Introduction

The p53 tumor-suppressor protein, encoded by the TP53 gene, is a key transcription factor that regulates the expression of several cellular stress response genes involved in processes that include DNA repair, cell cycle arrest, apoptosis and senescence.1, 2 However, approximately 50% of all human cancers harbor missense mutation in TP53 tumor-suppressor gene giving rise to the full-length mutated p53 proteins with single-amino-acid substitution.3 Missense mutation in the p53 protein frequently occur at six discrete hotspot codons within the DNA-binding domain: codons R248, R175, G245, R249, R273 and R282.4 Such mutated p53 proteins not only lose their tumor-suppressive functions, but also gain additional oncogenic properties by a gain-of-function mechanism, which can endow the incipient cancer cells with migratory, invasive and metastatic properties.5, 6 Interestingly, genetically engineered mice that harbor missense p53 mutations develop more aggressive and metastatic tumors in vivo, compared with p53 null or heterozygous mice,7, 8, 9, 10 suggesting that the gain-of-function activity is a characteristic of mutant p53. In humans, the gain-of-function hypothesis is further supported by the association between specific p53 mutations with poor patient prognosis in number of cancers, including those of breast,11, 12 colon13, 14 and lung.15, 16

As p53 mutations are predominantly found in the early invasive stage of lung carcinomas,17 the identification of key regulatory pathways that mediate the mutant p53-induced invasion may uncover new targets for cancer therapy. Although pre-clinical compounds like PRIMA-1 and PRIMA-MET that specifically target mutant p53 tumors have been extensively studied, the target specificity of these compounds remains to be further defined.18, 19, 20, 21 Hence, as a possible alternative approach to successfully target mutant p53 tumors, efforts are now being focused on exploiting the underlying downstream targets through which mutant p53 mediates oncogenic properties. Indeed, such an approach has identified several proteins involved in mevalonate pathway,22 integrin recycling23 and microRNA biogenesis24 as an essential mediator of mutant p53-driven oncogenic function.

Although mutant p53-regulated genes and their effect in invasion have been extensively studied,25 the global impact of mutant p53 on cancer cell proteome are unclear. A possible mechanism of mutant p53-driven oncogenic activity is via regulation of secreted proteins that functions in cell autocrine/paracrine signaling. In fact, the study on mutant p53 proteome of breast cancer cell lysates investigated using inducible knockdown of endogenous mutant p53 R273H have reported a transcription-independent role for mutant p53 in regulating key proteins that are involved in DNA replication and repair pathways.26 In addition, three independent proteomic studies evaluating mutant p53-driven alterations in cancer cell models and embryonic stem cells using whole-cell lysates have identified factors involved in regulating cell invasion and conformational reversion of mutant p53.27, 28, 29 Similarly, the mutant p53-induced extracellular proteome also called ‘secretome’ may contain signal transduction targets of mutant p53 that can offer both therapeutic and diagnostic benefit. Indeed, our laboratory has previously highlighted a role for mutant p53 in the induction of invasive cancer cell secretome.30 Similarly, previous studies have also reported the involvement of mutant p53 in the regulation of cancer cell secreted enzymes, proteins and inflammatory cytokines that alter the tumor microenvironment.31 However, the impact of mutant p53 on cancer cell secretome is still unclear.

In this study, we used quantitative proteomics approach using conditioned medium from inducible EI-H1299 (non-small cell lung cancer (NSCLC)) cell lines expressing p53 R175H or R248Q mutants,32 to identify and characterize the mutant p53-induced extracellular factors. Our findings discovered a significant overlap of mutant p53-induced secreted proteins with its corresponding gene targets and a robust increase in expression of alpha-1 antitrypsin (A1AT). We demonstrate that A1AT is a key target of mutant p53 that drives in vitro and in vivo invasion in lung cancer cells via induction of epithelial–mesenchymal transition (EMT) markers expression. Furthermore, elevated A1AT expression is prognostic in adenocarcinoma (ADC) of the lung and a potential therapeutic target for treatment of mutant p53-expressing tumors.

Results

Proteomic quantification of mutant p53-induced secretome identifies A1AT as a downstream mediator

Although mutant p53 has widely been shown to regulate cellular invasion, it is not clear whether this is primarily mediated through intracellular means or it also involves the extracellular secreted factors. Hence, to explore the possibility that mutant p53 enhance invasion through secreted factors, we used conditioned medium from EI-H1299-p53 null human NSCLC cell line, which has been engineered to inducibly re-express various p53 mutants following treatment with the ecdysone analog, ponasterone A.32 Incubation of p53 null (H1299) and p53 wild-type (SBC-3) lung cancer cell lines with the conditioned medium from cultures of EI-H1299 with induced expression of p53 R248Q mutant enabled the cells to acquire the invasive potential as compared with the conditioned media from the un-induced EI-H1299-R248Q cells (P<0.005; Figure 1a), indicating the presence of invasive extracellular factors in mutant p53-induced secretome.

Figure 1
figure 1

Identification of A1AT as a secreted target of mutant p53. (a) EI-H1299 cells with inducible expression of p53 mutant R248Q was cultured in the presence of ponasterone A (PonA; 2.5 μg/ml) or vehicle control (ethanol) for 96 h. Independent cultures of H1299 or SBC-3 were grown in a dilution (50:50) of this conditioned media (CM) and their invasion into the matrigel was quatified as described in the Materials and Methods section. Data presented as means±s.e.m. from three independent replicates. **P<0.005. Representative z-stack images of invading cells are shown (right). (b) Schematic workflow of iTRAQ liquid chromatography tandem-mass spectrometry (LC-MS/MS) quantitative proteomics performed. (c) Two-way Venn diagram representing the number of proteins that were significantly activated or repressed between two p53 mutants (R175H and R248Q), with criteria of P<0.05 and fold change >1.6 or <1.6. Differentially expressed proteins were further compared against their genes identified in previous microarrays30 and two commonly activated proteins were identified whose expression was determined by qRT–PCR. (d) Ingenuity pathway analysis (Ingenuity Systems, http://www.ingenuity.com). Bars (P<0.05) representing molecular and cellular functions significantly altered (activated and repressed) in cell secretome following mutant p53 induction. (e) Real-time qRT–PCR and western blot analysis of A1AT secreted protein levels in conditioned media and p53 expression in the panel of inducible p53 mutant cell lines. In all, 10 μg of protein samples were loaded on 8% polyacrylamide gels. β-Tubulin was used as a loading control.

Full size image

Using isobaric tag for relative and absolute quantification (iTRAQ) based quantitative proteomic approach, we set out to identify mutant p53-induced secreted factors in the conditioned medium of EI-H1299 NSCLC cell lines expressing p53 R175H or R248Q mutants (Figure 1b). When R175H was expressed, a total of 49 proteins were significantly up or downregulated out of 85 identified proteins (58%), compared with the un-induced control (with criteria of P <0.05 and fold change >1.6 or <1.6; Figure 1c; Supplementary Table 1). Similarly, when R248Q was expressed, 46/106 (43%) identified proteins were significantly up or downregulated compared with the un-induced control. Ingenuity pathway analysis revealed that ~50% of the significantly altered (activated and repressed) proteins in the secretome of induced EI-H1299 expressing p53 R175H and p53 R248Q mutants belonged to the functional class of ‘cellular movement’ (Figure 1d; Supplementary Table 2) consistent with our experimental observation in Figure 1a, which implicates for a role of mutant p53 in the induction of invasive secretome.

Using a panel of inducible p53 mutant H1299 cell lines, our laboratory previously performed an expression microarrays to decipher the global gene regulation network of mutant p53.30 In this study, p53 R175H expression was associated with activation of 44 genes and repression of 6 genes (P<0.05; 1.6-fold) compared with un-induced control, whereas R248Q induction activated 91 genes and repressed 8 genes. Reassuringly, the majority of the secreted proteins identified in mutant p53-induced secretome overlapped with the mutant p53 gene targets from expression array analysis (Supplementary Figure S1A), indicating that the mutant p53 modulates its extracellular environment through changes in the expression of its gene targets. Comparing the proteins identified in secretome of two inducible p53 mutants identified 2 activated and 10 repressed proteins in common (Figure 1c). Interestingly, the comparison between proteins detected in the secretome versus genes showing expression changes following induction of mutant p53, showed no commonly repressed secreted proteins/genes (Supplementary Figure S1B). However, two proteins and their transcripts, BIGH3 and A1AT, were commonly and abundantly increased in both the secretome and expression array, indicating that these proteins could be critical effectors for mutant p53-driven oncogenic function (Figure 1c; below; Supplementary Figure S1B). BIGH3, encoded by TGFBI (Transforming growth factor beta induced), is an extracellular matrix protein that modulate cell adhesion;33 and A1AT encoded by SERPINA1, is a secreted serine protease inhibitor that neutralizes the effect of proteases.34 On average, both induced mutants EI-H1299-R175H and EI-H1299-R248Q increased BGH3 transcript and protein levels by 4.2-fold and 2.3-fold compared with the un-induced control, whereas A1AT gene and protein abundance was highly increased by 25.2-fold and 12.3-fold, respectively (Figure 1c).

Recent studies have linked A1AT expression to invasion and metastasis in several human cancers,35, 36, 37 however, to date no studies have linked A1AT expression in cancer with the presence of mutant p53. As A1AT protein was most abundantly and commonly secreted by both p53 mutants, with an average of >10-fold increase upon mutant p53 induction, we hypothesized A1AT may mediate oncogenic function of mutant p53 in lung cancer. First, we verified by quantitative real-time PCR (qRT–PCR) and western blotting that the expression of A1AT mRNA and protein levels were upregulated following induction of mutant p53. Expression of A1AT was consistently upregulated upon induction of eight different hotspot p53 mutants: R249S, R175H, R282W, G245S, R248W, R248Q, R273H and R273C, in EI-H1299-inducible cell lines (Figure 1e). The expression of A1AT and other identified mutant p53 targets (PLK1, DKK1, LAMC2, SFRP1, DDIT4 and BHLHB3) were also confirmed to be significantly higher in inducible p53 mutant cell lines compared with the expression in induced EI-H1299 cell line with no mutant p53 (Supplementary Figures S1C and D).

A1AT mediates mutant p53-induced invasion in human lung cancer cells

As mutation in TP53 is considered an early event in lung cancer progression, we speculated that A1AT may be a downstream mediator of mutant p53-driven oncogenic function. Indeed, the expression profiling data from the Cancer Cell Line Encyclopedia in a panel of 162 different lung cancer cell lines38 revealed that A1AT mRNA levels were significantly higher in lung cancer cells expressing p53 missense mutations compared with those with an intact or non-functional wild-type p53 allele (P<0.05; Supplementary Figure S2A). Mutant p53 protein with hotspot mutations at codons: 285, 163, 162, 141, 245, 47, 273 and 157 had the highest expression of A1AT mRNA (>10-fold) compared with the wild-type or non-functional p53 allele. We next examined if A1AT expression was also regulated by endogenous mutant p53. Knockdown of endogenous p53 R273L in H2009 lung ADC cells resulted in a significant reduction in expression of A1AT mRNA and protein levels, indicating a role for mutant p53 in aberrant activation of A1AT (P<0.05; Figure 2a).

Figure 2
figure 2

A1AT enhances mutant p53-mediated cell migration and invasion concomitantly with EMT markers alteration. (a) qRT–PCR and western blot analysis for A1AT and p53 in H2009 cells with two p53 siRNA (si.p53-1 and si.p53-2) or control (si.Ctrl). In all, 20 μg of protein samples were loaded on 8% polyacrylamide gels. Data presented as mean±s.e.m. from three replicates. *P<0.05. (b) H2009 cells transfected with si-p53-2 or si.Ctrl were placed onto matrigel for 72 h, and invasion was analyzed by confocal microscope and quatified as described in Materials and methods section. Data presented as means±s.e.m. from three replicates. **P<0.005. (c) The invasion of H2009 cells stably expressing control (sh.ctrl) or a shRNA against A1AT (sh.A1AT) was measured and quatified. Data presented as means±s.e.m. from three replicates. **P<0.005. (d) quantification of wound confluence in scratch wound assays at 0 and 24 h after wounding of human H2009 cells stably expressing a non-targeting control (sh.ctrl) or a shRNA targeting A1AT (sh.A1AT) (right). Data presented as mean±s.e.m. **P<0.005. Representative phase contrast images from the live cell recordings of two conditions are shown at 0 and 24 h (below). (e) qRT–PCR of A1AT mRNA expression in double inducible EI-H1299 p53 R248Q cells transfected with doxycycline (2 μg/ml) inducible shRNA targeting A1AT. Induction of p53 protein was determined by western blot analysis. β-Tubulin was used as a loading control (left). (f) Invasion (left) and migration (left) assay analysis of double inducible EI-H1299 R248Q sh.A1AT cells. Data presented as means±s.e.m. from three replicates compared with un-induced. **P<0.005, *P<0.05. (g) Chorio-allantoic membrane (CAM) invasion assay was performed using H2009 cells expressing sh.Ctrl or sh.A1AT. The layers ectoderm (ECT), mesoderm (MES) and endoderm (END) of CAM, 5 days post implantation is evident. Hematoxylin and eosin and pan-cytokeratin IHC of the CAM with transfected cells are shown (right). Quantification analysis of invasion of H2009 cells with sh.Ctrl or sh.A1AT into the CAM (left). Data presented as means±s.d. from indicated replicates. **P<0.005. (h) qRT–PCR analysis of mesenchymal markers expression in H2009 expressing sh.ctrl or sh.A1AT (right). Data presented as a fold change in expression relative to cells expressing scrambled RNA sequence (sh.ctrl), which has been set at 1. The protein level of A1AT and vimentin were determined by western blot analysis (left) using 10 μg of protein samples.

Full size image

A1AT mediates mutant p53-induced cell migration and invasion in vitro

Mutant p53 is associated with invasive properties in vivo, which prompted us to investigate whether A1AT functions in promoting mutant p53-driven malignant transformation. Using inverted invasion assay, the invasion of H2009 cells that expresses an endogenous p53 mutant R273L was confirmed to be dependent on the expression of mutant p53 (P<0.005; Figure 2b). Consistent with this observation, the shRNA (A1AT-short hairpin RNA) mediated A1AT knockdown also significantly decreased the ability of H2009 cells to invade through matrigel (P<0.005; Figure 2c). This was further confirmed using a scratch wound assay where wound closure, reflecting cellular migration was significantly decreased in A1AT knockdown cells compared with its control (P<0.005; Figure 2d).

To further confirm whether A1AT specifically mediates mutant p53-driven invasion, a double inducible EI-H1299-R248Q cells with inducible sh.A1AT cell line was constructed. Knockdown of A1AT is facilitated through the treatment with doxycycline, whereas induction of mutant p53 is achieved through addition of ponasterone A (Figure 2e). Treatment of this cell line with ponasterone A triggered an increase in mutant p53-dependent migration (P<0.005; Figure 2f, right) and invasion (P<0.005; Figure 2f, left). Remarkably, this response was significantly attenuated when A1AT was silenced upon treatment with doxycycline. Importantly, knockdown of A1AT did not alter the basal level of migration and invasion in the absence of induced mutant p53. Furthermore, silencing of A1AT in H2009 cells had no effect on cellular proliferation (P<0.05; Supplementary Figure S2B).

A1AT drives chick chorio-allantoic membrane invasion in vivo

Next, we used the chorio-allantoic membrane invasion assay to investigate whether mutant p53-induced A1AT expression enhances the lung cancer cell invasion in vivo. H2009 cells with stable knockdown of A1AT or control were mixed with matrigel and placed onto the chorio-allantoic membrane of 11-day-old chick embryos. The invasion of cells through the ectoderm into the mesoderm was assessed by immunohistochemical analysis using a pan-cytokeratin antibody (Figure 2g). Consistent with our in vitro observation, A1AT knockdown significantly reduced the invasive potential of H2009 cells (P<0.005; Figure 2g). Collectively, these findings indicate that A1AT is a key downstream target of mutant p53 that promotes cellular invasion.

A1AT alters EMT marker expression

Next, we examined whether A1AT induced suppression of migration and invasion involved alteration in expression of key drivers of mesenchymal transformation. Indeed, silencing of A1AT expression in H2009 cell line resulted in significant repression of several key target genes, including TGFB1, TGFBR2, SNAI1, SNAI2, ZEB1, FN, VIM and TNC (P<0.05; Figure 2h; Supplementary Figure S2C). Collectively, these observations indicate that elevated A1AT levels promote cellular transformation and invasion by altering key drivers of metastasis.

Secreted A1AT enhances mutant p53-induced cell migration and invasion

To determine whether A1AT was secreted by mutant p53-expressing lung cancer cell line, the conditioned medium was collected and concentrated by ultra centrifugal filter device. Consistent with the observation in A1AT mRNA expression profiling (Supplementary Figure S2A), A1AT protein was abundantly secreted in p53 mutant (H2009) cells compared with p53 wild-type or null cells (Supplementary Figure S3A). To investigate whether secreted A1AT was a driving factor, H2009-sh.A1AT cells was incubated with the conditioned media from H2009-sh.Control cells and the degree of invasion was assessed using inverted invasion assay. Conditioned media harvested from H2009-sh.A1AT contained less secreted A1AT (Figure 3a) and induced lesser invasion through matrigel, compared with conditioned media from H2009-sh.Control cells (P<0.005; Figure 3b). In addition, the loss of migratory ability of H2009 cells following A1AT knockdown was rescued by the growth in concentrated conditioned medium from H2009-shControl cells in a dose-dependent manner (P<0.05; Supplementary Figure S3B). Furthermore, the incubation of SBC-3 and H1299 cells with concentrated conditioned medium from induced EI-H1299 R248Q cell line showed increased cell invasion compared with un-induced counterparts (P<0.05; Supplementary Figure S3C). Interestingly, treatment of H2009 (p53 R273L; P<0.005; Figures 3c and d) and EI-H1299 R248Q cells (P<0.005; Figures 3e and Supplementary Figure S3D) with an A1AT-blocking antibody attenuated the mutant p53-induced cell migration and invasion. Collectively, these observations support the notion of A1AT as a critical mediator of mutant p53-driven migration and invasion.

Figure 3
figure 3

Blockade of secreted A1AT attenuates mutant p53-induced cell migration and invasion. (a) Western blot analysis of A1AT secreted protein in conditioned medium from H2009 cells expressing sh.Ctrl or sh.A1AT. (b) Invasion assay analysis following incubation of H2009-sh.A1AT cells with the conditioned media from H2009-sh.ctrl or sh.A1AT transfected cells. Data presented as mean±s.e.m. from triplicate experiments. **P<0.005. (c) Invasion assay analysis and migration (left) following incubation of H2009 cells with the A1AT1-blocking antibody (40 μg/ml A1AT IgG) or control IgG. Data presented as mean±s.e.m. from triplicate experiments. **P<0.005. (d) Migration assay analysis following incubation of un-induced or induced EI-H1299 R248Q cells with the A1AT-blocking antibody (A1AT IgG) or control IgG. Data presented as mean±s.e.m. from triplicate experiments. **P<0.005. (e) Invasion assay analysis following incubation of un-induced or induced EI-H1299 R248Q cells with A1AT-blocking antibody or control IgG. Data presented as means±s.e.m. from triplicate experiments. **P<0.005, ***P<0.0005.

Full size image

Mutant p53 interacts with p63 to regulate A1AT expression

Mutant p53 exerts its oncogenic function through interaction with other transcription factors such as ETS proto-oncogene 1 (ETS1), vitamin D receptor (VDR) and nuclear transcription factor Y subunit alpha (NFYA), and also through the suppression of its own family members p63 and p73.30, 39, 40, 41 Furthermore, the genes activated by mutant p53 were shown to frequently share promoter sequences with p63 and wild-type p53.30 Interestingly, A1AT (SERPINA1 gene) was also shown to have promoter-binding sites for p63,42 which prompted us to examine a role for p63 in mutant p53-regulated A1AT expression. In silico analyses identified three putative p63 response elements (REs) in the regulatory elements of A1AT gene (Supplementary Figure S4A). Next, we investigated if endogenous p63 could associate with the identified REs in the A1AT gene. Chromatin immunoprecipitation (ChIP) analyses demonstrated that knockdown of endogenous p63 significantly reduced the amount of p63 bound to p63 REs of A1AT gene in H2009 cells (P<0.05; Figure 4a). In addition, knockdown of p63 in H2009 cells resulted in a significant decrease in expression of A1AT, implicating A1AT as a target for constitutive upregulation by p63 (Figure 4a, right).

Figure 4
figure 4

Mutant p53 regulates A1AT gene expression through p63. (a) H2009 (p53 R273L) cells were transfected with either p63-specific siRNA (si-p63) or non-targeting control (si.Ctrl) and subjected to ChIP analysis using either a p63-specific antibody or IgG control. The relative expression of A1AT was determined by qRT–PCR analysis (right). p63 knockdown was confirmed using western blot analysis (right). Data represented as mean±s.e.m. *P<0.05, **P<0.005. (b) EI-H1299 cells with inducible p53 R175H, R248Q or R282W were cultured in the presence of ponasterone A (PonA; 2.5 μg/ml) or vehicle control for 24 h before ChIP analysis using a p53-specific antibody. Fold enrichment of a putative p53 RE within the A1AT gene (as compared with un-induced control) was determined for three independent experimental replicates. The p53 RE within PLK2 promoter was used as a positive control. *P<0.05, **P<0.005 (right). In addition, cells were lysed with 1 × lysis buffer as described in Supplementary Materials and Methods under IP section. Expression of p53 was determined by immunoblotting 3 μl of inputs with anti-p53 antibodies (mAb DO-1, Santa Cruz (Santa Cruz Biotechnology, Inc., Dallas, TX, USA); top right). (c) H2009 (p53 R273L) cells were transfected with either p63-specific siRNA (si.p63) or non-targeting control and was subjected to ChIP analysis using either a p53-specific antibody or IgG control. Fold enrichment of a putative p53 and p63-RE within the A1AT gene (as compared with IgG control) was determined. **P<0.005. (d) Diagram representing the mechanism of action of mutant p53 in regulating A1AT expression.

Full size image

Based on these results, we speculated that mutant p53 is recruited to these REs through its interaction with p63. Data from ChIP analyses were consistent with this hypothesis, as induced p53 mutants R175H, R248Q and R282W were found to be associated with the identified p63 REs of A1AT gene in EI-H1299 cells (P<0.05; Figure 4b). The human PLK2 promoter served as a positive control, as it has been previously shown that p53 mutants can directly bind this promoter and drive transcription.30 These findings provide firm evidence that mutant p53 and p63 are co-recruited to these p63 REs. Furthermore, silencing of p63 in H2009 cells resulted in a complete disassociation of the endogenous mutant p53 (R273L) from the REs of A1AT (P<0.005; Figure 4c). Collectively, these data suggest that mutant p53 recruits p63 as a molecular chaperone to regulate A1AT expression (Figure 4d).

A1AT is also a target of wild-type p53 with no possible role in p53-mediated cell cycle function

Mutant p53 transcriptional targets are nested within a small subset of wild-type p53-responsive genes.30 Thus, we sought to examine whether the regulation of A1AT was restricted to mutant p53. Indeed, induction of wild-type p53 in EI-H1299 cells significantly upregulated the expression of A1AT (Figure 5a). Although the basal level of A1AT mRNA expression was low in p53 wild-type cell (SBC-3; Supplementary Figure S2A), treatment with the p53 activator, nutlin-3a (10 μm) significantly upregulated A1AT mRNA levels (Figure 5b). This increase following nutlin-3a treatment was directly related to the activation of wild-type p53 as increased A1AT mRNA expression was not observed in p53 null (H1299) or p53 mutants cell lines; H1466 (frame shift; codon 89) and H2009 (missense; R273L; Figures 5b and c). Nutlin-3a treatment was previously shown to induce the expression of mutant p53 in osteosarcoma cells,43 however, our data show that nutlin-3a have no effect on the expression of mutant p53 in H2009 lung cancer cell line. Furthermore, treatment of the p53 isogenic cell lines HCT116 p53−/− and HCT116 p53+/+ with nutlin-3a upregulated the A1AT expression in HCT116 p53+/+ but did not change A1AT expression in HCT116 p53−/− cells (Supplementary Figure S5A).

Figure 5
figure 5

A1AT is also a target of wild-type p53 with no possible role in p53 mediated cell cycle function. (a) Western blot analysis of A1AT secreted protein levels in conditioned media and p53 protein in EI-H1299 p53 wild-type cells. β-Tubulin was used as a loading control. (b) qRT–PCR for A1AT expression in panel of lung cancer cell lines: H1299 (p53 null), SBC-3 (p53 WT), H1466 (p53 frame shift) and H2009 (p53 R273L) incubated with 10 μM of nutlin-3a for 24 h. p21 mRNA expression was used as a positive control (right). (c) Western blot analysis of p53 proteins levels in the panel of lung cancer cell lines: H1299 (p53 null), SBC-3 (p53 WT), H1466 (p53 frame shift) and H2009 (p53 R273L) incubated with 10 μM of nutlin-3a for 24 h. In all, 10 μg of each protein samples were loaded on 8% polyacrylamide gels. β-Tubulin was used as a loading control.

Full size image

We next examined the functional role of A1AT in wild-type p53-mediated cell cycle control. The knockdown of A1AT had no effect in p53-mediated cell cycle function in SBC-3 cells following nutlin-3a treatment (Supplementary Figure S5B). Furthermore, the incubation of H1299 cells with concentrated conditioned medium from induced EI-H1299 wild-type cell line had no effect on the invasive potential of H1299 cells129H compared with un-induced counterparts (P<0.05; Supplementary Figure S5D). Interestingly, knockdown of A1AT in both p53 wild-type (SBC-3) and p53 mutant (H2009) resulted in significantly reduced colony size and numbers when cultured in soft agar, suggesting a potential oncogenic role of A1AT to support anchorage-independent growth (P<0.005; Supplementary Figure S5C).

A1AT is upregulated in human lung ADC tissues and correlates with poor prognosis

To investigate the prognostic significance of A1AT gene expression in human lung cancer tissues, publicly available data sets from two histological subtypes of NSCLC patients were analyzed.44 High expression of A1AT was correlated with decreased overall survival of lung ADC patients (Supplementary Figure S6A). In contrast, A1AT expression had no association with the overall survival of squamous cell carcinoma patients (Supplementary Figure S6B). Similarly, A1AT mRNA expression levels were significantly higher in lung ADC patient (n=129) tumors than in squamous cell carcinoma (n=178) as analyzed using TCGA data sets (P<0.0005; Figure 6a).45, 46

Figure 6
figure 6

High A1AT expression in human lung ADC correlates with reduced overall survival. (a) Box plot of A1AT mRNA expression in two histological subtypes of NSCLC patients exported from TCGA data sets—lung ADC (n=129) and squamous cell carcinoma (SCC, n=178). Expression of A1AT was sourced from publicly available data sets. ****P<0.00001. (b) A1AT intensity in lung ADC tumors versus matching normal lung tissues. A1AT protein levels were assessed by IHC on tissue microarray from patients with completely resected stages I–III lung ADC patients and scored quantitatively as described in Materials and Methods section. Each colored dots represent individual patient in the cohort. Scored values are presented as log base 10. Data presented as means±s.e.m. ****P<0.0001. (c) Kaplan–Meier survival curves of 105 lung ADC patients as a function of A1AT-low, versus A1AT-high expressing tumors. Survival comparison and P-values were calculated by log-rank test. Tumors below the median expression level of the cohort were defined as ‘low expression’ and those tumors above the median expression level of the cohort defined as ‘high expression’. (d) Box plot of A1AT expression versus tumor stages of lung ADC. Data calculated using chi-square test (P=0.003). ADC tumors were staged using the AJCC tumor, node and metastasis (TNM) classification. (e) Stratification of human ADC samples (n=105) based on high and low A1AT and p53 expression levels. The levels of p53 and A1AT were determined by IHC and representative images are shown (right). Data calculated using chi-square test (P=0.003).

Full size image

These findings were further confirmed in an independent patient cohort. A1AT expression was evaluated by immunohistochemistry (IHC) on a tissue microarray containing 105 lung ADC tumor samples (stages I–III) and matching normal lung tissue. The IHC analyses showed strong upregulation of A1AT expression in lung ADC tumors compared with matching normal tissues (P<0.0001; Figure 6b). A1AT expression was predominant in the cytoplasm and at the boundary between tumor and stroma. Strikingly, we observed that lung ADC tumors with elevated expression of A1AT were significantly correlated with poor overall survival (P=0.001; Figure 6c). In addition, A1AT expression highly associated with increased propensity for tumor invasion in distant organs (P=0.003; Figure 6d; Supplementary Table 4); thus demonstrating a role for A1AT in tumor cell dissemination and invasion in vivo.

To investigate the correlation of A1AT expression with p53 expression, tissue microarray of lung ADC samples were stained for p53 protein and compared with the levels of A1AT. We used high p53 staining expression as a surrogate marker for missense p53 mutation. In contrast to loss-of-function mutations, missense mutations in TP53 are frequently associated with the expression of high levels of the mutated form of p53 protein that accumulate in the nucleus.6 Consistent with our model, significantly higher levels of A1AT were observed in those tumors that showed strong nuclear accumulation of p53 (P=0.003; Figure 6e). Interestingly, A1AT overexpression showed no correlation with activating alterations of lung ADC oncogenes; EGFR, KRAS and ALK (Supplementary Table 4),47 suggesting that A1AT activation in lung ADC in specific to mutant p53 pathway.

Based on our evidence, we propose that lung cancer cells expressing mutant p53 acquire invasive and metastatic potential through upregulation of A1AT. This study provides first supporting evidence on functional role of oncogenic mutant p53-induced secretome and demonstrates that A1AT targeted therapies may provide an attractive approach to treat mutant p53-expressing tumors.

Discussion

Mutations in the p53 tumor suppressor inactivate its wild-type functions and endow the cancer cells with newly acquired oncogenic properties that contribute to tumor malignancy. The studies on the complex mutant p53 transcriptome have uncovered several mediators of mutant p53 oncogenic action in invasion and metastasis. Importantly, our laboratory has previously highlighted that the induction of mutant p53 releases the secretion of extracellular factors with invasive property.30 The proteomic study for mutant p53-driven effect in breast cancer cells using whole-cell lysates have identified protein targets critical for DNA replication and repair pathways,26 however, to date the global influence of mutant p53 on cancer cell secretome has not been investigated.

In this study, we characterized the mutant p53-induced extracellular factors using an isogenic set of inducible H1299 NSCLC cell line in conjunction with iTRAQ quantitative proteomics. The induction of wild-type p53 has been shown to drive a potent G1 arrest and cease proliferation in H1299 cell line.30 Thus, it is possible that secretomic differences between the wild-type and mutant p53 as determined by iTRAQ may be masked by the underlying differences in the biological outcomes of the cells induced with wild-type or mutant p53. To eliminate this possibility, we used the isogenic paired-matching un-induced H1299 cell lines as a control to accurately detect changes to the cells secretome in the presence of mutant p53. Our study has discovered for the first time that mutant p53 drives its oncogenic pathways through modulating the expression of numerous genes that are subsequently secreted from the cells. Of these genes, we identified A1AT as a critical and indispensable target of mutant p53 that drives cellular invasion both in vitro and in vivo. These results were further demonstrated using novel double inducible system, which demonstrates that A1AT-mediated invasion is specific to the mutant p53 pathway (Figure 2). In agreement with this finding, the IHC staining of lung ADC showed strong correlation between elevated A1AT staining and high expression of p53 (an indicator of presence of mutant p53), which is mutually exclusive with other activating alteration of lung ADC oncogenes: EGFR, KRAS and ALK.47 In this study, we have used p53 histocore as a surrogate marker for potential oncogenic p53 mutants in the absence of TP53 sequencing data (Figure 6). Although this is a method consistent with several previous studies,23 however, this approach may not be entirely accurate because of the presence of unstable mutants or other unexplained reasons for increased wild-type p53 stability. Nevertheless, none of the patients whose tumors were used for this study had received any neoadjuvant chemotherapeutic agents, which eliminates the possibility of chemo-induced effect on expression of p53 in this tumor cohorts. Furthermore, A1AT expression was predominant in tumor cells at boundary between tumor and stroma and showed strong correlation with decreased overall survival in lung ADC and tumor stages (Figure 6; Supplementary Figure S6). These results suggest that lung ADC tumors expressing mutant p53 acquire the invasive and metastatic potential through upregulation of A1AT expression.

Our finding from antibody-mediated inhibition of secreted A1AT led to complete inhibition of mutant p53-induced invasion, providing substantial evidence that A1AT targeted therapies may provide an attractive approach to treat mutant p53-expressing lung tumors (Figure 3; Supplementary Figure S3). Studies have reported high levels of secreted A1AT in the blood and urine samples of cancer patients compared with healthy individuals.48, 49 In addition, serum A1AT level has also been reported to provide diagnostic benefit in bladder cancer.50 Our study shows that secreted A1AT enhances the invasive ability of p53 mutant in lung tumors, hence it is worth testing if serum or plasma A1AT may be used as a predictive diagnostic biomarker in lung ADC.

Despite the evidence of clinical relevance in cancer,48 the regulation of A1AT and its functional role and mechanistic behavior are not yet fully understood. Our studies demonstrate a new role for A1AT as a regulator of EMT-related gene: TGFB1, TGFBR2, SNAI1, SNAI2, ZEB1, FN, VIM and TNC. Furthermore, our findings suggest that the role of A1AT as a downstream mediator of mutant p53 is tumor cell specific, and is mediated by both intracellular and extracellular functions of A1AT. Hence, the oncogenic functions of A1AT are unlikely to be attributed to its physiological function as a serum trypsin inhibitor in lung tissues. Although the mechanism of A1AT in tumorigenesis is not fully understood, it is likely that the regulation of EMT markers by A1AT involves an indirect mechanism, as A1AT is not a transcription factor. Previous studies have reported an anti-apoptotic effect of A1AT on lung endothelial cells via inhibition of caspase-3 activity.51 Studies also indicate that A1AT is an irreversible inhibitor of extracellular serine proteases, such as trypsin and kallikreins52 and matripase.53, 54 In addition, A1AT has been shown to inhibit the activity of natural killer cells against tumor cells,55, 56 indicating that A1AT primarily functions via inhibiting the activity of enzymes present in the extracellular environment. Two important proteases, mast cell chymase and leukocyte elastase, are known to activate TGFB (Transforming growth factor beta) signaling that can induce an EMT in cancer.57 Hence, further study on chymase activity and A1AT may help uncover the mechanism on how A1AT regulates the expression of EMT network genes. Besides its function as a protease inhibitor, A1AT has been reported to induce the release the angiopoietin-like protein 4 in an adherent human blood monocytes and in human lung microvascular endothelial cells via a proliferator-activated receptor-dependent pathway,58 which indicates that A1AT has multiple functions in cancer.

Through ChIP-qPCR, we have also confirmed that A1AT is transcriptionally regulated by mutant p53 through interaction with p63 (Figure 4). p63 is a transcription factor that has major role in normal development and can be expressed in two different isoforms that either have an intact (TAp63) or deleted (deltaNp63) amino terminal transactivation domain.59 The isoforms of p63 have been reported to have differing roles in gene regulation mediated by mutant p53.59, 60, 61, 62 However, the p63 antibody used in this study targets both TA and deltaNp63 isoforms and hence cannot make any conclusions over the isoforms responsible for this regulation of A1AT. In addition, the expression of mutant p53 was found to be comparable across all p53 mutants suggesting that the differences in the binding are likely to be the result of differing promoter affinities between different p53 mutants (Figure 4b top right; Supplementary Figure S4B). Previous studies have shown that genes activated by mutant p53 overlap with wild-type p53 activated targets30 and may represent an oncogenic ‘dark side’ of p53 with pro-survival function.63, 64, 65 In fact, our findings confirm that A1AT is an oncogenic target of mutant p53, which is also upregulated by activated wild-type p53. We further show that A1AT have no role in p53-mediated cell cycle function, rather it supports the anchorage-independent growth potential of cancer cells irrespective of p53 status (Figure 5; Supplementary Figure S5). Interestingly, our finding also shows that the conditioned medium from induced wild-type p53 have no significant impact on the invasive ability of H1299 p53 null cell line (Supplementary Figure S5). In normal unstressed cells, p53 protein is maintained at very low levels66 and the basal expression of A1AT is also limited. However, when p53 expression is triggered by stress or exogenous factor, it activates multitude of genes67 including A1AT. Questions remain as to whether the tumor-suppressive targets of wild-type p53 possess any masking effect on this small subset of overlapping p53 target genes; an effect, which is otherwise absent in p53 mutant tumors. Alternatively, these overlapping genes may favor p53 to select against tumor-suppressive function and promote cancer development as evident through tumors with high levels of wild-type p53.29, 65 Hence, further studies will be required to explore the role of these overlapping genes in p53 pathway.

In conclusion, through quantitative proteomics we have identified A1AT as a secreted mediator of mutant p53-induced invasion activities. Elevated A1AT enhances the invasive and metastatic potential of mutant p53 in lung ADC tumors. Our study provides new insights into the invasive behavior of mutant p53-expressing cells that are manifested through an aberrant secretion of extracellular proteins and offers a potential therapeutic target for treatment of mutant p53-expressing lung tumors.

Materials and methods

Cell culture and constructs

Cell cultures were performed as described in the Supplementary Materials and Methods. The identity of lung cancer cell lines were authenticated by Garvan Molecular Genetics (Darlinghurst, NSW, Australia).

Quantitative real-time PCR

Briefly, total RNA was isolated using RNAeasy mini kit Sigma Aldrich, St. Louis, MO, USA with qRT–PCR performed as described previously.68 Specific primers for real-time PCR are listed in Supplementary Table 3.

iTRAQ Liquid chromatography tandem-mass spectrometry (LC-MS/MS) analysis

iTRAQ Liquid chromatography tandem-mass spectrometry (LC-MS/MS) quantitative proteomic analysis was performed as described in Supplementary Materials and Methods.

In silico analysis and ChIP assays

p53 and p63 scan were used to identify quantification putative p53-REs or 63-REs on A1AT gene. Gene sequence was derived from NCBI (NIH, Bethesda, MD, USA), with Aceview (developed at NCBI) used to define the classical promoter region (10-kB upstream), intronic regions or 3’ UTR. ChIP analysis was performed as described previously.30 The levels of enrichments were determined by qRT–PCR using primers spanning the p53 and p63 REs (Supplementary Table 3).

Lung tissue microarrays and IHC

IHC was performed to determine A1AT and p53 expression and are described in Supplementary Materials and Methods.

Image capture and quantification of A1AT and p53 immunostaining

Slides were digitally scanned using NanoZoomer Digital Pathology System (Hamamatsu Photonics, Hamamatsu, SZK, Japan). The digitized images were edited to exclude stromal regions. Brown pixels (A1AT) were extracted from each image using Adobe Photoshop and the color range tool and a fuzziness factor of 25 as described previously.69 Extracted pixels were converted to greyscale format and a consistent threshold was applied. The intensity of A1AT at the cytoplasm and epithelial–stromal interface was determined using ImageJ software (developed at NIH, Bethesda, MD, USA). An average of the integrated density were quatified per sample. The quantitative scores were converted into log base 10 values and the average values of each patient sample were grouped into two categories based on A1AT staining intensity: low and high. Semiquantitative analysis was performed for nuclear staining of p53 and scored blind as positive and negative. Comparison analysis and Kaplan–Meier survival analysis was done using IBM SPSS statistics 22 software (IBM, Armonk, NY, USA).

Additional Materials and Methods are described in Supplementary Materials and Methods.