Treatment of colon cancer cells using the cytosine deaminase/5-fluorocytosine suicide system induces apoptosis, modulation of the proteome, and Hsp90β phosphorylation

Introduction

Suicide gene therapy consists of the transfer into tumor cells of a “killer gene,” which converts a nontoxic compound into a lethal drug. This nontoxic compound can be systemically administered without any side effect. This strategy has proven to be effective in several tumor models in animals (1, 2). The bacterial cytosine deaminase (CD) gene, which is not expressed in eukaryotic cells, encodes a protein capable of converting cytosine into uracil. When expressed in mammalian cells, this protein transforms the nontoxic antifungal agent 5-fluorocytosine (5FC) into the widely used chemotherapeutic drug 5-fluorouracil (5FU), and expression of the CD gene within cancer cells results in a tumor-targeted chemotherapy. Although bypassing the toxic side effects of a systemic chemotherapy is in itself an interesting topic, the key effects of this strategy lie in the existence of two associated bystander effects. The first one, called the local bystander effect, is known to induce a complete tumor regression even if only a small proportion of the tumor cells express the suicide gene (3, 4). This effect is due to the toxicity of 5FU in cells expressing the suicide gene and to the diffusion of 5FU to the neighboring cells. The second effect, the distant bystander effect, results in the regression of distant wild-type tumors, mediated by the triggering of an immune reaction after the destruction of the tumor expressing the suicide gene (5–8).

In previous studies, we have developed a CD/5FC-based gene therapy model against colon cancer (9). A plasmid vector containing the Escherichia coli CD gene has been introduced into a BDIX rat colon carcinoma cell line (PROb). Intrahepatic injection of the modified cells (PRObCD) into the liver of syngeneic animals generates a single experimental “suicide tumor.” Treatment of animals with 5FC induced the regression of CD-expressing tumors and rendered the treated animals resistant to challenge with wild-type tumor cells (PROb). Immunohistologic analysis of experimental tumors indicates that natural killer (NK) cells are the major immune components involved in this antitumor effect (7–9).

To dissect the mechanisms involved in the regression of the tumor in our gene therapy model, we have examined changes in protein expression that occurred during cell death in the PRObCD cells exposed to 5FC in vitro. Using Annexin V staining, we first show that 5FC induced apoptosis in CD-expressing cells. Proteomic analysis of 5FC-treated cells identified several proteins involved in protein chaperoning, detoxification, and the antioxidative network. Attentive examination of the function of the identified proteins revealed that most of them are involved in the resistance to anticancer drugs and apoptosis. In addition, Hsp60 and Hsp90β, which have been described to be involved in the activation of NK cells by stressed apoptotic tumoral cells and in antigen presentation are, respectively, up-regulated and phosphorylated on serine 254 (10–14). Our data corroborate previous works indicating that the functions of Hsp90 are regulated by reversible posttranslational modifications (15, 16) and suggest that activation of Hsp90β by phosphorylation contributes to tumor regression and activation of the immune system.

Materials and Methods

Cell Culture

DHD/K12/PROb (PROb) cells are a colon carcinoma cell line originating from a chemically induced colon cancer in BDIX rats (17). These cells are poorly immunogenic and induce progressive and metastatic tumors in syngeneic hosts. The cells were maintained in DMEM (Bio Whittaker) supplemented with 10% FCS. PRObCD cells, generated by the transfection of pCDβgeo plasmid expressing the CD gene (8) in the PROb cell line, were maintained in the same medium in the presence of G418 at 100 μg/mL. 5FC was dissolved in DMEM (15 mg/mL) and added in the culture medium at a concentration of 500 μg/mL.

Annexin V-FITC Assay

Following 12, 24, 48, and 72 h of 5FC treatment, culture media containing any floating cells were collected from the drug-treated or control cells and centrifuged. The remaining monolayer cells were detached using trypsin-EDTA. The floating and adherent cells were washed once with cold PBS and centrifuged. The two cell pellets (adherent and floating cells) were stained with FITC-labeled Annexin V (AnV) and propidium iodide (PI) according to the recommendations of the manufacturer. Ten thousand events were collected using a FACScan flow cytometer (Becton Dickinson). Flow cytometry data were analyzed using Cell Quest software. The distribution of cells was placed into the following three groups: viable (AnV−/PI−), apoptotic (AnV+/PI− and AnV+/PI+), and necrotic (AnV−/PI+).

Reagents

5FC was from ICN. The following antibodies were used: clone 4G10 antiphosphotyrosine antibody (Upstate Biotechnology), clone PY20 antiphosphotyrosine antibody (Zymed Laboratories), clone SPA-845 anti-Hsp90 monoclonal antibody (StressGen Biotechnologies), clone Ac-K-103 antiacetylated lysine monoclonal antibody (Cell Signalling Technology), antiphosphoserine polyclonal antibodies (Zymed Laboratories), and goat anti-mouse and goat anti-rabbit polyclonal antibodies conjugated to horseradish peroxidase (Dako). Trypsin, proteinase K, and endoprotease GluC were from Promega.

Western Blot Analysis

Following two-dimensional gel electrophoresis, proteins were transferred to polyvinylidene difluoride membranes (Millipore Corporation). Immediately after transfer, membranes were soaked in amido black (0.1% naphtol blue black, 10% methanol, 2% acetic acid), washed in 50% methanol and 7% acetic acid, and then scanned. Membranes were blocked with 5% (w/v) bovine serum albumin in TBST [20 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, and 0.1% Tween 20] for 1 h at room temperature and incubated with primary antibodies overnight at 4°C. The primary antibody concentrations used were 1:1,000 for antiphosphotyrosine clone 4G10 and clone PY20, 1:1,000 for the monoclonal antiacetylated lysine, 1:250 for the polyclonal antiphosphoserine, and 1:1,000 for the monoclonal anti-Hsp90. The membranes were washed thrice for 5 min with TBST and incubated in TBST containing the appropriate secondary antibody conjugated to peroxidase for 1 h at room temperature. The membranes were washed as described above, and the labeled proteins were detected using an enhanced chemiluminescence kit (Amersham International) according to the manufacturer's instructions. Detection was done with autoradiography films (BioMax Light Film; Eastman Kodak).

Sample Preparation for Proteome Analysis

Treated and nontreated cells were rinsed thrice with an ice-cold sucrose solution [250 mmol/L sucrose, 10 mmol/L Hepes (pH 7.5), 0.2 mmol/L CaCl₂]. Rehydration buffer (urea 7 mol/L, thiourea 2 mol/L, CHAPS 4%, and Triton X-100 0.24%) was added to the cell monolayer and incubated for 15 min at room temperature. To remove genomic DNA, spermidine (10 mmol/L) was added for an additional 15 min. After centrifugation (22,000 × g, 30 min, 15°C), the supernatant and the pellet were separated. The supernatant was precipitated with the Ready Prep 2D clean-up kit (Bio-Rad Laboratories). Proteins were resuspended with rehydration buffer and the protein concentration was determined by the Bradford assay (Bio-Rad Laboratories) using bovine serum albumin in rehydration buffer as standard curve.

Two-Dimensional Gel Electrophoresis and Protein Staining

Each sample was thawed, adjusted to a protein concentration of 2.3 mg/mL, and supplemented with 100 mmol/L of 2-hydroxyethyl disulfide (Acros Organics). The samples (700 μg) were loaded onto IPG ReadyStrips (pH 3–10; 17 cm, linear gradient; Bio-Rad), and rehydrated for 7 h at 20°C. Isoelectric focusing on the Protean IEF Cell from Bio-Rad was carried out as follows: 15 min at 250 V, a slow (15 h) voltage ramping to 10,000 V and a final step at 10,000 V up to 500,000 Vh. Immediately after focusing, immobilized pH gradient strips were wrapped in plastic trays (rehydration trays; Bio-Rad) and stored at −80°C. Prior to SDS-PAGE, proteins in the immobilized pH gradient were reduced in 4% SDS in 1.5 mol/L Tris (pH 8.8) containing 130 mmol/L of DTT and 30% glycerol for 15 min, alkylated in 4% SDS in 1.5 mol/L of Tris (pH 8.8) containing 130 mmol/L of iodoacetamide and 30% glycerol for 20 min. Finally, strips were equilibrated in 4% SDS, 1 mol/L of Tris (pH 6.8) and 30% glycerol for 90 min. The second dimension separation was run overnight at room temperature on a 1-mm-thick 10% polyacrylamide gel at a constant voltage up to 1,300 Vh (Protean II XL cell). Immediately after electrophoresis, gels were fixed in ethanol/phosphoric acid (40%:3%) for 2 h, rinsed with deionized water thrice for 30 min, and then stained for 16 h with Bio-Safe Coomassie Stain (Bio-Rad). Gels were scanned using a GS-800 calibrated densitometer (Bio-Rad). Gel patterns were compared using ImageMaster 2D Platinum (version 5.0, Amersham Bioscience). Relative quantification of the detected spots was made as a percentage of the total spot volume for every gel (% vol). After these normalizations, based on the total quantity of valid spots on each gel, heuristic clustering was used to blindly classify similar gels in order to confirm the two classes (control and 5FC-treated cells). Lastly, statistical differences in expression between spots from control and 5FC-treated cells were assessed through Kolmogorov and Wilcoxon tests. The selected spots were subsequently treated for identification by mass spectrometry.

In-gel Tryptic Digestion and Protein Identification by Mass Spectrometry

The protein spots were manually excised from the two-dimensional gel and digested according to a standard protocol. Briefly, after a washing step with 50% ethanol, proteins were reduced and alkylated with 10 mmol/L of DTT and 50 mmol/L of iodoacetic acid, respectively. The spots were washed twice with 50% ethanol, shrunken with acetonitrile and then subsequently digested with 100 ng of enzyme in 50 mmol/L of ammonium bicarbonate and 15% acetonitrile. The peptides were extracted with 0.5% trifluoroacetic acid, 50% acetonitrile, and 100% acetonitrile. Matrix-assisted laser desorption ionization-time of flight mass spectrometry used a Voyager DE-PRO (PerSeptive Biosystems) set in positive reflectron mode. Samples were spotted on a thin layer of nitrocellulose (5 mg/mL solution in acetone/isopropanol, 1:1), washed with 2 μL of cold water, and then overlaid with α-cyano-4-hydroxycinnamic acid (5 mg/mL solution in 50% acetonitrile and 0.3% trifluoroacetic acid). Spectra were obtained with standard settings and external calibration. Protein identification was done with Aldente software on the SwissProt web site.⁵ For the posttranslational modification analysis of Hsp90, LC-MS/MS mass spectrometry was done with a Surveyor system coupled with a LCQ DECA XP ion trap mass spectrometer (ThermoQuest). Peptides were separated on a C18 column (200 μm × 10 cm, Hypersil; ThermoQuest) at 5 μL/min with an acetonitrile gradient from 1% to 40% acetonitrile in 0.1% formic acid. A 3 kV voltage was applied on the micro-ESI needle. The automatic acquisition was set as previously described (18). Data processing was done with Bioworks 3.3. The main search variables were methionine oxidation, serine, threonine, and tyrosine phosphorylation or lysine acetylation as differential modification, no enzyme as cleavage specificity. Peptide identification was considered significant when Xcorr > 1.7, 2.2, and 3.3 for mono-, di- and tri-charged peptides, respectively; δCn > 0.1, Resp < 3, and peptide probability < 5E-3.

Results and Discussion

Before performing the proteomic studies, we verified that 5FC induced apoptosis in our CD-expressing colorectal adenocarcinoma cell line in a time-dependent manner. 5FC (500 μg/mL) was added to actively growing PRObCD cells. As shown in Fig. 1A , the number of PRObCD cells rapidly declined and 25% of the cells remained attached to the plastic vessel after 72 h of treatment, whereas nontreated PRObCD cells proliferated rapidly. To assess whether 5FC induced apoptosis, PRObCD-detached cells were stained with Annexin V and propidium iodide. Annexin V binds to phosphatidylserine, which shifts to the cell surface from the inside of the cell membrane during the early stage of apoptosis, whereas propidium iodide penetrates only in cells which are in the late stage of apoptosis and intercalates into genomic DNA (19). As shown in Fig. 1B, ∼90% of the detached cells were AnV+, and 10% were AnV+PI+ at 12, 24, 48, and 72 h of treatment, which shows that 5FC induces apoptosis in CD-expressing cells. The number of detached PRObCD cells without treatment was not significant (<0.01% compared with treated PRObCD cells).

To determine the apoptosis stage of cells which were still attached after 5FC treatment, we stained cells with Annexin V and propidium iodide. We detected three populations in the treated and untreated PRObCD cells: AnV+, AnV+PI+, and PI+, which corresponded to the early stage of apoptosis, late stage of apoptosis, and necrosis, respectively. Unstained cells were considered as living cells. As shown in Fig. 1D, AnV+ cells accounted for 23% and 16% of total cells in the presence and in the absence of 5FC, respectively. The absence of staining in 80% of cells and the small difference between treated and untreated cells for AnV+ and AnV+PI+ stainings suggest that detachment of treated PRObCD cells from the plastic vessel precedes the appearance of phosphatidyl serine in the outer leaflet of the plasma membrane. Finally, we show that 5FC, after being transformed into 5FU by CD, induces apoptosis in the colon adenocarcinoma cell line PRObCD.

We then analyzed PRObCD's proteome modification which was induced after 48 h of 5FC treatment. Two-dimensional gel electrophoresis of total proteins extracted from treated and untreated PRObCD cells was carried out to identify differentially expressed proteins.

Two additional controls were done: (a) PROb cells were treated with 5FC. We did not observe any cellular toxicity, apoptosis, and modification of the proteome (data not shown), and (b) PRObCD cells were treated with 5FU for 48 h. This treatment induced the apoptosis of PRObCD cells and modified the proteome in a similar extent compared with CD/5FC. This result suggests that the two killing mechanisms (i.e., CD/5FC and 5FU) are the same. However, only major proteins can be observed and analyzed on two-dimensional gels. We do not exclude that minor differences occurred between the two treatments.

A representative Coomassie blue–stained gel of proteins from 5FC-treated PRObCD cells is depicted in Fig. 2A . Replicates were carried out for both treated (5FC, n = 3) and nontreated cells (CT, n = 2). Only reproducible spots were considered as significant. Image analysis with ImageMaster Platinum identified 353 spot groups (i.e., the same spot on different gels) and Kolmogorov/Wilcoxon tests were used to discriminate the up-regulated and down-regulated proteins. Mass spectrometry analysis identified the corresponding proteins. Fourteen up-regulated proteins, six down-regulated proteins, and one protein exhibiting a significant shift in its isoelectric point were reproducibly detected from treated PRObCD cells compared with nontreated cells.

Matrix-assisted laser desorption ionization-time of flight mass spectrometry analysis of the tryptic fragments and a data search allowed for the identification of these proteins, as shown in Table 1 . The up-regulated and down-regulated proteins include cytoskeletal proteins, chaperone proteins, and proteins involved in protein synthesis, protein folding, the antioxidative network, and detoxification. The increase of proteins involved in the antioxidative network and detoxification are likely due to the conversion of 5FC into 5FU by the CD. PRObCD cells need to eliminate 5FU and its metabolic derivatives, FdUMP, 5-FUTP, and 5-FdUTP, which are responsible for the cytotoxic effects in cancer cells (20). The thioredoxin-like 2 protein is involved in the mitochondrial thioredoxin system, which controls redox balance, cell growth, and apoptosis. When thioredoxin levels are elevated, there is increased cell growth and resistance to the normal mechanism of programmed cell death (21). In addition, an increase in the thioredoxin-like 2 protein level seen in many human primary cancers compared with normal tissue seems to contribute to increased cancer cell growth and resistance to chemotherapy (22). The mitochondrial aldehyde dehydrogenase has been implicated in drug catabolism and has been identified as a major actor in the resistance to the anticancer molecule cyclophosphamide. This has been shown in the medulloblastoma cell line (23), and particularly, in leukemic cells in which an elevated level of aldehyde dehydrogenase is found (24). The genetic variability of aldehyde dehydrogenase is also linked to resistance in cyclophosphamide (25). The dihydropyrimidinase-related protein 2 belongs to the dihydropyrimidinase family, which is involved in the pyrimidine metabolism and in the catabolism of 5FU (26). Patients with a partial dihydropyrimidine deficiency proved to be at risk of developing severe toxicity after 5FU administration (27).

Besides proteins which are directly involved in detoxification, we also found that three cytoskeletal proteins involved in cell survival and drug resistance were increased. α-Actinin 4 binds to the actin cytoskeleton and integrins whose roles in matrix adhesion induced–cell survival are largely documented (28). α-Actinin is also up-regulated by many cytotoxic chemicals in embryonic stem cells and is now used as a biomarker of toxicity in in vitro tests (29). In addition, α-actinin is up-regulated in osteosarcoma cell lines (30). Ezrin is a cytoskeletal protein which has a major role in cell polarization through actin binding and the role of ezrin in metastasis behavior has been well documented (31). Ezrin binds to Fas and contributes to the Fas-related resistance of tumors to apoptosis (32). Furthermore, ezrin is connected to P-glycoprotein, which regulates the efflux pumps that are responsible for some of the multidrug resistance mechanisms of tumors (33). The glial fibrillary acidic protein which belongs to the intermediate filament family also binds to the P-glycoprotein, and therefore, may have a role in 5FU resistance (34).

We also found that three proteins involved in protein synthesis were up-regulated after 5FC treatment. These proteins (eIF2β, eIF4I, and eEF1G) belong to the initiation and elongation factor families. They are not just housekeeping proteins. For instance, eIF4E, which is structurally very close to eIFA1 and eIF2β, plays an important regulatory role in the control of cellular stress–induced apoptosis (35). Similarly, eEF1A2, which is structurally close to eEF1G, is antiapoptotic and has been identified as a tumorigenic oncogene regulating cell growth (36).

In addition, we found four up-regulated proteins which belong to the huge family of chaperone proteins. Chaperones limit aggregation, facilitate protein refolding, and are involved in signal transduction. Under conditions of cellular stress such as chemotherapy, intracellular chaperone levels increase in order to provide cellular protection and maintain homeostasis (37). The 14-3-3ε protein belongs to a highly conserved cellular protein family that can interact with >200 targets proteins. 14-3-3 proteins control cell cycle, cell growth, survival, and apoptosis (38, 39). In addition, 14-3-3σ has been identified as a marker of potential clinical interest for breast cancer detection and contributes to drug resistance in human breast cancer cells (40). We also observed that two chaperones, Hsp60 and Hsp90β, which are heat shock proteins (37), are up-regulated and subject to posttranslational modification, respectively. These two proteins have roles in protein chaperoning and in signal transduction during tumorigenesis (41). In addition, it has been described that Hsp60 and Hsp90 have roles in tumor immunogenicity and antigen presentation, respectively (10–12), and thus, might contribute to the activation of the NK cells that we observed in our gene therapy model.

Hsp90β was further analyzed in order to identify the posttranslational modification responsible for the acidification of the isoelectric point which shifted from 5.5 to 5.1 (Fig. 2B). Two major posttranslational modifications of Hsp90 have been described: phosphorylation and lysine acetylation. It has been shown that phosphorylation of Hsp90 on serine was linked to its chaperoning function. Serine phosphorylation triggers the release of the chaperone from the target protein (16). In addition, Hsp90 is phosphorylated in tyrosines during mammalian sperm capacitation, the process by which spermatozoa gain the ability to fertilize the oocyte (42). Finally, acetylation of Hsp90 regulates chaperone-dependent activation of the glucocorticoid receptor, lysine acetylation of Hsp90, leading to the loss of chaperone activity (15).

We did Western blot analyses on two-dimensional gel electrophoresis for the detection of phosphorylation and acetylation of Hsp90β. After transfer to a polyvinylidene difluoride membrane, the area corresponding to Hsp90β spots were hybridized with antibodies specific for phosphotyrosine, phosphoserine, and acetylated lysine. The membranes were also hybridized with anti-Hsp90 monoclonal antibodies as controls (data not shown). As shown in Fig. 3 , we did not detect any signal for phosphorylated tyrosine on Hsp90β from 5FC-treated and nontreated cell extracts. Hybridizations were done with two antiphosphotyrosine monoclonal antibodies, clones 4G10 and PY20, which are the most widely used antiphosphotyrosine antibodies. In contrast, phosphorylation of the acidic form of Hsp90β was detected with antiphosphoserine antibodies (Fig. 3). Finally, hybridization with antiacetylated lysine antibodies revealed the acetylation of the basic and acidic forms of Hsp90β. In conclusion, we show that 5FC treatment induces serine phosphorylation of Hsp90β and that Hsp90β is constitutively acetylated in PRObCD cells.

A large proteomics survey previously revealed that Hsp90β is acetylated on lysine 623 and phosphorylated on serines 254 and 260 in the nuclear protein fractions of HeLa cells (43, 44). Phosphorylation of Hsp90β on serine 254 is found in the mitotic spindle during division of HeLa cells (45), in synaptosomes from human cerebral cortex (46), and in the inner medulla of rat kidneys (47). In addition, Hsp90β is phosphorylated in vitro by casein kinase II on serines 254 and 225 (48). Finally, Hsp90β is phosphorylated on tyrosine 483 in Jurkat cells (49). Complementary mass spectrometry analysis with LC-MS/MS (ion trap) was done to determine the acetylation and phosphorylation sites of Hsp90β. In order to increase the coverage of the protein sequence, three different digestions were done: trypsin, gluC, and proteinase K. This allowed for the identification of 71% of the protein sequence (Fig. 4 ). Tandem mass spectrometry identified a phosphorylation on serine 254 and showed that serine 260 was not modified (Fig. 5 ). Tyrosine 483 was also identified as unmodified, confirming the negative results obtained with antiphosphotyrosine antibodies.

Regarding acetylation, Western blot analysis showed that the two forms of Hsp90β were acetylated. One acetylation has been previously described on lysine 623 but our MS/MS data clearly shows that this lysine was not modified. Two different peptides containing lysine 623 without acetylation were identified after trypsin and proteinase K digestion. The absence of detection of acetylated lysine with MS/MS, despite the positive signal obtained in Western blot analysis, suggests that acetylation was uncovered with mass spectroscopy (Fig. 4).

Finally, serine 225 was not covered with our MS/MS analysis. However, it is possible that this serine is also phosphorylated during 5FC treatment. We cannot exclude that in addition to phosphorylation on serine 254, one additional serine phosphorylation occurred on Hsp90β during 5FC treatment for the following reasons: (a) the pH shift that we found for Hsp90β is 0.4 pH units, whereas one phosphorylation usually induces an isoelectric point shift of 0.2 pH units on two-dimensional gels (50). (b) It has been shown that three phosphorylations on Hsp27 and one phosphorylation on Hsp70 induce an isoelectric point shift of 0.6 and 0.15 pH units, respectively (50).

The assessment of protein expression profiles by proteomic analysis after treatment with chemotherapeutic agents has the potential to identify novel signaling pathways involved in mediating the downstream response to these therapies, and could greatly facilitate the discovery of novel potential therapeutic targets and/or markers of chemoresistance. Our results show that treatment of PRObCD cells with 5FC induces the up-regulation and down-regulation of several proteins, most of them having a role in resistance to anticancer drugs and apoptosis. In this context, determination of the expression level of these proteins in different tumors might be of interest. It is possible that one of these proteins will constitute a very useful tool to predict resistance to CD/5FC and/or 5FU treatments.

In addition, we show the phosphorylation of Hsp90β on serine 254 during 5FC treatment. This protein may have a role in apoptosis which occurred in our gene therapy model. In this light, it has been described that Hsp90 has roles in protein chaperoning and resistance to chemotherapeutic drug–induced apoptosis (41). The resistance to apoptosis is mediated via the binding of Hsp90 to prosurvival client proteins such as Bcr-ABl, c-Raf1, and AKT, which promote their proper folding, assembly, and transportation across different cellular compartments. Failure of Hsp90 activity leads to misfolding of client proteins, which leads to ubiquitination and proteasome degradation (15). Therefore, the chaperone activity of Hsp90β should be tidily regulated. Although it has been known for some time that Hsp90 is a phosphoprotein, the significance of this phosphorylation event in the function of Hsp90 is entirely unknown. In this study, we have shown for the first time that Hsp90β is phosphorylated on serine 254 after anticancer treatment. The phosphorylation of serine 254, very likely, has an important role in the regulation of Hsp90β activity. It is possible that this phosphorylation induces a modification of the affinity of Hsp90β toward its client proteins. The phosphorylation of serine 254 might induce the release of client proteins, or inversely, the binding to new client proteins. In this context, it would be interesting to identify proteins bound to Hsp90β before and after 5FC treatment. Ablation of serine 254 using site-directed mutagenesis would also be interesting to dissect the role of this phosphorylation in its resistance to chemotherapeutic agents.

Besides its role in protein chaperoning and apoptosis, it has been described that Hsp90 has a role in antigen presentation (10–12). Therefore, it is possible that Hsp90β has a role in NK cell activation, which occurred in our gene therapy model (7–9). Another protein which may have a role in NK cell activation is Hsp60, which has been described to induce tumor immunogenicity (10–12). We have observed that this protein was up-regulated during 5FC treatment. Furthermore, stressed apoptotic tumor cells are capable of providing the necessary danger signals by increasing the surface expression of Hsp60, resulting in the activation/maturation of dendritic cells and the generation of potent antitumor T cell responses (12). Concerning Hsp90, it has been shown that this protein is able to present antigens to T lymphocytes via MHC class II molecules (13). In addition, surface expression of Hsp90 promotes γδ T cell proliferation in EBV-infected B cells (14). It is possible that phosphorylation of serine 254 occurring during 5FC treatment capacitates Hsp90β to stimulate NK cells after being released from dying PRObCD cells. Phosphorylation of serine 254 might also capacitate Hsp90β to present antigenic peptides from apoptotic PRObCD cells to T lymphocytes. To check our assumption, it would be interesting to evaluate the capacity of Hsp90β purified from 5FC-treated and untreated PRObCD cells to activate dendritic cells.