Duck lncRNA lnc455 enhances RIG-I/MAVS type I interferon signaling by modulating hnRNPAB-mediated regulation of MAVS signaling

Abstract

Long non-coding RNAs (lncRNAs) regulate antiviral immunity against influenza A viruses, yet their roles in natural reservoir hosts like ducks remain unclear. In ducks, Retinoic acid-inducible gene I (RIG-I) contributes to their resistance to infection. To identify duck lncRNAs involved in RIG-I-mediated antiviral responses, we analyzed transcriptomic data from highly pathogenic avian influenza (HPAI)-infected duck lung tissues and identified lnc455 co-expressed with IFNβ. In-silico predictions suggested lnc455 could interact with RIG-I and MAVS, two key signaling proteins in the Type I IFN pathway. Functional assays in chicken cells showed that lnc455 enhances IFNβ promoter activity when co-expressed with either RIG-I or MAVS, even in the absence of viral ligand. However, our data did not show direct binding between lnc455 and RIG-I or MAVS. Instead, lnc455 was associated with Type I IFN negative regulators, including Pur-α, Pur-β, YB-1, HNRNPDL, and HNRNPAB. Notably, overexpression of HNRNPAB reduced MAVS-induced IFNβ activity and MAVS protein abundance, effects that were significantly rescued by lnc455. These findings suggest that lnc455 enhances Type I IFN signaling indirectly by modulating MAVS-associated repression rather than through direct interaction with upstream sensors. Together, our results identify lnc455 as a duck-specific lncRNA that fine-tunes antiviral Type I IFN responses through indirect regulation of the RIG-I/MAVS signaling pathway.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-42849-6.

Introduction

Long noncoding RNAs (lncRNAs) are RNA molecules over 200 nucleotides long which have garnered attention for their essential cellular functions despite lacking protein-coding capability¹. LncRNAs play critical roles in immune responses, operating through diverse mechanisms as signals, decoys, guides, or scaffolds^1–3.

LncRNAs have emerged as significant regulators of influenza A virus (IAV) infection, influencing viral replication, pathogenicity, and immune response^2–4. For instance, lncRNA negative regulator of antiviral response (NRAV) enhances IAV replication and pathogenicity by controlling the initial transcription of interferon-stimulated genes (ISGs), while lncRNA nuclear paraspeckle assembly transcript 1(NEAT1) promotes the expression of antiviral genes like interleukin-8^2,4,5. In mammalian systems, some lncRNAs modulate different stages of the RIG-I signaling pathway during viral infections, acting as molecular scaffolds or decoys^2,6,7.

RIG-I comprises two caspase activation and recruitment domains (CARDs) at the N-terminus, a DExD/H box RNA helicase domain, and a C-terminal domain (CTD) with a repressor domain (RD) at the C-terminus^2,8,9. In its inactive state, the RD shields the RNA-binding helicase domain, and the CARDs are folded inward^10,11. Upon viral infection, the CARD in RIG-I becomes exposed, enabling its interaction with E3-ubiquitin ligases like RIPLET^12–14 and TRIM25^10,15. This interaction leads to the recruitment of RIG-I to the mitochondrial antiviral signaling (MAVS) protein situated on the outer mitochondrial membrane. MAVS oligomerization activates kinases that phosphorylate the transcription factors IRF3 and IRF7, which induce the expression of interferons (IFNs) and proinflammatory cytokines^16–18. Specific lncRNAs regulate this pathway directly. For example, LncRNA Lnczc3h7a binds to the RIG-I helicase domain and TRIM25 to stabilize the activation complex, while lnc-Lsm3b dampens signaling by preventing conformational activation of RIG-I^6,7. MAVS also interacts with endogenous cellular RNAs¹⁹, further showing the importance of lncRNA-protein interactions in regulating this pathway.

Waterfowl and ducks are the natural reservoirs and hosts of IAV^20,21. In birds, IAV is classified as either highly pathogenic avian influenza (HPAI) or low pathogenic avian influenza (LPAI) based on the severity of infection in chickens^20,22,23. Ducks typically exhibit resistance to strains that are lethal to chickens. Previous evidence indicates that the presence of RIG-I in ducks may explain their reduced susceptibility to IAV compared to chickens²⁴.

In this study, we aimed to bridge the knowledge gap by identifying and describing lncRNAs potentially involved in regulating the RIG-I signaling pathway as a response to IAV infection in their natural host, ducks.

Results

Lnc455 is a duck-specific lncRNA with interferon-stimulated gene-like features and predicted involvement in the RIG-I/MAVS signaling pathway

To identify lncRNAs associated with the RIG-I signaling and Type I IFN pathway, we mined our published differential gene expression dataset from lung samples of ducks infected with H5N1 and mock infection²⁵. We compared the differential expression pattern of lncRNAs to IFNβ expression over three days post-infection. We identified 15 lncRNAs with a log₂ fold change (log₂fc) ≥ 2.5 on 1 dpi that exhibit a differential expression pattern consistent with IFNβ (Fig. 1a; S1a-S1d Tables). For clarity, each lncRNA is referred to by a simplified name using the prefix “lnc” (for long non-coding) and the last three digits of its NCBI ID (e.g., LOC101794455 as lnc455). Full ID-to-name mappings are provided in Supplementary Table S1a.

Fig. 1.

Lnc455 is a duck-specific lncRNA with interferon-stimulated gene-like features and predicted involvement in the RIG-I/MAVS signaling pathway. a Differential gene expression profiles of lncRNAs at 1, 2, and 3 days post-infection (dpi) in highly pathogenic avian influenza (HPAI)-infected lung samples, following the IFNβ expression pattern. As edgeR reports model-based log₂FC without associated standard deviations or confidence intervals, variability is not shown. Significance was defined as adjusted p < 0.05. The dotted line indicates significant upregulation (log₂FC ≥ 2.5) based on 1 dpi expression values. b RPISeq average interaction probability scores between lncRNAs and duck RIG-I, computed from both Random Forest (RF) and Support Vector Machine (SVM) classifiers. Scores are deterministic outputs of the algorithm and therefore do not include error bars (n = 1 per lncRNA-protein pair). The dotted line represents the selection threshold (≥ 0.85). c STRING analysis of predicted vertebrate transcription factors within 500 bp upstream of lnc455 and their enriched Gene Ontology biological processes. d Differential gene expression profiles of interferon-stimulated genes (ISGs) at 1, 2 and 3 days post-infection (dpi) in lung samples infected with highly pathogenic avian influenza (HPAI). Values represent log₂ fold-changes relative to mock-infected controls for significantly differentially expressed genes (adjusted p < 0.05; n ≥ 3). ISGs are shown in black, lnc455 in red, and non-ISG control genes in blue. As edgeR reports model-based log₂FC without associated standard deviations or confidence intervals, variability is not shown. lnc455 exhibits a temporal expression pattern that resembles that of ISGs, with maximal induction at 1 dpi followed by a decline toward 3 dpi. e Prediction of disorder in MAVS of human (blue), duck (orange), and chicken (grey) using IUPred3.

In mammals, lncRNAs can bind RIG-I to regulate its signaling activity. To predict whether any of the 15 duck lncRNAs could interact with RIG-I, we used RPISeq²⁶ which employs Random Forest (RF) and Support Vector Machine (SVM) classifiers trained on experimentally validated RNA-protein complexes. Each classifier outputs a probability (0–1) representing the likelihood of interaction based on sequence features. RPISeq achieved AUC values > 0.8 in benchmark datasets, indicating high predictive reliability. We averaged the RF and SVM scores for each lncRNA-RIG-I pair and applied a stringent cutoff of ≥ 0.85, consistent with high-confidence predictions in the original study. Six lncRNAs exceeded this threshold and were considered strong candidates for RIG-I binding (Fig. 1b; S1a Table). To further refine these candidates, we next examined regulatory features and expression patterns associated with Type I interferon signaling. Transcription factors (TFs) related to Type I IFN signaling bind to evolutionarily conserved ISREs located upstream of their associated genes²⁷. To identify potential transcription factors regulating lncRNA expression, we employed the MEME suite FIMO program²⁸, scanning 500 bp upstream of these genes for 879 vertebrate TF-binding motifs obtained from JASPAR²⁹. Occurrence scores with a q-value < 0.05 were considered significant. Among the 6 lncRNAs with potential binding to RIG-I, only lnc455 and lnc673 showed at least one occurrence of a TF-binding motif for any of the IRFs or STATs (S1e Table). STRING analysis of the predicted TFs showed enrichment for Type I interferon production and influenza A virus-related pathways (Fig. 1c; S1f Table), supporting ISG-like regulation. Consistent with this regulatory profile, RT-qPCR analysis showed that lnc455 was strongly upregulated at 1 day post-infection and declined thereafter, consistent with an ISG-like expression pattern that was also observed in the RNA-seq data (Fig. 1d and S1a).

To evaluate whether these predictions were supported by independent datasets, we examined published transcriptomic data from duck embryonic fibroblasts infected with Duck Hepatitis A Virus type I, given that lncRNAs may function across related viral infections engaging RIG-I-like receptor signaling. In this dataset, lnc455, but not lnc673, was significantly induced at 12 h post-infection compared with uninfected controls (log₂FC ≈ 11.03, FDR ≈ 0.04). Although both lnc455 and lnc673 exhibited predicted ISG-like regulatory features, only lnc455 showed reproducible induction across independent viral infection datasets and was therefore prioritized for downstream analyses. Supporting this prioritization, nhmmer-based homology analysis of a de novo assembly of 72 duck transcriptomes showed that lnc455 was consistently reconstructed while lnc673 displayed variable length and structure and did not match the duck RefSeq genome (ZJU1.0) predicted transcript, reducing confidence in its in silico predictions.

Lnc455 was initially predicted to associate with RIG-I; however, robustness testing using 100 dinucleotide-preserving shuffled lnc455 sequences showed that both the native sequence and all shuffled controls met RPISeq “binding” criteria (scores ≥ 0.5) with indistinguishable scores, indicating that the prediction reflects general sequence composition rather than a specific binding interface. Consistent with this, UV-crosslinking immunoprecipitation followed by RT-qPCR (UV-CLIP-qPCR) did not detect a physical association between lnc455 and RIG-I under the conditions tested (S2 Fig; S2b Table).

Given the absence of detectable lnc455 and RIG-I binding under the conditions tested, we examined MAVS as the immediate downstream adaptor of RIG-I and a potential point of pathway engagement. A recent study showed that MAVS interacts with cellular RNA through its conserved central disordered domain¹⁹. Disorder conservation analysis revealed the presence of the disordered region in duck MAVS, similar to human MAVS (Fig. 1e; S1g Table).

Lnc455 enhances type I IFN signaling

To study the effect of overexpressing lnc455 on Type I IFN signaling via the RIG-I or MAVS pathway and minimize direct binding to genomic DNA, we used chicken fibroblast cells (DF-1). Due to the evolutionary loss of RIG-I in chickens²⁴, DF-1 cells also lack endogenous RIG-I, RIPLET^30,31, and lnc455 enabling selective expression of pathway components. Supporting this, RNA central nhmmer homology analysis³² did not identify any significant homologs of lnc455 in chickens, indicating low sequence similarity between lnc455 and chicken ncRNAs (S2a Table). Additionally, the predicted secondary structure of lnc455 shows multiple stem-loops (Fig. 2a) that may mediate protein interactions^7,33–35.

Fig. 2.

Lnc455 enhances Type I IFN signaling. a Predicted RNA secondary structure of lnc455 predicted and visualized using ViennaRNA RNAfold and forna, respectively. Colors represent nucleotide positions: green = 5′ end, yellow = middle, red = 3′ end. b Overexpression of lnc455, but not empty vector (pcDNA3.1) or lnc527, enhances chicken chIFN-β (chIFN-β) promoter activity when co-transfected with duck RIG-I (dRIG-I) and duck RIPLET (dRIPLET) in DF-1 cells. Cells were stimulated with 5’-ppp-dsRNA (RIG-I ligand) for 12 h, 12 h post-transfection. Upon 5′ppp-dsRNA stimulation, lnc527 also induces a modest increase in promoter activity, although this effect is weaker than that observed for lnc455. c When stimulated at 18 h post-transfection, chIFN-β promoter activity was measured across increasing doses of 5′ppp-dsRNA (left) and summarized by construct (right). Lnc455 maintains enhanced promoter activity across ligand doses, whereas the enhancement previously observed for lnc527 at 12 h is no longer evident. At higher ligand doses (0.5–1.0 ng/µL), lnc527 significantly reduces promoter activity, consistent with non-specific effects under strong stimulation. d lnc455 enhances chIFN-β promoter activity only in the presence of activated dRIG-I. e lnc455 enhances promoter activity with dRIG-I and dRIPLET, and duck MAVS (dMAVS). Data are shown as mean ± SD from three independent experiments. (*, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; ****, p ≤ 0.0001).

To determine if lnc455 affects the RIG-I/RIPLET signaling pathway in chicken DF-1 cells, we performed dual luciferase reporter assays measuring chicken interferon-beta (chIFN-β) promoter activity. At 12 h post-transfection, cells were stimulated with 5′ppp-dsRNA, and chIFN-β promoter activity was subsequently measured (Fig. 2b). Under these conditions, lnc455 enhanced RIG-I signaling in both ligand-stimulated and unstimulated cells. Duck RIPLET^36,37, when overexpressed in chicken cells, can activate the Type I IFN pathway through RIG-I even in the absence of the viral ligand, with 5′ppp-dsRNA further enhancing the activation. To assess if this enhancement is specific to lnc455, we compared its activity with equimolar amounts of lnc527, a lncRNA that did not exhibit the IFNβ expression pattern described in the in-silico analysis. When 5′ppp-dsRNA was added, lnc527 also produced a modest increase in chIFN-β promoter activity, but this effect was weaker than that observed with lnc455 (Fig. 2b, right). When stimulated at 18 h post-transfection, the enhancement observed for lnc527 at 12 h was no longer evident (Fig. 2c, left), and at higher ligand doses (0.5 to 1.0 ng/µL), lnc527 showed a significant reduction in IFN-β promoter activity (Fig. 2c, left). This decrease likely reflects non-specific effects on protein expression or cellular stress under strong stimulation rather than targeted pathway inhibition. The increase observed at 12 h therefore appears transient, possibly reflecting effects on early signaling or feedback regulation, although the specific negative regulators involved remain unknown. Across all constructs, increasing 5′ppp-dsRNA doses continued to enhance IFN-β promoter activity (Fig. 2c, right), while lnc455 maintained higher activation at both timepoints, supporting its role as a more consistent enhancer of RIG-I/RIPLET-mediated signaling.

To determine whether lnc455 is dependent on activated RIG-I or the pathway, we co-transfected it with RIG-I or RIPLET individually. The results demonstrated that lnc455 did not enhance the activity of RIG-I or RIPLET alone (Fig. 2d). Together, these findings demonstrate that lnc455 can enhance the signaling activity of the activated RIG-I pathway. However, there is a possibility that lnc455 interacts with other components downstream of the activated pathway rather than with activated RIG-I alone. To investigate whether lnc455 interacts with other components downstream of the activated pathway, we compared chIFN-β activity when lnc455 was co-transfected with MAVS or with RIG-I and RIPLET. The results indicate a significant enhancement in MAVS signaling activity by lnc455, suggesting that lnc455 may also interact with chicken MAVS or proteins downstream (Fig. 2e).

Lnc455 shows putative associations with proteins that function as type I IFN negative regulators in other species

To investigate the molecular function of lnc455 and how it affects Type I IFN signaling, we performed Multiple Oligo-assisted RNA Pulldown with Hybridization followed by Mass Spectrometry (MORPH-MS)³⁸ to identify its potential RNA-binding protein interactors (Fig. 3a). RT-qPCR confirmed the enrichment of lnc455 in the biotinylated probe group, which consisted of ten 25nt tiling oligos targeting lnc455 (Fig. 3a,b; S3a Table). Using MORPH-MS, we identified ten chicken proteins that are significantly associated with lnc455 following stimulation with duck MAVS. These proteins, along with their corresponding gene names, include YBX1, HNRNPAB, HNRNPDL, NCL, YBX3, PURB, HMGB3, HISTONE H1, PURA, and DLST (Fig. 3c and d; S4 Fig; S3b Table). Additionally, eight non-significant hits (NPM1, HMGB1, HNRNPA2B1, H3-I, TKTL1, RPSA, RPS7, VIM) with log₂FC ≥ 1.5 were detected (S2 Fig; S3b Table). Despite overexpression, MORPH-MS did not identify duck MAVS as an interactor of lnc455 (Fig. 3a; S3c Table).

Fig. 3.

Lnc455 shows putative associations with proteins that function as Type I IFN negative regulators in other species. a Workflow of UV fixation followed by MORPH-MS and design of antisense oligos to identify proteins directly associated with lnc455. The arrows mark the target sites of the antisense oligos within lnc455. The numbered line (1–2375 bp) shows the nucleotide positions along the lnc455 sequence where the arrows align. b RT-qPCR confirms lnc455 enrichment in the group treated with antisense oligos attached to a universal biotinylated probe (ubp). c Volcano plot showing the significantly enriched proteins in the biotinylated probe group (log₂FC ≥ 1.5; adjusted p value < 0.05); data represent three independent experiments. d STRING physical subnetwork and functional pathways enriched by the proteins. e Chicken HNRNPAB (chnrnpab) overexpression reduced chIFN-β promoter activity when co-transfected with duck MAVS (dMAVS) in DF-1 cells, whereas lnc455 reversed this inhibition. Data are shown as mean ± SD from three independent experiments. (*, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; ****, p ≤ 0.0001).

Physical association subnetwork of the significant hits revealed that YBX1- Pur-α/β likely forms a physical complex with three other proteins (Fig. 3d). This complex has been described in several studies and is implicated in the negative transcriptional regulation of some genes^39,40. Moreover, Pur-α/β binds to 500 bp upstream of the human IFNB promoter^41,42. Heterogeneous nuclear ribonucleoprotein AB, HNRNPAB, also binds to lnc455. In carp, HNRNPAB was shown to act as a negative regulator of type I interferon (IFN) response by interacting with MITA, TBK1 and IRF3 to initiate their degradation⁴³. Overexpression of the chicken HNRNPAB (chnrnpab) downregulated the IFNB promoter activity induced by duck MAVS. This downregulation was rescued by the addition of lnc455 (Fig. 3e). Pairwise sequence alignment using EMBOSS Needle showed that duck (XP_038042596) and chicken (NP_001383760) HNRNPAB share 94.2% identity and 95.1% similarity, with RNA recognition motifs RRM1 and RRM2 being fully conserved, which indicates preserved RNA-binding potential.

Lnc455 modulates HNRNPAB-associated effects on MAVS signaling

Since lnc455 is a duck lncRNA, we next sought to determine whether the observed effect could be recapitulated using duck HNRNPAB. To assess the functional relevance of duck HNRNPAB, we examined its effect on Type I IFN signaling. Like chicken HNRNPAB, overexpression of duck HNRNPAB (dhnrnpab) was associated with a significant reduction in IFN-β promoter activity when co-transfected with duck MAVS in DF-1 cells (Fig. 4a). Similar to chnrnpab, co-expression of lnc455 partially restored MAVS-driven IFN-β promoter activity in the presence of duck HNRNPAB. These indicate that lnc455 can mitigate HNRNPAB-associated reductions in signaling output at the MAVS level.

Fig. 4.

Lnc455 modulates HNRNPAB-associated effects on MAVS signaling. a Duck HNRNPAB (dhnrnpab) is associated with reduced chIFN-β promoter activity when co-expressed with duck MAVS (dMAVS) in DF-1 cells, and lnc455 partially restores this activity. b Lnc455 does not enhance IFN-β promoter activity when co-transfected with duck TBK1 (dTBK1), and the reduction associated with dHNRNPAB remains unchanged. c Western blots showing dMAVS and dTBK1 expression in the presence or absence of dHNRNPAB. Lnc455 co-expression increases detectable dMAVS protein levels but not dTBK1. d Lnc455 does not alter dMAVS or dhnrnpab protein levels compared with the pcDNA3.1 control. e Lnc455 partially restores dMAVS expression when co-expressed with chnrnpab or dhnrnpab. f Western blot confirming dhnrnpab enrichment in α-FLAG pull-down compared to IgG. g RT-qPCR showed a significant difference in lnc455 copy number between α-FLAG pull-down and IgG control. h A schematic summary illustrating how dHNRNPAB is associated with reduced MAVS signaling output and how lnc455 modulates these effects at the MAVS level. Western blot images shown are representative of independent experiments: c (n = 3), d (n = 3), e (n = 2), f (n = 3). Data are shown as mean ± SD from three independent experiments. (*, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; ****, p ≤ 0.0001).

To determine whether this modulatory effect extended downstream of MAVS, we next evaluated IFN-β promoter activation driven by duck TBK1 (dTBK1). In contrast to the MAVS experiment, lnc455 did not enhance TBK1-driven IFN-β promoter activity, and the reduction associated with duck HNRNPAB co-expression remained unchanged (Fig. 4b). These findings suggest that the effect of lnc455 on Type I IFN signaling is evident upstream of, or at the level of, MAVS rather than downstream signaling nodes.

Since carp HNRNPAB was shown to interact with MITA, TBK1 and IRF3 to initiate their degradation⁴³, we next examined whether duck HNRNPAB influences the abundance of MAVS signaling components. Immunoblot analysis revealed that duck HNRNPAB co-expression was associated with reduced detectable MAVS protein levels, whereas TBK1 protein levels were unaffected (Fig. 4c). Co-expression of dhnrnpab with lnc455 did not alter its expression (Fig. 4d) and partially restored MAVS abundance in the presence of chicken or duck HNRNPAB (Fig. 4e).

To examine whether lnc455 is associated with duck HNRNPAB, we performed UV-CLIP-qPCR. Under the conditions tested, lnc455 was enriched in 3x-FLAG-HNRNPAB pull-downs compared with controls (Fig. 4f,g). While this enrichment supports an association between lnc455 and HNRNPAB in cells, it does not establish a direct interaction. Confirming direct binding would require additional approaches such as native RIP-RT-qPCR, in vitro RNA-protein binding assays (e.g., EMSA or pulldown with purified HNRNPAB), and/or mapping of binding sites using eCLIP/iCLIP or mutational analyses.

Together, these results indicate that duck HNRNPAB is associated with reduced Type I IFN signaling and decreased MAVS protein abundance, and that lnc455 can partially counteract these effects at the MAVS level (Fig. 4h).

Discussion

By integrating transcriptome profiles of HPAI‑infected duck lung with DF‑1 cell assays and MORPH‑MS pulldown, we identified lnc455 as a duck‑specific lncRNA that enhances RIG‑I/MAVS‑dependent IFNβ expression potentially through interactions with regulatory proteins Pur‑α/β, YB‑1, HNRNPDL and HNRNPAB. We report evidence that HNRNPAB expression is associated with reduced MAVS pathway activity, which lnc455 reverses in vitro, highlighting its role as an indirect enhancer of antiviral innate immunity in birds.

Type I interferons (IFNs) are crucial for the antiviral response to highly pathogenic avian influenza (HPAI), as they induce interferon-stimulated genes (ISGs) that inhibit viral replication and spread^25,44,45. Regulating IFN signaling is vital for controlling infections, especially given the rapid evolution of influenza viruses and their immune evasion strategies. Recent research highlights long non-coding RNAs (lncRNAs) as important regulators of immune responses, including Type I IFN signaling⁴⁶. We identified lnc455 as one of the most upregulated lncRNAs in the lungs of ducks infected with HPAI. This upregulation matches supplementary data from a published study showing high lnc455 expression in duck embryonic fibroblasts infected with Duck Hepatitis A Virus Type I⁴⁷. Moreover, we found that in a comprehensive RNA‑Seq-based annotation of 4,094 duck lncRNAs, XLOC_038561, which partially overlaps with lnc455 ranked among the top ten upregulated transcripts (log₂ FC) in lungs on days 2 and 3 after infection with HPAI A/duck/Hubei/49/05 (H5N1) compared to controls⁴⁸. These data suggest that lnc455 may play a role in the immune response to both influenza A and other viruses that activate RIG-I-like receptor signaling. We note, however, that other lncRNAs displaying ISG-like expression patterns were not functionally tested in this study. While this focused approach enabled detailed functional investigation of lnc455, systematic evaluation of additional candidate lncRNAs in IFNβ reporter assays may reveal other regulators of Type I IFN signaling and represents an important direction for future work. In addition, further in vivo and infection-based functional studies will be required to determine whether lnc455 is necessary for ISG induction or for controlling viral replication during physiological infection. Together, our data show that lnc455 enhances Type I IFN signaling in RIG-I and MAVS-activated cells and likely interacts with components downstream of these pathways.

Our data show that lnc455 does not directly bind to either RIG‑I or MAVS. This finding contrasts with other lncRNAs known to modulate RIG‑I signaling pathway proteins through direct interaction^2,6,7,19. For example, MAVS directly interacts with cellular RNA to activate its signalosome independently of RIG-I or MDA5, highlighting the role of host RNAs in MAVS activation¹⁹. Certain lncRNAs, like lnc-Lsm3b, directly bind RIG-I, functioning as a decoy for the ligand, preventing downstream signaling and regulating the duration of the immune response⁷. Similarly, Lnczc3h7a enhances RIG-I signaling by stabilizing the RIG-I–TRIM25 complex, facilitating K63-linked ubiquitination, and promoting antiviral immune responses⁶.

MORPH-MS revealed that lnc455 associates with several proteins involved in Type I IFN regulation, including Pur-α and YB-1. A recent study showed that the lncRNA TM4SF1-AS1 activates interferon signaling in gastric cancer cells in a Pur-α– and RIG-I-dependent manner, and is functionally associated with Pur-α, YB-1, and RIG-I, suggesting an indirect mechanism of RIG-I pathway activation⁴⁹. Given the similarity in interactors, lnc455 may perform a similar molecular function, though it is important to note that lnc455 and TM4SF1-AS1 have low sequence similarity and few conserved motifs. In addition to Pur-α, our MORPH-MS results also identified Pur-β, which can form a heterotrimer with Pur-α and YB-1⁴⁰. The Pur-α/β heterodimer represses transcription^39,40. In humans, the IFNB promoter contains a Pur-α/β binding site 500 bp upstream of the transcription start site^41,42, and while this binding site does not affect promoter activation, its role in repression remains to be explored.

Our MORPH-MS also revealed that lnc455 interacts with two heterogeneous nuclear ribonucleoproteins (HNRNPs), HNRNPAB and HNRNPDL. HNRNPAB is a protein known to attenuate Type I IFN production by promoting autophagy-mediated degradation of MITA, TBK1, and IRF3 through its two RNA-recognition motifs (RRM) domains in teleost fishes⁴³. Similarly, HNRNPDL negatively regulates interferon activation at the TBK1 level in chicken cells⁵⁰. We demonstrated that in chicken cells, HNRNPAB also inhibits Type I IFN promoter induction. In contrast with fish, our data in ducks indicates that HNRNPAB is associated primarily with reduced MAVS protein abundance, while TBK1 protein levels remain unchanged, suggesting a species-specific difference in the point of pathway regulation. How HNRNPAB mediates this reduction in MAVS remains unresolved; however, the rescue of MAVS-dependent signaling by lnc455 suggests that lnc455 counteracts HNRNPAB activity, potentially by stabilizing MAVS, modulating HNRNPAB function, or acting as a decoy RNA. By contrast, the functional interaction between lnc455 and HNRNPDL has not yet been experimentally tested. Together, these results suggest that HNRNPAB is a promising candidate for understanding how lnc455 enhances Type I IFN signaling.

Importantly, while these findings identify multiple candidate regulators associated with lnc455, only HNRNPAB was functionally evaluated in this study. The roles of Pur-α, YB-1, and HNRNPDL in DF-1 cells therefore remain speculative. Taken together, our data suggests that lnc455 modulates Type I IFN signaling through putative interactions with downstream regulatory proteins rather than through direct binding to RIG-I or MAVS. This aligns with the emerging understanding that lncRNAs fine-tune immune responses by interacting with signaling pathway components rather than primary immune sensors⁵¹.

In this study, in silico screening identified lnc455 as a highly upregulated long non-coding RNA in ducks infected with HPAI. Functional analyses showed that lnc455 enhances Type I IFN signaling, with HNRNPAB emerging as a key candidate mediator of this effect. While MORPH-MS identified several putative interactors, including Pur-α, YB-1, and HNRNPDL, only HNRNPAB was functionally validated in DF-1 cells, and its inhibition of MAVS-driven signaling was partially reversed by lnc455. The roles of the remaining candidate proteins in lnc455-mediated IFN regulation remain to be established.

Materials and methods

In silico screening

We mined our published differential gene expression dataset²⁵ to identify lncRNAs with a log₂ fold change (log₂FC) ≥ 2.5 at one day post-infection, exhibiting a pattern consistent with IFNβ over three days in lung samples of ducks infected with rgVN1203 A/Vietnam/1203/04 (H5N1) or mock infection. The expression of lnc455 was validated by RT-qPCR using cDNA synthesized from RNA samples prepared according to the viral infection and RNA collection protocols described previously²⁵. RNA-protein binding probabilities for the lncRNAs with duck RIG-I (NP_001297309.1) were predicted using RPISeq. lncRNA-duck RIG-I interactions were considered if the average of the RPISeq RF and SVM scores were greater than ≥ 0.85. Intrinsically disordered regions in MAVS proteins were identified using IUPred3 with no smoothing for disorder conservation analysis. Vertebrate transcription factor (TF) binding profiles from the JASPAR database were used to identify potential TFs binding 500 bp upstream of lncRNA genes. Motif matches were identified using MEME FIMO. Significant predicted TFs (q-value < 0.05) were analyzed using the STRING webserver against the human database to identify differentially enriched gene ontology biological processes. Chicken ncRNA homologs of lnc455 were searched using nhmmer sequence search of RNAcentral. The hits were aligned with lnc455 using Clustal Omega⁵² and visualized using NCBI MSA Viewer 1.25.0. RNA secondary structure of lnc455 was predicted and visualized using ViennaRNA RNAfold and forna, respectively⁵³.

Overlap extension polymerase chain reaction

Total RNA was extracted from a 6-week-old Pekin duck lung sample collected one day after infection with A/Vietnam/1203/2004 (H5N1) using TRIzol^® (Invitrogen, Waltham, MA, USA) as described previously²⁵. RNA samples were DNase I (Invitrogen, Waltham, MA, USA) treated to remove any DNA contamination and then used for cDNA synthesis with SuperScript™ III using Oligo (dT) primer, and random hexamers according to the manufacturer’s instructions (Invitrogen, Waltham, MA, USA). Amplification utilized primers incorporating EcoRI and NotI restriction enzyme sites, with cloning and sequencing revealing a 28 bp deletion in exon 2. The de novo assembly of 72 duck transcriptomes verified the existence of the full-length lnc455 without the 28 bp deletion. To clone the entire sequence, internal forward and reverse primers with the length and sequence of the missing 28 bp were used with the original primers to amplify two fragments from the same sample. Utilizing the 28 bp overlap, these fragments underwent overlap extension PCR to generate the full-length lnc455.

Cells and constructs

DF-1 cells were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA; CRL-12203)⁵⁴. Cells were maintained in DMEM plus 10% fetal bovine serum (FBS) and incubated at 39 °C. The pcDNA3.1/Hygro(+) (Invitrogen, Waltham, MA, USA) served as the backbone plasmid for the constructs in this work: C-terminal 3x-FLAG-tagged duck RIG-I (dRIGI), C-terminal HA-tagged duck RIPLET (dRIPLET), N-terminal V5-tagged duck MAVS (dMAVS), lnc527, N-terminal 3x-FLAG-tagged chicken HNRNPAB (chnrnpab), N-terminal 3x-FLAG-tagged duck HNRNPAB (dhnrnpab), and N-terminal 3x-FLAG-tagged duck TBK1 (dTBK1). PCR products were first cloned into pCR2.1-TOPO (Invitrogen, Waltham, MA, USA), then subcloned into the pcDNA3.1/Hygro(+) expression vector. LncRNA lnc527, chnrnpab, dhnrnpab, and dTBK1 constructs were synthesized by GenScript (Nanjing, China).

Dual luciferase reporter assays

The chicken IFN-β promoter luciferase reporter plasmid (pGL3-chIFN-β), originating from the chicken IFN2 gene²⁴, was used for luciferase activity measurement via the dual luciferase reporter assay system (Promega, Madison, WI, USA), following the manufacturer’s guidelines. DF-1 cells were seeded in either 24-well or 96-well plates. For 24-well plates, cells were seeded at 3 × 10⁵ cells per well, and transfections were performed using the amounts specified below. For 96-well plates, transfections were scaled proportionally. After 24 h, cells were transfected with fixed amounts of the IFNβ reporter plasmid (150 ng), synthetic Renilla luciferase internal control construct phRTK (10 ng), dMAVS construct (60 ng), dRIGI and dRIPLET constructs (100 ng each), and lnc455 and lnc527 (100 fmol each), chnrnpab (150 ng), dhnrnpab (150 ng), and dTBK1 (60 ng), as specified. To ensure a consistent amount of transfected DNA, varying amounts of pcDNA3.1/Hygro(+) backbone vector were added. Cells were stimulated with 5′ppp-dsRNA (InvivoGen, San Diego, CA, USA) at 12 h post-transfection using 250 ng per well (0.5 ng/µL final). For 18-hour stimulation experiments, a dose range of 0–1.0 ng/µL (0, 0.01, 0.05, 0.20, 0.5, 1.0) was used. After 24 h, cells were lysed with 100 µL/well passive lysis buffer for 30 min at room temperature. Subsequently, 20 µL of cell lysate was transferred to a 1.5-mL Eppendorf tube with 100 µL luciferase assay reagent II for firefly luciferase activity measurement using the GloMax 20/20 luminometer (Promega, Madison, WI, USA). Finally, 100 µL Stop & Glo reagent was added to the lysate mix, and Renilla luciferase activity was measured.

UV-crosslinking Immunoprecipitation quantitative PCR (UV-CLIP-qPCR)

DF-1 cells were seeded at 5 × 10⁶ cells in a 100 mm petri dish. After 24 h, cells were transfected with 3.0 µg of dRIGI, 3.0 µg of dRIPLET, 3.0 µg of dhnrnpab, and 5 µg of lnc455 using Lipofectamine 2000. After 24 h post-seeding, cells were washed with PBS and irradiated with 200 mJ of UV using Stratalinker 2400. Cells were washed twice with PBS and lysed with IP lysis buffer (20 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1.5 mM MgCl₂, 1% Triton X-100, 1 mM DTT, 200 U/ml RNaseOUT [ThermoFisher Scientific, Waltham, MA, USA]) containing a proteinase inhibitor cocktail (Roche, ThermoFisher Scientific, Waltham, MA, USA). Cell lysates were then incubated with anti-Flag-M2 monoclonal (F3165; Sigma-Aldrich, St. Louis, MO, USA) and mouse IgG2b isotype (02–6300; ThermoFisher Scientific, Waltham, MA, USA) overnight at 4 °C, followed by Dynabeads protein G magnetic bead pulldown. The beads were washed 4x with IP Lysis Buffer. Washed beads were mixed with RIP buffer (150 mM KCl, 25 Tris-HCl [pH 7.4], 5 mM EDTA, 0.5 mM DTT, 0.5% Triton X-100, 200 U/ml RNaseOUT [ThermoFisher Scientific, Waltham, MA, USA]) containing Proteinase K, followed by 40 min shaking at 37 °C to digest proteins and reverse the crosslinks^55–57. RNA was isolated using TRIzol extraction (ThermoFisher Scientific, Waltham, MA, USA).

Western blotting

Samples were subjected to sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and were subsequently transferred onto a PVDF membrane. RIG-I-3x-FLAG was detected with anti-Flag-rabbit monoclonal (1:1,000) (701629; ThermoFisher Scientific, Waltham, MA, USA) and rabbit IgG (H + L) secondary antibody (1:40,000) (31460; ThermoFisher Scientific, Waltham, MA, USA, and was visualized by chemiluminescence using Pierce ECL Western blotting substrate (ThermoFisher Scientific, Waltham, MA, USA) and iBright 1500 imager (Invitrogen, Waltham, MA, USA).

Multiple oligo-assisted RNA pulldown via hybridization followed by mass spectrometry (MORPH-MS)

Antisense (complementary) probes against lnc455 were designed using the ChIRP probe designer tool (LGC Biosearch) with a masking level of 2, minimum spacing length of 60 nt and oligo length of 25 nt. To ensure specificity, all the oligos were manually screened for non-specific binding using NCBI BLAST. A common binding sequence 5’- TTACACTCGGACCTCGTCGACATGCATT-3’ was added towards the 3’ end of all oligos. The probes with standard desalting purification were ordered from Integrated DNA Technologies (IDT, Coralville, IA, USA). The biotinylated universal oligo, which had TEG-biotin attached to both the 5’ and 3’ ends, was designed to be complementary to the common sequence was HPLC-purified and ordered from Integrated DNA Technologies (IDT, Coralville, IA, USA). Oligos were dissolved in IDT Duplex buffer to a final concentration of 100 µM. To achieve an equimolar concentration of all the complementary probes with the biotinylated oligo, 5 µL from each was pooled together, resulting in a final concentration of 100 µM. The probes were annealed to the universal biotinylated oligo at 94 °C for 2 min, followed by cooling. Annealing was confirmed using 5% agarose gel electrophoresis.

DF-1 cells were seeded at 7 × 10⁶ cells in a 100 mm petri dish. After 24 h, cells were transfected with 1.0 µg of dMAVS using Lipofectamine 2000. After 24 h post-seeding, cells were washed with PBS and irradiated with 200 mJ of UV. Cells were washed twice with PBS and lysed with IP lysis buffer (Pierce™ MS-Compatible Magnetic IP Kit) containing a proteinase inhibitor cocktail (Roche, ThermoFisher Scientific, Waltham, MA, USA). Cell lysates were then incubated with 1000 pmol of the annealed probes at 37 °C overnight at 4 °C, followed by streptavidin magnetic bead pulldown. The beads were washed according to the manufacturer’s instructions. Final wash was done with 25 mM ammonium bicarbonate, and 10% of the beads were subjected to RT-qPCR. The rest of the samples were sent to Alberta Proteomics and Mass Spectrometry Facility for mass spectrometry analysis. Differential protein enrichment, along with data filtering and normalization, was performed using the R Bioconductor package, DEP⁵⁸. The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via the PRIDE⁵⁹ partner repository with the dataset identifier PXD067094 and 10.6019/PXD067094.

Reverse transcriptase quantitative PCR (RT-qPCR)

MORPH-MS beads were subjected to Proteinase K-SDS digestion at 37 °C for 40 min while shaking at 1100 rpm followed by TRIzol extraction. RNA from UV-CLIP and MORPH-MS were treated with DNase I (Invitrogen, Waltham, MA, USA). cDNA synthesis was conducted using Superscript III. qPCR was done using the QuantStudio 3 Real-Time PCR System (Applied Biosystems, ThermoFisher Scientific, Waltham, MA, USA) and the Dynamite 2x qPCR mastermix (Molecular Biology Service Unit, University of Alberta). Primers and probes were generated through IDT Technologies.

Statistical analysis

Statistical analysis was performed using GraphPad Prism. Normality was tested with the D’Agostino and Pearson Test. Data passing normality were analyzed with Welch’s ANOVA; otherwise, the Kruskal-Wallis Test followed by Dunn’s Test was used. For experiments involving two independent variables, two-way ANOVA with appropriate multiple-comparison correction was applied. For two-group comparisons, a two-tailed unpaired t-test with Welch’s correction was applied. Data are presented as mean ± SD from three independent experiments, with p < 0.05 considered significant.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

RJL and KEM conceived the study; RJL performed the experiments, created the figures, and wrote the first draft, and KEM funded, supervised, validated, and edited the manuscript.

Funding

This work was funded by the Canadian Institutes of Health Research (CIHR) project grant PS 159442 (KEM), and CIHR project grants-PA (Pandemic Preparedness and Health Emergencies Research) PS 192122 and PS 196069 (KEM).

Data availability

All published differential expression datasets are accessible in the Supplementary Material of reference number 25 and as raw reads in the NCBI Sequence Read Archive (BioProject PRJNA767080). Mass-spectrometry proteomics data are available via the ProteomeXchange Consortium (PRIDE dataset PXD067094; DOI:10.6019/PXD067094; access with project accession PXD067094 and token fCQfDUC7dV9Q). Nucleotide sequences for lnc455 (PX068892), lnc527 (XR_005264706.1), duck RIG-I (NM_001310380.1), duck MAVS (NM_001310828.1), duck RIPLET (PV972220), duck TBK1 (XM_027456350.3), chicken HNRNPAB (NM_001396831), and duck HNRNPAB (XM_038186668.2) can be retrieved from NCBI. All codes used in this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.