C/EBPδ-induced epigenetic changes control the dynamic gene transcription of S100a8 and S100a9

  1. Saskia-Larissa Jauch-Speer
  2. Marisol Herrera-Rivero
  3. Nadine Ludwig
  4. Bruna Caroline Véras De Carvalho
  5. Leonie Martens
  6. Jonas Wolf
  7. Achmet Imam Chasan
  8. Anika Witten
  9. Birgit Markus
  10. Bernhard Schieffer
  11. Thomas Vogl
  12. Jan Rossaint
  13. Monika Stoll
  14. Johannes Roth
  15. Olesja Fehler

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    The manuscript by Jauch-Speer and colleagues uses a CRISPR/Cas9 screening approach in a myeloid cell line to identify C/EBP-delta as a regulator of the alarmins S100A8 and S100A9, which amplify inflammation. This paper is of great interest to macrophage biologists studying macrophage function in inflammatory diseases. In an elegant series of gene targeting and sequencing studies, the authors also characterized epigenetic mechanisms regulating the expression of these pro-inflammatory mediators. Furthermore, human monocytes from cardiovascular patient cohorts showed correlative changes, indicating possible clinical relevance.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

The proinflammatory alarmins S100A8 and S100A9 are among the most abundant proteins in neutrophils and monocytes but are completely silenced after differentiation to macrophages. The molecular mechanisms of the extraordinarily dynamic transcriptional regulation of S100a8 and S100a9 genes, however, are only barely understood. Using an unbiased genome-wide CRISPR/Cas9 knockout (KO)-based screening approach in immortalized murine monocytes, we identified the transcription factor C/EBPδ as a central regulator of S100a8 and S100a9 expression. We showed that S100A8/A9 expression and thereby neutrophil recruitment and cytokine release were decreased in C/EBPδ KO mice in a mouse model of acute lung inflammation. S100a8 and S100a9 expression was further controlled by the C/EBPδ antagonists ATF3 and FBXW7. We confirmed the clinical relevance of this regulatory network in subpopulations of human monocytes in a clinical cohort of cardiovascular patients. Moreover, we identified specific C/EBPδ-binding sites within S100a8 and S100a9 promoter regions, and demonstrated that C/EBPδ-dependent JMJD3-mediated demethylation of H3K27me 3 is indispensable for their expression. Overall, our work uncovered C/EBPδ as a novel regulator of S100a8 and S100a9 expression. Therefore, C/EBPδ represents a promising target for modulation of inflammatory conditions that are characterized by S100a8 and S100a9 overexpression.

Article activity feed

  1. Version published to 10.7554/elife.75594 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    The authors elegantly use the CRISPR/Cas9 screening approach to perform an unbiased analysis of which genes regulate the expression of the alarmins S100A8 and S100A9. As pointed out by the authors, these alarmins amplify inflammation and can thus contribute to tissue injury during excessive inflammation. Understanding the regulators of alarmin expression could lead to a better understanding of myeloid cell differentiation and hyperinflammatory activation.

    We thank the reviewer very much and we appreciate the critical evaluation and suggestions for improvement of our manuscript. We are pleased about the assessment of the strengths of our study and we will respond to the specific comments below.

    1. Unclear relevance of the C/EBP-delta regulation of alarmins in a disease model. The C/EBP-delta knockout cells are studied ex vivo but that does not address whether C/EBP-delta absence would dampen alarmin expression and inflammatory injury in vivo.

    We agree with the reviewer’s statement. We have now added an experimental model of LPS-induced acute lung injury comparing wild type and C/EBPδ KO mice which supports our data obtained in vitro and the functional role of C/EBPδ for S100-expression in vivo as well (see Figure 2, E – J and Figure 2 – figure supplement 2 and lines 190 – 202).

    1. The phenotyping of the myeloid cells following C/EBP-delta deletion is very limited and does not provide a clear assessment of whether C/EBP-delta affects monocyte-to-macrophage differentiation and their polarization to specific monocyte and macrophage phenotypes.

    As suggested by the reviewer we added data regarding polarization of monocytes comparing expression of M1 and M2 markers in wild type with C/EBPδ KO cells (Figure 3 – figure supplement 2, lines 215 - 224). In addition, we performed a genome wide ATAC-seq also in C/EBPδ KO monocytes which allows comparison of general differences of genome organisation between wild type and C/EBPδ KO monocytes for interested researchers (Figure 7, A – C, Figure 7 – figure supplement 1 and Figure 7 – source data 1, lines 278 – 287).

    Reviewer #2 (Public Review):

    In this work, Jauch-Speer et al. examine the epigenetic mechanisms regulating the expression of S100A8 and S100A9, two prevalent DAMPs released by monocytes and neutrophils during acute inflammation and tissue injury. S100A8/9 are highly expressed by monocytes but downregulated in mature macrophages, and thus their transcription is temporally controlled. While helpful in anti-microbial responses, S100A8/9 have been associated with a variety of inflammatory diseases, including autoimmune and cardiovascular diseases, in which their expression may become dysregulated. However, the mechanisms regulating their expression are poorly understood. Using ER-Hoxb8 cells, a series of gene targeting and sequencing studies were employed to characterize the dynamics of s100a8 and s100a9 transcription and regulation. First, a genome-wide screening approach using the CRISPR/Cas9 system identified the transcription factor C/EBPδ as a direct regulator of s100a8/9, which were co-expressed by differentiating monocytes. Accordingly, S100A8 and S100A9 expression was decreased in C/EBPδ-deficient cells and increased with C/EBPδ overexpression. Next, ChIP- and ATAC-seq sequencing determined the C/EBPδ binding sites within the promoters for s100a8 and s100a9. Furthermore, the presence of H3K27me3 (silencing) markers on the s100a8/9 promoters were elevated in C/EBPδ KO cells relative to WT monocytes, as well as decreased expression of the demethylase-encoding gene jmjd3, which could mediate the removal of H3K27me3 markers. Thus, C/EBPδ-dependent JMJD3 was essential for the demethylation and expression of S1008/9 in monocytes. Finally, S100A8, S100A9, and C/EBPδ expression in classical monocytes was positively associated with stable coronary artery disease and MI in a cohort of cardiovascular disease patients, demonstrating clinical relevance.

    Presented here are a logical series of studies that demonstrate a previously unknown epigenetic mechanism through which S100A8/9 expression is regulated in monocytes. Furthermore, the authors make use of a variety of current sequencing technologies to sufficiently support their conclusions. This work has important implications for diseases in which S100A8/9 expression is altered, and provides clinically relevant targets for future therapeutic studies. Overall, this study would be of great interest to macrophage biologists studying macrophages in a broad variety of disease models.

    We thank the reviewer for this friendly comment.

    However, certain aspects of the paper require further clarification or were not sufficiently investigated.

    1. The mechanism(s) through which S100A8/9 expression is subsequently downregulated in mature macrophages was a concept that was introduced, but not explored. Whether C/EBPδ and JMJD3 are also involved in the downregulation of S100A8/9 in monocytes during later stages of differentiation would be important to fully understand the temporal dynamics of this regulatory network.

    We agree with the reviewer that our approach and experimental setup addressed the induction of S100 expression during monocyte differentiation. To identify mechanisms of down-regulation of S100 expression during the later course of differentiation one would have to perform an additional GeCKO screen to analyse cells that express S100A9 in high amounts at later stages of differentiation, such as day 5, in comparison to reference cells where S100A9 expression is diminished. At least as far as C/EBPδ is concerned, we assume that the parallel decrease of C/EBPδ and S100 expression points to a functional link as well. However, the late S100-kinetics in wild type and C/EBPδ KO cells clearly indicate the presence of additional relevant factors. We discussed these points in our revised manuscript (lines 371 - 376).

    1. The authors cite a differentiation protocol using estrogen-regulated Hoxb8 cells (Wang et al., 2006) to produce monocytes or neutrophils in culture over a span of 5 days, and utilizes differentiating monocytes as early as day 3. However, the original paper states that the precursor cells can differentiate into macrophages after 6 days - not monocytes. Macrophages and monocytes are functionally distinct, and will have different gene expression profiles as well as different epigenetic mechanisms regulating them. Hence it is unclear whether the cells at varying days of culture are monocytes or macrophages, or if transitioning from one to the other, what stage of differentiation they are in. Following this, the authors should characterize the exact cell composition of the culture at each day of differentiation using flow cytometry and more clearly validate that it was monocytes and not macrophages that were being analyzed. Additionally, cells grown in culture lack the complex cues and factors provided by the tissue environment, and thus additional studies ought to be performed on myeloid cells within tissues of interest to confirm these findings. The data in the cell line show a direct relationship between C/EBPδ and S100A8/A9, however the data in primary cells is only a correlation.

    The ER-Hoxb8 system is a model of differentiation of bone marrow derived monocyte-macrophage differentiation. Differentiation of macrophages of different origin (e. g. yolk sac of fetal liver) cannot be analysed with this technique. There is no defined step for the transition of monocytes to macrophages in this culture system. In humans, monocytes express high amounts of S100A8 and S100A9 whereas monocyte-derived macrophages show no expression of these molecules any more after differentiation in vitro or in vivo. In analogy we used these terms in our manuscript. We made this point clear in our revised manuscript (lines 315 - 319).

    We tested S100a8 and S100a9 expression after differentiation of primary bone-marrow derived monocytes from WT and C/EBPδ KO mice (n = 3) and found again decreased s100 levels upon C/EBPδ deletion (Figure 2, D), confirming our ER-Hoxb8 data.

    To show the impact of C/EBPδ-deficiency on S100A8 and S100A9 expression in vivo we performed a murine mouse model for acute lung inflammation. Not only at baseline conditions (NaCl-exposure), but also after LPS-exposure S100A8/A9 levels are systemically (serum) and locally (bronchoalveolar lavage fluid, BALF) decreased in C/EBPδ KO mice compared to WT mice (Figure 2, E and F).

    1. ATAC-Seq was not adequately utilized to characterize s100a8 and s100a9 chromatin (Figure 6). While the peaks for s100a8 and s100a9 were provided, the difference between day 0 and day 3 was not quantified, nor were peaks for any other genes shown despite over 20,000 gene regions with differential peaks between the two time points. What genes these differential peaks were annotated to aside from s100a8/9 would help paint a more comprehensive picture of the differences between day 0 and day 3. Furthermore, C/EBPδ KO cells were not analyzed by ATAC-seq despite being included in subsequent ChIP-seq experiments, creating a gap in the data analysis.

    We addressed the recommendations of the reviewer and performed additional ATAC-seq experiments with C/EBPδ KO day 0 and day 3 (n = 3) in the same methodical manner as for the WT samples and re-analysed all data. This approach revealed over 1,000 regions with differential peaks for all comparisons (Figure 7, A and Figure 7 – figure supplement 1). Openness of chromatin between precursor and differentiated cells, as shown already for WT, was also highly different in C/EBPδ KO cells. Among the regions with significantly higher ATAC-seq reads in differentiated samples were the S100a8 and S100a9 promoter and enhancer locations in both, WT and C/EBPδ KO, which reflects the H3K27ac-ChIP data (Figure 7, B – E). As expected, comparison of peaks in this S100-associated regions at day 3 reveals a weaker chromatin accessibility in C/EBPδ KO in relation to WT cells, which in turn mirrors the H3K27me3-ChIP data (Figure 7, B – C and F - G) and overall reflects the findings of less S100a8 and S100a9 expression upon C/EBPδ deletion (Figure 2). For a comprehensive picture of all regions with differential peaks between all four conditions, see Figure 7 – source data 1. In addition to the visualisation of ATAC-seq peaks between regions, we now statistically analysed differential chromatin accessibility between the conditions by specifying the adjusted p-values (padj) < 0.05 and log2 fold changes in our revised version (Figure 7, C).

    1. In the section that deals with human data, it states "RNA-seq in monocyte subpopulations of BioNRW participants (n=26,from 3 individuals in each of the sCAD, MI and Ctrl diagnostic groups)". N value is unclear. If its n=3 from 3 groups, is that not 9? What is 26 in reference to?

    We apologise for the confusion. The monocytes dataset contained read counts of classical, intermediate and non-classical monocyte subpopulations from 9 male individuals (3 MI, 3 sCAD and 3 controls = 27 samples). One sCAD non-classical monocyte sample had to be excluded from analysis due to low mapping rate; therefore, the monocytes dataset used for analysis contained 26 samples (27 -1 = 26, see Methods section ,,RNA seq’’).

    Reviewer #3 (Public Review):

    Major strengths:

    1. Unique design of genome-wide CRISPR screen by focusing on S100A9 expression. As presented, S100A9 expression, as detected by fluorescent antibody, is a robust reporter signal to unbiasedly screen for regulatory genes of S100A9.
    1. The impact of C/EBPD knockout on S100A8/A9 expression is highly significant, as observed in multiple cell models. Importantly, the authors demonstrated that C/EBPD directly binds to the promoters of s100a8/a9 genes, and this gene activation is further regulated by chromatin accessibility, a process regulated by JMJD3-mediated demethylation.
    1. The expression correlation between S100A8/A9 and C/EBPD has been observed in cardiovascular patients, implying clinical significance of this pathway in disease development.

    We appreciate the detailed review, careful evaluation and constructive criticism, which has helped us to improve the manuscript and its value. We are pleased that our general findings of C/EBPδ as a novel regulator of S100A8 and S100A9 are convincing. We thank you for highlighting the major strengths and will discuss the potential weaknesses below in the corresponding paragraphs.

    Potential Weaknesses:

    1. The genome-wide screen suggested that about 28% of all genes (estimated from the cell percentage) are involved in S100A9 regulation (Fig. 1B). This seems unlikely and probably reflects a high level of false positives in the screening results. The substantial level of noise could easily mask signals from other true regulators of S100A8/A9.

    We agree that there is in fact a high level of false positive hits. We therefore collected the S100A9high expressing cells as reference/control. The MaGECK algorithm that was used for analysis of gRNA abundancy considers genes as hits for whose gRNAs are mathematically overrepresented in our hit pool over the reference pool. This way, the risk of false positives is reduced but still significant. We addressed the limitations of sensitivity and specificity of this approach in our revised manuscript (lines 352 - 365). However, despite limitations this approach can be successfully used for an unbiased search of new candidates of gene regulation. However, identification of all members of a regulatory complex or pathway seems unrealistic in our opinion. On the other hand, several transcription factors which have been reported to regulate S100-expression in more or less artificial systems, did not pop up in our screen and our independent validation experiments confirmed the screening results (Figure 1 – figure supplement 1, B). We discuss this point in our revised manuscript (lines 377 - 382).

    1. There were many hits from the CRISPR screen. The authors picked C/EBPD because three distinct gRNAs were identified from this gene. However, there are many other interesting candidates with highly significant p-values (better than C/EBPD) and with two distinct gRNAs. There is no description of these candidates, nor discussion of their potential relevance to S100A8/A9 regulation.

    We agree with this point. See also our response to point #1. Due to the high number of significant hits and the risk of false positive candidates we selected C/EBPδ targeted by three independent gRNAs. However, we present now additional experimental evidence supporting our strategy. We chose 3 additional targets known to be involved in gene regulation and analysed their effects on S100-expression in independent knock-out approaches (Phf8, Hand1, and Csrp1). As shown in our revised Figure 1 – figure supplement 1, A C/EBPδ deletion has the greatest effect on S100a8 and S100a9 expression confirming the special relevance of C/EBPδ.

  3. eLife

    Evaluation Summary:

    The manuscript by Jauch-Speer and colleagues uses a CRISPR/Cas9 screening approach in a myeloid cell line to identify C/EBP-delta as a regulator of the alarmins S100A8 and S100A9, which amplify inflammation. This paper is of great interest to macrophage biologists studying macrophage function in inflammatory diseases. In an elegant series of gene targeting and sequencing studies, the authors also characterized epigenetic mechanisms regulating the expression of these pro-inflammatory mediators. Furthermore, human monocytes from cardiovascular patient cohorts showed correlative changes, indicating possible clinical relevance.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    The authors elegantly use the CRISPR/Cas9 screening approach to perform an unbiased analysis of which genes regulate the expression of the alarmins S100A8 and S100A9. As pointed out by the authors, these alarmins amplify inflammation and can thus contribute to tissue injury during excessive inflammation. Understanding the regulators of alarmin expression could lead to a better understanding of myeloid cell differentiation and hyperinflammatory activation.

    The strengths of the study are:

    1. Unbiased CRISPR/Cas9 screening which identifies C/EBP-delta as a regulator of alarmins

    2. Rigorous analysis of the mechanistic link between C/EBP-delta and S100A8 and S100A9 using C/EBP-delta-knockout mouse cells as well as rescue of S100A8 and S100A9 expression with an inducible C/EBP-delta construct.

    3. Showing translational relevance of the C/EBP-delta link by demonstrating a correlation between S100A8 and S100A9 levels with C/EBP-delta in patient samples of peripheral blood mononuclear cells

    4. Promoter analysis and epigenetic assessment of how C/EBP-delta binds to the promoters of the alarmins and which epigenetic modulators contribute to their expression.

    The weaknesses of the study are:

    1. Unclear relevance of the C/EBP-delta regulation of alarmins in a disease model. The C/EBP-delta knockout cells are studied ex vivo but that does not address whether C/EBP-delta absence would dampen alarmin expression and inflammatory injury in vivo.

    2. The phenotyping of the myeloid cells following C/EBP-delta deletion is very limited and does not provide a clear assessment of whether C/EBP-delta affects monocyte-to-macrophage differentiation and their polarization to specific monocyte and macrophage phenotypes.

  5. eLife

    Reviewer #2 (Public Review):

    In this work, Jauch-Speer et al. examine the epigenetic mechanisms regulating the expression of S100A8 and S100A9, two prevalent DAMPs released by monocytes and neutrophils during acute inflammation and tissue injury. S100A8/9 are highly expressed by monocytes but downregulated in mature macrophages, and thus their transcription is temporally controlled. While helpful in anti-microbial responses, S100A8/9 have been associated with a variety of inflammatory diseases, including autoimmune and cardiovascular diseases, in which their expression may become dysregulated. However, the mechanisms regulating their expression are poorly understood. Using ER-Hoxb8 cells, a series of gene targeting and sequencing studies were employed to characterize the dynamics of s100a8 and s100a9 transcription and regulation. First, a genome-wide screening approach using the CRISPR/Cas9 system identified the transcription factor C/EBPδ as a direct regulator of s100a8/9, which were co-expressed by differentiating monocytes. Accordingly, S100A8 and S100A9 expression was decreased in C/EBPδ-deficient cells and increased with C/EBPδ overexpression. Next, ChIP- and ATAC-seq sequencing determined the C/EBPδ binding sites within the promoters for s100a8 and s100a9. Furthermore, the presence of H3K27me3 (silencing) markers on the s100a8/9 promoters were elevated in C/EBPδ KO cells relative to WT monocytes, as well as decreased expression of the demethylase-encoding gene jmjd3, which could mediate the removal of H3K27me3 markers. Thus, C/EBPδ-dependent JMJD3 was essential for the demethylation and expression of S1008/9 in monocytes. Finally, S100A8, S100A9, and C/EBPδ expression in classical monocytes was positively associated with stable coronary artery disease and MI in a cohort of cardiovascular disease patients, demonstrating clinical relevance.

    Presented here are a logical series of studies that demonstrate a previously unknown epigenetic mechanism through which S100A8/9 expression is regulated in monocytes. Furthermore, the authors make use of a variety of current sequencing technologies to sufficiently support their conclusions. This work has important implications for diseases in which S100A8/9 expression is altered, and provides clinically relevant targets for future therapeutic studies. Overall, this study would be of great interest to macrophage biologists studying macrophages in a broad variety of disease models.

    However, certain aspects of the paper require further clarification or were not sufficiently investigated.

    1. The mechanism(s) through which S100A8/9 expression is subsequently downregulated in mature macrophages was a concept that was introduced, but not explored. Whether C/EBPδ and JMJD3 are also involved in the downregulation of S100A8/9 in monocytes during later stages of differentiation would be important to fully understand the temporal dynamics of this regulatory network.

    2. The authors cite a differentiation protocol using estrogen-regulated Hoxb8 cells (Wang et al., 2006) to produce monocytes or neutrophils in culture over a span of 5 days, and utilizes differentiating monocytes as early as day 3. However, the original paper states that the precursor cells can differentiate into macrophages after 6 days - not monocytes. Macrophages and monocytes are functionally distinct, and will have different gene expression profiles as well as different epigenetic mechanisms regulating them. Hence it is unclear whether the cells at varying days of culture are monocytes or macrophages, or if transitioning from one to the other, what stage of differentiation they are in. Following this, the authors should characterize the exact cell composition of the culture at each day of differentiation using flow cytometry and more clearly validate that it was monocytes and not macrophages that were being analyzed. Additionally, cells grown in culture lack the complex cues and factors provided by the tissue environment, and thus additional studies ought to be performed on myeloid cells within tissues of interest to confirm these findings. The data in the cell line show a direct relationship between C/EBPδ and S100A8/A9, however the data in primary cells is only a correlation.

    3. ATAC-Seq was not adequately utilized to characterize s100a8 and s100a9 chromatin (Figure 6). While the peaks for s100a8 and s100a9 were provided, the difference between day 0 and day 3 was not quantified, nor were peaks for any other genes shown despite over 20,000 gene regions with differential peaks between the two time points. What genes these differential peaks were annotated to aside from s100a8/9 would help paint a more comprehensive picture of the differences between day 0 and day 3. Furthermore, C/EBPδ KO cells were not analyzed by ATAC-seq despite being included in subsequent ChIP-seq experiments, creating a gap in the data analysis.

    4. In the section that deals with human data, it states "RNA-seq in monocyte subpopulations of BioNRW participants (n=26,from 3 individuals in each of the sCAD, MI and Ctrl diagnostic groups)". N value is unclear. If its n=3 from 3 groups, is that not 9? What is 26 in reference to?

  6. eLife

    Reviewer #3 (Public Review):

    Major strengths:

    1. Unique design of genome-wide CRISPR screen by focusing on S100A9 expression. As presented, S100A9 expression, as detected by fluorescent antibody, is a robust reporter signal to unbiasedly screen for regulatory genes of S100A9.
    2. The impact of C/EBPD knockout on S100A8/A9 expression is highly significant, as observed in multiple cell models. Importantly, the authors demonstrated that C/EBPD directly binds to the promoters of s100a8/a9 genes, and this gene activation is further regulated by chromatin accessibility, a process regulated by JMJD3-mediated demethylation.
    3. The expression correlation between S100A8/A9 and C/EBPD has been observed in cardiovascular patients, implying clinical significance of this pathway in disease development.

    Potential Weaknesses:

    1. The genome-wide screen suggested that about 28% of all genes (estimated from the cell percentage) are involved in S100A9 regulation (Fig. 1B). This seems unlikely and probably reflects a high level of false positives in the screening results. The substantial level of noise could easily mask signals from other true regulators of S100A8/A9.
    2. There were many hits from the CRISPR screen. The authors picked C/EBPD because three distinct gRNAs were identified from this gene. However, there are many other interesting candidates with highly significant p-values (better than C/EBPD) and with two distinct gRNAs. There is no description of these candidates, nor discussion of their potential relevance to S100A8/A9 regulation.
  7. Version published to 10.1101/2021.12.05.471309v1 on bioRxiv