CRISPR screen identifies CEBPB as contributor to dyskeratosis congenita fibroblast senescence via augmented inflammatory gene response

Abstract Aging is the consequence of intra- and extracellular events that promote cellular senescence. Dyskeratosis congenita (DC) is an example of a premature aging disorder caused by underlying telomere/telomerase-related mutations. Cells from these patients offer an opportunity to study telomere-related aging and senescence. Our previous work has found that telomere shortening stimulates DNA damage responses (DDRs) and increases reactive oxygen species (ROS), thereby promoting entry into senescence. This work also found that telomere elongation via TERT expression, the catalytic component of the telomere-elongating enzyme telomerase, or p53 shRNA could decrease ROS by disrupting this telomere–DDR–ROS pathway. To further characterize this pathway, we performed a CRISPR/Cas9 knockout screen to identify genes that extend life span in DC cells. Of the cellular clones isolated due to increased life span, 34% had a guide RNA (gRNA) targeting CEBPB, while gRNAs targeting WSB1, MED28, and p73 were observed multiple times. CEBPB is a transcription factor associated with activation of proinflammatory response genes suggesting that inflammation may be present in DC cells. The inflammatory response was investigated using RNA sequencing to compare DC and control cells. Expression of inflammatory genes was found to be significantly elevated (P < 0.0001) in addition to a key subset of these inflammation-related genes [IL1B, IL6, IL8, IL12A, CXCL1 (GROa), CXCL2 (GROb), and CXCL5]. which are regulated by CEBPB. Exogenous TERT expression led to downregulation of RNA/protein CEBPB expression and the inflammatory response genes suggesting a telomere length-dependent mechanism to regulate CEBPB. Furthermore, unlike exogenous TERT and p53 shRNA, CEBPB shRNA did not significantly decrease ROS suggesting that CEBPB's contribution in DC cells’ senescence is ROS independent. Our findings demonstrate a key role for CEBPB in engaging senescence by mobilizing an inflammatory response within DC cells.


Introduction
Senescence is the result of cellular programming to prevent aged, damaged, or abnormal cells from continued proliferation.Cell intrinsic/extrinsic factors include telomere attrition (due to steady-state proliferation), DNA damage (oxidative stress, UV, etc.), and oncogene-induced senescence (e.g.RAS V12 ).These pathways can alter gene expression and arrest the cell cycle [e.g.senescence-associated secretory phenotype (SASP), p16 INK4A ].
Elongated telomeres suppress senescence by forming a stable secondary structure called a T-loop that prevents activation of the double-strand DNA break (DSB) repair pathway (d'Adda di Fagagna et al. 2003;Griffith et al. 1999;de Lange 2018).However, systematic telomere attrition leads to critically shortened telomeres thus initiating a DSB response.This DSB pathway involves activation of ATM/ATR followed by p53 to arrest the cell cycle via various p53-responsive genes including the cyclin-dependent kinase inhibitor p21 CDKN1A (Stein et al. 1999;Engin and Engin 2021).Sustained telomere dysfunction leads to replicative senescence via p53-p21 CDKN1A in addition to p16 INK4A -RB (d'Adda di Fagagna  et al. 2003; Smogorzewska and de Lange 2002).Furthermore, the SASP pathway can promote senescence to neighboring cells via paracrine/autocrine effects by releasing proinflammatory cytokines, chemokines, and growth factors within a niche (Acosta et al. 2013;Kuilman et al. 2008;Lopes-Paciencia et al. 2019).SASP alters the local environment through extracellular matrix remodeling and is stimulated independent of p53 and p16 activation (Coppé et al. 2011;Rodier et al. 2009).
Dyskeratosis congenita (DC) is characterized by telomere shortening due to mutations in telomerase-and telomere-related genes.A partial list of these genes includes TERC (H/ACA boxcontaining noncoding RNA template; component of the telomere-elongating enzyme, telomerase), TERT (protein/catalytic component of telomerase), DKC1 (dyskerin protein, binds H/ACA RNAs), and TINF2 (telomere-stabilizing protein interacts with telomere-binding proteins TRF1 and TRF2).The clinical DC phenotype is variable and classically consists of leukoplakia, abnormal skin pigmentation, and nail dystrophy in addition to pulmonary fibrosis and bone marrow failure.While hematopoietic stem cell (HSC) transplantation is the only curative therapy for bone marrow failure in DC, it is associated with significant posttransplant complications.Using several different cell types, including HSCs, fibroblasts, and lymphocytes from varying DC genotypes, we have found these cells display premature DNA damage responses (DDRs), resulting in elevated reactive oxygen species (ROS) (Pereboeva et al. 2016(Pereboeva et al. , 2013;;Sorge et al. 2017;Westin et al. 2011Westin et al. , 2007)).The increase in ROS appears to be related to telomere shortening/p53 activation; experimentally elongating telomeres or disrupting p53 led to a concomitant decrease in ROS and an increased cellular life span (Westin et al. 2011).
One strategy to better understand a pathway like telomeredependent senescence is to deploy a genome-wide screen.The GeCKO (genome-scale CRISPR-Cas9 knockout) lentiviral library delivers CRISPR/Cas9 to cut gRNA-targeted genes, knocking them out via insertions/deletions created by nonhomologous end joining (NHEJ) (Shalem et al. 2014).
We designed a forward genetic experiment whereby primary fibroblasts from a DC patient (TINF2 mutation) were transduced with the GeCKO library.We identified a number of candidate genes, including the transcription factor CEBPB, that regulate senescence in telomere-shortened cells.RNA sequencing (RNA-seq) revealed a significant inflammatory gene expression response (IL1B, IL6, IL8, IL12A, CXCL1 (GROa), CXCL2 (GROb), and CXCL5) in DC cells that could be corrected upon exogenous TERT (exogTERT) expression.Cells expressing exogTERT also had a decrease of both CEBPB (RNA and protein) and CEBPB-SASP-related gene expression.These findings identify new genes and downstream pathways that could promote senescence in a background of shortened telomeres.

Cell culture
This study was approved by the University of Alabama at Birmingham Internal Review Board (F100512004).Skin punch biopsies were collected from DC patients and healthy volunteers after written informed consent.DC mutations included TINF2 (R282C) and TERT (R631W) as previously reported (Pereboeva et al. 2016) in addition to DKC1 (A353V).Primary skin fibroblasts were expanded and grown in DMEM supplemented with 10% fetal bovine serum under atmospheric oxygen conditions, as previously described (Westin et al. 2011).Fibroblast colonies requiring cloning were ringcloned to permit characterization of gRNAs expressed.To this end, sterile silicone grease (Fisher Science) was applied to the bottom of sterile cloning cylinders (Fisher Science).These cloning cylinders were adhered to 10-cm cell culture plates to seal off clones from other cells.Media within cylinder was aspirated and cells PBS washed and trypsinized.Clones were further expanded in 24-well and 6-well plates to permit DNA isolation.

DNA cloning
DNA was procured from ring-cloned colonies using QIAquick DNA extraction kit (Qiagen).Isolated genomic DNA was PCR-amplified (Ex Taq, Takara) with primers flanking the gRNA common to each GeCKO.PCR-amplified DNA was cloned using the TA cloning system (Invitrogen) and transformed into DH5α bacteria.Individual bacterial colonies were isolated, DNA-extracted, and submitted for Sanger sequencing (4× bacterial colonies per cellular clone).

Oxidative stress
ROS analysis was performed as previously reported (Westin et al. 2011).Briefly, for FACS-based dihydroethidium (DHE, Sigma) ROS detection, fibroblasts were incubated in the presence of 10 µM DHE/PBS/2% FBS/5 mM sodium pyruvate for 15 min at 37°C.After 2× PBS washing, cells were trypsinized and placed on ice.ROS levels were quantified by recording the mean fluorescent intensity (MFI).As a positive control, antimycin A (10 µM, Sigma) was used to stimulate ROS production.

qRT-PCR
Expression of CEBPB was examined in cells expressing 5 different CEBPB shRNAs (Dharmacon).Isolated RNA was converted to cDNA via iScript cDNA synthesis (Bio-Rad).CEBPB and β-actin (ACTB gene) were amplified from cDNA with Power SYBR Green PCR Master Mix (Life Tech) to assess relative expression between control (scrambled shRNA) and CEBPB shRNA cells.Analysis was performed on the ViiA Real-Time PCR expression analyzer (Life Tech) using equation 2 −ΔΔCT (Livak and Schmittgen 2001).

Statistical analyses
Statistical significance was evaluated using the Student's t-test between 2 groups of data under the null hypothesis and reported by calculated P-values.Error bars found within graphs illustrate the standard deviation of DC or control samples in each experiment.

CRISPR knockout screen
We set out to identify human genes that promote senescence due to telomere dysfunction using primary skin fibroblasts from a DC patient with a single TINF2 mutation (R282C).TINF2-mutated fibroblasts were transduced with the GeCKO lentiviral CRISPR knockout library at a low multiplicity of infection (MOI) to prevent transduction of cells with multiple viruses/gRNAs.In total, 38 colonies were isolated and submitted for Sanger sequencing of the incorporated gRNAs, uncovering ∼42 gRNAs with an average 2.0 unique gRNAs per colony.To verify that the gRNAs were indeed cutting the locus of interest, genomic DNA was PCR-amplified flanking the cut site and Sanger-sequenced to verify genomic targeting.A representative figure of typical CRISPR verification is presented in Fig. 1.Sequencing of a single MED28 clone revealed an insertion/deletion (indel) at the site targeted by the gRNA.Sequencing of a CEBPB clone uncovered a 600-bp exonic deletion on 1 allele, and the inability to PCR amplify the other allele suggests a larger deletion encompassing the primers within this region.In addition to CEBPB and MED28, 2 other genes were also found repeatedly knocked out and deemed high-value.As noted in Table 1, these included gRNAs for CEBPB (13 clones), WSB1 (3 clones), MED28 (3 clones), and TP73 (3 clones).

CEBPB and inflammatory responses
The knockout screen provided evidence that short telomeres activate CEBPB to promote senescence in primary DC cells.Past research by our lab demonstrated DC senescence could be overcome upon telomere elongation by exogTERT or p53 knockdown (Westin et al. 2011).We tested whether protein levels of CEBPB were altered in DC whole-cell lysates when expressing either exogTERT or p53 shRNA in TINF2-mutated cells.Expression of exogTERT, but not p53 shRNA, caused near complete loss of CEBPB protein expression (Fig. 2a).That p53 knockdown did not reduce CEBPB suggests that upregulation of CEBPB in DC cells may occur independent of the DDR pathway.
CEBPB and CEBPB-regulated genes have been associated with a proinflammatory response and SASP (Salotti and Johnson 2019).To investigate whether CEBPB-regulated genes were altered in primary DC cells, we performed RNA-seq analyses on DC fibroblasts with different genotypes (DKC1, TINF2, and TERT).Each genotype was transduced with vector control, exogTERT or p53 shRNA, and examined for gene expression changes by pooling genotypes together.DC cells expressing exogTERT showed a 40% decrease in CEBPB expression (P < 0.01; Fig. 2b) but not shp53.A significant increase in SASP-related, CEBPB-dependent genes were found in primary DC cells compared to controls (white vs black bars; Fig. 2c).Furthermore, this inflammatory profile was downregulated with exogTERT (black vs gray bars).Of interest, the inflammatory response profile persisted when p53 shRNA was expressed (black vs dotted bars).Expression of another SASP factor, CSF3 (G-CSF), was the most variable among all genotypes in the targeted analysis of CEBPB-regulated genes.Its expression among the 3 DC genotypes averaged a 37-fold Table 1.Key genes identified from knockout library.increase (P < 0.001) and decreased an average of ∼20× in exogTERT cells (P < 0.05; Fig. 2d).

Gene
GSEA was performed on the RNA-seq data sets and provided further evidence that primary DC cells have a significant underlying inflammatory response (Fig. 3a; positive enrichment score: 1.54, P < 0.0001).Using all 3 genotypes, a pairwise comparison of DC cells and those expressing exogTERT found a statistically significant difference (Fig. 3b; positive enrichment score: 1.53, P < 0.0001).However, knocking down p53 in DC cells created a more robust inflammatory response than untreated DC cells (Fig. 3c; negative enrichment score: −1.98, P < 0.0001).Together these data suggest that DC cells mobilize CEBPB to augment the expression of SASP-related genes to promote an inflammatory response.The concomitant decrease in CEBPB gene/protein expression and genes regulated by CEBPB in exogTERT cells suggests the reduced SASP/inflammatory response is likely telomere length-dependent.

CEBPB knockdown does not impact ROS
Telomere elongation or p53 knockdown can both extend the life span of primary DC cells and decrease ROS levels (Westin et al. 2011).We hypothesized that if CEBPB was integral to the short telomere-p53-ROS pathway, then decreasing CEBPB expression would also reduce ROS.We expressed CEBPB shRNA in DC cells and analyzed ROS using DHE staining and FACS analysis.As noted in Fig. 4a, although we found a significant decrease in shRNA-targeted CEBPB expression by qRT-PCR (83% reduction, P < 0.001), there was no significant decrease in ROS within these same cells (Fig. 4b).Decreases in ROS could be found in both DC-exogTERT and DC-shp53-expressing cells (P < 0.001), as previously reported.These results provide evidence that primary DC cells may have multiple, redundant, prosenescence pathways that are either DDR/ROS-dependent or CEBPB-dependent.

Discussion
The mechanism by which telomere shortening promotes senescence is not entirely understood.DC, a well-described telomere biology disorder, consists of pathologies related to premature aging/cellular senescence.Causative DC mutations are found in genes related to telomere stability and maintenance (Mason and Bessler 2011;Tummala et al. 2015;Dokal et al. 2015;Revy, Kannengiesser, and Bertuch 2023).Since telomere dysfunction is the critical factor in promoting senescence, DC cells are useful to study this relationship.We have previously found proliferative defects, activated DDR, and increased ROS in several cell types and mutations from DC patients (Pereboeva et al. 2016(Pereboeva et al. , 2013;;Sorge et al. 2017;Westin et al. 2011Westin et al. , 2007)).We showed that shortened telomeres activate DDR/p53 leading to increased ROS.In addition, telomere elongation (exogTERT) or blocking DDR (p53 shRNA) rescued proliferation and decreased ROS (Westin et al. 2011).
The GeCKO library provided a tool to systematically knock out genes in DC cells to recover those critical for senescence.The CEBPB gRNA repeatedly appeared in this CRISPR KO screen, signifying its importance (Table 1).CEBPB is a member of a family of CEBP proteins that contain a basic leucine zipper required for binding and initiating/repressing gene transcription (Ramji and Foka 2002).CEBPB is well characterized and key to immune/ inflammatory responses and when overexpressed has potent cytostatic activity (Johnson 2005;Freund et al. 2010).CEBPB has a known role in regulating genes involved in the inflammatory Fig. 3. Elevated inflammatory response in DC cells, suppressed DC cells expressing exogenous TERT.RNA-seq analysis provided expression changes to perform GSEA (Mootha et al. 2003, Subramanian et al. 2005) on control and DC skin fibroblasts.The enrichment score analysis presents the ranked order of genes from overexpressed (left) to underexpressed (right).The more or less enriched a particular gene set is compared to a control set, the more positive/ negative the normalized enrichment score (NES) and thus the greater the P-value.DC samples from patients with 3 different genotypes (TINF2, DKC1, and TERT mutations) were individually analyzed and averaged to determine a "DC expression signature".a) DC cells had a positive enrichment score for inflammatory response when compared to control or b) TERT-expressing DC fibroblasts (exogTERT; NES 1.54/P < 0.0001 and NES 1.53/P < 0.001 respectively, also from TINF2, DKC1, and TERT mutations).c) DC cells had a negative enrichment score when compared to p53 shRNA-expressing DC cells (NES −1.98/P < 0.0001, all 3 genotypes).
response [e.g.IL6, IL8, IL1B, IL12A, CXCL1 (GROa), CXCL2 (GROb), CXCL5, CSF3 (G-CSF)] and others (Akira et al. 1990;van der Krieken et al. 2015;Salotti and Johnson 2019;Freund et al. 2010).Our data also found CEBPB RNA and protein expression were suppressed in response to exogTERT.This could provide an explanation and a link for the concomitant suppression of the underlying inflammatory response in DC cells (Fig. 2a).Our previous research identified that both DDRs and ROS respond to telomere length (Westin et al. 2011).This previous research found (1) DDR and ROS were ameliorated with telomere elongation, bypassing senescence, and (2) ROS could be decreased by blocking p53 signaling in a background of shortened telomeres (also bypassing senescence).This allowed the design of a simple hypothesis: shortened telomeres increase DDR and therefore ROS to promote and enforce senescence.In this research, we have amended our hypothesis with the finding that telomere elongation decreases inflammatory response, independent of DDR and ROS.These data suggest that (1) CEBPB functionality lies downstream of telomere signaling, but (2) CEBPB activity is independent of p53 activation.Finally, the fact that similar RNA expression was found among all DC genotypes, including with exogTERT and shp53 expression, suggests that CEBPB and the inflammatory phenotype are telomere length-dependent and independent of DC genotype.Aging-related knockout screens have also been performed elsewhere but did recover CEBPB (Liu et al. 2019).This screen was designed a priori to knockout genes related to senescence, however, and it is not clear if CEBPB was part of the screen.Consistent with our findings, this screen did find the involvement of inflammation.The proinflammatory SASP includes gene and protein expression changes of secreted proteins (i.e.growth factors, interleukins, and chemokines) that are responsible for alterations in the local microenvironment through paracrine effects (Coppé et al. 2010).Another study examined 9 cytokine levels in serum of patients diagnosed with inherited bone marrow failure disorders.They found that, in general, these proteins were not elevated in DC patients (Giri et al. 2015).However, upon stratification of DC patients based on the presence of severe bone marrow failure, significant levels of CSF3 (G-CSF) were found, consistent with our results (Fig. 2d).The presence of an intrinsically upregulated inflammatory response suggests that current antiinflammatory approaches could be repurposed to attenuate disease pathologies found in DC patients.
The GeCKO screen also recovered a gRNAs targeting MED28, p73, and WSB1 gRNAs.MED28 is a subunit of the Mediator complex, a large ∼26 submit complex that regulates transcription.Mediator acts as a bridge between the activation domains of transcription factors and RNA polymerase II in response to intracellular signaling (Allen and Taatjes 2015).CEBPB differentially binds the Mediator complex depending on whether it is transcriptionally active (Mo et al. 2004), suggesting a common mechanism for CEBPB and MED28 knockouts.p73 is a homolog to p53 and has significant sequence and functional overlap.p73 can bind canonical p53 binding sites, oligomerize with p53 protein (Dötsch et al. 2010), and induce cell-cycle arrest, apoptosis, and senescence (Dötsch et al. 2010).Based on our previous experiments using p53 shRNA, we predicted knockout of p53 would be recovered in this screen.Of note, 1 clone included a p53-targeting gRNA (data not shown) confirming the capacity of the screen to return credible gene candidates.Finally, knockout of WSB1 may have imparted and extended life span via an E3 ubiquitin ligase that targets and destabilizes ATM through ubiquitination in 293T and MEF cells (Kim et al. 2017).
We cannot rule out that alternative explanations may account for increased life span in colony knockouts.Gene knockouts may have led to the activation of endogenous TERT expression or blocked a functional DDR from taking place.It is also possible that gRNAs may have targeted and knocked out unanticipated genes; however, off-target minimization was a consideration designing the library (Shalem et al. 2014).Each gRNA had possible off-targets though no off-target sequence had 100% homology to the gRNA.These alternative explanations will require further investigation.
The recovery of CEBPB and other genes within our CRISPR screen indicates that additional mediators of senescence exist that may have not been previously appreciated.CEBPB and the downstream SASP response appear to be influenced by TERT activity and not DDR according to our p53 shRNA/ROS experiments.Although initiated by dysfunctional telomeres, the CEBPB and DDR/ROS pathways likely belong to independent but redundant pathways each contributing to senescence.A model for the relationship among shortened telomeres, DDR, and CEBPB/inflammation is presented in Fig. 5. Whether TP73, MED28, and WSB1 exist within this pathway or elsewhere is a question for future research.It is possible DC disease pathology could arise from an excessive or chronic inflammatory response.Limited therapeutic options exist in DC, and novel approaches are needed.Links have been found between inflammation and telomere dysfunction in mice (Jurk et al. 2014) and zebrafish (Lex et al. 2020), but future research will need to investigate inflammation in DC animal models and assess how pathogenic mutations impact various cell compartments and organ systems.If DC phenotypes do in fact arise as a result of CEBPB/SASP activation from short telomere signaling, manipulation of this pathway may be an interesting approach toward disease amelioration.Deposition of 53BP1 and γ-H2AX is found on critically shortened telomeres leading to a DDR signaling pathway.This pathway activates ATM/ATR and p53 and ultimately leads to increased ROS and senescence.Alternatively, data supported by this research support the activation of a second pathway that includes mobilization of CEBPB and CEBPB-regulated genes related to SASP.When telomeres are elongated or CEBPB is knocked out, SASP gene expression is likely suppressed affording an extension of life span to these cells.Possible interplay between CEBPB, p73, and other genes found in the knockout screen are possible.
knockouts from 38 cellular clones found multiple occurrences of the same gene.For example, CEBPB was found 13 times within 38 clones, 8 times of which it was the only gRNA uncovered (8 clones with single gRNA + 5 clones with CEBPB/addition gRNA(s) = 13).

Fig. 4 .
Fig. 4. CEBPB downregulation does not impact ROS.TINF2-mutated DC cells were infected with 5 different CEBPB shRNAs and examined for an expression decrease and presence of ROS (representative shRNA knockdown presented in figure).a) Expression of CEBPB was determined by qRT-PCR (8.5× decrease bars).b) ROS levels were examined by staining cells with dihydroethidium (DHE) and quantified by FACS analysis using the MFI.Exogenous TERT (exogTERT) and p53 shRNA-expressing DC cells provided as controls to compare to CEBPB shRNA-expressing cells.Levels were normalized to the empty vector control.Significance was determined by Student's t-test; ***P < 0.001; N.S.: not significant.

Fig. 5 .
Fig. 5. CEBPB increases inflammation/SASP independent of DDR pathway to promote senescence.Schematic of telomere-related senescence in DC cells.Deposition of 53BP1 and γ-H2AX is found on critically shortened telomeres leading to a DDR signaling pathway.This pathway activates ATM/ATR and p53 and ultimately leads to increased ROS and senescence.Alternatively, data supported by this research support the activation of a second pathway that includes mobilization of CEBPB and CEBPB-regulated genes related to SASP.When telomeres are elongated or CEBPB is knocked out, SASP gene expression is likely suppressed affording an extension of life span to these cells.Possible interplay between CEBPB, p73, and other genes found in the knockout screen are possible.