TP53 drives invasion through expression of its Δ133p53β variant

1) Removal of the p63 chapter. The reviewers agreed that the paper would be better served by removing the last chapter on mechanism - basically Figures 6 and 7 - and strengthening the main observations throughout Figures 1–5 by addressing the various technical aspects listed below. Though the association with p63 is of interest, the experiments presented lack major controls and depth to draw solid conclusions. Reviewers concluded that developing the p63 aspect of the paper would require an amount of work not feasible within the review process and that it could become the subject of a follow up study.

We agree that the description of the mechanism involving p63 isoforms could be more appropriate in a follow-up of the present study. As suggested by reviewers, we have removed the results related to p63 (Figures 6 and 7) in the Results and Discussion sections of the revised manuscript.

2) Improved expression analysis in tumor samples:

2.1) The data on the expression of the different p53 isoforms in tumors only rely on a nested RT-PCR analysis. This could lead to misleading interpretation especially when used to compare expression of different isoforms. For example, are the tumors 5 and 8 in Figure 1B negative or positive for the γ isoform? It is important to verify these data with an orthogonal approach.

We agree with the reviewers that it is essential in such study to minimise the number of false positive and false negative data by improving detection threshold and by identifying precisely the different Δ133p53 isoforms. In addition, it is important to take account of the tumor heterogeneity as some p53 isoforms may be only expressed in a small population of tumor cells. It is well-established that only a small percentage of tumor cells can leave the primary tumor site and spread to other organs to form metastasis. Therefore, the methodology to analyse tumor samples should be highly sensitive and specific.

Two methods could be used: PCR that detects specifically each Δ133p53 mRNA variants and/or immunohistochemistry (IHC) of tumor sections using antibodies specific of the different Δ133p53 protein isoforms.

Currently, there is no antibody specific of Δ133p53α or Δ133p53β or Δ133p53γ. Therefore, the detection and identity of each p53 isoforms in tumor samples can only be ascertained by PCR using different sets of primers that specifically amplify Δ133p53α, Δ133p53β or Δ133p53γ mRNA variants respectively. The reviewers’ comment made us realise that we did not explain well enough that the RT-PCR method used in this manuscript was an improved version of the RT-PCR method previously described (Khoury et al., 2013. Detecting and quantifying p53 isoforms at mRNA level in cell lines and tissues. Methods in molecular biology, 962, 1-14). The RT-PCR method used in this manuscript has a higher sensitivity and specificity that enable to detect and accurately identify the expression of Δ133p53α, Δ133p53β and Δ133p53γ mRNA in each tumor sample. It allows thus to compare their respective association with the clinico- pathological markers and/or patient clinical outcome.

We corrected this in the revised manuscript. We wrote a detailed protocol and described the points that increase the specificity and sensitivity of the nested RT-PCR analysis in the “extended experimental procedures”. The primers sequences are now provided in Table 1.

Reviewers suggested to use an orthogonal approach to verify RT-PCR data, notably regarding the expression of Δ133p53γ (For example, are the tumors 5 and 8 in Figure 1B negative or positive for the γ isoform?) We cannot verify expression of the γ p53 protein isoforms by immunoblot or IHC as there is no antibodies specific of the γ isoforms. However, we verified by immunoblot the expression of Δ133p53β protein in some tumor samples (please see below).

2.2) Since an antibody specific for the β isoform is available, it will be important also to verify expression at protein level.

The rabbit polyclonal antibody KJC8 was raised and affinity-purified against the peptide TLQDQTSFQKENC corresponding to the C-terminal region of the different β p53 isoforms. It detects all β p53 protein isoforms. Therefore, it is not currently possible to specifically detect Δ133p53β by IHC in tumors.

To address the reviewers’ comments, we determined the expression of p53 isoforms by western- blot. However, this technique is rudimentary as it destroys large amount of a tumor sample preventing its use for on-going and future clinical studies. We were thus authorised to extract protein only from few large breast tumors. The western-blot is included in the new Figure 1—figure supplement 2 and shows that Δ133p53β protein is detected only in the samples expressing Δ133p53β at the mRNA level. Actin was used as a loading control and as an indicator of protein degradation that could have occurred during extraction. Only one band was observed for Actin indicating that no protein degradation occurred during extraction. This immunoblot validates the nested RT-PCR analysis used in this study. Furthermore, it shows for the first time that Δ133p53β protein is expressed in human tumor tissues. This is an important observation as it demonstrates its biological relevance and justifies the characterisation of its biological activities. Thank you for the suggestion, it strengthens the manuscript.

However, in future studies, we would not recommend to verify the nested RT-PCR analysis by immunoblot because it uses large shunk of precious tumor samples. In addition, sensitivity of immunoblot is much lower than PCR analysis; this could lead to misleading interpretation notably for small tumors.

2.3) It is not clear if the analysis in Figure 1—figure supplement 1E-I was limited to Luminal A tumors. Conducting the analysis without considering breast cancer subtypes could in fact be problematic given the difference p53 isoforms and mutations have different distribution across breast cancer sub-types and given lineage specificities in the behavior of tumors.

Figure 1—figure supplement 1E-I shows by using Kaplan-Meier univariate analyses that several markers (Breast cancer subtypes, TP53 mutation status, Δ133p53β expression, tumor size, tumor grade, Number of invaded lymph node) are associated with patient clinical outcomes. The analyses were performed on the entire cohort without segregating tumors according to breast cancer subtypes, TP53 mutation status and p53 isoform expression. It was not limited to the luminal A tumors. We clarified this in the figure legend.

The purpose of Figure 1—figure supplement 1E-I is to justify the use of the multivariate Cox- regression analysis to identify, among all the markers (Δ133p53β, the breast cancer subtypes (triple negative, Luminal A, Luminal/HER2+, Luminal B, and HER2+), TP53 mutation status, tumor grade, lymph node metastasis and tumor size (>20mm)), the independent markers significantly associated with the patient’s clinical outcome. The multivariate Cox’s regression analysis led to the identification of Luminal-A as the most significant independent predictor of disease-free survival and overall survival (Table 4).

The Cox’s regression analysis was then reiterated and limited to the luminal-A breast cancer to identify, among all the markers (Δ133p53β, TP53 mutation, tumor grade, lymph node metastasis and tumor size (>20mm)), the independent markers significantly associated with the clinical outcome of luminal-A patients. Only Δ133p53β was then identified as the most significant independent predictor of cancer progression and death in the luminal A Breast cancer patients (Table 5). Δ133p53α and Δ133p53γ were also tested in the multivariate Cox’s regression analysis but it did not lead to significant association with clinical outcome. Therefore, the statistical analysis was performed in considering all markers when they were included in the Cox’s regression multivariate analysis. This point has been clarified in the revised manuscript.

(Of note, Figure 1—figure supplement 1I shows that Δ133p53β expression is associated with poor disease free survival in WT TP53 breast cancer patients using Kaplan-Meier univariate analysis. The Kaplan-Meier univariate analysis was also performed for Δ133p53α and Δ133p53γin the WT TP53 breast cancer patient cohort but no significant association was found.)

3) Another major concern is related to the specificity of the siRNA and their association with the observed phenotypes. A rescue experiment with the appropriate controls will be required to definitely prove that the described differences are due to δ-133 p53 β isoform. Also to have a better sense of the specific functional role of the different p53 isoforms, it will be important that all experiments be performed with all the described siRNAs and upon over expression of the different p53 isoforms.

To confirm the specificity of the siRNA used, we performed a rescue experiment to assess whether the difference in invasive phenotype was due to the loss of expression of ∆133p53ß. We analyzed breast cancer MDA-MB-231 D3H2LN cells depleted of all Δ133p53 isoforms (by using the siRNAs si133-1 or si133-2, which specifically target the 5’UTR of Δ133 mRNAs (Aoubala et al., 2011)). The MDA-MB-231 D3H2LN cells express mutant TP53 gene mutated at codon 280 (R280K). Expression of the si133-1- and si133-2- resistant mutant Δ133p53β-R280K cDNA rescued invasion (Figure 3C and Figure 3—figure supplement 1B). In addition, we show that transfection of mutant Δ133p53β-R280K rescued invasion when MDA-MB-231 D3H2LN cells were previously depleted of all β isoforms by siβ (Figure 3C). We also verified that expression of si133- resistant WT Δ133p53β cDNA rescued invasion in the LoVo colon cancer cells previously depleted of all Δ133p53 isoforms. The LoVo cells express WT TP53 gene. These two experiments confirm the specificity of the siRNAs si133-1, si133-2 and siβ. They confirm that the differences in invasion are mainly due to the ∆133p53ß isoform (whether it is WT or mutated).

Another question raised by the reviewers is the specific functional role of the different p53 isoforms. This question is important since the siRNAs si133-1 or si133-2 used in this study deplete all Δ133p53 isoforms, i.e. Δ133p53α, Δ133p53β and Δ133p53γ. In a rescue experiment, we compared the specific contribution of each Δ133p53 isoforms in rescuing the invasive phenotype promoted by the siRNAs si133-1 or si133-2. As shown in the new Figures 3C and Figure 3—figure supplement 1B, all three Δ133p53 isoforms rescued invasion in cells depleted of all ∆133p53 variants.

However, Δ133p53β conferred a significantly higher invasive potential compared to Δ133p53α or Δ133p53γ. The results indicate that all Δ133p53 isoforms promote invasion. However Δ133p53β is significantly more potent than Δ133p53α or Δ133p53γ.

Altogether, the experimental data corroborate the clinical data and support the conclusion that the TP53 gene promotes cancer cell invasion through expression of its Δ133p53β variant.