Abstract
Lysine-specific demethylase 1 (LSD1) is a histone modifier that is highly overexpressed in lung adenocarcinoma, which results in aggressive tumor biology. Tumor cell proliferation and migration analysis after LSD1 inhibition in the lung adenocarcinoma cell line PC9, using the LSD1 inhibitor HCI-2509 and siRNA, demonstrated that LSD1 activity was essential for proliferation and migration capacities of tumor cells. Moreover, reduced proliferation rates after LSD1 inhibition were shown to be associated with a cell-cycle arrest of the tumor cells in the G2–M-phase. Expression profiling followed by functional classification and pathway analysis indicated prominent repression of the polo-like kinase 1 (PLK1) pathway upon LSD1 inhibition. In contrast, transient overexpression of exogenous PLK1 plasmid rescued the LSD1 inhibition–mediated downregulation of PLK1 pathway genes. Mechanistically, LSD1 directly regulates expression of PLK1 by binding to its promoter region that subsequently affects expression of its downstream target genes. Notably, using lung adenocarcinoma TCGA datasets a significant correlation between LSD1 and PLK1 along with its downstream targets was observed. Furthermore, the LSD1/PLK1 linkage was confirmed by IHC analysis in a clinical lung adenocarcinoma cohort (n = 43). Conclusively, this is the first study showing a direct transcriptional link between LSD1 and PLK1.
Implications: These findings point to a role of LSD1 in regulating PLK1 and thus efficient G2–M-transition–mediating proliferation of tumor cells and suggest targeting the LSD1/PLK1 axis as a novel therapeutic approach for lung adenocarcinoma treatment.
Introduction
Lung cancer is ranked first among cancer-related deaths worldwide (1). The most common subtype of non–small cell lung cancer (NSCLC), lung adenocarcinoma, is associated with a poor 5-year survival rate of less than 20% after diagnosis, and is defined as a disease with genetic and cellular heterogeneity (2, 3). Comprehensive molecular characterization of lung adenocarcinoma has revealed a huge complexity of not only genetic alterations but also epigenetic alterations (4, 5). Importantly, tumor progression and development in lung adenocarcinoma has been shown to be linked with epigenetic silencing of crucial tumor suppressor genes (6, 7).
Posttranslational modification of histone tails is a complex mechanism regulating gene expression changes. Lysine-specific demethylase 1 (LSD1/KDM1A) is one such epigenetic modifier that significantly affects transcriptomic changes by demethylating histone tails. LSD1 belongs to the amine oxidase superfamily of proteins, and catalyzes FAD-dependent oxidative demethylation (8). Specifically, it demethylates mono/di-methylated histone 3 lysine 4 (K4) and lysine 9 (K9) (H3K4 and H3K9) residues, thereby altering the epigenetic marks and inducing gene expression changes (8, 9). In addition, LSD1 also demethylates lysine residues of nonhistone proteins like E2F1, TP53, and DNMT1, resulting in a change in their cellular properties (10, 11). LSD1 performs these functions in association with different histone modifying complexes and transcription coactivating or corepressing complexes (12). LSD1 is also crucial for the development and maintenance of normal tissue homeostasis and for stem cell differentiation, and has been shown to be involved in embryogenesis (13–15).
Furthermore, high expression of LSD1 has been linked to poor prognosis in various cancer types, including breast cancer, hepatocellular carcinoma, esophageal cancer, lung cancer, neuroblastoma, as well as acute myeloid leukemia (13). As a result, many compounds targeting LSD1 are being employed in preclinical studies (16, 17). Among these inhibitors, HCI-2509 a reversible LSD1 inhibitor identified by extensive screening approaches has been shown to be highly specific for LSD1 (18). This compound has demonstrated to significantly inhibit tumor growth in prostate and Ewing sarcoma tumor cell lines and is expected to enter phase I clinical trials within the near future (19–21).
In this study, we used HCI-2509 to disrupt LSD1 activity in lung adenocarcinoma cells. Inhibition of LSD1 reduced cell proliferation, downregulated major cell growth regulatory pathways, and most predominantly led to downregulation of the polo-like kinase (PLK) mitotic pathway. We demonstrate that PLK1 is a direct transcriptional target of LSD1, and that depletion of LSD1 activates G2–M arrest. Moreover, we show that LSD1 binds to the PLK1 promoter region and thereby regulates its expression and tumor cell proliferation. In summary, we propose that therapeutic targeting of LSD1 offers a mechanistic rationale for the treatment of lung adenocarcinoma and other cancers that exhibit PLK1 addiction.
Materials and Methods
Cell culture
Information about all cell lines used in this study is listed in Supplementary File S1; Supplementary Table S1. Cell lines were either cultured in DMEM or RPMI media, supplemented with 10% FCS, in presence or absence of l-Glutamine. All cell cultures were maintained in 5% CO2 at 37°C. Cells were passaged when 80% confluency was reached.
Cell lines used in this study are not listed in the International Cell Line Authentication Committee's misidentified cell lines database. All cell lines used in the study were tested for Mycoplasma using Venor GeM OneStep Kit (Minerva Biolabs; last test, November 2017). On an average all the cell lines used were in between 15–40 passage number and were used for cellular assays after 5–8 passages postthawing.
Cell treatment with the LSD1 inhibitor
One day prior to treatment, cells were plated on 20 cm2 plates and then supplemented with 2 μmol/L HCI-2509 (Xcess Biosciences). For microarray analysis and the corresponding validation experiments, and for the expression analysis in different cancer lines, HCI-2509 treatments were performed for 48 hours, while for all other experiments it was for 72 hours. Cells were harvested after 48 or 72 hours for either protein or RNA isolation.
Transfection with silencing RNA and plasmids
Transfections were performed using control siRNA (SIC001; Sigma) and three different siRNA sequences against LSD1 at a final concentration of 50 nmol/L for 72 hours using Lipofectamine 2000 as described earlier (Thermo Fisher Scientific; ref. 22). The following siRNA oligonucleotides were used in the knockdown experiments: siLSD1-S1: sense 5′-CUGCAGUUGUGGUUGGAUAtt-3′ and antisense 5′-UAUCCAACCACAACUGCAGtg-3′; siLSD1-S2: sense 5′-CACAAGGAAAGCUAGAAGAtt-3′ and antisense 5′-UCUUCUAGCUUUCCUUGUGtt-3′; and siLSD1-S3: sense 5′-AGGCCUAGACAUUAAACUGaa-3′ and antisense 5′-CAGUUUAAUGUCUAGGCCUaa-3′. Plasmid transfections were performed using a PLK1 overexpression construct [mCherry-PLK1-N-16 was a gift from Michael Davidson, Florida State University, Tallahassee, FL (Addgene plasmid catalog no. 55119; http://n2t.net/addgene:55119; RRID: Addgene_55119)] and a mock control, [pSicoR-Ef1a-mCh-Puro was a gift from Bruce Conklin, Gladstone Institute of Cardiovascular Disease, San Francisco, CA (Addgene plasmid catalog no 31845; http://n2t.net/addgene:31845; RRID: Addgene_31845)] using Lipofectamine 2000 according to the manufacturer's instructions.
RNA isolation and quantitative PCR
Isolation of RNA from cells was performed using the Maxwell LEV simplyRNA Tissue Kit (Promega) according to the manufacturer's instructions. For real-time PCR, RNA was reverse transcribed into cDNA using TaqMan Reverse Transcription Kit (Applied Biosystems). GoTaq QPCR Master Mix (Promega) was used to perform real-time PCR. Primers used are listed in Supplementary File S1; Supplementary Table S2. The real-time PCR was run on CFX96 Thermo Cycler (Bio-Rad) or Roche Lightcycler 480 (Roche).
Microarray
Gene expression–profiling analysis was performed on untreated PC9 cells and PC9 cells treated with the LSD1 inhibitor HCI-2509. RNA isolation from cells was performed using the AmbionPureLink RNA Mini Kit (Life Technologies) following the manufacturer's instructions. The Human Gene 2.0 ST Array Platform (Affymetrix) was used for performing the microarray analysis. Significance parameters used: P < 0.05 and fold change > 1.5. All details regarding the Gene Ontology and pathway analysis can be found in the Supplementary File S1; Supplementary Methods section. The microarray data generated are available in the Gene Expression Omnibus repository under accession number GSE117702.
Protein isolation and Western blot analysis
Protein lysates were extracted from cells using the Pierce RIPA Buffer (Thermo Fisher Scientific) and blotted as described previously (23). For studying phosphorylated proteins by Western blot analysis, protein lysates were extracted from cells using Cell Lysis Buffer 10X (New England Biolabs) as per the manufacturer's instructions. The membranes were incubated overnight with primary antibodies in blocking solution (5% milk powder in PBST) at 4°C using the antibodies listed in Supplementary File S1; Supplementary Table S3. Developed blots were imaged using ChemiDoc XRs+ System (Bio-Rad) and processed using the associated Image Lab software v 4.0.
Cell-cycle analysis
Cycletest Plus DNA Kit (BD Biosciences) was used to determine percentage of cells in different phases of the cell cycle following the manufacturer's instructions. Cell-cycle measurement analysis was performed on FACSVerse Flow Cytometer (BD Biosciences). All experiments were performed in triplicates. Percentage of cells in different phases of cell cycle was calculated by plotting propidium iodide area histogram.
Immunofluorescence
Cells were grown in 12-well plates on coverslips and cell treatments were performed as mentioned above. Seventy-two hours later cells were fixed using 100% ice cold methanol for 10 minutes and blocked using 0.1% fish gelatin (G7041; Sigma) in PBS for 1 hour at room temperature. Cells were then sequentially stained with primary (overnight at 4°C) and fluorescent secondary antibodies (2 hours at room temperature) with intermediate washing steps using the blocking buffer. Antibodies used have been listed in Supplementary File S1; Supplementary Table S3. Confocal images were collected on Olympus Fluoview FV 1000 scanning confocal microscope using a 60 × oil objective and 1 × magnification. Olympus Fluoview FV 1000 software was used for data acquisition. Image files were processed using Fiji v 2.0.0 (ImageJ) and Adobe Photoshop.
TCGA lung adenocarcinoma RNA-seq analysis
A processed dataset (z-transformation) of RNA-seq clinical data from lung adenocarcinoma (TCGA, Provisional) was downloaded from the cancer genomics repository cBioPortal (24). The dataset were grouped in two by filtering for LSD1 expression; LSD1 levels above the 90th percentile was defined as a high LSD1–expressing group, while levels below the 10th percentile were defined as a low LSD1–expressing group. Differentially expressed genes after LSD1 inhibition were plotted against the grouped dataset in an unsupervised way following Spearman correlation with the help of R (25). The association of LSD1 and PLK1 mRNA expression in the TCGA lung adenocarcinoma data were calculated by Spearman correlation coefficient.
IHC staining of tissue microarray
Staining was performed as described previously (26) on paraffin-embedded tumor sections of prepared tissue microarrays (TMA) consisting of 43 lung adenocarcinoma patients, using α-LSD1 and PLK1 antibodies as per the manufacturer's instructions. Staining intensities for LSD1 and PLK1 were calculated using QuPath (27), where the DAB channel intensity was extracted for each represented core of the TMA. In addition, immunostaining results for LSD1 and PLK1 were scored on a scale of 1–3 (1 = low, 2 = medium, and 3 = high), by experienced pathologists (S.C. Schaefer and S. Klein). The association between LSD1 and PLK1 protein levels was calculated by Spearman correlation coefficient.
The study was approved by the local ethics committee (KEK No. 200/2014) and informed consent was obtained from all patients. All methods were performed in accordance with the relevant guidelines and regulations of the institution.
Chromatin immunoprecipitation
Chromatin immunoprecipitation (ChIP) was performed using SimpleChIP Enzymatic Chromatin IP Kit (Cell Signaling Technology) as per the manufacturer's instructions. Two steps of the protocol were optimized; sonication and amount of nuclease, by using different setups for sonication and different concentrations of nuclease to obtain appropriate fragmentation of chromatin. For sonication, eight cycles of 30 seconds ON and 30 seconds OFF each was determined to be optimal and 0.5 μg nuclease per 10 million cells. ChIP was performed using 20 μg lysate with 5 μg each of LSD1 antibody and nonspecific rabbit IgG antibody. Eluted DNA was tested for successful working of protocol by analyzing binding of LSD1 to known gene targets by qPCR. Primers specific to different regions of PLK1 promoter were designed; for CDKN1A, promoter region sequences were obtained from previous reports (22). Each promoter region was analyzed in triplicate qPCR reactions per experiment and three independent experiments were performed. Data are presented relative to percent of input. List of primers used for ChIP qPCR is provided in Supplementary File S1; Supplementary Table S2.
Statistical analysis
Student t test (two-tailed) and two-way ANOVA tests were used to analyze the in vitro and the lung adenocarcinoma clinical data by GraphPad Prism v5 and “R” (25). Significance parameters used: P < 0.05.
Results
LSD1 inhibition results in a differential gene expression signature
Previous studies have shown that inhibition or silencing of LSD1 leads to reduced tumor cell proliferation and migration capacities in vitro, as well as in vivo (28–30). Here, we could confirm these results using the LSD1 inhibitor HCI-2509 or with siRNA, that also led to reduced cell proliferation and migration capacities in the lung adenocarcinoma cell line PC9 (Supplementary File S1; Supplementary Fig. S1). To identify the underlying molecular mechanisms of LSD1 inhibition in lung adenocarcinoma, we first examined gene expression patterns of the EGFR-mutated (ΔE746-A750) lung adenocarcinoma cell line PC9, in presence of the LSD1 inhibitor HCI-2509 using hybridization arrays. Comparison of the transcriptome of untreated PC9 cells and LSD1-inhibited PC9 cells identified 2,570 (>1.5-fold change, P < 0.05) genes to be differentially regulated (Fig. 1A). Gene ontology analysis using PANTHER classification system, revealed enrichment of genes involved in multiple biological processes, including cellular and metabolic processes, as well as immune response–related processes (Fig. 1B). Next, we subjected the dataset for gene set enrichment analysis (GSEA) to analyze the differential expression of biologically relevant sets of genes that share common biological functions after LSD1 inhibition. The top 25 up- and downregulated genes, exhibited several genes associated with cell cycle to be significantly depleted after LSD1 inhibition (Fig. 1C). Among them the most prominent genes were PLK1, BORA, KIF18A, HISTH2AC, and HIST1H3I.
LSD1 inhibition in PC9 cells results into differential gene expression. A, Pie chart indicating the total number of differentially expressed genes upon LSD1 inhibition in PC9 cells using HCI-2509. Microarray data from n = 3 biological replicates. B, Bar chart depicting the enriched biological processes on y axis and the total number of genes annotated to each biological process on x axis. C, Heatmap representing the top 50 genes identified by GSEA to be deregulated in the PC9 cells after HCI-2509 treatment (red, relatively high gene expression and blue, relatively low gene expression). D, GSEA enrichment plots showing negative enrichment of different cell-cycle–related gene sets, as represented by negative NES (normalized enrichment score). NES reflects the degree to which the gene set was changed after treatment as compared with the untreated control. Significant gene sets were defined as those with a nominal P < 0.05. E, Bar chart depicting predicted canonical pathways on the y axis and the corresponding significance of expression indicated as −log10(P) on the x axis. Indicated in black boxes on the right are activation z-scores. A positive z-score indicates an increased activity of the canonical pathway whereas a negative z-score indicates the decreased activity of the indicated canonical pathway.
Gene expression profiling suggests a role of LSD1 in cell-cycle regulation
An unbiased transcriptome profiling using GSEA showed that the untreated group was negatively enriched in cell-cycle progression genes (Fig. 1D). In contrast, the HCI-2509–treated group was positively enriched with genes associated with cell-cycle arrest. GSEA revealed that the G2–M cell cycle and PLK1 pathways were among the most enriched gene sets in the untreated control group as compared with the HCI-2509–treated group. To further validate these results, we used ingenuity pathway analysis (IPA) to examine the relationship between the highly significant genes and to determine the key canonical pathways which are dysregulated by LSD1. In line with the results of the GSEA analysis, IPA analysis revealed two growth regulatory pathways playing a central role in cell cycle to be significantly altered. Both pathways, namely, the G2–M DNA damage checkpoint regulation pathway and the mitotic role of PLKs (Fig. 1E) pathway had an activation z-score of >±2. Together, a consistent finding using both GSEA and IPA was that multiple genes related to cell-cycle regulation were significantly altered after LSD1 inhibition using HCI-2509. These results indicated a reduced cell growth through downregulation of cell-cycle regulatory genes.
LSD1 inhibition represses the PLK1 mitotic pathway
With an in depth analysis of transcriptional changes using GSEA and IPA, we recognized that the PLK1 mitotic pathway was most significantly downregulated upon LSD1 inhibition. By pharmacologically disrupting LSD1 function due to HCI-2509 treatment, several of the PLK1 target genes, including ECT2, CDC20, CDC25C, KIF11, TERF11, TOP2A, BUB1, AURKA, ORC2, CCNB1, KIF20A, BORA, and PLK1 itself were downregulated (Fig. 2A). To further verify these results, the expression level of PLK1 target genes was quantified by qPCR. Indeed, the qPCR results were in accordance with the microarray analysis (Fig. 2B). Furthermore, we examined the effect of LSD1 inhibition on PLK1 expression on a panel of 23 different cancer cell lines from breast, liver, esophageal, neuroblastoma, and lung cancer by treating these cell lines with the LSD1 inhibitor HCI-2509. qPCR analyses revealed that the majority of the cell lines showed reduced PLK1 expression (Fig. 2C). Notably, 13 of the 23 cell lines showed at least 50% downregulation of PLK1 mRNA expression in response to LSD1 inhibition by HCI-2509.
LSD1 inhibition downregulates expression of PLK1 and its target genes. A, Heatmap showing the expression profile of PLK1 target genes after LSD1 inhibition in PC9 cells using HCI-2509. B, Validation of microarray results of the differentially expressed PLK1 target genes in PC9 cells after LSD1 inhibition using HCI-2509 by qPCR in an independent experiment. C, PLK1 gene expression in different human cancer cell lines after LSD1 inhibition using HCI-2509 measured by qPCR. Data in B and C are plotted as mean ± SD for n = 3 biological replicates; *, P < 0.05; **, P < 0.01; and ***, P < 0.001 as compared with untreated group. D, Immunoblot showing protein levels of PLK1, AURKA, CDKN1A, and H3K4me2 in untreated PC9 cells and cells treated with HCI-2509. E, Bar graph showing mRNA expression levels of LSD1 and PLK1 target genes quantified after knockdown of LSD1 in PC9 cells using three different siRNA sequences. Data are presented as mean ± SD for n = 3 biological replicates; *, P < 0.05 as compared with the scrambled siRNA control. F, Immunoblot showing protein levels of LSD1, PLK1, H3K4me2, and CDKN1A in scrambled transfected PC9 cells and cells transfected with siLSD1.
The methylation levels of LSD1 substrate H3K4me2 were altered after LSD1 inhibition (Fig. 2D). An increase in H3K4me2 levels was detected using HCI-2509, confirming its on-target activity. Compared with the control cells, LSD1 inhibition showed substantial downregulation of PLK1 at the protein level. Similarly, the protein expression of PLK1's positively regulated target aurora kinase A (AURKA) was significantly reduced, whereas the expression of its negatively regulated target cyclin-dependent kinase inhibitor 1A (CDKN1A) was significantly increased, suggesting that LSD1 inhibition hampered PLK1 activity. Similar effects on the expression of PLK1 and its target genes at transcript and protein levels was observed upon LSD1 knockdown using three different siRNAs (Fig. 2E and F).
To further control for an antiproliferative effect of LSD1 inhibition in lung adenocarcinoma, we analyzed a publicly available dataset containing gene expression profile of the lung adenocarcinoma cell line A549 after treatment with the anticancer drug cisplatin (GSE6410; ref. 31) for the top hits of the differentially expressed genes. Particularly, we noticed no difference in the expression of the corresponding genes using a multivariate analysis (Supplementary File S1; Supplementary Fig. S2). The gene expression patterns of the genes evaluated did not overlap with the effect observed after LSD1 inhibition using HCI-2509, suggesting that related downregulation of PLK1 target genes is LSD1 specific and not a general antiproliferative effect of the inhibitor.
Transient overexpression of PLK1 overcomes LSD1 inhibition–mediated alteration of its target genes
To demonstrate that the altered expression of PLK1 and its target genes in response to LSD1 inhibition using HCI-2509 is a LSD1-specific effect, we transiently overexpressed PLK1 plasmid in the lung adenocarcinoma cell line PC9 in combination with HCI-2509. As shown in Fig. 3A and B, ectopic expression of PLK1 construct in the cells followed with LSD1 inhibition using HCI-2509, did no longer affect PLK1 protein or mRNA levels. Ectopic expression of PLK1 by transient transfection not only blocked LSD1 inhibition–induced alteration of PLK1, but also its target genes. The positively regulated targets of PLK1, BUB1, AURKA, and CCBN1, showed a trend of increasing mRNA levels after PLK1 overexpression (Fig. 3C). Whereas, HCI-2509–mediated LSD1 inhibition had relatively less or no effect on the expression of BUB1, AURKA, and CCBN1 in presence of transgenic PLK1 overexpression. Likewise, the negatively regulated target gene of PLK1, CDKN1A, showed reduced mRNA levels after PLK1 overexpression, whereas, this effect was minimized in combination with HCI-2509. Overall, transient overexpression of exogenous PLK1 in PC9 cells rescued the LSD1 inhibition–mediated downregulation of PLK1 target genes, indicating the positive regulation of PLK1 by LSD1.
Ectopic overexpression of PLK1 rescues LSD1 inhibition–mediated effect. Immunoblot showing PLK1 protein levels (A) and bar graph showing PLK1 mRNA expression levels of PC9 cells transfected with mock and PLK1 overexpression plasmid (B), in absence and presence of HCI-2509. C, Bar graph showing transcript levels of PLK1 target genes in untreated PC9 cells, cells treated with HCI-2509 and PC9 cells transfected with mock and PLK1 overexpression plasmid, and overexpression constructs in combination with HCI-2509. Data in B and C are plotted as mean ± SD for n = 3 biological replicates.
Inhibition of LSD1 results into mitotic arrest
PLK1 is highly expressed in cancer cells and its major role is implemented in cell proliferation by regulating the G2–M-phase transition of cell cycle. To investigate whether the repressed PLK1 levels caused by LSD1 inhibition lead to abnormalities in cell-cycle progression, potentially due to the inability of cells to transit from G2–M-phase, we performed cell-cycle analysis after treatment of PC9 cells with HCI-2509 using flow cytometry (Fig. 4A). LSD1 inhibition in PC9 cells led to accumulation of cells in the S- and G2–M-phase of cell cycle, while it led to decrease of cells in the G1-phase as compared with the control cells (Fig. 4B). In addition, LSD1 inhibition using HCI-2509 and knockdown using siRNA induced depletion of the mitotic marker H3S28ph in PC9 cells (Fig. 4C). Furthermore, we performed immunofluorescence (IF) staining to observe the effect of LSD1 inhibition on mitosis. The most visible effect was reduced LSD1 and PLK1 levels in the cells after HCI-2509 treatment and perinuclear localization of LSD1 (Fig. 4D). Noticeably, with nuclear counterstaining no mitotic cells were detected after LSD1 inhibition. These results are in accordance with our experimental observations showing that LSD1 inhibition using HCI-2509 or siLSD1 results into less cell growth and proliferation in PC9 cells.
LSD1 inhibition leads to reduction in mitotic activity. A, Representative histograms showing cell-cycle distribution of untreated and PC9 cells treated with HCI-2509, measured by propidium iodide staining followed by flow cytometry (n = 3). B, Bar graph representing the percent cell-cycle distribution in each condition. Data presented are mean ± SD for n = 3 biological replicates; #, P < 0.001 as compared with untreated group by two-way ANOVA. C, Immunoblot showing protein levels of H3S28ph in untreated PC9 cells, cells treated with HCI-2509 and cells transfected with scrambled control and siRNA against LSD1. D, Representative microscopic images of untreated PC9 cells (top) and HCI-2509 treated PC9 cells (bottom), stained with DAPI (blue), LSD1 (green), and PLK1 (magenta). White arrows indicate mitotic cells. Scale bar, 25 μm; n = 3 biological replicates.
LSD1 and PLK1 expression positively correlates at mRNA and protein levels in lung adenocarcinoma
While the results of LSD1 inhibition showed a picture of differentially expressed genes, we next asked whether this could be translated into patient-related data. To identify, whether there is a correlation between LSD1 and PLK1, we first analyzed RNA-seq dataset of TCGA from patients with lung adenocarcinoma (517 lung adenocarcinoma cases). Impressively, we observed a statistically significant correlation between LSD1 and PLK1 mRNA expression (r = 0.42; P < 0.05; Fig. 5A). To further determine, whether the correlation that we found at transcript level also exists at protein level, we inspected 43 lung adenocarcinoma clinical samples for LSD1 and PLK1 protein expression. Strikingly, in the studied lung adenocarcinoma specimens we detected a strong correlation (r = 0.53; P = 1.1e−06) between the two proteins immunohistochemically on the TMA (Fig. 5B and C). In a previous study, we showed that LSD1 levels correlate with tumor grade in lung adenocarcinoma (26). As evident from Fig. 5C, on similar lines, we observed that PLK1 expression was higher in high LSD1–expressing tumors (higher tumor grade, g3) and lower in low LSD1–expressing tumors (lower tumor grade, g1). Nuclear LSD1 expression was observed all over the tumor area, however a scattered PLK1 expression was observed in the tumor cells, with mitotic cells expressing the most (Supplementary File S1; Supplementary Fig. S3).
LSD1 and PLK1 expression correlates at mRNA and protein levels in clinical lung adenocarcinoma samples. A, Figure shows Spearman correlation matrix between LSD1 and PLK family members in a TCGA lung adenocarcinoma mRNA expression dataset. Correlation between LSD1 and PLK1 is indicated with a red box. Values that reach statistical significance (P < 0.05) have been represented with a circle (bigger the circle, higher is the significance). Blank spaces indicate a nonsignificant correlation defined by a P > 0.01. The color of the circles in the matrix represent the level of correlation, with 1 indicating positive correlation (dark blue) and −1 indicating negative correlation (dark red) between the two protein families. B, Correlation plot between LSD1 and PLK1 protein expression analyzed by IHC on the lung adenocarcinoma TMA. C, Exemplary images of IHC staining performed on lung adenocarcinoma patient samples; top, tumor grade 3 (poorly differentiated) lung adenocarcinoma sample, showing corresponding high LSD1 and PLK1 levels, bottom, tumor grade 1 (well differentiated) showing corresponding low LSD1 and PLK1 levels. Scale bar, 250 μm.
Dysregulation of LSD1 in lung adenocarcinoma affects tumor-associated activation of the PLK1 pathway
Next, we evaluated a TCGA lung adenocarcinoma dataset to examine whether the LSD1/PLK1 axis in lung adenocarcinoma governs the expression of PLK1 target genes involved in cell-cycle transition. Using an unsupervised hierarchical clustering approach, we identified a cluster of genes that was differentially expressed in LSD1-high versus LSD1-low–expressing tumors. Interestingly, among the clustered genes many targets of LSD1 and PLK1 were detected (Fig. 6), whose expression positively correlated with that of LSD1 and PLK1.
PLK1 target genes are highly expressed in high-LSD1–expressing lung adenocarcinoma tumors. Expression levels for differentially expressed genes after LSD1 inhibition plotted in an unsupervised way following Spearman correlation against a grouped (high- and low-LSD1–expressing group) TCGA lung adenocarcinoma dataset. Columns represent samples and rows represent genes ordered by hierarchical clustering analysis. (Red, relatively high gene expression and blue, relatively low gene expression). The cluster of PLK1 target genes that were found to be differentially expressed between the two groups is highlighted.
LSD1 occupies the promoter region of PLK1
To assess the direct involvement of LSD1 in the regulation of PLK1, we examined the occupancy of LSD1 on PLK1 genomic region by ChIP followed by qPCR. The ChIP assay showed that LSD1 binds to the PLK1 promoter region and that inhibition of LSD1 using HCI-2509 significantly reduced this effect in PC9 cells (Fig. 7A). As reported earlier, binding of LSD1 to CDKN1A promoter region could also be shown, which significantly decreased after LSD1 inhibition (Fig. 7B). We also validated the effect of change in LSD1 occupancy on PLK1 and CDKN1A promoter regions in PC9 cells after inhibition of LSD1 using siRNA (Fig. 7C and D). These data demonstrate that LSD1 binds at the PLK1 promoter site and thereby potentially regulates PLK1 expression.
LSD1 binds to PLK1 promoter region. Shown are representative results of IgG and LSD1 ChIP performed on DNA samples from untreated and HCI-2509–treated PC9 cells (top), and siControl and siLSD1-transfected PC9 cells (bottom). ChIP qPCR analysis of IgG and LSD1 at PLK1 (A and C) and CDKN1A (B and D) gene promoter regions. Data are presented as mean ± SD of n = 2 technical replicates from one representative experiment are shown [entire ChIP experiment performed three times (one experiment = two technical replicates)].
Discussion
LSD1 was the first histone lysine demethylase discovered, and since then it has been attributed to play various biological roles in different cellular processes (8, 32). LSD1 is remarkably overexpressed in a variety of tumor types and its expression correlates with aggressive tumor biology (33, 34). Therefore, because of its widespread functions in various cancers, disrupting LSD1 signal effect in human malignancies is a potential novel therapeutic option. In this study, our findings support a previously undescribed role of LSD1 in cell-cycle progression, by regulating the key cell-cycle kinase PLK1.
Previous work from our laboratory showed that LSD1 is overexpressed in lung adenocarcinomas and its expression correlates with tumor grade and invasive phenotypes (26). To identify the underlying molecular and cellular mechanisms associated with higher tumor cell proliferation in presence of LSD1, we investigated the impact of LSD1 inhibition in lung adenocarcinoma cells using the LSD1 inhibitor HCI-2509. Microarray profiling revealed that LSD1 inhibition led to differential expression of genes, and further comprehensive analysis showed a significant downregulation of several genes associated with cell-cycle progression and therefore leading to a cell-cycle arrest. The gene expression profiles are consistent with the work of Liang and colleagues, who also showed deregulation of cell-cycle–related genes upon LSD1 inhibition (35). Interestingly, classification of the differentially expressed genes identified the G2–M DNA damage checkpoint regulation pathway to be significantly activated, and the mitotic roles of PLK pathway to be significantly downregulated after LSD1 inhibition. These results are in agreement with the known role of LSD1 in DNA damage response. Predicted activation of the DNA damage checkpoint regulatory pathway implied the repression of cell-cycle progression, a mechanism known to be regulated by PLKs (36).
PLKs are vital cell-cycle modulators that are overexpressed in various tumors and are regarded as important prognostic indicators for antitumor therapy (37–39). PLK1 is the major player of the PLK family, which regulates diverse cell-cycle activities, including centrosome maturation, mitotic entry, spindle assembly, chromosomal alignment, and segregation and cytokinesis (40). To execute the different mitotic events it interacts with and regulates the function of numerous cell-cycle proteins (41–44). In our study, PLK1 and its downstream cell-cycle regulatory genes were differentially expressed after LSD1 inhibition with HCI-2509, as well as after siLSD1-mediated knockdown. Furthermore, PLK1 overexpression was sufficient to rescue LSD1 inhibition–mediated alteration of PLK1 target genes. Mitotic entry and exit is coordinated by PLK1, therefore repression of downstream targets genes of PLK1 upon LSD1 inhibition should lead to a decrease in mitotic activity of the tumor cells (45). Significantly, cell cycle, H3S28ph mitotic assay and IF analysis confirmed reduced mitotic activity and arrest of cells in S-G2–M-phase.
The diminished levels of LSD1 observed by IF can be very well reasoned with the experimental work of Sehrawat and colleagues, who could clearly show that HCI-2509 reduces LSD1 stability and targets it for proteasomal degradation (21). Whereas the altered localization of LSD1 could be due to lower nuclear import rates or protein sequestration in the cytoplasm. A number of studies have indicated similar effects on cell-cycle arrest after inhibition or knockdown of LSD1 (46–48). Taken together, the current findings suggest that LSD1 inhibition blocked cell-cycle advancement by deregulating the key mitotic mediators.
To rule out the possibility of a cell line–specific effect of LSD1 inhibition on PLK1 expression using HCI-2509, we examined the effect of LSD1 inhibition on different cancer cell lines. The majority of the cell lines revealed distinct downregulation of PLK1 mRNA expression in response to LSD1 inhibition. The varying effects between cancer cell lines could potentially be explained by their genetic diversity, as well as their distinct epigenetic landscape. In addition, decreased protein levels of PLK1 and AURKA, and increased levels of CDKN1A following inhibition and knockdown of LSD1 were observed. Downregulation of PLK1 and AURKA disrupts the PLK1/AURKA axis necessary for G2–M transition (49). Overall, these data indicate a prominent role of LSD1 in controlling cell proliferation via PLK1.
Analyzing a lung adenocarcinoma TCGA dataset revealed that PLK1 target genes which were downregulated after LSD1 inhibition had elevated levels in high-LSD1–expressing tumors in contrast to the low-LSD1–expressing tumors. These results suggest that LSD1 and PLK1 coexpress simultaneously in high levels in patients with lung adenocarcinoma. Strikingly, we found a statistically significant correlation between LSD1 and PLK1 at mRNA, as well as at protein level in two separate patient cohorts. Interestingly, two independent previous studies showed individual correlation of PLK1 and LSD1 with poor tumor differentiation and advanced clinical stage in NSCLC (26, 50). This is a first study reporting evidence for correlation of LSD1 and PLK1 in NSCLC. Together, these results point out to a direct transcriptional regulation of PLK1 expression by LSD1.
On the basis of these findings, we can confirm that PLK1 upregulation by LSD1 is a potential mechanism through which LSD1 drives tumorigenesis and invasiveness in cancer. To date, several studies have shown that inhibition or silencing of LSD1 leads to reduced cell proliferation, G2–M-arrest, mitotic defects, nonetheless the underlying rationale behind remained unidentified. In conclusion, this study sheds new light on to a molecular mechanism, why disruption of LSD1 function inhibits cell proliferation in tumor cells. Importantly, our data also define downregulation of PLK1 as a new biological tumor marker, which could be used to monitor tumors in vivo with respect to their responses to LSD1 inhibitors.
Disclosure of Potential Conflicts of Interest
R. Buettner is co-founder and chief scientific officer at Targos Molecular Pathology, Inc and has received speakers bureau honoraria from BMS, AZ, Roche, MSD, Illumina, and Boehringer. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: P.S. Dalvi, S.-Y. Lim, R. Buettner, S. Klein, M. Odenthal
Development of methodology: P.S. Dalvi, M. Müller, R. Buettner, M. Odenthal
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.S. Dalvi, I.F. Macheleidt, H. Eischeid-Scholz
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.S. Dalvi, I.F. Macheleidt, M. Müller, S.C. Schaefer, R. Buettner, S. Klein
Writing, review, and/or revision of the manuscript: P.S. Dalvi, I.F. Macheleidt, S. Meemboor, S. Klein, M. Odenthal
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.S. Dalvi, S. Meemboor, M. Müller, H. Eischeid-Scholz, S.C. Schaefer, R. Buettner
Study supervision: S. Klein, M. Odenthal
Other (conducted enlisted experimental procedures): P.S. Dalvi
Acknowledgments
The authors are thankful for the excellent technical support of Olivia Kaesgen and Ursula Rommerscheidt-Fuss. This work was supported by the funding received from the Center for Molecular Medicine Cologne (2635 8000 01) and Federal German Ministry of Science and Education (3632 1407 21 all to R. Buettner and M. Odenthal) and the North Rhine-Westfalia Ministry of Innovation and Sciences as part of e-Med SMOOSE and PerMed-initiatives. In addition, the project was supported by the Else Kröner-Fresenius Stiftung (2016-Kolleg-19 received to S. Klein).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Note: Supplementary data for this article are available at Molecular Cancer Research Online (http://mcr.aacrjournals.org/).
Mol Cancer Res 2019;17:1326–37
- Received September 12, 2018.
- Revision received January 16, 2019.
- Accepted February 8, 2019.
- Published first February 13, 2019.
- ©2019 American Association for Cancer Research.
References
Volume 17, Issue 6
