Transcriptional repressor Capicua is a gatekeeper of cell-intrinsic interferon responses

collaborative

The Capicua and ATXN1L repressor complex binds an eight-nucleotide motif at interferon and interferon-stimulated gene loci to hold them repressed during homeostasis in human and mouse cells, and is degraded through EGFR-MAPK signaling early in respiratory viral entry, which relieves that repression.

Senthamizharasi Manivasagam; Julianna Han; Athmane Teghanemt; Henry Keen; Boopathi Sownthirarajan; Boyang Cheng; Abhiraj Singh; Abigail Lewis; Olivia A. Vogel; Gayathri Loganathan; Lei Huang; Maryline Panis; David K. Meyerholz; Benjamin tenOever; Jasmine T. Perez; Santhakumar Manicassamy; Priya D. Issuree; Balaji Manicassamy

2025 · Cell Host & Microbe · primary research

Senior authors

Priya D. Issuree; Balaji Manicassamy

Correspondence

Priya D. Issuree; Balaji Manicassamy

Research areas & themes

Citation

Manivasagam S, Han J, Teghanemt A, Keen H, Sownthirarajan B, Cheng B, Singh A, Lewis A, Vogel OA, Loganathan G, Huang L, Panis M, Meyerholz DK, tenOever B, Perez JT, Manicassamy S, Issuree PD, Manicassamy B. Transcriptional repressor Capicua is a gatekeeper of cell-intrinsic interferon responses. Cell Host & Microbe. 2025. Volume 33, issue 4, pages 512-528.e7. DOI 10.1016/j.chom.2025.02.017. PMID 40132591. PMCID PMC11985295.

One-sentence contribution

The Capicua and ATXN1L repressor complex binds an eight-nucleotide motif at interferon and interferon-stimulated gene loci to hold them repressed during homeostasis in human and mouse cells, and is degraded through EGFR-MAPK signaling early in respiratory viral entry, which relieves that repression.

Executive summary

Interferon and interferon-stimulated gene promoters were generally held to sit in a default off state until bound by activated interferon regulatory factors or by ISGF3, with negative regulation during homeostasis acting upstream of transcription on the sensing and signaling machinery. FOXO3 acting at the Irf7 promoter was the noted exception. The question here is whether a DNA-binding transcriptional repressor holds these loci off directly.

Starting from a prior genome-wide CRISPR screen in which Capicua emerged as a host factor for influenza A virus, the authors compared control, Capicua knockout and ATXN1L knockout human lung epithelial cells, then extended to primary human airway basal cells, conditional knockout mice and three primary mouse cell types. Loss of either complex member raised interferon and interferon-stimulated gene transcripts under mock conditions and during infection, and restricted influenza A virus replication.

RNA sequencing showed a global rise in innate immune genes rather than a few loci, ATAC sequencing showed matching increases in chromatin accessibility, and integrated analysis found CIC binding site motifs at the great majority of affected loci. Reporter assays with native and mutated motifs, and a synthetic transactivating Capicua, established motif dependence. During influenza A virus infection the complex was lost within 40 minutes in a proteasome-dependent manner, an effect reproduced by recombinant hemagglutinin or by EGF and blocked by MEK or ERK inhibitors. Capicua knockout mice lost less weight, carried 5 to 50-fold lower lung viral burden and showed smaller areas of inflammation.

Scientific context

Cell-intrinsic antiviral responses begin with RIG-I-like receptor detection of viral RNA, signaling through MAVS to activate IRF3 or IRF7, induction of type I interferon, and amplification through the interferon receptor, JAK-STAT signaling and ISGF3 acting at interferon-stimulated response elements. Aberrant activation in the absence of infection causes inflammatory disease, and the known safeguards act largely before transcription, through direct suppression of sensing components or post-translational modification, with ADAR1 editing endogenous double-stranded RNA to prevent MDA5 activation. FOXO3 repression of the Irf7 promoter was the one reported transcriptional-level exception. Capicua is an evolutionarily conserved HMG-box transcriptional repressor characterized in Drosophila, where it binds an eight-nucleotide site and is degraded through EGFR-MAPK signaling during development, and in humans it is studied mainly in cancer and neurodegeneration. The immediate precedent for this study is the authors' own genome-wide CRISPR screen for influenza A virus host factors, published by Han and colleagues in 2018 with tenOever as a co-author, in which Capicua knockout cells showed elevated interferon and interferon-stimulated genes and restricted replication of several RNA viruses.

Central question

Does a DNA-binding transcriptional repressor act directly at interferon and interferon-stimulated gene promoters to prevent their expression during homeostasis, and if the Capicua and ATXN1L complex does so, how is that repression lifted when a virus arrives?

Experimental strategy

The design proceeds from phenotype to mechanism to organism. Knockouts of Capicua and of its obligate partner ATXN1L in a human lung epithelial line establish that the phenotype belongs to the complex rather than to one protein, and complementation with each Capicua isoform confirms specificity. Separating whether the repressor acts on RIG-I-like receptor signaling or on interferon receptor signaling is done by stimulating with transfected viral RNA versus with recombinant interferon, and then genetically by making double knockouts with MAVS and with STAT1, which pins the source of the basal signal.

Whether the effect is direct is addressed on three axes that are then intersected. Transcript changes come from RNA sequencing, chromatin changes from ATAC sequencing, and candidate direct targets from the presence of the Drosophila-defined CIC binding site motif. Promoter reporters with native and with mutated motifs test sufficiency, and a synthetic Capicua in which the repressor domain is swapped for VP16 transactivator repeats tests the converse, that the same motifs can be driven rather than silenced.

The relief mechanism is approached by timing the loss of the complex after synchronized infection, by pharmacology with proteasome, MEK and ERK inhibitors, by substituting recombinant hemagglutinin or EGF for virus, and by asking whether blocking degradation blunts interferon induction in control but not in knockout cells. Conservation and physiological relevance come from a tamoxifen-inducible conditional knockout mouse, three primary mouse cell types, lung RNA sequencing and a sublethal influenza A virus challenge.

Key findings

1. Capicua knockout and ATXN1L knockout human lung epithelial cells expressed higher IFNB1 and interferon-stimulated gene transcripts under mock conditions and after stimulation with viral RNA or with type I interferon, and restricted influenza A virus replication. Complementation with either Capicua isoform restored replication and lowered the transcripts (Figure 1C and 1D, Supplementary Figures 1 and 2). Knockdown in primary human airway basal cells reproduced the effect.

2. RNA sequencing identified 800 differentially expressed genes under mock conditions and 3,608 during infection, with over 60 percent elevated in knockout cells and broad upregulation of restriction factors, antiviral transcription factors, nucleic acid sensors, cytokines and chemokines, and enrichment of IRF and ISRE motifs among the regulatory sequences (Figure 1E to 1G).

3. Capicua knockout cells showed increased expression and nuclear localization of IRF2 and IRF9 without stimulation (Figure 2A). Deleting MAVS alongside Capicua returned basal IFNB1 to control levels, while deleting STAT1 lowered it only partially (Figure 2D). The authors interpret the MAVS dependence as tonic RIG-I-like receptor activation by endogenous double-stranded RNA ligands. The MAVS requirement is demonstrated. That the ligands are endogenous retroelement-derived double-stranded RNA is an inference drawn from the literature and is not directly shown here.

4. Interferon treatment induced higher MX1 in the Capicua and MAVS double knockout than in control cells, indicating that Capicua also restrains genes induced through interferon receptor signaling (Figure 2E).

5. ATAC sequencing found 6,992 differentially accessible peaks over 4,400 genes under mock conditions, with 3,904 showing increased accessibility in knockout cells at loci for sensors, transcription factors, interferons, interferon receptors, cytokines and restriction factors, and 11,004 peaks over 5,698 genes during infection. Peaks were predominantly intronic and distal intergenic (Figure 3A to 3C).

6. Over 93 percent of the genes with differential accessibility carried the Drosophila-derived CIC binding site motif, and integrated analysis across transcript change, accessibility change and motif presence yielded 324 common genes under mock conditions and 1,108 during infection, with IRF and ISRE motifs co-enriched (Figure 3D and 3E). Motif presence is a computational assignment and is not by itself evidence of occupancy.

7. Capicua with ATXN1 repressed luciferase from a synthetic interferon beta promoter carrying six copies of the motif under RIG-I or Sendai virus stimulation, and from native MX1, IFIT1 and TRIM22 promoters, while mutation of the motifs abolished repression (Figure 4A to 4D). A synthetic Capicua with VP16 transactivator repeats in place of the repressor domain activated IRF1, IRF7, MX1, IFIT1 and TRIM22 promoters without any stimulus, and failed to do so when the motifs were mutated (Figure 4E). Together these are the strongest evidence in the paper for motif-dependent action at these promoters.

8. Capicua and ATXN1L protein fell from 40 minutes after synchronized influenza A virus infection, was blocked by MG132 or bortezomib, and occurred in primary human airway basal cells on the same timescale, before genome replication. Levels recovered gradually between 8 and 24 hours (Figure 5C to 5F). Type I interferon treatment caused transient loss at 40 to 60 minutes with recovery by 120 minutes (Figure 5G). The authors suggest the later recovery participates in restoring homeostasis after clearance. That is an interpretation.

9. Degradation occurred with multiple influenza A virus subtypes and with respiratory syncytial virus, human parainfluenza virus type 3 and Sendai virus, but not with Zika virus, encephalomyocarditis virus or Ebola glycoprotein pseudotyped vesicular stomatitis virus (Figure 6A and 6B).

10. MEK or ERK inhibition blocked infection-induced degradation, recombinant H1 or H5 hemagglutinin alone was sufficient to trigger it, and recombinant EGF caused degradation within five minutes (Figure 6C, 6D and 6F).

11. MEK or ERK inhibition reduced IFNB1 and MX1 induction by Sendai virus, interferon or viral RNA in control cells but not in Capicua knockout cells, which places the inhibitor effect on the Capicua axis (Figure 6E, Supplementary Figure 4). Adding EGF during encephalomyocarditis virus or pseudotyped vesicular stomatitis virus infection increased interferon and interferon-stimulated gene induction (Figure 6G).

12. In conditional knockout mice, mouse embryonic fibroblasts, bone marrow derived dendritic cells and macrophages all showed higher Ifnb1 and interferon-stimulated gene induction, and lung RNA sequencing found 5,296 differentially expressed genes under mock conditions with antiviral gene ontology enrichment, but only 61 differentially expressed genes at day 5 of infection, since control mice had by then induced these genes (Figure 7A to 7C, Supplementary Figures 5 and 6).

13. Following sublethal influenza A virus challenge, Capicua knockout mice lost less weight and recovered earlier, carried 5 to 50-fold lower lung viral burden, expressed 5 to 10-fold more Ifnb1 and interferon-stimulated genes, and showed smaller areas of inflammation on histopathology (Figure 7D to 7G). In mice, MEK inhibition reduced poly(inosinic-cytidylic) acid-induced interferon-stimulated gene expression in control but not in knockout animals.

Mechanistic model

The model the data support is that the Capicua and ATXN1L complex occupies CIC binding site motifs at and near interferon and interferon-stimulated gene loci, holding chromatin at those loci less accessible and basal transcription low, so that tonic RIG-I-like receptor signaling driven by host-derived ligands does not escalate into an interferon response. When a respiratory virus engages its receptor, EGFR-MAPK signaling is activated within minutes, the complex is degraded by the proteasome, and the repression is lifted before the viral genome has replicated, which primes the cell for a faster and larger response.

Several links in this chain are inferred rather than demonstrated. Direct occupancy of endogenous loci by Capicua was not measured, for example by chromatin immunoprecipitation. Motif presence was assigned computationally from a Drosophila consensus, and the functional evidence for motif dependence comes from transfected reporters rather than from the native chromatin context. The kinase or ubiquitin ligase that connects ERK activity to degradation of the complex is not identified, and the model that MAPK activation is the direct cause rests on inhibitor pharmacology plus the EGF and hemagglutinin sufficiency experiments. Why non-respiratory viruses fail to trigger degradation is attributed to differences in engagement with the MAPK pathway, which the authors state as a likely explanation rather than a finding. The identity of the endogenous double-stranded RNA ligands driving basal signaling is taken from prior literature. Finally, the causal chain from lower viral burden to milder disease in knockout mice is a correlation the authors draw, and the contribution of any developmental or immune-cell-intrinsic effects of Capicua loss is not separated.

Conceptual or technical advance

A negative regulator of the interferon system is placed at the DNA, rather than on the sensing and signaling proteins. That changes the default-off account of interferon and interferon-stimulated gene promoters into an actively repressed state with an identified repressor, a binding motif and a defined removal mechanism. It also connects an RTK-MAPK derepression circuit conserved from Drosophila development to vertebrate antiviral defense, and it supplies a rationale, which the authors offer as speculation, for why MEK and ERK inhibitors limit inflammation in severe respiratory viral infection, namely that they preserve the repressor. The observation that Capicua loss improves outcome in a sublethal influenza A virus challenge makes the axis a candidate for pharmacological tuning in either direction.

Relationship to the broader research program

The tenOever contribution is listed under investigation, alongside Maryline Panis of the same laboratory, and the study descends from a collaboration between the Manicassamy and tenOever groups, the genome-wide CRISPR screen for influenza A virus host factors published as Han and colleagues 2018, which is where Capicua first appeared. The substance of the paper touches a question the tenOever laboratory has pursued independently, namely which transcription factors set the interferon-stimulated gene program and how much of that program can run without interferon signaling. Category 3 synthesis, visible only when this paper is set beside Schmid and colleagues 2010 from the tenOever laboratory, is that the regulatory logic at interferon-stimulated response elements involves both redundancy among activating factors and, as shown here, an independent repressive layer acting through a distinct motif, so that the output at a given promoter reflects both which activators are present and whether the repressor has been removed. That combination is not drawn in either paper alone.

Related publications

• Han and colleagues 2018, Cell Reports, Genome-wide CRISPR/Cas9 screen identifies host factors essential for influenza virus replication. Predecessor. The screen, co-authored by tenOever, that identified Capicua and reported the elevated interferon phenotype followed up here.
• Vogel and colleagues 2020, PLoS Pathogens, on the p150 isoform of ADAR1 and sustained RLR signaling. Conceptual extension. From the same group, addressing the parallel safeguard against endogenous double-stranded RNA, cited here for that role.
• Schmid and colleagues 2010, Journal of Biological Chemistry. Conceptual extension. Addresses the activating side of the same promoters through IRF7 and ISGF3 motif specificity.

Limitations and boundaries

Direct binding of Capicua to the proposed motifs in native chromatin is not shown, so the mechanism rests on motif prediction, accessibility changes and transfected reporters. The motif itself is the Drosophila consensus applied to mammalian genomes, and 93 percent of differentially accessible genes carrying it raises the question of how discriminating the motif is. The human cellular work is centered on one lung epithelial line, with primary airway basal cells used only for knockdown and degradation timing. Degradation kinetics are inferred from western blots and immunofluorescence at a small number of time points, and no ubiquitin ligase or ERK substrate site is identified. The MAPK conclusion depends on two inhibitors whose selectivity is not established here, although the absence of effect in knockout cells substantially strengthens the interpretation. The virus panel that fails to trigger degradation is small and mechanistically unexplained. In mice, Capicua deletion is whole-body and tamoxifen-induced rather than lung-restricted, so cell-intrinsic and systemic contributions are not separated, and the infection data come from one strain at one sublethal dose, with group sizes of two to three animals for some RNA sequencing comparisons. Improved outcome in knockout mice was measured over a sublethal challenge and does not address whether chronic loss of this repression is harmful, which the authors themselves raise in connection with autoimmunity and cancer as speculation. The sequencing-based claims of global change are from n equal to two per group in the cell line experiments.

Audience summaries

25 words

A repressor protein sits on interferon gene promoters and keeps them quiet, and respiratory viruses inadvertently destroy it within minutes of touching the cell surface.

75 words

Interferon genes were thought to stay silent until activating factors switch them on. Here the repressor Capicua, with its partner ATXN1L, is shown to bind a short motif near these genes and hold them closed, in human and mouse cells. Respiratory viruses activate EGFR-MAPK signaling during entry, which destroys the complex within 40 minutes and releases the brake. Mice lacking Capicua controlled influenza better, with lower lung virus and less inflammation.

150 words

Aberrant interferon expression causes inflammatory disease, and the safeguards described previously act on sensing and signaling proteins rather than at the DNA. Building on a CRISPR screen that flagged Capicua as an influenza host factor, the authors show that Capicua with ATXN1L represses interferon and interferon-stimulated genes during homeostasis. Knockout cells upregulated these genes globally, in a manner dependent on MAVS, indicating tonic sensing of host-derived ligands, and showed increased chromatin accessibility at the same loci. Most affected loci carried the eight-nucleotide CIC binding site motif, and reporters bearing native motifs were repressed by Capicua and activated by a synthetic Capicua carrying a transactivation domain, with motif mutation abolishing both. Influenza A virus, respiratory syncytial virus and parainfluenza virus triggered proteasomal degradation of the complex within 40 minutes through EGFR-MAPK signaling, reproduced by hemagglutinin or EGF alone. Capicua knockout mice showed lower viral burden and less pulmonary inflammation.

Discoveries supported by this paper

Documented publication relationships

• Genome-wide CRISPR/Cas9 Screen Identifies Host Factors Essential for Influenza Virus Replication — predecessor. Direction: this paper → linked paper. stated in the Related publications section of 2025-manivasagam-transcriptional-repressor-capicua-
• Transcription Factor Redundancy Ensures Induction of the Antiviral State — conceptual extension. Direction: this paper → linked paper. stated in the Related publications section of 2025-manivasagam-transcriptional-repressor-capicua-

Pathogens

Technologies