Effect of p53 and its N‐terminally truncated isoform, Δ40p53, on breast cancer migration and invasion

Breast cancer is the most diagnosed malignancy in women, with over half a million women dying from this disease each year. In our previous studies, ∆40p53, an N‐terminally truncated p53 isoform, was found to be upregulated in breast cancers, and a high ∆40p53 : p53α ratio was linked with worse disease‐free survival. Although p53α inhibits cancer migration and invasion, little is known about the role of ∆40p53 in regulating these metastasis‐related processes and its role in contributing to worse prognosis. The aim of this study was to assess the role of ∆40p53 in breast cancer migration and invasion. A relationship between Δ40p53 and gene expression profiles was identified in oestrogen‐receptor‐positive breast cancer specimens. To further evaluate the role of Δ40p53 in oestrogen‐receptor‐positive breast cancer, MCF‐7 and ZR75‐1 cell lines were transduced to knockdown p53α or Δ40p53 and overexpress Δ40p53. Proliferation, migration and invasion were assessed in the transduced sublines, and gene expression was assessed through RNA‐sequencing and validated by reverse‐transcription quantitative PCR. Knockdown of both p53α and ∆40p53 resulted in increased proliferation, whereas overexpression of ∆40p53 reduced proliferation rates. p53α knockdown was also associated with increased cell mobility. ∆40p53 overexpression reduced both migratory and invasive properties of the transduced cells. Phenotypic findings are supported by gene expression data, including differential expression of LRG1, HYOU1, UBE2QL1, SERPINA5 and PCDH7. Taken together, these results suggest that, at the basal level, ∆40p53 works similarly to p53α in suppressing cellular mobility and proliferation, although the role of Δ40p53 may be cell context‐specific.

Breast cancer is the most diagnosed malignancy in women, with over half a million women dying from this disease each year. In our previous studies, Δ40p53, an N-terminally truncated p53 isoform, was found to be upregulated in breast cancers, and a high Δ40p53 : p53a ratio was linked with worse disease-free survival. Although p53a inhibits cancer migration and invasion, little is known about the role of Δ40p53 in regulating these metastasis-related processes and its role in contributing to worse prognosis. The aim of this study was to assess the role of Δ40p53 in breast cancer migration and invasion. A relationship between D40p53 and gene expression profiles was identified in oestrogen-receptor-positive breast cancer specimens. To further evaluate the role of D40p53 in oestrogen-receptorpositive breast cancer, MCF-7 and ZR75-1 cell lines were transduced to knockdown p53a or D40p53 and overexpress D40p53. Proliferation, migration and invasion were assessed in the transduced sublines, and gene expression was assessed through RNA-sequencing and validated by reverse-transcription quantitative PCR. Knockdown of both p53a and Δ40p53 resulted in increased proliferation, whereas overexpression of Δ40p53 reduced proliferation rates. p53a knockdown was also associated with increased cell mobility. Δ40p53 overexpression reduced both migratory and invasive properties of the transduced cells. Phenotypic findings are supported by gene expression data, including differential expression of LRG1, HYOU1, UBE2QL1, SERPINA5 and PCDH7. Taken together, these results suggest that, at the basal level, Δ40p53 works similarly to p53a in suppressing cellular mobility and proliferation, although the role of D40p53 may be cell context-specific.

Introduction
Breast cancer is the most commonly diagnosed malignancy in females and accounts for over half a million deaths in women worldwide [1]. TP53 is the most frequently mutated gene in cancer. Wild-type p53 suppresses tumorigenesis through multiple pathways, such as inducing cell-cycle arrest, apoptosis, senescence and directly responding to oncogenic stress, whereas mutant p53 can gain dominant negative functions, contributing to tumorigenesis [2]. Intriguingly, p53 mutations are not ubiquitous in breast cancers (less than 25% of all cases) [3], indicating that other mechanisms are involved in ablating the canonical function of p53.
In 2005, it was discovered that p53 can be expressed as distinct protein isoforms other than the wild-type full-length p53 (wtp53a) [4]. Truncation can occur at the N terminus (D40p53, D133p53 and D160p53) or C terminus (a, b and c) or both, giving rise to twelve isoforms retaining different proportions of the three main functional domains: transactivation domain (including TADI and II), DNA-binding domain (DBD) and oligomerisation domain (OD) [4][5][6]. These domains are critical for p53a to function. The regulatory function of the truncated isoforms is not limited to their interaction with p53a [7]; D40p53a [8] and D133p53a [9] can function independently of p53a.
D40p53a (Δ40p53 from here on) lacks the first 40 amino acids of p53a and can arise via two mechanisms. It has been reported by us and others that Δ40p53 is primarily generated by alternative splicing, which retains intron 2 encompassing several stop codons, thereby preventing p53a translation [10,11]. D40p53 can also be produced by alternative translation at AUG40 [5]. The missing region encodes for TADI, resulting in immunity to human double minute-2 (HDM2)-mediated degradation and partial loss of transactivation ability [12]. The function of D40p53 is difficult to interpret, and different roles have been reported so far (for review, [13]).
Early studies showed that D40p53 heterotetramerisation led to elevated nuclear export of p53a, a reduction in the transactivation of certain p53a target genes and inhibition of p53a-mediated apoptosis [5]. Hafsi et al. co-transfected p53-null cell lines with expression vectors for D40p53 and p53a at different amounts and measured the transactivation activity of a reporter gene containing the p53a response element of HDM2. They found that if D40p53 was expressed at levels higher than p53a (≥ 3 fold), D40p53 could inhibit p53a's function, whereas lower or equivalent levels of D40p53 compared to p53a, had diverse effects that were cell line specific [12]. It is deducible that high levels of D40p53 compete with p53a in tetramer formation, and therefore, the canonical p53a function was compromised.
Other studies showed that overexpression of D40p53 induced apoptosis in melanoma cell lines [14] and reduced proliferation regardless of p53a status in hepatocellular carcinoma cell lines [15], where D40p53 stimulated the canonical function of p53a. Bourougaa et al. (2010) reported that D40p53 induced G2 arrest, while p53a induced G1 arrest, further establishing an independent role for D40p53 [16]. Using point mutations to inactivate phosphorylation residues, it was shown that TADI and TADII induce the expression of distinct genes, and hence, it is likely that the subset of genes transactivated by D40p53, which lacks TADI, is distinct from that of p53a [17]. Additionally, D40p53 has been shown to play a role in development. In mice, overexpression of D40p53 led to the activation of insulin growth factor 1 receptor (IGF-1R), maintainence of proliferation and self-renewal potential in embryonic stem cells during early development [18].
D40p53 is overexpressed in multiple cancers such as melanoma and breast cancer, and its expression has been correlated with prognostic and therapeutic outcomes [19,20], suggesting a role for D40p53 in disease progression. Our group demonstrated the importance of D40p53 expression in breast cancer. It is the most highly expressed p53 isoform in breast cancer at the mRNA level, and it is significantly upregulated in tumours and cell lines compared to the normal breast. A high D40p53 : p53a ratio (> 0.7) is significantly associated with worse disease-free survival (HR 2.713) [19]. These studies suggest that D40p53 may be involved in cellular functions that promote the aggressiveness of breast cancer, but functional studies demonstrating this are lacking, particularly on the endogenously expressed isoform.
As a tumour suppressor, p53a regulates the expression of a plethora of genes and microRNAs linked to the inhibition of proliferation, migration and invasion. For instance, p53a is known to regulate the expression of cell cycle regulatory genes CDKN1A and RB1, growth factor receptors EGFR and MET, and matrix metalloproteinases MMP2 and MMP9 [21], highlighting p53a's regulatory role in proliferation and migration. P53a also induces the expression of miR-34, which anatagonises NOTCH, ZEB1 and SNAI1, further exemplifying how p53a promotes cell adhesion and reduces proliferation [22]. At the functional level, p53a's role in repressing migration and invasion has also been shown [23]. Hence, there is a growing body of evidence suggesting a role for p53a in metastasisrelated processes; however, the role of Δ40p53 in this context has not been examined. In this study, we examined the role of D40p53 in migration, invasion and in the regulation of gene expression to determine whether this could provide an explanation for the association of high D40p53 : p53a with worse survival outcomes in breast cancer as identified in our previous study [19]. Our results showed that molecular inhibition of p53a was associated with increased cell mobility, confirming the previous association of loss of p53a function and increased metastatic potential. Overexpression of Δ40p53 reduced both migratory and invasive properties of the transduced cells. Inhibition of both p53a and Δ40p53 resulted in increased proliferation, while overexpression of Δ40p53 reduced proliferation rates. These phenotypic findings are supported by gene expression data. Taken together, these results suggest that at the basal level Δ40p53 works similarly to p53a in suppressing cellular mobility and proliferation.

Breast cancer samples
Breast cancer samples were acquired from the Australian Breast Cancer Tissue Bank and have previously been described [19,24]. All patients whose tissue samples were used in this study had an understanding of and provided written consent for the use of their tissue in research. mRNA that had been previously extracted from these samples was used in our studies. The study methodologies conformed to the standards set by the Declaration of Helsinki and were approved by the Hunter New England Human Research Ethics Committee (Approval number: 09/05/20/5.02).

Human gene 1.0 array
100 ng of total RNA from FFPE samples was amplified (Ovation FFPE WTA kit) and biotinylated (Encore Biotin module) according to the manufacturers' instructions (Nugen, San Carlos, CA, USA). The arrays were scanned on an Affymetrix GeneChip Scanner 3000 7G (Affymetrix, Santa Clara, CA, USA), the data were imported to Genomic Suite 6.6 (Partek, St. Louis, MO, USA), and a robust multi-array analysis (RMA) was performed, which included log 2 transformation, background correction, quantile normalisation and summarisation of the probe features resulting in a set of expression signal intensities. Unsupervised hierarchical clustering was performed on genes that were found to be differentially expressed in invasive ductal carcinomas (IDCs) with high Δ40p53 compared to IDCs with low Δ40p53 (P < 0.5; fold change > |1.5|). Correction for multiple testing was performed using the Benjamini-Hochberg procedure.

Cell culture
The human breast cancer cell lines MCF-7 and ZR75-1 (with wtp53a) were kind gifts from Professor Christine Clarke (Westmead Millennium Institute, University of Sydney, Australia) and Dr Judith Weidenhofer (The University of Newcastle, Australia), respectively. Before the beginning of the experiments, the cell lines were authenticated by the Australian Genome Research Facility (Fine Mapping and Custom Genotyping, 6173, 100% for MCF-7 and 84.62% for ZR75-1). Briefly, 1 9 10 6 cells were collected and DNA was extracted by using Promega Genomic DNA purification Kit (Promega, A1120, Madison, WI, USA). GenePrint 10 system (Promega, B9510) was used to amplify 9 human loci from the extracted DNA and the amplified PCR products were used to generate a genetic profile, which was then compared to the profile provided by the supplier of the cell line. Cells were routinely cultured and passaged (up to the 20th passage after thawing) using DMEM [Life Technologies, 21063029), with 10% fetal bovine serum (FBS, Sigma-Aldrich, F9423, St. Louis, MO, USA)], 200 mM Lglutamine (Life Technologies, 25030081) and 2 µgÁmL À1 insulin (Sigma-Aldrich, 19278). Cells were kept in a cell culture incubator at 37°C with 5% CO 2 and humidity and routinely tested for mycoplasma according to the manufacture's recommendations (MycoAlert PLUS, Lonza, LT07-701).

Vector transductions
D40p53-overexpressing MCF-7 cells were created by transducing MCF-7 cells with a lentiviral expression construct containing D40p53. To generate the lentiviral construct, the empty vector LeGO-iG2-puro+ was linearised by digestion with BamH1 and Sbf1. D40p53 cDNA was amplified by PCR using primers, which incorporated 14 bps of sequence homologous to the LeGO-iG2-puro+ vector. The insert was then incorporated into the LeGO-iG2-puro+ vector by recombination (In-Fusion, Clontech). The D40p53 lentivirus was then produced in 293T cells and used to transduce MCF-7 cells. The resulting MCF-7 cells were maintained in puro-media as described above. Cells with stable D40p53-overexpression are referred as MCF-7-D40p53 and MCF-7-LeGO (empty vector control).

Proliferation assay
All MCF-7 sublines were plated into 96-well plates at a seeding density of 3000 cellsÁwell À1 and monitored using the IncuCyte TM ZOOM (Essen Bioscience, Ann Arbor, MI, USA). Confluence was calculated with integrated algorithms. All ZR75-1 sublines were plated at a seeding density of 1 9 10 4 cellsÁwell À1 into four individual 96-well plates to be processed every 24 h until 96 h. Cell Titer Glo Ò 2.0 reagent (Promega, G9241) was applied to each plate according to the manufacturer's instructions and luminescence was measured using a Cytation 3 (BioTek, Winooski, VT, USA). All experiments were repeated three times on separate days with three technical replicates.

Migration/Invasion assay based on the wound healing method
Migration and invasion were measured with the wound healing assay as previously described [27]. Briefly, a coated (100 lLÁmL À1 matrigel, Sigma-Aldrich, E6909) 96-well plate (invasion assay) or an uncoated 96-well plate (migration assay) was seeded with cells overnight to achieve a confluent cell monolayer for the next day prior to the scratch. Plates for the migration/invasion assay were scratched using the 96-well WoundMaker TM (Essen Bioscience). Dislodged cell sheets were washed away with prewarmed media and fresh media was added (migration assay). For the invasion assay, matrigel was diluted to 125 lgÁmL À1 with cold fresh media as previously described [28] and 50 lL was added to wells before gelation. Wounded cells were placed into the IncuCyte (Essen Bioscience) and monitored until the wounds closed.

Transwell migration assay
All cell lines were cultured to 80% confluence and resuspended in DMEM. 1 9 10 5 cells were seeded into transwell inserts (polycarbonate membrane with 8-lm pores, 24-well format, Sigma-Aldrich, CLS3422) in triplicate, and allowed to migrate into the lower chamber with puro-media (10% FBS) for 24 h (MCF-7 sublines) or 48 h (ZR75-1 sublines) under standard cell culture conditions. Cells were fixed with 3.7% formaldehyde at room temperature for 15 min and cells on the upper side of the transwell inserts were removed before the inserts were stained with crystal violet. Migrated cells were counted under a light microscope with a 10x objective. Cell numbers in five random fields of view were averaged per insert. All experiments were repeated three times on separate days.

Transwell invasion assay
Transwell inserts were coated with 125 lgÁmL À1 matrigel overnight under standard cell culture conditions. Transduced MCF-7 and ZR75-1 sublines were cultured to 80% confluence and resuspended in DMEM. 1 x 10 5 cells were seeded into transwell inserts in triplicate and allowed to migrate into the lower chamber with puro-media (10% FBS) for 24 h (MCF-7 sublines) or 48 h (ZR75-1 sublines) under standard cell culture conditions. Transwell inserts and invaded cell numbers were processed as described in the migration assay. All experiments were repeated three times on separate days.

RNA-seq
All sublines were seeded in triplicate into 6-well plates for 24 h (3-4 9 10 5 to achieve around 70% confluence the next day and mapped to Human GRCh37 Assembly using STAR [29]. Differential expression was computed in DESeq2. Genes of ≥ 50 counts, log 2 (fold change) ≥ |1|, and a false discovery rate (FDR)-adjusted P-value ≤ 0.05 were deemed to be differentially expressed.

Gene set enrichment analysis
Gene set enrichment analysis (GSEA) was performed by Enrichr [30] using GO Biological Process 2018.

Statistical analyses
For cell line experiments, unpaired student t-tests were performed for two comparisons, and one-way ANOVA or two-way ANOVA for multiple comparisons, corrected for multiple comparisons using the Dunnett's or Sidak's test, respectively. All results are the mean of three independent experiments, and error bars represent the standard deviation (SD) or are otherwise indicated. All statistical analyses were performed using GRAPHPAD Prism v. 6.0. An adjusted P-value of < 0.05 was considered statistically significant.

Differential gene expression in ER+ IDCs with high vs low Δ40p53
Using RT-qPCR, D40p53 expression levels were measured in 38 oestrogen-receptor-positive (ER+) and 16 ER-IDCs (Grade 1 and 2) for which gene expression data has been previously published by our group [24].
To determine genes that are affected by endogenous Δ40p53, both ER+ and ER-IDCs were classified based on high or low Δ40p53 expression (as compared to the median expression level of all IDC cases) (Fig. 1A).  Table 1) were differentially expressed in tumours expressing high Δ40p53 vs low Δ40p53 (> | 1.5|-fold, P < 0.05, FDR 5%) (Fig. 1B). The same pattern of differential expression was not observed in ER-breast cancer cases (Fig. 1C). Thus, the transcriptional effects of Δ40p53 may be ER-dependent. GSEA revealed GO terms associated with immune responses mediated by cytokines, which is not surprising given the fact that the breast cancer specimens contain all cell types including the tumour cells, stroma, epithelium and the lymphatic cells. Additionally, genes involved in extracellular matrix organisation, such as ACTN1 (actinin 1), FBLN1 (fibulin 1) and ITGB2 (integrin 2), were highlighted by GSEA, indicating endogenously higher levels of D40p53 are associated with downregulated cell mobility ( Table 2).

Establishment of Δ40p53 and p53a
knockdown in the ER+ cell lines MCF-7 and ZR75-1 To further investigate the role of Δ40p53 in ER+ breast cancer, knockdown sublines were established through shRNA transduction. ER+ MCF-7 and ZR75-1 cell lines were transduced with shRNA vectors against p53a (-shp53a), Δ40p53 (-shΔ40p53) and a nontargeting control (-shNT) ( Fig. 2A). The region targeted by the p53 shRNA (exon 2/3 junction) will inhibit the expression of all p53 isoforms with the exception of Δ40p53 (and other N-terminal variants, i.e. Δ133p53 and Δ160p53, that are transcribed from the P2 promoter). The region targeted by the Δ40p53 shRNA (intron 2) will result in the inhibition of all Δ40p53 transcripts (regardless of whether their C terminus is full-length (a) or truncated (b/c)) that are generated by alternative splicing. Additionally, a Δ40p53 overexpression model was also established by transfecting MCF-7 cells with a lentiviral construct containing the open reading frame of Δ40p53 (MCF-7-Δ40p53), or an empty lentiviral construct (MCF-7-LeGO), which served as a control. The D40p53 cDNA lacks introns; hence, only the full-length D40p53a will be overexpressed in the MCF-7 cell line. shp53a was able to inhibit the mRNA expression of p53a, but not D40p53 in both MCF-7 and ZR75-1 cells by approximately 80 and 75%, respectively (Fig. 2B, C). This reduced expression was confirmed at the protein level with the DO-1 antibody (Fig. 2D,F). Transduction with shD40p53 did not change the mRNA expression level of p53a but knocked down D40p53 specifically by 65% in MCF-7 cells and 55% in ZR75-1 cells (Fig. 2B,C). These results were confirmed at the protein level with the KJCA40 antibody (Fig. 2D,F). Interestingly, in the MCF-7-shp53a subline, the D40p53 mRNA level was increased by 1.4-fold compared to the nontargeting control (Fig. 2B).
In the overexpression model (MCF-7-Δ40p53), increased levels of p53a and Δ40p53 were detected (Fig. 2E), consistent with the stabilising effects of D40p53 on p53a [14]. Hierarchical clustering was performed on 72 transcripts found to be differentially expressed in high (red branches) vs low (blue branches) Δ40p53-expressing ERa+ tumours. (C) The 72 differentially expressed transcripts were hierarchically clustered in ER-breast cancers in high (red) vs low (blue) Δ40p53-expressing tumours. Similarity in the expression between genes (branches on left) and between samples (branches on top) was measured using Euclidean correlation. Distances between clustered branches represent the average distance. Upregulated expression is represented by red, downregulated expression is represented by blue, and equal expression is represented by grey.

Δ40p53 affects proliferation in a cell contextspecific manner
Proliferation was firstly evaluated in all sublines to determine the role of D40p53 and p53a in cell growth.
In MCF-7 cells, D40p53-overexpression led to an indicative but not significant reduction in proliferation compared to the control subline (Fig. 3I), while D40p53 knockdown led to significant increased  (Fig. 3J). Similarly, knockdown of p53a in MCF-7 cells led to significant increased proliferation, consistent with its role as a tumour suppressor [2]. In comparison, similar growth rates were observed between ZR75-1-shNT and ZR75-1-shp53a sublines, but the ZR75-1-shD40p53 subline had a significantly slower growth rate when compared by the confluencebased assay (Fig. 3K). However, confluence-based proliferation assays rely on the ability of cells of identical morphology to form a confluent monolayer. The unique morphologies amongst the ZR75-1 sublines, especially the shD40p53 subline, may therefore have affected confluence-based proliferation assays. Therefore, a metabolic proliferation assay (Cell Titer Glo Ò 2.0 end-point assay) was used to measure proliferation in ZR75-1 sublines. The metabolic proliferation assay indicated that there was no difference in proliferation between ZR75-1 sublines (Fig. 3L). The above results showed that D40p53 had cell context-specific effects on proliferation, decreasing proliferation in MCF-7 cells, but not ZR75-1 cells, even though both cell lines contain wtp53 and are ER+.

P53a knockdown enhances cell mobility, while Δ40p53 overexpression impairs it
Cell mobility was assessed through the scratch wound assay and the Transwell assay, both of which can be modified to analyse migration (no matrigel coating) or invasion (with matrigel coating). MCF-7 cells overexpressing D40p53 migrated slower compared to MCF-7-LeGO cells (empty control vector, Fig. 4A-C). Both sublines exhibited enlarged cells at the migratory front and protruding edges stretching towards the cell-free area, but the wound remained significantly larger in MCF-7-Δ40p53 cells (Fig. 4B,C). Moreover, MCF-7-D40p53 cells had greatly impaired invasion capacity and phenotype compared to MCF-7-LeGO cells (Fig. 4D-F).
We then sought to assess cell migration/invasion in the shRNA-transduced MCF-7 sublines. However, shRNA knockdown of Δ40p53 and p53a altered the cells' ability to form a uniform confluent monolayer overnight (which was achievable in MCF-7-LeGO and MCF-7-Δ40p53, Fig. 3A,B). Hence, achieving similar confluence required for wound healing assays was challenging. To circumvent this issue, transwell assays were used, which are based on equal seeding density and do not require similar confluence to be established. The number of cells that migrated (Fig. 4G) or invaded (Fig. 4H) through the membrane of the transwell inserts was not significantly different between MCF-7-shD40p53 cells and MCF-7-shNT cells. In contrast, MCF-7-shp53a cells had acquired significantly increased cell mobility, indicating p53a as the major modulator of migration and invasion in MCF-7 cells (Fig. 4G,H).
Thus, knockdown of p53a resulted in significantly increased cell migration and invasion in both MCF-7 and ZR75-1 cells, demonstrating a dominant role of the full-length isoform in these processes. In contrast, D40p53-overexpression led to reduced cell mobility, while D40p53 knockdown had no significant effect on mobility in MCF-7 or ZR75-1 cell lines.

Molecular profiles support a proliferative, migratory and invasive phenotype
To evaluate the molecular mechanisms driving differences in proliferation, migration and invasion between  ). The p53 gene has 11 exons. D40p53 lacks part of the TAD but includes part of intron 2, which shD40p53 targets. Shp53a targets the sequence that spans across exon 2/3, therefore generating isoform-specific knockdown. Knockdown of p53a and D40p53 was quantitated at the mRNA level in MCF-7 (B) and ZR75-1 (C) derived cell lines. mRNA expression levels were measured using semiquantitative real-time PCR. All real-time PCR results were normalised to the housekeeping gene GAPDH, and transduction conditions were compared to the nontargeting shRNA control (shNT). Relative expression was calculated using 2 ÀDDCt method as described [25]. Experiments were repeated three times in three technical replicates. Results are the mean of three independent experiments, and error bars represent the standard deviation (SD). Significant differences are indicated with brackets and stars by one-way ANOVA. **P < 0.01, ****P < 0.0001. p53a and D40p53 protein levels were detected by Western blot in three independent experiments using DO-1 (detecting p53a) and KJCA40 (detecting D40p53) antibodies, respectively. The protein expression levels of p53a and D40p53 are shown by representative Western blots in the MCF-7 sublines (D) including the preestablished D40p53-overexpression cells (MCF-7-D40p53 and its control MCF-7-LeGO (E) and the shRNA-transduced sublines, as well as the transduced ZR75-1 sublines (F).   Fig. S1. Cell proliferation was measured by confluence using the IncuCyte in D40p53-overexpressing MCF-7 sublines (I) and D40p53/p53a knockdown MCF-7 sublines (J). Cell proliferation of ZR75-1 sublines was measured by confluence using the IncuCyte (K) and by metabolism using CellTiter Glo â (L), normalising to the value of 24 h within each subline. Results are the mean of three independent experiments in triplicate and error bars indicate the standard deviation (SD). Unpaired t-tests and one-way ANOVA were used to identify significance. ****P < 0.0001. D40p53 knockdown/overexpression, p53a knockdown and control sublines, RNA was extracted from each subline and subjected to RNA-seq. In both MCF-7 and ZR75-1-derived sublines, knockdown of p53a altered the expression of less than 1% of the genes detected by RNA-seq, and the overlap of differentially expressed genes (DEGs) between the two sublines was limited to eight genes ( Fig. 5A-C, Table S1). Of the shared DEGs between the p53a-knockdown sublines, five of the downregulated genes (EIF4A1, SENP3-EIF4A1, AC016876.2, CD68 and SENP3) are in proximity of each other on Chr17.p13. Several putative p53 response elements (RE) have been identified upstream of CD68 in an in silico analysis [31], indicating that p53a may control the expression of these genes via a shared promoter. Three of these overlapping DEGs, EIF4A1, SENP3 and snoRNA SNORA67 (AC016876.2) are involved in translation, indicating that changes driving increased migration and invasion in p53a knockdown sublines may be accentuated at the protein level.
Several of the DEGs support the invasive-migratory phenotype of the MCF-7 p53a knockdown subline. In MCF-7-shp53a, increased expression of LRG1 and HYOU1 (Fig. 5A, Table S1; expression trends confirmed by RT-qPCR: Fig. 5I) support increased migration and invasion of the subline (Fig. 4G,H), as both genes have been linked to tumorigenicity and cell migration/invasion [32,33]. These findings also highlight the tumour suppressing function of p53a. Additionally, increased HYOU1 [32] and decreased SESN2 [34,35] expression in MCF-7-shp53a (Fig. 5A, Table S1) support the increased proliferation observed in the subline (Fig. 3J). SESN2 is a repressor of mTOR signalling and the reduction in its expression increases the activity of the pro-proliferative signalling pathway [34].
In comparison to only nine DEGs in the MCF-7-shΔ40p53 subline (Fig. 5D, Table S2), 246 differentially expressed genes were detected in the ZR75-1-shΔ40p53 subline when compared to its' vector control (Fig. 5E, Table S2), representing around 3.4% (246/7159) of the genes detected and offering a possible explanation for the morphological changes observed in this subline (Fig. 3F), and highlighting cell line specific effects of D40p53 on gene expression. GSEA did not yield any significant results (Table S3). DEGs in the ZR75-1-shΔ40p53 subline include downregulation of genes involved in cell adhesion and extracellular matrix interaction (e.g. COL16A1, AMOTL1, ADAMTSL5, SGCD), as well as downregulation of genes linked to plasma membrane structure and cytoskeletal organisation (e.g. BIN3, PICK1, PHACTR1, RHOQ). This differential expression provides further support for the altered morphology observed in these cells. Only four of the DEGs were common between the two sublines (Fig. 5F, Table S2). The lack of commonality in DEGs highlights that Δ40p53 acts in a cell context-specific manner. Downregulation of UBE2QL1 (log 2 (fold change): À1.59; FDR-adj. P-value: 0.0001; Fig. 5D,I, Table S2), a negative regulator of mTOR pathway activity, offers a possible explanation for increased proliferation (Fig. 3J), as well as the trend towards increased migration and invasion observed in the MCF-7-shΔ40p53 subline (Fig. 4G,H). While the same changes in cell behaviour (migration and invasion, Fig. 4I,J) were not observed in the ZR75-1-shΔ40p53 subline, deregulation of additional genes involved in proliferation, migration and invasion, such as decreased expression of LRG1, ZMYND8, GNA13, DHX29 and increased expression of EMILIN2, and RECK may be counteracting the reduced levels of UBE2QL1 (Fig. 5E, Table S2) by inhibiting proliferation, migration and invasion [36][37][38][39][40][41].
Overall, several of the differentially expressed genes in Δ40p53 knockdown sublines support the hypothesis that Δ40p53 acts as a tumour suppressor. Potential oncogenes upregulated in the Δ40p53 knockdown sublines include EDN1 and CCDC28B in the MCF-7 subline; and LIFR, HOXA11-AS, NACC1, FLT4, AQP3 and FAM129A in ZR75-1-shΔ40p53 cells. Similarly, decreased expression of tumour suppressor genes was In MCF-7 cells, transwell migration (G) and invasion (H) showed no significant increase in cell mobility when D40p53 was knocked down but increased cell mobility when p53a was knocked down. In ZR75-1 cells, transwell migration (I) and invasion (J) showed no significant increase in cell mobility when D40p53 was knocked down but increased cell mobility when p53a was knocked down. Results are the mean of three independent experiments, and error bars represent the standard deviation of the mean (SD). Experiments were repeated three times in triplicate. Significant differences are indicated with brackets and stars by one-way ANOVA. *P < 0.05, **P < 0.01, ****P < 0.0001.  Table S2).
Notably, no common DEGs were found between isoform-specific knockdown in MCF-7 sublines at the basal level (Fig. 5H), demonstrating loss of D40p53 or p53a alone affected separate gene sets. Together with the fact that D40p53-overexpression uniquely induced the expression of 15 genes (Fig. 5H), these results highlight a p53a-independent function of D40p53, which has also been reported by others [8]. Contrastingly, 19 DEGs were common to both ZR75-1-shp53a and ZR75-1-shD40p53 sublines, indicating some similarity in transactivation capacity between the isoforms, yet even in the ZR75-1 cells, knockdown of D40p53 affected the expression of 227 genes that were not affected by p53a knockdown (data not shown). Further, only two genes (DICER and HLTF) overlapped between MCF-7-shD40p53 and MCF-7-D40p53 (Fig. 5H), indicating that overexpression of D40p53 may have different effects on gene expression than physiological levels of the p53 isoform.

Discussion
Wtp53 is present in most breast cancers, suggesting that the canonical tumour suppressing function is compromised. We have shown previously that Δ40p53 is the mostly highly expressed p53 isoform in breast cancer, besides the full-length p53a isoform, and that a high D40p53 : p53a ratio is associated with worse disease-free survival in breast cancer patients, unveiling a link between Δ40p53 expression and p53 modulation endogenously [11]. This lead to the hypothesis that D40p53 may play a role in breast cancer progression. Gene expression array analysis identified distinct clustering of differentially expressed genes by higher or lower D40p53 expression in ER+ cases, but not in ERcases (Fig. 1B). DEGs were mostly associated with immune responses ( Table 2), indicating that D40p53 may participate in modulating p53a-mediated immune responses in ER+ tumours. Indeed, another p53 isoform, D133p53, has been shown to be associated with immunity, interfering with p53a-mediated anti-viral responses, and inducing inflammation and autoimmunity in mouse models [46]. In contrast, downregulated genes in tumours with high D40p53 expression were mostly cytoskeletal components such as ACTN1 and FBLN1, supporting D40p53's regulation of cell motility.
When D40p53 was overexpressed, p53a protein expression was also enhanced (Fig. 2E), suggesting a role for D40p53 in stabilising p53a, potentially by forming a heterotetramer and thus attenuating HDM2-mediated degradation. This protection phenomenon has been reported in Saos-2 cells by cotransfection of Δ40p53, p53a and HDM2 [12].
The canonical function of p53a is to monitor DNA integrity by inducing repair, cell cycle arrest and apoptosis. Loss of p53a or mutation of TP53 induces proliferation [2]. Consistently, proliferation was accelerated when p53a was knocked down in MCF-7 cells. Similarly, Δ40p53 knockdown enhanced proliferation, though to a lesser extent, while overexpression of Δ40p53 slightly reduced proliferation in MCF-7 sublines. These findings propose similar roles for  D40p53 and p53a in proliferation suppression in MCF-7 cells (Fig 3I,J). This proliferation suppression by D40p53 has been previously demonstrated through transfection of p53-null cells with Δ40p53 vectors [14]. In ZR75-1 cells, no differences in proliferation were detected, indicating that these findings are likely cell context-specific.
p53a regulates cell migration and invasion, mostly indirectly via other cofactors, but the function of D40p53 in cell mobility has never been examined [23]. Epithelia-like breast cancer migration and invasion is generally through a migratory front, which passively drags the following cells [47]. Therefore, scratch wound assays were employed. MCF-7 cells overexpressing D40p53 were less migratory and invasive than MCF-7-LeGO cells (Fig. 4A-F), implying a role of D40p53 in inhibiting cell mobility. Due to changes in the ability to form confluent monolayers and morphological changes in shRNA-transduced ZR75-1 cells (Fig. 3D,F,H), scratch wound assays were considered inaccurate. Transwell assays were therefore performed on shRNA-transduced MCF-7 and ZR75-1 sublines for consistency. Increased migration and invasion were observed in both cell lines transduced with p53a-shRNA, indicating that p53a is a critical safeguard preventing cell mobility. D40p53-shRNA, on the other hand, mildly impaired ZR75-1 but not MCF-7 cell mobility ( Fig. 4G-J). The reason could be that there are more extracellular matrix-associated genes being affected by Δ40p53 in ZR75-1 cells than in MCF-7 cells (discussed below).
RNA-seq analysis at the basal level showed that genes associated with increased proliferation potential and decreased tumour suppression were differentially expressed following p53a-knockdown ( Fig. 5A-C), supporting the functional assays. Taken together, these results suggest that loss of p53a in MCF-7 cells enhanced tumorigenicity as expected. Δ40p53 was found to differentially regulate genes linked to migration and invasion as well as upregulate tumour suppressor genes and downregulate oncogenes (Fig. 5D-H). Together with the functional data, this suggests that at the basal level, the N terminally truncated isoform Δ40p53 may inhibit migration and invasion, while also controlling proliferation. D40p53 has been previously reported to retain tumour suppressor function under stress due to the presence of the second transactivation domain [14]. As such, D40p53-overexpression at the basal level assimilated the function of p53a. Whether this is a result of the stabilising effect Δ40p53 exhibited on p53a is still unclear at this point. Few of the genes differentially expressed in D40p53 knockdown and overexpression sublines are known p53 target genes, which is supported by the lack of overlap with p53 knockdown sublines. However, DICER1, which was found to be downregulated when Δ40p53 was knocked down (log 2 (fold change) À1.768 and À4.538 in MCF-7 and ZR75-1, respectively) as well as when Δ40p53 was overexpressed in MCF-7 cells (log 2 (fold change) À1.299) contains a p53RE in its promoter [48]. DICER1 is critical for microRNA (miRNA) maturation, which could post-transcriptionally regulate mRNA expression [49]. As reviewed by Boominathan et al. [48], the p53 family mediates a complicated tumour suppressor miRNA network through DICER1, controlling tumour suppressor genes such as PTEN as well as metastasis-associated genes including ZEB1. The fact that DICER1 is downregulated regardless of Δ40p53 knockdown or overexpression implies a ratio of Δ40p53 to p53a could be the key to altering DICER1 and the associated miRNA network. Additionally, D40p53 expression has been linked with enhanced stemness in mouse embryos [18], while as reviewed by Molchadsky et al [50], p53 is known to promote differentiation and development. Hence, altering the levels of the isoforms may play a role in regulating cell differentiation. The RNA-seq results showed that the FGFR3 (Fibroblast growth factor receptor 3, promoting differentiation) gene was downregulated when D40p53 was knocked down in both MCF-7 and ZR75-1 (Fig. 5F), highlighting the relationship between D40p53 and stemness/differentiation. Inconsistent results from MCF-7 and ZR75-1derived sublines may result from differences in the expression of other endogenous regulators. For example, HDM2 expression was reported to be higher in ZR75-1 cells than MCF-7 cells [51] and thus may suppress p53a function in this cell line. D40p53 is HDM2insensitive and, therefore, may have taken over the function of p53a to a greater extent in these cells and provides a possible explanation for greater differential expression observed in ZR75-1-shD40p53 cells compared to MCF-7-shD40p53 cells (Fig. 5F). Additionally, DEGs associated with cell adhesion and extracellular matrix organisation in ZR75-1-shD40p53 cells may explain the altered morphology observed in this subline (Fig. 3F). These DEGs include increased EMILIN2, RECK and ADMTSL1, as well as decreased LRG1, COL16A1, QSOX1, KIAA0319, GNA13, AMOTL1, ADAMTSL5, SGCD and PXDN ( Fig. 5E and Table S2). The discrepancy in induction of morphological changes may therefore be celldependent. This study examined two breast cancer cell lines, and this is not representative of all breast cancer cases and subtypes. In particular, our previous studies have demonstrated that D40p53 expression was found to be highest in triple negative breast cancers in which TP53 was frequently mutated [3,19]. However, the purpose of the current study was to define the function of D40p53 in a wtp53a setting. The role of D40p53 in the context of mutant p53a needs to be further investigated through other breast cancer cell lines. The custom shRNAs established as part of this study will be a very useful tool for this.

Conclusion
In summary, examining the role of D40p53 in two ER+ breast cancer cell lines revealed differential effects on cell motility in MCF-7 cells but not in ZR75-1 cells. In contrast, p53a, acted to restrain cell motility in both cell lines, suggesting it plays a more dominant regulatory role in this context. Together with the downregulation of putative oncogenes and upregulation of tumour suppressor genes, Δ40p53 stunted proliferation, migration and invasion in MCF-7 cells, highlighting its cell context-specific function as a tumour suppressor.