Mutant KRAS induces intrinsic ISG expression

To explore the biological pathways that are perturbed by oncogenic RAS signaling, we performed gene set enrichment analysis (GSEA) (Powers et al., 2018) using genes that were differentially expressed in our mutant KRAS AALE cells. GSEA revealed that the three most significantly enriched pathways were the IFN-α and -γ responses, as well as the hallmark inflammatory response (Figure 1B), along with increased KRAS signaling from mutant KRAS(G12D), increased metabolic gene expression, and decreased expression of epithelial-to-mesenchymal transition (EMT) genes (Figure 1B).

To further validate the connection observed between mutant KRAS and ISG expression, we compared mutant KRAS-induced ISGs in AALE cells to those that were induced in human bronchial epithelial cells (HBECs) in response to mutant KRAS(G12V) (Figure S1C). We observed a strong concordance between mutant KRAS-induced ISGs in AALE and HBEC cells (Figure 1C), confirming our previous results. We then examined the promoter regions (±500 bp) of upregulated ISGs and identified motifs enriched in comparison to non-differentially expressed (DE) ISGs (Figure 1D), including the key IFN response regulators IRF1, IRF7, and STAT2 (Jefferies, 2019). To determine the in vivo relevance of our findings in both mutant KRAS AALE and HBEC cells, we examined ISG expression in mutant KRAS(G12D) lung adenocarcinomas (LUAD) from The Cancer Genome Atlas (TCGA), which revealed a subset of ISGs that were upregulated in KRAS(G12D) tumors when compared to lung cancer samples with wild-type (WT) KRAS (Figure 1E). These results indicate that mutant KRAS signaling activates an intrinsic ISG response in lung cells both in vitro (AALE, HBEC) and in vivo (TCGA LUAD).

Epigenetic reprogramming of ISGs by mutant KRAS

To elucidate potential mechanisms involved in inducing ISG signatures in mutant KRAS AALE cells, we performed the assay for transposase-accessible chromatin using sequencing (ATAC-seq) (Buenrostro et al., 2015). In mutant KRAS AALEs, open chromatin was significantly enriched at gene promoters for upregulated ISGs (Figure 2A). Open chromatin peaks were uniquely present in mutant KRAS AALEs when compared to control AALEs at 183 transcriptional start sites (TSSs), including 11 ISGs that were specifically and significantly upregulated by mutant KRAS signaling (Figure 2B). In addition, we observed strong enrichment of ATAC signal at the TSS of the significantly upregulated IRF9 gene, which forms the ISGF3 transcription factor (TF) complex with STAT1 and STAT2 (MacMicking, 2012), and also strong enrichment at the TSS of IRF7, IFI27, OAS2, IFI44, and MX1 (Figure 2C). In conjunction with the motif enrichment analysis (Figure 1D), these results show that oncogenic KRAS signaling induces the epigenetic activation of ISG TFs and their downstream ISG targets.

Open in a separate window

Figure 2.

Mutant KRAS signaling mediates epigenomic reprogramming of ISGs

(A) Mean ATAC-seq counts per million (CPM) (95% confidence interval [CI]) in promoter regions of upregulated ISGs (log2 fold change >1.5) in both mutant KRAS and control (CTRL) AALEs.

(B) Differential expression of ISGs with unique peaks near TSS (only present in mutant KRAS or control AALEs).

(C) ATAC-seq coverage in both mutant KRAS and CTRL AALEs for subset of ISGs with unique peaks detected near TSS.

(D) Significant Gene Ontology (GO) term enrichment over unique peaks detected in mutant KRAS AALEs as determined by genomic regions enrichment of annotations tool (GREAT) analysis.

The genome-wide effects of mutant KRAS-mediated epigenomic reprogramming were further assessed with the Genome Regions Enrichment of Annotations Tool (GREAT) (McLean et al., 2010). GREAT analysis orthogonally confirmed the enrichment of accessible chromatin regions near ISGs and showed the enrichment of related molecular functions, including double-stranded RNA binding. Notably, the cellular components most enriched were extracellular in nature, including extracellular vesicle and extracellular exosome (Figure 2D).

Mutant KRAS reprograms the extracellular transcriptome

To test whether extracellular RNAs secreted from mutant KRAS cells also exhibit differential expression relevant to intracellular reprogramming events, we isolated extracellular vesicles from the culture media of control and mutant KRAS AALEs (Enderle et al., 2015; Liu et al., 2017). Extracellular vesicles isolated from mutant KRAS AALEs comprised different sized vesicles that were ~90, ~150, and ~213 nm in diameter, while vesicles from control AALE media were predominantly ~196 nm in size (Figure 3A).

Open in a separate window

Figure 3.

Mutant KRAS signaling induces secretion of TE RNAs and ISGs in EVs

(A) Size distribution of extracellular vesicles (EV) isolated from control (CTRL) and mutant KRAS AALEs.

(B) Volcano plot of differentially secreted GENCODE protein-coding RNAs and lncRNAs between mutant KRAS and CTRL AALE EVs.

(C) Scatterplot comparing differentially expressed genes between intracellular and extracellular mutant KRAS AALE RNA-seq libraries; linear regression fit with formula and goodness of fit displayed.

(D) Upset plot summarizing overlap of differentially expressed upregulated (up) and downregulated (dn) genes across in and ex contexts.

(E) Significantly enriched gene sets detected in both in and ex contexts.

(F) Differential secretion of TE RNAs in EVs from mutant KRAS AALEs when compared to control AALE EVs.

Ex, extracellular; in, intracellular.

RNA isolated and sequenced from these vesicles exhibited mutant KRAS-dependent differential expression of both protein-coding genes (n = 17 upregulated, n = 140 downregulated) and lncRNA (n = 5 upregulated, n = 8 downregulated) (Figures 3B and S2A). We also observed significant correlation between differentially expressed ISGs in our intracellular and extracellular RNA-seq datasets that largely agreed with intracellular epigenetic changes (IFI6, MX1, IFI27, and OASL) (Figures 3C and ​and3D).3D). Furthermore, GSEA showed that IFN-α and -γ signatures were enriched in both intracellular and extracellular RNA (Figure 3E), indicating that extracellular RNAs reflect intracellular ISG changes due to mutant KRAS signaling.

To determine the effects of oncogenic KRAS on noncoding RNA secretion, we also characterized the TE RNAs that were preferentially packaged and released in extracellular vesicles. We found significant upregulation of predominantly long terminal repeat (LTR) RNAs such as LTR12, MER11C, and LTR27C, along with LINE, DNA, and Satellite repeat RNAs in mutant KRAS AALE extracellular vesicles (Figure 3F). Moreover, TE RNAs represented approximately 50% of the extracellular RNA released from mutant KRAS AALE cells, suggesting their preferential secretion in extracellular vesicles (Figures S2C and S2D).

Regulation of TE RNAs by mutant KRAS

Given the prevalence of secreted TE RNAs, we investigated intracellular TE RNA dynamics in response to mutant KRAS signaling in AALE cells. Analogous to extracellular RNAs, LTR RNAs were among the most significantly upregulated TE RNAs in response to oncogenic KRAS signaling, including LTR12C RNAs (Figure S3A). In addition, LINE RNAs such as L1MEc and DNA element RNAs such as Tigger5 were also significantly enriched in mutant KRAS AALEs (Figure S3A). Furthermore, we examined TE RNAs in mutant KRAS HBEC cells, which similarly exhibited significant upregulation of TE RNAs in response to mutant KRAS when compared to control HBECs (Figure S3A). LTR12C RNAs were again the most significantly upregulated TE RNAs in mutant KRAS HBEC cells (Figure S3A), further validating our intracellular and extracellular RNA analyses in mutant KRAS AALE cells.

Based on the functions of KZNF genes in silencing TE RNAs in other contexts (Imbeault et al., 2017), we examined whether KZNFs could be involved in TE RNA regulation in both mutant KRAS AALE and HBEC cells. Given the broad downregulation of KZNFs in mutant KRAS AALEs (Figure 1A), we also analyzed KZNF expression in mutant KRAS HBECs, which similarly exhibited significant downregulation of KZNFs, many of which overlap with KZNFs downregulated in mutant KRAS AALEs (Figure S3B). To determine the potential relationship between our upregulated TE RNAs and our downregulated KZNFs, we looked for significantly enriched motifs in TE RNAs using a previously described KZNF-specific motif set (Barazandeh et al., 2018), which confirmed the presence of binding motifs for significantly downregulated KZNFs in the significantly upregulated TE RNAs (Figure S3C). We also used the KNZF binding scores generated from previous chromatin immunoprecipitation sequencing (ChIP-seq) experiments (Imbeault et al., 2017) to rank TE RNAs targeted by KZNFs, finding that many of the upregulated TEs were among the top 10–20 targets of downregulated KZNFs in mutant KRAS AALEs (Figure S3D), their extracellular vesicles (Figure S4A), and in mutant KRAS HBECs (Figure S4B). We then computed the average log2 fold change of downregulated ZNFs with putative binding sites within upregulated TE RNAs, which confirmed a negative association across all three contexts of mutant KRAS transcriptional profiling (Figure S3E). These analyses point to a coordinated, TE-KZNF axis that is dysregulated by mutant KRAS.

KZNFs repress TE RNAs and ISGs activated by mutant KRAS

To explore the mechanistic relationship between KZNFs and TE RNA expression, we examined mutant KRAS A549 lung cancer cells that overexpress ZNF257 or ZNF682 (Ito et al., 2020), both of which we found to be significantly downregulated by mutant KRAS signaling in AALE cells and putative regulators of dysregulated TE families (Figures 1A and S3). Differential expression analysis of RNA-seq data indicated significant downregulation of ISGs OAS1 and IRF9 in mutant KRAS A549 cells overexpressing either ZNF257 or ZNF682 (Figure S5A), as well as significant downregulation of TE RNAs that were upregulated by mutant KRAS in AALE cells (Figure S5B). These findings directly connect mutant KRAS-regulated KZNFs with control of TE RNA and ISG expression.

Epigenetic silencing of KZNFs regulated by mutant KRAS signaling

To determine the extent to which mutant KRAS signaling epigenetically silences KZNF expression, we examined ATAC-seq data for all significantly downregulated KZNF loci. We found that mutant KRAS signaling substantially reduces chromatin accessibility at TSS regions (Figure 4A). When we examined genes with ‘‘unique’’ ATAC peaks that were only present in control AALEs but disappeared in mutant KRAS AALEs, we found that many of these genes were KZNFs that were significantly downregulated (Figure 4B). Six of these downregulated KZNFs, ZNF90, ZNF826P, ZNF736, ZNF471, ZNF682, and ZNF853, had peaks unique to control AALEs (Figure 4C). Downregulated KZNF TSS regions were enriched in motifs for ETS (ETV1) and ELK (ELK1) TFs (Figure 4D), known downstream effectors of the RAS signaling pathway (Simanshu et al., 2017).

Open in a separate window

Figure 4.

Mutant KRAS signaling epigenetically silences KZNFs in vitro

(A) Mean ATAC-seq CPM (95% CI) in promoter regions of downregulated KZNFs (<–4.5 log2 fold change) in both mutant KRAS and control (CTRL) AALEs.

(B) Differential expression of KZNFs with unique peaks near TSS (only present in mutant KRAS or control AALEs).

(C) ATAC-seq coverage in both KRAS and CTRL AALEs for subset of KZNFs with unique peaks detected near TSS.

(D) Volcano plots of differentially expressed TFs in mutant KRAS AALEs with significant TF motif enrichment in downregulated KZNF gene promoters. chr, chromosome.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Immortalized lung epithelial cells (AALE cells; XX), derived at Dana-Farber and immortalized by SV40 large-T antigen (Lundberg et al., 2002) were obtained as a gift from the laboratory of Eric Collison (University of California, San Francisco). The AALE stable cell lines pBABE-mCherry Puro (control) (Lu et al., 2017) and pBABE-FLAG-KRAS(G12D) Zeo (mutant KRAS) were generated using retroviral transduction, followed by selection in puromycin or zeocin. Cells were cultured at 37°C and 5% CO₂ in SABM Basal Medium (Lonza SABM basal medium, CC-3119) with supplements and growth factors (Lonza SAGM SingleQuots Kit Suppl. & Growth Factors, CC-4124).

HBEC3kt cell lines (HBEC cells; XX) were obtained as a gift from the laboratory of Harold Varmus (National Human Genome Research Institute and Weill Cornell Medicine). The HBEC stable cell lines pLenti6/V5-GW/lacZ (control) (Vikis et al., 2007) and pLenti-KRASV12 (mutant KRAS) were generated using lentiviral transduction, followed by selection in blasticidin. Lentiviral plasmids were obtained as a gift from the laboratory of John Minna (The University of Texas Southwestern Medical Center) (Sato et al., 2013; Vikis et al., 2007). Cells were cultured at 37°C and 5% CO₂ in Keratinocyte Serum-Free Media (KSFM) with supplements (Invitrogen, #17005042).

METHOD DETAILS

QUANTIFICATION AND STATISTICAL ANALYSIS

All quantitative data for functional assays have been reported as means ± standard deviation. Statistical significance for these were calculated using a Wilcox-test (R – wilcox.test()) unless otherwise noted and p values <0.05 were considered significant. All statistical analyses were performed with R (version 4.1.1) running from the Rocker ‘Tidyverse’ Docker container (rocker/tidyverse:4.1.1). Linear regression was carried out with the lm() function.

We thank members of the Kim lab, the Demirci lab, the Brooks lab, the Collisson lab, the Haussler lab, and the Carpenter lab for helpful discussions. This work was supported by funds from the Baskin School of Engineering at University of California, Santa Cruz (to D.H.K.), the Ken and Gloria Levy Fund for RNA Biology (to D.H.K.), the Department of Defense Congressionally Directed Medical Research Program (W81XWH-20–1-0746 to U.D. and D.H.K.), and the National Cancer Institute (R01CA227807, R01CA239604, R01CA230263 to E.C.). R.E.R. is supported by the National Institutes of Health (1F99DK131504–01), S.V.M. is supported by the California Institute for Regenerative Medicine (EDUC4–12759), and V.P. (T32DT4904) and D.C. (T30DT0997) are supported by the Tobacco-Related Disease Research Program.