1. Chromosomes and Gene Expression
  2. Genetics and Genomics

Mouse B2 SINE elements function as IFN-inducible enhancers

  1. Isabella Horton
  2. Conor J Kelly
  3. Adam Dziulko
  4. David M Simpson
  5. Edward B Chuong  Is a corresponding author
  1. Department of Molecular, Cellular, and Developmental Biology and BioFrontiers Institute, University of Colorado Boulder, United States
https://doi.org/10.7554/eLife.82617
Share this article
Cite this article
  1. Isabella Horton
  2. Conor J Kelly
  3. Adam Dziulko
  4. David M Simpson
  5. Edward B Chuong
(2023)
Mouse B2 SINE elements function as IFN-inducible enhancers
eLife 12:e82617.
https://doi.org/10.7554/eLife.82617
  2. Download BibTeX
  3. Download .RIS

Abstract

Regulatory networks underlying innate immunity continually face selective pressures to adapt to new and evolving pathogens. Transposable elements (TEs) can affect immune gene expression as a source of inducible regulatory elements, but the significance of these elements in facilitating evolutionary diversification of innate immunity remains largely unexplored. Here, we investigated the mouse epigenomic response to type II interferon (IFN) signaling and discovered that elements from a subfamily of B2 SINE (B2_Mm2) contain STAT1 binding sites and function as IFN-inducible enhancers. CRISPR deletion experiments in mouse cells demonstrated that a B2_Mm2 element has been co-opted as an enhancer driving IFN-inducible expression of Dicer1. The rodent-specific B2 SINE family is highly abundant in the mouse genome and elements have been previously characterized to exhibit promoter, insulator, and non-coding RNA activity. Our work establishes a new role for B2 elements as inducible enhancer elements that influence mouse immunity, and exemplifies how lineage-specific TEs can facilitate evolutionary turnover and divergence of innate immune regulatory networks.

Editor's evaluation

This important paper will be of interest to scientists studying evolutionary divergence of immune responses and those studying how transposable elements rewire transcriptional regulatory networks. Using a combination of computational and experimental approaches, this work describes a new class of rodent-specific transposons that can act as enhancers of immune genes in mice.

https://doi.org/10.7554/eLife.82617.sa0

Introduction

The cellular innate immune response is the first line of defense against an infection and is initiated by the activation of transcriptional networks that include antiviral and pro-inflammatory genes. While innate immune signaling pathways are generally conserved across mammalian species, the specific transcriptional networks are increasingly recognized to show differences across lineages (Chuong et al., 2017; Shaw et al., 2017). These differences are widely attributed to independent evolutionary histories and continual selective pressures to adapt to new pathogens (Daugherty and Malik, 2012). Understanding how innate immune systems have evolved in different host genomes is critical for accurately characterizing and modeling responses that are related to autoimmunity or involved in disease susceptibility.

Transposable elements (TEs) are increasingly recognized as a source of genetic elements that shape the evolution of mammalian innate immune responses (Chuong et al., 2017; Cordaux and Batzer, 2009). TE-derived sequences constitute roughly half of the genome content of most mammals, and are the predominant source of lineage-specific DNA. While most TE-derived sequences are degraded and presumed nonfunctional, TEs have occasionally been co-opted to function as genes or regulatory elements that benefit the host organism. In the context of host innate immunity, there are several reported examples of species-specific restriction factors that are encoded by TEs co-opted for host defense, including Friend Virus 1, Syncytin, Suppressyn, and Jaagsiekte sheep retrovirus (JSRV) (Arnaud et al., 2008; Aswad and Katzourakis, 2012; Frank et al., 2022; Lavialle et al., 2013). In many cases, TEs derived from ancient viral infections are poised for co-option since they already have the ability to bind to receptors, therefore blocking infection as a dominant negative mechanism (Frank and Feschotte, 2017).

More recently, TEs have also been identified as a source of non-coding regulatory elements that control inducible expression of cellular innate immunity genes (Buttler and Chuong, 2022). In the human genome, we previously showed that MER41 elements have been co-opted as enhancer elements to regulate multiple immune genes in human cells, including the AIM2 inflammasome genes (Chuong et al., 2016). Elements belonging to other transposon families, including LTR12, MER44, and THE1C, have also been co-opted to regulate inducible expression of immune genes (Bogdan et al., 2020; Donnard et al., 2018; Srinivasachar Badarinarayan et al., 2020). Notably, the majority of these families are primate-specific, supporting the co-option of TEs as a driver of primate-specific divergence of immune regulatory networks.

A key open question is whether the co-option of TEs as immune regulatory elements is evolutionarily widespread as a mechanism driving divergence of innate immune responses. Most research in this area has focused on human cells and primate-specific TE families, but different mammalian species harbor highly distinct and lineage-specific repertoires of TEs in their genomes. Due to the independent origin of most of these TEs in different species, it remains unclear whether the co-option of TEs is a rare or common mechanism contributing to the evolution of immune gene regulatory networks.

Here, we focused on the role of TEs in regulating murine innate immune responses. Mice are a commonly used model for human diseases but their immune system is appreciated to have significant differences. Transcriptomic studies have revealed that mouse and human immune transcriptomes show substantial divergence (Shaw et al., 2017; Shay et al., 2013), consistent with functional differences in inflammatory responses (Seok et al., 2013). The rodent and primate lineages diverged roughly 90 million years ago (Hedges et al., 2006; Mestas and Hughes, 2004) and 32% of the mouse genome consists of rodent-specific repeats (Waterston et al., 2002). Therefore, we sought to define the potential role of TEs in shaping lineage-specific features of the murine innate immune response.

In our study, we re-analyzed transcriptomic and epigenomic datasets profiling the type II interferon (IFN) response in primary mouse macrophage cells. We screened for TEs showing epigenetic signatures of inducible regulatory activity, and identified a rodent-specific B2 subfamily as a substantial source of IFN-inducible regulatory elements in the mouse genome. As a case example, we used CRISPR to characterize a B2-derived IFN-inducible enhancer that regulates mouse Dicer1. These findings uncover a novel cis-regulatory role for the SINE B2 element in shaping the evolution of mouse-specific IFN responses.

Results

Species-specific TEs shape the epigenomic response to type II IFN in mouse

To examine how TEs contribute to mouse type II IFN signaling regulation, we re-analyzed two independent transcriptomic and epigenomic datasets of primary bone-marrow-derived macrophages (BMDMs) that were stimulated with recombinant interferon gamma (IFNG) or untreated for 2 or 4 hr (Platanitis et al., 2019). These datasets included matched RNA-seq and ChIP-seq for STAT1 and H3K27ac. The STAT1 transcription factor mediates the type II IFN response (Platanias, 2005) by binding to enhancers and promoters containing the Gamma-IFN activation site (GAS) motif, and the H3K27ac modification is strongly associated with active enhancers (Creyghton et al., 2010). Using these datasets, we mapped both IFNG-inducible enhancers and IFNG-stimulated genes (ISGs).

Our analysis of the RNA-seq data identified a total of 1,896 ISGs (FDR adjusted p-value <0.05, log2 fold change (log2FC)>1), which enriched for canonical genes associated with the IFNG response (GO:0034341, adjusted p-value = 9.394 × 10–35; Supplementary file 1). We predicted IFNG-inducible enhancers based on occupancy by the enhancer-associated histone mark H3K27ac and the transcription factor STAT1, which mediates type II IFN signaling. We identified 22,921 regions bound by STAT1 in IFNG-induced cells, 18,337 (80.0%) of which also resided within H3K27ac-enriched regions, indicating they are putative enhancers. Specificity of pulldown was confirmed by enrichment of canonical STAT1 binding motifs including the Gamma-IFN activation site (GAS; E-value=1.11 × 10–746) and IFN-stimulated response element (ISRE; E-value=1.01 × 10–442) motifs within the STAT1 ChIP-seq peaks (Supplementary file 2).

Using this set of STAT1 binding sites, we next asked what fraction of binding sites were derived from mouse TEs. Using the summits of the STAT1 ChIP-seq peaks, we found that 26.6% resided within TEs, 71.1% of which contain significant matches (p-value <1 × 10–4) to either ISRE or GAS motifs (Supplementary file 3). These TEs likely represent direct binding sites of STAT1 with potential regulatory activity. We next asked whether any TE families were overrepresented within the set of predicted IFNG-inducible binding sites, using GIGGLE colocalization analysis (Layer et al., 2018). We identified three subfamilies enriched for STAT1 binding sites, including the rodent-specific B2_Mm2 subfamily (p-value = 7.18 × 10–201) as well as the RLTR30B_MM (p-value = 9.61 × 10–77) and RLTR30E_MM (p-value = 7.32 × 10–31) endogenous retrovirus subfamilies (Figure 1A, Supplementary file 4). This indicates that the expansion of rodent-specific TE families has shaped the innate immune regulatory landscape in mouse.

Figure 1 with 2 supplements see all
Download asset Open asset
Identification of transposon-derived enhancers in innate immunity.

(A) Bubble plot showing family-level enrichment of transposons within ChIP-seq peak regions. TE families enriched for STAT1 and H3K27ac ChIP-seq peaks are sorted by descending median Kimura distance. GIGGLE enrichment score is a composite of the product of both -log10(p-value) and log2(odds ratio). (B) Frequency histogram of absolute distances from STAT1-bound B2_Mm2 (blue, n=2,122), randomly shuffled B2_Mm2 (black, n=2,122), and randomly subset unbound B2_Mm2 (green, n=2,122), B2_Mm1a (red, n=2,122), and B2_Mm1t (yellow, n=2,122) elements to the nearest ISG (n=750), IRG (n=750), or nonresponsive gene (n=750). Data shown for (Piccolo et al., 2017) comparing expression in BMDMs stimulated with IFNG for 4 hr relative to untreated. (C) DESeq2 normalized counts showing immune-stimulated expression of unbound B2_Mm2 (top, n=78,419) and STAT1-bound B2_Mm2 (bottom, n=2122) elements in murine BMDMs. Data shown for untreated (n=2) BMDMs and BMDMs stimulated with IFNG for 2 hr (n=3) or 4 hr (n=3). Treatments are indicated by color. ***DESeq2 FDR adjusted p-value <0.0001. Error bars designate SEM. Data shown for (Piccolo et al., 2017). ISG: Interferon-stimulated gene; IRG: Interferon-repressed gene. BMDMs: Bone marrow derived macrophages. SEM: Standard error of mean.

We previously identified enrichment of RLTR30 elements within STAT1-binding sites in IFNG and IFNB-stimulated mouse macrophages based on analysis of a different ChIP-seq dataset (Chuong et al., 2016; Ng et al., 2011). However, our previous analysis did not capture enrichment of B2_Mm2, likely because the dataset was generated using 36 bp short reads. In contrast, the more recent datasets analyzed here used 50 bp reads (Platanitis et al., 2019), which improves mappability to individual copies of evolutionarily young TE families such as B2_Mm2 (Sundaram et al., 2014).

B2_Mm2 elements contain STAT1 binding sites and show inducible enhancer activity

B2_Mm2 is a murine-specific subfamily of the B2 short interspersed nuclear element (SINE) family, which is highly abundant in the mouse genome. B2 SINE elements are divided into three subfamilies, including B2_Mm2 (80,541 copies), B2_Mm1a (16,321 copies), and B2_Mm1t (35,812 copies). B2 SINE elements have been characterized to show a wide range of regulatory activities in mice, including acting as promoters (Ferrigno et al., 2001), insulator elements bound by CTCF (Ichiyanagi et al., 2021; Lunyak et al., 2007; Schmidt et al., 2012), or regulatory non-coding RNAs (Hernandez et al., 2020; Karijolich et al., 2017; Karijolich et al., 2015; Schaller et al., 2020; Yakovchuk et al., 2009). As the potential for B2_Mm2 SINEs to act as inducible enhancers has not yet been investigated, we decided to further investigate B2 SINEs in this context.

The B2_Mm2 subfamily showed strong evidence of enrichment within regions bound by STAT1, providing 2,122 total binding sites (odds ratio = 4.85). These B2_Mm2 elements show significantly higher localization near ISGs (p-value = 5.03 × 10–52, odds ratio = 9.13) than interferon-repressed genes (IRGs) or nonresponsive genes, compared to unbound B2 elements or random genomic regions (Figure 1B, Figure 1—figure supplement 1A–B). We did not observe consistent enrichment of the B2_Mm1a and B2_Mm1t subfamilies over STAT1-bound regions (Supplementary file 4). Additionally, STAT1-bound B2_Mm2 elements are transcriptionally upregulated at the family level in response to IFNG stimulation (Figure 1C, Figure 1—figure supplement 2, Supplementary file 5). Although unbound B2_Mm2 elements are also transcriptionally active, we did not observe a significant increase in expression in response to IFNG stimulation. Taken together, these data indicate that thousands of B2_Mm2 elements show epigenetic and transcriptional evidence of IFNG-inducible regulatory activity in primary murine bone marrow derived macrophages.

We investigated the sequence features of each B2 SINE subfamily to determine the basis of IFNG-inducible activity. Given that B2 SINE elements have previously been associated with CTCF binding due to the presence of a CTCF motif harbored by most copies (Schmidt et al., 2012), we subdivided elements from each family based on occupancy by STAT1, CTCF, both factors, or neither factor based on ChIP-seq. Across each of these subsets, we looked for the presence of GAS or CTCF motifs (Figure 2A, Figure 2—figure supplement 1). As expected, all B2 subfamilies showed extensive ChIP-seq binding evidence of CTCF and the RAD21 cohesin subunit, coinciding with a CTCF motif (Figure 2—figure supplement 1). In contrast, only a subset of elements from the B2_Mm2 subfamily showed inducible binding of STAT1 (Figure 2A). Consistent with ChIP-seq evidence, STAT1-bound B2_Mm2 elements contain both GAS and CTCF motifs, while B2_Mm1a/t elements only harbor CTCF motifs (Figure 2B, Figure 2—figure supplement 2). Within B2_Mm2, elements that are bound by STAT1 are significantly enriched for GAS motifs when compared against unbound elements (E-value 2.08×10–71, Supplementary file 6). In addition, STAT1-bound B2_Mm2 elements contain stronger sequence matches to GAS motifs compared to unbound B2_Mm2 elements and B2_Mm1a/t elements (Figure 2A, Figure 2—figure supplement 3). While B1_Mm1a elements show a partial match to the GAS motif, they show no evidence of STAT1 binding (Figure 2B, Figure 2—figure supplement 2). Therefore, elements of the B2_Mm2 subfamily are uniquely characterized by containing strong matches to GAS motifs that are associated with STAT1 binding activity.

Figure 2 with 3 supplements see all
Download asset Open asset
Epigenomic profiling of B2_Mm2.

(A) Heatmaps showing CPM-normalized ChIP-seq signal and motif signal centered over B2_Mm2 elements bound only by STAT1 (n=1,203); B2_Mm2 bound by both STAT1 and CTCF (n=893); B2_Mm2 bound only by CTCF (n=4,143); and a random subset of unbound B2_Mm2 (n=1,000). Regions are sorted by descending mean CPM signal. Signal intensity is indicated below. CTCF track derived from Gualdrini et al., 2022. RAD21 track derived from Cuartero et al., 2018. (B) Schematic of GAS (blue) and CTCF (green) motifs present within extant B2_Mm2 (left, n=80,541), B2_Mm1a (middle, n=16,321), and B2_Mm1t (right, n=35,812) sequences. Heatmap intensity corresponds to motif matches based on the log likelihood ratio. Heatmaps are sorted by descending mean signal. Position weight matrices were obtained from JASPAR (Fornes et al., 2020). CPM: Counts per million. GAS: Gamma-IFN activated sequence.

Notably, we found that STAT1-bound B2_Mm2 elements are the only B2 elements that show an inducible H3K27ac signal associated with enhancer activity. In contrast, B2_Mm2 elements bound only by CTCF or unbound elements show minimal H3K27ac signal (Figure 2A). B2_Mm1a/t elements also show minimal STAT1 signal regardless of CTCF binding (Figure 2—figure supplement 2). This suggests that the binding of STAT1 to B2_Mm2 elements causes activation of enhancer activity including acetylation of H3K27. Thus, B2_Mm2 elements represent a distinct subclass of B2 SINE elements that exhibit IFNG-inducible enhancer activity.

An intronic B2_Mm2 element functions as an inducible enhancer of Dicer1

Having established that B2_Mm2 elements are an abundant source of IFNG-inducible STAT1 binding sites in the mouse genome, we asked whether any of these elements have been co-opted to regulate expression of individual ISGs. We first assigned predicted enhancers to their predicted targets using the Activity by Contact (ABC) model (Fulco et al., 2019), which incorporates both epigenomic signal and Hi-C 3D interaction data to predict enhancer-gene targets. As input into the ABC model, we used publicly available ATAC-seq and Hi-C data from murine BMDMs stimulated with IFNG for 2 hours (Platanitis et al., 2022) and H3K27ac ChIP-seq and RNA-seq data (Piccolo et al., 2017). Focusing on a permissive set of 2,720 ISGs (FDR adjusted p-value <0.05, log2FC >0), we identified 530 B2_Mm2 elements predicted to interact with 457 mouse ISGs (16.8% of the set of 2720 in this analysis; Supplementary file 7). Compared with a set of human ISGs from human CD14+ monocytes (Qiao et al., 2016), 393 of these 457 (86%) genes were only ISGs in mouse, and the remaining 64 (14%) were ISGs in both mouse and human.

The 393 mouse-specific ISGs predicted to be regulated by B2_Mm2 elements were significantly enriched for multiple immune related functions (GO:0002376, adjusted p-value = 6.023 × 10–9; Supplementary file 7). We identified multiple examples of predicted B2_Mm2 target genes with established immune functions that showed mouse-specific IFNG-inducible expression (Figure 3—figure supplement 1, Figure 3—figure supplement 2), including dicer 1 ribonuclease III (Dicer1) (Poirier et al., 2021; Figure 3A), SET domain containing 6, protein lysine methyltransferase (SETD6) (Levy et al., 2011), DOT1-like histone lysine methyltransferase (Kealy et al., 2020), fumarate hydratase 1 (Fh1) (Zecchini et al., 2023), heat shock protein family A (Hsp70) member 1B (Hspa1b) (Jolesch et al., 2012), and NFKB inhibitor delta (Nfkbid) (Souza et al., 2021).

Figure 3 with 5 supplements see all
Download asset Open asset
Identification of a putative B2_Mm2-derived enhancer for Dicer1.

(A) DESeq2 normalized counts showing immune-stimulated expression of Dicer1 in primary murine BMDMs. Data shown for untreated (n=3) BMDMs and BMDMs stimulated with IFNG for 2 hr (n=3) or 4 hr (n=3). ***DESeq2 FDR adjusted p-value <0.001, *DESeq2 FDR adjusted p-value <0.05. Error bars designate SEM. (B) Genome browser screenshot (http://genome.ucsc.edu) of the Dicer1 locus (Chr12:104,681,897–104,757,421) showing CPM-normalized ChIP-seq tracks for primary murine BMDMs and immortalized macrophage line J774A.1. B2_Mm2.Dicer1 (Chr12:104,742,467–104,742,646) is highlighted in gray. Values on the right of each track correspond to signal maxima. Bottom inset shows B2_Mm2.Dicer1 with accompanying STAT1 (IFNG 2 h Piccolo et al.) ChIP-seq signal and conservation tracks for rat, rabbit, human, cow, and dog. CTCF track derived from Gualdrini et al., 2022. RAD21 track derived from Cuartero et al., 2018. (C) RT-qPCR of wild type untreated (gray, n=3) J774A.1 cells and J774A.1 cells stimulated with IFNG (blue, n=3), IFNA (green, n=3), IL-2 (orange, n=3), IL-4 (light blue, n=3), or IL-15 (purple, n=3) for 4 hr. Treatments are indicated by color. *p-value <0.05, Student’s paired two-tailed t-test. BMDM: Bone-marrow-derived macrophage. SEM: Standard error of mean. CPM: Counts per million.

From this set, we decided to focus on a specific B2_Mm2 element located on Chromosome 12 within the first intron of Dicer1, which is an endonuclease responsible for recognizing and cleaving foreign and double stranded RNA that has been linked to innate immunity (Chiappinelli et al., 2012; Gurung et al., 2021; MacKay et al., 2014; Poirier et al., 2021). While the human ortholog DICER1 does not show IFNG-inducible expression in human primary macrophages (Qiao et al., 2016; Figure 3—figure supplement 3, Supplementary file 1), mouse Dicer1 shows a significant 50% upregulation in response to IFNG in primary mouse BMDMs (Figure 3A). This indicates that Dicer1 is a mouse-specific ISG and likely acquired IFNG-inducible expression in the mouse lineage, potentially due to the co-option of the B2_Mm2 element as a species-specific IFNG-inducible enhancer. The intronic B2_Mm2 element (B2_Mm2.Dicer1) shows biochemical hallmarks of enhancer activity including inducible STAT1 and H3K27ac signal as well as constitutive binding by CTCF and RAD21 (Figure 3B). The element provides the only prominent nearby STAT1 binding site and is not present in rat or other mammals (Figure 3B). Therefore, we hypothesized that the B2_Mm2.Dicer1 element was co-opted as an IFNG-inducible enhancer of mouse Dicer1.

To experimentally test the potential enhancer activity of B2_Mm2.Dicer1, we used the mouse J774A.1 macrophage-like cell line, a commonly used model of murine immunity (Lam et al., 2009; Ralph and Nakoinz, 1975). We first confirmed using RT-qPCR that Dicer1 shows 30–40% upregulation after 4 hr of IFNG treatment (Figure 3C). Given the STAT1 binding site and motif present in B2_Mm2.Dicer1, we also tested other cytokines that act through STAT-family transcription factors. We found that 4 hr treatment of J774A.1 cells with IFNA, IL6, and IL4 all induced Dicer1 expression to similar levels (30–40%), consistent with inducible regulation of Dicer1 by JAK-STAT signaling, potentially through the STAT binding site present in B2_Mm2.Dicer1.

We next used CRISPR to generate clonal J774A.1 lines harboring homozygous deletions of the B2_Mm2.Dicer1 element. We delivered guide RNAs targeting the flanking boundaries of B2_Mm2.Dicer1 along with recombinant Cas9 by electroporation (Figure 3—figure supplement 4A–C), isolated clones by limiting dilution, and screened clonal lines for homozygous deletions by PCR (Figure 3—figure supplement 4D–F). We isolated two clonal cell lines with a homozygous knockout of B2_Mm2.Dicer1, along with multiple wild-type (WT) J774A.1 clonal lines that were not electroporated to control for potential effects of clonal expansion (Figure 3—figure supplement 4F). We used RT-qPCR to compare Dicer1 expression levels and inducibility by IFNG in knockout and WT clones. WT clones showed consistent inducible expression, while both knockout clonal lines showed a complete lack of inducible expression (Figure 4A). These experiments demonstrate that B2_Mm2.Dicer1 acts as an IFNG-inducible enhancer of Dicer1 in J774A.1 cells.

Figure 4
Download asset Open asset
B2_Mm2 in the genomic landscape.

(A) RT-qPCR of Dicer1 expression in untreated and IFNG-treated cell lines across 3 clonal WT lines J774A.1 cells and 2 B2_Mm2.Dicer1 knockout (KO) lines, with three replicate treatments per cell line. Dicer1 expression was normalized relative to CTCF. Treatments are indicated by color. *p-value <0.05, **p-value <0.01, ***p-value <0.001, Student’s paired two-tailed t-test. (B) DESeq2 normalized counts of Dicer1 expression in each clonal WT and B2_Mm2.Dicer1 KO J774A.1 cell line. Treatments are indicated by color. ***DESeq2 FDR adjusted p-value <0.001. (C) Distance plot visualizing changes in gene expression in wild type J774A.1 cells in response to IFNG over a 5 Mb window centered on B2_Mm2.Dicer1. Significantly downregulated (log2FC <0, FDR adjusted p-value <0.05) genes are shown in red while significantly upregulated (log2FC >0, FDR adjusted p-value <0.05) genes are shown in blue. Dicer1 is labeled, as well as significantly IFNG-regulated genes within 1 Mb. (D) Same as in (C) but visualizing changes in gene expression in KO J774A.1 cells in response to IFNG.

B2_Mm2.Dicer1 impact on the genomic regulatory landscape

We used RNA-seq to study the genome-wide effects of the B2_Mm2.Dicer1 element in both knockout clones and three control wild-type clones which were also isolated by limiting dilution. Consistent with the RT-qPCR results, we found that Dicer1 showed significant IFNG-inducible upregulation in all WT clones but that this induction was completely ablated in B2_Mm2.Dicer1 KO clones (Figure 4B). Notably, the RNA-seq normalized count data revealed that expression of Dicer1 were also significantly reduced in untreated KO cells (log2FC = –0.43, FDR-adjusted p-value = 9.246 × 10–4) (Figures 4B and 5A, Figure 3—figure supplement 5B–C). Focusing on the IFNG-treated condition, Dicer1 was significantly downregulated in KO cells compared to WT cells (log2FC = –0.80, FDR-adjusted p-value = 3.393 × 10–15). The deletion of the element did not affect inducibility of other ISGs in the 5 Mb locus including the nearby highly induced SerpinA3 ISGs locus (Figure 4C–D). This indicates that the B2_Mm2.Dicer1 element specifically regulates both basal expression levels of Dicer1 and inducible expression by IFNG. While the enhancer deletion on Dicer1 expression levels had a modest downregulating effect, this effect was specific and consistent across individually edited clones, particularly under IFNG-stimulated conditions (Figure 4C–D, Figure 3—figure supplement 5D–E).

Figure 5
Download asset Open asset
B2_Mm2 impacts local chromatin profile.

(A) Genome browser screenshot of the Dicer1 locus visualizing CPM-normalized expression in WT and B2_Mm2.Dicer1 KO J774A.1 cells. Values on the right of each track correspond to signal maxima. B2_Mm2.Dicer1 is represented as a black box (not drawn to scale). (B) Genome browser screenshot of the Dicer1 locus showing CUT&TAG data from bulk WT and B2_Mm2.Dicer1 KO J774A.1 cells. CPM: Counts per million.

Genome-wide, there were 101 genes that showed greater significance than Dicer1 when testing for differential expression between IFNG-treated KO and WT cells (out of 3,567 genes with FDR-adjusted p-value <0.05; Supplementary file 8). Out of these 101 genes, 24 showed higher variability than Dicer1 (based on DESeq2 log2FC standard error) between individual clones (Figure 3—figure supplement 5B–E), consistent with intrinsic clonal transcriptional variation revealed by the limiting dilution and/or CRISPR editing process (Nahmad et al., 2022; Westermann et al., 2022). Thirty-five of these genes showed upregulation in the KO cells, suggesting that they could be silencing targets of Dicer1 that become upregulated upon Dicer1 downregulation. However, given the relatively modest effect of Dicer1 especially in the untreated condition, further experiments would be necessary to establish these genes as targets of Dicer1.

We confirmed the absence of intronic enhancer activity in B2_Mm2.Dicer1 knockout cells. We used CUT&TAG (Kaya-Okur et al., 2020) to profile H3K27ac, phosphorylated RNA polymerase II subunit A (POLR2A), and STAT1. In IFNG-stimulated WT cells, the B2_Mm2.Dicer1 element shows prominent H3K27ac, STAT1, and POLR2A signal. However, these signals are completely lost in the knockout clone (Figure 5B). Collectively, these experiments confirm that B2_Mm2.Dicer1 has been co-opted to function as an IFNG-inducible enhancer that regulates Dicer1.

Given the B2_Mm2.Dicer1 element is bound by POLR2A in WT cells, we asked whether the B2_Mm2.Dicer1 element may alter transcription by affecting usage of different splice sites or polyadenylation sites, which would be consistent with pause site activity (Jonkers and Lis, 2015). We examined transcript isoform-level expression changes in KO cells in both untreated and treated conditions and found multiple transcripts that showed the same trend as the gene-level analysis, where most expressed transcripts are downregulated in both basal and IFNG-treated conditions and show lack of inducibility in KO cells (Supplementary file 9). These findings are consistent with the element acting primarily as an IFNG-inducible enhancer without any major effect on alternative splicing. However, further experiments such as CDK9 inhibition and profiling of nascent transcription in stimulated conditions (Gressel et al., 2017; Laitem et al., 2015) are necessary to establish whether the element affects transcriptional elongation.

Considering that the B2_Mm2 enhancer is specific to rodents, we examined the regulatory landscape of the human DICER1 locus. Our analysis of RNA-seq data from human primary monocytes treated with IFNG for 24 hr (Qiao et al., 2016) indicated that human DICER1 expression is not induced by IFNG (Supplementary file 1, Figure 3—figure supplement 3). However, ChIP-seq data from IFN-treated monocytes from the same group (Qiao et al., 2013) showed multiple inducible STAT1 binding sites within the human DICER1 locus, including one originating from a primate-specific TE (LTR27) (Figure 3—figure supplement 3). Although these binding sites do not correlate with inducible DICER1 expression in the matched RNA-seq dataset, they suggest human DICER1 may be inducible under different conditions. An analysis of an independent dataset generated from another donor (McCann et al., 2022) supported the inducible expression of DICER1 (log2FC = 0.91 and FDR adjusted P-value = 3.0 × 10–4; Supplementary file 1). Thus, while the evidence for inducible human DICER1 expression is inconsistent, our analyses indicate that human DICER1 has independently evolved primate-specific binding STAT1 binding sites, which may also confer inducible regulation.

Discussion

B2 SINE elements are abundant in the mouse genome and they have been widely studied due to their substantial influence on genome regulation and evolution. B2 elements have contributed non-coding RNAs inducible by stress or infection (Allen et al., 2004; Karijolich et al., 2017; Karijolich et al., 2015; Li et al., 1999; Schaller et al., 2020; Walters et al., 2009; Wick et al., 2003; Williams et al., 2004), splicing signals (Kress et al., 1984), promoter elements (Ferrigno et al., 2001), and CTCF-bound insulator elements (Ichiyanagi et al., 2021; Lunyak et al., 2007; Schmidt et al., 2012). Our study reveals a new subclass of B2 elements that have IFNG-inducible enhancer activity. These elements, which belong to the B2_Mm2 subfamily, contain strong binding sites for both STAT1 and CTCF, are marked by H3K27ac, and have the potential to exert inducible enhancer activity on nearby genes.

Given the abundance of B2 elements and their potential to cause pathological regulatory rewiring, many B2 elements are targeted for SETDB1/H3K9me3-mediated epigenetic repression which inhibits their regulatory potential (Gualdrini et al., 2022). Therefore, the functional impact of B2 elements on the mouse epigenome remains unclear. By using CRISPR to generate knockout cells of a B2_Mm2 element, we demonstrated that B2 elements can be co-opted to act as inducible enhancer elements in the context of IFNG stimulation. While our experiments were conducted in the J774.A1 immortalized cell line, we confirmed that thousands of B2_Mm2 elements including B2_Mm2.Dicer1 show strong transcriptional and epigenetic signatures of inducible enhancer activity in multiple primary macrophage epigenomic datasets.

We identified hundreds of B2_Mm2-derived enhancers that are predicted to regulate genes displaying IFNG-inducible expression in mouse cells but not human cells, which supports their role in facilitating lineage-specific evolution of the IFNG-inducible regulatory network. However, we also identified a subset of target genes that show inducible expression in both species, suggesting independent evolution or turnover of regulatory elements that could serve similar regulatory functions as the B2_Mm2 enhancers. For instance, we identified an intronic STAT1 binding site derived from a primate-specific TE in the human DICER1 locus. While we did not uncover consistent evidence supporting IFNG-inducible regulation of human DICER1, these observations align with the concept of convergent regulatory evolution, in which similar expression patterns are perpetuated by the co-option of lineage-specific TEs (Choudhary et al., 2020; Sundaram et al., 2014). Therefore, as B2_Mm2 elements shaped the evolution of rodent immune regulatory networks, individual co-option events may have mediated either divergence or preservation of gene expression patterns.

Our identification of a B2_Mm2 element as an intronic enhancer of Dicer1 potentially uncovers a novel regulatory feedback loop that controls Dicer1 function related to TE silencing. Previous studies have demonstrated that Dicer1 cleaves double-stranded RNAs including those derived from B2 SINE transcripts (Fan et al., 2021). In Dicer1 knockout embryonic stem cells, TE-derived transcripts are upregulated (Bodak et al., 2017), and upregulation of B2-derived double-stranded RNAs causes activation of the IFN response (Gurung et al., 2021). Therefore, Dicer1 is important for defense against aberrant TE upregulation. We speculate that co-option of the B2_Mm2 element as an enhancer of Dicer1 facilitates upregulation of Dicer1 in response to conditions that drive TE upregulation, such as infection or stress.

In human, dysregulation of DICER1 is associated with a wide range of pathologies ranging from DICER1 syndrome, cancer, neurological diseases such as Parkinson’s disease, and autoimmune disorders such as rheumatoid arthritis (Theotoki et al., 2020). While DICER1 has a highly conserved function as an endonuclease involved in RNA interference, our work highlights that orthologs of DICER1 have undergone lineage- or species-specific regulatory evolution that may drive underappreciated differences in function across species. While we found that both human and mouse orthologs of DICER1 have STAT1 binding sites, the underlying cis-regulatory architecture is not conserved and likely results in different expression patterns. This could have significant implications when developing and testing RNA-based therapeutics in mouse genetic models, which may elicit distinct Dicer1-mediated responses due to species-specific enhancers such as B2_Mm2.Dicer1.

In conclusion, our work adds to a growing body of evidence highlighting the co-option of lineage-specific TEs for the regulation of ISGs. Previous genomic and experimental studies in human (Chuong et al., 2016) and cow cells (Kelly et al., 2022) have revealed independent co-option of TEs as IFNG-inducible enhancer elements. Interestingly, while endogenous retroviruses have been described as a prevalent source of TE-derived inducible enhancers in human, their contribution is relatively minor in cows and mice. Instead, we found that the B2_Mm2 SINE subfamily is the predominant source of TE-derived inducible enhancers in mouse, and the Bov-A2 SINE subfamily is the predominant source in cow (Kelly et al., 2022). These findings indicate that the acquisition of STAT1-associated GAS motifs and enhancer activity is not limited to a specific type of TE. In human cells, Alu SINE sequences are transcriptionally activated upon infection (Jang and Latchman, 1989; Panning and Smiley, 1993), although their epigenetic impact remains unexplored. It remains unclear whether the infection-inducible regulatory activity promotes SINE replication, and it is possible that the emergence of these motifs is coincidental and does not affect TE fitness. Nevertheless, our work supports the idea that TEs have been repeatedly co-opted as IFNG-inducible enhancers throughout mammalian evolution, contributing to the rewiring of immune regulatory networks.

Materials and methods

Sequences

Request a detailed protocol

A list of all primer sequences and gRNA sequences can be found in Supplementary file 10.

RNA-seq re-analysis

Request a detailed protocol

RNA-seq data (single-end reads) from primary murine BMDMs stimulated with 100 ng/mL IFNG for 2 or 4 hr (Piccolo et al., 2017) or 10 ng/mL IFNG for 2 hr (Platanitis et al., 2019) were downloaded from SRA using fasterq-dump v2.10.5 (NCBI, 2022 ). Adapters and low-quality reads were trimmed using BBDuk v38.05 (Bushnell, 2018) using options ‘ktrim = r k=34 mink = 11 hdist = 1 qtrim = r trimq = 10 tpe tbo’. Library quality was assessed using FastQC v0.11.8 (Andrews, 2018) and MultiQC v1.7 (Ewels et al., 2016), and trimmed reads were aligned to the mm10 assembly using HISAT2 v2.1.0 (Kim et al., 2019) with option ‘--no-softclip’. Only uniquely aligned fragments (MAPQ ≥ 10) were retained using samtools v1.10 (Li et al., 2009). Aligned fragments were assigned to the complete mm10 Gencode vM18 (Frankish et al., 2021) annotation in an unstranded manner using featureCounts v1.6.2 (Liao et al., 2014) with options ‘-p -O -s 0 t exon -g gene_id’, and differentially expressed genes between IFNG-stimulated and unstimulated cells were called using DESeq2 v1.26.0 (Love et al., 2014). For most analyses, ISGs and IRGs were defined as genes with a false discovery rate (FDR) adjusted p-value of at least 0.05 and log2FC greater than 0 and less than zero, respectively. Nonresponsive genes were defined using the following cutoffs: baseMean greater than 100; FDR adjusted p-value greater than 0.90; and absolute log2FC less than 0.10. Interferon stimulation was confirmed by gene ontology analysis using gProfiler (last updated 05/18/2022) with FDR adjusted p-value <0.05 (Raudvere et al., 2019). We additionally aligned RNA-seq data from human CD14+ monocytes stimulated with 100 U/mL IFNG for 24 hr to the hg38 assembly and identified differentially expressed genes using Gencode v38 (Frankish et al., 2021) with the methods described above.

ChIP-seq re-analysis

Request a detailed protocol

ChIP-seq data (single-end reads) from primary murine BMDMs (Cuartero et al., 2018; Gualdrini et al., 2022; Piccolo et al., 2017; Platanitis et al., 2019) were downloaded from SRA using fasterq-dump v2.10.5 (NCBI, 2022). Adapters and low-quality reads were trimmed using BBDuk v38.05 (Bushnell, 2018) using options ‘ktrim = r k=34 mink = 11 hdist = 1 qtrim = r trimq = 10 tpe tbo’. Library quality was assessed using FastQC v0.11.8 (Andrews, 2018) and MultiQC v1.7 (Ewels et al., 2016), and trimmed reads were aligned to the mm10 assembly using BWA-MEM v0.7.15 (Li, 2013). Low quality and unmapped reads were filtered using samtools v1.10 (Li et al., 2009), and duplicates were removed with Picard MarkDuplicates v2.6.0 (Broad Institute, 2016). Peak calling was performed with MACS2 v2.1.1 (Liu, 2014) using options ‘--gsize mm –pvalue 0.01 –bdg –SPMR –call-summits’. bigWigs corresponding to read pileup per million reads for visualization on the UCSC Genome Browser (Kent et al., 2002). Where possible, only peaks overlapping more than one replicate were retained for further analysis. To confirm whether STAT1 peaks were enriched for their associated binding motifs, we ran XSTREME v5.4.1 (Grant and Bailey, 2021) using options ‘--minw 6 --maxw 20 –streme-nmotifs 20 –align center’ querying against the JASPAR CORE 2018 vertebrates database (Fornes et al., 2020).

Transposable element analysis

Request a detailed protocol

To identify TE families enriched for STAT1 and H3K27ac peaks, we used GIGGLE v0.6.3 (Layer et al., 2018) to create a database of all TE families annotated in the mm10 genome according to Dfam v2.0 (Storer et al., 2021) annotation. STAT1 and H3K27ac ChIP-seq peaks were then queried against each TE family in the database. GIGGLE applies the Fisher’s exact test to assess family-level enrichment, attributing an odds ratio, Fisher’s two tailed p-value, and a GIGGLE combo score combining the two values. We only retained TE families that met the following criteria: (1) number of total annotated elements >100; (2) number of elements overlapping a ChIP-seq peak >30; (3) odds ratio >3; and (4) a GIGGLE combo score >100. Results were visualized as a bubble plot where the filtered TE families were sorted by ascending Kimura divergence according to RepeatMasker (Smit et al., 2019) output. Reported odds ratios and p-values are derived from Fisher’s exact test. For further analysis, we intersected STAT1 peaks with the full TE annotation or B2 elements specifically using BEDTools v2.28.0 (Quinlan and Hall, 2010). For the heatmap visualizations using deepTools v3.0.1 (Ramírez et al., 2016), signal from counts per million- (CPM) normalized bigWigs was plotted over a subset of B2_Mm2 elements that are bound only by STAT1, CTCF, both, or neither. We additionally visualized ChIP-seq signal over all B2_Mm2, B2_Mm1a, and B2_Mm1t elements by descending average signal, excluding elements with zero overlapping signal.

To assess whether STAT1-bound B2_Mm2 elements are enriched near ISGs, we sorted all ISGs and IRGs by descending and ascending log2FC, respectively, and retained the top 750 genes. We additionally randomly subset for 750 nonresponsive genes. The absolute distance to the nearest ISG, IRG, or nonresponsive gene was determined for all STAT1-bound B2_Mm2 elements using BEDTools v2.28.0 (Quinlan and Hall, 2010). Randomly shuffled STAT1-bound B2_Mm2 as well as randomly subset, unbound B2_Mm2, B2_Mm1a, and B2_Mm1t were included as controls. Statistical significance was determined for the first 20 kb bin using Fisher’s exact test with BEDTools v2.28.0 (Quinlan and Hall, 2010).

To identify TE families that are differentially expressed in response to IFNG in primary murine BMDMs, we realigned the RNA-seq data to the mm10 reference genome using HISAT2 v2.1.0 (Kim et al., 2019) with options ‘-k100 –no-softclip’. Aligned reads were assigned to TE families using TEtranscripts v2.1.4 (Jin et al., 2015) with options ‘--sortByPos –mode multi –iteration 100 –stranded no’. TEtranscripts allows for quantification of TE expression at the family level and does not discriminate between individual elements within a family. To differentiate unbound and STAT1-bound B2_Mm2 elements, we generated a custom TE annotation file compatible with TEtranscripts that includes all TEs annotated in Dfam v2.0 (Storer et al., 2021) but annotates STAT1-bound B2_Mm2 elements as an independent subfamily. Differentially expressed TE families between IFNG-stimulated and unstimulated cells were identified using DESeq2 v1.26.0 (Love et al., 2014). TE families with an FDR less than 0.05 and an absolute log2FC greater than 0.50 were considered as differentially expressed. These relaxed thresholds were used to better enable the identification of differentially expressed TE families where only a subset of elements are inducibly expressed and the majority are only lowly expressed.

We identified putative STAT1 and CTCF binding sites genome-wide using FIMO v5.0.3 (Grant et al., 2011) with a p-value cutoff of 1×10–4 (heatmaps) or 1 (B2 box-and-whisker). For all motif analyses, binding motif position-weight matrices for STAT1 and CTCF were obtained from the JASPAR CORE 2018 vertebrate database (Fornes et al., 2020). To visualize motif presence over all B2 elements, repeat 5’ start coordinates were recalculated based on their alignment to the consensus according to RepeatMasker annotations. Motif presence was visualized as a heatmap using deepTools v3.0.1 (Ramírez et al., 2016), and elements were sorted by descending average signal. We additionally aligned the consensus sequences for B2_Mm2, B2_Mm1a, and B2_Mm1t from Repbase v24.02 and the sequence for B2_Mm2.Dicer1 using MUSCLE v3.8.1551 (Edgar, 2004). Predicted STAT1 and CTCF motifs were identified using FIMO v.5.0.3 (Grant et al., 2011), and base changes relative to the canonical binding motifs were highlighted according to the weight of each individual base in the position-weight matrices. Finally, we filtered for STAT1-bound B2_Mm2 elements that were non-overlapping, non-nested, and unique and ran AME v5.4.1 (McLeay and Bailey, 2010) using a subset of unbound B2_Mm2 elements as the background control with options ‘--kmer 2 –method fisher –scoring avg’.

ATAC-seq re-analysis

Request a detailed protocol

Paired-end ATAC-seq data from primary murine BMDMs stimulated with 10 ng/mL IFNG for 2 hr (Platanitis et al., 2022) were downloaded from SRA using fasterq-dump v2.10.5 (NCBI, 2022). Adapters and low-quality reads were trimmed using BBDuk v38.05 (Bushnell, 2018) with options ‘ktrim = r k=34 mink = 11 hdist = 1 tpe tbo qtrim = r trimq = 10’. Library quality was assessed using FastQC v0.11.8 (Andrews, 2018) and MultiQC v1.7 (Ewels et al., 2016), and trimmed r eads were aligned to the mm10 assembly using Bowtie 2 v2.2.9 (Langmead and Salzberg, 2012) with options ‘--end-to-end --very-sensitive -X 1000 --fr’, and only uniquely mapping reads with a minimum MAPQ of 10 were retained. Fragments aligning to the mitochondrial genome were removed, and duplicates were removed using Picard MarkDuplicates v2.6.0 (Broad Institute, 2016). Aligned fragments were shifted +4/–5 using deepTools alignmentSieve v3.0.1 with option ‘--ATACshift’ (Ramírez et al., 2016) and used to call ATAC-seq peaks with an FDR <0.05 using MACS2 v2.1.1 (Liu, 2014) with options ‘--SPMR -B --keep-dup all --format BAMPE –call-summits’.

Hi-C re-analysis

Request a detailed protocol

Paired-end Hi-C data from primary murine BMDMs stimulated with 10 ng/mL IFNG for 2 hr (Platanitis et al., 2022) were downloaded from SRA using fasterq-dump v2.10.5 (NCBI, 2022). Library quality was assessed using FastQC v0.11.8 (Andrews, 2018) and MultiQC v1.7 (Ewels et al., 2016), and reads were aligned to the mm10 assembly using BWA-MEM 0.7.17 (Li, 2013) with arguments ‘-SP’, and the resulting bam file was converted to pairsam format pairtools parse v0.2.2 (https://github.com/mirnylab/pairtools; Goloborodko, 2019). Technical replicates were merged using pairtools merge v0.2.2. Duplicate reads were marked using pairtools dedup v0.2.2, and only aligned fragments with pairtools classification ‘UU’ or ‘UC’ were retained using pairtools filter v0.2.2 resulting in approximately 600 M pairs. A Knight Ruiz (KR)-normalized, Arima restriction site-aware Hi-C matrix was prepared using juicer pre v1.22.01 (Dudchenko et al., 2017) at 5 kb resolution and fitted to a powerlaw distribution in preparation for running the Activity-by-Contact (ABC) model (Fulco et al., 2019).

Gene-enhancer target prediction by Activity-by-Contact (ABC) analysis

Request a detailed protocol

We applied the Hi-C matrix (Platanitis et al., 2022) in conjunction with ATAC-seq (Platanitis et al., 2022) and H3K27ac (Piccolo et al., 2017) data from IFNG-stimulated BMDMs to predict enhancer activity using the Activity-by-Contact (ABC) model (Fulco et al., 2019). The ABC model predicts enhancer-gene contacts by leveraging epigenomic and chromatin interaction capture data. Each potential enhancer-gene contact is assigned an ABC interaction score that depends on the activity of the enhancer by ATAC-seq and H3K27ac ChIP-seq in addition to the likelihood of contact by Hi-C. We ran the ABC model as previously described (Fulco et al., 2019). In brief, we first identified candidate enhancer elements using makeCandidateRegions.py with options ‘--peakExtendFromSummit 250 –nStrongestPeaks 150000’ and quantified activity using run.neighborhoods.py. Predicted enhancer-gene pairs were attributed an ABC interaction score using predict.py with options ‘--hic_type juicebox –hic_resolution 5000 –scale_hic_using_powerlaw –threshold 0.02 –make_all_putative’. Only enhancer regions with an ABC interaction score over 0.001 were considered for subsequent analysis.

Orthology analysis

Request a detailed protocol

We used BioMart with human Ensembl v105 annotation (Cunningham et al., 2022) to identify high confidence, one-to-one orthologs in mouse for each identified human ISG (log2FC >0, FDR adjusted p-value <0.05). A union set consisting of mouse and human-to-mouse ISGs was generated, and each ISG was identified as mouse-specific, human-specific, or shared according to induction status. To determine how broadly B2_Mm2 has shaped murine innate immune responses, we identified 344 unique B2_Mm2 elements fully overlapping enhancers predicted to interact with 706 ISGs using the ABC model with a minimum ABC score of 0.001. The proportion of ISGs predicted to interact with a putative B2_Mm2 enhancer were plotted according to species status. We independently identified 926 ISGs with at least one of 655 STAT1-bound B2_Mm2 elements (irrespective of ABC) within 50 kb of the transcriptional start site.

Cell line passing and interferon treatments

Request a detailed protocol

J774A.1 mouse cells were purchased from ATCC and were cultured in DMEM supplemented with 1 X penicillin-streptomycin and 10% fetal bovine serum. J774A.1 cells were routinely passaged using 0.25% Trypsin-EDTA and cultured at 37 °C and 5% CO2. All IFNG treatments were performed using 100 ng/mL recombinant mouse IFNG (R&D Systems #485-MI-100). Cells were confirmed to be Mycoplasma-free by the Barbara Davis Center for Childhood Diabetes BioResources Core Molecular Biology Unit at the University of Colorado Anschutz Medical Center. The identity of the cells was verified using the ATCC Cell Line Authentication Service (STR profiling).

Cytokine panel

Request a detailed protocol

All treatments were carried out using a 4 hr time period. IL-4 (Sigma-Aldrich #I1020-5UG) was added to a final concentration of 1 ng/mL, recombinant IFNA (R&D Systems #12100–1) to a final concentration of 1000 U/mL, recombinant IL-2 (R&D Systems #402 ML-020) to a final concentration of 100 ng/mL, and recombinant IL-6 (Sigma-Aldrich #I9646-5UG) to a final concentration of 20 ng/mL in accordance with manufacturers’ recommendations. RNA was extracted using an Omega Mag-Bind Total RNA Kit (Omega Bio-Tek #M6731-00) and analyzed via RT-qPCR.

Design of gRNA constructs

Request a detailed protocol

Two gRNA sequences were designed to flank each side of B2_Mm2.Dicer1 in order to delete the element and generate knockout J774A.1 cells via SpCas9 (Integrated DNA Technologies #1081060). All gRNA sequences were also verified to uniquely target the locus of interest using the UCSC BLAT tool (Kent, 2002) against the mm10 genome assembly.

Generation of CRISPR KO cell lines

Request a detailed protocol

After gRNAs were designed, we used Alt-R Neon electroporation (1400 V pulse voltage, 10 ms pulse width, 3 pulses total) with four different combinations of the gRNAs to target B2_Mm2. One set of guides was found to produce the expected doubles-stranded cuts on both sides of the element in the bulk electroporated cell populating using gel electrophoresis. Clonal lines were isolated using the array dilution method and screened for the expected homozygous deletion using primers flanking the B2_Mm2.Dicer1 element. Clonal lines homozygous for the deletion were further validated using one flanking primer and one primer internal to B2_Mm2.Dicer1. To determine deletion breakpoint sequences, PCR products flanking each deletion site were cloned into a sequencing vector using the CloneJET PCR Cloning Kit (Thermo Fisher Scientific #K1231) and transformed into 5-alpha Competent E. coli (New England Biolabs #C2987H). Plasmid DNA was harvested using the EZNA Omega Plasmid DNA Mini Kit I (Omega Bio-Tek #D6942-02), and the sequence of each construct was verified by Sanger sequencing (Quintara Biosciences, Fort Collins, CO; Genewiz, South Plainfield, NJ). Sequencing results were visualized by aligning to the mm10 reference genome using BLAT (Kent, 2002). We identified two clonal lines homozygous for the B2_Mm2.Dicer1 deletion for further experimentation.

Quantifying Dicer1 expression using RT-qPCR

Request a detailed protocol

Real-time quantitative polymerase chain reaction (RT-qPCR) was used to quantify Dicer1 expression in WT and knockout cell lines. WT J774A.1 cells used in RT-qPCR are all biological replicates that underwent the same single cell seeding process as the KOs to serve as a control. Forward and reverse primers were designed for CTCF, Dicer1, and Gbp2b. Each set of primers was designed using a combination of tools from NCBI Primer BLAST (Ye et al., 2012), Benchling (Benchling), and IDT RT-qPCR Primer Design. The final primer sequences chosen were confirmed to uniquely bind to the desired target sequence using BLAT (Kent, 2002). RT-qPCR reactions were prepared using the Luna Universal One-Step RT-qPCR Kit (New England Biolabs #E3005S) according to the manufacturer’s instructions. RT-qPCR data were analyzed using CTCF as a housekeeping gene. A Cq, deltaCq, and deltaDeltaCq value were obtained for each well. These values were averaged to arrive at a mean deltaDeltaCq expression value for each treatment and genotype condition. Standard deviation and a two-tailed Student’s t-test were then calculated for each treatment and genotype condition. A p-value of less than 0.05 demonstrates there is a statistically significant difference in gene expression levels between two treatment and/or genotype conditions.

J774A.1 RNA-seq library preparation

Request a detailed protocol

The Zymo Quick RNA Miniprep Plus Kit (Zymo Research #R1504) was used to extract RNA from J774A.1 cells for all treatments except for the cytokine panel which used the Omega RNA Extraction Kit (Omega Bio-Tek #M6731-00). WT J774A.1 cells used in RNA-seq and all downstream analysis were clones that underwent single-cell expansions of from wild-type J774A.1 cells. All RNA lysates and single-use aliquots of extracted RNA were stored at –80 °C until library preparation. RNA integrity was quantified with High Sensitivity RNA TapeStation 4200 (Agilent). Libraries were generated using the KAPA mRNA HyperPrep Kit (KAPA Biosystems #08098123702) according to the manufacturer’s protocol. The final libraries were pooled and sequenced on a NovaSeq 6000 as 150 bp paired-end reads (University of Colorado Genomics Core).

J774A.1 RNA-seq analysis

Request a detailed protocol

Adapters and low quality reads were first trimmed using BBDuk v38.05 (Bushnell, 2018). Library quality was assessed using FastQC v0.11.8 (Andrews, 2018) and MultiQC v1.7 (Ewels et al., 2016) and trimmed reads were aligned to the mm10 assembly using HISAT2 v2.1.0 (Kim et al., 2019) with option ‘--rna-strandness RF’. Only uniquely aligned fragments were retained, and technical replicates were merged using samtools v1.10 (Li et al., 2009). CPM-normalized, stranded bigWigs were generated using deepTools bamCoverage v3.0.1 (Ramírez et al., 2016) and visualized using the UCSC Genome Browser (Kent et al., 2002). Aligned fragments were assigned to the mm10 refseq gene annotation in a reversely stranded manner using featureCounts v1.6.2 (Liao et al., 2014) with options ‘-t exon -s 2’, and differentially expressed genes were called using DESeq2 v1.26.0 (Ramírez et al., 2016). We analyzed every individual pairwise comparison to determine the effects due to both treatment and genotype. Untreated and wild type conditions were defined as the reference level. Log2FC values were shrunken using the apeglm function v1.8.0 (Zhu et al., 2019) for visualization across Chromosome 12 as a distance plot.

To determine whether the B2_Mm2.Dicer1 element acts as a regulator of splicing, we conducted transcriptome-guided transcript assembly on all WT and B2_Mm2.Dicer1 KO J774 RNA-seq alignments individually using Stringtie v1.3.3b with options ‘--rf -j 5’. Individual GTF files were merged into a single file with option ‘--merge’. Reads were subsequently aligned against the merged transcriptome using Salmon v1.9.0 with options ‘--validateMappings –rangeFactorizationBins4 –gcBias’. Transcript-level quantification data for each WT and KO sample were used for differential expression analysis with DESeq2 v1.26.0 (Ramírez et al., 2016).

CUT&Tag

Request a detailed protocol

CUT&Tag datasets were generated using a protocol from Kaya-Okur and Henikoff, 2020; Kaya-Okur et al., 2020 with an input of 100–500 k cells and the following modifications: pAG-Tn5 (EpiCypher #15–1017) was diluted 1:40 in nuclease-free water containing 20 mM HEPES pH 7.5, 300 mM NaCl, 0.5 mM spermidine, 0.01% digitonin, and a protease inhibitor tablet, and libraries were amplified for 14 cycles. The following primary antibodies were used: rabbit IgG (1:1000, EpiCypher #13–0042), rabbit anti-H3K27ac (1:100), rabbit anti-pRPB1-Ser5 (1:100, Cell Signaling Technology #13523 S), rabbit anti-STAT1 (1:100, Cohesion Biosciences #CPA3322), rabbit anti-pSTAT1-Ser727 (1:100, Active Motif #39634), rabbit anti-CTCF (1:100, EMD Millipore #07-729-25UL). Guinea pig anti-rabbit IgG (1:100, Antibodies-Online #ABIN101961) was used as a secondary antibody. CUTANA pAG-Tn5 (EpiCypher #15–1017) was added to each sample following primary and secondary antibody incubation. Pulldown success was measured by Qubit dsDNA High Sensitivity (Invitrogen) and TapeStation 4200 HSD5000 (Agilent) before proceeding to library preparation. Pulldowns were concentrated and pooled using KAPA Pure Beads (Roche). The final pooled libraries were quantities with TapeStation 4200 HSD5000 and sequences on an Illumina NovaSeq 6000 as 150 bp paired-end reads (University of Colorado Genomics Core).

CUT&Tag analysis

Request a detailed protocol

Adapters and low quality reads were first trimmed using BBDuk v38.05 (Bushnell, 2018). Library quality was assessed using FastQC v0.11.8 (Andrews, 2018) and MultiQC v1.7 (Ewels et al., 2016). Trimmed reads were then aligned to the mm10 assembly using BWA-MEM v0.7.15 (Li, 2013) and samtools (Li et al., 2009) retained only uniquely aligned fragments (MAPQ ≥ 10). Peaks were called without a control file using MACS2 v2.1.1 (Liu, 2014). bigWigs corresponding to read pileup per million reads for visualization on the UCSC Genome Browser (Kent et al., 2002).

External datasets

Request a detailed protocol

Publicly available data were downloaded from public repositories using fasterq-dump from the NCBI SRA Toolkit. RNA-seq datasets were obtained from GSE84517, GSE115434, GSE84691, GSE176562, and GSE43036. ChIP-seq datasets were obtained from GSE84518, GSE108805, GSE189971, and GSE115433. Hi-C and ATAC-seq datasets were obtained from SRA using accession PRJNA694816.

Data access

Request a detailed protocol

Raw and processed sequencing data generated in this study have been submitted to the NCBI Gene Expression Omnibus (GEO) with accession number GSE202574.

Code Availability

Request a detailed protocol

UCSC Genome browser sessions and all code available at https://genome.ucsc.edu/s/coke6162/B2_SINE_enhancers_Horton_et_al and https://github.com/coke6162/B2_SINE_enhancers (copy archived at Kelly, 2023).

Data availability

Raw and processed sequencing data generated in this study have been submitted to the NCBI Gene Expression Omnibus (GEO) with accession number GSE202574.

The following data sets were generated
    1. Horton I
    2. Kelly CJ
    3. Simpson DM
    4. Chuong EB
    (2022) NCBI Gene Expression Omnibus
    ID GSE202574. Mouse SINE B2 elements function as IFN-inducible enhancer elements.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE202574
The following previously published data sets were used
    1. Piccolo V
    2. Curina A
    3. Genua M
    4. Ghisletti S
    5. Simonatto M
    6. Sabò A
    7. Amati B
    8. Ostuni R
    9. Natoli G
    (2017) NCBI Gene Expression Omnibus
    ID GSE84517. Opposing macrophage-polarization programs show extensive epigenomic and transcriptional cross-talk [RNA-Seq].
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE84517
    1. Platanitis E
    2. Decker T
    (2019) NCBI Gene Expression Omnibus
    ID GSE115434. Transcription profile analysis of wild type and Irf9-/- bone marrow derived macrophages in response to type I and type II interferons.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE115434
    1. Qiao Y
    2. Giannopoulou E
    3. Zhang T
    (2016) NCBI Gene Expression Omnibus
    ID GSE84691. RNAseq to profile IFNg response in human primary monocytes.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE84691
    1. Qiao Y
    2. Li Y
    3. Giannopoulou E
    (2013) NCBI Gene Expression Omnibus
    ID GSE43036. Synergistic Activation of Inflammatory Cytokine Genes by Priming of Regulatory DNA Elements for Increased Transcription in Response to TLR Signaling.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE43036
    1. Piccolo V
    2. Curina A
    3. Genua M
    4. Ghisletti S
    5. Simonatto M
    6. Sabò A
    7. Amati B
    8. Ostuni R
    9. Natoli G
    (2017) NCBI Gene Expression Omnibus
    ID GSE84518. Opposing macrophage-polarization programs show extensive epigenomic and transcriptional cross-talk [ChIP_narrow].
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE84518
    1. Merkenschlager M
    2. Dharmalingam G
    3. Cuartero S
    (2013) NCBI Gene Expression Omnibus
    ID GSE108805. Transcriptional control of macrophage inducible gene expression by cohesin [ChIP-Seq II].
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE108805
    1. Gualdrini F
    2. Polletti S
    3. Simonatto M
    4. Prosperini E
    5. Natoli G
    (2022) NCBI Gene Expression Omnibus
    ID GSE189971. H3K9 trimethylation in active compartments optimizes stimulus-regulated transcription by restricting usage of CTCF sites in SINE-B2 repeats [ChIP-Seq].
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE189971
    1. Platanitis E
    2. Decker T
    (2019) NCBI Gene Expression Omnibus
    ID GSE115433. STAT1, STAT2 and IRF9 transcription factor binding analysis in wild type and Irf9-/- bone marrow derived macrophages in response to type I and type II interferons.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE115433
    1. Max Perutz Labs
    (2021) NCBI Sequence Read Archive
    ID PRJNA694816. 3D chromatin rearrangements in response to Interferon treatment.
    https://www.ncbi.nlm.nih.gov/sra/PRJNA694816

References

    1. Waterston RH
    2. Lindblad-Toh K
    3. Birney E
    4. Rogers J
    5. Abril JF
    6. Agarwal P
    7. Agarwala R
    8. Ainscough R
    9. Alexandersson M
    10. An P
    11. Antonarakis SE
    12. Attwood J
    13. Baertsch R
    14. Bailey J
    15. Barlow K
    16. Beck S
    17. Berry E
    18. Birren B
    19. Bloom T
    20. Bork P
    21. Botcherby M
    22. Bray N
    23. Brent MR
    24. Brown DG
    25. Brown SD
    26. Bult C
    27. Burton J
    28. Butler J
    29. Campbell RD
    30. Carninci P
    31. Cawley S
    32. Chiaromonte F
    33. Chinwalla AT
    34. Church DM
    35. Clamp M
    36. Clee C
    37. Collins FS
    38. Cook LL
    39. Copley RR
    40. Coulson A
    41. Couronne O
    42. Cuff J
    43. Curwen V
    44. Cutts T
    45. Daly M
    46. David R
    47. Davies J
    48. Delehaunty KD
    49. Deri J
    50. Dermitzakis ET
    51. Dewey C
    52. Dickens NJ
    53. Diekhans M
    54. Dodge S
    55. Dubchak I
    56. Dunn DM
    57. Eddy SR
    58. Elnitski L
    59. Emes RD
    60. Eswara P
    61. Eyras E
    62. Felsenfeld A
    63. Fewell GA
    64. Flicek P
    65. Foley K
    66. Frankel WN
    67. Fulton LA
    68. Fulton RS
    69. Furey TS
    70. Gage D
    71. Gibbs RA
    72. Glusman G
    73. Gnerre S
    74. Goldman N
    75. Goodstadt L
    76. Grafham D
    77. Graves TA
    78. Green ED
    79. Gregory S
    80. Guigó R
    81. Guyer M
    82. Hardison RC
    83. Haussler D
    84. Hayashizaki Y
    85. Hillier LW
    86. Hinrichs A
    87. Hlavina W
    88. Holzer T
    89. Hsu F
    90. Hua A
    91. Hubbard T
    92. Hunt A
    93. Jackson I
    94. Jaffe DB
    95. Johnson LS
    96. Jones M
    97. Jones TA
    98. Joy A
    99. Kamal M
    100. Karlsson EK
    101. Karolchik D
    102. Kasprzyk A
    103. Kawai J
    104. Keibler E
    105. Kells C
    106. Kent WJ
    107. Kirby A
    108. Kolbe DL
    109. Korf I
    110. Kucherlapati RS
    111. Kulbokas EJ
    112. Kulp D
    113. Landers T
    114. Leger JP
    115. Leonard S
    116. Letunic I
    117. Levine R
    118. Li J
    119. Li M
    120. Lloyd C
    121. Lucas S
    122. Ma B
    123. Maglott DR
    124. Mardis ER
    125. Matthews L
    126. Mauceli E
    127. Mayer JH
    128. McCarthy M
    129. McCombie WR
    130. McLaren S
    131. McLay K
    132. McPherson JD
    133. Meldrim J
    134. Meredith B
    135. Mesirov JP
    136. Miller W
    137. Miner TL
    138. Mongin E
    139. Montgomery KT
    140. Morgan M
    141. Mott R
    142. Mullikin JC
    143. Muzny DM
    144. Nash WE
    145. Nelson JO
    146. Nhan MN
    147. Nicol R
    148. Ning Z
    149. Nusbaum C
    150. O’Connor MJ
    151. Okazaki Y
    152. Oliver K
    153. Overton-Larty E
    154. Pachter L
    155. Parra G
    156. Pepin KH
    157. Peterson J
    158. Pevzner P
    159. Plumb R
    160. Pohl CS
    161. Poliakov A
    162. Ponce TC
    163. Ponting CP
    164. Potter S
    165. Quail M
    166. Reymond A
    167. Roe BA
    168. Roskin KM
    169. Rubin EM
    170. Rust AG
    171. Santos R
    172. Sapojnikov V
    173. Schultz B
    174. Schultz J
    175. Schwartz MS
    176. Schwartz S
    177. Scott C
    178. Seaman S
    179. Searle S
    180. Sharpe T
    181. Sheridan A
    182. Shownkeen R
    183. Sims S
    184. Singer JB
    185. Slater G
    186. Smit A
    187. Smith DR
    188. Spencer B
    189. Stabenau A
    190. Stange-Thomann N
    191. Sugnet C
    192. Suyama M
    193. Tesler G
    194. Thompson J
    195. Torrents D
    196. Trevaskis E
    197. Tromp J
    198. Ucla C
    199. Ureta-Vidal A
    200. Vinson JP
    201. Von Niederhausern AC
    202. Wade CM
    203. Wall M
    204. Weber RJ
    205. Weiss RB
    206. Wendl MC
    207. West AP
    208. Wetterstrand K
    209. Wheeler R
    210. Whelan S
    211. Wierzbowski J
    212. Willey D
    213. Williams S
    214. Wilson RK
    215. Winter E
    216. Worley KC
    217. Wyman D
    218. Yang S
    219. Yang S-P
    220. Zdobnov EM
    221. Zody MC
    222. Lander ES
    223. Mouse Genome Sequencing Consortium
    (2002) Initial sequencing and comparative analysis of the mouse genome
    Nature 420:520–562.
    https://doi.org/10.1038/nature01262

Decision letter

  1. Mia T Levine
    Reviewing Editor; University of Pennsylvania, United States
  2. Molly Przeworski
    Senior Editor; Columbia University, United States
  3. Kenji Ichiyanagi
    Reviewer; Nagoya University, Japan

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Mouse B2 SINE elements function as IFN-inducible enhancers" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Molly Przeworski as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Kenji Ichiyanagi (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

There was consensus among the reviewers that the topic is of broad interest and the data were rigorously generated. The reviewers, however, also felt that some of the claims require additional support. A request for new analyses of the current dataset in hand could potentially blunt these concerns. Depending on the results of these additional analyses, a single new experiment would be required to support the inference that the focal B2 element acts as a Dicer1 enhancer. Please see below for specific revisions requested.

Essential revisions:

1) It is not clear whether the Dicer1 locus is unique. Are there other loci that have both a nearby B2_Mm2 element and a binary difference between inducibility in mouse versus human cells? It would be helpful to understand whether there other B2_Mm2 insertions that could contribute to a mouse-specific IFGN-sensitivity. Adding DEseq values from human RNAseq data the authors already use (current references 10 and/or 37) for identifiable human orthologs to Table S7 would thus strengthen their conclusions. If there are many potential candidates, the authors should discuss the rationale for selecting Dicer1 in particular.

2) The results of Serpina3g and Serpina3F gene expression in the authors' knockout cells are very interesting. However, the authors focus almost exclusively on Serpina3g and Serpina3F, which makes it difficult to understand what is happening genome wide. Are other IFNΓ-induced genes (including those not on chromosome 12) similarly affected at the level of basal or induced transcription? How many genes are different in WT versus KO cells, both at basal and induced states? Does this correlate with their CUT&TAG data shown in Figure 5? By focusing only on nearby genes (Serpina3g and Serpina3F), the authors hint that this may be a long-range regulatory effect, "potentially mediated by the CTCF binding activity of the element" that they removed. But by only focusing on two nearby IFNΓ-induced genes, their data do not rule out the (also potentially quite interesting) possibility that there may be a more indirect role for this TE site or Dicer1 in basal transcription of IFNΓ-induced genes or IFNΓ-mediated gene expression.

3) The specificity of the KO effect on the Dicer1-Serpina region of chromosome 12 is not clear. Without analysis of the complete RNA-seq and CUT&RUN datasets, it is difficult to rule out a more global effect (i.e., beyond chromosome 12). If these new analyses yield evidence of specificity of the KO lines, the reviewers will be satisfied. If not, the reviewers request an additional manipulation: KO of an intron element of equivalent size to the original deletion, KO of a different B2 element – apparently there is one 2kb away, or even better, replacement of the B2 element with one that lacks cGAS motifs (though the final suggestion is likely too technically challenging). Determining whether there are changes in the basal or induced levels of Dicer1/Serpina genes in this additional line would serve as an important control for the KO experiment. Providing more data on other genes throughout the genome in WT and KO cells, which the authors have generated but do not include in the manuscript, would help distinguish between these models.

4) There are high levels of POLR2A occupancy at the B2_Mm2.Dicer1 element in induced WT cells. Could this be a Pol2 pause site? Could deletion of this element lead to a change in Pol2 occupancy and change Dicer1 expression independent of enhancer activity? To probe such questions, the reviewers requested that the authors directly test the possibility that the intronic B2 element actually acts as a regulator of splicing or transcriptional elongation. Careful analysis of the Dicer1-mapping reads from the RNA-seq data – or RT-qPCR – could resolve this concern.

5) Figure 4F – The authors claim that "deletion of B2_Mm2.Dicer1 also has a significant repressive effect on the IFNΓ-inducible expression of Serpina genes." However, the basal levels of Serpina3f/Serpina3g are significantly reduced upon this deletion compared to WT. Furthermore, expression of Serpina genes in the KO cell lines significantly increase upon IFNΓ stimulation, suggesting that they still show inducible expression despite the B2_Mm2.Dicer1 deletion. The authors should compare the magnitude of induction before and after stimulation between the WT and KO cell lines to determine if there is indeed a repressive effect on inducible expression of Serpina genes.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Mouse B2 SINE elements function as IFN-inducible enhancers" for further consideration by eLife. Your revised article has been evaluated by Molly Przeworski (Senior Editor) and a Reviewing Editor.

The reviewers appreciated your many additional experiments/analyses as well as the edits to the text. The manuscript has been improved substantially but there are some remaining issues that need to be addressed.

1. All three reviewers felt that the response to Critique#1 was insufficient. There are several places in the manuscript referring to the significance of the Mouse B2 SINE elements contributing to lineage-specific immune response. For example, the introduction highlights that "a key open question is whether the co-option of TEs as immune regulatory elements is evolutionarily widespread as a mechanism driving divergence of innate immune responses."

The reviewers (or at least two of them) did not expect to see additional KO experiments but did expect the authors to highlight a handful of other examples where STAT1 binding to a proximal B2 SINE element is found in a mouse-specific ISG only. If no other such examples are available to highlight, then it's currently difficult to discern how generalizable this relationship is between rodent-specific B2 SINE elements and mouse-specific ISGs.

Furthermore, the authors highlight a limitation of the current human cell dataset – specifically the limitation of sampling gene expression only 24 hours after interferon treatment (new paragraph starting line 310) in human cells. However, in line 232, the same dataset is used to support the earlier statement that "the human ortholog DICER1 does not show IFNΓ-inducible expression in human primary macrophages (Qiao et al. 2016)…mouse DICER1 shows a significant 50% upregulation in response to IFNΓ in primary mouse BMDMs…."

This earlier statement is substantially weakened by the final Results paragraph. This issue needs to be resolved to support the major claim of the paper that a lineage-specific TE is responsible for a lineage-specific immune response, echoed in the abstract: "B2 elements…exemplifies how lineage-specific TEs can facilitate evolutionary divergence of innate immune regulatory networks."

2. In the response to reviewer comments under critique#2, it was stated that "…the other differentially expressed genes in the KOs most likely represent off-target/stochastic changes, that are commonly seen across separate clonal isolations." And as communicated in the manuscript: Line 288: "The other dysregulated genes showed no discernable physical or functional pattern and also showed high variability between individual clones…consistent with intrinsic clonal transcriptional variation…" Whether this is true for all 101 genes should be clarified. It seems surprising given that at least some genes should be indirect targets of Dicer1. Finally, does Dicer1 show less variability than these other genes? If so, that should be stated.

3. Finally, one reviewer was concerned that the intronic location of B2 SINE means that the authors cannot delineate whether the element acts as an enhancer or instead a regulator of transcriptional elongation.

https://doi.org/10.7554/eLife.82617.sa1

Author response

Essential revisions:

1) It is not clear whether the Dicer1 locus is unique. Are there other loci that have both a nearby B2_Mm2 element and a binary difference between inducibility in mouse versus human cells? It would be helpful to understand whether there other B2_Mm2 insertions that could contribute to a mouse-specific IFGN-sensitivity. Adding DEseq values from human RNAseq data the authors already use (current references 10 and/or 37) for identifiable human orthologs to Table S7 would thus strengthen their conclusions. If there are many potential candidates, the authors should discuss the rationale for selecting Dicer1 in particular.

We agree with the reviewers that it would be useful to investigate whether other B2 elements contribute to mouse-specific IFN-inducible expression, which may indicate whether the Dicer1 example is unique or part of a larger species-specific network of genes regulated by B2 elements. As suggested, we have re-analyzed a publicly available human RNA-seq dataset to identify human ISGs in CD14+ monocytes stimulated with IFNΓ for 24 hours. A more detailed description of our analysis is provided below. In summary, we classified mouse ISGs based on whether those genes were also ISGs based on the human RNA-seq data, and integrated mouse Hi-C data to predict enhancer/gene interactions. We found that B2_Mm2 elements are associated nearby hundreds of mouse-specific ISGs, consistent with a role in regulating a network of mouse ISGs. However, we note several caveats below, including the observation that B2_Mm2 elements are also associated with many non mouse-specific ISGs, and cannot be causally tied to gene regulation without additional experimental perturbation.

Using RNA-seq data of IFNΓ-stimulated human CD14+ monocytes (GEO accession GSE84691) (Qiao et al., 2016), we identified a total of 2314 human ISGs (FDR adjusted p-value < 0.05, log2 fold change > 0), 1453 of which had a 1:1 ortholog assignment to a mouse gene based on ENSEMBL Biomart (Cunningham et al., 2022). We integrated the mouse and human datasets to classify mouse genes into 3 categories:

i) "mouse-specific ISGs" (i.e., genes that have a 1:1 mouse/human ortholog but are only ISGs in mouse OR mouse ISGs with no direct ortholog in human)

ii) "shared ISGs" (i.e., genes that have a 1:1 mouse/human ortholog and are ISGs in both species)

iii) "human-specific ISGs" (i.e., genes that have a 1:1 mouse/human ortholog but are only ISGs in human).

The DESeq2 values for this dataset are now reported in Supplementary file 1. Notably human DICER1 does not appear to be induced (Supplementary file 1, Figure 3 —figure supplement 1). To compare this set to mouse ISGs, we used ISGs defined from BMDMs stimulated with IFNΓ for 4 hours from the Piccolo et al. dataset (GSE84517) analyzed in our previous submission (Piccolo et al., 2017). From this dataset, we defined 2720 mouse genes as ISGs (FDR adjusted p-value < 0.05, log2FC > 0). We compared this set of genes to the 1453 human ISGs defined above (i.e., genes that are an ISG in human CD14+ monocytes and have a direct 1:1 human:mouse orthology relationship). Out of the 2720 mouse ISGs, 2266 genes (83.3%) were only ISGs in mouse (including 1467 genes with a 1:1 human ortholog and 799 genes with no 1:1 ortholog), and 454 (16.6%) were ISGs in both mouse and human. Conversely, out of the 2314 human ISGs, 1860 genes (80.4%) were only ISGs in human (including 999 genes with a 1:1 mouse ortholog and 861 genes with no 1:1 ortholog). The relatively low fraction of ISGs shared across species is consistent with a previous study comparing ISGs induced by type I interferon α across vertebrates (Shaw et al., 2017), where a core of 62 ISGs were conserved across species (in our IFNΓ datasets, 56 of these are induced in either mouse or human, 40 of which are induced in both species).

We next sought to determine how broadly B2_Mm2 has shaped murine innate immune responses using our list of mouse-specific, human-specific, and shared ISGs. We first assigned predicted enhancers to their predicted targets using the Activity-by-Contact (ABC) model (Fulco et al., 2019), which incorporates both epigenomic signal and Hi-C 3D interaction data to predict enhancer-gene targets. As input into the ABC model, we used publicly available ATAC-seq and Hi-C data (PRJNA694816) (Platanitis et al., 2022) from murine BMDMs stimulated with IFNΓ for 2 hours along with the H3K27ac ChIP-seq and RNA-seq datasets included in the manuscript to identify putative enhancer-gene contacts. We identified 50264 putative enhancer-ISG contacts attributed a minimum ABC likelihood score of 0.01, with an average of 14 enhancers for each gene

represented in our list of mouse- and human-to-mouse ISGs. We subsequently identified 393 "mouse-specific" 142 "human-specific", and 64 "shared" ISGs associated with at least one B2_Mm2 enhancer. A list of all mouse and human-to-mouse ISGs along with the top predicted B2_Mm2-derived enhancer can be found in Supplementary file 7.

Of 2266 mouse-specific ISGs, 393 (17.3%) were predicted to be regulated by at least one B2_Mm2 element. We confirmed that the B2_Mm2.Dicer1 element was predicted by ABC to target Dicer1 (a mouse-specific ISG) with an ABC likelihood score of 0.127508 (Supplementary file 7). This finding is consistent with the idea that B2_Mm2 elements have facilitated the evolution of the mouse-specific ISG regulatory network. However, we also found that B2_Mm2 elements are predicted to regulate 64 out of 454 (14.1%) genes that are shared ISGs in both mouse and human, and 142 out of 999 (14.2%) of genes that are only ISGs in human (Author response image 1). Additionally, we did not observe a difference in the distribution of ABC scores between mouse- and human-specific ISGs for B2_Mm2-derived enhancers as well as nonresponsive TE families (Author response image 2). This indicates that the association with a B2_Mm2-derived enhancer is not strongly predictive that a gene is a mouse-specific ISG.

There are a few notable caveats with this analysis. (1) the human RNA-seq dataset was generated by a different lab and derived from cells stimulated with IFNΓ for 24 hours, whereas the two mouse RNA-seq datasets collected cells 2 or 4 hours after IFNΓ stimulation. Previous studies have shown that transcriptional responses to type I and II IFNs vary across timepoints (Piccolo et al., 2017; Sumida et al., 2022). Genes that are not induced in human cells after 24 hours may potentially be induced at 4 hours. As such, the number of shared ISGs is likely an underestimate. Careful analysis of human and mouse datasets derived from cells collected at similar timepoints may improve identification of species-specific ISGs and potentially result in a larger difference in the number of B2_Mm2 elements predicted to drive species-specific ISG expression. (2), we used a relaxed ABC likelihood score threshold of 0.01 such that the number of B2_Mm2-derived enhancer-gene contacts identified likely represent the maximum possible using these data. These predictions also did not distinguish between elements that show constitutive or IFN-inducible regulation, so it is possible that some B2 elements may have act as constitutive rather than IFN-inducible enhancers. (3) we were able to confidently identify only 1453 (62.8%) human-to-mouse, one-to-one ISG orthologs. Careful analysis of all potential orthologs and paralogs (such as species-specific gene family expansions) may better reveal the full extent to which B2_Mm2 has contributed to species-specific immune regulatory evolution.

In summary, while our reanalysis is indeed consistent with a slightly increased enrichment of B2_Mm2 elements near mouse-specific ISG (e.g., 17.3% of mouse-specific ISGs are predicted to be regulated by B2_Mm2 elements compared to 14.1% of shared human/mouse ISGs), there are many caveats that we feel are outside the scope of this study to address and therefore we have opted not to include these analyses in the study.

We initially decided to study the B2_Mm2 within the Dicer1 intron for several reasons that were outlined in the main text, including the species-specific IFN-inducible expression of Dicer1 and strong epigenomic signature of inducible enhancer activity (STAT1, H3K27ac, CTCF) of the specific B2_Mm2 element. We have clarified our process for selecting the element as below, incorporating the revised ABC analysis to assign enhancers to genes through 3D interactions as described above.

“Having established that B2_Mm2 elements are an abundant source of IFNΓ-inducible STAT1 binding sites in the mouse genome, we asked whether any of these elements have been co-opted to regulate expression of individual ISGs. We first assigned predicted enhancers to their predicted targets using the Activity by Contact (ABC) model (Fulco et al., 2019), which incorporates both epigenomic signal and Hi-C 3D interaction data to predict enhancer-gene targets. As input into the ABC model, we used publicly available ATAC-seq and Hi-C data from murine BMDMs stimulated with IFNΓ for 2 hours (PRJNA694816) (Platanitis et al., 2022) and H3K27ac ChIP-seq and RNA-seq data from (Piccolo et al. 2017). Focusing on a permissive set of 2720 ISGs from (Piccolo et al. 2017) (FDR adjusted p-value <0.05, log2 fold change > 0), we identified 530 B2_Mm2 elements predicted to interact with 457 mouse ISGs (16.8% of the set of 2720 in this analysis) (Supplementary file 7). Compared with a set of human ISGs from human CD14+ monocytes (Qiao et al., 2016), 393 of these 457 (86%) genes were only ISGs in mouse, and the remaining 64 (14%) were ISGs in both mouse and human.

From this set, we decided to focus on a specific B2_Mm2 element located on Chromosome 12 within the first intron of Dicer1, which is an endonuclease responsible for recognizing and cleaving foreign and double stranded RNA that has been linked to innate immunity (MacKay et al. 2014; Poirier et al. 2021; Chiappinelli et al. 2012; Gurung et al. 2021). While the human ortholog DICER1 does not show IFNΓ-inducible expression in human primary macrophages (Qiao et al., 2016) (Figure 3 —figure supplement 1, Supplementary file 1), mouse Dicer1 shows a significant 50% upregulation in response to IFNΓ in primary mouse BMDMs (Figure 3A). This indicates that Dicer1 is a mouse-specific ISG and likely acquired IFNΓ-inducible expression in the mouse lineage, potentially due to the co-option of the B2_Mm2 element as a species-specific IFNΓ-inducible enhancer. The intronic B2_Mm2 element (B2_Mm2.Dicer1) shows biochemical hallmarks of enhancer activity including inducible STAT1 and H3K27ac signal as well as constitutive binding by CTCF and RAD21 (Figure 3B). The element provides the only prominent nearby STAT1 binding site and is not present in rat or other mammals (Figure 3B). Therefore, we hypothesized that the B2_Mm2.Dicer1 element was co-opted as an IFNΓ-inducible enhancer of mouse Dicer1.”

Author response image 1
Download asset Open asset
Activity-by-Contact (ABC) analysis of B2_Mm2 enhancers and ISGs.

(A) Proportions of ISGs (mouse-specific, human-specific, shared) predicted to be regulated by a B2_Mm2 enhancer. (B) Proportions of ISGs within 50 kb of a STAT1-bound B2_Mm2 element.

Author response image 2
Download asset Open asset
Distribution of Activity-by-Contact scores in mouse IFNΓ-stimulated bone-marrow derived macrophages between various TE families and mouse-specific/human-specific/shared ISGs.

(A) B2_Mm2. (B) PB1D7. (C) ID_B1. (D) MT2B2. (E) MIR1_AMN. (F) MIR3.

2) The results of Serpina3g and Serpina3F gene expression in the authors' knockout cells are very interesting. However, the authors focus almost exclusively on Serpina3g and Serpina3F, which makes it difficult to understand what is happening genome wide. Are other IFNΓ-induced genes (including those not on chromosome 12) similarly affected at the level of basal or induced transcription? How many genes are different in WT versus KO cells, both at basal and induced states? Does this correlate with their CUT&TAG data shown in Figure 5? By focusing only on nearby genes (Serpina3g and Serpina3F), the authors hint that this may be a long-range regulatory effect, "potentially mediated by the CTCF binding activity of the element" that they removed. But by only focusing on two nearby IFNΓ-induced genes, their data do not rule out the (also potentially quite interesting) possibility that there may be a more indirect role for this TE site or Dicer1 in basal transcription of IFNΓ-induced genes or IFNΓ-mediated gene expression.

We agree that a more global assessment of the transcriptome-wide changes is necessary to fully understand the direct and indirect effects of the B2_Mm.Dicer1 element. However, the RNA-seq dataset for WT and KO cells generated in our initial submission was not suited for this comparison because (i) they were generated in different batches, and (ii) the WT control was a pool of WT J774 cells rather than clonal controls. Therefore, we generated new RNA-seq data where WT and KOs were treated in the same batch, using 3 additional control WT/unedited clones that were carried through the limited dilution process. For each WT (N=3) and KO (N=2) independently derived clonal line, we carried out 3 experimental replicates of untreated cells and cells treated for IFNΓ for 4 hrs, and harvested RNA for RNA-seq. These were all conducted in the same round of experiments to avoid potential batch effects.

We re-analyzed our new RNA-seq data to address the concerns raised by reviewers' comment (2) and comment (3). Consistent with our previous analysis, KO cells showed significantly reduced expression of Dicer1 in both untreated and IFNΓ-induced conditions. We first conducted DESeq2 pairwise differential expression analysis considering all replicates, WT (N=9) vs KO (N=6) in untreated cells showed Dicer1 downregulation with log2FC = -0.43, FDR adjusted p-value = 9×10-4 (see Author response image 3). In IFNΓ treated cells, the WT (N=9) vs KO (N=6) comparison showed Dicer1 downregulation with log2FC = -0.80, FDR adjusted p-value = 3.393×10-15 (see Author response image 4). As before, all KO lines showed the same non-inducible expression of Dicer1. Within all WT clones, Dicer1 showed significant IFNΓ-inducibility (log2FC = 0.35; FDR adjusted p-value = 1.7×10-6) while this inducibility was ablated in KO cells (log2FC = 0; FDR adjusted p-value = 0.98).

Author response image 3
Download asset Open asset
MA plot of all WT replicates vs all KO replicates in the untreated condition.

Dicer1 is labeled and is highly expressed and modestly but significantly downregulated (log2FC = -0.43, FDR adjusted p-value = 9×10-4).

Author response image 4
Download asset Open asset
MA plot of all WT replicates vs all KO replicates in the IFNΓ-treated condition.

Dicer1 is labeled and is highly expressed and significantly downregulated (log2FC = -0.43, FDR adjusted p-value = 9×10-4).

We next examined how Dicer1 compares to other genes affected genome-wide by the knockout. Comparing IFNΓ-treated WT to KO cells, Dicer1 was the 102nd most dysregulated gene (out of 3567 total genes with FDR adjusted p-value < 0.05). In untreated cells, Dicer1 was ranked 1741 out of differentially expressed genes based on FDR adjusted p-value (out of 3992 total genes with FDR adjusted p-value < 0.05). In other words, in untreated KO cells, there are 1740 genes that show a more significant change than Dicer1. In contrast, in IFNΓ-treated KO cells, there are just 101 genes that show a more significant change than Dicer1. In both conditions, the other dysregulated genes are located across all chromosomes and did not show any clear functional patterns or enrichment. We did not observe correlations between these genes and our CUT&TAG data, with the caveat that our CUT&TAG was performed on just one KO clone (KO1) and the pooled WT cells. We expect that CUT&TAG data would correspond to RNA-seq data if derived from the same clone.

Based on the global distribution across the genome, the other differentially expressed genes in the KOs most likely represent off-target/stochastic changes, that are commonly seen across separate clonal isolations. We routinely use CRISPR KO and observe similar levels of off-target effects by RNA-seq even across control lines, presumably due to the process of clonal isolation. We have observed greater on-target gene expression differences when using CRISPRi (dCas9-Krab) to target enhancers in pooled cells (avoiding clonal isolation), but we decided not to use CRISPRi for this element, reasoning that the element is located within the gene intron and dCas9-Krab binding could impede Dicer1 transcription. While it is a potentially interesting scenario that the observed dysregulation of other genes is an indirect effect of Dicer1 downregulation, the untreated KO cells showed only a modest downregulation of Dicer1 (log2FC = -0.43) which is may not sufficiently alter Dicer1 protein levels or function. The functional consequence of IFNΓ-inducible regulation on Dicer1 would be interesting to follow up in a future study.

Summary of changes

– Figure 4B-E is updated to use numbers from the NA-seq reanalysis

– Main text updated as follows

“We used RNA-seq to study the genome-wide effects of the B2_Mm2.Dicer1 element in both knockout clones and 3 control wild-type clones which were also isolated by limiting dilution. Consistent with the RT-qPCR results, we found that Dicer1 showed significant IFNΓ-inducible upregulation in all WT clones but that this induction was completely ablated in B2_Mm2.Dicer1 KO clones (Figure 4B). Notably, the RNA-seq normalized count data revealed that levels of Dicer1 were also significantly reduced in untreated knockout cells (Figure 4B, 5A). This indicates that the B2_Mm2.Dicer1 element regulates both basal expression levels of Dicer1 and inducible expression by IFNΓ. Focusing on the IFNΓ-treated condition, Dicer1 was significantly downregulated in KO cells compared to WT cells cells (log2FC = -0.80, adjusted p-value = 3.393×10-15). Genome-wide, there were 101 genes that showed greater significance than Dicer1 when testing for differential expression between IFNΓ-treated KO and WT cells (out of 3567 total genes with FDR adjusted p-value < 0.05; Supplementary file 8). The other dysregulated genes showed no discernable physical or functional pattern and also showed high variability between individual clones, consistent with clonal transcriptional variation typical of the limiting dilution process (Lee et al., 2019).”

3) The specificity of the KO effect on the Dicer1-Serpina region of chromosome 12 is not clear. Without analysis of the complete RNA-seq and CUT&RUN datasets, it is difficult to rule out a more global effect (i.e., beyond chromosome 12). If these new analyses yield evidence of specificity of the KO lines, the reviewers will be satisfied. If not, the reviewers request an additional manipulation: KO of an intron element of equivalent size to the original deletion, KO of a different B2 element – apparently there is one 2kb away, or even better, replacement of the B2 element with one that lacks cGAS motifs (though the final suggestion is likely too technically challenging). Determining whether there are changes in the basal or induced levels of Dicer1/Serpina genes in this additional line would serve as an important control for the KO experiment. Providing more data on other genes throughout the genome in WT and KO cells, which the authors have generated but do not include in the manuscript, would help distinguish between these models.

As described in our response above for Reviewer Critique #2, we have generated RNA-seq data for 3 new control lines generated by taking wild-type cells through clonal isolation. We agree with the reviewers that a "negative control KO" would be the best control, however in our past experience with CRISPR we find that most off-target/non-specific expression changes arise due to the process of clonal isolation. We also repeated the experiments and RNA-seq from our 2 KO lines such that our comparisons would not be confounded by batch effects.

Our new RNA-seq analysis of the 3 clonal WT lines and 2 KO lines confirmed that Dicer1 is consistently IFNΓ-inducible in the controls but not inducible in the KOs, with lower basal and induced levels (updated Figure 4A-B). However, and importantly, we found that Serpina3f and Serpina3g did not show the same pattern of regulation we observed previously (significantly reduced untreated and IFNΓ-induced expression in KO cells). In the new round of RNA-seq, we found that Serpina3f/g did not show significantly or consistently reduced expression in untreated KO cells (Supplementary file 8). These results conflict with our previous claim that Serpina3f and Serpina3g are regulated by the B2_Mm2.Dicer1 element, and instead suggest that our previous observation was an artifact of clonal variation and the extremely strong (log2fc > 10) IFNΓ-inducibility of Serpina3g within each genotype (in WT, inducible log2FC = 11.3, FDR adjusted p-value = 4.27×10-183; in KO, inducible log2FC 10.4; FDR adjusted p-value ≈ 0) and Serpina3f (WT log2FC = 10.4, FDR adjusted p-value = 4.8E×10-99; KO log2FC 10.3, FDR adjusted p-value = 9.3×10-100). It is likely that our previous observation was due to the fact that our KO1/2 RNA-seq datasets were generated in a separate experimental batch than the WT control, which was a pooled set of cells. Furthermore, our chromatin interaction analysis (described in response to Reviewer Critique #1) predicts that the B2_Mm2.Dicer1 element targets Dicer1, but does not target the Serpina genes. Taken together, our new controls support a specific regulatory effect of the element on Dicer1 but not the Serpin genes (as previously claimed). As such, we have removed our claims that Serpin genes are affected by the B2_Mm2.Dicer1 knockout from the main text. While our re-analysis indicates that the B2_Mm2.Dicer1 element does not have long-range enhancer activity as we previously concluded, this does not affect the major claims of our study. While the long-range distal enhancer activity was potentially an interesting observation, the fact that B2_Mm2.Dicer1 has a clear and reproducible regulatory effect on the more proximal gene Dicer1 is consistent with our central claim that B2_Mm2 elements act as IFNΓ-inducible enhancers.

The following paragraph was removed from the main text:

Our RNA-seq analysis revealed that deletion of B2_Mm2.Dicer1 also had a significant repressive effect on the IFNΓ-inducible expression of Serpina genes. In mouse, Serpina is a family of inflammatory genes represented by a large cluster of 14 murine paralogs, located 330 – 1000 kb from B2_Mm2.Dicer1. In particular, we found that Serpina3f and Serpina3g (approximately 500 kb from B2_Mm2.Dicer1) show a 1-2 orders of magnitude reduction in basal and induced gene expression in knockout cells compared to WT cells (Figure 4E-F, 5B). The relatively long range of this regulatory activity is not uncommon in vertebrate genomes (Schoenfelder and Fraser, 2019) and is potentially mediated by the CTCF binding activity of the element.

4) There are high levels of POLR2A occupancy at the B2_Mm2.Dicer1 element in induced WT cells. Could this be a Pol2 pause site? Could deletion of this element lead to a change in Pol2 occupancy and change Dicer1 expression independent of enhancer activity? To probe such questions, the reviewers requested that the authors directly test the possibility that the intronic B2 element actually acts as a regulator of splicing or transcriptional elongation. Careful analysis of the Dicer1-mapping reads from the RNA-seq data – or RT-qPCR – could resolve this concern.

We agree that it would be interesting if the B2_Mm2.Dicer1 altered Dicer1 transcription separately from its function as an inducible enhancer, and thank the reviewer for the suggestion. As shown in Figure 5, deletion of the B2_Mm2.Dicer1 element does remove Pol2 occupancy over the element and reduces overall signal over the locus, consistent with enhancer activity but not mutually exclusive with pausing activity. If the intronic B2_Mm2.Dicer1 element acts as a regulator of splicing, we would expect changes in isoform expression. As suggested, we manually inspected aligned RNA-seq reads from all WT and KO untreated/treated samples. Qualitatively, we observed no evidence of differential splicing or exon skipping when looking at aligned fragments from WT J774 and B2_Mm2.Dicer1 KO J774 cells (see below).

To more formally analyze this possibility with the data available, we conducted transcript assembly on all the samples and tested for differential isoform usage focusing on Dicer1 transcripts. As suggested by the reviewers, if the B2_Mm2.Dicer1 element acts as a pause site, it may alter splice site or polyadenylation site usage and cause measurable differential expression of specific Dicer1 splice variants in KO cells. To test this, we ran Stringtie v2 to conduct transcript assembly using the RNA-seq reads from each WT and KO individual sample, then used the ‘--merge’ flag to generate a merged gene annotation file representing isoforms from all samples. In parallel we also generated merged GTF files created using alignment files only from WT or B2_Mm2.Dicer1 KO samples. All exons identified in the fully merged GTF file are represented in the WT- and KO-specific GTF files. At the transcript level, all Dicer1 isoforms in the fully merged GTF file are also represented in the WT- and KO-specific GTF file. Together, this indicates that there are no isoforms that are unique to either KO and WT cells (see Author response image 5).

Author response image 5
Download asset Open asset
UCSC Snapshot of the Dicer1 locus in mouse, with custom Stringtie transcript-level annotation (top three tracks), GENCODE gene-level annotation (fourth track), and alignments from clonal WT (fifth track) and B2_Mm2.

Dicer1 KO (sixth track) J774.A1 cells.

To quantify differences in isoform expression, we used DESeq2 to measure differences in isoform expression across WT and B2_Mm2.Dicer1 KO samples using the merged Stringtie isoform information as input (gene-level analysis was used for the manuscript). This revealed several transcripts that all show similar trends to the gene-level analysis (see Supplementary file 9), where expression is reduced in both basal and IFN-treated conditions in the KO, and inducible expression is lost. If the element acted as a pause site affecting elongation, we would have expected unrestricted Pol2 elongation in both conditions and therefore greater expression levels in the KO in either or both basal and treated conditions. However, we only observed reduced expression of transcripts in all conditions in the KO, which is most consistent with activity as an enhancer.

We have uploaded these alignments and accompanying Stringtie annotation files to an updated UCSC Genome Browser session (https://genome.ucsc.edu/s/coke6162/B2_SINE_enhancers_Horton_et_al). These analyses did not provide strong evidence supporting splicing changes driven by the KO, although a major caveat is that the isoforms were analyzed using short-read sequencing and therefore may not be accurately assembled or quantified, and high-coverage long-read sequencing may be required for more accurate isoform quantification.

In addition to analyzing our RNA-seq data as suggested by the reviewers, we also looked for evidence of ChIP-seq binding at the B2 element by pausing associated factors including NELF and CDK9 using published datasets atalogued by the mouse Cistrome ChIP-seq binding site database. We did not find evidence for binding by any canonical pausing factors at either the B2 element or the Dicer1 promoter.

B2 element: http://dbtoolkit.cistrome.org/?specie=mm10&factor=tf&interval=chr12%3A104742467-104742646#plot_result

Dicer1 promoter: http://dbtoolkit.cistrome.org/?specie=mm10&factor=tf&interval=chr12%3A104%2C751%2C526-104%2C751%2 C993#plot_result

We also downloaded signal coverage tracks from a recently published NELF KO PRO-seq and NELF/CDK9 ChIP-seq dataset from mouse macrophages either untreated or stimulated with LPS from (Yu et al., 2020) and inspected the B2_Mm2.Dicer1 region (see Author response image 6). We did not find any evidence for activity over the B2 element in this dataset in either untreated or treated conditions, with the caveat that the study used LPS treatment instead of IFNΓ.

Author response image 6
Download asset Open asset
IGV Snapshot of the Dicer1 locus in mouse, with PRO-Seq of WT and NELF KO (pause-release deficient) mouse bone-marrow derived macrophages from (Yu et al., 2020).

Overall, these analyses do not provide strong support for the element acting as a pause site, but do not completely rule out the possibility. For example, it is possible that the element serves as a pause site in vitro, and is only released by NELF and other factors specifically under IFNΓ conditions, driving inducible elongation of Dicer1 in WT cells. However, the fact that KO of the element reduced basal transcript levels of Dicer1 more strongly supports enhancer activity. Direct ChIP-weq investigation of NELF/P-Tefb (CDK9/Cyclin T1) in our IFNΓ-treated WT and KO cells, as well as use of nascent transcription (PRO-seq), are necessary to directly investigate Pol2 pausing. However, given that did we did not observe any clear differences at the transcript level, we feel that our findings are most consistent with the element acting as a typical inducible cis-regulatory enhancer, and investigation of pausing would be outside the scope of the study.

We have added our analysis of isoform-level changes to the main text as follows:

“Given the B2_Mm2.Dicer1 element is bound by POLR2A in WT cells, we examined whether the B2_Mm2.Dicer1 element may alter transcription by affecting usage of different splice sites or polyadenylation sites, which would be consistent with pause site activity. We examined transcript isoform-level expression changes in KO cells in both untreated and treated conditions and found multiple transcripts that showed the same trend as the gene-level analysis, where most expressed transcripts are downregulated in both basal and IFNΓ-treated conditions and show lack of inducibility in KO cells (Supplementary file 9). These findings are consistent with the element acting primarily as an IFNΓ-inducible enhancer without any major effect on alternative splicing.”

5) Figure 4F – The authors claim that "deletion of B2_Mm2.Dicer1 also has a significant repressive effect on the IFNΓ-inducible expression of Serpina genes." However, the basal levels of Serpina3f/Serpina3g are significantly reduced upon this deletion compared to WT. Furthermore, expression of Serpina genes in the KO cell lines significantly increase upon IFNΓ stimulation, suggesting that they still show inducible expression despite the B2_Mm2.Dicer1 deletion. The authors should compare the magnitude of induction before and after stimulation between the WT and KO cell lines to determine if there is indeed a repressive effect on inducible expression of Serpina genes.

We thank the reviewers for encouraging us to further validate the regulatory effect on the Serpina genes. In our original study, the WT and KO cell lines were treated and sequenced in different batches, which confounded our comparison. As we describe in our answer to reviewer comment (3), we have re-generated and re-analyzed RNA-seq data from multiple WT and KO clonal lines in both untreated/treated conditions, with all data generated in the same experimental batch. This reanalysis confirmed the specificity of the enhancer's inducible regulation of Dicer1, but revealed no consistent effect on Serpina3f/a3g. This suggests that our previous finding of SerpinA3g was an artifact likely due to clonal variation and/or batch effect, as our reanalysis using new control clones that were isolated by limited dilution no longer showed a consistent effect on the Serpina genes. Therefore we have removed our findings on the enhancer's long-range regulation of Serpina3g and focus on the effect of enhancer activity on Dicer1. We do not believe this alters our central claims, which is that B2_Mm2 elements are an abundant source of IFNΓ-inducible enhancers, which typically regulate proximal genes within 50 kb of the gene.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The reviewers appreciated your many additional experiments/analyses as well as the edits to the text. The manuscript has been improved substantially but there are some remaining issues that need to be addressed.

1. All three reviewers felt that the response to Critique#1 was insufficient. There are several places in the manuscript referring to the significance of the Mouse B2 SINE elements contributing to lineage-specific immune response. For example, the introduction highlights that "a key open question is whether the co-option of TEs as immune regulatory elements is evolutionarily widespread as a mechanism driving divergence of innate immune responses."

The reviewers (or at least two of them) did not expect to see additional KO experiments but did expect the authors to highlight a handful of other examples where STAT1 binding to a proximal B2 SINE element is found in a mouse-specific ISG only. If no other such examples are available to highlight, then it's currently difficult to discern how generalizable this relationship is between rodent-specific B2 SINE elements and mouse-specific ISGs.

We thank the reviewers for the suggestion to highlight examples of species-specific mouse ISGs predicted to be regulated by B2_Mm2 elements. We identified 393 B2_Mm2 elements predicted regulate mouse-specific ISGs based on chromatin state and interaction (ABC model (Fulco et al., 2019)). We have added the information on these candidate elements and their target gene as a new sheet in Supplementary File 7. We also highlighted 5 additional examples in new Figure 3 supplements (Figure 3 —figure supplement 1, Figure Supplement 2), based on high ABC predicted interaction scores and published associations with immunity. We also conducted GO enrichment on these 393 genes and identified significant enrichment of terms related to immune responses, which we have also reported in a new sheet in Supplementary File 7. We have added the following paragraph to the main text:

“The 393 mouse-specific ISGs predicted to be regulated by B2_Mm2 elements were significantly enriched for multiple immune related functions (GO:0002376, adjusted p-value = 6.023×10-9; Supplementary file 7). We identified multiple examples of predicted B2_Mm2 target genes with established immune functions that showed mouse-specific IFNΓ-inducible expression (Figure 3 Figure Supplement 1, Figure Supplement 2), including dicer 1 ribonuclease III (Dicer1) (Poirier et al., 2021) (Figure 3A), SET domain containing 6, protein lysine methyltransferase (SETD6) (Levy et al., 2011), DOT1-like histone lysine methyltransferase (Kealy et al., 2020), fumarate hydratase 1 (Fh1) (Zecchini et al., 2023), heat shock protein family A (Hsp70) member 1B (Hspa1b) (Jolesch et al., 2012), NFKB inhibitor δ (Nfkbid) (Souza et al., 2021).”

Furthermore, the authors highlight a limitation of the current human cell dataset – specifically the limitation of sampling gene expression only 24 hours after interferon treatment (new paragraph starting line 310) in human cells. However, in line 232, the same dataset is used to support the earlier statement that "the human ortholog DICER1 does not show IFNΓ-inducible expression in human primary macrophages (Qiao et al. 2016)…mouse DICER1 shows a significant 50% upregulation in response to IFNΓ in primary mouse BMDMs…."

This earlier statement is substantially weakened by the final Results paragraph. This issue needs to be resolved to support the major claim of the paper that a lineage-specific TE is responsible for a lineage-specific immune response, echoed in the abstract: "B2 elements…exemplifies how lineage-specific TEs can facilitate evolutionary divergence of innate immune regulatory networks."

During the first revision, we found that the human DICER1 locus harbors STAT1 binding sites, despite not showing IFNΓ-inducible expression in primary monocytes (Qiao et al. 2013, 2016). While it is possible the STAT1 binding sites are nonfunctional, this prompted us to analyze RNA-seq data of IFNΓ-stimulated monocytes from another group (McCann et al. 2022) and saw that DICER1 did show IFNΓ-inducible expression in this other dataset. Thus, the evidence for inducible expression of human DICER1 was inconsistent but we felt it was important to include this additional analysis. There are several factors that could drive this observation, including the fact that these studies were conducted by different groups from cells from different donors. We recognize that this observation weakens our original claim stating that B2 elements drive lineage-specific expression and "evolutionary divergence of innate immune regulatory networks." However, our finding is consistent with conclusions from other groups (Choudhary et al., 2020; Sundaram et al., 2014) that lineage-specific TEs can contribute to regulatory turnover of transcription factor binding sites and/or convergent evolution.

We emphasize that B2 elements are still driving lineage-specific evolution of the cis-regulatory architecture underlying the IFNΓ regulatory network. But, at the gene expression level, different B2 elements may drive lineage-specific divergence or maintenance/turnover (replacement of existing sites) (Dermitzakis and Clark, 2002; Moses et al., 2006) or convergent regulatory evolution, which are important mechanisms of regulatory evolution (Long et al., 2016). Distinguishing the last two possibilities would require examining Dicer1 inducibility by IFNΓ in monocytes from other species, which would be interesting but outside the scope of this study. Furthermore, given that we identified 393 other B2 elements predicted to act as species-specific enhancers (based on our comparison of the Piccolo et al. and Qiao et al. 2016 datasets), we expect that a substantial fraction are driving lineage-specific divergence of inducible gene expression.

We have rewritten the final results paragraph to clarify that this idea:

“Considering that the B2_Mm2 enhancer is specific to rodents, we examined the regulatory landscape of the human DICER1 locus. Our analysis of RNA-seq data from human primary monocytes treated with IFNΓ for 24 hrs (Qiao et al., 2016) indicated that human DICER1 expression is not induced by IFNΓ (Supplementary file 1, Figure 3 —figure supplement 3). However, ChIP-seq data from IFN-treated monocytes from the same group (Qiao et al., 2013) showed multiple inducible STAT1 binding sites within the human DICER1 locus, including one originating from a primate-specific TE (LTR27) (Figure 3—figure supplement 3). Although these binding sites do not correlate with inducible DICER1 expression in the matched RNA-seq dataset, they suggest human DICER1 may be inducible under different conditions. An analysis of an independent dataset generated from another donor (McCann et al., 2022) supported the inducible expression of DICER1 (log2FC = 0.91 & FDR adjusted p-value = 3.0×10-4) (Supplementary file 1). Thus, while the evidence for inducible human DICER1 expression is inconsistent, our analyses indicate that human DICER1 has independently evolved primate-specific binding STAT1 binding sites, which may also confer inducible regulation.”

We added the following paragraph to the discussion:

“We identified hundreds of B2_Mm2-derived enhancers that are predicted to regulate genes displaying IFNΓ-inducible expression in mouse cells but not human cells, which supports their role in facilitating lineage-specific evolution of the IFNΓ-inducible regulatory network. However, we also identified a subset of target genes that show inducible expression in both species, suggesting independent evolution or turnover of regulatory elements that could serve similar regulatory functions as the B2_Mm2 enhancers. For instance, we identified an intronic STAT1 binding site derived from a primate-specific TE in the human DICER1 locus. While we did not uncover consistent evidence supporting IFNΓ-inducible regulation of human DICER1, these observations align with the concept of convergent regulatory evolution, in which similar expression patterns are perpetuated by the co-option of lineage-specific TEs (Choudhary et al., 2020; Sundaram et al., 2014). Therefore, as B2_Mm2 elements shaped the evolution of rodent immune regulatory networks, individual co-option events may have mediated either divergence or preservation of gene expression patterns.”

We have also edited the last sentence of the abstract as follows to be consistent with our Dicer1 findings:

“Regulatory networks underlying innate immunity continually face selective pressures to adapt to new and evolving pathogens. Transposable elements (TEs) can affect immune gene expression as a source of inducible regulatory elements, but the significance of these elements in facilitating evolutionary diversification of innate immunity remains largely unexplored. Here, we investigated the mouse epigenomic response to type II interferon (IFN) signaling and discovered that elements from a subfamily of B2 SINE (B2_Mm2) contain STAT1 binding sites and function as IFN-inducible enhancers. CRISPR deletion experiments in mouse cells demonstrated that a B2_Mm2 element has been co-opted as an enhancer driving IFN-inducible expression of Dicer1. The rodent-specific B2 SINE family is highly abundant in the mouse genome and elements have been previously characterized to exhibit promoter, insulator, and non-coding RNA activity. Our work establishes a new role for B2 elements as inducible enhancer elements that influence mouse immunity, and exemplifies how lineage-specific TEs can facilitate evolutionary turnover and divergence of innate immune regulatory networks.”

2. In the response to reviewer comments under critique#2, it was stated that "…the other differentially expressed genes in the KOs most likely represent off-target/stochastic changes, that are commonly seen across separate clonal isolations." And as communicated in the manuscript: Line 288: "The other dysregulated genes showed no discernable physical or functional pattern and also showed high variability between individual clones…consistent with intrinsic clonal transcriptional variation…" Whether this is true for all 101 genes should be clarified. It seems surprising given that at least some genes should be indirect targets of Dicer1. Finally, does Dicer1 show less variability than these other genes? If so, that should be stated.

We agree with the reviewers that some of these genes are potentially indirect targets of Dicer1 which would be interesting. Given that the role of Dicer1 is mRNA degradation, we expect any indirect targets to increase in expression after Dicer1 knockout, and 35/101 of the genes were. We have included these numbers in the revision. However, we emphasize that CRISPR KO experiments often result in large unintended transcriptional changes and we cannot exclude the possibility that these changes are due to clonal variation or off-target effects.

We have added the following paragraph to the results

“Genome-wide, there were 101 genes that showed greater significance than Dicer1 when testing for differential expression between IFNΓ-treated KO and WT cells (out of 3567 genes with FDR-adjusted p-value < 0.05; Supplementary file 8). Out of these 101 genes, 24 showed higher variability than Dicer1 (based on DESeq2 log2foldchange standard error) between individual clones (Figure 3 – figure supplement 5B-E), consistent with intrinsic clonal transcriptional variation revealed by the limiting dilution and/or CRISPR editing process (Nahmad et al., 2022; Westermann et al., 2022). 35 of these genes showed upregulation in the KO cells, suggesting that they could be silencing targets of Dicer1 that become upregulated upon Dicer1 downregulation. However, given the relatively modest effect of Dicer1 especially in the untreated condition, further experiments would be necessary to establish these genes as targets of Dicer1.”

3. Finally, one reviewer was concerned that the intronic location of B2 SINE means that the authors cannot delineate whether the element acts as an enhancer or instead a regulator of transcriptional elongation.

We agree that we cannot exclude the possibility that the element acts as a regulator of transcriptional elongation due to its position in the intron. We have added this caveat in our results at the end of the paragraph describing this analysis:

“However, further experiments such as CDK9 inhibition and profiling of nascent transcription in stimulated conditions (Gressel et al., 2017; Laitem et al., 2015) are necessary to establish whether the element affects transcriptional elongation.”

References

Choudhary MN, Friedman RZ, Wang JT, Jang HS, Zhuo X, Wang T. 2020. Co-opted transposons help perpetuate conserved higher-order chromosomal structures. Genome Biol 21:16.

Dermitzakis ET, Clark AG. 2002. Evolution of transcription factor binding sites in Mammalian gene regulatory regions: conservation and turnover. Mol Biol Evol 19:1114–1121.

Fulco CP, Nasser J, Jones TR, Munson G, Bergman DT, Subramanian V, Grossman SR, Anyoha R, Doughty BR, Patwardhan TA, Nguyen TH, Kane M, Perez EM, Durand NC, Lareau CA, Stamenova EK, Aiden EL, Lander ES, Engreitz JM. 2019. Activity-by-contact model of enhancer-promoter regulation from thousands of CRISPR perturbations. Nat Genet 51:1664–1669.

Gressel S, Schwalb B, Decker TM, Qin W, Leonhardt H, Eick D, Cramer P. 2017. CDK9-dependent RNA polymerase II pausing controls transcription initiation. ELife 6. doi:10.7554/eLife.29736

Jolesch A, Elmer K, Bendz H, Issels RD, Noessner E. 2012. Hsp70, a messenger from hyperthermia for the immune system. Eur J Cell Biol 91:48–52.

Kealy L, Di Pietro A, Hailes L, Scheer S, Dalit L, Groom JR, Zaph C, Good-Jacobson KL. 2020. The Histone Methyltransferase DOT1L Is Essential for Humoral Immune Responses. Cell Rep 33:108504.

Laitem C, Zaborowska J, Isa NF, Kufs J, Dienstbier M, Murphy S. 2015. CDK9 inhibitors define elongation checkpoints at both ends of RNA polymerase II-transcribed genes. Nat Struct Mol Biol 22:396–403.

Levy D, Kuo AJ, Chang Y, Schaefer U, Kitson C, Cheung P, Espejo A, Zee BM, Liu CL, Tangsombatvisit S, Tennen RI, Kuo AY, Tanjing S, Cheung R, Chua KF, Utz PJ, Shi X, Prinjha RK, Lee K, Garcia BA, Bedford MT, Tarakhovsky A, Cheng X, Gozani O. 2011. Lysine methylation of the NF-κB subunit RelA by SETD6 couples activity of the histone methyltransferase GLP at chromatin to tonic repression of NF-κB signaling. Nat Immunol 12:29–36.

Long HK, Prescott SL, Wysocka J. 2016. Ever-Changing Landscapes: Transcriptional Enhancers in Development and Evolution. Cell 167:1170–1187.

McCann KJ, Christensen SM, Colby DH, McGuire PJ, Myles IA, Zerbe CS, Dalgard CL, Sukumar G, Leonard WJ, McCormick BA, Holland SM. 2022. IFNγ regulates NAD+ metabolism to promote the respiratory burst in human monocytes. Blood Adv 6:3821–3834.

Moses AM, Pollard DA, Nix DA, Iyer VN, Li X-Y, Biggin MD, Eisen MB. 2006. Large-scale turnover of functional transcription factor binding sites in Drosophila. PLoS Comput Biol 2:e130.

Nahmad AD, Reuveni E, Goldschmidt E, Tenne T, Liberman M, Horovitz-Fried M, Khosravi R, Kobo H, Reinstein E, Madi A, Ben-David U, Barzel A. 2022. Frequent aneuploidy in primary human T cells after CRISPR-Cas9 cleavage. Nat Biotechnol 40:1807–1813.

Poirier EZ, Buck MD, Chakravarty P, Carvalho J, Frederico B, Cardoso A, Healy L, Ulferts R, Beale R, Reis e Sousa C. 2021. An isoform of Dicer protects mammalian stem cells against multiple RNA viruses. Science 373:231–236.

Qiao Y, Giannopoulou EG, Chan CH, Park S-H, Gong S, Chen J, Hu X, Elemento O, Ivashkiv LB. 2013. Synergistic activation of inflammatory cytokine genes by interferon-γ-induced chromatin remodeling and toll-like receptor signaling. Immunity 39:454–469.

Qiao Y, Kang K, Giannopoulou E, Fang C, Ivashkiv LB. 2016. IFN-γ Induces Histone 3 Lysine 27 Trimethylation in a Small Subset of Promoters to Stably Silence Gene Expression in Human Macrophages. Cell Rep 16:3121–3129.

Souza SP, Splitt SD, Sànchez-Arcila JC, Alvarez JA, Wilson JN, Wizzard S, Luo Z, Baumgarth N, Jensen KDC. 2021. Genetic mapping reveals Nfkbid as a central regulator of humoral immunity to Toxoplasma gondii. PLoS Pathog 17:e1010081.

Sundaram V, Cheng Y, Ma Z, Li D, Xing X, Edge P, Snyder MP, Wang T. 2014. Widespread contribution of transposable elements to the innovation of gene regulatory networks. Genome Res 24:1963–1976.

Westermann L, Li Y, Göcmen B, Niedermoser M, Rhein K, Jahn J, Cascante I, Schöler F, Moser N, Neubauer B, Hofherr A, Behrens YL, Göhring G, Köttgen A, Köttgen M, Busch T. 2022. Wildtype heterogeneity contributes to clonal variability in genome edited cells. Sci Rep 12:18211.

Zecchini V, Paupe V, Herranz-Montoya I, Janssen J, Wortel IMN, Morris JL, Ferguson A, Chowdury SR, Segarra-Mondejar M, Costa ASH, Pereira GC, Tronci L, Young T, Nikitopoulou E, Yang M, Bihary D, Caicci

F, Nagashima S, Speed A, Bokea K, Baig Z, Samarajiwa S, Tran M, Mitchell T, Johnson M, Prudent J, Frezza C. 2023. Fumarate induces vesicular release of mtDNA to drive innate immunity. Nature 615:499–506.

https://doi.org/10.7554/eLife.82617.sa2

Article and author information

Author details

  1. Isabella Horton

    Department of Molecular, Cellular, and Developmental Biology and BioFrontiers Institute, University of Colorado Boulder, Boulder, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing - original draft
    Contributed equally with
    Conor J Kelly
    Competing interests
    No competing interests declared

  2. Conor J Kelly

    Department of Molecular, Cellular, and Developmental Biology and BioFrontiers Institute, University of Colorado Boulder, Boulder, United States
    Contribution
    Conceptualization, Investigation, Visualization, Methodology, Writing - original draft
    Contributed equally with
    Isabella Horton
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2410-7892

  3. Adam Dziulko

    Department of Molecular, Cellular, and Developmental Biology and BioFrontiers Institute, University of Colorado Boulder, Boulder, United States
    Contribution
    Methodology
    Competing interests
    No competing interests declared

  4. David M Simpson

    Department of Molecular, Cellular, and Developmental Biology and BioFrontiers Institute, University of Colorado Boulder, Boulder, United States
    Contribution
    Investigation, Methodology, Project administration
    Competing interests
    No competing interests declared

  5. Edward B Chuong

    Department of Molecular, Cellular, and Developmental Biology and BioFrontiers Institute, University of Colorado Boulder, Boulder, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Project administration, Writing - review and editing
    For correspondence
    edward.chuong@colorado.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5392-937X

Funding

National Institutes of Health (1R35GM128822)

  • Edward B Chuong

David and Lucile Packard Foundation

  • Edward B Chuong

Boettcher Foundation

  • Isabella Horton
  • Edward B Chuong

Alfred P. Sloan Foundation

  • Edward B Chuong

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the University of Colorado Genomics Shared Resource and BioFrontiers Computing core for technical support during this study. Funding EBC was supported by the National Institutes of Health (1R35GM128822), the Alfred P Sloan Foundation, the David and Lucile Packard Foundation, and the Boettcher foundation. IH was supported by the Boettcher foundation.

Senior Editor

  1. Molly Przeworski, Columbia University, United States

Reviewing Editor

  1. Mia T Levine, University of Pennsylvania, United States

Reviewer

  1. Kenji Ichiyanagi, Nagoya University, Japan

Publication history

  1. Preprint posted: August 4, 2022 (view preprint)
  2. Received: August 11, 2022
  3. Accepted: May 8, 2023
  4. Accepted Manuscript published: May 9, 2023 (version 1)
  5. Version of Record published: May 30, 2023 (version 2)

Copyright

© 2023, Horton, Kelly et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 294
    Page views
  • 83
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Isabella Horton
  2. Conor J Kelly
  3. Adam Dziulko
  4. David M Simpson
  5. Edward B Chuong
(2023)
Mouse B2 SINE elements function as IFN-inducible enhancers
eLife 12:e82617.
https://doi.org/10.7554/eLife.82617

Categories and tags

Research organism

Further reading

    1. Biochemistry and Chemical Biology
    2. Chromosomes and Gene Expression

    An acetylation-mediated chromatin switch governs H3K4 methylation read-write capability

    Kanishk Jain, Matthew R Marunde ... Brian D Strahl
    Short Report Updated

    In nucleosomes, histone N-terminal tails exist in dynamic equilibrium between free/accessible and collapsed/DNA-bound states. The latter state is expected to impact histone N-termini availability to the epigenetic machinery. Notably, H3 tail acetylation (e.g. K9ac, K14ac, K18ac) is linked to increased H3K4me3 engagement by the BPTF PHD finger, but it is unknown if this mechanism has a broader extension. Here, we show that H3 tail acetylation promotes nucleosomal accessibility to other H3K4 methyl readers, and importantly, extends to H3K4 writers, notably methyltransferase MLL1. This regulation is not observed on peptide substrates yet occurs on the cis H3 tail, as determined with fully-defined heterotypic nucleosomes. In vivo, H3 tail acetylation is directly and dynamically coupled with cis H3K4 methylation levels. Together, these observations reveal an acetylation ‘chromatin switch’ on the H3 tail that modulates read-write accessibility in nucleosomes and resolves the long-standing question of why H3K4me3 levels are coupled with H3 acetylation.

    1. Chromosomes and Gene Expression

    Hypermetabolism in mice carrying a near-complete human chromosome 21

    Dylan C Sarver, Cheng Xu ... G William Wong
    Research Article

    The consequences of aneuploidy have traditionally been studied in cell and animal models in which the extrachromosomal DNA is from the same species. Here, we explore a fundamental question concerning the impact of aneuploidy on systemic metabolism using a non-mosaic transchromosomic mouse model (TcMAC21) carrying a near-complete human chromosome 21. Independent of diets and housing temperatures, TcMAC21 mice consume more calories, are hyperactive and hypermetabolic, remain consistently lean and profoundly insulin sensitive, and have a higher body temperature. The hypermetabolism and elevated thermogenesis are likely due to a combination of increased activity level and sarcolipin overexpression in the skeletal muscle, resulting in futile sarco(endo)plasmic reticulum Ca2+ ATPase (SERCA) activity and energy dissipation. Mitochondrial respiration is also markedly increased in skeletal muscle to meet the high ATP demand created by the futile cycle and hyperactivity. This serendipitous discovery provides proof-of-concept that sarcolipin-mediated thermogenesis via uncoupling of the SERCA pump can be harnessed to promote energy expenditure and metabolic health.

    1. Chromosomes and Gene Expression

    Interplay between PML NBs and HIRA for H3.3 dynamics following type I interferon stimulus

    Constance Kleijwegt, Florent Bressac ... Armelle Corpet
    Research Article

    Promyelocytic leukemia Nuclear Bodies (PML NBs) are nuclear membrane-less organelles physically associated with chromatin underscoring their crucial role in genome function. The H3.3 histone chaperone complex HIRA accumulates in PML NBs upon senescence, viral infection or IFN-I treatment in primary cells. Yet, the molecular mechanisms of this partitioning and its function in regulating histone dynamics have remained elusive. By using specific approaches, we identify intermolecular SUMO-SIM interactions as an essential mechanism for HIRA recruitment in PML NBs. Hence, we describe a role of PML NBs as nuclear depot centers to regulate HIRA distribution in the nucleus, dependent both on SP100 and DAXX/H3.3 levels. Upon IFN-I stimulation, PML is required for interferon-stimulated genes (ISGs) transcription and PML NBs become juxtaposed to ISGs loci at late time points of IFN-I treatment. HIRA and PML are necessary for the prolonged H3.3 deposition at the transcriptional end sites of ISGs, well beyond the peak of transcription. Though, HIRA accumulation in PML NBs is dispensable for H3.3 deposition on ISGs. We thus uncover a dual function for PML/PML NBs, as buffering centers modulating the nuclear distribution of HIRA, and as chromosomal hubs regulating ISGs transcription and thus HIRA-mediated H3.3 deposition at ISGs upon inflammatory response.