A CRISPR activation screen identifies CH25H as a restriction factor against influenza viruses by targeting accessible cholesterol

ABSTRACT

Influenza A viruses (IAVs) cause severe outbreaks with high mortality in birds and humans. A deeper understanding of cell-intrinsic defense mechanisms against influenza viruses is therefore crucial for developing novel antiviral strategies. Herein, we perform a genome-wide CRISPR activation screen to systematically elucidate host restriction factors against influenza A (H7N9) virus. Among multiple candidates, cholesterol 25-hydroxylase (CH25H) is shown to be induced by influenza virus infection and inhibit viral membrane fusion. Notably, our previous work demonstrated that CH25H blocks the entry of plasma membrane-fusing viruses such as coronaviruses. This inhibition occurs by relocating accessible cholesterol from the plasma membrane (PM) to the endoplasmic reticulum (ER). Here, we extend this finding and show that the same mechanism works against endocytosis-dependent viruses such as influenza viruses. The exogenous supplementation of cholesterol can restore depleted accessible cholesterol and reverse the CH25H-mediated restriction. Additionally, we prove that acyl-CoA:cholesterol acyltransferase (ACAT) is required to recruit the accessible cholesterol in this process. However, how hydrophobic accessible cholesterol is transported remains unclear. Here, we demonstrate that GRAMD1/Aster-mediated non-vesicular cholesterol transport is utilized to mobilize accessible cholesterol upon stimulation of CH25H. 25-hydroxycholesterol (25HC), the catalytic product of CH25H, is a natural metabolite that potently inhibits influenza virus infection both in vitro and in vivo. These findings underscore the promising therapeutic potential of 25HC against influenza viruses.

Introduction

Influenza A viruses (IAVs) undergo frequent genetic recombination and antigenic drift in both wildlife and domestic animals, resulting in widespread seasonal and annual outbreaks across the globe [1]. Since 2016, China has reported five successive waves of human infections with highly pathogenic avian influenza A(H7N9) virus, with a case-fatality rate as high as approximately 40% [2]. Mandatory vaccination has effectively curtailed the transmission of H7N9. However, its persistent circulation in poultry and evolutionary potential necessitate ongoing surveillance. H7N9 continues to pose a pandemic risk and therefore warrants sustained monitoring. Currently, high-throughput influenza virus screens predominantly use gene loss-of-function methods to define host factors essential for infection [3]. In recent years, some host factors with anti-influenza virus activity have also been identified, most notably interferon-stimulated genes such as IFITM3, BTN3A3, and MX1 [4–6]. The CRISPR synergistic activation mediator (CRISPR SAM) is designed to recruit transcriptional activators. This system provides a highly effective platform for genome-wide gain-of-function screens, complementing standard loss-of-function studies [7]. Significantly, the low overlap rate of common hits across diverse high-throughput screens indicates that many antiviral genes may still be undiscovered. This underscores the critical need for further research to deepen our understanding of host antiviral mechanisms. Notably, there is a lack of systematic screening studies focused on restriction factors against highly pathogenic H7 subtype influenza viruses.

In this study, we performed a genome-wide CRISPR activation screening with a human origin of H7N9 influenza virus. CH25H is an interferon-stimulated gene (ISG) known to restrict a broad range of viral infections [8]. CH25H catalyzes the conversion of cholesterol to 25HC, an oxysterol with broad antiviral activity [9,10]. However, it has been reported that CH25H knockout effectively protects mice against the H1N1 challenge. Interestingly, we previously reported that CH25H blocks plasma membrane-fusing viruses like coronaviruses by moving accessible cholesterol from the PM to the ER [11]. In contrast, influenza virus is a classic endocytosis-dependent virus. Therefore, the exact association between CH25H and influenza virus has not been fully elucidated to date.

Here, we showed that CH25H inhibits viral membrane fusion to restrict entry of endocytosis-dependent viruses, including influenza virus. We also demonstrated that acyl-CoA:cholesterol acyltransferase (ACAT) is required for recruiting accessible cholesterol in this process. Additionally, we have refined our model to show that 25HC exploits GRAMD1/Aster proteins for non-vesicular cholesterol transport. By mobilizing accessible cholesterol through this pathway, 25HC effectively inhibits viral membrane fusion. Furthermore, we demonstrated the in vivo antiviral efficacy of 25HC against H7N9 using a mouse model. These findings highlight the potential of 25HC and its underlying mechanism for future therapeutic development.

Material and methods

Cells

Human lung adenocarcinoma cells (A549, ATCC, #CVCL_0023), human embryonic kidney cells (HEK-293 T, ATCC, #CVCL_0063), Madin-Darby canine kidney cells (MDCK, ATCC, #CVCL_0460), human hepatocellular carcinoma cells (Huh7, ATCC, #CVCL_0336), African green monkey kidney cells (Vero, ATCC, #CVCL_0574), and immortalized murine bone marrow-derived macrophage (iBMDM, ATCC, #CVCL_M688) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, #11965092). Mouse alveolar macrophage cells (MHS, ATCC, #CVCL_2744) were grown in Roswell Park Memorial Institute medium (RPMI 1640, Gibco, #11875093). All cells were grown in medium containing 10% fetal bovine serum (v/v, FBS, ExCell Bio, #FSP500) and 1% penicillin–streptomycin. To establish CH25H-deficient A549 cells, sgRNAs in Table S1 were inserted into the lentiGuide-Puro backbone. The efficiency of CH25H knockout was confirmed by western blot using anti-CH25H antibody (Bioss, #bs-6480R). For generation of A549 cell lines stably overexpressing CH25H, the encoding region of CH25H was amplified into Myc tag lentiviral expression vector (pCDH) and confirmed with anti-Myc antibody (Proteintech, #16286). Stable cell lines were generated through puromycin selection.

Viruses

The influenza virus strains used were H1N1 (A/Puerto Rico/8/1934), H3N2 (A/Hong Kong/1/1968), H3N8 (A/Henan/4-10/2022), H5N6 (A/Duck/Sicuan/21957-1/2021), H7N9 (A/Guangdong/Th008/2017), and H9N2 (A/Chicken/Guangdong/V/2008). Viruses were grown in 9–11 d-old chicken embryos, then titrated by TCID₅₀ assay on MDCK cells [12]. ZIKV was produced in Vero cells, and supernatant was harvested after 48 h. Viral titer was assessed using plaque assay on Vero cells. The experiments with H1N1, H3N2, H3N8, and H9N2 were done in a BSL-2 lab, but those involving H5N6 and H7N9 were performed in a BSL-3 lab.

Plasmids

Plasmids encoding HA of strains A/Puerto Rico/8/1934, A/Duck/Sicuan/21957-1/2021, A/Guangdong/Th008/2017, and A/Chicken/Guangdong/V/2008, as well as NA of strain A/Guangdong/Th008/2017, were constructed previously. This RNA was reverse transcribed, followed by the amplification of viral genomes with universal primers as described by Hoffmann [13]. CH25H sequence was amplified by CH25H gene-specific primers (Fw: 5’-ATGAGCTGCCACAACTGCTC-3’, Rv: 5’-TGCATCTGTCCCAGCGCGGTGA-3’) from cDNA of A549 cells and cloned into the Myc-tag pCAGGS expression vector.

A549-CRISPR activation library preparation and screening with H7N9

The A549-CRISPR activation library cells were generated using three vector lentiviral CRISPR SAM systems [14]. Briefly, 8 × 10⁸ A549-dCas9 polyclonal cell lines were infected with lentivirus containing a CRISPR activation library at a multiplicity of infection (MOI) of 0.3, covered at 1000-fold. Next, the A549-CRISPR activation library was then challenged with H7N9 (MOI = 2), until all negative control cells died after infection. After enrichment, the second infection screen was continued and repeated once. Three independent genome-wide screens were performed. In survival and control cells, sgRNAs were tagged with barcodes, and subjected to deep sequencing (Novogene Co., Ltd.). NGS enrichment data were analyzed using the MAGeCK algorithm. The fold enrichment of individual sgRNAs was quantified through a direct comparison of their read frequencies in experimental versus control samples.

Functional and signaling pathway enrichment analyses based on Gene Ontology (GO) terms were performed using the Metascape online platform. Functional enrichment analysis was performed by mapping enriched genes to the GO database, followed by identification of significantly enriched GO terms with P < 0.05.

Validation of the candidate genes

Full-length cDNAs of the candidate genes were PCR-amplified from HEK-293 T total RNA-derived cDNA and subcloned into a Myc-tagged pCAGGS vector. HEK-293 T cells transfected with empty vector (EV)-Myc or candidate genes were infected with H7N9 for 1 h at MOI = 1; then, cell lysates were harvested at 12 h p.i. to assess mRNA expression of AIV-NP.

siRNAs directed against the candidate genes, along with non-targeting control siRNAs (siNC), were obtained from Ribo Bio (Guangzhou, China). HEK-293 T cells were transfected with 50 nM siRNAs using Lipofectamine RNAiMAX (Invitrogen, #13778075) following the manufacturer’s protocol. For each target gene, a mixture of three independent siRNAs was utilized. Following 48 h of incubation, the cells were inoculated with H7N9 influenza virus at an MOI of 1. Cellular samples were collected at 12 h p.i. for the analysis of AIV-NP mRNA expression.

Virus infection

Cells were infected with H1N1, H3N2, H3N8, H5N6, H7N9, or H9N2 with different MOIs for 1 h [15]. After removing viruses, cells were maintained in maintenance medium (DMEM, 0.2% BSA). The cells were collected for TCID₅₀ or immunofluorescence.

Lung organoid infection

Human iPSC-derived lung organoids were produced following previously reported methods [16,17]. Human lung organoids were infected with viruses at MOI = 2. Upon removal from the viral inoculum, encapsulate the cells in Matrigel and maintain them in CKDCI medium. Cell lysates were collected at 24 h p.i. to assess mRNA expression of CH25H.

Multi-omics in situ pairwise sequencing (MiP-seq)

4–6 week-old female BALB/c mice were intranasally infected with 10⁶ TCID₅₀ in 40 μL of H7N9. Lungs were harvested for the analysis of MiP-seq at 4 d p.i. Mouse lungs were dissected and fixed in 4% PFA, then stored in 30% sucrose in PBS for cryopreservation. Subsequently, MiP-seq probe design and in situ RNA MiP-seq were kindly completed by Gang Cao (State Key Laboratory of Agricultural Microbiology, Huazhong Agricultural University) [18].

Immunofluorescence

Cells were fixed with 4% PFA and then incubated with PBS containing 0.5% Triton X-100 and 5% BSA for blocking. Antibody staining was performed using anti-M1 antibody (ATCC, #M2-1C6-4R3), followed by incubation with anti-mouse IgG 488 AlexaFluor (Invitrogen, #A56869). Nuclear staining was achieved using DAPI (Beyotime Biotech, #C1005).

RNA isolation and qRT-PCR

Total RNA was isolated using the RNAfast200 kit (Shanghai Feijie, #220011). The cDNA was generated using a reverse transcription kit (TaKaRa, #2641A) with random primers for influenza virus NP or M1 mRNA (5’-AGCAAAAGCAGG-3’) and with specific primers for influenza virus NP vRNA (5’-GGCCGTCATGGTGGCGAATGAATGACAATGGACGGAAAACAAG GATTGC-3’). Quantitative RT–PCR was conducted with the ChamQ Universal SYBR qPCR Master Mix (Vazyme, #Q711) with one cycle at 95°C for 30 s, followed by 40 cycles of 95°C for 10 s and 60°C for 30 s, on the Applied Biosystems 7500 qPCR cycler. Primers used in qRT-PCR were listed in Table S2.

CH25H knockdown rescue assays

A549 cells were transduced with siCH25H (Table S3) for 48 h, then treated with medium containing DMSO or 20 μM 25HC (MedChemExpress, #HY-113134) for 12 h. Cells were infected with H7N9 influenza virus, followed by cell lysis and qPCR analysis after 12 h.

In vivo virus infection

BALB/c female mice aged 4–6 weeks were intraperitoneally injected with 25HC (30 mg/kg) or DMSO vehicle. After 12 h, they were intranasally infected with H7N9 virus (A/Guangdong/Th008/2017) titrated to 10⁶ TCID₅₀ [19]. Daily treatment continued for 7 days. Body weight and survival rate were monitored daily after infection. Animals were sacrificed if their body weight fell below 75% of baseline, adhering to ethical animal welfare standards. Serum was collected from seven randomized mice in each group on the day before infection and on day 2 after infection. Lungs were harvested for the analysis of the viral titer, mRNA abundance, immunohistochemistry and H&E staining at 4 d p.i.

Measurement of total cholesterol concentration

Total cholesterol was assessed via a COP-PAP based enzymatic assay (Solarbio, #BC1985). Serum or standard samples (10 μL) were reacted with 90 μL of assay reagent for 30 min at 37°C. Measure the absorbance at 500 nm [20].

Binding and internalization assay

For virus binding, A549 cells transfected with EV-Myc or CH25H-Myc were infected with H7N9 (MOI = 10) for 1 h at 4°C. Unattached virus was removed by washing with cold PBS. The virus binding was determined by measuring vRNA level by qRT-PCR [21]. Following viral attachment at 4°C for 1 h, cells were shifted to 37°C for an additional 1 h to permit viral entry. The unattached and noninternalized viruses were removed by incubation with 0.25% trypsin at 4°C for 15 min. The virus internalization was determined by measuring vRNA level by qRT-PCR.

Acid-bypass assay

A549 cells were treated with 20 μM 25HC or DMSO for 12 h. Then cells were infected with H7N9 (MOI = 10) for 1 h at 4°C. After viral attachment, cells were washed with cold PBS, and exposed for 2 min at 37°C to either an acidic fusion buffer (50 mM citrate, 154 mM NaCl, pH 5.0) or a neutral control buffer (20 mM HEPES, 154 mM NaCl, pH 7.4). Virus stop solution (50 mM HEPES, 20 mM NH₄Cl, pH 7.4) was subsequently applied, and cells were maintained at 37°C for 6 h. Viral fusion efficiency was determined by quantifying viral mRNA levels using qPCR [22].

Pseudovirus production and infection

HEK 293 T cells were transfected with pCAGGS-HA, pCAGGS-NA, pSPAX, and pWXLD-Luc, and then harvested 60 h [23,24]. After pseudovirus incubation for 2 h, cells were lysed with lysis buffer to measure the relative luminescence units (RLU) of Fluc activity using the Luciferase Assay Kit (YEASEN, #11401ES80).

Influenza viral polymerase activity assay

HEK 293 T cells were initially transfected with EV-Myc or CH25H-Myc for 12 h, followed by co-transfection with PB2, PB1, PA, NP, the firefly luciferase reporter (pPolI-NP-Luc), and a Renilla luciferase control. Luciferase signals were quantified by a dual-luciferase reporter assay system (Promega, #E1910) [15].

Cell–cell fusion assay

A549 cells were co-transfected with EGFP and HA, and subsequently exposed to DMSO or 25HC (20 μM). Cells were then treated with either an acidic fusion buffer (50 mM HEPES, 154 mM NaCl, pH 5.0) or a neutral control buffer (50 mM HEPES, 154 mM NaCl, pH 7.4) at 37°C for 5 min [25]. For H1N1-HA and H9N2-HA mediated cell–cell fusion assays, cells were additionally incubated with TPCK-trypsin (2 μg/mL) for 20 min [26]. Cells were allowed to recover in complete medium for 15 min, followed by fixation with 4% paraformaldehyde. Membrane fusion was quantified by measuring the mean area of fluorescent puncta from 20 randomly selected fields per group using ImageJ.

ALOD4 protein purification, labeling, and incubation

ALOD4-His (Addgene, #111026) was affinity-purified using a Capturem His-tag kit (Beyotime Biotech, #P2226) [27]. For ALOD4 binding assay, cells pretreated with DMSO, 25HC, or HP-β-CD (2%, TargetMol, #T19609) were incubated with 4 μM ALOD4 at 37°C for 30 min.

Cholesterol and K-604 dihydrochloride rescue assays

A549 cells treated with DMSO or 20 μM 25HC for 12 h were subjected to cholesterol replenishment using PBS or 40 μM water-soluble cholesterol for 30 min, followed by exposure to H7N9 pseudovirus or ZIKV for 2 h [11]. In the K-604 dihydrochloride rescue experiment, A549 cells subjected to 25HC treatment with DMSO or 20 μM K-604 dihydrochloride (TargetMol, #T11733) for 12 h were washed and maintained with DMSO or K-604 dihydrochloride for an additional 4 h before exposure to H7N9 pseudovirus. Then, the cells were analysed for luciferase activity.

GRAMD1A/B (Aster-A/B) knockdown rescue assays

A549 cells were transduced with siAster-A/B (Table S3) for 48 h, then pretreated with DMSO or 5 μM 25HC for 8 h. Aster-A/B knockdown cells were infected with H7N9 and quantified by qPCR measurement at 6 h p.i. Aster-A/B knockdown cells were co-transfected with pCAGGS expressing EGFP and HA for 24 h , and treated with DMSO or 20 μM 25HC for 12 h. Then cells were incubated in fusion buffer to observe syncytium formation.

Cell viability assay

Cells were exposed to increasing concentrations of 25HC, K-604 dihydrochloride, or U18666A. Evaluate cell viability using the Cell Counting Kit-8 after 48 h (Beyotime Biotech, #C0043).

Ethics statement

Experiments involving animals were approved by the Animal Care and Use Committee of South China Agricultural University and handled according to the applicable guidelines (2017A002).

Statistical analysis

All datasets were processed and visualized with Prism version 8.0.2 (GraphPad). Differences between groups were assessed by two-tailed unpaired Student’s t tests, while survival analysis was conducted using the Kaplan-Meier approach. P values < 0.05 are considered significant and denoted as *P < 0.05, **P < 0.01, and ***P < 0.001. Nonsignificant values are denoted as ns.

Results

Genome-wide CRISPR activation screening revealed host factors limiting H7N9 infection

As previously detailed, a gain-of-function library was used to produce an A549 cell pool for identifying genes involved in H7N9 infection and replication [7] (Figure 1A). Briefly, A549-dCas9 cells were transduced with a human CRISPR activation lentiviral library (70,290 sgRNAs) at an MOI of 0.3 [28]. Cells were challenged with H7N9 (MOI = 2). Surviving cells from both the infected and mock control groups were harvested and expanded for sgRNA sequencing. The enrichment of target genes was determined by comparing sgRNA reads from resistant cells (infection group) to those from uninfected cells. We observed robust enrichment of approximately 957 sgRNAs in the A549 activation library cells (P < 0.05) (Figure 1B, Dataset S1). Previous studies have reported that B4GALNT2 exhibits broad-spectrum inhibitory activity against influenza virus infection [7,29]. Notably, B4GALNT2 was also identified as a top candidate in our functional screening, which further underscores the reliability and robustness of our screening system.

Figure 1.

CRISPR activation screen identified host restriction factors for involved in influenza virus infection in A549 cells. (A) Schematic of CRISPRa library construction in A549 cells and screening strategy for influenza virus infection. (B) Gene enrichment analysis in A549 cells transduced with the CRISPRa library after H7N9 infection. (C) Gene Ontology functional analysis of highly enriched genes (P < 0.05). (D–E) H7N9 infection in HEK 293 T cells overexpressing (D) or with knockdown of (E) candidate genes. (F) H7N9 infection led to upregulation of CH25H in human cells, mouse cells and human lung organoids. (G) Spatial multi-omics profiling of H7N9-infected mouse lungs by in situ pairwise sequencing. Blue, Ch25 h. Yellow, Ifit2. (H) A549 cells were infected with different subtypes of influenza viruses at MOI of 2. Cells were subsequently harvested at 12 h p.i. to measure viral M1 mRNA expression levels by qRT-PCR. Mean ± SD of n = 3. ****P < 0.0001, by Student’s t-test.

Gene Ontology (GO) analysis of the candidates revealed 957 genes (P < 0.05) enriched in specific biological processes (Figure 1C). Enrichment was observed among genes linked to inflammatory signaling pathways and the type I interferon pathway. Furthermore, genes involved in lipid glycosylation and protein phosphorylation were also enriched. According to the gene list enriched from our CRISPR activation screen, we selected 20 candidate genes for further functional validation by combining their enrichment rankings and our previous research background (Figure 1D,E). As presented in Figure 1D,E, overexpression of CH25H markedly inhibited influenza virus replication, whereas knockdown of CH25H enhanced viral replication. In our previous study, CH25H blocked SARS-CoV-2 and MERS-CoV membrane fusion via 25HC-mediated cholesterol mobilization from the PM. Notably, influenza virus is a classic endocytosis-dependent virus, which is distinct from these plasma membrane-fusing viruses. Accordingly, dissecting the antiviral mechanism of CH25H will deepen our understanding of whether antiviral targets against viral membrane fusion are broad-spectrum and reliable [30–32].

Here we analyzed the data from H7N9 infection in A549 cells, human lung organoids, and two mouse cell lines (MH-S and iBMDM). CH25H expression was significantly upregulated upon H7N9 influenza virus infection (Figure 1F). Similar results were observed in MiP-seq spatial transcriptomics of infected mouse lungs. We measured spatial RNA expression of Ch25 h and Ifit2 genes at the in situ level. Influenza virus infection significantly upregulated the transcriptional levels of Ifit2, a well-characterized host restriction factor against influenza, as well as those of the Ch25 h gene in mouse lungs [33] (Figure 1G). Furthermore, infection with multiple influenza A viruses (H1N1, H3N2, H3N8, and H9N2) resulted in a robust induction of endogenous CH25H transcription (Figure 1H). Collectively, we conducted a comprehensive genome-wide gain-of-function screen targeting highly pathogenic influenza H7N9 virus and further validated multiple candidates as restriction factors. Notably, CH25H exhibited significant potential and was induced by H7N9 infection.

CH25H acted as a restriction factor against influenza virus infection

CH25H is a critical mediator of cholesterol metabolism with important roles in cellular homeostasis and immune regulation. Our recent work has also elucidated a novel antiviral mechanism of CH25H against plasma membrane-fused viruses such as coronaviruses [11]; however, it remains to be determined whether this mechanism is also effective against endocytosis-dependent viruses like influenza virus. Thus, exploring the mechanism of CH25H will clarify whether antiviral targets against viral membrane fusion are broad-spectrum and reliable.

To investigate the role of CH25H in influenza virus infection, we established a stable CH25H-expressing A549 cell line (Figure S1A). Subsequently, viral titers were measured in CH25H-expressing and wild-type (WT) A549 cells following H7N9 influenza virus infection (Figure 2A). The results showed that CH25H expression markedly inhibited viral infection. Consistent with this observation, overexpression of CH25H-Myc (Figure S1B) resulted in a significant reduction in viral mRNA (Figure 2B) and protein (Figure 2C). Conversely, depletion of CH25H led to a marked elevation in viral titers (Figure 2D, S1C). Also, siRNA-mediated knockdown of CH25H (Figure S1D) led to increased viral infection (Figure 1E,F).

Figure 2.

Identification of CH25H as a host antiviral factor against influenza viruses. (A) Multicycle replication of H7N9 influenza virus in A549 cells or cell lines stably expressing CH25H (MOI = 0.1). Viral titers were measured by TCID₅₀ assay at 48 h p.i. (B–C) A549 cells were transfected with CH25H-Myc plasmids or empty vector control for 24 h and infected with H7N9 influenza virus at MOI = 1. The relative level of NP RNA was detected by qPCR (B). NP proteins were detected by western blot (C). (D) Viral titers were measured by TCID₅₀ assay in WT and CH25H-deficient A549 cells at 48 h p.i. (E–F) A549 cells were transfected with CH25H-targeting siRNA or negative control (NC) siRNA for 48 h. Then cells were infected with H7N9 influenza virus at MOI = 1. NP and M1 were measured by qPCR (E) and western blot (F). (G) Catalytic reaction scheme of CH25H and 25HC. (H–I) Cells were exposed to DMSO or 25HC (10 μM) for 12 h before being infected with H7N9 influenza virus at MOI = 1. NP RNA (H) and NP protein (I) was analyzed. (J–K) A549 cells were treated with DMSO or 25HC, and then challenged with H7N9 influenza virus at MOI = 1, NP RNA (J) and protein (K) were quantified after 12 h. (L) A549 cells were treated with 20 μM 25HC for 0, 3, 6, 12, and 18 h prior to H7N9 infection (MOI = 0.3). Cells were subsequently harvested at 12 h p.i. to measure viral NP mRNA expression levels by qRT-PCR. (M) A549 cells transfected with CH25H or control siRNA were treated with 20 μM 25HC for 12 h prior to infection with H7N9 influenza virus. (N) A549 cells were challenged with H1N1, H5N6, H7N9 and H9N2 influenza viruses at MOI = 0.3 for 6 h after treatment with 25HC. M1 proteins (green) were probed by immunofluorescence. Scale bars, 200 μm. Fluorescence dots inhibition was analyzed by high-content cell imaging system. Inhibition efficiency =   $100-\left({\displaystyle \frac{\mathrm{total}\mathrm{fluorescent}\mathrm{dots}\mathrm{in}25\mathrm{HC}}{\mathrm{total}\mathrm{fluorescent}\mathrm{dots}\mathrm{in}\mathrm{DMSO}}}\right)\ast 100$. Mean ± SD of n =   3. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ^nsP > 0.05, by Student’s t-test.

CH25H catalyzes the conversion of cholesterol to 25-hydroxycholesterol (25HC) (Figure 2G) [34]. To investigate whether CH25H inhibited H7N9 infection via 25HC production, A549 cells were treated with 25HC at a non-cytotoxic concentration (CC₅₀ = 147.1 μM) (Figure S1E). As expected, both viral RNA transcription (Figure 2H) and protein expression (Figure 2I) were significantly reduced. 25HC treatment suppressed H7N9 infection with a degree of dose-dependent efficacy (Figure 2J,K). A time-course experiment for 25HC pretreatment demonstrated that, compared to the 0-hour control, the viral inhibition rates for the 3, 6, 12, and 18-hour pretreatment groups were 35.8%, 53.9%, 85.1%, and 89.4%, respectively (Figure 2L). To confirm that CH25H exerted its antiviral effect through 25HC, we examined H7N9 infection in CH25H-knockdown A549 cells treated with either DMSO or 25HC. CH25H knockdown significantly enhanced viral genome transcription, whereas 25HC supplementation effectively diminished this enhancement (Figure 2M). Consistent with these findings, viral protein levels also exhibited a similar trend.

We further investigated the effect of 25HC on inhibiting other influenza virus subtypes. The green fluorescence signal was significantly diminished for each influenza virus following 25HC treatment (Figure 2N). Quantitative analysis using a high-content imaging system revealed that 25HC reduced infections by 68.96%, 72.24%, 80.61%, and 75.89% for H1N1, H5N6, H7N9, and H9N2, respectively. In brief, our findings demonstrated that CH25H broadly restricts influenza A virus infection through the production of 25HC.

25HC inhibited influenza virus replication in vivo

25HC is a natural metabolite produced by the immune system. Importantly, its effects are transient and fully reversible upon removal, which contributes to its favorable safety profile in both murine and non-human primate models [32,35]. To evaluate the antiviral efficacy of 25HC, we conducted experiments using a murine influenza virus infection model. Female mice (4-6 weeks old) were administered 25HC (30 mg/kg) or vehicle via intraperitoneal injection and subsequently challenged with 10⁶ TCID₅₀ of H7N9 influenza virus through intranasal inoculation [36]. Daily monitoring of survival and body weight was performed over a 14-day period (Figure 3A). The 25HC-treated group showed a significantly higher survival rate (∼66.67%) compared to the vehicle-treated group (Figure 3B). In both uninfected groups, the mice exhibited a slight increase in body weight over time. Conversely, infected mice in the untreated group underwent a ∼30% reduction in body weight. Notably, the body weight of infected mice in the 25HC-treated group initially declined but showed a marked recovery by day 11 post-infection (Figure 3C).

Figure 3.

25HC suppressed H7N9 infection in a murine model. (A) Experimental schematic of 25HC administration and influenza virus infection in mice. (B) Survival rates of H7N9 infection with 25HC treatment in mice (n = 6, with a log-rank test [Mantel-Cox], representative of 6 mice). (C) Body weight changes of H7N9 infection with 25HC treatment in mice (n = 6, representative of 6 mice). (D) Lung indices in mouse lungs at 4 d p.i. were calculated as lung wet weight/body weight*100 (n = 4, representative of 4 mice). (E) Virus titers in lungs of mice at 4 d p.i. were determined by TCID₅₀ assay (n = 4, representative of 4 mice). (F) H&E staining of lung tissues from mice at 4 d p.i. Scale bars, 1000 and 200 μm. (G) IHC staining (anti-NP) of lung tissues from mice at 4 d p.i. Scale bars, 1000 and 200 μm. (H) Total cholesterol levels in mouse serum at day-1 and day 2. Mean ± SD of n = 7.

Viral titers and lung index from each group on day 4 post-infection indicated that 25HC treatment significantly reduced viral load and alleviated weight loss in lungs (Figure 3D,E). Hematoxylin and eosin (H&E) staining revealed decreased inflammatory infiltration in the lungs of 25HC-treated mice (Figure 3F), accompanied by lower expression levels of several innate immune factors compared to the untreated group, such as IFN-β, TNF-α, IL-6 and MX1 (Figure S2A, S2B, S2C and S2D). These findings suggested that 25HC treatment mitigated inflammatory damage. Additionally, immunohistochemistry showed a marked reduction in viral distribution in the lungs of 25HC-treated mice (Figure 3G).

25HC has been shown to modulate cholesterol metabolism. To assess its potential toxicity in mice, we measured serum cholesterol levels in mice receiving daily intraperitoneal injections of 25HC. High-dose 25HC treatment did not significantly affect serum cholesterol levels (Figure 3H). These findings indicated that daily high-dose administration of 25HC caused no obvious adverse effects or significant alterations of serum cholesterol levels. Overall, 25HC effectively inhibited viral replication without evident toxicity in vivo.

25HC broadly inhibited influenza virus entry by blocking HA-mediated membrane fusion

To investigate how CH25H restricts H7N9 infection, firstly we evaluated whether CH25H regulates influenza virus entry using a pseudovirus system. We confirmed that hemagglutinin (HA) was successfully incorporated into the surface of HIV-based pseudoviral particles encoding luciferase (Figure S3A, S3B) [37]. 25HC treatment significantly reduced pseudovirus entry (Figure 4A). Following viral entry, virions are transported to the cytoplasm, where vRNPs subsequently enter the nucleus to initiate viral replication. Using a mini-replicon assay, we found that CH25H did not affect viral genome amplification (Figure 4B).

Figure 4.

25HC broadly inhibited HA-mediated membrane fusion of influenza viruses. (A) A549 cells were treated with DMSO or 25HC, then challenged with H7N9 influenza virus pseudovirus. (B) 293 T cells received CH25H-Myc or vector control together with RNP, NP-Luc, and RL-TK plasmids. Luciferase activity was measured 24 h post-transfection. (C) A flowchart of the binding and internalization assay. (D) For binding assay, A549 cells were transfected with CH25H-Myc plasmids and infected with H7N9 influenza virus (MOI = 10) at 4°C. For internalization assay, cells were maintained at 37°C for 1 h after viral binding. The relative level of vRNA was quantified by qPCR. (E) A flowchart of the acid-bypass assay. (F) After viral binding to 25HC treated A549 cells at 4°C, cells were exposed to either pH = 5.0 or pH = 7.0 at 37°C, then replaced in stop medium. The relative level of vRNA was quantified by qPCR. Mean ± SD of n = 3. (G) 24 h after co-transfection of H7N9-HA and EGFP plasmids in A549 cells, cells were treated with DMSO or 25HC for 12 h, followed by pH = 5.0 or pH = 7.0 treatment for 5 min. The formation of syncytium was observed using fluorescence microscopy. Mean ± SD of n = 20. (H) A549 cells were co-transfected with the indicated HA subtypes and EGFP plasmids. At 24 h post-transfection, the cells were subjected to DMSO or 25HC treatment for 12 h, and were incubated with TPCK-trypsin, followed by exposure to pH = 5.0. Mean ± SD of n = 20. Scale bars, 200 μm. ***P < 0.001, ****P < 0.0001, ^nsP > 0.05, by Student’s t-test.

Then we investigated whether viral binding to the cell surface was inhibited. Wild-type (WT) and CH25H-expressing A549 cells were incubated with H7N9 influenza virus at 4°C for 1 h, followed by viral RNA quantification to detect surface-bound virus (Figure 4C). Our results indicated that CH25H exerted no effect on viral binding (Figure 4D). Next, we performed a viral internalization assay. No obvious difference in viral internalization was observed between CH25H-expressing and WT A549 cells.

After viral attachment and internalization, influenza virus is trafficked to the acidified endosomes, where low pH triggers the fusion between the viral and endosomal membrane, and releases vRNPs into the cytoplasm [38]. To investigate the role of 25HC during endosomal fusion, we employed an acid-bypass assay, which allows viral fusion exclusively at the PM by applying exogenous low pH stimulus, while canonical endosomal fusion is blocked by subsequent treatment with NH₄Cl (Figure 4E) [39]. The results demonstrated that treatment with 25HC significantly inhibited viral fusion under low pH conditions, indicating that 25HC targets the viral-host membrane fusion process (Figure 4F).

The effect of 25HC on viral membrane fusion was further validated using a cell–cell fusion assay in A549 cells (Figure S3C) [40]. While DMSO-treated cells formed multinucleated syncytia, syncytium formation was remarkably inhibited in the 25HC-treated group (Figure 4G). Given the broad-spectrum antiviral activity of 25HC, we hypothesized that it inhibits HA-mediated fusion across multiple influenza subtypes. Indeed, 25HC effectively reduced HA-mediated syncytium formation for H1N1, H5N6, and H9N2 (Figure 4H). These results suggested that 25HC blocks viral entry by preventing HA-mediated membrane fusion, independent of viral attachment and endosomal internalization.

25HC blocked HA-mediated membrane fusion through mobilizing accessible cholesterol by activating ACAT

Accessible cholesterol in the PM is critical for maintaining cholesterol homeostasis [41]. 25HC rapidly depletes accessible cholesterol from the cell membrane, acting as a host defense mechanism against bacterial infections [42] and viral pathogens [31].

We hypothesized that 25HC inhibits influenza virus entry by depleting accessible cholesterol from the PM (Figure 5A). The accessible cholesterol can be specifically monitored by domain 4 of anthrolysin O (ALOD4) [41], and 2-hydroxypropyl-β-cyclodextrin (HP-β-CD) serving as a positive control [43]. Our results showed that 25HC nearly completely depleted accessible cholesterol from the PM (Figure 5B). The addition of external cholesterol reversed the inhibitory effect of 25HC on influenza pseudovirus entry (Figure 5C,D) and preserved hemagglutinin (HA)-mediated membrane fusion (Figure 5E). Interestingly, the result of another endocytosis-dependent enveloped virus, ZIKA, was similar to that of influenza virus (Figure 5F). These findings provide solid evidence that 25HC inhibits viral membrane fusion by depleting membrane accessible cholesterol.

Figure 5.

25HC blocked viral fusion through accessible cholesterol by activating ACAT. (A) 25HC induces the depletion of accessible cholesterol on the PM. (B) A549 cells were treated with DMSO, 25HC or HP-β-CD and subsequently incubated with ALOD4. The quantities of bound ALOD4 were measured by western blot. (C–D) A549 cells were exposed to DMSO or 25HC, then treated with PBS or cholesterol. ALOD4 were measured by western blot (C). Detection of H7N9 influenza virus pseudovirus (D). Mean ± SD of n = 3. (E) A549 cells were co-transfected with H7N9-HA and EGFP plasmids, then were subjected to DMSO or 25HC treatment. After cholesterol incubation, then stimulated at pH = 5. Mean ± SD of n = 20. (F) Huh7 cells were pretreated with DMSO or 25HC, then incubated with PBS or cholesterol to infection with ZIKA. Mean ± SD of n = 3. (G) Inhibition of cholesterol transport by the ACAT inhibitor K-604 dihydrochloride. (H–I) A549 cells were treated with DMSO or 25HC in the presence of DMSO or K-604 dihydrochloride, then incubated with DMSO or K-604 dihydrochloride. Cells were then subjected to incubation with ALOD4 (H). Cells were infected with H7N9 influenza virus pseudovirus (I). Mean ± SD of n = 3. (J) A549 cells were treated with DMSO or 25HC in the presence of DMSO or K-604 dihydrochloride, followed by stimulated at pH = 5. Mean ± SD of n = 20. Scale bars, 200 μm. *P < 0.05, ***P < 0.001, ****P < 0.0001, by Student’s t-test.

We next investigated the mechanism through which 25HC facilitates accessible cholesterol transport. Previous studies demonstrated that the rapid depletion of accessible cholesterol from the PM by 25HC is primarily driven by activating ACAT (Figure 5G) [11,44]. Our results indicated that inhibition of ACAT reversed the 25HC-induced reduction in accessible cholesterol (Figure 5H, S4A), infection by influenza pseudovirus (Figure 5I) and HA-mediated syncytium formation (Figure 5J). Collectively, our findings suggested that ACAT activation by 25HC mobilizes accessible cholesterol from the PM, thereby limiting influenza virus entry.

25HC modulated cholesterol transport through GRAMD1/Aster-mediated non-vesicular pathway

A critical gap remains in our understanding of how accessible cholesterol is transported. Retrograde cholesterol transport from the PM to the ER is mediated by non-vesicular mechanisms involving transporter proteins such as Aster-A/B/C in mammals [45,46]. We next aimed to investigate how cholesterol mobilized from the PM is transported to the ER. We performed ALOD4 binding assays using a cholesterol transport inhibitor U18666A (3-β-[2-(diethylamino) ethoxy] androst-5-en-17-one), which interferes with cholesterol trafficking by targeting the structural domain of Aster proteins [47] (Figure S5A). Co-treatment with 25HC and U18666A resulted in a significant increase in accessible cholesterol at the PM (Figure 6A). Additionally, syncytium formation was markedly restored in U18666A-treated cells (Figure 6B).

Figure 6.

25HC modulated accessible cholesterol by Aster-mediated non-vesicular transport. (A–B) A549 cells received DMSO or 25HC together with DMSO or U18666A before being exposed to ALOD4 (A). After incubation in pH = 5, syncytium formation was observed (B). Mean ± SD of n = 20. (C–D) CH25H-knockdown cells were treated with 25HC. Analysis of ALOD4 bound (C) and syncytia formation (D). Mean ± SD of n = 20. (E) Cells were transfected with siRNA, then after treatment with DMSO or 25HC infected with H7N9 influenza virus to quantify NP RNA. Mean ± SD of n = 3. Scale bars, 200 μm. *P < 0.05 ***P < 0.001, ****P < 0.0001, by Student’s t-test.

Recently, Aster proteins are recruited to the ER, functioning to regulate cholesterol transport between the PM and the ER via non-vesicular pathways [48,49]. We observed that 25HC-induced depletion of accessible cholesterol was partially restored in Aster-deficient cells (Figure 6C, S5B), and HA-mediated membrane fusion was also rescued (Figure 6D). To verify the critical role of Aster in 25HC-mediated inhibition, we demonstrated that H7N9 replication was reduced to 36.39% following 25HC treatment in wild-type A549 cells, whereas viral replication was restored to 73.34% in Aster-deficient A549 cells (Figure 6E). These findings indicated that the ability of 25HC to deplete accessible cholesterol and block influenza virus membrane fusion is partially dependent on Aster-facilitated non-vesicular cholesterol transport.

Discussion

Elucidating the mechanisms of cell-intrinsic defenses can facilitate the development of innovative antiviral strategies for therapeutic purposes. Our objective is to systematically identify host-restricting factors targeting highly pathogenic H7N9 influenza virus using a genome-wide CRISPR activation screen. We showed that CH25H expression is upregulated upon influenza virus infection and limits viral propagation by blocking membrane fusion. The catalytic product of CH25H, 25HC, is a small molecule and natural compound that has demonstrated significant anti-influenza activity both in vitro and in vivo, indicating its potential for further therapeutic development. Furthermore, we refined this model and demonstrated that 25HC produced by CH25H restricts influenza virus infection through a sequential mechanism: (1) CH25H hydroxylates cholesterol to generate 25HC; (2) 25HC activates ACAT to deplete ER cholesterol, triggering the compensatory recruitment of accessible cholesterol from the PM; (3) Accessible cholesterol is transported from the PM to the ER via the Aster-mediated non-vesicular transport, causing a decrease in PM accessible cholesterol levels; (4) The resulting decrease in PM cholesterol subsequently depletes cholesterol in PM-derived endosomes, thereby restricting viral membrane fusion (Figure 7).

Figure 7.

A hypothetical model for how CH25H and the product 25HC blocked viral membrane fusion by mobilizing accessible cholesterol.

Genome-wide gain-of-function screen using CRISPR activation is a powerful tool to uncover restriction factors against multiple viruses. We performed this screen to identify potential factors restricting H7N9. Our findings revealed several previously unreported influenza-associated host restriction factors, including Rab3A and FBXL14. Rab is a key regulator of endocytosis and intracellular membrane transport and belongs to the Ras GTPases superfamily [50]. The FBXL family, serving as receptors for E3 ubiquitin ligases, may exert antiviral activity via the ubiquitin-proteasome system [51].

CH25H exhibits antiviral activity through multiple mechanisms: (1) its natural product, 25HC, interferes with virus-host membrane fusion; (2) CH25H reprograms cholesterol metabolism and inhibits viral replication through diverse mechanisms; (3) CH25H modulates the innate immune response, including enhancing IFN production [8,9,30,52,53]. Although multiple studies have confirmed that CH25H inhibits the entry of various viruses such as HIV, VSV, ZIKV and SARS-CoV, the underlying mechanism of the interaction between CH25H and influenza virus remains poorly defined. Gold et al. demonstrated CH25H’s dual role as an in vitro antiviral factor and an in vivo proinflammatory mediator that aggravates influenza-associated pathology in mice [9]. Herein, we elucidate the anti-influenza mechanisms of CH25H and confirm that its metabolite 25HC harbors in vivo antiviral activity in murine influenza infection models. The in vivo antiviral activity of 25HC is presumably underpinned by its intrinsic cholesterol-regulatory properties. Our current study confirms that 25HC blocks endocytosis-dependent influenza viruses by decreasing the accessible cholesterol. Serving as a vital structural and signaling molecule, PM cholesterol modulates membrane rigidity, fluidity, and curvature. Since endosomes derive their membranes from the PM, the depletion of PM accessible cholesterol inevitably leads to a corresponding reduction in endosomal cholesterol. Altered biophysical properties of the endosomal membrane compromise the conformational changes of viral fusion proteins and the ensuing lipid mixing. Both steps are fundamental prerequisites for successful fusion inside acidified endosomes. This hypothesis is further supported by our functional rescue experiments, where exogenous supplementation of cholesterol and inhibition of ACAT could reverse the phenotype induced by 25HC. Collectively, these findings suggest that 25HC could be utilized for the development of broad-spectrum therapeutic agents. We also found that 25HC induces the expression of ISGs such as MX1. Viral activation of IFN signaling induces CH25H expression, and 25HC forms a positive feedback loop that enhances IFN signaling and STAT1/2 nuclear translocation, thereby upregulating MX1 and amplifying the antiviral cascade.

Cholesterol is esterified by ER-localized ACAT into cholesteryl esters, which regulates intracellular cholesterol levels [54]. However, the mechanism by which 25HC mobilizes cholesterol transport remains poorly understood. Retrograde transport of intracellular cholesterol from the plasma membrane is primarily mediated by non-vesicular proteins [55]. In mammals, elevated cholesterol levels recruit GRAMD1A/B/C proteins to ER-PM junctions, facilitating the transfer of HDL cholesterol [56]. In this study, 25HC regulates viral membrane fusion by disrupting Aster-mediated non-vesicular transport. We hypothesize that there are still undiscovered cholesterol transport mechanisms involved in various steps of viral infection.

Previously, 25HC inhibited ZIKV infections in both non-human primate and murine models, confirming its safety at effective doses [32]. The excellent safety profile of 25HC can be attributed to its unique antiviral mechanism. Accessible cholesterol constitutes only 1-10% of total cholesterol, and depleting this fraction does not cause significant toxicity to host cells [57]. Moreover, this depletion is transient and reversible once 25HC undergoes metabolic degradation. In addition to 25HC, other oxysterols such as 27HC and 24HC have been shown to inhibit HIV and vesicular stomatitis virus (VSV) in vitro [58–60]. Therefore, it is feasible to identify new derivatives of 25HC that exhibit enhanced antiviral activity and improved pharmacokinetics. Moreover, 25HC acts as a pro-inflammatory oxysterol that can amplify innate immune signaling during systemic administration, potentially increasing cytokine production and inflammatory responses. Therefore, careful dose optimization and targeted delivery strategies may be required to minimize inflammation-associated side effects. Given that 25HC targets the viral entry step, a process for which few FDA-approved drugs are currently available, combining it with virus-targeting therapies could potentially enhance existing clinical antiviral treatments.

Funding Statement

This project was supported by National Natural Science Foundation of China (82472247, 32302956, 82302520 and 32461120064), the Specific University Discipline Construction Project (Grant2023B10564003), the Guangdong Basic and Applied Basic Research Foundation (2023A1515110946), the Major Project of Guangzhou National Laboratory (SRPG22-002 and GZNL2024A01008), and the Pearl River Talent Recruitment Program (2023QN10Y308).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Source data in CRISPR Activation screening are available in the Supplementary Dataset. Complementary data related to the conclusions of this study are available from the corresponding author.

Supplemental Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/22221751.2026.2651464.