Preventive Cancer Stem Cell-Based Vaccination Reduces Liver Metastasis Development in a Rat Colon Carcinoma Syngeneic Model§

Authors


  • Author contributions: S.D. and D.M.: data acquisition (in vitro and in vivo) and analysis and interpretation; P.B. and V.C.: data acquisition (in vivo) and analysis and interpretation; A.L.: data acquisition (flow cytometry experiments) and analysis and interpretation; M.S. and J.-M.G.: data acquisition (mass spectrometry) and analysis and interpretation; P.S.: statistical analysis; M.-C.S.-P.: histologic analysis and interpretation; M.P.d.L. and G.F.C.: data analysis and interpretation and financial support; V.P.-C.: conception and design, data acquisition, analysis and interpretation, financial support, manuscript writing, and final approval of manuscript. S.D. and D.M. contributed equally to this article.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS November 29, 2012.

Abstract

Cancer stem cells (CSCs) represent a minor population of self-renewing cancer cells that fuel tumor growth. As CSCs are generally spared by conventional treatments, this population is likely to be responsible for relapses that are observed in most cancers. In this work, we analyzed the preventive efficiency of a CSC-based vaccine on the development of liver metastasis from colon cancer in a syngeneic rat model. We isolated a CSC-enriched population from the rat PROb colon carcinoma cell line on the basis of the expression of the aldehyde dehydrogenase-1 (ALDH1) marker. Comparative analysis of vaccines containing lysates of PROb or ALDHhigh cells by mass spectrometry identifies four proteins specifically expressed in the CSC subpopulation. The expression of two of them (heat shock protein 27-kDa and aldose reductase) is already known to be associated with treatment resistance and poor prognosis in colon cancer. Preventive intraperitoneal administration of vaccines was then performed before the intrahepatic injection of PROb cancer cells. While no significant difference in tumor occurrence was observed between control and PROb-vaccinated groups, 50% of the CSC-based vaccinated animals became resistant to tumor development. In addition, CSC-based vaccination induced a 99.5% reduction in tumor volume compared to the control group. To our knowledge, this study constitutes the first work analyzing the potential of a CSC-based vaccination to prevent liver metastasis development. Our data demonstrate that a CSC-based vaccine reduces efficiently both tumor volume and occurrence in a rat colon carcinoma syngeneic model. STEM CELLS2013;31:423–432

INTRODUCTION

Colon carcinoma represents one of the most frequent cancers in westernized countries and is the second most common cause of cancer-related mortality, due to the appearance of metastases preferentially located to the liver in 50%–60% of the patients [1]. Resection of liver metastases constitutes the only curative treatment and survival following surgery now approaches 35%–50% [2]. However, approximately 65% of patients will have a recurrence in 5 years [2].

It has recently been suggested that tumor persistence and recurrence could be due to the presence of a rare population of cells within the tumor mass with stem cell-like properties called “cancer stem cells” (CSCs). These cells exhibit extensive capability of self-renewal, are able to fuel tumor growth and are likely to be more resistant to therapy [3]. CSCs have recently been evidenced in solid tumors from various origins including colon [3–5] and appear as a new therapeutic target. The prerequisite and challenging step is to identify CSCs on the basis of the expression of specific markers and the ability of a very small number of cells to generate a tumor reconstituting the cellular heterogeneity of the original lesion. Cell surface antigens such as CD133 [4, 5] or CD44 and EpCAM [6] have been successfully used to identify colon cancer cells with CSC-like properties. Alternatively, colon CSCs can be isolated on the basis of the expression of the aldehyde dehydrogenase-1 (ALDH1) enzymatic activity, which has been recognized as a promising CSC marker, including in rat [7, 8].

Several strategies can be developed to target CSCs, including blocking their proliferation [9, 10] or inducing their differentiation [11, 12], thus enhancing sensitivity to conventional treatments. In this study, we planned to evaluate the benefit of vaccination to prevent the appearance of liver metastasis. Postulating that CSCs are at the origin of tumor development and could differentially express some antigens compared to their differentiated progeny, we analyzed the protective efficiency of a vaccine against CSCs in a rat syngeneic model of liver metastasis from colon carcinoma. Using ALDH1-based selection and serial in vivo transplantations, we isolated a CSC-enriched population from the PROb rat colon carcinoma cell line. Comparative analysis of vaccines originating from this CSC-enriched population or from the parental cell line by mass spectrometry allowed the identification of CSC-expressed proteins. In a rat syngeneic model, we further showed that preventive vaccination using a CSC lysate allowed the protection of half of the animals against the development of an experimental liver metastasis, while no significant difference in tumor occurrence was observed between control and PROb-vaccinated groups. In addition, CSC-based vaccination induced a 99.5% reduction in tumor volume compared to the control group. Taken together, these results suggest that CSC-based vaccination could be an efficient treatment to reduce tumor relapse in colon carcinoma.

MATERIALS AND METHODS

Cell Culture

DHD/K12/PROb (PROb) cells constitute a colon carcinoma cell line originating from a chemically induced colon cancer in BDIX rats [13]. These cells are poorly immunogenic and induce progressive and metastatic tumors in syngeneic hosts [14]. The cells were maintained in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Cergy Pontoise, France, http://fr-fr.invitrogen.com/site/fr/fr/home.html) supplemented with 10% fetal calf serum (FCS) (Dutscher, Brumath, France, http://www.dutscher.com).

Populations enriched in CSCs were cultured in ultra-low attachment flasks (Sigma-Aldrich, St Quentin-Fallavier, France, http://www.sigmaaldrich.com/france.html) in DMEM-F12 (Invitrogen) containing 6 mg/ml glucose (Sigma-Aldrich), 1 mg/ml NaHCO3 (Sigma-Aldrich), 5 mM HEPES (Sigma-Aldrich), 4 μg/ml heparin (Sigma-Aldrich), 4 mg/ml bovine serum albumin (BSA) (Sigma-Aldrich), 20 μg/ml insulin (Sigma-Aldrich), and N2 supplement (Invitrogen) and supplemented with 10 ng/ml basic fibroblast growth factor (bFGF) (Peprotech, Neuilly sur Seine, France, http://www.peprotech.com/en-US) and 20 ng/ml epidermal growth factor (EGF) (Peprotech). Sphere dissociation was performed upon incubation with Accumax (Sigma-Aldrich), a mix of mild proteases, at 37°C for 20 minutes.

ALDH1 Selection

The identification of ALDH1 activity from tumor cells was carried out using the Aldefluor assay (Stem Cell Technologies, Grenoble, France, http://www.stemcell.com) and fluorescence-activated cell sorting (FACS) (Aria cell sorter, BD Biosciences, Le Pont de Claix, France, http://www.bdbiosciences.com). Cells were resuspended in Aldefluor assay buffer and 10 μl of ALDH substrate was added to the cell suspension. Samples were then incubated for 40 minutes at 37°C. As a negative control, for each cell sample, an aliquot was treated with 10 μl of diethylaminobenzaldehyde (DEAB), a specific ALDH inhibitor. After the reaction, cells were centrifuged, resuspended in PBS buffer containing 0.5% BSA (Sigma-Aldrich) and 10% FCS, and maintained on ice. Flow cytometry experiments were performed to identify populations of cells with high (ALDHhigh) or low (ALDHlow) ALDH activity. The sorting gates were established to exclude dead cells and aggregates and the specific flow gates for ALDHhigh cells were set using the control sample in which ALDH activity was inhibited with DEAB.

Tumor Dissociation

Tumors were washed twice first in PBS containing 500 U/ml penicillin (Lonza, Verviers, France http://www.lonza.com) and 500 U/ml streptomycin (Lonza) and then in DMEM-F12 containing 500 U/ml penicillin, 500 U/ml streptomycin, and 1.25 μg/ml amphotericin B (Sigma-Aldrich) at 4°C. Specimens were then mechanically dissociated in a DMEM-F12 digestion medium containing 1.6 mg/ml collagenase type I (Invitrogen), 20 μg/ml hyaluronidase (Sigma-Aldrich), and 60 μg/ml DNase I (Roche Molecular Biochemicals, Mannheim, Deutschland http://www.roche-applied-science.com/index.jsp) for 30 minutes at 37°C under gentle agitation. A second similar digestion step with a 2× digestion medium was then performed before crushing the tumor pieces in a 100 μm Cell Stainer (Dutscher, Brumath, France) to obtain a cell suspension. The cell suspension was centrifuged at 1,000 rpm for 5 minutes and then filtered through a 40-μm nylon sieve and centrifuged again. The pellet was resuspended in supplemented DMEM/F12 medium.

Vaccine Preparation

Immediately after sorting, ALDH1high xv1 cells or PROb cells were seeded in culture in their respective media and 2-methoxyoestradiol (2-ME) (Sigma-Aldrich) was then added to each culture at a 0.1 mM final concentration and incubated for 15 hours. Cells were then centrifuged, resuspended in PBS at 5 × 106 cells/2 ml (first vaccination) or 4 × 106 cells/2 ml (second vaccination), and submitted to 10 cycles of freeze-thaw for lysate preparation. Mass spectrometry analysis was performed on these samples. Phosphorothioate-modified CpG oligodeoxynucleotide (Eurogentec, Seraing, Belgium, http://www.eurogentec.com/eu-home.html) (sequence: 5′-TCCATGACGTTCCTGACGTT-3′) was used as adjuvant and 50 μg was added in each vaccine dose containing 0.5 × 106 lysed cells/0.2 ml (first vaccination) or 0.4 × 106 lysed cells/0.2 ml (second vaccination).

Mass Spectrometry Analysis

One hundred micrograms of proteins from cell lysates was incubated with 5 μg of trypsin (Sequencing Grade Modified Trypsin, Promega Corporation, Madison, WI, http://www.promega.com) in 250 μl of ammonium bicarbonate (100 mM) at 37°C for 20 hours. Five micrograms of fresh trypsin was further added 4 hours before stopping the proteolysis with formic acid (5% final concentration). The mixture was centrifuged at 10,000 g for 15 minutes and the supernatant was collected and evaporated under vacuum (SpeedVac Concentrator SVC200H, Savant Instruments INC, Farmingdale, NY). Dried pellets were rehydrated in 20 μl of methanol/formicacid (20%:5%) and further processed for mass spectrometry analysis. All samples were analyzed by capillary-LC/ESI/MS/MS on an LTQ/FT-Orbitrap mass spectrometer (Thermo Fisher, Waltham, MA, http://www.thermofisher.com/global/en/home.asp) coupled with pumps and autosampler under standard conditions: capillary temperature, 275°C; and source voltage, 4,500 V. Helium was used as the collision gas. Experiments were done in parallel mode (survey at 30,000 resolution and five data-dependent ion trap MS/MS scans [Top 5]). The MS/MS parameters were as follows: isolation width, 3; collision energy, 35%; capillary-HPLC Surveyor system (Thermo Fisher); 120-minute gradient; and BioBasic C18 (Thermo Scientific, Courtaboeuf, France, http://www.thermo.com) column (100 0.32 mm). Such high mass accuracy on the precursor ion allowed the elimination of virtually any false-positive peptide identifications, suggesting that peptides that do not match the specificity of the protease used in the digestion should not automatically be considered as false positives. Acquired MS/MS spectra were interpreted using Mascot version 2.2.0 (Matrix Science, London, U.K., http://www.matrixscience.com) in-house software. Search parameters were set as follows: enzyme specificity, trypsin; one missed cleavage permitted; variable modification, methionine oxidation; mass tolerance for precursor ions, 6 ppm; mass tolerance for fragment ions, 0.6 Da; significance threshold, p < .05; and expect value threshold, 0.001 (to select ions with the highest individual score). Both b and y ion series were used to search against UniProt KB/Swiss-Prot/TrEMBL (database version 51.6; 257,964 sequence entries).

Animals and Tumor Model

For all the experiments, we used adult BDIX male rats weighing 180–250 g (Charles River, L'arbresle, France, http://www.criver.com/fr-CA/pages/home.aspx). All the surgical procedures and the care given to the animals were in accordance with institutional guidelines. All the animals were randomly assigned to treatment.

Induction of Subcutaneous Tumors

After anesthetization of the rat, 1,000 or 500 ALDH1high or ALDH1low cells (50 μl) mixed with 50 μl of matrigel (BD Biosciences) were injected under the skin using a 25-gauge needle.

Induction of Liver Tumors

After anesthetization and analgesia of the rat, the liver was surgically exposed and 1,000 or 250 ALDH1high or ALDH1low cells (100 μl) or 1.5 × 106 PROb cells were injected under the Glisson's capsule using a 27-gauge needle.

Vaccination

Each vaccine dose (200 μl) was administered by intraperitoneal injection using a 25-gauge needle.

Evaluation of Tumor Volume and Histologic Analysis

Tumors were removed, sliced, measured with calipers in the two perpendicular diameters (a and b). The tumor volume was calculated according to the formula ab2π/6, as previously described [15]. Hematoxylin and eosin staining was performed with Mayer hematoxylin and eosin solution (LaboNord, Templemars, France, http://www.labonord.com) on paraffin-embedded sections.

Statistical Analysis

The results are expressed as median (95% confidence intervals [CI]; range = min-max) and comparison of tumor volumes between vaccinated and control rats was performed using the Mann-Whitney U test, which is a nonparametric, two-tailed probability test. Qualitative analysis for the presence of a tumor was performed using the two-tailed Fisher's test. All statistics were computed with MINITAB Inc. V12.2 (State College, PA). Differences were considered statistically significant at p < .05.

RESULTS

Isolation of an ALDH1high Cell Population from the Rat PROb Colon Carcinoma Cell Line and Tumorigenicity Analysis in a Syngeneic Subcutaneous Tumor Model

Using the Aldefluor assay and FACS analysis, we isolated ALDH1high and ALDH1low cells from the PROb cell line. As shown in Figure 1A, the entry of the fluorescent Aldefluor substrate within cells makes them shift to the P6 gate. In the presence of the ALDH1 enzyme, the substrate cleavage increases its fluorescence, shifting the cells to the P5 gate. This is not observed in the presence of the ALDH1 inhibitor DEAB. As expected, the majority of PROb cells had low ALDH activity with 3% average (three independent experiments) of ALDH1high cells within the population. As shown in Figure 1B, the ALDH1high cells were able to grow as spheres under serum-free medium with bFGF and EGF. ALDH1low cells were also able to generate spheres under these conditions, although those spheres were substantially smaller and less numerous (data not shown).

Figure 1.

Isolation of an ALDH1high cell population from the PROb colon carcinoma cell line and tumorigenicity analysis. (A): Selection of ALDH1high cells from the PROb cell line using the Aldefluor assay. The flow cytometry profiles of cells treated with ALDH1 substrate in the presence or in the absence of DEAB are presented. The P5 gate contains the ALDH1high cells. (B): Sphere formation of the ALDH1high cells cultured in serum-free medium containing basic fibroblast growth factor and epidermal growth factor. (C): Tumorigenic potential of ALDH1high cells. Representative photographs of subcutaneous tumors generated in 30 days by the injection of 500 or 1,000 ALDH1high cells in the presence of matrigel. (D): Histological analysis of tumors generated by PROb or ALDH1high cell injection. The morphology of the tumor generated by ALDH1high cell injection closely resembles that of the original tumor, as judged by hematoxylin and eosin staining of paraffin-embedded sections. Abbreviations: ALDH, aldehyde dehydrogenase; DEAB, diethylaminobenzaldehyde.

To evaluate the tumorigenicity of ALDH1high cells, 500 or 1,000 of these cells were injected subcutaneously with matrigel in syngeneic BDIX rats immediately after sorting. Although 1.5 × 106 PROb cells are generally required to generate a tumor in 30 days [16], 500 and 1,000 ALDH1high cells were able to give rise to a tumor in the same period (Fig. 1C). In addition, histologic analysis indicates that the lesion induced by ALDH1high cells is a phenocopy of the parental tumor (Fig. 1D). Surprisingly, the frequency of tumor formation after ALDH1low cell implantation was not significantly different from that obtained with ALDH1high cells in two independent experiments (data not shown).

Isolation of a CSC-Enriched Population from the ALDH1high-Originating Tumors and Tumorigenicity Analysis in a Syngeneic Intrahepatic Tumor Model

Since the ability to generate a tumor that can be serially transplanted constitutes one of the criteria that define a CSC population, we next dissociated ALDH1high-originating tumors and cultured the cell suspension in serum-free medium with bFGF and EGF (Fig. 2A). At day 1, the culture contained nontumoral cells that progressively disappeared (day 20) and tumor cells, which we called xv1 (for ex vivo passage 1), were able to exponentially grow as spheres (day 50). ALDH1high and ALDH1low cells were then isolated from the xv1 culture. As shown in Figure 2B, we observed a 10-fold increase in the percentage of ALDH1high cells in this culture (30% average from three independent experiments) compared to the one obtained for the PROb cell line (3% average from three independent experiments). To evaluate the tumorigenicity of ALDH1high and ALDH1low xv1 cells, 250 or 1,000 cells of each of these populations were injected within the left and right liver lobes of syngeneic rats immediately after sorting. As liver is the natural developing site for these cells, injections were performed in the absence of matrigel to be closer to physiological conditions. In this case, tumor development was substantially longer and the animals were sacrificed as soon as the first clinical sign appeared. For the rats injected with 250 ALDH1high xv1 cells, these clinical signs appeared after 4.5 months. As shown in Figure 2C, these animals exhibited an extended peritoneal carcinomatosis, associated with mesenteric tumor nodules and diffuse liver metastasis. In contrast, the corresponding control rats, injected with 250 ALDH1low xv1 cells and sacrificed at the same time, did not present any signs of tumor development. For the rats injected with 1,000 ALDH1high or ALDH1low xv1 cells, similar results were obtained 3 months after tumor cell injection (Fig. 2D). Taken together, these results suggest that ALDH1high xv1 cell population is enriched in CSCs.

Figure 2.

Isolation of a cancer stem cell-enriched population from the ALDH1high-originating tumors and tumorigenicity analysis. (A): Culture in suspension and in serum-free medium containing basic fibroblast growth factor and epidermal growth factor of cells obtained after dissociation of ALDH1high-originating subcutaneous tumors. (B): Selection of ALDH1high xv1 cells using the Aldefluor assay. The flow cytometry profiles of cells treated with ALDH1 substrate in the presence or in the absence of diethylaminobenzaldehyde are presented. The P5 gate contains the ALDH1high cells. (C, D): Tumorigenic potential of ALDH1high xv1 cells. Representative photographs of tumors generated by the intrahepatic injection of 250 (C) or 1,000 (D) ALDH1high or ALDH1low cells in the absence of matrigel. Abbreviation: ALDH, aldehyde dehydrogenase.

Preparation and Analysis of PROb or ALDH1high xv1 Vaccines

Whole tumor cell vaccination requires a method to kill and, at the same time, to enhance the immunogenicity of tumor cells [17]. PROb or ALDH1high xv1 vaccines were prepared with freeze-thaw cell lysates. Repeated freeze-thaw cycles constitute a widely used method to prepare necrotic whole tumor cell lysate as an immunogenic source of tumor-associated antigens [17, 18]. However, it was demonstrated that cell lysates can inhibit the maturation and function of dendritic cells, although this suppression is partially reversed in vitro by induction of tumor cell stress before lysis [18]. Therefore, prior to cell lysis, cells were submitted to an oxidative stress upon treatment with the superoxide dismutase inhibitor 2-ME. Some cells appeared to be dead but the majority of them were still alive before lysis (data not shown). We then analyzed the content of PROb and ALDH1high xv1 cell lysates by mass spectrometry. As shown in Figure 3A, the chromatography profile of each lysate is different, suggesting that CSCs could express specific antigens. Among the 142 identified proteins, 123 were detected both in ALDH1high xv1 and in PROb lysate, 14 were observed only in PROb lysates, and four only in ALDH1high xv1 lysates. The proteins detected only in ALDH1high xv1 lysate or only in PROb lysate are presented in Tables 1 and 2, respectively.

Figure 3.

Chromatography profile of the cell lysates. (A): ALDH1high xv1 cell lysate. (B): PROb cell lysate. Abbreviation: ALDH, aldehyde dehydrogenase.

Table 1. Proteins detected in the aldehyde dehydrogenase-1high xv1 cell lysate and absent in the PROb cell lysate
inline image
Table 2. Proteins detected in the PROb cell lysate and absent in the aldehyde dehydrogenase-1high xv1 cell lysate
inline image

Comparison of the Protective Efficiency of PROb or ALDH1high Xv1 Vaccines

To determine whether vaccination with ALDH1high xv1 cells can elicit a superior protection than vaccination with PROb cells against the development of a PROb tumor, we designed the protocol presented in Figure 4A. In the control PBS-vaccinated group, all rats developed a tumor (10 out of 10) and there was no significant difference in tumor occurrence compared to the PROb-vaccinated group (eight out of 10) (p = .47). Importantly, half of the animals did not exhibit a tumor (five out of 10) in ALDH1high xv1-vaccinated group (p = .03). Regarding tumor volumes, both vaccinations induced a significant decrease with a median of 25.9 mm3 (95% CI: 0.33–50.43; range = 0–207.0 mm3) in PROb-vaccinated rats (p = .006) and 0.8 mm3 (95% CI: 0–85.3; range = 0–117.7 mm3) (p = .0015) in ALDH1high xv1-vaccinated rats, compared to 169.5 mm3 (95% CI: 104.6–265.1; range = 10.5–423.6 mm3) in the control group (Fig. 4C). No statistically significant difference was observed in the tumor volumes between PROb and ALDH1high xv1-vaccinated groups (p = .61).

Figure 4.

Comparison of PROb and ALDH1high xv1 cell-based vaccines for prevention of experimental liver metastasis development. (A): Protocol. At days 0 and 7, the rats received an intraperitoneal injection of PBS (n = 10), or PROb vaccine (n = 10), or ALDH1high xv1 vaccine (n = 10). At day 14, 1.5 × 106 PROb colon cancer cells were injected in the left lobe of the liver. Thirty days later, the rats were sacrificed and volume left lobe tumor was measured. (B): Representative photographs of the liver in one animal of each group. Tumors are delimited by a dashed line. (C): Median tumor volume in each group of rats. The 95% confidence intervals are indicated. *, p = .006; **, p = .0015. Abbreviations: ALDH, aldehyde dehydrogenase; CSC, cancer stem cell; PBS, phosphate-buffered saline.

DISCUSSION

CSC resistance to conventional therapies may explain why tumor recurrence is often inevitable. The development of new therapeutic strategies based on CSCs represents now a major challenge. Although most studies have been performed with human tumor-derived CSCs analyzed in immunocompromised mice, these xenotransplantation assays have been shown to result in a great variability in the frequency of cells with tumorigenic potential, depending on the degree of host immunocompetence, which constitutes a technical limitation [19]. Therefore, in this work, we assessed the effect of CSC-based vaccination in the prevention of tumor development and recurrence in a syngeneic rat model. To this aim, we first isolated a CSC-enriched population from the rat colon cancer PROb cell line based on the expression of the ALDH1 marker. The isolated cells, representing 3% of the whole population, were able to grow as spheres in serum-free medium and to generate subcutaneous tumors following injection of 500 or 1,000 cells in the presence of matrigel. ALDH1low cells were also able to generate tumors in these conditions. Tumor formation from ALDH1low cell injection was previously observed, especially for high doses of transplanted CSCs [20–22]. This is also the case for another widely used CSC marker, CD133, for which several studies have shown that CD133 cells are also able to give rise to tumors in immunodeficient animals [23–29]. In this regard, Kim et al. suggested a potential contamination with more tumorigenic cells during high-volume cell sorting [21]. Additionally, the use of matrigel, which may provide tumor cells with additional survival factors generally absent from their normal environment, is known to facilitate tumorigenesis [3, 30]. It is also reasonable to assume that, in some conditions, and probably when the selection marker is weakly expressed and the dose of injected tumor cells is high, marker-negative cells are able to develop in favorable environment, that is, in the skin in the presence of matrigel. In our case, working in a syngeneic system could further increase the permissivity for tumor development.

After tumor dissociation and culture in serum-free conditions (xv1 cells), the percentage of ALDH1high cells was significantly increased as previously demonstrated for the CD133 marker in colon CSC cells derived from human tumors [5]. To more closely mimic physiological conditions, these ALDH1high xv1 cells were then injected in the absence of matrigel in the liver which may represent their natural environment during metastasis formation. As few as 250 or 1,000 cells were able to generate highly invasive tumors invading peritoneum and mesentere, while no tumor could be observed with ALDH1low cells. It is interesting to note that, in these conditions, tumor development is delayed compared to subcutaneous injection of the cells in the presence of matrigel, indicating that tumor growth from CSC is particularly slow in their physiologic environment. However, current hypotheses suggest that metastatic lesions originate from disseminated tumor stem cells that, after undergoing a period of dormancy, subsequently resume growth [31, 32]. This dormancy period could be related with the existence of a nonpermissive CSC niche [33]. In this context, our model could represent an adequate system to study the transition between a minimal residual disease state to a metastatic disease. Altogether, these data suggest that ALDH1high xv1 cells represent a CSC-enriched population that can be used as a vaccine source.

The way in which tumor cells die is critical for the immune antitumoral response. While freeze-thaw tumor cell lysates were shown to inhibit dendritic cell maturation and function, this lysate-induced dendritic cell suppression can be partially reverted in vitro by induction of tumor cell stress before lysis [18]. Similar to normal stem cells, CSCs are supposed to have a high antioxidant capacity to keep cellular reactive oxygen species at a moderate level and maintain both stemness and cancer-forming capabilities [34, 35]. Therefore, before lysis, tumor cell stress was induced using 2-ME that inhibits superoxide dismutase and leads to superoxide accumulation [35]. The content of each lysate was then analyzed by mass spectrometry. Among the four proteins detected only in ALDH1high xv1 lysate, we found the heat shock protein 27-kDa (HSP27), which has been described to be overexpressed in several types of human cancers and associated with treatment resistance and poor prognosis [36]. More recently, HSP27 was shown to be involved in self-renewal and maintenance of breast CSCs. Indeed, it was demonstrated in breast cancer that knockdown of HSP27 decreased CSCs characters, such as ALDH-positive population percentage, mammosphere formation ability, migration, and in vivo CSC frequency [37]. In colorectal cancer (CRC), HSP27 expression is correlated in vitro with 5-FU [38, 39] and oxaliplatin [40] resistance and HSP27 downregulation restores 5-FU sensitivity [39]. HSP27 expression is significantly increased in human primary CRCs [41, 42] and there is a significant correlation between HSP27 expression and TNM stage [42]. In addition, HSP27 overexpression is associated with poorer overall survival and adverse outcomes in CRC patients [41, 42]. These findings suggest that HSP27 could be a colon CSC marker as already described for lung CSCs [43]. Aldose reductase, a member of aldo-keto reductase superfamily of proteins, is a mediator of inflammatory signals induced by various stimuli [44]. Overexpression of aldose reductase was shown to be involved in drug resistance in different tumor cells such as colon [45, 46], liver [47], and medulloblastoma [48]. In colon cancer, aldose reductase was shown to regulate cancer cell adhesion, invasion, and migration events which initiate metastasis [49] and to mediate hypoxic signals, leading to tumor progression and invasion [50].

In addition, HSP27 and aldose reductase are already known as antigens eliciting a humoral immune response in several cancers. Indeed, antibodies against HSP27 were identified in the sera from breast cancer patients [51] and in the genital tracts of patients exhibiting gynecologic cancers [52]. Regarding aldose reductase, this protein was previously demonstrated to be immunogenic and has been identified as a serologically defined tumor-associated antigen in renal [53] and gingivo-buccal [54] cancers as well as in cutaneous T-cell lymphoma [55] and melanoma [56].

Among the 14 proteins detected only in PROb lysates, two annexin proteins were identified. The first one, annexin A3, is known to be upregulated in premalignant adenoma lesions compared to normal mucosa [57]. The second one, Annexin A6 as well as clathrin, was described to correlate with sensitivity to 5-fluoro-2′-deoxyuridine-5′-monophosphate, a 5-fluorouracil metabolite [58]. T complex protein 1 is a molecular chaperone assisting in the folding of several cytosolic proteins which has been shown to be upregulated in colon carcinoma compared to normal tissue [59]. Heat shock 70 kDa protein 4, also known as ischemia-responsive protein-94 (Irp94), is a ligand for natural killer cell receptor protein-2 on dendritic cells, which in turn plays a role in both innate and adaptive immunity [60]. DJ-1 or PARK7 is known as a tumor biomarker in several kinds of cancer [61] and has been shown to be increased in CRC compared to normal mucosa [62].

This study represents a proof of principle of using CSCs as a source of immunogenic antigens to trigger an antitumor immune response, strengthening the recent work of Ning et al. on melanoma and squamous cell carcinoma [63]. Although we did not observe any macroscopic abnormalities in treated animals, we cannot exclude the occurrence of unexpected side effects.

Regarding the protective effect of vaccination, our results show that the ALDH1high xv1 vaccine rendered 50% of the animals resistant to tumor development while no significant difference in tumor occurrence was observed between control and PROb-vaccinated groups. In addition, ALDH1high xv1 vaccination induced a 99.5% reduction in tumor volume, compared to a 85% decrease in the PROb-vaccinated group. Taken together, these results suggest that CSC-based vaccination could elicit a superior protection than whole tumor cell vaccine.

CONCLUSIONS

Patients exhibiting an advanced-stage colon cancer without liver metastasis are known to have a high-risk to later develop liver metastases. Similarly, in patients exhibiting liver metastasis treated by surgery, 65% will have a tumor recurrence. All these patients could be eligible for an adjuvant treatment inhibiting tumor development. In this context, our data suggest that CSC-based preventive vaccination could be a potential attractive adjuvant treatment to decrease tumor relapses.

Acknowledgements

This study was supported by the CNRS, Fondation de l'Avenir (n° ET8-487) and the PESSOA-EGIDE Program (2010–2011), the Portuguese Foundation for Science and Technology and France Cancer. S.D. is recipient of a fellowship from the Portuguese Foundation for Science and Technology (SFRH/BD/39727/2007).

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors indicate no potential conflicts of interest.

Ancillary