Research ArticleCytokines

IFN-γ–inducible antiviral responses require ULK1-mediated activation of MLK3 and ERK5

See allHide authors and affiliations

Science Signaling  20 Nov 2018:
Vol. 11, Issue 557, eaap9921
DOI: 10.1126/scisignal.aap9921

IFN-γ activates alternative pathways

The cytokine IFN-γ has potent antiviral activity, and the kinase ULK1 is a key initiator of cellular catabolism through autophagy. Saleiro et al. found that IFN-γ receptor stimulation activated an alternative ULK1-dependent pathway required for antiviral activity. IFN-γ activated a ULK1-dependent cascade that involved activation of the kinases MLK3 and ERK5. The kinase function of ULK1 was required for the transcription of IFN-stimulated genes independently of the autophagy mediators Beclin-1 and VPS34. These data suggest that autophagy-independent functions of ULK1 are important for IFN-γ–mediated resistance to infection.

Abstract

It is well established that activation of the transcription factor signal transducer and activator of transcription 1 (STAT1) is required for the interferon-γ (IFN-γ)–mediated antiviral response. Here, we found that IFN-γ receptor stimulation also activated Unc-51–like kinase 1 (ULK1), an initiator of Beclin-1–mediated autophagy. Furthermore, the interaction between ULK1 and the mitogen-activated protein kinase kinase kinase MLK3 (mixed lineage kinase 3) was necessary for MLK3 phosphorylation and downstream activation of the kinase ERK5. This autophagy-independent activity of ULK1 promoted the transcription of key antiviral IFN-stimulated genes (ISGs) and was essential for IFN-γ–dependent antiviral effects. These findings define a previously unknown IFN-γ pathway that appears to be a key element of the antiviral response.

INTRODUCTION

Interferons (IFNs) are secreted cytokines that act as the first line of defense against viral infections (1). IFN-inducible activation of signal transducer and activator of transcription 1 (STAT1) and STAT2 is important for an antiviral response associated with type I and type III IFNs (2). By contrast, there is little information on the signaling effectors that mediate the antiviral effects of the type II IFN, IFN-γ (3). IFN-γ initiates its biological activities by binding to its cognate heterodimeric cell surface receptor, composed of two ligand-binding IFN-γ receptor 1 (IFNGR1) subunit chains and two signal-transducing IFNGR2 chains (4). This interaction triggers activation of several downstream signaling events, including the canonical Janus kinase (JAK)–STAT1 pathway, and non-STAT pathways, such as mitogen-activated protein kinase (MAPK), phosphoinositide 3-kinase (PI3K), mechanistic target of rapamycin complex 2 (mTORC2), calcium/calmodulin-dependent protein kinase II (CaMKII), and nuclear factor κB (NF-κB) cellular pathways (3, 58). Although IFN-γ–mediated activation of STAT1 is required for several IFN-γ–dependent antiviral responses, some of these activities are STAT1 independent, which implies that there may be activation of alternate and complementary pathways by IFN-γ stimulation during a viral infection (3).

The human serine/threonine kinase ULK1 (Unc-51–like kinase 1) is the human homolog of Saccharomyces cerevisiae autophagy-related protein kinase Atg1, which promotes autophagy (9). However, ULK1 activity is required for type I IFN–induced activation of p38 MAPK and for its antiproliferative effects in normal hematopoiesis and myeloproliferative neoplasms (9). In light of these findings, we questioned whether ULK1 could also be activated downstream of the type II IFN receptor (IFNR), IFNGR, and whether ULK1 is required for IFN-γ–mediated antiviral activity. We found that IFN-γ–dependent engagement of ULK1 promoted activation of downstream IFN-γ–dependent signaling events and transcription of key antiviral IFN-stimulated genes (ISGs). We identified that mixed lineage kinase 3 (MLK3) interacted with and was phosphorylated by ULK1 after engagement of the IFNGR. This study defines the molecular mechanism by which ULK1 activates IFN-γ–mediated antiviral responses.

RESULTS

IFN-γ treatment promotes ULK1-dependent MLK3 activation

We speculated that IFN-γ activation of the IFNGR may engage ULK1, which could, in turn, interact with other cellular signaling effectors. To identify potential IFN-γ–dependent binding partners of ULK1, we performed nano–liquid chromatography (nLC) and tandem mass spectrometry (MS/MS) analysis of endogenous protein-ULK1 complexes immunoprecipitated from untreated and IFN-γ–treated KT-1 leukemia cells. We found that ULK1 interacted with 35 proteins under untreated conditions, 75 proteins under both untreated and IFN-γ–treated conditions, and 34 unique proteins after IFN-γ treatment for 10 min (Fig. 1A). Pathway and process enrichment analysis of the putative ULK1 interactors in KT-1 cells identified that six known IFN-γ signaling pathway–related proteins bound to ULK1 after IFN-γ stimulation (Fig. 1B, red asterisk, and table S1). In addition, a separate pathway and process enrichment analysis of the 34 putative ULK1 interactors after IFN-γ treatment revealed enrichment of 20 proteins associated with viral processes (Fig. 1C, red asterisk, and table S2) and indicated that MAP kinase kinase kinase 11 (MAP3K11, also MLK3) was involved in both IFN-γ signaling and virus-associated processes (tables S1 and S2). When we probed the binding interaction between ULK1 and MLK3 by coimmunoprecipitation followed by immunoblotting in IFN-γ–sensitive KT-1 and U937 cells, we found that ULK1 and MLK3 interacted in both unstimulated and IFN-γ–treated cells (Fig. 2, A and B). Furthermore, in vitro kinase reactions between ULK1 and heat-inactivated MLK3 increased the adenosine triphosphate (ATP) consumption in vitro (Fig. 2C), and loss of Ulk1/2 in mouse embryonic fibroblasts (MEFs) inhibited IFN-γ–induced phosphorylation of MLK3 within its activation loop (10) at Thr277 and Ser281 (Fig. 2D). These data suggest that MLK3 is activated downstream of ULK1 during engagement of the IFNGR.

Fig. 1 Putative binding partners of ULK1 after engagement of the IFNGR.

(A to C) MS/MS analysis of protein-ULK1 complexes from untreated (UT) or IFN-γ–treated KT-1 cells. Venn diagram (A) indicates the number of proteins that interact with endogenous ULK1 under untreated (blue), both (black), and after 10-min IFN-γ treatment (red) conditions. Heatmap analysis (B) is of the putative binding partners of ULK1 identified in untreated, IFN-γ–treated (IFN-γ), and both groups (see also table S1). Ontology analysis (C) is of putative ULK1 interactors identified only after IFN-γ stimulation (see also table S2). Note that red rectangles and asterisks highlight (B) IFN-γ signaling pathway and (C) viral process terms. Data are from one experiment. ncRNA, noncoding RNA; IP, immunoprecipitation.

Fig. 2 ULK1 interacts with and phosphorylates MLK3 during engagement of the IFNGR.

(A and B) Coimmunoprecipitation analysis of ULK1 interaction with MLK3 in KT-1 (A) or U937 (B) cells left untreated or treated with IFN-γ for 10 min, as indicated. Blots are representative of three independent experiments. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; RIgG, rabbit immunoglobulin G. (C) ADP-Glo kinase assay analysis of adenosine diphosphate (ADP) concentration produced by in vitro kinase reaction of recombinant human ULK1 and heat-inactivated MLK3 alone or in combination. Data are means ± SEM of three independent experiments performed in triplicate. RLU, relative luminescence units. (D) Western blot analysis of pMLK3 in lysates from Ulk1/2+/+ and Ulk1/2−/− MEFs treated with IFN-γ for 10 or 30 min, as indicated. Blots (top) are representative of five independent experiments. Quantified data (bottom) are means ± SEM pooled from all experiments. *P < 0.05, ***P < 0.001, and not significant (NS) by one-way analysis of variance (ANOVA), followed by Tukey’s multiple comparisons test (C) or two-way ANOVA using Bonferroni correction (D). a.u., arbitrary units.

IFN-γ–dependent signaling events require ULK1/2 activity

MLK3, a member of the MAP3K family, is a Ser/Thr protein kinase essential for activation of several MAPK signaling cascades in response to different stimuli (1113). Specifically, in response to IFN-γ, MLK3 is required for the activation of the transcription factor CCAAT/enhancer-binding protein-β (14). However, it is not known whether MLK3 promotes activation of MAP kinases after engagement of the IFNGR. Thus, we evaluated the importance of MLK3 for IFN-γ–induced phosphorylation of the MAP kinases ERK5 (extracellular signal–regulated kinase 5), ERK1/2, and JNK (c-Jun N-terminal kinase), using Mlk3+/+ and Mlk3−/− MEFs (12). IFN-γ–mediated activation of ERK5 by phosphorylation of the Thr-Glu-Tyr motif within the activation loop of its kinase domain (15) (Thr218/Tyr220) was defective in Mlk3−/− MEFs when compared with Mlk3+/+ MEFs (Fig. 3A). In contrast, IFN-γ–induced phosphorylation of ERK1/2 and JNK were unaffected in Mlk3−/− cells (Fig. 3, B and C). When we examined whether ULK1 is required for IFN-γ–dependent phosphorylation of ERK5 kinase in Ulk1/2+/+ and Ulk1/2−/− MEFs (16), we found that ERK5 phosphorylation in MEFs is defective in the absence of Ulk1/2 (Fig. 4A). However, there was some increased baseline phosphorylation of ERK5 in Ulk1/2+/+ MEFs compared to Ulk1/2−/− MEFs (Fig. 4A). Furthermore, IFN-γ stimulation resulted in phosphorylation of both ERK1/2 and JNK kinases, as well as STAT1, independently of Ulk1/2 presence (Fig. 4, B and C, and fig. S1). Together, our results suggest that ULK1/2 activity is required for IFN-γ–mediated phosphorylation of MLK3, which, in turn, is required for IFN-γ–induced activation of ERK5.

Fig. 3 IFN-γ stimulates MLK3-mediated activation of ERK5, but not ERK1/2 and JNK.

(A to C) Western blot analysis of pERK5 (A), pERK1/2 (B), and pJNK (C) in lysates from Mlk3+/+ and Mlk3−/− MEFs treated with IFN-γ for 10 or 30 min, as indicated. Blots (left) are representative of three (C) or four (A and B) independent experiments. Quantified data (right) are means ± SEM pooled from all experiments. *P < 0.05, ***P < 0.001, and not significant by two-way ANOVA using Bonferroni’s correction.

Fig. 4 ULK1/2 promotes IFN-γ–induced phosphorylation of ERK5 and p90RSK1.

(A to D) Western blot analysis of pERK5 (A), pERK1/2 (B), pJNK (C), and p-p90RSK1 (D) in lysates from Ulk1/2+/+ and Ulk1/2−/− MEFs treated with IFN-γ for 10 or 30 min, as indicated. Blots (left) are representative of three independent experiments. Quantified data (right) are means ± SEM pooled from all experiments. *P < 0.05, **P < 0.01, ***P < 0.001, and not significant by two-way ANOVA using Bonferroni’s correction.

Once activated, ERK5 plays a critical role as a master regulator of transcription by promoting the activation of many transcription factors and protein kinases, including p90 ribosomal S6 kinase 1 (p90RSK1) (16, 17). The p90RSK1 Ser/Thr kinase modulates several biological processes by increasing both transcription and mRNA translation of specific genes (1820). IFN-γ–induced phosphorylation of p90RSK1 was defective in Ulk1/2−/− MEFs compared to Ulk1/2+/+ MEFs (Fig. 4D). Together, these results suggest that ULK1/2 may play a role in the regulation of IFN-γ–induced activation of ERK5 and p90RSK1.

ULK1/2 activity is essential for transcription of key IFN-γ–inducible antiviral genes

To determine whether ULK1/2 activity was required for IFN-γ–dependent gene transcription, we performed high-throughput single-end RNA sequencing (RNA-seq) of total RNA isolated from untreated and IFN-γ–treated Ulk1/2+/+ and Ulk1/2−/− MEFs. When we used multidimensional scaling (MDS) on the expression of the top 500 genes in all samples to determine how distinct and reproducible the samples were, we found that all four biological replicates in each group clustered together (Fig. 5A). We identified those genes that were differentially expressed after IFN-γ treatment using edgeR (21). This analysis revealed that the expression of 793 genes was decreased and that of 644 genes was increased in IFN-γ–treated Ulk1/2+/+ MEFs compared with IFN-γ–treated Ulk1/2−/− MEFs (Fig. 5B). Moreover, IFN-γ treatment altered the expression of 338 genes in Ulk1/2+/+ MEFs (Fig. 5C, green ellipse, and fig. S2A), and 376 genes in Ulk1/2−/− MEFs (Fig. 5C, blue ellipse, and fig. S2B). From these, 268 common genes were differentially expressed in both groups after IFN-γ treatment (Fig. 5C, fig. S3, and table S3). We also identified 70 IFN-γ–stimulated genes that were expressed only in Ulk1/2+/+ MEFs (Fig. 5C, fig. S4, and table S4) and 108 IFN-γ–stimulated genes that were expressed only in Ulk1/2−/− cells (Fig. 5C, fig. S5, and table S5). Gene ontology (GO) analyses were performed to classify the differentially expressed genes detected after IFN-γ treatment in both genotypic groups, those only in Ulk1/2+/+ cells and those only in Ulk1/2−/− cells among ontology clusters (fig. S6, A to C, and tables S6 to S8). The 268 commonly differentially expressed genes are mainly associated with processes involved in the defense response to virus infection, namely, the innate immune response and antigen processing and presentation (fig. S6A and table S6). The 70 genes differentially expressed only in Ulk1/2+/+ MEFs are predominantly associated with processes involved in immune cell differentiation, regulation of the JAK-STAT cascade, and regulation of inflammatory responses (fig. S6B and table S7). By contrast, the 108 genes differentially expressed only in Ulk1/2−/− MEFs were predominantly associated with regulation of leukocyte proliferation and migration and positive regulation of the MAPK cascade (fig. S6C and table S8). To closely evaluate whether the absence of Ulk1/2 affects IFN-γ–mediated expression of antiviral genes, we selected the IFN-γ–inducible genes depicted in our GO analysis (tables S6 to S8) associated with antiviral processes (a total of 66 genes) (22, 23). From these genes, we found that expression of 46 genes was increased by IFN-γ more in Ulk1/2+/+ MEFs than in Ulk1/2−/− MEFs (Fig. 5D) and, overall, this difference was statistically significant (Fig. 5E).

Fig. 5 Ulk1/2 augments IFN-γ–mediated transcription of ISGs.

(A to E) RNA-seq analysis of transcript expression in Ulk1/2+/+ and Ulk1/2−/− MEFs untreated or treated with mouse IFN-γ for 6 hours, as indicated. MDS plot of all groups (A) and Volcano plot of differentially expressed (DE) genes after IFN-γ treatment in Ulk1/2+/+ versus Ulk1/2−/− MEFs (B). Venn diagram (C) indicates the gene expression overlap existing between differentially expressed genes after IFN-γ treatment in Ulk1/2+/+ (green ellipse) and Ulk1/2−/− (blue ellipse) MEFs. Heatmap analysis (D) of antiviral genes induced in Ulk1/2+/+ MEFs compared to Ulk1/2−/− MEFs by IFN-γ treatment. Box plot (E) of log2 fold change (FC) distribution of IFN-γ–inducible antiviral genes in Ulk1/2+/+ and Ulk1/2−/− MEFs. Data are from four biological replicates per group (see also figs. S2 to S6 and tables S3 to S8). Statistical analyses were performed using Wilcoxon unpaired rank sum test.

To evaluate whether ULK1/2 activity was required for optimal IFN-γ–dependent transcriptional activation of ISGs, we generated Ulk1/2+/+ and Ulk1/2−/− MEFs stably expressing an IFN-γ activation site (GAS) element-luciferase reporter gene (GAS-LUC) and performed luciferase reporter assays. IFN-γ–induced transcriptional activation of GAS elements was significantly reduced in Ulk1/2−/− cells compared with Ulk1/2+/+ cells (Fig. 6A). In addition, when we examined the specific role of ULK1/2 on the expression of select IFN-γ–inducible antiviral genes Cxcl10 (24, 25), Oasl2, and Ifit3 (26, 27) by quantitative reverse transcription polymerase chain reaction (qRT-PCR), we found that expression of these genes after IFN-γ treatment was significantly decreased in the absence of Ulk1/2 (Fig. 6, B to D). Although the activation of ULK1 promotes induction of autophagy (28), treatment of Ulk1/2+/+ MEFs with the autophagy inhibitor chloroquine (CQ) did not significantly affect IFN-γ–mediated mRNA expression of Cxcl10, Oasl2, and Ifit3 (fig. S7, A to C). Moreover, small interfering RNA (siRNA)–mediated knockdown of vacuolar protein sorting 34 (VPS34; PIK3C3) and Beclin-1 (BECN1), which are necessary for ULK1-mediated autophagy (29), did not reduce IFN-γ–induced expression of antiviral ISGs in KT-1 cells (fig. S7, D to I). These results suggest that the key role of ULK1/2 related to IFN-γ–mediated transcription of antiviral genes is independent of the activation of autophagy.

Fig. 6 ULK1/2 is required for IFN-γ–dependent transcriptional induction of antiviral ISGs.

(A) Luciferase reporter assay analysis of GAS promoter activity in Ulk1/2+/+ and Ulk1/2−/− cells untreated or treated with IFN-γ for 6 hours. Data are means ± SEM from four independent experiments. (B to D) qRT-PCR analysis of Cxcl10 (B), Oasl2 (C), and Ifit3 (D) mRNA expression in Ulk1/2+/+ and Ulk1/2−/−MEFs untreated or treated with IFN-γ for 6 hours. Data are means ± SEM from three (B) or four (C and D) independent experiments. (E) Western blot analysis of ULK1 in lysate from KT-1 ULK1 KO cells generated by CRISPR-Cas9 genome editing. Blots are representative of three independent experiments. (F to I) qRT-PCR analysis of CXCL10 (F), OAS1 (G), IFIT3 (H), and IRF1 (I) mRNA expression in KT-1 ULK1 WT and KT-1 ULK1 KO cells that were incubated in the presence or absence of IFN-γ for 6 hours. Data are means ± SEM from three (F) or four (G to I) independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 by unpaired t test (two-tailed) with Welch’s correction.

To determine whether ULK1 activity was also required for IFN-γ–induced transcription of select antiviral genes in human cells, we generated KT-1 ULK1 knockout (KO) cells using CRISPR (clustered regularly interspaced short palindromic repeats)–Cas9 genomic editing. Plasmids encoding Cas9 and single-guide RNA (sgRNA) targeting the ULK1 gene (fig. S8) were transduced into KT-1 cells, and stably transduced cells were selected with blasticidin and puromycin antibiotics. The complete absence of ULK1 protein was confirmed using Western blot analysis and by comparing ULK1 abundance between the KT-1 parental cell line [ULK1 wild-type (WT)] and KT-1 ULK1 KO cells (Fig. 6E). We verified that IFN-γ–induced gene expression of the antiviral effectors CXCL10, OAS1, IFIT3, and IRF1 was significantly reduced in ULK1 KO cells compared with ULK1 WT cells by qRT-PCR analysis (Fig. 6, F to I). Together, our data suggest that ULK1 promotes antiviral ISG transcription.

ULK1/2 kinase activity is required for IFN-γ–induced antiviral responses

Because ULK1/2 expression was required for the optimal transcription of select IFN-γ–induced antiviral genes, we tested whether ULK1 was required for IFN-γ–stimulated antiviral responses. For this, we compared the protective effects of mouse IFN-γ on protection from encephalomyocarditis virus (EMCV)–induced cell death in Ulk1/2+/+ and Ulk1/2−/− MEFs. We found that cells that lacked Ulk1/2 were less responsive to IFN-γ treatment than WT cells (Fig. 7A). Moreover, human fibrosarcoma 2fTGH cells were less sensitive to the antiviral effects of human IFN-γ treatment against EMCV infection when treated with the ULK1 kinase inhibitor MRT68921 (Fig. 7B) (30). Similarly, selective inhibition of ERK5 activity with XMD8-92 (31) resulted in a reduction of IFN-γ–induced expression of the antiviral genes Cxcl10 and Ifit3 in Ulk1/2+/+ MEFs (Fig. 7C) and of IFN-γ–induced antiviral effects in human 2fTGH cells (Fig. 7D). Together, these data suggest that engagement of the ULK1/2-ERK5 pathway augments IFN-γ–dependent antiviral responses (Fig. 7E).

Fig. 7 Requirement of ULK1/2 for the generation of IFN-γ–dependent antiviral effects.

(A) Crystal violet viability analysis of EMCV-induced cytopathic effects (CPEs) in Ulk1/2+/+ and Ulk1/2−/− MEFs pretreated with IFN-γ at the indicated doses for 16 hours and subsequently challenged with EMCV for 24 hours. Data are means ± SEM of quadruplicate assays from three independent experiments. (B) Crystal violet viability analysis of EMCV-induced CPEs in human fibrosarcoma 2fTGH cells pretreated for 2 hours with the ULK1 kinase inhibitor MRT68921 (MRT), as indicated, and then exposed to IFN-γ for 16 hours before challenge with EMCV for 24 hours. Data are means ± SEM of quadruplicate assays from three independent experiments. (C) qRT-PCR analysis of Cxcl10 (left) and Ifit3 (right) mRNA expression in Ulk1/2+/+ MEFs were treated with dimethyl sulfoxide (DMSO) (vehicle-control, C), XMD8-92 (XMD), and/or IFN-γ, as indicated. Data are means ± SEM from four independent experiments. (D) Crystal violet viability analysis of EMCV-induced CPEs in human fibrosarcoma 2fTGH cells pretreated for 2 hours with the ERK5 inhibitor XMD8-92, as indicated, and then exposed to IFN-γ for 16 hours before challenge with EMCV for 24 hours. Data are means ± SEM of quadruplicate assays from three independent experiments. (E) Schematic illustration of the potential role of ULK1/2 in IFN-γ signaling. *P < 0.05, **P < 0.01, and ****P < 0.0001 by one-way ANOVA analysis, followed by Tukey’s multiple comparisons test (C) or two-way ANOVA with post hoc t tests (A, B, and D).

DISCUSSION

IFN-γ is a critical immunomodulator, regulating the innate and adaptive immune responses to pathogen infection (32). Both IFN-γ KO and IFNGR KO mice are more susceptible to bacterial and viral infections than their WT counterparts (3336). Moreover, mice injected with a monoclonal antibody against IFN-γ succumb to Toxoplasma gondii infection (37) and Listeria monocytogenes at higher rates (38). Distinct from the ability of IFN-γ to influence the differentiation and activation of specific immune cell subsets required for pathogen clearance, IFN-γ promotes apoptosis of pathogen-infected cells and induces the expression of type I IFNs and of several antiviral and antimicrobial genes (1, 27, 39, 40). However, a comprehensive understanding of the signaling pathways activated downstream of the IFNGR and how these affect IFN-γ–driven responses remains unclear. Herein, we provide evidence that IFN-γ–induced engagement of ULK1 was required to inhibit virus infection. We found that lack of Ulk1/2 genes or pharmacological inhibition of the ULK1/2 kinase activity abrogated IFN-γ–induced antiviral activity. This correlated with defective IFN-γ–mediated transcription of antiviral genes in the absence of ULK1/2 activity, although we cannot exclude that additional pathways may participate in the induction of antiviral responses. Our work suggests that IFN-γ signaling involves the interaction of ULK1 with MLK3, which was required for IFN-γ–dependent phosphorylation/activation of MLK3. These data suggest that ULK1 likely acts upstream, either as a kinase or a scaffold protein, of IFN-γ–activated MAPK signaling cascades. In human fibroblasts, MLK3 is required for both cytokine- and mitogen-induced activation of several MAPK signaling cascades (ERK1/2, JNK, and p38 MAPK) (41). Moreover, tumor necrosis factor (TNF)–induced stimulation of JNK, but not p38 MAPK or ERK1/2, is defective in Mlk3−/− MEFs when compared with WT cells (12). Similarly, MLK3 is required for IFN-γ–driven gene expression, independently of ERK1/2 activation (15). Here, we showed in MEFs that both Ulk1/2 and Mlk3 were required for IFN-γ–mediated phosphorylation of ERK5, but not JNK and ERK1/2. These data suggest that ULK1 and MLK3 may act upstream of MEKK2, MEKK3, or MEK5 kinases, which are known to activate ERK5 in IFN-γ–stimulated cells (16, 42). Our results support a model in which IFN-γ–dependent engagement of ULK1 leads to phosphorylation of MLK3 and ERK5, which promotes transcription of antiviral ISGs and consequent IFN-γ–induced antiviral responses that may complement the antiviral effects induced by the canonical JAK1/JAK2-STAT signaling.

Autophagy is a degradative process that aids the host immune response against viral infections, including induction of type I IFN production and activation of IFN-γ–induced JAK-STAT signaling (43, 44). Assembly of the nondegradative Atg5-Atg12/Atg16L1 autophagy protein complex is also required for IFN-γ–dependent antiviral activity (45). In contrast, ULK1-dependent autophagy activates a different pathway of phagophore nucleation that requires phosphorylation of Beclin-1 and activation of VPS34 (29, 30, 46). Using pharmacologic and genetic inhibition, we found that autophagy is not required for the effects of ULK1 on IFN-γ–dependent gene expression. Thus, multicellular organisms may have developed ways to use autophagy-related proteins for additional immune functions. In support of this, here we report a nondegradative role of ULK1 that is required for transcription of IFN-γ–inducible antiviral genes.

The precise mechanisms by which ULK1 or MLK3 may stimulate ERK5 phosphorylation remain undetermined. MEK5 is the direct upstream activator of ERK5 (47), and MEKK2/3 (4850), Cot/TPL-2 (51), and MLTK (52) are MAP3Ks that activate ERK5 in response to multiple stimuli. In addition, another MAP3K, transforming growth factor β–activated kinase 1 (TAK1), interacts with MEKK3 and promotes its autophosphorylation in TNFα signaling (53). Similarly, ULK1 and/or MLK3 could act as scaffold proteins, or MAP3Ks of ERK5 upstream kinases after IFN-γ treatment. Although p38 MAPK activity is not required for IFN-γ–induced transcription of ISGs in MEFs (54), it is involved in the generation of IFN-γ–inducible biological effects in different human cell types (5558). Given that ULK1 promotes p38 MAPK activity downstream of the type I IFNR (9), in future studies, it would be of interest to evaluate whether ULK1-MLK3 interactions are required for IFN-γ–dependent activation of p38 MAPK in human cells. Viewed altogether, our studies identify an IFN-γ−stimulated signaling cascade that is critical for expression of IFN-γ–inducible ISGs and antiviral responses. These findings may suggest new points of therapeutic intervention for the treatment of several diseases where IFN-γ has proven clinical activity (5961) or how to enhance immune checkpoint blockade where IFN-γ responses are involved (62, 63).

MATERIALS AND METHODS

Cell culture

KT-1 (64) and U937 (CRL-1593.2; American Type Culture Collection) cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) and antibiotics. 2fTGH cells (65) were provided by G. R. Stark (Cleveland Clinic) and were cultured in DMEM (Dulbecco’s modified Eagle’s medium) high-glucose medium supplemented with 10% FBS and antibiotics. The immortalized Ulk1/2+/+ and Ulk1/2−/− MEFs (17) were provided by C. B. Thompson (Memorial Sloan-Kettering Cancer Center) and were cultured in DMEM supplemented with 10% FBS and antibiotics. Primary Mlk3+/+ and Mlk3−/− MEFs were described previously (12) and were cultured in DMEM high-glucose medium supplemented with 10% bovine growth serum, 1× l-glutamine, 1× penicillin/streptomycin solution, and 8 μl of 2-mercaptoethanol per liter of medium. All experiments were performed using primary MEFs between passages 2 and 4. The other cell lines were frozen at low passage in liquid nitrogen and were kept in culture for no longer than eight passages. All cells were cultured at 37°C and 5% CO2. All cells were tested for mycoplasma contamination using a MycoAlert PLUS mycoplasma detection kit, following the manufacturer’s instructions (Lonza).

Reagents

IFN-γ recombinant human (no. PHC4033) and mouse (no. PMC4033) proteins were from Gibco (Life Technologies). MRT68921 dual autophagy kinase ULK1/2 inhibitor was purchased from Selleckchem (no. S7949). CQ was purchased from Sigma-Aldrich. XMD8-92 was purchased from Tocris (no. 4132). The Edit-R human lentiviral ULK1 sgRNAs (no. VSGH10142-246477203) and Edit-R Lentiviral hEF1α-Blast-Cas9 nuclease plasmid DNA (no. CAS10138) were purchased from GE Healthcare Dharmacon. ON-TARGETplus Non-targeting Control Pool siRNA (no. D-001810-10-05), ON-TARGETplus PIK3C3 siRNA (SMARTpool: no. L-005250-00-0005), and ON-TARGETplus BECN1 siRNA (SMARTpool: no. L-010552-00-0005) were purchased from GE Healthcare Dharmacon. Antibodies against ULK1 (D8H5; no. 8054), MLK3 (no. 2817), phospho-ERK5 (Thr218/Tyr220; no. 3371), ERK5 (D3I5V; no. 12950), phospho-p44/42 MAPK (ERK1/2) (Thr202/Tyr204) (D13.14.4E; no. 4370), p44/42 MAPK (ERK1/2; no. 9102), phospho-SAPK/JNK (Thr183/Tyr185) (81E11; no. 4668), SAPK/JNK (no. 9252), phospho-STAT1 (Ser727) (D3B7; no. 8826), phospho-STAT1 (Tyr701) (D4A7; no. 7649), phospho-p90RSK (Thr359/Ser363) (no. 9344), and RSK1 (D6D5) (no. 8408) were purchased from Cell Signaling Technology. Anti-MLK3 (phospho Thr277 + Ser281) antibody (no. ab191530) was purchased from Abcam, and STAT1 p84/p91 antibody (E-23; no. sc-346) was from Santa Cruz Biotechnology. Antibody against GAPDH (6C5) (no. MAB374) was purchased from EMD Millipore.

Mass spectroscopy analyses

To identify proteins that interact with endogenous ULK1, KT-1 cells were left untreated or were treated with human IFN-γ (5 × 103 IU/ml) for 10 min and then lysed in NP-40 buffer [20 mM Hepes (pH 7.4), 180 mM KCl, 0.2 mM EGTA, and 0.1% NP-40] supplemented with protease and phosphatase inhibitors. Three milligrams of protein (total cell lysates) from untreated and IFN-γ–treated samples was used for IP of endogenous protein-ULK1 complexes using ULK1 (D8H5) rabbit monoclonal antibody–conjugated to magnetic beads (custom order, Cell Signaling Technology). As control, the same procedure was followed for IFN-γ–treated lysates, but using rabbit (DA1E) monoclonal antibody immunoglobulin G (IgG) XP Isotype control–conjugated to magnetic beads (no. 8726, Cell Signaling Technology) instead of the ULK1 antibody. After incubating the samples overnight with rotation at 4°C, the beads were washed two times with NP-40 buffer and one time with washing buffer [20 mM Hepes (pH 7.4), 180 mM KCl, and 0.2 mM EGTA]. Protein-ULK1 complexes were eluted from the beads by incubation with Lane Marker Reducing Sample Buffer (Pierce) at 95°C for 10 min, and proteomic analyses were performed in the Northwestern Proteomics Core Facility (Northwestern University, Chicago). IP-eluted proteins were initially separated using SDS–polyacrylamide gel electrophoresis (PAGE) and cut into 10 equivalent height bands before standard in-gel digestion (66). Resulting peptides were extracted from the gel pieces and desalted using solid-phase extraction on a Pierce C18 Spin column, before elution in 40 μl of 80% acetonitrile in 0.2% formic acid. After lyophilization, peptides were reconstituted with 0.1% formic acid in water and injected onto a trap column {150 μm [inner diameter (ID)] by 3 cm} coupled with a nanobore analytical column [75 μm (ID) by 15 cm; both ReproSil-Pur C18-aQ, 3 μm]. Samples were separated using a linear gradient of solvent A (95% water, 5% acetonitrile, and 0.1% formic acid) and solvent B (5% water, 95% acetonitrile, and 0.1% formic acid) over 60 min. nLC-MS/MS data were obtained on a Velos Orbitrap (Thermo Fisher Scientific) mass spectrometer. Data were searched using Mascot 2.5 (Matrix Science) against the human SwissProt database, and results were reported at 1% false discovery rate in Scaffold 4.5 (Proteome Software). Proteins identified by nLC-MS/MS analysis in the control group (rabbit IgG) were excluded from our data analysis.

Protein function enrichment analysis

Protein lists identified in untreated and/or IFN-γ–treated groups were converted to gene lists that were submitted to the Metascape database (67, 68), a gene annotation and analysis resource (http://metascape.org/), for pathway and process enrichment analysis. For each given gene list, Metascape carries pathway and process enrichment analysis using the following ontology sources: GO biological processes, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, and Reactome gene sets. All genes in the genome were used as the enrichment background. Terms with P < 0.01, a minimum count of 3, and an enrichment factor of >1.5 are collected and grouped into clusters based on their membership similarities. More specifically, P values are calculated on the basis of accumulative hypergeometric distribution, and q values are calculated using the Benjamini-Hochberg procedure to account for multiple testing. κ scores were used as the similarity metric when performing hierarchical clustering on the enriched terms, and then, subtrees with similarity > 0.3 are considered a cluster. The most statistically significant term within a cluster is chosen as the one representing the cluster.

Coimmunoprecipitation of protein-ULK1 complexes

To confirm the physical interaction between ULK1 and MLK3 identified by nLC-MS/MS analysis, KT-1 cells were left untreated or were treated with human IFN-γ (5 × 103 IU/ml) for 10 min, and U937 cells were cultured overnight in serum-free RPMI 1640 medium and then were left untreated or were treated with human IFN-γ (5 × 103 IU/ml) for 10 min. After treatment, cell pellets were lysed in NP-40 buffer [20 mM Hepes (pH 7.4), 180 mM KCl, 0.2 mM EGTA, 10% glycerol, and 0.1% NP-40] supplemented with protease and phosphatase inhibitors. For IP of endogenous protein-ULK1 complexes, 200 μg of protein (total cell lysates) from each sample was incubated overnight at 4°C with rotation with ULK1 (D8H5) rabbit monoclonal antibody (1:100; no. 8054, Cell Signaling Technology), followed by incubation for 1 hour at 4°C with rotation with protein G Sepharose 4 Fast Flow beads (GE Healthcare Life Sciences). As control, the same procedure was followed using rabbit (DA1E) monoclonal antibody IgG XP Isotype control (no. 3900, Cell Signaling Technology) instead of ULK1 antibody. After IP, the beads were washed three times with NP-40 buffer without glycerol. Protein-ULK1 complexes were eluted from the beads by incubation with Lane Marker Reducing Sample Buffer (Pierce) at 95°C for 10 min. Eluates were resolved by SDS-PAGE and processed for immunoblotting analyses.

In vitro kinase assay

Recombinant human MLK3 active protein (no. M19-11G, SignalChem) was heat-inactivated at 65°C for 20 min in a sonicating bath. To confirm heat inactivation of MLK3 recombinant protein, kinase reactions between MLK3 heat-inactive protein (1 μg) and myelin basic protein substrate (5 μg; no. M42-51N, SignalChem) were performed. To determine whether ULK1 can directly phosphorylate MLK3, kinase reactions between recombinant human ULK1 active protein (100 ng; no. U01-11G, SignalChem) and MLK3 heat-inactive protein (1.66 μg) were performed. As controls, the same kinase reactions were carried out, but using each recombinant protein alone. For each kinase reaction, we used dithiothreitol (no. D86-09B, SignalChem), ATP (no. V915A, Promega), and 5× kinase buffer (no. K03-09, SignalChem), following the manufacturer’s instructions. Kinase mixtures were incubated for 40 min shaking at 300 rpm and 30°C. The ADP formed from the kinase reactions was measured using the ADP-Glo Kinase Assay Kit (no. V6930, Promega), following the manufacturer’s instructions.

MEFs treatment and lysis for immunoblotting analyses

Before IFN-γ treatment, MEFs were either cultured in 10% FBS-containing medium (for detection of phosphorylation of ERK5, STAT1, and ERK1/2) or starved overnight (for detection of phosphorylation of JNK, MLK3, and p90RSK1). MEFs were then treated with mouse IFN-γ (5 × 103 IU/ml) for 10 or 30 min, as indicated. After treatment, cell pellets were lysed with lysis buffer [50 mM Hepes (pH 7.3), 150 mM NaCl, 1.5 mM MgCl2, 1 mM EDTA (pH 8.0), 100 μM sodium fluoride, 10 μM sodium pyrophosphate, 0.5% Triton X-100, and 10% glycerol) supplemented with protease and phosphatase inhibitors.

Immunoblotting analyses

Equal amounts of total cell lysates were resolved by SDS-PAGE and transferred to a Immobilon-P PVDF membrane (Millipore) using the Trans-Blot Turbo transfer system (Bio-Rad). For immunoblotting analyses, the membranes were probed with primary antibodies, followed by horseradish peroxidase–conjugated secondary antibodies, and antibody binding was detected by enhanced chemiluminescence using an Amersham ECL prime Western blotting detection reagent (GE Healthcare Life Sciences). Bands corresponding to proteins of interest were scanned and quantified by densitometry using ImageJ software.

Library construction and RNA-seq

Ulk1/2+/+ and Ulk1/2−/− MEFs were plated as four biological replicates and were either left untreated or were treated for 6 hours with mouse IFN-γ (2.5 × 103 IU/ml). Total RNA was isolated using the RNeasy Mini Kit (Qiagen), following the manufacturer’s instructions. Library construction and stranded mRNA sequencing were conducted at the NUSeq Core Facility of Northwestern University. Briefly, RNA quality and quantity were first determined with the Agilent Bioanalyzer 2100 and Qubit fluorometer, and all samples presented an RNA integrity number of 10. Sequencing libraries were prepared from 1 μg of high-quality RNA samples using the Illumina TruSeq Stranded mRNA Library Preparation Kit (Illumina), as per the manufacturer’s instructions. This procedure includes mRNA purification and fragmentation, complementary DNA (cDNA) synthesis, 3′ end adenylation, Illumina adapter ligation, and library PCR amplification and validation. An Illumina NextSeq 500 Sequencer was used to sequence the libraries with the production of single-end, 75-bp reads.

RNA-seq reads mapping and transcript abundance estimation

Raw reads in the FASTQ format were controlled for quality using Trimmomatic (69). Trimmed reads were then aligned to the mouse genome (University of California Santa Cruz Mouse Genome Browser mm9) using TopHat v.2.1.0 (70). Only uniquely mapped reads with a maximum of two mismatches over the whole length of the gene were considered for ensuing analyses. Gene annotations came from Ensembl release 75. Exonic reads were assigned to specific genes from Ensembl release 75 using the htseq-count script from HTSeq-0.6.1 (71).

Differential analysis of gene regulation at transcript resolution

Gene counts computed by HTSeq were used as input for edgeR (version 3.18.1) (22, 72), which was used to calculate the differential expression and analyze the significances of observed changes between two groups using the standard edgeR workflow. Genes with Benjamini-Hochberg–adjusted P values of less than 0.01 were considered to be differentially expressed unless otherwise specified.

Gene clustering

RNA-seq results are shown as a heatmap and include transcripts whose expression was the most significantly altered by IFN-γ treatment in Ulk1/2+/+ and Ulk1/2−/− MEFs. Colored rectangles represent normalized total reads per million (RPM) mRNA abundance of the transcript. Hierarchical clustering was performed with R version 3.4.0 software using the heatmap.2 function and distance correlation. The intensity of the color is proportional to the RPM values from each RNA-seq measurement, as indicated on the bar next to the heatmap image.

Gene set enrichment analyses

The gene list of the differentially induced genes by IFN-γ found only in Ulk1/2+/+ MEFs was submitted to the Metascape database (http://metascape.org) (68, 69). Metascape carries pathway and process enrichment analysis using the following ontology sources: GO biological processes, KEGG pathway, and Reactome gene sets. GO functional analysis was conducted using default parameters.

Generation of Ulk1/2+/+ and Ulk1/2−/− cell lines stably expressing GAS luciferase reporter gene elements

pGF1-GAS-LUC, which expresses a puromycin selection cassette and firefly luciferase reporter under the control of a minimal cytomegalovirus promoter followed by four tandem consensus GAS elements (5′-AGTTTTCATATTACTCTAAATC-3′) was purchased from System Biosciences (SBI). Reporter vector carrying viral particles was produced by cotransfection of the lentiviral plasmid and the packaging vectors into the Lenti-X 293T cell line (Clontech). Ulk1/2+/+ and Ulk1/2−/− MEFs were infected with virus-containing fresh supernatant using Transdux reagent (SBI). Stably transduced cells were selected using puromycin (2 μg/ml; Gibco, Life Technologies).

Luciferase assays

Ulk1/2+/+ and Ulk1/2−/− MEFs stably expressing pGF1-GAS-LUC were plated in a 96-well plate (three or five replicates of 5000 cells per well) and, 24 hours later, were maintained in serum-free RPMI 1640 medium overnight. Serum-starved cells were left untreated (control) or were treated with mouse IFN-γ (2.5 × 103 IU/ml) for 6 hours and then lysed using 1× Reporter Lysis Buffer (Promega). Luciferase assay substrate buffer (Luciferase Assay System no. E4030, Promega) was used as per the manufacturer’s instructions, and luciferase activities were measured using a Cytation 3 cell imaging multimode microplate reader (BioTek).

qRT-PCR analyses

To determine the effect of ULK1/2 on IFN-γ–induced expression of antiviral ISGs, Ulk1/2+/+ and Ulk1/2−/− MEFs or KT-1 WT and KT-1 ULK1 KO cells were left untreated or were treated for 6 hours with mouse or human IFN-γ (2.5 × 103 IU/ml), respectively. To determine whether ERK5 is required for IFN-γ–induced expression of antiviral ISGs, Ulk1/2+/+ MEFs were preincubated for 1 hour with 0.5 μM ERK5 inhibitor XMD8-92, followed by 6 hours of cotreatment with mouse IFN-γ (2.5 × 103 IU/ml). Cells were treated with vehicle-control (DMSO) or each compound alone (controls). To define whether inhibition of autophagy would affect antiviral ISG expression, KT-1 cells were transfected with control, PIK3C3 (VPS34), or BECN1 (Beclin-1) siRNAs using the Amaxa Cell Line Nucleofector Kit V (Lonza), following the manufacturer’s instructions and, 18 hours later, cells were either left untreated or were treated with human IFN-γ (2.5 × 103 IU/ml) for 6 hours. In addition, Ulk1/2+/+ MEFs were preincubated for 1 hour with 6 μM CQ, followed by 6 hours of cotreatment with mouse IFN-γ (2.5 × 103 IU/ml). As controls, cells were left untreated or were treated with each compound alone. Total RNA was isolated using the RNeasy Mini Kit (Qiagen) following the manufacturer’s instructions. Two micrograms of the total cellular mRNA was reverse-transcribed into cDNA using a Omniscript RT kit (Qiagen) and oligo(dT)12–18 primers (Life Technologies). qRT-PCR was carried out using an ABI7500 sequence detection system (Applied Biosystems) or a Bio-Rad CFX96 Real Time System (Bio-Rad) using commercially available fluorescein amidite (FAM)–labeled primer/probe sets (Thermo Fisher Scientific) to determine mouse Cxcl10 (Mm00445235_m1), Ifit3 (Mm01704846_s1), Oasl2 (Mm00496187_m1), and human CXCL10 (Hs01124252_g1), OAS1 (Hs00973635_m1), IFIT3 (Hs01922752_s1), IRF1 (Hs00971965_m1), PIK3C3 (Hs00176908_m1), and BECN1 (Hs01007018_m1) mRNA expression. Mouse Gapdh (Mm99999915_g1) and human GAPDH (Hs02758991_g1) were used for normalization. The mRNA amplification was calculated as previously (9), and the data were plotted as the increase of fold change as compared with control samples.

CRISPR-Cas9 approach to generate ULK1 KO KT-1 cells

KT-1 cells were transduced with Edit-R Lentiviral hEF1α-Blast-Cas9 Nuclease Plasmid DNA and Edit-R Human ULK1 Lentiviral sgRNA (no. CAS10138 and no. VSGH10142-246477203, GE Healthcare Dharmacon). Transduction was carried out by spinoculation (750g) for 90 min at room temperature in the presence of Transdux reagent (SBI), followed by 48 hours of incubation. Stably transduced cells were selected using blasticidin (5 μg/ml) and puromycin (2.5 μg/ml) antibiotics. One week later, antibiotic-resistant cells were separated from dead cells using a BD FACSAria 5-Laser Cell Sorter (BD Biosciences) and performed at the Northwestern University Flow Cytometry Core Facility. Lack of ULK1 protein in KT-1 ULK1 KO cells was confirmed using Western blot analysis.

Antiviral assays

Ulk1/2+/+ and Ulk1/2−/− MEFs were seeded in quadruplicate in 96-well plates overnight and then treated with the indicated doses of mouse IFN-γ for 16 hours. Cells were subsequently challenged with EMCV, and EMCV-induced CPEs were determined 24 hours later. Briefly, medium was aspirated, and cells were fixed in 95% ethanol for 30 min and then stained with 0.1% crystal violet in 2% ethanol for a further 30 min. Cells were then destained in 0.5 M NaCl in 50% ethanol for 1 hour, and then, absorbance readings at 570 nm were taken to determine the extent of uninfected, viable cells. For those assays where the ULK1 pharmacological inhibitor MRT68921 or the ERK5 inhibitor XMD8-92 was included, human fibrosarcoma 2fTGH cells were seeded in quadruplicate in 96-well plates and then either treated with human IFN-γ for 16 hours or pretreated for 2 hours with MRT68921 (1 μM) or with XMD8-92 (5 μM) and then, the medium was removed, and the cells were treated with MRT68921 or XMD8-92 and human IFN-γ, as indicated, for a further 16 hours. Cells were subsequently challenged with EMCV, and EMCV-induced CPEs were determined 24 hours later, as described above.

Statistical analyses

GraphPad Prism v.6.0 software (GraphPad Software) and SAS 9.4. software (SAS Institute Inc.) were used for statistical analyses. Unpaired two-tailed t test with Welch’s correction was used for comparison of means between two groups, one-way ANOVA was used to compare means among more than two independent groups followed by Tukey’s multiple comparisons test, and two-way ANOVA with Bonferroni-corrected post hoc t tests were used to compare the mean differences among groups classified by two factors. Differences were considered statistically significant when P values were less than 0.05.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/557/eaap9921/DC1

Fig. S1. ULK1/2 activity is not required for IFN-γ–induced phosphorylation of STAT1.

Fig. S2. Effects of targeted disruption of Ulk1/2 gene expression on IFN-γ–dependent gene transcription.

Fig. S3. Genes differentially expressed in both Ulk1/2+/+ and Ulk1/2−/− MEFs after IFN-γ treatment.

Fig. S4. Genes differentially expressed only in Ulk1/2+/+ MEFs after IFN-γ treatment.

Fig. S5. Genes differentially expressed only in Ulk1/2−/− MEFs after IFN-γ treatment.

Fig. S6. GO analysis of IFN-γ–dependent differentially expressed genes.

Fig. S7. Inhibition of autophagy does not affect transcription of antiviral ISGs.

Fig. S8. Genomic location for the ULK1 sgRNA in the human ULK1 open reading frame.

Table S1. GO analysis of putative ULK1 interactors in untreated versus IFN-γ–treated KT-1 cells identified by mass spectrometry analysis.

Table S2. GO analysis of putative ULK1 interactors identified after IFN-γ stimulation by mass spectrometry analysis.

Table S3. Genes differentially expressed in both Ulk1/2+/+ and Ulk1/2−/− MEFs after IFN-γ treatment.

Table S4. Genes differentially expressed only in Ulk1/2+/+ MEFs after IFN-γ treatment.

Table S5. Genes differentially expressed only in Ulk1/2−/− MEFs after IFN-γ treatment.

Table S6. GO analysis of differentially expressed genes identified in both Ulk1/2+/+ and Ulk1/2−/− MEFs after IFN-γ treatment.

Table S7. GO analysis of differentially expressed genes identified only in Ulk1/2+/+ MEFs after IFN-γ treatment.

Table S8. GO analysis of differentially expressed genes identified only in Ulk1/2−/− MEFs after IFN-γ treatment.

REFERENCES AND NOTES

Acknowledgments: We thank C. B. Thompson (Memorial Sloan-Kettering Cancer Center) for the Ulk1/2+/+ and Ulk1/2−/− MEFs. We thank A. Shilatifard (Northwestern University, Feinberg School of Medicine) for helping with the analysis of the RNA-seq data. Funding: This work was supported, in part, by NIH grants CA161196, CA77816, and CA155566 and by grant I01CX000916 from the Department of Veterans Affairs. A.D.A. was supported, in part, by NIH/NCI grant T32 CA070085, and D.S. was supported, in part, by NIH/NCI grant T32 CA080621. E.N.F. is a Tier 1 Canada Research Chair in Women’s Health and Immunobiology. Proteomic services were performed by the Northwestern Proteomics Core Facility, supported by NCI CCSG P30 CA060553 awarded to the Robert H. Lurie Comprehensive Cancer Center and the National Resource for Translational and Developmental Proteomics supported by P41 GM108569. Flow cytometry cell sorting was performed on a BD FACSAria SORP system, purchased through the support of NIH 1S10OD011996-01, and this work was supported by the Northwestern University–Flow Cytometry Core Facility supported by the Cancer Center Support Grant (NCI CA060553). Sequencing of RNA libraries were performed by the NUSeq Core of Northwestern University. Author contributions: D.S., G.T.B., and B.M.-K. performed experiments; E.M.K., A.D.A., M.F., and N.K.R. assisted in conduction of experiments; D.S., E.N.F., and L.C.P. designed experiments; D.S., E.M.K., C.M.H., E.N.F., and L.C.P. analyzed data; P.A.O. performed RNA-seq bioinformatic analyses; R.J.D. provided key materials; D.S., L.C.P., and E.N.F. wrote the manuscript; all authors reviewed and approved the final version of the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Raw RNA-seq data have been uploaded to the NIH GEO web portal, and the accession number is GSE118766. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD009018. All other data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Stay Connected to Science Signaling

Navigate This Article