Expression of TP53 Isoforms p53β or p53γ Enhances Chemosensitivity in TP53null Cell Lines

The carboxy-terminal truncated p53 alternative spliced isoforms, p53β and p53γ, are expressed at disparate levels in cancer and are suggested to influence treatment response and therapy outcome. However, their functional role in cancer remains to be elucidated. We investigated their individual functionality in the p53null background of cell lines H1299 and SAOS-2 by stable retroviral transduction or transient transfection. Expression status of p53β and p53γ protein was found to correlate with increased response to camptothecin and doxorubicin chemotherapy. Decreased DNA synthesis and clonogenicity in p53β and p53γ congenic H1299 was accompanied by increased p21(CIP1/WAF1), Bax and Mdm2 proteins. Chemotherapy induced p53 isoform degradation, most prominent for p53γ. The proteasome inhibitor bortezomib substantially increased basal p53γ protein level, while the level of p53β protein was unaffected. Treatment with dicoumarol, a putative blocker of the proteasome-related NAD(P)H quinone oxidoreductase NQO1, effectively attenuated basal p53γ protein level in spite of bortezomib treatment. Although in vitro proliferation and clonogenicity assays indicated a weak suppressive effect by p53β and p53γ expression, studies of in vivo subcutaneous H1299 tumor growth demonstrated a significantly increased growth by expression of either p53 isoforms. This study suggests that p53β and p53γ share functionality in chemosensitizing and tumor growth enhancement but comprise distinct regulation at the protein level.


Introduction
The gene of the tumor suppressor p53 is shown to encode at least 12 different p53 protein isoforms through alternative splicing, promoter and translational initiation (reviewed in [1,2]) ( Figure 1A). The differential expression of several of these isoforms has recently been established in cancer, [3,4] though their functional role is not fully understood. Their structural characteristics may indicate isoform specific mechanisms. p53b and p53c lack the oligomerization domain ( Figure 1A) that is required for p53 tetramerization and thus influence p53 DNA binding and transcriptional activity. However, p53b has been shown to bind certain p53 promoters and form protein complexes with full-length p53. Furthermore, p53b and p53c is expressed in a tissue-specific manner, which may suggest diverse tissue-determined functions that may be reflected in cancer [5]. This complicates a simple understanding of p53 function, but may support future use of p53 isoform profiles in prediction of outcome and drug sensitivity in cancers.
We have previously reported that acute myeloid leukemia patients with high expression of p53b and p53c protein relative to full-length p53 protein respond better to intensive chemotherapy and have a significant longer survival after treatment [6]. Similarly, in chronic lymphocytic leukemia, there is a strong correlation between an accumulation of full-length p53 protein and inferior outcome [7]. In breast cancer, patients with mutated p53 have more than three times increased risk of recurrence and death compared to patients with wild-type p53, but co-expression of p53c and mutated p53 leads to similar beneficial prognostic outcomes as those expressing wild-type p53 only [8]. p53b is overexpressed in renal cell carcinomas compared to normal tissue and the p53b mRNA level is significantly associated with tumor stage in these cancers [9]. In addition the p53b concentration is associated with poorly differentiated ovarian cancer and in patients with functionally active p53, expression of p53b correlated with worse recurrence-free survival [10]. Furthermore, a frequent loss of p53b and p53c has been reported in head and neck squamous carcinoma [11]. These studies suggest that p53b and p53c may influence carcinogenesis and drug-sensitivity in an organ-and ratio-dependent manner, and emphasize the need to discern their individual function and regulation.
It is not known if p53b and p53c direct biological effects alone, or if they require full-length p53 to function [14]. Therefore, we expressed p53b and p53c individually in a p53 null background using the H1299 lung carcinoma and SAOS-2 osteosarcoma cell lines. We compared functional implications of individual expressed isoforms on clonogenicity, examined the proteasomal route of degradation, and tested functional impact on chemosensitivity in vitro and tumor growth in a xenograft model. p53b and p53c were found to have a chemosensitizing effect as well as an increased tumor growth potential in vivo.

Results
Stable expression of p53b and p53c in p53 null H1299 lung carcinoma cells In order to study isoform specific biology, retroviral constructs/ vectors containing p53b, p53c or full-length p53 ( Figure 1B) were generated to either retrovirally transduce or transiently transfect p53 null cancer cell lines. The p53 null lung carcinoma H1299 cell line was retrovirally transduced and FACS-sorted to generate stably expressing p53b, p53c or full-length p53 H1299 cell lines. Employing a retroviral vector containing the p53 isoform and a fluorescent protein marker (tdTomato), transduction and sorting of tdTomato + cells was performed twice ( Figure 1B; see Materials and Methods for experimental details). Sorted cells were evaluated for tdTomato expression by flow cytometry and fluorescence microscopy ( Figure 1C), and re-sorted for tdTomato expression if needed. Considerably fewer tdTomato + fluorescent p53c + cells were observed following all transductions when compared to transduction efficiencies obtained with p53b + cells (not shown). Full-length p53 + congenic H1299 cells could not be established, presumably due to the cytotoxic effect of p53 expression. p53 immunofluorescence showed a predominantly nuclear localization of p53b and both nuclear and cytoplasmic localization of p53c ( Figure 1C). p53b or p53c in H1299 cells was confirmed by PCR of TP53 segment sequencing (exon 1-12) of both strands and immunoblot of p53 protein isoforms (Figure 2A; for details see Materials and Methods section). Immunoblot showed that p53b was expressed at considerably higher levels compared to p53c (Figure 2A). 3 H-thymidine DNA-incorporation was measured to investigate the proliferative capacity of the transduced cells. A small, but significant reduction in proliferation was noted in H1299 p53b and H1299 p53c compared with H1299 transduction control (p,0.01; Figure 2B). The reduced proliferation of both p53b + and p53c + cells was also perceived by electric cell-substrate impedance sensing (ECIS) when compared to proliferation of vector control cells 30 hours after plating (p,0.05; Figure 2C). Performing a colony formation assay of H1299 p53b and H1299 p53c under hypoxic conditions also demonstrated a reduced tendency of colony formation compared with vector control (mean 6 SEM for vector control + cells; 93613, p53b + cells; 7264, p53c + cells; 6864 (n = 3) (not shown)). Immunoblots illustrated an increased basallevel of p21 (CIP1/WAF1) in both H1299 p53b and H1299 p53c cells, while little change in Bax was noted ( Figure 2E, 2D). A p53 promoter reporter assay was performed to determine if p53b or p53c would be able to activate the genes DDI2, ARG2, CDKN1A, E2F7, SERPINE1, TP53INP1 or TP73. However, no significant activation was detected in untreated or treated (camptothecin, doxorubicin) H1299 p53b or H1299 p53c cells (not shown). However, a transient transfection of H1299 p53b and H1299 p53c cells with a 136p53 responsive element (RE) coupled to a GFP reporter gene, and subsequent analysis for GFP expression by flow cytometry, indicated that p53b and p53c both may activate this p53 consensus responsive element ( Figure 2F).

Both p53b and p53c sensitize cells to chemotherapy
Following the observation that the isoforms may activate p53responsive genes, we examined the response of the H1299 p53b and H1299 p53c cell lines to chemotherapeutics with a colonyformation assay. The pyrimidine antagonist arabinofuranosyl cytidine (cytarabine, Ara-C), the cytotoxic antibiotic and topoisomerase II inhibitor doxorubicin (Dox) and the topoisomerase I inhibitor camptothecin (CPT) were tested. In both H1299 p53b and H1299 p53c, a significantly decreased colony formation compared to vector control was observed when treated with doxorubicin ( Figure 3A-C). Treatment with Ara-C and camptothecin (not shown) showed less effect on colony formation than doxorubicin, but nevertheless revealed a tendency towards reduced number of colonies. Decreased proliferation of H1299 p53b and H1299 p53c after treatment with doxorubicin and camptothecin (especially at higher dose) was also identified by a 3 H-thymidine incorporation assay ( Figure 3D). Transient transfection of p53 null SAOS-2 osteosarcoma cell line with p53b-and p53c-tdTomato construct followed by treatment with 0.5 mM doxorubicin for 24 hours, also showed significant reduced 3 Hthymidine DNA incorporation in p53b + cells (data not shown). This was not seen with the p53c or full-length p53, and we propose that this lack of doxorubicin toxicity may be caused by high cell death caused by the cell death induction potential of p53c or full-length p53.
Immunoblots of camptothecin-and doxorubicin-treated H1299 p53b, H1299 p53c and H1299 vector control cells showed a significant decrease in p53c and p53b expression indicating degradation of p53b and p53c protein ( Figure 4A). Camptothecin and doxorubicin treatment in H1299 p53b cells amplified the Bax and p21 (CIP1/WAF1) response ( Figure 4B,C). p53c + cells showed a significant decrease in Bax after doxorubicin treatment while p21 (CIP1/WAF1) was significantly elevated after camptothecin treatment ( Figure 4B,C). Furthermore, an increase in basal level of the kinase Chk1 was detected with p53b. Chk1 was significantly reduced upon camptothecin treatment ( Figure 4D) and no clear changes in Puma protein levels were seen between the groups ( Figure 4E). Decreased protein levels of Tigar (TP53-induced glycolysis and apoptosis factor) protein in p53b + and p53c + cells after doxorubicin treatment were observed ( Figure 4F). Increased basal Mdm2 levels were observed in both p53b and p53c, and reduced in all cells after chemotherapy treatments ( Figure 4G). microscopy confirms tdTomato expression (red) of FACS sorted H1299 cells. Scale bar: 100 mm. p53 (DO-1) immunofluorescence staining (green) show mainly nuclear localization of p53b and both nuclear and cytoplasmic localization of p53c. DAPI (blue) DNA stain visualize the nucleus. Scale bar: 20 mm. doi:10.1371/journal.pone.0056276.g001 p53b or p53c Enhances Chemosensitivity PLOS ONE | www.plosone.org

Proteasomal inhibition increases p53c but not p53b
The p53c protein was considerably degraded after doxorubicin exposure ( Figure 4A). To investigate the mechanism of degradation, H1299 p53b and H1299 p53c cells were treated with either the proteasome inhibitor bortezomib (Bzm) or the lysosome inhibitor chloroquine (Chq). Immunoblotting demonstrated that while p53c levels were elevated considerably subsequent to treatment with bortezomib, p53b displayed stable protein levels after bortezomib and chloroquine treatment ( Figure 5A). These findings was further confirmed at the subcellular level through immunofluorescence of the H1299 p53b and H1299 p53c cells ( Figure 5B) where an increase in fluorescence was observed for p53c after 8 hours of bortezomib treatment while the level of p53b appeared unaffected ( Figure 5B (i)). An incubation time of 8 hours with bortezomib was needed in order to inhibit p53c degradation since treatment for 1 hour, 2 hours and 4.5 hours did not result in an inhibition of p53c degradation and also no difference in the expression level of p53b was observed using these time points (data not shown). To further investigate the subcellular localization of p53c and p53b after bortezomib treatment as compared to untreated cells, the cells was investigated using a higher magnification where the untreated p53c cells were overexposed in order to capture the localization ( Figure 5B ii)). It is evident in both untreated and bortezomib treated cells that p53c is concentrated in the nucleus, excluded from the nucleoli and only observed diffusely in the cytoplasm. Without treatment, the p53b protein was concentrated in the nucleus, localized in the nucleoli and in a speckled nucleoplasmic pattern, in addition to diffusely in Sensing. Impedance values are normalized after initial cell stabilization and shown as ratio of normalization value. The graph shows results from four measurements of vector control cells and six measurements of p53b + and p53c + cells from two separate experiments. Standard error of mean is denoted by dotted lines. Highest variation in cell proliferation occurred after 30 hours after initiation, *P-value,0.05 calculated by paired Students ttest. (D) and (E) show immunoblot of basal level of Bax and p21, in both p53b + and p53c + H1299 cells. GAPDH act as loading control, and the ratio of p21 or Bax to loading control with control value set to 1.0 is indicated. (F) Transient transfection of H1299 p53b and H1299 p53c with the 136p53RE-GFP construct and H1299 wt cells with both p53FL-construct and 136p53RE-GFP were analyzed by flow cytometry (n = 2). Results are presented as a ratio of GFP positive cells in H1299 p53b, H1299 p53c and H1299 cells transiently transfected with full-length p53 to H1299 vector control. Error bars: standard error of mean. Student's t-test give P-value 0.053 of p53b cells versus vector control, and P-value 0.21 of p53c versus vector control. doi:10.1371/journal.pone.0056276.g002 p53b or p53c Enhances Chemosensitivity PLOS ONE | www.plosone.org the cytoplasm. After bortezomib treatment the nuclear localization of p53b appeared more diffuse. No change in p53c and p53b stability or subcellular localization was observed after exposure to chloroquine at the time points described above (data not shown).
To investigate if Mdm2 could have a role in degradation of p53b or p53c, the H1299 cells were co-treated with the p53-Mdm2 inhibitor nutlin-3 in addition to doxorubicin ( Figure 5C). The nutlin-3 treatment alone had no effect on p53c stability and the doxorubicin-induced degradation of p53c was not rescued by treatment with nutlin-3, indicating a Mdm2 independent proteasomal degradation of p53c ( Figure 5C). The p53b protein, on the other hand, was to a higher degree degraded after treatment with nutlin-3 compared to doxorubicin, indicating that Mdm2 is involved in p53b stability ( Figure 5C). To investigate whether p53c could be regulated by NQO1, as reported for full-length p53, p53c + -cells were treated with the NQO1-inhibitor dicoumarol. A dose dependent degradation of p53c after dicoumarol exposure demonstrates that NQO1 is indeed involved in the regulation of p53c. p53b was less affected by dicoumarol-treatment and only the high-dose treatment resulted in a minor reduction in the p53b protein level ( Figure 5D). In order to further map the degradation pathway of p53c after doxorubicin treatment, p53c + cells were  Figure 5E). This degradation was partly rescued by bortezomib treatment.

Growth advantage of p53b and p53c expressing H1299 cells in vivo but not in vitro
To examine the function of p53b and p53c protein expression in a more realistic cancer environment exhibiting hypoxia and low nutrition, we examined the growth of subcutaneous H1299 in NSG mice. Surprisingly, an early tumor growth initiation followed by a significantly increased tumor growth was found of both p53c and p53b congenic H1299 cells, compared to vector control ( Figure 6A). This was in contrast to in vitro findings whereby the 3 H-thymidine-incorporation assay showed only a minor decrease in proliferation of p53b + and p53c + cells ( Figure 2B). To evaluate if the growth factors present in the matrigel (for example TGF-b, epidermal growth factor, insulin-like growth factor and fibroblast growth factor) injected with the tumor cells could influence growth of p53b + and p53c + tumors, the same assay was performed without matrigel. However, the same pattern was observed with these tumors (Figure 6A, top left graph). Immunohistochemistry of   Figure 1C) demonstrating strong p53b signals (bottom images) with a major localization to the nucleus (insert), and p53c signals of predominantly cytoplasmic origins.

Discussion
Stably transduced H1299 cells demonstrated enhanced chemosensitivity to doxorubicin and camptothecin after introduction of p53b and p53c (Figures 3, 4). This was particularly evident in the colony formation assays, which reflects the total sum of all proliferative, differentiation, senescence and cell death effects [16]. Immunoblot analysis demonstrated an upregulation of p21 (CIP1/ WAF1) and Bax after exposure to doxorubicin and camptothecin, apparently through a p53-independent mechanism, but with an enhanced p21 (CIP1/WAF1) response in p53b + and p53c + cells especially in response to camptothecin. It has previously been reported that p53b may bind to the p21 (CIP1/WAF1) promoter sequences [5]. p21 (CIP1/WAF1) promotes cellular arrest, but may also promote apoptosis through both p53-dependent and independent mechanisms under certain cellular stress (reviewed in [17]), dependent upon upregulation of pro-apoptotic Bax [18]. This may also explain the reduced 3 H-thymidine incorporation observed in CPT-treated p53b cells. We found that p53b and p53c may have an effect on an optimized p53-responsive element (Fig. 2F), but a direct activation of p21 (CIP1/WAF1) promoter assay was only indicated and not significant (data not shown). However, the basal level of p21 (CIP1/WAF1) protein was found to be increased by p53b and p53c expression ( Figure 2E). Furthermore, the basal levels of Mdm2 were also found to be increased by p53b and p53c expression ( Figure 4G). Although a p53 promoter reporter assay did not detect activation of the genes tested (data not shown, see Materials and Methods section for details), a weak positive signal by a 136p53 responsive element reporter was observed ( Figure 2E). Thus, we cannot rule out p53b and p53c modulation of these p53 targeted genes at the posttranslational level or a weak gene induction not detected by the promoter reporter assay.
We find that resting cells stably transduced with p53c only express low levels of p53 isoform protein, corresponding with previous reports that p53c may be cytotoxic [19]. We propose that this low level is sufficient to induce muted levels of p21 (CIP1/WAF1) protein and is accompanied by a tendency towards decreased proliferation and clonogenicity in vitro. Further experiments are needed to determine if the isoforms respond specifically to different chemotherapeutics.
Protein levels of p53b and p53c decreased after treatment with camptothecin or doxorubicin, and p53c was particularly attenuated following therapy with doxorubicin. Through treatment with proteasome-, lysosome-and a NQO1-inhibitor, we suggest that the stability of p53b and p53c are differentially regulated ( Figure 5). Treatment with nutlin-3, a Mdm2-binding inhibitor of the Mdm2-p53 interaction, did not result in an increased level of p53b or p53c, suggesting that Mdm2 is not a negative regulator of p53b/c and consistent with previous reports [13,19]. However, our experiments indicate that p53c is degraded by the proteasome (Figure 5). Conflicting reports exist on proteasomal degradation of p53b and p53c [13] [20]. It is also reported that Mdm2 interact with both isoforms, but only promote ubiquitination of p53b. However, Mdm2-promoted stabilization of p53b is suggested through neddylation [13]. This could explain the decreased level of p53b we observed after treatment with nutlin-3 ( Figure 5C). It was recently suggested that p53 may be proteasomally degraded by default in an Mdm2 and ubiquitin-independent manner, and that p53 is stabilized by the NAD(P)H quinone oxidoreductase NQO1 [15]. Treatment of H1299 p53c and p53b cells with the NQO1 inhibitor dicuomarol ( Figure 5) resulted in a dose dependent degradation of p53c but not p53b, suggesting that NQO1 may be an important enzyme in proteasomal processing of p53c. This was further confirmed with an increased degradation of p53c after combining doxorubicin and dicuomarol treatment ( Figure 5E). Together, these observations emphasize the importance to further characterize the route of degradation of p53b and p53c.
p53 is proposed to play a role in metabolic regulation of tumor growth, including tumor responses to hypoxia and nutritional deprivation [21,22]. Therefore, we compared tumor growth in a subcutaneous xenograft. Both H1299 p53c and H1299 p53b lead to significantly increased size of tumors compared to vector control cells. A critical stage of early tumor progression is an adaptation to hypoxic and acidic conditions and a change to aerobic glycolysis to promote further tumor expansion [23,24]. Thus, the finding that p53b and p53c cells caused a significantly earlier initiation of tumor growth indicates that they play a role in the adaptive response to metabolic stress. It has recently been suggested that p53 isoforms may have more specific metabolic functions that promote metabolic adaptation [21], and tumor growth advantage in serum-nutrient starvation is suggested through modulation of p21 (CIP1/WAF1) [25]. Engraftment without the use of matrigel resulted in the same tumor size profile, eliminating interference by the matrigel basement membrane matrix injected with the cells (Figure 6, insert). Further investigation of the role of the various isoforms and their response to metabolic stress may contribute to additional understanding of their role in cancer.
In summary, we suggest that p53b and p53c individually imply functional effects in cancer cell lines. Future studies are needed to investigate if the function of p53b and p53c at defined expression levels, and to delineate the mechanisms of p53 isoform regulation in cancer growth and chemotherapy.

Design of p53isoform-tdTomato constructs
Expression vectors for p53 isoforms p53FL, p53b, and p53c were generously provided by Dr. Bourdon (University of Dundee, Scotland, UK). The p53 segments from these plasmids were cloned into an MMLV retroviral vector (L335, D.R. Micklem, unpublished) upstream of an IRES-tdTomato reporter gene. This vector drives constitutive transcription of a bicistronic mRNA comprising the cloned gene followed by an internal ribosome entry site and the red fluorescent protein tdTomato. The predicted sequence of the vector is available upon request. p53 isoforms were amplified by PCR using a forward primer containing EcoRI and SfiI sites (at gaa ttc ggc cat tac ggc cac acc ATG GAG GAG CCG CAG TCA GAT) and reverse primers containing BamHI and SfiI sites (p53FL: ta gga tcc ggc cga ggc ggc cat ata TCA GTC TGA GTC AGG CCC TTC; p53b: ga gga tcc ggc cga ggc ggc cat cta AGG CAA AGT CAT AGA ACC ATT; p53c: gc gga tcc ggc cga ggc ggc cga ata CAC GGA TAA TAT TTT CAA CTT; overlap with the p53 gene in capitals). After digestion with SfiI, the p53 isoforms were cloned into matching SfiI sites upstream of the IRES-tdTomato reporter gene. Correct inserts were confirmed by sequencing.

Retroviral transduction of NCI-H1299 cells
NCI-H1299 (p53 2/2 ) cells made to stably express p53b, p53c and control by retroviral transduction with the p53b-tdTomato vector, p53c-tdTomato vector, and tdTomato only vector (transfection control). Production of infectious retroviral vector particles in 293-based Phoenix A packaging cells and infection of cells were carried out as described [26].

Sequencing
The p53 isoform-tdTomato construct were sequenced both prior to transduction and after transduction into H1299 cells to confirm correct TP53 isoform sequence. DNA was purified from the cells using DNeasy Blood and Tissue Kit (Qiagen Inc., Valencia, CA, USA), and the concentration was calculated by NanoDrop UV-Vis Spectrophotometer (Thermo Scientific, Wilmington, DE, USA). By PCR using a forward primer for p53 (59-39) a reverse primer for p53b (59-39) and a reverse primer for p53c (59-39) [5] the p53 isoform product were amplified. This product was characterized by agarose gel separation to confirm segment, and purified using ExoSAP-IT according to suppliers instructions (USB Corporation, Cleveland, Ohio USA). Sequencing PCR was performed using the abovementioned primers in addition to primers towards the middle of the p53 sequence to make sure that the sequencing reaction detects the whole segment: p53 forward primer (59 gg ccc atc ctc acc atc atc-39) and reverse primer (59-c agg gga gta cgt gca agt-39), with the BigDye Terminator v1.1 Cycle sequencing Kit (Applied Biosystems, Foster City, CA, USA). Sequences were analyzed using DNA sequencing chromatogram p53b or p53c Enhances Chemosensitivity trace viewer FinchTV v1.4.0 (Geospiza Inc., Seattle, WA, USA) and EMBOSS Pairwise Sequence Alignment Matcher.

Colony Formation Assay
A total of 500 cells were seeded per 10 cm 2 dish in 8 ml of RPMI supplemented with 10% FBS, 1% penicillin/streptomycin and 1% L-glutamine. Treatment with 5 nM camptothecin, 25 nM doxorubicin, and 0.1 mM arabinofuranosyl cytidine (cytarabine, Ara-C) was initiated following 3 days of culture, and the assay terminated at day 10. Colonies were washed twice in 16PBS before 3 ml of 95% methanol was added for 3 minutes, washed in 16PBS and stained with 1:5 solution of tryphan blue stain solution (Thermo Fisher Scientific, Hanover Park, IL, USA) in ddH 2 O. Colony assay was also performed at hypoxic conditions with 1.5% O 2 .

H-thymidine DNA incorporation
H1299 cells (2610 3 or 5610 3 ) were seeded in 96 well plates and left to settle for 20 hours prior to treatment. The cells were treated for 8 or 12 hrs, and 3 H -thymidine (1 mCi per well; TRA310, Amersham International, Amersham, UK) was added the last 6 or 10 hours of the treatment period, respectively. For basal proliferation experiments the cells were incubated with 3 H-thymidine for 10 hours. Cells were harvested and DNA synthesis was determined by 3 H-thymidine incorporation assays as described [31]. Statistical analysis was performed using GraphPad PRISM (version 5.0b, GraphPad Software, Inc., La Jolla, CA, USA) software. Groups were compared using paired Student's t-test.

Electric-Substrate Impedance Sensing cell proliferation assay
Growth potential of the H1299 cells were assessed using by Electric-Substrate Impedance Sensing (ECIS; Applied Biophysics, Troy, NY) [32,33]. 2610 4 cells were seeded at 5610 4 /ml concentration in 8W10E+ plates coated with cysteine, and cultured in RPMI supplemented with 10% FBS. Impedance was measured at 64 kHz every 30 seconds for two days.
p53 Immunofluorescence 2610 5 H1299 p53b, H1299 p53c and H1299 control cells (tdTomato only) were grown on coverslips immersed in 0.5 ml RPMI medium with 10% FBS and 1% L-glutamine. The cells were fixed and permeabilized with 4% paraformaldehyde and icecold 99% methanol, respectively, before blocking with 0.5% BSA (Roche Diagnostics GmbH) in 16PBS. Next, cells were treated with primary p53 antibody (1:100 or 1:50, mouse anti-human p53, cat. 554293 BD Pharmingen) diluted in 16PBS with 0.5% BSA and incubated at 4uC over night before incubation with secondary antibody (1:5,000 of Alexa 488 goat anti-mouse (Invitrogen Molecular Probes)) diluted in 16PBS with 0.5% BSA, and incubated in the dark for 1 hour at room temperature. Last, the coverslip was washed 16PBS and mounted in 5 ml Fluoro-gel II with DAPI (Electron Microscopy Sciences, PA, USA). Images of cell fluorescence were acquired with a Zeiss Axio Observer Z1 inverted microscope (Carl Zeiss Microimaging GmbH, Germany) and analyzed by the AxioVision 4.8.2 software.
Flow cytometric analysis 1-5610 6 transduced H1299 cells were washed twice in 16PBS and suspended in 16PBS at a concentration of 5610 6 cells/ml. Stably expressing tdTomato + cells were isolated by a Fluorescence Activated Cell Sorter (FACSAria, BD Biosciences) using a 532 nm laser. TdTomato expression was regularly evaluated on an Accuri (Accuri Cytometers Ltd., St. Ives, Cambs UK) flow cytometer to keep TdTomato expression equal between the three H1299 subclones, and cells were re-sorted if needed.

General animal care and ethics statement
All experiments were approved by The Norwegian Animal Research Authority under study permit number 2008 070, and conducted according to The European Convention for the Protection of Vertebrates Used for Scientific Purposes. Mice capable of engrafting human cancer cell lines NOD/LtSz-Prkdc scid /IL2Rc null mice (abbreviated as NSG) [34,35] originally from Dr. Leonard Schultz, Jackson Laboratories, Bar Harbour, ME, USA) were housed in groups of five or less in individually ventilated cages (Techniplast, Buguggiate, Italy) and kept on a 12 hrs dark/light schedule at a constant temperature of 21uC and 50% relative humidity. The mice had continuous access to food and autoclaved water. p53b or p53c Enhances Chemosensitivity Subcutaneous tumor model 5610 6 NCI-H1299 p53b-tdTomato, NCI-H1299 p53c-tdTomato, NCI-H1299 tdTomato and NCI-H1299 wt cells were suspended in 100 ml sterile 16PBS with 10% Matrigel (BD Matrigel TM Basement Membrane Matrix, BD Biosciences) for s.c. inoculation, and injected with a 28 G syringe. Animals were monitored closely for tumor growth, and tumor volume was measured twice weekly using digital calipers. The mice were euthanized when tumor size reached 1000 mm 3 .

Histology
Tumor samples collected following euthanasia were transferred to a tube containing 4% formalin for paraffin embedding, cryosectioning and subsequent immunohistochemistry of the samples. Sections were stained with hematoxylin and eosin (H&E) and p53 (DO-7, Dako) (appearing as brown stain). Results were analyzed by standard light microscopy (Olympus BX51, Olympus Corp., Tokyo 163-0914, Japan).