RBM39 shapes innate immunity through transcriptional and splicing control of key factors of the interferon response

Article activity feed

1. 

   Review Commons

   Jul 18, 2024

   Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Reply to the reviewers

   1. General Statements

   We are grateful for the valuable, constructive comments of the reviewers, which helped to substantially improve the quality of our manuscript. We particularly agree that the original structure of the manuscript was confusing and in parts misleading, since we followed the history of the project, which first identified the RBM39 mediated impact on IRF3 expression, whereas the -omics studies, identifying additional factors, were done at a far later point. Many discrepancies further arose from the low sensitivity of our initial proteomics analysis, which we now repeated, thereby obtaining far more sensitive detection of the key factors …

   Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Reply to the reviewers

   1. General Statements

   We are grateful for the valuable, constructive comments of the reviewers, which helped to substantially improve the quality of our manuscript. We particularly agree that the original structure of the manuscript was confusing and in parts misleading, since we followed the history of the project, which first identified the RBM39 mediated impact on IRF3 expression, whereas the -omics studies, identifying additional factors, were done at a far later point. Many discrepancies further arose from the low sensitivity of our initial proteomics analysis, which we now repeated, thereby obtaining far more sensitive detection of the key factors we also found in the transcriptomics data.

   We have re-structured the entire manuscript by moving the -omics data from the end of the paper towards the middle and provide similar depth downstream analysis of all relevant key factors identified (RIG-I/MDA5, IFN receptors, STAT1/2), to reduce the focus on IRF3, as suggested. We further changed the title and abstract to reflect this major conceptual change. Thanks to this helpful comment, we think that our manuscript is now conceptually much clearer.

   We further added new data to support the central claims of our manuscript, including a repetition of the proteomics study. Proteomics and transcriptomics now consistently demonstrate the impact of RMB39 knockdown as well as indisulam treatment on several key factors of innate immunity, including IRF3, STAT1/2, RIG-I and MDA5 (now in Fig. 5), with IFNAR2 and IL10RB additionally found in transcriptomics. We provide additional functional evidence that IRF3 is the key factor affected in the TLR3 pathway (IRF3 overexpression, Fig. 6B, C), whereas diminished abundance of RIG-I/MAD5 is equally important in the respective pathway, thereby also affecting NF-κB response (Fig. 6F-I). We further show the functional significance of IFN-receptor/STAT downregulation on type I and III IFN responses (Fig. 7E-G).

   The reviewers also pointed to some datasets showing the expected trends, but in some cases lacking statistical significance, due to variability in knockdown efficiency. We repeated all mentioned datasets with new batches of siRNA with sufficient biological replicates (n=3). We thereby obtained consistent, statistically significant data in all cases. Importantly, all experiments implementing the RMB39.esc control now show consistent rescue (Fig2. A-E).

   To generate a homogenous experimental design for virus infections, we further added new data showing a comparable impact of siRNA knockdown (Fig. 3F) and indisulam treatment (new Fig. 3G) on Sendai virus infection in A549 cells and took this as a rationale to consistently use indisulam for all other infections.

   2. Point-by-point description of the revisions

   __Reviewer #1 (Evidence, reproducibility and clarity (Required)): __ This manuscript by Li and colleagues examines the role of RBM39 in innate immune signaling. Splicing factor RBM39 was identified through a genome wide screen with a death reporter under control of the IFIT1 promoter that got stimulated with pIC in a TLR3-dependent manner. Besides IFIT1, further experiments showed that RBM39 is also involved in optimal expression of other innate immunity genes like IFNB, CXCL10, RIG-I or MDA5. While NFkB-dependent genes seem not to depend on RBM39, for IRF3 it was shown that protein levels decrease under conditions of RBM39 depletion, because IRF3 mRNAs are (slightly) reduced and spliced differently. The sulfonamid Indisulam could largely recapitulate the phenotype of RBM39 depletion. Further analyses using proteomics and transcriptomics showed that RBM39 is required for mRNA splicing and expression of a large set of other proteins. Altogether, this well designed and written study highlights the fundamental role played by RBM39 in in maintaining the pathways of immunity and metabolism. The key conclusions are convincing but some additional experiments would strengthen them further.

   We are grateful for the very positive general comments of this reviewer.

   Major comments:

   • For the statistics, authors seem not to have done multiple tests but rather tested individual datasets within larger graphs against each other. Please explain where this is the case and use corrections if multiple testing was done

   We apologize for not have been clearer here, we indeed used multiple testing. In the proteomics, statistical significance was evaluated by "two-sample tests" (Student's T-test with permutation-based FDR 0.05 and 250 number of randomizations). For the analysis of RNAseq data, p values were calculated with the Wald test and corrected for multiple testing according to Benjamini-Hochberg. We have now included this information in the materials and methods section and in the respective figure legends.

   • Fig. 4 shows that RBM39 depletion reduces IFIT expression in virus infected cells and slightly increases virus replication. RBM39 has a major effect on IRF3 levels, but also on other players in innate immunity. What happens if IRF3 is ectopically expressed as in figure 5? With this experiment one could measure how high the contribution of IRF3 miss-splicing is to innate immunity.

   We thank this reviewer for the valuable suggestion. We restructured the entire manuscript, to address several reviewer comments regarding the focus on IRF3 and the lack of data on other factors in the pathway. We now clearly demonstrate that ectopic IRF3 expression entirely rescues the TLR3 response to poly(I:C) in PH5CH cells (Fig. 6B-C), which also explains the lack of impact on the NF-κB pathway (Fig. 2G-H). In contrast, overexpression of IRF3 does not rescue the RIG-I/MDA5 response in A549 cells (new data, Fig. 6F-I). Here, also the NF-κB pathway is affected by knockdown of RBM39, suggesting that reduced RIG-I/MDA5 abundance upon RMB39 knockdown substantially contributed to the diminished innate immune response.

   • Fig. 4 A uses siRNAs but B, C and D only indisulam treatment. It would be better if siRNAs would also be used for the other viruses.

   We agree that a homogenous setup for virus infection would be favorable, however, the use of different cell lines was authorative due to limited permissivess of the used cell types towards virus infection and it appeared challenging to achieve similar knockdown efficiencies. To generate a homogenous experimental design, we now added new data showing a comparable impact of siRNA knockdown (Fig. 3F) and indisulam treatment (new Fig. 3G) on Sendai virus infection in A549 cells and took this as a rationale to consistently use indisulam for all other infections.

   • RBM39 depletion strongly reduces IRF3 levels in the WB, but not so much in RT-PCR and not at all in proteomics. Is the antibody used for WB perhaps recognizing a domain that is underrepresented in isoforms after disturbed splicing? Please clarify.

   Our previous proteomics data suffered from a very low sensitivity, therefore we missed clear detection of many factors, including IRF3. We repeated the whole proteomics analysis with siRNA and indisulam treatment (new Fig. 5A, B) and now found significantly reduced IRF3 protein levels in both conditions (new Fig. S5C), in agreement with the WB data. The lower impact on IRF3 mRNA abundance is due to the additional contribution of alternative splicing (Fig. 6A, Fig. S6A-D), which both in combination affect protein abundance.

   • Volcano plots in figure 7 show a lot of hits obtained after both RBM38 siRNA and indisulam (green dots), and some that are additionally identified in transcriptomes and in proteomes (red dots). Nonetheless only innate immunity and stress response genes are marked, although they do not belong to these highly conserved classes. Please elaborate more on the most RBM39-dependent genes, e.g. by presenting them in a heat map.

   To our knowledge, our study is the first with a comprehensive comparison on the impact of RBM39 knockdown and indisulam treatment on the host cell proteome and transcriptome. However, several studies already did -omics studies on individual conditions/readouts (e.g. (Coomar* et al, 2023; Dou et al, 2023; Mai et al, 2016; Nijhuis et al*, 2022)). These studies already identified and described in detail key changes in transcriptome and proteome e.g. affecting genes involved in cell cycle control and metabolism, which we find as well. However, the novelty of our paper is the impact on innate immune response, we therefore rather decided to put an even stronger focus on these genes and to omit other factors, like stress response pathway components, etc.. This strategy is supported by the higher sensitivity of our new proteome analysis, which now generated a far better overlap with the transcriptomics, favoring a display setting on highlighting only those factors that were further analyzed in detail in the volcano blots (Fig. 5). Still, interested readers will find the comprehensive list of data in the supplementary Excel-datasheets as well as in our primary data in online depositories.

   Minor comments:

   • Some abbreviations are not explained, like PGK, siNT, siVTN

   We apologize and have added the missing explanation of abbreviations.

   • Welsch should read Welch

   Corrected.

   • Fig. 2H: were cells also stimulated and if yes, how?

   These were unstimulated conditions, to show the impact of RBM39 on basal expression of the IFNlambda receptor chains. However, we deleted this dataset due to the re-organisation of the manuscript. The analysis of the type I and type III receptor and STAT1/2 expression is now comprehensively shown in Fig. 7/S6E, F, solely based on the transcriptomic data for consistency reasons, along with the functional impact on the IFN response.

   • Fig. 6E: I cannot see a difference between to IRF3-203 and 228 isoforms. And what are the white boxes?

   • Also 6E: Location of the primers is barely visible

   Due to the re-organization of the manuscript these data are now shown in Fig. S6D. Both isoforms are indeed very similar and only differ by a very small (16nt) additional exon in isoform 228. The white boxes are exons not translated in the respective isoforms. We have included this important information in the legend to Fig. S6 and increased the arrows indicating the positions of the primer.

   • Some materials are not properly referenced, like the death reporter, the lentiviral system, or the Rift Valley fever luciferase virus

   We are sorry for the missing information, which has now been added to the materials and methods section.

   • Supplement has no page numbers

   We have added page numbers to the supplementary information.

   Reviewer #1 (Significance (Required)):

   The study advances our knowledge about the regulation of innate immunity. Strengths are the discovery of a novel layer of innate immunity regulation by splicing and the in-depth analysis of the importance of RBM39 for cellular gene expression. A potential weakness might be the focus on innate immunity as other biological functions seem even more dependent on RBM39. However, this reviewer sees the necessity that covering all aspects of RBM39 finction would be beyond the scope of a single study. The relevant literature is appropriately cited (except for some materials, see minor comments). Results will be of interest not only to people doing basic research on innate immunity, but also to those interested in gene regulation in general or to cancer researchers using indisulam

   __Reviewer #2 (Evidence, reproducibility and clarity (Required)): __ The authors performed a CRISPR-based screen for genes required for TLR3-mediated signaling and gene expression in Hepatoma cells. Interferon-stimulated expression of an apoptosis inducer was used as a read-out system. A number of candidate genes were identified and one of these, RBM39, investigated in detail. The protein has previously been linked to both transcriptional control and RNA processing. Validation studies confirm that reduction of cellular RBM39 results in less TLR3-mediated IFN-beta synthesis and lower levels of ISG mRNA synthesis. Initial studies suggest a role of RBM39 in regulating of IRF3 levels, the transcription factor activated by TLR3 signaling to induce IFN-beta synthesis. However, the effect is variable and poorly supported by transcriptomic and proteomic data. Moreover, only one out of four cell-based viral infection models reports a substantial effect of the RBM39 knockdown.

   We apologize for the lack of consistency among several datasets, which was mainly due to the low sensitivity of the proteomic analysis. This has been repeated and now fully confirms all other data. In part due to the comments of this reviewer, we further broadened the scope of the manuscript away from IRF3, including a change of the title.

   Major comments:

   1. The data do not support the claim that RBM39 is a broadly acting player in innate immune responses. In addition, they suggest that IRF3 may not be the only relevant RBM39 target. The most informative knockdown control in this regard would be IRF3 siRNA.

   We have re-structured the entire manuscript and added new data to support the central claims of our manuscript, including a repetition of the proteomics study. Proteomics and transcriptomics now consistently demonstrate the impact of RMB39 knockdown as well as indisulam treatment on several key factors of innate immunity, including IRF3, STAT1/2, RIG-I and MDA5 (now in Fig. 5), with IFNAR2 and IL10RB additionally found in transcriptomics. We further provide functional evidence that IRF3 is the key factor affected in the TLR3 pathway (IRF3 overexpression, Fig. 6B, C), whereas diminished abundance of RIG-I/MAD5 is equally important in the respective pathway, thereby also affecting NF-κB response (Fig. 6F-I). We further show the functional significance of IFN-receptor/STAT downregulation on type I and III IFN responses (Fig. 7E-G). We hope this reviewer now agrees with our claim that RBM39 is a broadly acting player in innate immune responses.

   1. The structure of the manuscript is rather confusing because IRF3 is presented as the main RBM39 target in figures 3-6, but the -omics data in figures 7 and 8 do not support this view. The authors argue different sensitivities of the experimental approaches, but I think few people would agree that western blots are more sensitive than MS. To my opinion a narrative with less focus on IRF3 and a broader integration of candidates of the -omics approaches would be preferable.

   We are grateful for this valuable comment and fully agree that the original structure of the manuscript was confusing and in parts misleading, which was mainly due to the fact that we followed the history of the project, which first identified the RBM39 mediated impact on IRF3 expression, whereas the -omics studies, identifying additional factors, were done at a far later point. Many discrepancies further arose from the low sensitivity of our proteomics analysis, which we now repeated, thereby obtaining far more sensitive detection of the key factors we also found in the transcriptomics data. We now moved the -omics data from the end of the paper towards the middle and provide similar depth downstream analysis of all relevant key factors identified (RIG-I/MDA5, IFN receptors, STAT1/2, to reduce the focus on IRF3, as suggested. We further changed the title and abstract to reflect this major conceptual change. Thanks to this helpful comment, we think that our manuscript is now conceptually much clearer.

   Investigating the role of RBM39 by RNA-seq in pIC-treated cells would further strengthen the manuscript. It will yield a broader view of the protein's role in induced innate immunity.

   We did not add pIC treatment to the RNA-seq analysis, since, based on own experience and numerous papers, this will change the expression of literally thousands of genes. Based on the key factors of the pIC response modulated by RBM39 (RLRs and IRF3), this would very likely simply result in reduced induction of the whole ISG panel (as exemplified for IFIT1, ISG15, MxA and CXCL10 in Fig. 2B-E).

   3.The results in figures 6A-C are confusing for two reasons. First, the siRNA-mediated knockdown should result in reduced RBM39 protein as well (as shown in Fig. 3A) and, therefore, in an increase in RBM39 levels. Second, why was this effect not noted in the experiments shown in figs. 1-5? To avoid this confusion it might be good to mention which IRF3 splice isoforms are detected by the primers and antibodies used in these figures.

   Unfortunately, the reviewer seems to have conceptually misinterpreted Fig. 6A-C of the original paper, which did not show protein, but transcriptome data. We now added the corresponding data of the proteomic analysis in the new Fig. S5, for all detectable, relevant candidates, showing consistency to all previous data. The confusing point in previous Fig. 6B, which the reviewer appears to refer to, is the upregulation of RBM39 transcript levels upon indisulam treatment, which was not apparent in previous experiments, since we always used WB to show diminished RBM39 protein levels upon indisulam treatment. This increase in RBM39 mRNA is due to an autoregulation of RBM39 mRNA by protein abundance, which has been reported in literature (Campagne* et al*, 2023). Since this is rather confusing and not relevant for our study, we removed previous Fig. 6B and show this aspect only in the volcano blot in Fig. 5D, mentioning and citing the paper on autoregulation.

   Minor comments.

   1. Fig S1: the figure panels and legend are inconsistent. IFIT1 is labeled as ISG56 in panel S1A.

   We apologie for this inconsistency and now use IFIT1 throughout the paper.

   1. Data with the siRNA escape mutant of RBM39 are inconsistent. For example, why is its effect significantly different only in 1 out of 4 ISG in figures S2A-D?

   We apologize for the inconsistency, which is due to variability of silencing efficiency. We repeated the entire set of experiments (n=3) with a new batch of siRNA and obtained comparable, significant differences for all ISGs analyzed (new Fig. 2B-E).

   1. Line 164: the statement that TRIF and RBM39 siRNAs produce effects of similar magnitude is incorrect for the IFIT1 gene in figure S2A.

   This experiment was repeated (see previous point), now obtaining significant, more homogenous data. We have modified the text accordingly.

   4.Fig. 2H: In absence of additional evidence for functional implications, the data showing reduced IL10RB expression should be omitted.

   We omitted the data, as suggested by the reviewer, however, we provide a more in depth analysis of the type I and III IFN response in Fig. 7, based on the transcriptomic data and a functional analysis.

   5.Fig. 3: More datapoints would be needed in panel A to sustain the lack of significant difference between the untreated and escape mutant samples. Are the viability data in panels B and C normalized to untreated cells to control for Indisulam toxicity? In figure S3A the effect of the mutant is rather small. To allow for comparison, the Indisulam titration curves should be adapted to the concentrations used in Fig. 3.

   Fig. 3 (now Fig. 4) was replaced by another representative experiment, now also containing the quantification of the shown western blots, however, the statistical analysis shown in the previous version was and is based on three independent biological replicates, as indicated in the figure legend. Viability data was normalized to controls and this information is now added to the figure lengend as well. The mutant analyzed in Fig. S3A (now S4A) confers only partial resistance, which explains the limited but clear rescue. We did not include higher indisulam concentrations here due to the increased cytotoxicity of concentration above 5 µM in PH5CH, in the absence of pronounced additional effects on RBM39 abundance (Fig. 4B).

   6.RNA-seq measures steady-state RNA, not transcription.

   This is of course correct, we changed all sentences, where our wording might have indicated that we are measuring transcription by RNAseq. However, we still need to differentiate between the role of RBM39 in transcriptional regulation and splicing, where changes in RNA abundance found in RNAseq rather point to transcriptional regulation.

   Reviewer #2 (Significance (Required)):

   The identification of RBM39 as a candidate player in innate immune responses is of interest to a large scientific community with interest in signalling by pattern recognition receptors. Its role should be strengthened with additional infection models. It is puzzling that three out of four viruses don't benefit from the reduced IFN-beta synthesis in the RBM39 knockdown. Moreover, the data are not convincing (or too diverse) to nail down IRF3 as a major, or the most relevant, RBM39 target.

   __Reviewer #3 (Evidence, reproducibility and clarity (Required)): __ CRISPR Screen for factors that are required for dsRNA-dependent ISG production. Found a large number of hits but most did not validate in subsequent assays. The authors follow up the one candidate that did pass secondary screening criteria, RBM39, although re-expression of RBM39 only rescues the phenotype of the siRNAs against RBM39 (siRBM39) in one of the two cell lines tested. Additionally, siRBM39 impacts only a subset of polyIC-induced ISGs and does not regulate NFkB-driven gene expression. They go on to attempt to investigate the impact of siRBM39 on other key innate immune genes and proteins, although many key controls and appropriate methods are missing.

   We thank this reviewer for pointing at inconsistencies and missing controls in our manuscript. We have critically re-evaluated the respective datasets.

   Major comments:

   1. The authors propose some rationale for the limited success of the screen, however, while RBM39 may have a role in dsRNA-induced innate immunity, in general the screen seems to have limited value.

   The aim of our CRISPR/Cas9 death reporter screen was the identification of so far unknown contributors to innate immune response. This was achieved by identifying a critical role of RBM39, followed by an in depth validation focusing on RBM39. We further found known components of the TLR3 pathway in our candidate list (e.g. TRIF and UNC93B1), pointing to the overall quality of the experimental setup. At no point of the manuscript we claim that our screen aimed for or delivered a comprehensive overview on innate immunity pathways. Honestly, no comparable screen (e.g. on cytopathic viruses) has delivered such data.

   1. Given that the siRBM39 clearly has off-target effects (since expression of a resistant RBM39 cDNA only gives limited rescue in many cases - Fig S2), each of the experiments in which siRBM39 is used (i.e. Fig 2) should have the RBM39.esc control - especially those that drive subsequent experiments such as the expression of IFNbeta and IFNLR1 (Fig 2a, h)

   The inconsistency in some datasets, showing all the same trends, but in some cases lacking statistical significance was due to variability in knockdown efficiency. We repeated all mentioned datasets with new batches of siRNA with sufficient biological replicates (n=3) with now all of them revealing consistent, statistically significant data. Importantly, all experiments implementing the RMB39.esc control now show consistent rescue.

   1. Since RBM39 reduction has an apparent impact even if IFNLR1-deficient cells (although need the rescue control to know if this is real) the authors conclude that RBM39 regulates the initial wave of dsRNA signaling-events, but this should be tested with the use of Ruxilitinib to block JAK-STAT signaling.

   Due to the general major re-organization of the manuscript, aiming for a less confusing data presentation and consistency towards depth of candidate evaluation, we have removed the data on the IFNLR-deficient cell line. The claim that RBM39 affects the initial wave of ISG responses is based on reduced IFNb expression, which is exclusively induced by the initial wave of ISG response and by the general impact on ISG expression, which we measure at 6h after induction, too early for autocrine IFN stimulation (Burkart* et al*, 2023). However, we further demonstrate that downregulation of type I and type III IFN receptors in conjunction with STAT1/2 affect the type I and the type III IFN response as well (Fig. 7E-G, in part new data). Therefore, RBM39 affects both, the intial wave and the auto-/paracrine IFN response, and we therefore undertook no further efforts to separate these effects.

   1. IRF3 expression in the Indisulam-treated cells more closely tracks cell viability than RBM39 expression. For example in Fig 3C 10 microM gives 50% IRF3 expression and 50% viability but still 95% RBB39 expression - arguing that the impact of siRBM39 on IRF3 might be very indirect (and error bars on rescue are large so unclear if the rescue really worked in Fig 3A).

   Based on this reviewer comment we re-evaluated the quantification in previous Fig. 3C (now Fig. 4C), which combines data from three independent experiments. We deeply apologize, but the initial quantification proved to be wrong, due erroneous background subtraction, which was relatively high in one of the PHH-replicates (Replicate 1, see Reviewer Fig. 1 in uploaded file). The re-evaluated quantification revealed 55% for the RBM39 abundance at 10µM indisulam, which better reflects the data shown and is now in line with the impact on cytotoxicity and IRF3 abundance.

   1. It is unclear in Fig 4 why some cell/virus combinations are tested with siRBM39 and others are tested with Indisulam. Also the conclusion that RBM39 "substantially contributes to the cell intrinsic innate immune response to viral infections" is greatly overstated given that the differences are between ~3 fold and non-significant.

   We agree that a homogenous setup for virus infection would be favorable, however, the use of different cell lines was authoritave due to limited permissivess of the used cell types towards virus infection and it appeared challenging to achieve similar knockdown efficiencies. To generate a homogenous experimental design, we now added new data showing a comparable impact of siRNA knockdown (Fig. 3F) and indisulam treatment (new Fig. 3G) on Sendai virus infection in A549 cells and took this as a rationale to consistently use indisulam for all other infections. Overall, the aim of the virus infection experiments was using a variety of natural triggers of innate immunity beyond synthetic poly(I:C). Here we found indeed significant reductions of ISG induction for all viruses tested, similar to poly(I:C), this is the basis for the statement that RBM39 contributes the cell intrinsic innate immune response to viral infections. Our experimental design did not intend to see pronounced effects on viral replication, this was only measured to secure that reduced ISG induction was not due to inhibition of viral replication. We have explained this strategy now clearer and tuned down corresponding statements, to exclude potential overinterpretation of the data.

   1. Neither DTU/DRIMseq or qPCR are valid methods to measure splice isoform differences. The authors need to use rMATS or MAJIQ and validate by gel-based RT-PCR.

   Output generated by modern alignment algorithms like salmon is suitable for studies on an isoform level (Love* et al, 2018) and has been used in a variety of studies (e.g.(Jabs et al, 2020; Xiong et al, 2023). MAJIQ and rMATS are only superior tools if the detection of so far unknown isoforms is of interest (Love et al.*, 2018), which is beyond the scope of this project. We have validated the data for IRF3 in RT-qPCR, showing close to identical results to the DTU analysis (compare Fig. 6A and S6D). We disagree that a gel-based RT-PCR analysis would be superior here, due to the lack of quantification.

   1. The conclusions from the proteomic and transcriptomic analyses should be treated with extreme caution given the caveats of methodology and controls discussed above.

   We are aware of the caveats of these technologies. The previous proteomic analysis indeed suffered from low sensitivity, failing to detect essential candidates like IRF3. The repetition of the experiment (new Fig. 5A, B, new Fig. S5) now revealed data very consistent with the transcriptomic data. Overall, the strength of our approach is the direct comparison of siRNA based RBM39 knockdown and RBM39 depletion by indisulam throughout transcriptomics and proteomics analyses. The wide overlap argues for the validity of our data and suggests that we thereby circumvented many caveats.

   Reviewer #3 (Significance (Required)):

   Innate immune signaling is a complex and essential pathway for maintaining health. While much is known about key components of this pathway, additional regulators are likely to exist. This manuscript describes an attempt to identify new regulators of dsRNA-mediated gene expression.

   References

   Burkart SS, Schweinoch D, Frankish J, Sparn C, Wust S, Urban C, Merlo M, Magalhaes VG, Piras A, Pichlmair A* et al* (2023) High-resolution kinetic characterization of the RIG-I-signaling pathway and the antiviral response. Life Sci Alliance 6

   Campagne S, Jutzi D, Malard F, Matoga M, Romane K, Feldmuller M, Colombo M, Ruepp MD, Allain FH (2023) Molecular basis of RNA-binding and autoregulation by the cancer-associated splicing factor RBM39. Nat Commun 14: 5366

   Coomar S, Mota P, Penson A, Schwaller J, Abdel-Wahab O, Gillingham D (2023) Overlaid Transcriptional and Proteome Analyses Identify Mitotic Kinesins as Important Targets of Arylsulfonamide-Mediated RBM39 Degradation. Mol Cancer Res 21: 768-778

   Dou Z, Zhang X, Su W, Zhang T, Ye F, Zhao D, Chen X, Li Q, Zhang H, Di C (2023) Indisulam exerts anticancer effects via modulation of transcription, translation and alternative splicing on human cervical cancer cells. Am J Cancer Res 13: 2922-2937

   Jabs S, Biton A, Becavin C, Nahori MA, Ghozlane A, Pagliuso A, Spano G, Guerineau V, Touboul D, Giai Gianetto Q* et al* (2020) Impact of the gut microbiota on the m(6)A epitranscriptome of mouse cecum and liver. Nat Commun 11: 1344

   Love MI, Soneson C, Patro R (2018) Swimming downstream: statistical analysis of differential transcript usage following Salmon quantification. F1000Res 7: 952

   Mai S, Qu X, Li P, Ma Q, Cao C, Liu X (2016) Global regulation of alternative RNA splicing by the SR-rich protein RBM39. Biochim Biophys Acta 1859: 1014-1024

   Nijhuis A, Sikka A, Yogev O, Herendi L, Balcells C, Ma Y, Poon E, Eckold C, Valbuena GN, Xu Y* et al* (2022) Indisulam targets RNA splicing and metabolism to serve as a therapeutic strategy for high-risk neuroblastoma. Nat Commun 13: 1380

   Xiong L, Liu J, Han SY, Koppitch K, Guo JJ, Rommelfanger M, Miao Z, Gao F, Hallgrimsdottir IB, Pachter L* et al* (2023) Direct androgen receptor control of sexually dimorphic gene expression in the mammalian kidney. Dev Cell 58: 2338-2358 e2335

   Read the original source
2. 

   Review Commons

   Jul 18, 2024

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #3

   Evidence, reproducibility and clarity

   CRISPR Screen for factors that are required for dsRNA-dependent ISG production. Found a large number of hits but most did not validate in subsequent assays. The authors follow up the one candidate that did pass secondary screening criteria, RBM39, although re-expression of RBM39 only rescues the phenotype of the siRNAs against RBM39 (siRBM39) in one of the two cell lines tested. Additionally, siRBM39 impacts only a subset of polyIC-induced ISGs and does not regulate NFkB-driven gene expression. They go on to attempt to investigate the impact of siRBM39 on other key innate immune genes and proteins, although many key controls and …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #3

   Evidence, reproducibility and clarity

   CRISPR Screen for factors that are required for dsRNA-dependent ISG production. Found a large number of hits but most did not validate in subsequent assays. The authors follow up the one candidate that did pass secondary screening criteria, RBM39, although re-expression of RBM39 only rescues the phenotype of the siRNAs against RBM39 (siRBM39) in one of the two cell lines tested. Additionally, siRBM39 impacts only a subset of polyIC-induced ISGs and does not regulate NFkB-driven gene expression. They go on to attempt to investigate the impact of siRBM39 on other key innate immune genes and proteins, although many key controls and appropriate methods are missing.

   Major comments:

   1. The authors propose some rationale for the limited success of the screen, however, while RBM39 may have a role in dsRNA-induced innate immunity, in general the screen seems to have limited value.
   2. Given that the siRBM39 clearly has off-target effects (since expression of a resistant RBM39 cDNA only gives limited rescue in many cases - Fig S2), each of the experiments in which siRBM39 is used (i.e. Fig 2) should have the RBM39.esc control - especially those that drive subsequent experiments such as the expression of IFNbeta and IFNLR1 (Fig 2a, h)
   3. Since RBM39 reduction has an apparent impact even if IFNLR1-deficient cells (although need the rescue control to know if this is real) the authors conclude that RBM39 regulates the initial wave of dsRNA signaling-events, but this should be tested with the use of Ruxilitinib to block JAK-STAT signaling.
   4. IRF3 expression in the Indisulam-treated cells more closely tracks cell viability than RBM39 expression. For example in Fig 3C 10 microM gives 50% IRF3 expression and 50% viability but still 95% RBB39 expression - arguing that the impact of siRBM39 on IRF3 might be very indirect (and error bars on rescue are large so unclear if the rescue really worked in Fig 3A).
   5. It is unclear in Fig 4 why some cell/virus combinations are tested with siRBM39 and others are tested with Indisulam. Also the conclusion that RBM39 "substantially contributes to the cell intrinsic innate immune response to viral infections" is greatly overstated given that the differences are between ~3 fold and non-significant.
   6. Neither DTU/DRIMseq or qPCR are valid methods to measure splice isoform differences. The authors need to use rMATS or MAJIQ and validate by gel-based RT-PCR.
   7. The conclusions from the proteomic and transcriptomic analyses should be treated with extreme caution given the caveats of methodology and controls discussed above.

   Significance

   Innate immune signaling is a complex and essential pathway for maintaining health. While much is known about key components of this pathway, additional regulators are likely to exist. This manuscript describes an attempt to identify new regulators of dsRNA-mediated gene expression.

   Read the original source
3. 

   Review Commons

   Jul 18, 2024

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #2

   Evidence, reproducibility and clarity

   The authors performed a CRISPR-based screen for genes required for TLR3-mediated signaling and gene expression in Hepatoma cells. Interferon-stimulated expression of an apoptosis inducer was used as a read-out system. A number of candidate genes were identified and one of these, RBM39, investigated in detail. The protein has previously been linked to both transcriptional control and RNA processing. Validation studies confirm that reduction of cellular RBM39 results in less TLR3-mediated IFN-beta synthesis and lower levels of ISG mRNA synthesis. Initial studies suggest a role of RBM39 in regulating of IRF3 levels, the transcription …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #2

   Evidence, reproducibility and clarity

   The authors performed a CRISPR-based screen for genes required for TLR3-mediated signaling and gene expression in Hepatoma cells. Interferon-stimulated expression of an apoptosis inducer was used as a read-out system. A number of candidate genes were identified and one of these, RBM39, investigated in detail. The protein has previously been linked to both transcriptional control and RNA processing. Validation studies confirm that reduction of cellular RBM39 results in less TLR3-mediated IFN-beta synthesis and lower levels of ISG mRNA synthesis. Initial studies suggest a role of RBM39 in regulating of IRF3 levels, the transcription factor activated by TLR3 signaling to induce IFN-beta synthesis. However, the effect is variable and poorly supported by transcriptomic and proteomic data. Moreover, only one out of four cell-based viral infection models reports a substantial effect of the RBM39 knockdown.

   Major comments:

   1. The data do not support the claim that RBM39 is a broadly acting player in innate immune responses. In addition, they suggest that IRF3 may not be the only relevant RBM39 target. The most informative knockdown control in this regard would be IRF3 siRNA.
   2. The structure of the manuscript is rather confusing because IRF3 is presented as the main RBM39 target in figures 3-6, but the -omics data in figures 7 and 8 do not support this view. The authors argue different sensitivities of the experimental approaches, but I think few people would agree that western blots are more sensitive than MS. To my opinion a narrative with less focus on IRF3 and a broader integration of candidates of the -omics approaches would be preferable. Investigating the role of RBM39 by RNA-seq in pIC-treated cells would further strengthen the manuscript. It will yield a broader view of the protein's role in induced innate immunity.
   3. The results in figures 6A-C are confusing for two reasons. First, the siRNA-mediated knockdown should result in reduced RBM39 protein as well (as shown in Fig. 3A) and, therefore, in an increase in RBM39 levels. Second, why was this effect not noted in the experiments shown in figs. 1-5? To avoid this confusion it might be good to mention which IRF3 splice isoforms are detected by the primers and antibodies used in these figures.

   Minor comments.

   1. Fig S1: the figure panels and legend are inconsistent. IFIT1 is labeled as ISG56 in panel S1A.
   2. Data with the siRNA escape mutant of RBM39 are inconsistent. For example, why is its effect significantly different only in 1 out of 4 ISG in figures S2A-D?
   3. Line 164: the statement that TRIF and RBM39 siRNAs produce effects of similar magnitude is incorrect for the IFIT1 gene in figure S2A.
   4. Fig. 2H: In absence of additional evidence for functional implications, the data showing reduced IL10RB expression should be omitted.
   5. Fig. 3: More datapoints would be needed in panel A to sustain the lack of significant difference between the untreated and escape mutant samples. Are the viability data in panels B and C normalized to untreated cells to control for Indisulam toxicity? In figure S3A the effect of the mutant is rather small. To allow for comparison, the Indisulam titration curves should be adapted to the concentrations used in Fig. 3.
   6. RNA-seq measures steady-state RNA, not transcription.

   Significance

   The identification of RBM39 as a candidate player in innate immune responses is of interest to a large scientific community with interest in signalling by pattern recognition receptors. Its role should be strengthened with additional infection models. It is puzzling that three out of four viruses don't benefit from the reduced IFN-beta synthesis in the RBM39 knockdown. Moreover, the data are not convincing (or too diverse) to nail down IRF3 as a major, or the most relevant, RBM39 target.

   Read the original source
4. 

   Review Commons

   Jul 18, 2024

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #1

   Evidence, reproducibility and clarity

   This manuscript by Li and colleagues examines the role of RBM39 in innate immune signaling. Splicing factor RBM39 was identified through a genome wide screen with a death reporter under control of the IFIT1 promoter that got stimulated with pIC in a TLR3-dependent manner. Besides IFIT1, further experiments showed that RBM39 is also involved in optimal expression of other innate immunity genes like IFNB, CXCL10, RIG-I or MDA5. While NFkB-dependent genes seem not to depend on RBM39, for IRF3 it was shown that protein levels decrease under conditions of RBM39 depletion, because IRF3 mRNAs are (slightly) reduced and spliced differently. …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #1

   Evidence, reproducibility and clarity

   This manuscript by Li and colleagues examines the role of RBM39 in innate immune signaling. Splicing factor RBM39 was identified through a genome wide screen with a death reporter under control of the IFIT1 promoter that got stimulated with pIC in a TLR3-dependent manner. Besides IFIT1, further experiments showed that RBM39 is also involved in optimal expression of other innate immunity genes like IFNB, CXCL10, RIG-I or MDA5. While NFkB-dependent genes seem not to depend on RBM39, for IRF3 it was shown that protein levels decrease under conditions of RBM39 depletion, because IRF3 mRNAs are (slightly) reduced and spliced differently. The sulfonamid Indisulam could largely recapitulate the phenotype of RBM39 depletion. Further analyses using proteomics and transcriptomics showed that RBM39 is required for mRNA splicing and expression of a large set of other proteins.

   Altogether, this well designed and written study highlights the fundamental role played by RBM39 in in maintaining the pathways of immunity and metabolism. The key conclusions are convincing but some additional experiments would strengthen them further.

   Major comments:

   • For the statistics, authors seem not to have done multiple tests but rather tested individual datasets within larger graphs against each other. Please explain where this is the case and use corrections if multiple testing was done
   • Fig. 4 shows that RBM39 depletion reduces IFIT expression in virus infected cells and slightly increases virus replication. RBM39 has a major effect on IRF3 levels, but also on other players in innate immunity. What happens if IRF3 is ectopically expressed as in figure 5? With this experiment one could measure how high the contribution of IRF3 miss-splicing is to innate immunity.
   • Fig. 4 A uses siRNAs but B, C and D only indisulam treatment. It would be better if siRNAs would also be used for the other viruses.
   • RBM39 depletion strongly reduces IRF3 levels in the WB, but not so much in RT-PCR and not at all in proteomics. Is the antibody used for WB perhaps recognizing a domain that is underrepresented in isoforms after disturbed splicing? Please clarify.
   • Volcano plots in figure 7 show a lot of hits obtained after both RBM38 siRNA and indisulam (green dots), and some that are additionally identified in transcriptomes and in proteomes (red dots). Nonetheless only innate immunity and stress response genes are marked, although they do not belong to these highly conserved classes. Please elaborate more on the most RBM39-dependent genes, e.g. by presenting them in a heat map.

   Minor comments:

   • Some abbreviations are not explained, like PGK, siNT, siVTN
   • Welsch should read Welch
   • Fig. 2H: were cells also stimulated and if yes, how?
   • Fig. 6E: I cannot see a difference between to IRF3-203 and 228 isoforms. And what are the white boxes?
   • Also 6E: Location of the primers is barely visible
   • Some materials are not properly referenced, like the death reporter, the lentiviral system, or the Rift Valley fever luciferase virus
   • Supplement has no page numbers

   Significance

   The study advances our knowledge about the regulation of innate immunity. Strengths are the discovery of a novel layer of innate immunity regulation by splicing and the in-depth analysis of the importance of RBM39 for cellular gene expression. A potential weakness might be the focus on innate immunity as other biological functions seem even more dependent on RBM39. However, this reviewer sees the necessity that covering all aspects of RBM39 finction would be beyond the scope of a single study.

   The relevant literature is appropriately cited (except for some materials, see minor comments). Results will be of interest not only to people doing basic research on innate immunity, but also to those interested in gene regulation in general or to cancer researchers using indisulam

   Read the original source
5. 

   Version published to 10.1101/2023.10.13.562221v2 on bioRxiv

   May 22, 2024
6. 

   Version published to 10.1101/2023.10.13.562221v1 on bioRxiv

   Oct 15, 2023