Aminoglycosides induce immune activation but is insufficient for systemic viral control: a double-blind randomized-controlled clinical trial Authors: Punitha Sivanandam, Ayesa Syenina, Yvonne FZ Chan, Dorothy HL Ng, Esther S Gan, Jia Xin Yee, Christine YL Tham, Hwee Cheng Tan, Eugenia Z Ong, Eng Eong Ooi, Shirin Kalimuddin Published: 2026 ABSTRACT AbstractBackgroundMucosally-applied aminoglycosides have demonstrated host-directed antiviral activity in preclinical models through induction of interferon-stimulated genes (ISGs). However, it remains unknown if systemically administered aminoglycosides elicit comparable effects in humans.MethodsWe conducted a single-centre randomized, double-blind, placebo-controlled trial in Singapore in 32 healthy adults who were randomized 1:1 using sequentially sealed envelopes, to receive single-dose intravenous gentamicin 5 mg/kg or placebo. Participants were subsequently challenged with live-attenuated yellow fever 17D (YF17D) virus and followed for 30 days. The primary outcome was the proportion of participants with detectable YF17D RNAemia. Secondary outcomes included RNAemia levels, symptom rates, neutralizing antibody titers, and innate immune transcriptomic changes.ResultsWe found no significant differences between the gentamicin and placebo groups in detectable RNAemia (62.5% [n = 10/16] vs. 68.8% [n = 11/16], p > 0.99), mean RNAemia levels (3.93 log10 area under the curve [AUC] vs. 3.51 log10 AUC, p = 0.81), symptom rates (56% [n = 9/16] vs. 25% [n = 4/16], p = 0.15) or antibody titers (mean log10 PRNT50 titer: 3.89 vs. 3.71, p = 0.37). However, transcriptomic profiling demonstrated upregulation of ISGs RSAD2 and OASL and the transcription factor IRF3 in gentamicin-treated individuals, with concomitantly higher protein expression of chemokines CCL2, CXCL12 and IL-27.ConclusionsThis study represents, to our knowledge, the first in humans to evaluate the host-directed antiviral effects of systemic aminoglycosides. Our findings suggest that, whilst modest and insufficient to control systemic viral infection, the immunomodulatory effects of aminoglycosides warrant continued exploration as a potential preventive strategy against mucosal viral infections.Subject terms: Viral infection, Antivirals Introduction As clearly illustrated by the recent COVID19 pandemic, antivirals remain an important tool in mitigating disease burden from viruses with epidemic potential1. However, the development of direct-acting antivirals for emergency use is both resource-intensive and technically challenging – this is especially pertinent for RNA viruses which are prone to rapid mutation and emergence of resistance2. An alternative strategy would be to exploit off-target antiviral properties of existing licensed drugs. Recent studies have shown that the aminoglycoside class of antibiotics exhibits broad-spectrum host-directed antiviral activity. Aminoglycoside antibiotics exert their antibiotic effects through binding bacterial ribosomal RNA (rRNA) and disrupting protein translation, but they are also able to bind mitochondrial and mammalian RNA3–5. Preclinical studies have suggested that aminoglycosides synergise with host double-stranded RNA and render them more “visible” to pattern recognition receptors such as toll like receptors (TLRs) in neighbouring dendritic cells, thereby initiating downstream innate immune signalling and upregulation of interferon stimulated genes (ISGs)6. In mouse models of infection, topically administered aminoglycosides such as neomycin and kasugamycin induced expression of ISGs via activation of TLR3-interferon regulatory factor 3 and 7 (TLR-IRF3/7) signalling pathways, conferring broad-spectrum antiviral resistance against viruses such as Zika, influenza A, SARS-CoV-2 and herpes simplex6,7. In humans, intranasal application of the aminoglycoside neomycin induced local ISG expression7, although its antiviral efficacy in humans has yet to be conclusively demonstrated. While mucosal application of aminoglycosides can induce antiviral immune responses to control respiratory viruses such as influenza or SARS-CoV-2, the possibility that aminoglycosides could also control systemic viral disease remains unknown. We therefore aimed to investigate whether parenteral administration of aminoglycosides could elicit comparable responses in systemic compartments, and thus confer systemic viral control in humans. Herein, we conducted a double blind, randomized, placebo-controlled clinical trial to evaluate the immunomodulatory and antiviral effects of a systemically administered aminoglycoside in healthy human volunteers who were subsequently challenged with live-attenuated yellow fever (YF17D) virus. Previous studies demonstrating the antiviral effects of mucosally applied aminoglycosides primarily used neomycin and kasugamycin; however, these agents are not licensed for parenteral use in humans. Gentamicin was therefore selected for our study because it is licensed for parenteral administration and has a well-established safety and tolerability profile8. Structurally, gentamicin shares similarities to neomycin, with both containing a central 2-deoxystreptamine ring glycosidically linked to amino sugar moieties9. Although other aminoglycosides such as amikacin and streptomycin are also licensed for parenteral use, prior work has shown that they do not induce local ISG expression in murine models6. Importantly, prior to trial initiation we demonstrated that gentamicin upregulated innate immune gene expression and inhibited viral replication in primary human monocytes ex vivo, providing additional rationale for evaluating its host-directed antiviral effects in this clinical model. YF17D was chosen as the challenge virus as infection results in detectable RNAemia, as well as mild systemic symptoms such as fever, lethargy and myalgia, in the majority of recipients10–13. The primary objective of our study was to assess whether a single parenteral dose of gentamicin administered prior to YF17D challenge, could reduce the proportion of participants with detectable YF17D RNAemia post-challenge. Secondary outcomes included innate immune transcriptomic changes, mean RNAemia levels and duration of RNAemia, symptomatic outcomes and post-challenge neutralizing antibody titers. Exploratory outcomes included post-challenge YF17D-specific T cell responses. Although parenteral gentamicin administration did not result in measurable viral control, we were able to observe significant modulation of both peripheral blood innate immune gene expression as well as chemokine production. Whilst our findings suggest that parenteral aminoglycosides are unlikely to confer benefit as a host-directed strategy against non-mucosally transmitted systemic viral infections, they lend support for further clinical investigation of this antibiotic-class as a preventative strategy against mucosal or respiratory tract infections. Methods Study participant details and ethics The study was approved by the SingHealth Centralised Institutional Review Board (CIRB) (Ref: 2022/2090) and registered on clinicaltrials.gov (NCT05303909) on 21 March 2022. The study sponsor was the Singapore General Hospital, Singapore, in collaboration with Duke-NUS Medical School, Singapore. The study was carried out in accordance with the principles of the Singapore Good Clinical Practice guidelines and in compliance with the Helsinki Declaration. Participants were compensated for their time and inconvenience. Study enrolment, follow-up and data collection were conducted between 01 May 2022 to 31 July 2023 at the SingHealth Investigational Medicine Unit, Singapore. Data was collected using REDCap (v13.1.30). Study participants Healthy adult volunteers aged 21–50 years were enrolled into the study, after written informed consent. Other inclusion criteria included a satisfactory baseline medical assessment and stable health status, willingness to use adequate and reliable contraceptive measures or practice abstinence for 10 days after vaccination, and a negative urine pregnancy test (for female volunteers of child-bearing potential) at screening and day of vaccination. Key exclusion criteria included a history of cardiovascular, respiratory, hepatic, renal, gastrointestinal, neuropsychiatric, hematological, endocrine or immunosuppressive disorders that would be a risk factor for study vaccine administration, known allergy to aminoglycosides, previous receipt of YF17D vaccination, known allergy to YF17D vaccines or aminoglycosides, thymus gland disease, acute infection in the preceding 7 days or presence of a temperature ≥ 38.0 °C on the day of the first vaccination, pregnant or breast feeding women, receipt of anti-inflammatory drugs (such as non-steroidal anti-inflammatory drugs or systemic steroids) in the past 7 days, and receipt of any licensed vaccine in the past 30 days before the first study vaccine dose. Study design Pharmacokinetic assessment of intravenous gentamicin Participants were allocated, based on order of study enrolment, into 3 dosing groups. Group 1 received single dose IV gentamicin at a dose of 1 mg/kg, Group 2 at a dose of 3 mg/kg and Group 3 at a dose of 5 mg/kg. In all three groups, the gentamicin was diluted in 100 mls 0.9% normal saline and administered over 30 min via a cannula inserted into a peripheral vein. Blood samples were collected for measurement of serum gentamicin levels at 30 min, 3 h, 6 h and 24 h post-infusion. Randomized-controlled clinical trial Participants were randomized in a 1:1 ratio to receive either single-dose IV gentamicin 5 mg/kg diluted in 100 mls 0.9% normal saline or 100 mls 0.9% normal saline (placebo) administered over 30 min via a cannula inserted into a peripheral vein, on day 0. Randomization was performed using sequentially numbered, opaque, sealed envelopes to ensure allocation concealment. The allocation sequence was prepared by an independent study team member not involved in participant enrolment or assessment. The study was conducted in a double-blind manner; both participants and investigators were blinded to treatment allocation. Study infusions were prepared by an unblinded pharmacist not involved in participant assessment to maintain blinding. One-hour after the end of the gentamicin or placebo infusion, participants received YF17D vaccination administered subcutaneously in the deltoid region of the participant’s right or left arm. In order to ensure consistency in the antigen-load delivered, all vaccines administered were from the same manufacturing lot. The dose of 5 mg/kg of gentamicin was chosen as this dose is well within the recommended safe maximum once-daily dose of 7 mg/kg in healthy adults14. Participants were followed up for 30 days from the first vaccination, and blood samples were collected at multiple time-points (Day 0; prior to both gentamicin/placebo and YF17D challenge and Days 1, 4, 6, 14 and 30 post-YF17D challenge) for viral RNA quantification, serological and T cell analysis, as well as transcriptomic analysis. Participants were also monitored for vaccine-related symptoms post-vaccination and challenge, and these symptoms were recorded. Study outcomes The primary outcome was the proportion of participants with detectable YF17D RNAemia after YF17D challenge. Secondary outcomes included innate immune transcriptomic changes, mean RNAemia levels and duration of RNAemia, symptomatic outcomes and post-challenge neutralizing antibody titers. Exploratory outcomes included post-challenge YF17D-specific T cell responses. Laboratory assays Serum gentamicin concentrations Serum gentamicin concentrations were quantified using a validated immunoassay performed in an accredited clinical laboratory. Venous blood samples were collected into serum-separator tubes at predefined time points (30 min, 3 h, 6 h, and 24 h following completion of infusion). Samples were allowed to clot at room temperature and were centrifuged within 2 h of collection. Serum was aliquoted and analyzed immediately or stored at 2–8 °C for up to 24 h prior to analysis, in accordance with manufacturer recommendations. Gentamicin concentrations were measured using a cloned enzyme donor immunoassay on the Beckman-Coulter analyser, following the manufacturer’s instructions. The assay was calibrated using traceable gentamicin standards, and quality control samples at low and high concentration ranges were included in each analytical run. The lower limit of quantification (LLOQ) of the assay was 0.5 μg/ml. Samples with concentrations below the LLOQ were reported as undetectable. Viral RT-qPCR Viral RNAemia was measured using reverse-transcriptase quantitative polymerase chain reaction (RT-qPCR) directed against the NS5 gene of YF17D in an ISO 15189 compliant laboratory15. Viral RNA was extracted from 200 μl of plasma samples using the Roche MagNA Pure 24 Total NA Isolation Kit and elution volume was 50 μl. The YF17D qPCR was conducted using the SuperscriptTM III PlatinumTM One-Step qRT-PCR Kit (Invitrogen). Briefly, 5 μl of extracted RNA was added to 20ul master mix containing 10 μM of YF17D specific primers and probe as shown below: YF-17D Forward: GAACAGTGATCAGGAACCCTCTCT YF-17D Reverse: GGATGTTTGGTTCACAGTAAATGTG YFV-17D Probe: 5’ HEX – CTACGTGTC/ZEN/TGGAGCCCGCAGCAAT – IABkFQ 3’ YF17D RNA standards were generated by first synthesizing cDNA from YF17D viral RNA using SuperScript® First-Strand Synthesis System (Invitrogen), subsequently amplifying the YF17D amplicon using Q5 Polymerase (New England Biolabs) according to manufacturer’s instructions. Primers to generate YF17D amplicons are as follows: T7 YF17D forward (TAATACGACTCACTATAGACTCTTGGA AGAGACGGCC) and T7 YF17D reverse (TTGGTGCAGCTTCCTTTCTT). Finally, amplicon was in vitro transcribed using MEGAScriptTM T7 Transcription Kit (Life Technologies) according to manufacturer’s instructions. RNA copy numbers were calculated as previously described16. The limit of quantification (LOQ) for RNAemia measurement was 12.5 copies/μl, any measurements below the LOQ were considered negative. The human RNase P gene17 was amplified as an internal control for successful nucleic acid extraction using the same master mix containing 10 μM of RNase P specific primers and probe as shown below: RNase P Forward: AGA TTT GGA CCT GCG AGC G RNase P Reverse: GAG CGG CTG TCT CCA CAA GT RNase P Probe: 5’ HEX – TTCTGACCT/ZENGAAGGCTCTGCGCG – IABkFQ 3’ Reverse transcription was performed at 50 °C for 15 min and then at 95 °C for 2 min, followed by 45 cycles of PCR amplification in a Roche LC96 RT-PCR System at an annealing temperature of 54 °C. Samples were tested in technical duplicates for YF17D and singlicates for RNase P. RNA extracted from pooled human serum matrix (Seracare) was used as an extraction control; a positive signal for RNase P from the extraction control demonstrated the integrity of the RNA extraction. An in vitro transcribed RNA of the YF17D NS5 gene was used to construct a standard curve ranging from 106 copies/μl to 1 copy/μl, which was used to quantitate vaccine RNAemia in plasma samples. Raw Cq values were subsequently converted into log10 genome copies/ml for reporting purposes. Plaque reduction neutralization test Neutralizing antibody titers induced by YF17D vaccination were measured by plaque reduction neutralization test (PRNT) as described previously18. In brief, serial two-fold dilutions of sera (starting at 1:10, followed by two-fold serial dilutions up to 1:20,480) in RPMI maintenance media (RPMI-MM) were incubated with 40 plaque forming units (pfu) of YF17D virus for one hour at 37 °C followed by addition to BHK-21 cells (100 µl per well). The mixture was aspirated after incubation for one hour at 37 °C. Cells were then overlaid with 1% carboxymethylcellulose (CMC) in RPMI-MM and incubated for five days at 37 °C. Subsequently, cells were fixed with 20% formaldehyde and stained with 1% crystal violet. Assays were considered valid when wells yielded 35-60 plaques, ensuring reliable quantification. The positive control was defined as the empirically determined infectious virus input per well, obtained by back-titration. Serially diluted virus was added to cells without serum, and plaque counts after staining were used to calculate the actual pfu per well. This value was used as the reference for neutralisation analysis. PRNT50 titers were calculated using a sigmoid dose-response curve and reported as reciprocal values PBMC isolation Peripheral blood mononuclear cells (PBMCs) derived primary monocytes were isolated from flavivirus-naïve healthy donors with approval from the National University of Singapore’s Institutional Review Board (reference number B-15-227) as previously described19. Briefly, blood was collected in BD vacutainer containing sodium heparin or K2EDTA and diluted with 2–3 volumes of phosphate buffered saline (PBS) + 0.5% bovine serum albumin (BSA). The diluted blood was then layered over Ficoll (GE healthcare) and centrifuged at 750 x g for 30 min at 22 °C without brakes. After centrifugation, the interphase cells containing PBMCs were aspirated and transferred to a new tube. PBS + 0.5% BSA was then added and centrifuged at 300 x g for 10 min at 22°C. Supernatant was subsequently discarded, and cell pellet washed thrice with PBS + 0.5% BSA. Isolated PBMCs were cryopreserved and stored in liquid nitrogen until used in infection and T cell assays. MTS cell viability assay Cell viability after treatment with 200, 50, 12.5, and 3.125 µg/ml of gentamicin were assessed using the CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega) in accordance with the manufacturer’s protocol. Briefly, 20 μl of CellTiter 96 AQueous One Solution Reagent was added to gentamicin-treated cells in a 96-well plate and incubated for 1–4 h at 37°C, 5% CO2 atmosphere. Absorbance was recorded at 490 nm. Virus quantification in primary human monocytes by plaque assay Plaque assay was performed to quantify viral titres as previously described19 using RPMI-MM on BHK21 cells. Briefly, ten-fold serial dilutions of culture supernatants were infected on BHK cells for 1 h in 37°C, shaking every 15 min. Samples were then removed and replaced with an overlay of 1% carboxymethyl- cellulose in RPMI with 3% FBS and incubated in 37°C for 5 days until formation of plaques were observed. Cells were then fixed with 20% formaldehyde and stained with 1% crystal violet to visualize the plaques. Viral titres were calculated as plaque forming unit (PFU)/ml. Quantification of IL-6 and IRF3 in primary human monocytes by RT-qPCR Quantification of IL-6 and IRF3 gene expression levels was performed by RT-qPCR. Briefly, RNA was isolated from cell lysates using the RNeasy Mini Kit (Qiagen) according to manufacturer’s manual. Next, cDNA synthesis was performed using qScript cDNA Synthesis Kit (Quantas Biosciences). IL-6 and IRF3 transcripts were measured by RT-qPCR using the SYBR Green Supermix Kit (Roche) and normalised to GAPDH levels. All reactions were run on a Roche Lightcycler 480 (Roche) and data analysis was carried out with the Lightcycler 480 software. Primer sequences used are as below: IL-6 Forward: TGTGAAAGCAGCAAAGAGGCACTG IL-6 Reverse: ACAGCTCTGGCTTGTTCCTCACTA IRF3 Forward: TCTGCCCTCAACCGCAAAGAAG IRF3 Reverse:TACTGCCTCCACCATTGGTGTC GAPDH Forward: TTGATTTTGGAGGGATCTCG GAPDH Reverse: GAGTCAACGGATTTGGTCGT Peptide pools 15-mer peptides that overlapped by 10 amino acids and covering each of the following YF17D proteins were synthesized (Genscript): capsid, pre-membrane, envelope, NS1, NS2A/B, NS3 NS4A/B and NS5. IFN-γ ELISpot assay YF17D-specific T responses were assessed using an IFN-γ ELISpot Assay. Cryopreserved PBMCs were thawed and directly tested by IFN-γ ELISpot for reactivity to peptides covering the YF17D proteome. ELISpot plates (Millipore) were coated with human IFNγ antibody overnight at 4 °C. 200,000 PBMCs were seeded per well in singlicate and stimulated for 18 h with YF17D peptides at 2 µg/ml. The plates were then incubated with human biotinylated IFNγ detection antibody, followed by Streptavidin-AP and developed using the KPL BCIP/NBT Phosphatase Substrate. Unstimulated wells containing cells and medium alone were included as negative controls, while PMA/Ionomycin-stimulated positive controls were included to confirm cell viability and functional responsiveness. To quantify positive peptide-specific responses, spot forming cells (SFC) were quantified with ImmunoSpot. To quantify positive peptide-specific responses, 2 × mean spots of the unstimulated wells were subtracted from the peptide-stimulated wells, and the results expressed as SFC/106 PBMCs. Results were excluded if negative control wells had >30 SFC/106 PBMC or positive control wells (PMA/Ionomycin) were negative. RNA sequencing Whole blood was obtained in Tempus Blood RNA tubes, which contain an RNA-stabilising agent, enabling transport of samples to be conducted at room temperature without the need for cold chain conditions. Tubes were then stored at −80 °C until RNA extraction. RNA extraction was performed with Tempus Spin RNA Isolation Kit, eluted in 90 μl elution buffer and stored at −80 °C. Total RNA quality check was assessed using ThermoFisher NanoDrop ND-2000 and the LabChip GX Touch HT before library preparation by Azenta Life Sciences. 1 μg total RNA was used as input for library preparation. Poly(A) mRNA isolation was performed using mRNA capture beads before cDNA synthesis using the VAHTS Universal V8 RNA-Seq Library Prep Kit for Illumina according to manufacturer’s instructions. End repair and dA-tailing of the purified cDNA was performed, followed by a T-A ligation to add adaptors on both ends. Size selection of adaptor-ligated DNA was then performed using VAHTS DNA Clean Beads. Samples were PCR amplified using P5 and P7 primers and the PCR products were validated. Libraries were multiplexed using VAHTS RNA Multiplex Oligos Set 1 for Illumina and library quality was validated using capillary electrophoresis on Agilent 2100 Bioanalyzer, before sequencing on an Illumina Novaseq instrument using a 2 x 150bp paired-end (PE) configuration according to manufacturer’s instructions, yielding 150 million reads per sample. The DESeq2 (v1.42.0) R package was used to normalize raw counts from RNA sequencing data. Raw counts were averaged for duplicate gene reads and genes present in a minimum of 16 samples with at least 10 reads were included for subsequent analysis. This was determined by the minimum number of samples present in each comparison group and yielded 16,374 genes for downstream analysis. The DESeq function in DESeq2 was used to normalize filtered genes, which fits values to a negative binomial distribution model and conducts a Wald statistical test20. DESeq2’s results function was used to obtain log2 fold-changes and p-values. To identify positively and negatively enriched blood transcription modules (BTM), gene expression fold-changes were ranked, and analyzed by pre-ranked gene set enrichment analysis (GSEA) using BTMs as gene sets21,22. Significantly enriched modules (FDR < 0.05) were identified using GSEApy (https://github.com/zqfang/GSEApy), a Python wrapper for GSEA and Enrichr. Olink cytokine profiling Plasma cytokine levels were measured using Olink Target 48 Cytokine panel23. Briefly, plasma samples were incubated with antibody pairs tagged with DNA reporter oligonucleotides. Upon binding to target proteins, the antibody pairs come into proximity, leading to hybridization and extension of the DNA reporter barcode. The double-stranded DNA barcode was then amplified and quantified using the Olink Signature Q100. Biological replicates rather than technical replicates were used, as is the standard recommended practice for Olink Proximity Extension Assay (PEA)-based platforms. Data was calibrator-normalised with values reported as pg/ml. Statistical analyses were performed using GraphPad Prism v9.0. Cytokine concentrations were compared between groups at each timepoint independently using Student’s two-sided t-test. Each cytokine was treated as a separate, independent analysis; no correction for multiple comparisons across the cytokine panel was performed Statistics and reproducibility A sample size of 22 per arm would provide 80% power at 5% two-sided type I error rate, to detect a 50% reduction in the proportion of participants with detectable YF17D RNAemia between participants who receive gentamicin and placebo. This was based on an estimated rate of detectable RNAemia of 40% in the gentamicin group and 80% in the placebo group. However, a decision was made to terminate enrolment after the first 33 participants due to significant delays in trial conduct, primarily as a result of the COVID-19 pandemic. All analyses were conducted on a per-protocol basis, as one participant was withdrawn due to non-attendance at study visits on days 4 and 6 post-YF17D challenge, which were predefined critical time-points for RNAemia assessment. No formal intention-to-treat (ITT) analysis was performed. Missing data were handled by complete-case analysis without imputation, and no imputation methods were applied for missing outcomes. Participants with incomplete primary endpoint data (n = 1) were excluded from the per-protocol analysis population. The difference in proportions of participants with undetectable YF17D RNAemia levels and vaccine-induced systemic symptoms in the two groups were analysed using a two-tailed Fisher’s exact test. Analyses of continuous variables were performed using a two-tailed Student’s t-test. Statistical analyses were performed using GraphPad Prism v9.0; details are provided in the figure legends. Results Gentamicin upregulates innate immune gene expression and inhibits YF17D replication in human primary monocytes Prior to conduct of the clinical trial, we first investigated whether gentamicin could induce innate immune responses and antiviral effects in an ex-vivo model using human primary monocytes. Human primary monocytes from two healthy donors were treated with varying doses of gentamicin, ranging from 0.01 μg/ml to 50 μg/ml, for 24 h. There was no evidence of gentamicin-induced cytotoxicity on MTS assay (Figure S1a). Gene expression levels of intereleukin-6 (IL-6) and interferon regulatory factor 3 (IRF3) were measured using RT-qPCR. IL-6 is a pro-inflammatory cytokine that is induced via nuclear factor-kB (NF-kB) transcription factor upon TLR signalling24, while IRF3 plays an essential role in the TLR3 mediated IFNβ response to viral infection24. We found that gentamicin treatment indeed upregulated gene expression of both IL-6 and IRF3 in a dose dependant manner (Figure S1b). Next, to determine if these innate immune transcriptional changes would be sufficient for inhibition of YF17D infection, gentamicin-treated primary monocytes from both healthy donors were infected with YF17D and viral replication was measured 48 h post-infection. Corresponding to the induction of innate immune genes, primary monocytes from both donors treated with a concentration at or above 12.5 μg/ml gentamicin showed significant decrease in YF17D plaque titers as compared to non-treated cells, although there was some variability in response between the two donors (Figure S1c). Critically, a serum concentration of 12.5 μg/ml is routinely achieved in human subjects who receive 5 mg/kg gentamicin administered parenterally as a single dose, which is within the recommended safe dosing guidelines for gentamicin14. Overall, we found, using an ex vivo model of primary human monocytes, that gentamicin induces innate immune gene expression and results in antiviral activity against YF17D infection, at clinically achievable concentrations. Clinical studies Pharmacokinetic assessment and dose selection of intravenous gentamicin We next enrolled 21 healthy adult volunteers in a pharmacokinetic study to determine the optimal intravenous (IV) gentamicin dose for use in the subsequent randomized-controlled clinical trial. Participants received a single IV infusion of gentamicin at 1 mg/kg (n = 7), 3 mg/kg (n = 7) or 5 mg/kg (n = 7), administered over 30 min via a peripheral vein. Blood samples for measurement of serum gentamicin concentrations were collected at 30 min, 3 h, 6 h and 24 h after completion of the infusion. As anticipated, serum gentamicin concentrations at 30 min, 3 h and 6 h demonstrated a clear dose to serum concentration relationship, with the 5 mg/kg group achieving the highest serum levels (Fig. 1a). Notably, all participants in the 5 mg/kg group attained serum gentamicin concentrations > 20 μg/ml at 30 min, with levels remaining detectable in all participants at 6 h post-infusion. By 24 h, serum gentamicin concentrations were undetectable in all participants, irrespective of dose received. On the basis of this pharmacokinetic data, and our previous ex-vivo findings of innate immune gene activation and control of YF17D replication at gentamicin concentrations above 12.5 μg/ml, a dose 5 mg/kg was selected for evaluation in the subsequent clinical trial. Study overview. a Pharmacokinetics of gentamicin were measured for gentamicin doses 1, 3, and 5 mg/kg at 30 min, 3, 6, and 24 h. b Overall study design. Participants were randomized to receive either IV gentamicin or normal-saline placebo. One-hour post intervention, all participants were challenged with YF17D and followed up for 30 days. Blood was sampled for viral RNAemia, immune gene expression, cytokine and chemokine profiling, and antibody and T cell responses at pre-defined study time points on days 0 (prior to both gentamicin/placebo administration and YF17D challenge), 1, 4, 6, 14 and 30 post-YF17D challenge (c) CONSORT diagram showing the flow of study participants through the randomized controlled-trial. Clinical trial 33 healthy adult volunteers aged 21–50 were enrolled, and randomized in a 1:1 ratio to receive either intravenous gentamicin 5 mg/kg (n = 16) or placebo (n = 17) on day 0, followed by subcutaneously-administered YF17D challenge infection one-hour later (Fig. 1b). One participant in the placebo arm was withdrawn due to non-attendance at study visits on days 4 and 6 post-YF17D challenge (Fig. 1c), time-points critical for RNAemia measurement. Baseline demographics such as age, sex, body mass index and race were similar between the two arms, as was dengue serostatus, a surrogate marker for prior orthoflaviviral immunity (Table 1). Table 1Baseline demographics of study participants in the randomized controlled clinical trialGentamicin (n = 16)Placebo (n = 16)Median age in years (range)43(22–52)36(26–48)Sex, n (%)Male14(88)12(75)Ethnicity, n (%)Chinese12(75)9(56)Indian3(19)4(25)Other1(6)3(19)Body mass index, BMI (kg/m2)Mean (± SD)26.9 (7.5)27.5 (5.8)Anti-dengue IgG ELISA, n (%)Positive8(50)7(44)Negative8(50)9(66)n number of participants; SD standard deviation Baseline demographics of study participants in the randomized controlled clinical trial n number of participants; SD standard deviation Parenteral gentamicin does not reduce RNAemia or rates of systemic symptoms after YF17D challenge infection The primary end point of the study was the proportion of participants with detectable RNAemia post- YF17D challenge infection. Viral RNAemia was measured on days 1, 4, 6 and 14 post-challenge infection. Here, we found no differences in the proportion of participants with detectable RNAemia between the gentamicin and placebo arms (62.5% [n = 10/16] versus 68.8% [n = 11/16], p > 0.99) (Fig. 2a). There were also no differences in mean RNAemia levels (3.93 log10 area under the curve [AUC] versus 3.51 log10 AUC, p = 0.81) (Fig. 2b) or kinetics of viral replication (Fig. 2c). Consistent with the RNAemia findings, overall rates of systemic symptoms reported by participants post-YF17D challenge were similar between the gentamicin and placebo arms (56.3% [n = 9/16] versus 25% [n = 4/16], p = 0.149) (Table 2), although gentamicin-treated individuals experienced increased rates of central nervous system symptoms such as headache, dizziness and facial paraesthesia (31% [n = 5/16] versus 0% [n = 0/16], p = 0.04) (Table 2). It is likely that these CNS-related symptoms, which are known adverse effects of gentamicin8, were related to gentamicin rather than YF17D infection. Comparable RNAemia, antibody and T cell responses after YF17D challenge between gentamicin and placebo group. a Proportion of participants in each intervention group with detectable YF17D RNAemia post-challenge. b Dot plot of individual RNAemia AUC values in the gentamicin (red circles; n = 16) and placebo (blue squares; n = 16) groups. Black horizontal bar represents mean. c Kinetics of viral RNAemia post-challenge in gentamicin (n = 16) and placebo (n = 16) group. d Dot plot of neutralizing antibody titers as measured by plaque-reducing neutralizing test (PRNT) at day 30 post-YF17D challenge in gentamicin (red circles; n = 16) and placebo (blue squares; n = 16) group. Black horizontal bar represents mean. e Total antigen-specific T cell response post-YF17D challenge as measured by IFN-gamma ELISpot in gentamicin (red bars; mean, n = 16) and placebo (blue bars; mean, n = 16) group. Black open circles and squares represent individual participants. Two-sided Fisher’s exact test was used to compare proportions of participants with detectable RNAemia, while unpaired Student’s two-sided t-test was used for all other comparisons. ns = non-significant. Table 2Symptomatic outcomes following YF17D challengeSymptomatic outcome by system organ classGentamicin (n = 16), no. of participants with symptoms (%)Placebo (n = 16), no. of participants with symptoms(%)P-value Any symptom9(56)4(25)0.15 Systemic4(25)3(19)>0.99 Fever01 Malaise30Swollen lymph node01 Rashes10 Hot Flushes10 Chills11Central Nervous System5(31)00.04* Headache30 Dizziness10 Facial Paresthesia10Musculoskeletal System3(19)2(13)>0.99 Myalgia21 Arthralgia01 Backache10Respiratory System01(6)>0.99 Sore throat01 Nasal congestion/discharge01 Cough01Cardiovascular System1(6)0>0.99 Chest Pain10Gastrointestinal System2(13)00.48 Diarrhoea10 Oral ulcer10 Abdominal Discomfort/bloating20Number of participants in each group who reported systemic or organ-class symptoms post-YF17D challenge. Two-sided Fisher’s exact test. n: number of participants. Symptomatic outcomes following YF17D challenge Number of participants in each group who reported systemic or organ-class symptoms post-YF17D challenge. Two-sided Fisher’s exact test. n: number of participants. As viral replication after YF17D infection is known to drive both humoral and cellular immune responses18,25, we measured both neutralizing antibody titres against YF17D, and the YF17D-specific T cell response. Neutralizing antibody titres were quantified using plaque-reduction neutralization assay (PRNT50) on day 30 post-challenge, while T cell responses against YF17D capsid, pre-membrane, envelope, non-structural 1 (NS1), NS2A/B, NS3, NS4A/B and NS5 proteins were assessed by interferon gamma (IFN-γ) ELISpot on days 14 and 30 post-challenge. No difference in post-challenge neutralizing antibody titres (mean log10 PRNT50 titer: 3.89 versus 3.71, p = 0.37) (Fig. 2d), the magnitude of the total antigen-specific T cell responses (mean SFC/106 cells: 2463 versus 2390, p = 0.91) (Fig. 2e), or the magnitude of the T cell responses against each individual YF17D protein (Figure S2), was found between the two study arms. Parenteral gentamicin modulates systemic innate immune gene expression and chemokine expression Despite no observed differences in post-challenge YF17D RNAemia, symptom rates, or humoral and cellular immune responses between the gentamicin and placebo arms, we asked if systemic gentamicin was nevertheless able to exert early immunomodulatory effects at a transcriptional level. Here, we performed peripheral blood whole genome RNA sequencing to interrogate changes in innate immune gene profiles of participants in both arms. Gene expression profiles were analysed at baseline (day 0; prior to both gentamicin/placebo administration and YF17D challenge), and on days 1, 4 and 6 post-YF17D challenge. Given the short one-hour interval, we did not assess gene expression changes between gentamicin (or placebo) administration and subsequent YF17D challenge. We focused our analysis on genes within the TLR3-IRF3/7 signalling pathway, as prior studies have shown that these genes, particularly ISGs, are induced by topically administered aminoglycosides6,7. Despite randomisation of study participants, several of these genes such as MX1, OASL, RSAD2, USP18 and ZBP1, were significantly enriched at baseline in the placebo group compared to the gentamicin group (Fig. 3a). To account for this baseline difference, we analysed the mean log2fold change (FC) of gene counts on days 1, 4, and 6 compared to baseline (day 0) (Fig. 3b–j). As early as day 1, IRF3 genes and the ISG RSAD2 were significantly enriched in the gentamicin but not the placebo arm (Fig. 3c, h). By day 6, although there was overall upregulation of most genes within the TLR-IRF3/7 signalling pathway, likely driven by YF17D infection, the ISGs OASL and RSAD2 were significantly more enriched in the gentamicin arm compared to placebo (Fig. 3g, h). In contrast, upregulation of IRF3 genes was no longer observed in both arms at day 6 (Fig. 3c). TLR-IRF3/7 signalling pathway-related gene expression after YF17D challenge. a Mean log2signal of CXCL10, IRF3, IRF7, MX1, OAS1, OASL, RSAD2, USP18, and ZBP1 genes in gentamicin (n = 16, red bars; mean) and placebo (n = 16, blue bars; mean) arms at day 0. b-j Mean log2fold change (FC) of (b) CXCL10 (c) IRF3 (d) IRF7 (e) MX1 (f) OAS1 (g) OASL (h) RSAD2 (i) USP18 (i) ZBP1 genes in gentamicin (n = 16, red bars; mean, whiskers; standard deviation) and placebo (n = 16, blue bars; mean, whiskers; standard deviation) arms. Expression of selected genes at days 1, 4, and 6 post-YF17D challenge were compared with the participants’ baseline transcripts on day 0 (pre-challenge). Red circles and blue squares represent individual participants. Unpaired Student’s two-sided t test. P-values are presented. ns = not significant. To corroborate our gene expression findings, we compared changes at the protein expression level in both arms. Here, Olink chemokine/cytokine profiling of peripheral blood was performed at baseline (day 0), day 1, 4, 6 and 14 post-YF17D challenge (Fig. 4, Figure S3). CCL2 (Fig. 4a), CXCL12 (Fig. 4b), and IL-27 (Fig. 4d) levels were significantly higher in the gentamicin arm at day 4 post-challenge, although the differences were no longer significant by day 6. In contrast, FLT3LG cytokine levels on day 1 post-challenge were significantly lower in the gentamicin arm (Fig. 4c). Overall, our results suggest that parenteral gentamicin is able to transiently induce systemic innate immune responses at a transcriptional level, as well as influence systemic chemokine and cytokine production post-YF17D challenge. Differences in plasma cytokine and chemokine concentrations post-YF17D challenge in gentamicin versus placebo group. Plasma concentration of (a) CCL2, (b) CXCL12, (c) FLT3LG and (d) IL27 in gentamicin (n = 16; red bars; mean) and placebo (n = 16; blue bars; mean) groups. Olink analysis was performed on days 0 (pre-challenge) and days 1, 4, 6, and 14 post-YF17D challenge. Red circles and blue squares represent individual participants; horizontal dotted line depicts assay limit of quantitation (LOQ). Student’s two-sided t-test, P-values are presented. ns = not significant. Discussion Previously published animal studies have demonstrated that mucosally-applied aminoglycosides can prevent viral infection through immunomodulatory mechanisms to induce a localized antiviral state6,7. This study represents, to our knowledge, the first investigation in humans of the systemic immunologic and antiviral effects of parenterally administered aminoglycosides. Although, we observed upregulation of a number of ISGs, along with increased protein expression of selected chemokines and cytokines in gentamicin -treated individuals, these responses were transient and of modest magnitude and were insufficient to confer systemic control of YF17D challenge infection. The mechanisms underlying aminoglycoside-induced antiviral responses have been explored in preclinical studies. Aminoglycosides are able to bind host dsRNA and potentially exert a synergistic effect to render them more visible to TLRs, triggering a signalling cascade that results in production of type I interferons, including IFN-α and IFN-β3–6. These cytokines subsequently bind interferon-α/β receptor subunits 1 and 2 (IFNAR1/2) on neighbouring cells, leading to induction of ISGs that inhibit viral infection. Murine studies have shown that aminoglycoside-induced ISG activation is dependent on TLR3 pathway signalling via TRIF and IRF3/76; neomycin- or kasugamycin-treated mucosa recruit XCR1⁺ conventional dendritic cells, which uniquely express high levels of TLR326, and ISG induction is abrogated in TLR3⁻/⁻, TRIF⁻/⁻, or IRF3/IRF7-deficient mice, demonstrating pathway dependence. Moreover, neomycin-mediated ISG induction can occur in the absence of type I interferon receptor signalling6, suggesting interferon-independent activation of IRF3/7 and direct ISG transcription, a mechanism that has also been described other antiviral contexts27,28. Our observation of early IRF3-related gene upregulation in peripheral blood following parenteral gentamicin is therefore mechanistically concordant with this TLR3–TRIF–IRF3/7 axis, but contrasts with preclinical data in that these effects did not culminate in a protective antiviral state. The lack of difference in our a priori selected endpoints was likely due to the modest magnitude and short duration of ISG induction, which may reflect limitations of the intravenous route or the pharmacokinetic profile of gentamicin in immune-relevant tissues8. Mucosal administration of aminoglycosides in mice achieves high local exposure at epithelial barriers and mucosa-associated lymphoid tissues6,29, where XCR1+ dendritic cells and other TLR-3 expressing cells are abundant29,30, whereas parenteral dosing in humans may not reach comparable levels in these compartments, or in the draining lymph nodes where viral replication of YF17D initially occurs25,31. Additionally, species- and tissue-specific differences in TLR3 expression patterns and downstream IRF modulation are well-described and may further affect the translatability of murine mucosal findings to systemic human responses32,33. Although our ex vivo experiments in primary human monocytes demonstrated gentamicin-induced upregulation of IL-6 transcripts, this was not observed at the protein expression level in vivo. Unlike ex vivo studies which are conducted under controlled conditions with defined gentamicin concentrations, systemic responses in vivo are influenced by pharmacokinetics, host regulatory mechanisms, and inter-individual variability, further emphasizing the importance of conducting human studies to validate findings from in vitro and even preclinical models. In addition, serum cytokine measurements in the clinical trial were obtained at predefined timepoints, which may not coincide with the peak of cytokine induction and could vary between individuals. Interestingly, we observed increased levels of IL-27 was only reported as being increased in the gentamicin group on day 4 post-challenge. IL-27 is known to promote proliferation, differentiation and maintenance of CD8 + T cells34. Consistent with this, topical application of the neomycin to vaginal mucosa of mice has been shown to increase recruitment of tissue-resident CD8 + T cells29. Although we did not observe an increase in the overall magnitude of the YF17D-specific T cell response in gentamicin-treated individuals, we did not specifically assess CD8⁺ T cell phenotype or function. It therefore remains plausible that gentamicin-induced immune modulation may influence qualitative aspects of the CD8⁺ T cell response that were not captured in our analyses. Our study has a number of limitations. First, we did not assess gene expression or cytokine levels between drug administration and viral challenge due to the short one-hour interval, precluding a clearer distinction between gentamicin versus virus-induced responses. Secondly, unlike studies in mice in which aminoglycosides were dosed repeatedly, participants in our study received only a single dose of gentamicin prior to challenge, with complete drug clearance by 24 h post administration; it remains possible that a prolonged dosing regimen, might have resulted in measurable clinical or virological effects. However, repeated parenteral administration of gentamicin is unlikely to be practical as a prophylactic strategy, particularly in the context of a pandemic, due to logistical constraints and the potential for unacceptable renal and vestibular toxicity8. Thirdly, serum gentamicin levels were not measured in participants during the clinical trial, precluding assessment of a dose-response relationship and limiting our ability to directly relate individual systemic gentamicin exposure to the observed transcriptomic and cytokine responses. Lastly, we focused our transcriptomic analyses on a predefined set of innate immune genes based on findings in prior published pre-clinical studies6,7, and broader immunologic effects of gentamicin, may have gone unrecognized. In conclusion, this study extends upon prior preclinical findings, and suggests that parenteral administration of the aminoglycoside gentamicin in humans can induce modest activation of systemic innate immune responses, but these responses were insufficient to confer virologic protection against systemic YF17D infection. These findings emphasize the importance of route, timing, and context when considering repurposing aminoglycosides for antiviral indications. Although unlikely to be of use as a prophylactic agent against systemic viral infection, further investigation into mucosal delivery strategies and optimized dosing regimens may yet support the use of aminoglycosides as prophylactic immunomodulators against mucosally-transmitted viral respiratory infections, especially in a pandemic setting. Supplementary information Transparent Peer Review file Supplementary Figs. Description of Additional Supplementary Data Supplementary Dataset 1 Acknowledgements The authors would like to acknowledge all study coordinators and staff of the SingHealth Investigational Medicine Unit, for their dedication to the study, as well as all study volunteers. This study was funded by the National Medical Research Council (NMRC), Singapore (Award No: MOH-000337) awarded to SK. SK and EEO hold career awards from the NMRC, Singapore (Award No: MOH-001869-00 and MOH-001271-00 respectively). The funders had no role in the design, conduct, analysis or reporting of the trial. Author contributions SK and EEO conceived of and/or designed the study. SK, YFZC and DLN ran the clinical trial. ESG performed the ex vivo experiments on human primary monocytes. CYT, JXY and HCT conducted the assays. PS performed the analysis under the supervision of SK, EEO, EZO and AS. SK obtained funding for the project. PS wrote the first draft of the manuscript. All authors reviewed and approved of the manuscript. Peer review Peer review information Communications Medicine thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available. Data availability The data that support the findings in this study are contained in the paper. Source data underlying Figs. 1–4 can be found in Supplementary Data 1. Raw and processed bulk RNA-sequencing data have been deposited in ArrayExpress under accession E-MTAB-16858. Competing interests SK has served in advisory capacities for Takeda and Janssen Pharmaceuticals on dengue. EEO has served in various advisory capacities for Sanofi Pasteur, Takeda, MSD, Janssen Pharmaceuticals, Novartis and Arcturus Therapeutics on dengue/vaccines. All other authors have no competing interests to declare. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Punitha Sivanandam, Ayesa Syenina. Supplementary information The online version contains supplementary material available at 10.1038/s43856-026-01678-4. References Associated Data Supplementary Materials Transparent Peer Review file Supplementary Figs. Description of Additional Supplementary Data Supplementary Dataset 1 Data Availability Statement The data that support the findings in this study are contained in the paper. Source data underlying Figs. 1–4 can be found in Supplementary Data 1. Raw and processed bulk RNA-sequencing data have been deposited in ArrayExpress under accession E-MTAB-16858.