Introduction

TP53, the human gene for tumor suppressor p53, encodes several isoforms owing to distinct promoters, alternative splicing and multiple translation initiation sites (Bourdon et al., 2005; Courtois et al., 2002; Flaman et al., 1996; Yin et al., 2002). p53 alternative isoforms can be abnormally expressed in cancer cells and some may regulate the canonical p53 protein (Anbarasan and Bourdon, 2019; Bourdon et al., 2005; Mondal et al., 2013; Senturk et al., 2014). However, aberrant RNA splicing is a common feature of cancer cells (Graubert et al., 2012; Martin et al., 2013; Pajares et al., 2007; Sette and Paronetto, 2022) and to which extent alternative splicing generates functionally relevant proteins is controversial (Abascal et al., 2015; Blencowe, 2017; Tress et al., 2017a, 2017b; Ule and Blencowe, 2019; Weatheritt et al., 2016). Thus, the biological importance of many p53 isoforms remains elusive.

Like its human TP53 homolog, the murine Trp53 gene encodes multiple isoforms differing in their N– or C-termini (Arai et al., 1986; Marcel et al., 2011). Mouse models to evaluate the role of p53 isoforms differing in their N-terminus revealed that Δ40-p53 overexpression leads to accelerated ageing (Maier et al., 2004; Steffens Reinhardt et al., 2020). However, the potential role of p53 isoforms with an alternative C-terminus was not analyzed in vivo. p53 isoforms with distinct C-termini result from the splicing of two mutually exclusive final exons: exon 11, encoding the canonical “α” C-terminal domain, and the Alternatively Spliced (AS) exon, encoding another C-terminus (Arai et al., 1986). In adult mice, isoforms with the canonical C-terminus are predominant in all tissues (Figure S1A). Two models (p53^Δ31 and p53^ΔCTD), designed to study the consequences of a loss of the canonical p53 C-terminus, exhibited signs of increased p53 activity leading to a rapidly lethal anemia (Hamard et al., 2013; Simeonova et al., 2013). To determine the role of p53-AS isoforms in vivo, we created p53^ΔAS, a mouse model with a specific deletion of the AS exon (Figure S1B). In mouse embryonic fibroblasts (MEFs), the Trp53^ΔAS allele prevented the expression of isoforms with the AS C-terminus whereas it did not affect RNA levels for p53 isoforms with the canonical C-terminus (Figure S1C). We previously used this model to show that p53-AS isoforms had no role in the anemia affecting p53^Δ31/Δ31 mice (Simeonova et al., 2013). However, a detailed phenotyping of p53^ΔAS/ΔAS mice remained to be performed. The detailed phenotyping, presented here, yielded surprising information on lymphomagenesis.

Results

Stress responses in WT and p53^ΔAS/ΔAS cells

Thymocytes undergo a p53-dependent apoptosis upon irradiation (Lowe et al., 1993). We analyzed thymocytes from irradiated wild-type (WT) and p53^ΔAS/ΔAS mice. In WT thymocytes, isoforms with the AS C-terminus were 5 times less abundant than isoforms with the α C-terminus at the RNA level (Figure 1A), and in western blots the p53-AS protein appeared as a faint band running just ahead of, and often hard to separate from, the band specific for p53-α, the canonical full-length p53 (Figure 1B). In p53^ΔAS/ΔAS thymocytes, mRNA levels for α isoforms were slightly decreased, if at all (Figure 1A), whereas p53-α protein levels appeared markedly decreased (Figure 1B), raising the possibility that p53-AS isoforms might contribute to p53-α abundance. Nevertheless, the transactivation of classical p53 target genes (Figure 1C) and apoptotic response (Figures 1D and S1D) were not significantly altered by the loss of AS isoforms. Likewise, no significant difference was observed between WT and p53^ΔAS/ΔAS fibroblasts in assays for cell cycle control (Figures 1E and S1E), expression of well-known p53 target genes (Figure 1F and S1F-G), proliferation under hyperoxic conditions (Figure1G), or the growth of tumor xenografts (Figure 1H).

Figure 1.

Lymphomagenesis in WT and p53^ΔAS/ΔAS mice

We compared spontaneous tumor onset in WT and p53^ΔAS/ΔAS littermates for over 2 years and observed no significant difference in tumor-free survival (Figure 1I). Because lymphoma is a common neoplasm in C57BL6/J WT mice (Brayton et al., 2012) and our mouse cohorts resulted from >10 generations of backcrosses with C57BL6/J mice, we searched for evidence of lymphoma in the lymph nodes and spleen, by macroscopic examination at autopsy and histological analyses. B-cell lymphomas were observed in about 30% of mice of either genotype (Figure 1J). In p53^+/+ mice, a higher incidence of B-cell lymphomas was observed in females, in agreement with previous observations (Brayton et al., 2012). By contrast, no obvious sex-specific bias was observed for B-cell lymphomas in p53^ΔAS/ΔAS mice (Figure 1J), raising the possibility that the loss of p53-AS isoforms affected B-cell lymphomagenesis. However, the numbers of lymphoma-bearing mice were too small to be conclusive.

We next used Eμ-Myc transgenic mice, prone to highly penetrant B-cell lymphomas (Adams et al., 1985). p53^+/+ Eμ-Myc and p53^ΔAS/ΔAS Eμ-Myc mice developed B-cell lymphomas (Figure S2A) with similar survival curves when sexes were not considered (Figure S2B). Importantly however, death was accelerated, and tumor lymph nodes were larger, in p53^ΔAS/ΔAS Eμ-Myc males compared to their p53^+/+ Eμ-Myc male counterparts, whereas no difference in lymphomagenesis was noticeable between p53^ΔAS/ΔAS Eμ-Myc and p53^+/+ Eμ-Myc female mice (Figure 2A-B). Our data (Figures 2A-B and S2C), together with the fact that B-cell lymphomas occur with a higher incidence in WT C57Bl6/J female mice (Brayton et al., 2012), suggested that p53^+/+ Eμ-Myc male mice are more refractory to B-cell lymphomas, and that p53-AS isoforms might confer this male-specific protection against lymphomagenesis.

Figure 2.

Cause for accelerated lymphomagenesis in p53^ΔAS/ΔAS Eμ-Myc males

We next aimed to determine the mechanisms underlying the accelerated lymphomagenesis in p53^ΔAS/ΔAS Eμ-Myc males. Inactivating p53 mutations were not more frequent in tumors from p53^ΔAS/ΔAS Eμ-Myc males than in those from p53^+/+ Eμ-Myc males, ruling out additional mutations at the Trp53 locus as potential causes for accelerated lymphomagenesis in p53^ΔAS/ΔAS Eμ-Myc males (Figures S2D-E). We next analyzed a subset of tumors with no detectable Trp53 mutation in males of both genotypes, and found that Myc was expressed at similar RNA and protein levels in all tumors (Figure 2C). No difference in p53-α mRNA levels was observed in tumors from both genotypes, although a decrease at the protein level was detected in most tumors from p53^ΔAS/ΔAS Eμ-Myc males (Figure 2D). Nevertheless, similar transcript levels for classical p53 target genes were observed in tumor cells of both genotypes (Figure 2E). To test whether a higher tumor volume in p53^ΔAS/ΔAS Eμ-myc males might result from lower apoptosis and/or higher cell proliferation, we next analyzed tumors by immunohistochemistry with antibodies against cleaved caspase-3 or ki-67, respectively. Similar apoptotic and proliferation indexes were observed for both genotypes (Figure 2F-G). In sum, classical assays for p53 activity in tumors failed to account for differences between the two male genotypes.

The speed of lymphomagenesis in Eμ-Myc mice correlates with the extent of B-cell expansion in the first stages of B cell differentiation (Langdon et al., 1986) and p53 was proposed to control the pool of early B cells (Slatter et al., 2010). Therefore, we determined the levels of the early pre-B / immature B cells in 6 weeks-old mice, before any sign of tumor. We analyzed the spleen, a preferential site of B-cell expansion (Langdon et al., 1986) with a relatively high AS/α isoform ratio (Figure S1A). Flow cytometry with a combination of markers was used to discriminate the pre-B, immature, transitional and mature B subpopulations. As expected (Langdon et al., 1986), we observed high numbers of pre-B and immature B cells in Eμ-Myc mice. In males, pre-B and immature B cells were more abundant in p53^ΔAS/ΔAS Eμ-Myc animals, while no difference was observed for transitional and mature B cells (Figures 3A and S3A). By contrast, in the spleen of p53^+/+ and p53^ΔAS/ΔAS 6 weeks-old male mice without the Eμ-Myc transgene, most B cells were mature B cells (Figure S3B). In Eμ-Myc females, the numbers of pre-B and immature B cells were similar between genotypes, as were the numbers of transitional and mature B cells (Figure 3A). Interestingly, p53^+/+ Eμ-myc males, which develop lymphomas less rapidly, exhibited the lowest number of immature B cells (Figure 3A), suggesting a direct correlation between the level of immature B cell expansion and the speed of lymphomagenesis. Together, these data suggested that p53-AS isoforms may not be required to control the pool of early B cells under normal conditions, but that in an Eμ-Myc context they would limit the expansion of pre-tumor early B cells, specifically in males.

Figure 3.

Transcriptomes from p53^+/+ Eμ-Myc and p53^ΔAS/ΔAS Eμ-Myc male spleens

We next performed bulk RNA-seq and differential expression analyses comparing the spleens from 4-6 weeks-old p53^ΔAS/ΔAS Eμ-Myc males to spleens from age-matched p53^+/+ Eμ-Myc males. This revealed a limited number of significantly deregulated genes (Figure 3B), including 13 protein-coding genes and 11 pseudogenes (Figure 3C). Out of the 13 protein-coding genes, we focused on the 10 genes not encoding an immunoglobulin and analyzed the same samples by RT-qPCR (Figure 3D). For 6 of the 10 genes, expression levels were too low to be quantified (Tcstv3), or differences in expression were not statistically significant (Masp2, Akr1c19, Cd300lh, Il5ra, Slc26a1). Of note, Il5ra belonged to this group, although it is regulated by p53 (Zhu et al., 2022), which illustrates the difficulty to analyze subtle effects in our experiments. Taking this into account, we considered as potentially interesting the 4 remaining genes, exhibiting differences in mRNA levels with statistical significance (p<0.05) or borderline statistical significance (p=0.057): Ackr4, less expressed in p53^ΔAS/ΔAS Eμ-Myc males, and Fam132a, Mt2 and Prss50, with an increased expression in p53^ΔAS/ΔAS Eμ-Myc males (Figure 3D). Importantly, survival curves indicated that the Myc-induced lethality was delayed in p53^+/+ Eμ-Myc males compared to p53^ΔAS/ΔAS Eμ-Myc males, p53^+/+ Eμ-Myc females and p53^ΔAS/ΔAS Eμ-Myc females (Figure S2C). Thus, we quantified transcripts for Ackr4, Fam132a, Mt2 and Prss50 in the spleen of 4-6 weeks-old p53^+/+ Eμ-Myc and p53^ΔAS/ΔAS Eμ-Myc females, then compared mRNA levels in p53^+/+ Eμ-Myc males versus the 3 other groups. Significantly higher expression of Ackr4 and lower expression of Mt2 were found in p53^+/+ Eμ-Myc males (Figure 3E).

Ackr4 encodes the atypical chemokine receptor 4, a decoy receptor promoting the degradation of chemokines that modulate cancer cell proliferation and metastasis (Chow and Luster, 2014; Marcuzzi et al., 2018; Müller et al., 2001). Our data suggested that Ackr4 might be a gene transactivated by p53-α and/or p53-AS isoforms. Consistent with this, by extracting data from a transcriptome-wide study in MEFs (Younger et al., 2015) we found evidence for a p53-dependent transactivation of Ackr4 in response to doxorubicin (Figure 3F). Furthermore, ChIP-Atlas, the database of chromatin immunoprecipitation experiments (Oki et al., 2018), indicated p53 binding to sequences within the intron 1 of Ackr4 in doxorubicin-treated MEFs, and we identified a candidate p53 responsive element in this intron (Figure 3G). We next used luciferase assays to show that this p53 responsive element can be bound and regulated by both p53-α and p53-AS (Figures 3G-H and S3C). Together, these data show that Ackr4 is indeed a p53 target gene, although RNAseq data indicated that it is expressed at much lower levels than classical p53 targets like Cdkn1a or Mdm2 in the splenocytes of Eμ-Myc mice (Table S1). Furthermore, Ackr4 was shown to be regulated by Foxl2 and estrogen signalling in ovarian cells (Georges et al., 2014) and 17-β estradiol was recently found to regulate ACKR4 expression in meniscal cells from both sexes, albeit differentially (Knewtson et al., 2020). Accordingly, we observed, in both WT and p53^ΔAS/ΔAS MEFs, that p53 activation with the Mdm2 antagonist Nutlin led to the transactivation of Ackr4, but that a concomitant treatment with 17-β estradiol markedly decreased, or completely abrogated, Ackr4 transactivation (Figure 3I). By contrast, Cdkn1a was efficiently transactivated under both conditions in mutant and WT cells (Figure 3I). These data indicate that Ackr4 is a p53 target gene whose p53-mediated transactivation can be inhibited by estrogens.

The Mt2 gene, encoding the potentially oncogenic metallothionein-2 (Si and Lang, 2018), was less expressed in p53^+/+ Eμ-Myc male pre-tumoral splenocytes, which raised the possibility of its direct or indirect repression by p53, potentially through the binding of p53 or the DREAM complex at its promoter (Engeland, 2018; Peuget and Selivanova, 2021). However, ChIP-Atlas reported no binding of these proteins at the Mt2 promoter. Alternatively, evidence that Myc may impact on Mt2 expression was obtained previously (Qin et al., 2021), and ChIP-Atlas reported Myc binding at the Mt2 promoter in primary B-cells from lymph nodes of Eμ-Myc mice as well as Eμ-Myc-induced lymphoma cells (Figure S3D). This may suggest that the lower expression of Mt2 in pre-tumoral splenic cells from p53^+/+ Eμ-Myc males (Figure 3E) might result from a subtle difference in Myc signalling.

Relevance of ACKR4 expression in Burkitt lymphomas

Murine and human alternatively spliced p53 isoforms exhibit structural differences (Marcel et al., 2011) and the ChIP-Atlas database (Oki et al., 2018) does not report p53 binding to the human ACKR4 intron 1. Nevertheless, we found that p53 activation in human cells also led to an increased ACKR4 expression abrogated by 17-β estradiol, whereas 17-β estradiol had no significant effect on the p53-mediated transactivation of CDKN1A (Figure 4A). This led us to investigate the potential relevance of ACKR4 expression in human B-cell lymphomas. We analyzed public databases of B-cell lymphoma patients with clinical and gene expression information. We first analyzed #GSE4475 (Hummel et al., 2006), a dataset of mature aggressive B-cell lymphomas previously used to define Burkitt lymphoma-like specific transcriptomes, comprising 159 patients (91 men, 68 women) with clinical follow-up. Overall, ACKR4 gene expression was not significantly different between male and female patients (Figure 4B, left). However, average mRNA levels appeared higher in males when we considered the 30% patients of each sex with the highest ACKR4 expression (Figure 4B, right). Strikingly, when we compared the survival of 30% patients with the highest ACKR4 mRNA levels to the survival of the 30% patients with the lowest ACKR4 mRNA levels, high ACKR4 expression correlated with a better prognosis in men, but not in women (Figure 4C). By contrast, for MT2A, the human homolog of Mt2, differences in mRNA levels did not correlate with significant differences in survival for either sex (Figure S4A).

Figure 4.

We also analyzed dataset #GSE181063 (Lacy et al., 2020), comprising mostly diffuse large B-cell lymphomas (DLBCL; 613 men, 536 women) and a few Burkitt lymphomas (65 men, 18 women). We found no difference in survival curves of DLBCL patients with low versus high ACKR4 levels, neither in men nor in women. However, there was again an increased survival for Burkitt lymphoma male patients with high ACKR4 expression, but not for women (Figure S4B). Finally, we analyzed #phs000235 (Morin et al., 2011), a Burkitt lymphoma-specific dataset (65 men, 37 women) comprising mostly patients diagnosed at 0-17 years of age, hence providing cohorts homogeneous for both tumor type and age of onset. Again, ACKR4 was expressed at higher levels in a subset of male patients (Figure 4D) and high ACKR4 expression correlated with a better prognosis only in males (Figure 4E). We thus conclude that, as in Eμ-Myc mice, ACKR4 is a male-specific positive prognostic factor in Burkitt lymphoma, the archetype of MYC-driven B-cell lymphomas.

Discussion

In this report we analyzed a mouse model with a specific deletion of the alternatively spliced (AS) exon of the Trp53 gene. Despite a subtle phenotype, this model revealed that a male-specific protective effect against Eμ-Myc-induced B-cell lymphomas is lost in the absence of p53-AS isoforms. p53^ΔAS/ΔAS males also appeared more prone to develop spontaneous lymphomas, suggesting that the sex-specific protective effect conferred by p53-AS isoforms might not be restricted to the Eμ-Myc model.

Our transcriptomic data from splenocytes of p53^+/+ Eμ-Myc and p53^ΔAS/ΔAS Eμ-Myc males disclosed very few differentially expressed genes, and highlighted Ackr4 as a male-specific positive prognostic factor in Eμ-Myc-induced lymphomas. Mechanistically, we identified Ackr4, expressed at low levels in splenocytes, as a p53 target gene that may be transactivated by p53-α and/or p53-AS according to luciferase assays. Because the loss of p53-AS isoforms correlated with an apparent decrease in p53-α levels in both thymocytes and tumor lymph nodes, the decreased transactivation of Ackr4 in splenocytes of p53^ΔAS/ΔAS Eμ-Myc males might result from the loss of p53-AS isoforms per se, but also possibly from a decrease in p53-α levels. Furthermore, we observed that 17-β estradiol can inhibit the p53-mediated transactivation of Ackr4. Together, our data suggest that Ackr4 may be regulated by p53-α, p53-AS and estrogens, likely accounting for sex-specific and p53-status dependent differences in gene expression.

Our analyses reveal that in both mice and humans, Ackr4/ACKR4 is a male-specific prognostic factor in Burkitt-like lymphomas. In fact, Ackr4 most likely acts as a tumor suppressor of Myc-driven B-cell lymphomas because: 1) in lymph nodes, Ackr4 creates a gradient of the chemokine Ccl21, a ligand of the chemokine receptor Ccr7 (Bastow et al., 2021; Ulvmar et al., 2014); 2) the Ackr4-mediated sequestration of Ccl21 impairs the Ccr7 signalling cascade, which may lead to decreased Myc activity (Shi et al., 2015); 3) Ccr7 is also required for lymphoma cell lodging to secondary lymphoid organs, and Ccr7 deficiency was shown to delay Eμ-Myc-induced lymphomagenesis (Rehm et al., 2011). Thus, in p53^+/+ Eμ-Myc male pre-tumoral splenocytes, an increased Ackr4 expression might delay lymphomagenesis in part by attenuating Myc signalling. Indeed, the observed lower expression of Mt2, known to be regulated by Myc (Qin et al., 2021), appears consistent with this hypothesis. Importantly, Ackr4 was previously found to inhibit the growth and metastasis of breast, cervical, colorectal, hepatocellular and nasopharyngeal cancer cells (Feng et al., 2009; Hou et al., 2013; Ju et al., 2019; Shi et al., 2015; Zhu et al., 2014), although no report mentioned any sex-specific bias for cancers occurring in both sexes. Our data provide evidence that sex-specific differences in Ackr4 expression may have prognostic value. This suggests that measuring ACKR4 gene expression in male patients with Burkitt lymphoma could be useful to identify the patients at higher risk, for whom specific therapeutic regimens might be required.

Interestingly, our data suggested that Mt2 might be a male-specific negative prognostic factor in murine Eμ-Myc induced lymphomas, but MT2A expression levels had no prognostic value in human lymphomas. A possible explanation for this discrepancy is suggested by the fact that Mt2 is regulated by Myc. A translocation leading to MYC overexpression drives oncogenesis in all Burkitt lymphomas, but half of them exhibit additional missense MYC mutations enhancing its tumorigenicity (Chakraborty et al., 2015). The transcriptional program of a WT and two lymphoma-associated Myc mutants were recently compared, and we noticed that one of the mutants led to an alteration in Mt2 expression (Mahani et al., 2021), which would abrogate any potential prognostic value.

Finally, a polymorphism in the MDM2 gene promoter provided evidence that sex-specific hormones may affect p53 signalling and tumorigenesis (Bond and Levine, 2007). More recently, a higher frequency of TP53 mutations in men, together with an increased vulnerability to alterations of X-linked genes encoding p53 regulators, were proposed to explain a higher cancer incidence and death in male patients (Haupt et al., 2019). Here, on the contrary, male mice and a subset of male patients were more efficiently protected against Burkitt-like lymphomas, which adds another layer of complexity to sex-specific differences in tumorigenesis. The p53 pathway thus underlies cancer sex-disparities through multiple mechanisms, that may notably include variations in p53 isoforms or Ackr4 expression.

Materials and Methods

Mice

Design and construction of the p53^ΔAS mouse model was previously described (Simeonova et al., 2013). A minimum of 10 backcrosses with C57Bl6/J mice of both sexes (Charles River Laboratories) were performed before establishing the cohorts of p53^+/+ and p53^ΔAS/ΔAS littermate mice used in this study. Mouse genotyping with multiple primer sets confirmed >99% C57Bl6/J genetic background after 10 backcrosses (primer sequences available upon request). Cohorts of p53^+/+ Eμ-Myc and p53^ΔAS/ΔAS Eμ-Myc mice were established with identical parental origin of the Eμ-Myc transgene. For all experiments, mice housing and treatment were conducted according to Institutional Animal Care and Use Committee of the Institut Curie.

Cells and cell culture reagents

Mouse embryonic fibroblasts (MEFs) were isolated from 13.5 days post-coitum embryos and cultured in a 5% CO₂ and 3% O₂ incubator, in DMEM Glutamax (GIBCO), with 15% FBS (PAN Biotech), 100 μM 2-mercaptoethanol (Millipore), 0.1 mM Non-Essential Amino-Acids and Penicillin/Streptomycin (GIBCO) for less than 5 passages, except for 3T3 experiments, performed in a 5% CO₂ incubator for 9 passages. Cells were treated for 24h with 0.5 μg/ml Doxorubicin (Sigma-Aldrich), 15 μM Etoposide (Sigma-Aldrich), 10 μM Nutlin 3a (Vassilev et al., 2004) (Sigma-Aldrich) and/or 5 μg/ml 17β-Estradiol (Merck). At least 3 independent experiments with at least 2 independent littermate MEF clones of each genotype and each sex were performed to measure DNA damage responses. For estradiol assays, four independent experiments with three independent MEF male clones of each genotype were performed. Human lung fibroblast MRC5 and its SV40-transformed derivatives were cultured in a 5% CO2 and 3% O2-regulated incubator in MEM medium without Phenol Red (Gibco), completed with 10% FBS, 2 mM L-glutamine (Gibco), 1 mM pyruvate, 0.1 mM Non-Essential Amino-Acids, and penicillin/streptomycin, and treated for 24h with 10 μM Nutlin 3a (Sigma-Aldrich) and/or 5 μg/ml 17β-Estradiol (Merck). Four independent experiments were performed.

Quantitative RT-PCR

Total RNAs were extracted using nucleospin RNA II (Macherey-Nagel), reverse-transcribed using superscript IV (Invitrogen), and real-time quantitative PCRs were performed on an ABI PRISM 7500 using Power SYBR Green (Applied Biosystems) as previously described (Simeonova et al., 2012). For quantification of p53 isoforms in healthy tissues, a forward primer in exon 10 and a reverse primer encompassing the boundary between exons 10 and 11 were used for p53-α amplification, whereas the same forward primer and a reverse primer located in exon AS were used for p53-AS amplification (see Table S2 for primer sequences). To determine the AS/α mRNA ratios, expression levels were compared with a standard curve generated by serial dilutions of a plasmid containing both p53-AS and p53-α cDNAs.

Western blots

Thymocytes were lysed in RIPA buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 5 mM EDTA, 0.5% deoxycholic acid, 0.1% SDS, 1% NP-40) with a cocktail of protease inhibitors (Roche) and 1 mM PMSF (Sigma). Whole-cell extracts were sonicated three times for 10 s and centrifuged at 13 000 r.p.m. for 30 min to remove cell debris. MEFs or B-cell lymphomas were lysed in Giordano’s buffer (50 mM Tris-HCl pH 7.4, 250 mM NaCl, 5 mM EDTA, 0.1% Triton X-100) with a cocktail of protease inhibitors (Roche) and 1 mM PMSF (Sigma). Protein lysate concentration was determined by BCA assay (Thermo Scientific) and 30 μg of each lysate was fractionated by SDS–PAGE on a 4-12% polyacrylamide gel and transferred onto PDVF membrane (Amersham). Membranes were incubated with antibodies against p53 (CM5, Novocastra), myc (9E-10, Santa Cruz), p21 (F-5, Santa Cruz) and actin (actin-HRP sc47778, Santa Cruz) and revealed with SuperSignal West femto detection reagent (Thermo Scientific).

Apoptosis Assays

Six weeks-old p53^+/+ and p53^ΔAS/ΔAS male mice were whole-body irradiated with 5 Gy of ψ-irradiation. Mice were sacrificed 4h later and thymocytes were recovered, stained with AnnexinV-FITC Apoptosis detection kit (Abcam), then analyzed by flow cytometry using FlowJo.

Cell-Cycle assays

Log phase MEFs were irradiated at room temperature with a CS ψ-irradiator at doses of 3 or 10 Gy, incubated for 24 h, then pulse-labeled for 1h with 10 μM BrdU, fixed in 70% ethanol, double-stained with FITC anti BrdU and propidium iodide, and sorted by flow cytometry using a BD FACSort. Data were analyzed using FlowJo.

Oncogene-induced tumor xenografts

MEFs with the indicated genotypes were sequentially infected with pWZL-E1A12S and pBABE-Hrasv12 viruses as previously described (Toledo et al., 2006). In total, 5×10⁶ E1A– and Ras– (E1A+Ras) expressing MEFs of each genotype were injected subcutaneously into the flanks of 7-weeks-old female athymic nude mice (at least 4 mice per genotype) and tumor volumes were determined 1, 8, 15, 21 and 25 days after injection. Importantly, populations of (E1A+Ras)-expressing cells were used to minimize potential differences in expression levels that could result from independent viral insertion sites.

Cell sorting of B-cell sub-populations

Splenocytes were recovered from 6 weeks-old asymptomatic mice and incubated with DAPI and the following antibodies: APC rat anti-mouse CD45R/B220, FITC rat anti-mouse CD43, PE rat anti-mouse IgM and BV605 rat anti-mouse IgD (BD Pharmingen). First, the B220+CD43-cells were selected by flow cytometry from DAPI negative living cells, yielding subsequently 4 different B subpopulations based on IgM and IgD labeling: IgM-/IgD-preB lymphocytes, IgM low/IgD-immature B lymphocytes, IgM high/IgD-transitional B lymphocytes and IgM+/IgD+ mature B lymphocytes.

RNA-Seq analysis

Total RNA was extracted from the spleen of 4-6 weeks-old asymptomatic mice using nucleospin RNA II (Macherey-Nagel). The quality of RNA was checked with Bioanalyzer Agilent 2100 and RNAs with a RIN (RNA integrity number) > 6 were retained for further analysis. RNA was depleted from ribosomal RNA, then converted into cDNA librairies using a TruSeq Stranded Total Library preparation kit (Illumina). Paired-end sequencing was performed on an Illumina MiSeq platform. Reads were mapped to the mouse genome version GRCm38 and counted on gene annotation gencode.vM18 with featureCounts (Liao et al., 2014). Differentially expressed genes of C57Bl6/J genetic background with an adjusted p-value < 0.05 were identified using the DESeq2 R package (Love et al., 2014).

Luciferase assays

The candidate p53 responsive element (p53 RE) in the Ackr4 promoter was identified by using the JASPAR database of binding profiles (Fornes et al., 2020) with the position weight matrix scanner PWMscan (Ambrosini et al., 2018). A 1.5 kb fragment from Ackr4 intron 1, containing a WT or mutant p53 RE at its center, was cloned upstream a SV40 minimal promoter and a luciferase reporter gene in the backbone of a PGL3 plasmid (Promega). We used lipofectamine 2000 to transfect p53^-/- MEFs with 2 μg of either luciferase expression vector, 2 μg of an expression vector for p53^WT, p53^AS or the DNA-binding mutant p53^R270H, and 30 ng of a renilla luciferase expression plasmid (pGL4.73, Promega) for normalization. Transfected cells were incubated for 24 h then trypsinized, resuspended in 75 μl culture medium with 7.5% FBS and transferred into a well of an optical 96-well plate (Nunc). The dual-glo luciferase assay system (Promega) was used according to the manufacturer’s protocol to lyse the cells and read firefly and renilla luciferase signals. Results were normalized, then the average luciferase activity in cells transfected with the WT p53RE luciferase reporter and the p53^R270H expression plasmid were assigned a value of 1.

Statistical Analyses

Student’s unpaired t tests were used in most Figures to analyze differences between WT vs ΔAS values. Log-Rank Mantel-Cox were used to analyze Kaplan-Meier tumor-free survival curves. Analyses were performed using Graphpad Prism 5, and values of p<0.05 were considered significant.

Data availability

RNA sequencing data have been deposited in the Gene Expression Omnibus (GEO) under the accession code GSE209708.

Acknowledgements

This project was supported by grants from the Comité Tumeurs of Fondation de France (to F.T.), the Comité Ile-de-France and Comité national (Labellisation 2014-18) of the Ligue Nationale Contre le Cancer (to F.T.), and the Fondation ARC pour la recherche sur le Cancer (to F.T.). I.S., E.E. and J.R. were PhD fellows of the Ministère de la Recherche; J.L. and A.M. were post-doctoral fellows of Cancéropôle Ile-de-France and Institut National du Cancer. M.G. was paid by European Research Council 875532-Prostator-ERC-2019-PoC attributed to A.M.; J.C.B. was supported by Cancer Research-UK. We thank K. Fernandes for his participation in making the ΔAS mutation, and members of the Institut Curie platforms: I. Grandjean, H. Gautier, C. Daviaud, M. Garcia, M. Verlhac, A. Fosse and P. Bureau (animal facility); S. Baulande and S. Lameiras (NGS); M. Huerre, A. Nicolas and R. Leclere (histopathology); Z. Maciorowski, A. Viguier and S. Grondin (flow cytometry).

Author Contributions

F.T, J.C.B. and I.S. initiated the project; F.T., B.B. and A.F. designed the research and analyzed data; B.B., A.F., I.S., J.L., A.M., V.L., F.T., A.H., W.V.--Z., E.E. and J.R. performed the research; M.G., M.P. and A.M. analyzed RNAseq data; F.T. acquired funding; F.T., A.F. and B.B. wrote the paper.

Declaration of interests

The authors declare no competing interest.

Supplemental Information

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Table S1.

Table S2.