HSP DNAJB8 Controls Tumor-Initiating Ability in Renal Cancer Stem–like Cells Free

Renal cell carcinoma (RCC) is one of the most resistant forms of cancers to both radiotherapy and chemotherapy. In recent years, molecular targeted therapies have been developed and have shown significant objective responses (1–3), and they have been incorporated into current standard therapies of metastatic RCC; however, these molecular targeted therapies have not provided durable responses. RCC is regarded as an immunogenic malignancy and has well-documented responses to some cytokines such as interleukin-2 (IL-2) and IFN-α, and some patients have shown significant responses to treatments with these cytokines (4–6). However, the results have been limited by such nonspecific immunotherapy; therefore, cancer-specific immunotherapy may become a new modality for patients with metastatic RCC. Tumor-associated antigens (TAA) that can be recognized by CTLs have been investigated (7, 8), and some of these antigens induced objective tumor regression in patients with malignant melanoma (9, 10). In advanced cases, however, complete elimination of tumor cells by limited numbers of effector CTLs is difficult because a solid tumor contains an enormous number of tumor cells (about 5 × 10⁸ cells per gram; ref. 11).

Cancer cells in solid carcinomas display heterogeneity in many aspects of their phenotypes, and only a small population of cells, called cancer stem–like cells/cancer-initiating cells (CSC/CIC), express stem cell phenotype and have high tumor-initiating ability (cancer stem cell hypothesis; refs. 12–14). CSCs/CICs are resistant to chemotherapy and radiotherapy by various mechanisms, and these characteristics of CSCs/CICs are thought to be related to posttherapeutic recurrence (15). Thus, therapy targeting the small population of CSCs/CICs might be a reasonable approach for treatment of resistant and advanced cancers. Some immunologic effector cells, including natural killer (NK) cells and γδT cells, have been reported to be able to efficiently recognize the CSC/CIC population (14). Furthermore, we have reported that CTLs can efficiently recognize human colon CSCs/CICs (16). We have categorized TAAs that can be recognized by CTLs into 3 groups: (i) CSC antigens, which are expressed in CSCs/CICs but not in non-CSCs/CICs (e.g., MAGE-A3 and MAGE-A4); (ii) shared antigens, which are expressed in both CSCs/CICs and non-CSCs/CICs (e.g., CEP55, SURVIVIN); and (iii) non-CSC antigens, which are expressed in non-CSCs/CICs but not in CSCs/CICs (e.g., AMACR, HIFPH3; ref. 17).

In this study, we found that a HSP40 family protein, DnaJ (Hsp40) homolog, subfamily B, member 8 (DNAJB8), is expressed preferentially in the CSC/CIC population cancer cells, including RCC and the testis among normal tissues, indicating that DNAJB8 is a novel cancer-testis (CT) antigen and also a novel CSC antigen. DNAJB8 has a role in the maintenance of RCC CSCs/CICs. We compared the potency of DNAJB8 as an immunologic target with that of Survivin, which is expressed in both CSC/CIC and non-CSC/CIC populations (shared antigen), and we found that DNAJB8 is more effective than Survivin. Our results suggest that targeting a CSC antigen is more effective than targeting a shared antigen and that DNAJB8 is a candidate for CSC/CIC-targeting immunotherapy.

RCC cell lines ACHN, Caki-1, SMKTR2, and SMKTR3 and the murine RCC cell line RenCa of BALB/c mouse origin were maintained in RPMI1640 (Sigma) supplemented with 10% FBS. HEK293T cells and the murine fibroblast cell line BALB/3T3 of BALB/c mouse origin were grown in Dulbecco's Modified Eagle's Medium (DMEM; Sigma) supplemented with 10% FBS. PLAT-E and PLAT-A cells (kind gifts from Dr. T. Kitamura, Division of Stem Cell Signaling, The Institute of Medical Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108-8639, Japan) were grown in DMEM supplemented with 10% FBS, 10 μg/mL blasticidin, and 1 μg/mL puromycin.

All mouse procedures were carried out in accordance with institutional protocol guidelines at Sapporo Medical University School of Medicine. BALB/c female mice were purchased from Clea Japan and nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice were purchased from Charles River Laboratory Japan at the age of 6 to 8 weeks.

Reverse transcriptase PCR (RT-PCR) analysis was carried out as described previously (18). Human Multiple Tissue cDNA Panels I and II (Clontech) were used as templates of normal adult tissue cDNAs. Except for SOX2 and Sox2, PCR amplification was done in 20 μL of PCR mixture containing 0.25 μL of the cDNA mixture, 0.1 μL of Taq DNA polymerase (Qiagen), and 12 pmol of primers. The PCR mixture was initially incubated at 94°C for 2 minutes, followed by 35 cycles of denaturation at 94°C for 15 seconds, annealing at 58°C for 30 seconds, and extension at 72°C for 30 seconds. PCR amplification of SOX2 and Sox2 was carried out in 20 μL of PCR mixture containing 0.4 μL of the cDNA mixture, 0.2 μL of PrimeSTAR HS DNA polymerase (Takara), and 12 pmol of primers. The PCR mixture was initially incubated at 94°C for 2 minutes, followed by 35 cycles of denaturation at 94°C for 15 seconds and annealing and extension at 68°C for 30 seconds. Primers used in experiments are summarized in Supplementary Table.

A monoclonal antibody (mAb) against DNAJB8 (clone #EMR-DNAJB8.214-8) was generated, as described previously (19), by immunizing mice 4 times weekly with recombinant His-tag DNAJB8 protein produced and purified by a Ni-NTA agarose column (Qiagen). Cell lysate with SDS sample buffer was separated by denaturing SDS-PAGE. Separated proteins were transferred onto nitrocellulose membranes and probed with mouse anti-FLAG M2 antibody (Sigma) or anti-DNAJB8 antibody. β-Actin was used as a loading control and was detected with a mouse mAb (Sigma). Anti-DNAJB8 antibody was used at 200 times dilution. Anti-FLAG antibody and anti–β-actin antibodies were used at 2,000 times dilution.

Side population (SP) cells were isolated as described previously using Hoechst 33342 dye (Lonza) with some modifications (20). Briefly, cells were resuspended at 1 × 10⁶/mL in prewarmed DMEM supplemented with 5% FBS. Hoechst 33342 dye was added at a final concentration of 2.5 μg/mL in the presence or absence of verapamil (50 μmol/L; Sigma-Aldrich), and the cells were incubated at 37°C for 90 minutes with intermittent shaking. Analyses and sorting were carried out with an FACSAria II cell sorter (Becton Dickinson).

Transduction of genes into cells was carried out by a retrovirus-mediated method as described previously (21). PLAT-A and PLAT-E cells, which are amphotropic and ecotropic packaging cells, respectively, were transiently transduced with a pMXs-puro (kind gift from Dr. T. Kitamura, Tokyo, Japan) retroviral vector expressing FLAG-tagged DNAJB8, Dnajb8, and Survivin, V5-tagged DNAJB8 ΔSSF-SST mutant, and a control plasmid using FuGENE HD transfection reagent (Roche) following the manufacturer's protocol. Retroviral supernatants were harvested 48 hours after transfection. The supernatant was used for infection of BALB/3T3 cells, ACHN cells, or RenCa cells in the presence of 8 μg/mL of polybrene (Sigma) overnight. For the generation of stable transformants, the infected cells were selected with 1 μg/mL puromycin. DNAJB8, Dnajb8, or Survivin expression was confirmed by Western blot analysis.

DNAJB8 siRNAs (HSS136480, HSS136482, and HSS176060) were purchased from Invitrogen and transfected using Lipofectamine RNAi MAX reagent (Invitrogen) according to the protocol of the manufacturer. Cells were transfected with siRNA 72 hours before analysis. Nontargeting siRNA (Stealth RNAi Negative Control; Invitrogen) was used as a negative control. DNAJB8 knockdown was confirmed by Western blot analysis.

Plasmids including the target site for mutation as a template and primers that contain the desired mutation encoding the same amino acids as those of the wild-type sequence were used for site-directed mutagenesis. The sequences of primers are listed in Supplementary Table. Mutated plasmids were generated by 20 cycles of PCR amplification with KOD plus DNA polymerase (TOYOBO) and then digested by restriction enzyme DpnI for 2 hours. The generated nicked dsDNA was transformed into BL21 strain of Escherichia coli and amplified. The mutations were verified by DNA sequencing.

Sorted SP and non-SP cells, and gene-transfected ACHN cells were injected into the subcutaneous space of the back of syngeneic BALB/c mice (for RenCa cells) or NOD/SCID mice (for ACHN cells). Tumor growth was monitored weekly, and tumor volume was calculated by XY²/2 (X = long axis, Y = short axis).

Murine Dnajb8 and Survivin cDNAs were amplified by PCR from RenCa cells and subcloned into BamHI and XhoI cloning sites of the modified pcDNA3.1(+) vector (Invitrogen), which contains the murine immunoglobulin kappa chain signal sequence at the N terminus [inserted into NheI and HindIII cloning sites of pcDNA3.1(+)] and a FLAG epitope at the C terminus [inserted into XbaI and PmeI cloning sites of pcDNA3.1(+)] of the expression antigen. A blank vector, modified pcDNA3.1(+), was used as control. The sense sequences and antisense sequences of the inserted oligo DNA are listed in Supplementary Table.

One week after the DNA vaccination, mice were sacrificed and a single-cell suspension was prepared from the spleen. Erythrocytes were lysed by osmotic shock (0.1 mol/L NH₄Cl), and CD8⁺ T cells were isolated from splenocytes by using a MACS separation system (Miltenyi Biotech). A total of 5 × 10⁶ CD8⁺ or CD8⁻ cells per well were, respectively, seeded in a 24-well plate with RPMI-1640 medium (Sigma) supplemented with 10% FBS. Phytohemagglutinin (PHA) blasts were generated from CD8⁻ cells by using 50 IU/mL IL-2 and 2 μg/mL PHA for 3 days, and then Dnajb8 or Survivin genes were transduced with a retrovirus vector. On day 7, 5 × 10⁵Dnajb8- or Survivin–expressing PHA blasts were irradiated (100 Gy), washed once, and added onto CD8⁺ cells. On day 8, IL-2 was added to each well at a concentration of 50 IU/mL. On day 14, the antigen-specific IFN-γ release of CD8⁺ T cells was assessed by an ELISPOT assay as described previously (19).

To measure protective immunity, BALB/c female mice were weekly immunized with DNA plasmid 4 times by injection into their footpad followed by injection of 1 × 10⁵ RenCa tumor cells. Tumor growth was monitored weekly. RenCa/Dnajb8 cells (1 × 10⁶) were injected subcutaneously into the back of BALB/c female mice that had rejected a previous injection of 1 × 10⁵ RenCa cells.

Mice were injected intraperitoneally with 500 μg of either anti-CD4 (clone: GK 1.5, rat IgG) or anti-CD8 (clone: 2.43, rat IgG) mAb 2 days before immunization and then immunized with cDNA plasmid vaccines once weekly for 4 weeks. Tumor cells were challenged 1 week after the fourth immunization. Depletion of CD4⁺ and CD8⁺ T cells was verified by flow cytometry.

Data are presented as means ± SD. Differences in variables were assessed using Student t test. Survival curves were constructed according to the Kaplan–Meier method. Statistical significance was determined by the log-rank test. P < 0.05 was considered significant. Statistical analysis was done with Stat Mate III (ATMS Co., Ltd.).

DNAJB8 has been reported to inhibit the aggregation of missfolding proteins and to protect cells from cell death (22); however, its role in cancer cells is not clear. We first investigated DNAJB8 mRNA distributions in human normal adult tissues and cancer cells by RT-PCR. DNAJB8 mRNA is barely expressed in normal mature tissues except for the testis (Fig. 1A), and it is expressed in all cells of human cancer lines (Fig. 1B).

We generated a DNAJB8-specific mAb (clone: #EMR-DNAJB8.214-8) to investigate DNAJB8 protein expression. The mAb #EMR-DNAJB8.214-8 showed DNAJB8-specific reaction using FLAG-tagged DNAJB8-overexpressed ACHN cells by Western blotting (Fig. 1C). Interestingly, the mAb #EMR-DNAJB8.214-8 was also reactive for FALG-tagged mouse Dnajb8-transduced RenCa cells (Fig. 1C). Therefore, the mAb #EMR-DNAJB8.214-8 is reactive for both human DNAJB8 and mice DNAJB8 proteins. DNAJB8 proteins were detectable in human sperm and mice testis tissues (Fig. 1D). Immunohistochemical staining also confirmed protein expression in human testis and mouse testis. Matured sperm and spermatid were positive for #EMR-DNAJB8.214-8, whereas spermatogonia and spermatocytes were negative for #EMR-DNAJB8.214-8 (Fig. 1E). DNAJB8 protein was also detectable in human RCC (clear cell carcinoma) tissues, whereas it was undetectable in counterpart human renal tubules (Fig. 1F). These expression profiles strongly indicated that DNAJB8 is a novel cancer-testis antigen (23).

To address DNAJB8 functions in RCC CSCs/CICs, we carried out SP analysis using the human RCC cell lines ACHN, Caki1, SMKTR2, and SMKTR3 and the mouse RCC cell line RenCa. SP cells were detectable in ACHN, Caki1, and RenCa cells, and those SP cells were completely inhibited by verapamil (Fig. 2A and Supplementary Fig. S1), indicating that SP cells were specific for ABC transporter activity. The ratios of SP cells in ACHN, Caki1, and RenCa cells were 2.6%, 0.2%, and 18%, respectively. The ratios of SP cells in Caki1, SMKTR2, and SMKTR3 cells were too low for further analysis, and we therefore used ACHN and RenCa in the following experiments. It has been shown that SP cells were not enriched with CSCs/CICs in some types of cancer cells (24), and thus it is essential to validate that SP cells are enriched with CSCs/CICs. ACHN SP cells initiated tumor formation with 10³ cells, whereas ACHN main population (MP) cells needed 10⁴ cells to initiate tumor formation (Fig. 2B and Table 1). RenCa SP cells initiated tumor formation with 10² cells, whereas RenCa MP cells needed 10³ cells to initiate tumor formation (Fig. 2B and Table 1). These data indicated that SP cells derived from ACHN and RenCa cells were enriched with CSCs/CICs.

We carried out RT-PCR analysis to evaluate genetic characteristics of SP and MP cells derived from ACHN and RenCa cells (Fig. 2C). SP cells derived from ACHN and RenCa cells showed higher expression levels of SOX2/Sox2 and POU5F1/Pou5f1 than did MP cells, indicating that they present stem cell–like phenotypes (Fig. 2C). Recently, induction of epithelial to mesenchymal transition (EMT) in transformed mammary epithelial cells has been shown to result in the creation of populations of cells that are highly enriched in CSCs/CICs (25); therefore, we analyzed EMT-related gene expression in SP and MP cells. Both ACHN and RenCa SP cells showed repressed CDH1/Cdh1 expression. ACHN SP cells preferentially expressed SNAI2 and RenCa SP cells preferentially expressed Snai1 and Twist1, representative EMT-inducing transcription factors. These findings suggested that SP cells of RCC are enriched with CSCs/CICs and have a partial EMT feature.

DNAJB8/Dnajb8 mRNA was predominantly expressed in SP cells derived from both ACHN and RenCa cells (Fig. 2D). DNAJB8 mRNA was also predominantly expressed in SP cells derived from colon carcinoma cells (KM12LM and SW480 cells), lung carcinoma cells (LHK2 cells), and breast carcinoma cells (MCF7 cells; Supplementary Fig. S2). Western blotting and immunostaining using SP cells and MP cells derived from ACHN and RenCa cells also revealed preferential expression of DNAJB8 protein in SP cells (Fig. 2E and F and Supplementary Fig. S3A and B). DNAJB8 protein was also preferentially expressed in SP cells derived from LHK2 and SW480 cells (Supplementary Fig. S4). These findings indicated that DNAJB8/Dnajb8 is preferentially expressed in the CSC/CIC population and that DNAJB8 is therefore a novel CSC antigen.

To address the functions of DNAJB8 in RCC CSCs/CICs, we generated stable transformants expressing DNAJB8 (Fig. 1C). The percentages of SP cells in DNAJB8-transduced ACHN and RenCa cells (ACHN/DNAJB8 and RenCa/DNAJB8) were increased compared with those in control cells (ACHN/Mock and RenCa/Mock cells; Fig. 3A). Similar phenomena were observed in several other cell lines including RCC (Caki1), prostate carcinomas (LNCaP and DU145), and bladder carcinoma cells (T-24; Supplementary Fig. S5). Furthermore, DNAJB8-transduced ACHN cells showed higher tumor-initiating ability than that of Mock-transduced ACHN cells (Fig. 3B and Table 1). DNAJB8 contains an N-terminal J-domain, which is responsible for association with HSP70 families, and a C-terminal serine-rich region, which is capable of interacting with histone deacetylases (HDAC; HDAC4, HDAC6, and SIRT2) and has a role in suppression of protein aggregation. A DNAJB8 mutant that lacks the C-terminal serine-rich region (ΔSSF-SST) showed a smaller SP augmentation effect (Supplementary Fig. S6), suggesting that the C-terminal serine-rich region has a role induction of CSCs/CICs.

DNAJB8 gene knockdown experiments were carried out using gene-specific siRNAs. Three different DNAJB8-specific siRNAs (siRNA A, B, and C) could suppress the expression of DNAJB8 (Fig. 3C); therefore, we used siRNA A in the following experiments. Transfection of siRNA A almost completely eliminated the SP cell population in both wild-type ACHN and ACHN/DNAJB8 cells (Fig. 3D). Furthermore, siRNA A transfection significantly inhibited the tumorigenicity of ACHN cells in NOD/SCID mice (Fig. 3E). Gene knockdown of DNAJB8 by siRNA transfection almost completely decreased the numbers of SP cells. Thus, we generated a DNAJB8 mutant (DNAJB8 mt) by site-directed mutagenesis to confirm the siRNA results (Fig. 3F). DNAJB8 mt contains 6 DNA point mutations within the iRNA target sequence and codes the same amino acid sequence as that of the wild type. DNAJB8 mt was insensitive to siRNA A transfection, whereas DNAJB8 wt was inhibited by siRNA transfection (Fig. 3G). The siRNA transfection into ACHN/DNAJB8 cells decreased the numbers of SP cells (Fig. 4D); however, the DNAJB8 mt stable transformant showed no reduction of SP cells by siRNA transfection (Fig. 3H). These results indicated that the effects of siRNA transfection were specific for targeting DNAJB8 mRNA, and DNAJB8 therefore has a role in the maintenance of RCC CSCs/CICs.

Because DNAJB8 is one of the cancer-testis antigens and one of the CSC antigens and has a role in the maintenance of RCC CSCs/CICs, we hypothesized that DNAJB8 is a suitable immunologic target of RCC CSCs/CICs-targeting immunotherapy. However, it is not clear which will bring about a better antitumor effect: targeting CSC antigens or targeting shared antigens. On the basis of this point of view, we compared the immunogenicity of DNAJB8 with that of the well-characterized TAA Survivin (26–29), as Survivin is expressed in both CSCs/CICs and non-CSCs/CICs at the same levels (Fig. 5A) and it is a shared antigen.

We subcloned FLAG-tagged Dnajb8 and Survivin into the pcDNA3.1 expression vector and confirmed the expression by Western blot analysis (Fig. 4B). We carried out an ELISPOT assay to verify the immunization of DNA vaccine. Anti-Dnajb8- or anti-Survivin–specific IFN-γ spots could be observed with CD8⁺ T cells derived from Dnajb8- or Survivin-immunized mice, respectively (Fig. 4C). To address the antitumor effects, we injected RenCa cells into Dnajb8- and Survivin-immunized mice. Antitumor effects were observed in both Dnajb8- and Survivin-immunized mouse compared with control vector- or PBS-immunized mouse (Fig. 4D; P values of 0.002 and 0.026, respectively). Dnajb8-immunized mouse showed significantly greater antitumor effect than that in Survivin-immunized mouse (P = 0.03; Fig. 4D).

To evaluate the subtype of T cells for antitumor activities, we depleted murine CD4⁺ or CD8⁺ T cells in immunized mice by 4 intraperitoneal injections of anti-CD4 or CD8 antibody (Fig. 5A). CD4⁺ T-cell depletion significantly inhibited the antitumor effect of Dnajb8 immunization (P = 0.046), whereas CD8⁺ T-cell depletion did not (Fig. 5B). However, the survival period of RenCa cell-injected mice was significantly shorter for both CD4⁺ T-cell–depleted mice and CD8⁺ T-cell–depleted mice than the survival period of Dnajb8-immunized mice (Fig. 5C). These observations indicated that both CD8⁺ T cells and CD4⁺ T cells might have a role in the antitumor effect. Three of 5 Dnajb8-immunized mice showed complete inhibition of tumor formation initiation, and the mice were therefore rechallenged with 10 times larger numbers of higher tumorigenic RenCa/DNAJB8 cells. Initiation of tumor formation was completely inhibited in those mice, suggesting a strong tumor-inhibitory effect of Dnajb8 immunization (Fig. 5D).

In this study, we investigated the distributions of DNAJB8 mRNA and protein and found that DNAJB8 is expressed only in the testis among normal tissues. DNAJB8 protein expression was detected in postmeiotic sperm and spermatid. Previous studies have shown that DNAJB6, DNAJB1, and DNAJB13 are also expressed in the testis (30–32), and Dnaja1 was reported to have a role in spermatogenesis by regulating androgen receptor signals (33). Therefore, DNAJB8 is another testis-expressing HSP40 family protein and may have a role in spermatogenesis, especially in the postmeiotic stage.

We isolated RCC CSCs/CICs as SP cells. SP cells have been reported to be enriched with CSCs/CICs in several kinds of malignancies (34–36). SP cells derived from RCC and normal renal tubule epithelia have also been reported (37, 38); however, the tumor-initiating ability of those RCC SP cells has not been characterized yet. Tumor-initiating ability is one of major characteristics of CSCs/CICs, and this characteristic makes CSCs/CICs a reasonable target of cancer therapy. In this study, SP cells derived from ACHN human RCC cells and RenCa mouse RCC cells showed higher tumor-initiating ability than that of MP cells, and these SP cells are suitable for analysis of RCC CSCs/CICs.

HSPs, especially HSP90, are overexpressed in a wide range of human cancers, allowing mutant proteins to be retained in cancer cells and to confer resistance to cytotoxic therapies (39). On the other hand, most of the HSP40 family proteins have been reported to be inversely correlated with high-grade malignancy, and HSP40 family proteins have been reported to function as tumor suppressors (40). In this study, we found that DNAJB8 was expressed in several kinds of cancer cells, including RCCs, and that it was preferentially expressed in the CSC/CIC population. DNAJB8 has a role in maintenance of RCC CSCs/CICs as shown by siRNA and mutant DNAJB8 canceling experiments, and DNAJB8 is thus an HSP40 family member that has oncogenic potential. CSCs/CICs are known to be resistant to several stresses, including chemotherapy and radiotherapy (15), and overexpression of DNAJB8 might be related to the antistress ability and resistance to treatments. We observed that overexpression of DNAJB8 increased the percentage of SP cells; however, we still detected MP cells in ACHN/DNAJB8 cells. With regard to tumor-initiating ability, ACHN/DNAJB8 SP cells showed higher tumor-initiating ability than that of ACHN/DNAJB8 MP cells (Table 1). Because both ACHN/DNAJB8 SP cells and ACHN/DNAJB8 MP cells expressed DNAJB8 at almost same levels (data not shown), these observations suggest the existence of other cofactor for induction of CSCs/CICs. In a previous study, DNAJB8 was shown to have role in inhibition of cytotoxic protein aggregation (22), and it is important for association with HDACs through the C-terminal serine-rich region. We also found that the serine-rich region of DNAJB8 has a role in maintenance of CSCs/CICs. These observations suggest that association with HDACs (HDAC4, HDAC6, and SIRT2) might be important for DNAJB8 functions in CSCs/CICs, and that HDACs are possible candidates of cofactors of DNAJB8.

In this study, we showed for the first time that DNAJB8 can be a target of immunotherapy. Mutated and wild-type HSP70 proteins have been reported to be targets of CTLs (41–43), whereas other HSP family proteins have never been reported to be targets of immunocytes. HSP family proteins are often expressed in several kinds of normal organs, and immunocytes might be immunologically tolerated to those proteins. On the other hand, DNAJB8 is expressed only in the testis, and it is thus a novel CT antigen that is regarded as an ideal immunologic target.

HSP family proteins work as molecular chaperones and bind to their client proteins. Antigenic peptide-bound HSP 70 and 90 family proteins are known to work as immune modulators that enhance the cross-priming pathway and enhance antitumor immunity (44–47). In this study, we confirmed anti-DNAJB8 immunity with DNAJB8-immunized mouse spleen cells. However, there remains the possibility that DNAJB8 also binds to another antigen protein/peptide and induces antigen-specific immunity, and further investigation should be carried out.

A DNA vaccination experiment revealed that targeting CSC antigen (Dnajb8) was more effective than targeting a shared antigen (Survivin). In a previous study, we found that colon CSCs/CICs can be recognized by an established CTL clone at the same level as non-CSCs/CICs both in vitro and in vivo (16). We therefore suggest that CTLs are promising tools to target CSCs/CICs. Expression of TAAs is essential for CTL recognition, and we categorized TAAs into 3 groups (CSC antigens, shared antigens, and non-CSC antigens; ref. 17). In a recent study, we found that TAA expression in the CSC/CIC population is necessary to achieve an antitumor effect and that the antitumor effect of a shared antigen is greater than that of non-CSC antigens (48). Then, which is the best target, CSC antigens or shared antigens? In this study, we showed for the first time that a CSC antigen is more effective than a shared antigen. Our results indicate that a CSC antigen has the greatest antitumor effect and a non-CSC antigen has the smallest antitumor effect and that evaluation of the distribution in CSCs/CICs and non-CSCs/CICs is important to predict the efficiency of antitumor effects of novel TAAs.

We showed that targeting CSCs/CICs is an effective approach for cancer immunotherapy. On the other hand, glioma-associated CSCs/CICs have been reported to be related to immunosuppression through B7-H1 and soluble Galectin-3 and STAT3 signaling (49, 50). These observations suggest that CSCs/CICs might suppress the CTL induction phase but might not affect the CTL effector phase. In this study, we observed a significant antitumor effect with immunization using Dnajb8-coding plasmid. This is a prophylactic model and CTL induction might not be affected by CSCs/CICs included in RenCa cells and thus bring about preferable results. Therefore, immunotherapy targeting CSCs/CICs might be useful for suppression of posttreatment tumor recurrence and/or adoptive transfer of established CTLs.

In summary, we identified an HSP40 family protein, DNAJB8, as one of the CT antigens and also as a CSC antigen. DNAJB8 has a role in maintenance of RCC CSCs/CICs. Targeting a CSC antigen is more effective than targeting a shared antigen, and DNAJB8 is a possible candidate for CSC/CIC-targeting immunotherapy.

No potential conflicts of interest were disclosed.

Conception and design: Y. Hirohashi, T. Torigoe

Development of methodology: Y. Hirohashi, T. Torigoe, A. Takahashi, K. Kamiguchi, A. Sokolovskaya, R. Fujii

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Nishizawa, Y. Hirohashi, T. Torigoe, A. Takahashi, Y. Tamura, H. Asanuma, R. Morita, A. Sokolovskaya, J. Matsuzaki, R. Yamada

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Nishizawa, Y. Hirohashi, T. Torigoe, A. Takahashi, Y. Tamura, H. Asanuma

Writing, review, and/or revision of the manuscript: S. Nishizawa, Y. Hirohashi, T. Torigoe, T. Kanaseki, R. Fujii

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Hirohashi, T. Torigoe, T. Mori, H.H. Kampinga, T. Kondo, I. Hara

Study supervision: Y. Hirohashi, T. Torigoe, T. Hasegawa, N. Sato

The authors thank Ms. Emiri Nakazawa and Ms. Eri Saka for technical assistance; Drs K. Imai, K. Itoh, and P.G. Coulie for kindly providing cell lines; and Dr. T. Kitamura for kindly providing a retrovirus system.

This study was supported by a grant-in-aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to N. Sato), program for developing the supporting system for upgrading education and research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to N. Sato) and Takeda Science Foundation (to Y. Hirohashi).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.