MICAL2 Promotes Pancreatic Cancer Growth and Metastasis

Characterization of the epigenomic landscape of pancreatic cancer to identify early drivers of tumorigenesis uncovered MICAL2 as a super-enhancer–associated gene critical for tumor progression that represents a potential pharmacologic target.

Abstract

Pancreatic ductal adenocarcinoma (PDAC) remains one of the deadliest solid cancers; thus, identifying more effective therapies is a major unmet need. In this study, we characterized the super-enhancer (SE) landscape of human PDAC to identify drivers of the disease that might be targetable. This analysis revealed MICAL2 as an SE-associated gene in human PDAC, which encodes the flavin monooxygenase enzyme that induces actin depolymerization and indirectly promotes serum response factor transcription by modulating the availability of serum response factor coactivators such as myocardin-related transcription factors (MRTF-A and MRTF-B). MICAL2 was overexpressed in PDAC, and high-MICAL2 expression correlated with poor patient prognosis. Transcriptional analysis revealed that MICAL2 upregulates KRAS and epithelial–mesenchymal transition signaling pathways, contributing to tumor growth and metastasis. In loss- and gain-of-function experiments in human and mouse PDAC cells, MICAL2 promoted both ERK1/2 and AKT activation. Consistent with its role in actin depolymerization and KRAS signaling, loss of MICAL2 also inhibited macropinocytosis. MICAL2, MRTF-A, and MRTF-B influenced PDAC cell proliferation and migration and promoted cell-cycle progression in vitro. Importantly, MICAL2 supported in vivo tumor growth and metastasis. Interestingly, MRTF-B, but not MRTF-A, phenocopied MICAL2-driven phenotypes in vivo. This study highlights the multiple ways in which MICAL2 affects PDAC biology and provides a foundation for future investigations into the potential of targeting MICAL2 for therapeutic intervention.

Significance: Characterization of the epigenomic landscape of pancreatic cancer to identify early drivers of tumorigenesis uncovered MICAL2 as a super-enhancer–associated gene critical for tumor progression that represents a potential pharmacologic target.

Introduction

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies worldwide, characterized by early metastasis, resistance to conventional therapies, and a dismal prognosis (1). Despite notable progress in comprehending the genetics of PDAC, discerning transcriptional subtypes, and an increased understanding of its tumor microenvironment, efficacious treatment modalities remain elusive (2). It is discouraging that our most effective therapeutic approaches still predominantly comprise traditional cytotoxic therapies, which often entail significant adverse effects on patients’ quality of life.

In recent years, extensive research has shed light on large, highly active chromatin regions known as “super-enhancers” (SE), which play a critical role in defining cell identity and cell state in both normal and malignant cells (3, 4). Previous studies have indicated that histone 3 lysine 27 acetylation (H3K27ac) marks serve as a reliable indicator for demarcating SE regions. These chromatin regions can regulate key genes that govern the cell phenotype. In tumor cells, this regulatory mechanism may encompass both oncogene and nononcogene drivers of the transformed state. We hypothesized that PDAC is driven and sustained by SE-associated genes and that delineating these genes could reveal novel promising therapeutic targets for drug development. To investigate this hypothesis, we characterized the epigenetic landscape of the normal pancreas and PDAC tissues obtained from primary patient samples. Using chromatin immunoprecipitation and sequencing (ChIP-seq), we identified MICAL2 (microtubule associated monooxygenase, calponin and LIM domain containing 2) as a putative SE-associated gene in human PDAC samples and confirmed its overexpression (OE) at the RNA and protein levels in both human tissues and cell lines as well as in murine models.

MICAL2 is a member of the MICAL (molecules interacting with CasL) protein family, consisting of evolutionarily conserved flavin monooxygenases, the canonical function of which is the oxidation and resultant depolymerization of actin (5). Unique to its other family members MICAL1 and MICAL3, MICAL2 has no autoinhibitory domain and is thus constitutively active. MICAL2, which is present in both the cytoplasm and the nucleus, was previously shown to indirectly regulate serum response factor (SRF)–mediated transcription through its modulation of nuclear G actin levels (6). G actin acts to sequester myocardin-related transcription factors A and B (MRTF-A and MRTF-B), coactivators of SRF. Nuclear accumulation of MRTFs is associated with the upregulation of genes associated with cell migration, fibrosis, and epithelial–mesenchymal transition (EMT), although MRTF-A has been the subject of most cancer-related studies (7). Interestingly, the role of MRTF-B in oncogenesis is much more uncertain, with some studies linking it to mesenchymal and hepatocellular tumor progression, whereas a recent study concluded that it acts as a tumor suppressor in human and murine colorectal cancers (8). MICAL2 was first linked to malignant disease when its splice variants were found to be overexpressed in prostate cancer (9, 10). More recently, studies have revealed that MICAL2 may promote EMT, migration, and invasion in lung (11), gastric (12), and breast cancers (13), yet no prior studies have experimentally probed the role of MICAL2 in pancreatic cancer biology, nor comprehensively characterized MICAL2-regulated pathways and the specific roles of MRTF-A and MRTF-B in oncogenic phenotypes. After we identified MICAL2, we found that its expression correlated with poor prognosis in patients with pancreatic cancer who had undergone surgical resection. We then determined that MICAL2 promotes PDAC growth and metastasis and that it is associated with patterns of gene expression associated with KRAS. We have identified a MICAL2/MRTF-B/SRF transcriptional program that underpins these malignant hallmarks. Our findings support the hypothesis that MICAL2 drives pancreatic cancer progression and that as an enzyme, it is worthy of further study as a potential therapeutic target for PDAC.

Materials and Methods

Human specimens

Normal and tumor pancreatic tissues were obtained from patients undergoing surgical resection or tissue biopsy at the University of California San Diego Health. All tissue donations and experiments were reviewed and approved by the Institutional Review Board of the University of California San Diego Health. Written informed consent was obtained prior to the acquisition of tissue from all patients. The studies were conducted in accordance with recognized ethical guidelines (Declaration of Helsinki). Samples were confirmed to be tumor or normal based on pathologist assessment (Supplementary Table S1).

Animals and in vivo procedures

All mouse protocols were reviewed and approved by the Institutional Animal Care and Use Committee of the University of California, San Diego. Five- to eight-week-old age-matched and sex-balanced male and female F1 hybrid and nude mice were purchased from The Jackson Laboratory and used for each experiment. We used murine and human cell lines such as KPC46, AsPC1, and BxPc3 for in vivo experiments. KPC46 cells were injected in F1 hybrid mice, and AsPC1 and BxPc3 cells were injected in NOD/SCID gamma (NSG) mice (JAX). For orthotopic and subcutaneous tumor challenges, mice were administered intrapancreatic and flank injections of 50 k KPC46, 500 k AsPC1, and BxPc3 cells suspended in Matrigel. For the liver metastases model, we injected 50 k KPC46 and 500 k BxPC3 cells mixed in PBS via the intrasplenic route. Additionally, 150,000 KPC46-inducible shControl or shMICAL2 cells were injected into the spleens of F1 hybrid mice. After 48 hours, all mice were started on doxycycline (dox) chow and continuously received dox chow for 20 days until euthanasia. At the time of euthanasia, liver weights were measured. Liver and spleen specimens were paraffin fixed and underwent hematoxylin and eosin staining to observe metastases. Liver metastases were quantified using QuPath. Animals were euthanized 3 weeks after implantation for tumor weight and volume analyses. For the subcutaneous tumor model, we injected 125,000 inducible shControl or shMICAL2 KPC 46 cells in F1 hybrid mice and enrolled them for dox chow when tumors reached 3 to 5 mm in diameter and saced them 44 days later.

Cell lines

The KPC46 cell line was developed in our laboratory from LSL-Kras^G12D/+; LSL-Trp53^R172H/+; and Pdx-1-Cre (KPC) mice, which develop spontaneous tumors (14). AsPC1 and BxPc3 cells were obtained from the ATCC. All cell lines were maintained in RPMI 1640 (Sigma-Aldrich, R8758) with 10% FBS and penicillin/streptomycin at 37°C with 5% CO_2, and all short hairpin RNA (shRNA) lines were maintained in RPMI 1640 (Sigma-Aldrich, R8758) with 10% FBS, penicillin/streptomycin, and 5 µg puromycin at 37°C with 5% CO₂. We routinely checked for Mycoplasma contamination every 2 weeks. siRNAs and scramble control shRNA against MICAL2, MRTFA, and MRTFB were obtained from Dharmacon Horizon Discovery, and human and murine cell lines were infected using siRNA and shRNA using the protocol provided by the manufacturer. shRNA and siRNA sequences used in this study are summarized in Supplementary Tables S2 and S3.

To establish BxPc3 OE and empty vector lines, the coding sequences of MICAL2 (ENSG00000133816) and EGFP were cloned and inserted into the CSII-CMV-MCS-IRES2-Bsd vector, which was generously provided by the RIKEN BioResource Research Center through the National BioResource Project of the Ministry for Education, Culture Sports, Science and Technology (MEXT), Japan (catalog number RDB04377; ref. 15). Following vector construction, the plasmid containing the lentiviral genome was transfected into HEK 293T cells along with helper plasmids, pCMV-VSV-G and pCAG-HIVgp, encoding viral envelope and packaging proteins. The viral supernatant was collected 2 days after transfection and concentrated using Lenti-X lentivirus concentrator (Clontech). Cells were then infected with the concentrated lentivirus, and positive clones were selected by culturing cells in a medium containing 10 μg/mL blasticidin. The sequences of oligonucleotides used for coding sequence cloning can be found in Supplementary Table S4. MICAL2 knockdown (KD) and OE efficiencies were evaluated by qRT-PCR.

qRT-PCR

Total RNA from cell lines was extracted as per the manufacturer’s instructions (RNeasy Mini Kit, Qiagen). Total RNA (1 μg) was reverse transcribed using QuantiTect Reverse Transcription Kit (Qiagen). Subsequently, specific transcripts were amplified by SYBR Green PCR Master Mix (USB) using a Bio-Rad CFX96 Real-Time System thermocycler. For cases in which fold expression is specified, the comparative CT method was used to quantify gene expression. Expression was normalized to 18S. Primers used for qPCR are detailed in Supplementary Table S5.

Western blotting

Western blotting was conducted following established protocols with slight adjustments. Initially, cell pellets were lysed in 2% SDS buffer supplemented with cOmplete protease (Roche, 11697498001) and PhosSTOP phosphatase (Roche, 4906845001) inhibitors after thorough washing with Dulbecco PBS. Subsequently, cell lysates were prepared using RIPA buffer, and protein concentrations were determined using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23225). The proteins were denatured by adding NuPAGE LDS Sample Buffer (4×; Thermo Fisher Scientific, NP0008) and heating at 98°C for 10 minutes. For human and murine 2D cell lines, 10 µg of proteins were separated by SDS-PAGE and transferred onto 0.2-μm polyvinylidene difluoride membranes (Bio-Rad Laboratories, Inc., 1704156) using the Trans-Blot Turbo Transfer System (Bio-Rad Laboratories, Inc.). Following blocking with 5% skim milk or 5% BSA in TBS with 5% TBS containing 1% Tween 20 (TBST) for 1 hour at room temperature, membranes were incubated overnight at 4°C with the primary antibody diluted in the blocking solution. After washing three times with TBST, membranes were incubated with the corresponding secondary anti-mouse IgG (Thermo Fisher Scientific, 31430) or anti-rabbit IgG antibody (Thermo Fisher Scientific, 31460) in a blocking solution. Signals were visualized using Immobilon Western Chemiluminescent HRP substrate (MilliporeSigma, WBKLS0500) and captured using the ChemiDoc Imaging System (Bio-Rad Laboratories, Inc.). Data were analyzed using ImageJ and Image Lab software (16). The list of antibodies used for Western blotting is detailed in Supplementary Table S6.

Histology and IHC

Tissues were fixed in 10% neutral-buffered formalin, embedded in paraffin, and sectioned at a thickness of 5 μm. Standard hematoxylin and eosin staining procedures were used for histologic examination. For IHC, heat-induced antigen retrieval was achieved by immersing the tissue sections in citric acid buffer (pH 6.8) and subjecting them to pressure cooking for 20 minutes. Endogenous peroxidases were blocked using 3% H₂O₂ in TBS for 20 minutes at room temperature. To minimize nonspecific binding, tissue sections were incubated with a blocking solution containing 10% goat and 2.5% horse sera. Subsequently, the sections were exposed to the primary anti-MICAL2 antibody (13965-1-AP) overnight at 4°C. Following three washes with TBST, the sections were treated with a goat anti-rabbit IgG ImmPRESS secondary antibody for 30 minutes at room temperature. After additional washing steps with TBST, the signals were developed using ImmPACT AEC or DAB HRP Substrate. Finally, the slides were counterstained, dehydrated, and coverslipped according to standard protocols.

In vitro assays

Cell viability and proliferation rates were assessed using a CCK-8 proliferation kit. Initially, cells were seeded in six-well plates at a density of 500 k cells per well and transfected with specific siRNAs (siControl, siMical-2, siMRTF-A, and siMRTF-B). After 72 hours, the cells were split and seeded in 96-well plates at a density of 1k cells per well for KPC46 and 5 × 10³ cells per well for AsPC1 in 5% FBS (Omega Scientific). Following incubation for 3, 4, 5, 6, 7, and 8 days, 10 μL of CCK-8 reagent (Abcam, ab228554) was added to each well at each time point. Subsequently, the 96-well plate was incubated in the dark for 1 hour, and the absorbance was measured at 460 nm using a microplate reader (BioTek, Synergy HT). For the in vitro wound healing assay, human and mouse PDAC cell lines were cultured in six-well plates until 80% to 90% confluency was reached. A straight-line scratch was created across the cell monolayer using a 1,000-µL pipette tip. The cells were then rinsed three times with 1× PBS to eliminate unattached cells and debris before adding RPMI supplemented with 10% FBS. Following the acquisition of the initial image (0 hour), the plate was placed in an incubator at 37°C and 5% CO₂. Subsequent images were captured at various time points (12, 16, 21, 32, and 72 hours) using a phase-contrast microscope. For the cell-cycle assay, 1 million cells of shControl and shMICAL2 were seeded in a six-well plate. The following day, cultures were washed with PBS and harvested by trypsinization. The cells were then washed again with ice-cold PBS to remove excess media and fixed with 6 mL of 70% ice-cold ethanol for 24 hours at −20°C. After fixation, tubes were centrifuged at 2,000 RPM for 10 minutes, followed by washing with ice-cold PBS. The cells were suspended in 250 µL of TBST and treated with 50 µL of a 100 µg/mL stock of RNase and 200 µL of propidium iodide from a 50 µg/mL stock solution. Tubes were then incubated at 37°C in the dark for 15 minutes before analysis by FACS. For the SRF reporter assay, cells were transduced with Serum Response Factor (RhoA/ELK1/SRF) response element luciferase reporter lentiviral particles according to the manufacturer’s protocol (G and P Biosciences). Positive cells were selected using the antibiotic Zeocin, and luciferase reporter assay was performed using Dual-Glo luciferase assay kits (Promega) according to the manufacturer’s instructions. Firefly luminescence was normalized by the viability and number of cells. BxPC3 cells were used for active GTPase pulldown assay. This assay was done according to the manufacturer’s protocol (Active Ras Detection Kit, Cell Signaling Technology, #8821).

Gelatin degradation assay and immunofluorescence

The gelatin degradation assay was performed essentially as described previously (17). Briefly, 12 mm diameter glass coverslips were presoaked in 70% ethanol for 20 minutes before use, transferred to a 24-well plate (Corning), and incubated at room temperature with 500 μL poly-L-lysine (Sigma, P8920): PBS (Wisent) at a 1:1 ratio for 20 minutes. After incubation and washing thrice with 500 μL PBS, 500 μL of freshly prepared 0.5% glutaraldehyde in PBS was added to the wells and incubated for 20 minutes at room temperature to activate the poly-L-lysine surface for further protein binding. The coverslips were washed thrice with PBS and coated with a mixture of prewarmed (37°C) 0.2% gelatin and FITC-gelatin (Thermo Fisher Scientific, G13187) at a 6:1 ratio. Afterward, the coverslips were washed and incubated with a fresh solution of 5 mg/mL sodium borohydride (Sigma, 452882) for 20 minutes to quench free aldehydes. After extensive washing, the coverslips were incubated in 24-well plates containing 70% ethanol and stored at 4°C in the dark until used. The ethanol was removed, and the coverslips were washed thrice with PBS before use. The coverslips were mounted with the ProLong gold antifade reagent (Thermo Fisher Scientific) before image acquisition.

Confocal microscopy and image analysis

Images were acquired using the Zeiss LSM780 laser scanning confocal microscope equipped with ×63 oil objective. Five representative images were captured per cell type. Gelatin degradation appeared as punctate or diffuse “dark clearings” (depending on the incubation time) in the bright fluorescent gelatin because of loss of fluorescence signal. Gelatin matrix degradation was quantified using ImageJ. Data processing was performed using GraphPad Prism 6, and statistical analyses were performed using one-way ANOVA.

Macropinocytosis assay and quantification

Macropinocytosis assay and quantification were performed as previously described (18). Briefly, cells transfected with a scrambled siRNA (siCTRL) or siRNA targeting MICAL2 (siMICAL2) were seeded on acid-washed glass coverslips in 24-well plates and subjected to serum and glutamine starvation for 24 hours. Media were then removed from the wells, the same media containing 1 mg/mL FITC-dextran (Thermo Fisher Scientific) were added back to the cells, and plates were incubated for 30 minutes at 37°C. Thereafter, 4 to 5 washes with cold PBS were done, and cells were fixed in 3.7% formaldehyde for 15 minutes. Then the nuclei were stained with 2 μg/mL of 4′,6-diamidino-2-phenylindole (DAPI; MilliPore), and coverslips were mounted on glass slides with Dako fluorescent media (Agilent Technologies). Images were automatically captured at ×40 magnification using a BioTek Cytation 5 system (Agilent Technologies) and were subjected to automated analysis using the BioTek Gen5 software to calculate the relative macropinocytic index.

RNA sequencing

AsPC1 shControl and shMICAL2 cells were cultured in 10-cm dishes in triplicates. Total RNA was extracted using TRIzol following the manufacturer’s instructions. RNA samples were submitted to Azenta Life Sciences for further analysis. Quality testing was carried out by measuring RNA integrity and optical density (OD) readings (260/280 and 260/230) to have RNA integrity number (RIN) scores of 9.4 or higher, and OD readings were within the 1.8 to 2.2 range. Differential expression analyses were analyzed by gene set enrichment analysis (GSEA) for Hallmarks.

H3K27ac ChIP-seq

Frozen tissue samples were homogenized using the Covaris CP02 cryoPREP Automated Dry Pulverizer. Homogenized tissue was fixed by rotating at room temperature for 8 minutes in 1% formaldehyde (Thermo Fisher Scientific, cat. #28906) and then diluted in PBS. Fixation was quenched with 1/20th volume of 2.5 mol/L glycine. All buffers after fixation and prior to wash buffer before elution included HALT protease inhibitor (Life Technologies, cat. #78438) and 1 mmol/L sodium butyrate to inhibit histone deacetylase activity.

Samples were pelleted and washed twice with cold PBS. Samples were then lysed by rotating at 4°C with ChIP Lysis Buffer 1 (Boston BioProducts, cat. #CHP-126) for 10 minutes and then with ChIP Lysis Buffer 2 (Boston BioProducts, cat. #CHP-127). Samples were then washed with sonication buffer (CHP-133). Samples were resuspended in sonication buffer and sonicated to an average size of 100 to 500 bp on the Covaris E220 focused ultrasonicator in a volume of 750 µL sonication buffer. Following sonication, samples were collected and centrifuged at 20,000 × g for 10 minutes at 4°C to pellet insoluble material. The supernatant was collected and diluted in one volume of ChIP Dilution Buffer (Boston BioProducts, #CHP-143), and input controls were collected.

Chromatin was immunoprecipitated overnight (12–16 hours) by rotating at 4°C with anti-H3K27ac antibody (Abcam, cat. #4729) conjugated to protein G Dynabeads (Life Technologies, cat. #10004D). The following day, immunoprecipitated chromatin was washed twice with low-salt wash buffer (Boston BioProducts, #CHP-146), once with high-salt wash buffer (Boston BioProducts, #CHP-147), once with LiCl wash buffer (Boston BioProducts, #CHP-149), and once with Tris-based wash buffer (Boston BioProducts, #CHP-148). Chromatin was eluted with ChIP Elution Buffer (Boston BioProducts, #CHP-153) for 1 hour at 65°C with agitation at regular intervals. Eluted chromatin was reverse cross-linked for 16 hours at 65°C. Chromatin was then diluted in one volume of TE buffer and then treated with RNase (Thermo Fisher Scientific, cat. #AM2286) for 2 hours at 37°C and proteinase K (Thermo Fisher Scientific, cat. #AM2548) for 1 hour at 55°C. An equal volume of phenol/chloroform (Thermo Fisher Scientific, cat. #15593031) was added to each sample, and the aqueous phase was collected in a Phase Lock Gel tube (VWR, cat. #10847-802). Each sample received 1 mL ethanol, 20 µL 5 mol/L NaCl, and 30 µg GlycoBlue (Thermo Fisher Scientific, cat. #AM9516), which were then mixed by inversion and stored at −20°C overnight. The following day, the samples were centrifuged at 20,000 × g for 20 minutes to precipitate the purified DNA. Pellets were washed with 800 µL cold 80% ethanol, and a second centrifugation was performed at 20,000 × g for 10 minutes. The pellets were briefly air-dried and resuspended in 50 µL water.

H3K27ac sequencing

Libraries were generated using Swift Accel-NGS 2S Plus DNA Library Kit (Swift Biosciences, cat. #21024). Samples were multiplexed six libraries per lane on an Illumina HiSeq 2500 sequencer.

Differential H3K27ac enrichment analysis

H3K27ac ChIP-seq reads were aligned to the hg19 human reference genome with bowtie2 (v2.0.5) using the option “—sensitive” (19). Regions enriched for H3K27ac were called with MACS2 (v2.0.10; ref. 20) using input control data as background. To calculate differences in the H3K27ac signal between PDAC samples and normal pancreas samples, we performed the following analysis steps: (i) peaks with a MACS2 score <20 were removed, and the remaining peaks were merged across samples to generate a universal peak set; (ii) ChIP-seq reads and input control reads overlapping each peak were tallied for each sample; (iii) after applying a normalization factor generated using normalization of ChIP-seq (NCIS; ref. 21), input control reads were subtracted from ChIP-seq reads for each peak and sample; (iv) the resulting peaks were further filtered to remove those that did not have at least 20 normalized reads in at least two samples; (v) batch effects were estimated and removed with a negative binomial generalized linear model via DESeq2; (vi) normalized read counts across technical replicates were averaged (mean) and rounded to the nearest integer; and (vii) these values were then used to calculate differences between PDAC and normal pancreas using DESeq2. The data are available in Supplementary Table S7 (22).

H3K27ac gene set analyses

To determine whether differential H3K27ac regions were enriched near genes belonging to specific biological processes or modes of regulation, we followed the framework described by McLean and colleagues (23). Briefly, we performed the following analysis steps: (i) regulatory domains were defined using GENCODE v19 protein-coding annotations with level 1 or 2 support. (ii) Test peak sets were defined as peaks with an adjusted P value ≤0.01 and a log₂ (fold change) of either >0 or <0. The background peak set was defined as any region tested by DESeq2 as described above. (iii) The width of each peak was adjusted to 1,000 bp by extending the boundaries ± 500 bp from the peak’s midpoint. (iv) The enrichment of the test peak sets in the regulatory domains of a given gene was calculated using “phyper” in R. (v) Multiple hypothesis testing correction was performed by the Benjamini–Hochberg method with “P.adjust.”

Differential SE analysis

To call SEs, we followed previously described methods (3, 24). Briefly, we performed the following analysis steps: (i) MACS2 peaks were filtered using a P value threshold of ≤1 × 10⁻⁹; (ii) MACS2 peaks located within 12.5 kb of one another were merged to yield a set of “stitched” regions; (iii) stitched regions were merged across samples to generate a universal region set; (iv) ChIP-seq reads and input control reads overlapping each region were tallied for each sample; (v) after applying a normalization factor generated using NCIS (21), input control reads were subtracted from ChIP-seq reads for each region and sample; vi) a negative binomial distribution was fit to these normalized region counts for each sample; and vii) regions were called SEs for a given sample if their counts were ≥ the 97.5th percentile of the abovementioned estimated distribution. To determine which SEs had differential H3K27ac enrichment in PDAC compared with the normal pancreas, we performed the following analysis steps: (i) a universal set of SEs was defined as the union of regions called SEs in four of six normal pancreas samples and five of seven PDAC samples; (ii) using the quantifications for these SEs, batch effects were estimated and removed with a negative binomial generalized linear model via DESeq2; (iii) normalized read counts across technical replicates were averaged (mean) and rounded to the nearest integer; and (iv) these values were then used to calculate differences between PDAC and the normal pancreas using DESeq2 (22). Library sizes were determined using the full-stitched region set.

H3K27ac enrichment visualization

To visualize the input-normalized H3K27ac signal across the genome, we generated bigwig files for each sample using “bamCompare” from deepTools2 (25). More specifically, we used the options “—operation subtract,” “—scaleFactors,” “—binSize 10,” “—smoothLength 30,” “—normalize,” “using none,” and “—extendReads.” To determine scaling factors, we used NCIS (21) and further adjusted these factors by the number of mapped reads (in millions) of the ChIP-seq sample. We estimated the fragment size to extend the reads using “phantompeakqualtools” (26). A signal was averaged (mean) for technical replicates of the same biological sample.

Statistical analysis

GraphPad Prism version 9 was used for the graphical representation of data. The results are expressed as the mean and SEM unless otherwise indicated in the figure legend. For the tumor growth studies, the data shown represent independent experiments with biological replicates. A Student two-tailed unpaired t test was used for the experiments with two groups unless otherwise indicated. For multiple group comparisons, a one-way or two-way ANOVA was performed followed by a Tukey test unless otherwise indicated in the figure legend. Data were considered significant if P < 0.05; exact values are shown in each figure legend.

Data availability

H3K27ac ChIP-seq data generated in this study are publicly available in the Gene Expression Omnibus database (GSE277337). Raw RNA sequencing (RNA-seq) data for this study were generated by Azenta Life Sciences. Derived RNA-seq read counts supporting the findings of this study are available in Supplementary Table S8. The MICAL2 expression survival data were obtained from The Cancer Genome Atlas data hosted on the Protein Atlas (Expression of MICAL2 in pancreatic cancer - The Human Protein Atlas). All other data are available upon request from the corresponding authors.

Results

MICAL2 is a putative SE-associated gene in human PDAC

We sought to identify tumor-specific active SE regions in human PDAC. We obtained tumors (n = 7) from surgical resections of PDAC and normal pancreatic tissues (n = 6) from patients undergoing resection for nonadenocarcinoma histology (Supplementary Table S1). Bulk tissues were lysed, and we performed H3K27ac ChIP-seq to identify regions of active transcription in human PDAC. We identified SE regions and SE-associated genes computationally for both normal and tumor samples (Fig. 1A). Top enriched tumor SE-associated genes included DAD1 and MBP (associated with perineural invasion); LIMK2, MSN, and CAPZB (associated with cytoskeleton dynamics); and FCGR3A, FCGR2A, and RUNX1 (associated with immune regulation). Normal pancreas tissue was enriched for SE-associated genes such as CEL and PLA2G1B (associated with pancreatic exocrine function). We then performed hierarchical clustering of the SE-associated genes that were enriched in tumor and normal tissues (Supplementary Fig. S1A; Supplementary Table S7). We used the Genomic Regions Enrichment of Annotations Tool to determine which pathways were enriched in tumor versus normal adjacent tissues. Overall, we found that immune, KRAS, and EMT pathways were upregulated in tumor samples, whereas pancreatic endocrine genes and metabolic pathways were correlated with normal tissues (Supplementary Fig. S1B). Among the tumor-specific putative SE-associated genes, we found the leukemia inhibitory factor [log₂-fold change (FC), 1.20; adjusted P value, 6.20 × 10⁻⁵], which was previously identified as a potential therapeutic target in PDAC and is a target of ongoing clinical trials (27).

Figure 1.

H3K27ac ChIP-seq in human PDAC identifies MICAL2 as an SE-associated gene. A, Differentially expressed SE-associated genes in tumor compared with normal tissue. The top 10 upregulated and downregulated genes as well as the MICAL2 gene are annotated. B, qPCR of MICAL2 using RNA extracted from the same patient samples used for ChIP-seq. Two-tailed t test P value is shown. C, H3K27ac ChIP-seq occupancy upstream and within the MICAL2 loci in normal and tumor samples. Norm, normal.

We next sought to identify genes that were upregulated in PDAC and that encoded proteins with the potential to be targeted by small molecules or antibody-based therapeutics. We chose to focus on MICAL2, a member of the molecules interacting with CasL family, which indirectly promotes SRF transcription. MICAL2 contains several functional protein domains, including its flavin adenine dinucleotide enzymatic domain. MICAL2 was significantly enriched (log₂ FC, 1.07; adjusted P value, 5.39 × 10⁻⁴) in tumor samples compared with normal samples (Fig. 1A). Furthermore, qPCR analysis revealed that MICAL2 mRNA was enriched (P value 0.009) in tumor samples, suggesting that the increased H3K27ac signal at the promoter and coding regions leads to increased transcriptional activation of the MICAL2 locus (Fig. 1B and C). We next assessed the enrichment of transcription factor–binding motifs using GSEA, and we found significant enrichment of development and inflammation-promoting transcription factor motifs such as HOX, NFKB, and STAT1 (Supplementary Fig. S1C). Interestingly, SRF motifs were the most prevalent in tumor samples. Transcription factor motifs associated with normal pancreatic development such as PAX were decreased in tumor tissues. We further interrogated the MICAL2 epigenetic regulation in available human PDAC datasets. We found that the H3K27ac peaks at the MICAL2 locus aligned with published assay for transposase-accessible chromatin (ATAC) peaks from EPCAM-sorted epithelial cells (Supplementary Fig. S2A; ref. 28). We then leveraged single-cell multiomic analyses (ATAC and RNA) of human PDAC to show that epithelial clusters demonstrate ATAC peaks that match our H3K27ac peaks (Supplementary Fig. S2B; ref. 29). These datasets together show putative SE regions upstream of the MICAL2 locus in epithelial cells.

Investigation of The Cancer Genome Atlas solid cancers datasets revealed that PDAC is the fourth highest expressor of MICAL2 (Supplementary Fig. S3A). Furthermore, using the TNMplot dataset (30), we determined that MICAL2 expression was significantly enriched in PDAC primary tumors (2.52 FC; Dunn test P value, 4.60 × 10⁻²¹) and metastases (1.56 FC; Dunn test P value, 2.58 × 10⁻³) compared with normal pancreatic tissues (Supplementary Fig. S3B). To ascertain whether the higher level of MICAL2 transcription resulted in greater MICAL2 protein levels within PDAC tumors, we performed IHC on the human normal pancreas and PDAC tissues. We found a striking tumor-specific increase in MICAL2 staining across four patient tumors (Fig. 2A). We next observed that high-MICAL2 protein levels and tumor specificity are preserved in the commonly used KPC genetically engineered mouse model (Fig. 2B; ref. 31). In addition, we found that MICAL2 was significantly overexpressed in murine KC and KPC organoids (Supplementary Fig. S3C; ref. 32). We then investigated the expression of MICAL2 across commonly used PDAC cell lines and found that most cell lines express MICAL2; ASPC1 is a high-MICAL2 expressor, whereas BxPc3 and MIAPaCa2 express low to no MICAL2, likely recapitulating the heterogeneity of PDAC (Supplementary Fig. S3D). Finally, we assessed three KPC cell lines and found that all expressed similar levels of MICAL2 (Supplementary Fig. S3E).

Figure 2.

MICAL2 is overexpressed in human and mouse PDAC. A, IHC for MICAL2 in normal pancreas and PDAC human tissues. B, IHC for MICAL2 in normal pancreas and KPC-derived PDAC mouse tissues. C and D, Survival analysis of patients with PDAC segregated by MICAL2 expression (high vs. low) in two datasets: PanCuRx (C) and COMPASS (D). panc, pancreas.

We investigated the association of MICAL2 expression with outcomes in patients with PDAC who were eligible for surgical resection (PanCuRx; ref. 33) and in patients with advanced disease (COMPASS; ref. 34). We found that high MICAL2 expression correlated with worse outcomes in the surgical cohort but was not prognostic in the advanced cohort (Fig. 2C and D). This suggested a possible role for MICAL2 in the progression from primary to advanced disease. Overall, we found that PDAC has a distinct landscape of SE-associated genes that are linked to known PDAC biology. We found MICAL2 among these genes and determined that MICAL2 transcription is higher in PDAC compared with normal adjacent tissues both in our study as well as in other independent datasets and that this is associated with increased expression at the protein level. Importantly, high MICAL2 expression is associated with a poorer prognosis in patients whose tumors were surgically removed, indicating that expressing high MICAL2 may mark tumors with increased recurrence and progression.

MICAL2 expression is associated with KRAS and EMT signaling pathways

As MICAL2 is known to canonically regulate MRTF/SRF activity, we first sought to determine whether this was occurring in the setting of PDAC. We checked the expression of common SRF target genes by qPCR following constitutive shRNA KD of MICAL2 in ASPC1 cells and found that many of the MRTF/SRF target genes were downregulated, as we expected (Supplementary Fig. S4A). RhoA expression was dramatically reduced in MICAL2-KD cells, suggesting that MICAL2 may regulate a key promoter and target of SRF signaling. Notably, constitutive shRNA KD of MICAL2 leads to a decrease in MRTF-A and MRTF-B expression, suggesting that prolonged loss of MICAL2 activity depresses SRF transcription (Supplementary Fig. S4A). Interestingly, in ASPC1 and KPC46 cells, siRNA KD of MICAL2 led to small increases in MRTF-A and MRTF-B expression, which may reflect acute compensation (Supplementary Fig. S4B and S4C). To further understand the impact of MICAL2 on pancreatic cancer cell biology, we performed RNA-seq. Using differential gene expression analysis comparing siRNA targeting MICAL2 and scramble control, we found, as expected, that MICAL2 was the most significantly repressed gene (Supplementary Fig. S4D; Supplementary Table S8). To investigate pathways likely to be regulated by MICAL2, we performed GSEA. Interestingly, we found that KRAS signaling pathways were dramatically reduced in ASPC1 cells lacking MICAL2 (Fig. 3A; Supplementary Fig. S4E). Additional prosurvival pathways such as TNFα and HIF1α signaling were lost in the MICAL2-KD cells, suggesting that MICAL2 may act as a proto-oncogene in PDAC. Importantly, we found that EMT signaling was also significantly reduced in MICAL2-KD cells (Fig. 3A). We leveraged a recently published study from the Der laboratory that comprehensively analyzed KRAS-dependent transcriptional programs and found that upon silencing of MICAL2, KRAS-regulated genes were downregulated, further suggesting that MICAL2 modulates KRAS signaling (Fig. 3B; ref. 35). As an orthogonal pathway analysis method, we identified the cell states, elucidated as transcriptional activity downstream of KRAS activation via the OncoGPS methodology (36), that were associated with MICAL2 expression in other tumor models of the Cancer Cell Line Encyclopedia (37). Similar to GSEA, we found that the EMT state was associated with higher MICAL2 gene expression in the Cancer Cell Line Encyclopedia (Fig. 3C). To further investigate the EMT phenotype, we used a murine PDAC organoid line derived from a KPC liver metastasis, KPC46, which has mesenchymal features and a high propensity for metastasis (Fig. 3D). In MICAL2-KD cells, we observed mesenchymal to epithelial changes, marked by highly uniform, polarized, compact epithelial cell structures that express increased E-cadherin (Fig. 3E). Overall, these experiments revealed that MICAL2 drives MRTF/SRF transcription, EMT, and pro-oncogenic pathways in both human and mouse models of PDAC.

Figure 3.

MICAL2 promotes a KRAS and EMT phenotype in PDAC cells. A, GSEA enrichment plots in AsPC1 cells with MICAL2-KD compared with scramble control indicating normalized enrichment score (NES), adjusted P value, and FDR. B, Enrichment analysis of the Klomp and colleagues (35) KRAS upregulated (KRAS-UP) and KRAS downregulated (KRAS-DOWN) genes in ASPC1 MICAL2-KD compared with scramble control. P.adj <0.01. C, OncoGPS plot of Cancer Cell Line Encyclopedia cell lines showing MICAL2 expression. Red dots, cell lines with high MICAL2 expression; blue dots, low expression. D, Hematoxylin and eosin stain of mouse PDAC organoids derived from control KPC46 cells (SCR) and MICAL2-KD. E, Immunoblots of organoids shown in D. Quantification is shown with two-tailed t test P value. E-cad, E-cadherin; Vinc, vinculin. F and G, Immunoblot analysis of AsPC1 (F) and KPC46 (G) cells treated with SCR, MICAL2, MRTF-A, and MRTF-B siRNAs at 72 hours. Quantification is shown with ANOVA P values. H and I, Immunoblot analysis of BxPc3 cells expressing empty vector (EV) or MICAL2-OE vector at 72 hours for AKT and ERK signaling (H) and KRAS–GTP loading (I). Total KRAS and GAPDH were loaded at 2% input. Quantification is shown with two-tailed t test P value. J, Representative immunofluorescent images of AsPC1 cells transfected with siRNA control (SCR) or MICAL2-KD. DAPI, blue, cell nuclei. FITC-conjugated dextran (green) was used to label macropinosomes. Quantitation of the relative macropinosome index with two-tailed t test P value.

MICAL2 promotes KRAS signaling

Our transcriptomic analyses indicated that KRAS signaling is altered in MICAL2-deficient PDAC cells. Therefore, we evaluated the activation of KRAS effector pathways PI3K/AKT and MAPK/ERK signaling in cells with loss and gain of function of MICAL2. As MICAL2 loss leads to altered expression of the SRF coactivators MRTF-A and MRTF-B, we also investigated the effect of KD of these two genes. We generated human and mouse models silencing MICAL2, MRTF-A, and MRTF-B (Supplementary Fig. S5A and S5B). For gain-of-function studies, we overexpressed MICAL2 in BxPc3 human PDAC cells, as BxPc3 cells do not endogenously express MICAL2 (Supplementary Figs. S3D and S5C). In the human PDAC cell line, AsPC1, loss of MICAL2 led to a marked decrease in pAKT and no change in pERK1/2 (Fig. 3F; Supplementary Fig. S5D). The negative cell-cycle regulator P27 was dramatically increased in MICAL2-KD cells, suggesting a possible mechanism by which MICAL2 could modulate cell proliferation. The siRNA-mediated silencing of MRTF-B phenocopied the pAKT decrease observed in MICAL2-silenced cells. Notably, MICAL2 KD led to a partial loss of MRTF-B RNA and protein (Supplementary Fig. S5A and S5D). Interestingly, the loss of MRTF-A did not recapitulate this phenotype, and the loss of function of MRTFs did not phenocopy the P27 increase. We next evaluated the loss of MICAL2 and MRTFs in KPC46 cells. We found that mouse PDAC cells recapitulated the decrease of pAKT and pERK1/2 in both MICAL2-silenced and MRTF-B–silenced cells; however, P27 was unchanged in this cell line (Fig. 3G; Supplementary Fig. S5E). These results demonstrate that MICAL2-deficient and MRTF-B–deficient cells have reduced phosphorylation of PI3K and MAPK consistent with reduced KRAS signaling. When we examined BxPc3 cells modified to express MICAL2, we found that pAKT and P27 expressions were significantly increased (Fig. 3H; Supplementary Fig. S5F). To directly determine whether KRAS signaling was altered by MICAL2 expression, we characterized KRAS–GTP loading after MICAL2 OE in BxPc3 cells. We found a significant increase in the levels of KRAS–GTP loading in these cells, thus directly connecting MICAL2 expression to KRAS activity (Fig. 3I).

To further assess the impact of MICAL2 on KRAS signaling, we examined the effects of MICAL2 KD on macropinocytosis, an extracellular nutrient-scavenging process driven by KRAS. We observed a dramatic reduction in macropinocytosis in MICAL2-deficient cells that is also consistent with a decrease in KRAS signaling activity (Fig. 3J). To summarize, loss- and gain-of-function studies revealed that MICAL2 promotes signaling through KRAS, as shown by KRAS–GTP binding, phosphorylation of MAPK and PI3K, and macropinocytosis. Although there is a model-to-model difference, these observed biochemical results were consistent with our pathway analyses of the cellular transcriptome.

MICAL2 promotes PDAC cell proliferation and migration

We next sought to investigate how MICAL2 loss and gain of function would impact oncogenic phenotypes in PDAC cells. As MICAL2 seemed to drive EMT, we first measured the motility of MICAL2-, MRTF-A-, and MRTF-B–deficient cells. We found that migration in both ASPC1 and KPC46 was reduced after MICAL2 KD (Fig. 4A and B; Supplementary Fig. S6A and S6B). We also found that MRTF-A and MRTF-B KD decreased migration in the human ASPC1 cells, whereas in mouse, KPC46 only MRTF-A KD resulted in reduced migration. Conversely, in a gain-of-function experiment, BxPc3 cells overexpressing MICAL2 had increased migration compared with empty vector control cells (Fig. 4C; Supplementary Fig. S6C). Using an invasion assay, we further determined that KPC46 cells with loss of MICAL2 were less capable of invading a gelatin matrix (Supplementary Fig. S6D).

Figure 4.

MICAL2 drives cancer cell migration and proliferation in vitro. A and B, Quantification of wound healing assay of AsPC1 (A) and KPC46 (B) transfected with SCR, MICAL2, MRTF-A, and MRTF-B siRNAs at the time points indicated. C, Quantification of wound healing assay of BxPc3 cells expressing empty vector (EV) or MICAL2-OE vector at the time points indicated. D and E, Proliferation assay of AsPC1 (D) and KPC46 (E) transfected with SCR, MICAL2, MRTF-A, and MRTF-B siRNAs at the time points indicated. F, Proliferation assay of BxPc3 cells expressing empty vector or MICAL2-OE vector at the time points indicated. G–I, Cell-cycle analysis of AsPC1 (G) and KPC46 (H) transfected with SCR, MICAL2, MRTF-A, and MRTF-B siRNAs at 72 hours, and BxPc3 (I) cells expressing empty vector or MICAL2-OE vector. ANOVA was used for statistical analysis.

We then assessed the impact of MICAL2 on cell proliferation. We found that MICAL2 KD led to a significant decrease in proliferation in ASPC1 and KPC46 cells, whereas OE of MICAL2 in BxPc3 led to an increase in proliferation rate compared with control cells (Fig. 4D–F). Interestingly, KD of either MRTF in the human or mouse PDAC cells led to only a partial decrease in the cell proliferation rate (Fig. 4D and E). To better understand which part of the cell cycle was affected by MICAL2 expression, we used flow cytometry to examine cell-cycle progression. In both AsPC1 and KPC46, MICAL2- and MRTF-B–deficient cells had a significant shift toward arrest in the G₀–G₁ phase and concomitant reduction in the proportion of cells in the S phase and G₂$\mathit{–}$M phase (Fig. 4G and H). MRTF-A–depleted KPC46 cells had no change in their cell-cycle profile, but AsPC1 cells lacking MRTF-A surprisingly had a block in the S phase rather than in the G₀–G₁ phase, hinting at differences between models and possibly between human and mouse PDAC cells. Conversely, BxPc3 MICAL2-OE cells progressed faster through the G₀–G₁ phase and had an increased S phase proportion compared with the control, indicating that the increase in MICAL2 expression was sufficient to increase cell division (Fig. 4I). Overall, these experiments show that MICAL2 and MRTF-A/B expression promote cell migration, invasion, and proliferation. Furthermore, these findings are consistent with the RNA-seq and biochemical results we observed after the genomic KD of MICAL2 in PDAC cells.

MICAL2 and MRTF-B promote heterotopic and orthotopic growth in vivo

When we silenced MICAL2 in vitro, we observed reduced cell proliferation, migration, invasion, and a reversal of the EMT phenotype. Therefore, we next sought to determine how the loss of MICAL2 and MRTFs impact tumorigenesis initially using heterotopic subcutaneous mouse transplant models. We first transplanted AsPC1 cells with constitutive KD of MICAL2 or MRTF-A/B into immunodeficient NSG mice. We observed a dramatic decrease in tumor size when MICAL2 and MRTF-B were lost but no significant differences after MRTF-A silencing (Fig. 5A). Transplantation of KPC46 cells in syngeneic mice recapitulated the results observed with human PDAC cells, although with an even more profound reduction in tumor formation (Fig. 5B). In one experiment, no tumors formed from the MICAL-KD cells, whereas in a repeat experiment, 1/6 tumors did not form, and the remaining tumors were markedly inhibited in growth relative to those in the control. Conversely, BxPc3 MICAL2-OE cells grew larger than control cells when injected into the flank of NSG animals (Fig. 5C). To determine whether the reduced growth of MICAL2-KD cells was due to failure of engraftment, we established dox-inducible KPC46 cells with shRNA targeting MICAL2 (ish-MICAL2) or control (ish-control). We implanted the cells into the flank of syngeneic animals and let tumors form over 10 days prior to dox induction as documented by both palpation and ultrasound. Tumor growth over 44 days was abrogated in the ish-MICAL2–implanted mice compared with ish-control mice when treated with dox but not when kept on a regular dox-free diet (Fig. 5D and E).

Figure 5.

MICAL2 and MRTF-B promote tumor growth. A, Representative images and weight quantification with ANOVA testing of subcutaneous AsPC1 tumors grown in immunocompromised mice. AsPC1 cells express shRNA vectors to silence MICAL2, MRTF-A, and MRTF-B. B, Representative images and weight quantification with ANOVA testing of subcutaneous KPC46 tumors grown in syngeneic mice. KPC46 cells express shRNA vectors to silence MICAL2, MRTF-A, and MRTF-B. C, Representative images and weight quantification with ANOVA testing of subcutaneous BxPc3 tumors grown in immunocompromised mice. BxPc3 parental (PT) cells express empty vectors (EV) or MICAL2-OE vectors. D, Tumor growth over time of subcutaneous KPC46 tumors grown in syngeneic mice. KPC46 cells express dox-inducible shRNA vectors to silence MICAL2. E, Quantification and t test of tumor sizes from D. Tumors grown with and without dox chow (regular diet control) are shown. F, Representative images and weight quantification with ANOVA testing of orthotopic AsPC1 tumors grown in immunocompromised mice. AsPC1 cells express shRNA vectors to silence MICAL2, MRTF-A, and MRTF-B. G, Representative images and weight quantification with ANOVA testing of orthotopic KPC46 tumors grown in syngeneic mice. KPC46 cells express shRNA vectors to silence MICAL2, MRTF-A, and MRTF-B. RegDiet, regular diet.

We investigated how MICAL2 and MRTFs impact orthotopic tumor growth by implanting cells into the pancreatic tail. We found that only the mice implanted with AsPC1 MICAL2-KD cells, not the MRTF-KD cells, had a significantly decreased tumor burden, whereas in the KPC46 model, both the MICAL2- and MRTF-B–silenced cells had reduced in vivo growth compared with scramble control (Fig. 5F and G). MRTF-A again had no impact on tumor growth. To evaluate whether MRTF-B could compensate for the loss of MICAL2, we used SRF reporter and flow cytometry to assess the cell-cycle state in KPC46 MICAL2-KD cells with and without the OE of MRTF-B. We found that, as expected, SRF signaling was downregulated in the KPC46 MICAL2-KD cells, and we observed a decrease in cells in the S phase, but each of these effects were rescued by the OE of MRTF-B, demonstrating that MICAL2-dependent SRF signaling in PDAC is likely driven by modulating the levels of MRTF-B (Supplementary Fig. S7A and S7B). In summary, the expression of MICAL2 and MRTF-B in PDAC cells promoted tumor growth in both heterotopic and orthotopic locations whereas MRTF-A expression did not affect tumor growth. These results are consistent with our in vitro findings that MICAL2 promotes PDAC cell proliferation and suggest that MRTF-A and MRTF-B play distinct roles in promoting pancreatic tumor growth.

MICAL2 promotes metastasis in mice

To determine how MICAL2 and MRTFs expression impact the competency of PDAC cells to metastasize to the liver, we injected PDAC cells into the spleen of mice. We first injected KPC46 cells with KD of MICAL2, MRTF-A, and MRTF-B (Fig. 6A). Grossly, we observed a dramatic decrease in liver metastatic burden in the MICAL2, MRTF-A, and MRTF-B KD models compared with scramble control cells. Histologically, we found only small and often merely microscopically detectable liver metastases in the KD models compared with extensive gross metastatic disease when the control cells were injected (Fig. 6B). To determine whether the engraftment of dox-inducible MICAL2-KD cells contributed to this phenotype, we injected the ish-MICAL2 or ish-control KPC46 cells into the spleens of syngeneic mice and provided dox to the mice on the third day after injection. In this model, we found a dramatic reduction in gross and histologic metastases in the ish-MICAL2 group (Fig. 6C and D). Finally, we injected BxPc3 MICAL2-OE cells or vector control into the spleen of NSG mice. Although we did not observe gross metastatic disease in either group, microscopic analysis revealed numerous metastatic foci in the livers of mice implanted with BxPc3 MICAL2-OE cells whereas there was no metastatic disease detectable in the control group (Supplementary Fig. S8A and S8B). These results suggest that MICAL2, MRTF-A, and MRTF-B promote liver metastasis of PDAC cells.

Figure 6.

MICAL2, MRTF-A, and MRTF-B promote metastatic spread in vivo. A, Representative images of liver metastatic burden after splenic injection of KPC46 cell into syngeneic mice. KPC46 cells express shRNA vectors to silence MICAL2, MRTF-A, and MRTF-B. B, Histologic quantification of liver metastasis area normalized to total liver area. ANOVA was used for statistical analysis. C, Representative images of liver metastatic burden after splenic injection of KPC46 cell into syngeneic mice. KPC46 cells express dox-inducible shRNA vectors to silence MICAL2. D, Histologic quantification of liver metastasis area normalized to total liver area. One-tailed t test was used for statistical analysis.

Discussion

There remains a critical need to identify better pharmacologic targets that drive pancreatic cancer. Although commonly mutated, effectively targeting mutations in KRAS, CDKN2A, SMAD4, and TP53 has not been possible until the recent development of KRAS inhibitors, and no drugs have yet gained approval for pancreatic cancer therapy. Importantly, these and other known genetic alterations do not sufficiently explain the vast tumor heterogeneity and unique biology of PDAC (38, 39). Patient outcomes, including survival, associated with PDAC subtypes have been strongly linked to epigenomic alterations in oncogenes and tumor suppressors (40–43).

SEs, as marked by H3K27ac, are clusters of enhancers with aberrantly high levels of transcription factor binding, which are central to driving the expression of genes that control cell identity and stimulate oncogenic transcription (4). Certain environmental or patient-specific events during tumorigenesis likely modulate specific SEs to influence target gene expression associated with PDAC phenotypic outcomes (43). We hypothesized that characterizing the landscape of H3K27 acetylation in pancreatic cancer tissue would lead us to identify genes critical in driving tumor progression and maintenance. Although there have been efforts to target global epigenetic changes (44–46) and thereby mitigate PDAC-promoting pathways, our study is one of the first to directly perform unbiased discovery using an epigenetic screen on primary tumor and normal pancreatic tissues.

Our characterization of SE-associated regions through H3K27ac ChIP-seq identified multiple differentially acetylated regions of the genome between tumor and normal tissues. Among them included many genes associated with PDAC phenotypes such as perineural invasion and cytoskeleton dynamics. Interestingly, we also found the leukemia inhibitory factor, previously identified as a putative target in pancreatic cancer and a subject of ongoing clinical trials (27). We chose to focus on the MICAL2 gene as it was a differentially acetylated gene between PDAC and the normal human pancreas. Furthermore, the enrichment of the SRF transcriptional motif in the PDAC-associated SE also provided a strong rationale to pursue MICAL2. MICAL2 is a flavin monooxygenase, and families of flavoproteins have been successfully targeted in human disease, i.e., monoamine oxidase inhibitors (47) for depression and neurodegenerative disorders (48), as well as inhibitors of flavo-containing monooxygenases like methimazole (49) to target excess thyroid hormone synthesis. Hence, there is significant foundation to posit that it may be possible to inhibit the enzymatic domain of MICAL2 pharmacologically. MICAL2 canonical function is linked to the transcriptional control of MRTF/SRF signaling. MRTF/SRF signaling has been shown to regulate the transcription of genes that promote tumor EMT, fibrosis, and metastasis, which are hallmarks of PDAC biology (50). In previous studies, MICAL2 has been primarily shown to promote EMT and migration in malignancies such as lung (11), gastric (12), and breast cancer (13) cell carcinomas. In addition, a single prior publication suggested that MRTF-A and MRTF-B were oncogenic based on their modulation in PDAC cell lines (51). Importantly, no prior studies have investigated the impact of MICAL2 on PDAC cells or studied its relationship with MRTF-A/B/SRF signaling in this disease. At the time of article submission, using bioinformatic approaches, Liu and colleagues (52) reported that MICAL2 is overexpressed in PDAC and that its expression was associated with EMT and poor prognosis, findings entirely consistent with our own. They further noted that MICAL2 is expressed in PDAC-associated fibroblasts and an immunosuppressive microenvironment, findings that only serve to heighten our interest in MICAL2 as a putative therapeutic target.

After identifying MICAL2, we confirmed its OE at the RNA and protein levels in both human and mouse PDAC cells. In vitro, we found MICAL2 to be a driver of cell proliferation, cell-cycle progression, matrix invasion, macropinocytosis, and migration whereas in vivo, we determined that MICAL2 promotes tumor growth and metastasis in several model systems. We have shown that MICAL2 expression regulates MRTF–SRF transcriptional targets and that the MICAL2-regulated factors MRTF-A and MRTF-B also play a critical role in PDAC biology. These findings, for the first time to the best of our knowledge, clearly implicate SRF transcription as an important driver of PDAC gene expression. The silencing of MICAL2 and MRTF-B led to a marked decrease in tumor growth in both heterotopic and orthotopic mouse models, suggesting their important role in tumorigenesis. We found that MICAL2, MRTF-A, and MRTF-B promote PDAC metastasis. Our findings that link MICAL2-promoted tumor progression specifically to MRTF-B are novel, as is the identification of a regulatory loop whereby they reciprocally promote each other’s transcription. Yet, our study also reveals a clear distinction between the impact of MRTF-A and MRTF-B on PDAC tumorigenesis and progression, as MRTF-A promoted metastasis but not tumor growth, whereas MRTF-B was necessary for both metastasis and tumor growth, demonstrating a differential role of MRTF isoforms in PDAC.

We made some unexpected and novel observations that further heightened our interest in MICAL2 as a putative target for pancreatic cancer therapy, namely, our RNA-seq data, suggesting that the loss of MICAL2 expression downregulates genes associated with KRAS signaling. We further found that the loss of MICAL2 expression in vitro resulted in decreased ERK1/2 and AKT phosphorylation. Although ERK1/2 and PI3K are activated downstream of various signals, the impact of MICAL2 may be linked to its effects on KRAS. This hypothesis is bolstered by our findings that the loss of MICAL2 expression significantly reduced macropinocytosis, a KRAS-driven nutrient-scavenging activity, and increased the GTP-bound KRAS associated with MICAL2 OE in BxPc3 cells. Although the mechanisms by which MICAL2 may regulate KRAS are likely complex, a previous report found that MICAL2 promoted stabilization of EGFR (13). Thus, the regulation of receptor tyrosine kinase signaling is at least one possible mechanism for the link between MICAL2 and KRAS signaling worthy of further study. Interestingly, MTRF-A loss did not consistently result in a profound decrease in ERK1/2 and AKT activation. This suggests that the mechanism by which MICAL2 promotes KRAS signaling may not be simply related to its role in promoting MRTF/SRF transcription.

There are several limitations to our study, including that we did not specifically perform experiments to define the specific enhancer region most responsible for MICAL2 expression. Furthermore, as it promotes several oncogenic phenotypes that can affect metastatic competency, our experiments do not clearly distinguish the stage of metastasis most affected by MICAL2. Despite these limitations, our study expands upon previous reports on the role of MICAL2 in cancer by providing a comprehensive analysis of its role in PDAC. It further reveals, for the first time, that MICAL2 affects KRAS-related gene expression as well as KRAS-regulated macropinocytosis. However, many questions remain, most specifically related to the mechanisms by which MICAL2 promotes these oncogenic phenotypes and KRAS function. The canonical function of MICAL2 in promoting MRTF/SRF expression likely underpins many of these effects; however, it is also localized in the cytosolic protein with other functional domains that may play an important role in MICAL2-related phenotypes.

In conclusion, we have studied the epigenomic landscape of PDAC to identify MICAL2 as a unique driver of PDAC progression through its regulation of MRTF/SRF signaling and determined the differential effects of MRTF-A versus MRTF-B in promoting PDAC growth. MICAL2 and MRTF-SRF biology represent attractive, novel, and potentially tractable targets for PDAC therapy given their regulation of transcriptional programs fundamental to its tumor biology.

Supplementary Material

Supplementary Tables S1-S6

supplementary tables S1-S6

Supplementary Table S7

Supplementary Table S7

Supplementary Table S8

Supplementary Table S8

Supplementary Legends

Supplementary Figures and Tables legends.

Supplementary Figures

Supplementary Figures S1-S8

Authors’ Disclosures

A.S. Courelli reports grants from the NIH during the conduct of the study. A.T. Wenzel reports grants from the NIH during the conduct of the study. D. Jaquish reports grants from the NIH during the conduct of the study. K. Jaque reports grants from the NCI of the NIH during the conduct of the study. A. D’Ippolito reports personal fees from Syros Pharmaceuticals during the conduct of the study and from Syros Pharmaceuticals and Precede Biosciences outside the submitted work, as well as a patent for US-20230210852-A1 issued. D.A. Orlando reports personal fees from Syros Pharmaceutical during the conduct of the study, as well as a patent for US9181580B2 licensed to Syros Pharmaceuticals. C. Commisso reports a patent for US9983194B2 issued. H. Tiriac reports a patent for “Targeting MICAL2” pending. A.M. Lowy reports grants from The Lustgarten Foundation, Stand Up to Cancer, the Pancreatic Cancer Action Network, and the NIH/NCI during the conduct of the study, as well as a patent for “MICAL2 as a target for pancreatic cancer therapy” pending. No disclosures were reported by the other authors.

Authors’ Contributions

B. Garg: Conceptualization, data curation, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. S. Khan: Data curation, visualization, methodology, writing–original draft, writing–review and editing. A.S. Courelli: Data curation. P. Panneerpandian: Data curation. D. Sheik Pran Babu: Data curation, formal analysis, validation, methodology. E.S. Mose: Data curation, validation, investigation, visualization, methodology. K.C.M. Gulay: Data curation, validation, investigation, methodology. S. Sharma: Conceptualization, data curation, validation, methodology. D. Sood: Data curation, formal analysis, validation, visualization, methodology. A.T. Wenzel: Data curation, software, formal analysis, validation, investigation. A. Martsinkovskiy: Data curation, software, formal analysis, investigation, visualization, methodology. N. Rajbhandari: Data curation. J. Patel: Data curation, formal analysis, visualization, methodology. D. Jaquish: Data curation, validation, investigation, visualization, methodology. E. Esparza: Data curation. K. Jaque: Data curation. N. Aggarwal: Data curation. G. Lambies: Data curation, validation, visualization, methodology. A. D’Ippolito: Data curation, formal analysis, validation, visualization, methodology, writing–review and editing. K. Austgen: Data curation, formal analysis, validation, visualization, methodology. B. Johnston: Data curation, formal analysis, validation, investigation, methodology. D.A. Orlando: Data curation, validation, investigation, methodology, writing–review and editing. G.H. Jang: Data curation, formal analysis. S. Gallinger: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. E. Goodfellow: Data curation, visualization, methodology. P. Brodt: Conceptualization, data curation, validation, visualization, methodology, writing–review and editing. C. Commisso: Conceptualization, data curation, validation. P. Tamayo: Data curation, investigation, visualization, methodology. J.P. Mesirov: Data curation, investigation, visualization, methodology, writing–review and editing. H. Tiriac: Conceptualization, resources, data curation, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. A.M. Lowy: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.