Transcriptional repression of SOX2 by p53 in cancer cells regulates cell identity and migration

Abstract

During cancer development and progression, many genetic alterations lead to the acquisition of novel features that confer selective advantage to cancer cells and that resemble developmental programs. SRY‐box transcription factor 2 (SOX2) is one of the key pluripotency transcription factors, expressed during embryonic development and active in adult stem cells. In cancer, SOX2 is frequently dysregulated and associated with tumor stemness and poor patient survival. SOX2 expression is suppressed in differentiated cells by tumor suppressor proteins that form a transcriptional repressive complex. We previously identified some of these proteins and found that their absence combined with deficiency in Trp53 leads to maximal dysregulated expression of Sox2. Using cancer cell lines of different origin and with different p53 status, we show here that manipulating TP53 to restore or decrease its activity results in repression or induction of SOX2, respectively. Mechanistically, we observed that the regulation of SOX2 expression by TP53 is transcriptional and identified Trp53 bound to the promoter region and the Sox2 Regulatory Region 2 enhancer of Sox2. Forcing high levels of SOX2 in cancer cells leads to morphological changes that molecularly correspond to the acquisition of a more mesenchymal phenotype, correlating with an increased migratory capacity. Finally, the analysis of human breast cancer samples shows that this correlation between TP53 status, levels of expression of SOX2, and a more metastatic phenotype is also observed in cancer patients. Our results support the notion that lack of TP53 in tumor cells results in deregulated expression of developmental gene SOX2 with phenotypic consequences related to increased malignization.

What's new?

SOX2 is one of the key developmental genes with a reported altered expression in many tumors. The mechanisms underlying SOX2 regulation in cancer, however, remain elusive. This study shows that TP53 transcriptionally regulates SOX2 through binding to its promoter and the SRR2 enhancer. Increased expression of SOX2 promotes cancer cell epithelial‐to‐mesenchymal transition and increased migration. This correlation between TP53 status, levels of SOX2 expression, and a more metastatic phenotype is also observed in cancer patients. This work demonstrates an additional effect of TP53 genetic alterations, leading to de‐repression of SOX2 and alteration of cancer cell plasticity with important phenotypic consequences.

Abbreviations

ChIP

chromatin immunoprecipitation

DMEM

Dulbecco's modified eagle medium

EMT

epithelial‐to‐mesenchymal transition

FBS

fetal bovine serum

GFP

green fluorescent protein

PEI

polyethylenimine

PFA

paraformaldehyde

RT‐qPCR

quantitative real‐time PCR

SCID

severe combined immunodeficiency

shRNA

short hairpin RNA

SOX2

SRY‐box transcription factor 2

SRR1

Sox2 Regulatory Region 1

SRR2

Sox2 Regulatory Region 2

STR

short tandem repeat

TMA

tissue microarray

1. INTRODUCTION

Cancer progression is driven by genetic mutations in genes that confer a selective advantage to tumor cells. Along tumor development, cancer cells acquire remarkable biological properties that allow them to sustain robust proliferation, escape tumor suppressor control, resist cell death, replicate indefinitely, induce angiogenesis, and ultimately migrate and invade distant organs. ¹ Some of these properties are reminiscent of features that are typical of embryonic development, and the reactivation of developmental programs has been identified during tumor progression. Alternatively, these new malignant attributes could emerge from the acquisition of adult stem/progenitor features by cancer cells.

SRY‐box transcription factor 2 (SOX2) is an essential gene to maintain pluripotency in early embryonic cells of the inner cell mass of the blastocyst, and it is a key factor to reprogram differentiated adult somatic cells back into a pluripotent stem cell stage. ² , ³ , ⁴ Its expression can be found in progenitors of several epithelia during fetal development, while after birth, small populations of Sox2‐expressing cells seem to play crucial roles regulating endodermal and ectodermal tissue regeneration and homeostasis. ⁵

The role of SOX2 has also been investigated in the context of cancer. SOX2 expression is high in tumors of different origins, and its gene amplification has even been identified as driving lung and esophageal squamous cell carcinomas. ⁶ , ⁷ Several studies indicate that SOX2 overexpression provides tumor cells with advantageous properties for tumor development, including increased cellular proliferation, invasion, migration, anchorage‐independent growth, evasion of apoptosis, and drug resistance, while SOX2 knockdown has been linked to reduced tumor progression and decreased resistance to cancer treatments. ⁸ , ⁹

Little is known, however, regarding the regulation of SOX2 in cancer. In previous work from our group, while investigating the transcriptional regulation of SOX2 by the CDK inhibitor p27^{Kip1 10} and the retinoblastoma family of pocket proteins, ¹¹ we realized that a low‐level expression of Sox2, causing dramatic phenotypic consequences, was sufficient to activate its own endogenous transcription provided that Trp53 was downregulated. These results pointed to a potential novel function for Trp53 repressing the expression of Sox2 in differentiated cells and a novel mechanism for deregulated Sox2 expression in cancer.

To further investigate this potential link between the tumor suppressor activity of p53 and de‐regulation of SOX2 in cancer, we decided to manipulate TP53 levels in cancer cells that express wild type TP53 or that lack expression of this tumor suppressor. We found that enforced expression of exogenous TP53, or DNA damage induction of the endogenous gene, leads to a transcriptional repression of SOX2. This regulation seems to be mediated, at least in part, by binding of TP53 to the SOX2 promoter and to its proximal enhancer Sox2 Regulatory Region 2 (SRR2). Conversely, downregulation of endogenous TP53 seems to alleviate SOX2 repression, leading to an increased expression of SOX2. Phenotypically, we found that cancer cells with higher level expression of SOX2 undergo a morphological change accompanied by expression of markers and mediators of the epithelial‐to‐mesenchymal transition (EMT). Functionally, this altered phenotype caused by increased levels of SOX2 correlated with an increase in the migratory capacity of cancer cells, suggesting a link between high level expression of SOX2 in cancer and increased migration, and possibly a more metastatic capacity. This hypothesis was further supported by data from breast cancer patient‐derived samples that showed a positive correlation between higher expression of SOX2, mutation in TP53, and a more metastatic behavior.

2. MATERIALS AND METHODS

2.1. Cell line generation

NCI‐H1299 (RRID:CVCL_0060), U‐373MG ATCC (RRID:CVCL_2219), T‐47D (RRID:CVCL_0553), A‐549 (RRID:CVCL_0023), U‐87MG ATCC (RRID:CVCL_0022), MCF‐7 (RRID:CVCL_0031), and HEK293T (RRID:CVCL_0063) cell lines were obtained from the ATCC. T653 and T731 are tumor cell lines derived from gliomas originating from inoculation of HRas ^V12 and Rb1 ^−/− HRas ^V12 cells in mice with severe combined immunodeficiency (SCID) (a kind gift from José Antonio Costoya, CiMUS, USC). All cell lines were cultured in Dulbecco's Modified Eagle Medium (DMEM) (Sigma‐Aldrich) supplemented with 10% fetal bovine serum (FBS) (Sigma‐Aldrich), 1% glutamine (Sigma‐Aldrich), and 1% penicillin/streptomycin (Sigma‐Aldrich) and maintained in an incubator at 37°C and 5% CO₂ atmosphere. All cell lines were authenticated using short tandem repeat (STR) profiling within the last 3 years. All experiments were performed with mycoplasma‐free cells.

To generate TP53‐overexpressing cells, target cells were transfected with expression plasmid pcDNA3‐p53 (a gift from R. T. Hay), ¹² that encodes human TP53 cDNA, or pcDNA3 (Invitrogen) as a control, in the presence of 1 mg/mL polyethylenimine reagent (PEI) (Polysciences). To generate TP53‐knockdown cell lines, human cancer cells were transduced with pRetroSuper‐Blast‐shp53, and murine cancer cells were transduced with pRetroSuper‐puro‐shp53, or with the respective controls pRetroSuper‐Blast and pRetroSuper‐puro (all of them a kind gift from Reuven Agami).

To obtain cells with increased endogenous SOX2 levels, cells were transduced with PL‐SIN‐EOS‐C(3+)‐EiP (a gift from James Ellis; Addgene #21313) and treated with 1 μg/mL of puromycin (Millipore) to enrich for naturally expressing high levels of SOX2. ¹³ SOX2‐overexpressing cells were generated by retro or lentiviral transduction with pMXs‐Sox2‐IP (a gift from Shinya Yamanaka; Addgene #15919), ⁴ which encodes mouse SOX2 cDNA, and pBabe‐puro‐IRES‐EGFP (a gift from L. Miguel Martins; Addgene #14430) as a control, or pLV‐tetO‐Sox2 (a gift from Konrad Hochedlinger; Addgene #19765), ¹⁴ which also encodes mouse SOX2 cDNA, and the transactivator FUW‐M2rtTA (a gift from Rudolf Jaenisch; Addgene #20342), ¹⁵ followed by treatment with 0.5 μg/mL of doxycycline. To silence SOX2 expression, we cloned the specific shRNA_1983 ¹⁶ targeting SOX2 into the LT3GEPIR vector (a gift from Johannes Zuber, Addgene #111177) ¹⁷ and TP53‐silenced A‐549 cells were transduced with the LT3GEPIR‐shSOX2 vector and treated with 1 μg/mL of doxycycline for 48 h to induce the expression of the short hairpin RNA (shRNA).

For retroviral and lentiviral transduction, HEK293T cells were co‐transfected with the plasmid of interest and the corresponding packaging plasmids in equal proportions, in the presence of 1 mg/mL PEI reagent. Third‐generation lentiviral packaging plasmids were pLP1, pLP2, pLP‐VSVG (ViraPower Lentiviral Packaging Mix, K497500, Invitrogen); second‐generation lentiviral packaging plasmids were psPAX2 (a gift from Didier Trono; Addgene #12260) and pLP‐VSVG, and retroviral packaging plasmids were pCL‐Eco (a gift from Inder Verma; Addgene #12371) ¹⁸ and pLP‐VSVG. Thirty‐six hours after transfection, we collected viral supernatants every 12 h for three consecutive rounds. The supernatants were filtered through a 0.45 μm filter, supplemented with 8 μg/mL of polybrene (Sigma‐Aldrich) and added to target cells. Cell cultures transduced with plasmids with antibiotic resistance genes were selected with 1 μg/mL puromycin (Millipore) or 4–10 μg/mL blasticidin (Sigma‐Aldrich).

2.2. Quantitative real‐time PCR

To measure gene expression, total RNA was extracted using the NucleoSpin RNA Plus kit (Macherey‐Nagel) following the manufacturer's protocol, and retrotranscribed into cDNA using the High‐Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer's protocol. Quantitative real‐time PCR (RT‐qPCR) was performed using the NZYSpeedy qPCR Green Master Mix reagent (2X) ROX (NZYTech) and the QuantStudio 3 Real‐Time polymerase chain reaction (PCR) System (Applied Biosystems). Each reaction was performed in triplicate and consisted of 33 ng of cDNA, 0.25 μM primers, 5 μL of Master Mix reagent in a final volume of 10 μL per reaction. RNA expression values were relativized to the expression levels of the housekeeping gene GAPDH. The results were analyzed with the QuantStudio 3 Real‐Time PCR Software (Applied Biosystems). Primer sequences were as follows:

GAPDH: 5′‐TCCATGACAACTTTGGCATCGTGG‐3′, 5′‐GTTGCTGTTGAAGTCACAGGAGAC‐3′;

TP53: 5′‐CCGCAGTCAGATCCTAGCG‐3′, 5′‐AATCATCCATTGCTTGGGACG‐3′;

CDKN1A: 5′‐CCTGTCACTGTCTTGTACCCT‐3′, 5′‐GCGTTTGGAGTGGTAGAAATCT‐3′;

SOX2: 5′‐GGGAAATGGGAGGGGTGCAAAAGAGG‐3′, 5′‐TTGCGTGAGTGTGGATGGGATTGGTG‐3′;

Cdkn1a: 5′‐GTGGGTCTGACTCCAGCCC‐3′, 5′‐CCTTCTCGTGAGACGCTTAC‐3′;

Sox2: 5′‐TAGAGCTAGACTCCGGGCGATGA‐3′, 5′‐TTGCCTTAAACAAGACCACGAAA‐3′;

Trp53: 5′‐TGAAACGCCGACCTATCCTTA‐3′, 5′‐GGCACAAACACGAACCTCAAA‐3′;

CDH1: 5′‐CGAGAGCTACACGTTCACGG‐3′, 5′‐GGGTGTCGAGGGAAAAATAGG‐3′;

CDH2: 5′‐AGCCAACCTTAACTGAGGAGT‐3′, 5′‐GGCAAGTTGATTGGAGGGATG‐3′;

SNAI1: 5′‐TCGGAAGCCTAACTACAGCGA‐3′, 5′‐AGATGAGCATTGGCAGCGAG‐3′ and

TWIST1: 5′‐GTCCGCAGTCTTACGAGGAG‐3′, 5′‐GCTTGAGGGTCTGAATCTTGCT‐3′.

2.3. Western blot

To analyze protein levels, cell extracts were resuspended in Radio‐Immunoprecipitarion Assay buffer (150 mM NaCl, 10 mM Tris–HCl pH 7.5, 0.1% sodium dodecyl sulfate, 1% Triton X100, 5 mM ethylenediaminetetraacetic acid pH 8.0, 1% deoxycholate and sodium salt) containing protease inhibitors (1 mM sodium orthovanadate; 1 mM phenylmethylsulfonyl fluoride; 1 mM dithiothreitol (DTT); 4 mM NaF; protein inhibitor cocktail, P8340 Sigma‐Aldrich). Protein concentration was quantified with the DC Protein Assay Kit (Bio‐Rad) and 25 μg of total cellular protein were electrophoresed in 12% polyacrylamide gels. After transferring to a polyvinylidene fluoride membrane, the blot was blocked with 5% milk solution in Tween‐Tris‐buffered saline (20 mM Tris–HCl pH 7.5, 150 mM NaCl, 0.05% Tween‐20). Membranes were incubated at 4°C overnight with primary antibodies (TP53: sc‐6243, Santa Cruz Biotechnology, 1:500; P21: sc‐6246, Santa Cruz Biotechnology, 1:500; SOX2: #2748, Cell Signaling, 1:1000; β‐TUBULIN: #2146, Cell Signaling, 1:1000; GAPDH: sc‐32,233, Santa Cruz Biotechnology, 1:2000; β‐ACTIN: sc‐8432, Santa Cruz Biotechnology, 1:1000). Secondary antibodies (anti‐mouse IgG‐horseradish peroxidase (HRP): #31430, Invitrogen, 1:10,000; anti‐rabbit IgG‐HRP: #31460, Invitrogen, 1:10,000) were incubated for 1 h. Chemiluminescence detection was performed using SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Scientific).

2.4. Luciferase reporters

To analyze the activity of regulatory regions, cells were co‐transfected with luciferase reporter plasmids PG13‐luc (wt p53 binding sites) (a gift from Bert Vogelstein; Addgene #16442) ¹⁹ and pGL3‐Sox2 luc (a gift from Richard G. Pestell) in the presence of PEI. After 48 h, the cultures were lysed with lysis buffer (Roche) and luciferase assay buffer (25 mM glycylglycylglycine pH 7.8, 15 mM phosphate buffer pH 7.8 [KH₂PO₄ and K₂HPO₄], 15 mM MgSO₄, 4 mM ethyleneglycoltetraacetic acid, 2 mM adenosine triphosphate, 1 mM DTT). Luciferase activity was measured using the Mithras LB 940 luminometer (Berthold Technologies) after adding D‐luciferin (Sigma‐Aldrich) with the automatic injector. To ensure transfection efficiency, cells were co‐transfected with the plasmid pBP‐LacZ (a gift from Paul Freemont; Addgene #72948) ²⁰ and β‐galactosidase activity was measured. Cell extracts were added to Z‐buffer (40 mM NaH₂PO₄, 60 mM Na₂HPO₄, 10 mM KCl, 1 mM MgSO₄, β‐mercaptoethanol and 2‐nitrophenyl) for this purpose. Β‐D‐galactopyranoside (Sigma‐Aldrich) was incubated at 37°C for 8 h. The optical density was measured at 450 nm using Mithras LB 940 (Berthold Technologies).

2.5. Chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP) was performed as described in King and Klose, ²¹ with minor adjustments. After chromatin extraction, lysates were sonicated at 20% amplitude, 15 s ON, 30 s OFF for 7 min using a Branson Sonifier 550 (Branson Ultrasonics). Antibodies used for ChIP were anti‐goat IgG (G4018, Sigma‐Aldrich) and p53 (FL‐393) (sc‐6243, Santa Cruz Biotechnology). DNA samples were purified with ultraPure phenol:chloroform:isoamyl alcohol (25:24:1, v/v) (15593‐031, Invitrogen) and eluted in 80 μL H₂O. RT‐qPCR was performed in triplicates and each reaction consisted of 2 μL sample, 0.25 μM primers, and 5 μL of NZYSpeedy qPCR Green Master Mix reagent (2X) ROX (NZYTech) in a final volume of 10 μL per reaction, using the QuantStudio 3 Real‐Time PCR System (Applied Biosystems). Primer sequences were following:

Cdkn1a promoter: 5′‐CCAGAGGATACCTTGCAAGGC‐3′, 5′‐TCTCTGTCTCCATTCATGCTCCTCC‐3′;

Sox2 promoter: 5′‐CCTAGGAAAAGGCTGGGAAC‐3′, 5′‐GTGGTGTGCCATTGTTTCTG‐3′;

Sox2‐SRR1: 5′‐TCCCCCAATACTGGTGGTCGTCA‐3′, 5′‐GAAGGCGAACGGCAGGGGAC‐3′;

Sox2‐SRR2 ¹: 5′‐ATTTATTCAGTTCCCAGTCCAAGC‐3′, 5′‐CCCTCTCCCCCCACGC‐3′ and

Sox2‐SRR2 ²: 5′‐CGTGGTAATGAGCACAGTCG‐3′, 5′‐AGGCTGAGTCGGGTCAATTA‐3′.

2.6. Proliferation and clonogenicity assays

To analyze cell proliferation, A‐549 and MCF‐7 cells were seeded at a density of 8 × 10³ and 1 × 10⁴ cells, respectively, in 24‐well plates in triplicate for each condition. Cells were counted every 2 days using the LUNA II Automated Cell Counter (Logos Biosystems). For the clonogenicity assay, 2 × 10³ A‐549 or MCF‐7 cells were seeded in six‐well plates, in triplicate for each condition. After 12–14 days, cell colonies were fixed with 4% paraformaldehyde (PFA) and stained with 0.05% crystal violet solution. Images were obtained after scanning the plates with the Epson Perfection V550 Photo scanner. For quantification, the staining was eluted with 10% acetic acid solution and the optical density was measured at 570 nm in the spectrophotometer Epoch 2 (Biotek).

2.7. Migration assay

Cell migration capacity was assessed by migration through 8 μm Transwell inserts (Falcon) in 24‐well plates. A‐549 and MCF‐7 cells were seeded in the upper chamber at a density of 2.5 × 10⁴ cells in serum‐free medium, in triplicate. In the lower chamber, the medium was supplemented with 10% FBS to promote cell migration. After 16 h for A‐549 or 48 h for MCF‐7, cells were fixed with 4% PFA, permeabilized with 100% methanol, and stained with 0.05% crystal violet solution. After removing the cells in the upper chamber, we took images with Axio Vert.A1 Microscope (Zeiss) or SZ61 Microscope (Olympus) and the number of migrated cells was quantified using ImageJ.

2.8. Immunohistochemistry

This study utilized tissue microarray (TMA) and breast carcinoma samples obtained from a previously described cohort. ²² Tumor sections, two micrometers thick, were deparaffinized using BOND Dewax Solution (Leica Biosystems) and subsequently rehydrated with a series of ethanol solutions. Immunohistochemistry was performed using a BOND RXm autostainer (Leica Biosystems), employing BOND Epitope Retrieval Solutions and the BOND Polymer Refine Detection kit (Leica Biosystems) following the manufacturer's protocols. Images were captured using Aperio eSlide Manager software (Leica Biosystems). SOX2 and TP53 levels were categorized according to Rodriguez‐Pinilla et al. ²³

2.9. Statistical analysis

All data are represented as mean ± standard deviation. Statistical significance was calculated using the two‐tailed Student's t‐test for most experiments, except for immunohistochemistry experiments, which were analyzed using the two‐tailed Fisher's exact test. p‐Values less than .05 were considered statistically significant: ***p < .001; **p < .01; * p < .05; ns not significant.

Additional experimental procedures are described in Supporting Information S1.

3. RESULTS

3.1. Tumor suppressor TP53 regulates SOX2

To explore the contribution of tumor suppressor TP53 to the regulation of the expression of SOX2 in cancer cells, we selected lung (NCI‐H1299 and A‐549), glioma (U‐373MG and U‐87MG), and breast (T‐47D and MCF‐7) tumor cell lines with defective or wild‐type TP53, respectively; and transfected them with a plasmid expressing TP53 or a shRNA against TP53 to restore or decrease the expression of the tumor suppressor, accordingly. When we transfected TP53‐deficient NCI‐H1299, U‐373MG, and T‐47D cells with a plasmid coding for TP53, we observed different levels of expression of the tumor suppressor protein and of its target gene, CDKN1A (coding for p21). Following the increased expression of TP53 in all the tumor cell lines, we observed a reduction of SOX2, both at mRNA (Figure 1A) and protein levels (Figure 1B). Conversely, when we introduced a shRNA targeting TP53 in wild‐type A‐549, U‐87MG, and MCF‐7 cells, the mRNA (Figure 1C) and protein levels (Figure 1D) of TP53 and CDKN1A (and of its encoded protein, p21) were reduced, as expected, while at the same time SOX2 was increased (Figure 1C,D).

FIGURE 1.

Effects of altered TP53 expression on SOX2 levels. (A) TP53 (left), CDKN1A (middle) and SOX2 (right) relative mRNA levels by RT‐qPCR in TP53 defective NCI‐H1299, U‐373MG and T‐47D tumor cell lines overexpressing TP53. (B) TP53, P21 and SRY‐box transcription factor 2 (SOX2) protein levels by western blot in TP53 defective NCI‐H1299, U‐373MG and T‐47D tumor cell lines overexpressing TP53. (C) TP53 (left), CDKN1A (middle) and SOX2 (right) relative mRNA levels by RT‐qPCR in TP53 wild type A‐549, U‐87MG and MCF‐7 tumor cell lines with silenced TP53 expression. (D) TP53, CDKN1A and SOX2 protein levels by western blot in TP53 wild type A‐549, U‐87MG and MCF‐7 tumor cell lines with silenced TP53 expression. P‐values less than 0.05 were considered statistically significant: ***p < .001; **p < .01; *p < .05; ns not significant.

In order to corroborate these findings without recurring to genetic manipulation, we decided to investigate the effect of using a chemotherapeutic drug that causes DNA damage and consequently activates TP53, namely doxorubicin, on the expression of SOX2. For this, we used two tumor cell lines derived from gliomas originated by inoculation of HRas ^V12 or Rb1 ^−/− HRas ^V12 cells in SCID mice, T653 and T731. Addition of doxorubicin (0.5 μg/mL) to these cells caused a time‐dependent increase in Trp53 protein that was accompanied by an increase of its target, p21, and a concomitant decrease in the levels of Sox2 protein in both cell lines (Figure 2A). Again, the analysis of Sox2 mRNA revealed that these reduced levels of Sox2 were caused by a downregulation of its mRNA after activation of Trp53 and Cdkn1a by doxorubicin (Figure 2B).

FIGURE 2.

The activation of Trp53 by DNA damage decreases Sox2 levels. (A) Trp53, P21 and Sox2 protein levels assessed by western blot in T653 (top) and T731 (bottom) mouse glioma cell lines treated with 0.5 μg/mL doxorubicin (doxo) at different times. (B) Relative mRNA levels by quantitative real‐time PCR (RT‐qPCR) of the Trp53 target gene Cdkn1a and Sox2 in T653 and T731 cells treated with 0.5 μg/mL doxo at different times. (C) Trp53, P21 and Sox2 protein levels by western blot in T653 and T731 cells previously transduced with an shTrp53 and treated with 0.5 μg/mL doxo for 10 h. (D) Relative mRNA levels by RT‐qPCR of Trp53 and Sox2 in T653 and T731 cells previously transduced with an shTrp53 and treated with 0.5 μg/mL doxo for 10 h. EV, empty vector. P‐values less than 0.05 were considered statistically significant: ***p < .001; **p < .01; *p < .05; ns not significant.

To prove that this reduction in Sox2 mRNA and protein after doxorubicin treatment of the glioma cells is due to the activation of Trp53 and not the result of an off‐target effect of doxorubicin, we repeated the experiment adding the drug to cells previously transduced with an shRNA construct targeting Trp53. Again, we observed that doxorubicin caused a sharp increase in the levels of Trp53, as expected, and this change correlated with a reduction in the levels of Sox2, both at the protein and mRNA levels (Figure 2C,D, respectively). Interestingly, the introduction of the shRNA against Trp53 impeded the increase in Trp53 after doxorubicin addition, and this in turn prevented the repression of Sox2 expression (Figure 2C,D).

These results clearly show that increased genetic expression of TP53 or its induction after DNA damage in cancer cells results in downregulation of SOX2 expression and that, conversely, inactivation of TP53 in cancer cells retaining a wild type tumor suppressor protein alleviates SOX2 transcriptional repression.

3.2. The regulation of SOX2 by tumor suppressor TP53 is transcriptional

To address if the effect of TP53 on SOX2 expression was produced at the transcriptional level, we transfected lung and glioma tumor cell lines carrying wild type or defective TP53 with reporter constructs expressing luciferase under the control of a promoter containing 13 copies of the Tp53‐binding consensus sequence (pG13‐luc) as a positive control for Tp53 activity, or with a reporter expressing luciferase under the control of the SOX2 promoter (pGl3‐Sox2‐luc). Using this system, we observed that the overexpression of TP53 caused an increase in luciferase activity in the positive control pG13‐luc cells, as expected, and a decrease in luciferase from the SOX2 promoter reporter plasmid pGl3‐Sox2‐luc (Figure 3A, upper panels). Conversely, the knockdown of TP53 reduced the expression of luciferase from the control plasmid pG13‐luc and stimulated luciferase expression from the SOX2 promoter, confirming that TP53 expression negatively affects the transcription of SOX2 (Figure 3A, lower panels).

FIGURE 3.

The alteration of TP53 levels modulates SOX2 activity transcriptionally in lung cancer and glioma cell lines. (A) Luciferase activity from the CDKN1A promoter (pG13‐luc) or the SOX2 promoter (pGl3‐Sox2‐luc) in TP53 defective tumor cell lines NCI‐H1299 and U‐373MG overexpressing TP53 (top) and in TP53 wild type cell lines A‐549 and U87MG (bottom). (B) Analysis of TP53 binding by Trp53 induction with 0.5 μg/mL doxorubicin for 10 h followed by chromatin immunoprecipitation with a Trp53 antibody and quantitative real‐time PCR of the Cdkn1a promoter, the Sox2 promoter, the Sox2 Regulatory Region 1 (SRR1) and the Sox2 Regulatory Region 2 (SRR2) (with two different primer pairs). EV, empty vector. P‐values less than 0.05 were considered statistically significant: ***p < .001; **p < .01; *p < .05; ns not significant.

Next, we wondered if we could identify Trp53 bound to regulatory regions on the Sox2 locus. For this, we performed ChIP followed by RT‐qPCR with primers amplifying the promoter region or two previously described enhancers surrounding the Sox2 open reading frame, namely the Sox2 Regulatory Region 1 (SRR1) and the SRR2. ChIP using specific anti‐Trp53 antibodies allowed the detection of the Cdkn1a promoter, a well‐known transcriptional target of Trp53, in T731 cells after doxorubicin treatment. Similarly, ChIP after DNA damage led to the detection of the Sox2 promoter as well as the SRR2 using two different pairs of primers for this enhancer region, but not the SRR1 enhancer (Figure 3B).

Altogether, our results clearly show that TP53 represses SOX2 gene expression in cancer cells acting at the transcriptional level, at least through binding to the SOX2 promoter and its SRR2 enhancer.

3.3. Increased levels of SOX2 in cancer cells do not alter proliferation

Next, we wondered what the effect would be of expressing higher levels of SOX2 in a cancer cell. To address this question, we set up three different systems to achieve higher gradual expression levels of SOX2. For the first one, we took advantage of a SOX2 reporter plasmid carrying tandem repeats of an enhancer element containing Sox2/Oct4 binding sites upstream of a promoter derived from a murine early transposon active in embryonic stem cells that drives the expression of green fluorescent protein (GFP) and puromycin resistance (EOS‐GFP). Introduction of this construct in A‐549 and MCF‐7 cells renders a small fraction of these cell cultures positive for GFP and resistant to puromycin (Figure S1A,B). Antibiotic selection of this population enriches for tumor cells showing the highest natural level of expression of SOX2 within the cancer cell line, as we confirmed by western blot and RT‐qPCR (Figure S1C). As a second approach, we used retroviral expression of Sox2 using pMXs‐Sox2‐IP. After transduction and selection of the cell population carrying the plasmid, we confirmed the elevated levels of Sox2 in these cells by western blot and RT‐qPCR (Figure S1D). Finally, we used a tetracycline‐inducible system to overexpress Sox2, pLV‐tetO‐Sox2, showing the highest level of expression of Sox2 protein and mRNA of the three systems (Figure S1E).

Using these cell models with different degrees of expression of SOX2, we observed that cells expressing higher levels of SOX2 did not show increased proliferation when followed for 8 days in culture (Figure S2A). On the contrary, when Sox2 expression was maximal, cells appeared to have an impaired proliferative rate (Figure S2A, bottom panels). Moreover, when we tested the capacity of SOX2‐overexpressing tumor cells to form colonies at low density after 14 days in culture, the results indicated that SOX2 expression does not confer a proliferative advantage to tumor cells (Figure S2B) or, if anything, it represents an obstacle for efficient proliferation when SOX2 is very highly expressed (Figure S2B, bottom panels).

3.4. Increased levels of SOX2 in cancer cells induce features of epithelial‐to‐mesenchymal transition

The microscopic inspection of these cell cultures revealed, however, morphological changes in the cells expressing higher levels of SOX2, with cells showing a more elongated fusiform shape, resembling a more mesenchymal phenotype in the epithelial cancer cell lines A‐549 (Figure S3A,C,E) and MCF‐7 (Figure S3B,D,F), and with a stronger effect caused by the two overexpression systems, pMXs‐Sox2‐IP (Figure S3C,D) and pLV‐tetO‐Sox2 (Figure S3E,F). This prompted us to evaluate the possibility of an induced EMT caused by the forced expression of SOX2 in A‐549 and MCF‐7. For this, we first checked the expression of a classical epithelial marker, CDH1 (coding for E‐Cadherin), and a mesenchymal marker, CDH2 (coding for N‐Cadherin), using the three systems of increased expression of SOX2. We confirmed that, for the most part, the levels of CDH1 were reduced, while for CDH2 they were increased when SOX2 was overexpressed (Figure 4A,B), confirming the induction of a change in the phenotype of the epithelial cancer cell lines A‐549 and MCF‐7 to a more mesenchymal one. However, we also observed some variability in these responses, with certain conditions showing a less pronounced or inconsistent regulation of CDH1 or CDH2. This heterogeneity suggests that SOX2‐mediated EMT induction may be influenced by additional context‐dependent factors and that cells with increased SOX2 levels could develop a partial EMT induction, resulting in an epithelial‐mesenchymal hybrid phenotype. Apart from the expression of cell identity markers, we decided to check the expression of some of the most commonly described mediators of the process of EMT, transcription factors SNAIL1 and TWIST1. Both A‐549 and MCF‐7 cells expressing high levels of SOX2 from the three overexpression systems showed higher level expression of these EMT mediators (Figure 4C,D).

FIGURE 4.

Increasing SOX2 levels promotes the expression of epithelial‐to‐mesenchymal transition (EMT) mediators. mRNA expression by quantitative real‐time PCR of the epithelial marker CDH1 (A) the mesenchymal marker VIM (B) and the EMT effectors SNAI1 and TWIST1 (C, D) in A‐549 and MCF‐7 cell lines transduced with EOS‐GFP and selected with 1 μg/mL puromycin (puro; top), with pMXs‐Sox2‐IP or the control pBP‐GFP (middle) and pLV‐tetO‐Sox2 followed by 0.5 μg/mL doxycycline (dox) treatment (bottom). P‐values less than 0.05 were considered statistically significant: ***p < .001; **p < .01; *p < .05; ns not significant.

Together, the change in cellular morphology, the increased expression of mesenchymal markers and reduced expression of epithelial markers, and the expression of EMT transcriptional regulators all point to a switch in tumor cell identity that favors a more mesenchymal phenotype.

3.5. Increased levels of SOX2 in cancer cells increase migration

One of the effects derived from the acquisition of an EMT state in cancer cells is the stimulation of cell migration. To test whether this is the case when cancer cells undergo EMT due to increased expression of SOX2, we tested the migration capacity of A‐549 and MCF‐7 cells forced to express higher levels of SOX2 using our three systems of gradual increased levels. For this, we performed migration assays using transwells. A‐549 and MCF‐7 cells expressing higher levels of SOX2 were seeded in cell culture medium lacking FBS, and these upper chambers were placed in wells containing medium with 10% FBS to create a chemoattractant gradient that stimulates migration. After incubation, migrating cells were quantified by removing the filter at the bottom of the top compartment, staining, and counting. Again, using the three cellular systems of higher expression of SOX2, we observed a clear and consistent increase in the number of cells migrating in the transwells when SOX2 was overexpressed (Figure 5A–C).

FIGURE 5.

Increasing SOX2 levels promotes cell migration capacity. Representative images and quantification of the transwell migration assay of A‐549 (top) and MCF‐7 cells (bottom) transduced with EOS‐GFP and selected with 1 μg/mL puromycin (puro) (A), with pMXs‐Sox2‐IP or the control pBP‐GFP (B) and pLV‐tetO‐Sox2 followed by 0.5 μg/mL doxycycline (dox) treatment (C). Relative mRNA levels of TP53 and SOX2 by quantitative real‐time PCR in A‐549 cells previously transduced with empty vector (EV), shTP53, or shTP53 shSOX2 (D). Representative images and quantification of the transwell migration assay of A‐549 EV, shTP53, or shTP53 shSOX2 (E). C, control; NT, not treated. P‐values less than 0.05 were considered statistically significant: ***p < .001; **p < .01; *p < .05; ns not significant.

We also investigated whether SOX2 de‐repression resulting from TP53 inactivation in cancer cells directly influences cell migration and whether this effect could be reversed by SOX2 silencing. To evaluate this, we transduced TP53‐silenced A‐549 cells with a shRNA targeting SOX2 to suppress its expression (Figure 5D) and assessed cell migration through transwells. TP53‐silenced A‐549 cells exhibited a significant increase in migration, an effect that was abrogated upon SOX2 silencing (Figure 5E).

These assays corroborated the altered cancer cell phenotype observed when SOX2 expression was increased and that is typically observed when epithelial cancer cells increase their plasticity to become more mesenchymal and migratory.

3.6. SOX2 expression is higher in TP53 mutant human breast cancer patients, correlating with a more metastatic phenotype

To further substantiate our molecular findings in cancer patients, we decided to analyze the expression of SOX2 in a set of breast cancer patient‐derived tissue samples by immunohistochemistry. First, we observed a statistically significant (p = .036) positive correlation between a TP53+ staining and a concomitant SOX2+ staining (Table 1). Notably, TP53+ immunohistochemical staining is recognized as an indicator of mutant TP53 status, as mutant TP53 proteins often accumulate due to impaired degradation, whereas wild‐type TP53 remains undetectable by this method. ²⁴ , ²⁵ , ²⁶ To further validate these findings, we analyzed public datasets from lung adenocarcinoma (LUAD), glioblastoma multiforme (GBM), and breast invasive carcinoma (BRCA) (TCGA, Firehose Legacy). In line with our observations in patient‐derived tissues, SOX2 expression was significantly higher in tumors harboring TP53 mutations across the three cancer types (Figure S4A). These data reinforce our observations of an inverse correlation between TP53 functionality and SOX2 expression in cancer cells, with SOX2 protein being expressed at higher levels in tumors with defective TP53.

TABLE 1.

Correlation between SOX2 expression with TP53 status and tumor metastatic state in human breast cancer samples.

	SOX2 negative	SOX2 positive	p
TP53
Negative (n = 40)	32 (80.0%)	8 (20.0%)	p = .036
Positive (n = 43)	25 (58.1%)	18 (41.9%)
Metastasis
Negative (n = 58)	48 (82.8%)	10 (17.2%)	p = .003
Positive (n = 21)	10 (47.6%)	11 (52.4%)

In addition, we observed a clear and statistically significant (p = .003) difference in SOX2 expression according to the metastatic phenotype of breast tumors, with higher level expression of SOX2 correlating with tumors showing a more metastatic behavior (Table 1). Moreover, in BRCA tumors, SOX2 expression levels were also associated with higher tumor grade and lymph node metastasis status (Figure S4B), reinforcing the link between dysfunctional TP53, SOX2 upregulation, and tumor progression.

Together, our data in breast cancer patient tissues and cancer datasets corroborate our in vitro findings, suggesting that another tumor suppressor function of TP53 is to restrict SOX2 expression in cancer cells carrying other genetic mutations that could lead to a change in tumor cell identity and the acquisition of a more migratory and invasive activity.

4. DISCUSSION

The reactivation of developmental programs is a frequent event during cancer progression that confers tumor cells with advantages in cell competition. SOX2 is one of the key developmental genes whose expression has been reported to be altered in many tumors. Adult somatic cells typically maintain the SOX2 locus completely repressed. However, deletion of some key tumor suppressor genes, such as CDK inhibitors Cdkn1a and Cdkn1b, or the retinoblastoma members, Rb1 and Rbl2, results in a partial de‐repression and low‐level expression of Sox2 capable of causing a dramatic phenotype. ¹⁰ , ²⁷ When these genetic alterations are combined with deletion of Trp53, the deregulated expression of Sox2 is maximal. ¹¹

These observations led us to explore the role of TP53 repressing the expression of SOX2 in cancer. For this, we set up a number of cancer cell lines of different origins and with defective TP53 function or retaining a wild type tumor suppressor protein and genetically reversed the status of TP53 on them. This allowed us to prove an inverse correlation between active TP53 and repressed SOX2 expression and, conversely, decreased TP53 and induction of SOX2 expression (Figure 1A–D). Although our results followed a consistent trend, we observed some heterogeneity that could be, at least in part, due to a variable dependency of TP53 on the regulation of SOX2 in the different cell lines. In the context of pluripotency and the transition to a differentiated state, Trp53 has already been shown to be capable of regulating pluripotency factors. ²⁸ , ²⁹ For example, mouse ES cell differentiation caused by the DNA damaging agent doxorubicin causes a shift in whole genome occupancy of Trp53, resulting in its recruitment to regions in which pluripotency core transcription factors reside and showing a repressive activity acting through distal enhancers, while at the same time it binds promoter regions of differentiation‐associated genes to activate them. ²⁹ Interestingly, we have recently described new distal developmental enhancers controlling the expression of Sox2 (namely SRR124‐134) that can be reprogrammed during tumorigenesis that merit further detailed investigation. ³⁰

Our results using luciferase reporter plasmids suggest that this repressive control over SOX2 expression by TP53 is transcriptionally mediated (Figure 3A). The ChIP analysis we performed points to the binding of Trp53 to the Sox2 promoter region and to the enhancer SRR2 (Figure 3B). This SRR2 was shown to be bound by Sox2 together with Oct3/4 to activate its own transcription and was originally identified together with SRR1 as regulatory regions stimulating Sox2 transcription in ES cells and promoting the undifferentiated state. ³¹ Interestingly, an activated SRR2 showing increased Sox2 binding capacity in cancer cells under oxidative stress has been reported to confer resistance to this type of stress, contributing to tumorigenesis. ³² This link between increased levels of expression of Sox2 and the activity of the SRR2 was also reported in triple negative breast cancer cells. A highly responsive SRR2 is a feature of a highly tumorigenic subpopulation in triple negative breast cancers showing increased capacity to form tumor spheres. ³² Thus, having a transcriptional repressive factor such as TP53 acting directly at this location could have profound effects on SOX2 expression in cancer cells. There are, however, other ways to repress SOX2 expression in a TP53‐dependent manner different from a direct effect of TP53 that we have not explored in the context of cancer. For example, microRNA miR‐34a has been shown to represent a barrier for somatic cell reprogramming and this suppression of reprogramming was shown to be due, at least in part, to repression of pluripotency genes, including SOX2. ³³

The role of SOX2 in cancer has been addressed by many laboratories around the world and found to be important in the development, maintenance of stemness, cancer progression, and resistance toward cancer therapies. ⁸ , ⁹ , ³⁴ SOX2 was originally identified as an oncogene amplified in squamous cell carcinomas of the lung and esophagus where it drives the expression of markers of both squamous differentiation and pluripotency. ⁶ Interestingly, in prostate cancer, mutations in TP53 and RB1 lead to increased expression of SOX2 that results in increased cellular plasticity, allowing the escape from antiandrogen therapy by lineage switching. Prostate cells with high levels of SOX2 shift from an androgen receptor‐dependent luminal epithelial phenotype to a basal‐like state that is independent of the androgen receptor, making them resistant to the antiandrogen drug enzalutamide. ¹⁶ , ³⁵ In another report, using a reporter plasmid based on the activation of GFP expression driven by genetic elements bound and activated by SOX2, the authors showed how selecting GFP+ cells from different breast cancer cell lines allows the isolation of a subpopulation with higher tumor‐ and metastasis‐initiating activity, and resistance to chemotherapy. ³⁶ In our case, using systems that allow the generation of cancer cells showing gradually increased levels of SOX2, we observed a phenotypic change that morphologically results in an EMT (Figures S1 and S2). This was further confirmed molecularly by analyzing the levels of epithelial and mesenchymal markers, and by measuring the levels of EMT executers (Figure 4A–D). Despite some variability, our data indicate that cancer cells with high levels of SOX2 tend to acquire a phenotype that resembles a more mesenchymal state. Since EMT is a process shared by development and cancer that is linked with the capacity to switch cellular states, a process that is related to cell migration capacity, ³⁷ we decided to test whether SOX2‐expressing cancer cells show an increased capacity to migrate. Our results clearly show that higher levels of SOX2 in different cancer cells cause increased migration. This enhanced migratory activity is also observed after TP53 downregulation through the increase in SOX2 levels since the knockdown of SOX2 abrogates this effect (Figure 5A–C). Reinforcing our experimental results, the link between altered expression of SOX2 caused by mutation of TP53 and a more migratory and metastatic phenotype could also be substantiated with patient‐derived samples in which we observed a correlation between these parameters (Table 1), and through the analyses of public datasets (Figure S4).

Altogether, our results show that among the range of mutations and genetic alterations occurring to cancer cells, those that impact the regulation of SOX2 expression and that act in concert with TP53 deficiency lead to increased expression of this developmental factor, conferring new properties to cancer cells related to a more plastic cell identity and increased migration. A better understanding of the regulation of SOX2 expression might provide clues for therapeutic interventions aimed at reducing cancer cell plasticity, contributing to an improved management of neoplastic diseases.

AUTHOR CONTRIBUTIONS

Patricia Lado‐Fernández: Data curation; formal analysis; writing – original draft; investigation; methodology. Jéssica M. Vilas: Investigation; methodology; formal analysis; data curation. Tânia Fernandes: Investigation. Carmen Carneiro: Investigation. Sabela Da Silva‐Álvarez: Investigation. Valentín Estévez‐Souto: Investigation. Pablo Pedrosa: Investigation. Miguel González‐Barcia: Investigation; resources. Luis E. Abatti: Investigation. Jennifer A. Mitchell: Resources; investigation. Carmen Rivas: Resources; investigation. Gema Moreno‐Bueno: Resources; investigation. Anxo Vidal: Conceptualization; funding acquisition; writing – original draft; writing – review and editing; project administration; supervision; resources. Manuel Collado: Conceptualization; funding acquisition; writing – original draft; writing – review and editing; project administration; supervision; resources.

FUNDING INFORMATION

Work in the laboratory of Manuel Collado is funded by grants from MCINN/AEI/FEDER, UE (PID2021‐125479OB‐I00) and GAIN, Xunta de Galicia (IN607D2021/08 and IN607B2024/13). The laboratory of Anxo Vidal is funded by MCINN/AEI/FEDER, UE (PLEC2022‐009476), and Xunta de Galicia (ED431C 2023/10). Patricia Lado‐Fernández was supported by a predoctoral fellowship from GAIN, Xunta de Galicia (ED481A‐2020/066) and funds from Deputación da Coruña (BINV‐CS/2023 #2023000020568); Jéssica M. Vilas and Pablo Pedrosa were supported by predoctoral fellowships from GAIN, Xunta de Galicia (PRE/2012/317 and IN606A‐2017/002, respectively); Sabela Da Silva‐Álvarez was supported by a postdoctoral fellowship from GAIN (IN606B‐2021/011), Xunta de Galicia, and Juan de la Cierva contract Grant (JDC2022‐049032‐I) funded by MICIU/AEI/10.13039/501100011033 and by European Union NextGenerationEU/PRTR.

CONFLICT OF INTEREST STATEMENT

The authors have no relevant financial or non‐financial interests to disclose.

Supporting information

Data S1. Supporting Information.

Lado‐Fernández P, Vilas JM, Fernandes T, et al. Transcriptional repression of SOX2 by p53 in cancer cells regulates cell identity and migration. Int J Cancer. 2025;157(5):980‐992. doi: 10.1002/ijc.35490

DATA AVAILABILITY STATEMENT

All data generated or analyzed during this study are included in this published article. The materials used in this study are available from the corresponding authors, Anxo Vidal and Manuel Collado, upon reasonable request.