There is increasing experimental evidence that therapy-induced tumor cell modification (cell-based gene and ionizing radiation therapies) may promote the immune system or induce the expression of cytokines or other factors that may be inhibitory for local or distant unmodified tumor cells.1, 2, 3, 4, 5, 6, 7, 8
Cell-based gene therapy after cytokine or suicide gene transfer has been investigated for autologous and allogeneic vaccination in cancer therapy. Tumor cells engineered to produce cytokines such as interleukin-12 (IL-12) induced antitumor immunity capable of inhibiting a challenge with unmodified parental tumor cells.1 Recently, in a model of allogeneic tumor cells transduced with IL-12 mixed with autologous lymphoma cells, IL-12 induced microenvironmental changes that, through the induction of an immune response, yielded eradication of distant preestablished lymphoma.2 Cells transfected with a suicide gene such as that of thymidine kinase from herpes simplex virus3, 4, 5, 6 or of cytosine desaminase,7, 8 enzymes that activate a drug or prodrug, inhibited distant tumors and provoked the regression of tumor cell mixtures with few suicide gene-transfected cells. These studies demonstrated that gene-modified cells communicated with nearby or distant nonmanipulated tumor cells, a phenomenon known in ionizing radiation studies, the so-called bystander effect. Bystander effects require either direct cell–cell contact or the presence of soluble circulating factors that might contribute to a distant antitumor effect as demonstrated by medium transfer experiments.9, 10, 11, 12, 13, 14, 15
At present there is no report of alkylating agent induction of such bystander effects. It is known that chloroethylnitrosourea (CENU) causes DNA damage.16, 17 However, residual tumors after CENU treatment may exhibit genomic instability as shown to be associated with ionizing radiation-induced bystander effects,14 phospholipid metabolism alterations18, 19 and other alterations.20 Also, in clinical studies, sustained favorable responses may occur after a delay from chemotherapy, suggesting that therapy-induced tumor cell modifications may play a role in tumor regression.15 Recently, it was reported that when a CENU-treated residual tumor was challenged by parental tumor cells, the untreated secondary tumor implanted at a remote location from the treated tumor underwent growth inhibition,20 suggesting that primary treated tumors could release bioactive factors with inhibitory effects.
Therefore, the present study was undertaken to test the hypothesis of whether CENU-treated melanoma released soluble factors mediating the secondary tumor response. First, starting with in vivo experiments, we showed that serum from CENU-treated tumors was able to arrest proliferation of untreated tumors, decrease vessel formation and induce specific tumor metabolite changes. These experiments demonstrated that primary tumors were at the origin of the release in the serum of the bioactive soluble factors responsible for the phenotypes changes. Second, turning to cell culture medium transfer experiments, we showed that the conditioned medium from CENU-treated cells cultures provoked similar phenotype changes, growth inhibition and specific tumor metabolite alterations. The latter data confirmed the presence of bioactive soluble factors in vitro. The modified cell phenotype had close similarities with that observed in vivo on secondary tumors. Proteomics of conditioned media using two-dimensional gel electrophoresis followed by mass spectrometry analysis demonstrated that CENU-treated cells had an altered protein secretory activity. However, the role of protein in the observed bystander effect was questionable. In conclusion, this article provides evidence of the release by CENU-treated tumors of soluble factors responsible for a bystander effect, involving growth inhibition, differentiation, reduced neovasculature and tumor metabolite alterations. This finding shows persistent biological activity of residual tumors after chemotherapy that may account for the delayed responses to treatment, and may represent a relay for chemotherapy.
A member of the CENU drug family, cystemustine (N′-(2-chloroethyl)-N-(2-(methylsulphonyl)-ethyl)-N′-nitrosourea) (Orphachem, Clermont-Ferrand, France), was used in experiments and was prepared as 5 mM solution in 0.9% NaCl before being used in vitro in fresh culture medium or in vivo by injection to mice.
Cell culture and cell treatments
The transplantable B16 (F1) melanoma cells originating from C57BL6/6J Ico mice were obtained from ICIG (Villejuif, France) and adapted to grow in culture.12 The melanocytes were maintained as monolayers in culture flasks using culture medium consisting of Eagle's MEM-glutaMAX medium (Life technologies). Cells were harvested by trypsinization and plated 20 hr before treatment for 2 hr with 200 μM of CENU. After treatment, melanocytes were rinsed in PBS and maintained in fresh culture medium for 60 days. At different time, cells were harvested and prepared for cell analysis.
In vivo models
Six- to 8-week-old C57BL6/6J male mice were purchased from IFFA CREDO, France. All procedures were approved by the Animal Experimental Ethical Committee. C57BL6/6J recipients were inoculated with 5 × 105 syngeneic B16 melanoma cells on 1 flank at day 1 of the experiment. Tumors became palpable at days 7–9 after cell inoculation. Primary tumors were treated by 15 μg/g CENU at days 11–14–18 intratumorally. At day 32, mice were challenged s.c. with 5 × 105 parental melanoma cells on the opposite flank (n = 11). Primary and secondary tumors were followed until day 65. A naive group of mice (n = 8) was injected with B16 cells at the same time. Tumor surface areas were calculated by measuring the length and width with a caliper square. Mean tumor growth curves were fitted the Gompertz shape.21 At different times of evolution, a subset of mice was sacrificed according to institutional guidelines for animal's welfare. Tumors were removed, dissected, weighted and prepared for histology and ex vivo NMR spectroscopy examinations.
Administration of serum from CENU-treated recipients
Two groups of mice were used, one as donors of serum and the other one as recipients of serum. Sera from tumor-bearing untreated (Tum+ CENU−) and CENU-treated (Tum+ CENU+) donors were taken at day 30 of tumor evolution. Mice were killed and total mixed blood (arterial + venous) was collected into a tube. Sera were sampled after centrifugation and were kept frozen until use. As controls, sera from untreated (Tum− CENU−) and CENU-treated tumor-free (Tum− CENU+) donors were collected. Fifty microliters of each serum was administered intratumorally to melanoma tumor-hosting recipients at days 11, 12 and 13 from cells inoculation (n = 10 mice for each type of serum).
Tumor pieces were fixed in formol solution. Paraffin-embedded sections were cut into 4 μm sections, and tissue section was prepared for hematoxylin-eosin staining and routine pathological analysis.
Challenged cell culture model
Melanoma B16 cells were treated for 2 hr with 200 μM CENU. After treatment, cells were maintained in culture for 60 days. Three times a week, the supernatant (the conditioned medium) of treated cells was collected, centrifuged and kept frozen until use. The conditioned medium was collected during 3 sequential phases after CENU treatment, from day 1 to 10, day 13 to 35 and after day 35, during long-term cell culture. Supernatants were mixed with fresh medium (1:2 v/v) to protect against serum deprivation effects. Untreated melanoma cells were exposed to the conditioned medium. As controls, supernatants from long-term culture of untreated cells were collected and processed in the same conditions.
Protein synthesis inhibitor test
Untreated and CENU-treated cells were cultured in a medium supplemented with 1.5 μg/ml of cycloheximide (Sigma) for 2 days. Also, the supernatants were denatured by heating at 100°C for 1 hr.
Protein content of cells and supernatants was measured as follows. Intact cell pellets were lysed using a lysis buffer (50 mmol/l Tris-HCl pH 8, 100 mmol/l NaCl) containing a protease inhibitor mixture (Roche) by ultrasonication (3 times for 15 sec in ice). After centrifugation (14,000g, 10 min at 4°C), the supernatants were kept at −80°C until analysis. Protein content was determined with Commassie Blue (Pierce) at 595 nm with bovine serum albumin as standard.
Cell growth inhibition assay
Following exposure of cells to a conditioned medium or to a standard medium, cell viability was determined by the trypan blue exclusion assay. Cell proliferation under conditioned media was followed during 2 weeks.
Flow cytometry was used to evaluate cell cycle and reactive oxygen species (ROS) generation.
Cell cycle analyses of control and treated cells was performed after the cells were washed in PBS and frozen in nitrogen for a few seconds, and then stained with 4 μg/ml propidium iodide. The cell cycle distribution was determined from histograms of DNA content using the MULTICYCLE software.
ROS generation was determined after treatment by CENU or exposure to a conditioned medium by staining cells with 20 μM of 2′,7′-dichlorofluorescin diacetate (DCFH-DA) for 30 min at 37°C in the dark; then the cells were washed in HBSS solution. The analysis of samples was carried out by flow cytometry. The specific fluorescence signal corresponding to DCFH-DA was collected with a 485 and 530 nm band pass filter. As positive control, cells were treated with either tert-butylhydroperoxyde (tBh) used at 0.2 mM or ethanol used at 20 mM.
Proton high resolution magic angle spinning NMR spectroscopy
Analysis was performed on a small bore Bruker DRX 500 magnet, equipped with a high resolution magic angle spinning probe. Samples, consisting of a piece of intact melanoma tissue below 50 mg or of pellet of intact cells, were inserted into 4 mm diameter zirconia rotors and rotated at 4 kHz.
Metabolite profiling was performed based on a conventional one-dimensional 1H saturation recovery sequence (repetition time, 10 sec; spectral width, 10 ppm; complex data points, 16 K; number of samples, 64; water signal presaturation at low power). Quantification was performed on an O2 workstation (Silicongraphics) using the built-in deconvolution software (XWIN-NMR v2.6, Bruker), and normalization was performed relative to the water signal as already reported.22 Measured derivatives included phosphoethanolamine (PE) (CH2 signal at 3.99 ppm), phosphatidylcholine (PtdCho) (N(CH3)3 broad signal at 3.26 ppm), taurine (Tau) (CH2 signal at 3.43 ppm), glycine (Gly) (CH2 signal at 3.56 ppm), lysine (Lys) (ε-CH2 signal at 3.00 ppm), total creatine (tCr) (CH2 signal at 3.93 ppm), lactate (Lac) (CH3 signal at 1.33 ppm), alanine (Ala) (CH3 signal at 1.47 ppm). Also total glutathione (GSx = reduced + oxidized glutathione) (CH2 signal at 2.55 ppm) was quantified. Because the latter signal was common to GSH and GSSG, no differentiation was made between the reduced and oxidized forms of glutathione. Polyunsaturated fatty acid (PUFA) content was quantified from its CHCH2CH broad signal centered at 2.85 ppm, after normalization to the ε-CH2 signal of Lys, a quite constant metabolite in all tumor groups.
One-dimensional NMR spectroscopy metabolite profiling was completed by a two-dimensional proton total correlation spectroscopy (TOCSY) sequence (repetition time, 1.5 sec; spectral width, 6 ppm × 6 ppm; complex data points, 2 K × 256; number of samples, 16; spin mixing time, 75 msec; water signal presaturation at low power). TOCSY spectra were processed using the XWIN-NMR software. Cross-peaks originating from the nJHH coupling between intramolecular moieties were highly specific of the corresponding metabolite. Metabolites followed by TOSCY spectra were GSx (cross-peaks at 2.17–2.55 ppm and 2.96–4.58 ppm), PE (cross-peak at 3.23–3.99 ppm) and PUFA (cross-peak at 2.85–5.33 ppm).
Two-dimensional gel electrophoresis and mass spectrometry of media
A volume of 0.5 ml of serum or a conditioned medium was precipitated with trichloroacetic acid and resolubilized with 500 μl of lysis buffer containing 8 M urea, 4% CHAPS, 0.5% IPG buffer, 20 mM DTT. Proteins were quantified using bovine albumin serum as a standard. First dimension isoelectric focusing (IEF) was carried out using the IPGphor system. Precast (18 cm, pH 4–7) IPG dry strips were rehydrated overnight with 350 μl of buffer containing 8 M urea, 4% (v/v) CHAPS, 20 mM DTT, 2% (v/v) IPG Buffer (pH 4–7) and bromophenol blue. For analytical separations, 50 μg of proteins were included in the rehydration buffer and IEF proceeded for up to 60,000 Vhr. For micropreparative separations, 250 μg of proteins were used and focusing was extended to 100,000 Vhr. IPG strips were stored at −80°C. Second dimension SDS-PAGE was performed using the Ettan DALT system (Amersham Biosciences). Electrophoresis was run overnight at 4°C and 7 W per gel, and analytical gels were silver stained and gel images were acquired with a laser image scanner. Molecular weights, isoelectric points and spot variations were measured using the Image Master 2D Elite software (Amersham Biosciences). Up- and downregulations were determined by statistical comparison of normalized spot volumes Student's t test, p < 0.05. Micropreparative gels were stained with colloidal Coomassie blue. Protein spots were excised and analyzed on a BiFlex III MALDI-TOF mass spectrometry (Brucker Daltonics). The procedure was performed in triplicate.
In all experiments, data are given as mean ± standard deviation. Comparison between groups was performed using the two-tailed unpaired Student's t test or the Mann-Whitney test.
In vivo phenotype alteration of secondary tumors and coupling with primary tumors
We used the murine B16 melanoma tumor cell model, an aggressive, poorly immunogenic cell line, grafted s.c. into syngeneic C57BL6 mice. We used the F1 cell line, the nonmetastatic one, to avoid interference between spontaneous secondary tumor (metastasis) and implanted secondary tumor. CENU was administered intratumorally at days 11, 14 and 18.
CENU-treated primary tumors underwent proliferation arrest for 17 ± 3 days, after the end of treatment, then resumed growth and, in compliance with a Gompertz-like growth curve21 (Fig. 1a), reached a steady-state mass of 1.8 ± 0.33 vs. 6.1 ± 0.7 g (p < 0.001, treated vs. control tumors). Untreated secondary-challenged tumors implanted 32 days after the primary tumor did not show growth retardation but their mass leveled to 1.6 ± 0.4 g (p < 0.001 untreated secondary vs. control tumors) (Fig. 1a).
As shown in Figure 1b, histology of secondary tumors revealed necrotic areas, increased melanin content, decreased microvessel content and number of mitoses, anisocytosis and loss of aggressiveness. These features indicated that primary and secondary tumors acquired a differentiation phenotype.
Phenotype characterization was further performed by proton NMR spectroscopy tumor metabolite profiling. First, PUFAs expression was obviously increased in primary and secondary tumors in comparison with control tumors (p < 0.05) (Fig. 1c and 1d). Second, glutathione (GSx) level decreased (p < 0.05, primary or secondary tumors vs. control tumors) and PE tumor content increased (p < 0.01, primary or secondary tumors vs. control tumors) (Fig. 1c and 1d). Other small metabolites such as lactate (Lac) and alanine (Ala) mildly decreased in primary and secondary tumors, while glycine (Gly), lysine (Lys), taurine (Tau) and creatine (tCr) levels were not modified.
Primary and secondary tumors thus exhibited the same phenotype (Fig. 1a–1d), suggesting a coupling between tumors.
Primary and secondary tumor coupling is mediated by a tumor factor
To investigate the origin of tumor coupling, we tested whether the administration of serum from CENU-treated tumor-bearing donor mice could inhibit untreated melanoma tumor growth. Sera were obtained from donors that did or did not harbor melanoma tumor, and either had or had not been treated with CENU. Sera were administered intratumorally to tumor-bearing recipients on 3 consecutive days from day 11 of tumor evolution. Serum from either nontumor-bearing CENU-treated (Tum− CENU+) donors or tumor-free untreated (Tum− CENU−) donors moderately increased tumor growth in comparison to untreated tumor-bearing (Tum+ CENU−) donors. In contrast, serum from tumor-bearing CENU-treated (Tum+ CENU+) donors markedly impaired tumor growth in recipients, with tumor weights of 0.7 ± 0.2 vs. 5.2 ± 1.1 g at day 28 (p < 0. 001, Tum− CENU+vs. Tum+ CENU+) (Fig. 2a).
Histologic analyses of Tum+ CENU+ serum-treated tumors showed decreased vasculature, areas of necrosis, decreased number of mitoses and loss of aggressiveness (Fig. 2b). These features were consistent with a differentiation pattern of treated tumors.
Tumor metabolite profiling revealed significant accumulation of PUFAs (p < 0.05, Tum+ CENU−vs. Tum+ CENU+) (Fig. 2c and 2d). Besides PUFA increase, these tumors showed an increase in PE and a decrease in GSx level as shown in one-dimensional and two-dimensional proton NMR spectra (Fig. 2c). To evaluate the dose–effect relationship of serum, we administered serum on days 11, 14 and 18 after tumor cell inoculation. This protocol induced poor alterations of the phenotype. Tumor mass regression was less marked and tumor metabolite profile showed smaller PE and GSx variations than with the used protocol (data not shown).
Since treatment with Tum+ CENU+ serum induced growth inhibition and differentiation, i.e. phenotype changes mimicking those of secondary tumors, it may be concluded that CENU-treated melanoma tumors released bioactive molecules into the serum. These factors mediate a “bystander effect” on distant secondary tumors.
In vitro model for bystander factor investigation
To further explore the induction by CENU-treated tumor cell of factors with growth inhibitory and differentiation effects, we turned to in vitro experiments of medium transfer.
Long-term cultures of CENU-treated cells were first established. Briefly, a few days after CENU treatment, cells were characterized by growth inhibition, changes in cell morphology and cell cycle G2 phase accumulation. About 2 weeks after treatment, cells maintained a growth arrest that was accompanied by an accumulation of polyploid cells with large cytoplasm and an increase in melanin content. After 30 days, treated cells resumed growth with purely diploid cells and G1 phase accumulation (Fig. 3a and 3b). On the basis of cytological observations, these cells acquired a differentiation pattern. The metabolite profile of CENU-treated cells revealed an increase in PE content and a decrease in GSx level (Fig. 3c). CENU-treated cells were analyzed for ROS content. Oxidation was determined by measuring the DCF-DA fluorescence, and showed that CENU treatment increased ROS production (Fig. 3d).
Supernatants harvested from CENU-treated cell cultures and exposed to untreated cells provoked about 1:50 growth inhibition at day 10 (conditioned medium from treated vs. untreated cells) (Fig. 4a). Conditioned medium from the sequential phases of long-term CENU-treated cell cultures provoked similar cell growth inhibition. Thus, the conditioned medium used for the following experiments on cell phenotype changes was a mixture of supernatants collected over long-term cultures. As early as day 1 after exposure to a conditioned medium from CENU-treated cells, melanoma cells exhibited morphology changes with cell enlargement, increase of melanin content and cytokinesis retardation as attested by a number of daughter cells unseparated or linked together by a cytoplasmic bridge (Fig. 4b). The conditioned medium induced an accumulation in the G1 phase of the cell cycle: 75% ± 1% treated vs. 54% ± 1% control (p < 0.01) (Fig. 4c). As in CENU-treated cells and based on cytological observations, cells exposed to the conditioned medium have acquired differentiation features. In contrast with CENU treatment, no increase of DCF-DA fluorescence occurred in cells exposed to the conditioned medium. This point is discussed below.
The metabolite profile of primary and secondary cultures is presented in Figure 4d. One-dimensional proton NMR spectra showed a rapid and strong increase in PE cell content and a decrease in GSx level during the 10 days of exposure to the conditioned medium. These variations were as large as those found on CENU-treated cell cultures.
In conclusion, conditioned medium transfer experiments demonstrated the release by CENU-treated cells and their progeny of soluble factors that provoked growth inhibition, cytokinesis retardation and tumor cell metabolite alterations.
Effect of protein inhibitors
To investigate whether proteins released into the external medium were involved in the bystander effects, untreated cells and CENU-treated cells were incubated with cyclohexidine for 48 hr. After the treatment by cyclohexidine, cells were washed with PBS and maintained in standard medium during 7 days. The supernatants were collected. Then, naïve cell cultures were exposed to these conditioned media.
Cell protein content was increased by CENU treatment (+75% in CENU-treated cells vs. control cells). Cyclohexidine treatment reduced protein content by 17% (cyclohexidine-treated cells vs. control cells). In CENU–cyclohexidine-treated cells, protein content dropped by 53% in comparison with control cells (Fig 4e). In supernatants from either cyclohexidine-treated cells or CENU–cyclohexidine-treated cells, protein release was reduced by 23% and 41% in comparison with control cell supernatant or CENU-treated cell supernatant, respectively (Fig 4e). CENU-treated supernatant showed a weak increase in protein secretion in comparison with untreated supernatant. Naïve cells exposed to these supernatants for 7 days were examined for metabolite profile: PE and GSx levels varied similarly in cells exposed to the supernatant of CENU-treated cells additionally treated by the protein inhibitor or not. Furthermore, the heat-denatured supernatant from CENU-treated cells did not block the changes in PE and GSx. (Fig 4d)
Proteomic profile of CENU-treated cell culture supernatant and sera
Despite the questionable role of proteins in the observed bystander effects, we investigated the proteomic profile of conditioned media and sera to search for protein expression that may correlate with bystander factor expression. The proteomic study was performed using two-dimensional electrophoresis followed by MALDI-TOF mass spectrometry analysis. First supernatants were sampled over long-term follow-up of CENU-treated cell cultures (SCENU). In comparison with the supernatant of untreated cell cultures (SCTL), 69 modifications were observed, 19 proteins were significantly upregulated, 17 proteins were downregulated, 28 proteins were de novo expressed and 5 proteins were lost (Table I). Among the spots upregulated or de novo expressed, 7 spots had a highly significant expression in conditioned medium from CENU-treated cells, and had continuous expression during the 3 sequential phases of long-term cell cultures after CENU treatment. These spots were selected and analyzed by mass spectrometry. On the basis of a statistical score analysis, we identified the following proteins (Table II).
Table I. Proteomic Analysis of Supernatants
pI, isoelectric point; MW, molecular weight (kDa); Norm. vol., normalized volume; factor, ratio of normalized volumes (SCENU/SCTL); Student, Student's test;
Among the spots up-regulated or de novo expressed, the 7 spots with the strongest and more significant expression were retained for MALDI-TOF analysis.
Mice proteins (Mus musculus) in the MSDB.
MW, molecular weight (kDa); pI, theoretical isoelectric point of the identified molecule.
The score was equal to −10 log(P), P represented the probability that the identification proposed was the result of the hazard. In our conditions a score superior at 61 was significant at p < 0.05. The values given in parentheses indicates number of matched peptides/total number of peptides.
Hippocampal cholinergic neurostimulating peptide precursor protein(phosphatidylethanolamine-binding protein (PEBP))
First, a 20.8 kDa protein was identified as phosphatidylethanolamine-binding protein (PEBP), also known as Raf kinase inhibitory protein (RKIP). The function of this cytosolic and membrane protein is to bind phosphatidylethanolamine, a component of the cell membrane, and to inhibit the Raf-1 protein, thus activating the mitogen activating phospho-kinase (MAPK) signaling pathway.23 This bioactive protein may be involved in membrane remodeling and may be related to PE accumulation as observed in our models. Then, a small heat shock protein with 18.6 kDa, called cardiovascular Hsp (cHsp), was identified. This protein was shown to play a role of chaperone and to modulate actin polymerization.24 We also found a Rho-associated coiled-coil forming kinase 2 (ROCK) of 88.2 kDa. ROCK belongs to the Rho-Rac family, and is mainly involved in cytoskeleton regulation, cell motility and division.25, 26, 27 Then, we had 3 actin isoforms (43, 44 and 45 kDa). The release of actin filaments testify to cytoskeleton and cell coat remodeling. Lastly, one among the 7 proteins was not identified.
Proteomics were also performed on sera from Tum+ CENU− and Tum+ CENU+ donors. Among the observed spot variations, 2 proteins were de novo expressed in sera from Tum+ CENU+ donors and were absent in sera from Tum+CENU− donors. According to their MW and pI, these spots should correspond to PEPB and cHsp (Fig. 5c and 5d). These data confirmed the findings of the secretion spectrum of CENU-treated cells.
Few studies relate to anticancer drug-induced genotype or phenotype modifications of tumor cells escaping cell death. Our data provide evidence that a chemotherapeutic DNA-alkylating agent was able to induce in vivo and in vitro paracrine factors that caused growth inhibition, histopathological alterations consistent with differentiation and antiangiogenesis. Production and distal effect of these factors had analogies with the bystander effects reported in ionizing radiation studies.14, 15 A summary of our findings is given in Figure 6. We first discussed the phenotype changes associated with chemotherapy-induced bystander effects in vivo and in vitro; and second, the data obtained from proteomics of conditioned media of CENU-treated cell cultures.
Phenotype changes associated with chemotherapy-induced bystander effects
The phenotype mimetism between primary CENU-treated tumors and secondary untreated tumors speaks in favor of an autocrine-paracrine factor. The reason why secondary tumors modify their phenotype in response to the factors remains unknown, possibly to escape cell death or resist environmental stress. In this case, the factors would be cytoprotective; alternatively, the treated tumor might have secreted “suicide” cytotoxic factors.
The factor-related effect cannot be due to residual CENU molecule derivatives, since the half-life of CENU derivatives in vivo was about 30 min in mice,20 and since secondary tumors were implanted several weeks after the inoculation of primary tumors. As well, sera were taken 30 days after primary tumor evolution.
In our in vivo double tumor model, distant untreated tumors exhibited a differentiation pattern similar to that of CENU-treated tumors consisting in increased melanin content, decreased microvessel content and number of mitoses, anisocytosis and loss of aggressiveness. These tumors acquired their differentiation pattern under the pressure of the bystander factors. Metabolite profiling using NMR spectroscopy provided new information and putative markers of the bystander factors. In our model, a feature shared by challenged tumor cells both in vivo and in vitro was an increase in PE and a decrease in GSx levels. The observed large and rapid increase in PE would require the rapid inflation of a substrate breakable (phosphatidylethanolamine or sphingosine-phosphate) or synthesizable (ethanolamine) into PE. A rise in PE is a feature found in other types of tumor cells as they undergo therapy-escaped cell death and could be a mediator of cell survival.18, 28 In our experiments, the increased expression of PE correlates with the release of PEBP in the extracellular microenvironment. The presence of PEBP in the environment was shown to induce a loss of membrane asymmetry with cell surface exposure of membrane components such as phosphatidylethanolamine that are normally restricted to the inner leaflet of the lipid bilayer.23
A feature found exclusively in vivo was an increase in PUFA. PUFAs seem to be characteristic of programmed cell death in tumors.29 Some reports argue that damaged mitochondrial membranes are the source for PUFAs that inhibit superoxide dismutase and cause an increase of O2− in tumor cells leading to cell death.30 Similar observations were reported with ionizing radiation, anticancer agents and in therapy using suicide genes.29, 30 Others works indicate that the low PUFA content in hepatomas can be attributed to the high rate of proliferation and the poorly differentiated pattern, and reciprocally, the increase of PUFAs was related to the low rate of proliferation and differentiation pattern as in normal liver membranes.30 In addition, PUFAs have a protective role in tumor development by inhibiting tumor progression or favoring tumor regression. Thus, high PUFA expression, as induced in our models, might indicate a differentiation status.
Besides PE and PUFAs, GSx was another metabolic common denominator for several experiments. The formation of excessive amounts of ROS is toxic for the cell. Hence, metabolizing and scavenging systems to remove them are critical for cell survival. In treated tumor cells in vivo and in vitro, we found a decrease in GSx in comparison with control tumor cells. In tumors, the decrease in GSx content may be related to the loss of resistance to chemotherapy,31 or it may testify to the induction of stress-activated signal transduction pathway by alkylating agents.32 The depletion in GSx content also correlated with the loss of metastatic activity of B16 melanoma and cell arrest.33
Thus, PE and GSx alterations were a common denominator of a “controlled” tumor cell proliferation pattern resulting from direct CENU targeting or bystander factor impact, in vivo and in vitro. Their values as tumor response biomarkers for “controlled” tumor cell proliferation are under investigation. These changes can be considered as a phenotype change for 3 major reasons. First, these variations were maintained during the recovery phase; thus, daughter cells had gene expression inducing these modifications. Second, these tumors became indifferent to a new administration of CENU (data not shown). Third, secondary tumor regression was irreversible, since it persisted even when the primary tumor was obviously necrosed.
An antiangiogenic effect induced by the paracrine factors was suspected in vivo. In our study, secondary tumors showed poor neovasculature. Since these tumors were inoculated as paracrine mediators were circulating, the angiogeneic switch may have remained “off” from the onset of tumor growth.34 In contrast, in serum transfer experiments, tumors, once having developed neovasculature, were exposed to the factors. These tumors were more prone to thrombosis.34, 35
Chemotherapeutic DNA-alkylating agents are known to induce DNA damage and long-term genomic alterations in the progeny of treated cells.19, 36 In our CENU-treated cell cultures, we found the presence of diploid and polyploid subpopulations to be an indication of genomic instability.36 During this inhibition period, we showed that these cells produced ROS. In contrast, during the proliferation recovery, the progeny of the CENU-treated cells returned to a diploid status with an accumulation in the G1 phase of the cell cycle. Dying cells have been suggested to be a source of oxidative stress for the remaining cells and induced genomic instability.14 Genomic instability may explain the reprogramming of cell functions at the origin of the secretion of factors with bystander effects.
Proteins of the conditioned medium coexpressed with bystander factors: the issue of the bystander factor nature
To our knowledge, few studies have investigated the conditioned medium of chemotherapy-treated tumor cells. Because the global protein tests were in disfavor of factors of protein nature, we investigated proteomics of conditioned media to identify proteins coexpressed with the factors. In supernatant of CENU-treated cell cultures, we found a de novo release of PEBP, cHsp, ROCK and 3 isoforms of actin. Several of these proteins are involved in cytoskeletal rearrangement or cytokinesis.24, 25, 26, 27 Cellular PEBP overexpression was proposed as one of the factors responsible for resistance to cell death, and was shown to have an inhibitory effect on the MAPK signaling pathway and to interact with Raf-1 and MEK1.37 PEBP, also known as RKIP, is a modulator of the Raf/MAPK signaling cascade and a suppressor of metastatic cancer. It has been shown that PEBP inhibits MAPK by regulating Raf-1 activation.38 In this study, the release of PEBP may reflect a means for the cell to modulate the MAPK pathway. The blockage of MAPK may have the consequences of counteracting cytoskeletal rearrangement, affecting cell invasion, angiogenesis, cell differentiation39 and altering membrane lipid metabolism.23
Expression of small stress proteins (sHsp) have been shown to enhance survival of mammalian cells exposed to heat or oxidative injuries. The overexpression of sHsp(s) decreased the intracellular level of ROS. Also, sHsp expression abolished the burst of intracellular ROS induced by TNFα.40 Small Hsps have been shown to inhibit proliferation in vitro and to delay tumor development in human melanoma tumors in vivo.41 In conclusion, the presence of sHsp in the conditioned medium may explain why we did not find the ROS production in cells exposed to this medium. The modulatory effects of sHps on apoptosis are well documented. However, the mechanisms of Hsp-mediated protection against apoptosis have yet to be fully defined.40 PEBP and cHsp were also found in the serum of recipients bearing CENU-treated tumors. The presence of these 2 proteins in the serum validates in vitro studies to interpret in vivo data and presents PEBP and cHsp as proteins coexpressed with the factors.
ROCK and actin fragments were secreted in large amounts during the CENU-treated cell culture, and testified to cytoskeletal and Rho-Rac signaling alterations in the progeny of treated cells. Actin filaments constitute the cytoskeleton whose regulation is crucial for cell motility and growth. The relationship between inhibition of cytokinesis and formation of actin clots is unclear, but recent observations suggested that membrane lipids affect actin organization.27 In summary, the identified proteins might be involved in cytoplasm architecture disorders, in the decrease of cell proliferation, in metabolite alterations, in differentiation and in the inhibition of angiogenesis in vivo.
Actually, the direct role of the released proteins in the observed bystander effects was questionable, since the inactivation of proteins, in a qualitative and a quantitative manner, did not block the metabolite alterations.
However, factors with a protein nature that would be thermostable cannot be excluded. Also, bystander factors could be of another nature, such as small peptides or lipids possibly with oxidized or alkylated residues.
Further analyses are required to screen for nonprotein factors. In comparison, bystander factors due to ionizing radiation are well established but their molecular nature remains unclear. They may be soluble factors such as cytokines (IL-8, TGFβ) or oxidative metabolism products.42, 43 In the present study, features of the response to the CENU-induced bystander factors, not reported with ionizing radiation, involved differentiation, antiangiogenic effect and metabolic disorders consistent with lipid membrane remodeling. Thus, the nature of chemotherapy-induced bystander factors can differ from bystander factors in ionizing radiation studies.
In conclusion, this is the first demonstration that a chemotherapeutic DNA-alkylating agent may induce soluble factors with growth inhibitory bystander effects. These effects are found in vitro and in vivo long after tumor treatment and at locations remote from the treated tumor. The demonstrated bystander effects may account for delayed responses to treatment and may relay chemotherapy.
The authors thank Monique De Latour for histopathology analyses, Yves Communal for cytometry analyses, Gwenaelle Bestel-Corre for proteomic analysis and Janine Papon for technical assistance.