PRMT6 Promotes Lung Tumor Progression via the Alternate Activation of Tumor-Associated Macrophages Free

This article is featured in Highlights of This Issue, p. 1

Lung cancer is the leading cause of cancer-related deaths in the world, and non–small cell lung cancer (NSCLC) accounts for approximately 85% of all lung cancers with a 5-year survival rate of only approximately 16% (1, 2). Lung cancer, a genetically heterogeneous disease, is most often diagnosed at an advanced inoperable stage where systemic therapies show only a modest benefit. However, the molecular mechanisms in the initiation and the progression of lung cancer remain poorly understood. Hence, identification of novel gene/s, which drive tumor initiation and progression, is required for future development of more effective therapies to treat advanced lung cancers. Posttranslational modifications (PTM) on proteins expand the functional diversity of the proteome, favoring robust dynamic modulation of protein behavior for precise temporal and spatial control of gene expression and/or signaling. One such PTM is protein arginine methylation, which is deregulated in cancers (3).

Protein arginine methylation is catalyzed by a class of enzymes known as protein arginine methyl transferases (PRMT). Depending on the position of the guanidino nitrogens being methylated, PRMTs are classified as monomethylated or dimethylated, and the latter can be either symmetric or asymmetric (4, 5). Increased expression of PRMT isoforms, for example, PRMT1, CARM1, PRMT5, PRMT6, and PRMT9, in several tumor types has been correlated with poor overall survival (6–11). In recent years, several small-molecule compounds and peptide inhibitors that target the catalytic/substrate binding domains of PRMTs have been developed and tested (12–14). However, because some PRMT isoforms display very limited substrate specificity, the development of catalytic inhibitors of PRMTs should not be the sole approach to treat cancers. Therefore, the identification of catalytic-independent functions of PRMTs may help in the development of new therapies with increased specificity and efficacy. Interestingly, among the members of arginine methyl transferase family that are upregulated in several cancers, including lung cancer, PRMT6 is known to display a narrow substrate specificity (15). However, studies investigating PRMT6 function in vivo, and in lung cancer in particular, have lagged behind other genes, leaving several open questions about their role.

In this study, to closely mimic PRMT6 amplification events in human lung tumors, we developed a lung-targeted PRMT6 gain-of-function (GOF) mouse model. Overexpression of PRMT6 was sufficient to drive de novo hyperproliferation in the lungs. Furthermore, PRMT6 overexpression potentiated chemical carcinogen (urethane)-induced lung tumor growth. We also demonstrate that PRMT6 overexpression promotes lung tumor progression via the alternate activation of tumor-associated macrophages (TAM). To explore the molecular mechanisms by which PRMT6 promotes alternate activation of TAMs and lung tumor growth, we used proteomics-based approaches and identified a protein–protein interaction (PPI) between PRMT6 and interleukin-enhancer binding protein 2 (ILF2), which is critical for the regulation of macrophage migration inhibitor factor (MIF). Collectively, our findings reveal a unique role for PRMT6 in potentiating lung tumor progression via the alternate activation of TAMs. Therefore, targeting the newly identified PRMT6/ILF2/MIF axis may open new possibilities for the therapeutic intervention of lung cancer.

Animal experiments were conducted in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH (Bethesda, MD). The animals were housed in Biologics Research laboratory vivarium, University of Illinois at Chicago (UIC). All the animal experiments were approved by the Institutional Animal Care and Use Committee.

The full-length cDNA sequences of open reading frame (ORF) for human PRMT6 were inserted into the multiple cloning site (MCS) of the pCAG-floxed STOP-3XFlag-MCS plasmid backbone. The constructed plasmid encoding the PRMT6 ORF was sequenced to confirm that it was mutation-free, and PRMT6 protein overexpression was verified through transient transfection into 293T cells with the PRMT6 STOP plasmid and Cre plasmid, followed by Western blot analysis using a Flag antibody. To generate the PRMT6 overexpression transgenic mice, the STOP plasmid was linearized and introduced into the pronucleus of day 1 fertilized embryos (FVB/N), by microinjection. Injected embryos were then transferred into day 1 plugged pseudopregnant female mice. Founder pups were genotyped by PCR, using the tail clip to examine the germline transmission.

PRMT6^Tg founder mice were backcrossed with the Sftpc-CreER^T2 mouse line to generate PRMT6^Tg; Sftpc-Cre^tm mice. Parental stocks of Sftpc-CreER^T2 were a generous gift from Dr. Brigid Hogan (Duke University; ref. 16). The genetic background of the mice was determined using PCR of DNA from tail biopsies.

Viable frozen splenocytes from PRMT6^Tg mouse were used and processed according to the CerGentis' targeted locus amplification (TLA) protocol (17). Briefly, two primer sets were designed on the basis of the hPRMT6 transgene and were used in individual TLA amplifications. PCR products were purified, and a library was prepared using the Illumina Nextera flex protocol, followed by sequencing on an Illumina sequencer. Sequence reads were mapped using Burrows–Wheeler Aligner's Smith-Waterman Alignment (BWA-SW; ref. 18), and NGS reads were aligned to the transgene sequence and host genome (mouse). Integration sites were detected based on the coverage peaks in the genome and the identification of fusion reads between transgene sequence and host genome.

Tumors were initiated in tamoxifen-treated PRMT6^Tg; Sftpc-Cre^tm mice and PRMT6^Tg mice via four weekly intraperitoneal injections (i.p.) of either 0.9% saline or urethane 1 g/kg body weight. The mice were euthanized and dissected after 20 weeks to assess the formation of lung tumors. Lung tumors were counted and measured using a digital caliper (Thermo Fisher Scientific).

H2122 parental cells and H2122 PRMT6 sgRNA cells (2 × 10⁶ viable cells/flank) were injected subcutaneously (s.c.) into the lower flanks of 6- to 8-week-old athymic nude mice (The Jackson Laboratory). Tumor growth was assessed by weekly caliper measurements. Four weeks after the implantation of the cells, mice were sacrificed; flank tumors were removed and weighed.

Human bronchial epithelial (Beas2B) cells and the NSCLC cell lines (H2122 and H1299) were obtained from ATCC. All cell lines were cultured in RPMI-1640 medium (10-040-CV, Cellgro, Mediatech Inc.) supplemented with 10% FBS in a humidified 5% CO₂ incubator at 37°C. Cell lines were cultured biweekly, and stocks of cell lines were passaged no more than 10 times for use in experiments. The cell lines were routinely tested for Mycoplasma contamination.

Guide RNAs (sgRNA) targeting the first exon of PRMT6 (GAAAAGAAAGCTTGAGTCGG) were cloned into the BbsI site of px330 plasmid (Addgene; plasmid # 42230). H2122 cells were transfected with the guide plasmids using Lipofectamine reagent (Invitrogen) according to the manufacturer's recommendations. A day after transfection, cells were trypsinized and were diluted to a concentration of 1 cell/100 μL, seeded into 96-well plates, and were grown for 3 weeks. Thirty single clones were isolated and assayed by Western blot analysis for PRMT6 expression. Multiple clones that showed complete loss of PRMT6 expression were selected for downstream analysis.

Double-stranded RNAs (siRNA) targeting the 3′UTR of human ILF2 (AGCUGCCUGCUACUGGGCUUUACAU) were purchased from Invitrogen, while siRNAs targeting MIF were purchased from Santa Cruz Biotechnology (sc-37137). Control siRNAs were purchased from Santa Cruz Biotechnology. NSCLC cells were treated with 100 nmol/L siRNAs by using Lipofectamine 2000 reagent according to the manufacturer's protocol.

Lung tissue protein extracts were obtained from mouse lungs homogenized in tissue protein extraction lysis buffer containing a cocktail of 1 mol/L HEPES, 5 mol/L NaCl, 10% Triton X-100, 1 mol/L DTT, 0.5% EDTA, 20 mmol/L NaVO3, 10 mmol/L PMSF, 0.5 mol/L Na-β-glycerophosphate and protease inhibitors using Qiagen TissueLyser LT (Qiagen), while NSCLC cells were lysed in a lysis buffer (0.5% Triton X-100, 50 mmol/L β-glycerophosphate, pH 7.2, 0.1 mmol/L sodium vanadate, 2 mmol/L MgCl₂, 1 mmol/L EGTA, 1 mmol/L dithiothreitol, 2 μg/mL leupeptin, and 4 μg/mL aprotinin), and the Western blotting analysis was carried out as previously described (6, 19). Briefly, aliquots of various protein extracts were resolved on SDS-PAGE gels and transferred to nitrocellulose. The filters were blocked in Tris-buffered saline (10 mmol/L Tris-HCl, pH 7.4, 140 mmol/L NaCl, containing 0.1% Tween-20 (TTBS) and 3% nonfat dry milk and then incubated with the same blocking solution containing the indicated antibodies at 0.5 μg/mL for 16 hours. Filters were extensively washed in TTBS, and bound antibodies were visualized with horseradish peroxidase (HRP)–coupled secondary antibodies. The following antibodies were used for immunoblotting: PRMT6 (14641) and GAPDH (5174) antibodies were from Cell Signaling Technology, FLAG (F3165) and actin (A3853) were from Sigma-Aldrich, ILF2 (ab28772) antibodies were from Abcam, and MIF (FL-115) antibodies were from Santa Cruz Biotechnology.

For 5-day cell growth assay, 25,000 cells were seeded in duplicates in a 24-well plate. Cell proliferation was measured for 5 days by trypsinization of cells, followed by cell counting using a hemocytometer. Cell numbers were represented in the graphs. For clonogenic cell proliferation assays, 1,000 cells per well were seeded in a 12-well culture plate followed by incubation at 37°C in a 5% CO₂ incubator. After 5 to 7 days, colonies were stained using a staining solution (0.5% crystal violet, 12% glutaraldehyde, 87.5% H₂O) for 1 hour at room temperature. After destaining in water and drying, colonies were quantified using Bio-Rad Chemidoc Imaging System and Quantity One Software. Cloning efficiency represents the mean number of colonies formed per well.

For assessing cell migration, 30,000 cells in serum-free media were seeded into the transwell inserts (Corning) containing 8-μm permeable pores and allowed to migrate toward 10% FBS containing medium. Later, the cells in the transwell inserts were removed, and the inserts were washed in PBS for three times. The migrated cells on the bottom of the insert were fixed with 2% glutaraldehyde solution followed by crystal violet (1%) staining. After washing the inserts three times with PBS, the inserts were allowed to air dry, and pictures were taken using an inverted microscope. Ten independent fields were counted for each transwell, and the average number of cells/field was represented in the graphs.

Soft-agar assays were performed as described previously (20). Briefly, 5,000 cells were plated in duplicates in a 6-well plate in growth medium containing 0.3% noble agar. After 14 days, colonies were stained and visualized with nitroblue tetrazolium chloride.

To identify PRMT6-associated proteins, lung lysates from FVB/NJ mice were incubated with GST or GST-PRMT6. Following SDS-PAGE analysis of PRMT6 immunocomplexes, the gel bands corresponding to the molecular range of 50 to 80 kDa were excised, destained, reduced, aklyated, and digested with trypsin. The resulting concentrated peptide extract was analyzed by automated microcapillary liquid chromatography tandem mass spectrometry. Full mass (MS) spectra were recorded on the peptides over a 400 to 2000 m/z range. Only peptides with a P value of at least 0.02 were analyzed further.

Cell lysates derived from Beas2B cells expressing empty vector of pCMV-HA-hILF2 were subjected to immunoblotting-based cytokine profiling using the Proteome Profiler Human XL Cytokine Array kit (ARY022B; R&D Systems).

Lungs from the PRMT6^Tg; Sftpc-Cre^tm and PRMT6^Tg mice 10 days after last tamoxifen treatment were harvested, and cells were isolated using a mouse tumor cell isolation kit (130-110-187; Miltenyi Biotec) according to the manufacturer's recommendations. The cell suspension was incubated in a 100-mm cell culture dish precoated with mouse IgG for 2 hours. Later, the nonadherent cells were collected, centrifuged, resuspended in DMEM complete medium (10% FBS, 1% penicillin/streptomycin, 25 mmol/L HEPES, and 10 ng/mL keratinocyte growth factor). The cells were plated on culture dishes coated with fibronectin, collagen, and BSA, and the medium was changed every 2 days. On day 5, the medium was collected, centrifuged to remove the cells, aliquoted, and stored at −80°C.

Bone marrow–derived monocytes were isolated from femurs and tibias of wild-type FVB/NJ mice and cultured in the presence of macrophage-colony stimulating factor (M-CSF; R&D Systems) to promote macrophage maturation as previously described (21). After 3 to 7 days, these cells display the morphology of mature (M0) macrophages. M0 macrophages were later treated with the conditioned medium isolated from PRMT6^Tg; Sftpc-Cre^tm and PRMT6^Tg mice for 24 hours. Later, total RNA was isolated from the macrophages, and markers of M1 and M2 polarization were assayed by qPCR.

Lungs from PRMT6^Tg; Sftpc-Cre^tm and PRMT6^Tg mice 20 weeks after tamoxifen and urethane treatment were enzymatically digested, and CD11b⁺ cells were positively selected using autoMACs ProSeparator (Miltenyi Biotec). Purity of the isolated cells was later confirmed by flow cytometry using CD11b and F4/80 macrophage markers. RNA from the isolated cells was isolated, and the expression of macrophage M1 and M2 markers was determined by quantitative PCR (qPCR).

Total RNA from the lung tissues, NSCLC cells, CD11b⁺ cells, and macrophages was obtained using TRIzol reagent (Invitrogen) as per the manufacturer's recommendations. For quantitative RT-PCR, 3 μg of total RNA was reverse transcribed using random primers, and real-time PCRs were performed using the QuantiTect SYBR Green PCR kit (204050, Qiagen) and the Bio-Rad CFX qPCR detection system. The primers utilized in the PCR experiments were shown in Supplementary Table S2.

Lung tumor sections from PRMT6^Tg; Sftpc-Cre^tm and PRMT6^Tg mice 20 weeks after tamoxifen and urethane treatment were used in immunostaining procedure using the following antibodies: CD68 (FA-11, Bio-Rad) for identifying macrophages, iNOS (ab15323, Abcam) to detect M1 polarization, Arginase I (sc-18351, Santa Cruz Biotechnology) to detect M2 polarization, EPCAM (VU1D9, Cell Signaling Technology), alpha smooth muscle actin (ab5694, Abcam), and CD31 (550274, BD Biosciences) using procedures described previously (22). Briefly, lung sections were fixed with 3% paraformaldehyde and permeabilized with 0.01% Triton X-100 followed by blocking with 3% normal horse serum. The samples were then exposed to primary antibodies and detected with fluorescent dye–conjugated secondary antibodies. Image acquisition was performed using an Evos digital inverted microscope (Thermo Fisher) with a 40× objective.

In vitro methylation assays were performed as described previously (6). Briefly, GST-ILF2 or histone 3 (1 μg) was incubated with HA-affinity matrix containing bound PRMT6 and 1 μCi of S-adenosyl-L-[methyl-3H] methionine (NEN radiochemicals, 250 μCi, 9.25 MBq), at 30 °C for 1 hour. After 1 hour, the reactions were stopped and separated on an SDS-PAGE gel. The separated proteins were later transferred to nitrocellulose membrane, amplified (Autofluor, National Diagnostics, 2 hours), dried, and fluorography was performed.

Data were compiled from at least three independent, replicate experiments, each performed on separate cultures and on separate occasions. The responses were displayed as “fold changes.” Comparisons of data among experimental groups were performed using Student t test to assess variance. Increase in statistical significance (P < 0.05) is denoted with an “asterisk” symbol, while a decrease in statistical significance (P < 0.05) is denoted with a “hash” symbol.

PRMT6 was shown to be upregulated in lung tumors (11). Consistent with the publication by Yoshimatsu and colleagues, (11), mining The Cancer Genome Atlas (TCGA) lung adenocarcinoma (LUAD) data sets revealed a significant upregulation of PRMT6 in lung tumors when compared with normal lung tissues (Supplementary Fig. S1A). To further assess the prognostic significance of high PRMT6 expression in lung cancer patients, we analyzed TCGA LUAD data sets by using Kaplan–Meier plotter (www.kmplot.com). Our analyses indicated that high PRMT6 expression is significantly correlated with poor overall survival (Supplementary Fig. S1B). To determine whether PRMT6 expression is also elevated at the protein level, we used formalin-fixed paraffin-embedded (FFPE) human tissue samples and fresh frozen lung tumor tissues (Supplementary Fig. S1C and S1D). IHC staining of FFPE sections with PRMT6 antibody showed a strong nuclear staining of PRMT6, while no such staining was observed in secondary antibody alone control (Supplementary Fig. S1C). The slides were later scanned using Aperio imagescope scanner (Leica Biosystems), followed by the scoring of the tissue slides by an H-score method that combines the staining intensity with the percentage of cells stained (24). A significant increase in PRMT6 staining was observed in the tumor tissues when compared with adjacent uninvolved lung tissues (N = 8; Supplementary Fig. S1C). Similarly, immunoblotting of fresh frozen lung tumor tissues also revealed a robust increase in PRMT6 expression in the tumors when compared with uninvolved lung tissues of the same patient (N = 5; Supplementary Fig. S1D). Together, these results indicate that PRMT6 might be a significant contributor to LUAD and provide evidence that PRMT6 may represent a novel prognostic indicator for this disease.

PRMT6 was shown to be upregulated in lung tumors (11). However, mouse models to investigate PRMT6 function in vivo, and in lung cancer in particular, have lagged behind other genes, leaving several open questions about their role. To closely mimic PRMT6 amplification events in human lung tumors, we created a PRMT6 GOF mouse model by cloning the ORF for human PRMT6 isoform into the MCS of the pCAG-Lox-STOP-Lox-3XFlag-MCS plasmid backbone (LSL-hPRMT6) that keeps the LSL-hPRMT6 allele silent until the “stop” cassette is deleted by a Cre recombinase. The LSL-hPRMT6 plasmid was later linearized and introduced into the pronucleus of day 1 fertilized embryos (FVB/N), by microinjection. Injected embryos were then transferred into day 1 plugged pseudopregnant female mice. Founder pups were genotyped by PCR, using the tail clip to examine the germline transmission. In order to identify the exact transgene integration sites, we performed TLA analysis on splenocytes from LSL-hPRMT6 mice (hereafter referred to as PRMT6^Tg). TLA analysis mapped the transgene insertion site to mouse chromosome 11 at intronic region between exons 10 and 11 and ending in the intronic region between exons 24 and 25, which results in a 21-kb deletion of Spag9 (Fig. 1A). The copy number is estimated to be around 5 to 8 copies. To avoid possible phenotypes arising due to random insertion of PRMT6 transgene, transgenic mice were maintained in heterozygous genotype.

For lung-targeted expression of PRMT6, we generated an inducible hPRMT6 GOF transgenic by crossing PRMT6^Tg to a tamoxifen-regulated surfactant protein C-driven Cre-recombinase mouse line (Sftpc-Cre^tm; ref. 16). Daily injections of tamoxifen (100 mg/kg body weight) for 5 days led to a robust induction of hPRMT6 expression in the lungs of PRMT6^Tg; Sftpc-Cre^tm, but not in PRMT6^Tg control mice within a week after the last tamoxifen administration (Fig. 1B). PRMT6 overexpression caused hyperproliferation of the lung epithelium as demonstrated by an increase in ki-67 immunostaining (Fig. 1C and D). Consistent with our observations, a recent report demonstrated hyperbranching of the mammary glands and increased ki-67 staining in the breast epithelium of PRMT6 GOF transgenic mice (25).

Because PRMT6 and KRas mutations are co-occurring in 24% (272 patients) of queried patients/samples (cBioPortal; www.cbioportal.org), we determined whether PRMT6 overexpression can potentiate mutant KRas-driven lung tumor formation. For these studies, we interrogated the effects of urethane (ethyl carbamate) on PRMT6^Tg and PRMT6^Tg; Sftpc-Cre^tm mice, a prototypical model to study lung tumorigenesis (23, 26, 27). Urethane is a chemical carcinogen, which causes activating mutations in KRas (23, 26, 27), leading to the formation of lung tumors in mice. Several studies have highlighted a strain-specific response to urethane; FVB/NJ showed lung tumorigenesis within 20 weeks after 1 dose (1 g/kg body weight) of urethane administration, while C57Bl/6J, being resistant, require 6 weekly doses (1 g/kg body weight) of urethane and develop lung tumors within 40 weeks (23, 26, 27). Because the PRMT6^Tg; Sftpc-Cre^tm mice were in mixed background, we determined the urethane administration regimen empirically that results in 5 to 10 lung tumors in control mice after intraperitoneal administration. Strikingly, 4 weekly injections (intraperitoneal, 1 g/kg body weight) of urethane to PRMT6^Tg; Sftpc-Cre^tm mice after tamoxifen administration (for PRMT6 induction; Fig. 2A) resulted in a more severe phenotype in terms of tumor size (Fig. 2B) and number (Fig. 2C) when compared with PRMT6^Tg control mice treated with tamoxifen and urethane. Masked histologic examination by a lung pathologist revealed that while control mice mostly developed microscopic adenomas, the PRMT6Tg; Sftpc-Cre^tm mice, on the other hand, developed macroscopic aggressive LUADs. Furthermore, indirect immunofluorescence staining of the tumor sections with CD31 antibodies consistently revealed more mature blood vessels in the tumors of PRMT6^Tg; Sftpc-Cre^tm mice, when compared with the tumors of PRMT6^Tg control mice (Fig. 2D). Taken together, we show that PRMT6 is an important contributor to tumor progression and metastasis in mutant KRas-driven lung tumors.

To investigate the function of PRMT6 in NSCLC, we used CRISPR-Cas9–mediated gene editing of PRMT6 in NSCLC cell lines: H2122 (Fig. 3A–3E) and H1299 (Fig. 3F–3J). For generating PRMT6 knockout cell lines, we transfected H2122 and H1299 cells with single-guide RNAs (sgRNA) targeting PRMT6 followed by clonal isolation via limiting dilutions. H2122 and H1299 PRMT6 knockout clones displayed a consistent reduction in cell proliferation when compared with their parental cells, as determined by cell growth curves (Fig. 3B and 3G) and clonogenic cell proliferation assays (Fig. 3C and 3H). We next investigated the effects of PRMT6 knockout on cell migration by using transwell assays. H2122 and H1299 PRMT6 knockout clones displayed reduced cell motility when compared with their respective parental cells (Fig. 3D and 3I). To further explore the functional consequence of PRMT6 gene knockout, we evaluated anchorage-independent growth capabilities of H2122 and H1299 PRMT6 knockout clones. Clearly, depletion of PRMT6 in H2122 and H1299 cells resulted in a reduced number of colonies in soft agar when compared with their corresponding parental cells (Fig. 3E and 3J).

We also tested the role of PRMT6 in lung tumor growth in vivo by performing xenograft studies in athymic nude mice (Fig. 3K–3M). For these studies, H2122 parental cells and H2122 PRMT6 sgRNA cells were injected s.c. into the lower flanks of athymic nude mice. Tumor growth was assessed by weekly caliper measurements. It was striking to observe that PRMT6 depletion resulted in a dramatic reduction in tumor growth within 4 weeks of implantation of cells (Fig. 3K–3M). Collectively, these data support an important role for PRMT6 in promoting cell proliferation, cell migration, anchorage-independent growth, and in vivo tumor growth of NSCLC cells.

Because the molecular functions of PRMT6 are governed by the proteins it interacts with and/or its substrates, we sought to identify PRMT6-associated proteins of the lung. For these studies, mouse lung lysates were used in pulldowns by using GST-PRMT6 as a “bait” followed by proteomics. Affinity pulldown reactions resulted in the identification of several proteins that are specific to GST-PRMT6, but not GST pulldowns. Of the identified proteins, only the proteins with four or more unique peptides were considered as potential PRMT6-associated proteins (Fig. 4A; Supplementary Table S1). The STRING database provides critical assessment and integration of protein–protein interactions that are both direct and indirect (28). We created a STRING-curated protein interaction network (Fig. 4B) for the proteins that met our selection criteria (Supplementary Table S1). With the exception of 3 proteins, the remaining proteins had at least one interaction with another protein (either predicted or experimentally determined), indicating an interconnectivity among the identified proteins (Fig. 4B). We elected to focus on interleukin-enhancer binding factor 2 (ILF2) as it is identified as an arginine methylated protein in a proteomic screen (29), and given its previous association with malignant tumors of diverse origins (30–35).

To validate the interaction of ILF2 with PRMT6, affinity pulldowns were performed. For these studies, human bronchial epithelial cells (Beas2B) were engineered to express HA-PRMT6. Later, cell lysates were used in pulldown assays with anti-HA antibodies. The presence of endogenous ILF2 in the HA-PRMT6 immunocomplex was made possible by probing the immunoblots with anti-ILF2 antibodies. A strong PRMT6–ILF2 interaction was detected (Fig. 4C). Furthermore, to determine whether the PRMT6–ILF2 interactions were direct or indirect, we purified recombinant His-PRMT6 and GST-ILF2 in E. coli. Later, the two proteins were mixed, and pulldowns were performed with GST-agarose beads. A strong PRMT6–ILF2 interaction was detected in GST–ILF2 pulldowns, but not in GST pulldowns (Fig. 4D). Collectively, these data provide evidence for a direct interaction between PRMT6 and ILF2.

To evaluate if ILF2 is a PRMT6 substrate, we performed in vitro methylation assays (Fig. 4E). For these studies, HA-PRMT6 was expressed and isolated from Beas2B cells via HA-affinity pulldowns. However, recombinant GST–ILF2 was purified from E. coli. Later, the abilities of HA-PRMT6 to methylate GST-ILF2 were evaluated in the presence of radiolabeled (³H) S-adenosyl L-methionine (SAMe), a methyl group donor. To our surprise, PRMT6 failed to catalyze ILF2 methylation (Fig. 4E). The fact that PRMT6 could methylate histone 3, a known PRMT6 substrate (36), indicates that the PRMT6 used in in vitro methylation reactions was catalytically active (Fig. 4E). Taken together, these data indicate that ILF2 is a PRMT6-interacting protein, but not a catalytic substrate of PRMT6.

It was shown earlier that increased expression of ILF2 is associated with poor overall survival in tumors of diverse origin (31, 32, 34). However, the upstream regulators of ILF2 expression remain to be identified. In this study, we have identified ILF2 as a novel PRMT6-interacting protein (Fig. 4). Interestingly, Spearman rank correlation analysis of the TCGA lung cancer data sets revealed a significant positive correlation between PRMT6 and ILF2 (Fig. 5A). Furthermore, mutually dependent expression of PRMT6 and ILF2 proteins in the lung tumors of the patient samples was also detected (Fig. 5B). These observations suggest that PRMT6 is a novel regulator of a protumorigenic protein ILF2 in the lung. To explore this exciting possibility, we performed overexpression of HA-tagged hPRMT6 in nontransformed bronchial epithelial cells (Beas2B). Overexpression of PRMT6 resulted in a dramatic increase in ILF2 protein levels, as determined by Western blotting with anti-ILF2 antibodies (Fig. 5C). To fully complement our overexpression studies, we also evaluated ILF2 levels in PRMT6 knockout H2122 and H1299 clones. It was striking to observe a complete loss of ILF2 protein levels, but not mRNA levels (data not shown), upon PRMT6 depletion in both H2122 (Fig. 5D) and H1299 (Fig. 5E). Consistent with the effects of PRMT6 modulation on ILF2 protein levels in vitro, probing mouse lung lysates of PRMT6^Tg; Sftpc-Cre^tm mice (post-tamoxifen and urethane administration) also resulted in a robust induction of ILF2 levels when compared with the PRMT6^Tg control mice (Fig. 5F). In vitro methylation assays demonstrate that ILF2 is not a catalytic substrate of PRMT6 (Fig. 4E). To evaluate whether PRMT6 catalytic activity is required for ILF2 regulation, we performed overexpression of HA-hPRMT6 and HA-hPRMT6-KLD (methyl transferase dead mutant) in nontransformed bronchial epithelial cells (Beas2B). Overexpression of either PRMT6 or PRMT6-KLD mutant resulted in a dramatic increase in ILF2 protein levels (Fig. 5G). Taken together, these observations establish that PRMT6 regulates ILF2 protein levels in a catalytic-independent manner.

To determine if ILF2 mediates the proproliferative effects of PRMT6, we used RNA interference by designing siRNAs targeting the 3′ untranslated region (UTR) of ILF2 (Fig. 5H). Treatment of H1299 cells with ILF2 siRNAs resulted in a reduction in cell proliferation as determined by clonogenic cell proliferation assays (Fig. 5H). Furthermore, the effects of ILF2 depletion on cell proliferation could be rescued by the expression of siRNA-resistant HA-ILF2 plasmid (Fig. 5H), indicating the specificity of the siRNAs to ILF2. To determine the biological significance of PRMT6-mediated ILF2 regulation, we performed overexpression of PRMT6 together with ILF2 siRNAs in Beas2B cells (Fig. 5I). As anticipated, overexpression of PRMT6 resulted in a robust increase in cell growth, as determined by hemocytometer cell count (Fig. 5I). Depletion of ILF2 in PRMT6-overexpressing cells resulted in a reduction in PRMT6-induced cell proliferation (Fig. 5I), indicating that the biological effects of PRMT6 were indeed ILF2 dependent.

Because ILF2 is an important regulator of the T-cell expression of IL2 (37), we hypothesized that PRMT6/ILF2 signaling potentiates lung tumor growth by modulating cytokine levels. To identify the cytokines regulated by ILF2, we used whole-cell lysates of bronchial epithelial cells (Beas2B) expressing either empty vector or HA-ILF2 on a 105-cytokine antibody array (Proteome profiler Human XL Cytokine Array kit, R&D Systems). Interestingly, forced expression of ILF2 resulted in a robust induction of four cytokines, e.g., ICAM1, IL8, IL32, and MIF (Fig. 6A). Of note, these cytokines were not previously associated with either ILF2 or PRMT6. Macrophage MIF was of interest due to its protumorigenic role in a number of cancer types (38–41), and due to its most prominent induction by ILF2 (Fig. 6B).

To validate the regulation of MIF by ILF2, we performed overexpression of HA-tagged ILF2 in Beas2B cells (Fig. 6C). Overexpression of ILF2 resulted in a robust increase in MIF expression (Fig. 6C). Complementary studies involving siRNA-mediated knockdown of ILF2 in H2122 (Fig. 6D) and H1299 (Fig. 6E) cells resulted in a dramatic reduction in MIF expression. We also determined if PRMT6, an upstream regulator of ILF2 (Fig. 5), could also regulate MIF. For these studies, we probed the cell lysates of H2122 and H1299 PRMT6 knockout clones. Depletion of PRMT6 in H2122 and H1299 cells resulted in a dramatic reduction in MIF expression (Fig. 6F and G). In agreement with the effects of PRMT6 depletion on MIF expression in vitro, probing mouse lung lysates of PRMT6^Tg; Sftpc-Cre^tm mice (post-tamoxifen and urethane administration) also resulted in a robust induction of MIF expression when compared with the PRMT6^Tg control mice (Fig. 6H). Taken together, these observations indicate that PRMT6 and ILF2 are novel regulators of MIF. To further investigate whether PRMT6 stimulated MIF expression was ILF2 dependent, we performed overexpression of PRMT6 in Beas2B cells (Fig. 6I). As anticipated, overexpression of PRMT6 in Beas2B cells resulted in a robust increase in MIF expression (Fig. 6I), which was blocked by depletion of ILF2 (Fig. 5I). Taken together, these data indicate that the PRMT6/ILF2 signaling axis is a novel regulator of MIF expression in NSCLC.

To determine the biological significance of ILF2-mediated MIF expression, we performed overexpression of ILF2 together with MIF siRNAs in Beas2B cells (Fig. 6J). Overexpression of ILF2 resulted in a significant increase in cell growth, as determined by hemocytometer cell count (Fig. 6K) and clonogenic assays (Fig. 6L), which was blocked by the depletion of MIF (Fig. 6K and L), indicating that the biological effects of ILF2 are MIF dependent.

Prior studies highlighted an important role for MIF in the alternate (M2) activation of macrophages within the tumor microenvironment (42). Because the PRMT6/ILF2 signaling axis regulates MIF (Fig. 6), we hypothesized that increased expression of PRMT6 potentiates lung tumor growth through the alternate activation of macrophages. To explore this possibility, we probed the expression of classical and alternate activation markers within the macrophages via indirect immunofluorescence staining of lung tumor sections of PRMT6^Tg; Sftpc-Cre^tm mice and PRMT6^Tg control mice post-tamoxifen and urethane treatment (Fig. 7A). TAMs of PRMT6^Tg control mice displayed classical activation of macrophages with a proinflammatory phenotype [high M1 marker (iNOS) and low M2 (Arginase I) marker expression; Fig. 7A], while the TAMs of PRMT6^Tg; Sftpc-Cre^tm mice were skewed toward an M2-like tumor-promoting phenotype [low M1 marker (iNOS) and high M2 marker (Arginase I) expression; Fig. 7A]. To gain more insights into PRMT6-mediated macrophage polarization, CD11b⁺ cells were isolated from total lungs of PRMT6^Tg; Sftpc-Cre^tm and PRMT6^Tg mice treated with tamoxifen and urethane. Eighty-three percent (83%) of CD11b⁺ cells were positive for F4/80, an alveolar macrophage-specific marker, displaying the purity of the TAM population (Fig. 7B). Later, RNAs from the CD11b⁺ cells were isolated, and the expression of M1 and M2 macrophage markers was determined by quantitative PCR (qPCR). Consistent with the indirect immunofluorescence data (Fig. 7A), CD11b⁺ cells isolated from PRMT6^Tg; Sftpc-Cre^tm lung tumors showed increased expression of M2 markers (Arg I and IL10) and decreased expression of M1 markers (TNFα and iNOS; Fig. 7B). In contrast, CD11b⁺ cells isolated from PRMT6^Tg control lung tumors displayed increased expression of M1 markers (Fig. 7C). These studies demonstrate for the first time that PRMT6 is an important modulator of the lung tumor microenvironment via the alternate activation of TAMs.

To determine if the secreted factors from PRMT6-overexpressing alveolar epithelial cells (AEC) induce macrophage polarization toward the M2 phenotype, we isolated and cultured AECs from PRMT6^Tg; Sftpc-Cre^tm and PRMT6^Tg mice treated with tamoxifen only (Fig. 7D). Later, conditioned medium from AECs was collected. We also cultured bone marrow–derived monocytes in the presence of macrophage-colony stimulating factor (mCSF) to promote macrophage maturation (M0). Treatment of mouse bone marrow–derived macrophages (M0) with the conditioned medium from PRMT6-overexpressing AECs resulted in an increase in M2 markers (Arginase 1 and IL10), and a decrease in an M1 marker (TNFα), when compared with control conditioned medium–treated macrophages (Fig. 7D). As a complementary strategy, we used conditioned medium collected from H2122 PRMT6 knockout clones (Fig. 3) on THP1 monocytic cell–derived macrophages (5). Macrophages treated with the conditioned medium collected from PRMT6 knockout 2122 clone showed a significant reduction in M2 macrophage polarization as detected by Mannose receptor 1 (MRC1) expression when compared with macrophages treated with the conditioned medium collected from parental H2122 cells (Fig. 7E). These observations indicate that PRMT6-induced paracrine signaling products promote the alternate activation of macrophages.

Alternate activation of TAMs is known to promote tumor progression by supporting tumor neoangiogenesis (42). To determine whether PRMT6-stimulated M2 macrophage polarization promotes macrophage angiogenic potential, we measured the levels of vascular endothelial growth factor (VEGF) and matrix metalloproteinase 9 (MMP9), which are the markers of angiogenesis. Mouse bone marrow–derived macrophages treated with the conditioned medium from PRMT6-overexpressing AECs, which were M2 polarized (Fig. 7C), showed a significant increase in the expression of VEGF and MMP9 transcripts (Fig. 7F) when compared with the control conditioned medium–treated macrophages (Fig. 7E). Furthermore, we also detected increased VEGF and MMP9 expression in the lung tumors of PRMT6-overexpressing mice post-tamoxifen and urethane treatment (Fig. 7G), suggesting increased angiogenesis in the tumors of PRMT6-overexpressing mice. In agreement with the increased expression of VEGF, there were more mature blood vessels in the tumors of PRMT6^Tg; Sftpc-Cre^tm mice when compared with the tumors of control (PRMT6^Tg), as determined by CD31 immunostaining (Fig. 7H). Collectively, these findings indicate a new role for PRMT6 in potentiating lung tumor progression and tumor angiogenesis via the alternate activation of TAMs.

Despite several studies highlighting the importance of PRMTs in multiple tumor types (6–11), pharmacologic inhibitors of PRMTs have not reached the clinic. This is in part due to the current strategies aimed at targeting the catalytic domains of PRMTs, which may not work for PRMTs that display narrow substrate specificities. Therefore, the identification of new roles, apart from their classical catalytic roles, of PRMTs in tumor development would be beneficial to develop new therapeutics with increased specificity. To our knowledge, PRMT6, which is overexpressed in lung cancer, is the only member of arginine methyl transferase family that displays narrow substrate specificity (15). These observations hint at previously undisclosed noncatalytic functions of PRMT6 in lung tumor development.

In the current study, we made several observations that will further our knowledge of lung tumor progression. First, to closely mimic PRMT6 amplification events in human LUAD, we created a lung-targeted PRMT6 GOF mouse model. Induction of PRMT6 expression resulted in hyperproliferation of the alveolar epithelium, as determined by ki-67 staining (Fig. 1B and C). Consistent with our observations, a recent report demonstrated hyperbranching of the mammary glands and increased ki-67 staining in the breast epithelium of PRMT6 GOF transgenic mice (25). These observations are indicative of a genetic predisposition of PRMT6 GOF mice to increased carcinogen-induced lung tumor growth. Indeed, when we interrogated the effects of a common oncogenic driver in lung cancer, e.g., mutant KRas in PRMT6 GOF mice, we observed a striking increase in lung tumor number in mouse tumor models (Fig. 2). We also observed differences in tumor histology; while control mice mostly developed microscopic adenomas, the PRMT6^Tg; Sftpc-Cre^tm mice, on the other hand, developed macroscopic aggressive lung adenocarcinomas (Fig. 2B). Furthermore, the tumors of PRMT6^Tg; Sftpc-Cre^tm mice (experimental) also showed more mature blood vessels, an indication of neoangiogenesis (Fig. 2D). Taken together, we show that PRMT6 is an important contributor to tumor progression and metastasis in mutant KRas-driven lung tumors. Of note, this mouse model can also serve as an important preclinical tool to evaluate the efficacy of PRMT6 inhibitors.

Second, we have also identified a PPI between PRMT6 and ILF2, which contributes to lung tumor development (Fig. 4B and C). We also demonstrate that ILF2 is not a catalytic substrate of PRMT6 (Fig. 4D). A similar role for PRMT6-mediated regulation of proteins independent of arginine methylation has been reported earlier (15). These findings suggest that PRMT6 displays noncatalytic functions, probably as a scaffold that contributes to lung tumor growth. However, how PRMT6 recruits, organizes, and regulates their binding partners in molecular terms awaits further study.

Third, ILF2 is a known regulator of T-cell expression of cytokines (37, 43–45). By using cytokine arrays, we have identified macrophage MIF, a proinflammatory cytokine, as a downstream target of ILF2. We also demonstrate that the proproliferative effects of the PRMT6–ILF2 signaling axis were MIF dependent (Fig. 6J–L). However, the other known functions of ILF2 in DNA metabolism (46), transcription (47), RNA export (48), and microRNA biogenesis (30) in lung cancer development remain to be discerned.

Fourth, we have identified that PRMT6 promotes lung tumor growth by reprogramming macrophages to a tumor-promoting (M2) phenotype. Earlier studies highlighted an important role for MIF in alternate activation of macrophages to a tumor-promoting phenotype (42). In this study, we have identified that TAMs in PRMT6^Tg; Sftpc-Cre^tm mice (post-tamoxifen + urethane) were observed to be alternatively activated to M2 phenotype (Fig. 7). We also show the PRMT6–ILF2 axis as a novel regulator of MIF expression (Fig. 6). Taken together, these observations suggest that the PRMT6–ILF2–MIF signaling axis contributes to the alternate activation of TAMs during lung adenocarcinoma development. However, more detailed studies using MIF-neutralizing antibodies and/or inhibitors are needed to fully demonstrate the new paradigm, which is clearly a major contributing factor for immunosuppression and neoangiogenesis (Fig. 7H) during lung tumor progression. Furthermore, the PRMT6 GOF lung tumor model can also serve as a preclinical tool for the investigation of immunotherapy efficacy and function.

In conclusion, we have identified a unique role for the PRMT6/ILF2/MIF signaling axis in promoting lung tumor growth. We also demonstrate that PRMT6 promotes lung tumor growth via its scaffold functions, but not its catalytic function (Fig. 4), offering new therapeutic opportunities for the development of inhibitors of PRMT6 and ILF2 heterodimerization as antineoplastic agents. Furthermore, PRMT6 is the only methyl transferase that displays automethylation, which is critical for its stability and regulation (49). Another therapeutic approach would be to target the PRMT6 automethylation. Therapeutic targeting based on these unique features, e.g., scaffold and/or automethylation, would not only provide us novel drugs to treat lung cancers, but also provide us new tools to study the role of PRMT6 in alternate activation of macrophages.

M.T. Bedford has ownership interest (including patents) in Epicypher. V. Gupta is cofounder of and has ownership interest (including patents) in Adhaere Pharmaceuticals, Inc. R.A. Winn is a member of Genentech external council and is an unpaid consultant/advisory board member for NIH. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Avasarala, V. Gupta, R.A. Winn, R.K. Bikkavilli

Development of methodology: S. Avasarala, S.Q. Khan, A. Di Lorenzo, V. Gupta, R.K. Bikkavilli

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Avasarala, P.-Y. Wu, S.Q. Khan, S. Yanlin, M.V. Scoyk, J. Bao, M.T. Bedford, R.K. Bikkavilli

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Avasarala, P.-Y. Wu, S.Q. Khan, O. David, R.K. Bikkavilli

Writing, review, and/or revision of the manuscript: S. Avasarala, S.Q. Khan, R.A. Winn, R.K. Bikkavilli

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.-Y. Wu, M.V. Scoyk

Study supervision: R.K. Bikkavilli

We thank UIC genome editing core (GEC) for designing PRMT6 guide RNAs, Research Histology and Tissue Imaging core at UIC Research Resources Center for IHC staining of lung tumor FFPE sections with PRMT6 antibody, and UIC Biorepository for providing human lung tumor biopsy tissues. We thank Biological Mass Spectrometry Core, University of Colorado, Denver, for their help in in-gel tryptic digestion and protein identification. We would also like to thank CerGentis for TLA analysis. We thank Dr. Juel Chowdhury for Spearman rank correlation analysis of lung cancer TCGA data sets. This study was supported in part by A Breath of Hope Lung Foundation (ABOHLF) fellowship to R.K. Bikkavilli, by a Merit Award from the U.S. Department of Veterans Affairs, and NIH grant R01CA138528 to R.A. Winn. M.T. Bedford is supported by NIH grant R01GM126421. V. Gupta is supported by a grant from Bears Care.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.