Therapeutic Efficacy of Cancer Stem Cell Vaccines in the Adjuvant Setting Free

Although surgical resection has been a standard treatment for solid malignancies, therapeutic efficacy is limited by both local and distant recurrence (1–3). There are many factors associated with tumor recurrence (4, 5). Several reports have described strategies to eliminate residual tumor cells after surgery (3, 6). However, effectively preventing local tumor recurrence remains a significant challenge. The existence of micrometastasis at the time of tumor resection represents an even greater therapeutic challenge, as 90% of tumor-related deaths are due to tumor metastasis. There is increasing evidence that many cancers are driven and maintained by a subpopulation of cells that display stem cell properties. Cancer stem cells (CSC) can self-renew, mediate tumor growth, and contribute to tumor recurrence and metastasis (7–9). Targeting CSCs may thus increase the therapeutic efficacy of current cancer treatment.

ALDEFLUOR/ALDH (aldehyde dehydrogenase) activity has been successfully used as a marker to enrich CSC populations in a variety of cancers (10–17). We previously demonstrated that ALDH^high murine squamous carcinoma SCC7 and D5 melanoma cells were highly enriched for tumor-initiating capacity (15). Their protective immunogenicity was evaluated by administering CSC-based dendritic cell (DC) vaccines in syngeneic immunocompetent hosts (15). In a recent study (17), we demonstrated significant therapeutic efficacy conferred by an ALDH^high CSC-DC vaccine in the treatment of established tumors after localized radiotherapy.

Eliminating microscopic residual disease in the tumor bed is important in preventing local disease recurrence. Administration of CSC-based vaccines after surgical excision of tumor, where local recurrence is high, may reduce local tumor relapse and distant metastasis and possibly improve survival. Furthermore, as CSCs mediate tumor metastasis, targeting this cell population in the adjuvant setting may eliminate micrometastasis prolonging survival. In this study, we evaluated the potential therapeutic efficacy of this approach in the adjuvant setting using CSC-DC vaccination following surgical resection of the tumor, or by treatment of minimal disease.

We developed a vaccination strategy utilizing cell lysates from ALDH^high SCC7 or D5 CSCs to pulse dendritic cells (CSC-DC). DCs pulsed with ALDH^low SCC7 or D5 non-CSC lysate (ALDH^low-DC), or with heterogeneous, unsorted cell lysate (H-DC) served as controls. Vaccination with ALDH^high CSC-DC in immunocompetent mice significantly inhibited SCC7 local tumor recurrence after surgery, and inhibited minimal D5 tumor growth with prolonged survival significantly more than either ALDH^low-DC or H-DC vaccination. Furthermore, this effect was accentuated by simultaneous PD-L1 immune checkpoint blockade.

Female C3H/HeNCr MTV (C3H) mice and C57BL/6 (B6) mice were purchased from The Jackson Laboratory and Charles River Laboratories (15). The University of Michigan Laboratory of Animal Medicine approved all animal protocols.

The squamous carcinoma cell line, SCC7, produces a poorly immunogenic tumor and is syngeneic to C3H mice. D5 is a clone of the melanoma cell line B16, which is syngeneic to B6 mice, and was originally established in our laboratory. The cell lines were grown in complete medium consisting of RPMI1640 and supplements (15).

The ALDEFLUOR Kit (StemCell Technologies) was used to isolate ALDEFLUOR⁺/ALDH^high CSCs from the SCC7 and D5 cells (15).

Tumor cell lysates of unsorted, ALDEFLUOR⁺/ALDH^high, or ALDEFLUOR⁻/ALDH^low SCC7 and D5 cells were prepared as described previously (15). Bone marrow–derived murine cells were cultured in 10-mL complete medium (CM) supplemented with 20 ng/mL GM-CSF at a concentration of 0.2–0.4 × 10⁶ cells/mL in non-tissue culture petri dishes (Corning) on day 0. Fresh medium supplemented with 20 ng/mL GM-CSF was added on days 3 (10 mL). On day 6 and 8, 10 mL of cultured cell suspension was taken from each dish, respectively, centrifuged, and the pellet resuspended in 10 mL of fresh CM with 20 ng/mL of GM-CSF, and added back to each dish. On day 10, DCs were harvested by dispenser and enriched by Opti-Prep density gradient medium. Lysates of unsorted, ALDH^low, or ALDH^high cells were added to DCs at a 1:3 cell equivalent ratio. The DCs were then incubated at 37°C for 24 hours with 5% CO₂. After incubation, the unsorted tumor cell lysate–pulsed DCs (H-DC), ALDH^low lysate–pulsed DCs (ALDH^low-DC), or ALDH^high lysate–pulsed DCs (ALDH^high-DC, e.g., CSC-DC) were used as vaccine as specified in the subsequent experiments.

C3H mice were inoculated subcutaneously with 0.5 × 10⁶ SCC7 cells on day 0. On day 21, the mice were subjected to surgical tumor resection except for one group serving as control. The animals with the subcutaneous tumor removed were then divided into 4 groups (n = 5), and administrated with PBS, H-DC, ALDH^low SCC7-DC, and ALDH^high SCC7 CSC-DC vaccine, respectively, 24 hours after tumor resection. The vaccination was repeated on day 29 and day 36, respectively. Each mouse was inoculated subcutaneously with 2 × 10⁶ DCs per vaccine. In additional experiments, when SCC7 CSC-DC was used in combination with an anti-PD-L1 antibody (MedImmune Inc.), the vaccination was only repeated once on day 29 with anti-PD-L1 administration. In the minimal tumor model, B6 mice were inoculated subcutaneously with 5,000 D5 cells. The first vaccine was administered subcutaneously 24 hours after tumor inoculation, followed by a second vaccine on day 8. Each vaccine comprised 2 × 10⁶ DCs. The long and short diameters of tumor masses, as well as the tumor volumes, were measured three times per week. The volumes were calculated as: tumor volume = (width² × length)/2. Survival was monitored and recorded as the percentage of survivors after tumor inoculation.

At the conclusion of the experiments, the lungs were harvested and stained with hematoxylin and eosin (H&E) to discern the histopathologic response.

Freshly harvested subcutaneous D5 tumors were disaggregated into single-cell suspensions (18). Unsorted, ALDH^high, and ALDH^low D5 cells were then, respectively, incubated with PE-anti-CCR10 for flow cytometry analysis with a BD LSR-cytometer. To evaluate the PD-L1 levels in the CSC and non-CSC populations after treatment, D5 tumors were harvested at the end of therapy to prepare tumor cell suspensions. These tumor cells were then incubated with PE-anti-PD-L1 (BioLegend), followed by staining with ALDEFLUOR (FITC) for ALDH^high and ALDH^low population isolation as described in the section “ALDEFLUOR assay.” The ALDH^high (CSC) and ALDH^low (non-CSC) D5 tumor cells were then examined by flow cytometry for PD-L1 expression.

The mRNA expression levels of chemokines CCL27 and CCL28 in lung tissues were analyzed using real-time quantitative PCR (qRT-PCR; ref. 17). The preparations of the total RNA and cDNA were described previously (19). The data were expressed as the relative fold change.

Equal doses of Ccr10 siRNA and negative siRNA (Qiagen) were used according to the manufacturer's instructions to transfer unsorted, ALDH^high, and ALDH^low D5 cells for 48 hours to inhibit CCR10 expression. Cells (10⁶) were then resuspended and RNA extracted using RNeasy Mini Kit (Qiagen). Five milligrams of total RNA was reverse transcribed (M-MLV, Invitrogen) to generate cDNA for subsequent RT-PCR. Platinum SYBR Supermix (Invitrogen) was used to amplify sequences for Ccr10 (forward: CAGTCTTCGTGTGGCTGTTGTC; reverse: TCACAGTCTGCGTGAGGCTTTC) and GAPDH (forward: TGAAGCAGGCATCTGAGGG; reverse: CGAAGGTGGAAGAGTGGGAG) using a standard three-step protocol (35 cycles of 30 seconds each at 95°C, 58°C, and 72°C). Melting point analysis verified the presence of single products.

RPMI1640 (500 μL) containing 1 × 10⁶ D5 cells or Ccr10 siRNA–transferred D5 cells were added to the top chamber of a Transwell (insert pore size, 8 μm; Corning). Chemokines CCL27 and CCL28 (R&D Systems) were added to the bottom chamber in a volume of 750-μL RPMI1640, which contained 20% FBS. After incubation at 37°C for 27 hours, the cells that migrated to the bottom surface of membrane were stained with Diff-Quik set (Siemens Healthcare Diagnostics Inc). Cells were photographed under the microscope at 200× magnifications, and counted in 5 fields of triplicate membranes.

Spleens were harvested from animals subjected to various treatments at the end of the experiments. Splenic B cells were purified and activated in CM supplemented with lipopolysaccharide (LPS, Sigma), anti-CD40 (AdipoGen), and IL2 (Prometheus Laboratories Inc.; ref. 15). The culture supernatants were collected and stored at −20°C for future experiments. Splenic T cells were purified and activated to generate CTLs that were analyzed in LDH cytotoxicity assays (15).

Sorted ALDH^ihgh or ALDH^low D5 cells were incubated with the immune supernatants collected from the cultured B cells with equal quantities of IgG followed by incubation with the second FITC–conjugated anti-mouse IgG. The binding of supernatant antibody to ALDH^high versus ALDH^low D5 cells was assessed using flow cytometry (15). Antibody and complement-mediated cytotoxicity against CSCs was measured as described previously (15).

Survival analysis was determined by the log-rank test. Analysis for the presence of lung metastasis was performed using the Fisher exact test. Other data were evaluated by unpaired Student t test (2 cohorts) or one-way ANOVA (>2 cohorts).

We previously demonstrated that administration of ALDH^high SCC7 CSC-DC vaccines in immunocompetent mice induces protection against subsequent SCC7 challenge (15). In this study, we examined the therapeutic potential of CSC-DC vaccination to prevent local tumor recurrence, reduce metastasis, and prolong survival when deployed in the adjuvant/early disease settings. The first model employed surgical excision of SCC7 subcutaneous head and neck squamous carcinomas, a tumor in which local recurrence contributes to patient mortality and morbidity (20, 21). C3H mice were inoculated subcutaneously with 0.5 × 10⁶ SCC7 tumor cells. Resulting tumors were surgically excised 21 days after inoculation, followed by vaccination with DCs pulsed with lysates of heterogeneous unsorted SCC7 cells (H-DC), ALDH^low SCC7 cells (ALDH^low-DC), or ALDH^high SCC7 cells (ALDH^high –DC). Vaccines were administrated once per week for 3 weeks starting on the second day postsurgery. Mice were subsequently monitored for local tumor recurrence and survival.

As shown in Fig. 1, there was 100% mortality in tumor-bearing mice without tumor resection by day 40 due to progressive tumor growth. In PBS control mice, tumor recurrence was noted beginning on day 30 and all mice ultimately died by day 55 due to tumor growth. The H-DC and ALDH^low-DC vaccination delayed tumor recurrence, resulting in prolonged animal survival compared with control mice. More importantly, the ALDH^high-DC (CSC-DC) vaccine significantly reduced tumor recurrence compared with the PBS control (P < 0.0001), H-DC (P = 0.0221), and ALDH^low-DC (P = 0.0495) vaccination, respectively (Fig. 1A). As a result, the ALDH^high-DC treatment significantly increased animal survival compared with the other treatments or control mice (Fig. 1B). While only 50% of the mice in H-DC and ALDH^low-DC–treated groups survived until day 65, all of the mice treated with the ALDH^high-DC vaccine survived until that timepoint. These results demonstrate the ability of the ALDH^high-DC vaccine to reduce local recurrence and prolong survival in this model of SCC.

One of the major recent advances in tumor immunotherapy has been the development of strategies to block the immunosuppressive components of the tumor microenvironment (22, 23). We next performed experiments where SCC7 subcutaneous tumors were surgically excised as in Fig. 1A, and animals were treated as indicated in Fig. 1C with or without anti-PD-L1 administration. SCC7 ALDH^high-DC (CSC-DC) vaccination plus anti-PD-L1 administration significantly inhibited tumor relapse (Fig. 1C) and prolonged animal survival (Fig. 1D) compared with either treatment alone. These experiments clearly demonstrate that immunologically targeting CSCs, while simultaneously blocking PD-1/PD-L1–mediated immune suppression, has the potential to significantly enhance the efficacy of cancer immunotherapies.

To evaluate the therapeutic efficacy of the CSC-DC vaccine in the setting of micrometastatic disease, we utilized the highly metastatic D5 mouse melanoma model. To test the efficacy of the CSC-DC vaccine in treating micrometastatic disease, it was administered 24 hours after inoculation of tumor cells. Syngeneic B6 mice were inoculated with 5,000 D5 melanoma cells subcutaneously followed by vaccination 24 hours later (day 1) with DCs pulsed with the ALDH^high D5 CSCs (CSC-DC) cell lysate, ALDH^low D5 cell lysate (ALDH^low-DC), heterogeneous unsorted D5 cell lysate (H-DC), or with PBS, respectively. The treatment was repeated on day 8. As shown in Fig. 2A, no significant difference in primary tumor growth was observed among PBS, H-DC, or ALDH^low-DC–treated mice. However, administration of the CSC-DC vaccine treatment resulted in significant inhibition of tumor growth compared with controls (P < 0.02 vs. all other groups). The CSC-DC–treated mice also survived longer than controls (Fig. 2B). These data indicate that the treatment of subcutaneous tumor-bearing mice in the setting of minimal tumor with CSC-DC vaccination generated significant antitumor immunity, resulting in inhibited subcutaneous tumor growth and prolonged survival of the tumor-bearing hosts.

To investigate the effect of these treatments on the development of lung metastases, we harvested the lungs at the end of the experiments and accessed lung metastases. Representative histology of the lungs is shown in Fig. 2C. Mice subjected to PBS treatment, H-DC, or ALDH^low-DC vaccine all displayed numerous large lung metastases. In contrast, there were significantly reduced lung metastases detected in the lungs harvested from ALDH^high CSC-DC–vaccinated hosts (Fig. 2C). The ALDH^high-DC vaccine significantly inhibited tumor metastasis in the lung compared with PBS, H-DC, and ALDH^low-DC vaccine treatments (P < 0.05, Fig. 2D). Only 2 of 11 total mice developed lung metastasis after ALDH^high-DC vaccination, while 9 of 11 mice treated with PBS or ALDH^low-DC, and 8 of 11 mice treated with H-DC developed lung metastases (Fig. 2D). Together, these results indicated that CSC-DC vaccine significantly inhibited tumor growth and lung metastases, resulting in increased animal survival. In addition, as shown in Fig. 2E, after ALDH^high-DC vaccination, PD-L1 expression on ALDH^high cells (CSC) was decreased to 5.3% compared with PBS (17.2%), H-DC (8.3%), or ALDH^low-DC (11.4%) treatment. Similarly, PD-L1 expression on ALDH^low cells (non-CSC) was decreased after ALDH^high-DC vaccination to 4.5% (Fig. 2F) compared with PBS (8.2%), H-DC (6.2%), or ALDH^low-DC (7.6%) treatment. Statistically, when the mean ± SE of the three experiments' data were compared (Fig. 2G), ALDH^high-DC vaccination significantly (P < 0.05) reduced the PD-L1 expression on both ALDH^high cells (CSC) and ALDH^low (non-CSC) cells.

Chemokines play a significant role in tumor metastasis (24–26). We examined the expression of CCR10 on tumor cells from mice treated in the minimal disease setting, and compared its expression in ALDH^high CSCs versus ALDH^low non-CSCs. CCR10 expression was accessed by flow cytometry in D5 tumors harvested from animals subjected to various vaccines (Fig. 3A and B). CSC-DC vaccination significantly decreased expression of CCR10 in unsorted bulk tumor cells (P < 0.01 vs. all other groups). With CSC-DC vaccination, the expression of CCR10 on D5 tumor cells was significantly decreased to approximately 3% compared with PBS treatment (>20%), or with H-DC and ALDH^low-DC vaccination (both around 15%; Fig. 3A and B). We then sorted ALDH^high and ALDH^low cells from freshly harvested D5 tumors subjected to vaccine therapy, and assessed their CCR10 expression. We found that the expression of CCR10 was significantly (P < 0.0001) higher on D5 ALDH^high CSCs (>60%) than on ALDH^low non-CSCs (<20%; Fig. 3C and D, PBS groups). ALDH^high CSC-DC vaccination significantly decreased the expression of CCR10 on D5 ALDH^high as well as on ALDH^low cells (Fig. 3C and D). Finally, using qRT-PCR, we found that mRNA for the corresponding chemokine ligands for CCR10 in the lung tissues, CCL27 and CCL28, were both significantly decreased after ALDH^high CSC-DC vaccine treatment (P < 0.01 vs. all other groups; Fig. 3E). Collectively, these data suggest that CSC-DC vaccination may inhibit pulmonary metastasis of the local tumor by significantly downregulating the expression of CCR10 on primary tumor cells, particularly on the ALDH^high CSCs in the primary tumor, as well as reducing the production of CCR10 ligands, CCL27 and CCL28, in the lung tissues.

To substantiate the role for CCR10 and its ligands in tumor metastasis, we used Ccr10 siRNA to inhibit Ccr10 gene expression as described in Materials and Methods. To test the effect of Ccr10 siRNA on the inhibition of D5 cell migration in vitro, we carried out a chemotaxis assay.

As shown in Fig. 4A, Ccr10 siRNA–treated D5 cells demonstrated significantly (P < 0.001) less migration ability than nontreated D5 cells toward the CCL27 and CCL28 added to the bottom of the transwell at the concentrations as indicated. To test the effect of Ccr10 siRNA on the inhibition of ALDH^low and ALDH^high D5 cell migration in vivo, we compared the metastasis of nontreated ALDH^low and ALDH^high D5 cells with that of Ccr10 siRNA–transferred ALDH^low and ALDH^high D5 cells when 1 × 10⁶ cells of each group were intravenously injected into the normal B6 mice. As expected, ALDH^high D5 cells generated significantly (P = 0.03) more metastasis than ALDH^low D5 cells (Fig. 4B). Importantly, Ccr10 siRNA–treated ALDH^low and ALDH^high D5 cells generated significantly less metastasis than nontreated ALDH^low (P = 0.0002) or ALDH^high (P = 0.0003) D5 cells, respectively. These experiments strongly suggest that CCR10 plays an important role in the migration and therefore the metastasis of D5 tumor cells.

To confirm the efficacy of Ccr10 siRNA in Ccr10 gene silencing, equal doses of Ccr10 siRNA and control siRNA were used to transfer unsorted D5 cells for various time periods, for example, 24, 48, and 72 hours. Figure 4C shows that transfer of unsorted D5 cells for 48 hours begins to demonstrate significantly (P = 0.0005) silenced Ccr10 gene. We therefore transferred unsorted D5 cells for 48 hours in Fig. 4A as well as for ALDH^high and ALDH^low D5 cells in Fig. 4B. In addition, Fig. 4D revealed that Ccr10 gene expression in ALDH^high D5 cells is higher (P < 0.0001) than that in ALDH^low D5 cells. Importantly, Ccr10 siRNA–treated ALDH^low and ALDH^high D5 cells showed significantly downregulated Ccr10 gene expression compared with nontreated ALDH^low (P < 0.05) and ALDH^high (P < 0.05) D5 cells, respectively.

To provide experimental evidence that CSC-DC vaccination induces specific anti-CSC immunity, we collected the spleens after the full treatment course in the minimal D5 tumor model, purified splenic B cells, and activated them in vitro with LPS and anti-CD40. We then accessed the specificity of CSC-DC vaccine–primed antibody by binding assays of the B-cell culture supernatants to ALDH^high D5 CSCs versus ALDH^low D5 non-CSCs, respectively. Immune supernatants produced by B cells from mice that received ALDH^high-DC treatment bound to ALDH^high D5 CSCs (60.8%; Fig. 5A) much more effectively than the immune supernatants collected from PBS-treated (12.3%), H-DC vaccinated (29.8%), or ALDH^low-DC–treated (15.7%) mice. In contrast, the immune supernatants produced by B cells harvested from H-DC or ALDH^low-DC–vaccinated mice bound to the ALDH^low non-CSCs (45.8% and 50.2%, respectively) significantly more than the immune supernatants produced by B cells harvested from the CSC-DC–vaccinated mice (6.8%) or from PBS-treated controls (18.8%). Figure 5B shows the results of multiple binding assays, indicating that the immune supernatants produced by CSC-DC vaccine–primed B cells bound to the ALDH^high D5 CSCs much more effectively (P < 0.01 vs. all other groups). In contrast, the ALDH^low-DC vaccine–primed immune supernatants bound to the ALDH^low non-CSCs similar to the binding by H-DC vaccine–primed immune supernatants, but significantly more than the binding of PBS or CSC-DC vaccine–primed immune supernatants (Fig. 5C), demonstrating CSC-DC vaccine induced CSC-specific humoral immunity.

To examine the functional consequence of CSC-specific antibody induced by CSC-DC vaccination, we performed antibody and complement-dependent cytotoxicity (CDC) assays (Fig. 5D). ALDH^high CSC-DC vaccine–primed immune supernatant killed ALDH^high D5 CSCs significantly more than the immune supernatants collected from other groups (P < 0.001 vs. all other groups). In contrast, the immune supernatant harvested from H-DC or ALDH^low non-CSC vaccine–treated host resulted in significant ALDH^low D5 cell lysis, while the immune supernatant from the ALDH^high CSC-DC–vaccinated hosts produced minimal lysis of the ALDH^low targets. Together, these data support the conclusion that ALDH^high D5 CSC-DC vaccine confers significant host anti-CSC humoral immunity by producing D5 CSC–specific antibodies that specifically bind and kill D5 CSCs.

We next examined the ability of CSC-DC vaccination to generate host CSC–specific CTL activity. As shown in Fig. 5, we collected the spleens at the end of the treatment in the minimal D5 tumor model, purified splenic T cells, and activated them in vitro with anti-CD3/anti-CD28 followed by expansion in IL2. This activation procedure generates cytotoxic T cells (>95 CD3 cells; ref. 15). We then measured the CTL activity of these T cells on ALDH^high D5 CSCs versus ALDH^low D5 non-CSCs, respectively. D5 ALDH^high-DC–primed CTLs mediated significantly greater cytotoxicity in D5 ALDH^high CSCs at all effector to target (E:T) ratios compared with the CTLs generated from PBS, H-DC, or ALDH^low DC–primed CTLs (P < 0.05; Fig. 6A). In contrast, CTLs generated from splenocytes of mice subjected to ALDH^low DC and H-DC vaccination selectively killed ALDH^low D5 cells (Fig. 6B). These experiments indicate that CSC-DC vaccination conferred host CTL reactivity as well as humoral responses against CSCs in the treatment of minimal tumor disease.

As described above, ALDH^high CSC-DC vaccination induced significant host cellular and humoral immune responses against CSCs. To confirm that CSCs are effectively targeted by CSC-induced immunity, we determined the effect of CSC-DC vaccination on the proportion of ALDH^high CSC in vivo. Assessment of the ALDH^high population was performed by flow cytometry using the Aldefluor assays as described previously (15). We mixed the tumor cells from mice of each experimental group, and generated representative flow cytometric graphs to demonstrate the ALDH^high populations in each group (Fig. 7A). Subcutaneous tumors harvested from CSC-DC–treated mice contained only 1.7% ALDH^high cells, which was significantly less than that present in the subcutaneous tumors subjected to PBS (13.4%), H-DC (7.5%), or ALDH^low-DC (8.3%) treatments. As shown in Fig. 7B, CSC-DC vaccination significantly reduced the percentage of ALDH^high populations compared with PBS, H-DC, or ALDH^low-DC treatments (P = 0.0002, 0.0002, and 0.0029, respectively) in this tumor model. Together, these studies demonstrate that in the D5 minimal disease model, CSC-DC vaccine elicits both humoral and cellular immune responses reducing the proportion of CSC, resulting in decreased tumor growth, lung metastases, and prolonged survival.

Utilizing two tumor models, we demonstrate the efficacy of a CSC-DC vaccine when used to treat minimal disease in the adjuvant setting. Several reports have described the generation of CSC-specific CD8 T effector cells in vitro (27–30); the killing of CSCs via nonspecific immune effector cells (31–34) as well as by oncolytic viruses (35) and antibodies (36). We previously reported therapeutic efficacy of CSC-DC vaccination in the treatment of established tumors after localized radiotherapy (17). However, as CSCs may be responsible for local tumor recurrence after resection (37) as well as mediating tumor metastasis (38–41), CSC-targeted therapeutics may have their greatest utility when they are utilized in the adjuvant-minimal disease setting. We examined this utilizing two different mouse tumor models.

The SCC7 squamous carcinoma model was designed to determine the efficacy of CSC-DC vaccination after surgical removal of the primary tumor. This model is clinically relevant, as in squamous carcinoma of the head and neck resection of bulky SCC primary tumors has been associated with a high rate of local tumor recurrence associated with significant morbidity and mortality (20, 21). Using the murine SCC7 tumor model, we found that the SCC7 ALDH^high CSC-DC vaccine significantly inhibited tumor recurrence and prolonged animal survival following surgical resection compared with SCC7 H-DC or SCC7 ALDH^low-DC vaccinations. Simultaneous administration of an anti-PD-L1 mAb significantly enhanced the therapeutic efficacy of SCC7 CSC-DC vaccine in the adjuvant setting.

The second model involved treatment of D5 murine melanoma in an early disease setting 24 hours after tumor inoculation. While the H-DC vaccination and the ALDH^low-DC had minimal effects on the local tumor growth and only modestly prolonged survival, the ALDH^high CSC-DC vaccine was significantly more effective in inhibiting tumor growth, resulting in prolonged survival.

To date, the mechanisms that are involved in CSC-DC vaccine–mediated therapeutic efficacy have not been fully defined, and limited experimental evidence was provided for direct targeting of CSCs by CSC-DC vaccine–induced anti-CSC immunity. CSCs are responsible for tumor metastasis and progression (38–41). In this study, we found that the therapeutic efficacy of CSC-DC vaccine was associated with significantly inhibited metastasis of the subcutaneous tumor to the lung. A number of studies have suggested that tumor cell metastasis is determined by the expression level of chemokine receptors on the malignant tumor cells and the expression of corresponding chemokine ligands in the target organs (24–26, 42–46).

We demonstrated high levels of CCR10 (>20%) in tumor cells isolated from control mice. CSC-DC vaccination significantly reduced the expression of CCR10 to 3%. More importantly, we found that the expression of CCR10 was significantly higher on D5 ALDH^high CSCs (>60%) than on ALDH^low non-CSCs (<20%), and ALDH^high CSC-DC vaccine significantly decreased the expression of CCR10 on D5 ALDH^high cells to <15%. In a group of experiments, we found that Ccr10 siRNA treatment to inhibit Ccr10 gene expression significantly blocked tumor cell migration in vitro and metastasis in vivo. In addition, ligands for CCR10, including CCL27 and CCL28, were significantly decreased in the lung tissues harvested from the animals treated with CSC-DC vaccination. These data suggest that decreased CCR10, CCL27, and CCL28 may play an important role in CSC-DC vaccination–induced inhibition of tumor metastasis. Chemokine receptors can activate downstream effectors, such as MAPKs, by complex mechanisms (47). The molecular and biochemical signaling pathways by which CSC-DC vaccination induces downregulation of CCR10, CCL27, and CCL28 remain to be identified.

We examined the ability of CSC-DC vaccines to elicit CSC-specific humoral and cellular immune responses. Using splenocytes collected from the treated mice, we generated CTLs. D5 ALDH^high-DC–primed CTLs significantly killed the D5 ALDH^high CSCs compared with the CTLs generated from PBS, H-DC, or ALDH^low DC–primed CTLs. In contrast, CTLs generated from splenocytes of mice subjected to ALDH^low DC and H-DC vaccination selectively killed ALDH^low D5 cells. These experiments indicate that CSC-DC vaccination confers host CTL activity that specifically target CSCs. In addition, the ALDH^high CSC-DC vaccine–primed host B cells produced antibody specifically bound to ALDH^high CSCs (>60%), which was significantly higher than the binding by antibodies produced PBS, H-DC, or ALDH^low-DC–primed B cells (10%–30%). In contrast, H-DC or ALDH^low-DC vaccine–primed B cells produced antibody preferentially bound to ALDH^low D5 cells (45%–50%), which was significantly higher than the binding by antibodies produced by PBS or ALDH^high CSC-DC vaccine–primed B cells (18% and 6%, respectively). The immunologic consequence of antibody binding of the CSCs was the lysis of the CSCs in the presence of complement, demonstrating that CSC-DC vaccines elicit significant humoral immune responses against CSCs as well as CSC-specific cellular immune responses. While we demonstrated that antibodies and CTLs were induced against CSCs, the identity of any recognized antigens has yet to be elucidated. Identification of CSC antigen(s) represents an active research focus in our laboratory and warrants further investigation.

The induction of cytotoxic T cells to CSCs has been observed in two different animal histologies using the CSC lysate vaccine both in our previous protection study (15) and in this therapeutic study. To date, we have not observed immune tolerance in our model system of CSC-DC vaccination. However, an immune adjuvant may enhance the induction of a tumor lysate–DC vaccine. We previously reported that the use of a second signal agent such as anti-4-1BB mAb augmented the antitumor efficacy of DC-based vaccines (48). We did not use this approach in this report to focus on the use of CSC-DC vaccines by themselves. Nevertheless, the use of adjuvant agents may enhance T, B-cell activation as a method to improve CSC-DC vaccine–induced anti-CSC immunity.

Our experiments provide direct evidence that CSC-DC vaccine can induce anti-CSC immunity by targeting CSCs. As a result, ALDH^high CSC populations in the residual tumor of the mice subjected to CSC-DC vaccine were significantly decreased to <2% compared with the PBS-treated control (∼15%), and was significantly lower than those of the animals subjected to H-DC or ALDH^low-DC treatment (7%–9%). In our previous publications, we have demonstrated that a reduction in ALDH expression is strongly associated with reduction in tumor-initiating capacity as accessed by extreme limiting dilution analysis (49, 50). The values of reduction of ALDH associated with treatments shown in this study are highly statistically significant. Future studies accessing the ability of CSC-DC vaccines to reduce tumor-initiating capacity of treated tumor cells transplanted into secondary animals are warranted. We propose that the significant reduction of the residual CSCs after CSC-DC immunotherapy is due to CSC-DC vaccine-induced cellular and humoral targeting of CSCs. Together, these studies suggest the potential clinical efficacy of utilizing CSC-DC vaccines in the adjuvant/early tumor setting, a strategy that may be augmented by PD-l/PD-L1 immune checkpoint blockade.

R.E. Hollingsworth is a senior director, oncology and has ownership interest (including patents) in MedImmune. No potential conflicts of interest were disclosed by the other authors.

Conception and design: L. Lu, A.E. Chang, J.S. Moyer, M.E. Prince, S. Huang, Q. Li

Development of methodology: Y. Hu, L. Lu, Y. Xia, A.E. Chang, J.S. Moyer, F. Dai, Y. Bao, Q. Li

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Hu, L. Lu, X. Chen, R.E. Hollingsworth, J.H. Owen, M.E. Prince, J. Whitfield, M.S. Wicha

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Hu, L. Lu, Y. Xia, X. Chen, A.E. Chang, J.S. Moyer, F. Dai, Y. Bao, J. Whitfield

Writing, review, and/or revision of the manuscript: Y. Hu, X. Chen, A.E. Chang, E. Hurt, J.S. Moyer, M.E. Prince, J. Whitfield, S. Huang, M.S. Wicha, Q. Li

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.S. Moyer, Y. Bao, J. Xia, S. Huang, Q. Li

Study supervision: E. Hurt, M.E. Prince, S. Huang, Q. Li

We thank Jill Granger for valuable assistance in editing the manuscript.

This work was supported by the Elsa U. Pardee Foundation, partially supported by the Gillson Longenbaugh Foundation and the University of Michigan MICHR Grant UL1TR000433, as well as the National Science Fund of China (81072170 and 81202093) and NCI research grant 1R-35CA 197585 (M.S. Wicha).

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.