eosin staining of a negative control showing no remarkable brain lesions.

https://doi.org/10.1371/journal.ppat.1014093.g007

The 28-day treatment course prevented NiV-induced systemic inflammation: Circulating cytokines were quantified at the time of euthanasia, since they contribute to NiV pathogenesis [40,41] and can indicate changes in blood brain barrier (BBB) integrity in hamsters [42]. A 9-plex panel for IL (interleukin)-2, IL-4, IL-6, IL-10, Interferon (IFN)-γ, MCP-1, MIP-1α, TNFα & VEGF was tested with samples of the mock, vehicle, and pooled 28-day 4′-FlU treatment groups (S5 Fig). IFN-γ was significantly increased in hamsters in the vehicle group at peak disease compared to mock-infected and 4′-FIU-treated animals (p < 0.0001). Elevated IL-10 was present in the vehicle group compared to 4′-FIU-treated animals (p < 0.05), but not the mock group (p > 0.05). These results are consistent with the absence of pronounced lung lesions and viremia in animals treated for 28 days, indicating that treated and mock-infected animals did not develop systemic inflammation.

rNiV-B Gluc did not require adaptation to replicate in the lung of hamsters: To assess genetic integrity of compound-experienced virus populations, the rNiV-B Gluc genome of the virus stock was first deep sequenced and compared to reference sequence (human isolate, AY988601, [31]), which was appropriately modified to reflect the insertion of reporter genes between the nucleocapsid and phosphoprotein viral genes. Three single-nucleotide polymorphisms (SNP) with allele frequency (AF) over 99% were identified in the whole genome of rNiV-B Gluc, which corresponded to two nonsynonymous mutations in the fusion protein at amino acids (aa) 207 and 252 (S3 Table). No other SNPs with AF higher than 25% were detected. Then, NiV genome-positive tissues from the second cohort were used for sequencing and the same two nonsynonymous mutations in the fusion protein were found in infected tissues. Note that no coverage for NiV genome was found in mock-infected samples, confirming that these animals had not experienced virus. Of 11 lung tissue samples positive for virus RNA, 6 samples from the vehicle group and 3 samples from the 21-day 4′-FIU-treated group provided sufficient reads and coverage for reliable interpretations (S2 and S4 Tables). No nonsynonymous or complex mutations with AF higher than 10% were found in the 3 lung samples of the 21-day 4′-FIU-treated animals compared to genomes recovered from the vehicle group (S6 Fig), and even a lower AF threshold (5%) did not reveal any changes in the RNA-dependent RNA-polymerase (RdRp) gene (L).

rNiV-B Gluc did not require major adaptation to enter the CNS and 4′-FIU pressure induced adaptative changes to NiV replication complex: Only one SNV with AF > 25% was recovered from one of the seven sequenced untreated NiV-infected brain specimens (Matrix, G122*, 29%), suggesting that CNS-invasion does not require major NiV evolution in Syrian golden hamsters. Sequence analysis performed on the NiV genome-positive brain samples across the 21 (n = 6), and the 28-day (n = 5) 4′-FIU-treated groups (S4 Table) showed five nonsynonymous and nonsense single-nucleotide variants (SNV) with AF higher than 25% that were solely found in viral sequences from four hamsters that had received 4′-FIU (S5 Table). However, the corresponding lung tissues of these animals were negative for viral RNA and it could, therefore, not be determined whether these changes represent adaptation to the specific tissues or emerged in response to exposure to 4’-FlU. Three of the five SNVs were nonsynonymous, carried AF values higher than 96%, and were reported in three distinct samples from animals that succumbed to disease or met euthanasia criteria. Changes occurred either in the PRNTase domain of the RdRp (L gene, residue T1341I) [43] (Fig 8A–8C), in the nucleocapsid (N) gene (residue T191S) near a region directly interacting with vRNA [44] (Fig 8D–8E), or in the N-terminal domain of the phosphoprotein (P) gene (residue I286V) [43,45] (Fig 8F). The threonine at position 1341 of L was the most conserved of the 3 reported SNVs among henipaviruses (Fig 8C). A fourth specimen had two linked nonsense mutations in the P gene (E162*, E171*) at 53% and 38% AF, respectively. Due to lack of statistical power (n = 1 for each of these mutations), we inserted the 3 nonsynonymous SNVs individually into the rNiV-B Gluc backbone to assess whether either of these substitutions may represent a compound-induced change. Susceptibility of the recovered rNiV-B Gluc mutants to 4’-FlU was compared to that of the genetic parent virus in vitro (Fig 9A). Inhibitory concentrations were slightly increased by 2.5- to 3-fold compared to parental rNiV-B Gluc (EC₅₀ ranging from 1.59 µM to 1.85 µM versus 0.6 μM, F test, p < 0.0001), altogether suggesting that these SNVs could provide a minor advantage to the virus population in the presence of 4’-FlU inhibitory pressure.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Nonsynonymous single-nucleotide variations are in components of the viral replicase.

(A) Structural modeling depicting the viral replicase complex, consisting of the vRNA wrapped in N forming the vRNP, and displaced N where the RdRp, L, is threaded along the template vRNA; L is associated with a tetramer of P. PDB structures 7NT5 (NiV-N bound to RNA) and 9CGI (NiV-L and NiV-P) were manipulated in UCSF ChimeraX to generate these visualizations. NiV-N is depicted in light green, vRNA is depicted in black, NiV-P is depicted in blue, and NiV-L is depicted in yellow. (B) Visualization of the internal cavity of NiV-L with regions corresponding to the GDNE motif, priming loop, and intrusion loop indicated in pink, purple, and yellow, respectively. The T1341 residue identified in one virus as T1341I is highlighted in red. The PRNTase domain is depicted in green, whereas the vRdRp domain is depicted in light blue. (C) Amino acid alignment of the region of the intrusion loop from L proteins of NiV-B, NiV-M, HeV (Hendra virus), CedV (Cedar virus), GhV (Ghana virus), AngV (Angavokely virus), and MeV (Measles virus); T1341 and corresponding residues are highlighted in red, and the highly conserved HR motif is highlighted in orange. (D) Structural depiction of NiV-N bound to vRNA, with the T191 residue highlighted in red. Adjacent residue R192, which is involved in RNA binding, is shown in blue. (E) Amino acid alignment of the region of N containing T191 from NiV-B, NiV-M, HeV, CedV, GhV, AngV, and MeV. T191 and corresponding residues are highlighted in red. (F) Amino acid alignment of the region of P containing I286 from NiV-B, NiV-M, HeV, CedV, GhV, AngV, and MeV. I286 and corresponding residues are highlighted in red. For all sequence alignments, ClustalW was implemented in MEGA-X; phylogenetic trees depict topology only and were constructed for each alignment using the neighbor-joining method in MEGA-X. Conserved residues are shown with an * below each respective alignment. Numbering and mutants for each alignment refer to NiV Bangladesh sequences for each respective protein.

https://doi.org/10.1371/journal.ppat.1014093.g008

Download:

• PNG
  larger image
• TIFF
  original image

Fig 9. In vitro efficacy of 4’-FIU against rNiV-B Gluc incorporating T191S (N gene), I286V (P gene), or T1341I (L gene) change and viral fitness in human astrocytes.

(A) Antiviral assays were concomitantly conducted in Vero CCL81 and on 4 individual 96-well plates with each mock and virus control (8 biological replicates). Four biological replicates were used per 4’-FIU concentration per plate. Determination of half-maximal effective concentration (EC₅₀) against each virus at day 2 post-infection was done using the luminescence reporter on a Cytation 5 (Agilent). Values indicate the percentage of virus replication based on intensity of reporter gene expression from 4’-FIU- versus vehicle-treated infected cells. Results are expressed as the average of biological replicates, and error bars represent standard deviations. (B) Growth kinetic of rNiV-B Gluc and mutants in primary human astrocytes. Supernatants were collected 6 times between 1 and 48 hours post infection and were then tittered by plaque assay. Results are expressed as the average of biological triplicates, and error bars represent standard deviations. Horizontal dotted line shows the limit of detection.

https://doi.org/10.1371/journal.ppat.1014093.g009

The viral fitness of rNiV-B Gluc mutants is reduced compared to parental virus in human brain astrocytes

Viral genome and proteins have been detected in the CNS, including in astrocytes, across several animal models of Nipah virus disease [46–49] and these glial cells may also be affected in humans [50]. To evaluate the fitness of rNiV-B Gluc mutants in the brain, a 48-hour kinetic was performed using primary human astrocytes derived from the cerebral cortex (Fig 9B). The titers from rNiV-B Gluc mutants in the supernatants remained comparable from one another through the time course and tended to be slightly lower compared to the parent virus from 16 to 36 hours post infection (Extra Sum-of-Squares F test; p = 0.0004). Specifically, only the rNiV-B Gluc titer increased from below the limit of detection (LOD 67pfu/ml) to an average of 5.6 x 10² pfu/ml by the 16-hour mark suggesting a minimum of about 10-fold difference compared to rNiV-B Gluc mutants. This slight advantage was still evident at 36 hours, and by the final time point, titers from the parent virus and two out of the three rNiV-B Gluc mutants exceeded 10⁶ pfu/ml. Together, these results indicate that human astrocytes are permissive to NiV and that the nonsynonymous SNVs in L, N, and P gene did not improve growth kinetics or increased maximal titers in vitro.