Transcriptomic and functional analysis of ANGPTL4 overexpression in pancreatic cancer nominates targets that reverse chemoresistance

Abstract

Background

Pancreatic ductal adenocarcinoma (PDAC) is one of the deadliest cancers based on five-year survival rates. Genes contributing to chemoresistance represent novel therapeutic targets that can improve treatment response. Increased expression of ANGPTL4 in tumors correlates with poor outcomes in pancreatic cancer.

Methods

We used statistical analysis of publicly available gene expression data (TCGA-PAAD) to test whether expression of ANGPTL4 and its downstream targets, ITGB4 and APOL1, were correlated with patient survival. We measured the impact of ANGPTL4 overexpression in a common pancreatic cancer cell line, MIA PaCa-2 cells, using CRISPRa for overexpression and DsiRNA for knockdown. We characterized global gene expression changes associated with high levels of ANGPTL4 and response to gemcitabine treatment using RNA-sequencing. Gemcitabine dose response curves were calculated on modified cell lines by measuring cell viability with CellTiter-Glo (Promega). Impacts on cell migration were measured using a time course scratch assay.

Results

We show that ANGPTL4 overexpression leads to in vitro resistance to gemcitabine and reduced survival times in patients. Overexpression of ANGPTL4 induces transcriptional signatures of tumor invasion and metastasis, proliferation and differentiation, and inhibition of apoptosis. Analyses revealed an overlapping signature of genes associated with both ANGPTL4 activation and gemcitabine response. Increased expression of the genes in this signature in patient PDAC tissues was significantly associated with shorter patient survival. We identified 42 genes that were both co-regulated with ANGPTL4 and were responsive to gemcitabine treatment. ITGB4 and APOL1 were among these genes. Knockdown of either of these genes in cell lines overexpressing ANGPTL4 reversed the observed gemcitabine resistance and inhibited cellular migration associated with epithelial to mesenchymal transition (EMT) and ANGPTL4 overexpression.

Conclusions

These data suggest that ANGPTL4 promotes EMT and regulates the genes APOL1 and ITGB4. Importantly, we show that inhibition of both targets reverses chemoresistance and decreases migratory potential. Our findings have revealed a novel pathway regulating tumor response to treatment and suggest relevant therapeutic targets in pancreatic cancer.

Background

Pancreatic ductal adenocarcinoma (PDAC) is among the deadliest cancers with a five-year survival of only 11% (SEER) [1]. The high mortality of pancreatic cancer patients is explained by the lack of early diagnostic markers leading to late-stage diagnosis of the disease and the high rate of treatment resistance [2]. Standard of care includes resection for about 20% of patients with localized disease who are eligible and treatment with cytotoxic chemotherapies such as gemcitabine or combination agents like FOLFIRINOX (5-FU, oxaliplatin, folinic acid, and irinotecan). This combination, when tolerated, improves progression-free survival by 11 months [3]. Despite the aggressive treatment of this disease with combinations of cytotoxic drugs, the vast majority of patients do not have an enduring response to the treatment. Through improvements in precision cancer treatments based on genetic mutations, implementation of combination therapies like FOLFIRINOX, and advances in immunotherapy over the last decade the five-year survival has increased from 6 to 11%, nevertheless, pancreatic cancer remains extremely deadly. These factors combined with the increased incidence of risk factors like diabetes and obesity. It is predicted that pancreatic cancer will be the second leading cause of cancer deaths by 2030 [4].

Doubling the five-year survival rate is remarkable progress, but novel treatment strategies are still needed. One effective strategy is to identify cellular markers or pathways that are enriched in pancreatic cancer cells. For example, over 90% of PDAC tumors have a mutation in the KRAS gene, however, no KRAS targeting treatments are available and existing targeted therapies like PARP inhibitors [5] and EGFR inhibitors [6] are limited to a subset of patients. It is clear that targeted therapies are impactful for the eligible patients with significant improvement in progression-free survival among patients receiving a targeted treatment [7]. Unfortunately, patients who do not respond to standard chemotherapeutics and are not eligible for existing targeted therapies are left with few options. Thus, there is significant interest in understanding the development of resistance, and in pursuing targets that might sensitize resistant cells to existing treatments.

Our previous transcriptomic study identified genes whose expression was positively or negatively correlated with patient survival and also linked to in vitro response to gemcitabine, a common PDAC treatment. We identified ANGPTL4 among those genes whose expression in patient tumors is associated with poor survival and whose knockdown in PDAC cell lines could increase sensitivity to gemcitabine [8]. ANGPTL4 is a member of the family of angiopoietin-like proteins that were first described for their role in angiogenesis [9, 10]. ANGPTL4 can be proteolytically cleaved [11] and the two products each have their own functions. The N-terminal domain plays a role in lipid metabolism and genetic variation in this domain is linked to cardiovascular disease risk. The C-terminal domain has been implicated in tumorigenesis, the promotion of proliferation, and wound healing [11,12,13].

Overall, the role of ANGPTL4 in cardiovascular disease and lipid metabolism has been much better described than its roles in cancer. ANGPTL4 has a described role in known cancer pathways including its ability to regulate CREB, FOS, and STAT3 via ERK signaling [14, 15]. ANGPTL4’s ability to alter metabolism and ATP abundance can also impact drug transport [15]. The picture, however, is not perfectly clear since ANGPTL4 expression seems to have different impacts in cancers of different primary sites. For example, methylation and downregulation of ANGPTL4 are associated with progression and metastasis in colon cancer [16] but overexpression is associated with progression and poor prognosis in breast cancer [17]. Breast cancers of the triple negative subtype may be different since ANGPTL4 overexpression in that context has been associated with inhibition of invasion [18]. In pancreatic cancer, ANGPTL4 overexpression has been associated with tumorigenesis [19], cellular resistance to chemotherapy [8], hypoxia response, and poor patient outcomes [20]. The complicated role for ANGPTL4 motivates our further exploration of its function in pancreatic cancer.

Here, we describe the role of ANGPTL4’s in PDAC by exploring the molecular and cellular changes associated with altered activity of ANGPTL4, the impact of ANGPTL4 on chemoresistance, and the potential for inhibiting downstream pathways driven by ANGPTL4 activation to sensitize tumor cells to treatment. We show that overexpression of ANGPTL4 leads to chemoresistance, increased migratory potential, and proliferation. Our transcriptomic analysis revealed altered gene expression signatures downstream of ANGPTL4 overexpression that are linked to epithelial to mesenchymal transition (EMT) and predict patient outcomes. We showed that knockdown of downstream effectors including APOL1 and ITGB4 reversed resistance to treatment and reduced migratory potential. The expression of these genes is also associated with patient survival. These data support the hypothesis that ANGPTL4 and its downstream pathways are potential therapeutic targets for the reversal of treatment resistance in pancreatic cancer.

Methods

Cell culture

HEK293FT cells (ThermoFisher #70,007), MIA PaCa-2 cells (ATCC #CRM-CRL-1420), and Panc-1 (ATCC #CRL-1469) were cultured in D10 media: DMEM (Lonza #12-614Q) supplemented with 10% FBS, and 0.5% Penicillin-Streptomycin. All cell lines were maintained at 37 °C and 5% CO₂. Cells were cryopreserved with the addition of 10% DMSO (EMD #MX1458-6) to D10 media.

Plasmids

LentiSAMv2 (Addgene #92,062) and lenti-MS2-p65-HSF1-Hygro (Addgene #89,308) were used to generate stable cell lines for gene activation. pMD2.G (Addgene #12,259) and psPAX2 (Addgene #12,260) were used to facilitate viral packaging of sgRNA vector plasmids.

sgRNA cloning

sgRNA oligos were designed and cloned into their respective plasmids as described previously (ANGPTL4: 5’-CACGGCCCTGGGGATGCCAAACTGTGG-3’ and NTC: 5’-ACGGAGGCTAAGCGTCGCAA-3’) [21].

DsiRNA

IDT TriFECTa RNAi kit was used per manufacturer’s protocol. The DsiRNA sequences used were as follows: ANGPTL4 (IDT hs.Ri.ANGPTL4.13.1), APOL1 (IDT hs.Ri.APOL1.13.3), and ITGB4 (IDT hs.Ri.ITGB4.13.3). 100,000 cells were seeded per well of a 12 well tissue culture treated plate 24 h prior to transfection. Cells were transfected using RNAiMax (ThermoFisher #13778-030) following manufacturer’s recommended protocol. As directed in the TriFecta kit (IDT #hs.Ri.HDAC1.13), TYE 563 transfection efficiency control, positive HPRT-S1 control, and negative (DS NC1) scrambled sequence control were utilized. Functional assays were performed 48 h after transfection. Expression was validated with each transfection with IDT PrimeTime qPCR Assay system according to manufacturer’s recommendations on an Agilent QuantStudio 6 Flex Real-Time PCR system (ANGPTL4: Hs.PT.58.40146104, ACTB: Hs.PT.39a.22,214,847, GAPDH: Hs.PT.39a.22,214,836, and HPRT: Hs.PT.58v.45,621,572).

Protein quantification using AlphaLISA

Briefly, cells were seeded at 25,000 cells in 50uL of media per well of a 96-well tissue-culture treated plate in triplicate per cell type. The following day, the media from each well was transferred to a non-treated 96-well plate and 25ul of AlphaLISA Lysis Buffer (PerkinElmer, #AL001C) was added per well. Plates were then shaken at 250 RPM at room temperature for 10 min. Lysates were immediately used. Samples or standards were transferred to a 384 well AlphaPlate (PerkinElmer, #600,535, lot:8220–21,331) and the AlphaLISA ANGPTL4 manufacturer’s kit protocol (PerkinElmer, #AL3017HV, lot:2,905,318) was followed for standards, lysate, and supernatant (media) except for the following: lids were covered in foil to prevent light exposure when possible and during incubation periods plates were shaken at 150 RPM for the recommended time. The AlphaLISA plate was read on a BioTek Synergy H5 following previously published protocols [22]. Data were analyzed in R (version 4.1.2) and GraphPad Prism 9.

RNA-sequencing

Cells were seeded at a density of 0.6 × 10^6 cells per well of a 6-well plate in triplicate. The following day media was removed and 3ml of media per well with or without 1.5nM of gemcitabine (Sigma #G6423) was added and incubated for 24 hours. Media was removed, cells were washed with PBS (Gibco #10010072), and released from the plate with 2 ml of TrypLE (Gibco #12604-021) per well. After centrifugation, pellets were frozen at -80 C until RNA extraction. For RNA extraction, of cell pellets 350 ul of RL Buffer plus 1% BME from the Norgen Total RNA extraction kit and extraction proceeded per manufacturer’s instructions including use of the DNase kit (Norgen # 37500, 25720). RNA quality was verified with the Agilent BioAnalyzer RNA Nano 600 kit (cat# 5067 − 1512) with the RIN range between 9.2–10. RNA-sequencing libraries were made using Lexogen QuantSeq 3’ mRNA-Seq Library Prep Kit FWD for Illumina kit (cat# 015.24) with 250 ng of RNA input. They were pooled and sequenced on an Illumina NextSeq 500 instrument with 75 bp single-end reads. Read counts averaged 4 million reads and 93.65% of bases exceeding Q30. Lexogen’s BlueBee integrated QuantSeq data analyses pipeline was used for trimming, mapping, and alignment and DESeq2 was used for differential expression [23]. Heatmaps were generated using iDEP.95 [24].

Scratch-wound assay

Cells were seeded at 100,000 cells per well of a 96-well plate. After 24 h uniform scratches were made across the diameter of the wells using a multichannel pipette with 200ul pipette tips and even pressure applied across the wells to cause a wound. The media was then vacuumed off and 200ul of new media was added. Cells were imaged on a Lionheart FX Automated Microscope every 8 h for 48 h. Cell culture growth conditions of 37 °C and 5% CO₂ were maintained throughout. Forty images were taken of each scratch (4 wide by 10 long) with no overlap, autofocus, and auto brightness.

Images were integrated using R and the “tiff” package. Images were further processed using GIMP 2.10. ImageJ and the MRI_wound_Healing_Tool.ijm macro plugin [25] were used to process the images and calculate the area of the wound the cells have not covered. Relative wound closure over time using time 0 for each condition as the control was plotted and a curve line equation was formed by fitting the curve to a non-linear fit one phase decay least squares fit with Yo = 0 as a constraint. This base equation (Y=(Y0 - Plateau)*exp(-K*X) + Plateau) where Y0 is the Y value when X (time) is zero, Plateau is the Y value at infinite times, and K is the rate constant, was used to determine the value of X or half the time it takes to close the wound. Data were analyzed using GraphPad Prism 9.

Drug resistance screening

Cells were seeded in 96-well plates at 2000 cells/well. Seeded cells were treated with a range of gemcitabine concentrations. Cells were treated again 48 h later. The number of viable cells surviving drug treatment were measured with CellTiter-Glo (Promega #G7571) 24 h after the last drug treatment per the manufacturer’s protocol using a BioTek Synergy H5 plate reader. Sample size ranged from 4 to 11.

Pathway analysis

1198 DEG from the ANGPTL4 OE vs. KD analysis were imported into the KEGG Mapper-Color [26] and Panther GO Enrichment Analysis [27]. The following parameters were used for KEGG: Search mode: hsa, used uncolored diagrams, included aliases. The following parameters were used for PANTHER: overrepresentation test, homo sapiens reference list, GO biological process complete, Fisher’s exact test, and calculate FDR. Top pathways were merged manually and drawn using BioRender.

Overall survival analysis (OS)

To conduct survival analysis, clinical and RNA-seq expression data was retrieved from The Cancer Genome Atlas for 178 PDAC (TCGA-PAAD) patients (https://portal.gdc.cancer.gov/). Data were normalized using the R package DESeq2 and differentially expressed genes with an FDR < 0.1 were used to generate Kaplan-Meier survival curves. We classified tissues based on their average expression of a given gene set (bottom 25%, middle 50%, and top 25% of gene expression). We compared the patients with the lowest and highest quartile of average gene expression and performed survival analysis. Survival curves and analyses were generated using the “ggplot2”, “survminer”, and “survival” R packages. P values were generated using a log rank test.

Recurrence free survival (RFS)

A Kaplan Meier curve of recurrence-free survival survival plots for ANGPTL4 was created using GEPIA (http://gepia.cancer-pku.cn/) single gene analysis. The relevant parameters were as follows: Disease-free Survival (RFS), Group Cutoff: Quartile (75% high, 25% low), Hazards Ratio: Yes, 95% Confidence Interval: Yes, Axis Units: Months, and datasets Selection: PAAD.

Correlation analysis

To compute correlation matrices, RNA-seq expression data was retrieved from The Cancer Genome Atlas for 178 PDAC (TCGA-PAAD) patients (https://portal.gdc.cancer.gov/).

Rank-based correlation coefficients were computed for each of the DEG with log₂ fold change greater than 0.7 and baseMean > 10 (MP2_ANGPTL4_KD vs. MP2_ANGPTL4_OE with or without treatment) for (1) all TCGA-PAAD data DEG per gene vs. ANGPTL4 expression and (2) all TCGA-PAAD data DEG gene versus OS time. These correlation values were used to generate a list of genes that are co-expressed with ANGPTL4 is overexpressed.

Correlation coefficients and p-values were computed using the “Hmisc” (version 4.6-0) R package. We classified tissues based on their ANGPTL4 expression of a given gene set (bottom 25%, middle 50%, and top 25% of gene expression). We computed correlation matrices using the expression data for patients with the lowest and highest quartile of ANGPTL4 expression which was used to divide samples into quantiles with highest and lowest average gene expression; based on expression of these 42 genes. (40 genes were measured in TCGA: KDM7a and STN1 were omitted).

Statistical analysis

Statistical analysis was conducted in R (version 3.6.1 and R version 4.1.2 for RNAseq analysis). The following R packages and software were used for analysis:

survival (version 1.2.1335) [28]

survminer (version 0.4.9) [29]

ggplot2 (version 3.3.6) [30]

DESeq2 (version 1.24.0) [31]

pheatmap (version 1.0.12) [32]

Hmisc” (version 4.6-0) [33]

tiff (version 0.1–11) [34]

Results

ANGPTL4 overexpression increases gemcitabine resistance genes in PDAC

We previously showed that increased ANGPTL4 expression in tumors is associated with poor survival in PDAC patients (n = 51) [8]. Here, we confirm the association with survival in an independent cohort of 178 PDAC patients from The Cancer Genome Atlas (TCGA, PAAD expression dataset). Patients with the highest (top 25%) expression of ANGPTL4 in tumor tissue had reduced overall survival (OS) and disease-free survival, (p = 0.043 and p = 0.059 (log-rank test) (Fig. 1A, Supplementary Fig. S1A).

Fig. 1

Expression of ANGPTL4 is associated with patient outcomes and modifies expression of resistance-associated genes. (a) A Kaplan Meier curve for overall survival using TCGA-PAAD data. The patients with the top (red) and bottom (blue) 25% average ANGPTL4 gene expression have significantly different OS. (p = 0.043, log-rank test) (b) Relative expression of ANGPTL4 transcript in MP2_ANGPTL4_KD (black circles), MP2 (pink squares), and MP2_ANGPTL4_OE (turquoise triangles) as measured by qPCR. P-values from unpaired, two-tailed parametric t-tests performed with 95% confidence intervals are displayed. Values are normalized with ACTB as the housekeeper. (P-values ≤ 0.0001=****) (c) Heatmap displaying RNA-seq results from MP2_ANGPTL4_KD (black), MP2_NTC (pink), and MP2_ANGPTL4_OE (turquoise) cell lines. Displayed are 15 genes associated with both ANGPTL4 overexpression and gemcitabine resistance. Normalized expression for each of the 15 genes is plotted in the heatmap. Red (*) indicates ITGB4 used for later experiments

Our previous findings motivated the current study to understand possible mechanisms by which ANGPTL4 contributes to patient survival and specifically how it alters survival outcomes through a possible role in drug resistance. We measured the impact of ANGPTL4 overexpression and knockdown in the MIA PaCa-2 cell line using CRISPRa (MP2_ANGPTL4_OE) [21] and siRNA knockdown (MP2_ANGPTL4_KD). qPCR showed CRISPRa successfully increased expression of the ANGPTL4 transcript by 8-fold compared to a non-targeting control guide. Similarly, siRNA knockdown reduced expression by over 90% of the control line (Fig. 1B, Supplementary Table S1). An ELISA assay showed that protein levels were affected similarly (Supplementary Fig. S1B). We measured global gene expression changes using RNA-sequencing on the modified cell lines and controls which revealed 1198 differentially expressed genes (DEG) with a mean read count greater than 10, absolute log₂Fold Change (log2FC) of at least 0.7, and adjusted p-value (padj) less than 0.05) when comparing overexpressed (OE) ANGPTL4 cells with the knockdown (KD) ANGPTL4 cells (Supplementary Fig. S1C, Supplementary Table S2). Given the observed association with drug resistance, we asked whether these DEGs were also associated with gemcitabine resistance by overlapping the 1198 DEG with 114 known gemcitabine resistance genes (Supplementary Table S3) [35,36,37,38]. This revealed 15 gemcitabine resistance genes that are altered with overexpression of ANGPTL4, 10 of which overlap genes that are part of the EMTome (Fig. 1C) [39]. Several of these genes are linked to activation of TGFβ and its downstream signaling pathways: ERK, PI3K/AKT, and JNK (SH3RF2, TGFBR3, PPP1R15A, INPP4B, WWTR1, and FGF2), contributors to EMT (SOX9, CUL7, and TGFBR3), and transcriptional regulation of pluripotent stem cells (STAT3, KLF4, and FGF2) [26, 27]. Additional pathway analysis showed a connection between the expression of ANGPTL4 and EMT (GO:0001837) in the presence (3.24-fold enrichment, 0.0273 FDR) or absence (3.04-fold enrichment, 0.195 FDR) of gemcitabine treatment (Supplementary Table S4-S5).

Increased ANGPTL4 expression alters cellular response to gemcitabine

We assessed whether altering ANGPTL4 expression impacted response to gemcitabine to explore the potential impact of ANGPTL4 overexpression on treatment response in patients. We measured cell viability in MP2_ANGPTL4_KD and MP2_ANGPTL4_OE cell lines before and after treatment with gemcitabine. As expected, treatment with 12.5nM gemcitabine reduces viability compared to untreated cells. Comparing cells with overexpression of ANGPTL4 to controls, we observed an increase in cell viability with the same gemcitabine concentrations. Similarly, there was a significant reduction in viability with the ANGPTL4 knockdown compared to cells overexpressing ANGPTL4. This difference in viability is not observed in the untreated cells (Fig. 2A, Supplementary Table S1). These results demonstrate a greater sensitivity to gemcitabine with reduction of ANGPTL4. To understand the link between ANGPTL4 expression and gemcitabine response, we measured transcriptional response to gemcitabine in MP2_ANGPTL4_OE cells and MP2_ANGPTL4_KD cells using RNA-sequencing. We were not surprised to note that 955 genes were differentially expressed upon treatment with the cytotoxic gemcitabine alone. We explored the intersection of the 1198 DEGs identified upon ANGPTL4 overexpression and the 955 DEGs associated with gemcitabine treatment. This analysis revealed 42 genes associated with gemcitabine response and ANGPTL4 overexpression (Fig. 2B, Supplementary Table S6). This strategy nominated genes that might contribute to altered response to gemcitabine in cells overexpressing ANGPTL4 and are potentially important for patient outcomes. We tested whether these genes were predictive of patient overall survival using data from TCGA. Since all 42 transcripts were positively correlated with ANGPTL4 expression, we calculated the mean gene expression of these genes and found that patients whose tumors had the highest average expression had significantly reduced overall survival (p = 0.0073, Fig. 2C). The fraction of patients surviving 1500 days or more is dramatically increased in the individuals with low expression of these 42 ANGPTL4 and gemcitabine response genes. This compares favorably to the average 3-year relative survival for PDAC patients in the United States with resectable stage I/II disease- 17% [40].

Fig. 2

Gemcitabine resistance signature in PDAC cells with high ANGPTL4. (a) Cell viability was measured for MP2_ANGPTL4_KD (black), MP2_NTC (pink), and MP2_ANGPTL4_OE (turquoise) cells treated with 0, 5nM, or 12.5nM gemcitabine. Relative viability reflects normalization to cells not treated with gemcitabine. P-value asterisks: ≤0.01=**, ≤ 0.001=*** & ≤0.0001=****, unpaired, two-tailed t-test (b) Heatmap of gene expression measured by RNA-seq in MP2_ANGPTL4_KD (black/gray), MP2 (pink/light pink), and MP2_ANGPTL4_OE (turquoise/aqua) cell lines both untreated and treated with 1.5nM gemcitabine. Normalized expression of the 42 genes that resulted from an intersection of ANGPTL4 expression DEGs and gemcitabine treatment response DEGs are included in the heatmap. (c) Kaplan-Meier curve for overall survival using TCGA-PAAD data where patients were divided into quartiles with highest (red) and lowest (blue) using mean expression of the 42 genes in the intersection (40 genes were available in TCGA: KDM7a and STN1 were omitted). P-values were derived from a log-rank test

ITGB4 and APOL1 inhibition sensitizes high ANGPTL4 cells to gemcitabine

In vitro overexpression of ANGPTL4 alters cellular drug response and correlates with patient survival but the pathway(s) by which this resistance is achieved are unknown. We reasoned that genes co-regulated with ANGPTL4 may function with ANGPTL4 and contribute to resistance and found those genes by identifying genes that are co-expressed with ANGPTL4 in patient tissues. For each of the 1198 DEGs identified when ANGPTL4 was overexpressed in vitro, we calculated rank-based correlation coefficients and identified co-expressed genes in patient tissues (TCGA-PAAD). For each gene with a significant correlation with ANGPTL4 in patient tissues, we determined whether expression of that gene alone was significantly correlated with patient overall survival (Supplementary Table S7). This allowed further filtering of the list based on the assumption that true resistance-associated genes would impact patient survival. From this list, ITGB4 and APOL1 are both positively correlated with ANGPTL4 expression in vitro and in patients, altered with gemcitabine response in vitro, and correlated with patient survival in tissues. While there are several genes of interest, we focused on ITGB4 and APOL1 in this study, in part based on their drugability (pharos.nih.gov) and availability of efficient siRNA knockdown reagents. Based on the co-expression with ANGPTL4 and the link to patient survival, we tested whether expression of these genes in combination with ANGPTL4 was relevant for patient outcomes. Comparing tissues that were among the top 25% expression of both ANGPTL4 and APOL1 (Fig. 3A) or ITGB4 (Fig. 3B) to those that were among the bottom 25% expression of both ANGPTL4 and APOL1 or ITGB4 we saw significant differences in overall survival, with overall survival being worse for TCGA-PAAD patients with high ANGPTL4 combined with high APOL1 (p = 0.014) or ITGB4 (p = 0.01) expression. Tumors with high levels (top 25%) of all three genes; APOL1, ITGB4, and ANGPTL4, also had reduced survival compared with the bottom 25%, but when APOL1 and ITGB4 are combined with ANGPTL4 expression the reduction in survival did not become more significant (p = 0.015) (Supplementary Fig. S1D).

Fig. 3

ITGB4 and APOL1 inhibition sensitizes cells overexpressing ANGPTL4 to gemcitabine. a-b) Kaplan-Meier plot for overall survival using TCGA-PAAD data. Patients with the top (red) and bottom (blue) 25% average ANGPTL4 gene expression and APOL1 or ITGB4 gene expression (p = 0.01 and p = 0.014, respectively, log-rank test). c) ANGPTL4 expression measured by qPCR normalized to MP2_ANGPTL4_OE + DsiRNA_NTC (turquoise) control: MP2 (pink), MP2_ANGPTL4_OE_ITGB4_KD (lavender), and MP2_ANGPTL4_OE_APOL1_KD (purple) compared to mean of MP2_ANGPTL4_OE + DsiRNA_NTC (MP2_ANGPTL4_OE cells transduced with non-targeting control siRNA). Expression is normalized to the housekeeping gene ACTB. P-values ≤ 0.0001=****; unpaired two-tailed parametric t-test d) Relative viability is plotted by normalizing all data to cells not treated with gemcitabine. MP2_ANGPTL4_KD (black), MP2_NTC (pink), MP2_ANGPTL4_OE + DsiRNA_NTC (turquoise), MP2_ANGPTL4_OE_ITGB4_KD (lavender), and MP2_ANGPTL4_OE_APOL1_KD (purple) cells treated with 0, 5nM, or 12.5nM gemcitabine. Sample size ranges from 4 to 11 per condition. P-value asterisks: ≤0.001=*** & ≤0.0001=****, unpaired, two-tailed parametric t-tests

Given the association of ANGPTL4 with resistance and patient outcome, we tested whether, like ANGPTL4, loss of ITGB4 or APOL1 expression increased sensitivity to gemcitabine in vitro. Using siRNAs, we knocked down APOL1 and ITGB4 in the MP2_ANGPTL4_OE cell line. We showed that treatment with the siRNAs reduced mRNA levels of the target genes (Supplementary Figure S2A-B), and we also found that knockdown of ITGB4 or APOL1 reduced expression of ANGPTL4 significantly compared to a non-targeting siRNA control with ANGPTL4 overexpression background; expression was reduced to 2% in MP2, 5% in ITGB4 knockdown, and 25% in APOL1 knockdown (Fig. 3C, Supplementary Table S1). Given our previous finding that overexpression of ANGPTL4 leads to resistance, we hypothesized that knockdown of APOL1 and ITGB4 might reverse resistance given the impact on ANGPTL4 expression. We tested the impact of ITGB4 and APOL1 knockout on gemcitabine sensitivity in the MP2_ANGPTL4_OE line. After gemcitabine treatment, cells with knockdown of ITGB4 or APOL1 in the ANGPTL4_OE background (MP2_ANGPTL4_OE_ITGB4_KD or MP2_ANGPTL4_OE_APOL1_KD) showed reversal of drug resistance when compared to the ANGPTL4 OE cells (Fig. 3D). There was no difference in growth rate among the lines that explained the differential viability observed with gemcitabine treatment (Supplementary Fig. S2C).

ANGPTL4 overexpression increases cell migration

To further understand the pathways that might be involved in resistance, we performed pathway enrichment analysis on the 1198 DEG associated with overexpression of ANGPTL4 (OE v KD). Those included tumor invasion and metastasis, proliferation, differentiation, inhibition of apoptosis, and cell membrane formation. Genes participating in invasion and metastasis included RhoEGF which was increased with ANGPTL4 expression. RhoEGF is linked to tumor invasion and metastasis [41]. Many genes in the Ras-ARF6 and Ras-MEK2 signaling pathways are also upregulated (e.g. FGF, APOL1, ITGB4, EGFR, MPK, CREB, cFOS, JUN, and VEGF) impacting proliferation and differentiation. In addition, ANTPGL4 overexpression is associated with the upregulation of anti-apoptotic pathways including JAK-STAT, Bcl_XS, and SOCS pathways triggered by upregulation of cytokine, cytokine receptors, CNTF, and IL22RA2. The formation of cellular structure was also affected by ANGPTL4 and increase in Claudin via the decrease in cell permeability (GEF-H1, Claudin2), increased actin assembly (ARP2/3), and cell polarity (AMPK, DLG1) (Fig. 4A, Supplementary Table S8).

Fig. 4

ANGPTL4 overexpression correlation with EMT. (a) Pathway analysis of 1198 DEG revealed common cancer pathways. Red = log₂FC > 0.7 and Blue = log₂FC < 0.7. (b) Wound healing assay results at 0 h (after initial wound) and 48 h MP2_NTC (pink), MP2_ANGPTL4_OE + DsiRNA_NTC (turquoise), MP2_ANGPTL4_OE_ITGB4_KD (lavender), and MP2_ANGPTL4_OE_APOL1_KD (purple). Each set was performed in triplicate or quadruplicate. (c) Time to close half the gap (t1/2 gap) for each of the four cell types is plotted. t1/2: MP2 = 57.7 h, MP2_ANGPTL4_OE = 31.6 h., MP2_ANGPTL4_OE_ITGB4_KD = 192 h., and MP2_ANGPTL4_OE_APOL1_KD = 52.96 h. P-values ≤ 0.01=**, unpaired parametric t-test

Others have shown that on their own ANGPTL4, APOL1, and ITGB4 are each associated with cell proliferation and contribute to PDAC progression [17, 42, 43]. We measured migration using a wound healing time course assay to determine if the increased migration associated with ANGPTL4 overexpression could be reversed with the KD of ITGB4 or APOL1 (Fig. 4B). We observed that knockdown of either APOL1 or ITGB4 in lines overexpressing ANGPTL4 reduced the migratory potential of these cells. This was apparent 48 h after the scratch was made and was quantified by calculating the time to close half of the gap (t1/2 gap) for each of the four cell types is plotted. (Fig. 4C, Supplementary Table S1).

Discussion

Here we have described how overexpression of ANGPTL4 in pancreatic cancer contributes to disease progression and resistance. We have shown that overexpression of this gene and protein leads to cellular resistance to gemcitabine, one of the most commonly used chemotherapeutics in PDAC. It follows that if overexpression leads to chemoresistance, increased expression of this gene would also be associated with poor patient outcomes, and we confirmed that in independent patient cohorts. The goal of our study is to further understand the role of this gene in chemoresistance in the hopes that a mechanistic understanding will facilitate the development of new treatment strategies. Our transcriptomic analysis revealed that overexpression of ANGPTL4 broadly impacts transcriptomic profiles in pancreatic cancer cells. This is perhaps not surprising given a rather large body of literature describing not only ANGPTL4’s role in cancer but also cardiovascular disease risk and metabolism [9]. In fact, ANGPTL4’s role in cancer can be difficult to summarize because it has been shown to have both protective and promoting effects in different cancer types [13]. Our initial findings highlighted the potential importance of this gene and existing literature was not sufficient to understand how this gene functions in pancreatic cancer. Using our transcriptomic data, we narrowed our analysis to a list of ANGPTL4-impacted genes that are also responsive to gemcitabine treatment in vitro. We identified 42 genes associated with ANGPTL4 overexpression and gemcitabine response and determined that the expression of those genes effectively predicted patient outcomes.

Further narrowing our focus, we identified APOL1 and ITGB4 among the genes that are associated with ANGPTL4 overexpression, gemcitabine resistance, and patient survival and explored them further. We found that knockdown of either APOL1 or ITGB4 increased sensitivity to gemcitabine in cells overexpressing ANGPTL4. This finding confirms that inhibition of APOL1 or ITGB4 can reverse resistance associated with ANGPTL4 expression in pancreatic cancer cells. One unexpected finding was that knockdown of these genes also reduced expression of ANGPTL4 itself. Rather than our initial hypothesis that expression of ANGPTL4 was correlated with APOL1 or ITGB4 because they were regulated by ANGPTL4, this data supports the hypothesis that ANGPTL4 is regulated, directly or indirectly, by APOL1 and ITGB4 and these genes could be involved in a feedback loop that explains our results.

Considering APOL1 first, APOL1 (apolipoprotein L1) is part of the HDL lipid complex which plays a key role in lipid metabolism [44]. APOL1 can activate NOTCH1 signaling leading to reduced proliferation and increased apoptosis [42]. NOTCH1 also upregulates ABCC1, a transporter that can also lead to multidrug resistance [45]. In other contexts, NOTCH1 also has been shown to stabilize and activate PPARγ, a transcription factor that promotes transcription of ANGPTL4 [46]. These results suggest a possible mechanism by which knockdown of APOL1 may reduce ANGPTL4 expression and further suggests that inhibition of APOL1, especially in patients with high expression of ANGPTL4 might reduce cellular capacity for migration and proliferation. Additional studies would need to be performed to demonstrate whether the link between ANGPTL4 and APOL1 requires NOTCH signaling.

ITGB4 knockdown also reduces ANGPTL4 overexpression. ITGB4 (integrin β4, ɑ6β4) is an integrin protein that plays a role in cell-extracellular matrix adhesion. It is upregulated in several cancers including pancreatic cancer and has been linked to poor prognosis, high tumor grade, lymph node metastasis, and drug response [47,48,49]. Knockdown of ITGB4 increases cisplatin sensitivity in lung cancer models [47] suggesting that ITGB4 is relevant for resistance to multiple chemotherapeutics. ITGB4 may promote receptor tyrosine kinase (RTK) activation through ERBB2 [50]. RTKs generally promote proliferation, invasion, and sustained cell migration, and ITGB4 has a synergistic effect in combination with increased RTKs expression [50, 51]. In our study, we observed increased expression of CSF3, VEGFA, HBEGF, and DDR2, which all play a role in the epithelial-to-mesenchymal transition in cells overexpressing ANGPTL4. An alternate hypothesis is that ITGB4 promotes cell migration and invasion through regulation of the MEK1-ERK1/2 signaling cascade [49]. Reduction of ITGB4 in the context of overexpressed ANGPTL4 could also affect this pathway, explaining the reduced migratory potential of ITGB4 KD cells in the cells overexpressing ANGPTL4. Additional transcriptomic and biochemical studies are needed to determine which of these pathways are most relevant for the chemoresistance phenotypes.

Conclusions

Our study provides evidence that ANGPTL4 overexpression in pancreatic cancer impacts several key signaling and metabolic pathways associated with patient survival, in part through controlling cellular resistance to chemotherapy and cellular migration. ITGB4 and APOL1 knockdown are both able to reverse drug resistance and migration increases associated with overexpression of ANGPTL4 but appear to do so through different mechanisms. These findings emphasize the complexity of developing treatments to target resistance and metastasis; a single approach may not be sufficient. Nonetheless, both APOL1 and ITGB4 represent potential novel targets and our data support the further exploration of these genes and the pathways in which they function.