The lung-enriched p53 mutants V157F and R158L/P regulate a gain of function transcriptome in lung cancer

Abstract

Lung cancer is the leading cause of cancer-related deaths in the USA, and alterations in the tumor suppressor gene TP53 are the most frequent somatic mutation among all histologic subtypes of lung cancer. Mutations in TP53 frequently result in a protein that exhibits not only loss of tumor suppressor capability but also oncogenic gain-of-function (GOF). The canonical p53 hotspot mutants R175H and R273H, for example, confer upon tumors a metastatic phenotype in murine models of mutant p53. To the best of our knowledge, GOF phenotypes of the less often studied V157, R158 and A159 mutants—which occur with higher frequency in lung cancer compared with other solid tumors—have not been defined. In this study, we aimed to define whether the lung mutants are simply equivalent to full loss of the p53 locus, or whether they additionally acquire the ability to drive new downstream effector pathways. Using a publicly available human lung cancer dataset, we characterized patients with V157, R158 and A159 p53 mutations. In addition, we show here that cell lines with mutant p53-V157F, p53-R158L and p53-R158P exhibit a loss of expression of canonical wild-type p53 target genes. Furthermore, these lung-enriched p53 mutants regulate genes not previously linked to p53 function including PLAU. Paradoxically, mutant p53 represses genes associated with increased cell viability, migration and invasion. These findings collectively represent the first demonstration that lung-enriched p53 mutations at V157 and R158 regulate a novel transcriptome in human lung cancer cells and may confer de novo function.

Introduction

Cancers of the lung and bronchus, the leading cause of cancer mortality, are responsible for nearly 30% of all cancer deaths in the USA (1). The primary risk factor for lung cancer is cigarette smoking, which is estimated to account for 80% to 90% of lung cancer cases (2,3). In the USA, smoking rates have been on the decline since the 1980s, followed ~20 years later by decreasing lung cancer incidence (2). Other regions of the world continue to have a high prevalence of cigarette smoking, as well as exposures to additional etiologic agents, commensurate with high rates of lung cancer (2).

Tobacco smoke contains ≥60 vapor and particulate phase carcinogens, including polycyclic aromatic hydrocarbons such as benzo(a)pyrene, which are metabolically activated and ultimately induce mutations in DNA (3–5). Years of exposure to cigarette smoke results in high cumulative doses of these carcinogens, leading to preneoplasia in the airway epithelium and ultimately tumorigenesis (6). Benzo(a)pyrene in particular reacts preferentially with codon 157, among other loci, in the tumor suppressor gene TP53 to form bulky DNA adducts to generate a G:C to T:A transversion mutation (7,8).

The TP53 gene, located on the short arm of chromosome 17, encodes the tumor suppressor protein p53 and is the most commonly mutated gene in human cancer (9). Wild-type p53 is a tetrameric transcription factor typically found at very low levels with a short half-life until it is activated by cellular insults such as DNA damage, oncogene activation, hypoxia, oxidative stress, ribonucleotide depletion and nutrient deprivation (10). Following phosphorylation and displacement from its negative regulators MDM2 and MDM4, p53 binds to specific response elements and transcriptionally activates a myriad of target genes including CDKN1A, BAX and PUMA, ultimately leading to cell cycle arrest, senescence, or apoptosis, depending on the cellular context (11). In addition to these canonical tumor suppressive functions, p53 may also modulate additional cellular processes including inhibiting metabolic reprogramming in cancer cells (12), limiting accumulation of reactive oxygen species (13), preventing stem cell self-renewal (14), and promoting autophagy (15) and tumor microenvironment signaling (10,16).

Somatic mutations in TP53 are most commonly single nucleotide missense mutations that allow production of a full-length protein (17). Some of these mutant p53 proteins not only lose tumor suppressive function but also can acquire oncogenic gain-of-function (GOF) properties including increased invasion (18), proliferation (19), and chemoresistance (20), among others (21–24). GOF have been identified and validated most frequently in canonical p53 hotspot mutations such as R175H and R273H, both in vitro and in vivo (19,25,26). It has been suggested that GOF activity may depend upon a variety of contexts including tissue type and tumor microenvironment (27,28).

Of particular interest are mutations in p53 at V157, R158 and A159, which occur with increased frequency in lung cancer, surpassing the frequency of several of the traditional hotspots (29–32). The V157F mutation is driven by exposure to benzo(a)pyrene, and the V157F and R158L mutants exhibit defective transactivation ability in yeast functional assays, with ˂20% of wild-type activity on p53 response elements (30). To the best of our knowledge, the biological consequences of mutations at V157, R158 and A159 as a lung-enriched cluster remain uncharacterized.

Cigarette smoking is, without question, the major cause of lung cancer through induction of a mutagenic environment leading to deleterious gene alterations (33). Given the relative abundance of the lung-enriched p53 mutations at V157, R158 and A159, however, we propose an additional, non-competing, hypothesis by which this lung mutant cluster (LMC) imparts an oncogenic GOF, promoting their preferential selection in the lung. Here, we examine clinical characteristics of these lung-specific mutations in a human lung cancer dataset from The Cancer Genome Atlas (TCGA) and identify a novel transcriptome in vitro with mutant p53-dependent alterations not previously linked to mutant p53 function.

Materials and methods

The Cancer Genome Atlas Data

Data from TCGA were accessed via the cBioPortal for Cancer Genomics (34,35). Frequency of nucleotide changes and clinical characteristics including age, sex, smoking history, cancer stage at diagnosis and histology were extracted for human lung adenocarcinoma and lung squamous cell cancer cases. Statistical analysis was performed using Stata (version 12.1, College Station, TX).

Cell culture

Human NCI-H460, NCI-H2087, NCI-H441 and NCI-H2110 cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA) in May 2016 (H460), August 2016 (H441 and H2110), and January 2017 (H2087). A549 cells (passage 4) were obtained in August 2016 and NCI-H2009 cells were obtained in August 2017; both cell lines were authenticated by ATCC in October 2018. ATCC uses cytochrome C oxidase I gene analysis and short tandem repeat profiling to confirm the identity of human cell lines. Following receipt of all cell lines, aliquots of passages 1–6 were frozen in liquid nitrogen. New aliquots were thawed every 4–6 months for use in these experiments. Cells were maintained in RPMI 1640 (Corning) supplemented with 10% fetal bovine serum (Gemini) at 37°C and 5% CO₂. Cells were passaged twice weekly and monitored for Mycoplasma contamination by PCR detection (LookOut Mycoplasma PCR detection kit, Sigma–Aldrich) within 6 months of being thawed. The chemotherapeutic agent cisplatin was used at a concentration of 10 μM (US Pharmacopeia).

Transfection of TP53 siRNA

Transient knockdown of p53 was performed using ON-TARGETplus Human TP53 siRNA SMARTpool (Dharmacon), MISSION siRNA against TP53 (Sigma NM_000546), or MISSION siRNA Universal Negative Control (Sigma SIC001). Cells were transfected with 25nM siRNA using DharmaFECT 1 Transfection Reagent (Dharmacon) and harvested 72 h following transfection.

Quantitative RT-PCR

Cell lysates were harvested and RNA was extracted using TRIzol (Invitrogen) according to the manufacturer’s instructions. Complementary DNA (cDNA) was generated using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quantitative reverse transcription polymerase chain reaction (qRT–PCR) was performed using the StepOnePlus detection system (Applied Biosystems) and SYBR Green PCR Master Mix (Applied Biosystems). Primer sequences for each gene are indicated in Supplementary Table 2, available at Carcinogenesis Online.

Western blotting

Cells were lyzed with E1A lysis buffer (20 mM NaH₂PO₄, 150 mM NaCl, 0.5% IGEPAL, 5 mM ethylenediaminetetraacetic acid, 50 mM NaF, 30 mM sodium pyrophosphate, 10% glycerol) to obtain whole cell extracts. The following antibodies were used to detect protein expression by western blot: p53 (DO-1, sc-126; Santa Cruz), p21 (C-19, sc-397, Santa Cruz) and uPA (AF1310, R&D systems).

RNA-sequencing and bioinformatics analysis

Total messenger RNA (mRNA) was extracted from cells transfected with control or p53-targeted small interfering RNA (siRNA) using the RNeasy Plus Mini Kit (Qiagen) according to the manufacturer’s instructions. Samples with experimental duplicates were sequenced by the Genomics Facility at The Wistar Institute. An Agilent 2100 Bioanalyzer (Agilent Technologies) was used to analyze quality and quantity of mRNA, and DNase-treated samples with RNA integrity number >8 were identified for high-throughput sequencing. Libraries were prepared using the Lexogen mRNA kit and sequencing was performed by NextSeq 500 (Illumina) in a 75 bp paired end run with high output option. Bioinformatics analysis was carried out by the Bioinformatics Facility at The Wistar Institute. RNA-sequencing (RNA-seq) data were aligned using Bowtie2 (36) against hg19 version of the human genome and RSEM v1.2.12 software (37) was used to estimate raw read counts and RPKM using ensemble transcriptome. DESeq2 (38) was used to estimate significance of differential expression between groups pairs. Overall gene expression changes were considered significant if passed false discovery rate (FDR) < 5% thresholds for H2087 cell line. Additional significance criteria of nominal P < 0.05 for H441 and H2110 cell lines were used for overlap. Significance of overlap between cell lines was estimated with hypergeometric test using 18 091 genes detected with at least 10 counts as a reference set. Enrichment analysis was performed with Ingenuity Pathway Analysis (Qiagen, http://www.qiagen.com/ingenuity) using ‘Functions & Diseases’ and ‘Canonical pathways’ options. Only results with the predicted activation Z-score > 3 were considered. Top enriched functions that passed P < 10^–5 significance threshold and selected pathways that passed FDR < 10% thresholds were reported. Relationship of sip53/siControl log₂ ratios for expression from RNA-seq and RT-qPCR that passed at least 1.2-fold threshold were reported. The RNA-seq data were submitted to the NCBI Gene Expression Omnibus database under accession number GSE120534.

ChIP sequencing

ChIP was performed as described previously. Formaldehyde (1%) was added to ~5 × 10⁷ cells for cross-linking, followed by glycine for quenching. Cells were rinsed with phosphate-buffered saline and harvested, lyzed and sonicated (QSonica Q800R2). Technical adequacy of sonication was evaluated by gel electrophoresis to confirm DNA fragmentation to 100–300 bp. The supernatant was subjected to immunoprecipitation overnight with p53 or IgG antibody (FL-393; Santa Cruz; Rabbit IgG sc-2027, Santa Cruz) using magnetic beads (Dynabeads; Invitrogen). Beads were washed and DNA was reverse-crosslinked and purified. Sequencing was performed on an Illumina NextSeq 500, followed by alignment to the hg19 reference genome using Bowtie2. Significant overlapping and differentially bound peaks among p53, IgG, and input conditions for each cell line were identified using HOMER. Overall peaks were considered significant for FDR < 5% with p53 signal >2, fold change over control >, and control signal <1. The ChIP-seq data were submitted to the NCBI Gene Expression Omnibus database under an accession number that is pending.

Proliferation and migration assays

Cell proliferation was measured by 3-(4,5-dimethylthiazole-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay. Cells were plated in triplicate and sterile MTT (Sigma) was added. Absorbance at 562 nm was measured on a plate reader. Cell migration was measured through a standard wound healing assay. A549 and H2087 cells were seeded in 24-well plates at 150 000 cells per well and 200 000 cells per well, respectively, and were fully confluent at 48 h when a scratch was made. Images were taken at regular intervals and scratch diameter was calculated in multiple separate regions as percentage of 100% scratch closure.

Results

The tumor suppressor gene TP53 exhibits a lung-specific pattern of missense mutations

Analysis of human tumors in 147 non-lung cancer studies available on the cBioPortal for Cancer Genomics (34,35) reveals 5972 samples with 8434 somatic alterations in the tumor suppressor gene TP53. Of these alterations, 5461 were missense mutations and occurred most frequently in the DNA-binding domain of TP53 at traditional ‘hotspot’ residues, codons R175, R273, G245, R248 and R282, among others (Figure 1A, top and 1C, left). The TCGA Pan-Lung Cancer study (4), also publicly available, includes 1144 patient samples (660 lung adenocarcinoma and 484 squamous cell lung cancer) of which 776 (68%) had alterations in TP53 (Figure 1B). The frequency of TP53 alterations, including truncating, inframe, and missense mutations in lung adenocarcinoma was 54.2% (358 of 660) and in squamous cell lung cancer was 86.4% (418 of 484). Across both histologic subtypes of lung cancer, 505 missense mutations were found (Figure 1A, bottom). As in non-lung cancers, the most frequent site of missense mutations occurred at the amino acid residue R273, although the frequency was reduced from 10.2% to 6.1% in lung cancers. (Figure 1C). The frequencies of missense mutations at R158 and V157 increased greater than 5-fold, to 5.9% and 4.6%, respectively, making these the second and fourth most commonly mutated residues. Among lung adenocarcinomas and squamous cell cancers, the combined frequency of missense mutations in the LMC was 12.3%. The American Association of Cancer Research Project Genomics, Evidence, Neoplasia, Information, Exchange dataset similarly reveals a relative increase in frequency of V157, R158 and A159 mutations among 6346 non-small cell lung cancer samples compared with non-lung tumors (Supplementary Figure 1, available at Carcinogenesis Online) (39). These studies confirm (5,29–31) that a specific pattern of somatic mutations in TP53 exists in human tumors of the lung.

Figure 1.

Somatic mutations in TP53 exhibit a specific mutational pattern in lung cancers. (A) The TP53 gene is shown here, with transactivation domain (TAD), DNA-binding domain (DBD) and tetramerization domain (TD) labeled. Top, number of mutations at each amino acid out of 5461 total somatic missense mutations in non-lung cancers in the cBioPortal for Cancer Genomics (34,35). Bottom, number of mutations at each amino acid out of 505 total somatic missense mutations in the Pan-Lung Cancer study (4). (B) Oncoprint (34,35) showing cancer and genetic alteration type for the Pan-Lung Cancer study (4), with a TP53 somatic mutation rate of 68%. (C) Frequency of mutations at TP53 hotspots (unshaded) and lung-specific amino acid residues (shaded in blue) among non-lung versus lung cancers.

We examined the frequency of specific nucleotide changes at traditional hotspots and lung-enriched hotspots (Supplementary Table 3, available at Carcinogenesis Online). As expected, given the well-described ‘smoking signature’ of G>T transversions (5), these alterations demonstrated an increased frequency at G245, R248, R273, V157 and R158 in the Pan-Lung Cancer study compared with non-lung cancer studies available on cBioPortal (Figure 2) (4,34,35). In fact, at G245, V157 and R158, greater than 80% of the missense mutations seen were G>T transversions in the Pan-Lung Cancer Study. Accordingly, the proportion of G>A transitions was decreased among samples in the Pan-Lung Cancer study compared with non-lung cancers. In addition, G>C transversions occurred at higher rates at the lung-enriched hotspots 157, 158, and 159 (ranging from 3.9% to 26.3%) compared with traditional hotspots 245, 248 and 273 (ranging from 0.6% to 3.8%) in non-lung cancers. Finally, despite the presence of a guanine within the A159 codon, this nucleotide participated exclusively in G>C transversions (A159P, 44.4%) or C>T transitions (A159V, 55.6%) in the Pan-Lung Cancer Study. Taken together, these data confirm the presence of a G>T smoking signature but also suggest an additional mechanism for tumor progression given the diversity of nucleotide alterations that occur in TP53. This supports our hypothesis that the lung-enriched mutations may, in part, be imparted by biological selection separate from environmental mutagenic mechanisms.

Figure 2.

Lung cancers exhibit single nucleotide mutations in TP53 hotspots in patterns distinct from non-lung cancers. Graphs depict the frequency of nucleotide transversions (G>T, G>C, C>G, C>A, T>A, T>G) and transitions (G>A, C>T) among somatic missense mutations from non-lung cancers in the cBioPortal for Cancer Genomics and from the Pan-Lung Cancer study (4,34,35).

Single nucleotide changes at V157, R158 and A159 are associated with histologic subtype

We also analyzed the clinical characteristics of the 65 patients in the Pan-Lung Cancer study (4) with alterations (including 62 missense and 3 frameshift mutations) at V157, R158 or A159 (Supplementary Table 1, available at Carcinogenesis Online). The mean age of this cohort was 67.0 ± 10.1 years, with 95.4% (62 of 65 patients) current or former smokers. There was a statistically significant association between nucleotide change (G>T versus G>C versus other) and American Joint Committee on Cancer Tumor, Node, and Metastasis (TNM) seventh edition stage groupings (stage I–IV), with a higher incidence of advanced stage lung cancer in the group with G>C transversion at nucleotide 157, 158 or 159 (P = 0.024). A statistically significant association between nucleotide change and tumor histology was also observed, with 62.2% of patients with G>T transversion were found to have squamous cell histology, whereas 87.5% of patients with G>C transversion had adenocarcinoma (P = 0.03). Notably, extent of cigarette smoking history, in pack-years, did not differ among subgroups. These results, while obtained from a small cohort, suggest that specific nucleotide changes at the codons for the mutations enriched in lung cancer be linked to different tumor histology and stage of disease.

Cell lines carrying the LMC exhibit loss of expression of canonical wtp53 target genes

To determine whether the lung-enriched mutant p53 proteins retain any wild-type p53 transactivational function, we carried out qRT–PCR for canonical wtp53 target genes to determine transcript levels for canonical wtp53 target genes in three human lung cancer cell lines: A549 (wild-type p53), H2087 (p53-V157F), H441 (p53-R158L) and H2110 (p53-R158P) (Figure 3A). Transcript levels of wtp53 target genes involved in apoptosis (BAX, NOXA and PUMA) and cell cycle arrest (CDKN1A, or p21), among others, had lower fold induction in mutp53 cell lines even after treatment with the DNA damaging agent cisplatin (Figure 3B). We also noted a lack of induction of p21 at the protein level in mutp53 cell lines relative to cisplatin-treated A549 cells (Figure 3C). These data show that human lung cancer cell lines harboring mutant p53 (V157F, R158L and R158P, specifically), while retaining some wild-type activity, generally induce canonical p53-regulated gene targets to a lower level than that induced by wtp53.

Figure 3.

Human lung cancer cell lines with endogenous mutant p53 exhibit loss of expression of canonical wild-type p53 target genes. (A) qRT–PCR for p53 mRNA expression after depletion via transfection of siRNA targeting p53, with and without cisplatin treatment (10 uM for 24 h) in A549 (wtp53), H2087 (mutp53(V157F)), H441 (mutp53(R158L)), and H2110 (mutp53(R158P)) cells. (B) qRT–PCR was performed for canonical target genes of wtp53 in RNA isolated from A549, H441, and H2110 cell lines. (C) Western blot for p53 and p21 protein expression with and without cisplatin treatment.

Analysis of lung-enriched mutp53-driven transcriptomes identifies differentially expressed genes

To assess the effects of the lung mutants on gene expression patterns, we transfected H2087 (V157F), H441 (R158L) and H2110 (R158P) cells with siRNA targeted against endogenous mutant p53 and performed RNA sequencing (Figure 4A and B). Gene expression in mutant p53-expressing versus mutant p53-depleted conditions were compared within each cell line. The most effect was observed in H2087 (V157F) cells with 1719 genes significantly affected (FDR < 5%) by p53 depletion (Figure 4C). Among this group of genes, 108 were also significantly affected by p53 depletion in H441 (R158L) cells (P < 0.05), a significant overlap (3.42-fold more than expected by chance, P < 10^–13), and 157 were also significantly affected by p53 depletion in H2110 (R158P) cells (P < 0.05), also a significant overlap (3.38-fold more than expected by chance, P < 10^–13). Overall, 51 differentially expressed genes were commonly affected in all three mutant p53 cell lines (Figure 4D, Supplementary Table 4, available at Carcinogenesis Online)

Figure 4.

The p53 mutants V157F, R158L and R158P exhibit differential expression of genes associated with cell migration and invasion. (A) Western blot of cell lysates. (B) qRT–PCR for mRNA expression of TP53 in cells expressing endogenous mutant p53 or depleted of p53. (C) Overlap of differentially expressed genes identified among siCtrl versus sip53 comparisons within each cell line. (D) Heatmap of 53 differentially expressed genes among all three mutant alleles (V157F, R158L and R158P). (E). mRNA expression, in representative experiments, of enriched genes identified by RNA-sequencing. mRNA expression values are normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and A549 siCtrl condition. (F) Selected top biological functions significantly altered by mutant p53 depletion in H2087, H441 and H2110 cells. (G) Selected top signaling pathways significantly altered by mutant p53 depletion in H2087 cells.

Expression at the mRNA level of all 51 commonly altered genes was measured in subsequent experiments. We validated expression changes of nine genes repressed in all three cell lines by mutant p53 at least 2-fold and four genes activated at least 2-fold by mutant p53 with RT-qPCR in independent experiments in A549, H460, H2087, H441 and H2110 cells (Figure 4E, Supplementary Figure 2A, available at Carcinogenesis Online). In addition, from a group of genes most significantly altered (FDR < 1%) in H2087 cells (36 repressed at least 6-fold and 15 activated at least 10-fold by mutant p53, Supplementary Table 5, available at Carcinogenesis Online), we also validated expression changes of 29 genes by qRT–PCR in this panel of cell lines (Supplementary Figure 2B, available at Carcinogenesis Online).

Studies were undertaken to assess whether these target genes were specific to mutant p53 or also regulated by wild-type p53. A number of genes were differentially repressed or activated by mutant p53 when compared with expression in wild-type p53 cell lines—specifically, we found increased expression of EFHD2 and ANXA2 with knockdown of mutant p53 in H2087, H441 and H2110 cell lines, but no change in mRNA level with knockdown of wild-type p53 in H460 cells. A subset of genes, including PLAU, had a greater magnitude of change in mRNA expression in mutant p53 cell lines compared with that in wild-type p53 cell lines. We found reduced expression of CMTM6 and EFCAB14 with knockdown of p53 across all cell lines, suggesting that this gene is activated by p53. In contrast, there was increased expression of TNS4 with knockdown of p53 in A549, H2087 and H441 cells, suggesting that this gene is repressed by both wild-type and mutant p53. Finally, some genes including TULP3 and CDH1 appeared to have differential expression patterns between cell lines. Together, these data suggest that a number of gene targets may be commonly affected by all lung-specific p53 mutations, whereas others may be mutation- or cell line-dependent.

V157F and R158P mutant p53 lack the ability to directly bind DNA

To determine whether lung-enriched mutant p53 regulates the transcriptome changes identified in our RNA-seq by directly binding these candidate target genes, we performed ChIP sequencing in H2087 (V157F mutp53) and H2110 (R158P mutp53) cells using p53 antibody and IgG as a control. We confirmed binding of the polyclonal p53 antibody to mutant p53 and DNA binding by wild-type p53 at canonical target genes including CDKN1A by ChIP-qPCR in ChIP experiments performed in parallel in the wild-type p53 cell line A549 (Supplementary Figure 3A and B, available at Carcinogenesis Online). Upon sequencing, 93% of reads aligned to hg19. All identified peaks were very low relative to expected binding by wild-type p53 at canonical p53 target genes, with fold-change (over IgG control) signal <5 and uniform distribution across the genome. No overlap was seen between differential peaks among H2087 cells versus H2110 cells (Supplementary Figure 3C, available at Carcinogenesis Online). Moreover, no binding was seen in gene-proximal regions for any of the 51 activated or repressed genes identified through RNA-seq (Supplementary Figure 3D, available at Carcinogenesis Online). Taken together, these findings suggest that V157F and R158P mutant p53 do not directly bind DNA as a mechanism for regulating gene expression changes.

Biological processes repressed by mutp53 may include cell migration and invasion

We carried out enrichment analysis of the differentially expressed genes to identify biological functions and pathways associated with genes regulated by mutant p53 using Ingenuity Pathway Analysis. First, a set of 382 significantly altered genes (FDR < 5% from a combined statistical model including H2087, H441 and H2110 cells) was analyzed. Across all three cell lines, the top associated biological functions repressed by mutant p53 included cell viability, cell migration and cell invasion (Figure 4F). Analysis of pathways associated with significantly altered genes in H2087 (V157F) cells only (1719 genes, FDR < 5%) revealed repression of integrin signaling, RhoA/Rac signaling, and vascular endothelial growth factor signaling (Figure 4G). The significance of these paradoxical data, which run counter to the pro-invasive GOF phenotypes imparted by the canonical p53 mutations, will be addressed in the Discussion.

To extend our functional analysis, we evaluated cell migration in lung cancer cells expressing wild-type or lung-enriched mutant p53. Migration levels were compared with those of cells depleted of p53. We first confirmed that loss of p53 did not alter cell proliferation (Figure 5A). A wound healing assay performed in A549 (wild-type) and H2087 (V157F) cells showed significantly increased migration upon depletion of mutant p53, whereas no difference in scratch wound closure was observed in cells harboring wild-type p53 (Figure 5B and C). These findings are in line with our transcriptome analysis, suggesting that lung-enriched mutant p53 ‘represses’ cell migration.

Figure 5.

V157F mutp53 impedes migration of lung cancer cells. Cells harboring endogenous wild-type or lung-enriched mutp53 was transfected with siRNA targeting p53. (A). Cell proliferation, measured by MTT assay, in A549 (wild-typet), H2087 (V157F), H441 (R158L) and H2110 (R158P) cells. (B and C) Cell migration in a wound healing assay in A549 and H2087 cells.

Urokinase-type plasminogen activator, a regulator of the plasminogen activation system, is regulated by mutant p53

To determine whether the p53-regulated changes in mRNA expression are also present at the protein level, we transfected the following human lung cancer cell lines with siRNA directed against p53 or with a control siRNA: A549 (wild-type p53); cell lines with endogenous lung-enriched p53 mutants V157F (H2087), R158L (H441) and R158P (H2110); and the canonical p53 mutant R273L (H2009). Protein expression of urokinase-type plasminogen activator (uPA), encoded by the gene PLAU, was assessed by immunoblotting of cell lysates from human lung cancer cell lines expressing endogenous wild-type or mutant p53, or depleted of p53. In A549 cells expressing wtp53 compared with those where p53 had been depleted, we observed uniformly low protein expression of uPA. In H2087 cells (V157F), uPA expression increased with depletion of mutant p53 by siRNA (Figure 6A, D and E). No p53-dependent change was seen in uPA expression in H2009 cells (R273L) despite a similar relative quantity of uPA protein to that of H2087 cells. qRT–PCR showed similarly increased mRNA expression of PLAU with repression of mutant p53 in H2087 cells (Figure 6B). uPA, a key regulator of the extracellular plasminogen activation system (PAS), is secreted for extracellular proteolysis. To elucidate whether secreted uPA is regulated by mutant p53, cells expressing or depleted of p53 were incubated overnight in low-serum media. Conditioned media harvested from H2087 (V157F) cells showed an increase in uPA secretion with depletion of mutp53 (Figure 6C).

Figure 6.

uPA protein expression is mutp53-dependent. Protein expression of uPA encoded by the RNA-seq identified gene PLAU, was assessed in p53-depleted and control conditions. (A) Western blot of A549, H2087, H441, H2110 and H2009 cells transfected with negative control or siRNA directed against p53 (25 nM). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) is shown as a loading control. (B) qRT–PCR was performed to evaluate mRNA expression of PLAU after depletion of p53 by siRNA. mRNA expression values are normalized to GAPDH and siCtrl condition within each cell line. (C) Protein and conditioned media were harvested from cells after overnight incubation in reduced-serum conditions. Nitrocellulose membranes were stained with Ponceau S dye to visualize protein bands for loading control. Protein expression of pro-uPA (49 kDa) and its cleaved and activated forms B-chain (33 kDa) and A-chain (19 kDa) are shown in cell lysates and conditioned media. (D) Protein and E. mRNA expression of uPA (PLAU) following depletion of V157F mutp53 by siRNA.

Discussion

The mutant p53 field has increasingly focused on acquisition of GOF phenotypes, with in vitro and murine knock-in models confirming the ability of mutant p53 to drive enhanced invasion and tumor progression when harboring mutations at the canonical hotspots R175, R273 and R248, among others (24,26,40,41). Additional GOF activities described in vitro include chemoresistance and regulation of cancer metabolism, also driven by the traditionally altered mutp53-R175H, -G245C and -R282W (24,42). To the best of our knowledge, the LMC hotspots V157, R158 and A159 have not been studied as a group for mutp53-dependent GOF phenotypes specific to lung cancer. Recent studies of the global landscape of TP53 mutations, however, have allowed prediction of structural and functional impact among alterations in p53 which are less prevalent across all tumor types (27,43,44). Giacomelli et al. (43)demonstrated that the TP53 mutational spectrum reflects the impact of both tissue specificity and functional advantage conferred by mutp53. Similarly, Kotler et al. (44) showed that TP53 mutation prevalence in human tumors (specifically among the 10 most frequent hotspots) was correlated with a ‘relative fitness score’ (RFS) based on mutation retention in an in vitro lentiviral infection model. However, there were additional mutations with equally elevated RFS which occurred with a very low frequency. The authors reported that V157F (the 12th most frequent mutation in the IARC database, with an RFS in the top 7%), R158H/L/P, and A159P all exhibited a high RFS (>−1), predicting loss of wtp53 anti-proliferative effect. Furthermore, the authors found no proliferative advantage among mutations with a high RFS in traditional cell culture or spheroid culture models despite enrichment for these mutations in in vivo heterotopic mouse tumors (44). These results suggest that although large-scale systematic screens may identify functional advantages favoring selection of hotspot mutants, more detailed studies are necessary. Specifically, the less prevalent mutations should be interrogated for direct functional measurement using a context-dependent approach.

GOF activities are in some cases initiated through interactions between the mutant p53 protein and other transcription factors that increase or reduce the transactivation of gene targets (45). Common cofactors and effector proteins have been identified for several mutants in genome occupancy studies, suggesting that a core set of target genes may drive common GOF processes. For example, p53-R175H and -R273H bind to common transcription factors including PML and SP1, among others, and may transactivate similar sets of target genes (28). However, murine models have demonstrated varying tumor spectra depending on expression of different mutant p53 proteins, and it has been proposed that the mutation position and nature (contact mutants which alter amino acids critical for DNA binding or structural mutants which cause unfolding of the p53 protein) may affect the spectrum of factors interacting with mutant p53, leading to mutation- or context-specific phenotypes (28). Our ChIP-seq findings suggest that lung-enriched mutp53 regulates the GOF transcriptome through mechanisms other than direct DNA binding. Additional mechanisms by which mutant p53 regulates GOF phenotypes are through indirect DNA binding via mutp53 interactions with other proteins, chromatin remodeling and upregulation of proteasome levels (46,47). These studies raise the possibility that there are lineage-dependent and allele-specific roles for p53 mutations.

We have shown here that specific p53 mutations occur more frequently in human lung cancer compared with other smoking-related solid tumors, suggesting that they may undergo biological selection and confer an oncogenic phenotype in lung cancer cells. We also found that specific nucleotide changes are associated with differences in tumor histology, with G>T transversions significantly more common in squamous cell lung cancer. Although this difference appears to be independent of smoking status and pack-years of smoking, we did not assess other potential confounders. These clinical findings require confirmation in larger cohorts but may raise the possibility that DNA repair mechanisms may differ among specific cell types. Furthermore, we have demonstrated that the mutant p53 harboring V157F and R158L/P mutations regulates a novel transcriptome in human lung cancer cells. In particular, the serine protease uPA is repressed by mutp53-V157F, and depletion of the traditional hotspot mutp53-R273L has no effect on uPA expression in the lung cancer cell line H2009. This suggests an allele-specific consequence of V157F in lung cancer, although differences in genetic background between the H2087 and H2009 cell lines may confound these results. Repression of uPA by V157F mutp53 represents one example of an overall transcriptional program which may inhibit cell migration in lung tumors harboring V157 or R158 mutant p53. Much additional work is needed to define the pathways and molecular mechanisms regulating this phenotype.

uPA is a key regulator of the PAS. The uPA molecule is secreted as a pro-enzyme that upon activation converts plasminogen to plasmin, which in turn proteolytically cleaves and activates pro-uPA through a positive-feedback loop (48). Plasmin also cleaves extracellular matrix (ECM) components, degrades fibrin blood clots and activates matrix metalloproteases to facilitate breakdown of basement membranes to promote cell migration (49). This extracellular proteolytic cascade, when activated, coordinates ECM cleavage, cell-ECM interactions, and cell signaling through the uPA receptor, activating the Ras–mitogen-activated protein kinase pathway, the tyrosine kinases focal adhesion kinase and Src, and the Rho family small GTPase Rac (50). The uPA protein is highly expressed in inflammatory states and solid tumors of the prostate, breast and lung, among other cancer types, where high expression is frequently associated with poor clinial prognosis (48,51). Inhibition of uPA activity has been shown to diminish tumor growth, angiogenesis, and metastasis in prostate and lung cancer cells (52–54).

Activation of the PAS and its regulatory components uPA and uPAR is considered oncogenic, given their ability to cleave the ECM and induce cell motility and invasion. Paradoxically, we found that uPA is ‘repressed’ by mutant p53 in lung cancer, thereby leading to enhanced stability of the ECM. Other investigators have noted that matrix architecture and fibrotic tumor microenvironments might subvert tumor immunity in part through chemokine alteration or physical exclusion of T-cell infiltration, allowing for tumor immune evasion in the lung (55,56). Increased ECM stiffness may also influence other components of the tumor microenvironment including activation of cancer-associated fibroblasts and tumor-associated macrophages, or lead to hypoxia and depletion of micronutrients, thereby regulating metabolic reprogramming and enhancing tumor survival (57,58). In line with these mechanisms, our pathway analysis revealed repression of genes associated with cell migration and invasion upon expression of lung-enriched mutant p53.

Notably, uPA is inhibited by plasminogen activator inhibitor 1 (PAI1, also known as SERPINE1), which is a target gene of wild-type p53 and induced via phosphorylation at Ser 15 (59). Specifically in the lung, Shetty et al. have shown that wild-type p53 binds in a sequence-specific fashion to the 3′ untranslated region of uPA, uPAR and PAI-1 mRNA, leading to increased mRNA turnover and decreased cellular uPA, uPAR and PAI-1 expression in lung epithelial cells and in p53-null lung cancer cells expressing exogenous wild-type p53 cDNA (60–62). This is a possible mechanistic explanation for the effects we have reported here in lung cancer cells harboring mutant p53.

In summary, our findings are the first to describe cellular alterations regulated by the lung-enriched mutant p53-V157 and R158 alleles. These data support the hypothesis that the lung-preferred mutations may confer novel and lung-specific gain of function. Further mechanistic studies will be critical to understanding whether mutant p53 affects PLAU transcript through direct interaction with mRNA or through its more traditional role as a transcription factor.

Funding

This study was supported by NCI R01 CA164834 (SBM) and the American Cancer Society 1300042-IRG-16-244-10 (JAB).

Glossary

Abbreviations

ATCC

American Type Culture Collection

ECM

extracellular matrix

GOF

gain-of-function

LMC

lung mutant cluster

mRNA

messenger RNA

PAS

plasminogen activation system

RFS

relative fitness score

RNA-seq

RNA-sequencing

siRNA

small interfering RNA

uPA

urokinase-type plasminogen activator