1. Genetics and Genomics
  2. Immunology and Inflammation

Inflammation drives alternative first exon usage to regulate immune genes including a novel iron-regulated isoform of Aim2

  1. Elektra K Robinson
  2. Pratibha Jagannatha
  3. Sergio Covarrubias
  4. Matthew Cattle
  5. Valeriya Smaliy
  6. Rojin Safavi
  7. Barbara Shapleigh
  8. Robin Abu-Shumays
  9. Miten Jain
  10. Suzanne M Cloonan
  11. Mark Akeson
  12. Angela N Brooks  Is a corresponding author
  13. Susan Carpenter  Is a corresponding author
  1. Department of Molecular, Cell and Developmental Biology, University of California Santa Cruz, United States
  2. Department of Biomolecular Engineering, University of California Santa Cruz, United States
  3. Division of Pulmonary and Critical Care Medicine, Joan and Sanford I. Weill Department of Medicine, Weill Cornell Medicine, United States
https://doi.org/10.7554/eLife.69431
Share this article
Cite this article
  1. Elektra K Robinson
  2. Pratibha Jagannatha
  3. Sergio Covarrubias
  4. Matthew Cattle
  5. Valeriya Smaliy
  6. Rojin Safavi
  7. Barbara Shapleigh
  8. Robin Abu-Shumays
  9. Miten Jain
  10. Suzanne M Cloonan
  11. Mark Akeson
  12. Angela N Brooks
  13. Susan Carpenter
(2021)
Inflammation drives alternative first exon usage to regulate immune genes including a novel iron-regulated isoform of Aim2
eLife 10:e69431.
https://doi.org/10.7554/eLife.69431
  2. Download BibTeX
  3. Download .RIS

Abstract

Determining the layers of gene regulation within the innate immune response is critical to our understanding of the cellular responses to infection and dysregulation in disease. We identified a conserved mechanism of gene regulation in human and mouse via changes in alternative first exon (AFE) usage following inflammation, resulting in changes to the isoforms produced. Of these AFE events, we identified 95 unannotated transcription start sites in mice using a de novo transcriptome generated by long-read native RNA-sequencing, one of which is in the cytosolic receptor for dsDNA and known inflammatory inducible gene, Aim2. We show that this unannotated AFE isoform of Aim2 is the predominant isoform expressed during inflammation and contains an iron-responsive element in its 5′UTR enabling mRNA translation to be regulated by iron levels. This work highlights the importance of examining alternative isoform changes and translational regulation in the innate immune response and uncovers novel regulatory mechanisms of Aim2.

Introduction

Macrophages are critical cells in the innate immune system that combat infection by initiating acute inflammatory responses. Acute inflammation is tightly coordinated and begins with the detection of pathogen-associated molecular patterns (PAMPs) by pattern recognition receptors (PRRs), which include toll-ike receptors (TLRs) (Medzhitov and Janeway, 1998; Pai et al., 2016). These initial steps are followed by the activation of sequestered transcription factors (TFs), such as nuclear factor of kappa-B (NF-κB) and interferon regulatory factors (IRFs), which orchestrate pro-inflammatory and antiviral response signals involved in pathogen clearance (Pai et al., 2016). Once pathogens are cleared, macrophages express genes involved in the resolution of inflammation to return the host to homeostasis (Hamidzadeh et al., 2017). Dysregulation of these pro-inflammatory pathways can have devastating consequences, leading to unresolved inflammation and chronic inflammatory diseases (Zhou et al., 2016).

Recently, the process of alternative splicing has emerged as another key mechanism by which the immune system is regulated. Alternative splicing is a regulated process enabling a single gene to produce many isoforms, thus increasing the complexity of gene function and the proteome (Boudreault et al., 2016; Ivanov and Anderson, 2013; Pai et al., 2016; Wang et al., 2015). Much of this occurs in a cell-type-specific and signal-induced manner (Ergun et al., 2013; Wells et al., 2006). Previous studies have shown that mouse and human macrophages exposed to inflammatory stimuli undergo alternative splicing (Beyer et al., 2012; Bhatt et al., 2012; de Bruin et al., 2016; Haque et al., 2018; Janssen et al., 2020; Lin et al., 2016; Liu et al., 2018; O’Connor et al., 2015; Pai et al., 2016; Pandya-Jones et al., 2013). Alternative splicing within the immune system can affect the type and magnitude of the inflammatory response, such as the production of a soluble form of TLR4 that is expressed upon lipopolysaccharide (LPS), which leads to inhibition of TNFα and NF-κB serving as a negative feedback mechanism (Lynch, 2004; Schaub and Glasmacher, 2017). Additionally, this mechanism has been characterized within signaling molecules (Blumhagen et al., 2017; Shakola et al., 2015), including TBK1 (Deng et al., 2008) and MyD88 (De Arras and Alper, 2013), that produce the alternative RNA splice forms, TBK1s and MyD88s, respectively, which function to limit the extent of the pro-inflammatory response. Alternative splicing can also result in the production of inflammatory signaling molecules, such as TRIF (Han et al., 2010) and the proteins in the NFκB family (Wells et al., 2006) with altered activity or stability. Beyond changing the ORF of an mRNA molecule, elongating or shortening the first or last exon can impact post-transcriptional gene regulation and are important to consider when elucidating the regulatory mechanisms of immune genes (Carpenter et al., 2014; Ghiasvand et al., 2014; Leppek et al., 2018), specifically underlying motifs in 5′ untranslated regions (UTRs) (Kramer et al., 2013; Resch et al., 2009) and 3′UTRs (Mariella et al., 2019; Mayr, 2016).

While inflammation-induced alternative splicing in both human and mouse macrophages has been investigated on a genome-wide scale (Beyer et al., 2012; Bhatt et al., 2012; de Bruin et al., 2016; Haque et al., 2018; Janssen et al., 2020; Lin et al., 2016; Liu et al., 2018; O’Connor et al., 2015; Pai et al., 2016; Pandya-Jones et al., 2013), to our knowledge, long-read RNA-sequencing has not been utilized to generate a de novo transcriptome for primary murine macrophages. Such an approach is necessary to fully appreciate the extent of alternative transcript isoform usage as we know most transcriptome annotations are incomplete (Workman et al., 2019) and isoforms generated are cell-type and treatment specific (Sapkota et al., 2019; Weirather et al., 2017; Workman et al., 2019; Wu et al., 2018).

Here we used both long- and short-read RNA-sequencing to uncover novel isoforms and classes of alternative splicing events following inflammation in human and murine macrophages. Interestingly the dominant conserved class of alternative isoform usage observed following inflammation is alternative first exon (AFE) usage, which involves alternative transcription start sites (TSS) usage coupled with alternative splicing. AFE events can have multilevel effects on protein diversity, regulating genes through alterations of the 5′UTR region, and directing the locality of proteins through alternative N-termini (Landry et al., 2003). We identified 95 unannotated AFE events in mice from native RNA-sequencing, one of which is in the cytosolic receptor for dsDNA and known inflammatory inducible gene, Aim2. We show that this unannotated AFE isoform of Aim2 is the predominant isoform produced during inflammation and contains an iron-responsive element (IRE) in its 5′UTR, enabling mRNA translation to be controlled by iron levels. This work reveals that alternative transcript isoform usage plays a crucial role in shaping the transient nature of the inflammatory response. Isoform expression is an additional layer of regulation within the immune response and therefore a possible contributing factor to the development of auto-immune and inflammatory diseases. Understanding the exact isoforms of genes that are expressed during an inflammatory response will enable us to design better targets for therapeutic intervention of these diseases.

Results

Global profiling of the cellular alternative splicing landscape in human and mouse macrophages post-inflammation

To identify alternative splicing events following inflammation, we performed whole transcriptome analysis on human monocyte-derived macrophages (MDMs) and murine bone marrow-derived macrophages (BMDMs) with and without LPS treatment (Figure 1A). We found that ~50% of splicing changes (corrected p-value ≤ 0.25 and |ΔPSI| ≥ 10) were classified as AFE events following LPS activation in both human and murine macrophages (Figure 1B, Figure 1—figure supplement 1). A ranking analysis of the significant events, from both mouse and human data sets, revealed that AFE events consistently comprised a large proportion of the top splicing changes (Figure 1—figure supplements 2 and 3). Additionally, analysis of previously published primary human macrophages stimulated with either Listeria or Salmonella, using our bioinformatic pipeline and with more stringent thresholds than the previous study, also revealed AFE events to be amongst the most significant prevalent alternative splicing events (corrected p-value ≤ 0.05 or 0.25 and |ΔPSI| ≥ 10) (Figure 1—figure supplement 4; Supplementary files 4-5–; Pai et al., 2016).

Figure 1 with 9 supplements see all
Download asset Open asset
Global profiling of the cellular alternative splicing landscape in human and mouse macrophages post-inflammatory.

(A) Diagram of RNA-seq library generation. (B) Categorization of significant splicing events in human and mouse macrophage. (C) Categorization of significant splicing events found in mouse bone marrow-derived macrophages (BMDM) ±6 hr lipopolysaccharide by using either the Gencode annotation or the GENCODE+ de novo annotation. (D) Venn diagram representing unique and common genes with alternative first exon (AFE) events found in RNA-seq of primary BMDMs post-inflammatory stimulation using the Gencode annotation or the GENCODE+ de novo annotation. Volcano plots of all differentially expressed genes from RNA-seq of either human (E) or mouse (F) macrophages. Genes highlighted in red undergo significant AFE changes following inflammation. Schematic of AFE inclusion and exclusion isoforms, followed by RT-PCR gel results and percent spliced in calculation for Argehf7 (G) Denr (H), and Aim2 (I), was performed in biological triplicates, p-value assessed using Student’s t-test.

We next identified 11 conserved AFE splicing events between human and mouse (Figure 1—figure supplement 5A,B), the largest number of conserved event types amongst all alternative splicing event types (Figure 1—figure supplement 5). We validated the AFE changes upon stimulation on the already characterized Ncoa7 (Figure 1—figure supplement 6A–C; Singer et al., 2008; Yu et al., 2015) and Rcan1 (Figure 1—figure supplement 6D–F; Pang et al., 2018), as well as a previously uncharacterized inflammatory-specific isoform of Ampd3 (Figure 1—figure supplement 6G–I), in human and mouse primary macrophages using RT-PCR. Taken together, these results show the high prevalence and conservation of AFE usage following inflammatory activation.

A caveat to our analysis thus far was the reliance on annotated transcriptome assemblies to identify first exons of genes (Brooks et al., 2015). In order to determine if there are additional splicing events that are not captured using the publicly available GENCODE M18 transcriptome annotation (Garalde et al., 2018; Hsieh et al., 2019; Pollard et al., 2018; Workman et al., 2019), we performed native RNA-sequencing of murine macrophages with and without LPS treatment to build a de novo murine macrophage specific transcriptome with an average read depth of 1 million. We identified isoforms using full-length alternative isoform analysis of RNA (FLAIR) (Tang et al., 2020; Workman et al., 2019) that also had promoter support identified from accessible chromatin (ATAC-seq) (Atianand et al., 2016; Tong et al., 2016). The FLAIR isoforms were then merged with the GENCODE M18 assembly (mm10) (Frankish et al., 2019) as a transcript reference to identify and quantify alternative splicing events using short-read sequencing, which has increased read depth, denoting this transcriptome as ‘GENCODE+ de novo’ (Figure 1—figure supplement 7A). Overall, the incorporation of long-read sequencing to generate a novel transcriptome led to the identification of 95 novel and statistically significant AFE events that occur following inflammation (Figure 1C,D). A comparison of significant AS events between splicing events identified by using GENCODE M18 annotation and de novo FLAIR transcriptome shows an overall increase in the number of unique events (Figure 1—figure supplement 7B,H). To gain confidence of the novel AFE events, we compared a comprehensive, 339 mouse sample, FANTOM CAGE-seq data (FANTOM Consortium and the RIKEN PMI and CLST (DGT), 2014) or annotated and unannotated TSSs as defined by the GENCODE M18 assembly. The caveat to this FANTOM data set is that there are no inflammatory stimulated samples. Even with this caveat, the analysis revealed the similarity of the distributions between CAGE scores associated with CAGE peaks overlapping known and those overlapping novel TSSs of FLAIR isoforms, further supporting the validity of the novel TSSs identified (Figure 1—figure supplement 8, Figure 3—figure supplement 1B). Additionally, we utilized a data set generated through application of machine learning methods to classify CAGE-seq peaks as true or false, where true corresponds to CAGE-seq peaks that overlap true TSSs (Kanamori-Katayama et al., 2011). Novel TSSs identified with FLAIR showed a 45% overlap with CAGE-seq peaks classified as true, providing us with confidence in the novel AFE events identified (Figure 1—figure supplement 8C). Together these approaches provide us with confidence in the novel AFE events identified by direct RNA-seq.

Interestingly, when identifying gene expression changes, we found that ~50% of genes with AFE usage were not differentially expressed following inflammation (Figure 1E,F; Supplementary files 6–7), highlighting the importance of studying isoform usage for control of gene expression. Among the most statistically significant novel AFE first exon events were Denr, Arhgef7, and Aim2, which we validated using RT-qPCR (Figure 1G–I, Figure 1—figure supplement 9).

Identification of an unannotated promoter for Aim2

To better understand the potential functional consequence of AFE changes, we further examined the novel first exon event upregulated upon inflammatory activation in Aim2. Aim2 is an interferon-stimulated gene (ISG), localized to the cytosol. Aim2 is a dsDNA sensor that upon recognition induces the formation of an inflammasome complex releasing IL1β and IL18 from the cell as a defense mechanism to control infection (Wang and Yin, 2017). Chromatin immunoprecipitation (ChIP)-seq for the myeloid pioneering TF PU.1 in primary BMDMs (Figure 2A,B, top track, in black) (Lam et al., 2013) supported the presence of an additional promoter upstream of the canonical isoform for Aim2 (NM_001013779.2). Predominant isoforms (≥10% of total gene expression in a sample) assembled from native RNA-sequencing with FLAIR identified the canonical isoform and five unannotated isoforms that use the inflammatory-activated promoter, revealing a new longer 5′UTR (Figure 2C). Native RNA-sequencing-based quantification provided additional support that the unannotated promoter usage is upregulated upon LPS stimulation. At steady state, approximately 20% of reads map to Aim2’s transcript with the canonical promoter and 65% of reads map to transcripts with the upstream promoter, while following inflammatory activation, 14% of the reads map to transcripts with the canonical promoter and 81% of reads map to transcripts with the upstream promoter (Figure 2D). To validate the change in Aim2 AFE usage upon LPS stimulation, RT-qPCR was performed using exon spanning primers that were either specific to the annotated AFE or the unannotated AFE, in BMDMs (Figure 2E). The expression profile of the annotated first exon is not induced by LPS stimulation (Figure 2F), while the unannotated first exon and the CDS of Aim2 are equally induced by LPS stimulation (Figure 2G,H). Therefore, these data show that it is the novel isoform of Aim2 that is inflammatory-regulated and not the canonical isoform defined in GENCODE annotation, nor isoforms from the RefSeq annotation.

Figure 2
Download asset Open asset
Identification of an unannotated promoter in Aim2.

(A, B) The top track, in black, represents ChIP-seq data for a macrophage-specific transcription factor, PU.1. Peaks represent possible promoter regions; two distinct peaks of equal height are present at the annotated transcriptional start site (TSS) for Aim2 and about 1 kb upstream of the transcriptional start site. The middle track, in orange, represents basal transcription in bone marrow-derived macrophages (BMDMs), while the bottom track, in purple, represents active transcription in BMDMs 6 hr lipopolysaccharide (LPS) post treatment. (C) Aim2 transcript isoforms identified in BMDMs by native RNA long-read sequencing through FLAIR analysis. Transcripts are categorized by promoter, denoted by gray, orange, or purple. (D) The bar chart represents data from long-read sequencing showing the abundance of each transcript isoform from BMDMs ±6 hr LPS. (E–H) qRT-PCR was performed in biological triplicate, on primary BMDM RNA extracts that had been stimulated with LPS for indicated time points.

The novel inflammatory promoter of Aim2 is regulated by IRF3 and P65

To gain insights into potential regulatory mechanisms controlling the expression of the 152 significant AFE events, we assessed changes in chromatin accessibility during inflammatory activation in BMDMs. Analysis of ATAC-Seq (Tong et al., 2016) revealed differential peaks at the promoter regions for 25% of genes with significant AFE events, suggesting that chromatin remodeling is one mechanism driving the expression of the AFE events (Figure 3A). This regulation is not what controls isoform usage for Aim2. The annotated promoter is accessible in all cells while the novel Aim2 promoter is specific to myeloid progenitors and monocytes (Lara-Astiaso et al., 2014; Tong et al., 2016; Figure 3—figure supplement 1A). In addition, the accessibility of both the annotated and unannotated promoters remains open despite the cell’s inflammatory status (Figure 3—figure supplement 1B). Therefore, the expression of the new isoform is not due to chromatin remodeling of either promoter region (Figure 3—figure supplement 1C).

Figure 3 with 2 supplements see all
Download asset Open asset
Novel inflammatory promoter of Aim2 is regulated by IRF3 and p65.

(A) UpSet plot showing the number of alternative first exon (AFE) events out of 77 total that have differential transcription factor (TF) binding and differential chromatin accessibility and all combinations of these sets. (B) Pie chart representing the gene ontology of TF motifs identified through analysis motif enrichment (AME) tool of all AFE promoter sequences. (C) Venn diagram of all motifs defined using HOMER analysis within the annotated and unannotated promoter regions. (D, E) DAVID analysis examining the gene ontology of TFs at the annotated and unannotated promoters of Aim2.

Another potential mode of regulation that can drive AFE usage is TF binding. We next analyzed ChIP-seq data of two major TFs that drive inflammation downstream of LPS, NF-κB (p65) and interferon response factor 3 (IRF3) (Tong et al., 2016). We found that p65 and IRF3 specifically account for another 25% of the AFE events, including the novel Aim2 isoform, which we confirmed using multiple ChIP-seq data sets (Figure 3A, Figure 3—figure supplement 2).

Further bioinformatic analysis of promoters associated with all AFEs, using HOMER TF motif enrichment, shows that there are 304 potential TFs that bind these promoters. By gene ontology analysis, we see that the majority of the TFs are associated with metabolism, as well as the immune system (Figure 3B; Supplementary file 10). When specifically assessing the two promoter regions driving the annotated and unannotated isoforms of Aim2, we identified 106 individual TF motifs within the annotated promoter and 121 motifs within the unannotated promoter (Supplementary file 11). Of these predicted motifs, there were 44 motifs unique to the unannotated promoter (Figure 3C). Gene ontology (Huang et al., 2007) analysis of TFs specific to the annotated and unannotated promoter regions of Aim2 confirms that the unannotated promoter is driven by inflammatory-specific TFs including NF-κB and IRFs (Figure 3—figure supplement 2L). Use of ATAC-seq and ChIP-seq for specific TFs has enabled us to determine the regulatory pathways driving 50% of the AFEs in our data. For the remaining 50%, there could be additional TFs, RNA-binding proteins (Lynch, 2004), or differential RNA stability driving their expression. Further work will be required to fully understand the complex regulation of all AFE events.

Unannotated 5′UTR of Aim2 negatively regulates translation through a single iron-responsive element

The novel inflammatory isoform identified here for Aim2 acquired a longer 5′UTR compared to the canonical isoform; however, globally there are no differences in AFE size between conditions (Figure 4A, Figure 4—figure supplement 1A,B). Previous studies have shown that longer 5′UTRs can affect the translation of a gene (Kramer et al., 2013; Senanayake and Brian, 1999). Using a GFP reporter system, the translational efficiency of the unannotated Aim2 5′UTR (767 bp) was compared to the annotated 5′UTR (489 bp) (Figure 4B). The unannotated 5′UTR showed significantly lower mean GFP fluorescence units, suggestive of lower translational efficiency, as assessed by flow cytometry 72 hr post-transient transfection in 293T cells (Figure 4C,D), while equal mCherry fluorescence was observed for all co-transfected control constructs (Figure 4E). To explore the mechanism of how the unannotated 5′UTR results in decreased translational efficiency, we used RegRNA2.0 to predict RNA regulatory motifs in the 5′UTRs (Chang et al., 2013). We identified a single IRE within the unannotated 5′UTR, while Musashi binding element (MBE) motifs were identified in both 5′UTRs (Figure 4F; Supplementary file 14). Globally, we find that there are more predicted motifs in inflammatory-regulated first exons (inclusion exons) in comparison to basal exons (excluded exons), but the IRE motif is found only in Aim2 (Figure 4—figure supplement 1C). The finding that there are structured motifs in the unannotated 5′UTR of Aim2 is also supported by the RNAfold Vienna package (Gruber et al., 2008), which predicts the hairpin structure of the IRE in the alternative 5′UTR (Figure 4—figure supplement 2). Since the IRE motif is unique to the unannotated 5′UTR of Aim2, we hypothesized that this motif is critical in regulating translational efficiency.

Figure 4 with 4 supplements see all
Download asset Open asset
Unannotated 5′ untranslated region (UTR) of Aim2 negatively regulates translation through a single iron-responsive element (IRE).

(A) Schematic of annotated and unannotated 5′UTR of most prevalent Aim2 isoforms in mouse macrophages. (B) Diagram of cloning strategy of Aim2’s 5′UTR in GFP plasmid. (C) Transfection strategy of 5′UTR-GFP plasmids co-transfected with an mCherry control plasmid at a 1–1 ratio in 293T cells. (D, E) Flow cytometry of 293T cells 72 hr post-transfection with control annotated and unannotated 5′UTR of Aim2 to measure GFP and mCherry (Ctl) protein fluorescence. (F) Using RegRNA2.0, a single IRE was found in the alternative 5′UTR, in addition to multiple Musashi binding elements (MBEs). (G) Diagram of how an IRE functions in the cytoplasm of a cell within a 5′UTR. With low or normal levels of iron, iron binding proteins (IRP1 or IRP2) bind to IRE and block translation. During high levels of iron, within a cell, IRP1 is sequestered by iron-sulfur (Fe-S) clusters and IRP2 is degraded, therefore allowing translation of the protein. (H, I) Flow cytometry of 293T cells 72 hr post-transfection of mCherry (Ctl), along with an annotated 5′UTR-GFP plasmid, unannotated 5′UTR-GFP plasmid, or a GFP plasmid containing the unannotated 5′UTR without the defined IRE. (J, K) Flow cytometry of 293T cells ± 100 µM ferric ammonium citrate (FAC) 72 hr post-transfection of mCherry (Ctl), along with an annotated 5′UTR-GFP plasmid or unannotated 5′UTR-GFP plasmid. (L) Overview of the polysome profiling protocol to analyze translation activity. (M) Cytoplasmic lysates from ±lipopolysaccharide (LPS)-treated cells were fractionated through sucrose gradients. Global RNA polysome profiles generated by the density gradient fractionation system are shown. A representative plot from stimulated primary bone marrow-derived macrophage (BMDM)-fractionated samples is shown. The experiment was performed three times. (N–Q) The relative distribution of Gapdh mRNA, encoding a housekeeping protein, Neat1, long non-coding RNA (lncRNA), annotated and unannotated Aim2 mRNA, was measured by RT-qPCR analysis of RNA. Each of the gradient fractions are calculated as relative enrichment when compared to unfractionated input mRNA; standard deviation represents technical triplicate. (R) Protein lysates of time-course LPS stimulation of 0 hr, 6 hr, 24 hr, 48 hr, and 72 hr without and with 100 µM of FAC (iron) added to immortalized WT BMDMs. Western blot performed on AIM2 and B-ACTIN. (S) Western blot quantification performed in FIJI, standard deviation represents biological triplicates, and p-value assessed using Student’s t-test.

Figure 4—source data 1

Supplemental WB Uncrop Primary BMDM +/- Iron.

https://cdn.elifesciences.org/articles/69431/elife-69431-fig4-data1-v2.pdf

When cells are at homeostasis, iron binding proteins (IRP1/2) bind to IREs located within the 5′UTR (e.g., ferritin) and can block translation, while IREs in the 3′UTR (e.g., transferrin receptor) can promote translation (Rouault, 2006; Wang et al., 2004). However, iron repletion results in the inactivation of IRP1/2 (Outten, 2017; Figure 4G). To experimentally test if the IRE motif within the unannotated 5′UTR acts as a translational repressor, we removed the element using site-directed mutagenesis, which led to an increase in GFP expression by ~20% compared to the annotated 5′UTR (Figure 4H,I). Next, we exogenously added 100 µM ferric ammonium citrate (FAC) to overload the cells with iron and determined if this can rescue the observed decrease in translational efficiency from our unannotated 5′UTR. Upon FAC administration, the relative GFP expression of the unannotated 5′UTR plasmid increased by ~50%, while mCherry control was unchanged, suggesting that the translational efficiency of the unannotated 5′UTR can be rescued with iron supplementation (Figure 4J,K). From these results, we conclude that the predicted IRE motif within the unannotated 5′UTR of Aim2 functions as an IRE to control translation.

To test if the IRE motif in the unannotated 5′UTR of Aim2 acts as a translational regulator endogenously, we performed polysome profiling followed by RT-qPCR on primary BMDMs in the presence and absence of LPS for 18 hr to determine the translational competency of the isoforms of Aim2 (Figure 4L,M). As a negative control, we examined Neat1 (Nakagawa et al., 2014), a long non-coding RNA (lncRNA) that is not detected in polysomes, nor translated, and this is not dependent on LPS (Figure 4N). The relative distribution of our positive control gene, Gapdh, which encodes a highly expressed housekeeping protein, is enriched in the high polysome fraction, with or without LPS, as expected (Figure 4O). Using isoform-specific primer sets (Figure 2E) for the annotated and unannotated Aim2 isoforms, we find that the annotated isoform is enriched in the high polysome fraction with and without LPS treatment (Figure 4P), while the novel isoform is enriched in the low polysome fraction with and without inflammatory stimulation (Figure 4Q). Additionally, from our RNA-seq data we find that IRP1 and IRP2 are not transcriptionally regulated by inflammation, but based on Weiss et al. inflammation can affect the binding affinity of IRPs in macrophages (Weiss et al., 1997; Figure 4—figure supplement 3A–D). These data show that the unannotated Aim2 isoform has a lower translational efficiency compared to the canonical form.

To further validate the effect the post-transcriptional mechanism of the unannotated Aim2 isoform, we performed time-course LPS stimulations. In primary BMDMs, a 72 hr, time-course stimulation with and without the treatment of iron (FAC) protein lysates was generated and Aim2 expression was measured by western blot (Figure 4R-SFigure 4—source data 1). Aim2 is expressed basally and significantly decreases upon LPS treatment at the 48 hr time point, most likely as a control mechanism to return the pathway to homeostasis and limit the inflammatory stimulation. When FAC is added to cells, Aim2 expression does not decrease at the 48 hr time point suggesting that it is indeed the IRE that is driving this decrease in Aim2 observed in the wild type cells (Figure 4R-SFigure 4—source data 1). Finally, we utilized the CRISPR/Cas9 technology to remove the Aim2 IRE endogenously (Figure 4—figure supplement 4A). In these immortalized BMDMs, RT-qPCR was performed with an LPS time course. The annotated Aim2 isoform shows no induction by LPS (Figure 4—figure supplement 4C), while the unannotated Aim2 isoform and the CDS sequence are induced by twofold (Figure 4—figure supplement 4D,E). There is no significant difference in transcriptional expression at the Aim2 locus between the WT and IRE KO cell lines (Figure 4—figure supplement 4D,E). Using these newly characterized WT and IRE KO clonal cell lines, we performed a 72 hr time course of LPS stimulation (Figure 4—figure supplement 4F,G, Figure 4—figure supplement 4—source data 1). At the 48 hr time point, there is a significant decrease in expression of Aim2 protein in WT cells, but not in the IRE KO cells (Figure 4—figure supplement 4H), further indicating that endogenous Aim2 is regulated through the IRE motif located in the AFE.

Discussion

While we have come a long way in determining the transcriptomes of immune genes to better understand signaling pathways, very little work has focused on the role that mRNA isoforms play (Akira et al., 2006; Medzhitov and Janeway, 1998; Pai et al., 2016). Over 95% of genes have more than one mRNA transcript due to alternative splicing, but the regulatory importance of these splicing events is not fully understood (Carninci et al., 2006; Tian and Manley, 2017; Wang et al., 2008). On a gene-by-gene approach, alternative splicing has been shown to play a role in health and disease by shaping the proteome (Nishiyama et al., 2000; Yang et al., 2011; You et al., 2016). Globally, a number of labs have tackled the prevalence of alternative splicing in vitro and in vivo, showing that alternative splicing can affect both the nature and duration of inflammation (Beyer et al., 2012; Bhatt et al., 2012; Janssen et al., 2020). To date, no one has examined the conservation of these mechanisms using primary cells or utilized long-read sequencing to build the transcriptome de novo to obtain a complete understanding of the extent of alternatively expressed isoforms generated following an immune response.

In our study, we demonstrate a conservation of splicing, specifically AFE events in both human and murine macrophages. We found that there are 11 genes that have AFE in both human and mouse macrophages (Figure 1—figure supplement 5A). Most studies to date have focused on isoform changes linked to genes that are differentially expressed following inflammation, and interestingly these 11 genes would have been previously overlooked because many of them are not differentially expressed, emphasizing the importance in studying isoform expression in all conditions (Figure 1E,F). Of the 11 conserved genes, 7 AFE isoforms have been previously studied in some context including Rps6ka1 (Qin et al., 2018), Ncoa7 (Singer et al., 2008; Yu et al., 2015), Rcan1 (Pang et al., 2018), Wars (Liu et al., 2004; Miyanokoshi et al., 2013), Arap1 (Kulzer et al., 2014), Tsc22d1 (PMID:21448135), and Sgk1 (Arteaga et al., 2007; Kobayashi et al., 1999; Lang et al., 2006). While this validates our technique, it is important to note that none of these genes had been connected to inflammation or formally shown to be conserved mechanisms of regulation, besides Wars (Liu et al., 2004; Miyanokoshi et al., 2013). This also highlights our method’s ability to accurately identify inflammatory regulated RNA isoforms, in addition to the uncharacterized AFE events of Ampd3, Snx10, Shisa5, and Tspan4. Furthermore, Snx10 is studied outside the context of inflammation, has been implicated in chronic inflammatory disease, and our study may suggest new insights into how alternative splicing could be regulating these genes (You et al., 2016). We further validated Ncoa7, Rcan1, and Ampd3 in human and murine macrophages using RT-PCR (Figure 1—figure supplement 6).

To overcome the current limitations of any transcriptome build, we used direct RNA nanopore technology on primary murine macrophages to build our own transcriptome de novo with the aim of identifying novel full-length transcriptional isoforms that are expressed with and without inflammation (Tang et al., 2020; Workman et al., 2019). The limitation of direct RNA-sequencing is read depth, and our average read depth was about 1 million reads per treatment. In a previous study of direct RNA-sequencing, we found that increased sequencing depth will result in additional isoforms detected (Workman et al., 2019). Therefore, we expect to see additional unannotated AFE and other AS events with increased sequencing depth. However, even with 1 million reads, we were able to build a transcriptome that led to the identification of 154 novel AS events. Following this, we identified hundreds of novel isoforms resulting in 95 novel AFE events, supported by CAGE-seq (Figure 1, Figure 1—figure supplements 7 and 8), including an unannotated mRNA isoform of the well-studied gene protein absent in melanoma 2 (Aim2).

Aim2 is characterized as an interferon-inducible gene (Johnstone and Trapani, 1999) (PMID:10454530), functioning as a cytoplasmic dsDNA sensor leading to the formation of an inflammasome and eventual cleavage and release of pro-inflammatory cytokines of IL1β and IL18 (Bürckstümmer et al., 2009; Fernandes-Alnemri et al., 2009; Hornung et al., 2009). Our study highlights that it is an alternative mRNA isoform of Aim2 that is inducible, and that this upregulated transcript is translated less efficiently compared to the canonical isoform. This novel finding goes against the existing assumption that induced gene expression results in induced protein expression (Figures 1I, 2 and 4, Figure 4—figure supplement 4). FLAIR-identified transcripts (Workman et al., 2019) show three clear AFEs for Aim2, and only one of those first exons is inflammatory inducible (Figure 2C,D). RT-qPCR further confirms that the annotated TSS of Aim2 is not inducible, while the unannotated AFE of Aim2 is LPS inducible (Figure 2E–H). This result highlights the need for cell-type and treatment-specific transcriptome annotations if one is to have a complete understanding of the transcriptome and proteome of a given cell.

We further investigated what drove the expression of this novel Aim2 isoform, as well as what drove the expression of all the AFE genes. Using ATAC-seq (Lara-Astiaso et al., 2014; Tong et al., 2016) and ChIP-seq (Tong et al., 2016) data sets, we were able to determine that 50% of the AFE events were driven partially by chromatin accessibility and inflammatory-specific TFs (Figure 3A). Using HOMER, a TF motif enrichment tool, we identified that the majority of TFs that regulate the expression of the AFE genes are involved in metabolism pathways (Figure 3B). Additionally, another mechanism that can drive AFE usage is the splicing of internal exons that activate proximal upstream promoters (Fiszbein et al., 2019). Further analysis incorporating more TFs and coupling multiple splicing events will be necessary to determine the definitive regulatory mechanism of all AFEs. Interestingly, the annotated Aim2 promoter accessibility is constitutively open across all hematopoietic cells, while the unannotated Aim2 promoter is only accessible in myeloid progenitors or terminally differentiated cells (Figure 3—figure supplement 1A), meaning that the novel Aim2 isoform may be specific to myeloid cells. Furthermore, Aim2 annotated and unannotated promoter usage is not driven by chromatin accessibility (Figure 3—figure supplement 1B,C) but is driven by the activation of inflammatory-specific TFs (Figure 3C–E, Figure 3—figure supplement 2L).

There is no difference in the open reading frame of the novel isoform of Aim2 when compared to the annotated transcript using NCBI ORFfinder; therefore, we turned our attention to a possible regulatory mechanism within the 5′UTR (Mignone et al., 2002). Broadly, UTRs play crucial roles in the post-transcriptional regulation of gene expression, including alteration of the mRNA translational efficiency (van der Velden and Thomas, 1999), subcellular localization (Jansen, 2001), and stability (Bashirullah et al., 2001). Post-transcriptional regulatory mechanisms of Aim2 have not been previously studied. Using RegRNA2.0 (Chang et al., 2013) globally, we identify that LPS-activated AFEs contain more regulatory elements than basal first exons (Figure 4—figure supplement 1). More importantly, we identified an IRE, a regulatory motif unique to only the novel 5′UTR of Aim2 (Figure 4F). Utilizing a GFP reporter plasmid, we were able to determine that the IRE motif was functional, by recapitulating the same experiments used on the protein Ferritin, the first functional IRE motif ever studied (Leedman et al., 1996). Then, we showed the inflammatory-specific mechanism regulating Aim2 protein expression by performing a western blot on primary BMDMs with and without FAC during a 72 hr LPS time-course experiment. Aim2 protein is basally expressed, and while the transcript is inducible, specifically the novel isoform, we do not observe an increase in expression of Aim2 protein by western blot. In fact, we find that Aim2 protein decreases following inflammation, and this can be reversed by iron supplementation. Finally, we confirmed the IRE mechanism by using CRISPR/Cas9 to generate an IRE KO clonal cell line. Once our cell line was characterized (Figure 4—figure supplement 4A–E), we performed an LPS time-course stimulation and measured Aim2 protein expression. At the 48 hr time point, we see a significant rescue of Aim2 protein expression in the IRE KO cell line (Figure 4—figure supplement 4F–H). This could be a critical regulatory step that has evolved to ensure the Aim2 pathway is switched off following its formation and activation of the inflammasome.

These results demonstrate that the inflammatory-specific mRNA isoform of Aim2 has lower translational efficiency than the canonical form and that protein translation can be increased by the addition of iron. Crane et al., 2014 demonstrate that ROS can contribute to activation of Aim2 inflammasome in mouse macrophages. Our proposed mechanism of translational regulation of Aim2 is through an IRE, which is known to directly interact with IRP proteins (Outten, 2017; Rouault, 2006; Wang et al., 2004). Interestingly, IRP2 degradation can be driven not only through iron, but also through ROS and RNS (Hanson and Leibold, 1999), further supporting this novel IRE regulatory mechanism of Aim2 protein expression. Finally, Cheng et al. have shown that AIM2 is regulated by oxidative stress and show that overactivation of AIM2 inflammasome can contribute to pancreatic tumorigenesis, all within the environment of mitochondrial iron overload (Li et al., 2018). This newly identified isoform, with an IRE-specific translational mechanism, provides mechanistic understanding to these recent studies of Aim2 (Crane et al., 2014; Li et al., 2018). These findings could have significance for better understanding the mechanisms driving pathology in inflammatory disease such as systemic lupus erythematosus (SLE) (Corbett, 2018). AIM2 expression levels have been correlated with severity of inflammation in SLE patients (Zhang et al., 2013), and it is well known that iron is dysregulated in this disease (Vanarsa et al., 2012). It is possible that AIM2 levels remain high in SLE patients due to dysregulated iron; therefore, homeostasis in macrophages cannot be maintained.

In summary, signaling within macrophages enables us to fight infection but also can contribute to pathological inflammation associated with a wide variety of diseases. While there are multiple regulatory checkpoints in place to control inflammation, we propose that alternative splicing and translational regulation play critical roles in maintaining this type of control. A better understanding of the molecular mechanisms that control inflammatory-regulated genes, including Aim2, could provide new targets for therapeutic intervention of autoimmune and inflammatory diseases.

Materials and methods

Maintenance of mice

Request a detailed protocol

UCSC and the Institutional Animal Care and Use Committee maintained mice under specific pathogen-free conditions in the animal facilities of University of California Santa Cruz (UCSC) in accordance with the guidelines.

Human PBMC-derived macrophage differentiation and in vitro stimulation

Request a detailed protocol

Human peripheral blood mononuclear cells (PBMCs) were enriched by density gradient centrifugation of peripheral blood from healthy human donors through a Ficoll-Paque PLUS (GE Healthcare) gradient. Monocytes were isolated from PBMC by negative selection using the EasySep Human Monocyte Isolation Kit (STEMCELL Technologies) according to the manufacturer’s instructions. To differentiate monocytes into macrophages, recombinant human M-CSF (50 ng/mL) was used in RPMI-1640 medium with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 10 mM HEPES, 1 mM sodium pyruvate, 100 U/mL penicillin, and 100 µg/mL streptomycin. The culture medium that contained fresh recombinant human M-CSF was replaced every 2 days.

Cell culture, mouse macrophage differentiation and stimulation

Request a detailed protocol

Cells were cultured in DMEM with 10% FBS supplemented with penicillin/streptomycin or ciprofloxacin. Primary BMDMs were generated by cultivating erythrocyte-depleted bone marrow cells in the presence of 30% L929 supernatant, and the cells were used for experiments 6–9 days after differentiation. J2Cre virus (Blasi et al., 1985) was used on day 3/4 after isolation of bone marrow cells to establish transformed BMDM cell lines. BMDMs were cultivated in the presence of J2Cre virus for 48 hr, and L929 was then gradually tapered off over 6–10 weeks depending on the growth pattern of transformed cells.

In vitro stimulation of macrophages

Request a detailed protocol

Bone-derived macrophage cells were primed with 100 µM of FAC for 24 hr prior to TLR stimulation. BMDM cells were stimulated with TLR ligands for the indicated time points using the following concentrations: LPS 200 ng/mL (TLR4). For RNA and protein isolation, 1–2 × 106 cells were seeded in 12-well format.

RT-PCR validation

Request a detailed protocol

RT-PCR validation was completed using three biological replicates. KAPA HiFi HotStart ReadyMix PCR Kit (Kapa Biosystems) and the manufacturer’s suggested cycling protocol were used to complete the PCR reaction with the following primers:

  • Mse_Denr_F1: ATCGCGATAAAGGCTCATTG

  • Mse_Denr_F2: GCTACCTGTCCTTTTCCCCA

  • Mse_Denr_R: AACTTGGCACTGTTCTTCGT

  • Mse_Arhgef7_F1: TGTTGTTCTGGGGTTTGTGA

  • Mse_Arhgef7_F2: CTGTGTGTTGCAGGTCTACC

  • Mse_Arhgef7_R: GTGTCACCAAGGAGCTGAGG

  • Mse_Ncoa7_F1: GTGGTGGAGAAGGAAGAGCT

  • Mse_Ncoa7_F2: TTCTATTGTGCCAGGCCTGA

  • Mse_Ncoa7_R: GCATGTTTTCCAGGAGTGCA

  • Mse_Ampd3_F1: CCCTACTGTAGATGAATCCCCTTA

  • Mse_Ampd3_F2: GCTGAGCTTTGTGTCTGTGT

  • Mse_Ampd3_R: GGGGACAGTAAACAGGGACA

  • Mse_Rcan1_F1: ACTGGAGCTTCATCGACTGC

  • Mse_Rcan1_F2: GACTGAGAGAGCGAGTCGTT

  • Mse_Rcan1_R: CATCGGCTGCAGATAAGGGG

  • Hu_NCOA7_F1: TGTTCAGTGGTCTCCCGATGTCTATGG

  • Hu_NCOA7_R: GGGCCGTAGGACAGGCAGCA

  • Hu_NCOA7_R2: AGCGTGGCTACAAGTAACTGTGGTGT

  • Hu_AMPD3_F1: TATGCAAAACAGAGACCTCC

  • Hu_AMPD3_R: CACTTCAGAGATGTTCAGCT

  • Hu_AMPD3_F2: CCTGCTTGGTTTTAGAGGAT

  • Hu_RCAN1_F1: GACTGGAGCTTCATTGACTG

  • Hu_RCAN1_R: ATTCTGACTCGTTTGAAGCT

  • Hu_RCAN1_F2: TAGCGCTTTCACTGTAAGAA

Band intensities were measured for each band in each condition and sample using FIJI (Schindelin et al., 2012). The relative abundance of each isoform was calculated using the equation to calculate percent spliced in (PSI) (PSI = inclusion/(inclusion + exclusion)) in each condition and sample to validate the computationally derived delta PSI values. A gel extraction was completed for each band using the PCR clean-up Gel extraction kit (Machery-nagel). The PCR product was confirmed using Sanger Sequencing.

RNA isolation, RT-qPCR

Request a detailed protocol

Total cellular RNA from BMDM cell lines or tissues was isolated using the Direct-zol RNA MiniPrep Kit (Zymo Research) according to the manufacturer’s instructions. RNA was quantified and controlled for purity with a nanodrop spectrometer (Thermo Fisher). For RT-qPCR, 500–1000 ng were reversely transcribed (iScript Reverse Transcription Supermix, Biorad) followed by RT-PCR (iQ SYBRgreen Supermix, Biorad) using the cycling conditions as follows: 50°C for 2 min, 95°C for 2 min followed by 40 cycles of 95°C for 15 s, 60°C for 30 s, and 72°C for 45 s. The melting curve was graphically analyzed to control for nonspecific amplification reactions. Quantitative RT-PCR analysis was performed with the following primers:

  • Mse_Aim2_F_Annotated: CCGCCATGCTTCCTTAACTA

  • Mse_Aim2_F_Unannotated: AGGCGGATGGTTTGAACTCT

  • Mse_Aim2_R_Exon2: TTGAAGCAACTTCCATCTGC

  • Mse_Aim2_CDS_F: AGTACCGGGAAATGCTGTTG

  • Mse_Aim2_CDS_R: GAGTGTGCTCCTGGCAATCT

  • Mse_Gapdh_F: CCAATGTGTCCGTCGTGGATC

  • Mse_Gapdh_R: GTTGAAGTCGCAGGAGACAAC

  • Mse_Neat1_F: TTGGGACAGTGGACGTGTGG

  • Mse_Neat1_R: TCAAGTGCCAGCAGACAGCA

Cloning strategy for 5′UTR GFP plasmid

Request a detailed protocol

The GFP reporter plasmid was CMV-Zeo-t2A-GFP. Zeocin is flanked by NheI and AgeI. The sequence of the annotated and unannotated 5′UTR was used as defined by the UCSC RefSeq and our sequencing results to be. Using KAPA HiFi HotStart ReadyMix PCR Kit (Kapa Biosystems), the two 5′UTRs were amplified from cDNA.

5′UTRcDNA sequence
AnnotatedTTCCTGTCCTGTCTGCCGCCATGCTTCCTTAACTAGCTGCTAGGTTTTTTCCTTGTCGTGATGAAATCCACCCTCATGGACCTGTAAGTAAAATGTAGACTTGCATAGAGTGCTGTAATCTTACGGCCGAGGTTTCTTTTCAGGCTGATCCTGGGACTGTGAG
UnannotatedTATATCTAAAATACCTCTGGTTGAGACCTCACAGCTGGAGGAGAAACTCTGCTGAGGCTTGTAAAAAGGAAACTGAAAACTAGCATTTGCTTGGGCAGAGCCTTAATATATAATTATTTTGCCCCAGCATCAGGGTTTAGGACTCAGCTATAGGGCCAGGACTAGCCAAGCTTCAAAGTGAAAGAAGATAGTTGAGAGTACTTTCTGCTTTCTGTCTCCCAAGACCTGATTTTCATGATTTTCATGTCCTACTACTCATAGTGAAAATCTTTGTGAGGCGGATGGTTTGAACTCTCAGGACATACACCAGTCCCTGAGTTGAGAACTAAGGCTGCTTTGGAGAGAAGAAAATCCCCTGAGGTAAGTAGACTTGCATAGAGTGCTGTAATCTTACGGCCGAGGTTTCTTTTCAGGCTGATCCTGGGACTGTGAG

Primers

Request a detailed protocol

  • cDNA_F_Annotated: TTCCTGTCCTGTCTGCCG

  • cDNA_F_Unannotated: TATATCTAAAATACCTCTGGTTGAGACCTC

  • cDNA_R_5'UTR: CTCACAGTCCCAGGATCAGC

The 5′UTRs were then PCR amplified with primers containing restriction enzyme sites for AgeI and NheI.

  • NheI_F_Annotated: GGTGCTAGCTTCCTGTCCTGTCTGCC

  • NheI_F_Unannotated: GGTGGTGCTAGCTATATCTAAAATACCTCTGGTTGAGA

  • AgeI_R_5'UTR: GGTACCGGTCTCACAGTCCCAGGATCAGC

PCR products, as well as the GFP plasmid, were then processed using AgeI and NheI restriction enzymes overnight. These samples were run on a 1% agarose gel. A gel extraction was completed for each band using the PCR clean-up Gel extraction kit (Machery-nagel). The PCR product was confirmed using Sanger Sequencing.

Site-directed mutagenesis

Request a detailed protocol

Set up PCR reaction with 1 µL 279 plasmid, with unannotated Aim2 5′UTR, 5 µL 10 × Phu buffer, 1 µL F primer (0.1 µg/µL) [Remove_IRE_F - CCCTGATGCTGGGGCAAAATAATTATAAATGCTAGTTTTCAGTTTC], 1 µL R primer (0.1 µg/µL) [Remove_IRE_R - GAAACTGAAAACTAGCATTTATAATTATTTTGCCCCAGCATCAGGG], 1 µL dNTP (10 nM), 1 µL Phu polymerase, and 40 µL dH2O. PCR program: 95℃ 1 min, 18 cycles of 95℃ 30 s, 55℃ 1 min, 72℃ 1 min, then end PCR with 72℃ 1 min and 4℃ hold. Add 0.5 µL of Dpn1 (NEB) to 25 µL PCR reaction. Incubate at 37℃ for 1 hr to digest parental DNA. Transform digested and undigested plasmid into DH5α competent cells. Pick ~10–15 colonies and start overnight cultures. Colony PCR plasmids using Dpn1_Colony_PCR_F - TTGGCTAGTCCTGGCCCTAT and Dpn1_Colony_PCR_R - GCTGGTTTAGTGAACCGTCAG to check for 20 bp deletion on a 3% agarose gel. Grow up colonies that have deletion, miniprep plasmids and send to Sequetech for sequence verification.

Transfection of 5′UTR GFP and mCherry plasmid

Request a detailed protocol

A 1:1 ratio of the GFP vector containing the mature sequence of Aim2 5′UTR (annotated or unannotated) or zeocin and a plasmid containing mCherry were transfected into 293T cells for 48–72 hr. A 6-well plate of HEK293Ts was plated the night before with a concentration of 2 × 105. HEK293Ts cells were primed with 100 µM of FAC for 24 hr prior to transfection. Transfection was performed on HEK293Ts (±100 µM FAC) using lipofectamine 2000, serum-free OPTI-MEM media was used as a transfection reagent according to the manufacturer’s instructions, and a (1:1) concentration of the 5′UTR GFP reporter plasmid and the mCherry control plasmid. HEK293Ts were visualized via flow cytometry 48–72 hr post transfection.

Polysome profiling

Request a detailed protocol

Prior to lysis, cells were treated with cycloheximide (100 mg/mL), 10 min at 37°C, 5% CO2. Cells were washed three times with ice-cold PBS and lysed in ice cold buffer A (0.5% NP40, 20 mM Tris HCl pH 7.5, 100 mM KCl, and 10 mM MgCl2). Lysates were passed three times through a 23G needle and incubated on ice for 7 min. Extracts were then centrifuged at 10K rpm for 7 min at 4°C. The supernatant was collected as crude cytosolic extract. Cytosolic extracts were overlaid on 10–50% sucrose gradients prepared in 20 mM Tris HCl pH 7.5, 100 mM KCl, and 10 mM MgCl2 buffer (prepared using the Gradient Station, Biocomp Instruments). Gradients were then ultracentrifuged at 40K rpm for 1 hr 20 min at 4°C using an SW41 in a Beckman ultracentrifuge. Individual polyribosome fractions were subsequently purified using a Gradient Station (Biocomp Instruments) and stored in (1:3) TRI Reagent.

IRE KO cell line generation

Request a detailed protocol

The gRNA construct was constructed from a pSico lentiviral backbone driven by EF1a promoter expressing T2A flanked genes: puro and cherry. gRNAs were expressed from a mouse U6 promoter. Cloning of 20 nt gRNA spacer forward/reverse oligos was annealed and cloned via the AarI site.

  • IRE_gRNA1_F: TTGGACTGAAAACTAGCATTTGCT

  • IRE_gRNA1_2: AAACAGCAAATGCTAGTTTTCAGT

  • IRE_gRNA2_F: TTGGCTGAAAACTAGCATTTGCTT

  • IRE_gRNA2_2: AAACAAGCAAATGCTAGTTTTCAG

  • IRE_gRNA3_F: TTGGGGCAAAATAATTATATATTA

  • IRE_gRNA3_2: AAACTAATATATAATTATTTTGCC

3 μg of pooled gRNAs were electroporated using the Lonza Amaxa Mouse Macrophage Nucleofector kit (VPA-1009). Electroporated cells were then selected using puromycin 5 µg/mL and grown to 80% confluency. Limited dilution series were seeded in 96-well plate, let to grow for 3 weeks. Then the clonal cell lines were genotyped using: F:GCAGGAAATAACTTTTGTGGAGT and R:TGGGAGACAGAAAGCAGAAAG. Then this PCR product was sequenced, and KO efficiency was assessed using ICE Synthego (https://ice.synthego.com/#/). Then the hairpin structure was assessed by RNAfold (http://rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi).

Protein lysate and western blot

Request a detailed protocol

Cell lysates were prepared in RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, 1 % [v/v] NP-40, 0.5% [w/v] sodium deoxycholate, 0.1% [w/v] SDS) containing protease inhibitor cocktail (Roche) and quantified by the Bicinchoninic Acid Assay (BCA) assay (Thermo Fisher). Equivalent masses (20 μg) of each sample were resolved by SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane and western blotted with either Aim2 (1:1,000; Cell Signaling #63660) or horseradish peroxidase-conjugated b-actin monoclonal antibody (1:5000, Santa Cruz Biotechnology) used as a loading control. HRP-conjugated goat anti-rabbit (1:1500, Biorad) secondary antibodies were used. ImageJ (Schindelin et al., 2012) was used for quantification of western blots.

Statistical analysis

Request a detailed protocol

Error bars represent the standard deviation of biological triplicates. Student’s t-tests were performed using GraphPad Prism. Asterisks indicate statistically significant differences between mouse lines (*p>0.05, **p>0.01, ***p>0.005).

Illumina RNA-sequencing (human)

Request a detailed protocol

RNA-seq libraries were prepared with the Illumina TruSeq RNA Sample Preparation kit (Illumina) according to the manufacturer’s protocol. Libraries were validated on an Agilent Bioanalyzer 2100. Indexed libraries were equimolarly pooled and sequenced on a SE50 (single-end 50 base pair) Illumina HiSeq2500 lane, which yielded an average of about 30 × 106 reads/sample.

Illumina RNA-sequencing (mouse)

Request a detailed protocol

For generation of RNASequencing libraries, RNA was isolated as described above and the RNA integrity was tested with a BioAnalyzer (Agilent Technologies) or FragmentAnalyzer (Advanced Analytical). For RNASequencing, target RIN score of input RNA (500–1000 ng) usually had a minimum RIN score of 8. RNASequencing libraries were prepared with TruSeq stranded RNA sample preparation kits (Illumina), and depletion of ribosomal RNA was performed by positive selection of polyA+ RNA. Sequencing was performed on Illumina HighSeq or NextSeq machines. RNA-seq 50 bp reads were aligned to the mouse genome (assembly GRCm38/mm10) using TopHat (Trapnell et al., 2009). The Gencode M13 gtf was used as the input annotation. Differential gene expression-specific analyses were conducted with the DESeq (Anders and Huber, 2010) R package. Specifically, DESeq was used to normalize gene counts, calculate fold change in gene expression, estimate p-values and adjusted p-values for change in gene expression values, and perform a variance-stabilizing transformation on read counts to make them amenable to plotting. Data was submitted to GEO GSE141754.

Gene expression analysis

Request a detailed protocol

DESeq2 v1.22.2 (Love et al., 2014) was used to create counts tables (Supplementary files 6-7) and complete differential gene expression analysis on RNA-seq data from human monocyte-derived macrophage ± 18 hr LPS and mouse BMDM ± 6 hr LPS experiments. The sample conditions used were ‘control’ and ‘LPS.’ Data was plotted using ggplot2 v3.1.1 (Wickham, 2009). Significance thresholds were set to |log2FC| ≥ 2 and adjusted p-value≤0.05. The list of genes with significant AFE events was then highlighted on the appropriate graphs.

Alternative splicing quantification (PSI)

Request a detailed protocol

JuncBASE (Brooks et al., 2011) was used to identify and quantify alternative splicing events. After the identification of each alternative splicing event, JuncBASE counts reads supporting the inclusion and exclusion isoform of each event. Isoform abundances are then calculated by dividing the read counts for the isoform by the length of the isoform. Ψ-values for each splicing event are derived from the isoform abundances:

PSI formula:

PSI=InclusionIsoformAbundance/(InclusionIsoformAbundance+ExclusionIsoformAbundance)

Nanopore direct RNA-sequencing

Total RNA extraction

Request a detailed protocol

Total RNA was extracted according to Workman et al., 2019. 5 × 107 frozen macrophages were resuspended in 3 mL of TRI-Reagent (Invitrogen AM9738) and vortexed for 5 min. The mixture was incubated at RT for 5 min, transferred to 1.5 mL tubes, and spun down to remove debris. Supernatant was transferred to fresh tubes and chloroform extracted. The aqueous portion was mixed with an equal volume of isopropanol, incubated for 15 min at RT, and centrifuged at 12,000 g at 4°C. Pellet was washed twice with 75% ethanol, air dried, and resuspended in nuclease-free water.

Poly(A) RNA isolation

Request a detailed protocol

100 μg aliquots of total RNA preparation were brought to 100 μL in nuclease-free water and poly-A selected using NEXTflex Poly(A) Beads (BIOO Scientific Cat#NOVA-512980) according to the manufacturer’s instructions. The resulting poly-A RNA solution was stored at –80°C.

Library preparation and MinION run

Request a detailed protocol

A native RNA-sequencing library was prepared following the ONT SQK-RNA001 using Superscript IV (Thermo Fisher) for the reverse transcriptase step. Sequencing was performed using ONT R9.4 flow cells and the standard MinKNOW protocol.

Basecalling and sequence alignments

Request a detailed protocol

ONT albacore version 2.1.0 was used to baseball Nanopore direct RNA raw signal. We used minimap2 (Li, 2016) with default parameters to align reads to the mm10 mouse genome reference. Following alignment, we used SAMtools (Li et al., 2009) to filter out reads with mapping quality (MAPQ) less than 30.

Alignment of paired-end mouse RNA-seq data

Request a detailed protocol

Bowtie2-build v2.3.1 (Langmead and Salzberg, 2012) was used to build the index files from GRCm38.p6 mouse (mm10 assembly) genome sequence obtained from Gencode. The index files were then used for completing paired-end alignment of each sample using TopHat2 v2.1.1 (Kim et al., 2013) with parameters: --segment-length 20, --library-typ fr-firststrand, --no-coverage-search.

Identification of splicing events and calculating PSI

Request a detailed protocol

Human monocyte-derived macrophage ± LPS and mouse BMDM ± LPS were each run through JuncBASE v1.2 (Brooks et al., 2011) to calculate PSI values and identify splicing events. The JuncBASE parameters used for the identification of splicing events and calculation of PSI in human monocyte-derived macrophage ± LPS are -c 1.0 j [introns from Gencode v24 (hg19 assembly) (Frankish et al., 2019) --jcn_seq_len 88]. The JuncBASE parameters used for the identification of splicing events and calculation of PSI in mouse bone marrow-derived macrophage ± LPS are: -c 1.0 j [introns from Gencode M18 (mm10 assembly) (Frankish et al., 2019) --jcn_seq_len 88].

Differential splicing analysis

Request a detailed protocol

Differential splicing analysis was completed using DRIMSeq v1.10.1 (Nowicka and Robinson, 2016) and the compareSampleSets.py script within JuncBASE. CompareSampleSets.py applies the statistical t-test and DRIMSeq applies the framework of the Dirichlet-multinomial distribution for differential analysis. Each tool was used to apply the respective statistical method in order to determine significant differentially spliced events between control (-LPS) and LPS (+LPS) conditions. The AS_exclusion_inclusion_counts_lenNorm.txt JuncBASE output table from the identification and quantification analysis of each experiment was used as the input table for both compareSampleSets.py and DRIMSeq.

For all experiments, compareSampleSets.py was run using parameters: --mt_correction BH --which_test t-test --thresh 10 --delta_thresh 5.0. The following parameters were used for the differential splicing analysis of data from human monocyte-derived macrophage ± LPS with DRIMSeq: min_samps_gene_expr = 8, min_samps_feature_expr = 4, min_gene_expr = 10, min_feature_expr = 0. The following parameters were used for the differential splicing analysis of data from mouse BMDM ± LPS with DRIMSeq: min_samps_gene_expr = 6, min_samps_feature_expr = 3, min_gene_expr = 10, min_feature_expr = 0. Following differential splicing analysis using each tool, genes with significant differential splicing events were filtered for using thresholds of a corrected/adjusted p-value≤0.25 and a |Δ PSI| ≥ 10. Within each category of event type, the union of genes with significant events identified using compareSampleSets.py and DRIMSeq within each experiment was used for further comparison. Novel intron retention events, annotated with a ‘N,’ were removed for further analyses.

Ranking analysis

Request a detailed protocol

Following differential splicing analysis with JuncBASE, events were ordered by p-value. A subset of the top 5, 10, 15, 20, and 25% events were quantified by event type. Only events having a |Δ PSI| ≥ 10 were considered. The same analysis was completed following differential splicing analysis with DRIMSeq2.

Analysis of Listeria and Salmonella data sets following 24 hr stimulation

Request a detailed protocol

Data for the 24 hr time point was downloaded from GEO (GSE73502) for control and experimental conditions: Listeria and Salmonella (Pai et al., 2016). JuncBASE and differential alternative splicing analysis was run using each pair of control and experimental samples and parameters as described above. Following differential alternative splicing analysis, alternative splicing events were categorized by event type using two significance thresholds: |Δ PSI| ≥ 10 and adjusted p-value≤0.05 or |Δ PSI| ≥ 10 and adjusted p-value≤0.25. Only known (K) events were considered for intron retention events. Jcn_only_AD and jcn_only_AA events were not considered.

Identification of high-confidence isoforms from nanopore data

Request a detailed protocol

FLAIR (Tang et al., 2020) was used to assemble the high-confidence isoforms from native RNA sequencing of mouse BMDM ± 6 hr LPS. FLAIR modules align, correct, and collapse were used for the assembly. Corresponding short-read data was used when running the correct module in order to help increase splice-site accuracy. Putative promoter regions were obtained using ATAC-seq data from Atianand et al., 2016; Tong et al., 2016 converted to mm10 coordinates using liftOver (Hinrichs et al., 2006), and used when running the collapse module.

Creating merged reference annotation files and incorporating nanopore

Request a detailed protocol

The isoforms.gtf output file from FLAIR collapse was combined with the Gencode M18 (mm10 assembly) basic annotation using cuffmerge from Cufflinks v2.2.1 (Trapnell et al., 2010) with parameter: -s GRCm38.p6.genome.fa. Similarly, the isoforms.gtf output file was combined with the Gencode M18 (mm10 assembly) comprehensive annotation with parameter: -s GRCm38.p6.genome.fa. The resulting comprehensive annotation output file was used to generate an intron coordinate file for the identification of splicing events and calculating PSI of splicing events found in mouse BMDM ± 6 hr LPS using JuncBASE with parameters: -c 1.0, -j [intron coordinates from merged comprehensive annotation], --jcn_seq_len 238. Parameters used for finding significantly differentially spliced events using compareSampleSets.py from JuncBASE are: --mt_correction BH --which_test t-test --thresh 10 --delta_thresh 5.0. Parameters used for finding significantly differentially spliced events using DRIMSeq are: min_samps_gene_expr = 6, min_samps_feature_expr = 3, min_gene_expr = 10, min_feature_expr = 0.

Creating and comparing gene lists

Request a detailed protocol

For each experiment, a table with the union of significant events found using DRIMSeq and compareSampleSets.py was created. A list of genes with significant events was generated for each experiment using this table. BioVenn (Hulsen et al., 2008) and DrawVenn (Li, 2016) were then used to remove duplicate gene names and compare the lists of genes to find unique and common genes between experiments.

Identification of genes with conserved alternative splicing events

Request a detailed protocol

Following differential splicing analysis with JuncBASE and DRIMSeq2, gene names of significant events (|Δ PSI| ≥ 10 and adjusted p-value≤0.25) were curated by event type. Overlaps were determined between human and mouse JuncBASE and DRIMSeq2 events determined with and without the support of Nanopore-identified transcripts.

Comparison of all alternative splicing events identified in mouse using Gencode or de novo + Gencode annotations

Request a detailed protocol

Following differential splicing analysis with JuncBASE and DRIMSeq2, gene names of significant events (|Δ PSI| ≥ 10 and adjusted p-value≤0.25) were curated by event type. Overlaps were determined between mouse JuncBASE and DRIMSeq2 events determined with and without the support of Nanopore-identified transcripts.

Validation of FLAIR-identified TSS with CAGE data

Request a detailed protocol

Coordinates corresponding to mouse CAGE peaks (mm9.cage_peak_phase1and2combined_coord.bed) were downloaded from the FANTOM5 database (FANTOM Consortium and the RIKEN PMI and CLST (DGT), 2014). Coordinates were converted from mm9 to mm10 using liftOver. TSS of FLAIR-identified isoforms were extracted using the pull_starts.py script included in the FLAIR software. TSS were extracted from the Gencode M18 annotation using a custom script and used as the set of known TSS. TSS of FLAIR-identified isoforms were annotated as known if they overlapped with known TSS using bedtools (v 2.25.0) tool intersectBed. TSS of FLAIR-identified isoforms were annotated as being novel if they did not overlap with any known TSS. These were also identified using bedtools tool intersectBed with parameter: -v. CAGE peaks that overlapped with known and novel FLAIR-identified TSSs were identified using bedtools tool intersectBed, and the distribution of CAGE scores was plotted for each group.

Coordinates corresponding to CAGE peaks identified as true TSS by TSS classifier (TSS_mouse.bed) were downloaded from the FANTOM 5 database. Coordinates were converted from mm9 to mm10 using liftOver. The proportion of true known and novel FLAIR-identified TSS sites was identified by determining TSSs that overlapped with the CAGE peaks. Bedtools (v 2.25.0) tool intersect was used to complete this analysis.

Differential chromatin accessibility

Request a detailed protocol

Raw ATAC-seq fastq sequence files were published in Tong et al., 2016 and pulled from the GEO accession number GSE67357. A bowtie2 (Langmead and Salzberg, 2012) index file was created from the GENCODE mm10 version M18 genome annotation file, and the untreated and LPS-treated ATAC-seq reads were aligned using the created index file with the default bowtie2 parameters. Peaks were then called separately by treatment type on untreated and treated samples using the ENCODE published ATAC-seq peak calling pipeline (ENCODE Project Consortium, 2012, Lee et al., 2021a, https://github.com/ENCODE-DCC/atac-seq-pipeline) using the aligned reads as sequence input. Parameters that were used followed the basic JSON input file template, using an IDR threshold of 0.05. Peaks from both conditions were then merged using bedtools merge (Quinlan and Hall, 2010) if the tail ends were less than 10 bp away from each other in order to create a set of consensus peaks from both conditions. A GFF file was created from the merged peaks, assigning a unique ID to each peak. This GFF file was provided to HTSeq-count (Anders et al., 2015) along with the aligned reads for each condition in each replicate to count reads aligning to the unique peaks. The read count matrix was provided to DESeq2 (Love et al., 2014) to call differential peaks. All peaks were considered significant if log2FC > 0.8 and p-value<0.15.

Differential transcription factor binding

Request a detailed protocol

ChIP-seq fastq sequencing files for the NF-κB subunit p65 and interferon transcription factor Irf3 were downloaded from the GEO accession number GSE67357 published by Tong et al., 2016. ChIP-seq samples for input control, untreated, and LPS treatment were aligned using bowtie2 (Langmead and Salzberg, 2012) to the mm10 version M18 mouse genome annotation with default parameters. Peaks were separately called between untreated and treated conditions using the ENCODE published ChIP-seq peak calling pipeline (ENCODE Project Consortium, 2012, Lee et al., 2021b, https://github.com/ENCODE-DCC/chip-seq-pipeline2) from the aligned reads. The aligned input control reads were input as genomic background to account for noise in ChIP-seq experiments. The basic JSON input template file was used using an IDR threshold of 0.05. Differential peak analysis was done using the HOMER suite designed for ChIP-seq data (Heinz et al., 2010). Consensus peaks from both conditions were merged using mergePeaks within HOMER, reporting the direct overlap between peaks. Tag directories were created to count reads for each aligned sequence file with TagDirectory. The merged consensus peaks were then annotated for raw read counts using the tag directories for each replicate and condition with the annotatePeaks.pl tool. Annotated consensus peaks were provided to getDiffExpression.pl, normalizing to total read counts. Peaks were considered significant if they had a corrected p-value<0.25 and log2FC > 1.

Alternative splicing event overlap

Request a detailed protocol

To identify differential TF binding and chromatin remodeling at the promoters of the observed AFE events, the coordinates of the alternate first exon were determined from the statistical testing results. Significant (p-value<0.05) AFE events were first filtered out from all results. For all significant results, if the inclusion exon had a Δ PSI >10, the inclusion exon coordinates from the JuncBASE table (Supplementary files 1-3) were used as the coordinates for that splicing event. If the inclusion exon had a Δ PSI < –10, all other inclusion exons for that splicing event from the statistical testing (DRIMSeq or t-test) were considered, and any inclusion exon with Δ PSI >10 was used. Redundant events with the same exon coordinates were then filtered out, leading to a final set of exon coordinates. The coordinates were then extended to include 10 kbp upstream of the exon. Overlap of differential chromatin accessibility and different TF binding was done using bedtools intersect (Quinlan and Hall, 2010) with the significant differential peak coordinates and the AFE 10 kbp upstream region, returning the coordinates of the exon that show differential chromatin accessibility or TF binding.

HOMER promoter analysis

Request a detailed protocol

HOMER (Boeva, 2016) was used to search for sequence motifs in promoter regions of LPS-specific AFEs. The start coordinates from the BED file were extended to include the 3 kb upstream of the exon. These extended regions were used as input for HOMER’s findMotifsGenome.pl script, along with the mm10 reference genome from within HOMER and the size parameter set to ‘given.’ findMotifsGenome.pl was run initially with the -preparse flag to parse the reference genome based on the size of the input sequences and then was ran after without the preparse flag in order to generate the motif output file (Supplementary files 10-11).

BED file generation of inclusion and exclusion AFE

Request a detailed protocol

Significant (corrected p-value<0.25) AFE events were first filtered out from all results. For all significant results, if the inclusion exon had a Δ PSI >10, the inclusion exon coordinates from the JuncBASE table were used as the coordinates for that splicing event. If the inclusion exon had a Δ PSI < –10, all other exons for that splicing event from the statistical testing (DRIMSeq or t-test) were considered, and any other exon with Δ PSI >10 was used (Supplementary files 12-13).

Data availability

Sequencing data have been deposited in GEO under accession codes GSE141754.

The following data sets were generated
    1. Cattle M
    2. Robinson EK
    3. Jagannatha P
    4. Covarrubias S
    (2021) NCBI Gene Expression Omnibus
    ID GSE141754. Inflammation drives alternative first exon usage of critical immune genes including Aim2.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE141754
The following previously published data sets were used
    1. Pai AA
    2. Baharian G
    3. Pagé Sabourin GA
    4. Nédélec GY
    5. Grenier GJ
    6. Siddle GKJ
    7. Dumaine GA
    8. Yotova GV
    9. Burge GCB
    10. Barreiro GLB
    (2016) NCBI Gene Expression Omnibus
    ID GSE73502. Widespread shortening of 3' untranslated regions and increased exon inclusion characterize the human macrophage response to infection.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE73502
    1. Tong AJ
    2. Liu X
    3. Thomas BJ
    4. Lissner MM
    (2016) NCBI Gene Expression Omnibus
    ID GSE67357. A Stringent Systems Approach Uncovers Gene-Specific Mechanisms Regulating Inflammation.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE67357
    1. Tong A
    2. Thomas BJ
    (2016) NCBI Gene Expression Omnibus
    ID GSE74191. A Stringent Systems Approach Uncovers Gene-Specific Mechanisms Regulating Inflammation.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE74191
    1. Link VM
    2. Glass CK
    (2018) NCBI Gene Expression Omnibus
    ID GSE109965. Analysis of genetically diverse macrophages reveals local and domain-wide mechanisms that control transcription factor binding and function.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE109965
    1. Tong AJ
    2. Liu X
    3. Thomas BJ
    4. Lissner MM
    (2016) NCBI Gene Expression Omnibus
    ID GSE67343. A Stringent Systems Approach Uncovers Gene-Specific Mechanisms Regulating Inflammation.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi
    1. Song R
    2. Dozmorov I
    3. Malladi V
    4. Liang C
    5. Arana C
    6. Wakeland B
    7. Pasare C
    8. Wakeland EK
    (2021) NCBI Gene Expression Omnibus
    ID GSE147310. IRF1 governs the differential Interferon-Stimulated Gene responses in human monocytes and macrophages by regulating chromatin accessibility.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE147310

References

    1. Ghiasvand S
    2. Bakhshinejad B
    3. Mowla SJ
    4. Sadeghizadeh M
    (2014)
    Potential roles of 5´ utr and 3´ utr regions in post-trans-criptional regulation of mouse oct4 gene in BMSC and p19 cells
    Iranian Journal of Basic Medical Sciences 17:490–496.
    1. Hanson ES
    2. Leibold EA
    (1999)
    Regulation of the iron regulatory proteins by reactive nitrogen and oxygen species
    Gene Expression 7:367–376.
    1. Weiss G
    2. Bogdan C
    3. Hentze MW
    (1997)
    Pathways for the regulation of macrophage iron metabolism by the anti-inflammatory cytokines IL-4 and IL-13
    Journal of Immunology 158:420–425.

Decision letter

  1. Timothy W Nilsen
    Reviewing Editor; Case Western Reserve University, United States
  2. James L Manley
    Senior Editor; Columbia University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Inflammation Drives Alternative First Exon usage to Regulate Immune Genes including a New Iron Regulated Isoform of Aim2" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work cannot be considered further for publication in eLife at this time.

There was significant enthusiasm for the work. However, it seems that considerable effort including additional experiments will be required to firm up the conclusions and make the paper suitable for eLife. Accordingly we must reject the paper in its current form. Nevertheless, we encourage you to resubmit if and when you can address the majority of the reviewers' concerns. In this regard, addressing both functional relevance and mechanism would be ideal, but addressing one of these topics will be sufficient.

Reviewer #2:

Our understanding of the transcriptomic impact of innate immune signaling remains incomplete. Here Robinson et al., use both long and short read RNA sequencing to gain further insight into LPS-induced changes to mRNA isoform expression in human and mouse macrophages. Their studies report the novel observation that the most common change in isoform expression is alternative use of the first exon. Such changes are indicative of transcriptional regulation, and is thus consistent with the known impact of innate immune signaling on activation of multiple transcription factors. Despite some minor concerns with details of the study, this is a well-executed and important study that will be of interest and importance to many studying innate immunity, as well as those interested in gene regulation.

1. In some ways this is minor, but the authors should be careful to not describe alternative first exon use as alternative splicing. While a novel splice junction is created, mechanistically this is driven by changing transcriptional regulation, and then splicing occurs in the only pattern available to that TSS. In general this is described appropriately in the manuscript, but at a few points there is confusing terminology.

2. An interesting and somewhat surprising point in the manuscript is that 50% of the AFE events don't show an overall change in gene expression. For Aim2, which does change, the authors show that the AFE change is due to activated use of the unannotated TSS in LPS-stimulated cells. For those genes for which AFE use doesn't correlate with a change in gene expression (e.g. Ncoa7, Rcan1, Ampd3 – Figure S3) is there still transcriptional activation of one TSS and transcriptional silencing of the other? In other words, is there coordinated regulation of the two TSSs to ensure overall message abundance doesn't change, or does activation of one TSS inherently shut off the other (more akin to splice site competition in traditional AS)?

3. The data suggesting that an IRE regulates translation of the induced 5'UTR is compelling, but more work should be done to confirm. Most importantly, the experiment in Figure 4J should be repeated with the deltaIRE version of the unannotated UTR. Also is the IRE regulation controlled upon LPS-stimulation, or just the presence of the IRE element? In other words, what is the distribution of the annotated and unannotated isoforms in the polysome in the absence of LPS (i.e. repeat 4P without LPS)? Can the authors comment on whether the level of iron or the activity of IRP1/2 change in LPS-stimulated cells?

Reviewer #3:

This manuscript by Robinson et al. presents an interesting and timely analysis of a wealth of transcriptome data upon immune stimulation. The unique combination of long-read Oxford Nanopore and short-read Illumina high-throughput sequencing across both human and mouse samples presents an opportunity many interesting inter-species immune response comparisons, as well as elucidation of full-length transcript information. This paper is well-written and has interesting validation and discussions regarding Aim2. My major concern is that the paper seems to narrow in on the characterization of Aim2 and class of RNA processing changes (alternative first exons) quite quickly without really delving into the rest of the data and how they arrived there. Below are my major comments and suggestions:

1. I would have liked the authors to provide more insight into how they honed-in on specifically talking about first exon changes, by discussing more of the other RNA processing changes they found. There is cursory mention in the text and figures of other alternative exon or splice site changes. Firstly, other studies (including those referenced by the authors) have found hundreds RNA processing changes genome-wide upon immune stimulation – especially of cassette exons, alternative splice sites, and last exon/3'UTR changes. However here, the authors only find tens of changes (Figure 1B). Are they underpowered to identify changes and can they do any sort of analyses to show that they are sufficiently powered (# of sequencing reads and junctions, complexity of reads, etc)?

2. Similarly, I would also be interested in seeing an analysis indicating whether the 50 AFE events that overlap between the long-read and short-read sequencing analyses is a statistically significant overlap. Particularly, how many overlapping events would be expected given the difference in quantification power between the two methods? How many real AFE differences might the authors be missing because the long-read sequencing methods often do not have the power to identify them (ie. lower expressed genes in one or the other condition, thus dropout of isoforms and perhaps fewer isoform differences for differentially expressed genes).

3. Second, for the non-AFE changes that they did find, there is very little discussion about what those changes might represent. Specifically: (a) how many changes are validated with long-read data?, (b) is there any insight into specific domains being included/changed, especially using the long-read data?, (c) how many of these non-AFE changes overlap between species? and (d) which types of genes show higher overlap between species and what are their characteristics (binding sites, etc)? To my knowledge, this is the first study that is really designed to properly really look at the conservation of splicing or RNA processing changes after immune activation, so I would love to see more analysis and discussion of this aspect genome-wide.

4. The authors define significant splicing changes as those with a p-value <= 0.25 and |dPSI| >= 10. I'd like some more clarification on whether this is an adjusted p-value (BH, FDR, or some other multiple test-corrected p-value). Especially if this is adjusted, I find it surprising that the authors are choosing such a liberal statistical confidence level and that even with such a liberal threshold, they are only getting tens of significant events. I would like the authors to at least show these same trends across multiple p-value thresholds or with rank threshold analysis (top 5%, top 10%, top 20%) to show biological trends.

5. The authors introduce their long-read sequencing data by mentioning that they wanted to identify "additional splicing events that are not captured using short-read sequencing." They then go onto to only talk about novel first exon events identified with the long-read sequencing data. Did they identify any other non-AFE events in using the long-read that could then be quantified with the short read data? And second, how do they quantify confidence for novel AFE isoforms, when long-read data seems to have lots of issues with properly sequencing the terminal ends of transcripts (particularly the 5' end when polyA primed, as occurs in ONT DirectRNA sequencing)? They mention the use of ATAC-seq data to show putative promoter support, but mention at one point in their methods that ATAC regions within 10kb of AFEs are considered. These seems like it could be a rather large region to be sure that the ATAC peak is specific to a novel AFE – what is the average distance between AFEs? Finally, I would love to also see the incorporation of CAGE-seq data (or other 5'end data) to validate the specific AFEs sites – which I believe the FANTOM consortium has across many human and mouse tissues.

Reviewer #4:

In this manuscript, Robinson et al., identified alternative first exon (AFE) switching events conserved between mouse and human following macrophage inflammation. Using short and long-read sequencing, the authors identified a few unannotated transcription initiation sites (TSS) that are specific of an inflammatory response. Among those, they centered on an unannotated TSS in the Aim2 gene that drives expression of a novel isoform regulated by an iron-responsive element in its 5′UTR.

While previous work had documented crucial AFE switching events in many other biological contexts, Robinson et al. presents here an interesting AFE switching event that can have potential implications for our understanding of the molecular regulation of the innate immune response. For publication in eLife, I would expect further progress on global mechanisms and biological relevance of these AFE switching events, as well as evidence that the AFE are truly first exons/TSSs.

1. Are the AFEs truly first exons/TSS? While both short-read and long-read sequencing detected changes in alternative splicing choices, neither of those are optimal methodologies to analyze first exons. Therefore, I suggest to use a more specialized method to identify (and quantify) more accurately the usage of first exons. Globally, cap analysis of gene expression (CAGE) would be ideal. For validation of specific AFE changes, the qPCR technique has a few issues. First, it does not have nucleotide resolution, so the authors should not refer to TSSs if they used this technique for validation. Second, many downstream first exons are also used as internal exons in other isoforms. There is not a direct technology to analyze specifically first exons/TSSs here. Also, RNA-sequencing technologies, depending on their depth, can definitely miss specific isoforms. Considering a low coverage in 5'end of genes in RNA-seq analysis, this is particularly important for first exons. A qPCR would only analyze the well-known TSSs. Thus, 5'RACE or a similar technology should be perform to assess the relative usage of AFE specifically.

2. Global mechanism. The authors assumed that the mechanism of AFE switching is generated by transcription initiation and looked for transcription factors binding and chromatin structure modifications in promoters. However, they did not ruled out the possibility that the global switching effect is a post-transcriptional regulation, such as differential mRNA stability. A transcription initiation measurement (e.g., 4SU metabolic labelling) is necessary to demonstrate that the changes in AFE usage are co-transcriptional. In addition, in terms of their ATAC-Seq analysis, the chromatin structure changes in promoters can be cause or consequence of transcription initiation. Thus, it should not be listed as one mechanism driving the expression of AFE events (line 145). Also, to demonstrate a mechanism based on transcription factor binding more than 2 transcription factors should be considered. In any case, the expression patterns of the transcription factors considered are not clear. As a minor note, the bioinformatic analysis of the two promoters regions driving the isoforms of Aim2 (line 156) is not explained in the method section.

3. Biological relevance. Could the authors evaluate whether the translation regulation of Aim2 based on its AFE switching is a more generalized phenomenon? Are there any global gene regulation changes triggered by the other genes with significant changes is AFE usage?

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

Congratulations, we are pleased to inform you that your article, "Inflammation Drives Alternative First Exon usage to Regulate Immune Genes including a Novel Iron Regulated Isoform of Aim2", has been accepted for publication in eLife. Your revised article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Reviewer #1:

I appreciate the authors' efforts to address the issues of global mechanism and biological relevance, as well as other points raised. This new version of the manuscript is significantly stronger and it represents an important contribution to the literature. However, I still have concerns about the evidence the authors present to demonstrate that the AFE are "true" first exons/TSSs.

In my previous report, I suggested the authors use CAGE instead of RNA-seq (either short or long-reads) and 5'RACE instead of qPCR, both specific techniques to analyze first exons/TSSs. In this version, the authors used CAGE data from the FANTOM project to classify their first exon calls as "true" TSS. They found that only 45% of the novel TSS identified overlap with CAGE peaks. In my opinion, this raise the possibility that the other 55% are not "true" TSSs. Further, before publication, I would expect 5'RACE or any other direct method to validate that specific TSS are "true" TSS, at least for Denr, Arhgef7 and Aim2 which they validated using RT-qPCR.

Reviewer #2:

1. While the authors have now included a brief discussion about whether AFEs are regulated by transcription, splicing, or both in their discussion, I still think the authors should tone down the mechanistic implications of the language throughout the rest of the text, perhaps use the term "alternative RNA processing changes" instead of "alternative splicing changes"?

2. The authors mention that they implemented a new parameter in their FLAIR analysis to incorporate a bed file of ATAC annotations to calibrate the identification of novel AFEs. Do the authors plan to release this new parameter as an update to their existing software? I think this would be of great use to the community that analyzes long-read RNA data.

Reviewer #3:

The authors have fully addressed all of the concerns raised by adding additional data. This enhances the significance of the work and strengthens the conclusions.

https://doi.org/10.7554/eLife.69431.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #2:

Our understanding of the transcriptomic impact of innate immune signaling remains incomplete. Here Robinson et al., use both long and short read RNA sequencing to gain further insight into LPS-induced changes to mRNA isoform expression in human and mouse macrophages. Their studies report the novel observation that the most common change in isoform expression is alternative use of the first exon. Such changes are indicative of transcriptional regulation, and is thus consistent with the known impact of innate immune signaling on activation of multiple transcription factors. Despite some minor concerns with details of the study, this is a well-executed and important study that will be of interest and importance to many studying innate immunity, as well as those interested in gene regulation.

1. In some ways this is minor, but the authors should be careful to not describe alternative first exon use as alternative splicing. While a novel splice junction is created, mechanistically this is driven by changing transcriptional regulation, and then splicing occurs in the only pattern available to that TSS. In general this is described appropriately in the manuscript, but at a few points there is confusing terminology.

Many thanks for this discussion point. There are a couple of nuances that we would like to address and that we hope clarifies our use of language throughout this manuscript. We now clearly define that AFEs involve alternative TSS usage coupled with alternative splicing and thus changed the terminology to “Alternative Isoform Usage”, where more appropriate. Not all of the AFE events are only changed due to transcriptional regulation. In the literature, we do see evidence of alternative splicing coupled with a new TSS and upstream exon. A recent study by Fiszbein et al. challenges the notion that promoter choice is first, but rather the splicing of internal exons can impact promoter choice (PMID: 31787377). This study shows that internal exons can dictate which first exon is spliced in the mature RNA molecule. We hope that using “Alternative Isoform Usage” strikes a balance that is clearer in this version of the manuscript.

2. An interesting and somewhat surprising point in the manuscript is that 50% of the AFE events don't show an overall change in gene expression. For Aim2, which does change, the authors show that the AFE change is due to activated use of the unannotated TSS in LPS-stimulated cells. For those genes for which AFE use doesn't correlate with a change in gene expression (e.g. Ncoa7, Rcan1, Ampd3 – Figure S3) is there still transcriptional activation of one TSS and transcriptional silencing of the other? In other words, is there coordinated regulation of the two TSSs to ensure overall message abundance doesn't change, or does activation of one TSS inherently shut off the other (more akin to splice site competition in traditional AS)?

Thank you for your interest in the other AFE genes. We have now included further characterization information of inclusion and exclusion isoforms of Ncoa7, Rcan1, Ampd3, Denr and Arhgef7 in supplemental figure 11. We show that of the 5 candidates, only Denr is significantly differentially expressed, with a log2 fold change more than 2 or -2. While Ncoa7, Rcan1, Ampd3 and Arhgef7 are not significantly differentially expressed with a log2FC more than 2. However, more importantly, these normalized counts show that all genes are expressed with or without LPS stimulation. Furthermore, looking at figure 1 (G-H) it is clear that Arhgef7 exclusion TSS is shut off by LPS, while both the Denr isoforms, at low amounts, are expressed with or without LPS.

Supplemental figure 11G shows that Ncoa7 inclusion TSS isoform is solely expressed with LPS, 11H shows that Rcan1 exclusion TSS closes with LPS, and 11I shows Ampd3 has both isoforms expressed with and without LPS.

Now to take a step further to determine the potential mechanism of how the promoters for these AFE are regulated, we have utilized publicly available tracks tracks including ATACseq (GSE74191), PU.1 ChIPseq (GSE109965), p65 ChIPseq (GSE67343) and IRF3 ChIP seq (GSE67343). In supplemental figure 11F for Ncoa7, we cannot conclude a mechanism driving from inflammatory TFs since there are no p65 or IRF3 binding driving the inclusion isoform and the PU.1, myeloid specific transcription factor is most enriched to the exclusion isoform. Interestingly, The ATAC peak for the inclusion isoform for Ncoa7 is more open for the LPS BMDM sample, which could conclude that the promoter is opened by LPS. Rcan1, SupFig11H, shows ATAC opening by inflammation for the inclusion isoform, this promoter also shows enrichment of IRF3 for the inclusion isoform promoter. SupFig11I for Ampd3, it is clear that the inclusion isoform is driven by p65 and IRF3 binding, however the promoter of the exclusion is less clear since the ATACseq shows the promoter to be closed. Arhgef7, SupFig11J, shows a clear strong promoter for the novel inclusion isoform. This is shown by the strong PU.1 binding, ATACseq reads, as well as IRF3 and p65 binding. The exclusion isoform is open as shown by ATACseq reads and has PU.1 binding, but much less in comparison to the inclusion isoform. Finally, SupFig11K for Denr, shows a convoluted promoter region. These two isoforms TSSs are about 250bp away from one another, therefore the ATACseq reads, p65 and IRF3 binding are not distinct between promoters, but to give confidence to the promoter regions it is clear that there are unique promoters by examining the PU.1 binding in BMDMs +/-LPS. In conclusion the mechanisms governing TSS choice and isoform usage are complex and gene and context specific. (Manuscript Lines 199-201, 209-212, 363-366)

3. The data suggesting that an IRE regulates translation of the induced 5'UTR is compelling, but more work should be done to confirm. Most importantly, the experiment in Figure 4J should be repeated with the deltaIRE version of the unannotated UTR.

Thank you for the suggestion of this important experiment. We have generated a ΔIRE BMDM cell line data in supplemental figure 15, which supports our findings and strengthens our new mechanism of posttranscriptional regulation of Aim2. To generate a ΔIRE we used our lab’s previously published CRISPR/Cas9 immortalized BMDM cell line (PMID: 29051223) and gRNAs were chosen based off of the CRISPR/Cas9 target tracks on the genome browser (PMID: 27380939, https://genome.ucsc.edu/cgi-bin/hgTrackUi?hgsid=1041360567_lXenOcRdA3sNMjAw8PODB1MiRgWj&c=chr1&g=crisp).

After careful selection of 3 gRNAs, we electroporated them in our BMDM Cas9 cell line, and through immediate cellular dilution we generated a KO IRE cell line. Performing LPS time course experiments, similar to that of Fig4R-S, we determine that the Aim2 protein is regulated by this IRE motif. Our WT BMDM Cas9 cell line shows a loss of protein at the 48 hr LPS time point while our ΔIRE BMDM Cas9 cell line shows a significant increase in Aim2 protein at this time point. With these novel cell lines and data, we are confident in our proposed IRE mechanism of regulating

Aim2. (Manuscript Lines 278-288, 384-390).

Also is the IRE regulation controlled upon LPS-stimulation, or just the presence of the IRE element? In other words, what is the distribution of the annotated and unannotated isoforms in the polysome in the absence of LPS (i.e. repeat 4P without LPS)?

The post-transcriptional regulation is completely controlled by the IRE hairpin present in the novel isoform of Aim2. Figure 2D-H, clearly shows that the annotated Aim2 isoform expression is not regulated by inflammation, while the IRE containing unannotated isoform is highly upregulated during inflammation. Additionally, we have generated a new cell line that lacks the IRE hairpin, genomically, and the Aim2 protein is no longer regulated by LPS, shown in supplemental figure 15. Finally, in figure 4Q, we redid the polysome profiling experiment to include unstimulated macrophages. In these data, we see that the annotated transcript is still enriched in the high polysome fraction and the unannotated transcript is enriched in the low polysome fraction, and these data are not dependent on LPS. (Manuscript Lines 257-266)

Can the authors comment on whether the level of iron or the activity of IRP1/2 change in LPS-stimulated cells?

We have now added the expression of IRP1 and IRP2, found in supplemental figure 14, showing that they are not significantly differentially expressed by LPS in primary BMDMs. Therefore, the amount of IRP1 or IRP2 would likely not be affected by LPS. (Manuscript Lines 265-268).

Reviewer #3:

This manuscript by Robinson et al. presents an interesting and timely analysis of a wealth of transcriptome data upon immune stimulation. The unique combination of long-read Oxford Nanopore and short-read Illumina high-throughput sequencing across both human and mouse samples presents an opportunity many interesting inter-species immune response comparisons, as well as elucidation of full-length transcript information. This paper is well-written and has interesting validation and discussions regarding Aim2. My major concern is that the paper seems to narrow in on the characterization of Aim2 and class of RNA processing changes (alternative first exons) quite quickly without really delving into the rest of the data and how they arrived there. Below are my major comments and suggestions:

1. I would have liked the authors to provide more insight into how they honed-in on specifically talking about first exon changes, by discussing more of the other RNA processing changes they found. There is cursory mention in the text and figures of other alternative exon or splice site changes. Firstly, other studies (including those referenced by the authors) have found hundreds RNA processing changes genome-wide upon immune stimulation – especially of cassette exons, alternative splice sites, and last exon/3'UTR changes. However here, the authors only find tens of changes (Figure 1B). Are they underpowered to identify changes and can they do any sort of analyses to show that they are sufficiently powered (# of sequencing reads and junctions, complexity of reads, etc)?

Our lab was interested in the global alternative splicing profiles of primary macrophages during inflammation. To determine this, we utilized both primary human and mouse macrophages and analyzed them using both JuncBASE and DRIMSeq to determine the significant differential splicing events. It was clear that in both species Alternative First Exon (AFE) changes are the most prevalent event-type of the significant events. To ensure that our AFE signature is inflammatory specific, we reanalyzed the datasets generated by Pai et al. using methods mirroring those we used for analyzing our own data. Following this, in supplemental figure 4 A-D we see that the results show a higher prevalence of AFE events using either adjusted p-value thresholds of.05 or.25 (PMID: 27690314). On the other hand, in the Pai publication, their figure 2A and C shows the AFE exons are prevalent amongst the significant events in both experimental conditions, even more prevalent than cassette exons (PMID: 27690314). (Manuscript Lines 110-114).

While AFE events are not the most prevalent amongst all event types in their analysis, this may be due to other factors such as difference in methodologies applied. We focused on alternative isoform changes that are coupled with alternative splice site usage; therefore, we did not examine TandemUTR events. To define alternative splicing events, we used JuncBASE which has features that allows for defining alternative first exons with higher confidence (PMID: 26294686). For statistical testing, we used a t-test through the JuncBASE package or DRIMSeq, which applies a Dirichlet multinomial framework. On the other hand, Pai et al., evaluated changes in RNA processing events with MISO, which uses Bayesian inference in order to detect differentially expressed isoforms. MISO identifies differentially expressed isoforms based on annotation, while JuncBASE allows for the detection of unannotated splicing changes. In addition, significance was defined differently between both studies. In our method, significance was generally defined as AS events with a corrected p-value <= 0.25 and |ΔPSI| >= 10%. In Pai et al., significance is designated to events following infection were defined as those with at least at least 10% of individuals having a Bayes Factor (BF) > = 5 and the | mean ΔPSI| ≥ 5%. The smaller threshold of | mean ΔPSI| ≥ 5% attributes to the increase in the number of splicing changes reported in Pai et al. compared to those reported in our study.

2. Similarly, I would also be interested in seeing an analysis indicating whether the 50 AFE events that overlap between the long-read and short-read sequencing analyses is a statistically significant overlap. Particularly, how many overlapping events would be expected given the difference in quantification power between the two methods?

To address this comment, we have revised the manuscript (Manuscript Lines 133-147, 324-334) to more clearly indicate that an alternative splicing identification and quantification analysis was not completed using the long-read sequencing data. The long-read sequencing data was only used to assemble high confidence isoforms to supplement the existing gencode transcript annotation to more accurately identify alternative splicing events in the Mouse data with JuncBASE and DRIMSeq. Figure supplement 5 indicates overlap of significant alternative splicing events found using either the gencode annotation file or our ‘Gencode + de novo’ (Manuscript Lines 115-117).

How many real AFE differences might the authors be missing because the long-read sequencing methods often do not have the power to identify them (ie. lower expressed genes in one or the other condition, thus dropout of isoforms and perhaps fewer isoform differences for differentially expressed genes).

For our primary BMDM direct RNA sequencing, we had an average read depth of 1 million reads per condition (+/- LPS). In a previous study of direct RNA sequencing, we found that increased sequencing depth will result in additional isoforms detected; therefore, we expect to see additional unannotated AFE with increased sequencing depth (PMID: 31740818). We have made sure to include this point in our Discussion. (Line 125-136, 327-333).

3. Second, for the non-AFE changes that they did find, there is very little discussion about what those changes might represent. Specifically: (a) how many changes are validated with long-read data?

To address this point, we have now generated venn diagrams for all other splicing events (SFigure 7). The venn diagrams show the overlap of differential alternative splicing events from short-read RNA-seq data using either the publicly available gencode.gtf file or our combined gencode and de novo.gtf file generated from our direct RNA nanopore sequencing data. (Manuscript Lines 132-141, 333-334).

(b) Is there any insight into specific domains being included/changed, especially using the long-read data?

Looking deeper in the motifs of the mouse AFE genes, in SFigure 12 (line 232-234, 373-375), we see that overall there are more motifs in LPS (inclusion) first exons in comparison to the CTL (exclusion) first exons. Musashi binding element, MBE, was the most prevalent motif identified in the first exons. Because AFE events were the predominant type of alternative isoform usage, we focused on motif analysis for AFE events.

(c) How many of these non-AFE changes overlap between species?

We generated 8 new venn diagrams to identify the overlap between mouse and human alternative splicing events during acute inflammation (SFigure 5). Interestingly, we see that alternative first exons have the most conservation of all other splicing events. (Manuscript Lines 115-117)

(d) Which types of genes show higher overlap between species and what are their characteristics (binding sites, etc)? To my knowledge, this is the first study that is really designed to properly really look at the conservation of splicing or RNA processing changes after immune activation, so I would love to see more analysis and discussion of this aspect genome-wide.

We appreciate this inquiry, and hope to take a deeper look into this on our next project. The scope of this study was primarily focused on alternative first exons since they are the most prevalent event type across species after inflammatory activation. Our first step to take a global approach in assessing the exclusion and inclusion of AFE was to look at length. It is known that longer 5’UTRs lead to a decrease in translational efficiency of a transcript (PMID: 11897027); therefore we looked at the length of all inclusion and exclusion exons (SFigure 12A). There is no significant difference between the length of inclusion and exclusion exons. However, it is important to see that the average ΔAFE was 37, meaning there is a slight shift in a longer 5’UTR in the inclusion AFEs. Furthermore, we took a deeper approach to investigate the possible motifs in the mouse inclusion (LPS) and exclusion (CTL) exons (SFigure 12C) by using RegRNA2.0. This is a web server that takes an unbiased approach to scan for 44 potential murine motifs. We show that there are more motifs identified in inclusion (LPS) exons, and the most prominent motif is the musashi binding element (MBE). (Manuscript Lines 232-234, 372374).

4. The authors define significant splicing changes as those with a p-value <= 0.25 and |dPSI| >= 10. I'd like some more clarification on whether this is an adjusted p-value (BH, FDR, or some other multiple test-corrected p-value). Especially if this is adjusted, I find it surprising that the authors are choosing such a liberal statistical confidence level and that even with such a liberal threshold, they are only getting tens of significant events. I would like the authors to at least show these same trends across multiple p-value thresholds or with rank threshold analysis (top 5%, top 10%, top 20%) to show biological trends.

Adjusted p-values from JuncBASE are adjusted using Benjamini and Hochberg (BH) by implementing the option --mt_correction BH. The adjusted p-values from DRIMSeq are also adjusted using Benjamini and Hochberg (BH) by default (Nowicka, M, et al.). A 0.05 significance threshold is somewhat of an arbitrary significance threshold that has been designated in the field and an adjusted p-value < 0.25 has been used in a variety of other studies in genomics (e.g. PMID: 30124010, PMID:27818134, PMID: 16199517) We felt that a less stringent threshold of 0.25 would be appropriate to use in this study as the identification of alternative splicing events following LPS stimulation began as an exploratory analysis and the implementation of a 0.05 adjusted p-value threshold yielded few results. As requested by the reviewer we performed a rank threshold analysis of the top 5%, 10%, 15%, 20%, and 25% based on uncorrected p-values and they show that alternative first exon events remains amongst the most prevalent significant events (SFigure 2-3). (Manuscript Lines 107-114).

(5) The authors introduce their long-read sequencing data by mentioning that they wanted to identify "additional splicing events that are not captured using short-read sequencing." They then go onto to only talk about novel first exon events identified with the long-read sequencing data. Did they identify any other non-AFE events in using the long-read that could then be quantified with the short read data?

Thank you for asking this question. This is a very critical point, and we hope our corrections make this point clearer throughout the manuscript. With our study we first did a differential splicing analysis, using the publicly available gencode gtf file. We took this a step further, by then performing long-read direct RNA sequencing to generate a new macrophage specific gtf file using FLAIR. We then reanalyzed our short-read RNAseq data with our de novo transcriptome. From this new transcriptome file, we did find other differential splicing events that were novel to macrophages, as well as events that were identified in the short read sequencing that were supported by our new gft file. This data can be found in both Figure 1C, as well as in the venn diagrams of supplemental figure 7.

And second, how do they quantify confidence for novel AFE isoforms, when long-read data seems to have lots of issues with properly sequencing the terminal ends of transcripts (particularly the 5' end when polyA primed, as occurs in ONT DirectRNA sequencing)? They mention the use of ATAC-seq data to show putative promoter support, but mention at one point in their methods that ATAC regions within 10kb of AFEs are considered.

FLAIR collapse, the software tool used in order to assemble high confidence isoforms from nanopore reads, is designed to handle truncated reads (Tang, AD., et al. 2020). In addition, we implemented an optional parameter of FLAIR collapse (-p) which instructs the tool to use an input file of coordinates (a BED file) corresponding to promoter regions to aid with distinguishing isoform start sites. When implementing this parameter, we only used promoter regions that were also supported by ATAC-seq data. The ATAC regions within 10kb of AFEs are considered for a separate analysis looking into chromatin accessibility. Methods found in the manuscript lines 786-793.

These seems like it could be a rather large region to be sure that the ATAC peak is specific to a novel AFE – what is the average distance between AFEs? Finally, I would love to also see the incorporation of CAGE-seq data (or other 5'end data) to validate the specific AFEs sites – which I believe the FANTOM consortium has across many human and mouse tissues.

We appreciate the inquiry. To address this comment, we elected to use the

FANTOM5 CAGE data and complete two analyses. In both analyses, “known” TSS are those that overlap TSS in the Gencode M18 annotation and “novel” TSS are those that do not. In the first analysis, we plotted the distribution of CAGE scores of the CAGE peaks that overlap known and novel TSS. The similarity in the distribution between the known and novel CAGE score distribution indicates to us that the novel TSS identified using Nanopore sequencing coupled with the bioinformatics tool, FLAIR, are truly TSS. A second analysis to further support that claim was completed in order to observe the proportion of known and novel TSS that overlap CAGE data corresponding to “true” TSS as classified with machine learning methods (PMID: 21596820). 45% of the novel and ~62% of known TSS overlap with CAGE peaks corresponding to “true” TSS, respectively. Because FANTOM5 CAGE data is an aggregation of data collected from multiple studies using different cell-types and experimental conditions, we expect that many of the non-overlapping CAGE peaks classified as “true” TSS are a result of these experimental differences. (Main text: 141-153, 333, Methods: 825-842).

Reviewer #4:

In this manuscript, Robinson et al., identified alternative first exon (AFE) switching events conserved between mouse and human following macrophage inflammation. Using short and long-read sequencing, the authors identified a few unannotated transcription initiation sites (TSS) that are specific of an inflammatory response. Among those, they centered on an unannotated TSS in the Aim2 gene that drives expression of a novel isoform regulated by an iron-responsive element in its 5′UTR.

While previous work had documented crucial AFE switching events in many other biological contexts, Robinson et al. presents here an interesting AFE switching event that can have potential implications for our understanding of the molecular regulation of the innate immune response. For publication in eLife, I would expect further progress on global mechanisms and biological relevance of these AFE switching events, as well as evidence that the AFE are truly first exons/TSSs.

1. Are the AFEs truly first exons/TSS? While both short-read and long-read sequencing detected changes in alternative splicing choices, neither of those are optimal methodologies to analyze first exons. Therefore, I suggest to use a more specialized method to identify (and quantify) more accurately the usage of first exons. Globally, cap analysis of gene expression (CAGE) would be ideal. For validation of specific AFE changes, the qPCR technique has a few issues. First, it does not have nucleotide resolution, so the authors should not refer to TSSs if they used this technique for validation.

We appreciate the inquiry about the confidence of our isoforms that have novel TSS. To address this comment, we elected to use the FANTOM5 CAGE data and complete two analyses. In both analyses, “known” TSS are those that overlap TSS in the Gencode M18 annotation and “novel” TSS are those that do not. In the first analysis, we plotted the distribution of CAGE scores of the CAGE peaks that overlap known and novel TSS. The similarity in the distribution between the known and novel CAGE score distribution indicates to us that the novel TSS identified using Nanopore sequencing coupled with the bioinformatics tool, FLAIR, are truly TSS. A second analysis to further support that claim was completed in order to observe the proportion of known and novel TSS that overlap CAGE data corresponding to “true” TSS as classified with machine learning methods (PMID: 21596820). 45% of the novel and ~62% of known TSS overlap with CAGE peaks corresponding to “true” TSS, respectively. Because FANTOM5 CAGE data is an aggregation of data collected from multiple studies using different cell-types and experimental conditions, we expect that many of the non-overlapping CAGE peaks classified as “true” TSS are a result of these experimental differences. (Manuscript Lines 141-153, 333, Methods: 825-842).

Second, many downstream first exons are also used as internal exons in other isoforms. There is not a direct technology to analyze specifically first exons/TSSs here. Also, RNA-sequencing technologies, depending on their depth, can definitely miss specific isoforms. Considering a low coverage in 5'end of genes in RNA-seq analysis, this is particularly important for first exons. A qPCR would only analyze the well-known TSSs. Thus, 5'RACE or a similar technology should be perform to assess the relative usage of AFE specifically.

While it is true that there are limitations of these RNA sequencing technologies to determine the true first nucleotide, we still are able to determine nearly all splice junctions depending on read depth. Our illumina sequencing run was very deep, around 30million per library. Our nanopore sequencing libraries were much lower, at 1million reads per library, but they were used to generate a new transcriptome using FLAIR. The limitations of nanopore, as discovered by Mulroney et al. show that direct-RNA long read nanopore sequencing is unable to detect the first ~11nt of a transcript (DOI: 10.1101/2020.11.18.389049). Then we re-analyzed our illumina RNA-sequencing libraries using our new transcriptome. Thus the exact transcriptional start site is not critical in the definition of alternative first exon. We also used orthogonal ATAC-seq data to provide additional confidence that the first exon was at a putative promoter.

2. Global mechanism. The authors assumed that the mechanism of AFE switching is generated by transcription initiation and looked for transcription factors binding and chromatin structure modifications in promoters. However, they did not ruled out the possibility that the global switching effect is a post-transcriptional regulation, such as differential mRNA stability. A transcription initiation measurement (e.g., 4SU metabolic labelling) is necessary to demonstrate that the changes in AFE usage are co-transcriptional.

Thank you for this astute point. While we did not assess differential mRNA stability of the identified isoforms, we can confirm many of the events by using our ATAC-seq and ChIP-seq analysis. For instance, by examining the p65 and IRF3 transcription factors, we were able to show that 37 of the promoters are regulated by either p65 or IRF3. Additionally, 26 promoters of the AFE genes undergo chromatic accessibility changes during LPS stimulation (Figure 3). While 47 of the 95 AFE events are unaccounted for, this can be explained by our global assessment of promoters of the AFE events (Figure 3B), showing that many of the transcription factors enriched are regulated by metabolism and other biological processes (Line 203-205, 353-355). Assessing additional mechanisms that may be driving these AFE events we fell are outside the scope of this current study.

In addition, in terms of their ATAC-Seq analysis, the chromatin structure changes in promoters can be cause or consequence of transcription initiation. Thus, it should not be listed as one mechanism driving the expression of AFE events (line 145). Also, to demonstrate a mechanism based on transcription factor binding more than 2 transcription factors should be considered. In any case, the expression patterns of the transcription factors considered are not clear.

Thank you for this astute point, we have changed the text to address this concern. We investigated transcription factor binding as another potential contributing factor to the changes in AFE usage. The transcription factors we chose, p65 and IRF3, are two of the key inflammatory transcription factors driving thousands of immune response genes (described line 4346). Additionally, there are not many ChIP-seq datasets available for transcription factors of interest in primary BMDMs. The datasets chosen are for an inflammatory time course, 0-120min, with 120min chosen for analysis. This is the time point in which the highest binding of IRF3 or p65 is seen. While additional examination of other TFs could provide more mechanistic understanding we feel those studies are outside the scope of this paper. However, as an alternative approach to analyzing more ChIP-seq data from primary BMDMs, we performed a HOMER analysis of all AFE promoters (Figure 3B). We found that the majority of TFs are regulated by metabolism (Manuscript Lines 203-205, 353-355).

As a minor note, the bioinformatic analysis of the two promoters regions driving the isoforms of Aim2 (line 156) is not explained in the method section.

Thank you for this point, we have edited our methods to include the promoter analysis on Aim2 (Manuscript Lines 892-899).

3. Biological relevance. Could the authors evaluate whether the translation regulation of Aim2 based on its AFE switching is a more generalized phenomenon?

The general phenomenon of AFE usage has been seen previously in a handful of plant or Drosophila studies, but in the context of inflammation in mammalian cells is a novel finding (PMID: 17941993, PMID: 16344560, PMID: 19178699). For our results within this manuscript, we can definitively say that the AFE of Aim2 downregulates the protein expression through an IRE motif by our GFP expression experiments, polysome profiling experiments, western blot analysis of primary BMDMs and our Cas9 IRE KO experiments in immortalized BMDMs (Figure 4 and SFigure 15). (Line 374-389).

Furthermore, we took a deeper approach to investigate the possible motifs in the mouse inclusion (LPS) and exclusion (CTL) exons (SFigure 12C) by using RegRNA2.0. This is a web server that takes an unbiased approach to scan for 44 potential murine motifs. We show that there are more motifs identified in inclusion (LPS) exons, and the most prominent motif is the musashi binding element (MBE). (Line 232-234, 372-374) IRE motifs were not found in any other AFE event.

Are there any global gene regulation changes triggered by the other genes with significant changes is AFE usage?

After performing a global assessment of all inclusion and exclusion AFEs, using RegRNA2.0, we found that the inclusion AFE events had more motifs than the exclusion exons (SFigure 12, line 232-234). This could mean that there are additional post-transcriptional regulations affecting the translation of the genes during inflammation. It was beyond the scope of the paper to do another global sequencing experiment to assess gene regulation of the isoforms, such as polysome profiling or ribosome profiling. While our study does not look at the global translational implication of all AFE events, it does unearth a novel mechanism of an AFE event in the Aim2 protein coding gene. Aim2 was originally identified in a seminal 1997 study (PMID: 9242382) as an interferon inducible gene. From our work we now know that it is this alternative isoform that is inducible and it leads to a decrease in protein expression most likely as a way to control the pathway and return it to homeostasis.

https://doi.org/10.7554/eLife.69431.sa2

Article and author information

Author details

  1. Elektra K Robinson

    Department of Molecular, Cell and Developmental Biology, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing
    Contributed equally with
    Pratibha Jagannatha
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1427-3897

  2. Pratibha Jagannatha

    1. Department of Molecular, Cell and Developmental Biology, University of California Santa Cruz, Santa Cruz, United States
    2. Department of Biomolecular Engineering, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Conceptualization, Formal analysis, Methodology, software, Visualization, Writing – review and editing
    Contributed equally with
    Elektra K Robinson
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0257-3139

  3. Sergio Covarrubias

    Department of Molecular, Cell and Developmental Biology, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Conceptualization, Methodology
    Competing interests
    No competing interests declared

  4. Matthew Cattle

    Department of Biomolecular Engineering, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Methodology, software, Visualization
    Competing interests
    No competing interests declared

  5. Valeriya Smaliy

    Department of Molecular, Cell and Developmental Biology, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Validation
    Competing interests
    No competing interests declared

  6. Rojin Safavi

    Department of Biomolecular Engineering, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared

  7. Barbara Shapleigh

    Department of Molecular, Cell and Developmental Biology, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Methodology
    Competing interests
    No competing interests declared

  8. Robin Abu-Shumays

    Department of Biomolecular Engineering, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Methodology, Writing – review and editing
    Competing interests
    No competing interests declared

  9. Miten Jain

    Department of Biomolecular Engineering, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Methodology
    Competing interests
    No competing interests declared

  10. Suzanne M Cloonan

    Division of Pulmonary and Critical Care Medicine, Joan and Sanford I. Weill Department of Medicine, Weill Cornell Medicine, New York, United States
    Contribution
    supervision, Writing – review and editing
    Competing interests
    No competing interests declared

  11. Mark Akeson

    Department of Biomolecular Engineering, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Funding acquisition, Resources
    Competing interests
    holds options in Oxford Nanopore Technologies (ONT), is a paid consultant to ONT, received reimbursement for travel, accommodation and conference fees to speak at events organized by ONT,received research funding from ONT and is an inventor on 11 UC patents licensed to ONT (6,267,872, 6,465,193, 6,746,594, 6,936,433, 7,060,50, 8,500,982, 8,679,747, 9,481,908, 9,797,013, 10,059,988, and 10,081,835)

  12. Angela N Brooks

    Department of Biomolecular Engineering, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Writing – review and editing
    For correspondence
    anbrooks@ucsc.edu
    Competing interests
    received reimbursement for travel, accommodation and conference fees to speak at events organized by Oxford Nanopore Technologies (ONT)
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7898-3073

  13. Susan Carpenter

    Department of Molecular, Cell and Developmental Biology, University of California Santa Cruz, Santa Cruz, United States
    Contribution
    Funding acquisition, Project administration, supervision, Writing – review and editing
    For correspondence
    sucarpen@ucsc.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5600-5404

Funding

NIH Office of the Director (HG010053)

  • Angela N Brooks
  • Mark Akeson

Oxford Nanopore Technologies (SC20130149)

  • Mark Akeson

University of California, Santa Cruz (Committee on Research)

  • Angela N Brooks
  • Susan Carpenter

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

Acknowledgements

We thank Eric Martin for sharing his time and knowledge on HOMER. We would also like to thank Alison Tang for sharing her knowledge and assisting with FLAIR isoform analysis. We also thank Dr. Vijay Rathinam and Dr. Kate Fitzgerald for their critical insights into this manuscript. Finally, we would like to thank Kevin S Johnson and Dr. Karen Ottemann who provided reagents for the IRE mutagenesis experiments. Funding for this work was partially supported by a Special Research Grant/Collaborative Research Grant from the UCSC Committee on Research (COR) to ANB and SCa. Additional funding support from NIH HG010053 (ANB and MA) and Oxford Nanopore Research Grant SC20130149 (MA).

Ethics

All animal work was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee at the University of California Santa Cruz (Protocol number CARPS1810).

Senior Editor

  1. James L Manley, Columbia University, United States

Reviewing Editor

  1. Timothy W Nilsen, Case Western Reserve University, United States

Publication history

  1. Received: April 15, 2021
  2. Accepted: May 21, 2021
  3. Accepted Manuscript published: May 28, 2021 (version 1)
  4. Version of Record published: July 6, 2021 (version 2)

Copyright

© 2021, Robinson 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

  • 2,893
    Page views
  • 338
    Downloads
  • 7
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, 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. Elektra K Robinson
  2. Pratibha Jagannatha
  3. Sergio Covarrubias
  4. Matthew Cattle
  5. Valeriya Smaliy
  6. Rojin Safavi
  7. Barbara Shapleigh
  8. Robin Abu-Shumays
  9. Miten Jain
  10. Suzanne M Cloonan
  11. Mark Akeson
  12. Angela N Brooks
  13. Susan Carpenter
(2021)
Inflammation drives alternative first exon usage to regulate immune genes including a novel iron-regulated isoform of Aim2
eLife 10:e69431.
https://doi.org/10.7554/eLife.69431

Categories and tags

Research organism

Further reading

    1. Evolutionary Biology
    2. Genetics and Genomics

    The lingering effects of Neanderthal introgression on human complex traits

    Xinzhu Wei, Christopher R Robles ... Sriram Sankararaman
    Research Article

    The genetic variants introduced into the ancestors of modern humans from interbreeding with Neanderthals have been suggested to contribute an unexpected extent to complex human traits. However, testing this hypothesis has been challenging due to the idiosyncratic population genetic properties of introgressed variants. We developed rigorous methods to assess the contribution of introgressed Neanderthal variants to heritable trait variation relative to that of modern human variants. We applied these methods to analyze 235,592 introgressed Neanderthal variants and 96 distinct phenotypes measured in about 300,000 unrelated white British individuals in the UK Biobank. Introgressed Neanderthal variants have a significant contribution to trait variation consistent with the polygenic architecture of complex phenotypes (contributing 0.12% of heritable variation averaged across phenotypes). However, the contribution of introgressed variants tends to be significantly depleted relative to modern human variants matched for allele frequency and linkage disequilibrium (about 59% depletion on average), consistent with purifying selection on introgressed variants. Different from previous studies (McArthur 2021), we find no evidence for elevated heritability across the phenotypes examined. We identified 348 independent significant associations of introgressed Neanderthal variants with 64 phenotypes . Previous work (Skov 2020) has suggested that a majority of such associations are likely driven by statistical association with nearby modern human variants that are the true causal variants. We therefore developed a customized statistical fine-mapping methodology for introgressed variants that led us to identify 112 regions (at a false discovery proportion of 16%) across 47 phenotypes containing 4,303 unique genetic variants where introgressed variants are highly likely to have a phenotypic effect. Examination of these variants reveal their substantial impact on genes that are important for the immune system, development, and metabolism. Our results provide the first rigorous basis for understanding how Neanderthal introgression modulates complex trait variation in present-day humans.

    1. Genetics and Genomics

    Native functions of short tandem repeats

    Shannon E Wright, Peter K Todd
    Review Article

    Over a third of the human genome is comprised of repetitive sequences, including more than a million short tandem repeats (STRs). While studies of the pathologic consequences of repeat expansions that cause syndromic human diseases are extensive, the potential native functions of STRs are often ignored. Here, we summarize a growing body of research into the normal biological functions for repetitive elements across the genome, with a particular focus on the roles of STRs in regulating gene expression. We propose reconceptualizing the pathogenic consequences of repeat expansions as aberrancies in normal gene regulation. From this altered viewpoint, we predict that future work will reveal broader roles for STRs in neuronal function and as risk alleles for more common human neurological diseases.

    1. Cell Biology
    2. Genetics and Genomics

    Identification of a weight loss-associated causal eQTL in MTIF3 and the effects of MTIF3 deficiency on human adipocyte function

    Mi Huang, Daniel Coral ... Sebastian Kalamajski
    Research Article Updated

    Genetic variation at the MTIF3 (Mitochondrial Translational Initiation Factor 3) locus has been robustly associated with obesity in humans, but the functional basis behind this association is not known. Here, we applied luciferase reporter assay to map potential functional variants in the haplotype block tagged by rs1885988 and used CRISPR-Cas9 to edit the potential functional variants to confirm the regulatory effects on MTIF3 expression. We further conducted functional studies on MTIF3-deficient differentiated human white adipocyte cell line (hWAs-iCas9), generated through inducible expression of CRISPR-Cas9 combined with delivery of synthetic MTIF3-targeting guide RNA. We demonstrate that rs67785913-centered DNA fragment (in LD with rs1885988, r2 > 0.8) enhances transcription in a luciferase reporter assay, and CRISPR-Cas9-edited rs67785913 CTCT cells show significantly higher MTIF3 expression than rs67785913 CT cells. Perturbed MTIF3 expression led to reduced mitochondrial respiration and endogenous fatty acid oxidation, as well as altered expression of mitochondrial DNA-encoded genes and proteins, and disturbed mitochondrial OXPHOS complex assembly. Furthermore, after glucose restriction, the MTIF3 knockout cells retained more triglycerides than control cells. This study demonstrates an adipocyte function-specific role of MTIF3, which originates in the maintenance of mitochondrial function, providing potential explanations for why MTIF3 genetic variation at rs67785913 is associated with body corpulence and response to weight loss interventions.