Glioma regression requires the recruitment of potent antitumor immune cells into the tumor microenvironment. Dendritic cells (DC) play a role in immune responses to these tumors. The fact that DC vaccines do not effectively combat high-grade gliomas, however, suggests that DCs need to be genetically modified specifically to promote their migration to tumor relevant sites. Previously, we identified extracellular signal–regulated kinase (ERK1) as a regulator of DC immunogenicity and brain autoimmunity. In the current study, we made use of modern magnetic resonance methods to study the role of ERK1 in regulating DC migration and tumor progression in a model of high-grade glioma. We found that ERK1-deficient mice are more resistant to the development of gliomas, and tumor growth in these mice is accompanied by a higher infiltration of leukocytes. ERK1-deficient DCs exhibit an increase in migration that is associated with sustained Cdc42 activation and increased expression of actin-associated cytoskeleton-organizing proteins. We also demonstrated that ERK1 deletion potentiates DC vaccination and provides a survival advantage in high-grade gliomas. Considering the therapeutic significance of these results, we propose ERK1-deleted DC vaccines as an additional means of eradicating resilient tumor cells and preventing tumor recurrence. Mol Cancer Ther; 15(8); 1975–87. ©2016 AACR.
High-grade gliomas, particularly glioblastoma multiforme, remain the least curable tumors with a high rate of recurrence and low survival rates (1). Glioblastoma multiforme evade elimination in many ways, particularly by promoting complex neovascularization networks (2) and prohibiting antitumor immunologic surveillance (3). Therapeutic strategies will likely need to target these features from many possible angles to successfully eliminate tumors and prevent their recurrence. Feasible approaches include restoring, promoting, and maintaining natural antitumor immune responses through immune cell therapies such as dendritic cell (DC) vaccines (4).
DCs are potent antigen-presenting cells that have been shown to initiate adaptive immune responses. DC-based immunotherapies are being pursued as potential tools in curing glioblastoma multiforme. DCs loaded with patient-derived tumor antigens have been investigated in early clinical trials that have yielded a therapeutic proof-of-principle (5). However, multiple phase III trials have not yet demonstrated a clear superiority of DC vaccination over standard chemotherapy, even though they have prolonged the median and overall survival (OS) rate of patients (4). Taking advantage of the full therapeutic potential of DC vaccines will require a better understanding of their role in eliminating tumor tissue. A major limiting factor in DC vaccinations is the small proportion of DCs (fewer than 5%) that reach draining lymph nodes (LN) following administration (6). This might hamper their therapeutic efficacy. Noninvasive methods such as MRI are, therefore, necessary to study DC migration as measure of therapy outcome and might help identify strategies overcoming the poor DC migration observed in patients (6).
ERK MAP kinases have been shown to control cell migration by regulating cytoskeletal and focal contact dynamics (7). In myeloid DCs, persistent ERK activation leads to a reduced migration of DCs, while ERK inhibition is associated with a higher migratory phenotype (8). Notably, ERK activation prevented immune cell infiltration in the articular capsule in autoimmune arthritis (9). Most studies have validated the role of ERK1 in combination with ERK2; it is however becoming clearer that these isoforms have explicitly different functions (10). When we specifically studied ERK1 in a model of neuroinflammation using bone marrow chimeras, we observed an increased infiltration of CD11c+ cells close to inflamed areas in the brain (11). ERK1 thus appears to be an important regulator of DC immunogenicity and its absence precipitates severe brain autoimmunity (11).
Taking these findings into consideration, we hypothesized that ERK1 is an important regulator of DC migration and that its deletion might increase the migratory properties and vaccination efficacy of DCs toward high-grade gliomas. We used high-resolution proton (1H) and fluorine (19F) MR methods to study the role of ERK1 in tumor progression as well as DC migration. ERK1-deficient mice were more resistant to the development of WT gliomas, and their tumors were highly infiltrated with Fascin+ and CD11c+ cells. We observed a marked increase in the in vitro and in vivo migration of Erk1−/− bone marrow derived DCs (BMDC). We also showed a marked reduction in tumor size and increase in survival following treatment of WT glioblastoma multiforme mice with Erk1−/− DC, which suggests ERK1 deletion in DC vaccines as additional therapeutic strategy against high-grade gliomas.
Materials and Methods
Preparation of retroviral vectors and transduction of murine cells
The production of retroviral vectors MP71-OVA-IRES-GFP and the transduction of GL261 cells and murine T cells are presented in Supplementary Methods.
Mice and glioma animal model
All mice were handled according to the Berlin State review board at the Landesamt für Gesundheit und Soziales (LAGeSo) and internal (MDC) rules and regulations. The following mice were used: Erk1−/− mice on C57BL/6 background (12), their wild-type litermates, and Rag1−/−/OTI TCR transgenic mice recognizing the OVA-derived cognate peptide SIINFEKL on H2kb. Erk1+/+ × tdRFP and Erk1−/− × tdRFP mice were generated by intercrossing Erk1−/− knockout mice (12) to ROSA26tdRFP reporter mice (13) and the F1 generation. Following anesthesia and using stereotactic (David Kopf Instruments) coordinates referenced from bregma, glioma cells (2 × 104 cells in 1 μL) were implanted in the striatum (anteroposterior 1 mm; mediolateral ±1.5 mm; dorsoventral −4 mm).
Mouse DCs were prepared from adult C57BL/6 mice as described (11). Briefly, BMDC were generated from femurs of WT or Erk1−/− mice. Cells were grown in RPMI1640 medium supplemented with 10% FCS and 30 ng/mL GM-CSF (eBioscience). Cells were replenished with fresh GM-CSF medium every 3 days. On day 9, the fully differentiated BMDC were ready for further experiments.
In vitro migration assays
Murine CD4+ T cells were sorted and activated with anti-CD3/anti-CD28 antibodies for 72 hours. For both murine BMDC and T cells, 600 μL of RPMI-1640 with or without recombinant stromal-derived factor CXCL12 (125 ng/mL, R&D Systems) was added to 24-well plates (lower chamber). In some case, MEK inhibitor U0126 at 10 μmol/L was applied 30 minutes before the assay. Hundred μL of cell suspension (2 × 105) was added to an upper chamber consisting of a Transwell polycarbonate insert with 5-μm pore size (6.5 mm diameter; Costar Corning). The cells were then allowed to migrate at 37°C for 3 hours. Cells in the lower chamber were then collected and counted by flow cytometry.
Another method involved an agarose spot assay (14). Briefly, agarose spots with or without 2 μg/mL CCL19 (Peprotech) were placed onto 35-mm glass dishes (MatTek) and BMDC were added for 3 hours. The migration of BMDC under the spots were determined by acquiring and fusing the microscope images for all the fields of views (FOV). Image processing and analysis were done with Fiji (Image J v1.47p).
Tumor infiltrating leukocyte (TIL) single-cell suspensions (5 × 105) were stained for 25 minutes at 4°C with fluorescently conjugated antibodies in FACS buffer after blocking low affinity Fcγ receptor with CD16/CD32 antibody (BD Biosciences). The following antibodies were from Biolegend: APC-anti–mouse CD3 (clone: 145-2C11), PE-Cy7-anti–mouse CD11c (clone: N418), Pacific blue-anti–mouse CD8a (clone: RM4-5). For fascin1 (antibodies-online Inc.) staining, cells were fixed 10 minutes with 4% PFA then permeabilized 30 minutes with methanol. Cytometry acquisition was on an LSRFortessa cell analyzer (BD Biosciences) and analysis was done using FlowJo (Tree Star).
Cdc42 GTP loading and immunoblotting
BMDC were treated with CXCL12 (100 ng/mL) for 1 minutes and Cdc42 GTPase activity determined using a pull down and detection kit (Thermo Scientific). Total lysate (20 μg protein) was incubated with a GSTfusion protein, and then analyzed by 12% SDS-PAGE and immunoblotting according to the manufacturer's instructions. Guanosine 5′-O-[γ-thio] triphosphate (GTPγS) nucleotide was used as positive control. Membranes were incubated with secondary antibody and detected as previously described (15).
All in vivo MR measurements were performed on a 9.4 Tesla small animal MR system (Biospec 94/20, Bruker Biospin). Mice were anesthetized by inhalation narcosis using 0.5% to 1.5% isoflurane (Baxter), pressurized air and oxygen. Core body temperature was maintained at 37°C. Respiration rate and temperature were monitored using a remote monitoring system (Model 1025, SA Instruments Inc.). For further details on MR methods, please refer to Supplementary Methods.
Isolation and analysis of immune cells from glioma tissue
TILs were isolated as described previously (16). TILs obtained from 3 mice were pooled due to the limited cell number. Briefly, mice were transcardially perfused with PBS. Brain tumor tissue was extracted and mechanically dissociated using glass potter homogenizers (model TT57.1 Roth) into a crude suspension in RPMI 1640 complete media followed by filtering with 40-μm cell strainer (BD Falcon). Cell suspensions were spun on Percoll gradient, centrifuged at 800 g for 20 minutes, collected from the cell interface and washed two times with FACS buffer. For FACS analysis, please refer to Supplementary Methods.
Mouse brains for sectioning were prepared as described previously (14). Sixteen μm thick brain sections were mounted on glass slides. After blocking, primary antibodies were added overnight at a dilution of 1:50 for CD11c or 1:25 for CD8a (BD Pharmingen) at 4°C. Alexa 488-conjugated goat anti-guanine pig IgG (Invitrogen) or Rhodamine Red-X (Jackson ImmunoResearch) was subsequently applied. All images were taken using a confocal microscope (LSM 710, Zeiss) with a ×20 objective.
For detecting f-actin in BMDC, 1.5 × 105 cells were plated on 1 mg/mL poly-d-lysine (Sigma-Aldrich) coated glass cover slips. Cells were fixed with 4% PFA, and stained with rhodamine-coupled phalloidin (1:50; Molecular Probes, Invitrogen). Images were taken by inverse fluorescence microscope (Leica) and digitized by Leica DC Viewer 3.2. For detecting fascin1 in BMDC, 2 × 104 cells were plated on glass cover slips. Fixed cells were permeabilized with 0.01% Triton X100. The nuclei were counterstained with Hoechst 33342 (1:1,000, Sigma-Aldrich). Images were taken using a confocal microscope (LSM 710 Zeiss) with a ×40 objective.
Quantification of CD8+ tumor-infiltrating T cells
Immunofluorescent staining sections from two animals of each group (PBS, WT-DC, and Erk1−/− DC treated) of glioma-bearing mice were used for quantification of CD8+ tumor infiltrating T cells. For each section, CD8+ T cells from at least five random fields within the tumor region were counted, each field measuring 1 to 2 mm2. Cell number from each group were normalized with tumor area. Cell quantification was performed with Fiji (ImageJ v1.47p).
BMDC–T-cell coculture and ELISA
BMDC from WT and Erk1−/− mice (5 × 104), preloaded with either SIINFEKL (1 μg/mL, Biosynthan), full-length EndoGrade ovalbumin protein (1 μg/mL; Hyglos) or OVA-GL261 cell lysate (50 μg protein/ml), were cocultured with 5 × 104 OT1-TCR–transduced T cells (OT1-T cell). PMA (20 ng/mL) and ionomycin (1 μg/mL) were used for positive controls. IFNγ concentration in was determined 24 hours later by enzyme-linked immunosorbent assay (BD Biosciences).
DC vaccination protocol
BMDC from WT and Erk1−/− mice were incubated with GL261 cell lysate (50 μg protein/ml) and after 3 hours, DCs were further matured with 0.5 μg/mL LPS and incubate for 18 hours. On day 3, 7, 10, and 14 following tumor implantation, WT mice were divided into 3 groups and treated by intraperitoneal injections of either PBS or 2 × 106 BMDC (WT BMDC, n = 10; Erk1−/− BMDC, n = 10; PBS, n = 10). Survival studies were performed between different groups and the survival rate was calculated by the Kaplan–Meier method (MedCalc) using log-rank analysis.
Statistical group analysis
All data represent the average of at least triplicate samples. Error bars represent SEM. All data are presented as mean ± SEM. Data were analyzed by the Student t test when compared between 2 groups. When comparing more than 2 groups, one way ANOVA was used in Microsoft Office Excel 2010. For multiple comparisons the Bonferroni correction was applied. The Kaplan–Meier method was used for survival rate using log-rank analysis. The differences were considered statistically significant at *, P < 0.05; **, P < 0.01; and ***, P < 0.001.
Glioma growth is diminished in ERK1-deficient mice
The role of ERK1 in the glioma tumor microenvironment or its indirect impact on tumor progression is unknown. We stereotactically implanted WT GL261 glioma cells in the striatum of either WT or ERK1-deficient (Erk1−/−) mice. Using a previously established, high spatial resolution MRI protocol (14), we observed a marked reduction in glioma tumor growth in Erk1−/− compared with WT mouse brains 14 days post implantation (dpi; Fig. 1A, left). Tumor volume was significantly smaller in Erk1−/− mice (9.6 ± 1.1 mm3) compared with WT mice (18.6 ± 2.1 mm3; Fig. 1A, right, P = 0.00172). The survival rate in Erk1−/− glioma-bearing mice was also markedly improved compared with WT glioma mice (Fig. 1B, P = 0.0002). Notably, 2 of 11 Erk1−/− glioma-bearing mice exhibited complete tumor rejection. Altogether, these data indicate that the deletion of ERK1 in the stroma significantly attenuates glioma progression.
Glioma-induced immune cell infiltration is regulated by ERK1
High-grade gliomas contain brain-infiltrating peripheral immune cells among other cell types (17). We used a 19F MR method (18) to study the way leukocytes leave the circulatory system and infiltrate the tumor site in WT and Erk1−/− mice harboring WT gliomas. Before 19F/1H MRI, rhodamine-19F nanoparticles (19) were intravenously injected into WT and Erk1−/− glioma-bearing mice(18 dpi) to label leukocytes traveling within the blood circulation (Fig. 2A). These cells take up nanoparticles, migrate into the brain, infiltrate glioma tissue (tumor-infiltrating leukocytes or TILs) and can then be visualized using 19F MRI. Upon segmentation of the whole tumor region, we calculated a 1.7-fold increase of total 19F signal within tumors in Erk1−/− mice compared with WT. Considering the significantly smaller size of these tumors in the Erk1−/− mice, we next calculated the 19F signal per tumor volume, which reflects the density of 19F-labeled TILs within the tumor and observed a 2.4-fold (P = 0.0098) increase in Erk1−/− versus WT glioma-bearing mice (Fig. 2B, right). Ex vivo immunofluorescence staining also revealed that higher numbers of CD11c+/19F rhodamine–labeled cells infiltrated the tumor region in Erk1−/− brain sections compared with WT controls (Fig. 2C). Flow cytometry of TILs extracted from dissected whole tumor tissue revealed a significant increase in the frequency of CD11c+ cells in Erk1−/− mice (28.1% in WT and 42.1% in Erk1−/− mice, P = 0.0473; Fig. 2D). More than 10% of this CD11c+ population took up rhodamine-19F particles in vivo, representing cells that had infiltrated tumor tissue from the periphery. Notably, the number of CD11c+/19F rhodamine+ leukocytes was significantly higher in Erk1−/− mice (2.7 ± 0.4% from whole TIL fraction in WT and 4.9 ± 0.4% in Erk1−/− mice, P = 0.0057; Fig. 2D). Of note, a large proportion of CD11c+ cells was rhodamine negative (Fig. 2D), one reason being that CD11c is also expressed on brain resident cells such as microglia; these are abundantly present around the glioma tissue and could still be present after preparation of the tumor tissue for FACS analysis. Other than CD11c+/19F rhodamine+ cells, the CD3+/CD8+ T cells from whole TIL fraction was also significantly higher in Erk1−/− mice compared to WT controls (3.1% ± 0.6% in WT and 10.7% ± 2.2% in Erk1−/− mice, P = 0.0094; Fig. 2E), although the proportion of CD3+/CD8+ T cells that took up 19F rhodamine particles in vivo was very low (1%–2% of whole CD3+/CD8+ population). This can be explained by the low phagocytic properties of T cells.
Overall, we demonstrate a significantly increased infiltration of CD11c+ immune cells and CD3+/CD8+ T cells from the periphery into glioma tissue in Erk1−/− mice. The recruitment of CD11c+ immune cells and cytotoxic T cells into glioma tissue is necessary for an efficient antitumor response (20, 21). Therefore, an increased proportion of immune cells in tumor tissue of Erk1−/− mice (Fig. 2D and E) might explain the increased resilience of these mice toward developing gliomas (Fig. 1A) and their significantly increased rate of survival (Fig. 1B).
ERK1 deficiency enhances DC migration
The previous data (Fig. 2) show that a deficiency in ERK1 increases the accumulation of TILs. This is in line with previous findings that ERK activation after pharmacologic inhibition of MEK-1 reduces DC migration toward the chemokines CCL21 (8), CCL19 and CXCL12 (22). However, MEK-1 phosphorylates both ERK1 and ERK2, and these isoforms—despite their striking homology—have explicitly different roles. Considering the role of ERKs, in general, in controlling cell migration (7), we next investigated the role of ERK1 in DC and T-cell migration through in vitro migration assays. CXCL12 is highly expressed in brain tumors and is a potent chemoattractant of CXCR4-expressing immune cells, including DCs and T cells (23). A modified Boyden chamber assay revealed that in comparison to WT T cells, Erk1−/− T cells exhibit a slight but not significant decrease in migration toward CXCL12 (Fig. 3A). MEK inhibition by UO126 (which inhibits both ERK1 and ERK2 activation) did not affect T-cell migration in WT or Erk1−/− T cells. In contrast, Erk1−/− BMDC revealed significantly increased migration toward CXCL12 compared with WT BMDC (Fig. 3B). With the application of UO126, BMDC migrated to a slightly but not significantly lower extent toward CXCL12 (Fig. 3B). Migration inhibition following UO126 occurred at equal levels in WT and Erk1−/− BMDC.
The chemokine CCL19 plays a key role in the trafficking of DCs into LNs (24). To confirm the role of ERK1 in DC migration, we applied another in vitro assay. We observed a significant increase in the numbers (>3 folds, ***P < 0.001) of Erk1−/− BMDC migrating toward CCL19-containing agarose spots 3 hours after plating (Fig. 3C). Considering the increased migration of Erk1−/− BMDC toward CCL19, we then studied expression of CCR7 on these cells to determine possible mechanisms for the increased migratory properties. However, we did not observe an increase in CCR7 expression in immature and mature Erk1−/− BMDC when compared with WT BMDC (Fig. 3D).
Given that migration is an important feature for DC homing to lymphoid organs (24), we next developed a noninvasive in vivo DC migration assay involving 19F/1H MRI and a labeling of BMDC with 19F nanoparticles in vitro before their application in vivo (25, 26). In a WT mouse, one limb received 5 × 106 WT BMDC intradermally, and the other limb 5 × 106Erk1−/− BMDC. Using this method, we observed an increased migration of 19F-labeled Erk1−/− BMDC (compared with WT BMDC) toward WT popliteal LN as shown by the 19F-signal in 19F/1H MRI (Fig. 4A). Ex vivo19F MR spectroscopy of the extracted popliteal LNs (Fig. 4A) revealed at least twice as many 19F-labeled Erk1−/− BMDC (26.1 × 103 cells) reach the corresponding LNs compared with WT BMDC (11.8 × 103 cells; P = 0.028; Fig. 4B).
We attributed the increase in the in vivo19F signal following Erk1−/− BMDC application (Fig. 4A and B) and in Erk1−/− gliomas (Fig. 2B and C) to an increase in cell mobilization in tissue. However, an increased 19F-signal could also be the result of an increased 19F nanoparticle uptake by Erk1−/− DC, inspite of our previous observations showing that Erk1−/− BMDC are more mature than WT BMDC (11). We then measured phagocytosis in BMDC following overnight incubation with 19F nanoparticles (same protocol as for the in vivo experiments) by performing 19F MR spectroscopy. We did not observe any significant differences between WT and Erk1−/− BMDC to take up 19F nanoparticles (Fig. 4C). We also did not observe any influence of the 19F nanoparticles themselves to significantly alter the phagocytic or migratory properties of both WT and Erk1−/− BMDC (data not shown).
ERK1 controls cytoskeletal changes in DCs
The Rho GTPase Cdc42 is a master regulator of DC polarity, and is thus necessary for processes such as the reorientation of the microtubule-organizing center during endocytosis (27) and leading-edge coordination as decisive factors for DC motility in vivo (28). Interestingly, DCs have been shown to downregulate Cdc42 activity following activation (27). To determine ERK1's effects on Cdc42 regulation during DC migration, we performed Cdc42 activity assays in Erk1−/− and WT BMDC. Surprisingly, although Cdc42 activity was downregulated in WT BMDC, this was not the case for Erk1−/− BMDC following stimulation with CXCL12 chemokine (Fig. 5A).
Altogether our findings (Figs. 3 and 5A and B) suggest that DC polarization is increased in the absence of ERK1. We thus measured filamentous actin (f-actin) in BMDC by FACS and found larger quantities of f-actin in Erk1−/− BMDC compared with WT BMDC (Fig. 5B).
The large cytoskeletal protein talin is an adaptor protein associated with actin filaments and is required for actin polymerization. Immunofluorescence microscopy revealed an increase in cytoskeletal protrusions and a denser expression of f-actin and talin in Erk1−/− BMDC, particularly within the intracellular compartment (Fig. 5C). This indicates a higher level of cytoskeleton organization in Erk1−/− BMDC and might explain our observations of an increase in their motility and migratory potential in vitro (Fig. 3) and in vivo (Fig. 4). Furthermore, the actin-bundling protein fascin1 is selectively expressed on DCs upon maturation (29) and is critical for the assembly of filopodia, thereby increasing cell motility (30). As shown in Fig. 5D (top row), immature BMDC expressed lower levels of fascin1 in WT and Erk1−/− BMDC. Upon LPS-induced maturation, fascin1 expression was upregulated, an effect that was more pronounced in Erk1−/− compared with WT BMDC (Fig. 5D, bottom row).
Our data indicate that in contrast with WT BMDC, Erk1−/− BMDC maintain a high level of Cdc42 activity following activation and exhibit an abundantly polymerized cytoskeleton with fully assembled filopodia. This might aid the infiltration of DCs into tumor tissue (Fig. 2C and D). We also investigated the expression of fascin1 in CD11c+ cells that infiltrated glioma tissue. In glioma-bearing Erk1−/− mice, the number of fascin1-expressing CD11c+ DCs in TILs isolated from dissected glioma tissue (18 dpi) was significantly higher than that from glioma tissue from WT mice (Fig. 5E, P = 0.00334). These data provide further evidence of increased DC migration in the absence of ERK1, which could indicate a more potent antitumor response in ERK1-deficient DCs.
ERK1 deficiency does not influence DC–T-cell interaction
A successful immune response to tumors depends on an effective crosstalk between DCs and T cells in priming tumor-specific CD8+ T cells (31). Since Cdc42-associated pathways are also necessary for microtubule-organizing center reorientation in DCs during antigen presentation to T cells (32), we investigated whether Erk1−/− BMDC mount a more potent antitumor T-cell response. To explore this, we generated OVA-GL261 and transduced T cells with a T-cell receptor (TCR) that specifically recognizes OVA-derived peptide in the context of MHC I (H-2Kb; please refer to Supplementary Methods). To test the specificity of the OVA-GL261 system, we stereotactically implanted OVA-GL261 cells into the striatum of WT mice and OT1 mice (CD8 T-cell TCR-transgenic recognizing OVA). Fourteen days following tumor cell inoculation, MRI revealed OVA-GL261-induced glioma formation in the WT mice (Fig. 6A, left) but no sign of tumor formation in the OT1 mice (Fig. 6A, right). This confirmed the tumorigenicity of the OVA-GL261 glioma cells and efficient antigen recognition by OT1 T cells.
We next investigated the capacity of Erk1−/− BMDC to cross-prime T cells. BMDC were isolated from WT and Erk1−/− mice and then pulsed with various antigens (LPS, OVA peptide, or OVA-GL261 lysate). After coculture with OT1-transduced T cells, we measured the secretion of IFNγ by ELISA. Secretion of IFNγ has been proven to reflect both the extent of DC-mediated T-cell cross-priming as well as cytotoxic properties of tumor antigen-specific CD8+ T cells (33). We did not observe any differences between the capacity of WT and Erk1−/− BMDC to prime IFNγ secretion in tumor-specific T cells (Fig. 6B).
Antitumor activity of ERK1 deficient DC vaccines against high-grade glioma
Brain tumor regression requires the recruitment of functional DCs into the glioma microenvironment (20). Because we observed a higher migratory potential of Erk1−/− BMDC in vivo, we next asked whether ERK1 deficiency might have a beneficial impact on DC vaccination in glioblastoma multiforme. To examine the therapeutic efficacy of Erk1−/− BMDC, we vaccinated glioma-bearing mice twice a week with WT or Erk1−/− BMDC, which were pulsed with GL261 lysate, and performed 1H MRI 18 dpi (Fig. 7A). Tumors in both WT BMDC-treated mice and Erk1−/− BMDC-treated mice were smaller than those in PBS-treated glioma-bearing mice (Fig. 7B, right), although tumors in the Erk1−/− BMDC-treatment group showed a greater difference in size (8.4 ± 2.9 mm3, compared with 30.9 ± 2.8 mm3 for the PBS-treated group; P = 0.000035) than the Erk1−/− BMDC treatment group (20 ± 5 mm3, compared with 30.9 ± 2.8 mm3 for the PBS-treated group; P = 0.0803). No significant difference in size was observed between the Erk1−/− and WT BMDC-treated mice. Consistent with previous observations made in glioma-bearing mice (34), mice receiving WT BMDC pulsed with GL261 lysate exhibited significantly enhanced survival when compared with untreated PBS controls (P = 0.0019; Fig. 7C). Furthermore, in line with a markedly decreased tumor size, we documented significantly improved survival probability in glioma-bearing mice receiving Erk1−/− BMDC compared with those of untreated mice (P = 0.0005; Fig. 7C). However, no significant difference in survival was observed between Erk1−/− and WT BMDC-treated mice.
We next wanted to study the distribution of both Erk1−/− and WT BMDC in the glioma model following intraperitoneal application (Fig. 7D). For this, we used BMDC that were derived from either (Erk1−/− × tdRFP) mice or [Erk1+/+(WT) × tdRFP] littermate control mice. Although we could not detect any of the injected RFP BMDC (both WT and Erk1−/−) in the brain of glioma mice 21 dpi, we observed a significant increase in CD11c+ RFP+ BMDC in the lymph nodes of mice treated with (Erk1−/− × tdRFP) BMDC when compared to mice treated with [Erk1+/+(WT) × tdRFP] BMDC (Fig. 7D). These results in the glioma model are in line with our in vitro assays (Fig. 3) and in vivo experiments in WT healthy mice (Fig. 4A).
Although we could not detect any of the intraperitoneally administered (WT or Erk1−/−) BMDC in the brain of glioma-bearing mice, we observed differences in the expression of tumor-infiltrating CD8+ T cells between the different treatment groups (Fig. 7E). Although tumor sections demonstrated visibly smaller tumors when treated with BMDC, the number of CD8+ T cells was significantly increased in sections derived from mice treated with WT BMDC compared with PBS-treated groups (1,359 ± 39 cells/mm2 and 716 ± 24 cells/mm2, respectively; P = 0.001). The infiltration of CD8+ cells was further enhanced when treating mice with Erk1−/− BMDC (1,920 ± 49 cells/mm2; P = 0.02 when compared with WT BMDC; Fig. 7E).
In this report, we identify ERK1 as a regulator of DC migration and show that the introduction of DCs lacking ERK1 lead to a significant reduction in tumor growth as well as improved survival of glioma-bearing mice.
DCs need to actively migrate between lymphatic tissue and interstitial spaces to initiate adaptive immune responses. Here, we demonstrate that ERK1 plays a crucial role in the migratory capacity of DCs. This key ERK–MAPK intracellular signaling pathway transduces a broad range of extracellular stimuli into important biologic responses, including cell survival, proliferation, differentiation, and the regulation of DCs (9, 35) and T cells (36). The specific role of ERK1 has been studied in T cells and it was shown that ERK1 does not appear to play a direct role on the effector function of antigen-specific T cells, notwithstanding increased susceptibility toward T-cell–mediated autoimmunity (37). In DCs, ERK1 inhibits cytokine production following their stimulation by TLRs (38) and reduces surface receptors (11), which in turn affects the priming of naïve T cells toward an effector phenotype during autoimmunity (11). Most studies of tumor pathology have not distinguished ERK1 from ERK2 function, due to the assumption that they have compensatory effects (39). However, in some biologic contexts the distinction between these two kinases appears to be crucial, considering that ERK2-deficient mice are embryonically lethal (40) in contrast with ERK1-deficient mice, which are viable and fertile (12). Here we report for the first time a role for ERK1 in negatively regulating the migration specifically of DCs by showing a significant increase in migration in Erk1−/− BMDC, both in in vitro and in vivo experiments. In vitro, we distinguished between the impact of ERK1 on BMDC and effector T cells. Although ERK1 is expressed in both cell types, its role in migration was more pronounced in DC than in T cells. Furthermore, while ERK1 deletion promoted DC migration, application of a MEK inhibitor (U0126) inhibited DC migration. As U0126 inhibits both ERK1 and ERK2 activation, this might have masked the increase in migration we observe in the absence of ERK1 and could furthermore explain discrepancies in earlier studies using inhibitors to ascertain the signals' roles in migration-related processes (41).
Migratory phenotypes typically involve mechanisms related to cytoskeletal organization and structure, so we studied this aspect of ERK1's influence. We showed that ERK1 influences the cytoskeleton and associated adaptor and actin-bundling proteins such as talin and fascin1, another strong piece of evidence for a role of ERK1 in the regulation of DC migration. We chose fascin1 as a marker for DCs in the tumor tissue, because CD11c+ is not an exclusive marker for these cells in brain gliomas: CNS resident and nonresident immune cells such as peripheral macrophages (42) are also CD11c+. Importantly, fascin1 is not expressed on activated microglia (43) and peripheral blood cells, including macrophages and neutrophils (44).
A role for ERK1 in DC migration is additionally supported by our finding that ERK1 deletion promotes Cdc42 activation, because Cdc42 is responsible for leading edge coordination in vivo (28). Notwithstanding the role of Cdc42 in DC polarity and orientation during antigen presentation to T cells (32), Erk1−/− DCs did not present OVA-GL261 (as model–tumor antigen) to T cells more effectively than WT DCs. This excludes tumor antigen presentation as an underlying mechanism for the decreased susceptibility of Erk1−/− mice to develop gliomas and for the increased infiltration of T cells in WT gliomas in these mice compared with WT mice. The observation indicates that the benefits of ERK1 deficiency in reducing the growth of gliomas likely derive from an ultimate increase in immune cell surveillance in the glioma tissue as shown by an increase in CD11c+ leukocytes, fascin1+ DCs and CD8+ T cells in Erk1−/− glioma-bearing mice as well as increase in CD8+ T cells in Erk1−/− BMDC-treated glioma-bearing mice. From our results, we believe that ERK1 deficiency is better adapted for providing a higher availability/localization of BMDC in the periphery rather than increasing tumor antigen presentation, and therefore priming of antigen-specific cytotoxic T cells. Although we did not detect the administered Erk1−/− BMDC in the tumor tissue following therapeutic application in the glioma model, we did observe a significantly increased proportion of Erk1−/− RFP+ cells in the LNs and an increased CD8+ population of T cells infiltrating the tumor tissue.
Ultimately, improved immune cell surveillance would also augment the reactivation and expansion of cytotoxic T cells within tumor-relevant sites. Our results suggest that the significant reduction in glioma growth by Erk1−/− BMDC in contrast with WT BMDC could be the result of an improvement in the delivery of Erk1−/− BMDC to the lymphatic tissue, which eventually results in a larger infiltration of CD8+ T cells into the tumor tissue. A recent NIH-funded clinical trial found that improving the LN homing of DCs in glioblastoma multiforme patients significantly enhances the therapeutic efficacy of tumor antigen-loaded DCs (45). Their finding and our study suggest that monitoring DC migration might deliver a predictive readout for the effectiveness of DC vaccines.
Here, we used 19F MR to do this, and although still in early clinical development (46), it might offer significant advantages in the future over other methods to track DC migration such as scintigraphy combined with radioisotope 111In-oxine (47). MRI in general was shown to be better at localizing DC vaccines in vivo (48); paramagnetic iron oxide nanoparticles were used to label and follow cells in melanoma patients (48). Even though this represented a major step forward in the tracking of DCs, lower image contrast often makes it difficult to distinguish MRI signal hypointensities originating from magnetically labeled cells from hypointensities caused by endogenous factors (e.g., deoxygenated blood). 19F MR-based methods were simultaneously introduced in animal models to overcome these hurdles in cellular MR imaging (25).
The elegance of 19F MR lies in the fact that it provides background-free images in mammals and permits a highly selective detection of cells in vivo throughout the organism's body (49). The main hurdle is signal sensitivity, which can pose a challenge when imaging 19F-labeled DCs in cancer patients (46). In a recent report, we aimed to promote the signal achieved per unit cell; we enriched 19F nanoparticles with the phosphatidylethanolamine DPPE and observed a stronger enhancement of 19F signal (from 74 to 771 nmol) per 106 cells, which equates to an order of magnitude increase in 19F spins from 0.89 × 1012 19F spins (control nanoparticles) to 0.93 × 1013 19F spins (in DPPE-enriched nanoparticles) per DC unit (19). In the present study, we made use of the recently developed 19F marker (19) to label and track DCs in vivo with greater sensitivity and to determine the role of ERK1 in DC migration.
We can follow the 19F signal from DCs that we had administered and showed that Erk1−/− DCs migrate more readily than WT DCs into popliteal LNs. We also quantified the 19F signal and thereby the number of DCs in draining LNs by ex vivo19F MRS and observed a significant increase in the number of Erk1−/− DCs compared with WT DCs that appeared in the corresponding draining LNs. Although further developments in 19F MR will be needed to expand its use in clinical applications (50), it is a valuable tool to monitor DC homing in cancer therapy and to effectively interpret the in vivo distribution of DCs as a means of assessing the effectiveness of these therapies in glioblastoma multiforme patients. The advent of cryogenically cooled MR detectors (51) will help boost 19F MR sensitivity and lower detection levels of 19F-labeled cells to facilitate in vivo cell tracking within shorter scan times.
Here, we identified ERK1 as a molecular player that negatively regulates the mobility of DCs and their capacity to eliminate malignant gliomas. The results were demonstrated in a preclinical mouse glioblastoma multiforme model and further studies will be required to predict the safety, efficacy and potential benefits of Erk1−/− DC vaccines in human high-grade or recurrent gliomas. It should be noted that Erk1−/− BMDC, although significantly reducing glioma growth in comparison with untreated controls, were not significantly different to WT BMDC (P = 0.05196), even though WT BMDC did not show significant differences to untreated controls. Furthermore, statistically significant treatment effects with regard to survival were observed by both Erk1−/− BMDC (P < 0.001) and WT BMDC (P < 0.01) groups. One caveat in this study was that treatment was terminated at day 14 and not later as has been done in other glioma studies (52). Future studies, perhaps using more advanced 19F MR methods, should aim at studying changes in therapeutic efficacy when using different treatment schedules that extend over more than two weeks. The present study was restricted to the role of ERK1 against glioblastoma multiforme. Effective therapies will likely need to target the tumor from many different angles, employing several approaches toward tumor cell elimination while preserving sufficient natural antitumor immune responses. Other strategies such as the preconditioning of application sites with recall antigen to promote DC vaccination in humans and mice (45), as recently reported further emphasizes the need to promote DC migration as a means of optimizing DC vaccines. As negative regulator of DC migration, ERK1 appears to be an attractive target for deletion in genetically modified vaccines. Other possible molecular targets include molecules such as IRF4 (53) and CD37 (54), which favor DC migration. It has now become practical to apply simultaneous gene silencing and retroviral transgenic insertion; such combinations have proven to be highly effective (55, 56). Ex vivo gene silencing in immune cell therapeutics using both viral- and nonviral-based approaches has been an area of intensive research in clinical trials during the last decade (57)
In summary, our study presents ERK1 as one target for manipulating DC vaccines to promote their migration in an animal model of glioblastoma multiforme. Furthermore, this study underscores the benefits and needs of using 19F MRI to measure in vivo cell migration and localization, which are critical parameters in the effectiveness of such vaccines. ERK1 and other molecular targets that enhance DC migration could be important tools in developing next generation of DC vaccines (45). The OS of glioblastoma multiforme patients could well be improved by patient-specific treatment regimens that use immunotherapy based on genetically modified cells in conjunction with standard treatment protocols (surgery, radiotherapy, chemotherapy, and antiangiogenic therapy).
Disclosure of Potential Conflicts of Interest
T. Niendorf has ownership interest (including patents) in MRI TOOLS GmbH. S. Waiczies reports receiving a commercial research grant from Novartis. No potential conflicts of interest were disclosed by the other authors.
Conception and design: M.-C. Ku, M. Günther, C. Martin, S. Waiczies
Development of methodology: M.-C. Ku, I. Edes, I. Bendix, A. Pohlmann, H. Waiczies, C. Martin, S. Waiczies
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.-C. Ku, I. Edes, T. Prozorovski, M. Günther, C. Martin, G. Pagès, S.A. Wolf, T. Niendorf, S. Waiczies
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.-C. Ku, I. Edes, I. Bendix, A. Pohlmann, M. Günther, C. Martin, H. Kettenmann, T. Niendorf, S. Waiczies
Writing, review, and/or revision of the manuscript: M.-C. Ku, I. Edes, I. Bendix, G. Pagès, S.A. Wolf, H. Kettenmann, W. Uckert, T. Niendorf, S. Waiczies
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.-C. Ku, I. Edes, I. Bendix, W. Uckert, S. Waiczies
The authors thank S. Kox and Y. Balke for assistance with MR measurements and cell culturing, and M. Naschke for assistance with ELISA measurements. The authors also acknowledge the support of the Animal Facilities and the technology platforms for Advanced Light Microscopy and Preparative Flow Cytometry of the Max Delbrück Center for Molecular Medicine. The authors are also grateful to Prof. K. Rajewsky for valuable discussions and scientific writer R. Hodge for support in article editing.
This study was funded by the Deutsche Forschungsgemeinschaft (to S. Waiczies; DFG/WA2804).
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.