ZNF185 is a p53 target gene following DNA damage

The transcription factor p53 is a key player in the tumour suppressive DNA damage response and a growing number of target genes involved in these pathways has been identified. p53 has been shown to be implicated in controlling cell motility and its mutant form enhances metastasis by loss of cell directionality, but the p53 role in this context has not yet being investigated. Here, we report that ZNF185, an actin cytoskeleton-associated protein from LIM-family of Zn-finger proteins, is induced following DNA-damage. ChIP-seq analysis, chromatin crosslinking immune-precipitation experiments and luciferase assays demonstrate that ZNF185 is a bona fide p53 target gene. Upon genotoxic stress, caused by DNA-damaging drug etoposide and UVB irradiation, ZNF185 expression is up-regulated and in etoposide-treated cells, ZNF185 depletion does not affect cell proliferation and apoptosis, but interferes with actin cytoskeleton remodelling and cell polarization. Bioinformatic analysis of different types of epithelial cancers from both TCGA and GTEx databases showed a significant decrease in ZNF185 mRNA level compared to normal tissues. These findings are confirmed by tissue micro-array IHC staining. Our data highlight the involvement of ZNF185 and cytoskeleton changes in p53-mediated cellular response to genotoxic stress and indicate ZNF185 as potential biomarker for epithelial cancer diagnosis.

In addition to its roles in cell death, p53 has also been implicated in cytoskeleton assembly, cell motility and mechanosignaling, as negative regulator of cancer cell mobility, invasion and metastasis [18][19][20]. Integrin expression and signalling pathways, which play a key role in tumour cell invasion and metastasis, have been reported to be regulated indirectly by p53 [18]. For instance, Nutlin-3a, an MDM2 antagonist that acts as p53 activator, decreases the expression of integrin alpha5 in colorectal cancer and glioma cells [21,22]; also the expression of integrin beta3 decreases upon DNA-damage in wild-type p53 expressing cells [23]. p53 also regulates focal adhesion and Rho signalling pathways by regulating Rho GTPase activity [24] and effector protein genes of RhoA/RhoC and Cdc42 pathways [25][26][27][28]. In addition, F-actin formation is negatively or positively regulated by p53 in response to DNA damage depending on the anti-tumour drug used and cell type. For instance, while doxorubicin increases the expression of RhoC and LIM kinase 2 in a p53dependent manner promoting actin stress fibers formation [29], etoposide and camptothecin attenuate this process through p53-dependent expression of RhoE [30]. It has been also reported that upon etoposidemediated DNA damage, p53 alters actin cytoskeleton by transcriptionally induction of the expression of the cytoskeleton adaptor protein ankyrin-1 [31]. The relevance of cytoskeleton remodelling and cell mobility in tumours is evidenced by the fact that mutant p53 promotes tumour cell invasion and results in loss of directionality during migration [32]. Cytoskeleton remodelling and cell migration in cancer is a complex process and is controlled by many proteins and pathways, the specific role of p53 in these mechanisms is not yet completely understood.
Here, we describe a novel p53 target gene, ZNF185, which codifies for a Zn-finger protein belonging to LIM-family, activated upon genotoxic stress caused by DNA-damaging drug etoposide. ZNF185 itself is not necessary for p53-dependent cell cycle arrest and apoptosis, yet its silencing affects actin cytoskeleton changes and cell polarity upon etoposide treatment. At mRNA and protein level, ZNF185 is strongly reduced in different types of epithelial tumours, including skin and head and neck squamous cell carcinomas, suggesting that depletion of ZNF185 in cancer cells facilitate cancer cell migration and spreading.

ZNF185 is a p53 target gene
We have previously shown that the p53 family member p63, using a novel promoter region and a specific enhancer, directly regulated ZNF185 expression in keratinocytes [33]. To investigate whether also p53 could regulate ZNF185 expression and expand the p53 target genes involved in cytoskeleton regulation and cell polarity, we further analysed ZNF185 promoter region using UCSC genome browser (Fig 1A). We observed several regions showing high accessibility and conservation between the species, and an enrichment in different transcription factors (TF) binding. Analysis of the publicity available ChIP-seq data for p53 performed in MCF7 cells after p53 stabilization by nutlin (GSE86164, [34]), revealed a strong peak within ZNF185 promoter only in nutlin-treated cells (Fig 1B), suggesting p53 involvement in regulation of ZNF185 transcription. Using previously described bioinformatic tool for p53 binding site (bs) prediction [35], we identified a putative binding site for p53 within the genomic region corresponding to the peak from the ChIP seq shown in Fig 1B (Fig 1C). Interestingly, this region is conserved only in primates and is absent in other species (Fig 1C). To confirm physical binding of p53 on ZNF185 promoter, we performed ChIP assay in p53 Tet-On inducible SaOs-2 cells previously generated in the laboratory [36]. As a positive control, we used the promoter of CDKN1A, the gene coding for p21 ( Fig  1D). To confirm if p53 could directly regulate ZNF185 expression binding its promoter, we cloned the genomic locus harbouring p53 bs up-stream of the luciferase reporter gene. Luciferase activity assay showed a strong activation (120-fold, P<0.01) upon p53 overexpression. Interestingly, the overexpression of the two different p53 mutants frequently found in human cancers (R175H and R273H) did not show any strong activation compared to the control (Fig 1E), indicating that ZNF185 is target of wild-type p53. Furthermore, the substitution of cytosines and guanines to adenines within ZNF185 promoter sequence led to dramatic decrease of the luciferase activity (83% reduction, P<0.01) upon p53 overexpression (Fig 1F). To investigate if p53 is able to regulate ZNF185 transcription, we induced p53 expression by doxycycline in SaOs-2 Tet-On cells and measured by RT-qPCR a significant increase of ZNF185 mRNA after p53 induction paralleling CDKN1A increases (15-fold for ZNF185 and 5-fold for CDKN1A at 24 h of induction, P<0.05; Fig 1G). We also confirmed this result changing cellular system. Indeed, overexpression of p53 in H1299 also led to a 20-fold increase of ZNF185 mRNA, meanwhile the overexpression of two p53 mutants didn't show any significant modulation ( Fig 1H). Altogether, these data indicate ZNF185 as a bona fide transcriptional target of wild-type p53.

ZNF185 is up-regulated upon DNA damage
We investigated whether ZNF185 is transcribed as consequence of p53 activation following DNA damage.
Using two different carcinoma cell lines harbouring wild-type p53 (HCT116 and MCF7), we analysed ZNF185 expression after 0, 8, 16, and 24 hours of etoposide treatment. In both cases, we saw p53 stabilization as indicated by the western blots ( AGING B) and, as a consequence of p53 activation, significant up-regulation of ZNF185 mRNA (3-4-fold over control at 24 h of etoposide treatment, P<0.05), and p21 as positive control, both at mRNA and protein levels (Fig  2A-B). Interestingly, analysis of publicity available ChIP seq data (GSE56674, [37]) for p53 performed in keratinocytes showed that p53 binds to the locus within ZNF185 promoter identified by us in this study. Moreover, this binding is observed only upon cisplatin or doxorubicin treatment ( Fig 2C). As a model of basal layer keratinocytes, we used the commercial cell line of immortalized keratinocytes, Ker-CT. We confirmed that also in Ker-CT cells etoposide treatment leads to p53 stabilization and ZNF185 up-regulation both at mRNA (3-fold, P<0.01) and protein levels ( Fig 2D). To confirm that ZNF185 up-regulation is p53-dependent, we performed siRNA-mediated knock-down of p53 in Ker-CT cells. As expected, depletion of p53 abolished up-regulation of ZNF185 upon etoposide treatment ( Fig  2E). Since the major source of DNA damage in the human keratinocytes is UV irradiation, we irradiated Ker-CT cells and analysed ZNF185 level. Also in this case, we saw an up-regulation of ZNF185 at protein level. Altogether, these findings show that upon DNA damage we detected up-regulation of ZNF185 expression in p53-dependent manner both in tumour cell lines and in normal human keratinocytes.

ZNF185 is involved in the cytoskeleton remodelling upon DNA damage
Since the major functions of p53 activation upon DNA damage relate to the cell cycle arrest and apoptosis, we asked whether depletion of ZNF185 could alter cell AGING cycle content under this specific stress condition. We performed siRNA mediated knock-down of ZNF185 in Ker-CT cells with two different siRNAs and treated the cells with etoposide. Cytofluorimetric analysis did not reveal any significant modulation in cell cycle distribution and apoptosis respect to the control ( Fig  3A). It was previously reported that ZNF185 regulates proliferation of prostate cancer cells [38], to further investigate this point we generated Ker-CT cell line, stably expressing shRNA against ZNF185 (shZNF185).

AGING
We performed the EdU-incorporation assay to evaluate the number of cells in S-phase, but we did not observe any significant difference in cell proliferation respect to the control (Fig 3B). Given that several LIM-domain Zn-fingers can migrate into the nucleus under stress conditions [39], we asked whether DNA damage can alter ZNF185 localisation. We found that ZNF185 localised in the cytoplasm and at the cell periphery ( Fig  3C) also after etoposide treatment. Due to the presence of the actin-interacting domain within ZNF185 protein, we hypothesised that ZNF185 could be involved in cytoskeleton remodelling upon DNA damage. To this aim, we performed immunofluorescence analysis using phalloidin as a marker of filamentous actin and vinculin as a marker of focal adhesion. Under normal conditions, most of the cells had migratory phenotype showing vinculin accumulation on the leading edge. After etoposide treatment, cells lost planar polarity as visualised by homogeneous vinculin distribution on the cell periphery (percentage of polarized cells from 100% to 35%). Surprisingly, this phenotype was abolished in the shZNF185 cells which retained planar polarization also upon etoposide treatment (percentage of polarized cells from 100% to 82%) ( Fig 3D). Altogether, these results suggest that ZNF185 is involved in the loss of the planar polarity of cells upon DNA damage.

ZNF185 is down-regulated in epithelial cancers
p53 is frequently mutated in human cancers and we have shown that ZNF185 is positively regulated by wild-type p53, therefore we asked if ZNF185 level is decreased in epithelial cancers and particularly in the skin carcinomas. Firstly, we analysed ZNF185 mRNA expression in different types of epithelial cancers from TCGA collection. Five cancer types -prostate adenocarcinoma (PRAD), chromophobe renal cell carcinoma (KICH), head and neck squamous cell carcinoma (HNSC), oesophageal carcinoma (ESCA), and adenoid cystic carcinoma (ACC) -showed a significant decrease in ZNF185 mRNA level respect to the normal tissues from both TCGA and GTEx database (Fig 4A). Furthermore, we analysed correlation between the expression of ZNF185 and two distinct targets of p53 -PERP and CDKN1A. Interestingly, a strong positive correlation was observed only in the cancers AGING arising from squamous epithelia -oesophageal and head and neck carcinomas (Fig. 4B). Since there are only few datasets of skin cancer with a very low number of samples, we decided to analyse ZNF185 expression in skin cancer by immunohistochemistry using tissue microarray, containing 42 samples of the cutaneous squamous cell carcinoma (cSCC), 14 samples of the cutaneous basal cell carcinoma (cBCC), 12 samples of cutaneous malignant melanoma (cMM), and 10 samples of the normal skin. As a marker of proliferation, we used Ki67. Analysis of ZNF185 expression pattern at protein level in the normal skin confirmed previously published data from our laboratory [33], in which ZNF185 highest expression occurs in the differentiated spinous and granular layers ("SS/SG") of the epidermis with low expression in the proliferating basal layer AGING ("SB"). Cornified layer ("SC") and dermis ("D") were found negative for ZNF185 ( Figure 5A). Analysis of skin cancer samples revealed that ZNF185 expression is dramatically down-regulated in the cutaneous squamous and basal cell carcinoma ("cSCC" and "cBCC") and malignant melanoma ("cMM") samples ( Figure 5B). Furthermore, ZNF185 was found only in welldifferentiated subpopulations of squamous cell carcinoma ("WD" of cSCC) in contrast to poorlydifferentiated basal-like subpopulations ("PD" of cSCC) ( Figure 5C). All the tumour samples showed a significant decrease (P<1x10 -5 ) of ZNF185 H-score respect to the differentiated layers of the normal epidermis ( Figure 5D). These findings reveal a dramatic down-regulation of ZNF185 at protein level in the skin cancer and suggest that ZNF185 could be a potential biomarker for epithelial cancer diagnosis and prognosis.
Recently, the importance of actin-cytoskeleton remodelling and cell polarity during cancer cell spreading and metastasis has emerged [114][115][116][117], and p53 is in part involved in counteracting this specific aspect. Indeed, wild-type p53 can influence actin cytoskeleton dynamics controlling integrin and cadherin signalling and extracellular matrix degradation, suppressing EMT via different pathways [20,23,[118][119][120]. Interestingly, also tumour microenvironment influences actin cytoskeleton [121,122], in part repressing wild-type p53 functions [123]. In fact, when p53 is inactivated, cancer cells invasion increases [124]. Here, we demonstrated that p53 wild-type transcriptionally activates ZNF185 in cells upon DNA damage, which could make part of p53 negative regulation of cancer cell mobility, invasion and metastasis. Interestingly, similarly to another p53 target gene Rap2B [125], down-regulation of ZNF185 does not affect cell cycle progression or cell death, but its silencing abolishes the actin cytoskeleton rearrangements and cell polarity changes upon etoposide treatment.
ZNF185 is an actin-cytoskeleton-associated Lin-l 1, Isl-1 and Mec-3 (LIM) domain-containing protein [126]. The domain interacting with actin is located at the Nterminus, and it is necessary to mediate actincytoskeleton targeting of ZNF185, while the C-terminus LIM domain is dispensable for actin binding [38]. The LIM domain is a protein-protein interaction domain found in a wide range of proteins whose functions are related to the dynamics of the cytoskeleton [39,127,128]. In keratinocytes and epidermis ZNF185 has been described highly expressed in differentiating conditions, physically interacting with E-cadherin, a component of the adherens junctions, one of the critical cell-cell adhesive complexes crucial in the pluristratified epithelia [33]. ZNF185 involvement in pathologies, such as cancer [38], has not been completely investigated yet. Few studies reported ZNF185 as an unfavourable prognostic marker in ductal carcinoma of pancreas [129]. Its expression was found upregulated in colon cancer and likely correlated with liver metastasis [130]. On the other hand, other studies described epigenetic silencing of ZNF185 associated with high grade and metastatic prostate tumours [131], lung tumours and head and neck squamous cell carcinomas [33, [132][133][134]. Recently, it was reported that ZNF185 expression is negatively correlated with lymph node metastasis of lung adenocarcinoma and its overexpression leads to down-regulation of p-AKT, p-GSK3β, VEGF and MMP-9 expression [135]. These studies suggest a possible tumour-specific contribution of ZNF185 expression in tumour formation.
We confirmed ZNF185 down-regulation in different epithelial tumours and, by analysing the expression of ZNF185 at protein level, we found a significant decrease of ZNF185 in all the tumour samples analysed. Moreover, we found ZNF185 positive signal only in well-differentiated subpopulations of squamous cell carcinoma in contrast to poorly-differentiated basal-like aggressive subpopulations, suggesting a tumoursuppressor role of ZNF185. The possible involvement of ZNF185 in cytoskeleton remodelling upon DNA damage suggests its role in the metastasis promotion which is in line with previous reports [135]. The identification of p53-ZNF185 axis could contribute to determine how p53 controls cell spreading by actin cytoskeletal remodelling, in which both the mechanical properties of the cytoskeleton of the cell as well as the microenvironment of the tumour cells seem to play an important role. Further investigation on the mechanisms by which p53 controls actin cytoskeleton reorganization and cell polarity, including the identification of novel target genes and pathways, would possibly be useful in developing new anti-cancer strategies and therapies.

Western blotting
The cells were collected by trypsinization, washed in PBS and lysed in RIPA buffer (50 mM Tris-cl pH 7.4, 150 mM NaCl, 1 % NP40, 0.25 % Na-deoxycholate, 1 mM AEBSF, 1 mM DTT). 20-50 µg of total protein extracts were resolved in SDS polyacrylamide gel using the Mini-PROTEAN Tetra cell System (Bio-Rad, Hercules, CA, USA) and blotted onto a Hybond PVDF membrane (GE Healthcare, Chicago, IL, USA) using the Bio-Rad Mini Trans-Blot Cell system Bio-Rad). Membranes were blocked with 5 % non-fat dry milk (Bio-Rad) in PBS/0.1 % Tween-20 buffer, for 1 h at room temperature in agitation. Membranes were incubated with primary antibodies over night at +4 °C, washed and hybridized for 1 h at room temperature with the appropriate horseradish peroxidase-conjugated secondary antibodies (goat anti-rabbit and goat antimouse antibodies, Bio-Rad). Detection was performed with the ECL chemiluminescence kit (

RNA extraction and RT-qPCR analysis
Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany) following the manufacturer's protocol. Total RNA (1 µg) was used for cDNA synthesis by GoScript Reverse Transcription System kit (Promega, Madison, WI, USA). RT-qPCRs were performed using the GoTaq Real-Time PCR System (Promega) in Applied Biosystems 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) using appropriate qPCR primers (Supplementary Table 1). TBP was used as housekeeping gene for normalization. The expression of each gene was defined from the threshold cycle (Ct), and relative expression levels were calculated using the 2 −ΔΔCt method. All reactions were run in triplicate.

Analysis of ZNF185 genomic locus
To analyse ZNF185 genomic locus, different publicity accessible high-throughput sequencing data from ENCODE database (ChIP seq for H3K4me3 in different cell lines, CpG islands, DNase clusters, vertebrate conservation, TF binding) were visualised in the UCSC Genome Browser. Several ChIP-seq data from NCBI GEO database were analysed to assess p53 binding to the ZNF185 promoter locus (GSE56674 ( [37]) and AGING GSE86164 ( [34])). To identify putative p53 binding sites was used the "p53 scan" software [136]. The conservation analysis of ZNF185 promoter locus was performed within UCSC genome browser.
Chromatin immunoprecipitation assay 1x10 6 of SaOs Tet-On p53 cells, induced to overexpress p53 for 16 h, were used for ChIP assay. Cells were collected, fixed in 1% formaldehyde, and subjected to sonication for DNA shearing. The chromatin immunoprecipitation was performed with HA antibody (BioLegend) or unspecific immunoglobulin G (IgG, Invitrogen) using the MAGnify ChIP Kit (Invitrogen). Specific primers were used to amplify the putative p53 response element identified within ZNF185 promoter region (Supplementary Table 1).

Luciferase activity assay
Promoter region of ZNF185 containing the putative p53 binding site was amplified from human genomic DNA using specific primers (Supplementary Table 1). PCR products were digested by Kpn1/Nhe1 restriction enzymes (New England Biolabs, Ipswich, MA, USA) and subcloned into the pGL3-Promoter reporter vector (Promega). The constructs were completely sequenced. For luciferase activity assay, a total of 1.2x10 5 H1299 cells were seeded in 12-well dishes 24 h before transfection. 100 ng of pGL3 reporter vector, 2 ng of pRL-CMV-Renilla luciferase vector (Promega) and 300 ng of either pcDNA-HA-p53, pcDNA-HA-p53-R175H, pcDNA-HA-p53-R273H, or empty pcDNA-HA vector (as a control) were cotransfected using Effectene transfection reagent according to the manufacturer's instructions (Qiagen). The luciferase activity was measured 24 h after transfection using a Dual Luciferase Reporter Assay System (Promega). The light emission was measured over 10 sec using a Lumat LB9507 luminometer (EG&GBerthold, Bad Wildbad, Germany). The transfection efficiency was normalized to Renilla luciferase activity. The overexpression of p53 was confirmed by western blotting.

Mutagenesis
For mutagenesis of p53 binding site was performed a PCR on 100 ng of pGL3 vector carrying p53 binding site using specific primers (Supplementary Table 1). PCR product was digested with DpnI restriction enzyme (New England Biolabs). The presence of mutated site was confirmed by sequencing.

Cell proliferation
The incorporation of EdU during DNA synthesis was evaluated using the Click-iT EdU flow cytometry assay kit according to the manufacturer's protocol (Thermo Fisher Scientific). The cell cycle was analysed using an Accuri C6 flow cytometer (BD Biosciences). Fifteen thousand events were evaluated using the Accuri C6 (BD Biosciences) software. For cell cycle analysis, cells were fixed in 50% methanol/acetone 4:1 mix for 30 min at +4 °C, then treated with 13 Kunitz U/mL RNase for 15 min and stained with 50 µg/mL of propidium iodide for 20 min. Twelve thousand events were acquired using FACScalibur (BD Biosciences). Cell cycle distribution was calculated using FloxJo software.

Bioinformatic analysis
Analysis of ZNF185 expression in normal and tumour samples from TCGA/GTEx databases was performed using GEPIA [137]. ZNF185, PERP, and CDKN1A AGING expression data in ESCA and HNSC samples from TCGA collection were obtained using R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl/).

Statistical analysis
The significance of differences between two experimental groups was calculated using unpaired, two-tailed Student's t-test. Values of P < 0.05 were considered significant. For RT-qPCR and luciferase assay, values reported are the mean ± SD. For statistical analysis of TMA scoring was used Mann-Whitney U test. All statistical analyses were performed using GraphPad Prism 7.0 Software. Violin plots were generated in R using ggplot2 package.

AUTHOR CONTRIBUTIONS
AS, AC, AML and LA performed the research, EC designed the research, EC, AM, NDD, MAP and GM analysed the data, EC wrote the paper and all the authors read the paper and made comments.

CONFLICTS OF INTEREST
The authors declare no conflict of interest.

FUNDING
This work has been supported by grants from Associazione Italiana per la Ricerca contro il Cancro (AIRC): IG15653 (to G.M.), and by Istituto Dermopatico dell'Immacolata, IRCCS, RC to EC and joint research program Italia-Cina. This work has been also partially supported by the Medical Research Council, UK.