Inhibition of glucosylceramide synthase eliminates the oncogenic function of p53 R273H mutant in the epithelial-mesenchymal transition and induced pluripotency of colon cancer cells

Missense mutation of tumor suppressor p53, which exhibits oncogenic gain-of-function (GOF), not only promotes tumor progression, but also diminishes therapeutic efficacies of cancer treatments. However, it remains unclear how a p53 missense mutant contributes to induced pluripotency of cancer stem cells (CSCs) in tumors exposed to chemotherapeutic agents. More importantly, it may be possible to abrogate the GOF by restoring wild-type p53 activity, thereby overcoming the deleterious effects resulting from heterotetramer formation, which often compromises the efficacies of current approaches being used to reactivate p53 function. Herewith, we report that p53 R273H missense mutant urges cancer cells to spawn CSCs. SW48/TP53 cells, which heterozygously carry the p53 R273H hot-spot mutant (R273H/+, introduced by a CRISPR/Casp9 system), were subchronically exposed to doxorubicin in cell culture and in tumor-bearing mice. We found that p53-R273H (TP53-Dox) cells were drug-resistant and exhibited epithelial-mesenchymal transition (EMT) and increased numbers of CSCs (CD44v6+/CD133+), which resulted in enhanced wound healing and tumor formation. Inhibition of glucosylceramide synthase with d-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP) sensitized p53-R273H cancer cells and tumor xenografts to doxorubicin treatments. Intriguingly, PDMP treatments restored wild-type p53 expression in heterozygous R273H mutant cells and in tumors, decreasing CSCs and sensitizing cells and tumors to treatments. This study demonstrated that p53-R273H promotes EMT and induced pluripotency of CSCs in cancer cells exposed to doxorubicin, mainly through Zeb1 and β-catenin transcription factors. Our results further indicate that restoration of p53 through inhibition of ceramide glycosylation might be an effective treatment approach for targeting cancers heterozygously harboring TP53 missense mutations.


INTRODUCTION
The p53 protein, encoded by human gene TP53, functions as a key tumor suppressor that stabilizes the genome with respect to propensity for tumorigenesis and cancer progression. The TP53 gene is somatically mutated in over half of all cancer cases. More than 80% of TP53 alterations are missense mutations, encoding full-length and dysfunctional proteins [1,2]. Alterations at codons 175, 248, and 273 constitute 19% of all TP53 mutations reported, and are considered to be mutation hotspots in human cancers, including those occurring in colon and lungs [1][2][3] (http://p53.free.fr/Database/p53_cancer/all_ cancer.html). Missense versions of p53 that lack the tumor suppression activity of wild-type p53 (wt p53) instead often exhibit oncogenic gain-of-function (GOF) [4].

Research Paper
Knock-in mouse models that express hotspot mutant alleles R172H or R270H (R175H or R273H in the human versions) manifest GOF by conferring a broader tumor spectrum and more tumor metastases, as compared with wt p53-expressing mice [5,6]. TP53 mutants are observed more frequently in tumors diagnosed at advanced stages, or with more metastases, and in recurrences of cancer in colon, ovaries and breasts [7][8][9].
Despite the well-known fact that expression of p53 mutants correlates strongly to poor prognosis in cancer patients, the exact roles in the promotion of cancer progression played by p53 mutants, which vary in type as well as position, remain as yet unclear. Recent reports document that inactivation of p53 function enhances the production efficiency, and decreases the latency for emergence of induced pluripotent stem cells (iPSCs) in cell culture [10,11]. iPSCs can be generated from somatic cells of mouse and of human by introduction of Oct4, Sox2, Klf4 and c-Myc transcription factors [12]. Suppression of p53 with small interfering RNA (siRNA) increased the efficiency of iPSC generation from human fibroblasts, indicating that the p53-p21 pathway serves as a barrier to iPSC generation [13]. With Oct4 and Sox2 reprogramming, p53-knockout cells merely maintained their pluripotent capacity in vivo, whereas mutant p53 (R172H) mouse cells gave rise to malignant tumors, with inherent oncogenic GOF attributable to the involvement of Klf4 [14].
Metastasis and drug resistance are major limitations to the survival and management of patients with cancers. Cancer stem cells (CSCs), which possess malignant capacities of self-renewal and pluripotency, not only lead tumorigenesis, but also drive tumor progression and are responsible for treatment failure [15,16]. Most cytotoxic agents in chemotherapy damage DNA or disrupt mitosis to induce cell death in highly proliferative populations of cancer cells. Chemotherapy often eliminates differentiated cells in tumors, but surviving CSCs (which do not rapidly proliferate) may cause disseminated metastases or recurrence of aggressive tumors upon treatment failure. Recent studies have convincingly shown enrichment with CSCs in populations of breast, ovarian and colon cancer cells that have become drug resistant following sequential exposure to anticancer drugs, including doxorubicin, paclitaxel and 5-fluorouracil (5FU) [17][18][19]. Enrichment with CSCs is also observed in xenogeneic tumors following chemotherapies [20,21]. Similarly, after systemic chemotherapy of cancer patients, CSCs are increased in breast and lung cancers [22,23].
Aiming to understand the roles played by p53 mutants in promoting tumor progression and in chemotherapy failure, we studied EMT and CSC populations pursuant to various treatments in cancer cells heterozygously carrying a p53 missense mutation, in cell culture and in tumor-bearing mice.

R273H p53 mutant is a promoting factor for drug resistance and induced EMT of colon cancer cells
It has been reported that p53 missense mutants, including those at "hotspot" codons 248 and 273, conferred oncogenic GOF in cells and in transgenic mice [24,25]. We examined cell response to doxorubicin in human COLO 320DM (COLO) and WiDr colon cancer cells, which carry homozygous p53-R248W and p53-R273H mutations [26,27], respectively. COLO and WiDr cells were significantly resistant to doxorubicin as compared to wt p53 SW48 colon cancer cells, or MCF-12A noncancerous epithelial cells ( Figure 1A left-panel). The IC 50 values for doxorubicin in COLO and WiDr are 36-fold (1.50 vs. 0.04 μM, p<0.001) and 18-fold (0.78 vs. 0.04, p<0.001) higher than in SW48 cells. Other missense mutant SW48/TP53 (TP53) cells, which heterozygously carry p53-R273H knocked in by using a CRISPR/Cas9 genome editing system [28], however, showed responses to doxorubicin similar to those of its parental SW48 colon cancer line (wt p53) ( Figure 1A rightpanel). To characterize the association of GOF with acquired drug resistance during chemotherapy, we cultured TP53 as well as SW48 cells in 10% FBS medium with sub-lethal concentrations of doxorubicin (5-25 nM) for approximately 26 passages. As shown in Figure 1A (right-panel), exposure to doxorubicin induced drug resistance in heterozygous p53-R273H mutant cells. The IC 50 value for doxorubicin in TP53-Dox cells increased by 24-fold (1255 vs. 49.2 nM, p<0.001) over that seen for naïve SW48/TP53 cells; however, the IC 50 values in SW48-Dox cells did not change significantly (45 vs. 50 nM) versus naïve SW48 cells ( Figure 1A right panel).
In addition to drug resistance, EMT plays critical roles in modulating cancer stem-like cell phenotype [34]. To test possible roles played by p53 R273H mutant in acquisition of cancerous stemness, we further investigated the effects of R273H on the self-renewal potential of cells using a sphere formation assay. We found that the numbers of enlarged tumor spheres increased in TP53-Dox cells, comparing to SW48-Dox cells. PDMP treatments substantially decreased sphere numbers, by twofold (33 vs. 67, p<0.001) in TP53-Dox cells ( Figure 3A); however, PDMP treatment did not significantly alter tumor spheres in SW48-Dox cells. We next looked for CSCs with CD44v6 + /CD133 + phenotype among cells in these lines, as a CD44v6 + /CD133 + phenotype has been associated with CSCs in tumors and in cells of colon cancers [35]. It was found that CSC populations increased fourfold in TP53-Dox cells (1.2 vs. 4.6%, p<0.001) as compared to in SW48-Dox cells ( Figure 3B). With PDMP treatment, CSC populations in TP53-Dox cells significantly decreased, by approximately 3.5-fold (4.6% vs. 1.23%, p<0.001), whereas PDMP treatments did not have a significant effect in SW48-Dox cells ( Figure 3B).

p53 R273H mutant promotes the induced pluripotency of cancer cells through Zeb1 and c-Myc transcription factors
To understand how R273H impacts pluripotency of cancer cells, we assessed the protein expression of p53 and p53-responsive genes. A previous report showed that isogenic TP53 cells expressed the R273H mutant in response to a presence of 5-fluorouracil [28]. In the present study, p53 expression, in TP53 and TP53-Dox cells, in each case assessed either as pan-p53 or as phosphorylated p53 only (pp53, Ser15), was not altered in response to DNA damage induced by doxorubicin, in contrast to behavior of SW48 and SW48-Dox cells, which carry wt p53 ( Figure 4A, 4C). Significantly, the protein levels of p53, pp53, p21 and Puma were decreased in TP53-Dox cells as compared to SW48-Dox cells. Long-term Dox exposure (26-passages) significantly enhanced the protein levels of β-catenin, TGF-β, and Oct4 in both SW48-Dox and TP53-Dox cell sublines, as compared to naïve SW-48 and TP53 cells, respectively ( Figure 4B, 4C). However, Dox exposure increased c-Myc and Zeb1 levels exclusively in TP53-Dox, but not in SW48-Dox cells ( Figure 4B, 4C).
Our previous studies showed that ceramide can restore wt p53 expression and functional activity in ovarian cancer OVCAR-8 and NCI/ADR-RES cells that carry p53 deletion mutations in codons 126-132 and 126-133, respectively [32,33]. Until now, it remained unknown whether or not cells harboring p53 missense mutants could be restored to express wt p53 protein. Intriguingly, we found that PDMP treatments, which inhibited GCS and increased cellular ceramide ( Figure 2B, 2C, 4D), restored wt p53 expression as well as p53-responsive proteins in TP53-Dox cells ( Figure 4A, 4B, 4C, 4D). The levels of pp53, and of several proteins encoded by p53-responsive genes (p21, Bax and Puma), are substantially enhanced in TP53-Dox cells treated with PDMP, and those levels are almost equal to those observed in SW48 cells that carry wt p53 ( Figure 4A). Moreover, PDMP treatments exclusively decreased the protein levels of β-catenin, TGF-β, c-Myc, Klf4, and Zeb1 in TP53-Dox cells ( Figure 4B, 4C), whereas PDMP treatment significantly lowered Oct4 in both SW48-Dox and TP53-Dox cells. These results, taken in aggregate with the observed array of expression level changes between cell lines and sub-lethal Dox exposure, indicate that increased ceramide levels brought about by inhibition of GCS restores wt p53 expression in cells harboring a p53-R273H mutant allele. Also, the promoting effects of p53 R273H mutation on the spawning of induced cancer stem cells are seen to be highly associated with Zeb1, c-Myc, β-catenin, and Klf4.
Western blot analysis indicated that TGF-β, c-Myc, Zeb1 and Oct4 are pronouncedly increased in TP53-generated tumors after doxorubicin treatments, as compared to SW48-tumors ( Figure 7B). PDMP treatments dramatically decreased the protein levels of β-catenin, c-Myc, Klf4, Zeb1 and Oct4 in TP53-tumors ( Figure 7B), while these combined treatments significantly decreased only β-catenin, Klf4, and Oct4 in SW48-tumors. These findings further corroborate the results of cell-culture studies, demonstrating that GOF associated with p53 R273H mutation induces the production of iPSCs via Zeb1, c-Myc and TGF-β transcription factors.

DISCUSSION
We demonstrated that the presence of heterozygous p53 R273H mutant is an underlying factor promoting the occurrence of iPSCs in colon cancer cells, and in xenografted tumors exposed to chemotherapeutic agents. GOF associated with mutant p53 proteins can enhance the ability of cancer cells to invade, metastasize, and poorly respond to chemotherapies, all of which attributes to CSC presence and pluripotent stem-like phenotypic character [38][39][40]. p53 clearly plays a major role in iPSC generation from noncancerous cells, in both attenuating reprogramming and controlling the qualities of reprogrammed cells [41]. p53 function in iPSCs exerts an indirect effect on proliferation arrest and on the restriction of mesenchymal-epithelial transition (MET) during its early phases [42]. Inactivation of p53 function is associated with the acquisition of stem-like phenotypic character in reprogrammed cells and in cancers [14,[43][44][45][46]. Heterozygous p53 missense mutations are more common than any other mutants in cancers. This is particularly true for sarcomas of Li-Fraumeni syndrome patients. Patients with p53 mutant cancers often have poor prognoses in treatments; however, to date it remains unclear which particular p53 mutants cause this to be so, how significant the impacts are for each mutant, and in each case the exact mechanistic bases for the clinical observations with respect to chemotherapy resistance. The studies presented herein, carried out in cell culture and in xenografted tumor-bearing mice, showed that low-dose doxorubicin induced EMT (Figure 1), augmented CSC numbers, promoted tumor growth, and conferred drug resistance (Figures 1, 3, 5, 7, 8) for cancer cells carrying a heterozygous p53 R273H mutation.
R273 is a mutation hotspot codon, with variants R273H, R273C and R273G occurring most commonly in tumor samples from patients. R273H and R273C, for which expression levels do rise in response to DNA damage and repair, bind less tightly to DNA, and tend to stimulate production of proteins having protective and repair functions, including BRCA1, TOPBP1 and MDC1, thereby leading to a more-aggressive phenotype [47]. We found that the presence of the heterozygous R273H allele led to inadequate levels (or even complete lack of appearance) of the p53-responsive proteins p21, Bax and Puma, which are normally induced in response to p53 after DNA damage. The response failure is especially noteworthy for p21, which strongly constrains cell proliferation in normally differentiated, non-cancerous cells [13]. More importantly, in the R273H mutantharboring cells, upregulated expression of Zeb1, β-catenin, TGF-β, c-Myc, and Oct4 was observed; these transcription factors reportedly can de-differentiate adult cells [12] and augment the malignant potential of reprogrammed cells after transduction [14,48]. Furthermore, the heterozygous presence of R273H mutant alone did not intrinsically confer drug resistance or directly induce EMT in cultured SW48/TP53 cells (Figures 1, 8); these characteristics arose only following chronic low-dose doxorubicin exposure, and extended to other chemotherapeutic agents besides doxorubicin. These observations indicate that the presence of the p53 missense mutation acts as an underlying promoting factor for the generation of iPSCs in tumors exposed chronically to anticancer drugs.
Resuscitating normal function of the p53 mutant protein when its mutant forms are extant constitutes an attractive therapeutic strategy for cancer treatments [49], and restoration of wt p53 protein expression levels and functional normalcy holds promise for targeting the majority of p53-mutant cancers more effectively [50]. The tumor-suppressing functions of p53 mainly rely on binding of its homotetrameric form to DNA, thereupon suitably activating or repressing the expression of p53responsive genes. Restoration of wt p53 expression can switch mutant phenotype to wt by reducing the formation of heterotetramers (wt proteins with mutant proteins) that bind to DNA in place of normal homotetramer, bringing about oncogenic GOF in p53-mutant-harboring cancer cells [50].
Our previous studies showed that inhibition of ceramide glycosylation, either by silencing of GCS expression with MBO-asGCS or by inhibiting GCS activity with PDMP, increased cellular ceramide levels and restored wt p53 expression as well as p53-dependent apoptosis in cancer cells carrying p53 deletion mutants [32,33]. In the study we report herein, suppression of ceramide glycosylation restored wt p53 expression in TP53 cells that carry a R273H missense mutation allele, with dramatic increases in the protein levels of pp53 and p21 ( Figures 4A, 6A). As a downstream consequence, When mutant-heterozygous (HZ) cancer cells are exposed to a sub-lethal dose of doxorubicin (Dox), DNA damage induces overexpression of p53 R273H, and the presence of this missense mutant protein upregulates Zeb1 and β-catenin (β-Cat) stem-like factors. The consequent gain-of-function (GOF) presents as epithelial-mesenchymal transition (EMT) and induced pluripotency of stem cells (iPSCs), lending to tumor growth. Inhibition of glucosylceramide synthase (GCS)-catalyzed ceramide glycosylation with PDMP increases cellular ceramide levels, bring about restoration of wt p53 protein expression, in turn increasing p53-responsive proteins, including p21 and Puma, thereby suppressing tumor progression. restoration of p53 function further prevented GOFassociated induced pluripotency of CSCs ( Figures 1B,  3, 5, 7A), by bring about decreased expression of Zeb1, β-catenin, TGF-β, cMyc and Oct4 (Figures 4, 7). Previous studies showed that targeting GCS overexpression enhanced the sensitivity of cancer cells to anticancer drugs and decreased tumor formation [36,[51][52][53][54][55]. In drugresistant cancer cells, elevated ceramide glycosylation led to increased levels of globotriaosylceramide (Gb3), which can upregulate multidrug resistant gene 1 (MDR1) [52,53] through activation of the β-catenin signaling pathway. Increased levels of Gb3 and gangliosides GD2 caused enrichment with CSCs in breast cancers; accordingly, inhibition of GCS lowered CSC numbers, as well as tumor formation [54,55]. The novel finding in present study is that the increased levels of ceramide ( Figures 2B, 2C, 8A, 8B), rather than decreased levels of glycosphingolipids, eliminated oncogenic GOF of p53 in iPSC, via reactivation of wt p53 expression in cancer cells carrying a heterozygous p53 missense mutation.
To our knowledge, this might be the first study showing that normal p53 expression and function can be restored in cancer cells heterozygously harboring a p53 missense mutant allele, in this case a R273H. A recent study documented that activation of chaperone-mediated autophagy degrades mutant p53, including R175H and R273H, and sensitizes cancer cells to treatment-responsive death [56]. Ceramide can mediate phosphorylation of SRSF1 to select wt p53 mRNA for protein expression in cancer cells carrying p53 deletion mutation [33]. Yu et al. recently reported that NSC319726, a thiosemicarbazone compound, can reactivate an R175 p53 mutant so as to upregulate the expression of p21, and thereby induce apoptosis for cancer treatment [57]. All these indicate it is feasible to reactivate wt p53 expression and anticancer activity in cancer cells carrying missense mutation, although further studies are required to figure out how post-transcriptional processing modulates p53 expression.

Cell viability assay
Cell viability was determined by quantitation of ATP, an indicator of live cells, using the CellTiter-Glo luminescent cell viability assay (Promega, Madison, WI) kit, as described previously [32,58]. Briefly, cells (4000 cells/well) were grown in 96-well plates with 10% FBS supplemented RPMI-1640 medium. Cells were treated with test agents in 5% FBS medium for 72 hours.
Tumors were fixed and maintained in paraffin blocks. Microsections (5 μm) of tumors were stained with hematoxylin and eosin (H&E), and characterized by a pathologist. For immunostaining, antigens were retrieved in steaming sodium citrate buffer (10 mM, 0.05% Tween-20, pH 6.0). After blocking with 5% goat serum in PBST, slides were immunostained, as described above.
To analyze CSCs in mice, resected tumors (~60 mg) were immediately dispersed in RPMI-1640 medium with collagenase IV (500 units/mL) at 37°C for 120 min with shaking (20 rpm). After filtration through a 70-μm cell strainer, cells were incubated with APC-CD44 v6 and PE-CD133/2 IgG for flow cytometry analysis as described above. Collected bone marrow cells (BMCs) from each mouse were counted with a hemocytometer. The ABCG2 + BMCs were analyzed by flow cytometry, following the incubation of BMCs with anti-ABCG2 antibody, as described previously [21].

ESI/MS/MS analysis of ceramides
Speciation of endogenous sphingolipids was accomplished with a Thermo-Fisher TSQ Quantum triple quadrupole mass spectrometer, operated in a Multiple Reaction Monitoring (MRM) positive ionization mode, as described previously [33,62,63]. Total cells, fortified with internal standards, were extracted with ethyl acetate/isopropanol/water (60/30/10 v/v). These extracts were evaporated to dryness, and reconstituted in 100 μl of methanol. Portions of the reconstituted samples were injected on the Surveyor/ TSQ Quantum LC/MS system, and gradient-eluted from the BDS Hypersil C8 column (150 × 3.2 mm, 3 μm) with a 1.0 mM methanolic ammonium formate/2 mM aqueous ammonium formated mobile phase system. The peaks for the target analytes and internal standards were identified and processed using the Xcalibur software. Calibration curves were constructed by plotting peak area ratios of synthetic standards, representing each target analyte, to the corresponding internal standard. Concentrations for samples were obtained from these calibration curves by linear regression, and cellular sphingolipids levels calculated by normalization against total cellular protein.

Tumor-bearing mice and treatments
All animal experiments were approved by the Institutional Animal Care and Use Committee, University of Louisiana at Monroe (ULM), and were carried out in strict accordance with good animal practice as defined by NIH guidelines. Athymic nude mice (Foxn1 nu /Foxn1 + , 4-5 weeks, female) were purchased from Harlan (Indianapolis, IN) and maintained in the vivarium at ULM. Animal studies were conducted as described previously [32,36,53]. Briefly, a cell suspension of SW48 and SW48/ TP53 (2-3 passages, 1×10 6 cells in 20 μl/mouse) was subcutaneously injected in the left flank of the mice. Mice were monitored by measuring tumor growth and body weight, under clinical observation. Once tumors were visible (2 mm in diameter), mice were randomly allotted to different treatment groups (5 mice/group). For treatments, PDMP (4.0 mg/kg once every 3 days) was administered intraperitoneally alone or with doxorubicin (200 or 300 μg/kg once every 6 days) for 32 days. Tumor volume was approximated by the formula L/2 × W 2 (where L is the length and W is the width). Tumors and metastases were examined and characterized by pathologist following H&E staining of tissue sections at Louisiana State University Health Sciences Center (Shreveport, LA).

Data analysis
All experiments were repeated 2 or 3 times. Data are expressed as mean ± SD. Two-tailed Student's t tests and ANOVA tests were used to compare the continuous variables in groups, using the Prism v5 program (GraphPad, San Diego, CA). All p<0.05 comparisons were regarded as statistically significant.